WATER RESEARCH A Journal of the International Water Association
Editor-in-Chief Mark van Loosdrecht Delft University of Technology and KWR Watercycle Research The Netherlands E-mail:
[email protected]
Editors Jean-Claude Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai University of Tokyo Japan Xiaodi Hao Beijing University of Civil Engineering and Architecture China Gregory Korshin University of Washington USA Anna Ledin Formas Sweden Eberhard Morgenroth ETH Zürich and Eawag, Swiss Federal Institute of Aquatic Science and Technology Switzerland Wolfgang Rauch University of Innsbruck Austria Maria Reis Universidade Nova de Lisboa/FCT Portugal Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea Thomas Ternes Bundesanstalt für Gewässerkunde Germany Stefan Wuertz University of California, Davis USA
Associate Editors Andrew Baker University of New South Wales Australia Damien Batstone University of Queensland Australia
Guang-Hao Chen Hong Kong University of Science & Technology China Tom Curtis University of Newcastle upon Tyne UK Francis de los Reyes III North Carolina State University USA Ana Deletic Monash University USA Rob Eldridge Glen Waverley Australia Monica B. Emelko University of Waterloo Canada Rosina Girones University of Barcelona Spain Stephen Gray Victoria University Australia Kate Grudpan Chiang Mai University Thailand Edwin Herricks University of Illinois at Urbana-Champaign USA Peter Hillis AECOM Australia Hong-Ying Hu Tsinghua University China Bruce Jefferson Cranfield University UK Ulf Jeppsson Lund University Sweden Sergey Kalyuzhnyi RUSNANO Russian Federation Jae-Hong Kim Georgia Institute of Technology USA Günter Langergraber University of Natural Resources and Applied Life Sciences, Vienna (BOKU) Austria
Yoshihiko Matsui Hokkaido University Japan Armand Maul Université de Metz France Max Maurer Eawag Switzerland How Yong Ng National University of Singapore Singapore Satoshi Okabe Hokkaido University Japan Jong Moon Park Pohang University of Science & Technology Korea
Shang-Lien Lo National Taiwan University Taiwan Dionisis Mantzavinos Technical University of Crete Greece
Anastasios Zouboulis Aristotle University of Thessaloniki Greece
Susan Richardson US Environmental Protection Agency USA Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Jean-Philippe Steyer LBE-INRA France Masahiro Takahashi Hokkaido University Japan Nathalie Tufenkji McGill University Canada Kai Udert Eawag Switzerland Vayalam Venugopalan BARC Facilities India Eduardo von Sperling Federal University of Minas Gerais Brazil Julie Zilles University of Illinois at Urbana-Champaign USA
Editorial Office E-mail:
[email protected]
Publication information: Water Research (ISSN 0043-1354). For 2011, volume 45 is scheduled for publication. Subscription prices are available upon request from the publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/watres). Further information is available on this journal and other Elsevier products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon request. Claims for missing issues should be made within six months of the date of dispatch. Author Enquiries: For enquiries relating to the submission of articles (including electronic submission) please visit this journal’s homepage at http://www.elsevier.com/ locate/watres. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. You can track accepted articles at http://www.elsevier.com/trackarticle. You can also check our Author FAQs at http://www.elsevier.com/authorFAQ and/or contact Customer Support via http://support.elsevier.com. Orders, claims, and journal enquiries: Please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 3251 Riverport Lane, Maryland Heights, MO 63043, USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4478878 [outside the USA]; fax: (+1) (314) 4478077; e-mail:
[email protected] Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865) 843970; e-mail:
[email protected] Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3) 5561 5037; fax: (+81) (3) 5561 5047; e-mail:
[email protected] Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510; e-mail:
[email protected] Application for membership of International Water Association should be made to: Executive Director, IWA, Alliance House, 12 Caxton Street, London SW1H 0QS, U.K. (Tel.: +44 207 654 5500; Fax: +44 207 654 5555; e-mail:
[email protected]; website: http://www.IWAhq.org.uk). Registered Charity (England) No. 289269. Individual membership is available from £30 upwards. For details contact IWA.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Model selection, identification and validation in anaerobic digestion: A review Andres Donoso-Bravo a,*, Johan Mailier a, Cristina Martin b, Jorge Rodrı´guez c,d, Ce´sar Arturo Aceves-Lara e,f,g, Alain Vande Wouwer a a
Automatic Control Laboratory, University of Mons 31 Boulevard Dolez, B-7000 Mons, Belgium modelEAU, De´partement de ge´nie civil et genie des eaux, Universite´ Laval, 1065 av. de la Me´decine, Que´bec (QC) G1V 0A6, Canada c Department of Chemical Engineering, University of Santiago de Compostela, Spain d Masdar Institute of Science and Technology, Abu Dhabi, United Arab Emirates e Universite´ de Toulouse, INSA, UPS, INP, LISBP, 135 Avenue de Rangueil, F-31077 Toulouse, France f INRA, UMR792, Inge´nierie des Syste`mes Biologiques et des Proce´de´s, F-31400 Toulouse, France g CNRS, UMR5504, F-31400 Toulouse, France b
article info
abstract
Article history:
Anaerobic digestion enables waste (water) treatment and energy production in the form of
Received 13 June 2011
biogas. The successful implementation of this process has lead to an increasing interest
Received in revised form
worldwide. However, anaerobic digestion is a complex biological process, where hundreds
26 August 2011
of microbial populations are involved, and whose start-up and operation are delicate
Accepted 29 August 2011
issues. In order to better understand the process dynamics and to optimize the operating
Available online 3 September 2011
conditions, the availability of dynamic models is of paramount importance. Such models have to be inferred from prior knowledge and experimental data collected from real plants.
Keywords:
Modeling and parameter identification are vast subjects, offering a realm of approaches
Anaerobic digestion
and methods, which can be difficult to fully understand by scientists and engineers
Modeling
dedicated to the plant operation and improvements. This review article discusses existing
Identification
modeling frameworks and methodologies for parameter estimation and model validation
Kinetic parameters
in the field of anaerobic digestion processes. The point of view is pragmatic, intentionally
Sensitivity analysis
focusing on simple but efficient methods. ª 2011 Elsevier Ltd. All rights reserved.
Contents 1. 2. 3. 4.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modeling procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mathematical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Available measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Experimentation mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1. Batch operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2. Continuous operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3. Fed-batch operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
* Corresponding author. Tel./fax: þ32 065 374. 130. E-mail address:
[email protected] (A. Donoso-Bravo). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.059
5348 5348 5349 5350 5350 5352 5352 5353 5353
5348
5. 6.
7.
8.
9.
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
Structural adequacy and parameter identifiability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods for parameters estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Cost function selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2. Optimization techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1. Local methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2. Global methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. What about optimization constraints? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4. Alternative methods: The Bayesian inference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5. Some considerations for parameter estimation in AD models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.1. Steady states analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.2. Mass continuity (conservation laws) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.3. Initial conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Parameter uncertainty estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1. Error covariance matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2. Confidence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3. Joint posterior distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Model validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1. Direct validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2. Cross validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction
Anaerobic Digestion (AD) is a chain of interconnected biological reactions, where the organic matter (in the form of carbohydrates, proteins, lipids or more complex compounds), is transformed into methane, carbon dioxide and anaerobic biomass, in an oxygen-free environment. This biological process is used to simultaneously treat waste and wastewater and to produce biogas. AD is now considered as a consolidated technology with more than 2200 high-rate reactors implemented worldwide (Van Lier, 2008). In Europe’s case, between 1995 and 2010, the number of plants installed increased from 15 to 200, which implies an installation capacity rise of nearly 6,000,000 tons per year (from 200,000 to 6,000,000 tons per year) (de Baere et al., 2010). Moreover, the number of AD reactors is expected to increase due to both climate change awareness and the significant boost in the use of renewable energy. The main characteristics of the process, such as reactor design issues, microbial aspects, inhibition phenomena, and of course, the strategic advantages of this process, are well described in the literature (Appels et al., 2008; Chen et al., 2008; Ward et al., 2008). Likewise, a thorough description of the role of the microorganisms in the bioenergy production, with a special emphasis in the anaerobic digestion process, can be found in Rittmann (2008). Mathematical models enable the representation of the main aspects of a biological system. They improve the understanding of the system, the formulation and validation of some hypothesis, the prediction of the system’s behavior under different conditions, reducing, consequently, the experimental information requirements, costs, risk and time. The proper evaluation and application of mathematical models in bioprocesses must follow several stages if the final goal of the approach is to generate useful tools to improve the
5353 5354 5354 5355 5355 5356 5356 5356 5357 5357 5357 5358 5358 5358 5358 5359 5359 5359 5359 5359 5360 5360
understanding of the process or to predict the behavior of the system. This issue has been addressed, in a more general context of environmental models, by Jakeman et al. (2006), from a in-depth theoretical point of view by Walter and Pronzato (1997) and specifically for wastewater treatment processes by Dochain and Vanrolleghem (2001). Several mathematical models of anaerobic digestion have been proposed in the last two decades and a variety of methods have been used for parameter estimation and model validation. AD process is characterized by its high complexity and non-linearity and by the difficulty to collect large amounts of informative experimental data for modeling purposes. One of the consequences of the latter is variety of approaches to modeling and parameter identification is the important variability in values reported for the kinetic parameters, even when the same operational and environmental conditions have been evaluated. This paper presents an overview of the main procedures that can be used for developing and assessing dynamic models of the anaerobic digestion process. It is structured according to a step-by-step approach of the modeling task, i.e., from model selection up to model validation.
2.
Modeling procedure
The whole modeling process (selecting a model structure, identifying the parameter values, and planning the experimental measurements) should be in coherence with the objective pursued. In general, the three most common objectives of using a model are: understanding the system’s behavior and interaction of components; quantitatively expressing or verifying our hypothesis and predicting the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
behavior of the system in the future or under other similar circumstances. Adequate model structures should be chosen according to four principles (Spriet, 1985): (i) simplicity, the model should be as simple as possible; (ii) causality, the model should represent the most relevant causeeeffect relationships; (iii) identifiability, the values of the unknown parameters should be identifiable from the available data; and (iv) predictive capability, the model should remain valid under future or alternative reasonable conditions. As stated by Flotats et al. (2003) model identification and parameter estimation have not been given the same attention in anaerobic digestion processes, as it does with activated sludge systems in which considerable efforts have been rightfully devoted (Ossenbruggen and Stevens, 1996; Weijers, 2003; Liwarska-Bizukojc and Biernacki, 2010). Fig. 1 shows a schematic view of the parameter estimation and model validation procedure. At first, it is of course very important to define the purpose of the modeling exercise. An explicative (mechanistic) model intended for process investigation and hydraulic/chemical/biological analysis will likely include a detailed description of specific mechanisms and phenomena, which would probably be irrelevant for a global dynamic analysis or the design of controllers, for instance (Jakeman et al., 2006). Therefore, the level of details of the description has to be selected with care depending on the targeted application of the model (physical/chemical/biological investigation, process design, dynamic simulation, optimization, control, supervision). Once an appropriate model structure has been selected (usually a system of non-linear differential equations including a number of unknown or uncertain parameters), a simulator can be implemented using a platform of choice (a programming language such as Fortran or C, or an environment such as Matlab or its open-source counterparts Octave and Scilab). Local and global parametric sensitivity analysis can then be used to assess, on the one hand, the most influential parameters, and on the other hand, the parameters
Fig. 1 e Parameter estimation procedure.
5349
with weaker influences on the measured outputs (at least in the scenario under consideration), possibly involved in correlation with other parameters. This first analysis can lead to a model reformulation or simplification, eliminating correlated parameters, and it can also lead to the reformulation of a new experimental design and data collection methods in order to have more informative data in relation with the parameters to be estimated. Next, the experimental data must be examined in terms of potential errors (outliers, missing data, etc.) and the deviation between the model prediction and the measured outputs. In this last part, special attention has to be paid to the selected cost function. Finally, the model has to be evaluated with regard to the experimental data used so far (direct validation), as well as fresh data (unseen in the identification process e cross validation). These last two steps, if unsuccessful, can lead back to model reformulation and/or experiment design and data collection.
3.
Mathematical models
Mathematical modeling of the anaerobic digestion process was motivated by the need for efficient operation of anaerobic systems in the early 70’s. The first models were relatively simple due to the limited knowledge about the process. Experimental investigation, further system analysis and the increase in computing capacity lead to the development of much more detailed models in recent years. As it is not the goal of this review to list the available models in anaerobic digestion, a brief overview is given in the next paragraphs. The first modeling approaches focused on describing the limiting step of the process, considering that anaerobic digestion is a multistep process where one slower step controls the global rate (Hill and Barth, 1977). Such limiting step can, however, be different under different operating conditions (Speece, 1996). Some authors considered methanogenesis as the limiting step or the conversion of fatty acids into biogas or the hydrolysis of suspended solids (Eastman and Ferguson, 1981). These series of models were simple and easy to use but were unable to adequately describe the process performance, especially under transient conditions. A second generation of models considered the concentration of volatile fatty acids as the key parameter, incorporating acidogenesis and acetogenesis separately (Hill, 1982). The hydrogen partial pressure, as a key regulatory parameter influencing the redox potential in the liquid phase and more bacterial groups, with differentiated acetoclastic and hydrogenotroph methanogens, was included in several models (Costello et al., 1991; Ruzicka, 1996). The redox potential (as NADH/ NAD þ ratio) is a function of the hydrogen partial pressure and determines the VFA production in this family of models. Further microbiological studies led to another generation of models (Angelidaki et al., 1993, 1999; Siegrist et al., 1993; Vavilin et al., 1994, 1995; Kalyuzhnyi and Davlyatshina, 1997; Kalyuzhnyi, 1997; Kalyuzhnyi and Fedorovich, 1998; Batstone et al., 2000; Tartakovsky et al., 2002; Keshtkar et al., 2003; Haag et al., 2003). These models incorporated additional processes and species, more detailed kinetics with inhibition and considered different substrates.
5350
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
As a response to the need for a generic model of anaerobic digestion, the IWA Task Group for Mathematical Modeling of Anaerobic Digestion Processes developed the generic Anaerobic Digestion Model No.1 (ADM1) (Batstone et al., 2002) in order to reach a common basis for further model development and validation studies with comparable results. The ADM1 model describes the dynamics of 24 species and includes 19 bioconversion processes. The latter, makes ADM1 a model with a large number of parameters. In view of its general purpose, the ADM1 neglects some processes and species, which are related to more specific applications, in order to avoid extreme complexity. Still, the large number of parameters and identifiability difficulties are the major drawbacks of ADM1, as well as, some structural weaknesses (Kleerebezem and Van Loosdrecht, 2004). Many applications based on the ADM1 have been published in recent years. Some authors applied the model to stirred tank systems while others considered distributed parameter systems (Batstone et al., 2004a, 2005). Extensions have been developed to incorporate processes that were absent in the original model. Also, reports of the applications of the ADM1 to particular types of wastewater have been published (Batstone and Keller, 2003; Fedorovich et al., 2003; Batstone et al., 2004b; Fezzani and Ben Cheikh, 2008; Fezzani and Ben Cheikh, 2009; Derbal et al., 2009; Gali et al., 2009; Lee et al., 2009; Ramirez et al., 2009; Ozkan-Yucel and Gokcay, 2010). The framework provided by the ADM1 is useful especially for process design and dynamic simulation. Due to its fixed stoichiometry approach, its applicability would, however, require important structural modifications for some processes. Implications of structural changes in some processes of the ADM1 toward a variable stoichiometry structure have been recently analyzed (Rodriguez et al., 2006). Efforts have also been directed in recent years to simplify this model (Siegrist et al., 2002; Rodriguez et al., 2008). In order to ease the application of the ADM1 some methodologies have been developed (Zaher et al., 2004; Kleerebezem and Van Loosdrecht, 2006), as well as, some structural simplifications of the model under certain conditions (Bernard et al., 2006). Among the simplified models of the AD process, the one developed by Bernard et al. (2001) has been used in different applications. This model considers two-reactions (acidogenesis and methanogenesis) and has been widely applied for control purposes, for AD process optimization (Dalmau et al., 2010) and for mathematical analysis (Dimitrova and Krastanov 2009; Rincon et al., 2009; Sbarciog et al., 2010). However, only few applications with data from lab- or fullscale plants have been reported (Donoso-Bravo et al., 2009a; Lopez and Borzacconi, 2009). Several reviews of the existing models have been published in the last two decades. Husain (1998) made a brief review of steady state and dynamic models of the kinetics of anaerobic digestion. Later on, Gavala et al. (2003) presented a comprehensive review, describing from the simplest to the most complex models. Following these general reviews, more specific studies appeared in response to the increasing number of available models. For instance, Tomei et al. (2009) focused on models developed for the anaerobic treatment of sewage sludge. Likewise, Batstone (2006) addressed anaerobic digestion modeling in the framework of domestic sewage
systems. Other studies have focused on the type of reactor, instead. For instance, Saravanan and Sreekrishnan (2006) described the different available models for UASB (Up-flow Anaerobic Sludge Blanket), AF (Anaerobic Filter) and EGSB (Expanded Granular Sludge Blanket) reactors, in which the biomass is attached to either a support or forming granules. In Table 1, a summary of different studies where modeling and optimization have been performed is shown.
4.
Experimental information
4.1.
Available measurements
Accurate and reliable measurements of key variables of the process are very important. These data will be used for model identification and validation therefore they should contain the most relevant information at the lowest possible cost of monitoring. Counting on several measured variables will increase the possible parameters that can be reliably estimated; however, and particularly in the case of large and complex models as ADM1, identifying all the parameters and coefficients is not feasible mainly due to the extreme difficulty of separately identifying specific biomass concentrations from the maximum specific uptake rates. Methanogenic applications, however, might require only very accurate identification of a limited number of key parameters to provide good results due to the dynamics of AD in which acidogenesis and acetogenesis are much faster than methanogenesis and hydrolysis (Rodriguez et al., 2006). For any AD model in general, a proper structural identification (discussed in Section 5), for instance using sensitivity analysis methods together with knowledge of the process dynamics, can play a key role in the success of the optimization process to select the most relevant parameters. In general, two types of data can be considered, those that come from off-line or on-line measurements. Off-line sensors are those in which the sample is taken usually manually (low sample frequency) and analyzed by an operator; the data will be available after hours or days. On-line measurements are attached to the process and the analysis is automatic. The analysis is performed under a high frequency, so data produced by these sensors are considered continuous compared to processes that use a time scale. These above-mentioned aspects must be carefully taken into account since the quality of experimental data, in terms of measurement error and sampling frequency, will have a substantial influence on the parameter estimation of the model (Guisasola et al., 2006). To characterize the substrates and intermediates characterization, the organic matter content (i.e. all the organic compounds present in the solution) is usually calculated through the chemical oxygen demand (COD), which is the most common (off-line) measured variable. This measure, and in a lesser extent, along with the total organic carbon (TOC) analysis has been also used for this purpose. In order to recognize the dynamic of specific variables, off-line tests are usually used for volatile fatty acids (VFAs) and for the main macromolecular compounds such as carbohydrates, proteins and lipids. New on-line sensors have been developed which are starting to be used more often (Molina et al., 2009; Boe
Table 1 e Summary and brief description of the studies found in literature about modeling and kinetic parameters identification in AD systems (IC: Initial conditions, KP: Kinetic parameters, YC: Yield coefficients, PCC: physico-chemical constants, CF: Conversion factors). Reference Batch Batstone et al. (2009) Lopez and Borzacconi (2010) Palatsi et al. (2010)
Model
Estimates
Measurements
Estimation method
Uncertainty
ADM1 Model for complex substratesa
2 KP (hydrolysis) 7 KP
Biogasb Methaneb
Gradient search technique Multiple shooting
Confidence region Monte Carlo
ADM1
3 IC, 4 KP
Non-linear weighted square minimization
n.d.
3-reaction model
2 KP, 2 YC
Methane, Acetic, butyric, propionic acidc Biogasb
n.d.
1 IC, 2 KP, 1 YC
Biogasb
Lokshina et al. (2001)
Monod and Non-competitive model Monod and Haldane Equations
Hooke and Jeeves optimization method Least-square Non-linear weighted square minimization
1 ratio (IC/YC), 1 YC, 3 KP
Methanec
Covariance matrix-FIM
Flotats et al. (2003)
ADM1
3 KP, 2 YC
Acetate, propionate, valerate, methanec
Non-linear regression with the Marquardte Levenberg algorithm Combination of random direct search and gradient methods
Continuous Batstone et al. (2009) Bernard et al. (2001)
ADM1 2-reaction model
2 KP (hydrolysis) 4 KP, 6 YC, 1 PCC
Gradient search technique Linearization at different steady states
Confidence region n.m.
Haag et al. (2003)
3-reaction model
26 IC, 8 KP, 4 YC, 6 CF
Batstone et al. (2003)
ADM1
6 KP
Directed search method followed by a gradient-based method Secant method
Covariance matrix-FIM and confidence interval Confidence region
Lopez and Borzacconi (2009) Ghaniyari-Benis et al. (2010) Bhunia and Ghangrekar (2008) Kalfas et al. (2006)
2-reaction model
6 YC
Least-squares criterion
n.d.
1-reaction model
1 KP
Biogas,b VSSc Methane,b Carbon dioxide,c COD, VFA, Z, ICc CODt, CODs, TOC, MiS, TDE, ODE, VFAc VFA, Biogas, pH, methane contentb Biogas,b COD, VFA, gas compositionc CODc
n.m.
Monod, Grau-2nd order and Haldane equations ADM1
3 KP, 1 YC, 1 CF
COD, VSSc
Non-linear regression using least-squares criterion Linearization at different steady states
2 KP, 2 YC (mesothermophilic conditions)
Secant method using unweighted least-square criterion
Confidence region and linear confidence intervals
Koch et al. (2010)
ADM1
5 KP, 1 CF
TSS, VSS, COD, VFA, BIogas, gas composition, pHc Biogas,b gas composition, NH4, NKT, VFA, alkalinity, TSc
Evaluation of the modified Nash-Sutcliffe coefficient
n.d.
Initial rate Donoso-Bravo et al. (2011) Donoso-Bravo et al. (2009b)
1 reaction model (Monod Kinetic) 1st order, Monod, Haldane equations.
2 KP
Methanec
Covariance-FIM
5 KP
Carbohydrates, VSS, VFAc
Non-linear regression using least-squares criterion Non-linear regression using least-squares criterion
Noykova and Gyllenberg (2000) Muller et al. (2002)
Covariance matrix-FIM
n.d.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
n.d.
5351
n.d. not determined, n.m. determined but not mentioned the used method. a Proposed by Angelidaki et al. (1993). b on-line. c off-line.
Monte Carlo
5352
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
et al., 2010; Ward et al., 2011; Jacobi et al., 2011). Specific species concentration of the anaerobic biomass (state variables are unknown variables due to difficulties in performing measurements of the concentration of each population. Molecular biological techniques have been executed for these purposes (Sanchez et al., 1994; Pobeheim et al., 2010)); however, these are costly and usually only qualitative information can be drawn. This issue may trigger some identification problems since some parameters cannot be determine independently (Bernard et al., 2001; Noykova et al., 2002), and may cause some inaccurate model predictions (Batstone et al., 2004b). The volatile suspended solid content (VSS/l), which is normally known, has been used to approximately estimate the total biomass concentration, i.e., the sum of all the population involved (Bernard et al., 2001; Lopez and Borzacconi, 2009). In any case, this measure has not been used for parameter fit but for model validation. A typical model application, in this context, has been the use of software sensors to estimate the concentration of each population (Bernard et al., 2000; Lopez and Borzacconi, 2009). In regards to the product of the reaction, the sum of all gas compounds (formed during the process) is the most common on-line performed measurements and consequently used widely in modeling applications. The biogas measurement has been used as the only measurement for parameters estimation in plenty of articles. Independent measurement of the different gases (CO2, H2 and H2S) is generally done off-line. Depending on the model, the biogas production can be considered as a state variable (Batstone, 2006; Keshtkar et al., 2003) or as a dependent variable (Bernard et al., 2001). The biogas cumulative volume is normally used in the case of using data from batch test and biogas flow rate in the case of continuous system. Biogas flow rate contains more information than cumulative biogas volume and the latest can be derived from the previous, by numerical integration. Instantaneous gas flow rate is typically however more costly and for many applications cumulative volume can also be used to derive the instantaneous flow rate by derivation but the accuracy will be affected by the frequency of sampling points available of accumulated volume. Among other measurements, for instance, the pH is a variable which is easily measured by online sensor. In most of the model is used as an input in the form of an inhibition function (Angelidaki et al., 1999; Batstone, 2006; Haag et al., 2003; Keshtkar et al., 2003), but also can be found as an output of the model to be used for validation (Bernard et al., 2001). So far, no models have used pH for parameter identification; firstly, because it is used as an input and secondly, because pH may present a low sensitivity in well-buffered systems.
4.2.
Experimentation mode
Understanding the nature of the experimental data is a crucial point when making good use of the modeling procedure. This section analyses the different experimental conditions in which anaerobic tests are carried out. Two main issues have to be considered: 1. culture history, and 2. the selected operation mode of the anaerobic digestion process. The culture or inoculum history, involves the specific characteristics of the anaerobic biomass used for the assay,
which in the case of anaerobic digestion systems is normally a mixed culture. The crucial factor is the manner in which the culture has been developed since it determines which species are predominantly present which also influences the physiological state. It is clearly different if the anaerobic inoculum comes from a continuous reactor (where organisms with higher affinity enzymes are favored), if it comes from a batch reactor, or if the inoculum has been exposed to specific conditions for some time (microorganisms have the ability to change their macromolecular composition as a result of physiological adaptation). Usually, the culture history cannot be easily modified and in most cases, the tests are simply carried out with the available inoculum. However, it is of prime importance to document the specific conditions of the inoculum culture since the results are likely to be valid only under these experimental culture conditions. The operation mode of the assay has a paramount influence on the information content of the collected data, and thus, on the quality of the estimated parameters. The variety of operational conditions explain the variability of the reported parameter values (Grady et al., 1996), and implicitly shows that either the studies have considered too limited experimental data information or have not properly applied the parameter estimation procedure (otherwise, more reference parameter sets would have been published and validated). Batch assays are commonly used in AD for kinetic parameter determination, even though other types of operation modes have also been employed for these purposes. This situation is rather unfortunate since it has been demonstrated that the parameters of a simple Monod law cannot be uniquely determined from a batch experiment (Baltes et al., 1994). It is therefore very unlikely that the complex kinetics of AD could be determined from batch tests only, which will become clearer after the next section’s explanation on parameter sensitivity analysis.
4.2.1.
Batch operation
Batch operation can be defined as a biological process in which there is no interchange of mass with the environment, i.e. there are no input or output flow (except for the gas stream). All substrates and nutrients are added at the beginning of the reaction cycle. Several advantages have been stated with regards to the use of batch tests (often known as the biochemical methane potential test, BMP) for kinetic parameter determination in AD, such as: (1) the possibility to easily record the time evolution of several variables (2) the relatively short time span as compared to continuous operations and (3) simplicity and popularity as reflected in a wide acceptance (Noykova et al., 2002; Flotats et al., 2003; Batstone et al., 2009; Lopez and Borzacconi, 2010). However, the main drawback of batch tests, stems from the lack of input excitation (since the only input is the initial condition) resulting in a lack of parameter sensitivity (Lokshina et al., 2001). This can be partly alleviated using different sets of initial conditions (Flotats et al., 2003, 2006) and determining a proper range of substrate/biomass (S/X ) ratio (Grady et al., 1996). The S/X factor, whose inverse is also used in some studies (X/S: inoculumesubstrate ratio or IRS, (Raposo et al., 2009), may influence the correlation between some parameters, for instance, mm and Ks of the Monod-
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
equation, and thus, affect the results of parameter identification. Despite its importance, the value of S/X is seldom mentioned in the studies where batch tests are used for parameter estimation. Whereas BMP tests are widely used for parameter estimation, most full-scale reactors operate in continuous or semi continuous conditions which are drastically different conditions. This explains the proposal to use alternative procedures such as the initial rate reaction measurement, which has been widely applied in enzymatic processes. It is used to obtain either production or consumption rates of a specific compound involved in a reaction on a short time span (Illanes, 2008). Donoso-Bravo et al. (2009b) used the initial rate measurement of the substrate degradation (starch, glucose and VFAs) to evaluate the influence of the temperature on the main reactions of the anaerobic digestion, just as Flotats et al. (2003) estimated kinetic parameters of anaerobic degradation of gelatin using the same technique. According to DonosoBravo et al. (2011) this method can be an interesting alternative to classical batch tests, as it allows to alleviate inhibitory effects of byproducts or substrate limitation (which are more likely to occur in batch than in continuous operation). The main drawback of this technique is the lack of research and validation since only a few studies have used initial rate tests in AD applications.
4.2.2.
Continuous operation
In this operation, spent medium or digestate is continuously replaced with an equal volume of fresh medium (substrate solution) and therefore a continuous discharge of biomass also occurs. Continuous systems also offer a proper platform for kinetic analysis, as long as a series of experiments at different dilution rates (D) are carried out. Overall, it is a more time-consuming method than a batch test and, therefore, the kinetic parameter calculation is usually performed with data from continuous anaerobic reactors which are already operating. For instance, more than 3 months of a pilot up-flow fixed-bed anaerobic reactor operation were required by Bernard et al. (2001) or 1.5 year for full-scale anaerobic digester by Batstone et al. (2009) in order to perform parameter estimation. A less time consuming and appealing alternative is to evaluate the dynamic response of a continuous reactor after specific substrate pulses (Batstone et al., 2003; Kalfas et al., 2006). This method allows the estimation of the kinetic parameters of specific compounds since the pulses provide some decoupling of the biological phenomena, and in turn lower parameter correlation, and thus better identifiability. The pulse amplitude has to be selected so that the substrate concentration crosses the affinity-saturation constant (Batstone et al., 2003). The main drawback of this pulse-based methodology is that it cannot be applied at full-scale and it may be expensive at laboratory scale. New approaches in parameters estimation in this field try to combine data from both continuous and batch experiments (Girault et al., 2011).
4.2.3.
Fed-batch operation
In this process, substrate and nutrients are added continuously or intermittently into the reactor, without an output stream from it, so that the volume of the reaction media increases during the cycle of operation. Fed-batch operations
5353
are scarcely used in AD, neither at full-scale nor at a pilot/ bench/lab-scale. Hence, only a few studies have used these systems for kinetic parameter determination. Rodrigues et al. (2003) evaluated the behavior of a fed-batch reactor by fitting some apparent parameters with a simple global model of AD. Likewise, Redzwan and Banks (2004) used a simplified model in order to assess the kinetic of methane production in a fedbatch reactor. Effective volume of a fed-batch reactor always increases, whereas it remains constant in batch or continuous operation; this may make the mathematical analysis of the system more complicated. In addition, the measurement of the biogas flow has to be corrected to take this volume change into account. These issues may represent significant drawbacks for the use of fed-batch operation in kinetic parameter determination.
5. Structural adequacy and parameter identifiability After the formulation of the modeling objectives and data collection, a double question arises: a) Is the selected model structure able to fit the data? To achieve this objective, the model has to include the necessary degrees of freedom, but not too many as there is a risk of overparametrization. This risk is linked to the other side of the question. b) Once a model structure is selected, is it possible to determine a unique optimal set of parameters based on the experimental data at hand? This double-sided analysis can lead to model simplifications or, on the contrary, to the introduction of additional terms or equations, and in turn to the elimination or the introduction of some parameters. In general, AD models are mechanistic models that synthesize extensive scientific research work dedicated to understand most of the physical, chemical and biological mechanisms of the processes involved. As consequence, most of the model parameters have some physical meaning and generally some default values are available (Batstone et al., 2002). The identifiability problem is then a delicate issue where the modeler should calibrate only those parameters necessary to explain the observed mechanisms without “overfitting” the data, i.e.: an “overcalibrated” model would reproduce the experimental data pretty well but would lose predictive or exploration capability (Reichert, 2010). These structural and parametric identifiability questions are relatively seldom addressed in the reported AD modeling studies. Taylor Series Expansion is the most reported technique. It consists of calculating the successive derivatives of the output function with respect to the unknown model parameters, so as to obtain a system of independent equations in these parameters. Flotats et al. (2003) applied this approach with four of the state variables of ADM1 (Acetate, propionate, valerate and methane) so as to design an experiment for the identification of parameters related to the anaerobic degradation of valerate and the initial biomass concentrations. Noykova et al. (2002), using a more simplified
5354
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
model, evaluated identifiability with respect to three unknown parameters based on the measurement of the biogas production. In this case, the vector of parameters was reduced in order to decrease the computational requirement of this method. In fact, this method is mostly applicable to simple models, whereas it leads to complex systems of equations in more general cases (Muller et al., 2002). Another approach to the study of identifiability is based on local and global sensitivity analysis. The goal of sensitivity analysis is to explore the change in model output resulting from a change in model parameters (kinetic or stoichiometric coefficients, parameters, input conditions, initial conditions, etc.) (Sin et al., 2011). Most of the sensitivity analysis techniques encountered in literature, for AD process, are of local nature and use a differential analysis of outputs with respect to parameters (eq. (1)): vyj yj ðqi Þ yj ðqi þ Dqi Þ ¼ Dqi vqi
(1)
Where yj corresponds to the jth output and qi to the ith parameter. Examples of local sensitivity analysis in AD modeling can be found at Tartakovsky et al. (2008) and Noykova and Gyllenberg (2000). The main drawback of this method is that it is based on the linearization of the model equations at a given set of parameter values and therefore it only describes local model behavior at this point. Other authors attempt to get a more global picture of the sensitivity by varying the parameters (one at a time) and aggregating the relative difference observed in the outputs (Vavilin et al., 2003), either by integrating the errors in time (Bernard et al., 2001), or by estimating a weighted sum of them (Wichern et al., 2009; Lin and Wu, 2011). However, none of these methods are able to detect correlations among the parameters and aggregating the errors can lead to erroneous conclusions when compensations between negative and positive terms occur. An alternative definition of sensitivity analysis is the so called “global sensitivity analysis (GSA)” and relates to uncertainty analysis. It can be viewed as an analysis of variance (ANOVA) problem (Sobol, 2001; Helton and Davis, 2003; Saltelli et al., 2006). Hence the output variance is decomposed into fractions which are attributed to the single model inputs. Examples of such sensitivity analysis methods include Morris Screening (Morris, 1991), the spectral information of measurements to characterize parameter interactions (Tarantola et al., 2006), linear regression of Monte Carlo outputs (Helton and Davis, 2003) and variance decomposition (Saltelli et al., 2008). Global sensitivity analysis has recently been applied to biological models of plant cell cultures (Mailier et al., 2011) and to activated sludge systems (Sin et al., 2011) and it offers promising perspectives in AD models.
6.
Methods for parameters estimation
Unfortunately, AD models are not universal enough and some parameters need to be estimated for each particular case study. Traditionally, they have been calibrated by a trial and error approach. However, this method is very time consuming
and does not provide any information about the uncertainty associated to the parameter values nor any guarantee about its uniqueness. The selection of the objective function (usually also called cost function), which can play a crucial role in the result of the optimization, will be reviewed in the first part of this section and will then follow with the most used techniques for parameter estimation.
6.1.
Cost function selection
In order to find the best-fit of a model to given experimental data, an appropriate criterion for the optimal solution of the model parameter vector must be selected. Several cost functions have been used for parameter identification in AD models mostly in the form of output-error criteria, i.e., measuring the deviation between the model and real system outputs. The type of selected function may influence how the optimization procedure acts and how it adjusts the parameters (Batstone et al., 2003). The most popular cost function is the sum of least squares (OLS, eq. (2)) (Noykova and Gyllenberg, 2000; Bhunia and Ghangrekar, 2008; Batstone et al., 2009; Donoso-Bravo et al., 2010; Lopez and Borzacconi, 2010), where it is implicitly assumed that the standard deviation of the measurement errors, which can be known or unknown, is constant. In eq. (2), J is the objective function, nexp are the collected measurements, nsim are the model-predicted outputs, q represents the parameters to be determined (which can include the stoichiometry, the kinetic parameters and also the unknown initial conditions of some experiments) and N is the number of measurements. Minimizing the cost function has been acknowledged as an important issue for prediction purposes or process stability (Batstone et al., 2003). JðqÞ ¼ min
N X
2 vexp ðtÞ vsim ðt; qÞ
(2)
t¼1
When the measurement errors do not have a constant standard deviation, then it is generally required to introduce weighting factors wt into eq. (2), leading to a weighted leastsquare criterion JðqÞ ¼ min
N X
2 wt vexp ðtÞ vsim ðt; qÞ
(3a)
t¼1
in scalar form, or more generally, when vectors of measurements are considered, JðqÞ ¼ min
N X
vexp ðtÞ vsim ðt; qÞ W vexp ðtÞ vsim ðt; qÞ
(3b)
t¼1
where W is a N N weighting matrix to be selected. If the measurement errors are white and normally distributed, i.e. ε w N(0,Q), then the best choice of the weighting matrix W, in a maximum likelihood sense, is the inverse of the covariance matrix of the measurement noise, i.e. W ¼ Q1. If these assumptions do not hold or a deterministic approach is preferred, some other weighting could be used. A variety of weightings have been used in published studies (Smith et al., 1998; Lokshina et al., 2001; Noykova et al., 2002;
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
Flotats et al., 2003; Palatsi et al., 2010). For instance, the weighting factors have been estimated by calculating a local slope of the output variation (Lokshina et al., 2001), by the difference between maximum and minimum values (Palatsi et al., 2010) or by using the deviation with respect to the mean (Flotats et al., 2003). A frequent situation corresponds to constant, possibly unknown, relative errors. In this case, eq. (3a) can simply be formulated as: JðqÞ ¼ min
N X vexp ðtÞ vsim ðt; qÞ 2 t¼1
vexp ðtÞ
(3c)
or, noting that a constant absolute error on a logarithm, is equivalent to a constant relative error on its argument: JðqÞ ¼ min
N X 2 ln vexp ðtÞ ln ðvsim ðt; qÞÞ
(3d)
t¼1
as it has been used successfully in several studies, e.g. (Batstone et al., 2003; Haag et al., 2003; Vande Wouwer et al., 2006) When a cost function has been formulated, a numerical procedure has to be used to minimize it with respect to the unknown parameters.
6.2.
Optimization techniques
In order to avoid the tedious trial and error approach, several algorithms have been developed. They are search techniques that numerically approach the optimum parameter values by optimizing an objective function. These algorithms can be divided into local algorithms and global algorithms.
6.2.1.
Local methods
The vast majority of optimization methods are local in nature, i.e. they assume the convexity of the cost function. When this condition is not fulfilled, and a local method is applied, there is a high risk that the algorithm will get trapped into a local minimum. To alleviate this problem, it is recommended to start the search from several randomly selected initial parameter values so as to explore the parameter space. This procedure, called multi-start strategy (Kocsis and Gyo¨rgy, 2009), allows the assessment of the problem multimodality and, for instance by drawing a histogram of the frequency of occurrences of the different minima, to determine the global minimum and its basin of attraction (i.e., the ball-sized region containing the initial guesses leading to the global optimum). As initialization is of paramount importance, it is advised to decompose a complex optimization problem into several simpler ones, whenever possible, and to use the solution of these intermediate problems as initial guesses for the next step. This type of procedure has been successfully applied in the identification of bioprocess models by (Hulhoven et al., 2005). Among local methods, one basically distinguishes gradient-based methods, which make use of the first- and, in some cases, the second-order derivatives of the cost function, and gradient-free methods which do not require the cost function differentiability such as the direct-search methods. Another important feature of the optimization problem is the
5355
presence of equal and inequal constraints. These methods are briefly described in the sequel with respect to their application to the estimation of parameters in AD models.
6.2.1.1. Simple unconstrained optimization: steepest descent, GausseNewton and LevenbergeMarquardt methods. The steepest descent is an iterative method which uses first-order information to move downhill in the gradient direction. A large number of iterations are often required to achieve convergence. GausseNewton method is particularly well suited to the minimization of sum of squares (and therefore to the least-squares approach) and avoids the costly evaluation of the Hessian (second-order information) by building an approximation based on the Jacobian while preserving the quadratic convergence of the original method of Newton. The LevenbergeMarquard method (LMA) (Marquardt, 1963) blends the two previous methods, steepest descent and GausseNewton. LMA usually starts using a steepest descent method and progressively becomes a GausseNewton method as it gets closer to the optimum. This way, the algorithm is more robust than GausseNewton but achieves better convergence than steepest descent. LMA has been commonly applied to parameter identification in AD models for the treatment of livestock manure (Garcia-Ochoa et al., 1999), raw industrial wine distillery vinasses (Aceves-Lara et al., 2005; Martin et al., 2002), baker’s yeast effluents (Deveci and Ciftci, 2001), low temperature acetoclastic methanogenesis (Lokshina et al., 2001) and animal wastes from calf farms (Simeonov 1999). Aceves-Lara et al. (2005) combined this algorithm with an asymptotic observer to evaluate the parameters kinetics.
6.2.1.2. Non-linear constrained optimization: sequential quadratic programming. Sequential Quadratic Programming (SQP) is one of the most successful methods for the numerical solution of constrained non-linear optimization problems. SQP (Nocedal and Wright, 2006) is an iterative method which solves, for each iteration, a quadratic problem (QP), i.e., a quadratic approximation of the objective function subject to a linearization of the constraints. If the problem is unconstrained, then the method reduces to the Newton method. For solving the QP problem under inequality constraints, a variety of methods are commonly used, including among others interior point, active set. SQP has been used for parameter estimation in AD models (Sales-Cruz and Gani, 2004; AcevesLara et al., 2005).
6.2.1.3. Multiple shooting. In the previous methods, a sequential optimization approach has always been assumed, i.e., the optimization algorithm repeatedly evaluates the cost function by a call to a time integrator which numerically solves the dynamic equations of the process model (for instance, an LMA algorithm evaluates the cost function through the solution of the model differential equations e which depends on the current values of the model parameters e using a Runge-Kutta method). There is another family of methods which rather use a simultaneous approach, i.e., discretize the model differential equations and uses the resulting set of algebraic equations as constraints to the optimization algorithm. Multiple
5356
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
shooting (Ascher et al., 1995) discretizes the time span into time intervals (ti, tiþ1) and new optimization variables are introduced which correspond to initial conditions of the state variables on each interval. A set of boundary conditions ensures the continuity of the solution. As time integration is performed over short time intervals, the numerical stability property of the algorithm is improved. In addition, state constraints can be easily incorporated. Recently, multiple shooting has been used for parameter estimation in AD (Muller et al., 2002; Lopez and Borzacconi, 2010).
6.2.1.4. Direct-search methods. Direct-search methods (Lewis et al., 2000) are derivative-free methods, which do not even require numerical function values since the relative rank of objective values is sufficient. Several classes of methods exist, such as pattern search methods, simplex methods, and methods with adaptive sets of search directions. These methods date back to the 60’s and have since been replaced by more sophisticated techniques. However, they still have an undisputed popularity due to their simplicity of use and good performance in practical use. In engineering applications, the simplex methods have always been in wide use. The basic idea of simplex search is to construct a nondegenerate simplex in the parameter space and use the simplex to drive the search (a simplex is a set of n þ 1 points in the n dimensional space, e.g. a triangle in 2D; a nondegenerate simplex is one for which any point in the domain of the search can be constructed by taking linear combinations of the edges adjacent to any given vertex). Not only does the simplex provide a frugal design for sampling the space, it has the added feature that if one replaces a vertex by reflecting it through the centroid of the opposite face, then the result is also a simplex. It means that one can proceed parsimoniously, reflecting one vertex at a time, in the search for an optimizer. The simplex algorithm is usually less sensitive to local minima than the gradient-based methods, such as the LevenbergeMarquardt method. However, the convergence is usually slower and closer to the optimum, and the algorithm of course does not provide any sensitivity information (Jacobian) that could be used as a byproduct to estimate the Fisher Information Matrix; as will be introduced in the sequel. The simplex algorithm has been widely applied to parameter estimation in AD models (Mosche and Jordening, 1999; Simeonov 1999; Ruel et al., 2002; Haag et al., 2003; Guisasola et al., 2009; Lopez and Borzacconi, 2010)
6.2.2.
Global methods
Non-linear parameter identification problems are often characterized by the presence of various local minima. Global optimization is aimed at finding the best solution to these kinds of problems. This is a very active research area and two main families of methods have emerged: deterministic algorithms, including for instance grid search and branch and bound, and stochastic algorithms including for instance simulated annealing, tabu search, genetic algorithms, differential evolution, ant colony optimization and particle swarm optimization. Among all these methods, Simulated Annealing (SA), Genetic Algorithms (GA), Particle Swarm Optimization (PSO)
have been quite popular in engineering applications. SA is a probabilistic algorithm, whose initial idea comes from annealing in metallurgy, a technique involving heating and controlled cooling of a material to increase the size of its crystals and reduce their defects. The objective function to be minimized is analogous to the internal energy of the system. GA belong to the larger class of evolutionary algorithms (EA), which generate solutions to optimization problems using techniques inspired by natural evolution, such as inheritance, mutation, selection, and crossover. PSO are search algorithms based on the simulation of the animal social behavior in a group. The advantages of these algorithms is that they do not require the objective function to be differentiable as in classic local (gradient-based) optimization algorithms, which make few assumptions about the problem to be solved, and can explore a large space of candidate solutions. Although, these algorithms produce better solutions they do not guarantee that an optimal solution will ever be found at the price of large amounts of computation. These algorithms have found applications in the identification of AD models, for instance Simulated Annealing (Haag et al., 2003), Genetic Algorithms (Jeong et al., 2005; Abu Qdais et al., 2010; Wichern et al., 2009), and Particle Swarm Optimization (Wolf et al., 2008).
6.3.
What about optimization constraints?
Constraints in the estimated parameter values are usually employed in the optimization process as long as they are allowed by the selected technique. Simple and logic constraints are the most used ones, such as: positive values and within certain reasonable ranges (i.e., parameters between some minimum and maximum values that the experimenter could determine from past experience). Nevertheless, sometimes linear and non-linear constraints can be useful during parameters estimation since they enable the inclusion of conservations laws (i.e. yields) and avoid some mathematical uncertainties linked to the model’s structure (i.e. pH and gas transfers). In other cases, it is practical to estimate separately some parameters, e.g. volumetric coefficient of mass transfer (kLa), with iterative linear approximations in order to simplify the estimation task (Batstone, 1999). On the other hand, when using sophisticated non-linear constrained algorithms is not necessary, and simple methods such as the simplex can be used, some transformation of the parameter space can be done. For instance, positivity constraints can be imposed using a logarithmic transformation, i.e. if q ¼ ln (l) is a positive unknown parameter, then the optimization algorithm explore the full real parameter space (positive and negative) for l. This technique has been applied for instance in Vande Wouwer et al. (2006).
6.4.
Alternative methods: The Bayesian inference
The Bayesian approach opens a new calibration framework because it abandons the idea of believing that the model parameters are fixed and not known, and treats them as probabilistic (or random) variables having a probability density function. This function is called joint posterior
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
distribution of parameters and defines subjective beliefs of parameter values by summarizing the state of knowledge about the system performance (Omlin and Reichert, 1999). Equation (4) shows a very general expression of the Bayes formula: pðqjyÞ ¼
pðyjqÞ pðqÞ pðyjqÞ pðqÞ fpðyjqÞ pðqÞ ¼Z pðyÞ pðyjqÞ pðqÞdq
(4)
q
where q is the model’s parameters vector and y represents experimental data. The Bayes theorem states that the posterior probability function, p(qjy), is proportional to the multiplication of the prior beliefs, p(q), and the likelihood function of observations, p( yjq). The prior probability distribution, p(q), expresses modeler prior beliefs about possible parameter values while the posterior, p(qjy), expresses the posterior beliefs after having evaluated the model residuals. The likelihood function p( yjq) plays a very important role in Bayes’ formula because it is the function through which the data y modifies prior knowledge of q. Finally, the probability of the observations, p( y), is the expected value of the likelihood function over the parameter space, and acts as a normalizing constant. Qian et al. (2003) explained that the most important limitation of using Bayesian methods for scientific inference was that analytical solutions of the posterior distributions are available for fairly limited combinations of model forms and probability distributions (such as the linear model leading to a normal distribution of the residuals). For most non-linear models, or models with a large number of parameters to be calibrated, the estimation of the posterior likelihood becomes intractable. Fortunately, advent of fast and inexpensive computing has promoted the implementation of numerical techniques. Particularly important are the Markov Chain Monte Carlo (MCMC) techniques that construct a Markov chain which asymptotically converges into the posterior distribution. The Bayesian techniques are especially advisable in the case of poor parameter identifiability because subjective prior knowledge about possible parameter values can be used. Omlin and Reichert (1999) stress that in environmental modeling the use of complex model structures and limited experimental data makes the Bayesian techniques very important. An interesting application of Bayesian calibration of an AD model has been recently proposed by Martin et al. (2011). In this example, a digester model (De Gracia et al., 2009) able to work under anaerobic and aerobic conditions is calibrated by using a complete so-called Integrated Monte Carlo Methodology (Martin and Ayesa, 2010).
6.5. Some considerations for parameter estimation in AD models The complexity and particularities of AD processes give rise to some considerations when trying to calibrate a model.
6.5.1.
Steady states analysis
When the selected model is simple and some convenient assumptions are considered (mainly that the system has
5357
reached a steady state condition) some mathematical modifications can be done in order to get rid of the ODE and to express the outputs of the systems as a functions of other variables. Once these expressions are obtained, simple linear regression may be used to estimate the model parameters. Bernard et al. (2001) used this approach to draw some expressions in order to estimate several kinetic parameters, stoichiometric coefficients and one mass transfer constant of a 2-reaction model in the AD of wine distillery wastewater. Likewise, Simeonov et al. (1996) employed steady state analysis to estimate some kinetic parameters of a 3-reaction model in the AD of different type of animal waste at labscale. Both studies evaluated the models with 7 steady states varying either the dilution rate or the organic matter concentration in the influent. The main drawback of this method is that reaching steady state conditions requires a highly controlled reactor, which is quite difficult to achieve in a full-scale reactor because of the flow and concentration disturbances. In this context, Bhunia and Ghangrekar (2008) assessed the application of the linearized form of the non-linear expression drawn from the steady states analysis with three simple models (Monod, Haldane and second-order functions) in the treatment of synthetic sucrose-based wastewater using labscale UASB reactors. Overall, non-linear optimization showed better performance than linearization.
6.5.2.
Mass continuity (conservation laws)
The problem of mass continuity in the context of AD has been analyzed (Banks et al., 2011; Ekama et al., 2007; De Gracia et al., 2006; Huete et al., 2006). Tracking the mass trajectories of the elemental compounds (C, H, O, N and P) can uncover hidden processes when studying experimental data or identify inadequate model structures when analyzing model results. Concerning the latter objective, De Gracia et al. (2006) proposed a mass and charge conservation check methodology to verify the consistency of AD models. On the other hand, the equations of the mass continuity can represent by themselves biochemical transformation models. This is the case of Zaher et al. (2009) where a simple AD model is presented to study the microbial activity in the treatment of dairy manure. The study of the energy balance is also crucial when analyzing the process’s performance. Banks et al. (2011) presented a complete energy study to assess the performance of the anaerobic digestion of source-segregated domestic food waste. In the modeling field, De Gracia et al. (2009) incorporated the energy balance equation to model the digestion of sludge generated in wastewater treatment plants. Lu¨bken et al. (2007) proposed a modified version of ADM1 model to simulate energy production in the digestion of cattle manure and renewable energy crops. A thermodynamic analysis of the acidogenic reaction was performed by Bastidas-Oyanedel et al. (2008), where was demonstrated that the energy transfer efficiency is influenced by some operational conditions, such as pH and hydraulic retention time. The mass continuity equations have also been used to characterize input conditions in terms of substrate characteristics (Huete et al., 2006). In the case of sludge produced by wastewater treatment plants, and when trying to characterize it in terms of the ADM1 model, Huete et al. (2006) points out
5358
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
that “the biggest uncertainty comes from the elemental composition of the organic composites and the inert soluble and particulate components, since the elemental mass fractions depend on the specific case under study”. When approaching the same problem, Ekama et al. (2007) found that the non-biodegradable particulate organics, including both (a) those originated in influent wastewater, and (b) those generated in the activated sludge endogenous process, are also nonbiodegradable under anaerobic conditions. Grau et al. (2007a) proposed a new model building philosophy called plant wide modeling which indirectly addresses the characterization problem of sludge. It proposes to generate a list of non-redundant model components for the biochemical processes and to define them in terms of C, H, N, O, P, charge and other possible elements. This extensive description allows for a straightforward relationship between the model components and the most common analytical measurements carried out in the wastewater sludge. Using this approach De Gracia et al. (2011) has presented a new tool to characterize the wastewater sludge in terms of the model components. This estimation is carried out by minimising a cost function (Grau et al., 2007b) and taking into account the physical and chemical restrictions due to the components’ elemental composition.
6.5.3.
Initial conditions
Establishing the initial conditions of the state variables of the selected model, which in many articles is merely omitted, is one the first issues that have to be defined before the optimization process itself. As previously mentioned, several experimental methods for determining substrate characterization may be used, thus knowing the initial conditions for the different organic compounds considered in the model should not de considered a big issue. However, setting the initial concentration of all the microbial populations of the model, especially in large model such as ADM1, is quite complicated considering the difficulty of its experimental estimation. On the other hand, the type of operation mode has an influence in how accurate the initial conditions has to be established. In batch tests, the initial conditions are the sole inputs of the system, hence the initial values of the state variables will exert great influence in the model behavior. By contrast, in continuous systems the initial condition had a negligible effect especially in the case of long-term operation evaluations, as long as an initial simulation period is not taken into account, allowing convergence (Batstone et al., 2003). In other cases, initial values of the microbial population have been either arbitrarily fixed (Knobel and Lewis, 2002; Noykova et al., 2002; Nopharatana et al., 2003; Ozkan-Yucel and Gokcay, 2010) or estimated through a prior simulation evaluation (Batstone et al., 2004b). A typical strategy to achieve this is to run, for a very long time, a steady state simulation of a similar system to that from where the sludge comes from and use the biomass relative composition from the results of such simulation and the total biomass from an available measurement in the real system to be simulated. In some cases, the initial concentration of each population has been considered as an unknown value, thus it has been estimated in the optimization process by using adequate experimental information
(Haag et al., 2003; Flotats et al., 2006; Palatsi et al., 2010); however, some important identification problems have been encountered in other studies (Flotats et al., 2003; Jeong et al., 2005). Regardless, estimating the initial conditions of the biomass composition is the most recommended choice, especially in the case of batch test, since besides being valuable information, it would not be appropriate to force the bestfit by fixing these initial values.
7.
Parameter uncertainty estimation
If the model has passed direct validation, and whenever possible cross validation (this later step can be made difficult by the scarcity of experimental data), it is interesting to analyze further the accuracy of the model parameters, and to provide confidence intervals for the parameters and in turn, for the model prediction.
7.1.
Error covariance matrix
The Cramer-Rao bound (Walter and Pronzato, 1997), which corresponds to the inverse of the Fisher Information Matrix (FIM), provides an optimistic estimate of the parameter error covariance matrix, i.e., information on the parameter standard deviation and correlation. The FIM can be computed using the output parameter sensitivity matrix and the inverse of the covariance matrix of measurement noise. In AD systems, FIM has been evaluated, for instance in Noykova et al. (2002), Flotats et al. (2003) and Haag et al. (2003). In the case of simple models, the parametric sensitivity can be calculated analytically as in Lokshina et al. (2001). For more complex models, a numerical procedure is preferred, e.g. finite differences. When using a gradient-based method, e.g. LevenbergeMarquardt, it is usually possible to extract the sensitivities at the optimum (as a byproduct of the algorithm computation). It is important to keep in mind that the inverse of FIM just provides a (maybe too) optimistic estimate of the parameter error covariance matrix, as has been pointed out in the case of large and non-linear biochemical systems by Schenkendorf et al. (2009), Joshi et al. (2006) and Lopez and Borzacconi (2010).
7.2.
Confidence intervals
The covariance matrix of the parameter errors, as evaluated in the previous subsection, together with a model linearization, allows the computation of confidence intervals for the model prediction. These intervals are of course only approximations, whose quality depends on the model non-linearity in the parameters, but they provide a first view of the model’s uncertainty. More accurate, but also more computationally demanding, approaches to the estimation of confidence intervals have been proposed (Dochain and Vanrolleghem, 2001). Confidence regions have been obtained by Batstone et al. (2003), Batstone et al. (2004b), Kalfas et al. (2006) and Batstone et al. (2009), in the case of the estimation of kinetic parameters or stoichiometric coefficients of ADM1 for the description of the AD of several substrates.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
7.3.
Joint posterior distribution
In the Bayesian approach the model parameters are defined as random variables, and therefore, the uncertainty of the parameters is defined by the joint posterior distribution. In this case, the posterior distribution of parameters is generally derived by a Markov Chain Monte Carlo technique, i.e., using a numerical implementation. This is a great advantage in the case of dynamic non-linear models because: first, the calibration method is independent of the model structure; and second, the uncertainty of the parameters is estimated by the distribution of their most feasible values, without making any assumptions about the model structure.
8.
Model validation
Once a set of parameters has been obtained, it is necessary to question the predictive quality of the resulting model and to assess the parameter accuracy. This will determine the confidence behind the model, and tell the modeler if he needs to revise the model’s identification. The overall procedure, called model validation, consists of several steps.
8.1.
Direct validation
The first test is to check whether the model is able to reproduce the experimental data that has been used for parameter identification. Otherwise, there is obviously something wrong in the identification procedure (see details in Section 2) and it has to be adjusted and repeated. There are different ways of checking the model’s adequacy. One of the best, even if it cannot be cast into mathematical formulas, is the visual inspection: the model has to follow well the data evolution while smoothing off the noise (a model that tends to reproduce noise is overparametrized and will fail later on in cross-validation tests). Model performance by visual inspection is widely used in AD system, and has even been the only applied method in many cases. On a more mathematical basis, a good test is based on residuals analysis. If the model is predicting the data well, the residual can be directly related to the measurement error. Different types of information may be drawn from the residuals, such as the determination coefficient (R2), an estimation of the variance of the data, analysis of randomness, etc. Actually, the determination coefficient has been the sole tool used to evaluate the model fit in several studies (Redzwan and Banks, 2004; Flotats et al., 2006; Palatsi et al., 2010). This parameter was also used by Flotats et al. (2003) to evaluate the model fit quality as part of a more detailed model analysis. Other statistical tests have been used to compare several models, such as the Fisher test (Aceves-Lara et al., 2005), the sum of normalized errors (Bhunia and Ghangrekar, 2008) or the determination coefficient (Donoso-Bravo et al., 2010). Nevertheless, few studies have estimated the variance of the experimental data through the residuals (Haag et al., 2003). Analysis of randomness in the residual (another method to evaluate the fit quality) is seldom performed, even though it
5359
may represent a good criterion to determine autocorrelation or the need to change the formulation of the cost function. A particular case is the study developed by Barampouti et al. (2005) who used this type of statistical analysis to generate an empirical model to predict the biogas production and to evaluate goodness-of-fit with the simulated data.
8.2.
Cross validation
Direct validation is a necessary condition, but by no means a sufficient condition to accept a model as being one that can reproduce the behavior of the system under consideration. It may well be that the model fits the data that has been used for identification adequately, but performs poorly with new unseen data. To this end, enough data must be available and divided these into two subsets, one for parameter identification (and afterward direct validation), and the other for cross validation. Notice that cross validation can in turn imply the identification of the initial conditions of the experiments under consideration (i.e., the initial conditions of the unseen experiments are sometimes unknown, or at least, know but no well, and have to be estimated before checking that the model with the previously identified parameters fits well the new data). This procedure has been applied to check the AD model validity, especially when complex models, such as ADM1, are considered (Ozkan-Yucel and Gokcay, 2010; Fezzani and Ben Cheikh, 2009; Fezzani and Ben Cheikh, 2008; Tartakovsky et al., 2008; Siegrist et al., 2002; Lu¨bken et al., 2007). In the same context, short calibration steps may also be performed regularly during the validation of the model (Batstone et al., 2009; Bernard et al., 2001), in order to take the possible variations of the substrate characteristics as well as the changes in the anaerobic population into account especially when long-term operation data are used.
9.
Conclusion
Anaerobic digestion is a very complex process involving various bacterial populations and substrates. With the progresses in instrumentation and in computer science, the development of mathematical models, predicting the dynamic process behavior has attracted considerable attention in the last two decades. ADM1 is undoubtedly one of the milestones of this research era. However, modeling is always a goal-driven exercise, and many alternative models have been proposed in the literature, depending on the aim, e.g., process understanding, dynamic simulation, optimization, or control. Models contain unknown parameters, e.g., initial conditions, stoichiometry, and kinetic parameters which have to be estimated from experimental data. Parameter identification is a delicate task due the potentially large number of parameters and the scarcity of informative experimental data. This review attempts to summarize the efforts that have been accomplished in the parameter estimation of models of anaerobic digestion processes and highlights the critical steps of the identification procedure. In general, the literature shows a lack of systematic and clear procedure for modeling AD processes. Also, sets of parameter values are reported without
5360
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
thorough analysis of the model’s validity and parameter accuracy, which makes it difficult to exploit all of the published information. This situation will certainly improve with more awareness of these important issues. As a general recommendation, the development of benchmarks, and the availability of data bases, as open resources on the internet, would certainly speed up these developments and consolidate knowledge in the field.
Acknowledgments This paper presents research results of the Belgian Network DYSCO (Dynamical Systems, Control, and Optimization), funded by the Interuniversity Attraction Poles Programme, initiated by the Belgian State, Science Policy Office. The scientific responsibility rests with its author(s). This study is also supported by a grant from Belspo (Belgian Science Policy) through its Postdoc fellowships to non-EU researchers program. Johan Mailier is a research fellow supported by the FNRS (Belgian National Fund for Scientific Research).
references
Abu Qdais, H., Hani, K.B., Shatnawi, N., 2010. Modeling and optimization of biogas production from a waste digester using artificial neural network and genetic algorithm. Resources Conservation and Recycling 54 (6), 359e363. Aceves-Lara, C.A., Aguilar-Garnica, E., Alcaraz-Gonzalez, V., Gonzalez-Reynoso, O., Steyer, J.P., Dominguez-Beltran, J.L., Gonzalez-Alvarez, V., 2005. Kinetic parameters estimation in an anaerobic digestion process using successive quadratic programming. Water Science and Technology 52 (1e2), 419e426. Angelidaki, I., Ellegaard, L., Ahring, B.K., 1993. A mathematical model for dynamic simulation of anaerobic digestion of complex substrates: focusing on ammonia inhibition. Biotechnology and Bioengineering 42 (2), 159e166. Angelidaki, I., Ellegaard, L., Ahring, B.K., 1999. A comprehensive model of anaerobic bioconversion of complex substrates to biogas. Biotechnology and Bioengineering 63 (3), 363e372. Appels, L., Baeyens, J., Degreve, J., Dewil, R., 2008. Principles and potential of the anaerobic digestion of waste-activated sludge. Progress in Energy and Combustion Science 34 (6), 755e781. Ascher, U.M., Mattheij, R.M.M., Russell, R.D., 1995. Numerical Solution of Boundary Value Problems for Ordinary Differential Equations. Classics in Applied Mathematics. SIAM, Philadelphia, PA. Baltes, M., Schneider, R., Sturm, C., Reuss, M., 1994. Optimal experimental design for parameter estimation in unstructured growth models. Biotechnology Progress 10 (5), 480e488. Banks, C.J., Chesshire, M., Heaven, S., Arnold, R., 2011. Anaerobic digestion of source-segregated domestic food waste: performance assessment by mass and energy balance. Bioresource Technology 102 (2), 612e620. Barampouti, E.M.P., Mai, S.T., Vlyssides, A.G., 2005. Dynamic modeling of biogas production in an UASB reactor for potato processing wastewater treatment. Chemical Engineering Journal 106 (1), 53e58. Bastidas-Oyanedel, J.R., Aceves-Lara, C.A., Ruiz-Filippi, G., Steyer, J.P., 2008. Thermodynamic analysis of energy transfer
in acidogenic cultures. Engineering in Life Sciences 8 (5), 487e498. Batstone, D., 1999. High Rate Anaerobic Treatment of Complex Wastewater. PhD thesis, Department of Chemical Engineering, University of Queensland. Batstone, D., Keller, J., Newell, R., Newland, M., 2000. Modelling anaerobic degradation of complex wastewater. I: model development. Bioresource Technology 75, 67e74. Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A., Sanders, W.T., Siegrist, H., Vavilin, V.A., 2002. Anaerobic Digestion Model No. 1. (ADM1). Iwa Task Group, IWA Task Group for Mathematical Modelling of Anaerobic Digestion Processes, IWA Publishing, London. Batstone, D.J., Pind, P.F., Angelidaki, I., 2003. Kinetics of thermophilic, anaerobic oxidation of straight and branched chain butyrate and valerate. Biotechnology and Bioengineering 84 (2), 195e204. Batstone, D.J., Keller, J., 2003. Industrial applications of the IWA anaerobic digestion model No. 1 (ADM1). Water Science and Technology 47 (12), 199e206. Batstone, D., Keller, J., Bonnet, B., 2004a. The influence of substrate kinetics on the microbial community structure in granular anaerobic biomass. Water Research 38 (6), 1390e1404. Batstone, D.J., Torrijos, M., Ruiz, C., Schmidt, J.E., 2004b. Use of an anaerobic sequencing batch reactor for parameter estimation in modelling of anaerobic digestion. Water Science and Technology 50 (10), 295e303. Batstone, D., Hernandez, J.L.A., Schmidt, J.E., 2005. Hydraulics of laboratory and full-scale upflow anaerobic sludge blanket (UASB) reactors. Biotechnology and Bioengineering 91 (3), 387e391. Batstone, D.J., 2006. Mathematical modelling of anaerobic reactors treating domestic wastewater: rational criteria for model use. Re-views in Environmental Science and Biotechnology 5 (1), 57e71. Batstone, D.J., Tait, S., Starrenburg, D., 2009. Estimation of hydrolysis parameters in full-scale anerobic digesters. Biotechnology and Bioengineering 102 (5), 1513e1520. Bernard, O., Zakaria, H., Dochain, D., 2000. Software sensors to monitor the dynamics of microbial communities: application to anaerobic digestion. Acta Biotheoretica 48, 197e205. Bernard, O., Hadj-Sadok, Z., Dochain, D., Genovesi, A., Steyer, J.P., 2001. Dynamical model development and parameter identification for an anaerobic wastewater treatment process. Biotechnology and Bioengineering 75 (4), 424e438. Bernard, O., Chachuat, B., Helias, A., Rodriguez-, J., 2006. Can we assess the model complexity for a bioprocess: theory and example of the anaerobic digestion process. Water Science and Technology 53 (1), 85e92. Bhunia, P., Ghangrekar, M.M., 2008. Analysis, evaluation, and optimization of kinetic parameters for performance appraisal and design of UASB reactors. Bioresource Technology 99 (7), 2132e2140. Boe, K., Batstone, D.J., Steyer, J.P., Angelidaki, I., 2010. State indicators for monitoring the anaerobic digestion process. Water Research 44 (20), 5973e5980. Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: a review. Bioresource Technology 99 (10), 4044e4064. Costello, D.J., Greenfield, P.F., Lee, P.L., 1991. Dynamic modeling of A single-Stage high-rate anaerobic reactor .1. Model derivation. Water Research 25 (7), 847e858. Dalmau, J., Comas, J., Rodriguez-Roda, I., Pagilla, K., Steyer, J.P., 2010. Model development and simulation for predicting risk of foaming in anaerobic digestion systems. Bioresource Technology 101 (12), 4306e4314. de Baere, L. A., Mattheeuws, B., Velghe, F., 2010. State of the art of anaerobic digestion in Europe. 12th World Congress on
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
Anaerobic Digestion (AD12), Guadalajara, Mexico, October 31st to November 4th, 2010. De Gracia, M., Sancho, L., Garcı´a-Heras, J.L., Vanrolleghem, P., Ayesa, E., 2006. Mass and charge conservation check in dynamic models: application to the new ADM1 model. Water Science and Technology 53 (1), 225e240. De Gracia, M., Grau, P., Huete, E., Go´mez, J., Garcı´a-Heras, J.L., Ayesa, E., 2009. New generic mathematical model for WWTP sludge digesters operating under aerobic and anaerobic conditions: model building and experimental verification. Water Research 43 (18), 4626e4642. De Gracia, M., Huete, E., Beltra´n, S., Grau, P., Ayesa, E., 2011. Automatic characterisation of primary, secondary and mixed sludge inflow in terms of the mathematical generalised sludge digester model. Water Science and Technology 64 (3), 557e567. Derbal, K., Bencheikh-Lehocine, M., Cecchi, F., Meniai, A.-H., Pavan, P., 2009. Application of the IWA ADM1 model to simulate anaerobic co-digestion of organic waste with waste activated sludge in mesophilic condition. Bioresource Technology 100, 1539e1543. Deveci, N., Ciftci, G., 2001. A mathematical model for the anaerobic treatment of baker’s yeast effluents. Waste Management 21 (1), 99e103. Dimitrova, N., Krastanov, M., 2009. Nonlinear stabilizing control of an uncertain bioprocess model. International Journal of Applied Mathematics and Computer Science 19 (3), 441e454. Dochain, D., Vanrolleghem, P., 2001. Dynamical Modelling & Estimation in Wastewater Treatment Processes. IWA Publishing, London. Donoso-Bravo, A., Carballa, M., Ruiz-Filippi, G., Chamy, R., 2009a. Treatment of low strength sewage with high suspended organic matter content in an anaerobic sequencing batch reactor and modelling application. Electronic Journal of Biotechnology 12 (3), 13e14. Donoso-Bravo, A., Retamal, C., Carballa, M., Ruiz-Filippi, G., Chamy, R., 2009b. Influence of temperature on hydrolysis, acidogenesis and methanogenesis in anaerobic digestion: parameter identification and modeling application. Water Science and Technology 60 (1), 9e17. Donoso-Bravo, A., Perez-Elvira, S., Fdz-Polanco, F., 2010. Application of simplified models for anaerobic biodegradability tests. Evaluation of pre-treatment processes. Chemical Engineering Journal 160, 607e614. Donoso-Bravo, A., Garcia, G., Perez-Elvira, S.I., Fdz-Polanco, F., 2011. Initial rate technique as a procedure to predict the anaerobic digester operation. Biochemical Engineering Journal 53, 275e280. Eastman, J.A., Ferguson, J.F., 1981. Solubilization of particulate organic-carbon during the acid phase of anaerobic digestion. Journal of Water Pollution Control Federation 53 (3), 352e366. Ekama, G.A., So¨temann, S.W., Wentzel, M.C., 2007. Biodegradability of activated sludge organics under anaerobic conditions. Water Research 41 (1), 244e252. Fedorovich, V., Lens, P., Kalyuzhnyi, S., 2003. Extension of anaerobic digestion model No. 1 with processes of sulfate reduction. Applied Biochemistry and Biotechnology 109 (1e3), 33e45. Fezzani, B., Ben Cheikh, R., 2008. Implementation of IWA anaerobic digestion model No. 1 (ADM1) for simulating the thermophilic anaerobic co-digestion of olive mill wastewater with olive mill solid waste in a semi-continuous tubular digester. Chemical Engineering Journal 141 (1e3), 75e88. Fezzani, B., Ben Cheikh, R., 2009. Extension of the anaerobic digestion model No. 1 (ADM1) to include phenolic compounds biodegradation processes for the simulation of anaerobic co-digestion of olive mill wastes at thermophilic temperature. Journal of Hazardous Materials 162 (2e3), 1563e1570.
5361
Flotats, X., Ahring, B.K., Angelidaki, I., 2003. Parameter identification of thermophilic anaerobic degradation of valerate. Applied Biochemistry and Biotechnology 109 (1e3), 47e62. Flotats, X., Palatsi, J., Ahring, B.K., Angelidaki, I., 2006. Identifiability study of the proteins degradation model, based on ADM1, using simultaneous batch experiments. Water Science and Technology 54 (4), 31e39. Gali, A., Benabdallah, T., Astals, S., Mata-Alvarez, J., 2009. Modified version of ADM1 model for agro-waste application. Bioresource Technology 100, 2783e2790. Garcia-Ochoa, F., Santos, V.E., Naval, L., Guardiola, E., Lopez, B., 1999. Kinetic model for anaerobic digestion of livestock manure. Enzyme and Microbial Technology 25 (1e2), 55e60. Gavala, H.N., Angelidaki, I., Ahring, B.K., 2003. Kinetics and modeling of anaerobic digestion process. Advances in Biochemical Engineering/biotechnology 81, 57e93. Ghaniyari-Benis, S., Martin, A., Borja, R., 2010. Kinetic modelling and performance prediction of a hybrid anaerobic baffled reactor treating synthetic wastewater at mesophilic temperature. Process Biochemistry 45 (10), 1616e1623. Girault, R., Rousseau, P., Steyer, J., Bernet, N., Beline, F., 2011. Combination of batch experiments with continuous reactor data for ADM1 calibration: application to anaerobic digestion of pig slurry. Water Science and Technology 63 (11), 2575e2582. Grady, C.P.L., Smets, B.F., Barbeau, D.S., 1996. Variability in kinetic parameter estimates: a review of possible causes and a proposed terminology. Water Research 30 (3), 742e748. Grau, P., de Gracia, M., Vanrolleghem, P.A., Ayesa, E., 2007a. A new plant-wide modelling methodology for WWTPs. Water Research 41 (19), 4357e4372. Grau, P., Beltra´n, S., de Gracia, M., Ayesa, E., 2007b. New mathematical procedure for the automatic estimation of influent characteristics in WWTPs. Water Science and Technology 56 (8), 95e106. Guisasola, A., Baeza, J.A., Carrera, J., Sin, G., Vanrolleghem, P.A., Lafuente, J., 2006. The influence of experimental data quality and quantity on parameter estimation accuracy. Education for Chemical Engineers 1, 139e145. Guisasola, A., Sharma, K.R., Keller, J., Yuan, Z.Q., 2009. Development of a model for assessing methane formation in rising main sewers. Water Research 43 (11), 2874e2884. Haag, J.E., Vande Wouwer, A., Queinnec, I., 2003. Macroscopic modelling and identification of an anaerobic waste treatment process. Chemical Engineering Science 58, 4307e4316. Helton, J.C., Davis, F.J., 2003. Latin hypercube sampling and the propagation of uncertainty in analyses of complex systems. Reliability Engineering and System Safety 81, 23e69. Hill, D.T., Barth, C., 1977. A dynamical model for simulation of animal waste digestion. Journal of Water Pollution Control Association 10, 2129e2143. Hill, D.T., 1982. A comprehensive dynamic model for animal waste methanogenesis. Transactions of the ASAE 25 (5), 1374e1380. Huete, E., de Gracia, M., Ayesa, E., Garcia-Heras, J.L., 2006. ADM1based methodology for the characterisation of the influent sludge in anaerobic reactors. Water Science and Technology 54 (4), 157e166. Hulhoven, X., Vande Wouwer, A., Bogaerts, P., 2005. On a systematic procedure for the predetermination of macroscopic reaction schemes. Journal of Bioprocess and Biosystems Engineering 27, 283e291. Husain, A., 1998. Mathematical models of the kinetics of anaerobic digestion e a selected review. Biomass and Bioenergy 14 (5e6), 561e571. Illanes, A., 2008. Enzyme Biocataysis. Principles and Applications. Springer Science, Valparaiso.
5362
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
Jacobi, H.F., Moschner, C.R., Hartung, E., 2011. Use of near infrared spectroscopy in online-monitoring of feeding substrate quality in anaerobic digestion. Bioresource Technology 102 (7), 4688e4696. Jakeman, A.J., Letcher, R.A., Norton, J.P., 2006. Ten iterative steps in development and evaluation of environmental models. Environmental Modelling and Software 21 (5), 602e614. Jeong, H.S., Suh, C.W., Lim, J.L., Lee, S.H., Shin, H.S., 2005. Analysis and application of ADM1 for anaerobic methane production. Bioprocess and Biosystems Engineering 27 (2), 81e89. Joshi, M., Seidel-Morgenstern, A., Kremling, A., 2006. Exploiting the bootstrap method for quantifying parameter confidence intervals in dynamical systems. Metabolic Engineering 8 (5), 447e455. Kalfas, H., Skiadas, I.V., Gavala, H.N., Stamatelatou, K., Lyberatos, G., 2006. Application of ADM1 for the simulation of anaerobic digestion of olive pulp under mesophilic and thermophilic conditions. Water Science and Technology 54 (4), 149e156. Kalyuzhnyi, S.V., 1997. Batch anaerobic digestion of glucose and its mathematical modeling .2. Description, verification and application of model. Bioresource Technology 59 (2e3), 249e258. Kalyuzhnyi, S.V., Davlyatshina, M.A., 1997. Batch anaerobic digestion of glucose and its mathematical modeling .1. Kinetic investigations. Bioresource Technology 59 (1), 73e80. Kalyuzhnyi, S.V., Fedorovich, V.V., 1998. Mathematical modelling of competition between sulphate reduction and methanogenesis in anaerobic reactors. Bioresource Technology 65 (3), 227e242. Keshtkar, A., Meyssami, B., Abolhamd, G., Ghaforian, H., Khalagi Asadi, M., 2003. Mathematical modeling of non-ideal mixing continuous flow reactors for anaerobic digestion of cattle manure. Bioresource Technology 87 (1), 113e124. Kleerebezem, R., Van Loosdrecht, M.C.M., 2004. Criticizing some concepts of ADM1. 10th IWA World Congress Anaerobic Digestion, Montreal, Canada. Kleerebezem, R., Van Loosdrecht, M.C.M., 2006. Waste characterization for implementabon in ADM1. Water Science and Technology 54 (4), 167e174. Knobel, A.N., Lewis, A.E., 2002. A mathematical model of a high sulphate wastewater anaerobic treatment system. Water Research 36 (1), 257e265. Koch, K., Lubken, M., Gehring, T., Wichern, M., Horn, H., 2010. Biogas from grass silageemeasurements and modeling with ADM1. Bioresource Technology 101 (21), 8158e8165. Kocsis, L., Gyo¨rgy, A., 2009. Efficient Multi-start Strategies for Local Search Algorithms. Lecture Notes in Computer Science, 5781/2009 doi:10.1007/978-3-642-04180-8_63705e720. Lee, M.Y., Suh, C.W., Ahn, Y.T., Shin, H.S., 2009. Variation of ADM1 by using temperature-phased anaerobic digestion (TPAD) operation. Bioresource Technology 100 (11), 2816e2822. Lewis, R.M., Torczon, V., Trosset, M.W., 2000. Direct-Search Methods: Then and Now. ICASE Report No. 2000e26. Lin, Y.H., Wu, C.L., 2011. Sensitivity analysis of phenol degradation with sulfate reduction under anaerobic conditions. Environmental Modeling & Assessment 16 (2), 213e225. Liwarska-Bizukojc, E., Biernacki, R., 2010. Identification of the most sensitive parameters in the activated sludge model implemented in BioWin software. Bioresource Technology 101 (19), 7278e7285. Lokshina, L., Vavilin, V., Kettunen, H., Rintala, J., Holliger, C., Nozhevnikova, A., 2001. Evaluation of kinetic coefficients using integrated Monod and Haldane models for lowtemperature acetoclastic methanogenesis. Water Research 35 (12), 2913e2922.
Lopez, I., Borzacconi, L., 2009. Modelling a full scale UASB reactor using a COD global balance approach and state observers. Chemical Engineering Journal 146, 1e5. Lopez, I., Borzacconi, L., 2010. Modelling of slaughterhouse solid waste anaerobic digestion: determination of parameters and continuous reactor simulation. Waste Management 30, 1813e1821. Lu¨bken, M., Wichern, M., Schlattmann, M., Gronauer, A., Horn, H., 2007. Modelling the energy balance of an anaerobic digester fed with cattle manure and renewable energy crops. Water Research 41 (18), 4085e4096. Mailier, J., Delmotte, A., Cloutier, M., Jolicoeur, M., Wouwer, A.V., 2011. Parametric sensitivity analysis and reduction of a detailed nutritional model of plant cell cultures. Biotechnology and Bioengineering 108 (5), 1108e1118. Marquardt, D.W., 1963. An algorithm for least squares estimation of nonlinear parameters. Journal of the Society for Industrial and Applied Mathematics 11, 431e441. Martin, C., Ayesa, E., 2010. An integrated Monte Carlo methodology for the calibration of water quality models. Ecological Modelling 221, 2656e2667. Martin, C., de Gracia, M., Ayesa, E., 2011. Bayesian Calibration of the Disintegration process in WWTP Sludge Digesters. 8th IWA Symposium on Systems Analysis and Integrated Assessment (WaterMatex), June 20e22, San Sebastia´n, Spain. Martin, M.A., Raposo, F., Borja, R., Martin, A., 2002. Kinetic study of the anaerobic digestion of vinasse pretreated with ozone, ozone plus ultraviolet light, and ozone plus ultraviolet light in the presence of titanium dioxide. Process Biochemistry 37 (7), 699e706. Molina, F., Castellano, M., Garcia, C., Roca, E., Lema, J., 2009. Selection of variables for on-line monitoring, diagnosis, and control of anaerobic digestion processes. Water Science and Technology 60 (3), 615e622. Morris, M.D., 1991. Factorial sampling plans for preliminary computational experiments. Technometrics 33 (2), 161e174. Mosche, M., Jordening, H.J., 1999. Comparison of different models of substrate and product inhibition in anaerobic digestion. Water Research 33 (11), 2545e2554. Muller, T.G., Noykova, N., Gyllenberg, M., Timmer, J., 2002. Parameter identification in dynamical models of anaerobic waste water treatment. Mathematical Biosciences 177, 147e160. Nocedal, J., Wright, S., 2006. Numerical Optimization. Springer Verlag. Nopharatana, A., Pullammanappallil, P.C., Clarke, W.P., 2003. A dynamic mathematical model for sequential leach bed anaerobic digestion of organic fraction of municipal solid waste. Biochemical Engineering Journal 13 (1), 21e33. Noykova, N., Gyllenberg, M., 2000. Sensitivity analysis and parameter estimation in a model of anaerobic wastewater treatment processes with substrate inhibition. Bioprocess Engineering 23, 343e349. Noykova, N., Muller, T.G., Gyllenberg, M., Timmer, J., 2002. Quantitative analyses of anaerobic wastewater treatment processes: identifiability and parameter estimation. Biotechnology and Bioengineering 78 (1), 89e103. Omlin, M., Reichert, P., 1999. A comparison of techniques for the estimation of model prediction uncertainty. Ecological Modelling 115 (1), 45e59. Ossenbruggen, P.J., Stevens, D.K., 1996. Mechanistic model identification for nitrification in activated sludge. Environmetrics 7 (6), 579e600. Ozkan-Yucel, U.G., Gokcay, C.F., 2010. Application of ADM1 model to a full-scale anaerobic digester under dynamic organic loading conditions. Environmental Technology 31 (6), 633e640. Palatsi, J., Illa, J., Prenafeta-Boldu, F.X., Laureni, M., Fernandez, B., Angelidaki, I., Flotats, X., 2010. Long-chain fatty acids
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
inhibition and adaptation process in anaerobic thermophilic digestion: batch tests, microbial community structure and mathematical modelling. Bioresource Technology 101 (7), 2243e2251. Pobeheim, H., Munk, B., Mueller, H., Berg, G., Guebitz, G., 2010. Characterization of an anaerobic population digesting a model substrate for maize in the presence of trace metals. Chemosphere 80 (8), 829e836. Qian, S.S., Stow, C.A., Borsuk, M.E., 2003. On Monte Carlo methods for bayesian inference. Ecological Modelling 159 (2e3), 269. Ramirez, I., Mottet, A., Carrere, H., Deleris, S., Vedrenne, F., Steyer, J.P., 2009. Modified ADM1 disintegration/hydrolysis structures for modeling batch thermophilic anaerobic digestion of thermally pretreated waste activated sludge. Water Research 43, 3479e3492. Raposo, F., Borja, R., Martı´n, M.A., Martı´n, A., de la Rubia, M.A., Rinco´n, B., 2009. Influence of inoculumesubstrate ratio on the anaerobic digestion of sunflower oil cake in batch mode: process stability and kinetic evaluation. Chemical Engineering Journal 149, 70e77. Reichert, P., 2010. Environmental Systems Analysis Lecture Notes Summer Course, June 21e25, Eawag Du¨bendorf, Switzerland. Redzwan, G., Banks, Ch, 2004. The use of a specific function to estimate maximum methane production in a batch-fed anaerobic reactor. Journal of Chemical Technology and Biotechnology 79, 1174e1178. Rincon, A., Angulo, F., Olivar, G., 2009. Control of an anaerobic digester through normal form of fold bifurcation. Journal of Process Control 19 (8), 1355e1367. Rittmann, B.E., 2008. Opportunities for renewable bioenergy using microorganisms. Biotechnology and Bioengineering 100 (2), 203e212. Rodrigues, J.A.D., Ratusznei, S.M., Zaiat, M., 2003. Fed-batch and batch operating mode analysis of a stirred anaerobic sequencing reactor with self-immobilized biomass treating low-strength wastewater. Journal of Environmental Management 69 (2), 193e200. Rodriguez, J., Lema, J.M., Van Loosdrecht, M.C.M., Kleerebezem, R. , 2006. Variable stoichiometry with thermodynamic control in ADM1. Water Science and Technology 54 (4), 101e110. Rodriguez, J., Roca, E., Lema, J.M., Bernard, O., 2008. Determination of the adequate minimum model complexity required in anaerobic bioprocesses using experimental data. Journal of Chemical Technology and Biotechnology 83 (12), 1694e1702. Ruel, S.M., Comeau, Y., Ginestet, P., Heduit, A., 2002. Modeling acidogenic and sulfate-reducing processes for the determination of fermentable fractions in wastewater. Biotechnology and Bioengineering 80 (5), 525e536. Ruzicka, M., 1996. The effect of hydrogen on acidogenic glucose cleavage. Water Research 30 (10), 2447e2451. Sales-Cruz, M., Gani, R., 2004. Aspects of modelling and model identification for bioprocesses through a computer-aided modelling system. In: Barbosa-Po´voa, Ana Paula, Matos, Henrique (Eds.), Computer Aided Chemical Engineering European Symposium on Computer-Aided Process Engineering14, 37th European Symposium of the Working Party on Computer-Aided Process Engineering. Elsevier, pp. 1123e1128. Saltelli, A., Ratto, M., Tarantola, S., Campolongo, F., 2006. Sensitivity analysis practices: strategies for model-based inference. Reliability Engineering and System Safety 91, 1109e1125. Saltelli, A., Ratto, M., Andres, T., Campolongo, F., Cariboni, J., Gatelli, D., Saisana, M., Tarantola, S., 2008. Global Sensitivity Analysis. The Primer. John Wiley & Sons, West Sussex, England. Sanchez, J.M., Arijo, S., Mun˜oz, M.A., Morinigo, M.A., Borrego, J.J., 1994. Microbial colonization of different support materials
5363
used to enhance the methanogenic process. Applied Microbiology and Biotechnology 41 (4), 480e486. Saravanan, V., Sreekrishnan, T.R., 2006. Modelling anaerobic biofilm reactors e a review. Journal of Environmental Management 81 (1), 1e18. Sbarciog, M., Loccufier, M., Noldus, E., 2010. Determination of appropriate operating strategies for anaerobic digestion systems. Biochemical Engineering Journal 51 (3), 180e188. Schenkendorf, R., Kremling, A., Mangold, M., 2009. Optimal experimental design with the sigma point method. IET Systems Biology 3 (1), 10e23. Siegrist, H., Renggli, D., Gujer, W., 1993. Mathematical-Modeling of anaerobic mesophilic sewage-sludge treatment. Water Science and Technology 27 (2), 25e36. Siegrist, H., Vogt, D., Garcia-Heras, J., Gujer, W., 2002. Mathematical model for meso- and thermophilic anaerobic sewage sludge digestion. Environmental Science and Technology 36 (5), 1113e1123. Simeonov, I., Momchev, V., Grancharov, D., 1996. Dynamic modeling of mesophilic anaerobic digestion of animal waste. Water Research 30 (5), 1087e1094. Simeonov, I.S., 1999. Mathematical modeling and parameters estimation of anaerobic fermentation processes. Bioprocess Engineering 21 (4), 377e381. Sin, G., Gernaey, K.V., Neumann, M.B., Van Loosdrecht, M.C.M., Gujer, W., 2011. Global sensitivity analysis in wastewater treatment plant model applications: prioritizing sources of uncertainty. Water Research 45 (2), 639e651. Sobol, I.M., 2001. Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates. Mathematics and Computers in Simulation 55 (1e3), 271e280. Smith, L.H., McCarty, P.L., Kitanidis, P.K., 1998. Spreadsheet method for evaluation of biochemical reaction rate coefficients and their uncertainties by weighted nonlinear least-squares analysis of the integrated monod equation. Applied and Environmental Microbiology 64 (6), 2044e2050. Speece, R.E., 1996. Anaerobic Biotechnology: For Industrial Wastewater. Archae Press. Spriet, J.A., 1985. Structure characterization e an overview. In: Barker, H.A., Young, P.C. (Eds.), Identification and System Parameter Estimation 1985 e Proceedings of the seventh IFAC/ IFORS Symposium. Pergamon Press, Oxford, pp. 749e756. Tarantola, S., Gatelli, D., Mara, T.A., 2006. Random balance designs for the estimation of first order global sensitivity indices. Reliability Engineering and System Safety 91 (6), 717e727. Tartakovsky, B., Morel, E., Steyer, J.P., Guiot, S.R., 2002. Application of a variable structure model in observation and control of an anaerobic digestor. Biotechnology Progress 18 (4), 898e903. Tartakovsky, B., Mu, S.J., Zeng, Y., Lou, S.T., Guiot, S.R., Wu, P., 2008. Anaerobic digestion model no. 1-based distributed parameter model of an anaerobic reactor: II. Model validation. Bioresource Technology 99 (9), 3676e3684. Tomei, M.C., Braguglia, C.M., Cento, G., Mininni, G., 2009. Modeling of anaerobic digestion of sludge. Critical Reviews in Environmental Science and Technology 39 (12), 1003e1051. Vande Wouwer, A., Renotte, C., Queinnec, I., Bogaerts, P., 2006. Transient analysis of a wastewater treatment biofilterdistributed parameter modeling and state estimation. Mathematical and Computer Modelling of Dynamical Systems 12, 423e440. Van Lier, J.B., 2008. High-rate anaerobic wastewater treatment: diversifying from end-of-the-pipe treatment to resourceoriented conversion techniques. Water Science and Technology 57 (8), 1137e1148.
5364
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 4 7 e5 3 6 4
Vavilin, V.A., Vasiliev, V.B., Ponomarev, A.V., Rytow, S.V., 1994. Simulation-model methane as a tool for effective biogas production during anaerobic conversion of complex organicmatter. Bioresource Technology 48 (1), 1e8. Vavilin, V.A., Rytow, S.V., Lokshina, L.Y., 1995. Modelling hydrogen partial pressure change as a result of competition between the butyric and propionic groups of acidogenic bacteria. Bioresource Technology 54 (2), 171e177. Vavilin, V., Rytov, S.V., Pavlostathis, S.G., Jokela, J., Rintala, J., 2003. A distributed model of solid waste anaerobic digestion: sensitivity analysis. Water Science and Technology 48 (4), 147e154. Walter, E., Pronzato, L., 1997. Identication of Parametric Models from Experimental Data. Springer-Verla, Berlin. Ward, A.J., Hobbs, P.J., Holliman, P.J., Jones, D.L., 2008. Optimisation of the anaerobic digestion of agricultural resources. Bioresource Technology 99 (17), 7928e7940. Ward, A.J., Hobbs, P.J., Holliman, P.J., Jones, D.L., 2011. Evaluation of near infrared spectroscopy and software sensor methods for determination of total alkalinity in anaerobic digesters. Bioresource Technology 102 (5), 4083e4090.
Weijers, S.R., 2003. Model reduction of activated sludge model no. 1 and bioprocess models for identification and control. Biotechnology for the environment: wastewater treatment and modeling. Waste Gas Handling 3C, 11e63. Wichern, M., Gehring, T., Fischer, K., Andrade, D., Lubken, M., Koch, K., Gronauer, A., Horn, H., 2009. Monofermentation of grass silage under mesophilic conditions: measurements and mathematical modeling with ADM 1. Bioresource Technology 100 (4), 1675e1681. Wolf, C., McLoone, S., Bongards, M., 2008. Plant optimization using genetic algorithms and particle swarm optimization. In: Signals and Systems Conference, 208. (ISSC 2008). IET Irish. Zaher, U., Vanrolleghem, P., Rodriguez, J., Franco, A., 2004. Conceptual approach for ADM1 applications. In: Ujang, Z., Henze, M. (Eds.), Environmental Biotechnology: Advancement in Water and Wastewater Applications in the Tropics. IWA Publishing. Zaher, U., Pandey, P., Chen, S., 2009. A simple elemental continuity based model application to study the anaerobic microbial activity for the treatment of dairy manure. Applied Mathematical Modelling 33 (9), 3553e3564.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Evaluation of a laboratory-scale bioreactive in situ sediment cap for the treatment of organic contaminants David W. Himmelheber a,*, Kurt D. Pennell b, Joseph B. Hughes a,c a
School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA Department of Civil and Environmental Engineering, Tufts University, Medford, MA, USA c School of Material Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA b
article info
abstract
Article history:
The development of bioreactive sediment caps, in which microorganisms capable of
Received 13 December 2010
contaminant transformation are placed within an in situ cap, provides a potential remedial
Received in revised form
design that can sustainably treat sediment and groundwater contaminants. The goal of
7 April 2011
this study was to evaluate the ability and limitations of a mixed, anaerobic dechlorinating
Accepted 17 June 2011
consortium to treat chlorinated ethenes within a sand-based cap. Results of batch exper-
Available online 30 June 2011
iments demonstrate that a tetrachloroethene (PCE)-to-ethene mixed consortium was able to completely dechlorinate dissolved-phase PCE to ethene when supplied only with sedi-
Keywords:
ment porewater obtained from a sediment column. To simulate a bioreactive cap,
Sediment remediation
laboratory-scale sand columns inoculated with the mixed culture were placed in series
In situ capping
with an upflow sediment column and directly supplied sediment effluent and dissolved-
Microbial processes
phase chlorinated ethenes. The mixed consortium was not able to sustain dechlorina-
Bioremediation
tion activity at a retention time of 0.5 days without delivery of amendments to the sediment effluent, evidenced by the loss of cis-1,2-dichloroethene (cis-DCE) dechlorination to vinyl chloride. When soluble electron donor was supplied to the sediment effluent, complete dechlorination of cis-DCE to ethene was observed at retention times of 0.5 days, suggesting that sediment effluent lacked sufficient electron donor to maintain active dechlorination within the sediment cap. Introduction of elevated contaminant concentrations also limited biotransformation performance of the dechlorinating consortium within the cap. These findings indicate that in situ bioreactive capping can be a feasible remedial approach, provided that residence times are adequate and that appropriate levels of electron donor and contaminant exist within the cap. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The management and remediation of contaminated aquatic sediments pose major technical and economic challenges. Treatment of contaminated sediment sites with in situ caps has become an established practice that can provide advantages over alternative methods in certain settings
(Reible et al., 2003). Clean sand has traditionally been employed as capping material, and remains a large component of many field-scale capping applications. Sand-based caps have the potential to delay contaminant breakthrough when diffusive transport dominates (Go et al., 2009; Thoma et al., 1993), but eventual contaminant breakthrough remains a source of concern. Additionally, traditional sand
* Corresponding author. Geosyntec Consultants, 10220 Old Columbia Road, Suite A, Columbia, MD 21046, USA. Tel.: þ1 410 381 4333; fax: þ1 410 381 4499. E-mail address:
[email protected] (D.W. Himmelheber). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.022
5366
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
caps are less effective at sites where groundwater seepage or mobile contaminants (i.e., low Koc) are present (Go et al., 2009). Research studies have focused on in situ sequestration (Cho et al., 2007; Zimmerman et al., 2004), in situ transformation (Krumins et al., 2009; Lowry and Johnson, 2004), and the development of active caps which incorporate reactive and/or sorptive constituents designed to reduce contaminant and bioavailability (Choi et al., 2009; Hyun et al., 2006; Jacobs and Fo¨rstner, 1999; McDonough et al., 2007; Murphy et al., 2006; Reible et al., 2007). Ideally, active caps eliminate the risk of contaminant breakthrough into the overlying water column, and can potentially be implemented at sediment sites with groundwater seeps and relatively mobile contaminants. The employment of physicochemical-based active caps appears promising, but possible limitations (e.g., high material costs, sorption and reaction capacities) have stimulated the consideration of in situ bioreactive caps, in which contaminant biotransformations are designed to occur within the cap matrix to produce environmentally-acceptable reaction products. Enhanced in situ bioremediation, through biostimulation and bioaugmentation, has proven to be a successful groundwater remediation technology for a diverse range of contaminants (Lo¨ffler and Edwards, 2006). Adaptation of these principles to subaqueous sediment remediation has not been demonstrated, prompting the recent identification of in situ bioremediation as a priority research and development need (SERDP/ESTCP, 2008). Biologically-based active caps have the potential to maintain reactivity over long periods of time and could serve as a sustainable remedial option if microorganisms capable of biotransformation are present and necessary metabolic requirements are met. Previous studies that investigated the activity of microbial populations within a sediment cap demonstrated that microorganisms indigenous to underlying sediment, including organisms capable of contaminant biotransformation, are able to colonize the overlying cap and possibly participate in contaminant bioattenuation processes (Himmelheber et al., 2009). Bioaugmentation of microorganisms within a cap, as opposed to intrinsic colonization (defined here as the natural redistribution of microorganisms native to the sediment into the cap matrix), could provide enhanced degradation capacity and minimize the potential for contaminant release to benthic and aqueous receptors. Such a bioaugmentation strategy was recently evaluated by the US Geological Survey (USGS) as a means to reductively dechlorinate a mixture of chlorinated ethenes, ethanes, and methanes present in a groundwater seep discharging into a tidal wetland (Majcher et al., 2007). A mixed, anaerobic culture was enriched from the site (Lorah et al., 2008) and incorporated into an organic-based matrix that was placed at the sediment-water interface. This bioreactive mat successfully treated the chlorinated contaminants prior to discharge (Majcher et al., 2009). Although the bioreactive mat was constructed on the banks of a tidal wetland (i.e., not completely subaqueous) and the design is not immediately suitable for submergence (e.g., buoyancy restrictions, delivery of bioaugmentation culture), the success of the approach supports the concept of bioreactive capping as an in situ remedial technique.
The USGS bioreactive mat was designed in part because the chlorinated organics present in the groundwater were undergoing only partial dechlorination in the sediment prior to discharge, a phenomenon commonly reported at sediment sites (Abe et al., 2009; Conant et al., 2004; Hamonts et al., 2009; Himmelheber et al., 2007; Lendvay et al., 1998; Lorah and Voytek, 2004; Majcher et al., 2007). Additionally, recent studies have demonstrated that anaerobic conditions develop within sediment caps subject to diffusive and upflow conditions (i.e., groundwater seeps) (Himmelheber et al., 2008, 2009). It is therefore expected that contaminated groundwater seeps will carry partially-degraded contaminants into the overlying anaerobic cap, thereby providing an opportunity for treatment by reductive biotransformations. Detailed assessment of bioreactive in situ sediment caps has not been previously undertaken and little is currently known about the feasibility of bioreactive caps, particularly their limitations and maintenance requirements. The objective of this work was to establish an actively dechlorinating microbial consortium within a simulated overlying cap and to determine how contaminant mass flux and electron donor amendments influenced bioreactive cap performance. More specifically, the bioreactive cap experiments were designed to determine whether or not amendments are necessary to sustain complete reductive dechlorination by an active microbial community. Chlorinated ethenes were utilized as the contaminants due to their frequent occurrence as groundwater contaminants, their presence in groundwater seeps, and their greater mobility relative to other sediment contaminants (e.g., chlorinated benzenes, polychlorinated biphenyls). Batch reactor and bioaugmented column studies were conducted to assess bioreactive cap performance over a range of electron donor and contaminant conditions.
2.
Materials and methods
2.1.
Chemicals
PCE (99þ%, SigmaeAldrich, St. Louis, MO), TCE (99.5%, SigmaeAldrich), cis-DCE (97%, Acros Organics, Morris Plains, NJ), trans-DCE (99.7%, Acros Organics), and 1,1-DCE (99.9%, Acros Organics) were obtained in neat liquid form. Vinyl chloride (8%/N2 balance), ethene (99.5%), ethane (99.5%), and methane (99%) were obtained from Matheson Tri-Gas (Parsippany, NJ). Sodium bicarbonate, potassium chloride, magnesium chloride, and calcium chloride were used in the preparation of simulated groundwater and were purchased from Fisher Scientific (Pittsburgh, PA). Sodium lactate syrup (60% vol/vol, Fisher Scientific) was used during the preparation of stock lactate solutions. Potassium bromide, calcium sulfate, and potassium phosphate were purchased from Fisher Scientific and used for IC standards and non-reactive tracer studies.
2.2.
Batch reactors
Batch reactors were established in triplicate and consisted of Anacostia River (Washington, D.C., USA) sediment porewater, dissolved-phase PCE, and a mixed PCE-to-ethene dechlorinating consortium. A PCE-to-ethene dechlorinating mixed
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
consortia referred to as OW served as the inoculum. The OW consortia, which is capable of complete reductive dechlorination of PCE to ethene, has been described previously (Daprato et al., 2006). The OW culture has been found to contain multiple dechlorinating microorganisms, including Dehalococcoides species, and known reductive dehalogenases including tceA, vcrA, and bvcA (Daprato et al., 2006). Three 25 mL aliquots of OW culture were transferred to 70 mL serum bottles pre-capped with Teflon-faced butyl septa and sparged with N2 gas for 15 min to remove oxygen from the empty bottles. The collected OW aliquots were then sparged with N2 gas for 15 min in attempt to remove residual chlorinated ethenes, methanol, and volatile fatty acids from the batch reactors. Sediment effluent was collected from a sediment column that was supplied only with simulated groundwater and dissolved-phase PCE (Himmelheber et al., 2007). The composition of simulated groundwater was slightly modified from that described by Dries et al. (2004) and consisted of 3.5 mM NaHCO3, 0.1 mM KCl, 0.25 mM MgCl2, 0.75 mM CaCl2, and resazurin as a redox indicator. Sediment effluent was collected under anoxic conditions and 25 mL of effluent were added directly to batch reactors containing the OW consortium. Dissolved-phase PCE was obtained from a saturated stock solution containing neat PCE in contact with anaerobic, sterilized simulated groundwater. Stock PCE concentrations were quantified immediately prior to injection into the batch reactors. Approximately 16 mmol of dissolvedphase PCE was added to each microcosm using a 10 mL Hamilton glass syringe. All reactors were wrapped in foil and incubated at 20 C on an orbital shaker operated at 150 rpm. Chlorinated ethenes, ethene, ethane, and methane concentrations were determined from headspace samples of the microcosms.
2.3.
Bioreactive cap operation
Two one-dimensional (1-D) columns (designated herein as Bioreactive Cap A and Bioreactive Cap B) were constructed using 2.5 cm inside diameter (I.D.) glass chromatography columns 30 cm in length (Spectrum Chromatography, Houston, TX) and equipped with custom-built stainless steel end plates (Dutton & Hall, Atlanta, GA). A 2.5 cm diameter disc of 80 mesh stainless steel (Small Parts, Inc., Miami Lakes, FL) was placed on the column end plates to retain sand grains within the column. A fabricated glass reservoir (15 mL) fitted with a stopcock was placed at the column effluent to allow for aqueous effluent sampling. The columns were packed with ASTM C-33 grade concrete sand (U.S. Silica, Mauricetown, NJ). This particular sand was selected because it is representative of the solids used for submerged sediment caps and was utilized in the Anacostia River Capping Demonstration Project (Reible, 2005). An elemental analysis of the sand was performed at the University of Georgia Laboratory for Environmental Analysis (see Supplementary Information, Table S.1). The dry, autoclaved sand was packed into the bioreactive columns under aerobic conditions in 5-cm increments with vibration along the outside wall of the column. Three pore volumes of N2-sparged, autoclaved simulated groundwater were flushed through the columns to check for leakage and to ensure anaerobic conditions. The columns were then
5367
inoculated by flushing the columns with three pore volumes of the OW culture suspension. Following inoculation, the two sand cap columns were wrapped in foil to avoid exposure to light then connected in series with an upflow column packed with Anacostia River sediment as depicted in Fig. 1. The sediment column effluent, which was provided only with simulated groundwater and dissolved-phase PCE, served as the influent for the bioreactive sand columns. Therefore, the influent for the sand columns consisted of sediment effluent and a mixture of partial PCEdechlorination products, similar to the conditions that would be anticipated in a submerged sediment capping scenario subject to a PCE-contaminated groundwater seep. Table 1 provides a summary of experimental conditions employed for each bioreactive sand column. Chlorinated ethene and ethene concentrations in the effluent of Bioreactive Caps A and B were normalized on a molar basis to total chlorinated ethenes and ethene eluted per sample to reduce scatter in concentration data and to monitor product distribution.
2.3.1.
Bioreactive Cap A
Bioreactive Cap A was designed to assess the ability of sediment effluent to maintain an external dechlorinating community in a cap, simulating a bioreactive cap inoculated with a mixed dechlorinating consortia and operating under reducing conditions. Prior to inoculating Bioreactive Cap A, a tracer test was conducted with a pulse injection of 100 mg L1 (1.25 mM) bromide obtained from an autoclaved, sparged stock solution of potassium bromide in simulated groundwater. A total of 1.2 pore volumes were flushed through the column, collected with a fraction collector, and analyzed via ion chromatography. Three pore volumes of simulated groundwater were then flushed through the column following the tracer test to remove residual bromide prior to inoculation. A 200 mL aliquot of aqueous OW culture was obtained for inoculation and stored in a 160 mL serum bottle that had previously been capped with a Teflonfaced butyl septum and sparged with N2 for 15 min to remove oxygen. The 200 mL aliquot was tested for its dechlorination ability in batch conditions by spiking with PCE and methanol. After successfully dechlorinating PCE to ethene (Supplementary Information, Fig. S.1A), 1.5 pore volumes of the OW culture were supplied to the column at a flow rate of 2.2 mL h1 (1-day residence time). Following a 24-h attachment period during which there was no flow, Bioreactive Cap A was connected in series with the sediment column from 67 to 83 sediment pore volumes. The unamended sediment column effluent served as the influent for the duration of the Bioreactive Cap A experiment.
2.3.2.
Bioreactive Cap B
The Bioreactive Cap B experiment was designed to simulate a dechlorinating bioreactive cap operating under reducing conditions, but differed from Cap A in that the influent for this experiment was supplied at various flow rates and periodically spiked with amendments. Thus, Bioreactive Cap B demonstrates the impact of contaminant influx and the presence of reducing equivalents on the capacity of sediment column effluent to maintain an external dechlorinating
5368
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
Gas purge following sample collection
Anacostia Sediment
30 cm
Sand incoluated with PCF, to ethene mixed consortia
(C) (A) Anoxic Simulated Groundwater + Dissolved PCE
(B)
Upflow
30 cm
Effluent Samples
Upflow
Effluent Samples
Amendments and Flow Rate Control to Bioreactive Cap B
Captured Sediment Effluent for Bioreactive Cap B
Fig. 1 e Conceptual schematic of laboratory simulation of a bioreactive sand cap placed in series with an anaerobic sediment bed subject to a PCE-contaminated groundwater seep. (A) Sediment effluent was directly supplied to Bioreactive Cap A. (B) Sediment effluent was initially captured via a syringe for Bioreactive Cap B, spiked with amendments, then (C) supplied to Bioreactive Cap B at select flow rates.
community. An aliquot of OW culture was retrieved and sparged with nitrogen prior to inoculation as described for Bioreactive Cap A. The aliquot of OW culture again demonstrated the ability to completely dechlorinate PCE to ethene in batch culture (Supplementary Information, Fig. S.1B). A total of 1.7 pore volumes of OW culture was then supplied to the column at a flow rate of 2.6 mL h1 (1-day residence time), followed by a no-flow attachment period of one day. Unlike Bioreactive Cap A, Bioreactive Cap B was not immediately connected to the sediment column effluent, but rather positive-control experiments were conducted to ensure the inoculated column could completely dechlorinate cis-DCE to ethene when provided DCB-1 media, Wolin vitamins, and 5 mM lactate as an electron donor and carbon source. Following this demonstration of complete dechlorination in the cap under optimal conditions (Supplementary Information, Fig. S.2), one pore volume of anaerobic simulated groundwater was flushed through the column to remove these constituents from the system prior to the introduction of sediment effluent. Sediment column effluent was obtained
from 146 to 180 sediment pore volumes to serve as Bioreactive Cap B influent. For Bioreactive Cap B, sediment column effluent was captured under anoxic conditions by connecting an empty, gas-tight syringe to sediment effluent tubing and allowing the aqueous flow to gradually fill the syringe at the same rate of sediment column influent (5.5 mL h1). Once the effluent syringe had been filled, it was immediately transferred to a separate syringe pump and introduced into the sand column as the influent. This method allowed for manipulation of flow rates within the sand column and for addition of electron donor and acceptor to the influent prior to connection with the sand column. The electron donor used for this study was lactate, which was obtained from a 100 mM stock solution in autoclaved, sparged simulated groundwater. Lactate was supplied to the bioreactive sand column (Cap B) at a concentration of 5 mM from 0 to 13.3 pore volumes (Table 1). The experimental conditions employed for Bioreactive Cap B were designed to gradually decrease aqueous residence times, as well as electron donor concentrations, to determine
5369
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
Pore volume (PV)a (mL) Porosity (n)a (cm3 void (cm3 total)1) Connected in series to sediment column (sediment pore volumes) Experimental flow rate (Q) (mL h1) Porewater velocity (v) (cm day1) (Darcy velocity (cm day1)) Peclet number (Pe)b (dimensionless) Alterations to influent
Influent chloroethene concentration (mM total chlorinated ethenes)
Bioreactive Cap A
Bioreactive Cap B
62.72 0.41
61.82 0.41
67.0e83.2
146.0e180.0
5.46
1.29; 2.58; 5.46
62.67 (25.88)
14.99; 29.98; 63.59 (6.10); (12.20); (25.88) N/Ac
80.5 None
16.19 11.06d
Addition of Lactate Addition of cisDCE Decrease of flow rate 0e3.44 PV: 200 42d 3.44 PV to end: 34 3.6d
a Estimated from mass difference between dry and wet packed columns. b Obtained with the CFITM3 breakthrough curve fitting program under equilibrium constraints. c Tracer test not performed. d Average one standard deviation.
limitations on dechlorination (Table 1). The influent flow rate for Bioreactive Cap B was increased step-wise from 1.3 mL h1 (2-day retention time), to 2.6 mL h1 (1-day retention time) to 5.5 mL h1 (0.47-day retention time). From 0 to 3.4 sand pore volumes, additional cis-DCE was provided to the influent to ensure chlorinated ethenes were present due to complete dechlorination of PCE to ethene in the sediment column effluent prior to connecting Bioreactive Cap B. cis-DCE was chosen assuming partial, intrinsic PCE dechlorination would occur in sediment beds, based on prior research findings (Himmelheber et al., 2007). The cis-DCE was obtained from a saturated stock solution of cis-DCE (i.e., NAPL present) in autoclaved, sparged simulated groundwater and supplied to the influent at a concentration of 200 42 mM. After 3.4 pore volumes, however, the only source of chlorinated ethenes to Bioreactive Cap B was the sediment effluent. Lactate (5 mM) was provided from 0 to 13.3 pore volumes, at which point it was removed from the influent and no electron donor was provided for the remainder of the experiment.
2.4.
Analytical methods
PCE, TCE, DCE isomers, VC, ethene, ethane, and methane concentrations were determined from the headspace of 5 mL aqueous effluent samples, which were analyzed using an Agilent 6890 gas chromatograph (GC) equipped with a flame ionization detector (FID), as described previously (Carr and Hughes, 1998). Bromide was measured using a Dionex DX-
3.
Results and discussion
3.1.
Batch reactors
The OW culture successfully dechlorinated PCE to ethene when provided only sediment effluent and dissolved-phase PCE (Fig. 2). Complete PCE dechlorination to ethene was achieved after 19 days of incubation. Chlorinated ethene mass balance was within 10% for each time point except day 12, when chloroethene mole totals were 124% of the initial dissolved-phase PCE introduced to the batch reactors. This discrepancy arose because one of the triplicate reactors recorded unusually high concentrations of VC, despite balanced ethene concentrations at the end of the experiment. This is reflected in the relatively high standard deviation of VC at day 12. Duplicate analysis at the same time point yielded similar results. Regardless of this isolated analytical discrepancy, the presence of VC and ethene indicates that dechlorinating species within the OW consortium, specifically Dehalococcoides, remained active for at least one dechlorination cycle when provided only sediment effluent and a dissolved-phase electron acceptor (PCE). Thus, the sources of carbon, electron donor, and micronutrients were provided by the sediment effluent or from microbial biomass (Adamson and Newell, 2009). Methane concentrations rose steadily during the dechlorination of PCE (Fig. 2), indicating that methanogenic populations were also able to remain active when provided only sediment effluent. These data suggest that dechlorinating species within a bioreactive cap inoculated with a methanogenic mixed consortia may have to 20
120 Methane PCE TCE DCE VC ETH
16
12
100
80
60
8 40
4
Total Methane (mM)
Parameter
100 ion chromatograph with a Dionex AG4A IonPac guard column and Dionex AS4A IonPac column at a flow rate of 1.5 mL/min and an ED40 electrochemical detector.
Total Chloroethene (µmoles)
Table 1 e Summary of experimental conditions for sand column experiments.
20
0
0
0
3
6
9 12 Time (days)
15
18
Fig. 2 e Batch microcosm results of OW culture provided only PCE and sediment effluent. Chlorinated ethenes are reported as the sum of aqueous and gas phases within the microcosms. Error bars represent one standard deviation calculated from triplicate reactors. The shaded background area corresponds to methane production (mM) at each time point and is referenced to the right vertical axis. Total methane is calculated as the sum of aqueous and gas phase methane.
5370
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
compete with methanogens for electron donors, which could result in reduced dechlorination efficiency over time. Methanogens and other microbial populations indigenous to the sediment are also expected to populate the cap material (Himmelheber et al., 2009) and may, therefore, compete for electron donor and other nutrients.
3.2.
Bioreactive Cap A
The pore volume of Bioreactive Cap A was estimated to be 62.7 mL from differences between wet and dry column mass and assuming complete water saturation (Table 1). The nonreactive tracer test conducted at the onset of Bioreactive Cap A operation yielded a symmetrical breakthrough curve, indicative of the absence of immobile regions of water (see Supplementary Information, Fig. 2). The measured tracer BTC, expressed as the relative concentration versus number of dimensionless pore volumes applied, was fit to an analytical solution of the one-dimensional advectivedispersive reactive (ADR) transport equation using the CXTFIT model (van Genuchten, 1981). As anticipated, the fitted retardation factor (RF) obtained from the tracer BTC was approximately equal to 1.0, indicating no detectable interactions between the solid phase and tracer during transport through the sand column. The fitted Peclet number (Pe) was approximately 81, yielding a hydrodynamic dispersion coefficient (DH) of 2.8 108 m2 s1 and a hydrodynamic dispersivity (aD) of 0.37 cm. These data are consistent with values reported for similar water-saturated columns packed with graded sands, and indicate that advective flow and transport through the column was normal and not subject to physical nonequilibrium. Chlorinated ethene effluent product distributions, normalized to moles of chlorinated ethenes and ethene eluted, are shown for Bioreactive Cap A (Fig. 3B). The applied influent flow rate of 5.5 mL h1 corresponded to a column residence time of 0.47 days. When Bioreactive Cap A was connected in series to the sediment column, cis-DCE was the predominant chlorinated ethene present in influent solution. The bioreactive sand column was initially able to dechlorinate cis-DCE to VC, but ethene was not detected (Fig. 3B). This dechlorination activity disappeared prior to 5 pore volumes, and eventually only 5% of the cis-DCE was dechlorinated to VC, indicating that Dehalococcoides activity was impaired. Methane data collected during the Bioreactive Cap A experiment reveal that microbes other than dechlorinators also lost activity, suggesting microbial impairment in the system as a whole and not just for the dechlorinating population (Fig. 3C). Based on data presented in Fig. 3B, the sediment column effluent was not able to sustain the dechlorinating consortium OW without additional amendments. Data were not collected to determine if non-contaminant stressors (e.g., ammonia) were present in the sediment, which could suppress microbial activity. However, previous research (Himmelheber et al., 2007) has demonstrated that microorganisms, specifically Dehalococcoides strains, can be stimulated in the Anacostia sediment with the addition of electron donor, suggesting that non-contaminant stressors were not a major concern in the system. Previous research (Himmelheber et al., 2007) has also demonstrated that
microbial activity in the sediment column was limited by electron donor availability. It was therefore hypothesized that the levels of electron donor eluting from the sediment column effluent prevented the dechlorinating community in the sand cap from maintaining sufficient activity to achieve complete reductive dechlorination of the cis-DCE introduced to the sand cap column. A second possibility is that the relatively high flow rates through the sand cap column did not provide sufficient contact time between the contaminants and the dechlorinating community to achieve complete reductive dechlorination.
3.3.
Bioreactive Cap B
To address the hypotheses raised above, the second sand column, Bioreactive Cap B, was operated at three different flow rates with and without the addition of lactate as an electron donor and cis-DCE as an electron acceptor (Table 1). The experimental conditions associated with Bioreactive Cap B are presented in Fig. 4A, while normalized chloroethene product distributions are shown in Fig. 4B. The pore volume for Bioreactive Cap B was estimated to be 61.8 mL from differences between wet and dry column mass and assuming complete water saturation (Table 1). Prior to supplying Bioreactive Cap B with sediment effluent, the inoculated column was able to completely dechlorinate cis-DCE to ethene when provided electron donor, carbon sources, vitamins, and reduced media; confirming the ability of the OW culture to achieve complete dechlorination within the column (Supplementary Information, Fig. S.2). After applying one pore volume of simulated groundwater, sediment effluent was supplied to Bioreactive Cap B, indicated as pore volume 0 in Fig. 4A and B. The influent for Bioreactive Cap B was the sediment column effluent from 146 to 180 sediment pore volumes, which contained a mixture of cis-DCE, VC, and ethene. The influent solution provided to Bioreactive Cap B was initially augmented with cis-DCE to yield a total influent chloroethene concentration of approximately 200 mM and 5 mM lactate, operated at a residence time of 2 days (flow rate ¼ 1.29 mL h1) (Table 1). Incomplete dechlorination was observed during this period, with a mix of VC and ethene in the sand column effluent. From 3.4 to 5.7 pore volumes, only lactate was provided to the influent porewater (i.e., no cis-DCE was added) and the sediment effluent served as the sole source of chlorinated ethenes (ca. 34 mM). The sand column successfully achieved complete reductive dechlorination of the applied chlorinated ethenes to ethene during this period, demonstrating that with lactate addition and a residence time of 2 days the sand cap could detoxify the flux of chlorinated ethenes exiting the sediment column. These data, coupled with the lack of complete dechlorination during the previous condition (0e3.4 sand pore volumes) when additional cis-DCE was provided to the influent, suggests that high chloroethene concentrations entering the sand column limited the extent of dechlorination. The results obtained from Bioreactive Cap B indicate that electron donor concentrations and contaminant residence times within the cap can impact dechlorination activity. Complete dechlorination was observed between 8.0 and 13.3 pore volumes when lactate was provided to the sand column
5371
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
A
Bioreactive Cap Experimental Conditions Sand Cap connected in series with mud effluent No lactate provided R = 0.47 days 0
B
5
10
15
20
25
30
35
Product Distribution Normalized to Total Chloroethenes + Ethenes Eluted
Bioreactive Sand Cap Effluent Product Distribution 1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
DCE VC ETH Ethane
0.0 0
5
10
15
20
25
30
35
Pore Volumes Eluted from Cap
C
Bioreactive Sand Cap Effluent Methane Production
Cummulative Methane (mg/L)
4
4
3
3
2
2
1
1
0
0 0
5
10
15
20
25
30
35
Pore Volumes Eluted from Cap Fig. 3 e AeB. (A) Operating conditions for Bioreactive Cap A. The effluent of the sediment column served as the influent of the sand column and was not amended with exogenous electron donors, electron acceptors, carbon sources, minerals, nor vitamins. (B) Effluent product distribution of Bioreactive Cap A inoculated with a PCE-to-ethene dechlorinating mixed consortia and connected in series with sediment column effluent between 68 and 83 sediment pore volumes. (C) Cumulative aqueous methane concentration in samples collected from Bioreactive Cap A effluent.
despite relatively fast flow rates (0.47 residence time). When lactate was removed from the influent at 13.3 pore volumes, however, a mixture of chlorinated ethenes was observed in the effluent, indicating the importance of exogenous reducing equivalents to the sand column. Delivery of external electron donor is a common technique used to stimulate and enhance reductive dechlorination in groundwater aquifers (Anderson et al., 2003; Haas and Trego, 2001; Lendvay et al., 2003; Scow and Hicks, 2005) and may be necessary for bioreactive caps employing anaerobic biotransformations. Contaminant mass entering the cap also dictated performance, as noted above, since incomplete dechlorination was observed when additional cis-DCE was supplemented into the influent (0e3.4 PV) while complete dechlorination was observed when the cap was only treating sediment effluent (5e10 PV).
3.4.
Implications for capping
The combined results from the batch study and the two sand columns suggest that the sediment effluent alone could not sustain complete dechlorination in a bioreactive cap over the range of residence times (0.5e2 days) examined in this study. Batch results showed complete dechlorination occurred after 19 days when the OW consortium was provided only sediment effluent, much longer than the 2 day retention times utilized for these column studies. However, at sites where diffusive conditions exist, or where groundwater seepage rates are significantly slower than those employed here, complete dechlorination could be achieved. For instance, at the USGS biomat pilot test described by Majcher et al. (2009), a bioreactive layer successfully dechlorinated a range of
5372
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
0
B
5
5 mM Lactate Rt = 0.47 day
5 mM Lactate Rt = 1 day
R = 2 days
5 mM Lactate Rt = 2 days
5 mM Lactate
DCE to influent
Bioreactive Cap Experimental Conditions
A
10
No Lactate R = 0.47 day
15
20
25
30
35
Product Distribution Normalized to Total Chloroethenes + Ethene Eluted
Bioreactive Sand Cap Effluent Product Distribution 1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
DCE VC ETH Ethane
0.0 0
5
10
15
20
25
30
35
Pore Volumes Eluted from Cap
C Cummulative Methane (mg/L)
Bioreactive Sand Cap Effluent Cummulative Methane 80
80
60
60
40
40
20
20
0
0 0
5
10
15
20
25
30
35
Pore Volumes Eluted from Cap Fig. 4 e AeB. (A) Operating conditions for Bioreactive Cap B. The effluent of the sediment column served as the influent of the Bioreactive Cap and was not amended with minerals nor vitamins. Exogenous electron donor (lactate), carbon sources (lactate), and electron acceptor (cis-DCE) were added where indicated. (B) Effluent product distribution of Bioreactive Cap B inoculated with a PCE-to-ethene mixed dechlorinating consortia and connected in series with sediment column effluent between 146 and 185 sediment pore volumes. (C) Cumulative aqueous methane concentration in samples collected from Bioreactive Cap B effluent.
chlorinated aliphatics at a site where average hydraulic residence times in the reactive mat were assumed to be 8e14 days. Thissystem also included an organic layer composed of a mixture of peat, compost, and chitin to provide long-term electron donor.
4.
Conclusions
Based on the results presented herein, Engineered controls may be needed to maintain microbial dechlorination activity, reduce contaminant flux, or increase contaminant residence time for bioreactive caps to achieve
complete reductive dechlorination of dissolved chlorinated ethenes to ethene. Incorporation of electron donor was required to stimulate and sustain long-term contaminant biotransformations in a bioreactive cap under the conditions tested. At sites with lower seepage velocities, allowing for greater residence time in the cap, complete dechlorination without electron donor may be possible. The need for electron donor delivery in bioactive design could support greater cell numbers of degrading populations, resulting in greater degradation rates and possible deployment at sites with reasonably high contaminant flux (e.g, high concentrations, high flow rates). This is supported
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
by data in this study during Bioreactive Cap B, where the addition of electron donor, albeit at relatively high concentrations, supported complete dechlorination under relative short residence times (1e2 days). Careful attention should be provided to accurately characterize seepage rates and contaminant concentrations at sites where contaminated groundwater seeps are present. In summary, this study examined the conditions governing the implementation of novel subaqueous bioreactive in situ caps. Experimental results suggest that the process is feasible provided that sufficient electron donor and contaminant mass fluxes exist in the bioactive cap.
Acknowledgments Funding for this research was provided by the Hazardous Substance Research Center-South and Southwest, the National Institute of Environmental Health Sciences, and a fellowship to D.W.H from the Georgia Institute of Technology.
Appendix. Supplementary information Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.06.022.
references
Abe, Y., Aravena, R., Zopfi, J., Parker, B., Hunkeler, D., 2009. Evaluating the fate of chlorinated ethenes in streambed sediments by combining stable isotope, geochemical and microbial methods. Journal of Contaminant Hydrology 107 (1e2), 10e21. Adamson, D.T., Newell, C.J., 2009. Support of source zone bioremediation through endogenous biomass decay and electron donor recycling. Bioremediation Journal 13 (1), 29e40. Anderson, R.T., Vrionis, H.A., Ortiz-Bernad, I., Resch, C.T., Long, P.E., Dayvault, R., Karp, K., Marutzky, S., Metzler, D.R., Peacock, A., White, D.C., Lowe, M., Lovley, D.R., 2003. Stimulating the in situ activity of Geobacter species to remove uranium from the groundwater of a uranium-contaminated aquifer. Applied and Environmental Microbiology 69 (10), 5884e5891. Carr, C.S., Hughes, J.B., 1998. Enrichment of high rate PCE dechlorination and comparative study of lactate, methanol, and hydrogen as electron donors to sustain activity. Environmental Science & Technology 32 (12), 1817e1824. Cho, Y.-M., Smithenry, D.W., Ghosh, U., Kennedy, A.J., Millwarde, R.N., Bridges, T.S., Luthy, R.G., 2007. Field methods for amending marine sediment with activated carbon and assessing treatment effectiveness. Marine Environmental Research 64 (5), 541e555. Choi, H., Agarwal, S., Al-Abed, S.R., 2009. Adsorption and simultaneous dechlorination of PCBs by GAC impregnated with ZVI/Pd bimetallic particles: mechanistic aspects and reactive capping barrier concept. Environmental Science & Technology 43 (2), 488e493. Conant, B., Cherry, J.A., Gillham, R.W., 2004. A PCE groundwater plume discharging to a river: influence of the streambed and
5373
near-river zone on contaminant distributions. Journal of Contaminant Hydrology 73 (1e4), 249e279. Daprato, R.C., Lo¨ffler, F.E., Hughes, J.B., 2006. Comparative analysis of three tetrachloroethene to ethene halorespiring consortia suggests functional redundancy. Environmental Science & Technology 41 (7), 2261e2269. Dries, J., Bastiaens, L., Springael, D., Agathos, S.N., Diels, L., 2004. Competition for sorption and degradation of chlorinated ethenes in batch zero-valent iron systems. Environmental Science & Technology 38 (10), 2879e2884. Go, J.L., Lampert, D.J., Stegemann, J.A., Reible, D.D., 2009. Predicting contaminant fate and transport in sediment caps: mathematical modeling approaches. Applied Geochemistry. 24 (7), 1347e1353. Haas, J.E., Trego, D.A., 2001. A field application of hydrogenreleasing compound (HRC) for the enhanced bioremediation of methyl tertiary butyl ether (MTBE). Soil & Sediment Contamination 10 (5), 555e575. Hamonts, K., Kuhn, T., Maesen, M., Bronders, J., Lookman, R., Kalka, H., Diels, L., Meckenstock, R.U., Springael, D.D., Dejonghe, W., 2009. Factors determining the attenuation of chlorinated aliphatic hydrocarbons in eutrophic river sediment impacted by discharging polluted groundwater. Environmental Science & Technology 43 (14), 5270e5275. Himmelheber, D.W., Pennell, K.D., Hughes, J.B., 2007. Natural attenuation processes during in situ capping. Environmental Science & Technology 41 (15), 5306e5313. Himmelheber, D.W., Taillefert, M., Pennell, K.D., Hughes, J.B., 2008. Spatial and temporal evolution of biogeochemical processes following in situ capping of contaminated sediments. Environmental Science & Technology 42, 4113e4120. Himmelheber, D.W., Thomas, S.H., Lo¨ffler, F.E., Taillefert, M.T., Hughes, J.B., 2009. Microbial colonization of an in situ sediment cap and correlation to stratified redox zones. Environmental Science & Technology 43, 66e74. Hyun, S., Jafvert, C.T., Lee, L.S., Rao, P.S.C., 2006. Laboratory studies to characterize the efficacy of sand capping a coal tarcontaminated sediment. Chemosphere 63 (10), 1621e1631. Jacobs, P.H., Fo¨rstner, U., 1999. Concept of subaqueous capping of contaminated sediments with active barrier systems (ABS) using natural and modified zeolites. Water Research 33 (9), 2083e2087. Krumins, V., Park, J.-W., Son, E.-K., Rodenburg, L.A., Kerkhof, L.J., Ha¨ggblom, M.M., Fennell, D.E., 2009. PCB dechlorination enhancement in Anacostia River sediment microcosms. Water Research 43 (18), 4549e4558. Lendvay, J.M., Dean, S.M., Adriaens, P., 1998. Temporal and spatial trends in biogeochemical conditions at a groundwatersurface water interface: implications for natural bioattenuation. Environmental Science & Technology 32 (22), 3472e3478. Lendvay, J.M., Lo¨ffler, F.E., Dollhopf, M., Aiello, M.R., Daniels, G., Fathepure, B.Z., Gebhard, M., Heine, R., Helton, R., Shi, J., Krajmalnik-Brown, R., Major, C.L., Barcelona, M.J., Petrovskis, E., Hickey, R., Tiedje, J.M., Adriaens, P., 2003. Bioreactive barriers: a comparison of bioaugmentation and biostimulation for chlorinated solvent remediation. Environmental Science & Technology 37 (7), 1422e1431. Lo¨ffler, F.E., Edwards, E.A., 2006. Harnessing microbial activities for environmental cleanup. Current Opinion in Biotechnology 17 (3), 274e284. Lorah, M.M., Voytek, M.A., 2004. Degradation of 1,1,2,2tetrachloroethane and accumulation of vinyl chloride in wetland sediment microcosms and in situ porewater: biogeochemical controls and associations with microbial communities. Journal of Contaminant Hydrology 70 (1e2), 117e145. Lorah, M.M., Majcher, E.H., Jones, E.J., Voytek, M.A., 2008. Microbial Consortia Development and Microcosm and Column
5374
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 6 5 e5 3 7 4
Experiments for Enhanced Bioremediation of Chlorinated Volatile Organic Compounds, West Branch Canal Creek Wetland Area, Aberdeen Proving Ground, Maryland. USGS Scientific Investigations Report. 2007-5165, pp. 1e79. Lowry, G.V., Johnson, K.M., 2004. Congener specific PCB dechlorination by microscale and nanoscale zero-valent iron in a methanol/water solution. Environmental Science & Technology 38 (19), 5208e5216. Majcher, E.H., Lorah, M.M., Phelan, D.J., McGinty, A.L., 2009. Design and performance of an enhanced bioremediation pilot test in a tidal wetland seep, West Branch Canal Creek, Aberdeen Proving Ground, Maryland. USGS Scientific Investigations Report 2009-5112, 1e70. Majcher, E.H., Phelan, D.J., Lorah, M.M., McGinty, A.L., 2007. Characterization of Preferential Ground-water Seepage from a Chlorinated Hydrocarbon-contaminated Aquifer to West Branch Canal Creek, Aberdeen Proving Ground, Maryland, 20022004. USGS Scientific Investigations Report. 2006-5233, pp. 1e193. McDonough, K.M., Murphy, P., Olsta, J., Zhu, Y.W., Reible, D., Lowry, G.V., 2007. Development and placement of a sorbentamended thin layer sediment cap in the Anacostia River. Soil & Sediment Contamination 16 (3), 313e322. Murphy, P., Marquette, A., Reible, D., Lowry, G.V., 2006. Predicting the performance of activated carbon-, coke-, and soilamended thin layer sediment caps. Journal of Environmental Engineering e ASCE 132 (7), 787e794. Reible, D., 2005. Personal communication.
Reible, D., Hayes, D., Lue-Hing, C., Patterson, J., Bhowmik, N., Johnson, M., Teal, J., 2003. Comparison of the long-term risks of removal and in situ management of contaminated sediments in the Fox River. Soil & Sediment Contamination 12 (3), 325e344. Reible, D.D., Lampert, D., Constant, D.W., Mutch, R.D., Zhu, Y., 2007. Active capping demonstration in the Anacostia River, Washington, DC. Remediation 17 (1), 39e53. Scow, K.M., Hicks, K.A., 2005. Natural attenuation and enhanced bioremediation of organic contaminants in groundwater. Current Opinion in Biotechnology 16 (3), 246e253. SERDP and ESTCP, 2008. Expert Panel Workshop on Research and Development Needs for Understanding and Assessing the Bioavailability of Contaminants in Soils and Sediments. Thoma, G.J., Reible, D.D., Valsaraj, K.T., Thibodeaux, L.J., 1993. Efficiency of capping contaminated sediments in situ. 2. Mathematics of diffusion adsorption in the capping layer. Environmental Science & Technology 27 (12), 2412e2419. van Genuchten, M., 1981. Non-equilibrium Transport Parameters from Miscible Displacement Experiments. Research Report 119, U.S. Salinity Lab, USDA, pp. 1e94. Zimmerman, J.R., Ghosh, U., Millward, R.N., Bridges, T.S., Luthy, R.G., 2004. Addition of carbon sorbents to reduce PCB and PAH bioavailability in marine sediments: physicochemical tests. Environmental Science & Technology 38 (20), 5458e5464.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 7 5 e5 3 8 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Preparation of ion exchanger layered electrodes for advanced membrane capacitive deionization (MCDI) Ju-Young Lee, Seok-Jun Seo, Sung-Hyun Yun, Seung-Hyeon Moon* School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), 261 Cheomdan-gwagiro, Buk-gu, Gwangju 500-712, Republic of Korea
article info
abstract
Article history:
A noble electrode for capacitive deionization (CDI) was prepared by embedding ion
Received 18 February 2011
exchanger onto the surface of a carbon electrode to practice membrane capacitive deion-
Received in revised form
ization (MCDI). Bromomethylated poly (2, 6-dimethyl-1, 4-phenylene oxide) (BPPO) was
30 April 2011
sprayed on carbon cloth followed by sulfonation and amination to form cation exchange
Accepted 22 June 2011
and anion exchange layers, respectively. The ion exchange layers were examined by
Available online 3 July 2011
Scanning electron microscopy (SEM) and Fourier transform infrared spectrometer (FT-IR). The SEM image showed that the woven carbon cloth was well coated and connected with
Keywords:
BPPO. The FT-IR spectrum revealed that sulfonic and amine functional groups were
Capacitive deionization (CDI)
attached on the cationexchange and anionexchange electrodes, respectively. The advan-
BPPO
tages of the developed carbon electrodes have been successively demonstrated in a batch
Carbon cloth
and a continuous mode CDI operations without ion exchange membranes for salt removal
Spraying
using 100 mg/L NaCl solution. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Recently, capacitive deionization (CDI) has been used as a water treatment technology due to the simple principle and low operating potential, without the need for chemicals (Ito et al., 2007). Therefore, it is considered as an environmentally-friendly and economical system (Foo and Hameed, 2009). A CDI system operation consists of adsorption and desorption periods for obtaining purified water and concentrated water, respectively (Lee et al., 2010). When an electric potential is applied to CDI cells, charged ions in contaminant water are adsorbed onto the surface of charged electrodes, and formed an electric double layer due to the charged electrode and adsorbed ions, producing purified water. After the adsorption of ions, the saturated electrode undergoes regeneration by desorption of the adsorbed ions under zero electrical potential or reversed electric field (Seo
et al., 2010). Hereby utilization of the adsorption ability of an electrode is the key parameter for the CDI operation. In order to maintain acceptable operation efficiencies, the complete adsorption and desorption of charged ions should be accomplished within appropriate periods. Practically, however, when a potential is applied to a CDI cell, counter ions are attracted onto the electrode surface, simultaneously co-ions expelled from the counter electrode (Kim and Choi, 2010). It leads to a higher energy consumption and a lower operation efficiency due to mobility of unwanted ions. To avoid this phenomenon, a membrane-CDI (MCDI) is employed with the help of ion selective membranes in the CDI cell. A MCDI has two types of ion exchange membranes, i.e. anion exchange and cation exchange membranes (AEM & CEM, respectively). The AEM and CEM are positioned in front of the positively and negatively charged electrodes, respectively (Lee et al., 2006). The ion exchange membrane has the
* Corresponding author. Fax: þ82 62 715 2434. E-mail address:
[email protected] (S.-H. Moon). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.028
5376
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 7 5 e5 3 8 0
ability to selectively permeate ions, i.e., a CEM permit the passage of cations only, while an AEM allow the passage of anions only. The selectivity of ion exchange membranes prevent reverse adsorption and prohibit the mobility of unwanted ions. However, a MCDI requires strong physical pressure for smooth contact between the membrane and electrode material. Also the diffusion layer on the membrane surface will become thick when the concentration of contaminants in the feed water is low (Dlugolecki et al., 2010). Accordingly, this phenomenon induces a decreased mobility of wanted ions and a high interfacial resistance of the membrane. In this study to solve the problems of contact resistance and the diffusion layer, an advanced-MCDI (A-MCDI) was developed by adhering the ion exchanger onto the carbon electrode surface as a thin layer, which reduces the contact resistance between the ion exchanger and electrode of a MCDI. Therefore, an A-MCDI is expected to exhibit a high removal efficiency and a low current consumption compared to a conventional MCDI.
The amination of the BPPO embedded electrode was conducted by immersion of the electrode in 25% TMA for 20 min. The embedded electrode was maintained in distilled water (Tang et al., 2005).
2.3.
Characterization of the embedded electrodes
The surface morphology of the BPPO embedded electrodes were characterized using Scanning electron microscopy (SEM, JEOL, Japan). Both the bare and embedded carbon electrodes were scanned to compare the morphologies before and after coating with the BPPO slurry, at magnifications of 2,000 and 10,000 times. The sulfonated and aminated electrodes were examined using Fourier transform infrared spectrometry (FT-IR, 460 Plus, Jasco Japan) within the wavelength range 400e4000 cm1. FT-IR is used to confirm the functional groups via the bond vibration and stretching energies between atoms. The electrode with the base polymer, sulfonated and aminated electrodes were analyzed to compare the peaks corresponding the functionalization.
2.
Experimental
2.4.
Salt removal test
2.1.
Materials
2.4.1.
Comparison of the CDI, MCDI and A-MCDI
Bromomethylated poly (2, 6-dimethyl-1, 4-phenylene oxide) (BPPO) was used as the base polymer, and N-methyl-2pyrrolidone (NMP, Fluka, Japan) was purchased as the solvent to dissolve BPPO. Carbon cloth (Kuraray, Japan), 3 cm 8.5 cm in size, was employed as the carbon electrode material. The BPPO embedded electrode was sulfonated using sulfuric acid (H2SO4 99%, DC chemical, Korea), while amination was performed using trimethylamine (TMA, 25wt % in D.I.water, Aldrich). In the salt removal test, sodium chloride (NaCl, Dongyang Chemical, Korea) was used to prepare the feed water.
2.2. Preparation of the ion exchanger embedded electrodes Embedding the ion exchanger onto the electrode surface was carried out in two steps; coating the carbon electrode with a base polymer and then attaching functional groups onto the polymer. BPPO (donated by laboratory of fundamental membranes in USTC) slurry was prepared by mixing 1 g of BPPO with 5 ml of NMP at room temperature for 1 day. The BPPO slurry was then sprayed onto the surface of one side of the carbon cloth using an air brush. In order to evaporate the NMP, the electrode was dried in an oven at 40 C for 12 h, forming the BPPO coated electrodes for sulfonation and amination. The sulfonation of the BPPO embedded electrode was conducted using four different solution concentrations, i.e. 99, 80, 50 and 30% sulfuric acid solutions. The BPPO embedded carbon cloth was initially immersed in 99% sulfuric acid solution for 20 min, and subsequently moved to each of the 80, 50 and 30% sulfuric acid solutions for 1 min. The embedded electrode was finally maintained in distilled water (Liu et al., 2006).
To compare the salt removal efficiency of the A-MCDI, experiments were performed using 3 cm 8.5 cm CDI, MCDI and A-MCDI unit cells, with simultaneous recording of the ion conductivity and current in a batch system (Brose´us et al., 2009). Fifty mililiters of feed water was continuously circulated by a pump (Masterflex Cole-parmer, USA). The initial conductivity of feed water, 100 mg/L sodium chloride, was 190 mS/cm. The variation in the ion conductivity was measured every 30 s in the reservoir using a TDS conductivity meter (OAKTON, Japan). A potential of 1.8 V was applied to the CDI cells using an Agilent 6613C (Agilent, USA) power supply. Feed water was continuously circulated at a flow rate of 4 ml/min, which was determined with respect to the surface area of the electrode. Fig. 1 shows schematic diagrams of the assembled CDI, MCDI and A-MCDI cells. In the structure of the CDI cell, the electrodes were located between the current collectors; whereas, in case of the MCDI cell, the positively and negatively charged electrodes were located behind the anion exchange (AMX, Tokuyama, Japan) and cation exchange membranes (CMX, Tokuyama, Japan), respectively. In the A-MCDI cell, the sulfonated BPPO embedded electrode was positioned in front of negatively charged current collector, while the aminated BPPO embedded electrode was positioned in front of positively charged current collector. The arrangement of the ion exchangers in A-MCDI is the same as MCDI, because even though the ion exchangers have the ion exchange capacity for the selected ions, the system still requires electrical potential as a driving force between the cathode and anode. The three systems were compared in terms of their removal efficiencies and energy consumptions.
2.4.2.
Cyclic testing of the A-MCDI
Cyclic testing of the A-MCDI was performed to demonstrate its ability of repeated operation in a continuous mode (Dai et al.,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 7 5 e5 3 8 0
5377
Fig. 1 e Schematics of the CDI, MCDI and A-MCDI cell structures.
2005). The experiment was carried out under the same conditions as the previous salt removal test and subjected to repeated periods of adsorption and desorption. The adsorption and desorption periods were determined to be 5 and 5 min, respectively, to allow sufficient time for adsorption and desorption.
3.
Results and discussion
3.1.
Morphologies of the electrodes
Fig. 2 shows the SEM images of the surfaces of the bare electrode and ion exchanger embedded electrodes. The left and right sides of the figure are the bare carbon cloth and ion exchanger embedded electrode surfaces, respectively. Each sample was magnified by 2,000 and 10,000 times, respectively. Originally the woven carbon cloth material was complicatedly interlinked (Ahn et al., 2007). The bare carbon
electrode surface is observed to have clean surface, with only a tiny amount of dust, whereas, in case of the ion exchanger embedded electrode, the polymer is filled within the carbon cloth and interconnects the individual fibers. In a higher magnification, the bare electrode shows porous carbon cloth surface. However, the ion exchanger embedded electrode shows the pores blocked by the base polymer. Consequently, it was found that the ion exchanger layer was well formed on the surface of the carbon cloth, so that the ion-conducting surface of the embedded electrode may provide good ion selectivity and enhanced conductivity.
3.2.
FT-IR analyses
To impart ion exchange capabilities, the sulfonic and amine functional groups should be attached to the base polymer. The sulfonic and amine groups have the abilities to transport cations and anions, respectively in accordance with the Grotthuss mechanism (Agmon, 1995) and vehicle mechanism
Fig. 2 e SEM images of (a) bare carbon cloth surface at 2,000 times, (b) embedded carbon electrode surface at 2,000 times, (c) bare carbon cloth surface at 10,000 times, (d) embedded carbon cloth surface at 10,000 times.
5378
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 7 5 e5 3 8 0
Fig. 3 e FT-IR analyses of the untreated, sulfonated and aminated electrode surfaces.
(Schuster et al., 2008). In order to examine the sulfonic and amine groups on the BPPO surface, three types of electrodes prepared with the base polymer, sulfonated and aminated polymers were analyzed using FT-IR. Fig. 3 shows the existence of functional groups, indicated by the unique peaks in the sulfonated and aminated BPPO. When the sulfonated BPPO was compared with the untreated BPPO, the sulfonic group peak appeared around 1165e1120 cm1 (Panicker et al., 2006). Similarly, when the aminated BPPO was compared with the untreated BPPO, the amine group peak appeared around 850e750 cm1 (Volkov et al., 1980). The results show that the functional groups were successfully attached to the BPPO.
3.3.
Salt removal test
3.3.1.
Comparison of the CDI, MCDI and A-MCDI
CDI, MCDI, and A-MCDI tests were performed to observe the initial adsorption capability of each system. The results
Fig. 4 e Variations in the ion conductivity of the CDI, membrane-CDI (MCDI) and advanced-MCDI (A-MCDI) during their operation.
obtained for the CDI, MCDI and A-MCDI were compared in terms of salt removal and current consumption, as shown in Figs. 4 and 5, respectively, at an applied potential of 1.8 V and flow rate of 4 ml/min over a 30 min period. As shown in Fig. 4, the CDI and A-MCDI exhibited higher degrees of removal than the MCDI. It means that the removal efficiency of MCDI was lower than the CDI and A-MCDI at the same voltage, because the MCDI suffered problems, such as contact resistance and membrane resistance. Fig. 5 shows that the MCDI and A-MCDI allowed lower currents than that of the CDI. This phenomenon is due to the fact that the current depends on the total ionic flux. In other words, the ion selectivity increases the ion mobility resistance at a constant voltage; therefore, a high resistance will decrease the current in accordance with Ohm’s law (I ¼ V/R) (Kim and Choi, 2010). The operation parameters, the ion removal efficiency and power consumption of each system, are listed in Table 1. These results were obtained based on the measured voltages and currents at 30 min after starting the operation.
Fig. 5 e Operating currents of the CDI, MCDI and A-MCDI.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 7 5 e5 3 8 0
Table 1 e Numerical results of the ion removal efficiencies, currents and power consumptions during the operation of each system.
Applied electric potential Initial conductivity Conductivity at 30 min Removal efficiency Current at 30 min Power consumption
CDI
MCDI
A-MCDI
1.8 V 190.8 mS/cm 39.9 mS/cm 79.1% 235.7 mA 212.3 mWh
1.8 V 185.1 mS/cm 168 mS/cm 9.23% 2.575 mA 2.318 mWh
1.8 V 188.9 mS/cm 31.5 mS/cm 83.4% 25.64 mA 23.07 mWh
The variation in the feed water conductivity was measured to get the removal efficiency, h which was determined using the following equation (Dermentzis and Ouzounis, 2008): hð%Þ ¼
Ci C 100 Ci
(1)
where Ci is the initial conductivity [mS/cm] of the feed water and C is the conductivity of the diluted feed water [mS/cm] at 30 min after starting the operation. According to Eq. (1), the removal efficiencies of the CDI, MCDI and A-MCDI were approximately 79, 9.2 and 83%, respectively. Obviously, the removal efficiency of A-MCDI was the highest. The power consumption of each system was calculated based on the current measured at the same applied potential according to the following equation (Masiuk, 1999): E ¼ VIt
(2)
where E is the power consumption [mWh], V the applied potential [V], I the current [A] and t the time [h]. As a result, the power consumptions of the CDI, MCDI and A-MCDI were approximately 210, 2.3 and 23 mWh, respectively. Considering the removal efficiency and energy consumption, the A-MCDI showed better performance among the systems tested.
3.3.2.
5379
concentration of the charged ions in the effluent was reduced. When the saturated electrode underwent regeneration with zero electrical potential, the adsorbed ions were desorbed. Accordingly, the charged ions in the effluent increased again. The results showed that the repeated adsorption and desorption occurred in a regular pattern of a CDI system. This implies that the A-MCDI system would have stable performance over repeated operation; thus, demonstrated the practical applicability as a water treatment system.
4.
Conclusion
An A-MCDI has been developed by introducing an ion exchanger layer on electrode surface for a CDI system. The electrodes were prepared by coating a base polymer on the carbon cloth followed by functionalization of sulfonic and amine groups on the polymer structure. The noble electrode enables to overcome the drawbacks of a membrane-CDI system by reducing the interfacial resistance between the ion exchanger layer and the carbon electrode. Practically the A-MCDI is operated without ion exchange membranes while the system performs better than a conventional MCDI in terms of salt removal efficiency. This research may contribute significantly in application of CDI in various water treatment systems such as desalination, hardness removal, and drinking water treatment.
Acknowledgment This research was supported by a grant (07seaheroB02-02-01) from the Plant Technology Advancement Program, funded by the Ministry of Land, Transport and Maritime Affairs.
Cyclic testing of the A-MCDI
A cyclic test was performed for a continuous operation of AMCDI. Fig. 6 shows the variation of the ion conductivity in the effluent stream. During the initial adsorption period, charged ions in feed water were adsorbed onto the electrode; therefore, the
Fig. 6 e Continuous mode operation of the A-MCDI system by repeated cycles of adsorption and desorption.
references
Agmon, N., 1995. The Grotthuss mechanism. Chemical Physics Letters 244, 456e462. Ahn, H.-J., Lee, J.-H., Jeong, Y., Lee, J.-H., Chi, C.-S., Oh, H.-J., 2007. Nanostructured carbon cloth electrode for desalination from aqueous solutions. Materials Science and Engineering: A 449e451, 841e845. Brose´us, R., Cigana, J., Barbeau, B., Daines-Martinez, C., Suty, H., 2009. Removal of total dissolved solids, nitrates and ammonium ions from drinking water using charge-barrier capacitive deionisation. Desalination 249, 217e223. Dai, K., Shi, L., Fang, J., Zhang, D., Yu, B., 2005. NaCl adsorption in multi-walled carbon nanotubes. Materials Letters 59, 1989e1992. Dermentzis, K., Ouzounis, K., 2008. Continuous capacitive deionization-electrodialysis reversal through electrostatic shielding for desalination and deionization of water. Electrochimica Acta 53, 7123e7130. Dlugolecki, P., Anet, B., Metz, S.J., Nijmeijer, K., Wessling, M., 2010. Transport limitations in ion exchange membranes at low salt concentrations. Journal of Membrane Science 346, 163e171. Foo, K.Y., Hameed, B.H., 2009. A short review of activated carbon assisted electrosorption process: an overview, current stage and future prospects. Journal of Hazardous Materials 170, 552e559.
5380
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 7 5 e5 3 8 0
Ito, E., Mozia, S., Okuda, M., Nakano, T., Toyoda, M., Inagaki, M., 2007. Nanoporous carbons from cypress II. Application to electric double layer capacitors. New Carbon Materials 22, 321e326. Kim, Y.-J., Choi, J.-H., 2010. Enhanced desalination efficiency in capacitive deionization with an ion-selective membrane. Separation and Purification Technology 71, 70e75. Lee, J.-B., Park, K.-K., Eum, H.-M., Lee, C.-W., 2006. Desalination of a thermal power plant wastewater by membrane capacitive deionization. Desalination 196, 125e134. Lee, J.Y., Seo, S.J., Park, J.W., Moon, S.H., 2010. A study on the cell structure for capacitive deionization system. Korean Chemical Engineering Research 48, 791e794. Liu, J., Xu, T., Han, X., Fu, Y., 2006. Synthesis and characterizations of a novel zwitterionic hybrid copolymer containing both sulfonic and carboxylic groups via sulfonation and zwitterionic process. European Polymer Journal 42, 2755e2764. Masiuk, S., 1999. Power consumption measurements in a liquid vessel that is mixed using a vibratory agitator. Chemical Engineering Journal 75, 161e165.
Panicker, C.Y., Varghese, H.T., Philip, D., Nogueira, H.I.S., 2006. FT-IR, FT-Raman and SERS spectra of pyridine-3-sulfonic acid. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 64, 744e747. Schuster, M., Kreuer, K.-D., Steininger, H., Maier, J., 2008. Proton conductivity and diffusion study of molten phosphonic acid H3PO3. Solid State Ionics 179, 523e528. Seo, S.-J., Jeon, H., Lee, J.K., Kim, G.-Y., Park, D., Nojima, H., Lee, J., Moon, S.-H., 2010. Investigation on removal of hardness ions by capacitive deionization (CDI) for water softening applications. Water Research 44, 2267e2275. Tang, B., Xu, T., Gong, M., Yang, W., 2005. A novel positively charged asymmetry membranes from poly(2,6-dimethyl-1,4phenylene oxide) by benzyl bromination and in situ amination: membrane preparation and characterization. Journal of Membrane Science 248, 119e125. Volkov, A., Tourillon, G., Lacaze, P.-C., Dubois, J.-E., 1980. Electrochemical polymerization of aromatic amines: IR, XPS and PMT study of thin film formation on a Pt electrode. Journal of Electroanalytical Chemistry 115, 279e291.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Electrochemical sulfide oxidation from domestic wastewater using mixed metal-coated titanium electrodes Ilje Pikaar, Rene´ A. Rozendal, Zhiguo Yuan, Ju¨rg Keller, Korneel Rabaey* The University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia
article info
abstract
Article history:
Hydrogen sulfide generation is a major issue in sewer management. A novel method based
Received 8 April 2011
on electrochemical sulfide oxidation was recently shown to be highly effective for sulfide
Received in revised form
removal from synthetic and real sewage. Here, we compare the performance of five
11 July 2011
different mixed metal oxide (MMO) coated titanium electrode materials for the electro-
Accepted 25 July 2011
chemical removal of sulfide from domestic wastewater. All electrode materials performed
Available online 6 August 2011
similarly in terms of sulfide removal, removing 78 5%, 77 1%, 85 4%, 84 1%, and 83 2% at a current density of 10 mA/cm2 using Ta/Ir, Ru/Ir, Pt/Ir, SnO2 and PbO2,
Keywords:
respectively. Elevated chloride concentrations, often observed in coastal areas, did not
Electrochemical oxidation
entail any significant difference in performance. Independent of the electrode material
Oxygen generation
used, sulfide oxidation by in situ generated oxygen was the predominant reaction mech-
Sulfide oxidation
anism. Passivation of the electrode surface by deposition of elemental sulfur did not occur.
Sewer
However, scaling was observed in the cathode compartment. This study shows that all the MMO coated titanium electrode materials studied are suitable anodic materials for sulfide removal from wastewater. Ta/Ir and Pt/Ir coated titanium electrodes seem the most suitable electrodes since they possess the lowest overpotential for oxygen evolution, are stable at low chloride concentration and are already used in full scale applications. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Hydrogen sulfide is ubiquitously present in industrial and domestic waste streams, and is of special concern in sewer systems since it is responsible for odor issues in urban areas, is toxic to sewer workers, and is the main cause for sewer pipe corrosion (Zhang et al., 2008). Repair and/or replacement of corroded sewer pipes results in considerable costs (Sydney et al., 1996; Vincke et al., 2002; Kaempfer and Berndt, 1998), and therefore measures for mitigating sulfide production and emission are required. Current strategies to prevent sewer corrosion come with substantial costs due to both chemical consumption and system maintenance (Zhang et al., 2008; Hvitved-Jacobsen, 2001).
Recently, we described the electrochemical oxidation of aqueous sulfide using Ta/Ir coated titanium electrodes from domestic wastewater (Pikaar et al., 2011). At the used current densities of 5 mA/cm2 sulfide could be oxidized, producing elemental sulfur, thiosulfate and sulfate as the final products. Indirect oxidation of sulfide with in situ generated oxygen rather than the direct oxidation of sulfide at the electrode was shown to be the predominant reaction mechanism due to the low sulfide concentrations (i.e. w10 mg/L) normally observed in sewers. In comparison to sulfide control with conventional oxygen injection, the method does not require any transport and storage of oxygen, whereas in situ generated oxygen is expected to have a much higher efficiency due to the fine dispersion (i.e. 1e30 mm) of the generated oxygen (Chen, 2004).
* Corresponding author. Tel.: þ61 7 3365 7519; fax: þ61 7 3365 4726. E-mail address:
[email protected] (K. Rabaey). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.07.033
5382
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
Electrode materials possess different selectivity and catalytic activities, and may hence produce different intermediates such as oxygen, chlorine/hypochlorous acid, hydroxyl radicals and other reactive oxygen species (Chen, 2004). These intermediates may have a different product spectrum for sulfide oxidation, and may increase/decrease overall oxidation efficiency. In this study, the feasibility of anodic sulfide oxidation in domestic wastewater using titanium electrodes coated with 5 different types of electrocatalyst materials, namely Ta/Ir, Pt/Ir, Ru/Ir, PbO2 and SnO2, was investigated. Ta/Ir coated titanium electrodes are known for their low overpotential for oxygen and their stability. They are therefore commonly used as oxygen evolving electrodes (Chen, 2004). Ru/Ir electrodes find widespread use in the chloro-alkali industry for the in situ oxidation of chloride to chlorine/ hypochlorite (Feng and Li, 2003; Takasu et al., 2010). Both Ru/Ir and Pt/Ir electrodes also possess a low overpotential for oxygen evolution, and therefore especially at low chloride concentrations oxygen evolution may become a predominant reaction mechanism. It is important to note that these electrodes have a low overpotential for the production of chlorine from chloride. In coastal areas (e.g., Queensland, Australia) the chloride concentrations in domestic wastewater can be a high as >1 g/L due to marine infiltrations into the sewer system (Taylor and Gardner, 2007). Hence, in situ generation of hypochlorous acid/hypochlorite may become an important reaction mechanism for sulfide oxidation from domestic wastewater. Hypochlorous acid/hypochlorite can oxidize sulfide to elemental sulfur at pH 7.5, values normally observed in sewer systems. Disadvantage of in situ chlorine generation is the possible formation of toxic organochlorine derivatives (Sun et al., 2009), which needs to be prevented. Contrarily to the three aforementioned materials, PbO2 and SnO2 have a high overpotential for oxygen evolution. Just like boron doped diamond (BDD), the electrocatalysts PbO2 and SnO2 are known to generate hydroxyl radicals from the oxidation of water (Panizza and Cerisola, 2008; Zhu et al., 2008; Panizza et al., 2008). However, contrarily to BDD, PbO2 and SnO2 are made of inexpensive materials which are readily available in practical mesh geometries and at scale, and have a low electrical resistivity. This makes them suitable materials for applications on large scale (i.e. industrial applications and sewer systems). Considering the above, the main aims of this study are to assess the impact of electrode coating on electrochemical sulfide oxidation in wastewater, to assess the differences in energy requirement for the different electrode materials and to assess the impact of chloride concentrations on the sulfide oxidation process.
2.
Materials and methods
2.1.
Electrochemical cell and operation
Fig. 1 gives a schematic diagram of the electrochemical cell. The two-chambered electrochemical cell consisted of two parallel Perspex frames (internal dimensions 20 4.8 1.2 cm) separated by a cation exchange membrane (Ultrex CM17000, Membranes International Inc., USA) to create an anode and
cathode compartment each with a volume of about 100 mL. In the anode chamber, mesh shaped Ta/Ir (TaO2/IrO2: 0.35/0.65), Ru/Ir (RuO2/IrO2: 0.70/0.30), Pt/Ir (PtO2/IrO2: 0.70/0.30), PbO2 and SnO2 coated titanium electrodes (diameter: 240 mm; thickness: 1 mm; specific surface area: 1.0 cm2/cm2) were used (Magneto Anodes BV, The Netherlands). Stainless steel fine mesh (24 cm2) with a stainless steel current collector (6 mm mesh size, 0.8 mm wire connected via a 6 mm stainless steel rod) was used as electrode material in the cathode chamber. Both the anode and cathode had a projected electrode surface area of 24 cm2. In all experiments, an Ag/AgCl (RE-5B, Bio Analytical, USA) was used as the reference electrode (þ197 mV versus SHE). The anode liquid medium was constantly recirculated over a 5 L vessel, allowing a total anode liquid volume of 5 L. The influent flow rate through the anode chamber was maintained at 3.6 L/h using a peristaltic pump (Watson Marlow, UK). An additional recirculation flow in the anode chamber, which was kept at 22 L/h using a peristaltic pump (Watson Marlow, UK), to obtain a good mixing rate in the anode chamber was used. PVC tubing with an internal diameter of 8 mm was used for the feeding and recirculation lines. The off-gas coming from the external buffer vessel was captured in a gas collection bag (TKC Tedlar bags, Air-Met Scientific Pty Ltd, USA). An external buffer flask of 2 L was used in the recirculation of the cathode chamber. A 0.10 M NaCl solution in the cathode chamber was used in all experiments. The recirculation flow of the cathode solution was kept at 22 L/h using a peristaltic pump (Watson Marlow, UK). The anode liquid medium, domestic wastewater, was collected weekly from a local pumping station and stored at 4 C. Prior to use, 5 L of the domestic wastewater was heated up to ambient temperatures (24.3 0.5 C) and fed to the influent buffer tank. pH in the influent was measured using a pH probe (Ionode Pty Ltd., AU) and maintained at 7.5 during the experiment through a PLC controlled dosage of a 0.5 M NaOH solution. The wastewater was continuously recirculated. This fed batch system was used to minimize the amount of domestic wastewater needed (continuous feeding would have required significant amounts of wastewater per day). A concentrated sodium sulfide (Na2S$9H2O) stock solution (5.5 g/L sulfide-S), prepared according to Dutta et al. (2008), was continuously supplied to the incoming line of the anode chamber via a syringe pump (NE-1600, New Era Pump Systems, Inc., USA) at a dosing rate of w36 mg sulfide-S/h, which was sufficient to give an anode influent concentration of w10.0 mg S/L by assuming that the recirculated wastewater contained no sulfide.
2.2.
Measurements and calculations
Galvanostatic measurements and control were performed using a Wenking potentiostat/galvanostat (KP07, Bank Elektronik, gmbH, Germany). The anode, cathode potentials and the current were recorded every 60 s using an Agilent 34970A data acquisition unit. The amount of sulfide dosed to the reactor can be expressed as a charge quantity expressed in Coulomb: Q ¼ nFcadded =M
(1)
With F the Faraday constant (96,485 C/mol, n the number of electrons involved (i.e. 8 electrons for the oxidation of 1 mol
5383
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
13 Power supply
14
14
7 12 15
10 2
+
H
+
Na K
3
11
8
+
1 5
Sulfide solution
9
6
4
Fig. 1 e Schematic overview experimental setup. 1. Influent buffer (5 L); 2. Influent anodic compartment; 3. Recirculation flow anodic compartment; 4. Sulfide feeding line; 5. Anode compartment; 6. Cathode compartment; 7. Effluent anode compartment; 8. Influent cathode compartment; 9. Cathode buffer (2 L); 10. Effluent cathode compartment; 11. Cathode water-lock; 12. Cathode vent gas (H2); 13. Potentiostat/galvanostat; 14. Sampling points; 15. Anode vent gas (H2S, O2, CO2).
sulfide to sulfate), c the amount of sulfide added (g) and M the molar weight of sulfide (i.e. 32 g/mol). Hence, by determining the coulombic efficiency (CE) based on the conversion of sulfide to sulfate the coulombic efficiency can be calculated as follows: CE ¼
2.3.
nFcremoved =M cremoved ¼ cadded nFcadded =M
(2)
Chemical analyses
Sulfide, sulfite, thiosulfate and sulfate concentrations were measured by Ion Chromatography (IC), using a Dionex 2010i system, according to Keller-Lehmann et al. (2006). Samples were immediately filtered using a 0.22 mm syringe filter (Millipore, USA) and preserved in previously prepared Sulfide Antioxidant Buffer (SAOB) solution prior to ion chromatography analysis. SAOB solution was prepared using Helium purged MilliQ (18MU) water, 3.2 g/L NaOH and 2.8 g/L aascorbic acid. After preparation, the solution was kept refrigerated, shielded from light and not used beyond 24 h. Elemental sulfur was assumed to be the difference between the total sulfide added and the soluble sulfur species (i.e. sulfide, sulfite, thiosulfate and sulfate). COD (range 25e1500 mg/L) and free chlorine (range 0e4 mg/L) concentrations were determined by means of cuvette tests (Merck, Germany). COD concentrations were corrected for the soluble sulfur species in the solution. The conductivity was measured using a hand-held meter
(Cyberscan PC 300, Eutech Instruments). The produced gas (i.e. in situ oxygen generation) was analyzed using a Gas Chromatography (Shimadzu, Japan).
2.4.
Experimental procedures
Experiments were divided into 2 different sets. The first set of experiments compared the performance of five different mixed metal-coated titanium electrode materials (Ta/Ir, Ru/Ir, Pt/Ir, PbO2, SnO2,). Key focal points were (a) the kinetics of the sulfide oxidation, (b) possible oxidation of organics (i.e. COD), (c) the amount of excess in situ generated oxygen and (d) the required energy input (i.e., based on the obtained cell potential) of the different electrode materials. The second set of experiments investigated the influence of chloride concentrations on (a) the kinetics of the sulfide oxidation, (b) possible oxidation of organics (i.e. COD) and (c) the required energy input (i.e. obtained cell potential) using Ru/Ir and Ta/Ir coated titanium as electrode material. Each time, the performance was assessed during 6-h experimental runs using galvanostatic control at a fixed current density of 10 mA/cm2. This current density level was enough to oxidize all sulfide added to the system to sulfate. All experiments were performed in triplicate. Prior to each experiment, the headspace of the influent buffer vessel was flushed with nitrogen for at least 5 min to obtain a headspace that consisted of 100% nitrogen. In this way, any excess in situ
5384
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
generated oxygen could be measured using Gas Chromatography after every experiment.
3.
Results and discussion
3.1.
Influence of electrode material on sulfide oxidation
The influence of the electrode material on the kinetics of anodic sulfide oxidation was investigated during 6-h experiments at a fixed current density of 10 mA/cm2. During the experimental runs, sulfide concentrations increased to approximately 7e10 mg/L due to the recirculatory mode of operation. The typical profiles of the sulfide concentration are presented in the supplementary information S1. In Table 1 an overview of the results using Ta/Ir, Ru/Ir, Pt/Ir, PbO2 and SnO2 as electrode material is presented. The obtained removal and Coulombic efficiencies using Ta/Ir, Ru/Ir, Pt/Ir, PbO2 and SnO2 were 78 5, 77 1, 85 4, 83 2 and 84 1%, respectively. This is equal to sulfide removal rates of 10.8 0.2, 11.7 1.1, 1 12.4 0.4, 12.4 0.4 and 12.9 0.8 g S m2 electrode surface h . The obtained Coulombic efficiencies were calculated based on the oxidation of sulfide to sulfate (see Equation (2)). The values observed for the sulfide removal rates expressed in mg S/L wastewater h1 7.5 0.3 (Ta/Ir) to 7.8 0.6 (SnO2) are in agreement with the chemical sulfide oxidation rates found under high dissolved oxygen concentrations in domestic wastewater (Sharma and Yuan, 2010). Furthermore, GC analysis of the produced gas at the end of every experiment showed that in all experiments similar amount of excess oxygen was generated and transferred to the headspace. The obtained excess oxygen generation using Ta/Ir, Ru/Ir, Pt/Ir, PbO2 and SnO2 was 65 13, 60 17, 69 8, 66 17 and 47 8 mg (Table 1). In a previous study, we showed that direct
sulfide oxidation at the electrode surface was negligible under the given operational conditions (Pikaar et al., 2011). If reactive oxygen species were primarily responsible for sulfide oxidation, we would have expected sulfate being the primary product of oxidation, contrarily to the results (see Table 1). If oxygen would be the predominant reaction mechanism, a mixture of sulfur species would be expected. The results suggest that in situ oxygen generation is significant and could be primarily responsible for the sulfide oxidation observed. The materials used are known to have different overpotentials for oxygen evolution. This however does not necessarily mean that they will produce different amounts of oxygen. The formation of reactive oxygen species such as OH radicals is intermediate products during the oxidation of water to oxygen. In the first step, adsorbed OH radicals are formed on the electrode surface. In the second step, the adsorbed OH interacts with the oxygen already present in the oxide anode to form physisorbed or chemisorbed ‘active oxygen’. In absence of any oxidizable pollutant this ‘active oxygen’ subsequently produces oxygen (Comninellis, 1994). Thus, the efficiency of the formation of reactive oxygen species does not only depend on the electrode material but also on the concentration of the pollutant and its reactivity toward oxidation. Especially at low pollutant concentrations, low Coulombic efficiencies for the formation of reactive oxygen species have been observed (Martinez-Huitle and Brillas, 2009), which means that high Coulombic efficiencies for oxygen evolution can be expected. Considering the low sulfide (i.e. w10 mg/L) and organics concentrations (i.e. 380 140 mg/L) during the experiments, this could explain the similar sulfide removal and excess oxygen production rates for all electrode materials. It should be noted that in this study a laboratory reactor was used rather than a real rising main system. Hence, our
Table 1 e Sulfide oxidation (n [ 3) from domestic wastewater using MMO coated titanium electrodes at a current density of 10 mA/cm2. Parameter a
Coulombic efficiency Removal efficiency Removal rate Removal rate Total S added (mg) Final sulfide conc. S0 produced S2O2 3 produced SO2 3 produced SO2 4 produced COD removed COD removal rate O2 produced Temperature Conductivity Chloride concentration pHb Average anode potential Average cell voltage
Unit
Ta/Ir
Ru/Ir
Pt/Ir
SnO2
PbO2
% % 1 g S m2 electrode surface h mg S L1 h1 mg mg/L mg mg mg mg mg mg COD h1 mg C mS/cm mg/L e V V
78 5 78 5 10.8 0.2 7.5 0.3 199 9 7.9 2.6 100 9 63 9 0.9 0.2 10 5 199 43 60 17 24.5 0.5 1.11 0.01 114 9 7.5 1.41 0.08 5.3 0.4
77 1 77 1 11.7 1.1 7.7 0.5 219 21 10.0 1.2 97 4 64 13 1.5 0.5 14 15 256 33 65 13 24.1 0.4 1.15 0.01 117 9 7.5 1.42 0.06 5.2 0.4
85 4 85 4 12.4 0.4 7.7 0.2 210 18 6.2 2.3 108 19 52 6 1.1 0.7 34 16 204 34 69 8 24.6 0.4 1.17 0.06 115 9 7.5 1.53 0.12 5.0 0.2
84 1 84 1 12.9 0.8 7.8 0.6 220 12 6.6 0.6 136 25 43 10 0.8 0.5 28 8 274 46 47 8 25.0 0.6 1.16 0.04 109 10 7.5 2.2 0.11 9.0 0.4
83 2 83 2 12.4 0.1 7.8 0.7 214 6 6.9 1.8 101 36 50 8 1.1 0.7 30 17 117 26 66 17 24.3 0.4 1.09 0.09 114 6 7.5 1.8 0.04 5.0 0.2
a Coulombic efficiency is based on the oxidation of sulfide to sulfate. b pH is controlled online.
5385
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
system differed from a real sewer system as such that in our system (a) a headspace was present and (b) in situ generated oxygen was dissolved into only 5 L. Hence, we produced much more oxygen per unit wastewater. The above caused that oxygen was transferred to the headspace instead of remaining in the water phase. The anode potentials remained constant during the course of all the experiments. The anode potentials for Ta/Ir, Ru/Ir and Pt/Ir (low overpotential for oxygen evolution) were 1.41 0.08, 1.42 0.06 and 1.53 0.12 V vs. SHE, whereas the anode potentials for PbO2 and SnO2 were 1.80 0.04 and 2.22 0.11 V vs. SHE, respectively. The overall cell voltage for all electrode materials, except for SnO2 (9.0 0.4 V), was approximately 5 V (Table 1). As yet, we have no explanation for the unexpectedly high cell voltage for SnO2, the use of this electrode apparently caused a higher ohmic resistance or higher cathodic overpotential. The mechanism of the generation of oxygen active species is very complex and can involve the generation of several radical species containing oxygen and/or halogen atom (e.g. C C OC 2 , HO2 , HClO ) other than hydroxyl radicals which can occur at lower potentials. Hence, reactive oxygen species other than hydroxyl radicals might have been formed but did not have a significant impact on the sulfide removal process. Fig. 2 shows the electron distribution among the different electron sinks (i.e. sulfur, thiosulfate, sulfite, sulfate, excess oxygen and COD (i.e. organics)) during the oxidation of sulfide using the different electrode materials. A significant part of the charge supplied to the system was used for the oxidation of organics. The charge used for the oxidation of organics was 46%, 60%, 47%, 64% and 36% for Ta/Ir, Ru/Ir, Pt/Ir, PbO2 and SnO2, respectively. Concentrations of the different dissolved sulfur species (i.e. sulfide, sulfite, thiosulfate and sulfate) indicate that 43 2%,
39 7%, 47 8%, 42 11% and 57 10% of the total sulfide added was oxidized to elemental sulfur using Ta/Ir, Ru/Ir, Pt/ Ir, PbO2 and SnO2, respectively. This is further supported by the reasonably small gap in the electron balances of the experiments (see Fig. 2). No elemental sulfur was visually observed on the electrode surface. This suggests that the elemental sulfur was formed in the bulk by means of indirect oxidation with oxygen. In summary, the results highlight that under the applied operational conditions the sulfide removal process in terms of removal efficiency, excess oxygen generation, overall cell potential (except SnO2) and electron distribution is not significantly influenced by the electrode material used.
3.2. Influence of chloride concentration on sulfide oxidation The impact of elevated chloride concentration was investigated using Ta/Ir and Ru/Ir coated titanium electrodes; electrodes with a low and high reported catalytic activity toward chlorine generation, respectively. Table 2 shows that similar sulfide removal efficiencies and removal rates as in the experiments at low chloride concentrations were obtained (i.e. 7.7 0.5 versus 7.0 1.0 and 7.5 0.3 versus 7.7 0.3 mg S/L h). Fig. 3 shows the electron distribution among the different electron sinks (i.e. sulfur, thiosulfate, sulfite, sulfate, excess oxygen and organics) during the oxidation of sulfide using of Ta/Ir and Ru/Ir electrodes, respectively. Similar to the experiments at low chloride concentrations a large part of the charge supplied to the system was used for the oxidation of organics (i.e., 50 10, 50 8% for Ta/Ir and Ru/Ir, respectively). Ru/Ir coated titanium electrodes are well-known for their low overpotential for chlorine generation. Hence, at high chloride concentrations in situ chlorine generation may have
organics excess oxygen sulfate sulfite thiosulfate sulfur
electron distribution (%)
100
80
60
40
20
0 Ta/Ir
Ru/Ir
Pt/Ir
SnO2
PbO2
Fig. 2 e Electron distribution (%) among the different electron sinks (i.e. sulfur, thiosulfate, sulfite, sulfate, excess oxygen and organics) during the oxidation of sulfide using Ta/Ir, Ru/Ir, Pt/Ir, PbO2 and SnO2 electrodes at a current density of 10 mA/ cm2 (n [ 1).
5386
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
h i JL ¼ nFkm Cl
(3)
where JL is the mass transfer limited current density (mA/ cm2), n is the number of electrons involved (i.e. 1 for the oxidation of chloride to chlorine), F the Faraday constant (96,485.3 C/mol), km the mass transport rate coefficient (m/s) and Cl the chloride concentration (mg/L). Assuming a km of 0.8 105 m/s (Szpyrkowicz et al., 2005) and an average chloride concentration of 114 mg/L (i.e. the chloride concentration present in the sewage wastewater used) the mass transfer limited current is only 0.25 mA/cm2. In addition, a migration current due to the migration in the chloride transport can be expected as result of the low conductivity of the wastewater (Bergmann and Koparal, 2005). However, at the applied current density (10 mA/cm2), the chloride concentration and the applied electrical field the relative importance of the migration current density is small. The applied current density in all experiments is 10 mA/cm2 and thus sulfide oxidation by means of in situ chlorine production is expected to be small. Hence, it is expected that at low chloride concentrations Ru/Ir coated titanium electrodes mainly will result in the in situ generation of oxygen and hence similar kinetics for the anodic sulfide oxidation are expected. However, at high chloride concentrations up to 1100 mg/L, which are often observed in sewers in many coastal areas,
Table 2 e Sulfide oxidation (n [ 3) from domestic wastewater at elevated chloride concentrations using Ta/ Ir and Ru/Ir coated titanium electrodes at a current density of 10 mA/cm2. Parameter Coulombic efficiencya Removal rate Removal rate Total S added (mg) Final sulfide conc. S0 produced S2O2 3 produced SO2 3 produced SO2 4 produced COD removed COD removal rate O2 produced Temperature Conductivity Chloride concentration pH Average anode potential Average cell voltage a n ¼ 2.
Unit %
Ta/Ir
Ru/Ir
87 4
86 5
1 g S m2 electrode surface h mg S L1 h1 mg mg/L mg mg mg mg mg mg COD h1 mg C mS/cm mg/L
12.9 1.0 12.2 1.2 7.7 0.3 7.0 1.0 213 8 203 8 5.4 1.3 5.7 1.8 139 0 103 20 29 14 33 5 0.1 0.1 0.3 0.1 20 12 39 14 213 16 216 33 36 6 37 7 77 19 65 10 24.0 0.5 24 0.4 3.73 0.05 3.73 0.05 1117 37 1119 25
e V
7.5 7.5 1.51 0.18a 1.43 0.03
V
4.21 0.9a
4.64 0.2
70
60
elecetron distribution (%)
an impact on the sulfide oxidation process and might also affect the removal of organics. The maximum chlorine generation under the mass transfer limiting conditions at a planar electrode (under optimum mixing conditions) can be described according to:
50
40
30
20
10
0 sulfur
thiosulfate
sulfite
sulfate
oxygen
COD
Fig. 3 e Electron distribution (in %) among the different electron sinks (i.e. sulfur, thiosulfate, sulfite, sulfate, excess oxygen and COD (i.e. organics)) during the oxidation of sulfide using Ta/Ir and Ru/Ir, electrodes at a current density of 10 mA/cm2 at high chloride concentrations (n [ 3).
chlorine generation can play a significant role in the anodic sulfide removal process. Depending on the pH, sulfide can be oxidized by chlorine either to elemental sulfur or sulfate. At pH values 7.5, sulfide is oxidized by chlorine to elemental sulfur (Chwirka and Satchell, 1990). Thus, under the conditions normally observed in sewer systems (i.e. pH 7.5) the addition of chlorine would result in the oxidation of sulfide (HS) to elemental sulfur. The amount of sulfide dosed in the experiments was equivalent to 2.5 mA/cm2 (when oxidized to sulfur), whereas the mass transport limited current for the generation of chlorine at chloride concentrations of 1100 mg/L is approximately 2.4 mA/cm2. Thus, under elevated chloride levels sufficient chlorine could have been produced to oxidize the sulfide added while this might not have been the case at low chloride concentrations. The results showed that at elevated chloride concentrations similar sulfide and organic removal rates were obtained (Table 2). Moreover, similar electron distribution among the electron sinks was observed. Hence, it appears that elevated chloride concentrations did not entail any significant differences in sulfide removal rate as well as in the organic removal. Analysis of the free chlorine concentration revealed that in all experiments no free chlorine or other reactive oxygen species (which are also detected with the used method) were present. It cannot be excluded that chlorine was formed and instantly reacted with the sulfide and/or organics present. This would be in line with several studies earlier reported (Chen, 2004; Bergmann and Koparal, 2005).
3.3.
Implications for practice
The results could indicate that sulfide removal by in situ generated oxygen was the predominant reaction mechanism, independent of the electrode material used. Other reaction mechanisms including the formation of radical species
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
C containing oxygen and/or halogen atom (e.g. OC 2 , HO2 , HClOC) cannot be excluded but did not have a significant impact on the sulfide removal process. In perfectly mixed conditions, such as in a rotating disc setup, the different mechanisms for the electrodes would be observed to a higher extent. However, in the reactors here, which are more amenable to wastewater treatment, limitations exist on the mixing intensity. Oxygen injection is presently considered as an attractive option for sulfide abatement in sewer systems. It is less expensive than most other chemicals, and can target rising mains where the SRB activity is the highest (Hvitved-Jacobsen, 2001). However, transport and storage of pure oxygen carries serious safety issues and precise control of dosing is not straightforward. By generating oxygen in situ the requirement for transport and storage are avoided, thus mitigating safety concerns. Other advantages of in situ oxygen generation compared to traditional methods for oxygen supply are the fine dispersion, high controllability and the ease to monitor. The disadvantage is the cost of the oxygen per unit weight. All electrode materials performed similarly in terms of sulfide removal. Therefore, Ta/Ir, Ru/Ir and Pt/Ir coated titanium electrodes seem the most suitable electrodes since they posses the lowest overpotential for oxygen evolution and they are already used in full scale applications. However, the life time of Ru/Ir electrodes for the oxygen evolution reaction, which is the predominant reaction at low chloride concentrations, is low (Hine et al., 1979). Hence, Ta/Ir and Pt/Ir coated titanium electrodes appear the most suitable electrodes for sulfide oxidation from domestic wastewater in sewer systems. Based on a cell potential of 5 V, a Coulombic efficiency of 95% (for oxygen generation) and a cost of $0.06 per kWh, the estimated delivery cost is $1.06 per kg, relative to a delivery cost of $0.54e0.82 per kg for standard oxygen purchases (de Haas et al., 2008). However, standard oxygen injection in sewer systems has low efficiency (i.e. w20 to 40%) (de Haas et al., 2008) since oxygen is often dosed in an inefficient way (i.e., coarse bubbles) which results in a significant loss of undissolved gas from air in gas release valves downstream. The latter is avoided when oxygen is generated in situ due to the high transfer efficiency and fine dispersion of in situ generated oxygen. The deployment of a sulfide or dissolved oxygen sensor before or after the electrochemical system respectively would also allow further fine-tuning of the oxygen dosing. While entailing a higher cost, sustainable electricity solutions such as photovoltaic power would allow total independence of the dosing system of transport or utility requirements. An in-depth life cycle analysis would provide more insight into the sustainability of the existing dosing methods. In all experiments excess oxygen was produced, which in a practical situation would be used for chemical oxidation or by the biofilm in the sewer pipes for biological sulfide oxidation or the removal of organic matter. Nonetheless, in this study, the impact of electrode material and chloride concentration on the kinetics of sulfide oxidation from real domestic wastewater was successfully investigated. Over the course of the experiments, none of the electrode materials deteriorated or demonstrated changing potential over time. Earlier studies at lower current densities indicated
5387
electrode fouling with sulfur (Dutta et al., 2009; Ateya et al., 2003; Rabaey et al., 2006), it appears that at the higher current densities used here sulfur either flakes off or is further oxidized to its soluble forms. However, in realistic conditions of the sewer, other forms of fouling such as ragging, particle settling and scaling may occur. Long term trials on site are presently underway to investigate these possible impacts, which were not observed here. Scaling of the membrane and the electrode surface, on the other hand, was observed in the cathode chamber. This is caused by transport of bivalent cations such as calcium through the cation exchange membrane causing precipitation of inorganics such as calcium hydroxide or calcium carbonate (Jeremiasse et al., 2010). To overcome problems with scaling the cathode needs to be cleaned periodically. This can be done either by chemical cleaning (i.e. the addition of citric or hydrochloric acid) or by periodic switching of the polarity of the electrodes (anode becomes cathode/cathode becomes anode), the latter option is attractive from an operational perspective but does restrict the types of electrodes that can be used.
4.
Conclusions
In this study, we investigated the kinetics of sulfide oxidation from real domestic wastewater using five different types of mixed metal-coated titanium electrodes (Ta/Ir, Ru/Ir, Pt/Ir, SnO2 and PbO2) at a current density of 10 mA/cm2. The obtained sulfide removal efficiencies were not significantly influenced by the electrode material used. The results indicate that, independent of electrode material used, sulfide was removed by means of chemical oxidation with in situ generated oxygen. Ta/Ir and Pt/Ir appear the most suitable electrodes since they have a low overpotential for oxygen evolution and are known for their stability even at low chloride concentrations. The obtained sulfide removal rates were in the same order of chemical rates found under high oxygen concentrations whereas in all experiments excess oxygen was produced. Elevated chloride concentrations did not entail any significant difference in sulfide removal rate. Analysis revealed that no free chlorine was present even at high chloride concentrations.
Acknowledgments Ilje Pikaar, Rene´ Rozendal and Korneel Rabaey thank the University of Queensland for scholarship (University of Queensland Research Scholarship) and fellowship support (RR UQ Postdoctoral Fellowship). KR is supported by the Australian Research Council (APD, DP0879245). This work was funded by the Australian Research Council (ARC Linkage project: LP0882016 “Optimal Management of Corrosion and Odor Problems in Sewer Systems”). The authors also want to acknowledge Dr. Beatrice Keller-Lehmann and Ms. Susan Cooke for their helpful collaboration with the chemical analyses.
5388
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 1 e5 3 8 8
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.07.033.
references
Ateya, B.G., AlKharafi, F.M., Al-Azab, A.S., 2003. Electrodeposition of sulfur from sulfide contaminated brines. Electrochemical and Solid-State Letters 6 (9), C137eC140. Bergmann, M.E.H., Koparal, A.S., 2005. Studies on electrochemical disinfectant production using anodes containing RuO2. Journal of Applied Electrochemistry 35 (12), 1321e1329. Chen, G., 2004. Electrochemical technologies in wastewater treatment. Separation and Purification Technology 38 (1), 11e41. Chwirka, J.D., Satchell, T.T., 1990. A 1990 guide for treating hydrogen sulfide in sewers. Water Engineering & Management 137, 32e35. Comninellis, C., 1994. Electrocatalysis in the electrochemical conversion/combustion of organic pollutants for waste water treatment. Electrochimica Acta 39, 1857. de Haas, D.W., Corrie, S., O’Halloran, K., Keller, J., Yuan, Z., 2008. Odour control by chemical dosing: a review. Journal of the Australian Water Association 35 (02), 138e143. Dutta, P.K., Rabaey, K., Yuan, Z., Keller, J., 2008. Spontaneous electrochemical removal of aqueous sulfide. Water Research 42 (20), 4965e4975. Dutta, P.K., Rozendal, R.A., Yuan, Z., Rabaey, K., Keller, J., 2009. Electrochemical regeneration of sulfur loaded electrodes. Electrochemistry Communications 11 (7), 1437e1440. Feng, Y.J., Li, X.Y., 2003. Electro-catalytic oxidation of phenol on several metal-oxide electrodes in aqueous solution. Water Research 37 (10), 2399e2407. Hine, F., Yasuda, M., Noda, T., Yoshida, T., Okuda, J., 1979. Electrochemical behavior of the oxide-coated metal anodes. Journal of the Electrochemical Society 126 (9), 1439e1445. Hvitved-Jacobsen, T., 2001. Sewer processes: microbial and chemical process engineering of sewer networks. xi þ 237 pp. Jeremiasse, A.W., Hamelers, H.V.M., Buisman, C.J.N., 2010. Microbial electrolysis cell with a microbial biocathode. Bioelectrochemistry 78 (1), 39e43. Kaempfer, W., Berndt, M., 1998. Polymer modified mortar with high resistance to acid to corrosion by biogenic sulfuric acid. In: Proceedings of the IX ICPIC Congress, Bologna, Italy, pp. 681e687. Keller-Lehmann, B., Corrie, S., Ravn, R., Yuan, Z., Keller, J., 2006. Preservation and simultaneous analysis of relevant soluble sulfur species in sewage samples. In: 2nd International IWA Conference on Sewer Operation and Maintenance, Vienna, Austria.
Martinez-Huitle, C.A., Brillas, E., 2009. Decontamination of wastewaters containing synthetic organic dyes by electrochemical methods: a general review. Applied Catalysis B: Environmental 87 (3e4), 105e145. Panizza, M., Cerisola, G., 2008. Electrochemical degradation of methyl red using BDD and PbO2 anodes. Industrial and Engineering Chemistry Research 47 (18), 6816e6820. Panizza, M., Kapalka, A., Comninellis, C., 2008. Oxidation of organic pollutants on BDD anodes using modulated current electrolysis. Electrochimica Acta 53 (5), 2289e2295. Pikaar, I., Rozendal, R.A., Yuan, Z., Keller, J., Rabaey, K., 2011. Electrochemical sulfide removal from synthetic and real domestic wastewater at high current densities. Water Research 45 (6), 2281e2289. Rabaey, K., Van de Sompel, K., Maignien, L., Boon, N., Aelterman, P., Clauwaert, P., De Schamphelaire, L., Pham, H.T. , Vermeulen, J., Verhaege, M., Lens, P., Verstraete, W., 2006. Microbial fuel cells for sulfide removal. Environmental Science & Technology 40 (17), 5218e5224. Sharma, K., Yuan, Z., 2010. Kinetics of chemical sulfide oxidation under high dissolved oxygen levels. Submitted for Oral Presentation, 6th International Conference on Sewer Processes and Networks, 7e10 November 2010. Sun, Y.-X., Wu, Q.-Y., Hu, H.-Y., Tian, J., 2009. Effects of operating conditions on THMs and HAAs formation during wastewater chlorination. Journal of Hazardous Materials 168 (2e3), 1290e1295. Sydney, R., Esfandi, E., Surapaneni, S., 1996. Control concrete sewer corrosion via the crown spray process. Water Environment Research 68, 338e347. Szpyrkowicz, L., Kaul, S.N., Neti, R.N., Satyanarayan, S., 2005. Influence of anode material on electrochemical oxidation for the treatment of tannery wastewater. Water Research 39 (8), 1601e1613. Takasu, Y., Sugimoto, W., Nishiki, Y., Nakamatsu, S., 2010. Structural analyses of RuO2eTiO2/Ti and IrO2eRuO2eTiO2/Ti anodes used in industrial chlor-alkali membrane processes. Journal of Applied Electrochemistry, 1e7. Taylor, B., Gardner, T., 2007. Southeast Queensland recycled water aspects and soil impacts, Sunshine Coast, Australia. Vincke, E., Wanseele, E.V., Monteny, J., Beeldens, A., Belie, N.D., Taerwe, L., Gemert, D.V., Verstraete, W., 2002. Influence of polymer addition on biogenic sulfuric acid attack of concrete. International Biodeterioration & Biodegradation 49 (4), 283e292. Zhang, L., De Schryver, P., De Gusseme, B., De Muynck, W., Boon, N., Verstraete, W., 2008. Chemical and biological technologies for hydrogen sulfide emission control in sewer systems: a review. Water Research 42 (1e2), 1e12. Zhu, X., Tong, M., Shi, S., Zhao, H., Ni, J., 2008. Essential explanation of the strong mineralization performance of boron-doped diamond electrodes. Environmental Science and Technology 42 (13), 4914e4920.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Biological iron oxidation by Gallionella spp. in drinking water production under fully aerated conditions W.W.J.M. de Vet a,c,d,*, I.J.T. Dinkla b,1, L.C. Rietveld d, M.C.M. van Loosdrecht c,a a
Oasen Drinking Water Company, PO Box 122, 2800 AC Gouda, The Netherlands Bioclear bv. Rozenburglaan 13, 9727 DL Groningen, The Netherlands c Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands d Department of Water Management, Delft University of Technology, Stevinweg 1, 2628 CN Delft, The Netherlands b
article info
abstract
Article history:
Iron oxidation under neutral conditions (pH 6.5e8) may be a homo- or heterogeneous
Received 7 December 2010
chemically- or a biologically-mediated process. The chemical oxidation is supposed to
Received in revised form
outpace the biological process under slightly alkaline conditions (pH 7e8). The iron
8 July 2011
oxidation kinetics and growth of Gallionella spp. e obligatory chemolithotrophic iron
Accepted 25 July 2011
oxidizers e were assessed in natural, organic carbon-containing water, in continuous lab-
Available online 29 July 2011
scale reactors and full-scale groundwater trickling filters in the Netherlands. From Gallionella cell numbers determined by qPCR, balances were made for all systems. The homo-
Keywords:
geneous chemical iron oxidation occurred in accordance with the literature, but was
qPCR
retarded by a low water temperature (13 C). The contribution of the heterogeneous
Gallionella spp.
chemical oxidation was, despite the presence of freshly formed iron oxyhydroxides, much
Groundwater trickling filtration
lower than in previous studies in ultrapure water. This could be caused by the adsorption
Biological and chemical
of natural organic matter (NOM) on the iron oxide surfaces. In the oxygen-saturated
iron oxidation
natural water with a pH ranging from 6.5 to 7.7, Gallionella spp. grew uninhibited and biological iron oxidation was an important, and probably the dominant, process. Gallionella growth was not even inhibited in a full-scale filter after plate aeration. From this we conclude that Gallionella spp. can grow under neutral pH and fully aerated conditions when the chemical iron oxidation is retarded by low water temperature and inhibition of the autocatalytic iron oxidation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The existence and relevance of iron-oxidizing bacteria (FeOB) in drinking water treatment has been well established from the very beginning of central water supply. Berger and Berger
(1928) mentioned that only five years after start-up, all Berlin Water Works were forced to switch from groundwater to surface water in 1882 due to a so-called ‘Eisenkalamita¨t’ (iron calamity). Biological essays demonstrated that Crenothrix polyspora and probably also Leptothrix ochracea caused
Abbreviations: FeOB, iron-oxidizing bacteria; NOM, natural organic matter; qPCR, (quantitative) real-time polymerase chain reaction; WTP, water treatment plant. * Corresponding author. Oasen Drinking Water Company, PO Box 122, 2800 AC Gouda, The Netherlands. Tel.: þ31 610927947; fax: þ31 152782355. E-mail addresses:
[email protected] (W.W.J.M. de Vet),
[email protected] (I.J.T. Dinkla),
[email protected] (L.C. Rietveld),
[email protected] (M.C.M. van Loosdrecht). 1 Tel.: þ31 505718455; fax: þ31 505717920. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.07.028
5390
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
pollution of the unfiltered, distributed water with ‘ocheryellow, dirty brownish up to coffee brown flocky deposits’ (Ibid.). Similar problems with iron-containing groundwater occurred in Rotterdam, The Netherlands (de Vries, 1890). Iron-containing groundwater could only be used for drinking water production once properly working de-ironing filters were developed. Since the establishment of de-ironing filters, there has been an ongoing discussion with regard to the importance of chemical versus biological iron oxidation (Czekalla et al., 1985; Sharma et al., 2005, and references therein). In drinking water filters in northern Germany, at least four species of FeOB have been reported (Gallionella sp., L. ochracea, Toxothrix trichogenes and an unknown bacterium), next to five species of manganese-oxidizing bacteria. From these observations it was concluded that iron and manganese removal was a bacterial process (Czekalla et al., 1985). Søgaard et al. (2000) studied precipitates from backwash sludge from three water treatment plants (WTP) in Denmark. They suggested low oxygen content of the raw water, poor aeration and relatively low pH as the determining prerequisites for biological iron oxidation; however, they did not provide consistent data from the WTPs to substantiate these presumptions. The presence of ferrous iron in combination with low dissolved oxygen and/ or slightly acidic pH is also regarded by other researchers as prerequisites for growth of FeOB (Hallbeck and Pedersen, 1990; Emerson and Floyd, 2005). The distinction between heterogeneous chemical and biological iron oxidation is, however, hard to make (Sharma et al., 2005; Tekerlekopoulou and Vayenas, 2008). Only in some cases, the distinguishable characteristic forms of iron deposits e like the twisted stalks formed by Gallionella spp. e indicate biological action, but in many other cases, particulate amorphous iron oxyhydroxides, very similar to chemical precipitates, are shown to be of biological origin as well (Emerson and Weiss, 2004). In recent studies the catalysis of iron oxidation by excreted RedOx-enzymes like flavins (Degre´mont, 2007) or exopolymers (Søgaard et al., 2000) has been reported, however this chemical process does not yield energy for bacterial growth. The chemo-lithotrophy of some FeOB is still under dispute (Spring and Ka¨mpfer, 2005). Gallionella spp. are, however, generally regarded as strictly lithotrophic, unable to catabolize organic matter (Lu¨tters-Czekalla, 1990), so the growth of Gallionella spp. can be seen as direct proof of biological iron oxidation. For this reason, this paper focuses on Gallionella spp., even though other FeOB such as Leptothrix spp. were found to be growing in the studied systems as well (data not shown). New molecular techniques provide powerful tools to assess and quantify the role of FeOB in full-scale treatment systems. For this paper, the kinetics of iron oxidation and the growth of the iron-oxidizing Gallionella bacteria were assessed in continuous lab- and full-scale reactors and trickling filters. The results of these studies were used to discuss the competition of biological iron oxidation with chemical iron oxidation at different pH’s in groundwater filtration. We hypothesize that Gallionella spp. can also grow under fully aerated and slightly alkaline pH conditions when chemical iron oxidation is retarded.
2.
Methods and materials
2.1.
Lab-scale experiments
The oxidation and removal of iron were investigated in two lab-scale setups at WTP Lekkerkerk of the Oasen drinking water company in The Netherlands. The lab-scale research consisted of oxidation column and filtration column experiments, which are described separately in the next two sections. Both experimental setups were fed with drinking water locally produced from riverbank groundwater. This water is moderately hard (Ca2þ w2 mM), well buffered (HCO 3 w3.0 mM), has a constant temperature of 13 C, a pH of 7.8 0.1. The dissolved oxygen content of the feed water was 9.9 0.6 mg L1 (74 data points in the period April 2008eAugust 2009 by potentiometric measurement in accordance with NEN-ISO 5814, NEN, 1993). Ferrous iron (FeSO47H20, Merck 103965 5000) was added to the feed water of all but the reference filter columns in a concentration of 3.3 mg L1 Fe, resembling the groundwater quality at WTP Lekkerkerk. A nutrient solution, containing phosphorus (0.6 mM PO4-P), nitrogen (3.8 mM NH4-N) and trace elements (Zn, Co, Cu and Mo), was added to prevent bacterial growth limitation. All water and chemical flows were controlled by tube pumps and all flow rates were checked weekly by mass measurements. All columns had an internal diameter of 0.089 m resulting in a water velocity of about 2.2 m h1, similar to the full-scale filters. While the desired flow rate was 14.0 L h1, the realized flow rates for the oxidation and filter columns were 13.9 0.4 and 13.5 0.8 L h1, respectively. The flow direction was upwards for the oxidation columns and downwards for the filter columns. Additional information on the columns’ setup, including schemes and pictures, is given in the Supplementary Material A.
2.1.1.
Oxidation columns’ setup
The chemical iron oxidation strongly depends on pH (Sung, 1980; Tamura et al., 1976), it is therefore supposed that, with a decreasing pH, biological oxidation might outcompete chemical oxidation processes. In order to determine which rates of chemical oxidation still allow simultaneous biological oxidation by Gallionella spp., different pH conditions allowing different rates of chemical oxidation were applied. The influence of feed water pH on the oxidation rate of the ferrous iron and the growth of Gallionella spp. in the natural water of WTP Lekkerkerk was studied in six oxidation columns. With an overflow level of 0.58 m, the residence time calculated from mass balances was 16 min. Determination of the residence time by NaCl spiking and corresponding conductivity measurements (not presented) showed no short-circuiting, indicating mainly plug flow conditions in the oxidation columns. To set the pH, HCl or NaOH was added to the column influent. The required doses of HCl and NaOH were determined in triplicate by titration of the feeding drinking water and checked by offline potentiometric pH measurement in conformity with the Dutch NEN-ISO 10523 protocol (NEN, 2009) of the columns’ influent, also in triplicate (Figure B.1 of Supplementary Material B). Although the titrations and control measurements were executed at 20 C, operational pH
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
values at 13 C will not have differed much because of the good buffering of the feed water. The deviation of the control measurements at the more extreme pH values was probably caused by gas exchange during sampling and offline measurements and calcium carbonate precipitation during storage. As the values determined by the titrations best represented the actual system during the experiments, these pH values will be used in the results section, with the uncertainty range calculated from mass measurements of the acid and base dosing. The realized iron dose determined by mass balances and total iron analysis (see Section ‘Iron analyses’) was 3.3 0.7 mg L1 Fe and is shown per column in Figure B.2 of the Supplementary Material B. In total, 58 4 g Fe was dosed over 7 weeks. The oxidation columns were run for seven weeks from July 8 to August 26, 2009. The concentrations of ferrous and ferric iron in the column effluents were determined weekly. The Gallionella spp. cell numbers were determined after three and seven weeks in the column influents and effluents and after seven weeks in the accumulated sludge in the oxidation columns. A picture of the oxidation columns’ setup at the end of the experiment is given as Figure A.2 in the Supplementary Material A.
2.1.2.
Filter columns’ setup
To model groundwater trickling filtration, iron removal was studied in a lab-scale filter columns’ setup, consisting of seven trickling filter columns in duplicate, in total 14 columns. All columns were filled with standard filter sand (1.7e2.5 mm). The pH of the feed water was lowered by HCl to resemble the groundwater before filtration (7.25 0.15). One column was used as a reference with no iron removal. In three columns, ferrous iron was dosed just before the filter top; in three other columns, ferrous iron passed a pre-oxidation column before filtration. These columns are referred to as “without preoxidation” and “with pre-oxidation", respectively. All trickling filter columns had forced ventilation which raised the pH in the filter effluents to 7.67 0.07. Each filter column was automatically backwashed every 24 h with a fixed volume of 30 L drinking water and under expansion (fluidization) of the filter bed. The filtration columns were run for six months from April 2 to October 8, 2008. The Gallionella spp. cell numbers were determined after six months in the influents and effluents, the backwash water and the filter material of seven columns (one of each duplicate).
2.2.
Full-scale groundwater trickling filters
The growth of Gallionella spp. and their role in iron oxidation was verified in three full-scale trickling filters at two Oasen WTPs. All three full-scale filters treated moderately hard and well-buffered anoxic groundwater. The filters were backwashed automatically after a filter runtime of 48 h, to prevent clogging by removal of inorganic precipitates and excess biomass. At WTP Lekkerkerk, the filter material of a trickling filter was externally washed and the filter performance and growth of Gallionella spp. were monitored extensively for nine months after restart of the filters from December 12, 2007 to September 19, 2008. At WTP De Hooge Boom, the growth of Gallionella spp. in two groundwater trickling filters was
5391
assessed in a quick scan on March 1, 2010. In one of these filters, the anoxic groundwater was sprayed directly on the trickling filter, while in the other the groundwater was intensively aerated on a plate aerator prior to spraying on top of the trickling filter. Both plate aeration and trickling filtration raised the dissolved oxygen content to nearly a saturated level and the pH by the stripping of carbon dioxide (see Table 1). All iron present in the groundwater was virtually completely removed in the filters (>95%). Table 1 gives an overview of the groundwater and filtrate qualities as well as the characteristics of the three studied filters.
2.3.
DNA extraction
Samples for detection and identification of Gallionella spp. were taken in sterilized glass bottles at different points in the Oasen WTP Lekkerkerk. All samples were stored at 4 C. Gallionella spp. cell numbers were determined by qPCR. Groundwater, influent and effluent water and backwash water samples were filtered over 0.2 mm polycarbonate membranes to concentrate the cells prior to DNA extraction. DNA was extracted from a volume of 100e150 ml water per sample. The filter was subsequently subjected to DNA extraction by bead beating in a Fast DNA spin kit for soil (MP Biomedicals, Soton, Ohio, United States). The DNA was purified using a silica-based column and eluted in 100 ml TE. DNA from approximately 10 g of filter sand was extracted as described by de Vet et al. (2009). In all cases, an internal control was used to determine the extraction efficiency.
2.4.
Quantification of Gallionella spp.
In order to quantify the number of Gallionella spp. cells in the systems, a specific PCR was developed to detect these bacteria, including the Gallionella spp. sequence that was previously found in the drinking water filters (Ibid.). PCR primers were developed for the detection of the 16S rRNA gene from Gallionella spp. using ARB software. One forward primer, GALFER0218-F 50 -GCTTTCGGAGTGGCCGATA-30 -, and one reverse primer, GALFER1408-R 50 - CAGATTCCACTCCCATGGTG -30 were designed. Amplification was performed by initial denaturation for 3 min at 94 C, followed by 35 cycles of amplification (30 s denaturation at 94 C; 30 s annealing at 62 C; 1 min elongation at 72 C), and 5 min at 72 C to complete elongation. Quantification was based on a comparison of the sample Ct value to the Ct value of a calibration curve using standard numbers of 16S rDNA fragments of Gallionella (see Supplementary Material C). It was assumed that Gallionella cells contain 1 16S gene copy per cell. An internal control was added to all samples to correct for the efficiency of the PCR reaction. The specificity of the qPCR method was checked through the construction and sequencing of clone libraries of the PCR products from filtrate and backwash water samples of the WTP Lekkerkerk full-scale filter (de Vet, 2011). From the qPCR enumeration results, balances for the labscale columns and the full-scale filters were calculated. To assess the role of biological iron oxidation, the following assumption was made: when the cell numbers entering and leaving the filters are constant in time, no net accumulation occurs and the net washout measured by qPCR balances the
5392
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
Table 1 e Groundwater and filtrate quality and filter characteristics of the full-scale trickling filters at the Oasen WTPs. WTP Lekkerkerk
WTP De Hooge Boom
Direct trickling filtration Filter sand d10-d90, mm Filter bed area, m2 Average production flow, m3 h1 Filter runtime, h Groundwater quality Temperature, C pH, 1 HCO 3 , mg L Total iron, mg L1 TOC, mg L1 C Effluent plate aerator pH, O2, mg L1 Filtrate quality pH, O2, mg L1 Total iron removed per filter runtime, kg Fe
Direct trickling filtration
1.7e2.5 18.0 37 48 Average St. Dev. Jan.eSept. 2008 11.6 0.3 7.33 0.04 216 7 5.5 0.6 2.2 0.1
Plate aeration and trickling filtration
2.0e3.15 28.0 64 48 Average St. Dev. Jan. 2008eMar. 2010 11.5 0.2 7.10 0.05 387 12 8.5 0.8 8.3 0.2
e e
7.7a 10.1a
e e 7.8a 9.7a
7.69 0.10 9.7 0.6 (10 data points) 9.6 1.1
7.80 0.03 10.2 0.4 (3 data points) 26.1 2.3
a Indicative local measurement on March 1, 2010.
growth of Gallionella spp., during one filter run. For every water flow entering or leaving a system, the totalized values for the cell numbers were calculated by multiplying the measured concentration with the flow rate and duration of the phase. For the full-scale filter at WTP Lekkerkerk, the groundwater was sampled in duplicate; the filtrate in duplicate five and nine months after external washing with four samples per filter runtime of 48 h; the backwash water was sampled three, six and nine months after external washing; control backwash samples were taken in quintuplet 15e16 months after external washing. At WTP De Hooge Boom, the influent and backwash water of each filter and the effluent of the plate aerator were sampled once; the filtrate water of each filter was sampled twice, at the beginning and at the end of the filter runtime of 48 h. The filtrate of the filter columns was sampled two hours after backwash. Filter column sand samples were taken from the top half of the bed during expansion backwashing.
2.5.
Iron analyses
Samples for iron analysis were taken directly into acid containing bottles to set the pH below 2. Nitric and hydrochloric acid were used to stabilize the samples for total and ferrous iron analysis, respectively. All samples were stored cool and analyzed within 24 h after sampling. Total iron in water samples was determined by inductively coupled plasma mass spectrometry (ICP-MS). Ferrous iron was determined by the 1,10-phenanthroline method according to the Dutch NEN 6482 protocol, based on Standard methods (1975). The iron concentration in the sludge was measured by atomic emission spectroscopy after sample destruction in a microwave. The mass of the filter coating was determined by measurements of the dry mass before and after acidification with 4 M
hydrochloric acid and oxalic acid. The iron concentration in the decanted acid solution was measured by ICP-MS.
3.
Results
3.1. pH effect on growth of Gallionella and iron oxidation rate in oxidation columns The pH dependency of Gallionella growth and iron oxidation was examined in six lab-scale oxidation columns. During the first week after start-up of the experiment, the degree of oxidation was lower than the average for the following six weeks for all oxidation columns except the one with pH 8.25 (Fig. 1). Apart from the first week, the oxidation degree of iron in the columns’ effluent e calculated from the ferrous iron concentrations in filter effluent (by 1,10-phenanthroline method; Figure B.3 of Supplementary Material B) and added iron concentrations (from mass balances) e was constant in time. During the first week after start-up, virtually no iron oxyhydroxides or FeOB had formed in the columns yet, and the measured iron oxidation was accounted for mainly by the homogeneous chemical process. After the start-up period, both iron oxyhydroxides and FeOB accumulated in the columns and influenced the oxidation kinetics. The numbers of Gallionella cells determined by qPCR in the influents and effluents of the columns after 3 and 7 weeks and the total iron and Gallionella cell numbers accumulated as sludge in the oxidation columns after 7 weeks are shown in Fig. 2. The cell concentrations in the influent, effluent and sludge show that Gallionella grew and accumulated in all oxidation columns. Statistical analysis (Supplementary Material D) shows that Gallionella grew equally fast in all columns with a pH between 7.0 and 7.73, and slightly (but not significantly)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
5393
present in the effluent of the columns in dissolved ferrous or colloidal ferric form. Growth of Gallionella spp. was confirmed based on the morphology of the deposits. Fig. 3 shows a phase contrast picture of deposits in the oxidation column with pH 7.73. In the oxidation column experiment, the specific growth rate m of Gallionella spp. can be approached by Equation (1). ln ðDtY0Þm ¼
Fig. 1 e Oxidation degree of iron after 16 min’ passage through oxidation columns; green solid bars, averages and standard deviations for 6e7 data points for whole period except first week after start-up, calculated from the ferrous iron concentrations in filter effluent (by 1,10phenanthroline method) and total iron concentrations (from mass balances); blue striped bars, oxidation degree of iron during first week after start-up; calculated with [OHL]0.6, calculated with [OHL]2 (see Discussion section for explanation). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
faster at pH 6.5. The increasing rate of chemical iron oxidation did not inhibit the growth of Gallionella up to a pH of 7.73. Only in the column with pH 8.25 was Gallionella growth significantly slower. The total iron accumulated in the oxidation columns had a maximum at pH 7.00, and was for all columns between 1.8 and 3.6 g (3e6 102 mol). This was between 3 and 6% of the iron loading. The majority of the iron, therefore, was
Fig. 2 e Gallionella spp. concentrations (left axis, in cells mLL1) in the influent and the effluents of the oxidation columns operated at different pH values after 3 weeks (yellow solid bars) and after 7 weeks (green striped bars); total Gallionella spp. numbers (red dotted bars on left axis, in cells) and total iron (white bars on right axis, in g Fe) accumulated in the oxidation column after 7 weeks; error bars show uncertainty of qPCR method (between 0.5*N and 2*N). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Xt;column þ Q Dt xt;effluent Xt;column Dt
(1)
where, Xt,column ¼ total cells in column (cells), Q ¼ flow rate (m3 h1), [x]t,effluent ¼ cell concentration washed out of column (cells m3), t ¼ experimental time (h). At the end of the oxidation column experiment, m was 0.08 0.06 h1, corresponding to a doubling time of 8.4 h on average. Hallbeck and Pedersen (1990) found a generation time of 8.3 h in vitro at the optimal temperature of 20 C. This is comparable to the growth rate observed in our experiments at 13 C, which suggests slightly more favorable growth conditions in situ.
3.2. Effect of pre-oxidation on Gallionella growth in trickling filters The effect of pre-oxidation on the number of Gallionella spp. in trickling filters was studied in the combined oxidation and filtration column experiment. The pre-oxidation caused an oxidation degree of 29 6% before the water entered the trickling filters. At the applied pH (7.25 0.15) this corresponded with the oxidation degree measured in the oxidation column experiments (Fig. 1). The total numbers of Gallionella spp. for the water flows cumulated over one filter runtime (24 h) and for the filter beds at the end of the test period of 6 months are shown in Fig. 4. This figure clearly shows the growth of Gallionella spp. in all filter columns spiked with ferrous iron, but none in the reference column. No significant difference was found in the water samples from filter columns without and with pre-oxidation and only marginally more
Fig. 3 e Phase contrast microscopic picture (4003 magnification) of a sludge sample from the bottom of the oxidation column with pH set to 7.73 after three weeks of operation.
5394
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
Fig. 4 e Iron deposition in filter coating and cumulative numbers of Gallionella spp. by qPCR determined for a reference column, three columns without and three with pre-oxidation in the filter columns’ setup at the end of 6-months’ trial; numbers in water flows (solid and striped bars) cumulated over one filter run of 24 h, yellow solid bar in feed water, green vertically striped bar in filtrate water; and blue horizontally striped bars backwash water; total present in filter material (dotted bars); iron deposition in filter coating ( after 91 days, after 187 days, on the right axis); error bars show uncertainty of qPCR method for the outflow measurements of the reference and standard deviation for the other measurements. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Gallionella spp. in the filter samples from filter columns without pre-oxidation (shown in Supplementary Material D). Fig. 4 also shows iron deposition in the filter coating. After 91 days, the amount of iron deposited in the filter coatings was comparable regardless of pre-oxidation or not, and on average 60% of the loaded iron (Supplemental Material C) was encapsulated in the filter coating. After 187 days, however, the amount of iron in the filter coatings of columns with pre-oxidation had not increased, while it had in the columns without preoxidation. This suggests that the growth of attached FeOB in the filters may enhance the formation of iron coating. In the filter column experiments, the absence of preoxidation resulted, on average, in slightly higher cell numbers attached to the filter material, but due to error margins it is not possible to judge if this is significant. Although this is consistent with the higher ferrous iron loading in the columns without pre-oxidation, there was no significant difference in Gallionella spp. numbers in the water flows from columns with and without pre-oxidation. With the approach according to Equation (1), m was calculated as 0.01 0.005 and 0.03 0.01 h1 for the filter columns without and with pre-oxidation, respectively (with equal distribution of the cells washed out during backwash periods over the filter runtime). This suggests that the Gallionella cells in the columns without pre-oxidation were better attached, more encapsulated in the iron oxyhydroxide filter coating, and less active than in the columns with pre-oxidation.
3.3.
Full-scale groundwater trickling filters
In order to determine the potential role of biological oxidation in the groundwater trickling filters, the abundance and growth of the iron-oxidizing Gallionella species were assessed in the three full-scale filters by qPCR. The balances for Gallionella spp.
in duplicate calculated over one filter run of 48 h from the qPCR cell numbers, water flows, and time are shown in Fig. 5. Repeated measurements over the trial period of nine months at WTP Lekkerkerk (see Supplemental Material C) showed no trend in cell numbers, indicating a stable population. The measurements show that significant numbers of Gallionella cells were found in all three full-scale filters despite the fact that these filters were very well aerated and the oxygen content of the filtrate water was close to saturation level. This condition is usually associated with chemical iron oxidation (Sharma et al., 2005). The cell numbers leaving the filter through the filtrate and backwash water were much higher than in the groundwater feeding the filter. This indicates a strong growth of Gallionella spp. in these full-scale trickling filters. The plate aeration prior to the filtration at WTP De Hooge Boom did not inhibit the growth of Gallionella spp. in the filter, despite the oxygen saturation and elevated pH of the effluent water. Gallionella spp. started to grow in the plate aerator, were filtered off in the trickling filter and continued growing there.
4.
Discussion
4.1. Chemical versus biological iron oxidation in groundwater filtration Iron oxidation under aerobic, neutral pH conditions may be homogeneous, heterogeneous or biologically mediated. At the start of the oxidation column experiment, a negligible amount of iron oxyhydroxides and FeOB was present, and the iron oxidation was predominantly homogeneous. The general kinetic equation for homogeneous iron oxidation is given by Equation (2): n m (2) dFe=dt ¼ k OH PO2 Fe2þ
5395
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
Fig. 5 e Gallionella spp. balances for the WTP Lekkerkerk trickling filter (multiple measurements, graph A, red solid bars) and for the WTP De Hooge Boom trickling filters (singular measurements, graph B, green square bars, trickling filtration after plate aeration; blue line bars, direct trickling filtration); cumulative cell numbers inoculated directly from groundwater or via plate aerator, washed out to effluent and to backwash are calculated for one filter runtime of 48 h; Influent filter values are for effluent plate aerator if present and equal to groundwater for the other filters; error bars show standard deviation (graph A) and uncertainty of qPCR method (graph B, between 0.5*N en 2*N). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
where m ¼ 1 and n ¼ 2, as found by Sung (1980); only for the first week measurements did the model reasonably match the measured data (blue squares in Fig. 1; 13 C, ionic strength 0.01 M) with a rate constant of k ¼ 4$1012 M2 atm1 min1. This is approximately 10 times lower than the rate reported by Sung (Ibid.) in water with similar salinity but at 25 C. The temperature difference between 25 C and 13 C, explains this difference for the larger part. The lower temperature reduces the rate constant by a factor of 7, not because of changes in the activation energy (almost zero), but by the decline in Kw and thus [OH] activity (Stumm and Lee, 1961). This strongly indicates that the iron oxidation was mostly a homogeneous chemical reaction in the first week of the oxidation column experiment. The measurements during the rest of the experimental period can only be fitted to the model by reducing the order of [OH], i.e. n, to 0.6 (green triangles in Fig. 1). Tamura et al. (1976) found that rate of heterogeneous chemical iron oxygenation was proportional to the first order of the reciprocal [Hþ]. The general kinetic equation for heterogeneous chemical iron oxidation is given by Equation (3): dFe= ¼ k1 þ k2 Fe3þ Fe2þ dt
(3)
where 2 k1 ¼ khom OH PO2 Homogeneous Oxidation rate Constant; (3a)
1 Heterogenous oxidation rate constant; k2 ¼ ks;O ½O2 K Hþ (3b) 1
ks;O ¼ 4380 M1 min K ¼ 104:85
Surface rate ;
Adsorption constant of Fe2þ on FeOOH;
(3c) (3d)
The iron sludge that accumulated at the end of the experimental period of seven weeks in the oxidation columns was assumed to be equally distributed in the oxidation columns, leading to a ferric iron concentration of 9e18 mM (see Fig. 2; sludge volume per column was 4.0 0.1 L). Under these conditions, the heterogeneous oxidation rate constant k2 Fe3þ according to Equation (3) would be in the range 1e35 min1 (for pH 6.5 up to 8.25, respectively). As this means an oxidation half-life (t½) of less than 1 min, it would implicate a nearly complete chemical oxidation of iron after the average residence time of 16 min in all the oxidation columns, which was not the case in our experiments. The reason for this reduced heterogeneous oxidation rate cannot be deduced from our experiments, but is probably related to the composition of the natural water. In many studies on chemical iron oxidation (Stumm and Lee, 1961; Sung, 1980; Tamura et al., 1976) the oxidation rates were determined with ultrapure water. Some studies determined the effects of natural organic matter (NOM) on the chemical iron oxidation. Davison and Seed (1983) and Liang et al. (1993)
5396
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
found that the rate constant for homogeneous iron oxidation in natural freshwater under oxygen-saturated conditions was comparable to the one in synthetic water, as we did. Other researchers found a significant effect of NOM complexation on iron removal and oxidation, but that effect could be either accelerating (Ninh Pham et al., 2004) or inhibiting (Theis and Singer, 1974). All these studies were confined to homogeneous chemical iron oxidation. Sung (1980) stated that catalytic iron oxidation was only noticeable at pH 7 and above because of the slow surface formation at a lower pH. Our research with natural water indicates that the heterogeneous iron oxidation rate was strongly reduced even at a higher pH. This reduced heterogeneous chemical iron oxidation rate may be caused by surface complexation of inorganic and (natural) organic compounds in water. Complexation of inorganic ions had little influence on the adsorption capacity of ferrous iron (Sharma, 2001). Tipping (1981) showed that the surface charge and adsorption capacity of iron oxyhydroxides could be influenced by the complexation of humic substances. Gallionella growth may have contributed to the surface complexation and stabilization of the iron (oxy)hydroxides by the excretion of polysaccharides, comparable to the stalk formation described by Chan et al. (2011). Analysis of the growth of Gallionella spp. by qPCR demonstrates the significance of bacterial iron oxidation in the fullscale filter and laboratory filter columns. The direct enumeration of Gallionella spp. by this method combined with the biomass yield on iron oxidation makes it possible to quantify the share of biological iron oxidation. The maximum biomass yield reported in the literature is low (0.006 g DW g1 Fe oxidized (Lu¨tters and Hanert, 1989) and 0.013 g DW g1 Fe oxidized (Neubauer et al., 2002). Thermodynamically, a maximum theoretical yield of 0.012 g DW g1 Fe can be expected, based on the anabolic reaction energy of 3500 kJ C mol1 biomass (Heijnen and Van Dijken, 1992) and the catabolic reaction energy (Hanselmann, 1991): Fe2þ þ ¼ O2 þ 1½ H2O / FeOOH þ 2Hþ with ΔGr ¼ 83.8 kJ mol1 Fe at pH 7.73 and 1 mM Fe2þ. During one filter runtime, 0.9 g of iron was removed in the laboratory filter system (Supplementary Material B) and 9.6 1.1 kg in the full-scale filter at WTP Lekkerkerk. During one runtime, in total 4.7 2.9 1011 and 3.4 2.9 1015 Gallionella cells were washed out of these filters, respectively. When biomass accumulation, maintenance and decay were not considered, the observed yield was 5.1 3.3 1011 and 3.6 3.4 1011 Gallionella cells g1 Fe oxidized, respectively.
This assumes that the iron oxidation was completely biological and exclusively by Gallionella spp.. These maximum Gallionella cell yields can be related to dry weight (DW) by using the cell dimensions (mean volume of 0.4 mm3) determined by Hallbeck and Pedersen (1991). With a specific cell DW of 1.2 1013 g, the yield equals 0.062 0.040 and 0.043 0.041 g DW g1 Fe oxidized, for the filter columns and the full-scale filter, respectively (Table 2). Although the standard deviations are large, the average yield was higher than reported in the literature and the theoretical maximum. The high cell yields found suggest that biological iron oxidation by Gallionella spp. played a dominant role in both the full-scale filter and in the filter columns.
4.2.
Growth conditions of Gallionella spp.
The results reported in this manuscript show that Gallionella spp. may grow under broader conditions than generally assumed. No growth inhibition was found in the natural water under fully aerated conditions and at a pH ranging from 6.5 to 7.73. This finding contrasts with the general perception of Gallionella being strictly microaerophilic (Emerson, 2000). The oxygen concentration at the microsites where Gallionella grew in our experiments may have been reduced compared to the bulk water. We did not measure the oxygen concentration in those microsites. When iron oxidation is the dominant oxygen-consuming process, however, only a limited reduction in oxygen concentration may be expected from the bulk water into microsites, based on the stoichiometry of the iron oxidation, the measured oxygen and ferrous iron concentrations in the bulk water and the diffusion coefficients for both compounds from the literature (Broecker and Peng, 1974; Li and Gregory, 1974). According to Degre´mont (2007), biological iron oxidation will only prevail under conditions where physico-chemical iron oxidation is not possible: oxygen concentration between 0.2 and 0.5 mg L1, pH 6.3, oxidation reduction potential þ100 mV and rH2 between 14 and 20 (whereas rH2 ¼ log ( pH2) ¼ Eh/0.0296 V þ 2 pH). Under rH2 of 14, the biological oxidation should be inhibited, while over 20, the bacteria would lose the competition with the physico-chemical iron precipitation. At pH 7.73, the upper limit of rH2 indicates a maximum redox potential of 135 mV and an oxidation degree of less than 98%. It was stated that the boundaries are not strictly defined and can shift e.g. by chelation. Hanert (2006) listed the broad array of the environments where Gallionella spp. have been
Table 2 e Overview of iron conversion, net Gallionella cells washout and calculated yield for filter column experiment and full-scale trickling filter at WTP Lekkerkerk. Parameters Iron removed per filter runtime Gallionella cells washed out per filter runtime Cell yield Biomass yielda
Unit g Fe cells cells g1 Fe g DW g1 Fe
Filter columns
Full-scale trickling filter
0.92 0.04 4.7 2.9 1011 5.1 3.3 1011 0.062 0.040
9.6 1.1 103 3.4 2.9 1015 3.6 3.4 1011 0.043 0.041
a Calculated with 1.2 1013 g DW cell1, mean cell volume 0.4 mm3, Hallbeck and Pedersen (1991).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
found and concluded that the stability of ferrous iron in combination with oxygen is crucial for their existence, more than mere pH or Eh. This paper substantiates this claim by showing the growth of Gallionella spp. on ferrous iron under fully aerated and slightly alkaline circumstances, when the chemical iron oxidation is slow. In the oxidation column experiment with a fixed pH ranging from 6.5 to 7.73 and oxygen-saturated natural water, the initial iron oxidation was homogeneous with rates consistent with the literature. After the start-up period, FeOB and iron oxyhydroxydes accumulated in the columns but the oxidation rate increased less than theoretically expected from heterogeneous chemical oxidation. Heterogeneous chemical iron oxidation may be seriously hampered in natural water compared to synthetic water by complexation of natural organic matter on iron oxyhydroxide surfaces. The specific growth of Gallionella spp. was in accordance with the values found in culture experiments. The comparable Gallionella cell growth and the increase in iron oxidation degree indicate that, for pH ranging from 6.5 to 7.73, the increased iron oxidation rate had to be attributed to the growth and activity of FeOB, rather than to chemical catalysis. Yield calculations for the biological iron oxidation by Gallionella spp. in lab- and fullscale trickling filters, indicate that the dominant iron oxidation mechanism in groundwater filtration is biological under wider process conditions (pH and oxygen content) than previously thought.
5.
Conclusions
The quantitative PCR approach targeting the 16S rRNA of Gallionella spp. was successfully used to determine the significance of biological versus chemical oxidation in full-scale groundwater trickling filters and lab-scale column experiments. Gallionella spp. grew in fully aerated full-scale groundwater trickling filters and lab-scale oxidation columns and trickling filters at neutral pH (up to pH 7.7) and at a moderate temperature of 13 C. Biological oxidation by Gallionella spp. was the dominant process for iron oxidation in this type of groundwater, and heterogeneous chemical iron oxidation in natural water was substantially reduced, compared to experimental results from the literature for synthetic water.
Acknowledgements The authors gratefully acknowledge the contribution of Peter Dijkstra for assistance in the full-scale research, Petra Lafeber for support in the column experiments and Sabine Doddema and Paul van der Wielen for the DNA-isolation for the qPCR.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.07.028.
5397
references
Berger, H., Berger, E., 1928. Biologie der Trink- und Brauchwasseranlagen. Verlag von Gustav Fischer, Jena. BerlinDahlem, Germany. Broecker, W.S., Peng, T.H., 1974. Gas exchange rates between air and sea. Tellus 26 (1e2), 21e35. Chan, C.S., Fakra, S.C., Emerson, D., Fleming, E.J., Edwards, K.J., 2011. Lithotrophic iron-oxidizing bacteria produce organic stalks to control mineral growth: implications for biosignature formation. ISME Journal 5 (4), 717e727. Czekalla, C., Mevius, W., Hanert, H., 1985. Quantitative removal of iron and manganese by microorganisms in rapid sand filters (in situ investigations). Water Supply 3 (1), 111e123. Davison, W., Seed, G., 1983. The kinetics of the oxidation of ferrous iron in synthetic and natural waters. Geochimica et Cosmochimica Acta 47 (1), 67e79. Degre´mont (Ed.), 2007. Water Treatment Handbook. Lavoisier SAS. Emerson, D., Floyd, M.M., 2005. Enrichment and isolation of ironoxidizing bacteria at neutral pH. Methods in Enzymology 397, 112e123. Emerson, D., Weiss, J.V., 2004. Bacterial iron oxidation in circumneutral freshwater habitats: findings from the field and the laboratory. Geomicrobiology Journal 21 (6), 405e414. Emerson, D., 2000. Environmental microbe-metal interactions. In: Lovley, D.R. (Ed.). ASM, Washington, pp. 31e52. Hallbeck, L., Pedersen, K., 1990. Culture parameters regulating stalk formation and growth rate of Gallionella ferruginea. Journal of General Microbiology 136 (9), 1675e1680. Hallbeck, L., Pedersen, K., 1991. Autotrophic and mixotrophic growth of Gallionella ferruginea. Journal of General Microbiology 137 (11), 2657e2661. Hanert, H.H., 2006. The Prokaryotes: a Handbook on the Biology of Bacteria. Springer, pp. 990e995. Hanselmann, K.W., 1991. Microbial energetics applied to waste repositories. Experientia 47 (7), 645e687. Heijnen, J.J., Van Dijken, J.P., 1992. In search of thermodynamic description of biomass yields for the chemotrophic growth of microorganisms. Biotechnology and Bioengineering 39 (8), 833e858. Li, Y.-H., Gregory, S., 1974. Diffusion of ions in sea water and in deep-sea sediments. Geochimica et Cosmochimica Acta 38 (5), 703e714. Liang, L., Andrew McNabb, J., Paulk, J.M., Gu, B., McCarthy, J.F., 1993. Kinetics off Fe(II) oxygenation at low partial pressure of oxygen in the presence of natural organic matter. Environmental Science and Technology 27 (9), 1864e1870. Lu¨tters, S., Hanert, H.H., 1989. The ultrastructure of chemolithoautotrophic Gallionella ferruginea and Thiobacillus ferrooxidans as revealed by chemical fixation and freezeetching. Archives of Microbiology 151 (3), 245e251. Lu¨tters-Czekalla, S., 1990. Lithoautotrophic growth of the iron bacterium Gallionella ferruginea with thiosulfate or sulfide as energy source. Archives of Microbiology 154 (5), 417e421. Nederlands Normalisatie instituut (NEN), 1993. NEN-ISO 5814: 1993 en, Water e Bepaling van het gehalte aan opgeloste zuurstof e Elektrochemische methode. http://www.nen.nl/ web/Normshop/Norm/NENISO-58141993-en.htm. Nederlands Normalisatie instituut (NEN), 2009. NEN-ISO 10523: 2008 en, Water e Bepaling van de pH. http://www.nen.nl/web/ Normshop/Norm/NENISO-105232008-en.htm. Neubauer, S.C., Emerson, D., Megonigal, J.P., 2002. Life at the energetic edge: kinetics of circumneutral iron oxidation by lithotrophic iron-oxidizing bacteria isolated from the wetlandplant rhizosphere. Applied and Environmental Microbiology 68 (8), 3988e3995.
5398
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 8 9 e5 3 9 8
Ninh Pham, A., Rose, A.L., Feitz, A.J., Waite, T.D., 2004. The Effect of Dissolved Natural Organic Matter on the Rate of Removal of Ferrous Iron in Fresh Waters, pp. 213e219. Sharma, S.K., Petrusevski, B., Schippers, J.C., 2005. Biological iron removal from groundwater: a review. Journal of Water Supply: Research and Technology e AQUA 54 (4), 239e247. Sharma, S.K. (2001) Adsorptive iron removal from groundwater. Dissertation Wageningen University/International Institute for Infrastructural, Hydraulic and Environmental Engineering, Delft, the Netherlands. Søgaard, E.G., Medenwaldt, R., Abraham-Peskir, J.V., 2000. Conditions and rates of biotic and abiotic iron precipitation in selected Danish freshwater plants and microscopic analysis of precipitate morphology. Water Research 34 (10), 2675e2682. Spring, S., Ka¨mpfer, P., 2005. Bergey’s Manual of Systematic Bacteriology, pp. 740e746. Standard methods for the Examination of Water and Waste Water, fourteenth ed., 1975 American Public Health Association, Washington, pp. 208e213. Stumm, W., Lee, G.F., 1961. Oxygenation of Ferrous Iron, pp. 143e146. Sung, W., 1980. Kinetics and product of ferrous iron oxygenation in aqueous systems. Environmental Science and Technology 14 (5), 561e568.
Tamura, H., Goto, K., Nagayama, M., 1976. Effect of ferric hydroxide on the oxygenation of ferrous ions in neutral solutions. Corrosion Science 16 (4), 197e207. Tekerlekopoulou, A.G., Vayenas, D.V., 2008. Simultaneous biological removal of ammonia, iron and manganese from potable water using a trickling filter. Biochemical Engineering Journal 39 (1), 215e220. Theis, T.L., Singer, P.C., 1974. Complexation of iron(II) by organic matter and its effect on iron(II) oxygenation. Environmental Science and Technology 8 (6), 569e573. Tipping, E., 1981. The adsorption of aquatic humic substances by iron oxides. Geochimica et Cosmochimica Acta 45 (2), 191e199. de Vet, W.W.J.M., Dinkla, I.J.T., Muyzer, G., Rietveld, L.C., van Loosdrecht, M.C.M., 2009. Molecular characterization of microbial populations in groundwater sources and sand filters for drinking water production. Water Research 43 (1), 182e194. de Vet, W.W.J.M. (2011). Biological drinking water treatment of anaerobic groundwater in trickling filters. Dissertation. Technical University Delft, The Netherlands. http://repository. tudelft.nl. de Vries, H., 1890. Die Pflanzen und Thiere in den dunklen Ra¨umen der Rotterdamer Wasserleitung; Bericht u¨ber die biologischen Untersuchungen der Crenothrix-Commission zu Rotterdam, vom Jahre 1887. Verlag von Gustav Fischer, Jena.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Temporal variability of pharmaceuticals and illicit drugs in wastewater and the effects of a major sporting event Daniel Gerrity a,b,*, Rebecca A. Trenholm b, Shane A. Snyder b,c a
Trussell Technologies, Inc., 6540 Lusk Blvd., Suite C274, San Diego, CA 92121, United States Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, P.O. Box 99954, Las Vegas, NV 89193-9954, United States c Department of Chemical and Environmental Engineering, University of Arizona, 1133 E. James E. Rogers Way, Harshbarger 108, Tucson, AZ 85721-0011, United States b
article info
abstract
Article history:
Diurnal variations in wastewater flows are common phenomena related to peak water use
Received 23 April 2011
periods. However, few studies have examined high-resolution temporal variability in trace
Received in revised form
organic contaminant (TOrC) concentrations and loadings. Even fewer have assessed the
23 June 2011
impacts of a special event or holiday. This study characterizes the temporal variability
Accepted 17 July 2011
associated with a major sporting event using flow data and corresponding mass loadings of
Available online 23 July 2011
a suite of prescription pharmaceuticals, potential endocrine disrupting compounds (EDCs), and illicit drugs. Wastewater influent and finished effluent samples were collected during
Keywords:
the National Football League’s Super Bowl, which is a significant weekend for tourism in
Pharmaceutical
the study area. Data from a baseline weekend is also provided to illustrate flows and TOrC
Endocrine disrupting
loadings during “normal” operational conditions. Some compounds exhibited interesting
compound (EDC)
temporal variations (e.g., atenolol), and several compounds demonstrated different loading
Illicit drug
profiles during the Super Bowl and baseline weekends (e.g., the primary cocaine metabolite
Temporal variation
benzoylecgonine). Interestingly, the influent mass loadings of prescription pharmaceuti-
Loading
cals were generally similar in magnitude to those of the illicit drugs and their metabolites.
Reuse
However, conventional wastewater treatment was more effective in removing the illicit
Wastewater
drugs and their metabolites. Total influent and effluent mass loadings are also provided to summarize treatment efficacy and environmental discharges. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Pharmaceuticals and personal care products (PPCPs) and endocrine disrupting compounds (EDCs) are often considered “emerging contaminants,” but researchers have been aware of their presence in water for decades. However, the occurrence of PPCPs and EDCs in water did not become a mainstream research topic until the late 1990s and early 2000s. The spike in scientific interest stemmed from demonstrated impacts on
aquatic ecosystems (Snyder et al., 2001, 2004; Lange et al., 2009), potential human health effects (Snyder et al., 2008; Schriks et al., 2010; Stanford et al., 2010), and increased media coverage (Donn et al., 2008), which ultimately led to increased public awareness. This increased interest was coupled with the development of extremely sensitive analytical methods such as liquid chromatographyetandem mass spectrometry (LCeMS/MS) that allowed researchers to approach parts-per-quadrillion (sub-ng/L) detection limits for
* Corresponding author. Trussell Technologies, Inc., 6540 Lusk Blvd., Suite C274, San Diego, CA 92121, United States. Tel.: þ1 858 458 1030; fax: þ1 626 486 0571. E-mail address:
[email protected] (D. Gerrity). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.07.020
5400
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
a variety of trace organic contaminants (TOrCs) (Ternes et al., 2002; Snyder et al., 2003; Vanderford and Snyder, 2006; Postigo et al., 2011). Each of these factors led to more thorough scientific investigations into the presence, fate, and transport of TOrCs in natural and engineered systems. With respect to organic compounds intended for human consumption, contamination of water supplies stems from their release during manufacturing, excretion after personal use, and public disposal of unused quantities (Daughton and Ternes, 1999). Considering that each of these routes directly impacts wastewater, it is reasonable to assume that discharged wastewater is a major source of these contaminants in environmental waters. In the past, wastewater treatment trains were generally not designed for TOrC removal. However, the prevalence of indirect potable reuse, whether “planned” or “unplanned”, and demonstrated impacts on aquatic ecosystems now justify some consideration of TOrCs in the design process. In fact, expansion and optimization of wastewater treatment processes may be the most efficient strategy to mitigate the potential effects of these contaminants (Nelson et al., 2011). To aid in this effort, one design factor that must be studied in greater detail is the temporal variability of TOrC occurrence in wastewater. Diurnal variations in wastewater flows are common phenomena related to peak water use periods (Nelson et al., 2011). Sewers and wastewater treatment plants must be designed to account for the maximum and minimum flows and associated loadings each day (Ort et al., 2010). However, few studies have examined temporal fluctuations in TOrC concentrations and mass loadings (Joss et al., 2005; Takao et al., 2008; Nelson et al., 2011; Plosz et al., 2010; Postigo et al., 2011). In the existing studies, the temporal resolution was typically limited to sampling intervals of 8 h or longer. However, one recent study evaluated finished effluent samples for a suite of TOrCs on an hourly basis (Nelson et al., 2011). These high-resolution finished effluent samples characterize the temporal variability of environmental discharges, but they do not indicate the temporal variability of the influent mass loadings due to attenuation during treatment. Another recent study indicated that influent TOrC concentrations may vary on extremely short time scalesdeven as short as 2 mindand this may bias many of the recent wastewater monitoring studies (Ort et al., 2010). The authors emphasized that influent wastewater is “composed of a number of intermittently discharged, individual wastewater packets from household appliances, industries, or subcatchments” (Ort et al., 2010). The authors demonstrated that such temporal variation exists by monitoring TOrCs at time scales that could capture a single toilet flush. As noted in their study, such resolution is often limited by the costly, labor-intensive analytical methods necessary to detect trace concentrations of organic contaminants in wastewater. In fact, the authors indicated that only two previous studies had reported TOrC concentrations with sufficient temporal resolution (Ort et al., 2005; Ort and Gujer, 2006). The extent of temporal variation is dependent on the characteristics of the target compounds and the size of the service area for a particular wastewater treatment plant, thereby accounting for the number of toilet flushes containing the compounds of interest (Ort et al., 2010). In a small
catchment, contrast media used for magnetic resonance imaging will demonstrate much more temporal variability than compounds with more widespread human consumption (Joss et al., 2005; Ort et al., 2010; Nelson et al., 2011), such as non-steroidal anti-inflammatory drugs (Ternes, 1998; Joss et al., 2005). In particular, X-ray contrast media are more prevalent on weekdays when most scheduled medical appointments occur (Nelson et al., 2011). Ort et al. (2010) also emphasized that sampling uncertainty will even be a factor for large systems. Therefore, additional studies are necessary to characterize the temporal variation that is generally lost in large composite samples. Days of the week, seasons, and even special events may warrant design or operational considerations given their potential for unusual flow patterns and contaminant loadings. Nelson et al. (2011) reported significant concentration spikes for the insect repellant N,N-diethyl-meta-toluamide (DEET) during the warmer months when mosquitoes are most prevalent. With respect to special events, a recent article documented spikes in wastewater flows caused by different stages of a major auto race in Speedway, Indiana, USA (Enfinger and Stevens, 2011). Another study evaluated days of the week, seasons, and winter holidays for their effects on illicit drug concentrations in Spanish surface water (Huerta-Fontela et al., 2008). The authors observed higher illicit drug concentrations, including amphetamine-type stimulants, cocaine, and cocaine metabolites, on weekends compared to weekdays, and the authors also observed the highest concentrations in the winter, particularly after the Christmas and New Year holidays (Huerta-Fontela et al., 2008). These weekly fluctuations and holiday-specific spikes are supported by other studies of illicit drug use in Canada and Spain (Metcalfe et al., 2010; Postigo et al., 2011). The current study addresses some of these issues, including high-resolution temporal variability and the effects of a special event. This study presents flow data and corresponding influent and effluent mass loadings of a suite of prescription pharmaceuticals, potential EDCs, and illicit drugs at a wastewater treatment plant in a major metropolitan area in the United States (U.S.). Samples were collected during the National Football League’s Super Bowl weekend in addition to a baseline weekend to compare flows and mass loadings during “special event” versus “normal” operational conditions. Although the game was not held in the study area, the Super Bowl causes a tremendous spike in tourism and associated wastewater flows. The additional wastewater flows pose potential operational issues for local wastewater treatment plants, including fluctuations in the loadings of prescription and illicit drugs. This study characterizes these issues and provides further evidence of the importance of sample collection strategies in accurately characterizing TOrC concentrations in wastewater, as emphasized in Ort et al. (2010).
2.
Materials and methods
2.1.
Sampling location
Samples were collected from a municipal wastewater treatment plant with an average daily flow of approximately
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
380,000 m3/day (100 million gallons per day (MGD)). The service area of the wastewater treatment plant is approximately 505 km2 (195 mi2) with a total population of approximately 1 million people. More than 99% of the flow is delivered to the treatment plant by 3206 km (1992 mi) of gravity sewer lines, while the remaining portion is delivered by continuously operated lift stations and 64 km (40 mi) of pressurized lines. The flow rates at the lift stations remain relatively constant throughout the day. The principal treatment train consists of bar screens; grit removal; primary clarification with ferric chloride addition; activated sludge with full nitrification (NH3,eff <0.1 mg-N/L), partial denitrification, and biological phosphorus removal (TPeff < 100 mg/L); secondary clarification; dual-media filtration with alum addition; and UV disinfection (UV dose ¼ 40 mJ/cm2). The activated sludge process is generally operated with a solids retention time (SRT) of 7 days. Approximately 50% of the secondary effluent is diverted to an advanced water treatment plant, which includes coagulation/flocculation, sedimentation, dualmedia filtration, and UV disinfection or chlorination (depending on the final use). A simplified treatment schematic for the principal treatment train is provided in Fig. S1 (Supplementary data). The chlorine-disinfected effluent from the advanced water treatment plant, which constitutes an extremely small portion of the overall flow, is used for reclaimed water applications (e.g., golf course irrigation). The UV-disinfected effluent from the principal treatment train (z50% of the flow) and the advanced water treatment plant (z50% of the flow) is discharged into a local wash and ultimately a drinking water reservoir. Since the reservoir is the immediate drinking water source for the study area and millions of people downstream, this facility and the other local wastewater treatment plants are contributors to indirect potable reuse.
2.2.
Sample collection and preservation
Thirty-minute, composite samples were obtained by diverting a constant, continuous side stream of primary clarifier effluent. Although the sampling was not flow-proportional, the relative standard deviation (RSD) of the flow rate over the 30-min sampling periods was less than 5%. These wastewater “influent” samples were collected over 12-h periods on February 7th/8th, 2010 (Super Bowl), and March 7th/8th, 2010 (baseline), for a total of 24 samples per sample event. The samples were collected from 3:00 PM Sunday afternoon to 3:00 AM Monday morning to capture any wastewater effects related to Super Bowl viewing (all times are local and based on Pacific Standard Time (PST)). The 12-h time period was intended to account for the duration of pregame, Super Bowl, and postgame activities in addition to potential lag times in the sewer system. March 7th/8th was selected as a baseline weekend since there were no large events occurring in the area. In order to assess treatment efficacy, corresponding finished effluent samples from the principal treatment train were also collected during the Super Bowl and baseline weekends, but the sampling times were adjusted to account for the hydraulic retention time (HRT) of the wastewater treatment plant (z9 h).
5401
Finished effluent samples were collected from 12:00 AM Monday morning to 12:00 PM Monday afternoon. All samples were collected with refrigerated composite samplers in amber glass bottles preserved with 1 g/L of sodium azide as a biocide and 50 mg/L of ascorbic acid to quench any potential oxidant residual. Samples were stored at 4 C before laboratory filtration with 0.7-mm GF/F filters (Whatman, Piscataway, NJ, USA), on-line solid phase extraction (for PPCPs/ EDCs), and analysis. Prior to analysis, a separate aliquot of sample was prepared for each analytical method, spiked with an appropriate stock solution of isotopically labeled standards, and placed into 2-mL autosampler vials.
2.3.
Target compounds
The following pharmaceuticals and potential EDCs were monitored during the study: atenolol, atrazine, carbamazepine, DEET, meprobamate, phenytoin, primidone, sulfamethoxazole, tris-(2-chloroethyl)-phosphate (TCEP), and trimethoprim. Except for atrazine, which was purchased from ChemService (West Chester, PA, USA), all of these compounds were obtained from SigmaeAldrich (St. Louis, MO, USA). Meprobamate-d3, sulfamethoxazole-d4, and trimethoprim-d9 were obtained from Toronto Research Chemicals (Ontario, Canada). Phenytoin-d10, atrazine-d5, and atenolol-d7 were obtained from C/D/N Isotopes (Pointe-Claire, Canada). DEET-d6, carbamazepine-d10, and primidone-d5 were obtained from Cambridge Isotope Laboratories (Andover, MA, USA). TCEP-d12 was synthesized by Isotec (St. Louis, MO, USA). All concentrated stock solutions were prepared in methanol and stored at 20 C. Working stock solutions were prepared frequently in either reagent water or methanol and stored at 4 C. All solvents were trace analysis grade from Burdick and Jackson (Muskegon, MI). Reagent water was obtained using a Milli-Q Ultrapure Water Purification System (Millipore, Bedford, MA, USA). The following illicit drugs and metabolites were monitored during the study: methamphetamine and its metabolite amphetamine; cocaine and its metabolites ecgonine, ecgonine methyl ester, benzoylecgonine (BZE), and norcocaine; 3,4methylenedioxymethamphetamine (MDMA) and its metabolite 3,4-methylenedioxyamphetamine (MDA); heroin and its metabolites 6-acetylmorphine and morphine; and D-9tetrahydrocannabinol (THC) and its metabolite 11-hydroxyD-9-THC. All illicit drug standards and isotopically labeled standards were obtained from Cerilliant (Austin, TX, USA) as individual concentrated stock solutions in either methanol or acetonitrile.
2.4.
Analysis of pharmaceuticals and potential EDCs
Extraction and analysis for the pharmaceuticals and potential EDCs were performed using on-line solid phase extraction and liquid chromatographyetandem mass spectrometry (SPEeLCeMS/MS). Following laboratory filtration with 0.7-mm GF/F filters, the extraction was performed using a Symbiosis (Spark Holland, Emmen, the Netherlands) automated on-line solid phase extractor and a 4000 QTRAP triple quadrupolelinear ion trap hybrid mass spectrometer (ABSCIEX, Foster City, CA, USA). Full method descriptions and parameters, including details for the on-line SPE method, have been
5402
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
described previously (Trenholm et al., 2009). Briefly, for each extraction, a 1-mL sample volume was loaded onto a conditioned HLB cartridge. The analytes were eluted from the cartridge with 100 mL of methanol directly into the LC mobile phase just prior to the LC column. Separation was performed on a 150 4.6-mm Luna C18(2) column with a 5-mm particle size (Phenomenex, Torrance, CA, USA) and a mobile phase consisting of 5 mM ammonium acetate in DI water (A) and methanol (B) gradient. The flow rate was maintained at a constant 800 mL/min. All samples were analyzed using positive electrospray ionization (ESI) and tandem mass spectrometry with multiple reaction monitoring (MRM). Two MS/ MS transitions were used for each compound for quantitation and confirmation. Quantitation was performed using isotope dilution, which involves the correction of target compound concentrations based on the recovery of spiked isotopically labeled standards (Vanderford and Snyder, 2006). The target compounds, method reporting limits (MRLs), and MRM transitions are listed in Table S1 (Supplementary data). Fig. S2 (Supplementary data) is an example chromatogram of a 100ng/L calibration standard; for clarity, isotopically labeled standards and confirmation transitions were not shown.
The effluent mass loading rates were actually estimates because effluent samples were only collected from the principal treatment train (z50% of the overall flow). Effluent concentrations were not determined for the advanced water treatment plant (z50% of the overall flow). However, coagulation/flocculation and sedimentation are the only additional processes at the advanced water treatment plant when considering the water discharged to the wash (i.e., UVdisinfected effluent). Coagulation/flocculation and sedimentation have been shown to be highly ineffective for TOrC removal so it is reasonable to assume the TOrC concentrations in both effluents were similar (Westerhoff et al., 2005; Snyder et al., 2007). As a result, the concentrations in the principal treatment train effluent were extended to the entire flow to provide an estimate of the effluent mass loading rate, or environmental discharge rate. Total influent and effluent mass loadings over the 12-h sampling period were calculated as follows (with appropriate unit conversions for time):
2.5.
3.
Results and discussion
3.1.
Wastewater flow data
Analysis of illicit drugs and metabolites
The illicit drugs and their major metabolites were analyzed by liquid chromatographyetandem mass spectrometry (LCeMS/ MS) using a CTC Autosampler (CTC Analysis, Zwingen, Switzerland), an Agilent 1100 LC Binary Pump (Palo Alto, CA, USA), and an ABSCIEX 4000 QTRAP mass spectrometer. Following laboratory filtration with 0.7-mm GF/F filters, samples were analyzed using direct injection without solid phase extraction. All analytes were monitored using positive ESI with MRM. A 100-mL sample loop (sample volume) was used for each injection. Separation was performed on a 150 4.6-mm Allure Biphenyl column with a 5-mm particle size (Restek, Bellefonte, PA, USA) at a temperature of 40 C. A 0.1% formic acid in reagent water solution (A) and methanol (B) gradient was used for LC mobile phases at a constant 800 mL/min flow rate. Full method details (with slight modifications) are described elsewhere (Trenholm and Snyder, 2011). The target compounds, MRLs, and MRM transitions used for quantitation and confirmation are listed in Table S2 (Supplementary data). Fig. S3 (Supplementary data) is an example chromatogram of a 500-ng/L calibration standard.
2.6. Mass loading rates and environmental discharge calculations Mass loading rates provide a valuable indication of the treatment needs and environmental discharges from wastewater treatment plants due to their incorporation of both flow rate and contaminant concentrations. This study presents temporal variations in mass loading rates (g/day) for influent and effluent samples from the study site. The mass loading rates were calculated as follows, where Ci is the 30-min composite concentration for each sample and Qi is the average flow rate over each 30-min sampling period: Mass loading rateðg=dayÞ ¼ Ci Qi
(1)
Total mass loadingðgÞ ¼
24 X ðMass loading rateÞi 30 min
(2)
i¼1
Flow data from January 31 to February 1 (baseline), February 7 to 8 (Super Bowl), and March 7 to 8 (baseline) 2010 are provided in Fig. 1 to illustrate the effects of Super Bowl tourism on wastewater influent flow rates. Flow data from January 25 to 26 (baseline), February 1 to 2 (Super Bowl), and February 8 to 9 (baseline) 2009 are also provided to determine whether flow phenomena are consistent between years. The graphs indicate that the flow patterns for both baseline weekends (i.e., before and after the Super Bowl) were relatively consistent between 2009 and 2010. The baseline flows fluctuated between 95 and 120 MGD between 3:00 PM and 11:00 PM and then dropped to 60 MGD between 11:00 PM and 3:00 AM. In both years, the flows on Super Bowl Sunday were consistently higher (maximum differential of approximately 10 MGD) between 3:00 PM and 5:00 PM, but they were also consistently lower (maximum differential of approximately 10 MGD) between 7:00 PM and 9:00 PM, as indicated by the boxes in Fig. 1. Therefore, restroom use likely peaked immediately before the game started and then reached a minimum immediately after the game ended. Outside of these particular time periods, the Super Bowl and baseline flow data were relatively similar.
3.2.
Pharmaceuticals and potential EDCs
3.2.1.
Wastewater influent
Fig. 2 illustrates the influent mass loading rates for the target pharmaceuticals and potential EDCs over the sampling period. Atrazine was not detected in any of the samples at a MRL of 10 ng/L so it was excluded from the results. Average loading rates, standard deviations, and RSDs are also provided in Table 1. Relative to the influent concentrations (Fig. S4 and Table S3; Supplementary data), the influent mass loading
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
A
5403
2009
140
Flow Rate (MGD)
120 100 80 60 40 20
3: 00 P 3: M 30 P 4: M 00 P 4: M 30 P 5: M 00 PM 5: 30 P 6: M 00 P 6: M 30 PM 7: 00 P 7: M 30 PM 8: 00 P 8: M 30 P 9: M 00 PM 9: 30 10 P M :0 0 10 PM :3 0 11 PM :0 0 11 PM :3 0 12 PM :0 0 12 AM :3 0 A 1: M 00 A 1: M 30 A 2: M 00 A 2: M 30 A 3: M 00 A M
0
Time Baseline Before Super Bowl
B
Super Bowl
Baseline After Super Bowl
2010 140
Flow Rate (MGD)
120 100 80 60 40 20
3: 00 P 3: M 30 P 4: M 00 P 4: M 30 P 5: M 00 P 5: M 30 P 6: M 00 P 6: M 30 P 7: M 00 P 7: M 30 P 8: M 00 P 8: M 30 P 9: M 00 P 9: M 30 10 PM :0 0 10 PM :3 0 11 PM :0 0 11 PM :3 0 12 PM :0 0 12 AM :3 0 A 1: M 00 A 1: M 30 A 2: M 00 A 2: M 30 A 3: M 00 A M
0
Time Baseline Before Super Bowl
Super Bowl
Baseline After Super Bowl
Fig. 1 e Influent wastewater flow data for the study site during the (A) 2009 and (B) 2010 Super Bowls. The boxes represent flow periods during which the Super Bowl appears to have a discernable effect in comparison to the baseline weekends.
rates were slightly higher at the beginning of the sampling period and slightly lower at the end of the sampling period due to the change in flow rate. The steady decrease at the end of the day was expected due to decreased water use (i.e., toilet flushing) while many people were sleeping. There was a distinct difference in the magnitudes of the influent mass loading rates for the various compounds, and some compounds experienced more dramatic fluctuations than others on an absolute basis. In fact, the influent mass loading rate for atenolol nearly doubled several times over the two sampling weekends. DEET, primidone, sulfamethoxazole, TCEP, and trimethoprim also varied greatly throughout the sampling period. With respect to magnitude, the target compounds could be classified as follows: atenolol was always present at the highest influent mass loading rate (400e1100 g/ day); sulfamethoxazole, meprobamate, and trimethoprim
were grouped together in the middle (200e600 g/day); and carbamazepine, phenytoin, primidone, DEET, and TCEP were present at the lowest influent mass loading rates (0e200 g/ day). However, there were several instances where the influent mass loading rates for phenytoin, DEET, and TCEP spiked above 200 g/day. Furthermore, some compounds spiked at similar times (e.g., atenolol and primidone during the Super Bowl; atenolol, TCEP, and phenytoin during the baseline weekend), but the trends were not entirely consistent. Nelson et al. (2011) reported their temporal variability in terms of RSD. The RSDs for the current study are provided in Table 1, and it is important to note that the temporal fluctuation in flow rate accounted for 13e15% of the total variability in the mass loading rates. The RSDs actually contradict the visual observations in that the compounds with the highest
5404
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
Fig. 2 e Influent loading rates of pharmaceuticals and EDCs during the (A) Super Bowl and (B) baseline weekends in 2010.
influent mass loading rates appeared to show the greatest variability (e.g., atenolol and sulfamethoxazole), but it was actually the compounds with the lowest influent mass loading rates that had the highest RSDs (e.g., primidone, phenytoin, and DEET). This also occurred in the Nelson et al. (2011) study as the authors noted that higher baseline concentrations were often linked to lower RSDs despite significant changes in concentration. Therefore, RSD may be more appropriate for TOrCs present at similar concentrations, while standard deviation may be more appropriate for TOrCs with a wide range of concentrations. With respect to the special event effect, there was not a distinct difference in influent mass loading rates for the Super Bowl and baseline weekends. This is apparent in the
time series plots and the average influent mass loading rates over the 12-h sampling period (Table 1). As shown in Fig. S8C (Supplementary data), the influent mass loading rate for atenolol was slightly elevated prior to 10:00 PM in the Super Bowl samples, while the baseline weekend demonstrated higher influent mass loading rates after 10:00 PM. Although atenolol spiked several times during the Super Bowl period, there was a similar spike for both weekends near 10:00 PM, which was followed by a steady decline thereafter. The trends for the other compounds were not as dramatic. For compounds that are prescribed or administered by medical personnel, the total wastewater loadings were generally not affected by the Super Bowl, but the game may have affected dosing schedules and their arrival times at the
5405
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
Table 1 e Summary of flow rate and mass loadings during sampling periods. Compound
Super Bowl
Baseline d
Influent
Effluent
Effluentd
Influent
Average RSDb Total Average RSD Total Average RSD Total Average RSD Total loading loading loading loading loading loading loading loading rate (g/day)a (g)c rate (g/day) (g) rate (g/day) (g) rate (g/day) (g) Pharmaceuticals and potential EDCs Atenolol 713 247 Carbamazepine 38 16 DEET 63 52 Meprobamate 313 62 Phenytoin 30 10 Primidone 68 54 Sulfamethoxazole 348 68 TCEP <94e Trimethoprim 248 54
35% 43% 82% 20% 33% 79% 19% N/Ae 22%
349 18 30 153 15 33 170 <45e 121
Illicit drugs and metabolites (indented and italicized) Methamphetamine 806132 16% 393 Amphetamine 114 17 14% 56 Cocaine 294 70 24% 142 Ecgonine 271 52 19% 132 Ecgonine methyl ester 161 27 16% 78 Benzoylecgonine 718 142 20% 349 Norcocaine 72 34% 3 MDMA 106 30 28% 51 MDA 17 3 17% 8 Heroin <10e N/Ae <5e Morphine 231 112 48% 113
Flow rate a b c d e f
Average flow (MGD)a 101 15
RSD
15%
Total volume (MG) 49
42 25 60 10 57 9 139 39 47 10 51 9 455 101 157 42f 45 19
59% 17% 17% 28% 22% 18% 22% 27% 43%
21 29 28 68 23 25 222 77 22
700 152 37 12 71 38 301 59 37 57 42 18 414 90 <128e 262 45
22% 33% 53% 19% 156% 43% 22% N/Ae 17%
340 18 35 146 18 20 202 <62e 128
48 22 74 9 71 16 94 21 42 7 112 21 504 91 118 24f 23 7
46% 13% 22% 22% 18% 19% 18% 21% 31%
23 36 35 46 20 55 246 57 11
<18e <10e <4e <19e <10e <10e <4e 30 12 <19e <10e <19e
N/Ae N/Ae N/Ae N/Ae N/Ae N/Ae N/Ae 40% N/Ae N/Ae N/Ae
<9e <5e <2e <9e <5e <5e <2e 15 <9e <5e <9e
930 154 125 23 295 47 266 40 135 23 494 82 72 97 43 18 5 <10e 269 61
17% 18% 16% 15% 17% 17% 28% 44% 29% N/Ae 23%
451 61 143 129 66 240 3 47 9 <5e 131
<10e <10e <4e <19e <10e <10e <4e 23 8 <19e <10e <19e
N/Ae N/Ae N/Ae N/Ae N/Ae N/Ae N/Ae 37% N/Ae N/Ae N/Ae
<5e <5e <2e <9e <5e <5e <2e 11 <9e <5e <9e
Average Flow (MGD) 101 15
RSD
Average flow (MGD) 103 13
RSD
Total volume (MG) 50
Average flow (MGD) 103 13
RSD
Total volume (MG) 15% 49
13%
Total volume (MG) 13% 50
Average of 24 sampling periods one standard deviation. RSD ¼ relative standard deviation ¼ (one standard deviation)/(average). Total over 12-h sampling period. Estimated loading rates and loadings (see Section 2.5 for full explanation). Some or all samples were <MRL. All samples >MRL in effluent.
wastewater treatment plant. In particular, atenolol appears to be a relatively consistent outlier in terms of magnitude and variability, which may be partially attributable to dosing schedules. In contrast to the other prescribed pharmaceuticals, which are administered in 2e4 doses throughout the day, atenolol is administered in a single dose. Therefore, the large atenolol spikes may have been attributable to initial metabolism of the daily dose, while the other loads were related to excretion of residual concentrations throughout the day. Furthermore, some prescribed pharmaceuticals, including atenolol, primidone, and phenytoin, spiked at similar times, which might be indicative of large point source discharges (e.g., hospitals). The spikes in TCEP might also be linked to hospital wastewater (Stapleton et al., 2011) or industrial manufacturing, but the available literature is insufficient to develop explanations with confidence. Similar to Nelson et al. (2011), this study was only intended to characterize the temporal variability of certain compounds so the exact reason(s) for the variability is unclear. A study focused on pharmacokinetics and consumer behavior would be necessary to develop more definitive explanations.
3.2.2.
Wastewater effluent
Fig. 3 provides estimated (see Section 2.5) effluent mass loading rates for the target compounds during the Super Bowl and baseline weekends. Average loading rates, standard deviations, and RSDs are also provided in Table 1. Fig. S5 and Table S3 (Supplementary data) provide actual effluent concentrations. With the exception of sulfamethoxazole, Fig. S5 and Fig. 3 indicate that nearly all of the target TOrCs were present at less than 300 ng/L in the finished effluent, which corresponds to an estimated effluent mass loading rate of <150 g/ day for each compound. The aforementioned discrepancy between RSD and absolute variability was also apparent in the effluent samples, particularly for sulfamethoxazole. Considering the special event effect, the Super Bowl effluent mass loading rates were slightly higher for TCEP and meprobamate, while the baseline effluent mass loading rates were slightly higher for primidone and sulfamethoxazole. Whether due to mixing or treatment, the treatment train seemed to reduce the temporal variability for many of the target compounds. However, the concentrations of some compounds, particularly sulfamethoxazole, actually increased
5406
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
Fig. 3 e Estimated effluent loading rates of pharmaceuticals and EDCs during the (A) Super Bowl and (B) baseline weekends in 2010.
through the treatment plant. For the more biologically recalcitrant compounds, that may have been partially attributable to sampling error. In addition, the removal of sulfamethoxazole during secondary treatment is reported to be highly variable (Suarez et al., 2010), and some studies indicate that sulfamethoxazole metabolites may cleave conjugated functional groups during biological treatment and subsequently be detected as the parent compound (Joss et al., 2005; Radjenovic et al., 2009). This may provide a partial explanation for the high effluent concentrations for sulfamethoxazole, which was clearly an outlier compared to the other target compounds. However, the steady decline in sulfamethoxazole concentrations coupled with the decreasing flow rate resulted in a steep decline in effluent mass loading rates over the sampling period.
With respect to TOrC treatment efficacy, the study site currently employs one treatment processdactivated sludgedcapable of achieving substantial reductions in the concentrations of certain compounds. This was apparent in the effluent concentrations and mass loading rates as the recalcitrant compounds (e.g., sulfamethoxazole, carbamazepine, phenytoin, primidone, DEET, and TCEP) (StevensGarmon et al., 2011, submitted for publication) remained relatively stable through the treatment train, whereas the compounds susceptible to either biotransformation or sorption (e.g., atenolol and trimethoprim) (Stevens-Garmon et al., 2011, submitted for publication) experienced high removals. The generally recalcitrant compound meprobamate (StevensGarmon et al., 2011, submitted for publication) experienced
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
relatively high removals, but this particular compound still had one of the higher effluent concentrations. Therefore, biologically amenable compounds like atenolol and trimethoprim are useful indicators of secondary treatment efficacy, whereas recalcitrant compounds like meprobamate, primidone, DEET, sulfamethoxazole, and carbamazepine may be useful indicators of anthropogenic contamination in the environment. The most appropriate indicators of anthropogenic contamination may differ between sites depending on each facility’s unit treatment processes. If some form of oxidation is employed, the list of environmental indicators would have to be altered (e.g., destruction of sulfamethoxazole with chlorination and carbamazepine with ozonation). Primidone (Guo and Krasner,
5407
2009) and meprobamate appear to be conservative indicators in nearly all applications, while compounds like DEET might be affected by geographic and seasonal variability due to climate and usage patterns. Treatment efficacy, environmental discharges, and the associated indicator concept are reflected in Table 1, which provides total influent and effluent mass loadings over the 12-h sampling periods. A total influent loading is not provided for TCEP because some samples were <MRL. All other pharmaceuticals and potential EDCs had reportable concentrations in the influent and effluent over the entire sampling period so total loading estimates could be developed. As mentioned earlier, different unit treatment processes, particularly those related to disinfection, will have
Fig. 4 e Influent loading rates of illicit drugs and their metabolites during the (A) Super Bowl and (B) baseline weekends in 2010.
5408
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
a significant effect on effluent mass loadings or environmental discharges. For example, the effluent mass loading for sulfamethoxazole and trimethoprim could be reduced dramatically with chlorination. The effluent mass loadings of sulfamethoxazole, carbamazepine, trimethoprim, and, to a lesser extent, atenolol, DEET, meprobamate, phenytoin, and primidone could be reduced with ozonation. Cost effective mitigation strategies for TCEP are still lacking.
3.3.
Illicit drugs and metabolites
3.3.1.
Wastewater influent
Fig. 4 illustrates the influent mass loading rate for each of the illicit drugs and metabolites over the sampling period.
Average loading rates, standard deviations, and RSDs are also provided in Table 1. Heroin, acetylmorphine, THC, and hydroxy-THC were <MRL (i.e., 25, 25, 100, and 100 ng/L, respectively) in all samples so they were excluded from the results. According to Postigo et al. (2011), these particular compounds and metabolites are characterized by extremely low excretion rates (<3%) relative to the other illicit drugs. Relative to the influent concentrations (Fig. S6 and Table S3; Supplementary data), the influent mass loading rates were slightly higher at the beginning of the sampling period and slightly lower at the end of the sampling period due to the change in flow rate. With the exception of methamphetamine, BZE, and isolated spikes of morphine and cocaine, the illicit drugs and metabolites were present at influent mass loading
Fig. 5 e Estimated effluent loading rates of illicit drugs and their metabolites during the (A) Super Bowl and (B) baseline weekends in 2010.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
rates less than 400 g/day. The higher influent mass loading rates for methamphetamine and BZE (400e1,100 g/day) may be partially attributable to their higher excretion rates (43% and 45%, respectively) (Postigo et al., 2011). Methamphetamine was similar to the pharmaceutical atenolol in that it was the most prevalent compound (600e1200 g/day) and experienced the most dramatic fluctuations in concentration and influent mass loading rate. It is also important to note that some metabolites are consumed directly or are metabolites of other compounds, specifically in the case of morphine (Postigo et al., 2011). Despite being a major metabolite of heroin, morphine can also be administered directly in medical facilities for pain mitigation. However, the spikes in morphine did not always coincide with the dramatic spikes in atenolol, primidone, and phenytoin, which may negate the aforementioned hospital discharge theory. However, it is not possible to differentiate illicit and prescribed morphine use in the current study. Similar to the pharmaceuticals and potential EDCs, there was no discernable difference between the two weekends for most of the illicit drugs and metabolites. Of the 24 pharmaceuticals, potential EDCs, illicit drugs, and metabolites monitored during this study, only the cocaine metabolite BZE experienced a conspicuous difference in influent loading rates between the two weekends. Influent mass loading rates for BZE during the Super Bowl ranged from 500 to 1100 g/day with an average of 718 g/day, whereas the influent mass loading rates for the baseline weekend ranged from only 300 to 700 g/ day with an average of 494 g/day. Although not as apparent as the difference for BZE, influent mass loading rates for methamphetamine were slightly elevated in the baseline weekend as compared to the Super Bowl. The trends for BZE and methamphetamine are illustrated in Fig. S8A and B (Supplementary data), respectively. A larger sample size would be necessary to reach any definitive conclusions, but the data suggest that drug abuse may shift from methamphetamine to cocaine during the Super Bowl.
3.3.2.
Wastewater effluent
Fig. 5 provides estimated (see Section 2.5) effluent mass loading rates for the illicit drugs and metabolites during the Super Bowl and baseline weekends. Average loading rates, standard deviations, and RSDs are also provided in Table 1. Fig. S7 and Table S3 (Supplemental data) provide actual effluent concentrations. Methamphetamine and MDMA were the only illicit drugs detected in the effluent samples, but methamphetamine was only present in a subset of those samples. The Super Bowl effluent samples contained slightly higher concentrations of methamphetamine and MDMA in comparison to the baseline weekend, and both compounds demonstrated a relatively constant decline in both effluent mass loading rates and concentrations over the sampling period. Methamphetamine and MDMA were detected at maximum concentrations of 86 and 118 ng/L, respectively, in the Super Bowl samples and 32 and 83 ng/L, respectively, in the baseline samples. The corresponding maximum effluent mass loading rates were 41 and 56 g/day, respectively, for the Super Bowl samples and 14 and 36 g/day, respectively, for the baseline samples. With respect to treatment efficacy, conventional wastewater treatment was highly effective in removing the
5409
illicit drugs and metabolites based on the number of compounds present at concentrations <MRL. Methamphetamine and MDMA were slightly more resistant to conventional wastewater treatment, which may be related to their similar structures (e.g., branched amines). Metcalfe et al. (2010) reported similar removal trends for illicit drugs, and their study also indicated that methamphetamine and MDMA were slightly more resistant to secondary treatment than the other target compounds. Treatment efficacy and environmental discharges are reflected in Table 1, which provides total influent and effluent mass loadings over the 12-h sampling periods. Total influent loadings for heroin and total effluent loadings for all of the compounds except MDMA are not provided because some or all of the samples were <MRL. Due to the extremely low effluent loads to the environment, this set of target compounds does not provide any suitable candidates for indicators of anthropogenic contamination.
4.
Conclusion
Temporal variations in wastewater flows are common phenomena, but the corresponding effects on TOrCs are not entirely understood. The intent of this study was to evaluate high-resolution temporal variability in TOrC concentrations and mass loading rates during a baseline weekend and a special event. In support of the available literature, this study suggests that temporal variability can be a significant factor for TOrC mitigation efforts in that treatment designs may be based on composite samples that do not accurately characterize the discrete nature of wastewater matrices. In other words, unit processes that target a specific effluent concentration may consistently fail to achieve their goals if the designs are based on composite samples where significant spikes are attenuated. With respect to the current study, the Super Bowl flows were consistently higher than the baseline weekends during the early part of the sampling period but consistently lower during the middle part of the sampling period. Although there were interesting temporal variations for some compounds, particularly atenolol, the data did not indicate any significant effect of the Super Bowl on the loadings or loading rates of many of the compounds. The unique loading profile for atenolol may have been affected by its once-daily dosing schedule compared to the other prescription pharmaceuticals that are dosed multiple times throughout the day. Based on the BZE data, limited evidence suggests that cocaine use was elevated during Super Bowl weekend as compared to the baseline, whereas methamphetamine use was slightly lower. It is also interesting to note that the influent loadings of prescription pharmaceuticals were generally similar in magnitude to those of the illicit drugs and metabolites. However, conventional wastewater treatment was more effective in removing the illicit drugs and metabolites targeted in this study. In order to more effectively evaluate the effects of holidays, seasons, special events, and other unusual circumstances, a more comprehensive database of “normal” temporal variability must be developed for a variety of compounds. These
5410
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
expanded studies should determine whether “normal” temporal variability is consistent between sites for certain compounds. These studies should also identify strategies to predict temporal variability. For example, two studies suggested that influent nitrogen or ammonia spikes might coincide with spikes in prescription pharmaceuticals (Joss et al., 2005; Nelson et al., 2011). Future studies should evaluate this correlation as a potential modeling tool and determine whether the correlation is suitable for compounds that are directly related to consumption and excretion, such as prescription pharmaceuticals, and compounds from other sources, such as TCEP. Similarly, future studies should determine whether over-the-counter medications with more widespread use exhibit similar temporal variability to prescription pharmaceuticals. Finally, the current data set indicates that some compounds show definitive spikes throughout the day, but additional studies are needed to link these spikes to particular aspects of human behavior.
Acknowledgments The authors would like to thank members of the Applied Research and Development Center at the Southern Nevada Water Authority, including Josephine Chu, Shannon Ferguson, Jasmine Koster, Roxanne Phillips, Janie Holady, Yongrui Tan, and Brett Vanderford, for all of their efforts during this study. The authors would also like to thank personnel at the study site for their assistance with experimental design, scheduling, and sampling efforts. Finally, the authors would like to thank Dr. Christoph Ort from Eawag for reviewing the study and providing valuable comments.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2011.07.020.
references
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health Perspect. 107, 907e938. Donn, J., Mendoza, M., Pritchard, J., 2008. Pharmaceuticals Lurking in U.S. Drinking Water. Associated Press. Enfinger, K.L., Stevens, P.L., 2011. Pit stop. Water Environ. Technol. 23, 88. Guo, Y.C., Krasner, S.W., 2009. Occurrence of primidone, carbamazepine, caffeine, and precursors for Nnitrosodimethylamine in drinking water sources impacted by wastewater. J. Am. Water Resour. As. 45, 58e67. Huerta-Fontela, M., Galceran, M.T., Ventura, F., 2008. Stimulatory drugs of abuse in surface waters and their removal in a conventional drinking water treatment plant. Environ. Sci. Technol. 42, 6809e6816. Joss, A., Keller, E., Alder, A.C., Gobel, A., McArdell, C.S., Ternes, T., Siegrist, H., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Res. 39, 3139e3152.
Lange, A., Paull, G.C., Coe, T.S., Katsu, Y., Urushitani, H., Iguchi, T., Tyler, C.R., 2009. Sexual reprogramming and estrogenic sensitization in wild fish exposed to ethinylestradiol. Environ. Sci. Technol. 43, 1219e1225. Metcalfe, C., Tindale, K., Li, H., Rodayan, A., Yargeau, V., 2010. Illicit drugs in Canadian municipal wastewater and estimates of community drug use. Environ. Pollut 158, 3179e3185. Nelson, E.D., Do, H., Lewis, R.S., Carr, S.A., 2011. Diurnal variability of pharmaceutical, personal care product, estrogen and alkylphenol concentrations in effluent from a tertiary wastewater treatment facility. Environ. Sci. Technol. 45, 1228e1234. Ort, C., Gujer, W., 2006. Sampling for representative micropollutant loads in sewer systems. Water Sci. Technol. 54, 169e176. Ort, C., Lawrence, M.G., Reungoat, J., Mueller, J.F., 2010. Sampling for PPCPs in wastewater systems: comparison of different sampling modes and optimization strategies. Environ. Sci. Technol. 44, 6289e6296. Ort, C., Schaffner, C., Giger, W., Gujer, W., 2005. Modeling stochastic load variations in sewer systems. Water Sci. Technol. 52, 113e122. Plosz, B.G., Leknes, H., Liltved, H., Thomas, K.V., 2010. Diurnal variations in the occurrence and the fate of hormones and antibiotics in activated sludge wastewater treatment in Oslo, Norway. Sci. Total Environ. 408, 1915e1924. Postigo, C., Lopez de Alda, M., Barcelo, D., 2011. Evaluation of drugs of abuse and trends in a prison through wastewater analysis. Environ. Int. 37, 49e55. 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. Schriks, M., Heringa, M.B., van der Kooi, M.M.E., de Voogt, P., van Wezel, A.P., 2010. Toxicological relevance of emerging contaminants for drinking water quality. Water Res. 44, 461e476. Snyder, E.M., Snyder, S.A., Kelly, K.L., Gross, T.S., Villeneuve, D.L., Fitzgerald, S.D., Villalobos, S.A., Giesy, J.P., 2004. Reproductive responses of common carp (Cyprinus carpio) exposed in cages to influent of the Las Vegas Wash in Lake Mead, Nevada, from late winter to early spring. Environ. Sci. Technol. 38, 6385e6395. Snyder, S., Vanderford, B., Pearson, R., Quinones, O., Yoon, Y., 2003. Analytical methods used to measure endocrine disrupting compounds in water. Pract. Periodical of Haz., Toxic, and Radioactive Waste Mgmt. 7, 224e234. Snyder, S.A., Trenholm, R.A., Snyder, E.M., Bruce, G.M., Pleus, R.C. , Hemming, J.D.C., 2008. Toxicological Relevance of EDCs and Pharmaceuticals in Drinking Water. American Water Works Association Research Foundation, IWA Publishing. Snyder, S.A., Villeneuve, D.L., Snyder, E.M., Giesy, J.P., 2001. Identification and quantification of estrogen receptor agonists in wastewater effluents. Environ. Sci. Technol. 35, 3620e3625. Snyder, S.A., Wert, E.C., Lei, H., Westerhoff, P., Yoon, Y., 2007. Removal of EDCs and Pharmaceuticals in Drinking and Reuse Treatment Processes. American Water Works Association Research Foundation, IWA Publishing. Stanford, B.D., Snyder, S.A., Trenholm, R.A., Holaday, J.C., Vanderford, B.J., 2010. Estrogenic activity of US drinking waters: a relative exposure comparison. JAWWA 102, 55e65. Stapleton, H.M., Klosterhaus, S., Keller, A., Ferguson, P.L., van Bergen, S., Cooper, E., Webster, T.F., Blum, A., 2011. Identification of flame retardants in polyurethane foam collected from baby products. Environ. Sci. Technol. 45, 5323e5331. Stevens-Garmon, J., Drewes, J., Khan, S., McDonald, J., Dickenson, E. Biotransformation of emerging trace organic compounds by wastewater activated sludge. Bioresour. Technol., submitted for publication.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 3 9 9 e5 4 1 1
Stevens-Garmon, J., Drewes, J.E., Khan, S.J., McDonald, J.A., Dickenson, E.R., 2011. Sorption of emerging trace organic compounds onto wastewater sludge solids. Water Res. 45, 3417e3426. Suarez, S., Lema, J.M., Omil, F., 2010. Removal of pharmaceutical and personal care products (PPCPs) under nitrifying and denitrifying conditions. Water Res. 44, 3214e3224. Takao, Y., Shimazu, M., Fukuda, M., Ishibashi, H., Nagae, M., Kohra, S., Tabira, Y., Ishibashi, Y., Arizono, K., 2008. Seasonal and diurnal fluctuations in the concentrations of pharmaceuticals and personal care products (PPCPs) in residential sewage water. J. Health Sci. 54, 240e243. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Res. 32, 3245e3260. Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F., Brauch, H.J., Haist-Gulde, B., Preuss, G., Wilme, U., ZuleiSeibert, N., 2002. Removal of pharmaceuticals during drinking water treatment. Environ. Sci. Technol. 36, 3855e3863.
5411
Trenholm, R.A., Snyder, S.A., 2011. Analysis of illicit drugs in water using direct-injection liquid chromatographyetandem mass spectrometry. In: Castiglione, S., Zuccato, E., Fanelli, R. (Eds.), Illicit Drugs in the Environment: Occurrence, Analysis, and Fate Using Mass Spectrometry. John Wiley & Sons, Inc, Hoboken. Trenholm, R.A., Vanderford, B.J., Snyder, S.A., 2009. On-line solid phase extraction LC-MS/MS analysis of pharmaceutical indicators in water: A green alternative to conventional methods. Talanta 79, 1425e1432. 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. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environ. Sci. Technol. 39, 6649e6663.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 2 e5 4 1 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Fluid shear influences on the performance of hydraulic flocculation systems Ian C. Tse, Karen Swetland, Monroe L. Weber-Shirk*, Leonard W. Lion School of Civil and Environmental Engineering, Cornell University, Ithaca, NY 14853, USA
article info
abstract
Article history:
Gravity driven hydraulic flocculators that operate in the absence of reliable electric power
Received 26 May 2011
are better suited to meet the water treatment needs of green communities, resource-poor
Received in revised form
communities, and developing countries than conventional mechanical flocculators.
7 July 2011
However, current understanding regarding the proper design and operation of hydraulic
Accepted 29 July 2011
flocculation systems is insufficient. Of particular interest is the optimal fluid shear level
Available online 23 August 2011
needed to produce low turbidity water. A hydraulic tube flocculator was used to study how fluid shear levels affect the settling properties of a flocculated alum-kaolin suspension. A
Keywords:
Flocculation Residual Turbidity Analyzer (FReTA) was used to quantitatively compare the
Hydraulic flocculation
sedimentation velocity distributions and the post-sedimentation residual turbidities of the
Fluid shear
flocculated suspensions to see how they were affected by varying fluid shear, G, and
Sedimentation velocity
hydraulic residence time, q, while holding collision potential, Gq, constant. Results show
Velocity gradient
that floc breakup occurred at all velocity gradients evaluated. High floc settling velocities were correlated with low residual turbidities, both of which were optimized at low fluid shear levels and long fluid residence times. This study shows that, for hydraulic flocculation systems under the conditions described in this paper, low turbidity water is produced when fluid shear is kept at a minimum. Use of the product Gq for design of laminar flow tube flocculators is insufficient if residual turbidity is used as the metric for performance. At any Gq within the range tested in this study, best performance is obtained when G is small and q is long. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Surface water resources such as rivers and lakes contain considerable amounts of colloidal matter which cannot easily be removed because the parameters determining separation performance such as particle size, concentration and surface properties are often unfavorable for aggregation and sedimentation. Sustainable treatment technologies that are both economical and robust are needed in many resource-poor communities where turbid surface waters are often not treated prior to consumption, and where limited financial
resources and a lack of reliable electric power prevent the implementation of conventional water treatment plants. AguaClara, a program at Cornell University, has collaborated with the NGO Agua Para el Pueblo of Honduras to design and implement gravity powered water treatment plants in rural communities and small cities. The process train used in these plants includes a sequence of flocculation, up-flow sedimentation, and chlorination (additional information about AguaClara can be found at http://aguaclara.cee.cornell.edu). In AguaClara treatment plants, conventional mechanical flocculators are replaced with hydraulic flocculators in which the
* Corresponding author. Tel.: þ1 607 255 8445; fax: þ1 607 255 9004. E-mail address:
[email protected] (M.L. Weber-Shirk). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.07.040
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 2 e5 4 1 8
water is agitated by being routed around baffles rather than mixed by motor driven paddles. The goal of any flocculation process is to transform suspended colloidal particles into flocs that can be removed by sedimentation. The design of sedimentation tanks is dictated by the settling velocity of the flocs; as floc capture requires that the fluid residence time in a sedimentation tank or in lamellar plate or tube settlers be greater than the time required for flocs to settle to a surface. Therefore, one design goal of flocculators is to produce flocs with sufficiently high sedimentation velocities. Unfortunately, guidelines for proper design and operation of hydraulic flocculators are incomplete. Conventional designs utilize the product of the average velocity gradient (G) and hydraulic residence time (q) as a measure of the extent of flocculation in a reactor. In principle, any combination of G and q that gives the same product should work equally well (Ives, 1981). The appropriate fluid shear levels (measured as the average energy dissipation rate, ε) required at different points along a flocculator that will produce the best flocs are not well understood. It is expected that a high energy dissipation rate will increase the collision frequency and hence create large floc aggregates quickly; but on the other hand, a high energy dissipation rate will breakup large flocs. In addition, higher ε may form denser flocs (Gregory, 1998). This study evaluated the effect of ε and q on the sedimentation velocity and residual turbidity of the resulting floc suspension.
S-shaped channels that may wind horizontally or vertically. The magnitude of the energy dissipation rate (and thus the magnitude of fluid shear) can be controlled by adjusting either the flow rate through the flocculator or the spacing between the baffles. Since water treatment plants are designed to operate within a target range of flow rates determined by the needs of the communities they serve, varying the spacing between baffles is the primary method of controlling the magnitude of fluid shear in the flocculator. Hydraulic flocculators have narrow, long flow passages and thus approach plug flow. Floc growth occurs as the suspension moves through the flocculator, and thus the extent of flocculation at any position is a function of the local energy dissipation rate, ε, and the hydraulic residence time, q, required to reach that position. In laminar flow, the collision potential, a measure of the ability of the reactor to produce collisions, can be quantified by the product of the mean velocity gradient (G) and q (Ives, 1981). As noted below, the mean velocity gradient is related to ε. In a comparison of two laminar flow flocculators where one has a higher G value but shorter q, the collision potential should be equal as long as the dimensionless Gq terms are identical. The goal of this study was to evaluate the relative importance of G and q in hydraulic flocculation.
3. 2.
Theoretical considerations
Colloidal particles present in natural waters generally have negatively charged surfaces causing inter-particle repulsion that inhibits aggregation into larger particles which can be removed by gravity. Coagulants are normally added to enhance the kinetics for particle aggregation into flocs. When a coagulant such as aluminum sulfate (alum) is added to water, soluble positively charged hydrolysis species are formed that adsorb onto the colloids. In addition, precipitation of Al(OH)3(am) can occur on colloid surfaces and this solid phase is positively charged at circumneutral pH values. The neutralization of negative particle surface charge that ensues after coagulant addition is typically reflected in rapid formation of colloid aggregates or flocs. Contact of colloidal sized particles is primarily facilitated by diffusion (known as perikinetic flocculation). Fully destabilized particles aggregate as soon as they come into contact with one another (Serra et al., 2008). As flocs grow, diffusion is replaced by differential fluid velocities as the dominant particle to particle transport mechanism in orthokinetic flocculation. It has been shown that the frequency of particle collisions in orthokinetic flocculation is related to the magnitude of the energy dissipation rate, ε (Ives, 1981; Cleasby, 1984). As flocs grow larger, they become more susceptible to breakup. Eventually the particle size distribution can reach a pseudo-steady state during which breakup balances aggregation (Spicer and Pratsinis, 1996). In hydraulic flocculators, differential fluid velocities are generated by the flow of water around baffled channels. Most hydraulic flocculators have staggered baffles which form
5413
Materials and methods
Experiments were conducted using an apparatus comprised of synthetic raw water and coagulant metering systems, a coiled tube hydraulic flocculator, and a flocculation residual turbidity analyzer (FReTA) (see Fig. 1). Tse et al. (2011) provide a complete description of the experimental apparatus and methods; only flocculator length and flow rate have been changed for the data presented here. The synthetic raw water (SRW) metering system consisted of a concentrated stock suspension of kaolinite clay (R.T. Vanderbilt Co., Inc., Norwalk, CT) mixed with tap water to produce a feedback-regulated constant turbidity raw water source. Tap water characteristics are: total hardness z 150 mg/L as CaCO3, total alkalinity z 113 mg/L as CaCO3, pH z 7.7 and dissolved organic carbon z 1.9 mg/L (Bolton Point Municipal Water System, 2009). The concentrated stock and the SRW feedstock were each stirred by a variable speed electric mixer to ensure homogeneous suspensions. A float valve regulated the flow of temperature controlled (25 C) tap water into the SRW tank to maintain a constant water level. For all of the experiments performed in this study, the SRW was maintained at a constant turbidity of 50 5 NTU, which corresponded to a clay concentration of approximately 50 mg/L. Technical grade aluminum sulfate (Al2(SO4)312-16H2O) from Fisher Scientific was used as the coagulant for all experiments. The alum stock was prepared with distilled water at a concentration of 200 mg/L as Al. Based on initial settling experiments performed with a tube flocculator at a G ¼ 40 s1 and Gq ¼ 19700,an alum dose of 3.1 mg/L Al was determined by to be optimal for a SRW with a turbidity of 50 NTU. Both the SRW and the alum were metered with Cole Parmer MasterFlex L/S digital computer controlled peristaltic pumps.
5414
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 2 e5 4 1 8
clay
alum FReTA synthetic raw water
actuated ball valve
peristaltic pumps
NTU NTU
turbidimeter (feedback loop )
backwash effluent
solenoid valves
infrared turbidimeter
tight coil rapid mix settling column
tube flocculator
solenoid valve pressure sensor (measure head loss)
effluent discharge
high-pressure backwash
Fig. 1 e Schematic of the experimental assembly.
SRW (turbidity ¼ 50 NTU) and the alum stock were combined to give an alum concentration of 3.1 mg/L Al. The mixture was passed through a rapid mix unit comprised of a 120 cm segment of 4.3 mm (0.17”) ID tubing coiled around a cylinder with an outer diameter of 5 cm to ensure thorough mixing of the SRW and the alum. The mixed flow entered a 28 m flocculator divided into 6 equal segments of 466 cm each. The velocity gradient was varied from 40 to 250 s1by varying the volumetric flow rate for each of the six lengths; while the upper end of this range exceeds G values typically used for flocculation, it was chosen to ensure that breakup by floc shear would be observed. Gq values remained within a non-overlapping range at each of the flocculator lengths, because as G would increase q would decrease. In initial experiments, a minimum G value of 40 s1 was shown to be needed to keep lose flocs in suspension within the flocculator. As Owen et al., (2008) note, flocculation is frequently studied in batch reactors with offline size measurements for aggregation processes, resulting in poor control over reaction time and questionable size measurements. A tube flocculator was used because it can be idealized as a high Peclet number reactor much like a hydraulic flocculator and also because the average velocity gradient (G) in laminar tube flow is well defined (Equation (1)) (Gregory, 1981). Gs ¼
8Q 3pr3
(1)
where: Q is the volumetric flow rate and r is the inner radius of the tube. The tube flocculator consisted of a 28 m segment of 9.5 mm (3/8”)inner diameter transparent plastic tubing wrapped in
a figure eight shape around two 11 cm outer diameter parallel cylinders for structural support. The length of tubing was chosen based on Camp and Stein’s (1943) recommendation that a Gq of 20,000 is typically needed for sufficient flocculation. The diameter of the tubing was chosen to be large relative to floc diameter and to reduce the fraction of the overall residence time spent in the settling column. The settling column in the turbidimeter has a 1” diameter and a length of 12”. A large diameter tube for flocculation requires a high flow rate, which keeps the residence time in the settling column low. A differential pressure sensor was attached at each end of the tube flocculator to monitor head loss. The flocculator was coiled to reduce the effects of sedimentation. In laminar flows there are no turbulent eddies to resuspend flocs that settle on the bottom surface of the flocculator tube. In coiled tubes fluid inertia causes secondary flow patterns that consist of two vortical cells with the line of symmetry being the radius of curvature of the coil (Berger et al., 1983). The secondary flows helped resuspend flocs because the secondary flows have a velocity component that moves flocs away from the bottom of the tube. A simple helical coil was observed to concentrate the flocs into two zones that correspond to the two circulating cells. Therefore, the coils were configured into a tight figure eight pattern to disrupt the two circulating cells and cause particles to move throughout the cross-section of the tube. The secondary circulation caused by coiling acted to increase the magnitude of the average velocity gradient inside the tube. Tse et al. (2011) provide a derivation of the velocity gradient in a coiled tube, the result of which is shown in Equation (2).
5415
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 2 e5 4 1 8
Data analysis
Turbidity data was collected over 30 min at a 1 Hz sampling frequency for each flocculated suspension. Turbidity fluctuations were observed when large flocs (high sedimentation velocities) moved past the measurement area and refracted more light into the light sensor. Cheng et al. (2010) confirm the correlation between turbidity standard deviation and floc size. The large fluctuations were often problematic for data fitting routines and required smoothing. The data was first averaged over 9 s intervals and then a filter that reported the moving median value over a set of 5 averaged 9 s intervals was used to smooth the data. This smoothing technique acted to exclude extreme fluctuations while preserving the shape characteristics of the Vs and particle size distributions. The data was normalized to range between 0 and 1 by dividing turbidity values by the initial turbidity of the settling period. Normalized turbidity, recorded as a function of time, was converted into a settling velocity, Vs, by dividing the length of the settling column (16 cm) by the elapsed time. The resulting normalized turbidity vs. Vs curves were then fit with the following modified gamma distribution’s cumulative distribution function (CDF) in order to determine the mean and variance of the data: Zx F0ðx; a; b; gÞ ¼ ð1 gÞ 0
xa1
ex=b dx þ g ab GðaÞ
(3)
where: the independent variable is a base 10 logarithm of Vs, a and b are parameters of the distribution that are fitted, g is an offset parameter that accounts for the non-zero residual
40 1 30
20 0.5
Normalized Turbidity
Raw Turbidity (L axis) Normalized & smoothed (R axis)
10
0
0 0
500
1000
1500
Settling Time (sec)
B
Normalized & Smoothed Modified Gamma CDF 1
0.5
0 0.01
0.1
1
10
100
Settling Time (sec)
C 0.8
Modified gamma PDF (L axis) Modified gamma CDF (R axis)
1
0.6
0.4
0.5
Normalized Turbidity
4.
A 50
Turbidity (NTU)
where: the subscripts c and s refer to coiled and straight tubes, pffiffiffiffiffiffiffiffiffi respectively. De is the Dean number: e ¼ r=Rc Red , r is the inner radius of the tube, Rc is the radius of curvature, Red ¼ Ud=v, U is the average axial velocity, d is the inner diameter of the tube, and n is the kinematic viscosity of the fluid. The flocculation residual turbidity analyzer (FReTA) described by Tse et al. (2011)was used to measure both the sedimentation rate and the post-sedimentation residual turbidity of the effluent from the coiled tube flocculator. FReTA is capable of optically measuring floc sedimentation velocities (Vs) and residual turbidity and does so without altering or damaging the structure of the flocs in suspension. FReTA consists of three primary components: an HF Scientific MicroTOL 2 infrared turbidimeter, a transparent glass settling column, and an electrically actuated ball valve. The valve is situated atop the glass column that has been inserted through the measurement chamber of the infrared turbidimeter. For this study, a distance of 16 cm separated the bottom of the closed ball valve and the middle of the 5 mm zone monitored by the turbidimeter. Any floc contained within this length of the settling column that settled past the measurement area and crossed the beam of infrared light had its turbidity detected.
normalized turbidity, and G(a) is the gamma function. The mean Vs is equal to the product ab and the variance is equal to the product ab2. A Gamma function was selected because it better accounted for skewed distributions when observed, however it is also capable of describing a log normal
Normalized Turbidity
(2)
Normalized Turbidity/(mm/s)
1 4 2 Gc ¼ Gs 1 þ 0:033ðlogðDeÞÞ
0.2
0 0.01
0.1
1
10
0 100
Vs (mm/s)
Fig. 2 e A: raw time series turbidity data obtained from FReTA and the same data set after being normalized and median smoothed. B: normalized turbidity-Vs curve on a semi-log graph, and cumulative distribution function fitted to the data using Equation (3). C: modified cumulative distribution function and probability distribution function for the modified Gamma distribution, where the data has mean Vs [ 0.94 mm/s.
5416
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 2 e5 4 1 8
distribution that is often attributed to particle suspensions (Aksoy, 2000). Curve fitting was performed using the genfit function in Mathsoft’s MathCAD 14.0. Genfit is capable of fitting a user defined equation to a set of data points using LevenbergeMarquardt method for minimization.
5.
Length
2796 cm
2330 cm
1864 cm
1398 cm
932 cm
High Gq Low Gq
26,000 19,700
21,600 16,400
17,300 13,100
13,000 9800
8700 6600
Results
Fig. 2A shows an example of a typical data set obtained using FReTA before and after smoothing and normalization. Fig. 2B shows the smoothed and normalized data transformed into normalized turbidity as a function of settling velocity and the modified gamma distribution fit of this data. Fig. 2C shows the cumulative distribution function (CDF) and the resulting probability density function (PDF) of settling velocities obtained from the fitted gamma distribution to the data shown in Fig. 2A. The area under each PDF curve in Fig. 3 is unity as expected; however, this is visually obscured by the semi-log plot used to display the PDFs. The sedimentation velocity distributions and residual turbidities of flocs formed inside several tube flocculators (see Table 1) were measured with FReTA. For each of the tube flocculator lengths tested, the mean sedimentation velocities were obtained from the PDFs of the gamma distribution fits. Fig. 3 shows an exemplary set of PDFs obtained from the modified gamma distribution fits of experiments performed in a flocculator with Gq ¼ 19,000. Fig. 4 shows the family of Vs values obtained over the range of Gq values tested as a function of G. Each curve in Fig. 4 represents a set of flocculation experiments performed under conditions of almost identical collision potential (or Gq). If Gq accurately predicted the extent of flocculation and if floc breakup were not significant, the curves in Fig. 4 should not vary with G, but only be affected by the magnitude of the product Gq. However, each of the plots in Fig. 4 has
a significant negative slope indicating that flocs were able to grow to a larger size at lower G values as evidenced by the larger Vs. These results suggest that floc breakup was a significant factor in limiting floc size for each of the five tube flocculator lengths and for all of the velocity gradients that were tested. Results from Fig. 4 exhibit similarities to the findings of Serra et al. (2008), who showed that shear-induced breakup limited the size of latex flocs formed in various types of reactors when G was greater than 30 s1. As expected, floc sedimentation velocities generally increased with Gq (see Fig. 4), indicating that many of the flocs continued to grow when given more time to flocculate. However, floc size did not increase when additional flocculation time was provided for G > 100 s1 at Gq 15,000, as shown by the convergence of the curves for Gq equal to 15,000, 19,000 and 23,000. Shear-induced breakup prevented further growth at high G values. Floc size was independent of Gq for G > 200 s1 and Gq > 11,000 suggesting that for these conditions floc growth was limited by floc breakup. A similar convergence was observed for G > 100 s1 for Gq 15,000. The curve corresponding to Gq ¼ 7500 did not converge with any of the higher Gq curves, indicating that flocs produced under those conditions had not yet reached a breakup limited size. Nevertheless, floc breakup may have retarded floc growth at Gq ¼ 7500. Fig. 5 shows the observed residual turbidity at a sedimentation capture velocity of 0.09 mm/s as a function of G. Data sets correspond to experiments with nearly constant Gq. Residual turbidity increased with higher fluid shear. Thus, increased fluid shear not only decreased the average size and sedimentation velocity of flocs as shown above, but it resulted
G = 40 /s G = 52 /s G = 103 /s G = 175 /s G = 237 /s
2
3
Mean Vs (mm/s)
Normalized turbidity/(mm/s)
Table 1 e Gq range for each flocculator length used in this study.
1
G G G G G
2
= 23000 = 19000 = 15000 = 11000 = 7500
1
0 0.01
0.1
1
10
Vs (mm/s) Fig. 3 e Modified gamma distribution PDFs of floc sedimentation velocities from experiments performed using a tube flocculator with an average Gq of 19,000. Mean and variance statistics for each data set were calculated from similar PDF curves.
0 0
100
200
300
G (1/s) Fig. 4 e Mean sedimentation velocities plotted vs. average velocity gradients for various flocculator lengths (listed with their mean Gq values).
5417
G G G G G
30
100
= 23000 = 19000 = 15000 = 11000 = 7500
Residual Turbidity (% of initial turbidity)
Residual Turbidity (% of initial turbidity)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 2 e5 4 1 8
20
10
0 0
100
200
in higher residual turbidities as well. The observed relationships could be explained by two distinct hypotheses: (1) the breakup of flocs due to fluid shear released smaller floc fragments with poor settling properties, or (2) larger flocs that are created under low fluid shear may be more effective in sweeping up smaller particles. Fig. 5 also shows that use of the product Gq for design of laminar flow tube flocculators is insufficient if residual turbidity is used as the metric for performance. Over the range of Gq studied, best performance is obtained when G is small and q is long. The velocity gradient (G) term used in describing the magnitude of fluid shear in laminar flow tube flow was converted into energy dissipation rate (ε) using Equation (4) and the resulting data is shown in Figs. 6 and 7. 10
Vs (mm/s)
G G G G G
= 23000 = 19000 = 15000 = 11000 = 7500
1
10
10
1
10
100
(mW/kg)
Fig. 5 e Residual turbidity corresponding to a capture velocity of 0.09 mm/s (16 cm height in 30 min of sedimentation and given as percentage of the turbidity at the start of each settling period) vs. average velocity gradient for various flocculator lengths (listed with their mean Gq values).
1
= 23000 = 19000 = 15000 = 11000 = 7500
1
300
G (1/sec)
0.1
G G G G G
100
(mW/kg) Fig. 6 e Mean sedimentation velocities are plotted against the energy dissipation rate.
Fig. 7 e Residual turbidity after a 30 min settling period (given as percentage of the turbidity at the start of each settling period) vs. energy dissipation rate for various flocculator lengths (listed with their mean Gq values).
G¼
rffiffiffi ε v
(4)
where n is the kinematic viscosity. Consistent with prior reports, a power law relationship between sedimentation velocity and the energy dissipation rate is evident for hydraulic flocculators after this analysis. The curves corresponding to the three largest flocculator lengths are superimposed on top of one another in Fig. 6 at ε 3 m W/kg. Thus, increasing Gq appears to have no effect on mean floc sedimentation velocity for a given ε once breakup from shear limits the steady state floc size. Residual turbidity data from Fig. 5 are plotted against ε in Fig. 7. Best results for residual turbidity require a combination of low ε and long q. The more efficient removal of colloids at low ε could be due to the creation of larger flocs that have a higher sedimentation velocity which would increase their ability to collide with other colloids. Larger flocs also have a more porous structure as manifested by a lower fractal dimension (Li et al., 2007), which could make them effective in filtering other colloids. In Figs. 4e7 it is apparent that the Gq ¼ 7500 plot is unique since it does not converge to the same values at high energy dissipation rates or high velocity gradients. The difference may be that, at Gq of 7500, floc growth was affected but not yet completely limited by shear-induced breakup. The negative slope in Fig. 4 suggests that breakup was occurring. Floc growth was also still occurring at all measured velocity gradients because larger flocs appeared by a Gq of 11,000.
6.
Conclusions
The results shown above demonstrate that increased collision potential (increasing Gq from 11,000 to 23,000) did not increase
5418
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 2 e5 4 1 8
mean particle settling velocities when the velocity gradient was greater than 200 s1. Increased collision potential (increasing Gq from 11,000 to 23,000) also did not reduce the residual turbidity when the velocity gradient was at 250 s1. Both quickly-settling flocs and low turbidity water are best produced when the energy dissipation rates are low. If the best way to produce low turbidity water is to produce very large flocs, baffle spacing and the flow conduits between the flocculator and the sedimentation tank must be designed to ensure they do not breakup large flocs. Since large flocs settle very quickly, it is difficult to reduce the energy dissipation rate sufficiently to reduce their breakup while still maintaining flow velocities high enough to prevent sedimentation in the flow conduits. Thus, the flocculator to sedimentation tank transition may ultimately set a practical limit to the goal of creating quickly-settling flocs in flocculation. The data presented in this report were obtained under conditions of laminar tube flow. In practice, most hydraulic flocculators operate under conditions of turbulent flow. It is reasonable to expect that turbulent eddies promote orthokinetic flocculation and floc breakup in a manner that is similar to the effect of fluid shear in laminar flow. While similar trends are expected with respect to the effect of energy dissipation rate and residence time on floc characteristics under turbulent flow conditions, the temporal and spatial variability in the energy dissipation rate in turbulent flow is expected to influence the specific relationships that are observed.
Acknowledgments This work was made possible through the generous support of the Sanjuan Foundation. Partial support was also obtained from the National Science Foundation under Grant CBET0604566. Additional information on AguaClara can be found at: http://aguaclara.cee.cornell.edu
references
Aksoy, H., 2000. Use of gamma distribution in hydrological analysis. Turkish Journal of Engineering and Environmental Science 24 (2000), 419e428. Berger, S.A., Talbot, L., Yao, L., 1983. Flow in curved pipes. Annual Review of Fluid Mechanics 15, 461e512. Bolton Point Municipal Water System, 2009. City of Ithaca Water System, Cornell University Water System. Drinking Water Quality Report. Camp, T.R., Stein, P.C., 1943. Velocity gradients and internal work in fluid motion. Journal of the Boston Society of Civil Engineers 30, 219. Cheng, W., Chang, J., Chen, P., Ruey Yu, R., Huang, Y., Hsieh, Y., 2010. Monitoring floc formation to achieve optimal flocculation in water treatment plants. Environmental Engineering Science 28 (6), 523e530. Cleasby, J., 1984. Is velocity gradient a valid turbulent flocculation parameter? Journal of Environmental Engineering 110 (5), 875e897. Gregory, J., 1998. The role of floc density in solid-liquid separation,". Filtration and Separation 35 (4), 366e371. Gregory, J., 1981. Flocculation in laminar tube flow. Chemical Engineering Science 36 (11), 1789e1796. Ives, K., 1981. Orthokinetic flocculation. In: Svarovsky, L. (Ed.), Solid-Liquid Separation. Butterworths, London, pp. 86e119. Li, T., Zhu, Z., Wang, D., Yao, C., Tang, H., 2007. The strength and fractal dimension characteristics of alumekaolin flocs. International Journal of Mineral Processing 82 (1), 23e29. Owen, A.T., Fawell, P.D., Swift, J.D., Labbett, D.M., Benn, F.A., Farrow, J.B., 2008. Using turbulent pipe flow to study the factors affecting polymer-bridging flocculation of mineral systems. International Journal of Mineral Processing 87 (3e4), 90e99. Serra, T., Colomer, J., Logan, B.E., 2008. Efficiency of different shear devices on flocculation. Water Research 42 (4e5), 1113. Spicer, P.T., Pratsinis, S.E., 1996. Shear-induced flocculation: the evolution of floc structure and the shape of the size distribution at steady state,". Water Research 30 (5), 1049e1056. Tse, I.C., Weber-Shirk, M.L., Lion, L.W., 2011. Method for quantitative analysis of flocculation performance. Water Research 45 (10), 3075e3084.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Quantifying quagga mussel veliger abundance and distribution in Copper Basin Reservoir (California) using acoustic backscatter Michael A. Anderson a,*, William D. Taylor b a b
Department of Environmental Sciences, University of California, Riverside, CA 92521, USA Metropolitan Water District of Southern California, 700 Moreno Ave, La Verne, CA 91750, USA
article info
abstract
Article history:
Quagga mussels (Dreissena bugensis) have been linked to oligotrophication of lakes, alter-
Received 14 May 2011
ation of aquatic food webs, and fouling of infrastructure associated with water supply and
Received in revised form
power generation, causing potentially billions of dollars in direct and indirect damages.
29 July 2011
Understanding their abundance and distribution is key in slowing their advance, assessing
Accepted 1 August 2011
their potential impacts, and evaluating effectiveness of control strategies. Volume back-
Available online 27 August 2011
scatter strength (Sv) measurements at 201- and 430-kHz were compared with quagga mussel veliger and zooplankton abundances determined from samples collected using
Keywords:
a Wisconsin closing net from the Copper Basin Reservoir on the Colorado River Aqueduct.
Quagga mussel
The plankton within the lower portion of the water column (>18 m depth) was strongly D-shaped
quagga mussel veligers, comprising up to 95e99% of the
Veliger
dominated by
Distribution
community, and allowed direct empirical measurement of their mean backscattering
Hydroacoustics
cross-section. The upper 0e18 m of the water column contained a smaller relative
Backscatter
proportion of veligers based upon net sampling. The difference in mean volume backscatter strength at these two frequencies was found to decrease with decreasing zooplankton abundance (r2 ¼ 0.94), allowing for correction of Sv due to the contribution of zooplankton and the determination of veliger abundance in the reservoir. Hydroacoustic measurements revealed veligers were often present at high abundances (up to 100e200 ind L1) in a thin 1e2 m layer at the thermocline, with considerable patchiness in their distribution observed along a 700 m transect on the reservoir. Under suitable conditions, hydroacoustic measurements can rapidly provide detailed information on the abundance and distribution of quagga mussel veligers over large areas with high horizontal and vertical resolution. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Dreissenid mussels have been linked to oligotrophication of Lake Michigan and Lake Huron (Evans et al., 2011), alteration of aquatic food webs (MacIsaac et al., 1995; Wong et al., 2003), and fouling of infrastructure associated with water supply and
power generation, causing hundreds of millions of dollars of economic costs to these facilities through 2004 (Connelly et al., 2007). Economic loss in the U.S. due to the invasion by quagga and zebra mussels is thought to be as high as $1B per year (Pimentel et al., 2005). As a result, considerable interest exists in the U.S., Canada and elsewhere concerning the distribution,
* Corresponding author. Tel.: þ1 951 827 3757; fax: þ1 951 827 3993. E-mail address:
[email protected] (M.A. Anderson). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.018
5420
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
fate and control of quagga and zebra mussels. Quagga mussels (Dreissena bugensis) have recently invaded the western U.S., being first identified in Lake Mead in January 2007 (LaBounty and Roefer, 2007), and have quickly spread throughout the lower Colorado River, Southern California, Arizona and Nevada (Wong et al., 2010). Their occurrence in the Colorado River system thus threaten a complex and vast conveyance system that delivers water to 7 states and has a hydroelectric generating capacity exceeding 4 GW. Quagga and zebra mussels are r-strategists, with lifehistory characteristics promoting rapid population growth (Vanderploeg et al., 2002). They accomplish rapid population growth through broadcast spawning of very large numbers of eggs and sperms; a single female can release up to a million eggs to the water column that can then be externally fertilized to form trochophore larvae (Ackerman et al., 1994). The trochophore larvae quickly develop to the planktonic larval veliger stage that secrete an unadorned D-shaped CaCO3 shell that are generally 70e160 mm in height. Within 7e9 days postfertilization, a second more ornamented shell is secreted from the mantle tissue that has a more pronounced umbonal region near the hinges and is round or clam-like in profile. This umbonal veliger is somewhat larger in size (120e280 mm) and is the last free-swimming veliger stage routinely found in the plankton, although a developmental bottleneck significantly limits the number D-shaped dreissenid veligers that develop into umbonal veligers (Schneider et al., 2003). Thus the D-shaped veligers are the dominant larval form of quagga and zebra mussels present in invaded lakes. Sampling using a Wisconsin closing net revealed a strong vertical gradient in quagga mussel veliger abundance in Copper Basin Reservoir on the Colorado River Aqueduct near Lake Havasu and the California-Arizona border (Reid et al., 2010). Veligers dominated the plankton community in the lake, with very high abundances present near the thermocline (Reid et al., 2010). Sampling was restricted to 3 locations on the lake, with samples collected at 3-6 discrete depth intervals, however, thus providing information on a coarse vertical and lateral scale. The abundance and distribution of zooplankton in lakes vary widely in space and time, with patchy distributions and large gradients in populations over very short length scales (Hembre and Megard, 2003; Rinke et al., 2009), marked diel migration for many species (Masson et al., 2001), and finescale heterogeneities within communities due to fish predation and other factors (Benoit-Bird, 2009a). Much of the detailed information about the distribution of zooplankton has been determined through use of hydroacoustic measurements (Holliday and Pieper, 1995; Hembre and Megard, 2003). Sound is reflected off objects within the water column based upon their size, shape, and density and soundspeed contrasts with water. As a result, hydroacoustic measurements are commonly used in fishery assessments (Simmonds and MacLennan, 2006; Godlewska et al., 2009), although acoustic backscatter is also used to quantify abundance and distribution of zooplankton (Hembre and Megard, 2003), larval aquatic insects (Kubecka et al., 2000), and submerged aquatic vegetation (Winfield et al., 2007). Volume backscattering strength measured using hydroacoustic methods has provided detailed information about
the distribution and densities of zooplankton not possible from net sampling (Hembre and Megard, 2003), although such methods do not provide direct information about species composition within a diverse community. Acoustic backscatter by zooplankton is strongly dependent upon acoustic frequency, organism size, shape, and other factors, with stronger backscatter generally found at higher frequencies. Multifrequency backscatter measurements can, under some circumstances, provide information about mean size(s) and type(s) of zooplankton, however (Greenlaw, 1979; Greene et al., 1989; Mitson et al., 1996). Hydroacoustic measurements were conducted at Copper Basin Reservoir to test the ability of the technique to (i) provide remotely-sensed population estimates of quagga mussel veligers, (ii) provide fine-scale spatial information about their distribution, and (iii) compare day and night distributions of veligers (and other zooplankton) in the reservoir.
2.
Materials and methods
2.1.
Field sampling and measurements
Coupled net sampling and volume backscattering strength measurements were conducted on August 10-11, 2009 at Copper Basin Reservoir (Fig. 1). Copper Basin Reservoir is a 172 ha reservoir on the Colorado River Aqueduct and located near the California-Arizona border (34.2876 N, 114.2331 W) (Fig. 1). The reservoir has a mean depth of 17.5 m, high Secchi
Fig. 1 e Study site: Copper Basin Reservoir on the Colorado River Aqueduct near Lake Havasu, Bathymetric map showing sampling site for net and hydroacoustic measurements, and survey transect.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
depth (9.4 m), low mean chlorophyll concentration (5 mg L1) and short hydraulic detention time (about 6e10 d depending upon flow) (Reid et al., 2010). The Colorado River Aqueduct is operated by the Metropolitan Water District of Southern California and is one of the primary sources of drinking water for 18M residents in Southern California. Following Reid et al. (2010), a 0.3 m diameter, 63 mm mesh Wisconsin closing net was used to collect 6 m-interval samples from the surface to 36 m depth at the deepest location on the lake (Zmaxw38 m) (Fig. 1). Sampling was conducted at approximately 9:30 p.m. on August 10th (about 1 h after sunset), 12 h later the following morning (about 9:30 a.m., August 11th), and later that afternoon at approximately 1:30 p.m. Each depth interval represented a 424 L sample of water; total plankton (zooplankton and veligers) retained on the net were quantitatively transferred by rinsing with 70% ethanol into 125-mL wide mouth polypropylene bottles. Samples were stored on ice until returned to the laboratory. Hydroacoustic measurements were made immediately prior to net sampling with a BioSonics DT-X Echosounder multiplexed to a 430-kHz 7 single-beam transducer with an orientation sensor and a 201-kHz 6.6 split-beam transducer (Table 1). Both transducers were mounted to a common swivel mount secured to the bow of a 17-ft Boston Whaler and lowered 0.6 m below the water surface. A JRC 212W real-time differential global-positioning satellite (DGPS) receiver was mounted directly over the transducers and recorded differentially-corrected positions every 1-s. Data were acquired using BioSonics VisualAcquisition software on a Dell ATG laptop at a rate of 5 pps and 0.4 ms pulse duration (Table 1). The 430- and 201-kHz transducers were calibrated prior to data collection using 17 and 33 mm tungsten-carbide calibration spheres, respectively, with known target strengths. Temperature, pH, conductivity and dissolved oxygen (DO) profiles were measured using a Hydrolab DataSonde4a and Surveyor 4 to quantify thermal structure, availability of DO, and basic chemical properties in the water column. Temperature, conductivity and pH values were also used in sound speed and range (depth) calculations.
2.2.
Laboratory measurements and data analyses
Veliger enumeration was conducted using a Nikon E600 compound microscope fitted with cross-polarization and Whipple grid micrometer. One-mL samples were removed from gently inverted, mixed samples of known volume and placed on a gridded Sedgewick-Rafter counting cell. The sample was first analyzed for veligers under crosspolarization, and then inspected under phase-contrast or
Table 1 e Transducer configurations used in this study. Property Frequency (kHz) Beam angle ( ) Source level (dB/mPa) Receive sensitivity (dBC/mPa) Pulse length (ms) Pings per second (pps)
DTX-201
DTX-420
201 6.6 221.3 63.6 0.4 5
430 7.0 220.0 62.9 0.4 5
5421
brightfield for identification of other zooplankton. Size was determined on at least 10 randomly selected individuals from each taxa or group when available. Volume backscattering strength (Sv) was determined from echograms using echointegration with a 20logR time-varied gain after manual removal of fish echoes using Sonar5Pro (Balk and Lindem, 2007). Echogram analyses were conducted by averaging across layers corresponding to net-sampled depth intervals, except for the 0e6 m depth interval, where the acoustically-sampled depth interval was restricted to 1.5e6 m depth as a result of the transducer depth and nearfield effects. Volume backscatter strength is a function of the abundance of scatterers and their backscattering cross-section (Hembre and Megard, 2003). That is, Sv ¼ 10log
n X
sbs;i Ni
(1)
i¼1
where sbs,i is the backscattering cross-section of species i and N is the abundance of species i (Benoit-Bird, 2009b). Backscattering cross-sections have been empirically measured (Hembre and Megard, 2003) or theoretically calculated from scattering models (e.g., Greenlaw, 1979; Stanton, 1989) for a limited number of zooplankton. The target strength (TS, in dB) of a scatterer can in turn be calculated from sbs as (Hembre and Megard, 2003): TS ¼ 10logsbs
3.
(2)
Results
Total plankton (zooplankton and veligers) exhibited a strong vertical gradient in abundance during the 9:30 p.m. sampling on August 10, 2009 (Fig. 2a). Quagga mussel veligers were numerically the dominant plankton below 6 m depth and comprised 70e99% of the total invertebrate plankton recovered in the deeper net samples, while the surface (0e6 m) interval exhibited low abundances of all taxa (Fig. 2a). The non-veliger zooplankton community was quite modest in density in the reservoir, reflecting both low productivity (chlorophyll concentrations approximately 5 mg L1) (Reid et al., 2010), and the hydraulics of the system. That is, Copper Basin Reservoir has a short hydraulic detention (6 d when operated at the full flow capacity of the Colorado River Aqueduct), with water pumped up about 100 m from Gene Wash Reservoir through the pumps, surge chambers and penstocks of the Gene Pumping Plant. Inspection of zooplankton (chiefly Bosmina, small adult copepods and nauplii) at the inlet and outfall of the pumps indicated >90% mortality associated with pumping (unpubl. data). The short hydraulic residence coupled with the lack of viable zooplankton delivered with inflows is thought to be a chief reason for the low abundance of zooplankton in the reservoir. In contrast, mortality to veligers, as determined from ciliar or internal organelle motion, was low (<10% mortality) (unpubl. data). Mean volumetric backscatter strength measured at 201kHz (Sv201) followed trends in overall plankton abundance (Fig. 2b). Sv201 of about 101 dB was measured in the surface
5422
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
a
b
Concentration (ind L-1) 0
20
40
60
80
c
Sv201 (dB re 1 m-1)
-102
-100
-98
-89
-96
Sv430 (dB re 1 m-1) -88
-87
-86
-85
0
6
Depth (m)
12
Veligers Bosmina Cyclopoid Calanoid Nauplii Rotifers
18
24
30
36
Fig. 2 e Results from measurements made at approximately 9:00 p.m. on August 10, 2009: a) zooplankton and quagga mussel veliger abundances from net sampling, b) volume backscatter strength at 201-kHz, and c) volume backscatter strength at 430-kHz.
a general increase in Sv201 with depth and veliger abundance to a maximum value of 96.5 dB in the depth interval immediately above the thermocline (Fig. 4b). Backscatter strength at 430-kHz yielded a somewhat different distribution with depth than found the previous evening, however, with Sv430 decreasing from a maximum value of 85.6 dB in the surface layer to 87.6 dB in the 12e18 m depth range (Fig. 4c). The Sv430 values did not follow veliger abundance quite as well as Sv201, and reflected stronger contributions from zooplankton. These results suggest that Sv at 201-kHz offers some potential utility in estimating veliger abundance, although significant overestimates would be expected for surface samples or elsewhere where veligers comprise a minority of the total plankton community.
a
Temperature (oC) 0
16
20
24
28
b
Dissolved O2 (mg L-1) 0
2
4
6
8
10
6
12
Depth (m)
layer and corresponded to low overall plankton abundance, while Sv201 increased to a maximum value ofe96.9 dB at the 24e30 m depth interval (Fig. 2b) that corresponded to a total plankton abundance of 67 ind L1 (Fig. 2a). Veligers comprised >97% of the plankton community at that depth. Abundance and Sv201 both declined below that depth at the deepest sampled layer in the profile. The mean volumetric backscatter strength at 430-kHz (Sv430) was uniformly high at approximately 87.5 dB over the 12e30 m depth range, and lower both near the surface (0e12 m) and lowest depth interval (30e36 m) (Fig. 2c). Temperature profile measurements at that time revealed the presence of stratification, with a deep thermocline located at approximately 30 m (Fig. 3a). Dissolved oxygen (DO) concentrations were high throughout the epilimnion, with reductions across the metalimnion and hypolimnion (Fig. 3b). Sampling was repeated the following morning to inspect vertical distribution at that time and further test use of the hydroacoustic method to probe veliger distribution in the lake. Veligers once again dominated the plankton community deeper in the water column, with maximum abundance (65 ind L1) again present from 24 to 30 m depth (Fig. 4a). The overall trend in veliger abundance with depth was broadly similar to that found 12 h earlier (Fig. 2a), although higher numbers of zooplankton were present in the upper water column at the sampling site in the morning than found the previous night (Fig. 4a). The relative composition of the plankton community was broadly similar in the upper 12 m for the two sampling times, with about 10% Bosmina (about 300 mm in length) and 20% small (300e400 mm) adult copepods, although the morning sampling did have a larger fraction of nauplii (100e150 mm) and smaller proportion of veligers. Veligers in all samples were predominantly D-shaped, with lengths generally 80e90 mm. Volume backscatter strength measurements over these same depth intervals yielded similar trends at 201-kHz, with
18
24
30
36
Fig. 3 e Vertical profiles of a) temperature and b) dissolved oxygen at Copper Basin Reservoir on August 11, 2009.
5423
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
Determination of veliger abundance from Sv requires information about their backscattering cross-section (sbs) (eq (1)). Values for sbs for the veligers and zooplankton in Copper Basin Reservoir are not known, although the very high abundance of veligers relative to other plankton, especially near the thermocline (Figs. 2a and 4a), suggest that they would dominate the backscatter at those depths. The mean backscattering cross-section for quagga mussel veligers was thus calculated from Sv for those depth intervals where they were present at high concentrations and comprised 95e99% of the total plankton populations to minimize contributions to backscatter from other zooplankton. Rearranging eq (1) and solving for sbs, we calculated a value of 3.66 0.73 1015 m2. The empirically-derived sbs value was further used to calculate, from eq (2), a target strength of 144.4 0.8 dB for a D-shaped quagga mussel veliger at 201-kHz. In a similar way, we calculate at 430-kHz the sbs and TS to be 4.16 1.28 1014 m2 and e133.9 dB 1.3, respectively. Using this information, we can estimate veliger concentrations for the deeper parts of the water column where veligers dominated the plankton community, although contributions to backscatter from zooplankton would result in significant overpredictions of veliger abundance in the upper water column (e.g., Fig. 4a). Since the acoustical response of an organism is a complex function of its size, shape, density and soundspeed contrasts with water, and other factors, as well as the frequency of the soundwave (Stanton, 1989), differential frequency response has been used in some studies to resolve different scatterers (Mitson et al., 1996; Godlewska et al., 2009). The presence of several different groups of zooplankton make it impossible in this study to resolve the contribution to Sv for each group of zooplankton, although it is relevant to note that the difference in volume backscatter strength at 430- and 201kHz (Sv430e201) varied linearly with log zooplankton abundance (dashed line, R2 ¼ 0.94) over the 0e18 m depth interval for the 2 sampling events (Fig. 5, solid symbols). Strong
a
b
Concentration (ind L-1) 0
20
40
60
80
deviation from this linear trend was found at lower depths when zooplankton abundances were low and veligers dominated the total plankton community (Fig. 5, open symbols). The empirical regression line from Fig. 5 was thus used to correct Sv201 for the contribution from (non-veliger) zooplankton. That is, the log zooplankton concentration in each depth interval was estimated from the difference in mean volume backscatter strength (Sv430e201) and, based upon an ensemble-average backscattering cross-section calculated for zooplankton from Sv measured at 201-kHz (1.26 1014 m2), used to correct Sv201 for the contributions from zooplankton. The mean backscattering coefficient for veligers at 201-kHz (3.66 1015 m2) was then used to estimate the veliger concentrations from corrected Sv201 values over the entire depth range (Fig. 6). Fig. 6c shows predicted and observed veliger concentrations for measurements collected from Copper Basin Reservoir at approximately 1:30 p.m. on August 11, 2009 (this data set was not included in any of the prior figures or calculations). One sees generally fair agreement between predicted concentrations (Fig. 6, dashed line) and observed concentrations (Fig. 6, solid symbols) of veligers. This approach yielded an average absolute error in predicted veliger abundance of 9.9 ind L1 (mean relative error of 37.6% for veliger concentrations greater than 5 ind L1). Relative error was smaller when veliger concentrations were high and zooplankton abundances were low (e.g., mean relative error dropped to 16.8% when veliger concentrations exceeded 20 ind L1 and zooplankton abundances were less than 2 ind L1). The net sampling results shown in Fig. 6 were necessarily integrated over finite depth intervals and thus potentially mask finer-scale heterogeneities within the water column. Moreover, even with the relatively large net used (0.3 m diameter opening) the net samples a very small crosssectional area of the water column. Hydroacoustic measurements can provide far greater vertical resolution and rapidly sample a much greater area and volume of the lake. For
c
Sv201 (dB re 1 m-1) -102
-100
-98
-96
-89
Sv430 (dB re 1 m-1) -88
-87
-86
-85
0
6
Depth (m)
12
Veligers Bosmina Cyclopoid Calanoid Nauplii Rotifers
18
24
30
36
Fig. 4 e Results from measurements made at approximately 9:00 a.m. on August 11, 2009: a) zooplankton and quagga mussel veliger abundances from net sampling, b) volume backscatter strength at 201-kHz, and c) volume backscatter strength at 430-kHz.
5424
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
log Zooplankton Concentration (ind L-1)
2
1.5
1
0.5
0
Depth 0-18 m Depth 18-36 m
-0.5
-1 10
12
14
16
Sv430-201 (dB re 1 m-1) Fig. 5 e Log zooplankton abundance (excluding veligers) vs. the mean difference in volume backscatter strength at 430- and 201-kHz (Sv430e201).
example, dual frequency echograms from a short 40 m section of survey collected about 2 p.m. on August 11, 2009 were binned into 1 m 1 m cells and Sv values corrected for the contribution of fish by removal of echoes with greater than 75 dB target strength using the cross-filter detector for noise reduction in Sonar5. The difference in mean volume backscatter strength (Sv430e201) was used to estimate the abundance of zooplankton in the water column, while Sv201 values after correction for zooplankton were used with the veliger scattering cross-section at this acoustic frequency to estimate
veliger concentrations with depth in three adjacent profiles (Fig. 7). Similar to findings from net sampling made earlier in the day (Fig. 4a), the highest abundances of zooplankton were present in the upper 12 m of the water column, with generally very few zooplankton below this depth (Fig. 7). Veliger abundances were very low in the uppermost 6 m of water and present at a maximum concentration at 30 m (Fig. 7), which coincides with the depth of the thermocline (Fig. 3a). The hydroacoustic measurements reveal that this layer of veligers is quite thin, on the order of 1e2 m thick. Some lateral variability was also seen in the three adjacent profiles (each representing the average of 5 pings) (Fig. 7), although as a result of beam spreading, the ensonified volumes increasingly overlap with depth. Lateral and vertical heterogeneities in veliger abundances from a 700 m transect from Copper Basin Reservoir on August 10, 2009 (Fig. 1) can be seen in Fig. 8. Veliger concentrations generally increased with depth, with highest values frequently near the thermocline, although substantial variability was present at that depth as well as elsewhere. Consistently high concentrations were found where the thermocline intercepts the bottom sediments (about 300 m on transect, Fig. 8); this suggests that internal waves and local resuspension may play some role in high observed veliger concentrations there. Previous studies have shown that settlement rates of quagga mussel veligers were extremely low below the thermocline (Mueting et al., 2010) and water velocity is critical in determining quagga mussel veliger settlement (Chen et al., 2011).Very high concentrations were also present near the surface extending down to about 10 m depth at approximately 420 m from the start of the transect, and also above the lake bottom at approximately 670 m on the transect that may represent local spawning inputs, aggregation by mixing, settling, transport or other processes (Fig. 8). Veliger abundance was thus found to be strongly heterogeneous in both lateral and vertical directions.
Veliger Concentration (ind L-1) 0
20
40
60
80
0
20
40
60
80
0
20
40
60
80
0
a
b
c
Depth (m)
6
12
18
24
30
36 Fig. 6 e Measured (solid symbols) and predicted (dashed lines) veliger concentrations in Copper Basin Reservoir on: a) August 10, 2009, 9 p.m.; b) August 11, 2009, 9 a.m.; and c) August 11, 2009, 1:30 p.m.
5425
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
Veliger Concentration (ind L-1) 0
50
100
150
200
0
0
50
100
a
200 0
150
50
100
b
150
200
c
6 Zooplankton (lower x-axis)
Depth (m)
12
18
24 Veligers (upper x-axis)
30
36 0
10
20
30
40
50
0
10
20
30
40
50
0
10
20
30
40
50
Zooplankton Concentration (ind L-1) Fig. 7 e Vertical distribution of a) zooplankton and b) quagga mussel veligers estimated from volume backscatter strength measurements on Copper Basin Reservoir made on August 11, 2009, 2:00 p.m. Each panel represents the average of 5 consecutive pings on a transect at a survey speed of 1 m sL1.
4.
Discussion
Volume backscatter strength has been used in a number of previous studies to quantify the abundance and distribution of zooplankton in lakes (e.g., Hembre and Megard, 2003; Holbrook et al., 2006). Single-frequency measurements can provide reasonable estimates under certain conditions where a single known size/target strength organism dominates the acoustic backscatter (Holliday and Pieper, 1995), although two
or more frequency measurements are necessary when size and abundance are not known (Mitson et al., 1996; Lavery et al., 2007). Multiple frequencies can also help separate fish from other scatterers (Jurvelius et al., 2008). Empirical measurements of backscattering cross-section have been made (e.g., Hembre and Megard, 2003), as well as scattering models developed that include size, shape and density and sound speed contrasts with water (Stanton et al., 1994; Lavery et al., 2007).
Fig. 8 e Distribution of veligers determined from corrected volume backscatter strength measurements across a 700 m transect on Copper Basin Reservoir on August 10, 2009, 10 p.m.
5426
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
This study represents the first using hydroacoustic methods to infer the abundance and distribution of quagga (or zebra) mussel veligers in an invaded lake. Veligers are smaller than most other zooplankton that have been studied and as a result are relatively weak scatters of sound. The effective diameter (d) of a typical D-shaped veliger is estimated to be about 60 mm, based upon the volume calculated for a scalene ellipsoid that is 85 mm 65 mm x 40 mm in size. As a result, veligers are Rayleigh scatterers at all typical echosounder frequencies, and thus scatter sound at an intensity that is proportional to (d/l)4. The wavelengths at the 2 frequencies used in this investigation (430-kHz and 201-kHz) are 3.49 and 7.46 mm, respectively; as a result, we expect the veligers to scatter sound at an intensity that is about 20 greater at 430kHz than at 201-kHz, although the empirically-derived backscattering cross-section was about 10 greater. Nonlinear frequency response in backscatter has been seen in several other studies and is thought to result from resonance or other factors (Benoit-Bird, 2009b). Perhaps more interesting, the ensemble-average backscattering cross-section for the somewhat larger zooplankton present in the reservoir that included adult and nauplii copepods and Bosmina, was inferred to be only 3e4 greater than that of D-shaped veligers at 201-kHz (1.24 1014 vs. 3.66 1015 m2, respectively). At effective diameters near 100e200 mm for the copepods and Bosmina present, one would expect backscattering crosssections 10e100 greater. The dense CaCO3 shell of the quagga mussel veligers provide a significant density contrast with water, and is thus thought to help make veligers comparatively strong scatterers of sound for their size. Stanton et al. (1994) previously noted that the hard aragonite shell of marine gastropods was responsible for their much higher echo energy per unit biomass when compared with fluid-like zooplankton. The difference in mean volume backscatter strength was found to be strongly correlated with log zooplankton concentration in the upper (0e18 m) water column (r2 ¼ 0.94), although a multiple linear regression of log zooplankton concentration with Sv201 and Sv430 yielded a similarly strong correlation (r2 ¼ 0.96). The response in Sv across these 2 frequencies thus provided a convenient way to estimate zooplankton abundance in this reservoir and also correct for their contributions to overall acoustic backscatter. Changes in difference in mean backscatter strength were also found to track changes in composition of micronekton (Benoit-Bird, 2009b). Our findings support the previous observations of Reid et al. (2010) that quagga mussel veligers are often most abundant at the thermocline, with the hydroacoustic measurements reported here indicating their accumulation to high concentrations within a thin layer that can be 1e2 m thick (Fig. 7). At the same time, strong patchiness is present (Fig. 8), as previously found in high resolution measurements of other zooplankton communities (e.g., Hembre and Megard, 2003). The patchiness is presumed to reflect spatially and temporally heterogeneous spawning inputs, turbulent horizontal and vertical transport, and complex ecological factors. This patchiness is also thought to account for some of the deviations between measured and predicted veliger concentrations shown in Fig. 6. That is, the 0.3 m diameter Wisconsin
net necessarily sampled a different volume of water than the acoustic beam owing to the time required to complete the plankton tows, spreading of the acoustic beam, and related factors (Hembre and Megard, 2003). Additional work is nonetheless needed to better understand the acoustical properties of quagga mussel veligers and measurement of their abundance and distribution using hydroacoustic techniques. Copper Basin Reservoir provided favorable conditions for acoustical measurements owing to the very large abundance of veligers in the lower portion of the water column with few other scatterers present (often comprising 95e99% of the total plankton there). The small size and low overall abundances of zooplankton also provided favorable conditions for acoustical measurements, with a relatively simple way to correct for their contribution to overall Sv. Additional studies are planned for other reservoirs on the Colorado River Aqueduct where veliger abundances are somewhat lower, and populations of other larger zooplankton (e.g., Daphnia) are much higher. Measurements that include volume backscatter strength at a third frequency will further define the acoustical response of quagga mussel veligers. Notwithstanding, results of this study indicate that hydroacoustic measurements can offer a way to rapidly estimate veliger abundance and distribution over large areas with high horizontal and vertical resolution in invaded lakes and reservoirs. Such information can increase our understanding of their ecology, relationships to other zooplankton, effects on food webs, and potential impacts to infrastructure associated with water supply and power generation in invaded systems such as the Colorado River system in the western USA.
5.
Conclusions
1. Volume backscatter strengths measured at 430- and 201kHz were related to abundances of quagga mussel veligers and zooplankton in Copper Basin Reservoir. 2. The difference in mean volume backscatter strength at 430and 201-kHz was strongly correlated with log zooplankton abundance and allowed for correction of volume backscatter strength and determination of veliger abundance in the water column. 3. Hydroacoustic measurements indicate that high abundances of veligers were often present within a thin 1e2 m layer at the thermocline. 4. Veliger distribution was nonetheless strongly heterogeneous in both the horizontal and vertical directions, and thought to reflect spatially and temporally heterogeneous spawning and other inputs, and complex transport, lifehistory and settling processes.
Acknowledgments This research was conducted under agreement no. 95835 between the University of California-Riverside and the Metropolitan Water District of Southern California. Special thanks to Michelle Tobin for her assistance in the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 1 9 e5 4 2 7
identification and enumeration of zooplankton in the net samples. A warm thanks also for Jim Nafsey and the staff at Gene Camp for their assistance and hospitality.
references
Ackerman, J.D., Sim, B., Nichols, S.J., Claudi, R., 1994. A review of the early life history of zebra mussels (Dreissena polymorpha): comparisons with marine bivalves. Canadian Journal of Zoology 72, 1169e1179. Balk, H., Lindem, T., 2007. Sonar4 and Sonar5-Pro post processing systems operator manual version 5.9.8. Lindem Data Acquisition, Oslo, 455 pp. Benoit-Bird, K.J., 2009a. Dynamic 3-dimensional structure of thin zooplankton layers is impacted by foraging by fish. Marine Ecology Progress Series 396, 61e76. Benoit-Bird, K.J., 2009b. The effects of scattering-layer composition, animal size, and numerical density on the frequency response of volume backscatter. ICES Journal of Marine Science 66, 582e593. Chen, D., Gerstenberger, S.L., Mueting, S.A., Wong, W.H., 2011. Environmental factors affecting settlement of quagga mussel (Dreissena bugensis) veligers in Lake Mead, Nevada-Arizona, USA. Aquatic Invasions 6, 149e156. Connelly, N.A., O’Neill, C.R., Knuth, B.A., Brown, T.L., 2007. Economic impacts of zebra mussels on drinking water treatment and electric power generation facilities. Environmental Management 40, 105e112. Evans, M.A., Fahnenstiel, G., Scavia, D., 2011. Incidental oligotrophication of North American Great lakes. Environmental Science and Technology 45, 3297e3303. Godlewska, M., Colon, M., Doroszczyk, L., Dlugoszewski, B., Verges, C., Guillard, J., 2009. Hydroacoustic measurements at two frequencies: 70 and 120 kHz e consequences for fish stock estimation. Fisheries Research 96, 11e16. Greene, C.H., Wiebe, P.H., Burczynski, J., 1989. Analyzing zooplankton size distributions using high-frequency sound. Limnology and Oceanography 34, 129e139. Greenlaw, C.F., 1979. Acoustical estimation of zooplankton populations. Limnology and Oceanography 24, 226e242. Hembre, L.K., Megard, R.O., 2003. Seasonal and diel patchiness of a Daphnia population: an acoustic analysis. Limnology and Oceanography 48, 2221e2233. Holbrook, B.V., Hrabik, T.R., Branstrator, D.K., Yule, D.L., Stockwell, J.D., 2006. Hydroacoustical estimation of zooplankton biomass at two shoal complexes in the Apostle Islands region of Lake Superior. Journal of Great Lakes Research 32, 680e696. Holliday, D.V., Pieper, R.E., 1995. Bioacoustical oceanography at high frequencies. ICES Journal of Marine Science 52, 279e296. Jurvelius, J., Knudsen, F.R., Balk, H., Marjomaki, T.J., Peltonen, H., Taskinen, J., Tuomaala, A., Viljanen, M., 2008. Echo-sounding can discriminate between fish and macroinvertebrates in fresh water. Freshwater Biology 53, 912e923. Kubecka, J., Frouzova, J., Cech, M., Peterka, J., Ketelaars, H.A.M., Wagenwoort, A.J., Papacek, M., 2000. Hydroacoustic assessment of pelagic stages of freshwater insects. Aquatic Living Resources 13, 361e366. LaBounty, J.F., Roefer, P., 2007. Quagga mussels invade Lake Mead. Lake Line 27, 17e22. Lavery, A.C., Wiebe, P.H., Stanton, T.K., Lawson, G.L., Benfield, M. C., Copley, N., 2007. Determining dominant scatterers of
5427
sound in mixed zooplankton populations. Journal of the Acoustical Society of America 122, 3304e3326. MacIsaac, H.J., Lonnee, C.J., Leach, J.H., 1995. Suppression of microzooplankton by zebra mussels. Importance of mussel size. Freshwater Biology 34, 379e387. Masson, S., Angeli, N., Guillard, J., Pinel-Alloul, B., 2001. Diel vertical and horizontal distribution of crustacean zooplankton and young of the year fish in a sub-alpine lake: an approach based upon high frequency sampling. Journal of Plankton Research 23, 1041e1060. Mitson, R.B., Simard, Y., Goss, C., 1996. Use of a two-frequency algorithm to determine size and abundance of plankton in three widely spaced locations. ICES Journal of Marine Science 53, 209e215. Mueting, S.A., Gerstenberger, S.L., Wong, W.H., 2010. An evaluation of artificial substrates for monitoring quagga mussel (Dreissena bugensis) in Lake Mead, NV-AZ. Lake and Reservoir Management 26, 283e292. Pimentel, D., Zuniga, R., Morrison, D., 2005. Update on the environmental and economic costs associated with alieninvasive species in the United States. Ecological Economics 52, 273e288. Reid, N.J., Anderson, M.A., Taylor, W.D., 2010. Distribution of quagga mussel veligers, Dreissena bugensis, in the reservoirs of the Colorado River Aqueduct. Lake and Reservoir Management 26, 328e335. Rinke, K., Huber, A.M.R., Kempke, S., Eder, M., Wolf, T., Probst, W. N., Rothhaupt, K.O., 2009. Lake-wide distribution of temperature, phytoplankton, zooplankton, and fish in the pelagic zone of a large lake. Limnology and Oceanography 54, 1306e1322. Schneider, D.W., Stoeckel, J.A., Rehmann, C.R., Blodgett, K.D., Sparks, R.E., Padilla, D.K., 2003. A developmental bottleneck in dispersing larvae: implications for spatial population dynamics. Ecology Letters 6, 352e360. Simmonds, E.J., MacLennan, D.N., 2006. Fisheries Acoustics: Theory and Practice, second ed.. Blackwell, Oxford, 437 pp. Stanton, T.K., 1989. Simple approximate formulas for backscattering of sound by spherical and elongated bodies. Journal of the Acoustical Society of America 86, 1499e1510. Stanton, T.K., Weibe, P.H., Chu, D., Benfield, M.C., Scanlon, L., Martin, L., Eastwood, R.L., 1994. On acoustic estimates of zooplankton biomass. ICES Journal of Marine Science 51, 505e512. Vanderploeg, H.A., Nalepa, T.F., Jude, D.J., Mills, E.L., Holeck, K.T., Liebig, J.R., Grigorovich, I.A., Ojaveer, H., 2002. Dispersal and emerging ecological impacts of Ponto-Caspian species in the Laurentian Great lakes. Canadian Journal of Fisheries and Aquatic Sciences 59, 1209e1228. Winfield, I.J., Onoufriou, C., O’Connel, M.J., Godlewska, M., Ward, R.M., Brown, A.F., Yallop, M.L., 2007. Assessment in two shallow lakes of a hydroacoustic system for surveying aquatic macrophytes. Hydrobiology 584, 111e119. Wong, W.H., Levinton, J.S., Twining, B.S., Fisher, N., 2003. Assimilation of micro- and mesozooplankton by zebra mussels: a demonstration of the food web link between zooplankton and benthic suspension feeders. Limnology and Oceanography 48, 308e312. Wong, W.H., Tietjen, T., Gerstenberger, S., Holdren, G.C., Mueting, S., Loomis, E., Roefer, P., Moore, B., Turner, K., Hannoun, I., 2010. Potential ecological consequences of invasion of the quagga mussel (Dreissena bugensis) into Lake Mead, NevadaeArizona. Lake Reservoir Management 26, 306e315.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Seawater pretreatment for reverse osmosis: Chemistry, contaminants, and coagulation James K. Edzwald a,*, Johannes Haarhoff b a b
Department of Civil and Environmental Engineering, University of Massachusetts, Amherst, MA 01003-9293, USA Department of Civil Engineering Science, University of Johannesburg, Box 524, Auckland Park, 2006 Johannesburg, South Africa
article info
abstract
Article history:
The paper addresses the effects of salinity and temperature on the chemistry of
Received 8 May 2011
important parameters affecting coagulation pretreatment including the ion product of
Received in revised form
water, acid-base chemistry, dissolved metal speciation, and precipitation reactions for
11 July 2011
aluminum and iron coagulants. The ion product of seawater is greater than for fresh-
Accepted 1 August 2011
waters and affects chemical hydrolysis and metal-hydroxide solubility reactions. Inor-
Available online 16 August 2011
ganic carbon is the main cause of seawater alkalinity and buffer intensity but borate BðOHÞ1 4 also contributes. Buffer intensity is an important parameter in assessing coag-
Keywords:
ulation pH adjustment. Mineral particles are relatively unstable in seawater from elec-
Algae
trical double layer compression, and when present these particles are easily coagulated.
Seawater chemistry
Algal-particle stability is affected by steric effects and algal motility. Dissolved natural
Desalination
organic matter from algae and humic substances causes fouling of RO membranes and
Coagulation
pretreatment removal is essential. Aluminum coagulants are not recommended, and not
Membrane fouling
used, because they are too soluble in seawater. Ferric coagulants are preferred and used.
Particles
The equilibrium solubility of Fe with amorphous ferric hydroxide in seawater is low over
Pretreatment
a wide range of pH and temperature conditions. Ferric chloride dosing guidelines are
Reverse osmosis
presented for various raw seawater quality characteristics. The effect of pH on coagulant dose and the role of buffer intensity are addressed. A dual coagulation strategy is recommended for treating seawater with moderate to high concentrations of algae or seawater with humic matter. This involves a low and constant dose with high chargedensity cationic polymers using Fe as the main coagulant where it is varied in response to raw water quality changes. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
It is essential to have pretreatment prior to reverse osmosis (RO) desalination (Voutchkov, 2010a, 2010b). Pretreatment serves to remove contaminants that can foul RO membranes, and the integration of pretreatment with RO makes for more efficient plant performance with respect to water quality and energy usage. There are numerous pretreatment configurations. All involve coagulation, flocculation, and particle
separation. The particle separation processes can consist of (1) direct filtration with granular media, (2) sedimentation and granular media filtration, (3) sedimentation and low-pressure membrane filtration, (4) dissolved air flotation (DAF) and granular media filtration, or (5) DAF and low-pressure membrane filtration. Many desalination plants are faced with removing algae in pretreatment, and some plants face the problem of dealing with harmful algae (red and browntide algae).
* Corresponding author. 4 Hillcrest Dr., Potsdam, NY 13676, USA. Tel.: þ1 315 261 4186. E-mail addresses:
[email protected] (J.K. Edzwald),
[email protected] (J. Haarhoff). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.014
5429
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Seawater is high in total dissolved solids (TDS) and often expressed as salinity (S ) in ppt on a mass basis (or & for ppt). The salinity of the open ocean is about 35 ppt (g/kg) but varies from about 31 to 45 ppt. Higher salinity seawater is found in the Mediterranean Sea, Red Sea, and the Persian Gulf, while lower salinities are found near the mouths of rivers. High temperatures of 35 C, and even slightly greater, occur for seawater in the Middle East. This paper does not address flocculation and particle separation processes. Rather, it addresses the effects of seawater chemistry and contaminants on coagulation of seawater. The purposes of the paper follow: first, to summarize the effects of seawater salinity (ionic strength) and temperature on important chemical parameters: water ion product, boron, silica, inorganic carbon, alkalinity, buffer intensity, and the chemistry of metal ferric (Fe) and aluminum (Al) coagulants; second, to examine important seawater contaminants that affect coagulation such as turbidity, algae, and TOC; and finally, to examine the potential use of various coagulants and why coagulation with Fe coagulants is preferred. In the latter part of the paper, ferric dosages and optimum pH conditions are addressed for various seawater quality characteristics. Dissolved chemical speciation and solubility plots for Fe in seawater are presented for S of 35 ppt and two temperatures: 10 C to represent temperate regions of the world and 35 C to represent summer seawater conditions for the Middle East and other warm-water regions.
2.
Chemistry of seawater
Ionic strength (I ) is the measure used to account for the effects of the ionic content of seawater on chemical reactions. It is defined by Eq. (1) where Ci is the ion concentration in mol/ kg and zi is the charge of the ion. I¼
1 X Ci z2i 2 i
! (1)
Seawater has an average ionic strength of 0.7 mol/kg compared to freshwaters at about 5 104 to 102 (expressed as mol/L). Brackish and estuarine waters have I starting at about 102 increasing to just less than seawater. There are several water quality parameters that are important in considering seawater chemistry and
pretreatment as summarized in Table 1. Some of these are discussed next while others are discussed under Section 3 on contaminants.
2.1.
pH and the dissociation of water
The pH of seawater is mainly between 8.1 and 8.3 but varies from about 7.5 to 8.3 (Table 1). It affects dissolved chemical speciation, other chemical reactions, coagulation, other pretreatment processes, and RO desalination. Because of the high ionic strength of seawater, the dissociation of water or ion product (Kw) is affected and differs from that for freshwaters. Eq. (2) in Table 2 presents the ion product ðKsw w Þ as a function of salinity and temperature. 13.21 in For 298.15 K (25 C) and S of 35 ppt, the Ksw w is 10 14 sw for freshwater. The Kw values are 1013.83 contrast to 10 and 1012.84 at 10 C and 35 C, respectively, and greater than those for freshwaters of 1014.53 and 1013.67 at 10 and 35 C. This effect on the ion product affects, in particular, dissolved metal-hydrolysis and metal-hydroxide precipitation reactions such as occur in coagulation processes (covered in Section 4). Here, it is demonstrated with the simple illustration. For coagulation processes in water treatment, we use pH as a control-operating variable; however, it is the concentration of [OH] that affects the chemical reactions cited above. For example, precipitation of ferric hydroxide in coagulation in seawater carried out at pH 7 and 35 C, [OH] is 105.84 compared to 106.67 for freshwater (correspondingly at 10 C in seawater, [OH] is 106.83 compared to 107.53 for freshwater). In summary, [OH] is much greater for seawater than what occurs for freshwaters for the same pH conditions.
2.2.
Boron
The chemistry of boron is of interest. First, it affects the alkalinity (Alk) and buffer intensity of seawater therefore affecting pH pretreatment control. Second, boron can have detrimental health effects (developmental and reproductive) as reported by Post et al. (2010). In seawater the average total boron (BT) concentration is 4.45 103 g/kg, and it is composed of boric acid (H3BO3) and borate ðBðOHÞ 4 Þ. The speciation depends on pH and is accounted for through its dissociation (acidity) constant. The boron acidity constant ðKsw B Þ, as a function of salinity and
Table 1 e Some important seawater quality parameters. Ion pH CT, mol/kg Alk, mg/L as CaCO3 SiT as SiO2, mg/L Turbidity, NTU TOC, mg/L UV254, cm1 SUVA, m1 per mg/L Algae Counts, #/mL
Concentration/Value
References or Comments
8.1e8.3 (mainly) 7.5e8.3 (range) 2.2 103 110e135 0.12e6 0.5 to 2 (dry weather conditions) 0.1 to 100 (range) 2e5 (typically, can be greater e see text) 0.01 or less (typically, can be greater e see text) <2 >2 <103 >103e106
Eby (2004) Stumm and Morgan (1996) Eby (2004) See text See text Freeman (2009) Voutchkov (2010a; 2010b) Freeman (2009) Voutchkov (2010a) See text Little humic matter Some humic matter Low concentrations Moderate to bloom conditions
5430
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Table 2 e Key equilibrium constants as a function of salinity and temperature. Constants Water: ion product
Boron: acidity
Silica: acidity
Inorganic carbon:
:
Equations
References
13847:26 lnKsw 23:6521 ln T w ¼148:9802 T 118:67 þ 5:977 þ þ 1:0495 ln T S0:5 0:01615 ðSÞ T lnKsw B ¼
1 8966:902890:51 S0:5 77:942ðSÞ T þ1:726 S1:5 0:0993 S2 þ 148:0248þ137:194 S0:5 0:5 þ1:62247ðSÞ þlnT 24:434425:085 S 0:2474ðSÞ þ 0:053105 S0:5 T
8904:2 ln Ksw 19:334 ln T Si ¼ 117:40 T 458:79 0:5 188:74 I þ 1:5998 þ I þ 3:5913 T T 12:1652 2 I þ 0:07871 T
(2)
Millero (1995), Stumm and Morgan (1996)
Millero (1995)
(3) Millero (1995)
(4)
2329:1378 ln Ksw 1:597015 lnðTÞ 1 ¼3:17537 T 5:79495 þ 0:210502 ðSÞ0:5 þ0:0872208ðSÞ 0:00684651ðSÞ1:5 T
(8)
3403:8782 ln Ksw 0:352253 lnðTÞ 2 ¼ 8:19754 T 25:9316 þ 0:088885 ðSÞ0:5 þ0:1106658ðSÞ 0:00840155ðSÞ1:5 T
(9)
Millero (1995), Stumm and Morgan (1996)
SW SW SW Notation: KSW and KSW w , KB , KSi , K1 2 : are the water ion product constant, boron acidity constant, silica acidity constant and inorganic carbon acidity constants for seawater; S is the salinity in &, T is the absolute temperature (T ), and I is the ionic strength.
temperature, is given by Eq. (3) in Table 2. For S of 35 ppt, the pKsw B is 8.76 at 10 C and 8.47 at 35 C. The fractions of boric acid and borate as a function of pH are presented in Fig. 1. Boron removal by RO depends on pH and is favored when the anion ðBðOHÞ 4 Þ dominates. Therefore, some RO plants practice enhanced removal of boron by increasing the pH to achieve treated water 0.5 mg/L (Voutchkov, 2010a), which has been a WHO (January 2011) provisional guideline. From Fig. 1 the pH should be adjusted above pH 8.5 for warm waters (e.g., 35 C), and above pH 8.8 for lower temperature waters (e.g., 10 C).
2.3.
Silica
Dissolved Si is of interest because it can theoretically affect Alk and because many solids (SiO2 and aluminum silicates) are affected by its solubility. Furthermore, Si is incorporated and released by algae, particularly diatoms. The concentration of dissolved Si thus varies in seawater from about 0.12 to 6 mg/L (2 106 to 1.0 104 mol/L) as SiO2. Like boron, it is a weak acid and occurs in acid (H4SiO4) sw and base (H3SiO 4 ) forms. Its acidity constant ðKSi Þ, as a function of ionic strength and temperature, is given by Eq. (4) in Table 2. For average seawater I of 0.7 mol/kg, the pKsw Si is 9.64 at 10 C and 9.21 at 35 C. Fig. 2 shows the distribution of dissolved Si as a function of pH. H4SiO4 dominates in seawater unless in pretreatment you increase the pH above
9.2 for warm waters (35 C) and above 9.6 for lower temperature waters (10 C).
2.4.
Inorganic carbon
Inorganic carbon is important in several chemical and biochemical reactions in seawater including CO2 exchange with the atmosphere, solubility reactions with carbonate minerals, algal growth and respiration, and bacterial decay of organic matter. Inorganic carbon is the main contributor to alkalinity and buffer intensity, both of which are discussed below. Total inorganic carbon (CT) is defined by Eq. (5) where ½H2 CO3 is the sum of dissolved carbon dioxide: [CO2(aq)] and [H2CO3]. ½H2 CO3 is mostly [CO2(aq)] (see Stumm and Morgan, 1996). The equilibrium concentration of H2 CO3 is set by Henry’s law. For seawater in equilibrium with atmospheric CO2(g) at 380 ppm yields ½H2 CO3 concentrations of 1.67 105 M and 0.80 105 M for 10 and 35 C, respectively. Surface seawater CT reported in Eby (2004) is about 2.3 103 mol/kg or 2.36 103 M assuming a seawater density of 1025 kg/m3. This agrees closely with Stumm and Morgan (1996) who report an average CT of 2 103 M. In this paper, a CT of 2 103 M is used in calculations. 2 CT ¼ H2 CO3 þ HCO 3 þ CO3
(5)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Fraction of Species
Recognizing that [OH] is small for seawater unless you add strong base and increase the pH to above 9, we can neglect it. [Hþ] is very small and can be neglected. Other bases in seawater are small, compared to inorganic carbon and borate, and can be neglected. These are: (1) orthophos2 3 phate ðH2 PO 4 ; HPO4 ; PO4 Þ; (2) base form of dissolved Si (H3 SO4 is small as discussed above for pH <w9); (3) base forms of humic and fulvic acids, small except at the mouths of some rivers; and (4) others such as NH3, HS, and HF. Neglecting all these minor bases, we obtain Eq. (11a). This equation can be rewritten in a practical and useful form for seawater as Eq. (11b). i 2 h (11a) þ BðOHÞ1 Alk ¼ þ HCO 3 þ 2 CO3 4
pKBSW = 8.76
1.0
H3BO3 B(OH)4-
0.8 0.6 0.4
10oC, S 35 ppt
0.2 0.0
pKBSW = 8.47
1.0
Fraction of Species
H3BO3
8 <
!1 þ þ þ 2 !1 9 = H H H KSW 2 Alk ¼ 1þ SW þ C þ2 1þ SW þ SW SW : ; T ½H K1 K2 K1 K2 ! KSW B BT þ þ KSW B þ H
B(OH)4-
0.8 0.6 0.4
o
35 C, S 35 ppt
0.2 0.0 4
6
8
10
pH Fig. 1 e Fraction of H3BO3 and BðOHÞ1L as a function of pH 4 for seawater S& of 35: top figure at 10 C, bottom figure at 35 C.
Inorganic carbon participates in two acid-base reactions shown by Eqs. (6) and (7). þ H2 CO3 #HCO 3 þH
(6)
2 þ HCO 3 #CO3 þ H
(7)
sw The acidity constants ðKsw 1 and K2 Þ as a function of salinity and temperature are presented in Eqs. (8) and (9) in Table 2. The first acidity constant ðKsw 1 Þ for seawater salinity of 35 ppt, is 105.99 at 10 C and 105.76 at 35 C. These differ greatly from freshwater (I approaching 0 M), which are 106.46 and 106.34. The second acidity constant ðKsw 2 Þ for seawater salinity of 35 ppt, is 109.18 at 10 C and 108.75 at 35 C compared to freshwater constants of 1010.49 and 1010.25.
2.5.
5431
From earlier, CT is about 2 103 M. BT averages 4.45 103 g/kg or 4.12 104 mol/kg. For consistent units, the BT molar concentration assuming a seawater density of 2 1025 kg/m3 is 4.2 104 M. Some HCO 3 and CO3 are complexed by metals in seawater and would not contribute to alkalinity. However the fraction of HCO 3 complexed is small and since it is the main contributor of alkalinity for inorganic carbon and other bases were neglected, complexation is ignored here. Using the above CT and BT values and appropriate equilibrium constants found from Eqs. (3), (8) and (9) in Table 2, we can use Eq. (11b) to calculate alkalinities for seawater at pH of 8.1e8.3. They are: (1) 2.2e2.3 meq/L (110e115 mg/L as CaCO3) at 10 C where inorganic carbon accounts for 95 to 97 percent and boron the remainder; and (2) 2.5e2.7 meq/L (125e135 mg/L as CaCO3) at 35 C where inorganic carbon accounts for 93 to 95 percent and boron the remainder. Equation (11b) is the theoretical basis for Alk, but for seawater water quality and pretreatment monitoring it is not normally used. To do so requires measurements of CT, BT, pH, and temperature. Rather Alk is measured directly by titrating a sample with strong acid to reach a pH end-point of about 4.5. Hence, it is a capacity measurement and rightly called the acid neutralizing capacity. Alkalinity is used as a water quality parameter as an indicator of a water’s resistance to pH decrease from acid addition, but it is a capacity measurement. To properly assess the resistance for a specific change in pH from acid or base addition, one should consider the buffer intensity.
2.6.
Alkalinity
Alkalinity (Alk) is the acid neutralizing capacity of water and is due to the presence of bases in seawater that can accept a proton from acid addition. Alk for seawater is defined by Eq. (10). i 2 h Alk ¼ þ HCO þ BðOHÞ þotherbases Hþ 3 þ2 CO3 4 þ OH (10)
(11b)
Buffer intensity
Buffer intensity (b) is more useful than alkalinity in evaluating changes in seawater pH over specified pH changes of interest. b is the resistance to a unit change in pH for a unit addition of strong acid (Ca) or strong base (Cb), and it is defined by Eq. (12) (Stumm and Morgan, 1996). It is the inverse of the slope of alkalinity titration curves, thus a negative sign for strong acid addition and a decrease in Alk and positive sign if base is added increasing Alk.
5432
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Fraction of Species
1.0
a
pKSiSW = 9.64
H4SiO4
0.8
H3SiO4
1-
0.6 o
10 C, S 35 ppt 0.4
0.2
0.0
pKSiSW = 9.21
Fraction of Specis
1.0
H4SiO4
H3SiO41-
0.8
0.6
b
35oC, S 35 ppt
0.4
0.2
0.0 2
4
6
8
10
12
pH 1L
Fig. 2 e Fraction of H4SiO4 and H3 SiO4 as a function of pH for seawater S& of 35: top figure at 10 C, bottom figure at 35 C.
b¼
DCa DAlk DCb DAlk ¼ ¼ ¼ DpH DpH DpH DpH
(12)
Applying this to Eq. (11b), we can obtain Eq. (13) for b as a function of pH for seawater (Eby, 2004) for a closed system, meaning the dissolved carbon dioxide concentration is not forced to be in equilibrium with carbon dioxide in the atmospheric gas phase. This is applicable because dissolved carbon dioxide in seawater is usually out of equilibrium with the atmosphere, and because pretreatment-detention times are short so with pH changes equilibrium with atmospheric carbon dioxide is not established. The causes of buffer intensity are identified below the terms in Eq. (13). Buffer intensity is due to borate, inorganic carbon, OH at high pH, and Hþ at low pH.
Fig. 3 e (a) Buffer intensity of seawater as a function of pH for temperatures at 10 and 35 C. (b) Individual buffer intensity terms at 35 C (Conditions: CT at 2.0 3 10L3 M, BT at 4.2 3 10L4 M).
percent at 10 C. Fig. 3(a) shows the total buffer intensity, which is instructive. If we wish to decrease the pH of seawater in coagulation, then small amounts of acid either from Fe dosing or from a strong acid (HCl or H2SO4) are required to reach target pH conditions near 7 because of the relatively low b. To decrease the pH to 5.5 to 6.5, the buffer intensity increases significantly requiring larger amounts of Fe or strong acid. The effect of buffer intensity on coagulation is discussed further in Section 4.
3. (13)
Fig. 3 presents buffer intensity as a function of pH for seawater at 10 and 35 C. Fig. 3(b) shows that the primary contributor to buffer intensity is inorganic carbon. Boron contributes to buffer intensity for pH >7, and for seawater pH at about 8 boron contributes about 18 percent at 35 C and 12
Contaminants
The presentation below focuses on the commonly occurring contaminants that must be removed in pretreatment to prevent fouling of RO membranes and to enhance the operational efficiency of RO. These contaminants are (1) mineral particles, (2) algae, and (3) TOC and natural organic matter (NOM). The occurrence and concentrations of these contaminants depend on the location of the desalination plants. Most plants are in coastal areas and the waters in these areas are subject to variation in water quality. Other contaminants that may occur are oil and grease (O&G) and hydrocarbons, and
5433
The presence of particles is normally measured as turbidity. Seawater turbidity is usually fairly low as indicated in Table 1, but can be greater at some coastal locations and when algal blooms occur. Particles can cause colloidal/particle fouling of RO membranes in which a cake of particles form reducing the permeate flux. Algae can also cause biofouling e see Section 3.2.2. One pretreatment criterion for the water applied to RO membranes is the turbidity of the water following pretreatment. A turbidity goal of 0.1 NTU is desired while some membrane companies accept a higher turbidity such as 0.5 NTU (Freeman, 2009; Voutchkov, 2010a, 2010b). Another widely used pretreatment criterion of the acceptability of water applied to RO membranes is the SDI (Silt Density Index). It is a measure of colloidal/particle fouling. The SDI is a laboratory filtration test using a 0.45 mm membrane filter at a pressure of 207 kPa (30 psi) using three time intervals to collect filtered water (ASTM, 2007). Samples that contain greater colloid concentrations foul the membrane filter and take longer times to filter; hence higher SDI values. Guidelines for RO feedwater are: SDI >5 is not acceptable, SDI <2 is good quality, and SDI maximum of 3e4 (Freeman, 2009; Duranceau and Taylor, 2010; Voutchkov, 2010a, 2010b).
3.1.2.
Mineral particles
These particles can be (1) clays (aluminum silicates) carried by rivers and direct runoff to coastal waters; (2) other aluminumsilicate minerals carried by rivers and runoff or produced in the ocean by precipitation processes; (3) SiO2(s) (sand) particles of small sizes, present in the ocean and coastal areas, and (4) CaCO3(s) solids produced in the ocean by precipitation. Mineral particles are stable (tendency not to flocculate) in freshwaters. The primary cause of particle stability is the negative charge on particle surfaces. Other possible causes are hydration and steric effects, but surface charge is the primary one. As particles approach each other, a repulsive force occurs from electrical double-layer interaction from the particlesurface charge and an attractive force occurs from van der Waals forces at close particleeparticle separation distances. In freshwaters, the electrical-double layer thickness and repulsive interaction extends far enough from the particle surfaces that there is a net repulsive force causing stability. In seawater, the particle zeta potential is reduced as illustrated by clay particle electrophoretic particle mobility (EPM) data for freshwater versus seawater shown in Fig. 4. In seawater the electrical double layer is compressed toward the particle surface allowing for clay particles to approach close enough to each other that the attractive van der Waals forces can dominate and clays flocculate. This occurs for mineral particles in seawater, and in estuaries where mineral particles carried by rivers flocculate and undergo deposition at the mouths of rivers (Edzwald et al., 1974). Based on surface charge, mineral particles carried by rivers and runoff into the sea should flocculate and settle and therefore be at relatively low concentration at the intakes to desalination plants, but some mineral particles can be present
-1 -2 -3
Seawater
Seawater
Fresh Water
0
Fresh Water
Turbidity and silt density index
Seawater
3.1.1.
1
Auerococcus anophagefferens
Particles
EPM x 108 (m/s per V/m)
3.1.
2
Seawater
these can be removed via coagulation and separation and are discussed briefly.
Karenia brevis
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Kaolinite Montmorillonite
-4
Fig. 4 e Electrophoretic mobility (EPM) for two algae in seawater (Karenia brevis and Auerococcus anophagefferens) and two clays (montmorillonite and kaolinite) in freshwater and seawater (data from Sengco (2001).
in which particle stability is attributed to adsorbed NOM and hydration effects at particle surfaces. Other mineral particles can be present because they are produced in seawater by precipitation of CaCO3 or aluminum silicates. In all cases, however, the stability due to charge should be low because of the high ionic strength of seawater and these particles are easily treated by coagulation.
3.1.3.
Algae
Algae are often a major problem for desalination plants and must be removed by pretreatment. For all RO installations, algae are present at low to moderate concentrations (Table 1), but many plants are located in areas subject to algal blooms where concentrations may greatly exceed 103 cells per mL, especially those plants located in warm, shallow seas. Almost all groups of phytoplankton-type algae (floating or suspended) are found in seawater. Unlike mineral particles, algae remain suspended while undergoing growth and do not flocculate until they are in the later stages of their growth phase. With the high ionic strength of seawater, the algal-particle stability is not due to surface charge, although algae do carry a negative charge with low negative zeta potentials or EPM as shown for two algae in Fig. 4. Algal-particle stability can be due to (1) steric effects, (2) algal motility, and (3) hydration effects. Steric effects are attributed to their surface structure that prevents aggregation. Motility or swimming refers to the fact that many algae have flagella (such as dinoflagellates and others) or cilia (hair-like structure around the exterior of the cells) that provide a means of swimming and steric repulsive interaction. Hydration effects mean the presence of cell-surface groups with bound water inhibiting flocculation. Two extensive reviews of algae in freshwaters by Henerdson et al. (2008) and Ghernaout et al. (2010) cover algal properties, dissolved organic matter excreted by algae, and coagulation of algae. In many coastal areas and in warm-shallow seas, one can get harmful algal blooms often referred to as red tides and brown tides depending on the pigments associated with the algae. Some of these harmful algae can also bloom in colder waters. Without proper pretreatment, the algae and associated toxins can shut down desalination plants. Many dinoflagellates are associated with red tide; for example, Karenia
5434
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
brevis and Noctiluca scintillans (Fig. 5). Also shown in Fig. 5 is brown-tide algae, Auerococcus anophagefferens (a chrysophyte). Some algae are large (e.g., dinoflagellates of 10s to 100s of mm), while other algae are small, roughly 1 to several mm (e.g., diatoms, green algae, cyanobacteria, and chrysophytes). Removal of algae by pretreatment is essential in preventing algae from accumulating on RO membranes causing biofouling (see Section 3.2.2). Algae can be coagulated effectively with good separation by DAF clarification (Edzwald, 2010) and granular media filtration prior to RO membranes.
3.2.
Natural organic matter
The concentrations and types of NOM present in seawater are highly dependent on the water plant intake location. Higher levels are found at the mouths of river carrying NOM and in waters subject to algal blooms. This discussion begins with collective and surrogate parameters for NOM and then is followed by consideration of types of NOM.
3.2.1.
Collective and surrogate parameters
TOC is a collective parameter, and it is the most widely used measurement to characterize the concentration of organic matter. TOC is the sum of particulate organic carbon (POC) and dissolved organic carbon (DOC). In the absence of algal blooms, TOC is approximately the DOC. The TOC of seawater used at desalination plants is typically 2e5 mg/L (see Table 1), but it can be greater when algal blooms occur or for seawater affected by aquatic humic matter from rivers e for example, the Tampa Bay (Florida, United States) facility has raw water TOC averaging 6.2 mg/L with a range of 4.1e12 mg/L (Schneider, 2011). A useful surrogate parameter for DOC (or TOC when w DOC) is the UV absorbance at 254 nm (UV254) - see Edzwald et al. (1985), Edzwald and Tobiason (2010). Quite often UV254 is low for seawater at 0.01 cm1 or less (see Table 1); consequently, in measuring UV254 for raw seawater and especially across the treatment plant, spectrophotometer cells with a path-length greater than the typical 1 cm should be used to increase accuracy. Where seawater is affected by
Fig. 5 e Photos of examples of algae that occur in seawater (a) Karenia brevis, a dinoflagellate, size 25 mm, associated with red tide (Source: Florida Fish and Wildlife Conservation Commission, http://research.myfwc.com/images/articles/23559/23559_ 5513.jpg&imgrefurl); (b) Auerococcus anophagefferens a chrysophyte, size 2e3 mm, associated with brown tide, (Source: U.S. Department of Energy Joint Genome Institute, http://www.jgi.doe.gov/); (c) Noctiluca scintillans, a dinoflagellate, size 200e2000 mm, associated with red tide (Source:Australian Government, Department of the Environment, Water, Heritage, and the Arts, http://www.tafi.org.au/zooplankton/imagekey/dinophyta/) (all accessed 12 January 2011).
5435
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
the presence of aquatic humic matter and by algal blooms, raw seawater will have higher UV254 than 0.01 cm1.
3.2.2.
Types of organic matter and biofouling
A variety of types of organic matter can be present in seawater. For seawater that contains aquatic humic matter, the NOM is a mixture of aquatic humic and fulvic acids (especially, fulvic acids). Aquatic humic matter has properties of fairly high molecular weight (100se1000s) and while it carries a negative charge it has hydrophobic character, is reactive and can be removed in pretreatment by coagulation and separation of the precipitated solids. Other NOM can come from the decay of plant and algal matter. In addition algae, while undergoing growth and respiration, impart soluble organic matter to the water called extracellular organic matter (EOM). The algae can also impart intracellular organic matter (IOM). EOM and IOM are referred to in some literature as algogenic organic matter (AOM). The compounds composing AOM consist of acids, proteins, simple sugars, anionic polymers, negatively charged and neutral polysaccharides. The negatively charged compounds and some neutral compounds exert a coagulant demand as discussed further in Section 4.2.2. Another concern is that, without good pretreatment, algae such as those in Fig. 5 that are transported to RO membranes undergo cell lysis, due to the transmembrane pressure, releasing IOM. The compounds released are soluble, biodegradable, and a major concern regarding biofouling of RO membranes. Measurements for the presence of these compounds are not common, but fluorescence type measurements may have promise (Para et al., 2010; Baghoth et al., 2011). Ultraviolet light at 254 nm is absorbed by a variety of organic molecules, but especially organic compounds with an aromatic structure. Aquatic humic matter has this structural feature so it absorbs more light per unit concentration of DOC (called the specific UV absorbance or SUVA) than other types of NOM such as hydrophilic acids, hydrophobic bases (e.g., proteins and aromatic amines), and hydrophobic neutrals (e.g., aldehydes) (Edzwald and Van Benschoten, 1990; Edzwald, 1993; Bose and Reckhow, 1998; Edzwald and Tobiason, 2010). SUVA is a simple method for characterizing the nature of NOM. Applying the concept to seawater, the following guidelines are provided: (1) SUVA of about 4 or greater indicates the NOM is dominated by aquatic humic matter, (2) SUVA of 2e4 indicates the NOM is composed of a mixture of aquatic humic matter, EOM and IOM from algae, and (3) SUVA <2 indicates the NOM is composed largely of algal derived EOM and IOM. Coagulation of organic matter is an important pretreatment step to prevent fouling of RO membranes. Voutchkov (2010a; 2010b) reports that if the TOC is reduced to 0.5 mg/L or less then biofouling is unlikely, while seawater with TOC above 2 mg/L is most likely to cause biofouling. TOC of 0.5 mg/ L is difficult to achieve without treatment by granular activated carbon or nanofiltration. In assessing good coagulation pretreatment, for seawater containing several mg/L of TOC, a guideline of 1e2 mg/L is recommended.
membranes. The sources of O&G can be from wastewater discharge or from storm drains in the vicinity of desalination plant intakes. Voutchkov (2010a; 2010b) reports that O&G > 0.02 mg/L can cause membrane fouling. Removal of O&G can be achieved by coagulation followed by separation through sedimentation or more effectively by DAF. Hydrocarbons from ship spills or seawater oil wells must also be removed through pretreatment. The hydrocarbons are incorporated into the O&G measurement and can be removed as indicated above. Depending on the magnitude of the oil spill, removal of the oil by ferric chloride coagulation followed by DAF may be feasible.
4.
Coagulation
Coagulation is a chemical pretreatment step. It concerns treating seawater so that particles in the supply and particles produced within pretreatment through precipitation processes (e.g., metal hydroxides from coagulation) are destabilized (flocculate readily) for downstream flocculation and particle removal processes. Metal coagulants (Al and Fe) are used commonly to treat freshwaters under dose and pH conditions in which they precipitate as aluminum or ferric hydroxides. This precipitation process produces new particles that can incorporate the raw water particles into flocs, called sweep-floc coagulation in water treatment. Coagulation can also remove NOM by a phase change converting the dissolved NOM into particles directly by precipitation or by adsorption onto particles produced by the coagulant. Coagulation is an essential pretreatment process because it affects all downstream pretreatment processes. It affects flocculation (particles do not aggregate well without coagulation), DAF (bubble attachment to flocs and removal of floc-bubble aggregates), sedimentation, and granular media filtration. Ferric salts, particularly ferric chloride, are the most common coagulants used in the pretreatment of seawater, but there have been numerous studies of the potential use of other coagulants. In this section, the chemistry of several coagulants (Table 3) is examined, and it is explained why ferric coagulants are preferred for seawater pretreatment.
Table 3 e Potential coagulants for treating seawater. Coagulants Alum and PACls Concerns about precipitative scaling Ferric Salts Widely used, especially ferric chloride Organic Cationic Polymerelctrolytes Low Molecular Weight, High Charge-Density Cationic Some RO membrane manufacturers allow; case made in text for their use in dual-coagulant strategy with ferric salts Flocculant, Flotation, and Filter Aids
3.3.
Other contaminants
Removal of oil and grease (O&G) and oil-based hydrocarbons in pretreatment is essential to prevent fouling of RO
High Molecular Weight Nonionic and Anionic Polymers Not primary coagulants: possible use as flocculant, flotation, and filter aids Concerns about fouling RO membranes
5436
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
4.1.
Aluminum coagulants
While alum and polyaluminum chlorides (PACls) have been studied in laboratory and pilot-scale work for RO pretreatment of seawater, they are not used in full-scale plants. The primary reason is because of the relatively high solubility of Al. The residual soluble Al would carry over to the RO membrane leading to concentration and precipitative scaling. The problem is demonstrated by showing solubility curves for amorphous Al(OH)3(am), which would be formed in coagulation. Four soluble Al species are considered in equilibrium with Al(OH)3(am) for seawater ionic strength: Al3þ, Al(OH)2þ, AlðOHÞþ 2 , and AlðOHÞ4 . The solubility product constant was taken from Van Benschoten and Edzwald (1990) and the hydrolysis constants were taken from Nordstrom and May (1996) for infinitely dilute solutions and adjusted for temperature. They were then adjusted to account for the ionic strength of seawater as explained in Section 4.2 for ferric coagulants. For the solubility constants of Al(OH)3(am), pKSW s1 of 12.27 and 10.83 were used for 10 and 35 C, respectively, as defined by the stoichiometry in Eq. (14). AlðOHÞ3 ðamÞ þ 3Hþ #Al
3þ
þ 3H2 O
(14)
Fig. 6 shows the solubility curves. Al is more soluble in seawater compared to freshwater because of the ionic strength of seawater. Comparing the two temperatures, Al is more soluble at 35 C above about pH 6 but less soluble below pH 6. When using Al coagulants, optimum pH conditions occur near the pH of minimum solubility at Al doses that minimize turbidity and NOM. This also produces the maximum amount of precipitated solids for sweep-floc coagulation and minimizes residual soluble Al. The pH of minimum solubility at 10 C is pH 6.8 yielding a soluble Al of about 106 mol/L (27 mg/L). At 35 C, the pH of minimum solubility is lower at pH 6 and the soluble Al is much greater at about 106 mol/L (270 mg/L). To practice alum coagulation for seawater near the pH of minimum solubility, high alum dosages are required to overcome the buffer intensity (see Fig. 3(a)) or the addition of strong acid with alum would be
0 -1
Al(OH) (am)
-3 -4
2700
-5
270
Al ( g/L)
precipitation for dosing above curves
27
-6 -7 -8
Minimum Solubility
Minimum Solubility
Log Al (M)
-2
o
10 C 35oC
-9 4
5
6
7
8
9
10
pH 3
Fig. 6 e Solubility of Al(OH) (am) in seawater as a function of pH at 10 and 35 C (pH of minimum solubility shown by arrows).
required. The pH of raw seawater is generally near 8, if one were to practice alum coagulation at pH 7.5 to 8, the soluble Al would be much greater than at the pH of minimum solubility. PACls act as coagulants by charge neutralization and by precipitation as an aluminum hydroxide solid. These modes of coagulation are highly dependent on dose and pH. If used at pH >7, precipitation of an aluminum hydroxide solid occurs leaving a high soluble Al concentration at the tenths to 1 mg/L or greater depending on water temperature (Pernitsky and Edzwald, 2003). If used at pH <7, soluble Al can be greater. In short, Al is too soluble when using Al coagulants and would be carried to the RO membranes where it can concentrate producing aluminum hydroxide and aluminum silicate solids causing precipitative scaling. The best choice of a metal coagulant is ferric salts.
4.2.
Ferric coagulants
Ferric coagulants are the best choice for seawater coagulation (Table 3). Ferric chloride is commonly used, but there has been some consideration of the use of other ferric salts such as ferric sulfate and FeClSO4 as well the production of Fe by electrocoagulation employing Fe electrodes (Timmes et al., 2009). Ferric chloride is very insoluble leaving little residual dissolved Fe in the water after pretreatment (shown below) and thus avoiding precipitative scaling problems. We discuss next the dissolved Fe speciation, the solubility of ferric hydroxide, and effects of seawater buffer intensity on reaching target pH coagulation conditions. The presentation below considers coagulation with ferric chloride, but the principles also apply to ferric sulfate.
4.2.1.
Dissolved Fe speciation and ferric hydroxide solubility
Thermodynamic data (Drever, 1997; Stumm and Morgan, 1996) were used to calculate equilibrium constants at 10 and 35 C for infinitely dilute solution. The equilibrium constants were then adjusted for seawater ionic strength conditions using activity coefficients for the ions: monovalent 0.7, divalent 0.25, and trivalent 0.05 (Stumm and Morgan, 1996). The calculated seawater equilibrium constants are presented in Table 4. Figs. 7(a) and 8(a) present the fractions of dissolved Fe species as a function of pH for 10 and 35 C, and Figs. 7(b) and 8(b) present solubility curves for amorphous ferric hydroxide. The dissolved Fe species is discussed first with emphasis on pH 7.5 to 8 representing raw seawater or coagulation with little pH adjustment, and then for pH 6 to 7 representing coagulation with pH adjustment. For pH 7.5 to 8 and 10 C as shown in Fig. 7(a), there would be fairly high fractions (80e97 percent) of positively charged Fe as FeðOHÞþ 2 available for charge neutralization coagulation reactions; however, as shown in Fig. 8(a) there would be much less (10e55 percent) for the same pH range for seawater at 35 C. For pH 6 to 7 and 10 C (Fig. 7(a)), the FeðOHÞþ 2 fraction is maximized near 99 percent. At 35 C (Fig. 8(a)), the FeðOHÞþ 2 fraction is still maximized near 99 percent at pH of about 6, but decreases to about 90 percent as the pH increases to 7. In general, then the fraction of positively charge Fe varies with pH and temperature with less of it with increasing pH.
5437
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Table 4 e Ferric chemistry and equilibrium constants for seawater. 10 C 35 C
Equations FeðOHÞ3 ðamÞ þ 3Hþ #Fe3þ þ 3H2 O
4.95 3.63
(15) pKSW S1
4.2.2.
pbSW 11
3.15
2.49
þ Fe3þ þ 2H2 O #FeðOHÞþ 2 þ 2H
(17) pbSW 12
6.84
6.32
þ Fe3þ þ 4H2 O #FeðOHÞ 4 þ 4H
23.38 21.39 (18) pbSW 14
þ H2 O#FeðOHÞ þH
þ
(16)
The solubility curves for 10 and 35 C are presented in Fig. 7(b) and Fig. 8(b), If we coagulate with the addition of Fe at dosages of 1 mg/L (1000 mg/L) or greater (not accounting for coagulant demand by complexation of NOM, to be discussed in Section 4.2.2), then initially the system is oversaturated with Fe and precipitation of amorphous ferric hydroxide occurs, which is called sweep-floc coagulation. The soluble Fe concentration is reduced and at equilibrium drops to the concentrations shown on the curves. Fe is very insoluble over a wide range of pH conditions making it a better coagulant than Al salts because the residual soluble concentrations would be very low preventing precipitative scaling of RO membranes. For pH of 6e8 at 10 and 35 C (Fig. 7(b) and Fig. 8(b)), the residual soluble Fe is much less than 1 mg/L.
b
1.0
Fraction of Dissolved Species
Fe(OH)4-
Fe3+
0.8
0.6
0.4 o
Fe(OH)2+
10 C
0.2
0.0
b
-2
Fe(OH)3(am)
-4
Log Fe (M)
a 1.0
Fe(OH)2+
5600
precipitation for dosing above curve -6
56
-8
0.56
-10
10oC
Fe ( g/L)
Fraction of Dissolved Species
a
Ferric coagulation dosing
Ferric chloride doses range typically from 1 to 10 mg/L as Fe depending on (1) raw water quality characteristics, (2) pH of coagulation, and (3) whether a strong acid (e.g. HCl or H2SO4) is used to adjust the pH of coagulation. For raw seawater of good quality, say low to moderate turbidity (<10 NTU), low algae concentrations (<103 cells/mL), low TOC (3 mg/L), low UV254 (<0.03 cm1) and low SUVA (<1.5 m1 per mg/L), then ferric chloride doses should be low to moderate (several mg/L or less depending on how pH is adjusted). Because TOC is low and there is little organic matter of aromatic character (low SUVA) with negative charge, then there would little coagulant demand for positively charged Fe. Thus, coagulation can be carried out at relatively high pH values with some temperature dependence as discussed above with Figs. 7 and 8. For warm waters (say 20e35 C) or even greater, coagulation pH of 7e7.5 is recommended. This may be achieved with low Fe doses (few mg/L or less) especially if strong acid is used to achieve the target pH (see the b curve for 35 C in Fig. 3(a)). For lower temperature seawater (say <20 C), then coagulation pH conditions of 7.5e8 should be effective. Fe doses should be low (few mg/L or less) and addition of strong acid should not be necessary assuming raw seawater pH is about 8 e see b curve for 10 C Fig. 3(a) the fairly low buffer intensity at pH 7.5 to 8.
0.0056
-12
Fe(OH)2+ 0.8
Fe3+
Fe(OH)4-
0.6
0.4
35oC
Fe(OH)2+ 0.2
0.0
-2
Fe(OH)3(am)
-4
Log Fe (M)
Fe
2þ
5600
precipitation for dosing above curve -6
56
-8
0.56
-10
35oC
Fe ( g/L)
3þ
These are equilibrium concentrations and assume no complexation reactions with anions that can increase soluble Fe concentrations, but these effects should be small and in practice the soluble Fe is expected to be higher but still low.
0.0056
-12
-14 2
4
6
8
10
12
pH Fig. 7 e (a) Fraction of dissolved Fe species as a function of pH for seawater at 10 C, (b) solubility diagram for ferric hydroxide for seawater at 10 C.
-14 2
4
6
8
10
12
pH
Fig. 8 e (a) Fraction of dissolved Fe species as a function of pH for seawater at 35 C, (b) solubility diagram for ferric hydroxide for seawater at 35 C.
5438
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Raw seawater with higher turbidity, if caused by minerals, should not have much effect on ferric chloride doses and coagulation pH as long as algae concentrations, TOC, UV254, and SUVA are low as presented above. The effect of algae concentration on coagulation is important. With moderate algae concentrations and algal blooms (>103 to 106 cells/mL), turbidity and TOC will increase affecting ferric chloride dosing. If algal blooms occur increasing the background TOC by 2 mg/L or more, then the algae and especially the dissolved organic matter associated with the algae create a coagulant demand that must be satisfied by adding sufficient Fe in coagulation to first complex with the NOM compounds and then to precipitate ferric hydroxide under sweep-floc conditions at the pH conditions specified below. The concept is to maximize the fraction of dissolved cationic Fe species that would be available when the ferric coagulant is added to react with negatively charged dissolved organic matter and solids. For warm waters (say 20e35 C) or even greater, coagulation pH of 6e6.5 is recommended to maximize FeðOHÞþ 2 (Fig. 8(a)). The buffer intensity increases below pH 7 (see the b curve for 35 C in Fig. 3(a)) so ferric chloride doses could be high (several mg/L to say 10) to achieve pH 6 to 6.5 without the use of strong acid. For lower temperature seawater (say <20 C), then coagulation pH conditions of 6.5e7 should be effective to maximize the availability of FeðOHÞþ 2 (Fig. 7(a)). Because the buffer intensity for lower temperature seawater increases below pH 7.5, like for the warm water case ferric chloride doses could be high (several mg/L to say 10) to achieve pH 6.5 to 7 without the use of strong acid. Coagulation followed by downstream flocculation and particle separation thus removes the particles originally in the raw water, the particles produced by precipitation, and some NOM that was complexed by positively charged Fe and adsorbed to flocs. Another raw water quality characteristic that affects coagulant dose and optimum coagulation pH conditions is the presence of humic substances. SUVA is a good and practical indicator of whether the DOC of seawater contains humic substances or not, and it is discussed in Section 3.2.2. Seawater with aquatic humic matter (SUVA >2) requires ferric coagulation at certain pH conditions because of the fact there is little positively charged Fe available to react with negatively charged humic matter that creates a coagulant demand unless the pH is 6e6.5 for warm waters (Fig. 8(a)) or about pH 6.5 or slightly greater for cooler waters (Fig. 7(a)). Dosages of 5e10 mg/L may be needed depending on the raw seawater TOC and whether strong acid is used to adjust pH. In a laboratory jar-test study at 24e25 C, Duan et al. (2002) found that ferric chloride at 5e10 mg/L as Fe with pre-adjustment to pH 6.1 to 6.2 (final pH of 5.7e6.1 after addition of ferric chloride) produced good removals (85e90 percent UV254) of humic acid from seawater. For seawater in which NOM from algae or humic substances is exerting a coagulant demand, then a dual-coagulant strategy should be considered in which both a low molecular weight high charge-density cationic polymer (Table 3) and ferric coagulant are used. The cationic polymer provides positive charge to satisfy partially the negative charge associated with particles and more importantly the dissolved NOM. How to do this properly and avoid the possibility of overdosing the cationic polymer is addressed next.
4.3.
Organic polymers
Polymers are classified into two categories as summarized in Table 3: (1) low molecular weight, high charge-density cationic polymers, and (2) high molecular weight polymers. The former are primary coagulants that accomplish coagulation of negatively charged particles by adsorption on particle surfaces through charge neutralization. The cationic polymers can also react with a small fraction of dissolved negatively charged humic substances and lead to charge neutralization and precipitation of a solid phase, cationic polymer-humate precipitate. However, the degree of reaction and removal of dissolved organic matter is low compared to metal coagulants. Some RO suppliers allow the use of cationic polymers, while others do not because they are concerned about overdosing of the cationic polymers leading to carry-over to RO membranes where they would adsorb and foul the membrane. However, as is discussed below the cationic polymers would not be used as the sole coagulant but used at low dosages in a dual-coagulant strategy with ferric coagulant. High molecular weight nonionic and anionic polymers are not primary coagulants. They have several other potential uses. One is a flocculant aid in which they are added after coagulation for the purposes of increasing floc sizes by bridging flocs together and of strengthening flocs. This application applies to plants with sedimentation where large flocs are desired. Another use is as a flotation aid for plants with DAF. Their function is to strengthen flocs and to retain floated sludge. A third use is as a filter aid for granular media filtration. Here they aid attachment of flocs to filter grains. Voutchkov (2010a) reports that these anionic and nonionic polymers are an option in RO pretreatment; however, some RO manufacturers do not favor their use. If overdosing occurs and granular media filtration is included in pretreatment, then the filters would perform poorly often in the form of plugging near the surface producing high head loss. Overdosing can occur where poor filter performance is not always so obvious and control of the polymer dose can be problematic. There has been some use of these type polymers as flocaids at dosages at tenths of a mg/L. If these polymers reach RO membranes, then they adsorb strongly on the membranes producing scaling and fouling. Their use should be avoided, but if used then use with caution. High charge-density cationic polymers are frequently used in a dual-coagulant strategy for treating freshwaters. Their positive charge can neutralize the negative charge of particles and can satisfy some of the negative charge associated with aquatic humic matter and algae thus reducing metal coagulant dosages. Their use in seawater coagulation to complement ferric coagulation can be advantageous, especially because seawater pH is fairly high limiting the fraction of positively charged Fe species available for charge neutralization e Figs. 7(a) and 8(a). Some RO membrane manufacturers allow their use as stated earlier, while others may not because of concern of overdosing and fouling negatively charged RO membranes. This concern is without sound basis and cationic polymers should be considered in a dual-coagulant strategy with ferric chloride. Overdosing of cationic polymers in a dual-coagulant strategy is almost impossible. In a dual-coagulant strategy the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
cationic polymer dosage is held constant at a low concentration (typically a mg/L or less on a liquid product basis) where overdosing cannot occur because ferric chloride serves as the primary coagulant. With changes in raw water quality, the ferric chloride dosage is adjusted not the cationic polymer. One can use a streaming-current monitor to insure no overdosing of the cationic polymer. Bench-scale and pilot-scale studies can easily evaluate the low and constant cationic polymer dose that is beneficial in conjunction with Fe dosing. This type of dual-coagulant strategy has been reported in use for the Al Jubail desalination plant in Saudi Arabia (Baig and Al Kutbi, 1998). Acid is added to control pH at about 6.5, ferric chloride dose is low (about 1 mg/L as Fe), and the cationic polymer dosage is 0.2e0.4 mg/L.
5.
Summary
Chemical equilibrium constants are presented as a function of salinity (S ) and temperature for the ion product of water and for important seawater quality parameters. The latter include: the acid-base chemistry of boron, silica, and inorganic carbon. The salinity of seawater shifts pKSW s to lower values (higher equilibrium constants (KSW s)) than for freshwaters making chemical speciation in seawater much different. For the Middle East and tropical areas, water temperatures can be quite high (w35 C). High water temperatures yield lower pKSW s. While alkalinity is often used as an indicator of buffering from addition of acids, it is a capacity measurement. Buffer intensity (b) is a better measure of resistance to specified changes in pH and is discussed with regard to pH control and adjustment in coagulation. Inorganic carbon is the main cause of seawater alkalinity and buffer intensity, but borate ðBðOHÞ 4 Þ also contributes. Commonly occurring contaminants for seawater are mineral particles, algae, and dissolved organic matter that should be removed by pretreatment coagulation and particle separation to minimize RO membrane fouling. The concentrations of these contaminants vary and are influenced by river inputs and algal blooms. To avoid fouling from mineral particles, turbidity following pretreatment should be low (<0.5 NTU with goal of <0.1 NTU) and the SDI should be low (<2). Dissolved natural organic matter and algae can cause biofouling. A collective measure of organic matter is TOC, while UV254 is a good surrogate measure of DOC. The nature of DOC can be characterized by SUVA where SUVA >2 indicates the presence of humic matter. Guidelines for good quality water following coagulation pretreatment for raw seawater with moderate to high TOC are: TOC 1e2 mg/L, UV254 < 0.01 cm1, and SUVA<2. Algae are a particular problem because if carried to RO units, they will undergo lysis due to the transmembrane pressure releasing intracellular organic matter. Algae in seawater may occur at relatively low concentration of <103 cells/mL or at moderate to bloom concentrations (<103 to 106 cells/mL). Coagulation pretreatment and particle separation should maximize removals of algae with removals of 99 percent or greater. Aluminum based coagulants are too soluble for use in pretreatment of seawater. The residual dissolved Al can carry to RO membranes and concentrate causing precipitative scaling. Ferric coagulants are the best choice and they are widely used.
5439
The solubility of Fe in seawater is low over a wide range of pH and temperature conditions for equilibrium conditions with amorphous ferric hydroxide. Ferric chloride doses range typically from 1 to 10 mg/L as Fe depending on (1) raw water quality characteristics, (2) pH of coagulation, and (3) whether a strong acid (e.g., HCl or H2SO4) is used to adjust the pH of coagulation. For raw water containing mineral particle turbidity and low concentrations of algae and TOC, ferric chloride is effective at relatively low dosages and at pH conditions of 7e7.5 for warm waters and 7.5 to 8 for cooler waters. Algae at high concentrations produce AOM and some of these soluble organic compounds create a coagulant demand. Thus, seawater with high algae can require higher Fe doses and lower coagulation pH. For these waters, the optimum pH conditions are 6e6.5 for warm waters and 6.5 to 7 for lower temperatures. For waters influenced by humic matter (SUVA>2), then the coagulant doses and pH conditions are controlled by the humic matter (exert a coagulant demand). Fe dose depends on the TOC concentration and increases with TOC. The optimum pH conditions are 6e6.5 for warm waters and about 6.5 or slightly greater for lower temperatures. A dual-coagulant strategy is recommended for treating seawater containing moderate to high concentrations of algae or seawater containing humic matter. This strategy involves low and constant dosing (1 mg/L or less) with high charge-density cationic polymers and Fe as the main coagulant where it is varied in response to raw water quality changes. The positive charge from the polymer can neutralize the negative charge of particles and can satisfy some of the coagulant demand from the aquatic humic matter or from the algae and AOM, thus reducing metal coagulant dosages.
references
ASTM D4189-07, 2007. Standard Test Method for Silt Density Index (SDI) of Water. doi:10.1520/D4189-07. www.astm.org, Conshohocken, PA. Baghoth, S.A., Sharma, S.K., Amy, G.L., 2011. Tracking natural organic matter (NOM) in a drinking water treatment plant using fluorescence excitation-emission matrices and PARAFAC. Water Research 45 (2), 797e809. Baig, M.B., Al Kutbi, A.A., 1998. Design features of a 20 migd SWRO desalination plant, Al Jubial, Saudi Arabia. Desalination 118, 5e12. Bose, P., Reckhow, D.A., 1998. Adsorption of organic matter on preformed aluminum hydroxide flocs. Journal of Environmental Engineering 124 (9), 803e811. Drever, J.I., 1997. The Geochemistry of Natural Waters, third ed. Prentice Hall, Englewood Cliffs, NJ. Duan, J., Graham, N.J.D., Wilson, F., 2002. Coagulation of humic acid by ferric chloride in saline (marine) water conditions. Water Science and Technology 47 (1), 41e48. Duranceau, S.J., Taylor, J.S., 2011. Membranes, chapter 11 in water quality and treatment. In: . Edzwald, J.K. (Ed.), A Handbook on Drinking Water, sixth ed. McGraw-Hill, New York. Eby, G.N., 2004. Principles of Environmental Geochemistry. Thomson/Brooks Cole, Pacific Grove, CA. Edzwald, J.K., 1993. Coagulation in drinking water treatment: particles, organics, and coagulants. Water Science and Technology 27 (11), 21e35. Edzwald, J.K., 2010. Dissolved air flotation and me. Water Research 44 (7), 2077e2106.
5440
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 2 8 e5 4 4 0
Edzwald, J.K., Tobiason, J.E., 2011. Chemical principles, source water composition, and watershed protection, chapter 3 in water quality and treatment. In: Edzwald, J.K. (Ed.), A Handbook on Drinking Water, sixth ed. McGraw-Hill, New York. Edzwald, J.K., Van Benschoten, J.E., 1990. In: Hahn, H.H., Klute, R. (Eds.), Aluminum Coagulation of Natural Organic Matter, Chapter in Chemical Water and Wastewater Treatment. Springer-Verlag, New York. Edzwald, J.K., Upchurch, J.B., O’Melia, C.R., 1974. Coagulation in estuaries. Environmental Science and Technology 8 (1), 58e63. Edzwald, J.K., Becker, W.C., Wattier, K.L., 1985. Surrogate parameters for monitoring organic matter and THM Precursors. Journal of the American Water Works Association 77 (4), 122e131. Freeman, S., 2009. Seawater Pretreatment Operations Presentation at the workshop On Coagulation for Seawater and Reuse applications, AWWA, water quality Technology Conference, November 15, 2009, Seattle, WA. Ghernaout, B., Ghernaout, D., Saiba, A., 2010. Algae and cyanotoxins removal by coagulation/flocculation: a review. Desalination and Water Treatment 20 (1e3), 133e143. Henerdson, R., Parsons, S.A., Jefferson, B., 2008. The impact of algal properties and pre-oxidation on solideliquid separation of algae. Water Research 42, 1827e1845. Millero, F.J., 1995. Thermodynamics of the carbon dioxide system in the ocean. Geochimica et Cosmochimica Acta 59 (4), 661e677. Nordstrom, D.K., May, H.M., 1996. Aqueous equilibrium data for mononuclear aluminum species. In: Sposito, G. (Ed.), The Environmental Chemistry of Aluminum. CRC Press, Boca Raton, pp. 39e80. Para, J., Coble, P.G., Tedetti, M., Sempe´re´, R., 2010. Fluorescence and absorption properties of chromophoric dissolved organic
matter (CDOM in coastal surface waters of the Northwestern Mediterranean Sea (Bay of Marseilles, France). Biogeosciences Discussions 7 (4), 5675e5718. Pernitsky, D.J., Edzwald, J.K., 2003. Solubility of polyaluminum coagulants. Journal of Water Supply: Research and Technology e Aqua 52 (6), 395e406. Post, G.B., Atherholt, T.B., Cohn, P.D., 2011. Health and aesthetic aspects of drinking water, chapter 2 in water quality and treatment. In: Edzwald, J.K. (Ed.), A Handbook on Drinking Water, sixth ed. McGraw-Hill, New York. Schneider, O., 2011. Personal Communication. American Water, Voorhees, NJ. Sengco, M.R., 2001. The Aggregation of clay minerals and marine Microalgal cells: Physicochemical Theory and Implications for controlling harmful algal blooms. Doctoral dissertation, Massachusetts Institute of Technology and Woods Hole Oceanographic Institution. Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry, third ed. WileyInterscience, NY. Timmes, T.C., Kim, H., Dempsey, B.A., 2009. Electrocoagulation pretreatment of seawater prior to ultrafiltration: bench-scale applications for military water purification systems. Desalination 249, 895e901. Van Benschoten, J.E., Edzwald, J.K., 1990. Chemical aspects of coagulation using aluminium salts. I. Hydrolytic reactions of alum and polyaluminium chloride. Water Research. 24 (12), 1519e1526. Voutchkov, N., 2010a. Seawater Pretreatment. Water Treatment Academy, Bangkok, Thailand. Voutchkov, N., 2010b. Considerations for selection of seawater filtration pretreatment system. Desalination 261, 354e364. WHO (accessed January 2011) www.who.int/water_sanitation_ health/dwq/en.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Partitioning of dissolved organic matter-bound mercury between a hydrophobic surface and polysulfide-rubber polymer Eun-Ah Kim, Richard G. Luthy* Department of Civil and Environmental Engineering, Stanford University, Stanford, CA 94305-4020, USA
article info
abstract
Article history:
This study investigated the role of dissolved organic matter on mercury partitioning
Received 23 April 2011
between a hydrophobic surface (polyethylene, PE) and a reduced sulfur-rich surface (pol-
Received in revised form
ysulfide rubber, PSR). Comparative sorption studies employed polyethylene and poly-
23 June 2011
ethylene coated with PSR for reactions with DOM-bound mercuric ions. These studies
Accepted 3 August 2011
revealed that PSR enhanced the Hg-DOM removal from water when DOM was Suwannee
Available online 11 August 2011
River natural organic matter (NOM), fulvic acid (FA), or humic acid (HA), while the same amount of 1,3-propanedithiol-bound mercuric ion was removed by both PE and PSR-PE.
Keywords:
The differences for Hg-DOM removal efficiencies between PE and PSR-PE varied depend-
Dissolved organic matter
ing on which DOM was bound to mercuric ion as suggested by the PE/water and PSR-PE/
Polysulfide-rubber polymer
water partition coefficients for mercury. The surface concentrations of mercury on PE
Mercury
and PSR-PE with the same DOM measured by x-ray photoelectron spectroscopy were
Addition reaction
similar, which indicated the comparable amounts of immobilized mercury on PE and PSR-
Adsorption reaction
PE being exposed to the aqueous phase. With these observations, two major pathways for
Mercury complexation
the immobilization reactions between PSR-PE and Hg-DOM were examined: 1) adsorption of Hg-DOM on PE by hydrophobic interactions between DOM and PE, and 2) addition reaction of Hg-DOM onto PSR by a complexation reaction between Hg and PSR. The percent contribution of each pathway was derived from a mass balance and the ratios among aqueous mercury, PE-bound Hg-DOM, and PSR-bound Hg-DOM concentrations. The results indicate strong binding of mercuric ion with both dissolved organic matter and PSR polymer. The FT-IR examination of Hg-preloaded-PSR-PEs after the reaction with DOM corroborated a strong interaction between mercuric ion and 1,3-propanedithiol compared to Hg-HA, Hg-FA, or Hg-NOM interactions. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Methylmercury is a potent neurotoxin and has high bioaccumulation factors (Driscoll et al., 2007). A recent study showed an enhanced mercury methylation rate with Geobacter sulfurreducens when mercuric ion is bound to cysteine
(Schaefer and Morel, 2009), a small organic molecule with a thiol group. This finding implies enhanced methylation of mercury in the presence of certain organic molecules in sediments, which may depend on the strains of the mercury methylating bacteria. Therefore, it is crucial to have an accurate estimation of the amount of mercury that could be
* Corresponding author. E-mail address:
[email protected] (R.G. Luthy). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.003
5442
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
transformed into a source for methylation when there are perturbations of the sediment organic carbon speciation due to erosion, runoff, changes of pH/redox potentials, or sudden input of sediment organic matter. Since the mobile mercury is more readily bioavailable for mercury methylation compared with solid-bound mercury species (Benoit et al., 2001a,b; Skyllberg et al., 2009), the mobile portion of mercury is a better proxy for methylmercury concentration than the total mercury concentration. Mercury mobilization is related the presence of other metal binding ligands such as dissolved organic matter (DOM) and organic or inorganic thiols/sulfides (Ravichandran, 2004; Skyllberg, 2008). The latter would include DOM with reduced sulfur functional groups that strongly bind mercuric ion and prevent the mercury from precipitating as mercuric sulfide. Dissolution of Hg-S(s) by DOM (Ravichandran et al., 1998; Waples et al., 2005; Slowey, 2010) indicates a strong interaction between DOM and Hg, and critical roles of DOM in mercury mobilization. The implications of strong Hg-DOM interaction are yet unclear with respect to mercury partitioning in the presence of a strong binding ligand for mercuric ion on a solid surface. For example, with polysulfide-rubber (PSR) polymer, the DOM may compete with PSR for mercury binding, or PSR may simply provide additional binding sites for DOM-bound mercuric ion without competition. These interactions may be examined by preparing polysulfide-rubber polymer-coated polyethylene (PSR-PE) and conducting competitive sorption studies with various forms of DOM. Assuming Hg-DOM interaction mainly comprises a HgeS bond, immobilization of the Hg-DOM species on the reduced sulfur-rich sites can occur via multiple HgeS bond formation (Hesterberg et al., 2001). In this case, the Hg ion would be encapsulated in DOM and the PSR polymer simultaneously. An exchange of mercuric ion between DOM and PSR is also possible when DOM-bound mercuric ion is transferred to PSR, subsequently reaching a new equilibrium between Hg-DOM and Hg-PSR. In summary, the possible reaction pathways between PSR-PE and Hg-DOM can be classified as 1) adsorption of Hg-DOM on PE via hydrophobic interaction giving a Hg-DOM-PE species, and 2) additional bond formation between PSR and Hg-DOM giving a Hg-PSR species. The purpose of this study is to assess the role of dissolved organic matter on mercury partitioning between a hydrophobic surface (i.e., polyethylene) and a reduced sulfur-rich surface (i.e., polysulfide-rubber polymer). Better understanding of the partitioning behavior of Hg-DOM with PSR-PE or PE is expected to provide clues to delineate the Hg-DOM immobilization processes on PSR-PE. The results indicate that both PE and PSR participate in Hg-DOM removal from water, and mercury complexation with PSR polymer (6.7% of the total PSR-PE surface area) and DOM contributes greatly to the overall immobilization reaction on PSR-PE in the presence of Suwannee River natural organic matter, fulvic acid, and humic acid. Depending on the relative affinities of a hydrophobic surface and a polysulfide-rich surface for DOM-bound mercuric ion, the mercury removal efficiency of a multifunctional sorbent, for instance, PSR-coated activated carbon (Kim et al., 2011), can be optimized by adjusting the coverage of PSR polymer on activated carbon.
2.
Materials and methods
2.1.
Polymer coating on polyethylene strips
Polyethylene (PE) strips were pre-cleaned with methylenechloride, methanol, and DI-water consecutively for one day at each step. The PE was dried in a convection oven at 60 C for 4 h and cut into 2 cm 2 cm (18 0.5 mg) pieces. The polysulfide-rubber polymer was synthesized following the procedure described by Kalaee et al. (2009). Condensation polymerization between sodium tetrasulfide and 1,2dichloroethane, using methytributyllammonium chloride as a phase transfer catalyst, produced a yellowish elastic solid in water. One hundred mg of the PSR polymer in 40 mL toluene was refluxed until the polymer block was completely dissolved. The solution was cooled to room temperature before it was used for coating PE strips. A piece of PE was dipped into the PSR solution for less than 1 min and taken out for an immediate drying in air for 10 s. The PSR-PE was dried again under vacuum for 1 h. The sulfur content of the polymer-coated polyethylene strip was determined in duplicate by elemental analysis (Atlantic Microlab, GA).
2.2.
Surface reactions of Hg-DOM on PE or PSR-PE
Suwannee River natural organic matter (NOM), and fulvic acid (FA) were obtained from the International Humic Substance Society (IHSS). Humic acid (HA) and 1,3propanedithiol were purchased from Sigma Aldrich and Alfa Aesar. Because the source and the isolation method for Sigma Aldrich HA are different from those for IHSS HA, the differences between FA and HA reported in this paper do not necessarily represent any FA and HA properties in a specific natural system. Suwannee River NOM, Suwannee River fulvic acid (FA) and Sigma Aldrich humic acid (HA) were dissolved in 250 mL borosilicate glass bottles to make 10 mg DOM L1 solutions. Suwannee River NOM and FA dissolved in water readily, but the HA solution was sonicated until its complete dissolution. The solutions were filtered through 0.45 mm polyvinylidene fluoride (PVDF) membrane filters to remove particulate matter. An aqueous solution of 1,3-propanedithiol was freshly prepared by dissolving 1,3-propanedithiol in 1 M NaOH and diluted 400 fold to make 5.41 mg L1 solution. HgCl2 stock solution in 1 M HCl was added to each solution of dissolved organic matter (DOM) to make 10 ppb (50 nM) Hg solutions. Because Suwannee River NOM, FA, and HA have 0.65 wt%, 0.44 wt% and 0.96 wt% sulfur respectively, according to the elemental analysis results reported by IHSS and Pitois et al. (2008), 10 mg L1 DOM solutions have 1.4e3.0 mM sulfur, which is an excess amount for mercury binding in 50 nM Hg solutions. The pH was adjusted to 7 with 0.01M potassium monophosphate buffer. The mixtures of mercuric ion and DOM were shaken for one week to allow sufficient reaction time for forming Hg-DOM. One piece of PE or PSR-PE was in contact with 40 mL of these solutions for 4 weeks. The PE or the PSR-PE strip was taken out, washed with DI-water, and gently pressed on Kimwipes to remove water on a PE or a PSR-PE strip. PSR-PEs after the reaction with Hg-DOM were
5443
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
washed with MilliQ water followed by drying in vacuum for 1 h. The remaining aqueous solution was preserved by adding 400 mL of BrCl.
2.3.
Total mercury analysis
Four mL of preserved duplicate samples was filtered through 0.45 mm polyvinylidene fluoride (PVDF) membrane filter. The filtrate was diluted to ensure mercury concentrations were within the detection range of 0.5e400 ng L1. The total mercury concentrations were measured by Tekran 2600 cold vapor atomic fluorescent spectrometry (CVAFS) following the US EPA (Environmental Protection Agency) method 1631 revision E.
2.5.
Results and discussion
3.1. Synthesis of PSR-PE and proposed reaction pathways
DOM contact with Hg pre-loaded PSR-PE
HgCl2 stock solution in 1 M HCl was used to make 10 ppm Hg solution in 0.01M potassium monophosphate buffer to maintain the pH at 7. One piece of PSR-PE was in contact with 40 mL of 10 ppm Hg solutions for 1 week. The PSR-PE strips were taken out, washed with DI-water, and gently pressed on Kimwipes to remove water on the PSR-PE strip. The Hgloaded PSR-PE was then in contact with 40 mL of 10 mg L1 NOM, FA, HA, or 5.41 mg L1 1,3-propanedithiol for 4 weeks. The PSR-PE strips were taken out after the reactions, washed with DI-water, and gently pressed on Kimwipes to remove water on the PE strips. The PSR-PE strips were dried for 1 h in vacuum and analyzed by Fourier transform infra-red spectroscopy (FT-IR) and x-ray photoelectron spectroscopy (XPS). The remaining aqueous solution was preserved by adding 400 mL of BrCl to the reaction vessels.
2.4.
3.
X-ray photoelectron spectroscopy (XPS)
XPS techniques were used to determine the surface concentrations of sulfur and mercury on PSR-PE, PE, or Hg-preloaded PSR-PE before and after the reactions with Hg-DOM or DOM. Three spots for each sample were analyzed to account for non-uniform distributions of sulfur and mercury atoms on the PSR-PE surface. PHI 5000 Versa-Probe scanning XPS microprobe with Al Ka x-ray radiation (1486 eV) was used under high vacuum condition (below 105 Pa). Charging effects by the poor surface conductivity were minimized by applying 10 eV argon ions. Analytical sample size for the survey scans was 1 1 mm. An averaged spectrum from five survey scans over 0e1000 eV was obtained with a resolution of 1 eV.
The acronyms, definitions, and units for the terms used in this paper are summarized in Table 1. The synthesis procedure for PSR-PEs involves a solution casting of PE strips with PSR polymer solution in toluene, and drying in vacuum to produce a thin layer of PSR polymer over the PE surface as depicted in Fig. 1. The surface areas of PSR and PE are assumed to be proportional to the atomic counts of the constituting atoms (C and S for PSR, and C for PE) based on the XPS analysis of PSRPE before the reactions with DOM-bound Hg. The PSR is constituted of the repeating eC2S4- segments, and with the average atomic % of sulfur on PSR-PE at 4.0 atom%, the contribution of PSR to the total carbon atomic count is approximately 2.0 atom%. Accounting for the different atomic radii of sulfur and carbon, 0.109 nm and 0.091 nm respectively, the average fractions of PSR (fPSR) and PE (fPE) on a PSR-PE strip were determined as fPSR ¼ 0.067, and fPE ¼ 0.933. The possible reactions of DOM-bound mercuric ion with a PE strip or a PSRPE strip are illustrated in Fig. 2. Each Hg-DOM species (NOM, FA, HA, or 1,3-propanedithiol) was reacted with PE or PSR-PE separately for comparison. As illustrated schematically in Fig. 2, Hg-PSR and Hg-DOM-PE are proposed as the products of the two possible reaction pathways after the DOM-bound mercuric ion encounters PSR-PE. These reactions represent the sorption of Hg-DOM to the surface via direct Hg-sulfur interactions (Hg-PSR) or indirectly by DOM-PE hydrophobic interactions (Hg-DOM-PE). The reaction represented by Hg-PSR differs from that for Hg-DOM-PE in terms of where a mercuric ion is situated. Mercuric ion binds both with DOM and polysulfide in Hg-PSR, after which DOM may be detached from mercuric ion. However, in the case of the Hg-DOM-PE pathway, which utilizes DOM hydrophobic interaction with polyethylene,
Table 1 e Summary of the definitions and units for the terms that describe the reaction of mercury species with polyethylene (PE) and polysulfide-rubber coating on polyethylene (PSR-PE). Acronym
Definition
Unit
[Hg-DOM] [Hg-PSR]
total aqueous mercury concentration mercury associated with PSR per unit area of PSR-PE mercury associated with PE via DOM per unit area of PSR-PE total mass of mercury sorbed per unit area of PSR-PE total mass of mercury in the water and PSR-PE volume of water PE area on PSR-PE PSR area on PSR-PE area of PSR-PE SAPSR/SAPSR-PE SAPE/SAPSR-PE
mg Hg L1 mg Hg m2
[Hg-DOM-PE] [Hg]sorbed
2.6.
FT-IR analysis of PSR-PE
Far infra-red (FIR) spectra of Hg-pre-loaded PSR-PE (Hg-PSRPE) after the reaction with DOM were obtained with Bruker Vertex 70 FT-IR spectrometer using a deuterated triglycine sulfate (DTGS) detector. A piece of PSR-PE was placed where the IR beam passes through perpendicularly. Forty scans were averaged for each spectrum.
Hgtot Vaq SAPE SAPSR SAPSR-PE fPSR fPE
mg Hg m2 mg Hg m2 mg Hg L m2 m2 m2 e e
5444
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
a
Synthesis ofpolysulfide-rubber-coated polyethylene (PSR-PE)
2cm X 2cm X 51µm
b
Reactions with dissolved organic matter (DOM)-bound mercury
Fig. 1 e a) Synthesis of polysulfide-rubber-coated polyethylene (PSR-PE) with a piece of PE and PSR polymer solution in toluene, showing a schematic distribution of sulfur atoms on PSR-PE, b) reactions between Hg-DOM (NOM, FA, HA, and 1,3propanedithiol) and PE or PSR-PE. The mercury ion is not necessarily situated at the exterior of NOM, FA or HA.
mercuric ion does not form a chemical bond with polysulfide. The total aqueous mercury concentration measured after the reaction between PSR-PE and Hg-DOM is defined as [HgDOM]. Due to the 1000-fold mass ratio of DOM to Hg and the high stabilization constants for the association reactions between mercuric ion and DOM (Benoit et al., 2001a,b; Khwaja et al., 2006; Miller et al., 2009; Dong et al., 2010), DOM-bound mercuric ion would be the major constituent of total mercury concentration in the aqueous phase. [Hg]sorbed can be defined by using a mass balance with the total aqueous mercury concentrations before and after the reaction with DOM-bound mercuric ion (equation (1), (2)).
the reaction vessel (glass vial) because BrCl, a strong oxidant, was directly spiked to the vial after PSR-PE strip was removed so that BrCl would oxidize and extract any trace mercury from the glass wall and the lid. However, such mercury partitioning on the glass vial was negligible because the mercuric ion loss by the same glass vial with the same or higher DOM concentration was not significant according to the previous experimental data not reported here. Hg-DOM-PE.
Hgtot ¼ Vaq ,½Hg DOM þ SAPSRPE ,½Hgsorbed
(1)
½Hgsorbed ¼ ½Hg PSR þ Hg DOM PE
(2)
In order to estimate overall Hg-DOM removal efficiencies by PSR-PE, dissolved mercury concentrations after the reactions were measured and compared with the Hg-DOM solutions without any sorbents. Because dissolved humic substances, especially those with reduced sulfur groups are the main competitors for mercuric ion, we tested dissolved natural organic matter, fulvic acid, humic acid, and 1,3propanedithiol (PDT) as representatives of reactive constituents in sediment pore water. The results are depicted in Fig. 3, showing a strong partitioning of PDT-bound mercuric ion as well as HgCl2 by PSR-PE. The partition coefficient K1 is defined as the ratio of the total mercury concentration on PSR-PE surface to the total aqueous mercury concentration (equation (3)).
By our experimental methods, [Hg-DOM] includes the amount of aqueous phase mercury and any mercury loss by
Fig. 2 e Possible interactions between DOM-bound mercury (NOM, FA, HA, or 1,3-propanedithiol) and PSR-PE, which results in either PSR-bound Hg (Hg-PSR) via an addition reaction as illustrated by the first pathway, or PEbound Hg-DOM (Hg-DOM-PE) via hydrophobic adsorption reaction on the PE surface as illustrated by the second pathway.
3.2. Surface reaction of DOM-bound mercuric ion on PSR-PE
K1 ¼
Hgtot Vaq,½Hg DOM ½Hgsorbed ½Hgsorbed ¼ Lm2 h ½Hgaqueous ½Hg DOM SAPSRPE ,½Hg DOM (3)
Thus, the overall removal processes incorporated in K1 represent the combination of the two proposed reaction pathways. As shown in Fig. 3, the partition coefficients vary from 13 to 115 L m2.
5445
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
150
200
K2 [L m-2 ]
K1[L m-2]
150 100
50
50
0
0 NONE
NOM
FA
HA
NONE
PDT
Fig. 3 e Partition coefficients (K1) of mercury between PSRPE and water as results from the reactions between HgDOM and PSR-PE. These K1 values correspond to 21e70% removal of the initial Hg-DOM in 40 mL vials with 8 cm2 PSR-PE having 6.7% polysulfide rubber.
3.3. Reactions between PE strip and DOM-bound mercuric ion Because the complexation of mercuric ion with DOM itself transforms ionic mercury into a more hydrophobic species, namely Hg-DOM, the PE surface can provide sorption sites for Hg-DOM via a favorable DOM-PE hydrophobic interaction, i.e., the second pathway in Fig. 2. The PE/water partition coefficient with Hg-DOM gives a good criterion to estimate how much Hg-DOM sorption on PE alone contributes to the overall reaction between PSR-PE with Hg-DOM. A hydrophobic partition coefficient assigned for this reaction equals the ratio of [Hg-DOM-PE] to the total aqueous mercury species (equation (4)) ½Hg DOM PE fPE ½Hg DOM PEPEonly ¼ ½Hg DOM ½Hg DOM
NOM
FA
HA
PDT
Additives (DOM)
Additives (DOM)
K2 h
100
(4)
where Hg-DOM-PE denotes PSR-PE-bound Hg-DOM due to a favorable hydrophobic interaction between DOM and PE. As shown in Fig. 4 and 1,3-propanedithiol-bound mercuric ion has a high affinity for PE, which indicates a significant encapsulation of mercuric ion with a hydrophobic bidentate ligand, and a large contribution of the adsorption reaction to the overall reaction between PSR-PE and Hg-PDT. In contrast, the mercuric ion complexed with Suwannee River natural organic matter, fulvic acid, or humic acid as well as HgCl2 does not exhibit high removal efficiency by PE alone (Fig. 4) compared with that by PSR-PE (Fig. 3). These differences suggest the overall reactions between PSR-PE and Hg-DOM (NOM, FA, or HA) comprise other reactions than the hydrophobic adsorption reactions between DOM and the PE itself.
3.4. Comparison on the atomic % of mercury on PSR-PE and PE surfaces after the reactions with Hg-DOM The surface concentrations of mercury on PSR-PE and PE were measured after the reaction with Hg-DOM. The XPS
Fig. 4 e Partition coefficients (K2) of mercury ion between PE alone and water after the reaction with Hg-DOM. In comparison to PSR-PE/water partitioning of Hg-DOM shown in Fig. 3, these data indicate relatively weak hydrophobic interactions between Hg-DOM and PE except for 1,3-propanedithiol. The columns represent the average values of the triplicates, and the error bars correspond to the standard deviations. These K2 values correspond to 5e72% removal of the initial DOM-bound Hg.
analyses were used to estimate the proportion of the easily accessible (i.e., not significantly covered by organic matter) mercury among the total immobilized mercury on PSR-PE. Because XPS is a surface sensitive technique that measures the atomic compositions in the top w10 nm layer, any deposition of organic substances over mercuric ion will shield the escaping electron and reduce the signal. Therefore, we can qualitatively estimate the extent of the shielding effect from DOM or PSR by comparing the total immobilized mercury concentrations on PSR-PE or PE as exhibited in Fig. 3 or Fig. 4 and the XPS results as shown in Fig. 5. Whereas the Hg-DOM removal efficiency of PSR-PE (Table S1) ranges from 96% to 531% of the same Hg-DOM removal by PE, the estimated surface concentrations of mercury on PSR-PE and PE (Fig. 5) after the reactions with the same Hg-DOM species are not significantly different with each other. This indicates that Hg-DOM removal by PSR-PE involves partial covering of Hg ions with DOM, or migration of mercuric ion into the inner PSR layer.
3.5. Overall Hg-DOM immobilization reaction pathways with PSR-PE Table 2 summarizes two partition coefficients obtained from the two sets of surface reactions, one with PSR-PE and HgDOM, and the other with PE and Hg-DOM. Since K1 and K2 formulate two equations for three unknowns, the two classes of sorbed mercury concentrations, [Hg-PSR] and [Hg-DOM-PE], can be expressed as functions of the aqueous mercury concentration, [Hg-DOM]. With these constants and the following reaction model equations (5)e(7), f add : PSR PE þ Hg DOM/Hg PSR
(5)
5446
Atomic %
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
0.5 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0
Hg on PE Hg on PSR-PE
NOM
FA
HA
PDT
Additives (DOM) Fig. 5 e X-ray photoelectron spectroscopic analysis of the surface concentrations (atomic %) of mercury on PE or PSR-PE after reaction with Hg-DOM. The columns represent the average values of the triplicates, and the error bars correspond to the standard deviations.
f ads : PSR PE þ Hg DOM/Hg DOM PE
(6)
Overall reaction: PSR PE þ Hg DOM/f add Hg PSR þ f ads Hg DOM PE
(7)
we can calculate how much each reaction pathway contributes to the overall reaction. By definition, a contribution factor (fadd, or fads) means the fractions of addition (via pathway 1) or adsorption (via pathway 2) reaction to the overall reaction, which can be defined as a ratio of each corresponding reaction product concentration to the total surface mercury concentration (equations (8)e(10)).
f add ¼
½Hg PSR ½Hgsorbed
(8)
f ads ¼
½Hg DOM PE ½Hgsorbed
(9)
f add þ f ads ¼ 1
(10)
The definitions of the reaction constants and the solutions for fadd and fads are summarized in Table 3. The results shown in Table 4 reveal that the reaction of Hg-DOM on PSR-PE is mainly by the addition reaction, i.e., via pathway 1, with complexation of mercuric ion with PSR, except for the case with 1,3-propanedithiol. The hydrophobic partitioning reaction has the highest importance in the PDT-mediated mercury sorption on PSR-PE. The dependence of the major reaction pathway on DOM may stem from the Hg-DOM binding strength or the bulkiness of Hg-DOM, which varies with DOM. Suwannee River NOM, FA, and HA solutions have excess amounts of sulfur (1.4e3.0 mM) for mercury binding in 50 nM Hg solutions, and the proportion of reduced sulfur groups in river DOM ranges from 13 to 36% of the total sulfur (Ravichandran, 2004). Therefore, the amount of reduced sulfur atoms in the DOM solutions would not affect the overall affinity of DOM for mercuric ion. Instead, the interaction of mercuric ion with
Table 2 e Summary of the partition coefficients K1 and K2. DOM None Suwannee River natural organic matter Suwannee River fulvic acid Humic acid 1,3-propanedithiol
K1 (errora) [L m2]
K2 (error) [L m2]
86.6 (4.3) 13.2 (6.4)
6.61 (0.66) 4.15 (0.50)
38.2 (17.7) 14.8 (2.2) 115 (5.4)
4.15 (0.53) 2.71 (0.89) 122 (32.1)
a The errors in K1 and K2 are sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi error in½Hgsorbed 2 error in½Hg DOM 2 , K1 þ ½Hgsorbed ½Hg DOM sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi 2 2 error in½Hg PE error in½Hg DOM K2 þ ½Hg PE ½Hg DOM (Valca´rcel, 2000).
respectively
Table 3 e Summary of the partition coefficients and the solutions of the contribution factors, fadd, and fads, expressed in terms of the defined constants, K1, and K2.a Definitions of K1, K2
Solutions of fadd, fads
½Hgsorbed hK1 ½Hg DOM
½Hg PSR ðK1 K2 Þ fadd ¼ ¼ ¼ 1 fads K1 Hg sorbed
½Hg DOM PE hK2 ½Hg DOM
fads ¼
½Hg DOM PE K2 ¼ K1 ½Hgsorbed
a K1 is derived from the reactions of Hg-DOM with PSR-PE; K1 is the ratio of the total surface mercury concentration to the total aqueous mercury concentration. K2 is a partition coefficient of HgDOM between water and PE obtained from the Hg-DOM adsorption study with the PE strips; K2 from the Hg-DOM reaction with pure PE is multiplied by fPE to reflect the PE surface area on PSR-PE.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
Table 4 e Relative significance of the addition via pathway 1 (fadd), and the hydrophobic adsorption via pathway 2 (fads) reactions for Hg-DOM partitioning to PSRPE having 6.7% polysulfide. DOM
a
b
fadd (error )
fads (error )
None Suwannee River natural organic matter Suwannee River fulvic acid Humic acid 1,3-propanedithiol
0.924 (0.008) 0.686 (0.157)
0.076 (0.008) 0.314 (0.157)
0.891 (0.052) 0.817 (0.066) 0 (0.283)
0.109 (0.052) 0.183 (0.066) 1.0c (0.283)
a The error in fadd ¼ 1 e fads equals to the corresponding error in fads. rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi error in K1 error in K2 b The error in fads ¼ fads $ ð Þ2 þð Þ2 (Valca´rcel, K1 K2 2000). c The algebraic value for fads ¼ K2/K1 is 1.06. It was approximated to the maximum fads value, 1.0.
a reduced sulfur group in a DOM molecule and other adjacent reactive binding sites in DOM would determine Hg-DOM binding strength. The different DOM molecule sizes can also explain the difference in fadd because the limited PSR surface area (0.536 cm2) can only accommodate 4.5 10105.6 1016 mol of Hg-DOM out of 2 109 mol of the total Hg-DOM when Hg-DOM is assumed to be a sphere with diameter 0.5e450 nm, where 0.5 nm corresponds to the approximate size of one 1,3-propanedithiol molecule, and 450 nm corresponds to the filter pore size that the NOM, FA and HA molecules passed in preparation of the DOM solutions. Therefore, when the surface area is the limiting factor, a smaller Hg-DOM molecule would be able to react more extensively with PSR regardless of the Hg-DOM bond strength, and fadd may not provide information about the relative binding strength of Hg-DOM among the tested DOM. In this case, an analogous experiment with excess surface area of PSR for accommodating Hg-DOM may reveal the relative Hg-DOM bond strength.
5447
3.6. Changes in the chemical bonding between PSR and mercuric ion in the presence of DOM The reaction between Hg-preloaded PSR-PE with DOM provides information about the Hg-PSR bond strength, as DOM can dissolve Hg ions from PSR. Mercuric chloride was preloaded on PSR-PE, which produced Hg-PSR-PE. This surface was then reacted with DOM for one month. As the addition reaction proceeds, the PSR-PE surface would be covered with DOM, and Hg-DOM could be dissolved into the water depending on the Hg-DOM binding strength. It is also possible to have a subsequent migration of DOM-bound mercuric ion toward the hydrophobic PE surface. Our data on the aqueous mercury concentration after the reaction with each DOM (Table S2) reveal that the total amount of the immobilized mercury remaining on PSR-PE is close to that without DOM, which indicates an insignificant amount of remobilization of PSR-PE-bound mercuric ion by DOM. One way to assess the amount of Hg-PSR bond breakage or weakening by DOM is to analyze FT-IR spectra around 350e390 cm1 for the characteristic IR peaks for Hg-S (AlJeboori et al., 2010). As shown in Fig. 6, without any DOM, the major peak for Hg-S appears at 354 cm1, which is lower than the emerging peaks at around 376 cm1 in the presence of DOM. This change indicates that new Hg-S bonds are formed with DOM and existing bonds between polysulfide and mercuric ion are weakened or broken. The extent of the change is most prominent with 1,3-propanedithiol, which has two highly reactive thiol groups per molecule to compete with PSR for mercuric ion. According to the decrease in the absorbance intensities at 354 cm1 with respect to those at 376 cm1 as shown in Fig. 6, the extent of perturbation to Hg-PSR bond is comparable among NOM, FA, and HA. A diminished binding strength of Hg-PSR does not necessarily imply a lower stability of the immobilized mercuric ion unless Hg-DOM formation results in its transport into the water. The favorable interaction between DOM and hydrophobic PE surface or fouling of Hg-PSR by the DOM
0.35
Absorbance
0.3
Hg-PSR-PE-buffer
0.25
Hg-PSR-PE-fulvic acid
0.2
Hg-PSR-PE-humic acid
0.15
Hg-PSR-PE-NOM
0.1 Hg-PSR-PEpropanedithiol
0.05 0 320
340
360
380
400
420
Wave number (cm-1) Fig. 6 e FT-IR spectra (far-infrared region) of Hg-PSR-PE after the reaction with various DOM for one month. Hg-PSR-PEbuffer represents a control group that has no dissolved organic matter.
5448
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 1 e5 4 4 8
could compensate for any diminishment of Hg-PSR bonding and keep the immobilized mercury from being released into the solution.
4.
Conclusion
A comparison of Hg-DOM interactions with a hydrophobic surface (PE) and a sulfur-rich surface (PSR-PE) showed the stronger interactions of Hg-DOM with PSR-PE. A greater amount of Hg-DOM was removed by PSR-PE than PE when NOM, FA, or HA was present in the solution, during which DOM accumulation over mercury or migration of mercuric ion into the inner PSR layer occurred. This implies a significant contribution of the PSR-mediated reaction to the overall HgDOM immobilization reaction. The organic compound, 1,3propanedithiol, was examined as a strong competing ligand for mercuric ion, and 1,3-propanedithiol-bound mercuric ion was most effectively removed via an adsorption reaction pathway. The changes in the relative peak intensities at 354 cm1 and 376 cm1 in the FT-IR spectra of HgCl2-preloaded PSR-PE after the reaction with DOM suggest that partial breakage or weakening of PSR-Hg bonds took place with additional complexation with DOM. The magnitude of the change is most prominent with 1,3-propanedithiol. Since both PSR and PE participate in Hg-DOM removal from water, it is beneficial to develop a multi-functional sorbent that has high affinities for both DOM and mercuric ions in order to achieve high mercury removal efficiencies in sediments. Depending on the fadd and fads values for the Hg-DOM of concern, different ratios of hydrophobic/reduced-sulfur-rich sorbent can be determined for an optimum Hg-DOM removal efficiency.
Acknowledgments The authors acknowledge the National Institute of Environmental Health Sciences (NIEHS), grant number R01 ES016143-02, for the financial support of this study.
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.003.
references
Al-Jeboori, M.J., Al-Tawel, H.H., Ahmad, R.M., 2010. New metal complexes N2S2 tetradentate ligands: synthesis and spectral studies. Inorganica Chim. Acta 363 (6), 1301e1305. Benoit, J.M., Gilmour, C.C., Mason, R.P., 2001a. The influence of sulfide on solid-phase mercury bioavailability for methylation
by pure cultures of Desulfobulbus propionicus (1pr3). Environ. Sci. Technol. 35 (1), 127e132. Benoit, J.M., Mason, R.P., Gilmour, C.C., Aiken, G.R., 2001b. Constants for mercury binding by dissolved organic matter isolated from the Florida Everglades. Geochimica et Cosmochimica Acta 65 (24), 4445e4451. Dong, W., Liang, L., Brooks, S., Southworth, G., Gu, B., 2010. Roles of dissolved organic matter in the speciation of mercury and methylmercury in a contaminated ecosystem in Oak Ridge, Tennessee. Environ. Chem. 7 (1), 94e102. Driscoll, C.T., Lambert, K.F., Kamman, N., Holsen, T., Han, Y.-J., Chen, C., Coodale, W., Butler, T., Clair, T., Munson, R., 2007. Mercury matters-linking mercury science and public policy in the Northeastern United States. Sci. Links Publ. 1 (3), 11. Hesterberg, D., Chou, J.W., Hutchison, K.J., Sayers, D.E., 2001. Bonding of Hg(II) to reduced organic sulfur in humic acid as affected by S/Hg ratio. Environ. Sci. Technol. 35 (13), 2741e2745. Kalaee, M.R., Famili, M.H.N., Mahdavi, H., 2009. Synthesis and characterization of polysulfide rubber using phase transfer catalyst. Macromol. Symp. 277 (1), 81e86. Khwaja, A.R., Bloom, P.R., Brezonik, P.L., 2006. Binding constants of divalent mercury (Hg2þ) in soil humic acids and soil organic matter. Environ. Sci. Technol. 40 (3), 844e849. Kim, E.A., Seyfferth, A.L., Fendorf, S., Luthy, R.G., 2011. Immobilization of Hg(II) in water with polysulfide-rubber (PSR) polymer-coated activated carbon. Water Res. 45 (2), 453e460. Miller, C.L., Southworth, G., Brooks, S., Liang, L., Gu, B., 2009. Kinetic controls on the complexation between mercury and dissolved organic matter in a contaminated environment. Environ. Sci. Technol. 43 (22), 8548e8553. Pitois, A., Abrahamsen, L.G., Ivanov, P.I., Bryan, N.D., 2008. Humic acid sorption onto a quartz sand surface: a kinetic study and insight into fractionation. J. Colloidal Interface Sci. 325 (1), 93e100. Ravichandran, M., 2004. Interaction between mercury and dissolved organic matter e a review. Chemosphere 55 (3), 319e331. Ravichandran, M., Aiken, G.R., Reddy, M.M., Ryan, J.N., 1998. Enhanced dissolution of cinnabar (mercuric sulfide) by dissolved organic matter isolated from the Florida Everglades. Environ. Sci. Technol. 32 (21), 3305e3311. Schaefer, J.K., Morel, F.M.M., 2009. High methylation rates of mercury bound to cysteine by Geobacter sulfurreducens. Nat. Geosci. 2, 123e126. Feb. Skyllberg, U., 2008. Competition among thiols and inorganic sulfides and polysulfides for Hg and MeHg in wetland oils and sediments under suboxic conditions: illumination of controversies and implications for MeHg net production. J. Geophys. Res. 113, G00C03. Skyllberg, U., Westin, M.B., Meili, M., Bjo¨rn, E., 2009. Elevated concentrations of methylmercury in streams after forest clear-cut: a consequence of mobilization from soil or new methylation? Environ. Sci. Technol. 43 (22), 8535e8541. Slowey, A.J., 2010. Rate of formation and dissolution of mercury sulfide nanoparticles: the dual role of natural organic matter. Geochim. Cosmochim. Acta 74 (16), 4693e4708. Valca´rcel, M., 2000. Principles of analytical chemistry, a textbook: 2.5 Basic analytical properties. Springer-Verlag, Berlin Heidelberg. Waples, J.S., Nagy, K.L., Aiken, G.R., Ryan, J.N., 2005. Dissolution of cinnabar (HgS) in the presence of natural organic matter. Geochim. Cosmochim. Acta 69 (6), 1575e1588.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Seawater quality and microbial communities at a desalination plant marine outfall. A field study at the Israeli Mediterranean coast Dror Drami a,b, Yosef Z. Yacobi c, Noga Stambler d, Nurit Kress a,* a
Israel Oceanographic and Limnological Research, National Institute of Oceanography, P.O. Box 8030, Tel Shikmona, Haifa 31080, Israel The Porter School of Environmental Studies, Tel Aviv University, Israel c Israel Oceanographic and Limnological Research, Kinneret Limnological Laboratory, Israel d Mina & Everard Goodman Faculty of Life Sciences, Bar Ilan University, Israel b
article info
abstract
Article history:
Global desalination quadrupled in the last 15 years and the relative importance of seawater
Received 10 April 2011
desalination by reverse osmosis (SWRO) increased as well. While the technological aspects
Received in revised form
of SWRO plants are extensively described, studies on the environmental impact of brine
2 August 2011
discharge are lacking, in particular in situ marine environmental studies. The Ashqelon
Accepted 5 August 2011
SWRO plant (333,000 m3 d1 freshwater) discharges brine and backwash of the pre-
Available online 11 August 2011
treatment filters (containing ferric hydroxide coagulant) at the seashore, next to the cooling waters of a power plant. At the time of this study brine and cooling waters were
Keywords:
discharged continuously and the backwash discharge was pulsed, with a frequency
Seawater desalination
dependent on water quality at the intake. The effects of the discharges on water quality
Reverse osmosis
and neritic microbial community were identified, quantified and attributed to the different
Environmental impact
discharges. The mixed brine-cooling waters discharge increased salinity and temperature
Brine discharge
at the outfall, were positively buoyant, and dispersed at the surface up to 1340 m south of the outfall. Nutrient concentrations were higher at the outfall while phytoplankton densities were lower. Chlorophyll-a and picophytoplankton cell numbers were negatively correlated with salinity, but more significantly with temperature probably as a result of thermal pollution. The discharge of the pulsed backwash increased turbidity, suspended particulate matter and particulate iron and decreased phytoplankton growth efficiency at the outfall, effects that declined with distance from the outfall. The discharges clearly reduced primary production but we could not attribute the effect to a specific component of the discharge. Bacterial production was also affected but differently in the three surveys. The combined and possible synergistic effects of SWRO desalination along the Israeli shoreline should be taken into account when the three existing plants and additional ones are expected to produce 2 Mm3 d1 freshwater by 2020. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ972 4 8515202; fax: þ972 4 8511911. E-mail addresses:
[email protected],
[email protected] (N. Kress). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.005
5450
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
Introduction
Desalination to provide potable water is increasing worldwide. The total global production capacity increased from 17.3 to 68 million m3day1 (Mm3 d1) from 1994 to 2009 and is projected to reach 130 Mm3 d1 by 2016 (Water-world, 22ndGWI/IDA; Yermiyahu et al., 2007). The relative contribution of desalination by Reverse Osmosis (RO) of brackish and seawater (SW) stands at 44% (Greenlee et al., 2009). A combination of large plants, modern membrane technology and improved energy systems have reduced the cost of SWRO desalinated water by ca. 3 times in the last 15 years (Campbell and Jones, 2005; Service, 2006; Tal, 2006; Fritzmann et al., 2007). Therefore, the relative importance of SWRO is projected to increase globally. In Israel, three SWRO plants produce ca. 700,000 m3 d1 freshwater, 17% of the total potable water resources in the country. By 2020, SWRO is planned to supply more than 30% of Israel’s freshwater needs (Dreizin et al., 2008). The RO process uses high pressure to force water molecules through a semi-permeable membrane that retains the salts (up to a maximum of 50% conversion factor for SWRO), producing freshwater and brine. However, the process is not just a simple separation between water and salts. Chemicals are used at the various stages of the desalination process and may include: a) coagulants in the pre-treatment stage (iron or aluminum salts, polymers); b) biocides (such as chlorine) and neutralizers (sodium sulfite); c) antiscalants to prevent fouling of the membranes (polyphosphates, polyphosphonates, polyacrylic acid, polymaleic acid); d) cleaning solutions for RO membranes (acidic and alkaline solutions and detergents); and e) pH and hardness adjustors for the product water (lime) (NRC, 2008; UNEP, 2008). These chemicals are disposed off mostly with the brine. Until recently, the extensive literature on desalination did not address the environmental impacts associated with the process but focused on plant planning, site selection and construction, operational aspects, technological improvements, and energetic cost. It was assumed that when properly engineered and constructed, brine discharge is environmentally safe (Ahmed et al., 2001; Campbell and Jones, 2005; Alpert et al., 2007; Lattemann and Hopner, 2008; Safrai and Zask, 2008; Shannon et al., 2008). Environmental impacts are now addressed in the literature but mostly as a theoretical analysis that include entrainment and impingement of organisms at the intake; and at the outfall, increased salinity and stratification, reduced vertical mixing, decrease in oxygen concentration, increased turbidity, euthrophication, decreased or increased production, toxicity, mitigation techniques (UNEP/MAP/MEDPOL, 2003; Fritzmann et al., 2007; NRC, 2008; UNEP, 2008; Khan et al., 2009; Lattemann, 2010). A large proportion of the published work is descriptive and provides little quantitative data. The number of published articles with actual measurements of effects in situ or in lab experiments is small and limited in scope (Roberts et al., 2010). Most of the publications emphasize the effects of salinity on the benthic communities and those are site and organism specific (Ferna´ndez-Torquemada et al., 2005; Raventos et al., 2006; Gacia et al., 2007; Sa´nchezLizaso et al., 2008).
In this study we aim to show the effects of the discharges of Ashqelon’s (Israel) SWRO plant on seawater quality and on the marine microbial community at the discharge site. To the best of our knowledge this is the first paper to present the effects of desalination effluents on the neritic marine environment. The motivation of this research was an unexpected red plume observed at the outfall site, attributed to the pulsed discharge of ferric hydroxide used as a coagulant in the pre-treatment stage. This notable esthetic effect motivated the regulator (Safrai and Zask, 2008) to initiate this research despite claims that iron discharge has no ecological consequences.
2.
Study site
The Ashqelon SWRO plant is located at the southern Mediterranean coast of Israel (Fig. 1). It started to operate in 2005 and now produces (2011) 330,000 m3 d1 of freshwater, one of the largest SWRO plants in the world. The detailed planning and operation of the plant were described previously (Kronenberg, 2004; Sauvet-Goichon, 2007) as were some of the possible adverse environmental aspects (Einav et al., 2002; Einav and Lokiec, 2003; Safrai and Zask, 2008). At the time of this study (2008e2009) the average seawater intake, conversion factor and salinity were: 737,000 m3 d1, 43.5%. and 75.3, respectively. The brine (18,000 m3 h1), containing chemicals used in the process, is discharged at the shoreline, next to four discharge points of cooling waters (308,000 m3 h1) of a power plant adjacent to the SWRO plant (Fig. 1). The water temperature at the discharge point was higher by up to 10 C compared to the temperature at the intake (Glazer, 2009, 2010b). At the time of this study, backwash of the sand filters, containing iron hydroxide used as coagulant in the pre-treatment step, was discharged in pulses (up to 6500 m3 h1), with a frequency dependent on the seawater quality at the intake. As a result, a “red plume” formed during the pulsed discharges (Safrai and Zask, 2008; UNEP, 2008). In 2007, 535 ton of iron was discharged at the outfall, decreasing to 270 and 175 ton in 2008 and 2009, respectively (I. Safrai, personal communication). The concentration of iron in the brine ranged from 200 to 1100 mg L1 in 2008e2009 (Glazer, 2010b). Since May 2010, the backwash is mixed and discharged continuously with the brine, in an attempt to mitigate the esthetics effect of the “red plume”. Additional chemicals used in the process and discharged with the brine during 2009 were: polyphosphonate anti scalant (34 t as P), HCl (15 t) and NaOH (20 t) for cleaning the RO membranes neutralized to NaCl, Sodium bisulfite (NaHSO3 e 70 t) used for membrane preservation and wash water of the CaCO3 reactor used for alkalinity and pH adjustment of the desalinated water (I. Safrai, personal communication). It should be noted that brine from well amelioration RO treatment is discharged with the SWRO brine. At the time of this study the discharge rate was 130 m3 h1 and the concentration of nitrate and silicic acid high (47 mg L1 NO3eN and 62 mg L1 Si(OH)4eSi) (I. Safrai, Ministry of Environmental Protection (MEP), personal communication).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
5451
Fig. 1 e Location of Ashqelon at the Mediterranean coast of Israel and schematic map of the study site. The Ashqelon seawater desalination plant (A) is adjacent to a power plant (B). The brines are discharged at the shoreline through one outfall (2) that is located next to cooling waters discharges (1). Line 3 represents the coal unloading quay. Sampling was performed along the broken line 4 (see Table 1) and at a background station (W), 650 m from the outfall. The water intake for the desalination plant (SWRO) and the cooling waters (CW) are also depicted.
3.
Methods
3.1.
Sampling rationale
This study took place while the backwash discharge was pulsed creating a transient signal (the iron containing backwash) on top of a constant influence (brine from the SWRO and well amelioration plants and cooling waters from the power plant). Therefore, the sampling was designed to differentiate among these environmental signals, looking at time-dependent gradients from backwash discharge and distance-dependent gradients relative to the outfall. In the field, the backwash discharge was easily distinguished by the red tinge of the seawater that served as a marker of the dispersion path. This was confirmed by the temperature and salinity of the samples. Seawater samples were collected near the outfall before and during the backwash discharge (Stations O and OR, respectively), during dispersion (Stations DX, where X is the sampling order and increased with time from the backwash discharge or with distance from the outfall) and at a background station located west to the outfall (W). The sampling scheme differentiated between samples under brine and cooling water influence only (Stations O), under brine, cooling waters and backwash discharge influence (OR and DX) and background (W) (Table 1). The best sampling scheme was executed at the end of winter when the same geographical position was occupied, before, during and at the wake of the backwash discharge (Owi, ORwi, D1wi).
3.2.
Sampling
Three surveys were conducted on board the R/V “Dolphin” from the Marine and Coastal Environment Division (MCED) of
the MEP. The surveys took place in Spring (sp)-April 15th, 2008, Summer (su)-August 19th, 2008, and April 2nd, 2009. Although the last survey took place at the beginning of April, we considered it as end of winter (wi) because from mid February to the end of March 2009 winter storms delayed the establishment of spring conditions. Sampling stations are described in Table 1 and Fig. 1. In situ depth profiles of temperature, salinity, turbidity and dissolved oxygen were measured with a Yellow Spring Instruments YSI 6000 probe. Seawater for the analysis of nutrients, particulate matter, chlorophyll-a (chl-a), and microscopic community (picophytoplankton cell numbers, primary and bacterial production) were collected with a FLOJET pump, whose inlet was attached to the YSI. Five replicate seawater samples for nutrient determinations were collected into 15 ml acid-washed plastic scintillation vials, brought refrigerated to the laboratory (within 4 h from collection) were they were immediately frozen and kept frozen until laboratory analysis. Samples for chl-a and suspended particulate matter (SPM) determination were collected into plastic bottles, kept refrigerated and in the dark and brought to the laboratory. At the same day, samples for chl-a determination were filtered through GF/F filters that were folded, wrapped in aluminum paper, and frozen. Filters were kept frozen for 1 week until laboratory analyses. Samples for SPM determination were filtered into pre-weighted 0.45 mm Nuclepore membrane filters, washed with MilliQ water and frozen. Duplicate samples of 1.8 cm3 for picophytoplankton enumeration were fixed immediately upon collection with 20 mL of 25% glutaraldehyde and transported to the laboratory in the dark. There, the samples were immediately frozen in liquid nitrogen and stored at 80 C until analysis within 2 months. Samples for the measurement of primary and bacterial production were collected into plastic bottles, kept
5452
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
Table 1 e Station’s position and variables measured in discrete water samples in spring (sp, 15 April 2008), summer (su, 15 August 2008) and at the end of winter (wi, 2 April 2009). Station names with d as a suffix indicate samples collected 0.5 m above the bottom, and the sampling depth is indicated by *. All other samples were collected at the surface. Background stations are given in bold. Station Distance Temp Salinity (*sampling depth)a m Osp ORsp D1sp D2sp D2dsp D3sp D3dsp Wsp Wdsp O(1)su ORsu D1su D2su D2dsu D3su D3dsu D4su O(2)su Wsu O(2)wi Owi ORwi D1wi D2wi D3wi D4wi D4dwi D5wi D5dwi Wwi Wdwi
230 230 410 550 3.9* 1000 4.2* 650 w 5.6* 310 300 190 400 4.1* 590 2.4* 1340 240 650 w 330 280 280 280 410 630 1340 2.7* 1810 2.8* 650 w 5.3*
C
23.53 24.47 25.95 27.6 20.18 24.81 20.43 23.43 19.89 37.01 34.19 34.86 35.26 29.9 33.7 29.48 33.39 31.26 30.96 22.26 26.35 26.23 26.34 26.2 26.12 19.01 19.03 19.15 19.12 22.11 18.83
40.11 40.23 40.55 40.51 39.91 40.33 39.92 40.13 39.91 40.38 40.5 40.93 40.46 39.5 40.11 39.35 39.84 39.58 39.56 39.72 41.14 40.84 41.16 40.91 40.36 39.77 39.77 39.77 39.76 40.01 39.77
SPM
NO PO3 3 4 Si(OH)4 þ NO 2
mg L1
mM
3.05 5.98 7.82 2.71 2.9 1.56 3.76 3.06 6.15 7.85 10.9 14.8 13.2 6.73 8.13 10.97 7.74 6.84 5.93 3.09 3.63 28.8 3.67 4.39 5.48 4.40 4.6 4.43 6.27 4.90 4.63
0.68 0.59 1.77 0.96 0.08 0.33 0.08 0.31 0.08 1.76 2.21 2.75 1.82 0.35 1.33 0.08 0.34 1.14 0.29 0.14 5.75 4.61 2.18 1.92 1.22 0.06 0.14 0.13 0.07 0.97 0.07
0.08 0.06 0.09 0.09 0.05 0.05 0.07 0.06 0.07 0.11 0.11 0.10 0.13 0.12 0.13 0.09 0.04 0.10 0.10 0.09 0.16 0.13 0.14 0.14 0.14 0.09 0.08 0.08 0.09 0.10 0.08
3.47 2.98 3.71 3.11 2.77 3.02 2.77 2.84 2.61 3.72 3.84 4.15 3.81 2.97 3.52 2.61 2.5 2.57 2.65 0.33 4.20 2.20 0.89 0.89 0.72 0.12 0.17 0.20 0.21 0.64 0.17
SPM-suspended particulate matter. a South south west of the outfall. Distance of the W stations are toward the west.
refrigerated and in the dark, brought to the laboratory in 4 h and then processed immediately.
3.3.
Laboratory analysis
Nutrients (nitrate þ nitrite, phosphate and silicic acid) were determined using a segmented flow SANplus SYSTEM from Skalar (Kress and Herut, 2001). Samples were thawed at the day of analysis. The precision of nitrate þ nitrite, phosphate and silicic acid measurements was 0.02, 0.003 and 0.06 mM, respectively. The limit of detection (2 times the standard deviation of the blank) was 0.075 mM for nitrate þ nitrite, 0.008 mM for phosphate and 0.03 mM for silicic acid. The SPM filters were dried by lyophylization and re-weighed and the concentration calculated from the difference and the volume
filtered. The filters were digested with a mixture of hydrofluoric acid and aqua-regia (ASTM, 1983). The concentrations of Fe and Al in the digest were determined by Atomic Absorption Spectrometry using a Varian SpectrAA 220 FS spectrometer. Quality control and quality assurance of the results were performed with standard reference materials (NIST e Estuarine Sediment 1646 and NRCC MESS-3) that were digested and analyzed in the same manner as the samples. Chl-a was determined fluorometrically after overnight extraction with 95% acetone, in the dark at 4 C (Holm-Hansen et al., 1965). The pigment extract was subsequently measured following acidification by 1 N HCl to estimate chla degradation products. Picophytoplankton was enumerated by flow cytometry. Prior to the analysis, samples were fast thawed at 37 C, and excited in a flow cytometer e FACScan Becton Dickinson, fitted with an Argon laser (488 nm). 0.93 mm beads (Polysciences) served as standards (Marie et al., 2005; Stambler, 2006). Taxonomic discrimination was based on: cell side-scatter e a proxy of cell volume; forward scatter e a proxy of cell size; and orange and red fluorescence of phycoerythrin and of chl-a (585 nm and 630 nm, respectively). Primary production (PP) was measured with the 14C technique (Steemann-Nielsen, 1952) in duplicated 50 cm3 subsamples. A spike of approximately 3*105 Bq of [14C] bicarbonate was added to each bottle. The bottles were incubated in the laboratory, at 20 C and provided with light flux of 60 mmol photon m2 s1. After incubation, of approximately 3 h, the samples were filtered onto poly-acetate 25 mm 0.45 mm membrane filters under light vacuum (about 100 mg Hg), rinsed with filtered seawater and left overnight in the presence of HCl vapor to eliminate any remaining traces of inorganic 14C. Control samples poisoned by Lugol’s solution at zero time were run in each experimental series to compensate for non-biological absorption to filters. The total added 14C was checked for each sampling series by counting 0.1 cm3 portions withdrawn directly from each of the incubated bottles. Total radioactivity in the particulate fraction retained on the filters was determined by liquid scintillation with quench correction. The average difference between duplicates was w9%. Assimilation number (AN) was calculated by normalizing PP with chla concentration. AN is an important ecological indicator of growth efficiency and/or shift in phytoplankton composition (Karl et al., 1995). Bacterial production (BPP) was determined by the modified (Smith and Azam, 1993) leucine uptake method (Kirchman et al., 1985; Simon and Azam, 1989). Zero time controls were run for all samples. Leucine uptake was converted to carbon uptake using the conversion factors of Simon and Azam (1989) with an isotope dilution factor of 2. Regression analysis was performed with the Addinsoft XLStat software under the assumption of 95% confidence level.
4.
Results
The brine and cooling waters mixed at the nearshore, following their discharge at the shoreline. The pulsed backwash discharge tinted the water red and showed that the discharges dispersed toward the southesouth west during the three surveys. The mixing was very heterogeneous close to the outfall, as visualized by patches of water with different
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
coloration. The frequency of the discharge in the spring and at the end of winter was every 40e60 min, while in the summer the frequency was every 20e30 min. The higher backwash discharge frequency in the summer probably caused the area to be chronically affected by it as will be shown hereafter.
4.1.
Temperature and salinity
Temperature was seasonally dependent, as expected (Table 1, Fig. 2a). At the background stations, surficial temperatures were 23.43, 30.96 and 19.01 C in sp, su and wi, respectively and decreased with increased water depth to 19.89, 29.93 and 18.83 C in sp, su and wi, respectively. The affected stations in all surveys were warmer than the background. In spring and summer all stations were stratified. In spring, maximum stratification was at D2sp (DT ¼ 7.3 C) and minimal at Wsp
Fig. 2 e Representative depth profiles of temperature (A) and salinity (B) during the three surveys: spring (squares), summer (triangles) and end of winter (circles). Background stations are drawn with solid lines and filled symbols and affected stations with dashed lines and open symbols. The water column was stratified in the spring and the summer, and the temperature and salinity at the upper layers were higher at the affected stations. At the end of winter the water column was mixed and the affected station more saline and warmer than background. The dispersion stations (exemplified by D3wi) were stratified.
5453
(DT ¼ 3.2 C). In the summer, maximum stratification was at D2su (DT ¼ 4.8 C). The warmest station was O(1)su (37 C) while the western station, Wsu, was the coldest (30.96 C) and mixed down to 3 m. At the end of winter most of the stations were mixed. There was a clear separation between the colder background (D4wi and D5wi with 19.5 C) and the affected stations (26.2 C). The dispersion station D3wi was highly stratified (DT ¼ 6.7 C). The western station, Wwi, that served as background in the previous surveys, was stratified as well indicating influence of the discharges down to 2 m depth, where temperature reached background. Salinity was independent of sampling season, ranging from 39.32 to 41.54 (39.9e40.64, 39.32e40.93 and 39.55e41.54, in sp, su and wi, respectively) (Table 1, Fig. 2b). The water column was stratified with respect to salinity, similarly to the thermal stratification. There was a significant positive increasing correlation between temperature and salinity in spring and summer (Fig. 3). The correlation at the end of the winter is similar to that of the spring except for stations Owi, ORwi, D2wi where salinity varied while temperature remained essentially constant. Although it is assumed that the coefficient of turbulent (eddy) mixing for temperature and salinity are equal, here, because the brine and cooling waters are discharged from different outfalls with different discharge rates (cooling waters ca. 10 times more than brine), there may be an initial different stratification of the discharges that make the mixing of the warm waters with coastal waters different from that of the brine. Heterogeneity of the dispersal can be shown by the differences in salinity and temperature at outfall stations O(2)su/O(1)su and Owi/O(2)wi located 50 m apart (Table 1). During the three surveys the discharge was positively buoyant and dispersed close to the surface, traced up to 1340 m south of the outfall. Similar dispersion patterns were
Fig. 3 e Temperature versus salinity during the three surveys: spring (squares), summer (triangles) and end of winter (circles). The more saline and warmer waters were found closer to the outfall. Linear regressions were significant for spring and summer (R2 [ 0.902 and 0.804, respectively, p < 0.0001). At the end of winter some of the points were similar to the spring while close to the outfall salinity varied at constant temperature, probably due to non-homogeneous mixing.
5454
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
documented during bi-annual monitoring studies performed by the Israel Electric Corporation (IEC) at the discharge site (Glazer, 2009, 2010b). The maximal differences in temperature between affected and background stations were similar at the three surveys: 7.7, 7.8 and 7.3 C in sp, su and wi, respectively. However, at the surface only, the maximal difference was measured at the end of winter (4.2, 6.0 and 7.3 C in sp, su and wi, respectively). The differences in salinity between the affected stations and background were 0.74, 1.61 and 1.84 in sp, su and wi, respectively, and at the surface 0.38, 0.90 and 0.79 in sp, su and wi, respectively. The IEC monitoring in 2008e2009 found similar differences in temperature (7.4 C and 6.2 C in spring and winter 2008) and in salinity (2.08 and 1.89 in spring and winter 2008) between the stations at the vicinity of the outfall and the reference station located ca. 2.5 km south of the outfall.
4.2.
Turbidity and suspended particulate matter (SPM)
Seawater turbidity without the backwash discharge was similar at all stations at the three surveys and ranged from <0.1 to 1.5 NTU. Backwash discharge increased water turbidity rapidly following its discharge as exemplified by the time series measurement at the end of winter (Fig. 4). At the three surveys, turbidity returned to background values at ca. 500 m from the outfall or after 20 min from the discharge at the outfall. SPM ranged from 1.56 to 7.82 mg L1 in spring, from 5.93 to 14.8 mg L1 in summer and from 3.09 to 6.27 mg L1 at the end of winter, with a very high concentration, 28.8 mg L1, measured at station ORwi (Table 1). The concentrations measured at the deep samples at the same station were lower than the surficial samples. In the summer, SPM concentrations were the highest among the surveys, probably caused by the high frequency of the backwash discharge. In the spring and summer the SPM concentration doubled immediately following the backwash discharge (stations ORsp and ORsu) and increased 8 times at the end of winter.
Fig. 4 e Seawater turbidity at the outfall increased steeply following the backwash discharge and returned to background values within 20 min. The horizontal line represents the time the discharge reached the outfall station.
As expected, turbidity increased with increased SPM concentration. During spring and end of winter the relations were similar while in the summer more SPM was present at the same turbidity and the data more scattered.
4.3.
Dissolved oxygen and nutrients
The seawater was saturated or close to saturation with dissolved oxygen during the three surveys, and not influenced by the discharges. Therefore, dissolved oxygen concentrations will not be further discussed. The background concentrations of the nutrients (nitrate þ nitrite, phosphate, silicic acid) were similar during the three surveys, except for silicic acid, that was about 2 times lower at the end of winter. The concentrations measured at the deep samples at the same station were similar or lower than the surficial samples (Table 1). Nutrient concentrations were higher at the outfall prior and after the backwash discharge (O and OR) and at dispersion stations, compared to background, more noticeable in the nitrate þ nitrite concentrations. While in spring and summer the concentrations were higher at the affected stations, and similar to background in the others, at the end of the winter concentrations decreased as a function of distance from the outfall and time from the discharge.
4.4.
Particulate iron (Fe)
Iron in particulate matter ranged from to 35e1611 mg L1, similar in the three surveys (Fig. 5). As Fe is supplied both with the coagulant and by natural processes we normalized the concentrations to aluminum (Al) concentrations as done for the sediments in the area (Goldsmith et al., 2001). The relationship between Fe and Al was similar during the three
Fig. 5 e Iron versus aluminum concentration in suspended particulate matter during the three surveys (spring (squares), summer (triangles) and end of winter (circles)) shows six samples with higher than natural iron concentration. Those were found at the stations at the outfall with backwash presence and at the first dispersion station. Regression line solid (R2 [ 0.751, p < 0.0001), 95% confidence and prediction intervals in dotted and dashed lines, respectively.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
5455
surveys, indicating a common source and no seasonal variation. Fe concentrations were higher than natural at stations ORsp, D1sp, D2sp, ORsu, D1su and ORwi, namely, the outfall stations with the backwash discharge and the first and second dispersion stations (Fig. 5, Table 1).
4.5.
Chlorophyll-a and picophytoplankton cells
Chl-a concentrations during the three surveys were similar and ranged from 0.26 to 0.94 mg m3 in spring, from 0.42 to 0.74 mg m3 in summer and from 0.23 to 1.03 mg m3 at the end of winter (Supplementary material 1). In the spring and in particular at the end of winter there was a clear separation into two groups of stations: the affected stations group with low chl-a and higher temperature and salinity and the background group, with high chl-a and lower temperature and salinity. The average chl-a concentrations at the affected, low chl-a group were 0.32 0.08 and 0.29 0.06 mg m3 in spring and end of winter, respectively. These concentrations were significantly lower (more than 60%) than the background average concentrations of 0.80 0.11, and 0.89 0.04 mg m3 in spring and end of winter, respectively. It should be noted that we did not find detectable concentrations of degraded chl-a (data not shown), indicating that the phytoplankton cells at the discharge site were probably intact. In the summer, the concentrations were more dispersed and the concentration at the affected stations (0.45 0.03 mg m3) 32% lower than background (0.66 0.06 mg m3). There was a significant linearly decreasing correlation between chl-a and temperature at each survey, and also between chl-a and salinity, but with lower correlation coefficients than with temperature (Fig. 6). The total number of picophytoplankton cells in spring (1.83e10.2 107 cells L1) and summer (3.74e15.5 107 cells L1) were similar (Supplementary material 1) as well as the relative contribution of the groups (Fig. 7). Synechococcus comprised of an average 86 and 93% of the total cells in spring and summer, respectively. At the end of winter, the number of cells was 10 times lower (1.15e8.81 106 cells L1) with a different relative contribution: pico-eukaryotes dominated (50% of the total cells), Synechococcus and Prochlorococcus contributed 18% each to the total and an unexpected 4th group comprised of 13% of the total cells. This 4th group belonged probably to the picoeukaryotes, slightly larger than the “classical” group. During each survey the total cell numbers were the lowest at the affected stations lower by 3.8, 2 and 6 times from background in sp, su and wi, respectively. In the three surveys, cell numbers decreased with increased temperature and salinity and were positively correlated to chl-a concentration (Fig. 8). At the end of winter the stations could be grouped into two: higher and lower cell numbers, similar to chl-a. Moreover, even though the cell numbers were 10 times lower at the end of winter the chla concentration was similar to the other surveys. That could be due to photo-acclimation of the cells, increasing chl-a per cell and/or the presence of cells larger than 10 mm, not enumerated by flow cytometry. The latter is indicated by the large contribution of the pico-eukaryotes and the low concentration of Si(OH)4 in seawater at the end of winter
Fig. 6 e Chlorophyll-a concentrations decreased with increased salinity (A) (R2 [ 0.497, 0.694 and 0.666 in sp, su and wi, respectively) and more significantly with increased temperature (B) ( R2 [ 0.594, 0.794 and 0.718 in sp, su and wi, respectively) during the three surveys.
suggesting utilization by larger cells, possible diatoms. The two pico-eukaryotes groups contributed up to 97% to the total biomass during the winter (calculation not shown).
4.6. Primary production (PP) and bacterial production (BPP) rates The average primary production rates (PP) at the non affected stations during the three surveys were similar: 4.41 1.4, 4.79 0.85 and 5.53 0.37 mgC m3 h1 the sp, su and wi, respectively. PP was significantly lower at the affected stations: 0.60 0.56, 1.84 0.67 and 1.43 0.31 mgC m3 h1 in sp, su and wi, respectively. During each survey, PP was negatively correlated with temperature, similar in the spring and at the end of winter and slightly different in the summer. PP was also negatively correlated with salinity; however, the correlation was significant only when all data were pooled. PP was positively correlated with chl-a concentrations, showing a fairly uniform relationship in all surveys (Fig. 9). The ranges of assimilation number (AN ¼ PP/Chl-a) in the spring and at end of winter were similar (3.19e6.53 (median
5456
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
100
Percentage
80 60 40 20
O O sp R D sp 1 D sp D 2sp 2d D sp D 3sp 3d s Wp W sp O dsp (1 O )su Rs D u 1 D su D 2su 2d D su D 3su 3d D su D 4su 4d s Wu W su O dsu (2 O )su (2 )w O i O wi Rw D i 1 D wi 2 D wi 3 D wi D 4wi 4d D wi D 5wi 5d w W i W wi dw i
0
Station Syn
Pro
Pico-euk
4th
Fig. 7 e Relative picophytoplankton distribution at the different stations during the three surveys show that Synechococcus dominated in spring and summer while at the end of winter the pico-eukaryotes constituted 50% of the total cells, Synechococcus and Prochlorococcus contributed 18% each to the total and an unexpected 4th group comprised of 13% of the total cells. In addition, the relative distribution at the affected stations was different from background at the end of winter.
4.88) and 2.19e5.69 (median 4.97)) mgC (mg Chl)1 h1. AN was lower in the summer (1.07e3.77 (median 2.46) with a high value of 6.79 mgC (mg Chl)1 h1 at station D2dsu), showing that the area in the summer was chronically affected probably due to the high frequency of the backwash discharge. AN was negatively correlated to particulate Fe concentration, indicating that the backwash caused a decrease in phytoplankton growth efficiency (Fig. 10). The average rates of bacterial production (BPP) at the non affected stations were similar in summer and at the end of winter (0.58 0.25 and 0.46 0.05 mgC m3 h1, respectively) and lower in the spring (0.19 0.10 mgC m3 h1). The rates were much lower at the affected stations in spring and end of winter (0.03 0.03 mgC m3 h1 during both surveys) compared to background (Supplementary material 1). However, in the summer BPP at the affected stations was higher than background with an average rate of
Fig. 8 e Chlorophyll-a increased with picophytoplankton cells numbers during the three surveys (R2 [ 0.647, 0.422 and 0.809 in sp, su and wi, respectively) despite the 10 times lower number of cells at the end of winter.
1.64 0.73 mgC m3 h1. BPP decreased at surface at the dispersion stations D2su and D3su (0.79 mgC m3 h1) and at the western station Wsu. The relationship between BPP and temperature and salinity was inconsistent, being positive in the summer, and negative in the spring and at the end of winter (Fig. 11).
5.
Discussion
5.1.
Cause e effect identification and quantification
Seawater at the discharge site was warmer and more saline than the background due to the constant discharge of brine and cooling waters, while the pulsed backwash discharge added an additional pressure to the already impacted environment. The mixed discharge was positively buoyant, and dispersed at the surface, mainly toward the south. This is in
Fig. 9 e Primary production increased with increasing chlorophyll-a similarly during the three surveys (R2 [ 0.820, p < 0.0001).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
Fig. 10 e The assimilation number (a proxy of phytoplankton growth efficiency) decreased with increased particulate iron, a marker of the backwash presence, during the three surveys (R2 [ 0.585, p < 0.0001. Two extreme points in the summer were neglected).
Fig. 11 e Bacterial production increased with decreased salinity in spring and at the end of winter (R2 [ 0.558) and was not correlated to salinity in the summer due to the high rates measured at the stations located close to the outfall (A). Bacterial production increased with increased temperature in the summer (R2 [ 0.519), decreased with temperature at the end of winter (R2 [ 0.838) and was not correlated to temperature in the spring (B).
5457
agreement with the findings of monitoring studies at the site (Glazer, 2009, 2010b) but in contrast to the dominant northward current direction at the inner shelf (26 m depth) region (Rosentraub and Brenner, 2007). The difference is probably due to the localized conditions at the outfall, with high cooling water discharge rate and the presence of the breakwater north to the outfalls (Fig. 1). As the brine dispersed at the surface we will concentrate the discussion on the upper layer and identify and quantify the effects based on the background values. To do so, we redefined the background stations as those with minimal temperature and salinity at each survey. In the spring and summer the background stations were the western ones as planned (Wps and Wsu, Table 1). At the end of winter, the western station (Wwi) was affected by the discharges therefore the southernmost station (D5wi) was used as the background (Table 1). In addition, we calculated the fractional change of a parameter (F ¼ value at a sampling point divided by the background value), essentially an enrichment (F > 1) or depletion factor (0 < F < 1) from background as a mean to quantify and compare among the effects (Supplementary material 2). The effects were classified into three groups: 1) Possibly attributed to brine and cooling waters discharge; 2) Possibly attributed to backwash discharge, and 3) Inconclusive. In the first group, brine and cooling waters affected salinity, temperature, nutrients, chl-a, and picophytoplankton cell numbers. The salinity and temperature increases were obviously attributed to brine and cooling waters, respectively. The maximal enrichment factor of temperature (F ¼ 1.39) was larger than for salinity (F ¼ 1.04) probably due to the 10 times larger discharge of cooling waters. This may also explain why temperature was a more accurate proxy of the dispersion in this study. The enrichment factors for both the temperature and salinity were larger at the end of the winter than at the other two surveys. Similar enrichment factors were also calculated with the IEC monitoring data (Glazer, 2009, 2010b) showing that background conditions were reached during this study. For example, in the spring of 2008 the maximal salinity measured was 41.50 compared to the background of 38.42 (F ¼ 1.05) and the maximal temperature was 30.4 C compared to the background of 22.3 C (F ¼ 1.32). Numerical modeling of the dispersion run prior to the operation of the plant simulated a larger effect of the brine, with 49 salinity at 100 m from the shore (Sladkevich and Kit, 2004). Lower salinities were measured in this study, however the stations were located at least 250 m from the discharge point. Nutrient concentrations were higher at the outfall and decreased with increased distance. As for salinity and temperature, the effect was more evident at the end of winter, with the better sampling scheme. Maximal enrichment factors for nitrate and silicic acid were F ¼ 96 and 35, respectively, at the outfall station (Owi) and decreased with increased time from the discharge and distance from the outfall (Supplementary material 2). Maximal enrichment factor in spring and summer occurred at the D1 dispersion stations, with F ¼ 5.7 and 9.4 for nitrate in the spring and the summer, respectively and F ¼ 1.31 and 1.6 for silicic acid. Nitrate and silicic acid were introduced with the brines but they originate mostly from the well amelioration plant and not at the desalination plant. Therefore, based on the discharge rate, the pulsed backwash would dilute the
5458
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
concentrations and decrease them by 20%. And indeed, the enrichment factor for nitrate decreased by 20% at ORwi, following the backwash discharge, but more (48%) for silicic acid. Based on the concentration of nitrate in the well amelioration brines (47 mg L1) and the discharge rates we would expect the concentration at the outfall to be 15.5 mM if mixing only with the SWRO brines or 1.45 mM if mixing with the brine and cooling waters. However, the maximal concentration detected was 5.75 mM at the end of the winter (Table 1). That shows that at our sampling point, ca. 250 m from the brine discharge point, the brines were mixed with just one fifth of the cooling waters. As the cooling waters are discharged through 4 separate outfalls this is reasonable, although mixing was lower than expected. Chl-a in general was lower at the affected stations during the three surveys, with maximal depletion factors of 0.59, 0.71 and 0.26 in sp, su and wi, respectively (Supplementary material 1 and 2). The phytoplankton near the outfall did not benefit from the increased concentrations of nutrients and the chl-a and PP rates were lower than in the unaffected, nutrient poor, stations. The effect was larger at the end of the winter, followed by the summer, while in the spring the effect was more variable. The concentrations decreased with increasing salinity but more significantly with increasing temperature (Fig. 6). There were two exceptions. In the spring, chla concentration decreased only after the backwash discharge and in the winter, the concentration increased following the backwash discharge and decreased drastically at the start of the dispersion (Supplementary material 1). However, thermal pollution is known to cause a decrease in chl-a concentrations and change primary production and community structure (Langford, 1990; Langford and John, 2001). The effects may vary and are season-dependent (Choi et al., 2002; Chuang et al., 2009). Therefore, we hypothesize that the decrease in chl-a in this study was a result of increased water temperature and not salinity. This hypothesis is supported by the results of monitoring studies at the brine only outfall of the SWRO plant “Via Maris” off Palmahim (Israel, Fig. 1), located at ca. 10 m depth and 1 km from the coast which showed no effect on chl-a concentrations (Kress et al., 2010). Picophytoplankton cell numbers were lower at affected stations and increased with decreasing temperature and salinity. Moreover, the depletion factors were larger than for chl-a, which represents all components of the phytoplankton, with F ¼ 0.27, 0.49 and 0.13 in the spring, summer and end of winter, respectively (Supplementary material 2). The relative distribution of the 4 picoplankton groups in spring and summer were similar at all stations. This may be due to the fact that more than 85% were Synechococcus and small changes may have been undetectable. At the end of winter, when the relative distribution of the groups was more balanced, the affected stations had a lower contribution of pico-eukaryotes and a higher contribution of Synechococcus and Prochlorococcus cell numbers than at the background stations. Temperature affects size distribution and metabolic rates of phytoplankton (Finkel et al., 2010) therefore we assume that was the case in this study as well. In the second group, discharge of the pulsed backwash increased turbidity, SPM, particulate Fe and reduced AN. Turbidity and SPM concentrations were similar to the
background at the outfall (O) stations and increased following the backwash discharge. Maximal enrichment factors were found in the summer, followed by the spring and the lowest found at the end of winter, except for a very high enrichment factor at station ORwi (Supplementary material 2). The maximal enrichment factors were not found at the same surveys as the first group and may be connected to the frequency of the backwash discharge. During the summer, the backwash was discharged every 20e30 min, probably causing a constant influence. During the spring and at the end of winter, the frequency was lower (every 40e60 min), while in the winter the desalination plant was working only at half capacity. Particulate Fe was highest at the backwash affected stations, with no gradual decrease with distance from the outfall or time from backwash discharge. AN was negatively correlated with particulate Fe concentration in seawater (Fig. 10), a proxy for coagulant presence. It is reasonable to assume that the presence of the coagulant decreased the growth efficiency of the phytoplankton. As with turbidity and SPM, the depletion factors were lower in the summer and spring and less affected at the end of winter. In the third group, the rates of primary and bacterial production could not be definitely attributed to one of discharges. PP seemed to follow chl-a and picophytoplankton population, and it may be that the decrease could be also attributed to thermal pollution. The most affected survey was the end of winter, with maximal depletion factors of 0.21 (Supplementary material 2). The correlation of BPP with temperature and salinity varied: negative in the spring and at the end of winter and positive in the summer. Maximal enrichment factor in the summer was 11.9 while depletion was much higher at the end of winter than in the spring (down to 0.02, Supplementary material 2). The highest rates were measured in the summer in the vicinity of the outfall (Supplementary material 1), while at the other stations the rates were similar to those found in the other surveys at similar salinities (Fig. 11).
5.2. Comparison to other areas along the coast and to environmental quality guidelines The area of study is located at the shoreline of the ultraoligotrophic Eastern Mediterranean that has exceptionally low nutrient concentrations, chl-a, PP and BPP and the phytoplankton community is dominated by the pico and nano fractions (Krom et al., 1991; Yacobi et al., 1995; Zohary and Robarts, 1998; Kress and Herut, 2001; Pitta et al., 2005; Psarra et al., 2005; Yogev et al., 2011). However, the neritic waters are characterized by somewhat higher nutrient and chla concentrations, higher PP and a greater abundance of larger size phytoplankton (Azov, 1986; Berman et al., 1986; Kimor et al., 1987; Gitelson et al., 1996; Herut et al., 2000), and particularly in estuaries of coastal streams (Herut et al. 2010) and at locations influenced by anthropogenic sources. Our study was performed in a nearshore location, and by definition modified by anthropogenic activity, and its characteristics manifest that fact. For example, surficial nitrate and phosphate increased from <0.03 to <0.008 mM, respectively at the open sea (Krom et al., 2005), through 0.02 and 0.03 mM in the summer at 120 m bottom depth (Herut et al.,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
2000), to 0.31 and 0.05 mM at the spring background station in this study. Chl-a at the same stations increased from 0.018, 0.03 and 0.44 mg m3. A gradual decrease in chl-a was shown along a westward transect (0.5e22 km distance from the shore) from 0.36 to 0.06 mg m3 in another study, while SPM changed from 3 to 0.9 mg L1 (Gitelson et al., 1996). Primary production and bacterial production in the open sea in the spring were 0.091 and 0.043 mg C m3 h1 (Pitta et al., 2005; Psarra et al., 2005) while in this study at the background station in the spring the rates were higher: 1.68 and 0.07 mg C m3 h1 for PP and BPP, respectively. The concentration of chl-a and the PP and BPP rates at the affected stations in this study resembled the lower values found at the open sea stations. However, the lower values in this study were accompanied by high concentrations of nutrients, SPM and higher turbidity, contrary to those found in the open sea. A different comparison can be made to data collected during monitoring studies conducted along the Mediterranean coast of Israel: The National monitoring program (Herut et al., 2010) performed in the summer, monitoring at the Via Maris SWRO brine discharge site off Palmahim (Kress et al., 2010) and monitoring at the outfall of power stations performed by the Israel Electric Corp (IEC) (Glazer, 2010a,b). These studies concentrate on water quality and benthic community while the biological component of the water column is small. In general, salinity, temperature, nutrients, chl-a and particulate Fe concentrations measured during this study at the background stations represented the non polluted waters along the Mediterranean coast. Temperatures higher than background, by up to 6 C, are found at the outfall of three additional power plants located along the coast, similar to the increase found in this study. Regular monitoring performed at those plants did not show a correlation between seawater temperature and chl-a concentrations. However, at one instance (Glazer, 2010a) the sampling points were not located at the maximal temperatures, so the sampling may have missed the affected areas. Moreover, using only the data for the surficial samples collected during 2009 at the Ashqelon power plant (Glazer, 2010b), chl-a decreased with increased temperature similar to the findings of this study. Salinities measured during this study were higher than those measured along the coast, except for the area of the SWRO Via Maris plant (Kress et al., 2010) where values up to 43.5 were measured near the bottom. This plant is smaller (123,000 m3 d1) than the Ashqelon’s SWRO Plant and is not located next to a power plant so the brines are discharged through a marine outfall. Chl-a concentrations were not affected by the brine discharge and were natural for the area and seasonally dependent. Much higher concentrations of nutrients and chl-a were measured at the mouth of coastal streams (Herut et al., 2010) and at other polluted areas along the coast then at the effected stations during this study. We attempted to evaluate the environmental significance of the findings of this study based on available marine environmental quality guidelines (Bricker et al., 1999; ANZECC, 2000; MCED, 2002; Buchman, 2008). Most guidelines do not address salinity. However, increased salinity may have an adverse environmental effect, as recorded, for instance, in the case of Posedonia oceanica seagrass meadows off the Mediterranean coast of Spain (Sa´nchez-Lizaso et al., 2008). Most
5459
guidelines do not address also temperature, iron, PP and BPP as well. Chl-a is usually given an upper limit concentration to prevent eutrophication, in contrast to the decrease in concentrations detected in this study. On the other hand, guidelines for turbidity, suspended particulate matter (SPM) concentration, water color, total nitrogen and phosphorus are usually present in marine water environmental quality guidelines. In Israel, the increase in turbidity should not be higher than 10% of the seasonal average, the increase in SPM should not exceed the seasonal average by more than 10 mg L1 and the color of the water should not be affected outside the mixing zone (MCED, 2002). Turbidity exceeded the Israeli guidelines at the end of the winter and SPM exceeded the guidelines at station ORwi, but the effect of water color was more extensive and well outside the mixing zone. We think that the Israeli allowable upper limits for total nitrogen and total phosphorus, 71 and 3.2 mM, are too high for the area, and therefore used NOAA’s criteria for oligotrophic areas of 7.1 and 0.32 mM, respectively (Bricker et al., 1999) and assumed that nitrate and phosphate were about half of the total N and P (3.5 and 0.16 mM) (N. Kress, unpublished results). The ANZECC guidelines are stricter for nitrate (0.6 mM) and similar for phosphate (0.17 mM). Nitrate concentrations higher than half of NOAA’s criteria were measured at stations Owi, ORwi, while at 3 samples in the spring, 6 in the summer and 5 at the end of the winter, the concentrations were higher than the ANZECC guideline limits (Table 1). The concentrations of phosphate were lower than the guideline limits at all samples except at Owi, that was similar to the guideline. The water quality criteria for British Columbia for iron is 300 mg L1 (cited in Buchman, 2008). In this study, we measured concentrations up to 1611 mg L1, with 9 samples above 300 mg L1, mostly in the summer (Fig. 5) when the discharge frequency was the highest (every 20e30 min). In summary, even though the stations in this study were located outside the mixing zone, some of the measured parameters exceed accepted environmental guidelines, indicating possible effects at the discharge site. The infringement of the water color guideline gave the regulator the means to request improvements in brine management, because of the lack of relevant legislation (Drami, 2011). The study we present herein indicates that such a simple criterion as color, which ostensibly is only of an esthetic value, is affiliated with a change in water components that may potentially affect microbial life in the marine environment.
5.3.
Conclusions and future outlook
This research was prompted by the adverse esthetic effect of red backwash discharge on the marine environment. The first steps taken by the Ministry of Environmental Protection (MEP) were to require a reduction of iron utilization and discharge, followed by a requirement to stop the pulsed backwash discharge, mix it with the brine and discharge it continuously. Since May 2010, the backwash at the Ashqelon plant is mixed and discharged continuously with the brine. Moreover, The Hadera desalination plant, that started operating on March 2010, discharges the backwash continuously, while future SWRO desalination plants will be required to dispose of the coagulant on land. In addition, assessment of the environmental and economic implications of positioning the SW intake and brine outfall at different
5460
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
distances from the shore are now required at the planning stages of future SWRO plants. We were able to show that in addition to the esthetic effect (increased turbidity and SPM) the backwash affected the phytoplankton growth efficiency. Moreover, increased salinity, increased temperature or both together affected water quality and the microbial community at the disposal site. It is still not clear if the mentioned changes have a detectable impact beyond the immediate vicinity of the discharge location, however, the very existence of adverse environmental impact call for further specific research in situ and in controlled laboratory experiments. Marine pollution was not taken into account when the government decided to increase SWRO desalination from ca. 700,000 m3day1 freshwater today to ca. 2 Mm3day1 by 2020 (Dreizin et al., 2008). Three SWRO desalination plants are operational now at the 200 km long Mediterranean coast of Israel, two more passed the tender/biding stage while others are at the planning stage. While we approve the steps taken to reduce the environmental impact of desalination, the combined and possible synergistic effects of the plants at the Israeli shoreline as well as those at neighboring countries were not considered. While SW desalination is a viable option for freshwater production, we caution the decision makers to continue to study and understand the risks to the marine environment.
Acknowledgments We are grateful to Itzhak Kodovizky (Kodo) and Galia Pasternak of the Marine and Coastal Environmental Division of the Ministry of Environmental Protection for the cruises. We thank Yaron Gertner, Semion Kaganovsky, Benayahu Sulimani and Lora Izraelov for field and laboratory assistance. The research was funded by Israel’s Ministry of Environmental Protection (Marine and Coastal Environmental Division) and Israel’s Nature and Parks Authority. The thorough comments by three reviewers helped improve this manuscript and are greatly appreciated. This paper was written during N. Kress sabbatical at UC Santa Cruz (with A. Paytan) and MBARI (with K. Johnson).
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.005.
references
Ahmed, M., Shayya, W.H., Hoey, D., Al-Handaly, J., 2001. Brine disposal from reverse osmosis desalination plants in Oman and the United Arab Emirates. Desalination 133 (2), 135e147. Alpert, H., Borrowman, C., Haddad, B., 2007. Evaluating Environmental Impacts of Desalination in California. Center for Integrated Water Research.
. ANZECC, 2000. Australian and New Zealand Guidelines for Fresh and Marine Water Quality. http://www.mfe.govt.nz/
publications/water/anzecc-water-quality-guide-02/anzeccwater-quality-guide-02-pdfs.html. ASTM, 1983. American Society for Testing and Materials Designation -D 3683-78. Standard Test Method for Trace Elements in Coal and Coke Ash by Atomic Absorption, pp. 472e475. Azov, Y., 1986. Seasonal patterns of phytoplankton productivity and abundance in nearshore oligotrophic waters of the Levant Basin (Mediterranean). Journal of Plankton Research 8, 41e53. Berman, T., Azov, Y., Schneller, A., Walline, P.D., Townsend, A.R., 1986. Extent, transparency, and phytoplankton distribution of the neritic waters overlying the Israeli coastal shelf. Oceanologica Acta 9, 439e447. Bricker, S.B., Clement, C.G., Pirhalla, D.E., Orlando, S.P., Farrow, D. R.G., 1999. National estuarine eutrophication assessment effects of nutrient enrichment in the nation’s estuaries. NOAA, National Ocean Service, Special Projects Office and the National Centers for Coastal Ocean Science. Silver Spring, MD, 71 pp. Buchman, M.F., 2008. NOAA Screening Quick Reference Tables. NOAA OR&R Report 08e1. Office of Response and Restoration Division, National Oceanic and Atmospheric Administration, Seattle WA, 34 pp. Campbell, R.L., Jones, A.T., 2005. Appropriate disposal of effluent from coastal desalination facilities. Desalination 182 (1e3), 365e372. Choi, D.H., Park, J.S., Hwang, C.Y., Huh, S.H., Cho, B.C., 2002. Effects of thermal effluents from a power station on bacteria and heterotrophic nanoflagellates in coastal waters. Marine Ecology-Progress Series 229, 1e10. Chuang, Y.L., Yang, H.H., Lin, H.J., 2009. Effects of a thermal discharge from a nuclear power plant on phytoplankton and periphyton in subtropical coastal waters. Journal of Sea Research 61 (4), 197e205. Drami, D., 2011. Ashkelon’s Desalination Plant and the Red Brine Phenomenon: Chemo-oceanographic and Role of Scientific Knowledge on Environmental Policy Making Concerning Desalination. M.Sc Thesis. The Porter School of Environmental Studies, Tel Aviv University, Israel. Dreizin, Y., Tenne, A., Hoffman, D., 2008. Integrating large scale seawater desalination plants within Israel’s water supply system. Desalination 220 (1e3), 132e149. Einav, R., Harussi, K., Perry, D., 2002. The footprint of the desalination processes on the environment. Desalination 152 (1e3), 141e154. Einav, R., Lokiec, F., 2003. Environmental aspects of a desalination plant in Ashkelon. Desalination 156 (1e3), 79e85. Ferna´ndez-Torquemada, Y., Sa´nchez-Lizaso, J.L., Gonza´lezCorrea, J.M., 2005. Preliminary results of the monitoring of the brine discharge produced by the SWRO desalination plant of Alicante (SE Spain). Desalination 182 (1e3), 395e402. Finkel, Z.V., Beardall, J., Flynn, K.J., Quigg, A., Rees, T.A.V., Raven, J.A., 2010. Phytoplankton in a changing world: cell size and elemental stoichiometry. Journal of Plankton Research 32 (1), 119e137. Fritzmann, C., Lo¨wenberg, J., Wintgens, T., Melin, T., 2007. Stateof-the-art of reverse osmosis desalination. Desalination 216 (1e3), 1e76. Gacia, E., Invers, O., Manzanera, M., Ballesteros, E., Romero, J., 2007. Impact of the brine from a desalination plant on a shallow seagrass (Posidonia oceanica) meadow. Estuarine, Coastal and Shelf Science 72 (4), 579e590. Gitelson, A., Karnieli, A., Goldman, N., Yacobi, Y.Z., Mayo, M., 1996. Chlorophyll estimation in the Southeastern Mediterranean using CZCS images: adaptation of an algorithm and its validation. Journal of Marine Systems 9 (3e4), 283e290. Glazer, A., 2009. Monitoring the Coastal and Marine Environment at the Discharge Site of the Rutenberg Power Plant, VID
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
Desalination Plant and Mekorot’s Well Amelioration Plant, Results from 2008. Israel Electric Corp. R-ELP 21e2009. 151 pp. (In Hebrew). Glazer, A., 2010a. Monitoring the Coastal and Marine Environment at the Discharge Site of the Orot Rabin Power Plant, Results from 2009. Israel Electric Corp. 115 pp. (In Hebrew). Glazer, A., 2010b. Monitoring the Coastal and Marine Environment at the Discharge Site of the Rutenberg Power Plant, VID Desalination Plant and Mekorot’s Well Amelioration Plant, Results from 2009. Israel Electric Corp. R-ELP 04e2010. 216 pp. (In Hebrew). Goldsmith, S.L., Krom, M.D., Sandler, A., Herut, B., 2001. Spatial trends in the chemical composition of sediments on the continental shelf and slope off the Mediterranean coast of Israel. Continental Shelf Research 21 (16e17), 1879e1900. Greenlee, L.F., Lawler, D.F., Freeman, B.D., Marrot, B., Moulin, P., 2009. Reverse osmosis desalination: water sources, technology, and today’s challenges. Water Research 43 (9), 2317e2348. Herut, B., Almogi-Labin, A., Jannink, N., Gertman, I., 2000. The seasonal dynamics of nutrient and chlorophyll a concentrations on the SE Mediterranean shelf-slope. Oceanologica Acta 23 (7), 771e782. Herut, B., Shefer, E., Gordon, N., Galil, B., Tibor, G., Tom, M., Rilov, G., 2010. Environmental Quality of Israel’s Mediterranean Coastal Waters in 2009, IOLR Report H64/2010. 179 pp. (in Hebrew). Holm-Hansen, O., Lorenzen, C.J., Holmes, R.W., Strickland, J.D.H., 1965. Fluorometric determination of chlorophyll. Journal du Conseil, Permanent International pour l’Exploration de la Mer 30, 3e15. Karl, D.M., Letelier, R., Hebel, D., Tupas, L., Dore, J., Christian, J., Winn, C., 1995. Ecosystem changes in the North Pacific subtropical gyre attributed to the 1991e92 El Nino. Nature 373 (6511), 230e234. Khan, S.J., Murchland, D., Rhodes, M., Waite, T.D., 2009. Management of concentrated waste streams from highpressure membrane water treatment systems. Critical Reviews in Environmental Science and Technology 39 (5), 367e415. Kimor, B., Berman, T., Schneller, A., 1987. Phytoplankton assemblages in the deep chlorophyll maximum layers off the Mediterranean coast of Israel. Journal of Plankton Research 9 (3), 433e443. Kirchman, D., K’Nees, E., Hodson, R., 1985. Leucine incorporation and its potential as a measure of protein synthesis by bacteria in natural aquatic systems. Applied and Environmental Microbiology 49 (3), 599e607. Kress, N., Galil, B., Shoham-Frider, E., 2010. Marine Monitoring at the Brine Disposal Site of the Via Maris Desalination Plant off Palmahim, Results from May and September 2009. IOLR Report H27/2010. 71 pp. (In Hebrew). Kress, N., Herut, B., 2001. Spatial and seasonal evolution of dissolved oxygen and nutrients in the Southern Levantine Basin (Eastern Mediterranean Sea): chemical characterization of the water masses and inferences on the N:P ratios. DeepSea Research Part I-Oceanographic Research Papers 48 (11), 2347e2372. Krom, M.D., Kress, N., Brenner, S., Gordon, L.I., 1991. Phosphorus limitation of primary productivity in the eastern Mediterranean sea. Limnology and Oceanography 36 (3), 424e432. Krom, M.D., Woodward, E.M.S., Herut, B., Kress, N., Carbo, P., Mantoura, R.F.C., Spyres, G., Thingstad, T.F., Wassmann, P., Wexels-Riser, C., Kitidis, V., Law, C.S., Zodiatis, G., 2005. Nutrient cycling in the south east Levantine basin of the eastern Mediterranean: results from a phosphorus starved
5461
system. Deep Sea Research Part II: Topical Studies in Oceanography 52 (22e23), 2879e2896. Kronenberg, G., 2004. The largest SWRO plant in the world e Ashkelon 100 million m3/y BOT project. Desalination 166, 457e463. Langford, T.E.L., 1990. Ecological Effects of Thermal Discharges. Elsevier. Langford, T.E.L., John, H.S., 2001. Encyclopedia of Ocean Sciences, pp. 2933e2940. Academic Press, Oxford. Lattemann, S., 2010. Development of an Environmental Impact Assessment and Decision Support System for Seawater Desalination Plants. Ph.D. Dissertation, Delft University of Technology, The Netherlands, 294 pp. Lattemann, S., Hopner, T., 2008. Environmental impact and impact assessment of seawater desalination. Desalination 220 (1e3), 1e15. Marie, D., Simon, N., Vaulot, D., 2005. Phytoplankton cell counting by flow cytometry. In: Andersen, R.A. (Ed.), Algal Culturing Techniques. Phycological Society of America, Burlington, pp. 1e15. MCED, 2002. Seawater Quality Environmental Criteria for Israeli Mediterranean Waters. Marine and Coastal Environmental Division, Ministry of Environmental Protection (In Hebrew). NRC, 2008. Desalination, a National Perspective National Research Council of the National Academies. The National Academies Press, Washington, D.C. Pitta, P., Stambler, N., Tanaka, T., Zohary, T., Tselepides, A., Rassoulzadegan, F., 2005. Biological response to P addition in the eastern Mediterranean sea. The microbial race against time. Deep Sea Research Part II: Topical Studies in Oceanography 52 (22e23), 2961e2974. Psarra, S., Zohary, T., Krom, M.D., Mantoura, R.F.C., Polychronaki, T., Stambler, N., Tanaka, T., Tselepides, A., Thingstad, T.F., 2005. Phytoplankton response to a Lagrangian phosphate addition in the Levantine Sea (Eastern Mediterranean). Deep-Sea Research Part Ii-Topical Studies in Oceanography 52 (22e23), 2944e2960. Raventos, N., Macpherson, E., Garcı´a-Rubie´s, A., 2006. Effect of brine discharge from a desalination plant on macrobenthic communities in the NW Mediterranean. Marine Environmental Research 62 (1), 1e14. Roberts, D.A., Johnston, E.L., Knott, N.A., 2010. Impacts of desalination plant discharges on the marine environment: a critical review of published studies. Water Research 44 (18), 5117e5128. Rosentraub, Z., Brenner, S., 2007. Circulation over the southeastern continental shelf and slope of the Mediterranean Sea: direct current measurements, winds, and numerical model simulations. Journal of Geophysical Research 112 (C11), C11001. Safrai, I., Zask, A., 2008. Reverse osmosis desalination plants e marine environmentalist regulator point of view. Desalination 220 (1e3), 72e84. Sa´nchez-Lizaso, J.L., Romero, J., Ruiz, J., Gacia, E., Buceta, J.L., Invers, O., Ferna´ndez Torquemada, Y., Mas, J., RuizMateo, A., Manzanera, M., 2008. Salinity tolerance of the Mediterranean seagrass Posidonia oceanica: recommendations to minimize the impact of brine discharges from desalination plants. Desalination 221 (1e3), 602e607. Sauvet-Goichon, B., 2007. Ashkelon desalination plant e A successful challenge. Desalination 203 (1e3), 75e81. Service, R.F., 2006. Desalination freshens up. Science 313 (5790), 1088e1090. Shannon, M.A., Bohn, P.W., Elimelech, M., Georgiadis, J.G., Marinas, B.J., Mayes, A.M., 2008. Science and technology for water purification in the coming decades. Nature 452 (7185), 301e310.
5462
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 4 9 e5 4 6 2
Simon, M., Azam, F., 1989. Protein content and protein synthesis rates of planktonic marine bacteria. Marine Ecology Progress Series 51, 201e213. Sladkevich, M., Kit, E., 2004. VID Desalination Plant at Ashkelon. Mathematical Model for Spreading of Brine Outlet at Sea Report No. P.N. 612/04. Report No. P.N. 612/04. Smith, S., Azam, F., 1993. A simple economical method for measuring bacterial protein synthesis rates using 3H leucine. Marine Microbial Food Webs 6, 107e114. Stambler, N., 2006. Light and picophytoplankton in the Gulf of Eilat (Aqaba). Journal of Geophysical Research 111 (C11), C11009. Steemann-Nielsen, E., 1952. The use of radioactive carbon (14C) for measuring organic production in the sea. Journal des Conseil International Pour Exploration de la Mer 18, 117e140. Tal, A., 2006. Seeking sustainability: Israel’s evolving water management strategy. Science 313 (5790), 1081e1084. UNEP, 2008. Desalination Resource and Guidance Manual for Environmental Impact Assessments. United Nations Environment Programme, Regional Office for West Asia, Manama, and World Health Organization, Regional Office for the Eastern Mediterranean. Cairo Ed. S. Lattemann, 168 pp.
UNEP/MAP/MEDPOL, 2003. Sea Water Desalination in the Mediterranean: Assessment and Guidelines. MAP Technical Reports Series No. 139 UNEP/MAP Athens. Water-world, 22nd-GWI/IDA. International Desalination Association (IDA) and Global Water Intelligence (GWI). http:// www.waterworld.com/index.html. Yacobi, Y.Z., Zohary, T., Kress, N., Hecht, A., Robarts, R.D., Waiser, M., Wood, A.M., 1995. Chlorophyll distribution throughout the southeastern Mediterranean in relation to the physical structure of the water mass. Journal of Marine Systems 6 (3), 179e190. Yermiyahu, U., Tal, A., Ben-Gal, A., Bar-Tal, A., Tarchitzky, J., Lahav, O., 2007. Rethinking desalinated water quality and agriculture. Science 318 (5852), 920e921. Yogev, T., Rahav, E., Bar-Zeev, E., Man-Aharonovich, D., Stambler, N., Kress, N., Be´ja`, O., Mulholland, M.R., Herut, B., Berman-Frank, I., 2011. Is dinitrogen fixation significant in the Levantine Basin, East Mediterranean Sea? Environmental Microbiology 13 (4), 854e871. Zohary, T., Robarts, R.D., 1998. Experimental study of microbial P limitation in the eastern Mediterranean. Limnology and Oceanography 43 (3), 387e395.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Nitrate removal, communities of denitrifiers and adverse effects in different carbon substrates for use in denitrification beds So¨ren Warneke a,*, Louis A. Schipper a, Michael G. Matiasek b, Kate M. Scow b, Stewart Cameron c, Denise A. Bruesewitz d, Ian R. McDonald d a
Department of Earth and Ocean Sciences, University of Waikato, Hamilton, New Zealand Department of Land, Air and Water Resources, University of California Davis, Davis, United States c Wairakei Research Centre, GNS Science, Taupo, New Zealand d Department of Biological Sciences, University of Waikato, Hamilton, New Zealand b
article info
abstract
Article history:
Denitrification beds are containers filled with wood by-products that serve as a carbon
Received 7 June 2011
and energy source to denitrifiers, which reduce nitrate (NO3) from point source
Received in revised form
discharges into non-reactive dinitrogen (N2) gas. This study investigates a range of
1 August 2011
alternative carbon sources and determines rates, mechanisms and factors controlling
Accepted 6 August 2011
NO3 removal, denitrifying bacterial community, and the adverse effects of these
Available online 17 August 2011
substrates. Experimental barrels (0.2 m3) filled with either maize cobs, wheat straw, green
Keywords:
27.1 C (outlet temperature), and received NO3 enriched water (14.38 mg N L1 and
Denitrification
17.15 mg N L1). After 2.5 years of incubation measurements were made of NO3eN
waste, sawdust, pine woodchips or eucalyptus woodchips were incubated at 16.8 C or
Controlling factors
removal rates, in vitro denitrification rates (DR), factors limiting denitrification (carbon
Bioreactor
and nitrate availability, dissolved oxygen, temperature, pH, and concentrations of NO3,
Denitrification genes
nitrite and ammonia), copy number of nitrite reductase (nirS and nirK ) and nitrous oxide
nir
reductase (nosZ ) genes, and greenhouse gas production (dissolved nitrous oxide (N2O) and methane), and carbon (TOC) loss. Microbial denitrification was the main mechanism for NO3eN removal. NitrateeN removal rates ranged from 1.3 (pine woodchips) to 6.2 g N m3 d1 (maize cobs), and were predominantly limited by C availability and temperature (Q10 ¼ 1.2) when NO3eN outlet concentrations remained above 1 mg L1. The NO3eN removal rate did not depend directly on substrate type, but on the quantity of microbially available carbon, which differed between carbon sources. The abundance of denitrifying genes (nirS, nirK and nosZ ) was similar in replicate barrels under cold incubation, but varied substantially under warm incubation, and between substrates. Warm incubation enhanced growth of nirS containing bacteria and bacteria that lacked the nosZ gene, potentially explaining the greater N2O emission in warmer environments. Maize cob substrate had the highest NO3eN removal rate, but adverse effects include TOC release, dissolved N2O release and substantial carbon consumption by non-denitrifiers. Woodchips removed less than half of NO3 removed by maize cobs, but provided ideal conditions for denitrifying bacteria, and adverse effects were not observed. Therefore we
* Corresponding author. Tel.: þ64 7 858 3700; fax: þ64 7 858 4964. E-mail address: [email protected] (S. Warneke). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.007
5464
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
recommend the combination of maize cobs and woodchips to enhance NO3 removal while minimizing adverse effects in denitrification beds. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Anthropogenic production of reactive nitrogen (N), through the Haber Bosch process, cultivation of N-fixing crops, and combustion of fossil fuels, contributes 45% of global N fixation (Canfield et al., 2010). This human impact on the nitrogen cycle leads to N enrichment of surface waters, with consequences including eutrophication, hypoxia, harmful algae blooms and habitat degradation in lakes, rivers and coastal zones, and an increase in N2O emissions (Howarth et al., 2002; Rabalais, 2002; Phoenix et al., 2006). Denitrification beds are a promising approach to reduce reactive N release from point source discharges into waterways. These denitrifying bioreactors are containers filled with wood by-products, where the wood acts as carbon and energy source for denitrifying microorganisms (Schipper et al., 2010), which convert NO 3 to unreactive N gas via microbial denitrification (Warneke et al., 2011b). A wide range of carbon substrates have been trialled in column studies to find appropriate media for bioreactors (Volokita et al., 1996a,b; Soares and Abeliovich, 1998; Della Rocca et al., 2005, 2006; Saliling et al., 2007; Gibert et al., 2008; Cameron and Schipper, 2010). Nitrate removal rates in column studies range from 3 g N m3 d1 (woodchips; Cameron and Schipper, 2010) to 96 g N m3 d1 (rice husk; Shao et al., 2008). The exceptionally high NO 3 removal rates of many carbon substrates (e.g., rice husks, wheat straw, cotton) were attributed to a large organic carbon release in the startup phase of the columns, and were not sustainable over a longer time period (Cameron and Schipper, 2010). In a longterm study, barrels filled with maize cobs removed 3e6.5 times more NO 3 eN than wood substrate, but also had higher carbon leaching in the effluent (Cameron and Schipper, 2010). Greenan et al. (2006) also reported that maize stalks produced greater NO 3 removal than woodchips. However, little is known about the mechanism responsible for NO 3 removal, the controlling factors, denitrifying bacterial communities or adverse effects, such as greenhouse gas release, when using different carbon substrates than woodchips. Warneke et al. (2011a, b) demonstrated that the mechanism responsible for NO 3 removal in a full-scale woodchip bioreactor was microbial denitrification, and the removal process was limited by microbially available carbon and temperature. Smaller-scale studies have also determined that microbial denitrification is the dominant N removal mechanism, rather than dissimila tory NO 3 reduction to ammonium DNRA or NO3 immobilization (Robertson, 2010; Greenan et al., 2006, 2009; Gibert et al., 2008). Greenhouse gas (GHG) production during denitrification is an important issue to address when studying denitrification beds. An in field woodchip bioreactor study by Warneke et al. (2011a) yielded total N2O release of 4.3% of removed NO 3 eN, whereas Greenan et al. (2009) reported negligible release of dissolved N2O in a woodchip column study.
However, there have been no studies examining GHG production in denitrification beds containing different carbon sources. So far, the population of denitrifying bacteria has not been investigated in substrates for use in denitrification beds. The abundance of denitrifying communities can be estimated by quantifying the functional gene copy numbers for nitrite reductase, nirS and nirK, and nitrous-oxide reductase, nosZ. These denitrification genes express reductase enzymes involved in denitrification. NirS expresses the cytochrome cd1containing nitrite reductase (which catalyses the reduction of nitrite to nitric-oxide), nirK expresses the copper containing nitrite reductase, and nosZ expresses nitrous oxide reductase (which catalyses the reduction of N2O to N2) (Zumft, 1997; Braker et al., 1998). The two different genes for nitrite reductase, nirS and nirK, have coevolved to produce two independent pathways and no denitrifier is known to contain both pathways (Philippot, 2002). Interestingly many denitrifying organisms have been shown to reduce NO 3 only to nitrous oxide (Cheneby et al., 1998, 2004) and some, such as Agrobacterium tumerfaciens C58 do not possess nitrous oxide reductase (nosZ ) (Wood et al., 2001). Many studies have shown that differences in the diversity and abundance of denitrifying bacterial genes were correlated to a variety of physical and chemical conditions; organic carbon in glacier foreland (Kandeler et al., 2006), temperature in constructed wetlands (Chon et al., 2010), water logging in rice paddy soils (Yoshida et al., 2009), organic or conventional fertilizer in agricultural soils (Dambreville et al., 2006; Enwall et al., 2005), native and cultivated soils (Stres et al., 2004), soil pH in grassland soils (Cuhel et al., 2010), nitrous-oxide emissions (Philippot et al., 2009) and NO 3 concentration in woodlands with different vegetation (Lindsay et al., 2010). However, the diversity and abundance of denitrifying bacteria under consistent environmental conditions (e.g., same temperature, NO 3 concentration, DO concentration, flow rate), but with different carbon substrates are poorly known. This study followed a 2.5-year trial by Cameron and Schipper (2010), where different C substrates were compared for their ability to remove NO 3 from water at two temperatures. The main objectives of the present study were to determine the limiting factors and the microbial mechanisms of the NO 3 removal for different C substrates such as woodchips (Pine and Eucalyptus), sawdust, green waste, maize cobs and wheat straw in these barrels. The abundance of the denitrification functional genes nirS, nirK and nosZ were compared across replicate barrels, different temperatures and substrates. The factors affecting denitrifying communities were examined and whether NO 3 removal could be predicted from the copy number of denitrification genes. Adverse effects, including production of N2O and methane (CH4), and total organic carbon (TOC) release, were also determined to evaluate the benefit of the different C
5465
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
substrates. These findings can be used to help select the appropriate carbon substrate for denitrifying bioreactors (denitrification beds and walls) to optimise NO 3 removal, reduce GHG production, and maximize the lifetime of the bioreactor.
2.
Materials and methods
2.1.
Study site and substrate
The design of the experimental setup was fully described in Cameron and Schipper (2010). In this study, 24 experimental barrels (0.2 m3) filled with six different carbon substrates and placed in a 7 m long shipping container were continuously solution (in average loaded with a self-prepared NO 3 about15.8 mg L1; Table 1). Barrels were divided between a cold treatment (16.8 C average outlet temperature) and a warm treatment (27.1 C average outlet temperature), and every carbon source had two replicate barrels at each temperature. The selected substrates were: woodchips of Pinus radiata (soft wood), woodchips of Eucalyptus “Red Duke” (hardwood), sawdust (P. radiata), maize cobs, wheat straw and green waste (shredded and chipped miscellaneous shrubbery leaves and stems). The barrels had been loaded with NO 3 solution for 2.5 years before samples were taken for this study.
2.2.
Solute concentrations and NO 3 removal rate
Water was sampled from the inflow and outflow tubing of the cold and warm barrels. Samples were filtered through disposable membrane filters (0.45 mm) and analysed for NO 3, NHþ 4 and NO2 using a flow injection analyser (Lachat Instruments; Loveland, USA) (APHA, 1992). TOC was determined from unfiltered water samples using a Shimadzu TOC-5000 analyser (Shimadzu Corp.; Kyoto, Japan). Temperature and DO of the inlet and outlet of the barrels were measured with an InLab 605 O2-Sensor (Mettler Toledo, Switzerland). Nitrate removal rate was calculated as follows: NO 3 eN 1 removal rate ¼ DNO 3 eN FR V , where DNO3 eN was the difference of inflow and outflow NO 3 eN concentration, FR was the flow rate of the NO 3 solution, and V was the volume of the barrel.
2.3.
Greenhouse gas production
Water from the inlet and outlet of the barrels was collected in 3.7 mL exetainers (Labco, UK) for analysis of dissolved N2O and CH4 concentrations. The exetainers contained 0.2 mL H2SO4 (20%) to prevent further bacterial activity. After 12 h headspace equilibrium at room temperature, headspace gas samples were analysed for N2O and CH4 concentration using a gas chromatograph equipped with an electron capture detector and flame ionisation detector, respectively (Varian;
Table 1 e Solute concentrations and temperature at the inlet and outlet of barrels filled with different carbon substrates under warm and cold incubation. Barrela (cold line)
NO 3 eN (mg L1)
NO 2 eN (mg L1)
NHþ 4 eN (mg L1)
pH
Temp ( C)
DO (mg L1)
TC (mg L1)
TOC (mg L1)
Inlet Outlet PW1 Outlet PW2 Outlet MC1 Outlet MC2 Outlet WS1 Outlet WS2 Outlet GW1 Outlet GW2 Outlet SD1 Outlet SD2 Outlet EW1 Outlet EW2 Inlet Outlet PW1 Outlet PW2 Outlet MC1 Outlet MC2 Outlet WS1 Outlet WS2 Outlet GW1 Outlet GW2 Outlet SD1 Outlet SD2 Outlet EW1 Outlet EW2
14.4 10 10.5 0.4 0.1 0.5 1 6.2 2.2 8.3 5.7 9.7 9.6 17.2 12 11.2 3.7 6.1 8.5 9.3 4.3 7.2 8.5 8.6 9.6 10.9
0.023 0.080 0.179 0.033 0.003 0.060 0.025 0.153 0.024 0.164 0.020 0.082 0.424 0.007 0.234 0.171 0.079 0.094 0.410 0.562 0.088 0.234 0.444 0.494 0.816 0.516
<0.001 <0.001 <0.001 <0.001 <0.001 0.134 0.065 0.178 0.785 0.364 0.344 0.033 <0.001 <0.001 0.027 <0.001 0.081 0.364 0.259 0.194 <0.001 <0.001 0.307 0.084 0.024 0.123
7.7 6.9 6.9 6.2 5.9 6.9 6.9 6.8 6.6 6.8 7.1 7.0 7.0 8.3 7.6 7.5 7.6 7.7 7.6 7.6 7.5 7.6 7.6 7.6 7.6 7.6
19.2 18 17.2 17.2 16.2 17.1 16.6 16.6 16.3 16.9 16.7 16.6 16.3 36 26 27.3 26.3 29 29.1 27 27.4 25 27.2 28.4 27.7 25.3
7.1 1.9 1.3 1.1 0.5 0.7 0.5 0.9 0.5 0.4 0.3 0.6 0.5 5.9 1 0.6 1 1 0.4 1.1 0.9 1.1 1.3 0.6 0.4 1.1
13.5 18.3 26.6 100.4 86.8 56.1 49.3 25.1 60 23.2 14.8 21.6 29 14 19.5 18.2 51.4 53.7 33.71 10.9 52.1 46.4 23.8 13.7 30 20.9
5.4 6.8 9.1 70.2 76.8 16.5 11 7 14.7 5.9 4.6 4.7 7.4 6.0 5.8 6.2 9.5 9.7 7.8 0.6 8.3 7.9 5.8 3.8 5.8 5.1
a PW1 and PW2, soft woodchips (pine); MC1 and MC2, maize cobs; WS1 and WS2, wheat straw; GW1 and GW2, green waste; SD1 and SD2, sawdust; EW1 and EW2, hard woodchips (eucalyptus).
5466
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
Palo Alto, USA). Dissolved N2O and dissolved CH4 gas concentrations were calculated after Weiss and Price (1980) and Yamamoto et al. (1976) using the Bunsen coefficients. For N2O analyses, the gas chromatograph was fitted with a Hayesep D column (3.6 m 1/800 2.0 mm). The column oven temperature was 80 C, the ECD detector temperature was 300 C, and the flow rate of the carrier gas (10% methane in argon) was 40 mL min1. For CH4 analyses, the GC was equipped with a Hayesep Q column (80 1/800 SS; Q 80e100). The column oven temperature was 90 C, the FID detector temperature was 150 C, and the flow rate of the N2-carrier gas was 30 mL min1.
2.4.
Denitrification rates
Denitrification rates (DR) of the different carbon substrates in the barrels were determined using a modification of the denitrifying enzyme activity (DEA) method of Tiedje et al. (1989). Carbon substrate (600 g wet weight) from each barrel was collected using a gloved hand from the centre of the barrel and stored in plastic bags at 4 C. Rubber gloves were changed after each sampling. Water samples (500 mL) from the outlet of each barrel were stored in 1 L plastic bottles at 4 C. In the laboratory, the substrate and water samples were equilibrated to room temperature in a water bath. Carbon substrate (100 g wet weight) and water (60 g) from each barrel were added to four airtight bottles (600 mL). The headspace of the bottles was flushed with N2 gas for 10 min prior to injection of 40 mL of acetylene (10% of the headspace volume), to inhibit reduction of N2O to N2. Each assay was amended with one of four solutions (all 5 mL): i) glucose (8 g L1; DR þ C), ii) potassium 1 1 and 4 g L1 NO 3 (4 g L ; DR þ N), iii) glucose and KNO3 (8 g L respectively; DR þ C/N), and iv) no amendment (DR), to identify whether DR was C and/or NO 3 limited. After bottles were incubated at 27 C on a shaker table (100 rpm), headspace gas samples were collected through a rubber septum after 30, 40, 50 and 60 min using a syringe. Gas samples were stored in 3.7 mL exetainers (Labco, UK) until analysis for N2O concentration within 7 days via GC-ECD (see above).
2.5.
DNA extraction
Carbon substrates (400 mL) were sampled from the centre of each barrel, sealed in 500 mL airtight plastic containers and stored at 24 C until frozen samples were vacuum freeze dried. Several trial DNA extractions were performed on the 6 types of reactor bed material. It was determined that the corn cobs, green waste and sawdust, performed best with the FastDNA SPIN Kit for Soil (MP Biomedicals, Solon, OH) whereas the bulkier samples, woodchips and wheat straw performed better with the Mo Bio Ultra Clean Mega Prep Soil DNA kit (Mo Bio Laboratories, Inc., Carlsbad, CA). The criteria for selecting an extraction method was based on the amount of DNA extracted per amount of material extracted and the total number of 16S rRNA genes per gram dry material as determined by quantitative PCR (data not shown). The corn cobs, woodchips and wheat straw were reduced in size with a sterile scalpel and or scissors, so that they could fit in the initial extraction tube. The FastDNA SPIN Kit for Soil was used to extract 0.05e0.1 g of corn cobs, 0.13e0.22 g green waste
and 0.1e0.14 g of sawdust as per manufacturer instructions. The Ultra Clean Mega Prep Soil DNA kit was used to extract 2.27e2.65 g of pine woodchips, 0.45e0.69 g of wheat straw and 3.32e4.04 g of Eucalyptus woodchips as per manufacturer instructions. All samples were extracted in duplicate. The quantity of DNA extracted was quantified with a Qubit fluorometer (Life Technologies, Carlsbad, CA).
2.6.
Quantitative PCR
Thermal cycling, fluorescent data collection, and data analysis were performed on an ABI Prism 7300 sequence detection system (Life Technologies, Carlsbad, CA) according to the manufacturer’s instructions using SYBR-green based detection. Initially, the DNA extractions for each sample type were diluted from 20- to 1000-fold to determine the optimum DNA concentration for QPCR. It was determined that a 200-fold dilution was required for all samples to dilute past PCR inhibitors that were coextracted (data not shown). QPCR reactions for nirK, nirS and nosZ and 16S rRNA contain 5 uL of template DNA, 0.5 mM of each forward and reverse primer except nosZ which used 1.5 uM of primer, 12.5 mL of 2 SYBR GreenER QPCR Super Mix (Life Technologies, Carlsbad, CA), in a total volume of 25 mL. The primers (50 -30 ) used to detect the nirK, nirS and nosZ and 16S rRNA genes are nirK876 (ATY GGC GGV AYG GCG A) and nirK1040 (GCC TCG ATC AGR TTR TGG TT) (Henry et al., 2005) nirSCd3aF (AAC GYS AAG GAR ACS GG) and nirSR3cd (GAS TTC GGR TGS GTC TTS AYG AA) (Kandeler et al., 2006), nosZ2F (CGC RAC GGC AAS AAG GTS MSS GT) and nosZ2R (CAK RTG CAK SGC RTG GCA GAA) (Henry et al., 2006), 341F (CCT ACG GGA GGC AGC AG) and 534R (ATT ACC GCG GCT GCT GGC A, also referred to as 515R) 16S rRNA primers (Lopez-Gutierrez et al., 2004) respectively. The conditions for nirK and nirS real-time PCR are 10 min at 95 C for enzyme activation; afterwards six touchdown cycles are performed: 15 s at 95 C for denaturation, 30 s at 63 C for annealing, and 30 s at 72 C for extension. The annealing temperature is progressively decreased by 1 C down to 58 C. Finally, a last cycle with an annealing temperature of 58 C is repeated 40 times with the addition of a data acquisition step of 30 s at 80 C after the extension phase. One last step of 95 C for 15 s, 60 C for 30 s and 95 C for 30 s is added to obtain a specific denaturation curve. The thermal cycling conditions for nosZ are similar except for the annealing temperature, which is 65 C for 30 s for the first 6 cycles and 60 C for 15 s for the 40 cycles. 16S rRNA QPCR was performed with no touchdown cycle, just one annealing temperature at 60 C for 30 s and only 35 cycles instead of 40. Purity of amplified products was checked by the observation of a single peak during the dissociation analysis. Copy Numbers were determined by using a standard curve obtained with serial plasmid dilutions of a known amount of plasmid DNA containing a fragment of the 16S rRNA gene, nirK, nirS and nosZ gene. Each DNA extraction was analyzed for each gene in triplicate along with three non-template controls. Denitrification gene copy numbers are reported as copies per gram dry material and also reported as normalized to 16S rRNA gene copies. The nitrite reductase to nitrous oxide reductase ratio (Snir/nos) was determined by summing the nir genes (nirS þ nirK ) and dividing the sum by the nos genes and was used as an
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
indication of nitrous-oxide producing potential. To determine if each environment selected more for nirS or nirK, the ratio of nirS/nirK was also calculated. The authors acknowledge that novel bacterial sequences were likely missed by the 16S rRNA primers used in this study which might have resulted in an underestimation of the community size in our soil, which subsequently led to the calculation of higher relative abundances of nirS functional genes.
2.7.
Respirable C
Respirable C was measured, as an index of the availability of C to microorganisms, using a modified alkali trap method of Cheng and Coleman (1989). Carbon substrates (100 g wet weight) and effluent (60 g) from each barrel were added to airtight bottles (600 mL). Small beakers (30 mL) filled with 10 mL of 0.5 M KOH were placed into the jars to trap CO2. After sealing the bottles, the headspaces were flushed with N2 gas for 10 min and incubated at room temperature (22 C) for 4 days. After incubation, 5 mL of the CO2 trapping solution were removed from the bottles and mixed with 10% BaCl2 solution (10 mL) and phenolphthalein (pH indicator) in 100 mL flasks. After back-titration of these solutions against the standard 0.1 M HCl to determine the amount of trapped CO2, respirable carbon was expressed as CO2eC g1 carbon substrate (dry weight).
2.8.
Statistical analysis
Similarities and differences of nitrite reductase gene copies (Snir) per gram carbon substrate, nirS/nirK and nir/nosZ were evaluated calculating the Wald confidence interval (95%) of these gene copy numbers for each barrel (data not shown). Associated errors of the results are reported as standard errors.
3.
Results
3.1.
Solute concentrations
The average temperature of the cold and warm incubation outlet was 16.8 C and 27.1 C respectively, and used as calculation basis for determining the Q10. Q10 is the factor of the reaction rate increase with every 10 C rise in temperature. The inlet NO 3 eN concentration of the cold barrels was 14.4 0.6 mg L1, and for the warm barrels was 17.2 1 mg L1. The average flow rates of the cold and warm barrels were 48.3 2.0 ml min1 and 58.5 2.3 ml min1, respectively. NitriteeN concentrations in the outflow were always below 0.2 mg L1 for cold barrels and ranged from 0.08 to 0.82 mg L1 for warm barrels (Table 1). In the cold incubations wheat straw, green waste and sawdust, and in the warm incubations all the carbon substrates, except green waste, 1 (Table 1). All the released NHþ 4 ranging from 0.03 to 0.79 mg L barrels showed a slight decrease in pH at the outflow (Table 1). DO decreased from 7.1 mg L1 (inlet concentration) to below 1.9 mg L1 (outlet concentration) in cold barrels, and from 5.9 mg L1 to below 1.3 mg L1 in warm barrels (Table 1). TOC was released in high concentrations from the cold incubated
5467
maize cobs (70.2 and 76.8 mg L1) and wheat straw (16.5 and 11 mg L1), and in the warm incubated maize cobs (9.5 and 9.7 mg L1). However other carbon substrate barrels released either low concentrations, or consumed, TOC (Table 1).
3.2. Nitrate removal and controlling factors of denitrification NitrateeN removal rates ranged from 1.3 (soft woodchip barrel 2) to 6.2 g N m3 d1 (maize cobs barrel 2), and were dependent on temperature with a Q10 of 1.2 0.13 (Fig. 1). Maize cobs, wheat straw and green waste showed the highest NO 3 eN removal rates, ranging from 4.3 g N m3 d1 (green waste) to 5.7 g N m3 d1 (maize cobs) in cold barrels, and from 4.5 g N m3 d1 (wheat straw) to 6.0 g N m3 d1 (maize cobs) in warm barrels (Fig. 1). NitrateeN removal increased linearly with the in vitro denitrification rate DR þ C/N for cold and warm incubation ( y ¼ 0.16x þ 1.6; R2 ¼ 0.63; p ¼ 0.002 and y ¼ 0.24x þ 2.9; R2 ¼ 0.65; p ¼ 0.001 respectively; where y ¼ NO 3 eN removal rate in g N m3 d1 and x ¼ DR þ C/N in mg N h1 g1) (Fig. 2). Furthermore, the NO 3 eN removal rate depended on the available carbon content as shown in three ways. NitrateeN removal rate was linearly correlated with respirable carbon for both cold and warm incubated carbon substrates ( y ¼ 0.08x þ 1.6; R2 ¼ 0.82; p < 0.001 and y ¼ 0.15x þ 1.2; R2 ¼ 0.62; p ¼ 0.002 respectively; where y ¼ NO 3 eN removal rate in g N m3 d1 and x ¼ respirable carbon in mg C g1 d1) (Fig. 3). In vitro measured DR could be enhanced with a glucose amendment for all carbon substrates, except for maize cobs and wheat straw in cold and warm barrels. DR in cold incu bated maize cobs and wheat straw were NO 3 limited (NO3 eN 1 concentration <1 mg L ) (Fig. 4; Table 1) and DR in warm incubated maize cobs and wheat straw were not limited by glucose or NO 3 , except for one warm barrel of maize cobs (MC1), which was also limited by glucose (Fig. 4). Nitrate amended DR (DR þ N) was also significantly correlated with
Fig. 1 e Nitrate removal rates for different carbon substrates in cold (16.8 C) and warm (27.1 C) barrels. PW1 and PW2, soft woodchips (pine), replicates; MC1 and MC2, maize cobs; WS1 and WS2, wheat straw; GW1 and GW2, green waste; SD1 and SD2, sawdust; EW1 and EW2, hard woodchips (eucalyptus).
5468
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
Fig. 2 e NitrateeN removal rate as a function of in vitro DR amended with glucose and nitrate (DR D G/N) for cold and warm incubated substrate. Linear regression statistics are reported in text.
Fig. 4 e In vitro denitrification rates (DR) at 27 C for different carbon substrates in cold (A) and warm (B) barrels. DR assays were amended with glucose (DR D C), NOL 3 (DR D N), glucose and NOL 3 (DR D C/N), and none amended (DR). PW1 and PW2, soft woodchips (pine); MC1 and MC2, maize cobs; WS1 and WS2, wheat straw; GW1 and GW2, green waste; SD1 and SD2, sawdust; EW1 and EW2, hard woodchips (eucalyptus).
respirable carbon for cold and warm incubations ( y ¼ 0.38x 1.3; R2 ¼ 0.70; p < 0.001 and y ¼ 0.37x e 4.3; R2 ¼ 0.48; p ¼ 0.013 respectively; where y ¼ DR þ N in mg N h1 g1 and x ¼ respirable carbon in mg C g1 d1) (Fig. 3B).
3.3. Copies of denitrification genes (nirS, nirK and nosZ )
Fig. 3 e NitrateeN removal rate (A) and in vitro DR amended with nitrate (DR D N) (B) as a function of respirable carbon for cold and warm incubated substrate. Linear regression statistics are reported in text.
The abundance of nirS, nirK and nosZ ranged from 8.7 0.8 106 (pine woodchips) to 1.6 0.01 1010 (green waste) copies of nirS g1 dry substrate, 0.7 0.1 106 (pine woodchips) to 6.8 0.1 109 (maize cobs) copies of nirK g1 dry substrate, and 1.2 0.1 106 (pine woodchips) to 9.0 0.2 109 (maize cobs) copies of nosZ g1 dry substrate for cold incubations (Table 2). Abundance of nirS, nirK and nosZ in warm incubated substrate ranged from 2.0 0.1 107 (pine woodchips) to 1.3 0.04 1011 (maize cobs) copies of nirS g1 dry substrate, 7.4 0.4 106 (pine woodchips) to
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
5469
Table 2 e Average copy number (3106) of denitrification genes (nirS, nirK and nosZ ) and 16S rRNA isolated from different carbon substrates used in denitrifying barrels under cold and warm incubations. Treatment Cold
Warm
Carbon substratea
nirS copies g1 dry substrate
nirK copies g1 dry substrate
PW1 PW2 MC1 MC2 WS1 WS2 GW1 GW2 SD1 SD2 EW1 EW2 PW1 PW2 MC1 MC2 WS1 WS2 GW1 GW2 SD1 SD2 EW1 EW2
10.1 2.9 8.7 0.8 4718 299.2 7474.5 160.2 388.6 29.4 195 17.9 7296.5 368.3 16163.2 137.3 3412.4 43.5 3319.6 53.3 41.5 6.3 57.9 1.3 29.2 4 19.9 1.1 70533.5 481.5 126400 3695.2 1596.6 135.9 610.8 3.9 17237.9 488.2 13763.3 385.2 2716.2 35.8 2967.5 98.4 67.1 4.4 72.1 3
12 0.2 0.7 0.1 4217 301.8 6815.9 147 211.3 4.8 148.4 3.3 2919 72.4 5664.1 667.7 2227.5 185.9 1998.5 126.2 22.7 1.7 34 1.6 22.8 0.6 7.4 0.4 13860 504.3 14926.2 719.7 393.6 33.5 456.3 5044.3 88.3 3957.4 158.3 1224.2 48.2 1321.6 25.4 18.7 0.4 37.5 2.4
nosZ copies g1 dry substrate
16S rRNA copies g1 dry substrate
2.5 0.1 1.2 0.1 5760.1 783.8 8990.3 180.4 138.3 0.2 77.2 1.6 4183.6 109.3 7168.5 143.8 1064.2 50.4 927 25.2 15.9 0.5 23.3 1.3 14 1.1 6.5 0.3 19230.2 171.3 18745.9 365 200 9.9 230.9 5 7415.2 182.7 2670.5 157.7 696.8 18.5 974.1 7.7 14.8 0.6 25.1 0.8
3.1 0.1 1.2 0.1 96761.9 2649.3 43138.8 2076.5 351.1 28.1 1.2 0.1 22755.1 596.1 22660.2 1595.2 4745.9 253.5 3648.9 338.6 30.3 1.0 40.8 1.6 23.9 1.2 59.6 1.2 45035.9 1781.8 50774.6 4753.5 3870.4 19.1 476.7 33.4 35076.9 1550.5 18592.3 919.1 2217.6 187.1 2710.5 61.5 29.6 0.8 263.8 18.8
a PW1 and PW2, soft woodchips (pine); MC1 and MC2, maize cobs; WS1 and WS2, wheat straw; GW1 and GW2, green waste; SD1 and SD2, sawdust; EW1 and EW2, hard woodchips (eucalyptus).
1.5 0.07 1010 (maize cobs) copies of nirK g1 dry substrate, and 6.5 0.3 106 (pine woodchips) to 1.9 0.02 1010 (maize cobs) copies of nosZ g1 dry substrate (Table 2). The NO 3 removal rate increases exponentially with the total copy number of nitrite reductase genes (Snir) per gram substrate and was significantly linearly correlated with the ln (Snir g1 substrate) in cold and warm barrels ( y ¼ 0.45x 5.62; R2 ¼ 0.48; p ¼ 0.012 and y ¼ 0.38x 3.68; R2 ¼ 0.73; p < 0.001 respectively; 3 d1 and x ¼ ln where y ¼ NO 3 eN removal rate in g N m 1 (copies Snir g substrate) (Fig. 5). Generally, the copies of Snir were greater in warm than in cold barrels, except for sawdust. A temperature increase of 10 C yielded 4-fold increases in Snir (Fig. 6A). The carbon substrates maize cobs and green waste had the greatest bacterial population ranging from 18592.3 919.1 106 copies of 16S rRNA g1 dry substrate (warm incubated green waste) to 96761.9 2649.3 106 copies of 16S rRNA g1 dry substrate (cold incubated maize cobs), and the greatest Snir per gram carbon substrate (Table 2). In contrast, nitrite reductase gene copies (Snir) normalized to total bacteria (16S rRNA genes) of these substrates (maize cobs and green waste) were at the lower end of the data generated in this study, ranging from 0.1 0.00 (cold incubated maize cobs) to 2.82 0.11 copies Snir copies1 16S rRNA g1 dry substrate (warm incubated maize cobs) (Fig. 6B). Cold incubated pine woodchips had the highest Snir copy number normalized to total bacteria (7.29 0.58 and 7.89 0.07 copies Snir copies1 16S rRNA g1 dry substrate), followed by eucalyptus woodchips for cold incubations (Fig. 6B).
In order to estimate how the abundance of the different genes in the denitrifying pathway changed with respect to the other steps in denitrification, the ratios of copies of nirS/nirK, and Snir/nosZ (nitrous oxide reductase) were determined (Fig. 7). Increasing temperature increased the ratio of nirS/nirK, and Snir/nosZ, except for pine woodchips.
Fig. 5 e NitrateeN removal rate as a function of total nitrite reductase gene (Snir) copies for cold and warm incubated substrates. Linear regression statistics are reported in text.
5470
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
Fig. 6 e Total number of nitrite reductase genes (Snir) normalized per gram carbon substrate (A) and normalized to total bacteria (16S rRNA) (B) of the different carbon substrates used in the barrels under cold and warm incubation. PW1 and PW2, pine woodchips; MC1 and MC2, maize cobs; WS1 and WS2, wheat straw; GW1 and GW2, green waste; SD1 and SD2, sawdust; EW1 and EW2, eucalyptus woodchips. Error bars are one standard error (n [ 3).
For cold incubations the ratios of nirS/nirK within the same carbon substrate (replicates) were not different from each other applying the Wald confidence interval (95%), except for pine wood. The same was observed for the ratios of Snir/nosZ within the same carbon substrate in cold barrels, whereas in warm barrels differences in ratios of nirS/nirK, or Snir/nosZ were shown for each carbon source, except for nirS/ nirK ratios of warm incubated green waste and sawdust barrels (Fig. 7).
3.4.
Greenhouse gases
The inlet concentrations of dissolved N2OeN were below the detection limit (<1.1 mg L1). Therefore the measured dissolved N2OeN and CH4 concentrations in the outlet water of the barrels are the net dissolved N2OeN release from the barrels in the outlet water. The dissolved N2OeN release from the cold barrels in the outlet ranged from
Fig. 7 e Ratios of gene copies of nirS/nirK (A) and total nitrite reductase (Snir) to nitrous oxide reductase (nosZ ) (B). PW1 and PW2, pine woodchips; MC1 and MC2, maize cobs; WS1 and WS2, wheat straw; GW1 and GW2, green waste; SD1 and SD2, sawdust; EW1 and EW2, eucalyptus woodchips. Error bars are one standard error (n [ 3).
below detection limit (sawdust) to 214.5 mg L1 (wheat straw) and from the warm barrels from below detection limit (sawdust) to 1472.5 mg L1 (wheat straw). Wheat straw was the largest source of N2O for both cold and warm incubations, followed by green waste in warm incubations. Warm wheat straw barrels released almost 10% of the removed NO 3 eN as dissolved N2OeN in the outlet water. All substrates at the warmer temperature released on average about seven times more dissolved N2OeN in the outlet than cold barrels (Fig. 8). The inlet concentration of dissolved CH4 was 5.4 mg CH4 L1 for cold and 16.8 mg CH4 L1 for warm barrels. There was little net dissolved CH4 release in the outlet of woodchips (hard and soft wood) and sawdust (<40 mg L1) detected. Wheat straw released some dissolved CH4 in the outlet water at cold incubations (139 mg L1 and 1201 mg L1) and maize cobs released large amounts of dissolved CH4 at cold incubations (10,600 mg L1 and 7375 mg L1) in the outlet of the barrels, but less dissolved CH4 at warm incubation. Barrels of green waste released dissolved CH4 in the outlet from cold and warm barrels, with an average of 2970 mg L1 and 3870 mg L1, respectively (Fig. 8).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
5471
N2O and CH4 concentrations along the length of a field-scale woodchip bioreactor during a sampling period of one year.
4.1.
Fig. 8 e Dissolved nitrous oxide (A) and methane (B) concentrations in the outlet water of different carbon substrates in cold and warm barrels. PW1 and PW2, soft woodchips (pine); MC1 and MC2, maize cobs; WS1 and WS2, wheat straw; GW1 and GW2, green waste; SD1 and SD2, sawdust; EW1 and EW2, hard woodchips (eucalyptus).
4.
Discussion
In this study, several different carbon substrates (maize cobs, wheat straw, green waste, sawdust, hardwood and softwood) receiving NO 3 from a simulated household effluent (inlet NO3 1 concentration between 14 and 18 mg L ) were examined to determine factors controlling NO 3 removal and the extent of possible adverse effects. The denitrifying bacterial communities in the different barrels were also examined to determine whether microbial community structure could account for differences in activity (NO 3 removal, dissolved GHG concentrations). The experimental barrels had been operating for 2.5 years prior to these measurements, thereby eliminating short term study effects (i.e., high TOC release coupled with high NO 3 removal rates), as have been described in other column and barrel studies (Greenan et al., 2009; Cameron and Schipper, 2010; Soares and Abeliovich, 1998). In our study a single sampling was taken. However, we consider that steady state had been reached in the microbial community, which allow comparisons between substrates; e.g., Warneke et al. (2011a) found only very small differences in dissolved
Nitrate removal and microbial processes
The mean NO 3 eN removal rates of the experimental barrels were less than the NO 3 eN removal rates reported by Cameron and Schipper (2010) in the same experimental barrels for the previous 2.5 years, and less than the reported rates of most other column studies with alternative carbon substrates (Gibert et al., 2008; Saliling et al., 2007; Greenan et al., 2006; Della Rocca et al., 2005; Shao et al., 2008; Soares and Abeliovich, 1998). These lower NO 3 removal rates were most likely due to the age of the carbon material (>2.5 years in use) and the 10-fold lower NO 3 eN inlet concentration than used by Cameron and Schipper (2010). For example, in this study, NO 3 eN removal rates of cold incubated maize cobs and wheat straw were clearly limited by NO 3 eN concentrations (NO3 eN 1 outlet concentrations <1 mg L ; Table 1). Nitrate removal rates of pine and eucalyptus woodchip and sawdust ranged from 1.3 to 4.4 g N m3 d1 and were at the lower end of removal rates determined for woodchip bioreactors in the field (Schipper et al., 2010). Maize cobs, followed by wheat straw and green waste, exhibited a higher NO 3 removal rate than wood substrates in this study, as also reported by Cameron and Schipper (2010) for the same experimental system. However, the NO 3 eN removal rates for wood substrates in this study were in the same range as the NO 3 eN removal rates (3.9 g N m3 d1) measured by Greenan et al. (2009) in a column study. Other column studies with woodchips showed NO 3 eN removal rates 2e10 times higher than this study (Robertson, 2010; Saliling et al., 2007). As expected, there was good evidence that the mechanism for NO 3 eN removal in the substrates was most likely microbial denitrification, because the measured in vitro DR þ C/N of each experimental barrel were higher than many other NO 3 reducing ecosystems e.g., denitrification walls (Schipper et al., 2005; Moorman et al., 2010), forested land-based wastewater treatment system (Barton et al., 2000), riparian forest sites (Groffmann et al., 1992), a natural wetland and a constructed wetland (Duncan and Groffmann, 1994). Additionally, nitrite reductase genes (nirS and nirK ), which are responsible for the second step of denitrification, were on average more abundant in this study (Table 2) than in constructed wetlands (Chon et al., 2010), or rice fields (Yoshida et al., 2009). Furthermore, the significant linear relationship of the increase of NO 3 removal, and the increase of measured DR þ C/N, indicated that microbial denitrification was responsible for the NO 3 eN removal, regardless of the carbon substrate in the experimental barrels and showed that the acetylene inhibition method was a good measure for comparative NO 3 removal estimations between C substrates (Fig. 2). Although seven of the 12 cold barrels, and eight of the warm barrels produced small amounts of NHþ 4 , neither anammox or DRNA appeared to be significant contributors to þ NO 3 removal, because of the low NH4 eN concentration 1 (<0.8 mg L ) at the outlet. Both Gibert et al. (2008) and Greenan et al. (2006) also suggested that DNRA is a minor process involved in NO 3 removal (less than 5%).
5472
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
As NO 3 was depleted in the cold incubated maize cobs and wheat straw barrels, methanogenic bacteria were able to compete successfully with denitrifiers for carbon as suggested by the high dissolved CH4 production of cold incubated maize cobs and wheat straw barrels. Although NO 3 eN concentrations were above 2 mg L1 in the outlet of cold green waste barrels and warm maize cobs and green waste barrels, we observed dissolved CH4 production (Table 1, Fig. 8), which suggests that methanogenes may occur even at relatively moderate NO 3 concentrations. It is likely that once the microbial consumption of NO 3 exceeded diffusion of NO3 within the carbon substrate, methanogenes could develop in the interior of the substrate.
4.2.
Factors controlling NO 3 removal
In general, denitrification is primarily controlled by carbon availability, NO 3 , NO2 , sulphide, temperature, DO, and the number of denitrifiers (Firestone and Davidson, 1989; Seitzinger et al., 2006). In this study, carbon availability and temperature were identified as the main factors limiting nitrate removal in the experimental barrel systems, when 1 NO 3 concentrations were more than 1 mg L ; below this limited denitrification. concentration NO 3 The warm barrels removed more NO 3 than the cold barrels, with a Q10 factor of 1.2 0.13 (Fig. 1). Cameron and Schipper (2010) found a greater temperature dependence of NO 3 removal (Q10 ¼ 1.6) in the same experimental system, but these measurements were made with 10 times higher NO 3 inlet concentrations, whereas in the present study NO 3 limited in some barrels the NO 3 removal. Studies of woodchip bioreactors by Robertson et al. (2008), Elgood et al. (2010) and Warneke et al. (2011a), also determined higher Q10s than in the present study. In most of the other experimental barrels, carbon amendment (glucose) increased the denitrification activity (Fig. 4), as reported by Warneke et al. (2011a) for a field-scale woodchip bioreactor. Furthermore, NO 3 removal and the denitrification rate (DR þ N; removing NO 3 limitation) were found to increase linearly with the availability of carbon (measured as respirable carbon, Fig. 3). Therefore, nitrate removal in the experimental barrels was most likely limited by carbon availability, except for cold maize cobs and cold wheat straw barrels. Nitrate removal in cold maize cobs and cold wheat straw barrels was limited by NO 3 likely due to low NO3 eN outlet concentrations 1 below 1 mg L (Fig. 4; Table 1). These findings confirm that in anaerobic, NO 3 rich environments, carbon limits microbial denitrification (Knowles, 1982; Reddy et al., 1982). This study shows that respirable carbon measurements could also be used to make comparative estimations of NO 3 removal in carbon limited systems (Fig. 3). In this study, the pH decreased slightly from inlet to outlet as found in other studies (Van Driel et al., 2006; Robertson et al., 2005; Robertson and Merkley, 2009), but was still in the optimal range for denitrifiers (Bremner and Shaw, 1958; Knowles, 1982). In contrast Warneke et al. (2011a) reported an increase in pH along the length of a field-scale woodchip bioreactor. DO concentrations decreased from above 6 mg L1 at the inlet, to below 2 mg L1 at the outlet. Robertson (2010) also
measured a similar decrease in DO in a woodchip column study and that a substantial portion of microbially available carbon was consumed by aerobic respiration. However, Gibert et al. (2008) measured declines in DO from 4 to 1.2 mg L1 in the first 10 cm of a 90 cm long woodchip column. This finescale work suggested that most of the substrate close to the inlet served to provide anaerobic conditions for denitrifiers. The NO 3 removal rate was significantly correlated to the copy number of nitrite reductase genes (nirS and nirK ) (Fig. 5). Furthermore the average nitrite reductase gene copies per gram dry substrate increased 4-fold with a temperature increase of 10 C (Fig. 6A), but the NO 3 eN removal rate increased 1.2 times. This temperature dependence of denitrification genes corresponds with seasonal measurements of nitrite reductase gene copies in wetlands (Chon et al., 2010). The copies of 16S rRNA genes also increased with temperature, with the exception of the sawdust barrel (Table 2), so the greater copy number of denitrification genes in the substrate at higher temperature was also partially due to an increase in bacterial biomass.
4.3.
Denitrifying bacterial communities
Abundance of nirS, nirK and nosZ genes in maize cob, green waste, sawdust and wheat straw ranged from 107 to 1011 copies g1 dry substrate (Table 1, Fig. 6A), and these values were on average greater than those measured in constructed wetlands or rice fields (Chon et al., 2010; Yoshida et al., 2009). However, the abundance of denitrification genes in pine and eucalyptus woodchips were slightly lower, but in the same range as the wetland and rice field studies (Chon et al., 2010; Yoshida et al., 2009). But woodchips, especially those from cold incubations, showed the greatest abundance of nitrite reductase genes as a proportion of total bacterial DNA (16S rRNA), coupled with low 16S rRNA gene copies (Table 2, Fig. 6B). Green waste and maize cobs, particularly cold incubated maize cobs, had a low copy number of denitrification genes as a proportion of total bacteria, and gave high 16S rRNA gene copies (Table 2, Fig. 6B). Therefore, the bacterial community in green waste and maize cob barrels had a low ratio of denitrifying genes per copy number of 16S rRNA genes even though green waste and maize cobs had on average more denitrifiers per gram substrate than woodchips (Table 2, Fig. 6). Consequently, a substantial proportion of carbon in green waste and maize cob barrels was likely consumed by non-denitrifying bacteria, fungi and/or yeasts, whereas a greater proportion of C released from woodchips appeared to be consumed by denitrifiers. The ratios of nirS/nirK, and Snir/nosZ, were similar between replicate barrels in cold incubations, except for pine wood barrels (Fig. 7). In warm incubations, there was much greater variation in replicates, and the ratios of nirS/nirK, and nir/nosZ, varied significantly among carbon substrates (Fig. 7). Therefore we assume that it was likely that the composition of denitrifying bacteria in replicate barrels under cold incubation was very similar, but in warm barrels the denitrifying population varied greatly between replicates. Furthermore it is likely that the composition of denitrifier was also very distinct in different carbon substrates, in both warm and cold barrels.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
At warm temperatures, the nirS/nirK ratio increased (except in one pine woodchip barrel), suggesting that higher temperature enhanced growth of nirS containing bacteria, or did not encourage the growth of nirK containing bacteria (Fig. 7). The nirS/nirK has been shown to be greater in unfertilized soils, compared to those that were fertilized (Hallin et al., 2009). The ratio also decreased with the presence of cattle and increased with increasing nitrate, pH and soil moisture (Philippot et al., 2009). Similar temperature dependence was observed with the nitrite reductase/nitrous oxide reductase gene ratio (Snir/nosZ ). The Snir/nosZ was significantly higher in warm barrels than in cold barrels (Fig. 7). This finding corresponded with the higher N2O concentrations in warm barrels compared to cold barrels, and the observed increase in N2O emission at higher temperatures in previous studies (Warneke et al., 2011a; Teiter and Mander, 2005; Johansson et al., 2003). High N2O fluxes have been shown to correlate with a low ratio of nosZ/narG, where narG is the gene responsible for nitrate reduction the first step in the denitrification pathway (Philippot et al., 2009). Similarly, a high ratio of N2O/N2O þ N2 has also been shown to correlate with the Snir/nosZ ratio (Cuhel et al., 2010).
4.4.
Evaluation of the different carbon substrates
Maize cobs, wheat straw and green waste barrels removed more NO 3 than wood substrates. The dissolved N2OeN production of maize cobs, green waste and wood-filled barrels was moderate and the dissolved N2OeN outlet concentrations ranged from 7 to 110 mg L1 for cold barrels, and from 207 to 566 mg L1 for warm barrels. Wheat straw produced on average about three times more dissolved N2O (Fig. 8) than other carbon substrates. This corresponded with the relatively high ratio of nitrite reductase gene copies to P nitrous oxide reductase gene copies ( nir/nosZ ) in the wheat straw barrels (Fig. 7), which lead likely to more N2O production than N2O consumption. The N2OeN release from wheat straw in the effluent was almost 10% of the removed NO 3 eN, which is also about three times greater than the dissolved N2OeN release of a field-scale wood chip denitrification bed (Warneke et al., 2011a). Only sawdust showed no N2O release. Maize cobs had the highest NO 3 removal rate and were concentration. Therefore, additionally limited by NO 3 a higher NO 3 removal rate could be expected for maize cobs if it was loaded with more NO 3 as shown by Cameron and Schipper (2010). However, maize cobs also released high concentrations of TOC and dissolved CH4. It would be expected that CH4 release from maize cobs in the outlet water would decrease with a higher NO 3 concentration in inlet water because denitrification would outcompete methanogenesis. Additionally maize cobs had a low denitrifier/ bacteria ratio, which would probably yield substantial carbon loss due to carbon consumption by non-denitrifiers, whereas woodchips seemed to be an ideal substrate for denitrifying bacteria. Furthermore, wood substrate showed moderate NO 3 removal rates, with almost no adverse effects. As demonstrated in previous studies (Warneke et al., 2011a; Schipper et al., 2010; Robertson, 2010; Long et al., 2011)
5473
woodchips provide sustained NO 3 removal due to slow decomposition of wood in the bioreactor.
5.
Conclusions
This study suggested that microbial denitrification was the main mechanism for nitrate removal for all carbon sources tested, due to the high in vitro DR, the linear relationship between NO 3 removal and in vitro DR þ C/N, high abundance of nitrite reductase genes, and uniformly low NHþ 4 concentrations. The denitrification process in the experimental barrels was limited by carbon availability and temperature, except when 1 NO 3 eN outlet concentrations were below 1 mg L , when eN limitation occurred. The NO eN removal rate was NO 3 3 dependent on the quantity of microbially available carbon, which varied between carbon sources. Both the acetylene inhibition method for measuring denitrification activity, and the quantification of denitrification genes were good approaches for determining comparative NO 3 removal in carbon limited systems (Figs. 3 and 5). It would be useful to determine and compare the slope of the linear regressions between NO 3 removal and Ln (Snir g1 substrate) in different ecosystems to estimate the nitrate removal rates only by the copy number of nitrite reductase genes in similar ecosystems (Fig. 5). Greatest dissolved N2O release in the outlet water was detected for wheat straw and was about 10% of the removed NO 3 eN, which was much greater than reported in previous studies for wood substrates. Methanogenesis could compete with denitrification when NO 3 eN concentrations were below P 2 mg L1 and nir/nosZ ratio was high. Maize cobs had the highest NO 3 eN removal rate, but released elevated amounts of TOC, and substantial carbon consumption by non-denitrifiers was likely. Wood substrates removal, and exhibited moderate and sustained NO 3 appeared to be ideal for denitrifiers under anaerobic, high NO 3 conditions. Therefore it may be useful to combine maize cobs with woodchips, to enhance C availability and increase the denitrifying activity in the woodchip material. This approach would possibly generate higher NO 3 eN removal rates than woodchips alone, with only moderate adverse effects. Furthermore, findings in this study suggest that increased temperatures enhance the growth of nirS-containing and nosZ-lacking bacteria, but further research is needed to understand this effect.
Acknowledgments This research was made possible by funding from WaikatoLink Ltd. (New Zealand) and Hans-Sauer-Foundation (Germany). Additional support was provided by grant number 5 P42 ES04699 from the National Institute of Environmental Health Sciences, NIH. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIEHS, NIH. Many thanks to the technicians of the Science and Engineering department of the University of Waikato for their help.
5474
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
references American Public Health Association, 1992. Standard Methods for the Examination of Water and Wastewater, eighteenth ed. USA. Barton, L., Schipper, L.A., Smith, C.T., 2000. Denitrification enzyme activity is limited by soil aeration in a wastewaterirrigated forest soil. Biol. Fertil. Soils 32, 385e389. Braker, G., Fesefeldt, A., Witzel, K.P., 1998. Development of PCR primer systems for amplification of nitrite reductase genes (nirK and nirS ) to detect denitrifying bacteria in environmental samples. Appl. Environ. Microbiol. 64, 3769e3775. Bremner, J.M., Shaw, K., 1958. Denitrification in soil: II. Factors affecting denitrification. J. Agr. Sci. 51, 40e52. Cameron, S.C., Schipper, L.A., 2010. Nitrate removal and hydraulic performance of carbon substrates for potential use in denitrification beds. Ecol. Eng. 36, 1588e1595. Canfield, D.E., Glazer, A.N., Falkowski, P.G., 2010. The evolution and future of earth’s nitrogen cycle. Science 330, 192e196. Cheneby, D., Hartmann, A., Henault, C., Topp, E., Germon, J.C., 1998. Diversity of denitrifying microflora and ability to reduce N2O in two soils. Biol. Fertil. Soils 28, 19e26. Cheneby, D., Perrez, S., Devroe, C., Hallet, S., Couton, Y., Bizouard, F., Iuretig, G., Germon, J.C., Philippot, L., 2004. Denitrifying bacteria in bulk and maize-rhizospheric soil: diversity and N2O-reducing abilities. Can. J. Microbiol. 50, 469e474. Cheng, W., Coleman, D.C., 1989. A simple method for measuring CO2 in a continuous airflow system: modifications to the substrate-induced respiration technique. Soil Biol. Biochem. 21, 385e388. Chon, K., Kim, K., Chang, N.I., Cho, J., 2010. Evaluating wastewater stabilizing constructed wetland, through diversity and abundance of the nitrite reductase gene nirS, with regard to nitrogen control. Desalination 264, 201e205. Cuhel, J., Simek, M., Laughlin, R.J., Bru, D., Cheneby, D., Watson, C.J., Philippot, L., 2010. Insights into the effect of soil pH on N(2)O and N(2) emissions and denitrifier community size and activity. Appl. Environ. Microbiol. 76, 1870e1878. Dambreville, C., Hallet, S., Nguyen, C., Morvan, T., Germon, J.-C., Philippot, L., 2006. Structure and activity of the denitrifying community in a maize-cropped field fertilized with composted pig manure or ammonium nitrate. FEMS Microbiol. Ecol. 56, 119e131. Della Rocca, C., Belgiorno, V., Meric, S., 2005. Cotton-supported heterotrophic denitrification of nitrate-rich drinking water with sand filtration post treatment. Water SA 31, 1022e1028. Della Rocca, C., Belgiorno, V., Meric, S., 2006. Heterotrophic/ autotrophic denitrification (HAD) of drinking water: prospective use for permeable reactive barrier. Desalination 210, 194e204. Duncan, C.P., Groffmann, P.M., 1994. Comparing microbial parameters in natural and constructed wetlands. J. Environ. Qual. 23, 298e305. Elgood, Z., Robertson, W.D., Schiff, S.L., Elgood, R., 2010. Nitrate removal and greenhouse gas production in a stream-bed denitrifying bioreactor. Ecol. Eng. 36, 1575e1580. Enwall, K., Philippot, L., Hallin, S., 2005. Activity and composition of the denitrifying bacterial community respond differently to long-term fertilization. Appl. Environ. Microbiol. 71, 8335e8343. Firestone, M.K., Davidson, E.A., 1989. Microbiological basis of NO and N2O production and consumption in soil. In: Exchange of Trace Gases Between Terestrial Ecosystems and the Atmosphere. John Wiley, USA, pp. 7e21. Gibert, O., Pomierny, S., Rowe, I., Kalin, R.M., 2008. Selection of organic substrates as potential reactive materials for use in
a denitrification permeable reactive barrier (PRB). Bioresour. Technol. 99, 7587e7596. Greenan, C.M., Moorman, T.B., Kaspar, T.C., Parkin, T.B., Jaynes, D.B., 2006. Comparing carbon substrates for denitrification of subsurface drainage water. J. Environ. Qual. 35, 824e829. Greenan, C.M., Moorman, T.B., Parkin, T.B., Kaspar, T.C., Jaynes, D.B., 2009. Denitrification in wood chip bioreactors at different water flows. J. Environ. Qual. 38, 1664e1671. Groffmann, P.M., Gold, A.J., Simmons, R.C., 1992. Nitrate dynamics in riparian forests: microbial studies. J. Environ. Qual. 21, 666e671. Hallin, S., Jones, C.M., Schloter, M., Philippot, L., 2009. Relationship between N-cycling communities and ecosystem functioning in a 50-year-old fertilization experiment. ISME J. 3, 597e605. Henry, S., Baudoin, E., Lopez-Gutierrez, J.C., Martin-Laurent, F., Brauman, A., Philippot, L., 2005. Quantification of denitrifying bacteria in soils by nirK gene targeted real-time PCR (vol. 59, pp. 327, 2004). J. Microbiol. Methods 61, 289e290. Henry, S., Bru, D., Stres, B., Hallet, S., Philippot, L., 2006. Quantitative detection of the nosZ gene, encoding nitrous oxide reductase, and comparison of the abundances of 16S rRNA, narG, nirK, and nosZ genes in soils. Appl. Environ. Microbiol. 72, 5181e5189. Howarth, R.W., Sharpley, A.W., Walker, D., 2002. Sources of nutrient pollution to coastal waters in the United States: implications for achieving coastal water quality goals. Esturaries 25, 656e676. Johansson, A.E., Kasimir Klemedtsson, A., Klemedtsson, L., 2003. Nitrous oxide exchanges with the atmosphere of a constructed wetland treating wastewater. Tellus 55B, 737e750. Kandeler, E., Deiglmayr, K., Tscherko, D., Bru, D., Philippot, L., 2006. Abundance of narG, nirS, nirK, and nosZ genes of denitrifying bacteria during primary successions of a glacier foreland. Appl. Environ. Microbiol. 72, 5957e5962. Knowles, R., 1982. Denitrification. Microbiol. Rev. 46, 43e70. Lindsay, A.E., Colloff, M.J., Gibb, N.L., Wakelin, S.A., 2010. The abundance of microbial functional genes in grassy woodlands is influenced more by soil nutrient enrichment than by recent weed invasion or livestock exclusion. Appl. Environ. Microbiol. 76, 5547e5555. Long, L.M., Schipper, L.A., Bruesewitz, D.A., 2011. Long-term nitrate removal in a denitrification wall. Agr. Ecosyst. Environ. 140, 514e520. Lopez-Gutierrez, J.C., Henry, S., Hallet, S., Martin-Laurent, F., Catroux, G., Philippot, L., 2004. Quantification of a novel group of nitrate-reducing bacteria in the environment by real-time PCR. J. Microbiol. Methods 57, 399e407. Moorman, T.B., Parkin, T.B., Kaspar, T.C., Jaynes, D.B., 2010. Denitrification activity, wood loss, and N2O emissions over 9 years from a wood chip bioreactor. Ecol. Eng. 36, 1567e1574. Philippot, L., 2002. Denitrifying genes in bacterial and Archaeal genomes. Biochim. Biophys. Acta 1577, 355e376. Philippot, L., Cuhel, J., Saby, N.P.A., Cheneby, D., Chronakova, A., Bru, D., Arrouays, D., Martin-Laurent, F., Simek, M., 2009. Mapping field-scale spatial patterns of size and activity of the denitrifier community. Environ. Microbiol. 11, 1518e1526. Phoenix, G.K., Hicks, W.K., Cinderby, S., Kuylenstierna, J.C.I., Stock, W.D., Dentener, F.J., Giller, K.E., Austin, A.T., Lefroy, R. D.B., Gimeno, B.S., Ashmore, M.R., Ineson, P., 2006. Atmospheric nitrogen deposition in world biodiversity hotspots: the need for a greater global perspective in assessing N deposition impacts. Glob. Change Biol. 12 (3), 470e476. Rabalais, N.N., 2002. Nitrogen in aquatic ecosystems. Ambio 31 (2), 102e112.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 6 3 e5 4 7 5
Reddy, K.R., Rao, P.S.C., Jessup, R.E., 1982. The effect of Carbon mineralization on denitrification kinetics in mineral and organic soils. Soil Sci. Soc. Am. J. 46, 62e68. Robertson, W.D., Merkley, L.C., 2009. In-stream bioreactor for agricultural nitrate treatment. J. Environ. Qual. 38, 230e237. Robertson, W.D., Ford, G.I., Lombardo, P.S., 2005. Wood-based filter for nitrate removal in septic systems. Am. Soc. Agr. Eng. 48 (1), 121e128. Robertson, W.D., Vogan, J.L., Lombardo, P.S., 2008. Nitrate removal rates in a 15-year old permeable reactive barrier treating septic system nitrate. Ground Water Monit. Remediat. 28, 65e72. Robertson, W.D., 2010. Nitrate removal rates in woodchip media of varying age. Ecol. Eng. 36, 1581e1587. Saliling, W.J.B., Westerman, P.W., Losordo, T.M., 2007. Wood chips and wheat straw as alternative biofilter media for denitrification reactors treating aquaculture and other wastewaters with high nitrate concentrations. Aquacult. Eng. 37, 222e233. Schipper, L.A., Barkle, G.F., Vojvodic-Vukovic, M., 2005. Maximum rates of nitrate removal in a denitrification wall. J. Environ. Qual. 34, 1270e1276. Schipper, L.A., Robertson, W.D., Gold, A.J., Jaynes, D.B., Cameron, S. C., 2010. Denitrifying bioreactors e An approach for reducing nitrate loads to receiving waters. Ecol. Eng. 36, 1532e1543. Seitzinger, S., Harrison, J.A., Bo¨hlke, J.K., Bouwman, A.F., Lowrance, R., Peterson, B., Tobias, C., Van Drecht, G., 2006. Denitrification across landscapes and waterscapes: a synthesis. Ecol. Appl. 16, 2064e2090. Shao, L., Xu, Z.X., Jin, W., Yin, H.L., 2008. Rice husk as carbon source and bilofilm carrier for water denitrification. Polish J. Environ. Stud. 18 (4), 693e699. Soares, M.I.M., Abeliovich, A., 1998. Wheatstraw as substrate for water denitrification. Water Res. 32 (12), 3790e3794. Stres, B., Mahne, I., Avgustin, G., Tiedje, J.M., 2004. Nitrous oxide reductase (nosZ ) gene fragments differ between native and cultivated Michigan soils. Appl. Environ. Microbiol. 70, 301e309.
5475
Teiter, S., Mander, U., 2005. Emissions of N2O, N2, CH4 and CO2 from constructed wetlands for wastewater treatment and from riparian buffer zones. Ecol. Eng. 25, 528e541. Tiedje, J.M., Simkins, S., Groffmann, P.M., 1989. Perspectives on measurement of denitrification in the field including recommended protocols for acetylene based methods. Plant Soil 115, 261e284. Van Driel, P.W., Robertson, W.D., Merkley, L.C., 2006. Denitrification of agricultural drainage using wood based reactors. Am. Soc. Agr. Biol. Eng. 49 (2), 565e573. Volokita, M., Abeliovich, A., Soares, M.I.M., 1996a. Denitrification of groundwater using cotton as energy source. Water Sci. Technol. 34, 379e385. Volokita, M., Belkin, S., Abeliovich, A., Soares, M.I.M., 1996b. Biological denitrification of drinking water using newspaper. Water Res. 30, 965e971. Warneke, S., Schipper, L.A., Bruesewitz, D.A., McDonald, I., Cameron, S., 2011a. Rates, controls and potential adverse effects of nitrate removal in a denitrification bed. Ecol. Eng. 37, 511e522. Warneke, S., Schipper, L.A., Bruesewitz, D.A., Baisden, T.W., 2011b. A comparison of different approaches for measuring denitrification rates in a nitrate removing bioreactor. Water Res.. doi:10.1016/j.watres.2011.05.027. Weiss, R.F., Price, B.A., 1980. Nitrous oxide solubility in water and seawater. Mar Chem. 8, 347e359. Wood, D.W., et al., 2001. The genome of the natural genetic engineer Agrobacterium tumefaciens C58. Science 294, 2317e2323. Yamamoto, S., Alcauskas, J.B., Crozier, T.E., 1976. Solubility of methane in distilled water and seawater. J. Chem. Eng. Data 21, 78e80. Yoshida, M., Ishii, S., Otsuka, S., Senoo, K., 2009. Temporal shifts in diversity and quantity of nirS and nirK in a rice paddy field soil. Soil Biol. Biochem. 41, 2044e2051. Zumft, W., 1997. Cell biology and molecular basis of denitrification. Microbiol. Mol. Biol. Rev. 61, 533e616.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Fine-scale bacterial community dynamics and the taxaetime relationship within a full-scale activated sludge bioreactor George F. Wells a, Hee-Deung Park a,1, Brad Eggleston b, Christopher A. Francis c, Craig S. Criddle a,* a
Department of Civil and Environmental Engineering, Stanford University, Stanford, CA 94305, USA Palo Alto Regional Water Quality Control Plant, 2501 Embarcadero Way, Palo Alto, CA 94303, USA c Department of Environmental Earth System Science, Stanford University, Stanford, CA 94305, USA b
article info
abstract
Article history:
In activated sludge bioreactors, aerobic heterotrophic communities efficiently remove
Received 27 April 2011
organics, nutrients, toxic substances, and pathogens from wastewater, but the dynamics of
Received in revised form
these communities are as yet poorly understood. A macroecology metric used to quantify
18 July 2011
community shifts is the taxaetime relationship, a temporal analog of the speciesearea
Accepted 6 August 2011
curve. To determine whether this metric can be applied to full-scale bioreactors, activated
Available online 16 August 2011
sludge samples were collected weekly over a one-year period at a local municipal wastewater treatment plant. Bacterial community dynamics were evaluated by monitoring 16S
Keywords:
rRNA genes using Terminal Restriction Fragment Length Polymorphism (T-RFLP), corrob-
Activated sludge
orated by clone libraries. Observed taxa richness increased with time according to a power
Microbial community dynamics
law model, as predicted by macroecological theory, with a power law exponent of
Taxaetime relationship
w ¼ 0.209. The results reveal strong long-term temporal dynamics during a period of stable
T-RFLP
performance (BOD removal and nitrification). Community dynamics followed a gradual
Multivariate statistics
succession away from initial conditions rather than periodicity around a mean “equilib-
16S rRNA
rium”, with greater within-month then among-month community similarities. Changes in community structure were significantly associated via multivariate statistical analyses with dissolved oxygen, temperature, influent silver, biomass (MLSS), flow rate, and influent nitrite, cadmium and chromium concentrations. Overall, our results suggest patterns of bacterial community dynamics likely regulated in part by operational parameters and provide evidence that the taxaetime relationship may be a fundamental ecological pattern in macro- and microbial systems. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biological wastewater treatment is, in essence, an endeavor in microbial ecological engineering. The goal of this endeavor is to manage microbial communities for the good of society by
promoting degradation of oxygen-depleting organics, transformation of toxic substances, and removal of nutrients from water; thus, a firm understanding of the microbial ecology of wastewater treatment bioreactors is essential (Rittmann et al., 2006). Performance changes observed at the treatment plant
* Corresponding author. Environmental Engineering and Science Program, Department of Civil and Environmental Engineering, Yang & Yamazaki Environment & Energy Bld., Rm. 151, 473 Via Ortega MC-4020, Stanford University, Stanford, CA 94305, USA. Tel.: þ1 650 723 9032; fax: þ1 650 725 3164. E-mail address: [email protected] (C.S. Criddle). 1 Present address: School of Civil, Environmental and Architectural Engineering, Korea University, Seoul 136-713, South Korea. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.006
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
scale (hundreds of cubic meters) are the emergent properties of an incredibly diverse assembly of w1018 individual microbial cells (1e2 cubic micrometers) (Curtis et al., 2003). Many of the critical process failures (i.e. unacceptable reduction in final effluent quality) that arise during biological wastewater treatment are likely attributable to variations in the relative abundance or activity of these cells (Graham and Smith, 2004). Such variations in microbial community structure are thought to be influenced by a combination of deterministic (reactor design, environmental and operational variables) and stochastic (probability of microbial dispersal into or out of a reactor) properties (Curtis and Sloan, 2006). Critical questions important to rational design and operation of these systems remain unanswered (Rittmann et al., 2006): What community structures, and what range of community dynamics, are optimal for different applications? What environmental conditions trigger optimal community assembly with its desired function? Are the conditions and outcomes predictable, reproducible, and controllable? Understanding, and eventually predicting, dynamics in community composition and its relationship to ecosystem function is one of the key engineering problems that might be solved through a more quantitative understanding of the microbial ecology of wastewater treatment processes (Gentile et al., 2007; Graham and Smith, 2004). Removal of COD is catalyzed in biological wastewater treatment plants (WWTPs) by activated sludge communities composed of a diverse assemblage of chemoorganoheterotrophic microorganisms. Numerous studies in recent years have characterized activated sludge microbial community structure via culture-independent studies (Wagner et al., 2002; Xia et al., 2010), and others have investigated succession and dynamics in these communities, particularly within specific microbial subpopulations such as nitrifiers (Wells et al., 2009), phosphorus-accumulating organisms (Slater et al., 2010), denitrifiers (Gentile et al., 2007), and methanogens (Fernandez et al., 1999). Many of these studies have occurred in lab-scale systems, however, where selective pressures likely differ dramatically from those in full-scale plants (Seviour and Nielsen, 2010). Indeed, few studies have directly characterized long-term, finescale microbial population dynamics in activated sludge from full-scale WWTPs. Moreover, we lack a quantitative understanding of the relative influence of specific environmental or operational drivers of community dynamics in full-scale systems, and it is not clear whether these dynamics are random or predictable. Activated sludge bioreactors are excellent test beds for fundamental questions in microbial ecology (Daims et al., 2006). A longstanding area of inquiry in ecology as a whole focuses on patterns of generation and maintenance of species diversity through time and space. In macroecology, spatial patterns in species diversity have long been recognized to follow a SpecieseArea Relationship (SAR) of the form S ¼ cAz, where S is species richness, A is the spatial scale of observation, c is an empirically derived constant, and z is a scaling exponent that reflects species turnover. Accumulation of microbial taxa richness with increasing spatial scales follow a similar pattern (Horner-Devine et al., 2004). Conversely, the speciesetime relationship (herein referred to as the taxaetime relationship [TTR]) has received comparatively little
5477
attention. First proposed by Preston (1960), the TTR describes the increase in observed taxa richness, S, for increasing time of observation, T, by a power law model, similar in form to the SAR: S ¼ cTw, where the scaling exponent w is a reflection of species turnover. The TTR has been documented in a rapidly expanding literature in macroecology (White et al., 2006), and has very recently been tested in a limited number of microbial systems (Redford and Fierer, 2009; van der Gast et al., 2008). The applicability of the power-law TTR to full-scale activated sludge bioreactors, and a comparison of the associated temporal scaling exponent (w) to previously reported values, has not yet been explored. The primary objectives of this study were thus two-fold: 1) to monitor long-term, fine-scale temporal dynamics in the overall bacterial community structure in activated sludge and identify operational or environmental factors that most significantly correlate with those dynamics; and 2) to assess the applicability of a power law taxaetime relationship to activated sludge microbial communities in full-scale bioreactors. To this end, we collected weekly samples and concurrently monitored 20 operational or environmental variables over a 1-year period at a well-mixed, full-scale municipal wastewater treatment bioreactor. We analyzed shifts in bacterial community structure in this time series via 16S rRNA gene-based Terminal Restriction Fragment Length Polymorphism (T-RFLP) corroborated by clone libraries. We then employed multivariate statistical tools to identify explanatory variables most significantly associated with these dynamics, and we explored the relationship between time and taxa in this critical application of environmental biotechnology.
2.
Materials and methods
2.1.
Sample collection and site description
The Palo Alto Regional Water Quality Control Plant (PARWQCP) is a WWTP located in Palo Alto, California (37.2631 N, 122.0835 W) that treats primarily municipal wastewater (100,000e170,000 m3 day1) for a population of w225,000 people. Effluent is released to San Francisco Bay after primary and secondary treatment (with nitrification), tertiary treatment (dual media filtration), and disinfection. Approximately 65% of total BOD removal is accomplished in two trickling filters operated as roughing reactors; minimal ammonia is removed in this unit. Immediately downstream of the trickling filters are four parallel 6880 m3 aeration basins comprising the activated sludge bioreactor, operated for removal of remaining organics and for nitrification. The average hydraulic residence time (HRT) in the activated sludge bioreactor during the sampling period was 6.2 h, and the solids residence time (SRT) was 21 days. The aeration basins are well mixed with respect to microbial community structure, based on a baseline study in which biomass from three locations within the bioreactor yielded essentially identical microbial community composition via 16S rRNA T-RFLP (data not shown). For a one-year period from February 2005 to February 2006, a Sigma 900 refrigerated automatic sampler (Hach, Loveland, CO) was used to obtain 24-h composite
5478
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
activated sludge samples (100 ml every 30 min) weekly from the combined outlet of all four aeration basins. 1.5 ml composite activated sludge samples were centrifuged onsite for 3 min at 10,000 g, decanted, and stored at 20 C. Concurrently, 20 operational and environmental variables (influent flow rate; reactor temperature; reactor MLSS; dissolved oxygen (DO); influent and effluent COD, ammonium, and nitrite; and primary influent chromium, nickel, copper, zinc, silver, cadmium, lead, arsenic, selenium, and mercury) were monitored, as described previously (Wells et al., 2009).
2.2. DNA extraction and PCR amplification of the 16S rRNA gene Activated sludge from 53 weekly samples (1 February 2005 to 2 February 2006) and a baseline sample from 29 September 2004 were washed using 1 ml TriseEDTA (TE) buffer (1 mM, pH ¼ 7.0), concentrated at 10,000 g for 3 min, decanted, and resuspended in 200 ml TE buffer. Duplicate genomic DNA extractions from each sampling date were performed with the MoBio Ultraclean Soil DNA Extraction Kit (Carlsbad, CA) as per the manufacturer’s directions, with the exception that the initial bead-beating step was performed with a MiniBeadbeater (Biospec Products, Bartlesville, OK) at 5000 rpm for 1 min. Near full-length fragments (w1350 bp) of the 16S rRNA gene were amplified using bacterial-specific primer set 8F (50 -AGAGTTTGATCCTGGCTCAG-30 ) and 1392R (50 ACGGGCGGTGTGTRC-30 ). For T-RFLP analyses, the forward primer was labeled with the fluorophore 6-carboxyfluorescein (FAM). Each 50 ml PCR mixture consisted of 0.25 mM of each primer, 1 Fail-Safe PCR buffer F (Epicentre, Madison, WI), 1.25 U AmpliTaq LD Taq Polymerase (Applied Biosystems, Foster City, CA), and 20e50 ng of genomic DNA. The PCR temperature profile was as follows: 94 C for 5 min, then 35 cycles of 94 C for 45 s, 55 C for 30 s, and 72 C for 90 s, followed by a final extension at 72 C for 10 min. Amplicon presence and quality was verified via 1.5% agarose gel electrophoresis.
2.3.
Cloning, sequencing, and phylogenetic analyses
16S rRNA gene clone libraries were generated from four dates of sampling (29 September 2004, 9 June 2005, 10 August 2005, and 1 December 2005) to facilitate in silico choice of restriction enzymes for T-RFLP and to provide a coarse baseline phylogenetic analysis of activated sludge microbial community structure. Triplicate PCR products were pooled and purified via the QIAEX II gel extraction kit (Qiagen, Valencia, CA) and cloned using the TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA) or pGEM-T Easy Vector System with JM109 competent E. Coli cells (Promega, Madison, WI), as per the manufacturer’s instructions. Randomly picked clones were sequenced using T7 and SP6 or M13 primers on ABI 3100 or 3730 automated sequencers by MCLab (San Francisco, CA) or Elim Biopharmaceuticals, Inc. (Hayward, CA), generating 172 near full-length 16S rRNA gene sequences. Putative chimeric sequences identified using BELLEROPHON (Huber et al., 2004) were removed (47 total). Remaining 16S rRNA gene sequences were aligned via the NAST utility (http://greengenes.lbl.gov/ NAST) (DeSantis et al., 2006) and inserted via parsimony to an existing phylogenetic tree in ARB v5.1 (Ludwig et al., 2004)
containing all 236,469 sequences in the greengenes database as of 18 November 2008. Following manual adjustment of the alignment and removal of 15 sequences displaying low identity (<90%) to the closest related Genbank sequence, phylogenetic reconstruction via the neighbor-joining method was performed in ARB for the 110 remaining sequences with the Jukes-Cantor correction and Lane mask. A neighbor-joining bootstrap analysis with 1000 replicates was also performed in ARB under the same conditions. Consensus phylogenetic assignments of 16S rRNA gene sequences were performed by manual inspection of the phylogenetic tree and via the online classifier at http://greengenes.lbl.gov/cgi-bin/nph-classify.cgi by separately employing the RDP, NCBI, Hugenholtz, and G2_Chip taxonomic nomenclatures. Nonchimeric 16S rRNA gene sequences were submitted to Genbank and assigned accession numbers HQ385507eHQ385631.
2.4.
T-RFLP analyses
Temporal dynamics in activated sludge bacterial community structure were monitored via Terminal Restriction Fragment Length Polymorphism (T-RFLP) analyses targeting the 16S rRNA gene. T-RFLP PCR conditions were identical to endpoint PCR. Duplicate PCR products from each of 53 weeks of sampling were pooled and purified using Montage PCR filter units (Millipore, Billerica, MA). Purified products (400 ng/ digest) were digested in separate reactions with 0.25 U/ml of RsaI or MspI restriction endonuclease (Promega, Madison, WI) in buffer provided by the manufacturer for 3 h at 37 C, providing two independent fingerprints of the bacterial community structure for each time point. Digests were repurified with Montage PCR filter units, loaded onto a capillary electrophoresis machine (ABI 310 genetic analyzer, Foster City, CA) with the MapMarker 1000 internal size standard (BioVentures, Murfreesboro, TN), and analyzed at the Genomics Technology Support Facility at Michigan State University. T-RF lengths (bps) and peak heights were extracted from the resulting electropherograms using Genotyper v3.1 (ABI, Foster City, CA) and standardized based on previously described methods (Gentile et al., 2007). Briefly, total fluorescence signal was calculated as the sum of peak heights between 50 and 1000 bp. Profiles were standardized to a total sum of peak heights of 10,000 fluorescence units (FU). Peaks with adjusted heights lower than the threshold of 50 FU were removed, yielding a detection limit of 0.5% relative abundance. The relative of abundance of each OTU (T-RF) was calculated as the ratio of the peak height for that OTU to the sum of peak heights for all T-RFs in a given profile and expressed as a percentage.
2.5.
Statistical analyses
Nonmetric Multidimensional Scaling (NMDS) analyses were performed in PC-ORD v5.0 (MJM Software Design, Gleneden Beach, OR). Pearson productemoment correlation coefficients were calculated in R v2.5.1 (http://www.r-project.org/). Mantel correlograms and Analyses of Similarity (ANOSIM) were performed in PAST v2 (http://folk.uio.no/ohammer/past/). Redundancy Analyses (RDA) were performed in Canoco for Windows v4.53 (Plant Research International, Netherlands).
5479
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
Nonlinear regression analyses were performed in SPSS v19 (IBM, Somers, NY). A detailed discussion of statistical methods is provided in the Supplementary Materials.
3.
Results
3.1.
Bioreactor performance and operational conditions
Concurrent to weekly activated sludge sampling, we monitored 20 operational or environmental variables that we hypothesized may be significant drivers of temporal dynamics in overall bacterial community structure. Variations in these parameters are summarized in Table 1. Generally stable influent flow rates led to maintenance of the hydraulic residence time within a narrow range (6.2 0.6 h). Reactor temperature varied from a summer maximum of 25.4 C in August to a winter minimum of 18.2 C in January. The bioreactor was also operated with relatively high MLSS during the winter of 2005, which led to a strong artificial correlation between biomass and temperature in our time series. Reactor performance was relatively stable across the sampling period. The COD removal rate averaged 57 12%. However, 5 of 53 time points displayed COD removal rates less than 40%, with a minimum of 22% on 24 January 2006. Moderate variation in effluent COD was also observed, from a minimum of 18 mg/L to a maximum of 67 mg/L. Nitrification performance was significantly higher and more stable across the entire year of sampling (95% 3% NHþ 4 removal rate), and accordingly, was low (0.9 0.5 mg N/L). Secondary effluent effluent NHþ 4 NO 2 was similarly maintained at a low and nearly constant
level (0.1 mg NO 2 -N/L). Interestingly, while influent NO2 to the activated sludge bioreactor was low relative to influent NHþ 4 , it was highly variable, ranging from a minimum of 0.05 mg N/L to a maximum of 1.18 mg N/L. The observation of low but significant and highly variable levels of NO 2 in the influent to the PARWQCP bioreactor was unexpected, given the lack of an explicitly designed nitrification function in the upstream trickling filter and routinely negligible NO 2 concentrations in the raw sewage influent to the plant. While aeration basin pH was not routinely monitored as part of this study, 21 separate measurements between December 2004 and March 2005 were nearly constant (6.7 0.1). All 10 metals monitored in the primary influent remained at ppb levels throughout the sampling period, and ranged from 0.4 0.21 mg/l for cadmium to 136 27 mg/l for zinc.
3.2. Phylogenetic analysis of bioreactor bacterial community structure To phylogenetically characterize the bacterial community structure in activated sludge at the PARWQCP and to inform choice of restriction enzymes for T-RFLP analysis, nearly fulllength (w1350 bp) 16S rRNA genes from four time points were cloned and sequenced. Phylogenetic inferences for these sequences are summarized in Fig. 1, based on a neighborjoining phylogenetic tree (Fig. S1) and consensus classifications via the greengenes classification utility. All four dates of sampling were dominated by sequences affiliated with the Proteobacteria. 28% of pooled sequences clustered within the Alphaproteobacteria, while 22% affiliated with the Betaproteobacteria. Ten of the latter sequences (9% of total pooled
Table 1 e Operational conditions and bioreactor performance of the Palo Alto Regional Water Quality Control Plant. Unless otherwise specified, influent and effluent refer to the entry and exit points to the activated sludge bioreactor. Characteristics Wastewater inflow rate,b m3/day Mixed liquor temperature,b C Biomass (mixed liquor suspended solids),c mg/L Dissolved Oxygen,d mg/L Influent chemical oxygen demand (filtered),c mg/L Effluent chemical oxygen demand (filtered),c mg/L c Influent ammonia (NH3 þ NHþ 4 ), mg N/L þ c Effluent ammonia (NH3 þ NH4 ), mg N/L Influent nitrite,c mg N/L Effluent nitrite,c mg N/L Primary Influent Silver,c mg/L Primary Influent Arsenic,c mg/L Primary Influent Cadmium,c mg/L Primary Influent Chromium,c mg/L Primary Influent Copper,c mg/L Primary Influent Mercury,c mg/L Primary Influent Nickel,c mg/L Primary Influent Lead,c mg/L Primary Influent Selenium,c mg/L Primary Influent Zinc,c mg/L a b c d
Rangea
Average
Standard deviation
82,000e181,000 18.2e25.4 2600e4900 3.08e4.50 54e214 18.0e67.0 14.0e33.0 0.20e2.10 0.05e1.18 0.01e0.10 0.6e5.0 0.6e2.2 0.2e1.0 1.8e9.0 43e110 0.12e0.44 5e12 2.6e9.0 0.6e1.3 84e220
105,000 22.4 3340 3.88 88.8 37.0 18.9 0.84 0.41 0.03 1.6 1.1 0.4 4.4 66.6 0.22 7.2 5.0 0.8 135.9
11,500 1.8 520 0.37 22.9 9.9 2.8 0.52 0.22 0.02 0.75 0.26 0.21 1.40 15.2 0.07 1.37 1.53 0.16 26.7
Range indicates minimum to maximum value. data based on daily measurements. data based on weekly measurements. data based on the 24-h average of measurements collected every 15 min in all four aeration basins.
5480
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
Fig. 1 e Phylum or class-level (for Proteobacteria) phylogenetic affiliations for 110 near full-length (w1350 bp) 16S rRNA gene sequences retrieved from four activated sludge sampling time points. The pie chart displays the percentage of clones from the pooled libraries affiliated with each phylogenetic division, and the total number of clones affiliated with each division is shown in parentheses.
sequences) affiliated with the Comamonadaceae, a family of bacteria often reported in wastewater treatment systems (Sadaie et al., 2007; Seviour and Nielsen, 2010). Eight Alphaproteobacterial sequences (7.3% of pooled sequences) were affiliated with the Rhodobacterales order, and five putative sphingomonad sequences (4.5% of pooled sequences) clustered within the Sphingosinicella genus. The large majority (7, or 6.3% of the total) of sequences within the Deltaproteobacteria phyla affiliated with the order Myxococcales, which are gliding bacteria commonly found in soils and activated sludge that are thought to significantly impact biomass carbon turnover due to their role as ‘micropredators’ (Lueders et al., 2006). The aeration basins are operated for simultaneous BOD removal and nitrification and, as expected, we recovered sequence-based evidence for the latter process in this environment. Six percent of sequences clustered within the Nitrospira genus and can thus be considered putative nitriteoxidizing bacteria (NOB). Nitrospira have been found to be the most abundant NOB in most nitrifying wastewater treatment plants (Daims et al., 2006). No sequences affiliated with the Nitrobacter genus, previously viewed as the dominant NOB in activated sludge, were recovered in our clone libraries. Moreover, strong bootstrap support indicated that two sequences (1.8% of total sequences) in the betaproteobacterial grouping clustered within the Nitrosomonas genus and are thus putatively ammonia-oxidizing bacteria (AOB). Seven percent of sequences affiliated with the Bacteroidetes, of which five
sequences (4.5%) clustered in the family Saprospiraceae. Members of the Bacteroidetes phylum are thought to degrade complex organic materials, and recent evidence suggests that many Saprospiraceae are epiphytic protein-hydrolyzers commonly observed attached to filamentous bacteria (Xia et al., 2008). The Actinobacteria and deeply branching Planctomycetes accounted for 6% and 5% of sequences, respectively, and representatives from both phyla have been reported in high numbers in activated sludge (Seviour and Nielsen, 2010). Filamentous Chloroflexi, particularly those associated with the class Anaerolineae as the majority of 9 (8.1%) Chloroflexi clones did in this study, are also relatively cosmopolitan in wastewater treatment processes and have been implicated in sludge bulking events (Nielsen et al., 2009). The phyla Acidobacteria, Verrucomicrobia, and TM7 were each represented by a single sequence (0.9% of the total) in our clone libraries.
3.3.
Tracking bacterial community dynamics
Based on in-silico analyses of a baseline 16S rRNA gene clone library from 29 September 2004 with 16 restriction enzymes, we identified MspI and RsaI as optimal for resolving bacterial community diversity and dynamics in PARWQCP activated sludge via T-RFLP analyses. Previous analyses in the primary literature (Engebretson and Moyer, 2003; Liu et al., 1997) and insilico digests of clone libraries from three separate sampling dates during our year-long time series (9 June 2005, 5 August
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
2005, and 1 December 2005) corroborated our choice of restriction enzymes. These two enzymes were employed in parallel digests to provide an increased level of resolution. 318 total OTUs were observed, of which 192 associated with MspI digestions and the remaining 126 with RsaI analyses (Fig. S2 and S3). Within the concatenated MspI and RsaI T-RFLP datasets, we observed an average of 102 OTUs (standard deviation ¼ 8) in each time point. However, only 12 of these were represented in all 53 time points examined, and less than 10% of OTUs were above the detection limit (>0.5% relative abundance) in 50 or more weeks. Within any given time point, the most abundant OTU (T-RF with highest peak height) averaged 11.5% of the T-RFLP profile (sum of T-RF peak heights). Interestingly, the 12 OTUs represented at all time points were also among the most abundant OTUs we measured; their average relative abundances across the time series ranged from 1.6% to 6.2% of the total fluorescence signal. Our results demonstrate strong variation in the relative abundance of a highly diverse assemblage of ribotypes throughout the 53 week sampling period. To further explore temporal variability within the overall bacterial community, T-RF patterns over the sampling period were assessed via the indirect gradient ordination technique Nonmetric Multidimensional Scaling (NMDS). Community structures in 53 weekly samples from the PARWQCP aeration basin are represented as colored points in the NMDS ordination (Fig. 2). The proximity of two points is an estimate of the pairwise BrayeCurtis similarity index between the two samples. The coefficient of determination (R2) between distances in the original unreduced BrayeCurtis distance matrix and Euclidean distances in ordination space was 0.76, indicating that distances in ordination space are a reasonably
5481
accurate representation of those in unreduced space. A stress value of 18.4 provides additional verification that the reduced NMDS ordinates preserve patterns in activated sludge bacterial community dynamics. In general, samples collected in similar time periods (e.g. the same month) cluster closely together in the ordination. This ordination pattern suggests a gradual succession within the overall bacterial community over time. We constructed a Mantel correlogram to further characterize this successional pattern and assess potential periodicities in our time series (Fig. 3). The correlogram displays average mean similarity (BrayeCurtis distance) between bacterial community structure as a function of the lag time between sampling events. A corresponding Mantel scalogram, which color codes similarities between all pairs of points along the time series, is available in the Supplementary Materials (Fig. S4). A strong negative trend in community similarity as a function of increasing lag time is evident, while periodicity was not detected. This negative trend was confirmed via correlation analysis (Spearman’s rank correlation coefficient ¼ 0.998, p < 0.001). In agreement with the NMDS analysis, this result provides evidence that, rather than undergoing variation (periodicity) around a mean “equilibrium” community structure, activated sludge bacterial community dynamics follow a gradual succession away from initial conditions. To test the hypothesis that within-month T-RFLP profile similarities were greater than among-month similarities, an analysis of similarity (ANOSIM) was conducted. Results are summarized in Table 2. Global ANOSIM revealed strong and significant variation in T-RFLP profiles as a function of month of sampling ( p < 0.001). Moreover, 10 of 11 pairwise ANOSIM tests demonstrated significantly higher within-group than
Fig. 2 e Nonmetric Multidimensional Scaling (NMDS) analysis calculated from normalized peak heights for T-RFLP electropherograms for concatenated MspI and RsaI T-RFs. Circles and associated numbers indicate sampling points analyzed in temporal order (1e4: February 2005, 5e9: March 2005, 10e13: April 2005, 14e17: May 2005, 18e22: June 2005, 23e26: July 2005, 27e30: August 2005, 31e35: September 2005, 36e39: October 2005, 40e43: November 2005, 44e48: December 2005, 49e52: January 2006, 53: February 2006). Circles are color-coded according to month of sampling. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
5482
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
one-year time series between relatively stable COD or NHþ 4 removal efficiencies and bacterial community dynamics. 64 of 66 ANOSIM tests for all pairwise between-month groupings (not limited to consecutive months) indicated significant separation in bacterial community structure at the p ¼ 0.1 level (data not shown). Our analysis reveals successive shifts in bacterial community structure throughout the year of sampling and demonstrates that bacterial community structure is dynamic during periods of relatively stable operation.
3.4. Correlations between operational parameters and bacterial community structure
Fig. 3 e Mantel correlogram displaying mean pairwise similarity (BrayeCurtis distance metric) between bacterial community structures (via T-RFLP profiles) as a function of lag time between activated sludge sampling events.
among-group similarity in T-RFLP profiles at the p ¼ 0.1 level for consecutive months of sampling, although the strength of this association was somewhat weaker than measured global variation. T-RFLP profiles from November and October 2005 generated the sole pairwise assessment for sequential months of sampling that did not indicate significant separation in community structure (R ¼ 0.157, p ¼ 0.267). Interestingly, average COD removal efficiencies during these two months (65.5% 2.0% and 65.3% 7.6%, respectively) were nearly identical and represented the two highest monthly average removal efficiencies over this one-year time series. It is tempting to speculate that overlap in bacterial community structure between October and November is associated with similarities in this performance metric, despite the lack of a significant correlation via RDA or NMDS analyses over our
Table 2 e ANOSIM test for significant differences between monthly groupings in activated sludge overall bacterial community structure, as measured via concatenated MspI and RsaI T-RFLP patterns. Scale or Comparison Global, with months nested February vs. March March vs. April April vs. May May vs. June June vs. July July vs. August August vs. September September vs. October October vs. November November vs. December December vs. January (2006)
We interpreted NMDS ordination axes in terms of explanatory variables by calculating Pearson productemoment correlation coefficients between sample axis scores and 20 measured operational or environmental variables (Table 3). DO concentrations displayed the strongest correlation to both axis 1 and 2. This correlation was positive and highly significant ( p 0.001) for both axes. Axis 2 was also significantly negatively associated with influent silver concentrations, and significantly positively associated with influent arsenic and reactor temperature. To directly assess the influence of monitored operational parameters on temporal dynamics in bacterial community structure, we employed the direct gradient ordination method Redundancy Analysis (RDA). Axes in RDA ordinations are constrained against input environmental variables. The final RDA model is shown in Fig. 4, and results are summarized in Tables S1 and S2. The model explained the majority of variance in the species-environment correlations (60.7%), but only encompassed a minority of the variance in species (T-RF) data (17.5%). Both axes displayed high specieseenvironment correlation values, indicating strong correlations between T-RF patterns and operational data. Of 20 input explanatory variables, 7 were identified as significantly linked to bacterial community variability at the p ¼ 0.05 level. These seven
Table 3 e Significant Pearson productemoment correlation coefficients between NMDS ordination axes and 20 operational or environmental variables. Environmental or Operational Parameters Dissolved Oxygen (r ¼ 0.457, P ¼ 0.001)
Sample Statistic R
P-Value
Axis 1 (R2 ¼ 0.413) Axis 2 (R2 ¼ 0.345)
0.742 0.256 0.205 0.427 0.375 0.563 0.521 0.788 0.488 0.156 0.456 0.408
<0.0001 0.045 0.077 0.055 0.025 0.022 0.058 0.018 0.007 0.267 0.016 0.008
Only parameters significantly correlated at an adjusted P-value of 0.05 after application of the Benjamini and Hochberg correction for multiple comparisons are shown. Correlation coefficients and associated P-values are indicated in parentheses. Parameters in bold were significant at an adjusted P-value of 0.01 after application of the Benjamini and Hochberg correction for multiple comparisons. R2 values beneath axis labels indicate the coefficient of determination for correlations between ordination distances and distances in the original unreduced space. The sum of coefficients of determination for both axes (0.758) is the overall (cumulative) correlation between ordination and unreduced distances.
Dissolved Oxygen (r ¼ 0.625, P ¼ 0.000001) Influent Silver (r ¼ L0.440, P ¼ 0.001) Influent Arsenic (r ¼ 0.377, P ¼ 0.005) Temperature (r ¼ 0.370, P ¼ 0.006)
5483
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
parameters are indicated in Fig. 4 by arrows. Arrow length indicates the strength of the relationship between that parameter and bacterial community composition, and arrow direction indicates the parameter’s approximate direction of increase. The strongest correlate was DO ( p < 0.005), in line with Pearson correlation coefficients to NMDS axes. Also in agreement with the NMDS analysis, reactor temperature and primary influent silver concentration were identified as strong correlates to variation in T-RFLP profiles. Biomass (MLSS) levels, primary influent cadmium, flow rate, and bioreactor influent nitrite concentrations rounded out the explanatory variables included in the final RDA model, in order of addition to the model.
300 200
250
Cumulative Observed OTUs
150
A taxaetime relationship 100
3.5.
Observed Power Fit
To assess the applicability of a taxaetime relationship (TTR) to temporal dynamics in the overall bacterial assemblage in a full-scale activated sludge WWTP, we used nonlinear regression analysis to describe the cumulative observed OTUs (S ) as a function of time of observation (T ) via a power law model, S ¼ cTw. Here, the scaling exponent w can be considered a measure of temporal turnover of microbial taxa, in the same way that the exponent x of an ecological speciesearea relationship is a measure of spatial turnover. We found that the best-fit power law model displayed a strong (R2 ¼ 0.989) and significant ( p < 0.001) fit to our molecular characterization of bacterial community dynamics (Fig. 5). The best-fit TTR exponent was w ¼ 0.209 0.007 (95% confidence interval) for the concatenated MspI and RsaI T-RFLP profiles. Power law models for MspI and RsaI datasets analyzed independently yielded similar scaling exponents (w ¼ 0.213 and 0.203, respectively). The power-law fit to our data was superior to eight other models (linear, logarithmic, inverse, quadratic, cubic, compound, S, growth, exponential, and logistic) tested,
0
100
200
300
Time (days)
Fig. 5 e Taxaetime relationship and least-squares nonlinear regression power law fit between time and cumulative observed OTUs in the activated sludge bioreactor.
as measured by the proportion of variation in the data explained by each model (i.e. the R2 value).
4.
Discussion
0.6
16S rRNA gene clone libraries from four time points revealed a highly diverse community, with sequences clustering within
25
Temperature
27
Axis 2
33 38
39 36
43 DO
6
12 10
44
Influent NO Influent 45 46 Cd
4
3 MLSS 50
September 2005
1
53
5
47
2
8
Flowrate
52
October 2005 November 2005
December 2005
January 2006
11
February 2006
9
49 51
-0.6
48
July 2005 August 2005
IInfluent fl Ag
7
April 2005
June 2005
15 14 13
42
March 2005 May 2005
19 21 18 16
40
37
17
20
30 34 29 28 41 23 22
32 31 35
February 2005
24 26
-0.8
0.8
Axis 1 Fig. 4 e RDA ordination calculated from normalized peak heights for T-RFLP electropherograms of concatenated MspI and RsaI T-RFs. Variables that contributed significant improvement ( p < 0.05) to the explanatory power of the RDA models are indicated by arrows in the ordination. Circles and associated numbers [ samples analyzed in temporal order; Ag [ silver; Cd[Cadmium; MLSS [ Mixed Liquor Suspended Solids. Circles are color-coded according to month of sampling. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
5484
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
9 separate phyla. Xia et al. (2010) employed a 16S rRNA microarray (PhyloChip) to characterize diversity in single time points from five lab or full-scale bioreactors. In line with our results, they found a high representation of Proteobacteria, Actinobacteria, and Bacteroidetes, but also high abundance of Firmicutes. In contrast, we observed no 16S rRNA gene sequences showing high similarity to Firmicutes. This discrepancy may be due to the fact that some of the reactors sampled by Xia and colleagues contained anaerobic or anoxic zones, thereby yielding niches for anaerobic Firmicutes such as the Clostridia, whereas the bioreactor targeted in this study is highly aerated. Nonetheless, the observed bacterial community structure in this study largely mirrored previous observations in activated sludge based on 16S rRNA gene clone library analyses (Seviour and Nielsen, 2010; Wagner et al., 2002), suggesting that results of our investigation into temporal dynamics and associated drivers of bacterial community structure may be widely representative of municipal activated sludge bioreactors. Further investigation is warranted to test this hypothesis, particularly in additional bioreactor process configurations, under varying environmental conditions, and in different geographic locations. As with many ecologically relevant methodologies, interpretation of microbial community fingerprinting methods requires caution and a healthy awareness of methodological limitations. T-RFLP is subject to well-known biases associated with PCR-based methods. However, all samples are subject to the same biases, and it is thus valid to compare between samples on a relative basis after standardization of profiles. Indeed, T-RFLP has been shown to be a highly reproducible and sensitive means of assessing fine-scale changes in microbial community structure at a variety of temporal and spatial scales (Schu¨tte et al., 2008). While it is also possible in some cases to track specific phylogenetic lineages in microbial communities by linking peaks in T-RFLP profiles to in-silico analyses of clone libraries, phylogenetic inference of T-RFs is uncertain in activated sludge due to the highly complex nature of this system. This is an acknowledged limitation of T-RFLP (Marsh and Jared, 2005; Osborne et al., 2006) and other microbial community fingerprinting techniques, and, in this study, precludes identification of the exact phylogenetic scale (e.g. family, genus, species) at which observed shifts in community structure occur. However, our intention was to characterize overall bacterial community dynamics in activated sludge at a full-scale WWTP and the relative influence therein of a multitude of deterministic environmental drivers. Employment of T-RFLP profiles over time for wholecommunity ecological comparisons was fully sufficient for this purpose. Molecular signatures of the activated sludge bacterial community revealed surprisingly strong temporal dynamics during a period of relatively stable reactor performance. We previously characterized ammonia-oxidizing bacterial population dynamics in the same full-scale bioreactor (Wells et al., 2009), and found strong oscillations in several lineages during a period of stable nitrification. Taken together, our combined results support the argument that dynamic behavior is probably an innate property of biological systems, including engineered bioreactors (Briones and Raskin, 2003). Steady-state equilibrium-based concepts of stability most likely apply only
to aggregate community or ecosystem properties, and true steady-state conditions may not exist in individual microbial populations (Briones and Raskin, 2003; Huisman and Weissing, 2002). In fact, Fernandez et al. (1999) suggested that less stable microbial community structure in lab-scale bioreactors may actually be correlated with more stable aggregate ecosystem function. Similar to our results, Nadarajah et al. (2007) and Kaewpipat and Grady (2002) noted strong temporal bacterial community dynamics in lab-scale reactors during periods of stable operation. Among the few studies available assessing long-term community dynamics in full-scale systems, Lapara et al. (2002) noted moderate dynamics in microbial community structure based on DGGE profiles in industrial wastewater treatment bioreactors, particularly in response to changing influent conditions, and Werker (2006) recorded significant microbial community dynamics via microbial fatty acid analyses during stable operation in a full-scale municipal WWTP during stable operating conditions. In contrast, Smith et al. (2003) reported relatively stable community structure based on ribosomal intergenic spacer length polymorphism analyses over a 55-day period in an industrial WWTP during a period of stable operation. Interestingly, other recent studies have noted strong associations between microbial community dynamics and process performance. Gentile et al. (2007) reported that nitrite appearance and nitrous oxide emissions were linked to population dynamics in lab-scale denitrifying bioreactors, and Zhang et al. (2005) characterized strong changes in microbial community structure during a complete loss of phosphorus removal functionality in a lab-scale EBPR reactor. The results we present here in concert with these previous studies suggest that activated sludge microbial communities exhibit a baseline of moderate population dynamics not necessarily reflected in system performance, with periodic dramatic shifts in community structure that can impact functional stability. Concurrent to weekly activated sludge sampling, we monitored 20 operational and environmental variables that we hypothesized could be deterministic drivers of overall bacterial community dynamics. We identified these variables based on previous reports in the primary literature, expected variability over the year-long sampling period, and importance to process performance goals. DO emerged in both RDA (Fig. 4) and NMDS (Table 3) analyses as the most important explanatory variable associated with bacterial community dynamics. This finding dovetails with our previous assessment of the influence of operational parameters on AOB community structure in the PARWQCP (Wells et al., 2009). In activated sludge systems, large variations in DO have been implicated in the proliferation of bacteria involved in filamentous bulking (Gaval and Pernelle, 2003). Not surprisingly, increases in bacterial populations capable of anoxic or anaerobic metabolisms, such as denitrifying Comamonadaceae, have also been observed in activated sludge at low DO levels (<1 mg/L) (Sadaie et al., 2007). High bulk DO levels (>3 mg/L) were maintained at all times in the PARWQCP bioreactor during our year-long time series. It is thus striking that variations in this parameter were closely associated with bacterial community dynamics. It is possible that the strong correlation between bulk DO and bacterial community dynamics is linked to the presence of anoxic microniches in activated sludge flocs. Schramm et al. (1999) used O2 microsensors to verify the presence of anoxic
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
zones in large flocs from two separate activated sludge bioreactors at 2 mg/L bulk DO concentration, although flocs from an additional four bioreactors showed no evidence of anoxia. Li and Bishop (2004) provided additional evidence for anoxic zones within activated sludge flocs, albeit only at DO 1.2 mg/L. In addition, biomass decay rates have previously been suggested to be a continuous increasing function of DO concentration (Yuan and Blackall, 2002), and it is possible that community dynamics associated with fluctuations in high levels of DO are in fact linked to variations in these decay rates and associated predation or grazing processes. Reactor temperature was also strongly and significantly associated with bacterial community dynamics via both RDA and NMDS analyses. The impact of temperature on bioreactor community structure has been noted previously in lab-scale studies. Nadarajah et al. (2007) detailed strong divergence in bacterial community structure based on DGGE profiles following a temperature increase from 30 C to 45 C in labscale reactors treating an industrial waste. Using culturebased techniques, Alawi et al. (2009) demonstrated a temperature-dependent population shift in NOB in activated sludge incubated at 10 C, 17 C, and 28 C. Moreover, in a full-scale municipal WWTP, Werker (2006) suggested that microbial community dynamics were linked to seasonal temperature variations between 12 C and 28 C. Interestingly, temperature variation in the PARWQCP over our time series was moderate (18.2e25.4 C) and lower than in studies discussed above. The observed correlation to temperature may be a reflection of seasonal periodicity in bacterial community structure. Further investigation based on multi-year time series of activated sludge samples is warranted to test this hypothesis. Influent silver was the only metal of 10 assessed that was significantly associated with bacterial community dynamics in both RDA and NMDS analyses. Silver ions are a known highly effective bacteriocide and have been observed to cause inhibition in activated sludge systems treating photoprocessing wastewater, albeit at levels far exceeding those observed at the PARWQCP (Pavlostathis and Maeng, 1998). We previously documented a moderate association between AOB population dynamics in activated sludge and low influent silver concentrations, similarly with no apparent impact on process performance (Wells et al., 2009). While heavy metals (including copper, zinc, and nickel) have previously been shown to strongly impact microbial community structure in soils upon application of sewage sludge (Baath et al., 1998), we are not aware of previous reports implicating low levels of silver ions in activated sludge microbial community dynamics. It is tempting to speculate that variations in both the overall bacterial community and the AOB subpopulation are examples of community-level adaptation in response to perturbations in influent characteristics that facilitate maintenance of high effluent quality. This hypothesis requires substantial follow-up testing. Surprisingly, overall bacterial community dynamics were also significantly linked in our RDA to influent nitrite levels to the bioreactor, despite low levels (0.05e1.18 mg N/L) and low and stable effluent nitrite. This observation dovetails with previously documented associations in this bioreactor of influent nitrite to AOB community dynamics (Wells et al., 2009) and to variations in broad-scale taxonomic and functional gene diversity based on high-density oligonucleotide microarray
5485
analyses (Wells, 2011). Under conditions observed in this study, the concentration of free nitrous acid (FNA, the conjugate acid of nitrite) is expected to be two orders of magnitude lower than levels implicated in inhibition of AOB or phosphorusaccumulating organisms (Anthonisen et al., 1976; Zhou et al., 2007). It is thus unlikely that variable levels of FNA due to fluctuations in nitrite significantly influenced the observed population dynamics. Nitrite is generally present in municipal wastewater at exceptionally low levels (Tchobanoglous et al., 2002), and the observed fluctuations in this parameter are unexpected. These fluctuations suggest that influent nitrite to the aeration basins may be due to small levels of partial nitrification in the upstream BOD-removal trickling filter. Indeed, elsewhere we detail converging lines of evidence suggesting that bioreactor influent nitrite may be a surrogate measure of microbial immigration from the trickling filter to the downstream activated sludge bioreactor (Wells, 2011). While bacterial community dynamics were also linked via RDA to biomass levels and influent flow rate, both of these parameters were significantly correlated to reactor temperature (r ¼ 0.65, p < 0.001 and r ¼ 0.31, p ¼ 0.02, respectively). Their association to community dynamics may be an artifact of this correlation. Nonetheless, it is feasible that both links represent biologically relevant processes. SRT, which is proportional to biomass concentration at constant HRT, has previously been shown to correlate with shifts in microbial community structure in lab-scale reactors (Saikaly et al., 2005). Moreover, flow rate fluctuations may be associated with varying levels of immigration of autochthonous microorganisms to activated sludge bioreactors, either from the influent or from upstream reactors. We provide evidence elsewhere that such immigration may be a significant driver of activated sludge community dynamics (Wells, 2011). While significant, our RDA explained a relatively low percentage (17.5%) of the variance in T-RFLP profiles. This result is not uncommon in ecological studies (see, for example (Yannarell and Triplett, 2005).) Nonetheless, it is likely that overall bacterial community dynamics in activated sludge are mediated by additional factors beyond the parameters we focused on, including stochastic influences, predation, unmonitored inhibitory chemicals, and fluctuations in levels of specific organic substrates (such as micropollutants) in the plant influent. Indeed, Ofiteru et al. (2010) recently employed a portion of the dataset presented here to show that both neutral (random immigration and births/deaths) and nichebased (deterministic) drivers were associated with population turnover in activated sludge. It also appears likely that microbial community structure is shaped in part by phage predation (Wu and Liu, 2009), protozoan grazing (Petropoulos and Gilbride, 2005), and chaotic behavior (Graham et al., 2007). Nonetheless, our results provide strong evidence that specific deterministic environmental selection processes are significantly linked to dynamics within the overall bacterial community in full-scale activated sludge bioreactors. We demonstrated that observed bacterial taxa richness in activated sludge from a full-scale WWTP increased with the length of time censused according to a power law model, as predicted by macroecological theory, with a best-fit scaling exponent of w ¼ 0.209. White et al. (2006) gathered 984 macroecological community time series to evaluate the generality
5486
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
of the SpecieseTime Relationship (STR) across ecosystems and taxonomic groups, and found that power law exponents typically ranged between 0.2 and 0.4. They further postulated that this surprisingly narrow range of exponents suggests that the STR is a fundamental ecological pattern, analogous to the wellknown SAR. Our observed scaling exponent falls in the lower end of the range recorded by White and colleagues, suggesting that species turnover in microbial and macrobial systems follow similar ecological patterns. Our observed scaling exponent is also well in line with the few previous studies that have examined taxaetime relationships in microbial systems. Redford and Fierer (2009) detailed a power law exponent of w ¼ 0.312 for bacterial communities on leaf surfaces during one growing season, based on bacterial 16S rRNA gene clone libraries. van der Gast et al. (2008) employed lab-scale reactors to test the impact of varying percentages of industrial and municipal wastewater on turnover in microbial community composition based on DGGE analyses, and observed a pronounced decrease in power law exponent (from 0.512 to 0.162) as the percentage of industrial wastewater increased. Interestingly, while the power law exponent we observed in a full-scale bioreactor (w6900 m3) fell within this range, it was well below the reported value for a lab-scale (5 L) reactor with 100% municipal wastewater as influent (w ¼ 0.512). It is tempting to speculate that this discrepancy is associated with the more than six order of magnitude variation in volume between these reactors. Indeed, the exponent of the STR is known in macroecology to be negatively correlated with the spatial scale of observation (White et al., 2006), and van der Gast (2008) suggests that this may be the case in microbial systems as well. Moreover, Adler et al. (2005) suggested that the SAR and STR are components of a unified speciesetimeearea relationship. Additional research is warranted to establish a firm understanding of how taxa turnover in space and time are related in microbial systems. This question is of fundamental importance to the wider field of microbial ecology, and has important implications for the management of complex microbial communities in engineered systems to promote adaption and resilience in the face of process perturbations and variations in environmental parameters.
5.
Conclusions
To our knowledge, the year-long weekly time series described in this report is the most detailed record of bacterial community dynamics in full-scale bioreactors to date and the first assessment of the applicability of a power-law taxaetime relationship to a full-scale wastewater treatment plant. Our results reveal successive shifts in bacterial community structure, with temporal dynamics significantly correlated to temperature, dissolved oxygen, influent silver, and influent nitrite concentrations. These correlations suggest that community dynamics are likely regulateddat least in partdby specific deterministic operational and environmental parameters. Activated sludge bacterial community dynamics followed a gradual succession away from initial conditions, with no apparent impact on process performance.
Accumulation of observed taxa over time followed a powerelaw relationship, in line with predictions from macroecological theory, with a power-law exponent of w ¼ 0.209. Multi-year time series are needed to determine whether this relationship extends over longer time periods. Taken together, our results provide a baseline assessment of bacterial community temporal dynamics in a fully functional activated sludge bioreactor and point to the utility of integrating ecological theory with engineered systems. A future challenge is to elucidate the relationship between these temporal dynamics (i.e. the taxaetime powerelaw exponent) and system stability and resilience, and to employ this information to rationally design and manage complex microbial communities.
Acknowledgments We thank the PARWQCP staff for weekly sampling. This study was funded by the Stanford Woods Institute for the Environment, NSF SGER Grant CBET-0630092 (to C.S.C. and C.A.F.), and the PARWQCP. G.F.W. was supported by EPA STAR and NSF Graduate Research Fellowships.
Appendix. Supplementary material The supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2011.08.006.
references
Adler, P.B., White, E.P., Lauenroth, W.K., Kaufman, D.M., Rassweiler, A., et al., 2005. Evidence for a general speciesetimeearea relationship. Ecology 86, 2032e2039. Alawi, M., Off, S., Kaya, M., Spieck, E., 2009. Temperature influences the population structure of nitrite-oxidizing bacteria in activated sludge. Environmental Microbiology Reports 1, 184e190. Anthonisen, A.C., Loehr, R.C., Prakasam, T.B.S., Srinath, E.G., 1976. Inhibition of nitrification by ammonia and nitrous acid. Journal Water Pollution Control Federation 48, 835e852. Baath, E., Diaz-Ravina, M., Frostegard, A., Campbell, C.D., 1998. Effect of metal-rich sludge amendments on the soil microbial community. Applied and Environmental Microbiology 64, 238e245. Briones, A., Raskin, L., 2003. Diversity and dynamics of microbial communities in engineered environments and their implications for process stability. Current Opinion in Biotechnology 14, 270e276. Curtis, T., Sloan, W., 2006. Towards the design of diversity: stochastic models for community assembly in wastewater treatment plants. Water Science & Technology 54, 227e236. Curtis, T.P., Head, I.M., Graham, D.W., 2003. Theoretical ecology for engineering biology. Environmental Science & Technology 37, 64A. Daims, H., Taylor, M.W., Wagner, M., 2006. Wastewater treatment: a model system for microbial ecology. Trends in Biotechnology 24, 483e489.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
DeSantis, T.Z., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E.L., et al., 2006. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Applied and Environmental Microbiology 72, 5069e5072. Engebretson, J.J., Moyer, C.L., 2003. Fidelity of select restriction endonucleases in determining microbial diversity by terminal-restriction fragment length polymorphism. Applied and Environmental Microbiology 69, 4823e4829. Fernandez, A., Huang, S.Y., Seston, S., Xing, J., Hickey, R., et al., 1999. How stable is stable? Function versus community composition. Applied and Environmental Microbiology 65, 3697e3704. Gaval, G., Pernelle, J.-J., 2003. Impact of the repetition of oxygen deficiencies on the filamentous bacteria proliferation in activated sludge. Water Research 37, 1991e2000. Gentile, M.E., Lynn Nyman, J., Criddle, C.S., 2007. Correlation of patterns of denitrification instability in replicated bioreactor communities with shifts in the relative abundance and the denitrification patterns of specific populations. ISME Journal 1, 714e728. Graham, D.W., Knapp, C.W., Van Vleck, E.S., Bloor, K., Lane, T.B., et al., 2007. Experimental demonstration of chaotic instability in biological nitrification. ISME Journal 1, 385e393. Graham, D.W., Smith, V.H., 2004. Designed ecosystem services: application of ecological principles in wastewater treatment engineering. Frontiers in Ecology and the Environment 2, 199e206. Horner-Devine, M.C., Lage, M., Hughes, J.B., Bohannan, B.J.M., 2004. A taxaearea relationship for bacteria. Nature 432, 750e753. Huber, T., Faulkner, G., Hugenholtz, P., 2004. Bellerophon: a program to detect chimeric sequences in multiple sequence alignments. Bioinformatics 20, 2317e2319. Huisman, J., Weissing, F.J., 2002. Oscillations and chaos generated by competition for interactively essential resources. Ecological Research 17, 175e181. Kaewpipat, K., Grady, C.P.L., 2002. Microbial population dynamics in laboratory-scale activated sludge reactors. Water Science and Technology 46, 19e27. LaPara, T.M., Nakatsu, C.H., Pantea, L.M., Alleman, J.E., 2002. Stability of the bacterial communities supported by a sevenstage biological process treating pharmaceutical wastewater as revealed by PCR-DGGE. Water Research 36, 638e646. Li, B., Bishop, P.L., 2004. Micro-profiles of activated sludge floc determined using microelectrodes. Water Research 38, 1248e1258. Liu, W.T., Marsh, T.L., Cheng, H., Forney, L.J., 1997. Characterization of microbial diversity by determining terminal restriction fragment length polymorphisms of genes encoding 16S rRNA. Applied and Environmental Microbiology 63, 4516e4522. Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., et al., 2004. ARB: a software environment for sequence data. Nucleic Acids Research 32, 1363e1371. Lueders, T., Kindler, R., Miltner, A., Friedrich, M.W., Kaestner, M., 2006. Identification of bacterial micropredators distinctively active in a soil microbial food web. Applied and Environmental Microbiology 72, 5342e5348. Marsh, T.L., Jared, R.L., 2005. Culture-independent Microbial Community Analysis with Terminal Restriction Fragment Length Polymorphism, Methods in Enzymology. Academic Press, pp. 308e329. Nadarajah, N., Grant Allen, D., Fulthorpe, R.R., 2007. Effects of transient temperature conditions on the divergence of activated sludge bacterial community structure and function. Water Research 41, 2563e2571. Nielsen, P.H., Kragelund, C., Seviour, R.J., Nielsen, J.L., 2009. Identity and ecophysiology of filamentous bacteria in activated sludge. FEMS Microbiology Reviews 33, 969e998.
5487
Ofiteru, I.D., Lunn, M., Curtis, T.P., Wells, G.F., Criddle, C.S., et al., 2010. Combined niche and neutral effects in a microbial wastewater treatment community. Proceedings of the National Academy of Sciences USA 107, 15345e15350. Osborne, C.A., Rees, G.N., Bernstein, Y., Janssen, P.H., 2006. New threshold and confidence estimates for terminal restriction fragment length polymorphism analysis of complex bacterial communities. Applied and Environmental Microbiology 72, 1270e1278. Pavlostathis, S.G., Maeng, S.K., 1998. Aerobic biodegradation of a silver-bearing photoprocessing wastewater. Environmental Toxicology and Chemistry 17, 617e624. Petropoulos, P., Gilbride, K.A., 2005. Nitrification in activated sludge batch reactors is linked to protozoan grazing of the bacterial population. Canadian Journal of Microbiology 51, 791e799. Preston, F.W., 1960. Time and space and the variation of species. Ecology 41, 612e627. Redford, A., Fierer, N., 2009. Bacterial succession on the leaf surface: a novel system for studying successional dynamics. Microbial Ecology 58, 189e198. Rittmann, B.E., Hausner, M., Loffler, F., Love, N.G., Muyzer, G., et al., 2006. A vista for microbial ecology and environmental biotechnology. Environmental Science & Technology 40, 1096e1103. Sadaie, T., Sadaie, A., Takada, M., Hamano, K., Ohnishi, J., et al., 2007. Reducing sludge production and the domination of comamonadaceae by reducing the oxygen supply in the wastewater treatment procedure of a food-processing factory. Bioscience, Biotechnology, and Biochemistry 71, 791e799. 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. Schramm, A., Santegoeds, C.M., Nielsen, H.K., Ploug, H., Wagner, M., et al., 1999. On the occurrence of anoxic microniches, denitrification, and sulfate reduction in aerated activated sludge. Applied and Environmental Microbiology 65, 4189e4196. Schu¨tte, U., Abdo, Z., Bent, S., Shyu, C., Williams, C., et al., 2008. Advances in the use of terminal restriction fragment length polymorphism (T-RFLP) analysis of 16S rRNA genes to characterize microbial communities. Applied Microbiology and Biotechnology 80, 365e380. Seviour, R.J., Nielsen, P.H., 2010. Microbial communities in activated sludge plants. In: Seviour, R.J., Nielsen, P.H. (Eds.), Microbial Ecology of Activated Sludge. IWA Publishing, London, UK. Slater, F.R., Johnson, C.R., Blackall, L.L., Beiko, R.G., Bond, P.L., 2010. Monitoring associations between clade-level variation, overall community structure and ecosystem function in enhanced biological phosphorus removal (EBPR) systems using terminal-restriction fragment length polymorphism (T-RFLP). Water Research 44, 4908e4923. Smith, N.R., Yu, Z., Mohn, W.W., 2003. Stability of the bacterial community in a pulp mill effluent treatment system during normal operation and a system shutdown. Water Research 37, 4873e4884. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2002. Wastewater Engineering: Treatment and Reuse. McGrawHill, New York. van der Gast, C.J., 2008. Chapter 6-Islands Shaping Thought in Microbial Ecology. In: Allen, I., Laskin, S.S., Geoffrey, M.G. (Eds.), Advances in Applied Microbiology. Academic Press, pp. 167e182. van der Gast, C.J., Ager, D., Lilley, A.K., 2008. Temporal scaling of bacterial taxa is influenced by both stochastic and
5488
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 7 6 e5 4 8 8
deterministic ecological factors. Environmental Microbiology 10, 1411e1418. Wagner, M., Loy, A., Nogueira, R., Purkhold, U., Lee, N., et al., 2002. Microbial community composition and function in wastewater treatment plants. Antonie Van Leeuwenhoek 81, 665e680. Wells, G.F., 2011. Reexamining the Engineered Nitrogen Cycle: Microbial Diversity, Community Dynamics, and Immigration in Nitrifying Activated Sludge Bioreactors. PhD Thesis. Civil and Environmental Engineering. Stanford University, Stanford, CA. Wells, G.F., Park, H.-D., Yeung, C.-H., Eggleston, B., Francis, C.A., et al., 2009. Ammonia-oxidizing communities in a highly aerated full-scale activated sludge bioreactor: betaproteobacterial dynamics and low relative abundance of Crenarchaea. Environmental Microbiology 11, 2310e2328. Werker, A.G., 2006. An evaluation of full-scale activated sludge dynamics using microbial fatty acid analysis. Water Science & Technology 54, 11e19. White, E.P., Adler, P.B., Lauenroth, W.K., Gill, R.A., Greenberg, D., et al., 2006. A comparison of the speciesetime relationship across ecosystems and taxonomic groups. Oikos 112, 185e195. Wu, Q., Liu, W.-T., 2009. Determination of virus abundance, diversity and distribution in a municipal wastewater treatment plant. Water Research 43, 1101e1109.
Xia, S., Duan, L., Song, Y., Li, J., Piceno, Y.M., et al., 2010. Bacterial community structure in geographically distributed biological wastewater treatment reactors. Environmental Science & Technology 44, 7391e7396. Xia, Y., Kong, Y., Thomsen, T.R., Halkjaer Nielsen, P., 2008. Identification and ecophysiological characterization of epiphytic protein-hydrolyzing Saprospiraceae (“Candidatus Epiflobacter” spp.) in activated sludge. Applied and Environmental Microbiology 74, 2229e2238. Yannarell, A.C., Triplett, E.W., 2005. Geographic and environmental sources of variation in lake bacterial community composition. Applied and Environmental Microbiology 71, 227e239. Yuan, Z., Blackall, L.L., 2002. Sludge population optimisation: a new dimension for the control of biological wastewater treatment systems. Water Research 36, 482e490. Zhang, T., Liu, Y., Fang, H.H., 2005. Effect of pH change on the performance and microbial community of enhanced biological phosphate removal process. Biotechnology and Bioengineering 92, 173e182. Zhou, Y., Pijuan, M., Yuan, Z., 2007. Free nitrous acid inhibition on anoxic phosphorus uptake and denitrification by polyphosphate accumulating organisms. Biotechnology and Bioengineering 98, 903e912.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Morphological architecture of dual-layer hollow fiber for membrane distillation with higher desalination performance Peng Wang, May May Teoh, Tai-Shung Chung* Department of Chemical & Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore 117576, Singapore
article info
abstract
Article history:
A new strategy to enhance the desalination performance of polyvinylidene fluoride (PVDF)
Received 26 January 2011
hollow fiber membrane for membrane distillation (MD) via architecture of morphological
Received in revised form
characteristics is explored in this study. It is proposed that a dual-layer hollow fiber con-
21 June 2011
sisting of a fully finger-like macrovoid inner-layer and a sponge-like outer-layer may
Accepted 7 August 2011
effectively enhance the permeation flux while maintaining the wetting resistance. Dual-
Available online 23 August 2011
layer fibers with the proposed morphology have been fabricated by the dry-jet wet spinning process via careful choice of dopes composition and coagulation conditions. In
Keywords:
addition to high energy efficiency (EE ) of 94%, a superior flux of 98.6 L m2 h1 is obtained
Membrane distillation
during the direct contact membrane distillation (DCMD) desalination experiments. More-
Uniform finger-like macrovoids
over, the liquid entry pressure (LEP) and long-term DCMD performance test show high
Fully sponge-like structure
wetting resistance and long-term stability. Mathematical modeling has been conducted to
Liquid entry pressure
investigate the membrane mass transfer properties in terms of temperature profile and
Heat and mass transfer model
apparent diffusivity of the membranes. It is concluded that the enhancement in permeation flux arises from the coupling effect of two mechanisms; namely, a higher driving force and a lower mass transfer resistance, while the later is the major contribution. This work provides an insight on MD fundamentals and strategy to tailor making ideal membranes for DCMD application in desalination industry. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Water, a necessity to daily life, is currently in great deficiency because of the rapid growth of global population and the accelerated industrialization and urbanization (Escobar, 2010). To mitigate the water crisis, the desalination and wastewater reclamation processes have been studied extensively in past years (Shannon et al., 2008). Among the developed technologies, membrane distillation (MD) shows a promising perspective credited to its large membrane contact area, high salt rejection, small foot print and mild operation conditions (Song et al., 2008). Furthermore, as compared with other desalination technologies like multi-stage flash distillation (MSFD) and reverse osmosis
(RO), MD is capable of integrating with various renewable energy sources such as solar energy, geothermal energy and waste heat source (Curcio and Drioli, 2005). However, at present, the application of MD process is still challenging in the real desalination industry application, mainly due to the low permeation flux, high risk of membrane wetting and long term operation instability (Khayet, 2011). MD is a thermally driven process based on the principle of vaporeliquid equilibrium and coupled heat and mass transfer. During MD process, the volatile compounds in the hot feed solution evaporate at the feed/membrane interface, diffuse through the membrane pores and condense into liquid via various approaches (Sua´rez et al., 2010). The non-volatile
* Corresponding author. Tel.: þ65 6516 6645; fax: þ65 6779 1936. E-mail address: [email protected] (T.-S. Chung). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.012
5490
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
compounds are left in the feed solution. In direct contact membrane distillation (DCMD), the most widely MD configuration in desalination application, the preheated brine feed and cooled permeate are brought in direct contact with two surfaces of micro-porous membrane, respectively. Owing to the hydrophobic characteristic, MD membrane acts as barrier between two phases. Even though in MD process, membrane does not directly contribute the selectivity, its structure, material and geometrical properties play an important role in the production rate, mechanical property and wetting resistance (Bonyadi et al., 2007). As a result, properties of MD membranes have been investigated by many researchers in recent years. As a coupled mass and heat transfer process, the effective driving force of DCMD process is the vapor pressure difference between the feed and permeate interfaces. Unlike other membrane separation processes, improvement of permeation flux for a DCMD membrane can be ascribed to two possible mechanisms: namely, improvement of mass transfer coefficient and enhancement of effective driving force. Via the first mechanism, DCMD membrane structure should be tailored in an attempt to reduce the mass transfer resistance. This can be attained by fabricating membranes with large mean pore size and porosity, open-cell pore structure, thin functional layer and small tortuosity. Through the second mechanism, the thermal conductivity of the membrane needs to be minimized so as to reduce the temperature polarization and achieve a higher driving force. It can be reduced by increasing the membrane porosity, since the conductivity of filled air in membrane pores is much lower than the polymer matrix (Wang et al., 2009). Recently, by fabricating the single-layer hollow fiber membranes with finger-like macrovoids, Wang et al. reported that the fabrication of membrane with macrovoids is an effective way for flux enhancement attributed to high porosity and low tortuosity which allows fast diffusion of water vapor (Wang et al., 2009). A remarkably high permeation flux of 79 kg m2 h1 was achieved at 80 C feed. Apart from improving permeation flux, the macrovoid structure also has potential on energy recovery improvement ascribed to its excellent thermal insulation because of air-filled macro pores. However, the impact of the macrovoid structure on membrane wetting and long-term stability are still concerned (Gryta and Barancewicz, 2010; Wang et al., 2009). During the continuous DCMD operation, there is a risk of membrane pore wetting by vapor condensation and liquid penetration. At a worst situation, entirely wetted pores would result in the loss of membrane selectivity when the feed and permeate solutions at both streams are in direct contact (Tomaszewska, 1999). Therefore, wetting of membrane pores should always be avoided. To improve the wetting resistance, a membrane of high hydrophobicity, small pore size and high tortuosity is highly desired (Phattaranawik et al., 2003). Qtaishat et al. modified the polyetherimide (PEI) asymmetric flat membrane by using fluorinated surface modifying macromolecules (SMM) to increase the membrane hydrophobicity (Qtaishat et al., 2009). The contact angle of their modified membrane increased by more than 10 and the wetting resistance was also found to be improved greatly. On top of material hydrophobicity, a membrane matrix with a fully sponge-like structure can amplify the wetting resistance because of
narrow pore size distribution and higher tortuosity (Khayet, 2011). Teoh and Chung (2009) have produced composite polyvinylidene fluoride polytetrafluoroethylene (PVDF/PTFE) hollow fiber membranes with a fully sponge-like structure. The resultant fiber membranes showed a higher resistance to pore wetting and stable performance in long-term tests (Gryta and Barancewicz, 2010; Teoh and Chung, 2009). On one hand, membranes with smaller tortuosity, higher porosity, larger pore size, interconnected pore structure and macrovoid structure may improve permeation flux. On the other hand, larger tortuosity, smaller porosity and pore size as well as sponge-like structure are desired for high wetting resistance. The trade-off relationship makes it difficult for traditional single-layer MD membranes to achieve both high permeation flux and wetting resistance in a single configuration. Hence, dual-layer hollow fiber membranes comprising dual morphologies offer the opportunity to optimize these two properties. The objective of this study is to elucidate a novel approach to enhance the permeation flux and energy efficiency (EE ) without compromising the wetting resistance via morphological architecture. The inner-layer is designed to be full of finger-like macrovoids whereas the outer-layer comprises sponge-like structure. We also aim to measure liquid entry pressure (LEP) using a homemade set-up as a quantitative indication of membrane wetting resistance. In addition, a mathematical model will be established based on the DCMD permeation flux and operation properties to obtain and analyze the essential transportation parameters including temperature profiles and apparent diffusivity. This information could provide the implicit understanding of vapor permeation mechanism and relative contributions of aforementioned two possible mechanisms for flux enhancement through these newly developed MD membranes. Overall, this work may provide profound implications for the strategies to enhance MD performances and attempts to make the application of this process more practical.
2.
Materials and methods
2.1.
Materials
The working polymer, Kureha PVDF#1300 resin, was kindly provided by Kureha Corp. N-methyl-1-pyrrolidone (NMP, >99.5%), ethylene glycol (EG, >99.5%) and isopropanol (IPA, >99.5%) used in the hollow fiber fabrication were purchased from Merck. The hydrophobic clay particles Closite 20A were purchased from Southern Clay (Gonzales) while PTFE particles (<1 mm) were supplied by SigmaeAldrich. The ultra-pure water used in DCMD tests was produced by a Milli-Q unit (MilliPore) with the resistivity of 18 MU cm.
2.2.
Dope preparation
Prior to dope preparation, the PVDF resin and additive particles were dried in a vacuum oven (Model 282A, Thermo Fisher Scientific Inc.) at 80 C to remove trapped moisture. A typical procedure to prepare a dope is described as follows: The PVDF polymer resin and the clay or PTFE particles with predetermined amounts were added into the NMP/EG solvent
5491
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
mixture. After complete dissolution, the dope was maintained at a slow stirring rate to prevent the sedimentation of the particles before spinning. The compositions for each component in all the dopes are listed in Table 1.
2.3.
Hollow fiber fabrication
The dual-layer PVDF hollow fiber membranes were fabricated via a dry-jet wet phase inversion spinning process by a triorifice spinneret as documented elsewhere (Jiang et al., 2004). During spinning, the inner-layer dope was held at 50 C to reduce the dope viscosity. Single-layer hollow fiber with comparable overall thickness with the dual-layer fiber D3 was spun and denoted as S-out to investigate the effect of sponge-like selective layer thickness on the DCMD performance. Membrane S-out is produced by feeding the outerlayer dope to the inner annulus of the spinneret and employing a bore fluid contains water/NMP 60/40 wt%. Details of the spinning conditions can be found in Table 1. After spinning, the as-spun fibers were immersed in tap water around 4 days to remove solvent residue. Subsequently, the wet fibers were frozen in a refrigerator and dried overnight in a freeze drier (S61-Modulyo-D, Thermo Electron Corp.). All the spinning conditions have been repeated for three times to ensure the reproducibility.
2.4.
Membrane characterizations
2.4.1.
Morphology study
Hollow fiber samples were observed by a field emission scanning electron microscope (FESEM; JEOL JSM-6700). The samples were prepared by being immersed and fractured in liquid nitrogen. Porosity of hollow fibers was obtained by Eq. (1). 3
¼
mfiber =rfiber 100% 1 Vfiber
(1)
where Vfiber is the fiber volume, mfiber is the fiber weight and rfiber is the density of the fiber material. Vfiber was estimated from inner diameter, outer diameter and fiber length. mfiber was measured by an accurate beam balance (A&D, GR-200). rfiber was measured by a multi pycnometer (Quantachrome MVP-D160-E) under helium purging at 20 psi. For each sample, 10 measurements were carried out and an average was adopted.
2.4.2.
Contact angle measurements
The dynamic contact angle, q, was measured with a KSV Sigma 701 tensiometer (0.01 , KSV Instruments Ltd.) through a force tensiometry method at 25 C. During the tensiometry measurement, a high sensitivity micro balance was utilized to measure the interfacial force when the sample fiber was brought into contact with water. Then, the dynamic contact angle was calculated by a software module with the fiber geometry.
2.4.3.
Mechanical property measurements
The mechanical properties of fibers including Young’s modulus, strain at break and maximum extension were measured by an Instron tensiometer (Model 5542, Instron Corp.). A constant elongation rate of 10 mm min1 with a starting gauge length of 50 mm was applied. For each spinning condition, ten fiber samples were tested so as to ensure the accuracy.
2.4.4.
LEP measurements
The liquid entry pressure (LEP) was measured to examine the membrane wetting resistance. The homemade set-up for LEP test is shown in Fig. 1. A stainless steel test tube, which can sustain high pressure, was used as the reservoir for testing solution (3.5 wt% NaCl solutions). A digital pressure gauge (Range: 0e2 bar, 0.001 bar, Wika) was installed on top of the tube. To conduct the measurement, a hollow fiber module was prepared by sealing one end of the fiber with epoxy whereas leaving the other end open. During testing, compressed N2 was connected to the tube to generate the hydraulic pressure. The outlet of the tube was connected to the lumen side of the hollow fiber module. The whole module was put in a beaker of deionized water. The hydraulic pressure was increased with a step of 0.1 bar and each pressure was maintained for 10 min. The conductivity of the water in the beaker was monitored by a conductivity meter Lab 960 (0e500 ms cm1, 0.1 ms cm1, SCHOOT instrument).
2.5.
DCMD desalination experiments
The DCMD desalination experiments were carried out to evaluate the permeation flux at different conditions. Before the experiments were conducted, the testing modules were fabricated by assembling pre-determined number of fibers into a plastic tube of 3/8 in. diameter, with both ends sealed by
Table 1 e Spinning conditions of single- and dual-layer PVDF hollow fibers. Membrane ID Inner-layer dope composition (wt%) Inner-layer dope flow rate (ml min1) Outer-layer dope composition (wt%) Outer-layer dope flow rate (ml min1) Bore fluid (wt%) Bore flow rate (ml min1) External coagulant (wt%) Air gap (cm) Take up speed
D1
D3
S-outa
0.3 water/NMP: 100/0 1.5 IPA/water: 60/40 2 Free fall
Outer-layer dope 1.6 NA NA water/NMP: 60/40 1.5 IPA/water: 60/40 2 Free fall
D2
PVDF/NMP/EG/Closite 20A:10/77/10/3 2.0 1.6 PVDF/NMP/EG/PTFE:12/77/8/3 0.3 0.3 water/NMP: 100/0 water/NMP: 100/0 1.5 1.5 IPA/water: 60/40 IPA/water: 60/40 2 2 Free fall Free fall
a S-out is produced by feeding the outer dope to the inner annulus of the spinneret.
1.2
5492
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
Fig. 1 e Experimental set-up for the LEP measurements.
epoxy. The number of fibers in each module was determined by attaining a packing fraction of 30 5% and a filtration area around 100 cm2. A laboratory-scale DCMD set-up is shown in Fig. 2. A 3.5 wt % NaCl solution was used as the feed whereby ultrapure water was employed as the permeate. The inlet temperature of the feed and permeate solutions were maintained at the target value by a temperature circulator (F12, Julabo) and a water cooler (RT7, Thermal Scientific), respectively. The feed solution was circulated by a centrifugal pump to the shell side of the membrane module when a rotary pump circulates the permeate solution to the lumen side of membrane module. The linear velocities of feed and permeate solutions were kept at 1.4 and 0.7 m s1 respectively. To monitor the temperature variation during the test, digital thermal couples with accuracy of 0.1 C were installed at the inlets and outlets of feed and permeate streams, respectively. The ionic conductivity of the permeate stream was measured before and after the
DCMD test to ensure there is no occasion of leakage. For each test, the membrane module was tested for 30 min to ensure the accuracy. The permeation flux for each feed temperature was calculated based on the outer surface of the fiber using Eq. (2): Nw ¼
DW Ao t
(2)
where Nw is the permeation flux, DW is the permeation weight collected over a pre-determined time duration (t) and Ao is the effective permeation area calculated based on the outer diameter of hollow fiber. The long-term DCMD desalination performance of the fabricated dual-layer hollow fiber was conducted with fiber D3 and similar operation conditions with short-term DCMD experiment. During the test, the inlet temperatures of feed and permeate solutions were maintained at 60.0 C and 15.0 C, respectively. The produced water from the permeate
Fig. 2 e Laboratory-scale set-up for the DCMD experiment.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
5493
3.
Results and discussion
fluid and inner-layer dope flow rates of 2, 1.6 and 1.2 ml min1, respectively. The respective cross sections of the aforementioned fibers are shown in Fig. 4. Shown in Fig. 4 and Table 2, a sponge-like layer as thin as w39 mm can be produced in fiber D3, which is within the range of the optimum membrane thickness as proposed by Lagana` et al. (2000). This thin spongelike layer is expected to enhance the vapor permeation effectively without sacrificing in wetting resistance. As compared with dual-layer fibers, single-layer fiber S-out exhibits smaller macrovoids and sponge-like layer with thickness w104 mm. The morphology is similar as the typical MD hollow fibers reported previously (Teoh and Chung, 2009).
3.1.
Membrane characterizations
3.1.2.
tank was recycled to the feed tank to maintain the salinity of feed solution. The separation factor (R) is calculated with Eq. (3) (Wang et al., 2009): R¼
1
cp cf
100%
(3)
where cf and cp are the NaCl concentration of feed and permeate solutions, respectively.
3.1.1.
Hollow fiber membrane morphology
Fig. 3 displays the morphology of dual-layer hollow fiber membrane D3. It can be seen that a novel morphology consisting of an inner-layer full of finger-like macrovoids and an outer-layer made of sponge-like structure has been attained. From the enlarged images, large macrovoids of regular fingerlike shapes fully occupy the entire porous substrate layer, whilst open cell sponge-like pores form at the outer selective layer. As PVDF was used as the organic phase for the inner- and outer-layer dopes, there is no delamination or interfacial resistance between these two layers. These morphological properties are in good agreement with the proposed morphology. Based on the modeling result reported by Lagana` et al. (2000), an optimum selective layer thickness of the MD membranes is about 30e60 mm by assuming the thermal conductivity of the polymeric material to be 0.1e0.3 W m1 K1 and a typical pore tortuosity of 1.3. Applied into our proposed morphology, the thickness of the sponge-like functional layer should be within the recommended range. Hence, the spinning condition was varied in order to optimize the membrane morphology. Fibers D1, D2, D3 were spun using water as bore
Porosity and mechanical properties measurements
Generally, membranes with high porosity have the benefits of lower thermal conductivity and higher mass transfer coefficient, which are favorable for MD processes. In fact, due to the high concentration of non-solvent (EG), the as-spun dual-layer hollow fibers have elevated bulk porosities larger than 80%, as shown in Table 2. Furthermore, the dual-layer hollow fibers (D1eD3) exhibit superior properties than the single-layer one (S-out), owing to the large amount of macrovoids formed in the inner-layer. The measured Young’s modulus, tensile stress at break and strain at break are summarized in Table 2. Membranes with macrovoid structure often suffer from the weak mechanical strength (Widjojo and Chung, 2006). The Young’s modulus and tensile stress of the self-fabricated dual-layer hollow fiber are comparable with other microporous membranes (Tsai et al., 2001), which make it acceptable for low pressure membrane operation. Compared with the single-layer membrane (S-out), the tensile stress at break of dual-layer hollow fibers is slightly lower. This may be attributed to existence of long finger-like macrovoids in the membrane inner-layer. On the other hand, the observed mechanical properties for dual-layer membrane decreases with a reduction in membrane overall thickness.
Fig. 3 e Morphologies of the hollow fiber membrane D3 (IF [ 1.2 ml minL1). The dashed arrow shows the spinning direction.
5494
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
Fig. 4 e Morphology of the cross-sections for hollow fibers: D1, D2, D3 and S-out.
Since MD membranes are not operated at high pressure and the sponge-like structure provides sufficient wetting resistance (as discussed in Section 3.1.1), the newly developed membranes can be further tested in LEP measurements and DCMD experiments.
3.1.3.
LEP measurements
Wetting resistance is one of essential performance indicators to assess the long-term prospective of any newly developed membranes for MD applications. It is known that the LEP measurement is a quantitative indication of wetting resistance. During the LEP tests, a 3.5 wt% NaCl testing solution is directly in contact with the feed side of the respective membranes. Once the applied pressure reaches the LEP value, some of the membrane pores will be wetted and the testing solution will permeate through those wetted pores. Thus the conductivity of the deionized water will increase gradually. With the pre-
Table 2 e Characteristic properties of single- and duallayer PVDF hollow fibers. Membrane ID
D1
D2
OD (mm) 1192 21 1049 19 Wall thickness 198 13 164 17 (mm) Sponge-layer 78 6 70 7 thickness (mm)* 106 5.0 107 4.6 Contact angle ( ) Porosity (%) 89.4 1.9 86.7 2.0 LEP (bar) 0.7 0.7 Strain at 116 4 92 3 break (%) Tensile stress 0.78 0.03 0.66 0.05 at break (MPa) Young’s modulus 30.1 1.4 26.9 1.4 (MPa)
D3
S-out
961 22 141 24
982 12 143 6
39 1
104 5
110 2.9 84.0 1.1 0.7 74 7
113 5.5 73.6 1.6 1.0 104 3
calibrated correlation between conductivity and NaCl concentration, the amount of testing solution permeates across the wetted membrane can be calculated. Fig. 5(a) illustrates the calculated permeation fluxes under different pressures for hollow fiber D1. Under low hydraulic pressures (0e0.6 bar), no obvious wetting can be observed through the study. Nonetheless, when the applied pressure increases to 0.7 bar, there is a trace amount of testing solution permeating through the membrane, which clearly indicates the wetting of bigger pores. Interestingly, a further increase in pressure not only results in partial wetting, but also promotes the wetting of smaller pores. Since LEP is defined as the minimum pressure and the sustainability of the membrane toward liquid penetration, the LEP value for hollow fiber D1 is determined as 0.7 bar. A similar procedure has been repeated for other fibers and the LEP values for the aforementioned membranes are tabulated in Table 2. It can be shown that LEP values for all tested membranes are higher than the typical operation pressure of DCMD experiments, which is within the range of 0.1e0.3 bar. It is worth noting that dual-layer membranes (D1, D2 and D3) with similar sponge-like layer but different macrovoids layer thicknesses exhibit the same LEP values of 0.7 bar. This implies that the sponge-like layer, rather than the macrovoid layer, is the key factor in determining the capability of membrane wetting resistance. A comparison of LEP values for single- and dual-layer membranes further reveals the impact of the sponge-like layer thickness on the membrane wettability. According to Garcia-Payo et al. (2000), more tortuous pore structure and thicker sponge-like layer provide greater resistance to pore wetting. Thus, membrane S-out with a thicker sponge-like layer of around 106 mm can provide a slightly higher LEP and better wetting resistance.
0.62 0.09 1.08 0.02 25.4 1.7
33.5 2.3
* Sponge-like layer thickness is measured as distribution of silicon element in Energy-dispersive spectroscopy (EDX).
3.2.
DCMD performance
The permeation fluxes obtained for both single- and dual-layer hollow fibers vary with feed inlet temperatures are shown in Fig. 5 (b). For all the membrane samples, the conductivity of the
5495
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
permeate solutions is <5 ms cm1 with separation factor >99.98%, which indicates the membranes are not wetted. Among these dual-layer hollow fibers, fiber D3 which was spun at an inner-layer flow rate of 1.2 ml min1 demonstrates a superior permeation flux as high as 98.6 L m2 h1 at an inlet feed temperature of 80.2 C. Since an increase in inner-layer flow rate during spinning results in a thicker inner-layer and decays the flux dramatically, the values of the flux obtained for fibers D1 and D2 are about 9.7% and 28.3% lower than fiber D3, respectively. This may be owing to the fact that larger overall membrane and sponge-like layer thicknesses lead to decay in membrane mass transfer coefficient. With a similar overall wall thickness, the permeation flux obtained for dual-layer hollow fiber D3 is w96% higher than the
b
Permeation Flux (L P L m-2 hr-1)
6.0 5.0
0.9bar
4.0 3.0
0.8bar 2.0 1.0
0-0.6bar
0.7bar
Permeation Flux, (L P L m-2 hr-1)
a
single-layer fiber (S-out). This evidently discloses the noteworthy function of finger-like macrovoids at the membrane inner-layer for the enhancement of DCMD flux. As proposed earlier, the dual-layer fiber consisting of a macrovoids supporting layer has a lower mass transfer resistance and a higher driving force resulting from the smaller tortuosity and higher porosity. Further details about the relative contributions of these two effects will be discussed in Section 3.3. Generally, the permeation flux increases almost exponentially with increasing feed temperature owing to the exponential dependence of vapor pressure on temperature (Bonyadi and Chung, 2007). The normalized permeation with a cylindrical coordinate and hollow fiber configuration has been calculated with Eq. (4) (Hou et al., 2009):
0.0
D1 D2 D3 S-out
100 80 60 40 20 0
0
3
6
9
12
Time (min)
c
45
5.0E-06 4.0E-06 3.0E-06 2.0E-06 1.0E-06
45
55
65
75
85
Tf Tf,m Tp Tp,m
80
60
40
20
85
45
f
2.0E-05
1.6E-05
Energy Efficiency, EE E
App parant Diffusivity, KD ,( m2 s-1)
75
0
0.0E+00
1.2E-05
8.0E-06
D1
4.0E-06
65
100
D1 D2 D3 S-out
Feed Inlet Temperature (°C)
e
55
Feed Inlet Temperature (°C)
d
6.0E-06
Temperaature Profile (°C)
Normalized Permeation Flux, (L m-1 s-1)
120
1µmD2 D3 S-out
0.0E+00
55
65
75
85
Feed Inlet Temperature (°C) 1.00
D1 D2 D3 S-out
0.80
0.60
100nm
0.40
0.20
45
55
65
75
Feed Inlet Temperature (°C)
85
45
55
65
75
85
Feed Inlet Temperature (°C)
Fig. 5 e (a) Permeation fluxes obtained from LEP measurements for hollow fiber D1. (b) DCMD Permeation fluxes obtained for PVDF hollow fibers. (c) Normalized DCMD Permeation fluxes obtained for PVDF hollow fibers. (d) Temperature profile for PVDF hollow fiber D1: Tf and Tp are the average temperatures for feed and permeate sides; Tf,m and Tp,m are temperatures calculated for membrane surfaces facing the feed and permeate sides, respectively. (e) Calculated apparent Diffusivities (Da) for PVDF hollow fibers. (f) Calculated energy efficiencies for PVDF hollow fibers.
5496
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
ro Nw;normal ¼ Nw ro ln ri
(4)
Fig. 5(c) demonstrates the normalized flux for all the membranes. The results show that all the dual-layer membranes have the similar normalized fluxes, which may be attributed to the similar structure (finger-like macrovoids structure and sponge-like structure). The values of normalized fluxes of dual-layer membranes are much higher than the single-layer membrane.
3.3.
Modeling of mass transfer in DCMD
Due to the complexity of DCMD processes, the vapor transportation across a MD membrane is often determined by many parameters including membrane micro-structure, membrane module configuration and operation conditions. In order to investigate the detailed mass transport of the self-fabricated hollow fiber membranes, a mathematical model is proposed. This model aims to acquire essential mass transfer parameters as temperature profile and apparent diffusivity with the operation parameters and permeation flux data obtained from the DCMD experiments. The key information and results of the modeling is shown in this section and detailed modeling equations can be found in the Supplementary material.
3.3.1.
Temperature profile
Fig. 6 demonstrates the temperature changes and heat transfer across the hollow fiber membrane. As a coupled heat and mass transfer process, the vapor pressure difference between the bulk feed and permeate solutions cannot be fully utilized as the effective driving force. A considerable proportion of heat loss (i.e., temperature polarization) is observed at the boundary layers. Therefore, investigation of the temperature profile across MD membrane is a critical approach to understand the heat transfer during the DCMD process.
An attempt to estimate the temperature profile across the MD hollow fiber was performed as follows: (1) the steady state heat flux (Qss) is equal to the enthalpy change between the inlet and outlet of both feed and permeate streams, and (2) the convective heat transfer coefficients (hf and hp) are obtained from heat transfer correlations. Thus, the temperatures at membrane surfaces can be calculated by Eqs. (5) and (6). Qss hf Ao
(5)
Qss Tp hp Ai
(6)
Tf ;m ¼ Tf
Tp;m ¼
where Tf,m and Tp,m are the average temperatures at membrane surfaces toward feed and permeate, respectively. Tf and Tp are the average temperatures of bulk feed and permeate solutions, respectively. hf and hp are the heat convection coefficients of feed and permeate sides, respectively. Ao and Ai are the area of outer and inner surface of hollow fiber, respectively, and Qss is the average heat flux at steady state. Fig. 5(d) illustrates the temperature profile of fiber D1. The temperatures difference of bulk and membrane surface at the feed side (i.e., 3e4 C) is relatively lower than the gap at the permeate side (i.e., 10e20 C). This may be due to the fact that the heat transfer coefficient and mass flow rate are much higher at the feed as compared to the permeate side. At steady state, where the heat fluxes at feed and permeate sides are equal, the resultant heat transfer boundary layer at the permeate side is thicker than the feed side. As a result, a dramatic temperature drop is expected from the membrane surface to the permeate bulk solution. In addition, the temperature difference for the permeate side are more significant at a higher feed temperature, but no noticeable changes can be observed at the feed side. The increase of the permeate side is owing to a higher total heat flux (Qss) arisen from superior heat conduction flux and permeation flux. In
Fig. 6 e (a) Temperature profile and heat transfer across a DCMD hollow fiber and (b) total heat transfer along the hollow fiber module.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
contrast, the heat transfer coefficient for the feed side increases simultaneously as the feed temperature and total heat flux increase, which results in a smaller change for the temperature gap between Tp and Tp,m.
3.3.2.
Apparent diffusivity
Usually, the permeation flux obtained in the DCMD is strongly correlated by the fiber’s structure, such as wall thickness, inner and outer diameters. In order to investigate the function of the proposed morphology, the parameter of apparent diffusivity is introduced which exclude the impacts of membrane thickness, fiber diameters and MD operation parameters. It is defined as Eq. (7): Da ¼
3 Dwa
s
(7)
where Dwa is the diffusion coefficient of water vapor through stagnant air, s is the tortuosity of vapor diffusion path in the membrane. For industrial processes, apparent diffusivity is one of the most important modeling parameters to examine the diffusion of gas or liquid through a porous structure (Merdas et al., 2002). Hereby, it is introduced as a quantitative measurement of the intrinsic diffusion property. The apparent diffusivity can be obtained from the equation that describes the diffusion of vapor through a porous hollow fiber membrane as following (Fernandez-Pineda et al., 2002): Nw ¼
1 Mww Da Pf ;m Pp;m Ylm ro lnðro =ri Þ
(8)
where Nw is the permeation flux, Mww is the molecular weight of water, Ylm is the logarithm mean fraction of air in membrane pores, ro and ri are the outer radius and inner radius of the hollow fiber respectively, Pf ;m and Pp;m are the vapor pressure at the membrane surfaces facing feed and permeate sides respectively. (The effect of concentration polarization (CPC) on Pf ;m is taken into consideration with calculated mass transfer coefficient. The detailed equations are included in Supplementary material.) The calculated Da values for all PVDF follow fibers are presented in Fig. 5(e). For dual-layer fibers, the Da values are fairly comparable, which is possibly attributing to their similarity in terms of membrane morphology. From Eq. (7), it can be concluded that the value of Da is only determined by the membrane characteristics such as mean pore size, porosity and tortuosity. Owing to the similar coagulation conditions, duallayer membranes have a similar pore size and porosity. Besides, Table 2 displays that the dual-layer fibers also show similar percentages of sponge-like layer thickness to the overall thickness (0.3e0.35). This result suggests a comparable average tortuosity and consequently obtained similar Da values. A comparison of Da values between dual-layer and single layer provides solid evidence that the enhancement in mass transfer rate is mainly due to a lower mass transfer resistance of fingerlike macrovoids. At the feed temperature of 80 C, the Da value of the dual-layer membrane (D3) is w80% higher than that of the single-layer fiber (S-out). As discussed in Section 3.2, the finger-like macrovoid structure in the MD membrane enhances the permeation flux through two possible mechanisms: namely, increased mass transfer rate and enhanced driving
5497
force. In this regards, about 1-fold permeation flux improvement is attained by membrane D3 as compared to the fiber Sout, as shown in Fig. 5(b). Therefore, a conclusion can be drawn that superior permeation fluxes obtained for the proposed MD fibers is a coupling effect of both mechanisms, while the enhancement of mass transfer rate is the major contribution.
3.4.
Energy efficiency
As a coupled mass and heat transfer process, the total heat transfer in the DCMD process consists of two mechanisms: (1) the latent heat associated with the vaporization and condensation of water (Ql), and (2) the heat conduction across the membrane which usually indicated as undesirable heat loss (Qc). Hence, the energy efficiency, EE, can be denoted to the percentage of latent heat flux over the total heat flux across the membrane (Smolders and Franken, 1989), and calculated using Eq. (9) for hollow fibers: N w l m Ao km Tf Tp Ao þ Nw lm Ao ro lnðro =ri Þ Nw lm Ao ¼ mp Cp;p Tp;out Tp;in
h¼
Ql ¼ Q c þ Ql
(9)
where lm is the latent heat of water vaporization evaluated at average membrane temperature, Tf and Tp are average temperatures at feed and permeate sides, respectively. km is the thermal conductivity of the membrane, mp is the mass flow rate of the permeate solution, Cp,p is the average specific heat capacity of the permeate solution, Tp,in and Tp,out are inlet and outlet temperatures of the bulk permeate solution, respectively. The calculated EE against feed inlet temperature for all membranes are shown in Fig. 5(f). The EE increases with the feed inlet temperature. This is attributed to the fact that the driving force for vapor transport increases exponentially, whereas the conduction heat increases linearly with the feed temperatures. The EE of fiber D3 are almost twice of the fiber S-out; this improvement is mainly caused by the co-action of two factors: namely, enhanced mass transfer coefficient and reduced thermal conductivity. As referring to Fig. 5(e), the dual-layer fiber D3 exhibits a much higher apparent diffusivity and a less mass transport resistance, and consequently larger permeation fluxes (Nw) and latent heat flux (Ql). On the other hand, because the thermal conductivity of the stagnant air in pore structure is much lower than the membrane polymer matrix (Curcio and Drioli, 2005). The unique micro-structure with finger-like macrovoids with a high porosity in fiber D3 effectively reduces the heat conduction loss (Qc). Therefore, compared with single-layer configuration, the unique duallayer morphology allows the combined advantages offered by the macrovoid supporting layer (i.e., higher permeation flux and energy efficiency) and a fully sponge-like outer selective layer (i.e., withstand membrane wetting). For dual-layer hollow fibers, the EE values increase gradually with a decrease in the overall fiber thickness. This is mainly attributed to the adverse effects of membrane thickness on thermal conduction and evaporation, where a thinner membrane may enhance both conduction heat flux (Qc) and latent heat flux (Ql). In other words, a decrease in membrane
5498
100
100.00%
80
99.99%
60
99.98%
40
99.97%
20
99.96%
0
Separation factor (%)
Permeation Flux (L m-2 hr-1)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
99.95% 0
20
40
60
80
100
120
140
160
180
200
Time (hr)
Fig. 7 e Variation of permeation flux and separation factor during the long-term DCMD experiment.
thickness could increase both the de-numerator and numerator in Eq. (9). Hence, the changes of EE values are depending on the relative changing rate of de-numerator and numerator. Due to the high apparent diffusivities of dual-layer membranes, the rate of enhancement in latent heat flux (Ql) is much greater than the conduction heat flux (Qc). As a result, the latent heat flux is the dominant factor on the calculated EE for all the tested MD membranes.
can be seen that the permeation flux and separation factor are relatively stable during the whole test. The separation factor is higher than 99.9%, which indicates a high wetting resistance of the tested membranes. A slight decline of permeation flux and separation factor can be observed at w around 80 h. This may be attributed to the scaling of inorganic salts on the membrane surfaces (Gryta and Barancewicz, 2010; Song et al., 2008).
3.6. 3.5.
Comparison with other MD membranes
Long-term desalination performance
The long-term DCMD desalination performance of the fabricated dual-layer hollow fiber was conducted for 200 h. The permeation flux and separation factor are illustrated in Fig. 7. It
Table 3 shows a comparison of DCMD performance between this work and the literature data. For a better comparison, the literature data consists of commercial and laboratory fabricated flat sheet and hollow fiber membranes. A noticeable flux
Table 3 e DCMD performances of MD membranes for desalination. Membrane
GVHP commercial flat sheet (Khayet, 2011) PVDF flat sheet (Tomaszewska, 1996)a SMM/PEI flat sheet (Qtaishat et al., 2009) Accurel Membrana GmbH PP commercial hollow fiber (Li and Sirkar, 2004)b PVDF single-layer hollow fiber (Teoh and Chung, 2009) PVDF single-layer hollow fiber (Hou et al., 2009) PVDF single-layer hollow fiber (Wang et al., 2009) PVDF dual-layer hollow fiber (Bonyadi and Chung, 2007) PVDF dual-layer hollow fiber (Su et al., 2010) PVDF dual-layer hollow fiber D3 (this work)
Thickness
Feed solution property
Permeate solution property
Permeation flux
vp (m s1)
(L m2 h1)
19.7
500 rpm
48.6
e
20
e
10
50
500 rpm
15
500 rpm
20.9
1.0 wt% NaCl
90
2.29
15e17
1.66
41.4
110
3.5 wt% NaCl
79.5
1.9
17.5
0.9
40.4
130
3.5 wt% NaCl
81.8
0.5
20
0.15
40.5
190
3.5 wt% NaCl
81.3
1.8
17
1.2
79.2
340
3.5 wt% NaCl
90.3
1.6
17
0.8
55.2
271
3.5 wt% NaCl
78.2
1.8
16.6
0.72
66.9
141
3.5 wt% NaCl
79.8
1.4
17
0.7
98.6
(mm)
Solution
Tf,in ( C)
vf (m s1)
118
Deionized water
90.7
500 rpm
e
1.0 wt% NaCl
60
160a
0.5 M NaCl
150
Tp,in ( C)
Boldface denotes result of the presented research. All other data are from the literature. a The thickness was measured from the SEM image. b The original data was reported based on the inner diameter, we had converted the permeation flux based on outer diameter, which is more convenient for the comparison.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
enhancement has been achieved in this work. This is denoted to the morphological architecture with the aid of the dopes formulation and spinning conditions.
4.
Conclusions
We have demonstrated that morphological architecture of PVDF dual-layer hollow fiber membranes can be utilized to enhance the DCMD performance. The dual-layer hollow fiber with a fully finger-like inner-layer and a totally sponge-like outer-layer is a preferred structure. The resultant dual-layer membranes show enhanced MD performance with minimum sacrifice of wetting resistance. The aforementioned morphology can be manipulated and tailor made via careful manipulations of inner- and outer-dope compositions as well as coagulation conditions. As a result, a permeation flux of 98.6 L m2 h1 has been obtained. Credited to the high thermal resistance and excellent permeation flux, the dual-layer fiber exhibits a remarkably high EE value of w94%. Furthermore, dual-layer membranes with sponge-like layer thicknesses between 30 and 40 mm reveal higher membrane wetting resistance with the measured LEP values of 0.7 bar. The mass transfer modeling shows that the apparent diffusivity values for dual-layer fibers are remarkably higher than the single-layer configuration. This implies that the former has a much smaller resistance for vapor transfer than the latter. The as-spun fibers membranes exhibit a better permeation flux due to the coupling effects of lower mass transfer resistance and higher effective driving force. However, a lower mass transfer resistance is the dominant factor for the higher performance.
Acknowledgments The authors would like to acknowledge A*STAR and National University of Singapore for funding the research through the grant number R-279-000-291-305. The authors also appreciate Kureha Corp., Japan for the provision of the Kureha PVDF resin. Special thanks are due to Prof. E.L. Cussler (University of Minnesota), Dr. K. Y. Wang, Dr. N. Widjojo, Dr. J. C. Su, Miss H. Wang and Mr. Y. H. Sim for their valuable suggestions.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.012.
references
Bonyadi, S., Chung, T.S., 2007. Flux enhancement in membrane distillation by fabrication of dual layer hydrophilichydrophobic hollow fiber membranes. Journal of Membrane Science 306 (1e2), 134e146.
5499
Bonyadi, S., Chung, T.S., Krantz, W.B., 2007. Investigation of corrugation phenomenon in the inner contour of hollow fibers during the non-solvent induced phase-separation process. Journal of Membrane Science 299 (1e2), 200e210. Curcio, E., Drioli, E., 2005. Membrane distillation and related operations e a review. Separation and Purification Reviews 34 (1), 35e86. Escobar, I.C., 2010. A summary of challenges still facing desalination and water reuse. In: Escobar, I.C., Scha¨fer, A.I. (Eds.), Sustainable Water for the Future e Water Recycling Versus Desalination. Elsevier Science, The Netherlands, pp. 389e397. Fernandez-Pineda, C., Izquierdo-Gil, M.A., Garcia-Payo, M.C., 2002. Gas permeation and direct contact membrane distillation experiments and their analysis using different models. Journal of Membrane Science 198 (1), 33e49. Garcia-Payo, M.C., Izquierdo-Gil, M.A., Fernandez-Pineda, C., 2000. Wetting study of hydrophobic membranes via liquid entry pressure measurements with aqueous alcohol solutions. Journal of Colloid and Interface Science 230 (2), 420e431. Gryta, M., Barancewicz, M., 2010. Influence of morphology of PVDF capillary membranes on the performance of direct contact membrane distillation. Journal of Membrane Science 358 (1e2), 158e167. Hou, D., Wang, J., Qu, D., Luan, Z., Ren, X., 2009. Fabrication and characterization of hydrophobic PVDF hollow fiber membranes for desalination through direct contact membrane distillation. Separation and Purification Technology 69 (1), 78e86. Jiang, L., Chung, T.S., Li, D.F., Cao, C., Kulprathipanja, S., 2004. Fabrication of matrimid/polyethersulfone dual-layer hollow fiber membranes for gas separation. Journal of Membrane Science 240 (1e2), 91e103. Khayet, M., 2011. Membranes and theoretical modeling of membrane distillation: a review. Advances in Colloid and Interface Science 164 (1e2), 56e88. Lagana`, F., Barbieri, G., Drioli, E., 2000. Direct contact membrane distillation: modelling and concentration experiments. Journal of Membrane Science 166 (1), 1e11. Li, B.A., Sirkar, K.K., 2004. Novel membrane and device for direct contact membrane distillation e based desalination process. Industrial & Engineering Chemistry Research 43, 5300e5309. Merdas, I., Thominette, F., Tcharkhtchi, A., Verdu, J., 2002. Factors governing water absorption by composite matrices. Composites Science and Technology 62 (4), 487e492. Phattaranawik, J., Jiraratananon, R., Fane, A.G., 2003. Effect of pore size distribution and air flux on mass transport in direct contact membrane distillation. Journal of Membrane Science 215 (1e2), 75e85. Qtaishat, M., Matsuura, T., Khayet, M., Khulbe, K.C., 2009. Comparing the desalination performance of SMM blended polyethersulfone to SMM blended polyetherimide membranes by direct contact membrane distillation. Desalination and Water Treatment 5 (1e3), 91e98. Shannon, M.A., Bohn, P.W., Elimelech, M., Georgiadis, J.G., Mar~ıas, B.J., Mayes, A.M., 2008. Science and technology for water purification in the coming decades. Nature 452 (7185), 301e310. Smolders, K., Franken, A.C.M., 1989. Terminology for membrane distillation. Desalination 72 (3), 249e262. Song, L., Ma, Z., Liao, X., Kosaraju, P.B., Irish, J.R., Sirkar, K.K., 2008. Pilot plant studies of novel membranes and devices for direct contact membrane distillation-based desalination. Journal of Membrane Science 323 (2), 257e270. Su, M., Teoh, M.M., Wang, K.Y., Su, J., Chung, T.S., 2010. Effect of inner-layer thermal conductivity on flux enhancement of dual-layer hollow fiber membranes in direct contact membrane distillation. Journal of Membrane Science.
5500
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 4 8 9 e5 5 0 0
Sua´rez, F., Tyler, S.W., Childress, A.E., 2010. A theoretical study of a direct contact membrane distillation system coupled to a salt-gradient solar pond for terminal lakes reclamation. Water Research 44 (15), 4601e4615. Teoh, M.M., Chung, T.S., 2009. Membrane distillation with hydrophobic macrovoid-free PVDF-PTFE hollow fiber membranes. Separation and Purification Technology 66 (2), 229e236. Tomaszewska, M., 1999. Membrane distillation. Environment Protection Engineering 25 (1), 37e47. Tomaszewska, M., 1996. Preparation and properties of flat-sheet membranes from poly(vinylidene fluoride) for membrane distillation. Desalination 104 (1e2), 1e11.
Tsai, H.A., Huang, D.H., Ruaan, R.C., Lai, J.Y., 2001. Mechanical properties of asymmetric polysulfone membranes containing surfactant as additives. Industrial and Engineering Chemistry Research 40 (25), 5917e5922. Wang, K.Y., Foo, S.W., Chung, T.S., 2009. Mixed matrix PVDF hollow fiber membranes with Nanoscale pores for desalination through Direct Contact Membrane Distillation. Industrial & Engineering Chemistry Research 48 (9), 4474e4483. Widjojo, N., Chung, T.S., 2006. Thickness and air gap dependence of macrovoid evolution in phase-inversion asymmetric hollow fiber membranes. Industrial & Engineering Chemistry Research 45 (22), 7618e7626.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effect of short-chain organic acids on the enhanced desorption of phenanthrene by rhamnolipid biosurfactant in soilewater environment Chun-jiang An 1, Guo-he Huang*, Jia Wei 1, Hui Yu 1 Faculty of Engineering and Applied Science, University of Regina, Regina, Saskatchewan, Canada S4S 0A2
article info
abstract
Article history:
This study investigated the effect of short-chain organic acids on biosurfactant-enhanced
Received 14 December 2010
mobilization of phenanthrene in soilewater system. The desorption characteristics of
Received in revised form
phenanthrene by soils were assessed in the presence of rhamnolipid and four SCOAs,
15 June 2011
including acetic acid, oxalic acid, tartaric acid and citric acid. The tests with rhamnolipid
Accepted 7 August 2011
and different organic acids could attain the higher desorption of phenanthrene compared
Available online 16 August 2011
to those with only rhamnolipid. Among the different combinations, the series with rhamnolipid and citric acid exhibited more significant effect on the desorption perfor-
Keywords:
mance. The removal of phenanthrene using rhamnolipid and SCOAs gradually increased
Phenanthrene
as the SCOA concentration increased up to a concentration of 300 mmol/L. The effects of
Biosurfactant
pH, soil dissolved organic matter and ionic strength were further evaluated in the presence
Short-chain organic acids
of both biosurfactant and SCOAs. The results showed that the extent of phenanthrene
Desorption
desorption was more significant at pH 6 and 9. Desorption of phenanthrene was relatively
Field soil
lower in the DOM-removed soils with the addition of biosurfactant and SCOAs. The presence of more salt ions made phenanthrene more persistent on the solid phase and adversely affected its desorption from contaminated soil. The results from this study may have important implications for soil washing technologies used to treat PAH-contaminated soil and groundwater. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Polycyclic aromatic hydrocarbons (PAHs) are frequently found in the environment as a result of various natural processes and anthropogenic activities (Nadal et al., 2004). During the past decades, large amounts of PAHs have been generated in incomplete combustion of organic materials and the refining of crude oil (Ferrarese et al., 2008). Contamination from the release of petroleum hydrocarbons into the environment by deliberate discharges, accidental spills and leakages is a particularly significant problem. These contaminants have
been demonstrated to be a threat to human health and ecological safety (Djomo et al., 2004; Yu et al., 2011). Pollution caused by these compounds is becoming a widespread problem for the ecosystem across the world. The U.S. Environmental Protection Agency (EPA) has already classified PAHs as priority pollutants (Kanaly and Harayama, 2000). Due to the low solubility and high hydrophobicity, a significant proportion of these organic contaminants can bind to soil sediments (Gomez et al., 2010; Wei et al., 2011). Mass transfer limitation from the solid phase to the liquid phase seriously impedes the bioavailability and reduces the bioremediation
* Corresponding author. Tel.: þ1 306 585 4095; fax: þ1 306 585 4855. E-mail addresses: [email protected] (C.-j. An), [email protected] (G.-h. Huang), [email protected] (J. Wei), [email protected] (H. Yu). 1 Tel.: þ1 306 585 4095; fax: þ1 306 585 4855. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.011
5502
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
efficiency of these compounds. PAHs often persists in soil and groundwater system, posing a long-term threat to the environment. Soil washing with extracting solutions has been proposed as a promising technique for the remediation of contaminated soils and groundwater in recent years (Deshpande et al., 1999; Chu and Chan, 2003; Gan et al., 2009). These washing solutions include different types of chemical agents, such as surfactants, solvents, and cosolvents (Khodadoust et al., 2000; Silva et al., 2005; Alcantara-Garduno et al., 2008). Among these compounds, surfactant has been extensively used to increase the desorption and mobilization of pollutants in soil due to its unique characteristic. Surfactant is composed of both a hydrophilic and hydrophobic moiety, which can form micelle aggregates in aqueous phase when surfactant concentration exceeds critical micelle concentration (CMC) (Deshpande et al., 2000). These micelles significantly facilitate the solubilization of hydrophobic organic compounds (HOCs) and their transport from contaminated soil to aqueous phase. However, concerns over some synthetic surfactants have increased because they are toxic and thus present a potential risk for environment and human health (Kanga et al., 1997). As a more prominent and less toxic alternative, biosurfactants, such as rhamnolipids, are starting to be widely used in soil remediation. Rhamnolipids are anionic biosurfactants produced by Pseudomonas sp. bacteria and they can show greater environmental compatibility and surface activity (Clifford et al., 2007). The application of rhamnolipid in field remediation of PAH-contaminated soil has been successfully demonstrated in many previous studies (Noordman et al., 1998; Bordas et al., 2005; Mulligan, 2005). However, a good understanding of the various factors involved in biosurfactant-enhanced remediation is still a challenge in many respects. Further study is necessary to define the role of complex system composition and heterogeneity in contaminant behaviors. The technology of soil washing with short-chain organic acids (SCOAs) has also been employed to remediate polluted soils. In natural environment, these organic acids can be produced from incomplete carbohydrate oxidation and chemical acidification within root exudates, microbial secretions, and decomposition processes (Blank et al., 1994; Wang et al., 2009). These simple carboxylic acids can act as chelating agents and facilitate the translocation of metals in soils (Jones, 1998; van Hees et al., 2003). Such properties have been used to increase the treatment efficiency for heavy metal contaminated soil in many applications (Yuan et al., 2007; Zhang et al., 2008). Recently, it was found the presence of SCOAs could have an influence on the behavior of HOCs. White et al. (2003) reported that the availability of weathered p,p0 -DDE increased in the presence of seven organic acids. It was also observed in our previous studies that the addition of SCOAs showed adverse effect on adsorption process of pyrene (An et al., 2010). Although these findings are encouraging, a number of important issues about the role of SCOAs remain poorly understood under either laboratory or field conditions. Moreover, with the prevalence and concern of biosurfactant, few studies have focused on the interacted influence of biosurfactants and organic acids on the behaviors and mechanisms of PAH translocations.
The present study therefore investigated the effect of SCOAs on the mobilization of phenanthrene by rhamnolipid biosurfactant in soilewater environment. Batch tests were conducted to determine the desorption of phenanthrene in two kinds of natural soil samples. The performances of different biosurfactant-SCOA combinations, as well as the effects of solution pH, dissolved organic matter and ionic strength were evaluated. The results from this study may have important implications for soil washing technologies used to treat PAH-contaminated soil and groundwater.
2.
Material and methods
2.1.
Biosurfactant
The biosurfactant used in this study was a rhamnolipid obtained from the Jeneil Biosurfactant Company (WI, USA). Jeneil product RECO-10, with a mono- to di-rhamnolipid ratio of 1:1, was used and the biosurfactant was supplied as a 10% aqueous solution. It is an extra-cellular natural substance produced during precisely controlled fermentation processes utilizing certain bacterial strains (Wei et al., 2005). The two major rhamnolipid components in this solution are a monorhamnolipid (a-L-rhamnopyranosyl-b-hydroxydecanoyl-bhydroxydecanoate), and a dirhamnolipid (2-O-a-L-rhamnopyranosyl-a-L-rhamnopyranosyl-b-hydroxydecanoyl-b-hydroxydecanoate). The molecular weight of monorhamnolipid (C26H48O9) and dirhamnolipid (C32H58O13) are 504 g/mol and 650 g/mol, respectively. The critical micellar concentration of this biosurfactant is 24 mg/L.
2.2.
Chemicals
Phenanthrene was selected as representative PAH, and was obtained from Aldrich Chemical Co. (WI, USA) with purity greater than 99%. Four species of SCOAs, including acetic acid, oxalic acid, tartaric acid and citric acid, obtained commercially as pure chemical reagents (Sigma Aldrich, MO, USA), were used in the study. Dichloromethane was obtained from Fisher Scientific (PA, USA). All other chemicals used were of reagent grade quality or higher.
2.3.
Soil preparation
The clean natural soil was derived from Coleville site in Saskatchewan, Canada. Two soils, under-plant soil and ditch soil which represent the predominant soil types of the site, were chosen for evaluation in this study. The physical and textural characteristics of these soils are given in Table 1. Soil samples were dried and sieved through a 2-mm sieve to remove coarse fragments. Contaminated soil was then prepared by dissolving an appropriate quantity of phenanthrene in dichloromethane and a known weight of soil was added with continuous mixing. The resultant mixture was placed in a ventilation hood to allow the complete evaporation of the solvent. The contaminated soil was stored in a vessel at room temperature for 7 days. Such soil had a final concentration of 100 mg/kg for phenanthrene and was used directly in the batch experiments.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
concentration for each SCOA was 100 mmol/L, and the rhamnolipid concentrations were 50 mg/L.
Table 1 e Properties of the selected soils. Sorbents
Organic pH Sand Silt Clay matter (%) (H2O) (%) (%) (%)a
Under-plant soil Ditch soil
8.84 4.57
7.2 7.2
Texture
14.3 11.9 28.7 Sandy loam 2.9 24.8 36.2 Clay loam
a Soil grain size classification is according to international criterion: clay < 0.002 mm, silt < 0.02 mm, 0.02 mm < sand < 0.2 mm.
2.4.
Desorption experiments
All the experiments were carried out in 25 mL glass centrifuge tubes with Teflon-lined screw caps. The effects of biosurfactant and SCOAs on the behaviors of phenanthrene were examined using the combinations of rhamnolipid and four organic acids including acetic acid, oxalic acid, tartaric acid and citric acid. For the desorption experiment, 0.5 g contaminated soil and 10 mL background solution. The background solution was comprised of appropriate rhamnolipid and organic acids, as well as 0.01 mol/L NaN3 as a biocide. System pH was adjusted to 6 at a constant ion concentration. The centrifuge tubes were vortexed for 20 s, and then placed on a reciprocal shaker at 20 1 C and 125 rpm for 24 h to reach the desorption equilibrium. Preliminary experiments showed that 24 h were enough for the desorption of phenanthrene to reach equilibrium, and the phenanthrene degradation or adsorption by the tubes was negligible. The tubes were then centrifuged at 5000 rpm for 25 min. Phenanthrene in the aqueous phase were extracted with dichloromethane, and their concentrations were analyzed by gas chromatography. The amount of phenanthrene adsorbed to the soil was the difference between the initial amount in the system and the amount of the phenanthrene remaining in the solution. All the tests were conducted in triplicate.
2.5.
Effect of pH
The nature of biosurfactant and SCOAs can vary at different pH, so combined effects of rhamnolipid and SCOAs under different pH conditions should be studied. Prior to the test, pH of background solution was adjusted to 3, 6 and 9 with appropriate HCl and NaOH, respectively. The ion concentration in the system was kept constant by adding NaCl. Then tests were performed in the same way as the batch desorption experiments described above. The initial concentration for each SCOA was 100 mmol/L, and the rhamnolipid concentrations were 50 mg/L.
2.6.
5503
Effect of DOM in the soil
To investigate the influence of soil dissolved organic matter (DOM) on phenanthrene desorption in the presence of biosurfactant and SCOAs, the DOM in the soils was removed, following the method as described by Cookson et al. (2005). The treated soils were contaminated and mildly grounded to pass through the sieve and homogenized. Then desorption tests were done in the presence of rhamnolipid and different SCOAs following the procedures above. The initial
2.7.
Effect of ionic strength
To investigate the influence of ionic strength, the desorption tests were done in the presence of rhamnolipid and different SCOAs following the procedures detailed above. NaCl was added at different concentrations (0.2, 0.3 and 0.4 mol/L). The initial concentration for each SCOA was 100 mmol/L and initial pH value was 6. When the pH level of the solution was adjusted with NaOH or HCl, the increment of ionic strength due to the pH-adjusting solution was corrected.
2.8.
Analytical methods
The concentrations of phenanthrene were determined with analysis by gas chromatography. The GC instrument, a Varian GC 3800-FID system coupled with a Varian 8200 autosampler (CA, USA), was equipped with a 25 m 0.32 mm ID (DB-5) column with 0.25 mm film thicknesses (J&W Scientific Inc., CA). Helium was used as the carrier gas (1.5 mL/min). The temperature program was as follows: the oven temperature was held at 40 C for 1.5 min then ramped to 175 C at a rate of 50 C/min. The temperature was held at 175 C for 1 min, then ramped to 220 C at 7 C/min, and held at this final temperature for 1 min. The injector temperature was 250 C, and the detector temperature was 230 C. Injection was made in the split mode with a split ratio of 1:50 (since 1.75 min).
3.
Results and discussion
3.1. Effect of biosurfactant on the phenanthrene desorption The effect of rhamnolipid on the desorption of phenanthrene in soilewater system was quantified through measuring the phenanthrene concentration in the bulk after desorption equilibrium. Various biosurfactant concentrations were applied to evaluate the performance of rhamnolipid in removing phenanthrene from the contaminated soil. As Fig. 1 shows, in the absence of rhamnolipid, the amount of phenanthrene desorbed was 3.37 mg/kg for the under-plant soil with an organic carbon content of 8.84%, and 7.98 mg/kg for the ditch soil with an organic carbon content of 4.57%. The amount of phenanthrene that could be desorbed in ditch soil was greater than that in under-plant soil. In the presence of rhamnolipid biosurfactant, the desorption of phenanthrene by two soils was noticeably affected. The phenanthrene removal amounts were of 8.76 and 12.24 mg/kg with 50 mg/L rhamnolipid for under-plant and ditch soils, respectively. In the presence of 100 and 200 mg/L rhamnolipid solution, the results showed superior desorption efficiency and the desorbed amounts of phenanthrene were 12.05 and 13.12 mg/kg for under-plant, respectively. The increase of rhamnolipid concentration for ditch soil would result in a similar enhancement of the phenanthrene desorption. Desorption is the predominant process controlling movement of HOCs through the soil matrix. The less desorbed
5504
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
amount of phenanthrene observed in under-plant soil, which has higher contents of organic matter and lower contents of clay, was due to the stronger binding of phenanthrene to under-plant soil. The removal efficiency often showed great dependence on the soil composition and PAH properties (Zhou and Zhu, 2007). The different performance between these two soils might be attributed mainly to their differences in soil texture and organic contents. Biosurfactant concentration is a critical factor for the removal of HOCs from soil. It was observed that increasing the concentration of rhamnolipid could enhance phenanthrene desorption regardless of the soil used (Fig. 1). At concentrations above the CMC, hydrophobic pollutants can readily partition into the hydrophobic core at the center of the micelle, thus increasing HOC aqueous concentration through micelle solubilization and promoting the desorption of HOCs from soils into aqueous phases (Pennell et al., 1997). In the present study, the rhamnolipid biosurfactant with concentration above 50 mg/L were applied to initiate the onset of micellization and avoid the significant adsorption of biosurfactant (Mata-Sandoval et al., 2002; Cheng et al., 2004). Phenanthrene removed from soil in the presence of 100 and 200 mg/L rhamnolipid could achieve a greater desorption efficiency when compared with that in the presence of 50 mg/ L rhamnolipid. The results in this study are similar to previous studies by Lai et al. (2009) who reported that increasing the concentration of rhamnolipid biosurfactant could enhance the removal of total petroleum hydrocarbons from contaminated soil. However, the excessive biosurfactant addition would make the remediation process less economically feasible. Furthermore, since the biodegradation of biosurfactant could compete with the biodegradation of PAHs, too high concentration of biosurfactant would result in the retardation of contaminant bioremediation (Vipulanandan and Ren, 2000). The potential sorption of biosurfactant on soils, which can also show an adverse effect on remediation efficiency, is mainly due to the influence of clays rather than soil organic matters (Ochoa-Loza et al., 2007). Rhamnolipid biosurfactant used in present study could facilitate the phenanthrene removal with high efficiency, thereby being
suitable for application for remediation and biostimulation of PAH-polluted soil.
3.2. Effect of SCOAs on the phenanthrene desorption by biosurfactant The desorption characteristics of phenanthrene by soils were assessed in the presence of rhamnolipid biosurfactant and four SCOAs. Fig. 2 illustrates the desorbed amount of phenanthrene in the soilewater systems under different combinations of agents. Performance with the combined treatments of rhamnolipid and organic acids exhibited an increase in phenanthrene desorption compared to those that had received a treatment of rhamnolipid alone. The data in Fig. 2 also indicate that desorption behaviors of phenanthrene in each soil in the presence of different biosurfactant-SCOA combinations are not similar. Among the combined treatments of biosurfactant and SCOAs, the series with rhamnolipid and citric acid exhibited more significant effect on the desorption performance. The maximum removal of phenanthrene was 10.95 and 16.69 mg/kg for the combination of rhamnolipid and citric acid with under-plant soil and ditch soil, respectively. The rhamnolipid biosurfactant was still effective for the significant removal of phenanthrene from the soils in the presence of SCOAs. The result indicates the presence of SCOAs other than rhamnolipid is responsible for the different behaviors of phenanthrene desorption. The mechanism by which rhamnolipid and SCOAs affected the desorption of phenanthrene was complex. Soil organic matter (SOM) functions as an important partition medium in the retaining of HOCs (Rutherford et al., 1992). The organic matter can bond with the metals strongly in the soil through forming surface complexes. The combined treatments with rhamnolipid and citric acid showed better metal extraction performance than individual application of rhamnolipid or citric acid by increasing ligand availability (Gunawardana et al., 2010). In the presence of biosurfactant, therefore, the use of SCOAs might further enhance the release of these metals and disrupt the linkage between organic matter and soil matrix.
16 Under-plant soil
Ditch soil
14
Desorbed amount (mg/kg)
12 10 8 6 4 2 0 0
50 100 Rhamnolipid concentration (mg/L)
200
Fig. 1 e Effect of rhamnolipid biosurfactant on the phenanthrene desorption.
Fig. 2 e Effect of SCOAs on the phenanthrene desorption by biosurfactant.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
Phenanthrene may be released from the soil along with the desorbed organic matter. This is in accordance with other studies that the presence of organic acid was correlated with the favorable desorption of organic pollutants (White et al., 2006; Zhang and Dong, 2008; An et al., 2010). Moreover, the rhamnolipid biosurfactant could lower the interfacial tension against soil organic fractions in water, facilitating the translocation of organic matter and bounded contaminants into aqueous phase. Then the surfactant molecule could bond with the unbound phenanthrene compounds and dissolve them. Besides soil organic matter, mineral surfaces also play a role in the distribution of PAHs, especially for the soil with low organic matter content. SCOAs are known to interact with inorganic soil particles via different mechanisms (Jones and Brassington, 1998). The adsorption sites on the soil surface might be occupied by organic acids and thus the adsorption of phenanthrene was hindered. In addition, the adsorption of SCOAs could retard the adsorption of the biosurfactant on the soil surface. More surfactant molecules, therefore, will be diffused in the solution and possess a stronger capacity to desorb phenanthrene. Several effects discussed above could all facilitate the desorption process of phenanthrene. Enhanced phenanthrene availability in these combinations could be attributed to the synergistic actions of rhamnolipid and organic acids through potentially different modes of action. The observed removal efficiencies of phenanthrene from soil were in the order of: rhamnolipid and citric acid > rhamnolipid and oxalic acid > rhamnolipid and tartaric acid > rhamnolipid and acetic acid. The differences are mainly due to different chemical structures and properties of organic acids. Among the SCOAs used, for example, citric acid is a ternary organic acid and can provide more anions for complexing than the other unary or binary acids. Such characteristic could lead to the greater efficacy of citric acid in desorbing metals and SOM from soils, carrying more adsorbed phenanthrene molecules meanwhile. Furthermore, ions of binary and ternary organic acids can form chelates with fiveor six-membered ring structure which are more stable (Qin et al., 2004). Thus it can be seen the extent of increment was less for the series with rhamnolipid and acetic acid than that with rhamnolipid and other organic acids. In some other studies, however, it was observed that oxalate acid and acetic acid could enhance more desorption of organic and metal contaminants, respectively (White et al., 2003; Maturi and Reddy, 2008). Such difference is perhaps ascribed to the presence of rhamnolipid biosurfactant, as well as the differences of contaminants and soil properties, which need to be elucidated in further research.
5505
were all higher than that where rhamnolipid had been applied alone. It can be seen that the desorbed phenanthrene increased significantly with the increase of SCOA concentration from 0 to 100 mmol/L, and then increased slightly when further increasing the SCOA concentration from 100 to 300 mmol/L in most series. The desorbed amount with underplant soil reached 9.25, 11.43, 10.89, 11.72 mg/kg in the presence of 100 mg/L rhamnolipid and 300 mmol/L acetic acid, oxalic acid, tartaric acid and citric acid, respectively. A similar trend was also found in other treatments with ditch soils. Additionally, the phenanthrene desorption with the combinations of biosurfactant and different SCOAs showed different increments when organic acid concentrations increased. At the same concentration range of SCOA with ditch soil, the desorbed phenanthrene with acetic acid varied from 12.29 to12.93 mg/kg, which was less remarkable than the series with rhamnolipid and other organic acids. It was observed that the removal of phenanthrene using rhamnolipid and SCOAs gradually increased as the SCOA concentration increased up to a concentration of 300 mmol/L. But the relative change in the amount of desorbed phenanthrene in the presence of both biosurfactant and SCOAs
3.3. Effect of SCOA concentrations on the phenanthrene desorption in the presence of biosurfactant The experiments were based on a series of batch tests with same initial concentration of rhamnolipid but different dosages of SCOAs. Considering the conditions in filed applications (Dermont et al., 2008), initial SCOA concentrations ranged from 0 to 300 mmol/L. As shown in Fig. 3, when rhamnolipid and SCOAs were applied in combination at different ratios, the desorption percentage of phenanthrene
Fig. 3 e Effect of SCOA concentrations on the phenanthrene desorption from (A) under-plant soil and (B) ditch soil in the presence of biosurfactant.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
3.4. Effect of pH on the phenanthrene desorption in the presence of biosurfactant and SCOAs The effect of pH on the desorption of phenanthrene in the presence of rhamnolipid and SCOAs was studied by conducting batch studies at various pH between 3 and 9. Fig. 4 depicts the amount of initial PAHs desorbed from soil into aqueous phase. The data indicates that the desorption behavior of phenanthrene in two soils could be affected by the pH conditions in the system. In the presence of biosurfactant and SCOAs, the extent of phenanthrene desorption was relatively lower at low pH range. At pH 3, the desorption amount was from 4.64 to 5.63 mg/kg for different series with underplant soil. Then the desorption increased sharply from pH 3 to pH 6, but decreased little with the change in pH from pH 6 to pH 9. With the ditch soil, the phenanthrene desorption was also less pronounced at pH 3. Above pH 3, similar trend was observed except that the addition of oxalic, tartaric and citric acids in the presence of rhamnolipid appeared to slightly enhance the desorption of phenanthrene at pH 9 compared with that at pH 6. The amounts of phenanthrene desorbed at pH 9 were 16.34, 15.57, and 17.89 mg/kg in the presence of rhamnolipid along with oxalic, tartaric and citric acids, respectively. At the same pH value, desorption amounts of phenanthrene differentiate for varied combinations. The use of rhamnolipid and citric acid showed better desorption performance than the combinations of rhamnolipid and other organic acids.
The results indicate that pH variation could influence the transfer of phenanthrene from soil to aqueous phase in the presence of surfactant and organic acids. The performance of biosurfactant may be related quantitatively to the surface tension and contaminant solubility, which are pH dependent (Zhang and Miller, 1992; Shin et al., 2004). As for the SCOAs, the possible existing forms of organic acids (dissociated and undissociated) are also strongly related to pH conditions (Jones and Brassington, 1998). Under the combined application of rhamnolipid and SCOAs, the series at pH 3 could achieve a least desorption of phenanthrene when compared with those at higher pH range. This was mainly due to limited ability of rhamnolipid biosurfactant to enhance surface activity and contaminant solubility at low pH. Also, less organic acids could be dissociated from their protons and existed as organic acid molecules. The maximum desorption was achieved when the solution pH increased to 6. At this pH range, both of the surface activity and contaminant solubility with biosurfactant are near to the optimum conditions. Organic acids could provide more ions, resulting in enhanced complexing tendency (Qin et al., 2004). Beyond pH 6, the increase of pH would result in a slight decrease for the desorption efficiency of phenanthrene in
A
12 pH 3
pH 6
pH 9
10 Desorbed amount (mg/kg)
decreased with an increase in the concentration of SCOAs. As the concentration of rhamnolipid was fixed, the enhancement was most likely due to the variation of SCOA amount. At low concentration of SCOAs, the effect of organic acids was not significant and there was no obvious difference in the desorption enhancement for various combinations. At this stage, the enhanced desorption of phenanthrene could be mainly ascribed to the performance of rhamnolipid biosurfactant. At higher concentration of SCOAs in the presence of rhamnolipid, organics acids became more available to destroy the barriers within soil matrix that normally functioned to impede the translocation of organic contaminants. More desorbed phenanthrene was observed at high SCOA concentrations, meaning that the SCOA concentrations in the presence of rhamnolipid at fixed concentration were positively correlated to desorbed amount of phenanthrene from soil. In the absence of biosurfactant, it has been previously noted that desorbed norfloxacin increased with increasing concentrations of citric acid, malic acid and salicylic acid (Zhang and Dong, 2008). In contrast, it was observed that desorbed p,p0 -DDE decreased with organic acids above 50 mmol/L (White et al., 2003). Although the increased SCOA concentrations were favorable for the enhanced desorption of phenanthrene, the increments for each SCOA were different, indicating the nature of organic acid is important for this process. When the amount of organic acid was higher than certain value, the low increasing rate of desorption was possibly due to excess quantities of organic acids approaching the maximum effectiveness to dissociate organic matter and desorb contaminant from soil.
8
6
4
2
0 Rhamnolipid
B
20
D esorbed amount (mg/kg)
5506
15
pH 3
Rhamnolipid + Acetic acid pH 6
Rhamnolipid Rhamnolipid + Oxalic Acid + Tartaric Acid
Rhamnolipid + Citric Acid
Rhamnolipid Rhamnolipid + Oxalic Acid + Tartaric Acid
Rhamnolipid + Citric Acid
pH 9
10
5
0 Rhamnolipid
Rhamnolipid + Acetic acid
Fig. 4 e Effect of pH on the phenanthrene desorption from (A) under-plant soil and (B) ditch soil in the presence of biosurfactant and SCOAs.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
most tests. Although the individual performance of rhamnolipid could be limited at pH above 6, the removal of phenanthrene still reached more than 7.61 and 12.19 mg/kg with combined use of rhamnolipid and SCOAs in our experiment with under-plant and ditch soil, respectively. The high removal efficiency at high pH could be attributable to facilitation by organic acids. Moreover, the system at high pH might lead to increased polarity of the organic material, showing lower affinity for hydrophobic compounds (Ping et al., 2006). At the same time, increases in pH could be accomplished with increased soil-particle dispersion, as well as more disassociated soil minerals and organic matter (You et al., 1999). These factors will contribute to the strong desorption even at high pH range. Furthermore, since soil is negatively charged at near-neutral or higher pH, rhamnolipid, present in anionic forms, are expected to sorb minimally to the particle surface. For both under-plant and ditch soils, elevated pH in combination with rhamnolipid and citric acid had the largest removal efficiency. The variations of phenanthrene desorption at different pH might relate to the different dissociation constants of organic acids involved. The complexing ability of organic acid will increase with enhanced pH, especially for the polyprotic weak acid such as citric acid. The different phenanthrene desorption performances at varied pH may be attributed to effects of biosurfactant and organic acids, as well as the soil constitutes and other system conditions. The relative importance of the different mechanisms depends on the physical and chemical properties of the interactive sorbateesorbent system. Furthermore, based on our results, the optimum pH values for desorption was relatively close to that for microbial growth. It is thus important to also include such environmental concern of pH for in situ application.
3.5. Effect of soil DOM on the phenanthrene desorption in the presence of biosurfactant and SCOAs The desorption amounts of phenanthrene in the presence of biosurfactant and SCOAs with DOM-removed soils are illustrated in Fig. 5. When compared with Fig. 2, the results showed that the phenanthrene desorption differentiated significantly between the treated and untreated soils. In general, the desorbed amounts of phenanthrene for treated soils were less than those for untreated soils. Without the amendments of SCOAs, addition of rhamnolipid enhanced the mobilization of phenanthrene with the desorption of 5.98 and 9.36 mg/kg for DOM-removed under-plant and ditch soils, respectively. The desorption capacities were greater in the presence of both rhamnolipid of SCOAs, irrespective of whether the soil DOM was removed. In the series with rhamnolipid and different organic acids, the desorbed amounts of phenanthrene in the soilewater systems were from 6.13 to 7.66 mg/kg for treated under-plant soils, which were lower than the values of corresponding untreated soils. Similar trend was observed for different combinations of biosurfactant and SCOAs with treated ditch soils, except the phenanthrene desorption was greater in the series with rhamnolipid and tartaric acid as compared with the tests with rhamnolipid and oxalic acid. The desorption of phenanthrene was adversely affected with the use of DOM-removed soil. This difference indicates that some soil properties other than biosurfactant and organic
5507
Fig. 5 e Effect of soil DOM on the phenanthrene desorption in the presence of biosurfactant and SCOAs.
acids can be responsible for the changes of phenanthrene desorption. This result is similar to previous study where adsorption of phenanthrene on soil increased after removal of DOM (Luo et al., 2008). The DOM in soil could exhibit a high affinity for HOCs and enhance the dissolution of organic compounds in soilewater system (Pignatello et al., 2006). With a decreased quantity of DOM in soil, the concentration of DOM in solution decreased so that less phenanthrene was transported into aqueous phase. Binding affinity of organic contaminates to the sediment increased after the removal of soil DOM. In contrast, if all the organic matter is removed, organic contaminants would deposit mainly on mineral surface and exhibits a stronger desorbing capacity (Gao et al., 1998). Although most DOM has been removed, larger amounts of phenanthrene were desorbed from soil in the soilewater systems containing both rhamnolipid and SCOAs than those from soil in the systems containing only rhamnolipid. This is in part due to the mobilization of phenanthrene by rhamnolipid biosurfactant, but also due to the influence of the organic acids on remaining organo-mineral linkages, all of which operate simultaneously. The variation of desorption percentage with different combinations is closely related to the speciation of the organic acids. The lower increases in phenanthrene desorption with rhamnolipid and acetic acid amendments are possibly due to weak complexing ability of organic acid, restricting the diffusion of resident contaminants in soils. These observations suggested that the soil DOM was a critical factor affecting the desorption processes by biosurfactant in presence of organic acids.
3.6. Effect of ionic strength on the phenanthrene desorption in the presence of biosurfactant and SCOAs The phenanthrene desorption influenced by combined use of rhamnolipid and SCOA at different ionic-strength conditions is shown in Fig. 6. The desorption of phenanthrene in the presence of rhamnolipid and different organic acids decreased with increasing concentrations of the ions. At the ionic strength of 0.3 mol/L, the phenanthrene desorption
5508
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
ranged from 7.92 to 10.16 mg/kg and 10.64e14.30 mg/kg for different combinations with under-plant and ditch soils, respectively. Compared with the tests without any amendments, the desorption amount of phenanthrene was also elevated with the addition of biosurfactant and SCOAs at higher ionic-strength range. However, the increase was not very significant as noticed at lower ionic strength. At the ionic strength of 0.4 M NaCl, the phenanthrene desorption efficiency with different combinations was from 6.52 to 9.10 and 8.45e13.77 mg/kg for under-plant and ditch soil, respectively. At the same range of NaCl concentration, ditch soil followed by under-plant soil, showed the greater desorption capacities with different combinations of rhamnolipid and SCOAs. Organic acids often coexist with metal ions in natural environment. The above results indicates that the variation of system ionic strength is likely to influence the overall phenanthrene desorption efficiency with the addition of rhamnolipid and SCOAs. A recent study reported that the surface activity of rhamnolipid was not negatively affected to any degree by high salinities (Abdel-Mawgoud et al., 2009). The size of both mono- and dirhamnolipid biosurfactant vesicles were reduced after the addition of NaCl, thereby improving the water solubility of the biosurfactant (Pornsunthorntawee et al., 2009). However, the findings of present study show
that the high ionic strength values were correlated with the low mobility of phenanthrene. In the solution with high ionic strength, solubility of HOCs might decrease due to the salt-out effect process (Brunk et al., 1996). It was also postulated that PAHs could be less partitioned to dissolved organic substances in the presence of high-concentration metals (Saison et al., 2004). The presence of more salt ions, therefore, can decrease the diffusion of phenanthrene contaminants from soil into aqueous phase. In addition, certain SCOAs could enhance the removal efficiency for phenanthrene which contains rhamnolipid biosurfactant and NaCl at high concentrations. The metal ions showed less inhibitory effect on phenanthrene desorption in biosurfactant-enhanced process when the organic acids were amended. This was of special interest for application to in situ soil washing with high salinity level. Among the different combinations at the same ionic strength, the maximum desorption for phenanthrene was observed in the series with rhamnolipid and citric acid. In the presence of NaCl, the desorption of phenanthrene by two soils was still different. Consequently, other factors, like soil fractions and organic acid characteristics, might also be important for deciding the combined role of biosurfactant and SCOAs in contaminant behaviors within metal ion enrichment.
4.
Fig. 6 e Effect of ionic strength on the phenanthrene desorption from (A) under-plant soil and (B) ditch soil in the presence of biosurfactant and SCOAs.
Conclusions
The study investigated the effect of SCOAs on the desorption of phenanthrene in the presence of rhamnolipid biosurfactant. The results supported the combined use of biosurfactant and SCOAs could further enhance the desorption of phenanthrene from soil into aqueous phase. The overall desorption behavior of phenanthrene could be assumed to be a sum of a large number of individual desorption events with different mechanisms. Desorption amounts of phenanthrene differentiated for varied combinations, in which the effect of ternary and binary SCOAs were more remarkable. SCOAs at high concentration caused a more significant increase in desorption from the soils with the amendment of biosurfactant. The quantity and species of organic acids could affect the tendency of phenanthrene distribution in the presence of biosurfactant. Furthermore, the phenanthrene desorption was higher in the ditch soil compared with the under-plant soil. The variation of desorption with different combinations was also closely related with the soil DOM. Soil property might be an important factor influencing the mobility behaviors of phenanthrene with combined use of biosurfactant and organic acids. In addition, adjusting the pH and ionic strength could control the effect of biosurfactant and organic acids, allowing higher remediation efficiency at specific values. The results presented here demonstrated that the effect of SCOAs on the biosurfactant-enhanced desorption might be directly influenced by the solution chemistry. The results suggest that the combined application of rhamnolipid and SCOAs can be regarded as a good alternative for the removal of PAH compounds from contaminated soil. Further study is also needed to help obtain more theoretical foundation for the PAH mobility with additional biosurfactant and organic acids.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
Acknowledgments This research was supported by the Major Science and Technology Program for Water Pollution Control and Treatment, the Canadian Water Network under the Networks of Centers of Excellence (NCE), and the Natural Science and Engineering Research Council of Canada. We are also grateful for anonymous reviewers for their helpful suggestions and advices.
references
Abdel-Mawgoud, A., Aboulwafa, M., Hassouna, N., 2009. Characterization of rhamnolipid produced by Pseudomonas aeruginosa isolate Bs20. Applied Biochemistry and Biotechnology 157 (2), 329e345. Alcantara-Garduno, M.E., Okuda, T., Tsai, T.Y., Nishijima, W., Okada, M., 2008. Experimental and mathematical evaluation of trichloroethylene removal from saturated soil using acetic acid with saturated ozone. Separation and Purification Technology 60 (3), 299e307. An, C.J., Huang, G.H., Yu, H., Wei, J., Chen, W., Li, G.C., 2010. Effect of short-chain organic acids and pH on the behaviors of pyrene in soilewater system. Chemosphere 81 (11), 1423e1429. Blank, R.R., Allen, F., Young, J.A., 1994. Extractable anions in soils following wildfire in a sagebrush-grass community. Soil Science Society of America Journal 58 (2), 564e570. Bordas, F., Lafrance, P., Villemur, R., 2005. Conditions for effective removal of pyrene from an artificially contaminated soil using Pseudomonas aeruginosa 57SJ rhamnolipids. Environmental Pollution 138 (1), 69e76. Brunk, B.K., Jirka, G.H., Lion, L.W., 1996. Effects of salinity changes and the formation of dissolved organic matter coatings on the sorption of phenanthrene: implications for pollutant trapping in estuaries. Environmental Science & Technology 31 (1), 119e125. Cheng, K.Y., Zhao, Z.Y., Wong, J.W.C., 2004. Solubilization and desorption of PAHs in soil-aqueous system by biosurfactants produced from Pseudomonas aeruginosa P-CG3 under thermophilic condition. Environmental Technology 25 (10), 1159e1165. Chu, W., Chan, K.H., 2003. The mechanism of the surfactantaided soil washing system for hydrophobic and partial hydrophobic organics. Science of the Total Environment 307 (1e3), 83e92. Clifford, J.S., Ioannidis, M.A., Legge, R.L., 2007. Enhanced aqueous solubilization of tetrachloroethylene by a rhamnolipid biosurfactant. Journal of Colloid and Interface Science 305 (2), 361e365. Cookson, W.R., Abaye, D.A., Marschner, P., Murphy, D.V., Stockdale, E.A., Goulding, K.W.T., 2005. The contribution of soil organic matter fractions to carbon and nitrogen mineralization and microbial community size and structure. Soil Biology and Biochemistry 37 (9), 1726e1737. Dermont, G., Bergeron, M., Mercier, G., Richer-Lafle`che, M., 2008. Soil washing for metal removal: a review of physical/chemical technologies and field applications. Journal of Hazardous Materials 152 (1), 1e31. Deshpande, S., Shiau, B.J., Wade, D., Sabatini, D.A., Harwell, J.H., 1999. Surfactant selection for enhancing ex situ soil washing. Water Research 33 (2), 351e360. Deshpande, S., Wesson, L., Wade, D., Sabatini, D.A., Harwell, J.H., 2000. DOWFAX surfactant components for enhancing contaminant solubilization. Water Research 34 (3), 1030e1036.
5509
Djomo, J.E., Dauta, A., Ferrier, V., Narbonne, J.F., Monkiedje, A., Njine, T., Garrigues, P., 2004. Toxic effects of some major polyaromatic hydrocarbons found in crude oil and aquatic sediments on Scenedesmus subspicatus. Water Research 38 (7), 1817e1821. Ferrarese, E., Andreottola, G., Oprea, I.A., 2008. Remediation of PAH-contaminated sediments by chemical oxidation. Journal of Hazardous Materials 152 (1), 128e139. Gan, S., Lau, E.V., Ng, H.K., 2009. Remediation of soils contaminated with polycyclic aromatic hydrocarbons (PAHs). Journal of Hazardous Materials 172 (2e3), 532e549. Gao, J.P., Maguhn, J., Spitzauer, P., Kettrup, A., 1998. Sorption of pesticides in the sediment of the Teufelsweiher pond (Southern Germany). I: equilibrium assessments, effect of organic carbon content and PH. Water Research 32 (5), 1662e1672. Gomez, J., Alcantara, M.T., Pazos, M., Sanroman, M.A., 2010. Remediation of polluted soil by a two-stage treatment system: desorption of phenanthrene in soil and electrochemical treatment to recover the extraction agent. Journal of Hazardous Materials 173 (1e3), 794e798. Gunawardana, B., Singhal, N., Johnson, A., 2010. Amendments and their combined application for enhanced copper, cadmium, lead uptake by Lolium perenne. Plant and Soil 329 (1e2), 283e294. Jones, D.L., 1998. Organic acids in the rhizosphere e a critical review. Plant and Soil 205 (1), 25e44. Jones, D.L., Brassington, D.S., 1998. Sorption of organic acids in acid soils and its implications in the rhizosphere. European Journal of Soil Science 49 (3), 447e455. Kanaly, R.A., Harayama, S., 2000. Biodegradation of highmolecular-weight polycyclic aromatic hydrocarbons by bacteria. Journal of Bacteriology 182 (8), 2059e2067. Kanga, S.A., Bonner, J.S., Page, C.A., Mllls, M.A., Autenrieth, R.L., 1997. Solubilization of naphthalene and methyl-substituted naphthalenes from crude oil using biosurfactants. Environmental Science & Technology 31 (2), 556e561. Khodadoust, A.P., Bagchi, R., Suidan, M.T., Brenner, R.C., Sellers, N.G., 2000. Removal of PAHs from highly contaminated soils found at prior manufactured gas operations. Journal of Hazardous Materials 80 (1e3), 159e174. Lai, C.C., Huang, Y.C., Wei, Y.H., Chang, J.-S., 2009. Biosurfactantenhanced removal of total petroleum hydrocarbons from contaminated soil. Journal of Hazardous Materials 167 (1e3), 609e614. Luo, L., Zhang, S.Z., Ma, Y.B., 2008. Evaluation of impacts of soil fractions on phenanthrene sorption. Chemosphere 72 (6), 891e896. Mata-Sandoval, J.C., Karns, J., Torrents, A., 2002. Influence of rhamnolipids and Triton X-100 on the desorption of pesticides from soils. Environmental Science & Technology 36 (21), 4669e4675. Maturi, K., Reddy, K.R., 2008. Extractants for the removal of mixed contaminants from soils. Soil & Sediment Contamination 17 (6), 586e608. Mulligan, C.N., 2005. Environmental applications for biosurfactants. Environmental Pollution 133 (2), 183e198. Nadal, M., Schuhmacher, M., Domingo, J.L., 2004. Levels of PAHs in soil and vegetation samples from Tarragona County, Spain. Environmental Pollution 132 (1), 1e11. Noordman, W.H., Ji, W., Brusseau, M.L., Janssen, D.B., 1998. Effects of rhamnolipid biosurfactants on removal of phenanthrene from soil. Environmental Science & Technology 32 (12), 1806e1812. Ochoa-Loza, F.J., Noordman, W.H., Jannsen, D.B., Brusseau, M.L., Maier, R.M., 2007. Effect of clays, metal oxides, and organic matter on rhamnolipid biosurfactant sorption by soil. Chemosphere 66 (9), 1634e1642.
5510
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 0 1 e5 5 1 0
Pennell, K.D., Adinolfi, A.M., Abriola, L.M., Diallo, M.S., 1997. Solubilization of dodecane, tetrachloroethylene, and 1,2dichlorobenzene in micellar solutions of ethoxylated nonionic surfactants. Environmental Science & Technology 31 (5), 1382e1389. Pignatello, J.J., Kwon, S., Lu, Y., 2006. Effect of natural organic substances on the surface and adsorptive properties of environmental black carbon (char): attenuation of surface activity by humic and fulvic acids. Environmental Science & Technology 40 (24), 7757e7763. Ping, L.F., Luo, Y.M., Wu, L.H., Qian, W., Song, J., Christie, P., 2006. Phenanthrene adsorption by soils treated with humic substances under different pH and temperature conditions. Environmental Geochemistry and Health 28 (1e2), 189e195. Pornsunthorntawee, O., Chavadej, S., Rujiravanit, R., 2009. Solution properties and vesicle formation of rhamnolipid biosurfactants produced by Pseudomonas aeruginosa SP4. Colloids and Surfaces B e Biointerfaces 72 (1), 6e15. Qin, F., Shan, X.Q., Wei, B., 2004. Effects of low-molecular-weight organic acids and residence time on desorption of Cu, Cd, and Pb from soils. Chemosphere 57 (4), 253e263. Rutherford, D.W., Chiou, C.T., Kile, D.E., 1992. Influence of soil organic matter composition on the partition of organic compounds. Environmental Science & Technology 26 (2), 336e340. Saison, C., Perrin-Ganier, C., Amellal, S., Morel, J.L., Schiavon, M., 2004. Effect of metals on the adsorption and extractability of 14 C-phenanthrene in soils. Chemosphere 55 (3), 477e485. Shin, K.H., Kim, K.W., Seagren, E.A., 2004. Combined effects of pH and biosurfactant addition on solubilization and biodegradation of phenanthrene. Applied Microbiology and Biotechnology 65 (3), 336e343. Silva, A., Delerue-Matos, C., Fiu´za, A., 2005. Use of solvent extraction to remediate soils contaminated with hydrocarbons. Journal of Hazardous Materials 124 (1e3), 224e229. van Hees, P.A.W., Vinogradoff, S.I., Edwards, A.C., Godbold, D.L., Jones, D.L., 2003. Low molecular weight organic acid adsorption in forest soils: effects on soil solution concentrations and biodegradation rates. Soil Biology and Biochemistry 35 (8), 1015e1026. Vipulanandan, C., Ren, X.P., 2000. Enhanced solubility and biodegradation of naphthalene with biosurfactant. Journal of Environmental Engineering-ASCE 126 (7), 629e634.
Wang, Y.J., Chen, J.H., Cui, Y.X., Wang, S.Q., Zhou, D.M., 2009. Effects of low-molecular-weight organic acids on Cu(II) adsorption onto hydroxyapatite nanoparticles. Journal of Hazardous Materials 162 (2e3), 1135e1140. Wei, J., Huang, G.H., Yu, H., An, C.J., 2011. Efficiency of single and mixed gemini/conventional micelles on solubilization of phenanthrene. Chemical Engineering Journal 168 (1), 201e207. Wei, Q.F., Mather, R.R., Fotheringham, A.F., 2005. Oil removal from used sorbents using a biosurfactant. Bioresource Technology 96 (3), 331e334. White, J.C., Mattina, M.I., Lee, W.Y., Eitzer, B.D., IannucciBerger, W., 2003. Role of organic acids in enhancing the desorption and uptake of weathered p, p0 -DDE by Cucurbita pepo. Environmental Pollution 124 (1), 71e80. White, J.C., Parrish, Z.D., Isleyen, M., Gent, M.P.N., IannucciBerger, W., Eitzer, B.D., Kelsey, J.W., Mattina, M.I., 2006. Influence of citric acid amendments on the availability of weathered PCBs to plant and earthworm species. International Journal of Phytoremediation 8 (1), 63e79. You, S.J., Yin, Y., Allen, H.E., 1999. Partitioning of organic matter in soils: effects of pH and water/soil ratio. The Science of the Total Environment 227 (2e3), 155e160. Yu, H., Huang, G.H., Wei, J., An, C.J., 2011. Solubilization of mixed polycyclic aromatic hydrocarbons through a rhamnolipid biosurfactant. Journal of Environmental Quality 40 (2), 477e483. Yuan, S.H., Xi, Z.M., Jiang, Y., Wan, J.Z., Wu, C., Zheng, Z.H., Lu, X. H., 2007. Desorption of copper and cadmium from soils enhanced by organic acids. Chemosphere 68 (7), 1289e1297. Zhang, J.Q., Dong, Y.H., 2008. Effect of low-molecular-weight organic acids on the adsorption of norfloxacin in typical variable charge soils of China. Journal of Hazardous Materials 151 (2e3), 833e839. Zhang, W., Tsang, D.C.W., Lo, I.M.C., 2008. Removal of Pb by EDTA-washing in the presence of hydrophobic organic contaminants or anionic surfactant. Journal of Hazardous Materials 155 (3), 433e439. Zhang, Y., Miller, R.M., 1992. Enhanced octadecane dispersion and biodegradation by a Pseudomonas rhamnolipid surfactant (biosurfactant). Applied and Environmental Microbiology 58 (10), 3276e3282. Zhou, W.J., Zhu, L.Z., 2007. Efficiency of surfactant-enhanced desorption for contaminated soils depending on the component characteristics of soil-surfactant- PAHs system. Environmental Pollution 147 (1), 66e73.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Nitrification and potential control mechanisms in simulated premises plumbing Mohammad Shahedur Rahman a,1, Gem Encarnacion a,b, Anne K. Camper a,c,* a
Center for Biofilm Engineering, Montana State University, 366 EPS Building, Bozeman, MT 59717, USA Department of Microbiology, Montana State University, Bozeman, MT 59717, USA c Department of Civil Engineering, Montana State University, Bozeman, MT 59717, USA b
article info
abstract
Article history:
Indigenous drinking water organisms were used to establish nitrification in glass reactors
Received 7 February 2011
containing copper or polyvinyl chloride (PVC) surfaces. The reactors were fed soil-derived
Received in revised form
humics as the organic carbon source and ammonium sulfate as the nitrogen source in
19 July 2011
biologically treated tap water. Water in the reactors was stagnant for 8 h and then flowed
Accepted 7 August 2011
for 5 min to simulate conditions in household plumbing. Following the establishment of
Available online 16 August 2011
complete nitrification (conversion of ammonia to nitrate) in both reactor types, various inhibitors of nitrification were tested followed by a period where recovery of nitrification
Keywords:
was observed. In one PVC reactor, copper was gradually introduced up to 1.3 ppm. To
Nitrification
ensure that most of the copper was in the ionic form, the pH of the influent was then
Biofilms
gradually lowered to 6.6. No significant change in nitrification was observed in the pres-
Chlorite
ence of copper. Chlorite was introduced into copper and PVC reactors at doses increasing
Chloramines
from 0.2 ppm to 20 ppm. There was limited effect on the PVC system and inhibition in the
Copper
copper reactor only at 20 ppm. Chloramine was tested at chlorine to ammonia ratios
Plumbing
ranging from 0.5:1 to 5:1. Nitrification activity was impacted significantly at a 5:1 ratio and ultimately stopped, with the fastest response being in the copper system. Whenever a control mechanism was tested, there was increased release of copper from the reactors with copper coupons. In all cases, nitrification recovered when inhibitors were removed but the rates of recovery differed depending on the treatment method and coupon surface. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
A topic of regulatory and public health concern in drinking water is the creation of potentially carcinogenic disinfection by-products (DBPs) when the water is chlorinated. As an alternative to chlorination, many utilities have shifted to the use of chloramines to reduce the levels of regulated DBPs to meet the Stage 2 Disinfectants/Disinfection By-Product Rule (USEPA, 2000). Although the use of chloramines as a secondary
disinfectant can reduce DBPs, there is the chance that increased levels of free ammonia in the distribution system may serve as an energy source for indigenous nitrifying organisms. Proliferation of these organisms can then result in nitrification in the distribution system. Not surprisingly, nitrification is one of the most frequent operational problems encountered by drinking water utilities that use chloramine for secondary disinfection (Skadsen, 1993; Wolfe and Lieu, 2001; Odell et al., 1996; Wilczak et al., 1996; Seidel et al., 2005).
* Corresponding author. Center for Biofilm Engineering, Montana State University, 366 EPS Building, Bozeman, MT 59717, USA. Tel.: þ1 406 994 4906; fax: þ1 406 994 6098. E-mail address: [email protected] (A.K. Camper). 1 Current address: Anderson Civil Consultants Nanaimo, BC V9R 2T5, Canada. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.009
5512
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
Nitrification is a microbial process by which reduced nitrogen compounds (primarily ammonia) are sequentially oxidized to nitrite and nitrate. This process can have a detrimental impact on water quality. Due to nitrification, chloramine residual, pH, alkalinity and dissolved oxygen of water decrease. As nitrification can cause a decrease in pH, some utilities may be susceptible to elevated levels of soluble metal contaminants such as lead (Zhang et al., 2009), leading to Lead and Copper Rule (LCR) violations. Nitrification can also cause biological instability through production of soluble microbial products which may support the growth of heterotrophic bacteria in low nutrient environments (Rittmann et al., 1994). Studies on nitrification in drinking water have mostly been done in distribution mains or at the treatment plant level and studies on premises plumbing are lacking. Premises plumbing not only has higher surface to volume ratios but also has about 10 times more length than water mains (NRC, 2006). In addition, it stands to reason that favorable conditions for nitrification such as low or no disinfectant, long water age and warmer temperatures exist in premises plumbing. This gap in knowledge led to investigations on possible control for nitrification in premises plumbing. The studies took place in laboratory reactors designed to simulate the surface area to volume ratios, flow conditions, and water quality that may be encountered in these systems. Ammonia concentrations chosen represented a worst case scenario where all the chloramine added at the regulatory limit (4 ppm) had decayed to release ammonia. Two commonly utilized premises plumbing materials, copper and polyvinyl chloride (PVC), were used. Copper is the most widely used metal for household plumbing systems and more than 90% of domestic plumbing material in the US is made of copper (Oskarsson and Norrgren, 1998) while PVC pipes are also a very common plumbing material (NSF, 2008). Three separate strategies based on realistic approaches for drinking water systems were evaluated. First, the inhibitory effect of copper was examined by (1) comparing nitrification in PVC vs copper reactors and (2) by introducing known amounts of copper (gradually increasing concentration) into a nitrifying reactor with PVC coupons. Previous work has shown that nitrification in pure cultures is either enhanced or inhibited by copper, depending on its concentration. Loveless and Painter (1968) found that 0.005e0.03 ppm of Cuþ2 stimulated the growth of the ammonia oxidizer Nitrosomonas while Skinner and Walker (1961) observed enhanced growth at higher concentrations of 0.1e0.5 ppm copper. However, these higher concentrations were found to be inhibitory by Loveless and Painter (1968). Zhang and Edwards (2005) observed slight inhibition of nitrification for pure cultures in the presence of 5 ppb copper while 25 ppb copper had a slightly stimulatory effect. In the same study, a much higher concentration of 500 ppb copper significantly inhibited nitrification. Zhang and Edwards (2010) also reported inhibition of nitrification at copper levels greater than 100 mg/L. In cases where there is free ammonia, coppereammonia complexes such as [Cu þ2 (NH3)2þ X ], (Sato et al., 1988) and cupric tetraamine [Cu(NH3)4 ] (Lee et al., 1997) may also be responsible for inhibition of nitrifiers. It is also important to know the form in which the copper exists because Cuþ2 ions are believed to be responsible for the inhibition of nitrifying bacteria (Braam and Klapwijk,
1981; Hu et al., 2003). Cupric ions in the vicinity of the cell membrane may cause damage by depolarization and impairment of receptors or transporter molecules (Alt et al., 1990), and may bind proteins and block the function of proteins in the exoploymeric substance (EPS) of the bacteria (Geesey and Jang, 1989). Ion speciation can be controlled by the pH (Braam and Klapwijk, 1981; Edwards et al., 1996), and this was considered as part of the experimental design. A second set of experiments was done to determine the impact of chlorite on nitrification. The use of chlorite as a control mechanism for nitrification has been proposed for full scale distribution systems and storage tanks. Hynes and Knowles (1983) showed that chlorite interfered with the first step of nitrification, specifically in the oxidation of ammonia to nitrite by Nitrosomonas europea. Several studies in both laboratory and full scale drinking water distribution systems have been conducted to investigate the effect of chlorite on nitrification. McGuire et al. (1999) found that low levels (0.2 ppm) of chlorite caused a significant reduction in the culturability of ammonia oxidizing bacteria (AOB). In the same study, the experience of the Gulf Coast (Texas) Water Authority (GCWA) which uses chlorine dioxide as the primary disinfectant and chloramine as the secondary disinfectant was reported. Chlorite was detected in their distribution system in the range of 0.25e0.35 ppm and although the conditions were conducive to nitrification, it did not occur suggesting the inhibitory effect of chlorite to nitrification. In another study conducted in plug flow reactors in Tucson AZ, continuous feed of chlorite at concentrations as low as 0.1 ppm was found to prevent nitrification (McGuire et al., 1999). Chlorite was also used to prevent nitrification in parts of the Glendale, CA distribution system (McGuire et al., 2009). Interestingly, chlorite has been found to be ineffective in controlling nitrification in other studies. McGuire et al. (1999) reported a nitrification episode in the Corpus Christi Texas distribution system that continued even after dosing with chlorite. Similarly, Karim and LeChevallier (2006) reported the recurrence of nitrification in a pilot system where initially the use of 0.5 ppm chlorite controlled nitrification. In another study the presence of chlorite in water reservoirs prevented the onset of nitrification, but once nitrification started, introducing chlorite was not effective and a 0.2 ppm dose of chlorite in a nitrifying reservoir inhibited nitrification for only two weeks (McGuire et al., 2006). To investigate the impact of chlorite on nitrification in the simulated premises plumbing system, doses were incrementally increased from 0.2 to 20 ppm. A third potential control mechanism was maintaining a chloramine disinfectant residual within the plumbing system. Chloramine, although considered as a weaker disinfectant than chlorine for suspended cells, is thought to be more effective for disinfecting biofilms (LeChevallier et al., 1988; Wahman et al., 2009), and since most nitrifying bacteria are present as biofilms rather than planktonic cells in both natural and engineered systems (Schramm et al., 1996; Okabe et al., 2005; Kindaichi et al., 2006), the use of chloramine was also examined. In an actual water distribution system, it may be possible to either distribute water with a stable residual or recreate chloramine through booster chlorination. To investigate this experimentally, chlorine was applied to attain different chlorine to ammonia ratios.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
All of these experiments took place in reactors that had been undergoing complete nitrification (conversion of ammonia to nitrate) for one year. Importantly, the reactors had undefined mixed population biofilms originating from Bozeman tap water and were not inoculated with specific nitrifying organisms. The operating conditions were carefully chosen to represent conditions in premises plumbing while allowing for meaningful sampling strategies and control of variables. During each test, effluent copper concentrations were measured to assess the influence of that strategy on copper release. As such, the results provide insights on how indigenous nitrifying biofilm communities respond to potential control strategies and how these strategies influence effluent water quality.
2.
Materials and methods
2.1.
Reactors
Domestic plumbing systems were simulated using a modified CDC reactor (Goeres et al., 2005). Modifications included adding parallel coupons to solid rods, adding a base plate, and changing the stir blades to either polyvinyl chloride (PVC) or copper (Fig. 1). The surface area of the coupons, base plate and stir blades was calculated to create the same surface to volume ratio as that of a six foot long 3/4 ” diameter domestic plumbing pipe. The PVC or copper was washed with 0.1 N NaOH three times prior to use. Volume of the reactors is 120 ml.
Fig. 1 e Modified CDC reactor showing the paired copper coupons (1.5 3 1.7 cm inside, 1.5 3 1.3 cm outside on the rods), base plate (bottom of the reactor) and the blade in the center.
2.2.
5513
Operational scheme
To simulate periods of stagnation in home plumbing the reactors were flushed with peristaltic pumps with shear created by the stirplate for 5 min followed by 8 h stagnation periods. The feed pumps and stirplates were controlled by timers that were offset by 1 min with the stirplate starting before the pumps. At the end of 5 min the stirplates stopped, followed by the pumps. The stirplates were set to create a rotational speed of the blade of 300 rpm, which was approximately equivalent to a velocity of 3 ft/s in the bulk water. The cycle was repeated three times per day. Two sets (four reactors each) of modified CDC reactors equipped with two different types of coupons (PVC and copper) were used in this investigation. All reactors had been in operation for more than one year and showed signs of stable, complete nitrification as measured by conversion of ammonia to nitrate.
2.3.
Stock/feed solution preparation
All reactors were fed with a combination of mineral amended reverse osmosis (RO) water, biologically treated Bozeman tap water, and a humic substances organic feed. The RO water was amended to create an alkalinity of 35 mg/L as CaCO3 and a stable pH of 8.15. Constituents of the RO water þ mineral feed were MgSO4 (39.6 mg/L), NaHCO3 (59.6 mg/L), CaSO4$2H2O (25 mg/L), Al2(SO4)$18H2O (0.62 mg/L), CaCl2$2H2O (20.80 mg/ L), and Na2SiO3$9H2O (26 mg/L). Ammonium sulfate was added to provide a final concentration in the reactor of 0.71 mg/L as N. Biologically treated water provided as a separate, parallel influent. It was created by flowing Bozeman tap water (surface water source, no background ammonia, chlorinated) through a granular activated carbon column followed by flow through a biologically active carbon column to provide a continuous inoculum of indigenous organisms (104 CFU/ml of heterotrophic plate count (HPC)) that was the only source of microorganisms to the reactors. This water also contained sufficient background phosphate for microbial growth at the organic carbon level used. Organic carbon was supplied to the reactors in a third, separate influent in the form of soil-derived humic substances. 50 grams of Elliot silt loam soil (International Humic Substances Society) was added to 500 ml of 0.1 N NaOH and mixed for 48 h. This solution was centrifuged at 10,000 g for 20 min. The supernatant was collected in carbon free glassware (baked at 390 C for 5 h) and stored at 4 C in the dark. Total organic carbon content of the humics was measured using a Dohrman DC-80 and subsequently diluted to the appropriate concentration using the RO water feeding the reactors to provide a concentration in the reactors of 4 mg/ L as dissolved organic carbon. The RO water/biologically treated Bozeman tap water/organic carbon feed ratio was 50:5:1. Other amendments and modifications of the influent feed are described below for each set of specific experiments.
2.4.
Sampling
Water was collected from the reactors at the end of the 8 h stagnation period three times weekly. Samples were analyzed
5514
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
for ammonia, nitrate, and nitrite. Weekly samples were analyzed for heterotrophic plate counts, ammonia oxidizing bacteria, and nitrite oxidizing bacteria. Free NH3eN was measured using a HACH 2000 spectrophotometer using the salicylate method (HACH method 10023) at 655 nm immediately after the samples were collected. Nitrite was analyzed using a HACH 2000 spectrophotometer and the diazotization method. Reaction of nitrite with sulfanilic acid and forms an intermediate diazonium salt that couples with chromotropic acid to produce a pink colored complex measured at 507 nm. Nitrate in filtered samples (0.2 mm pore size polyethersulfone) was measured using a Dionex ion chromatography system with a CD20 conductivity detector and GP40 gradient pump unit. An AS4A column and DS3 detection stabilizer was also used in this method. The Dionex ion chromatography system was calibrated using five sodium nitrate standards (1, 0.5, 0.2, 0.1, 0 ppm of NO3eN). To minimize experimental error, after every seven measurements a standard solution of nitrate was measured to check the accuracy of the measurement. If the obtained measurement of the standard was outside 90e110% of the standard value then the calibration was repeated and sample was measured again (Standard Methods, 1995). Heterotrophic plate counts were done according to Standard Methods (1995) 9215A using R2A agar plates. Plates were incubated at 20 C for 7 days, and then the number of colonies in the plates was counted using a Quebec colony counter. For the chloramine experiments, the disinfectant was neutralized with sodium thiosulfate prior to dilution and plating. Ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB) populations were enumerated using the most probable number (MPN) technique (Rowe et al., 1977) using Costar Clear-Bottom 96 well microtiter plates. The mineral medium used for AOB contained per liter: (NH4)2SO4, 330 mg; KH2PO4, 100 mg; MgSO4$7H2O 40 mg; CaCl2, 15 mg and 1 ml of a trace-element solution. The trace-element solution contained per liter: Na2EDTA, 4292 mg; FeCl2$4H2O, 1988 mg; MnCl2$H2O, 99 mg; NiCl2$6H2O, 24 mg; CoCl2$6H2O, 24 mg; CuCl2$2H2O, 17 mg; ZnCl2, 68 mg; Na2MoO4$2H2O, 24 mg and H3BO3, 62 mg. Bromothymol blue (5ml/L of 0.04% solution in water) was added as a pH indicator. The pH was adjusted to 8 using 1 M NaOH before autoclaving at 110 C for 15 min. The NOB medium had the same composition except that it did not contain (NH4)SO4 and bromothymol blue, and was supplemented with 34.5 mg/L NaNO2. The pH was adjusted to 6.5 with 1 M NaOH before autoclaving at 110 C for 15 min. After inoculation the microtiter plates were sealed with polyester tape to prevent evaporation and incubated for nine weeks at 20 C in the dark. After the incubation period AOB and NOB presence was determined by detecting nitrite and nitrate in the medium by adding 40 ml of 0.2% diphenylamine in H2SO4 in the well. The absorbance of the blue color was measured at 630 nm with a microplate reader (EL808 ultra microplate reader BioTek instruments). The blue color indicated nitrite or nitrate had formed and the well was scored as positive. Griess Ilosvay reagent (Alexander and Clark, 1965) was used to detect nitrite. A red color was produced within 5 min and absorbance measured after 5 min at 540 nm with the microplate reader. If nitrite in the well was detected it was scored as
negative for NOB. The difference between the two measurements then allowed for differentiation between AOB and NOB. The MPNs were calculated according to Rowe et al. (1977). Biofilm samples were collected at the end of each experiment’s test period prior to the recovery phase. One coupon was removed from the reactor and placed in a glass tray (baked at 390 C for 5 h) containing the autoclaved RO water. The coupon was then scraped using an autoclaved rubber policeman inside a laminar flow hood. The biomass with water was poured in a sterilized 50 ml Falcon tube, which was then homogenized with a homogenizer (Biohomogenizer Model M133/12810, ESGE) for 30 s. From the homogenized biomass, samples were taken for MPN and HPC analysis.
2.5.
Statistical analysis
Paired t-test analysis was done using Microsoft Excel on the data to see if there are significant differences between two treatments. The level of significance for all tests was a ¼ 0.05.
2.6.
Experimental approach
2.6.1.
Effect of copper ion on nitrification
To test the effect of copper on nitrification, copper in the form of CuSO4 was added to the influent of one nitrifying PVC reactor. Initially15 ppb copper was continually dosed and incrementally raised to 1.3 ppm, which is the action limit for copper according to the Lead and Copper Rule. Each concentration of copper was maintained for two weeks. Another PVC reactor was used as the control where no copper was added. Because nitrification is presumed to be more affected by free copper (Cuþ2) (Braam and Klapwijk, 1981), and the amount of Cuþ2 in solution is a function of pH, the pH of the influent with 1.3 ppm of copper was then reduced gradually from 8.15 to 6.6 by 0.3 units every two weeks. Another PVC reactor with the influent at the same pH but with no copper was used a control. When copper was added to the PVC reactor at lower concentrations (less than 400 ppm) an ICPMS (inductively coupled plasma mass spectrometer) was used to measure the copper. At higher concentrations a HACH 2000 spectrophotometer (method 8506,560 nm wavelength) was used. Both total and dissolved copper in the effluent water were measured. Dissolved copper is operationally defined as the portion of the copper which passes through a 0.45 mm pore size syringe filter. In the presence of colloidal species that can pass through the filter, the method represents an upper bound to truly soluble copper. When the pH was adjusted, a copper ion selective electrode (Cu-ISE), (Orion cupric electrode, model, 94e29, Boston, MA) was used to measure the free copper (Cuþ2) in the water. The electrode was calibrated using standard cupric ion solutions according to manufacturer’s direction before measuring the sample. Cupric ion at 1.3 ppm total copper and pH 6.6 was measured and was found to be in the 0.90 0.10 ppm range.
2.6.2.
Effect of chlorite on nitrification
Laboratory grade sodium chlorite was added to the influent of one PVC and one copper reactor. Initially 0.2 ppm of chlorite was added and gradually increased to 2.0 ppm and then
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
a shock load of 20 ppm of chlorite was dosed. Each concentration of chlorite was tested for two weeks. Chlorite was measured by a modification (McGuire et al., 1999) of the US Environmental Protection Agency (USEPA, 1993) method 300 with the Dionex ion chromatography system described above equipped with an AS9 column and 100 ml sampling loop. Chlorite was then discontinued and the recovery of nitrification in both the PVC and copper system was observed. Copper concentrations in the copper reactor effluent were also monitored.
2.6.3.
Effect of chloramine on nitrification
To represent different scenarios that can occur in a distribution system due to chloramine decay, different amounts of chlorine were added in the ammonia-containing RO water influent of two nitrifying copper and PVC reactors. The added sodium hypochlorite formed chloramine with the ammonia in the influent at different chlorine to ammonia ratios. Initially a 0.5:1 ratio was applied and gradually raised to 5:1. Both free and total chlorine were measured using a Lamotte DC1100 colorimeter with the DPD colorimetric method (Standard Methods, 1995, method 4500-Cl). At the beginning of the measurement phosphate buffer was added to the sample to maintain a pH between 6.2 and 6.5. Dissolved and total copper in the copper reactor effluents were also measured. At the end of chloramine exposure, chlorine was discontinued and recovery of nitrification in PVC and copper reactors was compared.
compared to the control until 600 ppb was reached. After 600 ppb added copper, the difference was significant ( p < 0.05) but actual measurements differed by a maximum of only 0.05 mg/L NH3eN (n ¼ 6 for each copper dose). A similar trend was seen with nitrate measurements; there were very small differences with a maximum variation of 0.04 mg/L NO3eN between the control and copper dosed systems. After the highest copper dose was reached, and to ensure all copper was as Cuþ2, the pH of the reactor was gradually lowered to 6.6. Another PVC reactor without copper where the pH was also adjusted was used as a control. There were statistically significant differences in effluent ammonia and nitrate levels at all pH values (7.8e6.6 at 0.3 increments) but the actual amounts were no more than 0.04 mg/L NH3eN or NO3eN. In all cases, there was no more than 0.26 mg/L NH3eN in the control or copper exposed reactor effluents (range 0.16e0.26) and NO3eN ranged from 0.65 to 0.43 mg/L. Only trace amounts of nitrite were measured (data not shown). There were no significant differences in AOB or NOB MPN counts in the reactor effluents from the copper vs control reactors. Values ranged from 7 to 35 MPN/ml for AOB and from 4 to 41 MPN/ml for NOB. These results are reported because utilities will typically only measure bulk water nitrifier numbers since biofilm interrogation is not feasible. Similarly, there were no differences in the HPC values, with the range from 4 104 to 6 105 per ml. Increasing copper doses were not statistically correlated with MPN or HPC values.
3.2.
3.
Results
All experiments were done in reactors where stable, complete nitrification (conversion of ammonia to nitrate) had been occurring for one year. Before initiation of the experiments, there was no difference in ammonia conversion between PVC and copper, and there was good reproducibility in effluent water quality between pairs of reactors with the same coupon materials (data not shown). All testing consisted of a phase where the reactors were exposed to a stepwise change in water quality (addition of copper/change of pH, addition of chlorite, or increase in chlorine to ammonia ratio) and compared to a control reactor with stable influent water quality followed by a recovery period where complete nitrification was again established. As such, the data demonstrated the efficacy of the treatment method in addition to the robustness of the nitrification process.
3.1.
5515
Effect of chlorite on nitrification
Because chlorite has been used in full scale distribution systems to control nitrification with varying effect, this chemical was added in increasing doses (0.2e20 ppm) to nitrifying PVC and copper reactors. Reactors that did not receive chlorite were maintained as controls. As seen in Fig. 2, low range chlorite (i.e., 0.2e2 ppm) did not affect nitrification in the PVC reactor. When the dose was increased to 20 ppm the PVC reactor was slightly affected as it temporarily had lower NH3eN utilization, but even in the presence of chlorite, complete ammonia loss rebounded. The temporary decreases in NH3eN utilization may have been the result of detaching biomass, adaptation to the chlorite, or other unknown factors.
Effect of copper on nitrification
To eliminate the influence of the copper substratum on the results, these experiments were conducted with copper ion added to a PVC reactor. Copper doses were incrementally increased from 15 to 1300 ppb. At the lower copper doses there was little difference in NH3 utilization measured by percent disappearance of influent to effluent concentrations in the effluent. At higher copper doses (i.e., 600e1300 ppb) intermittent small decreases in NH3 utilization were observed. Overall, there was no statistically significant difference in the effluent ammonia concentrations of the copper reactor
Fig. 2 e NH3eN utilization (%) in bulk water for chlorite dosed and control PVC reactors.
5516
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
The copper reactor was relatively unaffected by lower chlorite doses, but at 20 ppm chlorite, NH3eN utilization dropped to zero (Fig. 3). After chlorite was discontinued it took almost two months to re-establish complete nitrification. The bulk water NO2eN and NO3eN as percentages of the added NH3eN were calculated. For the both reactors, chlorite at the low range (0.2e2.0 ppm) did not significantly affect the NO2eN percentage. At 20 ppm chlorite, NO2eN percentage increased noticeably to about 1e3% of the added ammonia in both systems, but the magnitude of this change was only in the ppb range. In the case of NO3eN, the percent of ammonia converted decreased from near 100%e60% in the PVC reactor the day that the chlorite was increased to 20 ppm and rebounded to the original level of near 100% within 10 days even when this level of chlorite was maintained. In the case of the copper reactor, the percent conversion dropped to less than 5% after a week of exposure to 20 ppm chlorite and did not rebound for almost two months. Effluent total and dissolved copper concentrations in chlorite added and control copper were also measured. The 95% confidence interval for total and dissolved copper for the chlorite dosed reactor (0.8 0.04 and 0.69 0.03 ppm) was higher than that for control reactor (0.71 0.02 and 0.63 0.02 ppm). A paired t-test for both effluent total and dissolved copper was averaged for chlorite concentrations from 0.1 to 2 ppm and for 20 ppm. In all cases, there was significantly more total and dissolved copper in reactors that had received chlorite. The MPN values for AOB and NOB were comparable to those found in the previous copper experiment. The only difference was that at 20 ppm chlorite, no NOB were found in the effluent of both the PVC and copper reactors. HPC values were unaffected and similar to those in the previous copper experiment.
3.3.
Effect of chloramine on nitrification
The effect of chloramine on nitrification was investigated by gradually increasing the amount of free chlorine fed to the influent of nitrifying PVC and copper reactors. Initially chlorine was added at 0.5:1 chlorine to ammonia ratio and gradually raised to a 5:1 ratio, with a total/combined chloramine
Fig. 3 e NH3eN utilization (%) in bulk water for chlorite dosed and control copper reactors.
dose of 3.55 mg/L. NH3 utilization in the chloramine dosed PVC, copper and control reactors are shown in Fig. 4. Bulk NH3eN utilization decreased significantly only at the 5:1 chlorine to ammonia ratio, with occasional decreases at the lower ratios. Note that the 5:1 ratio was maintained for two months and there was a long period of acclimation where the ammonia utilization gradually declined. Even at the 5:1 ratio after two months of exposure, the copper reactor continued to utilize around 20% of the ammonia. When chloramine was discontinued, the copper reactor regained full nitrification after three weeks. In contrast, the PVC reactor ceased ammonia utilization at the higher ratio and it required approximately six weeks to recover its nitrifying ability. The percent conversion of ammonia in the influent to NO2eN and NO3eN in the reactors after 8 h of stagnation was calculated for all chlorine to ammonia ratios. Only in the case of the 5:1 ratio in the PVC reactor did the nitrite level initially increase from the detection limit and peaked at two weeks to approximately 1.4% of the added nitrogen. This level then again decreased to no detectable nitrite. There was no change in nitrite in the copper reactors as a result of increased chlorine to ammonia ratios; it remained at the limit of detection suggesting complete conversion to nitrate. For the copper reactor, the percent of ammonia converted to nitrate dropped one week after attaining the 5:1 ratio to an average of 18% conversion and it took two weeks to reach an average of 12% in the PVC reactor. When chlorine addition was terminated, there was a rebound of total conversion of ammonia to nitrate approximately three weeks later for the PVC reactor. This time was considerably longer (seven weeks) for the copper reactor, and corresponds to the complete loss of ammonia (Fig. 4). Total and dissolved copper concentrations in the effluent of the monochloramine-dosed and control copper reactors were measured at each chlorine to ammonia ratio. The 95% confidence intervals for total and dissolved copper for the monochloramine-dosed reactor (0.83 0.02 and 0.73 0.02) are higher than that for control reactor (0.75 0.02 and 0.63 0.01). A paired t-test showed that the p values for total and dissolved copper were significantly different. It appears that the presence of monochloramine significantly increased the copper concentration in the water.
Fig. 4 e NH3eN utilization (%) in bulk water for chloramine dosed (PVC, copper) and control (PVC, copper) reactors.
5517
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
The average MPN value for AOB and NOB in the reactor water are shown in Table 1. In general, there were fewer AOB and NOB in reactors that received chlorine. No NOB were detected in the presence of chloramine at the 5:1 ratio in either reactor which is consistent with the increased level of nitrite that was measured. HPC effluent counts were unaffected and comparable to those of the control reactors that did not receive chlorine (data not shown).
3.4. Effect of copper, chlorite and chloramine on biofilm cell numbers AOB/NOB and HPC numbers in the biofilm were determined at the end of each experiment and before the recovery phase (Table 2). AOB/NOB abundance is typically about 3 logs lower than the HPC with the exception of the highest chlorine to ammonia (5:1) ratio in the copper reactor, where the AOB were present at numbers greater than that of the heterotrophs. No NOB were detected in the copper reactor biofilms at the termination of the chlorite and chloramine experiments. A lack of detection of NOB in the biofilm and the bulk water is consistent with the increased levels of nitrite measured in these reactors.
4.
Discussion
In this project three different strategies for controlling nitrification in household plumbing were studied. The inhibitory effect of copper, chlorite, and a chloramine residual on nitrification were tested. The impact of each control measure on the release of total and ionic copper from copper coupons into solution was also evaluated. Throughout the experiments in the control reactors, there was total loss of ammonia, no detectable nitrite, and near complete conversion of ammonia to nitrate. This suggests that nitrite was likely formed but converted to nitrate within the 8 h stagnation period of the experimental system and that complete nitrification was occurring. Complete nitrification is contrary to what has often
Table 1 e Average (n [ 2) MPN per ml for ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB) at different Cl2 to NH3eN ratios. ND [ none detected. Chlorine to NH3eN ratio
NH2Cl added PVC reactor
Control PVC reactor
NH2Cl added copper reactor
Control copper reactor
0.5:1 1.0:1.0 1.5:1 2:01 5:01
AOB
31 27.6 9.7 7.2 12.7
37.7 31.5 52.2 9.5 31.4
10.7 10.4 6.9 8.7 13.2
38.1 14.5 27.6 17 21.2
0.5:1 1.0:1.0 1.5:1 2:01 5:01
NOB
48.8 9.7 21 8.4 ND
59.1 23.3 351.9 793.6 310.1
106.7 23.5 28.5 10.2 ND
30.3 51.5 170.9 61.5 48.3
Table 2 e Cell numbers for autotrophic and heterotrophic populations in the biofilm at the end of each experiment. AOB e ammonia oxidizing bacteria, NOB e nitrite oxidizing bacteria, HPC e heterotrophic plate counts. ND [ none detected. AOB (MPN/cm2) Copper added PVC pH Control PVC Chlorite added PVC NH2Cl added PVC PVC-control Chlorite added copper NH2Cl added copper Copper-control
5.7 2.8 3.5 9.2 2.4 1.3 1.4 1.1
103 103 102 101 103 101 102 103
NOB (MPN/cm2) 4.4 3.2 7.7 7.9 2.8 ND ND 8.0
103 103 101 102 103
102
HPC (CFU/cm2) 1.6 7.7 1.5 1.2 1.2 1.8 9.8 1.8
106 105 106 105 106 105 101 105
been observed in nitrifying water distribution systems where there is incomplete nitrification, nitrite accumulation, and high nitrite-N concentrations (Wolfe et al., 1988).
4.1.
Added copper
Copper introduced to the nitrifying PVC reactor did not have a significant effect on ammonia utilization, nor on nitrate concentration in the reactor. The numbers of HPC, AOB and NOB were also unaffected by copper. This observation contradicts the results from other researchers (Skinner and Walker, 1961; Loveless and Painter, 1968; Martin and Richard, 1982; Zhang and Edwards, 2005, 2010), who reported that low (5 ppbe0.56 ppm) copper concentrations inhibited nitrification. It is possible that inhibition may have occurred if concentrations had been increased to higher levels as those used by Tomlinson et al. (1966) and Meiklejohn (1950). The discrepancy in copper sensitivity between the results of this work and that of others may be because the previous work was done with pure cultures or activated sludge, and the copper tolerance for those biological systems may be much different from the biofilm grown in the reactors used in this project. Another possible explanation is that organic matter contains functional groups (Sarathy and Allen, 2005) that bind metals to form less bioavailable complexes (Loveless and Painter, 1968; Dodge and Theis, 1979; Crecelius et al., 1982) that are therefore less inhibitory (Kim et al., 2006). The humics used in this research may have acted in this capacity although the copper in the reactor water was detected in the ionic form. Added copper also had no impact on the HPC numbers in the reactors. This is in contrast to the findings of others (Thurman and Gerba, 1989; Artz and Killham, 2002; Kim et al., 2002; Teitzel and Parsek, 2003; Lehtola et al., 2004) who reported that copper has a toxic effect on heterotrophs in drinking water. Again this discrepancy may be due to the difference in experimental conditions and bacterial populations. Another potential explanation is that the reactors used in this study contained intact, natural biofilms that had been present for over a year while other studies used suspended organisms and/or fresh copper surfaces. Copper toxicity depends on the concentration of cupric ion (Cuþ2) (Braam and Klapwijk, 1981), and this ionic form
5518
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
becomes more abundant at a lower pH (Edwards et al., 1996). Even when the pH of the reactor was lowered to 6.6 to ensure that the added copper was in the cupric ion form, there was no effect on nitrification or the numbers of AOB, NOB or HPC in the reactor effluent or in the biofilm.
4.2.
Chlorite
Chlorite was chosen as another potential control mechanism because previous studies by McGuire et al. (1999) showed that chlorite ion (0.2e1.0 ppm) in distribution systems can inhibit nitrification. Results in this project contradict their reports. The PVC reactor was initially impacted only at the unrealistic dose of 20 ppm chlorite but then regained the ability to nitrify in the presence of this concentration. At this high concentration there was an impact on nitrification in the copper system that persisted after the cessation of chlorite addition. McGuire et al. (1999) mentions that chlorite did not inhibit nitrification in one system, which according to the author may be due to the presence of higher ammonia (1.4 mg/L). Karim and LeChevallier (2006) also reported that chlorite was unable to hinder nitrification. All these studies were done with low doses of chlorite (0.2e1.0 ppm) and their flow pattern, water quality and bacterial population/biofilm characteristics may be significantly different from this project. Another possible explanation is that chlorite inhibits the activity of Nitrosomonas europaea and Nitrobacter winogradski (Hynes and Knowles, 1983), but may be inactive toward other groups of AOB and NOB that may be predominant in the reactors. The planktonic and biofilm heterotrophic populations were unaffected by chlorite exposure. The trend in heterotrophic numbers supports previous work by Gagnon et al. (2005) where chlorite at 0.1e0.25 ppm was ineffective in inactivating heterotrophic bacteria. Similar to HPC, AOB values did not show any effect due to chlorite exposure. The planktonic NOB population remained unchanged for all concentrations of chlorite except at 20 ppm, when they could not be detected. NOB were also not detected in the copper reactor’s biofilm at the end of exposure period to the high chlorite concentration. Hynes and Knowles (1983) reported that pure cultures of AOB (N. europaea) and NOB (N. winogradski) were inhibited by chlorite, and that the AOB are 50 times more sensitive to chlorite inhibition than NOB. In our experiments it appeared that the NOB were more sensitive than the AOB. Again this may be due to the differences in experimental setup, pure cultures vs environmental biofilm, water quality and other factors. It is also possible that the methods used in this work did not enumerate all of the potential nitrifying organisms in the reactors. A potential impact of chlorite on copper plumbing could be an increased release of copper into the water. As chlorite (ClO2 2 ) could be transformed to chlorine dioxide (ClO2) in an acidic environment (Gates, 1989) that could created by oxidation of ammonia during nitrification by biofilm on the surfaces, copper corrosion could increase due to the oxidative nature of chlorine dioxide. Therefore, there could be an unintended result of elevated copper release when chlorite is applied, and this was seen in these experiments. For this reason, and because chlorite is a regulated compound, the
utilities should carefully evaluate chlorite before implementing it as a nitrification control strategy.
4.3.
Chloramine
According to a survey by Seidel et al. (2005), optimizing the chlorine to ammonia ratio is the most common nitrification control technique. The chlorine to ammonia-N weight ratio used to form monochloramine typically varies from 3:1 to 5:1 (Wilczak, 2006). Several studies in full scale chloraminated systems have determined that a minimum 2e3 mg Cl2/L chloramine residual should be maintained to prevent nitrification (Wolfe et al., 1990; Lieu et al., 1993; Kirmeyer et al., 1995; Odell et al., 1996; Harrington et al., 2002). A combination of chloramines does and optimizing the chlorine to ammonia ratio has been shown to be the easiest and most cost-effective way to control/prevent nitrification (Lieu et al., 1993). The results of this work can be compared to several studies where the chlorine to ammonia ratio has been measured in pilot and full scale systems and related to nitrification. McGuire et al. (2004) showed that nitrification occurred in a pilot system where chloramine was applied at a 3:1 ratio. Similarly, for two covered finished water reservoirs in southern California which were initially chloraminated at a 3:1 ratio, nitrification was significantly reduced after raising the ratio to 5:1 (Wolfe et al., 1988). A Florida utility rarely experienced nitrification when the combined chlorine residual was above 1 mg/L at a chlorine to ammonia ratio of 5:1 (Liu et al., 2005). A study done by Karim and LeChevallier (2006) showed that monochloramine applied at a 3:1 ratio was not able to control nitrification, but it was effective at a 5:1 ratio. Based on the above information, the impact of an incremental increase in the chlorine to ammonia ratio on nitrification was tested. The intent was to test situations in (1) a system where chloramine residual was regained and/or (2) households downstream of booster chlorination that was used to recreate chloramines from free ammonia (Wolfe and Lieu, 2001; Wilczak, 2006). At the starting ratio of 0.5:1, about 0.35 mg/L of total chlorine was present. This was chosen because Holt et al. (1995) showed reduction of nitrification at total chlorine concentrations of more than 0.3 mg/L. However, at an 8 h stagnation time, this low ratio did not have an impact on nitrification and organisms responsible for nitrification could be found in the bulk water and biofilm. Very minor transient impacts were seen with increasing ratios of 1:1, 1.5:1, and 2:1. There are several possible explanations for these results. Nitrifying bacteria form protective layers (slime layer or capsules) mainly composed of polysaccharides (Prosser, 1986) as a defensive mechanism against unfavorable environmental conditions such as low pH. These capsules protect organisms so that they are more resistant to disinfectants (Stewart and Olson, 1996). Cunliffe (1991) detected nitrifiers in 64% of the samples collected in South Australia and of them 20.7% contained more than 5.0 mg Cl2/L of monochloramine. The author hypothesized that as the nitrifiers grow in aggregates or in biofilm attached to the surface, they remain unaffected by disinfectant and the nitrifiers detected in samples containing high chloramine residual may have been detached shortly before or during the sampling
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
period. Higher AOB have been detected in biofilm than water in the Metropolitan Water District of Southern California distribution system (Stewart and Lieu, 1997). Regan (2001) postulated on the protective mechanism of biofilms. Monochloramine is mass transport limited in biofilm, so bacteria inside the biofilm are not exposed to the disinfectant. Another protective mechanism may be the relative ratio of growth vs disinfection. Harrington et al. (2003) stated that if AOB growth rate driven by ammonia concentration exceeds the AOB inactivation rate by monochloramine, then theoretically AOB can grow in the presence of monochloramine. Fleming et al. (2005) proposed nitrification potential curves based on the relative concentration of chlorine (biocide) and free ammonia (food). According to them the threshold chlorine value is 1.6, above which nitrification would be prevented, without any influence from free ammonia concentration. At chlorine concentrations below 1.6, the nitrification potential depends on the ratio of chlorine and free ammonia. Only when the 5:1 Cl2 to NH3eN ratio was reached was there a consistent reduction of nitrification during the 8 h of stagnation, and the affect of this ratio agrees with results from Karim and LeChevallier (2006) and Lieu et al. (1993). Likewise, in work by Harrington et al. (2002), nitrification did not occur when the total chlorine concentrations were more than 2.2 mg/L and the biocide to food ratio was 1.9 mg Cl2/mg of N or more. The most pronounced effect was on the PVC reactor, which is in contrast to the chlorite effect which was most obvious on the copper reactor. Interestingly, the HPC numbers in the biofilm on the copper surfaces were lower than those of the AOB, suggesting that the disinfectant had greater activity on the HPC than the nitrifiers. No planktonic NOB were detected in monochloramine exposed reactors at the 5:1 ratio and they were also not detected in the biofilm of the copper reactor. A possible explanation is that the NOB were more vulnerable to disinfection than the AOB (Wolfe and Lieu, 2001). These results also support observations that once nitrification starts in full scale systems, higher levels of chloramine may not be an effective control method. Skadsen (1993) reported that a chloramine dose of 8 mg Cl2/L was not effective in controlling nitrification in the Ann Arbor, Michigan distribution system. This may be due to the fact that nitrite can degrade chloramine residuals before it can inactivate the nitrifying bacteria (Wolfe et al., 1988; Kirmeyer et al., 1995; Odell et al., 1996). Also, maintaining monochloramine at a high ratio close to 5:1 in full scale systems is not always easy, and is sometimes associated with dicholoramine formation, and taste/odor problems and higher DBP formation (Skadsen and Cohen, 2006). This project also provides some insight on effect of chloramine on copper corrosion. It has been found that chloramine increased copper corrosion (Ingleson et al., 1949). Enhanced copper solubility during periods of chloramination with excess ammonia present was observed for Champaign IL tap water (AwwaRF, 1990). Ammonia has a strong complexation constant for cupric ion (Schock et al., 1995). On the other hand, according to MacQuarrie et al. (1997) and Rahman et al. (2007), application of monochloramine results in a decrease in copper pipe corrosion. Current results found higher total and dissolved copper concentrations in the reactors where the disinfectant was added.
4.4.
5519
Copper vs PVC surfaces
Copper and PVC were chosen because of their use in plumbing systems and also because of the potential antimicrobial effect of copper vs PVC. One could also hypothesize that copper materials may be more prone to nitrification in chloraminated systems because monochloramine can rapidly decay through reactions with a copper plumbing system (Edwards and Nguyen, 2005), therefore providing more free ammonia for nitrification. Therefore, pipe material could significantly influence the nitrification process. In the conditioned control reactors with no added chlorite or chlorine, there was no difference in ammonia utilization between the two surfaces, and the HPC, AOB and NOB numbers were similar. The lack of efficacy of copper against these organisms was supported by the experiments where copper was added to the PVC reactor and no decrease in nitrification was seen. In the presence of chorite or combined chlorine, there were some differences between the two systems. In the case of chlorite, nitrification in the copper system was affected to a greater extent with a loss of nitrification at the highest concentration (20 ppm) and an extended recovery time after chlorite addition ceased. This could be because of the production of chlorine dioxide (Gates, 1989) on the metal surface. However, in the presence of combined chlorine, the PVC system was more greatly impacted; nitrification ceased and it required nearly six weeks for recovery. In the copper reactor, there was never a complete loss of nitrification, and complete nitrification was again attained only three weeks after chlorine addition stopped. This suggests that the mechanism of chloramine decay on the copper surface, as reported by Edwards and Nguyen (2005), could be at play.
4.5.
Relationship between HPC and nitrification
Some research has shown that there is a correlation between HPC values and nitrification in water systems with HPC increasing when nitrification occurs. Wolfe et al. (1990) reported that HPC and AOB population were highly correlated in distribution system water, and that this may be explained due to the dependence of HPC on AOB for carbon fixation. However, other researchers reported that the relationship is very site specific (Donnelly and Giani, 2005), likely because HPC can grow in response to other available organics and their numbers are also affected by residual disinfectant. Throughout these experiments there was no consistent relationship between HPC and AOB/NOB MPN values. This may be due to the presence of humic substances in the reactors.
5.
Conclusions
Copper surfaces or copper added to reactors at concentrations up to 1.3 ppm within a pH range of 6.6e8.15 did not inhibit nitrification. Chlorite was effective at inhibiting nitrification only at an unrealistic dose (20 ppm), on copper surfaces. There was limited, transient control on reactors with PVC surfaces at 20 ppm chlorite.
5520
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
Chloramine at a Cl2 to NH3eN ratio of 5:1 managed nitrification in the copper reactor and was able to control it in the PVC system. The addition of chlorite and chloramines may increase copper corrosion. No correlation between HPC and AOB/NOB was found. Time to recovery after cessation of addition of chlorite and chloramine varied but did occur, suggesting that nitrification in premises plumbing is a robust process.
Acknowledgments This material is based upon work supported by the National Science Foundation under Grant No. 0329474. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. The authors thank the MUSES team funded by this grant and led by Dr. Andrea Dietrich at Virginia Tech for their insights, with special appreciation to Dr. Marc Edwards for his assistance. The authors also acknowledge John Neuman and Daniel Swanson for their help in the laboratory.
references
Alexander, M., Clark, F.E., 1965. Nitrifying bacteria. In: Black, C.A. (Ed.), Methods of Soil Analysis. American Society of Agronomy, Madison, WI, pp. 1477e1483. Part 2. Alt, E.R., Sternlieb, I., Goldfischer, S., 1990. The cyto-pathology of metal overload. International Review of Experimental Pathology 31, 165e188. Artz, R.R.E., Killham, K., 2002. Survival of Escherichia coli O157:H7 in private drinking water wells: influences of protozoan grazing and elevated copper concentrations. FEMS Microbiology Letters 216 (1), 117e122. American Water Works Association Research Foundation, 1990. Lead Control Strategies. AWWA Research Foundation and AWWA, Denver, CO. Braam, F., Klapwijk, A., 1981. Effect of copper on nitrification in activated sludge. Water Research 15 (9), 1093e1098. Clescerl, L.S., Greenberg, A.E., Eaton, A.E. (Eds.), 1995. Standard Methods for Examination of Water & Wastewater, nineteenth ed. American Public Health Association. Crecelius, W.A., Hardy, J.T., Gibson, C.I., Schmid, T.R.L., Apts, C.W., Curtisen, J.M., Joyce, S.P., 1982. Copper bioavailability to marine bivalves and shrimps: relationship to cupric ion activity. Marine Environmental Research 6 (1), 13e26. Cunliffe, D.A., 1991. Bacterial nitrification in chloraminated water supplies. Applied and Environmental Microbiology 57 (11), 3399e3402. Dodge, E.E., Theis, T.L., 1979. Effect of chemical speciation on the uptake of copper by Chironomus tentans. Environmental Science Technology 13 (10), 1287e1288. Donnelly, M., Giani, R., 2005. Developing nitrification control limits from distribution system data. In: Proceedings of American Water Works Association Water Quality Conference, Denver CO. Edwards, M., Nguyen, C., 2005. Chemistry of rapid chloramine decay in water contacting copper plumbing. In: Proceedings of
American Water Works Association Water Quality Technology Conference, Quebec City, Canada. Edwards, M., Schock, M.R., Meyer, T.E., 1996. Alkalinity, pH and copper corrosion by-product release. Journal of American Water Works Association 88 (3), 81e88. Fleming, K.K., Harrington, G.W., Noguera, D.R., 2005. Nitrification potential curves: a new strategy for nitrification prevention. Journal of American Water Works Association 97 (8), 90e99. Gagnon, G.A., Rand, J.L., O’Leary, K.C., Rygel, A.C., Chauret, C., Andrews, R.C., 2005. Disinfectant efficacy of chlorite and chlorine dioxide in drinking water biofilm. Water Research 39 (9), 1809e1817. Gates, D., 1989. Chlorine dioxide generation technology and mythology. In: Proceedings of AWWA Water Quality Conference, Denver CO. Geesey, G.G., Jang, L., 1989. Interactions between metal ions and capsular polymers. In: Beveridge, T.J., Doyle, R.J. (Eds.), Metal Ions and Bacteria. Wiley & Sons, New York, pp. 325e357. Goeres, M.D., Loetterle, L.R., Hamilton, M.A., Murga, R., Kirby, D. W., Donlan, R.M., 2005. Statistical assessment of a laboratory method for growing biofilms. Microbiology 151 (3), 757e762. Harrington, G.W., Noguera, D.R., Kandou, A.I., Vanhoven, D.J., 2002. Pilot-scale evolution of nitrification control strategies. Journal of American Water Works Association 94 (11), 78e89. Harrington, G.W., Noguera, D.R., Bone, C.C., Kandou, A.I., Oldenburg, P.S., Regan, J.M., Hoven, D.V., 2003. Ammonia from Chloramine Decay: Effect on Distribution System Nitrification. AWWA Research Foundation, Denver, CO. Holt, D., Todd. R.D., Delanoue, A., Colborne, J.S., 1995. A study of nitrite formation and control in chloraminated distribution systems. In: Proceedings of American Water Works Association Water Quality Technology Conference, Denver, CO. Hu, Z., Chandran, K., Grasso, D., Smets, B.F., 2003. Impact of metal sorption and internalization on nitrification inhibition. Environmental Science Technology 37 (4), 728e734. Hynes, R.K., Knowles, R., 1983. Inhibition of chemoautotrophic nitrification by sodium chlorate and sodium chlorite: a reexamination. Applied and Environmental Microbiology 45 (4), 1178e1182. Ingleson, H., Sage, A.M., Wilkinson, R., 1949. Effect of chlorination of drinking water on brass fittings. Journal of Industrial Water Engineering 3 (1), 81e91. Karim, M.R., LeChevallier, M.W., 2006. Alternative treatment for nitrification control in chloraminating utilities. In: Proceedings of the AWWA Annual Conference, San Antonio TX. Kim, B.R., Andersson, J.E., Mueller, S.A., Gaines, W.A., Kendall, A. M., 2002. Literature reviewdefficacy of various disinfectants against Legionella in water systems. Water Research 36 (18), 4433e4444. Kim, K.T., Kim, I.S., Hwang, S.H., Kim, S.D., 2006. Estimating the combined effect of copper and phenol to nitrifying bacteria in wastewater treatment plants. Water Research 40 (3), 561e568. Kindaichi, T., Kawano, Y., Ito, T., Satoh, H., Okabe, S., 2006. Population dynamics and in situ kinetics of nitrifying bacteria in autotrophic nitrifying biofilms as determined by real-time quantitative PCR. Biotechnology and Bioengineering 94 (6), 1111e1121. Kirmeyer, G.J., Odell, L.H., Jacangelo, J.G., Wilczak, A., Wolfe, R.L., 1995. Nitrification Occurrence and Control in Chloraminated Water Systems. AWWARF and AWWA, Denver CO. LeChevallier, M.W., Cawthon, C.D., Lee, R.G., 1988. Inactivation of biofilm bacteria. Applied and Environvironmental Microbiology 54 (10), 2492e2499. Lee, Y.W., Ong, S.K., Sato, C., 1997. Effect of heavy metals on nitrifying bacteria. Water Science and Technology 36 (12), 69e74. Lehtola, M.J., Miettinen, I.T., Kienanen, M.M., Kekki, T.K., Laine, O., Hirvonen, A., Vartiainen, T., Martikainen, P.J., 2004. Microbiology, chemistry and biofilm development in a pilot
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
drinking water distribution system with copper and plastic pipes. Water Research 38 (17), 3769e3779. Lieu, N.I., Wolfe, R.L., Means, E.G., 1993. Optimizing chloramine disinfection for control of nitrification. Journal of American Water Works Association 85 (2), 84e90. Liu, S., Taylor, J.S., Randall, A.A., Dietz, J.D., 2005. Nitrification modeling in chloraminated distribution systems. Journal of American Water Works Association 97, 98e108. Loveless, J.E., Painter, H.A., 1968. The influence of metal ion concentration and pH value on the growth of a Nitrosomonas strain isolated from activated sludge. Journal of General Microbiology 52 (1), 1e14. MacQuarrie, D.M., Mavinic, D.S., Neden, D.G., 1997. Greater vancouver water district drinking water corrosion inhibitor testing. Canadian Journal of Civil Engineering 24 (1), 34e52. Martin, G., Richard, Y., 1982. Elements of inhibition of nitrifying bacteria. Water Science and Technology 14 (6e7), 781e794. McGuire, M., Lieu, J.N., Pearthree, M., 1999. Using chlorite ion to control nitrification. Journal of American Water Works Association 91 (10), 52e62. McGuire, M.J., Pearthree, M., Blute, N.K., Arnod, K.F., 2004. Tucson nitrification control pilot study: a significant advance in the state of the art of implementing chloramines in a desert environment. Arizona Water and Pollution Control Association Annual Conference. McGuire, M.J., Pearthree, M.S., Blute, N.K., Arnold, K.F., Hoogerwerf, T., 2006. Nitrification control by chlorite ion at pilot scale. Journal of American Water Works Association 98 (1), 95e105. McGuire, M.J., Wu, X., Blute, N.K., Askenaizer, D., Qin, G., 2009. Prevention of nitrification using chlorite ion: results of a demonstration project in Glendale, Calif. Journal of American Water Works Association 101 (10), 47e59. Meiklejohn, J., 1950. The isolation of Nitrosomonas europaea in pure culture. Journal of General Microbiology 4 (2), 185e191. National Research Council, 2006. Water Science and Technology Board, Science and Technology Board. Drinking Water Distribution Systems: Assessing and Reducing Risks. The National Academies Press, Washington D.C. National Sanitation Foundation, 2008. http://www.nsf.org/ consumer/plumbing/index.asp (accessed 18.06.08.). Odell, L.H., Kirkrmeyer, G.J., Wilczak, A., Jacangelo, J.G., Marcinko, J.P., Wolfe, R.L., 1996. Controlling nitrification in chloraminated systems. Journal of American Water Works Association 88 (7), 86e98. Okabe, S., Kinadaichi, T., Itoh, T., 2005. Fate of 14C-labeled microbial products derived from nitrifying bacteria in autotrophic nitrifying biofilms. Applied and Environmental Microbiology 71 (7), 3987e3994. Oskarsson, A., Norrgren, L., 1998. Copper pipes as a source of copper exposure in man and environment. Environmental Reviews 6, 139e150. Prosser, J.I., 1986. Experimental and theoretical models of nitrification. In: Prosser, J.I. (Ed.), Nitrification. IRL Press, Oxford, England, pp. 74e75. Rahman, S., McDonald, B.C., Gagnon, G.A., 2007. Impact of secondary disinfectants on copper corrosion under stagnation conditions. Journal of Environmental Engineering 133 (2), 180e185. Regan, J. M., 2001. Microbial Ecology of Nitrification in Chloraminated Drinking Water Distribution Systems, Ph.D Dissertation, University of Wisconsin, Madison, Wisconsin. Rittmann, B.E., Regan, J.M., Stahl, D.A., 1994. Nitrification as a source of soluble organic substrate in biological treatment. Water Science and Technology 30 (6), 1e8. Rowe, R., Todd, R., Waide, J., 1977. Microtechnique for most probable number analysis. Applied and Environmental Microbiology 33 (3), 675e680.
5521
Sarathy, V., Allen, H.E., 2005. Copper complexation by dissolved organic matter from surface water and wastewater effluent. Ecotoxicology and Environmental Safety 62, 337e344. Sato, C., Leung, S.W., Schnoor, J.L., 1988. Toxic response of Nitrosomonas europaea to copper in inorganic media and wastewater. Water Research 22 (9), 1117e1127. Schock, M., Lytle, D., Clement, J., 1995. Effect of pH, DIC, orthophosphate and sulfate on drinking water cuprosolvency. US EPA Report/600/R-95/085 June 1995. Schramm, A., Larsen, L.H., Revsbech, N.P., Ramsing, N.B., Amman, R., Schleifer, K.H., 1996. Structure and function of a nitrifying biofilm as determined by in situ hybridization and the use of microelectrodes. Applied and Environmental Microbiology 62 (12), 4641e4647. Seidel, C.J., McGuire, M.J., Summers, S., Via, S., 2005. Have utilities switched to chloramines? Journal of American Water Works Association 97 (10), 87e97. Skadsen, J., 1993. Nitrification in a distribution system. Journal of American Water Works Association 85 (7), 95e103. Skadsen, J., Cohen, Y.K., 2006. Fundamentals and Control of Nitrification in Chloraminated Drinking Water Distribution System. AWWA Manual M56. American Water Works Association, Denver, CO. Skinner, F.A., Walker, N., 1961. Growth of Nitrosomonas europaea in batch and continuous culture. Archives of Microbiology 38, 339e349. Stewart, M.H., Olson, B.H., 1996. Bacterial resistance to potable water disinfectants. In: Hurst, C.J. (Ed.), Modeling Disease Transmission and its Prevention by Disinfection. Cambridge University Press, Cambridge, England, pp. 140e192. Stewart, M.H., Lieu, N.I., 1997. Nitrification in chloraminated drinking water and its association with biofilms. In: Proceedings of the AWWA Water Quality Technology Conference, Denver, CO. Teitzel, G.M., Parsek, M.K., 2003. Heavy metal resistance of biofilm and planktonic Pseudomonas aeruginosa. Applied and Environmental Microbiology 69 (4), 2313e2320. Thurman, R.B., Gerba, C.P., 1989. The molecular mechanisms of copper and silver ion disinfection of bacteria and viruses. CRC Critical ReviewdEnvironmental Control 18 (4), 295e315. Tomlinson, T.G., Boon, A.G., Trotman, C.N.A., 1966. Inhibition of nitrification in the activated sludge process of sewage disposal. Journal of Applied Bacteriology 29 (2), 266e291. USEPA, 1993. Method for the Determination of Inorganic Substances in Environmental Samples. EPA/600/R-93/100 Washington D.C. USEPA, 2000. Microbial/Disinfection Byproducts (M-DBP) Federal Advisory Committee: Stage 2 M-DBP Agreement in Principle. EPA, Washington, D.C. Wahman, D.G., Wulfeck-Kleier, K.A., Pressman, J.G., 2009. Monochloramine disinfection kinetics of Nitrosomonas europaea by propidium monoazide quantitative PCR and live/ dead BacLight methods. Applied and Environmental Microbiology 75 (17), 5555e5562. Wilczak, A., 2006. Overview of causes and control of nitrification in water distribution systems. Chapter 4. In: Kosyra, M.K. (Ed.), Fundamentals and Control of Nitrification in Chloraminated Drinking Water Distribution Systems, AWWA Manual M56. American Water Works Association, Denver, CO. Wilczak, A., Jacangelo, J.G., Marcinko, J.P., Odell, L.H., Kirmeyer, G. J., Wolfe, R.L., 1996. Occurrence of nitrification in chloraminated distribution systems. Journal of American Water Works Association 88 (7), 74e85. Wolfe, R.L., Lieu, N.I., 2001. Nitrifying bacteria in drinking water. In: Bitton, G. (Ed.), Encyclopedia of Environmental Microbiology. John Wiley and Sons, New York, NY.
5522
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 1 1 e5 5 2 2
Wolfe, R.L., Means III, E.G., Davis, M.K., Barrett, S.E., 1988. Biological nitrification in covered reservoirs containing chloraminated water. Journal of American Water Works Association 80 (9), 109e114. Wolfe, R.L., Lieu, N.I., Izaguirre, G., Means, E.G., 1990. Ammoniaoxidizing bacteria in chloraminated distribution systems: seasonal occurrence, distribution and disinfection resistance. Applied and Environmental Microbiology 56 (2), 451e462.
Zhang, Y., Edwards, M., 2005. Water chemistry and growth of nitrifying bacteria in drinking water distribution systems. In: Proceedings of the American Water Works Association Water Quality Technology Conference, Denver, CO. Zhang, Y., Griffin, A., Rahman, M., Camper, A., Baribeau, H., Edwards, M., 2009. Lead contamination of potable water due to nitrification. Environmental Science and Technology 43 (6), 1890e1895.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 3 e5 5 2 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Disinfection of domestic effluents by gamma radiation: Effects on the inactivation of Ascaris lumbricoides eggs Gloria S.M.B. de Souza a, Ludmila A. Rodrigues a, Warllem J. de Oliveira a, Carlos A.L. Chernicharo b,*, Marcos P. Guimara˜es a, Cristiano L. Massara c, Pablo A. Grossi d a
Department of Parasitology, Universidade Federal de Minas Gerais, Av. Antoˆnio Carlos, 6627, 31270-901 Belo Horizonte, MG, Brazil Department of Sanitary and Environmental Engineering, Universidade Federal de Minas Gerais, Av. Antoˆnio Carlos, 6627, 31270-901 Belo Horizonte, MG, Brazil c Rene´ Rachou Research Center, Av. Augusto de Lima, 1715, 30190-002 Belo Horionte, MG Brazil d Center for Development of Nuclear Technology, CDTN, Av. Antoˆnio Carlos, 6627, 31270-901 Belo Horizonte, MG, Brazil b
article info
abstract
Article history:
This work investigated the inactivation of Ascaris lumbricoides eggs in domestic effluents by 60
Co source. Domestic wastewater was treated in a compact
Received 2 February 2011
gamma radiation from a
Received in revised form
demo-scale system consisting of a UASB reactor and a trickling filter; treatment was
30 July 2011
carried out at the Center for Research and Training on Sanitation (CePTS), Federal
Accepted 7 August 2011
University of Minas Gerais, in Belo Horizonte-MG, Brazil. One-liter of treated wastewater
Available online 23 August 2011
samples was artificially contaminated with an average of 1000 non-embryonated Ascaris lumbricoides eggs from human feces; samples were then irradiated in a multiple-purpose
Keywords:
irradiator at different doses (0.5e5 kGy). Eggs were recovered from the wastewater and
Disinfection
the viability of these irradiated eggs was evaluated; the description of the egg develop-
Gamma radiation
mental phases with each dose of gamma radiation was recorded. Radiation doses of
Wastewater
3.5 kGy effectively disinfected effluents with lower concentrations of A. lumbricoides eggs;
Inactivation
higher radiation doses of 5 kGy were necessary to disinfect effluents with higher eggs
Eggs Ascaris lumbricoides
concentrations. ª 2011 Elsevier Ltd. All rights reserved.
Viability
1.
Introduction
Although conventional wastewater treatment systems have the capacity to improve effluent quality, they are not sufficient to remove all contaminants (Von Sperling and Mascarenhas, 2005). Disinfectant agents act by inducing biochemical changes in both pathogenic agents and in the effluent. Different wastewater disinfection methods include chlorination, ozonation and ultraviolet radiation. However, none of these methods are effective in removing or inactivating more resistant pathogens such as nematode eggs (Al-Adawi et al., 2006; Galal-Gorchev, 1996; Tahri et al., 2010).
To protect the public, prior to its use, treated and disinfected effluent should have its contamination potential monitored by the examination of indicators such as the presence of coliforms and nematode eggs (WHO, 1989, 2004, 2006; Ayres and Mara, 1996). A parameter that has not yet been taken into consideration by legislation, but which is epidemiologically relevant, is the evaluation of nematode egg viability. The presence of nematode eggs alone does not assess egg viability or infective potential. Eggs from several nematode species, especially from geohelminths, undergo a development step in the environment that requires favorable temperature, humidity and
* Corresponding author. Tel.: þ55 3134091020; fax: þ55 3134091879. E-mail address: [email protected] (C.A.L. Chernicharo). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.008
5524
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 3 e5 5 2 8
oxygenation conditions to become viable and infective. Ascaris lumbricoides eggs are a good indicator of parasitological quality; they are the most numerous species present in wastewater and are more resistant to adverse environmental situations compared to many other enteric organisms. Therefore, nematode egg inactivation can be estimated by viability differences between A. lumbricoides eggs before and after the wastewater treatment process. Although currently expensive, gamma radiation use is justified by its capability of degrading organic matter and pathogenic organisms while retaining nutrients such as nitrogen and phosphorus, which are important in the use of wastewater for agricultural purposes (Bao et al., 2002; Basfar and Abdel Rehim, 2002; Borrely et al., 1998; Hill, 2003; Horak, 1994; Rawat et al., 1998; Wang and Wang, 2007). The literature is controversial regarding the effectiveness of gamma radiation on the inactivation of helminth eggs present in wastewater and sewage sludge. Higher radiation doses were necessary in the studies reported by Chmielewski et al. (1995) and Enigk et al. (1975) cited by Capizzi and Schwartzbrod (2001), who achieved 100% egg inactivation at 6 kGy and 4.8 kGy doses, respectively. Both irradiated the eggs in sludge, but neither provided sample characteristics. Chmielewski et al. (1995) used helminth eggs present in the sludge itself. Also in the study by Melmed and Comninos (1979) high doses were tested. They used Ascaris sp. eggs directly from sludge with an 85% viability percentage and determined that the irradiation of 1-l sample at a radiation dose of 5 kGy inactivated 99.5% of the eggs. Further, a radiation dose of 10 kGy inactivated 98.6e99.6% of the eggs present in the sludge. On the other hand, Horak (1994) showed that the development of Ascaris suum eggs contained in aerobic sludge was inhibited after irradiation at a dose of 1.1 kGy. However, in their study the sludge was inoculated with a concentration of 100e1000 eggs recovered directly from the uterus of an adult worm, and the initial egg viability ranged between 24.6 and 54.6%. Consistent with this report, Shamma and Al-Adawi (2002) demonstrated that a gamma radiation dose of 1.5 kGy inactivated 100% of the A. lumbricoides eggs present in sludge samples. In this study, samples of sludge without A. lumbricoides eggs were inoculated with approximately 100 eggs previously recovered from raw sewage sludge by the NaNO3 flotation method (WHO, 2004). The egg viability percentage in the control was 59.6%.
The conflicting studies differ on four important aspects of experimental design: species of helminth eggs used in the experiments; sources of the helminth eggs; percentages of viable eggs in the control groups; and type of liquid media for egg inoculation. Taking these factors into consideration, this paper investigates the disinfection of domestic effluents by the action of gamma radiation using the viability of A. lumbricoides eggs as an indicator of efficacy.
2.
Materials and methods
To clarify the existing differences found in the literature and to render the experimental conditions more realistic, this paper outlined its tests under the following conditions: Use of eggs of A. lumbricoides, which is the species more prevalent in domestic wastewater and poses more risks to human beings. Use of A. lumbricoides eggs with a high percentage of viable eggs (above 80%). Inoculation of eggs in treated wastewater samples, to better represent possible interactions between the eggs and the medium during irradiation.
2.1.
Area of study
The treated wastewater used in the disinfection experiments was obtained from a demo-scale treatment system comprised of an Upflow Anaerobic Sludge Blanket (UASB) reactor and a Trickling Filter (TF) operating in series, as depicted in Fig. 1. The system was installed at the Center for Research and Training on Sanitation (CePTS), Belo Horizonte, Brazil (coordinates 19 530 4200 S and 43 520 4200 W, altitude 800 m). The UASB reactor was fed on raw sewage taken from the Arrudas wastewater treatment plant, after being submitted to pretreatment for solids and grit removal, therefore representing a typical urban wastewater. The main characteristics and operational conditions of the treatment units are presented in Fig. 1; the main physical-chemical characteristics of the raw sewage and the final effluent from the treatment system are depicted in Table 1.
Fig. 1 e Flow sheet, sampling points and main characteristics of the demo-scale UASB/TF system.
5525
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 3 e5 5 2 8
Table 1 e Physicalechemical characteristics of the raw sewage and the final effluent of the UASB/TF system. Parameters (mg L1)
COD BOD TSS
Raw sewage
Final effluent (treated wastewater)
Median
Mean
Standard deviation
Median
Mean
Standard deviation
405 198 180
392 211 172
100.1 44.6 46.4
75 24 38
76 23 42
21.2 4.5 18.1
COD, chemical oxygen demand; BOD, biochemical oxygen demand; and TSS, total suspended solids.
2.2.
Preparation and collection of samples
The relatively low concentration of A. lumbricoides eggs in the wastewater samples being studied required artificial contamination with eggs in order to obtain results sufficient for statistical analysis. Eggs were obtained from human fecal sediments provided by a clinical analysis laboratory. To obtain concentrations higher than 100 non-embryonated eggs L1 after the recovery procedure, portions of feces containing approximately 1000 eggs were inoculated into the treated wastewater samples. Composite sampling was performed by an automatic sampler programmed to collect 1-l of wastewater at hourly intervals for 24 h. The samples were prepared from the homogenization of the collected wastewater; duplicate 1-l aliquots for every dose of irradiation were withdrawn. Subsequently, the samples were artificially inoculated with A. lumbricoides eggs. The samples were irradiated at a bench scale. Irradiation was performed at the Gamma Irradiation Laboratory of the Nuclear Technology Development Center. This Gamma Irradiation Facility is a Multiple-Purpose Panoramic Irradiator, classified as a Category II and manufactured by MDS Nordion in Canada; model/series number IR-214 and type GB-127, equipped with a dry-storage Cobalt-60 source with maximum radioactive activity of 2200 TBq or 60,000 Ci. The samples were irradiated at a distance of 15 cm from the radioactive source where the radiation field was simultaneously measured by using Fricke dosimeters under electronic equilibrium conditions (ASTM-E-1026-95, 2002). Dose homogenization techniques were taken into account and decay time corrections were performed to assure a reliable prediction of the radiation dose to be delivered to the samples. A total of seven tests were performed in duplicate; thus, a total of 14 samples were tested for each dose, except for doses of 0.5, 1.0 and 1.5 kGy, which corresponded to preliminary tests (see Table 3), when only two tests and four samples were tested for each dose. In parallel, non-irradiated eggs were incubated as controls. The radiation doses were calculated according to Eq. (1). Table 2 displays the gamma radiation doses tested. D¼RT
(1)
where D is the radiation dose (kGy); R, dose rate (Gy/h), measured by Fricke dosimetry at 15 cm from the radiation source; T, irradiation time (h). Eggs were recovered from the wastewater by the adapted incubation method (Zerbini and Chernicharo, 2001). Briefly, 1-l of each sample was put at rest for a 24-h period to allow
sedimentation. Supernatants were removed with the aid of a suction pump (FANEN-Model 089/CAL), leaving 200 mL volume in the container. Successive additions of saline solution (0.85% NaCl) to the remaining material, followed by centrifugation steps, were carried out until obtaining a clear sediment. Next, eggs were floated with zinc sulfate solution (ZnSO4$7H2O e Sinth) (1.18 g L1 density) and filtered out with 47 mm diameter, 0.45 mm porosity cellulose ester membranes (Millipore). The retained material was stored in 0.1 N sulfuric acid (H2SO4 e Sinth) solution in 50 mL Falcon tubes that were incubated at 28 C for 28 days and aerated daily (Massara et al., 1990). After the incubation period, the tubes were centrifuged, and the supernatants were carefully removed. The remaining 2 mL were thoroughly mixed and counted in a Sedgewick Rafter chamber with the aid of a microscope (Olympus, model CX31). Eggs with a formed larva were considered viable. The other developmental stages (morula and gastrula) were quantified; these forms were considered non-viable. The percentage of inactivation of A. lumbricoides eggs in the samples was obtained using the following equation: E¼
1
Nf 100 No
(2)
where E is the efficiency of inactivation of A. lumbricoides eggs; Nf, number of viable A. lumbricoides eggs after irradiation; No, total number of fertile A. lumbricoides eggs. The differences between the central tendency measurements (medians) of the percentages of egg inactivation by gamma ray-mediated disinfection of effluents were analyzed with the aid of the non-parametric test for multiple comparisons among independent samples (Kruskal-Wallis); Statistica 6.0 software was used. For this purpose, the hypothesis of equality between the central tendencies (medians) is denied when the p value is lower than a, where a is the level of significance ( p < 0.05).
Table 2 e Gamma radiation doses. Gamma radiation dose (kGy) 0.5 1.0 1.5 2.5 3.5 4.5 5.0
Time spent (min) 5.61e6.41 11.23e12.81 16.86e19.23 23.55e24.43 32.96e33.41 42.10e42.7 47.10e47.73
5526
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 3 e5 5 2 8
wastewater samples. Because the initial amount of eggs in the different experiments was not constant, the samples were compared from the calculation of the percentage of inactivation promoted by each dose of gamma radiation. In this way, the results of the amount of total and viable A. lumbricoides eggs allowed the percentage of inactivated eggs in each sample to be calculated by the application of Eq. (2). The calculation was individually made for each sample, and the results are presented in Fig. 4. No infective larvae were observed for the 5.0 kGy dose of gamma radiation. Figs. 3 and 4 demonstrate that the use of gamma radiation in the disinfection of domestic effluents is relatively effective, considering its effect on A. lumbricoides egg development.
3.3. Doses of gamma radiation recommended for disinfection Fig. 2 e Distribution of the developmental stages of A. lumbricoides eggs following gamma irradiation.
3.
Results
3.1.
Effects of radiation on egg development
Fig. 2 presents the percentage distributions of the different developmental stages of the A. lumbricoides eggs in the irradiated wastewater samples and controls. The different doses of gamma radiation had little effect on egg development up to the gastrula stage, as 60% of the eggs were able to achieve this stage. Fig. 3A and B shows that gamma radiation affected A. lumbricoides egg development and prevented the formation of infective larva. Non-viable eggs are no longer considered epidemiologically relevant.
3.2.
Efficiency of Ascaris lumbricoides egg inactivation
Table 3 displays the descriptive statistics of the amounts of total and viable eggs identified in the gamma-irradiated
The radiation doses were compared according to the statistical test Kruskal-Wallis. The p values calculated when the medians among percentages of inactivation in the doses of gamma radiation tested were compared. The hypothesis of equality between the central tendencies (medians) is denied when the p value is less than 0.05. Significant differences between the median percentages of inactivation of non-irradiated and irradiated samples with doses above 3.5 kGy were confirmed by the statistical tests ( p < 0.05). No significant difference between the radiation doses of 3.5 and 5.0 kGy was found ( p > 0.05).
4.
Discussion
4.1.
Effects of radiation on egg development
Gamma radiation has a short wavelength, on the order of 104 nm, making it highly penetrating. Thus, when gamma radiation acts on a pathogenic organism, it can cause both internal and external chemical changes (Hill, 2003; Horak, 1994). In the case of nematode eggs, the gamma irradiation most likely affects the biomolecules inside the eggs. If the radiation interacts with organelles such as the centrioles, it
Fig. 3 e (A, B) Non-viable A. lumbricoides eggs irradiated with 5.0 kGy of gamma radiation.
5527
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 3 e5 5 2 8
Table 3 e Descriptive statistics of the amount of A. lumbricoides eggs. Dose (kGy)
0.5a
0.0
1.0a
1.5a
2.5
3.5
4.5
5.0
Total Viable Total Viable Total Viable Total Viable Total Viable Total Viable Total Viable Total Viable eggs eggs eggs eggs eggs eggs eggs eggs eggs eggs eggs eggs eggs eggs eggs eggs
Mean (g L1) Max. (g L1) Min. (g L1) SD N
346
297
363
275
371
175
302
55
337
10
384
6
389
1
368
0
663
556
641
475
527
263
471
101
799
37
862
32
819
10
671
0
168
121
113
89
194
116
218
27
65
2
225
0
174
0
216
0
197 6
175 6
285 4
213 4
148 4
70 4
115 4
33 4
174 14
9 14
157 14
9 14
180 14
3 14
137 14
0 14
SD, standard deviation; N, number of samples. a Results of doses 0.5, 1.0 and 1.5 kGy refer to preliminary studies and used for the purpose of constructing the dose-response curve.
would damage the mitotic spindle that is essential for embryonic development. Thus, the radiation would prevent cell division and impact the initial developmental stages (morula). However, if the radiation damaged other organelles such as the ribosome or even the DNA itself, there may be deficient production of the enzymes required to coordinate egg development. Therefore, although the cells could divide, the lack of enzymes would prevent the continuation of egg development to the larva stage. In this scenario, egg development would stop at one of the more advanced stages (gastrula). As approximately 60% of the recovered eggs were found in the gastrula stage, these results strongly suggest that the gamma irradiation damaged the biomolecules and organelles required for enzyme synthesis and regulation. These effects interrupt the development of the egg, making it inactive and without epidemiological importance. No data were found in the literature that takes into consideration the action of the gamma radiation on the organelles within internal nematode egg cells.
4.2.
Efficiency of Ascaris lumbricoides egg inactivation
Although 84.8% of the eggs in the control samples of this study were viable, 100% of the A. lumbricoides eggs exposed to 5 kGy of gamma radiation were inactivated. These results are in agreement with the results of Bastos (2007), who suggested
Fig. 4 e Dose-response curve regarding the inactivation of A. lumbricoides eggs following gamma irradiation.
a dose between 2.5 and 5 kGy for the inactivation of A. lumbricoides eggs in domestic wastewater. A hypothesis to explain that the doses necessary for helminth eggs inactivation were higher for wastewater samples (present study) than for sludge samples (Horak, 1994; Shamma and Al-Adawi, 2002), which present much higher solids content and therefore should demand higher radiation doses, is the origin of the eggs used as an efficiency indicator in each study, as well as their viability percentage prior to irradiation with gamma rays. Another hypothesis involves the type of sample irradiated. It is known that some components, such as oxygen, which is potentially present in aerobic sludge samples, promote a synergetic effect with radiation and increase the action of the gamma radiation (Melmed and Comninos, 1979). This may explain, at least in some studies (Horak, 1994), the lower radiation doses reported to achieve 100% inactivation of Ascaris sp. eggs.
4.3. Doses of gamma radiation recommended for disinfection According to the statistical tests, there was no significant difference among the doses of gamma radiation of 3.5, 4.5 and 5.0 kGy; however, the 5 kGy dose resulted in 100% egg inactivation in each of the seven experiments. The quality of the effluent, in terms of concentration of nematode eggs, is directly related to the disease, endemicity and sanitation of the area as well as to the socioeconomic situation of the population (WHO, 2004). In this context, this research suggests that for effluents with lower concentrations of nematode eggs, such as 10e80 nematode eggs per liter, a 3.5 kGy dose of gamma radiation could be adequate to ensure adequate effluent quality. However, in the case of more concentrated effluents, such as those containing greater than 500 nematode eggs per liter, the recommended dose would be 5.0 kGy of gamma radiation for wastewater disinfection. Thus, the application of gamma radiation for the disinfection of domestic wastewater is promising because it fits the effluent within the limits recommended by the World Health Organization for use of treated wastewater (less than 1 egg per liter) (WHO, 1989, 2004, 2006; Ayres and Mara, 1996).
5528
5.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 3 e5 5 2 8
Conclusions
This work suggests that gamma radiation acts on the biomolecules and organelles of A. lumbricoides eggs, affecting protein production. Although this allowed eggs to develop to more advanced stages, the eggs were ultimately non-viable and without epidemiological importance. The radiation dose of 5 kGy hindered the development of infective larvae, whereas 85% of the non-irradiated eggs (control) developed into the infective form. The radiation gamma doses of 3.5 kGy were effective for the disinfection of effluents with lower concentrations of A. lumbricoides eggs, whereas higher radiation doses of 5 kGy were necessary for the disinfection of effluents with higher concentrations of eggs. Although it is still expensive, the technology for the disinfection of sewage with gamma rays is potentially applicable in the near future, especially in the case of worsening water shortage.
Acknowledgments The authors thank the following institutions for the support provided: Centro de Tecnologia de Desenvolvimento Nuclear e CDTN; Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´gico e CNPq; Financiadora de Estudos e Projetos e FINEP (PROSAB); Fundac¸a˜o de Amparo a` Pesquisa de Minas Gerais e FAPEMIG (project no 14.107); Instituto de Ana´lises Clı´nicas Hermes Pardini; Pro-Reitoria de Pesquisa da Universidade Federal de Minas Gerais e UFMG; Universidade Federal de Minas Gerais (Department of Parasitology and Department of Sanitary and Environmental Engineering).
references
Al-Adawi, M.A., Albarodi, H., Hammoudeh, A., Shamma, M., Sharabi, N., 2006. Accelerated larvae development of Ascaris lumbricoides eggs with ultraviolet radiation. Radiation Physics and Chemistry 75 (1), 115e119. ASTM-E-1026-95, 2002. Practice for Using the Fricke Reference Standard Dosimetry System. American Society for Testing and Materials, West Conshohocken, PA.; USA. Ayres, R.M., Mara, D.D., 1996. Analysis of wastewater for use in agriculture. A Laboratory Manual of Parasitological and Bacteriological Techniques. WHO e World Health Organization, Geneva, pp. 1e35. Bao, H., Liu, Y., Jia, H., 2002. A study of irradiation in the treatment of wastewater. Radiation Physics and Chemistry 63 (3e6), 633e636. Basfar, A.A., Abdel Rehim, F., 2002. Disinfection of wastewater from a Riyadh Wastewater Treatment Plant with ionizing radiation. Radiation Physics and Chemistry 65 (4, 5), 527e532.
Bastos, G.S.M., 2007. Application of gamma radiation for the disinfection of sewage treated in UASB reactor. In: 24 Proc. Brazilian Congress on Sanitary and Environmental Engineering, ABES (in Portuguese). Borrely, S.I., Cruz, A.C., Del Mastro, N.L., Sampa, M.H.O., Somessari, E.S., 1998. Radiation processing of sewage and sludge. A review. Progress in Nuclear Energy 33 (1, 2), 3e21. Capizzi, S., Schwartzbrod, J., 2001. Irradiation of Ascaris ova in sludge using an electron beam accelerator. Water Research 35 (9), 2256e2260. Chmielewski, A.G., Zimek, Z., Bryl-Sandelewska, T., Kosmal, W., Kalisz, L., Kaimierczuk, M., 1995. Disinfection of municipal sewage sludges in installation equipped with electron accelerator. Radiation Physics and Chemistry 46 (4e6), 1071e1074. Enigk, K., Holl, P., Dey-Hazra, A., 1975. Eradication of parasitic cysts and eggs in sewage sludge by irradiation with low energy electrons. Zbl. Bakt. Hyg 161, 61e71. Galal-Gorchev, H., 1996. Chlorine in water disinfection. Pure and Applied Chemistry 68 (9), 1731e1735. Hill, V.R., 2003. Prospects for pathogen reductions in livestock wastewaters: a review. Critical Reviews in Environmental Science and Technology 33 (2), 187e235. Horak, P., 1994. Experimental destruction of ascarid ova in sewage sludge by accelerated electron irradiation. Water Research 28 (4), 939e941. Massara, C.L., Costa, H.M.A., Carvalho, O.S., 1990. The study of Ascaris lumbricoides in the laboratory. Revista da Sociedade Brasileira de Medicina Tropical 23 (1), 43e47 (in Portuguese). Melmed, L.N., Comninos, D.K., 1979. Disinfection of sewage sludge with gamma radiation. Water SA 5 (4), 153e159. Rawat, K.P., Sharma, A., Rao, S.M., 1998. Microbiological and physicochemical analysis of radiation disinfected municipal sewage. Water Research 32 (3), 737e740. Shamma, M., Al-Adawi, M.A., 2002. The morphological changes of Ascaris lumbricoides ova in sewage sludge water treated by gamma irradiation. Radiation Physics and Chemistry 65 (3), 277e279. Tahri, L., Elgarrouj, D., Zantar, S., Mouhib, M., Azmani, A., Sayah, F., 2010. Wastewater treatment using gamma irradiation: Te´touan pilot station, Morocco. Radiation Physics and Chemistry 79 (4), 424e428. Von Sperling, M., Mascarenhas, L.C.A.M., 2005. Performance of very shallow ponds treating effluents from an UASB reactor. Water Science and Technology 51 (12), 83e90. Wang, J., Wang, J., 2007. Application of radiation technology to sewage sludge processing: a review. Journal of Hazardous Materials 143 (1, 2), 2e7. WHO e World Health Organization, 1989. Health Guidelines for the Use of Wastewater in Agriculture and Aquaculture. Geneva. Technical report 778, pp. 1e76. WHO e World Health Organization, 2004. Integrated Guide to Sanitary Parasitology. Regional Office for Eastern Mediterranean. Regional Centre for Environmental Health Activities, Amman Jordan, pp. 1-120. WHO e World Health Organization, 2006. Guidelines for the safe use of wastewater, excreta and greywater. Wastewater use in agriculture 2, p. 222. Zerbini, A.M., Chernicharo, C.A.L., 2001. Methodologies for quantification, identification and analysis of viability of helminth eggs in raw and treated sewage. In: Chernicharo, C. A.L. (Ed.), Post Treatment of Effluents from Anaerobic Reactors. Methodologic Aspects. PROSAB 2 Belo Horizonte, pp. 71e107 (in Portuguese).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 9 e5 5 3 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Bromate formation in a hybrid ozonation-ceramic membrane filtration system Mohammadreza Moslemi a, Simon H. Davies b, Susan J. Masten a,b,* a b
Department of Civil Engineering, McMaster University, Hamilton, Ontario, Canada Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48824, USA
article info
abstract
Article history:
The effect of pH, ozone mass injection rate, initial bromide concentration, and membrane
Received 30 March 2011
molecular weight cut off (MWCO) on bromate formation in a hybrid membrane
Received in revised form
filtrationeozonation reactor was studied. Decreasing the pH, significantly reduced bromate
16 July 2011
formation. Bromate formation increased with increasing gaseous ozone mass injection
Accepted 8 August 2011
rate, due to increase in dissolved ozone concentrations. Greater initial bromide concen-
Available online 16 August 2011
trations resulted in higher bromate concentrations. An increase in the bromate concentration was observed by reducing MWCO, which resulted in a concomitant increase in the
Keywords:
retention time in the system. A model to estimate the rate of bromate formation was
Bromide
developed. Good correlation between the model simulation and the experimental data was
Bromate
achieved.
Ozonation
ª 2011 Elsevier Ltd. All rights reserved.
Ceramic membrane Water treatment Modeling
1.
Introduction
Ozonation has gained widespread use in drinking water treatment over the past few decades (Gottschalk et al., 2000). Recently, a number of researchers have shown that when ozonation is used in combination with membrane filtration, problems of fouling resulting from the deposition of natural organic matter on the membrane surface or within the membrane pores can be overcome (Van Geluwe et al., 2011; Kim and Van der Bruggen, 2010; Mozia et al., 2006; Karnik et al., 2005). In such hybrid systems, aqueous ozone reacts with and degrades organic foulants accumulated on the membrane surface, thereby decreasing fouling. In this innovative method, ensuring a minimum dissolved ozone concentration enables the continuous treatment of
drinking water and eliminates or greatly reduces the need for membrane cleaning procedures (Kim et al., 2009; You et al., 2007; Mozia et al., 2006; Schlichter et al., 2003). A major concern in the treatment of bromide-containing waters using ozonation is the formation of brominated byproducts (Siddiqui et al., 1995). Bromate (BrO 3 ), an inorganic by-product of ozonation, is a human carcinogen, and the United States Environmental Protection Agency (USEPA), the Ontario (Canada) Safe Drinking Water Act, and the European Union have set the maximum contaminant level (MCL) for bromate at 10 mg/L (Ontario drinking-water quality standards, 2002; European Union, 1998; USEPA, 1998). Bromate is formed by the reaction of molecular ozone with the bromide ion (Br), which is naturally present in many waters. The reactions involved are complex and involve both
* Corresponding author. Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48824, USA. Tel.: þ1 517 355 2254; fax: þ1 517 355 0250. E-mail address: [email protected] (S.J. Masten). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.015
5530
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 9 e5 5 3 4
O3
Br• •
•
HO
HO O3
Br-
BrO• BrO2-
HOBr/OBr-
O3
BrO3-
O3
Fig. 1 e Bromate formation e molecular ozone and hydroxyl radical pathways. Adapted from Pinkernell and von Gunten (2001).
molecular ozone and OH radicals, as shown in Fig. 1 (Crittenden and Harza, 2005; Legube et al., 2004; Pinkernell and von Gunten, 2001; Haag and Hoigne´, 1983). Bromide ion reacts with molecular ozone to form hypobromous acid (HOBr), which is at equilibrium with hypobromite ion (OBr). Subsequently, hypobromite ion reacts with ozone to form bromite ion, which further reacts with molecular ozone to form bromate. Bromide can also react with the hydroxyl radical ( OH) to form the bromine radical, which then reacts with molecular ozone in a complex set of reactions to form bromate. As such, both direct and indirect pathways play a role in bromate formation; however the rate constants for the reaction of hydroxyl radical with bromide are appreciably greater than that of those involving molecular ozone and it has been also shown that the contribution of OH to bromate formation is more significant compared to that of molecular ozone (Mizuno et al., 2004; von Gunten and Oliveras, 1998). It is reported that in Milli-Q water, almost 70% of bromate formation occurs through OH mediated oxidation reactions and the remaining 30% depends on the molecular ozone reactions (Ozekin et al., 1998). While the formation of various bromine containing byproducts in conventional ozonation systems has been studied (Haag and Hoigne´, 1983; Krasner et al., 1993; von Gunten and Pinkernell, 2000), the formation of such
O3 generator
N2
disinfection by-products in a hybrid ozonation-membrane filtration has not been previously investigated. Hence, it is crucial to study bromate formation under various operating conditions in such hybrid systems. In this study, the effect of pH, inlet ozone mass rate, initial bromide concentration, and membrane molecular weight cut off on bromate formation in a membrane filtrationeozonation system was investigated.
2.
Materials and methods
2.1.
Experimental setup
Fig. 2 illustrates the schematic of the reactor that was used in this study. The system was designed to withstand pressures of up to 550 kPa (80 psi) and pressurized using nitrogen gas. The experimental apparatus was equipped with two highpressure stainless steel tanks with capacities of 5 and 8 L. The smaller tank was located in the recycle line to supply the system with the feed water. The second tank operated as reserve raw water supply to replace the water lost from the system through permeation and bleeding and was connected to the main tank through a solenoid valve (Type 6013, Bu¨rkert Corp., USA), which was programmed to maintain a constant water level in the main tank. The total water volume and the pressure inside the reactor were maintained at 1.5 L and 138 kPa (20 psi), respectively, throughout the experiments. All equipment, tubing and connections were made of ozone resistant materials and were either Teflon or 316-stainless steel. Water was circulated in the loop at a flow rate of 200 mL/min using a gear pump (Model 000-380, Micropump Inc., USA). The water inside the tank was mixed as a result of the turbulence produced by the jet of water as it was returned to the tank. Gaseous ozone was generated from pressurized oxygen gas (having a purity of 99.999%) using a corona discharge ozone
O3 monitor
Vent
O3 destruction
O2
Valve Pump
F
P
P
Pressure gauge
F
Flow meter
P
Retentate
Bleed
F Feed tank
Main tank
P F
P
Permeate
Fig. 2 e Schematic of the experimental setup.
F
Bleed
P
5531
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 9 e5 5 3 4
generator (Absolute Ozone AE15MC80P, Absolute System Inc., Edmonton, AB, Canada). The gaseous ozone concentration was adjusted by varying the ozone generator’s voltage. Ozone gas was injected into the water stream flowing in the recirculation loop through a Swagelok Tee fitting (Model SS-400-3, Swagelok, USA). Ozone was transferred from the gas phase to the aqueous phase, as a result of the hydraulic mixing inside the reactor. The inlet gas flow rate was adjusted to the desired values using a digital flow-controller (Model MC-500SCCM-D, Alicat Scientific Inc., USA). Any gas exhausted from the system was destroyed by purging the gas through a potassium iodide (2% KI) solution. A pressure regulator (Model KCB1F0A2A5P20000, Swagelok, USA) was employed to monitor and regulate the pressure in the recirculation loop. The water temperature was kept at 20 0.1 C by recycling water through a water coil which was installed inside the main tank. The temperature of the chilled water was controlled using a refrigerated circulator bath (NESLAB RTE10, Thermo Fisher Scientific, USA). In these experiments, tubular ultrafiltration (UF) membranes (TAMI North America, QC, Canada) with nominal molecular weight cutoffs of 5, 8 and 15 kilodaltons (kD) were used. UF membranes, rather than nanofiltration (NF) membranes, were chosen because the desired flux can be achieved at a lower operating pressure with a UF membrane, than with an NF membrane. Ceramic membranes were used as they are ozone resistant. The active length and the external diameter of the employed membranes were 25 cm and 10 mm, respectively. A stainless steel housing (TAMI North America, QC, Canada) was used to hold the membrane. A bleed line was installed in the recycle loop to allow the retentate to be sampled. The bleed flow rate was set to 5.0 mL/ min using a digital flow-controller (Model LC-50CCM-D, Alicat Scientific Inc., USA). A digital flow meter (Model L-5LPM-D, Alicat Scientific Inc., USA) and a digital pressure gauge (Model 2074, Ashcroft Inc., USA) were mounted in the loop to monitor the water flow rate and pressure within the system. The permeate flux was continuously measured using a balance (Adventurer Pro, Ohaus Corporation, USA). Temperature, pressure, and cross flow rate were continuously recorded using a data acquisition system (LabView, National Instruments, USA).
2.2.
of a high performance liquid chromatograph (Model 230, Varian ProStar, USA) with an anion-exchange column (IonPac AS23, Dionex Corp., USA) a guard column (IonPac AG23, Dionex Corp., USA), an anion micromembrane suppressor (AMMS III e 4 mm, Dionex Corp., USA) and a conductivity detector (CD25A, Dionex Corp., USA). The analysis was carried out in accordance with the method proposed by Dionex (Application Note 184, Dionex Corp., USA). A solution of 4.5 mM sodium carbonate (NaCO3) (99.5%, EMD, USA) and 0.8 mM sodium bicarbonate (NaHCO3) (99.7%, SigmaeAldrich, Germany) was used as the mobile phase. The column flow rate was 1 mL/min. pH was measured using a pH meter (pHTestr 30, Eutech Instruments, Illinois, USA).
2.3.
Reagents
Ultrapure water (Milli-Q water with a resistivity greater than 18 MU) was used to prepare all reagents and solutions. Milli-Q water was spiked with sodium bromide (NaBr) to obtain the desired concentrations. Subsequently, the pH of water was adjusted to the desired value by adding phosphoric acid (H3PO4) (99.99%, SigmaeAldrich, USA) and/or sodium hydroxide (NaOH) (98.0þ%, Fluka, Germany). The water was not buffered, since our experimental results showed that the pH did not vary over the course of the experiment. Tertiary butyl alcohol (Fisher Scientific, Germany) was used as hydroxyl radical scavenger. Bromide and bromate standards were prepared using sodium bromide (99.995% (metals basis), Fluka, Germany) and sodium bromate (NaBrO3) (99.7þ%, Fisher Scientific, USA).
3.
Results and discussion
The effects of pH, ozone mass injection rate, initial bromide concentration, and membrane molecular weight cut off (MWCO) on bromate formation were investigated. Steady state bromae concentrations observed under various operating conditions are presented in Table 1. All bromate concentrations reported are averages of triplicate analyses within experiments. In all cases, the relative standard deviation for the measurements was less than 3.6%.
Analytical methods
The concentrations of ozone in the permeate and the retentate were continuously monitored by measuring the absorbance at a wavelength of 258 nm using flow-through cells mounted in a UV/Vis spectrophotomer (Model 4054, Pharmacia LKB, Biochrom Ltd., UK). The aqueous ozone concentration was then determined using an extinction coefficient of 2900 M1 cm1 (Hoigne´ and Bader, 1976). The inlet ozone gas concentration was also continuously measured using a UV ozone monitor (Model 454H, Teledyne Instruments, USA). To quench further ozone reactions, the residual dissolved ozone in the samples for bromide and bromate analysis were purged by bubbling nitrogen gas through the sample for 2 min. Bromide and bromate concentrations in the samples were measured in triplicate using ion chromatography, consisting
Table 1 e Steady state bromate concentration in the permeate under various operating conditions. pH
3 6 8 6 6 6 6 6 6 6
Steady Inlet O3 Initial MWCO Temperature mass [Br] (kD) ( C) state [BrO 3] (mg/min) (mg/L) (mM) 1.5 1.5 1.5 0.5 1.0 0.75 2.25 1.5 1.5 1.5
1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.05 0.25 0.50
5 5 5 5 5 5 5 5 5 5
20 20 20 20 20 20 20 20 20 20
0.58 4.13 8.03 1.57 2.86 2.55 4.67 0.055 0.72 1.75
5532
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 9 e5 5 3 4
10
5
Inlet O3 mass=2.25 mg/min Inlet O3 mass=1.5 mg/min Inlet O3 mass=1.0 mg/min Inlet O3 mass=0.75 mg/min Inlet O3 mass= 0.5 mg/min
4
8
[BrO3-] (micromole/L)
[BrO3-] (micromole/L)
9
7 6 5 4
pH=8.0 pH=6.0 pH=3.0
3 2
3
2
1
1 0
0 0
30
60
90
120
150
0
180
30
60
90 Time (min)
Time (min)
Fig. 3 e Bromate concentration in the permeate vs. time at various pH values. Membrane: 7 channel e 5 kD, inlet O3 mass injection rate: 1.5 mg/min, initial [BrL]: 1 mg/L, temperature: 20 C, ionic strength: 512.5, 13.0 and 12.5 mM corresponding to pH 3, 6 and 8, respectively.
3.1.
120
150
180
Fig. 5 e Bromate concentration in the permeate vs. time at various inlet O3 mass injection rates. Membrane: 7 channel e 5 kD, initial [BrL]: 1 mg/L, temperature: 20 C, pH: 6.0, ionic strength: 13.0 mM.
is an effective way to control bromate formation, doing so is neither cost effective nor practical in terms of full-scale drinking water treatment (Crittenden and Harza, 2005).
Effect of pH
As shown in Fig. 3, the steady state concentration of bromate ion measured in the permeate line at a reactor pH of 3.0 is significantly less then that observed at pH 6.0 or 8.0. This is not surprising since the hydroxyl radical ( OH) is more reactive with bromide than is the ozone molecule (von Gunten and Oliveras, 1998) and at low pH, the ozone molecule is more stable and the concentration of hydroxyl radicals in the reactor would be expected to be less than that found at higher pH (Pinkernell and von Gunten, 2001; von Gunten, 2003). Furthermore, HOBr is much less reactive with ozone than is OBr (Mizuno et al., 2004). The pKa for the dissociation of hypobromous acid to hypobromite (HOBr 4 OBr þ Hþ) is 8.8 (Haag and Hoigne´ 1983; Siddiqui and Amy, 1993), so the concentration of OBr is low at low pH, hence the rate of formation of bromate is slower at low pH. While pH reduction
3.2.
Effect of ozone mass injection rate
The effects of the ozone mass injection rate (0.50, 0.75, 1.00, 1.50, 2.25 mg/min) on the dissolved ozone and bromate concentration are shown in Figs. 4 and 5, respectively. The aqueous ozone concentrations increased with time until steady state conditions were attained. The steady state aqueous ozone concentration ([O3]ss) was found to be proportional to the ozone mass injection rate (see inset in Fig. 4), which is also consistent with the experimental results presented by Moslemi et al. (2010). Higher ozone mass injection rates resulted in greater bromate concentrations in the permeate. As shown in Fig. 6, there is a strong linear relationship between the steady state bromate concentration in the permeate and the steady state dissolved ozone concentration
14
[O ]
Inlet O3 mass= 2.25 mg/min Inlet O3 mass=1.5 mg/min Inlet O3 mass=1.0 mg/min Inlet O3 mass=0.75 mg/min Inlet O3 mass=0.5 mg/min
R = 0.933
10 Inlet O mass
8
y = 0.480x + 0.114 2 R = 0.996
4
3
2
-
6
5
[BrO3 ]ss (micromole/L)
Aqueous [O ] (mg/L)
12
4 2 0
1
0 0
30
60
90
120
150
180
Time (min)
Fig. 4 e Dissolved ozone concentration in the permeate vs. time at various inlet O3 mass injection rates. Membrane: 7 channel e 5 kD, initial [BrL]: 1 mg/L, temperature: 20 C, pH: 6.0, ionic strength: 13.0 mM.
0
1
2
3
4
5
6
7
8
9
Aqueous [O3]ss (mg/L)
Fig. 6 e Steady state bromate concentration in the permeate vs. steady state aqueous O3 concentration. Membrane: 7 channel e 5 kD, initial [BrL]: 1 mg/L, temperature: 20 C, pH: 6.0, ionic strength: 13.0 mM.
10
5533
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 9 e5 5 3 4
Table 3 e Retention time in the membrane for various MWCOs.
4
y = 4.186x - 0.176 2 R = 0.992
3
2
-
[BrO3 ]ss (micromole/L)
5
MWCO (kD)
Vmp (mL)
Wm (g)
Vmd (mL)
VP (mL)
J (mL/s)
q (s)
5 8 15
14.1 14.1 14.1
38.7 38.8 40.2
10.2 10.2 10.6
3.96 3.93 3.57
0.116 0.169 0.198
34 23 18
1
0 0
0.25
0.5
0.75
1
-
Initial [Br ] (mg/L)
Fig. 7 e Steady state bromate concentration in the permeate vs. initial bromide concentration. Membrane: 7 channel e 5 kD, inlet O3 mass injection rate: 1.5 mg/min, temperature: 20 C, pH: 6.0.
(r2 ¼ 0.996). This result shows the importance of developing processes that effectively minimize DBP formation by operating at the lowest possible dissolved ozone concentrations.
there is a strong linear correlation (r2 ¼ 0.999) between the steady state bromate concentration and the retention times in the membrane (r2 ¼ 0.999) and the entire system (r2 ¼ 0.993). This implies that in hybrid systems, bromate formation is significantly affected by the contact times in the membrane and the entire system, although the latter likely predominates simply because of its greater magnitude. These results suggest that bromate formation can be minimized by employing membranes with the largest MWCO feasible for the particular operation.
4.
In another set of experiments, the initial bromide concentration was varied (0.05, 0.25, 0.5 or 1.0 mg/L) (see Fig. S1). As it is shown in Fig. 7, there is a strong correlation between the steady state bromate concentration in the permeate and the initial bromide concentration in the feed solution (r2 ¼ 0.992) indicating that as more bromide ion was available to react with molecular ozone and/or hydroxyl radicals, more bromate was produced.
3.4.
Effect of MWCO
As presented in Table 2, lower bromate concentrations were observed with membranes having higher MWCOs (also see Fig. S2). The permeability of the membranes tested increases with MWCO, so at a constant operating pressure, as the MWCO of the membrane increases, the permeate flux also increases, and the residence time inside the reactor decreases. Therefore, for membranes with greater MWCOs, the ozone exposure (the dissolved ozone concentration multiplied by retention time) is reduced. The retention times for each system and the three different membranes were calculated (see Supporting Documentation for details). The retention times for the membranes with MWCOs of 5, 8, and 15 kD were 34, 23, and 18 s, respectively (Table 3). As can be seen in Fig. 8
A kinetic-based model was developed to predict the rate of bromate formation in the hybrid ozone-membrane system. Experimental results reported herein were used to calculate the reaction rate constant. Bromate formation in the employed reactor can be expressed using the following mass balance equation: V
dCR ¼ Qin Cin QP CP QB CR þ kCR V dt
(1)
where V, total volume of water in the system (1500 mL); t, time (min); CR, concentration of bromate in the system (retentate) (mM); Cin, concentration of bromate in the inflow into the system (mM); CP, concentration of bromate in the permeate (mM); Qin, inlet flow rate (mL/min); QP, permeate flow rate (mL/ min); QB, bleed flow rate (mL/min); k, rate constant of bromate formation (1/min).
5
-
Initial bromide concentration
[BrO3 ]ss (micromole/L)
3.3.
Bromate formation model
y = 0.090x + 1.060 2 R = 0.999
4
3
2
1
Table 2 e Steady state [BrOL 3 ] at various MWCOs.
0 0
MWCO (kD)
pH
Inlet O3 mass (mg/min)
Initial [Br] (mg/L)
Steady state [BrO 3 ] (mM)
5 8 15
6.0 6.0 6.0
1.5 1.5 1.5
1.00 1.00 1.00
4.13 3.18 2.67
10
20 Retention time (s)
30
40
Fig. 8 e Steady state bromate concentration in the permeate vs. retention time in the membrane. Membrane: 7 channel, inlet O3 mass: 1.5 mg/min, initial [BrL]: 1 mg/L, temperature: 20 C, pH: 6.0, ionic strength: 13.0 mM.
5534
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 2 9 e5 5 3 4
Table 4 e Rate constants for the formation of bromate in the hybrid ozonation-membrane system. Inlet O3 mass (mg/min)
Rate constant (min1)
Relative standard deviation (%)
0.0061 0.0063 0.0064 0.0067 0.0068
4.8 6.7 2.4 4.3 5.6
0.5 0.75 1.0 1.5 2.25
At steady state, the bromate concentration in the system will not vary with time (dC/dt ¼ 0). Moreover, the feed water was free of bromate ion (Cin ¼ 0). Hence, after substitution into Eq. (1), the rate constant of bromate formation can be obtained using the following equation. k¼
QP CP þ QB CR CR V
(2)
Using Eq. (2), the rate constants for the experiments with varying ozone mass injection rate were calculated (see Table 4). At a pH of 6.0, the rate constants do not vary significantly with ozone mass injection rate. Hence, one can assume that the mechanism for bromate formation is independent of ozone mass injection rate at a constant pH. At an ozone mass injection rate of 1.5 mg/min, the rate constants for bromate formation at pH 3, 6, and 8 were determined to be 0.0061, 0.0067, and 0.0085 min1, respectively. Under these conditions, the aqueous ozone concentrations were 9.6, 8.3, and 4.9 mg/L, for solutions at pH values of 3, 6, and 8, respectively. As discussed earlier, the increased overall rate of reaction is likely due to an increase in the OH radical concentration at higher pH and the dissociation of hypobromous acid to OBr, which is more reactive with ozone than is HOBr.
Acknowledgements The authors acknowledge financial support of this work from the National Science Foundation Research (Grant No. CBET0506828) and from the Natural Sciences and Engineering Research Council of Canada Discovery Grant Program.
Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.015.
references
Crittenden, J., Harza, M.W., 2005. In: Water Treatment: Principles and Design, second ed. MWH, John Wiley, New Jersey. European Union, 1998. Official Journal of the European Communities, L 330, Council Directive 98/83/EC. Gottschalk, C., Libra, J.A., Saupe, A., 2000. Ozonation of Water and Waste Water: A Practical Guide to Understanding Ozone and
Its Application. Wiley-VCH, New York. von Gunten, U., Oliveras, Y., 1998. Advanced oxidation of bromide-containing waters: bromate formation mechanisms. Environ. Sci. Technol. 32, 63e70. von Gunten, U., Pinkernell, U., 2000. Ozonation of bromide-containing drinking waters: a delicate balance between disinfection and bromate formation. Water Sci. Technol. 41, 53e59. von Gunten, U., 2003. Ozonation of drinking water: part II. Disinfection and by-product formation in presence of bromide, iodide or chlorine. Water Res. 37, 1469e1487. Haag, W.R., Hoigne´, J., 1983. Ozonation of bromide-containing waters: kinetics of formation of hypobromous acid and bromate. Environ. Sci. Technol. 17, 261e267. Hoigne´, J., Bader, H., 1976. The role of hydroxyl radical reactions in ozonation processes in aqueous solutions. Water Res. 10, 377e386. Karnik, B.S., Davies, S.H.R., Chen, K.C., Jaglowski, D.R., Baumann, M. J., Masten, S.J., 2005. Effects of ozonation on the permeate flux of nanocrystalline ceramic membranes. Water Res. 39, 728e734. Kim, J., Van der Bruggen, B., 2010. The use of nanoparticles in polymeric and ceramic membrane structures: review of manufacturing procedures and performance improvement for water treatment. Environ. Pollut. 158, 2335e2349. Kim, J., Shan, W., Davies, S.H.R., Baumann, M.J., Masten, S.J., Tarabara, V., 2009. Interactions of aqueous NOM with nanoscale TiO2: implications for ceramic membrane filtrationeozonation hybrid process. Environ. Sci. Technol. 43, 5488e5494. Krasner, S.W., Glaze, W.H., Weinberg, H.S., Daniel, P.A., Najm, I.N. , 1993. Formation and control of bromate during ozonation of waters containing bromide. J. AWWA 85, 73e81. Legube, B., Parinet, B., Gelinet, K., Berne, F., Croue, J.P., 2004. Modeling of bromate formation by ozonation of surface waters in drinking water treatment. Water Res. 38, 2185e2195. Mizuno, T., Yamada, H., Tsuno, H., 2004. Formation of bromate ion through a radical pathway in a continuous flow reactor. Ozone Sci. Eng. 26, 573e584. Moslemi, M., Davies, S.H., Masten, S.J., 2010. Ozone mass transfer in a recirculating loop semibatch reactor operated at high pressure. Adv. Oxidation Technol. 13, 79e88. Mozia, S., Tomaszewska, M., Morawski, A.W., 2006. Application of an ozonation-adsorption-ultrafiltration system for surface water treatment. Desalination 190, 308e314. Ontario drinking-water quality standards, 2002. Safe Drinking Water Act, Ontario regulation 169/03. Ozekin, K., Westerhoff, P., Amy, G.L., Siqqiqui, M., 1998. Molecular ozone and radical pathways of bromate formation during ozonation. J. Environ. Eng. 124, 456e462. Pinkernell, U., von Gunten, U., 2001. Bromate minimization during ozonation: mechanistic considerations. Environ. Sci. Technol. 35, 2525e2531. Schlichter, B., Mavrov, V., Chmiel, H., 2003. Study of a hybrid process combining ozonation and membrane filtration filtration of model solutions. Desalination 156, 257e265. Siddiqui, M.S., Amy, G.L., 1993. Factors affecting DBP formation during ozone-bromide reactions. J. AWWA 85, 63e72. Siddiqui, M.S., Amy, G.L., Rice, R.G., 1995. Bromate ion formation: a critical review. J. AWWA 87, 58e70. USEPA, 1998. National primary drinking water regulations: disinfection and disinfection byproducts. Federal Reg. 63, 69390e69476. Van Geluwe, S., Braeken, L., Van der Bruggen, B., 2011. Ozone oxidation for the alleviation of membrane fouling by natural organic matter: a review. Water Res. 45, 3551e3570. You, S.H., Tseng, D.H., Hsu, W.C., 2007. Effect and mechanism of ultrafiltration membrane fouling removal by ozonation. Desalination 202, 224e230.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Prediction of contamination potential of groundwater arsenic in Cambodia, Laos, and Thailand using artificial neural network Kyung Hwa Cho a,1, Suthipong Sthiannopkao b,1, Yakov A. Pachepsky c, Kyoung-Woong Kim d, Joon Ha Kim d,e,* a
Department of Civil and Environmental Engineering, University of Massachusetts, Amherst, 130 Natural Resources Road, Amherst, MA 01003, USA b Department of Environmental and Occupational Health, National Cheng Kung University, Tainan City, Taiwan, ROC c USDA-ARS, Environmental Microbial & Food Safety Laboratory, 10300 Baltimore Ave., Beltsville, MD 20705, USA d School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), 261 Cheomdan-gwagiro, Buk-gu, Gwangju 500-712, Republic of Korea e Sustainable Water Resource Technology Center (SWRTC) at GIST, 261 Cheomdan-gwagiro, Buk-gu, Gwangju 500-712, Republic of Korea
article info
abstract
Article history:
The arsenic (As) contamination of groundwater has increasingly been recognized as
Received 10 February 2011
a major global issue of concern. As groundwater resources are one of most important
Received in revised form
freshwater sources for water supplies in Southeast Asian countries, it is important to
29 May 2011
investigate the spatial distribution of As contamination and evaluate the health risk of As
Accepted 8 August 2011
for these countries. The detection of As contamination in groundwater resources, however,
Available online 27 August 2011
can create a substantial labor and cost burden for Southeast Asian countries. Therefore, modeling approaches for As concentration using conventional on-site measurement data
Keywords:
can be an alternative to quantify the As contamination. The objective of this study is to
Multiple linear regression
evaluate the predictive performance of four different models; specifically, multiple linear
Principal component regression
regression (MLR), principal component regression (PCR), artificial neural network (ANN),
Artificial neural network
and the combination of principal components and an artificial neural network (PC-ANN) in
Principal component-artificial
the prediction of As concentration, and to provide assessment tools for Southeast Asian
neural network
countries including Cambodia, Laos, and Thailand. The modeling results show that the prediction accuracy of PC-ANN (NasheSutcliffe model efficiency coefficients: 0.98 (traning step) and 0.71 (validation step)) is superior among the four different models. This finding can be explained by the fact that the PC-ANN not only solves the problem of collinearity of input variables, but also reflects the presence of high variability in observed As concentrations. We expect that the model developed in this work can be used to predict As concentrations using conventional water quality data obtained from on-site measurements, and can further provide reliable and predictive information for public health management policies. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), 261 Cheomdan-gwagiro, Buk-gu, Gwangju 500-712, Republic of Korea. Tel.: þ82 62 970 3277; fax: þ82 62 970 243. E-mail address: [email protected] (J.H. Kim). 1 Cho and Sthiannopkao contributed equally to this work and are listed in alphabetical order. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.010
5536
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
Introduction
Groundwater resources are the most important components of drinking water supplies for Southeast Asian countries (Berg et al., 2007). Arsenic (As) contamination of groundwater has become a major problem on a worldwide scale because As is a carcinogenic element, which is mostly present as an inorganic species in natural water systems (Bagla and Kaise, 1996; AWWA, 2001; Berg et al., 2006; Kocar and Fendorf, 2009; Sthiannopkao et al., 2010). Naturally occurring As-enriched groundwater has been observed in tube well/hand pump drinking water supplies in South and Southeast Asia, including Thailand, Vietnam, Loa PDR, Cambodia, Myanmar, Bangladesh, India, Nepal, and Pakistan (Berg et al., 2001a,b; Smedley and Kinniburgh, 2002; Sun et al., 2002; Polya et al., 2003, 2005; Stanger et al., 2005; Kohnhorst, 2005; Tetsuro et al., 2006; Chiew et al., 2009). It is estimated that about 200 million people are at risk of harmful toxic effects of As in these countries (Sun et al., 2002). In Southeast Asian countries, monitoring strategies of As contamination need to be improved in order to quantify As concentrations in groundwater, which can then be used to provide further information to better assess and manage public health. The detection of As contamination of groundwater resources, however, is hampered by a substantial labor and cost burden for Southeast Asian countries; it requires sophisticated equipment, highly skilled technicians, and incurs a high maintenance cost. As such, dealing with local As-associated problems in Cambodia and Laos remains problematic, where facilities for determining arsenic concentrations as well as human resources and funding for analyzing arsenic are still very much lacking. Therefore, modeling approaches for As concentrations using on-site measurement data can be an alternative to characterizing the As contamination potential, to provide predictive information for better public health management. Indeed, obtaining on-site data such as pH, redox potential (Eh), and salinity are less expensive and much easier to perform than to detect and measure As contamination via graphite atomic absorption spectrophotometry (AAS) or inductively coupled plasma mass spectroscopy (ICP-MS). It is thought that these simple measurements can be used to fill the gap incurred by the missing facilities, human resources, and funding for arsenic determination using sophisticated equipment. The most popular modeling approaches currently in use are multiple linear regression (MLR) and principal component regression (PCR), both of which attempt to find the most appropriate predictive model by fitting a linear equation to multiple observed data (explanatory variables). However, MLR and PCR may not be successful due to their statistical assumptions, including absence of outliers, normality, and randomness, which can be easily violated (Gros, 1997). In particular, the existence of correlations among explanatory variables (referred to here as “collinearity”) diminishes the statistical stability (or robustness) and may cause significantly high prediction errors (Mac Nally, 2002). Moreover, incorporation of too many redundant (or insignificant), or too few, explanatory variables into models is not practical; thus,
a stepwise regression technique is often required to determine the optimal number of variables to use as explanatory variables (over- or underspecification) (Luan et al., 2008). Furthermore, even though the predictive power of MLR and PCR can be acceptable, the models often result in the poor accuracies when making predictions with new datasets in the validation step. Therefore, more complex nonlinear models have been developed that achieve better predictions, and have subsequently been applied to water resources problems such as hydrological processes, water quality problems, and dam operations (Maier and Dandy, 1996; Wen and Lee, 1998; Lee et al., 2003; Riad et al., 2004; Sarangi and Bhattacharya, 2005; Tayfur et al., 2005; Holmberg et al., 2006; Kuo et al., 2008). These nonlinear models such as artificial neural networks (ANNs), however, are not only difficult to construct, but also often cause over-fit problems in predictions; furthermore, the straightforward interpretation of relationships between explanatory and dependent variables cannot be achieved by nonlinear models. A few researchers have applied ANN to the prediction of As in groundwater (Purkait et al., 2008; Chang et al., 2010). For example, Purkait et al. (2008) used conventional water quality parameters to predict As contamination in Eastern India in an attempt to select the best model among linear and nonlinear models. Chang et al. (2010) applied an ANN model to recover missing values in As concentration datasets in an area of Taiwan. This study utilizes a comprehensive dataset from three different countries, having a wide range of As concentration levels and conventional water quality parameters. The objective of this study is to analyze and evaluate the predictive performance of MLR, PCR, ANN, and the sequential combination of principal component and ANN (PC-ANN) in the prediction of As concentrations in groundwater and thereby provide improved assessment tools for Southeast Asian countries. In addition, a sensitivity analysis is also applied to investigate the cause and effect relationship between input parameters and As concentrations.
2.
Materials and methods
2.1.
Field and sampling sites
Groundwater samples were taken in an attempt to investigate the As concentration and five in-situ parameters for three countries: Cambodia, Loas, and Thailand (see Fig. 1). Table 1 shows the information on the timing and study sites. In Cambodia, thirty groundwater samples were collected from six villages in Kandal Province in 2008 ((Prek Thom village, Kbal Kaoh commune, n ¼ 5), (Phoum Thom village, Phoum Thom commune, n ¼ 5), (Chounlork village, Korki commune, n ¼ 5), (Tuol Tnort village, Koki commune, n ¼ 5), (Doun Sor village, Koki commune, n ¼ 5), (Poul Pear Ker village, Khom Day Eth commune, n ¼ 5)). In 2010, 49 samples were taken from south of Phnom Penh. In Laos, a total of 62 tube well samples were collected in 2008 from households located in provinces of Champasack (n ¼ 27), Attapeu (n ¼ 10), Saravane (n ¼ 11), Savannakhet (n ¼ 4), Borikhamxay (n ¼ 7), and
5537
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
Vientiane (n ¼ 3). Vientiane is located in the upstream of the Mekong River, followed by Borikhamxay, Savannakhet, Saravane, Champasack and Attapeu, respectively. The areas along the Mekong River located in middle and southern parts of Laos starting from Borikhamxay downward are floodplain areas. In Thailand, the concentrations of As in groundwater were examined in Tambon Ongphra (n ¼ 10) in Suphan buri province in 2008. In the three countries, samples for total As were collected from tube wells by following this sequence: 1) pumping the tube well for several minutes, 2) washing a clean polyethylene bottle with the well water, and 3) taking water without filtering.
2.2.
Sample analysis and on-site measurement
As concentrations in groundwater samples were measured by graphite atomic absorption spectrophotometry (GF-AAS; Perkin Elmer 5100 PC, with a detection limit of 0.5 mg L1). GF-AAS was calibrated by an external standard technique in the range of 0e100 mg L1. For a quality control, standard reference material (SRM) for natural water (National Institute of Standards & Technology NIST 1640) was used to assure the precision of the measurement. After every tenth sample during analysis, the SRM sample and calibration standards were analyzed to check the analysis accuracy. All samples were measured at least twice in order to assess the measurement reliability; samples were reanalyzed if the error either from the SRM or from the calibration standards exceeded 10% or the
Table 1 e Timing and sites of Arsenic sampling in Southeast Asia. Study area Cambodia Laos Thailand
Sampling time
Number of samples
Number of villages or provinces
2008 2010 2008 2008
30 49 62 10
6 1 6 1
relative standard deviation of the measurement exceeded 5%. Dilution was made with 2% HNO3 when the concentration of the sample was over the upper limit of the standard range (100 mg L1). During the sample collection in three countries, a series of in-situ measurements were conducted: pH, Eh, water temperature (Wt) (HORIBA d-54 meter), electrical conductivity (EC), and total dissolved solids (TDS) (ORION 3 STAR, Thermo Electron Corporation).
2.3. ANN
Modeling approaches: MLR, PCR, ANN, and PC-
All predictive models were developed using on-site data for pH, EC, TDS, Wt, and Eh. These variables were logarithmically transformed in order to normalize them for four different models (Rawlings et al., 1998). A pattern search algorithm and the Latin hypercube-one-factor-at-a-Time (LH-OAT) method were used to optimize parameters and investigate parameter sensitivity analysis of ANN, respectively (see Supplementary Information). Strategies of training, validation, and testing for ANN are also addressed in Supplementary Information.
Table 2 e Results of MLR and PCR for As concentration of groundwater. Regression coefficients Independent variable (i) (Constant)b pH Electrical conductivity Total dissolved solids Temperature Redox potential
2.83 2.60 0.00
2.73 0.89 0.05
1.14 2.67
0.23
0.04
2.83
1.62 0.21
1.74 0.19
1.16 2.29
1.80 0.18 0.13 0.01 0.16 0.34
0.04 0.04 0.04 0.04 0.06 0.04
0.98 0.99 0.99 1.00 0.98
(Constant)b PC1 PC2 PC3 PC4 PC5
Fig. 1 e Map of study the site, showing sampling stations and Mekong River in Cambodia, Laos, and Thailand.
Collinearity statistics
bia
Std. Error SEbi
VIF
Mean VIF 2.02
0.99
a The subscript i indicates the water quality parameter, and bi the computed coefficients of the water qualities in the MLR and PCR models, log(As) ¼ b0 þ b1 pH þ b2 Conductivity þ b3 TDS þ b4 Water temperature þ b5 Redox, log(As) ¼ b0 þ b1 PC1 þ b2 PC2 þ b3 PC3 þ b4 PC4 þ b5 PC5. b (Constant) in the table represents b0 in the MLR and PCR models.
5538
2.3.1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
of ANN; other procedures for the optimization, training, and validation are the same as those for the ANN model.
Multiple linear regression (MLR)
The MLR model was developed as follows: logðyÞ ¼ b0 þ
n X
bi logðxi Þ
(1)
i¼1
where xi is the explanatory variable i, y is the dependent variable, bi is the regression coefficient of explanatory variables i, and b0 is the value of the intercept in the log-linear fitting.
2.3.2.
Principal component regression (PCR)
PCR combines a principal component analysis (PCA) decomposition with MLR (Jolliffe, 2002; Cho et al., 2009). As a result of PCA, a new set of variables (the principal components (PCs)) and PC scores are generated from the orthogonal linear transform of the original data. The PC scores are then used in the regression as explanatory variables.
2.3.3.
Artificial neural network (ANN)
ANN is a useful method for determining pattern classifications of multi-variable datasets as well as the prediction of complex processes. The multilayer perceptron ANN consists of two or more layers of nodes, including an input layer, a hidden layer, and an output layer, which are connected by links with varying weights. The nodal data are multiplied by the weights to compute the signal strength, and then are transferred to the next node in the network; the input layer nodes accept the input vectors and forward the signals to the next layer according to the connection. This process is continued until the signals reach the output layer.
2.3.4. ANN)
Principal component-artificial neural network (PC-
PC-ANN merges PCA decomposition with ANN (Sousa et al., 2007). The main difference between this approach and ANN is that PC scores generated from the orthogonal linear transformation of the original data are used as the input variables
3.
Results and discussion
3.1. Relationship between As concentrations and on-site measurements Fig. 2 shows the average values and standard deviations of As concentrations and conventional water quality parameters in Cambodia, Laos, and Thailand. Arsenic concentrations of groundwater were observed to be the highest in Cambodia and the lowest in Laos. Similarly, pH was the highest in Cambodia, and the lowest in Laos. Conversely, Eh was the highest in Laos, and the lowest in Cambodia. Significant positive (Pearson correlation: 0.25; p-value: 0.00) and negative correlations (Pearson correlation: 0.10, p-value: 0.12) were found between the As concentrations and pH and between the As concentrations and Eh, respectively. Even though the pvalue for Eh is greater than 0.05, the presence of this correlation is consistent with previous studies (Berg et al., 2007; Buschmann et al., 2007), indicating that high arsenic concentrations might be triggered by reducing conditions.
3.2.
Linear models for predicting As concentrations
3.2.1.
MLR for predicting As concentrations
Fig. 3(A) and (B) respectively shows the observed and the predicted As concentrations of groundwater in both the generation and validation steps of two different regression models (i.e., MLR and PCR). Overall, the MLR model did not reproduce the variations of observed As concentrations in either the generation or validation steps; the developed MLR model tends to underestimate As concentrations. Table 2 shows the regression coefficients bi, the corresponding
Fig. 2 e Mean values of As concentration, pH, redox potential, total dissolved solids, electrical conductivity, and temperature along with their standard deviations for Cambodia, Laos, and Thailand.
5539
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
Fig. 3 e Comparison results between the observed and predicted As concentrations of groundwater, (A) and (B): MLR and PCR, (C) and (D): ANN and PC-ANN, closed circles: MLR, open circles: PCR, closed squares: ANN, open squares: PC-ANN.
standard error SEbi ; and the collinearity statistics of the MLR model. Note that if the absolute value of bi in the MLR model is greater than twice its standard error (i.e., SEbi ), the ith variable can be regarded as a significant variable (Rawlings et al., 1998). Here, TDS and pH have regression coefficients greater than twice their standard errors. Furthermore, collinearity was found among the explanatory variables because the largest variance inflation factor (VIF) was greater than 10, with an average VIF value being substantially greater than 1 (Bowerman and O’Connell, 1990; Myers, 1990). Consequently, the computed values of bi and SEbi for a certain explanatory variable strongly rely on the degree of its correlation with the other variables in the MLR model. The modeling accuracy was also compared using the NasheSutcliffe model efficiency coefficient (NSE) computed from the predicted and observed As concentrations (Nash and Sutcliffe, 1970). In essence, the closer the NSE is to 1, the more accurate the model is. In particular, an acceptable NSE value needs to be greater than 0.5, while a good agreement value should be greater than 0.7 (Moriasi et al., 2007). As shown in Table 3, the NSE values of the MLR model (0.29 and 0.75) indicate that the MLR model developed in this study cannot be considered a suitable model for predicting the As concentration of groundwater using onsite measurement data. This result is similar to a previous study in Eastern India; Purkait et al. (2008) found that MLR showed good accuracy for low levels, but that it did not reproduce the high levels of As concentration.
3.2.2.
PCR for predicting As concentrations
In the PCR model, five PCs were generated from five water quality parameters and then used as input variables for
Table 3 e MAEs and NasheSutcliffe model efficiency coefficients of the model predictions for As. Generation/Training steps Validation step
MLR PCR ANN PC-ANN Group A ANN PC-ANN Group B ANN PC-ANN
MAEa
NSEb
MAEa
NSEb
141.59 93.68 79.68 72.08 24.93 22.63 49.68 35.10
0.29 0.14 0.71 0.61 0.83 0.84 0.96 0.98
345.26 102.54 88.93 101.17 24.78 44.09 111.21 75.15
0.75 0.19 0.47 0.66 0.34 0.52 0.74 0.71
a Mean absolute error (MAE) is a statistical approach used to assess the model performance, and its unit is mg L1 Pn 1 MAE ¼ ½n i¼1 jxobs xpre j where n indicates the number of observations of As. Here, xobs and xpre indicate the observed and predicted As concentrations, respectively. PT 2 t t t¼1 ðxobs xpre Þ b NSE ¼ 1 PT where xobs is the observed As 2 t t ðx x Þ t¼1 obs pre concentrations, xpre is the predicted As concentrations, and xtobs is the averaged value of the observed As.
5540
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
predicting As concentrations. Here, the cumulative percentage of the variations explained by the five PCs was 100%, implying that these PCs were able to explain all the variations in the original dataset. The five extracted PCs are significantly related to each water quality parameter, PC1 (Eh, 0.95), PC2 (conductivity, 0.93), PC3 (water temperature, 0.99), PC4 (pH, 1.00), and PC5 (TDS, 0.81). Fig. 3(A) and (B) respectively compares the observed and the predicted As concentrations of groundwater in both the generation and validation steps of PCR. Overall, however, the PCR model did not reproduce variations of the observed As concentrations, usually by underestimating the relatively high As concentrations similar to the MLR. As shown in Table 2, PC1, PC2, PC4, and PC5 had regression coefficients with absolute values being greater than twice their standard errors. In addition, all VIF values for the explanatory variables were equal to or less than 1, thereby implying that the collinearity problem in MLR was completely overcome through the PC application. As shown in Table 2, however, the calculated NSEs imply that the PCR model developed in this study still cannot be used as a reasonable model for predicting groundwater As concentrations using on-site measurement data. Consequently, it demonstrates that the linearity of MLR and PCR cannot reproduce the dynamic variations of observed As concentrations in groundwater.
concentrations of groundwater in Cambodia, Laos, and Thailand. As shown in Table 2, NSE values of ANN in training and validation step are respectively greater than 0.5 and approximately 0.5. In this case, it is clear that the MAE values for ANN are much less than those for MLR and PCR. Consequently, it can be seen that the optimized ANN model can be a useful tool in predicting groundwater As concentrations using on-site measurement data. This result is in good agreement with a previous study by Purkait et al. (2008), in which ANN showed better results than either a linear model.
3.3.3.
3.4. 3.3.
Nonlinear models for predicting As concentrations
3.3.1.
Optimization processes for ANN and PC-ANN models
The optimal momentum rate, the number of hidden nodes, and the learning rate were obtained by the pattern search algorithm. The pattern search was used to determine the optimal parameter set from the ranges of three parameters; hidden nodes were ranged from 5 to 20, and learning and momentum rates were ranged from 0.01 to 0.6. Finally, the pattern search provided the optimal parameter set of the learning and momentum rates and the number of hidden layers, resulting in a minimum objective function. Moreover, while too few hidden nodes in ANN results in a poor predictive power, too many hidden nodes may cause a large computational time and over-fitting. The optimized number of hidden nodes for the ANN and PC-ANN models were respectively set to 13 and 15, which is consistent with past works on ANN because they range between 2 and 3 times the number of input nodes (Brion and Lingireddy, 1999). Note that the “tansigmoid” transfer function was used in the neurons of the hidden and output layers. After the determining the learning rate, the momentum rate, and the number of hidden nodes in the ANN and PC-ANN models, the errors for the prediction of As concentrations were computed in terms of MAE and NSE, and then compared to those of MLR and PCR.
3.3.2.
ANN for predicting As concentrations
Fig. 3(C) and (D) compares the observed and the predicted As concentrations of groundwater in both the training and validation steps of ANN, where the horizontal and vertical axes respectively indicate the observed and the predicted As concentrations in groundwater. Overall, it can be seen that the ANN model well reproduced the variations of the observed As
PC-ANN for predicting As concentrations
Fig. 3(C) and (D) illustrates the observed and predicted groundwater As concentrations in both the training and validation steps of PC-ANN, where the open circles and squares indicate values from PC-ANN. The figure shows that the PC-ANN model also well reproduced the variations of observed groundwater As concentrations in Cambodia, Laos, and Thailand. As shown in Table 2, whereas the NSE value for PC-ANN in the training step was 0.61, the NSE in the validation step was 0.66; i.e., the NSE of PC-ANN was greater than that of ANN in the validation step. Consequently, it demonstrates that the PC-ANN model can be useful in predicting groundwater As concentrations using on-site measurement data.
Comparison between MLR, PCR, ANN, and PC-ANN
As shown in Table 2, the prediction accuracies of ANN and PCANN are better than those of MLR and PCR models in both the generation/training and validation steps. The table also demonstrates that the developed ANN and PC-ANN models show acceptable accuracies for predicting As concentrations in the groundwater in Cambodia, Laos, and Thailand, which can be explained by the fact that the nonlinearity of ANN can reproduce the vigorous variations of As concentrations. As mentioned above, the MLR and PCR models developed in this study are not deemed to be suitable models because linearity is not sufficient for explaining the dynamic variations of observed As concentrations. This study demonstrates, however, that the ANN model is a more acceptable approach than the MLR model in terms of modeling As concentrations using data from on-site measurements. In particular, it can be posited that PC-ANN is the best model because of its highest NSE in the validation step.
3.5.
Redox potential vs As concentration
Fig. 4 shows the As concentration versus Eh. Based on the Eh, the dataset was divided into two groups, Group A (<0.00) and Group B (0.00). The 89.23% of data for Group A are samples collected from Kandal Province in Cambodia and the 84.52% of data for Group B were measured from Lao PDR and Thailand. Kandal Province is located in the Mekong Delta, which receives a substantial volume of sediment (160 millon t yr1) from the Mekong River (Meybeck and Carbonnel, 1975; Ta et al., 2001); the delta is mainly composed of young alluvial soil (Nguyen et al., 2000). In addition, a reduced state of As (III) was found to be the dominant species in Kandal Province (Polya et al., 2003; Rowland et al., 2008; Polizzotto et al., 2008). Consequently, Group A in the figure can be characterized as
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
Fig. 4 e As concentration versus redox potential, showing Groups A and B.
5541
predominantly As in highly reducing aquifer regions having a low Eh level. In the figure, the As concentrations are mostly stable in the elevated level (Group A), but show a steep negative trend with increasing Eh in Group B (i.e., Lao PDR and Thailand). Hydrological conditions in Lao PDR, Cambodia, and Vietnam have many characteristics in common and As (III) was also found to be a dominant species in Lao PDR. Here, Group B can also be interpreted as As in a reducing region, though it showed a negative correlation with Eh as opposed to Group A. This correlation may result from a combination of low iron (ferrous) concentrations and high Eh in Lao PDR (Chanpiwat et al., 2011). This different type of dependence of As on Eh found may result in difficulty in using the entire dataset for a single modeling approach. Therefore, Groups A and B were utilized to train ANN and PC-ANN models, which showed superior accuracy than linear models, in an attempt to improve the prediction performance. Fig. 5 compares the observed and predicted values of As concentrations for the two datasets (Groups A and B). The two models show a better prediction for Group A, but relatively poor accuracies for Group B. Table 2 also demonstrates that the NSE of the models for Group A were lower than the models for Group B in the validation steps, and even worse than the
Fig. 5 e The prediction results of ANN and PC-ANN for Groups A and B, (A) ANN model for Group A, (B) ANN model for Group B, (C) PC-ANN model for Group A, and (D) PC-ANN model for Group B.
5542
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
Table 4 e Sensitivity rank of the conventional on-site measurement data in ANN and PC-ANN models from the LH-OAT sensitivity analysis. Group A ANN Rank Rank Rank Rank Rank
1 2 3 4 5
Group B PC-ANN
Temperature [ C] Total dissolved solids [g L1] Electrical conductivity [ms cm1] Redox potential [mV] pH [e]
pH [e] Total dissolved solids [g L1] Temperature [ C] Redox potential [mV] Electrical conductivity [ms cm1]
models trained with a whole dataset. This result may be attributed to the insignificant relationship between As concentration and the other water quality parameters, and the limited variability of the observed As concentration in Group A. In particular, the prediction for the lower As concentration in Group A is poor, resulting in a lower NSE. In contrast, the model accuracies for Group B in the training and validation steps are satisfactory in terms of NSE (Moriasi et al., 2007).
3.6.
Sensitivity analysis
The sensitivities of the model outputs (i.e., As concentrations) to water quality parameters were investigated in an attempt to optimize ANN and PC-ANN models using the LH-OAT method. Table 4 presents the sensitivity ranks for the five parameters of ANN and PC-ANN for Groups A and B. For Group A, the most significant parameters of ANN and PC-ANN are temperature and pH, respectively. It seems to be counterintuitive because pH, TDS, and Eh were identified as significant parameters in previous studies (Buschmann et al., 2007; Amini et al., 2008; Chanpiwat et al., 2011). Conversely, pH is the most sensitive parameter for PC-ANN, followed by TDS and temperature [ C]; this is an acceptable result because neutral to high pH conditions favor As release by promoting desorption processes compared to the predominantly acidic (Buschmann et al., 2007). In addition, pH 7 might possibly enhance the mobilization of As, as explained by Buschmann et al. (2007). The sensitivity results for Group B indicate that Eh is the most sensitive parameter for PC-ANN. This coincides with a previous study on As in Laos (Chanpiwat et al., 2011).
Table 5 e Comparison results between the observed and predicted mean As concentrations (PC-ANN) for each province. Countries
Province
Observed As [mg L1]
Predicted As [mg L1]
Thailand Cambodia
Suphan Buri Kandal Phnompen Vientiane Borikhamxay Savannakhet Saravance Champasack Attapeu
70.0 440.0 136.4 14.4 14.0 24.0 18.8 40.0 11.2
70.4 325.6 127.6 24.4 30.0 6.4 13.6 25.6 31.6
Lao PDR
ANN
PC-ANN 1
Total dissolved solids [g L ] Redox potential [mV] Temperature [ C] pH [e] Electrical conductivity [ms cm1]
Redox potential [mV] Electrical conductivity [ms cm1] pH [e] Temperature [ C] Total dissolved solids [g L1]
Also, it is clearly seen that As is negatively correlated with Eh, as shown in Fig. 4. Consequently, even though the predictive power of ANN is sufficient for following variations of the observed As concentration, the role of temperature is theoretically different from that in literaturedindeed, it may be caused by a collinearity problem among the input variables. Conversely, the roles of pH and Eh in PC-ANN coincide with preliminary studies, and thereby imply that PC-ANN is the superior model for predicting the As concentrations of groundwater in Cambodia, Laos, and Thailand. Table 5 compares the averaged observed concentrations and the As concentrations predicted by PC-ANN for each province. In general, the model performances for Cambodia showing a higher As concentration are more accurate than those for the others countries. This may be caused by the training process, which tends to follow a higher observed As, though may not be as useful for lower As concentrations. In other words, the model developed in this study would be more informative for a high-risk area, such as Kandal Province in Cambodia.
4.
Conclusions
Groundwater resources are one of the most important components of drinking water supplies, especially in rural areas of Southeast Asian countries including Cambodia, Laos, and Thailand. Over the years, several researchers have attempted to explore the levels of As contamination from various viewpoints. However, few studies have focused on a statistical modeling of As that involved other conventional water quality parameters. As such, the main conclusions drawn from this study are as follows: 1) The poor accuracies of MLR and PCR indicated that linear models for conventional water quality parameters cannot reproduce dynamic variations of observed groundwater As concentrations. 2) The prediction accuracies of ANN and PC-ANN were better than those of MLR and PCR models in both the generation/ training and validation steps, showing acceptable accuracies for predicting As concentrations. 3) The results of the sensitivity analysis demonstrated that the predictive power of ANN was satisfactory to follow variations of the observed As, but the roles of the input parameters are theoretically different from those in literature, which might be caused by a collinearity problem among the input variables.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
4) Conversely, the roles of pH and Eh in PC-ANN coincided with preliminary studies, and thereby imply that PC-ANN is the superior model for predicting As concentrations of groundwater in Cambodia, Laos, and Thailand. We expect the PC-ANN model developed in this study to be valuable to not only environmental scientists when designing an efficient As monitoring and removal plan, but also for policy makers and NGOs for establishing effective public health management policies. In particular, because the As testing in a laboratory is a complicated and costly process, the model can be better applied to establish an effective As monitoring and public health management in developing countries. The prediction of As, however, is still a challenging work, especially for developing a reliable model. In this study, the PC-ANN model developed for Cambodia, Laos, and Thailand will be more robust and reliable once new datasets are obtained from other countries. As clearly shown in this study, As contamination could not be explained by a linear combination of conventional water quality parameters. Therefore, this model still needs to be validated by using new datasets in order to more precisely investigate the relationship between As contamination and conventional water quality parameters.
Acknowledgments This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology, Korea (No. 2010-0011822). We would also like to acknowledge the assistance of the Ministry of Land, Transport and Maritime Affairs (MLTMA), and the Korea Meteorological Administration (KMA).
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.010.
references
American Water Works Association (AWWA), 2001. Arsenic Rule, vol. 45. Mainstream. Amini, M., Mueller, K., Abbaspour, K.C., Rosenberg, T., Afyuni, M., Møller, K.N., Sarr, M., Johnson, C.A., 2008. Statistical modeling of global geogenic fluoride contamination in groundwaters. Environmental Science and Technology 42 (10), 3662e3668. Bagla, P., Kaise, J., 1996. India’s spreading health crisis draws global arsenic experts. Science 274, 174e175. Berg, M., Tran, H.C., Nguyen, T.C., Pham, H.V., Schertenleib, R., Giger, W., 2001a. Arsenic contamination of groundwater and drinking water in Vietnam: a human health threat. Environmental Science and Technology 35 (13), 2621e2626. Berg, M., Tran, H.C., Nguyen, T.C., Pham, H.V., Schertenleib, R., Giger, W., 2001b. Magnitude of arsenic pollution in the Mekong and Red River DeltasdCambodia and Vietnam. Science of the Total Environment 372 (2e3), 413e425.
5543
Berg, M., Luzi, S., Kim, P.T., Viet, P.H., Giger, W., Stuben, D., 2006. Arsenic removal from groundwater by household sand filters: comparative field study, model calculations, and health benefits. Environmental Science and Technology 40 (17), 5567e5573. Berg, M., Stengel, C., Trang, P.T.K., Hung Viet, P., Sampson, M.L., Leng, M., Samreth, S., Fredericks, D., 2007. Magnitude of arsenic pollution in the Mekong and Red River DeltasdCambodia and Vietnam. Science of the Total Environment 372 (2e3), 413e425. Bowerman, B.L., O’Connell, R.T., 1990. Linear Statistical Models: An Applied Approach, second ed. Duxbury Press, Belmont, CA. Brion, G.M., Lingireddy, S., 1999. A neural network approach to identifying non-point sources of microbial contamination. Water Research 33 (14), 3099e3106. Buschmann, J., Berg, M., Stengel, C., Sampson, M.L., 2007. Arsenic and manganese contamination of drinking water resources in Cambodia: coincidence of risk areas with low relief topography. Environmental Science and Technology 41 (7), 2146e2152. Chang, F.J., Kao, L.S., Kuo, Y.M., Liu, C.W., 2010. Artificial neural networks for estimating regional arsenic concentrations in a blackfoot disease area in Taiwan. Journal of Hydrology 388 (1e2), 65e76. Chanpiwat, P., Sthiannopkao, S., Cho, K.H., Kim, K.W., San, V., Suvanthong, B., Vongthavady, C., 2011. Contamination by arsenic and other trace elements of tube-well water along the Mekong River in Lao RDR. Environmental Pollution 159 (2), 567e576. Chiew, H., Sampson, M.L., Huch, S., Ken, S., Bostick, B.C., 2009. Effect of groundwater iron and phosphate on the efficacy of arsenic removal by iron-amended biosand filters. Environmental Science and Technology 43 (16), 6295e6300. Cho, K.H., Kang, J.-H., Ki, S.J., Park, Y., Cha, S.M., Kim, J.H., 2009. Determination of the optimal parameters in regression models for the prediction of chlorophyll-a: a case study of the Yeongsan Reservoir, Korea. Science of the Total Environment 407, 2536e2545. Gros, P., 1997. Practical linear regression modelling. Basic principles. Validation methods. Oceanis 23 (3), 359e515. Holmberg, M., Forsius, M., Starr, M., Huttunen, M., 2006. An application of artificial neural networks to carbon, nitrogen and phosphorus concentrations in three boreal streams and impacts of climate change. Ecological Modelling 195 (1e2), 51e60. Jolliffe, I.T., 2002. Principal Component Analysis, second ed. Springer, New York. Kocar, B.D., Fendorf, C., 2009. Thermodynamic constraints on reductive reactions influencing the biogeochemical of Arsenic in soils and sediments. Environmental Science and Technology 43 (13), 4871e4877. Kohnhorst, A., 2005. Arsenic in groundwater in selected countries in South and Southeast Asia: a review. The Journal of Tropical Medicine and Parasitology 28 (2), 73e82. Kuo, J.T., Hsieh, P.H., Jou, W.S., 2008. Lake eutrophication management modeling using dynamic programming. Journal of Environmental Management 88 (4), 677e687. Lee, J.H., Yu, M.J., Bang, K.W., Choe, J.S., 2003. Evaluation of the methods for first flush analysis in urban watersheds. Water Science and Technology 48 (10), 167e176. Luan, F., Liu, H.T., Wen, Y., Zhang, X., 2008. Prediction of quantitative calibration factors of some organic compounds in gas chromatography. Analyst 133 (7), 881e887. Mac Nally, R., 2002. Multiple regression and inference in ecology and conservation biology: further comments on identifying important predictor variables. Biodiversity and Conservation 11 (8), 1397e1401.
5544
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 3 5 e5 5 4 4
Maier, H.R., Dandy, G.C., 1996. The use of artificial neural networks for the prediction of water quality parameters. Water Resources Research 32 (4), 1013e1022. Meybeck, M., Carbonnel, J.P., 1975. Chemical transport by Mekong river. Nature 255, 134e136. Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L., 2007. Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE 50 (3), 885e900. Myers, R., 1990. Classical and Modern Regression with Applications, second ed. Duxbury Press, Boston, MA. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part Ida discussion of principles. Journal of Hydrology 10 (3), 282e290. Nguyen, V.L., Ta, T.K.O., Tateishi, M., 2000. Late Holocene depositional environments and coastal evolution of the Mekong River delta, southern Vietnam. Journal of Asian Earth Sciences 18 (4), 427e439. Polizzotto, M.L., Kocar, B.D., Benner, S.G., Sampson, M., Fendorf, S., 2008. Near-surface wetland sediments as a source of arsenic release to ground water in Asia. Nature 454 (7203), 505e508. Polya, D.A., Gault, A.G., Bourne, N.J., Lythgoe, P.R., Cooke, D.A., 2003. Coupled HPLC-ICP-MS analysis indicates highly hazardous concentrations of dissolved arsenic species in Cambodian groundwaters. In: Holland, J.G., Tanner, S.D. (Eds.), Plasma Source Mass Spectrometry: Applications and Emerging Technologies. Royal Society of Chemistry, Cambridge, pp. 27e140. Polya, D.A., Gault, A.G., Diebe, N., Feldman, P., Rosenboom, J.W., Gilligan, E., Fredericks, D., Milton, A.H., Sampson, M., Rowland, H.A.L., Lythgoe, P.R., Middleton, C., Cooke, D.A., 2005. Arsenic hazard in shallow Cambodian groundwaters. Mineralogical Magazine 69 (5), 807e823. Purkait, B., Kadam, S.S., Das, S.K., 2008. Application of artificial neural network model to study arsenic contamination in groundwater of Malda District, Eastern India. Journal of Environmental Informatics 12 (2), 140e149. Rawlings, J.O., Pantula, S.G., Dickey, D.A., 1998. Applied Regression Analysis: A Research Tool. Springer, New York. Riad, S., Mania, J., Bouchaou, L., Najjar, Y., 2004. Rainfall-runoff model using an artificial neural network approach. Mathematical and Computer Modelling 40 (7e8), 839e846. Rowland, H.A.L., Gault, A.G., Lythgoe, P., Polya, D.A., 2008. Geochemistry of aquifer sediments and arsenic-rich
groundwaters from Kandal Province, Cambodia. Applied Geochemistry 23 (11), 3029e3046. Sarangi, A., Bhattacharya, A.K., 2005. Comparison of artificial neural network and regression models for sediment loss prediction from Banha watershed in India. Agricultural Water Management 78 (3), 195e208. Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behaviour and distribution of arsenic in natural waters. Applied Geochemistry 17 (5), 517e568. Sousa, S.I.V., Martins, F.G., Alvim-Ferraz, M.C.M., Pereira, M.C., 2007. Multiple linear regression and artificial neural networks based on principal components to predict ozone concentrations. Environmental Modelling and Software 22 (1), 97e103. Stanger, G., Truong, T.V., My Ngoc, K.S.L.T., Luyen, T.V., Tuyen, T.T., 2005. Arsenic in groundwaters of the lower Mekong. Environmental Geochemistry and Health 27 (4), 341e357. Sthiannopkao, S., Kim, K.W., Cho, K.H., Wantala, K., Sotham, S., Sokuntheara, C., Kim, J.H., 2010. Arsenic levels in human hair, Kandal Province, Cambodia: the influences of groundwater arsenic, consumption period, age and gender. Applied Geochemistry 25 (1), 81e90. Sun, G., Liu, J., Luong, T.V., Sun, D., Wang, L., 2002. Endemic Arsenicosis: A Clinical Diagnostic with Photo Illustrations. UNICEF East Asia and Pacific Regional Office, Bangkok, Thailand, ISBN 974-685-040-7. Ta, T.K.O., Nguyen, V., Tateishi, M., Kobayashi, I., Saito, Y., 2001. Sedimentary facies, diatom and foraminifer assemblages in a late PleistoceneeHolocene incised-valley sequence from the Mekong River Delta, Bentre Province, southern Vietnam: the BT2 core. Journal of Asian Earth Sciences 20 (1), 83e94. Tayfur, G., Swiatek, D., Wita, A., Singh, V.P., 2005. Case study: finite element method and artificial neural network models for flow through Jeziorsko Earthfill dam in Poland. Journal of Hydraulic Engineering 131 (6), 431e440. Tetsuro, A., Takashi, K., Junko, F., Reiji, K., Tu, B.M., Pham, T.K.T., Hisato, I., Annamalai, S., Pham, H.V., Shinsuke, T., 2006. Contamination by arsenic and other trace elements in tubewell water and its risk assessment to humans in Hanoi, Vietnam. Environmental Pollution 139 (1), 95e106. Wen, C.G., Lee, C.S., 1998. A neural network approach to multiobjective optimization for water quality management in a river basin. Water Resources Research 34 (3), 427e436.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Phytoaccumulation of antimicrobials from biosolids: Impacts on environmental fate and relevance to human exposure Niroj Aryal, Dawn M. Reinhold* Department of Biosystems and Agricultural Engineering, Michigan State University, 216 A.W. Farrall Agricultural Engineering Hall, East Lansing, MI 48824, USA
article info
abstract
Article history:
Triclocarban and triclosan, two antimicrobials widely used in consumer products, can
Received 5 January 2011
adversely affect ecosystems and potentially impact human health. The application of
Received in revised form
biosolids to agricultural fields introduces triclocarban and triclosan to soil and water
7 August 2011
resources. This research examined the phytoaccumulation of antimicrobials, effects of
Accepted 9 August 2011
plant growth on migration of antimicrobials to water resources, and relevance of phy-
Available online 23 August 2011
toaccumulation in human exposure to antimicrobials. Pumpkin, zucchini, and switch grass were grown in soil columns to which biosolids were applied. Leachate from soil columns
Keywords:
was assessed every other week for triclocarban and triclosan. At the end of the trial,
Triclosan
concentrations of triclocarban and triclosan were determined for soil, roots, stems, and
Triclocarban
leaves. Results indicated that plants can reduce leaching of antimicrobials to water
Phytoremediation
resources. Pumpkin and zucchini growth significantly reduced soil concentrations of tri-
Plant uptake
closan to less than 0.001 mg/kg, while zucchini significantly reduced soil concentrations of triclocarban to 0.04 mg/kg. Pumpkin, zucchini, and switch grass accumulated triclocarban and triclosan in mg per kg (dry) concentrations. Potential human exposure to triclocarban from consumption of pumpkin or zucchini was substantially less than exposure from product use, but was greater than exposure from drinking water consumption. Consequently, research indicated that pumpkin and zucchini may beneficially impact the fate of antimicrobials in agricultural fields, while presenting minimal acute risk to human health. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Antimicrobials, specifically triclocarban (3,4,40 -trichlorocarbanilide) and triclosan (5-chloro-2-(2,4-dichlorophenoxy)phenol), are widely used in personal care products (Ying et al., 2007). Triclocarban and triclosan primarily enter the environment through domestic sewage discharge to wastewater treatment plants, where removal is predominantly due to sorption (78 11% for triclocarban and 80 22% for triclosan) to wastewater particulate matter (Chu and Metcalfe, 2007; Heidler and Halden, 2007; Heidler et al., 2006; Sapkota et al., 2007). Consequently, digested municipal sludge accumulates
51 15 mg triclocarban and 30 11 mg triclosan per g dry sludge (Heidler and Halden, 2007; Heidler et al., 2006). Greater than 97% of triclocarban and triclosan in sewage is discharged to water resources and biosolids, leading to a 0.6 to 1 million kg/ year combined input into U.S. environment (EPA, 2003; Halden and Paull, 2005). The greatest discharge of triclocarban and triclosan into the environment is through municipal application of biosolids to fields, as more than 50% of biosolids are land applied (Heidler et al., 2006). Triclocarban is generally detected in U.S. surface waters at concentrations from 10 to 1550 ng/L (Halden and Paull, 2005; Sapkota et al., 2007), whereas triclosan is detected in U.S.
* Corresponding author. Tel.: þ1 517 432 7732; fax: þ1 517 432 2892. E-mail address: [email protected] (D.M. Reinhold). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.027
5546
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
rivers and streams from 3.0 to 75 ng/L (Bester, 2003; Ying and Kookana, 2007). Additionally, agricultural soils previously amended with biosolids can accumulate 1.2e65 ng/kg triclocarban and 0.16e1.0 ng/kg triclosan (Cha and Cupples, 2009). Once introduced into the environment, triclocarban and triclosan sorb to soils or sediments and are not predicted to readily degrade (Halden and Paull, 2005; Ying et al., 2007). Experimental half-lives of triclocarban and triclosan range from 87 to 231 and 18e58 days respectively in aerobic soil, with longer half-lives in anaerobic soils (Wu et al., 2009; Ying et al., 2007). Triclocarban and triclosan are lowly soluble in water with solubilities of 45 mg/L for triclocarban (Snyder et al., 2010) and 462 mg/L for triclosan (EPI Suite 4.0). Triclocarban and triclosan have high octanolewater partitioning coefficients (KOW) with log KOW of 4.90 for triclocarban and log KOW of 4.76 for triclosan (EPI Suite 4.0). Henry Law constants of 4.52 1011 atm-m3/mole for triclocarban and 2.13 108 atmm3/mole for triclosan (EPI Suite 4.0) indicate that they are nonvolatile. Release of antimicrobials into the environment may result in bioaccumulation of antimicrobials in aquatic organisms and humans. For example, up to 58 mg/kg triclosan and 299 mg/kg triclocarban was measured in snails near wastewater treatment plant (WWTP) effluent (Coogan and La Point, 2008). Furthermore, concentrations of 2.4e3790 mg/L triclosan were measured in 74.6% of urine samples collected from the U.S. general population (Calafat et al., 2008). Triclosan, at the concentration of 0.01e19 mg/kg, was also detected in plasma of Swedish women who did not use personal care products containing triclosan, indicating unintentional systemic exposure to antimicrobials through sources other than personal care products (Allmyr et al., 2006). Antimicrobials have the potential to adversely impact human and ecosystem health. In humans, disruption of endocrine activity by triclosan and triclocarban is expected at concentrations of 29e3150 mg/L (Ahn et al., 2008). At much lower concentrations, triclocarban and triclosan disrupt critical ecological processes. In rivers, antimicrobials adversely affect biofilm structure and function (Lawrence et al., 2009). Freshwater microbial communities are sensitive to 2.9 mg/L triclosan (Johnson et al., 2009) and concentrations as low as 150 ng/L triclosan can have physiological effects on thyroid hormone, body weight, and hind limb development in frogs (Fraker and Smith, 2004; Veldhoen et al., 2007). Consequently, current environmental concentrations of antimicrobials have the potential to disrupt aquatic ecosystems. In terrestrial ecosystems, triclocarban and triclosan can inhibit soil respiration, nutrient recycling, and plant growth (Liu et al., 2009). Development of microbial and drug resistance has also been reported (Heath et al., 1998; Walsh et al., 2003). Studies have demonstrated that plants such as pumpkin and zucchini have potential to accumulate hydrophobic chlorinated organic pollutants. In field experiments of polychlorinated biphenyls (Aroclor 1254/1260), Cucurbita pepo ssp pepo cv. Howden (pumpkin) plants took up, translocated, and accumulated 7.6 mg/kg PCBs in plant shoots (Aslund et al., 2007). Studies have also demonstrated that C. pepo species (pumpkin and zucchini) can extract and translocate dichlorodiphenyltrichloroethane (DDT) and its metabolites (White et al., 2003) and polychlorinated dibenzo-p-dioxins and
dibenzofurans (Hu¨lster et al., 1994). Pumpkin and zucchini showed more potential for phytoaccumulation than other crops. For example, pumpkin and zucchini translocated 1.8e36 times more 2,4,8-trichlorodibenzo-p-dioxin and 2e4.3 times more 1,3,6,8-tetrachlorodibenzo-p-dioxin than 10 other food crops (Zhang et al., 2009). Additionally, the capability of switch grass (Panicum variegatum L.) to reduce PCB concentration in soils through unidentified mechanism has been demonstrated (Dzantor et al., 2000). Recent studies have also detected triclocarban and triclosan in shoot tissues and beans of soybean plants that were planted in biosolids-amended soils (Wu et al., 2010). The study presented herein evaluates the hypothesis that triclocarban and triclosan, being chlorinated aromatic organic pollutants similar to DDT and PCBs, can be taken up by plants to reduce antimicrobial concentrations in soil and the migration of antimicrobials to water resources. Understanding the fate of triclocarban and triclosan in vegetated fields to which biosolids have been applied is crucial to understanding the entrance of antimicrobials into water resources and quantifying the human health effects of consuming food grown on land to which biosolids are applied. The aims of this study were to: (i) assess the effects of plant growth on leaching of antimicrobials from land applied biosolids, (ii) evaluate accumulation of antimicrobials in pumpkin, zucchini and switch grass, and (iii) preliminarily evaluate relevance of triclocarban and triclosan phytoaccumulation to human health.
2.
Materials and methods
2.1.
Soil, seeds and biosolids
The plant varieties used for this study were pumpkin (C. pepo cultivar Howden), zucchini (C. pepo cultivar Gold Rush), and switch grass (P. variegatum). The specific varieties of C. pepo were chosen based on their reported accumulation potential for hydrophobic organic pollutants (Aslund et al., 2008; Lunney et al., 2004; Wang et al., 2004; White et al., 2003). Switch grass was selected as a non-vegetable plant that has potential to stimulate rhizosphere microbial degradation of hydrophobic aromatic pollutants like PCBs (Dzantor et al., 2000). Pumpkin and zucchini seeds were obtained from Johnny Seeds, Maine, whereas switch grass seedlings were obtained from the Department of Crop and Soil Sciences, MSU. Soil for the study, a screened sandy clay loam, was obtained from East Lansing, MI. Biosolids were collected from a nearby wastewater treatment plant and were analyzed for triclocarban and triclosan prior to use by previously developed methods (Cha and Cupples, 2009) using pressurized solvent extraction and tandem mass spectrometry. Triclocarban and triclosan were present at concentrations of 8.18 0.56 mg/kg and 0.18 0.01 mg/kg dry mass of biosolids, respectively.
2.2.
Chemicals
Triclocarban [CAS 101-20-2] (>98%) was obtained from Tokyo Chemical Industry and triclosan [CAS 3380-34-5] from
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
Calbiochem. Mobile phase and extraction solvents were obtained from VWR, Inc. LCeMS solvents were of MS grade.
2.3.
Experimental columns
Plants were grown in experimental columns with diameters of 14.7 cm and lengths of 30 cm. The bottom 7.6 cm of the soil column was filled with soil without mixing biosolids. The next 15.2 cm of soil was thoroughly mixed with biosolids at the application rate of 0.73 dry Mg per 1000 m2 (Cha and Cupples, 2009). Solid content of biosolids was 4.8% (dry basis) and 200 g of wet biosolids were applied before seed sowing, with an additional 60 g of wet biosolids applied after 8 weeks to simulate a second field application. Experimental design included quadruple columns for pumpkin and zucchini and triplicate columns for switch grass and no plant controls. Seeds were sown at the rate of a seed per column except for switch grass, for which plants were transplanted directly. Plants were maintained at a constant temperature (23 2 C) using a light regime of 16 h light: 8 h dark.
2.4.
Sampling
For leachate sampling, columns were flooded with equal volumes of water so that a minimum of 100 mL water leached from each column. Leachate samples were collected in amber bottles under each soil column once every two weeks. Samples were stored immediately at 4 C until prepared for analysis. Plants were harvested at the end of 22 weeks. Plant tissues were separated into roots, leaves, and stems for pumpkin and zucchini and into roots and shoots for switch grass. Plant tissues were rinsed carefully to remove soil and dust particles, air-dried at room temperature (23 2 C), weighed and stored in amber bottles in the refrigerator until sample extraction. Soil samples from each column, at depths of 5 cm, 10 cm, 15 cm, and 20 cm, were collected in amber bottles, screened to remove plant roots, homogenized by mixing, and stored at 8 C until sample preparation.
2.5.
Sample preparation
Aqueous samples were prepared as published previously (Halden and Paull, 2004, 2005). Samples were passed through a solid phase extraction (SPE) cartridge (Oasis HLB 3 cc, Waters Corporation) and eluted with 4 ml of 50% methanol and 50% acetone containing 10 mM acetic acid. Elutes were dried under nitrogen, reconstituted in 1 ml of 50% methanol and 50% acetone, filtered through a 0.2 mm PTFE membrane, and analyzed by liquid chromatography mass spectrometry in negative electrospray ionization mode (LCeMS/ESI(e)). Soil samples were prepared and extracted as previously published (Cha and Cupples, 2009) by pressurized liquid extraction (PLE), using a Dionex ASE 200 accelerated solvent extractor. Triplicate subsamples of soil (5 dry g each) from each column sample were extracted. A fourth subsample was weighed and dried for at least 24 h at 105 C and again weighed for moisture determination. Extraction on the ASE utilized acetone with oven temperature of 100 C, extraction pressure of 1500 psi, static time of 5 min, and flush volume of 100%. The
5547
extracts were evaporated to dryness under nitrogen, reconstituted in 50% methanol and 50% acetone, spiked with 6 ppm triclosan and triclocarban, and filtered through a 0.2 mm filter before analysis by LCeMS. For quality assurance, all analytical runs included a blank, a control, and a sample triplicate. The recovery from the methods was 94.03 9.76% for triclocarban and 83.39 19.48% for triclosan. For plant samples, frozen plant tissues were oven dried, grounded in mortar and pestle, and extracted using the same method as for soil described above. All analytical runs included a blank, a control, and a sample triplicate except for roots, where sufficient biomass for replicate analysis was not available.
2.6.
LC/MS analysis
A Shimadzu LCeMS 2010 EV was used to analyze samples for triclocarban and triclosan. Samples were separated using Allure biphenyl column (5 mm, 150 2.1 mm) from Restek Cor. using a binary gradient of 75% methanol and 25% 5-mM ammonium acetate to 100% methanol. MS parameters were: curved desolvation line (CDL) of 1.5 V, block and CDL temperature of 30 C and nitrogen desolvation gas flow rate of 1.5 L/min. MS negative electrospray ionization mode with scan mode was used for method development and identification, whereas selected ion monitoring (SIM) mode was used for quantification. Retention time (tR 0.1 min), detection of characteristic molecular ions (m/z 313 for triclocarban and m/z 287 for triclosan), and detection of reference ions (m/z 315 and 317 for triclocarban and m/z 289 and 291 for triclosan) were used to identify the target molecules (Halden and Paull, 2005). Quantification was performed using external, linear calibration and a minimum of six calibration levels. The limits of detection were 10e100 ng/L for water and 0.1e1.0 ng/kg for soils and plants.
2.7.
Statistical analysis
All statistical analysis was performed in Sigma Plot (v 11.0). A one tailed t-test was used for all pair-wise comparisons and one-way ANOVA was used for all other comparisons. The reported values are in mean standard error.
2.8. Relevance of plant accumulation of antimicrobials to human health Accumulation of antimicrobials by pumpkin and zucchini represents the potential for direct exposure of humans to antimicrobial through ingestion of vegetables. To quantify the potential impacts of accumulation of antimicrobials by pumpkin and zucchini on human health, doses of exposure for different routes were compared. The resulting dose of triclocarban and triclosan from consumption of pumpkin and zucchini exposed to biosolids was predicted by assuming fruit concentrations equal to the range of observed stems and leaves concentrations. This assumption is conservative in that it likely over predicts actual risk because: (i) leaves had lower concentrations of antimicrobials than that of stems in this study and another similar study with soybeans (Wu et al., 2010) and (ii) decreases in concentrations of PCB and DDE were observed in the stem of pumpkin and zucchini as the distance
5548
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
from root increased (Aslund et al., 2007; White et al., 2003). Predicted doses were compared to doses from other routes of exposure: product use (EPA, 2002), drinking water exposure (EPA, 2002) and exposure from eating soybean produced from fields that receive biosolids (Wu et al., 2010). The dose calculations were based on 5 g/day average soybean (Reinwald et al., 2010) and 11.5 g/day average pumpkin and zucchini consumption per capita in USA (USDA, 2011).
3.
Results and discussions
3.1.
Leaching of antimicrobials
Both triclocarban and triclosan were detected in leachate from the experimental soil columns. As shown in Fig. 1, the concentrations increased initially for two weeks and then decreased. The lag in peak concentration was likely due to sorption of antimicrobials to bottom 7.6 cm of soil where biosolids were not applied. In saturated soil systems, sorption and biodegradation were primary removal mechanisms for triclocarban (Drewes et al., 2003; Essandoh et al., 2010). The maximum triclosan concentration in leached water from pumpkin, zucchini, switch grass and control were 530 180 mg/L, 1670 540 mg/L, 460 280 mg/L, 710 450 mg/mL respectively, all observed in second week. Triclocarban concentrations also followed a similar trend with maximum concentrations of 210 160 mg/L, 190 110 mg/L, 120 40 mg/L, and 370 360 mg/L for pumpkin, zucchini and switch grass and
Fig. 1 e Triclocarban and triclosan concentrations in leached water (mg/ml) with time (weeks). Points represent mean and error bars represent standard error. Volume of water leached each week was statistically similar between columns.
control, respectively. An initial peak with maximum concentration of 110 81 mg/L triclosan and 3.4 2.2 mg/L triclocarban, followed by a rapid decrease in concentrations, has previously been observed in runoff from agricultural field after application of biosolids (Sabourin et al., 2009). Higher concentrations in this experiment were likely due to the relatively short columns. The second addition of biosolids increased antimicrobial concentrations immediately. For triclocarban, the observed second peak was not statistically different from the first peak (P ¼ 0.6). In contrast, triclosan was observed at lower concentrations after the second biosolids application (P ¼ 0.03). Triclosan, despite being present at a lower initial concentration in the applied biosolids, was collected at higher concentrations than triclocarban in leachate during the first four weeks. The higher concentrations of triclosan in the leachate were attributed to its higher solubility and decreased affinity for sorption when compared to triclocarban. The aqueous solubility of triclosan is approximately 10 times that of triclocarban. Sorption of triclocarban to soils is also much stronger than sorption of triclosan, with sorption coefficients (Kd) of 1029 L/kg and 231 L/kg (respectively) for sandy loam soils (Wu et al., 2009). Despite triclosan and triclocarban being weak acids with acid dissociation constants (Ka) of 107.9 and 1012.7 (respectively), the role of soil pH, as affected by addition of biosolids, was not expected to greatly impact sorption of triclosan or triclocarban. While sorption of the anionic form of triclosan is less than the sorption of the neutral form of triclosan, sorption of the anionic form of triclosan is still considerable, resulting in only a slight decrease in net triclosan sorption in soils when the pH increased from 4 to 8 (Wu et al., 2009). Triclocarban sorption was almost unaltered when pH of soil was increased from 4 to 8 (Wu et al., 2009). As measurements indicated the soil pH after addition of biosolids to the experimental columns were between 7 and 8, the increase in pH from the addition of biosolids to the soil columns was expected to only minimally decrease the sorption of triclosan to the soil and was not expected to affect triclocarban sorption to the soil. In contrast to the first four weeks, more triclocarban leached from the columns than triclosan over the remaining 18 weeks. Reduction in the relative leaching of triclosan was most likely due to increased microbial degradation of triclosan. In aerobic soils, degradation of triclosan was characterized by a half-life of 18e58 days, while degradation of triclocarban exhibited a half-life of 87e231 days (Wu et al., 2009; Ying et al., 2007). Additionally, the initial concentrations of triclocarban in the biosolids were much higher than were the initial concentrations of triclosan. The total masses of antimicrobials leached from the vegetated soil columns over 22 weeks were not significantly different from the total masses leached from the control columns. However, the highest concentrations of antimicrobials were leached in week 2, prior to full establishment of the plants. When only considering the total antimicrobials leached after the second addition of biosolids, there were significant differences between the masses of antimicrobials leached from the control and vegetated columns, with P values of <0.01 for pumpkin, 0.01 for zucchini and 0.01 for switch grass. These results suggest that established plants can
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
5549
reduce leaching of antimicrobials, prompting additional analysis of the fate of antimicrobials in vegetated soils.
3.2.
Soil concentration of antimicrobials
Soil concentrations of antimicrobials at the end of 22 weeks are summarized in Table 1. Soil concentrations of triclocarban in zucchini columns and of triclosan in zucchini and pumpkin columns were substantially lower than were soil concentrations in control columns (0.02 < P < 0.05). However, soil concentrations of triclocarban in pumpkin columns were similar to those in control columns (P ¼ 0.09). Soil concentrations of triclosan and triclocarban in switch grass columns were similar to or greater than soil concentrations in control columns. The observed difference in final soil concentration of antimicrobials between pumpkin, zucchini, and switch grass columns may have resulted from difference in plant growth in the column. Shoot mass of switch grass plants was 0.49 0.24 g as opposed to 3.74 0.93 g for pumpkin and 5.39 0.28 g for zucchini. Root masses were 0.91 0.40 g for switch grass, 0.08 0.03 g for pumpkin and 0.10 0.03 g for zucchini. Development of an extensive root mass in the absence of robust above ground growth did not appear to influence soil concentration of triclosan in switch grass columns. However, results indicated pumpkin and zucchini may decrease soil concentrations of triclocarban and/or triclosan after land application of biosolids.
3.3.
Concentrations of antimicrobials in plant tissues
Triclocarban and triclosan were detected in roots, stems, and leaves of pumpkin, zucchini, and switch grass. Antimicrobials concentrations in plant tissues (Figs. 2 and 3) ranged from 1.10 mg/kg in leaves to 39.5 mg/kg in roots. Root concentrations of antimicrobials were generally higher than concentrations in stems and leaves, with the exception of triclocarban concentrations in pumpkin tissues. However, plant tissue concentrations were highly variable and higher root concentrations were usually not statistically significant compared to shoot concentrations. No consistent trends in change in concentration of antimicrobials in stem and leaf tissues in pumpkin and zucchini were observed. A general decrease in concentration from root to stem to leaves to fruits has been previously observed for pumpkin and zucchini
Fig. 2 e Plant concentrations of triclocarban. Points represent mean and error bars represent standard error.
accumulation of PCBs (Aslund et al., 2007) and soybean accumulation of triclocarban and triclosan (Wu et al., 2010). A similar decrease from stem to leaf was observed for the accumulation of antimicrobials by pumpkin; however, the decrease was only significant for triclosan. In contrast, concentration of antimicrobials in zucchini increased from stem to leaf, with a significant increase observed for triclocarban. Further studies with higher number of replicates could clarify the significance of the observed trends in plant concentrations of antimicrobials. The hydrophobicities of triclocarban (log KOW ¼ 4.9) and triclosan (log KOW ¼ 4.76) indicate the potential for bioaccumulation. Soil concentrations of antimicrobials were significantly less than the concentrations of antimicrobials in stems, leaves, and water for all plant species (P values from
Table 1 e Soil antimicrobial concentrations and P values for comparison with controls. ND refers to values which were detected but less than quantitation limit (<0.001 mg/kg). Triclocarban, mg/kg Pumpkin Zucchini Switch grass Control
0.055 0.003 (P ¼ 0.091) 0.038 0.004 (P ¼ 0.040) 0.241 0.026 (P ¼ 0.045) 0.097 0.012
Triclosan, mg/kg
Sum, mg/kg
ND (P ¼ 0.021)
0.055 0.003 (P ¼ 0.073) 0.038 0.004 (P ¼ 0.032) 0.166 0.014 (P ¼ 0.155) 0.105 0.013
ND (P ¼ 0.021) 0.001 0.000 (P ¼ 0.457) 0.007 0.001
Fig. 3 e Plant concentrations of triclosan. Points represent mean and error bars represent standard error.
5550
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
<0.05 except for between soil and zucchini roots). Triclocarban root bioaccumulation factors, the ratio of root concentration to soil concentrations (g/g), were 11.01 5.06 for pumpkin, 40.27 46.34 for zucchini, and 30.92 9.41 for switch grass. For triclosan, the root bioaccumulation factors (g/g) were 972 398 for pumpkin, 1822 260 for zucchini, and 874 706 for switch grass respectively. These values are comparably higher than 1.0e3.3 for DDT bioaccumulation for the same plant varieties (Lunney et al., 2004) and higher than root bioaccumulation factors of 2.2e5.8 for triclosan and 1.7e2.0 for triclocarban observed for soybean (Wu et al., 2010). Triclocarban translocation factors (the ratios of shoot concentration to root concentration) were 0.78 0.55 g/g for pumpkin, 0.27 0.19 g/g for zucchini, and 0.52 g/g for switch grass. For triclosan, translocation factors were 0.30 0.21 g/g for pumpkin, 0.16 0.05 g/g for zucchini and 0.81 g/g for switch grass. Low translocation factor imply limited transport from root to shoot. These translocation factors are comparable to those reported in Aslund et al. (2007) and Lunney et al. (2004) for accumulation of DDT, DDD, and DDE and PCBs by the same species. Wu et al. (2010) reported translocation factors of 0.01e0.28 and 0.16e1.77 for triclocarban and triclosan, respectively, in soybean plants, indicating the translocation of antimicrobials is comparable for pumpkin, zucchini, and soybean. However, as root bioaccumulation factors were greater for pumpkin and zucchini than for soybeans, results indicated that pumpkin and zucchini have greater potential to accumulate antimicrobials than soybeans. A mass balance for each column was completed (Fig. 4). A significant mass fraction of triclocarban was not accounted for in most of the columns, indicating other mechanisms of antimicrobial loss, such as microbial degradation, phytostimulation, or phytodegradation. After eight weeks, the largest portion of triclocarban remained in the soil, indicating that sorption plays an important role in fate of triclocarban in
Fig. 4 e Mass balance analyses of triclosan and triclocarban in vegetated and control columns.
vegetated soils. For zucchini, total mass accumulation in leaves was highest followed by stems and roots accumulation. In contrast, stems accumulated the greatest mass, followed by leaves and then by roots, in pumpkin. Accumulation of antimicrobials was greater in roots than in shoots for switch grass owing to its limited shoot production. With the exception of triclosan in zucchini columns (where all mass was accounted for in leachate and plant samples), the unaccounted mass of antimicrobials was greater in planted columns than in unplanted controls, supporting the observation that presence of plants may promote removal of antimicrobials by microbial degradation or other unidentified mechanisms. Analysis of coefficients of variation showed that there was minimal variation between the samples in the same column, but rather large variation between the columns. Additionally, large variability was observed in plant masses between columns. Consequently, the observed variability in the experiment was most likely due to inherent biological variability within plantbased systems, rather than variability due to sampling or procedure.
3.4. Relevance of plant accumulation of antimicrobials to human health The doses calculated for multiple routes of exposure to antimicrobials, shown in Fig. 5, are substantially less than the noobservable adverse effect level (NOAEL) of 25 mg/kg bw/d for triclocarban (EPA, 2002). Therefore, none of the examined exposure routes present concerns, on an acute basis, to human health. Exposure of triclocarban from eating pumpkin and zucchini grown in fields receiving biosolids is two orders less than exposure from using products containing triclocarban, about 35 times greater than exposure from drinking water, and about 250 times greater than exposure from eating soybeans grown in fields receiving biosolids. Assuming a per capita consumption of 534 g/day fresh vegetables (USDA, 2001e2002) and that the triclocarban concentrations in pumpkin shoots are representative of all vegetables, the
Fig. 5 e Comparison of triclocarban dose associated with multiple routes of exposure for an adult. Error bars represent maximum and minimum doses. The NOAEL (noobservable adverse effect level) for triclocarban is 25 mg/ kg-bw/day.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
resulting antimicrobial exposure from vegetable consumption is (at worst case) similar to the exposure from using personal care products. Therefore, the acute health risk from antimicrobial accumulation by vegetables in fields receiving biosolids is likely minimal. However, additional studies, utilizing a more diverse range of vegetable and fruit crops, are needed to confirm this assessment. However, there is also the issue of antibiotic resistance due to antimicrobial exposure. While there is dearth of information about the development of microbial resistance from using products containing triclocarban, many reports indicate development of antibiotic resistance due to the triclosan. For example, the susceptibility of Escherichia coli, Proteus mirabilis, and Staphylococcus aureus to antibiotics was reduced by 40e400 times from exposure to triclosan (Saleh et al., 2011; Stickler and Jones, 2008; Suller and Russell, 2000). Though reduction of soil antimicrobial concentrations by plants may help to reduce antimicrobial resistance in soils, the ingestion of antimicrobials through food might increase drug and antimicrobial resistance in humans. More research is needed to explore this potential ramification of the accumulation of antimicrobials by food crops.
4.
Conclusions
Research indicated that established plants reduce the migration of antimicrobials from biosolids to water resources through phytoaccumulation and additional unidentified mechanisms. The soil concentrations of triclocarban and/or triclosan in pumpkin and zucchini columns were less than soil concentrations in unplanted columns. Additionally, the masses of antimicrobials for which the study was unable to account were generally greater in columns with plants than in columns without plants. Pumpkin, zucchini, and switch grass plants accumulated triclosan and triclocarban in mg per kg (dry weight) concentrations. In general, roots had higher concentrations of antimicrobials than shoots, but less total mass accumulation of antimicrobials due to high production of shoots. Consequently, results indicate that plants (i) reduce leaching of antimicrobials from fields to which biosolids have been applied, (ii) directly impact the fate of antimicrobials through phytoaccumulation, and (iii) decrease the persistence of antimicrobials in soil systems through additional, unidentified mechanisms. There was no acute human risk to humans due to eating pumpkins and zucchini produced from field to which biosolids have been applied. However, additional fieldscale studies with fruit production and more food crops are necessary to more clearly understand human exposure to antimicrobials through food crops. More research is also needed to identify and quantify mechanisms that dictate the fate of antimicrobials in vegetated fields receiving biosolids.
Acknowledgments The authors are grateful to Reinhold Research group members, Department of Biosystems and Agricultural Engineering, Michigan State University. Also, special thanks to the
5551
funding agencies: Center for Water Science, MSU, Michigan Agriculture Experiment Station and College of Engineering, MSU. We would like to thank Dr. Alison Cupples and Dr. Jong M. Cha for assistance with extractions and biosolids analysis.
references
Ahn, K.C., Zhao, B., Chen, J., Cherednichenko, G., Sanmarti, E., Denison, M.S., Lasley, B., Pessah, I.N., Kultz, D., Chang, D.P.Y., Gee, S.J., Hammock, B.D., 2008. In vitro biologic activities of the antimicrobials triclocarban, its analogs, and triclosan in bioassay screens: receptor-based bioassay screens. Environmental Health Perspectives 116, 1203e1210. Allmyr, M., Adolfsson-Erici, M., McLachlan, M.S., SandborghEnglund, G., 2006. Triclosan in plasma and milk from Swedish nursing mothers and their exposure via personal care products. Science of the Total Environment 372, 87e93. Aslund, M.L.W., Rutter, A., Reimer, K.J., Zeeb, B.A., 2008. The effects of repeated planting, planting density, and specific transfer pathways on PCB uptake by Cucurbita pepo grown in field conditions. Science of the Total Environment 405, 14e25. Aslund, M.L.W., Zeeb, B.A., Rutter, A., Reimer, K.J., 2007. In situ phytoextraction of polychlorinated biphenyl (PCB) contaminated soil. Science of the Total Environment 374, 1e12. Bester, K., 2003. Triclosan in a sewage treatment process e balances and monitoring data. Water Research 37, 3891e3896. Calafat, A.M., Ye, X., Wong, L.Y., Reidy, J.A., Needham, L.L., 2008. Urinary concentrations of triclosan in the US population: 2003e2004. Environmental Health Perspectives 116, 303e307. Cha, J.M., Cupples, A.M., 2009. Detection of the antimicrobials triclocarban and triclosan in agricultural soils following land application of municipal biosolids. Water Research 43, 2522e2530. Chu, S.G., Metcalfe, C.D., 2007. Simultaneous determination of triclocarban and triclosan in municipal biosolids by liquid chromatography tandem mass spectrometry. Journal of Chromatography A 1164, 212e218. Coogan, M.A., La Point, T.W., 2008. Snail bioaccumulation of triclocarban, triclosan, and methyltriclosan in a North Texas, USA, stream affected by wastewater treatment plant runoff. Environmental Toxicology and Chemistry 27, 1788e1793. Drewes, J.E., Heberer, T., Rauch, T., Reddersen, K., 2003. Fate of pharmaceuticals during ground water recharge. Ground Water Monitoring and Remediation 23, 64e72. Dzantor, E.K., Chekol, T., Vough, L.R., 2000. Feasibility of using forage grasses and legumes for phytoremediation of organic pollutants. Journal of Environmental Science and Health Part A-Toxic/Hazardous Substances & Environmental Engineering 35, 1645e1661. EPA, 2002. High Production Volume (HPV) Chemical Challenge Program Data Availability and Screening Level Assessment for Triclocarban. The TCC Consortium. EPA, 2003. TSCA Chemical Substances Inventory Washington D.C. Essandoh, H.M.K., Tizaoui, C., Mohamed, M.H.A., Amy, G., Brdjanovic, D., 2010. Fate of triclocarban during soil aquifer treatment: soil column studies. Water Science and Technology 61, 1779e1785. Fraker, S.L., Smith, G.R., 2004. Direct and interactive effects of ecologically relevant concentrations of organic wastewater contaminants on Rana pipiens tadpoles. Environmental Toxicology 19, 250e256. Halden, R.U., Paull, D.H., 2004. Analysis of triclocarban in aquatic samples by liquid chromatography electrospray ionization mass spectrometry. Environmental Science and Technology 38, 4849e4855.
5552
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 4 5 e5 5 5 2
Halden, R.U., Paull, D.H., 2005. Co-occurrence of triclocarban and triclosan in US water resources. Environmental Science and Technology 39, 1420e1426. Heath, R.J., Yu, Y.T., Shapiro, M.A., Olson, E., Rock, C.O., 1998. Broad spectrum antimicrobial biocides target the FabI component of fatty acid synthesis. Journal of Biological Chemistry 273, 30316e30320. Heidler, J., Halden, R.U., 2007. Mass balance assessment of triclosan removal during conventional sewage treatment. Chemosphere 66, 362e369. Heidler, J., Sapkota, A., Halden, R.U., 2006. Partitioning, persistence, and accumulation in digested sludge of the topical antiseptic triclocarban during wastewater treatment. Environmental Science and Technology 40, 3634e3639. Hu¨lster, A., Mu¨ller, J.F., Marschner, H., 1994. Soil-plant transfer of polychlorinated dibenzo-p-dioxins and dibenzofurans to vegetables of the cucumber family (Cucurbitaceae). Environmental Science and Technology 28, 1110e1115. Johnson, D.R., Czechowska, K., Chevre, N., van der Meer, J.R., 2009. Toxicity of triclosan, penconazole and metalaxyl on Caulobacter crescentus and a freshwater microbial community as assessed by flow cytometry. Environmental Microbiology 11, 1682e1691. Lawrence, J.R., Zhu, B., Swerhone, G.D.W., Roy, J., Wassenaar, L.I., Topp, E., Korber, D.R., 2009. Comparative microscale analysis of the effects of triclosan and triclocarban on the structure and function of river biofilm communities. Science of the Total Environment 407, 3307e3316. Liu, F., Ying, G.G., Yang, L.H., Zhou, Q.X., 2009. Terrestrial ecotoxicological effects of the antimicrobial agent triclosan. Ecotoxicology and Environmental Safety 72, 86e92. Lunney, A.I., Zeeb, B.A., Reimer, K.J., 2004. Uptake of weathered DDT in vascular plants: potential for phytoremediation. Environmental Science and Technology 38, 6147e6154. Reinwald, S., Akabas, S.R., Weaver, C.M., 2010. Whole versus the piecemeal approach to evaluating soy. Journal of Nutrition 140, 2335Se2343S. Sabourin, L., Beck, A., Duenk, P.W., Kleywegt, S., Lapen, D.R., Li, H. X., Metcalfe, C.D., Payne, M., Topp, E., 2009. Runoff of pharmaceuticals and personal care products following application of dewatered municipal biosolids to an agricultural field. Science of the Total Environment 407, 4596e4604. Saleh, S., Haddadin, R.N.S., Baillie, S., Collier, P.J., 2011. Triclosan e an update. Letters in Applied Microbiology 52, 87e95. Sapkota, A., Heldler, J., Halden, R.U., 2007. Detection of triclocarban and two co-contaminating chlorocarbanilides in US aquatic environments using isotope dilution liquid chromatography tandem mass spectrometry. Environmental Research 103, 21e29. Snyder, E.H., O’Connor, G.A., McAvoy, D.C., 2010. Measured physicochemical characteristics and biosolids-borne concentrations of the antimicrobial triclocarban (TCC). Science of the Total Environment 408, 2667e2673.
Stickler, D.J., Jones, G.L., 2008. Reduced susceptibility of Proteus mirabilis to triclosan. Antimicrobial Agents and Chemotherapy 52, 991e994. Suller, M.T.E., Russell, A.D., 2000. Triclosan and antibiotic resistance in Staphylococcus aureus. Journal of Antimicrobial Chemotherapy 46, 11e18. USDA (United States Department of Agriculture), 2011. Economic Research Service. Vegetables and Melon Data. www.ers.usda. gov/Data/Vegetables/ (accessed 15.03.11.). USDA (United States Department of Agriculture), Office of Communications, 2001e2002. Agriculture Fact Book: 2001e2002. www.usda.gov/factbook, Washington, DC, p. 161, ISBN: 001-000-04709-4. Veldhoen, N., Skirrow, R.C., Osachoff, H., Wigmore, H., Clapson, D.J., Gunderson, M.P., Van Aggelen, G., Helbing, C.C., 2007. The bactericidal agent triclosan modulates thyroid hormone-associated gene expression and disrupts postembryonic anuran development (vol. 80, p. 217, 2006). Aquatic Toxicology 83, 84. Walsh, S.E., Maillard, J.Y., Russell, A.D., Catrenich, C.E., Charbonneau, D.L., Bartolo, R.G., 2003. Development of bacterial resistance to several biocides and effects on antibiotic susceptibility. Journal of Hospital Infection 55, 98e107. Wang, X.P., White, J.C., Gent, M.P.N., Iannucci-Berger, W., Eitzer, B.D., Mattina, M.J.I., 2004. Phytoextraction of weathered p,p0 -DDE by zucchini (Cucurbita pepo) and cucumber (Cucumis sativus) under different cultivation conditions. International Journal of Phytoremediation 6, 363e385. White, J.C., Wang, X.P., Gent, M.P.N., Iannucci-Berger, W., Eitzer, B.D., Schultes, N.P., Arienzo, M., Mattina, M.I., 2003. Subspecies-level variation in the phytoextraction of weathered p,p’-DDE by Cucurbita pepo. Environmental Science and Technology 37, 4368e4373. Wu, C., Spongberg, A.L., Witter, J.D., Fang, M., Czajkowski, K.P., 2010. Uptake of pharmaceutical and personal care products by soybean plants from soils applied with biosolids and irrigated with contaminated water. Environmental Science and Technology 44, 6157e6161. Wu, C.X., Spongberg, A.L., Witter, J.D., 2009. Adsorption and degradation of triclosan and triclocarban in soils and biosolids-amended soils. Journal of Agricultural and Food Chemistry 57, 4900e4905. Ying, G.G., Kookana, R.S., 2007. Triclosan in wastewaters and biosolids from Australian wastewater treatment plants. Environment International 33, 199e205. Ying, G.G., Yu, X.Y., Kookana, R.S., 2007. Biological degradation of triclocarban and triclosan in a soil under aerobic and anaerobic conditions and comparison with environmental fate modelling. Environmental Pollution 150, 300e305. Zhang, H., Chen, J., Ni, Y., Zhang, Q., Zhao, L., 2009. Uptake by roots and translocation to shoots of polychlorinated dibenzop-dioxins and dibenzofurans in typical crop plants. Chemosphere 76, 740e746.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Speciation of trace inorganic contaminants in corrosion scales and deposits formed in drinking water distribution systems Ching-Yu Peng*, Gregory V. Korshin Department of Civil and Environmental Engineering, University of Washington, Box 352700, Seattle, WA 98105-2700, United States
article info
abstract
Article history:
Sequential extractions utilizing the modified Tessier scheme (Krishnamurti et al., 1995) and
Received 27 December 2010
measurements of soluble and particulate metal released from suspended solids were used
Received in revised form
in this study to determine the speciation and mobility of inorganic contaminants (As, Cr, V,
21 July 2011
U, Cd, Ni, and Mn) found in corrosion scales and particles mobilized during hydraulic
Accepted 9 August 2011
flushing events. Arsenic, chromium and vanadium are primarily associated with the
Available online 30 August 2011
mobilization-resistant fraction that is resistant to all eluents used in this study and also bound in highly stable crystalline iron oxides. Very low concentrations of these elements
Keywords:
were released in resuspension experiments. X-ray absorbance measurements demon-
Arsenic
strated that arsenic in the sample with the highest As concentration was dominated by
Nickel
As(V) bound by iron oxides. Significant fractions of uranium and cadmium were associated
Manganese
with carbonate solids. Nickel and manganese were determined to be more mobile and
Uranium
significantly associated with organic fractions. This may indicate that biofilms and natural
Corrosion
organic matter in the drinking water distributions systems play an important role in the
Surface scales
accumulation and release of these inorganic contaminants. ª 2011 Elsevier Ltd. All rights reserved.
Inorganic contaminant Mobilization Fractionation
1.
Introduction
Accumulation of inorganic contaminants in corrosion solids and sediments commonly found in drinking water distribution systems (DWDSs) has been addressed in significant detail in prior research (Lytle et al., 2004; Schock, 2005; Schock et al., 2008; Friedman et al., 2010). These studies have demonstrated that several heavy elements can be found in scales formed on metals corroding in drinking water. For instance, Lytle et al. (2004) demonstrated that iron-based corrosion products and sediments adsorb and concentrate arsenic in DWDSs. Lead pipe scales have also been reported to have significant levels of As, Cd, Cr, Hg, and V (Schock et al., 2008; Gerke et al., 2009; Kim and Herrera, 2010).
These studies have demonstrated that while heavy metals and allied elements are typically present only at trace levels in treated potable water entering DWDS, they tend to accumulate in DWDS corrosion solids where their concentrations can exceed those in the influent treated water by several orders of magnitude (Valentine and Stearns, 1994; Reiber and Dostal, 2000; Lytle et al., 2004; Schock et al., 2008; Gerke et al., 2009). For instance, Reiber and Dostal (2000) and Lytle et al. (2004) determined that corrosion solids formed in treated water having arsenic levels <10 mg/L had arsenic concentrations of up to several hundred mg/kg. This corresponds to more than a hundred-fold enrichment of corrosion solids with As compared to its concentration in the ambient water.
* Corresponding author. Tel.: þ1 206 660 2233. E-mail address: [email protected] (C.-Y. Peng). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.017
5554
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
Our prior research (Friedman et al., 2010; Peng et al., submitted for publication) demonstrated similar trends for both radionuclides (radium-226, radium-228) and inorganic contaminants (antimony, arsenic, barium, chromium, lead, nickel, selenium, thallium, uranium, vanadium) present in stable surface scales and particles released during hydrant flushing that mobilizes relatively loosely bound DWDS solids (Vreeburg et al., 2009; Husband and Boxall, 2011). While the concentrations of antimony, selenium, cadmium and thallium in these samples were always low (<2 mg/g), levels of other toxic elements such as lead and arsenic were variable ranging from <0.1 mg/g to >1000 mg/g (Hill et al., 2010; Peng et al., submitted for publication). The accumulation of trace inorganic contaminants in DWDS solids can have consequences for the health of exposed populations if these contaminants are released from the scales thus resulting in their high levels at consumers’ tap (Fisher et al., 2000; Lytle et al., 2010). Remobilization of contaminants bound by corrosion solids can occur through several mechanisms, notably physical mobilization of particulates from the solid matrix (e.g., during hydrant flush events or due to the destabilization caused by changes of water chemistry) and/or chemical release via desorption and/or dissolution. Ultimately, the mobility and availability of contaminants accumulated in DWDS solids depend on their physicochemical speciation that encompasses states ranging from mobile species that can be released via desorption to those tightly bound by stable crystalline oxides of iron and, in lesser extent, Ca and Mg carbonates, silica and manganese oxides predominating typical corrosion scales (Linge, 2008). Thus, the determination of mobility and other aspects of the speciation of trace level inorganic contaminants in DWDS solids require that the modes of binding of these elements in the solid matrix be ascertained. That goal can be achieved via selective extractions that allow estimating contributions of the target elements bound by a priori defined types of solid phases. One sequential extraction procedure that has been extensively employed to speciate inorganic contaminants in soils and sediments was proposed by Tessier et al. (1979). It fractionates heavy metal species found in soils or sediments into five operationally defined fractions that correspond to exchangeable, carbonatebound, iron and manganese oxide-bound, organically bound and residual metal. The Tessier’s scheme was later modified to separate the fraction comprising species bound by iron and manganese oxides into narrower-defined fractions such as metaleorganic complex-bound, easily-reducible metal oxidebound, amorphous mineral colloid-bound, and crystalline iron oxide-bound species (Shuman, 1985; Krishnamurti et al., 1995). For this matter, iron and manganese minerals compounds, notably ferrihydrite, goethite, lepidocrocite, magnetite, hydrous manganese oxides, are commonly found in the corrosion scales and deposits of DWDSs (Benjamin et al., 1996; Sarin et al., 2001; Peng et al., 2010). Although the composition and properties of DWDS corrosion solids and deposits may be different from those of soils and sediments for which the Tessier’s and related schemes have been developed, it is reasonable to assume that the sequential extraction can be a powerful fractionation method to
investigate the speciation of heavy metals accumulated in DWDS solids. To our knowledge, sequential extraction methods have not been consistently applied to determine speciation of heavy metals found in DWDS corrosion solids and deposits. In this study, we applied this approach to examine the speciation and potential mobility of several inorganic contaminants (e.g., As, Cr, V, U, Cd, Ni, Mn) occurring at various levels in representative samples of DWDS corrosion scales. Of the elements targeted in this study, arsenic, chromium, uranium and cadmium are regulated by the National Primary Drinking Water Standards (NPDWSs). Manganese is regulated by the National Secondary Drinking Water Standards (NSDWSs), which are non-enforced guidelines concerned with contaminants that may cause aesthetic or cosmetic effects in drinking water (U.S. EPA, 2009a). Vanadium has been listed in the Drinking Water Contaminant Candidate List 3 (CCL3) (U.S. EPA, 2009b). Though currently unregulated by the U.S. EPA, nickel was regulated in the past due to its adverse health effects. The current WHO guideline for nickel is 70 mg/L (Guidelines for Drinking-Water Quality, 2008). In addition to sequential extractions, separate experiments were carried to determine concentrations of the target elements in soluble and particulate fractions (passing nominal pore sizes 5, 2, 1, and 0.4 mm) released to the ambient water. Finally, the structure-sensitive method of X-ray adsorption spectroscopy (XAS) was used to examine the speciation of As bound by one representative DWDS corrosion solids sample.
2.
Materials and methods
2.1.
Solid samples characteristics
Three pipe specimens (CC-A, CC-B, and CC-D) and two hydrant flush samples (J-E and J-J) were selected for the experiments based on higher concentrations of trace inorganic contaminants in them. These samples were obtained from the earlier study funded by Water Research Foundation (Friedman et al., 2010). Relevant characteristics and elemental compositions of the examined samples are shown in Table 1. Prior to all experimental procedures, the solids were dried at 103 C for 1 day, crushed using a mortar and pestle, passed through a number 50 sieve (300-mm mesh) and homogenized. Four samples of the samples that had a sufficient mass were analyzed by X-ray Diffraction (XRD) measurement to identify their mineralogical phases. XRD analyses were described in more detail in Peng et al. (2010) and Friedman et al. (2010). Specific surface area was carried out using a Quantachrome NOVA 4200e instrument and calculated using multi-point BET method.
2.2. Sequential extraction procedures and operationally defined metal fractions Speciation of the selected inorganic contaminants (As, Cr, V, U, Cd, Ni, Mn) in the solid samples was determined based on the sequential extraction scheme described by Krishnamurti et al. (1995). Important details of the scheme are summarized in
5555
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
Table 1 e Characteristics and elemental compositions of solid samples examined in this study. Sample ID
Sample type
Pipe materiala
Fe (mg/g)
Mn (mg/g)
As (mg/g)
CC-A
Pipe specimen Pipe specimen Pipe specimen Hydrant flush Hydrant flush
Cast iron
235,000
46,692
40.2
Cast iron
117,000
28,957
234
29.1
Cast iron
252,000
21,654
140
78.5
Cast iron
146,000
760
Cast iron
335,000
1,091
CC-B CC-D J-E J-J
a b c d
3.88
Cr (mg/g) 7.13
197
30.9
65.2
V (mg/g)
BETb (m2/g)
U (mg/g)
Cd (mg/g)
Ni (mg/g)
5.08
11.2
121
71.3
1.8
3.4
37.4
15.6
2.51
296
20.08
196
92.5
97.42 NMd
22.2
1.12
0.18
136
23.4
13.9
1.25
0.48
107
50.82
Mineralogyc Magnetite, quartz NM Siderite, quartz, hydroxyapatite Calcite, dolomite, quartz Goethite, magnetite, ferrihydrite, calcite, quartz
For hydrant flush samples, pipe material refers to the type of pipe used to distribute water in the flushing zone. Crushed samples were used to analyze the BET surface area. Mineralogy was based on X-ray diffraction (XRD) measurement. NM: inadequate mass for testing.
Table 2. The following fractions that correspond to different operationally defined modes of binding of inorganic contaminants in the solid phases were established: (1) Exchangeable metal: heavy elements associated with exchangeable binding sites. This mode of binding is expected to result in relatively easy displacement of the bound metals by competing ions. (2) Carbonate-bound metal: heavy elements associated with sorption on or occlusion into carbonates, typically calcite (CaCO3) and dolomite (CaMg(CO3)2). Elements coprecipitated with calcite can be dissolved by acidic sodium acetate. (3) Metaleorganic complexes: heavy elements associated with humic species present in the solid matrix. (4) Easily reducible metal oxide-bound metal: heavy elements associated with easily reducible metal oxides such as manganese (III, IV) oxides. Hydroxylamine hydrochloride (NH2OH$HCl) is specific to Mn oxides, leaving crystalline iron oxides unaffected and dissolving minimal amounts of amorphous iron oxides. (5) Organic-bound metal: heavy elements associated with organic matter other than humic substances.
(6) Amorphous mineral colloid-bound metal: heavy elements associated with amorphous iron oxides and poorly crystalline aluminosilicate mineral colloids. Acidic ammonium oxalate ((NH4)2C2O4) performed in dark is specific for dissolving amorphous minerals. (7) Crystalline iron oxide-bound metal: heavy elements associated with crystalline iron oxides. (8) Mobilization-resistant metal: heavy elements associated with mineral lattices resistant to the above successive sequential extractions (Krishnamurti and Naidu, 2002; Linge, 2008). Sequential extractions were carried out in duplicate using 1 g of each solid sample (except CC-B, in which case 0.4 g was used). Requisite amounts of the solids were placed in 50 mL metal free polypropylene centrifuge tubes (VWR). After each extraction, the tubes containing solids were centrifuged washing for 10 min at 4000 rpm to separate the supernatant. The residual solids were rinsed with 10 mL of de-ionized distilled water, centrifuged again for 10 min at 4000 rpm and the resulting supernatant collected for further analyses. All chemicals used in the sequential extraction procedure were all ACS grade. Their solutions were checked to determine the
Table 2 e Sequential extraction scheme for speciation used in this study (Krishnamurti et al., 1995). Step
Fractions
Reagent
1 2 3 4 5
Exchangeable Carbonate-bound Metaleorganic complex-bound Easily reducible metal oxide-bound Organic-bound
6 7
Amorphous mineral colloid-bound Crystalline iron oxide-bound
8
Mobilization-resistant
10 mL of 1 M Mg(NO3)2 at pH 7 25 mL of 1 M CH3CO2Na at pH 5 30 mL of 0.1 M Na4P2O7$10H2O at pH 10 20 mL of 0.1 M NH2OH$HCl in 0.01 M HNO3 5 mL of 30% H2O2 at pH 2, 3 mL of 0.02 M HNO3 3 mL of 30% H2O2 at pH 2 Cool, 10 mL Mg(NO3)2 in 20% HNO3 10 mL of 0.2 M (NH4)2C2O4 (adjusted to pH 3 with 0.2 M H2C2O4) 25 mL of 0.2 M (NH4)2C2O4 (adjusted to pH 3 with 0.2 M H2C2O4) in 0.1 M ascorbic acid Digestion with HNO3 and 30% H2O2
Reaction time and temperature 4 h at 25 C 6 h at 25 C 20 h at 25 C 30 min at 25 C 2 h at 85 C 2 h at 85 C 30 min at 25 C 4 h at 25 C (dark) 30 min at 95 C
5556
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
concentration of the target (Supplementary data Table S1).
trace
metals
in
them
2.3. Separation of soluble and particulate fractions of metal released from suspended corrosion solids To determine concentration of the target elements in solids assigned to soluble and particulate fractions released from suspended corrosion solids, 20 mg of each sample was dispersed into 100 mL water with alkalinity 100 mg/L as CaCO3 and pH 7.6 placed in 125 mL Nalgene metal free bottles. Samples were placed on a shaker and aliquots were taken for analyses after 30, 60 and 120 min of exposure. The samples were filtered through filters with nominal pore size of 5, 2, 1, and 0.4 mm (Whatman Nuclepore polycarbonate track-etched membranes). The filter units and syringes were prewashed with 1% HNO3 and then rinsed with de-ionized distilled water. Sample aliquots of 15 mL were filtered directly into 15 mL metal free tubes. The first 5 mL of filtrate was discarded. Filtered samples were acidified to 1% HNO3 with ultrapure HNO3 and stored at 4 C prior to analysis.
2.4.
X-ray absorption spectroscopy
X-ray absorbance near edge structure (XANES) and extended X-ray absorption fine structure spectra of three arsenic model compounds and CC-D sample were carried out at beamline X19-A of the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory (BNL), Upton, NY. The X-ray energy was varied from 200 eV below to 1000 eV above the absorption K-edges of As (Ek ¼ 11,868 eV) using a double crystal Si(111) monochromator. Samples were ground and mounted on a piece of 3M scotch tape via brushing; the tape was then folded several times until the intensity of the signal reached a suitable range for numerical analysis. Two model compounds (sodium arsenite NaAsO2 and sodium arsenate (Na2HAsO4$7H2O)) were measured in the transmission mode; while the adsorbed As(V)eFe (synthetic model compound) and CC-D were analyzed in the fluorescence mode. Fluorescence mode was carried out by using the Lytle detector with a Germanium filter and Soller-type slits to minimize the fluorescence background. To improve the signal-to noise ratio, at least 4 measurements were obtained and averaged for the diluted samples. Gold metal foil was always measured in the transmission mode simultaneously with all other samples and used as the reference for the alignment of energies. Energy calibration was performed with Athena software and achieved by assigning the first inflection point of the simultaneously measured Au foil to 11,919 eV. The edge jump of pre- and post-edge regions was then normalized. The function was then transformed from energy unit (eV) to photoelectron wave vector (k) unit ( A1) to produce the EXAFS 3 function (c(k)). The k -weighted c(k) function was then Fourier transformed (FT) using a Hanning window to create the radical structure function (RSF) in R-space ( A). The experimental EXAFS data were then fitted with coordination number (N ), interatomic distance (R), and the DebyeeWaller parameter (s2) using Artemis program. Linear least-squared fitting (LSF) was done on the XANES spectra after background subtraction and normalization with Athena program.
2.5.
Elemental analysis
Concentrations of several elements in samples generated using sequential extractions and filtration experiments that utilized filters with varying sizes were determined by inductively coupled plasma-mass spectroscopy (ICP-MS) (PerkinElmer ELAN DRC-e ICP-MS). The elements quantified in these measurements included arsenic (As), chromium (Cr), vanadium (V), uranium (U), cadmium (Cd), nickel (Ni) and manganese (Mn). Quality control samples, including laboratoryfortified blanks and laboratory-fortified samples, were performed for every ten samples analyzed. Average elemental recoveries ranged from 85.2 to 92.8% for the laboratoryfortified samples. A certified reference material (CRM) (Loamy Sand, CRM 024-050) was purchased from RTC (Laramie, Wyoming) to quality check the analytical method. Results obtained from ICP-MS analysis of certified reference material are summarized in Supplementary data Table S2.
3.
Results and discussion
Average extraction efficiencies for several elements examined in this study are presented in Table 3. They ranged from 95 to 114% with standard deviations shown in Table 3. The percent distribution of inorganic contaminants species extracted from individual solids can be found in Fig. 1. Three pipe specimens (CC-A, CC-B, and CC-D) originated from the same drinking water distribution system but were taken from three different locations. Very similar distributions of inorganic contaminants were found in samples CC-A and CC-D. However, the data for sample CC-B differed from them. Since these specimens originated from the utility that relies on groundwater from multiple wells, ambient water chemistry was likely to vary between the wells thus potentially resulting in variations of DWDS solids’ properties in the distribution system. Two hydrant flush samples (J-E and J-J) were provided by the utility that relies on groundwater from multiple wells and operates a larger water system than utility CC. Accordingly, water quality conditions of distribution system J are likely to vary spatially and temporally because of the different source waters and their blending in the system. Indeed, the compositions of sample J-E and J-J were very different. The hydraulically mobilized material in sample J-E had comparatively low levels
Table 3 e Average extraction efficiencies and corresponding standard deviations (SD) of five corrosion solids and deposits examined in this study.
Arsenic (As) Chromium (Cr) Vanadium (V) Uranium (U) Cadmium (Cd) Nickel (Ni) Manganese (Mn)
Average (%)
SD (%)
107.2 113.5 99.6 106.1 94.8 107.1 103.2
13.6 14.6 17.9 15.2 28.3 31.2 24.5
5557
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
a
b100%
100%
90%
90% 80% 70% 60% 50%
Mobilization-resistant
80%
Crystalline Fe oxide-bound
70%
Amorphous mineral colloid-bound
60%
H2O2 extractable organic-bound
50%
Easily reducible metal oxide-bound
40%
40%
Metal-organic complex-bound
30%
Carbonate-bound
20%
Exchangeable
Crystalline Fe oxide-bound Amorphous mineral colloid-bound H2O2 extractable organic-bound Easily reducible metal oxide-bound Metal-organic complex-bound
30%
Carbonate-bound Exchangeable
20% 10%
10%
0% As
0% As
c
Mobilization-resistant
Cr
V
U
Cd
Ni
Mn
Cr
U
Cd
Ni
Mn
d100%
100%
90%
90% Mobilization-resistant Crystalline Fe oxide-bound Amorphous mineral colloid-bound H2O2 extractable organic-bound Easily reducible metal oxide-bound Metal-organic complex-bound Carbonate-bound Exchangeable
80% 70% 60% 50% 40% 30% 20% 10%
Mobilization-resistant
80%
Crystalline Fe oxide-bound
70%
Amorphous mineral colloid-bound
60%
H2O2 extractable organic-bound
50%
Easily reducible metal oxide-bound
40%
Metal-organic complex-bound
30%
Carbonate-bound
20%
Exchangeable
10%
0% As
Cr
V
U
Cd
Ni
Mn
0% As
e
V
Cr
V
U
Cd
Ni
Mn
100% 90% Mobilization-resistant Crystalline Fe oxide-bound Amorphous mineral colloid-bound H2O2 extractable organic-bound Easily reducible metal oxide-bound Metal-organic complex-bound Carbonate-bound Exchangeable
80% 70% 60% 50% 40% 30% 20% 10% 0% As
Cr
V
U
Cd
Ni
Mn
Fig. 1 e Percent distribution of inorganic contaminants in pipe specimens (a) CC-A, (b) CC-B, (c) CC-D, and hydrant flush solids (d) J-E, (e) J-J examined following the sequential extraction scheme.
3.1.
Arsenic
Chemical fractionation data (Fig. 2) showed that on the average, the distribution of As in five corrosion solids was predominated by the mobilization-resistant fraction (59.3%) that is deemed to correspond to a very low mobility of the
retained contaminant. The fraction of arsenic associated with crystalline iron oxides was next in importance (19.8%), with contributions of amorphous mineral colloid-bound, organicbound, metaleorganic complex-bound fractions becoming progressively smaller (9.3%, 6.5% and 4%, respectively). Arsenic fractions assumed to have highest potential mobility had lowest, almost negligible contributions to the total arsenic, with the arsenic bound to easily reducible metal
Percentage of inorganic contaminants in various fractions
of iron (14.6 wt%) and manganese (760 mg/g), but a relatively high level of total carbon (6.1 wt%). By comparison, levels of Fe, total carbon and Mn in sample J-J were 33.5 wt%, 3.4 wt% and 1091 mg/g, respectively. The distributions of inorganic contaminants found in samples J-E and J-J were also different, especially for elements associated with the organic-bound fraction. Although pipe specimens and hydrant flush samples examined in this study were not from the same distribution system, intrinsic differences in the generation of these two types of samples warrant their comparison that can reveal potentially important differences in retention and release of inorganic contaminants from these substrates. Average percent distributions of inorganic contaminants species shown in Fig. 1(a)e(e) are present in Fig. 2. The following discussion of chemical fractionation data will focus on the data summarized in Fig. 2. A summary of average values and standard deviations of contributions of different fractions of heavy metals established using the sequential extraction scheme (Fig. 2) is provided in the Supplementary data Table S3.
100% 90%
Mobilization-resistant
80%
Crystalline Fe oxide-bound
70%
Amorphous mineral colloid-bound 60%
H2O2 extractable organic-bound
50%
Easily reducible metal oxide-bound
40%
Metal-organic complex-bound
30%
Carbonate-bound
20%
Exchangeable
10% 0%
As
Cr
V
U
Cd
Ni
Mn
Fig. 2 e Average percent distributions of inorganic contaminants species in the corrosion solids and deposits examined in this study following the sequential extraction scheme.
5558
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
Table 4 e Average concentrations (mg/L) of inorganic contaminants released from suspension of corrosion solids and deposits. Dissolved (<0.45 mm)
As U Cd Ni Mn
Particulate (0.45e1 mm)
Particulate (1e2 mm)
Particulate (2e5 mm)
PS
HF
PS
HF
PS
HF
PS
HF
0.170 0.019 0.051 3.487 377.132
0.132 0.006 0.024 0.837 7.294
0.039 0.001 0.002 0.114 14.038
0 0.001 0.004 0.116 0.111
0.067 0.003 0.005 0.122 15.594
0.001 0.001 0.004 0.162 0.515
0.163 0.019 0.037 0.194 39.542
0.002 0.004 0.022 0.868 1.947
PS: pipe specimen. Three pipe specimens include CC-A, CC-B and CC-D. HF: hydrant flush solid. Two hydrant flush solids include J-E and J-J.
oxides, carbonate-bound and exchangeable sites having concentrations of 0.6%, 0.3% and 0.2%, respectively. These results demonstrate that arsenic in the corrosion solids was dominated by the mobilization-resistant and crystalline iron oxide fractions contributing together to ca. 80% of the total. This may also indicate that As in these solids is mostly occluded in the stable mineral structures and, as a result, it is not likely to be easily mobilized. This is not surprising as As(V) is well known to have strong retention on mineral surfaces, especially iron minerals (Wilkie and Hering, 1996; Dixit and Hering, 2003). This also shows that the level of soluble arsenic release associated with possible chemical or
a
U
b
0.4 µm
Cd
1 µm
As
2 µm
Ni
5 µm
physical perturbation affecting corrosion solids is expected by largely negligible. This conclusion was confirmed by the data obtained when these samples were suspended in water having the pH 7.6 and alkalinity 100 mg/L as CaCO3, respectively. The average concentrations of inorganic contaminants released into ambient water from the five examined solids and belonging to different size fractions are shown in Table 4. Average concentrations (in log scale) of released inorganic contaminants of individual solids in different size fractions are provided in Fig. 3. For arsenic, 2.8% and 8.7%, respectively, of the total As were released from suspended pipe specimens and hydrant
0.4 µm
U
1 µm
Cd
Mn
As
2 µm
Ni
5 µm
Mn -3
-2
-1
0
1
2
3
4
c
-3
-2
-1
0
1
2
3
4
d U
U
0.4 µm
0.4 µm
Cd
1 µm
Cd
As
2 µm
As
2 µm
Ni
5 µm
5 µm
Ni
1 µm
Mn
Mn
-3
-3
-2
-1
0
1
2
3
-2
-1
0
1
2
3
4
4
e U
0.4 µm 1 µm
Cd As
2 µm
Ni
5 µm
Mn -3
-2
-1
0
1
2
3
4
Log (released inorganic concentrations, in µg/L) Fig. 3 e Concentrations of inorganic contaminants released from suspended corrosion solids (200 mg/L) at pH 7.6, alkalinity 100 mg/L passing filters with varying nominal pore sizes. (a) CC-A, (b) CC-B, (c) CC-D, (d) J-E, and (e) J-J solid samples.
5559
a
80% 2~5 µm
60%
1~2 µm 0.4~1 µm
40%
<0.4 µm
20%
Normalized µ(E)
100%
LCF 0% PS
HF As
PS
HF U
PS
HF Cd
PS
HF Ni
PS
As(V)-Fe (91.7%)
HF
NaAsO2 (8.3%)
Mn
Fig. 4 e Average contributions of inorganic contaminants released from suspended corrosion solids and deposits and associated with different size fractions. (PS: pipe specimen, including CC-A, CC-B and CC-D. HF: hydrant flush solid, including J-E and J-J.)
flush solids and passing through a 5 mm filter. The average As levels released from suspended pipe specimens and hydrant flush solids into water were 0.44 mg/L and 0.14 mg/L, respectively. Most of the arsenic released in the resuspension experiments from three pipe specimens was in the operationally defined soluble dissolved metal fraction (<0.4 mm) followed by the fraction assigned to particles with sizes between 2 and 5 mm. These two major size fractions contributed to ca. 80% of the total As passing a 5 mm filter while the 1e2 mm and 0.4e1 mm fractions constituted the rest, as shown in Fig. 4. The speciation of arsenic released from suspended samples exhibited a different distribution pattern for pipe specimens and hydrant flush solids. In all cases, actual concentrations of arsenic were very low. In the case of solids retained on the surface of corroding pipe, the average released As concentrations in dissolved fraction (<0.4 mm) and those corresponding to three ranges of varying particles sizes (2e5 mm, 1e2 mm and 0.4e1 mm) were 39%, 37%, 15% and 9%, respectively. In contrast, for hydrant flush solids, almost all (98%) released arsenic concentrations existed in the dissolved fraction. X-ray adsorption spectroscopy (XAS) was employed to further explore the chemical nature of arsenic in sample CC-D that had the highest As concentration among all studied solids. The white-line position of As in its X-ray absorbance spectra is known to be sensitive to the oxidation state of this element, with the white-line locations at 11,871.5 and 11,875 eV for As(III) and As(V), respectively (e.g., Manning et al., 2002; Cance`s et al., 2008). Accordingly, we examined the normalized As K-edge XANES spectrum of sample CC-D and processed the data using linear combinations of XANES spectra of model compounds with known oxidation states fitted by means of linear least-squared fitting (LSF). Results of this fitting are shown in Fig. 5(a). The LSF procedure indicated that the arsenic in sample CC-D is dominated by As(V) components (91%), especially by As(V) having Fe atoms in its chemical environment. Less than 9% of the total As present in sample CC-D appears to exist as As(III).
11840
11860
11880
11900
11920
11940
1
2
3
4
5
b Fourier Transform of X(k)k3
Percentage of inorganic contaminants in various fractions
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
0
Fig. 5 e (a) Linear combination fitting of XANES and (b) Fourier transforms of the EXAFS for As in corrosion solid CC-D. The solid bold line in (a) represents the XANES data for corrosion solid CC-D. The dotted line in (a) corresponds to the linear combination fitting (LCF) of two model compounds (As(V)eFe and NaAsO2). The solid and dashed lines in (a) correspond to the XANES data for As(V)eFe and NaAsO2 model compounds. The peak positions in (b) are uncorrected for phase shits. The dashed line in (b) is the model fits.
The local coordination environment of arsenic in corrosion solid CC-D was also examined by EXAFS spectroscopy. Fourier transformed spectra of As in CC-D are shown in Fig. 5(b). Peak positions shown in that figure are uncorrected for phase shifts so that they are slightly shifted from the true interatomic distances. The experimental spectrum was fitted with a theoretical model to yield interatomic distance (R), coordination number (CN), and the DebyeeWaller parameter (s2) based on the optimized fit. These calculations (their results are summarized in Table 5) showed that in the first As coordination shell, four oxygen atoms located at a 1.67 A distance from the central As atom were present. This AseO distance was statistically identical to the average AseO distances in the model compounds, such as scorodite (FeAsO4$4H2O) and sodium arsenate (Na2HAsO4$7H2O) (Waychunas et al., 1993), and arsenate adsorbed on or coprecipitated with goethite,
5560
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
Table 5 e EXAFS parameters defining local coordination environment of As in corrosion solid (CC-D). Shell Distance ( A) Coordination DebyeeWaller AseO AseFe
1.67 3.30
number (CN)
factor (s2)
4 2
0.001 0.002
lepidocrocite, hematite or ferrihydrite (Fendorf et al., 1997; Sherman and Randall, 2003). The second peak in the spectra shown in Fig. 5(b) appears to indicate the presence of two iron atoms located at distance of 3.3 A from the central As atom. This AseFe distance appears to correspond to As associated with the surfaces of iron oxides by double-corner sharing (Fendorf et al., 1997; Sherman and Randall, 2003; Cance`s et al., 2005). These data confirm that As present in the examined sample is in primarily þ5 oxidation state and it is at least partly associated with Fe-containing matrixes. This is in accord with data of prior research showing that As(V) tends to sorb very strongly on and be bound by mineral surfaces, especially iron minerals; while As(III) is considerably more mobile (Manning et al., 2002; Meng et al., 2002).
carbonate fraction was 16.2% (Fig. 2). The associations of uranium with these fractions are in agreement with the strong complexation of the uranyl ion (UO2þ 2 ) prevalent in the aerobic environment with carbonate and natural organic matter (e.g., Zhou and Gu, 2005; Bednar et al., 2007; Stewart et al., 2010). In experiments with suspended solids, 2.3% (0.042 mg/L) and 5.2% (0.012 mg/L) of U were released on the average from three pipe specimens and two hydrant flush solids, respectively. For the three pipe specimens, the average contributions of uranium associated with the 2e5 mm, 1e2 mm, 0.4e1 mm and <0.4 mm (soluble) fractions were 46.1%, 6.9%, 1.6% and 45.3%, respectively. The size fractions of uranium for the hydrant flush solids were similar, with 50% of released U found in the dissolved fraction (<0.4 mm), followed by 31.9%, 11.1% and 6.9% associated with the particular fraction of 2e5 mm, 1e2 mm and 0.4e1 mm, respectively (Fig. 4).
3.4.
Cadmium
Chromium and vanadium exhibited somewhat similar fractionation patterns (Fig. 2). The majority of Cr in the examined solids was associated with the mobilization-resistant (48.4%), crystalline iron oxide-bound (18.6%) and organic-bound (15.6%) fractions. The proportions of V occurring in the mobilizationresistant, crystalline iron oxide-bound and organic-bound fractions were 27.8%, 20.3% and 24.3%, respectively. A very low percentage of Cr (0%) and V (0.04%) were found in the exchangeable fraction. These findings suggest that Cr and V in these solids are located primarily in the stable matrixes and appear to be relatively immobile. This result is in agreement with the results of prior research demonstrating that ironbased oxides are highly effective sorbents that remove chromium and vanadium in a wide range of water conditions (Smith and Ghiassi, 2006; Naeem et al., 2007). This conclusion was supported by the data of determinations of soluble and particulate Cr and V concentrations released from suspended solids. In these experiments, the concentrations of Cr and V released were consistently below the method detection limits (Cr: 0.054 mg/L; V: 0.078 mg/L) and hence they were not included in Table 4, Figs. 3 and 4.
Chemical fractionation data (Fig. 2) showed that 25.3% of Cd in five corrosion solids was in the mobilization-resistant fraction. The next dominant fraction was cadmium associated with crystalline iron oxides (18.5%), followed by organicbound (16.1%) and carbonate-bound (12.6%) metal. The remaining fractions had smaller contributions that decreased in the order: easily reducible metal oxides (9%) > amorphous mineral colloid (8.2%) > metaleorganic complex (7.9%) > exchangeable cadmium (2.4%). Only 43.8% of Cd was retained in the stable mineral structures; while 31.9% of Cd, including exchangeable, carbonate, metaleorganic complex and easily reducible metal oxides, was associated with potential mobile fractions. The percentage of Cd released from the suspended solids was higher than that determined for As, V, Cr and U. Specifically, 8.8% (0.096 mg/L) and 59.7% (0.054 mg/L) of the total Cd were released from three pipe specimens and two hydrant flush solids, respectively. The dominant fraction of Cd released from three pipe specimens (53.8%) and two hydrant flush samples (43.8%) was in dissolved (<0.4 mm) fraction. For three pipe specimens, the next predominant fraction was in particular fraction of 2e5 mm (38.3%), followed by particular fraction of 1e2 mm (5.4%) and particular fraction of 0.4e1 mm (2.6%). For two hydrant flush samples, the particular fraction of 2e5 mm (41.3%) was also the second important fraction, followed by particular fraction of 0.4e1 mm (7.8%) and particular fraction of 1e2 mm (7.1%) (Fig. 4).
3.3.
3.5.
3.2.
Chromium and vanadium
Uranium
Sequential extractions showed that in the case of uranium the mobilization-resistant (31.2%), organic-bound (22.6%), and carbonate-bound (16.2%) fractions had the highest contributions. The sum of the contributions of the mobilizationresistant and crystalline iron oxide-bound fractions accounted for 41.4% of the total U, which is less compared to all other elements examined in this study (except Mn). The data also show that combined contributions of the organic-bound and metaleorganic complex-bound fractions were 34.6% of the total uranium, while the contribution of the
Nickel and manganese
Sequential extractions and filtration results demonstrated the existence of similar fractionation profiles for nickel and manganese, especially for pipe specimens (Figs. 2 and 4). This is possibly associated with a high affinity of hydrated manganese oxides to nickel and resulting co-accumulation of these two metals (Green-Pedersen et al., 1997; Trived and Axe, 2001; Trived et al., 2001). In the case of nickel, the mobilization-resistant fraction contained the largest percentage (39.1%) of the metal. Organically bound nickel accounted for 21.3% of the total followed
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
by other potentially mobile fractions including nickel associated with easily reducible metal oxides (9.1%), followed by the exchangeable and carbonate-bound Ni (8.8 and 7.6%, respectively). The organically bound manganese was the predominant fraction (28.1%), followed by the mobilization-resistant fraction (27.6%). Nominally mobile fractions of manganese accounted for 36.4% of the total. These fractions included easily reducible metal oxides (18.6%), carbonate-bound (9.1%), and exchangeable (8.7%). The least important fractions are still amorphous mineral colloid-bound (3.9%), crystalline iron oxide-bound (2.6%), and metaleorganic complex-bound (1.4%). Filtration data obtained using suspended solids showed that on the average 8.8% (3.92 mg/L) and 7.7% (1.98 mg/L) of the total Ni present in these samples were released into water from three pipe specimens and two hydrant flush samples, respectively. 89% of Ni released from the pipe specimens was in the dissolved (<0.4 mm) fraction while in case of hydrant flush samples only 42.2% of the released Ni passed through 0.4 mm filter. At the same time, a large part of released Ni (43.8%) from two hydrant flush samples is in particular fraction of 2e5 mm. On the other hand, 6.1% (446.3 mg/L) and 5% (9.87 mg/L) of total Mn were released into water from three pipe specimens and two hydrant flush samples, respectively. 84.5% and 73.9% of released Mn from three pipe specimens and two hydrant flush samples are in the dissolved (<0.4 mm) fraction. It needs to be recognized however, that manganese concentrations tend to be two or more magnitudes larger than those of nickel in the pipe specimens; although the release percentages of these two elements are very similar, manganese will have overwhelming concentrations compared to that of nickel.
3.6.
Practical implications
The experimental data presented above demonstrate the existence of pronounced differences in the mobility of inorganic contaminants retained by DWDS corrosion scales and, when released, potentially affecting human health. The data show that the physico-chemical properties of As, Cr and V retained by DWDS corrosion scales accumulated are similar. These three elements were found to be tightly bound by the solid matrixes and are expected to exhibit little mobility under most conditions typical for drinking water systems. However, because arsenic and, very likely, vanadium exist in anionic forms in these conditions, their mobility can hypothetically be increased via competition with and displacement by phosphate ions utilized for corrosion control (Jain and Loeppert, 2000; Copeland et al., 2007). Actual occurrence of such competition remains to be ascertained. On the other hand, arsenic release may be associated with particulate matter, as was observed by Lytle et al. (2010) for a small drinking water system in which high arsenic levels were associated with iron oxide particles carried in the system by the source water. Our speciation data similarly indicate that because As present in corrosion solids is dominated by the mobilization-resistant and crystalline iron oxide-bound fractions, its release can occur primarily via particulates physically dislodged by hydraulic events or via colloidal mobilization of iron-based corrosion solids caused by changes of water chemistry. Conditions possibly associated with such
5561
phenomena in drinking water distribution systems need to be investigated in the future. Similarly to our observations, high concentrations of vanadium (35 to 899 mg/g) in scales formed on surfaces of galvanized iron were reported by Gerke et al. (2010). In that study, the vanadium was determined to be present as discrete grains of the mineral vanadinite Pb5(V5þO4)3Cl that were formed “up stream” from lead pipe present in the examined systems; Gerke et al. (2010) concluded that the vanadinite solid phase was mostly embedded in near-surface regions of iron corrosion byproducts and, in the absence of physical dislodging or colloidal mobilization, it was not likely to increase V concentrations in the ambient water to dangerous levels. Our vanadium speciation data are in agreement with the above observations as they confirm that, the vanadium in the tested samples was primarily associated with the mobilization-resistant and crystalline iron oxide-bound fractions and the concentration of vanadium released from iron-based corrosion solids were very low. Similar observation was made in this study for chromium but, in contrast with vanadium, the accumulation and release of Cr have not been sufficiently addressed in prior research. In contrast with As, Cr and V, significant fractions of uranium and cadmium (ca. 16% and 13%, respectively) were associated with carbonate-type solids. Prior research (e.g., Rihs et al., 2004; Khaokaew et al., 2011) shows that calcite and other solid carbonate minerals can strongly sorb uranium and cadmium. Such minerals are common in corrosion scales and especially in hydrant flush samples where they can be dominant solid phases (Peng et al., 2010). The retention of cadmium by these phases appears to be at least partially reversible as evidenced by the results that ca. 60% of the total cadmium was released from the resuspended hydrant flush samples. While actual levels of U and Cd in DWDS solids are very low and per se not likely to be of concern, this finding suggests that changes of drinking water alkalinity can be accompanied by changes of levels of these contaminants. Nickel and manganese were determined to be the most mobile of all contaminants examined in our study. About 9% of the retained Ni and Mn were found to exist as the exchangeable fraction that is expected to be highly susceptible to variations of the ambient water quality. While estimation of health effects potentially associated with the release of nickel and manganese accumulated in DWDS solids goes beyond the scope of this paper, this finding highlights the importance of manganese removal from drinking water and control of its status within drinking water distribution networks. Notable contributions (more than 20%) of organic-bound fractions of vanadium, uranium, nickel and manganese may also indicate that biofilms and natural organic matter in the DWDSs play an important role in the accumulation of these inorganic contaminants. This is in agreement with prior findings showing that extracellular polymer substances (EPS) and biofilms per se can sorb inorganic contaminants (e.g., Flemming, 1995; Lalonde et al., 2007; Hitchcock et al., 2009). On the other hand, this finding indicates that changes of water chemistry causing destabilization of biofilms can be accompanied by release the sorbed contaminants. Further characterization of natural organic matter and biofilms found in DWDS solids is needed to provide more insight into the nature of binding of the examined and other inorganic contaminants.
5562
4.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
Conclusions
Results of sequential extractions and measurements of soluble and particulate metal fractions released from suspended corrosion solids indicate that all examined trace-level heavy metals (As, Cr, V, U, Cd, Ni, and Mn) exhibit specific features of their fractionation and mobility. Arsenic, chromium and vanadium are primarily associated with the mobilizationresistant fraction that was unaffected by all eluents used in this study. At the same time, low percentages of As, Cr and V were associated with mobile fractions (exchangeable, carbonate-bound, metaleorganic complex, and easily reducible metal oxide-bound). X-ray absorbance measurements demonstrated that the arsenic in the sample with the highest As concentration was dominated by As(V) bound by iron oxides. Measurements of soluble and particulate metal concentrations demonstrated that only in the case of Ni and Mn released from solids suspended at pH 7.6 and alkalinity 100 mg/L the majority of the released metal was in the dissolved (<0.4 mm) fraction while the other elements were mostly associated particles with sizes between 0.4 and 5 mm. However, in the case of arsenic, the concentrations of release As were very low and almost all (98%) As released from the hydrant flush solids was in the dissolved fraction. The data of the resuspension and sequential extraction experiments were in close agreement and showed that Ni and Mn are much more mobile than all the other inorganic contaminants examined in this study.
Acknowledgments This study was supported by National Science Foundation (Grant # 0931676) and partially by Water Research Foundation (Project #3118) and the USEPA. The content and conclusions are the views of the authors and do not necessarily reflect the views of the funding agencies. The authors would also like to thank Prof. Anatoly I. Frenkel and the Synchrotron Catalysis Consortium for the support and facilitation of the EXAFS experiments at Brookhaven National Laboratory.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2011.08.017.
references
Bednar, A.J., Medina, V.F., Ulmer-Scholle, D.S., Frey, B.A., Johnson, B.L., Brostoff, W.N., Larson, S.L., 2007. Effects of organic matter on the distribution of uranium in soil and plant matrices. Chemosphere 70 (2), 237e247. Benjamin, M.M., Sontheimer, H., Leroy, P., 1996. Corrosion of iron and steel. In: Internal Corrosion of Water Distribution Systems. Cooperative Research Report. AWWA Research Foundation, Denver, CO, pp. 29e70.
Cance`s, B., Juillot, F., Morin, G., Laperche, V., Polya, D., Vaughan, D.J., Hazemann, J.-L., Proux, O., Brown Jr., G.E., Calas, G., 2008. Changes in arsenic speciation through a contaminated soil profile: a XAS based study. Science of the Total Environment 397 (1e3), 178e189. Cance`s, B., Juillot, F., Morin, G., Laperche, V., Alvarez, L., Proux, O., Hazemann, J.-L., Brown Jr., G.E., Calas, G., 2005. XAS evidence of As(V) association with iron oxyhydroxides in a contaminated soil at a former arsenical pesticide processing plant. Environmental Science & Technology 39 (24), 9398e9405. Copeland, R.C., Lytle, D.A., Dionysiou, D.D., 2007. Desorption of arsenic from drinking water distribution system solids. Environmental Monitoring and Assessment 127 (3), 523e535. Dixit, S., Hering, J.G., 2003. Comparison of arsenic(V) and arsenic(III) sorption onto iron oxide minerals: implications for arsenic mobility. Environmental Science & Technology 37 (18), 4182e4189. Fendorf, S., Eick, M.J., Grossl, P., Sparks, D.L., 1997. Arsenate and chromate retention on goethite: I. Surface structure. Environmental Science & Technology 31 (2), 315e320. Fisher, E.L., Fuortes, L.J., Valentine, R.L., Mehrhoff, M., Field, R.W., 2000. Dissolution of 226Radium from pipe-scale deposits in a public water supply. Environment International 26 (1), 69e73. Flemming, H.C., 1995. Sorption sites in biofilms. Water Science & Technology 32 (8), 27e33. Friedman, M.J., Hill, A.S., Reiber, S.H., Valentine, R.L., Larsen, G., Young, A., Korshin, G.V., Peng, C.-Y., 2010. Assessment of Inorganics Accumulation in Drinking Water Systems Scales and Sediments. Water Research Foundation, Denver, CO. Gerke, T.L., Scheckel, K.G., Schock, M.R., 2009. Identification and distribution of vanadinite (Pb5(V5þO4)3Cl) in lead pipe corrosion by-products. Environmental Science & Technology 43 (12), 4412e4418. Gerke, T.L., Scheckel, K.G., Maynard, J.B., 2010. Speciation and distribution of vanadium in drinking water iron pipe corrosion by-products. Science of the Total Environment 408 (23), 5845e5853. Guidelines for Drinking-Water Quality, third ed., 2008 World Health Organization, Geneva. Green-Pedersen, H., Jensesn, B.T., Pind, N., 1997. Nickel adsorption on MnO2, Fe(OH)3, montmorillonite, humic acid and calcite: a comparative study. Environmental Technology 18 (8), 807e815. Hill, A.S., Friedman, M.J., Reiber, S.H., Korshin, G.V., Valentine, R. L., 2010. Behavior of trace inorganic contaminants in drinking water distribution systems. Journal of the American Water Works Association 102 (7), 107e118. Hitchcock, A.P., Dynes, J.J., Lawrence, J.R., Obst, M., Swerhone, G. D.W., Korber, D.R., Leppard, G.G., 2009. Soft X-ray spectromicroscopy of nickel sorption in a natural river biofilm. Geobiology 7 (4), 432e453. Husband, P.S., Boxall, J.B., 2011. Asset deterioration and discolouration in water distribution systems. Water Research 45 (1), 113e124. Jain, A., Loeppert, R.H., 2000. Effect of competing anions on the adsorption of arsenate and arsenite by ferrihydrite. Journal of Environmental Quality 29 (5), 1422e1430. Khaokaew, S., Chaney, R.L., Landrot, G., Ginder-Vogel, M., Sparks, D.L., 2011. Speciation and release kinetics of cadmium in an alkaline paddy soil under various flooding periods and draining conditions. Environmental Science & Technology 45 (10), 4249e4255. Kim, E.J., Herrera, J.E., 2010. Characteristics of lead corrosion scales formed during drinking water distribution and their potential influence on the release of lead and other contaminants. Environmental Science & Technology 44 (16), 6054e6061.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 5 3 e5 5 6 3
Krishnamurti, G.S.R., Huang, P.M., Van Rees, K.C.J., Kozak, L.M., Rostad, H.P.W., 1995. Speciation of particulate-bound cadmium of soils and its bioavailability. Analyst 120 (3), 659e665. Krishnamurti, G.S.R., Naidu, R., 2002. Solidesolution speciation and phytoavailability of copper and zinc in soils. Environmental Science & Technology 36 (12), 2645e2651. Lalonde, S.V., Amskold, L.A., Warren, L.A., Konhauser, K.O., 2007. Surface chemical reactivity and metal adsorptive properties of natural cyanobacterial mats from an alkaline hydrothermal spring, Yellowstone National Park. Chemical Geology 243 (1), 36e52. Linge, K.L., 2008. Methods for investigating trace element binding in sediments. Critical Reviews in Environmental Science and Technology 38 (3), 165e196. Lytle, D.A., Sorg, T.J., Frietch, C., 2004. Accumulation of arsenic in drinking water distribution systems. Environmental Science & Technology 38 (20), 5365e5372. Lytle, D.A., Sorg, T.J., Muhlen, C., Wang, L., 2010. Particulate arsenic release in a drinking water distribution system. Journal of the American Water Works Association 102 (3), 87e98. Manning, B.A., Hunt, M.L., Amrhein, C., Yarmoff, J.A., 2002. Arsenic(III) and arsenic(V) reactions with zerovalent iron corrosion products. Environmental Science & Technology 36 (24), 5455e5461. Meng, X., Korfiatis, G.P., Bang, S., Bang, K.W., 2002. Combined effects of anions on arsenic removal by iron hydroxides. Toxicology Letters 133 (1), 103e111. Naeem, A., Westerhoff, P., Mustafa, S., 2007. Vanadium removal by metal (hydr)oxide adsorbents. Water Research 41 (7), 1596e1602. Peng, C.-Y., Hill, A.S., Friedman, M.J., Valentine, R.L., Larson, G.S., Young, A.M., Reiber, S.H., Korshin, G.V. Occurrence of trace inorganic contaminants in drinking water distribution systems. Journal of the American Water Works Association, submitted for publication. Peng, C.-Y., Korshin, G.V., Valentine, R.L., Hill, A.S., Friedman, M.J., Reiber, S.H., 2010. Characterization of elemental and structural composition of corrosion scales and deposits formed in drinking water distribution systems. Water Research 44 (15), 4570e4580. Reiber, S., Dostal, G., 2000. Well water disinfection sparks surprises. Opflow 26 (3), 1e6. Rihs, S., Sturchio, N.C., Orlandini, K., Cheng, L., Teng, H., Fenter, P., Bedzyk, M.J., 2004. Interaction of uranyl with calcite in the presence of EDTA. Environmental Science & Technology 38 (19), 5078e5086. Sarin, P., Snoeyink, V.L., Bebee, J., Kriven, W.M., Clement, J.A., 2001. Physico-chemical characteristics of corrosion scales in old iron pipes. Water Research 35 (12), 2961e2969. Schock, M.R., 2005. Distribution systems as reservoirs and reactors for inorganic contaminants. In: Distribution System
5563
Water Quality Challenges in the 21st Century. AWWA Reserch Foundation, Denver, CO, pp. 105e140. Schock, M.R., Hyland, R., Welch, M., 2008. Occurrence of contaminant accumulation in lead pipe scales from domestic drinking water distribution systems. Environmental Science & Technology 42 (12), 4285e4291. Sherman, D.M., Randall, S.R., 2003. Surface complexation of arsenic(V) to iron(III) (hydr)oxides: structural mechanism from ab initio molecular geometries and EXAFS spectroscopy. Geochimica et Cosmochimica Acta 67 (22), 4223e4230. Shuman, L.M., 1985. Fractionation method for soil microelements. Soil Science 140 (1), 11e22. Smith, E., Ghiassi, K., 2006. Chromate removal by an iron sorbent: mechanism and modeling. Water Environment Research 78 (1), 84e93. Stewart, B.D., Mayes, M.A., Fendorf, S., 2010. Impact of uranylcalcium-carbonato complexes on uranium(VI) adsorption to synthetic and nature sediments. Environmental Science & Technology 44 (3), 928e934. Tessier, A., Campbell, P.G.C., Bisson, M., 1979. Sequential extraction procedure for the speciation of particulate trace metals. Analytical Chemistry 51 (7), 844e851. Trived, P., Axe, L., 2001. Predicting divalent metal sorption to hydrous Al, Fe, and Mn oxides. Environmental Science & Technology 35 (9), 1779e1784. Trived, P., Axe, L., Tyson, T.A., 2001. XAS studies of Ni and Zn sorbed to hydrous manganese oxide. Environmental Science & Technology 35 (22), 4515e4521. U.S. Environmental Protection Agency, 2009a. National Primary Drinking Water Standards and National Secondary Drinking Water Standards. EPA 816-F-09-004. U.S. EPA, Office of Water, Washington DC. U.S. Environmental Protection Agency, 2009b. Final Third Contaminant Candidate List 3. EPA 815-F-09-001. U.S. EPA, Office of Water, Washington DC. Valentine, R.L., Stearns, S.W., 1994. Radon release from water distribution system deposits. Environmental Science & Technology 28 (3), 534e537. Vreeburg, J.H.G., Blokker, E.J.M., Horst, P., Van Dijk, J.C., 2009. Velocity-based self-cleaning residential drinking water distribution systems. Water Science and Technology: Water Supply 9 (6), 635e641. Waychunas, G.A., Rea, B.A., Fuller, C.C., Davis, J.A., 1993. Surface chemistry of ferrihydrite: part 1. EXAFS studies of the geometry of coprecipitated and adsorbed arsenate. Geochimica et Cosmochimica Acta 57 (10), 2251e2269. Wilkie, J., Hering, J.G., 1996. Adsorption of arsenic onto hydrous ferric oxide: effects of adsorbate/adsorbent ratios and cooccurring solutes. Colloids and Surfaces A 107, 97e110. Zhou, P., Gu, B., 2005. Extraction of oxidized and reduced forms of uranium from contaminated soils: effects of carbonate concentration and pH. Environmental Science & Technology 39 (12), 4435e4440.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
QMRAspot: A tool for Quantitative Microbial Risk Assessment from surface water to potable water Jack F. Schijven a,*, Peter F.M. Teunis b, Saskia A. Rutjes c, Martijn Bouwknegt c, Ana Maria de Roda Husman c a
National Institute for Public Health and the Environment, Expert Centre for Methodology and Information Services, PO Box 1, 3720 BA, Bilthoven, The Netherlands b National Institute for Public Health and the Environment, Epidemiology and Surveillance, The Netherlands c National Institute for Public Health and the Environment, Laboratory for Zoonoses and Environmental Microbiology, The Netherlands
article info
abstract
Article history:
In the Netherlands, a health based target for microbially safe drinking water is set at less
Received 9 March 2011
than one infection per 10,000 persons per year. For the assessment of the microbial safety
Received in revised form
of drinking water, Dutch drinking water suppliers must conduct a Quantitative Microbial
1 August 2011
Risk Assessment (QMRA) at least every three years for the so-called index pathogens
Accepted 12 August 2011
enterovirus, Campylobacter, Cryptosporidium and Giardia. In order to collect raw data in the
Available online 23 August 2011
proper format and to automate the process of QMRA, an interactive user-friendly computational tool, QMRAspot, was developed to analyze and conduct QMRA for
Keywords:
drinking water produced from surface water. This paper gives a description of the raw data
Quantitative Microbial Risk
requirements for QMRA as well as a functional description of the tool. No extensive prior
Assessment
knowledge about QMRA modeling is required by the user, because QMRAspot provides
Tool
guidance to the user on the quantity, type and format of raw data and performs a complete
Drinking water
analysis of the raw data to yield a risk outcome for drinking water consumption that can be
Index pathogen
compared with other production locations, a legislative standard or an acceptable health based target. The uniform approach promotes proper collection and usage of raw data and, warrants quality of the risk assessment as well as enhances efficiency, i.e., less time is required. QMRAspot may facilitate QMRA for drinking water suppliers worldwide. The tool aids policy makers and other involved parties in formulating mitigation strategies, and prioritization and evaluation of effective preventive measures as integral part of water safety plans. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
1.1.
General
Drinking water may be derived from surface water that is often contaminated with human pathogens and that may cause disease upon ingestion (Leclerc et al., 2002). Applied treatment processes may efficiently remove such microbial
contamination depending on water, treatment and pathogen characteristics. Quantitative data on pathogen concentrations in the source water and on the efficiency of the treatment processes are required to assess the risks from drinking water consumption (ILSI, 1996; WHO, 2011). The World Health Organization (WHO) Guidelines for Drinking water Quality (WHO, 2011) outline a preventive management framework for safe drinking water entailing health based targets, system
* Corresponding author. Tel.: þ31 30 274 2994; fax: þ31 30 274 4434. E-mail address: [email protected] (J.F. Schijven). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.024
5565
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
assessment from source through treatment to the point of consumption, operational monitoring of the control measures in the drinking water production, management plans documenting the system assessment and monitoring plans and a system of independent surveillance that verifies that the above are operating properly.
1.2.
Dutch Drinking Water Act
In line with the WHO Guidelines, the Dutch Drinking Water Act of 2001 (Anonymous, 2001) requires risk assessment for waterborne pathogens to demonstrate microbially safe drinking water. The infection risk is set at complying with a provisional health based target of less than one infection per 10,000 individuals per year. If the assessed infection risk exceeds this target value, the drinking water company must consult with the national environmental inspector about necessary measures.
1.3.
Dutch Inspectorate Guideline 5318
In the Dutch Drinking Water Act, no specific directives were given on how to perform risk assessment. Therefore, the Inspectorate Guideline 5318 for the assessment of the microbial safety of drinking water (Anonymous, 2005) was drafted in close consultation between the government (Environmental Inspectorate), the National Institute of Public Health and the Environment (RIVM), Bilthoven, the Netherlands and the drinking water industry. Inspectorate Guideline 5318 specifies the demand for safe drinking water by ways of Quantitative Microbial Risk Assessment (QMRA) primarily using production locationspecific data and (inter)national knowledge mainly on infectivity of pathogenic microorganisms and treatment efficiency. Risk assessment for exposure to pathogenic microorganisms in drinking water was described by Teunis et al. (1997), Haas et al. (1999), Haas and Eisenberg (2001), the ILSI framework (Benford, 2001) and Medema et al. (2003). These publications and development of risk based approaches by the WHO (2011) served as the basis for the specification of QMRA in Inspectorate Guideline 5318. It is emphasized that risk assessment is an iterative process which is directed by practical and theoretical progress.
believed that if a drinking treatment is effective in removing these index pathogens, adequate safety is warranted against other waterborne pathogens.
1.5.
Monitoring of the surface water should be aimed at achieving a representative quantification of the numbers of pathogenic microorganisms in the source water, considering seasonal variability as well as short term fluctuations of pathogen concentrations (Westrell et al., 2006b). According to Inspectorate Guideline 5318, a QMRA needs to be repeated at least every three years and three years of monitoring for pathogens may be condensed into one year to allow a higher monitoring frequency and hence provide more information about variability in pathogen concentrations. Inspectorate Guideline 5318 defined the monitoring frequency for source waters as dependent on the drinking water production volume (Table 1). In addition to regular monitoring, a number of incidental samples must be collected and analyzed at so-called peak moments, when peak concentrations in pathogen counts are assumed to occur, for example due to heavy rainfall (Kistemann et al., 2002; Kay et al., 2007).
1.6.
Treatment efficiency
A treatment or production site should be designed, organized and operated such that under any circumstance safe drinking water is produced. In that regard, the so-called multiple barrier principle applies: at times or instances when a treatment stage performs less well the other treatment stages should still provide adequate removal in order to produce safe drinking water (WHO, 2011). To quantify the efficiency of the treatment, data should be collected for each drinking water production location, because, commonly treatment efficiency is highly locationspecific. Any changes in the treatment process require new collection of data. Monitoring programs for determining treatment efficiency should also account for any temporal variability (Kistemann et al., 2002; Westrell et al., 2006b).
1.7. 1.4.
Surface water monitoring
Indicator organisms
Index pathogens
The Dutch Drinking Water Act states that microorganisms should not be present in drinking water at such concentrations that they form a threat to public health (Anonymous, 2001). This demand essentially concerns all waterborne pathogens. Because it is not feasible to monitor each waterborne pathogen, four so-called index pathogens were selected to represent waterborne pathogenic viruses, bacteria and parasitic protozoa (Anonymous, 2005). The selection criteria were prevalence, disease outcome, and possibilities for prevention and/or treatment. Obviously, waterborne transmission should be a significant pathway for these pathogens. In addition, a detection method, possible pathogen sources and efficiency of water treatment processes have to be established. This way, enterovirus, Campylobacter, Cryptosporidium and Giardia were selected as the index pathogens. It is
Commonly, concentrations of index pathogens will decrease below detection limits by drinking water treatment. However, so-called indicator organisms that are assumed to have similar properties as the index pathogens, so that they are
Table 1 e Number of samples per period of three years for each index pathogen according to the drinking water production of a drinking water treatment location (Anonymous, 2005). Production (m3/day) <10,000 10,000e100,000 >100,000
Regular
Incidental
Total
6 (one every eight weeks) 13 (one every four weeks) 26 (one every two weeks)
3 6 9
9 19 35
5566
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
removed equally or less well by drinking water treatment, are used to characterize treatment. Appropriate indicator organisms not only resemble the corresponding index pathogens in their fate and behavior in water, but also occur in higher numbers and are easier to enumerate with higher recovery. Nevertheless, after each treatment stage, indicator numbers decrease and in many cases drop below detection limits as found for index pathogens. Inspectorate Guideline 5318 (Anonymous, 2005) prescribes by default F-specific or somatic bacteriophages as the indicator organisms for determining removal efficiency by drinking water treatment of enterovirus. Similarly, Escherichia coli is set as the indicator organism for Campylobacter and spores of sulphite reducing clostridia (SSRC) for both Cryptosporidium and Giardia.
2.
Data
1.8.
2.1.
Raw data
Data analysis and new QMRAs
The iterative process of risk assessment in the Netherlands started with the enforcement of Inspectorate Guideline 5318 in 2006 at the five Dutch drinking water companies that treat surface water for drinking water production at fourteen locations. For preparation of this first QMRA, they not only collected all available microbiological data on their source waters and on removal efficiency of their treatment, but also provided detailed descriptions on the dimensions and characteristics of all treatment processes. All this information was documented in reports that were sent to the Environmental Inspectorate, who forwarded these reports to the National Institute of Public Health and the Environment, Bilthoven, the Netherlands (RIVM). Using the reported data, the actual QMRA was conducted at RIVM. RIVM reported the outcomes of the QMRA to the Environmental Inspectorate encompassing conclusions on compliance with the allowed maximum infection risk as well as on the completeness of the data. Outcomes were also discussed in meetings with each drinking water company, the environmental inspector and RIVM. The outcomes of this study are summarized by Schijven and de Roda Husman (2009). Extracting all historic data from the first QMRA reports assembled by the Dutch drinking water companies as well as evaluating those reports appeared to be very laborious (Schijven and de Roda Husman, 2009), which led to the desire to process and analyze new data in a more efficient and standardized manner with a risk outcome in relation to the legislative health based target. To that aim, a spreadsheet was designed to be used by the drinking water companies for entering the data from the next round of monitoring and a computational tool was designed to automatically analyze the data and conduct the QMRA.
indicator organism data for determining drinking water treatment efficiency, all provided in a standard format; - Automatic analysis of those data, entailing fitting of appropriate probability distributions to quantify concentrations and treatment efficiencies; - A uniform and transparent way of conducting QMRA; - A risk outcome in relation to the legislative health based target. By describing the tool, this paper is intended to also serve as a user and reference manual of how to conduct QMRA for drinking water.
For a QMRA, it is essential to collect quantitative microbial data as raw unprocessed data. Raw data on enumerated microorganisms in water are the counted numbers of the microorganisms as well as the corresponding investigated volume of the sample. Commonly, counts are numbers of plaque-forming units (pfu) for viruses and colony-forming units (cfu) for bacteria (Schets et al., 2008; Teunis et al., 2005a). Oocysts of Cryptosporidium and cysts of Giardia may be counted manually or automatically under a microscope using fluorescent dye (Schets et al., 2008). Raw presence/absence data are presence or absence of micoorganisms in replicate dilutions of a water sample (De Roda Husman et al., 2009). Obviously, the concentration of, for example, 1 pathogen particle in 1 mL of water is the same as a 100 particles in 100 mL of water, but if 100 particles were observed that count is more accurate, hence it is essential to use counts and sample volumes as they were observed and not concentrations (ratios of count/volume). All raw data must include a sample date.
2.2.
Source water data are raw data of index pathogens in the source water (e.g. Rutjes et al., 2009; Lodder et al., 2010). At many production locations, river water first passes a storage reservoir before further treatment. For enterovirus, river water may appropriately be designated as source water. However, for Campylobacter, Cryptosporidium and Giardia, the storage reservoir should be considered the starting point of the QMRA, because of contamination of the storage reservoir water with Campylobacter from birds, wildlife, or runoff from agricultural land.
2.3. 1.9.
Source water data
Recovery data
Objectives
This paper presents and describes an interactive user-friendly computational tool, named QMRAspot, that was developed to analyze and conduct QMRA for a drinking water production chain from surface water to potable water. The objectives of developing this tool are the following: - Collection and automated reading of raw microbial data, entailing index pathogen data from source water and
In order to determine the recovery efficiencies of the detection method for index pathogens, ideally, each sample of source water, or a fraction used for analysis, is spiked with a sufficiently high number of, for example, a specific type of indicator organism. The spiked and recovered numbers can then be used to estimate the recovery efficiency. Therefore, raw recovery data consist of counts and samples sizes of the spiked and recovered microorganisms that are paired according to sampling date (Teunis et al., 1999; Rutjes and de
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Roda Husman, 2004; Schets et al., 2004). If the source water sample contains sufficiently high numbers of indicator organisms, like bacteriophages, to allow direct counting after plating, these indicator organisms can be used to estimate the recovery efficiency of a concentrating step needed for detection of the associated pathogen.
2.4.
Treatment data
Raw data of indicator organisms for a treatment process consist of counts and sample sizes of samples taken before and after treatment. If available, raw data of index pathogens can be used.
2.5.
Consumption data
QMRAspot offers four choices for consumption of unboiled drinking water per person per day. These are all lognormal distributions, with parameters defined by various studies. Parameters m ¼ 1085779 and s ¼ 1.07487 are for the Netherlands, corresponding to a mean of 0.27 L per person per day (Teunis et al., 1997), a lognormal distribution with parameters m ¼ 0.03598 and s ¼ 0.77218 for the USA, corresponding to a mean of 1.3 L per person per day (USEPA, 2006), a fixed volume of 2 L per person per day (WHO, 2011) and, finally, the possibility of putting in the parameter values for m and s for any other log normally distributed consumption data, if available for another country or for a specific subpopulation. Consumption data may differ between countries and also between subpopulations; climate may also play a role. For more data and a discussion about such variability, the reader is referred to USEPA (2006), Westrell et al. (2006a) and WHO (2011).
2.6.
Dose response data
The dose response relation for rotavirus has been published (Teunis et al., 1996), as well as for Giardia and Cryptosporidium (Teunis et al., 1996, 2002a, b). The dose response relation of rotavirus, an enteric virus, is applied as a worst case for virus infectivity (Regli et al., 1991). The hierarchical dose response relation for three isolates of Cryptosporidium parvum has been updated to include two additional isolates for which challenge studies have been published (Okhuysen et al., 2002). For Campylobacter, the human challenge dose response study updated with outbreak data has been used (Teunis et al., 2005b). For any index pathogen the same dose response data are used for each QMRA simulation.
3.
Standard data file: QMRAdata.xls
The tool reads the raw data from a standard Microsoft Excel spreadsheet file, here, for convenience, named QMRAdata.xls.
L r;
1 1 þ lVi
It contains three sheets: SCHEME, RAW DATA and HELP. The SCHEME sheet provides a description of the drinking water production location and defines a table with column headers that is used by the tool to make the appropriate data selections. Through the SCHEME sheet, the user has control over what data should be used for QMRA. Obviously, the RAW DATA sheet contains all raw data and the HELP sheet provides background information on how to fill the RAW DATA sheet with raw data in the required format. Table 2 shows a summarized setup of the SCHEME sheet. In the SCHEME sheet, the effluent data of a treatment step may be the influent data for the next treatment step. This need not be so if the data for the next treatment step concern other types of microorganisms, or were from a different location in the drinking water utility, or from pilot plant experiments. The SCHEME sheet can be modified easily. For example, two treatment steps may be combined using the influent data of the first of the two and the effluent data of the second of the two treatment steps. In the RAW DATA sheet, every row is a full record of raw data (Table 3). This simple design allows for automated filling from a Laboratory Information Management System (LIMS), for example, as records in the form of Comma Separated Values (CSV). The source of the data for the treatment steps can be selected from the following list: Plant scale, pilot plant scale, laboratory scale. According to Inspectorate Guideline 5318 (Anonymous, 2005), location-specific plant scale data are generally preferred, followed by pilot plant scale data, and if these are not available, data from laboratory experiments. In other words, location-specific data are recommended. If the use of data from other locations is desired, applicability should be verified by comparison of treatment conditions. References to data from literature should be listed in the accompanying QMRA reports.
4.
Fitting of distributions to the data
4.1.
General
All raw data sets should include three or more samples: smaller data sets are ignored and parameters are not estimated. Counts in QMRAdata.xls may only be integers.
4.2.
Source water concentration
Based on the assumption that counts n within each sample of size V are Poisson distributed, while the concentration (Poisson parameter) is gamma-distributed among samples, the counts n1 .nN of N samples with samples sizes V1 .VN have a Negative Binomial distribution (NegBin) with parameters r and 1=1 þ lVi (Teunis et al., 2009). Parameters r and l are estimated by minimizing the following deviance function:
0
11 B CC 1 B CC ¼ 2log B @f ðrÞNegBin@ni ; Vi ExpðlnrÞ; AA V ExpðlnlÞ i i¼1 1þ s 0
n B Y
5567
(1)
5568
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Table 2 e Summary of the SCHEME sheet in QMRAdata.xls. Row\column
A
1 2 3, 13, 23, 33
B
C
Drinking water utility Production location Index pathogen
Drinking water utility name Production location name Enterovirus, Campylobacter, Cryptosporidium, Giardia Source water name for each index pathogen Recovery indicator name: e F-specific RNA bacteriophage, somatic coliphage or enterovirus e E. coli or Campylobacter e SSRC or Cryptosporidium e SSRC or Giardia Treatment indicator Treatment indicator name: e F-specific RNA bacteriophage, somatic coliphage or enterovirus e E. coli or Campylobacter e SSRC or Cryptosporidium e SSRC or Giardia
4, 14, 24, 34
Source water
5, 15, 25, 35
Recovery indicator
6, 16, 26, 36 7e12 17e22 27e32 37e42
z1 .z6 for each of the four index pathogens
Treatment Treatment step names
where f ðrÞ ¼ 1 Fððx 20Þ=20Þ is a prior function for the shape factor r, with Fððx 20Þ=20Þ the cumulative normal distribution with a mean and variance of 20. This prior prevents extremely small values of r and facilitates robust parameter estimation without strongly affecting the estimates. s is a scaling factor. If the mean sample volume is less than 1 L, then s ¼ 0.001. This scaling avoids computational underflows. In QMRAdata.xls presence/absence data may be given for any microorganism, although this is usually only the case for
D
E
Spike
Recovery
Spike name
Recovery name
Influent Influent name of each treatment step
Effluent Effluent name of each treatment step
F
Data source For each treatment step: e Plant scale e Pilot plant scale e Laboratory scale
Campylobacter. These observations are used to calculate a concentration for each sample by minimizing the following deviance function: Lðc; VÞ ¼ 2log
n Y ni ¼ 0 0 Poisð0jcVi Þ ni > 0 0 1 Poisð0jcVi Þ
(2)
i¼1
where Pois denotes Poisson distribution, c is concentration and Vi is the sample size of the i-th sample.
Table 3 e The RAW DATA sheet in QMRAdata.xls. Column
Name
A B
Name Sampling code
C D E
Microorganism Date Count
F
Sample volume
G/J/M/P/S H/K/N/Q/T
V1/V2/V3/V4/V5 R1/R2/R3/R4/R5
I/L/O/R/U
MPN1/MPN2/MPN3/ MPN4/MPN5
Description Name of source water, "Spike", "Recovery", treatment influent, treatment effluent. Specific code included in RAW DATA for reference. Different codes may be included for different sampling points of the same influent or effluent. In the QMRA the data designated by different codes are combined. Index pathogens and indicator microorganisms. Date format: DD-MM-YY. Only whole positive numbers (integers) allowed. Maximum counts per plate according to the standard method. The actual sample volume (liter) used for counting of the microorganisms. Example 1: a 10 L sample was collected, concentrated to 100 mL, 5 ml was plated for counting, then the sample size is 10/20 ¼ 0.5 L. Example 2: 98 colonies were counted on a plate with 1 ml sample and 11 colonies were counted in the ten-fold dilution. Count is 109 and sample size is 0.0011 L V1.5 are the sample volumes (liter) of five dilution steps in the MPN scheme. R1.5 are the replicate numbers for each dilution step in the MPN scheme. Only whole numbers allowed. MPN1.5 are the Most Probable Numbers in the MPN scheme. Only whole numbers from 0 - the replicate number allowed. The tool does not use available MPN data if counts (column E) and sample sizes (column F) are available.
5569
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Subsequently, a Gamma distribution with parameters r and l is fitted to the concentration data by minimizing the following deviance function:
0 n Y Lðr; lÞ ¼ 2log @ i¼1
1 2 ci ¼ N 0 Gammacdf Lðc; ln l VÞ ¼ c95%2 ðdf ¼ 1Þ lnr; ci ¼ 0 0 1 Gammacdf Lðc; VÞ ¼ c95% ðdf ¼ 1Þlnr; ln l A 0 < ci < N 0 Gammaðlnr; ln lÞ
where ci is the concentration of the i-th sample. Gammacdf denotes a cumulative Gamma distribution. Lðc; VÞ ¼ c295% ðdf ¼ 1Þ is the root of the likelihood function equal to the 95-percentile of a X -squared distribution with one degree of freedom (df). If raw data of the index pathogen in the source water consist of nondetects only, a Gamma distribution with P parameters r ¼ 0.01 and l ¼ 1=ð Vi þ 1=100Þ is assumed.
4.3.
Recovery, R
In order to estimate the recovery efficiency of the detection method of the index pathogen, samples are spiked with a known number of the specified indicator organisms. After processing of the samples, a fraction of the spiked organisms will be recovered. The data on the initial spike and on the recovery are paired. The recovered fraction is assumed Betadistributed with parameters a and b. Estimation of these parameters by means of the paired Beta model is explained in detail by Teunis et al. (1999, 2009). If recovery data are lacking, then it is assumed that R ¼ 1.
4.4.
For the unboiled drinking water consumption, W liter, MC samples of lognormal distributions are generated, but in case of the WHO-data set, a fixed value of 2 L is used.
Treatment, z
Monte Carlo samples of the dose response parameters a and b are provided as pregenerated data and included in the tool in a packed form to save memory space.
6.
QMRAspot tool code
QMRAspot has been developed in Mathematica 8.0.0 (Wolfram, Inc., Champaing IL, USA). It can be run in as a Dynamic Module in Mathematica Player Pro versions 7 and 8. When the calculations are complete, the user can page through the results. A QMRA report can be generated for each of the index pathogens and saved as a Mathematica notebook and/or pdf file. Fig. 1 shows the splash screen of the tool.
7.
Exposure and infection risk
Exposure to the index pathogens is given as the dose D, the number of ingested index pathogens per person per day and is calculated by multiplying the MC samples of source concentration Csource, recovery R, treatment zi and consumption data W:
Here, we assume that treatment is in effect, implying that microorganisms are removed and thus 0 z 1. It is assumed that microorganisms passing treatment do so independently with a probability or fraction z. This may be modeled as a binomial process, either with paired or unpaired samples (Teunis et al., 1999, 2009). Collection of paired data from a treatment step requires exact timing of the sampling. The pairing may be lost if mixing occurs during treatment. Residence times in treatment may vary from a few hours to several days. In many cases, even with short residence times and samples of influent and effluent collected on the same day, pairing is not evident. Estimation of the parameters a and b by means of the unpaired Beta model is explained in detail by Teunis et al. (1999, 2009). In case the effluent of a treatment stage produces only nondetects, it is still possible to evaluate treatment.
D ¼ Csource
5.
8.
Monte Carlo simulation
From all distributions, 10,000 Monte Carlo (MC) samples are generated. The source water concentration of the index pathogens, Csource, is Gamma-distributed with parameters r and l/s. For recovery R, and treatment steps z1 .z6 , Betadistributed MC samples are generated.
(3)
6 1Y zi W R i¼1
(4)
Infection risk per person per day is calculated by applying the dose response relation (Teunis and Havelaar, 2000): Pinf;person;day ¼ 1 1 F1 ða; a þ b; DÞ
(5)
where a and b are infectivity parameters that are pathogen specific and 1 F1 is the confluent hypergeometric function. Parameters a and b are MC sample pairs (joint distribution), reflecting variability of infectivity. Infection risk per person per year is calculated from MC samples of daily infection risk by applying Eq. (5) 10,000 times for each day in a year to obtain 365 MC sample distributions, which are then multiplied with each other (Teunis et al., 1997): Pinf;person;year ¼ 1
365 Y
1 Pinf;person;day;i
(6)
i¼1
Results
As examples, a number of distributions will be shown of source water concentrations, treatment and infection risks based on data from a number of Dutch drinking water suppliers (Figs. 2e4). These are the raw data for the second round of QMRA, where data were delivered in the standard
5570
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Fig. 1 e Splash screen of QMRA spot. Only a QMRAdata.xls file with raw data needs to be opened. After pushing the “Run QMRA”-button, a full analysis of the data will be done and a QMRA will be conducted.
format in QMRAdata.xls. The names of the drinking water locations are kept anonymous, because it is not the intention of this paper to present the QMRA results for specific drinking water locations. Using fictive data could provide unrealistic examples. The presented data are a combination of data from different suppliers to form a fictive drinking water production location. Figs. 2e4 show the full QMRA for Cryptosporidium as example. Fig. 5 gives the box-whisker plots for the infection risk per person per year for all four index pathogens. Table 4 summarizes all QMRA steps for all four index pathogens by giving mean and 95-percentile values of all distributions on log10 scale, as well as the distribution parameter values.
8.1.
which appear to be left-skewed on a logarithmic scale. This was also observed for enterovirus by Teunis et al. (2009). For reference, mean concentration and 95-percentiles of these distributions are included in the time plot. Peak concentrations may be defined as those concentrations above the 95-percentile. In winter, lower temperatures and increased precipitation lead to shorter residence times between wastewater discharge locations and intake points for drinking water production, resulting in less inactivation than during summer (Schijven et al., 1996; Schijven and de Roda Husman, 2005). Note that for the index pathogens, except Campylobacter, in the majority of the samples, no index pathogens were detected, resulting in low mean concentrations (Table 4).
Index pathogen concentrations in source water 8.2.
QMRAspot provides time plots for inspecting variations of concentrations over time (Fig. 2). The histograms illustrate MC samples of the Gamma-distributed concentrations,
Recovery
Fig. 2 includes recovery data for Cryptosporidium. Given these observations, the need for including recovery data into QMRA
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
5571
Fig. 2 e QMRA of Cryptosporidium. Time plot of measured source water concentrations and histograms of MC samplings of fitted distributions to source water concentrations and recovery data, and histogram of calculated for recovery corrected source water concentration.
may be clear. In the Netherlands some data on recovery of index pathogens are available (e.g. Rutjes and de Roda Husman, 2004; Schets et al., 2004), but location-specific data on recovery of index pathogens are usually lacking. Moreover, recoveries may vary between samples, necessitating estimation of recovery efficiency for every single sample. Such an approach has been used for detection of Cryptosporidum oocysts and Giardia cysts (Quintero-Betancourt et al., 2003; Ferguson et al., 2004).
8.3.
Treatment
Fig. 3 shows the estimated removal of SSRC as indicator for removal of Cryptosporidium (and Giardia). Time plots of the influent and effluent concentrations are shown. The Beta distributions of the z-values on logarithmic scale are generally left-skewed. In this example, the effluent data of Z1 are the influent data for Z2. For all Z1 and Z2 influent and effluent samples, SSRC concentrations were above detection limit. For
Fig. 3 e QMRA of Cryptosporidium. Time plots of measured source influent and effluent concentrations of SSRC for treatment steps z1, z2 and z3 and histograms of MC samplings of fitted distributions the fractions of SSRC that pass treatment.
5572
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Fig. 4 e QMRA of Cryptosporidium. Histograms of total treatment and drinking water concentration as calculated from the MC samples of treatment steps z1, z2 and z3 and the source water concentration. Histogram of MC samples from the Dutch drinking water consumption data. Histogram of exposure to Cryptosporidium as calculated from the drinking water concentration and consumption MC samples.
Z3 a data set from another drinking water production location was used, showing a large number of samples, of which in the effluent, only a few samples had concentrations above the detection limit.
8.4.
Exposure
Fig. 4 shows the MC-histograms of SSRC removal from multiplying the MC-data of all treatment steps, the MChistograms of the drinking water concentrations as calculated from the estimated source water concentrations, recovery and treatment, the MC-histogram of the lognormal distributed drinking water consumption, and, finally, the MChistogram of the exposure or dose as calculated from the estimated drinking water concentrations and drinking water consumption data.
8.5.
Infection risk
Fig. 5 shows box-whisker plots of the annual infection risk for each of the index pathogens, with arithmetic mean (horizontal
line), quartiles (box) and 95% interval (whiskers). If the mean infection risk is above the target value of 104 per person per year, as in this example for Campylobacter, the box-whisker plot is red. If only the 95-percentile exceeds the target value, the boxwhisker plot is yellow, as is the case for Cryptosporidium. In all other cases the box-whisker plot is green. This is to emphasize that it is highly recommended that at least 95% of the time the infection risk is below the health based target value.
9.
Discussion
9.1.
Raw data
QMRAspot imposes no restrictions on microbial count ranges. Detection methods are restricted to a maximum count, usually 100 microorganisms per plate. Higher counts suffer large systematic errors because of nutrient exhaustion, overlapping colonies or plaques, or a higher probability that colonies or plaques did not originate from a single bacterial cell or virus particle (Teunis et al., 2005a). Counts per sample could
Fig. 5 e Box-whisker plots of the annual infection risk for each index pathogen, with mean (line), quartiles (box) and 95% interval (whiskers).
5573
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Table 4 e Summary of QMRA steps for all four index pathogens for a fictitious drinking water production location with three subsequent treatments Z1, Z2 and Z3. QMRA step Enterovirus Source water concentration Z1 Z2 Total treatment (somatic coliphage) Drinking water concentration Consumption Exposure Infection risk Infection risk Campylobacter Source water concentration Z1 Z2 Z3 Total treatment (E. coli) Drinking water concentration Consumption Exposure Infection risk Infection risk Cryptosporidium Source water concentration Recovery Corrected source water concentration Z1 Z2 Z3 Total treatment (SSRC) Drinking water concentration Consumption Exposure Infection risk Infection risk Giardia Source water concentration Total treatment (SSRC) Drinking water concentration Consumption Exposure Infection risk Infection risk
Dimensions
Log10 mean
Log10 95%
Distribution
N/litre Fraction Fraction Fraction N/litre Litre/person/day N/person/day /person/day /person/year
3.6 0.86 2.3 3.2 6.8 0.55 7.3 7.7 5.1
3.0 0.48 1.6 2.5 6.2 0.051 6.9 7.2 4.9
Gamma(0.63; 0.00044) Beta(1.7; 10) Beta(0.086; 18)
N/litre Fraction Fraction Fraction Fraction N/litre Litre/person/day N/person/day /person/day /person/year
2.1 0.64 2.3 4.8 1.9 2.9 0.55 3.5 3.9 1.3
2.8 0.26 1.5 1.1 5.0 3.4 0.051 4.1 4.8 0.81
N/litre Fraction N/litre Fraction Fraction Fraction Fraction N/litre Litre/person/day N/person/day /person/day /person/year
2.7 0.77 1.7 0.46 1.8 1.5 3.8 5.7 0.55 6.3 7 4.4
2.1 0.47 1.2 0.16 1.2 0.69 3.2 7 0.051 7.9 8.6 3.8
Gamma(0.033; 0.069) Beta(2.7; 13)
N/litre Fraction N/litre Litre/person/day N/person/day /person/day /person/year
2.9 3.8 6.7 0.55 7.4 9.1 6.5
2.2 3.2 6.7 0.051 7.4 9.1 6.1
Gamma(0.21; 0.0058)
Lognormal(1.9; 1.1)
Gamma(0.34; 380) Beta(1.2; 4.1) Beta(0.099; 21) Beta(0.068; 5.5)
Lognormal(1.9; 1.1)
Beta(1.9; 3.6) Beta(0.42; 28) Beta(0.076; 2.5)
Lognormal(1.9; 1.1)
nevertheless be high numbers if replicate plates were counted and totaled. High counts with trailing zeroes, for example 2200, are doubtful and may be processed data, for example, extrapolations of 22 in a 100 times diluted sample. Often, low counts per plate are assumed to be unreliable and discarded. Although accuracy of estimated concentrations may be low, the method in QMRAspot for fitting a distribution accounts for such uncertainty and discarding information should be discouraged. Ideally, volumes of source water should be large enough to enable detection of index pathogens. In some cases, analysis of a 10 L sample of source water is sufficient whereas other locations may require a 100- to 2000 L sample.
pathogen concentrations in source water for QMRA should be a continuous activity in order to evaluate possible trends, for example, due to climate changes or local changes such as in discharges of wastewater. Also, regular monitoring programs are inherently limited and may often not capture peak events; however, by continuating monitoring programs for index pathogen concentrations in source water, gradually more insight and knowledge on peak events will be collected (Kistemann et al., 2002; Kay et al., 2007). This may lead to sufficient data that enable adaptive dynamic filtering for early warning of peak concentrations (Westrell et al., 2006b) and timely and appropriate management actions may be undertaken to prevent waterborne disease.
9.2.
9.3.
Index pathogen concentrations in source water
In discussions with representatives of the drinking water companies, it became apparent that collection of index
Recovery, infectivity and typing
Location-specific data on the recovery of detecting index pathogens in the source water are still sparse. Recoveries may
5574
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
vary widely, especially for the parasitic protozoa (Schets et al., 2004), and, therefore, can have a strong impact on the risk assessment outcome. It should also be noted that recovery may not only be dependent on the detection method, the pathogen and the type of water, but also on microbial counts (Schmidt and Emelko, 2011). Schmidt and Emelko (2011) demonstrated this with a model that describes variability in microorganism counts as a function of sample volume and the analytical recovery of the enumeration method and that was expanded to include temporal concentration variability and sample-specific recovery information. Detection methods vary widely from counting of enterovirus plaques in a cell culture assay, determining most probable numbers of Campylobacter, to microscopic viability counting of Cryptosporidium oocysts and Giardia cysts. The performance characteristics of detection methods lead to specific meanings of the risk estimates regarding the infectivity of the counted microorganisms. This should be accounted for when comparing risks between different waterborne pathogens. More thought should go into the meaning of risk estimates in the light of newly developed, rapid molecular techniques such as PCR for significant waterborne pathogens, such as, for example, noroviruses that cannot be determined with the use of the abovementioned detection methods (Laverick et al., 2004). QMRAspot should be adapted to not only account for recovery but also for the infectious fraction of counted index pathogens. To correctly assess an infection risk for humans from drinking water consumption, one should limit quantification to human pathogens. Some pathogens may be zoonotic with an animal origin but nevertheless infectious to humans. If a substantial fraction of the pathogen is of animal origin, one may want to determine the contribution of the animal sources relative to human sources to the infection risk estimate. In this regard, quantitative microbial source tracking comes into play (Schijven and de Roda Husman, 2011).
9.4.
Treatment
If a treatment step is well-characterized and the process conditions remain the same, it is not necessary to collect new data before and after the treatment step for each new QMRA. Nevertheless, it is common practice for Dutch drinking water companies to constantly monitor treatment performance; hence it is obvious to include new data into a new QMRA. Moreover, information on treatment failure may accumulate over time. Obviously, if treatment process conditions have changed, which may include changes in source water quality or any of the other determinants of drinking water quality, a QMRA should be conducted to evaluate any effects of such changes. In that regard, specific treatment models that predict treatment efficiency for a range of process conditions are very useful and extensive monitoring to evaluate treatment efficiency under the altered conditions may not be necessary. Such a predictive model is in development for slow sand filtration (Schijven et al., 2008), which is not only applicable to altered process conditions, but also for other production locations. The default indicator organisms may not always be the best choice for representing the removal of the index
pathogens. For example, parasitic protozoa are removed much better by slow sand filtration than SSRC (Hijnen et al., 2007). However, currently, for most treatment processes, the abovementioned index organisms are used by default, awaiting research that provides data for more appropriate index organisms in particular cases. Although QMRA based on location-specific raw data for treatment at full scale is by far to be preferred, the tool provides the option of including distribution parameters values of fraction z of the microorganisms that were able to pass treatment instead of raw data. This option is not included to move away from collecting raw data, but often locationspecific data at plant scale are not available, because indicator organism levels were (expected to be) below detection limits. This often occurs with very efficient treatment steps and/or at the end of the production chain. In those cases, one has to rely on data from pilot plant experiments that mimic full scale conditions, or on data from laboratory scale experiments, or use treatment data that were collected at other plants operating under similar conditions. In all those cases, the applicability of the data needs to be verified. The option of including a treatment step by means of its distribution parameters can also be used to determine the required additional treatment if a drinking water location exceeds the health based targets. This option allows for scenario studies, and therefore, greatly increases the versatility of QMRAspot.
9.5.
Consumption data
QMRAspot provides the option of defining distribution parameter values for any set of consumption data. Also here, over time, a database of consumption data for different countries may emerge. It should be noted that in the current version of QMRAspot, the choice for drinking water consumption data encompasses the range of mean drinking water consumption per person per day from about a quarter of a liter to 2 L, which may cover all countries (Westrell et al., 2006a).
9.6.
Dose response relation
Applied dose response relations were generally derived from studies in which a specific strain of the index pathogen was given to human volunteers (Teunis et al., 1996, 2002a, 2002b). However, one pathogen strain does not represent the suite of strains that may occur in source waters for drinking water production. A hierarchical dose response relation, as was performed for multiple isolates of C. parvum, produced estimates that differed very much between isolates (Teunis et al., 2002a). Predictions based on multilevel dose response relations may aid probabilistic risk assessments such as presented here to properly reflect the variation in pathogen strains. Moreover, dose response data from outbreaks may inform the dose response relation as was shown for Campylobacter (Teunis et al., 2005b). Such additions, both hierarchical analysis and the use of outbreak data, could aid the estimation of the enterovirus dose response relation for which now the rotavirus dose response relation is used. For
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
such analysis data are currently not available, and additional research is required.
9.7.
Experience of Dutch drinking water companies
After the first round of QMRA using historic data, the Dutch drinking water companies reported their experiences with conducting QMRA at a workshop. There was a general consensus about the benefits of QMRA (Schijven and de Roda Husman, 2009): The QMRA framework provided integral insight into the robustness of the drinking water treatment. It enabled identification of weak links in the treatment and what additional or improved treatment would be needed. It also facilitated communication between management and operators, and, therefore, provided a basis for the appropriate implementation of Water Safety Plans (Summerill et al., 2010).
9.8.
Future developments
Because treatment efficiencies are described generically by a Beta distribution, the launch of the QMRAspot tool intends to initiate a database with treatment data in the form of the parameters of the Beta distribution for any specific indicator organism under any specific treatment process conditions. Such an effort was started for enterovirus removal by Teunis et al. (2009). Currently, the QMRA ends with finished water, i.e. effects of the distribution system are not included. Pathogen intrusion and interaction of pathogens with biofilms in distribution systems, for example, as described for viruses by Skraber et al. (2005), need to be evaluated for their significance to infection risks. QMRAspot is based on the Dutch Inspectorate Guideline 5318 (Anonymous, 2005) including the four index pathogens enterovirus, Campylobacter, Cryptosporidium and Giardia. Given the need to prioritize for emerging pathogens, this may change in the future and require extending the tool with QMRA of other pathogens.
10.
Conclusions
An interactive user-friendly computational tool, QMRAspot, has been developed for use without extensive prior knowledge about QMRA modeling to estimate a risk outcome for consumption of drinking water produced from surface water. This risk outcome for drinking water consumption can be compared to those of other production locations, to a legislative standard or to a generally acceptable health based target. Because of the ease of use and the standardization of data formats, the tool not only facilitates QMRA for Dutch drinking water companies, but, in fact, for any drinking water company or other interested party such as a policy maker, inspector or consumer. User guidance is provided on how to structure the required raw microbial data. Then, by a simple push on the button, data are analyzed and a QMRA report is produced for four index pathogens in less than a few minutes. This may be described as a quick-and-clean (not dirty) risk assessment. Although a QMRA based on actual raw data is highly preferred, it is also possible to conduct a QMRA using distribution parameters for pathogen concentrations in the source
5575
water, for each treatment step, and for drinking water consumption. Such parameters values may be taken from literature (e.g. Teunis et al., 2009). Moreover, QMRAspot can be used to estimate the distribution parameters for a specific (additional) treatment step to achieve the required pathogen removal. Generic information can be applied to define scenarios to answer a variety of what-if questions. QMRAspot has a strong educative character by providing guidance on how to structure data collection, analysis and risk assessment, how to report the results, and by aiding policy makers and decision makers in formulating mitigation strategies, preventive measures and prioritization of measures. The tool provides insight into the efficiency of the treatment steps. In all these regards, the tool facilitates risk communication.
Acknowledgements This work was performed by order and on the account of The Environmental Inspectorate, who is highly acknowledged for its contribution in defining of and observing compliance with Inspectorate Guideline 5318. The Dutch drinking water companies are highly acknowledged for their contribution with data and fruitful discussions as participants in the Infection Risk Working Group are the other participants.
references
Anonymous, 2001. Besluit van 9 januari 2001 tot wijziging van het waterleidingbesluit in verband met de richtlijn betreffende de kwaliteit van voor menselijke consumptie bestemd water. Staatsblad van het Koninkrijk der Nederlanden 31, 1e53. Anonymous, 2005. VROM-Inspectierichtlijn Analyse microbiologische veiligheid drinkwater. VROM-Inspectie. Artikelcode 5318. Benford, D., 2001. Principals of Risk Assessment of Food and Drinking Water Related to Human Health. ILSI Europe concise monograph SERIES. International Life Science Institute, Belgium, pp. 1e43. De Roda Husman, A.M., Lodder, W.J., Rutjes, S.A., Schijven, J.F., Teunis, P.F.M., 2009. Long-term inactivation study of three enteroviruses in artificial surface and groundwaters, using PCR and cell culture. Applied Environmental Microbiology 75 (4), 1050e1057. Ferguson, C., Kaucner, C., Krogh, M., Deere, D., Warnecke, M., 2004. Comparison of methods for the concentration of Cryptosporidium oocysts and Giardia cysts from raw waters. Canadian Journal of Microbiology 50, 657e682. Haas, C.N., Eisenberg, J.N.S., 2001. Risk assessment. In: Fewtrell, L., Bartram, J. (Eds.), Water Quality: Guidelines, Standards and Health. IWA Publishing, London, pp. 161e183. Haas, C.N., Rose, J.B., Gerba, C.P., 1999. Quantitative Microbial Risk Assessment. Wiley, New York, USA. Hijnen, W.A.M., Dullemont, Y.J., Schijven, J.F., HanzensBrouwer, A.J., Rosielle, M., Medema, G.J., 2007. Removal and fate of Cryptosporidium parvum, Clostridium perfringens and small-sized centric diatoms (Stephanodiscus hantzschii) in slow sand filters. Water Research 41 (10), 2151e2162. ILSI Risk Science Institute Pathogen Risk Assessment Working Group, 1996. A conceptual framework to assess the risks of human disease following exposure to pathogens. Risk Analysis 16, 841e848.
5576
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 6 4 e5 5 7 6
Kay, D., Watkins, J., Francis, C.A., Wyn-Jones, A.P., Stapleton, C.M., Fewtrell, L., Wyer, M.D., Drury, D., 2007. The microbiological quality of seven large commercial private water supplies in the United Kingdom. Journal of Water and Health 5 (4), 523e538. Kistemann, T., Classen, T., Koch, C., Dangendorf, F., Fischeder, R., Gebel, J., Vacata, V., Exner, M., 2002. Microbial load of drinking water reservoir tributaries during extreme rainfall and runoff. Applied Environmental Microbiology 68 (5), 2188e2197. Laverick, M.A., Wyn-Jones, A.P., Carter, M.J., 2004. Quantitative RT-PCR for the enumeration of noroviruses (Norwalk-like viruses) in water and sewage. Letters in Applied Microbiology 39 (2), 127e136. Leclerc, H., Schwartzbrod, L., Dei-Cas, E., 2002. Microbial agents associated with waterborne diseases. Critical Reviews in Microbiology 28 (4), 371e409. Lodder, W.J., van den Berg, H.H., Rutjes, S.A., de Roda Husman, A.M., 2010. Presence of enteric viruses in source waters for drinking water production in The Netherlands. Applied Environmental Microbiology 76 (17), 5965e5971. Medema, G.J., Hoogenboezem, W., van der Veer, A.J., Ketelaars, H.A.M., Hijnen, W.A.M., Nobel, P.J., 2003. Quantitative risk assessment of cryptosporidium in surface water treatment. Water Science & Technology 47, 241e247. Okhuysen, P.C., Rich, S.M., Chappell, C.L., Grimes, K.A., Widmer, G., Feng, X.C., Tzipori, S., 2002. Infectivity of a Cryptosporidium parvum isolate of cervine origin for healthy adults and interferon-gamma knockout mice. Journal of Infectious Diseases 185 (9), 1320e1325. Quintero-Betancourt, W., Gennaccaro, A.L., Scott, T.M., Rose, J.B., 2003. Assessment of methods for detection of infectious Cryptosporidium oocysts and Giardia cysts in reclaimed effluents. Applied Environmental Microbiology 69 (9), 5380e5388. Regli, S., Rose, J.B., Haas, C.N., Gerba, C.P., 1991. Modeling the risk from Giardia and viruses in drinking water. Journal AWWA 83 (11), 76e84. Rutjes, S.A. and de Roda Husman, A.M, 2004. Procedure for virus detection in water for use in the Dutch drinking-water guidelines. RIVM Report 330000007, 38 p. (in Dutch). Rutjes, S.A., Lodder, W.J., van Leeuwen, A.D., de Roda Husman, A. M., 2009. Detection of infectious rotavirus in naturally contaminated source waters for drinking water production. Journal of Applied Microbiology 107 (1), 97e105. Schets, F.M., Medema, G.J. and Schijven, J.F, 2004. The recovery of the detection method for Cryptosporidium and Giardia in water. RIVM Report 330000008, 71 p. in Dutch. Schets, F.M., van Wijnen, J.H., Schijven, J.F., Schoon, H., de Roda Husman, A.M., 2008. Monitoring of waterborne pathogens in surface waters in amsterdam, the Netherlands, and the potential health risk associated with exposure to cryptosporidium and giardia in these waters. Applied Environmental Microbiology 74 (7), 2069e2078. Schijven, J.F., de Roda Husman, A.M., 2011. QMST and QMRA (Chapter 25). Schijven J.F., Medema G.J., Nijs A.C.M. de and Elzenga J.G, 1996. Emissie en verspreiding van Cryptosporidium, Giardia en enterovirussen via huishoudelijk afvalwater. RIVM report no. 289202014. Schijven, J.F., de Roda Husman, A.M., 2005. Effect of climate changes on waterborne disease in the Netherlands. Water Science & Technology 51 (5), 79e87. Schijven, J.F. and de Roda Husman, A.M., 2009. Analysis of the microbiological safety of drinking water. Experiences with
handling records 2006e7. RIVM report 703719038, (in Dutch). Schijven, J.F., Colin, M., Dullemont, Y., Hijnen, W.A.M., MagicKnezev, A., Oorthuizen, W., Wubbels, G., Bruins, J., Rutjes, S.A. and de Roda Husman, A.M., 2008. Effect filtratiesnelheid, temperatuur en korrelgrootte op de verwijdering van microorganismen door langzame zandfiltratie, 2008 RIVM report 330204002, (in Dutch). Schmidt, P.J., Emelko, M.B., 2011. QMRA and decision-making: are we handling measurement errors associated with pathogen concentration data correctly? Water Research 45 (2), 427e438. Skraber, S., Schijven, J.F., Gantzer, C., de Roda Husman, A.M., 2005. Pathogenic viruses in drinking-water biofilms: a public health risk? Biofilms 2, 105e117. Summerill, C., Pollard, S.J., Smith, J.A., 2010. The role of organizational culture and leadership in water safety plan implementation for improved risk management. Science of the Total Environment 408 (20), 4319e4327. Teunis, P.F.M., Chappell, C.L., Okhuysen, P.C., 2002a. Cryptosporidium dose response studies: variation between isolates. Risk Analysis 22 (1), 175e183. Teunis, P.F.M., Rutjes, S.A., Westrell, T., de Roda Husman, A.M., 2009. Characterization of drinking water treatment for virus risk assessment. Water Resarch 43 (2), 395e404. Teunis, P.F.M., Havelaar, A.H., 2000. The Beta Poisson model is not a single hit model. Risk Analysis 20, 511e518. Teunis, P.F.M., Chappell, C.L., Okhuysen, P.C., 2002b. Cryptosporidium dose-response studies: variation between hosts. Risk Analysis 22 (3), 475e485. Teunis, P.F.M., Evers, E.G., Slob, W., 1999. Analysis of variable fractions resulting from microbial counts. Quantitative Microbiology 1, 63e88. Teunis, P.F.M., Lodder, W.J., Heisterkamp, S.H., de Roda Husman, A.M., 2005a. Mixed plaques: statistical evidence how plaque assays may underestimate virus concentrations. Water Research 39 (17), 4240e4250. Teunis, P.F.M., Medema, G.J., Kruidenier, L., Havelaar, A.H., 1997. Assessment of the risk of infection by Cryptosporidium or Giardia in drinking water from a surface water source. Water Research 31 (6), 1333e1346. Teunis, P.F.M., Van den Brandhof, W., Nauta, M., Wagenaar, J., Van den Kerkhof, H., Van Pelt, W., 2005b. A reconsideration of the Campylobacter dose-response relation. Epidemiology and Infections 133 (4), 583e592. Teunis, P.F.M., Van der Heijden, O.G., Van de Giessen, J.W.B., Havelaar, A.H., 1996. The dose-response relation in human volunteers for gastro-intestinal pathogens. RIVM Report 284550002. USEPA, 2006. Economic Analysis for the Final Ground Water Rule. United States Environmental Protection Agency. 2006. EPA 815-R-06e014. www.epa.gov/safewater. Westrell, T., Andersson, Y., Stenstro¨m, T.A., 2006a. Drinking water consumption patterns in Sweden. Journal of Water and Health 04.4 551-522. Westrell, T., Teunis, P., Van den Berg, H., Lodder, W., Ketelaars, H., Stenstro¨m, T.A., de Roda Husman, A.M., 2006b. Short- and longterm variations of norovirus concentrations in the Meuse river during a 2-year study period. Water Research 40, 2613e2620. WHO, 2011. Guidelines for Drinking-Water Quality, fourth ed. World Health Organisation, Geneve, Zwitserland, ISBN 978 92 4 154815 1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Embodied energy comparison of surface water and groundwater supply options Weiwei Mo a, Qiong Zhang a,*, James R. Mihelcic a, David R. Hokanson b a b
Civil and Environmental Engineering Department, University of South Florida, Tampa, FL 33620, USA Trussell Technologies, Inc., 232 North Lake Ave, Suite 300, Pasadena, CA, USA
article info
abstract
Article history:
The embodied energy associated with water provision comprises an important part of water
Received 16 February 2011
management, and is important when considering sustainability. In this study, an
Received in revised form
inputeoutput based hybrid analysis integrated with structural path analysis was used to
4 August 2011
develop an embodied energy model. The model was applied to a groundwater supply system
Accepted 9 August 2011
(Kalamazoo, Michigan) and a surface water supply system (Tampa, Florida). The two systems
Available online 16 August 2011
evaluated have comparable total energy embodiments based on unit water production. However, the onsite energy use of the groundwater supply system is approximately 27%
Keywords:
greater than the surface water supply system. This was primarily due to more extensive
Embodied energy
pumping requirements. On the other hand, the groundwater system uses approximately 31%
Hybrid analysis
less indirect energy than the surface water system, mainly because of fewer chemicals used
Life cycle assessment
for treatment. The results from this and other studies were also compiled to provide a rela-
Energy path
tive comparison of embodied energy for major water supply options. ª 2011 Elsevier Ltd. All rights reserved.
Groundwater Surface water Reclaimed water Water supply Water treatment
1.
Introduction
Global water withdrawals have increased rapidly over the past several decades, and are expected to continue to grow in the near future (Shah et al., 2003; Konikow and Kendy, 2005; USGS, 2010). Extensive groundwater and surface water withdrawals have led to environmental problems, such as groundwater depletion, land subsidence, seawater intrusion, and surface water quality deterioration, which have consequently impacted water availability in many regions (Taylor and Alley, 2001; Barlow, 2003; USGS, 2003; Konikow and Kendy, 2005). The environmental impacts associated with water supply are further compounded by energy requirements during
withdrawal, treatment, and distribution. The energy used onsite for constructing, operating, and maintaining water supply systems is referred to here as “direct energy.” It comprises around 33% of a typical city’s government energy budget for public utilities in California (CEC, 1992; AwwaRF, 2004) and around 2e3% of global energy demand (ASE, 2002). The energy associated with material use and administrative services is referred to here as “indirect energy.” Previous studies suggest that indirect energy of water supply is comparable to, or even greater than, direct energy (Mo et al., 2009). The embodied energy (direct and indirect energy) associated with water provision also increases with growing water demand. For instance, direct energy increases with
* Corresponding author. Tel.: þ1 813 974 6448; fax: þ1 813 974 2957. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.016
5578
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
a declining water table and well yield, while indirect energy increases when more sophisticated technologies and additional chemicals are used to treat water sources of poorer quality. Reduction of energy use and associated carbon emissions from water supply is also gaining increased attention. For example, in the US, states like California (under Assembly Bill 32) are requiring a reduction in carbon emissions from water supply and treatment. In light of global water management issues, consideration of the energy embodied in water systems should become more important in the future. Accordingly, this study focused on the energy embodiment in water supply systems. Other impact categories associated with material use were not considered as they are beyond the scope of the study. In the last decade, efforts have been made to evaluate the embodied energy of water importation, reclamation, and desalination, driven by the specific regional needs (Peters, 2005; Raluy et al., 2005; Tangsubkul et al., 2005; Stokes and Horvath, 2006; Lyons et al., 2009). The energy embodied in surface water systems has also been studied in countries such as Canada (Racoviceanu et al., 2007) and South Africa (Friedrich, 2002). Embodied energy values associated with specific water supply options are summarized in Table 1. Although environmental impacts such as greenhouse effects, acidification, and nutrient enrichment of groundwater and surface water supply have been compared (Godskesen et al., 2011), no direct comparison has been made in terms of energy embodiment between surface water and groundwater systems as shown in Table 1. Direct energy use associated with groundwater and surface water supply systems, on the other hand, has previously been examined on large scales (e.g., Wilkinson (2000) performed a study for the state of California; EPRI (2002) performed a study for the US). Specifically, the study published by the Electric Power Research Institution (EPRI, 2002) concluded that a groundwater supply system requires about 30% more electricity on a unit basis than a surface water supply system.
Neither of the studies, however, addresses indirect energy consumption. Three methods are primarily used by previous researchers for estimating embodied energy: (1) traditional life cycle assessment, (2) process based hybrid approach, and (3) inputeoutput based hybrid approach. The traditional life cycle assessment tends to underestimate the energy embodiments because of limited data sources and truncated system boundaries (Crawford, 2008). The process based hybrid approach sums the direct energy and the inputeoutput results of the energy embodied in each type of materials. It is more complete than the traditional life cycle assessment; however, it usually suffers from limited data sources for material use, and thus cannot be readily applied to other systems. Accordingly, an inputeoutput based hybrid approach was utilized in this study. This approach involves substituting available process data into an inputeoutput model in order to minimize the errors associated with the traditional life cycle assessment and the process based hybrid analysis (Crawford, 2008). Previous studies (Crawford, 2008; Mattila et al., 2010) have shown that the inputeoutput based hybrid approach is more comprehensive and less labor intensive than the traditional life cycle assessment. Additionally, the inputeoutput based hybrid approach enables flexibility by first providing a rough estimation, and then allowing detailed modifications based on site and system-specific data using structural path analysis. One weakness of this approach is that neither differences in water consumption patterns nor temporal differences associated with water supply systems can be reflected in the model results. The objective of this study was therefore to estimate the “cradle to gate” (source to customer) energy embodiment (direct and indirect energy) of one groundwater and one surface water supply system and to provide a relative comparison of embodied energy for major water supply options through the compilation of results from this and previous studies. The novelty of this study lies in the use of an inputeoutput based hybrid approach with structural path
Table 1 e Life cycle energy associated with water supply systems identified in previous studies. Water Sources
Embodied Energy (MJ/m3 of water)a
Methodology
Comments
Source
Imported water
18 5 42
Process based hybrid LCA Process LCA Process based hybrid LCA
Stokes and Horvath, 2009 Lyons et al., 2009 Stokes and Horvath, 2009
41
Process based hybrid LCA
27 24 17 3 3
Process Process Process Process Process
Conveyance pipe length: 575 km Conveyance pipe length: 261 km Reverse osmosis with conventional pretreatment Reverse osmosis with membrane pretreatment Brackish groundwater Reverse osmosis
Desalinated water
Recycled water Surface water
2
based hybrid LCA LCA based hybrid LCA LCA based hybrid LCA
Process LCA
Only considers operation phase of the treatment plant
Stokes and Horvath, 2009 Stokes and Horvath, 2009 Lyons et al., 2009 Stokes and Horvath, 2009 Lyons et al., 2009 Racoviceanu et al., 2007 Friedrich, 2002
a Energy was reported in the primary energy form, which includes the direct use of energy found in nature and the use of secondary energy such as electricity in forms of fossil fuels, nuclear energy and renewable energy.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
analysis to provide more comprehensive results with insights into the energy flow.
2.
Methodology
An inputeoutput based hybrid approach was used in this study for estimating embodied energy. Basic steps involved in the approach are presented in Fig. 1. This same approach can be used for estimating energy embodied in other water, wastewater, and industrial systems as long as the user has identified appropriate economic target sectors and has access to system-specific data. The system boundary in this study includes the construction and operation stages of water intake infrastructures (wells/exposed tower), treatment plants (administrative buildings included), water storage tanks, pipeline systems, and pumping stations. The end-of-life stage was not considered because the embodied energy associated with it has been shown to be insignificant in previous studies (Friedrich, 2002; Raluy et al., 2005). Among the 424 commodity sectors provided in the 2002 inputeoutput tables provided by the Bureau of Economic Analysis (BEA, 2007), the water-related sectors were identified as the “water, sewage and other systems” sector (WSOS), representing system operation and maintenance (O&M) phase, and the “other nonresidential structures” sector (NS), representing the system construction phase. Additionally, five energy supply sectors were identified within the 424 commodity sectors. They are: (1) oil and gas extraction, (2) coal
5579
mining, (3) power generation and supply, (4) natural gas distribution, and (5) petroleum refineries. The total embodied energy intensities (i.e., the total energy in primary energy form per dollar output of the water systems) for system construction, operation and maintenance phases were estimated using Eqs. (1)e(2) based on data from the 2002 inputeoutput tables. Additional details on the calculation of total embodied energy intensities can be found in Mo et al. (2010). P ¼ 3 t
0 1 N N X X @ dik1 ;jk 3 ik1 A k¼1
3 i0
¼
5 X
(1)
i;j¼1
dn;i0 tarriff n an
(2)
n¼1
In Eqs. (1) and (2), 3 P t ¼ total embodied energy intensity of the target sector “t”, TJ/$ output of sector “t”; k ¼ stage index; N ¼ number of sectors in stage k. dik1 ;jk ¼ direct coefficient from sector “i” at stage k 1 to sector “j” at stage k; 3 ik1 ¼ energy intensity of sector “i” at k 1 stage, TJ/$ output of sector “i”; 3 i0 ¼ direct energy intensity of sector “i” at stage 0, TJ/$ output of sector “i”; n ¼ energy supply sector index; dn;i0 ¼ direct coefficient from energy supply sector n into sector “i”; tarriffn ¼ energy tariff of the energy supply sector n, TJ/$ energy; and, an ¼ nationally averaged primary energy factor of energy supply sector n. After calculating the energy intensities of the water-related sectors, they were multiplied with their corresponding
Fig. 1 e Flow chart for the development of the embodied energy model.
5580
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
economic activities (expenses) to obtain the initial embodied energy. The initial embodied energy was then adjusted for individual systems through structural path analysis (Treloar et al., 2001; Lenzen and Crawford, 2009). This reduces errors caused by sector aggregation in the inputeoutput tables and provides an insight of the energy flow in water supply. The final embodied energy was obtained after the adjustments. The manuscript follows with an introduction to the methods in expense estimation, structural path analysis, and modification of the total embodied energy.
2.1.
Expense estimation
The WSOS sector represents the O&M activities in water supply systems. The monetary output of the WSOS sector is the annual expenses for operating and maintaining water supply systems, which were obtained from the selected water supply systems. The NS sector represents the activities in constructing water supply systems. The monetary output of the NS sector is the capital costs of water systems. Because it is very difficult to obtain the total capital costs directly from the water systems due to expansions and renovations over time, in this study, cost equations and curves were carefully selected to best estimate the capital costs of the existing systems including the capital costs of the treatment processes, equipments, and administrative buildings (Gumerman et al., 1979; Mickley, 2001; Traviglia and Characklis, 2008; McGivney and Kawamura, 2008). Costs estimated by the equations and curves from years other than 2002 were adjusted to 2002 $USD.
2.2.
Structural path analysis
In order to modify the initial embodied energy, energy paths (supply chains starting from the energy involved in one material or service supply sector, and ending at the waterrelated sector) representing high percentages of the total
embodied energy intensities were extracted through structural path analysis. The terms of energy paths and stages are illustrated in Fig. 2. To reduce the amount of calculations, up to 5-stage energy paths were checked. Threshold values were selected to determine the amount of energy paths to be extracted. The large amount of commodity sectors in the US inputeoutput tables leads to an extremely large number of energy paths to be extracted. Thus, in order to represent greater than 90% of the initial total embodied energy intensities, threshold values were selected to extract paths representing 90% of the initial total embodied energy intensities for the two water-related sectors in this study.
2.3.
Modification of the total embodied energy
To modify the initial direct energy, system-specific data from the water supply systems were substituted in to replace the initial model estimations. Due to data limitations, however, this adjustment was only performed for the WSOS sector. To modify the initial indirect energy, the method presented by Treloar (1997) and Lenzen and Crawford (2009) was used. For a certain 1-stage energy paths (from sector s1 to target sector), the energy involved can be calculated as: Es1 ;0 ¼ 3 s1 Cs1 ¼ 3 s1 ds1 ;t Ct
(3)
where: Es1 ;0 ¼ the initial energy for the energy path from sector “s1” (the sector in stage 1) to the target sector “t”, TJ; Es1 ¼ direct energy intensity of sector “s1”, TJ/$ output of sector “s1”; Cs1 ¼ direct purchase from sector “s1” by the target sector “t”, $; ds1 ;t ¼ direct coefficient from sector “s1” to the target sector “t”, $/$ output of the target sector “t”; and, Ct ¼ total monetary output of the target sector “t”, $. According to Eq. (3), the calculation of a 1-stage energy path contains two parts, the direct energy intensity of “s1” sector (3 s1 ) and the amount of “s1” commodity directly used by the target sector (Cs1 ). Both parts were adjusted based on available data. As shown in Eq. (3), Cs1 was calculated by multiplying the direct coefficient with the total monetary output of the water-
Fig. 2 e Description of Energy path, stages and relationship between different stages with a sample of a 3-stage energy path.
5581
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
related sector “t” in the inputeoutput analysis. It can be adjusted using detailed expenses associated with different items obtained from the selected water supply systems. To adjust 3 s1 energy use for manufacturing sectors in 2002 was obtained from the Energy Information Administration (EIA, 2010). The modified energy can be calculated using Eq. (4) 3
Es1 ;D ¼ Es1 ;0 rs1 ¼ Es1 ;0
adj s1
!
3 s1
adj
Cs1 Cs1
! (4)
In Eq. (4), rs1 ¼ ratio of the modified energy to the initial energy for the energy path from sector “s1” to the target sector “t”; Es1 ;D ¼ modified energy for the energy path from sector “s1” adj to the target sector “t”, TJ; 3 s1 ¼ adjusted direct energy 1 intensity of sector “s ”, TJ/$ output of sector “s1”; and, adj Cs1 ¼ adjusted direct purchase from sector “s1” by the target sector “t”, $. For energy paths with i stages, the initial energy involved can be determined using Eq. (5): Esi ;0 ¼
i Y
3 si dsk ;sk1 Ct
(5)
k¼1
where, Esi ;0 ¼ initial energy for the energy path from the sector in stage i “si” to the target sector “t” (target sector is the sector in stage 0, s0), TJ; 3 si ¼ direct energy intensity of the sector in stage i “si”, TJ/$ output of sector “si”; and, dsk ;sk1 ¼ direct coefficient from sector “sk” to sector “sk1”, (s0 represents the sector in stage 0 which is the target sector “t”). Similarly, the modified energy for energy path from the sector “si” to the target sector “t” can be calculated using Eq. (6) with rsi . Esi ;D ¼ Esi ;0 rsi ¼ Es1 ;0
3
adj si
!
3 si
adj
Cs1 Cs1
! (6)
For the commodity use at stage “i”, an assumption has been made that the change of direct commodity use will cause the upstream supply of this commodity to change proportionally. Also, the indirect energy was modified by substituting the original energy embodied in each energy path with the modified energy.
3. Description of water systems used in study One groundwater supply system (Kalamazoo Public Water Supply System, Michigan) and one surface water supply system (City of Tampa Waterworks, Florida) were studied. These two systems were chosen because: (1) both of them are classified as “very large” water supply systems by the US
Environmental Protection Agency according to the population they serve (both systems serve > 100,000 people) (EPA, 2010); (2) they represent typical groundwater and surface water treatment processes; and (3) data for these two systems are readily available to the authors. Geographic differences of the two systems were not considered in this study. A detailed comparison of the two systems is provided in Table 2.
3.1.
Kalamazoo Public Water Supply System
The Kalamazoo Public Water Supply System (referred to as Kalamazoo system) is the largest groundwater based water supply system in the Kalamazoo River watershed, serving over 121,000 customers. The Kalamazoo system pumps an average of 76.8 thousand m3 of water per day and deploys 1276 km of water mains. Raw water is withdrawn from 101 local wells with an average well depth of around 58 m. Limited treatment (disinfection) is provided in two of the total 18 pumping stations, after which the water is supplied to the end users (Kalamazoo, 2008). The annual O&M expense in the Kalamazoo system is approximately $11.1 million. Of the $11.1 million annual expense, $1.16 million are used for purchasing electricity, and $0.08 million are used for purchasing natural gas (CKWD, 2010). The commodity output of the NS sector (construction) was estimated based on the capital cost of the Kalamazoo system. Since the Kalamazoo system only has limited treatment within the pumping stations, the water treatment infrastructure was not considered separately. The well data of the Kalamazoo system were obtained from the “Water Well Viewer” (MDEQ, 2009a) and “Wellogic” (MDEQ, 2009b) managed by the Michigan Department of Environmental Quality.
3.2.
City of Tampa Waterworks
The City of Tampa Waterworks (referred to as Tampa system) is one of the largest water supply systems in Florida, serving a population of 657,000. The average daily flow in the system is approximately 287 thousand m3, about 3.7 times higher than the average flow in the Kalamazoo system. However, the impact of such differences on direct energy use per unit water produced is negligible at the production scale between 38 thousand m3 per day (10 MGD) and 380 thousand m3 per day (100 MGD) (EPRI, 2002). As a result, it is allowable for us to compare the total embodied energy of the two systems. The Tampa system has more than 3541 km of water mains. Raw water is withdrawn from the Hillsborough River, and treated with pre-ozonation and GAC filters in addition to a conventional process that consists of flash mix, flocculation and sedimentation. The raw water has a turbidity of 15e220
Table 2 e Key information of the Kalamazoo system and the Tampa system. Water supply systems
Water source
The Kalamazoo system The Tampa system
Groundwater Surface water
Daily flow (thousand m3/day)
Serving population
Percentage of chemical cost with total O&M cost
Length of the pipelines (km)
76.8 287
121,000 657,000
2% 13%
1276 3541
5582
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
NTU with an average of 117 NTU. The detected dissolved oxygen has a range of 1.9e14.3 mg/L with an average of 4.1 mg/L. The bromide detected ranges from 31 to 180 mg/L with an average of 85 mg/L. This is greater than the Maximum Daily Level of 0.5 mg/L. Total organic carbon ranges from 3.3 to 24.2 mg/L with an average of 15.1 mg/L. The annual O&M expense is $68.3 million. Of the $68.3 million annual expense, $3.95 million is used for purchasing electricity. The commodity output of the NS (construction) is estimated based on the capital cost of the Tampa system. Key information used for estimating capital cost was collected directly from the Tampa system.
4.
Results and discussion
4.1.
Expense estimation
Table 3 e Threshold value that represents 90% of the initial total embodied energy intensity for WSOS (the “water, sewage and other systems” sector) and NS (the “other nonresidential structures” sector). Sector Threshold Value (GJ/$) 1-stage 2-stage 3-stage 4-stage 5-stage
energy energy energy energy energy
paths paths paths paths paths
a
WSOS
NS
1.48E-07
2.40E-09
179 1031 240 8 0
190 8889 19,356 4819 499
a Energy path: Supply chains start from the energy involved in one material or service supply sector, and end at the water-related sector.
Kalamazoo system pipeline capital expense within the total capital expense is much larger than that of the Tampa system. This may result from the more distributed water intake infrastructure in the Kalamazoo system compared with the Tampa system and the lower population density in the City of Kalamazoo compared with the City of Tampa (USCB, 2010). For both systems, pipeline construction is the largest capital cost contributor. Overall, the results show that surface water supply systems may be more expensive to operate than the groundwater supply systems depending on the raw water quality, but may be less expensive to construct than the groundwater supply systems depending on the length of pipelines.
The estimated total capital expense in the Kalamazoo system is $118.4 million, and the total capital expense in the Tampa system is $416.0 million. The breakdowns of the capital costs in both systems are provided in Fig. 3. Assuming life spans for both systems of 100 years (Peters, 2005; Stokes and Horvath, 2006), the unit capital expense for the Kalamazoo system is around $42 per thousand m3 of water produced, and the unit O&M expense is around $394 per thousand m3 of water produced. The total cost (construction and O&M) for producing one thousand m3 of water in the Kalamazoo system is $436. Similarly, the unit capital expense for the Tampa system is around $40 per thousand m3 of water produced, and the unit O&M expense is around $653 per thousand m3 of water. The total cost for producing one thousand m3 of water in the Tampa system is $692. The unit O&M expense of the Tampa system is much larger than the Kalamazoo system. This may be because of the much greater use of water treatment chemicals in the Tampa system. On the other hand, the unit capital expenses of both systems are similar, even though the Tampa system has an additional water treatment plant. The percentage of the
4.2.
Top energy paths for water-related sectors
For the structural path analysis, the threshold values selected for both sectors and the numbers of energy paths in each of the five stages checked are provided in Table 3. Stage 2 has the most energy paths extracted for the WSOS sector; while stage 3 has the most energy paths for the NS sector. Overall, the NS sector has significantly more energy paths than the WSOS
0.9
28
0.8
10
0.1
Pipeline systems
Tampa System
Wells (Kalamazoo) Pumping stations 0 .1 2 .0
37
Water storage tanks
33.2 2
Water treatment plant Kalamazoo System
Exposed tower (Tampa))
0
5
10
15
20
25
30
35
40
45
Capital Cost (Dollar per thousand m3 in $ 2002 under 100-year lifetime)
Fig. 3 e Breakdown of capital costs per thousand cubic meter of water produced under 100 year life-time associated with the groundwater sourced Kalamazoo system and the surface water sourced Tampa system in $ 2002.
5583
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
Table 4 e Total embodied energy for groundwater sourced Kalamazoo system and the surface water sourced Tampa system. Water Supply Systems
The Kalamazoo system The Tampa system Differencesa
Direct Energy (MJ/m3)
Indirect Energy (MJ/m3)
Total Embodied Energy (MJ/m3)
O&M
Construction
Total
O&M
Construction
Total
O&M
Construction
Total
6.1 4.8 28%
0.2 0.2 6%
6.3 5.0 27%
3.7 5.5 32%
0.3 0.3 9%
4.0 5.8 31%
9.8 10.3 5%
0.5 0.5 8%
10.3 10.8 4%
a Differences ¼ [(Data from the Kalamazoo System -Data from the Tampa System)/Data from the Tampa System].
sector in the top 90% of the initial total embodied energy intensity, and the paths in the NS sector are more evenly distributed in different stages than in the WSOS sector. The top energy paths of the WSOS sector are mainly related to maintenance and engineering services, production of the treatment and maintenance materials, and transportation. The top energy path of the NS sector are mainly involved with the production of the building materials, such as asphalt, steel, cement, stone etc., engineering services and transportation. Engineering services have a large impact on the indirect energy use of water systems because large amounts of energy and materials are required to provide such services. Chemical and construction material production is another major contributor to the indirect energy because a large amount of energy is consumed during each stage of the production processes. Lastly, transportation plays a significant role in energy consumption for both constructing and operating water systems.
4.3. Modification and calculation of the total embodied energy The system-specific O&M direct energy use was estimated through the annual energy expenditures and local average energy prices. The average electricity retail price in Michigan is 9.18 cents/kWh, and the average price of natural gas is 6.1 dollars/GJ. Thus, the direct energy for operating and maintaining the Kalamazoo system was estimated to be 170 TJ. The direct energy for both O&M and construction amounts to 6.3 MJ per m3 of water produced at the Kalamazoo system. On the other hand, the average electricity retail price in Florida is 10.13 cents/kWh. Thus, the direct energy for the O&M of the Tampa system was estimated to be 497 TJ. The direct energy for both O&M and construction amounts to 5.0 MJ/m3 of water produced at the Tampa system. For indirect energy, the available system-specific data from the Kalamazoo system and the Tampa system were substituted to adjust the original embodied energy of the two systems. The direct energy intensities of 25 manufacturing sectors were also modified (EIA, 2010). Under a 100-year life
span, the indirect energy used for the Kalamazoo system to supply 1 m3 of water is 4.0 MJ, and the indirect energy used for the Tampa system to supply 1 m3 of water is 5.8 MJ after modification. After the modification of both direct and indirect energy, the total embodied energies for the two water supply systems are provided in Table 4. The total embodied energy in the Kalamazoo system for supplying 1 m3 of water is 10.3 MJ, and the total embodied energy in the Tampa system for supplying 1 m3 of water is 10.8 MJ. The unit total embodied energy in the Tampa system is slightly larger than that of the Kalamazoo system. Compared with initial total embodied energy, the modified total embodied energy of the Kalamazoo system increased by 68%, and the modified total embodied energy of the Tampa system increased by 10%. The differences show the necessity of the modification step using the system-specific data in the analysis. The unit direct energy consumption of the Kalamazoo system is 27% higher than the Tampa system, which is consistent with EPRI’s estimation. This result can be explained by the large pumping requirement for water delivery in the Kalamazoo system. It is also consistent with a previous result that the pipeline system in the Kalamazoo system accounts for a more important portion of energy consumption than the Tampa system. Groundwater supply systems usually have deep and widely distributed wells for water intake, which may increase their pumping energy requirements. Unlike the direct energy consumption, the unit indirect energy consumption at the Kalamazoo system is around 31% less than the Tampa system. This is primarily because of the greater use of chemicals and engineering services at the Tampa system. Groundwater supply systems typically have better raw water quality than surface water supply systems. Systems such as the Kalamazoo system require only limited treatment, which significantly reduces the amount of required chemicals. In contrast, the Tampa system uses a large quantity of chemicals to treat the lower quality raw water. In addition to disinfectants, other chemicals such as ferrous sulfate (for coagulation) and ozone (for pre-ozonation) are used. Manufacturing these chemicals is very energy intensive based on the data from the inputeoutput tables. Moreover, the
Table 5 e Breakdown of the major contributors to the total O&M embodied energy. Energy use categories Kalamazoo System Tampa System
Direct energy use
Chemicals
Maintenance
Engineering service
Customer service
61.9% 46.1%
5.7% 9.6%
12.6% 13.9%
0.7% 3.2%
0.4% 0.4%
5584
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
Table 6 e Sensitivity analysis of the Kalamazoo system and the Tampa system. Selected water supply systems
Direct energy
Infrastructure life span
Total chemical use
þ50%
50%
þ50%
Total embodied energy
Total embodied energy
Indirect energy
Total embodied energy
þ30% þ22%
þ5% þ4%
þ8% þ9%
þ3% þ5%
The Kalamazoo system The Tampa system
surface water supply systems are usually more complicated than the groundwater supply systems, thus more engineering services are involved, which also contributes to the large indirect energy demand of the Tampa system. Breakdown of the major contributors to the total O&M embodied energy of the two systems is provided in Table 5.
4.4.
Comparison with other studies
The results from this study are higher than the embodied energy provided by Racoviceanu et al. (2007) and Friedrich (2002) partly due to the different system boundaries selected. Unlike this study, Racoviceanu et al. (2007) only considered the operation phase of the treatment plant, while Friedrich considered all operation, construction, and decommission phases of the treatment plant. Furthermore, the estimated embodied energy varies a lot based on different estimation methods used, different raw water qualities and treatment technologies, and different geographical locations. For instance, as shown in Table 1, even the energy embodiments of the similar three water supply options studied by Stokes and Horvath (2009) and Lyons et al. (2009) differ by 2e4 fold. Although there is some variance in previous results, desalination consistently appears as the most energy intensive water supply option. Furthermore, the embodied energy of surface and groundwater supplies is comparable with options of water reclamation and importation. Additional studies are needed to compare groundwater, surface water, and reclaimed water supply options in a similar geographical area, with more details on raw water quality and treatment process characteristics, in order to better understand the energy and material use of these options.
4.5.
Uncertainty and sensitivity analysis
Uncertainties in this study are primarily from the inputeoutput tables, varied life span of different components, different geographical location of the selected systems, and capital expense estimation. Bullard and Sebald (1988) found a standard error of 1% for row sums in the US 1967 inputeoutput tables, while Lenzen (2000) assumed an error bound of 3% for the Australian inputeoutput tables. Because there is a lack of studies on the truncation errors and sensitivity of the recent US inputeoutput tables, uncertainty of our results was not quantified. A sensitivity analysis was carried out to determine how direct energy and different inputs used for the estimation would affect the results (Table 6). The analysis showed that the results are very sensitive to the direct energy consumption because it accounts for the largest portion of the total
embodied energy. Additionally, the Kalamazoo system is more sensitive to direct energy than the Tampa system, which is consistent with the previous discussion that the Kalamazoo system has higher unit direct energy use. The results are however not very sensitive to the change of the system life span. This is because the construction life stage only comprises a small portion of the total embodied energy. In regards to chemical use, the Tampa system is more sensitive to it than the Kalamazoo system. This observation is also consistent with the previous discussion that the Tampa system has a larger indirect energy requirement, primarily because of the greater use of chemicals.
5.
Conclusions
The results from this study show that Kalamazoo groundwater supply system that only employs disinfection with no additional treatment is more energy intensive than Tampa surface water supply system in terms of direct energy. This is caused by higher pumping requirements; however, the surface water supply system is more energy intensive in terms of indirect energy because of greater requirements for material use. The results from this study are also higher than previous life cycle studies performed on surface water systems due to different system boundaries selected and different estimation methods used. This study shows the flexibility of using the inputeoutput based hybrid analysis based on data availability. It can be easily used by researchers and utilities to evaluate embodied energy of water supply systems. This method, however, still has various uncertainties including errors propagated from inputeoutput tables and uncertainties in the capital cost estimation for the selected water supply systems. Additionally, this study did not consider the geographical differences between the two systems, which may also affect the total embodied energy. The sensitivity analysis indicated that the results are very sensitive to the direct energy use. However, the results are not very sensitive to the system life span. In addition, the embodied energy of the Tampa system is more sensitive to the chemical use than that of the Kalamazoo system. Although there is no significant difference on the total embodied energy consumption for the specific groundwater and surface water supply systems evaluated, the results suggest there is a trade-off between direct and indirect energy for different systems. It is thus important for water managers to differentiate direct and indirect energy in future life cycle or energy studies for water supply systems.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
Acknowledgments This material is based in part upon work supported by the National Science Foundation under Grant Numbers CBET 0725636. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. We would also like to thank the Kalamazoo Water Department and Mr. Skip Pierpont from the City of Tampa Waterworks for their assistance.
references
ASE, Alliance to Save Energy/US Agency for International Development, 2002. Watergy: taking advantage of untapped energy and water efficiency opportunities in municipal water systems. AwwaRF, Awwa Research Foundation, 2004. Water and wastewater industry energy efficiency: a research roadmap, Project #2923. Barlow, P.M., 2003. Ground water in fresh water salt water environments of the Atlantic coast. U.S. Geological Survey Circular 1262. BEA, Bureau of Economic Analysis, 2007. http://www.bea.gov/ industry/index.htm#benchmark_io (accessed 15.12.08.). Bullard, C.W., Sebald, A.V., 1988. Monte Carlo sensitivity analysis of inputeoutput models. The Review of Economics and Statistics 70 (4), 708e712. CEC, California Energy Commission, 1992. Energy efficiency programs for cities, counties, and schools. P400-91-030. CKWD, City of Kalamazoo Water Department, 2010. Personal communication. Crawford, R.H., 2008. Validation of a hybrid life-cycle inventory analysis method. Journal of Environmental Management 88 (3), 496e506. EIA, Energy Information Administration. http://www.eia.doe.gov/ emeu/mecs/mecs2002/data02/shelltables.html (accessed 15. 07.10.). EPA, US Environmental Protection Agency. http://water.epa.gov/ infrastructure/drinkingwater/pws/factoids.cfm (accessed 01. 04.10.). EPRI, Electric Power Research Institute, 2002. Water & Sustainability. In: U.S. Electricity Consumption for Water Supply & Treatment e The Next Half Century, vol. 4 No. 1006787, Palo Alto, California. Friedrich, E., 2002. Life-cycle assessment as an environmental management tool in the production of potable water. Water Science and Technology 46 (9), 29e36. Godskesen, B., Zambrano, K.C., Trautner, A., Johansen, N.B., Thiesson, L., Andersen, L., Clauson-Kaas, J., Neidel, T.L., Rygaard, M., Kloverpris, N.H., Albrechtsen, H.J., 2011. Life cycle assessment of three water systems in Copenhagen e a management tool of the future. Water Science and Technology 63 (3), 565e572. Gumerman, R.C., Culp, R.L., Hansen, S.P., 1979. Estimating water treatment costs: volume 2-cost curves applicable to 1 to 200 MGD treatment plants. EPA-600/2-79-162b. National Technical Information Service, Springfield, VA. Kalamazoo, The City of Kalamazoo, 2008. 2007 Water quality report. Environmental Service Division, Public Service Department.
5585
Konikow, L.F., Kendy, E., 2005. Groundwater depletion: a global problem. Hydrogeology Journal 13 (1), 317e320. Lenzen, M., 2000. Errors in conventional and input-output-based life-cycle inventories. Journal of Industrial Ecology 4 (4), 127e148. Lenzen, M., Crawford, R., 2009. The path exchange method for hybrid LCA. Environmental Science and Technology 43 (21), 8251e8256. Lyons, E., Zhang, P., Benn, T., Sharif, F., Li, K., Crittenden, J., Costanza, M., 2009. Life cycle assessment of three water supply systems: importation, reclamation and desalination. Water Science & Technology 9 (4), 439e448. Mattila, T.J., Pakarinen, S., Sokka, L., 2010. Quantifying the total environmental impacts of an industrial symbiosis e a comparison of process-, hybrid and input-output life cycle assessment. Environmental Science and Technology 44 (11), 4309e4314. McGivney, W., Kawamura, S., 2008. Cost estimating manual for water treatment facilities. John Wiley & Sons, Hoboken, N.J. MDEQ, Michigan Department of Environmental Quality, 2009a. Water Well Viewer. http://wellviewer.rsgis.msu.edu/ (accessed 31.08.09.). MDEQ, Michigan Department of Environmental Quality, 2009b. Welllogic. http://www.deq.state.mi.us/wellogic/main.html (accessed 31.08.09.). Mickley, M.C., 2001. Membrane Concentrate Disposal: Practices and Regulation. Department of the Interior: Bureau of Reclamation, Denver, CO, U.S, pp. 1e266. Mo, W., Zhang, Q., Mihelcic, J.R., Hokanson, D.R., 2009. Development and application of an embodied energy model for individual water supply systems in Great Lakes Region. In: Proceedings of the 80th Annual Water Environment Federation Conference & Exposition Orlando, Florida, October 10e14, 2009. Mo, W., Nasiri, F., Eckelman, M.J., Zhang, Q., Zimmermman, J.B., 2010. Measuring the embodied energy in drinking water supply systems: a case study in Great Lakes Region. Environmental Science and Technology 44 (24), 9516e9521. Peters, G.A., RK., 2005. Environmental sustainability in water supply planning e an LCA approach for the Eyre Peninsula, South Australia. In: 4th Australian Life Cycle Assessment Conference, Sydney. Racoviceanu, A.I., Karney, B.W., Kennedy, C.A., Colombo, A.F., 2007. Life-cycle energy use and greenhouse gas emissions inventory for water treatment systems. Journal of Infrastructure Systems 13 (4), 261e270. Raluy, R.G., Serra, L., Uche, J., 2005. Life cycle assessment of desalination technologies integrated with renewable energies. Desalination 183, 81e93. Shah, T., Roy, A.D., Qureshi, A.S., Wang, J., 2003. Sustaining Asia’s groundwater boom: an overview of issues and evidence. Natural Resource Forum 27 (2), 130e141. Stokes, J.R., Horvath, A., 2006. Life cycle energy assessment of alternative water supply systems. International Journal of Life Cycle Assessment 11 (5), 335e343. Stokes, J.R., Horvath, A., 2009. Energy and air emission effects of water supply. Environmental Science & Technology 43 (8), 2680e2687. Tangsubkul, N., Beavis, P., Moore, S.J., Lundie, S., Waite, T.D., 2005. Life cycle assessment of water recycling technology. Water Resource Management 19, 521e537. Taylor, C.J., Alley, W.M., 2001. Ground-water level monitoring and the importance of long-term water-level data. U.S. Geological Survey Circular 1217, 68. Traviglia, A.M., Characklis, G.W., 2008. An expert system for decision making in the use of desalination for augmenting
5586
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 7 7 e5 5 8 6
water supplies. Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill. DWPR Report No. 107. Treloar, G.J., 1997. Extracting embodied energy paths from inputoutput tables: towards an input-output based hybrid energy analysis method. Economic System Research 9 (4), 375e391. Treloar, G., Love, P., Faniran, O.O., 2001. Improving the reliability of embodied energy methods for project life-cycle decision making. Logistics Information Management 14 (5/6), 303e318. USCB, U.S. Census Bureau, 2010. State and County Quick Facts. http:// quickfacts.census.gov/qfd/index.html (accessed 15.12.10.).
USGS, U.S. Geographical Survey, 2003. Ground-water depletion across the nation. U.S. Geological Survey Fact Sheet 103e03. USGS, U.S. Geographical Survey, 2010. Groundwater Use in the United States. http://ga.water.usgs.gov/edu/wugw.html (accessed 15.07.10.). Wilkinson, R., 2000. Methodology for analysis of the energy intensity of California’s water systems and an assessment of multiple potential benefits through integrated water energy efficiency measures. Environmental Studies Program at the University of California, Santa Barbara. Agreement No. 4910110.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Nitrate reduction in a simulated free-water surface wetland system Teresa M. Misiti, Malek G. Hajaya, Spyros G. Pavlostathis* School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0512, USA
article info
abstract
Article history:
The feasibility of using a constructed wetland for treatment of nitrate-contaminated
Received 13 June 2011
groundwater resulting from the land application of biosolids was investigated for a site
Received in revised form
in the southeastern United States. Biosolids degradation led to the release of ammonia,
7 August 2011
which upon oxidation resulted in nitrate concentrations in the upper aquifer in the range
Accepted 11 August 2011
of 65e400 mg N/L. A laboratory-scale system was constructed in support of a pilot-scale
Available online 22 August 2011
project to investigate the effect of temperature, hydraulic retention time (HRT) and nitrate and carbon loading on denitrification using soil and groundwater from the biosolids
Keywords:
application site. The maximum specific reduction rates (MSRR), measured in batch assays
Denitrification kinetics
conducted with an open to the atmosphere reactor at four initial nitrate concentrations
Groundwater
from 70 to 400 mg N/L, showed that the nitrate reduction rate was not affected by the initial
Constructed wetland
nitrate concentration. The MSRR values at 22 C for nitrate and nitrite were 1.2 0.2 and
Carbon loading
0.7 0.1 mg N/mg VSSCOD-day, respectively. MSRR values were also measured at 5, 10, 15
Nitrate loading
and 22 C and the temperature coefficient for nitrate reduction was estimated at 1.13.
Temperature effect
Based on the performance of laboratory-scale continuous-flow reactors and model simulations, wetland performance can be maintained at high nitrogen removal efficiency (>90%) with an HRT of 3 days or higher and at temperature values as low as 5 C, as long as there is sufficient biodegradable carbon available to achieve complete denitrification. The results of this study show that based on the climate in the southeastern United States, a constructed wetland can be used for the treatment of nitrate-contaminated groundwater to low, acceptable nitrate levels. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
World-wide, nitrate is among the most common groundwater contaminants, mainly introduced into the environment from agricultural activities related to the excessive use of nitratecontaining fertilizers and manure (Burkart and Stoner, 2002; Murgulet and Tick, 2009; Rivett et al., 2008). In addition, as an alternative to landfilling, biosolids generated by the anaerobic digestion of municipal primary and waste activated sludge or other stabilization processes are in some cases land
applied to serve as a nutrient source for plant growth as well as to enrich the soil in organic matter. In the case of land application of biosolids, ammonia, which is either already present or produced as a result of biosolids degradation on site, is utilized as a nitrogen source for plant growth. When biosolids are land applied at recommended agronomic frequencies and rates, ammonia is not anticipated to be of environmental concern. However, when biosolids are applied in excess of recommended rates or during non-growing seasons, excess ammonia is oxidized to nitrate, which can
* Corresponding author. Tel.: þ1 404 894 9367; fax: þ1 404 894 8266. E-mail address: [email protected] (S.G. Pavlostathis). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.019
5588
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
then leach into the groundwater (Surampalli et al., 2008; US EPA, 2000a). A wastewater treatment plant located in the southeastern United States, which services an area with a total population of approximately 230,000, generates biosolids at approximately 5.4 103 dry metric tons/year. A portion of the biosolids had been land applied to an agricultural field adjacent to the wastewater treatment plant and served as a source of nutrients for crop production. As a result of longterm biosolids application, groundwater nitrate concentrations in the upper aquifer of this site reached levels between 65 and 400 mg NO 3 eN/L, which are above the regulated limit of 10 mg NO 3 eN/L (Maltais-Landry et al., 2009c; US EPA, 2009). Nitrate and nitrite concentrations higher than the regulated limit present health concerns as they are toxic to humans and livestock (US EPA, 2009). As a remediation approach, the wastewater treatment facility proposed groundwater pumping and development of an overland-flow wetland system for the treatment of the nitratecontaminated groundwater before being released into the nearby river. >Wetland-based treatment systems are commonly used for the biological removal of nitrogen, phosphorus, sulfur, heavy metals and other pollutants, acting as a buffer between the pollution source and the natural aquatic ecosystem (Bachand and Horne, 2000; Kadlec and Wallace, 2009; Kjellin et al., 2007; Maltais-Landry et al., 2009c). Among all the nitrogen transformation processes taking place in wetlands, the one related to nitrate removal is denitrification, i.e., the reduction of nitrate to dinitrogen (N2). The effectiveness of wetlands is largely affected by the biological activity and temperature of the wetland location. Treatment wetlands are often constructed in regions with moderate to cold climates that experience large seasonal temperature variations (Kadlec and Reddy, 2001). Wetlands are complex biological systems and their performance depends on a number of chemical, physical and biological processes (US EPA, 2000b). For the biologicallymediated nutrient removal in wetlands, both empirical and mechanistic models have been used. Kadlec and Reddy (2001) assessed the effect of temperature on treatment wetlands by using a simple model where all reaction mechanisms are grouped into a pseudo-first order removal rate. On the other hand, Kjellin et al. (2007) used Menten and first-order kinetics to model nitrate removal in wetland sediment. Denitrification kinetics have also been modeled using both the single- and dual-substrate Monod model (Hajaya et al., 2011; Heinen, 2006; Kornaros et al., 1996). A three-cell pilot-scale, free-water surface wetland system was constructed at the above-mentioned biosolids application site to test the effectiveness of treating the nitratecontaminated groundwater using various carbon sources during the system’s start-up period while vegetation was being established. One cell served as a control (i.e., no external carbon addition) and the other two cells used either MicroC G or hay as carbon source, respectively. The pilot-scale system was designed to demonstrate the efficiency of microbial nitrate reduction under conditions open to the atmosphere and as affected by various environmental and operational conditions. In particular, the high groundwater
nitrate concentrations and the presence of oxygen in the freesurface wetland system presented conditions which may affect nitrogen removal efficiency. In support of the pilot-scale constructed wetland demonstration project, a continuous-flow, laboratory-scale system was built to assess the treatment of nitratecontaminated groundwater as a function of carbon (i.e., electron donor) and nitrate loading, hydraulic retention time, and temperature. The objective of the laboratory study was to determine the feasibility of using a constructed wetland for treatment of the nitrate-contaminated groundwater through a series of batch and continuous-flow nitrate reduction tests using MicroC G as the electron donor, as well as soil and nitrate-contaminated groundwater from the biosolids application site. Nitrate and nitrite reduction rate estimates, resulting from the laboratory batch tests, were used in a mathematical model to simulate the nitrogen removal efficiency of continuous-flow, free-water surface systems as a function of both operational and environmental conditions.
2.
Materials and methods
2.1.
Sample collection and characterization
Groundwater and surface soil were collected at the biosolids application site located in the southeastern United States. The groundwater samples were stored in plastic containers under refrigeration (4 C). The soil sample was passed through a US No. 10 sieve, spread thin to air dry for 24 h at room temperature, and then stored in covered plastic containers at room temperature (22e24 C). All samples were characterized by measuring pH, soluble and total chemical oxygen demand (sCOD and tCOD), dissolved organic carbon (DOC), moisture content, NHþ 4 eN, NO3 eN, NO2 eN, and other ions. To measure soil pH, DOC, soluble COD, ammonia and ions, a soil filtrate solution was prepared by adding 5 g of dry soil to 300 mL of deionized (DI) water and mixing for 1 day at room temperature. The soil solution was then centrifuged at 10,000 rpm for 30 min. The results of soil and groundwater characterization are shown in Table 1. Both the soil and groundwater samples were slightly acidic, with pH values of 4.4 and 5.7, respectively. The soil sample was mostly inorganic matter (w95%) and did not contribute significant soluble COD or ions to the solution (Table 1). MicroC G, a plant-derived complex carbohydrate mixture, was obtained from Environmental Operating Solutions Inc. (Bourne, MA) and used as the electron donor and carbon source in all experiments. A MicroC G solution was prepared by a 1000-fold dilution in DI water and analyzed for pH, DOC, soluble COD and ions. The concentrated MicroC G stock and the diluted solution were stored in the dark at 22 and 4 C, respectively. The MicroC G solution was acidic with a pH of 3.9 and the measured COD of the undiluted solution was approximately 640 g/L, which agrees closely with the technical specifications provided by the manufacturer (Table 1). MicroC G did not contain any anions or ammonia, is soluble in water and has a freezing point of 8 C (EOS, 2008).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
Table 1 e Characteristics of soil, groundwater and MicroC G samples used in this study. Parameter pH Water content (%) Dry weight (%) Organic matter (% of dry) DOC (filtrate; mg C/L) Soluble COD (filtrate; mg/L) Total COD (mg/g dry weight) Ions (filtrate) Chloride (mg Cl/L) Nitrite (mg N/L) Nitrate (mg N/L) Sulfate (mg S/L) Phosphate (mg P/L) Ammonia (filtrate; mg N/L)
Soil 4.4 3.9 0.1a 96.1 0.1 4.9 0.1 9.8 0.4 14.5 7
Groundwater
MicroC G
5.7
3.9
9.1 1.5 58.1 4.8
411 32c 642 41c
14.4 0.8 ND 69.3 1.3 28.4 0.1 ND ND
ND ND ND ND ND ND
68.3 5.2
NDb ND 0.3 0.2 1.3 ND
a Mean standard deviation (n ¼ 3). b ND, not detected. c 1000-fold diluted solution.
2.2.
Continuous-flow laboratory-scale system
Three continuous-flow, laboratory-scale reactors were constructed in order to quantify nitrate removal under different operational conditions including, effect of initial nitrate concentration (Run 1), COD:N ratio (Run 2) and temperature (Run 3). Three 15-L cubic Plexiglas reactors were filled with 10.5 kg of soil and approximately 9 L of nitrate-bearing groundwater and were kept static for 1 day in order to expel all air from the soil and uniformly wet the soil. The water column depth was approximately 12 cm. Run 1 was conducted in a single-compartment continuous-flow reactor; however, in order to more closely simulate the flow regime of a full-scale wetland system, two baffles were inserted into the reactors for Run 2 and 3, thus dividing the liquid volume into three, equal-volume compartments, resulting in a flow regime that simulated 1.5e2 continuous-flow stir tank reactors (CSTRs) in series (see Supplementary Material, Text S1 and Fig. S1). Plastic reservoirs filled with groundwater were attached to peristaltic pumps (Masterflex; Cole-Parmer) and the nitratebearing groundwater was fed to the reactors continuously at a specific flow rate to achieve the target hydraulic retention times (HRTs). MicroC G was used as the electron donor and carbon source and a 200 g COD/L diluted solution was fed using a positive displacement pump (Fluid-Metering, Inc.) at predetermined flow rates to achieve the target COD:N ratios. The first continuous-flow test, Run 1, was designed to investigate the effect of HRT and initial nitrate concentration on system nitrate removal. Site groundwater, with a nitrate concentration of 69 mg NO 3 eN/L, was fed to the system and when the influent nitrate concentration was increased to 150 NO 3 eN/L, the site groundwater was amended with a stock solution of NaNO3. The MicroC G was fed every 2 h with the help of an electronic timer (ChronTrol) at a flow rate dependent on the HRT to maintain a COD:N ratio of 6. Run 1
5589
was conducted at room temperature (22e24 C) and lasted for a total of 121 days. The second continuous-flow test, Run 2, was conducted to investigate the effect of the COD:N ratio on nitrate removal in the system. Similarly to Run 1, site groundwater was fed with a nitrate concentration of 69 mg NO 3 eN/L. The MicroC G was fed every 2 h at a flow rate dependent on the HRT and target COD:N ratios of 6, 5, 4, 3, 2, 1 and 0.5. Run 2 was conducted at room temperature (22e24 C) and lasted for a total of 250 days. The last continuous-flow test, Run 3, was designed to investigate the effect of temperature on the nitrate reduction. The reactor was housed in a controlled temperature room and its temperature was stepwise decreased from 22 to 5 C. The rate of temperature change between the four target temperature values was 2 C/day. The system performance was assessed at four temperature values: 5, 10, 15 and 22 C. Site groundwater and MicroC G were fed at flow rates dependent on the HRT and to maintain a COD:N ratio of 6. At 5 C, the MicroC G had low solubility and a mixer was installed to incorporate the feed into the groundwater (see Section 3.1.3 for more details). Run 3 lasted for a total of 220 days. In all continuous-flow runs, overflow reactor effluent was periodically collected and analyzed for nitrate, nitrite, ammonia, pH, DOC, and soluble COD.
2.3.
Batch assays
A 15-L cubic Plexiglas reactor was used in all batch assays open to the atmosphere, filled with 10.5 kg of soil and approximately 9 L of nitrate-bearing groundwater. Batch assays were performed to investigate the effect of initial nitrate concentration and temperature on denitrification kinetics. To investigate the effect of initial nitrate concentration, the site groundwater, containing approximately 69 mg NO 3 eN/L, was used, and in selected batch assays was amended with a volume of a NaNO3 stock solution to achieve initial concentrations of 150, 300, and 400 mg NO 3 eN/L. This batch assay was conducted at room temperature (22e24 C). To investigate the effect of temperature on nitrate reduction, the site groundwater was amended with a volume of NaNO3 stock solution to achieve an initial concentration of 150 mg NO 3 eN/L at all temperature values tested: 5, 10, 15 and 22 C. The reactor was housed in a controlled temperature room and its temperature decreased stepwise from 22 to 5 C. The rate of temperature change between the four target temperature values was 2 C/day. This batch assay was conducted simultaneously with the above-described continuous-flow Run 3. MicroC G was used as the electron donor and carbon source in all batch assays at a COD:N ratio of 6. In order to maintain similar initial biomass concentrations in the soil layer, after each batch assay was complete, the reactor was drained, backflushed with deionized water three times and nitrate-bearing groundwater once before being refilled with nitrate-bearing groundwater for the next batch assay. After the first batch assay, which assessed the effect of initial nitrate concentration on nitrate reduction, was completed at room temperature, approximately 1 inch of the top soil layer was replaced with fresh site soil and the reactor was transferred to the controlled temperature room to assess the effect of temperature on denitrification kinetics. Similarly to the first batch assay, the
5590
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
reactor was drained and back-flushed in between each batch assay at the various target temperatures. In all open batch assays, liquid samples were taken daily and nitrate, nitrite, pH and periodic COD concentrations were measured. In order to assess the effect of oxygen on the denitrification kinetic rates measured in the open to the atmosphere reactor, a closed batch assay was conducted using duplicate 160-mL serum bottles sealed with rubber stoppers and aluminum crimps and flushed with helium. In each serum bottle, 5 g dry soil and 100 mL nitrate-containing groundwater were added. A volume of different NaNO3 stock solutions were added to each serum bottle resulting in initial nitrate concentrations ranging from 70 to 400 mg NO 3 eN/L. Different volumes of a MicroC G stock solution were added to each bottle resulting in a COD:N ratio of 6. Incubation was carried out at room temperature (22e24 C). During the incubation period, the following parameters were measured: nitrate, nitrite, gas production and gas composition (CO2, NO, N2O and N2).
2.4. Batch denitrification kinetics and parameter estimation For this work, a two-step denitrification model (nitrate to nitrite to dinitrogen) was used. Monod kinetic equations were used to describe microbial growth utilizing nitrate and nitrite in all batch assays. Assuming that nitrate and nitrite are the limiting substrates (electron donor in excess with a COD/N ratio of 6), the following differential equations were used: !
dSNO3 kNO3 SNO3 XNOx ¼ dt KSNO3 þ SNO3 dSNO2 ¼ dt
! kNO3 SNO3 XNOx KSNO3 þ SNO3
dXNOx ¼ dt
! YNO3 kNO3 SNO3 XNOx þ KSNO3 þ SNO3
(1) ! kNO2 SNO2 XNOx KSNO2 þ SNO2
(2)
! YNO2 kNO2 SNO2 XNOx bXNOx KSNO2 þ SNO2 (3)
where SNO3 , SNO2 , and XNOx are nitrate, nitrite and denitrifiers concentrations (mg NO 3 eN/L, mg NO2 eN/L and mg VSSCOD/L, respectively); t is time (days); kNO3 and kNO2 are the nitrate and nitrite maximum specific reduction rates (MSRR; mg N/ gVSSCOD day); KSNO3 and KSNO2 are the nitrate and nitrite halfsaturation constants (mg N/L); YNO3 and YNO2 are the theoretical yield coefficients (g VSSCOD/mg N); and b is the microbial decay coefficient (day1). Based on bioenergetic calculations, the yield coefficients for nitrate (YNO3 ) and nitrite (YNO2 ) used for all simulations were calculated to be 1.14 and 1.72 g VSSCOD/g N, respectively (Rittmann and McCarty, 2001). The decay rate values for denitrifiers are generally in the range of 0.05e0.15 day1 (Rittmann and McCarty, 2001; Tchobanoglous et al., 2003). A microorganism decay rate of 0.1 day1 was chosen for all simulations; however, preliminary simulations using values of 0.05 and 0.15 day1 resulted in small variations in nitrate concentration patterns. Given the fact that the initial, active denitrifiers concentration in the soil (XNOx ) was not measurable, for each set of
experimental data, an initial biomass concentration was chosen to fit the nitrate experimental data. Typical KS values for denitrification have been reported in the range of 0.8e153 mg N/L (Kjellin et al., 2007; Tugtas and Pavlostathis, 2007; Zumft, 1997). The half-saturation constants for nitrate and nitrite (KSNO3 and KSNO2 ) were estimated for each batch assay based on reported value ranges to best fit the experimental data. Parameter estimation was conducted following a previously reported procedure (Hajaya et al., 2011). Parameters sensitivity and identifiability analysis was performed using the Berkeley Madonna Software Version 8.3 (Macey and Oster, 2006) and Matlab ode15 solver (MATLAB 7.0.1; The Mathworks, Natick, MA) following the procedure described by Gujer (2008). The Fit ODE toolbox in Igor Professional v.5.057 (WaveMetrics, Inc., Lake Oswego, OR) was used to calculate the standard deviation values for the evaluated parameters. The effect of temperature on the MSRR was quantified by fitting the resulting MSRR values at each temperature to the modified Arrhenius model using nonlinear regression (SigmaPlot, Version 10.0 software; Systat Software Inc., San Jose, CA, USA): kT2 ¼ kT1 qðT2 T1 Þ T2
(4) T1
where k and k are MSRR values (mg N/mg VSSCOD-day) at two different temperatures ( C) and q is the dimensionless temperature coefficient.
2.5.
Continuous-flow model
The denitrification kinetic rates estimated in the batch assays can be used to simulate and predict the performance of the continuous-flow wetland system under various operational conditions. Based on a modified version of the dual-substrate Monod model presented by Kornaros et al. (1996) and a system mass balance, the continuous-flow system was modeled using the series of differential equations (5) through (8) for nitrate and nitrite reduction, cell growth and electron donor utilization as follows: ! SNO3 ;o SNO3 dSNO3 kNO3 SNO3 C ¼ XNOx s dt kSNO3 þ SNO3 KC þ C ! SNO2 ;o SNO2 dSNO2 kNO3 SNO3 C ¼ XNOx s dt kSNO3 þ SNO3 KC þ C ! kNO2 SNO2 C XNOx KSNO2 þ SNO2 KC þ C ! dC ðCo CÞ YNO3 kNO3 SNO3 C XNOx ¼ kSNO3 þ SNO3 dt s KC þ C ! YNO2 kNO2 SNO2 C XNOx KSNO2 þ SNO2 KC þ C ! dXNOx ðXNOx ;o XNOx Þ YNO3 kNO3 SNO3 C ¼ XNOx þ dt kSNO3 þ SNO3 s KC þ C ! YNO2 kNO2 SNO2 C XNOx bXNOx þ KSNO2 þ SNO2 KC þ C
(5)
(6)
(7)
(8)
5591
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
DOC, COD, pH, dissolved oxygen (DO), water content, NHþ 4 eN, NO 3 eN, NO2 eN, and other ions were measured following procedures outlined in Standard Methods (APHA, 2005). Soluble COD was measured by the HACH colorimetric method and the total COD was measured by the Open Reflux Method. DOC was measured using a Shimadzu Total Organic Carbon (TOC) Analyzer equipped with an infrared detector for CO2 measurement (Shimadzu Scientific Instruments, Inc., Columbia, MD). DO was measured using the polarographic method with a YSI Model 58 oxygen meter in conjunction with a YSI 5750 oxygen probe (Yellow Springs Instruments, Yellow Springs, OH). Anions were measured using a Dionex DX-100 ion chromatography unit (Dionex Corporation, Sunnyvale, CA) equipped with a conductivity detector, a Dionex IonPac AG14A (4 50 mm) precolumn, and a Dionex IonPac AS14A (4 250 mm) analytical column. The unit was operated in autosuppression mode with 1 mM NaHCO3/8 mM Na2CO3 eluent at a flow rate of 1 mL/min. The minimum detection limit for nitrate and nitrite was 0.05 and 0.1 mg N/L, respectively. Gas composition was determined by a gas chromatography (GC) unit (Agilent Technologies, Model 6890N; Agilent Technologies, Inc., Palo Alto, CA) equipped with two columns and two thermal conductivity detectors. Dinitrogen (N2) was separated with a 15 m HP-Molesieve fused silica, 0.53 mm i.d. column (Agilent Technologies, Inc.). Carbon dioxide (CO2), nitric oxide (NO) and nitrous oxide (N2O) were separated with a 25 m Chrompac PoraPLOT Q fused silica, 0.53 mm i.d. column (Varian, Inc., Palo Alto, CA). Helium was used as the carrier gas at a constant flow rate of 6 mL/min. The 10:1 split injector was maintained at 150 C, the oven was set at 40 C and the detector temperature was set at 150 C. The minimum detection limit for CO2, NO, N2O and N2 was 800, 500, 7 and 50 ppmv, respectively.
Effect of HRT and influent nitrate concentration
Run 1 was performed in a single-compartment, continuousflow reactor which was operated at HRT values of 2.8, 3.5 and 5 days, while being continuously fed with groundwater at 67 mg NO 3 eN/L and MicroC G at a COD:N of 6 after the first week, during which external carbon was not added. The effluent nitrate concentration over the entire run period is shown in Fig. 1, along with other operational parameters. Upon addition of MicroC G directly to the reactor on day 8, the effluent nitrate concentration decreased and reached non-detectable levels within 3 days. For the remainder of this run, the effluent nitrate concentration did not exceed 20 mg NO 3 eN/L at any of the three hydraulic retention times tested. To illustrate that the nitrate removal follows Monod kinetics, according to which the effluent nitrate concentration is not a function of influent nitrate concentration, the influent groundwater concentration was increased on day 88e130 mg NO 3 eN/L and the concentration of the MicroC G solution changed accordingly to maintain a COD:N ratio of 6. The effluent nitrate concentration increased slightly to approximately 7 mg NO 3 eN/L until steady-state was
INFLUENT NITRATE (mg N/L)
Analytical methods
3.1.1.
150 125 100 75 50 25 0 6
HRT (Days)
2.6.
The effluent pH ranged from 6.5 to 7.5 and the DO remained below 1 mg/L after addition of MicroC G. Ammonia was not detected in any effluent samples. In all reactors, the effluent COD and DOC remained constant and low (DOC below 40 mg/L and COD below 75 mg/L). Sulfate was also periodically measured and was consistently in the range of 20e30 mg S/L in both influent and effluent samples, indicating that significant sulfate reduction was not occurring in the open to the atmosphere reactors.
EFFLUENT NITRATE (mg N/L)
where SNO3 ;o , SNO2 ;o , and XNOx ;o are influent nitrate, nitrite and biomass concentrations (mg N/L, mg N/L and mg VSSCOD/L, respectively); Co and C are the electron donor (MicroC G) concentrations in the influent and effluent (mg COD/L); KC is the half-saturation constant for the electron donor (mg COD/L); and s is the HRT (¼V/Q) (days). The KC value used in all simulations was 20 mg COD/L, a value reported for MicroC G as the electron donor for denitrification (Cherchi et al., 2009). The effluent nitrate and nitrite concentrations were simulated in Matlab at various HRT, COD:N and temperature values using the previously estimated biokinetic constants (kNO3 , kNO2 , KSNO3 , KSNO2 , YNO3 , YNO2 , and b) and initial biomass concentrations based on the denitrification kinetics described in Section 2.4, above.
4 2 0 100 90 80 70 60 50 40 30 20 10 0 0
3.
Results and discussion
3.1.
Continuous-flow reactor system performance
The effluent streams from the three continuous-flow reactors were periodically analyzed for pH, ammonia, COD and DOC.
10
20
30
40
50
60
70
80
90 100 110 120 130
TIME (Days)
Fig. 1 e Effluent nitrate concentration in a continuous-flow reactor operated at room temperature (22e24 C) at various HRT values, influent nitrate concentrations and with MicroC G at a COD:N ratio of 6:1 at/after day 8 (MicroC G was not added from day 92 to 104).
5592
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
achieved, after which the effluent concentration quickly returned to non-detectable levels (Fig. 1). In order to qualitatively evaluate the effect of carbon source on the nitrate removal in this reactor, at approximately 92 days, the MicroC G pump was turned off and only the groundwater at approximately 130 mg NO 3 eN/L was fed to the reactor at an HRT of 5 days. As shown in Fig. 1, the effluent nitrate concentration increased and reached about 88 mg NO 3 eN/L within 10 days, further demonstrating the necessity of a continuous addition of a degradable carbon source. On day 104, the groundwater pump was turned off to simulate a batch system while the MicroC G pump continued supplying carbon. A gradual decrease of the nitrate concentration to about 20 mg NO 3 eN/L in 20 days was observed. Thus, even at an elevated influent nitrate concentration, continuous addition of biodegradable organic carbon in excess of stoichiometric levels achieved high nitrate removal efficiency.
3.1.2.
Effect of COD:N ratio
Run 2 was conducted in a three-compartment, continuousflow reactor with an influent groundwater nitrate concentration kept constant at 70 mg NO 3 eN/L, an HRT of 2 and then 5 days, while the COD:N ratio was stepwise decreased to lower values. Fig. 2 shows the reactor effluent nitrate concentration along with other operational parameters. For the first 18 days, the reactor was operated at an HRT of 2 days and a COD:N ratio of 6, during which the effluent nitrate concentration decreased sharply to less than 10 mg NO 3 eN/L. The HRT was then increased to 5 days to achieve more stable operation and consistent effluent nitrate concentration of less than 10 mg NO 3 eN/L. The COD:N ratio was stepwise decreased from an initial value of 6:1 to the lowest value of 0.5:1. On day
HRT (Days)
6 4 2
COD:N
4
EFFLUENT NITRATE (mg N/L)
0 6
70
2 0
25, the COD:N ratio was decreased to 5:1, during which the effluent nitrate concentration remained below 4 mg NO 3 eN/L. When on day 75, the COD:N ratio was further decreased to 4:1, the effluent nitrate concentration increased sharply to about 27 mg NO 3 eN/L. A decrease of the COD:N ratio to 3:1 and then to 2:1 resulted in an effluent nitrate concentration ranging between 32 and 40 mg NO 3 eN/L. A further decrease of the COD:N ratio to 1:1 and then to 0.5:1 resulted in a gradual increase of the effluent nitrate concentration to 42 mg NO 3 eN/L. On day 232, the COD:N ratio was increased to 5:1, which resulted in a rapid decrease of the effluent nitrate concentration to below 4 mg NO 3 eN/L. Based on these results, for a system open to the atmosphere at ambient temperature between 22 and 24 C, influent nitrate concentration of 67 mg NO 3 eN/L, and an HRT value of 5 days, the minimum COD:N ratio is approximately 5:1 in order to achieve an effluent nitrate concentration of less than 10 mg N/L. The theoretical requirement for complete denitrification, ignoring microbial growth, is 2.85 mg COD/mg nitrateeN reduced to N2. At relatively low COD:N values, incomplete denitrification is possible, which could lead to the formation of nitric oxide (NO) and nitrous oxide (N2O), both potent greenhouse gases (Maltais-Landry et al., 2009a, 2009b). It has been reported that N2O emissions in wetlands are highly dependent on the COD:N ratio, as well as the pH, dissolved oxygen, and temperature among other parameters (Inamori et al., 2008; Wu et al., 2009). Wu et al. (2009) found that significant amounts of N2O were released from constructed wetlands at very high and very low COD:N ratios (2 > COD:N > 10), with minimum emissions at a ratio of 5:1. In the present study, in order to investigate if NO and N2O were released in the laboratory reactor due to incomplete denitrification, on days 104, 143, 192 and 215 when the continuousflow reactor was operated with a COD:N ratio of 3:1, 2:1, 1:1 and 0.5:1, respectively, gas bubbles and water were collected biweekly from the soil/water interface by using an inverted glass vial fully submerged in the water and partially imbedded into the soil. Gas bubbles released from the surface soil were collected in the vial by water displacement. Then, the vial was sealed with a stopper while under water, positioned upright and its headspace analyzed by gas chromatography after 30 min equilibration at room temperature. NO and N2O were not detected at any of the COD:N ratios tested, confirming that complete denitrification occurred, leading to the production of nitrogen gas (N2) as the main nitrate reduction process in the laboratory reactor.
60
3.1.3.
50 40 30 20 10 0 0
20
40
60
80 100 120 140 160 180 200 220 240 260
TIME (Days)
Fig. 2 e Effluent nitrate concentration in a continuous-flow reactor operated at room temperature (22e24 C), mean influent groundwater nitrate concentration of 70 mg N/L and with MicroC G at several COD:N ratios.
Effect of temperature
The three-compartment, continuous-flow reactor was housed and operated in a temperature-controlled room to simulate the effect of temperature on nitrate reduction. For the first 15 days, the continuous-flow reactor was operated at 22 C with an HRT of 2 days, during which the effluent nitrate concentration decreased sharply to less than 13 mg NO 3 eN/L (Fig. 3). The HRT was then increased to 5 days to achieve more stable operation and an effluent nitrate concentration of less than 10 mg NO 3 eN/L. On day 30 the room temperature was decreased and by day 32 reached 15 C. While at 15 C, the reactor performance did not change and the effluent nitrate concentration was kept at non-detectable levels. On day 51,
EFFLUENT NITRATE (mg N/L)
HRT (Days)
o
TEMP. ( C)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
25 20 15 10 5 0 12 10 8 6 4 2 0 70 60 50 40 30 20 10 0 0
20
40
60
80
100
120
140
160
180
200
220
TIME (Days)
Fig. 3 e Effluent nitrate concentration in a continuous-flow reactor operated at a range of temperature (22e5 C), mean influent groundwater nitrate concentration of 70 mg N/L and with MicroC G at a COD:N ratio of 6:1 (arrow indicates start of complete MicroC G and groundwater mixing; see text).
5593
for approximately another 25 days, during which period the effluent nitrate concentration remained below 2 mg N/L (Fig. 3). During this period at 5 C, biofilm formation on the reactor walls was more noticeable than during operation at temperature values above 5 C. Based on the experimental results of the temperature study, an effluent nitrate concentration of 10 mg N/L or less can be achieved even at a water temperature as low as 5 C, as long as sufficient degradable carbon is provided and well incorporated into the groundwater. It is noteworthy that the conditions used in this test differed from what was typically observed at the biosolids application site where the groundwater temperature did not change significantly throughout the year (it ranged between 18 and 20 C). In addition, during winter with ambient air temperature between 4 and 15 C, the temperature at the sediment/water interface in the pilot-scale wetland ranged from 5 to 20 C. Therefore, even during winter, the impact of temperature on the wetland performance of the pilot-scale wetland at the biosolids application site, where the lowest average monthly air temperature, usually in January, is about 8e9 C, is expected to be less drastic. Therefore, the results of the laboratory study at a water temperature as low as 5 C are conservative, but show the resilience and efficiency of the denitrification process at low temperature values.
3.2.
Denitrification kinetics
3.2.1. Effect of initial nitrate concentration under open and closed conditions the room temperature was decreased again and reached 10 C by day 55. There was a slight increase in the effluent nitrate concentration at this time, but within 24 h it returned to nondetectable levels. After the room temperature was reduced to 5 C by day 82, the reactor effluent concentration increased rapidly to a maximum 47 mg NO 3 eN/L, and then started to decrease. However, for over 35 days at 5 C, the reactor’s performance was not stable and the effluent nitrate concentration fluctuated between 10 and 35 mg NO 3 eN/L, albeit with a downwards trend (Fig. 3). In an attempt to achieve a stable effluent concentration, on day 120 the HRT was increased to 10 days. Although the effluent nitrate concentration decreased significantly at an HRT of 10 days, it continued to fluctuate between 5 and 20 mg NO 3 eN/L. Upon further observation, it was realized that MicroC G was not well mixed with the groundwater in the reactor as it was delivered intermittently by a micro pump every 2 h at the point where the groundwater was constantly pumped into the reactor (head of reactor). It appears that MicroC G has a low solubility at 5 C. Therefore, the unstable and poor performance of the reactor was attributed to lack of uniform electron donor distribution and thus availability. On day 172, a mixer was installed in the influent portion of the reactor and turned on by an electronic timer every 2 h while the MicroC G was fed, and for an additional 10 min after feeding was stopped. With intermittent mixing, even at 5 C, MicroC G was well incorporated into the reactor groundwater, which resulted in stable reactor performance with non-detectable effluent nitrate concentrations. On day 190, the HRT was returned to 5 days and the reactor operated
Denitrification at initial nitrate concentrations of 70, 140, 300 and 400 mg N/L was tested under conditions open to the atmosphere. The nitrate and nitrite concentrations in each assay over the incubation period are shown in Fig. 4. A lag period of approximately 1 day was observed in the first batch assay performed with site soil and groundwater, which was attributed to the very low active denitrifying population size of the surface soil used in these assays. Nitrate reduction proceeded immediately in all subsequent assays, indicating that some active biomass was retained in the soil despite the rinsing procedure performed in between each batch assay. Transient nitrite concentrations were observed in all assays, but after the complete removal of nitrate, nitrite was completely removed in less than 4 days, except in the first assay, in which nitrite reduction was slower and nitrite was removed in approximately 5 days. Similarly to the open batch assays, closed batch assays were conducted in serum bottles in the absence of oxygen (data not shown). A relatively low nitrate removal rate was also observed in the first 20 h of incubation, which is attributed to the low population size of active denitrifying bacteria in the soil. Significant nitrite levels were observed in series with an initial nitrate concentration of 140 mg NO 3 eN/L and above. Nevertheless, the nitrite reduction rate was fast and all series achieved complete denitrification in less than 5 days. Using the nitrate and nitrite reduction data and assuming a two-step model (nitrate to nitrite to dinitrogen), the MSRR values were estimated. Based on the data from the open batch assays and applying Monod kinetics with the biokinetic parameter values described in Section 2.4 above, the MSRR
5594
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
80
values were determined to be very similar in all assays conducted with different initial nitrate concentrations (Table 2). The best model fit to the experimental data is shown in Fig. 4. The root mean squared deviation (RMSD) at initial nitrate concentration of 70, 140, 300 and 400 mg N/L was 7.0, 11.1, 36.4 and 44.8, respectively. Sensitivity and identifiability analysis (Section 2.4, above) showed that the highest degree of sensitivity was for the MSRR, while the lowest degree of sensitivity was for the halfsaturation constant (see Supplementary Material, Fig. S2). Although the sensitivity analysis indicates that the halfsaturation constant cannot be uniquely identified from the provided data sets, the KSNO3 and KSNO2 values were estimated at 20.7 6.3 and 18.8 7.5 mg N/L for nitrate and nitrite, respectively. These values are apparent half-saturation constants, as opposed to intrinsic, which take into consideration any mass transfer limitations (Tugtas and Pavlostathis, 2007). Such limitations are expected for a system where most of the microbial activity occurred at the lower water layers, soil/water interface and within the soil matrix. Visual inspection lead to the observation that most N2 gas production was occurring in the top soil layer, indicated by the entrapment of gas bubbles and thick biofilm layer on the soil/water interface compared to a relatively clear reactor water column. When estimating denitrification kinetics for each initial nitrate concentration of 70, 140, 300 and 400 mg N/L, the initial biomass was assumed to be 5, 10, 10 and 15 mg VSS/L, respectively, due to increased retention of active biomass in the sediment despite flushing of the reactor in between each batch assay. Similarly to the open assay, the MSRR values for the closed batch assays for nitrate and nitrite were estimated to be 1.5 0.2 and 0.4 0.1 mg N/ mg VSSCOD-day, respectively. The KSNO3 and KSNO2 values were 16.1 2.8 and 20.5 6.2 mg N/L, respectively. Oxygen affects (increases) the electron donor requirement for denitrification; however, comparing the nitrate reduction rate achieved in closed systems (1.5 0.2 mg NO 3 eN/ mg VSSCOD-day) to that achieved in open to the atmosphere systems (1.2 0.2 mg NO 3 eN/mg VSSCOD-day) at an initial COD:N ratio of 6, the kinetics of nitrate reduction were not severely affected in the open to the atmosphere reactor. Therefore, as long as a bioavailable carbon source is supplied in excess of that required for the complete nitrate reduction, the nitrate reduction kinetics are not impacted by other alternative electron acceptors (e.g., oxygen for open systems) for similar systems (e.g., low mixing intensity and reaeration).
A Nitrate
60
Nitrite
40
20
0 150
B
125 100 75
NITROGEN (mg N/L)
50 25 0 300
C
250 200 150 100 50 0 400
D
300
200
100
0 0
1
2
3
4
5
6
TIME (Days) Fig. 4 e Measured (data points) and simulated (lines) nitrogen species in batch assays conducted with an open to the atmosphere reactor at an initial nitrate concentration of (A) 70, (B) 140, (C) 300 and (D) 400 mg N/L using MicroC G as the carbon source at an initial COD:N ratio of 6:1.
Table 2 e Estimated maximum specific reduction rate (MSRR; k) and half-saturation constant (KS) values for batch assays conducted with an open to the atmosphere reactor and four initial nitrate concentrations (70, 140, 300 and 400 mg N/L). Parameter kNO3 (mg N/mg VSSCOD-day) KSNO3 (mg N/L) kNO2 (mg N/mg VSSCOD-day) KSNO2 (mg N/L) a Estimate standard deviation. b Range.
Value 1.2 0.2a 20.7 6.3 0.7 0.1 18.8 7.5
(1.0e1.5)b (16.5e30.0) (0.6e0.9) (15.0e30.0)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
A
The laboratory reactors were static and mixing was minimal. In both the continuous-flow and batch assays, most of the microbial activity was in the soil/water interface and top soil layer where dissolved oxygen concentrations were typically less than 1 mg/L and maintained only by diffusion from the free-water surface. Alternatively, under field conditions, a higher rate of aeration is expected (e.g., wind action), which may negatively impact the nitrate reduction rate as a result of a higher competition for the carbon/electron source between oxygen and nitrate-reducing processes.
3.2.2.
B
C
D
Fig. 5 e Measured (data points) and simulated (lines) nitrogen species in batch assays conducted with an open to the atmosphere reactor maintained at (A) 22, (B) 15, (C)
5595
Effect of temperature
In order to assess the effect of temperature on denitrification kinetics, batch assays were performed at target temperatures of 5, 10, 15 and 22 C in a reactor open to the atmosphere (Fig. 5). Similarly to the previous batch assays, a lag of approximately 1 day was observed for nitrate removal. The rates of both nitrate and nitrite reduction were very similar at 22 and 15 C, with a maximum transient nitrite concentration slightly higher at 15 C. At 10 C, a significantly lower rate of nitrate and nitrite reduction was observed; however, at 5 C, the time required for the complete removal of nitrate and nitrite was more than double of that under 10 C. The nitrate and nitrite reduction rates at 5 C were much lower than at the other three temperature values, and the maximum transient nitrite concentration was the lowest. Therefore, the nitrate reduction rates were not severely affected until the temperature dropped below 10 C, which agrees with the findings of other studies on denitrification at low temperature values (Burgoon, 2001; Darbi and Viraraghavan, 2004; Hajaya et al., 2011; Lee et al., 2009). Hajaya et al. (2011) reported a decrease of more than 50% in the nitrate and nitrite reduction rates for a temperature decrease from 15 to 10 C. Studies investigating the seasonal temperature effects on denitrification in wetland systems have reported significant decrease in denitrification rates at temperature values below 15 C (Kadlec and Reddy, 2001; Poe et al., 2003). Using the nitrate reduction data, estimated biokinetic parameters and assuming a two-step denitrification model as previously discussed in Section 2.4 above, the MSRR values were estimated and are summarized in Table 3. The half-saturation constants were similar at all temperature values; KSNO3 and KSNO2 were estimated as 57.0 4.8 and 16.5 2.4 mg N/L, respectively. As expected, the MSRR increased with increasing temperature; however, the MSRR values were very similar at 15 and 22 C, which agrees with previous reports stating that the optimum temperature for nitrate reduction is closer to 15 than 22 C, and that in this temperature range, variations in temperature only slightly affect denitrification rates (Hajaya et al., 2011; Kristiansen, 1983; Lee et al., 2009). The model accurately simulated the nitrate reduction at all four temperature values; however, as temperature decreased, the discrepancy between the model and experimental nitrite concentrations increased. At lower temperature values the model overestimated the 10, and (D) 5 C with groundwater at an initial nitrate concentration of 150 mg N/L using MicroC G as the carbon source at an initial COD:N ratio of 6:1 (note the different xaxis scale of panel D).
5596
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
Table 3 e Estimated maximum specific reduction rate (MSRR; k) values for batch assays conducted with an open to the atmosphere reactor maintained at different temperatures. Temperature ( C)
kNO3
kNO2
RMSDa
mg N/mg VSSCOD-day 5 10 15 22
0.41 0.87 1.61 1.64
0.01b 0.01 0.02 0.03
0.20 0.31 0.50 0.53
0.02 0.03 0.04 0.50
23.7 19.7 26.1 15.5
a Root mean square deviation qX ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ ðmeasured value estimated valueÞ2 : b Estimate standard deviation.
experimentally measured nitrite concentrations, with the deviation increasing with decreasing temperature, indicated by the increasing RMSD values (Table 3). The best model fit for the four temperature batch assays is shown in Fig. 5. Based on the MSRR values at the four temperature values tested and the modified Arrhenius model (equation (4)), the dimensionless temperature coefficient (q) was estimated. By taking 15 C as the basis, the temperature coefficient was 1.13 for nitrate reduction, a value which is within the temperature coefficient range of 1.04 and 1.16 reported by Kadlec and Reddy (2001) for removal of nitrate in wetlands. Cherchi et al. (2009) reported a temperature coefficient of 1.11 for nitrate reduction in a chemostat using MicroC G as the carbon source. Kadlec and Wallace (2009) also reported a mean temperature coefficient of 1.11 for nitrate reduction in wetland systems with temperatures ranging from 6 to 24 C. Temperature coefficient values of 1.07e1.14 were estimated for bacteria in biofilms and artificially encapsulated at temperatures as low as 3 C, indicating that the temperature dependence of microbial activity is not affected in immobilized bacteria (Vackova et al., 2011; Welander and Mattiasson, 2003). Thus, the q value found in the present study agrees well with those previously reported.
3.3.
Continuous-flow system simulation
Based on the estimated denitrification kinetic rates, the continuous-flow wetland system was simulated using the series of differential equations and parameters described in Section 2.5, above. Model conditions were chosen based on the results of both the batch and continuous-flow laboratoryscale experiments. The obtained MSSR values indicated that nitrate reduction rates were not affected by initial nitrate concentration and trial simulations showed that the effluent nitrate concentration at high COD:N ratios was not a function of influent nitrate concentrations, agreeing with Monod kinetics. As a result, for all simulations the value for SNO3 ;o was kept constant at 150 mg N/L. Nitrite was not detected in the site groundwater and the groundwater biomass was assumed to be negligible (i.e., SNO2 ;o and XNOx ;o were assumed to be 0). The influent electron donor concentration, Co, was adjusted to simulate different COD:N ratios.
In addition to HRT, COD:N ratio and temperature, microbial biomass retention is another variable parameter for a continuous-flow system. Full-scale wetland systems are generally rich with various types of biomass, such as trees, cattails, algae and bacteria, most of which is retained within the system, providing nutrients and serving as a carbon source to drive biological processes and cell growth (Kadlec and Wallace, 2009). Retention of the denitrifying biomass ultimately controls the solids retention time (SRT) of the system, which is usually very high and not easily controlled in fullscale wetland systems. To simulate a wetland system with a high degree of microbial biomass retention, the biomass retention factor (b) was introduced into the system biomass mass balance equation as follows: ! dXNOx ðXNOx ;o bXNOx Þ YNO3 kNO3 SNO3 C ¼ XNOx dt kSNO3 þ SNO3 s KC þ C ! YNO2 kNO2 SNO2 C XNOx bXNOx þ KSNO2 þ SNO2 KC þ C
(9)
The factor b takes values from 0 (i.e., 100% biomass retention) to 1 (i.e., 0% biomass retention). The pilot-scale wetland system at the biosolids application site was open to the atmosphere and plant and vegetation growth was significant in all cells, likely contributing significant biodegradable carbon to the system and increasing retention of the denitrifying biomass. Alternatively, the laboratory-scale system was constructed with only soil and groundwater, with MicroC G serving as the sole carbon source in all laboratory assays. Although there was no vegetation, most of the biomass was retained in the laboratoryscale continuous-flow reactors, with minimum amounts released in the effluent streams. Because the fraction of biomass retained could not be quantified accurately, a value of 75% biomass retention (i.e., b ¼ 0.25) was assumed for all continuous-flow simulations. Using this biomass retention value and an HRT of 5 days at 22 C, the effluent steady-state nitrogen concentration (nitrate and nitrite) was predicted to be approximately 5 mg N/L using model simulation. Alternatively, assuming no biomass retention (i.e., b ¼ 1) and operating under the same conditions (HRT of 5 days at 22 C), the effluent steady-state nitrogen concentration was predicted to be approximately 13.5 mg N/L. The steady-state effluent nitrate concentration in the laboratory reactor, Run 1, under the same conditions ranged from 0 to 3 mg N/L, which is slightly lower than the simulation results. Therefore, these results verify that the assumption for biomass retention is reasonable and that the model closely predicts the performance of the laboratory continuous-flow system. Using the continuous-flow design equations (equations (5) through (7) and (9)), the system performance was evaluated as a function of HRT, temperature and electron donor availability (i.e., COD:N ratio). System performance is expressed as nitrogen removal efficiency (%), which includes both nitrate and nitrite in the effluent. Other possible denitrification intermediates, NO and N2O, which are not included in the model and were not detected in any of the continuous-flow runs, are assumed to react rapidly and therefore cannot be detected (Kornaros et al., 1996). Thus, the nitrogen removal efficiency was evaluated using the following relationship:
5597
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
0
REMOVAL (%)
A 100 80 60 40 20 06
5
CO 4 3 2 D:N 1
NITROGEN REMOVAL (%) 20 40 60
80
B
C
100 80 60 40 20 06 5 15 CO 4 3 2 1113 9 7 D:N 1 ) 5 s ay 13 T (D
HR
100
100 80 60 40 20 06 5 15 1113 CO 4 3 2 9 7 D:N 1 5 ) s 13 (Day
HRT
23 1720 ) 1114 5 8 MP. (°C E
T
Fig. 6 e Model simulation of the effect of COD:N ratio and HRT on nitrogen removal in a continuous-flow system at (A) 5 and (B) 15 C; (C) effect of COD:N ratio (1e6) and temperature (5e22 C) on nitrogen removal at an HRT of 5 days (75% denitrifying biomass retention is assumed in all simulations).
Nitrogen Removal Efficiency ð%Þ ¼
SNO3 ;o SNO3 þ SNO2 100 SNO3 ;o (10)
Fig. 6A and B shows the effect of COD:N and HRT on nitrogen removal efficiency of the system at 5 and 15 C, respectively. Assuming that 75% of biomass is retained (i.e., b ¼ 0.25), changes in the HRT only slightly affect nitrogen removal efficiency. Due to the high SRT, only at low HRT values, below 3 days, is the system performance impacted. Similarly, Fig. 6C illustrates the system nitrogen removal efficiency (%) as a function of COD:N ratio and temperature for values ranging from 1 to 6 and 5 to 15 C, respectively. System performance is only slightly impacted at a decreased temperature and is more severely impacted at COD:N ratio values below 3. The experimental results of the laboratory system and simulations indicate that denitrification can be successful in free-water surface wetland systems at low temperature and HRT values as long as there is enough biodegradable carbon to achieve complete denitrification.
4.
Based on results obtained with the laboratory-scale continuous-flow system and model simulations, high nitrogen removal efficiencies (>90%) can be consistently achieved and maintained by free-water surface wetland systems while treating nitrate-contaminated groundwater with high nitrate levels with an HRT of 3 days or higher and at temperature values as low as 5 C, as long as there is sufficient biodegradable carbon available to achieve complete denitrification. Influent COD:N ratios should be maintained at 3:1 and higher. The results of this study show that constructed wetland technology is a technically feasible and attractive alternative for the treatment of nitrate-bearing groundwater at the biosolids application site.
Acknowledgments This research was supported by the Columbus Water Works (CWW), Columbus, GA through Jordan, Jones and Goulding, Inc. (JJG), Norcross, GA. Special thanks to Camp, Dresser and McKee, Inc. (CDM) for a graduate fellowship to T. Misiti.
Conclusions
A laboratory-scale system was designed and developed to simulate a pilot-scale, free-water surface constructed wetland system proposed for the treatment of nitrate-contaminated groundwater at a biosolids application site. Although oxygen can increase the electron donor requirement for denitrification, this study showed that fast nitrate reduction rates can be achieved even in systems open to the atmosphere, as long as the electron/carbon source is not limiting. The kinetics of nitrate reduction in open to the atmosphere reactors were not severely affected by oxygen competition at initial nitrate concentrations as high as 400 mg N/L. The rate of nitrate reduction was not affected by nitrate concentrations as high as 400 mg N/L, but decreased with decreasing temperature; however, even at temperature values as low as 5 C, complete denitrification occurred in both batch and continuous-flow systems as long as sufficient biodegradable carbon was bioavailable (dissolved).
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.08.019.
references
American Public Health Association, 2005. Standard Methods for the Examination of Water and Wastewater, 21st ed. APHAAWWA-WEF, Washington, D.C. Bachand, P.A.M., Horne, A.J., 2000. Denitrification in constructed free-water surface wetlands: I. Very high nitrate removal rates in a macrocosm study. Ecological Engineering 14 (1e2), 9e15. Burgoon, P.S., 2001. Denitrification in free water surface wetlands receiving carbon supplements. Water Science and Technology 44 (11e12), 163e169.
5598
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 8 7 e5 5 9 8
Burkart, M.R., Stoner, J.D., 2002. Nitrate in aquifers beneath agricultural systems. Water Science and Technology 45 (9), 19e28. Cherchi, C., Onnis-Hayden, A., El-Shawabkeh, I., Gu, A.Z., 2009. Implication of using different carbon sources for denitrification in wastewater treatments. Water Environment Research 81 (8), 788e799. Darbi, A., Viraraghavan, T., 2004. Effect of low temperature on denitrification. Fresenius Environmental Bulletin 13 (3B), 279e282. Environmental Operating Solutions, 2008. Material Safety Data Sheet e EOS MicroC GTM. http://www.eosenvironmental.com/ product/microc_g.htm (accessed 01.12.10). Bourne, Massachusetts. Gujer, W., 2008. Systems Analysis for Water Technology. Springer, Berlin, Germany. Hajaya, M.G., Tezel, U., Pavlostathis, S.G., 2011. Effect of temperature and benzalkonium chloride on nitrate reduction. Bioresource Technology 102 (8), 5039e5047. Heinen, M., 2006. Simplified denitrification models: overview and properties. Geoderma 133 (3e4), 444e463. Inamori, R., Wang, Y., Yamamoto, T., Zhang, J., Kong, H., Xu, K., Inamori, Y., 2008. Seasonal effect on N2O formation in nitrification in constructed wetlands. Chemosphere 73 (7), 1071e1077. Kadlec, R.H., Reddy, K.R., 2001. Temperature effects in treatment wetlands. Water Environment Research 73 (5), 543e557. Kadlec, R.H., Wallace, S.D., 2009. Treatment Wetlands, second ed. CRC-Press, Boca Raton, FL. Kjellin, J., Hallin, S., Worman, A., 2007. Spatial variations in denitrification activity in wetland sediments explained by hydrology and denitrifying community structure. Water Research 41 (20), 4710e4720. Kornaros, M., Zafiri, C., Lyberatos, G., 1996. Kinetics of denitrification by Pseudomonas denitrificans under growth conditions limited by carbon and/or nitrate or nitrite. Water Environment Research 68 (5), 934e945. Kristiansen, S., 1983. The temperature optimum of the nitrate reductase assay for marine-phytoplankton. Limnology and Oceanography 28 (4), 776e780. Lee, C.G., Fletcher, T.D., Sun, G.Z., 2009. Nitrogen removal in constructed wetland systems. Engineering in Life Sciences 9 (1), 11e22. Macey, R.I., Oster, G.F., 2006. Berkeley Madonna, 8.3. Berkeley Madonna Inc. Maltais-Landry, G., Maranger, R., Brisson, J., 2009a. Effect of artificial aeration and macrophyte species on nitrogen cycling and gas flux in constructed wetlands. Ecological Engineering 35 (2), 221e229. Maltais-Landry, G., Maranger, R., Brisson, J., Chazarenc, F., 2009b. Greenhouse gas production and efficiency of planted and artificially aerated constructed wetlands. Environmental Pollution 157 (3), 748e754.
Maltais-Landry, G., Maranger, R., Brisson, J., Chazarenc, F., 2009c. Nitrogen transformations and retention in planted and artificially aerated constructed wetlands. Water Research 43 (2), 535e545. Murgulet, D., Tick, G.R., 2009. Assessing the extent and sources of nitrate contamination in the aquifer system of southern Baldwin County, Alabama. Environmental Geology 58 (5), 1051e1065. Poe, A.C., Piehler, M.F., Thompson, S.P., Paerl, H.W., 2003. Denitrification in a constructed wetland receiving agricultural runoff. Wetlands 23 (4), 817e826. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology: Principles and Applications. McGraw-Hill, New York, NY. Rivett, M.O., Buss, S.R., Morgan, P., Smith, J.W.N., Bemment, C.D., 2008. Nitrate attenuation in groundwater: a review of biogeochemical controlling processes. Water Research 42 (16), 4215e4232. Surampalli, R.Y., Lai, K.C.K., Banerji, S.K., Smith, J., Tyagi, R.D., Lohani, B.N., 2008. Long-term land application of biosolids e a case study. Water Science and Technology 57 (3), 345e352. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. Wastewater Engineering: Treatment and Reuse, fourth ed. McGraw-Hill, Boston. Tugtas, A.E., Pavlostathis, S.G., 2007. Electron donor effect on nitrate reduction pathway and kinetics in a mixed methanogenic culture. Biotechnology and Bioengineering 98 (4), 756e763. United States Environmental Protection Agency (US EPA), 2000a. Biosolids Technology Fact Sheet e Land Application of Biosolids. Office of Water, Washington, D.C. United States Environmental Protection Agency (US EPA), 2000b. Constructed Wetlands Treatment of Municipal Wastewaters. Office of Research and Development, Cincinnati, Ohio. United States Environmental Protection Agency (US EPA), 2009. Drinking Water Contaminants. http://water.epa.gov/drink/ contaminants/index.cfm (accessed 31.05.11). Vackova, L., Srb, M., Stloukal, R., Wanner, J., 2011. Comparison of denitrification at low temperature using encapsulated Paracoccus denitrificans, Pseudomonas fluorescens and mixed culture. Bioresource Technology 102 (7), 4661e4666. Welander, U., Mattiasson, B., 2003. Denitrification at low temperatures using a suspended carrier biofilm process. Water Research 37 (10), 2394e2398. Wu, J., Zhang, J., Jia, W.L., Xie, H.J., Gu, R.R., Li, C., Gao, B.Y., 2009. Impact of COD/N ratio on nitrous oxide emission from microcosm wetlands and their performance in removing nitrogen from wastewater. Bioresource Technology 100 (12), 2910e2917. Zumft, W., 1997. Cell biology and molecular basis of denitrification. Microbiology and Molecular Biology Reviews 61 (4), 533e616.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Diversity and antibiotic resistance of Aeromonas spp. in drinking and waste water treatment plants Vaˆnia Figueira, Ivone Vaz-Moreira, Ma´rcia Silva, Ce´lia M. Manaia* CBQF/Escola Superior de Biotecnologia, Universidade Cato´lica Portuguesa, R. Dr. Anto´nio Bernardino de Almeida, 4200-072 Porto, Portugal
article info
abstract
Article history:
The taxonomic diversity and antibiotic resistance phenotypes of aeromonads were
Received 14 June 2011
examined in samples from drinking and waste water treatment plants (surface, ground
Received in revised form
and disinfected water in a drinking water treatment plant, and raw and treated waste
13 August 2011
water) and tap water. Bacteria identification and intra-species variation were determined
Accepted 13 August 2011
based on the analysis of the 16S rRNA, gyrB and cpn60 gene sequences. Resistance
Available online 22 August 2011
phenotypes were determined using the disc diffusion method.
Keywords:
water, and Aeromonas media and Aeromonas puntacta in waste water. No aeromonads were
Aeromonas
detected in ground water, after the chlorination tank or in tap water. Resistance to cefta-
Quinolone resistance
zidime or meropenem was detected in isolates from the drinking water treatment plant
Surface water
and waste water isolates were intrinsically resistant to nalidixic acid. Most of the times,
Waste water
quinolone resistance was associated with the gyrA mutation in serine 83. The gene qnrS,
Aeromonas veronii prevailed in raw surface water, Aeromonas hydrophyla in ozonated
but not the genes qnrA, B, C, D or qepA, was detected in both surface and waste water isolates. The gene aac(6’)-ib-cr was detected in different waste water strains isolated in the presence of ciprofloxacin. Both quinolone resistance genes were detected only in the species A. media. This is the first study tracking antimicrobial resistance in aeromonads in drinking, tap and waste water and the importance of these bacteria as vectors of resistance in aquatic environments is discussed. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The genus Aeromonas comprises ubiquitous bacteria, considered indigenous to aquatic environments (Janda and Abbott, 2010). Members of this genus are able to inhabit surface water (rivers, lakes), sewage, drinking water (tap and bottled mineral), thermal waters and sea water (Biscardi et al., 2002; Maalej et al., 2003; Pablos et al., 2009). Some species, mainly the psychrophilic Aeromonas salmonicida and the mesophilic Aeromonas hydrophila and Aeromonas veronii are recognized
causative agents of fish disease (Janda and Abbott, 2010). Aeromonas spp. are also important human opportunistic pathogens with ability to cause various types of diseases, which include intestinal, blood, skin and soft tissue and trauma-related infections (Aminov, 2009; Lamy et al., 2009; Janda and Abbott, 2010). Among the leading pathogenic species are A. hydrophila, Aeromonas caviae (later synonym of Aeromonas punctata) and A. veronii (Lamy et al., 2009). The environmental ubiquity associated with the potential pathogenicity of these bacteria has been illustrated also in recent
* Corresponding author. Tel.: þ351 22 5580059; fax: þ351 22 5090351. E-mail address: [email protected] (C.M. Manaia). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.021
5600
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
natural disasters (Dixon, 2008). Evidences for the waterhuman transmission of Aeromonas spp. are also available (Khajanchi et al., 2010). Over the last years, a greater public awareness and scientific understanding of antimicrobial resistance contributed to consider environmental reservoirs and paths of dissemination as critical points for antimicrobial resistance control. Among such reservoirs and paths of dissemination, water environments in which enter antibiotic resistant organisms from human and animal sources play a pivotal role (Baquero et al., 2008; Ku¨mmerer, 2009; Taylor et al., 2011). The importance of municipal waste water treatment plants as sources of antimicrobial resistance and the risks of contamination of surface waters has been demonstrated in numerous publications (e.g. Go˜ni-Urriza et al., 2000; Ferreira da Silva et al., 2007; Servais and Passerat, 2009; Novo and Manaia, 2010). As a consequence of surface and ground water contamination, emerges the hypothesis that antimicrobial resistance can reach drinking water, serving as a vehicle of resistance transfer to the water consumers. Indeed, antimicrobial resistance has been detected in drinking water (Schwartz et al., 2003; Faria et al., 2009; Xi et al., 2009; Vaz-Moreira et al., 2011). Considering this urban water cycle, ubiquitous bacteria, which can colonize different types of water, are of special interest to assess potential forms of antimicrobial resistance dissemination. Given their ubiquity and patterns of acquired antimicrobial resistance, members of the genus Aeromonas are good example of such bacteria. In a recent comprehensive review on the genus Aeromonas, Janda and Abbott (2010) alerted for the little attention given to the general low susceptibility of aeromonads to various classes and combinations of antimicrobial agents. Nevertheless, the potential of aeromonads to develop and disseminate antibiotic resistance either in clinical settings or in the environment has been demonstrated in numerous publications (Walsh et al., 1997; Go˜ni-Urriza et al., 2000; Huddleston et al., 2006; Blasco et al., 2008; Cattoir et al., 2008; Gordon et al., 2008; Lamy et al., 2009; Arias et al., 2010a,b). Moreover, recent and emerging antibiotic resistance seems to be common in different species of Aeromonas. For instance, different variants of the plasmid-mediated quinolone resistance qnr gene were detected in environmental isolates of the species A. punctata, Aeromonas media or Aeromonas allosaccharophila (Cattoir et al., 2008; Pica˜o et al., 2008; Xia et al. 2010). In spite the ubiquity of aeromonads in aquatic environments and the likelihood to develop antimicrobial resistance, the ecology and patterns of resistance of members of this genus present in drinking and waste water treatment plants has not been addressed in scientific literature. The current work aimed at filling this gap and was based on the hypothesis that Aeromonas spp. could serve as a vehicle for antibiotic resistance dissemination within the urban water cycle. According to this hypothesis, this work was designed to track aeromonads and their antibiotic resistance profiles in different parts of the urban water cycle. Our main goal was the identification of major sources of antibiotic resistant aeromonads and of critical points for their elimination. Specifically, it was intended to i) identify the different aquatic environments within the urban water cycle where aeromonads are more prevalent, and possible factors contributing
for their reduction; ii) determine the predominant species in each of those environments; iii) infer about the possible relationship Aeromonas species-antibiotic resistance pattern.
2.
Materials and methods
2.1.
Water sampling
Water samples were collected from different aquatic environments within an urban water cycle in the region of Northern Portugal (Fig. 1). These sites included: i) a drinking water treatment plant (WTP) and the respective distribution system, which supplies water to a population of about 1.5 million of inhabitants; ii) household tap water, served by the WTP and iii) a municipal waste water treatment plant (WWTP), serving 85 000 inhabitant equivalents, in the same geographical area (Ferreira da Silva et al., 2006; Vaz-Moreira et al., 2011). In the WTP, samples were collected from raw surface water (A), ground water (alluvial wells) (B), after sand filtration and ozonation (C) and after chlorination (D), the drinking water final treatment step. Four samples were collected downstream the WTP, in the drinking water distribution system respectively, after the first or the second re-chlorination stations (EH). Tap water samples were collected in eleven houses (I-S), from taps used 1e4 times a month, served by the WTP referred to above and situated within an area of 25 km (Vaz-Moreira et al., 2011). Waste water samples corresponded to raw (T) and treated waste water (U). Sites A to H (drinking water treatment plant and distribution system) were sampled in November 2007 and in September 2009, sites I to S (taps) were sampled in April, July and October 2009 and sites T and U (waste water treatment plant) were sampled nine times between November 2004 and November 2009.
2.2. Isolation, enumeration and preliminary identification This work was integrated in a wider study designed to assess the diversity and antibiotic resistance of culturable bacteria, belonging to different groups, present in selected niches within the urban water cycle. For the microbiological characterization of the water samples it was used the membrane filtration method, as described before for waste and surface water (Ferreira da Silva et al., 2006; Vaz-Moreira et al., 2011). Volumes of 10e500 ml (WTP, drinking water distribution system, taps) or of 1e10 ml (WWTP) of water samples or decimal dilutions thereof were filtered through cellulose nitrate membranes (0.45 mm pore size, 47 mm diameter, Albet), which were placed onto different culture media and incubated up to 7 days. No selective culture medium for aeromonads was used. Aeromonas spp. analysed in this study were isolated among the culturable heterotrophs recovered on different culture media - Plate Count Agar (PCA, Pronadisa), on m-endo-agar-LES (Difco), on mFC agar (Difco), on Tergitol-7 agar (TTC, Oxoid), on Pseudomonas Isolation Agar (PIA, Difco), on R2A agar (Difco) or on Bile Esculin Agar (BEA, Pronadisa). The culture media PCA, PIA, BEA and R2A were incubated at 30 C and mFC agar, m-endo-agar-LES and TTC were incubated at 37 C. Given this work was designed to
5601
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
B (101 CFU mL-1) 2
C (101-103 CFU mL-1) O3 4 O3 O3
3
5
6 Cl 7 Cl
1
-
-
A (103 CFU mL-1) 8
D (10-2-101 CFU mL-1)
WWTP 10 T 11 1 – Raw Surface Water 2 – Ground Water 3 – Filtration 4 – Ozonation
(105-106
9
CFU
Water Distribution E-H (10-2-102 CFU mL-1) System 4 1 -1 (10 -10 CFU mL )
mL-1)I-S
U (104-105 CFU mL-1)
5 – Coagulation/Flocculation 6 - Flotation and Filtration 7 – Chlorination 8 – Treated Water Reservoir
9 – Drinking Water 10 – Raw Waste Water 11 – Treated Waste Water
Fig. 1 e Schematic representation of the drinking and waste water treatment plants, indicating the treatment stages and sampled sites (AeU). Ground water disinfection involves only stages 6e7. CFU mLL1 corresponds to culturable heterotrophic counts in the plates from which the aeromonads were isolated.
recover culturable bacteria from different bacterial groups, representatives of all colony types were selected for further culture isolation and purification, according to the following criterion: all colonies of morphotypes represented by less than five colonies, half of the colonies of morphotypes represented by five to 10 colonies, and about one third of colonies with morphotypes represented by more than 10 colonies. Cytochrome c oxidase positive isolates with the morphology of Gram-negative rods and forming yellow colonies on GSP agar (Merck) at 30 C were presumptively identified as aeromonads (Corry et al., 2003; Abulhamd, 2009; Kivanc et al., 2011). This set of aeromonads included a total of 121 isolates, from which 72 and 8 were isolated, respectively, from the first and second sampling campaigns in the WTP; 1, 3, 2, 6, 1, 1, 5 and 13 were isolated, respectively, from each sampling date in the WWTP. A group of 9 WWTP isolates recovered on PCA or mFC agar supplemented with 4 mg L1 ciprofloxacin (AppliChem) (Novo and Manaia, 2010) were also examined in this study.
2.3. Identification at the species level and determination of intra-species variation Identification at the species level was based on the analysis of the 16S rRNA gene sequence and intra-species variation was assessed on basis of the comparison of two additional housekeeping genes, gyrB and cpn60 (Ya´n˜ez et al., 2003; Min˜ana-Galbis et al., 2009). PCR amplifications of fragments of the genes 16S rRNA, gyrB and cpn60 were conducted using the primers and the conditions described before (Table 1). PCR products were purified with GFX PCR DNA purification kit (GE Healthcare) and the nucleotide sequences were determined. Partial nucleotide sequences of the genes 16S rRNA, gyrB and cpn60 were aligned using Clustal W from MEGA 4.0 software (Tamura et al., 2007) and compared with the homologous
sequences of the type strains of the different Aeromonas species, available in the GenBank database. For the gene cpn60, the nucleotide sequences of the type strains of the species Aeromonas sanarellii and Aeromonas taiwanensis were not available in the GenBank database, and were determined in this study using the type strains provided by BCCM/LMG culture collection with the numbers LMG 24682T and LMG 24683T, respectively. These nucleotide sequences were deposited in the GenBank database with the accession numbers JF920655 and JF920656 for A. sanarellii and A. taiwanensis, respectively. Nucleotide sequence relatedness was estimated based on the model of Jukes and Cantor (1969) and dendrograms were created using the neighbour-joining method. The maximumlikelihood method was also applied to assess tree stability. In the analysis were used 1283, 779 and 555 nucleotide positions of the 16S rRNA, gyrB and cpn60 gene sequences, respectively. Non-repetitive nucleotide sequences were deposited in the GenBank database with the accession numbers JF920473JF920563, JF938599-JF938689, and JF920564-JF920654 for 16S rRNA, gyrB and cpn60 sequences, respectively. In an attempt to discriminate strains of the same species, each pair of isolates was compared based on the nucleotide sequence of each of the three genes. Strains differing at least in a nucleotide position were classified as representing distinct sequence types (ST). This comparison was represented in a dendrogram constructed based on 2617 nucleotide positions of the concatenated sequences of 16S rRNA, cpn60 and gyrB genes (Fig. 2).
2.4.
Determination of antibiotic resistance phenotypes
The susceptibility to 12 antibiotics was determined using the agar diffusion method (CLSI, 2007). The antibiotics tested were nalidixic acid (NA, 30 mg); ciprofloxacin (CIP, 5 mg); amoxicillin
5602
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
Table 1 e Primers and PCR conditions used. Gene 16S rRNA gyrB cpn60 gyrA parC qnrA qnrB qnrS qnrC qnrD aac(6’)-Ib qepA cphA blaTEM
Sequence
Fragment length (bp)
Annealing temp. ( C)
27F 1492R gyrB-3F gyrB-14R C175 C938 AsalgyrAF AsalgyrAR AcparCF EcparCR qnr Am F qnrAm R qnrBm F qnrBm R qnrSm F qnrSm R qnrC-F qnrC-R qnrD-F qnrD-R aac(6)-F aac(6)-R qepA-F qepA-R cphA-F cphA-R blaTEM fw blaTEM rv
GAGTTTGATCCTGGCTCAG TAC CTT GTT ACG ACT T TCCGGCGGTCTGCACGGCGT TTGTCCGGGTTGTACTCGTC GAAATYGAACTGGAAGACAA GTYGCTTTTTCCAGCTCCA TCCTATCTTGATTACGCCATG CATGCCATACCTACCGCGAT GTTCAGCGCCGCATCATCTAC TTCGGTGTAACGCATTGCCGC AGAGGATTTCTCACGCCAGG TGCCAGGCACAGATCTTGAC GGMATHGAAATTCGCCACTG TTTGCYGYYCGCCAGTCGAA GCAAGTTCATTGAACAGGGT TCTAAACCGTCGAGTTCGGCG GGGTTGTACATTTATTGAATC TCCACTTTACGAGGTTCT CGAGATCAATTTACGGGGAATA AACAAGCTGAAGCGCCTG TTGCGATGCTCTATGAGTGGCTA CTCGAATGCCTGGCGTGTTT TGGTCTACGCCATGGACCTCA TGAATTCGGACACCGTCTCCG TCTATTTCGGGGCCAAGGG TCTCGGCCCAGTCGCTCTTCA ATAAAATTCTTGAAGACGAAA GACAGTTACCAATGCTTAATCA
1465
55 C
Lane, 1991
1130
58 C
Ya´n˜ez et al., 2003
763
52 C
Min˜ana-Galbis et al., 2009
481
50 C
Gon˜i-Urriza et al., 2002
245
54 C
Gon˜i-Urriza et al., 2002
580
54 C
Cattoir et al., 2007
264
54 C
Cattoir et al., 2007
428
54 C
Cattoir et al., 2007
447
50 C
Wang et al., 2009
582
55 C
Cavaco et al., 2009
482
55 C
Park et al., 2006
1137
56 C
Pe´richon et al., 2007
230
55 C
Balsalobre et al., 2009
939
55 C
DiPersio et al., 2005
(AML, 25 mg); ticarcillin (TIC, 75 mg); cephalothin (CP, 30 mg); ceftazidime (CEF, 30 mg); streptomycin (STR, 10 mg); sulphamethoxazole/trimethoprim (SXT, 25 mg); tetracycline (TET, 30 mg); gentamicin (GEN, 10 mg); colistin sulphate (CT, 50 mg) and meropenem (MER, 10 mg). For the antibiotics AML and CT, which are not included in the CLSI list, were used the following criteria: S 21/R < 14 and S 10/R < 10, respectively. Whenever diameters larger than R but smaller than S were observed, were referred to as intermediary, and were excluded from the resistance percentage calculations. The strains Escherichia coli ATCC 25922 and Pseudomonas aeruginosa DSM 1117 (¼ATCC 27853) were included as quality controls.
2.5.
Reference
Primers
(qnrA1 þ), Klebsiella pneumoniae B1 (qnrB1 þ) and Enterobacter cloacae S1 (qnrB4þ and qnrS1þ) were used as positive controls for the presence of the determinants qnrA, qnrB and qnrS. The strains E. coli DH10B transformant pHS11 and E. coli DH10B transformant p2007057 were used as positive controls for qnrC and qnrD, respectively. Salmonella enterica serovar typhimurium GSS-HN-2007-003 was used as positive control (Xia et al., 2009) for the presence of gene aac(6’)-Ib. Strain E. coli TOP10 þ pAT851 was used as positive control for gene qepA. PCR products were purified and the nucleotide sequences were determined and compared. A representative of each distinct nucleotide sequence was deposited in the GenBank (JF938596-JF938598).
Screening of resistance genetic determinants 2.6.
Mutations in the chromosomal genes gyrA and parC and the presence of resistance genes qnrA, qnrB, qnrS, qnrC, qnrD, aac(6’)-Ib and qepA were screened in quinolone resistant isolates. The presence of genes cphA and blaTEM associated with beta-lactam resistance were screened in all isolates. The primers and PCR conditions used were described before (Table 1). Point mutations in the genes gyrA and parC were identified after comparison with homologous nucleotide sequences of quinolone susceptible strains available in the GenBank - A. punctata CIP 7616T (AY027899 and AF435418), A. hydrophila subsp. hydrophila CIP 7614T (AY027901 and AF435419) and Aeromonas sobria CIP 7433T (AY027900 and AF435420) as described before (Gon˜i-Urriza et al., 2002). Strains E. coli L0
Statistical analysis
The chi-squared test was used to compare the prevalence values of antibiotic resistance phenotypes or genotypes and sequence types in different water sampled sites (SPSS 19.0 for Windows, SPSS Inc., Chicago, IL).
3.
Results
3.1.
Species diversity and intra-species variation
A collection of 741 Gram-negative cytochrome c oxidasepositive isolates recovered on the culture media and conditions described above was screened for the presence of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
5603
Aeromonas. On basis of 16S rRNA gene sequence analysis, among these isolates, 121 (80 from the WTP, 32 from WWTP and nine from WWTP isolated in the presence of ciprofloxacin) were identified as belonging to 11 species of the genus Aeromonas. The other strains were affiliated to genera such as Pseudomonas, Ralstonia, Comamonas, Acidovorax, Brevundimonas, Cupriavidus, Chryseobacterium, Achromobacter and to the family Sphingomonadacae. The group of Aeromonas spp. isolates comprised 80 from the drinking water treatment plant (51 from raw surface water - SC and 29 recovered after water ozonation - PO) and 32 from the waste water treatment plant (17 from raw - RWW and 15 from treated waste water - TWW). Additionally, nine aeromonads were isolated in the presence of ciprofloxacin (five from RWW and four from TWW). Aeromonads were not detected in ground water samples, neither downstream the chlorination tank of the WTP, including in tap water (Table 2, Fig. 1). Aeromonas spp. identification to the species level (Table 2) was supported by the 16S rRNA gene sequence analysis. Most of the times (114/121), these identifications were the same as those determined based on the analysis of the genes gyrB or cpn60. Only for seven isolates (five from TWW and two from SC) of the species Aeromonas aquariorum, A. punctata and A. veronii, the genes gyrB or cpn60 would lead to a different identification. Among the eleven species, eight were detected in raw surface water (Fig. 1, site A) and only four after the
Fig. 2 e Neighbour-joining dendrogram based on 16S rRNA, cpn60 and gyrB concatenated nucleotide sequences. Bootstrap values (‡50%) generated from 1000 replicates are indicated at branch points. Bold circles indicate branches recovered by the maximum-likelihood method. Sequence types (ST) and16S rRNA GenBank accession numbers are indicated in parenthesis. GenBank accession numbers for cpn60 and gyrB sequences are, respectively, for A. allosaccharophila EU306795 and AY101777, A. aquariorum FJ936120 and EU268444, A. enteropelogenes EU306837 and EF465526, A. eucrenophila EU306803 and AY101776, A. hydrophila subsp. hydrophila EU306804 and AY101778, A. jandaei EU306807 and AY101780, A. media EU306808 and AY101782, A. punctata EU306800 and AY101783, A. sanarellii (JF920655) and FJ807277, A. taiwanensis (JF920656) and FJ807272, A. veronii EU306839 and AY101795.* Strains for which 16S rRNA based identification differed from that given by the genes gyrB and cpn60. Strains designation: Isolates from the drinking water treatment plant were generically designated as SxMn, with S standing for site of isolation (A, raw surface water; C, after ozonation), x for the sampling date (2, second sampling date), M, for the culture medium of isolation (F, mFC agar; T, Tergitol-7 agar; P, Pseudomonas isolation agar; R, R2A agar; E, Bile Esculin agar), and n for the number of the isolate. Isolates from the raw or treated wastewater were generically designated as AxMn or ExMn, respectively (A relative to RWW and E to TWW); x for the sampling date; M for the culture medium of isolation (P, PCA; PC, PCA with ciprofloxacin; EL, mendo-agar-LES; FC, mFC agar with ciprofloxacin); and n for isolate number of the isolate.
5604
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
Table 2 e Diversity and percentage (number of isolates, number of sequence types) of Aeromonas species in the different types of water. Species % (n)
Raw surface water (51)
After Ozonation (29)
Raw Waste water (22)
A. A. A. A. A. A. A. A. A. A. A.
2.0% (1, 1) e 2.0% (1, 1) 2.0% (1, 1) 13.7% (7, 7) 3.9% (2, 2) 19.6% (10, 10) 7.8% (4, 4) e e 49.0% (25, 22)
e 6.9% (2, 1) e e 58.6% (17, 4) 10.3% (3, 1) e e e e 24.1% (7, 3)
9.1% (2, 2) e e e 4.5% (1, 1) e 36.4% (8, 8) 36.4% (8, 6) 4.5% (1, 1) 4.5% (1, 1) 4.5% (1, 1)
allosaccharophila (3) aquariorum (5) enteropelogenes (1) eucrenophila (1) hydrophila subsp. hydrophila (25) jandaei (5) media (25) punctata (18) sanarellii (4) taiwanensis (1) veronii (33)
Treated Waste water (19) e 15.8% e e e e 36.8% 31.6% 15.8% e e
(3, 3)
(7, 7) (6, 6) (3, 2)
No aeromonads were isolated from ground water or from any sampling point after water chlorination, including in 11 household taps.
ozonation process (Fig. 1, site C). Whereas in raw surface water the species A. veronii and A. media predominated, after ozonation, the species A. hydrophila subsp. hydrophila represented more than half of the isolates. The species A. media and A. punctata were not detected after the ozonation process. These same two species, A. media and A. punctata, prevailed in raw and in treated waste water (Table 2). In order to infer about intra-species variability and to track bacteria in the different water samples, bacterial isolates were compared on basis of the nucleotide sequences of the genes 16S rRNA, cpn60 and gyrB. This procedure allowed the identification of 91 sequence types (Table 2). For sake of simplicity, the relationship of the isolates was represented in a dendrogram based on the comparative analyses of the 16S rRNA, cpn60 and gyrB concatenated sequences (Fig. 2). Not surprisingly, the same sequence type was not detected in nondirectly-communicating water compartments, i.e, in surface and waste water. In contrast, the same sequence types were observed occasionally in communicating zones, separated by water treatment, i.e. ozonation or waste water treatment. In the drinking water treatment plant, the same sequence type of A. veronii (V1) was detected in raw surface water and after ozonation. In spite of this, water ozonation seemed to impose a serious bottleneck on strain diversity with the reduction of 48 sequence types in raw surface water to only nine in ozonated water. Moreover, the sequence types of A. hydrophila subsp. hydrophila detected in ozonated water (isolates C of sequence type HH1, Fig. 2) were, most of the times, distinct from those detected in raw surface water, suggesting that minor population representatives may have gained advantage under the oxidative stress imposed by ozone. In the waste water treatment plant, the same sequence types of the species A. media (M1), A. punctata (P2) and A. sanarellii (S1) were detected in raw and in treated waste water (Fig. 2) and P2 and S1 were also present in waste water samples collected in different dates.
3.2.
Antibiotic resistance phenotypes
Amoxicillin resistance was observed in all isolates except in an Aeromonas enteropelogenes strain, suggesting that aeromonads are intrinsically resistant to this beta-lactam. For the
other antibiotics, in general, the highest resistance prevalence values were observed for the penicillin ticarcillin and for the cephalosporin cephalothin, reaching percentages superior to 40%, irrespective of the type of water. Resistance phenotypes to the cephalosporin ceftazidime and the carbapenem meropenem were observed at low rates, exclusively in surface water (Table 3). Ceftazidime resistance was observed in a single isolate of A. media of raw surface water. Meropenem resistance was observed in three isolates of A. veronii, before and after water ozonation. Colistin resistance was another rare phenotype, detected only after the ozonation process in two isolates of Aeromonas jandaei with the same sequence type, presumably representing the same strain. Nalidixic acid resistance was about five times more prevalent among waste water isolates (90.6%) than in surface water (17.6%, p < 0.001) and was not detected after ozonation (Table 3). In waste water, quinolone resistance was mainly related to the species A. media and A. punctata, which predominated in that type of water. Curiously, these two species were not observed after water ozonation, a fact that may explain the apparent efficiency of that disinfection process on quinolone resistance elimination. The potential of A. media as reservoir of quinolone resistance in waste water was confirmed by the fact that nine waste water strains (five from RWW and four from TWW) isolated in the presence of 4 mg L1 of ciprofloxacin were all members of this species. These nine isolates, represented by eight distinct sequence types, were, not surprisingly, resistant to nalidixic acid and had at least an intermediary resistance phenotype to ciprofloxacin (four resistant and five intermediary). Six out of these nine strains were resistant to at least three different classes of antibiotics (R3) (gentamycin, tetracycline and sulfamethoxazole/ trimethoprim), exhibiting resistance phenotypes rare among the aeromonads isolated in the absence of ciprofloxacin. Multi-resistance was also frequent among A. punctata (six out of 18 isolates) (Table 3). In contrast, none of the 33 A. veronii isolates presented resistance to three different classes of antibiotics, probably due to the fact that most of these isolates were recovered from raw surface or ozonated water. Multiresistance (R3) did not differ significantly between raw surface and ozonated water. In contrast, R3 was significantly ( p < 0.05) higher in waste water than in raw surface water.
Table 3 e Antibiotic resistance prevalence (%) in the different sampled sites and Aeromonas species. NA
CIP
TIC
CP
CEF
MER
STR
SXT
TET
GEN
CT
17.6
2.0
68.6
51.0
2.0
3.9
54.9
0
2.0
0
0
5.9
0
0
44.8
44.8
0
3.4
44.8
0
3.4
0
6.9
6.9
RWW (n ¼ 22) - without CIP (n ¼ 17) - with CIP (n ¼ 5)
94.1 100.0
0 40.0
52.9 100.0
70.6 80.0
0 0
0 0
23.5 100.0
5.9 20.0
5.9 20.0
0 40.0
0 0
23.5 100.0
TWW (n ¼ 19) - without CIP (n ¼ 15) - with CIP (n ¼ 4)
86.7 100.0
0 50.0
40.0 100.0
66.7 100.0
0 0
0 0
26.7 25.0
20.0 25.0
13.3 25.0
0 25.0
0 0
20.0 25.0
12.0
0
32.0
52.0
0
0
24.0
0
0
0
0
8.0
B. A. media (n ¼ 25) ewithout CIP (n ¼ 16) ewith CIP (n ¼ 9)
43.8 100.0
6.3 44.4
87.5 100.0
81.3 88.9
6.3 0
0 0
12.5 66.7
6.3 22.2
0 22.2
0 33.3
0 0
6.3 66.7
C. A. punctata (n ¼ 18)
88.9
0
22.2
72.2
0
0
44.4
16.7
11.1
0
0
33.3
3.0
0
81.8
21.2
0
9.1
78.8
0
0
0
0
0
55.0
0
50.0
75.0
0
0
35.0
0
15.0
0
10.0
15.0
Type of water SC (n ¼ 51)
PO (n ¼ 29)
Species A. A. hydrophila subsp. hydrophila (n ¼ 25)
D. A. veronii (n ¼ 33)
E. Other (n ¼ 20)
R3 Species distribution
Water types distribution
NA, nalidixic acid; CIP, ciprofloxacin; TIC, ticarcillin; CP, cephalothin; CEF, ceftazidime; STR, streptomycin; SXT, sulphamethoxazole/trimethoprim; TET, tetracycline; GEN, gentamicin; CT, colistin sulphate; MER, meropenem. R3 represents isolates resistant to 3 or more distinct antibiotic classes (except AML). SC, surface water captation; PO, post ozonation; RWW, raw waste water; TWW, treated waste water.
5606
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
3.3. Genetic determinants of quinolone resistance and cphA gene distribution Given the significantly higher prevalence of quinolone resistance in waste- than in surface water and in the species Aeromonas puntacta and A. media than in the others ( p < 0.001) it was decided to investigate if similar mechanisms of resistance were present in both types of water and in the different species. Irrespective of the type of water or Aeromonas species, nalidixic acid resistance was associated with mutations in the gene gyrA (n ¼ 45) and sometimes also on the gene parC (n ¼ 15 and a silent mutation) (Table 4). The most common gyrA mutations were transversions in the position 83 (AGC / ATC, in 31 isolates or AGT / ATT, in six isolates) corresponding to the substitution of a serine for an isoleucine residue. In two isolates, one of A. allosaccharophila and one of A. jandaei, it was not possible to achieve a successful amplification of the gene gyrA, even using alternative primer sets and protocols (n.d. in Table 4). Among the plasmid-mediated quinolone resistance, only the genes qnrS and aac(6’)-ib-cr were detected. Although being found exclusively in the species A. media (Table 4), these genes were observed in different strains (different sequence types). The gene qnrS was detected in strains of both surface and waste water, isolated either in the presence or in the absence of ciprofloxacin. The qnrS positive strains isolated in the presence of 4 mg L1 ciprofloxacin harboured also the gene aac(6’)-ib-cr which, in contrast to qnrS, was associated with a resistance or intermediary phenotype for ciprofloxacin. The gene aac(6’)-ib-cr was detected exclusively in strains isolated on ciprofloxacin-supplemented medium, suggesting that these strains represent a minor fraction of the bacterial population, which can gain advantage in the presence of selective pressure. One isolate from surface water harboured the gene aac(6’)-ib, but not the cr variant that confers resistance to ciprofloxacin. The cr variant of the gene aac(6’)-ib presented mutations in the position 102, with an arginine residue (AGG or, in one RWW isolate, CGG) instead of tryptophan (TGG), on position 117, with a leucine residue (TTA) instead of serine (TCA) and on position 179, with a tyrosine residue (TAT) instead of an aspartate (GAT). The most common metallo-beta-lactamase expressed by Aeromonas spp. is encoded by the chromosomal gene cphA, reported mainly in the species A. hydrophila, A. veronii and A. jandaei (Janda and Abbott, 2010). The presence and diversity of this gene was screened in an attempt to identify a differential pattern between isolates from the drinking and waste water treatment plants or between different species. The gene cphA was detected in the species Aeromonas allosacharophila (one isolate from SC and one isolate from RWW), A. aquariorum (2 isolates from PO), A. hydrophila subsp. hydrophila (in all except in one isolate from SC and one from PO), A. jandaei (in all isolates) and A. veronii (in all except in one isolate from SC, seven from PO and one from RWW). Unexpectedly, it was also detected in a raw waste water isolate of the species A. media, recovered from ciprofloxacin-supplemented medium. The nucleotide sequences of the gene cphA were different among these isolates. However, those differences corresponded to silent mutations, as the amino acid sequences were identical among the waste- and drinking
water treatment plant isolates (data not shown). Nevertheless, a noticeable contrast was found in terms of prevalence. The gene cphA was significantly ( p 0.001) more prevalent among surface water (65%) than in ozonated water (97%) isolates. It was also significantly ( p < 0.001) less prevalent (18%) in the waste water treatment plant than in the drinking water treatment plant (76%). None of the treated waste water isolates harboured the gene cphA (Fig. 3). The plasmid related beta-lactamase gene blaTEM was detected in a single raw waste water strain of A. media isolated on culture medium supplemented with 4 mg L1 of ciprofloxacin, and which harboured also the gene aac(6’)-ib-cr.
4.
Discussion
This study was based on the hypothesis that Aeromonas spp. can serve as vehicle for antibiotic resistance dissemination within the urban water cycle. The experimental planning comprised the detection, diversity typing and determination of antimicrobial resistance patterns of aeromonads within different parts of the urban water cycle. In respect to detection, Aeromonas spp. were isolated from raw and treated waste water, as well as, from surface water, including after ozone disinfection. In contrast, culturable aeromonads were not detected in locations with pristine or disinfected water, ground and tap water, respectively. Apparently the drinking water treatment, mainly water chlorination, could remove aeromonads to, at least, less than one CFU in 100 mL of water. Although the failure to detect aeromonads in tap water could be attributed to low bacterial densities, in some taps, heterotrophs reached 101e104 CFU mL1 (Fig. 1). The presence of Aeromonas spp. in drinking water is undesirable, as may have implications for user health, mainly via contact transmission (WHO, 2008). Nevertheless, aeromonads have been detected in different types of drinking water, namely tap, mineral bottled and wells (Ku¨hn et al., 1997; Biscardi et al., 2002; Pablos et al., 2009). Some authors referred to the seasonality of aeromonads, which increase may coincide with the raise in the environmental temperature (Janda and Abbott, 2010). In this study, a priori, the failure to detect Aeromonas spp. in ground and tap water cannot be attributed to such seasonality, given the fact that these samples were collected in Summer and Winter (ground water) or in Spring, Summer and Autumn (taps). It is noteworthy that the absence of culturable Aeromonas spp. in tap water contrasts to what was observed in the same samples for other bacterial groups. For example, it was observed that sphingomonads, pseudomonads, and Acinetobacter spp., in spite the sharp decrease of total heterotrophs observed after water chlorination, were present in tap water (Vaz-Moreira et al., 2011; our data unpublished). The fact that the examined taps had a low usage rate (one to four times a month) may be part of the possible explanation for the absence of Aeromonas in tap water, as stagnancy of water in pipes is described as promoting bacterial community rearrangements (Lautenschlager et al. 2010). However, a deeper study would be needed to confirm such hypothesis. Thus, the apparent reduced risk of Aeromonas spp. to contribute for
5607
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
Table 4 e Diversity, antibiotic resistance phenotypes, origin and genetic determinants of quinolone resistance of nalidixic acid resistant isolates.
Species (n) A. allosaccharophila (2) A. aquariorum (3)
A. hydrophila subsp. hydrophila (3) A. jandaei (1)
ST (n) A2 (1) A3 (1) Aq2 (1) Aq3 (1) Aq4 (1) HH7 (1) HH8 (1) HH12 (1) J2 (1) M1 (2)*
Origin (n)
CIP
TIC
Resistance phenotype CP STR Other
RWW (2)
TWW (3) TET SC (2) RWW (1) SC (1) RWW (1)
GEN
TET
TWW (1) M6 (1) M7 (1) M10 (1) M12 (1) M13 (1)* A. media (16)
CAZ
GEN SXT
M16 (1)* M18 (1) M19 (1)* M20 (1)* M21 (1)* M23 (1) M24 (1)
A. taiwanensis (1) A. veronii (1)
ATC (Ill)83
-
83
-
TET
AGA (Arg)83 ATC (Ill)83 ATC (Ill)83 ATC (Ill)83
ATC (Ill)80
ATC (Ill)83
-
GTC (Val)83
-
83
TWW (5)
P1 (3)
RWW (3)
P2 (2)
RWW (1) TWW (1)
P3 (1) P4 (1) P6 (1) P7 (1) P8 (1) P9 (1) P10 (1) P12 (1) P13 (1) P14 (1) P15 (1)
SC (3)
RWW (5)
TWW (4)
RWW (1) A. sanarellii (4)
AGT (Ser)80
RWW (6) M15 (1)*
A. punctata (16)
CGC (Arg)83
ATC (Ill)
SC (3)
M14 (1)*
Mutations gyrA parC GAA (Glu)87 n.d. AGA (Arg)80 ATC (Ill)83 ATC (Ill)83 ATC (Ill)80 ATC (Ill)83 ATT (Ill)83 ATT (Ill)80 ATT (Ill)83 ATC (Ill)83 ATC (Ill)80 n.d. -
S1 (3) S2 (1) T1 (1) V25 (1)
TWW (2) TWW (1) RWW (1) RWW (1)
ATC (Ill) ATC (Ill)83 ATC (Ill)83 ATC (Ill)83 SXT TET GEN ATC (Ill)83 SXT TET ATC (Ill)83 AGA (Arg)83 ATC (Ill)83 TET ATC (Ill)83 SXT TET ATC (Ill)83 ATC (Ill)83 ATC (Ill)83 ATC (Ill)83 AGA (Arg)83 ATC (Ill)83 AGG (Arg)83 SXT ATC (Ill)83 AGA (Arg)83 ATC (Ill)83 ATC (Ill)83 SXT ATC (Ill)83 SXT ATC (Ill)83 ATC (Ill)83 ATT (Ill)83 ATT (Ill)83 ATT (Ill)83 TET ATT (Ill)83 ATC (Ill)83 ATC (Ill)83
n.d., not determined; (shadowing; black, resistant; grey, intermediary; white, susceptible) . * isolated in medium supplemented with 4 mg L1 of ciprofloxacin.
ATC (Ill)80 ATC (Ill)80 CGC (Arg)80 ATC (Ill)80 AAA (Lys)84 AAA (Lys)84 ATC (Ill)80 ATC (Ill)80 AAA (Lys)84 ATC (Ill)80 -
Quinolone resistance genes aac(6’)-ib-cr blaTEM aac(6’)-ib-cr aac(6’)-ib qnrS aac(6’)-ib-cr aac(6’)-ib-cr qnrS aac(6’)-ib-cr qnrS aac(6’)-ib-cr aac(6’)-ib-cr aac(6’)-ib-cr aac(6’)-ib-cr qnrS -
5608
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
Fig. 3 e Percentage of isolates of each type of water that harbored the cphA gene or the gyrA mutation.
antimicrobial resistance dissemination via tap water, observed in this study, must be interpreted with precaution. In respect to diversity, it was observed that the pattern of Aeromonas species was distinct in the different types of water. Although A. media was abundant in both raw surface and waste water, these types of water differed on the predominance of A. veronii and A. punctata (Table 2). Nevertheless, the percentage of distinct sequence types was not significantly different in surface and waste water. Apparently, and in spite the low number of isolates, waste water treatment did not lead to a significant reduction of the number of sequence types. In contrast, water ozonation seemed to impose a bottleneck both in the number of species and of sequence types (Fig. 2). Indeed, the species Aeromonas hydrophyla subsp. hydrophyla became over represented after ozonation with a significant ( p < 0.001) reduction in the percentage of distinct sequence types. Moreover, in general, the sequence types detected after ozonation were different from those found in raw surface water, suggesting some kind of rearrangement in the aeromonads population due to water disinfection (Fig. 2). A. media and A. punctata were the species in which quinolone resistance presented the highest prevalence ( p < 0.001) and the predominance of these species in waste water contributed to explain the elevated rates of nalidixic acid resistance in waste water, significantly ( p < 0.001) higher than in surface water. Similarly, sulfamethoxazole/trimethoprim resistance found exclusively in those two species, was observed only in waste water. In contrast, ceftazidime and meropenem resistance were detected only in surface water, although it is acknowledged that these resistance phenotypes could have been detected also in waste water if a larger number of isolates had been examined. The search for genetic determinants related to quinolone resistance showed that the gyrA mutations were the primary, even not the unique, mechanism (Table 4). The higher prevalence of these mutations in waste water isolates in comparison with the prevalence values observed in surface water (Fig. 3), can be supported by previous studies which demonstrate that quinolone resistance may arise from the contact with mutagenic substances, widely found in the environment (Miyahara et al., 2011). In fact, in waste water the occurrence of such potential mutagens is much more probable than in
uncontaminated waters. Additionally, some effect of selective pressure may take place in waste waters, in which the detection of quinolones is common, with concentrations of ciprofloxacin up to 0.7 mg/L detected in Portuguese municipal waste water treatment plants (Seifrtova´ et al., 2008; our data for the same plant, unpublished). It is not possible to know the relevance of de novo mutation or of selection (vertical transmission) for the observed chromosomal mutations associated with quinolone resistance. But, although probably both forms can contribute for resistance spreading, de novo events may be frequent as gyrA or parC gene mutations were observed in different strains (sequence types). In any case, the species A. punctata and A. media seem to play an important role on this form of dissemination. These results are in agreement with the work of Gon˜i-Urriza et al. (2000), who assessed the impact of an urban effluent on antibiotic resistance of Aeromonas spp. in a riverine area. As in the current study, nalidixic acid resistance was observed in the majority of the aeromonads (72%), most of them of the species A. punctata (A. caviae), and was exclusively chromosomally encoded. The conclusion reached by Gon˜i-Urriza et al. (2000), applies also to the present study e urban effluents are responsible for the increase of quinolone resistance in the receptor water courses. Among the quinolone resistance determinants associated with mobile genetic elements, only the genes qnrS and aac(6’)ib-cr were detected and only in the species A. media. In both cases, these genetic determinants were found in different strains (sequence types), as expected if horizontal gene transfer is equated. Both determinants, and mainly aac(6’)-ib-cr which was detected only in isolates recovered in the presence of ciprofloxacin, were rare in the analysed samples. Nevertheless, the gene qnrS is apparently widespread in waters (waste water, rivers, aquaculture), detected not only from total DNA and but also from cultures of aeromonads and Enterobacteriaceae (Cattoir et al., 2008; Pica˜o et al., 2008; Szczepanowski et al., 2009; Ishida et al., 2010; Cummings et al., 2011). Similarly, the gene aac(6’)-ib-cr is found in different types of water (lake water, river sediments, aquaculture), either in total DNA or in bacterial isolates (aeromonads and Enterobacteriaceae) (Pica˜o et al. 2008; Ishida et al., 2010; Cummings et al., 2011). The presence of these genes, although conferring low-levels of resistance, can favour and complement the selection of other resistance mechanisms (Rodrı´guez-Martı´nez et al., 2010). The fact that the determinants qnrS and aac(6’)-ib-cr were detected only in A. media suggests that this species may represent an important vector of quinolone resistance. Ceftazidime resistance was also detected only in this same species in surface water. Recent evidences that water A. media can colonize humans (Khajanchi et al., 2010) may give additional relevance to this species on the dispersal of resistance. Although the number of isolates examined was too low to strongly support this conclusion, the data suggested that water ozonation may promote the reduction of A. media. For instance, no quinolone or ceftazidime resistance were observed downstream of this point. In contrast, meropenem resistance, in this study associated to the species A. veronii, was observed also in ozonated water. Nevertheless, the data gathered in this study suggests that water chlorination may contribute to control resistance propagation by aeromonads via drinking water.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
In general, the results obtained suggest that different aeromonads populations and antibiotic resistance determinants prevail in different parts of the urban water cycle. The clearest example of this was the distribution of the gene cphA and of the gyrA mutation, observed in the majority of the surface and waste water isolates, respectively (Fig. 3). As expected, the gene cphA was predominant among the species A. hydrophila subsp. hydrophila, A. veronii and A. jandaei, also the most prevalent in the drinking water treatment plant. In the same way, the gyrA mutations prevailed in A. media and A. punctata, the predominant species in waste water. The contrast observed in the distribution of both genetic determinants is mainly due to the patterns of species occurring in both types of water and which, probably, are due to the environmental conditions and selective pressures imposed in both types of habitat. This also demonstrates that, in each type of water, aeromonads may represent a source of distinct types of antibiotic resistance. The importance of a given Aeromonas species for the antimicrobial resistance patterns in each type of water, observed in the current work, confirm previous studies conducted with other bacterial groups. Figueira et al. (2011) studied different populations of waste water E. coli and concluded that variations on the prevalence of quinolone resistance were correlated with the dynamics of some population sub-sets. VazMoreira et al. (2011) characterizing the patterns of antimicrobial resistance in sphingomonads from tap water and cup fillers of dental chairs also concluded that antibiotic resistance patterns were often species- rather than site-related. Nevertheless, in the current work, and in contrast to what was suggested by other authors studying A. salmonicida from fish farms and environmental samples (Giraud et al., 2004; Kim et al., 2011), no clonal spreading of antibiotic resistance was observed. In contrast, rarely were observed the same sequence types in different water samples. This suggests that the acquisition of a specific resistance type, either by horizontal gene transfer or by adaptive mutation, may take place preferentially in a given habitat, in which a species is prevalent or has a higher fitness than the others. In other words, the success of resistance acquisition may depend on the fitness of the target bacterium (receptor of horizontal gene transfer or mutant) in a specific environment.
5.
Conclusions
The patterns of Aeromonas species and antimicrobial resistance varied over different parts of the urban water cycle; In each type of water, the antimicrobial resistance patterns were primarily function of the prevailing species; In raw surface and waste water no strong evidences for clonal dissemination of antimicrobial resistance were detected; Water ozonation imposed a bottleneck on species diversity, with evidences of clonal selection, and promoted a significant reduction of quinolone resistance and the increase of cphA metallo-beta-lactamase; Waste water aeromonads, particularly A. media and A. punctata, were confirmed as relevant environmental harbours
5609
of quinolone resistance, either chromosomally ( gyrA mutation) or plasmid encoded (qnrS and aac(60 )-Ib-cr). Water aeromonads were confirmed as relevant agents for antimicrobial resistance spreading in the environment, which presence in tap water could be significantly reduced by water chlorination.
Acknowledgements Authors gratefully acknowledge Prof. P. Nordmann, Dr. L.M. Cavaco, and Prof. M. Wang for the positive controls for the detection of the genes qnrA, qnrB and qnrS; qnrD and aac(6’)Ibcr; and qnrC, respectively. This study was financed by Fundac¸a˜o para a Cieˆncia e a Tecnologia (projects PTDC/AMB/70825/ 2006; PTDC/AMB/71236/2006, IVM grant SFRH/BD/27978/2006; MS grant Integration into Research).
references
Abulhamd, A.T., 2009. Characterization of Aeromonas hydrophila isolated from aquatic environments using phenotypic and genotyping methods. Research Journal of Agriculture and Biological Sciences 5, 923e931. Aminov, R.I., 2009. The role of antibiotics and antibiotic resistance in nature. Environmental Microbiology 11 (12), 2970e2988. Arias, A., Seral, C., Navarro, F., Miro, E., Coll, P., Castillo, F.J., 2010a. Plasmid-mediated QnrS2 determinant in an Aeromonas caviae isolate recovered from a patient with diarrhoea. Clinical Microbiology and Infectious Diseases 16, 1005e1007. Arias, A., Seral, C., Gude, M.J., Castillo, F.J., 2010b. Molecular mechanisms of quinolone resistance in clinical isolates of Aeromonas caviae and Aeromonas veronii bv. sobria. International Microbiology 13, 135e141. Balsalobre, L.C., Dropa, M., Lincopan, N., Mamizuka, E.M., Matte´, G.R., Matte´, M.H., 2009. Detection of metallo-blactamases-encoding genes in environmental isolates of Aeromonas hydrophila and Aeromonas jandaei. Letters in Applied Microbiology 49, 142e145. Baquero, F., Martı´nez, J.-L., Canto´n, R., 2008. Antibiotics and antibiotic resistance in water environments. Current Opinion in Biotechnology 19, 260e265. Biscardi, D., Castaldo, A., Gualillo, O., Fuscos, R., 2002. The occurrence of cytotoxic Aeromonas hydrophila strains in Italian mineral and thermal waters. Science of the Total Environment 292, 255e263. Blasco, M.D., Esteve, C., Alcaide, E., 2008. Multiresistant waterborne pathogens isolated from water reservoirs and cooling systems. Journal of Applied Microbiology 105 (2), 469e475. Cattoir, V., Poirel, L., Rotimi, V., Soussy, C.J., Nordmann, P., 2007. Multiplex PCR for detection of plasmid-mediated quinolone resistance qnr genes in ESBL-producing enterobacterial isolates. Journal of Antimicrobial Chemotherapy 60, 394e397. Cattoir, V., Poirel, L., Aubert, C., Soussy, C.J., Nordmann, P., 2008. Unexpected occurrence of plasmid-mediated quinolone resistance determinants in environmental Aeromonas spp. Emerging Infectious Diseases 14, 231e237. Cavaco, L.M., Hasman, H., Xia, S., Aarestrup, F.M., 2009. qnrD, a novel gene conferring transferable quinolone resistance in Salmonella enterica serovar Kentucky and Bovismorbificans
5610
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
strains of human origin. Antimicrobial Agents and Chemotherapy 53, 603e608. Clinical and Laboratory Standards Institute, 2007. M100eS17. Performance Standards for Antimicrobial Susceptibility Testing, vol. 27, Wayne, PA. Corry, J.E.L., Curtis, G.D.W., Baird, R.M., 2003. Handbook of Culture Media for Food Microbiology, vol. 34. Elsevier Science BV, Amsterdam, The Netherlands, pp. 331. Cummings, D., Archer, K.F., Arriola, D.J., Pieter, A.B., Faucett, K.G., Laroya, J.B., Pfeil, K.L., Ryan, C.R., Ryan, K.R.U., Zuill, D.E., 2011. Broad dissemination of plasmid-mediated quinolone resistance genes in sediments of two urban coastal wetlands. Environmental Science & Technology 45, 447e454. DiPersio, J.R., Deshpande, L.M., Biedenbach, D.J., Toleman, M.A., Walsh, T.R., Jones, R.N., 2005. Evolution and dissemination of extended-spectrum aˆ-lactamase-producing Klebsiella pneumoniae: epidemiology and molecular report from the SENTRY Antimicrobial Surveillance Program (1997e2003). Diagnostic Microbiology & Infectious Disease 51, 1e7. Dixon, B., 2008. Natural disaster microbiology. Microbe 3, 312e313. Faria, C., Vaz-Moreira, I., Serapicos, E., Nunes, O.C., Manaia, C.M., 2009. Antibiotic resistance in coagulase negative staphylococci isolated from wastewater and drinking water. Science of the Total Environment 407, 3876e3882. Ferreira da Silva, M., Tiago, I., Verı´ssimo, A., Boaventura, A.R., Nunes, O.C., Manaia, C.M., 2006. Antibiotic resistance of enterococci and related bacteria in an urban waste water treatment plant. FEMS Microbiololy Ecology 55, 322e329. Ferreira da Silva, M., Vaz-Moreira, I., Gonzalez-Pajuelo, M., Nunes, O.C., Manaia, C.M., 2007. Antimicrobial resistance patterns in Enterobacteriaceae isolated from an urban waste water treatment plant. FEMS Microbiology Ecology 60, 166e176. Figueira, V., Serra, E., Manaia, C.M., 2011. Differential patterns of antimicrobial resistance in population subsets of Escherichia coli isolated from waste- and surface waters. Science of the Total Environment 409, 1017e1023. Giraud, E., Blanc, G., Bouju-Albert, A., Weill, F.-X., DonnayMoreno, C., 2004. Mechanisms of quinolone resistance and clonal relationship among Aeromonas salmonicida strains isolated from reared fish with furunculosis. Journal of Medical Microbiology 53, 895e901. Gordon, L., Cloeckaert, A., Doublet, B., Schwarz, S., BoujuAlbert, A., Ganie`re, J.-P., Le Bris, H., Le Fle`che-Mate´os, A., Giraud, E., 2008. Complete sequence of the floR-carrying multiresistance plasmid pAB5S9 from freshwater Aeromonas bestiarum. Journal of Antimicrobial Chemotherapy 62 (1), 65e71. Gon˜i-Urriza, M., Capdepuy, M., Arpin, C., Raymond, N., Caumette, P., Quentin, C., 2000. Impact of an urban effluent on antibiotic resistance of riverine Enterobacteriaceae and Aeromonas spp. Applied and Environmental Microbiology 66 (1), 125e132. Gon˜i-Urriza, M., Arpin, C., Capdepuy, M., Dubois, V., Caumette, P., Quentin, C., 2002. Type II topoisomerase quinolone resistancedetermining regions and mutations asociated with quinolone resistance. Antimicrobial Agents and Chemotherapy 46, 350e359. Huddleston, J.R., Zak, J.C., Jeter, R.M., 2006. Antimicrobial susceptibilities of Aeromonas spp. isolated from environmental sources. Applied and Environmental Microbiology 72 (11), 7036e7042. Ishida, Y., Ahmed, A.M., Mahfouz, N.B., Kimura, T., El-Khodery, S. A., Moawad, A.A., Tadashi Shimamoto, T., 2010. Molecular analysis of antimicrobial resistance in Gram-negative bacteria isolated from fish farms in Egypt. Journal of Veterinary Medical Science 72 (6), 727e734.
Janda, J.M., Abbott, S.L., 2010. The genus Aeromonas: taxonomy, pathogenicity, and infection. Clinical Microbiology Reviews 23, 35e73. Jukes, T.H., Cantor, C.R., 1969. Evolution of protein molecules. In: Mammalian Protein Metabolism, vol. 3. Academic Press, Munro, pp. 21e132. Academic Press, NY. Khajanchi, B.K., Fadl, A.A., Borchardt, M.A., Berg, R.L., Horneman, A.J., Stemper, M.E., Joseph, S.W., Moyer, N.P., Sha, J., Chopra, A.K., 2010. Distribution of virulence factors and molecular fingerprinting of Aeromonas species isolates from water and clinical samples: suggestive evidence of water-to-human transmission. Applied and Environmental Microbiology 76, 2313e2325. Kim, J.H., Hwang, S.Y., Son, J.S., Han, J.E., Jun, J.W., Shin, S.P., Choresca Jr., C., Choi, Y.J., Park, T.H., Park, S.C., 2011. Molecular characterization of tetracycline- and quinoloneresistant Aeromonas salmonicida isolated in Korea. Journal of Veterinary Science 12 (1), 41e48. Kivanc, M., Yilmaz, M., Demir, F., 2011. The occurrence of Aeromonas in drinking water, tap water and the porsuk river. Brazilian Journal of Microbiology 42, 126e131. Ku¨hn, I., Huys, G., Coopman, R., Kersters, K., Janssen, P., 1997. A 4-year study of the diversity and persistence of coliforms and Aeromonas in the water of a Swedish drinking water well. Canadian Journal of Microbiology 43, 9e16. Ku¨mmerer, K., 2009. Antibiotics in the aquatic environment e a review e part II. Chemosphere 75, 435e441. Lamy, B., Kodjo, A., Laurent, F.colBVH Study Group, 2009. Prospective nationwide study of Aeromonas infections in France. Journal of Clinical Microbiology 47, 1234e1237. Lane, D.J., 1991. 16S/23S rRNA sequencing. In: Stackebrandt, E., Goodfellow, M. (Eds.), Nucleic Acid Techniques in Bacterial Systematics. Wiley, Chichester, pp. 115e175. Lautenschlager, K., Boon, N., Wang, Y., Egli, T., Hammes, F., 2010. Overnight stagnation of drinking water in household taps induces microbial growth and changes in community composition. Water Research 44 (17), 4868e4877. Maalej, S., Mahjoubi, A., Elazri, C., Dukan, S., 2003. Simultaneous effects of environmental factors on motile Aeromonas dynamics in an urban effluent and in the natural seawater. Water Research 37, 2865e2867. Min˜ana-Galbis, D., Urbizu-Serrano, A., Farfa´n, M., Fuste´, M.C., Lore´n, J.G., 2009. Phylogenetic analysis and identification of Aeromonas species based on sequencing of the cpn60 universal target. International Journal of Systematic and Evolutionary Microbiology 59, 1976e1983. Miyahara, E., Nishie, M., Takumi, S., Miyanohara, H., Nishi, J., Yoshiie, K., Oda, H., Takeuchi, M., Komatsu, M., Aoyama, K., Horiuchi, M., Takeuchi, T., 2011. Environmental mutagens may be implicated in the emergence of drug-resistant microorganisms. FEMS Microbiology Letters 317, 109e116. Novo, A., Manaia, C.M., 2010. Factors influencing antibiotic resistance burden in municipal waste water treatment plants. Applied Microbiology and Biotechnology 87, 1157e1166. Pablos, M., Rodrı´guez-Calleja, J.M., Santos, J.A., Otero, A., Garcı´aLo´pez, M.L., 2009. Occurrence of motile Aeromonas in municipal drinking water and distribution of genes encoding virulence factors. International Journal of Food Microbiology 135, 158e164. Park, C.H., Robicsek, A., Jacoby, G.A., Sahm, D., Hooper, D.C., 2006. Prevalence in the United States of aac(6’)-Ib-cr encoding a ciprofloxacin-modifying enzyme. Antimicrobial Agents and Chemotherapy 50, 3953e3955. Pe´richon, B., Courvalin, P., Galimand, M., 2007. Transferable resistance to aminoglycosides by methylation of G1405 in 16S rRNA and to hydrophilic fluoroquinolones by QepA-mediated efflux in Escherichia coli. Antimicrobial Agents and Chemotherapy 51, 2464e2469.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 5 9 9 e5 6 1 1
Pica˜o, R.C., Poirel, L., Demarta, A., Silva, C.S., Corvaglia, A.R., Petrini, O., Nordmann, P., 2008. Plasmid-mediated quinolone resistance in Aeromonas allosaccharophila recovered from a Swiss lake. Journal of Antimicrobial Chemotherapy 62, 948e950. Rodrı´guez-Martı´nez, J.-M., Nordmann, P., Ronco, E., Poirel, L., 2010. Extended-spectrum cephalosporinase in Acinetobacter baumannii. Antimicrobial Agents and Chemotherapy 54 (8), 3484e3488. 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 Microbiology Ecology 43, 325e335. Seifrtova´, M., Pena, A., Lino, C.M., Solich, P., 2008. Determination of fluoroquinolone antibiotics in hospital and municipal wastewaters in Coimbra by liquid chromatography with a monolithic column and fluorescence detection. Analytical Bioanalytical Chemistry 391, 799e805. Servais, P., Passerat, J., 2009. Antimicrobial resistance of fecal bacteria in waters of the Seine river watershed (France). Science of the Total Environment 408, 365e372. Szczepanowski, R., Linke, B., Krahn, I., Gartemann, K.-H., Gu¨tzkow, T., Eichler, W., Pu¨hler, A., Schlu¨ter, A., 2009. Detection of 140 clinically relevant antibiotic resistance genes in the plasmid metagenome of wastewater treatment plant bacteria showing reduced susceptibility to selected antibiotics. Microbiology 155, 2306e2319. Tamura, K., Dudley, J., Nei, M., Kumar, S., 2007. MEGA4: molecular 572 Evolutionary genetics analysis (MEGA) software version 4.0. Molecular Biology and Evolution 24, 1596e1599. Taylor, N.G.H., Verner-Jeffreys, D.W., Baker-Austin, C., 2011. Aquatic systems: maintaining, mixing and mobilising antimicrobial resistance? Trends in Ecology and Evolution 26 (6), 278e284.
5611
Vaz-Moreira, I., Nunes, O.C., Manaia, C.M., 2011. Diversity and antibiotic resistance patterns of Sphingomonadaceae isolates from drinking water. Applied and Environmental Microbiology 77 (16), 5697e5706. Walsh, T.R., Stunt, R.A., Nabi, J.A., MacGowan, A.P., Bennett, P.M., 1997. Distribution and expression of beta-lactamase genes among Aeromonas spp. Journal of Antimicrobial Chemotherapy 40, 171e178. Wang, M., Guo, Q., Xu, X., Wang, X., Ye, X., Wu, S., Hooper, D.C., Wang, M., 2009. New plasmid-mediated quinolone resistance gene, qnrC, found in a clinical isolate of Proteus mirabilis. Antimicrobial Agents and Chemotherapy 53, 1892e1897. World Health Organization, 2008. Guidelines for drinking water, third ed. WHO, Geneva. Xi, C., Zhang, Y., Marrs, C.F., Ye, W., Simon, C., Foxman, B., Nriagu, J., 2009. Prevalence of antibiotic resistance in drinking water treatment and distribution systems. Applied and Environmental Microbiology 75 (17), 5714e5718. Xia, S., Hendriksen, R.S., Xie, Z., Huang, L., Zhang, J., Guo, W., Xu, B., Ran, L., Aarestrup, F.M., 2009. Molecular characterization and antimicrobial susceptibility of Salmonella isolates from infections in humans in Henan Province, China. Journal of Clinical Microbiology 47, 401e409. Xia, R., Guo, X., Zhang, Y., Xu, H., 2010. qnrVC-like gene located in a novel complex class 1 integron harboring the ISCR1 element in an Aeromonas punctata strain from an aquatic environment in Shandong Province, China. Antimicrobial Agents and Chemotherapy 54, 3471e3474. Ya´n˜ez, M.A., Catala´n, V., Apra´iz, D., Figueras, M.J., Martı´nezMurcia, A.J., 2003. Phylogenetic analysis of members of the genus Aeromonas based on gyrB gene sequences. International Journal of Systematic and Evolutionary Microbiology 53, 875e883.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Long-term effect of ZnO nanoparticles on waste activated sludge anaerobic digestion Hui Mu, Yinguang Chen* State Key Laboratory of Pollution Control and Resources Reuse, School of Environmental Science and Engineering, Tongji University, 1239 Siping Road, Shanghai 200092, China
article info
abstract
Article history:
The increasing use of zinc oxide nanoparticles (ZnO NPs) raises concerns about their
Received 9 May 2011
environmental impacts, but the potential effect of ZnO NPs on sludge anaerobic digestion
Received in revised form
remains unknown. In this paper, long-term exposure experiments were carried out to
10 August 2011
investigate the influence of ZnO NPs on methane production during waste activated sludge
Accepted 14 August 2011
(WAS) anaerobic digestion. The presence of 1 mg/g-TSS of ZnO NPs did not affect methane
Available online 23 August 2011
production, but 30 and 150 mg/g-TSS of ZnO NPs induced 18.3% and 75.1% of inhibition respectively, which showed that the impact of ZnO NPs on methane production was
Keywords:
dosage dependant. Then, the mechanisms of ZnO NPs affecting sludge anaerobic digestion
Zinc oxide nanoparticles
were investigated. It was found that the toxic effect of ZnO NPs on methane production
Waste activated sludge
was mainly due to the release of Zn2þ from ZnO NPs, which may cause the inhibitory
Anaerobic digestion
effects on the hydrolysis and methanation steps of sludge anaerobic digestion. Further
Mechanisms
investigations with enzyme and fluorescence in situ hybridization (FISH) assays indicated that higher concentration of ZnO NPs decreased the activities of protease and coenzyme F420, and the abundance of methanogenesis Archaea. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
With the rapid development of nanotechnology, nanoparticles (NPs) are now widely used in some industrial products, such as antibactericide coatings, catalysts, biomedicine, skin creams and toothpastes because of their unique physicochemical properties of enhanced magnetic, electrical, optical, and etc (Maynard et al., 2006; Roco, 2005). It is inevitable for the release of NPs from discover source to environment receptor, and some NPs have been found in wastewater treatment plants (WWTPs) and waste sludge (Brar et al., 2010). It is therefore necessary to evaluate their impacts on the environment. Zinc oxide (ZnO) NPs, one of metal oxide NPs, have received increasing interest due to their widespread industrial, medical and military applications (Ellsworth et al., 2000; Miziolek,
2002; Serda et al., 2009). There are some publications discussing the toxicity of ZnO NPs on microbes. For example, Adams et al. (2006) reported that 500 mg/L of ZnO NPs significantly inhibited the growth of Bacillus subtilis up to 90%, but only induced 38% of the growth inhibition of Escherichia coli, meaning the different toxicity of ZnO NPs on different species of bacteria. Previous study also showed that ZnO NPs reduced the microbial biomass, and altered the diversity and composition of soil bacterial community (Ge et al., 2011). The release of ZnO NPs to WWTPs, which are usually operated with an activated sludge process, has been reported recently (Gottschalk et al., 2009). The released ZnO NPs were observed to be removed by activated sludge via adsorption, aggregation and settling in WWTPs (Kiser et al., 2010, 2009; Kiser et al., 2010; Limbach et al., 2008). Large amounts of WAS
* Corresponding author. Tel.: þ86 21 65981263; fax: þ86 21 65986313. E-mail address: [email protected] (Y. Chen). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.022
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
are produced in municipal WWTPs, which need to be treated before being discharged to the environment. It can be anticipated that most of ZnO NPs will enter into sludge treatment system. Among several sludge treatment methods anaerobic digestion for methane is a preferred one because of WAS being reused and the energy being recovered. Nevertheless, the toxicity of ZnO NPs to sludge anaerobic digestion has seldom been investigated. Some studies addressed that the toxicity of ZnO NPs came from the released zinc ions (Zn2þ) (Franklin et al., 2007; Wong et al., 2010; Xia et al., 2008), but others found that the toxicity of ZnO NPs to some microorganisms (such as E. coli and Pseudomonas fluorescens) was not caused by the released Zn2þ but ZnO NPs themselves (Jiang et al., 2009). Thus, the role of released Zn2þ from ZnO NPs on sludge anaerobic digestion should be taken into account. Moreover, oxidative stress induced by ZnO NPs was reported to cause the loss of cell viability, and the increase of intracellular reactive oxygen species (ROS) was found to be toxic to cytoplasmic lipids, proteins and other intermediates in cells (Sharma et al., 2009; Xia et al., 2008). This study was to evaluate the impact of ZnO NPs on methane production during sludge anaerobic digestion and to explore the mechanisms. Furthermore, fluorescence in situ hybridization (FISH) technique with 16S rRNAtargeted oligonucleotide probes was employed to monitor the quantity change of bacteria and Archaea community after WAS anaerobic digestion system long-term exposed to ZnO NPs.
2.
Materials and methods
2.1.
Nanoparticles and waste activated sludge
ZnO NPs were purchased from Sigma Aldrich (St. Louis, MO). The X-ray diffraction (XRD) pattern of ZnO NPs was measured using a Rigaku D/Max-RB (Rigaku, Japan) diffractometer equipped with a rotating anode and a Cu Ka radiation source and shown in Fig. S1 (Supplementary Information). In this study, stock dispersion of ZnO NPs was produced by adding 2 g ZnO NPs to 1.0 L distilled water (pH 7.0) containing 0.1 mM sodium dodecylbenzene sulfonate (SDBS) (Sigma Aldrich, St. Louis, MO) to enhance the stability of nano-suspension because the particles almost immediately aggregated in surrounding medium (Adams et al., 2006; Franklin et al., 2007; Ganesh et al., 2010; Keller et al., 2010; Simon-Deckers et al., 2009; Xia et al., 2008). The stock dispersion was sonicated (25 C, 250 W, 40 kHz) for 1 h to break aggregates before being diluted to the exposure concentrations. Analysis of the suspension by dynamic light scattering (DLS) (Franklin et al., 2007; Simon-Deckers et al., 2009) using a Malvern Autosizer 4700 (Malvern Instruments, UK) indicated that the average particle size of ZnO NPs was approximately 140 20 nm on the basis of the number distribution with more than five separate measurements per sample. The WAS used in this study was withdrawn from the secondary sedimentation tank of a municipal WWTP in Shanghai, China. The sludge was concentrated by settling at 4 C for 24 h, and its main characteristics (average data plus standard deviations of triplicate tests) are as follows: pH 6.7 0.2, total
5613
suspended solids (TSS) 10070 780 mg/L, volatile suspended solids (VSS) 7690 452 mg/L, soluble chemical oxygen demand (SCOD) 90 5 mg/L, total chemical oxygen demand (TCOD) 10710 220 mg/L, total carbohydrate 899 530 mg-COD/L, total protein 5685 149 mg-COD/L, and total zinc 0.8 0.2 mg/g-TSS.
2.2.
Determination of ZnO nanoparticles dissolution
In order to measure the concentration of released Zn2þ from ZnO NPs, three concentrations of ZnO NPs in 0.1 mM SDBS solutions were prepared with the stock dispersion, and the mixtures were maintained in an air-batch shaker (150 rpm) at 35 1 C for 48 h. At different time, the samples were withdrawn and centrifuged at 12000 rpm for 30 min, and the supernatant was collected, and filtered through 0.22 mm mixed cellulose ester membrane (Jiang et al., 2009; Li et al., 2011). The released zinc was determined by inductively coupled plasma optical emission spectrometry (ICP-OES, PerkinElmer Optima 2100 DV, USA) after acidified with 4% ultrahigh purity HNO3.
2.3. Experiments of effects of ZnO nanoparticles and their released Zn2þ on WAS anaerobic digestion for methane production Three dosages (1, 30 and 150 mg/g-TSS) of ZnO NPs were used to investigate the impact of ZnO NPs on WAS digestion in this paper. The dosage of 1 mg/g-TSS of ZnO NPs was chosen to be the environmentally relevant concentration according to the literature (EPA, 2009; Gottschalk et al., 2009). Also some scientists suggested that although lower nanomaterial content (50 mg C60/g-TSS in their study) showed almost no influence on anaerobic community, a much higher nanomaterial dosage should be investigated before the final conclusion regarding the toxicity of nanomaterial was reached (Nyberg et al., 2008). Moreover, since the environmental release of NPs might be increased due to their large-scale production, the potential effects of higher concentrations (30 and 150 mg/g-TSS) of ZnO NPs were also investigated in this study according to the reference (Adams et al., 2006). The influence of ZnO NPs long-term exposure on methane production was conducted in series of serum bottles (500 mL each), with a sludge volume of 300 mL each. As SDBS was used as the dispersing reagent in this study, two controls, one with only sludge, and another one with sludge plus 4 mg/g-TSS of SDBS, were conducted to investigate whether methane production was affected by SDBS addition. After flushed with nitrogen gas for 5 min to remove oxygen, all bottles were capped with rubber stoppers, sealed and placed in an air-bath shaker (150 rpm) at 35 1 C. Every day, 15 mL fermentation mixture was manually withdrawn from each serum bottle and the same amounts of raw sludge, SDBS and ZnO NPs were supplemented, which resulted in a hydrolytic retention time (HRT) or sludge retention time (SRT) of 20 d. The sampling was operated in a glove box. After the reactors were operated for 105 d, the daily methane production did not change significantly with time, and then the analyses of methane production, enzyme activity, biomass viability and microbial community were conducted. The total gas volume was measured by releasing the pressure in the bottles using a glass syringe (100 mL) to equilibrate with the room pressure according to our previous publication (Zhao et al., 2010).
5614
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
Stock solution (50 mg/L) of ZnCl2 (Sigma-Aldrich) was prepared in 0.1 mM SDBS solution (pH 7.0). The long-term experiments of released Zn2þ from ZnO NPs affecting WAS digestion were conducted with the same method described above except that the ZnCl2 was used to replace ZnO NPs, and the total amount of Zn2þ added to the serum bottles was 1.2, 11.6, and 17.6 mg/L, respectively.
2.4. Effects of ZnO nanoparticles on each step involved in methane production It is well known that sludge anaerobic digestion usually undergoes solubilization of sludge particulate organic-carbon, hydrolysis, acidification and methanation. The experiments of long-term effects of ZnO NPs on these four stages were conducted with the inoculum seeds from four long-term operated reactors with ZnO NPs dosage of 0, 1, 30 and 150 mg/g-TSS, respectively. The experiments of long-term effects of ZnO NPs on sludge particulate organic maters solubilization were conducted as follows. WAS of 300 mL and 30 mL inocula were added to each serum bottle. After flushed with nitrogen gas for 5 min to remove oxygen, all bottles were capped with rubber stoppers, sealed and placed in an air-bath shaker (150 rpm) at 35 1 C. The concentrations of soluble protein and carbohydrate were measured after fermentation for 2 d. As soluble protein and polysaccharide were the main sludge solubilized products, in order to investigate the longterm effects of ZnO NPs on the hydrolysis of sludge solubilized products, the following batch tests with synthetic wastewater containing bovine serum albumin (BSA, average molecular weight Mw 67000, model protein compound used in this study) and dextran (Mww23800, model polysaccharide compound) were conducted. The synthetic wastewater consisted of (mg/L of distilled water) 1000 KH2PO4, 400 CaCl2, 600 MgCl2$6H2O, 100 FeCl3, 0.5 ZnSO4$7H2O, 0.5 CuSO4$5H2O, 0.5 CoCl2$6H2O, 0.5 MnCl2$4H2O, 1 NiCl2$6H2O and 34.8 SDBS. After 4.8 g BSA and 1.2 g dextran (the mass ratio of protein to carbohydrate was almost the same as that in WAS) were dissolved in 1200 mL synthetic wastewater, the mixture liquid was divided equally into 4 bottles, and then 30 mL inocula, which was heat-pretreated at 102 C for 30 min to kill methanogens (Oh et al., 2003), was added before the pH in each bottle was adjusted to 7.0 by adding 4 M NaOH or 4 M HCl. After flushed with nitrogen gas to remove oxygen, all bottles were capped with rubber stoppers, sealed and placed in an air-bath shaker (150 rpm) at 35 1 C. By analyzing the degradation efficiencies of protein and dextran, the longterm effects of ZnO NPs on sludge hydrolysis were obtained. The same operations were conducted when the long-term effects of ZnO NPs on the acidification of hydrolyzed products were investigated except that the synthetic wastewater containing 4.8 g L-glutamate (model amino acid compound) and 1.2 g glucose (model monosaccharide compound). As acetic acid was the main short-chain fatty acid (SCFA) of sludge acidification product (Yuan et al., 2006) and the preferred substrate for methane production (Fig. 1), the longterm effect of ZnO NPs on methanation of the acidification product was conducted with 300 mL synthetic wastewater (see the above description) containing 0.72 g sodium acetate (model
SCFA compound) in each serum bottle. All other operations were the same as described above. By the analysis of methane production, the long-term effect of ZnO NPs on methanation was obtained.
2.5.
Analytical methods
Gas component was measured via a gas chromatograph (Agilent 6890N, USA) equipped with a thermal conductivity detector using nitrogen as the carrier gas. The zinc concentration was analyzed by ICP-OES (PerkinElmer Optima 2100 DV, USA). To measure the zinc content in sludge, the sample was digested according to EPA Method 200.2 prior to ICP analyses. The pH value was measured by a pH meter. The determinations of SCFA, protein, carbohydrate, TSS and VSS were the same as those described in the previous publication (Yuan, et al., 2006). The total SCFA was calculated as the sum of measured acetic, propionic, n-butyric, iso-butyric, n-valeric and iso-valeric acids. The COD (chemical oxygen demand) conversion factors of protein, carbohydrate and SCFA were performed according to Grady et al. (1999). The detailed analytical procedures of scanning electron microscopy (SEM), intracellular ROS, Cell counting kit-8 (CCK-8), FISH, protease, acetate kinase (AK) and coenzyme F420 activities are presented in Supplementary Information.
2.6.
Statistical analysis
All assays were conducted in triplicate and the results were expressed as mean standard deviation. An analysis of variance (ANOVA) was used to test the significance of results and p < 0.05 was considered to be statistically significant.
Particulate organic matters of waste active sludge
Solubilization
Solubilization
Soluble polysaccharide
Soluble protein protease
Hydrolysis
Amino acids
Monosaccharide
Hydrolysis
Pyruvic acid Acidification
Acetyl-CoA AK
Propionic acid
Acetic acid
Methanation F420
Methane
Butyric acid Carbon dioxide
Hydrogen
F420
Fig. 1 e Proposed metabolic pathway for methane production from WAS anaerobic digestion. Only the key enzymes assayed in this study are labeled.
Relativ e m eth an e p ro d uction ( % o f co ntro l)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
*
100
* 80
60
* 40
* 20
0 1
30
150
1.2
ZnO NPs (mg/g-TSS)
11.6
17.6
Zn2+ (mg/L)
Fig. 2 e Effects of different dosages of ZnO NPs (1, 30 and 150 mg/g-TSS) and the corresponding released Zn2D (1.2, 11.6 and 17.6 mg/L) on methane production during WAS digestion. Asterisks indicate statistical differences ( p < 0.05) from the control. Error bars represent standard deviations of triplicate tests.
3.
Results and discussions
3.1.
Effect of ZnO nanoparticles on methane production
In this study, the addition of dispersing reagent (SDBS) at a dosage of 4 mg/g-TSS in sludge digestion experiments or 0.1 mM in synthetic wastewater tests was not observed to affect the methane production. This observation is consistent with Garcia et al. (2006). In the coming text, the control represents the reactor without ZnO NPs addition but with an SDBS dosage of 4 mg/g-TSS in sludge digestion experiments or 0.1 mM in synthetic wastewater tests. As shown in Fig. 2, when ZnO NPs were added to sludge fermentation system, their influence on methane production was relevant to the dosage. At a lower ZnO NPs dosage (1 mg/g-TSS), no inhibitory effect was observed (Fobserved ¼ 0.05, Fsignificance ¼ 7.71, P (0.05) ¼ 0.83 > 0.05). When the dosage of ZnO NPs was 30 mg/gTSS, however, the average methane production decreased to 81.7% of the control, which was further decreased to 24.9% of the control as the dosage of ZnO NPs increased to 150 mg/g-
5615
TSS. Apparently, higher concentrations (30 and 150 mg/g-TSS) of ZnO NPs were capable of inhibiting the methane production. Some nanomaterials, such as fullerene, Au, Ag and Fe3O4, were reported to give marginal influence on anaerobic community (Barrena et al., 2009; Nyberg et al., 2008). Nevertheless, the Gram-positive B. subtilis was observed to be more sensitive to ZnO NPs than Gram-negative E. coli. (Adams et al., 2006), and ZnO NPs were found to negatively affect the soil bacterial community (Ge et al., 2011). It seems that it is difficult to figure out the toxicity of ZnO NPs on microorganism involved in WAS digestion according to the current ZnO NPs toxicology information as various species of bacteria are in sludge anaerobic digestion system, and WAS anaerobic digestion for methane production usually includes sludge solubilization, hydrolysis, acidification and methanation (Fig. 1). According to our knowledge, the effects of ZnO NPs on the microbial community and each step involved in anaerobic digestion have never been documented, which will be investigated in detail in the following text to understand the mechanisms of ZnO NPs affecting methane production during WAS anaerobic digestion.
3.2. Effects of ZnO nanoparticles on sludge surface and Zn2þ release as well as ROS change The SEM analysis has been applied in literature to investigate the adsorption of NPs to sludge (Kiser et al., 2009). As seen in Fig. 3, there were large numbers of ZnO NPs on the surface of sludge after long-term exposed to ZnO NPs. The same observations were reported by other researchers when the behavior of NPs in wastewater treatment system was studied (Kiser et al., 2010; Limbach et al., 2008). At ZnO NPs dosages of 1, 30 and 150 mg/g-TSS, respectively, the corresponding released Zn2þ concentrations were 1.2, 11.6 and 17.6 mg/L (Fig. S2, Supplementary Information). The longterm impact of released Zn2þ on methane production during WAS anaerobic digestion is shown in Fig. 2. The presence of 1.2 mg/L of Zn2þ did not give any significant impact on the methane production ( p > 0.05). It might be that some chemical compounds, such as sulfate (11.5 mg/L) in sludge, was bioconverted to sulfide by sulfate reducing bacterial under anaerobic conditions, and then the sulfide reacted with Zn2þ and thus reduced the toxicity of Zn2þ. However, the methane production was 90.6% of the control at a Zn2þ concentration of 11.6 mg/L. When the Zn2þ was 17.6 mg/L, a much lower methane production (36.2% of the control) was observed. It can be seen
Fig. 3 e Scanning electron micrographs imaging of sludge long-term exposed to 0 mg/g-TSS (A), 1 mg/g-TSS (B), 30 mg/g-TSS (C), and 150 mg/g-TSS (D) of ZnO NPs during WAS anaerobic digestion.
5616
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
160
*
120
120
* 80
80
*
40
40
0
Relativ e activ ity o f en zy m e ( % o f co nt r o l)
160
200
*
ROS production Biomass viability
Relative biomass viability (% of control)
Relative ROS production (% of control)
200
1 mg/g-TSS
30 mg/g-TSS
150 mg/g-TSS
*
100
* 80
*
60
40
20
0 1
30
150
0 Protease
ZnO NPs (mg/g-TSS)
Fig. 4 e Effects of different dosages of ZnO NPs (1, 30 and 150 mg/g-TSS) on the intracellular ROS production and biomass viability. Asterisks indicate statistical differences ( p < 0.05) from the control. Error bars represent standard deviations of triplicate tests.
A
Protein
1500
Individual SCFA concentration (mg-COD / L)
Concentrations (mg-COD/L)
80
60
40
20
Propionic n-Butyric
400
200
B
D
Dextran 200
*
*
40
20
0
61
30
150
Innoculum sludge long-term exposed to different dosages of ZnO NPs (mg/g-TSS)
Cumulative methane production (mL / g-COD )
Degradation efficiency (%)
Acetic iso-Butyric
0
BSA
60
0
C
1400
0 80
F 420
Fig. 6 e Comparisons of the activities of protease, AK and coenzyme F420 in the long-term operated reactors exposed to different dosages of ZnO NPs (1, 30 and 150 mg/g-TSS). Asterisks indicate statistical differences ( p < 0.05) from the control reactor. Error bars represent standard deviations of triplicate tests.
Polysaccharide
100
AK
*
150
100
* 50
0
0
16
30
150
Innoculum sludge long-term exposed to different dosages of ZnO NPs (mg/g-TSS)
Fig. 5 e Effects of ZnO NPs on each step of sludge anaerobic digestion. A: the concentrations of soluble protein and carbohydrate during the initial 2 d; B: the degradation of solubilized products (BSA and dextran) with time of 4 d; C: the concentrations of acidification products (individual SCFA) with time of 4 d; D: the methanation products (methane) at time of 14 d. Asterisks indicate statistical differences ( p < 0.05) from the control. Error bars represent standard deviations of triplicate tests.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
from Fig. 2 that the impact of ZnO NPs on methane production mainly resulted from the dissolved Zn2þ. In a recent publication Liu et al. (2011) also reported that the released Zn2þ from ZnO NPs played an important role on the adverse effect of ZnO NPs on the performance of biological wastewater treatment process. In the literature the toxicity of ZnO NPs to some microbes was also observed to come from the released Zn2þ, but those studies focused on cell growth instead of microbial function (Franklin et al., 2007; Wong et al., 2010; Xia et al., 2008).
5617
The ROS induced by ZnO NPs was reported to be one reason for their toxicity, which caused the loss of cell viability (Xia et al., 2008). ZnO NPs were regarded as an exogenous source of ROS for cells or organisms in some previous reports (Joshi et al., 2009; Xia et al., 2008). As seen in Fig. 4, an increase of the intracellular ROS production was observed with the increase of ZnO NPs. Usually, ROS, including superoxide (O2-), hydrogen peroxide (H2O2), and the hydroxyl radical (OH), are produced in the presence of oxygen (Murphy, 2009). However,
Fig. 7 e Fluorescence in situ hybridization of sectios of biomass long-term (more than 105 d) cultured respectively in the absence of ZnO NPs (A1-A3) and in the presence of 1 mg/gTSS (B1-B3), 30 mg/g-TSS (C1-C3) and 150 mg/g-TSS of ZnO NPs (D1-D3) viewed by CLSM and photographed at higher (362) magnification. The sections were simultaneously hybridized with Cy-3-labeled bacterial-domain probe (EUB338) (red) and FITC-labeled archaeal-domain probe (ARC915) (green). Overlay of ARC915 (green) and EUB338 (red) are shown in A3, B3, C3 and D3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
5618
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
it has been reported that H2O2 can also be produced under anaerobic conditions (Degli-Esposti and McLennan, 1998). The increase of ROS in the sludge exposed to higher dosages of ZnO NPs was a likely reason for their adverse effect on sludge anaerobic digestion. It can be seen from Fig. 4 that the result of biomass viability assay was consistent with the ROS production, which decreased from 97.3% of the control ( p > 0.05) to 88.7% of the control when ZnO NPs increased from 1 to 30 mg/ g-TSS. At ZnO NPs dosage of 150 mg/g-TSS, the relative biomass viability further decreased to 62.4% of the control.
the bio-conversion step of acetic acid to methane. At other time the same observations could be made (Fig. S3, Supplementary Information). By comparing the data in Fig. 5C and D, it seems that methanogens are more sensitive to the toxicity of ZnO NPs than acidogens. In the literature, some researchers reported that acidogens were more resistant to metal toxicity than methanogens (Zayed and Winter, 2000). In addition, the data in Fig. 5AeD suggested that the negative influence of ZnO NPs on the methanation step was the most serious one among the four steps.
3.3. Effects of ZnO nanoparticles on each step involved in methane production
3.4.
Protein and carbohydrate, the main constituents of WAS (accounting for 61.5% of sludge TCOD), are usually in particulate state. The batch experiments were conducted to investigate the long-term effects of ZnO NPs on sludge particulate protein and carbohydrate solubilization. The effects of ZnO NPs on solubilization of sludge particulate organic matters were expressed by the changes of soluble protein and carbohydrate production in this study. As seen from Fig. 5A, there were no significant differences in the concentrations of soluble protein and carbohydrate after the initial 2 d fermentation ( p > 0.05). It might be that the solubilization of sludge particulate organic matters was not a microbial process, which resulted in the influences of ZnO NPs on the concentrations of soluble protein and carbohydrate not being observed. The long-term effects of three dosages of ZnO NPs on the hydrolysis of sludge solubilized products (soluble protein and carbohydrate) with time of 4 d are shown in Fig. 5B. The degradation of dextran (model carbohydrate mater) in the control reactor was almost the same as those in other three ZnO NPs reactors. Nevertheless, the influence of ZnO NPs on the degradation of BSA (model protein) was dosage dependent. At dosages of 1 and 30 mg/g-TSS, the influences of ZnO NPs were insignificant ( p > 0.05), but the degradation of BSA at 150 mg/g-TSS of ZnO NPs was lower than that in the control (58.5% versus 65.1%). It might be one reason for the decreased methane production exposed to higher concentrations of ZnO NPs. Fig. 5C illustrates the long-term effects of different concentrations of ZnO NPs on the acidification of main hydrolyzed products (amino acid and monosaccharide) to SCFA during the initial 4 d. The influences of ZnO NPs on the composition of SCFA were insignificant (see Table S1 for statistical analysis, Supplementary Information). The total SCFA concentrations, which calculated from Fig. 5C, were 2078 80, 2057 80, 2045 69 and 2050 50 mg-COD/L in the reactors of control, and 1, 30 and 150 mg/g-TSS of ZnO NPs, respectively. Obviously, the acidification step involved in sludge digestion was not affected by ZnO NPs. As to the influence of ZnO NPs on the methanation step, the data in Fig. 5D indicated that there was no significant difference in the cumulative methane production between the control and the 1 mg/g-TSS of ZnO NPs reactors at time of 14 d ( p > 0.05). However, the methane productions were 83.0% and 28.1% of the control at 30 and 150 mg/g-TSS of ZnO NPs, respectively, suggesting that ZnO NPs significantly inhibited
Determination of key enzyme activity
Further investigation showed that ZnO NPs influenced the activities of enzymes relevant to sludge anaerobic digestion. Although large numbers of enzymes took part in methane production during sludge anaerobic digestion, in this study only three enzymes responsible respectively for sludge hydrolysis (i.e., protease), acidification (AK) and methanation (coenzyme F420) (Fig. 1) were assayed as examples. The relative activities of these enzymes in the long-term operated reactors are demonstrated in Fig. 6. The AK activity did not change significantly with ZnO NPs dosages ( p > 0.05). To the protease, the dosage of 150 mg/g-TSS of ZnO NPs remarkably reduced its activity. The coenzyme F420 activity, however, was ZnO NPs dosage dependent, which was respectively 99.3%, 89.8% and 66.2% of the control at ZnO NPs of 1, 30 and 150 mg/g-TSS. Apparently, not only the hydrolysis of soluble protein but the transformation activity of electron donors of the redox-driven proton translocation in methanogenic Archaea (expressed by coenzyme F420 (Deppenmeier, 2002)) was significantly restrained by higher concentrations of ZnO NPs. All these consisted well with the above observed synthetic wastewater experimental results.
3.5.
FISH analysis results
For the purpose of investigating the influence of ZnO NPs on the abundance of bacteria and Archaea, the FISH assay was further conducted (Fig. 7), and the results were analyzed with image analysis system (Image-Pro Plus, V6.0). It was found that there were 39.5% of Archaea and 52.6% of bacteria in the control reactor. In ZnO NPs reactors, the Archaea were 38.6% (1 mg/g-TSS), 27.1% (30 mg/g-TSS), and 3.5% (150 mg/g-TSS), and the corresponding bacteria were 51.3%, 60.8% and 87.4%, respectively. The ratios of Archaea to bacteria were 0.8:1, 0.9:1, 0.4:1 and 0.04:1, respectively, in the reactors of control, and 1, 30 and 150 mg/g-TSS of ZnO NPs, respectively. Obviously, the more ZnO NPs appeared in sludge anaerobic digestion system, the less Archaea remained, which was consistent with the observed methane production when WAS was long-term exposed to different dosages of ZnO NPs.
4.
Conclusions
The above studies indicated that the methane production during sludge anaerobic digestion was not affected by ZnO NPs of 1 mg/g-TSS. Nevertheless, due to large numbers of Zn2þ release the higher dosages (30 and 150 mg/g-TSS) of ZnO NPs
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
inhibited the production of methane. By investigating the four stages involved in sludge anaerobic digestion, i.e., solubilization, hydrolysis, acidification and methanation, it was found that the activities of protease and coenzyme F420 were negatively influenced by higher dosages of ZnO NPs, which suggested that only the steps of hydrolysis and methanation were inhibited. The molecular biology studies indicated that a lower abundance of methanogenesis Archaea was observed at higher ZnO NPs dosages exposure.
Acknowledgements This work was financially supported by the Foundation of State Key Laboratory of Pollution Control and Resource Reuse (PCRRK09002).
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.022.
references
Adams, L.K., Lyon, D.Y., Alvarez, P.J.J., 2006. Comparative ecotoxicity of nanoscale TiO2, SiO2, and ZnO water suspensions. Water Research 40 (19), 3527e3532. Barrena, R., Casals, E., Colon, J., Font, X., Sanchez, A., Puntes, V., 2009. Evaluation of the ecotoxicity of model nanoparticles. Chemosphere 75 (7), 850e857. Brar, S.K., Verma, M., Tyagi, R.D., Surampalli, R.Y., 2010. Engineered nanoparticles in wastewater and wastewater sludge-Evidence and impacts. Waste Management 30 (3), 504e520. Degli-Esposti, M, McLennan, H, 1998. Mitochondria and cells produce reactive oxygen species in virtual anaerobiosis: relevance to ceramide-induced apoptosis. FEBS Letters 430 (3), 338e342. Deppenmeier, U., 2002. Redox-driven proton translocation in methanogenic archaea. Cellular and Molecular Life Sciences 59 (9), 1513e1533. Ellsworth, D.K., Verhulst, D., Spitler, T.M., Sabacky, B.J., 2000. Titanium nanoparticles move to the marketplace. Chemical Innovation 30 (12), 30e35. EPA, 2009. Targeted National Sewage Sludge Survey Sampling and Analysis Technical Report. Environmental Protection Agency, Washington, DC: United States. http://www.epa.gov/ waterscience/biosolids/tnsss-tech.pdf. Franklin, N.M., Rogers, N.J., Apte, S.C., Batley, G.E., Gadd, G.E., Casey, P.S., 2007. Comparative toxicity of nanoparticulate ZnO, bulk ZnO, and ZnCl2 to a freshwater microalga (Pseudokirchneriella subcapitata): the importance of particle solubility. Environmental Science and Technology 41 (24), 8484e8490. Ganesh, R., Smeraldi, J., Hosseini, T., Khatib, L., Olson, B.H., Rosso, D., 2010. Evaluation of nanocopper removal and toxicity in municipal wastewaters. Environmental Science and Technology 44 (20), 7808e7813. Garcia, M.T., Campos, E., Sanchez-Leal, J., Ribosa, I., 2006. Effect of linear alkylbenzene sulphonates (LAS) on the anaerobic digestion of sewage sludge. Water Research 40 (15), 2958e2964.
5619
Ge, Y., Schimel, J.P., Holden, P.A., 2011. Evidence for negative effects of TiO2 and ZnO nanoparticles on soil bacterial communities. Environmental Science and Technology 45 (4), 1659e1664. Gottschalk, F., Sonderer, T., Scholz, R.W., Nowack, B., 2009. Modeled environmental concentrations of engineered nanomaterials (TiO2, ZnO, Ag, CNT, Fullerenes) for different regions. Environmental Science and Technology 43 (24), 9216e9222. Grady, C.P.L., Daigger, G.T., Lim, H.C., 1999. Biological Wastewater Treatment. Marcel Dekker Inc., New York. Jiang, W., Mashayekhi, H., Xing, B., 2009. Bacterial toxicity comparison between nano- and micro-scaled oxide particles. Environmental Pollution 157 (5), 1619e1625. Joshi, P., Chakraborti, S., Chakrabarti, P., Haranath, D., Shanker, V., Ansari, Z.A., Singh, S.P., Gupta, V., 2009. Role of surface adsorbed anionic species in antibacterial activity of ZnO quantum dots against Escherichia coli. Journal of Nanoscience and Nanotechnology 9 (11), 6427e6433. Keller, A.A., Wang, H.T., Zhou, D.X., Lenihan, H.S., Cherr, G., Cardinale, B.J., Miller, R., Ji, Z.X., 2010. Stability and aggregation of metal oxide nanoparticles in natural aqueous matrices. Environmental Science and Technology 44 (6), 1962e1967. Kiser, M.A., Ryu, H., Jang, H.Y., Hristovski, K., Westerhoff, P., 2010. Biosorption of nanoparticles to heterotrophic wastewater biomass. Water Research 44 (14), 4105e4114. Kiser, M.A., Westerhoff, P., Benn, T., Wang, Y., Perez-Rivera, J., Hristovski, K., 2009. Titanium nanomaterial removal and release from wastewater treatment plants. Environmental Science and Technology 43 (17), 6757e6763. Li, M., Zhu, L., Lin, D., 2011. Toxicity of ZnO nanoparticles to Escherichia coli: mechanism and the influence of medium components. Environmental Science and Technology 45 (5), 1977e1983. Limbach, L.K., Bereiter, R., Mueller, E., Krebs, R., Gaelli, R., Stark, W.J., 2008. Removal of oxide nanoparticles in a model wastewater treatment plant: influence of agglomeration and surfactants on clearing efficiency. Environmental Science and Technology 42 (15), 5828e5833. Liu, G.Q., Wang, D.M., Wang, J.M., Mendoza, C., 2011. Effect of ZnO particles on activated sludge: Role of particle dissolution. Science of the Total Environment 409 (14), 2852e2857. Maynard, A.D., Aitken, R.J., Butz, T., Colvin, V., Donaldson, K., Oberdorster, G., Philbert, M.A., Ryan, J., Seaton, A., Stone, V., Tinkle, S.S., Tran, L., Walker, N.J., Warheit, D.B., 2006. Safe handling of nanotechnology. Nature 444 (7117), 267e269. Miziolek, A., 2002. Nanoenergetics: an emerging technology area of national importance. AMPTIAC Quarterly 6 (1), 43e48. Murphy, M.P., 2009. How mitochondria produce reactive oxygen species. Biochemical Journal 417, 1e13. Nyberg, L., Turco, R.F., Nies, L., 2008. Assessing the impact of nanomaterials on anaerobic microbial communities. Environmental Science and Technology 42 (6), 1938e1943. Oh, S.E., Van Ginkel, S., Logan, B.E., 2003. The relative effectiveness of pH control and heat treatment for enhancing biohydrogen gas production. Environmental Science and Technology 37 (22), 5186e5190. Roco, M.C., 2005. The emergence and policy implications of converging new technologies integrated from the nanoscale. Journal of Nanoparticle Research 7 (2e3), 129e143. Serda, R.E., Ferrati, S., Godin, B., Tasciotti, E., Liu, X.W., Ferrari, M., 2009. Mitotic trafficking of silicon microparticles. Nanoscale 1 (2), 250e259. Sharma, V., Shukla, R.K., Saxena, N., Parmar, D., Das, M., Dhawan, A., 2009. DNA damaging potential of zinc oxide nanoparticles in human epidermal cells. Toxicology Letters 185 (3), 211e218. Simon-Deckers, A., Loo, S., Mayne-L’Hermite, M., HerlinBoime, N., Menguy, N., Reynaud, C., Gouget, B., Carriere, M.,
5620
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 1 2 e5 6 2 0
2009. Size-, composition- and shape-dependent toxicological impact of metal oxide nanoparticles and carbon nanotubes toward bacteria. Environmental Science and Technology 43 (21), 8423e8429. Wong, S.W.Y., Leung, P.T.Y., Djurisic, A.B., Leung, K.M.Y., 2010. Toxicities of nano zinc oxide to five marine organisms: influences of aggregate size and ion solubility. Analytical and Bioanalytical Chemistry 396 (2), 609e618. Xia, T., Kovochich, M., Liong, M., Madler, L., Gilbert, B., Shi, H., Yeh, J., Zink, J.I., Nel, A.E., 2008. Comparison of the mechanism of toxicity of zinc oxide and cerium oxide nanoparticles based on dissolution and oxidative stress properties. ACS Nano 2 (10), 2121e2134.
Yuan, H., Chen, Y., Zhang, H., Jiang, S., Zhou, Q., Gu, G., 2006. Improved bioproduction of short-chain fatty acids (SCFAs) from excess sludge under alkaline conditions. Environmental Science and Technology 40 (6), 2025e2029. Zayed, G., Winter, J., 2000. Inhibition of methane production from whey by heavy metals - protective effect of sulfide. Applied Microbiology and Biotechnology 53 (6), 726e731. Zhao, Y., Chen, Y., Zhang, D., Zhu, X., 2010. Waste activated sludge fermentation for hydrogen production enhanced by anaerobic process improvement and acetobacteria inhibition: the role of fermentation pH. Environmental Science and Technology 44 (9), 3317e3323.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Genetic potential for N2O emissions from the sediment of a free water surface constructed wetland Arantzazu Garcı´a-Lledo´ a, Ariadna Vilar-Sanz a, Rosalia Trias a, Sara Hallin b, Lluı´s Ban˜eras a,* a b
Molecular Microbial Ecology Group, Institute of Aquatic Ecology, Universitat de Girona, C/ Maria Aure`lia Capmany, 69, 17071 Girona, Spain Swedish University of Agricultural Sciences, Department of Microbiology, Box 7025, Uppsala 750 07, Sweden
article info
abstract
Article history:
Removal of nitrogen is a key aspect in the functioning of constructed wetlands. However,
Received 13 April 2011
incomplete denitrification may result in the net emission of the greenhouse gas nitrous
Received in revised form
oxide (N2O) resulting in an undesired effect of a system supposed to provide an ecosystem
22 July 2011
service. In this work we evaluated the genetic potential for N2O emissions in relation to the
Accepted 14 August 2011
presence or absence of Phragmites and Typha in a free water surface constructed wetland
Available online 30 August 2011
(FWS-CW), since vegetation, through the increase in organic matter due to litter degradation, may significantly affect the denitrification capacity in planted areas. Quantitative
Keywords:
real-time PCR analyses of genes in the denitrification pathway indicating capacity to
Constructed wetlands
produce or reduce N2O were conducted at periods of different water discharge. Genetic
Denitrification
potential for N2O emissions was estimated from the relative abundances of all denitrifi-
Genetic potential
cation genes and nitrous oxide reductase encoding genes (nosZ ). nosZ abundance was
N2O emission
invariably lower than the other denitrifying genes (down to 100 fold), and differences
qPCR
increased significantly during periods of high nitrate loads in the CW suggesting a higher
Vegetation
genetic potential for N2O emissions. This situation coincided with lower nitrogen removal efficiencies in the treatment cell. The presence and the type of vegetation, mainly due to changes in the sediment carbon and nitrogen content, correlated negatively to the ratio between nitrate and nitrite reducers and positively to the ratio between nitrite and nitrous oxide reducers. These results suggest that the potential for nitrous oxide emissions is higher in vegetated sediments. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Constructed wetlands (CWs) are recognized as feasible alternatives for removal of nitrogen in wastewater from agricultural, industrial or municipal activities and have been exploited as secondary or tertiary treatment alternatives to promote water reuse (Reed et al., 1995; DeBusk and DeBusk, 2000). There are two types of CWs, the subsurface flow systems (SSF) and the free-water surface systems (FWS) that
mainly differ in the presence of a free water flow over the sediment surface. Water treatment in CWs involves processes driven by the sediment, the vegetation and its associated microbial communities (EPA, 2000). It is generally assumed that areas planted with emergent macrophytes positively affect the water restoration capacity of the wetland (Zhu and Sikora, 1995; Lin et al., 2002; Ibekwe et al., 2007). This is due to stimulation of microbial growth at either the epiphyton or the root surface, e.g. by influencing the oxygen conditions and
* Corresponding author. Tel.: þ34 972 418 177; fax: þ34 972 418 150. E-mail addresses: [email protected] (A. Garcı´a-Lledo´), [email protected] (A. Vilar-Sanz), [email protected] (R. Trias), [email protected] (S. Hallin), [email protected] (L. Ban˜eras). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.025
5622
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
creating microenvironments with oxiceanoxic zones (Gersberg et al., 1986; Zhu and Sikora, 1995; Boudraa et al., 1999) or by the excretion of organic compounds from the roots (Prade and Trolldenier, 1990; Brix, 1997; Nguyen, 2003), as well as causing changes in hydraulic retention time (Whitney et al., 2003; Toet et al., 2005; Kjellin et al., 2007). All these processes influence nitrogen removal in CWs since they affect the nitrifying and denitrifying microorganisms, which are ultimately responsible for nitrogen transformations resulting in nitrogen removal (Johnston, 1991; Zhu and Sikora, 1995; Boudraa et al., 1999; Ko¨rner, 1999; Vymazal, 2001; Lin et al., 2002; Francis et al., 2007). The microbial process denitrification is especially interesting in CWs as it represents the net nitrogen loss from the system. Denitrifying bacteria use nitrate as an alternative electron acceptor, which is sequentially reduced to nitrogen gas. The first step is catalyzed by nitrate reductases, either a membrane-bound enzyme (Nar type) or a soluble periplasmic enzyme (Nap) (Zumft, 1997; Moreno-Vivian et al., 1999; Philippot, 2002). These two reductases are not exclusive, but can be found simultaneously in the same organism (Carter et al., 1995; Gregory et al., 2003; Roussel-Delif et al., 2005). Dissimilatory nitrite reductases (Nir) catalyze the second step of denitrification and two functionally equivalent enzymes have been described, a cytochrome cd1 type and a copper containing type, encoded by the nirS and nirK genes, respectively. They are mutually exclusive and have not been found in the same strain so far (Zumft, 1997), although different strains of the same species may contain different Nir genes (Coyne et al., 1989; Philippot, 2002). Nitric oxide reductases (Nor) catalyze the reduction of nitric oxide to nitrous oxide and the last step of denitrification is catalyzed by nitrous oxide reductases (Nos) which lead to the production of nitrogen gas (Zumft, 1997). All denitrifying genes described so far have been used as molecular markers for qualitative and quantitative studies of denitrifying bacteria in the environment. The expression of denitrifying genes is dependent on the presence of the respective enzymes substrates, i.e. sequential oxidized forms of inorganic nitrogen, and low oxygen concentration due to the facultative nature of this process (Tiedje, 1988). Most denitrifiers use organic compounds as electron donors and easily available carbon is another prerequisite. Emissions of the potent greenhouse gas N2O may be due to the activity of some denitrifying bacteria that are unable to perform the final step of denitrification, due to the lack of the nosZ gene (Wood et al., 2001; Kandeler et al., 2006; Vial et al., 2006; Jones et al., 2008; Abell et al., 2010; Philippot et al., 2011). Cheneby et al., 2004 compared denitrifying bacteria between non-planted and maize planted soil and found that nosZ lacking bacteria were dominant in the rhizosphere suggesting that plants can affect N2O emission by selecting for denitrifiers that do not have the capacity to reduce N2O. Plants increase easily available carbon compounds by root exudates and through the increase in organic matter due to litter degradation. In addition, plants cause radial oxygen gradients around the roots. These factors affect denitrifying bacteria, as well as other heterotrophic bacteria (Boudraa et al., 1999; Henry et al., 2008). Studies specifically directed to the analysis of denitrification in the rhizosphere have shown that changes in the activity, composition and the abundance of
denitrifying bacteria may be plant-specific (Cheneby et al., 2004; Henry et al., 2008; Ruiz-Rueda et al., 2009). In the sediment of planted areas the presence of decomposing litter may be an additional factor supporting and shaping the denitrifying community (Ingersoll and Baker, 1998). If emergent macrophytes in CWs also select for denitrifiers genetically incapable of reducing N2O is not known. If so, although promoting nitrogen removal in CWs, macrophytes potentially contribute to increase the N2O/N2 end-product ratio. We evaluated the genetic potential for N2O emissions of the sediment in relation to the presence or absence of the main plant species used in treatment wetlands, Phragmites australis and Typha latifolia. For that purpose, we determined the abundance of denitrifiers by quantitative real-time PCR (qPCR) of key functional genes in the denitrification pathway (narG, napA, nirS, nirK and nosZ ) in comparison to the total bacterial community size targeted by the 16S rRNA gene in the sediments of the Empuriabrava FWS-CWs at periods of different water discharge over the year.
2.
Material and methods
2.1.
Study site
The Empuriabrava FWS-CWs (Girona, NE Spain) were designed in 1998 as a tertiary treatment to increase the water quality of the effluent of a nearby located wastewater treatment plant (WWTP). The CWs are included in the natural preserved area of Els Aiguamolls de l’Emporda` and are designed to provide additional water to avoid excessive desiccation of the flooded area in summer (Ruiz-Rueda et al., 2007). Nitrogen is removed at the WWTP by a combined nitrificationedenitrification process using carrousel-type bioreactors and enhanced aeration, resulting in nitrate loads to the CWs changing between 251 and 1016 kg N per month. The Empuriabrava CWs consists of three parallel cells followed by a shallower lagoon and all measurements were conducted on treatment cell 3. The cell is planted with independent communities of reed (P. australis) and cattail (T. latifolia) showing an almost equivalent distribution of both plant species (49.3% of reed and 50.7% of cattail; Fig. 1). The average water depth in the cell was about 0.6 m and the sediment was water saturated during the sampling period. The sediment depths varied between 5 and 10 cm in non-vegetated areas and increased to 15e20 cm in planted surfaces. The location of the vegetation spots and the plant species were designed in the original project and planted shortly after construction of the CW (1998). Harvest of the aerial biomass is performed every one or two years during winter.
2.2.
Sampling and DNA extraction
Temperature, conductivity, oxygen and pH at the sampling locations were measured with a portable multiparametric probe (Yellow Spring Instruments 650MDS). Water samples (20 ml) from the water column in the wetland were collected and analyzed for nitrate, nitrite and ammonia concentration as described previously (Garcı´a-Lledo´ et al., 2011). Water samples (100 ml) were regularly also collected at the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
5623
Fig. 1 e Scheme of vegetation stands and sampling locations in the third cell of the Empuriabrava FWS-CWs. Sampling plots are indicated as black dots with the following two-letter codes: SE e bulk sediment, TY e areas planted with Typha latifolia, PH e areas planted with Phragmites australis.
secondary effluent of the WWTP and at the end point of the treatment cells for analyses of pH, conductivity, biological oxygen demand (BOD), chemical oxygen demand (COD), oxygen saturation, and concentrations of nitrate, nitrite and ammonium, using conventional standard methods for wastewater analyses (APHA, 1998). Total water flow into the wetland system was automatically monitored daily. Sediment samples were collected May 2008, August 2008, March 2009 and July 2009 near the inlet and outlet areas. Three sampling plots of one square meter each were defined at the two areas. Sampling plot 1 was not covered with any emergent plant and is referred as bulk sediment (SE). Plot 2 consisted of an area covered with T. latifolia (TY) and Plot 3 was covered with P. australis (PH). Three sediment replicates were taken in each plot (Fig. 1). Sediment cores were obtained from planted and unplanted areas using a 2-cm-diameter methacrylate tube mounted in a manual sampler. The upper 3 cm of the sediment core were aseptically transferred to a container and chilled on ice for transportation. Once in the laboratory, sediment was completely homogenized and triplicates of 2 g aliquots were stored at 80 C until processed. Total nucleic acids were extracted using a modified CTAB protocol previously described for the simultaneous recovery of DNA and RNA from soils (Hurt et al., 2001). Purification of DNA was done with AllPrep DNA/RNA Mini Kit (Qiagen) according to the manufacturer’s instructions. DNA extraction and quality was checked with a 0.8% agarose gel. DNA concentration was quantified with a NanoVue Plus Spectrophotometer (GE Healthcare).
2.3.
Sediment chemical analyses
The content of TC and TN in the sediment were analyzed by combustion of dried samples at 975 C in a Perkin Elmer AE SeriesII equipped with a TCD detector. The results were evaluated using the K factor method with cystine (C6H12N2O4S2) as a standard. Duplicates were always performed for all chemical determinations. Due to the low nitrogen content of the sediment, in most cases near the detection limit, total Kjeldahl nitrogen (TKN) was also
measured using standard methods for sediments (APHA, 1998). Sediment pH was measured from 1/5 (dry weight/ volume) sediment samples homogenized with double distilled water.
2.4.
Quantitative PCR of functional genes
Bacteria involved in the denitrification processes were estimated by the quantification of key functional genes using quantitative real-time PCR. The 16S rRNA gene was also quantified to estimate the total amount of bacteria. Primers and thermal cycling conditions used for each reaction have previously been described for narG and napA (Bru et al., 2007) and for nirS, nirK, nosZ and 16S rRNA genes (Hallin et al., 2009). All qPCR reactions were performed on a Bio-Rad IQ5 thermal cycler (Bio-Rad Laboratories, Inc.) in a total volume of 20 ml, containing 1X Phire Hot Start II DNA Polymerase (Finnzymes Oy. Espoo, Finland), 1 mM of each primer, 103 ng ml1 of Bovine Serum Albumin (BSA) and 1 ng of DNA template. Results were analyzed using Bio-Rad IQ5 software. Standard curves were obtained using serial dilutions from 102 to 108 copies of linearized plasmids containing the respective functional genes. Controls without templates gave null or negligible values. To ensure that sediment samples did not have inhibitory effects on PCR performance an inhibitory test were run with all samples at the working concentration together with a known amount of circular plasmid. The measured threshold cycle (Ct) values were compared with those of a control of the plasmid mixed with water. Despite the use of highly diluted DNA extractions (down to 1 ng), inhibition effects could not be removed from one sample of a Phragmites covered sediment collected in May 2008 and three samples from sediments covered with Typha collected in August 2008 and March 2009. These samples have been removed from the statistical analyses.
2.5.
Statistical analyses
All statistical analyses were performed using SPSS for Windows 15.0 (SPSS, Inc). Measured gene abundances were log transformed in order to ensure a normal distribution of
5624
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
data, which was checked by KolmogoroveSmirnov and ShapiroeWilk tests. Differences in gene abundances or gene ratios were tested for the effects of planted vs. non-planted sediments, sampling time, location of sampling plots and the corresponding interactions by one-way ANOVA and Welch tests. Differences between means of either gene were analyzed using GameseHowell test (for unequal variance data) post hoc analyses for multiple comparisons. To test for effects of sampling period, type of sediment, position inside the treatment cell and interactions on denitrifying community, univariate general linear model (GLM) analyses were performed. Correlation analysis of microbial abundances with physical and chemical variables was performed using nonparametric Spearman’s correlation test. Paired samples t-test was chosen to compare differences between genes coding for enzymes catalyzing the same reaction (i.e. narG vs. napA). The significance level for all tests was 0.05.
3.
Results
3.1.
Characterization of the Empuriabrava FWS-CWs
During the studied period, the water inflow to the wetlands presented stable pH conditions, with values around 7.6. BOD was low, usually under 3 mg O2/l, except for occasional dates when higher concentrations were recorded (Fig. 2). COD varied from 30 to 121 mg O2/l, but no seasonal trends were observed in the variation. Ammonia concentrations were low, with slightly higher values between July and September (up to 4.5 mg NeNHþ 4 /l). In contrast, nitrate concentrations showed a higher variation, between 0.1 and 15.8 mg NeNO 3 /l, with concentration peaks occurring intermittently throughout the year due to changes in nitrogen removal efficiencies in the WWTP. The water flowing through the wetlands had strong seasonal variations, from 1500 m3/day (February) to almost 6000 m3/day (August). This large variation is due to the intense tourism in a nearby located touristic area that increases water consumption during summer. Physicochemical parameters of the water and sediment measured at the sampling locations are presented in Table 1. Water conductivity values varied from 2.7 (August 2008) to 7.8 mV/cm (March 2009). There was a significant negative correlation between conductivity and water temperature (Spearman’s correlation coefficient r ¼ 0.641, p < 0.001, results not shown). The higher conductivity values during winter are due to the infiltration of sea water in some of the wastewater network collectors, which is diluted during summer due to a higher wastewater volume to be treated at the WWTP. The water flow to the wetland varied significantly during the four sampling periods, receiving 2437 and 2202 m3/ day in May 2008 and March 2009, and 6007 and 4379 m3/day in August 2008 and July 2009, respectively. Hydraulic retention times (HRT) for treatment cell 3 were calculated during the sampling periods by independent measurements of the water flow, with values ranging from 4.2 to 15.5 days. Unfortunately, the water flow was unexpectedly cut for a short period in July 2009 and these conditions are not indicative of the average situation in the cell. To circumvent this problem, the monthly
average water flow to the wetland system was used in the statistical analyses. Differences in both water and sediment pH were not extremely large and values ranged from 7.2 to 8.7 and from 6.9 to 7.9, respectively. Water nitrate and ammonium concentrations at the sampling locations ranged between 0.01 and þ 1.24 mg NeNO 3 /l and 0.06 and 0.49 mg NeNH4 /l, respectively. Nitrite concentrations were not detectable. Positive correlations were obtained between conductivity and nitrate (r ¼ 0.350, p < 0.01) or ammonium (r ¼ 0.275, p < 0.05). Higher TC and TKN values were usually obtained in sediments collected from areas planted with P. australis ( p < 0.001). Nitrogen removal efficiencies (%) were calculated on the basis on the total nitrogen load and changes in the inlet and outlet concentrations in treatment cell 3 (Garcı´a-Lledo´ et al., 2011). Efficiencies were estimated to be 48 and 61% in periods of low water discharge (May 2008 and March 2009, respectively) and increased up to 71% when higher inflows entered the CWS (July 2009). Nitrogen removal efficiencies correlated positively with ammonia concentration (r ¼ 0.685, p < 0.01) and negatively with nitrate concentrations (r ¼ 0.640, p < 0.01) at the influent. Moreover, oxygen and temperature showed a negative (r ¼ 0.484, p < 0.01) and positive (r ¼ 0.703, p < 0.01) correlation to nitrogen removal efficiencies, respectively.
3.2.
Abundance of 16S rRNA and denitrification genes
The abundance of 16S rRNA genes ranged from 1.3 1011 to 5.8 1012 copies/g dw sediment and appeared to be fairly stable among samples and over time (Fig. 3; Supplementary Table 1). These numbers were always higher than those obtained for any of the functional genes. Mean values for all determined functional gene abundances were between 1 and 4 logs below the 16S rRNA gene. The narG and napA genes ranged from 8.8 108 to 7.1 1010 and from 1.2 109 to 6.4 1010 copies/g dw sediment, respectively. Mean napA gene numbers were significantly higher than narG according to a paired sample t-test (t ¼ 2.578, p < 0.05), but these differences were exclusively due to dominance of napA genes in bulk sediments in March 2009 and no differences were detected when the March samples were excluded (t ¼ 1.116, p > 0.05). Sediment samples showed between 4.9 107 and 6.9 109 copies/g dw sediment for nirS and from 1.6 109 to 2.1 1011 copies/g dw sediment for nirK, suggesting a significantly greater abundance of nirK-type denitrifiers (t ¼ 28.828, p < 0.001). Finally, nosZ ranged from 3.5 107 to 1.3 109 copies/g dw sediment, and were significantly lower than both nirS and nirK (t ¼ 5.184, p < 0.001 and t ¼ 51.770, p < 0.001, respectively).
3.3. size
Factors determining the denitrifying community
Univariate general linear model tests (GLM) were used to analyze the effect of time (sampling period), type of sediment, sampling position in the treatment cell and their interactions, as defined factors to explain differences in gene abundances. Highly significant effects ( p < 0.001) were found when
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
5625
Fig. 2 e Time variation for the 2008e2009 period of the water flow, concentrations of BOD and COD and ammonia and nitrate at the inlet of the constructed wetland. Sediment sampling dates are indicated with arrows.
sampling period and types of sediment were considered (Table 3). The abundance of the nirS gene was the only exception and was not influenced by the presence of vegetation. Additionally, the sampling position only had an effect on 16S rRNA and narG ( p < 0.05). All possible interactions between factors had no significant effect on gene abundances, except for nitrate reductase genes in some cases. Nevertheless, partial Eta squared values (%) for these interactions were rather low, indicating a low significance in the overall variance of the considered genes. Accordingly, in subsequent analyses only the sampling time and the type of sediment were considered. In all cases, the lowest abundance of all analyzed genes was found in samples obtained during August 2008, except for
the nitrate reductase gene narG (Fig. 3AeC). In addition, the presence of vegetation, especially of Phragmites, yielded a significant increase for all genes but napA and nirS according to a non-parametric Welch test (F > 3.671, p < 0.05) (Fig. 3DeF). Differences in the abundance of napA could only be assessed by a post hoc analysis and resulted in higher values in sediments also covered with Phragmites. The presence of Typha did not cause any clear influence on the gene abundances and, depending on the genetic marker considered, was similar to either the bulk sediment or the one with Phragmites. In order to analyze differences in the gene abundances according to physicochemical characteristics of the sediment and water, pair-wise correlation tests were performed. nosZ abundance, being indicative of capacity for N2O reduction,
5626
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
Table 1 e Chemical properties of the sediment and overlying water of the different sampling locations and periods. Data from sediments correspond to mean values and standard deviations of three replicate samples. Type of sediment
Date
Water characteristics
Temp ( C) Cond (mS/cm) O2 (mg/l) pH
Sediment properties
NeNO 3
(mg/l)
NeNHþ 4
(mg/l)
pH
TC (%)
TKN (%)
SE inlet TY inlet PH inlet
15-May-08 21-May-08 20-May-08
20.7 20.4 20.0
3.8 3.6 3.5
6.1 7.6 8.8
7.6 8.0 8.0
0.45 1.07 1.24
0.37 0.21 0.21
7.4 0.1 4.0 0.2 0.21 0.02 7.7 0.1 5.6 1.1 0.30 0.06 7.4 0.1 13.5 1.9 1.23 0.08
SE inlet SE outlet TY inlet PH inlet PH outlet
20-Aug-08 11-Aug-08 21-Aug-08 18-Aug-08 12-Aug-08
24.8 25.9 25.3 24.9 26.6
2.7 2.8 2.8 3.1 2.9
4.3 7.8 5.4 5.2 3.2
7.7 8.0 7.9 7.9 8.1
0.08 0.01 0.12 0.21 0.01
0.21 0.08 0.27 0.12 0.06
7.7 0.0 7.7 0.1 7.9 0.2 7.9 0.0 7.8 0.1
3.8 0.4 3.3 0.4 4.2 0.7 4.3 0.3 5.3 0.2
0.16 0.09 0.18 0.23 0.32
0.02 0.02 0.10 0.04 0.07
SE inlet SE outlet TY inlet TY outlet PH inlet PH outlet
9-Mar-09 3-Mar-09 11-Mar-09 4-Mar-09 10-Mar-09 5-Mar-09
12.6 12.2 13.3 11.1 14.0 10.0
7.1 7.8 7.2 7.8 7.2 7.6
6.1 7.8 5.5 6.8 5.6 5.7
8.4 8.3 8.1 8.3 8.3 8.3
0.26 0.40 0.11 0.33 0.19 0.57
0.19 0.25 0.17 0.27 0.11 0.22
6.9 0.1 7.2 0.1 7.0 0.1 7.0 0.1 7.4 0.1 7.3 0.4
4.2 0.2 3.9 0.2 4.9 0.4 4.3 1.4 9.8 0.8 8.7 0.2
0.07 0.13 0.16 0.16 0.90 0.54
0.04 0.05 0.08 0.16 0.11 0.22
SE inlet SE outlet TY inlet TY outlet PH inlet PH outlet
8-Jul-09 7-Jul-09 13-Jul-09 14-Jul-09 10-Jul-09 16-Jul-09
22.6 25.5 26.4 26.1 22.4 27.6
4.5 4.7 4.6 4.7 4.4 4.5
3.6 3.9 14.9 10.2 3.0 3.8
7.3 7.7 8.7 8.3 7.7 8.2
0.18 0.08 0.03 0.03 0.03 0.03
0.32 0.49 0.28 0.15 0.25 0.11
7.7 0.2 5.7 1.0 7.2 0.1 3.8 0.1 7.6 0.3 5.0 1.3 7.2 0.1 4.8 0.9 7.1 0.2 12.0 2.9 7.5 0.2 9.0 0.7
0.47 0.18 0.29 0.29 1.04 0.83
0.19 0.04 0.22 0.18 0.67 0.21
SE, bulk sediment; TY, areas planted with Typha latifolia; PH, areas planted with Phragmites australis.
correlated positively with sediment TC (r ¼ 0.661, p < 0.01), TKN (r ¼ 0.612, p < 0.01) and negatively with the water flow and COD (r ¼ 0.330, p < 0.05; r ¼ 0.277, p < 0.05, respectively; Table 2). For the other genes, both sediment carbon and nitrogen content also correlated positively and significantly with abundances and negative correlations were observed when the water flow was considered, except for narG. The nitrate content in the water was also an important factor that specifically correlated with nitrite reductase genes, nirS and nirK, and the nitrate reductase napA. Surprisingly, no significant correlations were found for any of the genes and BOD in the influent water or sediment pH, with the exception of water pH and napA.
3.4.
Relative abundance of denitrification genes
The variation in the relative proportion of genes coding for enzymes catalyzing different steps in the denitrification can be used to assess removal efficiencies in terms of potential accumulation of intermediates in the denitrification pathway. Variation of calculated ratios according to time (sampling period), type of sediment, sampling position in the treatment cell and the corresponding interactions was estimated by GLM (Table 4). The relative contribution of the functional genes (narG, napA, nirS, nirK and nosZ ) to the total bacteria population, represented by 16S rRNA did not vary according to the presence of vegetation ( p > 0.05). When sampling period was considered, significant variations could be found for the relative amount of the functional genes narG, napA and nosZ ( p < 0.05) (results not shown). Interesting results can be deduced when ratios between two functional genes are calculated. The ratio between nitrate and nitrite reductases (qnarG þ qnapA)/(qnirS þ qnirK ) varied from 0.2 to 5.4 and was
significantly higher in samples obtained in August 2008 independently of the presence of vegetation or not ( p < 0.05). TC and TKN in the sediment were negatively correlated with (qnarG þ qnapA)/(qnirS þ qnirK ) (r ¼ 0.366, p < 0.01 and r ¼ 0.261, p < 0.05, respectively). The ratio (qnirS þ qnirK )/ qnosZ was significantly influenced by the sampling period ( p < 0.05) and the interaction between sampling period and type of vegetation. The highest difference between the abundance of nitrite reductases (qnirS þ qnirK ) and nitrous oxide reductase (qnosZ ) was found in March and May, when higher nitrate concentrations were measured, and accounted for almost two orders of magnitude. When qnirK/qnosZ and qnirS/ qnosZ ratios were analyzed separately, a significant effect of sampling time was found for both of them ( p < 0.05), but the interaction between sampling time and vegetation was only found significant for the former ( p < 0.05). Relevant variables affecting the ratio between nitrite reducers and nitrous oxide reducers were temperature (r ¼ 0.332, p < 0.05), nitrate content in water (r ¼ 0.314, p < 0.05) and total carbon in sediment (r ¼ 0.309, p < 0.05). Finally, the (qnarG þ qnapA)/ qnosZ ratio was fairly stable between the four sampling periods ( p > 0.05) and no effect of the a priori defined factors was detected.
4.
Discussion
The location of samples was a minor factor related to the abundance of denitrifying genes in the sediment of the treatment wetland. Instead, sampling period and type of sediment, mainly differing by nutrient availability, seems to be more important factors controlling the total abundance of denitrification genes in this system. Nevertheless, data from
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
5627
Fig. 3 e Mean values of the number of copies of the bacterial 16S rRNA and the functional genes, narG, napA, nirS, nirK and nosZ according to sampling period (AeC) and the type of sediment (DeF). Standard errors of the mean are indicated. Different letters above the bars indicate significant differences ( p < 0.05) between sampling period (left) or type of sediment (right).
qPCR must be examined carefully since bacteria may have more than one copy of the same gene per genome. Bacterial genomes can harbor up to 13 copies of the 16S rRNA gene (Fogel et al., 1999). The denitrification genes most often only exist in one copy, although two or three copies have been shown for all denitrification genes (Philippot, 2002; Jones et al., 2008, 2011). Despite these limitations, qPCR studies provide a realistic quantification of the size of the denitrification gene pool and ratios between gene pools can be used to infer community dynamics among samples of similar characteristics (Henry et al., 2006; Kandeler et al., 2006; Geets et al., 2007; Enwall et al., 2010). The relative abundance of functional gene densities in relation to the bacterial 16S rRNA gene abundance indicated
that the denitrifying and nitrate respiring bacteria in the sediment constituted a fraction of 1.6 0.4 and 1.2 0.7% of the bacterial communities, respectively. The values we obtained in the sediment of Empuriabrava FWS-CWs compared well with those found in the only wetland sediment characterized in terms of denitrifier abundances so far (Chon et al., 2011) and studies performed in soils (e.g. Henry et al., 2006; Kandeler et al., 2006; Hallin et al., 2009). The strong correlation between narG and napA genes and the fact that both have similar ratios to the 16S rRNA gene, suggests a high proportion of Proteobacteria in the sediment, since both narG and napA gene can be found in members of this phylum (Philippot and Hojberg, 1999; Roussel-Delif et al., 2005; Bru et al., 2007). Accordingly, quantitative analyses of bacteria at
5628
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
Table 2 e Spearman correlation coefficients (r) between environmental variables in the wetland sediment and water, influent water and abundances of the 16S rRNA and functional genes.
Wetland characteristics
Water Temp ( C) Cond (mS/cm) O2 (mg/l) RedOx (mV) pH NeNO 3 (mg N/l) NeNHþ 4 (mg N/l) Sediment TC (%) TKN (%) C:N pH
Influent characteristics
Water flow (m3/day) COD (mg O2/l) BOD (mg O2/l) NO 3 (mg N/l) NHþ 4 (mg N/l)
q16S rRNA
qnarG
qnapA
qnirS
qnirK
qnosZ
0.186 ns 0.356 ** 0.167 ns 0.114 ns 0.141 ns 0.308 * 0.179 ns
0.076 ns 0.174 ns 0.070 ns 0.043 ns 0.071 ns 0.205 ns 0.177 ns
0.282 * 0.407 ** 0.134 ns 0.002 ns 0.281 * 0.325 * 0.240 ns
0.230 ns 0.241 ns 0.102 ns 0.210 ns 0.073 ns 0.366 ** 0.200 ns
0.248 ns 0.387 ** 0.086 ns 0.124 ns 0.182 ns 0.351 ** 0.168 ns
0.139 ns 0.341 ** 0.113 ns 0.098 ns 0.195 ns 0.224 ns 0.155 ns
0.712 ** 0.653 ** 0.582 ** 0.251 ns 0.421 ** 0.376 ** 0.047 ns 0.243 ns 0.168 ns
0.544 ** 0.573 ** 0.546 ** 0.062 ns 0.160 ns 0.134 ns 0.078 ns 0.074 ns 0.120 ns
0.428 ** 0.395 ** 0.34 ** 0.219 ns 0.361 ** 0.254 ns 0.112 ns 0.285 * 0.235 ns
0.613 ** 0.539 ** 0.468 ** 0.143 ns 0.384 ** 0.346 ** 0.099 ns 0.282 * 0.123 ns
0.658 ** 0.600 ** 0.526 ** 0.234 ns 0.443 ** 0.356 ** 0.107 ns 0.310 * 0.230 ns
0.661 ** 0.612 ** 0.561 ** 0.225 ns 0.330 * 0.227 * 0.11 ns 0.182 ns 0.116 ns
ns, not significant; *p < 0.05; **p < 0.01.
the inlet and outlet of WWTPs in other studies show a clear dominance of Proteobacteria (Juretschko et al., 2002; Chouari et al., 2010; McLellan et al., 2010). The denitrifier community, assessed by the quantification of nirS and nirK genes, was dominated by NirK-type denitrifiers, which were up to 2 logs more abundant than the NirStype. This contrasts with previous studies in the same wetland using a PCR-TRFLP approach in which positive nirK PCR amplifications could not be obtained for any of the sediment samples (Ruiz-Rueda et al., 2007). This discrepancy is likely due to significant changes in the quality of the water entering the Empuriabrava FWS-CWs due to reconstructions in June 2007 (Garcı´a-Lledo´ et al., 2011). Even though the two nitrite reductases are functionally equivalent, denitrifiers harboring either nitrite reductase seem to show a preference for certain environments and are likely not under the same community assembly rules (Jones and Hallin, 2010). Recent studies in agricultural soils also suggested that the existence of the two types of nitrite reductase is due to differential niche preferences (Philippot et al., 2009; Enwall et al., 2010). However, the stability of the present physicochemical conditions of the influent to the Empuriabrava FWS-CWs and in the sediments prevents us from finding the environmental drivers
affecting the qnirS/qnirK ratio in this environment, other than that the sediment and water provided conditions that favor the NirK-type denitrifiers. In agreement with our results, the nosZ abundance in soils is often lower than that of other denitrifying genes (Henry et al., 2006; Hallin et al., 2009; Philippot et al., 2009). The nosZ gene encodes the final step of denitrification, which makes the net nitrogen removal from CWs all the way to N2 possible, but denitrifier communities with low nosZ to nir gene ratios can result in increased N2O/(N2 þ N2O) end-product ratios (Philippot et al., 2011). The lack of nitrous oxide reductase in some denitrifying bacteria was reported a decade ago when the Agrobacterium tumefaciens genome was sequenced (Wood et al., 2001) and recently it has been shown that approximately 1/3 of genome sequenced denitrifying bacterial isolates have this truncated pathway (Jones et al., 2008). The relatively low abundance of nosZ genes in the treatment wetland was especially pronounced in August 2008, coinciding with low nitrate levels and an increase in the nitrogen load in the form of ammonia to the FWS-CWs. In contrast to the other periods, the ammonium was also decreasing between the inlet and outlet of the different vegetation plots (Table 1). At the same time, the relative abundance of the
5629
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
Table 3 e Results of general linear model (GLM) analyses for the effect of the three factors established (sampling period, type of sediment and sampling location) and their interactions with the abundances of 16S rRNA and functional genes. Time
q16S rRNA qnarG qnapA qnirS qnirK qnosZ
Sed
F
p
Eta
F
p
Eta
F
p
Eta
23.21 4.27 6.76 11.13 14.42 10.48
*** * ** *** *** ***
0.65 0.25 0.35 0.47 0.53 0.45
16.49 11.93 10.04 2.87 10.96 14.52
*** *** *** ns *** ***
0.46 0.39 0.35 0.13 0.37 0.43
6.01 6.77 3.41 2.25 3.49 3.18
* * ns ns ns ns
0.14 0.15 0.08 0.06 0.08 0.08
Time Sed
q16S rRNA qnarG qnapA qnirS qnirK qnosZ
Loc
Time Loc
Sed Loc
Time Sed Loc
F
p
Eta
F
p
Eta
F
p
Eta
F
p
1.67 2.71 1.38 1.03 2.23 1.06
ns * ns ns ns ns
0.21 0.30 0.18 0.14 0.26 0.14
1.69 2.15 3.92 0.31 1.14 1.14
ns ns * ns ns ns
0.08 0.10 0.17 0.02 0.06 0.06
1.65 1.08 0.88 0.63 1.26 0.56
ns ns ns ns ns ns
0.08 0.05 0.04 0.03 0.06 0.03
1.11 2.69 3.00 0.84 0.27 2.54
Eta
ns ns * ns ns ns
0.08 0.18 0.19 0.06 0.02 0.17
ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 Time, sampling period; Sed, type of sediment; Loc, sample location in the treatment cell. Eta, partial Eta Squared.
nitrate reductases, assessed by narG and napA genes, was higher compared to the other sampling occasions. These results suggest that the conditions favored bacteria capable of dissimilatory nitrate reduction to ammonium in addition to denitrifiers that do not have the complete denitrification pathway. Higher potential rates of nitrate reduction to ammonia were measured during August 2008 in bulk sediments and areas covered with P. australis thus confirming the previous results (Garcı´a-Lledo´ et al., 2011). Conditions that invariably will affect denitrification activity, such as temperature and lower nitrate to carbon ratios varied during the studied period. Relatively high nitrate loads to the Empuriabrava FWS-CW occurred intermittently
but frequently throughout the year. Although only two of these periods, May 2008 and March 2009, were sampled in this study, the increase in the measured gene abundances let us hypothesize that the probability of N2O production in the Empuriabrava FWS-CWs would increase in relation to the nitrate concentration. However, additional measurements in summer coinciding with the increase of nitrate at the inlet would be needed to confirm this hypothesis. Other studies conducted on CWs mesocosms have reported significantly higher N2O emissions during summer (Sovik and Klove, 2007). In the Empuriabrava FWS-CWs significant differences of the genetic potential for N2O emissions according to plant species were only detected when interactions with sampling
Table 4 e Results of general linear model (GLM) analyses for the effect of the three factors established (sampling period, type of sediment and sampling location) and their interactions with the ratios between the functional genes involved in denitrification process. Time
(qnarG þ qnapA)/(qnirS þ qnirK ) (qnirS þ qnirK )/qnosZ qnirS/qnirK qnirS/qnosZ qnirK/qnosZ (qnarG þ qnapA)/qnosZ
Sed
F
p
Eta
F
p
Eta
F
p
Eta
3.64 3.81 4.16 4.14 3.76 1.17
* * * * * ns
0.23 0.24 0.25 0.25 0.23 0.09
1.06 0.62 4.79 1.57 0.73 0.31
ns ns * ns ns ns
0.05 0.03 0.21 0.08 0.04 0.17
0.15 0.27 0.00 0.18 0.29 1.08
ns ns ns ns ns ns
0.00 0.01 0.00 0.01 0.01 0.03
Time Sed
(qnarG þ qnapA)/(qnirS þ qnirK ) (qnirS þ qnirK)/qnosZ qnirS/qnirK qnirS/qnosZ qnirK/qnosZ (qnarG þ qnapA)/qnosZ
Loc
Time Loc
Sed Loc
Time Sed Loc
F
p
Eta
F
p
Eta
F
p
Eta
F
1.91 2.88 0.88 1.55 2.91 1.29
ns * ns ns * ns
0.24 0.32 0.13 0.20 0.32 0.17
1.12 0.11 1.85 0.34 0.17 2.82
ns ns ns ns ns ns
0.06 0.01 0.09 0.02 0.01 0.13
1.15 2.09 0.29 0.68 2.17 0.27
ns ns ns ns ns ns
0.06 0.10 0.02 0.04 0.11 0.01
1.74 1.68 2.22 0.50 1.83 0.29
p ns ns ns ns ns ns
Eta 0.12 0.12 0.15 0.04 0.13 0.02
ns, not significant; *, p < 0.05; Time, sampling period; Sed, type of sediment; Loc, sample location in the treatment cell. Eta, partial Eta Squared.
5630
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
time were considered. However, according to the calculated (qnirS þ qnirK )/qnosZ ratio, the highest genetic potential would be found mainly in vegetated sediments where higher nitrogen availability occurred due to leaf litter decomposition. Others have measured higher N2O fluxes in mesocosms planted with P. australis when compared to other macrophytes and unvegetated sediments (Maltais-Landry et al., 2009). Nevertheless, the effect of planted areas for the production of N2O in wetlands has been poorly studied, and further research in this direction is still needed to fully understand the dynamics of constructed wetlands in relation to greenhouse gas emissions.
5.
Conclusions
The Empuriabrava FWS-CWs showed variations in physicochemical parameters analyzed both in water and sediment between the different periods. These conditions favored the maintenance of a denitrifying community that significantly changed according to nutrient concentration and the types of vegetation. The low abundance of nosZ genes compared with the other denitrification genes is an indicative of the genetic capacity of the system to potentially accumulate the N2O intermediary and the relative proportion of nosZ genes decreased during periods of high nitrate content to the wetlands. Overall, the quantitative data on denitrification gene abundances provide evidence of a high potential for nitrous oxide emissions along the entire sediment surface and in particular during periods of high nitrate loading. The vegetation effect was mainly detected in combination with sampling time and resulted in an increase of the potential for nitrous oxide emissions in vegetated areas. The increase in the nitrite to nitrous oxide reductase genes ratio has been related to the higher total carbon content in these sediments.
Acknowledgments A.G-L and A.V-S. are recipients of pre-doctoral grants from the Ministerio de Ciencia y Educacio´n and the Universitat de Girona, respectively. The authors thank the contributions of Anna Huguet, Jordi Sala and Lluı´s Sala for field analyses. This research has been funded by the Spanish Ministerio de Ciencia y Educacio´n (grant CGL2009-08338).
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.025.
references
Abell, G.C., Revill, A.T., Smith, C., Bissett, A.P., Volkman, J.K., Robert, S.S., 2010. Archaeal ammonia oxidizers and nirS-type denitrifiers dominate sediment nitrifying and denitrifying
populations in a subtropical macrotidal estuary. The ISME Journal 4 (2), 286e300. APHA, A.a.W., 1998. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, Washington. Boudraa, A.O., Champier, J., Djebali, M., Behloul, F., Beghdadi, A., Bachand, P.A.M., Horne, A.J., 1999. Denitrification in constructed free-water surface wetlands: II. Effects of vegetation and temperature. Ecological Engineering 14, 17e32. Brix, H., 1997. Do macrophytes play a role in constructed treatment wetlands? Water Science and Technology 35, 11e17. Bru, D., Sarr, A., Philippot, L., 2007. Relative abundances of proteobacterial membrane-bound and periplasmic nitrate reductases in selected environments. Applied and Environmental Microbiology 73 (18), 5971e5974. Carter, J.P., Hsaio, Y.H., Spiro, S., Richardson, D.J., 1995. Soil and sediment bacteria capable of aerobic nitrate respiration. Applied and Environmental Microbiology 61 (8), 2852e2858. Cheneby, D., Perrez, S., Devroe, C., Hallet, S., Couton, Y., Bizouard, F., Iuretig, G., Germon, J.C., Philippot, L., 2004. Denitrifying bacteria in bulk and maize-rhizospheric soil: diversity and N2O-reducing abilities. Canadian Journal of Microbiology 50 (7), 469e474. Chon, K., Chang, J.-S., Lee, E., Lee, J., Ryu, J., Cho, J., 2011. Abundance of denitrifying genes coding for nitrate (narG), nitrite (nirS), and nitrous oxide (nosZ) reductases in estuarine versus wastewater effluent-fed constructed wetlands. Ecological Engineering 37 (1), 64e69. Chouari, R., Le Paslier, D., Daegelen, P., Dauga, C., Weissenbach, J., Sghir, A., 2010. Molecular analyses of the microbial community composition of an anoxic basin of a municipal wastewater treatment plant reveal a novel lineage of proteobacteria. Microbial Ecology 60 (2), 272e281. Coyne, M.S., Arunakumari, A., Averill, B.A., Tiedje, J.M., 1989. Immunological identification and distribution of dissimilatory heme cd1 and nonheme copper nitrite reductases in denitrifying bacteria. Applied and Environmental Microbiology 55 (11), 2924e2931. DeBusk, T., DeBusk, W., 2000. In: Kent, D.M. (Ed.), Applied Wetlands Science and Technology, second ed. CRC Press, Boca Raton, Florida (USA), pp. 241e279. Enwall, K., Throback, I.N., Stenberg, M., Soderstrom, M., Hallin, S., 2010. Soil resources influence spatial patterns of denitrifying communities at scales compatible with land management. Applied and Environmental Microbiology 76 (7), 2243e2250. EPA, 2000. Guiding Principles for Constructed Treatment Wetlands Providing for Water Quality and Wildlife Habitat. United States Environmental Protection Agency, Office of Water. Fogel, G.B., Collins, C.R., Li, J., Brunk, C.F., 1999. Prokaryotic genome size and SSU rDNA copy number: estimation of microbial relative abundance from a mixed population. Microbial Ecology 38 (2), 93e113. Francis, C.A., Beman, J.M., Kuypers, M.M., 2007. New processes and players in the nitrogen cycle: the microbial ecology of anaerobic and archaeal ammonia oxidation. The ISME Journal 1 (1), 19e27. Garcı´a-Lledo´, A., Ruiz-Rueda, O., Vilar-Sanz, A., Sala, L., Ban˜eras, L., 2011. Nitrogen removal efficiencies in a free water surface constructed wetland in relation to plant coverage. Ecological Engineering 37, 678e684. Geets, J., de Cooman, M., Wittebolle, L., Heylen, K., Vanparys, B., De Vos, P., Verstraete, W., Boon, N., 2007. Real-time PCR assay for the simultaneous quantification of nitrifying and denitrifying bacteria in activated sludge. Applied Microbiology and Biotechnology 75 (1), 211e221. Gersberg, R.M., Elkins, B.V., Lyon, S.R., Goldman, C.R., 1986. Role of aquatic plants in wastewater treatment by artificial wetlands. Water Research 20 (3), 363e368.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
Gregory, L.G., Bond, P.L., Richardson, D.J., Spiro, S., 2003. Characterization of a nitrate-respiring bacterial community using the nitrate reductase gene (narG) as a functional marker. Microbiology 149 (1), 229e237. Hallin, S., Jones, C.M., Schloter, M., Philippot, L., 2009. Relationship between N-cycling communities and ecosystem functioning in a 50-year-old fertilization experiment. The ISME Journal 3 (5), 597e605. Henry, S., Bru, D., Stres, B., Hallet, S., Philippot, L., 2006. Quantitative detection of the nosZ gene, encoding nitrous oxide reductase, and comparison of the abundances of 16S rRNA, narG, nirK, and nosZ genes in soils. Applied and Environmental Microbiology 72 (8), 5181e5189. Henry, S., Texier, S., Hallet, S., Bru, D., Dambreville, C., Cheneby, D., Bizouard, F., Germon, J.C., Philippot, L., 2008. Disentangling the rhizosphere effect on nitrate reducers and denitrifiers: insight into the role of root exudates. Environmental Microbiology 10 (11), 3082e3092. Hurt, R.A., Qiu, X., Wu, L., Roh, Y., Palumbo, A.V., Tiedje, J.M., Zhou, J., 2001. Simultaneous recovery of RNA and DNA from soils and sediments. Applied and Environmental Microbiology 67 (10), 4495e4503. Ibekwe, A.M., Lyon, S.R., Leddy, M., Jacobson-Meyers, M., 2007. Impact of plant density and microbial composition on water quality from a free water surface constructed wetland. Journal of Applied Microbiology 102 (4), 921e936. Ingersoll, T.L., Baker, L.A., 1998. Nitrate removal in wetland microcosms. Water Research 32 (3), 677e684. Johnston, C.A., 1991. Sediment and nutrient retention by freshwater wetlands: effects on surface water quality. Critical Reviews in Environmental Control 21 (5), 491e565. Jones, C.M., Hallin, S., 2010. Ecological and evolutionary factors underlying global and local assembly of denitrifier communities. The ISME Journal 4 (5), 633e641. Jones, C.M., Stres, B., Rosenquist, M., Hallin, S., 2008. Phylogenetic analysis of nitrite, nitric oxide, and nitrous oxide respiratory enzymes reveal a complex evolutionary history for denitrification. Molecular Biology and Evolution 25 (9), 1955e1966. Jones, C.M., Welsh, A., Throback, I.N., Dorsch, P., Bakken, L.R., Hallin, S., 2011. Phenotypic and genotypic heterogeneity among closely related soil-borne N2- and N2O-producing Bacillus isolates harboring the nosZ gene. FEMS Microbiology Ecology 76 (3), 541e552. Juretschko, S., Loy, A., Lehner, A., Wagner, M., 2002. The microbial community composition of a nitrifyingedenitrifying activated sludge from an industrial sewage treatment plant analyzed by the full-cycle rRNA approach. Systematic and Applied Microbiology 25 (1), 84e99. Kandeler, E., Deiglmayr, K., Tscherko, D., Bru, D., Philippot, L., 2006. Abundance of narG, nirS, nirK, and nosZ genes of denitrifying bacteria during primary successions of a glacier foreland. Applied and Environmental Microbiology 72 (9), 5957e5962. Kjellin, J., Hallin, S., Worman, A., 2007. Spatial variations in denitrification activity in wetland sediments explained by hydrology and denitrifying community structure. Water Research 41 (20), 4710e4720. Ko¨rner, S., 1999. Nitrifying and denitrifying bacteria in epiphytic communities of submerged macrophytes in a treated sewage channel. Acta hydrochimica et hydrobiologica 27 (1), 27e31. Lin, Y.F., Jing, S.R., Wang, T.W., Lee, D.Y., 2002. Effects of macrophytes and external carbon sources on nitrate removal from groundwater in constructed wetlands. Environmental Pollution 119 (3), 413e420. Maltais-Landry, G., Maranger, R., Brisson, J., Chazarenc, F., 2009. Greenhouse gas production and efficiency of planted and artificially aerated constructed wetlands. Environmental Pollution 157 (3), 748e754.
5631
McLellan, S.L., Huse, S.M., Mueller-Spitz, S.R., Andreishcheva, E.N., Sogin, M.L., 2010. Diversity and population structure of sewage-derived microorganisms in wastewater treatment plant influent. Environmental Microbiology 12 (2), 378e392. Moreno-Vivian, C., Cabello, P., Martinez-Luque, M., Blasco, R., Castillo, F., 1999. Prokaryotic nitrate reduction: molecular properties and functional distinction among bacterial nitrate reductases. Journal of Bacteriology 181 (21), 6573e6584. Nguyen, C., 2003. Rhizodeposition of organic C by plants: mechanisms and controls. Agronomie 23 (5e6), 375e396. Philippot, L., 2002. Denitrifying genes in bacterial and Archaeal genomes. Biochimica et Biophysica Acta 1577 (3), 355e376. Philippot, L., Andert, J., Jones, C.M., Bru, D., Hallin, S., 2011. Importance of denitrifiers lacking the genes encoding the nitrous oxide reductase for N2O emissions from soil. Global Change Microbiology 17, 1497e1504. Philippot, L., Cuhel, J., Saby, N.P., Cheneby, D., Chronakova, A., Bru, D., Arrouays, D., Martin-Laurent, F., Simek, M., 2009. Mapping fieldscale spatial patterns of size and activity of the denitrifier community. Environmental Microbiology 11 (6), 1518e1526. Philippot, L., Hojberg, O., 1999. Dissimilatory nitrate reductases in bacteria. Biochimica et Biophysica Acta 1446 (1e2), 1e23. Prade, K., Trolldenier, G., 1990. Denitrification in the rhizosphere of rice and wheat seedlings as influenced by plant K status, air-filled porosity and substrate organic matter. Soil Biology and Biochemistry 22 (6), 769e773. Reed, S.C., Crites, R.W., Middlebrooks, E.J., 1995. Natural Systems for Waste Management and Treatment. McGraw-Hill, New York. Roussel-Delif, L., Tarnawski, S., Hamelin, J., Philippot, L., Aragno, M., Fromin, N., 2005. Frequency and diversity of nitrate reductase genes among nitrate-dissimilating Pseudomonas in the rhizosphere of perennial grasses grown in field conditions. Microbial Ecology 49 (1), 63e72. Ruiz-Rueda, O., Hallin, S., Baneras, L., 2009. Structure and function of denitrifying and nitrifying bacterial communities in relation to the plant species in a constructed wetland. FEMS Microbiology Ecology 67 (2), 308e319. Ruiz-Rueda, O., Trias, R., Garcia-Gil, L.J., Baneras, L., 2007. Diversity of the nitrite reductase gene nirS in the sediment of a free-water surface constructed wetland. International Microbiology 10 (4), 253e260. Sovik, A.K., Klove, B., 2007. Emission of N2O and CH4 from a constructed wetland in southeastern Norway. Science of the Total Environment 380 (1e3), 28e37. Tiedje, J., 1988. Ecology of denitrification and dissimilatory nitrate reduction to ammonium. In: Zehnder, A.J.B. (Ed.), Biology of Anaerobic Microorganisms. Wiley, New York, pp. 179e244. Toet, S., Bouwman, M., Cevaal, A., Verhoeven, J.T., 2005. Nutrient removal through autumn harvest of Phragmites australis and Thypha latifolia shoots in relation to nutrient loading in a wetland system used for polishing sewage treatment plant effluent. Journal of Environmental Science and Health, Part A 40 (6), 1133e1156. Vial, L., Lavire, C., Mavingui, P., Blaha, D., Haurat, J., MoenneLoccoz, Y., Bally, R., Wisniewski-Dye, F., 2006. Phase variation and genomic architecture changes in Azospirillum. Journal of Bacteriology 188 (15), 5364e5373. Vymazal, J., 2001. Types of constructed wetlands for wastewater treatment: their potential for nutrient removal. In: Vymazal, J. (Ed.), Transformations of Nutrients in Natural and Constructed Wetlands. Backhuys Publishers, Leiden, The Netherlands, pp. 1e93. Whitney, D., Rossman, A., Hayden, N., 2003. Evaluating an existing subsurface flow constructed wetland in Akumal, Mexico. Ecological Engineering 20 (1), 105e111. Wood, D.W., Setubal, J.C., Kaul, R., Monks, D.E., Kitajima, J.P., Okura, V.K., Zhou, Y., Chen, L., Wood, G.E., Almeida Jr., N.F.,
5632
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 2 1 e5 6 3 2
Woo, L., Chen, Y., Paulsen, I.T., Eisen, J.A., Karp, P.D., Bovee Sr., D., Chapman, P., Clendenning, J., Deatherage, G., Gillet, W., Grant, C., Kutyavin, T., Levy, R., Li, M.J., McClelland, E., Palmieri, A., Raymond, C., Rouse, G., Saenphimmachak, C., Wu, Z., Romero, P., Gordon, D., Zhang, S., Yoo, H., Tao, Y., Biddle, P., Jung, M., Krespan, W., Perry, M., Gordon-Kamm, B., Liao, L., Kim, S., Hendrick, C., Zhao, Z.Y., Dolan, M., Chumley, F., Tingey, S.V., Tomb, J.F., Gordon, M.P., Olson, M.V., Nester, E.W., 2001. The
genome of the natural genetic engineer Agrobacterium tumefaciens C58. Science 294 (5550), 2317e2323. Zhu, T., Sikora, F.J., 1995. Ammonium and nitrate removal in vegetated and unvegetated gravel bed microcosm wetlands. Water Science and Technology 32 (3), 219e228. Zumft, W.G., 1997. Cell biology and molecular basis of denitrification. Microbiology and Molecular Biology Reviews 61 (4), 533e616.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Optimization of the coagulation-flocculation process for pulp mill wastewater treatment using a combination of uniform design and response surface methodology Jian-Ping Wang a,b, Yong-Zhen Chen a, Yi Wang b, Shi-Jie Yuan a, Han-Qing Yu a,* a b
Department of Chemistry, University of Science and Technology of China, Hefei, 230026, China Department of Agricultural and Biological Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA
article info
abstract
Article history:
Pulp mill wastewater was treated using the coagulation-flocculation process with
Received 7 June 2011
aluminum chloride as the coagulant and a modified natural polymer, starch-g-PAM-g-
Received in revised form
PDMC [polyacrylamide and poly (2-methacryloyloxyethyl) trimethyl ammonium chloride],
6 August 2011
as the flocculant. A novel approach with a combination of response surface methodology
Accepted 14 August 2011
(RSM) and uniform design (UD) was employed to evaluate the effects and interactions of
Available online 1 September 2011
three main influential factors, coagulant dosage, flocculant dosage and pH, on the treatment efficiency in terms of the supernatant turbidity and lignin removals as well as the
Keywords:
water recovery. The optimal conditions obtained from the compromise of the three
Coagulation-flocculation
desirable responses, supernatant turbidity removal, lignin removal and water recovery
Optimization
efficiency, were as follows: coagulant dosage of 871 mg/L, flocculant dosage of 22.3 mg/L
Pulp mill wastewater
and pH 8.35. Confirmation experiments demonstrated that such a combination of the UD
Response surface methodology
and RSM is a powerful and useful approach for optimizing the coagulation-flocculation
Uniform design
process for the pulp mill wastewater treatment. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The gross output of paper and paperboard is 79.8 million tons in 2008 in China, and about 22% of them are using straw pulp as staffs. The wastewater generated by the pulp mills is not easy to treat due to the presence of a large amount of chemicals, such as many sodium salts of organic acids. In addition, a large amount of lignin present in the wastewater, which causes colority, turbidity and high COD (chemical oxygen demand), is usually wasted and overburdens the treatment process. Thus, both the lignin reclamation and wastewater treatment are crucial. An efficient and cost-effective process for the treatment of pulp mill wastewater should be pursued. Coagulation-flocculation is a simple and efficient method for wastewater treatment, and has been widely used for the
treatment of palm oil mill effluent (Ahmad et al., 2005), textile wastewater (Meric et al., 2005) and abattoir wastewater (Amuda and Alade, 2006), etc. Recently, coagulationflocculation or flocculation processes have also been extensively used for the treatment of pulp mill wastewater. In such studies, polyaluminium chloride (PAC), chitosan, polymeric phosphate-aluminum chloride, cationic and anionic polyacrylamides (PAMs) and polydiallyldimethylammonium chloride (polyDADMAC) have all been tested as a flocculant in the flocculation process, and various levels of removal efficiency for turbidity and lignin have been achieved (Razali et al., 2011; Renault et al., 2009; Wong et al., 2006; Zheng et al., 2011). In the coagulation-flocculation process, the efficiency is governed by various factors, such as the type and dosage of coagulant/flocculant (Desjardins et al., 2002; Hu
* Corresponding author. Fax: þ86 551 3601592. E-mail address: [email protected] (H.-Q. Yu). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.023
5634
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
et al., 2005; Nandy et al., 2003; Spicer and Pratsinis, 1996; Wang et al., 2002), pH (Elmaleh et al., 1996; Miller et al., 2008; Rohrsetzer et al., 1998; Syu et al., 2003), mixing speed and time (Gurses et al., 2003; Rossini et al., 1999), temperature and retention time (Howe et al., 2006; Zhu et al., 2004). A proper optimization of these factors could significantly increase its treatment efficiency. Response surface methodology (RSM) is an efficient way to achieve such an optimization by analyzing and modeling the effects of multiple variables and their responses and finally optimizing the process. This method has been widely used for the optimization of various processes in food chemistry, material science, chemical engineering and biotechnology (Granato et al., 2010; Hong et al., 2011; Liu et al., 2010; Singh et al., 2010). In the traditional experimental design approaches used in RSM, such as central composite design, with an increase in experimental factors, the number of coefficients of the quadratic model equation increases exponentially and so does the number of experimental trials (Cheng et al., 2002). To overcome this shortcoming of RSM, uniform design (UD) can be used to investigate more factors with substantially fewer experimental trials, since it determines the number of the experimental trials only by the level of factors, rather than by the number of factors. The UD method was first proposed by Fang (1978). Compared to the traditional experimental design methods, UD is capable of selecting experimental points uniformly in the experimental region and highly representative in the experimental domain; it imposes no strong assumption on the model and may be used when the underlying model between the responses and factors is unknown or partially unknown; it can accommodates the largest possible number of levels for each factor among all experimental designs (Leung et al., 2000; Li et al., 2003; Wen et al., 2005; Zhang et al., 1998). A combination of UD and RSM would be able to achieve the optimization of a complex multivariate process with the fewest multilevel experiments. Therefore, in this study, UD and RSM were integrated to optimize the coagulation-flocculation process for pulp mill wastewater treatment. The selection of high efficient coagulants and flocculants is essential for a successful coagulation-flocculation process. In this work, on the basis of our previous studies (Wang et al., 2007, 2009), the conventional coagulant Al(OH)3 was chosen as the coagulant, while a novel modified natural polymer, starch-g-PAM-g-PDMC [polyacrylamide and poly (2-methacryloyloxyethyl) trimethyl ammonium chloride] with both strong charge neutralization and bridging abilities, was used as the flocculant. The main objective of this work was to treat the pulp mill wastewater using the coagulation-flocculation process, which was optimized using an integrative UD-RSM approach. Removal efficiencies of both supernatant turbidity and lignin, and recovery efficiency of clean water were chosen as the dependent output variables. The compromise optimal conditions for these three responses were obtained using the desirability function approach. The novel optimization strategy used for the pulp mill wastewater treatment process in this study is expected to provide valuable information for other complicated systems in environmental engineering and other fields.
2.
Materials and methods
2.1.
Chemicals and operation
Aluminum chloride, hydrochloric acid and sodium hydroxide, purchased from Shanghai Chemical Reagent Co., China, were of analytical reagent grade and used without further purification. The flocculant, starch-g-PAM-g-PDMC, was prepared as follows: starch, (2-methacryloyloxyethyl) trimethyl ammonium chloride and acrylamide were dissolved in water in Pyrex glass vessels, and were heated in a preset water bath using potassium persulphate as the initiator after deoxygenating. Thereafter, the sample solutions were precipitated in acetone and separated by filtration. Homopolymers formed in the reactions were removed using the soxhlet extraction method in ethanol. All the grafted samples were dried in a vacuum oven at 50 C until a constant weight. The grafting percentage and the cationic degree of the graft copolymer used in this experiment was 269% and 1.96 103 mol/g, respectively. The point of zero charge of the graft copolymer was measured as 7.80. Its molecular structure is as shown in Fig. 1 and its image is illustrated in Fig. S1 (Appendix). Pulp mill wastewater was blending black liquor from the primary sedimentation tank of Guoyang Paper and Pulp Mill Co., China. The stuffs of the paper were wheat straw. The initial pH, chemical oxygen demand (COD) and turbidity were 6.99, 1358 mg/L and 1209 NTU, respectively. The average sizes of the colloid particles in the wastewater were 544 nm. The coagulation-flocculation experiments were carried out using the jar test method in 1-L beakers. After the coagulant (stock solution of 50.0 g/L) was added with a dosage varying from nil to 2100 mg/L, the solution pH was adjusted to 2.5e11.5 by adding 0.1 mol/L HCl or NaOH solutions. Then, the flocculant at a concentration of 1.0 g/L was added with a dosage varying from nil to 48 mg/L. The sample was immediately stirred at a constant speed of 200 rpm for 2 min, followed by a slow stirring at 40 rpm for 10 min; thereafter, a settlement for 5 min was performed. After that, samples were taken from the water level around 2 cm underneath the surface for measuring the turbidity and lignin concentrations of the supernatant. Meanwhile, the volume of produced sludge was calculated directly from the reading on the beakers, and then the volume of the recovered clean water was calculated accordingly.
2.2.
Experimental design and data analysis
UD tables can be described as Un(qm), where U, n, q and m stand for the UD, the number of experimental trials, the number of levels and the maximum number of factors, respectively. For a given measure of uniformity M, a uniform design has the smallest M-value overall fractional factorial design with n runs and m q-level factors.
CH3 S O
CH2 C
CONH2 m
CH2
C n H
COOC2H4N(CH3)3Cl Fig. 1 e Molecular structure of the graft copolymer starch-gPAM-g-PDMC (S: Starch).
5635
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
In this work, coagulant dosage (X1), flocculant dosage (X2) and pH (X3) were chosen as three independent variables in the coagulation-flocculation process. Seven levels for each factor were selected to investigate the influence and interaction of the factors. In order to improve the accuracy, the experiment were carried out using U14(143), the range and levels of each factor, the guide for using (U14(143)) are listed in Tables 1 and 2, respectively. Efficiencies of turbidity removal, lignin removal and clean water recovery were selected as the dependent variables in order to represent the overall wastewater treatment efficiency. The response variable was fitted by a sufficient model, which is able to describe the relationship between the dependent output variable and the independent variables using the regression method. Y ¼ b0 þ
j X
bi Xi þ
i¼1
k X
bii X2i þ
i<j X X i
i¼1
bij Xi Xj
(1)
j
where Y is the response variable to be modeled; Xi and Xj are the independent variables which influence Ym; b0, bi, bii and bij are the offset terms, the ith linear coefficient, the quadratic coefficient and the ijth interaction coefficient, respectively. The actual design of this work is given in Table 3 (Fang, 1994). The parameters of the response equations and corresponding analysis on variations were evaluated using Uniform Design Software 2.1 (http://www.math.hkbu.edu.hk/ UniformDesign/software) and MATLAB 6.5, respectively. The interactive effects of the independent variables on the dependent ones were illustrated by three- and twodimensional contour plots. Finally, two additional experiments were conducted to verify the validity of the statistical experimental strategies.
3.
Results
The experimental results are listed in Table 3. The variance trend was discrepant for the three responses. Therefore, the operational conditions have been optimized respectively for different responses.
3.1.
Table 2 e Guide for selecting columns of generating vectors in U14(143). No. of factors studied 2 3 4
Table 1 e Levels of the variable tested in the U7(73) uniform designs. Variables
Range and levels 1
2
3
4
5
6
7
X1, coagulant 0 350 700 1050 1400 1750 2100 dosage (mg/L) X2, flocculant 0 8 16 24 32 40 48 dosage (mg/L) 2.5 4.0 5.5 7.0 8.5 10.0 11.5 X3, pH
Discrepancy in uniformity
1, 4 1, 2, 3 1, 2, 3, 5
0.0957 0.1455 0.2091
the regression model using the backward regression method with the experimental results: Y ¼ 29:9 þ 22:1X1 þ 11:6X2 2:5X21 0:6X22 0:7X23 1:5X1 X2 þ 1:2X1 X3 R2 ¼ 0:917; F ¼ 8:81
(2)
Statistical testing of the model was performed with the Fisher’s statistical test for analysis of variance (ANOVA). The quadratic regression shows that the model was significant because the value of Fstatistic (the ratio of mean square due to regression to mean square to real error) of 8.81 was greater than F0.001,7,6 (4.27). The value of the correlation coefficient (R2 ¼ 0.917) indicates that only 8.3% of the total variation could not be explained by the empirical model (Leung et al., 2000; Meric et al., 2005). The p-value ( p ¼ 0.05) of Eq. (2) also implies that the second-order polynomial model fitted the experimental results well. This is confirmed by Fig. 2a, in which the plots of predicted turbidity removal efficiencies versus measured ones are shown. Most points distributed near to the straight line where the measured and predicted removal efficiencies are the same, indicating that the regression model is able to predict these removal efficiencies. From Eq. (2), the optimal conditions for the supernatant turbidity removal efficiency were obtained as follows: coagulant dosage of 917 mg/L, flocculant dosage of 33.0 mg/L and pH
Table 3 e UD and response results for the study of three experimental variables in coded units. Run X1 X2 X3
Factors
Response
X1 X2 X3 Turbidity Lignin Water removal removal recovery (%) (%) efficiency (%)
Optimization for supernatant turbidity removal
The supernatant turbidity removal efficiency listed in Table 3 is an important denotation for the treatment efficiency of the coagulation-flocculation process. The following equation is
Columns to be used
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 2 3 4 5 6 7 8 9 10 11 12 13 14
4 8 12 1 5 9 13 2 6 10 14 3 7 11
7 14 6 13 5 12 4 11 3 10 2 9 1 8
Source: Fang, 1994.
1 1 2 2 3 3 4 4 5 5 6 6 7 7
2 4 6 1 3 5 7 1 3 5 7 2 4 6
4 7 3 7 3 6 2 6 2 5 1 5 1 4
63.9 56.9 95.4 54.3 92.7 92.9 95.8 87.8 92.6 98.7 71.2 95.1 64.8 67.2
36.0 34.5 71.6 24.5 80.9 70.9 74.3 47.6 76.2 82.5 40.1 59.5 45.3 32.5
66.0 50.0 60.0 62.0 68.0 60.0 48.0 68.0 50.0 66.0 50.0 64.0 42.0 64.0
5636
a
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
of the lignin removal efficiency in the coagulation-flocculation experiments:
100
Predicted (%)
90
Y ¼ 45:3 þ 35:6X1 þ 24:6X2 þ 8:2X3 4:0X21 2:4X22 1:3X23 1:1X1 X2
80
R2 ¼ 0:889; F ¼ 4:64
70 60 50 50
b
(3) 2
60
80 70 Measured (%)
90
10 0
90
The results of F ¼ 4.64 > F (0.05,7,6) ¼ 4.207 and R ¼ 0.889 for the lignin removal efficiency show that the second-order polynomial model was significant and fitted the experimental results well. Fig. 2b shows that the measured versus predicted plot values were distributed evenly near to the straight line. From Eq. (3), the optimal conditions for maximal efficiency of lignin removal were estimated to be: coagulant dosage of 1004 mg/L, flocculant dosage of 25.9 mg/L and pH 5.7. Under these conditions, the maximal lignin removal efficiency was estimated to be 84.1%.
Predicted (%)
80 70
3.3.
40
The regression model to describe the water recovery efficiency of the coagulation-flocculation experiments was obtained using the total regression method (Eq. (4)).
30
Y ¼ 15:3 4:0X1 þ 24:4X2 þ 22:4X3 0:4X21 2:4X22 1:8X23
50
þ1:0X1 X2 þ 0:8X1 X3 2:4X2 X3
20 20
c
30
40
50 60 Measured (%)
70
80
90
70 60
Predicted (%)
Optimization for water recovery efficiency
60
50
R2 ¼ 0:935; F ¼ 6:35
(4)
The regression results, i.e., F ¼ 6.35 > F (0.05,9,4) ¼ 5.999 and R2 ¼ 0.935, show that the second-order polynomial model was significant and fitted the experimental results for the water recovery efficiency well. Again, the experimental values were distributed near to the straight line (Fig. 2c). From Eq. (4), the optimal conditions for maximal water recovery efficiency were estimated to be: coagulant dosage of 1040 mg/L, flocculant dosage of 20.6 mg/L and pH 8.17, under which the maximal water recovery efficiency was estimated to be 73.4%.
40 3.4.
Confirmation experimental results
30 30
40
50 60 Measured (%)
70
Fig. 2 e Relationship between the predicted and measured (a) turbidity removal efficiency; (b) lignin removal efficiency; and (c) water recovery efficiency.
of 5.67. Under the optimal conditions, the maximal turbidity removal efficiency was estimated to be 99.7%.
To confirm the validity of the statistical experimental strategies, additional confirmation experiments were conducted in duplicates. The chosen conditions for the coagulant dosage, flocculant dosage and pH are all listed in Table 4, along with the predicted and measured results. As shown in Table 4, the measured efficiencies of the supernatant turbidity removal, lignin removal and clean water recovery were close to the predicted values using their respective regression models. This demonstrates that the UD-RSM approach was appropriate for optimizing the operational conditions of the coagulation-flocculation process.
3.5. 3.2.
Multiple-response optimization
Optimization for lignin removal
Lignin is the main pollutant in the pulp mill wastewater. The efficient removal of lignin from wastewater by the coagulation-flocculation method is crucial to reclaim water and lignin. The following equation (Eq. (3)) is the regression model using the backward regression method with the results
Removal efficiency of supernatant turbidity, removal efficiency of lignin and water recovery efficiency are three individual responses, and their optimizations were achieved under different optimal conditions. Thus, a compromise among the conditions for the three responses is desirable. The desirability function approach was used to achieve such a goal
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
Table 4 e Measured and calculated values for the confirmation experiments. Run 15
16
17
Conditions Coagulant dosage: 917 mg/L Flocculant dosage: 33.0 mg/L pH: 5.67 Coagulant dosage: 1004 mg/L Flocculant dosage: 25.9 mg/L pH: 5.7 Coagulant dosage: 1040 mg/L Flocculant dosage: 20.6 mg/L pH: 8.17
Parameter Measured Calculated Turbidity removal (%)
99.6 0.1
99.7
Lignin removal (%)
88.4 0.3
84.1
Water recovery (%)
74.0 2.0
73.4
(Zhang et al., 1998). The regression equation of the compromise was obtained as follows: Y ¼ 14:401 þ 0:0918X1 þ 0:923X2 þ 11:084X3
0:0000395X21
0:0463X22 0:818X23 þ 0:000401X1 X2 0:00759X1 X3 þ 0:156X2 X3
(5)
The optimal conditions calculated from the regression equation were as follows: coagulant dosage of 871 mg/L, flocculant dosage of 22.3 mg/L and pH 8.35, respectively. The corresponding removal efficiency of turbidity, removal efficiency of lignin and water recovery efficiency were 95.7%, 83.4% and 72.7%, respectively. The overlay plot for the optimal region is presented in Fig. 3. The shaded portion gave the permissible values of the two variables by defining the desired limits of removal efficiency of supernatant turbidity, removal efficiency of lignin and water recovery efficiency. A confirmation experiment under the compromised conditions was carried out in triplicates, and the average removal efficiencies of turbidity and lignin, and water recovery efficiency were obtained as 95.0%, 83.5% and 72.0%, respectively (Fig. S4 in Appendix). These results were in good
Fig. 3 e Overlay plot for the optimal region.
5637
agreement with the model predictions. After flocculation, the aluminum ion concentration in the supernatant was measured as 6.8 mg/L, and the final pH was 6.2. An optimal pH of 8.35 was predicted for multi-response optimization purpose, at which the optimal compromised removal efficiencies were obtained. However, it is worthwhile noting that, decent efficiencies can be obtained at neutral pH (the initial pH of the wastewater) based on our experiment results and the prediction by Eq. (5). Thus, in the practical application, this wastewater could be treated without pH adjustment when the treatment efficiency is fulfilled and the cost for pH adjustment is a big concern. This is a tradeoff between the efficiency of the process and the cost for pH adjustment.
4.
Discussion
4.1.
Supernatant turbidity removal
With the turbidity removal efficiency as the response, the response surfaces of the quadratic model with one variable kept at the optimal level and the other two varying within the experimental ranges are shown in Fig. 4. The obvious peak in the response surfaces indicates that the optimal conditions were exactly located inside the design boundary. In other words, there were significant interactive effects on turbidity between coagulant dosage and the flocculant dosage, coagulant dosage and pH, as well as flocculant dosage and pH. The turbidity removal efficiency was high when the coagulant dosage and the flocculant dosage were within the range of 700e1400 mg/L and 16e48 mg/L, respectively, at the optimal pH 5.3 (Fig. 4b). Charge neutralization and sweep-floc were the two main mechanisms leading to the aggregation of particles in the coagulation process. In general, the appropriate pH for the charge neutralization in the coagulation process is in a range of 4.0e5.5 (Chang et al., 1993). When the aluminum ions is used as a coagulant, pH 6.0e8.0 is suitable for the formation of amorphous Al(OH)3, which removes organic matters by adsorption on the precipitation of Al(OH)3(s) through the sweep-floc mechanism (Chang et al., 1993). Thus, pH 5.3 is favorable only for the charge neutralization. However, the acidic condition (pH 5.3) was in favor of the improvement of cationic charge density as well as the extension of the grafting chain in the solution. In this case, both the charge neutralization ability and the sweep-floc ability were improved. Taking into account the two factors, pH 5.3 was appropriate for the turbidity removal of the pulp mill wastewater. Likewise, at the optimal flocculant dosage, the coagulant dosage was within the range of 350e1050 mg/L and pH was 2.0e8.5 (Fig. 4b). Under these conditions, the coagulant AlCl3 exhibited good charge neutralization and sweep-floc abilities, and thus the flocculant, in addition to the two abilities above, had an improved adsorption bridging ability. In addition, at the optimal coagulant dosage, the acidic condition and flocculant dosage within the range of 24e48 mg/L were favorable for the supernatant turbidity removal. As mentioned above, for a given coagulant dosage, the acidic condition was appropriate for the removal of turbidity with the graft copolymer used as the flocculant.
5638
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
Under the experimental conditions, the lignin was indiscerptible and suspended in the wastewater as colloids (about 500 nm). Thus, the coagulation-flocculation method was appropriate for lignin removal. Acidic condition was in favor of the precipitation of lignin and thus its removal. However, the lignin removal efficiency was less than the turbidity removal efficiency under their respective optimal conditions. This is attributed to the fact that there is a great amount of soluble phenols in the wastewater and phenols could not be efficiently removed by the coagulation-flocculation method.
4.3.
Water recovery efficiency
The response surfaces of the quadratic model with one variable kept at the optimal level and the other two varying within the experimental ranges, with the water recovery efficiency as the response, are shown in Fig. S3 (Appendix). The elliptical contour plots indicate that there were significant interactive effects between coagulant dosage and flocculant dosage, coagulant dosage and pH, as well as flocculant dosage and pH, even the water recovery efficiency percentage increased at the center of the three regions. This was evidenced by the obvious peak in the response surfaces, in which the optimal conditions were exactly located inside the design boundary. In the coagulation-flocculation process, the coagulant was dispersed in the wastewater to destabilize the colloidal particles and the flocculant was used to agglomerate the destabilized colloidal particles into large particles and then precipitates. The flocculant used in our experiment was a modified natural polymer, which was synthesized by grafting two monomers onto starch backbone in order to improve its charge neutralization and bridging ability (Wang et al., 2009). The flexible grafting chain grafted onto the rigid starch backbone increased the chances for the flocculant to approach to the contaminant particles in the wastewater. Therefore, the graft copolymer used as the flocculant would conduce to the formation of larger and denser flocs, which are readily separated from the wastewater. As a result, better quality and higher water quantity could be obtained.
4.4. Fig. 4 e 3D surface graphs and contour plots of turbidity removal efficiency showing the effect of variables: (a) X1eX2; (b) X1eX3; and (c) X2eX3.
4.2.
Lignin removal
When the lignin removal efficiency was selected as the response, the response surfaces of the quadratic model with one variable kept at the optimal level and the other two varying within the experimental ranges are illustrated in Fig. S2 (Appendix). The peak in the response surfaces indicates that the optimal conditions were exactly located inside the design boundary. Furthermore, there were significant interactive effects on turbidity between coagulant dosage and flocculant dosage, coagulant dosage and pH, as well as flocculant dosage and pH.
Significance of the integrated UD-RSM approach
A combination of UD and RSM applied in this work has a rational statistical basis, and is demonstrated to be a powerful approach for the optimization of the coagulationflocculation process for pulp mill wastewater treatment in this study. With the UD method, the selected experimental points were distributed uniformly in the factor space for all the three key factors influencing the efficiency of this process, i.e., coagulant dosage, flocculant dosage and pH. This facilitated the acquisition of most response information through the fewest numbers of experiments. In addition, the application of number theory in the experimental design facilitates the computer statistical modeling and the subsequent regression analysis in the UD software, such as linear regression, non-linear regression and quadratic regression, etc. In addition, the number of the experimental trials in UD is determined only by the level of factors, not by the number of factors. The UD method can also be used when the levels of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
factors are different. These are the advantages of UD over other experiment design approaches. A combination of UD and RSM provided a straightforward way to evaluate the individual effects and interactions of the experimental factors for desirable responses. Following the RSM optimization, a desirability function approach could be employed to obtain the compromise optimal conditions. This study demonstrates that this integrated approach could optimize the coagulation-flocculation process effectively using few data sets, which becomes very attractive for the processes where data are obtained costly. Thus, this integrated optimization approach can be useful for other complex wastewater treatment processes and multivariate systems in other fields.
5.
Conclusions
A coagulation-flocculation process with aluminum chloride as the coagulant and a modified natural polymer, starch-g-PAMg-PDMC [polyacrylamide and poly (2-methacryloyloxyethyl) trimethyl ammonium chloride], as the flocculant was employed for pulp mill wastewater treatment. A novel approach combined response surface methodology (RSM) and uniform design (UD) was used to optimize the process and evaluate the effects and interactions of three main influential factors, coagulant dosage, flocculant dosage and pH, on the treatment efficiency in terms of the supernatant turbidity and lignin removals as well as the water recovery. An optimal condition of coagulant dosage 871 mg/L, flocculant dosage 22.3 mg/L and pH 8.35 was obtained from the compromise of the three desirable responses, i.e. supernatant turbidity removal, lignin removal and water recovery efficiency. Further confirmation experiments demonstrated that such a combination of the UD and RSM is an effective and powerful approach for the optimization of the coagulation-flocculation process for pulp mill wastewater treatment.
Acknowledgments We wish to thank the Key Special Program on the S&T for the Pollution Control (2008ZX07103-001 and 2008ZX07010-003) for the partial support of this study.
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.023.
references
Ahmad, A.L., Ismail, S., Bhatia, S., 2005. Optimization of coagulation-flocculation process for palm oil mill effluent using response surface methodology. Environmental Science & Technology 39 (8), 2828e2834.
5639
Amuda, O.S., Alade, A., 2006. Coagulation/flocculation process in the treatment of abattoir wastewater. Desalination 196 (1e3), 22e31. Chang, Q., Fu, J.Y., Li, Z.L., 1993. Principles of Flocculation. Lanzhou University Press, Lanzhou, China. Cheng, B.J., Zhu, N., Fan, R.L., Zhou, C.X., Zhang, G.N., Li, W.K., Ji, K.J., 2002. Computer aided optimum design of rubber recipe using uniform design. Polymer Testing 21 (1), 83e88. Desjardins, C., Koudjonou, B., Desjardins, R., 2002. Laboratory study of ballasted flocculation. Water Research 36 (3), 744e754. Elmaleh, S., Yahi, H., Coma, J., 1996. Suspended solids abatement by pH increase-upgrading of an oxidation pond effluent. Water Research 30 (10), 2357e2362. Fang, K.T., 1978. Probability and Statistics Bulletin. Institute of Mathematics, Academia Sinica, Beijing, China. Fang, K.T., 1994. Uniform Design and Uniform Tables. Science Publication Co, Beijing, China. Granato, D., Ribeiro, J.C.B., Castro, I.A., Masson, M.L., 2010. Sensory evaluation and physicochemical optimisation of soybased desserts using response surface methodology. Food Chemistry 121 (3), 899e906. Gurses, A., Yalcin, M., Dogar, C., 2003. Removal of Remazol Red RB by using Al (III) as coagulant-flocculant: effect of some variables on settling velocity. Water Air and Soil Pollution 146 (1e4), 297e318. Hong, F.L., Peng, J.C., Lui, W.B., 2011. Optimization of the process variables for the synthesis of starch-based bioadable resin using response surface methodology. Journal of Applied Polymer Science 119 (3), 1797e1804. Howe, K.J., Marwah, A., Chiu, K.P., Adham, S.S., 2006. Effect of coagulation on the size of MF and UF membrane foulants. Environmental Science & Technology 40 (24), 7908e7913. Hu, C., Liu, H., Qu, J., Wang, D., Ru, J., 2005. Coagulation behavior of aluminum salts in eutrophic water: significance of Al13 species and pH control. Environmental Science & Technology 40 (1), 325e331. Leung, S.Y.L., Chan, W.H., Luk, C.H., 2000. Optimization of fabrication conditions of high-efficiency ultrafiltration membranes using methods of uniform design and regression analysis. Chemometrics and Intelligent Laboratory Systems 53 (1e2), 21e35. Li, J.F., Liao, H., Normand, B., Cordier, C., Maurin, G., Foct, J., Coddet, C., 2003. Uniform design method for optimization of process parameters of plasma sprayed TiN coatings. Surface & Coatings Technology 176 (1), 1e13. Liu, B.B., Yang, M.H., Qi, B.K., Chen, X.R., Su, Z.G., Wan, Y.H., 2010. Optimizing L-(þ)-lactic acid production by thermophile Lactobacillus plantarum As. 1.3 using alternative nitrogen sources with response surface method. Biochemical Engineering Journal 52 (2e3), 212e219. Meric, S., Selcuk, H., Belgiorno, V., 2005. Acute toxicity removal in textile finishing wastewater by Fenton’s oxidation, ozone and coagulation-flocculation processes. Water Research 39 (6), 1147e1153. Miller, S.M., Fugate, E.J., Craver, V.O., Smith, J.A., Zimmerman, J.B., 2008. Toward understanding the efficacy and mechanism of Opuntia spp. as a natural coagulant for potential application in water treatment. Environmental Science & Technology 42 (12), 4274e4279. Nandy, T., Shastry, S., Pathe, P.P., Kaul, S.N., 2003. Pre-treatment of currency printing ink wastewater through coagulationflocculation process. Water Air and Soil Pollution 148 (1e4), 15e30. Razali, M.A.A., Ahmad, Z., Ahmad, M.S.B., Ariffin, A., 2011. Treatment of pulp and paper mill wastewater with various molecular weight of polyDADMAC induced flocculation. Chemical Engineering Journal 166 (2), 529e535.
5640
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 3 3 e5 6 4 0
Renault, F., Sancey, B., Charles, J., Morin-Crini, N., Badot, P.M., Winterton, P., Crini, G., 2009. Chitosan flocculation of cardboard-mill secondary biological wastewater. Chemical Engineering Journal 155 (3), 775e783. Rohrsetzer, S., Paszli, I., Csempesz, F., 1998. Colloid stability of electrostatically stabilized sols. Part III: the role of pH in hydration coagulation and peptization of SiO2- and Al2O3-sols. Colloid and Polymer Science 276 (3), 260e266. Rossini, M., Garrido, J.G., Galluzzo, M., 1999. Optimization of the coagulation-flocculation treatment: influence of rapid mix parameters. Water Research 33 (8), 1817e1826. Singh, K.P., Gupta, S., Singh, A.K., Sinha, S., 2010. Experimental design and response surface modeling for optimization of Rhodamine B removal from water by magnetic nanocomposite. Chemical Engineering Journal 165 (1), 151e160. Spicer, P.T., Pratsinis, S.E., 1996. Shear-induced flocculation: the evolution of floc structure and the shape of the size distribution at steady state. Water Research 30 (5), 1049e1056. Syu, M., Chen, B., Chou, S., 2003. A study on the sedimentation model and neural network online adaptive control of a benzoic acid imitated wastewater oxidation process. Industrial & Engineering Chemistry Research 42 (26), 6862e6871. Wang, D.S., Tang, H.X., Gregory, J., 2002. Relative importance of charge neutralization and precipitation on coagulation of kaolin with PACl: effect of sulfate ion. Environmental Science & Technology 36 (8), 1815e1820.
Wang, J.P., Chen, Y.Z., Ge, X.W., Yu, H.Q., 2007. Optimization of coagulation-flocculation process for a paper-recycling wastewater treatment using response surface methodology. Colloids and Surfaces A-Physicochemical and Engineering Aspects 302 (1e3), 204e210. Wang, J.P., Chen, Y.Z., Yuan, S.J., Sheng, G.P., Yu, H.Q., 2009. Synthesis and characterization of a novel cationic chitosanbased flocculant with a high water-solubility for pulp mill wastewater treatment. Water Research 43 (20), 5267e5275. Wen, S.H., Zhang, T., Tan, T.W., 2005. Optimization of the amino acid composition in glutathione fermentation. Process Biochemistry 40 (11), 3474e3479. Wong, S.S., Teng, T.T., Ahmad, A.L., Zuhairi, A., Najafpour, G., 2006. Treatment of pulp and paper mill wastewater by polyacrylamide (PAM) in polymer induced flocculation. Journal of Hazardous Materials 135 (1e3), 378e388. Zhang, L., Liang, Y.Z., Jiang, J.H., Yu, R.Q., Fang, K.T., 1998. Uniform design applied to nonlinear multivariate calibration by ANN. Analytica Chimica Acta 370 (1), 65e77. Zheng, H.L., Zhu, G.C., Jiang, S.J., Tshukudu, T., Xiang, X.Y., Zhang, P., He, Q.A., 2011. Investigations of coagulationflocculation process by performance optimization, model prediction and fractal structure of flocs. Desalination 269 (1e3), 148e156. Zhu, K., El-Din, M.G., Moawad, A.K., Bromley, D., 2004. Physical and chemical processes for removing suspended solids and phosphorus from liquid swine manure. Environmental Technology 25 (10), 1177e1187.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
In situ disinfection of sewage contaminated shallow groundwater: A feasibility study Morgan M. Bailey a, William J. Cooper b, Stanley B. Grant a,b,* a
Department of Chemical Engineering and Material Sciences, University of California, Henry Samueli School of Engineering, University of California, Irvine, CA 92697, USA b Department of Civil and Environmental Engineering, and Urban Water Research Center, Henry Samueli School of Engineering, University of California, Irvine, CA 92697, USA
article info
abstract
Article history:
Sewage-contaminated shallow groundwater is a potential cause of beach closures and
Received 18 June 2011
water quality impairment in marine coastal communities. In this study we set out to
Received in revised form
evaluate the feasibility of several strategies for disinfecting sewage-contaminated shallow
9 August 2011
groundwater before it reaches the coastline. The disinfection rates of Escherichia coli (EC)
Accepted 14 August 2011
and enterococci bacteria (ENT) were measured in mixtures of raw sewage and brackish
Available online 22 August 2011
shallow groundwater collected from a coastal community in southern California. Different disinfection strategies were explored, ranging from benign (aeration alone, and aeration
Keywords:
with addition of brine) to aggressive (chemical disinfectants peracetic acid (PAA) or per-
Disinfection
oxymonosulfate (Oxone)). Aeration alone and aeration with brine did not significantly
Shallow groundwater
reduce the concentration of EC and ENT after 6 h of exposure, while 4e5 mg L1 of PAA or
Sewage
Oxone achieved >3 log reduction after 15 min of exposure. Oxone disinfection was more
Remediation
rapid at higher salinities, most likely due to the formation of secondary oxidants (e.g.,
Water quality
bromine and chlorine) that make this disinfectant inappropriate for marine applications.
Peroxymonosulfate
Using a Lagrangian modeling framework, we identify several factors that could influence
Peracetic acid
the performance of in-situ disinfection with PAA, including the potential for bacterial
Oxone
regrowth, and the non-linear dependence of disinfection rate upon the residence time of
Brine
water in the shallow groundwater. The data and analysis presented in this paper provide a framework for evaluating the feasibility of in-situ disinfection of shallow groundwater, and elucidate several topics that warrant further investigation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Beach closures and advisories are generally caused by the coastal discharge of land-side sources of fecal pollution such as urban runoff or sewage effluent (Boehm et al., 2002; Grant and Sanders, 2010). However, contamination of shallow groundwater by aging sewage collection systems and failing onsite sewage treatment and disposal systems has also been
identified as a source of fecal pollution at some coastal beaches (Lipp et al., 2001). The latter typically involves a two-step process: (1) shallow groundwater is first contaminated by the surface or subsurface release of sewage, and (2) the shallow groundwater is then discharged to coastal waters under the influence of land hydraulic gradients, tidal pumping, and/or current induced pressure gradients (Burnett et al., 2003b). The coastal discharge of sewage-contaminated shallow
* Corresponding author. Tel.: þ1 949 824 8277; fax: þ1 949 824 2541. E-mail addresses: [email protected] (M.M. Bailey), [email protected] (S.B. Grant). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.020
5642
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
groundwater can impact near-shore water quality by vectoring fecal-oral pathogens (or their indicators) directly from sources of human sewage to receiving waters (Boehm et al., 2003). The discharge may also deliver nutrients that promote the survival, and perhaps growth, of fecal bacteria in near-shore waters (Boehm et al., 2004). To date, most studies on shallow groundwater discharge focus on its quantification using isotopic (McIlvin and Altabet, 2005), geophysical (Manheim et al., 2004; Santos et al., 2008; Swarzenski et al., 2006), or hydrological methods (Burnett et al., 2003a; Michael et al., 2005; Taniguchi et al., 2002), but to our knowledge there are no published reports on the very practical and important question of how to remediate shallow groundwater once it becomes contaminated with sewage. Given the advancing age and poor condition of sewage collection systems and onsite sewage treatment and disposal systems in many coastal areas of the U.S. (Food and Water Watch, 2008), a cost-effective strategy for remediating sewage-derived fecal pollution in brackish shallow groundwater is urgently needed for the sustainable use of coastal beaches and bays. While the best remediation strategy should always be to identify and eliminate the source(s) of sewage responsible for groundwater contamination, logistical and economic constraints may make such action infeasible in the near term, in which case interim solutions are needed. Further, even if a source of sewage contamination is identified and repaired, it may take some time before pathogens and indicator bacteria decay to acceptable concentrations, particularly when the sediments in question are subsurface and thus not exposed to sunlight (Mika et al., 2009). Subsurface injection of disinfectant (or “in-situ disinfection”) is one obvious interim strategy for remediating sewage-contaminated shallow groundwater, but several issues need to be resolved before such an approach can be implemented in practice. (1) Disinfectants can react with organic matter and/or trace anions (e.g., bromide and chloride ions) in brackish coastal groundwater to produce toxic disinfection by-products, and their discharge to coastal zones could pose a health risk to bathers along the shoreline and negatively impact sensitive marine ecosystems (Monarca et al., 2000). (2) The shallow groundwater adjacent to marine coastal areas can be thought of as “subterranean estuaries”, in which fresh (meteoric) waters mix with seawater before discharging to the ocean (Moore, 1999). Consequently, the flow in shallow groundwater can vary substantially in space and time, potentially affecting the disinfection contact times water parcels experience before discharging to coastal waters. (3) Sewage constituents, such as pathogens and their indicators, may not mix fully over the vertical dimension of the shallow groundwater, and therefore disinfection strategies may need to be tailored to target sewage constituents where they are located, for example near the top of the water table. (4) Planning and implementation of in-situ disinfection would greatly benefit from quantitative “design criteria” that account for the rate at which the target organisms decay with time upon exposure to disinfectant, the decay in disinfectant with time, the time scale over which water parcels reside in the shallow groundwater before discharge, and ecological processes that promote the subsurface regrowth of fecal bacteria.
In this paper we present laboratory data and modeling efforts intended to, at least preliminarily, address the issues raised above, using as a test case Avalon Bay, Catalina Island, California, where chronic fecal contamination of near-shore waters has been linked to sewage contamination of shallow groundwater by leaking sewage collection systems beneath the City (Boehm et al., 2003, 2009b). The goals of this study are to: (1) test three strategies for inactivating sewage-associated fecal bacteria from shallow groundwater, including aeration alone, aeration with addition of brine, and aeration with addition of either peracetic acid (PAA) or peroxymonosulfate (commercially known as Oxone); (2) characterize the salinity dependence and disinfection by-product formation potential of the latter two disinfectants; (3) evaluate several models for PAA disinfection; and (4) develop a Lagrangian modeling framework for evaluating, and potentially designing, in-situ disinfection of sewage-contaminated coastal shallow ground waters.
2.
Materials and methods
2.1.
Choice of disinfectant
Bench-scale experiments were carried out to measure the decay of sewage-associated fecal indicator bacteria in shallow groundwater upon exposure to: (1) aeration alone, (2) aeration with addition of brine, (3) aeration with addition of PAA (35% solution, CAS-79-21-0, Pfaltz & Bauer, CT), and (4) aeration with addition of Oxone (43% dry weight, CAS-37222-66-5, Alfa Aesar, Ward Hill, MA). The impact of aeration on bacterial dieoff was evaluated on the premise that, if efficacious, injection of ambient air into the subsurface would be a straightforward, economical, and environmentally benign approach for remediating sewage-contaminated shallow groundwater. The choice of brine as a candidate disinfectant was motivated by the fact that its injection into shallow groundwater is unlikely to cause environmental harm, and because the field site under consideration, Avalon Bay, has a desalination plant from which brine is produced as a waste product. PAA is a promising disinfectant for this application, because it produces little or no toxic disinfection by-products when mixed with seawater (it degrades into acetic acid, vinegar) (Kitis, 2004), is a strong biocide in marine waters as evidenced by its use as an antifouling agent in cooling water systems for coastal power plants, and is regarded as relatively safe for discharge to sensitive marine waters (Sanchez-Ruiz et al., 1995) including the highly regulated Italian Lagoon of Venice (Cristiani, 2005). To our knowledge, Oxone has not been tested as a biocide in coastal marine settings, but its application to this particular problem was motivated by the fact that, upon addition to water, it generates hydroxyl radicals and sulfate ions (Anipsitakis and Dionysiou, 2003); the former should accelerate the die-off of bacteria and viruses while the latter is already present at high concentrations in marine waters. Many commonly used disinfectants (e.g., ozone, chlorine gas, sodium hypochlorite) were excluded from consideration for logistical reasons and, although highly effective biocides, they would likely react with trace anions and organics in sewagecontaminated brackish groundwater to form toxic
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
disinfection by-products (Abarnou and Miossec, 1992; Allonier et al., 1999).
2.2.
Experimental characterization of disinfection kinetics
Raw (untreated) sewage was diluted 1:100 using various mixtures (with various final salinities) of three source waters: (1) relatively fresh (salinity ca. 1) water from an inland groundwater well, (2) Avalon Bay water (salinity ca. 32), and (3) desalination plant brine (salinity ca. 45). All source waters were aseptically collected from the City of Avalon using sterile polypropylene bottles and mixed with raw sewage collected from influent to the City of Avalon wastewater treatment plant. Source waters were not sterilized prior to mixing with sewage, and therefore bacteria populations present in the final mixtures of sewage and source waters probably included (minor) contributions from the latter. Bench-scale disinfection experiments commenced within 6 h of sample collection and were carried out in 4 L Nalgene polypropylene reaction vessels maintained in the dark and partially submerged in a temperature controlled and recirculating water bath at 15 1 C. Contents of the reactor vessel were continuously aerated using a frit through which ambient air was driven by an aquarium pump. To capture a range of salinities and disinfectant concentrations, a matrix design was adopted in which over 45 separate disinfection experiments (including 10 no-disinfectant-added control experiments) were conducted at five different disinfectant concentrations (ranging from 0 to 7.3 mg L1 for Oxone, and 0e8 mg L1 for PAA) and four different salinities (0, 15, 32, 45). The concentration range adopted for PAA was based on previously published studies with this disinfectant (Kitis, 2004; Santoro et al., 2007). To our knowledge, this is the first study to evaluate Oxone disinfection in saline mixtures, and thus the concentration range adopted for this disinfectant was based on pilot studies (not reported) with this disinfectant. Each batch reactor was sampled seven times over a period of 1e6 h (depending on experiment). This sampling schedule was chosen to resolve disinfection over a single ebb tide, which represents a theoretical minimum time a fluid parcel would be in contact with disinfectant before discharging to Avalon Bay, assuming that the discharge of shallow groundwater was under tidal control. Samples were extracted from the reactor using either a sterile syringe or pipet, analyzed for pH and conductivity, quenched by addition of approximately 0.4 mL of 0.1 N sodium thiosulfate (CAS-7772-98-7, Mallinckrodt Chemicals), immediately diluted either 1:10 or 1:100 in sterile deionized water (Hardy Scientific, California), and enumerated for Escherichia coli (EC) and enterococci bacteria (ENT) using Colilert-18 and Enterolert defined substrate tests implemented in a 97-Well Quanti-Tray format (IDEXX Laboratories, Maine). The adoption of IDEXX Colilert-18 and Enterolert was motivated based on the fact that both tests are approved by the U.S. Environmental Protection Agency for enumerating EC and ENT bacteria in ambient waters (USEPA, 2003) and, more to the point, are used by the Los Angeles Department of Health Services in their routine monitoring of recreational beach water quality in Avalon Bay, and as a basis for management decisions regarding, for example, the posting of Avalon beaches as unfit for swimming. For a subset of the disinfection experiments, dissolved organic carbon (DOC) and total organic carbon (TOC) concentrations were measured: (1) on the source water (groundwater or
5643
bay water) prior to addition of sewage and disinfectant; (2) on the source water plus sewage prior to the addition of disinfectant; and (3) at the end of the disinfection experiment. Samples analyzed for DOC and TOC were collected in 500 mL amber glass bottles and acidified with 2 mL of hydrochloric acid, and then analyzed by TestAmerica using Standard Method 5310B.
2.3.
Disinfection rate constants
Two different disinfection rate constants are reported in this study: (1) an effective first-order disinfection rate constant k0d , and (2) an intrinsic disinfection rate constant kd. Values of the effective first-order rate constant k0d were estimated by regressing log-transformed bacteria concentration against time based on Chick’s Law (Chick, 1908): NðtÞ ¼ k0d t ln N0
(1)
In Eq. (1), N(t) represents the concentration of bacteria in the reaction vessel at any time t and N0 represents the bacterial concentration present in the mixture at the start of the disinfection experiment. The regression excluded any bacterial measurements in the initial lag period and bacterial measurements that fell below the lower-limit of detection for the Colilert and Enterolert assays (in most cases, the lower limit of detection was 10 most probable number (MPN) per 100 mL of sample). In the case of PAA disinfection, an intrinsic disinfection rate constant was also calculated, which represents the susceptibility of bacteria to disinfectant independent of disinfectant concentration: kd ¼ k0d =C0
(2)
The variable C0 represents the initial (molar) concentration of peroxycompounds which, in the case of commercial preparations of PAA, includes both PAA and hydrogen peroxide (Wagner et al., 2002).
2.4.
Activation energy for PAA disinfection
Using the reactor set-up described above, four separate disinfection experiments were carried out at five temperatures (T ¼ 5, 10, 15, 20 and 30 C) and a fixed concentration of PAA (6 mg L1) to determine the temperature sensitivity of EC and ENT disinfection by PAA. Intrinsic disinfection rate constants kd calculated from these experiments (see Section 2.3) were used to determine activation energies for the PAA disinfection of EC and ENT, based on an Arrhenius plot of ln(kd) against 1/T.
3.
Disinfection results and discussion
3.1.
Measurements of pH, TOC, and DOC
Over aeration of the reactor vessel caused the pH to increase slightly from 7.9 0.2 to 8.3 0.2. No other systematic pH trends were observed across the different experiments and different source waters tested. The average DOC concentration in groundwater ranged from 1.6 0.14 mg L1 (N ¼ 7,
5644
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
foreshore well, salinity 20) to 2.9 0.2 mg L1 (N ¼ 4, inland well, salinity 1). The TOC/DOC ratio was near unity in water collected from the inland well (1.1 0.1, N ¼ 4) implying that most of the organic carbon was dissolved. Raw sewage collected from the Avalon Sewage Treatment Plant had much higher levels of DOC (25 mg L1, N ¼ 1). However, sewage contributed only background levels of DOC (<0.25 mg L1) to the final 1:100 mixtures of sewage and source waters used in the disinfection experiments. For disinfection experiments carried out with PAA, the largest source of DOC was frequently the PAA solution itself, which contained upwards of 280 mg L1 of carbon as acetic acid. The acetic acid was present as an equilibrium component of the commercial PAA preparation, and generated as a breakdown product of PAA.
3.2. Experimental characterization of disinfection kinetics First-order effective disinfection coefficients ðk0d Þ estimated from the 45 separate disinfection experiments are contoured using Delaunay triangulation (Igor Pro v 6.10, Lake Oswego, Oregon) against disinfectant concentration and salinity in Fig. 1. These results are described in the sections below. A more detailed analysis of k0d values, and an evaluation of different disinfection models, is presented later in the paper (Section 4).
3.2.1.
Aeration alone and aeration with addition of brine
We started by evaluating two environmental benign approaches for removing fecal indicator bacteria from sewagecontaminated shallow groundwater; namely, injection of ambient air and/or injection of brine produced from a local
desalination plant. Aeration alone and aeration with brine did not significantly reduce EC and ENT concentrations after 6 h of exposure; i.e., within the resolution of these experiments k0d ¼ 0. Thus these two strategies are unlikely to be effective against indicator bacteria in shallow groundwater over the time scale of a single ebb tide.
3.2.2.
Aeration with addition of PAA and oxone
EC and ENT concentrations were reduced by more than 1000 fold (3 log units) after 15 min of exposure to 4e5 mg L1 of PAA. The effective disinfection rate constants calculated for these experiments (Fig. 1A and B): (1) increased monotonically with 1 PAA dose up to the maximum rate ðk0d ¼ 0:2 min Þ resolvable with our experimental set-up; (2) do not depend, at least not dramatically, on the salinity of the disinfection mixture; and (3) are larger for EC than ENT at a fixed PAA dose. The last observation is consistent with the results reported in Stampi et al. (2002). Averaged across all experiments (and excluding any experiments where the effective disinfection rate constant exceeded 0.2 min1), the average intrinsic rate ¼ 2:3 0:57 and constants for PAA disinfection are kENT d 1 1 EC kEC d ¼ 3:7 0:2 mM min : The estimate for kd is similar to intrinsic rate constants estimated from previously published data for PAA disinfection of EC in secondary settled effluent (Dell’Erba et al., 2004) and PAA disinfection of fecal coliform (FC) in secondary-treated sewage effluent (Wagner et al., 2002) (Table 1). The fact that the intrinsic rate constants for PAA disinfection of EC and FC are similar across these three studies is notable, given that EC is a subset of FC, and the very different initial bacterial concentrations (and presumably sewage content) associated with the different source waters
Fig. 1 e Contour plots of the effective first-order rate constants k0d [minL1] for PAA disinfection of EC (panel A) or ENT (panel B), and Oxone disinfection of EC (panel C) or ENT (panel D).
5645
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Table 1 e Intrinsic disinfection rates and parameters from multiple studies. Value kd (mM1 min1)a
Source FC
EC
%H2 O2=%PAAb
Temp ( C)
N0 c
Solutiond
GW, BW, Br, TPI TPE TPE
ENT
This study
NA
3.7 0.2
2.3 0.57
0.183
15
104
Dell’Erba et al. Wagner et al.
NA 1.04 1.26
3.65 2.41 NA
NA NA
1.53 1.54
NR NR
102 105
NR-not reported. NA-not applicable. a Intrinsic disinfection rate. b Ratio of percent mass of PAA and hydrogen peroxide in PAA equilibrium solution. c Magnitude of initial bacteria concentration. d Sources waters for disinfection studies (GW-groundwater, BW-way water, Br-brine, TPI-treatment plant influent, TPE-treatment plant effluent).
used in these studies, ranging from 102 to 105 bacteria per 100 mL (Table 1). EC and ENT concentrations were reduced by more than 3 log units after 15 min of exposure to 4 mg L1 Oxone. However, unlike PAA, the disinfection rate depends on both Oxone dose and solution salinity (Fig. 1C and D). At low Oxone dose (<2 mg L1), the effective disinfection rate increases monotonically with Oxone dose, and exhibits no obvious salinity dependence (i.e., the contour lines are near vertical in this region of the plot). At higher Oxone dose, the effective disinfection rate increased monotonically with salinity, and exhibits no obvious dose dependence (i.e., the contour lines were near horizontal in this region of the plot). One possible explanation for this pattern is that Oxone may oxidize anions in the shallow groundwater and brine to yield, for example, the secondary oxidants chlorine and bromine. To explore this idea, a set of control experiments were carried out in which either PAA or Oxone was added to an aqueous solution consisting of 35 g L1 NaCl to mimic the background salinity of seawater and varying concentrations
(0e70 mg L1) of KBr to mimic the presence of trace anions. The formation of oxidized forms of chloride and bromide ions (e.g., chlorine and bromine) was monitored by measuring absorbance of the indicator dye N,N-diethyl-p-phenylenediamine (DPD) (Standard Methods, 4500-CL G). Fig. 2A shows DPD spectra measured after addition of 10 mg L1 of either Oxone (solid lines) or PAA (dashed lines) to distilled water (Solution 1), an aqueous solution consisting of 35 g L1 NaCl (Solution 2), or an aqueous solution consisting of 35 g L1 NaCl and 70 mg L1 KBr (Solution 3). In this set of experiments, the disinfectant was allowed to react in the solution for 5 min, whereupon DPD was added, and 1 min later the DPD absorbance spectrum was measured. Referring to Fig. 2A, the absorbance of DPD increased in the order Solution 1 < Solution 2 < Solution 3, consistent with the idea that Oxone oxidizes both chloride and bromide ions to form secondary oxidants. DPD absorbance did not increase when PAA was allowed to react for 5 min in Solution 2 and increased only slightly when PAA was allowed to react for 5 min in Solution 3. These latter results are consistent with the findings
Fig. 2 e Panel A: DPD absorbance spectra measured after 10 mg LL1 of either Oxone (solid lines) or PAA (dashed lines) is allowed to react for 5 min in: Solution 1 (DI water), Solution 2 (DI water and 35 g LL1 NaCl), or Solution 3 (DI water, 35 g LL1 NaCl, and 70 mg LL1 KBr). Panel B: DPD absorbance at 515 nm after 10 mg LL1 of either Oxone (solid lines) or PAA (dashed lines) are allowed to react for 5 min in Solution 2 with KBr concentrations shown. Panel C: the kinetics of oxidant formation by 10 mg LL1 of either Oxone (solid line) or PAA (dashed line) in Solution 3.
5646
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
of Booth and Lester (1995), who report that PAA could not oxidize chloride ion to hypochlorous acid, but could oxidize bromide ion to hypobromous acid. In the presence of Oxone, DPD absorbance increased with increasing KBr concentration in Solution 2 (Fig. 1B), and the oxidation kinetics in Solution 3 are relatively rapidly (<5 min of exposure time, Fig. 1C). In the presence of PAA, on the other hand, DPD absorbance increased only slightly with increasing KBr concentration in Solution 2 (Fig. 2B), and oxidization kinetics in Solution 3 are relatively slow (>150 min exposure time) (Fig. 2C). Collectively, the results presented above are consistent with the idea that Oxone quickly oxidized trace anions in brackish shallow groundwater to form secondary oxidants that act synergistically with Oxone to enhance disinfection rates. However, the reaction of Oxone with trace anions could also increase the toxicity of the shallow groundwater, by producing compounds that are carcinogenic (e.g., bromateion) and/or by producing secondary oxidants (e.g., bromine and chlorine) that subsequently react with organic compounds to form toxic disinfection by-products (e.g., trihalomethanes) (Guo and Lin, 2009). For these reasons, and despite its obvious utility as a disinfectant, Oxone should not be used for in-situ disinfection of sewage contaminated shallow groundwater in settings, like Avalon Bay, where the groundwater is under marine influence. PAA, on the other hand, does not appear to react strongly with chloride ion, or quickly with bromide ion, and thus is less likely to produce toxic disinfection by-products; a conclusion supported by several published studies (Liberti et al., 1999; Dell’Erba et al., 2007; Kitis, 2004). Although PAA does not appear to generate the quantity and spectrum of disinfection by-products associated with many disinfectants, some researchers have raised concern about the potential environmental impact of releasing disinfected effluents that contain a PAA residual (Antonelli et al., 2009; de Lafontaine et al., 2008), which can be toxic to crustaceans and microorganisms (Antonelli et al., 2009; de Lafontaine et al., 2008). However, Lafontaine et al. (2008) note that, because PAA breaks down rapidly in seawater, any potential impacts would be localized around the region where disinfection residual is discharged to the environment. Thus, if PAA is used for in-situ disinfection of sewage contaminated shallow groundwater, care should be taken to minimize the PAA residual discharged to coastal waters.
3.3.
Activation energy for PAA disinfection
Based on measurement of intrinsic disinfection rates over a range of temperatures, from 5 to 30 C, the following activation energies were estimated for PAA disinfection of EC and ENT: 37.5 7.8 kJ mol1 and 38.0 9.3 kJ mol1, respectively. These activation energies are used later in the paper to estimate intrinsic disinfection rates for PAA over a range of temperatures relevant to the shallow groundwater in Avalon Bay (see Section 5.1).
4.
Modeling PAA disinfection kinetics
Of the strategies evaluated above, PAA disinfection appears the most viable, given that it is both a potent biocide and less likely to form toxic disinfection by-products. In this section we
evaluate several published models for PAA disinfection, all of which are special cases of Hom’s Law (Hom, 1972): dN ¼ kd Ntm Cn dt
(3)
where, N, C, kd, and t represent bacterial concentration, disinfectant concentration, intrinsic disinfection rate constant, and time, respectively. Chick’s Law (Eqs. (1) And (2) of this paper) corresponds to a choice of exponent values m ¼ 0, n ¼ 1, and a constant disinfectant concentration, C ¼ C0. Wagner et al. (2002) suggested that, during the disinfection process, the concentration of peroxycompounds in commercial PAA mixtures decays with time in accordance with the following second-order rate law: dC ¼ k C2 dt
(4)
The rate constant k* depends on the initial peroxycompound concentration C0 (Wagner et al., 2002): k ¼ aCb 0
(5)
Where the pre-factor and power-law exponents are given by a ¼ 0.0093 mM(b 1) min1 and b ¼ 1.420. In their analysis of PAA disinfection, Santoro et al. (2007) suggest that peroxycompounds exhibit zero-order decay; however, zero-order decay models predict negative concentration in finite time, and therefore will not be considered further here. Combining Eqs. (4) and (5), and solving the differential equation, yields the following prediction for disinfectant concentration as a function of time: CðtÞ ¼
1 1=C0 þ k t
(6)
Given this time-dependence for the disinfectant concentration, Hom’s Law becomes: dN n ¼ kd Ntm ½1=C0 þ k t dt
(7)
Wagner et al. (2002) solved Eq. (7) for two different choices of the exponents n and m: (1) n ¼ 1 and m ¼ 0 (no-tailing model), and (2) n ¼ 1 and m ¼ 1 (tailing model). The tailing model provided a better empirical fit to the PAA disinfection data, although their intrinsic disinfection rate constant kd varied with the initial disinfectant concentration, and a new fitting parameter (the time t at which N(t) ¼ N0) was introduced. Here we opt for the more parsimonious no-tailing model, for which an exact solution can be derived: kd =k
N ¼ N0 ðk C0 t þ 1Þ
(8)
PAA disinfection data collected in this study were tested against Chick’s Model (Eq. (1)) and the no-tailing model (Eq. (8)) in Fig. 3. The data are plotted so that the intrinsic disinfection rate constant kd can be estimated directly from the slope b of the best-fit line: kd ¼ 2.303b (Chick’s Law) or kd ¼ b (no-tailing model). Also shown in Fig. 3 are two estimates of model performance, including the Pearson’s r2 correlation between logN/N0 and the x-axis (either C0t or log½k C0 t þ 1=k ), and the root mean square error (RMSE) between modeled and measured values of bacterial log reduction. In general, for a given choice of fecal bacteria group (either ENT or EC) the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
5647
Fig. 3 e PAA disinfection data for EC (panel A and B) and ENT (panel C and D) fit to either Chick’s model (Eq. (1), panel A and C) or Wagner et al.’s no-tailing model (Eq. (8), panel B and D). Values for kd are mML1 minL1 and units for RMSE are logN/N0.
two disinfection models have similar r2 and RMSE values (compare panels A and C with panels B and D in Fig. 3). Both models are a better description of ENT disinfection (r2 ¼ 0.68 to 0.72, RMSE ¼ 0.32e0.37) than EC disinfection (r2 ¼ 0.52 to 0.55, RMSE ¼ 0.6e0.62). The fact that ENT exhibited less dispersion around the no-tailing model may reflect less variability (compared to EC) in the disinfection resistance of enterococci bacteria populations present in the sewage and source waters. Intrinsic rate constants estimated from the slope b are also similar for the two models. Chick’s Law yields intrinsic rate constants for EC and ENT of 3.0 and 1.9 mM1 min1, respectively (panels A and C). The no-tailing model yields intrinsic rate constants for EC and ENT of 3.3 and 3.5 mM1 min1, respectively (panels B and D). The intrinsic disinfection rate for EC is also similar to values estimated by averaging rate constants obtained from our individual experiments (see Section 3), and from data reported in other studies of PAA disinfection (Table 1). In summary, Chick’s Law and the notailing model were both reasonably good predictors of bacterial decay caused by PAA disinfection, and both yield values of
the intrinsic disinfection rate constant kd that were consistent across models, and across studies. Relative to the modeling effort described in the next section, the primary benefit of the no-tailing model is that it accounts for the decay in disinfectant concentration that will inevitably occur following injection of PAA into the subsurface.
5.
Lagrangian model of in situ disinfection
In this section we develop a quantitative model that accounts for the physical, chemical, and biological factors that might influence the in-situ disinfection of sewage contaminated shallow groundwater with PAA. The model is developed in two stages. First, a Lagrangian framework is used to predict the concentration of both sewage constituents (fecal indicator bacteria) and PAA residual in a parcel of shallow groundwater as it travels from the point of injection to the point where it is discharged to the coastal ocean. Second, an analytical model is derived for fluid parcel residence time in shallow
5648
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
groundwater, for the situation (relevant to Avalon Bay) where shallow groundwater discharge is dominated by tidal pumping and meteoric flow.
5.1. Lagrangian framework for modeling disinfection kinetics Fig. 4 illustrates the physical transport processes that might affect fecal bacteria concentration in a parcel of groundwater as it moves from the point where it first encounters subsurface injected disinfectant (point A) to the point where it exits the shallow groundwater and is discharged to the coastal ocean (point B). As the parcel moves from A to B, the concentration of bacteria in the fluid parcel changes with time due to disinfection, non-disinfection related die-off, regrowth, and attachment to the porous matrix (filtration). Mass balance over a fluid parcel as it travels from A to B yields the following rate expression for the concentration of bacteria (N, bacteria L3), where s [min1] represents the time a sewage contaminated water parcel has been in contact with disinfectant (referred to here as “residence time”), C is the disinfectant concentration [mM], and the rate constants for disinfection, inactivation, filtration, and growth are kd [mM1 min1], ki [min1], kf [min1], and mg [min1]: dN 1 ¼ kd N½1=C0 þ k s ki þ kf N þ mg N ds
(9)
In formulating Eq. (9), we adopted the no-tailing version of Hom’s Law described earlier (Eq. (7) with m ¼ 0 and n ¼ 1) and assume that inactivation, filtration, and regrowth of bacteria all follow first-order kinetics. While we did not
where, mg,max and Ks represent the maximum growth rate and saturation constant, respectively. Based on the measurements of DOC presented in Section 3.1, potential sources of DOC include ambient groundwater (DOCGW), sewage (DOCSewage), and acetic acid associated with the PAA mixture, including acetic acid present as an equilibrium component of commercial PAA preparations (DOCAA,0), and acetic acid formed by the decomposition of PAA during disinfection (DOCA(s)), where the latter can be estimated from the loss of peroxycompound concentration with time: DOCAA ðsÞ ¼ 0:82MðC0 CðsÞÞ
(11)
Eq. (11) assumes stoichiometric conversion of PAA to acetic acid, taking into account the molar fraction (0.82) of the peroxycompound concentration that is PAA and the weight of carbon associated with every mole of acetic acid, M ¼ 25 g of carbon per mole. Combining Eqs. (10) and (11), we have the following prediction for the total DOC available for growth of fecal indicator bacteria: DOCT ðsÞ ¼ DOCGW þ DOCSewage þ DOCAA;0 þ 0:82MðC0 CðsÞÞ:
(12)
Combining Eqs. (9)e(12) and solving the resulting differential equation, yields the following formula for the concentration of bacteria in a parcel of shallow groundwater as a function of residence time (s):
kf ðC0 M þ Ks Þ þ C0 Mki C0 Mmg;max þ DOCT0 kf þ ki mg;max þ ki Ks N0 ðDOCT0 þ Ks Þ exp s C0 M þ DOCT0 þ Ks a
NðsÞ ¼
(e.g., Lazarova et al., 1998; Lefevre et al., 1992). Here we use the Monod equation (Levenspiel, 1980) to model the dependence of growth rate mg on dissolved organic carbon (DOCT): mg;max DOCT ðsÞ (10) mg ðsÞ ¼ Ks þ DOCT ðsÞ
kd =k
ðC0 k s þ 1Þ
½C0 k sðMC0 þ DOCT0 þ Ks Þ þ DOCT0 þ Ks
observe regrowth in the disinfection experiments presented earlier, it is a well-known that the acetic acid in commercial PAA mixtures can serve as a carbon source for the growth of heterotrophic bacteria, including fecal indicator bacteria
a¼
a
Ks mg M k ðC0 M þ DOCT0 þ Ks Þ2
(13a)
(13b)
Fig. 4 e A conceptual model for the in-situ disinfection of sewage-contaminated shallow groundwater in coastal marine environments.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Fig. 5A presents the bacterial log reduction predicted by Eq. (13) (solid curves) as a function of initial PAA concentration (vertical axis) and residence time (horizontal axis). This graph was generated using the parameter values listed in Table 2, which are based on the experimental measurements presented earlier and field conditions relevant to the Avalon Bay field site, including a shallow groundwater temperature of 15 C. For the range of peroxycompound concentrations considered in Fig. 5A, the dependence of bacteria concentration on residence time exhibits two patterns. At small residence times, very little of the PAA has been converted to acetic acid, disinfection dominates, and bacteria concentration declines rapidly with increasing residence time. At longer residence times, PAA has been mostly converted to acetic acid, regrowth of bacteria dominates, and bacterial concentration increases with increasing residence time. This decay/ regrowth pattern is illustrated for a single initial peroxycompound concentration (C0 ¼ 0.03 mM) and temperature (T ¼ 15 C) in Fig. 5B (dotted line). For this particular choice of parameter values, the model predicts that bacterial concentration falls approximately 2 log units within 12 h, and increases thereafter, eventually rising above the initial bacteria concentration at a residence time of around two days. As expected, bacteria removal increases monotonically with increasing initial peroxycompound concentration, as illustrated for a fixed residence time (s ¼ 5 days) and temperature (T ¼ 15 C) in Fig. 5C (dotted line). Using the activation energies for the intrinsic disinfection rate constant reported in Section 3.3, the model predicts similar trends over the range of temperatures (13e17 C) typically measured in Avalon shallow groundwater (compare the family of curves in Fig. 5B and C).
5649
In an ideal scenario, by the time a fluid parcel of shallow groundwater is discharged to the ocean, both its bacterial concentration and PAA residual would be very small. The dependence of PAA residual on initial peroxycompound concentration and residence time can be estimated from Eq. (6), by setting the left hand side equal to a fixed residual concentration, Cres and solving for residence time s: sres ¼
b 1 1 C0 C C1 0 a res
(14)
Curves of constant peroxycompound residual are plotted in Fig. 5A (dashed curves). As expected, peroxycompound residual declines with increasing residence time for a fixed C0. Interestingly, the curve for Cres ¼ 1 mM roughly demarcates the transition from bacteria disinfection to regrowth (compare solid and dotted lines in Fig. 5A).
5.2.
Residence time of fluid parcels
The Lagrangian model presented above reveals that both the bacterial concentration and PAA residual were sensitive to the residence time of water parcels in shallow groundwater, and in this section we describe some of the physical processes that can affect this key parameter. Here, residence time has precisely the same meaning as in estuarine systems: “how long a parcel, starting from a specified location within a waterbody, will remain in the waterbody before exiting” (Monsen et al., 2002). Provided that material diffusion can be neglected (a key assumption in the Lagrangian approach adopted here) (Deleersnijder et al., 2001), the residence time
Fig. 5 e Model predictions for EC concentration and PAA residual for the in-situ disinfection of sewage contaminated shallow groundwater with PAA. Panel A: Curves of constant EC log reduction predicted by Eq. (13) (solid lines labeled with numbers ranging from L1 to L9), and curves of constant peroxycompound residual predicted by Eq. (14) (dashed curves, labeled with numbers ranging from 0.2 to 1 mM). Panel B: Change in EC concentration with residence time predicted by Eq. (13) for C0 [ 0.03 mM and the ambient groundwater temperatures shown. (C) Change in EC concentration with increasing initial PAA concentration predicted by Eq. (13) for a fixed residence time of s [ 5 days and the ambient groundwater temperatures shown. Parameter values used to generate these curves are listed in Table 2.
5650
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Table 2 e Inputs parameters used to produce the disinfection contour plot (Fig. 5), residence time model (Fig. 6). Constant a b kd ki kf mg Ks DOCT0 M A Kp l
Name
Value (b 1)
k* fitting parameter k* fitting parameter EC disinfection rate (T ¼ 13,15,17 C) EC natural decay rate EC filtration rate Max growth rate Growth rate saturation constant Initial DOC Concentration Mass of carbon per mole PAA Wave amplitude Hydraulic conductivity Wave number
0.0093 mM min 1.42 3.3, 3.7, 4.1 mM1 min1 0.1 h1 0 min1 0.005 min1 0.07 mg L1 25 g 1m 5 104 m s1 0.05 m1
associated with the movement of a fluid parcel from point A to B in Fig. 4 can be written explicitly as follows: ZL sc ¼
dx v
(15)
0
where, v is the velocity experienced by the fluid particle as it moves toward the ocean. Here we have subscripted the residence time calculated from Eq. (15) (sc) to distinguish it from the actual residence time of a water parcel included in our disinfection model above (s). Fluid parcel velocity has contributions from tidal pumping (vt), wave set-up (vw), meteoric groundwater flow (vm), seasonal evapotranspiration (vs), and density driven flow (vd) (Burnett et al., 2003b; Michael et al., 2005): v ¼ vt þ vw þ vm þ vs þ vd
(16)
Given that tidal pumping appears to dominate the discharge of shallow groundwater to Avalon Bay (Boehm et al., 2009b), here we focus on the tidal pumping and meteoric terms in Eq. (16), vt and vm. Flow fields generated by tidal pumping can be approximated from the following expression (Nielsen, 1990):
vt ðx; tÞ ¼ Kp Alelx ðcos½ut lx sin½ut lxÞ
Source 1
(17)
Nielsen derived Eq. (17) from the Boussenesq equation after invoking a number of simplifying assumptions, including a homogeneous unconfined aquifer of hydraulic conductivity KP, a vertical beach face, and a one-dimensional flow field (parallel to the x-axis, see Fig. 4) characterized by a wave number l, and forced by a single harmonic tide with amplitude A and angular frequency u. The wave number is defined as l ¼ 2p/x0, where x0 represents the inland distance over which tidal fluctuations in the shallow groundwater are significant. Because the flow field predicted by Eq. (17) is tidally periodic and spatially variable, the residence time of a fluid parcel will depend not only on where in the aquifer it is released (referred to here as the setback distance, x ¼ L, see Fig. 4) but also on when in the tidal cycle that release occurs; a very similar phenomenon has been described for the residence time distributions in coastal estuaries (Monsen et al., 2002; Oliveira and Baptista, 1997). Despite these complications, a characteristic residence time can be estimated from Eqs. (15)e(17) by releasing the fluid parcel at the precise moment when the recessional tide wave (associated
Wagner et al. Wagner et al. Measured Estimated Estimated Surbeck et al. Surbeck et al. Measured Calculated Estimated Calculated via KozenyeCarman Estimated
with the falling tide) passes point A in Fig. 4, which is mathematically equivalent to assigning the value p to the quantity (ut lx) in Eq. (17). After invoking this simplification and allowing for the possibility of non-zero meteoric flow, Eqs. (15)e(17) can be combined to yield the following estimate for the characteristic residence time of a fluid parcel released at a setback distance x ¼ L from the beach: sc ¼
log AlKp elL vm log AlKp vm ; vm
sc ¼
1 lL e 1 ; Al2 Kp
vm ¼ 0
vm > 0
(18a)
(18b)
For the range of parameter values typical of the field site in Avalon Bay (see Table 2), the residence times predicted by Eq. (18) vary over one hundred thousand fold, from approximately 30 mine1000 days (Fig. 6). This variability in residence time derives, in part, from the non-linear dependence of residence time on both set-back distance and meteoric flow (Eq. (18a)). Furthermore, if studies of residence times in estuaries are any guide, the residence time sc of water parcels in shallow groundwater is best characterized by a probability distribution, not a single value. Studies in estuarine systems have noted that fluid parcel residence times tend to follow probability distributions characterized by long tails, implying that
Fig. 6 e Characteristic shallow groundwater residence times predicted by Eq. (18) for various set-back distances and meteoric flow velocities.
5651
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Table 3 e Inputs Parameters used in the uncertainty analysis (Eq. (19)). Constant k* kd ki mg Ks DOCT0 C0 s
Name
Value
Estimated uncertainty
PAA decay rate EC disinfection rate (T ¼ 15 C) EC natural decay rate Max growth rate Growth rate saturation constant Initial DOC concentration Initial peroxycompounds Residence time
1.35 L$mmol1$min1 3.7 L$mmol1$min1 0.1 h1 0.005 min1 0.07 0.006 g/ L1 0.03 mmol 2 day
0.2 0.2 0.2 0.2 0.2 0.2 0.2 1
a minority of fluid parcels spend a very long time in the estuary (Oliveira and Baptista, 1997).
6. Uncertainty analysis and practical implications One advantage of the analytical in-situ disinfection model derived earlier (Eq. (13)) is that the uncertainty associated with different independent variables can be assessed quantitatively using the Law of Uncertainty (Taylor and Kuyatt, 1993): u2 ðSÞ ¼
p X i¼1
2 vS u2 ðXi Þ vXi
(19)
In this equation, u2(S ) and u2(Xi) represents the variance of the dependent and independent variables, respectively, the dependent variable is the log-transformed bacteria concentration (S ¼ log(N )), the independent variables Xi include all variables appearing on the right hand side of Eq. (13) (i.e., C0, Ks, mg, DOCT0, k*, kd, ki, or s), vS/vXi is the sensitivity of the dependent variable to change in a particular independent variable (computed analytically from Eq. (13)), and the summation is taken over all independent variables (P ¼ 8). The uncertainty (or variance) associated with each independent variable was estimated from the relative uncertainty UR(Xi) and magnitude jXi j values listed in Table 3: uðXi Þ ¼ UR ðXi ÞjXi j
(20)
The form of the Law of Uncertainty adopted here assumes that independent variables do not co-vary, which is reasonable for most combinations of independent variables included in this analysis. When applied to Eq. (13), the Law of Uncertainty reveals that 99% of the variance in the log-transformed bacteria concentration can be attributed to variance in just three independent variables: residence time (s) (90%), maximum growth rate (mg) (8%), and the inactivation rate (ki) (1%). These results imply that the prediction (and optimization) of in-situ disinfection will depend strongly on the residence time of shallow groundwater which, in turn, depends non-linearly on the injection well set-back distance (L) and physical characteristics of the shallow groundwater system that can vary in time and space (l, A, KP, vm) (see Eq. (18)). Given the very approximate nature of the analysis that led to Eq. (18), experimental characterization of shallow groundwater residence times would be a fruitful topic for further investigation.
The disinfection model’s sensitivity to residence time is, in part, a consequence of the fact that fecal bacteria can grow in the environment, and thus dramatically different disinfection outcomes (e.g., from a net reduction in bacteria concentrations to a net increase in bacteria) can be caused by slight changes in residence time of water parcels in the shallow groundwater. Given that most recreational waterborne illnesses are caused by human viruses that cannot grow outside their host (Schoen et al., 2011), it is possible that in-situ disinfection of sewage contaminated shallow groundwater would reduce shoreline concentrations of human pathogens (and hence lower recreational waterborne illness rates), even if it did not substantially reduce fecal indicator bacteria concentrations. Indeed, the environmental growth of EC and ENT can lead to a decoupling between fecal indicator bacteria and human pathogens in recreational waters (Litton et al., 2010), and potentially nullify epidemiological relationships upon which current fecal indicator bacteria criteria are based (Colford et al., 2007). In light of these and other concerns the U.S Environmental Protection Agency is evaluating and possibly revising the current water quality criteria for marine recreational beaches (Boehm et al., 2009a).
7.
Conclusions
Aeration alone and aeration with brine did not significantly reduce EC and ENT concentrations in mixtures of raw sewage and shallow groundwater after 6 h of exposure, while 4e5 mg L1 of PAA and 4 mg L1 Oxone achieved >3 log reduction of EC and ENT after 15 min of exposure. Oxone disinfection is enhanced at higher salinities, most likely due to the formation of secondary oxidants (e.g., chlorine and bromine) that make this disinfectant inappropriate for marine applications. PAA disinfection of fecal bacteria in shallow groundwater in coastal settings depends non-linearly on residence time, and the “ideal” disinfection outcome (low bacterial concentration and low PAA residual) is achieved over a relatively narrow window of residence times and initial disinfection concentrations. By analogy to surface estuaries, the residence time of water parcels in shallow groundwater under the influence of marine tides (i.e., subterranean estuaries) is likely to exhibit broad (e.g., logenormal) probability distributions with long
5652
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
tails, and depend sensitively on local precipitation, meteoric flow, and tidal variability. Uncertainty calculations suggest 99% of the uncertainty associated with the log reduction of bacteria is caused by just three independent variables: residence time (s) (90%), maximum growth rate (mg) (8%), and the inactivation rate (ki) (1%). Given these results, further research into the residence time of water in shallow groundwater is needed before an in-situ disinfection schemes can be successfully designed and implemented.
Acknowledgments The authors thank R. Litton, L. Ho, and J. Monroe for assistance with the experiments, and C. Wagner and P. Woolson, and the City of Avalon staff for the use of City Hall for the disinfection studies. Funding was provided by the City of Avalon and State Water Resources Control Board Clean Beaches Initiative, under Agreement 07-582-550.This is publication 66 of the Urban Water Research Center, University of California, Irvine.
references
Abarnou, A., Miossec, L., 1992. Chlorinated waters discharged to the marine-environment chemistry and environmentalimpact e an overview. Science of the Total Environment 126 (1e2), 173e197. Allonier, A.S., Khalanski, M., Camel, V., Bermond, A., 1999. Characterization of chlorination by-products in cooling effluents of coastal nuclear power stations. Marine Pollution Bulletin 38 (12), 1232e1241. Anipsitakis, G.P., Dionysiou, D.D., 2003. Degradation of organic contaminants in water with sulfate radicals generated by the conjunction of peroxymonosulfate with cobalt. Environmental Science & Technology 37 (20), 4790e4797. Antonelli, M., Mezzanotte, V., Panouilleres, M., 2009. Assessment of peracetic acid disinfected effluents by microbiotests. Environmental Science & Technology 43 (17), 6579e6584. Boehm, A.B., Grant, S.B., Kim, J.H., Mowbray, S.L., McGee, C.D., Clark, C.D., Foley, D.M., Wellman, D.E., 2002. Decadal and shorter period variability of surf zone water quality at Huntington Beach, California. Environmental Science & Technology 36 (18), 3885e3892. Boehm, A.B., Fuhrman, J.A., Mrse, R.D., Grant, S.B., 2003. Tiered approach for identification of a human fecal pollution source at a recreational beach: case study at Avalon Cay, Catalina Island, California. Environmental Science & Technology 37 (4), 673e680. Boehm, A.B., Shellenbarger, G.G., Paytan, A., 2004. Groundwater discharge: potential association with fecal indicator bacteria in the surf zone. Environmental Science & Technology 38 (13), 3558e3566. Boehm, A.B., Ashbolt, N.J., Colford Jr., J.M., Dunbar, L.E., Fleming, L.E., Gold, M.A., Hansel, J.A., Hunter, P.R., Ichida, A.M., McGee, C.D., 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 & Technology 43 (21), 8046e8052. Booth, R., Lester, J., 1995. The potential formation of halogenated by-products during peracetic acid treatment of final sewage effluent. Water Research 29 (7), 1793e1801. Burnett, W.C., Bokuniewicz, H., Huettel, M., Moore, W.S., Taniguchi, M., 2003a. Groundwater and pore water inputs to the coastal zone. Biogeochemistry 66 (1), 3e33. Burnett, W.C., Bokuniewicz, H., Huettel, M., Moore, W.S., Taniguchi, M., 2003b. Groundwater and pore water inputs to the coastal zone. Biogeochemistry 66 (1e2), 3e33. Chick, H., 1908. An investigation of the laws of disinfection. Journal of Hygiene 8 (01), 92e158. Colford Jr., J.M., Wade, T.J., Schiff, K.C., Wright, C.C., Griffith, J.F., Sandhu, S.K., Burns, S., Sobsey, M., Lovelace, G., Weisberg, S.B., 2007. Water quality indicators and the risk of illness at beaches with nonpoint sources of fecal contamination. Epidemiology 18 (1), 27. Cristiani, P., 2005. Solutions to fouling in power station condensers. Applied Thermal Engineering 25 (16), 2630e2640. de Lafontaine, Y., Despatie, S.P., Wiley, C., 2008. Effectiveness and potential toxicological impact of the PERACLEAN (R) ocean ballast water treatment technology. Ecotoxicology and Environmental Safety 71 (2), 355e369. Deleersnijder, E., Campin, J.M., Delhez, E.J.M., 2001. The concept of age in marine modelling I. Theory and preliminary model results. Journal of Marine Systems 28 (3e4), 229e267. Dell’Erba, A., Falsanisi, D., Liberti, L., Notarnicola, M., Santoro, D., 2004. Disinfecting behaviour of peracetic acid for municipal wastewater reuse. Desalination 168, 435e442. Dell’Erba, A., Falsanisi, D., Liberti, L., Notarnicola, M., Santoro, D., 2007. Disinfection by-products formation during wastewater disinfection with peracetic acid1. Desalination 215 (1e3), 177e186. Grant, S.B., Sanders, B.F., 2010. Beach boundary layer: a framework for addressing recreational water quality impairment at enclosed beaches. Environmental Science & Technology, 4818e4824. Guo, G.M., Lin, F.K., 2009. The bromination kinetics of phenolic compounds in aqueous solution. Journal of Hazardous Materials 170 (2e3), 645e651. Hom, L.W., 1972. Kinetics of chlorine disinfection in an ecosystem. Journal of the Sanitary Engineering Division 98 (1), 183e194. Kitis, M., 2004. Disinfection of wastewater with peracetic acid: a review. Environment International 30 (1), 47e55. Lazarova, V., Janex, M., Fiksdal, L., Oberg, C., Barcina, I., Pommepuy, M., 1998. Advanced wastewater disinfection technologies: short and long term efficiency. Water Science and Technology 38 (12), 109e117. Lefevre, F., Audic, J., Ferrand, F., 1992. Peracetic acid disinfection of secondary effluents discharged off coastal seawater. Water Science and Technology 25 (12), 155e164. Levenspiel, O., 1980. The monod equation e a revisit and a generalization to product inhibition situations. Biotechnology and Bioengineering 22 (8), 1671e1687. Liberti, L., Lopez, A., Notarnicola, M., 1999. Disinfection with peracetic acid for domestic sewage re use in agriculture. Water and Environment Journal 13 (4), 262e269. Lipp, E.K., Farrah, S.A., Rose, J.B., 2001. Assessment and impact of microbial fecal pollution and human enteric pathogens in a coastal community. Marine Pollution Bulletin 42 (4), 286e293. Litton, R.M., Ahn, J.H., Sercu, B., Holden, P.A., Sedlak, D.L., Grant, S. B., 2010. Evaluation of chemical, molecular, and traditional markers of fecal contamination in an effluent dominated urban stream. Environmental Science & Technology.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Manheim, F.T., Krantz, D.E., Bratton, J.F., 2004. Studying ground water under Delmarva coastal bays using electrical resistivity. Ground Water 42 (7), 1052e1068. McIlvin, M.R., Altabet, M.A., 2005. Chemical conversion of nitrate and nitrite to nitrous oxide for nitrogen and oxygen isotopic analysis in freshwater and seawater. Analytical Chemistry 77 (17), 5589e5595. Michael, H.A., Mulligan, A.E., Harvey, C.F., 2005. Seasonal oscillations in water exchange between aquifers and the coastal ocean. Nature 436 (7054), 1145e1148. Mika, K.B., Imamura, G., Chang, C., Conway, V., Fernandez, G., Griffith, J.F., Kampalath, R.A., Lee, C.M., Lin, C.C., Moreno, R., Thompson, S., Whitman, R.L., Jay, J.A., 2009. Pilot- and benchscale testing of faecal indicator bacteria survival in marine beach sand near point sources. Journal of Applied Microbiology 107 (1), 72e84. Monarca, S., Feretti, D., Collivignarelli, C., Guzzella, L., Zerbini, I., Bertanza, G., Pedrazzani, R., 2000. The influence of different disinfectants on mutagenicity and toxicity of urban wastewater. Water Research 34 (17), 4261e4269. Monsen, N.E., Cloern, J.E., Lucas, L.V., Monismith, S.G., 2002. A comment on the use of flushing time, residence time, and age as transport time scales. Limnology and Oceanography 47 (5), 1545e1553. Moore, W.S., 1999. The subterranean estuary: a reaction zone of ground water and sea water. Marine Chemistry 65 (1e2), 111e125. Nielsen, P., 1990. Tidal dynamics of the water-table in beaches. Water Resources Research 26 (9), 2127e2134. Oliveira, A., Baptista, A.M., 1997. Diagnostic modeling of residence times in estuaries. Water Resources Research 33 (8), 1935e1946. ˆ n, I., 1995. Sanchez-Ruiz, C., MartI`nez-Royano, S., Tejero-MonzU An evaluation of the efficiency and impact of raw wastewater disinfection with peracetic acid prior to ocean discharge. Water Science and Technology 32 (7), 159e166. Santoro, D., Gehr, R., Bartrand, T.A., Liberti, L., Notarnicola, M., Dell’Erba, A., Falsanisi, D., Haas, C.N., 2007. Wastewater disinfection by peracetic acid: assessment of models for
5653
tracking residual measurements and inactivation. Water Environment Research 79 (7), 775e787. Santos, I.R., Niencheski, F., Burnett, W., Peterson, R., Chanton, J., Andrade, C.F.F., Milani, I.B., Schmidt, A., Knoeller, K., 2008. Tracing anthropogenically driven groundwater discharge into a coastal lagoon from southern Brazil. Journal of Hydrology 353 (3e4), 275e293. Schoen, M.E., Soller, J.A., Ashbolt, N.J., 2011. Evaluating the importance of faecal sources in human-impacted waters. Water Research. Stampi, S., De Luca, G., Onorato, M., Ambrogiani, E., Zanetti, F., 2002. Peracetic acid as an alternative wastewater disinfectant to chlorine dioxide. Journal of Applied Microbiology 93 (5), 725e731. Swarzenski, P., Burnett, W., Greenwood, W., Herut, B., Peterson, R., Dimova, N., Shalem, Y., Yechieli, Y., Weinstein, Y., 2006. Combined time-series resistivity and geochemical tracer techniques to examine submarine groundwater discharge at Dor Beach, Israel. Geophysical Research Letters 33 (24), L24405. Taniguchi, M., Burnett, W.C., Cable, J.E., Turner, J.V., 2002. Investigation of submarine groundwater discharge. Hydrological Processes 16 (11), 2115e2129. Taylor, B.N., Kuyatt, C.E., Standards N.I.o. and Technology, 1993. Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, Citeseer. USEPA, 2003. Guidelines Establishing Test Procedures for the Analysis of Pollutants; Analytical Methods for Biological Pollutants in Ambient Water, Final Rule US Federal Register, 40 CFR 68, 139, FRL-7529-7. Wagner, M., Brumelis, D., Gehr, R., 2002. Disinfection of wastewater by hydrogen peroxide or peracetic acid: development of procedures for measurement of residual disinfectant and application to a physicochemically treated municipal effluent. Water Environment Research 74 (1), 33e50. Watch F.a.W., 2008. Overloaded and Underfunded: Northwestern California’s Aging Infrstructure Needs A Clean Water Trust Fund. Food and Water Watch.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
In situ disinfection of sewage contaminated shallow groundwater: A feasibility study Morgan M. Bailey a, William J. Cooper b, Stanley B. Grant a,b,* a
Department of Chemical Engineering and Material Sciences, University of California, Henry Samueli School of Engineering, University of California, Irvine, CA 92697, USA b Department of Civil and Environmental Engineering, and Urban Water Research Center, Henry Samueli School of Engineering, University of California, Irvine, CA 92697, USA
article info
abstract
Article history:
Sewage-contaminated shallow groundwater is a potential cause of beach closures and
Received 18 June 2011
water quality impairment in marine coastal communities. In this study we set out to
Received in revised form
evaluate the feasibility of several strategies for disinfecting sewage-contaminated shallow
9 August 2011
groundwater before it reaches the coastline. The disinfection rates of Escherichia coli (EC)
Accepted 14 August 2011
and enterococci bacteria (ENT) were measured in mixtures of raw sewage and brackish
Available online 22 August 2011
shallow groundwater collected from a coastal community in southern California. Different disinfection strategies were explored, ranging from benign (aeration alone, and aeration
Keywords:
with addition of brine) to aggressive (chemical disinfectants peracetic acid (PAA) or per-
Disinfection
oxymonosulfate (Oxone)). Aeration alone and aeration with brine did not significantly
Shallow groundwater
reduce the concentration of EC and ENT after 6 h of exposure, while 4e5 mg L1 of PAA or
Sewage
Oxone achieved >3 log reduction after 15 min of exposure. Oxone disinfection was more
Remediation
rapid at higher salinities, most likely due to the formation of secondary oxidants (e.g.,
Water quality
bromine and chlorine) that make this disinfectant inappropriate for marine applications.
Peroxymonosulfate
Using a Lagrangian modeling framework, we identify several factors that could influence
Peracetic acid
the performance of in-situ disinfection with PAA, including the potential for bacterial
Oxone
regrowth, and the non-linear dependence of disinfection rate upon the residence time of
Brine
water in the shallow groundwater. The data and analysis presented in this paper provide a framework for evaluating the feasibility of in-situ disinfection of shallow groundwater, and elucidate several topics that warrant further investigation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Beach closures and advisories are generally caused by the coastal discharge of land-side sources of fecal pollution such as urban runoff or sewage effluent (Boehm et al., 2002; Grant and Sanders, 2010). However, contamination of shallow groundwater by aging sewage collection systems and failing onsite sewage treatment and disposal systems has also been
identified as a source of fecal pollution at some coastal beaches (Lipp et al., 2001). The latter typically involves a two-step process: (1) shallow groundwater is first contaminated by the surface or subsurface release of sewage, and (2) the shallow groundwater is then discharged to coastal waters under the influence of land hydraulic gradients, tidal pumping, and/or current induced pressure gradients (Burnett et al., 2003b). The coastal discharge of sewage-contaminated shallow
* Corresponding author. Tel.: þ1 949 824 8277; fax: þ1 949 824 2541. E-mail addresses: [email protected] (M.M. Bailey), [email protected] (S.B. Grant). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.020
5642
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
groundwater can impact near-shore water quality by vectoring fecal-oral pathogens (or their indicators) directly from sources of human sewage to receiving waters (Boehm et al., 2003). The discharge may also deliver nutrients that promote the survival, and perhaps growth, of fecal bacteria in near-shore waters (Boehm et al., 2004). To date, most studies on shallow groundwater discharge focus on its quantification using isotopic (McIlvin and Altabet, 2005), geophysical (Manheim et al., 2004; Santos et al., 2008; Swarzenski et al., 2006), or hydrological methods (Burnett et al., 2003a; Michael et al., 2005; Taniguchi et al., 2002), but to our knowledge there are no published reports on the very practical and important question of how to remediate shallow groundwater once it becomes contaminated with sewage. Given the advancing age and poor condition of sewage collection systems and onsite sewage treatment and disposal systems in many coastal areas of the U.S. (Food and Water Watch, 2008), a cost-effective strategy for remediating sewage-derived fecal pollution in brackish shallow groundwater is urgently needed for the sustainable use of coastal beaches and bays. While the best remediation strategy should always be to identify and eliminate the source(s) of sewage responsible for groundwater contamination, logistical and economic constraints may make such action infeasible in the near term, in which case interim solutions are needed. Further, even if a source of sewage contamination is identified and repaired, it may take some time before pathogens and indicator bacteria decay to acceptable concentrations, particularly when the sediments in question are subsurface and thus not exposed to sunlight (Mika et al., 2009). Subsurface injection of disinfectant (or “in-situ disinfection”) is one obvious interim strategy for remediating sewage-contaminated shallow groundwater, but several issues need to be resolved before such an approach can be implemented in practice. (1) Disinfectants can react with organic matter and/or trace anions (e.g., bromide and chloride ions) in brackish coastal groundwater to produce toxic disinfection by-products, and their discharge to coastal zones could pose a health risk to bathers along the shoreline and negatively impact sensitive marine ecosystems (Monarca et al., 2000). (2) The shallow groundwater adjacent to marine coastal areas can be thought of as “subterranean estuaries”, in which fresh (meteoric) waters mix with seawater before discharging to the ocean (Moore, 1999). Consequently, the flow in shallow groundwater can vary substantially in space and time, potentially affecting the disinfection contact times water parcels experience before discharging to coastal waters. (3) Sewage constituents, such as pathogens and their indicators, may not mix fully over the vertical dimension of the shallow groundwater, and therefore disinfection strategies may need to be tailored to target sewage constituents where they are located, for example near the top of the water table. (4) Planning and implementation of in-situ disinfection would greatly benefit from quantitative “design criteria” that account for the rate at which the target organisms decay with time upon exposure to disinfectant, the decay in disinfectant with time, the time scale over which water parcels reside in the shallow groundwater before discharge, and ecological processes that promote the subsurface regrowth of fecal bacteria.
In this paper we present laboratory data and modeling efforts intended to, at least preliminarily, address the issues raised above, using as a test case Avalon Bay, Catalina Island, California, where chronic fecal contamination of near-shore waters has been linked to sewage contamination of shallow groundwater by leaking sewage collection systems beneath the City (Boehm et al., 2003, 2009b). The goals of this study are to: (1) test three strategies for inactivating sewage-associated fecal bacteria from shallow groundwater, including aeration alone, aeration with addition of brine, and aeration with addition of either peracetic acid (PAA) or peroxymonosulfate (commercially known as Oxone); (2) characterize the salinity dependence and disinfection by-product formation potential of the latter two disinfectants; (3) evaluate several models for PAA disinfection; and (4) develop a Lagrangian modeling framework for evaluating, and potentially designing, in-situ disinfection of sewage-contaminated coastal shallow ground waters.
2.
Materials and methods
2.1.
Choice of disinfectant
Bench-scale experiments were carried out to measure the decay of sewage-associated fecal indicator bacteria in shallow groundwater upon exposure to: (1) aeration alone, (2) aeration with addition of brine, (3) aeration with addition of PAA (35% solution, CAS-79-21-0, Pfaltz & Bauer, CT), and (4) aeration with addition of Oxone (43% dry weight, CAS-37222-66-5, Alfa Aesar, Ward Hill, MA). The impact of aeration on bacterial dieoff was evaluated on the premise that, if efficacious, injection of ambient air into the subsurface would be a straightforward, economical, and environmentally benign approach for remediating sewage-contaminated shallow groundwater. The choice of brine as a candidate disinfectant was motivated by the fact that its injection into shallow groundwater is unlikely to cause environmental harm, and because the field site under consideration, Avalon Bay, has a desalination plant from which brine is produced as a waste product. PAA is a promising disinfectant for this application, because it produces little or no toxic disinfection by-products when mixed with seawater (it degrades into acetic acid, vinegar) (Kitis, 2004), is a strong biocide in marine waters as evidenced by its use as an antifouling agent in cooling water systems for coastal power plants, and is regarded as relatively safe for discharge to sensitive marine waters (Sanchez-Ruiz et al., 1995) including the highly regulated Italian Lagoon of Venice (Cristiani, 2005). To our knowledge, Oxone has not been tested as a biocide in coastal marine settings, but its application to this particular problem was motivated by the fact that, upon addition to water, it generates hydroxyl radicals and sulfate ions (Anipsitakis and Dionysiou, 2003); the former should accelerate the die-off of bacteria and viruses while the latter is already present at high concentrations in marine waters. Many commonly used disinfectants (e.g., ozone, chlorine gas, sodium hypochlorite) were excluded from consideration for logistical reasons and, although highly effective biocides, they would likely react with trace anions and organics in sewagecontaminated brackish groundwater to form toxic
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
disinfection by-products (Abarnou and Miossec, 1992; Allonier et al., 1999).
2.2.
Experimental characterization of disinfection kinetics
Raw (untreated) sewage was diluted 1:100 using various mixtures (with various final salinities) of three source waters: (1) relatively fresh (salinity ca. 1) water from an inland groundwater well, (2) Avalon Bay water (salinity ca. 32), and (3) desalination plant brine (salinity ca. 45). All source waters were aseptically collected from the City of Avalon using sterile polypropylene bottles and mixed with raw sewage collected from influent to the City of Avalon wastewater treatment plant. Source waters were not sterilized prior to mixing with sewage, and therefore bacteria populations present in the final mixtures of sewage and source waters probably included (minor) contributions from the latter. Bench-scale disinfection experiments commenced within 6 h of sample collection and were carried out in 4 L Nalgene polypropylene reaction vessels maintained in the dark and partially submerged in a temperature controlled and recirculating water bath at 15 1 C. Contents of the reactor vessel were continuously aerated using a frit through which ambient air was driven by an aquarium pump. To capture a range of salinities and disinfectant concentrations, a matrix design was adopted in which over 45 separate disinfection experiments (including 10 no-disinfectant-added control experiments) were conducted at five different disinfectant concentrations (ranging from 0 to 7.3 mg L1 for Oxone, and 0e8 mg L1 for PAA) and four different salinities (0, 15, 32, 45). The concentration range adopted for PAA was based on previously published studies with this disinfectant (Kitis, 2004; Santoro et al., 2007). To our knowledge, this is the first study to evaluate Oxone disinfection in saline mixtures, and thus the concentration range adopted for this disinfectant was based on pilot studies (not reported) with this disinfectant. Each batch reactor was sampled seven times over a period of 1e6 h (depending on experiment). This sampling schedule was chosen to resolve disinfection over a single ebb tide, which represents a theoretical minimum time a fluid parcel would be in contact with disinfectant before discharging to Avalon Bay, assuming that the discharge of shallow groundwater was under tidal control. Samples were extracted from the reactor using either a sterile syringe or pipet, analyzed for pH and conductivity, quenched by addition of approximately 0.4 mL of 0.1 N sodium thiosulfate (CAS-7772-98-7, Mallinckrodt Chemicals), immediately diluted either 1:10 or 1:100 in sterile deionized water (Hardy Scientific, California), and enumerated for Escherichia coli (EC) and enterococci bacteria (ENT) using Colilert-18 and Enterolert defined substrate tests implemented in a 97-Well Quanti-Tray format (IDEXX Laboratories, Maine). The adoption of IDEXX Colilert-18 and Enterolert was motivated based on the fact that both tests are approved by the U.S. Environmental Protection Agency for enumerating EC and ENT bacteria in ambient waters (USEPA, 2003) and, more to the point, are used by the Los Angeles Department of Health Services in their routine monitoring of recreational beach water quality in Avalon Bay, and as a basis for management decisions regarding, for example, the posting of Avalon beaches as unfit for swimming. For a subset of the disinfection experiments, dissolved organic carbon (DOC) and total organic carbon (TOC) concentrations were measured: (1) on the source water (groundwater or
5643
bay water) prior to addition of sewage and disinfectant; (2) on the source water plus sewage prior to the addition of disinfectant; and (3) at the end of the disinfection experiment. Samples analyzed for DOC and TOC were collected in 500 mL amber glass bottles and acidified with 2 mL of hydrochloric acid, and then analyzed by TestAmerica using Standard Method 5310B.
2.3.
Disinfection rate constants
Two different disinfection rate constants are reported in this study: (1) an effective first-order disinfection rate constant k0d , and (2) an intrinsic disinfection rate constant kd. Values of the effective first-order rate constant k0d were estimated by regressing log-transformed bacteria concentration against time based on Chick’s Law (Chick, 1908): NðtÞ ¼ k0d t ln N0
(1)
In Eq. (1), N(t) represents the concentration of bacteria in the reaction vessel at any time t and N0 represents the bacterial concentration present in the mixture at the start of the disinfection experiment. The regression excluded any bacterial measurements in the initial lag period and bacterial measurements that fell below the lower-limit of detection for the Colilert and Enterolert assays (in most cases, the lower limit of detection was 10 most probable number (MPN) per 100 mL of sample). In the case of PAA disinfection, an intrinsic disinfection rate constant was also calculated, which represents the susceptibility of bacteria to disinfectant independent of disinfectant concentration: kd ¼ k0d =C0
(2)
The variable C0 represents the initial (molar) concentration of peroxycompounds which, in the case of commercial preparations of PAA, includes both PAA and hydrogen peroxide (Wagner et al., 2002).
2.4.
Activation energy for PAA disinfection
Using the reactor set-up described above, four separate disinfection experiments were carried out at five temperatures (T ¼ 5, 10, 15, 20 and 30 C) and a fixed concentration of PAA (6 mg L1) to determine the temperature sensitivity of EC and ENT disinfection by PAA. Intrinsic disinfection rate constants kd calculated from these experiments (see Section 2.3) were used to determine activation energies for the PAA disinfection of EC and ENT, based on an Arrhenius plot of ln(kd) against 1/T.
3.
Disinfection results and discussion
3.1.
Measurements of pH, TOC, and DOC
Over aeration of the reactor vessel caused the pH to increase slightly from 7.9 0.2 to 8.3 0.2. No other systematic pH trends were observed across the different experiments and different source waters tested. The average DOC concentration in groundwater ranged from 1.6 0.14 mg L1 (N ¼ 7,
5644
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
foreshore well, salinity 20) to 2.9 0.2 mg L1 (N ¼ 4, inland well, salinity 1). The TOC/DOC ratio was near unity in water collected from the inland well (1.1 0.1, N ¼ 4) implying that most of the organic carbon was dissolved. Raw sewage collected from the Avalon Sewage Treatment Plant had much higher levels of DOC (25 mg L1, N ¼ 1). However, sewage contributed only background levels of DOC (<0.25 mg L1) to the final 1:100 mixtures of sewage and source waters used in the disinfection experiments. For disinfection experiments carried out with PAA, the largest source of DOC was frequently the PAA solution itself, which contained upwards of 280 mg L1 of carbon as acetic acid. The acetic acid was present as an equilibrium component of the commercial PAA preparation, and generated as a breakdown product of PAA.
3.2. Experimental characterization of disinfection kinetics First-order effective disinfection coefficients ðk0d Þ estimated from the 45 separate disinfection experiments are contoured using Delaunay triangulation (Igor Pro v 6.10, Lake Oswego, Oregon) against disinfectant concentration and salinity in Fig. 1. These results are described in the sections below. A more detailed analysis of k0d values, and an evaluation of different disinfection models, is presented later in the paper (Section 4).
3.2.1.
Aeration alone and aeration with addition of brine
We started by evaluating two environmental benign approaches for removing fecal indicator bacteria from sewagecontaminated shallow groundwater; namely, injection of ambient air and/or injection of brine produced from a local
desalination plant. Aeration alone and aeration with brine did not significantly reduce EC and ENT concentrations after 6 h of exposure; i.e., within the resolution of these experiments k0d ¼ 0. Thus these two strategies are unlikely to be effective against indicator bacteria in shallow groundwater over the time scale of a single ebb tide.
3.2.2.
Aeration with addition of PAA and oxone
EC and ENT concentrations were reduced by more than 1000 fold (3 log units) after 15 min of exposure to 4e5 mg L1 of PAA. The effective disinfection rate constants calculated for these experiments (Fig. 1A and B): (1) increased monotonically with 1 PAA dose up to the maximum rate ðk0d ¼ 0:2 min Þ resolvable with our experimental set-up; (2) do not depend, at least not dramatically, on the salinity of the disinfection mixture; and (3) are larger for EC than ENT at a fixed PAA dose. The last observation is consistent with the results reported in Stampi et al. (2002). Averaged across all experiments (and excluding any experiments where the effective disinfection rate constant exceeded 0.2 min1), the average intrinsic rate ¼ 2:3 0:57 and constants for PAA disinfection are kENT d 1 1 EC kEC d ¼ 3:7 0:2 mM min : The estimate for kd is similar to intrinsic rate constants estimated from previously published data for PAA disinfection of EC in secondary settled effluent (Dell’Erba et al., 2004) and PAA disinfection of fecal coliform (FC) in secondary-treated sewage effluent (Wagner et al., 2002) (Table 1). The fact that the intrinsic rate constants for PAA disinfection of EC and FC are similar across these three studies is notable, given that EC is a subset of FC, and the very different initial bacterial concentrations (and presumably sewage content) associated with the different source waters
Fig. 1 e Contour plots of the effective first-order rate constants k0d [minL1] for PAA disinfection of EC (panel A) or ENT (panel B), and Oxone disinfection of EC (panel C) or ENT (panel D).
5645
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Table 1 e Intrinsic disinfection rates and parameters from multiple studies. Value kd (mM1 min1)a
Source FC
EC
%H2 O2=%PAAb
Temp ( C)
N0 c
Solutiond
GW, BW, Br, TPI TPE TPE
ENT
This study
NA
3.7 0.2
2.3 0.57
0.183
15
104
Dell’Erba et al. Wagner et al.
NA 1.04 1.26
3.65 2.41 NA
NA NA
1.53 1.54
NR NR
102 105
NR-not reported. NA-not applicable. a Intrinsic disinfection rate. b Ratio of percent mass of PAA and hydrogen peroxide in PAA equilibrium solution. c Magnitude of initial bacteria concentration. d Sources waters for disinfection studies (GW-groundwater, BW-way water, Br-brine, TPI-treatment plant influent, TPE-treatment plant effluent).
used in these studies, ranging from 102 to 105 bacteria per 100 mL (Table 1). EC and ENT concentrations were reduced by more than 3 log units after 15 min of exposure to 4 mg L1 Oxone. However, unlike PAA, the disinfection rate depends on both Oxone dose and solution salinity (Fig. 1C and D). At low Oxone dose (<2 mg L1), the effective disinfection rate increases monotonically with Oxone dose, and exhibits no obvious salinity dependence (i.e., the contour lines are near vertical in this region of the plot). At higher Oxone dose, the effective disinfection rate increased monotonically with salinity, and exhibits no obvious dose dependence (i.e., the contour lines were near horizontal in this region of the plot). One possible explanation for this pattern is that Oxone may oxidize anions in the shallow groundwater and brine to yield, for example, the secondary oxidants chlorine and bromine. To explore this idea, a set of control experiments were carried out in which either PAA or Oxone was added to an aqueous solution consisting of 35 g L1 NaCl to mimic the background salinity of seawater and varying concentrations
(0e70 mg L1) of KBr to mimic the presence of trace anions. The formation of oxidized forms of chloride and bromide ions (e.g., chlorine and bromine) was monitored by measuring absorbance of the indicator dye N,N-diethyl-p-phenylenediamine (DPD) (Standard Methods, 4500-CL G). Fig. 2A shows DPD spectra measured after addition of 10 mg L1 of either Oxone (solid lines) or PAA (dashed lines) to distilled water (Solution 1), an aqueous solution consisting of 35 g L1 NaCl (Solution 2), or an aqueous solution consisting of 35 g L1 NaCl and 70 mg L1 KBr (Solution 3). In this set of experiments, the disinfectant was allowed to react in the solution for 5 min, whereupon DPD was added, and 1 min later the DPD absorbance spectrum was measured. Referring to Fig. 2A, the absorbance of DPD increased in the order Solution 1 < Solution 2 < Solution 3, consistent with the idea that Oxone oxidizes both chloride and bromide ions to form secondary oxidants. DPD absorbance did not increase when PAA was allowed to react for 5 min in Solution 2 and increased only slightly when PAA was allowed to react for 5 min in Solution 3. These latter results are consistent with the findings
Fig. 2 e Panel A: DPD absorbance spectra measured after 10 mg LL1 of either Oxone (solid lines) or PAA (dashed lines) is allowed to react for 5 min in: Solution 1 (DI water), Solution 2 (DI water and 35 g LL1 NaCl), or Solution 3 (DI water, 35 g LL1 NaCl, and 70 mg LL1 KBr). Panel B: DPD absorbance at 515 nm after 10 mg LL1 of either Oxone (solid lines) or PAA (dashed lines) are allowed to react for 5 min in Solution 2 with KBr concentrations shown. Panel C: the kinetics of oxidant formation by 10 mg LL1 of either Oxone (solid line) or PAA (dashed line) in Solution 3.
5646
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
of Booth and Lester (1995), who report that PAA could not oxidize chloride ion to hypochlorous acid, but could oxidize bromide ion to hypobromous acid. In the presence of Oxone, DPD absorbance increased with increasing KBr concentration in Solution 2 (Fig. 1B), and the oxidation kinetics in Solution 3 are relatively rapidly (<5 min of exposure time, Fig. 1C). In the presence of PAA, on the other hand, DPD absorbance increased only slightly with increasing KBr concentration in Solution 2 (Fig. 2B), and oxidization kinetics in Solution 3 are relatively slow (>150 min exposure time) (Fig. 2C). Collectively, the results presented above are consistent with the idea that Oxone quickly oxidized trace anions in brackish shallow groundwater to form secondary oxidants that act synergistically with Oxone to enhance disinfection rates. However, the reaction of Oxone with trace anions could also increase the toxicity of the shallow groundwater, by producing compounds that are carcinogenic (e.g., bromateion) and/or by producing secondary oxidants (e.g., bromine and chlorine) that subsequently react with organic compounds to form toxic disinfection by-products (e.g., trihalomethanes) (Guo and Lin, 2009). For these reasons, and despite its obvious utility as a disinfectant, Oxone should not be used for in-situ disinfection of sewage contaminated shallow groundwater in settings, like Avalon Bay, where the groundwater is under marine influence. PAA, on the other hand, does not appear to react strongly with chloride ion, or quickly with bromide ion, and thus is less likely to produce toxic disinfection by-products; a conclusion supported by several published studies (Liberti et al., 1999; Dell’Erba et al., 2007; Kitis, 2004). Although PAA does not appear to generate the quantity and spectrum of disinfection by-products associated with many disinfectants, some researchers have raised concern about the potential environmental impact of releasing disinfected effluents that contain a PAA residual (Antonelli et al., 2009; de Lafontaine et al., 2008), which can be toxic to crustaceans and microorganisms (Antonelli et al., 2009; de Lafontaine et al., 2008). However, Lafontaine et al. (2008) note that, because PAA breaks down rapidly in seawater, any potential impacts would be localized around the region where disinfection residual is discharged to the environment. Thus, if PAA is used for in-situ disinfection of sewage contaminated shallow groundwater, care should be taken to minimize the PAA residual discharged to coastal waters.
3.3.
Activation energy for PAA disinfection
Based on measurement of intrinsic disinfection rates over a range of temperatures, from 5 to 30 C, the following activation energies were estimated for PAA disinfection of EC and ENT: 37.5 7.8 kJ mol1 and 38.0 9.3 kJ mol1, respectively. These activation energies are used later in the paper to estimate intrinsic disinfection rates for PAA over a range of temperatures relevant to the shallow groundwater in Avalon Bay (see Section 5.1).
4.
Modeling PAA disinfection kinetics
Of the strategies evaluated above, PAA disinfection appears the most viable, given that it is both a potent biocide and less likely to form toxic disinfection by-products. In this section we
evaluate several published models for PAA disinfection, all of which are special cases of Hom’s Law (Hom, 1972): dN ¼ kd Ntm Cn dt
(3)
where, N, C, kd, and t represent bacterial concentration, disinfectant concentration, intrinsic disinfection rate constant, and time, respectively. Chick’s Law (Eqs. (1) And (2) of this paper) corresponds to a choice of exponent values m ¼ 0, n ¼ 1, and a constant disinfectant concentration, C ¼ C0. Wagner et al. (2002) suggested that, during the disinfection process, the concentration of peroxycompounds in commercial PAA mixtures decays with time in accordance with the following second-order rate law: dC ¼ k C2 dt
(4)
The rate constant k* depends on the initial peroxycompound concentration C0 (Wagner et al., 2002): k ¼ aCb 0
(5)
Where the pre-factor and power-law exponents are given by a ¼ 0.0093 mM(b 1) min1 and b ¼ 1.420. In their analysis of PAA disinfection, Santoro et al. (2007) suggest that peroxycompounds exhibit zero-order decay; however, zero-order decay models predict negative concentration in finite time, and therefore will not be considered further here. Combining Eqs. (4) and (5), and solving the differential equation, yields the following prediction for disinfectant concentration as a function of time: CðtÞ ¼
1 1=C0 þ k t
(6)
Given this time-dependence for the disinfectant concentration, Hom’s Law becomes: dN n ¼ kd Ntm ½1=C0 þ k t dt
(7)
Wagner et al. (2002) solved Eq. (7) for two different choices of the exponents n and m: (1) n ¼ 1 and m ¼ 0 (no-tailing model), and (2) n ¼ 1 and m ¼ 1 (tailing model). The tailing model provided a better empirical fit to the PAA disinfection data, although their intrinsic disinfection rate constant kd varied with the initial disinfectant concentration, and a new fitting parameter (the time t at which N(t) ¼ N0) was introduced. Here we opt for the more parsimonious no-tailing model, for which an exact solution can be derived: kd =k
N ¼ N0 ðk C0 t þ 1Þ
(8)
PAA disinfection data collected in this study were tested against Chick’s Model (Eq. (1)) and the no-tailing model (Eq. (8)) in Fig. 3. The data are plotted so that the intrinsic disinfection rate constant kd can be estimated directly from the slope b of the best-fit line: kd ¼ 2.303b (Chick’s Law) or kd ¼ b (no-tailing model). Also shown in Fig. 3 are two estimates of model performance, including the Pearson’s r2 correlation between logN/N0 and the x-axis (either C0t or log½k C0 t þ 1=k ), and the root mean square error (RMSE) between modeled and measured values of bacterial log reduction. In general, for a given choice of fecal bacteria group (either ENT or EC) the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
5647
Fig. 3 e PAA disinfection data for EC (panel A and B) and ENT (panel C and D) fit to either Chick’s model (Eq. (1), panel A and C) or Wagner et al.’s no-tailing model (Eq. (8), panel B and D). Values for kd are mML1 minL1 and units for RMSE are logN/N0.
two disinfection models have similar r2 and RMSE values (compare panels A and C with panels B and D in Fig. 3). Both models are a better description of ENT disinfection (r2 ¼ 0.68 to 0.72, RMSE ¼ 0.32e0.37) than EC disinfection (r2 ¼ 0.52 to 0.55, RMSE ¼ 0.6e0.62). The fact that ENT exhibited less dispersion around the no-tailing model may reflect less variability (compared to EC) in the disinfection resistance of enterococci bacteria populations present in the sewage and source waters. Intrinsic rate constants estimated from the slope b are also similar for the two models. Chick’s Law yields intrinsic rate constants for EC and ENT of 3.0 and 1.9 mM1 min1, respectively (panels A and C). The no-tailing model yields intrinsic rate constants for EC and ENT of 3.3 and 3.5 mM1 min1, respectively (panels B and D). The intrinsic disinfection rate for EC is also similar to values estimated by averaging rate constants obtained from our individual experiments (see Section 3), and from data reported in other studies of PAA disinfection (Table 1). In summary, Chick’s Law and the notailing model were both reasonably good predictors of bacterial decay caused by PAA disinfection, and both yield values of
the intrinsic disinfection rate constant kd that were consistent across models, and across studies. Relative to the modeling effort described in the next section, the primary benefit of the no-tailing model is that it accounts for the decay in disinfectant concentration that will inevitably occur following injection of PAA into the subsurface.
5.
Lagrangian model of in situ disinfection
In this section we develop a quantitative model that accounts for the physical, chemical, and biological factors that might influence the in-situ disinfection of sewage contaminated shallow groundwater with PAA. The model is developed in two stages. First, a Lagrangian framework is used to predict the concentration of both sewage constituents (fecal indicator bacteria) and PAA residual in a parcel of shallow groundwater as it travels from the point of injection to the point where it is discharged to the coastal ocean. Second, an analytical model is derived for fluid parcel residence time in shallow
5648
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
groundwater, for the situation (relevant to Avalon Bay) where shallow groundwater discharge is dominated by tidal pumping and meteoric flow.
5.1. Lagrangian framework for modeling disinfection kinetics Fig. 4 illustrates the physical transport processes that might affect fecal bacteria concentration in a parcel of groundwater as it moves from the point where it first encounters subsurface injected disinfectant (point A) to the point where it exits the shallow groundwater and is discharged to the coastal ocean (point B). As the parcel moves from A to B, the concentration of bacteria in the fluid parcel changes with time due to disinfection, non-disinfection related die-off, regrowth, and attachment to the porous matrix (filtration). Mass balance over a fluid parcel as it travels from A to B yields the following rate expression for the concentration of bacteria (N, bacteria L3), where s [min1] represents the time a sewage contaminated water parcel has been in contact with disinfectant (referred to here as “residence time”), C is the disinfectant concentration [mM], and the rate constants for disinfection, inactivation, filtration, and growth are kd [mM1 min1], ki [min1], kf [min1], and mg [min1]: dN 1 ¼ kd N½1=C0 þ k s ki þ kf N þ mg N ds
(9)
In formulating Eq. (9), we adopted the no-tailing version of Hom’s Law described earlier (Eq. (7) with m ¼ 0 and n ¼ 1) and assume that inactivation, filtration, and regrowth of bacteria all follow first-order kinetics. While we did not
where, mg,max and Ks represent the maximum growth rate and saturation constant, respectively. Based on the measurements of DOC presented in Section 3.1, potential sources of DOC include ambient groundwater (DOCGW), sewage (DOCSewage), and acetic acid associated with the PAA mixture, including acetic acid present as an equilibrium component of commercial PAA preparations (DOCAA,0), and acetic acid formed by the decomposition of PAA during disinfection (DOCA(s)), where the latter can be estimated from the loss of peroxycompound concentration with time: DOCAA ðsÞ ¼ 0:82MðC0 CðsÞÞ
(11)
Eq. (11) assumes stoichiometric conversion of PAA to acetic acid, taking into account the molar fraction (0.82) of the peroxycompound concentration that is PAA and the weight of carbon associated with every mole of acetic acid, M ¼ 25 g of carbon per mole. Combining Eqs. (10) and (11), we have the following prediction for the total DOC available for growth of fecal indicator bacteria: DOCT ðsÞ ¼ DOCGW þ DOCSewage þ DOCAA;0 þ 0:82MðC0 CðsÞÞ:
(12)
Combining Eqs. (9)e(12) and solving the resulting differential equation, yields the following formula for the concentration of bacteria in a parcel of shallow groundwater as a function of residence time (s):
kf ðC0 M þ Ks Þ þ C0 Mki C0 Mmg;max þ DOCT0 kf þ ki mg;max þ ki Ks N0 ðDOCT0 þ Ks Þ exp s C0 M þ DOCT0 þ Ks a
NðsÞ ¼
(e.g., Lazarova et al., 1998; Lefevre et al., 1992). Here we use the Monod equation (Levenspiel, 1980) to model the dependence of growth rate mg on dissolved organic carbon (DOCT): mg;max DOCT ðsÞ (10) mg ðsÞ ¼ Ks þ DOCT ðsÞ
kd =k
ðC0 k s þ 1Þ
½C0 k sðMC0 þ DOCT0 þ Ks Þ þ DOCT0 þ Ks
observe regrowth in the disinfection experiments presented earlier, it is a well-known that the acetic acid in commercial PAA mixtures can serve as a carbon source for the growth of heterotrophic bacteria, including fecal indicator bacteria
a¼
a
Ks mg M k ðC0 M þ DOCT0 þ Ks Þ2
(13a)
(13b)
Fig. 4 e A conceptual model for the in-situ disinfection of sewage-contaminated shallow groundwater in coastal marine environments.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Fig. 5A presents the bacterial log reduction predicted by Eq. (13) (solid curves) as a function of initial PAA concentration (vertical axis) and residence time (horizontal axis). This graph was generated using the parameter values listed in Table 2, which are based on the experimental measurements presented earlier and field conditions relevant to the Avalon Bay field site, including a shallow groundwater temperature of 15 C. For the range of peroxycompound concentrations considered in Fig. 5A, the dependence of bacteria concentration on residence time exhibits two patterns. At small residence times, very little of the PAA has been converted to acetic acid, disinfection dominates, and bacteria concentration declines rapidly with increasing residence time. At longer residence times, PAA has been mostly converted to acetic acid, regrowth of bacteria dominates, and bacterial concentration increases with increasing residence time. This decay/ regrowth pattern is illustrated for a single initial peroxycompound concentration (C0 ¼ 0.03 mM) and temperature (T ¼ 15 C) in Fig. 5B (dotted line). For this particular choice of parameter values, the model predicts that bacterial concentration falls approximately 2 log units within 12 h, and increases thereafter, eventually rising above the initial bacteria concentration at a residence time of around two days. As expected, bacteria removal increases monotonically with increasing initial peroxycompound concentration, as illustrated for a fixed residence time (s ¼ 5 days) and temperature (T ¼ 15 C) in Fig. 5C (dotted line). Using the activation energies for the intrinsic disinfection rate constant reported in Section 3.3, the model predicts similar trends over the range of temperatures (13e17 C) typically measured in Avalon shallow groundwater (compare the family of curves in Fig. 5B and C).
5649
In an ideal scenario, by the time a fluid parcel of shallow groundwater is discharged to the ocean, both its bacterial concentration and PAA residual would be very small. The dependence of PAA residual on initial peroxycompound concentration and residence time can be estimated from Eq. (6), by setting the left hand side equal to a fixed residual concentration, Cres and solving for residence time s: sres ¼
b 1 1 C0 C C1 0 a res
(14)
Curves of constant peroxycompound residual are plotted in Fig. 5A (dashed curves). As expected, peroxycompound residual declines with increasing residence time for a fixed C0. Interestingly, the curve for Cres ¼ 1 mM roughly demarcates the transition from bacteria disinfection to regrowth (compare solid and dotted lines in Fig. 5A).
5.2.
Residence time of fluid parcels
The Lagrangian model presented above reveals that both the bacterial concentration and PAA residual were sensitive to the residence time of water parcels in shallow groundwater, and in this section we describe some of the physical processes that can affect this key parameter. Here, residence time has precisely the same meaning as in estuarine systems: “how long a parcel, starting from a specified location within a waterbody, will remain in the waterbody before exiting” (Monsen et al., 2002). Provided that material diffusion can be neglected (a key assumption in the Lagrangian approach adopted here) (Deleersnijder et al., 2001), the residence time
Fig. 5 e Model predictions for EC concentration and PAA residual for the in-situ disinfection of sewage contaminated shallow groundwater with PAA. Panel A: Curves of constant EC log reduction predicted by Eq. (13) (solid lines labeled with numbers ranging from L1 to L9), and curves of constant peroxycompound residual predicted by Eq. (14) (dashed curves, labeled with numbers ranging from 0.2 to 1 mM). Panel B: Change in EC concentration with residence time predicted by Eq. (13) for C0 [ 0.03 mM and the ambient groundwater temperatures shown. (C) Change in EC concentration with increasing initial PAA concentration predicted by Eq. (13) for a fixed residence time of s [ 5 days and the ambient groundwater temperatures shown. Parameter values used to generate these curves are listed in Table 2.
5650
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Table 2 e Inputs parameters used to produce the disinfection contour plot (Fig. 5), residence time model (Fig. 6). Constant a b kd ki kf mg Ks DOCT0 M A Kp l
Name
Value (b 1)
k* fitting parameter k* fitting parameter EC disinfection rate (T ¼ 13,15,17 C) EC natural decay rate EC filtration rate Max growth rate Growth rate saturation constant Initial DOC Concentration Mass of carbon per mole PAA Wave amplitude Hydraulic conductivity Wave number
0.0093 mM min 1.42 3.3, 3.7, 4.1 mM1 min1 0.1 h1 0 min1 0.005 min1 0.07 mg L1 25 g 1m 5 104 m s1 0.05 m1
associated with the movement of a fluid parcel from point A to B in Fig. 4 can be written explicitly as follows: ZL sc ¼
dx v
(15)
0
where, v is the velocity experienced by the fluid particle as it moves toward the ocean. Here we have subscripted the residence time calculated from Eq. (15) (sc) to distinguish it from the actual residence time of a water parcel included in our disinfection model above (s). Fluid parcel velocity has contributions from tidal pumping (vt), wave set-up (vw), meteoric groundwater flow (vm), seasonal evapotranspiration (vs), and density driven flow (vd) (Burnett et al., 2003b; Michael et al., 2005): v ¼ vt þ vw þ vm þ vs þ vd
(16)
Given that tidal pumping appears to dominate the discharge of shallow groundwater to Avalon Bay (Boehm et al., 2009b), here we focus on the tidal pumping and meteoric terms in Eq. (16), vt and vm. Flow fields generated by tidal pumping can be approximated from the following expression (Nielsen, 1990):
vt ðx; tÞ ¼ Kp Alelx ðcos½ut lx sin½ut lxÞ
Source 1
(17)
Nielsen derived Eq. (17) from the Boussenesq equation after invoking a number of simplifying assumptions, including a homogeneous unconfined aquifer of hydraulic conductivity KP, a vertical beach face, and a one-dimensional flow field (parallel to the x-axis, see Fig. 4) characterized by a wave number l, and forced by a single harmonic tide with amplitude A and angular frequency u. The wave number is defined as l ¼ 2p/x0, where x0 represents the inland distance over which tidal fluctuations in the shallow groundwater are significant. Because the flow field predicted by Eq. (17) is tidally periodic and spatially variable, the residence time of a fluid parcel will depend not only on where in the aquifer it is released (referred to here as the setback distance, x ¼ L, see Fig. 4) but also on when in the tidal cycle that release occurs; a very similar phenomenon has been described for the residence time distributions in coastal estuaries (Monsen et al., 2002; Oliveira and Baptista, 1997). Despite these complications, a characteristic residence time can be estimated from Eqs. (15)e(17) by releasing the fluid parcel at the precise moment when the recessional tide wave (associated
Wagner et al. Wagner et al. Measured Estimated Estimated Surbeck et al. Surbeck et al. Measured Calculated Estimated Calculated via KozenyeCarman Estimated
with the falling tide) passes point A in Fig. 4, which is mathematically equivalent to assigning the value p to the quantity (ut lx) in Eq. (17). After invoking this simplification and allowing for the possibility of non-zero meteoric flow, Eqs. (15)e(17) can be combined to yield the following estimate for the characteristic residence time of a fluid parcel released at a setback distance x ¼ L from the beach: sc ¼
log AlKp elL vm log AlKp vm ; vm
sc ¼
1 lL e 1 ; Al2 Kp
vm ¼ 0
vm > 0
(18a)
(18b)
For the range of parameter values typical of the field site in Avalon Bay (see Table 2), the residence times predicted by Eq. (18) vary over one hundred thousand fold, from approximately 30 mine1000 days (Fig. 6). This variability in residence time derives, in part, from the non-linear dependence of residence time on both set-back distance and meteoric flow (Eq. (18a)). Furthermore, if studies of residence times in estuaries are any guide, the residence time sc of water parcels in shallow groundwater is best characterized by a probability distribution, not a single value. Studies in estuarine systems have noted that fluid parcel residence times tend to follow probability distributions characterized by long tails, implying that
Fig. 6 e Characteristic shallow groundwater residence times predicted by Eq. (18) for various set-back distances and meteoric flow velocities.
5651
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Table 3 e Inputs Parameters used in the uncertainty analysis (Eq. (19)). Constant k* kd ki mg Ks DOCT0 C0 s
Name
Value
Estimated uncertainty
PAA decay rate EC disinfection rate (T ¼ 15 C) EC natural decay rate Max growth rate Growth rate saturation constant Initial DOC concentration Initial peroxycompounds Residence time
1.35 L$mmol1$min1 3.7 L$mmol1$min1 0.1 h1 0.005 min1 0.07 0.006 g/ L1 0.03 mmol 2 day
0.2 0.2 0.2 0.2 0.2 0.2 0.2 1
a minority of fluid parcels spend a very long time in the estuary (Oliveira and Baptista, 1997).
6. Uncertainty analysis and practical implications One advantage of the analytical in-situ disinfection model derived earlier (Eq. (13)) is that the uncertainty associated with different independent variables can be assessed quantitatively using the Law of Uncertainty (Taylor and Kuyatt, 1993): u2 ðSÞ ¼
p X i¼1
2 vS u2 ðXi Þ vXi
(19)
In this equation, u2(S ) and u2(Xi) represents the variance of the dependent and independent variables, respectively, the dependent variable is the log-transformed bacteria concentration (S ¼ log(N )), the independent variables Xi include all variables appearing on the right hand side of Eq. (13) (i.e., C0, Ks, mg, DOCT0, k*, kd, ki, or s), vS/vXi is the sensitivity of the dependent variable to change in a particular independent variable (computed analytically from Eq. (13)), and the summation is taken over all independent variables (P ¼ 8). The uncertainty (or variance) associated with each independent variable was estimated from the relative uncertainty UR(Xi) and magnitude jXi j values listed in Table 3: uðXi Þ ¼ UR ðXi ÞjXi j
(20)
The form of the Law of Uncertainty adopted here assumes that independent variables do not co-vary, which is reasonable for most combinations of independent variables included in this analysis. When applied to Eq. (13), the Law of Uncertainty reveals that 99% of the variance in the log-transformed bacteria concentration can be attributed to variance in just three independent variables: residence time (s) (90%), maximum growth rate (mg) (8%), and the inactivation rate (ki) (1%). These results imply that the prediction (and optimization) of in-situ disinfection will depend strongly on the residence time of shallow groundwater which, in turn, depends non-linearly on the injection well set-back distance (L) and physical characteristics of the shallow groundwater system that can vary in time and space (l, A, KP, vm) (see Eq. (18)). Given the very approximate nature of the analysis that led to Eq. (18), experimental characterization of shallow groundwater residence times would be a fruitful topic for further investigation.
The disinfection model’s sensitivity to residence time is, in part, a consequence of the fact that fecal bacteria can grow in the environment, and thus dramatically different disinfection outcomes (e.g., from a net reduction in bacteria concentrations to a net increase in bacteria) can be caused by slight changes in residence time of water parcels in the shallow groundwater. Given that most recreational waterborne illnesses are caused by human viruses that cannot grow outside their host (Schoen et al., 2011), it is possible that in-situ disinfection of sewage contaminated shallow groundwater would reduce shoreline concentrations of human pathogens (and hence lower recreational waterborne illness rates), even if it did not substantially reduce fecal indicator bacteria concentrations. Indeed, the environmental growth of EC and ENT can lead to a decoupling between fecal indicator bacteria and human pathogens in recreational waters (Litton et al., 2010), and potentially nullify epidemiological relationships upon which current fecal indicator bacteria criteria are based (Colford et al., 2007). In light of these and other concerns the U.S Environmental Protection Agency is evaluating and possibly revising the current water quality criteria for marine recreational beaches (Boehm et al., 2009a).
7.
Conclusions
Aeration alone and aeration with brine did not significantly reduce EC and ENT concentrations in mixtures of raw sewage and shallow groundwater after 6 h of exposure, while 4e5 mg L1 of PAA and 4 mg L1 Oxone achieved >3 log reduction of EC and ENT after 15 min of exposure. Oxone disinfection is enhanced at higher salinities, most likely due to the formation of secondary oxidants (e.g., chlorine and bromine) that make this disinfectant inappropriate for marine applications. PAA disinfection of fecal bacteria in shallow groundwater in coastal settings depends non-linearly on residence time, and the “ideal” disinfection outcome (low bacterial concentration and low PAA residual) is achieved over a relatively narrow window of residence times and initial disinfection concentrations. By analogy to surface estuaries, the residence time of water parcels in shallow groundwater under the influence of marine tides (i.e., subterranean estuaries) is likely to exhibit broad (e.g., logenormal) probability distributions with long
5652
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
tails, and depend sensitively on local precipitation, meteoric flow, and tidal variability. Uncertainty calculations suggest 99% of the uncertainty associated with the log reduction of bacteria is caused by just three independent variables: residence time (s) (90%), maximum growth rate (mg) (8%), and the inactivation rate (ki) (1%). Given these results, further research into the residence time of water in shallow groundwater is needed before an in-situ disinfection schemes can be successfully designed and implemented.
Acknowledgments The authors thank R. Litton, L. Ho, and J. Monroe for assistance with the experiments, and C. Wagner and P. Woolson, and the City of Avalon staff for the use of City Hall for the disinfection studies. Funding was provided by the City of Avalon and State Water Resources Control Board Clean Beaches Initiative, under Agreement 07-582-550.This is publication 66 of the Urban Water Research Center, University of California, Irvine.
references
Abarnou, A., Miossec, L., 1992. Chlorinated waters discharged to the marine-environment chemistry and environmentalimpact e an overview. Science of the Total Environment 126 (1e2), 173e197. Allonier, A.S., Khalanski, M., Camel, V., Bermond, A., 1999. Characterization of chlorination by-products in cooling effluents of coastal nuclear power stations. Marine Pollution Bulletin 38 (12), 1232e1241. Anipsitakis, G.P., Dionysiou, D.D., 2003. Degradation of organic contaminants in water with sulfate radicals generated by the conjunction of peroxymonosulfate with cobalt. Environmental Science & Technology 37 (20), 4790e4797. Antonelli, M., Mezzanotte, V., Panouilleres, M., 2009. Assessment of peracetic acid disinfected effluents by microbiotests. Environmental Science & Technology 43 (17), 6579e6584. Boehm, A.B., Grant, S.B., Kim, J.H., Mowbray, S.L., McGee, C.D., Clark, C.D., Foley, D.M., Wellman, D.E., 2002. Decadal and shorter period variability of surf zone water quality at Huntington Beach, California. Environmental Science & Technology 36 (18), 3885e3892. Boehm, A.B., Fuhrman, J.A., Mrse, R.D., Grant, S.B., 2003. Tiered approach for identification of a human fecal pollution source at a recreational beach: case study at Avalon Cay, Catalina Island, California. Environmental Science & Technology 37 (4), 673e680. Boehm, A.B., Shellenbarger, G.G., Paytan, A., 2004. Groundwater discharge: potential association with fecal indicator bacteria in the surf zone. Environmental Science & Technology 38 (13), 3558e3566. Boehm, A.B., Ashbolt, N.J., Colford Jr., J.M., Dunbar, L.E., Fleming, L.E., Gold, M.A., Hansel, J.A., Hunter, P.R., Ichida, A.M., McGee, C.D., 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 & Technology 43 (21), 8046e8052. Booth, R., Lester, J., 1995. The potential formation of halogenated by-products during peracetic acid treatment of final sewage effluent. Water Research 29 (7), 1793e1801. Burnett, W.C., Bokuniewicz, H., Huettel, M., Moore, W.S., Taniguchi, M., 2003a. Groundwater and pore water inputs to the coastal zone. Biogeochemistry 66 (1), 3e33. Burnett, W.C., Bokuniewicz, H., Huettel, M., Moore, W.S., Taniguchi, M., 2003b. Groundwater and pore water inputs to the coastal zone. Biogeochemistry 66 (1e2), 3e33. Chick, H., 1908. An investigation of the laws of disinfection. Journal of Hygiene 8 (01), 92e158. Colford Jr., J.M., Wade, T.J., Schiff, K.C., Wright, C.C., Griffith, J.F., Sandhu, S.K., Burns, S., Sobsey, M., Lovelace, G., Weisberg, S.B., 2007. Water quality indicators and the risk of illness at beaches with nonpoint sources of fecal contamination. Epidemiology 18 (1), 27. Cristiani, P., 2005. Solutions to fouling in power station condensers. Applied Thermal Engineering 25 (16), 2630e2640. de Lafontaine, Y., Despatie, S.P., Wiley, C., 2008. Effectiveness and potential toxicological impact of the PERACLEAN (R) ocean ballast water treatment technology. Ecotoxicology and Environmental Safety 71 (2), 355e369. Deleersnijder, E., Campin, J.M., Delhez, E.J.M., 2001. The concept of age in marine modelling I. Theory and preliminary model results. Journal of Marine Systems 28 (3e4), 229e267. Dell’Erba, A., Falsanisi, D., Liberti, L., Notarnicola, M., Santoro, D., 2004. Disinfecting behaviour of peracetic acid for municipal wastewater reuse. Desalination 168, 435e442. Dell’Erba, A., Falsanisi, D., Liberti, L., Notarnicola, M., Santoro, D., 2007. Disinfection by-products formation during wastewater disinfection with peracetic acid1. Desalination 215 (1e3), 177e186. Grant, S.B., Sanders, B.F., 2010. Beach boundary layer: a framework for addressing recreational water quality impairment at enclosed beaches. Environmental Science & Technology, 4818e4824. Guo, G.M., Lin, F.K., 2009. The bromination kinetics of phenolic compounds in aqueous solution. Journal of Hazardous Materials 170 (2e3), 645e651. Hom, L.W., 1972. Kinetics of chlorine disinfection in an ecosystem. Journal of the Sanitary Engineering Division 98 (1), 183e194. Kitis, M., 2004. Disinfection of wastewater with peracetic acid: a review. Environment International 30 (1), 47e55. Lazarova, V., Janex, M., Fiksdal, L., Oberg, C., Barcina, I., Pommepuy, M., 1998. Advanced wastewater disinfection technologies: short and long term efficiency. Water Science and Technology 38 (12), 109e117. Lefevre, F., Audic, J., Ferrand, F., 1992. Peracetic acid disinfection of secondary effluents discharged off coastal seawater. Water Science and Technology 25 (12), 155e164. Levenspiel, O., 1980. The monod equation e a revisit and a generalization to product inhibition situations. Biotechnology and Bioengineering 22 (8), 1671e1687. Liberti, L., Lopez, A., Notarnicola, M., 1999. Disinfection with peracetic acid for domestic sewage re use in agriculture. Water and Environment Journal 13 (4), 262e269. Lipp, E.K., Farrah, S.A., Rose, J.B., 2001. Assessment and impact of microbial fecal pollution and human enteric pathogens in a coastal community. Marine Pollution Bulletin 42 (4), 286e293. Litton, R.M., Ahn, J.H., Sercu, B., Holden, P.A., Sedlak, D.L., Grant, S. B., 2010. Evaluation of chemical, molecular, and traditional markers of fecal contamination in an effluent dominated urban stream. Environmental Science & Technology.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 4 1 e5 6 5 3
Manheim, F.T., Krantz, D.E., Bratton, J.F., 2004. Studying ground water under Delmarva coastal bays using electrical resistivity. Ground Water 42 (7), 1052e1068. McIlvin, M.R., Altabet, M.A., 2005. Chemical conversion of nitrate and nitrite to nitrous oxide for nitrogen and oxygen isotopic analysis in freshwater and seawater. Analytical Chemistry 77 (17), 5589e5595. Michael, H.A., Mulligan, A.E., Harvey, C.F., 2005. Seasonal oscillations in water exchange between aquifers and the coastal ocean. Nature 436 (7054), 1145e1148. Mika, K.B., Imamura, G., Chang, C., Conway, V., Fernandez, G., Griffith, J.F., Kampalath, R.A., Lee, C.M., Lin, C.C., Moreno, R., Thompson, S., Whitman, R.L., Jay, J.A., 2009. Pilot- and benchscale testing of faecal indicator bacteria survival in marine beach sand near point sources. Journal of Applied Microbiology 107 (1), 72e84. Monarca, S., Feretti, D., Collivignarelli, C., Guzzella, L., Zerbini, I., Bertanza, G., Pedrazzani, R., 2000. The influence of different disinfectants on mutagenicity and toxicity of urban wastewater. Water Research 34 (17), 4261e4269. Monsen, N.E., Cloern, J.E., Lucas, L.V., Monismith, S.G., 2002. A comment on the use of flushing time, residence time, and age as transport time scales. Limnology and Oceanography 47 (5), 1545e1553. Moore, W.S., 1999. The subterranean estuary: a reaction zone of ground water and sea water. Marine Chemistry 65 (1e2), 111e125. Nielsen, P., 1990. Tidal dynamics of the water-table in beaches. Water Resources Research 26 (9), 2127e2134. Oliveira, A., Baptista, A.M., 1997. Diagnostic modeling of residence times in estuaries. Water Resources Research 33 (8), 1935e1946. ˆ n, I., 1995. Sanchez-Ruiz, C., MartI`nez-Royano, S., Tejero-MonzU An evaluation of the efficiency and impact of raw wastewater disinfection with peracetic acid prior to ocean discharge. Water Science and Technology 32 (7), 159e166. Santoro, D., Gehr, R., Bartrand, T.A., Liberti, L., Notarnicola, M., Dell’Erba, A., Falsanisi, D., Haas, C.N., 2007. Wastewater disinfection by peracetic acid: assessment of models for
5653
tracking residual measurements and inactivation. Water Environment Research 79 (7), 775e787. Santos, I.R., Niencheski, F., Burnett, W., Peterson, R., Chanton, J., Andrade, C.F.F., Milani, I.B., Schmidt, A., Knoeller, K., 2008. Tracing anthropogenically driven groundwater discharge into a coastal lagoon from southern Brazil. Journal of Hydrology 353 (3e4), 275e293. Schoen, M.E., Soller, J.A., Ashbolt, N.J., 2011. Evaluating the importance of faecal sources in human-impacted waters. Water Research. Stampi, S., De Luca, G., Onorato, M., Ambrogiani, E., Zanetti, F., 2002. Peracetic acid as an alternative wastewater disinfectant to chlorine dioxide. Journal of Applied Microbiology 93 (5), 725e731. Swarzenski, P., Burnett, W., Greenwood, W., Herut, B., Peterson, R., Dimova, N., Shalem, Y., Yechieli, Y., Weinstein, Y., 2006. Combined time-series resistivity and geochemical tracer techniques to examine submarine groundwater discharge at Dor Beach, Israel. Geophysical Research Letters 33 (24), L24405. Taniguchi, M., Burnett, W.C., Cable, J.E., Turner, J.V., 2002. Investigation of submarine groundwater discharge. Hydrological Processes 16 (11), 2115e2129. Taylor, B.N., Kuyatt, C.E., Standards N.I.o. and Technology, 1993. Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, Citeseer. USEPA, 2003. Guidelines Establishing Test Procedures for the Analysis of Pollutants; Analytical Methods for Biological Pollutants in Ambient Water, Final Rule US Federal Register, 40 CFR 68, 139, FRL-7529-7. Wagner, M., Brumelis, D., Gehr, R., 2002. Disinfection of wastewater by hydrogen peroxide or peracetic acid: development of procedures for measurement of residual disinfectant and application to a physicochemically treated municipal effluent. Water Environment Research 74 (1), 33e50. Watch F.a.W., 2008. Overloaded and Underfunded: Northwestern California’s Aging Infrstructure Needs A Clean Water Trust Fund. Food and Water Watch.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effect of some parameters on the rate of the catalysed decomposition of hydrogen peroxide by iron(III)-nitrilotriacetate in water Joseph De Laat a,*, Yen Hai Dao a, Nasma Hamdi El Najjar a, Claude Daou b a
Universite´ de Poitiers, Laboratoire de Chimie et Microbiologie de l’Eau (CNRS UMR 6008), Ecole Nationale Supe´rieure d’Inge´nieurs de Poitiers, 1, rue Marcel Dore´, 86 022 Poitiers Cedex, France b Universite´ Saint-Esprit de Kaslik (USEK), Faculte´ des Sciences, Jounieh, Lebanon
article info
abstract
Article history:
The decomposition rate of H2O2 by iron(III)-nitrilotriacetate complexes (FeIIINTA) has been
Received 24 May 2011
investigated over a large range of experimental conditions: 3 < pH < 11, [Fe(III)]T,0:
Received in revised form
0.05e1 mM; [NTA]T,0/[Fe(III)]T,0 molar ratios : 1e250; [H2O2]0: 1 mMe4 M) and concentrations
12 August 2011
of HO radical scavengers: 0e53 mM. Spectrophotometric analyses revealed that reactions
Accepted 15 August 2011
of H2O2 with FeIIINTA (1 mM) at neutral pH immediately lead to the formation of inter-
Available online 24 August 2011
mediates (presumably peroxocomplexes of FeIIINTA) which absorb light in the region
350e600 nm where FeIIINTA and H2O2 do not absorb. Kinetic experiments showed that the Keywords:
decomposition rates of H2O2 were first-order with respect to H2O2 and that the apparent
Fenton reaction
first-order rate constants were found to be proportional to the total concentration of
Aminopolycarboxylate ligands
FeIIINTA complexes, were at a maximum at pH 7.95 0.10 and depend on the [NTA]T,0/
Kinetics
[Fe(III)]T,0 and [H2O2]0/[Fe(III)]T,0 molar ratios. The addition of increasing concentrations of
Effect of pH
tert-butanol or sodium bicarbonate significantly decreased the decomposition rate of H2O2,
Tert-butanol
suggesting the involvement of HO radicals in the decomposition of H2O2. The decompo-
Bicarbonate ion
III
sition of H2O2 by Fe NTA at neutral pH was accompanied by a production of dioxygen and by the oxidation of NTA. The degradation of the organic ligand during the course of the reaction led to a progressive decomplexation of FeIIINTA followed by a subsequent precipitation of iron(III) oxyhydroxides and by a significant decrease in the catalytic activity of Fe(III) species for the decomposition of H2O2. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biological processes are widely used to treat industrial wastewater because they are very efficient and much more cost-effective than physical and chemical processes. However, the effectiveness of biological treatments can be limited when wastewaters contain non-biodegradable or toxic substances. In these cases, a chemical oxidation step prior to
a biological treatment can be used to achieve a partial oxidation of toxic and/or bio-recalcitrant organic pollutants into biodegradable by-products (Pera-Titus et al., 2004). Chemical oxidation can be performed by using advanced oxidation processes (AOPs) which involve the generation of highly reactive hydroxyl radicals. Among the AOPs, the Fenton and Fenton-like reagents (Fe(II)/H2O2 or Fe(III)/H2O2, pH 3) have been implemented on a number of industrial wastewater
* 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.2011.08.028
5655
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
Table 1 e Simplified reaction model of the FeIINTA/H2O2 and of the FeIIINTA/H2O2 systems. Reactions*
Rate or equilibrium constant
Acid-base equilibrium reactions þ 1 H2O2 $ HO 2 þ H 3NTA ¼ L ¼ N(CH2COO-)3 H3L $ H2L þ Hþ 2 H2L $ HL2- þ Hþ HL2- $ L3- þ Hþ 3 HO $ O þ Hþ 4 HO2 $ O2 þ Hþ
pKa ¼ 10.7 (Buxton et al., 1998) pKa1 ¼ 1.79 (Sanchiz et al., 1999) pKa2 ¼ 2.31 (Sanchiz et al., 1999) pKa3 ¼ 9.37 (Sanchiz et al., 1999) pKa ¼ 11.9 (Buxton et al., 1998) pKa ¼ 4.8 (Bielski et al., 1985)
Reactions involving FeIINTA or FeIIINTA complexes 5 FeðIIÞ þ NTA %FeII NTA
6
FeðIIIÞ þ NTA %FeIII NTA
7
FeIINTA þ H2O2 / FeIIINTA þ HO þ HO
8 9 10 11 12
FeIIINTA þ H2O2 $ {FeIIINTAH2O2} {FeIIINTAH2O2} / FeIINTA þ HO2/O2 FeIINTA þ HO2/O2 / FeIIINTA þ H2O2 FeIIINTA þ HO2/O2 / FeIINTA þ O2 HO þ FeIINTA / FeIIINTA þ HO
13
HO þ FeIIINTA / Products
Other reactions 14a 14b 15
HO þ H2O2 / HO2 þ H2O HO þ HO 2 / O2 þ H2O HO þ NTA / Products
16 17a 17b 18
HO þ tert-Butanol / Products þ H2O HO þ HCO 3 / CO3 þ H2O / CO HO þ CO23 3 þ HO CO3 þ H2O2 / HO2 þ HCO 3
log K for Fe2þ þ L3 % FeL : log K ¼ 7.47 (Demmink and Beenackers, 1997) log K ¼ 8.8 (Kurimura et al., 1968) log K ¼ 10.6 (MINEQLþ database) log K for Fe3þ þ L3 % FeL: log K ¼ 15.09 (Sanchiz et al., 1999) log K ¼ 18.6 (MINEQLþ database) log K ¼ 15.90 (Kurimura et al., 1968) log K ¼ 15.90 (Motekaitis and Martell, 1994) k ¼ 9.7 103 M1 s1 (Gilbert and Jeff, 1988) k ¼ 1.84 104 M1 s1 (Borggaard et al., 1971)
5.0 109 M1 s1 at pH ¼ 6.2 (Lati and Meyerstein, 1978) 2.3 109 M1 s1 (Cabelli and Bielski, 1990) k ¼ 1.6 108 M1 s1 at pH z2 (Sharma and Sahul, 1982) 2.7 107 M1 s1 (Buxton et al., 1998) 7.5 109 M1 s1 (Buxton et al., 1998) k ¼ 6.1 107 M1 s1 at pH ¼ 2.0; k ¼ 5.5 108 M1 s1 at pH ¼ 6.0; k ¼ 4.2 109 M1 s1 at pH ¼ 10 (Sahul and Sharma, 1987) k ¼ 7.5 108 M1 s1 at pH ¼ 4.0; k ¼ 2.5 109 M1 s1 at pH ¼ 9.0 (Lati and Meyerstein, 1978) k ¼ 2.1 109 M1 s1 at pH w0 (Borggaard, 1972) 6.0 108 M1 s1 (Buxton et al., 1998) 8.6 106 M1 s1 (Buxton et al., 1998) 3.9 108 M1 s1 (Buxton et al., 1998) 8 105M1 s1 (Behar et al., 1970) 4.3 105M1 s1 (Draganic et al., 1991)
*For simplicity, the terms FeIINTA and FeIIINTA will be used hereafter to represent all the forms of iron(II) and iron(III)-nitrilotriacetate complexes, unless specifically stated otherwise. Several reactions are not chemically balanced.
treatment plants (Bautista et al., 2008). The main benefits of the Fenton and Fenton-like reactions are the use of environmentally friendly and low cost reagents. However, these processes have also some limitations. They must be operated at low pH (pH z 3) in order to prevent the precipitation of ferric oxyhydroxides and in the absence of high concentrations of chloride or sulfate ions because of the formation of inactive chloro or sulfato-iron(III)-complexes (De Laat and Le, 2005, 2006). To overcome these drawbacks, the addition of organic and inorganic iron-chelating agents has been used to increase the solubility of iron species at neutral pH and therefore to enhance the efficiency of the homogenous Fenton-like oxidation processes (Sun and Pignatello, 1992; 1993; Li et al., 2007; Lee and Sedlak, 2009; Rastogi et al., 2009), or of the heterogenous systems involving ion bearing minerals (Xue et al., 2009) or granular zero-valent iron (Keenan and Sedlak, 2008; Lee et al.,
2008). The efficiency of 50 iron(III)-chelates for the decomposition of H2O2 and of 2,4-dichlorophenoxyacetic at pH 6 have been examined by Sun and Pignatello (1992) and nitrilotriacetic acid (NTA) was found to be one of the most active chelates. The ironNTA/H2O2 process 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 suggested 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 (HO radicals or/ and high-valent-oxoiron species) generated by the reaction of H2O2 with free and complexed Fe(II) and Fe(III) species remains a controversial issue (Walling et al., 1975; Yamazaki and Piette, 1991; Bossmann et al., 1998; Bamnolker et al.,
5656
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
1991; Pignatello et al., 2006). It is generally accepted that the reaction of H2O2 with free Fe2þ at low pH yields hydroxyl radicals whereas other oxidants such as the ferryl ion may be produced at neutral pH (Gallard et al., 1998; Rivas et al., 2001; Hug and Leupin, 2003; Keenan and Sedlak, 2008; Katsoyiannis et al., 2008). In the case of systems involving FeIINTA and FeIIINTA complexes, a recent competitive kinetic study showed that the degradation of three probe compounds (atrazine, fenuron, parachlorobenzoic acid) by FeIINTA/H2O2, FeIIINTA/H2O2 and FeIINTA/O2 at neutral pH could be attributed to HO radicals (Dao and De Laat, 2011). This was also confirmed by a decrease of the rates of degradation of the probe compounds in the presence of HO scavengers (tertbutanol and bicarbonate ion). These data do not exclude the formation of other oxidant species concurrently with the formation of HO radicals during the decomposition of H2O2 by FeIINTA and FeIIINTA complexes. Previous works on the FeIINTA/H2O2 and FeIIINTA/H2O2 systems showed that the rates of decomposition of H2O2 and of degradation of organic solutes were at a maximum at pH z 8 (Tachiev et al., 2000; Kim and Kong, 2001; Dao and De Laat, 2011) and decreased in the presence of increasing concentrations of hydroxyl radical scavengers such as tert-butanol and bicarbonate ions (Dao and De Laat, 2011). On the basis of the generally accepted mechanisms of the Fenton and Fenton-like oxidation processes, Table 1 reports the main reactions which can be proposed for the catalytic decomposition of hydrogen peroxide by iron(II)-NTA and iron(III)-NTA complexes. As shown in Table 1, most of the rate constants of elementary reactions which may be involved in the FeIINTA/H2O2 and FeIIINTA/H2O2 systems are unknown or not well known such as rate constants for the reactions of HO with the various FeIINTA and FeIIINTA complexes (reactions 12 and 13 in Table 1) and with the various acid-base forms of NTA (reactions 4 and 15 in Table 1). In addition, it has been demonstrated that the formation of iron(III)-peroxocomplexes represents the first step of the decomposition of H2O2 by free Fe(III) (Gallard et al., 1999) and by iron(III) complexes of ethylenediamine tetraacetate (EDTA). In the case of the FeIIIEDTA/H2O2 system, the reaction of H2O2 with FeIIIEDTA leads to the formation of purple peroxocomplexes and several studies have been conducted in order to determine equilibrium and rate constants for the formation of FeIIIEDTA peroxocomplexes (Walling et al., 1970; Francis et al., 1985). The formation of similar peroxocomplexes with FeIIINTA complexes has never been reported in literature. Table 1 also indicates that NTA would be consumed during the course of the reaction (reaction 15) and that this degradation would lead to a progressive decomplexation of iron(III) at neutral pH and therefore to a precipitation of ferric oxyhydroxides. Because of the large uncertainties on the rate constants of the chemical reaction model presented in Table 1, the rates of decomposition of H2O2 by FeIINTA and FeIIINTA complexes can not be predicted by computer simulations. In addition, the effects of various parameters on the decomposition rates of H2O2 are not well documented. Therefore, the main objective of the present work was to investigate the decomposition rates of H2O2 by FeIINTA and FeIIINTA complexes over a wide range of experimental conditions. Experiments were conducted at 25 C and the following parameters have been
investigated: pH, initial concentrations of Fe(III) and H2O2, [NTA]T,0/[FeIII]T,0 and [H2O2]0/[Fe(III)]T,0 molar ratios, concentrations of tert-butanol and bicarbonate ions.
2.
Materials and methods
2.1.
Reagents and preparation of solutions
All chemicals were reagent grade and were used without additional purification. All aqueous solutions were prepared using 18 MU Milli-Q water from a Millipore system. Glassware was washed with perchloric acid (pH z 2) and rinsed before use. Stock solutions of Fe(II) and of Fe(III) were prepared by dissolving the required amounts of iron(II) perchlorate (Fe(ClO4)2, 98%) and of iron(III) perchlorate (Fe(ClO4)3, 9 H2O) in 0.1 M HClO4. Stock solutions of NTA (from 5 to 20 mM) were prepared from nitrilotriacetic acid. Solutions of FeIINTA were prepared by mixing the required volumes of stock solutions of Fe(II) and of NTA in deoxygenated water to prevent spontaneous oxidation of Fe(II). The solutions were kept under dinitrogen gas atmosphere and the pH adjusted to the desired value with NaOH 1 M or 0.1 M. Solutions of FeIIINTA ([NTA]0/[Fe(III)]0 > 1 mol mol1) were prepared by mixing appropriate volumes of stock solutions of Fe(III) (10 mM) and of NTA. Under these conditions, Fe(III) was present only as FeIIINTA complexes at the beginning of the reaction (Motekaitis and Martell, 1994). The pH was then adjusted with NaOH 1 M or 0.1 M. All the FeIINTA and FeIIINTA solutions were freshly prepared each time before use and the concentrations of iron determined just before addition of H2O2.
2.2.
Reaction conditions
All kinetic experiments were carried out using a batch reactor. The batch reactor consisted of a 1.3-L cylindrical double-wall jacketed reactor to circulate thermostated water with an external circulating pump connected to a thermostated water bath (25.0 0.5 C) (Section S2.2 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-L of solution containing FeIINTA or FeIIINTA. 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 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. 1, 2 or 5 mL aliquots were withdrawn at selected time intervals and immediately analyzed for H2O2.
2.3.
Analytical methods
The concentration of H2O2 in stock solutions of H2O2 was determined by iodometric titration. Concentrations of H2O2 in solutions containing Fe(II) or Fe(III) were determined spectrophotometrically using the TiCl4 method (Eisenberg, 1943)
5657
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
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). Absorption spectra were measured with a SAFAS DES 190 double beam spectrophotometer and pH measurements were made with a Meter Lab PHM 240 pH meter calibrated with standard buffers. COD (chemical oxygen demand) was analyzed following the closed reflux - titrimetric method with potassium dichromate as oxidant using a TR320 thermoreactor (Merck, Germany). TOC and TN (total organic carbon and total nitrogen) were measured by using a carbon analyzer (Shimadzue TOC 5000).
decomposition of H2O2 by the FeIIINTA complexes formed from the oxidation of the initial FeIINTA complexes by H2O2. For this second stage of the reaction, the decomposition of H2O2 by FeIIINTA was first-order with respect to H2O2 concentration, with an apparent first-order rate constant equal to (1.83 0.05) 103 s1 (Section S3.1.1 in Supporting Material). The completion of the initial stage of the reaction within a reaction time of less than 20 s indicates that the secondorder rate constants for the reaction of H2O2 with FeIINTA complexes are greater than 103 M1 s1, in agreement with literature (Borggaard et al., 1971; Gilbert and Jeff, 1988). The rate constants for the reaction of H2O2 with FeIINTA could not be determined under the experimental conditions used in the present study. At the end of the initial fast phase of the reaction, the consumption of H2O2 was nearly 1.0e1.1 mol of H2O2/mol of FeIINTA (for [H2O2]0 ¼ 500, 700 or 930 mM). By assuming that the reaction of H2O2 with FeIINTA leads to the formation of HO radicals (reaction 7 in Table 1), the consumption of 1.0e1.1 mol of H2O2/mol of FeIINTA obtained in the presence of an excess of H2O2 suggests that the HO radicals formed by the initial attack of H2O2 with FeIINTA do not contribute to the oxidation of FeIINTA but are trapped by NTA. To examine the effect of the dose of H2O2 and of the [NTA]T,0/[Fe(II)]T,0 ratio on the overall stoichiometry between H2O2 and FeIINTA, an other set of experiments was performed by using FeIINTA in excess ([Fe(II)]T,0 ¼ 1.0 mM; 0.1 < [H2O2]0 < 0.7 mM, pH ¼ 7.0 0.1, [O2] < 0.1 mg O2/L) and with [NTA]0/[Fe(II)]T,0 ratios ranging from 0 to 10 mol/mol (0, 0.5, 1.1, 4 and 10 mol/mol) (Section S3.1.2 in Supporting Material). In the absence of NTA (FeII/H2O2 system), the overall stoichiometry was nearly equal to 1.85 0.05 mol Fe2þ/mol H2O2, in agreement with the expected value of 2 mol of H2O2/ mole of Fe(II) (reactions 7 and 12 in Table 1; Gallard et al., 1998). In the presence of NTA, the measured consumptions of H2O2 were equal to 1.5 mol H2O2/mole Fe(II) for [NTA]T,0/ [Fe(II)]T,0 ¼ 0.5 mol/mol (z50% of Fe(II) present as FeIINTA), 1.3 mol Fe(II)/mol H2O2 for [NTA]T,0/[Fe(II)]T,0 ¼ 1.1 mol/mol and 1.0 mol Fe(II)/mol H2O2 for [NTA]0/[Fe(II)]T,0 ¼ 4 and 10 mol/mol. These data demonstrate that NTA acts as an HO radical scavenger when NTA is in large excess.
3.
Results
3.1.
Decomposition rate of H2O2 by FeIINTA at pH 7
Ferrous species play a key role on the overall rate of decomposition of H2O2 and of degradation of organic pollutants by the Fenton and Fenton-like reactions because the reaction of H2O2 with Fe(II) represents the unique source of HO radicals. Limited data exist on the rate of decomposition of H2O2 by FeIINTA (reaction 7 in Table 1). A few papers indicate that FeIINTA complexes are readily oxidized by H2O2 (Borggaard et al., 1971; Gilbert and Jeff, 1988; Rush and Koppenol, 1988) at circumneutral pH. To illustrate the fast decomposition rate of H2O2 by FeIINTA, Fig. 1 presents typical decomposition profiles of H2O2 obtained after introduction of 200, 500, 700 or 930 mM of H2O2 into an oxygen-free aqueous solution of FeIINTA ([FeII]T,0 ¼ 200 mM; [NTA]T,0 ¼ 500 mM) at pH 7.0 0.1). The data showed a very fast decomposition of H2O2 during the first 20 s of the reaction. This initial stage was then followed by a slow decomposition of H2O2 which corresponds to the catalyzed
1000 First stage : very fast decomposition
3.2.
Decomposition rate of H2O2 by FeIIINTA
3.2.1.
Spectrophotometric study
800 Second stage : slow decomposition
600
400
200
0 0
200
400
600
800
Time (s) Fig. 1 e Decomposition of H2O2 by FeIINTA ([Fe(II)]T,0 [ 200 mM, [NTA]T,0 [ 0.5 mM, [H2O2]0 [ 200, 500, 700 or 930 mM, pH [ 7.0 ± 0.1, [O2]0 < 0.1 mg/L).
In our previous spectrophotometric studies on the Fenton-like reaction (FeIII/H2O2, pH < 3), we showed that the addition of H2O2 into a solution of Fe(III) immediately produces iron(III)peroxocomplexes which absorb in the region 350e600 nm (Gallard et al., 1999) and that their unimolecular decomposition represents the rate limiting step in the regeneration of the ferrous ion (De Laat and Gallard, 1999). It is also generally accepted that the formation of peroxocomplexes represents the first step of the decomposition mechanism of H2O2 by iron(III)-aminopolycarboxylate complexes (Walling et al., 1970; Francis et al., 1985; Tachiev et al., 2000). Contrary to the reaction of H2O2 with FeIIIEDTA complexes which leads to the formation of purple intermediates (Walling et al., 1970), no significant change of color was observed upon addition of
5658
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
H2O2 to FeIIINTA solutions. However, spectrophotometric analyses conducted in the present work demonstrate for the first time that the reaction of H2O2 (0.5e4 M) with FeIIINTA ([Fe(III)]T,0 ¼ 0.89 mM; [NTA]0/([Fe(III)]T,0 ¼ 1.2 mol/mol) leads to the formation of species which absorb UV/visible light and in particular in the region 350e600 nm where FeIIINTA complexes and H2O2 do not absorb (Fig. 2). Analyses also showed that the rise in the absorption occurred immediately after the addition of H2O2 and was pH dependant. As it has been previously demonstrated for uncomplexed iron(III) species at pH < 3 (Gallard et al., 1999), the change in the UV/ visible absorbance upon the addition of H2O2 can be attributed to the formation of complexes between H2O2 and FeIIINTA. Unfortunately, the complexation constants for the formation of peroxocomplexes of FeIIINTA could not be determined in the present work, presumably because the solutions contain several iron(III) species.
3.2.2. Effects of the total concentration of Fe(III) on the decomposition rate of H2O2 Fig. 3a presents typical data obtained for the decomposition of H2O2 by FeIIINTA in the presence of increasing concentrations of Fe(III) (from 0.05 to 0.5 mM) at pH 8.0 0.1 and for [NTA]T,0/[Fe(III)]T,0 ¼ 5 mol/mol. Similar data were also obtained at pH 7.0 and 9.0 (Fig. S3.5 in Supplementary Material). Under the conditions used, the rate of decomposition of H2O2 was first-order with respect to the concentration of H2O2: d½H2 O2 ¼ kapp ½H2 O2 dt
(1)
where kapp is the pseudo-first-order rate constant for the overall depletion of H2O2. The values of kapp have been reported in Table S2 and in Fig. 3b. Fig. 3b shows that the measured first-order rate constants (kapp) were found to increase linearly with the total concentration of FeIII. These
3
Absorbance (1-cm cell)
III
2.5 2 1.5
[Fe ]T,0 = 0.83 mM, [NTA]T,0 = 1.0 mM pH = 7.0 3.99 M [H2O2]0 : 2.43 M 0.83 M 0.41 M 0.00 M
0.5
350
400
450
d½H2 O2 ¼ kd ½FeðIIIÞT;0 ½H2 O2 dt
500
Wavelength (nm) Fig. 2 e Absorption spectra of solutions of FeIIINTA in the absence of H2O2 and in the presence of H2O2 recorded immediately after introduction of H2O2 in the spectrophotometric cell ([Fe(III)]T,0 [ 0.83 mM, [NTA]T,0 [ 1.0 mM, [H2O2]0 [ 0, 0.42, 0.83, 2.43 and 4.0 M, pH0 [ 7.0 ± 0.1, path length of the cell [ 10 mm).
(2)
where kd is the second-order rate constant for the overall rate of decomposition of H2O2 by FeIIINTA complexes. Under the conditions used in the present work ([H2O2]0 ¼ 5 mM; [Fe(III)]T,0 < 0.5 mM; [NTA]T,0/[Fe(III)]T,0 ¼ 5 mol/mol), the mean values of kd show that the reaction was faster at pH 8.0 (kd ¼ 27.1 1.6 M1 s1) than at pH 7.0 (kd ¼ 16.1 0.7 M1 s1) or pH 9.0 (kd ¼ 21.4 1.7 M1 s1).
3.2.3.
Effect of pH
To confirm the effect of pH on the rate of decomposition of H2O2 by FeIIINTA, a series of kinetic experiments has been carried out with pH values ranging from 2.7 to 10.5 and with initial concentrations of H2O2, FeIII and NTA equal to 1 mM, 0.2 mM and 1 mM, respectively. For all the pH studied, the decay of H2O2 was first-order with respect to H2O2. The plots of the apparent first-order rate constants (kapp) (Section S3.2.3 in Supplementary Material) and of the corresponding apparent second-order rate constants (kd ¼ kapp/[Fe(III)]T,0) as a function of pH demonstrate that the rate of decomposition of H2O2 was maximal at pH ¼ 7.95 0.10 (Fig. 3c). At this pH and for the conditions used in this work ([Fe(III)]T,0 ¼ 0.05e0.5 mM; [NTA]T,0/([Fe(III)]T,0 ¼ 5 mol/mol), the second-order rate constant for the reaction of H2O2 with FeIIINTA is equal to kd ¼ 28 2 M1 s1. The existence of an optimal pH for the decomposition of H2O2 by FeIIINTA complexes is in agreement with the data obtained by Francis et al. (1985), Tachiev et al. (2000) and Kim and Kong. (2001). As previously discussed by Tachiev et al. (2000), the effects of pH on the initial rates of decomposition of H2O2 have been attributed to the stability constants for the formation of the various peroxocomplexes from the different acid-base forms of FeIIINTA complexes (four FeIIINTA complexes, Fig. 3d) and to the unimolecular rates of decomposition of the various FeIIINTA peroxocomplexes in water. The increase in the rate of decomposition of H2O2 when the pH increased from to 2.7 to 8.0 may also be attributed to the dissociation of H2O2 (pKa ¼ 11.7) and the decrease in the reaction rates at pH > 8 to changes in the speciation of Fe(III).
3.2.4.
1
0 300
data suggest that the rate of decomposition of H2O2 can be described by a second-order kinetic expression:
Effects of [NTA]T,0/[Fe(III)]T,0 molar ratio
Fig. 4 presents the effects of the [NTA]T,0/[Fe(III)]T,0 molar ratios (2.5 < [NTA]T,0/[FeIII]T,0 < 100 mol/mol) on the apparent firstorder rate constants of decomposition of H2O2 obtained at pH ¼ 7.00 0.05 and 8.00 0.05. All the kinetic constants were determined for decomposition yields of H2O2 which did not exceed 50%. The data showed that the effects depend on the initial concentrations of reactants and on pH. For initial concentrations of Fe(III) and H2O2 equal to 0.2 mM and 5 mM, respectively, the rate of decomposition of H2O2 was not significantly affected at pH 7.0 by the [NTA]T,0/[Fe(III)]T,0 ratio for ratios less than 12.5 mol/mol (mean values of kapp z 2.86 103 s1) and decreased when the [NTA]T,0/[Fe(III)]T,0 molar ratio increased from 12.5 mol/mol (kapp z 2.86 103 s1) to 100 mol/mol (kapp z 1.4 103 s1). For an initial concentration of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
5659
Fig. 3 e Decomposition of H2O2 by FeIIINTA a) First-order plots for the decomposition of H2O2 at pH [ 8.0 and with various initial concentrations of FeIIINTA; b) Plots of the first-order rate constant (kapp) as a function of the concentration of FeIIINTA at pH 7.0, 8.0 and 9.0; c) Change of the first-order rate constant with pH and d) Distribution of FeIIINTA complexes as a function of pH ([Fe(III)]T,0 [ 0.05e0.5 mM; [NTA]T,0/[Fe(III)]T,0 [ 5.0 mol/mol, [H2O2]0 [ 5.0 mM, 25 C).
H2O2 of 40 mM, the pseudo-first-order rate constants increased from (2.7 0.1) 103 s1 to (3.2 0.1) 103 s1 when the [NTA]T,0/ [Fe(III)]T,0 molar ratio increased from 2.5 to 10 mol/mol. An increase was also observed for the series of experiments conducted at pH 8.0 ([Fe(III)]T,0 ¼ 0.05 mM, [H2O2]0 ¼ 3 mM) (Fig. 4). Several assumptions can be made to explain the observed effects of [NTA]T,0/[Fe(III)]T,0 molar ratio on the rate of decomposition of H2O2. Increasing [NTA]T,0/[Fe(III)]T,0 molar ratios would maintain Fe(III) in the form of FeIIINTA complexes during the course of the reaction, especially when high [H2O2]0/ [Fe(III)]T,0 molar ratios are used. Increasing [NTA]T,0/ [Fe(III)]T,0 molar ratios also affect the distribution of FeIIINTA complexes by increasing the proportion of the FeIII(NTA)2 complex. Therefore, the change in the distribution of FeIIINTA complexes may affect the reaction rates. In addition, at [NTA]T,0/[Fe(III)]T,0 higher than 2 mol/mol, NTA is mainly present as free NTA in solution and may act as an HO radical scavenger.
3.2.5.
Effects of [H2O2]0/[Fe(III)]T,0 molar ratio
Fig. 5 presents the effects of the initial concentration of H2O2 (0.7e50 mM) on the rate of decomposition of H2O2. Rate constants were determined from 36 experiments carried out at pH 7.0 and with initial concentrations of Fe(III) and NTA equal to 0.2 mM and 0.5 mM, respectively. Under our conditions, Fe(III) is only present as FeIIINTA complexes at the beginning of the reaction and the [H2O2]0/[FeIIINTA]T,0 molar ratios vary from 3.5 to 250 mol/mol. For all experiments, the
initial rates of decomposition of H2O2 ([H2O2]/[H2O2]0 < 0.5) could be accurately described by a first-order kinetic law with respect to H2O2 (Section S3.2.5 in Supporting Material). The change of the first-order rate constant (kapp) with the concentration H2O2 shows an increase of kapp (from 1.2 103 s1 to (3.2 0.1) 103 s1 when the concentration of H2O2 increased from 0.7 to 10 mM (Fig. 5a). These data can be explained by the fact that an increase of the concentration of H2O2 would increase the concentration of the peroxo FeIIINTA complexes (reaction 7 in Table 1) and therefore the decomposition rate of H2O2. For initial concentrations of H2O2 higher than 25e30 mM, a small decrease in the values of kapp could be observed because kapp decreased from (3.1 0.1) 103 s1 to (2.6 0.1) 103 s1 ([H2O2]0 ¼ 50 mM). As will be seen below (Section 3.3), the decrease of the kapp values observed at high [H2O2]0/[FeIIINTA]T,0 can probably be attributed to a degradation of the FeIIINTA complexes during the first 240 s of the reaction. It should also be noted that the initial rates of decomposition of H2O2 increased from 8 107 M s1 to 1.3 104 M s1 when the initial concentration of H2O2 increased from 0.7 mM to 50 mM (Fig. 5b).
3.2.6.
Effects of hydroxyl radical scavengers
As depicted in Fig. 6, the decomposition rate of H2O2 decreased in the presence of increasing concentrations of hydroxyl radical scavengers such as tert-butanol and bicarbonate ion at pH 8.3. These data are in agreement with those obtained in a previous work dealing with the effects of tert-butanol and
5660
a
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
III
-1
kapp (s )
0.005 [H2O2]0 = 5 mM, [Fe ]T,0 = 0.2 mM; pH = 7.0
0.003 0.001 III
[H2O2]0 = 40 mM; [Fe ]T,0 = 0.2 mM; pH = 7.0
-1
kapp (s )
0.005
0.003
0.002
-1
kapp (s )
0.001
0.001 III
[H2O2]0 = 5 mM, [Fe ]T,0 = 0.05 mM; pH = 8.0
0.000
0
20 40 60 80 III [NTA]T,0 / [Fe ]T,0 (mol/mol)
b 100
100
III
[Fe ]T,0 = 0.2 mM, pH 7.0
80
III
Fe (OH)NTA III
Fe (NTA)2
%
60 40 20 III
Fe (OH)2 NTA
0 0
20 40 60 80 [NTA]T,0 / [Fe(III)]T,0 (mol/mol)
100
Fig. 4 e Effects of the [NTA]T,0/[Fe(III)]T,0 molar ratio a) on the pseudo-first-order rate constant of decomposition of H2O2 at pH 7.0 and 8.0 and b) on the distribution of the FeIIINTA.
(reaction 9 in Table 1), the oxidation of H2O2 by HO radicals (reaction 14a and 14b in Table 1) and from the decomposition of peroxyl radicals formed from the oxidation of NTA by HO radicals. The decrease in the decomposition rate of H2O2 in the presence of tert-butanol can be explained by the fact that increasing concentrations of tert-butanol would decrease the steady-state concentrations of HO radicals and therefore would decrease the rate of decomposition of H2O2 by HO radicals (reactions 14a and 14b in Table 1) and the rate of formation of FeIINTA complexes by reduction of FeIIINTA by HO2/O2 radicals (reaction 11 in Table 1). The inhibiting effect of tert-butanol and bicarbonate ion for the oxidation of probe compound would depend on the relative rates of consumption of HO radicals by all the species present in solution. These relative rates of consumption of HO radicals are proportional to the term ki Ci where ki represents the second-order rate constant for the reaction of HO with a solute i (Table 1) and Ci, the concentration of the solute i. Under the conditions used for the experiments depicted on Fig. 6 ([NTA]0 ¼ 0.5 mM, [H2O2]0 ¼ 5 mM), the scavenging terms
bicarbonate ion on the degradation of HO probes (Dao and De Laat, 2011). In the case of H2O2, the overall rate of decomposition of H2O2 by FeIIINTA complexes depends on the rate of the initiation step of decomposition of H2O2 by FeIIINTA (reactions 8e10 in Table 1) and of the rates of all the secondary reactions that consume H2O2 or that can affect the steady-state concentration of FeIINTA. As for the FeIII/H2O2 system, the steady-state concentration of ferrous species in the FeIIINTA/ H2O2 system would play a key role in the overall rate of decomposition of H2O2. FeIINTA complexes can be formed by the decomposition of peroxocomplexes of FeIIINTA (reaction 9 in Table 1) and by reduction of FeIIINTA complexes by HO2/O2 radicals (reaction 11 in Table 1). The latter can be generated from the decomposition of peroxocomplexes of FeIIINTA
Fig. 5 e Effects of the [H2O2]0/[Fe(III)]T,0 molar ratio a) on the pseudo-first-order rate constant of decomposition of H2O2 and b) on the initial rate of decomposition of H2O2 ([Fe(III)]T,0 [ 0.2 mM, [NTA]T,0 [ 0.5 mM, pH [ 7.00 ± 0.05, 25 C).
5661
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
data are consistent with the fact that almost all the HO radicals would be scavenged by tert-butanol at concentrations of tert-butanol higher than 5 mM (ki Ci > 3 106 s1 for [tertbutanol] > 5 mM). In the case of bicarbonate ion, the data in Fig. 6b demonstrate that the inhibiting effect of bicarbonate ion is less important than the one to tert-butanol. This finding is not very surprising because the reaction of HO radicals with bicarbonate and carbonate ions leads to the formation of carbonate radicals (reactions 17a and 17b in Table 1) which are reactive toward H2O2 (reaction 18 in Table 1) and probably also toward NTA.
3.2.7.
Effects of chloride, sulfate and phosphate ions
Other kinetic experiments showed that the rate of decomposition of H2O2 by FeIIINTA at pH 8.3 ([Fe(III)]0 ¼ 0.2 mM, [NTA]0 ¼ 0.5 mM, [H2O2]0 ¼ 5 mM) was not affected by chloride ions (100 mM) and by sulfate ions (33 mM) whereas the presence of phosphate ions was found to decrease the decomposition rate of H2O2 (kapp z 5.7 103 s1, 4.97 103 and 2.7 103 for [PO34 ]T ¼ 0, 5 and 30 mM). The kinetic data obtained in the presence of phosphate ions might be explained by the formation of less reactive phosphate iron(III)-complexes but spectrophotometric analyses did not reveal a change of the UV/Visible spectra of FeIIINTA solutions ([H2O2] ¼ 0 mM) in the presence of phosphate ions (0e30 mM) as well as in the presence of chloride or sulfate ions.
3.2.8.
Effect of dissolved oxygen
The concentration of dissolved oxygen increased during the decomposition of H2O2 by FeIIINTA complexes. Under our experimental conditions ([Fe(III)]T,0 ¼ 0.2 mM, [NTA]T,0 ¼ 0.5 mM, [H2O2]0 ¼ 5 mM, pH ¼ 7 or 8.3), the production of dissolved oxygen at the beginning of the reaction was nearly equal to 0.25 mol O2/mol of H2O2 decomposed at pH 7 and 8.3 in the absence of tert-butanol and was roughly equal to 0.12 mol O2/mole of H2O2 decomposed in the presence of tertbutanol ([tert-butanol] > 0.5 mM, pH 8.3). These values represent net productions of dissolved oxygen because dissolved oxygen can be produced through various reactions (i.e. reaction 11 in Table 1) and consumed by organic radicals to produce organic peroxyl radicals.
Fig. 6 e Pseudo-first-order plots of H2O2 decomposition as a function of the concentration of a) bicarbonate ion and b) tert-butanol. c) Comparison of the pseudo-first-order rate constants of decomposition of H2O2 as a function of the value of the term ki Ci ([Fe(III)]T,0 [ 0.2 mM, [NTA]T,0 [ 0.5 mM, [H2O2]0 [ 5 mM, pH [ 8.30 ± 0.05, 25 C).
for NTA (ki z 109 M1 s1) and H2O2 (ki z 3 107 M1 s1 at neutral pH) are roughly equal to 5 105 s1 and 1.5 105 s1, respectively. The kinetic data obtained with tert-butanol showed a nearly complete inhibition of the decomposition rate of H2O2 at concentrations of tert-butanol higher than 5 mM (kapp z 5.7 103 s1 for [tert-butanol] ¼ 0 mM; kapp z 3.5 104 s1 for 10 mM < [tert-butanol] < 50 mM, Fig. 6a). These
3.3.
Degradation of the FeIIINTA complexes
It is known that NTA degradation by HO radicals leads to the formation of iminodiacetic acid (IDA), glycine, oxalic acid, ammonia and carbon dioxide (Chen et al., 1995). A decrease in the efficiency of the FeIIINTA/H2O2 process is therefore expected during the course of the reaction because the formation of weaker chelating by-products would lead to a progressive decomplexation and precipitation of Fe(III) after the complete depletion of free NTA in solution. In addition, complexes of Fe(III) with IDA are less active than FeIIINTA complexes for the decomposition of H2O2 (Fig S3.18). HPLC analyses for the determination of NTA and its by-products could not be done during this work. To illustrate the fate of NTA and FeIIINTA complexes, a series of flasks containing Fe(III) (1 mM) and NTA (3 mM) were treated with H2O2 doses ranging from 0 to 50 mM at pH 7. After a reaction time of 24 h
5662
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
1.0
[C] / [C]0
0.8
0.6
0.4 COD TOC TN Absorbance at 260 nm Total dissolved iron
0.2
0.0
0
10
20
30
40
50
H2O2 consumed (mM) Fig. 7 e Degradation of the ligand during the decomposition of H2O2 by FeIIINTA: Normalized values for the UV absorbance at 260 nm and for the concentrations of COD ([COD]0 [ 432 mg O2/L), TOC ([TOC]0 [ 216 mg C/L), TN ([TN]0 [ 42 mg N/L) and of total dissolved concentration of iron ([Fe(III)]T,0 [ 1.0 mM) as a function of the consumption of H2O2 (Batch experiments, 24-h reaction time, [Fe(III)]T,0 [ 1 mM, [NTA]T,0 [ 3 mM, [H2O2]0 [ 0e50 mM, H2O2 removal after 24-h reaction time : >97%, Initial pH [ 7.0, Final pH > 5.5e6.8).
in the dark, the solutions were filtered through 0.45 mm filters to eliminate the precipitate of iron(III) oxyhydroxides. The filtered samples were then analyzed by UV/Visible absorption spectroscopy and the residual concentrations of H2O2, dissolved iron, COD, TOC and TN were determined. The initial concentrations of COD, TOC and TN were equal to 432 mg O2/L, 216 mg C/L and 42 mg N/L, in agreement with the theoretical values. After a reaction time of 24 h, analyses showed that H2O2 was completely depleted in samples treated with H2O2 doses less than 10 mM and that the H2O2 removals were nearly equal to 97% for all samples treated with H2O2 doses higher or equal 1 III
H2 O2 (Fe /H2 O2 , pH 7 without NTA)
[C]t / [C] 0
0.8
0.6
H2O2 nd
2 injection
0.4
rd
3 injection Total dissolved iron
0.2
0 0
3000
6000
9000
12000
15000
18000
Time (s) Fig. 8 e Changes of the catalytic activity of FeIIINTA for the decomposition of H2O2 and of the total concentration of dissolved iron measured after 3 consecutive injections of 10 mM H2O2 at reaction times 0, 30 and 120 min ([Fe(III)]T,0 [ 1 mM, [NTA]T,0 [ 3 mM, pH [ 7.0 ± 0.1).
to 15 mM. Increasing doses of H2O2 led to a progressive decrease in the concentrations of COD and of TOC which demonstrate an oxidation of NTA. For the highest dose tested (50 mM), COD and TOC removals reached 50% and 40%, respectively. TN loss was less than 5% indicating that the degradation of free and complexed NTA by the FeIIINTA/H2O2 system did not lead to significant amounts of N2 or of N2O. The decrease in the concentrations of DOC and TOC was accompanied by a decrease in the concentration of dissolved iron in solution which is consistent with the formation of ferric oxyhydroxide precipitates and with the decrease in the UV absorbance of the solutions at the wavelength of 260 nm. Under the conditions used in the present work ([Fe(III)]T,0 ¼ 1 mM, [NTA]T,0 ¼ 3 mM, initial pH ¼ 7.0), the decrease of the UV absorbance and of the concentration of dissolved iron appeared at H2O2 doses higher than 5 mM. The decreases in the UV absorbance at 260 nm and in the total concentration of Fe(III) were nearly equal to 10, 30 and 70% for H2O2 doses of 10, 20 and 30 mM, respectively (Fig. 7). To illustrate the effects of the degradation of the ligand on the activity of the FeIIINTA for the decomposition of H2O2, Fig. 8 presents decomposition curves of H2O2 obtained after three consecutive additions of H2O2 (introduction of 10 mM of H2O2 at reaction times 0, 30 min and 2 h). As expected, the rate of decomposition of H2O2 and the total concentration of dissolved iron slowed down after each addition of H2O2 (Fig. 8). The initial rates of decomposition of H2O2 were equal to 92, 44 and 2 mM s1 after the first, the second and the third addition of H2O2, respectively. These data demonstrate that the FeIIINTA/H2O2 is not a true catalytic system because of the progressive degradation of the ligand during the course of the reaction. To avoid the decrease in the catalytic activity of FeIIINTA, NTA must be continuously added to the reactor at a feed rate that maintains FeIII in the form of soluble FeIIINTA complexes with a minimum concentration of uncomplexed NTA in solution.
4.
Conclusions
The data obtained in the present work showed that the catalyzed decomposition of H2O2 by FeIIINTA complexes represents the rate limiting step for the decomposition of H2O2 by FeIINTA complexes. Spectrophotometric studies reveal that addition of H2O2 to FeIIINTA solutions immediately leads to the formation of transient species (probably peroxocomplexes of FeIIINTA) which absorb light in the region 350e600 nm. All the kinetic data show that the initial rates of decomposition of H2O2 by FeIIINTA followed a pseudo-first-order kinetic law with respect to H2O2 concentration. Apparent first-order kinetic rate constants were found to be at a maximum at pH z 8 and to depend on the initial concentration of Fe(III) and on the molar ratios of reactants ([NTA]T,0/[Fe(III)]T,0 and [H2O2]0/ [Fe(III)]T,0). These data suggest that the kinetic modeling of the rate of decomposition of H2O2 by FeIIINTA over a wide range of experimental conditions should be rather complicated and that additional research is required prior the validation of a reaction kinetic model in order to determine rate constants or equilibrium constants of many elementary reactions involved in the FeIINTA/H2O2 or FeIIINTA/H2O2 systems.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
In addition, the data confirmed that the FeIIINTA/H2O2 system is not a true catalytic system because the progressive degradation of the organic ligand during the course of the reaction leads to the formation of much less active precipitates of oxyhydroxides at neutral pH. These data also suggest that the use of the FeIIINTA/H2O2 system for the oxidation of pollutants in industrial wastewater must be operated at pH z 8 and in continuous flow reactors with a continuous introduction of NTA to avoid precipitation of Fe(III) and of H2O2. For each application, the injection doses of reactants (NTA and H2O2) must be optimised in order to minimise the scavenging effects of an excess of NTA and H2O2 toward hydroxyl radicals and to minimise the consumption of reactants.
Appendix. Supplementary material Supplementary data related to this article can be found online at doi:10.1016/j.quascirev.2011.08.009.
references
Bamnolker, H., Cohen, H., Meyerstein, D., 1991. Reactions of low valent transition-metal complexes with hydrogen peroxide. Are they “Fenton-like” or not? 3. the case of Fe(II) [N(CH2CO2)3](H2O)2. Free Radical Research Communications 15 (4), 231e241. Bautista, P., Mohedano, A.F., Casas, J.A., Zazo, J.A., Rodriguez, J.J., 2008. An overview of the application of Fenton oxidation to industrial wastewaters treatment. Journal of Chemical Technology and Biotechnology 83, 1323e1338. Behar, D., Czapski, G., Duchovny, I., 1970. Carbonate radical in flash photolysis and pulse radiolysis of aqueous carbonate solutions. Journal of Physical and Chemical 74, 2206e2210. Bielski, B.H.J., Cabelli, D.E., Arudi, R.L., Ross, A.B., 1985. Reactivity of HO2/O2e radicals in aqueous solution. Journal of Physical and Chemical Reference Data 14 (4), 1041e1100. Borggaard, O.K., 1972. Polarographic determination of diffusion coefficients of hydrogen peroxide and iron chelates and rate constants of hydroxyl radical reactions. Acta Chemica Scandinavica 26 (8), 3393e3394. Borggaard, O.K., Farver, O., Andersen, V., 1971. Polarographic study of the rate of oxidation of iron(II) chelates by hydrogen peroxide. Acta Chemica Scandinavica 25 (9), 3541e3543. Bossmann, S.H., Oliveros, E., Go¨b, S., Siegwart, S., Dahlen, E.P., Payawan, A.L., Straub, M., Wo¨rner, M.B., Braun, A., 1998. A new evidence against hydroxyl radicals as reactive intermediates in the thermal and photochemically enhanced Fenton reactions. Journal of Physical and Chemical A 102, 5542e5550. 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. Journal of Physical and Chemical Reference Data 17 (2), 513e886. Cabelli, D.E., Bielski, B.H.J., 1990. In: Packer, L., Glazer, A.N. (Eds.), The Use of Polyaminocarboxylates (EDTA, DTPA, etc.) as Metal Chelators in methods in Enzymology, Oxygen Radicals in Biological systems. Academic Press, San Diego, CA, pp. 116e120. Chen, D., Martell, A.E., McManus, D., 1995. Studies on the mechanism of chelate degradation in iron-based, liquid redox H2S removal process. Canadian Journal of Chemistry 73, 264e274.
5663
Dao, Y.H., De Laat, J., 2011. Hydroxyl radical involvement in the decomposition of hydrogen peroxide by ferrous and ferricnitrilotriacetate complexes at neutral pH. Water Research 45, 3309e3317. De Laat, J., Gallard, H., 1999. Catalytic decomposition of hydrogen peroxide by Fe(II) in homogenous aqueous solution: mechanism and kinetic modelling. Environmental Science Technololgy 33, 2726e2732. De Laat, J., Le, T.G., 2005. Kinetics and modelling of the Fenton-like reaction in the presence of sulfate in acidic aqueous solutions. Environmental Science and Technololgy 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. Applied Catalysis B: Environmental 66, 137e146. Demmink, J.F., Beenackers, A.A.C.M., 1997. Oxidation of ferrous nitrilotriacetic acid with oxygen: a model for oxygen mass transfer parallel to reaction kinetics. Industrial and Engineering Chemistry Research 36, 1989e2005. Draganic, Z.D., Negron-Mendoza, A., Sehested, K., Vujosevic, S.I., Navarro-Gonzales, R., Albarran-Sanchez, M.G., Draganic, I.G., 1991. Radiolysis of aqueous solutions of ammonium bicarbonate over a large dose range. Radiation Physics and Chemistry 38, 317e321. Eisenberg, G.M., 1943. Colorimetric determination of hydrogen peroxide. Industrial and Engineering Chemistry 15, 327e328. Francis, K.C., Cummins, D., Oakes, J., 1985. Kinetic and structural investigation s of [FeIII(edta)] e [edta ¼ ethylenediamine-tetraacetate(4-)] catalysed decomposition of hydrogen peroxide. Journal of the Chemistry Society Dalton 3, 493e501. 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 Journal of Chemistry, 263e268. Gallard, H., De Laat, J., Legube, B., 1999. Spectrophotometric study of the formation of iron(III)-hydroperoxy-complexes in homogeneous aqueous solutions. Water Research 33, 2929e2936. Gilbert, B.C., Jeff, M., 1988. In: Rice- Evans, C., Dormandy, T. (Eds.), Free Radicals: Chemistry, Pathology and Medicine, pp. 25e49. Howsawkeng, J., Watts, R.J., Washington, D.L., Teel, A.L., Hess, T. F., Crawford, R.L., 2001. Evidence for simultaneous abioticbiotic oxidations in a microbial-Fenton’s system. Environmental Science and Technology 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. Environmental Science and Technology 37, 2734e2742. 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. Environmental Science and Technology 42, 7424e7430. Keenan, C.R., Sedlak, D.L., 2008. Ligand-enhanced reactive oxidant generation by nanoparticulate zero-valent iron and oxygen. Environmental Science and Technology 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. Journal of Korean Industry and Engineering Chemistry 12, 787e792. Kurimura, Y., Ochiai, R., Matsuura, N., 1968. Oxygen oxidation of ferrous ions induced by chelation. Bulletin of the Chemical Society of Japan 41, 2234e2239. Lati, J., Meyerstein, D., 1978. Oxidation of first-row bivalent transition-metal complexes containing ethylenediaminetetraacetate and nitrilotriacetate ligands by free radicals: a pulse-radiolysis study. Journal of the Chemical Society, 1105e1118.
5664
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 5 4 e5 6 6 4
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. Environmental Science and Technology 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. Journal Molecular Catalysis, A: Chemical 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. Industrial and Engineering Chemistry Research 46, 7984e7992. Motekaitis, R.J., Martell, A.E., 1994. The iron(III) and iron(II) complexes of nitrilotriacetic acid. Journal of Coordination. Chemistry 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. Science and Technology 40, 2778e2783. Pera-Titus, M., Garcıa-Molina, V., Ban˜os, M.A., Gime´nez, J., Esplugas, S., 2004. Degradation of chlorophenols by means of advanced oxidation processes: a general review. Applied Catalysis B: Environmental 47, 219e256. Pignatello, J., Oliveros, E., MacKay, A., 2006. Advanced oxidation process for organic contaminant destruction based on the Fenton reaction and related chemistry. Critical Reviews in Environmental Science and Technology 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 Research 43, 684e694. Rivas, F.J., Beltran, F.J., Frades, J., Buxeda, P., 2001. Oxidation of p-hydroxybenzoic acid by Fenton’s reagent. Water Research 35, 387e396. Rush, J.D., Koppenol, W.H., 1988. Reactions of FeIInta and FeIIedda with hydrogen peroxide. Journal of American Chemical Society 110 (15), 4957e4963. Sahul, K., Sharma, B.K., 1987. Gamma radiolysis of nitrilotriacetic acid (NTA) in aqueous solutions. Journal of Radioanalytical and Nuclear Chemistry 109, 321e327.
Sanchiz, J., Esparza, P., Dominguez, S., Brito, F., Mederosa, S., 1999. Solution studies of complexes of iron(III) with iminodiacetic, alkyl-substituted iminodiacetic and nitrilotriacetic acids by potentiometry and cyclic voltammetry. Inorganica Chimica Acta 291, 158e165. Sharma, B.K., Sahul, K., 1982. Co-60 gamma radiolysis of iron(III)nitrilotriacetate in aqueous solutions. Radiation Physics and Chemistry 20, 341e346. 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. Journal of Agricultural Food Chemistry 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. Journal of Agricultural Food Chemistry 41, 308e312. Tachiev, G., Roth, J.A., Bowers, A.R., 2000. Kinetics of hydrogen peroxide decomposition with complexed and “free” iron catalysts. International Journal of Chemical Kinetics 32, 24e35. Tamura, H., Goto, K., Yotsuyanagi, T., Nagayama, M., 1974. Spectrophotometric determination of iron(II) with 1.10phenanthroline in the presence of large amounts of iron(III). Talanta 21, 314e318. Walling, C., Kurz, M., Schugar, H.J., 1970. The iron(III)ethylenediaminetetraacetic acid-peroxide system. Inorganic Chemistry 9, 4931e4937. Walling, C., Partch, R.E., Weil, T., 1975. Kinetics of the decomposition of hydrogen peroxide catalyzed by ferric ethylenediaminetetraacetate complex. Proceedings of the National Academy Sciences of the United States of America 72, 140e142. 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. Journal of Molecular Catalysis A: Chemical 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. Journal of American Chemical Society 113, 7588e7593.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Operational aspects of the desulfurization process of energy gases mimics in biotrickling filters5 Marc Fortuny a,1, Xavier Gamisans b, Marc A. Deshusses c, Javier Lafuente a, Carles Casas a, David Gabriel a,* a
Department of Chemical Engineering, Universitat Auto`noma de Barcelona, Edifici Q, Campus de Bellaterra, 08193 Bellaterra, Spain Department of Mining Engineering and Natural Resources, Universitat Polite`cnica de Catalunya, Manresa, Spain c Department of Civil and Environmental Engineering, Duke University, Durham, NC, USA b
article info
abstract
Article history:
Biological removal of reduced sulfur compounds in energy-rich gases is an increasingly
Received 29 April 2011
adopted alternative to conventional physicochemical processes, because of economical
Received in revised form
and environmental benefits. A lab-scale biotrickling filter reactor for the treatment of high-
13 August 2011
H2S-loaded gases was developed and previously proven to effectively treat H2S concen-
Accepted 16 August 2011
trations up to 12,000 ppmv at gas contact times between 167 and 180 s. In the present work,
Available online 24 August 2011
a detailed study on selected operational aspects affecting this system was carried out with the objective to optimize performance. The start-up phase was studied at an inlet H2S
Keywords:
concentration of 1000 ppmv (loading of 28 g H2S m3 h1) and inoculation with sludge from
Desulfurization
a municipal wastewater treatment plant. After reactor startup, the inlet H2S concentration
Hydrogen sulfide
was doubled and the influence of different key process parameters was tested. Results
Biotrickling filter
showed that there was a significant reduction of the removal efficiency at gas contact times
Startup
below 120 s. Also, mass transfer was found to be the main factor limiting H2S elimination,
Process performance
whereas performance was not influenced by the bacterial colonization of the packed column after the initial startup. The effect of gas supply shutdowns for up to 5 days was shown to be irrelevant on process performance if the trickling liquid recirculation was kept on. Also, the trickling liquid velocity was investigated and found to influence sulfate production through a better use of the supplied dissolved oxygen. Finally, short-term pH changes revealed that the system was quite insensitive to a pH drop, but was markedly affected by a pH increase, affecting both the biological activity and the removal of H2S. Altogether, the results presented and discussed herein provide new insight and operational data on H2S removal from energy gases in biotrickling filters. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Energy rich off-gases such as biogas are sometimes not used for electric power generation due to the presence of corrosive
compounds, such as reduced sulfur compounds (RSC) (Ross et al., 1996; Tchobanoglous et al., 2003). Among those RSC, hydrogen sulfide (H2S) is one of the most commonly reported impurities. H2S concentrations in biogas can range from 0.1 to
5 We dedicate this article to the memory of Carles Casas Alvero, a valued colleague, proficient researcher and skilled educator who passed away on July 20, 2010. * Corresponding author. Tel.: þ34 935811587; fax: þ34 935812013. E-mail address: [email protected] (D. Gabriel). 1 Present address: Aeris Tecnologies Ambientals S.L., Edifici Eureka, Parc de Recerca de la UAB, 08193 Barcelona, Spain. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.029
5666
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
extremely high values of 2% v/v (1000e20,000 ppmv), whereas the specifications for the maximum content of H2S in typical biogas-burning engines are in the range of 0.02e0.05% v/v (200e500 ppmv). At present, biogas energy recovery is becoming more and more interesting due to increasing environmental and economical constraints associated to fossil fuels. Furthermore, an increasing number of solid and liquid wastes management facilities (biomethanation plants) are being installed with biogas production as the main economical benefit. So far, biological sulfide removal has usually been applied to odor control (Yang and Allen, 1994; Devinny et al., 1999; Gabriel and Deshusses, 2003; Gonza´lez-Sa´nchez et al., 2008). However, the growing interest in biological treatment alternatives has lead to an increasing number of studies in the recent years where these techniques are applied to the treatment of highly-loaded off-gases (Buisman et al., 1989; Bailo´n, 2005; van den Bosch et al., 2007; Fortuny et al., 2008). After absorption, treatment and biodegradation of H2S in bioreactors occur according to the following overall reactions (Kuenen, 1975): H2 S4HS þ Hþ
ðnon biologicalÞ
HS þ 0:5O2 /S0 þ OH
HS þ
2O2 /SO2 4
þH
þ
ðbiologicalÞ ðbiologicalÞ
(1)
ðbiologicalÞ
2 þ S2 O2 3 þ 2O2 þ H2 O/2SO4 þ 2H
ðnon biologicalÞ ðbiologicalÞ
Materials and methods
2.1.
Experimental setup
(3)
(4)
All four reactions above result in pH changes, as do parallel abiotic reactions such as oxidation of sulfide to thiosulfate (Eq. (5)), which in turn can also be biologically oxidized to sulfate (Eq. (6)). þ 2H2 S þ 2O2 /S2 O2 3 þ H2 O þ 2H
2. (2)
Also, depending on the redox conditions, further oxidation to sulfate can take place if sulfide is limited but elemental sulfur (S0) is present (Kuenen, 1975): þ S0 þ 3=2O2 þ H2 O/SO2 4 þ 2H
concentrations when validating an integrated analyzer for online process monitoring consisting of a Flow Injection Analyzer coupled to a Continuous Flow Analyzer with a previous Gas-Diffusion step (FIA/GD-CFA). The results showed that a slow drop in H2S removal was caused by progressive sulfide accumulation in the liquid phase. That system was operated at high H2S loadings (162 g H2S m3 h1) and the performance was limited by the biological oxidation of H2S. As a follow up, this study was directed toward optimizing the start-up phase using different inoculation methods and an improved reactor design. A second objective was to acquire a deeper knowledge of the influence of key process parameters when operating the system at less extreme conditions (lower H2S loading rates or inlet concentrations) than those previously tested (Fortuny et al., 2008). Thus, short-term experiments targeting the gas empty bed residence time (EBRT), the trickling liquid velocity (TLV), operating pH and H2S supply shutdowns were carried out and the response of the bioreactor was determined. Also, the influence of the H2S loading rate and operating conditions changes on the pH and biological activity were investigated.
(5)
(6)
Also, abiotic polysulfide formation and oxidation under alkaline conditions as described by van den Bosch et al. (2007) and Gonza´lez-Sa´nchez et al. (2008) may occur. These highlight the complex relationships between oxygen availability, pH and sulfide oxidation processes; a better understanding of the effects of these parameters is required for improved system design. In a preliminary study (Fortuny et al., 2008), the technical feasibility of using a single lab-scale biotrickling filter for the treatment of off-gases containing high concentrations of H2S was demonstrated. Preliminary results on the system robustness when exposed to short-term perturbations, and the relationship between the choice of packing type and the reactor long-term performance were discussed. Also, the inoculation procedure and start-up phase were studied. Furthermore, Montebello et al. (2010) studied the reactor performance when exposed to increased inlet H2S
A lab-scale prototype reactor described in details elsewhere (Fortuny et al., 2010) was used for this study. In short, the biotrickling filter had an inner diameter of 7.1 cm, a packed bed height of 50 cm and a total liquid volume of 2 L. The packing was HD-QPAC (Lantec Products Inc., Agoura Hills, CA, USA) with a 4 4 mm (0.1600 0.1600 ) grid opening cut to tightly fit inside the reactor. Except for the startup period (see Section 2.3), the bioreactor was continuously operated at an inlet H2S concentration of 2000 ppmv (corresponding to a loading of 56 g H2S m3 h1), an EBRT of 180 s, an average liquid hydraulic retention time (HRT) of 51 6 h and a TLV of 3.8 m h1 (liquid flow of 255 ml min1). A pH range of 6e6.5 was maintained by automated addition of NaOH 1 M as needed. Aerobic conditions in the liquid phase were ensured by continuous air addition at an O2/H2S supplied ratio of 23.6 (v v1) through a diffuser located in an oxygenation compartment installed in the recycle line (Fortuny et al., 2010). The conditions were only altered during the start-up phase (see Section 2.3) and for short-term exposures to higher loading rates (LR) up to 400 g H2S m3 h1 and for biomass sampling episodes. Metered amounts of H2S, N2 and air using digital mass flow controllers (Bronkhorst, The Netherlands) were used to simulate a controlled biogas inflow. Mineral medium (MM) as described elsewhere (Fortuny et al., 2010) and a solution of NaHCO3 (21 g L1) as inorganic carbon source were continuously fed at a rate of 0.8 and 0.4 L day1, respectively.
2.2.
Analytical methods
Continuous monitoring of outlet H2S concentration was performed using an electrochemical H2S sensor (Sure-cell, Euro-
5667
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
Gas Management Services LTD, UK) calibrated up to 300 2 ppmv. In order to measure higher H2S concentrations, a mass flow controller (Bronkhorst, The Netherlands) was used to precisely and continuously dilute the analyte flow with air. Dilution ratios between 1:20 and 1:5 were used. On-line liquid phase monitoring included pH and oxidationreduction potential (ORP) (PH 28, Crison Instruments, Spain) and dissolved oxygen (DO) (oxi340i, WTW, Germany) measurements. Also, daily samples of the purge flow were taken for ionic sulfur species and total inorganic carbon (TIC) analysis, using an ICS-1000 Ion Chromatography system with an IonPac AS9-HC column (Dionex Corporation) and a TIC-TOC 1020 analyzer (IO Analytical) respectively. Biomass concentration in the liquid phase was measured as mg N L1 according to van den Bosch et al. (2007). S0 concentration in the liquid recirculation was also measured according to Goehring and Helbing (1949).
2.3.
Inoculation and start-up
Reactor inoculation was carried out using aerobic sludge from a local municipal wastewater treatment plant (MWWTP) diluted 1:1 with MM, thus having a final volatile suspended solids (VSS) concentration of 1.9 g L1. During the start-up phase the inlet H2S concentration was set to 1000 ppmv (28 g H2S m3 h1), the pH setpoint to 6.5e7 in order to match the pH of the original inoculum, and the O2/H2S supplied ratio to 15.7 (v v1). During the first four days no new MM was supplied, though the NaHCO3 was fed to the reactor to avoid carbon limitation. 10% of the liquid volume needed to be removed twice (on the second and third days) in order to keep the liquid volume constant. Once the start-up phase was over (after 5 days), the operating conditions were set as previously described (see Section 2.1).
2.4.
Specific experiments
The maximum elimination capacity (EC) and the effect of reduced EBRTs were assessed 1 and 12 months after reactor
7.5
inoculation in order to asses the evolution of the bioreactor treatment capacity over time. The EBRT was decreased stepwise hourly from 180 s down to 30 s through gas flow increases at a constant inlet H2S concentration of 2000 ppmv and an O2/ H2S supplied ratio of 23.6 (v v1). This corresponds to LR increases from 55 to 334 g H2S m3 h1. Also, the effect of a stepwise increase of the trickling velocity at a constant EBRT (180 s), LR of 84 g H2S m3 h1 (3000 ppmv), HRT of 10 h and O2/ H2S supply ratio of 23.6 (v v1) was determined. The TLV was increased stepwise from 0.52 to 20 m h1 every 48 h (i.e., about 5 HRT) and the reactor response was monitored. To test the effect of gas supply shutdowns, the biogas mimic supply was stopped while the air flow, liquid recirculation, the purge and the make-up water flows were all kept constant. Since inorganic carbon in a full-scale system would be normally supplied via the gas, the HCO 3 supply in the make-up water was also discontinued while the gas supply was stopped. Finally, short-term, large pH variations were introduced in the system, using either HCl or NaOH 1 M and the pH controller, to assess the impact of pH shocks. Initially, a pH drop down to 2.5 was imposed and kept constant for a period of 34 h prior to resuming normal pH conditions (pH 6e6.5). Afterward, a pH increase up to 9.5 was imposed and maintained for 24 h before returning the pH to its normal setpoint. In both cases, a shorter HRT of 19 1 h was used in order to shorten the reactor response time.
3.
Results and discussion
3.1.
Inoculation and start-up
After 1 h of operation at 1000 ppmv inlet concentration, H2S was already detected in the outlet gas stream. This illustrates the low sorption capacity of the system, even when working at constant pH of 7 (Fig. 1). This rapid breakthrough pattern is consistent with the high pKa and relatively unfavorable gasliquid partition of H2S (pKa1 ¼ 6.9; pKa2 ¼ 12.8;
100 2000 90
1800
80 6.5
5.5 5.0 4.5 4.0
RE (%)
pH
6.0
1600
70 60
RE [H2 S]in
50
[H2 S]out
1400 1200 1000
pH
40
800
30
600
20
400
10
200
0
[H2 S] (ppmv )
7.0
0 0
1
2
3
4
5
6
7
8
9
10
Time (days) Fig. 1 e On-line monitoring of the pH, H2S inlet and outlet concentrations and removal efficiency (RE) during the start-up phase of the bioreactor.
5668
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
dimensionless H ¼ 0.39). Even so, some biological activity coupled to absorption of H2S into the fresh NaHCO3 solution allowed RE to stay within 55e65% during the first day of operation. The average DO concentration was 0.4 mg L1 (data not shown) which could have been limiting, and thus was increased to 2 mg L1 on day 2 by increasing the O2/H2S supplied ratio from 15.7 to 23.6 (v v1). After the first day, the RE progressively increased; H2S outlet concentrations were below the detection limit of the H2S sensor (30 ppmv, corresponding to RE > 97%) already by day 2. In parallel, a progressive shift in the sulfur species composition in the liquid phase, i.e. from primarily reduced S species (H2S (aq) and 2 HS) to more oxidized ones (S2 O2 3 and SO4 ) occurred due to increasing biological activity as well as to some chemical oxidation (Steudel, 2004). As shown in Fig. 2, there was an initial accumulation and subsequent depletion of thiosulfate and inorganic carbon during the first two days, concurrent with a progressive accumulation of sulfate in the liquid phase. This adds further evidence that hydrogen sulfide sorption and chemical oxidation (Eq. (5)) predominated during the first two days, but were rapidly surpassed by biological oxidation (Eqs. 2, 3 and 6), which became the main mechanism for removal from day 3 onwards. The TIC profile was also in agreement with the above explanation. For the first five days, the carbonate supply was kept constant at 0.9 0.1 g C-NaHCO3 g1 SeH2S with minimum purging of the trickling liquid (see methods). Thus, despite of the increased CO2 stripping because of the increase of the O2/H2S supplied ratio on day 2, the TIC decrease observed after day 2 (Fig. 2) was most likely linked to an increase in the biological uptake due to onset of H2S removal. Thus, a very short start-up phase of only 2 days when based on H2S gas concentrations or about 5 days when considering liquid phase concentrations was obtained. Moreover, even after doubling the LR up to 56 g H2S m3 h1 (i.e., inlet H2S of 2000 ppmv) on day six, the system performance in terms of RE remained high (Fig. 1). A start-up phase of 3e5 days is a much shorter than observed in previous experiments (Fortuny et al., 2008). Such a fast start-up was
50
1000
45
TIC SO42-
40
S2 O3
900 800 700
30
600
25
500
20
400
15
300
10
200
5
100
0
0 0
2
4
6
8
10
12
14
16
18
20
Time (days)
Fig. 2 e Sulfate, thiosulfate and total inorganic carbon profiles during the first 20 days of operation. Day zero concentrations correspond to the inoculum (MWWTP sludge diluted 1:1 with MM) concentrations. The arrow indicates the beginning of the liquid phase renewal.
-1
35
S-SO4 2- (mg L )
-1
TIC, S-S2 O3 2- (mg L )
2-
attributed to two main factors. First, the pH control ensured a constant operation at a pH between 6.5 and 7.0 (Fig. 1), which was the same pH as that of the original inoculum. The pH control also avoided any pH increase resulting from the addition of NaHCO3 in the early stages of the system operation, when there was little sulfate production to balance the pH. Second, heavy inoculation with the MWWTP sludge may have played an important role in the startup process. It has been previously described (Prado et al., 2005), that a high biomass concentration, like in MWWTP sludge, may facilitate biofilm formation onto a new packing material. Overall, this results show that specific sulfide oxidizing cultures are not needed in order to start-up a biological sulfide treatment system despite what has been reported by other authors (Koe and Yang, 2000; Sercu et al., 2004; Duan et al., 2006). Sludge from MWWTP works well as an inoculum because of its high microbial diversity (Maestre et al., 2010) and adding a high biomass concentration in the biotrickling filter under suitable conditions ensures that sulfide oxidizing organisms will rapidly establish a thriving community.
3.2.
Maximum EC and effect of the EBRT
Fig. 3 shows real time data for the first determination of the maximum EC carried out after one month of operation, while Fig. 4 (run 1) reports the elimination capacities obtained at the end of each step change as a function of the EBRT and the LR. The results show that even after decreasing the EBRT to 120 s (i.e., a LR of 84 g H2S m3 h1) the reactor was able to maintain an average RE of 97.7 0.3%. At an EBRT of 90 s, the RE only dropped to 88.6 0.5%. However, when the EBRT was decreased further, there was and important effect on H2S removal. The RE reached an average value of 39.7 0.9% at an EBRT of 30 s (i.e., a LR of 334 g H2S m3 h1). Examination of Figs. 3 and 4 shows that there was practically no RE reduction at LR up to 84 g H2S m3 h1 (or EBRT of 120 s). This would allow significant reduction of the EBRT without any effect on H2S removal at 2000 ppmv inlet concentration. Another important observation is the rapid recovery after returning the system to its original EBRT of 180 s (at time 288 min). The small ORP drop during the experiment (50 mV, Fig. 3) provides information on the possible accumulation of sulfide in the liquid. Montebello et al. (2010) showed that ORP values around 50 mV corresponded to very low sulfide liquid concentrations (<1 mg L1) whereas significant sulfide accumulation caused ORP to drop below 200 mV. Thus, the 50 mV drop suggests that there was not accumulation of sulfide in the liquid phase, indicating that sulfide was oxidized rather than being absorbed. It is worth mentioning that low concentrations of thiosulfate were detected in the liquid phase, but the amount corresponded to less than 1% of the total amount of H2S removed. Thus, abiotic oxidation was negligible, even during the high loading periods. The absence of sulfide accumulation during the high-load periods (LR up to 334 g H2S m3 h1) indicates that the rate limiting step was mainly mass transfer. This is an interesting finding which contrasts with previous results (Montebello et al., 2010) in which the same reactor became kinetically limited rather than mass-transfer limited when exposed to LR
5669
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
100
200
90
180
80
0
160 -25
70 60 RE
120
EBRT ORP
50
100
40
-50 -75
ORP (mV)
140
EBRT (s)
RE (%)
25
80
30
-100 60
20
-125
40
10 0 -200 -150 -100 -50
20 0
50
-150
100 150 200 250 300 350 400
Time (min) Fig. 3 e Removal efficiency (RE), redox potential (ORP) and empty bed gas residence time (EBRT) during run 1 of the EBRT experiment. The dashed line indicates beginning of experiment. above 162 g H2S m3 h1 by only increasing the inlet concentration at a constant EBRT. Therefore, the system becomes kinetically limited when high LR are achieved by an increasing concentration, but is mass transfer limited when high LR are achieved by reducing the EBRT. A maximum EC of 126 g H2S m3 h1 was achieved (Fig. 4B), which compares favorably to other reported maximum EC (Koe and Yang, 2000; Sercu et al., 2004; Duan et al., 2006). This is remarkable value for a bioreactor that had been operating for only a month. It demonstrates that reactor design, inoculation and operating conditions during startup were close to optimum. The same determination of EC was repeated after one year of reactor operation (run 2). A 14% increase in the maximum EC was observed, reaching an average value of 144 4 g H2S m3 h1 (Fig. 4B). Since no biological limitation was observed in run 1, the increase in the maximum EC of run 2 must be related to an improved H2S mass transfer. Pollutant transfer occurs through the gas-liquid and gas-biofilm 100
contact, therefore it is likely that increased bacterial coverage of the packing occurred after a year of operation, which resulted in a greater interfacial surface-to-volume ratio inside the reactor. Another possible contributing factor is that slight accumulation of (bio)solids onto the clean, openstructure of the packing material could cause a slight gas velocity increase, which in turn would increase pollutant mass transfer. The former hypothesis seems more likely than the latter, given the fact that wetting in biotrickling filters is often incomplete (Kim and Deshusses, 2008), which probably extends the time required for complete bacteria colonization of the packing.
3.3.
Effect of the trickling liquid velocity
The TLV is an important parameter for the attachment (and shear) of biomass onto the packing material, for proper gasliquid mass transfer and for S0 flushing in case of accumulation. As shown in Fig. 5, no difference in the performance was 200
A
B
175
EC (g H2 S m -3 h -1 )
80
60
40
Run 1 Run 2
150 125 100 75
Run 1 Run 2 EC = Load
50
20
25 0
0 0
20
40
60
80
100
120
EBRT (s)
140
160
180
200
0
50
100
150
200
250
300
350
400
LR (g H2 S m-3 h-1 )
Fig. 4 e Results for both experimental runs of the EBRT experiment. A) Removal efficiency (RE) versus applied empty bed residence time (EBRT). B) Elimination capacity (EC) versus applied loading rate (LR).
5670
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
110 100 90 80 70 60 50
100
18
80
RE Trickling liquid velocity Average O2 load
70
S-SO42-
90
16 14 12
60
10 50
8
40
6
30 20
4
10
2
Trickling liquid velocity (m h-1)
120
20
110
RE (%), O2 load (g O2 m-3h-1)
-3 sulfate production (g S-SO42-m h-1)
130
0
0
94
96
98
100
102
104
106
108
Time (days) Fig. 5 e Variation of the trickling liquid velocity (TLV), removal efficiency (RE), sulfate production and oxygen load supplied to the packed bed during the 14 days experiment.
observed when changing the TLV within the tested range (0.51e19 m h1) at a LR of 84 g H2S m3 h1. However, one effect of changing the TLV was on oxygen transfer, since changing the TLV altered the liquid retention time in the oxygenation compartment located in the recycle line (Fortuny et al., 2010). Both the DO in the liquid phase (data not shown) and thus the oxygen load supplied to the packed bed (in terms of g DO m3 h1actually available and supplied via the recirculation liquid) were increased with increases in the TLV (Fig. 5). At TLVs lower than the standard conditions (TLV < 3.8 m h1), the DO in the recirculation liquid increased up to 4.5 mg L1, even if the DO load supplied to the packed bed was significantly reduced (79% and 20% reduction for TLV ¼ 0.51 and 2.04 m h1, respectively, compared to the oxygen load under standard conditions). Conversely, TLVs greater than 3.8 m h1 which led to a DO reduction to an average concentration of 2.5 mg/L, corresponded to an oxygen load increase in proportion to the TLV. Consequently, raising the TLV caused a net increase in the O2 availability and resulted in increased sulfate production (Fig. 5). Montebello et al. (2010) already pointed out to the existence of DO gradients throughout the bed which could be partially reduced by increasing the TLV due to a larger penetration of DO throughout the bed depth when liquid and gas flows operate in counter-current mode. It is relevant to stress that this occurred without any change in the air gas flow rate or oxygen amount supplied to the system and thus, from an operational point of view, such increased sulfate production is an interesting strategy as it could be used to reduce S0 accumulation without the need for further dilution of the raw gas. The increase in TLV above typical values for biological systems (1e12 m h1 see Kim and Deshusses, 2008) had been expected to cause a direct effect on (bio)solids shear and wash out. Surprisingly, monitoring of the S0 and biomass concentrations in the recirculation liquid did not show a relationship between S0 or biomass concentration and TLV (data not
shown). These results indicate that increasing the TLV, in the short run (hours or days), may not be an efficient measure to scour the S0 accumulated on the packing material. Once S0 accumulates, it forms solid aggregates on the packing that are not easily removed. Further research is needed to establish optimum TLV or flushing methods to prevent S0 accumulation onto the packing.
3.4.
Effect of intermittent H2S supply
Starvation periods were implemented to simulate industrial operation with short-term shutdowns. The gas supply was discontinued for 2.5 and 5 days and the response of the reactor was monitored. Only a very small and brief reduction of the RE was detected during the first hours after resuming the gas supply. The maximum RE reduction was 4% after the 2.5-day stop (results not shown) and 7% after the 5-day shutdown (Fig. 6A) and a steady removal of over 99% was reestablished within 4 h. Shortly after the H2S supply was resumed (day 86.7, Fig. 6A) the ORP dropped to about 150 mV, which indicated a slight sulfide accumulation, in agreement with the transient drop in RE. However, ORP returned to pre-shutdown values (between 50 mV) within less than 24 h, indicating absence of dissolved sulfide and presence of oxygen (DO 0.5e2 mg L1). The ORP pattern after resuming H2S indicated a temporary lag of biological activity followed by a complete recovery. If the culture had been severely affected by the starvation, greater accumulation of sulfide and an ORP drop below 350 mV would have been observed (Montebello et al., 2010). Interestingly, a sulfur mass balance during the starvation period reveals that actual sulfate concentrations were 30e40% higher than expected from simple dilution by the make-up water (Fig. 6B). Because no dissolved sulfide had accumulated, the excess sulfate concentration measured indicates that biological activity did not completely stop during the
5671
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
8.0
A
7.5
150
7.0 100
80
50
6.0 60 5.5 5.0
40
0 -50
ORP (mV)
6.5
pH
LR (g H 2 S m -3 h -1 ), RE (%)
100
200
4.5 -100
LR RE pH ORP
20
4.0 -150
3.5
0
3.0 80
81
82
83
84
85
86
87
88
89
-200
90
Time (days) 800
B 100
LR
700
SO4 2- measured
-1
SO 4 2- (mg S L )
LR (g H 2 S m -3 h -1 )
600
SO4 2- expected
80
500 60 400 300
40
200 20 100 0
0 80
81
82
83
84
85
86
87
88
89
90
Time (days) Fig. 6 e Performance of the reactor during and after the 5-day H2S starvation period. The gas stream was discontinued, but the liquid feed was maintained. A) pH, redox potential (ORP), removal efficiency (RE) and inlet load. B) Inlet load, measured sulfate concentration and expected sulfate concentration (from washout calculation.
starvation period and that oxidation of accumulated S0 inside the reactor occurred, as previously reported (Fortuny et al., 2010). The source of oxygen for the aerobic oxidation of S0 was both via the liquid feed, and possibly diffusion of atmospheric O2 into the biotrickling filter. These results show that H2S shutdowns up to 5 days have no, or very little impact on the long-term reactor operation. Proper control of pH during starvation events is likely to be important for prompt recovery. It is also probable that biological oxidation of accumulated biological S0 during the starvation contributed to the fast recovery of the biotrickling filter, and could possibly allow longer periods of starvation without a greater impact on the reactor performance.
3.5.
Effect of short-term pH changes
Complete biological oxidation of sulfide leads to protons production and thus pH changes. However, it is not known to what extent, pH variations need to be controlled to ensure stable bioreactor operation. Since a pH control failure (lack or excess of control actuation) is relatively plausible during industrial operation, the reactor response to large but fast pH variations was studied. On day 135 the reactor pH was forced to drop to 2.5 for a period of 34 h. Later, after 60 h of operation at normal conditions (i.e., pH 6e6.5), the pH was increased to 9.5 for a 24 h period. As shown in Fig. 7A, the pH drop did not affect the RE of H2S. However, a sulfur mass balance indicated that
5672
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
Fig. 7 e Reactor response to pH drop and pH increase episodes. A) Redox potential (ORP), pH and removal efficiency (RE). B) Sulfate and thiosulfate concentration in the liquid phase.
a 25% reduction in the sulfate production occurred after decreasing the pH (Fig. 7B), probably indicating a reduction in the biological activity. A possibility is that under these conditions, sulfide chemical oxidation to thiosulfate, which is unstable at low pH and chemically reacts to produce S0 contributed to maintain the high RE. This is consistent with the sulfur speciation shown in Fig. 7B, in which neither thiosulfate (chemically converted to S0) nor other sulfurous ionic species accumulated. Inorganic polysulfides were not measured since they only occur at pH above 6 (Steudel, 2004). Sulfate production increased as soon as the pH was returned to its original value and continued to increase throughout the period at which the pH was kept at 6e6.5 (Fig. 7B). It is worth noticing that sulfate production rates greater than the sulfur load ðS SO2 4 =S H2 Sremoved > 100%Þ were encountered, probably due to the oxidation of accumulated S0. The high pH test was conducted next. Immediately after increasing the pH to 9.5 for 24 h, a sharp drop in the ORP
occurred indicating an accumulation of sulfide (Fig. 7A) while a shift in the sulfur mass balance from sulfate to thiosulfate was observed (Fig. 7B). These reveal that a significant slowdown of the biological activity as well as a shift in the metabolism occurred as a result of the pH change. However, at high pH, both physical absorption of H2S and chemical oxidation of dissolved sulfide to thiosulfate (van den Bosch et al., 2008) and possibly (poly)sulfide are favored. Therefore, although the biological oxidation was affected, the removal of H2S only slowly dropped to about 95% at the end of the high pH step. When the pH was returned to 6e6.5, an important transient phase of very low RE (<20%) was observed (Fig. 7A). This was attributed to H2S stripping and is consistent with the concurrent increase in ORP (i.e., decrease in dissolved sulfide) during this short phase. For the next two days, biological activity was probably severely inhibited and the primary means for H2S removal was likely sorption and chemical oxidation to thiosulfate. Thiosulfate concentration slowly
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
decreased because of dilution rather than to biological oxidation since biological oxidation would have led to a sulfate concentration increase. It was not until day 143, i.e., close to 3 days after the perturbation, that biological oxidation to sulfate became again the primary removal means, as shown by the ORP increase (Fig. 7A) and sulfate yields (Fig. 7B). The H2S RE reached over 99% indicating that the reactor had recovered and was operating normally about 3 days after the pH spike. This demonstrates that the bioreactor was much more susceptible to high pH exposure than to low pH changes. Such behavior can be explained by the fact that acidification is the natural consequence of sulfide oxidation and, therefore, most sulfide oxidizers are more tolerant to acidic conditions than to alkaline conditions (Bru¨ser et al., 2004; Syed et al., 2006). Also, a low pH results in a decrease of the dissolved sulfurous species concentration (H2S by stripping and polysulfides and thiosulfate by chemical destabilization), whereas a high pH leads to accumulation of dissolved sulfurous species (sulfides and polysulfides), which can negatively affect the activity of the sulfide degraders.
4.
Conclusions
A detailed study of the operational aspects affecting the biological sweetening of energy gases mimics was carried out in a lab-scale biotrickling filter. Results showed that: B
B
B
B
B
B
Detailed monitoring of the process with on-line H2S gas, ORP and pH sensors combined with off-line analysis of sulfurous species dissolved in the trickling liquid provided a detailed understanding of the phenomena involved during the treatment of H2S in the biotrickling filter. Inoculation of the biotrickling filter with MWWTP sludge led to a very fast (3 days) start-up therefore showing that, provided certain conditions, it is not necessary to obtain a specific culture of sulfide oxidizers to inoculate an H2S degrading system. The EBRT (so reactor volume) can be importantly reduced without significantly affecting the removal of H2S while treating 2000 ppmv H2S. However, excessive EBRT reduction lead to an important RE drop due to mass transfer limitation and not due to biological limitation as observed in our previously research when the LR was increased by increasing the H2S inlet concentration. Gas supply shutdowns for up to 5 days had little effect on the biotrickling filter performance after resuming normal operation. Possibly, controlling pH, addition of make-up water coupled with biological oxidation of S0 present in the biotrickling filter helped maintain a healthy biological activity during the perturbation. A high TLV is recommended because it favors sulfate production through a better use of the oxygen supplied. A short (34 h) but wide (down to 2.5) pH drop was much less aggressive to the biological activity and overall reactor performance than a short (24 h) but also wide (up to 9.5) pH rise. Even so, the robustness of the system allowed for a full recovery after about 48 h of normal operation.
5673
Acknowledgments The Spanish government (MEC) provided financial support through the project CICYT CTM2009-14338-C03-01. The Department of Chemical Engineering at UAB (Universitat Auto`noma de Barcelona) is a unit of Biochemical Engineering of the Xarxa de Refere`ncia en Biotecnologia de Catalunya (XRB), Generalitat de Catalunya. The study sponsors were not involved in: study design; collection, analysis, and interpretation of data; writing of the report; nor in the decision to submit the paper for publication.
references
Bailo´n, L., 2005. Development of a biotrickling filter for the removal of H2S from biogas. In: Proceedings of the II International Congress on techniques for Air Pollution Control. La Corun˜a, Spain, pp. 143e148. Bru¨ser, T., Lens, P.N.L., Tru¨per, H.G., 2004. The biological sulfur cycle. In: Lens, P., Hulshoff Pol, L. (Eds.), Environmental Technologies to Treat Sulfur Pollution. Principles and Engineering. IWA Publishing, London, UK, pp. 47e86. Buisman, C.J.N., Post, R., Ijspeert, P., Geraats, G., Lettinga, G., 1989. Biotechnological process for sulphide removal with sulphur reclamation. Acta Biotechnologica 9, 255e267. Devinny, J.S., Deshusses, M.A., Webster, T.S., 1999. Biofiltration for Air Pollution Control. CRC-Lewis Publishers, Boca Raton, Florida, USA. Duan, H., Koe, LC.C., Yan, R., Chen, X., 2006. Biological treatment of H2S using pellet activated carbon as a carrier of microorganisms in a biofilter. Water Research 40, 2629e2636. Fortuny, M., Baeza, J.A., Deshusses, M.A., Gamisans, X., Casas, C., Lafuente, J., Gabriel, D., 2008. Biological sweetening of energy gases mimics in biotrickling filters. Chemosphere 71, 10e17. Fortuny, M., Guisasola, A., Casas, C., Gamisans, X., Lafuente, J., Gabriel, D., 2010. Oxidation of biologically produced elemental sulfur under neutrophilic conditions. Journal of Chemical Technology and Biotechnology 85 (3), 378e386. Gabriel, D., Deshusses, M.A., 2003. Retrofitting existing chemical scrubbers to biotrickling filters for H2S emission control. Proceedings of the National Academy of Sciences of U.S.A 100 (11), 6308e6312. Gonza´lez-Sa´nchez, A., Revah, S., Deshusses, M.A., 2008. Alkaline biofiltration of H2S odors. Environmental Science & Technology 42 (19), 7398e7404. Goehring, M., Helbing, W., 1949. Journal of Analytical Chemistry V, 129e346. Kim, S., Deshusses, M.A., 2008. Determination of mass transfer coefficients for packing materials used in biofilters and biotrickling filters for air pollution control - 2: development of mass transfer coefficients correlations. Chemical Engineering Science 63 (4), 856e861. Koe, L.C.C., Yang, F., 2000. A bioscrubber for hydrogen sulphide removal. Water Science and Technology 41 (6), 141e145. Kuenen, J.G., 1975. Colourless sulphur bacteria and their role in the sulphur cycle. Plant and Soil 43, 49e76. ´ lvarez Hornos, J., Fortuny, M., Maestre, J.P., Rovira, R., A Lafuente, J., Gamisans, X., Gabriel, D., 2010. Bacterial community analysis of a gas-phase biotrickiling filter for biogas mimics desulfurization through the rRNA approach. Chemosphere 80, 872e880. Montebello, A.M., Baeza, M., Lafuente, F.J., Gabriel, D., 2010. Monitoring and performance of a desulphurization
5674
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 6 5 e5 6 7 4
biotrickling filter with an integrated continuous gas/ liquid flow analyser. Chemical Engineering Journal 165, 500e507. ´ .J., Veiga, M.C., Kennes, C., 2005. Treatment of gas-phase Prado, O methanol in conventional biofilters packed with lava rock. Water Research 39, 2385e2393. Ross, C.C., Drake, T.J., Walsh, J.L., 1996. Handbook of Biogas Utilization, second ed. SERBEP, c/o General Bioenergy, Florence, AL. Sercu, B., Nu´n˜ez, D., van Lagenhove, H., Aroca, G., Verstraete, W., 2004. Operational and microbiological aspects of a bioaugmented two-stage biotrickling filter removing hydrogen sulfide and dimethyl sulfide. Biotechnology and Bioengineering 90 (2), 259e269. Steudel, R., 2004. The chemical sulfur cycle. In: Lens, P.N.L., Hulshoff Pol, L. (Eds.), Environmental Technologies to Treat Sulfur Pollution. Principles and engineering. IWA Publishing, London, UK, pp. 1e31.
Syed, M., Soreanu, G., Falletta, P., Be´land, M., 2006. Removal of hydrogen sulphide from gas streams using biological processes a review. Canadian Biosystems Engineering 48, 2.1e2.14. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. Wastewater Engineering. Treatment and Reuse, fourth ed. McGraw-Hill Companies, New York, NY, USA, pp. 1505e1532. van den Bosch, P.L.F., van Beusekom, O.C., Buisman, C.J.N., Janssen, A.J.H., 2007. Sulfide oxidation at halo-alkaline conditions in a fed batch bioreactor. Biotechnology and Bioengineering 97 (5), 1053e1063. van den Bosch, P.L.F., Sorokin, D.Y., Buisman, C.J.N., Janssen, A.J. H., 2008. The effect of pH on thiosulfate formation in a biotechnological process for the removal of hydrogen sulfide from gas streams. Environmental Science and Technology 42 (7), 2637e2642. Yang, Y., Allen, E.R., 1994. Biofiltration control of hydrogen sulfide.1. Design and operational parameters. Journal of Air and Waste Management 44, 863e868.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 7 5 e5 6 8 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
The rheological behaviour of anaerobic digested sludge J.C. Baudez a,b,*, F. Markis b, N. Eshtiaghi b, P. Slatter b a b
Cemagref, UR TSCF, F-03150 Montoldre, France Rheology and Materials Processing Centre, Dept. of Chemical Engineering, RMIT University, Victoria 3001, Australia
article info
abstract
Article history:
Producing biogas energy from the anaerobic digestion of wastewater sludge is one of the
Received 20 May 2011
most challenging tasks facing engineers, because they are dealing with vast quantities of
Received in revised form
fundamentally scientifically poorly understood and unpredictable materials; while
3 August 2011
digesters need constant flow properties to operate efficiently. An accurate estimate of
Accepted 19 August 2011
sludge rheological properties is required for the design and efficient operation of digestion,
Available online 1 September 2011
including mixing and pumping. In this paper, we have determined the rheological behaviour of digested sludge at different concentrations, and highlighted common
Keywords:
features. At low shear stress, digested sludge behaves as a linear viscoelastic solid, but
Digested sludge
shear banding can occur and modify the apparent behaviour. At very high shear stress, the
Bingham model
behaviour fits well to the Bingham model. Finally, we show that the rheological behaviour
HerscheleBulkley model
of digested sludge is qualitatively the same at different solids concentrations, and depends
Shear banding
only on the yield stress and Bingham viscosity, both parameters being closely linked to the
Viscoelasticity
solids concentration. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Renewable energy is said to be one of the pillars of sustainable management. Biogas from the anaerobic digestion of sewage sludge can provide a clean, easily controlled source of renewable energy from sewage sludge, replacing fossil fuels. However, an accurate estimate of sludge rheological properties is required for the design and efficient operation of the pumping systems which surround anaerobic digesters (Slatter, 1997, 2003). Indeed Tarp and Melbinger (1967) showed the significant advantages of recycling and recirculating digested sludge to mix it with excess sludge, among them an increase in biogas production (Sperry, 1959). The mixture can be concentrated to a much higher solid content than would be possible for the excess sludge alone, and recirculation also facilitates improved mixing efficiency over mechanical stirring. However, the flow rate in the recirculation circuits has to be very large (Appels et al., 2008) and rheology is needed to calculate head losses and pumping power (Slatter, 2001).
Except for the work of Monteiro (1997) who showed that anaerobic digestion induces a decrease of the rheological characteristics of sludge, most investigations on sludge rheology were focused on activated sludge. No reliable data, at high shear rate (within recirculation pipes), can be found in the literature for digested sludge while at low shear rate (within the digester), results are scarce and not always usable. Most of these were obtained by applying shear rate ramps that gave distinct peaks in the flow curve (for example, Ayol, 2006), but Baudez (2006) clearly established that these peaks in the flow curves were principally instrument artefacts, and not material characteristics. However, the work of Ayol et al. (2006) pointed out that with very dilute sludge, the Ostwald model, i.e. a power-law model with no yield stress, gave the best fit. From a physical perspective, digested sludge appears to be a stable suspension with low settling rates (Namer and Ganczarczyk, 1993) and low surface charge (Forster, 2002), implying that interactions are more steric than electrostatic.
* Corresponding author. Cemagref, UR TSCF, F-03150 Montoldre, France. E-mail address: [email protected] (J.C. Baudez). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.035
5676
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 7 5 e5 6 8 0
Decreasing stress ramp to determine the flow curve, starting at a high stress corresponding to a shear rate of approximately 1000 s1 or lower for the less concentrated sludge (to avoid turbulent conditions).
The most important constituents in digested sludge are lipopolysaccharides (Forster, 1983) which are amphiphile lipids with both hydrophilic and hydrophobic heads. These molecules displayed a very intriguing rheological behaviour (Mun˜oz et al, 2000), showing linear viscoelasticity, nonNewtonian viscous flow and shear banding (Miller and Rothstein, 2007). In this paper, our intention is to establish the basic characteristics of the rheological behaviour of digested sludge, with the objective of industrial applications in digester mixing, pumping and pipe flows, meaning that we will focus on short-term behaviour. Short-term behaviour means we will not focus our research on eventual thixotropic effects. As predicted by the literature on amphiphile rheology, we show that digested sludge exhibits linear viscoelastic behaviour at low shear stresses, followed by shear banding phenomena at intermediate stresses, and finally a non-Newtonian fluid behaviour with a yield stress, modelled by a HerscheleBulkley model at intermediate shear rates and by a Bingham model at very high shear rates. We also highlight the fact that the rheological behaviour is qualitatively the same at different solid concentrations, allowing us to define a master curve for which the dimensionless parameters are the yield stress and the Bingham viscosity.
where g represents the strain, s the stress and l ¼ G=m with G and m the usual parameters of a Kelvin-Voigt model. Assuming that the sludge is flowing in its liquid regime above the critical shear stress following a HerscheleBulkley model (Monteiro, 1997), the additional strain can be expressed as:
2.
gðtÞ ¼
3.
Results and discussion
Starting from rest, the shear stress sweep first elicits a linear viscoelastic response from the digested sludge up to a critical shear stress s0 above which the material apparently starts to flow (Fig. 1). In the linear viscoelastic region, the behaviour is modelled by a generalised Kelvin-Voigt model, with a wide relaxation time spectrum modelled by a stretched exponential: 1 gðtÞ ¼ s$ $ 1 exp ðltÞm G
Zt
Material and methods
_ gdu ¼ t0
The digested sludge was sampled at the Mount Martha waste water treatment plant (Melbourne, Victoria, Australia) at the outlet of the digester number 1. Its initial solid concentration was at 18.5 g L1 and was also gently concentrated to 25.5, 32 and 49 g L1 by using a Buchner vacuum. Sludge samples were stored at 4 C for 30 days before experiments, in order to reduce temporal variability. Indeed, even after anaerobic digestion, sludge may not be fully stabilised and organic changes may still occur. By storing the sludge sample for such an extended period, the potential for composition changes is reduced; and we can assume that we used exactly the same material throughout all our experiments. Rheological measurements were performed with a DSR200 instrument from Rheometric Scientific, connected to a temperature controlled water bath. The rheometer was equipped with a cup and bob geometry (inner diameter: 29 mm, outer diameter: 32 mm, length: 44 mm). Temperature was kept at 25 C. To avoid evaporation, sludge was covered with a thin film of immiscible Newtonian oil. Before each measurement, sludge was presheared for 10 min at a shear rate of 1000 s1 then left at rest for 10 min. This procedure allowed us to erase material memory and to have reproducible measurements. Then, different tests were performed: Shear stress sweep, by applying a linear ramp of increasing stress over time. In this test, we changed the time of rest between preshear and shear, from 1 to 60 min in order to investigate structural changes occurring during rest; Creep test, by applying constant shear stress and measuring the corresponding shear strain, at different shear stresses in the linear viscoelastic regime and above;
(1)
1=n Z t 1=n Zt s s0 a$x a$t0 ¼ dx K K t0
t0
a$n ¼ ðt t0 Þ1þ1=n ðn þ 1ÞK
(2)
where a is the slope of the shear stress ramp and t0 the time such that the shear stress equals the yield stress of the HerscheleBulkley model s0 ¼ a$t0 . Thus, the total strain, which predicts the experimental data (Fig. 1), can be expressed as 1 gðtÞ ¼ s$ $ 1 exp ðltÞn þ b$ðt t0 Þ1þ1=n G
(3)
with b ¼ an=ðn þ 1Þ$K Applying this to the experimental data gave a flow behaviour index for the HerscheleBulkley model, n, greater than 1 (Fig. 1), meaning that the digested sludge could apparently be a shear-thickening liquid above s0 , which is unusual. Creep tests confirmed a change in the behaviour above s0 . Below s0 , the strain slowly increased with time, while above s0 , the increase is faster (Fig. 2), both following a power-law with time. However, even for stresses higher than s0 (Fig. 2) the shear strain follows a power-law with a power-law index less than 1, indicating that the shear rate is a decreasing function of time: there is no steady state and so, sludge is restructuring and not flowing (otherwise, the shear rate would have been constant over time for a constant shear stress). The value s0 cannot therefore be considered as a classical yield stress above which digested sludge flows in its liquid regime. These power-law relationships between strain and time are in fact a consequence of a structural relaxation process which occurs during creep (Baudez, 2008). When the time of rest between the preshear and the stress sweep increases, the behaviour is globally the same, with first a linear viscoelastic behaviour (Fig. 3) but the critical shear
5677
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 7 5 e5 6 8 0
Fig. 1 e Strain-stress behaviour of the 4.9% digested sludge. The dashed lined corresponds to the model of (1) with G [ 0.62 Pa, l [ 3.3.10L7 sL1, m [ 0.34, b [ 0.35 sL2, t0 [ 315.5 s, corresponding to a stress equals to 2.13 Pa and n [ 1.30.
stress, s0 decreases with increase of the time of rest, the global elasticity decreases, the mean relaxation time (inverse of l) increases and the strain corresponding to s0 decreases (Fig. 4): the longer the time of rest, the smaller the linear viscoelastic range. At rest, the digested sludge structure became weaker and weaker (decrease of s0 and elasticity) but concurrently the relaxation time increased, indicating an evolution from a viscoelastic material towards a more elastic solid (the decrease of m is faster than the decrease of G). Since this is physically impossible, this observed apparent behaviour is not
Fig. 3 e Strain-stress behaviour when a stress sweep is applied after different time of rest.
representative of the true material behaviour but derives from erroneous interpretation of raw data. Above s0 , experimental results showed the viscosity is globally decreasing, which is inconsistent with the apparent shear-thickening behaviour noted earlier, but oscillations of viscosity regarding shear stress are reported (Fig. 5). These oscillations indicated local minima in the flow curve where apparent shear rate occasionally decreased while shear stress increased. If we assume the relationship between local shear rate and local shear stress is monotonic, then we can write: g_ ¼ f ðsÞ where f is the inverse function of the behaviour law. In a Couette geometry, the shear rate can be expressed as: vðuÞ 5u ¼ g_ local ¼ r$ vr
Fig. 2 e Creep test below, above and equal to the critical shear stress. Here, the critical stress is 2.5 Pa for the 4.9% sludge. The insert is a focus on the strain at the highest strain at longer time, following a power-law with an index smaller than 1.
ZR2 R1
g_ local dr5u ¼ r
ZR
f ðslocal Þ dr r
(4)
R1
Fig. 4 e Evolution of the Kelvin-Voigt model parameters as a function of the time of rest.
5678
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 7 5 e5 6 8 0
Fig. 5 e Stress-viscosity variations highlighted oscillations with the 4.9% digested sludge.
Fig. 7 e Flow curves regarding the concentration of the digested sludge.
where u is the angular velocity, R1 the inner radius and R R1 the thickness of the sheared region. The maximum value of R R1 is R2 R1 , where R2 is the outer radius. The apparent shear rate is calculated from the measured angular velocity, the only raw data measured by the rheometer. The shear rate given by the rheometer is calculated with the assumption of a full shear within the gap. So, if the apparent shear rate decreased, it means the angular velocity decreased. However, because f is a monotonic function, this decrease of u is rather the consequence of a decrease of the effective gap R R1 , implying that shear banding has occurred during the measurement. Such behaviour (shear banding and viscoelastic behaviour) has to be taken into account in digester design and operation, because shear banding means that there is coexistence of both sheared and unsheared zones in the digester, these last being useless, unmixed, dead zones.
According to Moller et al. (2008), the width of the flowing band can be directly related to the macroscopically imposed shear rate. At high shear rates, the whole gap is sheared and when the applied stress is much higher than s0 , the sludge flows normally, with no apparent perturbation effects, allowing us to have achieve reproducible measurements (Fig. 6) with the corresponding smooth classical shape of the flow curve. As expected, the higher the concentration, the thicker the sludge (Fig. 7) but depending on the shear rate range, different well-known models can be used to describe the rheological behaviour of digested sludge. At high shear rates, a basic Bingham model is sufficient (Fig. 8) while at low and intermediate shear rates, HerscheleBulkley and power-law models are more appropriate (Fig. 9). They all represent the same material but can only be used in a specific range of validity, regarding the complexity of the process to be modelled. Thus,
60 Test 1
70
Test 2
50
60
40
Shear stress [Pa]
Shear stress [Pa]
Test 3
30 20
50 40 30 20 10
10
0 0
0 1
10
100
Shear rate [s-1] Fig. 6 e Repeatability of the measurements.
1000
500
1000
1500
Shear rate [s-1] Fig. 8 e At high shear rates, the rheological behaviour can be basically modelled with a Bingham plastic model.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 7 5 e5 6 8 0
Fig. 9 e At low and intermediate shear rates, the HerscheleBulkley model or power-law model are the most suitable. The dashed line represents the power-law model s[2:05,g_ 0:45 .
for pumping where shear rates are very high, a Bingham model would be appropriate since it deals with simple characteristics, i.e. a yield stress and a constant rheogram slope above it. From a more general point of view, in the liquid regime we can summarize the rheological behaviour of digested sludge as a shear-thinning yield stress fluid with a plateau viscosity at high shear rates: s ¼ sc þ h g_ $g_ with h g_ / a0 _ g/N
Moreover, at low and intermediate shear stresses, _ g5hð _ _ gÞzK$ g_ n1 s ¼ sc þ K$g_ n ¼ sc þ hðgÞ$ Thus, for the sake of simplicity, we define the rheological behaviour of digested sludge as follows: s ¼ sc þ K$g_ n1 þ a0 $g_
5679
Fig. 10 e Dimensionless flow curves of the digested sludge at different concentrations.
s K a0 s _ _ s ¼ sc þ K$g_ n þ a0 $g5 ¼ 1 þ $g_ n þ $g5 ¼ 1 þ b$Gn þ G s s s s c c c c n a0 K sc _ b¼ $ G ¼ $g; sc s c a0
(6)
In such a dimensionless form, all the flow curves are similar, independent of solids concentration (Fig. 10). From a physical point of view, this result means that there is some similarity of the network of interactions within the sludge at different concentrations, which is at the origin of the similarity of its macroscopic behaviour. In such suspensions, interactions can be classified into two main groups (Baudez and Coussot, 2001): hydrodynamic interactions (between solid particles and surrounding fluid, here basically
(5)
On our range of data, i.e. below 1000 s1, this model was pffiffiffiffiffiffiffiffiffiffiffi successful. However, if g_ << 1n K=a0 , the HerscheleBulkley model is sufficient to model the behaviour, which corresponds to a shear rate smaller than 565 s1 for the most concentrated sludge and smaller than 145 s1 for the less concentrated sludge as shown below (Table 1). Eq. (5) can also be expressed as:
Table 1 e Shear rate above which the HerscheleBulkley model is not suitable. Concentration [%] 1.85 2.56 3.17 4.89
Limit shear rate [s1] 145 280 470 565
Fig. 11 e Evolution of the yield stress and the Bingham viscosity regarding the concentration. The parameters of the Eqs. (7) and (8) are respectively a [ 0.19 Pa, f0 [ 1.17%, m [ 1.89 and m0 [ 0.0018 Pa s, b [ 0.604.
5680
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 7 5 e5 6 8 0
represented by the Bingham viscosity) and nonhydrodynamic interactions (between solid particles, basically represented by the yield stress). Increasing the concentration doesn’t change the nature of these interactions, but rather modifies their relative intensity. The dimensionless form smoothed these differences because both kinds of interactions in this form will approach unity. On our range of concentrations, yield stress and Bingham viscosity increase with the solid concentration (Fig. 11) respectively following a power-law and an exponential law of the following form, which is in agreement with the literature, both for the yield stress (Baudez, 2008) and the Bingham viscosity (Sanin, 2002): sc ¼ a$ðf f0 Þm
(7)
k2 ¼ m0 $expðb$fÞ
(8)
where f0 is the lowest concentration below which there is no yield stress, m is related to the fractal dimension of sludge flocs (Baudez, 2008) and m0 is the viscosity of the liquid medium. We found that the value m0 is twice that of pure water, which can be explained by the large amount of dissolved matter present, which may increase the supernatant viscosity.
4.
Conclusion
In this paper, we have shown that digested sludge is a shearthinning yield stress fluid, presenting flow instabilities at low shear rates, manifesting as shear banding. At low shear stress, below the yield stress, digested sludge behaved as a viscoelastic solid. When the applied stress is increased, above a critical shear strain, which decreases with the restructuring, shear banding appears. Then, at higher stresses, digested sludge behaves like a yield stress fluid and can be modelled using both the HerscheleBulkley and Bingham plastic models over a wide range of shear rates. This behaviour was similar at different concentrations and yield stress followed a power-law with the concentration while the Bingham viscosity followed an exponential law with concentration. By reducing the rheological parameters with the yield stress and the Bingham viscosity, which have to be measured separately, a master curve was obtained. This result means that the rheological behaviour of the digested sludge at any concentration can be deduced from this master curve. However, further work has to be done on shear banding. This behaviour will have to be taken into account in digester design and process operations, in order to avoid dead zones in the digester.
Acknowledgements The authors acknowledge the Cemagref-RMIT agreement for our collaboration.
references
Appels, L., Baeyens, J., Degre`ve, J., Dewil, R., December 2008. Principles and potential of the anaerobic digestion of waste-activated sludge. Prog. Energy Combust. Sci. 34 (6), 755e781. Ayol, A., Filibeli, A., Dentel, S.K., 2006. Evaluation of conditioning responses of thermophilic-mesophilic anaerobically and mesophilic aerobically digested biosolids using rheological properties. Water Sci. Technol. 54 (5), 23e31. Baudez, J.C., 2006. About peak and loop in sludge rheogram. J. Environ. Manage. 78, 232e239. Baudez, J.C., 2008. Physical aging and thixotropy in sludge rheology. Appl. Rheology 18 (13495), 1e8. Baudez, J.C., Coussot, P., 2001. Rheology of aging, concentrated, polymeric suspensions e Application to pasty sewage sludges. J. Rheol. 45 (5), 1123e1139. Forster, C.F., 1983. Bound water in sewage sludge and its relationship to sludge surfaces and sludge viscosities. J. Chem. Tech. Biol. 33B, 76e84. Forster, C.F., 2002. The rheological and physico-chemical characteristics of sewage sludge. Enzym. Microb. Tech. 30 (3), 340e345. Miller, E., Rothstein, J.P., 2007. Transient evolution of shear banding in wormlike micelle solutions. J. Non-Newtonian Fluid Mech. 143, 22e37. Moller, P.C.F., Rodts, S., Michels, M.A.J., Bonn, D., 2008. Shear banding and yield stress in soft glassy materials. Phys. Rev. E 77, 041507. Monteiro, P.S., 1997. The influence of the anaerobic digestion process on the sewage sludges rheological behaviour. Water Sci. Technol. 36 (11), 61e67. Munoz, J., Alfaro, M.C., 2000. Rheological and phase behaviour of amphiphilic lipids. Grasas y aceites 51, 6e25. Namer, J.J., Ganczarczyk, L., 1993. Settling properties of digested sludge particle aggregates. Water Res. 27, 1285e1294. Sanin, F.D., 2002. Effect of solution physical chemistry on the rheological properties of activated sludge. Water SA 28, 207e212. Slatter, P., 1997. The rheological characterisation of sludges. Water Sci. Technol. 36 (11), 9e18. Slatter, P., 2001. Sludge pipeline design. J. Water Sci. Technol. 44 (10), 115e120. Slatter, P., 2003. Pipeline transport of thickened sludges. Water 21, 56e57. Sperry, W.A., 1959. Gas recirculation at Aurora, Illinois. Sewage Ind. Waste 31 (6). Tropey, W.N., Melbinger, N.R., 1967. Reduction of digested sludge volume by controlled recirculation. J. Water Pollut. Control Fed. 39 (9), 1464e1474.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 1 e5 6 8 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Survival of environmental and clinical strains of methicillin-resistant Staphylococcus aureus [MRSA] in marine and fresh waters Emily Levin-Edens a, Natasha Bonilla b, J. Scott Meschke a, Marilyn C. Roberts a,* a
Department of Environmental and Occupational Health Sciences 357234, School of Public Health, 1959 NE Pacific St, University of Washington, Seattle, WA 98195-7234, USA b Department of Biology, University of Puerto Rico, San Juan, Puerto Rico
article info
abstract
Article history:
Recent studies have found variable levels of methicillin-resistant Staphylococcus aureus
Received 26 April 2011
[MRSA] in marine water from temperate and warmer climates suggesting that temperature
Received in revised form
may play a role in survival of MRSA in the environment. The aim of the study was to
19 August 2011
compare the survival of clinical and environmental MRSA and MSSA strains in fresh and
Accepted 20 August 2011
marine water incubated at 13 C and 20 C over 14 days. Seven different MRSA strains and
Available online 31 August 2011
the MSSA ATCC 25923 were tested. Individual strains were diluted in sterile saline to a 0.5 McFarland standard (108 cfu/ml), serially diluted in duplicate to a final concentration of
Keywords:
105 cfu/ml in pooled filter-sterilized marine or fresh water and incubated at 13 C or 20 C in
MRSA die-off
the dark. The results of this study found that temperature and salinity are important
Ambient and cold water tempera-
factors in MRSA and MSSA survival; the decay rate was w28% higher at 20 C versus 13 C
ture
and w34e44% higher in fresh water versus marine water. There was no statistical
Salt and fresh water
difference between environmental and clinical MRSA strain survival [P ¼ 0.138]. The study found that MRSA/MSSA survival was significantly longer in marine water at 13 C typical of the Pacific Northwest, which may have important implications for recreational beach visitors in colder climates. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Over the past decade community-acquired methicillin-resistant Staphylococcus aureus [CA-MRSA] has emerged as a major cause of disease in the general population with no health care exposure or known classical risk factors for methicillinresistant S. aureus [MRSA] infections. CA-MRSA causes skin and soft tissue infections, pneumonia and can lead to death (Bartlett, 2008; King et al., 2006) and the morbidity and mortality rate per 100,000 people is estimated to be 4.6 and 0.5, respectively (Klevens et al., 2007). Outbreaks of CA-MRSA in
previously healthy populations have been reported in prison populations, religious communities, sports teams, and military training camps (Coronado et al., 2007; David et al., 2008; Morrison-Rodriguez et al., 2010; Romano et al., 2006). Previously, contact with seawater has been associated with a four-fold increase in risk of infection from S. aureus (Charoenca and Fujioka, 1995). More recently, MRSA strains have been isolated and characterized from marine water and intertidal sand samples from five of ten Pacific Northwest [PNW] marine beaches (Soge et al., 2009), while S. aureus and MRSA have also been isolated from Florida (Abdelzaher et al.,
* Corresponding author. Tel.: þ1 206 543 8001; fax: þ1 206 543 3873. E-mail address: [email protected] (M.C. Roberts). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.037
5682
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 1 e5 6 8 6
2010), Hawaii (Tice et al., 2010) and California (Goodwin and Pobuda, 2009) marine water and sand. The sources of S. aureus and MRSA contamination in marine environments have yet to be characterized, though both potential point and non-point sources have been suggested. Two studies have shown that within the first 15 min of water immersion, the average person sheds 105e106 cfu S. aureus (Elmir et al., 2007; Plano et al., 2011), suggesting that bathers are a potential source of both S. aureus and MRSA contamination. MRSA has been isolated in wastewater prior to treatment (Bo¨rjesson et al., 2009) while S. aureus has been isolated from urban runoff from high density residential areas suggesting that combined sewer overflows [CSO] during storm events and urban runoff may contribute to MRSA and S. aureus loads at marine and fresh water recreational beaches (Selvakumar and Borst, 2006). In addition to the source of MRSA and S. aureus, the survivability of MRSA and S. aureus in beach environments is also an important factor that may impact the concentration level found in the beach environments. One study has previously investigated MRSA die-off kinetics in marine and fresh water using community- and hospital-acquired clinical strains, isolated from the hospital, and found an average of 7.4 days and 3.53 days for a 1-log reduction of MRSA from marine and fresh water, respectively, at ambient temperature which we estimate was between 20 and 22 C (Tolba et al., 2008). The authors found no statistical significance differences between survival of CA-MRSA and hospitalacquired methicillin-resistant S. aureus [HA-MRSA] strains. However, there is little research concerning the persistence of MRSA and S. aureus at lower temperatures (<20 C), that are normally found in Pacific Northwest marine beaches during summer. Furthermore, there is no information comparing survival of environmental vs and clinical MRSA strains. There has been limited research on survival of MRSA at the temperature of 13 C which is common at Pacific Northwest marine beaches during summer when the beaches are most frequented. The objective of this study was to compare survival of both environmental and clinical MRSA strains and a control MSSA ATCC 25923 strain for an assessment of MRSA survival in the Pacific Northwest beach environment [13 C] compared to warmer environments [20 C].
2.
Materials and methods
2.1.
Strain description
Seven MRSA strains [four environmental and three clinical] representing SCCmec type I [1 isolate], II [1 isolate], and IV [5 isolates] and MSSA ATCC 25923 were tested in this study (Table 1). The environmental strains were isolated from regional Pacific Northwest beaches in 2008 and 2010, and the clinical strains were isolated from skin and wound infections from hospitalized patients in 1945, 1993 and 2009. The presence of the mecA gene and the staphylococcal cassette chromosome mec [SCCmec] elements type were confirmed for each MRSA isolate by PCR (Soge et al., 2009) (Table 1).
Table 1 e MRSA strain characterization. Isolate
Straina collection date
Clinical ATCC 25923 MSSA MS361 MRSA MS1053 MRSA VA # 6 MRSA Environmental 9e48 MRSA L1 MRSA G1 MRSA M1 MRSA
Source
mecAb SCCmec typec
1949 1993 1993 2009
Clinical Clinical Clinical Clinical
e D D D
NAd II I IV
2008 2010 2010 2010
Marine water Marine water Fresh water Fresh water
D D D D
IV IV IV IV
Clinical strain ATCC 25923 was obtained from the American Type Culture Collection. Clinical strain VA #6 and the environmental strains 9-48, LI, GI, and MI were collected in Seattle, WA. a Methicillin-susceptible Staphylococcus aureus [MSSA]; Methicillinresistant Staphylococcus aureus [MRSA]. b mecA gene codes for an altered penicillin binding protein and all MRSA isolates carry this genes. c staphylococcal cassette chromosome mec elements. d NA: not applicable.
2.2.
Marine and stream water preparation
Marine and fresh water samples were collected in sterile 1-L Nalgene bottles from two Seattle marine beaches on the Puget Sound. Fresh water samples were collected from streams from surrounding upland drainage areas that traverse through the beach to the marine receiving waters. After collection, the fresh and marine water samples were respectively pooled and filter-sterilized through a 0.22 mM filter (Pall Corporation, Port Washington, NY, USA). Sterility of each water type was verified by plating 100 mL onto Brain Heart Infusion (BHI) agar (Difco Laboratories, Div. Becton Dickinson & Co., Sparks, MD, USA) in triplicate and incubated at 36.5 C for 24 h. The pooled fresh was had a salinity of <1 ppt and pH 7.43 and the pooled marine water had a salinity of 29 ppt and pH 7.77.
2.3.
Fresh and marine water inoculation
Overnight cultures from a Brucella agar (Difco Laboratories, Sparks, MD, USA) supplemented with 5% sterile sheep blood for each strain were diluted in 0.85% NaCl to a 0.5 McFarland standard (108 cfu/ml) and serially diluted to approximately 105 cfu/ml in 50 ml of pooled, filter-sterilized marine or fresh water in a sterile conical tube. Marine and fresh water microcosms were incubated at ambient temperatures (20 C 2 C) or 13 C 0.5 C in the dark. The starting concentration for each water microcosm was determined on day 0 by plating a 10-fold dilution series onto Brain Heart Infusion agar [BHI] (Difco Laboratories, Sparks, MD, USA). BHI plates were incubated at 36.5 C for 24 h and colony counts (cfu/ml) determined. Daily colony counts were determined at days 1e3 (13 C) or days 1e5 (20 C) as described for day 0 and then determined every second day for 14 days (Fig. 1). Prior to plating for cfu/ml determination each microcosm was
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 1 e5 6 8 6
5683
Fig. 1 e For each time point, spread plate counts were averaged and log-transformed. A linear regression was calculated for each microcosm to assess the time (days) needed for a 1-log reduction. A. Survival of environmental MRSA isolates in marine and fresh water at 13 C and 20 C. B. Survival of clinical MRSA isolates in marine and fresh water at 13 C and 20 C.
vortexed for approximately 30 s to disrupt adhered cells from the side of the tube and cell aggregates.
2.4.
Statistical analysis
For each microcosm, the average log-transformed spread plate counts were analyzed by linear regression in GraphPad Prism 5 (GraphPad Software, Inc, San Diego, CA, USA) to calculate the time elapse for a 1-log removal [T90]. The results of the seven MRSA strains were averaged to determine the 1log reduction (T90) of the MRSA strains. The significance of association between water type (marine vs fresh), temperature and strain classification (environmental vs. clinical) on survival was modeled by multivariable linear regression in Stata11 I.C (StataCorp LP, College Station, TX, USA). P < 0.05 was considered statistically significant.
3.
Results and discussion
The survival of the eight strains over 14 days is illustrated in Fig. 1. The cfu/ml plate counts were log-transformed and plated by day. On average, the seven MRSA strains in marine water at 20 C and 13 C, it took 7.89 1.62 days and 10.97 3.47 days, respectively, for a 1-log reduction (T90). The survival of 7.89 1.62 days in marine water at 20 C in the current study is very similar to 7.4 days previously reported by Tolba et al. (2008). Survival of all strains was reduced in fresh water with an average of 2.71 0.58 days and 4.84 1.11 days for a 1-log removal at 20 C and 13 C, respectively. For the ATCC 25923 MSSA strain, the T90 at 20 C and 13 C was 3.70 days and 13.29 days, respectively in marine water and 2.65
days and 4.32 days, respectively in fresh water (Fig. 1). The average survival results of 2.71 0.58, in the current study for fresh water at 20 C, was in the range of 3.5 days previously reported by Tolba et al. (2008). Overall the decay rate was w28% higher at 20 C versus 13 C and w34e44% higher in fresh water versus marine water. Table 2 describes individual strain correlation coefficient (R2), T90, die-off rate (k) and illustrates the variability found between strains. Unfortunately individual results were not available in the Tolba et al. (2008) for comparison. A multivariable linear regression model found that temperature [20 C vs 13 C] and water types [fresh vs marine] were significantly [P < 0.001] associated with MRSA and MSSA survival (Table 3). The source of the strain [environmental vs clinical] was not significantly associated with MRSA survival [P ¼ 0.138] when adjusted for temperature, time and water type even though the strains were collected between 1993 and 2010 when the ATCC 25923 MSSA strain was excluded from the analysis. When the multivariable linear regression model included the ATCC 25923 MSSA strain, the source of the strain became significantly associated with survival [P ¼ 0.011], suggesting that the ATCC 25923 MSSA strain is different from the other isolates (Table 3). This difference may be because the ATCC 25923 MSSA strain was isolated in 1949 and passed in the laboratory while the other isolates were collected more recently and had limited laboratory passage, or that there are genetic differences such as the lack of a SCCmec cassette. The MRSA strains reduction rate, in the absence of organic material and light, followed first-order kinetics ( y ¼ mx þ b), which were previously found (Tolba et al., 2008). There are several limitations to this study that may over or underestimate MRSA survival times, because survival was
5684
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 1 e5 6 8 6
Table 2 e Regression results, die-off rate (k), and T90 removal time for MRSA strains in fresh and marine water. Isolate
Marine water
Temp ( C) ATCC 25923
20 13 20 13 20 13 20 13 20 13 20 13 20 13 20 13 20 13
MS361 MS1053 Va#6 9e48 L1 G1 M1 T90 Average SDe
a b c d e
2 a
R
Fresh water 1 c
b
T90 (days)
k (day )
0.89 3.70 0.78 13.29 0.71 4.91 0.75 5.41 0.82 7.55 0.76 13.64 0.87 8.18 0.67 13.36 0.80 8.04 0.79 15.43d 0.80 9.90 0.83 8.74 0.82 9.38 0.82 11.11 0.76 7.30 0.90 9.07 7.89 1.62 days 10.97 3.47 days
0.27 0.08 0.20 0.18 0.13 0.07 0.12 0.07 0.12 0.06 0.10 0.11 0.11 0.09 0.14 0.11
R
2 a
T90 (days)b
0.89 2.65 0.87 4.32 0.99 2.15 0.81 4.13 0.92 2.25 0.93 6.84 0.90 2.43 0.94 5.77 0.98 2.36 0.96 4.57 0.87 2.92 0.94 3.97 0.86 3.78 0.96 4.89 0.89 3.06 0.98 3.73 2.71 0.58 days 4.84 1.11 days
k (day1)c 0.38 0.23 0.47 0.24 0.44 0.15 0.41 0.17 0.42 0.22 0.34 0.25 0.26 0.20 0.33 0.27
Regression correlation coefficient. Time needed for a 1-log reduction. Die-off rate: k ¼ 1/T90 Extrapolated beyond data range. Does not include the ATCC 25923 MSSA strain.
analyzed only in sterile microcosms in the dark. Intrinsic variables such as survival mechanisms and extrinsic factors such as nutritional sources, UV light, and antagonistic microflora were not examined. The study by Fuijoka and Unutoa (2006) showed that UV light and supplemented organic material such as sewage reduced or prolonged survival time, respectively, compared to survival time measured in the absence of these variables while without supplemented organic material, the proportion of culturable S. aureus populations declines relative to the total population in a seawater microcosm over time (Masmoudi et al., 2010). The potential role of algae wrack in the survival of MRSA and S. aureus is currently unknown, however data from Great Lakes studies show algal mats as reservoirs for Escherichia coli,
Enterococcus, Shigella, Campylobacter, and Salmonella (Ishii et al., 2006; Olapade et al., 2006; Verhougstraete et al., 2010). More recently marine wrack was shown to promote persistence of fecal indicator bacteria in laboratory marine microcosms (Imamura et al., 2011). Sand has also been identified as a protective from the bactericidal effects of UV light and dry sand near the high-tide line has been found to be a significant reservoir of pathogenic bacteria (Abdelzaher et al., 2010; Beversdorf et al., 2007; Bonilla et al., 2007). The current results are the first to examine MRSA survival in 13 C marine and fresh water. This is essential because the stability of MRSA in recreational beach environments is important for an accurate exposure assessment and ultimately risk characterization of human MRSA infection. This
Table 3 e Multivariable linear regression results for predictor variables water type, temperature and isolate type on MRSA survival.
Waterd Tempe Typef a b c d e f
ATCC 25923 Excluded
ATCC 25923 Included
95% CIb
95% CIb
ba
Lower
Upper
P-valuec
ba
Lower
Upper
P-valuec
0.88 0.41 0.14
0.70 0.59 0.36
1.07 0.23 0.05
<0.001 <0.001 0.138
e0.47 0.47 0.22
0.75 0.64 0.39
1.10 0.30 0.05
<0.001 <0.001 0.011
Coefficient. CI: confidence interval. P < 0.05 considered significant. Water: marine vs fresh. Temperature: 13 C vs 20 C. Type: environmental vs clinical; Model correlation coefficient including and excluding ATCC 25923 MSSA strain: R2 ¼ 0.57.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 1 e5 6 8 6
study found that there was an increased survival of MRSA and MSSA strains in marine vs fresh water. Survival was longer at 13 C vs 20 C suggesting that there may be differences in survival of MRSA at recreational beaches in the Pacific Northwest and those of subtropical and tropical climates. All four of the environmental strains were isolated in the Seattle area and it is unknown whether environmental MRSA strains isolated in warmer water (25 C) would respond the same as the strains in the current study.
4.
Conclusions
The study found that there was an increased survival of MRSA/MSSA strains in marine vs fresh water and survival was higher at 13 C vs 20 C in fresh and marine waters suggesting that there may be major differences between S. aureus survival at recreational beaches in the Pacific Northwest vs those of subtropical and tropical climates. Temperature [13 C vs 20 C] and salinity [marine water vs fresh water] were significantly [P < 0.001] associated with MRSA/MSSA survival. MRSA survival rate was not significantly associated with the isolate source [environmental vs clinical].
Acknowledgements The work was funded in part by NIH National Institute of Environmental Health and Science (NIEHS) 5 R25 ES016150-03 REVISED. Ms. Bonilla was supported by an NIH Minority Access to Research Careers (MARC) grant (# 5T34GM07821).
references
Abdelzaher, A.M., Wright, M.E., Ortega, C., Solo-Gabriele, H.M., Miller, G., Elmir, S., Newman, X., Shih, P., Bonilla, J.A., Bonilla, T.D., Palmer, C.J., Scott, T., Lukasik, J., Harwood, V.J., McQuaig, S., Sinigalliano, C., Gidley, M., Plano, L.R., Zhu, X., Wang, J.D., Fleming, L.E., 2010. Presence of pathogens and indicator microbes at a non-point source subtropical recreational marine beach. Appl. Environ. Microbiol. 76 (3), 724e732. Bartlett, J.G., 2008. Methicillin-resistant Staphylococcus aureus infections. Top. HIV Med. 16 (5), 151e155. Beversdorf, L.J., Bornstein-Forst, S.M., McLellan, S.L., 2007. The potential for beach sand to serve as a reservoir for Escherichia coli and the physical influences on cell die-off. J. Appl. Microbiol. 102 (5), 1372e1381. Bo¨rjesson, S., Matussek, A., Melin, S., Lo¨fgren, S., Lindgren, P.E., 2009. Methicillin-resistant Staphylococcus aureus (MRSA) in municipal wastewater: an uncharted threat? J. Appl. Microbiol. 108 (4), 1244e1251. Bonilla, T.D., Nowosielski, K., Cuvelier, M., Hartz, A., Green, M., Nwadiuto, E., McCorquodale, D.S., Fleisher, J.M., Rogerson, A., 2007. Prevalence and distribution of fecal indicator organisms in South Florida beach sand and preliminary assessment of health effects associated with beach sand exposure. Mar. Pollut. Bull. 54 (9), 1472e1482.
5685
Charoenca, N., Fujioka, R.S., 1995. Assessment of Staphylococcus bacteria in Hawaii recreational waters. Water. Sci. Technol. 31 (5e6), 11e17. Coronado, F., Nicholas, J.A., Wallace, B.J., Kohlerschmidt, D.J., Musser, K., Schoonmaker-Bopp, D.J., Zimmerman, S.M., Boller, A. R., Jernigan, D.B., Kacica, M.A., 2007. Community-acquired methicillin-resistant Staphylococcus aureus skin infections in a religious community. Epidemiol. Infect. 135 (3), 492e501. David, M.Z., Mennella, C., Mansour, M., Boyle-Vavra, S., Daum, R.S., 2008. Predominance of methicillin-resistant Staphylococcus aureus among pathogens causing skin and soft tissue infections in a large urban jail: risk factors and recurrence rates. J. Clin. Microbiol. 46 (10), 3222e3227. Elmir, S.M., Wright, M.E., Abdelzaher, A., Solo-Gabriele, H.M., Fleming, L.E., Miller, G., Rybolowik, M., Peter Shih, M.T., Pilla, S.P., Cooper, J.A., Quaye, E.A., 2007. Quantitative evaluation of bacteria released by bathers in marine water. Water. Res. 41 (18), 3e10. Fuijoka, R.S., Unutoa, T.M., 2006. Comparative stability and growth requirements of S. aureus and faecal indicator bacteria in seawater. Water. Sci. Technol. 54 (3), 169e175. 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 analysis. Water. Res. 43, 4802e4811. Imamura, G.J., Thompson, R.S., Boehm, A.B., Jay, J.A., 2011. Wrack promotes the persistence of fecal indicator bacteria in marine sands and seawater. FEMS. Microbiol. Ecol. 77, 40e49. Ishii, S., Yan, T., Shively, D.A., Byappanahalli, M.N., Whitman, R.L., 2006. Cladorphora (Chlorophyta) spp. harbor human bacterial pathogens in nearshore water of Lake Michigan. Appl. Environ. Microbiol. 72 (7), 4545e4553. King, M.D., Humphrey, B.J., Wang, Y.F., Kourbatova, E.V., Ray, S. M., Blumberg, H.M., 2006. Emergence of community-acquired methicillin-resistant Staphylococcus aureus USA 300 clone as the predominant cause of skin and soft-tissue infections. Ann. Intern. Med. 144 (5), 309e317. Klevens, R.M., Morrison, M.A., Nadle, J., Petit, S., Gersham, K., Ray, S., Harrison, L.H., Lynfield, R., Dumyati, G., Townes, J.M., Craig, A.S., Zell, E.R., Fosheim, G.E., McDougal, L.K., Carey, R.B., Fridken, S.K., 2007. Invasive methicillin-resistant Staphylococcus aureus infections in the United States. JAMA 298 (15), 1763e1767. Masmoudi, S., Denis, M., Maalej, S., 2010. Inactivation of the gene katA and sodA affects the transient entry into the viable but nonculturable response of Staphylococcus aureus in natural seawater at low temperatures. Mar. Pollut. Bull. 60 (12), 2209e2214. Morrison-Rodriguez, S.M., Pacha, L.A., Patrick, J.E., Jordan, N.N., 2010. Community-associated methicillin-resistant Staphylococcus aureus infections at an Army training installation. Epidemiol. Infect. 138 (5), 721e729. Olapade, O.A., Depas, M.M., Jensen, E.T., McLellan, S.L., 2006. Microbial communities and fecal indicator bacteria associated with Cladophora mats on beach sites along Lake Michigan shores. Appl. Environ. Microbiol. 72 (3), 1932e1938. Plano, L.R.W., Garza, A.C., Shibata, T., Elmir, S.M., Kish, J., Sinigalliano, C.D., Gidley, M.L., Miller, G., Withum, K., Fleming, L. E., Solo-Gabriele, H.M., 2011. Shedding of Staphylococcus aureus and methicillin-resistant Staphylococcus aureus from adult and pediatric bathers in marine waters. BMC Microbiol 11 (1), 5. Romano, R., Lu, D., Holtom, P., 2006. Outbreak of communityacquired methicillin-resistant Staphylococcus aureus skin infections among a collegiate football team. J. Athl Train. 41 (2), 141e145. Selvakumar, A., Borst, M., 2006. Variation of microorganism concentrations in urban runoff with land use and seasons. J. Water Health 4 (1), 109e124.
5686
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 1 e5 6 8 6
Soge, O.O., Meschke, J.S., No, D.B., Roberts, M.C., 2009. Characterization of methicillin-resistant Staphylococcus aureus and methicillin-resistant coagulase-negative Staphylococcus spp. isolated from US west coast public marine beaches. J. Antimicrob. Chemother. 64 (6), 1148e1155. Tice, A.D., Pombo, D., Hui, J., Kurano, M., Bankowski, M.J., Seifried, S.E., 2010. Quantification of Staphylococcus aureus in seawater using CHROMagar SA. Hawaii. Med. J. 69 (1), 8e12.
Tolba, O., Loughrey, A., Goldsmith, C.E., Millar, B.C., Rooney, P.J., Moore, J.E., 2008. Survival of epidemic strains of healthcare (HAMRSA) and community-associated (CA-MRSA) methicillinresistant Staphylococcus aureus (MRSA) in river-, sea- and swimming pool water. Int. J. Hyg. Environ. Health. 211 (3e4), 398e402. Verhougstraete, M.P., Byappanahalli, M.N., Rose, J.B., Whitman, R. L., 2010. Cladophora in the Great Lakes: impacts on beach water quality and human health. Water. Sci. Technol. 62 (1), 68e76.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
NDMA formation kinetics from three pharmaceuticals in four water matrices Ruqiao Shen*, Susan A. Andrews Department of Civil Engineering, University of Toronto, 35 St. George St., Toronto, Ontario, Canada M5S 1A4
article info
abstract
Article history:
N, N-nitrosodimethylamine (NDMA) is an emerging disinfection by-product (DBP) that has
Received 17 May 2011
been widely detected in many drinking water systems and commonly associated with the
Received in revised form
chloramine disinfection process. Some amine-based pharmaceuticals have been demon-
11 August 2011
strated to form NDMA during chloramination, but studies regarding the reaction kinetics
Accepted 20 August 2011
are largely lacking. This study investigates the NDMA formation kinetics from ranitidine,
Available online 27 August 2011
chlorphenamine, and doxylamine under practical chloramine disinfection conditions. The formation profile was monitored in both lab-grade water and real water matrices, and
Keywords:
a statistical model is proposed to describe and predict the NDMA formation from selected
NDMA
pharmaceuticals in various water matrices. The results indicate the significant impact of
Ranitidine
water matrix components and reaction time on the NDMA formation from selected
Chlorphenamine
pharmaceuticals, and provide fresh insights on the estimation of ultimate NDMA forma-
Doxylamine
tion potential from pharmaceutical precursors.
Chloramination
ª 2011 Elsevier Ltd. All rights reserved.
Kinetics
1.
Introduction
N, N-nitrosodimethylamine (NDMA) is a member of N-nitrosamines found in food, beer, cured meats, rubber products, tobacco smoke and more recently, drinking water. There is growing concern regarding the health effects associated with exposure to nitrosamines because of their potential carcinogenicity (EPA IRIS, 1993). The occurrence of NDMA in finished drinking water has been commonly associated with the application of chloramine as a final disinfectant. Recent surveys in Canada and the U.S. have revealed occurrence of NDMA in many chloraminated drinking water systems with concentration up to 630 ng/L (Blute et al., 2010; Charrois et al., 2007). The widespread detections of NDMA in source water and treated drinking water have spurred local governments and agencies to take actions. The Ontario Ministry of the Environment (MOE) has established a maximum acceptable
level of 9 ng/L for NDMA (MOE, 2003), and the California Department of Health Services has implemented an NDMA notification level of 10 ng/L (OEHHA, 2006). The USEPA has placed NDMA together with other four nitrosamines on the latest drinking water contaminant candidate list 3 (CCL3) (USEPA, 2009). More recently, Health Canada has proposed a maximum acceptable concentration for NDMA of 40 ng/L in drinking water (Health Canada, 2010). A number of research efforts have been invested in identifying potential NDMA precursors relevant to drinking water. Theoretically, any amine compounds containing dimethylamine (DMA) groups may react with chloramine to form NDMA. Typical precursors found in source water include some tertiary and quaternary amines (Kemper et al., 2010; Mitch et al., 2003; Mitch and Schreiber, 2008), and fractions of natural organic matter (NOM) (Chen and Valentine, 2007; Dotson et al., 2007; Gerecke and Sedlak, 2003; Mitch and
* Corresponding author. Tel.: þ1 4169783141; fax: þ1 4169783674. E-mail addresses: [email protected] (R. Shen), [email protected] (S.A. Andrews). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.034
5688
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
Sedlak, 2004). Some chemicals used in water treatment processes may also contribute to NDMA formation, such as certain amine-based polymers and anion exchange resins (Kohut and Andrews, 2003; Mitch and Sedlak, 2004; Najm and Trussell, 2001; Wilczak et al., 2003). More recently, dimethylsulfamide, a degradation product of the fungicide tolyfluanide, was newly identified as an NDMA precursor during ozonation (Schmidt and Brauch, 2008). Pharmaceuticals first came to attention as potential NDMA precursors when ranitidine was demonstrated to convert into NDMA at high conversion rate during chloramination (Sacher et al., 2008; Schmidt et al., 2006). Krasner (2009) has suggested that amine-based pharmaceuticals and their breakdown products might be part of the NDMA precursor pool. A recent study by the authors has demonstrated the formation of nitrosamines from twenty amine-based pharmaceuticals and personal care products (PPCPs) upon chloramine disinfection (Shen and Andrews, 2011). Up to now, studies on NDMA formation via pharmaceuticals have been mostly conducted using lab-grade water. Specifically, data regarding the reaction kinetics in real water matrices are largely lacking. Krasner et al. (2010) have investigated the NDMA formation over time from ranitidine under different pH and temperature, but only conducted the experiments in deionized water. Due to the lack of knowledge about the reactivity and chemistry, it is difficult to predict NDMA formation from pharmaceuticals using traditional kinetic models. In the literature, some kinetic models have been developed for the prediction of NDMA formation from DMA (Choi and Valentine, 2002; Kim and Clevenger, 2007) and from NOM (Chen and Valentine, 2006); however, these models use comparable concentrations of precursors and chloramines, and thus might not apply to pharmaceuticals which are usually present at trace levels in the source water and are at much lower concentrations relative to chloramine concentrations in real samples. This study demonstrates the NDMA formation kinetics from three amine-based pharmaceuticals in four different water matrices, and proposes a statistical model to describe and predict the NDMA molar conversion from selected pharmaceuticals during chloramination.
2.
Materials and methods
Three pharmaceuticals were selected to determine their NDMA formation potential (NDMA-FP) over time, including chlorphenamine, doxylamine, and ranitidine (Fig. 1). Stock solutions of pharmaceuticals were prepared in methanol and stored at 4 C until use. NDMA (reagent grade) and deuterated NDMA (d6-NDMA, 98 atom %D) were used as standard and internal standard, respectively. All chemicals were purchased from SigmaeAldrich Canada (Oakville, Ontario). Experiments were carried out under the Simulated Distribution System (SDS) conditions (pH ¼ 7.0 0.1; 21 C; Cl2: N mass ratio ¼ 4.2:1; chloramine dosage ¼ 2.5 0.2 mg/L after satisfying 24 hr chloramine demand). Further details concerning the experimental procedure and NDMA analysis have been described in Shen and Andrews (2011). NDMA formation from each pharmaceutical was monitored for up to 144 hr
Fig. 1 e Structures of selected pharmaceuticals.
based on preliminary results, depending on the compound and water matrix. At each time point samples were prepared in duplicate, together with one blank control to account for the potential background interference. Error bars in all the kinetic graphs represent the maximum and minimum values in the formation potential tests under the same reaction conditions (n ¼ 2). Experiments were conducted in four water matrices; the water sources and basic water quality parameters are summarized in Table 1. Lake and river water samples were taken from the influent of two drinking water treatment plants in June and October, 2010, respectively. The pH was determined using a pH meter (Model 8015, VWR Scientific Inc., Mississauga, Ontario). Alkalinity was measured based on an end-point titration, according to Standard Method 2320B (APHA, 2005). The total organic carbon (TOC) was analyzed with an Aurora 1030 TOC analyzer (O.I. Analytical, College Station, Texas). The ultraviolet absorbance at 254 nm (UV254) was determined by a CE3055 Reflectance Spectrophotometer (Cecil Instruments Ltd., Cambridge, England). The specific UV absorbance (SUVA) is calculated by normalizing the UV254 to the TOC.
3.
Results and discussion
3.1.
Formation kinetics in MQ water
Kinetic experiments in MQ (Milli-Q, Ultra Pure Water System, MilliPore, Etobicoke, Ontario) water were conducted for ranitidine, chlorphenamine, and doxylamine at two concentration levels (5 and 25 nM), as shown in Fig. 2. The markers in the figure are the measured NDMA molar conversion values, and the lines are model-estimated results. Details about the model development and estimation will be discussed in Section 3.3. NDMA formation via the three pharmaceuticals followed similar pattern over time. Generally, an initial lag period was observed, followed by a fast increase in NDMA concentration; the molar conversion then gradually leveled off and eventually reached a plateau (maximum molar conversion). Moreover, the formation kinetic behavior was observed to be relatively independent of the initial pharmaceutical concentration, except that NDMA formation from doxylamine in MQ water showed a more significant difference after 24 hr than did the other pharmaceuticals. Given the large excess of chloramine relative to the pharmaceuticals (mg/L vs. lower mg/ L), availability of chloramine was not a limiting factor at the concentration range of pharmaceuticals tested. These results also support observations made in an earlier study where the NDMA molar conversion at 24 hr for 20 selected
5689
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
Table 1 e Water matrix source and quality. Matrix Milli-Q (MQ) Tap water (TAP) Lake water (LW) River water (RW)
Source
pH
Alkalinity (mg/L)
Ultra Pure Water System (MilliPore, Etobicoke, Ontario) Toronto, Ontario Lake Ontario, Ajax, Ontario Otonabee River, Peterborough, Ontario
7.5 0.1
1.8 0.3
7.1 0.1 8.0 0.1 7.8 0.1
88.5 2.8 94.6 1.5 86.5 1.2
pharmaceuticals was found to be independent of their initial concentrations (Shen and Andrews, 2011).
3.2.
Formation kinetics in different water matrices
Kinetic experiments were also performed using real water samples dosed with selected pharmaceuticals. The same shape of NDMA formation curve was observed in real water matrices as in MQ water except that the initial lag phase was longer, especially for tests performed using lake or river water (Fig. 3). Similarly to Fig. 2, the markers are the measured NDMA molar conversion values, and the lines are the modelestimated results that will be discussed in detail in 3.3. Although the NDMA formation kinetics was shown to be unique to each water matrix tested, the kinetic behavior was relatively independent of the initial pharmaceutical concentration within a given water matrix, further confirming that observation in MQ water. The different NDMA formation profiles were likely influenced by the water matrix components, rather than by added reagents, since the pH of the water samples was controlled with a phosphate buffer and the same chloramine dosage was applied to all samples. Both bromide and NOM have been shown to influence NDMA formation, with bromide being reported to either catalyze NDMA formation (Mitch et al., 2003; Valentine et al., 2005) or have an inhibitory effect (Chen et al., 2010). However, bromide levels in the water sources that were tested are typically much lower than those for studies that have reported these effects, so the differences in the observed formation profiles were thought to be due to some aspect of the NOM. Since bromide is in higher concentration and so may be more of a concern in coastal waters due to saltwater intrusion, the potential impact from bromide was considered
TOC (mg/L)
UV254 (cm1)
SUVA (L,m/mg)
0.0
0.000
0.000
2.1 0.1 2.3 0.2 6.2 0.5
0.021 0.001 0.024 0.002 0.143 0.002
1.01 0.05 1.08 0.12 2.32 0.17
to be outside the scope of the present tests but would be of interest for future study. NOM may affect the NDMA formation in two ways. The influence of NOM’s competition for chloramine was considered to be minimal due to the small observed chloramine decay (data not shown) and the large excess of chloramine relative to the pharmaceuticals (mg/L vs. lower mg/L) at the end of the kinetic experiment. On the other hand, NOM may interact with the pharmaceuticals and then inhibit the reaction to form NDMA, and it is these interactions that were thought to better explain the observed results. NOM components can be at least partially described by the samples’ TOC and SUVA values. It was observed that water with higher TOC and SUVA levels tended to have a longer initial lag phase; all three pharmaceuticals exhibited their longest initial lag period in river water samples. However, while tap and lake water samples had similar TOC and SUVA values, ranitidine showed a longer initial lag phase in lake water samples. This suggests that some specific NOM fractions or moieties might be more relevant than would be indicated by simple bulk measurements of water quality, such as TOC and SUVA. Previous studies have demonstrated that aromatic amines undergo reversible covalent binding with carbonyls and quinones in soil humic substances in the environment (Parris, 1980; Thorn et al., 1996; Weber et al., 1996). Therefore, it is possible that certain fractions or functional groups in NOM may interact with these amine-based pharmaceuticals and thus hinder their initial contact with chloramine species. As the binding is reversible and chloramine is in large excess, eventually the NDMA conversion from pharmaceuticals can still reach the maximum level given enough reaction time. Currently, although no direct spectroscopic evidence exists for the NOM-pharmaceutical binding in aqueous phase, this
Fig. 2 e NDMA molar conversion over time for ranitidine, chlorphenamine, and doxylamine in MQ water (SDS conditions).
5690
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
Fig. 3 e NDMA molar conversion over time for (a) ranitidine; (b) chlorphenamine; and (c) doxylamine in different water matrices (SDS conditions).
theory is indirectly supported by some literature investigating the removal of pharmaceuticals during coagulation/flocculation process, where the removal of pharmaceuticals was likely due to the sorption onto particulate organic matter and coremoved through the settling process (Ballard and Mackay, 2005; Stackelberg et al., 2007; Vieno et al., 2006; Westerhoff et al., 2005). Stackelberg et al. (2007) also detected the target pharmaceuticals in the dried solids of settled sludge. In addition, De Ridder et al. (2011) observed enhanced removal of some positively charged pharmaceuticals using granular activated carbon preloaded with NOM. They attributed the enhancement to the electrostatic attraction since the surface of NOM is usually negatively charged due to abundant carboxyl groups. In the current study, the selected amine
pharmaceuticals are positively charged at neutral pH, therefore the possible electrostatic attraction may also lead to the formation of NOM-pharmaceutical complexes. Future studies are needed to further investigate the role of NOM components in the conversion of pharmaceuticals into NDMA, and alternative methods for the characterization of NOM components will be helpful, such as the application of size-exclusion chromatography with organic carbon detection (Huber et al., 2011).
3.3.
Kinetic model
The NDMA formation curves for the pharmaceuticals in this study have a sigmoidal shape that resembles the typical shape
Table 2 e Model parameter estimation and model verification. Compound
Concentration
Matrix
Parameter Estimation a
q Ranitidine
5 nM
25 nM
Chlorphenamine
5 nM
25 nM
Doxylamine
5 nM
25 nM
MQ TAP LW RW MQ TAP LW RW MQ TAP RW MQ TAP RW MQ TAP RW MQ TAP RW
0.912 0.902 0.729 0.822 0.906 0.847 0.769 0.841 0.027 0.023 0.037 0.033 0.030 0.043 0.062 0.068 0.059 0.106 0.092 0.060
(0.045) (0.045) (0.032) (0.056) (0.045) (0.039) (0.018) (0.044) (0.003) (0.003) (0.002) (0.002) (0.002) (0.002) (0.005) (0.006) (0.003) (0.007) (0.006) (0.002)
a
Lag (hr)
6.3 (0.5) 6.7 (0.8) 13.3 (0.7) 20.8 (1.8) 4.6 (0.5) 6.5 (0.7) 13.1 (0.4) 21.9 (1.4) 7.0 (1.3) 10.8 (3.0) 38.4 (3.4) 8.7 (0.6) 19.4 (1.8) 39.9 (1.9) 16.6 (2.2) 48.8 (5.3) 67.0 (4.3) 22.0 (2.0) 46.1 (4.1) 51.1 (1.8)
Model Verification 1 a
k (hr ) 0.225 0.169 0.251 0.086 0.313 0.177 0.252 0.083 0.143 0.065 0.039 0.234 0.070 0.056 0.068 0.027 0.026 0.069 0.030 0.051
(0.043) (0.045) (0.057) (0.026) (0.084) (0.043) (0.028) (0.021) (0.050) (0.024) (0.007) (0.079) (0.015) (0.009) (0.017) (0.005) (0.006) (0.018) (0.005) (0.011)
R
2
0.992 0.986 0.997 0.984 0.991 0.988 0.999 0.990 0.974 0.953 0.990 0.994 0.988 0.996 0.975 0.993 0.990 0.985 0.992 0.996
a Numbers in the bracket represent the 95% confidence interval of each model parameter. b Numbers in the bracket represent standard deviation from multiple tests (n ¼ 3).
Model-predicted conversion @ 24 h 91.2% 90.1% 72.7% 53.6% 90.6% 84.6% 76.8% 50.4% 2.7% 2.0% 0.8% 3.3% 2.0% 0.5% 4.7% 1.2% 0.4% 6.1% 1.6% 0.2%
Measured conversion @ 24 hb 85.2% 83.4% 64.1% 51.4% 82.7% 88.4% 70.1% 43.2% 2.9% 2.0% 1.0% 1.8% 1.5% 0.5% 3.8% 2.5% 1.1% 4.2% 3.2% 0.5%
(0.8%) (8.1%) (3.6%) (4.9%) (2.4%) (5.9%) (4.8%) (7.1%) (0.2%) (0.8%) (0.1%) (0.1%) (0.1%) (0.02%) (0.1%) (0.2%) (0.03%) (0.1%) (0.3%) (0.04%)
5691
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
of doseeresponse curves. A standard doseeresponse curve can be defined by a four-parameter logistic function, ab y¼þ 1 þ 10c$ðdxÞ where y is the response caused by certain dose of pharmaceuticals (x); a and b are the maximum and baseline response, respectively; c is the slope of the curve; and d is the dose which provokes a response halfway between the baseline and maximum (Motulsky and Christopoulos, 2003). Accordingly, the following model was proposed to describe the reaction kinetics for NDMA formation from selected pharmaceuticals, q Y¼ 1 þ 10k$ðLagtÞ where Y is the NDMA molar conversion at given reaction time (t); q is the ultimate NDMA molar conversion, i.e., the maximum molar conversion obtained at the plateau during kinetic testing; k is the pseudo-first order reaction rate constant; Lag is the time required to achieve 50% of the ultimate molar conversion, and thus is associated with the length of initial lag phase observed. Comparing this model with the four-parameter logistic function, the parameter b was set to zero because any possible NDMA in the background and any potential NDMA formed from
the matrix components were accounted for by the blank control samples. It is noted that the proposed model does not pass through the point of (0, 0), although, once background NDMA has been subtracted, there should be zero molar conversion at the beginning. Since the doseeresponse model is based on log (drug dose), it is always positive on the x-axis, and thus does not go through the point of (0, 0). Therefore, the proposed model was arbitrarily set to be: Y¼
8 < :
0 q 1 þ 10k$ðLagtÞ
ðt ¼ 0Þ ðt > 0Þ
The formation curve was fitted using GraphPad Prism 5 software, and the estimated model parameters for each compound in different matrices are summarized in the Parameter Estimation section of Table 2. The proposed model fit the experimental data very well, with correlation coefficients (R2) higher than 0.95 in all cases, and predicted accurately all three phases of the NDMA formation curve, as shown previously in Figs. 2 and 3. It is worth noting that the model requires data capturing all three phases of the NDMA formation curve in order to acquire reliable model parameters. For datasets lacking the plateau data, the model will arbitrarily assume the last point as the
Fig. 4 e Linear correlation between (a) Lag and TOC; (b) Lag and SUVA; (c) k and TOC; (d) k and SUVA for three pharmaceuticals (SDS conditions; [Pharmaceutical] [ 5 and 25 nM; error bars represent the 95% confidence interval for estimated model parameters).
5692
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
plateau. In this study, the NDMA formation curve for ranitidine in LW samples did not achieve the plateau within the 24hr of the experiment (Fig. 3a). Therefore, the calculated q may underestimate the ultimate NDMA molar conversion for ranitidine in LW samples. The estimated model parameters each well reflected the different aspects of the NDMA formation profiles that were observed in the different water matrices. Generally, the matrix had a minor impact on the ultimate NDMA molar conversion for ranitidine (q ¼ 84.1 6.7%), chlorphenamine (q ¼ 3.2 0.7%), and doxylamine (q ¼ 7.5 2.0%). Instead, the matrix components had a more profound impact on the initial lag phase (Lag) and the pseudo-first order rate constant (k). As summarized in Fig. 4, the Lag value is positively correlated with both TOC and SUVA values for all three pharmaceuticals; k value is negatively correlated with TOC and SUVA values for ranitidine and chlorphenamine, but not well related for doxylamine. The estimated model parameters and these correlations support the theory that water matrix components can affect NDMA formation from selected pharmaceuticals by inhibiting the initial reaction with chloramine and/or slowing down subsequent reactions. Chen and Westerhoff (2010) recently found NDMA-FP very difficult to predict based upon bulk water quality measurements such as DOC or UVA254. Results from the current study have
indicated that knowledge of the reaction kinetics is essential in the prediction of NDMA formation. While typical water quality measurements like TOC and SUVA can have significant impact on the reaction kinetics, they are not directly associated with the ultimate NDMA molar conversion, and thus are not appropriate for directly predicting NDMA-FP using empirical models. Moreover, knowledge of the reaction time employed is specifically crucial for compounds that react slowly with chloramine, such as doxylamine. Recently, more and more utilities have shown interest in conducting NDMA-FP tests. Because there is no standard protocol at the moment, many are considering adopting typical disinfection by-products formation potential tests (Summers et al., 1996) and have applied a 24 hr incubation time from a practical viewpoint; however, the NDMA formed after 24 hr from some compounds may only represent a small portion of their ultimate formation potential, especially in real water matrices where the initial reaction could be significantly inhibited. For example, the NDMA molar conversion at 24 hr for doxylamine in TAP and RW samples only accounted for less than 20% of its ultimate NDMA molar conversion. The results have suggested that typical bench-scale NDMA-FP tests may underestimate the ultimate NDMA-FP for some precursors. For water systems with higher water age, prolonged NDMA formation in the outreaches of the distribution system might be a potential risk and should be taken into consideration.
Fig. 5 e Linear correlation between the model-predicted and the independently measured NDMA molar conversion at 24 h for (a) ranitidine; (b) chlorphenamine; (c) doxylamine; and (d) three compounds together (SDS conditions; data from four matrices (MQ, TAP, LW, and RW) and two concentration levels ([Pharmaceutical] [ 5 and 25 nM) were included; the slope was reported as “the best fit value ± the standard error” at 95% confidence level).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
The model was verified by comparing the 24hr NDMA-FP predicted using the estimated model with that measured from independent 24 hr formation potential tests, as summarized in the last two columns of Table 2 (Model Verification). Linear regression was applied between the measured and predicted NDMA molar conversion for each compound individually and for three compounds all together (Fig. 5); the Student’s t-test was then conducted to determine whether the slope of each regression line differed significantly from 1.0. This goodness of fit test has suggested that there is significant correlation between the measured and model-predicted molar conversion (F-test, 95% confidence level), except for chlorphenamine, although even this correlation was determined to be significant at 90% confidence level. The t-test has indicated that the slope of the regression line for ranitidine and chlorphenamine did not differ from 1.0 ( p-value of 99.9% and 15.9%, respectively; 95% confidence level); yet the slope was determined to be smaller than 1.0 for doxylamine and three compounds altogether ( p-value of 3.3% and 0.1%, respectively; 95% confidence level). In general, however, the model-predicted molar conversion was within the 95% confidence interval of the measured value.
4.
Conclusions
NDMA formation kinetics from ranitidine, chlorphenamine, and doxylamine during chloramination was determined in four water matrices. The NDMA conversion over time followed a general three-phase formation curve: an initial lag phase was observed, followed by a fast increase in NDMA formation, and eventually a plateau was reached that represented the ultimate NDMA molar conversion. The NDMA formation profile was relatively independent of the initial pharmaceutical concentration in the same matrix. Water matrix components affected the NDMA conversion rates, most likely by inhibiting their initial contact with chloramine and slowing down the reaction, while they had less impact on the ultimate NDMA molar conversion. A three-parameter kinetic model was proposed to describe the NDMA formation over time during chloramination. The model accurately reflected all the three significant characteristics of the NDMA formation curve, and was able to predict the NDMA molar conversion from the selected pharmaceuticals to within the 95% confidence interval of the measured values. The model needs to be further verified using different potential precursors, water matrices, and reaction conditions. Bulk water quality measurements such as TOC and SUVA were found to correlate better with model parameters Lag and k than with the ultimate NDMA molar conversion (q), indicating interactions between the pharmaceuticals and NOM that might impact NDMA formation are not limited to those based on the general organic character or aromatic nature of either substance. Alternative methods are needed to better characterize the matrix components in order to further investigate their impact on NDMA formation from pharmaceuticals. Knowledge about the formation kinetics is essential in the prediction of NDMA formation from pharmaceuticals. Shortterm NDMA-FP tests (24 h), although practical, may underestimate the contribution of certain slow-reacting precursors.
5693
Thus, prolonged tests (perhaps >4 days for substances examined in this study) are required to determine the ultimate NDMAFP, especially in distribution systems with long water age.
Acknowledgment This research was supported by the Canadian Water Network, the Natural Sciences and Engineering Research Council of Canada, and the Ontario Research Fund. Special thanks are dedicated to Richard Jones and John Armour in the water treatment plants for their assistance in water sampling.
references
American Public Health Association (APHA), American Water Works Association (AWWA), Water Environment Federation (WEF), 2005. In: Eaton, A.D., Clesceri, L.S., Rice, E.W., Greenberg, A.E. (Eds.), Standard Methods for the Examination of Water and Wastewater, 21st ed. Washington, DC. Ballard, B.D., Mackay, A.A., 2005. Estimating the removal of anthropogenic organic chemicals from raw drinking water by coagulation flocculation. Journal of Environmental Engineering 131, 108e118. Blute, N., Russell, C., Chowdhury, Z., Wu, X., Via, S., 2010. Nitrosamine occurrence in the U.S. e analysis and interpretation of UCMR2 data. In: Proceedings of the AWWA Water Quality Technology Conference Savannah, GA, November 14-18, 2010. Charrois, J.W.A., Boyd, J.M., Froese, K.L., Hrudey, S.E., 2007. Occurrence of N-nitrosamines in Alberta public drinkingwater distribution systems. Journal of Environmental Engineering and Science 6, 103e114. Chen, B., Westerhoff, P., 2010. Predicting disinfection by-product formation potential in water. Water Research 44, 3755e3762. Chen, Z., Valentine, R.L., 2006. Modeling the formation of N-Nitrosodimethylamine (NDMA) from the reaction of natural organic matter (NOM) with monochloramine. Environmental Science & Technology 40, 7290e7297. Chen, Z., Valentine, R.L., 2007. Formation of N-Nitrosodimethylamine (NDMA) from humic substances in natural water. Environmental Science & Technology 41, 6059e6065. Chen, Z., Yang, L., Zhai, X., Zhao, S., Li, A., Shen, J., 2010. Nnitrosamine formation during chlorination/chloramination of bromide-containing water. Water Science and Technology Water Supply 10, 462e471. Choi, J., Valentine, R.L., 2002. A kinetic model of Nnitrosodimethylamine (NDMA) formation during water chlorination/chloramination. Water Science and Technology 46 (3), 65e71. De Ridder, D.J., Verliefde, A.R.D., Heijman, S.G.J., Verberk, Q.J.C., Rietveld, L.C., van der Aa, L.T.J., Amy, G.L., van Dijk, J.C., 2011. Influence of natural organic matter on equilibrium adsorption of neutral and charged pharmaceuticals onto activated carbon. Water Science and Technology 63, 416e423. Dotson, A., Westerhoff, P., Krasner, S.W., 2007. Nitrosamine formation from natural organic matter isolates and sunlight photolysis of nitrosamines. In: Proceedings of AWWA Annual Conference and Exposition Toronto, ON, Canada, June 24e28, 2007. EPA Integrated Risk Information System (IRIS), 1993. N-Nitrosodimethylamine; CASRN 62-75-9. www.epa.gov/iris/ subst/0045.htm.
5694
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 8 7 e5 6 9 4
Gerecke, A.C., Sedlak, D.L., 2003. Precursors of N-nitrosodimethylamine (NDMA) in natural waters. Environmental Science & Technology 37, 1331e1336. Health Canada, 2010. Guideline Technical Document on N-Nitrosodimethylamine (NDMA) in Drinking Water for Public Comment. http://www.hc-sc.gc.ca/ewh-semt/consult/_2010/ ndma/draft-ebauche-eng.php#a3. Huber, S.A., Balz, A., Abert, M., Pronk, W., 2011. Characterisation of aquatic humic and non-humic matter with size-exclusion chromatography e organic carbon detection e organic nitrogen detection (LC-OCD-OND). Water Research 45, 879e885. Kemper, J.M., Walse, S.S., Mitch, W.A., 2010. Quaternary amines as nitrosamine precursors: a role for consumer products? Environmental Science & Technology 44, 1224e1231. Kim, J., Clevenger, T.E., 2007. Prediction of Nnitrosodimethylamine (NDMA) formation as a disinfection byproduct. Journal of Hazardous Materials 145, 270e276. Kohut, K.D., Andrews, S.A., 2003. Polyelectrolyte age and Nnitrosodimethylamine formation in drinking water treatment. Water Quality Research Journal of Canada 38 (4), 719e735. Krasner, S.W., 2009. The formation and control of emerging disinfection by-products of health concern. Philosophical Transactions of the Royal Society A 367, 4077e4095. Krasner, S.W., Dale, M.S., Lee, C.F.T., Garcia, E.A., Wong, T.M., Mitch, W., Von Gunten, U., 2010. Difference in reactivity and chemistry of NDMA precursors from treated wastewater and from polyamine polymers. In: Proceedings of the AWWA Water Quality Technology Conference Savannah, GA, November 14e18, 2010. Mitch, W.A., Schreiber, I.M., 2008. Degradation of tertiary alkylamines during chlorination/chloramination: implications for formation of aldehydes, nitriles, halonitroalkanes, and nitrosamines. Environmental Science & Technology 42, 4811e4817. Mitch, W.A., Sedlak, D.L., 2004. Characterization and fate of N-nitrosodimethylamine precursors in municipal wastewater treatment plants. Environmental Science & Technology 38, 1445e1454. Mitch, W.A., Sharp, J.O., Trussell, R.R., Valentine, R.L., AlvarezCohen, L., Sedlak, D.L., 2003. N-nitrosodimethylamine (NDMA) as a drinking water contaminant: a review. Environmental Engineering Science 20 (5), 389e404. MOE, 2003. Ontario Regulation 268/03 made Under the Safe Drinking Water Act, 2002. http://www.e-laws.gov.on.ca/html/ source/regs/english/2003/elaws_src_regs_r03268_e.htm June 25, 2003. Motulsky, H.J., Christopoulos, A., 2003. Fitting Models to Biological Data using Linear and Nonlinear Regression. A Practical Guide to Curve Fitting. GraphPad Software Inc., San Diego, CA. www. graphpad.com. Najm, I., Trussell, R.R., 2001. NDMA formation in water and wastewater. Journal of the American Water Works Association 93 (2), 92e99. Office of Environmental Health Hazard Assessment (OEHHA), 2006. Public Health Goal for N-nitrosodimethylamine and Cadmium in Drinking Water. http://www.oehha.org/water/ phg/cadndma122206.html.
Parris, G.E., 1980. Covalent binding of aromatic amines to humates. 1. reactions with carbonyls and quinones. Environmental Science & Technology 14, 1099e1106. Sacher, F., Schmidt, C.K., Lee, C., von Gunten, U., 2008. Strategies for Minimizing Nitrosamine Formation during Disinfection. Water Research Foundation, Denver, C.O. AwwaRF Report 91209, August 15, 2008. Schmidt, C.K., Brauch, H.J., 2008. N, N-dimethylsulfamide as precursor for N-Nitrosodimethylamine (NDMA) formation upon ozonation and its fate during drinking water treatment. Environmental Science & Technology 42, 6340e6346. Schmidt, C.K., Sacher, F., Brauch, H.J., 2006. Strategies for minimizing formation of NDMA and other nitrosamines during disinfection of drinking water. In: Proceedings of the AWWA Water Quality Technology Conference Denvor, CO, November 5e9, 2006. Shen, R., Andrews, S.A., 2011. Demonstration of 20 pharmaceuticals and personal care products (PPCPs) as nitrosamine precursors during chloramine disinfection. Water Research 45, 944e952. Stackelberg, P.E., Gibs, J., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Lippincott, R.L., 2007. Efficiency of conventional drinkingwater-treatment processes in removal of pharmaceuticals and other organic compounds. Science of the Total Environment 377, 255e272. Summers, R.S., Hooper, S.M., Shukairy, H.M., Solarik, G., Owen, D. , 1996. Assessing DBP yield: uniform formation conditions. Journal of the American Water Works Association 88 (6), 80e93. Thorn, K.A., Pettigrew, P.J., Goldenberg, W.S., Weber, E.J., 1996. Covalent binding of aniline to humic substances. 2. 15N NMR studies of nucleophilic addition reactions. Environmental Science & Technology 30, 2764e2775. USEPA, 2009. Contaminant Candidate List 3 (CCL3). http://water. epa.gov/scitech/drinkingwater/dws/ccl/ccl3.cfm. Valentine, R.L., Choi, J., Chen, Z., Barrett, S.E., Hwang, C., Guo, Y.C. , Wehner, M., Fitzsimmons, S., Andrews, S.A., Werker, A.G., Brubacher, C., Kohut, K., 2005. Factors affecting the Formation of NDMA in Water and Occurrence, Denver, CO, pp. 81e90. Vieno, N., Tuhkanen, T., Kronberg, L., 2006. Removal of pharmaceuticals in drinking water treatment: effect of chemical coagulation. Environmental Technology 27, 183e192. Weber, E.J., Spidle, D.L., Thorn, K.A., 1996. Covalent binding of aniline to humic substances. 1. kinetic studies. Environmental Science & Technology 30, 2755e2763. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science & Technology 39, 6649e6663. Wilczak, A., Assadi-Rad, A., Lai, H.H., Hoover, L.L., Smith, J.F., Berger, R., Rodigari, F., Beland, J.W., Lazzelle, L.J., Kincannon, E.G., Baker, H., Heaney, C.T., 2003. Formation of NDMA in chloraminated water coagulated with DADMAC cationic polymer. Journal of the American Water Works Association 95 (9), 94e106.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Fate of N-nitrosodimethylamine, trihalomethane and haloacetic acid precursors in tertiary treatment including biofiltration Maria Jose´ Farre´*, Julien Reungoat, Francois Xavier Argaud, Maxime Rattier, Ju¨rg Keller, Wolfgang Gernjak The University of Queensland, Advanced Water Management Centre (AWMC), Qld 4072, Australia
article info
abstract
Article history:
The presence of disinfection by-products (DBPs) such as trihalomethanes (THMs), halo-
Received 5 May 2011
acetic acids (HAAs) and N-nitrosamines in water is of great concern due to their adverse
Received in revised form
effects on human health. In this work, the removal of N-nitrosodimethylamine (NDMA),
16 August 2011
total THM and five HAA precursors from secondary effluent by biological activated carbon
Accepted 20 August 2011
(BAC) is investigated at full and pilot scale. In the pilot plant two filter media, sand and
Available online 30 August 2011
granular activated carbon, are tested. In addition, we evaluate the influence of ozonation prior to BAC filtration on its performance. Among the bulk of NDMA precursors, the fate of
Keywords:
four pharmaceuticals containing a dimethylamino moiety in the chemical structure are
Adsorption
individually investigated. Both NDMA formation potential and each of the studied phar-
Biological activated carbon
maceuticals are dramatically reduced by the BAC even in the absence of main ozonation
Disinfection by-product
prior to the filtration. The low removal of NDMA precursors at the sand filtration in
NDMA
comparison to the removal of NDMA precursors at the BAC suggests that adsorption may
Sand filtration
play an important role on the removal of NDMA precursors by BAC. Contrary, the
Ozone
precursors for THM and HAA formation are reduced in both sand filtration and BAC indicating that the precursors for the formation of these DBPs are to some extent biodegradable. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Disinfection by-products in water are of great concern to public health since bladder and colorectal cancers have been associated with exposure to them in drinking water, and experimental evidence suggests that exposure also occurs through inhalation and dermal absorption (Villanueva et al., 2007). The US Environmental Protection Agency (USEPA) allows a maximum of 80 mg/L of total THMs (TTHMs) and 60 mg/L of five HAAs in drinking water based on its current regulation guidelines (Richardson et al., 2007). For NDMA,
even if it is not yet included in the drinking water regulation, the USEPA classifies it in the group B2, which includes compounds that are probably carcinogenic to humans (EPA, 2008). In Australia, NDMA has been included in the draft Australian Drinking Water Guidelines at a maximum concentration of 100 ng/L (ADWG, 2010). Moreover, NDMA was recently identified as one of the DBPs with the greatestpotential impact on public health (Hebert et al., 2010). While THMs and HAAs are mainly formed when water is disinfected with chlorine (Richardson et al., 2007), NDMA has been related to the presence of chloramines, specifically
* Corresponding author. Tel.: þ61 7 33463233; fax: þ61 7 33654726. E-mail address: [email protected] (M.J. Farre´). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.033
5696
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
dichloramines generated during the disinfection process (Schreiber and Mitch, 2006). Population increases, particularly in cities, and scarce water resources have increased the demand for use of highly treated municipal wastewater as a source of potable water (Shannon et al., 2008). Studying the fate of DBP precursors during secondary effluent treatment is of importance as an increasing number of municipal water treatment plants are engaged in the practice of potable reuse of treated wastewaters. A common method to measure the precursors of DBP in water is by means of formation potential tests. In these tests, chlorine or chloramines are added to a buffered sample at relatively high concentrations and kept reacting for at least seven days to achieve the maximum formation of the specific DBPs (Greenberg et al., 1992; Mitch et al., 2003). While THM and HAA precursors are difficult to characterize because different fractions of NOM may generate these DBPs upon disinfection (Xie, 2004), some particular NDMA precursors may be easier to monitor because the specific dimethylamino moiety is required to generate the nitrosamine upon chloramination. Traditionally, dimethylamine (DMA) was the first NDMA precursor considered during water chloramination since direct reaction between this molecule and chloramine produces the carcinogen. Dimethylamine is not only the typical catabolic product of proteins in animals and plants (implying a typical concentration in urine of 40 mg/L) (Tricker et al., 1994) but also amongst the most frequently produced amines by the chemical industry. Recently, it has also been shown that other pharmaceuticals and personal care products with substituted amino groups can serve as NDMA precursors during chloramine disinfection (Lee et al., 2007; Kemper et al., 2010; Shen and Andrews, 2010). In addition, natural organic matter is another typical source of NDMA (Westerhoff and Mash, 2002; Chen and Valentine, 2006, 2007, 2008). Wastewater treatment plants (WWTPs) remove 90% of dissolved organic nitrogen (DON) and the typical concentration of DON remaining after treatment at WWTPs is in the range from 1 to 3 mgN/L (Pehlivanoglu-Mantas and Sedlak, 2006). This remaining percentage consists of either difficult-to-remove DON species, or DON produced during biological treatment. Finding technologies and processes to reduce DBP precursors in water is thus of great interest. Adsorption on activated carbon (AC) is one of the proven methods for removing natural organic matter (NOM). However, the capacity for adsorption is limited and replacement and disposal is costly. AC with active biomass established on its surface is called biological activated carbon (BAC). A BAC filter consists of a fixed bed of granular AC supporting the growth of bacteria attached on the surface. This technology has been used for many years in drinking water treatment, usually after ozonation, and has proven to significantly remove NOM, pharmaceuticals and personal care products and ozonation by-products as well as odour and taste compounds (e.g. geosmin and 2-methylisoborneol) (Simpson, 2008; Reungoat et al., 2011). Although some work has already been published showing the efficiency of BAC to remove DBP precursors in drinking water (Simpson, 2008), to our knowledge, there are no reports regarding the degradation of THM, HAA and N-nitrosamine
precursors in secondary treated effluent using ozone/BAC. In the present study, we investigated the removal of NDMA, HAA and THM precursors in full and pilot-scale biofilters treating the effluent from a municipal WWTP. Additionally, the fate along the treatment train of specific pharmaceuticals containing tertiary amines (hence suspected to be NDMA precursors) is presented.
2.
Materials and methods
2.1.
Chemicals
All chemicals used for chemical analysis were of analytical grade and commercially available. NDMA (5000 mg/mL in methanol) had a purity of >99.9% and was obtained from Supelco. Deuterated d6-NDMA and d14-NDPA (N-nitrosodipropylamine) were used as surrogate and internal standard, respectively (1000 mg/mL in dichloromethane, >98.9%, supplied by Accustandard and Ultra Scientific, respectively). For the NDMA formation potential test, ammonium chloride (TraceSELECT, 99.9% purity), sodium hydroxide (SigmaUltra, 98%, pellets) and sodium hypochlorite solution (reagent grade, available chlorine 4%) were used. Potassium dihydrogenphosphate (KH2PO4, Fluka, puriss. p.a., 99.5%) and disodiumhydrogenphosphate (Na2HPO4$2H2O, Fluka, puriss. p.a., 99.5%) were used to prepare pH buffer solutions. To quench the chloramines solution, sodium sulphite (Fluka, puriss. p.a., 98.0%) was employed. Commercial DPD test kits (Hach) were used for the analysis of free and total chlorine (DPD Total and Free Chlorine Reagent, test tube vials 2105545 and 2105645). To standardize the chlorine solution, sodium thiosulphate (SigmaUltra, 99.5%), potassium dichromate (SigmaUltra, 99.5%), acetic acid (ReagentPlus, 99%), soluble starch (ACS reagent) and potassium iodide (ReagentPlus, 99%) were used. For solid phase extraction (SPE), EPA commercial charcoal optimized for NDMA analysis (Restek) was used. HPLC grade dichloromethane, methanol and water were used for conditioning and cleaning the SPE cartridges. Anhydrous sodium sulphate, granular 10e60 mesh from Mallinckrodt was used to remove water from the extracts. Finally 99% decane (SigmaeAldrich) was used as keeper in the final concentration step. Chemical standards of, venlafaxine hydrochloride, roxythromycin, tramadol hydrochloride and doxylamine were purchased from SigmaeAldrich (Steinheim, Germany) at analytical grade (99%).
2.2.
Full scale reclamation plant
The South Caboolture Water Reclamation Plant was designed to reduce riverine pollution from the 40,000 population equivalent wastewater treatment plant and to provide recycled water to industry and community consumers (van Leeuwen et al., 2003; Reungoat et al., 2010). The treatment process as detailed in Fig. 1a incorporates biological denitrification, pre-ozonation, coagulation/flocculation/dissolved air flotation-sand filtration (DAFF), main ozonation, granular activated carbon (GAC) filtration and final ozonation for disinfection. The GAC was replaced in March 2008, 19 months before the first sampling campaign, and the filter had treated
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
5697
Fig. 1 e South Caboolture Water Reclamation Plant with sampling locations S (a) and pilot-scale biofilters (b). WWTP [ wastewater treatment plant, HRT [ hydraulic residence time, SRT [ sludge residence time. ----> refers to pilot scale system.
approximately 45,000 bed volumes by that time. The properties of the GAC “Acticarb BAC GA1000N” (Activated Carbon Technologies Pty Ltd, Australia) are detailed in Table SI 1. The empty bed contact time (EBCT) is 18 min.19 months of operation have been shown to be sufficient for the development of bacteria on the AC filters (Simpson, 2008), as confirmed by its oxygen consumption (Reungoat et al., 2011). Therefore, the full-scale filter is assumed to be biologically active. However, in order to simplify the reading of the paper so as not confuse with the pilot plants, we have kept the AC nomenclature for the full scale unit in the manuscript.
2.3.
Pilot-scale biofilters
Three pilot-scale biofilters (Fig. 1b) were set up in December 2006 at the South Caboolture Water Reclamation Plant, Australia (Fig. 1a) parallel to the full scale AC. The biofilters are 3 m high and 22.5 cm internal diameter PVC columns; they consist of 80 1 cm filtering bed height supported by a 20 cm layer of gravel at the bottom, the top of the columns are filled with water. One column contains sand as the filtering medium and the other two are filled with the same GAC as the full-scale filter but with a slightly lower particle size. Details on the filtering media can be found in the supplementary information (Table SI 1). The filters were continuously fed with water from the main stream of the reclamation plant; BAC 1 received preozonated, whereas BAC 2 and SAND received water after main ozonation. Pre-ozonated water refers here and in the rest of the
manuscript to the denitrified effluent after pre-ozonation and dissolve air flotation and filtration. A prior study showed that the ozone dose added in the pre-ozonation is very low relative to the dissolved organic carbon (DOC) concentration at this stage (0.1 mgO3 mg1DOC) and does not lead to any significant removal of DOC or organic micropollutants (Reungoat et al., 2010). To support biological activity in the BAC filters, compressed air and later 90% oxygen gas was bubbled in the water column above the filtering bed to increase the dissolved O2 concentration. The EBCT was controlled by adjusting the effluent flow rate at the bottom of the columns and set at 60 min in all filters. The top layer of each filter bed (SAND and BAC filters) was stirred weekly while withdrawn from above the filter, to avoid clogging of the columns. This operation removed some of the biomass from the top of the filter; however no backwash of the entire filter was performed. A previous study showed that biological activity had developed on the filtering media and dissolved organic removal had reached a steady state by June 2007 (Pipe-Martin et al., 2010).
2.4.
Sample collection
During the first sampling campaign (October 2009), two sets of time proportional 24-h composite samples were collected to quantify NDMA formation potential. Sampling points are shown in Fig. 1a. As the flow rates in the reclamation plant (due to the presence of storage tanks) and in the pilot scale filters were constant at the time of sampling, representative samples
5698
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
were collected using continuous pumping at 7 ml/min. Samples were collected into glass bottles pre-washed with MilliQ water and HPLC grade methanol and rinsed with the water sampled directly before sampling. The samples were protected from light and refrigerated during collection. 1 L of each composite sample was then transferred into an amber glass bottle and transported on ice to the laboratory for nitrosamines quantification and NDMA formation potential test. During the second sampling campaign (July 2010), three sets of grab samples were collected to quantify instantaneous concentrations and formation potential of THMs and HAAs. Because the WWTP and the balance tank are located upstream of the reclamation plant, variations in the water quality were not expected to occur within the time of sampling. For THMs and HAAs quantification, 200 mL of sample were collected in 2 amber glass bottles supplied by Queensland Health Forensic and Scientific Services (QHFSS). In order to quench any chlorine residual, 200 mg of ammonium chloride were added to each bottle before sample collection. These samples were transported on ice to QHFSS for direct analysis. To perform the formation potential test, 1 L of sample was collected in amber glass bottles that had been soaked in soap overnight, then soaked in 10% nitric acid solution overnight, finally rinsed with MilliQ water and HPLC grade methanol and dried in an oven at 105 C. This sample was transported on ice to the laboratory. During all sampling procedures, particular care was taken to avoid stripping of the volatiles DBPs by slowly filling the bottles and leaving no head space. Simultaneously to other sample collection, 100 mL of sample were collected in MilliQ rinsed plastic bottles for DOC and nutrient concentration determination.
Bassett, 2004). N-nitrosodimethylamine (NDMA), N-nitrosodiethylamine (NDEA), N-nitrosomorpholine (NMOR), Nnitrosopiperidine (N-Pip), N-nirosodibuthylamine (NDBA) were included in the analysis. The details of the analysis are published elsewhere (Farre´ et al., 2011). In short, water is passed through a carbon solid phase extraction cartridge and the N-nitrosamines are eluted off with dichloromethane. The extracts are concentrated by evaporation under nitrogen to 1 mL and analysed by capillary GC-mass spectrometer in positive chemical ionisation (PCI) mode with anhydrous ammonia as the chemical ionisation gas (Finnigan Trace G.C. Ultra and Finnigan Trace DSQ Mass Spectrometer). The detection limit for the technique used was 5 ng/L for NDMA, 10 ng/L for NDEA and NMOR, and 20 ng/L for N-Pip and NDBA.
2.5.3.2. NDMA formation potential (FP) test. The NDMA FP test follows closely the procedure described as nitrosamine precursor test by Mitch et al. (2003) and is described elsewhere (Farre´ et al., 2011). A concentration of 2 mM (140 mg/L Cl2) was used to determine the NDMA FP of the selected samples to ensure that all precursors are reacting with chloramines to generate NDMA. The test was performed at pH 6.8 which was achieved by adding 700 mg/L KH2PO4 and 880 mg/L Na2HPO4$2H2O to the water sample (10 mM phosphate buffer). All experiments were performed in 1 L amber glass bottles which were stored at room temperature (23 2 C) in the dark for seven days. On the seventh day, the residual chloramine concentration was quenched with 2.5 g/L sodium sulphite (added in solid phase to the sample) to prevent further NDMA formation. The samples were then analysed for NDMA.
2.5.4. 2.5.
Analytical methods
2.5.1.
Dissolved oxygen
Dissolved oxygen (DO) concentration was measured with an YSI 6562 Dissolved Oxygen Probe connected to an YSI MDS 650 multi-parameter display system. An YSI 6560 conductivity and temperature probe connected to the same multiparameter display system allowed to simultaneously correct the DO concentration value and display it directly as a concentration.
2.5.2.
Dissolved organic carbon and nutrients
Prior to analysis, samples were filtered through a 0.45 mm PTFE membrane. The DOC was measured as non-purgeable organic carbon (NPOC) with an Analytik Jena multi N/C 3100 instrument. For each sample, 2e3 replicates were measured, giving a relative standard deviation of less than 3%. Ammonia, nitrite, total NOx and phosphate were measured on a Lachat flow injection analyzer as per the Lachat QuickChem method 31-107-06-1-A. Dissolved organic nitrogen (DON) was calculated to be the difference between total Kjeldahl nitrogen (TKN) and NH4eN nitrogen. TKN was measured using the Lachat QuickChem method 10-107-06-2-D.
2.5.3.
Nitrosamines
2.5.3.1. Nitrosamines quantification. The method used for N-nitrosamines was based on EPA Method 251 (Munch and
Trihalomethanes and haloacetic acids
2.5.4.1. Trihalomethanes quantification. Chloroform (TCM), bromodichloromethane (BDCM), dibromochlorometane (DBCM) and bromoform (TBM) were quantified by QHFSS by purging the volatile organic directly from the aqueous sample and subjecting the volatilized component to gas chromatography-mass spectrometry, in accordance with USEPA method 524.2 revision 4.1 (Munch, 1995). This was achieved using a using a Shimadzu QP2010 equipped with a purge and trap system (Tekmar Velocity). A column ZB-624 (20 m length 0.18 mm inner diameter 1.0 mm film thickness) was used for separation. Injection was done 1/1 split at 200 C. The carrier gas employed was helium at 42 mL/min and 118 kPa. Initial temperature of the oven was 40 C, held for 2 min and then increased to 200 C at a rate of 10 C/min rate. The mass spectrometer operating conditions were: ion source and interface line temperatures 200 C and 230 C; Scan 35e300 amu at 1111 amu/sec. The LOQ is 1 mg/L for all analytes. 2.5.4.2. Haloacetic acids quantification. Five HAAs (abbreviation 5HAAs; monochloroacetic acid-MCAA, dichloroacetic acid-DCAA, trichloroacetic acid-TCAA, bromochloroacetic acid-BCAA, monobromoacetic acid-MBAA and dibromoacetic acid-DBAA) were extracted from aqueous samples by portioning into methyl tert-butyl ether (MtBE) after adding sulphuric acid and sodium sulphate following USEPA method number 552.3 (USEPA., 2003).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
2.5.4.3. THMs and HAAs formation potential test. The formation potential test was performed following Standard Methods for the Examination of Water and Wastewater (Greenberg et al., 1992). Chlorine demand of each sample was determined with a 4 h test prior to starting the 7 days formation potential tests to calculate the chlorine dose necessary to ensure a residual of free chlorine at the end of the experiment. The 7 days test was performed in 1 L amber glass bottles washed following the same procedure as for the sample collection and filled completely to avoid the presence of a head space. Bottles were kept in the dark at a constant temperature of 23 2 C. At the end of the 7 days, the residual free chlorine was measured to ensure chlorine did not limit the reaction. Then 200 mL were collected in each of 2 amber glass bottles supplied by QHFSS for THMs and HAAs quantification. The residual chlorine was quenched by adding sodium sulfite (0.1 mL of 100 g/L solution per 25 mL of sample) and ammonium chloride (0.2 mL of 50 g/L solution per 250 mL of sample) in the THMs and HAAs bottles respectively, as recommended by the standard method. The bottles were sent immediately to QHFSS for analysis. A blank sample (MilliQ water) was included in each test batch for quality control purposes, they were all below LOQ for all THMs and HAAs.
3.1.2.
Micropollutants
Doxylamine, roxithromycin, tramadol and venlafaxine were quantified according to the method described in Reungoat et al. (2011) in the sampling campaigns done for NDMA precursors and in three additional sampling campaigns. The method consisted of solid phase extraction (SPE), elution, concentration, and analysis by liquid chromatography coupled with tandem mass spectrometry (LC/MS-MS).
NDMA formation potential
The same samples were subjected to a NDMA formation potential test for 7 days. Fig. 2 shows the result of the NDMA formation potential across the full-scale plant and the pilot plant. No other N-nitrosamines, considered in this work, were observed to be formed above the LOQ during the formation potential tests. The NDMA formation potential measured at the influent of the reclamation plant was 423 55 ng/L and remained in the same range after denitrification confirming that this treatment does not affect NDMA precursors (Mitch and Sedlak, 2004). The NDMA formation potential of the secondary effluent used in South Caboolture Water Reclamation Plant was found to be similar to other domestic wastewater treatment plants in South East Queensland (Farre´ et al., 2011) and in other countries (Pehlivanoglu-Mantas and Sedlak, 2006) supporting that no effluents with high NDMA formation potential were discharged to this specific WWTP. Pre-ozonation (2 mgO3/L) and DAFF reduced the NDMA formation potential by around 20% each bringing the concentration down to 260 31 ng/L. The main ozonation (5 mgO3/L with 15 min contact time) was the most effective step of the full scale treatment, reducing the NDMA formation potential by another 66% to levels below 100 ng/L (this corresponds to the 78% of total NDMA precursors removal from the beginning of the treatment as shown in Fig. 2). This data follows the trends observed by Lee et al. (2007) when measuring the effect of ozone treatment on NDMA precursors in natural waters. That study reported a NDMA formation potential reduction of 32e94% by applying up to 40 mM (1.9 mgO3/L) ozone, depending on the natural water and
NDMA FP (ng/L)
2.5.5.
and a BAC filter after main ozonation (BAC 2). NDMA and NMOR were detected beyond LOQ across the plant (i.e., 5 ng/L for NDMA and 10 ng/L for NMOR).
500
100
400
80
300
60
200
40
100
20
0
3.
Results and discussion
3.1.
Nitrosamines and NDMA formation potential
3.1.1.
Nitrosamines
NDMA, NDEA, NMOR, N-Pip and NDBA were analysed in all the samples taken from South Caboolture Water Reclamation Plant. Sampling points along the treatment train are indicated in Fig. 1. The selected N-nitrosamines were also analysed in the effluent of the three pilot scale columns, which are BAC filter placed after pre-ozonation and dissolve air flotation and filtration (BAC 1), sand filtration after main ozonation (SAND)
2 nt C 1 F on nt on on on AC st A fflue ati AF AC lue ati ati ati o e iltr ost B eff trific ozon ost D ost B ozon p f l y a r d p p p da fin eni ain t san pre on st d post st m pos sec po po
% NDMA FP removal
The analysis was carried out by QHFSS using a gas chromatography coupled with an electron capture detector (GCECD) at 300 C (Shimadzu GC2010). A column DB-1701 (15 m length 0.32 mm inner diameter 0.25 mm film thickness) was used for identification and a second column DB5 (30 m length 0.25 mm inner diameter 0.25 mm film thickness) was used for confirmation. 1 mL of sample was splitless injected at 250 C. The carrier gas employed was hydrogen at 68.1 mL/min and 45.3 kPa. Initial temperature of the oven was 50 C, held for 1 min and then increased to 75 C at a rate of 5 C/min rate. The temperature was finally increased to 280 C at a 40 C/min. The LOQ is 10 mg L1 for MCAA, DCAA and TCAA and 5 mg L1 for BCAA, MBAA and DBAA.
5699
0
Fig. 2 e Bar charts correspond to NDMA precursors measured by NDMA formation potential test (FP) across South Caboolture Water Reclamation Plant and pilot-scale biofilters. Error bars correspond to the standard deviation of two independent sampling campaigns (n [ 2). Dot points correspond to the percentage of NDMA precursor’s removal relative to the raw water. Striped bars represent results from the pilot plants while plain bars are results from the full-scale plant.
5700
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
oxidation conditions. In South Caboolture Water Reclamation Plant, the activated carbon filter reduced the NDMA precursors further, leaving a concentration of NDMA precursors in the final effluent of 53 6 ng/L. BAC 1 (pilot plant fed with water post DAFF) was able to reduce the NDMA formation potential by more than 85%, which is better than the main ozonation alone, and the effluent concentration was 48 4 ng/L, similar to the effluent of the main plant. The SAND filter was not able to reduce the NDMA FP after main ozonation but the BAC 2 decreased NDMA FP down to 39 9 ng/L, which is only slightly better than BAC 1. Ammonia, nitrate, DOC and dissolved organic nitrogen (DON) concentrations from the same sampling campaign are shown in Figure SI 1 of the supporting information. Overall, the average DON concentration measured across the treatment scheme decreased form an initial value of 1.6 mg/L to 1.0 mg/L after main ozonation with a further reduction to 0.8 mg/L after AC (18 min EBCT), while the concentration after BAC 1 (60 min EBCT) was 0.6 mg/L showing that this treatment is more effective than main ozonation plus AC at the full scale. From the residual 1.0 mg/L DON left after main ozonation, 0.15 mg/L could be degraded at the sand filtration while 0.5 mg/L could be removed at BAC 2 leaving a final 0.5 mg/L of DON when ozone plus BAC 2 are employed. Regarding DOC the same trend was observed. The initial concentration of DOC at the plant was 11 mg/L and this value was reduced to 6.5 mg/L after the main ozonation and to 4.7 mg/L after AC at the full-scale plant. BAC 1 was able to decrease DOC concentration to 4.7 mg/L even without the need of adding ozone in the system. From the residual 6.5 mg/L DOC left after main ozonation, 1.8 mg/L could be degraded at the sand filtration while 3.4 mg/L could be removed at BAC 2 leaving a final 3.1 mg/L of DOC when ozone plus BAC 2 are employed. It has been reported that NDMA formation potential cannot be predicted by simple measurements such as DOC because of the specific nature of NDMA precursors that is a very minor and specific fraction of DOC (Pehlivanoglu-Mantas and Sedlak, 2008). However, our data across the plant showed that this measurement may act as a good indicator of the NDMA formation potential for this specific scenario since the coefficient of regression when plotting these two parameters was higher than 0.9. For DON also a linear relation (R2 ¼ 0.8) could be observed even though the fitting was lower in this case. This data is plotted in the Figure SI 2 of the supporting information. The fraction of DOC removal in the BAC 1 (i.e., 56%) is not as high as the fraction of NDMA formation potential removal (85%); hence, it appears that the DOC removed at the BAC 1 contains a higher concentration of NDMA precursors, which means that the NDMA precursors are preferentially adsorbed or degraded in the BAC compared to DOC.
3.1.3.
NDMA model precursors
NDMA precursors contain a tertiary amino group with two methyl substituents except for dimethylamine which is the only secondary amine that can act as NDMA precursor. Dimethylamine concentrations in primary wastewater effluents are typically in the range of 20e80 mg/L. Since dimethylamine is degraded by bacteria, levels in secondary wastewater effluents are lower (<10 mg/L) (Mitch et al., 2003; Mitch and
Sedlak, 2004). This low concentration commonly present in secondary effluents could not explain the total concentration of NDMA precursors measured since the molar yield of dimethylamine conversion to NDMA is lower than 0.5% (Mitch et al., 2003). Although we did not measure dimethylamine in the samples, we considered other tertiary amines present in the secondary effluent to investigate the fate of specific NDMA precursors and along the treatment train. To this aim, four pharmaceuticals known to be NDMA precursors were analysed for and found at low mg/L levels in the secondary effluent used as influent in South Caboolture reclamation plant. Doxylamine, with a conversion yield to NDMA during chloramination of 8.0e9.8% (Shen and Andrews, 2010), was found in the influent to the reclamation plant at an average concentration of 289 ng/L. Roxithromycin, tramadol and venlafaxine were also found in the source water of the treatment plant at average concentrations of 240, 1300 and 1330 ng/L, respectively. However, the yield of conversion for these pharmaceuticals is around 0.5% (Shen and Andrews, 2010). The chemical structures of these NDMA precursors are plotted in Fig. 3 in conjunction with the logD and pKa values. The fate of the pharmaceuticals in conjunction with the NDMA formation potential is plotted in Fig. 4. Even though the presence of these four pharmaceuticals could only account for less than 4% of the NDMA precursors present in the inlet of South Caboolture Water Reclamation Plant based on the conversion yields published by Shen and Andrews (2010), they may be used as indicators of the fate of pharmaceuticals containing the dimethylamino moiety in the chemical structure across BAC. The high removal of both NDMA formation potential and the suspected pharmaceutical NDMA precursors observed in BAC 1 could be due to biodegradation activity of the biomass attached on the surface of activated carbon, adsorption of the compounds on the surface of the activated carbon or the combined effects of adsorption and biodegradation. Some authors have also hypothesized that the biodegradation continuously regenerates adsorption sites by degrading adsorbed molecules (Herzberg et al., 2003; Simpson, 2008). This effect is called bioregeneration. In a previous article Reungoat et al. (2011) observed that the concentration of pharmaceuticals and personal care products (PPCP) removed in the same BAC 1 used in the present study remained constant over more than two years of continuous operation. Therefore, they suggest that the removal of organic matter and PPCPs observed in the BAC filters was due to biodegradation (or adsorption followed by biodegradation) rather than adsorption alone as adsorption efficiency would typically decrease over time. When evaluating the fate of the NDMA precursors selected for this study we observed that the four pharmaceuticals, which have a logD(pH7)<2 and a pKa>8, remained at similar concentration before and after the SAND filter but decreased significantly after BAC 1 or BAC 2. The low removal of NDMA precursors during SAND filtration in comparison to the BACs suggests that adsorption may play a significant role in the removal of NDMA precursors. Hydrophobic interactions are the dominant mechanism in activated carbon adsorption of neutral organic compounds (Yoon et al., 2003). However, for charged molecules the charge interactions are of high importance. Activated carbon is usually negatively charged at pH typical for drinking water treatment, hence,
5701
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
Fig. 3 e Pharmaceuticals known to be NDMA precursors detected in the treatment plant. Log D is calculated using Advanced Chemistry Development (ACD/Labs) Software V11.02 (ª 1994e2011 ACD/Labs). pKa from Reungoat et al. (2011).
while in BAC 1 and BAC 2 this number increased to 9.6 and 9.9 mg/L, respectively. This oxygen consumption in both BAC 1 and BAC 2 confirms that there is higher biological activity in the BAC filters compared to the SAND filter which may explain
1000
10000 doxylamine roxithromycin tramadol venlafaxine NDMA FP
1000
100
100
10
10
1
1
0.1
0.1
1 2 n F C nt nt tion tion on AF AC atio ati AC st A fflue a a lu e o e eff trific ozon ost D ost B ozon filtr ost B p l y a r p p p D in da deni prefin AN on ma st st ost S sec po po p
Fig. 4 e Fate of specific NDMA precursors across South Caboolture Water Reclamation Plant and pilot-scale biofilters, error bars correspond to standard deviation of five independent sampling campaigns (n [ 5). The dot points correspond to the average NDMA formation potential.
NDMA FP (ng/L)
10000
ng/L
increasing the adsorption of positively charged compounds as the ones studied in this work. In accordance, Westerhoff et al. (2008), previously reported a good removal by adsorption on PAC of protonated compounds. Nevertheless, activated carbon typically has a surface area of several hundred square meters per gram due to its high porosity but most of this surface is not accessible to bacteria as it is located in micropores with a diameter smaller than 2 nm (Reungoat et al., 2011). Nonetheless, the external surface of the activated carbon grains is much rougher and uneven than the surface of sand grains and therefore potentially provides more sites for the bacteria to attach. Therefore, both a higher population of bacteria attached to the surface and a higher adsorption capacity could explain the enhanced removal of NDMA precursors observed at BAC in comparison to sand filtration for doxylamine, roxithromycin, tramadol and venlafaxine. It cannot be excluded that extracellular polymeric substances such as polysaccharides produced by bacteria attaching to the activated carbon grains could adsorb the pharmaceuticals with medium sorption potential in comparison to the GAC without biological activity (Zhang et al., 2010). In order to gain insight into understanding the hypothesis of biodegradation of NDMA precursors as mechanism for their removal, the dissolved oxygen was measured in the inlet and outlet of the three pilot scale filters (i.e., BAC 1, SAND filter and BAC 2). The average uptake of oxygen in the SAND filter was 2.7 mg/L
5702
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
the higher removal of NDMA precursors observed. However, adsorption cannot be completely discarded and more studies are being undertaken to distinguish biodegradation and adsorption effects in the BAC filters.
3.2.
Other disinfection by-products
3.2.1.
Trihalomethanes and haloacetic acids
TTHMs and five (5HAAs) are regulated DBPs in both drinking and recycled water. Therefore, we also evaluate the fate of those DBP precursors across Caboolture Water Reclamation Plant and the pilot plants. To this aim, TTHMs and five HAAs and their precursors were analysed in three different sampling campaigns. Sampling points included at the fullscale plant were: post DAFF, post main ozonation, post AC and final effluent. The DBPs were also analysed at the three pilot scale columns. No HAAs were measured above the LOQ for any of the sampling points during the different sampling campaigns. On the other hand low concentrations of THMs were measured across the treatment train but the TTHM concentration was always below 11 mg/L (see Table SI 2 of the supporting information).
3.2.2.
Trihalomethane and haloacetic acid precursors
The same samples were subjected to DBP formation potential for 7 days, as previously described. Fig. 5 shows the result of HAA and THM formation potential of the selected samples 400
8
150
350
6
100
4
50
2
300
250
5HAAs (µg/L)
DCAA TCAA BCAA DBAA DOC 5HAAs
DOC (mg/L)
individual HAA (µg/L)
200
200
150
0
100
TCM BDCM DBCM TBM DOC TTHMs
8
400
250 200 150
350 6
300
4
100
250
200
TTHMs (µg/L)
300
individual THM (µg/L)
0 AC st po
DOC (mg/L)
on on FF C1 C2 ati ati BA DA BA on iltr st st st oz df po po po n in a a s m st st po po
2
50
150
0
0
on FF C1 C2 ion ati rat BA BA DA on st st st filt oz o d t po po p s san po st po
100
AC st po
Fig. 5 e HAAs and THMs precursors and DOC across South Caboolture Water Reclamation Plant and pilot-scale biofilters, error bars correspond to standard deviation of three independent sampling campaigns (n [ 3).
across the full-scale plant and the pilot plant filters in conjunction with 5HAAs, TTHMs and DOC data. As seen in Fig. 5, monohalogenated acids were not formed in the formation potential test. Among the HAAs generated during the tests, DCAA was measured at the highest concentration (154 32 mg/L) followed by TCAA (143 17 mg/L) and BCAA (30 6 mg/L). The formation of chlorine-containing HAAs was significantly reduced after BAC 1 following the trend observed for NDMA precursors (76 36 mg/L and 60 36 mg/L for DCAA and TCAA, respectively), however this behavior was not observed with bromine-containing HAAs since the concentration of BCAA and DBAA remained unchanged by BAC 1. The same trend was observed for THMs. TCM formation (224 24 mg/L after DAFF) was reduced significantly in the BAC 1 (103 36 mg/L) but brominecontaining DBP formation remained constant along the plant. At the BAC 1 the organic matter is reduced from around 7 mg/L to 5 mg/L as seen in Fig. 5, either by adsorption or biodegradation. Nevertheless, the ion content remains constant as shown by the conductivity data presented in the supportive information (Table SI 3). Since we could not measure bromate formation either above the LOD (i.e., 10 mg/L) across the treatment plant we assumed the oxidation of Br to BrO 3 by ozone was minimal. Therefore all Br was available to be oxidised to HOBr by HOCl during the formation potential test. The rate constant of bromide with HOCl to generate HOBr is 1.5 103 M1 s1 (Kumar and Margerum, 1987) and the rate constant of THMs formation is in the range of 0.01 and 0.03 M1 s1 (Gallard and Von Gunten, 2002). It is known that once formed, bromine reacts about 10 times faster than chlorine with natural organic matter since the activities of electrophilic substitution for electron release to stabilize carbocation are more favourable for the Br atom due to its higher electron density and smaller bond strength relative to the Cl atom (Westerhoff et al., 2004; Hua et al., 2006). Hence, the formation of Br-DBPs is limited by the initial Br concentration whereas the Cl-DBPs would be limited by the organic matter. Therefore, when organic matter decreases along the treatment train, the formation of chlorine-containing DBPs is reduced while the formation of bromine-containing DBPs remains constant. Main ozonation removes the precursors for TCAA and TCM while biofiltration decreases also the concentration of DCAA precursors. The increase of DBCM observed by others (Chen et al., 2009) is also seen to a small degree in our data as the concentration of this DBP increases from 11 mg/L to 15 mg/L from after DAFF to after ozonation. Liang and Singer (2003) have suggested that bromide is more reactive with aliphatic precursors, such as hydrophilic organic material, than with aromatic precursors, such as hydrophobic organic material. Hence, the organic matter becoming more hydrophilic after ozonation may explain the increase of the formation of this specific DBP. The concentration of DCAA, TCAA and TCM precursors were further reduced in the sand filtration to 24, 18 and 22 mg/L, respectively showing that biodegradation indeed plays a role in their reduction since adsorption is considered minimal in the SAND filter. Further in the treatment, the formation of these DBPs is reduced in BAC 2 and AC. In general, the reduction of TTHM and 5HAA precursors was proportional to the decay of DOC across the plant with
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
R2 > 0.9, as seen in Figure SI 3 of the supporting information, confirming that this simple parameter is a good indicator to estimate the precursors for those specific DBPs in treated secondary effluents.
4.
Conclusions
Both, the bulk of NDMA precursors and individual pharmaceuticals that may form NDMA are dramatically reduced by the BAC filters even without preceding main ozonation. DOC removed by the BAC contains a higher concentration of NDMA precursors than the bulk DOC, which means that the NDMA precursors are preferentially adsorbed or degraded in the BAC. The percentage of removal in the BACs is unaffected by a main ozonation step when considering long filtration times (EBCT ¼ 60 min). On the contrary, under similar conditions, sand filtration does not significantly remove NDMA precursors. Since adsorption is considered minimal on the sand filtration while biodegradation may potentially take place and due to the different removal observed at the BACs and sand filter, adsorption is deemed likely to be important for the observed removal NDMA precursors during BAC filtration. However, more research is necessary to fully understand the removal mechanisms in order to quantify the adsorption versus biodegradation processes. THM and HAA precursors follow the trend of NDMA precursors since they are also considerably reduced by the BACs. However, the precursors for these specific DBPs are also removed in the sand filter. Hence, both adsorption and biodegradation may play a significant role on their removal. The formation potential for brominated THMs and HAAs remains constant across the different treatment steps due to increase on the bromine/DOC ratio along the filtration steps.
Acknowledgements The authors want to specifically acknowledge Urban Water Security Research Alliance for funding the “NDMA formation potential project” and “The Enhanced Treatment Project” and Unity Water and their staff for giving access to the plant. Thanks to Miss Hollie King for correcting the English.
Appendix. Supplementary material Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.033.
references
Australian Drinking Water Guideline, 2010. Public Draft Document for Consultation. http://www.nhmrc.gov.au/ guidelines/consult/consultations/draft_adwg_guidelines.htm. Chen, Z., Valentine, R.L., 2006. Modelling the formation of N-nitrosodimethylamine (NDMA) from the reaction of natural
5703
organic matter (NOM) with monochloramine. Environmental Science and Technology 40 (23), 7290e7297. Chen, Z., Valentine, R.L., 2007. Formation of N-nitrosodimethylamine (NDMA) from humic substances in natural water. Environmental Science and Technology 41 (17), 6059e6065. Chen, Z., Valentine, R.L., 2008. The influence of the pre-oxidation of natural organic matter on the formation of N-nitrosodimethylamine (NDMA). Environmental Science and Technology 42 (14), 5062e5067. Chen, K.C., Wang, Y.H., Chang, Y.H., 2009. Using catalytic ozonation and biofiltration to decrease the formation of disinfection by-products. Desalination 249, 929e935. Environmental Protection Agency, EPA, 2008. Emerging Contaminant- N-Nitrosodimethylamine (NDMA) Fact Sheet. Farre´, M., Keller, J., Holling, N., Poussade, Y., Gernjak, W., 2011. Occurrence of NDMA precursors in wastewater treatment plant effluent and their fate during UF-RO membrane treatment. Water Science and Technology 63, 605e612. Gallard, H., Von Gunten, U., 2002. Chlorination of natural organic matter: kinetics of chlorination and of THM formation. Water Research 36, 65e74. Greenberg, A., Clesceri, L., Deaton, A., 1992. Standard Methods for the Examination of Water and Wastewater. Water Environment Federation, Washington DC, USA. Hebert, A., Forestier, D., Lenes, D., Benanou, D., Jacob, S., Arfi, C., Lambolez, L., Levi, Y., 2010. Innovative method for prioritizing emerging disinfection by-products (DBPs) in drinking water on the basis of their potential impact on public health. Water Research 44, 3147e3165. Herzberg, M., Dosoretz, C.G., Tarre, S., Green, M., 2003. Patchy biofilm coverage can explain the potential advantage of BGAC reactors. Environmental Science & Technology 37, 4274e4280. Hua, G., Reckhow, D.A., Kim, J., 2006. Effect of bromide and iodide ions on the formation and speciation of disinfection byproducts during chlorination. Environmental Science and Technology 40, 3050e3056. Kemper, J.M., Walse, S.S., Mitch, W.A., 2010. Quaternary amines as nitrosamine precursors: a role for consumer products? Environmental Science and Technology 44, 1224e1231. Kumar, K., Margerum, D.W., 1987. Kinetics and mechanism of general-acid-assisted oxidation of bromide by hypochlorite and hypochlorous acid. Inorganic Chemistry 26, 2706e2711. Lee, C., Schmidt, C., Yoon, J., Von Gunten, U., 2007. Oxidation of Nnitrosodimethylamine (NDMA) precursors with ozone and chlorine dioxide: kinetics and effect on NDMA formation potential. Environmental Science and Technology 41, 2056e2063. Liang, L., Singer, P.C., 2003. Factors influencing the formation and relative distribution of haloacetic acids and trihalomethanes in drinking water. Environmental Science and Technology 37, 2920e2928. Mitch, W.A., Gerecke, A.C., Sedlak, D.L., 2003. A NNitrosodimethylamine (NDMA) precursor analysis for chlorination of water and wastewater. Water Research 37, 3733e3741. Mitch, W.A., Sedlak, D.L., 2004. Characterization and fate of NNitrosodimethylamine precursors in municipal wastewater treatment plants. Environmental Science and Technology 38, 1445e1454. Munch, J.W., 1995. Measurement of Purgeable Organic Compounds in Water by Capillary Column Gas Chromatography/Mass Spectrometry Cincinnati, OH. Munch, J.W., Bassett, M.V., 2004. Method Development for the Analysis of N-Nitrosodimethylamine and other NNitrosamines in Drinking Water at Low Nanogram/Liter Concentrations using Solid-Phase Extraction and Gas Chromatography with Chemical Ionization Tandem Mass Spectrometry Method 521. EPA/600/R-05/054.
5704
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 6 9 5 e5 7 0 4
Pehlivanoglu-Mantas, E., Sedlak, D.L., 2006. The fate of wastewater-derived NDMA precursors in the aquatic environment. Water Research 40, 1287e1293. Pehlivanoglu-Mantas, E., Sedlak, D.L., 2008. Measurement of dissolved organic nitrogen forms in wastewater effluents: concentrations, size distribution and NDMA formation potential. Water Research 42, 3890e3898. Pipe-Martin, C., Reungoat, J., Keller, J., 2010. Dissolved Organic Carbon Removal by Biological Treatment. Water Quality Research Australia, Adelaı¨de (Australia). Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F., Keller, J., 2010. Removal of micropollutants and reduction of biological activity in a full scale reclamation plant using ozonation and activated carbon filtration. Water Research 44, 625e637. Reungoat, J., Escher, B.I., Macova, M., Keller, J., 2011. Biofiltration of wastewater treatment plant effluent: effective removal of pharmaceuticals and personal care products and reduction of toxicity. Water Research 45, 2751e2762. Richardson, S.D., Plewa, M.J., Wagner, E.D., Schoeny, R., DeMarini, D.M., 2007. Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-products in drinking water: a review and roadmap for research. Mutation Research - Reviews in Mutation Research 636, 178e242. Schreiber, I.M., Mitch, W.A., 2006. Nitrosamine formation pathway revisited: the importance of chloramine speciation and dissolved oxygen. Environmental Science and Technology 40, 6007e6014. Shannon, M.A., Bohn, P.W., Elimelech, M., Georgiadis, J.G., Marin˜as, B.J., Mayes, A.M., 2008. Science and technology for water purification in the coming decades. Nature 452, 301e310. Shen, R., Andrews, S.A., 2010. Demonstration of 20 pharmaceuticals and personal care products (PPCPs) as nitrosamine precursors during chloramine disinfection. Water Research 45, 944e952. Simpson, D.R., 2008. Biofilm processes in biologically active carbon water purification. Water Research 42, 2839e2848.
Tricker, A.R., Pfundstein, B., Preussmann, R., 1994. Nitosable secondary amines: exogenous and endogenous exposure and nitrosation in vivo. ACS Symposium Series 553, 93e101. USEPA., 2003. Method 552.3. Determination of Haloacetic Acids and Dalapon in Drinking Water by Liquideliquid Microextraction, Derivatization, and Gas Chromatography with Electron Capture Detection Cincinnati, OH. van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water reclamation at South Caboolture, Queensland, Australia. Ozone: Science & Engineering 25, 107e120. Villanueva, C.M., Cantor, K.P., Grimalt, J.O., Malats, N., Silverman, D., Tardon, A., Garcia-Closas, R., Serra, C., Carrato, A., Castan˜o -Vinyals, G., Marcos, R., Rothman, N., Real, F.X., Dosemeci, M., Kogevinas, M., 2007. Bladder cancer and exposure to water disinfection by-products through ingestion, bathing, showering, and swimming in pools. American Journal of Epidemiology 165, 148e156. Westerhoff, P., Chao, P., Mash, H., 2004. Reactivity of natural organic matter with aqueous chlorine and bromine. Water Research 38, 1502e1513. Westerhoff, P., Mash, H., 2002. Dissolved organic nitrogen in drinking water supplies: a review. Journal of Water Supply: Research and Technology - AQUA 51 (8), 415e448. Westerhoff, P., Yoon, J., Snyder, S., Wert, E., 2008. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science & Technology 39, 6649e6663. Xie, Y.F., 2004. Disinfection by-Products in Drinking Water Formation, Analysis and Control. Lewis Publishers, Florida. Yoon, Y., Westerhoff, P., Snyder, S.A., Esparza, M., 2003. HPLCfluorescence detection and adsorption of bisphenol A, 17bestradiol, and 17a-ethynyl estradiol on powdered activated carbon. Water Research 37, 3530e3537. Zhang, Z., Wang, L., Shao, L., 2010. Study on Relationship between Characteristics of DOC and Removal Performance by BAC Filter 2010 4th International Conference on Bioinformatics and Biomedical Engineering, iCBBE 2010.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Continuous combined Fenton’s oxidation and biodegradation for the treatment of pentachlorophenol-contaminated water Julio A. Zimbron 1, Kenneth F. Reardon* Department of Chemical and Biological Engineering, 100 Glover Building, Colorado State University, Fort Collins, CO 80523-1370, USA
article info
abstract
Article history:
Pentachlorophenol (PCP) was studied as a model recalcitrant compound for a sequential
Received 5 July 2011
chemical oxidation and biodegradation treatment, in a continuous laboratory-scale system
Received in revised form
that combined a Fenton’s chemical reactor and a packed-bed bioreactor.
22 August 2011
PCP degradation and dechlorination were observed in the Fenton’s reactor at a resi-
Accepted 23 August 2011
dence time of 1.5 h, although no reduction of total organic carbon (TOC) was observed. Both
Available online 31 August 2011
PCP degradation and dechlorination were strongly dependent on the H2O2 dose to the chemical reactor. The PCP degradation intermediates tetrachlorohydroquinone and
Keywords:
dichloromaleic acid were identified in this reactor. Further treatment of the Fenton’s
Biodegradation
reactor effluent with a packed-bed bioreactor (operating at a residence time of 5.5 h)
Fenton’s reaction
resulted in partial biodegradation of PCP degradation intermediates and reduction in TOC,
Hydroxyl radical
although no further reduction of PCP or dechlorination was achieved in the bioreactor.
Kinetics model
Increased residence time in the bioreactor had no significant impact on degradation of
Pentachlorophenol
TOC. Recycle of the effluent from the bioreactor to the chemical reactor increased the TOC degradation, but not the extent of the PCP degradation or dechlorination. A mathematical model of the combined Fenton’s oxidation and biodegradation system supported the experimental results. While the model over-predicted the PCP and TOC degradation in the combined system, it adequately predicted the sensitivity of these parameters to different H2O2 doses and recycle rates. The model indicated that high recycle rates would improve TOC degradation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Pentachlorophenol is a common water contaminant, released into the environment from wood treating and biocide formulation operations. It often occurs in mixtures with other contaminants, constituting the most recalcitrant fraction of such mixtures (Graves and Joyce, 1994). PCP biodegradation has been reported both aerobically and anaerobically, although rates are too slow for practical treatment, owing to the need to induce specific enzymes (Hale
et al., 1994). Specialized cultures can achieve faster degradation rates (Puhakka and Melin, 1996), but might not be resilient when exposed to environmental disturbances (Scott and Ollis, 1995). Due to this reported recalcitrance of PCP and other compounds to biodegradation, there has been increased interest in alternative treatment technologies. Advanced oxidation processes (AOPs) rely on the generation of hydroxyl free radicals ($OH), a highly reactive chemical species (Venkatadri and Peters, 1993). Complete mineralization of
* Corresponding author. Tel.: þ1 970 491 6505; fax: þ1 970 491 7369. E-mail address: [email protected] (K.F. Reardon). 1 Present address: Department of Civil and Environmental Engineering, Colorado State University, Fort Collins, CO 80523-1320, USA. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.038
5706
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
List of symbols and abbreviations a
sBR sChRx mi mi,max $OH DCMA Fe(II) Fe(III) H2O2 ki kj mi,j
recycle rate, equal to the ratio of the flow rate of the effluent from the bioreactor that is recycled back to the chemical reactor with respect to the inlet flow rate to the combined system residence time (time units) in the bioreactor residence time (time units) in the chemical (Fenton’s) reactor microbial growth rate (1/h) Monod’s maximum microbial growth rate (1/h) hydroxyl free radical dichloromaleic acid ferrous iron ferric iron hydrogen peroxide Monod’s half saturation microbial degradation constant for compound i (mol/L) second-order kinetic constant for the Fenton’s system reaction j (L/mol$s) stoichiometric coefficient for reactant i in Fenton’s reaction j PCP pentachlorophenol
recalcitrant compounds can be achieved using these technologies (Scott and Ollis, 1995), although long treatment times and strong doses of the oxidizing reagents are required (Esplugas et al., 2004; Marco et al., 1997; Pera-Titus et al., 2004). Often, the chemically-oxidized intermediates are less recalcitrant than the parent compound (Comninellis et al., 2008). Since biological processes are typically less expensive than chemical processes, addition of a biodegradation stage can improve the economy of the overall process, particularly for low concentration wastewater (Pera-Titus et al., 2004). Examples of this combined treatment have been presented (Pera-Titus et al., 2004; Marco et al., 1997; Scott and Ollis, 1995), but the mechanistic information required for the combined reactor system analysis and design remains scarce (Comninellis et al., 2008; Esplugas et al., 2004; Mantzavinos and Psillakis, 2004). As a result of the complexity of the processes and the lack of mechanistic data, process integration has typically been experimentally evaluated on a case-bycase basis. For chemical oxidation alone, a widely accepted idea is that kinetics modeling requires a complete reaction pathway for the organic substrate (Duesterberg and Waite, 2006; Kang et al., 2002; Rivas et al., 2001). Although simplified kinetics models have provided significant insight into reactor design (Esplugas et al., 2004), they typically are based on first-order reactions. Such simplified analysis has the limitation of neglecting competitive effects of the chemically oxidized by-products on the target parent compound, an essential feature of AOPs that in practice precludes complete treatment of target contaminants. The purpose of this work was to study the combination of Fenton’s oxidation (the combination of H2O2 and ferrous iron) and biodegradation of PCP-contaminated water. Experiments were conducted in a continuous stirred-tank reactor (CSTR) in which Fenton’s degradation of PCP occurred. The effluent from this reactor was continuously
Ri,j ratei,Bio ratei,Ch Si,BioRx SiChRx Si,FbioRx
Si,FChRx TCHQ TOC XT YX/Si
rate law for Fenton’s reaction j (in which i is a reaction member) rate of reaction of chemical species i under biodegradation treatment rate of reaction of chemical species i under the Fenton’s reaction system concentration (mol/L) of the chemical species i in the bioreactor concentration (mol/L) of the chemical species i in the chemical reactor concentration (mol/L) of the chemical species i in the feed to the bioreactor (which in the combined model was equal to the concentration of this species i n the chemical reactor) concentration (mol/L) of the chemical species i in the feed to the chemical reactor tetrachlorohydroquinone total organic carbon packed bed biomass content (mg) yield of biomass to carbon substrate consumption (as TOC) in mg of biomass/mol of total organic carbon
treated in a second reactor system that included a packedbed bioreactor to degrade the PCP intermediates. The effects of H2O2 dose, residence time in the bioreactor, and recycle of the bioreactor effluent back to the chemical reactor on the combined system performance were studied. A mathematical model was developed for the combined system, based on a kinetics model simplified using a lumping approach, for both chemical oxidation and biodegradation. In this approach, a non-PCP fraction was defined to account for $OH scavenging effects of the by-products on the chemical oxidation of PCP, and as the biodegradable fraction on the biodegradation process. The model was used to test the consistency of the experimental data and provide insights into reactor design and operating strategies (i.e., the recycle of effluent from the second stage bioreactor back to the chemical reactor).
2.
Kinetics modeling
2.1.
Fenton’s oxidation kinetics model
A mathematical model was developed previously for the chemical oxidation kinetics and validated using data from batch experiments (Zimbron and Reardon, 2009). This Fenton’s kinetics model included 11 reactions between inorganic species involved in the Fenton’s system (i.e., H2O2, Fe(II)/ Fe(III) and initiation and propagation reactions involving radicals $OH, $O2H/$O 2 ). The mass balances were rewritten for the CSTR used in the combined Fenton’s-biodegradation system. The termination reaction for PCP with $OH (calculated using the competitive kinetics method) was included with a value of 4.4 1009 L/mol$s. The scavenging effects of the PCP by-products on the degradation of PCP (and the rest of the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
reaction system) were included as a lumped reaction with a kinetic constant (Zimbron and Reardon, 2009). This lumping approach has been successfully applied to estimate the degradation of contaminants by different AOPs, including photo-Fenton’s reactions (Zepp and Scholtzhauer, 1979), ultraviolet radiation (Haag and Hoigne, 1985), and soil slurries (Huling et al., 1998). Using the above mentioned set of reactions, the reaction rate for each reactant or product i in the Fenton’s reaction system is assumed to follow second-order kinetics, generalized as: ratei;Ch ¼
n X
mi;j Ri;j ¼
j¼1
n X
mi;j kj Si;jChRx Sh;jChRx
(1)
j¼1
in which Si,jChRx indicates the molar concentration of chemical species, i. For each specific Fenton’s reaction j, Ri,j and mi,j are the rate law and the stoichiometric coefficient of reactant i (negative for reactants, positive for products), respectively, kj is the reaction rate constant, and the subscript h indicates other chemical species involved in that particular reaction rate (e.g., $OH). The resulting mass balance for the chemical reactor (including a recirculation stream from the bioreactor to the chemical reactor) is: Si;ChRx ¼
Si;FChRx þ ratei;Ch sChRx þ aSi;BioRx 1þa
(2)
in which Si is the molar concentration of the reactive species i, the subscripts ChRx and FChRx indicate the chemical reactor and feed stream to this reactor, ratei is the rate of reaction of the chemical species i (mol/L$s), sChRx is the residence time (s) in the chemical reactor, and a is the ratio of recycle flow rate from the bioreactor to the chemical reactor to the total flow rate to the system.
2.2.
Biodegradation kinetics model
Kinetics data from the observed biodegradation of PCP byproducts was incorporated into a biodegradation kinetics model. The mass balances for PCP and TOC were calculated, based on the non-PCP TOC fraction (the difference of PCP concentrations in the feed and the chemical reactor, multiplied by 6, the PCP stoichiometric carbon molecular content). Monod-type biodegradation kinetic parameters were obtained based on the non-PCP TOC as substrate (at different feed concentrations), as supported by experimental data. The Monod model for the rate of biodegradation of substrate i, ratei,Bio, depends on the total concentration of biomass contained in the bioreactor (XT, assumed constant for an immobilized biomass reactor over the period of the experiment), the yield of biomass to substrate i (YX/Si), and the molar concentration of substrate (Si,BioRx): ratei;Bio ¼ mi
mi;max Si;BioRx XT XT ¼ YX=Si ki þ Si;BioRx YX=Si
(3)
The combined model assumed that chemical oxidation and biodegradation were mutually exclusive. Furthermore, upon recycle, biomass and excess nutrients incorporated into the
5707
bioreactor stream were assumed not to affect the degradation of PCP by Fenton’s reaction. These assumptions were verified in batch experiments (data not shown). The mass balance for the bioreactor is given by: Si;BioRx ¼ Si;ChRx þ
ratei;Bio sBioRx 1þa
(4)
in which sBioRx indicates the residence time (in time units) in the bioreactor. The combined reactor model results in a system of coupled algebraic equations that was solved using Engineering Equation Solver (F-chart Software). This model represents a large simplification of the combined system, for both the chemical oxidation process and the biodegradation process. However, the mass balance approach to model development has the potential to provide insights into the study of the process and chemical reactor design. The need for such simplified, yet mechanistic models for process integration has been highlighted by others (Esplugas et al., 2004; Mantzavinos and Psillakis, 2004).
3.
Material and methods
3.1.
Chemicals
Pentachlorophenol (99%), the extraction surrogate (dibromophenol, 95%) and the internal standard (dibromobenzene, 98%) were purchased from Sigma. FeSO4$7H2O (99%þ) was purchased from Baxter. H2O2 (non-stabilized, 31.4%), sodium chloride (99.9%) and sulfuric acid (conc.) were all purchased from Fisher. Solvents (chloroform and ethyl acetate) for the analysis of PCP and organic PCP byproducts were pesticide-grade (Fisher). Deionized and dechlorinated water was used for PCP solution and plate media preparation, while water used for preparation of all other reactants was Reagent Type I Grade water (Nanopure, 18 mU conductivity).
3.2.
Continuous system apparatus
The continuous combined Fenton’s and biodegradation system consisted of two CSTRs in series (2 L Bioflo C-30, New Brunswick Scientific) (Fig. 1). In the first reactor (with a liquid volume of 250 mL and residence time of 1.5 h), Fenton’s treatment of the PCP-contaminated water was performed at pH 3.5, within the reported optimum range for Fenton’s treatment (for example Venkatadri and Peters, 1993; Zepp and Scholtzhauer, 1979). A Markson controller with a combination pH electrode and H2SO4 0.1 N were used to control pH at 3.5. Ferrous iron and H2O2 were dosed to achieve specific concentrations as described in the next section. The chemical reactor and reactant feeds were covered with aluminum foil. The effluent from the Fenton’s reactor was fed to the bioreactor at pH 7, adding supplementary nutrient solution (phosphates, trace minerals and nitrogen; Section 3.3). pH control was achieved with NaOH 0.2 N solution and an Omega pH/ORP controller with a combination pH electrode. The bioreactor residence time was either 5.5 or 10 h, achieved by adjusting the bioreactor liquid volume (either 750 or 1500 mL,
5708
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
The supplementary nutrient medium was the PAS mineral salts medium (see Section 3.5), modified by replacing calcium chloride with calcium sulfate to eliminate background chloride concentrations. Nitrogen in the supplementary nutrient medium was stoichiometric to carbon in the PCP solution at a 1:5 N:C molar ratio (Shuler and Kargi, 1992). This nutrient solution was fed to the bioreactor at a rate of 85 mL/day (approximately 2.5% of the total flow rate to the system).
3.4.
Fig. 1 e Continuous combined reactor system for Fenton’s oxidation and biodegradation of PCP-contaminated water.
respectively). In some reactor configurations, the effluent from the bioreactor was recycled to the chemical reactor. The low organic carbon concentration in the feed stream (due to the low PCP solubility) resulted in oligotrophic (low nutrient) conditions, under which suspended cell cultures might be washed out of the reactor. Thus, a packed-bed column (glass, 3.8 cm OD, 14.5 cm long) filled with 38.8 g of Celite biocatalyst carrier R-635 (Manville) was provided, to which the contents of the main bioreactor vessel were recirculated (in an upflow configuration) at a rate 20 times higher than the feed to the stirred tank to ensure that the liquid contents were well mixed. The bioreactor was initially inoculated with 3 mL of Fort Collins wastewater treatment plant aerobic activated sludge. The reactor tubing consisted of chemically resistant PTFE when possible, or Norprene (Masterflex). All other parts in the system were stainless steel or glass. No PCP sorption or degradation was detected on these materials or the Celite biomass support through batch 24 h sorption experiments.
3.3.
Reactants to the continuous system
The PCP solution was prepared with deionized and dechlorinated water adjusted to pH 10 with 0.8 mL of 0.1 M NaOH to facilitate PCP dissolution. PCP was added to the reported solubility at neutral pH (14 ppm). The flow rate of this solution to the reactor system was 3.5 L/day. The flow rate of the H2O2 feed solution reactant was 85 mL/ day, equivalent to 2.5% of the total flow rate to the combined system.At this flow rate, the H2O2 feed of 880 mM and 1800 mM solutions yielded the low and high doses of [H2O2] ¼ 220 and [H2O2] ¼ 370 mM to the chemical reactor, respectively. The iron feed solution was prepared by dissolving 2.3 g of FeSO4 $7H2O in water, with 6.4 mL of concentrated H2SO4 to avoid Fe(II) oxidation, and diluted to 1.0 L to a final concentration of [Fe(II)] ¼ 8300 mM. The flow rate of this reactant was 85 mL/day, resulting in an actual dose of [Fe(II)] ¼ 200 mM to the chemical reactor.
Analytical methods
Concentrations of PCP and the identified PCP degradation byproducts tetrachlorohydroquinone and dichloromaleic acid were determined with a HP 5890 Series II GC-MS, after solvent extraction. Ferrous iron, chloride ion, and H2O2 analyses were done spectrophotometrically, using the 1,10-phenanthroline, thiocyanate (Hach Method 20635-00), and titanium sulfate methods, respectively (APHA et al., 1980). Additional details about these analyses are available from a previous report (Zimbron and Reardon, 2009). Dissolved oxygen was measured with a temperaturecompensated oxygen electrode (Phoenix Electrode), and a Cole Parmer 01971-00 analyzer. Purgeable total organic carbon (TOC) was analyzed with a Dohrmann DC-80 TOC Analyzer (detection limit lower than 1 mg/L).
3.5.
Biomass analysis
A protein quantitative assay was used to measure biomass indirectly, since iron precipitates precluded direct estimation by optical density (600 nm). Samples were collected and frozen until analysis. Samples were centrifuged in 250-mL Teflon bottles for 25 min at 13200g. The biomass was resuspended with phosphate 0.1 M pH 7 buffer and centrifuged again twice, then reconstituted to 3.0e10.0 mL. The biomass concentrate was lysed with a sonicator (UPXL, Heath Systems) for 20 min on ice and analyzed for protein using the micro-BCA assay (Pierce), comparing absorbance (562 nm) to that of bovine serum albumin standards. The ability of the bioreactor effluent biomass to degrade PCP was tested by plating on an agar medium with PCP as the only carbon source (PCP medium) at a concentration of 0.5 mg/ L (to prevent inhibitory effects).This medium included 7.5 g of Bacto-Agar with 38.5 mL PAS concentrate (13.8 g NH4Cl, 10.97 g KH2PO4 and 28.39 g K2HPO4 diluted to 500 mL) and 5 mL of PAS 100 salts solution (0.15 g CaCl2 $2H2O, 0. 5 g FeSO4 $7H2O, 2.5 g MnSO4 $H2O, 9.75 g MgSO4, 2 drops H2SO4 diluted to 500 mL), diluted to 500 mL (Bedard et al., 1986) and autoclaved. The ability of the biomass to grow on a readily available carbon source was tested by plating on a non-selective medium, Trypticase Soy Broth (TSB), diluted to 1:4 to simulate an oligotrophic (low nutrient) environment.
3.6.
Experiments
In addition to the basic sequential chemical oxidation and biodegradation reactor configuration, the effect of recycling part of the bioreactor effluent back to the chemical reactor was evaluated, at the low residence time of 5.5 h in the bioreactor, for both oxidation strengths. Two recycle rates
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
(0.2 and 0.4) were used for this purpose. This recycle rate (a) is defined as the ratio of the bioreactor effluent flow rate sent back to the chemical reactor (Fig. 1) to the total inlet flow to the system. After a new operating condition was established, the system ran for 48 h to achieve steady state before sampling. This represents 32 and 9 residence times for the chemical reactor and the bioreactor, respectively. A set of samples at a H2O2 dose of 220 mM taken at 36 h yielded similar PCP and TOC results to those at 48 h, suggesting that 36 h was sufficient for the system to achieve steady state.
4.
Results and discussion
4.1.
Experimental results
4.1.1.
Hydrogen peroxide consumption in side reactions
Control experiments without Fe(II) in the continuous Fenton’s reactor showed that H2O2 consumption occurred in that reactor in the absence of Fe(II), and that neither significant PCP degradation nor dechlorination resulted. The measured H2O2 concentrations were 29% and 30% lower than the mass balance around the chemical reactor at H2O2 doses of 178 and 406 mM, respectively, despite the lack of measured losses in the feed flask. These losses likely occurred at iron precipitatecoated reactor surfaces (due to the long term operation). The reactivities of such surfaces toward H2O2 have been documented (Watts et al., 1997). For these reasons, all reported H2O2 doses tested in this work were corrected to 30% lower than the actual dose, as estimates of the H2O2 available for Fenton’s mechanisms (i.e., resulting in PCP degradation). These are referred hereafter as the effective hydrogen peroxide doses.
4.1.2.
5709
1992). Biomass and protein measurements in the column and the bioreactor effluent confirmed that more than 99% of the biomass present in the bioreactor was located in the packed bed, where most of the microbial activity occurred.
4.1.3.
Effect of hydrogen peroxide dose
PCP degradation achieved in the chemical reactor (with a residence time of 1.5 h without recycle) and the resulting chloride production under a constant dose of Fe(II) of 200 mM and variable doses of hydrogen peroxide is shown in Fig. 2. Also shown are the model-predicted Fenton’s-driven degradation of PCP and dechlorination in the chemical reactor at these conditions. The model over-predicted PCP degradation in the continuous system, although it achieved considerably more accurate predictions for batch experiments data (Zimbron and Reardon, 2009). The lowest experimental PCP dimensionless concentrations (approximately 45%) were higher than the lowest model-predicted PCP dimensionless concentration (25%, corresponding to H2O2 doses of 200 mM or higher). The lack of agreement between experimental and modelpredicted PCP degradation is attributed to side reactions of free radicals at reactive surfaces in the continuous system. These side reactions consumed H2O2, but as shown by the noFe(II) control experiment with a high dose of H2O2, H2O2scavenging reactions did not yield significant PCP degradation. Detailed evaluation of the nature of the interactions of active surfaces with Fenton’s reactive species (i.e., free radicals) was beyond the scope of this work. Both PCP degradation and Cl release depended strongly on the H2O2 dose, up to a concentration of 200 mM (Fig. 2). The mean observed-to-theoretical chloride concentration ratio
Biomass characterization
The system operated for several months without recycle at constant conditions (Fe(II) dose of 200 mM and an effective H2O2 of 260 mM) to establish a stable microbial community. At these conditions, the PCP degradation (with respect to the feed concentration) achieved in the chemical reactor was 65% and the TOC degradation achieved in the bioreactor was 13%. A bioreactor effluent aliquot was analyzed for protein, yielding a concentration of 0.41 mg/L (c.v. ¼ 11%). This effluent contained 1.3 1005 CFU, obtained by plating on ¼-strength TSB medium (after 48 h of incubation). Upon plating the bioreactor effluent on PCP medium and incubating for one week, no colonies developed, although growth in the non-selective TSB medium occurred at 2 days, showing the lack of PCP degraders within this microbial population. Two samples of five pellets each were taken from the packed bed and analyzed for dry and organic matter (by drying at 35 C and incinerating). The estimated organic content of the pellets was 0.41% (0.13%), or 160 mg of biomass (dry weight) for the entire packed column. Three pellets from the column were crushed, sonicated, and analyzed for protein. The organic-free dry weight was estimated by incineration. The resulting approximate protein content of biomass (organic matter) in the bioreactor was 69%, within the reported range of 40%e70% (Shuler and Kargi,
Fig. 2 e Pentachlorophenol and chloride concentrations achieved at different effective H2O2 doses in the continuous chemical reactor. [Fe(II)] [ 200 mM. Error bars represent the standard deviation of duplicate samples. Filled (C) and open (B) circles represent dimensionless concentrations of PCP in the chemical reactor (with respect to the feed concentration) and measured Clconcentrations, respectively. The open triangle symbol is the result of a control experiment with a high dose of H2O2 and no Fe(II). The solid and dotted line represent the model-predicted PCP degradation and chloride production based on complete PCP dechlorination in the chemical reactor, respectively.
5710
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
Fig. 3 e Dimensionless concentrations of PCP (bottom) and TOC (top) at different hydrogen peroxide doses. Bars in white, light grey, and dark grey represent concentrations in the feed, chemical reactor, and bioreactor, respectively. Residence times in the chemical and bioreactor were 1.5 and 5.5 h, respectively. Error bars represent the standard deviation of duplicate samples.
(assuming complete dechlorination of the degraded PCP) was 84% (9%). Previous work on electrochemical oxidation of PCP in soil suspension reported about 80% dechlorination (Hanna et al., 2005), while studies on ozonation and UV/H2O2 have reported lower partial dechlorination (in the order of 30e60%) (Hirvonen et al., 2000). The PCP and TOC concentrations in the feed, chemical reactor, and bioreactor are presented in Fig. 3. The average TOC concentration of the saturated PCP feed was 4.7 mg/L (0.43), slightly higher than the theoretical TOC concentration of saturated PCP solution (4.0 ppm) (probably due to calibration error). Triplicate feed water blanks yielded values of 0.208 (0.07) mg/L, showing that the feed water contained very low levels of extraneous carbon. The chloride (at the same three points along the experimental apparatus) and protein concentrations in the bioreactor effluent are shown in Fig. 4.
In agreement with the plating test results, Figs. 3 and 4 indicate that microbial growth (based on protein measurements) was supported by the partial degradation of the nonPCP TOC as a carbon source. In addition to the lack of PCP biodegradation (beyond the level achieved by chemical oxidation), no further dechlorination occurred at the bioreactor. As indicated by the TOC reduction observed in the bioreactor, the bioreactor microbial population had a preference for non-chlorinated intermediates over chlorinated ones, consistent with previous findings for PCP treated by electroFenton’s (Hanna et al., 2005). No PCP losses in the chemical reactor (or the bioreactor) occurred in control experiments without H2O2. Cell wash-out due to the lack of nutrients at these conditions might have caused the non-zero protein concentration in the bioreactor effluent.
Fig. 4 e Concentrations of chloride (bottom) and protein (top) at different hydrogen peroxide doses. Bars in white, light grey, and dark grey represent concentrations in the feed, chemical reactor, and bioreactor, respectively. Residence times in the chemical and bioreactor were 1.5 and 5.5 h, respectively. Error bars represent the standard deviation of duplicate samples.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
Tetrachlorohydroquinone (TCHQ) and dichloromaleic acid (DCMA) were identified as PCP intermediates of Fenton’s oxidation in the continuous system, consistent with a previous batch reactor study (Zimbron and Reardon, 2009) and in agreement with previous reports of AOP treatment of PCP (Benitez et al., 2003; Hong and Zeng, 2002; Mills and Hoffman, 1993; Wong and Crosby, 1981; Zhao et al., 2006). DCMA was observed as a PCP by-product (and that of TCHQ and tetrachlorocatechol) under treatment with different AOPs (Hanna et al., 2005; Hong and Zeng, 2002; Sen Gupta et al., 2002; Shen et al., 2009; Wong and Crosby, 1981). The concentrations of TCHQ and DCMA in the chemical reactor and bioreactor are shown in Fig. 5. Attempts to identify other PCP intermediates were not successful. Owing to partial degradation of PCP in the Fenton’s system, the non-PCP TOC fraction was less than 55% of the PCPsaturated solution equivalent of 330 mM TOC. At a residence time of 5.5 h in the bioreactor, the maximum TOC reductions were about 14%. Further increases of the retention time in the bioreactor (with reactor conditions in the chemical reactor held constant) at the same effective H2O2 doses of 0, 150 mM, and 260 mM did not significantly increase TOC degradation (nor protein production) (95% confidence level, data not shown). Low biodegradation rates are typically observed at high substrate or at very low biomass concentrations (conditions that lead to zero-order biodegradation rates). These potential causes seem unlikely in this system, due to the low soluble TOC (limited by PCP solubility) and the large amount of biomass present in the bioreactor column. A better explanation is that a simple Monod-type biodegradation model might not adequately describe the complex mixture kinetics of the Fenton’s-treated PCP. The different biodegradability of the observed intermediates (Fig. 5) supports this. The observed lack of dechlorination, and reported toxicity of highly chlorinated compounds (such as PCP) (Pera-Titus et al., 2004), might explain the observed lack of sensitivity to extended residence time and the low extent of biodegradation of non-PCP TOC.
5711
Analysis of the biomass in the effluent indicated a yield of 7.5 g of biomass (dry weight)/mol TOC consumed. This measured yield was only 30% of the theoretical maximum of 25.5 g biomass dry weight/mol TOC, obtained assuming a theoretical biomass formula of CH2N0.25O0.5 (Shuler and Kargi, 1992). The remainder (70%) of the reduction of TOC in the bioreactor is an order of magnitude estimate of the mineralization extent to CO2 (rather than incorporation into biomass). The TCHQ yields upon PCP degradation in the continuous chemical reactor were 3.0 mol%, at both H2O2 doses (150 mM and 260 mM). The DCMA yields measured during continuous operation of the chemical reactor were 3.8 and 3.3 mol% at low and high H2O2 dose, respectively. These yields for TCHQ and DCMA were consistent with batch experiments results (Zimbron and Reardon, 2009).
4.1.4.
Effect of recycle
The effect of recycling the effluent from the bioreactor back to the chemical reactor for further treatment was tested at a residence time of 1.5 h in the chemical reactor and 5.5 h in the bioreactor. For this, 20% and 40% of the total flow rate to the system was recycled from the bioreactor back to the chemical reactor (a ¼ 0.2 and a ¼ 0.4, respectively). These experiments were conducted at both low (150 mM) and high (260 mM) effective H2O2 doses. Two-way ANOVA indicated that PCP and chloride concentrations in the effluent were not affected by the recycle level (20% or 40%), compared with no recycle (confidence level ¼ 95%). At the low hydrogen peroxide dose there were no significant effects of recycle (at both recycle levels tested) on TOC biodegradation. In contrast, the recycle rate effects (for both recycle ratios of 0.2 and 0.4) on the TOC biodegradation were significant at the high H2O2 dose (Fig. 6). At this high H2O2 dose, recycle had the effect of increasing the TOC reduction that occurred at the biodegradation stage with respect to no recycle. This improved TOC degradation upon recycling was confirmed by protein analysis that indicated increased biomass production (data not shown).
Fig. 5 e Concentration of two observed PCP intermediates in the combined system. Tetrachlorohydroquinone (TCHQ) concentrations (bottom) and dichloromaleic acid (DCMA) concentrations (top) at different hydrogen peroxide doses. Bars in white, light grey, and dark grey represent concentrations in the feed (not present), chemical reactor, and bioreactor, respectively. Residence times in the chemical and bioreactor were 1.5 and 5.5 h, respectively. Error bars represent the standard deviation of duplicate samples.
5712
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
Fig. 6 e Dimensionless concentrations of PCP (bottom) and TOC (top) at three different recycle ratios (0, 0.2, and 0.4) and a high hydrogen peroxide effective dose of 260 mM. Bars in white, light grey, and dark grey represent concentrations in the feed, chemical reactor, and bioreactor, respectively. Residence times in the chemical and bioreactor were 1.5 and 5.5 h, respectively. Error bars represent the standard deviation of duplicate samples (except for feed concentrations, for which error bars represent the standard deviation of 10 measurements).
4.2.
Kinetics modeling results
The biodegradation data presented in Figs. 3 and 6, in addition to data obtained at the longer bioreactor residence time (11 h, for a total of 6 data points) were used to estimate the Monod biodegradation kinetics of the non-PCP TOC. The yield coefficient was estimated as 5.2 (2.2) g/L of protein produced per mol/L of TOC consumed, based on the measured concentrations of protein in the effluent from the bioreactor. The linearized Monod equation (Shuler and Kargi, 1992) was used, resulting in the Monod values of ks ¼ 250 mM TOC and mmax ¼ 3.1 103 h1 (with a correlation coefficient of 0.74). The model-predicted TOC and PCP concentrations at different recycle ratios (0, 0.4, and 2.0) were estimated for two different scenarios:
back to the chemical reactor (in agreement with the experimental results). Model-predicted PCP and TOC dimensionless concentrations at a variable H2O2 dose, [Fe(II)] ¼ 200 mM, recycle ratios of 0.4 and 2.0, and constant residence time in each reactor (1.5 h in the chemical reactor and 5.5 h in the bioreactor) are shown in Fig. 8. At the three tested recycle rates (0, 0.4, and 2), the PCP degradation was not sensitive to additional H2O2 doses above 200 mM. Recycle rates of 0.4 and 2.0 achieve small increases in the degradation of PCP in the combined system. The observed lack of sensitivity of PCP degradation achieved by the experimental combined system at recycle rates of 0.2 and 0.4 can be explained in terms of the lack of sensitivity of the PCP and chloride analysis to detect the expected small changes. Model-predicted TOC and PCP degradation in the combined system increased with higher H2O2 doses (at constant
a) Variable biodegradation residence time, with constant residence time in the chemical reactor (1.5 h) and fixed H2O2 and ferrous iron doses (300 mM and 200 mM, respectively). b) Variable H2O2 dose, with constant residence times in both chemical and biodegradation reactors (1.5 h and 5.5, respectively), and Fe(II) ¼ 200 mM). At constant oxidizer dose and variable biodegradation residence time, the model-predicted PCP dimensionless concentrations (Fig. 7) were limited by the degradation of PCP achieved in the chemical reactor, because Fenton’s reaction was the only mechanism to degrade PCP in the combined system. Increases in recycle rate (from 0 to 0.4 and 2) achieved only small increases in the degradation of PCP in the chemical reactor. This is in agreement with the observed lack of sensitivity of PCP degradation in the experimental combined system at recycle rates of 0.2 and 0.4. In contrast, the modelpredicted degradation of TOC under the same conditions was more sensitive to the recycle of the bioreactor effluent
Fig. 7 e Model-predicted dimensionless degradation of PCP and TOC as a function of residence time in the bioreactor. The residence time in the chemical reactor was 1.5 h [H2O2] [ 300 mM and [Fe(II)] [ 200 mM. Thicker lines represent model solutions at lower recycle rates.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
Fig. 8 e Model-predicted dimensionless degradation of TOC and PCP in the combined system at variable hydrogen peroxide dose. The PCP degradation scale (right) is reversed to better separate the data. Residence time in the chemical reactor was 1.5 h, and 5.5 h in the bioreactor. [Fe(II)] [ 200 mM. Thicker lines represent model solutions at lower recycle rates.
residence time), as shown in Fig. 8. However, the magnitude of the increases was larger for TOC than for PCP degradation. This higher sensitivity of the model-predicted TOC degradation over that of PCP degradation can be explained in terms of the biodegradation kinetics constants for TOC and the scavenging effects of the PCP by-products on the chemical oxidation of PCP. The two limiting cases for the Monod-type kinetics are zero- and first-order kinetics, which occur under high and low substrate concentrations (compared to ki), respectively. The TOC degradation would be sensitive to increases in PCP degradation achieved in the chemical reactor only under a regime of first-order degradation kinetics (i.e., at low substrate concentrations) as determined by the magnitude of the ratio of the Monod’s constants mi,max and ki (which determines the value of the first-order constant).
5.
Conclusions
The combined system achieved both PCP and TOC degradation. All of the PCP degradation (which was correlated with dechlorination) occurred in the chemical reactor, while all of the TOC degradation occurred at the bioreactor. TOC biodegradation only occurred upon Fenton’s oxidation of PCP in the chemical reactor. Plating tests and bioreactor performance indicated that the microbial population could not grow on PCP as a single carbon source. During bioreactor operation, this population partially mineralized the non-PCP fraction of the TOC without further significant dechlorination of the Fenton’s reactor effluent. The observed Fenton’s oxidation intermediates differed in their biodegradability: TCHQ was completely biodegraded, while DCMA was only partially biodegraded. Increased bioreactor residence time did not yield higher TOC biodegradation extents, possibly because of (a) a limited amount of biodegradable intermediates (with respect to the non-PCP TOC) and/or (b) the partial biodegradability of these intermediates.
5713
Recycling of the waste from the bioreactor back to the chemical reactor proved useful in achieving additional TOC degradation, but achieved only marginal additional PCP degradation. The model developed in this work was useful to explain some of the key experimental findings: the sensitivity of PCP degradation to H2O2 doses up to 200 mM, the lack of the system sensitivity to increases in bioreactor residence time, and the increased sensitivity of TOC degradation to recycling effects, compared to PCP degradation. This supports the idea that within the model limitations, the lumped chemical approach presented in this work is a useful tool for design and study of these combined treatment systems.
Acknowledgments This research was possible by grants from Consejo Nacional de Ciencia y Tecnologı´a (Mexico), grant number 5P42ES0594905 from the National Institute for Environmental Health Sciences, the Colorado Institute for Research in Biotechnology, and the Colorado Section of the American Water Research Association.
references
APHA, AWWA and WPCF, 1980. Standard Methods for the Examination of Water and Wastewater. Bedard, D., Unterman, R., Bopp, L., Brenna, M., Haberl, M., Johnson, C., 1986. Rapid assay for screening and characterizing microorganisms for the ability to degrade polychlorinated biphenyls. Applied and Environmental Microbiology 51 (4), 761e768. Benitez, F.J., Acero, J.L., Real, F.J., Garcia, J., 2003. Kinetics of photodegradatio and ozonaton of pentachlorophenol. Chemosphere 51, 651e662. Comninellis, C., Kapalka, A., Malato, S., Parsons, S.A., Poulios, I., Mantzavinos, D., 2008. Perspective: advanced oxidation processes for water treatment: advances and trends for R&D. Journal of Chemical Technology and Biotechnology 83, 769e776. Duesterberg, C., Waite, D., 2006. Process optimization of fenton oxidation using kinetic modeling. Environmental Science and Technology 40, 4189e4195. Esplugas, S., Contreras, S., Ollis, D., 2004. Engineering aspects of the integration of chemical and biological oxidation: simple mechanistic models for the oxidation treatment. Journal of Environmental Engineering 130 (9), 967e974. Graves, J.W., Joyce, T., 1994. A critical review of the ability of biological treatment systems to remove chlorinated organics discharged by the paper industry. Water S.A 20 (2), 155e159. Haag, W., Hoigne, J., 1985. Photo-sensitized oxidation in natural water via. OH radicals. Chemosphere 14 (11/12), 1659e1671. Hale, D., Reineke, W., Wiegel, J., 1994. In: Chaudry, R. (Ed.), Biological Degradation and Bioremediation of Toxic Chemicals. Dioscorides Press, Portland, OR, USA, pp. 74e91. Hanna, K., Chiron, S., Oturan, M., 2005. Coupling enhanced water solubilization with cyclodextrin to indirect electrochemical treatment for pentachlorophenol contaminated soil remediation. Water Research 39, 2763e2773. Hirvonen, A., Trapido, M., Hentunen, J., Tarhanen, J., 2000. Formation of hydroxylated and dimeric intermediates during oxidation of chlorinated phenols in aqueous solution. Chemosphere 41, 1211e1218.
5714
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 0 5 e5 7 1 4
Hong, A., Zeng, Y., 2002. Degradation of pentachlorophenol by ozonation and biodegradability of intermediates. Water Research 36, 4243e4254. Huling, S., Arnold, R., Sierka, R., Miller, M., 1998. Measurement of hydroxyl radical activity in a soil slurry using the spin trap alpha-(4-pyridyl-1-oxide)-N-tert-butylnitrone. Environmental Science and Technology 32, 3436e3441. Kang, N., Lee, D.S., Yoon, J., 2002. Kinetic modeling of Fenton oxidation of phenol and monochlorophenols. Chemosphere 47, 915e924. Mantzavinos, D., Psillakis, E., 2004. Review: enhancement of biodegradability of industrial wastewaters by chemical oxidation pre-treatment. J Chem Technol Biotechnol 79, 431e454. Marco, A., Esplugas, S., Saum, G., 1997. How and Why combine chemical and biological processes for wastewater treatment. Water Science and Technology 35 (4), 321e327. Mills, G., Hoffman, M., 1993. Photocatalytic degradation of pentachlorophenol on TiO2 Particles: identification of intermediates and mechanism of reaction. Environmental Science and Technology 27, 1681e1689. Pera-Titus, M., Garcia-Molina, V., Ban˜os, M., Gime´nez, J., Esplugas, S., 2004. Degradation of chlorophenols by means of advanced oxidation processes: a general review. Applied Catalysis B: Environmental 47, 219e256. Puhakka, J., Melin, E., 1996. In: Crawford, R., Crawford, D. (Eds.), Bioremediation: Principles and Applications. Cambridge University Press, Cambridge, MA, USA, pp. 254e299. Rivas, F., Beltra´n, F.J., Frades, J., Buxeda, P., 2001. Oxidation of p-hydroxybenzoic acid by Fenton’s reagent. Water Research 35 (2), 387e396.
Scott, J., Ollis, D., 1995. Integration of chemical and biological oxidation processes for water treatment: review and recommendations. Environmental Progress 14, 88e103. Sen Gupta, S., Stadler, M., Noser, C., Ghosh, A., Steinhoff, B., Lenoir, D., Howitz, C., Schramm, K., Collins, T., 2002. Rapid total destruction of chlorophenols by activated hydrogen peroxide. Science 296, 326e328. Shen, X., Zhu, L., Liu, G., Tang, H., Liub, S., Lib, W., 2009. Photocatalytic removal of pentachlorophenol by means of an enzyme-like molecular imprinted photocatalyst and inhibition of the generation of highly toxic intermediates. New Journal of Chemistry 33, 2278e2285. Shuler, M., Kargi, S., 1992. Principles of Biochemical Engineering. Venkatadri, R., Peters, R., 1993. Chemical oxidation technologies: ultraviolet light/hydrogen peroxide, Fenton’s reagent, and titanium dioxide-assisted photocatalysis. Hazardous Waste and Hazardous Materials 10, 107e149. Watts, R., Jones, A., Cheng, P., Kelly, A., 1997. Mineral-catalyzed Fenton-like oxidation of sorbed chlorobenzenes. Water Environment Research 69 (3), 269e275. Wong, A.S., Crosby, D.G., 1981. Journal of Agricultural and Food Chemistry 29, 125. Zepp, R., Scholtzhauer, P., 1979. In: Leber, P.W.J.a.P. (Ed.), Polynuclear Aromatic Hydrocarbons. Ann Arbor Science Publishers, Ann Arbor, pp. 141e156. Zhao, L., Yu, Z., Peng, P.A., Huang, W., Feng, S., Zhou, H., 2006. Oxidation kinetics of pentachlorophenol by manganese dioxide. Environmental Toxicology & Chemistry 25, 2912e2919. Zimbron, J., Reardon, K., 2009. Fenton’s degradation of pentachlorophenol. Water Research 43, 1831e1840.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Spatio-temporal variation in trihalomethanes in New South Wales5 Richard J. Summerhayes a,e,*, Geoffrey G. Morgan a,d, Douglas Lincoln a, Howard P. Edwards a, Arul Earnest b, Md. Bayzidur Rahman c, Paul Byleveld f, Christine T. Cowie g, John R. Beard a a
University Centre for Rural Health, Northern Rivers, University of Sydney, PO Box 3074, Lismore, NSW 2480, Australia Centre for Quantitative Medicine, Duke-NUS Graduate Medical School, Singapore c School of Public Health and Community Medicine, University of New South Wales, Sydney, NSW 2052, Australia d North Coast Area Health Service, NSW, Australia e School of Health and Human Sciences, Southern Cross University, Lismore, NSW 2480, Australia f Water Unit, NSW Department of Health, PO Box 798, Gladesville, NSW 2111, Australia g Respiratory & Environmental Epidemiology, Woolcock Institute of Medical Research, 431, Glebe Point Road, Glebe, NSW 2037, Australia b
article info
abstract
Article history:
Aim: This paper describes the spatio-temporal variation of trihalomethanes in drinking
Received 16 December 2010
water in New South Wales, Australia from 1997 to 2007
Received in revised form
Method: We obtained data on trihalomethanes (THMs) from two metropolitan and 13 rural
22 August 2011
water utilities and conducted a descriptive analysis of the spatial and temporal trends in
Accepted 24 August 2011
THMs and the influence of season and drought.
Available online 1 September 2011
Results: Concetrations of monthly THMs in the two metropolitan water utilities of Sydney/
Keywords:
considerable variation between rural water utilities (range in mean THMs: 14.5e330.7 mg/L).
Trihalomethanes
Chloroform was the predominate THM in two-thirds of the rural water utilities. Higher
Chloroform
concentrations of THMs were found in chlorinated water distribution systems compared to
Bromodichloromethane
chloraminated systems, and in distribution systems sourced from surface water compared
Drought
to ground water or mixed surface and ground water. Ground water sourced supplies had
Disinfection by-products
a greater proportion of brominated THMs than surface water sourced supplies. There was
Australia
substantial variation in concentration of THMs between seasons and between periods of
Illawarra (mean 66.8 mg/L) and Hunter (mean 62.7 mg/L) were similar compared to the
drought or no drought. There was a moderate correlation between heavy rainfall and elevated concentrations of THMs. Conclusion: There is considerable spatial and temporal variation in THMs amongst New South Wales water utilities and these variations are likely related to water source, treatment processes, catchments, drought and seasonal factors. ª 2011 Elsevier Ltd. All rights reserved.
Abbreviations: THMs, trihalomethanes; THM4, total THM; NSW, New South Wales; DBP, disinfection by products; BDCM, bromodichloromethane; DBCM, dibromochloromethane. 5 Institution where this work was performed: University Centre for Rural Health, Northern Rivers, University of Sydney, Lismore, NSW 2480, Australia. * Corresponding author. Tel.: þ61 419249037. E-mail addresses: [email protected] (R.J. Summerhayes), [email protected] (G.G. Morgan), [email protected] (H.P. Edwards), [email protected] (A. Earnest), [email protected] (Md.B. Rahman), [email protected] (P. Byleveld), [email protected] (C.T. Cowie). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.045
5716
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Introduction
The provision of safe drinking water through disinfection to remove water-borne microbiological pathogens causing typhoid, cholera and gastroenteritis is one of the major achievements of public (Galal-Gorchev, 1996). An unexpected consequence of disinfection is the interaction of the disinfectant with natural organic matter (NOM) in the source water forming a range of chemicals collectively called disinfection by-products (DBPs) (Rook, 1974). DBP formation is not well understood but is influenced by the presence of bromide, iodide, pH, temperature, seasonal and climatic factors. Operational factors in water treatment also play a role in DBP formation including residency time within the distribution system, filtration methods used in removal of NOM, and disinfectant type and dose (Chowdhury et al., 2009). Trihalomethanes (THMs) were the first DBPs discovered in 1974 and are a group of organic compounds formed through reactions between methane (CH4) derivatives in NOM and chlorine or chloramines (Garrido and Fonseca, 2010). Since their discovery, more than 600 different DBPs have been reported, with THMs being the most frequently detected compounds (Richardson et al., 2007). In 1976, the US National Cancer Institute declared chloroform a carcinogen in animals and a suspected carcinogen in humans (National Cancer Institute, 1976). Exposure to DBPs is associated with bladder, rectal and colon cancer, ‘suggestive of a causal inference’ (Hrudey, 2009). While some reproductive outcomes such as small for gestational age and pre-term births have also been associated with DBP exposure, the current evidence is inconclusive (Grellier et al., 2010). In Australia, DBPs are not regulated but a guideline value of 250 mg/L for total THM is recommended and action to reduce THMs is encouraged, while not compromising disinfection as exposure to non-disinfected water poses substantially greater health risk than exposure to low level THMs (NHMRC, 2004). There is limited published data describing THMs or other disinfection by-products in Australia. A survey of several DBPs during 1994-95 in 16 cities throughout Australia found that some Australian drinking water supplies had high THM concentrations (up to 191 ug/L) (Simpson and Hayes, 1998). In New South Wales (NSW), Australia’s most populated state, more than 5 million residents (approximately 80% of the population) use public drinking water as their usual source of drinking water (Centre for Epidemiology and Research, 2002). The two largest public water suppliers provide drinking water to the Sydney/Illawarra (4.2 million people) and Hunter (516,000 people) metropolitan areas. In rural areas, local water utilities (largely through local government authorities) provide drinking water. While the NSW Health Department recommends that all water utilities collect monthly THM samples (NSW Health, 2000), only the large metropolitan utilities of Sydney/Illawarra and the Hunter, and a small number of rural water utilities conduct regular THM monitoring. This paper describes THM concentrations throughout New South Wales, Australia, covering large metropolitan water utilities, as well as small to medium size rural water utilities. Our study assesses geographic differences and temporal
trends in monthly THM data from metropolitan utilities over several years, and rural utilities for at least one year. During our study period, much of NSW experienced one of the longest droughts on record lasting nearly a decade from 1997 to 2006 (Bond et al., 2008) and we also investigated the influence of drought on THM’s.
2.
Methodology
2.1.
Water utility data
We obtained data on THMs for the periods 1998 to 2004 for Sydney/Illawarra region, 1997 to 2004 for Hunter region and various time periods between 1997 and 2007 for the rural water utilities summarized in Table 1.
2.1.1.
Sydney/Illawarra water utility data
The Sydney Water Corporation (SWC) supplies water from five surface catchments to the Sydney/Illawarra metropolitan region, covering an area of 12,700 km2. Sydney/Illawarra has a three level hierarchical structure, with 14 delivery systems (average area 241 km2) supplied with surface water treated at nine water filtration plants. Each delivery system contains from one to six distribution systems (average area 84.3 km2) which are treated either by chloramination or chlorination. Rechlorination occurs within the distribution system. Water is stored in 180 water supply zones (average area 16.4 km2) which supply water to homes within the distribution systems. Limited data was also available on a range of water quality factors and other DBP’s including: trichloracetonitrile (5 months in 1998 covering 18% of supply zones); six haloacetic acids (HAAs) (two periods of 12 and 7 months covering 69% of supply zones); and three HAAs (bromoacetic-, dibromoaceticand tribromoacetic acids, 5 months covering all supply zones). Due to the limited duration of sampling for these DBP’s, and the lack of comparable data from other water supplies in NSW, we have not reported these results in detail in the paper, but summary statistics are provided in the Supplementary Material AeC. There are some 3000 THM water sampling sites within the Sydney/Illawarra distribution systems. Monthly monitoring is generally conducted at these sites on a three to six-monthly rotational cycle with a minimum of 3e6 sites operating within each distribution system (SWC, 2002). Monthly THM data from all available monitoring sites within each supply zone was averaged to obtain zone/month THM concentrations for each month in the study period.
2.1.2.
Hunter water utility data
The Hunter Water Corporation supplies the Hunter metropolitan region via nine water distribution systems covering an area of 5400 km2. Six distribution systems received surface water and three received ground water for most of the study period although prior to 2003 blending of surface and ground waters from different sources occurred at various time periods for three distribution systems normally supplied by surface water. After 2003 two distribution systems, previously supplied by ground water, were continuously augmented with
5717
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Table 1 e Characteristics of NSW metropolitan and rural water utilities included in the study. Water Utility
Water Source
Disinfectant Process
Treatment Process (Catchment characteristics)
Sydney/ Illawarra Hunterb
Surface Surface
Chlorinateda Chloraminated Chlorinated
Ground
Chlorinated
Filtration, flocculation, coagulation, lime. Filtration, coagulation, flocculation, sedimentation, powdered activated carbon (PAC) Aeration, lime, coagulation, filtration
Surface
Chlorinated
2
Surface
Chlorinated
3
Surface
Chlorinated
4 5 6
Chlorinated Chlorinated Chlorinated
7
Surface Surface Surface blended Surface
Chlorinated
8ac
Ground
Chloraminated
8bc
Surface
Chlorinated
9 10
Surface Surface
Chlorinated Chlorinated
11
Surface
Chlorinated
12 13
Surface Surface
Chlorinated Chlorinated
Rural Utilities 1
Filtration, coagulation, sedimentation, PAC Filtration, flocculation, sedimentation, Filtration, coagulation, flocculation, sedimentation Filtration No treatment Lime, aeration Coagulation, flocculation, sedimentation Direct filtration, aeration, coagulation, flocculation, dissolved air flotation Filtration, coagulation, flocculation, sedimentation Direct filtration Direct filtration, coagulation, flocculation, sedimentation Filtration, dissolved air flotation, flocculation Filtration, flocculation Filtration, coagulation, flocculation, sedimentation
Approx. pop. (2002)
Number of samples
Study Period (months of data)
600,000 3,600,000 452,000
5341
Jan 1998eDec 2004 (84)
863
Jan 1997eDec 2004 (96)
4000
28
Feb 2003eNov 2003 (9)
21,000
156
Nov 1997eDec 2005 (89)
160,000
72
Nov 2003eJan 2004 (8)
26,000 65,000 2000
314 508 61
Jan 2001eMay 2007 (62) Jan 2001eDec 2004 (54) Jan 2001eMay 2006 (61)
20,000
30
Jul 2001eJan 2004 (30)
4500
1378
Dec 2000eMay 2006 (63)
11,000
660
Dec 2000eMay 2006 (66)
21,000 53,000
85 43
Jan 2000eOct 2006 (80) Jun 2001eAug 2005 (40)
74,000
1610
Jan 2000-Apr 2006 (61)
8000 135,000
43 159
FebeDec 2003 (11) Mar 1999eJan 2006 (86)
60,000
a In Sydney/Illawarra, 4 of the 33 distribution systems changed disinfection from chlorination to chloramination in June 2003. b In Hunter, two distribution systems using ground water were augmented with ground and surface waters from other sources after 2003, and prior to 2003, some systems supplied with surface water were occasionally augmented with blended ground and surface waters from other sources. c Water utility 8 includes a chloraminated distribution system with ground water supply (8a) and a chlorinated distribution system with a surface supply. Data for 8a should be treated with caution as majority of values were below <5 mg/L detection limit.
blended surface and ground water from different sources (HWC, 2004). A fixed sampling site is located within each distribution system towards the extremities of the systems, providing distribution/month THM values.
2.1.3.
Rural water utility data
We surveyed all 106 utilities in rural NSW requesting information and data on DBPs and water quality parameters. Ninety-four rural water utilities (89%) responded to the survey and 27 (26%) indicated they collected some data on THMs. We received data on THMs from 25 (24%) rural water utilities of which 13 (12%) had sufficient THM data to be included in our analysis. Five of these rural utilities supplied additional monitoring data on a range of other water quality parameters, however these data covered limited durations and were insufficient to assess seasonal trends and geographic differences and in keeping with the objectives of the paper are not
reported in detail, although they are briefly summarised in the Supplementary Material A. The size and complexity of rural water utilities varies, and THM sampling varied from one-off short term sampling periods to routine quarterly or monthly sampling covering various time periods between 1997 and 2007 (see Table 1). All sampling was conducted at random sites throughout the distribution systems. Water utility #6 provided data only on THM4 from sample points within the distribution system. A separate one-month survey in utility #6 including posttreatment samples indicated that chloroform comprised approximately 86% and BDCM 13% of the THM4 concentration entering the distribution system (these survey data not included in the final analysis but are provided in Supplementary Material D). Rural utility #8 includes a distribution system which is chloraminated and mainly sourced from ground water, and a chlorinated distribution system
5718
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
sourced from surface water. We designated the two distribution systems as #8a and #8b respectively for reporting. The source water, disinfection, treatment types, number of THM samples and sampling period for each of the water utilities included in our analysis are shown in Table 1. We included only those rural water utilities with monthly monitoring data covering all four seasons in at least one year (DecembereFebruary ¼ summer, MarcheMay ¼ autumn, JuneeAugust ¼ winter and SeptembereNovember ¼ spring) so that seasonal variation could be assessed. For each rural water utility with sufficient data to be included in the study we averaged available monthly THM concentrations to obtain utility/month concentrations.
2.1.4.
Trihalomethane monitoring
Water samples were analysed for THMs by Sydney Water Corporation laboratories using Standard Methods for the Examination of Water and Waste Water (20th Ed.), Method 6200B (modified) and USEPA Method 8260 (modified) (APHAAWWA-WPCF, 1998; USEPA, 2008); in the Hunter Water Corporation based on the USEPA standard Method 8260 (USEPA, 2008) and in rural water utilities using the Standard Methods for the Examination of Water and Waste Water (18th ed) Method 6232 (APHA-AWWA-WEF, 1992). Further details of the analytical techniques are provided in the Supplementary Material E. Multiple values below the detection limit were recorded within each water utility over the study period. The limit of detection for individual THMs was usually <1 mg/L, with occasional higher limits of detection reported. We used a similar approach to other studies and substituted samples flagged as below the detection limit with two-thirds of the detection limit value, which is the approximate mean of a lognormal distribution (Whitaker et al., 2003). Bromoform was found to be below the detection limit more than 75% of the time across most utilities (metropolitan and rural), and therefore we examined bromoform only as part of total brominated THMs (BrTHM) and the sum of the four trihalomethanes (THM4) (percentage of observations below detection values provided in Supplementary material F).
2.2.
Rainfall data
Data on rainfall was obtained from the Sydney Catchment Authority, and the Australian Bureau of Meteorology website (Bureau Of Meteorology, 2010). Data on monthly drought status for NSW districts containing the water utilities was obtained from the NSW Department of Primary Industries, with each month classified as ‘no drought’, ‘marginal’ or ‘drought’. Drought status is defined by a number of criteria including monthly rainfall, temperature, frost and evaporation, poor pasture biomass, soil moisture and livestock. Marginal conditions include rainfall for the previous 3 months within or below average for the three month rainfall decile and surface water supplies less than 50% of normal for the time of year. Drought conditions include rainfall for the previous 6 months within or below average for the 6-month rainfall decile and surface water supplies less than 30% of normal for the time of year (NSW Department of Primary Industries, 2010).
2.3.
Statistical analysis
We produced a range of descriptive statistics using log transformed THM concentrations due to the skewed THM distribution and back-transformed the results for reporting. Differences in the means for season and drought periods were assessed using KruskaleWallis tests. Correlations between THMs and rainfall lagged up to one month were conducted using the Spearman correlation coefficient. The selection of a one month lag is based on a 30e40 day residency time for Prospect Reservoir in the Sydney/Illawarra supply (Hamilton et al., 1995). Data management was conducted in Excel (version 2003) and Access (version 2003) and all statistical analyses were conducted using SAS version 9.1.3 (SAS Institute Inc., Cary North Carolina USA).
3.
Results
3.1.
Spatial variation
Table 2 summarises descriptive statistics for individual THMs and THM4 concentrations in treated water for all the selected NSW water utilities. The mean THM4 concentrations in Sydney/Illawarra (66.8 mg/L) and Hunter are similar (62.7 mg/l). Water treated by chlorination in Sydney/Illawarra had higher mean THM4 concentrations (81.1 mg/L) than water treated by chloramination (50.8 mg/L). The Prospect South delivery system in Sydney/Illawarra switched from chlorination to chloramination in July 2003 and the mean monthly chloroform concentration decreased substantially from 59.5 mg/L during the chlorination period (1998 to mid-2003) to 23.0 mg/L ( p < 0.001) during the chloramination period (mid2003e2004). In the Hunter, chlorinated surface water had higher mean THM4 concentrations (74.4 mg/L) than chlorinated ground water (36.1 mg/L). From 2003, Hunter augmented two ground supplies and three surface supplies continuously with blended ground and surface waters in response to drought conditions. The mean THM4 in the ground water supplies increased from 17.9 to 40.6 mg/L with augmentation and from 66.4 to 74.2 mg/L in the surface supplies. The majority of the rural water utilities reporting in the study use chlorination, one system used chloramination together with chlorination. Rural NSW water utilities generally use surface water, with some using ground water or a blend of ground and surface waters. During the study period there was large variability in the monthly mean THM4 concentration in treated water between rural water utilities. The THM4 concentration of the rural utility with the minimum (#8a: 15 mg/L) and maximum (#6: 331 mg/L) monthly concentrations were substantially different compared to the remaining 11 utilities which ranged from of 63.7 mg/L to 189.1 mg/L, although this trimmed range still represents a 3 fold difference in monthly THM4 concentrations. Chloroform is the main component of total THM in chlorinated (66%) and chloraminated (59%) surface water in Sydney/Illawarra. Chloroform is also the main component of total THM in chlorinated surface water in the Hunter (60%) and in most rural utilities with the exception of utilities #2 (29%), #3
5719
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Table 2 e Concentrations of trihalomethanes in treated drinking water in water utilities in New South Wales. Water Utility
Metropolitan Sydney/Illawarra Chloraminated Chlorinated Hunter Ground Blended Surface Rural Utilities 1a 2 3 4 5 6a 7 8a 8b 9 10 11 12a 13
N
Mean
THM4
Chloroform
BDCM
Min
P95
Max
Mean
Min
Max
Mean
DBCM
Min
Max
Mean
Min
Max
563 278 340 72 18 30 48
66.8 50.8 81.1 62.7 36.1 72.6 74.4
22.7 22.7 25.0 10.1 10.1 14.7 30.1
114.6 81.3 127.0 105.2 74.3 106.4 105.2
196.7 199.0 196.7 114.6 74.3 134.0 131.0
40.1 28.3 50.7 35.2 16.0 33.4 44.5
4.5 9.0 4.5 0.7 0.7 1.0 20.5
162.5 88.0 162.5 96.3 46.8 78.3 96.3
16.5 14.2 18.6 15.2 8.3 20.7 18.1
3.0 3.0 5.7 0.7 0.7 0.7 3.1
46.0 39.0 46.0 31.2 20.8 34.3 30.0
7.1 5.8 8.1 9.1 5.9 15.2 10.0
0.7 0.7 0.8 0.9 2.8 3.7 0.9
27.3 31.0 27.3 24.9 9.2 27.0 24.0
9 89 8 62 54 61 30 63 66 80 42 61 11 86
189.1 114.9 81.7 106.1 73.4 330.7 87.5 14.5 102.2 99.3 63.8 63.7 138.5 94.9
67.5 21.0 65.4 21.1 23.8 63.0 22.7 13.1 44.9 39.0 13.3 18.4 52.3 2.7
290.7 231.0 102.3 170.2 120.1 501.0 147.3 17.8 145.4 157.9 86.3 102.3 286.0 152.7
290.7 346.0 102.3 245.4 122.3 586.0 213.7 23.6 221.2 180.0 120.0 128.9 286.0 229.0
104.8 33.6 29.5 45.0 48.2 na 72.8 4.4 63.9 46.8 34.1 33.0 87.1 54.5
29.7 2.0 13.3 16.8 13.9 na 3.0 3.3 22.2 12.0 3.3 5.7 17.3 0.7
190.1 172.0 47.9 108.8 90.0 na 200.0 11.2 159.2 112.8 68.0 86.6 155.0 110.0
48.8 32.5 24.9 26.8 16.5 na 11.9 3.4 25.6 33.4 16.2 16.6 31.3 26.7
22.5 4.5 20.6 0.7 6.3 na 3.0 3.3 8.2 12.0 3.3 3.8 13.3 0.7
94.1 104.7 31.2 56.7 24.2 na 20.0 5.4 51.3 57.0 39.0 32.2 78.0 76.0
19.8 32.7 16.6 16.0 5.7 na 1.4 3.3 9.1 17.9 9.5 9.2 12.3 11.1
4.7 2.8 6.9 0.7 2.7 na 1.0 2.5 3.3 8.0 1.0 3.3 1.3 0.7
76.3 128.5 24.7 56.7 10.3 na 2.0 4.1 25.3 56.0 32.0 32.0 48.0 64.0
All values in mg/L ¼ micrograms per litre, N ¼ number of observations, Min ¼ minimum, P95 ¼ 95th percentile, Max ¼ maximum, na ¼ not available. Sydney/Illawarra mean: mean of annual zone means. Hunter mean: mean of annual distribution system means. Rural water utilities mean: mean of monthly utility means. Water utility 8 includes a chloraminated distribution system with ground water supply (8a) and a chlorinated distribution system with a surface supply. Data for 8a should be treated with caution as majority of values were below <5 mg/L detection limit. Water utility 6 only has data for THM4. a Utilities with 95th percentile THM4 concentrations above the Australian National Drinking Water Guideline value of 250 mg/L.
(36%) and in the ground water sources in the Hunter (56%) and #8a (30%) which had higher proportions of brominated THMs.
3.2.
Rainfall
Fig. 1 illustrates the temporal variation in Sydney/Illawarra chloroform and brominated THM concentrations and mean monthly rainfall. Prior to October 1998 brominated THM concentrations were generally higher than chloroform, then decreased until mid 1999 and have since remained relatively constant. Chloroform concentrations decreased steadily from mid-1998 but showed considerable variability, and from late 2003 concentrations were similar or lower than brominated species. Several peaks and troughs in chloroform concentration coincide with heavy rainfall events. We found a moderate correlation between overall monthly THM4 and rainfall (lag one month, r ¼ 0.35, p < 0.01) for the Sydney/Illawarra, with higher correlations within some of the five Sydney/Illawarra catchments (Blue Mountains/Cascade catchment: r ¼ 0.57, p < 0.001; Upper Nepean/Illawarra catchment r ¼ 0.47, p < 0.001). Fig. 2 shows the temporal variation in chloroform, brominated THMs and rainfall in the Hunter region. Chloroform concentrations are generally higher than brominated THMs,
with occasional large peaks in chloroform and brominated THMs from 2001. We found no correlation in the Hunter region between rainfall (lag one month) and mean THM concentration in distribution systems supplied by ground water, and a moderate correlation in distribution systems supplied by surface waters (r ¼ 0.59, p < 0.01). We generally found moderate correlations between rainfall and THM4 concentrations in the rural utilities located in subtropical coastal floodplains including utilities #11 (r ¼ 0.30, p < 0.02), #8b (r ¼ 0.58, p < 0.001) and #5 (0.67, p < 0.001). We generally found little correlation with mean monthly rainfall (lag 1 month) and THM in the other rural utilities (data not shown).
3.3.
Drought
We examined the difference in the concentration in THMs in periods of drought compared to marginal and no drought and the results are summarised in Table 3. Sydney/Illawarra was in drought for 18 months (21.4%) of the 84 month study period and continuously in marginal drought from September 2002 until mid-2007. There was a consistent significant decrease in mean THM4 concentrations during drought months compared to non-drought (THM4: 52.6ug/L compared to
5720
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
90
120.0 BrTHM
Chloroform
Rainfall
100.0 70 60
80.0
50 60.0 40
40.0
30 20
Mean Monthly Catchment Rainfall (mm)
Mean Monthly Chloroform and Brominated THM Concentration (ug/L)
80
20.0 10 0.0
0
Fig. 1 e Temporal variation in Sydney/Illawarra monthly chloroform and brominated THM concentration and monthly rainfall.
75.9ug/L, p < 0.001), and this was reflected in chloroform and BDCM, while DBCM concentration increased. The majority of sampling for HAAs in Sydney/Illawarra occurred in a nondrought period, however three HAAs (bromoacetic-; dichloroacetic- and trichloroacetic acids) showed significant decline during drought compared to no drought consistent with reductions in THMs, except DBCM (see Supplementary Material C). Hunter was in drought for 19 months of the 96 month study period and only the blended distribution systems showed a significant THM4 decrease during drought compared to no drought (THM4: 69.8 mg/L compared to 46.4 mg/L, p < 0.05), reflected in chloroform, BDCM and DBCM. THM concentrations 100
Seasonal variation
We found considerable seasonal variability in mean THM4 concentrations between locations and these results are summarised in Table 4. In Sydney/Illawarra there was a significant
BrTHM
Rainfall 400 350
70 300 60 250 50 200 40 150 30 20
100
10
50
0
0
Mean Monthly Catchment Rainfall (mm)
Mean Monthly Chloroform and Brominated THM Concentration (ug/L)
80
3.4.
450 chloroform
90
in rural water utilities during drought periods compared to non-drought periods were extremely variable. Five rural water utilities that were in drought continuously for 12 months or longer experienced significantly reduced chloroform concentrations compared to periods of no drought (#2, #4, #9, #10 and #13), and these same utilities experienced significantly increased DBCM concentrations.
Fig. 2 e Temporal Variation in Hunter monthly chloroform and brominated THM (BrTHM) concentration and monthly rainfall (mm).
Table 3 e Mean monthly trihalomethane concentrations during drought compared to no drought and marginal drought, NSW water utilities. Water Utility
Time (months
THM4
Chloroform
BDCM
DBCM
No Marginal Drought No Marginal Drought No Marginal Drought No Marginal Drought No Marginal Drought Drought n (%) n (%) Drought Mean Mean Drought Mean Mean Drought Mean Mean Drought Mean Mean n (%) Mean Mean Mean Mean 50 (59.5)
16 (19.1)
18 (21.4)
19 (19.8)
75.9 59.4 87.2 58.4 31.5 69.8 72.3
51.1 37.0 66.7 73.6 56.1 70.6 84.2
52.6** 36.5** 67.8** 66.4* 50.2** 46.4* 80.6*
49.0 36.0 57.8 33.2 12.5 36.7 46.4
27.5 17.4 38.7 41.6 22.9 35.4 54.5
27.6** 16.5** 37.9** 35.4 20.7** 19.0** 47.5
17.7 16.1 18.8 13.9 6.8 20.0 16.1
13.7 11.3 16.2 18.1 16.2 19.8 18.1
14.1** 11.4** 16.7** 17.0 14.4** 11.6* 20.0**
6.6 5.3 7.5 8.2 5.9 11.5 8.4
7.4 6.3 8.7 10.7 11.6 12.1 9.5
8.0** 6.6** 9.4** 10.6 9.8** 9.9 11.2*
60 (62.5)
17 (17.7)
26 (27.1) 1 (12.5) 10 (16.1) 24 (44.4) 23 (37.7) 12 (40.0) 24 (36.4) 29 (36.2) 10 (25.0) 38 (62.3) - (0.0) 47 (54.7)
19 (19.8) 4 (50.0) 5 (8.10) 16 (29.6) 20 (32.8) 6 (20.0) 7 (10.6) 16 (20.0) 7 (17.5) 15 (24.6) 1 (09.0) 22 (25.6)
51 (53.1) 3 (37.5) 47 (75.8) 14 (25.9) 18 (29.5) 12 (40.0) 35 (53.0) 35 (43.8) 23 (57.5) 8 (13.1) 10 (91.0) 17 (19.8)
149.9 67.4 151.2 74.8 287.5 86.0 97.1 100.4 63.7 63.4 na 83.9
97.6 86.7 134.5 81.3 379.3 89.2 108.8 97.3 56.9 69.0 157.9 112.7
113.5* 88.8 103.8* 82.6 331.9** 88.1 105.9 100.3 67.5 71.9 143.1 104.5*
55.6 14.3 92.3 49.2 a 74.3 56.8 54.8 48.3 33.5 na 50.7
36.8 30.8 92.9 56.9 a 74.3 70.9 40.5 29.5 39.7 135.0 67.5
25.3* 39.4 41.0* 57.7 a 70.6 68.9 43.7* 30.6* 34.0 94.1 48.5*
32.6 22.7 21.8 17.7 a 9.1 26.8 32.5 9.5 16.9 na 22.9
32.1 26.0 30.2 17.0 a 12.0 26.3 34.8 15.4 18.5 20.3 33.0
30.9 26.7 31.2 17.6 a 14.8** 25.3 33.6 19.7* 18.6 33.8 31.3*
32.4 20.2 5.8 7.2 a 1.3 10.1 12.7 2.6 9.6 na 9.4
23.2 21.9 11.6 6.7 a 1.6 8.3 20.2 8.7 8.8 1.3 10.4
36.6* 17.2 22.2* 7.2 a 1.5 8.7 21.1** 13.0** 14.8 14.1 18.7*
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Sydney/Illawarra Chloramination Chlorination Hunter Ground Blended Surface Rural Utilities 2 3 4 5 6 7 8b 9 10 11 12 13
* Significant at p < 0.05 level. ** Significant at p < 0.001 level, na e no data were available Means are reported in mg/L. Water utility #12 was in marginal or drought only. Water utilities #1 and #8a not shown as utility #1 was continuously in drought during the study period and water utility #8a values were below threshold and showed no variation between the means. a Water utility #6 reported THM4 data only.
5721
5722
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Table 4 e Seasonal variation in the mean trihalomethane concentration in water utilities in New South Wales. Summer Mean
THM4 Autumn Mean
Winter Mean
Spring Mean
69.9 54.8 80.0a,byy 60.8 34.2 43.1a,cy 75.4
72.6 58.0 79.3a,byy 66.8 39.6 74.8 80.6
64.3 54.0 71.9 60.5 39.5 66.7 77.4
60.3 52.0 69.8 62.5 39.8 63.1 71.5
þ5.6 þ0.8 þ8.1 þ0.3 5.3 23.6 þ2.0
165.4 90.9 90.2 87.4 46.2 324.3 77.1 79.8 99.5 60.1 48.7 89.5 97.3
189.3 104.8 79.7 86.0 62.0 345.1 97.8 91.3 97.7 64.2 58.1 126.1 94.7
þ68.3 þ55.1 8.2 þ23.4 þ43.5 þ16.6 þ3.4 þ44.3 1.9 þ4.0 þ23.4 þ123.0 3.6
Sydney/Illawarra Chloramination Chlorination Hunter Ground Blended Surface Rural Utilities 1 2 3 4 5 6 7 8b 9 10 11 12 13
233.7 146.0ay 82.0 110.8 89.7ayy 340.9 80.5 124.1ayy 97.6 64.1 72.1ay 212.5 93.7
183.6 112.3 66.2 122.5a,y 85.5ay 313.8 96.9 107.0ay 102.2 70.7 67.3 129.2 92.0
Summer-Winter
Difference (%) (9.0) (1.5) (11.3) (0.5) (15.5) (54.8) (3.0) (41.3) (60.6) (9.1) (26.8) (94.2) (5.1) (4.4) (56.5) (1.9) (6.7) (48.0) (137.4) (3.7)
Means and difference between summer and winter means are reported in mg/L. y significant at the p < 0.05 level. yy significant at the 0.001 level. Water utility 8a values were below threshold and showed no variation between the means. a Difference in means compared to winter. b Difference in means compared to spring. c Difference in means compared to autumn.
increase ( p < 0.001) in mean THM4 concentration in chlorinated water during summer (mean 80.0 mg/L) compared to winter (mean 71.9 mg/L) and spring (69.8 mg/L), but there was no difference in chloraminated distribution systems. In the Hunter THM4 concentrations in blended water increased significantly ( p < 0.05) by 23.6 mg/L in winter compared to summer, while there was generally a small increase in surface and ground water concentrations. There was substantial variation in the seasonality of THM4 concentrations between rural water utilities. Ten of the 13 rural water utilities had higher mean THM4 concentrations in summer compared to winter ranging in increases of 4e123 mg/L, and this increase was significant ( p < 0.05) in four utilities.
4.
Discussion
4.1.
Spatial variation
We found the concentration of THM4 and speciation to be similar for the Sydney/Illawarra (THM4 ¼ 67 mg/L, 60% chloroform) and Hunter (THM4 ¼ 63 mg/L, 56% chloroform) metropolitan water utilities. We found considerable variation in THM concentrations and speciation between rural water utilities, with mean monthly THM4 concentrations ranging from 15 to 331 mg/L (29e86% chloroform). The THM4 concentration of the rural utility with the minimum (#8a: 15 mg/L) and maximum (#6: 331 mg/L) in the range were substantially different compared to
the remaining 11 utilities and suggests that the factors influencing THM4 concentrations in utilities #8a and #6 are substantially different to the other 11 rural utilities. Differences in geography and catchment characteristics may contribute to variation in THM concentrations throughout NSW. Utilities in the semi-arid floodplains in the far-west of the State (#1, #2 and #12) had higher THM4 concentrations (range of means: 114e189 mg/L) compared to other water utilities (range of means: 15e106 mg/L) with the exception of water utility #6 (mean 331 mg/L) which is in a sub-tropical coastal floodplain area. The 95th percentile for THM4 concentration for three rural water supplies was above the Australian guideline value of 250 mg/L (Table 2) suggesting these utilities need to implement strategies to reduce THM concentrations (NHMRC, 2004). The Australian guideline value for THM4 of 250ug/L is higher than a number of other developed countries including United Kingdom, Japan (100 mg/L), USA (80 mg/L) France (30 mg/L) and Germany (10m/L) (Rizzo et al., 2005). The large variation in mean THM4 in rural NSW is similar to that reported in a survey of DBPs in North Virginia, USA prior to the introduction of the 1979 interim Disinfection By-Product Rule of <100 mg/L (USEPA, 1979) which found mean THM4 concentrations of 249 mg/L (range 40e531 mg/L) in 1975e76 and 173 mg/L (range 43e889 mg/L) in 1976e77 (Hoehn and Randall, 1979). In a study in Turkey, where THMs are not regulated, mean THM4 concentrations of 159 mg/L and 129 mg/L were reported (Rizzo et al., 2005). Similar large variation has also been found in rural communities in Alberta, Canada during
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Autumn 2000 with mean THM3 (excluding bromoform) ranging from 44m/L to 210m/L (Charrois et al., 2004). THM4 levels in NSW utilities were generally higher than levels found in countries where THMs are regulated. A 1988-89 survey reported median THM4 value of 39 mg/L for 35 utilities across the USA (Krasner et al., 1989). A survey of 113 systems in North Carolina in 2004-05 reported an average THM4 of 40.8 mg/L. In the United Kingdom, a study of data from 1992 to 1996 reported mean THM4 concentrations of 46 mg/L (Keegan et al., 2001). Chloroform generally comprised the majority of the total THM in treated surface water in NSW (52%e86%). The domination of brominated THMs in treated ground water in the Hunter (56%) and rural utility #8a (70%) is likely due to the coastal aquifers having higher levels of bromide compared to the surface supplies (Kampioti and Stephanou, 2002). The surface water supplies of utility #3 (64% brominated THMs) lies in a coastal floodplain area which may be affected by saltwater intrusion; while utility #2 (71% brominated THMs) lies in an ancient sea-bed which has been noted to have high bromide concentrations (Richardson, 2005). Chloraminated systems generally had lower THM4 concentrations (utility #8a mean THM4 14.5 mg/L, Sydney mean THM4 50.8 mg/L) than chlorinated systems. A change treatment from chlorination to chloramination in a delivery system in Sydney resulted in a 159% reduction in THM4. Bougeard et al. (2010) in a 2008 study in the United Kingdom reported a 92% reduction in THM4 with a shift in disinfection from chlorination to chloramination. A major strength of this study lies in the large number of observations covering a lengthy time period, incorporating periods of long-term drought, drought-free periods and heavy rainfall events. While we were only able to access THM data on a small proportion of the total number of rural NSW water utilities the surveyed utilities cover a wide range of geographic and climatic regions encountered across the Australian continent.
4.2.
Seasonal variation
We found substantial seasonal variation in THM4 concentrations in chlorinated water in Sydney/Illawarra and in many of the rural NSW water utilities, with generally higher concentrations in summer/autumn and lower concentrations in winter, especially in semi-arid locations in the far-west of NSW (utilities #1, #2 and #12). The ground water supplied systems in the Hunter showed little THM variation between seasons and these results are consistent with overseas studies. A Canadian survey of THMs showed a more than two-fold variation during summer and winter in chlorinated systems (62.5 mg/L compared to 33.5 mg/L) and chloraminated systems (32.8 mg/L compared to 13.7 mg/L), however in ground water supplies there was little variation between seasons (Williams et al., 1995). Another study in China in 2003 found THM levels to be 50 mg/L in autumn and around 10 mg/L in spring in chloraminated waters and noted variations in different organic matter concentrations and in the dynamics of algae/plankton production in the different seasons (Chen et al., 2008). Summer is considered the most challenging period for treating water and maintaining water quality and the
5723
magnitude of seasonal fluctuations may be dependent on the water source (McGuire and Meadow, 1988).
4.3.
Rainfall
Several peaks and troughs in chloroform concentration coincide with heavy rainfall events in Sydney/Illawarra. One such event was the rains in mid 1998 that broke a drought lasting from 1992 to 1998 and resulted in large inflows of contaminants and organic matter into the Sydney/Illawarra catchment (Cox et al., 2003). The moderate correlations found between rainfall and THM reported in our results are consistent with the correlations found in a North Virginia, USA study (Hoehn and Randall, 1979). Heavy drought breaking rainfall is associated with high turbidity, increased inflows, soil leaching of NOM and microbiological contamination from degraded catchments (Stein, 2000).
4.4.
Drought
Droughts lasting several months to several years are a regular feature of the Australian environment. Much of NSW experienced one of the longest droughts on record lasting nearly a decade from 1997 to 2006 (Bond et al., 2008). Sydney and parts of NSW experienced drought from 1992 to 98 with a significant wet period in mid-1998, then entered into drought again from 2002 until early 2011. During this period the Sydney catchment storage (2 million megalitres) fell from 91% in 2000 to 32.5% in mid-2007, in some rural catchments water levels fell below 30%, whereas the Hunter maintained a storage capacity above 60% by augmenting its water supply from 2002 with ground and surface water from additional sources in response to drought conditions (HWC, 2004). Inflows into major river systems in NSW were some of the lowest on record and some floodplains and wetlands had not been flooded during the decade of drought (Murphy and Timbal, 2008). While the effects of drought were variable, Sydney/Illawarra and five of the 13 rural utilities experienced consistent decrease in chloroform and an increase in DBCM during drought periods. The effect of drought was also evident in both chlorinated and chloraminated water as illustrated by decreased chloroform concentrations in the Prospect South distribution system during periods of drought when the system was chlorinated and when it changed to being chloraminated. The blended water distributions systems in the Hunter showed decreased THM concentrations during drought periods, while the ground and surface water systems showed increased THM concentration. Our findings in NSW are broadly similar to a UK study that also that found decreases in chloroform concentration in water supplies during drought (Whitaker et al., 2003). A study in Greece showed a lower mean chloroform concentration during an intense drought period (2.27 mg/L) compared to no drought (10.22 mg/L) and an increased mean DBCM concentration (drought 13.76 mg/L; no drought 1.38 mg/L) from increased bromide through saltwater intrusion and ground water augmentation (Kampioti and Stephanou, 2002). A North Virginia, UAS study reported a large variation in THMs during a period of no drought from 1975 to 76 (mean THM4 249 mg/L) compared to severe drought from June 1976 to November 1977
5724
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
(mean THM4 173 mg/L). The mean concentration of chloroform decreased from 246 to 152 mg/L in the drought compared to no drought while a small increase in BDCM from 19 to 21 mg/L was reported (Hoehn and Randall, 1979). Drought can cause longterm variations in DBP formation and the 1997-98 US Information Collection rule survey found that long-term seasonal factors such as drought, flood, hurricanes may account for seasonal variation in water quality both in the source water and in the distribution system although this was not assessed in detail (McGuire et al., 2002). Our results support previous findings that seasonal and long-term effects of drought can have complex effects on the nature and presence of NOM resulting in either an increase or decrease in NOM which may in turn affect the nature and concentration of THMs, suggesting that the effects of drought may be catchment specific (Williamson et al., 1999). Algal blooms, common in Australian waters may also play a large role during warmer months and periods of low rainfall. Algal blooms have been correlated with THM levels and regarded as an important precursor for DBP production (Hoehn and Randall, 1979; Chen et al., 2008). Although we had no data on algae levels, algal blooms occurred in several rural water utilities in coastal floodplain/estuarine areas (utilities #3, #6, #13) and sub-tropical coastal areas (#11, #8b) during warmer months and periods of low rainfall and drought. The concentration of chloroform in all these surface supplied distribution systems was higher during drought periods. Rural utility #6 suffered a number of major algal blooms during the study period and recorded the highest concentrations of THM4 amongst the study locations. A decrease in organic matter can occur during drought due to reduced run-off from catchments and inflows, which can lead to lower turbidity and settling of sediments and organic matter (Bond et al., 2008; Murphy and Timbal, 2008). Degradation of organic matter over time during lengthy dry periods and slower travel time of water compared to wet events can also reduce organics in the water (Personal communication Sydney Catchment Authority 2011). Drought can also influence ground water quality due to declines in aquifer levels and increased salinity causing bromide concentrations to rise which can affect THM speciation (Krasner et al., 1994). While there is much speculation on the effects of climate change on drinking water quality (Bates et al., 2008), increases in the frequency and extent of drought affected areas and flooding from drought breaking rains are expected in south eastern Australia (ABS, 2008; Dore, 2005). Water treatment authorities respond to the higher risk of microbial contamination of the raw source water during or shortly after heavy rainfall by using higher disinfectant concentrations which in turn can promote THM formation (Eikebrokk et al., 2004). Changes in climatic factors related to climate change are likely to create additional operational challenges for water utilities in Australia, particularly in rural areas of NSW with limited resources (Hurst et al., 2004; Soh et al., 2008).
5.
Conclusion
We found considerable variation in THM concentration between rural water utilities in NSW with some utilities
experiencing elevated concentrations above current national THM4 guideline values indicating that these utilities require action to reduce THM’s, while not compromising disinfection. We obtained THM data from a small proportion of rural NSW water utilities and our results suggests that elevated THM4 concentrations may occur in water utilities across rural NSW, and rural Australia. THM concentrations in the chloraminated systems in Sydney and the one rural utility were generally lower than the chlorinated water supply systems. We found considerable variation in THM’s between and within utilities associated with extended drought periods experienced during our study, with some utilities showing decreases in chloroform and BDCM and increases in DBCM. While we generally found higher concentration of THM during the warmer compared to cooler months the overarching influence of drought on THMs makes it difficult to identify the principal drivers. Unfortunately our data lacked detailed data on NOM, algae and catchment characteristics and inflows which may help understand the possible drivers for these variations. The Australian Drinking Water Guidelines are currently under review and at this time no change has been proposed to the guideline value of 250 mg/L for THM4 (NHMRC, 2009). The Guidelines note that a high concentration of THM4 is a good indicator that other DBPs may be present. However data on the occurrence, nature and concentrations of other DBPs are scarce due to the lack of comprehensive survey data in NSW and Australia. We recommend improved monitoring and reporting of DBP’s and the collection of good data on catchment characteristics and treatment to enable the assessment of contributing factors to elevated concentrations of DBPs, especially in rural Australia.
Competing financial interests None identified.
Acknowledgements The authors would like to acknowledge the following organisations and individuals for their work on this manuscript: Dr Mark Angles, Dr Peter Cox, Dr Vicky Whiffin, Mr David Holland from Sydney Water Corporation and Adam Lovell from Water Services Association of Australia for expert advice on the Sydney/Illawarra water utility. Mr Bruce Cole and Ms Pam O’Donoghue from Hunter Water Corporation for expert advice on the Hunter water utility, Mr Peter Littlejohns from the Sydney Catchment Authority for expert advice on drought effects in Sydney catchments in NSW. We also thank Dr Nel Glass and Dr Stephen Kermode from Southern Cross University, Ms Therese Dunn and Mr Paul Houlder for their valuable contributions to this manuscript. We wish to acknowledge the support from the Australian Research Council Linage Grant (LP0348628) and the Network for Spatially Integrated Social Science. This work is part of a PhD thesis by Richard Summerhayes.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Appendix. Supplementary material Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2011.08.045.
references
APHA-AWWA-WEF, 1992. Standard Methods for the Examination of Water and Waste Water, American Water Works Association. Water Environment Federation, Washington DC. APHA-AWWA-WPCF, 1998. Standard Methods for the Examination of Water and Waste Water, P. 1134, American Water Works Association. Water Pollution Control Federation, Washington DC. ABS, 2008. Year Book Australia, 2007. Australian Bureau of Statistics (ABS), Australian Government Publishers, Canberra, Australia. Bates, B.C., Kundzewicz, Z.W., Wu, S., Palutikof, J.P. (Eds.), 2008. Climate Change and Water. IPCC Secretariat, Geneva. Bond, N.R., Lake, P.S., Arthington, A.H., 2008. The impacts of drought on freshwater ecosystems: an Australian perspective. Hydrobiologia 600, 3e16. Bureau Of Meteorology, 2010. Climate Data Online. Australian Government, Canberra. URL: http://www.bom.gov.au/climate/ data/ (last accessed December 2010). Bougeard, C.M.M., Goslan, E.H., Jefferson, B., Parsons, S.A., 2010. Comparison of the disinfection by-product formation potential of treated waters exposed to chlorine and monochloramine. Water Research 44, 729e740. Centre for Epidemiology and Research, 2002. NSW Health Survey 2002. NSW Department of Health, Sydney. Charrois, J.W.A., Graham, D., Hrudey, S.E., Froese, K.L., 2004. Disinfection By-Products in small Alberta community drinking-water supplies. Journal of Toxicology and Environmental Health, Part A 67, 1797e1803. Chen, C., Zhang, X.-J., Zhu, L.-X., Liu, J., He, W.-J., Han, H.-D., 2008. Disinfection by-products and their precursors in a water treatment plant in North China: seasonal changes and fraction analysis. Science of the Total Environment 397 (1e3), 140e147. Chowdhury, S., Champagne, P., McLellan, P.J., 2009. Models for predicting disinfection byproduct (DBP) formation in drinking waters: a chronological review. Science of the Total Environment 407 (14), 4189e4206. Cox, P., Fisher, I., Kastl, G., Jegatheesan, V., Warnecke, M., Angles, M., Bustamante, H., Chiffings, T., Hawkins, P., 2003. Sydney 1998: lessons from a drinking water crisis. Journal of the American Water Works Association 95 (5), 147e161. Dore, M.H.I., 2005. Climate change and changes in global precipitation patterns: what do we know? Environment International 31 (8), 1167e1181. Eikebrokk, B., Vogt, R.D., Liltved, H., 2004. NOM increase in Northern European source waters: discussion of possible causes and impacts on coagulation/contact filtration processes. Water Science and Water Technology: Water Supply 44 (4), 47e54. Galal-Gorchev, H., 1996. Chlorine in water disinfection. Pure and Applied Chemistry 68 (9), 1731e1736. Garrido, S.E., Fonseca, M.G., 2010. Speciation and Kinetics of trihalomethane formation in drinking water in Mexico. Ground Water Monitoring & Remediation 30 (1), 77e84. Grellier, J., Bennett, J., Patelarou, E., Smith, R., Toledano, M., Rushton, L., Briggs, D., Nieuwenhuijsen, M., 2010. Exposure to disinfection by-products and Adverse birth outcomes related
5725
to Fetal Growth and Prematurity e A Systematic review and Meta-analysis. Epidemiology 21 (3), 300e313. Hamilton, D.P., Schladow, G., Fisher, I.H., 1995. Controlling the indirect effects of flow diversions on water quality in an Australian reservoir. Environment International 21 (5), 583e590. Hoehn, R.C., Randall, C.W., 1979. Viruses, organics, and other health-related Constituents of the Occoquan reservoir: Part I trihalomethanes, Pesticides, and Metals. Criteria and Standards Div., USEPA, Washington D.C. Hrudey, S.E., 2009. Chlorination disinfection by-products, public health risk tradeoffs and me. Water Research 43 (8), 2057e2092. HWC, 2004. Hunter Water Catchment Report 2003/4. Hunter Water Corporation (HWC), Newcastle Australia. Hurst, A.M., Edwards, M.J., Chipps, M., Jefferson, B., Parsons, S.A., 2004. The impact of rainstorm events on coagulation and clarifier performance in potable water treatment. Science of the Total Environment 321 (1e3), 219e230. Kampioti, A.A., Stephanou, E.G., 2002. The impact of bromide on the formation of neutral and acidic disinfection by-products (DBPs) in Mediterranean chlorinated drinking water. Water Research 36 (10), 2596e2606. Keegan, T., Whitaker, H., Nieuwenhuijsen, M.J., Toledano, M.B., Elliott, P., et al., 2001. Use of routinely collected data on trihalomethane in drinking water for epidemiological purposes. Occupational & Environmental Medicine 58 (7), 447e452. Krasner, S.W., McGuire, M.J., Jacangelo, J.G., et al., 1989. The occurrence of disinfection by-products in US drinking water. Journal of American Water Works Association 81, 41e53. Krasner, S.W., Sclimenti, M.J., Means, E.G., 1994. Quality degradation: implication for DBP formation. Journal of American Water Works Association 86 (6), 34e47. McGuire, M.J., Meadow, R.G., 1988. AWWARF trihalomethane survey. Journal of American Water Works Association 80, 61e68. McGuire, M.J., McLain, J.L., Obolensky, A., 2002. Information Collection Rule Data Analysis. AWWA Research Foundation, USA. Murphy, B.F., Timbal, B., 2008. A review of recent climate variability and climate change in southeastern Australia. International Journal of Climatology 28, 859e879. National Cancer Institute, 1976. Report on the Carcinogenesis Bioassay of Chloroform. National Cancer Institute, Bethesda, MD. NHMRC, 2004. Australian Drinking Water Guidelines 6. National Health & Medical Research Council. Australian Government, Canberra. NHMRC, 2009. Draft Australian Drinking Water Guidelines. National Health & Medical Research Council (NHMRC), Canberra Australia. NSW Department of Primary Industries, 2010. Drought. NSW Department of Primary Industries. NSW Government, Orange. URL: (Last accessed December 2010). NSW Health, 2000. Drinking Water Monitoring Program. NSW Health Department, pp. 1e39. Richardson, S., 2005. New disinfection by-product issues: emerging DBPs and alternative routes of exposure. Global NEST Journal 7 (1), 43e60. Richardson, S.D., Plewa, M.J., Wagner, E.D., Schoeny, R., DeMarini, D.M., 2007. Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection byproducts in drinking water: a review and roadmap for research. Mutation Research 636, 178e242. Rizzo, L., Selcuk, H., Nikolaou, A., Belgiorno, V., Bekbolet, M., Meric, S., 2005. Formation of chlorinated organics in drinking water of Istanbul (Turkey) and Salerno (Italy). Global NEST Journal 7 (1), 95e105. Rook, J.J., 1974. Formation of haloforms during chlorination of natural water. Water Treatment Exam 23, 234e243.
5726
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 1 5 e5 7 2 6
Simpson, K.L., Hayes, K.P., 1998. Drinking water disinfection byproducts: an Australian perspective. Water Research 32, 1522e1528. Stein, P., 2000. The great Sydney water crisis of 1998. Water, Air and Soil Pollution 123, 419e436. Soh, Y., Roddick, F., van Leeuwen, J., 2008. The future of water in Australia: the potential effects of climate change and ozone depletion on Australian water quality, quantity and treatability. Environmentalist 28, 158e165. SWC, 2002. Sydney Water Annual Report. Sydney Water Corporation (SWC), Sydney, Australia. USEPA, 1979. National Interim Primary Drinking Water Regulations; Control of Trihalomethanes in Drinking Water. Environmental Protection Agency, U.S.
USEPA, 2008. Test Methods for Evaluating Solid Waste, Physical. Chemical Methods: SW-846. Environmental Protection Agency, U.S. Whitaker, H., Nieuwenhuijsen, M.J., Best, N., Fawell, J., Gowers, A., Elliot, P., 2003. Description of trihalomethane levels in three UK water suppliers. Journal of Exposure Analysis and Environmental Epidemiology 13, 17e23. Williams, D.T., LeBel, G.L., Benoit, F.M., 1995. A National Survey of Chlorinated Disinfection By-Products in Canadian Drinking Water. Health Canada, Ottawa. Williamson, C.E., Morris, D.P., Pace, M.L., Olson, O.G., 1999. Dissolved organic Carbon and Nutrients as Regulators of Lake ecosystems: resurrection of a more Integrated Paradigm. Limnology and Oceanography 44 (3), 795e803.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
X-ray absorption and photoelectron spectroscopic study of the association of As(III) with nanoparticulate FeS and FeS-coated sand Young-Soo Han a,1, Hoon Y. Jeong b, Avery H. Demond a, Kim F. Hayes a,* a b
Department of Civil and Environmental Engineering, University of Michigan, Ann Arbor, MI 48109, United States Department of Geological Sciences, Pusan National University, Busan 609-735, South Korea
article info
abstract
Article history:
Iron sulfide (FeS) has been demonstrated to have a high removal capacity for arsenic (As) in
Received 7 February 2011
reducing environments. However, FeS may be present as a coating, rather than in nano-
Received in revised form
particulate form, in both natural and engineered systems. Frequently, the removal capacity
12 August 2011
of coatings may be different than that of nanoparticulates in batch systems. To assess the
Accepted 15 August 2011
differences in removal mechanisms between nanoparticulate FeS and FeS present as
Available online 26 August 2011
a coating, the solid phase products from the reaction of As(III) with FeS-coated sand and with suspensions of nanoparticulate (NP) FeS were determined using x-ray absorption
Keywords:
spectroscopy and x-ray photoelectron spectroscopy. In reaction with NP FeS at pH 5, As(III)
Arsenic
was reduced to As(II) to form realgar (AsS), while at pH 9, As(III) adsorbed as an As(III)
Sorption
thioarsenite species. In contrast, in the FeS-coated sand system, As(III) formed the solid
Mackinawite
phase orpiment (As2S3) at pH 5, but adsorbed as an As(III) arsenite species at pH 9. These
Redox
different solid reaction products are attributed to differences in FeS concentration and the
XAS
resultant redox (pe) differences in the FeS-coated sand system versus suspensions of NP
XPS
FeS. These results point to the importance of accounting for differences in concentration
Coatings
and redox when making inferences for coatings based on batch suspension studies. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The high removal capacity of mackinawite (FeS) for arsenic in reducing environments has been demonstrated in nanoparticulate suspensions (Wolthers et al., 2005; Gallegos et al., 2007). However, FeS often exists as a coating, both in natural and engineered environments. For example, mackinawite is found as a coating on the surfaces of minerals comprising soils and sediments (Rickard and Morse, 2005). Zero valent iron (ZVI), emplaced as a permeable reactive barrier (PRB)
material, develops an FeS coating under reducing conditions in the field (Beak and Wilkin, 2009). Recently, a procedure was developed to coat nanoparticulate (NP) FeS on a natural sand to create a PRB material for treating arsenic-contaminated groundwater (Han et al., 2011). One consequence of having FeS as a coating is that its effective concentration (g-FeS/L) is much lower than in batch reactor studies due to the practical limit of the amount of solid that can be added (e.g., 1e10 g/L solid suspensions results in 1e10 g/L for NP FeS compared to 4e40 mg/L for FeS-coated sand). Gallegos et al. (2008) recently
* Corresponding author. 1351 Beal Avenue, Department of Civil and Environmental Engineering, MI, United States. Tel.: þ1 734 763 9661; fax: þ1 734 763 2275. E-mail addresses: [email protected] (Y.-S. Han), [email protected] (H.Y. Jeong), [email protected] (A.H. Demond), ford@ umich.edu (K.F. Hayes). 1 Present address: Lawrence Berkeley National Laboratory, 1 Cyclotron Rd, Berkeley, CA, United States. Tel.: þ1 510 486 6950. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.026
5728
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
measured and modeled the redox potential of varying concentrations of mackinawite suspensions (from 0.01 to 10 gFeS/L) and illustrated that higher concentrations of mackinawite resulted in a lower redox potential. This decrease in redox potential with increasing concentrations of FeS was attributed to a surface-mediated reduction reaction with mackinawite. Given the importance of redox (and pH) in controlling the principal arsenic solid phases that may form in AseFeeS systems (O’Day et al., 2004), and the possible differences in redox conditions of NP FeS versus FeS coatings, this study was undertaken to assess whether the mechanisms of uptake of As(III) established from studies in NP FeS suspensions are representative of those occurring in FeScoated sand systems. In order to successfully predict the performance of FeS coatings, the difference in uptake mechanisms of such coatings relative to NP suspensions needs to be evaluated. Using suspensions of synthetic or isolated sulfide minerals, laboratory studies have previously established mechanisms of arsenic removal by various iron sulfides such as troilite, pyrite and mackinawite (Moore et al., 1988; Farquhar et al., 2002; Bostick and Fendorf, 2003; Wolthers et al., 2005; Gallegos et al., 2007; Jeong et al., 2010). As(III) sorption on troilite and pyrite surfaces was characterized as a FeAsS-like surface precipitation (Bostick and Fendorf, 2003). Farquhar et al. (2002) demonstrated that mackinawite is more efficient than other iron-oxide phases or pyrite in removing As(III), by adsorption of outer-sphere surface complexes and the formation of poorly crystalline arsenic sulfide precipitates. A disordered NP mackinawite, reacted within 1 h after formation, removed 0.012 mol As(III)/mol FeS at neutral pH by the formation of outer-sphere surface complexes on the mackinawite surface (Wolthers et al., 2005). In another study, nanoparticulate mackinawite that had been aged three days showed an As(III) removal capacity of 0.16, 0.018 and 0.004 mol As(III)/mol FeS at pH 5, 7, and 9, respectively (Han et al., 2011). The precipitation of realgar (AsS) was found to be responsible for the high As(III) removal observed at pH 5, while the formation of a thioarsenite surface species were identified at pH 9, based on x-ray absorption spectroscopy and thermodynamic calculations (Gallegos et al., 2007, 2008), and later supported by x-ray photoelectron spectroscopy and high resolution electron transmission microscopy (Renock et al., 2009). Yet, with the exception of the work reported in Gallegos et al. (2008), the redox state in those laboratory studies was not specifically controlled or reported. Even though the association of As(III) with NP mackinawite has been intensively studied by several research groups (Farquhar et al., 2002; Wolthers et al., 2005; Gallegos et al., 2007; Jeong et al., 2010), the association of As(III) with FeS coatings has not been studied. Yet, such coatings result in different As(III) uptake (Han et al., 2011). Therefore, to assess the potential impact caused by FeS attachment as a coating, this study characterized the solid phase reaction products of As(III) reacted with NP FeS and FeS-coated sand using x-ray absorption spectroscopy (XAS) and x-ray photoelectron spectroscopy (XPS). This comparison provides an opportunity to evaluate if laboratory-generated results with suspensions of NP FeS reasonably represent the As(III) uptake reactions of FeS-coated sand, and by implication, FeS coatings on other
minerals found environments.
in
natural
or
2.
Materials and methods
2.1.
Preparing FeS-coated sand
engineered
anoxic
Wedron 510 silica sand (Wedron Silica Co., Wedron, IL) with a geometric mean size of 0.15e0.22 mm was used as the substrate for the FeS coating. A batch of sand was washed several times with Milli-Q water, soaked in Milli-Q water overnight with mild shaking, and then dried at ambient temperature (around 25 C). FeS was synthesized inside an anaerobic chamber maintained at a 5% H2/95% N2 atmosphere by mixing 2.0 L of a 0.57 M FeCl2 (Fisher Chemical) with 1.2 L of 1.1 M Na2S solution (Butler and Hayes, 1998). As soon as the two chemicals were mixed, a black nanoparticulate suspension of mackinawite particles was obtained. The precipitated particles were stirred for three days, rinsed with distilled water to remove excessive salt and sulfide ions, and then separated from the supernatant using centrifugation. To coat the sand with FeS, freeze-dried mackinawite particles were resuspended to form a 2 g/L FeS suspension. The suspension was adjusted to a pH value of 5.5 and mixed with the sand for three days using an end-over-end rotator, at which point a clear supernatant was obtained. The FeS-coated sand was separated from the solution, dried in an anaerobic glovebox and stored in an air-tight container in the glovebox. The amount of FeS coating was determined to be 1.42 105 M FeS/g sand (1.24 103 g FeS/g sand) using an acid-extraction method. More details of the FeS coating methodology, and physical and chemical characterization can be found in Han et al. (2011).
2.2.
Spectroscopy sample preparation
X-ray absorption spectroscopy (XAS) and x-ray photoelectron spectroscopy (XPS) were used to determine the solid phase arsenic oxidation state and the relative proportions of different As species in As(III)-reacted NP FeS and FeS-coated sand samples. For the reaction of As(III) with FeS, 300 mL of a 5 g/L NP FeS suspension was placed in 400 mL glass reactors that allowed for the continuous measurement of pH in a closed system. The pH of a 5 g/L suspension was adjusted to pH 5 or 9 using HCl, and an aliquot of a 1.33 M NaAsO2 stock solution was added to achieve a concentration of 1.33 102 M As(III). The pH was monitored over a two-day equilibrium period and adjusted as necessary with acid to maintain the pH at 5 or 9. For the reaction of FeS-coated sand with As(III), 416 g/ L of FeS-coated sand (equivalent to 0.5 g/L FeS based on 1.24 103 g FeS/g sand) were reacted with a 1.33 103 M As(III) solution in 50 mL polypropylene tubes and mixed by an end-over-end rotating mixer for two days. Since the pH of this system could not be easily monitored continuously given the smaller volume reactor, multiple samples were prepared and a sample that gave a pH of 5 or 9 after two days of equilibration was selected for the spectroscopic analysis. The redox potential for the pH 5 samples was measured using an oxidation-reduction potential (ORP) combination platinum
5729
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
2.3.
X-ray absorption spectroscopy (XAS)
At SSRL, right before the XAS analysis, the wet paste samples were loaded, inside an anaerobic glove box, onto sample holders wrapped with a double layer of Kapton tape. XAS spectra were collected at SSRL on beamline 10-2 (3 GeV, w100 mA of maximum current) using an unfocused beam with a beam size of 2.0 20.0 mm at the sample holder. Arsenic K-edge XAS spectra were obtained using a Si(220) double-crystal monochromator with a 13-element solid-state Ge-array fluorescence detector or Lytle detector. The sample chamber was continuously purged with He gas to avoid potential oxidation. Based on a comparison of spectra, no oxidation occurred during the data collection. XAS spectra were also collected for reference compounds such as metallic arsenic (As(0)), amorphous AsS, amorphous As2S3 dissolved As(III), and dissolved As(V). All arsenic reference compounds were purchased from Alfa Aesar (Lancaster, UK). The XAS spectra were analyzed using the program package SixPACK (Webb, 2002). Individual spectra were first averaged, and the background absorbance was subsequently removed by a linear fit through the pre-edge region. X-ray absorption near-edge structure (XANES) spectra (e.g., 11,86011,890 eV) were obtained by normalizing the fluorescence signal to the edge jump height. The absorption edges (i.e., inflection energies) of XANES spectra were determined to compare the oxidation state of arsenic between the samples and reference compounds. Extended x-ray absorption fine structure (EXAFS) spectra were also obtained by fitting a quadratic spline function above the edge. EXAFS spectra were normalized using a Victoreen polynomial function and then transformed from ˚ 1) using E0 ¼ 11,885 eV. The resultant energy (eV) to k space (A EXAFS functions (c(k)) were weighted by k3 to amplify the higher k region, and Fourier-transformed to produce radial ˚ 1. structural functions (RSF) in R space over k ¼ 3.5e11.5 A 3 Structural parameters were obtained by fitting k -weighted EXAFS functions with the phase and amplitude functions derived from FEFF 8 (Ankudinov et al., 1998). The amplitudereduction factor (S2o ¼ 0.92) was optimized from the fitting of the reference compound spectra and kept constant for all EXAFS analysis. The Debye-Waller factors (s2) were also fixed based on the similarity between the sample spectra and the reference compound spectra or the optimization among the sample spectra to reduce the degrees of freedom during the fitting. Coordination number (N ), interatomic distance (R), and energy shift (DE0) were allowed to vary. The optimal fitting was obtained by minimizing the goodness of fit parameter (Rf).
2.4.
X-ray photoelectron spectroscopy (XPS)
Kratos Axis Ultra X-ray photoelectron spectrometer was used to examine the chemical composition and oxidation state of As species sorbed on NP FeS and FeS-coated sand. The reference compounds used for the XPS analyses were As(0), arsenic(II) sulfide, arsenic(III) sulfide, NaAsO2 and Na2HAsO4$7H2O, all purchased from Alfa Aesar (Lancaster, UK). The reference compounds and As(III) reacted samples were mounted on a sample bar in an anaerobic glove box and transferred using an air-tight container filled with 95% N2/5% H2 gas mixture to minimize the oxygen contact with the sample surface. The Al-Ka line (1486.6 eV) was used as the radiation source. Survey spectra were obtained using analyzer pass energies of 160 eV. Narrow XPS peaks were obtained primarily with a pass energy of 20 eV, but for the As(III)reacted FeS-coated sand samples, the higher pass energy of 160 eV was needed due to a low As loading. Energies were corrected for charging effects using the reference peak of adventitious carbon C ls with a binding energy of 284.6 eV. Raw spectra were smoothed using a linear base line for 1s peaks and a Shirley base line for 2p peaks, respectively, and then fitted using a GaussianeLorentzian function. To estimate the standard deviation of each of the component’s contribution to the overall XPS spectrum in the fitting procedure, Monte-Carlo analysis (CasaXPS, Casa Software Ltd., UK) was used. This program applies artificial noise to a spectrum and
8.8
Normalized absorbance
electrode with a Ag/AgCl reference (ColeeParmer). The measured potentials (in mV) were corrected for the standard hydrogen electrode (SHE) and temperature and converted to pe by multiplying by 0.0169. The samples for XAS analyses were filtered using 0.22 mm nylon filters and the wet filtered paste of particles was transferred into air-tight, crimp-sealed serum vials and then shipped to Stanford Synchrotron Radiation Lightsource (SSRL) (Stanford, CA) for analysis. For the XPS analysis, the samples were filtered, freeze-dried, crimp-sealed and stored in an anaerobic chamber until the analysis.
(j) (i)
6.8
(h) (g)
4.8
(f)
2.8
(e) (d) (c)
0.8
(b) (a)
-1.2 11860
11870
11880
11890
E (eV) Fig. 1 e Arsenic K-edge XANES spectra of (a) As(0) (grey), (b) FeAsS (green), (c) AsS (blue), (d) As2S3 (pink), (e) 5 g/L NP FeS reacted with 1.33 3 10L2 M As(III) for 2 days at pH 5, (f) 416 g/L FeS-coated sand reacted with 1.33 3 10L3 M As(III) for 2 days at pH 5, (g) 5 g/L NP FeS reacted with 1.33 3 10L2 M As(III) for 2 days at pH 9, (h) 416 g/L FeScoated sand reacted with 1.33 3 10L3 M As(III) for 2 days at pH 9, (i) dissolved NaAsO2 (yellow) and (j) dissolved Na2HAsO4$7H2O (red). The absorption edges correspond to the first derivative maxima of XANES spectra. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
5730
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
calculates an error matrix to give the variance of each fit based on the fitting constraints used.
3.
Results and discussion
3.1.
Arsenic loading
The supernatants of the samples used for the spectroscopic analyses were analyzed using ICP-MS to calculate the arsenic loading on the solid phase. NP FeS removed 99% (at pH 5) and 20% (at pH 9) of the initial 1.33 102 M As(III) present in solution, and FeS-coated sand removed 63% (at pH 5) and 23% (at pH 9) of the initial 1.33 103 M As(III). These data demonstrate that substantially higher As(III) sequestration occurs at pH 5 where the solubility of arsenic sulfides is lower than that of mackinawite. In terms of removal capacity (mg-As(III)/g-FeS), the FeS-coated sand had approximately 70% and 400% of the capacity of the NP FeS at pH 5 and pH 9, respectively, determined based on the results reported in Han et al. (2011).
3.2.
XAS analysis
The XAS spectra were subjected to both XANES and EXAFS analyses. While the oxidation state of arsenic can be obtained from XANES analysis, structural parameters such as interatomic distance (R) and coordination number (N ) on the near coordination environment around arsenic are gleaned from EXAFS analysis. In XANES spectra (Fig. 1), the absorption edges (i.e., inflection energies) of the samples are compared with those of reference compounds. While higher absorption edge energy is indicative of a higher oxidation state of arsenic, lower
absorption edge energy corresponds to a lower oxidation state. The absorption edge energies of As(0), arsenopyrite, AsS, As2S3, dissolved As(III), and dissolved As(V) were 11866.7, 11867.0, 11868.1, 11869.0, 11870.9, and 11874.4 eV, respectively. At pH 5, the NP FeS system reacted with As(III) had an absorption edge energy of 11868.4 eV, only slightly higher than that of AsS, indicating that the dominant oxidation state of As in the NP FeS sample was þII. The As(III)-reacted FeS-coated sand had an absorption energy of 11869.1 eV, close to that of As2S3, suggesting the formation of As2S3. At pH 9, the NP FeS reacted with As(III) had an absorption energy of 11868.06 eV, slightly lower than that of As2S3 but much lower than that of dissolved As(III), indicating the possible formation of thioarsenite species. The As(III)-reacted FeS-coated sand at pH 9 had an absorption energy of 11870.87 eV, close to that of dissolved As(III), suggesting the surface complexation of arsenite species. The EXAFS spectra and corresponding Fourier transforms of the experimental samples and reference compounds are compared in Fig. 2. The fitting results of the EXAFS analysis are shown in Table 1. For the NP FeS sample at pH 5, the first coordination shell around As is characterized by the AseS interaction, with the coordination number (NAseS) of 2.1 at ˚ , in good agreement with that of the AsS a distance of 2.27 A ˚ ). Also, the second referemce compound (NAseS ¼ 2.0 at 2.26 A coordination shell for the NP FeS sample is characterized by ˚ , with a coordination number the AseAs interaction at 3.49 A (NAseAs ¼ 0.95) twice that of the AsS reference compound ˚ ). Compared with the AsS reference (NAseAs ¼ 0.41 at 3.50 A compound, the NP FeS sample had a stronger second shell feature and a slightly higher absorption energy, indicating that another As phase, in addition to AsS, may form in the NP FeS system. Previously, Bostick et al. (2003) proposed a surface
Fig. 2 e k3-weighted arsenic K-edge EXAFS spectra (k3c(k)) and their FT transforms for (a) As(0) (grey), (b) FeAsS (green), (c) AsS (blue), (d) As2S3 (pink), (e) 5 g/L NP FeS reacted with 1.33 3 10L2 M As(III) for 2 days at pH 5, (f) 416 g/L FeS-coated sand reacted with 1.33 3 10L3 M As(III) for 2 days at pH 5, (g) 5 g/L NP FeS reacted with 1.33 3 10L2 M As(III) for 2 days at pH 9, (h) 416 g/L FeS-coated sand reacted with 1.33 3 10L3 M As(III) for 2 days at pH 9, (i) dissolved NaAsO2 (yellow) and (j) dissolved Na2HAsO4$7H2O (red). Solid lines are the experimental data; dashed lines are the numerical fits. The peak positions in the ˚ ). (For interpretation of Fourier transform (FT) are uncorrected for phase shift as indicated by the x-axis notation of R D D(A the references to colour in this figure legend, the reader is referred to the web version of this article.)
5731
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
Table 1 e EXAFS fit results for nanoparticulate FeS and FeS-coated sand reacted with As(III) at pH 5 and 9, and As model compounds. EXAFS fita
Pair N As(0)
AseAs AseAs
1.1
Crystallographic data 2
Reference
˚) R(A
˚ 2)b s (A
N
2.50
0.0058
3 3
2.5 3.13
O‘Day et al., 2004
2 1 2.5 1
2.24 2.57 3.44e3.51 3.41e3.52
Farquhar et al., 2002
3 1 3 2.5
2.24e2.31 3.19 3.22e3.57 3.52e3.64
Farquhar et al., 2002
˚) R(A
DE0 ¼ 6.95 eV, Rf ¼ 0.065 AsS
AseS AseAs AseAs AseS
2.0
2.26
0.003
0.41
3.50
0.006
AseS AseAs AseS AseAs
3.0 0.37
DE0 ¼ 9.80 eV, Rf ¼ 0.061 As2S3
2.28 3.54
0.0045 0.006
DE0 ¼ 7.75 eV, Rf ¼ 0.047 pH 5 NP-FeS
AseS AseAs
2.1 2.27 0.003 0.95 3.49 0.006 DE0 ¼ 8.24 eV, Rf ¼ 0.0412
This study
pH 9 NP-FeS
As(III)-O AseS AseAs (As(0)) AseAs (adsorbed)
0.69 1.96 1.10 1.73 DE0 ¼ -8.84
1.77 0.0045 2.25 0.0045 2.55 0.008 3.50 0.006 eV, Rf ¼ 0.0205
This study
pH 5 FeS coated-sand
AseS AseAs
2.9 2.26 0.0045 0.31 3.66 0.006 DE0 ¼ 10.09 eV, Rf ¼ 0.0974
This study
pH 9 FeS coated-sand
As(III)eO
2.88 1.78 0.0045 DE0 ¼ 2.78 eV, Rf ¼ 0.2119
This study
As(III)aq
As(III)eO
3.0 1.76 0.0045 DE0 ¼ 7.90 eV, Rf ¼ 0.069
3c
1.78c
Wolthers et al., 2005
As(V)aq
As(V)eO
4.0 1.69 0.0025 DE0 ¼ 5.01 eV, Rf ¼ 0.024
4c
1.69c
Yamauchi and Fowler, 1994
N: coordination number; R: interatomic distance; DE0: energy shift; Rf: goodness of fit parameter. a The amplitude-reduction factor (S2o) was set at 0.92. b The Debye-Waller factors (s2) were fixed during the numerical fit. c Structural data were obtained from EXAFS analysis.
precipitate in the form of trimeric arsenic sulfide for As(III) ˚) sorption by PbS and ZnS. The AseAs bonding distance (w3.6 A ˚ in their study is close to the value of 3.50 A observed here. Thus, the formation of surface precipitates as thioarsenites may explain the observed differences of the As coordination in the AsS reference compound and in the NP FeS system. The first coordination shell of the FeS-coated sand system at pH 5 is characterized by the AseS interaction with the ˚, coordination number (NAseS) of 2.9 at a distance of 2.26 A consistent with the coordination chemistry of As2S3 ˚ ). Unlike the NP FeS system, both the FeS(NAseS ¼ 3.0 at 2.28 A coated sand system and the As2S3 reference compound show ˚ very weak second coordination shells (NAseAs ¼ 0.31 at 3.66 A ˚ for for the FeS-coated sand system and NAseAs ¼ 0.37 at 3.54 A
spectrum compared to the pH 5 system. The best fit resulted from the inclusion of an AseS coordination shell (NAseS of 1.96 ˚ ) and an AseAs coordination shell at a distance of 2.25 A ˚ ), consistent with the (NAseAs of 1.73 at a distance of 3.50 A presence of surface precipitates or thioarsenites surface clusters (Jeong et al., 2010). A minor contribution of AseAs ˚ also bonding with NAseAs of 1.10 at a distance of 2.55 A improved the fit, indicating the possible formation of As(0) (Jeong et al., 2010). In contrast, the FeS-coated sand sample at pH 9 reacted with 1.33 102 M As(III) showed a first coordination shell comprised of As(III)-O with NAseO of 2.88 at ˚ , suggesting the surface complexation of a distance of 1.78 A arsenite species in the FeS-coated sand system at pH 9.
the As2S3 reference compound). Taken together, the formation of As2S3 is mainly responsible for As(III) uptake in the FeScoated sand system. At pH 9, the NP FeS system reacted with 1.33 102 M As(III) exhibited different characteristics in its EXAFS
3.3.
XPS analysis
Fig. 3 shows the As 3d spectra of arsenic reference compounds and NP FeS and FeS-coated sand samples reacted with As(III) (See Table 2 for fitting results). Each surface species in the
5732
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
Table 2 e XPS fit results nanoparticulate FeS reacted with As(III) at pH 5 and 9, and As model compounds using a low pass energy of 20 eV illustrated in Fig. 3. Percent peak area of each component
BE (g) (f) (e) (d) (c) (b) (a)
AseAs
As(II)eS
As(III)eS
As(III)eO
As(V)eO
41.8 (1) e e 19.78 e e e 100.0
42.8 (1) e e e 100.0 e 100.0 e
43.2 (1) e e e e 87.28 e e
43.0 (1) 15.76 89.52 56.62 e e e e
44.5 (1) 84.24 10.48 29.22 e 12.72 e e
BE: Binding energy (FWHM).
Fig. 3 e XPS spectra of As 3d peaks for (a) As(0), (b) AsS, (c) As2S3, (d) 5 g/L NP FeS reacted with 1.33 3 10L2 M As(III) for 2 days at pH 5, (e) 5 g/L NP FeS reacted with 1.33 3 10L2 M As(III) for 2 days at pH 9, (f) NaAsO2 salt and (g) Na2HAsO4$7H2O salt using a low pass energy of 20 eV (a, b, c, f and g are model compounds, and d and e are samples. The spectrum of sample (e) is enlarged for better viewing.).
As(3d ) spectrum is fitted with a doublet representing the spinorbit splitting of the As 3d5/2 and As 3d3/2 peaks. A higher binding energy is indicative of a higher oxidation state of arsenic and a lower binding energy corresponds to a lower oxidation state. The positions of the As 3d5/2 peaks for (a) As(0), (b) arsenic(II) sulfide, (c) arsenic(III) sulfide, (f) NaAsO2 and (g) Na2HAsO4$7H2O were determined to be 41.8, 42.8, 43.1, 43.5, and 44.5 eV, respectively. These peaks for the model
compounds reveal that each compound consisted of at least 83% of the expected dominant oxidation state, with minor contributions of other oxidation states of arsenic. The fullwidth-at-half-maximum (FWHM) values for all the model compounds were constrained to be 1.0 eV for low pass energy scans and 2.2 eV for the high pass energy scans. The binding energies determined by the As 3d5/2 peak positions and FWHM values were used to determine the predominant As oxidation states of the NP FeS samples reacted with 1.33 102 M As(III) and FeS-coated sand reacted with 1.33 103 M As(III) at pH 5 and 9. At pH 5, the As 3d5/2 peaks of NP FeS with 1.33 102 M As(III) were sharp, indicating that one prominent arsenic solid species precipitated as a result of the reaction between NP FeS and As(III) at pH 5 (Fig. 3 (d)). The peaks were well fitted with a doublet of As 3d5/2 and As 3d3/2 at binding energies of 42.8 and 43.5 eV, respectively. These peak positions indicate that the As oxidation state is primarily As(II) and that realgar is formed. This is consistent with the XAS analysis above (Table 1) and previous thermodynamic modeling of a similar system (Gallegos, 2007; Gallegos et al., 2008). As no AsS diffraction patterns were observed using XRD (Figure S-1 in Supplementary Material), the precipitated AsS is likely in an amorphous form. Renock et al. (2009) also reported a realgarlike precipitate as the primary product formed by As reaction with mackinawite under similar experimental conditions. However, their arsenic XPS peaks showed much broader features, representing a mixture of different species of various arsenic oxidation states, whereas in this study, the peak represents primarily a single contribution of As(II)-S. At pH 9, the intensity of the As 3d5/2 peaks is much smaller than that at pH 5 because only 20% out of the 1.33 102 M As(III) was solid-phase associated, compared with almost 100% at pH 5. The low solid phase arsenic resulted in a weak, broad peak, but with the center of the As 3d5/2 peak shifted to a higher binding energy and extending throughout the binding energy ranges of As(II)-S, As(III)-S and As(III)-O, possibly indicating the presence of a mixture of various thioarsenite species. Interpreted in this manner, the result is consistent with the EXAFS analysis. The XPS spectra for As(III) reacted with FeS-coated sand at pH 5 and 9 are shown in Fig. 4, with the peak fitting parameters
Relative Intensity
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
results indicate the presence of primarily As(III)-O on the FeScoated sand surface, suggesting arsenite adsorption. These results support those obtained with XAS. The broadening of the peak in the low and high binding region of the spectrum is likely caused by the use of higher pass energy of 160 eV. This is confirmed by the measurements presented in Figure S-2 (Supplementary Material), which show that the FWHM values increase with increasing pass energy. However, the highest peak positions still were located at the same binding energy. Consequently, the identification of the peak position using the higher pass energy for the samples with weak element intensity appears to be valid when used in this qualitative manner to verify the peak maximum.
(b)
(a)
50 49 48 47 46 45 44 43 42 41 40 39 38 37
Binding Energy (eV) As(5+): As(V)-O As(3+): As(III)-S Base line Measured point
As(3+): As(III)-O As(0): As-As Fitted envelope
Fig. 4 e XPS spectra of As 3d peaks for FeS-coated sand reacted with As(III) reacted with 1.33 3 10L3 M As(III) for 2 days at (a) pH 5 and (b) pH 9 using a high pass energy of 160 eV.
given in Table 3. Since the As(III) loading on the FeS-coated sand after reaction with As(III) was too low for sufficient spectral peak quality for quantitative analysis, a pass energy of 160 eV was used instead of the more widely used pass energy of 20 eV. A pass energy of 160 eV is sometimes used to identify an oxidation state of an element of interest (Su and Puls, 2008). In this study, the XPS As 3d spectrum obtained using a pass energy of 160 eV was used to qualitatively compare the peak position of the As(III) reacted FeS-coated sand samples. Fig. 4 shows the XPS analysis of the FeScoated sand sample reacted with As(III) at pH 5 and indicates that the solid phase reaction product is primarily As(III)S, consistent with the formation of orpiment. At pH 9, the
Table 3 e EXAFS fit results for nanoparticulate FeS and FeS-coated sand reacted with As(III) at pH 5 and 9, and As model compounds using a high pass energy of 160 eV illustrated in Fig. 4. Percent peak area of each component
BE (b) (a)
5733
AseAs
As(III)eS
As(III)eO
As(V)eO
41.8 (2.2) 11.37 31.05
43.2 (2.2) e 48.63
43.5 (2.2) 82.63 e
44.5 (2.2) e 20.31
BE: Binding energy (FWHM).
3.4. Comparison of NP FeS and FeS-coated sand systems reacted with As(III) NP FeS and FeS-coated sand reacted with As(III) at pH 5 resulted primarily in the precipitation of arsenic sulfides but different oxidation states were detected in each of the arsenic solid phases. The XANES and EXAFS analyses indicated the formation of AsS and a thioarsenite surface precipitate for As(III) uptake in the FeS system at pH 5, while the formation of As2S3 occurred in the FeS-coated sand system at pH 5. The XPS results also support the formation of these different arsenic sulfide solids in each system. In NP FeS systems, more reduced conditions favor realgar precipitation, while more oxidizing conditions favor orpiment precipitation. To assess the redox conditions in the NP FeS and Fe-coated sand, the redox potential was measured. In the systems consisting of 5 g FeS/L reacted with 1.33 102 M As(III) and 416 g FeS-coated sand/L (0.5 g-FeS/L) reacted with 1.33 103 M As(III), both at pH 5, the redox potential was measured to be 326 mV and 246 mV, respectively. The pe values calculated from these measured redox potentials are 2.15 for the NP FeS system and 0.91 for the FeS-coated sand system. A thermodynamic simulation of the FeeAseSeH2O batch system predicted amorphous As2S3 precipitation at pe values ranging from 1.0 to around 2.0, and AsS precipitation at pe values ranging from 2.0 to 5.0 (Gallegos et al., 2008). The different redox conditions were postulated to result from FeS surface redox reactions and the different total amounts of FeS in each system. The measured pe values in the present study match well with the thermodynamically-predicted arsenic sulfide species and those identified by XPS and XAS. In both the NP FeS and FeS-coated sand systems, the primary As(III) uptake process is precipitation at pH 5, while at pH 9, uptake is controlled by adsorption and/or surface precipitation reactions. However, the type of surface species formed is different. In the NP FeS system, thioarsenite surface species formed, but in the FeS-coated sand system, arsenite species were detected. Our previous study of FeS-coated sand showed that a small amount of an iron oxyhydroxide phase formed in the FeS-coated sand system due to the more oxidizing conditions that prevailed in this system (Han et al., 2011). The presence of this phase may also explain why arsenite surface complexes form on FeS-coated sand whereas thioarsenite species form on NP FeS and why there is greater adsorption of As(III) by FeS-coated sand at pH 9 (Han et al., 2011).
5734
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
3.5. Implications of As speciation differences in NP FeS and FeS-coated sand The different species that form when As(III) is reacted with FeS-coated sand compared to with NP FeS are thought to result from the different redox conditions: (1) the higher redox potential of the FeS-coated sand system at pH 5 results in the precipitation of orpiment rather than realgar as in the NP FeS system, and (2) the higher redox potential of the FeScoated sand system at pH 9 leads to the formation of a small amount of iron oxyhydroxide solid and the adsorption of arsenite rather than the adsorption of thiosarsenite as in the NP FeS system. The overall implication of these results is that the reaction products of arsenic with FeS in natural systems may be different from those determined using concentrated laboratory batch systems of NP FeS, in which artificially enhanced reducing conditions may result from concentrated NP FeS suspensions. This is a reminder that both pH and pe conditions need to be assessed in AseFeeS systems if results are to be applied more generally. Given that the extent of As removal by FeS-coated sand is similar to that by NP FeS, despite the differences in uptake mechanisms, these results support that FeS-coated sand may be a suitable sorbent material for in situ removal of arsenic in PRB applications where the emplacement of NP FeS is impractical.
4.
Conclusions
XAS and XPS analyses showed differences in the solid phase species that form during the reaction of As(III) with NP FeS and FeS-coated sand. At pH 5, the reaction of As(III) with NP FeS results primarily in the precipitation of realgar but the reaction with FeScoated sand results in orpiment precipitation. At pH 9, As(III) is removed through adsorption or surface precipitation of thioarsenite species by NP FeS but through the adsorption of arsenite surface species by FeS-coated sand. The differences in As speciation in the NP FeS system compared to the FeS-coated sand system are attributed to the lower redox potentials in the NP FeS suspensions. The mechanisms of As uptake determined in batch reactor systems with concentrated NP FeS may not be indicative of the uptake mechanisms that occur in systems in which FeS is present as a coating.
Acknowledgements The authors gratefully acknowledge the assistance of Tom Yavaraski (Department of Civil and Environmental Engineering, University of Michigan) for his help in developing the analytical methods for As and Fe analyses, and Udo Becker and Devon Renock (Department of Geological Sciences, University of Michigan) for their guidance on the XPS data collection and analysis. Portions of this research were carried out at the Stanford Synchrotron Radiation
Lightsource, a Directorate of SLAC National Accelerator Laboratory and an Office of Science User Facility operated for the U.S. Department of Energy Office of Science by Stanford University. The SSRL Structural Molecular Biology Program is supported by the DOE Office of Biological and Environmental Research, and by the National Institutes of Health, National Center for Research Resources, Biomedical Technology Program (P41RR001209). This research was supported by the Strategic Environmental Research and Development Program (SERDP) under Department of Defense, Department of Army, Contract Number W912HQ-04-C-0035. This paper has not been subject to agency review; it, therefore, does not necessarily reflect the sponsor’s view, and no official endorsement should be inferred.
Appendix. Supplementary Material Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.026.
references
Ankudinov, A.L., Ravel, B., Rehr, J.J., Conradson, S.D., 1998. Realspace multiple-scattering calculation and interpretation of xray-absorption near-edge structure. Physical Review BCondensed Matter 58 (12), 7565e7576. Beak, D.G., Wilkin, R.T., 2009. Performance of a zerovalent iron reactive barrier for the treatment of arsenic in groundwater: Part 2. Geochemical modeling and solid phase studies. Journal of Contaminant Hydrology 106 (1e2), 15e28. Bostick, B.C., Fendorf, S., 2003. Arsenite sorption on troilite (FeS) and pyrite (FeS2). Geochimica et Cosmochimica Acta 67 (5), 909e921. Bostick, B.C., Fendorf, S., Manning, B.A., 2003. Arsenite adsorption on galena (PbS) and sphalerite (ZnS). Geochimica et Cosmochimica Acta 67 (5), 895e907. Butler, E.C., Hayes, K.F., 1998. Effects of solution composition and pH on the reductive dechlorination of hexachloroethane by iron sulfide. Environmental Science and Technology 32 (9), 1276e1284. Farquhar, M.L., Charnock, J.M., Livens, F.R., Vaughan, D.J., 2002. Mechanisms of arsenic uptake from aqueous solution by interaction with goethite, lepidocrocite, mackinawite, and pyrite: an X-ray absorption spectroscopy study. Environmental Science and Technology 36 (8), 1757e1762. Gallegos, T.J., 2007. Sequestration of As(III) by synthetic mackinawite under anoxic conditions. Ph.D. Dissertation, The University of Michigan, Ann Arbor, MI. Gallegos, T.J., Hyun, S.P., Hayes, K.F., 2007. Spectroscopic investigation of the uptake of arsenite from solution by synthetic mackinawite. Environmental Science and Technology 41, 7781e7786. Gallegos, T.J., Han, Y.S., Hayes, K.F., 2008. Model predictions of realgar precipitation by reaction of As(III) with synthetic mackinawite under anoxic conditions. Environmental Science and Technology 42, 9338e9343. Han, Y.-S., Gallegos, T.J., Demond, A.H., Hayes, K.F., 2011. FeScoated sand for removal of arsenic(III) under anaerobic conditions in permeable reactive barriers. Water Research 45, 593e604. Jeong, H.Y., Han, Y.S., Hayes, K.F., 2010. X-ray absorption and xray photoelectron spectroscopic study of arsenic mobilization
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 2 7 e5 7 3 5
during mackinawite (FeS) oxidation. Environmental Science and Technology 44 (3), 955e961. Moore, J.N., Ficklin, W.H., Johns, C., 1988. Partitioning of arsenic and metals in reducing sulfidic sediments. Environmental Science and Technology 22 (4), 432e437. O’Day, P.A., Vlassopoulos, D., Root, R., Rivera, N., 2004. The influence of sulfur and iron on dissolved arsenic concentrations in the shallow subsurface under changing redox conditions. Proceedings of the National Academy of Sciences of the United States of America 101 (38), 13703e13708. Renock, D., Gallegos, T., Utsunomiya, S., Hayes, K., Ewing, R.C., Becker, U., 2009. Chemical and structural characterization of As immobilization by nanoparticles of mackinawite (FeSm). Chemical Geology 268 (1e2), 116e125. Rickard, D., Morse, J.W., 2005. Acid volatile sulfide (AVS). Marine Chemistry 97 (3e4), 141e197.
5735
Su, C.M., Puls, R.W., 2008. Arsenate and arsenite sorption on magnetite: relations to groundwater arsenic treatment using zerovalent iron and natural attenuation. Water Air and Soil Pollution 193 (1e4), 65e78. Webb, S.M., 2002. Sam’s Interface for XAS Package (SixPACK). Stanford Synchrotron Radiation Laboratory, Menlo Park, CA. Wolthers, M., Charlet, L., Van der Weijden, C.H., Van der Linde, P. R., Rickard, D., 2005. Arsenic mobility in the ambient sulfidic environment: sorption of arsenic(V) and arsenic(III) onto disordered mackinawite. Geochimica et Cosmochimica Acta 69, 3483e3492. Yamauchi, H., Fowler, B.A., 1994. Toxicity and metabolism of inorganic and methylated arsenicals. In: Advances in Environmental Science and Technology, Arsenic in the Environment, Part II. Human Health and Ecosystem Effects. Wiley, New York, pp. 35e53.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Defluoridation of water via electrically controlled anion exchange by polyaniline modified electrode reactor Hao Cui, Qin Li, Yan Qian, Rong Tang, Hao An, Jianping Zhai* State Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing 210093, PR China
article info
abstract
Article history:
A polyaniline (PANI) modified electrode reactor was designed for fluoride removal from
Received 13 June 2011
aqueous solutions. The innovative concept behind the reactor design is that the uptake and
Received in revised form
elute of fluoride could be well controlled by modulating the potential of the PANI film. The
24 August 2011
maximum fluoride removal capacity of PANI is more than 20 mg/g at a positive voltage
Accepted 25 August 2011
based on the electrically controlled anion-exchange mechanism. The results of batch tests
Available online 1 September 2011
showed that terminal potential values had a major impact on fluoride removal by this PANI, with optimal removal occurring at 1.5 V. The fluoride removal capacity (qe) increased
Keywords:
rapidly within 5 min and reached equilibrium within 10 min, which indicated a rapid
Fluoride
removal velocity of fluoride by PANI under this condition. The applicability of defluor-
Electrochemical
idation using the PANI reactor to treat fluoride-contaminated tap water was also tested
Anion exchange
through flow cell breakthrough studies. At initial fluoride concentrations of 5 mg/L and
Polyaniline
10 mg/L, the breakthrough capacities were 20.08 mg/g and 19.24 mg/g, respectively.
Water treatment
Moreover, during the first half of the period before the breakthrough point, the fluoride concentration of the treated solution was below the WHO’s recommended levels (1.5 mg/L). The results of the five consecutive treatment-regeneration studies also showed that the PANI films could be reused. Taken together, these results implied that the electrically controlled anion exchange by the PANI-modified electrode reactor may be an effective technique for the removal of fluoride from water. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Fluoride contamination in drinking water is a serious problem in several parts of East Asia and South America (Guo et al., 2010; Armienta and Segovia, 2008; Wang and Cheng, 2001). The acceptable safe limit of fluoride recommended by the World Health Organization (WHO) is 1.5 mg/L (World Health Organization, 2004). Current methods used to remove fluoride from water can be divided into three categories, precipitation, membrane techniques and adsorption. Precipitation of fluoride with calcium and aluminum salts has been used to remove fluoride from
industrial wastewater. Precipitation treatment usually reduces high concentrations of fluoride to 2 mg/L (Fan et al., 2003). Membrane techniques such as reverse osmosis (Ndiayea et al., 2005), nanofiltration (Diawara, 2008), donnan dialysis (Durmaz et al., 2005), electrocoagulation (Zhu et al., 2007) and electrodialysis (Kabay et al., 2008) have been developed to effectively remove the fluoride from water. Adsorption is a cost-effective and extensively used method that has recently received a great deal of attention. Accordingly, many studies have been conducted to identify effective, low-cost adsorbents (Chen et al., 2011; Sivasankara et al., 2010). However, there is still a great demand for identification of environmentally friendly,
* Corresponding author. Tel./fax: þ86 25 8359 2903. E-mail address: [email protected] (J. Zhai). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.049
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
simple, and low-cost technologies for the removal of fluoride from drinking water. In recent years, there has been considerable interest in conducting polymers owing to their extensive potential applications in areas such as microelectronics, composite materials, catalysts, and chemical sensing. In addition, conducting polymers used as anion exchanger materials have shown new potential applications in water and wastewater treatment. Use of adsorbents such as polyaniline has been reported for the removal of heavy metal ions such as Hg(II) (Wang et al., 2009) and Cr(VI) from water (Reza, 2006). Recently, studies have been conducted to investigate the adsorption of fluoride onto polyanile (PANI) and polypyrrole (PPy) (Karthikeyan et al., 2009a,b). Karthikeyan et al. found that fluoride ions could be removed from aqueous solutions by PANI via doping of the polymer matrix. However, batch tests showed that the adsorption capacity of pure PANI for fluoride was relatively low, being 0.78 mg/g at pH 7.0 when a 50 mg/ 50 ml dose was used to treat water with initial fluoride ion concentrations of 2e10 mg/l. One of the most effective approaches to enhancement of the defluoridation capacities of adsorbents such as polyaniline is development of PANI-based composites. For example, the alumina composites of PANI and PPy exhibited an appreciable capacity for the removal of fluoride ions from water, with defluoridation capacities of 6.6 and 8.0 mg/g, respectively (Karthikeyan et al., 2009c). Chitosan composites of PANI and PPy were also found to enhance the amounts of fluoride ions adsorbed to 5.9 and 6.7 mg/g, respectively (Karthikeyan et al., 2011). Polyaniline-tamarind seed biomaterial powders were fabricated for adsorption of fluoride (1e10 mg/L solutions) and found to have an adsorption capacity of 4.8 mg/g (Subramanian and Ramalakshmi, 2010). In another study, a novel nanocomposite with magnetic properties combining both polypyrrole and Fe3O4 was prepared and used for defluoridation. The fluoride adsorption capacity of the PPy/Fe3O4 nanocomposites was found to range from 17.63 to 22.31 mg/g. However, use of conducting polymers as adsorbents in batch-test mode could not fully utilize their remarkable advantage of electroactivity. Therefore, a novel technique, electrically switched ion exchange (ESIX), was recently investigated and applied for the removal of toxic ions from aqueous solutions. In the ESIX technique, the electrochemical oxidation or reduction of the electroactive species is conducted by applying an anodic potential to the film concurrently with ion uptake and elution. In a previous study, Weidlich et al. (2005) used the ion exchanger ability of PPy to develop an electrochemically switchable ion exchanger for softening drinking water. Lin et al. (2006) investigated the perchlorate removal process based on ESIX in a conventional three-electrode system using polypyrrole deposited carbon nanotubes as the working electrode. They demonstrated that the redox switching of polypyrrole was accompanied by the exchange of perchlorate ions into or out of the polymer. We previously reported the electrically controlled anion-exchange process for the removal and release of perchlorate and chromium using poly(aniline-co-oaminophenol) (Zhang et al., 2009, 2010). Taken together, the results of these previous studies indicate the potential for development of a green process for fluoride removal from water using polyaniline-based materials.
5737
In this study, a polyaniline (PANI) modified electrode reactor was designed and evaluated as an electrically switched ion exchanger for the removal of fluoride from aqueous solution. A series of batch tests were also conducted to identify the key parameters for electrochemical defluoridation by PANI (such as pH and terminal potential), and flow experiments were employed to optimize future reactor design.
2.
Experimental section
2.1.
Chemicals
Fluoride (NaF, >99%) was purchased from Fluka. Aniline was purified by distillation under reduced pressure, kept in the dark, and stored in a refrigerator until use. Ultrapure water (18.25 MU cm) was employed to prepare all solutions used in this study, and the solution pH was adjusted by the addition of HCl and NaOH (guaranteed grade). Other chemicals were of analytical grade and used without further purification. Tap water contaminated with fluoride was prepared by the addition of NaF to original tap water to obtain final F concentration of 5 and 10 mg L1. The pH was adjusted to 5.0 by the addition of HCl. The water quality of the original tap water is shown in Table 1.
2.2.
Preparation of the working electrodes
Electrochemical removal of fluoride was conducted in a 0.10 mol/L NaCl supporting electrolyte solution. Accordingly, the electrochemical polymerization of aniline was conducted in an HCl medium (200 mL, 1.0 M) containing 0.2 M distilled aniline. Two pieces of indium tin oxide (ITO) conductive glass were employed as the anode and cathode. Polyaniline was deposited onto the anode at 2.0 V for 5 min. After synthesis, the PANI films were washed with 0.1 M HCl solution to remove
Table 1 e Quality of original tap water. Item Cations Naþ Kþ Ca2þ Mg2þ Zn2þ Al3þ NHþ 4 Anions Fluoride Chloride Nitrate Sulfate Phosphate pH Electric conductivity Free chlorine Combined chlorine a N.D., not detected.
Value 20 mg/L 3.5 mg/L 15.6 mg/L 2.4 mg/L 0.06 mg/L 0.01 mg/L 2 mg/L 0.2 mg/L 38 mg/L 1.2 mg/L 15 mg/L N.D.a 7.2 252 mS/cm 0.3 0.05
5738
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
the unreacted aniline, after which they were rinsed with ultrapure water, and finally dried using a hairdryer. TG/DTA analysis showed that the average weight of PANI on the ITO conductive glass was about 15 0.5 mg/cm2.
2.3. Electrode reactor and electrochemical experiments procedure Two series of defluoridation experiments, batch tests and flow experiments were performed. The batch tests were conducted in reservoir tank containing 500 mL of fluoride solution, and 0.5 mL aliquots of the solution were sampled from the reservoir tank at intervals of 1 or 2 min. The effects of pH (pH 4e9) and terminal potential (0.5, 1, 1.5, 2.5, 3 V) on electrochemical defluoridation by PANI were investigated. Following treatment, the PANI films were regenerated at a negative voltage of 1.0 V in 100 mL of HCl solution (0.10 M) for 5 min. The polyaniline (PANI) modified electrochemical defluoridation reactor was employed for the flow experiments. During the flow experiments, potentiostatic anion exchange at a terminal voltage of 1.5 V was applied. Tap water contaminated with fluoride was passed through the flow cell of the electrode reactor once. The treated tap water flowing from the outlet was then sampled for chemical analysis. All runs of both the batch test and flow experiments were conducted at a velocity of 10 mL/min while maintaining room temperature at 25 C. Fig. 1 shows the schematic view of the experimental setup and the structure of the electrochemical defluoridation reactor. Two pieces of indium tin oxide (ITO) conductive glass modified with and without PANI film were employed to act as the anode and cathode, respectively. An insulated 5 mm thick silicon rubber spacer was inserted between the anode and cathode, and the effective surface area of each electrode was 3 5 cm2. Therefore, the working volume of the flow cell was 7.5 mL. A DC voltage-stabilized power supply (purchased from Shanghai Liyou Electrification Co., Ltd., China) was connected to both the anode and cathode to control the potential. The fluoride solution was fed into the flow cell from bottom to head using a syringe pump (Baoding Longer Precision Pump Co., Ltd., China).
2.4.
Analysis
The fluoride concentrations were evaluated by ion chromatography (IC) using a Dionex ICS-2000 ion chromatography with a conductivity detector and a 25 mL sample loop. The eluent was a 30 mM potassium hydroxide buffer solution and the total separation time was 10 min. The typical experimental error was lower than 5% for all experimental results. The XPS measurements were conducted on a Thermo ESCALAB 250 spectrometer with an Al Ka X-ray source (1486.6 eV). XPS analyses of three different samples (raw PANI, PANI after treatment and regenerated PANI) were conducted to demonstrate the electrically controlled anion exchange of the PANI films. All the binding energies were referenced to C1s neutral carbon peak at 284.6 eV. To assess the possible risk of aniline release from the electrode, the concentration of aniline in the effluent following treatment of fluoride-spiked tap water was analyzed by high-
Fig. 1 e A schematic view of the experimental setup of the PANI-modified electrolytic defluoridation indium tin oxide (ITO) conductive glass slide electrode for flow experiments (a), and structure of the electrochemical defluoride flow cell (b).
performance liquid chromatography (HPLC) as previously described (Tanaka et al., 2009).
3.
Results and discussion
3.1.
Electrically controlled anion-exchange mechanism
Fig. 2A shows three images of PANI films, PANI synthesized in the HCl solution in the oxidized state, PANI doped in a solution containing fluoride at pH 5 (PANI-F), and de-doped PANI in the reduced state (PANI-R). It is well known that polyaniline can exist in three different discrete oxidation states at the molecular level, both in the doped and undoped forms (Chiang and Macdiarmid, 1986). Accordingly, the color of the polyaniline
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
Fig. 2 e Images of PANI films (A), XPS spectra of survey scan (B) and XPS spectra of N1s (C) for PANI films: original PANI film (a); PANI film after electrochemical fluoride removal (b); PANI film after regeneration (c).
5739
film will change with the potential of the polyaniline film and the pH value of the electrolyte (Yuan et al., 1995). As shown in Fig. 2A, the raw deep green PANI film (Fig. 2A-a) turned violet blue after fluoride uptake (Fig. 2A-b) and turned light green after regeneration (Fig. 2A-c). To confirm the presence of fluoride ions in the doped PANI, XPS was used to characterize the three samples. Curves aec in Fig. 2B are the XPS spectra for PANI, PANI-F and PANI-R, respectively. It is obvious that, in the PANI film after fluoride removal (curve b), the F1s peak appeared with the decrease of the Cl2p peak when compared to the raw PANI film (curve a). Moreover, the F1s peak disappeared after regeneration of the PANI film (curve c). The contents of F1s, Cl1s and the three major species of N1s are listed in Table 2. Based on the data shown in Table 2, the Cl content in PANI films decreased from 4.68% to 0.15% after fluoride uptake, while the F content increased from 0 to 2.96%. The level of F dropped after regeneration of the PANI film. These results suggest that the uptake and elute of F ions can be controlled by modulating the potential of the PANI film. To further explain the anion-exchange mechanism, this study focused on the molecular conformational change of nitrogen atoms. The XPS core level spectra of N1s for the three samples are presented in Fig. 2C. The broad peaks in the spectra indicate the existence of several different structures in the PANI films. Since the nitrogen atoms in PANI film could be classified into three groups, nitride (]Ne), amine (eNHe), and doped imine (eNHþe), the N1s spectra are reasonably deconvoluted into three Gaussian peaks at 399.72, >400.00 and>402.00 eV, respectively (Chen et al., 2002). The nitride/imine ratio of the original PANI film obtained in HCl was 0.72, which is smaller than the theoretical value of 1 for the 50% intrinsically oxidized emeraldine. These findings imply that some imine species were converted into amine ones in the HCl solution. For the PANI film after fluoride removal, the ratio increased to 2.99, indicating that PANI was less protonated (Kang et al., 1998). The results also revealed that the binding energy of various N1s increased to some extents after fluoride removal. These findings could be attributed to the morphological changes in amine groups and the secondary doping of F ions on PANI chains, i.e. changing from compact coils to expanded coils (Reghu et al., 1993), which resulted in longer bond length and higher bond energy. Another interesting observation is that the nitride/imine ratio decreased to 0.26 after regeneration, indicating that the PANI film had been regenerated well and could be used repeatedly. Based on the XPS results, we proposed an ESIX mechanism of defluoridation on PANI films. The schematic illustration of the polymerization of PANI and fluoride ion intake and elution with the oxidation and reduction of the polyaniline film is shown in Fig. 3. First, polymerization of free aniline monomers occurred in HCl solution, during which time Cl was doped into the PANI film. Second, in the electrochemical anionexchange process, the loss of electrons from PANI chains was caused by morphological changes in the nitrogen atoms; therefore, the secondary doping of PANI by F ions occurred at a suitable anodic voltage. A portion of the Cl ions in the PANI film was then eluted into the solution via anion exchange, and F ions were extracted from the solution by secondary doping,
5740
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
Table 2 e Distribution of Cl, F, and N species on PANI at different stages.a Element PANI PANI-F PANI-Rb
Peak BE/eV Atom % Peak BE/eV Atom % Peak BE/eV Atom %
Cl1s
F1s
-N]
eNHe
199.47 4.36 199.73 0.15 200.45 2.74
685.99 N.D.c 685.99 2.96 685.99 N.D.c
399.39 0.93 399.57 2.06 399.39 0.43
400.48 5.56 400.83 5.55 400.57 6.94
eNHþ$e 402.38 0.94 402.46 0.49 402.43 1.28
403.72 0.36 403.98 0.2 403.75 0.37
a BE, binding energy. b PANI-R, PANI regenerated. c N.D., not detected.
thereby decreasing the fluoride concentration. Finally, the used PANI film could be regenerated in HCl solution at a certain negative voltage for F ion dedoping, which enabled the PANI electrode to be used repeatedly. A similar process was observed in a study of perchlorate removal by PPy (Lin et al., 2006).
3.2.
Effect of terminal potential on fluoride removal
Time-course changes in fluoride uptake by PANI films during potentiostatic batch tests conducted different terminal potentials are shown in Fig. 4. In this study, the fluoride removal capacity (qe) can be calculated by: qe ¼
F 0 F e V wS
where [F]0 is the initial fluoride concentration (mg/L), and [F]e is the equilibrium fluoride concentration (mg/L), V is the solution volume (0.1 L), w is the average weight of PANI on the ITO conductive glass (15 0.5 mg/cm2), and S is the effective surface area of each electrode (3 5 cm2). In the present study, the removal of fluoride by adsorption in the PANI film was 0.21 mg/g without terminal potential, which was similar to the results of Karthikeyan’s study (Karthikeyan et al., 2009a). As shown in Fig. 4, the external voltage resulted in a significantly enlarged fluoride removal capacity (qe) of the PANI films based on the electrically controlled anion-exchange mechanism. These results
showed that terminal potential values had a major impact on fluoride removal by PANI. The optimal removal was observed around 1.5 V at pH 6.0 and the amount of the fluoride removal reached 20.49 mg/g, which was higher than that observed for adsorbents. The fluoride removal capacity (qe) increased rapidly within 5 min and then continued to increase at lower speeds until equilibrium was reached at 10 min. These findings indicated a rapid removal of fluoride by PANI under these conditions. Both lower and higher potential leads to a slower reaction velocity and smaller fluoride removal capacity. Within the range of 0.5 Ve1.5 V, the increase in potential means the higher the current density, which leads to a higher fluoride removal capacity. However, due to the overoxidation reaction of PANI films, the terminal potentials above 2.0 V results in decreased the fluoride removal efficiency.
3.3.
Effect of pH on fluoride removal
For quantitative analysis of the fluoride removal capability of the PANI-modified electrode, a series of potentiostatic electrochemical experiments were performed at pH 5e9 in a solution containing 50 mg/L fluoride and 0.1 M NaCl. The effects of pH on the electrochemical fluoride removal capacities are presented in Fig. 5. The solution pH values had a significant impact on the fluoride removal by the PANI film, with apparent inhibition in the conditions at pH 7. For example, the final fluoride removal rate was more than 90% at
Fig. 3 e Schematic illustration of the polymerization of PANI and fluoride ion intake and elution with the oxidation and reduction of PANI film.
5741
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
3.4.
20
q e (m g /g )
15
10 0V 0.5 V 1.0 V 1.5 V 2.5 V 3.5 V
5
0 0
5
10 Time (min)
15
20
Fig. 4 e Effect of terminal potential on fluoride uptake by PANI films during potentiostatic treatment (C0 [ 10 mg/L; I [ 0.1 M NaCl; pH [ 6.0; volume of solution, 500 ml). The amount of PANI was approximated to 225 mg for each test.
pH 4e6, but decreased abruptly to 48.8% at pH 7 and to less than 25% at pH 8 and 9. The reason for the aforementioned phenomenon is that the electrochemical activity of PANI under acidic solutions is much better than under alkaline conditions. It is because of there is a much higher doped imine/nitride ratio (i.e. eNHþ$e/ ]Ne ratio) in the structure of the polyaniline chain. PANI contains nitrogen atoms that are easy to protonate, and the protonation of eNHþ$e/]N, which are both pH sensitive, is simultaneously achieved at a pKa in the range of 5.5e8 (Slim et al., 2008). Because the concentrations of Hþ played an important role in the doping process, the treatment of fluoride-contaminated water should be conducted under acidic conditions.
Treatment of fluoride-contaminated tap water
Fluoride retention in the defloration reactor at a potential of 1.5 V was investigated under different fluoride concentrations (5 and 10 mg/L). The characteristic shape and position of the breakthrough curves (BTC) on the time axis for different concentrations are shown in Fig. 6. As shown in Fig. 6, the values of Ct/C0 were very low and the breakthrough curves were smooth in the beginning. During this period, the fluoride concentration of treated tap water was below the WHO recommended level of 1.0 mg/L. As more fluoride-contaminated tap water passed through the flow cell, the BTC became sharper, rapidly reaching the breakthrough point. A sharp BTC was observed at higher feed fluoride concentrations because the increase in initial fluoride concentration caused the PANI films to become saturated with the fluoride ion earlier in the experiment. At initial concentrations of 5 mg/L and 10 mg/L, the breakthrough capacities were 20.08 mg/g and 19.24 mg/g, respectively, which indicates high fluoride removal by the reactor. To evaluate the reuse value of the PANI films, the consecutive treatment-regeneration process was conducted three times. HCl solution at 0.1 mol/L of was used as the desorbing agent. Fig. 7 shows the characteristic shape and position of BTC for fresh PANI film, first regenerated PANI film and second regenerated PANI film. The fluoride removal capacity of fresh PANI film was slightly higher than that of the other films, while the fluoride removal capacity of the first and second regenerated samples were similar. Specifically, the breakthrough capacities of the fresh PANI, first regenerated PANI and second regenerated PANI were 20.07 mg/g, 18.79 mg/g and 18.35 mg/g, respectively. In addition, the defluoridation capacity of PANI for five treatment-regeneration cycles is illustrated in Table S-1 (Support Information), and all tests were conducted three
1.0 C0 = 10 mg/L C0 = 5 mg/L
0.8 0.015
20
qe (mg/g)
15
0.010
Ct /C0
C t/ C 0
0.6 0.4
0.005
0.2
10
0.000 0
0.0
5
10 15 20 25 Time (min)
5 pH = 4 pH = 6 pH = 8
0 0
2
4
6
8
pH = 5 pH = 7 pH = 9
10
12
Time (min) Fig. 5 e Effect of pH on fluoride removal during potentiostatic treatment (C0 [ 10 mg/L; I [ 0.1 M NaCl; volume of solution, 100 ml; terminal potential, 1.5 V).
0
20
40
60
80
100
120
140
160
180
Time (min)
14
Fig. 6 e Fluoride breakthrough continuous flow experiments at initial fluoride concentrations of 5 and 10 mg/L (I [ 0.1 M NaCl; pH [ 6.0; flow rate, 10 ml/min; terminal potential, 1.5 V). The broken line (Ce/C0 [ 1) represents the breakthrough point. The inset shows details during the first 25 min.
5742
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
times to eliminate the error range. The results revealed that the defluoridation capacity of PANI was still 89% at the fifth cycle, which suggests that the PANI-modified electrode possessed the potential for regeneration and reuse.
used for water with an initial fluoride concentration ranging from 5 to 35 mg/L, pH 5e7 and 25 C, which is similar to the conditions that PANI-modified electrode reactors are usually used under. However, the applied voltages in these systems are much higher than the voltage used in the PANI-modified electrode reactor developed in this study, which was 5e30 V (Emamjomeh and Sivakumar, 2009; Khatibikamal et al., 2010; Lo et al., 2007). These findings indicated that the ESIX technique reduces energy use. Several conducting polymers and their composites were recently investigated for the ability to adsorb fluoride. The adsorption capacity (Qe) and specific conditions such as equilibrium time (te), initial concentration (C0), optimum pH, temperature (T ) and regeneration method of these polymers are listed in Table 3. Because of the special dopingededoping capacity of conducting polymers such as PANI and PPy under acidic conditions, the optimum pH for fluoride removal by these adsorbents was less than 7, which is similar to the results of the present study. When compared to these conducting polymer adsorbents, the ESIX technique using the PANI electrode was also a good method for fluoride removal. By adding a certain terminal voltage to the PANI-modified electrode, the defluoridation capability of the purified PANI was significantly enhanced from 0.78 to about 20 mg/g. These findings indicated that the ESIX technique can effectively enlarge the anionexchange capability of the conducting polymers. Further studies using conducting polymer composites such as the PPy/ Fe3O4 nanocomposite as electrode materials to improve fluoride removal efficiency are still necessary.
3.5.
3.6.
1.0 0.06
0.8 C /C
0.04
0.6 C t /C 0
0.02
0.00
0.4
0
10
20
30 40 50 Time (min)
60
70
fresh PANI first regeneration second regeneration
0.2
0.0 0
20
40
60
80
100
Time (min) Fig. 7 e Comparison of fluoride removal using fresh PANI, first regenerated sample and second regenerated sample in continuous flow experiments (C0 [ 10 mg/L; I [ 0.1 M NaCl; pH [ 6.0; flow rate, 10 ml/min; terminal potential, 1.5 V). The inset shows the details during the first 70 min.
Comparison of ESIX technique with other methods
Using this method of fluoride removal via electrically controlled anion exchange, the uptake and elute of fluoride could be easily controlled by modulating the potential of the PANI film. It is interesting to note that the removal of fluoride is a rapid process, and the PANI electrode was easily regenerated. Since the PANImodified electrode reactor removes fluoride by electrochemically controlled ion exchange, it is necessary to compare it with other electrochemical methods, such as electrocoagulation. In the electrocoagulation unit, aluminum electrodes are usually applied to produce aluminum ions as a coagulant for the removal of fluoride. These electrocoagulation units are usually
Safety and durability assessment
The concentration of aniline in the treated fluoride-spiked tap water was analyzed by HPLC after successive reactioneregeneration cycles and no aniline was detected in the effluent during five adsorptioneregeneration cycles. However, it is important to note that there is still a risk of release of aniline in future applications of this method. Furthermore, this risk may increase as the service time and number of adsorptioneregeneration cycles increases, and some small PANI particles may even enter the effluent. To avoid such a risk, additional studies are needed to design more stable PANI-based composites for electrode coatings or to combine
Table 3 e Adsorption capacity of other typical adsorbents. Qe, mg/g
te, min
C0, mg/L
Optimum pH
T, C
PANI PPy PANI/alumina PPy/alumina PANI/chitosan PPy/chitosan PANI/TS
0.78 6.37 5.8 6.7 5.5 5.8 10.7
5 10 20 20 10 10 30
4 10 10 10 10 10 10
3 <7 3 3 3e4 3e4 unknown
30 30 30 30 30 30 30
0.1 M NH4OH Unknown Unknown Unknown Unknown Unknown Unknown
PPy/Fe3O4 nanocomposite PANI-modified electrode
22.31
20
100
6.5
25
2 M HCl
Subramanian and Ramalakshmi (2010) Bhaumik et al. (2011)
20
10
10
6
25
Negative voltage
This study
Adsorbent
Regeneration
Reference Karthikeyan et al. (2009a) Karthikeyan et al. (2009b) Karthikeyan et al. (2009c) Karthikeyan et al. (2011)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
this reactor with other techniques such as membrane techniques in practical application. Overall, this defluoridation method via electrically controlled anion exchange by a polyaniline modified electrode reactor provides an alternative technique for the removal of fluoride from aqueous solutions.
4.
Conclusions
Fluoride removal via electrically controlled anion exchange by a polyaniline modified electrode reactor was investigated in this study. Based on the results of XPS measurements of the PANI films and ion chromatography analysis of the solutions before and after fluoride removal, the fluoride ion uptake and elution was well controlled by modulation of the potential of the PANI-modified electrode. Batch tests and flow experiments showed that the PANI-modified electrode reactor had good fluoride removal capability (about 20 mg/g) in acidic solutions at a positive voltage of 1.5 V, giving a high removal efficiency for trace fluoride ions in contaminated tap water. Moreover, the doped fluoride on the PANI film could be effectively eluted at a negative voltage of 1.0 V in 0.10 M HCl solution, and the regenerated PANI films still had high fluoride uptake and elute efficiencies in the three-time consecutive treatment-regeneration studies. Therefore, the results of this study suggested that electrically controlled anion exchange by a polyaniline modified electrode reactor has great potential for the removal of fluoride from drinking water. Accordingly, further studies investigating application in this area are warranted.
Acknowledgment This work was supported by the Natural Science Foundation of China (Grants 51008154), foundation of State Key Laboratory of Pollution Control and Resource Reuse of China, and the Scientific Research Foundation of Graduate School of Nanjing University (Grants 2010CL07).
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.08.049.
references
Armienta, M.A., Segovia, N., 2008. Arsenic and fluoride in the groundwater of Mexico. Environmental Geochemistry and Health 30, 345e353. Bhaumik, M., Leswifi, T.Y., Maity, A., Srinivasu, V.V., Onyango, M.S., 2011. Removal of fluoride from aqueous solution by polypyrrole/ Fe3O4 magnetic nanocomposite. Journal of Hazardous Materials 186, 150e159. Chen, L., Wang, T.J., Wu, H.X., Jin, Y., Zhang, Y., Dou, X.M., 2011. Optimization of a FeeAleCe nano-adsorbent granulation process that used spray coating in a fluidized bed for fluoride removal from drinking water. Powder Technology 206, 291e296.
5743
Chen, Y.J., Kang, E.T., Neoh, K.G., 2002. Electroless polymerization of aniline on platinum and palladium surfaces. Applied Surface Science 185, 267e276. Chiang, J.C., Macdiarmid, A.G., 1986. Polyaniline e protonic acid doping of the emeraldine form to the metallic regime. Synthetic Metals 13, 193e205. Diawara, C.K., 2008. Nanofiltration process efficiency in water desalination. Separation and Purification Reviews 37, 303e325. Durmaz, F., Kara, H., Cengeloglu, Y., Ersoz, M., 2005. Fluoride removal by Donnan dialysis with anion exchange membranes. Desalination 177, 51e57. Emamjomeh, M.M., Sivakumar, M., 2009. Fluoride removal by a continuous flow electrocoagulation reactor. Journal of Environmental Management 90, 1204e1212. Fan, X., Parker, D.J., Smith, M.D., 2003. Adsorption kinetics of fluoride on low cost materials. Water Research 37, 4929e4937. Guo, Q.H., Wang, Y.X., Guo, Q.S., 2010. Hydrogeochemical genesis of groundwaters with abnormal fluoride concentrations from Zhongxiang City, Hubei Province, central China. Environmental Earth Sciences 60, 633e642. Kabay, N., Arar, O., Samatya, S., Yuksel, U., Yuksel, M., 2008. Separation of fluoride from aqueous solution by electrodialysis: effect of process parameters and other ionic species. Journal of Hazardous Materials 153, 107e113. Kang, E.T., Neoh, K.G., Tan, K.L., 1998. Polyaniline: a polymer with many interesting intrinsic redox states. Progress in Polymer Science 23, 277e324. Karthikeyan, M., Kumar, K.K.S., Elango, K.P., 2011. Batch sorption studies on the removal of fluoride ions from water using eco-friendly conducting polymer/bio-polymer composites. Desalination 267, 49e56. Karthikeyan, M., Satheeshkumar, K.K., Elango, K.R., 2009a. Defluoridation of water via doping of polyanilines. Journal of Hazardous Materials 163, 1026e1032. Karthikeyan, M., Satheeshkumar, K.K., Elango, K.P., 2009b. Removal of fluoride ions from aqueous solution by conducting polypyrrole. Journal of Hazardous Materials 167, 300e305. Karthikeyan, M., Satheeshkumar, K.K., Elango, K.P., 2009c. Conducting polymer/alumina composites as viable adsorbent for the removal of fluoride ions from aqueous solution. Journal of Fluorine Chemistry 130, 894e901. Khatibikamal, V., Torabian, A., Janpoor, F., Hoshyaripour, G., 2010. Fluoride removal from industrial wastewater using electrocoagulation and its adsorption kinetics. Journal of Hazardous Materials 179, 276e280. Lin, Y.H., Cui, X.L., Bontha, J., 2006. Electrically controlled anion exchange based on polypyrrole and carbon nanotubes nanocomposite for perchlorate removal. Environmental Science & Technology 40, 4004e4009. Lo, S.L., Hu, C.Y., Kuan, W.H., 2007. Simulation the kinetics of fluoride removal by electrocoagulation (EC) process using aluminum electrodes. Journal of Hazardous Materials 145, 180e185. Ndiayea, P.I., Moulin, P., Dominguez, L., Millet, J.C., Charbit, F., 2005. Removal of fluoride from electronic industrial effluent by RO membrane separation. Desalination 173, 25e32. Reghu, M., Cao, Y., Moses, D., Heeger, A.J., 1993. Counterioninduced processibility of polyaniline e transport at the metalinsulator boundary. Physical Review B 47, 1758e1764. Reza, A., 2006. Application of polyaniline and its composites for adsorption/recovery of chromium (VI) from aqueous solutions. Acta Chimica Slovenica 53, 88e94. Slim, C., Ktari, N., Cakara, D., Kanoufi, F., Combellas, C., 2008. Polyaniline films based ultramicroelectrodes sensitive to pH. Journal of Electroanalytical Chemistry 612, 53e62. Sivasankara, V., Ramachandramoorthy, T., Chandramohan, A., 2010. Fluoride removal from water using activated and MnO2coated Tamarind Fruit (Tamarindus indica) shell: batch and column studies. Journal of Hazardous Materials 177, 719e729.
5744
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 3 6 e5 7 4 4
Subramanian, E., Ramalakshmi, R. Dhana, 2010. Pristine, purified and polyaniline-coated tamarind seed (Tamarindus indica) biomaterial powders for defluoridation: synergism and enhancement in fluoride-adsorption by polyaniline coating. Journal of Scientific and Industrial Research 69, 621e628. Tanaka, T., Hachiyanagi, H., Yamamoto, N., Iijima, T., Kido, Y., Uyeda, M., Takahama, K., 2009. Biodegradation of endocrinedisrupting chemical aniline by microorganisms. Journal of Health Science 55, 625e630. Wang, G.X., Cheng, G.D., 2001. Fluoride distribution in water and the governing factors of environment in arid north-west China. Journal of Arid Environments 49, 601e614. Wang, J., Deng, B.L., Chen, H., Wang, X.R., Zheng, J.J., 2009. Removal of aqueous Hg (II) by polyaniline: sorption characteristics and mechanisms. Environmental Science & Technology 43, 5223e5228. Weidlich, C., Mangold, K.-M., Ju¨ttner, K., 2005. Continuous ion exchange process based on polypyrrole as an
electrochemically switchable ion exchanger. Electrochimica Acta 50, 5247e5254. World Health Organization (WHO), 2004. Guidelines for Drinkingwater Quality, third ed., vol. 1. WHO, Geneva. Yuan, R.K., Gu, Z.P., Yuan, H., 1995. Studies of controllable colorchange properties of polyaniline film. Synthetic Metals 69, 233e234. Zhang, Y., Li, Q., Tang, R., Hu, Q.C., Sun, L., Zhai, J.P., 2009. Electrocatalytic reduction of chromium by poly(aniline-co-oaminophenol): an efficient and recyclable way to remove Cr(VI) in wastewater. Applied Catalysis B: Environmental 92, 351e356. Zhang, Y., Mu, S.L., Deng, B.L., Zheng, J.Z., 2010. Electrochemical removal and release of perchlorate using poly(aniline-co-oaminophenol). Journal of Electroanalytical Chemistry 641, 1e6. Zhu, J., Zhao, H., Ni, J., 2007. Fluoride distribution in electrocoagulation defluoridation process. Separation and Purification Technology 56, 184e191.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Optimization of capacity and kinetics for a novel bio-based arsenic sorbent, TiO2-impregnated chitosan bead Sarah M. Miller a, Matthew L. Spaulding a, Julie B. Zimmerman a,b,* a b
Department of Chemical and Environmental Engineering, Yale University, United States School of Forestry and Environmental Studies, Yale University, United States
article info
abstract
Article history:
The optimization of TiO2-impregnated chitosan beads (TICB) as an arsenic adsorbent is
Received 2 May 2011
investigated to maximize the capacity and kinetics of arsenic removal. It has been previ-
Received in revised form
ously reported that TICB can 1) remove arsenite, 2) remove arsenate, and 3) oxidize arsenite
24 August 2011
to arsenate in the presence of UV light and oxygen. Herein, it is reported that adsorption
Accepted 25 August 2011
capacity for TICB is controlled by solution pH and TiO2 loading within the bead and
Available online 1 September 2011
enhanced with exposure to UV light. Solution pH is found to be a critical parameter, whereby arsenate is effectively removed below pH 7.25 and arsenite is effectively removed
Keywords:
below pH 9.2. A model to predict TICB capacity, based on TiO2 loading and solution pH, is
Arsenic
presented for arsenite, arsenate, and total arsenic in the presence of UV light. The rate of
Water
removal is increased with reductions in bead size and with exposure to UV light. Phosphate
Chitosan
is found to be a direct competitor with arsenate for adsorption sites on TICB, but other
TiO2
relevant common background groundwater ions do not compete with arsenate for
Bio-based
adsorption sites. TICB can be regenerated with weak NaOH and maintain full adsorption
Sustainable
capacity for at least three adsorption/desorption cycles. ª 2011 Published by Elsevier Ltd.
1.
Introduction
Over 120 million people in Bangladesh and India rely upon groundwater with elevated arsenic levels as their primary source of drinking water (Ratnaike, 2003). This “mass poisoning” has resulted in many negative health outcomes, including toxicity to the liver, skin, kidney, and cardiovascular system, as well as multiple cancers (Agros et al., 2010). As community scale infrastructure for water treatment is not likely in the near future, point-of-use technologies, most relying on adsorption, have been advocated for arsenic removal (Petrusevski et al., 2008). Despite promising lab results for many of these technologies, no technology has adequately demonstrated long-term effectiveness and sustainability in the field (Petrusevski et al., 2008).
Arsenic in groundwater primarily exists as arsenite (As(III)) and arsenate (As(V)) (Bhattacharya et al., 2007). Arsenite is up to 60 times more toxic than arsenate (Ratnaike, 2003; Tien et al., 2004). Because arsenite is uncharged at environmentally relevant pH, it is also more difficult to remove than arsenate, which is negatively charged at environmentally relevant pH. Metal oxide sorption technologies, usually based on Ti oxides, Fe oxides, or Al oxides, have the unique ability to remove both As(III) and As(V). This is a major advantage relative to technologies like ion exchange, which can only remove As(V) and require a pre-oxidation step to achieve total arsenic removal. The synthesis of TiO2-impregnated chitosan bead (TICB), a bio-based adsorbent with promising arsenic removal capacity, has been previously reported (Miller and
* Corresponding author. Department of Chemical and Environmental Engineering, Yale University, United States. Tel.: þ1 203 432 9703. E-mail address: [email protected] (J.B. Zimmerman). 0043-1354/$ e see front matter ª 2011 Published by Elsevier Ltd. doi:10.1016/j.watres.2011.08.040
5746
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
Zimmerman, 2010). Previously reported results were for TICB comprised of 70% chitosan and 30% TiO2 by weight (Miller and Zimmerman, 2010). Chitosan, a derivative of chitin, has a positive environmental and economic profile because it is renewable, biodegradable (Muzzarelli and Muzzarelli, 2005), and can be isolated from the millions of tons of shellfish waste generated globally per year (Gerente et al., 2007). TiO2 is a nontoxic nanopowder (Deebar et al., 2009) that has shown promise as filtration media in a point-of-entry arsenic removal system (Bang et al., 2005). Like pure TiO2, TICB has demonstrated removal of both arsenite and arsenate and has demonstrated the ability to oxidize arsenite to arsenate in the presence of UV light (Miller and Zimmerman, 2010). Although arsenic removal by TICB is less than TiO2 on a sorbent weight basis, TICB exceeds TiO2 sorption on a sorbent surface area basis (Dutta et al., 2004; Bang et al., 2005; Miller and Zimmerman, 2010), and has the added advantage of selfseparation. The objective of this study was to optimize the arsenic adsorption capacity and kinetics of TICB and develop a predictive model to guide system conditions for implementation. Parameters affecting capacity and kinetics, including solution pH, bead surface area, bead size, TiO2 loading in the bead, UV irradiation, and background ions in the water matrix, are investigated. Optimization of the useful lifetime of the TICB adsorbent is also considered by examining the potential for regeneration and reuse of TICB.
2.
Materials and methods
2.1.
Standards and reagents
Experiments were conducted with either As(III) (arsenite) or As(V) (arsenate) as indicated. Stock solutions of As(III) and As(V) were prepared by dissolving NaAsO2 (Sigma Aldrich) and Na2HAsO.47H2O (Fisher Scientific) in deionized water, respectively. Stock solutions (10,000 mg/L) and appropriate dilutions were prepared daily, immediately before use. Chitosan was purchased from TCI America. TiO2 (anatase nanopowder, 99.7% trace metals basis, <25 nm particle size) was purchased from Sigma Aldrich. All other reagents were of standard laboratory grade. HNO3 (Fisher Scientific, trace metal grade) and HCl solutions were prepared from concentrated stock solutions; NaOH solutions were prepared from pellets.
2.2.
Bead preparation and characterization
TiO2-impregnated chitosan beads were prepared as reported in (Miller and Zimmerman, 2010). In brief, chitosan was dissolved in 0.1 M HCl (1 g/60 mL). TiO2 (0.4242 g TiO2/1 g chitosan) was added and mixed with a magnetic stir bar until a homogenous solution was achieved. Unless otherwise noted, TICB is 30% TiO2 on mass basis. Syringes with 18G1 needles were filled with the homogenous solution and loaded into a syringe pump; bead size was adjusted by varying the needle gage (19G1, 22G1). The syringe pump discharged the homogenous solution into 0.1 M NaOH (20 mL solution/100 mL 0.1 M NaOH), resulting in beads. Beads were rinsed in
deionized water until filtrate reached pH 6. After drying for > 18 h, beads were collected and stored at room temperature in the dark. BET surface area analysis was performed by Micromeritics Analytical Services (Norcross, Georgia). Samples were weighed at room temperature after degassing at 45 C for 16 h. Gas adsorption analysis was conducted with krypton gas using a TriStar II 3020 surface area and porosity system. Bead porosity was measured with Hg intrusion by Micromeritics Analytical Services (Norcross, Georgia). Bead diameter was measured with a Marathon Electronic Digital Micrometer (0e25 mm). Reported diameter values are the average and standard deviation of 10 randomly selected beads from the same batch.
2.3.
Adsorption experiments
Duplicate samples were prepared in 50 mL polypropylene falcon tubes into which 40 mL arsenic solution (either arsenite or arsenate) and the specified amount of TICB was added. Batch adsorption experiments were conducted in a shaking incubator (VWR 1575R), where temperature was maintained at 25 C and samples were agitated at 150 rpm. Unless otherwise indicated, experiments were conducted for >185 h (Miller and Zimmerman, 2010). Irradiated samples were continuously exposed to an 8 W, 365 nm lamp (UVP, UVL-28 EL Series 8) (Ferguson et al., 2005) suspended 1.5 feet from samples in a closed incubator. In identical experimental conditions to the batch experiments, kinetics of As(III) and As(V) were measured in the presence and absence of UV light. These studies were conducted by sacrificing duplicate samples at specified time intervals for up to 317 h. pH experiments were conducted in the presence and absence of UV light, where [As]0 ¼ 500 mg/L pH adjustments were made with 16 M HNO3 or 1 M NaOH. Synthetic groundwater was prepared based on a procedure by (Leupin and Hug, 2005). MgCO3 (112.7 mg), CaCO3 (367 mg), and KH2PO4 (11.0 mg) were added to 900 mL of deionized water and stirred, resulting in pH 9.06. Under rapid mixing, CO2 gas was bubbled through this solution for 42 min, resulting in pH 5.08. A 10 mL solution of Na2SiO.39H2O (205.9 mg) dissolved in deionized water was added to the bulk solution and stirred, resulting in pH 5.39. Compressed air (house air filtered through granular activated carbon) was bubbled through the solution for 15 min, resulting in pH 6.93. 40 mL deionized water and 50 mL of freshly prepared 10 ppm As(III) stock solution was added to the solution to achieve 500 mg As (III)/L initial concentration and pH 7.19. Groundwater was collected from a tubewell at Amhiribad High School in West Bengal, India on December 17, 2010. The tubewell was flushed for 5 min prior to collection. Water composition was analyzed by EnviroCheck Laboratory in West Bengal and is as follows: 62 ppb As, 0.83 ppm Fe, 0.37 ppm Mn, 0.66 ppm P, 30.66 ppm Si, 7.5 ppm SO24 , 33.18 ppm Mg, 101.0 ppm Ca, 408.2 ppm HCO 3 , 48.04 ppm Cl, 2.78 ppm K, and 14.44 ppm Na. Tubewell water was transported to Yale University where batch experiments and analyses were performed as previously described.
5747
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
2.4.
Analyte concentration
To measure arsenic concentration, all samples were diluted to a maximum of 100 mg As/L and acidified with 16 M HNO3 to reach a final concentration of 1% HNO3. Samples with neat TiO2 nanopowder were filtered through a 0.45 mm filter prior to dilution. Measurement of As, Mg, Ca, Si, and P was conducted on a Perkin Elmer DRC-e ICP-MS. An internal Germanium standard was used, and quality control standards were analyzed every 10 samples to verify instrument performance. Three readings for each sample were performed and an average and standard deviation for each sample was reported. Detection limit was determined to be 0.25 mg As/L. Arsenic speciation was measured using HPLCICP-MS (Perkin Elmer DRC-e) following published procedures (Neubauer et al., 2004).
3.
Results and discussion
3.1. Equilibrium adsorption capacity of TICB for dissolved arsenic 3.1.1.
pH
Solution pH is a critical variable in metal oxide chemistorptive processes. Fig. 1 shows data for arsenite and arsenate removal by TICB without UV irradiation as a function of pH, where arsenite and arsenate both exist. These results, obtained by deliberately altering solution pH, are consistent with a hydroxide exchange mechanism. Above pH 7.25, the pzc of TICB (Miller and Zimmerman, 2010), electrostatic repulsion between arsenate and hydroxyl groups on the bead surface prevents chemisorption. Neuturally charged arsenite is effectively sorbed up to its pKa, 9.2, where it becomes a negative oxyanion. Differences between arsenate and arsenite removal at pH > 9.2 can be attributed to the fact that arsenate (H2AsO2 4 ) is more negatively charged than arsenite (H2AsO 3 ) in this pH
100
As(III)
80
% As removal
For desorption/resorption experiments, TICB (25 mg) was saturated with As(III) or As(V), as indicated, in a batch experiment without the presence of UV light, where [As]0 varied from 100e10,000 mg/L as indicated. Samples were removed after > 185 h in the shaking incubator (25 C, 150 rpm), and the aqueous solution was immediately decanted from the falcon tube and measured for arsenic concentration. Deionized water (10 mL) was added to the falcon tube to rinse the beads. After gentle agitation, the rinse water was decanted, and the beads, contained in uncapped falcon tubes, were allowed to dry in the fumehood for 24 h. Beads were then reweighed and transferred to a new 50 mL polypropylene falcon tube. 40 mL of NaOH (0.07 M, 0.67 M, 1.67 M) was added to these falcon tubes and placed in the shaking incubator, where temperature was maintained at 25 C and samples were agitated at 150 rpm for 24 h. NaOH was then decanted and measured for arsenic concentration. Deionized water (10 mL) was added to the falcon tube to rinse the beads. After gentle agitation, the rinse water was decanted and the beads, contained in uncapped falcon tubes, were allowed to dry in the fumehood for 24 h. This process was repeated for a total of three cycles.
As(V)
60 40 20 0 4
6
8
10
12
Final pH Fig. 1 e Role of final solution pH in arsenite and arsenate removal by TICB, where data for pH < 4.4 has been omitted because of bead dissolution. Experiments conducted in absence of UV light, where [As]0 [ 500 mg/L.
range and, therefore, experiences more repulsion with net negatively charged TICB (Miller and Zimmerman, 2010).
3.1.2.
Bead size
Surface area measurements were performed across adjustable parameters in the synthesis and preparation of TICB, including bead size (Fig. 2a). TICB surface area increases as bead size decreases and with exposure to UV light. Because UV exposure enhances surface area (Miller and Zimmerman, 2010), surface area differences across bead size are more apparent after exposure to UV light. Given that materials with high surface area can achieve high adsorption capacities per unit mass, these results suggest that an optimized design of TICB might incorporate diameter reduction and UV irradiation. To maximize efficacy, that is removal of As(III) and As(V), there is an inherent tradeoff in that smaller particles (with higher surface area) are more effective but require posttreatment filtration. As such, there is likely an optimal bead size range that balances removal efficiency while maintaining density-separation. Fig. 2b shows the percent As(III) and As(V) arsenic removal across a range of bead sizes, where the final pH is <7.6 for As(III) and <7.2 for As(V), and where the dosing of TiO2: As ratio remains constant to isolate the impact of bead size on efficacy. The largest bead size tested (937 um) demonstrates similar removal capacities to neat TiO2 nanopowder. That is, bead size does not have an effect on the removal efficiency in terms of capacity for a given set of system conditions (i.e., arsenic oxidation state; irradiation). This is an unexpected result because the surface area of neat TiO2 powder is two orders of magnitude greater than that of TICB, and the TiO2 in TICB is bound to a chitosan matrix such that it may not all be available for arsenic removal. This suggests that TICB, in terms of capacity, acts like TiO2 powder given sufficient exposure time. However, this also suggests a need to assess the system kinetics, a critical factor for actual implementation in a field setting. Fig. 2 also illustrates improved performance (i.e., sorption of both As(III) and As(V)) for UV-exposed TICB relative to UVunexposed TICB, indicating that UV light is a variable that may be employed to enhance adsorption capacity.
5748
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
a
b 100 80
0.6
% As Removal
BET surface area (m2/g)
0.8
0.4
0.2
60 40 20
0
0 0
200
400
600
800
1000
0
200
bead diameter (µm)
400
600
800
1000
bead diameter (µm) As(III); 0 h UV irradiation As(III); 220 h UV irradiation As(V); 0 h UV irradiation As(V); 220 h UV irradiation
220 h UV irradiation 0 h UV irradiation
Fig. 2 e Relationship between bead size (where all beads are 30% TiO2 by weight and have been exposed to UV irradiation as noted) and a) BET surface area and b) % arsenic removal from [As]0 [ 1000 mg/L.
TiO2 loading
Chitosan beads impregnated with increasing amounts of TiO2 were synthesized, where it was empirically determined that 46% TiO2 by mass was the maximum amount of TiO2 that can be incorporated into a homogenous bead formulation. As shown in Fig. 3a, the surface area of these beads increases with increasing concentration of TiO2 in the bead. These beads were tested for As(III) and As(V) removal, in both the presence and absence of UV light to determine the relationship between TiO2 loading and functional performance. Arsenic removal increases as TiO2 loading increases for all experimental conditions, as illustrated with the representative plot in Fig. 3b; results from other experimental
BET surface area (m2/g)
a
0.8 0.6 0.4 0.2
conditions are included in Supplementary Information (SI.1). This suggests that all TiO2 in the bead contributes to arsenic adsorption, regardless of the mass fraction of TiO2 in the bead. To confirm that all TiO2 present provides adsorption capacity in TICB, regardless of relative TiO2 mass fraction in the bead, neat TiO2 nanopowder, equivalent in mass to the amount of TiO2 impregnated in the beads, was tested across the same experimental conditions. For removal of both As(III) and As(V), in the presence or absence of UV light, arsenic removal by TICB and TiO2 for a given mass of TiO2 are nearly identical (Fig. 3 and SI.1). This supports the finding that all of the TiO2 impregnated in the bead, not just the TiO2 on the
b
100
% As removal
3.1.3.
80 60 40 20
0
0 0
0.05
0.1
0.15
mmol TiO2 /25 mg TICB
0
0.05
0.1
0.15
mmol TiO2/ 25 mg TICB As(III)_UV As(V)_UV As(III)_UV, TiO2 powder As(V)_UV, TiO2 powder
Fig. 3 e Relationship between mmol TiO2 incorporated into TICB (875 mm) and a) BET surface area and b) % arsenic removal from [As]0 [ 1000 mg/L in the presence of UV light.
5749
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
bead surface, is forming chemical complexes with the arsenic in solution, given sufficient time to equilibrate.
3.1.4.
UV light
0.8
UV absent UV present
Final pH (<7)
As removal (mg/g) Experimental observation
6.88 0.04 6.42 0.03
1054 74 1397 11
binuclear complex forms between TiO2 and both arsenite and arsenate (Jegadeesan et al., 2010; Jing et al., 2005, 2009; Pena et al., 2006) (Fig. 5b). A similar complex, shown in Fig. 5c, where one of the TiO2 coordinating sites is replaced with a UVinduced COOH group may form. It has been previously reported that TICB can oxidize As(III) to As(V) in the presence of UV light under system conditions of 365 nm and sufficient oxygen (Miller and Zimmerman, 2010). Experiments conducted in sunlight confirm that sunlight can also induce TICB’s photooxidative process. As(III) samples exposed to sunlight showed 100% conversion to As(V) within 8 days, whereas samples not exposed to sunlight showed <16% conversion to As(V) over the course of 12 days (SI.3). As
100 80
0.6
60 0.4 40 0.2
% Porosity
BET surface area (m2/g)
Beads of identical size were exposed to different durations of UV irradiation. As shown in Fig. 4, there is a positive relationship between surface area and UV exposure, and SEM images of these beads (Miller and Zimmerman, 2010) reveal changes in surface morphology after exposure to UV light. This may be a result of a photooxidative process occurring at the bead surface. Although UV irradiation affects the surface of TICB, porosity measurements of beads with varying exposures to UV light are relatively constant and suggest that UV irradiation does not affect pores >3 nm within the dehydrated bead interior (Webb, 2001). The presence of UV light enhances arsenate sorption on TICB (Miller and Zimmerman, 2010). Because the pH of all samples, including controls, is lowered with UV irradiation (SI.2) and because lower solution pH is associated with greater arsenic removal (Fig. 1) (Miller and Zimmerman, 2010), pH was investigated as a potentially confounding variable. When controlling for the pH fluctuations caused by irradiation, enhanced arsenic removal for samples exposed to UV light is still observed. As shown in Table 1, more As(V) is removed when UV is present than when UV is absent, where final pH is <7. Based on a model developed to predict TICB performance in the presence of UV light (Section 3.3), the 7% difference in final pH would account for a 15% difference in final removal capacities, far less than the actual 28% difference observed between 1054 mg/g and 1397 mg/g, suggesting that UV light does in fact enhance sorption independent of its impact on solution pH. UV irradiation can result in scission of linkages in the chitosan backbone, producing carboxyl groups that do not significantly affect the chemical structure of chitosan (Zubieta et al., 2008). One likely oxidation product of chitosan, where a carboxy group is formed in the C6 position, is shown in Fig. 5a (Ahmed et al., 2003). The addition of new carboxyl groups may provide additional coordinating sites for arsenic oxyanions. Previous studies have reported that a bidentate,
Table 1 e % removal from batch experiment where [As(V)]0 [ 1000 mg/L and samples were analyzed at time [ 240 h. UV absent samples were spiked with HNO3 to match pH drop of UV present samples. Values reported are averages of four replicates.
20
0
0 0
100
200
300
h UV irradiation pretreatment Surface area
Porosity
Fig. 4 e Relationship between UV irradiation duration and BET surface area and porosity, where all beads are w875 mm and 30% TiO2 by weight.
Fig. 5 e Molecular structures relevant to the TICB-As system. a) Potential scheme for oxidation of chitosan (Ahmed et al., 2003). b) Bidentate binuclear complexation of arsenate by TiO2 (Jegadeesan et al., 2010; Jing et al., 2005; Jing et al., 2009; Pena et al., 2006). c) Potential bidentate binuclear complexation of arsenate by TiO2 and oxidation product of chitosan.
5750
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
decentralized treatment is the intended application, this is a significant result because sunlight is a free, non-resource intensive, and readily available source of UV light.
3.1.5.
Groundwater
Adsorbent behavior in a deionized water matrix may not be predictive of behavior in a natural water matrix, where background ions and pH can control adsorption processes (Schwartzenbach et al., 2003). For arsenic chemisorption by TiO2, in particular, researchers have reported that phosphate and silicate can act as competitive background ions (Jing et al., 2009; Mohan and Pittman Jr., 2007; Viraraghavan and Chowdhury, 2007). In a synthetic groundwater matrix, designed to model water from Bangladesh (Leupin and Hug, 2005), removal of As, Ca, Mg, P, and Si with TICB was tested (Fig. 6). These batch TICB adsorption tests were conducted in the presence and absence of UV light. In all cases, concentrations of Ca, Mg, and Si remained constant, and the concentration of As and P decreased. In the UV irradiated system, percentages of arsenic and of phosphate removed, with a range of TICB dosing, were nearly identical, suggesting that phosphate and arsenate are direct competitors for sorption sites. TICB performance was also tested in a natural arsenic-laden groundwater, collected from a tubewell in West Bengal, India. As with the synthetic groundwater, TICB removed nearly identical percentages of arsenic and phosphate in natural groundwater (SI.4). This suggests the TICB intended for use in the field will require enough TiO2 (either through increased mass per bead or an increased number of beads) to account for removal of both P and As.
3.1.6.
Capacity for arsenic desorption and resorption
From ease of use and sustainability perspectives, an adsorbent that can be simply regenerated and reused minimizes resource consumption and offers economic and environmental advantages. TICB was investigated for regeneration and reuse capacity over several adsorption/desorption cycles. A variety of solvents in which chitosan is insoluble, including base and some acids (Pillai et al., 2009), were evaluated for arsenic desorption from TICB. These regeneration solvents were evaluated at various concentrations to minimize the
a
Consistent with other literature reports (Bang et al., 2005; Pena et al., 2005), equilibrium between arsenic and TiO2 nanopowder in our batch system was reached rapidly, within 2 h (SI.6). Equilibrium between our “standard” TICB system (w875 um in diameter, 30% TiO2 loading by mass) and arsenic, however, was not reached for at least seven days. This was the case for As(III) and As(V), in the absence or presence of UV light. The adsorption kinetics of TICB systems with varying bead size and UV exposure (identical to those reported in Section 3.1.2 (Fig. 2) and Section 3.1.4 (Fig. 4), respectively)
% analyte removal
% analyte removal
60
Kinetics of adsorption
UV present 100
As Ca Mg P Si
80
3.2.
b
UV absent 100
amount of resources required for regeneration and to minimize hazard associated with aqueous arsenic-laden waste after regeneration. Of potential desorbing solvents screened, NaOH was the most effective. Kinetics of As(III)- and As(V)- desorption from saturated TICB were tested for a range of NaOH concentrations (SI.5). In all cases, >50% of arsenic adsorbed was desorbed within 2 h. Of the NaOH concentrations tested (0.07 M, 0.67 M, 1.67 M), 0.07 M NaOH demonstrated the most rapid and most complete desorption for both As(III) and As(V). For three cycles of adsorption/desorption, TICB achieved equilibrium with a non-buffered arsenic solution of either 10,000 mg As(III)/L or 10,000 mg As(V)/L. During desorption, Assaturated beads were equilibrated with NaOH (0.07 M, 0.67 M, 1.67 M). For all NaOH concentrations, > 60% of the arsenic adsorbed, whether As(III) or As(V), is desorbed; with 0.07 M NaOH, >85% of As(III) adsorbed and >78% of As(V) adsorbed was desorbed each of the three cycles. A representative plot of these adsorption/desorption cycles is shown in Fig. 7. In addition to investigating bead reuse and resorption from 10,000 mg/L arsenic solutions, other concentrations of greater relevance to source waters were tested. Resorption isotherms for up to three cycles, where initial As concentration ranged from 100 mg/L to 10,000 mg/L, are shown in Fig. 8. These isotherms show sustained and possibly improved resorption performance for As(III) but slightly varied resorption behavior for As(V), where the final solution pH can account for the changes in the observed amounts of As(V) resorbed.
40 20 0
As Ca Mg P Si
80 60 40 20 0
0
0.02
0.04
0.06
0.08
g TICB/ 40 mL groundwater
0.1
0
0.02 0.04 0.06 0.08
0.1
g TICB/ 40 mL groundwater
Fig. 6 e Ion removal in synthetic groundwater matrix by TICB where a) UV is absent, and b) UV is present.
5751
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
% As(III) Removal
100 80 60 40 20 0 0
100
200
300
400
Time (h) TICB ∅ ~875 μm TICB ∅ ~745 μm TiO2∅ ~0.025 μm Fig. 7 e Mass of As(III) sorbed onto TICB from a 10,000 mg/L stock solution and subsequently desorbed from TICB with 0.07 M NaOH for three cycles.
were evaluated to identify potential strategies to minimize this kinetic limitation which would likely be unacceptable in a field situation. These experiments were conducted without buffering to assess the kinetics of the system without interference from buffering ions. However, because pH can influence the rate of arsenic adsorption by TiO2 (Dutta et al., 2004), solution pH was closely monitored and was found to remain relatively constant throughout kinetic experiments.
3.2.1.
TICB size
The relationship between bead size and rate of removal was investigated. Kinetics of arsenic adsorption by TICB with
a
Fig. 9 e % As(III) removal in the absence of UV light for neat TiO2 nanopowder, 744 um TICB and 875 mm TICB, where the mass of TiO2 per sample is equal and [As]0 [ 500 mg/L.
diameter of 875 um and 744 um as well as neat TiO2 nanopowder are shown in Fig. 9. In these experiments, the mass of TiO2 in the system is constant for the two TICB sizes and the neat powder. Results indicate that a slight reduction (<15%) in the size of standard TICB can significantly increase the rate of arsenic removal, so much so that these slightly smaller TICB directly mimic neat TiO2 powder. Water and free molecules can travel within chitosan, which forms hydrogels in aqueous solution (Berger et al., 2004a, 2004b). Hydration of a hydrogel requires diffusion of water molecules into the polymer network, relaxation of
b
As(III)
µg As resorbed/g TICB
3000
µg As resorbed/g TICB
As(V) 3000
2500 2000 1500 1000 500 0
2500 2000 1500 1000 500 0
0
5000
10000
0
2000 4000 6000 8000 10000
[As]e (µg/L)
[As]e (µg/L)
Sorption cycle 1
Sorption cycle 1
Sorption cycle 2
Sorption cycle 2
Sorption cycle 3
Sorption cycle 3
Fig. 8 e Resorption isotherms over three adsorption/desorption cycles with 0.07 M NaOH where for a) As(III), where final pH [ 9.27, 8.85, 8.88 for sorption cycles 1, 2, and 3 respectively, and b) As(V), where final pH [ 8.74, 7.46, 7.72 for sorption cycles 1, 2, and 3, respectively.
5752
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
Table 2 e Kinetic calculations for TICB at 25 C, Orpm, where [As]0 [ 10,000 mg/L and bead diameter w875 mm. Pseudo 1st qe, calc (mg/g) As(III) removal As(V) removal
UV UV UV UV
absent present absent present
2861.6 4708.4 1940.8 3807.9
2892.9 3102.6 2540.2 3827.6
k (1/hour) 9.30 1.39 7.54 1.56
103 102 103 102
R2 0.904 0.868 0.886 0.946
polymer chains, and expansion of the polymer network into the surrounding bulk water medium (Okano, 1998). Enhanced kinetics observed for smaller beads reflects the fact that less diffusion is required for dissolved arsenic oxyanions to reach potential TiO2 binding sites when bead diameter is reduced.
3.2.2.
UV light
Several kinetic models, including power function, pseudo first order and pseudo second order, were tested to fit experimental data for arsenic sorption by TICB, models previously reported for similar metal-sorbent systems with photooxidation (Jing et al., 2009; Ofomaja et al., 2010). For all systems examined, that is regardless of arsenic species and the presence of UV light, the pseudo first order model best predicted the equilibrium sorption capacity, and best fit parameters for the system at 25 C are shown in Table 2. Not only does UV irradiation result in higher equilibrium capacities, but UV irradiation also results in higher rates of arsenic removal. The increased rate of arsenite and arsenate removal in the presence of UV light can be attributed to the availability of more arsenic binding sites as described above. Identical kinetic testing was conducted outdoors with fluctuating temperature, both with and without exposure to sunlight; a pseudo first order model also best predicts equilibrium sorption capacity for these data, and sunlight resulted in higher removal capacities as well as rates of removal (SI.7).
3.3.
Toward a predictive TICB sorption capacity model
Data presented in Section 3.1 indicate that arsenic removal in a given TICB system is directed by the solution pH and by the TiO2 content in the bead and is enhanced in the presence of
Table 3 e Semi-empirical model to predict TICB performance for pH range 4e11 and % TiO2 (by mass) range 0e46. Results apply to reference experiment, where equilibrium is reached after 220 h between 25 mg TICB and 40 mL 1000 mg As/L solution. a mg As=g TICB ¼ 1 þ ð20:8=%TiO2 Þ e^ðpH pKa Þ Experimental conditions As(III) As(V) AsUV(III) þ AsUV(V)
a
pka
599 880 1658
9.2 6.98 6.98
µ g As sorbed / g TICB
qe, exp (mg/g)
model 10% TiO2 model 30% TiO2 10% TiO2; R2=0.87 30% TiO2; R2=0.94
1600
1200
800
400
0 4
6
8
10
12
pH Fig. 10 e Arsenic removal (mg As/g TICB) across pH for beads of varying TiO2 loading. Experimental and model data are compared for beads loaded with 10% TiO2 by weight and 30% TiO2 by weight.
UV light. A semi-empirical model, based on the logistic function, was developed to predict TICB sorption capacity from solution pH and TiO2 content in the bead. Based on our mechanistic understanding that TICB-As complexes form through hydroxide exchange and can only form at pH values where electrostatic repulsion does not occur, a logistic function was chosen. The logistic function depicts the sigmoidal shape of the arsenic removal data across pH, analogous to the shape of a titration curve across pH. Fitting parameters were derived from batch equilibrium experiments between arsenic ([As]0 ¼ 1000 mg/L) and TICB (30% TiO2 by weight, 875 mm in diameter), across the pH range 4e11. This model, shown in Table 3, was optimized for three experimental conditions: As(III), As(V), and AsUV(III) þ AsUV(V). The different adsorption capacities for the experimental conditions are accounted for in parameter a. Different inflection points observed for As(III) and As(V) across pH (Fig. 1) are a result of repulsive charges between negatively charged TICB and negative arsenic oxyanions. These inflection points are determined by the relevant pKas for As(III), As(V), and AsUV(III) þ AsUV(V), where all arsenic exists as As(V). Experimental data is compared to this model for the experimental condition AsUV(III) þ AsUV(V) in Fig. 10; R2 values for TiO2 compositions of 10% and 30% by weight are 0.87 and 0.94, respectively.
4.
Conclusions
The optimization of TiO2-impregnated chitosan bead, TICB, as an arsenic sorbent was performed. As TiO2 loading increases, TICB surface area increases, and arsenic removal capacity of TICB loaded with a given amount of TiO2 is similar to the arsenic removal capacity of the neat TiO2 nanopowder equivalence. pH is a critical variable in the TICB sorption system, where As(III) removal is greatest below 9.2 and As(V) removal is greatest below 7.25. Sorption capacity by TICB is
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
determined to be a function of pH and % TiO2 loading, and a model was developed to predict arsenic sorption, for given experimental conditions, based on these two variables. Across pH 4e11 in the presence of UV light, this model predicts adsorption capacity with R2 values of 0.87 and 0.94 for 10% TiO2 beads and 30% TiO2 beads, respectively. Reduction in bead diameter does not influence sorption capacity, given sufficient equilibrium time. That is, the largest TICB (800 mme940 mm) results in arsenic removal that is similar to that of an equivalent dose of neat TiO2 nanopowder. However, bead diameter reductions increase bead surface area and increase rate of arsenic removal. Exposure to UV light also increases bead surface area and the rate of arsenic removal but does not affect the porosity of the bead. Long detention times are not practical for point-of-use systems, and future work will aim to minimize detention times through bead diameter reductions and incorporation of UV irradiation. Laboratory tests were performed to predict performance of TICB in a field setting. In synthetic and natural groundwater, phosphate is found to be a direct competitor for arsenic sorption sites on TICB. Other ions tested, including Ca, Mg, and Si, do not compete for sorption sites on TICB. In the presence of sunlight and TICB, dissolved As(III) is oxidized to As(V) within 8 days. Finally, TICB can be regenerated with weak NaOH, and TICB can be reused without losing effectiveness for at least three cycles in batch experiments.
Acknowledgments We are grateful to Jamila Yamani for her laboratory assistance and for her support of this project. Funding was provided by the National Science Foundation Graduate Research Fellowship and National Science Foundation Environmental Sustainability (CBET-0932060).
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.040.
references
Agros, M., Kalra, T., Rathouz, P.J., Chen, Y., Pierce, B., Farvez, F., Islam, T., Ahmed, A., Rakibuz-Zaman, R., Hasan, R., Sarwar, G., Slavkovich, V., van Geen, A., Graziano, J., Ahsan, H., 2010. Arsenic exposure from drinking water, and all-cause and chronic-disease mortalities in Bangladesh (HEALS): a prospective cohort study. The Lancet 376 (9737), 252e258. Ahmed, G.A.-W., Khairou, K.S., Hassan, R.M., 2003. Kinetics and mechanism of oxidation of chitosan polysaccharide by permanganate ion in aqueous perchlorate solutions. Journal of Chemical Research S, 182e183. Bang, S., Patel, M., Lippincott, L., Meng, X., 2005. Removal of arsenic from groundwater by granular titanium dioxide adsorbent. Chemosphere 60, 389e397. Berger, J., Reist, M., Mayer, J.M., Felt, O., Gurny, R., 2004a. Structure and interactions in chitosan hydrogels formed by
5753
complexation or aggregation for biomedical applications. European Journal of Pharmaceutics and Biopharmaceutics 57 (1), 35e52. Berger, J., Reist, M., Mayer, J.M., Felt, O., Peppas, N.A., Gurny, R., 2004b. Structure and interactions in covalently and ionically crosslinked chitosan hydrogels for biomedical applications. European Journal of Pharmaceutics and Biopharmaceutics 57 (1), 19e34. Bhattacharya, P., Mukherjee, A.B., Bundschuh, J., Zevenhoven, R., Loeppert, R.H. (Eds.), 2007. Arsenic in Soil and Groundwater Environment. Elsevier, Amsterdam. Deebar, N., Irfan, A., Ishtiaq, Q.A., 2009. Evaluation of the adsorption potential of titanium dioxide nanoparticles for arsenic removal. Journal of Environmental Sciences 21, 402e408. 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, 270e275. Ferguson, M.A., Hoffmann, M.R., Hering, J.G., 2005. TiO2photocatalyzed As(III) oxidation in aqueous suspensions: reaction kinetics and effects of adsorption. Environmental Science & Technology 39, 1880e1886. Gerente, C., Lee, V.K.C., Le Cloirec, P., McKay, G., 2007. Application of chitosan for the removal of metals from wastewaters by adsorption - mechanisms and models review. Critical Reviews in Environmental Science and Technology 37 (1), 41e127. Jegadeesan, G., Al-Abed, S.R., Sundaram, V., Choi, H., Scheckel, K.G. , Dionysiou, D.D., 2010. Arsenic sorption on TiO2 nanoparticles: size and crystallanity effects. Water Research 44, 965e973. Jing, C., Liu, S., Patel, M., Meng, X., 2005. Arsenic leachability in water treatment adsorbents. Environmental Science & Technology 39, 5481e5487. Jing, C., Meng, X., Calvache, E., Jiang, G., 2009. Remediation of organic and inorganic arsenic contaminated groundwater using a nanocrystalline TiO2-based adsorbent. Environmental Pollution 157, 2514e2519. Leupin, O.X., Hug, S.J., 2005. Oxidation and removal of arsenic (III) from aerated groundwater by filtration through sand and zero-valent iron. Water Research 39, 1729e1740. Miller, S.M., Zimmerman, J.B., 2010. Novel, bio-based, photoactive arsenic sorbent: TiO2-impregnated chitosan bead. Water Research 44, 5722e5729. Mohan, D., Pittman Jr., C.U., 2007. Arsenic removal from water/ wastewater using adsorbents - A critical review. Journal of Hazardous Materials 142, 1e53. Muzzarelli, R.A.A., Muzzarelli, C., 2005. Polysaccharides 1: Structure, Characterization and Use, pp. 151e209. Neubauer, K.R., Reuter, W., Perrone, P., Grosser, Z., 2004. In: Services, P.L.a.A. (Ed.), Simultaneous Arsenic and Chromium Speciation by HPLC/ICP-MS in Environmental Waters. PerkinElmer, Shelton, CT. Ofomaja, A.E., Baidoo, E.B., Modise, S.J., 2010. Kinetic and pseudosecond-order modeling of lead biosorption onto pine cone powder. Industrial & Engineering Chemistry Research 49, 2562e2572. Okano, T. (Ed.), 1998. Biorelated Polymers and Gels. Academic Press, San Diego. Pena, M., Meng, X., Korfiatis, G.P., Jing, C., 2006. Adsorption mechanism of arsenic on nanocrystalline titanium dioxide. Environmental Science & Technology 40, 1257e1262. Pena, M.E., Korfiatis, G.P., Patel, M., Lippincott, L., Meng, X., 2005. Adsorption of As(V) and As(III) by nanocrystalline titanium dioxide. Water Research 39, 2327e2337. Petrusevski, B., Sharma, S., van der Meer, W.G., Kruis, F., Khan, M., Barua, M., Schippers, J.C., 2008. Four years of development and field-testing of IHE arsenic removal family filter in rural Bangladesh. Water Science and Technology 58 (1), 53e58.
5754
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 4 5 e5 7 5 4
Pillai, C.K.S., Paul, W., Sharma, C.P., 2009. Chitin and chitosan polymers: Chemistry, solubility, and fiber formation. Progress in Polymer Science 34, 641e678. Ratnaike, R.N., 2003. Acute and chronic arsenic toxicity. Postgraduate Medical Journal 79, 391e396. Schwartzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. John Wiley and Sons, Inc., Hoboken, New Jersey. Tien, V.N., Chaudhary, D.S., Ngo, H.H., Vigneswaran, S., 2004. Arsenic in water: concerns and treatment technologies. Journal of Industrial and Engineering Chemistry 10 (3), 337e348.
Viraraghavan, T., Chowdhury, P., 2007. Recent trends in arsenic removal technologies. Progress in Environmental Science and Technology, 692e694. Webb, P.A., 2001. An Introduction to the Physical Characterization of Materials by Mercury Intrusion Porosimetry with Emphasis on Reduction and Presentation of Experimental Data Norcross, Georgia. Zubieta, C.E., Messina, P.V., Luengo, C., Dennehy, M., Pieroni, O., Schulz, P.C., 2008. Reactive dyes remotion by porous TiO2chitosan materials. Journal of Hazardous Materials 152, 765e777.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
One year environmental surveillance of rotavirus specie A (RVA) genotypes in circulation after the introduction of the Rotarix vaccine in Rio de Janeiro, Brazil Tulio Machado Fumian a,*, Jose´ Paulo Gagliardi Leite a, Tatiana Lundgreen Rose a, Tatiana Prado b, Marize Pereira Miagostovich a a
Laboratory of Comparative and Environmental Virology, Oswaldo Cruz Institute, Oswaldo Cruz Foundation (Fiocruz), Av. Brasil 4.365, Manguinhos, CEP 21040-360, Rio de Janeiro (RJ), Brazil b Laboratory of Technological Development in Virology, Oswaldo Cruz Institute, Oswaldo Cruz Foundation (Fiocruz), Av. Brasil 4.365, Manguinhos, CEP 21040-360, Rio de Janeiro (RJ), Brazil
article info
abstract
Article history:
Rotavirus specie A (RVA) infection is the leading cause of severe acute diarrhea among
Received 26 May 2011
young children worldwide. To reduce this major RVA health impact, the Rotarix vaccine
Received in revised form
(GlaxoSmithKline, Rixensart, Belgium) was introduced in the Brazilian Expanded Immu-
24 August 2011
nization Program in March 2006 and became available to the entire birth cohort. The aim of
Accepted 25 August 2011
this study was to evaluate the spread of RVA in the environment after the introduction of
Available online 1 September 2011
Rotarix in Brazil. For this purpose, a Wastewater Treatment Plant (WTP) in Rio de Janeiro was monitored for one year to detect, characterize and discriminate RVA genotypes and
Keywords:
identify possible circulation of vaccine strains. Using TaqMan quantitative PCR (qPCR),
Rotavirus A genotypes
RVA was detected in 100% (mean viral loads from 2.40 105 to 1.16 107 genome copies
Rotarix vaccine
(GC)/L) of sewage influent samples and 71% (mean viral loads from 1.35 103 to
Wastewater
1.64 105 GC/L) of sewage effluent samples. The most prevalent RVA genotypes were P[4],
Wastewater treatment plant
P[6] and G2, based on VP4 and VP7 classification. Direct nucleotide sequencing (NSP4 fragment) and restriction enzyme digestion (NSP3) analysis did not detect RVA vaccine-like strains from the sewage samples. These data on RVA detection, quantification and molecular characterization highlight the importance of environmental monitoring as a tool to study RVA epidemiology in the surrounding human population and may be useful on ongoing vaccine monitoring programs, since sewage may be a good screening option for a rapid and economical overview of the circulating genotypes. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Rotavirus specie A (RVA) is the main etiological agent of viral gastroenteritis in infants throughout the world and is associated with significant mortality in developing countries, where over 600,000 deaths occur annually (Parashar et al., 2006). In
developed countries, this virus remains a common cause of morbidity with significant economic burden (Charles et al., 2006; Parashar et al., 2006). RVA belongs to the Reoviridae family, Rotavirus genus, and possesses a double-stranded RNA (dsRNA) genome with 11 segments that encode six structural (VP) and six non-
* Corresponding author. Tel.: þ55 21 25621875; fax: 55 21 25621851. E-mail address: [email protected] (T.M. Fumian). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.039
5756
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
structural proteins (NSP) (Estes and Kapikian, 2007). A widely used binary classification scheme has been established based on the two genes that codify the outer capsid proteins, VP4 and VP7, defining G (from VP7, glycoprotein) and P (from VP4, protease-cleaved protein) genotypes. Currently 27 G and 35 P genotypes are recognized (Abe et al., 2009; Solberg et al., 2009; Ursu et al., 2009; Matthijnssens et al., 2011); however, only five RVA G genotypes (G1eG4 and G9) and two P genotypes (P[8] and P[4]) are prevalent worldwide (Santos and Hoshino, 2005; Ursu et al., 2009). RVA virions are shed in extremely high concentrations (up to 1010 virus/g) in the stool of infected children with acute gastroenteritis and can persist in the environment for long periods of time (Carter, 2005; Bosch et al., 2008). The features of the virions, including stability in aqueous environments and resistance to water treatment, may facilitate their transmission to humans via contaminated water (Ansari et al., 1991; Espinosa et al., 2008). Direct sewage discharge into environmental waters such as lagoons, rivers, beaches and coastal waters represents a public health problem mainly in developing countries. These contaminated waters have been broadly linked to the causation of several waterborne gastroenteritis outbreaks (Kukkula et al., 1997; Villena et al., 2003a; Schmid et al., 2005; Godoy et al., 2006). Despite the difficulty of determining the proportion of gastroenteritis cases due to contaminated water, it has been suggested that a significant percentage of the cases are related to the quality of the water (Bosch et al., 2008). As RVA is one of the most important causes of mortality in infants worldwide, two equally safe and efficacious live oral rotavirus vaccines, G1P[8] RVA vaccine (RV1 e Rotarix, GlaxoSmithKline, Rixensart, Belgium) and a pentavalent G1G4 and P[8] RVA vaccine (RV5 e RotaTeq, Merck and Co., Whitehouse Station, NJ, USA), were developed and are licensed for use in more than 100 countries worldwide (Jiang et al., 2010). The first one, RV1, was included in the Brazilian Expanded Immunization Program (PNI) in March 2006 and became available to the entire birth cohort. The impact of this vaccine on the circulating RVA genotypes is unknown and difficult to predict, so continuous genotype surveillance is needed to identify the effects of the vaccine program on circulating strains, particularly on genotype prevalence and the emergence of uncommon strains. The monitoring of the viruses circulating in sewage from a wastewater treatment plant (WTP) has been described as an appropriate model to understand the spread of RVA in the population served by the WTP, as influents may contain viruses shed from patients with sporadic or asymptomatic cases (Haramoto et al., 2006; Bosch et al., 2008). The main goal of this study was to evaluate the spread of RVA in the environment following the introduction of the Rotarix vaccine in Brazil. For this purpose, a WTP located in Rio de Janeiro was monitored for one year to detect, quantify and characterize RVA genotypes and to investigate the possible presence of the vaccine strain in sewage samples. RVA genomes were investigated in samples collected from raw and treated sewage using Taqman quantitative PCR (qPCR), and the P (VP4) and G (VP7) genotypes were characterized by nested PCR in a multiplex reaction (Gentsch et al., 1992; Gouvea et al., 1994). Protocols based on direct
nucleotide sequencing (NSP4) and restriction enzyme digestion (NSP3) analysis (Rose et al., 2010) were applied to discriminate between wild-type and vaccine strains.
2.
Materials and methods
2.1. Wastewater treatment plant (WTP) sample collection Sewage samples were collected from an urban WTP located in the metropolitan area of Rio de Janeiro, Brazil. The WTP receives sewage from around 1.5 million inhabitants leaving in both the central and north zone of the city and is one of the largest in Brazil. Sewage treatment employs a secondary treatment (aerobic process: activated sludge) with an inflow mean of 2500 L s1. Initial sewage treatment is composed of grid separation and primary sedimentation (five primary settling tanks with a volume of 7700 m3 each). There are four aeration tanks in parallel (volume: 11,500 m3 per tank) with a capacity to treat 625 L s1 of effluent. Secondary sedimentation is performed in four secondary settling tanks (volume: 8800 m3 per tank) with no chlorination before effluents are discharged into the water environment. A total of 48 sewage samples were collected bi-monthly (15 day interval) from August 2009 to July 2010, 24 of them were collected from raw sewage (influent) and 24 from the final treated sewage (effluent). At each sampling point, 50 ml of sewage was collected in sterile plastic bottles, kept at 4 C and transported to the laboratory for immediate analysis.
2.2.
Virus concentration
Viruses were concentrated using the ultracentrifugation method as described by Pina et al. (1998). To avoid false negative results and to evaluate the presence of inhibitors, sewage samples were inoculated with 500 ml of an internal control (bacteriophage PP7) before the concentration assay, and the extracted RNA was diluted 10-fold (Fumian et al., 2010).
2.3. Nucleic acid extraction, reverse transcription (RT) and quantitative PCR (qPCR) The viral dsRNA was extracted by the glass powder method (Boom et al., 1990), and the synthesis of cDNA was carried out by reverse transcription using a random primer (PdN6 e 50 A260 units e Amersham Biosciences, Chalfont St Giles, Buckinghamshire, UK). Multiplex qPCR to detect RVA and PP7 was performed as described previously (Fumian et al., 2010) using primers described by Zeng et al. (2008) and Rajal et al. (2007). RVA primers and probe were designed to target a highly conserved region of the non-structural protein 3 (NSP3), and the PP7 primers to amplify a region of the PP7 replicase gene. Both were synthesized by Applied Biosystems (CA, USA). qPCR was carried out using an ABI PRISM 7500 Sequence Detection System (Applied Biosystems, CA, USA). A standard curve (SC; 107, 105, 103 and 101 copies per reaction) was generated using 10-fold serial dilutions of a pCR2.1 vector (Invitrogen, USA) containing either the RVA NSP3 gene or the PP7 replicase gene.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
2.4.
VP4 and VP7 nested PCR amplification
Nested PCR was used for molecular characterization of RVA genotypes G and P, and it partially amplified VP7 and VP4 segments, respectively. In the first-round, RT-PCR was performed with VP7 and VP4 consensus primers 9con1e9con2 (Das et al., 1994) and 4con2e4con3 (Gentsch et al., 1992), respectively. Following the first-round, RVA G genotype classification was performed using specific primers for genotypes G1eG4, G5 and G9 (Das et al., 1994; Gouvea et al., 1994), and P genotype classification was carried out using primers for genotypes P[4], P[6] and P[8]-P[10], described by Gentsch et al. (1992).
2.5.
RVA molecular characterization
To discriminate RVA wild-type G1P[8] from the vaccine strain, the NSP4 gene was amplified according to the protocol described by Cunliffe et al. (1998). Two nucleotide mutations after the first initiation ATG codon at positions 100 and 134 were observed when the NSP4 nucleotide sequence of the Rotarix vaccine (patent number: PCT/EP2004/009725) was compared to sequences available in GenBank including reference strains. These two nucleotide shifts were used to classify RVA (data not shown). The NSP4 PCR amplicons were purified and sequenced using an ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit and an ABI Prism 3730 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). The chromatograms were analyzed using BioEdit (Hall, 1999). A phylogenetic dendrogram was constructed by the neighbor-joining method using a matrix of genetic distances established under the Kimura-two parameter model (Felsenstein, 1993) using MEGA V. 4.0 (Tamura et al., 2007). The robustness of each node was assessed by bootstrap analysis using 2000 pseudoreplicates. RVA NSP4 isolated from sewage samples was classified according to the most recent full genome-based classification proposed by Matthijnssens et al. (2011).
2.6.
Cloning and restriction endonuclease analysis
To characterize vaccine strains, another protocol based on BspHI restriction endonuclease analysis of the NSP3 gene was
5757
performed as previously described (Rose et al., 2010). Prior to endonuclease restriction analysis, PCR amplicons generated by NSP3 amplification (Matthijnssens et al., 2006) were cloned into the PCR4-TOPO vector (Invitrogen, USA) following the manufacturer’s recommendations.
2.7.
Statistical analysis
The total frequency of detection obtained in WTP, in both influent and effluent samples, using qPCR assay was compared by using a chi-square test and Fisher’s exact test at a significance level of 0.05. The same statistical analysis was performed to determine significant differences between VP4 and VP7 PCR detection in all of the 48 samples collected. Analysis of Variance (ANOVA) was performed to determine differences in mean levels of RVA, present in influent samples, during the four seasons (summer, fall, spring and winter), and a paired t-test was performed to verify differences between the mean levels of RVA in influent and effluent samples throughout the study. Statistical analyses were performed using GraphPad Prism software version 5.
3.
Results
3.1.
Rotavirus A detection and quantification
The RVA genome levels and genotypes were determined in a one-year monitoring study from influent and effluent streams at a WTP located in Rio de Janeiro city, Brazil. Using qPCR, 41 out of 48 (85%) of the samples were positive, corresponding 100% (24/24) of influent and 71% (17/24) of effluent samples. The difference in the total frequencies of RVA detection (qPCR) in WTP was significant between influent and effluent samples ( p ¼ 0.0042, Chi-square; p ¼ 0.0047, Fisher). Fig. 1 shows the monthly distribution of RVA genome copies (GC/L) and the standard deviation. For sewage influent, RVA concentrations ranged from 2.40 104 to 1.16 107 GC/L, and in effluent, positive sample concentrations, ranged from 1.35 103 to 1.64 105 GC/L. The differences in mean levels of RVA present in influent samples throughout the seasons were not significant (ANOVA/ NewmaneKeuls Multiple
Fig. 1 e Monthly distribution of rotavirus specie A (RVA) in influent (A) and effluent (B) samples from a WTP in Brazil. The plots show the geometric monthly mean values (GC/L). The upper and lower bars show the standard deviations of the mean values.
5758
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
Comparison Test), on the other hand, difference in the mean levels of RVA was significant between influent and effluent samples ( p ¼ 0.041, Paired t-test). Table 1 summarizes the results obtained from the genomic amplification protocols used for detection, quantification (qPCR) and molecular characterization of RVA genes (RT-PCR for NSP4, VP4 and VP7). The RVA genotype G2 was detected in 100% (24/24) of influent samples, and the genotypes P[4] and P [6] were detected in 33% (8/24) and 25% (6/24), respectively. Effluent samples showed a lower RVA detection rate, with genotypes G2 and P[4] being detected in 25% (6/24) and 4% (1/ 24) of samples, respectively. Genotypes G1 and P[8] were not identified in the samples tested. The difference in the frequency of RVA detection using VP4 and VP7 PCR was significant ( p ¼ 0.002, Chi-square; p ¼ 0.004, Fisher). RT-PCR based on the NSP4 gene was able to detect viruses in 92% (22/24) and 21% (5/24) of influent and effluent samples, respectively. No evidence of inhibitors was observed as bacteriophage PP7, inoculated as an internal control in all 48 sewage samples, was detected in 100% of samples tested using a multiplex qPCR. PP7 viral titers recovered ranged from 3.4 105 to 1.6 104 GC per 500 ml of PP7 suspension.
3.2.
Rotavirus A strain characterization
The sequence of gene segment 10 (encoding NSP4) of Brazilian waste samples were compared with NSP4 segments available
Table 1 e Rotavirus specie A (RVA) detection from influent (24) and effluent (24) sewage samples by quantitative (qPCR) and qualitative (NSP4, VP4, VP7) PCR protocols for genotyping. Year Month
Sewage influent
Aug Sep Oct Nov Dec
2010
Jan Feb Mar Apr May Jun Jul
þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ
þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ
þ: Positive; and : negative.
P[4] P[4] P[6] P[4] P[6] P[4] P[4] P[6] P[6] P[6] P[4] P[4] P[6] P[4]
G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2 G2
4.
Discussion
Sewage effluent
qPCR NSP4 VP4 VP7 qPCR NSP4 VP4 VP7 2009
in GenBank. The phylogenetic analysis of this segment classified the samples within two distinct genotypes: E1 genotype, with a single sequence (RJ-VA-550) and E2 genotype, in which 10 sequences clustered. Among the sequences that clustered in genotype E2, the sequence obtained from RJ-VA-575 sample clustered in a separate branch of the tree (Fig. 2). None of the 11 NSP4 sequences showed the two nucleotide mutations as in the vaccine pattern. RVA NSP4 nucleotide sequences obtained in the present study were deposited at the National Center for Biotechnology Information (GenBank, http://www.ncbi.nlm.nih.gov/) under the accession numbers: RVA/Env-wt/BRA/RJ-VA-500/2009/ GXP[X]: JF731369; RVA/Env-wt/BRA/RJ-VA-515/2009/GXP[X]: JF731370; RVA/Env-wt/BRA/RJ-VA-518/2009/GXP[X]: JF731371; RVA/Env-wt/BRA/RJ-VA-521/2009/GXP[X]: JF731372; RVA/Envwt/BRA/RJ-VA-523/2009/GXP[X]: JF731373; RVA/Env-wt/BRA/ RJ-VA-524/2009/GXP[X]: JF731374; RVA/Env-wt/BRA/RJ-VA526/2009/GXP[X]: JF731375; RVA/Env-wt/BRA/RJ-VA-527/2009/ GXP[X]: JF731376; RVA/Env-wt/BRA/RJ-VA-547/2009/GXP[X]: JF731377; RVA/Env-wt/BRA/RJ-VA-550/2009/GXP[X]: JF731378; and RVA/Env-wt/BRA/RJ-VA-575/2010/GXP[X]: JF731379. In order to accurately determine the presence of the vaccine components in the environment, six influent samples with high viral loads (1.7 107e3.7 105 GC/L) were subjected to PCR amplification of the region of the genome encoding NSP3, and the resulting products were cloned. Forty-seven colonies were screened for the appropriate banding pattern after BspHI restriction endonuclease analysis, and none of them demonstrated the vaccine pattern. A Rotarix NSP3 amplicon was analyzed in the same reaction as a positive control.
þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ
þ þ þ þ
P[4]
G2 G2 G2 G2 G2 G2
In this study, an environmental approach was used to evaluate the circulation of RVA genotypes in the city of Rio de Janeiro, Brazil, which has the second largest population in the country. Samples from a large WTP were analyzed using a concentration method (ultracentrifugation) and molecular techniques to detect, quantify and characterize the detected viruses (Pina et al., 1998; Fumian et al., 2010). This type of approach has been extensively employed to obtain information on circulating viruses in populations throughout the world, independently of single reported cases or outbreaks, and to assess virus circulation causing asymptomatic infections (Bosch et al., 2008; Gajardo et al., 1995; Haramoto et al., 2006; Clemente-Casares et al., 2009; Fumian et al., 2010; Kamel et al., 2010; Prado et al., 2011). Besides revealing the predominant genotypes circulating in Rio de Janeiro, this monitoring strategy also aimed to investigate the presence of the attenuated G1P[8] RVA vaccine Rotarix in the environment. The combination of virus concentration and molecular detection methods was successfully employed, and the results showed a high level of RVA contamination in sewage samples. The high recovery rate of RVA (47%) from sewage samples (Fumian et al., 2010), using the ultracentrifugation method, was fundamental for the success in RVA recovering. Another study using this ultracentrifugation method to
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
5759
Fig. 2 e Phylogenetic dendrogram based on partial NSP4 nucleotide sequences of rotavirus A strains isolated from sewage samples in this study. All sequences obtained from GenBank are named according to Matthijnssens et al. (2011), and G and P genotypes are indicated at the right. The Brazilian environmental samples are marked with a filled diamond (influent samples) and an unfilled diamond (effluent samples). The scale bar at the bottom of the tree indicates distance. Bootstrap values (2000 replicates) are shown at the branch nodes and values lower than 50% are not shown.
recover RVA from domestic sewage and polluted water river samples demonstrated a high percentage of positive samples: 67% and 83% (Rodrı´guez-Dı´az et al., 2009). Lower RVA detection rates have been observed when membrane-active charged filtration was used as a concentration method associated with organic or inorganic elution (Ferreira et al., 2009; Kamel et al., 2010). A pattern of seasonality of RVA-induced gastroenteritis has been demonstrated in Latin American countries, including Brazil, based on a higher incidence of infection occurring in winter months (Kane et al., 2004; Carvalho-Costa et al., 2011). However, this differential distribution was not observed by the analysis of sewage samples during the monitoring period, suggesting a high level of virus shedding occurring throughout the year.
The average reduction of 2 logarithms in viral load observed in effluent samples demonstrates that WTPs play an important role in reducing environmental contamination. However, as demonstrated in this study, the persistence of such viruses in treated effluents and in other studies from different regions, highlights the importance of evaluating the efficiency of different types of treatments used by WTPs in viral load reduction (Bofill-Mas et al., 2006; Haramoto et al., 2006; da Silva et al., 2007; Meleg et al., 2008; La Rosa et al., 2010). Despite the difficulties in associating virus infection to contact with contaminated water, the environmental dissemination of RVA, demonstrated by the high prevalence and concentration in the treated or untreated sewage samples, poses a risk to human health that must be considered and evaluated. Although the detection of nucleic acid does not
5760
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
directly indicate the presence of infectious viruses, it is strongly suggestive of an infectious particle (Girones et al., 2010). Different studies have demonstrated that signals generated after RT-PCR amplification of viral genomes correlated well with infectivity or that a great part of viral nucleic acid recovered from environmental samples corresponded to infectious virus particles (Bhattacharya et al., 2004; Espinosa et al., 2008; Barrella et al., 2009). The results obtained in this study regarding RVA dissemination, along with other studies conducted in developing countries, indicate RVA as a possible viral indicator of human fecal contamination in environmental samples, at least in countries where there is a high RVA prevalence (Ferreira et al., 2009; Miagostovich et al., 2008; Rodrı´guez-Dı´az et al., 2009; Prado et al., 2011; Sdiri-Loulizi et al., 2010). Data concerning virus genotyping provide significant epidemiological information necessary for the introduction and ongoing monitoring of vaccination programs (Villena et al., 2003b; Pinto et al., 2007; Bosch et al., 2008). The NSP4 segment analysis showed samples clustered with two genotypes. E1 sequence was close related to NSP4 from genotypes G1P[8], G3P[8] and G12P[6] and was probably associated with P [6] genotype detected. Within E2 genotype, nine samples formed a monophyletic group, and one sequence (RJ-VA-575) clustered in another group, including Brazilian G2P[4], isolated in 2008. The high detection frequency of E2 genotype, with G2 and P[4], is in agreement with trends described by Matthijnssens et al. (2011), where strains with a G2 and P[4] genotype presented an E2 profile and strains with a G1, G3, G12, P[6] and P[8] genotypes demonstrated an E1 profile. The prevalence of G2 and P[4] genotypes in sewage samples is in agreement with the results obtained in a previous survey using clinical samples from acute infantile gastroenteritis cases in the municipality of Rio de Janeiro after Rotarix introduction (Carvalho-Costa et al., 2009, 2011). RVA P[6] genotype detected in a lower prevalence than P[4] in sewage sample, reflects results obtained from clinical samples, showing that, in Brazil, the major circulation of G2P[4] and in a slight ratio, G2P[6] genotype. Data from the Laboratory of Comparative and Environmental Virology (LVCA), a Brazilian Regional Reference Laboratory for Rotaviruses, from 2009 to 2010, showed a higher percentage (78%) of RVA G2P[4] circulating in Rio de Janeiro when compared with other genotypes characterized as G4P[8] (6%); G9P[X] (6%); G2P[6], G2P[X] and G1P[X] (3%) (data not published). Although an increasing prevalence of genotype G2P[4] has also been reported in countries that have not established Rotarix vaccination programs (Ferrera et al., 2007; Antunes et al., 2009), it is important to note that in a smaller WTP sewage monitoring program also conducted in Rio de Janeiro in 2005, before RVA vaccine introduction, G1 and P[8] were the most prevalent RVA genotypes detected (Ferreira et al., 2009). This change in the RVA genotypes prevalence profile could be explained by a natural genotypic fluctuation, although the role of the Rotarix vaccine introduction cannot be ruled out (Go´mez et al., 2011). In Australia, where both vaccine types are used, it was observed a higher prevalence of G2P[4] genotypes in states that used exclusively Rotarix vaccine when compared with states that used Rotateq, showing a higher prevalence of G3P[8] strains (Kirkwood et al., 2011). Another important
aspect regarding the RVA vaccine is that Rotarix prevents around 90% of severe gastroenteritis cases caused by G1P[8], as well as other partially heterotypic strains; however, it is less effective (45%) in preventing diarrhea caused by fully heterotypic G2P[4] strains (Ruiz-Palacios et al., 2006). Linhares et al. (2008), when evaluating Rotarix efficacy against rotavirus gastroenteritis in a phase III study performed in Latin American infants, demonstrated a vaccine efficacy of 82% and 40% against G1P[8] and G2P[4], respectively. Genotypes G1 and P[8] were not found in these sewage samples, showing that these genotypes are no longer circulating or are circulating at a very low level, reinforcing data obtained via surveillance of clinical specimens (CarvalhoCosta et al., 2009, 2011). The high shedding of RVA antigens (up to 1010 virus/g) from naturally infected individuals may restrict the detection of the vaccine strain that would be less prevalent in the environment. In a study conducted in South Africa during 2003e2004 to evaluate the safety, reactogenicity and immunogenicity of the Rotarix vaccine, virus shedding was observed in healthy infants, ranging from 31% to 46% depending on the vaccination regimen used (Steele et al., 2010). The cloning of the gene NSP3 followed by restriction enzyme analysis was an alternative attempt to increase the probability of vaccine strain detection, as cloning of the PCR products enables detection of genotypes that are at lower abundance in the environment. This methodology based on NSP3 gene amplification followed by BspHI digestion was previously described for discrimination of the Rotarix vaccine (Rose et al., 2010). Epidemiological and laboratory surveillances to assess vaccine effectiveness and vaccine impact are currently significant concerns (WHO, 2008). As Brazil was the first Latin American country to introduce universal rotavirus vaccination, the evaluation of vaccine performance to examine possible changing strain patterns of RVA in circulation is a priority in this country. Sentinel RVA surveillance in selected pediatric settings has been recommended as part of the immunization program in Latin America (Carvalho-Costa et al., 2009). Environmental surveillance, as conducted in this study by investigating RVA in sewage samples, could be an alternative approach to support clinical monitoring of RVA infection. This kind of surveillance would allow continuous investigation of the genotypes circulating in the WTP service area, providing an overview of the prevalent genotypes and possibly discriminating between RVA vaccine and wild strains.
5.
Conclusion
(1) The high circulation of RVA in the population was measurable by environmental surveillance coupled with appropriate molecular tools. (2) Wastewater surveillance demonstrated that genotypes G2 and P[4] were the most prevalent, reflecting a natural fluctuation of RVA genotypes or a consequence of Rotarix vaccine introduction or even both. (3) This is the first study concerning RVA detection and discrimination between vaccine and wild-type strains
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
from environmental samples carried out in a WTP and may assist clinical epidemiological studies that will be essential in the post-vaccination era.
Acknowledgements This work was financially sponsored by the National Council for Scientific and Technological Development (CNPq e PROSUL 490292/2008-9; CNPq e PAPES V) and by CGVAM/Ministry of Health, Brazil. The authors thank the staff of PDTIS DNA Sequencing Platform at FIOCRUZ (RPT01A) for technical support in sequencing reactions and the WTP staff for supplying the samples, under the agreement between Fiocruz and the Water Company of Rio de Janeiro state (CEDAE). This research study is under the scope of the activities of Fiocruz as a collaborating center of PAHO/WHO of Public and Environmental Health.
references
Abe, M., Ito, N., Morikawa, S., Takasu, M., Murase, T., Kawashima, T., Kawai, Y., Kohara, J., Sugiyama, M., 2009. Molecular epidemiology of rotaviruses among healthy calves in Japan: isolation of a novel bovine rotavirus bearing new P and G genotypes. Virus Res. 144, 250e257. Ansari, S.A., Springthorpe, V.S., Sattar, S.A., 1991. Survival and vehicular spread of human rotaviruses: possible relation to seasonality of outbreaks. Rev. Infect. Dis. 13, 448e461. Antunes, H., Afonso, A., Iturriza, M., Martinho, I., Ribeiro, C., Rocha, S., Magalha˜es, C., Carvalho, L., Branca, F., Gray, J., 2009. G2P[4] the most prevalent rotavirus genotype in 2007 winter season in an European non-vaccinated population. J. Clin. Virol. 45, 76e78. Barrella, K.M., Garrafa, P., Monezi, T.A., Ha´rsi, C.M., Salvi, C., Violante, P.A.B.C., Mehnert, D.U., 2009. Longitudinal study on occurrence of adenoviruses and hepatitis A virus in raw domestic sewage in the city of Limeira, Sa˜o Paulo. Braz. J. Microbiol. 40, 102e107. Bhattacharya, S.S., Kulka, M., Lampel, K.A., Cebula, T.A., Goswami, B.B., 2004. Use of reverse transcription and PCR to discriminate between infectious and non-infectious hepatitis A virus. J. Virol. Methods 116, 181e187. Bofill-Mas, S., Albinana-Gimenez, N., Clemente-Casares, P., Hundesa, A., Rodriguez-Manzano, J., Allard, A., Calvo, M., Girones, R., 2006. Quantification and stability of human adenoviruses and polyomavirus JCPyV in wastewater matrices. Appl. Environ. Microbiol. 72, 7894e7896. Boom, R., Sol, C.J., Salimans, M.M., Jansen, C.L., Wertheim-van Dillen, P.M., van der Noordaa, J., 1990. Rapid and simple method for purification of nucleic acids. J. Clin. Microbiol. 28, 495e503. Bosch, A., Guix, S., Sano, D., Pinto´, R.M., 2008. New tools for the study and direct surveillance of viral pathogens in water. Curr. Opin. Biotechnol. 19, 295e301. Carter, M.J., 2005. Enterically infecting viruses: pathogenicity, transmission and significance for food and waterborne infection. J. Appl. Microbiol. 98, 1354e1380. Carvalho-Costa, F.A., Arau´jo, I.T., Santos de Assis, R.M., Fialho, A. M., de Assis Martins, C.M., Bo´ia, M.N., Leite, J.P., 2009. Rotavirus genotype distribution after vaccine introduction, Rio
5761
de Janeiro, Brazil. Emerg. Infect. Dis. 15, 95e97. Carvalho-Costa, F.A., Volota˜o Ede, M., de Assis, R.M., Fialho, A.M., de Andrade Jda, S., Rocha, L.N., Tort, L.F., da Silva, M.F., Go´mez, M.M., de Souza, P.M., Leite, J.P., 2011. Laboratorybased rotavirus surveillance during the introduction of a vaccination program, Brazil, 2005e2009. Pediatr. Infect. Dis. J. 30, S35eS41. Charles, M.D., Holman, R.C., Curns, A.T., Parashar, U.D., Glass, R.I., Bresee, J.S., 2006. Hospitalizations associated with rotavirus gastroenteritis in the United States, 1993e2002. Pediatr. Infect. Dis. J. 25, 489e493. Clemente-Casares, P., Rodriguez-Manzano, J., Girones, R., 2009. Hepatitis E virus genotype 3 and sporadically also genotype 1 circulate in the population of Catalonia, Spain. J. Water Health 7, 664e673. Cunliffe, N.A., Kilgore, P.E., Bresee, J.S., Steele, A.D., Luo, N., Hart, C.A., Glass, R.I., 1998. Epidemiology of rotavirus diarrhoea in Africa: a review to assess the need for rotavirus immunization. Bull. World Health Organ. 76, 525e537. da Silva, A.K., Le Saux, J.C., Parnaudeau, S., Pommepuy, M., Elimelech, M., Le Guyader, F.S., 2007. Evaluation of removal of noroviruses during wastewater treatment, using real-time reverse transcription-PCR: different behaviors of genogroups I and II. Appl. Environ. Microbiol. 73, 7891e7897. Das, B.K., Gentsch, J.R., Cicirello, H.G., Woods, P.A., Gupta, A., Ramachandran, M., Kumar, R., Bhan, M.K., Glass, R.I., 1994. Characterization of rotavirus strains from newborns in New Delhi, India. J. Clin. Microbiol. 32, 1820e1822. Espinosa, A.C., Mazari-Hiriart, M., Espinosa, R., Maruri-Avidal, L., Me´ndez, E., Arias, C.F., 2008. Infectivity and genome persistence of rotavirus and astrovirus in groundwater and surface water. Water Res. 42, 2618e2628. Estes, M., Kapikian, A.Z., 2007. Rotaviruses. In: Fields Virology, fifth ed. Lippincott Williams &Wilkins, Philadelphia. Felsenstein, J., 1993. Phylogeny Interference Package, Version 3. 5. Department of Genetics, University of Washington, Seattle, USA. Ferreira, F.F., Guimara˜es, F.R., Fumian, T.M., Victoria, M., Vieira, C.B., Luz, S., Shubo, T., Leite, J.P., Miagostovich, M.P., 2009. Environmental dissemination of group A rotavirus: P-type, G-type and subgroup characterization. Water Sci. Technol. 60, 633e642. Ferrera, A., Quan, D., Espinoza, F., 2007. Increased prevalence of genotype G2P(4) among children with rotavirus-associated gastroenteritis in Honduras. In: 17th European Congress of Clinical Microbiology and Infectious Diseases ICC, Munich, Germany. Fumian, T.M., Leite, J.P., Castello, A.A., Gaggero, A., Caillou, M.S., Miagostovich, M.P., 2010. Detection of rotavirus A in sewage samples using multiplex qPCR and an evaluation of the ultracentrifugation and adsorption-elution methods for virus concentration. J. Virol. Methods 170, 42e46. Gajardo, R., Bouchriti, N., Pinto, R.M., Bosch, A., 1995. Genotyping of rotaviruses isolated from sewage. Appl. Environ. Microbiol. 61, 3460e3462. Gentsch, J.R., Glass, R.I., Woods, P., Gouvea, V., Gorziglia, M., Flores, J., Das, B.K., Bhan, M.K., 1992. Identification of group A rotavirus gene 4 types by polymerase chain reaction. J. Clin. Microbiol. 30, 1365e1373. Girones, R., Ferru´s, M.A., Alonso, J.L., Rodriguez-Manzano, J., Calgua, B., Correˆa Ade, A., Hundesa, A., Carratala, A., BofillMas, S., 2010. Molecular detection of pathogens in waterethe pros and cons of molecular techniques. Water Res. 44, 4325e4339. Godoy, P., Nuı´n, C., Alseda`, M., Llovet, T., Mazana, R., Domı´nguez, A., 2006. Waterborne outbreak of gastroenteritis caused by Norovirus transmitted through drinking water. Rev. Clin. Esp. 206, 435e437.
5762
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
Go´mez, M.M., de Mendonc¸a, M.C., Volota˜o Ede, M., Tort, L.F., da Silva, M.F., Cristina, J., Leite, J.P., 2011. Rotavirus A genotype P [4]G2: genetic diversity and reassortment events among strains circulating in Brazil between 2005 and 2009. J. Med. Virol. 83, 1093e1106. Gouvea, V., de Castro, L., Timenetsky, M.C., Greenberg, H., Santos, N., 1994. Rotavirus serotype G5 associated with diarrhea in Brazilian children. J. Clin. Microbiol. 32, 1408e1409. Hall, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl. Acids Symp. Ser. 41, 95e98. Haramoto, E., Katayama, H., Oguma, K., Yamashita, H., Tajima, A., Nakajima, H., Ohgaki, S., 2006. Seasonal profiles of human noroviruses and indicator bacteria in a wastewater treatment plant in Tokyo, Japan. Water Sci. Technol. 54, 301e308. Jiang, V., Jiang, B., Tate, J., Parashar, U.D., Patel, M.M., 2010. Performance of rotavirus vaccines in developed and developing countries. Hum. Vaccin. 6, 532e542. Kamel, A.H., Ali, M.A., El-Nady, H.G., Aho, S., Pothier, P., Belliot, G., 2010. Evidence of the co-circulation of enteric viruses in sewage and in the population of Greater Cairo. J. Appl. Microbiol. 108, 1620e1629. Kane, E.M., Turcios, R.M., Arvay, M.L., Garcia, S., Bresee, J.S., Glass, R.I., 2004. The epidemiology of rotavirus diarrhea in Latin America. Anticipating rotavirus vaccines. Rev. Panam. Salud. Publica 16, 371e377. Kirkwood, C.D., Boniface, K., Barnes, G.L., Bishop, R.F., 2011. Distribution of rotavirus genotypes after introduction of rotavirus vaccines, Rotarix and RotaTeq, into the National Immunization Program of Australia. Pediatr. Infect. Dis. J. 30, 48e53. Kukkula, M., Arstila, P., Klossner, M.-L., Maunula, L., von Bonsdorff, C.-H., Jaatinen, P., 1997. Waterborne outbreak of viral gastro-enteritis. Scand. J. Infect. Dis 29, 415e418. La Rosa, G., Pourshaban, M., Iaconelli, M., Muscillo, M., 2010. Quantitative real-time PCR of enteric viruses in influent and effluent samples from wastewater treatment plants in Italy. Ann. Ist. Super. Sanita. 46, 266e273. Linhares, A.C., Vela´zquez, F.R., Pe´rez-Schael, I., Sa´ez-Llorens, X., Abate, H., Espinoza, F., Lo´pez, P., Macı´as-Parra, M., OrtegaBarrı´a, E., Rivera-Medina, D.M., Rivera, L., Pavı´a-Ruz, N., Nun˜ez, E., Damaso, S., Ruiz-Palacios, G.M., De Vos, B., O’Ryan, M., Gillard, P., Bouckenooghe, A., 2008. Human Rotavirus Vaccine Study Group. Efficacy and safety of an oral live attenuated human rotavirus vaccine against rotavirus gastroenteritis during the first 2 years of life in Latin American infants: a randomised, double-blind, placebo-controlled phase III study. Lancet 371, 1181e1189. Matthijnssens, J., Rahman, M., Martella, V., Xuelei, Y., De Vos, S., De Leener, K., Ciarlet, M., Buonavoglia, C., Van Ranst, M., 2006. Full genomic analysis of human rotavirus strain B4106 and lapine rotavirus strain 30/96 provides evidence for interspecies transmission. J. Virol. 80, 3801e3810. Matthijnssens, J., Ciarlet, M., McDonald, S.M., Attoui, H., Ba´nyai, K., Brister, J.R., Buesa, J., Esona, M.D., Estes, M.K., Gentsch, J.R., Iturriza-Go´mara, M., Johne, R., Kirkwood, C.D., Martella, V., Mertens, P.P., Nakagomi, O., Parren˜o, V., Rahman, M., Ruggeri, F.M., Saif, L.J., Santos, N., Steyer, A., Taniguchi, K., Patton, J.T., Desselberger, U., Van Ranst, M., 2011. Uniformity of rotavirus strain nomenclature proposed by the Rotavirus Classification Working Group (RCWG). Arch. Virol. 156, 1397e1413. Meleg, E., Ba´nyai, K., Martella, V., Jiang, B., Kocsis, B., Kisfali, P., Melegh, B., Szucs, G., 2008. Detection and quantification of group C rotaviruses in communal sewage. Appl. Environ. Microbiol. 74, 3394e3399. Miagostovich, M.P., Ferreira, F.F., Guimaraes, F.R., Fumian, T.M., Diniz-Mendes, L., Luz, S.L., Silva, L.A., Leite, J.P., 2008. Molecular
detection and characterization of gastroenteritis viruses occurring naturally in the stream waters of Manaus, central Amazonia, Brazil. Appl. Environ. Microbiol. 74, 375e382. Parashar, U.D., Gibson, C.J., Bresse, J.S., Glass, R.I., 2006. Rotavirus and severe childhood diarrhea. Emerg. Infect. Dis. 12, 304e306. Pina, S., Jofre, J., Emerson, S.U., Purcell, R.H., Girones, R., 1998. Characterization of a strain of infectious hepatitis E virus isolated from sewage in an area where hepatitis E is not endemic. Appl. Environ. Microbiol. 64, 4485e4488. Pinto, R.M., Alegre, D., Dominguez, A., El Senousy, W.M., Sanchez, G., Villena, C., Costafreda, M.I., Aragones, L., Bosch, A., 2007. Hepatitis A vı´rus in urban sewage from two Mediterranean countries. Epidemiol. Infect. 135, 270e273. Prado, T., Silva, D.M., Guilayn, W.C., Rose, T.L., Gaspar, A.M., Miagostovich, M.P., 2011. Quantification and molecular characterization of enteric viruses detected in effluents from two hospital wastewater treatment plants. Water Res. 45, 1287e1297. Rajal, V.B., McSwain, B.S., Thompson, D.E., Leutenegger, C.M., Kildare, B.J., Wuertz, S., 2007. Validation of hollow fiber ultrafiltration and real-time PCR using bacteriophage PP7 as surrogate for the quantification of viruses from water samples. Water Res. 41, 1411e1422. Rodrı´guez-Dı´az, J., Querales, L., Caraballo, L., Vizzi, E., Liprandi, F., Takiff, H., Betancourt, W.Q., 2009. Detection and characterization of waterborne gastroenteritis viruses in urban sewage and sewage-polluted river waters in Caracas, Venezuela. Appl. Environ. Microbiol. 75, 387e394. Rose, T.L., Miagostovich, M.P., Leite, J.P., 2010. Rotavirus A genotype G1P[8]: a novel method to distinguish wild-type strains from the Rotarix vaccine strain. Mem. Inst. Oswaldo Cruz 105, 1068e1072. Ruiz-Palacios, G.M., Pe´rez-Schael, I., Vela´zquez, F.R., Abate, H., Breuer, T., Clemens, S.C., Cheuvart, B., Espinoza, F., Gillard, P., Innis, B.L., Cervantes, Y., Linhares, A.C., Lo´pez, P., Macı´asParra, M., Ortega-Barrı´a, E., Richardson, V., Rivera-Medina, D. M., Rivera, L., Salinas, B., Pavı´a-Ruz, N., Salmero´n, J., Ru¨ttimann, R., Tinoco, J.C., Rubio, P., Nun˜ez, E., Guerrero, M.L., Yarza´bal, J.P., Damaso, S., Tornieporth, N., Sa´ez-Llorens, X., Vergara, R.F., Vesikari, T., Bouckenooghe, A., Clemens, R., De Vos, B., O’Ryan, M., 2006. Human rotavirus vaccine study group. Safety and efficacy of an attenuated vaccine against severe rotavirus gastroenteritis. N. Engl. J. Med. 354, 11e22. Santos, N., Hoshino, Y., 2005. Global distribution of rotavirus serotypes/genotypes and its implication for the development and implementation of an effective rotavirus vaccine. Rev. Med. Virol. 15, 29e56. Schmid, D., Lederer, I., Much, P., Pichler, A.M., Allerberger, F., 2005. Outbreak of norovirus infection associated with contaminated flood water, Salzburg, 2005. Euro. Surveill. 16, 10e16. Sdiri-Loulizi, K., Hassine, M., Aouni, Z., Gharbi-Khelifi, H., Chouchane, S., Sakly, N., Neji-Gue´diche, M., Pothier, P., Aouni, M., Ambert-Balay, K., 2010. Detection and molecular characterization of enteric viruses in environmental samples in Monastir, Tunisia between January 2003 and April 2007. J. Appl. Microbiol. 109, 1093e1104. Solberg, O.D., Hasing, M.E., Trueba, G., Eisenberg, J.N., 2009. Characterization of novel VP7, VP4, and VP6 genotypes of a previously untypeable group A rotavirus. Virology 385, 58e67. Steele, A.D., Reynders, J., Scholtz, F., Bos, P., de Beer, M.C., Tumbo, J., Van der Merwe, C.F., Delem, A., De Vos, B., 2010. Comparison of 2 different regimens for reactogenicity, safety, and immunogenicity of the live attenuated oral rotavirus vaccine RIX4414 coadministered with oral polio vaccine in South African infants. J. Infect. Dis. 202, S93e100. Tamura, K., Dudley, J., Nei, M., Kumar, S., 2007. MEGA4: molecular evolutionary genetics analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24, 1596e1599.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 5 5 e5 7 6 3
Ursu, K., Kisfali, P., Rigo, D., Ivanics, E., Erdelyi, K., Dan, A., Melegh, B., Martella, V., Ba´nyai, K., 2009. Molecular analysis of the VP7 gene of pheasant rotaviruses identifies a new genotype, designated G23. Arch. Virol. 154, 1365e1369. Villena, C., El-Senousy, W.M., Abad, F.X., Pinto, R.M., Bosch, A., 2003a. Group A rotavirus in sewage samples from Barcelona and Cairo: emergence of unusual genotypes. Appl. Environ. Microbiol. 69, 3919e3923. Villena, C., Gabrieli, R., Pinto, R.M., Guix, S., Donia, D., Buonomo, E., Palombi, L., Cenko, F., Bino, S., Bosch, A.,
5763
Divizia, M., 2003b. A large infantile gastroenteritis outbreak in Albania caused by multiple emerging rotavirus genotypes. Epidemiol. Infect. 131, 1105e1110. WHO, 2008. Generic Protocol for Monitoring Impact of Rotavirus Vaccination on Gastroenteritis Disease Burden and Viral Strains. Available at:. WHO www.who.int/vaccines-documents/. Zeng, S.Q., Halkosalo, A., Salminen, M., Szakal, E.D., Puustinen, L., Vesikari, T., 2008. One-step quantitative RT-PCR for the detection of rotavirus in acute gastroenteritis. J. Virol. Methods 153, 238e240.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Managed aquifer recharge of treated wastewater: Water quality changes resulting from infiltration through the vadose zone Elise Bekele a,*, Simon Toze b,c, Bradley Patterson a,d, Simon Higginson e a
CSIRO Water for a Healthy Country Flagship, CSIRO Centre for Environment and Life Sciences, Private Bag No 5, PO Wembley, Western Australia 6913, Australia b CSIRO Water for a Healthy Country Flagship, Ecosciences Precinct, 41 Boggo Road, Dutton Park, QLD 4102, Australia c School of Population Health, University of Queensland, Herston Road, Herston, QLD 4006, Australia d School of Biomedical, Biomolecular and Chemical Sciences, University of Western Australia, Crawley, WA 6009, Australia e Water Corporation of Western Australia, PO Box 100, Leederville, WA 6902, Australia
article info
abstract
Article history:
Secondary treated wastewater was infiltrated through a 9 m-thick calcareous vadose zone
Received 8 June 2011
during a 39 month managed aquifer recharge (MAR) field trial to determine potential
Received in revised form
improvements in the recycled water quality. The water quality improvements of the
23 August 2011
recycled water were based on changes in the chemistry and microbiology of (i) the recycled
Accepted 27 August 2011
water prior to infiltration relative to (ii) groundwater immediately down-gradient from the
Available online 3 September 2011
infiltration gallery. Changes in the average concentrations of several constituents in the recycled water were identified with reductions of 30% for phosphorous, 66% for fluoride,
Keywords:
62% for iron and 51% for total organic carbon when the secondary treated wastewater was
Managed aquifer recharge
infiltrated at an applied rate of 17.5 L per minute with a residence time of approximately
Wastewater infiltration
four days in the vadose zone and less than two days in the aquifer. Reductions were also
Natural attenuation processes
noted for oxazepam and temazepam among the pharmaceuticals tested and for a range of microbial pathogens, but reductions were harder to quantify as their magnitudes varied over time. Total nitrogen and carbamazepine persisted in groundwater down-gradient from the infiltration galleries. Infiltration does potentially offer a range of water quality improvements over direct injection to the water table without passage through the unsaturated zone; however, additional treatment options for the non-potable water may still need to be considered, depending on the receiving environment or the end use of the recovered water. Crown Copyright ª 2011 Published by Elsevier Ltd. All rights reserved.
1.
Introduction
An essential design question for proponents of managed aquifer recharge (MAR) schemes using recycled water in shallow aquifers is whether it is more beneficial to recharge the aquifer via direct well injection or to use infiltration with
either ponds, basins or shallow buried trenches. There is a range of different types of MAR as described in Dillon (2005). Ideally the design of a MAR scheme in an urban setting should have minimal surface footprint and minimal exposure potential for the community. Well injection offers advantages such as no evaporative loss, algae or mosquitoes, and no loss
* Corresponding author. Tel.: þ61 0 8 9333 6718; fax: þ61 0 8 9333 6211. E-mail address: [email protected] (E. Bekele). 0043-1354/$ e see front matter Crown Copyright ª 2011 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.058
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
of prime land (Pyne, 2006). Well injection, however, can suffer from clogging issues relating to the initial quality of the recycled water and loses the benefit of potential treatment processes provided by the unsaturated zone. In comparison, surface infiltration offers potential treatment during the migration of recycled water through the vadose zone, which can result in improved water quality via biological, chemical and physical processes before reaching the water table but can have the drawback of having a larger surface footprint and potential for exposed water, none of which is ideal in an urban environment. Improvements to water quality using infiltration have been demonstrated to reduce organic matter (Quanrud et al., 2003; Vanderzalm et al., 2010), trace organic compounds (Montgomery-Brown et al., 2003), nitrogen (Zhang et al., 2005) and bacteria (Schafer et al., 1998; Toze et al., 2004). However, a rigorous determination of the magnitudes of concentration reductions for a range of different chemical and biological contaminants in treated wastewater has not been conducted previously in MAR studies using unsaturated, calcareous sands such as the Spearwood sands of Swan Coastal Plain of Western Australia. The aim of this study was to determine the changes in recycled water quality after infiltrating vertically through a 9 m-thick vadose zone and migrating laterally through 2.3 m of aquifer using infiltration galleries as the MAR method during a 39 month MAR field trial. The focus of this work is on MAR in calcareous sand and limestone due to the prevalence of these deposits, which comprise unconfined aquifers in the Perth metropolitan region of Western Australia where there is also keen interest in enhancing the role of MAR (Scatena and Williamson, 1999; Smith and Pollock, 2010). The evaluation of water quality was based on measured concentrations of major ions, nutrients, trace metals, organic carbon, pharmaceutical compounds (carbamazepine, diazepam, oxazepam, phenytoin, temazepam) and numbers of faecal indicator microorganisms and selected enteric pathogens (thermotolerant coliforms, enterococci, bacteriophage, adenovirus) before and after recycled water passed through the subsurface.
2.
Materials and methods
2.1.
Facilities
The source of secondary treated wastewater was the Subiaco Wastewater Treatment Plant in Western Australia. After passage through a multi-media filtration system (AMIAD), the treated wastewater was pumped to the recharge site (Bekele et al., 2009). Two infiltration galleries were used at the CSIRO Centre for Environment and Life Sciences in Floreat, Western Australia for a pilot-scale investigation of MAR (Bekele et al., 2009). The east gallery was filled with 10 mm graded and washed granite gravel; the west gallery contained a series of modular polypropylene tanks, referred to as the Atlantis system by the manufacturer. The dimensions of each tank were 685 mm 408 mm 450 mm (length by width by height) and the tanks have a modular design so that they can be positioned in the trench and clipped together. The construction of each module resembles a milk crate that has
5765
large void spaces supported by matrix. In this study, each gallery trench was 25 m 1 m 0.5 m (length by width by height) and the width of the Atlantis gallery was 816 mm, consisting of two tank modules positioned side-by-side. Each gallery was buried to a depth of 1 m below ground and covered with 0.5 m of sediment backfill composed of Spearwood Dune sand from digging the trench (Bekele et al., 2009). The recharge site was located in a grassy paddock where sheep were allowed to graze. The infiltration galleries operated almost continuously for 39 months and infiltrated a total of 36.7 ML of treated wastewater to the aquifer supplied at a daily constant rate of 17.5 L per minute (Bekele et al., 2009).
2.2.
Local geology and hydraulic considerations
The geology of the site consisted of a 7 m-thick top layer of Spearwood Dune sand, overlying the Tamala Limestone aquifer. The Tamala Limestone is a calcareous aeolinite which has been weathered to produce the overlying Spearwood sands (Tapsell et al., 2003). The aquifer extends to a depth of 31 m below ground and is underlain by sediments from a regional aquitard. Regional investigations of the Tamala Limestone reveal high porosity zones from fractures and cavities (Davidson, 1995). The sand mineralogy from 0 to 7.5 m below ground was predominantly quartz (>80%), underlain by a more heterogeneous section from 7.5 to 11.6 m below ground with an average composition of quartz (60%), calcite (30%), microcline (5%) and anorthite (5%) from XRD analysis. Further mineral phase characterization using AutoGeoSEM (Robinson et al., 2000) on a sand sample revealed aluminum and iron oxides. The aluminum oxides are silicate weathering products (clay minerals) deposited as coatings on sand grains, which are colored yellow by the presence of hydrated iron oxides (Bastian, 1996). The depth to the water table below the galleries varied seasonally between 10 and 11 m. The regional groundwater flow direction was from east to west. For experimental purposes, an artificial hydraulic gradient was produced by continual pumping from a well located 50 m west of the infiltration galleries. The natural hydraulic gradient coincided with the imposed gradient. A series of monitoring wells were installed with slotted intervals positioned at different depths below ground as shown in Fig. 1. The migration rate of the infiltrated recycled water through the vadose zone was previously determined based on a bromide tracer experiment (Bekele et al., 2009). From this data, a minimum travel time of 3.7 days was estimated through the unsaturated zone. The travel time to BH1 after passage through the vadose zone was estimated to be an additional 0.5 day based on a comparison of the electrical conductivity of the recycled water relative to groundwater from BH1, indicating that the substantial time for processes to occur was in the vadose zone compared to the aquifer.
2.3.
Description of monitoring wells
To assess the potential infiltration benefits, improvements in quality of the recycled water immediately down-gradient from the infiltration galleries were compared to the quality
5766
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
general accordance with ASTM D 4448-01 Standard Guide for Sampling Ground-Water Monitoring Wells (2001). Wastewater samples were collected from the discharge chamber in the infiltration galleries. Water samples were analyzed for major ions, nutrients, trace metals, dissolved and total organic carbon, trace organics and faecal indicator microorganisms and pathogens. Further details regarding water sampling, analytical procedures and detection limits for the water chemistry are given in Bekele et al. (2009). Faecal indicator microbes E. coli and enterococci were detected using membrane filtration and selective media while the presence of enteric pathogens were determined from a concentrate of 40 L using PCR as described in Toze and Bekele (2009). Sampling water from the galleries and the wells for different chemicals and microbes was conducted during the first 25 months of the MAR trial at different time intervals, but generally on a weekly, fortnightly or monthly basis. Physical water parameters measurements were taken before water sampling occurred. These included pH, dissolved oxygen, electrical conductivity, temperature, oxidationreduction potential and turbidity measurements using a Troll 9000 multisensory meter housed within a flow cell.
2.5. Fig. 1 e The Floreat MAR site showing the positions of the two galleries, the monitoring wells (BH1, BH2 and BH5) and the recovery well in map view (a), and relative to the maximum height of the water table in cross-section (b).
of the infiltrated recycled water. Monitoring wells close to the infiltration galleries (BH1, BH2 and BH5) were selected to evaluate the benefits to water quality of infiltrating recycled water through the vadose zone. Three background groundwater monitoring wells located hydraulically up-gradient of the infiltration galleries (BGRND1, BGRND2 and BGRND3) were also monitored to provide an assessment of changes in groundwater quality due to infiltration. Monitoring well details are given in Table 1.
2.4.
Collection and analysis of water samples
Details of the groundwater sampling procedure are given in Bekele et al. (2009). Groundwater sampling procedures were in
Criteria to assess changes in recycled water quality
To assess the water quality changes as a result of the migration through the vadose zone, the quality of water samples collected from the shallow monitoring wells immediately down-gradient from the infiltration galleries were compared to the quality of the recycled water prior to infiltration. Sampling of the vadose zone water directly below the infiltration galleries and immediately above the water table was not undertaken. Thus, only gross changes after passage over the entire vadose zone could be quantified as changes during infiltration could not be determined. The significance of the effects of migration through the subsurface on recycled water quality was assessed using Student’s two-tailed t-test (unpaired) applied to datasets of measured nutrients, inorganic compounds, pharmaceutical and microbial pathogens, comparing the mean concentrations in recycled water prior to infiltration relative to the mean concentrations in recharged water sampled from BH1. For the Student’s t-test, the null hypothesis was that water quality was unchanged despite passage through the vadose zone. Microsoft Excel was used to
Table 1 e Monitoring well details. Well BH1 BH2 BH5b BGRND1 BGRND2 BGRND3
Ground elevation (m AHDa)
Distance and direction from west gallery
Total depth below ground (m)
Screened depth interval below the maximum height of the water table (m)
12.97 13.01 12.86 15.94 12.00 12.00
2.3 m west 2.5 m west 8.8 m east 185 m northeast 75 m east 75 m east
12.01 12.04 10.81 15.00 12.23 20.11
0.0e2.0 0.0e2.0 0.0e1.0 0.0e2.05 2.23e3.23 10.11e11.11
a Australian Height Datum. b BH5 is located up-gradient relative to the groundwater flow direction and a distance of 4 m east of the east gallery.
5767
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
calculate the probability (P) and the null hypothesis was rejected if P < 0.05.
3.
Results and discussion
3.1. Characterization of recycled water and ambient groundwater To understand what water quality changes were potentially achievable during passage of the recycled water through the vadose zone, it was also necessary to characterise and compare the ambient groundwater. Both the ambient groundwater and the recycled water were consistently aerobic. The average water temperature, pH, electrical conductivity, dissolved oxygen concentrations and sulfate concentrations of the ambient groundwater were similar to that of the recycled water (Table 2). The concentration of total dissolved solids was highly variable in the recycled water and the range of variation overlapped with that of the ambient groundwater. The major ions defining the water types were CaeNaeCleHCO3 for ambient groundwater and NaeCleHCO3 for recycled water. The recycled water and ambient groundwater had low concentrations of several inorganic species, namely arsenic, boron, cadmium, cobalt, chromium, copper, mercury, manganese, molybdenum, nickel, lead, selenium, uranium, vanadium and zinc. The impact of infiltration on these inorganic chemicals was not investigated as their measured concentrations were either very low or were below detection limits.
3.2. MAR impacts on groundwater quality near the galleries Site groundwater monitoring was undertaken to confirm that the shallow groundwater collected closest to the infiltration galleries did represent infiltrated recycled water and not ambient groundwater. Average concentrations of potassium, chloride and sodium in the recycled water were similar to the average concentrations from water samples collected from the monitoring well BH1 (Table 3). The recycled water had an average potassium concentration that was greater than fourfold more enriched in potassium compared to the ambient groundwater and therefore was an effective tracer for the
Table 2 e Mean values of water quality parameters of the recharge water that were similar to ambient groundwater with standard deviations in parentheses. Parameter (Unit of Measure)
Recycled water
Water temperature ( C) pH Electrical Conductivity (mS m1) Eh (mV-SHE) Dissolved Oxygen (mg L1) Sulfate as S (mg L1) Total Dissolved Solids (mg L1)
24 7.33 143 385 2.15 64.1 755
(4) (1.11) (53) (184) (1.82) (8) (179)
Ambient Groundwater 22 (1) 7.04 (0.88) 124 (63) 321 (189) 4.02 (2.56) 64.5 (18) 644 (25)
infiltrated recycled water as demonstrated by others in previous studies (Rueedi et al., 2009; Wolf et al., 2004). The recharged water quality sampled from monitoring wells BH1 and BH2 was indistinguishable from the recycled water prior to infiltration, but substantially different from ambient groundwater on the basis of potassium and chloride concentrations (Fig. 2). Water samples collected from the monitoring well BH5, 4 m up-gradient from the galleries, showed a greater proportion of ambient groundwater mixed with recycled water compared to the water samples from the monitoring wells down-gradient of the galleries. A theoretical line of mixing between the average concentrations of chloride and potassium in ambient groundwater and that of recycled water was generated (Fig. 2). During the first three months of infiltration, the composition of groundwater from BH5 was predominantly ambient groundwater (80%); thereafter, the composition of groundwater from BH5 had a higher proportion of recycled water relative to ambient groundwater. Since water sampled from well BH1 did not show temporal transition and it was frequently monitored, water collected from BH1 was used for evaluating influent water quality improvements as a result of its migration through the vadose zone.
3.3.
Water quality changes during infiltration
3.3.1.
Inorganic chemicals
The recycled water prior to infiltration had higher average concentrations of most species of nitrogen, phosphorous, and fluoride compared to ambient groundwater (Table 3). Recharged water sampled down-gradient from the galleries from BH1 had average concentrations for these chemicals that were generally between those for ambient groundwater and recycled water or within one standard deviation of these two water sources. To assess improvements due to infiltration, total nitrogen concentrations and a relatively non-reactive analyte (i.e.
Table 3 e Mean concentrations (mg/L) of selected chemicals in recycled water and groundwater sampled down-gradient from the galleries from BH1 compared to ambient groundwater. Standard deviations are provided in parentheses. Analtye
Recycled water
Chloride 245 (62) Potassium 22.9 (3.5) Sodium 194 (44) Ammonia as Nitrogen 0.64 (0.85) Nitrate as Nitrogen 2.16 (1.41) Total Kjeldahl Nitrogen 1.86 (0.98) Total Nitrogen 4.27 (1.9) Soluble reactive 6.31 (3.32) phosphorus as P Total organic carbon 9.98 (3.8) Fluoride 0.73 (0.2) Iron 0.14 (0.20) Calcium 28.6 (9.49) Aluminum 0.018 (0.011)
Ambient groundwater
BH1
162 (27) 4.96 (0.83) 92.6 (15) <0.01 0.16 (0.26) 0.066 (0.030) 0.302 (0.36) 0.0126 (0.007)
248 (50) 21.7 (4.1) 194 (42) 0.037 (0.046) 3.90 (1.61) 0.95 (0.51) 4.93 (1.8) 2.19 (1.53)
2.66 (2.72) 0.15 (0.1) 0.44 (0.72) 98.8 (23.5) 0.22 (0.48)
6.42 (3.57) 0.25 (0.05) 0.053 (0.054) 60.4 (7.28) 0.045 (0.048)
5768
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
30
Potassium (mg/L)
25 20 15 10
Ambient groundwater BH1 BH2 BH5, samples during first 3 months BH5, samples after first 3 months Recycled water
5 0 0
100
200
300 Chloride (mg/L)
400
500
Fig. 2 e Cross plot showing concentrations of chloride and potassium in ambient groundwater sampled from the background monitoring wells, recycled water and in water recovered from the monitoring wells (BH1, BH2 and BH5). A mixing line is shown between the average composition of recycled water and ambient groundwater.
potassium) were plotted (Fig. 3). No improvement in total nitrogen was demonstrated as shown by the cluster of data for BH1 in Fig. 3, which overlaps with that of recycled water and indicates a higher range of total nitrogen in contrast to ambient groundwater (<1.2 mg/L). Migration of recycled water through the vadose zone and aquifer to reach BH1 did not improve total nitrogen concentrations (P ¼ 0.24, Student’s ttest). Although passage through the vadose zone did not significantly alter the concentration of total nitrogen in the recycled water, there were changes to different nitrogen species (Table 3). The mean concentration of total Kjeldahl nitrogen in the recharged water sampled from BH1 was 49% lower relative to the mean concentrations in recycled water prior to infiltration by (P < 0.01, Student’s t-test), whereas the mean concentration of ammonia was 94% lower (P < 0 0.01). In comparison, the mean concentration of nitrate was 77% higher (P < 0.001) in the recharged water sampled from BH1 relative to the mean concentrations in recycled water prior to infiltration. This data suggests that nitrification was occurring through the vadose
zone with nitrate production from the ammonium and total Kjeldahl nitrogen, and provides a plausible explanation for the lack of change in total nitrogen concentrations. As conditions in the vadose zone were not conducive for reducing nitrate concentrations, potential treatment options to consider include manipulation of the oxygen content to drive denitrification and the addition of carbon amendments to promote nitrate removal within the aquifer (Patterson et al., 2004, 2011). Migration of the recycled water through the vadose zone and 2.3 m of aquifer aided removal of phosphorous (P < 0.0001, Student’s t-test). Further investigation revealed a timedependence for phosphorous removal. During the first 124 days of infiltration, phosphorous in recharged water sampled from the monitoring well BH1 down-gradient from the galleries remained low, similar to ambient groundwater (<0.02 mg/L), whereas the average phosphorous in the influent recycled water during this initial period was 10 mg/L (Fig. 4). After 271 days of infiltration, the average phosphorous concentration in the influent recycled water was 4.7 mg/L and more variable than previously (standard deviation of 2.1 mg/ L). In comparison, the concentration of phosphorous in recharged water collected from BH1 reached a maximum on day 271 and then stabilized with a mean concentration of 3.2 mg/L (standard deviation of 0.68 mg/L). Comparison of average phosphorous concentrations in BH1 and the influent recycled water reveals a reduction of 31% after 271 days of infiltration. Adsorption and precipitation are reported to be the main causes of phosphorous retention in calcareous sands and soils (von Wandruszka, 2006; Whelan, 1988; Whelan and Barrow, 1984). However, calcareous sands have a limited capacity to store phosphorous, thus caution may be warranted in anticipating consistent levels of phosphorous removal, particularly at sites recharging higher volumes of recycled water and after prolonged recharge. Desorption and remobilization of phosphorous should be considered in dealing with long-term MAR systems. Previous studies have demonstrated the transient nature of phosphorous retention in carbonate soils receiving wastewater. Examples include the release of sorbed phosphate caused by dissolution of carbonate, e.g. from the acidification of leachate during nitrification of ammonium
10 14
Ambient groundwater BH1 Recycled water
8 7
Phosphorous (mg/L)
Total Nitrogen (mg/L)
9
6 5 4 3 2 1
12 10 8 6 4 2
0 0
5
10
15 20 Potassium (mg/L)
25
30
35
Ambient groundwater BH1 Recycled water
Fig. 3 e Cross plot showing concentrations of total nitrogen and potassium in recycled water, ambient groundwater and recharged water sampled down-gradient from the galleries from monitoring well BH1.
0 0
100
200
300
400
500
600
700
Time since start of infiltration (days)
Fig. 4 e Temporal changes in concentrations of phosphorus in recycled water, ambient groundwater and water sampled down-gradient from the galleries from monitoring well BH1.
5769
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
3.3.2.
Organic carbon
Migration of recycled water through the subsurface to reach well BH1 reduced the total organic carbon concentrations (P < 0.01, Student’s t-test). Comparison of average TOC concentrations in BH1 and the influent recycled water reveals a reduction of 51% after 400 days of infiltration (Fig. 5). Probable mechanisms for the removal of organic carbon included filtration, adsorption and biodegradation (Drewes et al., 2003; Fox et al., 2001; Vanderzalm et al., 2010).
30 Total Organic Carbon (mg/L)
(Whelan, 1988), and soil saturated with phosphate becoming a source of phosphate if concentrations in the wastewater recharge decrease (Lin and Banin, 2005). Migration through the subsurface reduced fluoride concentrations in the infiltrated water (P < 0.0001, Student’s ttest). The comparison of average fluoride concentrations in BH1 and the influent recycled water reveals a reduction of 66% (Table 3). Similar to the reduction in phosphorous, reductions in fluoride concentrations were likely due to adsorption and precipitation in calcite-bearing layers (Turner et al., 2005). Adsorption of fluorite was more likely as both the recycled water and ambient groundwater were under-saturated with respect to fluorite. Concentrations of soluble iron in the ambient groundwater sampled from the three background wells were quite variable (mean of 0.44 mg/L; standard deviation of 0.72 mg/L) making it difficult to interpret changes to recycled water quality from passage through the subsurface. Infiltration of the recycled water sustained aerobic conditions favorable for the removal of soluble iron near the gallery. This may explain the nearly three-fold difference between the average concentration of soluble iron in water sampled from well BH1 and recycled water prior to infiltration (P ¼ 0.013, Student’s t-test). Comparison of average iron concentrations in BH1 and the influent recycled water reveals a reduction of 62%. While the pH of the recycled water varied over time during the trial, the pH of the groundwater at BH1 was consistently higher than the recycled water. During infiltration, the average pH of water sampled from well BH1 was 7.61 (standard deviation of 0.80), whereas the average pH of recycled water prior to infiltration was 7.33 (standard deviation of 1.11; Table 2). The increase in pH of the recycled water as it infiltrated through the calcareous sand and limestone was likely due to calcite dissolution in the vadose zone. Calcium and aluminium concentrations in water collected from well BH1 were higher relative to their concentrations in the recycled water; however ambient groundwater had the highest concentrations of these ions (Table 3). The average calcium concentration in the water sampled from BH1 was twice the average concentration in the recycled water prior to infiltration. The recycled water was under-saturated with respect to calcite whereas ambient groundwater was in equilibrium with calcite; hence conditions were favourable for calcite dissolution thereby increasing calcium in water sampled down-gradient from the galleries. The average concentration of aluminium in groundwater sampled from well BH1 was also twice the average aluminium concentration in the recycled water. It is likely that the mobility of aluminium is linked to dissolution of calcium carbonate coated with aluminium oxides.
Ambient groundwater BH1 Recycled water
25 20 15 10 5 0 0
100
200
300
400
500
600
700
Time since start of infiltration (days)
Fig. 5 e Temporal changes in concentrations of total organic carbon in recycled water, ambient groundwater and down-gradient from the galleries in water sampled from monitoring well BH1.
3.3.3.
Pharmaceuticals
All of the pharmaceuticals tested revealed concentrations in ambient groundwater that were below detection limits (i.e. <0.05 mg/L for carbamazepine; <0.01 mg/L for diazepam, oxazepam, phenytoin and temazepam). In comparison, diazepam and phenytoin were the only pharmaceuticals below the detection limit in the recycled water (Table 4). Migration of recycled water through the vadose zone produced reductions in oxazepam and temazepam (P < 0.01, Student’s t-test); however, concentrations of these pharmaceuticals in water collected from BH1 were highly variable. Additional treatment may be required to produce consistent levels of concentration reduction for oxazepam and temazepam, depending on the end use of the extracted water and the potential for large volumes to be consumed intentionally or unintentionally by the community. Sorption and biodegradation are common mechanism for the removal of pharmaceuticals under aerobic conditions in the vadose zone (Amy and Drewes, 2007; Conn et al., 2010; Scheytt et al., 2006; Tiehm et al., 2011), but further investigation of removal mechanisms for the pharmaceuticals was not undertaken in this study. An unexpected observation was that concentrations of carbamazepine in the water collected from well BH1 were consistently higher (80% higher on average; standard deviation of 37%) than carbamazepine detected in the recycled water. The higher carbamazepine concentrations from BH1 relative to recycled water may be due to changes in concentration in the recycled water over time, which varied between 0.13 and 0.33 mg/L (average of 0.21 mg/L, Table 4). In column experiments using aquifer sediment and recycled water from the MAR project and under fully saturated, aerobic conditions, no sorption or degradation were observed for carbamazepine and oxazepam (Patterson et al., 2009). The persistence of carbamazepine in the environment has been demonstrated previously and it has been suggested that it could be used as a potential anthropogenic marker in aquatic environments (Clara et al., 2004). These results may be important for wastewater reuse near wetlands or ecologically-sensitive areas where endocrinedisrupting chemicals could impact negatively on aquatic species in the receiving environment (Lister et al., 2009; Saaristo et al., 2010; Sulleabhain et al., 2009).
5770
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
Table 4 e Mean concentrations (mg/L) of pharmaceuticals in recycled water, water sampled from BH1, and ambient groundwater. Standard deviations are given in parentheses. Analtye
Carbamazepine Diazepam Oxazepam Phenytoin Temazepam
3.3.4.
Recycled water (n ¼ 25)
Ambient groundwater (n ¼ 33)
BH1 (n ¼ 16)
0.21 (0.067) <0.1 0.31 (0.090) <0.1 0.31 (0.084)
<0.05 <0.1 <0.1 <0.1 <0.1
0.38 (0.089) <0.1 0.21 (0.14) <0.1 0.17 (0.12)
Faecal indicator microorganisms and enteric pathogens
It is worth noting that thermotolerant coliforms and enterococci were detected in ambient groundwater from wells upgradient from the recharge site in the sheep paddock. With regard to thermotolerant coliforms, 13% of ambient groundwater samples tested (n ¼ 38) and 19% of water samples from BH1 (n ¼ 27) had numbers greater than 10 cfu 100 mL1. It is hypothesized that surface contamination was most likely due to excreta from the grazing sheep at the recharge site as the source of these microbes and not the treated wastewater being recharged to the aquifer. Despite the additional source of faecal indicator microorganisms from sheep excreta, migration of recycled water from the galleries through the vadose zone to well BH1 produced a reduction in thermotolerant coliforms (P < 0.0001, Student’s t-test). Thermotolerant coliforms were routinely detected in the recycled water with 80% of the samples tested (n ¼ 46), exceeding 10 colony forming units (cfu) 100 mL1. High numbers of enterococci were present in the recycled water with 79% of the water samples tested (n ¼ 48) exceeding 100 cfu 100 mL1, whereas 19% of ambient water samples (n ¼ 42) exceeded 100 cfu 100 mL1. Despite the high counts in the recycled water, there was some reduction in enterococci numbers comparing recycled water and water from BH1 with only 28% of samples from the BH1 (n ¼ 29) exceeding 100 cfu 100 mL1 (P < 0.001, Student’s t-test). Reductions in microbial pathogen numbers in the recycled water were shown by fewer detections of adenovirus and Fþ bacteriophage after migration through the subsurface. Adenovirus was detected in 68% of the samples of the recycled water (n ¼ 19) prior to infiltration but in only 6% of the samples from BH1 (n ¼ 18) and in none of the wells further down gradient. Adenovirus was not detected in any of the ambient groundwater samples. Fþ bacteriophage, commonly used as a surrogate enteric virus, were detected in 94% of recycled water samples tested (n ¼ 36). In comparison only 4% of the water samples from BH1 (n ¼ 24), and 6% of the ambient groundwater samples (n ¼ 33) were positive for this bacteriophage. Reductions in microbial pathogens in treated wastewater recharged to the Tamala Limestone aquifer over time were demonstrated using survival experiments conducted with selected faecal indicators (Cryptosporidium, adenovirus, rotavirus, coxsackievirus, MS2, E. coli, S. enterica and E. faecalis) in in-situ diffusion chambers (Toze et al., 2010). The reductions in
microbial pathogens were attributed to a combination of physical removal processes during filtration and the activity of indigenous groundwater microorganisms (Gordon et al., 2002; Toze and Hanna, 2002). The aquifer has an active treatment capacity to remove pathogens (Toze and Bekele, 2009), but a longer period of aquifer residence may be needed to allow for more inactivation of microbial pathogens (Toze et al., 2010).
4.
Conclusions
Water quality benefits were achieved by infiltrating secondary treated wastewater through weathered calcareous sands and limestone of the vadose zone in an urban area using infiltration galleries. Reductions in the average concentrations of several constituents in the recycled water before and after MAR were identified as follows: 30% for phosphorous, 66% for fluoride, 62% for iron and 51% for total organic carbon. Phosphorous removal declined over time, implying the “maximum” P adsorption capacity was reached. Reductions in the average concentrations of oxazepam and temazepam and the numbers of thermotolerant coliforms and enterococci in the infiltrated water were also identified, but not as easily quantified. The removal rates determined for chemical and biological species are particular to the MAR conditions in this study. The impact of a thicker unsaturated zone, anoxic conditions and different flow rates and contact times were not investigated and would give rise to different removal rates. The aerobic conditions present in the vadose zone were not conducive for denitrification to reduce nitrate concentrations in the water recharged to the aquifer, revealing that secondary treated wastewater recharged via infiltration cannot rely entirely upon processes in the vadose zone for nitrate removal. Geochemical conditions were favorable for calcite dissolution, suggesting that porosity in the calcareous vadose zone may increase over time and could cause faster breakthrough of contaminants to the water table. Future MAR in this aquifer type should consider the effects of prolonged recharge, higher rates of recharge and concentration-dependency of adsorption and precipitation reactions (e.g. controlling phosphorous mobility) to determine the long-term sustainability of recharging recycled water to a calcareous aquifer in urban environments.
Acknowledgements This research was funded by the Western Australian Government through the Water Foundation (gs1), the Water Corporation of WA and the CSIRO Water for a Healthy Country Flagship Program. Chemical analyses were conducted by the Chemistry Centre of Western Australia. The authors would like to thank Mr. Mark Shackleton and Mr. Sebit Gama from the CSIRO for their assistance with water sampling and microbial analyses. Drs Joanne
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
Vanderzalm and Grant Douglas (CSIRO) are gratefully acknowledged for discussions on the geochemistry. The content of this paper benefited from internal reviews by Drs Declan Page and Warish Ahmed at the Commonwealth Scientific and Industrial Research Organization.
references
Amy, G., Drewes, J., 2007. Soil aquifer treatment (SAT) as a natural and sustainable wastewater reclamation/reuse technology: fate of wastewater effluent organic matter (EfOM) and trace organic compounds. Environmental Monitoring and Assessment 129 (1e3), 19e26. ASTM, 2001. Standard Guide for Sampling Ground-water Monitoring Wells D 4448-01. edited. American Society for Testing and Materials (ASTM). Bastian, L., 1996. Residual soil mineralogy and dune subdivision, Swan coastal plain, Western Australia. Australian Journal of Earth Sciences 43 (1), 31e44. Bekele, E., Toze, S., Patterson, B., Devine, B., Higginson, S., Fegg, W., Vanderzalm, J., 2009. Design and operation of infiltration galleries and water quality guidelines - Chapter 1. In: Toze, S., Bekele, E. (Eds.), Determining the Requirements for Managed Aquifer Recharge in Western Australia (A Report to the Water Foundation). Clara, M., Strenn, B., Kreuzinger, N., 2004. Carbamazepine as a possible anthropogenic marker in the aquatic environment: investigations on the behaviour of Carbamazepine in wastewater treatment and during groundwater infiltration. Water Research 38 (4), 947e954. Conn, K.E., Siegrist, R.L., Barber, L.B., Meyer, M.T., 2010. Fate of trace organic compounds during vadose zone soil treatment in an onsite wastewater system. Environmental Toxicology and Chemistry 29 (2), 285e293. Davidson, W.A., 1995. Hydrogeology and groundwater resources of the Perth region, Western Australia, Bulletin. Western Australia Geological Survey 142. Dillon, P., 2005. Future management of aquifer recharge. Hydrogeology Journal 13 (1), 313e316. Drewes, J.E., Reinhard, M., Fox, P., 2003. Comparing microfiltrationreverse osmosis and soil-aquifer treatment for indirect potable reuse of water. Water Research 37 (15), 3612e3621. Fox, P., Narayanaswamy, K., Genz, A., Drewes, J., 2001. Water quality transformations during soil aquifer treatment at the Mesa Northwest water reclamation plant, USA. Water Science and Technology 43 (10), 343e350. Gordon, C., Wall, K., Toze, S., O’Hara, G., 2002. Influence of Conditions on the Survival of Enteric Viruses and Indicator Organisms in Groundwater. In: Paper Presented at Proceedings of the 4th International Symposium on Artificial Recharge of Groundwater Isar-4-Management of Aquifer Recharge for Sustainability. Balkema Publishers, Adelaide SA. Lin, C.Y., Banin, A., 2005. Effect of long-term effluent recharge on phosphate sorption by soils in a wastewater reclamation plant. Water Air and Soil Pollution 164 (1e4), 257e273. Lister, A., Regan, C., Van Zwol, J., Van Der Kraak, G., 2009. Inhibition of egg production in zebrafish by fluoxetine and municipal effluents: a mechanistic evaluation. Aquatic Toxicology 95 (4), 320e329. Montgomery-Brown, J., Drewes, J., Fox, P., Reinhard, M., 2003. Behavior of alkylphenol polyethoxylate metabolites during soil aquifer treatment. Water Research 37 (15), 3672e3681. Patterson, B., Pearce, J., Spadek, T., Linge, K., Busetti, F., Shackleton, M., Furness, A., Blair, P., Heitz, A., 2009. Fate of trace organics using laboratory column experiments - chapter
5771
2. In: Toze, S., Bekele, E. (Eds.), Determining the Requirements for Managed Aquifer Recharge in Western Australia (A Report to the Water Foundation). Patterson, B.M., Grassi, M.E., Robertson, B.S., Davis, G.B., Smith, A. J., McKinley, A.J., 2004. Use of polymer mats in series for sequential reactive barrier remediation of ammoniumcontaminated groundwater: field evaluation. Environmental Science & Technology 38 (24), 6846e6854. Patterson, B.M., Shackleton, M., Furness, A.J., Bekele, E., Pearce, J., Linge, K.L., Busetti, F., Spadek, T., Toze, S., 2011. Behaviour and fate of nine recycled water trace organics during managed aquifer recharge in an aerobic aquifer. Journal of Contaminant Hydrology 122 (1e4), 53e62. Pyne, R.D.G., 2006. Groundwater Recharge and Wells: a Guide to Aquifer Storage Recovery, second ed. CRC Press, Boca Raton, Florida, 401 pp. Quanrud, D., Hafer, J., Karpiscak, M., Zhang, H., Lansey, K., Arnold, R., 2003. Fate of organics during soil-aquifer treatment: sustainability of removals in the field. Water Research 37 (14), 3401e3411. Robinson, B.W., Hitchen, G.J., Verrall, M.R. 2000, The AutoGeoSEM: A programmable fully-automatic SEM for rapid grain-counting and heavy mineral characterisation in exploration, paper presented at International Mineralogical Association Commission on Ore mineralogy, Proceeding of the Modern Approaches to Ore and Environmental mineralogy, Short Course, p. 71e74. Rueedi, J., Cronin, A.A., Morris, B.L., 2009. Estimation of sewer leakage to urban groundwater using depth-specific hydrochemistry. Water and Environment Journal 23 (2), 134e144. Saaristo, M., Craft, J.A., Lehtonen, K.K., Lindstrom, K., 2010. An endocrine disrupting chemical changes courtship and parental care in the sand goby. Aquatic Toxicology 97 (4), 285e292. Scatena, M.C., Williamson, D.R. 1999, A potential role for artificial recharge within the Perth region: a pre-feasibility study. Center for Groundwater Studies report No. 84. Schafer, A., Ustohal, P., Harms, H., Stauffer, F., Dracos, T., Zehnder, A., 1998. Transport of bacteria in unsaturated porous media. Journal of Contaminant Hydrology 33 (1e2), 149e169. Scheytt, T.J., Mersmann, P., Heberer, T., 2006. Mobility of pharmaceuticals carbamazepine, diclofenac, ibuprofen, and propyphenazone in miscible-displacement experiments. Journal of Contaminant Hydrology 83 (1e2), 53e69. Smith, A.J., Pollock, D.W., 2010. Artificial Recharge Potential of the Perth Region Superficial Aquifer: Lake Preston to Moore River. CSIRO (Water for a Healthy Country National Research Flagship Report). Sulleabhain, C.O., Gill, L.W., Misstear, B.D.R., Johnston, P.M., 2009. Fate of endocrine-disrupting chemicals in percolating domestic wastewater effluent. Water and Environment Journal 23 (2), 110e118. Tapsell, P., Newsome, D., Bastian, L., 2003. Origin of yellow sand from Tamala limestone on the Swan Coastal Plain, Western Australia. Australian Journal of Earth Sciences 50 (3), 331e342. Tiehm, A., Schmidt, N., Stieber, M., Sacher, F., Wolf, L., Hoetzl, H., 2011. Biodegradation of pharmaceutical compounds and their occurrence in the Jordan Valley. Water Resources Management 25 (4), 1195e1203. Toze, S., Bekele, E., 2009. Determining Requirements for Managed Aquifer Recharge in Western Australia. A Report to the Water Foundation. CSIRO, Canberra, AU. http://www.csiro.au/files/ files/py7j.pdf. Toze, S., Bekele, E., Page, D., Sidhu, J., Shackleton, M., 2010. Use of static Quantitative Microbial Risk Assessment to determine pathogen risks in an unconfined carbonate aquifer used for Managed Aquifer Recharge. Water Research 44 (4), 1038e1049. Toze, S., Hanna, J., 2002. The Survival Potential of Enteric Pathogens in a Reclaimed Water ASR Project. In: Paper
5772
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 6 4 e5 7 7 2
Presented at Proceedings of the 4th International Symposium on Artificial Recharge of Groundwater ISAR-4-Management of Aquifer Recharge for Sustainability. Balkema Publishers, Adelaide, SA, pp. 139e142. Toze, S., Hanna, J., Smith, T., Edmonds, L., McCrow, A., 2004. Determination of water quality improvements due to the artificial recharge of treated effluent. In: Steenvoorden, J., Endreny, T. (Eds.), International Symposium on Wastewater Re-Use and Groundwater Quality. Int Assoc Hydrological Sciences, Wallingford, UK, pp. 53e60. Turner, B., Binning, P., Stipp, S., 2005. Fluoride removal by calcite: evidence for fluorite precipitation and surface adsorption. Environmental Science & Technology 39 (24), 9561e9568. Vanderzalm, J., Page, D., Barry, K., Dillon, P., 2010. A comparison of the geochemical response to different managed aquifer recharge operations for injection of urban stormwater in a carbonate aquifer. Applied Geochemistry 25 (9), 1350e1360.
von Wandruszka, R., 2006. Phosphorus retention in calcareous soils and the effect of organic matter on its mobility. Geochemical Transactions 7 (6). doi:10.1186/14674866-7-6. Whelan, B.R., 1988. Disposal of septic tank effluent in calcareous sand. Journal of Environmental Quality 17 (2), 272e277. Whelan, B.R., Barrow, N.J., 1984. The movement of septic-tank effluent through sandy soils near Perth .2. Movement of phosphorus. Australian Journal of Soil Research 22 (3), 293e302. Wolf, L., Held, I., Eiswirth, M., Hotzl, H., 2004. Impact of leaky sewers on groundwater quality. Acta Hydrochimica Et Hydrobiologica 32 (4e5), 361e373. Zhang, J., Huang, X., Liu, C., Shi, H., Hu, H., 2005. Nitrogen removal enhanced by intermittent operation in a subsurface wastewater infiltration system. Ecological Engineering 25 (4), 419e428.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Removal of mercury (II) by dithiocarbamate surface functionalized magnetite particles: Application to synthetic and natural spiked waters P. Figueira a, C.B. Lopes b,*, A.L. Daniel-da-Silva a, E. Pereira b, A.C. Duarte b, T. Trindade a a b
CICECO & Department of Chemistry, University of Aveiro, Campus de Santiago, 3810-193 Aveiro, Portugal CESAM & Department of Chemistry, University of Aveiro, Campus de Santiago, 3810-193 Aveiro, Portugal
article info
abstract
Article history:
In order to take advantage of the high affinity between mercury and sulphur, magnetite
Received 24 March 2011
(Fe3O4) particles functionalized with dithiocarbamate groups (CS 2 ), were synthesized to be
Received in revised form
used as a new type of sorbent to remove Hg (II) from synthetic and natural spiked waters.
22 July 2011
The effectiveness of this type of sorbent was studied, and its potential as cleanup agent for
Accepted 27 August 2011
contaminated waters was assessed.
Available online 3 September 2011
Batch stirred tank experiments were carried out by contacting a volume of solution with known amounts of functionalized Fe3O4 particles, in order to study the effect of sorbent
Keywords:
dose, salinity, and the kinetics and the equilibrium of this unit operation. A complete Hg (II)
Mercury
removal (ca. 99.8%) was attained with 6 mg/L of magnetic particles for an initial metal
Magnetite particles
concentration of 50 mg/L. It was confirmed that highly complex matrices, such as seawater
Dithiocarbamate functionalization
(ca. 99%) and river water (ca. 97%), do not affect the removal capacity of the functionalized
Isotherms
magnetic particles. Concerning isotherms, no significant differences were observed
Water remediation
between two- and three-parameter models (P ¼ 0.05%); however, Sips isotherm provided the lowest values of SS and Sx/y, predicting a maximum sorption capacity of 206 mg/g, in the range of experimental conditions under study. The solid loadings measured in this essay surmount the majority of the values found in literature for other type of sorbents. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The increasing awareness for the effects of pollution, has led to strict environmental regulations, such as the creation of a list of 33 priority hazardous substances by European Parliament and the Council of the European Union (EU 2008), and the instigation of the Member States to implement the necessary actions aiming at a progressive reduction of pollutants, and ceasing or phasing out emissions and discharges of priority hazardous substances, as considered in the Framework Directive (EU 2000).
Mercury and its compounds are one of the most dangerous contaminants in the environment, threatening the human health and natural ecosystems. They are included in the list of priority hazardous substances and consequently, the removal of Hg and its compounds, particularly, from aquatic systems is a major goal of wastewater treatment and cleanup technologies. Conventional techniques for Hg removal from aqueous solutions include sulphate or hydrazine precipitation, ionexchange, liquideliquid extraction, adsorption and solid phase extraction via activated carbon adsorption (Starvin and Rao, 2004).
* Corresponding author. Tel.: þ351 234 370 721; fax: þ351 234 370 084. E-mail address: [email protected] (C.B. Lopes). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.057
5774
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
Nomenclature C V M q k1 k2 kL qm kF n Nt a m kRP
Concentration of Hg (II) in bulk solution, mg/L Volume of solution, L Dry weight of Fe3O4/SiO2/NH/CS 2 particles, mg Amount of Hg (II) in the particles, mg/g First-order rate constant, h1 Second-order rate constant, g/mg h Langmuir constant, L/g Maximum loading of Fe3O4/SiO2/NH/CS 2 particles, mg/g Freundlich parameter, (mg11/n$L1/n/g) Freundlich parameter Total number of binding sites, mg/g Sips constant Heterogeneous index Redlich-Peterson constant, L/g
The high toxicity of Hg to organisms is generally attributed to the blockage of the enzyme binding sites and interference in protein synthesis (Starvin and Rao, 2004), due to the high affinity of Hg to bind with any molecule that has sulphur or a sulphurehydrogen combination in its structure. A promising strategy to achieve high-performance materials able to capture Hg (II) from aqueous solutions is to take advantage of its high affinity to sulphur, developing new materials with sulphur-containing functional groups such as diverse thiolates. This approach has been investigated by Antochshuk et al. (2003) and materials with high surface area have been used as an insoluble matrix for the attachment of sulphurcontaining groups (Antochshuk et al., 2003). Several materials such as mesoporous silica (Antochshuk et al., 2003; Venkatesan et al., 2003; Mattigod et al., 1999; Mercier and Pinnavaia, 1998; Feng et al., 1997; Mureseanu et al., 2010), silica gel (Venkatesan et al., 2002), activated carbon (Starvin and Rao, 2004) and organoceramic composites (Nam et al., 2003) have been used as matrices for the attachment of sulphur-containing groups, such as thiol (Mattigod et al., 1999; Mercier and Pinnavaia, 1998; Nam et al., 2003), dithiocarbamate (Venkatesan et al., 2003, 2002), 1-(2-thiazolylazo)-2naphthol (Starvin and Rao, 2004), mercaptopropylsilane (Feng et al., 1997), 1-furoyl thiourea urea (Mureseanu et al., 2010) and benzoythiourea (Antochshuk et al., 2003). Those materials have surface coverage of ligands typically between 2.5 104 and 3.7 103 mol/g, and some of them exhibit very high Hg (II) loading capacities (Antochshuk et al., 2003). We have been interested on magnetic particles, in particular iron oxides (e.g. Fe3O4), to develop new materials for environmental and bio-applications (Daniel-da-Silva et al., 2007). Iron oxides such as maghemite and magnetite offer convenient magnetic properties, low toxicity and price, high surface to volume ratios, and possibility for surface chemical modification (Girginova et al., 2010). Keeping in mind the interesting properties of magnetic magnetite and the higher values of the stability constants reported for Hgdithiocarbamate (Dtc) complexes (e.g. 1.2 1038 for Hg(Et2Dtc)2 (Venkatesan et al., 2002)), we have recently reported the synthesis and the preliminary application of silica
aRP b RL ARE SS CV-AFS DF RE RSD Sx/y
Redlich-Peterson constant, (L/mg)b Redlich-Peterson exponent Separation factor Average relative error Sum of squares Cold vapour atomic fluorescence spectroscopy Degrees of freedom Relative Error Relative Standard Deviation Standard Deviation of Residues
Subscripts 0 initial condition of experiment t intermediate condition of the experiment at a certain time e equilibrium condition of experiment
coated Fe3O4 particles functionalized with Dtc groups for the decontamination of synthetic waters with realistic Hg (II) levels (Girginova et al., 2010). However, this work also raised a number of important issues related with the performance of these materials as colloidal adsorbents in natural waters, particularly in strong salinity conditions. We wish to report here our research on the removal of Hg (II) at trace levels found in natural waters, and the effect of sorbent dose and salinity on the sorption efficiency. The results of this study on the sorption equilibrium and kinetics of Hg (II) onto the functionalized magnetic particles (MPs) were fitted to well-known kinetic and equilibrium equations. For the assessment of the real effectiveness of the developed sorbent, the functionalized magnetic NPs were applied in two natural waters (seawater and river water).
2.
Material and methods
2.1.
Chemicals
All chemicals used in this work were of analytical reagent grade, and obtained from commercial chemical suppliers and were used without further purification. The certified standard stock solution of mercury (II) nitrate was purchased from Merck (1000 2 mg/L).
2.2.
Adsorbent material
2.2.1.
Synthesis
Fe3O4/SiO2/NH/CS 2 magnetic particles were investigated for remediation of Hg (II) contaminated waters. The synthesis of these magnetic particles includes three distinct steps: (i) the synthesis of magnetite particles (Fe3O4), and their encapsulation in amorphous silica shells (Fe3O4/SiO2), (ii) further modification with 3-aminopropyltriethoxysilane (Fe3O4/SiO2/ NH2) and (iii) the grafting of dithiocarbamate groups at the surface of amine modified silica coated magnetite (Fe3O4/SiO2/ NH/CS 2 ).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
The first two steps of the synthesis were performed exactly as described by Girginova et al. (Girginova et al., 2010), while the procedure adopted for the grafting of dithiocarbamates groups was slightly different in order to graft a higher amount of particles. This procedure was performed as follows: the amine modified silica coated magnetite (50 mg) was added to 15 mL 2-propanol under mechanic stirring for 1 h. After, 0.75 mL 1 M NaOH, and 0.12 mL CS2 were added and the suspension was mechanically stirred for 4 h. The powder was then collected magnetically from the suspension formed, washed with 2-propanol and dried at room temperature.
2.2.2.
Structural and chemical characterization
Fourier Transform Infrared (FT-IR) spectra of Fe3O4/SiO2/NH/ CS 2 magnetic particles were recorded using a spectrometer Mattson 7000 with 256 scans and 4 cm1 resolution, using a horizontal attenuated total reflectance (ATR) cell. Elemental analysis for carbon, nitrogen, hydrogen and sulphur were performed using a LECO CHNS-932 elemental analyzer. The crystalline phase of the particles was identified by X-ray powder diffraction of the dried samples using a Philips X’Pert X-ray diffractometer equipped with a Cu Ka monochromatic radiation source. Transmission electron microscopy (TEM) was performed using a transmission electron microscope JEOL 200CX operating at 300 kV. The specific surface area of the magnetic particles was determined with nitrogen adsorption BET measurements performed with a Gemini Micromeritics instrument.
2.3.
Batch adsorption experiments
The ability of Fe3O4/SiO2/NH/CS 2 particles to capture Hg (II) from water was evaluated by contacting the particles with a Hg (II) solution for a required period of time. The sorption experiments were carried out in 500 mL batch reactors at 295 1 K, under mechanical stirring. Hg (II) solutions were prepared daily by diluting the corresponding standard solution, in high purity water (18 MUcm), to the desired initial concentration (50 mg/L), with a subsequent adjustment of pH to 7 with 0.1 M NaOH. The initial concentration of 50 mg/L was selected, as this is the current limit value for Hg discharges from industrial sectors other than the chloro-alkali electrolysis industry. All glassware used in these experiments was acid-washed prior to use with HNO3 25%, 12 h, and ultra-pure water. Experiments were performed to evaluate the time profile and the effects of sorbent concentration and salinity on the sorption of Hg (II) onto the functionalized magnetic particles. Accurately weighed amounts of Fe3O4/SiO2/NH/CS 2 particles were added to Hg (II) solutions, in glass batch reactors, which were immediately placed in an ultrasonic bath for ca. 10 s, for dispersing the magnetic particles. This time was considered the starting point for each experiment. For every experiment, the Hg (II) solution was continuously stirred, using a glass rod, and samples were withdrawn for analysis at several sampling times. Each sample was analyzed for the Hg (II) concentration after magnetic separation of the particles using a NdFeB magnet (1.48 T), and posterior pH adjustment (lower than 2) with concentrate HNO3 (Hg free, Merck). Mercury analyses were performed by cold vapour atomic fluorescence
5775
spectroscopy (CV-AFS), on a flow-injection-cold vapour atomic fluorescence spectrometer (Hydride/vapour generator PS Analytical Model 10.003, coupled to a PS Analytical Model 10.023 Merlin atomic fluorescence spectrometer; PS Analytical, Orpington, Kent, England) and using SnCl2 as reducing agent. An Hg (II) solution in the absence of magnetic particles was run as a control experiment. The time profile of Hg (II) sorption onto Fe3O4/SiO2/NH/CS 2 particles and the effect of sorbent dose, were determined by using four different amounts of the functionalized particles, namely 0.124, 0.256, 0.501 and 3.063 mg, which correspond to a sorbent dose of 0.2, 0.5, 1 and 6 mg/L. In order to study the effects of salinity changes on the sorption capacity of Fe3O4/SiO2/NH/CS 2 , 6 mg/L of particles were added to three different matrices spiked with 50 mg Hg (II)/L: ultra-pure water, 3 g/L NaCl solution and seawater collected from about 1 nautical mile from the Portuguese coast. In the following sections, those matrices will be denoted in terms of salinity (0% of salinity, 10% of salinity and 100% of salinity, respectively). The efficiency of Hg (II) removal by Fe3O4/SiO2/NH/CS 2 particles from different types of natural waters was also assessed by adding 6 mg/L of particles to river water collected from Vouga River (near the drinking water treatment plant) and spiked with 50 mg Hg (II)/L. Isotherms were obtained varying the amount of Fe3O4/ SiO2/NH/CS 2 particles from 0.124 to 3.063 mg, for an initial Hg (II) concentration of 50 mg/L and a temperature of 295 1 K. The amount of Hg (II) sorbed per unit of particles, at time t, qt (mg/g) was estimated from the mass balance between initial Hg (II) concentration and concentration at time t in solution, qt ¼ ðC0 Ct Þ
V M
(1)
where V is the volume of solution (L) and M is the dry weight of Fe3O4/SiO2/NH/CS 2 particles (mg), C0 (mg/L) is the initial Hg (II) concentration and Ct (mg/L) is its concentration at time t. The results were also compared by removal percentage, which at equilibrium time is defined by: Removal% ¼
ðC0 Ce Þ 100 C0
(2)
where, Ce (mg/L) is the equilibrium Hg (II) concentration in the solution.
2.4.
Kinetic and equilibrium models
Different theoretical models were applied to experimental data in order to find a model which adequately describes kinetic and equilibrium data.
2.4.1.
Kinetic models
The chemical kinetics gives relevant information about reaction pathways and rates, along with time to reach equilibrium. Moreover, the physical and chemical features of the sorbent have great impact on both sorption kinetics and sorption mechanism (Hasan and Srivastava, 2009).
5776
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
In order to evaluate the differences in the kinetic rates and to describe the kinetic removal of Hg (II) ions onto Fe3O4/SiO2/ NH/CS 2 particles in three matrices tested (ultra-pure water, NaCl solution and seawater), two simple and widely applied models, the pseudo-first- and pseudo-second-order models were used to describe the removal process. Numerous studies in literature have pointed out that one of those two models is able to describe the majority of the sorption studies (Wang and Chen, 2009; Ho and McKay, 1999).
2.4.1.1. Pseudo-first-order equation or Lagergren model. The first-order equation, firstly applied by Lagergren (Lagergren, 1898), is mathematically expressed by: dqt ¼ k1 qe qt dt
(3)
(4)
2.4.1.2. Pseudo-second-order equation. The pseudo-secondorder equation may be also applicable and, in contrast with the previous model, usually correlates the behaviour over the whole range of sorption (Ho and McKay, 1999). The kinetic rate equation is expressed as: 2 dqt ¼ k2 qe qt dt
(5)
where k2 (g/mg h) is the rate constant of pseudo-second-order. By applying the boundary conditions t¼0 to t¼t and qt¼0 to qt¼qe, the integrated form of Eq. (5) is: qt ¼
k2 q2e t 1 þ k2 qe t
2.4.2.2. Langmuir isotherm. This two-parameter model developed by Langmuir in 1916, relating the amount of gas sorbed on a surface to the pressure of the gas is probably one of the best known and widely used sorption isotherm (Ho et al., 2002). This model assumes that exist a fixed number of accessible sites available on the sorbent surface, all of them with the same energy and that the sorption is reversible. The Langmuir model is represented by the following equation: qe ¼
where k1 (h1) is the rate constant of pseudo-first-order and qe (mg/g) is the amount of solute sorbed per gram of sorbent at equilibrium. After integration and application of the boundary condition qt¼0 at t¼0, Equation (3) becomes qt ¼ qe 1 ek1 t
favourable adsorption, and is related to the non-linearity of the model.
(6)
qm kL Ce 1 þ kL Ce
(8)
where kL (L/mg) and qm (mg/g) are the Langmuir sorption equilibrium constant related to the energy of sorption and the maximum sorption capacity corresponding to complete monolayer coverage, respectively.
2.4.2.3. Sips or Langmuir-Freundlich isotherm. As the own name indicates, this three-parameter isotherm is a composite of the former isotherms. At high sorbate concentrations, it predicts a monolayer sorption capacity, characteristic of the Langmuir isotherm and at low sorbate concentrations it reduces to a Freundlich isotherm (Ho et al., 2002). The Sips equation can be describe as follows: qe ¼
Nt aCm e 1 þ aCm e
(9)
where Nt (mg/g) is the total number of binding sites, a is related to the median binding affinity (k), since a¼km and m is the heterogeneous index, which varies from 1 (homogeneous material) to 0 (heterogeneous material for m < 1) (Umpleby et al., 2001).
2.4.2.4. Redlich-Peterson 2.4.2.
Equilibrium models
Sorption equilibrium provides fundamental data for evaluating the applicability of the sorption process. Equilibrium isotherms give the equilibrium relationships between sorbent and sorbate, i.e. the quantity of solute sorbed and remaining in solution at a given temperature, at equilibrium. The equation parameters and the underlying thermodynamic assumptions of the equilibrium models often provide some insight into the sorption mechanism, the surface properties and the capacity and affinity of the sorbent (Ho et al., 2002). In this study, twoparameter isotherms (Langmuir and Freundlich) and threeparameter isotherms (Sips and Redlich-Peterson), in their nonlinear form were chosen to fit the experimental data.
2.4.2.1. Freundlich isotherm. This empirical model developed by Freundlich in 1906 (Freundlich, 1906), can be applied to multilayer sorption as well as non ideal sorption on heterogeneous surfaces and is represented by the following equation: qe ¼ kF Ce1=n
(7)
where kF (mg(11/n) L(1/n)/g) and n are the Freundlich parameters. n is usually between 1 and 10, which points out
isotherm. This three-parameter isotherm, which also incorporates characteristics of both the Langmuir and Freundlich isotherms, can be represented as follows: qe ¼
kRP Ce 1 þ aRP Cbe
(10)
where kRP (L/g) and aRP ((L/mg)b) are the Redlich-Peterson constants and b is the Redlich-Peterson exponent.
2.4.3.
Error analysis
The parameters of the kinetic and equilibrium models here considered were obtained by nonlinear regression analysis using GraphPad Prism 5 program, which uses the least-squares as fitting method and the method of Marquardt and Levenberg, which blends two other methods, the method of linear descent and the method of Gauss-Newton for adjusting the variables. In order to confirm which model presents the best fit to experimental data, the adjusted R squared (R0 2), the sum of squares (SS) and the standard deviation of residues (Sx/y) were analyzed. The standard error of the best fit parameters, the average relative error (ARE) of the fittings and the relative error (RE) between the experimental qe value and the models’ estimation value are also presented. The different statistical
5777
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
parameters were determined using the following mathematical expressions: (11)
^
yi y i
(13)
(440)
(511)
(400)
2000
(422)
2
Counts (a.u.)
X
(12)
(220)
3000
^ 2 yi y i R2 ¼ 1 P 2 yi yi
P
SS ¼
(311)
n1 np1
(111)
R02 ¼ 1 1 R2
1000
(P Sx=y ¼
^
yi y i n2
2 )1=
2
(14)
0 20
^ P jyi y i j yi 100 ARE ¼ n jyi y i j 100 yi
(15)
Results and discussion
3.1.
Characterisation of Fe3O4/SiO2/NH/CS 2 particles
In the present work, surface modified Fe3O4 particles were firstly prepared using methodologies previously described by us (Girginova et al., 2010). Fig. 1 and Table 1 summarizes relevant results obtained for the characterization of these materials and that have been collected here for monitoring the materials properties. In brief, the powder X-ray diffraction patterns of the materials match those of magnetite,1 whose identity was unequivocally proved by Mo¨ssbauer spectroscopy (Girginova et al., 2010). As expected, TEM analysis (Fig. 1) showed single cubic particles of magnetite coated with amorphous silica shells. As previously reported, the cubic particles exhibited a magnetization hysteresis loop at room temperature with a saturation magnetization of 62 emu/g (Girginova et al., 2010). Finally, the surface functionalization of the silica coated magnetite particles was monitored step by step using infrared spectroscopy and the results are summarized in Table 1. Due to the higher value of the stability constant for the HgDtc complex is expectable that Hg (II) loading onto the functionalized particles will depend of the extent of functionalization carried out on the sorbent. Based on surface area (14.6 m2/g), sulphur content (0.98%) and taking into account that each Dtc group has two sulphur atoms, the amount of Dtc 1
50
60
70
Fig. 1 e X-ray powder diffraction patterns of Fe3O4 and TEM image of silica coated Fe3O4 particles. The Miller indices corresponding to the most intense reflection peaks of Fe3O4 are indicated in brackets.
(16)
where n is the sample size, p is the number of adjustable parameters in the model and yi are the experimental or observed values, yi is the mean of the observed data and y^i are the modelled or predicted values. The absolute error for each experimental point from the kinetics and equilibrium experiments can be found in the complementary file.
3.
40
2 θ (°)
^
RE
30
Joint Committee for Powder Diffraction Studies, JCPDS 19-0629.
groups on the sorbent was found to be 1.53 104 mol/g or 1.05 105 mol/m2. This value is slightly lower but of the same order of magnitude than the density of Dtc groups found for other sorbents like mesoporous silica grafted with Dtc (Venkatesan et al., 2003) (2.5 104 mol/g) and silica gel grafted with Dtc (Venkatesan et al., 2002) (3.7 104 mol/g).
3.2.
Hg (II) removal by Fe3O4/SiO2/NH/CS 2 particles
3.2.1.
Time profile of Hg (II) removal
Kinetic curves corresponding to the sorption of Hg (II) onto different amounts of Fe3O4/SiO2/NH/CS 2 particles are shown in Fig. 2. The plots represent the normalized Hg (II) concentration that remained in the liquid phase vs. time in Fig. 2A and the Hg (II) concentration in the MPs vs. time in Fig. 2B, both for an initial concentration of 50 mg Hg (II)/L. For all the amounts of Fe3O4/SiO2/NH/CS 2 particles, a decrease with time on Hg (II) concentration in the liquid phase was observed, even when only 0.2 mg/L were used. The time profile curves show that equilibrium time is attained in 24 h for the highest amount of particles used (6 mg/L), and as expected, it increases as long as the amount of particles decreases. This increase is particularly notorious by comparing the equilibrium time for the lowest and the highest amount of Fe3O4/SiO2/NH/CS 2 ; for the lowest
Table 1 e Assignment of infrared diagnosis bands for Fe3O4 and surface modified Fe3O4 particles. Sample Fe3O4 Fe3O4/SiO2
Fe3O4/SiO2/NH2 Fe3O4/SiO2/NH/CS 2
Wave number (cm1)
Assignment
540, 310 (s) 1070 (vs) 789 (w) 943 (w) 786 (w) 1438 (s)
n(FeeO) n(SiOeSi) n(SieOH) n(SieOeFe) d(NeH) n(CeN)
(vs-very strong, s-strong, w-weak; n-stretching vibration, d-bending vibration).
5778
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
amount of MPs (0.2 mg/L) the equilibrium time was achieved only after 96 h, while for the highest amount of MPs (6 mg/L) the equilibrium was achieved in less than 24 h (Fig. 2). This difference suggests that equilibrium time can be drastically affected by small variations in sorbent amount, and it is mainly related with the large availability of the Dtc groups. The time profile curves also reveal that during the first 12 h, the Hg (II) content in the particles increased quickly and then slowed down approaching equilibrium (Fig. 2B).
3.2.2. The effect of Fe3O4/SiO2/NH/CS 2 concentration on Hg (II) removal The concentration of sorbent is an important parameter to obtain quantitative metal removal, since for a given initial concentration of metal, it influences both the contact time necessary to reach equilibrium and the sorption capacity (Lopes et al., 2009). Fig. 3 shows the equilibrium removal percentage and the equilibrium amount of Hg (II) sorbed by gram of Fe3O4/SiO2/ NH/CS 2 vs. concentration of Fe3O4/SiO2/NH/CS2 . Given the obtained results it can be concluded that the particles concentration strongly affects Hg (II) removal and Hg (II) uptake per gram of particles. It is noticeable, that at equilibrium, the removal percentage increases with the increasing of
A
1.0 6 mg/L 1 mg/L
0.8
0.5 mg/L 0.2 mg/L
Ct/C0
0.6
Fe3O4/SiO2/NH/CS 2 concentration, from 53.2 1.3% (for 0.2 mg/L of particles) to 99.8 1.5% (for 6 mg/L of particles) while the amount of Hg (II) sorbed per gram of Fe3O4/SiO2/NH/ CS 2 particles decreases drastically from112 3 mg/g to 9.2 0.1 mg/g. Thus, for an invariable initial Hg (II) concentration, the increase of the sorbent concentration provides greater contact surface area and increases the number of available sorption sites, promoting the removal of Hg (II). As the number of sorption sites increases, the concentration of Hg (II) in the liquid phase and the amount of Hg (II) per mass of sorbent decreases and more sorption sites remain unsaturated during the sorption process. Additionally, the increase of the amount of Hg (II) per mass of MPs will only occur as long as the maximum capacity of the MPs is not fulfilled. This fact, suggest that under the experimental conditions tested, the maximum capacity of the magnetic particles was not achieved. Another remarkable result was the achievement of a considerably low residual concentration (0.10 0.02 mg Hg (II)/L) with only the application of a few milligrams of sorbent (6 mg/L). The residual Hg (II) concentration achieved is ten times lower than the current guideline value of the European Union for water of drinking quality ([Hg (II)] 1 mg/L) (EU 1998). This result clearly evidence the huge potential of this magnetic material to decontaminate waters and to achieve totally Hg free effluents, as is required by the Water Framework Directive (EU 2000). Based on the obtained results, 6 mg of Fe3O4/SiO2/NH/CS 2 particles per litre of Hg (II) solution (50 mg/L) should be used for a completely effectiveness of the decontamination process, since any further increase in the particles dose will not significantly affect the equilibrium between the ions sorbed in the solid phase and those remaining in solution.
0.4
3.2.3. 0.2
0.0
B
The effect of salinity on Hg (II) removal
It is well know that in the presence of chloride ions, Hg forms chloro-complexes of Hg (II) and consequently some sorbents become useless in saline waters (Lopes et al., 2007). The removal of Hg (II) by Fe3O4/SiO2/NH/CS 2 particles was studied in the presence of two different Cl concentrations (ca. 10 and
120 120
120 Removal.. (%)
100
qe (mg/L)
100
100
80
80
60
60
40
40
20
20
qt , mg/g
80 60 40 20 0 0
24
48
72
96
120
t, h Fig. 2 e Normalized Hg (II) concentration in the liquid phase (A) and sorbed on the particles (B) as a function of time, for different amounts of Fe3O4/SiO2/NH/CSL 2 particles.
0
0 0.2
0.5
1.0
6.0
0.2
0.5
1.0
6.0
Fe 3O4/SiO2/NH/CS2- concentration (mg/L)
Fig. 3 e Effect of Fe3O4/SiO2/NH/CSL 2 concentration on Hg (II) removal (%) and sorption capacity (mg/g).
5779
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
A
1.0 0% 10% 100%
0.8
Ct /C0
0.6
0.4
0.2
0.0
B
10
qt , mg/g
8
6
4
2
0 0
24
48
72
96
120
t, h Fig. 4 e Normalized Hg (II) concentration in the liquid phase (A) and sorbed on the particles (B) as a function of time, for different percentages of salinity.
100% of salinity) and the results, which are shown in Fig. 4, were also compared with those obtained for ultra-pure water (0% of salinity). The plots in Fig. 4A represent the normalized Hg (II) concentration remaining in the liquid phase vs. time, while the Hg (II) concentration in the MPs vs. time is represented in Fig. 4B. According with the results, the presence of high Cl concentrations did not change significantly the equilibrium
values of the amount of Hg (II) sorbed per gram of Fe3O4/SiO2/ NH/CS 2 particles (qe values ranged from 9.08 to 9.35 mg/g, with a RSD ca. 1.5%), neither the equilibrium removal percentage, which was higher than 98% for all matrices (0, 10 and 100% of salinity). Even the residual concentration of Hg (II) in solution was lower than the current guideline value of the European Union for water of drinking quality (1 mg/L), for all three systems. For 0% of salinity, i.e. for ultra-pure water the residual concentration achieved was 0.10 0.02 mg Hg (II)/L, for 10% of salinity it was 0.97 0.01 mg Hg (II)/L and for 100% of salinity, i.e. for seawater the residual concentration achieved was 0.82 0.02 mg Hg (II)/L. Conversely, the presence of high Cl concentrations clearly increased the equilibrium time from 24h (0% of salinity) to ca. 96h (10 and 100% of salinity) (see Fig. 4) and the rate which Hg (II) is removed by the particles decreases with the increasing of the matrix’s complexity, i.e. from ultra-pure water to seawater. These results suggest that besides, the equilibrium time, the matrix of the system Hg (II)/Fe3O4/SiO2/NH/CS 2 particles also influences the kinetic rate of the sorption process, as confirmed by the rate constants (k1 and k2) of the pseudo-firstand pseudo-second-order kinetic models (see Table 2). The kinetic sorption rate constants k1 and k2 decrease with the increasing of the salinity of the matrix, which reinforces that the higher is the ionic strength, the slower equilibrium is reached. The obtained results also suggest that during the first hours not all Hg (II) is available for the active sites of the Fe3O4/ SiO2/NH/CS 2 particles, probably due to the formation of mercury chloro-complexes. However, as the equilibrium values of the amount of Hg (II) removed per gram of functionalized particles (qe) did not change significantly for the different matrices, it is expectable that as the concentration of the Hg (II) available in solution decreases the decomplexation of the chloro-complexes occurs, since they are less stable (Ks 107-1015) (Nam et al., 2003) than the complexes formed between mercury and the Dtc groups (Ks 1038) (Venkatesan et al., 2002), increasing the concentration of Hg (II) available in solution, for complexing with the Dtc groups present at the surface of the magnetic particles. The results also allow to conclude that Naþ do not compete with Hg (II) for the active þ sites of the Fe3O4/SiO2/NH/CS 2 particles, even when the Na concentration was much higher than that of Hg (II). Furthermore, the modelling of the kinetic process by the pseudo-first- and pseudo-second-order models allows to conclude that whatever the kinetic equation used, the description of the sorption kinetics was satisfactory for all
Table 2 e First- and second-order sorption rate constants obtained for the removal of Hg (II) from matrices with different percentage of seawater, together with experimental and fitted qe, and the goodness of the of the fittings. Kinetic model Pseudo - 1st order 0% 10% 100% Pseudo - 2nd order 0% 10% 100%
Model’s parameters k1 h1 0.397 0.183 0.154 k2 g/mg h 0.0667 0.0259 0.0224
SE 0.037 0.016 0.028 SE 0.0077 0.0032 0.0042
qe mg/g 9.10 8.89 8.78 qe mg/L 9.61 9.67 9.59
Goodness of the fit SE 0.19 0.22 0.45 SE 0.18 0.24 0.37
R0 2 0.98 0.98 0.92 R2 0.99 0.99 0.97
SS 1.53 1.69 6.85 SS 0.90 1.29 2.71
Sx/y 0.44 0.43 0.87 Sx/y 0.33 0.38 0.55
ARE % 4.4 10 18 ARE % 3.7 7.2 12
Experimental values DF 8 9 9 DF 8 9 9
qe mg/g 9.22 9.08 9.35 qe mg/g 9.22 9.08 9.35
RE % 1.3 2.1 6.1 RE % 2.6 2.3 2.2
5780
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
matrices, as confirmed by the high values of R0 2 (0.92e0.99) and low values of Sx/y (0.33e0.87) (see Table 2). It is clear that both kinetic models describe very well the quantity of Hg (II) removed in the early stages of sorption, although there are slightly deviations between the experimental and the fitted data, particularly in the inflexion zone (Fig. 5). However, both models are able to accurately estimate the qe value of the three systems, which is corroborated by the low values of RE (1.3e6.1%) found between the predicted and the experimental qe values in all matrices. Still, comparing the predicted and experimental qe values it is perceptible that the Langergren model underestimates the qe values while the pseudo-secondorder model overestimates them. Although these slight differences, for a confidence level of 95% there is no significant difference between the goodness of the fit of the two models (F-test), for all tested matrices.
3.3.
The Langmuir equation assumes that adsorption occurs at definite localized sites on the surface, each site being able to bind a single molecule of the adsorbing species. The energy of adsorption is equal for all sites and there are no interaction forces between adjacently adsorbed molecules (Cooney, 1999). Theoretically, a saturation value is reached, beyond which no further sorption can take place, which is represented by a plateau in the equilibrium isotherm and corresponds to the assumption of one complete monomolecular layer of coverage
Sorption equilibrium
An accurate mathematical description of the equilibrium data between the concentration of the sorbate in the liquid and the amount in the solid phase is essential for a consistent prediction of the sorption parameters and for quantitative comparison of the sorption capacity of different sorbents. This mathematical function, called isotherm, is a basic requirement for designing any sorption system (Marin et al., 2009), and is obtained for a specific temperature and initial sorbate concentration. Fig. 6 shows the sorption isotherm of Hg (II) onto Fe3O4/ SiO2/NH/CS 2 particles, as well as the fit to the isotherm models described in the experimental section. The parameters of the isotherm models obtained from the corresponding fittings are presented in Table 3. All isotherms are positive and concave to the concentration axis, and under the experimental conditions here used, the experimental equilibrium values of the amount of Hg (II) sorbed in the functionalized particles increases with the increasing of Hg (II) concentration in solution, without reaching a saturation plateau. The Freundlich isotherm is an empirical equation, which does not assume that the material coverage must approach a constant value corresponding to one complete solute monomolecular layer as Ce gets larger. This model predicts that qe monotonously increases with increasing Ce which, being physically impossible, means that the Freundlich equation should fail to describe the experimental data at high Ce values (Cooney, 1999). However, the concentration values in real sorption processes are considered sufficiently diluted, in order to avoid the process entering the region where the Freundlich equation breaks down (Cooney, 1999). According with the obtained results, the Freundlich model provides a good description of the experimental data (ARE¼5.8%, R0 2¼0.98), since in the range of experimental conditions used, the equilibrium data do not achieved a plateau at a limiting value of Ce, suggesting the existence of heterogeneous surface conditions. Freundlich isotherm allows to calculate two empirical constants, kF and n, which are related to adsorption capacity of the sorbent and sorption intensity, respectively. The magnitudes of kF (601 mg(11/n) L(1/n)/g) and n (2.26) indicate easy separation of Hg (II) from liquid phase and favourable sorption (1 < n < 10).
0 ,
Fig. 5 e Sorption kinetics modelling of Hg (II) on the particles, for different percentages of seawater.
5781
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
monolayer capacity (qm) of Fe3O4/SiO2/NH/CS 2 particles estimated by the Langmuir model is 142 mg/g, with a 95% confidence interval of [115e170 mg/g], and the Langmuir constant (kL), which represents the affinity between the sorbent and sorbate is 145 L/mg. Moreover, the isotherms can be classified as irreversible when the separation factor (RL) is 0; favourable when 0 < RL¼<1, linear when RL ¼ 1 and unfavourable when RL > 1. The separation factor (RL) can be calculated using the follow equation:
120
Two parameter models 100
qe, mg/g
80 60 40 20 0 0.000
RL ¼
Langmuir Freundlich
0.005
0.010
0.015
0.020
0.025
120
Three parameter models 100
qe, mg/g
80 60 40 Sips Redlich-Peterson
20 0 0.000
0.005
0.010
0.015
0.020
0.025
Ce , mg/L Fig. 6 e Equilibrium isotherms of Hg (II) on the particles at 21 ± 1 C.
of the adsorbing species on the adsorbent (Cooney, 1999; Kocaoba, 2007). Langmuir plot shows a higher difference between experimental equilibrium data and the predicted ones (ARE¼14%), although no significant differences were observed between the goodness of the fit of the Freundlich and Langmuir models (F-test for 95% confidence level). The
1 1 þ kL C0
The calculated RL value was 0.12, corroborating that Hg (II) sorption on Fe3O4/SiO2/NH/CS 2 particles is favourable. The Langmuir-Freundlich isotherm, also known as Sips isotherm, is an equation with three fitting coefficients with physical meaning (Nt, a and m), that describes the relationship of the equilibrium concentration of the sorbate between the solid and liquid phase in heterogeneous systems (Umpleby et al., 2001). As the name implies, this isotherm is a combination of the Langmuir and Freundlich isotherms. At low sorbate concentrations it can be reduced to the Freundlich isotherm, while at high sorbate concentrations, it predicts a monolayer sorption capacity characteristic of the Langmuir isotherm (Ho et al., 2002; Marin et al., 2009). This isotherm is capable of modelling both homogeneous and heterogeneous binding surfaces. The value of exponent m on Sips equation was 0.705, which means that Hg (II) sorption onto the functionalized particles is more of Langmuir type than that of Freundlich, since for m¼1 the Sips equation (Eq. (9)) reduces to the Langmuir equation (Eq. (8)) in which the variable a corresponds directly to binding affinity (k). The predicted value of Nt was higher than the corresponding value (qm) of the Langmuir model (Table 3). Redlich-Peterson isotherm also incorporate features of both Langmuir and Freundlich equations, approximating to Henry’s law at low concentrations (b¼0), while at higher concentrations its behaviour approaches that of the Freundlich isotherm (Ho et al., 2002). For b¼1 the RedlichPeterson isotherm reduces to the Langmuir form. In this study, the value of the exponent b approximates to 1 (0.860), suggesting like the Sips equation, that the equilibrium data can preferably be fitted by the Langmuir model rather the Freundlich. Among the three-parameter models, the Sips isotherm provides the highest value of R0 2 and the lowest values of ARE,
Table 3 e Isotherm constants of two- and three-parameter models for Hg (II) sorption on magnetic particles at 21 ± 1 C. Isotherm Freundlich Langmuir Sips Redlich-Peterson
Model’s parameters kF mg(11/n)L(1/n)/g 601 kL L/mg 145 a L/g 16.9 kRP L/g 26188
SE 109 SE 29 SE 30.7 SE 13485
n 2.26 qm mg/g 142 Nt mg/g 206 aRP (L/mg)b 114
Goodness of the fit SE 0.21 SE 11 SE 101 SE 44
m 0.705 b 0.860
SE 0.211 SE 0.227
R0 2i 0.98 R2 0.98 R2 0.98 R2 0.98
SS 148 SS 130 SS 102 SS 122
Sx/y 5.43 Sx/y 5.11 Sx/y 4.51 Sx/y 4.93
ARE 5.8 ARE 14 ARE 10 ARE 14
% % % %
DF 5 DF 5 DF 4 DF 4
5782
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
SS and Sx/y, suggesting that this model is more appropriate to describe the experimental equilibrium data than the RedlichPeterson. Among the two-parameter models, the Langmuir is in agreement with the experimental data and with the variables obtained from the three-parameter models and provides higher value of R0 2 and lower values of SS and Sx/y than the Freundlich. However, the comparison (F-test) between the values of Sx/y obtained for all models indicates that for 95% confidence level there are no significant differences between the goodness of the fit of the different models, in the range of experimental conditions studied (Ce [0e0.025 mg/L]).
3.4. Application of Fe3O4/SiO2/NH/CS 2 particles in natural waters Besides the application of Fe3O4/SiO2/NH/CS 2 particles to seawater, already discussed in a former section, the feasibility of the functionalized particles were also tested in river water. Likewise for the seawater, the time profile curve (plot not shown) reveals a decrease on Hg (II) concentration with time. However, probably due to a higher complexity of the matrix and higher levels of organic matter, the equilibrium time in the river water was attained in 240 h against the 96 h necessary to achieve equilibrium in seawater. The percentage of Hg (II) removal obtained was ca. 97% for river water and ca. 99% for seawater, while the residual Hg (II) concentration in the liquid phase was, respectively, 1.20 0.07 and 0.82 0.02 mg Hg (II)/L. Those results suggest that in the case of river water, the amount of particles employed should be slightly higher than 6 mg/L in order to achieve a completely effectiveness of the decontamination process in this type of natural waters; however, it must be highlight the high effectiveness of Fe3O4/ SiO2/NH/CS 2 particles to remove Hg (II) from water, even from
natural waters, where the high complexity of the matrix could undermine their performance.
3.5.
Comparison with other sorbents
The effectiveness of the Fe3O4/SiO2/NH/CS 2 particles by means of maximum Hg (II) sorption capacities, is quantitatively compared with other sorbents, particularly with algal biomass (Tuzun et al., 2005), Romanian clays (Hristodor et al., 2010), ETS-4 titanosilicate (Lopes et al., 2009) and furfural carbon (Yardim et al., 2003) (Table 4). The maximum Hg (II) sorption capacities of these sorbents range between 122 and 246 mg/g and are of the same order of magnitude of that found for Fe3O4/SiO2/NH/CS 2 particles. Compared to other sorbents containing Dtc ligands, the maximum Hg (II) sorption capacity of Fe3O4/SiO2/NH/CS 2 particles is considerable higher than that found for mesoporous silica (MCM-41-Dtc) (Venkatesan et al., 2003) and silica gel (Si-Dtc) (Venkatesan et al., 2002), despite the lower Dtc surface coverage (respectively 2.5 104 and 3.7 104 mol/g against 1.5 104 mol/g of the magnetic particles). Comparable binding capacities were observed between Fe3O4/SiO2/ NH/CS-2 particles and other sorbents, like mesoporous silica grafted with other sulphur ligands as thiol (Mercier and Pinnavaia, 1998), mercaptan (Mattigod et al., 1999) and 1furoyl thiourea urea (Mureseanu et al., 2010) (see Table 4). While sorbents like benzoylthiourea-modified mesoporous silica (Antochshuk et al., 2003) and thiol functional organoceramic composite (Nam et al., 2003), with higher sulphur surface coverage (ca. 103 mol/g) and/or multifunctional ligands, possess several “active” groups toward mercury ions and exhibit considerable higher sorption capacities, on the other hand they do not offer the possibility of magnetic separation.
Table 4 e Residual mercury concentration (Ce) and sorption capacity (qm) of other sorbents for Hg (II).
Biosorbents
Clays and zeolitic materials Carbons
Sorbents containing sulphur ligands
n.a. not available.
Adsorbent
qm (mg/g)
Ce (mg/L)
Ref.
Rice husk ash Bacillus sp. Eucalyptus bark Seaweed biomass Yeast cells Algal biomass Zeolitic mineral Clay ETS-4 titanosilicate Activated carbon Carbon aerogel Activated carbon Furfural carbon Thiol functional organoceramic composite Dithiocarbamate grafted on mesoporous silica Dithiocarbamate grafted on silica gel Thiol functionality grafted on mesoporous silica Mesoporous silica containing mercaptan groups Benzoylthiourea-modified mesoporous silica Mesoporous silica grafted with 1-furoyl thiourea urea Dithiocarbamate grafted on magnetite particles
6.72 7.94 33.1 84.7 93.4 122 10.1 152 246 25.8 34.9 43.8 174 726 40.1 61 110e301 26e270 1000 122 142e206
n.a 20 n.a n.a n.a n.a n.a n.a <1 n.a w5000 n.a n.a <1 n.a. n.a. n.a. 401 n.a. n.a. <1
(Feng et al., 2004) (Green-Ruiz, 2006) (Ghodbane and Hamdaoui, 2008) (Zeroual et al., 2003) (Yavuz et al., 2006) (Tuzun et al., 2005) (Gebremedhin-Haile et al., 2003) (Hristodor et al., 2010) (Lopes et al., 2009) (Rao et al., 2009) (Goel et al., 2005) (Ranganathan, 2003) (Yardim et al., 2003) (Nam et al., 2003) (Venkatesan et al., 2003) (Venkatesan et al., 2002) (Mercier and Pinnavaia, 1998) (Mattigod et al., 1999) (Antochshuk et al., 2003) (Mureseanu et al., 2010) This study
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
It must also be highlighted that the Fe3O4/SiO2/NH/CS 2 particles together with thiol functional organoceramic composite and ETS-4 titanosilicate were the only sorbents, mentioned in Table 4, able to reduce the initial mercury concentration to values lower than 1 mg/L (Table 4). This fact is due not only to their sorption capacity but also because the majority of the studies use initial Hg concentrations extremely high and nothing realistic of the degree of contamination found in the environment.
4.
Conclusions
The sorption capacity towards Hg (II) of silica coated magnetite particles derivatized with dithiocarbamate groups was studied in batch mode under different experimental conditions. It was confirmed that silica coated magnetite particles functionalized with Dtc groups are effective sorbents for Hg (II) removal from synthetic and natural waters; however the sorption process is strongly dependent on contact time and particles concentration. The presence of higher concentrations of Cl and Naþ ions did not affect the amount of Hg (II) removed per gram of sorbent at equilibrium but reduced the rate at which Hg ions were removed from solution, as confirmed by pseudo-first- and pseudo-second-order kinetic models. The Sips model provides a good description of the equilibrium data, predicting a maximum sorption capacity of 206 mg/g at 211 C, which is quite high compared with the sorption capacities found in the literature for other materials. Furthermore, this study confirmed the effectiveness of silica coated magnetite particles grafted with Dtc groups in two distinct types of natural waters, seawater and river water. However, the functionalized magnetite particles exhibited a slightly higher performance in seawater than in river water. It must be highlighted that in both ultra-pure water and seawater, only 6 mg/L of functionalized particles was sufficient to achieve a residual concentration lower than 1 mg/L, which is the current acceptable value for drinking water quality. Our results emphasize the advantages of these Dtc functionalized particles, such as high affinity towards mercury ions, selective removal, and large efficiency in high complex matrices. Additionally, the easy separation of the sorbent from solution due to the magnetite ferrimagnetic properties opens new prospects in the design of highperformance sorbents for environment remediation and mercury ions recovery.
Acknowledgements C.B. Lopes thanks Fundac¸a˜o para a Cieˆncia e Tecnologia for a Post-Doc grant (SFRH/BPD/45156/2008).
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.08.057.
5783
references
Antochshuk, V., Olkhovyk, O., Jaroniec, M., Park, I.S., Ryoo, R., 2003. Benzoylthiourea-modified mesoporous silica for mercury(II) removal. Langmuir 19 (7), 3031e3034. Cooney, D.O., 1999. Adsorption Design for Wastewater Treatment. Lewis Publishers, Boca Raton (FL). Daniel-da-Silva, A.L., Trindade, T., Goodfellow, B.J., Costa, B.F.O., Correia, R.N., Gil, A.M., 2007. In situ synthesis of magnetite nanoparticles in carrageenan gels. Biomacromolecules 8 (8), 2350e2357. EU, 1998. Council Directive 98/83/EC on the quality of water intended for human consumption. Official Journal 330, 0032e0054. EU, 2000. Directive 2000/60/EC of the European Parliament and of the Council of the European Union, establishing a framework for Community action in the field of water policy. Official Journal of the European Communities 327/1. EU, 2008. Directive 2008/105/EC of the European Parliament and of the Council of the European Union, on environmental quality standards in the field of water policy, amending and subsequently repealing Council Directives 82/176/EEC, 83/513/ EEC, 84/156/EEC, 84/491/EEC, 86/280/EEC and amending Directive 2000/60/EC of the European Parliament and of the Council. Official Journal of the European Communities 348/84. Feng, X., Fryxell, G.E., Wang, L.Q., Kim, A.Y., Liu, J., Kemner, K.M., 1997. Functionalized monolayers on ordered mesoporous supports. Science 276 (5314), 923e926. Feng, Q.G., Lin, Q.Y., Gong, F.Z., Sugita, S., Shoya, M., 2004. Adsorption of lead and mercury by rice husk ash. Journal of Colloid and Interface Science 278 (1), 1e8. Freundlich, H., 1906. Concerning adsorption in solutions. Zeitschrift Fur Physikalische ChemieeStochiometrie Und Verwandtschaftslehre 57 (4), 385e470. Gebremedhin-Haile, T., Olguin, M.T., Solache-Rios, M., 2003. Removal of mercury ions from mixed aqueous metal solutions by natural and modified zeolitic minerals. Water Air and Soil Pollution 148 (1e4), 179e200. Ghodbane, I., Hamdaoui, O., 2008. Removal of mercury(II) from aqueous media using eucalyptus bark: kinetic and equilibrium studies. Journal of Hazardous Materials 160 (2e3), 301e309. Girginova, P.I., Daniel-Da-Silva, A.L., Lopes, C.B., Figueira, P., Otero, M., Amaral, V.S., Pereira, E., Trindade, T., 2010. Silica coated magnetite particles for magnetic removal of Hg2þ from water. Journal of Colloid and Interface Science 345 (2), 234e240. Goel, J., Kadirvelu, K., Rajagopal, C., Garg, V.K., 2005. Investigation of adsorption of lead, mercury and nickel from aqueous solutions onto carbon aerogel. Journal of Chemical Technology and Biotechnology 80 (4), 469e476. Green-Ruiz, C., 2006. Mercury(II) removal from aqueous solutions by nonviable Bacillus sp from a tropical estuary. Bioresource Technology 97 (15), 1907e1911. Hasan, S.H., Srivastava, P., 2009. Batch and continuous biosorption of Cu2þ by immobilized biomass of Arthrobacter sp. Journal of Environmental Management 90 (11), 3313e3321. Ho, Y.S., McKay, G., 1999. Pseudo-second order model for sorption processes. Process Biochemistry 34 (5), 451e465. Ho, Y.S., Porter, J.F., McKay, G., 2002. Equilibrium isotherm studies for the sorption of divalent metal ions onto peat: copper, nickel and lead single component systems. Water Air and Soil Pollution 141 (1e4), 1e33. Hristodor, C., Copcia, V., Lutic, D., Popovici, E., 2010. Thermodynamics and kinetics of Pb(II) and Hg(II) ions removal from aqueous solution by Romanian clays. Revista De Chimie 61 (3), 285e289.
5784
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 7 3 e5 7 8 4
Kocaoba, S., 2007. Comparison of Amberlite IR 120 and dolomite’s performances for removal of heavy metals. Journal of Hazardous Materials 147 (1e2), 488e496. Lagergren, S., 1898. Zur theorie der sogenannten adsorption gelo¨ter stoffe. Kungliga Svenska Vetenskapsakademiens. Handlingar 24 (4), 1e39. Lopes, C.B., Otero, M., Coimbra, J., Pereira, E., Rocha, J., Lin, Z., Duarte, A., 2007. Removal of low concentration Hg2þ from natural waters by microporous and layered titanosilicates. Microporous and Mesoporous Materials 103 (1e3), 325e332. Lopes, C.B., Otero, M., Lin, Z., Silva, C.M., Rocha, J., Pereira, E., Duarte, A.C., 2009. Removal of Hg2þ ions from aqueous solution by ETS-4 microporous titanosilicate-kinetic and equilibrium studies. Chemical Engineering Journal 151 (1e3), 247e254. Marin, A.B.P., Aguilar, M.I., Meseguer, V.F., Ortuno, J.F., Saez, J., Llorens, M., 2009. Biosorption of chromium (III) by orange (Citrus cinensis) waste: batch and continuous studies. Chemical Engineering Journal 155 (1e2), 199e206. Mattigod, S.V., Feng, X.D., Fryxell, G.E., Liu, J., Gong, M.L., 1999. Separation of complexed mercury from aqueous wastes using self-assembled mercaptan on mesoporous silica. Separation Science and Technology 34 (12), 2329e2345. Mercier, L., Pinnavaia, T.J., 1998. Heavy metal lan adsorbents formed by the grafting of a thiol functionality to mesoporous silica molecular sieves: Factors affecting Hg(II) uptake. Environmental Science and Technology 32 (18), 2749e2754. Mureseanu, M., Reiss, A., Cioatera, N., Trandafir, I., Hulea, V., 2010. Mesoporous silica functionalized with 1-furoyl thiourea urea for Hg(II) adsorption from aqueous media. Journal of Hazardous Materials 182 (1e3), 197e203. Nam, K.H., Gomez-Salazar, S., Tavlarides, L.L., 2003. Mercury(II) adsorption from wastewaters using a thiol functional adsorbent. Industrial and Engineering Chemistry Research 42 (9), 1955e1964. Ranganathan, K., 2003. Adsorption of Hg(II) ions from aqueous chloride solutions using powdered activated carbons. Carbon 41 (5), 1087e1092.
Rao, M.M., Reddy, D., Venkateswarlu, P., Seshaiah, K., 2009. Removal of mercury from aqueous solutions using activated carbon prepared from agricultural by-product/waste. Journal of Environmental Management 90 (1), 634e643. Starvin, A.M., Rao, T.P., 2004. Removal and recovery of mercury(II) from hazardous wastes using 1-(2-thiazolylazo)-2-naphthol functionalized activated carbon as solid phase extractant. Journal of Hazardous Materials 113 (1e3), 75e79. Tuzun, I., Bayramoglu, G., Yalcin, E., Basaran, G., Celik, G., Arica, M.Y., 2005. Equilibrium and kinetic studies on biosorption of Hg(II), Cd(II) and Pb(II) ions onto microalgae Chlamydomonas reinhardtii. Journal of Environmental Management 77 (2), 85e92. Umpleby, R.J., Baxter, S.C., Chen, Y.Z., Shah, R.N., Shimizu, K.D., 2001. Characterization of molecularly imprinted polymers with the Langmuir-Freundlich isotherm. Analytical Chemistry 73 (19), 4584e4591. Venkatesan, K.A., Srinivasan, T.G., Rao, P.R.V., 2002. Removal of complexed mercury from aqueous solutions using dithiocarbamate grafted on silica gel. Separation Science and Technology 37 (6), 1417e1429. Venkatesan, K.A., Srinivasan, T.G., Rao, P.R.V., 2003. Removal of complexed mercury by dithiocarbamate grafted on mesoporous silica. Journal of Radioanalytical and Nuclear Chemistry 256 (2), 213e218. Wang, J.L., Chen, C., 2009. Biosorbents for heavy metals removal and their future. Biotechnology Advances 27 (2), 195e226. Yardim, M.F., Budinova, T., Ekinci, E., Petrov, N., Razvigorova, M., Minkova, V., 2003. Removal of mercury (II) from aqueous solution by activated carbon obtained from furfural. Chemosphere 52 (5), 835e841. Yavuz, H., Denizli, A., Gungunes, H., Safarikova, M., Safarik, I., 2006. Biosorption of mercury on magnetically modified yeast cells. Separation and Purification Technology 52 (2), 253e260. Zeroual, Y., Moutaouakkil, A., Dzairi, F.Z., Talbi, M., Chung, P.U., Lee, K., Blaghen, M., 2003. Biosorption of mercury from aqueous solution by Ulva lactuca biomass. Bioresource Technology 90 (3), 349e351.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Influence of operational parameters on nitrogen removal efficiency and microbial communities in a full-scale activated sludge process Young Mo Kim a, Hyun Uk Cho b, Dae Sung Lee c, Donghee Park d,*, Jong Moon Park b,** a
Department of Civil and Environmental Engineering, University of Massachusetts, Amherst, MA 01003, USA Department of Chemical Engineering, School of Environmental Science and Engineering, Division of Advanced Nuclear Engineering, Pohang University of Science and Technology, Pohang 790-784, Republic of Korea c Department of Environmental Engineering, Kyungpook National University, Daegu 702-701, Republic of Korea d Department of Environmental Engineering, Yonsei University, Wonju 220-710, Republic of Korea b
article info
abstract
Article history:
To improve the efficiency of total nitrogen (TN) removal, solid retention time (SRT) and
Received 5 May 2011
internal recycling ratio controls were selected as operating parameters in a full-scale
Received in revised form
activated sludge process treating high strength industrial wastewater. Increased biomass
17 August 2011
concentration via SRT control enhanced TN removal. Also, decreasing the internal recy-
Accepted 29 August 2011
cling ratio restored the nitrification process, which had been inhibited by phenol shock
Available online 3 September 2011
loading. Therefore, physiological alteration of the bacterial populations by application of specific operational strategies may stabilize the activated sludge process. Additionally, two
Keywords:
dominant ammonia oxidizing bacteria (AOB) populations, Nitrosomonas europaea and
Activated sludge
Nitrosomonas nitrosa, were observed in all samples with no change in the community
Operational parameter
composition of AOB. In a nitrification tank, it was observed that the Nitrobacter populations
Nitrogen removal
consistently exceeded those of the Nitrospira within the nitrite oxidizing bacteria (NOB)
qPCR
community. Through using quantitative real-time PCR (qPCR), nirS, the nitrite reducing
Nitrifying bacteria
functional gene, was observed to predominate in the activated sludge of an anoxic tank,
Denitrifying bacteria
whereas there was the least amount of the narG gene, the nitrate reducing functional gene. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Activated sludge process is one of the most widely used biological treatments of wastewaters containing carbon and nitrogen pollutants. Biological nitrogen removal has traditionally been accomplished using autotrophic nitrification and heterotrophic denitrification. Nitrification is carried out in two sequential steps via two distinct groups of bacteria: ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB). Denitrification consists of consecutive
reactions in which nitrate or nitrite is transformed into gaseous forms (N2 or N2O). Although efficient and reliable in treating industrial wastewater, activated sludge process is susceptible to disturbances and toxic loadings (Juliastuti et al., 2003; Mertoglu et al., 2008; Kim et al., 2009). In particular, the activity of nitrifying bacteria in wastewater treatment plants is sensitive to shifts in the process’s pH and temperature, ammonia/ nitrite concentrations, oxygen concentration and the presence of toxic compounds, often leading to process failure
* Corresponding author. Tel.: þ82 33 760 2435; fax: þ82 33 760 2571. ** Corresponding author. Tel.: þ82 54 279 2275; fax: þ82 54 279 2699. E-mail addresses: [email protected] (D. Park), [email protected] (J.M. Park). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.063
5786
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
(Eighmy and Bishop, 1989; Kim et al., 2007). Fluctuation inflow, organic loads, solid retention time (SRT) and lack of nutrient matter also inhibit the nitrification process (Hallin et al., 2005). Even though nitrifying bacteria have been fairly well studied, nitrification is difficult to maintain stably throughout the activated sludge process (Kim et al., 2009). Actually, many current full-scale activated sludge processes treating industrial wastewater have long experienced trouble with instability, but previous reports have simply focused on monitoring the emergence of the process failure (Kim et al., 2007, 2008, 2009). The causes of the process instability have not been clearly identified; thus appropriate solutions have not yet been suggested (Kim et al., 2011a,b). Moreover, research targeting a full-scale wastewater treatment process (WWTP) has rarely been attempted due to either sampling problems or difficulties in proper WWTP selection. Results from lab scale experiments have proved difficult to extrapolate to real WWTP conditions (Kim et al., 2007, 2009). In addition, little is known about the microbial ecology of nitrifying and denitrifying bacteria in full-scale activated sludge processes treating high strength industrial wastewater containing toxic compounds like thiocyanate, ammonia and phenol. Therefore, this study aimed to identify the causes underlying current difficulties achieving and maintaining the legal discharge of total nitrogen (TN) in the final effluent of a full-scale activated sludge process. To improve the efficiency of TN removal, SRT and internal recycling ratio controls as main operation parameters were attempted in an unstable process. Meanwhile, bacterial populations responsible for biological nitrogen removal were investigated in relation to changes in the efficiency of TN removal of a full-scale activated sludge process treating wastewater from a coke plant. Diversity surveys assessing the relationship between bacterial populations and activity to the overall processing conditions may lead to an understanding of the basis of process instability and be of help in designing better process monitoring while avoiding operational failure.
2.
Materials and methods
2.1.
Wastewater treatment plant operation and samples
nitrified effluent e the pre-denitrification activated sludge process. This system is composed of an anoxic tank (900 m3), an aerobic tank (3600 m3) and a nitrification tank (1800 m3) with a combined volume of 6300 m3 and treats about 300 m3 of cokes wastewater per hour (Fig. 1). Nitrified effluent was recycled from the nitrification tank to the anoxic tank at the rate of about 600 m3 per hour. Return activated sludge collected from the secondary clarifier was pumped back to the anoxic tank at the rate of about 300 m3 per hour. Both the aerobic and nitrification tanks were aerated. In the course of this study, the concentrations of pollutants in the raw wastewater were as follows: 2025e3150 mg/L of chemical oxygen demand (COD), 680e1070 mg/L of biochemical oxygen demand (BOD5), 642e916 mg/L of total organic carbon (TOC), 268e715 mg/L of phenol, 177e236 mg-N/L of total nitrogen (TN), 76e140 mg-N/L of ammonia, 190e297 mgSCN/L of thiocyanate (SCN) and 12.4e17.2 mg-CN/L of total cyanides. The mixed-liquor suspended solids (MLSS) of the system were controlled at about 1800 mg/L, the average hydraulic retention time (HRT) was 0.9 day and the average SRT was 15 days. The SRT was controlled by removal of excess sludge, resulting in different MLSS concentrations of the system. The temperature range of the tanks varied between 33 and 36 C. The pH of the influent, anoxic tank and nitrification tank was maintained at 9.0, 7.5 and 7.0, respectively. The dissolved oxygen (DO) concentration in the aerobic and nitrification tank was more than 4.0 mg/L, while the DO level of the anoxic tank was maintained below 0.3 mg/L. Functional stability of the system was defined and quantified by the effluent concentration of TN. MLSS samples for this study were taken from the last sections of the anoxic and nitrification tanks weekly for 3 months (AugusteOctober). Nitrification and denitrification activity was measured directly on fresh samples. For the DNA based studies, each sample of 1.0 mL was dispensed into a 1.5 mL sterile tube and centrifuged at 13000 g for 10 min. The supernatant was decanted and the pellet was stored at 20 C before being used.
2.2.
A full-scale WWTP of a coke manufacturing plant in Pohang, Korea employs a single sludge along with the recycling of
Process monitoring and chemical analysis
The collected samples were centrifuged at 3500 rpm for 3 min (MF550, Hanil Sci. Ind., Korea), and the supernatants were analyzed as follows: according to standard methods (APHA,
Fig. 1 e Schematic diagram of a full-scale activated sludge process treating high strength industrial wastewater.
5787
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
1998), chemical oxygen demand (COD), ammonia, phenol and thiocyanate (SCN) were analyzed by the colorimetric method with a spectrophotometer (Genesys TM-5, Spectronic Inc., USA). After distillation, the cyanide (CN) concentration was determined by the pyridine-pyrazolone method. Nitrite and nitrate ions were measured with an ion chromatograph (ICS-1000, Dionex Co., USA). Total oragnic carbon (TOC), inorganic carbon (IC), and TN were measured with a TOC/TN analyzer (TOC-V csu, TNM1, Shimadzu Co., Japan).
2.3.
2.4.
DNA extraction
The pellet was washed with 1 mL of deionized and distilled water (DDW) and centrifuged at 16,000 g for 5 min to ensure a maximal removal of residual medium. The supernatant was carefully removed and the pellet resuspended in 100 mL of DDW. All DNA in the suspension was immediately extracted using an automated nucleic acid extractor (Magtration System 6 GC, PSS, Chiba, Japan). Purified DNA was eluted with 100 mL of TriseHCl buffer (pH 8.0) and stored at 20 C for further analyses.
Microbial activity test 2.5.
To investigate microbial activities of nitrifiers and denitrifiers in each tank at the WWTP, nitrification and denitrification rates were estimated weekly through batch experiments with synthetic medium containing ammonia or nitrate. Batch experiments for nitrification and denitrification activity were carried out in 500 mL Erlenmeyer flasks filled with 100 mL of test solution containing 50 mg-N/L of ammonia and nitrate ion, respectively. Without any pretreatment each flask was inoculated with fresh activated sludge (about 2000 mg/L), which was sampled directly from each full-scale tank, and then agitated on a thermostatic shaker at 200 rpm and 35 C, maintaining the pH at 7.5. The specific nitrification and denitrification rates were calculated applying the equation provided by Kim et al. (2011a).
T-RFLP analysis
T-RFLP was used to analyze the nitrifying bacteria community in the pre-denitrification process reactor based on the known 16S rRNA genes of ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB), as described in the protocol of a previous study (Siripong and Rittmann, 2007). Because of the low concentration of DNA from the nitrifiers, it was amplified through nested PCR, using the universal primers 11f and 1492r (Table 1), followed by the specific amplification of nitrifier genes (Nitrifier-specific reverse primer: Nso1225r, NIT3r, Ntspa685r, Forward primer: Eub338f included phosphoramidite dye 6-FAM (Table 1)) (Siripong and Rittmann, 2007). 2 mL of template DNA was used for the universal amplification step and 1 mL of the universal amplification product as the
Table 1 e Primers and probes used in T-RFLP and qPCR. Target For T-RFLP Bacterial 16S rDNA Bacterial I6S rDNA AOB 16S rDNA Nitrobacter 16S rDNA Nitrospira 16S rDNA For qPCR Bacterial 16S rDNA
AOB 16S rDNA
Nitrospira spp. 16S rDNA
Nitrobacter spp. I6S1DNA
narG gene nirS gene nirK gene nosZ gene
Primer/probe
Sequence (5’-3’)
References
11f 1492r Eub338f Nso1225r NIT3r Ntspa685r
5’-GTTTGATCCTGGCTCAG-3’ 5’-TACCTTGTTACGACTT-3’ 5’-(6-FAM)-ACTCCTACGGGAGGCAGC-3’ 5’-CGCCATTGTATTACGTGTGA-3’ 5’-CCTGTGCTCCATGCTCCG-3’ 5’-CGGGAATTCCGCGCTC-3’
Kane et al., 1993 Lin and Stahl, 1995 Amann et al., 1990 Mobarry et al., 1996 Wagner et al., 1995 Regan et al., 2002
1055f 1392r 16STaq 1115 CTO 189fA/Ba CTO 1891Ca RTlr TMP1 NSR 1113f NSR 1264r NSR 1143Taq Nitro 1198f Nitro 1423r Nitro 1374Taq narG 1960m2f narG 2050m2r nirS If nirS 3r nirK 876 nirK 1040 nosZ 2f nosZ 2r
5’-ATGGCTGTCGTCAGCT-3’ 5’-ACGGGCGGTGTGTAC-3’ 5’-(6-FAM)-CAACGAGCGCAACCC-(TAMRA)-3’ 5’-GGAGRAAAGCAGGGGATCG-3’ 5’-GGAGGAAAGTAGGGGATCG-3’ 5’-CGTCCTCTCAGACCARCTACTG-3’ 5’-(6-FAM)-CAACTAGCTAATCAGRCATCRGCCGCT-(TAMRA)-3’ 5’-CCTGCTTTCAGTTGCTACCG-3’ 5’-GTTTGCAGCGCTTTGTACCG-3’ 5’-(6-FAM)-AGCACTCTGAAAGGACTGCCCAGG-(TAMRA)-3’ 5’-ACCCCTAGCAAATCTCAAAAAACCG-3’ 5’-CTTCACCCCAGTCGCTGACC-3’ 5’-(6-FAM)-AACCCGCAAGGAGGCAGCCGACC-(TAMRA)-3’ 5’-TAYGTSGGGCAGGARAAACTG-3’ 5’-CGTAGAAGAAGCTGGTGCTGTT-3’ 5’-TACCACCCSGARCCGCGCGT-3’ 5’-GCCGCCGTCRTGVAGGAA-3’ 5’-ATYGGCGGVCAYGGCGA-3’ S’-GCCTCGATCAGRTTRTGGTT-S’ 5’-CGCR ACGGC AAS AAGGTSM SSGT-3’ 5’-CAKRTGCAKSGCRTGGCAGAA-3’
Ferris et al., 1996 Ferris et al., 1996 Harms et al., 2003 Hermansson and Lindgren, Hermansson and Lindgren, Hermansson and Lindgren, Hermansson and Lindgren, Dionisi et al., 2002 Dionisi et al., 2002 Harms et al., 2003 Graham et al., 2007 Graham et al., 2007 Graham et al., 2007 Lo`pez-Gutie`rrez et al., 2004 Lo`pez-Gutie`rrez et al., 2004 Braker et al., 1998 Braker et al., 1998 Henry et al., 2004 Henry et al., 2004 Henry et al., 2006 Henry et al., 2006
a A mixture of CTO 189fA/B and CTO 189fC at the weight ratio of 2:1 was used as the forward primer.
2001 2001 2001 2001
5788
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
template for the nitrifier-specific amplification. Finally, the PCR products were purified and 16S rRNA gene amplicons were digested with MspI restriction endonuclease (Siripong and Rittmann, 2007). Digested PCR products were run through an ABI 3130XL Genetic Analyzer (Applied Biosystems, Foster City, USA) at the SolGent Company (Korea). The peak results were analyzed using the Peak Scanner software v 1.0 (https://products.appliedbiosystems.com; Applied Biosystems, Foster City, USA). Details of the PCR conditions, product purification and restriction digestion are provided elsewhere (Siripong and Rittmann, 2007).
2.6.
qPCR analysis
To investigate the changes in the nitrifying and denitrifying bacteria populations occurring during the process performance, all qPCR assays were performed using a 7300 Real Time PCR system (Applied Biosystems, Foster City, USA). To determine the amount of the nitrifying bacteria, four independent qPCR assays were conducted by quantifying total bacterial 16S rDNA, ammonia oxidizing bacterial 16S rDNA, Nitrospira spp. 16S rDNA, and Nitrobacter spp. 16S rDNA (Table 1). Each capillary tube was separately loaded with 2 mL of template DNA (at 14e26 ng/mL), followed by 4.0 pmol of the forward and reverse primers (1 mL), together with 2.0 pmol of the TaqMan probe (0.5 mL) corresponding to each primer and probe set, 12.5 mL of TaqMan Universal PCR Master Mix (No 4304437 Applied Biosystems, New Jersey, USA), and PCR-grade sterile water for a final volume of 25 mL. The amount of total bacterial 16S rDNA was amplified using primer (1055f and 1392r) (Ferris et al., 1996). The TaqMan probe (16STaq1115) was modified by the 1114f primer (Harms et al., 2003). The PCR program was 2 min at 50 C, 10 min at 95 C; 45 cycles of 30 s at 95 C, 60 s at 50 C, and 40 s at 72 C. To determine the amount of AOB 16S rDNA genes, two forward primers (CTO 189A/B and CTO 189C), one reverse (RT1r), and the TaqMan probe (TMP1) were used as described previously by Hermansson and Lindgren (2001). The PCR program for AOB 16S rDNA quantification included 2 min at 50 C, 10 min at 95 C; 40 cycles of 30 s at 95 C, 60 s at 60 C. The Nitrospira spp. 16S rDNA primers (NSR 1113f/NSR 1264r) (Dionisi et al., 2002) and the TaqMan probe (NSR 1143Taq) (Harms et al., 2003) were tested. PCR amplification consisted of 2 min at 50 C, 10 min at 95 C; 50 cycles of 30 s at 95 C, 60 s at 60 C. Lastly, the amount of Nitrobacter spp. from Graham et al. (2007) was amplified using primer (Nitro 1198f/Nitro 1423r) and TaqMan probe (Nitro 1374Taq). The program used for amplification was 2 min at 50 C, 10 min at 95 C; 50 cycles of 20 s at 94 C, 60 s at 58 C, and 40 s at 72 C. Meanwhile, the denitrifying functional genes were quantified with SYBR Premix Ex Tag (Takara, Japan). Amplification reactions were performed in a total volume of 25 mL containing 2 mL of template DNA (at 17.5e22.5 ng/mL), 4.0 pmol of the forward and reverse primers (1 mL), together with 0.5 mL (1X) of the ROX reference dye (50X), 12.5 mL of SYBR Premix, and PCR-grade sterile water. The qPCR program for 16S rDNA amplification using primer 1055f and 1392r was 30 s at 95 C; 30 cycles of 15 s at 95 C, 20 s at 55 C, and 31 s at 72 C. Primers designed by Lo´pez-Gutie´rrez et al. (2004) were used to determine the amount of narG gene. The PCR program for narG gene
quantification included 30 s at 95 C; 35 cycles of 15 s at 95 C, 30 s at 58 C, and 31 s at 72 C. The nirS gene PCR amplification using primers (nirS 1f and nirS 3r) (Braker et al., 1998) consisted of 30 s at 95 C; 30 cycles of 15 s at 95 C, 20 s at 60 C, and 31 s at 72 C. The PCR condition for nirK gene included 30 s at 95 C; 30 cycles of 15 s at 95 C, 30 s at 58 C, and 31 s at 72 C. Lastly, Primers (nosZ 2f and nosZ 2r) designed by Henry et al. (2006) were used to determine the amount of nosZ gene. The program used for amplification was 30 s at 95 C; 30 cycles of 15 s at 95 C, 30 s at 60 C, and 31 s at 72 C. All experiments were performed in duplicate per sample and all PCR runs included control reactions without the template. The specificity of each PCR assay was confirmed using both melting curve analysis and agarose gel electrophoresis. Gene copy numbers were calculated by comparing threshold cycles obtained in each PCR run with those of known standard DNA concentrations. Standards were prepared using serially diluted plasmid DNA with 103e108 gene copies/mL. Standard curves for the 16S rDNA, AOB, Nitrospira spp., Nitrobacter spp., narG, nirS, nirK, and nosZ assays were generated by plotting the threshold cycle values versus log10 of the gene copy numbers. The amplification efficiency (E ) was estimated using the slope of the standard curve through the following formula: E ¼ (101/slope) 1. The efficiency of PCR amplification for each gene was between 90% and 100%.
2.7.
Cloning and sequencing
Prior to cloning, the amplified unlabeled AOB 16S rRNA genes fragments were purified using the PCR purification kit (SolGent, Korea). Purified PCR products were ligated into pGEM-T Easy cloning vectors (Promega, USA) and transformed into competent Escherichia coli One-Shot Mach 1-T1 (Invitrogen, USA), as described in the manufacturer’s protocol. Transformants were selected by ampicillin resistance and blueewhite screening was performed to identify clones with inserts. Seventy-four white colonies were selected and cultivated. Primers T7 and SP6 were used to perform colony PCR and to verify that the insert size was correct. Following PCR confirmation of insert size, the amplified inserts were run on 2% (wt/vol) agarose gels. The samples containing inserts of the estimated size were used for subsequent sequencing. The 16S rRNA gene inserts were sequenced through an ABI 3130XL Genetic Analyzer (Applied Biosystems, Foster City, USA) at the SolGent Company (Korea). Database homology searches for these sequences were performed using the BLAST program in the National Center for Biotechnology Information (NCBI) database.
3.
Results and discussion
3.1. Functional performance of the full-scale activated sludge process The process performance of the full-scale wastewater treatment system during the study is presented in Fig. 2. Although the pre-denitrification activated sludge process is simple, various microbial reactions occur sequentially under both
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
5789
Fig. 2 e Variation of influent- and effluent-concentrations of (a) COD; (b) TOC; (c) phenol; (d) SCNL; (e) ammonia; (f) TN in the full-scale process during the study period (solid circles: influent, open circles: effluent, solid triangles: removal efficiency).
anoxic and aerobic conditions. During the study, 2025e3150 mg/L of COD was fed into the anoxic tank where 75e80% of it was subsequentially removed. Residual COD flowed into the aerobic tank and then further removed, resulting in a COD removal efficiency of 80e88%. The removal pattern of TOC was similar to that of COD. There was no further degradation of any residual COD or TOC in the nitrification tank. This implies that the remaining residual organic carbon was non-biodegradable. Phenol flowed into the fullscale process in the range of 268e715 mg/L and increased to be 75% higher than normal concentrations of about 400 mg/L in the ninth week (Fig. 2c). Most phenol was first degraded in the anoxic tank, with remaining phenol being almost completely degraded in the aerobic tank. Transformation of organic carbons such as phenol to inorganic carbon took place in the anoxic tank due to denitrification and fermentation reactions by various heterotorphes (Kim et al., 2009). The inorganic carbon was consumed by autotrophic nitrifiers and thiocyanate-degrading bacteria under aerobic conditions. Almost all SCN, which varied from 190 to 297 mg/L in the influent, was removed in the aerobic tank (Fig. 2d). Various autotrophic bacteria in activated sludge are known to degrade SCN under aerobic conditions ðSCN þ 2O2 þ 2H2 O/ 2 NHþ 4 þ SO4 þ CO2 Þ (Lee et al., 2008). Total cyanides in the range of 12.4e17.2 mg/L flowed into the anoxic tank and 1.7e3.8 mg/L of it flowed into the aerobic tank (data not shown). Additional degradation of cyanides did not take place under aerobic conditions. The incomplete removal of total cyanides was due to the existence of ferric cyanide ðFeðCNÞ3 6 Þ, which is known to undergo very slow biodegradation (Kim et al., 2008). The main role of the nitrification tank was to convert the ammonium ion into nitrite and/or nitrate. The average NHþ 4 N
concentration range in the influent was about 76e90 mg-N/L, except for shock loading of ammonia (Fig. 2e). Some amounts of ammonia were newly generated due to cell lysis and degradation of SCN in the anoxic and aerobic tanks, respectively. During stable operation, the final effluent concentration of NHþ 4 N in the nitrification tank remained less than 10 mg-N/L, while the NO 2 N concentration was in the range of 17e20 mgN/L. Detected NO 3 N concentration values were less than 3.0 mg-N/L. Denitrification was promoted by recycling the nitrite/nitrate formed via nitrification back to the anoxic tank. Complete denitrification was consistently achieved in the anoxic tank until 11 weeks regardless of ammonia and phenol shock loading. Excluding this shock loading of ammonia, the average TN in the influent was 190 mg-N/L (Fig. 2f). The effluent TN concentration remained below 40 mg-N/L, corresponding to an average TN removal efficiency of 85% (note that regulations stipulate a TN concentration less than 60 mg-N/L for discharge into surface water in South Korea). Also, the effluent TN concentration of less than 40 mg-N/L indicated that the system was functionally stable throughout the study.
3.2. Influence of operational parameters on nitrogen removal performance of the full-scale activated sludge process Nitrification of the full-scale activated sludge process treating high strength pollutants targeted in our study has been unstable during the past several summers due to an abnormal influx of pollutants such as phenol, SCN, ammonia and cyanide. Therefore, responsive system operations have been proposed to accomplish discharge level regulations under various situations.
5790
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
Fig. 3 shows the effect of operational parameters on the effluent nitrogen concentration of the full-scale activated sludge process. In the first week, although TN concentration remained below the legal discharge level (i.e., 60 mg-N/L), NO 2 N concentration, a by-product of nitrification, was only 7 mg-N/L (Fig. 3e). This indicated inadequate nitrification performance. The nitrification tank was operated at an MLSS concentration of about 1800 mg/L at 15 days of SRT (Fig. 3c), which was lower than the 2500e3000 mg/L MLSS level of general activated sludge processes treating cokes wastewater (Manekar et al., 2011). To increase the nitrifying biomass having a slow growth rate, the SRT was lengthened from 15 days to 20 days by reducing excess sludge removal. This resulted in an increase in MLSS concentration. For 2 weeks, the MLSS concentration in the nitrification tank increased
Fig. 3 e Variation of operational parameters: (a) SRT and internal recycling ratio; (b) MLSS concentration and MLVSS/ MLSS ratio in the nitrification tank; (c) MLSS concentration and MLVSS/MLSS ratio in the anoxic tank; (def) effluentconcentrations of ammonia, nitrite nitrate, TN in the nitrification tank during the study period.
from 1800 mg/L to 2000 mg/L (Fig. 3c). In the final effluent of the nitrification tank, the ammonia concentration was below 20 mg-N/L while the NO 2 N concentration gradually increased to 18 mg-N/L (Fig. 3d and e). TN concentration was consistently maintained under the legal discharge level of 60 mg-N/L. It is believed that this rather long SRT may lead to an increase in the slow growing nitrifying bacteria and support their dominance, resulting in better nitrification performance (Teck et al., 2009). At the beginning of the fourth week, however, an abrupt change in the effluent nitrogen concentration of the nitrification tank occurred (Fig. 3f). The ammonia concentration sharply increased to over 40 mg-N/L and the TN concentration exceeded its legal discharge level of 60 mg-N/L. Initially, it was doubted that a decrease in nitrification activity had occurred as a result of inhibition by cyanide or phenol. However, analysis results of the influent and effluent revealed that the shock loading of ammonia was to blame. In the fourth week, the ammonia concentration in the influent sharply increased to 140 mg-N/L, a level more than twice normal loading (data not shown). Meanwhile, the NO 2 N concentration in the effluent did not decrease, implying that the nitrification performance had not been inhibited. For high volumetric ammonia removal, the process was controlled as nearly infinite SRT without removal of excess sludge, resulting in an accumulation of MLSS in the system for 1 week. It is known that high volumetric loading can be achieved by maintaining a high MLSS (Rittmann and McCarty, 2001). At the end of 4 weeks, the MLSS concentration in the system significantly increased to 2800 mg/L (Fig. 3d). This led to a gradual decrease in both the ammonia and TN concentrations in the final effluent. In addition, as soon as the usual nitrogen concentration in the influent flowed into the system after the fifth week, the effluent nitrogen removal efficiency improved more than previously. The ammonia concentration was controlled at less than 10 mg-N/L and the TN concentration was maintained much lower than the legal level. The NO 2 N concentration of about 17 mg-N/L was produced by nitrification. Complete denitrification was consistently achieved in the anoxic tank. As a result, the increased MLSS concentration in the system by long SRT ensured stable nitrification of the full-scale activated sludge process during ammonia shock loading. These results implied that the selection of an adequate biomass concentration by SRT control can be vital in achieving the desired efficiency of the process. However, full-scale process performance took a sudden turn for the worse in the ninth week (Fig. 3f). Ammonia concentration increased to 87 mg-N/L while NO 2 N concentration decreased to 3.5 mg-N/L in the effluent of the nitrification tank (Fig. 3d and e). The decrease of nitrite concentration reflected incomplete nitrification in the nitrification tank. Since nitrification was significantly inhibited, TN concentration abruptly increased from 30 to 120 mg-N/L. The analysis of variations in the pollutants’ concentrations in the influent indicated that shock loading of phenol was one of the causes for the nitrification failure (Fig. 2c). As the influent concentration of phenol increased from 300e400 mg/L to 715 mg/L, more phenol flowed into the aerobic tank. The inhibitory effect of phenol on nitrification is well known. Previous research has shown that periods of nitrification
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
failure coincide with increased phenol concentration in the wastewater (Figuerola and Erijman, 2010). To prevent deterioration of the nitrification performance from phenol inhibition, the internal recycling ratio was controlled from 2.0 (600 m3/h) to 0.5 (150 m3/h) and long SRT was consistently maintained at 20 days. By reducing the internal recycling ratio, phenol removal performance was enhanced in the anoxic tank, resulting in inflow of less phenol into the nitrification tank. In addition, the decreased recycling ratio increased the retention time of both the biomass and effluent in the nitrification tank. The increased retention time, in turn, provided both adequate nitrification reaction time and contact stabilization, allowing nitrifiers to rebound from any prior inhibition. In addition, it is known that high internal recycling values have a negative effect on the maximum specific growth rate of nitrifiers (Jimenez et al., 2011). Meanwhile, the long SRT prevented the nitrifying biomass from washing out of the system. As a result, as soon as the phenol concentration in the influent decreased to below 500 mg/L after the tenth week (Fig. 2c), the TN removal efficiency improved and a discharge level less than 50 mg-N/L could be achieved (Fig. 3f). The effluent ammonia concentration in the tenth week quickly decreased from 87 mg-N/L to 25 mg-N/L and in the eleventh week was lower than 10 mg-N/L achieving the same functional stability period as between the fifth and eighth week (Fig. 3d). NO 2 N production also recovered from concentration slightly phenol inhibition and NO 3 N increased to 4.1 mg-N/L (Fig. 3e). This meant that nitrification performance had totally recovered. In addition, the increased retention time may be helpful in allowing NOB to generate nitrate ion from nitrite substrate. At the end of the twelfth week, although the effluent ammonia and TN concentrations were stably maintained, NO 2 N production began to decrease from 18 to 9 mg-N/L (Fig. 3e). MLSS concentration climbed to 3400e3600 mg/L in the system in spite of consistent SRT. These results indicated that an increase of organic matter may cause proliferation of heterotrophic microorganisms in the nitrification tank, increasing uptake ammonia for their growth. Contrary to the increase in MLSS concentration, MLVSS/MLSS ratio in the nitrification tank quickly fell to 0.65, resulting in a decrease in nitrification efficiency (Fig. 3b and d). In the anoxic tank, the MLVSS/MLSS ratio sharply decreased to 0.7 with NO 2 N concentration of 6.0 mg/L, resulting in a decrease in denitrification efficiency (data not shown). This decreased VSS concentration may result from an imbalance between feed and biomass, since influent concentrations like organic and nitrogen matter gradually decreased after the ninth week, corresponding to an increase of MLSS concentration by consistently long SRT. Consequently, these results in the twelfth week implied that nitrogen removal performance of the full-scale process in the future could be vulnerable to environmental factors.
3.3.
ammonia, nitrite and nitrate concentrations. As illustrated in Fig. 4, nitrifying bacteria activity was affected by process conditions. Until the third week, increased SRT did not influence the specific nitrification rate, maintaining at about 5.0 mg-N/g-VSS$h. In the fourth week, however, the specific nitrification rate decreased to 3.7 mg-N/g-VSS$h due to a sudden increase in the MLSS concentration. Between the fifth and eighth week, the specific nitrification rate gradually increased from 4.3 to 6.3 mg-N/g-VSS$h, irrespective of a small increase in MLSS concentration. In this period, nitrogen removal efficiency in the full-scale system achieved almost 90%. In the ninth week, nitrification performance of the fullscale process sharply decreased; on the other hand, the batch test revealed only slightly decreased nitrification activity, decreasing from 6.3 to 5.9 mg-N/g-VSS$h. This indicated that there was discordance between the potential rate (batch test) and the actual rate (full-scale performance) regarding nitrification performance. One may conclude that, although nitrifying bacteria possess good potential activity, difficult to identify environmental factors influence their actual performance in the full-scale process, leading to decreased nitrification. During the study, the specific denitrification rate pattern was similar to variations in nitrification activity. The specific denitrification rate remained in the range of 2.8e4.8 mg-N/gVSS h. Like the variation of the nitrification rate, the specific denitrification rate decreased in the fourth week, due to a considerable increase in the MLSS concentration, then gradually increased to 4.8 mg-N/g-VSS h. Even in the ninth week when phenol shock loading incidentally occurred, the specific denitrification rate was not affected and remained consistent. Contrary to the nitrification performance, denitrification in the full-scale process had a stable rate - similar to its batch tests. This implied that denitrification may be less affected than nitrification by environmental factors. Meanwhile, the MLVSS/MLSS ratio in both the anoxic and nitrification tanks gradually decreased. As shown in Fig. 4, however, both the nitrification and denitrification rates in the batch tests were consistent, regardless of the nitrogen removal performance of the full-scale process. As a result, the decrease of MLVSS did not lead to any loss of microorganism activity.
Microbial activity
Batch experiments to observe the variations of nitrification and denitrification activities in the activated sludge process were carried out. The specific nitrification and denitrification rate in each batch test was analyzed through variations of
5791
Fig. 4 e Variation of denitrification and nitrification activities by batch test during the study period.
5792
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
This may result from a consistent percentage of active bacterial populations among total microorganisms.
3.4.
T-RFLP analysis of AOB and NOB populations
We identified the nitrifying bacterial communities present in an activated sludge process using T-RFLP designed for the identification of AOB and NOB with terminal fragment (TF) lengths (Regan et al., 2002). Fig. 5 shows electropherograms of AOB, Nitrobacter-specific NOB, and Nitrospira-specific NOB present in the nitrification tank of the full-scale process, respectively. As shown in Fig. 5a, AOB-targeted T-RFLP allowed us to differentiate between AOB groups. All samples showed a peak at 164 bp, a signature of Nitrosomonas europaea/ eutropha and Nitrosomonas marina lineage (Table 2). Because the influents originate from industrial wastewater, marine AOB species need not be considered. Besides the major peak at 164 bp, we detected another peak at 273 bp, representing the potential presence of N. europaea/eutropha, Nitrosomonas oligotropha, Nitrosomonas cryotolerans, or Nitrosomonas communis lineage (Table 2). To better understand the AOB community present in the process, AOB 16S rRNA gene based cloning and sequencing was performed using the AOB-target primer (Nso1225r and Eub338f) without fluorescent dye (Table 1). Sixty-eight of total 74 AOB clones from the reactor were closely associated with N. europaea in the N. europaea/eutropha lineage and Nitrosomonas nitrosa in the N. communis lineage, but the AOB clone related to the Nitrosospira lineage was not detected. As a result, based on the 16S rRNA gene sequences, microorganisms corresponding to the peaks at 164 bp and 273 bp could be
identified as N. europaea and N. nitrosa, respectively (Fig. 5a). Thus, the high peak at 164 bp implies the dominance of the N. europaea within AOB in this wastewater treatment system, irrespective of variations in the full-scale process performance; N. nitrosa had a minor presence. N. europaea has been widely observed in WWTPs (Lydmark et al., 2007; Siripong and Rittmann, 2007), while N. nitrosa has previously been detected on occasion in activated sludge treating industrial wastewater (Layton et al., 2005). Despite variations in environmental conditions such as MLSS concentration, internal recycling ratio and influent characteristics for the AOB, the total selective pressure in the full-scale process has been insufficient to induce a population shift (Hallin et al., 2005; Kim et al., 2011b). This finding also indicated that low diversity of AOB populations may be conducive to nitrification failure. Based on Nitrobacter-specific T-RFLP, Fig. 5b shows a prominent peak at 137 bp, characteristic of the Nitrobacter species. Meanwhile, in the fifth week, the peak at 214 bp was dominant along with the 137 bp peak, but disappeared after the ninth week. This Nitrobacter sp. corresponding to 214 bp peak may be affected by phenol. We also found TF sizes at 92, 162, 245 and 273 in the samples. These unexpected peaks could be the result of incomplete digestion, uncharacterized Nitrobacter species or imperfectly matched primer (Siripong and Rittmann, 2007). The results of Nitrospira-specific T-RFLP showed four dominant peaks at 135, 192, 272 and 334 bp (Fig. 5c). The peak at 135, 192 and 272 corresponds to several Nitrospira clones in the database. The 334 TF belongs to one of the Nitrospira moscoviensis strains (Siripong and Rittmann, 2007). The Nitrospira species corresponding to the peak at 272 bp was a consistently dominant population, while
Fig. 5 e T-RFLP profiles of (a) AOB, (b) Nitrobacter, (c) Nitrospira in the nitrification tank during the study period.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
Table 2 e Expected TF sizes and their corresponding AOB and NOB groups based on T-RFLP of 16S rDNA gene (Siripong and Rittmann, 2007). TF size (bp) 164e166, 276 276 276 166 276 105e107 141, 196 133, 194, 265e267, 277,333
Nitrifying bacteria group Nitrosomonas europaea/eutropha lineage Nitrosomonas oligotropha lineage Nitrosomonas cryotolerans lineage Nitrosomonas marina lineage Nitrosomonas communis lineage Nitrosospira lineage Nitrobacter species Nitrospira species
N. moscoviensis sp. corresponding to the peak at 334 bp was found to be predominant when nitrification performance of the full-scale process began to stabilize in the fifth week. Despite production of small amounts of nitrate by NOB in the nitrification tank, harsh environmental conditions for NOB may stimulate diversity of the NOB species.
3.5. qPCR analysis of nitrifying and denitrifying bacteria populations Fig. 6a shows the changes in the 16S rRNA gene copies for the total bacteria, AOB, Nitrobacter, and Nitrospira, quantified using qPCR assays in the nitrification tank of the full-scale process. In all samples the total bacterial population in the nitrification tank ranged from 4.8 1012 to 3.8 1013 copies/L. These values are the same order of magnitude as those obtained from activated sludge samples of WWTPs (Limpiyakorn et al., 2005). As the MLSS concentration increased as a result of SRT control, the total bacterial population increased to 3.8 1013 copies/L in the seventh week, indicating an eightfold increase compared to that in the first week. The concentration of AOB identified using the AOB 16S rDNA assay gradually increased and the variation was not large, compared to that of the NOB population. In the tenth week, an approximately 9-fold increase was observed in the
5793
number of AOB 16S copies/L over the first week. This result indicated that newly controlled SRT and the internal recycling ratio may ensure continued AOB population. In addition, such a gradual increase in the AOB population may affect the increase in nitrification activity observed in the batch test. In the ninth week when the nitrification performance of the fullscale process was severely inhibited, there was no observed decrease in the AOB number. However, any change in the AOB number generally coincided with the variation of NO 2 N concentration produced in the nitrification tank. Meanwhile, the percentages of the AOB within the total bacteria varied from 1.07 to 3.29% in the nitrification tank. This result is similar to the values for activated sludge samples obtained from systems treating industrial wastewater (0.01e9.3%) (Layton et al., 2005). But we could not identify any correlation with the nitrification activity. We observed coexisting Nitrospira and Nitrobacter genera for NOB. The Nitrospira and Nitrobacter populations in the initial operating condition were similar, at 2.0 109 copies/L and 2.8 109 copies/L, respectively. However, a shift to Nitrobacter sp. in the NOB community was observed throughout the study. The 16S rRNA gene concentration of the Nitrobacter increased to a range of 2.8 109 to 4.8 1010 copies/ L, and the percentages of the Nitrobacter population within the total bacteria also sharply increased from 0.03 to 0.16% in the nitrification tank, as the MLSS concentration increased and more nitrite was produced. Finally, the Nitrobacter populations in the nitrifying system were consistently higher than the Nitrospira populations throughout the study. On the other hand, the number of Nitrospira gradually decreased from 2.0 109 to 1.1 109 copies/L, until the third week. When high concentration of ammonia in the influent flowed into the activated sludge process, a 5-fold increase was observed in the number of Nitrospira 16S copies/L along with an increase in the number of Nitrobacter sp. However, in the ninth week when high concentrations of phenol flowed into the system, a sharp decrease of Nitrospira population was observed. The percentage of the Nitrospira population among all bacteria shrank to 0.01%. Previous research has reported that Nitrospira
Fig. 6 e Changes in copies per liter of (a) the total bacteria, AOB, Nitrobacter, and Nitrospira, (b) the total bacteria, narG, nirS, nirK, nosZ in the full-scale activated sludge process during the study period.
5794
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
were far more sensitive to toxic compounds than Nitrobacter (Blackburne et al., 2007; Kim et al., 2011a). Meanwhile, despite coexisting Nitrospira and Nitrobacter genera for NOB (Fig. 6a) and generation of NO 3 N as the final product through nitrification activity in the batch test (data not shown), low concentrations of NO 3 N were produced in the nitrification tank of the full-scale process (Fig. 2). This may reflect the short retention time necessary for NOB to react with nitrite ions in the full-scale process. Meanwhile, the abundance of narG, nirS, nirK and nosZ genes of denitrifying bacteria was investigated during the study in the activated sludge process. The narG, nirS, nirK and nosZ target molecules were less abundant than the 16S rDNA gene copies for the total bacteria: total bacteria ranged from 9.9 1012 to 3.9 1013 copies/L; narG ranged from 2.0 109 to 9.2 109 copies/L; nirS ranged from 1.3 1012 to 1.7 1013 copies/L; nirK ranged from 1.1 1010 to 1.0 1011 copies/L; nosZ ranged from 2.1 1011 to 2.4 1012 copies/L (Fig. 6b). In the activated sludge process treating industrial wastewater, gene copy numbers per liter of the nirS gene exceeded those of the narG and nosZ genes for all sampling points. This trend implies that there is a greater abundance of genes for the nitrite reducing genes than for the nitrate and nitrous oxide reducing genes. Meanwhile, for nirS, the copy numbers of genes detected were much higher than those for nirK at all sampling points. It is known that more taxonomically diverse nirK denitrifiers are more sensitive to environmental changes than the nirS denitrifiers; however, the latter are more abundant (Yoshie et al., 2004). To evaluate the ratio of denitrifiers relative to total bacteria, the percentages of denitrification genes in proportion to 16S rDNA were calculated, resulting in proportions of approximately 0.02%, 26.9%, 0.30%, and 4.77% for narG, nirS, nirK, and nosZ genes, respectively. The maximum amount of nirS relative to 16S rDNA was 43%, confirming the high proportion of denitrifiers to total bacteria in this activated sludge process. On the other hand, the smallest amount of the narG gene, the nitrate reducing functional gene, in the anoxic tank for all samplings may result from low concentrations of NO 3 N in the nitrified effluent. Lastly, in the twelfth week, both nitrifying and denitrifying populations decreased along with a decrease in the MLVSS/ MLSS ratio, resulting in a slight decrease in both nitrification and denitrification efficiencies of the full-scale process. However, in the batch tests both the nitrification and denitrification rates remained steady. This may result from little loss of active bacterial populations among total microorganisms. Also, in actual environment conditions, there may be more factors inhibiting the activity of microorganisms in the full-scale process than in the batch test. Consequently, it was very difficult to identify any relationships between the bacterial populations, their activity and the process performance in the fullscale process. However, to prevent failure of the process performance, it is important to monitor any sudden decrease in the populations of important bacteria such as nitrifiers.
4.
Conclusions
In a full-scale activated sludge process, SRT and internal recycling ratio controls were selected as the main operating
parameters to improve TN removal efficiency. Increased biomass concentration via SRT control enhanced TN removal. In addition, decreasing the internal recycling ratio restored nitrification activity which had been inhibited by phenol shock loading. These results indicate that application of specific operational strategies can change the physiological state of the activated sludge process’s bacterial populations (Hallin et al., 2005). Therefore, proper operational strategies should be tailored to accommodate the possibility of erratic changes in the composition of the influent via consistent monitoring of its components to achieve and maintain a stable process.
Acknowledgments The authors thank David Nielsen for assistance during this work. This research was supported by WCU (World Class University) program through the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology (R31-30005) and the Advanced Biomass R&D Center (ABC) of Korea Grant funded by the Ministry of Education, Science and Technology (ABC-2010-0029800). Also, this work was financially supported by the second phase of the Brain Korea 21 Program in 2011 and Korea Ministry of Environment (MOE) as ’Human resource development Project for Energy from Waste & Recycling’.
references
Amann, R.I., Binder, R.J., Olson, S., Chisholm, S.W., Devereux, R., Stahl, D.A., 1990. Combination of 16S ribosomal RNA targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Applied and Environmental Microbiology 56, 1919e1925. APHA, 1998. Standard Methods for the Examination of Water and Wastewater. APHA, AWWA, WPCF, twentieth ed. American Public Health Association, Washington, DC, USA. Blackburne, R., Vadivelu, V.M., Yuan, Z.G., Keller, J., 2007. Kinetic characterisation of an enriched Nitrospira culture with comparison to Nitrobacter. Water Research 41, 3033e3042. Braker, G., Fesefeldt, A., Witzel, K.P., 1998. Development of PCR primer systems for amplification of nitrite reductase genes (nirK and nirS ) to detect denitrifying bacteria in environmental samples. Applied and Environmental Microbiology 64, 3769e3775. Dionisi, H.M., Layton, A.C., Harms, G., Gregory, I.R., Robinson, K.G. , Sayler, G.S., 2002. Quantification of Nitrosomonas oligotropha-like ammonia-oxidizing bacteria and Nitrospira spp. from full-scale wastewater treatment plants by competitive PCR. Applied and Environmental Microbiology 68, 245e253. Eighmy, T.T., Bishop, P.L., 1989. Distribution and role of bacterial nitrifying populations in nitrogen removal in aquatic treatment systems. Water Research 23, 947e955. Ferris, M.J., Muyzer, G., Ward, D.M., 1996. Denaturing gradient gel electrophoresis profiles of 16S rRNA-defined populations inhabiting a hot spring microbial mat community. Applied and Environmental Microbiology 62, 340e346. Figuerola, E.L., Erijman, L., 2010. Diversity of nitrifying bacteria in a full-scale petroleum refinery wastewater treatment plant
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 8 5 e5 7 9 5
experiencing unstable nitrification. Journal of Hazardous Matererials 181, 281e288. Graham, D.W., Knapp, C.W., Van Vleck, E.S., Bloor, K., Lane, T., Graham, C.E., 2007. Experimental demonstration of chaotic instability in biological nitrification. ISME Journal 1, 385e394. Hallin, S., Lydmark, P., Kokalj, S., Hermansson, M., So¨rensson, F., Jarvis, A., Lindgren, P.E., 2005. Community survey of ammonia-oxidizing bacteria in full-scale activated sludge processes with different solids retention time. Journal of Applied Microbiology 99, 629e640. Harms, G., Layton, A.C., Dionisi, H.M., Gregory, I.R., Garrett, V.M., Hawkins, S.A., Robinson, K.G., Sayler, G.S., 2003. Real-time PCR quantification of nitrifying bacteria in a municipal wastewater treatment plant. Environmental Science and Technology 37, 343e351. Henry, S., Baudouin, E., Lo´pez-Gutie´rrez, J.C., Martin-Laurent, F., Brauman, A., Philippot, L., 2004. Quantification of denitrifying bacteria in soils by nirK gene targeted real-time PCR. Journal of Microbiological Methods 59, 327e335. Henry, S., Bru, D., Stres, B., Hallet, S., Philippot, L., 2006. Quantitative detection of the nosZ gene, encoding nitrous oxide reductase, and comparison of the abundances of 16S rRNA, narG, nirK, and nosZ genes in soils. Applied and Environmental Microbiology 72, 5181e5189. Hermansson, A., Lindgren, P.E., 2001. Quantification of ammoniaoxidizing bacteria in arable soil by real time PCR. Applied and Environmental Microbiology 67, 972e976. Jimenez, J., Melcer, H., Parker, D., Bratby, J., 2011. The effect of degree of recycle on the nitrifier growth rate. Water Environment Research 83, 26e35. Juliastuti, S.R., Baeyens, J., Creemers, C., 2003. Inhibition of nitrification by heavy metals and organic compounds: the ISO 9509 test. Environmental Engineering Science 20, 70e90. Kane, M.D., Poulsen, L.K., Stahl, D.A., 1993. Monitoring the enrichment and isolation of sulfate-reducing bacteria by using oligonucleotide hybridization probes designed from environmentally derived 16s ribosomal RNA sequences. Applied and Environmental Microbiology 59, 682e686. Kim, Y.M., Park, D., Lee, D.S., Park, J.M., 2007. Instability of biological nitrogen removal in a cokes wastewater treatment facility during summer. Journal of Hazardous Matererials 141, 27e32. Kim, Y.M., Park, D., Lee, D.S., Park, J.M., 2008. Inhibitory effects of toxic compounds on nitrification process for cokes wastewater treatment. Journal of Hazardous Matererials 152, 915e921. Kim, Y.M., Park, D., Lee, D.S., Jung, K.A., Park, J.M., 2009. Sudden failure of biological nitrogen and carbon removal in the fullscale pre-denitrification process treating cokes wastewater. Bioresource Technology 100, 4340e4347. Kim, Y.M., Lee, D.S., Park, C., Park, D., Park, J.M., 2011a. Effects of free cyanide on microbial communities and biological carbon and nitrogen removal performance in the industrial activated sludge process. Water Research 45, 1267e1279. Kim, Y.M., Cho, H.U., Lee, D.S., Park, C., Park, D., Park, J.M., 2011b. Response of nitrifying bacterial communities to the increased thiocyanate concentration in pre-denitrification process. Bioresource Technology 102, 913e922.
5795
Layton, A.C., Dionisi, H., Kuo, H.W., Robinson, K.G., Garrett, V.M., Meyers, A., Sayler, G.S., 2005. Emergence of competitive dominant ammonia-oxidizing bacterial populations in a fullscale industrial wastewater treatment plant. Applied and Environmental Microbiology 71, 1105e1108. Lee, C., Kim, J., Do, H., Hwang, S., 2008. Monitoring thiocyanatedegrading microbial community in relation to changes in process performance in mixed culture systems near washout. Water Research 42, 1254e1262. Limpiyakorn, T., Shiohara, Y., Kurisu, F., Yagi, O., 2005. Communities of ammonia-oxidizing bacteria in activated sludge of various sewage treatment plants in Tokyo. FEMS Microbiology Ecology 54, 205e217. Lin, C., Stahl, D.A., 1995. Comparative analyses reveal a highly conserved endoglucanase in the cellulolytic genus Fibrobacter. Journal of Bacteriology 177, 2543e2549. Lo´pez-Gutie´rrez, J.C., Henry, S., Hallet, S., Martin-Laurent, F., Catrou, G., Philippot, L., 2004. Quantification of a novel group of nitrate-reducing bacteria in the environment by real-time PCR. Journal of Microbiological Methods 57, 399e407. Lydmark, P., Almstrand, R., Samuelsson, K., Mattsson, A., So¨rensson, F., Lindgren, P.E., Hermansson, M., 2007. Effects of environmental condition on the nitrifying population dynamics in a pilot wastewater treatment plant. Environmental Microbiology 9, 2220e2233. Manekar, P., Biswas, R., Karthik, M., Nandy, T., 2011. Novel two stage bio-oxidation and chlorination process for high strength hazardous coal carbonization effluent. Journal of Hazardous Matererials. doi:10.1016/j.jhazmat.2011.02.006. Mertoglu, B., Semerci, N., Guler, N., Calli, B., Cecen, B., Saatci, A.M. , 2008. Monitoring of population shift in an enrich nitrifying system under gradually increased cadmium loading. Journal of Hazardous Matererials 160, 495e501. Mobarry, B.K., Wagner, M., Urbain, V., Rittmann, B.E., Stahl, D.A., 1996. Phylogenetic probes for analyzing abundance and spatial organization of nitrifying bacteria. Applied and Environmental Microbiology 62, 2156e2162. Regan, J.M., Harrington, G.W., Noguera, D.R., 2002. Ammonia- and nitrite-oxidizing bacterial communities in a pilot-scale chloraminated drinking water distribution system. Applied and Environmental Microbiology 68, 73e81. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology: Principles and Applicatons. McGraw-Hill Science. Siripong, S., Rittmann, B.E., 2007. Diversity study of nitrifying bacteria in full-scale municipal wastewater treatment plants. Water Research 41 (5), 1110e1120. Teck, H.C., Loong, K.S., Sun, D.D., Leckie, J.O., 2009. Influence of a prolonged solid retention time environment on nitrification/ denitrification and sludge production in a submerged memebrane bioreactor. Desalination 245, 28e43. Wagner, M., Rath, G., Amann, R., Koops, H.P., Schleifer, K.H., 1995. In situ identification of ammonia-oxidizing bacteria. Systematic and Applied Microbiology 18, 251e264. Yoshie, S., Noda, N., Tsuneda, S., Hirata, A., Inamori, Y., 2004. Salinity decreases nitrite reductase gene diversity in denitrifying bacteria of wastewater treatment systems. Applied and Environmental Microbiology 70, 3152e3157.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Reduced microbial attachment by D-amino acid-inhibited AI-2 and EPS production Huijuan Xu a, Yu Liu a,b,* a
Division of Environmental and Water Resources Engineering, School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore b Advanced Environmental Biotechnology Centre, Nanyang Environment & Water Research Institute, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore
article info
abstract
Article history:
This study investigated the effect of D-tyrosine on microbial attachment to hydrophilic D-tyrosine
did not
Received 21 March 2011
glass and hydrophobic polypropylene surfaces. Results showed that
Received in revised form
influence microbial growth, ATP and substrate utilization, but significantly inhibited the
22 August 2011
synthesis of autoinducer-2 (AI-2), eDNA and extracellular polysaccharides and proteins,
Accepted 29 August 2011
and subsequently reduced microbial attachment onto glass and polypropylene surfaces
Available online 3 September 2011
was observed. It was shown that D-amino acid would be a non-toxic agent for control of microbial attachment.
Keywords:
ª 2011 Elsevier Ltd. All rights reserved.
Microbial attachment D-amino
acid
Autoinducer-2 ATP eDNA EPS
1.
Introduction
It had been considered that D-amino acids are excluded from living systems except for D-amino acids in the cell wall of microorganisms. D-amino acids have been discovered in many physiological processes. The best described of D-amino acids may be their involvement in the formation of the peptidoglycan. Both the thick cell wall of Gram-positive bacteria and much thinner cell wall of Gram-negative bacteria consist of peptidoglycan which contain D-amino acids. Besides components of bacterial cell wall, D-amino acid have been known to regulate bacterial germination and to be incorporated into peptides (Wood et al., 2011). The bacteria begin to synthesize
those D-amino acids in stationary phase, which may regulate the chemistry of the cell wall through slow production of peptidoglycan that is crucial for cell wall (Lam et al., 2009). It has been reported that many bacteria would produce various D-amino acids just before biofilm disassembly and the release of rapid diffused small molecule D-amino acid could be a signal to coordinate the whole population action to different environment (Kolodkin-Gal et al., 2010). During biofilm formation, microorganisms need first attach onto a solid surface, followed by secretion of extracellular polymeric substances (EPS) (Flemming and Wingender, 2010), whereas other factors have also been reported to be essentially involved in biofilm development. For example, both Gram-
* Corresponding author. Division of Environmental and Water Resources Engineering, School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore. Tel.: þ65 67 905 254; fax: þ65 67 910 676. E-mail address: [email protected] (Y. Liu). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.061
5797
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
negative and Gram-positive bacteria could communicate through small signaling molecules to coordinate population behavior (Bassler, 1999). Autoinducer-2 (AI-2) is a speciesnonspecific signal molecule found in both Gram-negative and Gram-positive bacteria, and can influence a mixed-species biofilm formation between Porphyromonas gingivalis and Streptococcus gordonii, and it was found that production of AI-2 by either species was sufficient for inter-species communication and biofilm formation (McNab et al., 2003). ATP provides a pool of energy for most energy-consuming microbial activities, such as signaling molecules secretion, EPS synthesis, motility and flagella etc. In addition, the role of EPS in the formation of biofilms has been well documented (Flemming and Wingender, 2010). Although the role of D-amino acids in pure culture biofilm dispersal has been reported (Kolodkin-Gal et al., 2010), little is currently known about the effects of exogenous D-amino acids on the formation of mixed-culture biofilm and synthesis of cellular ATP, AI-2, eDNA and EPS, which are all essential for biofilm development on a solid surface. Furthermore, the present study focused more on the interaction between D-amino acid and AI-2-mdeiated cellular communication. For this purpose, a typical D-amino acid, D-tyrosine was used due to its potent activity (Kolodkin-Gal et al., 2010). Therefore, this study aimed to investigate how D-tyrosine could affect attachment of mixed-culture microorganisms onto hydrophobic PP and hydrophilic glass surfaces through determining changes in surface charge, ATP, AI-2, eDNA and EPS. It is expected that this study can offer an alternative approach for biological control of microbial attachment on various solid surfaces including membrane.
2.
Materials and methods
2.1.
Carriers for microbial attachment
Glass slides with the dimension of 24 50 mm (CEP, SPD Scientific, Singapore) and 24 50 mm polypropylene (PP) coupons (Kinary, Singapore) were used as biocarriers in microbial attachment experiments. The PP coupons were cleaned with detergent and rinsed thoroughly with distilled water, whereas the glass slides were cleaned by being soaked in 10% nitric acid for 24 h, and were then thoroughly rinsed with distilled water and dried. The hydrophobicity of the carrier surface was characterized by contact angle that was measured using a contact angle goniometer (dataphysics OCA 20, Filderstadt, Germany). Eight measurements were made on triplicate samples. The average water drop contact angle for clean glass slide was 16.9 0.5 and 99.3 2.2 for PP coupons, i.e. the PP coupons are highly hydrophobic, and hydrophilic for glass slides.
2.2.
Microbial attachment assay
Activated sludge microorganisms were taken from a local wastewater treatment plant and acclimated with a synthetic substrate for one month. The synthetic substrate consisted of 690 mg l1 of sodium acetate and 240 mg l1 ethanol as carbon source, 200 mg l1 NH4Cl, 60 mg l1 K2HPO4, 15 mg l1
CaCl2·2H2O, 12.5 mg l1 MgSO4·7H2O and 20 mg l1 FeSO4·7H2O (Liu et al., 2003). Experiments were designed to investigate the effect of D-tyrosine on microbial growth and attachment potentials. Thus, microorganisms with and without exposure to D-tyrosine for different times were used in 1-h static microbial attachment experiments conducted under the same conditions, as detailed below: (i) two series of batch experiments were conducted: one served as control free of D-tyrosine, while the other was added with 6 mg l1 of D-tyrosine (SigmaeAldrich, St. Louis, MO, USA); (ii) suspended biomass cultivated with and without exposure to D-tyrosine was collected at different exposure times of 1e4 h for 1-h microbial attachment assay and determination of surface charge, cellular ATP, AI-2, eDNA and EPS. The static microbial attachment was conducted in Petri dishes mounted with one PP coupon and one glass slide on the bottom. Suspended microorganisms harvested from the batch reactors at different exposure times were resuspended in 30 ml of 10 mM phosphate buffered saline (PBS) solution with 100 mg dry biomass l1 and were made contact with carriers for 1 h in Petri dishes. After attachment, carriers were gently rinsed three times with distilled water to remove loosely attached microorganisms. Fixed biomass was quantified in terms of TOC by a TOC analyzer (ASI-V, TOCVcsh, Shimadzu, Japan).
2.3.
Surface charge
The colloid titration method was used to determine surface charge of suspended microorganisms with and without exposure to D-tyrosine (Wilen et al., 2003). Polybrene (SigmaeAldrich) was used as positive colloidal reagent and polyvinyl sulfate potassium salt (PVSK) (SigmaeAldrich) as negative reagent. For titrating negatively charged suspended microorganisms, 5 ml of 0.001 N polybrene was added to the sample. The excess polybrene was back titrated with 0.0005 N PVSK using 100 ml of 0.1% toluidine blue (SigmaeAldrich) as the endpoint indicator. Titration was terminated when the color changed from blue to pink, indicating that electrical neutrality was reached. Equal volumes of polybrene in distilled water were used as blanks. The surface charge expressed as mill equivalents per gram of dry biomass can be determined from the equation given below. Charge meq g1 SS
¼
1000ðA BÞN XV
(1)
Where A is the volume of PVSK added to the sample (ml), B is the volume of PVSK added to the blank (ml), N is the normality of PVSK solution used (0.0005 N), V is the volume of the sample (ml), X is the biomass concentration of the sample (g L1).
2.4.
Determination of cellular ATP
The cellular ATP were extracted from freshly collected biosamples according to the trichloroacetic acid (TCA) method (Chen and Leung, 2000) with some modifications. Five milliliter of the bacterial suspension was added into 5 ml of 5% TCA solution, and was homogenized with an ultrasonic homogenizer (Sonics & Materials, Newton, CT, USA) for 3 min. Aliquots of 0.5 ml homogenized suspension were diluted by ten times with Tris-Acetate-EDTA (TAE) buffer (Bio-Rad,
5798
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
Singapore) to adjust the pH of about 7.75. The mixture was filtered through 0.2 mm syringe filters and the collected filtrate was stored at 20 C for further use. ATP concentration was determined according to firefly luciferin-luciferase bioluminescence method with the FLAA Adenosine 50 -triphosphate (ATP) Bioluminescent Assay Kit (SigmaeAldrich, St. Louis, MO, USA) as recommended in the instruction. The intensity of luminescence was measured by a TD-20/20 Luminometer (Turner Designs, Sunnyvale, CA, USA).
2.5.
Autoinducer-2 measurement
For determination of AI-2, 10 ml suspended microorganisms were collected and resuspended in fresh autoinducer bioassay (AB) medium (Surette and Bassler, 1998). The resuspended sample was well mixed and filtered through 0.2 mm syringe filter. The filtrate was collected and stored at 20 C. The cellfree culture supernatant was thawed before determination of AI-2 concentration. The amount of AI-2 was measured by Vibrio harveyi BB170 (ATCC BAA 1117) bioluminescence reporter assay (Rickard et al., 2008). The reporter strain V. harveyi BB170 (ATCC, Manassas, VA, USA) was cultured in fresh AB medium for 13e16 h with shaking at 30 C and then diluted 1:5000 with fresh AB medium. One hundred and eighty microliter of the diluted cells was added to the well of 96-well plate containing 20 ml cell-free supernatant to be tested for AI-2 activity. The 96well plate was incubated in a rotary shaker at 30 C. The intensity of luminescence was measured hourly using a microplate reader. The fold induction was converted to the molar concentration of AI-2 by comparing the fold induction and 4,5-dihydroxy-2,3-pentanedione (DPD) concentration, as DPD (Omm Scientific, Dallas, USA) were used as the calibration standard. Each filtrated sample was assayed six times in parallel and the mean values reported.
2.6.
Response of AI-2 reporter strain to D-tyrosine
In order to further investigate the effect of D-tyrosine on AI-2 repression, an AI-2 bioluminescence assay in presence and absence of D-tyrosine was conducted. In this assay, five thousand times diluted reporter strain V. harveyi BB170 as described before was grown in fresh AB medium supplemented with 0.1e0.7 mM of DPD in the wells of a 96-well plate. Two series of experiments were conducted: for control, the wells were free of D-tyrosine; other wells were added with 6 mg l1 D-tyrosine. The 96-well plate was shaken in a rotary shaker at 30 C. The light intensity was assayed over time using a microplate reader until get the maximum fold induction of bioluminescence.
2.7.
Extraction and quantification of eDNA
Extracellular DNA was extracted from suspended microorganisms sample according to (Steinberger and Holden, 2005) with modification. Five milliliter of bacterial suspension was collected and resuspended in 0.9% NaCl solution. The resuspended sample was well mixed with a homogenizer (Sonics & Materials, CT, USA). Treated cell solution was filtered through 0.2 mm syring filter and the collected filtrate was stored at 20 C for determining eDNA concentration. The concentration of
DNA was measured by using PicoGreen dsDNA Quantification Kit (Molecular Probes, Invitrogen, Eugene, OR, USA) following the protocol provided by the kit. The calf thymus DNA was used as the standard. The fluorescence intensity was recorded by a microplate reader (BioTek, sygnergy 2, VT, USA).
2.8. Determination of extracellular polysaccharides and proteins Extracellular polysaccharides (PS) and proteins (PN) were extracted from biosample by modified cold aqueous technique method (Jia et al., 1996). Ten milliliter of suspended microorganisms was washed twice with distilled water and centrifuged at 3500 rpm for 10 min. The settled biomass was recovered and resuspended in 10 ml of 8.5% NaCl and 0.22% formaldehyde solution. The mixture was homogenized for 2 min using an ultrasonic homogenizer (Sonics & Materials, CT, USA) in an ice-water bath, and then was centrifuged at 10,000 rpm for 30 min to remove solid residues. The supernatant was harvested for PS, PN measurement and high performance size exclusion chromatography analysis. PS was determined by the phenol-sulfuric acid method (Dubois et al., 1956), whereas PN was analyzed by the modified Lowry method (Lowry et al., 1951). Glucose and bovine serum albumin (SigmaeAldrich) were used as the standards for PS and PN, respectively. High performance size exclusion chromatography (HPSEC) analysis was carried out with Series 200 HPLC system (Perkin Elmer, Waltham, MA, USA) equipped with a Series 200LC quaternary pump, Series 200 autosampler, a Perkin Elmer 600 interface and a UV/Vis detector (785A). A 300 7.8 mm size exclusion chromatography column BioSep SEC S2000 (Phenomenex, Torrance, CA) was used. The mobile phase consisted of 9.0 mM NaCl and 0.9 mM Na2HPO4 at pH 7.0 (Comte et al., 2007). Extracellular polymeric substances were extracted from suspended microorganisms as mentioned above and all samples were filtered through 0.20 mm filters prior to injection. All measurements were conducted at 25 C, mobile phase flow 1.0 ml min1, the sample injection 100 ml. The detection was carried out with a UV detector at 280 nm.
2.9.
Staining and visualization
In order to visualize microbial attachment, the adherent bacteria on glass slides and PP coupons surfaces were stained with LIVE/DEAD BacLight Bacterial Viability kits (Molecular Probes, Eugene, OR, USA), which consisted of two nucleic acid dyes staining on both live and dead cells: SYTO 9 and propidium iodide (PI). SYTO 9 is a green-fluorescent dye which stains both live and dead bacteria with intact and damaged cell membranes while the red-fluorescing PI only stains dead bacteria with damaged cell membranes. The excitation/emission maxima for these dyes are about 480/500 nm for SYTO 9 stain and 490/635 nm for PI. With an appropriate mixture of both dyes, viable bacteria with intact cell membranes are stained green, whereas bacteria with damaged cell membranes fluoresce red. The color assigned to the live and dead cells follows from the color at which the stained cells fluoresce under laser excitation. The sample staining procedure was carried out following the instructions in the manual. First, two
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
hundred microliters of the mixed solution (1000 times diluted SYTO 9 and PI from the stock solution) was added to each attachment sample on glass slides and PP coupons. The stained sample was then incubated in the dark at room temperature for 15 min. After that, the sample was gently rinsed two times with DI water to remove unbound dyes. Finally, the sample was covered with cover slip and viewed using an Olympus Fluoview FV300 confocal laser scanning microscopy (CLSM) (Olympus Optical, Tokyo, Japan) with a 100X objective.
2.10.
Statistical analysis
All tests were performed in triplicate otherwise stated. Results were expressed as mean value absolute deviation. Student ttests were employed for analyzing the significance of results at the level of P < 0.05.
3.
Results
3.1.
Microbial attachment on glass and PP
Fig. 1a shows that attachment of microorganisms exposed to 6 mg l1 D-tyrosine was reduced significantly on glass slides compared to the control free of D-tyrosine (Student’s t-test, P < 0.05). After 2-h culture, attachment of microorganisms without exposure to D-tyrosine was 10.8 mg TOC cm2 on the glass surface, while attachment of microorganisms with exposure to D-tyrosine was 8.5 mg TOC cm2, indicating 22% reduction in microbial attachment caused by D-tyrosine. Similar phenomenon was also observed in microbial attachment on PP surface (Fig. 1b). These results indeed are supported by the microscopic observations (Fig. 2). It should be noted that mixed-culture microorganisms with and without exposure to D-tyrosine were collected at different culture (exposure) times of 1e4 h, and used for 1-h microbial attachment assays, as shown in Fig. 1. According to substrate availability over 4-h culture, 1 h- and 2-h old microorganisms were basically in earlier exponential and post exponential growth phases, while 3 h- and 4 h-old microorganisms already entered into earlier stationary and post-stationary growth phases, respectively. It is reasonable to consider that microorganisms at different growth states would have different attachment abilities as observed in the control assays (Fig. 1). This view is strongly supported by the findings of Fletcher
5799
(1977), showing that the number of attached cells harvested from exponential growth phase was greatest, followed by those from stationary and decay phases. In addition, Fig. 3a shows that addition of D-tyrosine to the culture media had no negative effect on the TOC removal efficiency. It was shown in Fig. 3b that the suspended biomass concentration increased from 450 mg l1 to 630 mg l1 in the cultures supplemented with and without D-tyrosine. These suggest that D-tyrosine is not inhibitory to substrate utilization and microbial growth at the concentration studied. Fig. 4 shows that suspended microorganisms with and without exposure to D-tyrosine both carried negative surface charge, but microorganisms exposed to D-tyrosine carried more negative surface charge compared to that of control free of D-tyrosine.
3.2. Cellular ATP and AI-2 contents of suspended microorganisms Fig. 5a showed the possible effect of D-tyrosine on energy metabolism of suspended microorganisms. It can be seen in Fig. 5a that the cellular ATP content did not change significantly in the presence of D-tyrosine compared to that of control. D-tyrosine thus does not appear to inhibit the ATP synthesis. AI-2 as inter-species signaling molecules coordinates the formation of biofilm by various species (Rickard et al., 2008). To investigate the effect of D-tyrosine on cellular communication, Fig. 5b showed the respective AI-2 content of suspended microorganisms with and without D-tyrosine addition. After 1 h culture, the AI-2 content of suspended microorganisms without exposure to D-tyrosine was about 0.27 nmol mg1 in the control, while it decreased to about 0.19 nmol mg1 for suspended microorganisms with exposure to D-tyrosine, i.e. D-tyrosine could suppress the synthesis or secretion of AI-2. The response of reporter strain V. harveyi BB170 to D-tyrosine was further studied, and results were presented in Fig. 6, showing fold induction of luminescence in presence and absence of D-tyrosine, and obviously luminescence was suppressed significantly when the culture was supplemented with D-tyrosine (Student’s t-test, P < 0.05). For example, for 0.4 mM DPD, the fold induction of luminescence was 23.6 in the control free of D-tyrosine, whereas it decreased to 12.4 in the media with addition of D-tyrosine, indicating 48% reduction as compared to that of control. These imply that Dtyrosine at the concentration studied had an inhibitory effect on AI-2 expression.
Fig. 1 e Attachment of microorganisms with (-) and without (,) treatment by D-tyrosine on glass slides (a); on PP coupons (b). Each point represents the mean of triplicate measurements and error bar is absolute deviation from the mean.
5800
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
Fig. 2 e CLSM images of attachment of microorganisms without D-tyrosine treatment on glass slides (A) and PP coupons (C) at the exposure time of 1 h; with D-tyrosine treatment on glass slides (B) and PP coupons (D) at the exposure time of 1 h.
3.3.
EPS production of suspended microorganisms
EPS are composed of a variety of organic substances, in which polysaccharides and proteins are two major components, and play an important role in microbial attachment onto a solid surface (Flemming and Wingender, 2010). Fig. 7 shows the respective contents of extracellular polysaccharide (PS) and protein (PN) in microorganisms with and without exposure to D-tyrosine. As compared to the control, a 31% reduction in PN
and 17% decrease in PS were observed in microorganisms after 1 h exposure to D-tyrosine, leading to a lowered PN/PS ratio. eDNA is a unique component of the organic substances in the matrix of suspended microorganisms. Fig. 8 shows eDNA content of suspended microorganisms with and without exposure to D-tyrosine. After 1 h exposure to D-tyrosine, eDNA was reduced to 0.006 mg g1 biomass, i.e. a 68% reduction compared to the control. These results suggest that D-tyrosine would significantly inhibit eDNA secretion.
Fig. 3 e Profiles of TOC removal efficiency (a) and microbial growth (b) in the cultures supplemented with (C) and without (B) D-tyrosine. Each point represents the mean of triplicate measurements and error bar is absolute deviation from the mean.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
Fig. 4 e Surface charge of suspended microorganisms with (C) and without (B) exposure to D-tyrosine. Each point represents the mean of triplicate measurements and error bar is absolute deviation from the mean.
Fig. 9 shows the HPSEC spectra of the EPS extracted from microorganisms with and without exposure to D-tyrosine. In general, retention time in HPSEC spectrum reflects molecular weight of a target chemical, i.e. the peak of a chemical with higher molecular weight appears quicker. For EPS extracted from microorganisms exposed to D-tyrosine, the peaks showed a significant decrease in area compared with those of the control, indicating a lowered EPS production that is consistent with the results in Fig. 7. Two peaks were observed for the EPS extracted from microorganisms without exposure to D-tyrosine at the retention time of 10e13 min, while only one peak appeared for the EPS from microorganisms exposed to D-tyrosine. This implies that EPS with higher molecular weight was reduced due to exposure to D-tyrosine. These suggest that D-tyrosine would not only inhibit the EPS production, but also can alter the EPS composition.
4.
Discussion
Figs. 1 and 2 show that microbial attachments onto hydrophobic PP and hydrophilic glass surfaces were inhibited by
5801
Fig. 6 e Response of AI-2 reporter strain in the cultures supplemented with (C) and without (B) D-tyrosine. Each point represents the mean of six measurements in parallel and error bar is absolute deviation from the mean.
D-tyrosine. So far, little information is available for the role of Dtyrosine in control of mixed-culture biofilm development. Kolodkin-Gal et al. (2010) reported that D-tyrosine-triggered release of amyloid fibers from cell surface would suppress development of a pure culture biofilm. It has been known that the long amyloid fiber facilitates the anchoring of cells to various surfaces, which is essential for microbial attachment and biofilm formation. In addition, the PP coupons used in this study have a contact angle of 99.3 2.2 , while 16.9 0.5 for glass slides. As observed in Figs. 1 and 2, more microorganisms attached to hydrophobic PP than to hydrophilic glass slide. In fact, it has been well documented that higher hydrophobicity of a solid surface would favor microbial attachment (Liu et al., 2004); on the contrary, coating surfaces with non-charged hydrophilic polymers resulted in reduced cell adsorption on a variety of surfaces (Park et al., 1998). Fig. 3 showed that D-tyrosine did not appear to affect the biomass growth and substrate removal efficiency. Such observation is consistent with the results obtained from pure culture experiments (Kolodkin-Gal et al., 2010). It had been reported that D-amino acids at a concentration higher than 20 mg l1 would inhibit bacterial growth (Teeri and Josselyn,
Fig. 5 e Cellular ATP content (a) and AI-2 concentration (b) in the cultures supplemented with (-) and without (,) D-tyrosine. Each point represents the mean of triplicate measurements and error bar is absolute deviation from the mean.
5802
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
Fig. 7 e PS (a) and PN (b) contents of suspended microorganisms with (-) and without (,) exposure to D-tyrosine. Each point represents the mean of triplicate measurements and error bar is absolute deviation from the mean.
1953). Peptidoglycan is an important mesh-like polymer component of cell wall. In the peptidoglycan polymer, there are two unique amino acids at the terminal of a peptide side chain of peptidoglycan: D-alanine as opposed to its isomer L-alanine. D-tyrosine can replace D-alanine in the peptide side chain of cell wall (Lam et al., 2009), and further alter the cell wall-building protein so that the peptidoglycan production would be slowed down, i.e. D-amino acid negatively regulate the amount of peptidoglycan production. In the presence of D-methionine, peptidoglycan synthesis could be severely inhibited, whereas biomass continued to grow (Caparros et al., 1992). The amount of peptidoglycan per cell decreased significantly due to the increased biomass and decreased peptidoglycan synthesis. However, it had been reported that Escherichia coli would be able to grow properly with 60% decrease of the normal peptidoglycan content (Prats and De Pedro, 1989). The unchanged cellular ATP synthesis in microorganisms with and without exposure to D-tyrosine (Fig. 5a) strongly supports this. It appears from Fig. 5b that AI-2 content of suspended microorganisms decreased due to D-tyrosine in the culture. To exclude other factors affecting AI-2 quorum sensing, the response of reporter strain V. harveyi BB170 to D-tyrosine
Fig. 8 e eDNA of suspended microorganisms with (C) and without (B) exposure to D-tyrosine. Each point represents the mean of triplicate measurements and error bar is absolute deviation from the mean.
showed that the inhibition of AI-2 regulated bioluminescence was only observed in presence of D-tyrosine (Fig. 6), which further confirmed that AI-2 expression was suppressed due to the presence of D-tyrosine. In study of D-amino acids regulated cell wall remodeling, Lam et al. (2009) found that exogenous Dmethionine produced by Vibrio cholera were incorporated into E. coli at the same position in the peptide even though E. coli bacterium did not produce or release D-amino acids. Thus, these rapid diffused small D-amino acids molecules could regulate cells releasing them and the neighboring cells of different species. In study of biofilm inhibition by D-amino acids, Kolodkin-Gal et al. (2010) hypothesized that D-amino acid may play an important role of chemical signal, but opposite to quorum sensing signal molecule, to mediate interspecies communication for facilitating cell dispersion from biofilm. In addition, AI-2 is known to be a cellular communication signal molecule both for Gram-negative and Grampositive bacteria and has a positive effect on biofilm formation (Federle and Bassler, 2003). Due to the opposite effects of these two signal molecules on biofilm formation, D-amino acid
Fig. 9 e HPSEC chromatograms of EPS extracted from suspended microorganisms with (---) and without (d) exposure to D-tyrosine at the exposure time of 1 h.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
may co-coordinate AI-2 regulated quorum sensing as shown in Fig. 6. However, how these two signals may co-regulate each other needs further investigation. Furthermore, it is speculated that D-amino acids may play a coordinating role through regulating the gene expression to control biofilm community structure or induce biofilm dispersion. Fig. 7 shows that the PS and PN contents of suspended microorganisms tended to decrease with exposure to D-tyrosine. Tsuruoka et al. (1984) also observed that D-amino acid caused reduction of lipoprotein in study of D-amino acid incorporation into peptidoglycan. It has been reported that an incorporation of D-tyrosine into the cellular proteins of Bacillus subtilis (Champney and Jensen, 1970) and E. coli (Miyamoto et al., 2010). As the D-isomer has a similar shape and size to the L-isomer molecule, the D-analog incorporated into proteins in the place of the natural amino acid would modify the structure of the proteins and the enzymic activity (Richmond, 1962). These would eventually lead to the reduced production of PS and PN. As can be seen in Fig. 9, high molecular-weight EPS at the retention time of 10e13 min disappeared in microorganisms exposed to D-tyrosine. In fact, in study of the effect D-amino acid on structure and synthesis of peptidoglycan, Caparros et al. (1992) also found a direct inhibition of D-methionine on the production of high molecular-weight proteins. In addition, EPS have been believed to play an important role in microbial attachment. The reduced production of PS and PN would result in inhibited microbial attachment (Fig. 1). Oliveira et al. (1994) reported that extracellular polysaccharides could promote a preconditioning of the surface, making attachment more favorable, whereas Flint et al. (1997) found that treatment the cells with trypsin or sodium dodecyl sulfate to remove cell surface proteins resulted in a 100-fold reduction in the attachment of Thermophilic streptococci onto stainless steel. It had been shown that eDNA would play an important role in initial microbial adhesion to hydrophobic and hydrophilic surfaces (Das et al., 2010). Many studies have shown that eDNA is an important component of extracellular network that mediates cellecell and cellesurface interactions (Bockelmann et al., 2006; Das et al., 2010). As can be seen in Fig. 8, the presence of D-tyrosine in the culture media caused reduction of DNA in the extracellular network, as the result, less attachment was observed both on glass and PP surface (Fig. 1). These suggest that eDNA may facilitate microbial attachment onto both hydrophilic glass and hydrophobic PP surfaces. Further study is needed to elucidate how D-amino acid would regulate eDNA production. In study of the role of eDNA in Listeria monocytogenes attachment, Harmsen et al. (2010) observed that peptidoglycan, specifically N-acetylglucosamine, together with eDNA could induce adhesion. Inhibited production of peptidoglycan and subsequently eDNA by D-tyrosine would also be responsible for reduced microbial attachment (Fig. 1). It appears from Fig. 4 that microorganisms carried more negative surface charge in presence of D-tyrosine. Since EPS often have charged functional groups, the higher negative charge density would be associated with changes in the composition and quantity of EPS induced by D-tyrosine. According to DLVO theory, increased negative charge would lead to strong electrostatic repulsion between cell and approaching surface (Zita and Hermansson, 1994). Hence, the extent of attachment
5803
would be less (Fig. 1) as a result of increased surface charge or a greater level of electrostatic repulsion.
5.
Conclusions
This study showed that D-tyrosine as a typical D-amino acid could inhibit microbial attachment on both hydrophilic glass and hydrophobic PP surfaces, while no inhibitory effect on microbial growth, ATP synthesis and substrate utilization was observed. It was further found that the synthesis of AI-2, eDNA and EPS were all reduced in the presence of D-tyrosine in the culture media. These in turn provide a plausible explanation for the D-tyrosine-triggered reduction in microbial attachment and demonstrate a mean for biological control of microbial attachment on a solid surface.
references
Bassler, B.L., 1999. How bacteria talk to each other: regulation of gene expression by quorum sensing. Current Opinion in Microbiology 2 (6), 582e587. Bockelmann, U., Janke, A., Kuhn, R., Neu, T.R., Wecke, J., Lawrence, J.R., Szewzyk, U., 2006. Bacterial extracellular DNA forming a defined network-like structure. FEMS Microbiology Letters 262 (1), 31e38. Caparros, M., Pisabarro, A.G., Depedro, M.A., 1992. Effect of Damino acids on structure and synthesis of peptidoglycan in Escherichia-coli. Journal of Bacteriology 174 (17), 5549e5559. Champney, W.S., Jensen, R.A., 1970. Molecular events in the growth inhibition of Bacillus subtilis by D-tyrosine. Journal of Bacteriology 104 (1), 107e116. Chen, G.H., Leung, D.H.W., 2000. Utilization of oxygen in a sanitary gravity sewer. Water Research 34 (15), 3813e3821. Comte, S., Guibaud, G., Baudu, M., 2007. Effect of extraction method on EPS from activated sludge: an HPSEC investigation. Journal of Hazardous Materials 140 (1e2), 129e137. Das, T., Sharma, P.K., Busscher, H.J., van der Mei, H.C., Krom, B.P., 2010. Role of extracellular DNA in initial bacterial adhesion and surface aggregation. Applied and Environmental Microbiology 76 (10), 3405e3408. Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., 1956. Colorimetric method for determination of sugars and related substances. Analytical Chemistry 28 (3), 350e356. Federle, M.J., Bassler, B.L., 2003. Interspecies communication in bacteria. Journal of Clinical Investigation 112 (9), 1291e1299. Flemming, H.C., Wingender, J., 2010. The biofilm matrix. Nature Reviews Microbiology 8 (9), 623e633. Fletcher, M., 1977. Effects of culture concentration and age, time, and temperature on bacterial attachment to polystyrene. Canadian Journal of Microbiology 23 (1), 1e6. Flint, S.H., Brooks, J.D., Bremer, P.J., 1997. The influence of cell surface properties of Thermophilic streptococci on attachment to stainless steel. Journal of Applied Microbiology 83 (4), 508e517. Harmsen, M., Lappann, M., Knochel, S., Molin, S., 2010. Role of extracellular DNA during biofilm formation by Listeria monocytogenes. Applied and Environmental Microbiology 76 (7), 2271e2279. Jia, X.S., Furumai, H., Fang, H.H.P., 1996. Yields of biomass and extracellular polymers in four anaerobic sludges. Environmental Technology 17 (3), 283e291. Kolodkin-Gal, I., Romero, D., Cao, S.G., Clardy, J., Kolter, R., Losick, R., 2010. D-amino acids trigger biofilm disassembly. Science 328 (5978), 627e629.
5804
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 7 9 6 e5 8 0 4
Lam, H., Oh, D.-C., Cava, F., Takacs, C.N., Clardy, J., de Pedro, M.A., Waldor, M.K., 2009. D-amino acids govern stationary phase cell wall remodeling in bacteria. Science 325 (5947), 1552e1555. Liu, Y., Yang, S.F., Li, Y., Xu, H., Qin, L., Tay, J.H., 2004. The influence of cell and substratum surface hydrophobicities on microbial attachment. Journal of Biotechnology 110 (3), 251e256. Liu, Y., Yang, S.F., Tay, J.H., 2003. Elemental compositions and characteristics of aerobic granules cultivated at different substrate N/C ratios. Applied Microbiology and Biotechnology 61 (5-6), 556e561. 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 (1), 265e275. McNab, R., Ford, S.K., El-Sabaeny, A., Barbieri, B., Cook, G.S., Lamont, R.J., 2003. LuxS-based signaling in Streptococcus gordonii: autoinducer 2 controls carbohydrate metabolism and biofilm formation with Porphyromonas gingivalis. Journal of Bacteriology 185 (1), 274e284. Miyamoto, T., Sekine, M., Ogawa, T., Hidaka, M., Homma, H., Masaki, H., 2010. Detection of D-amino acids in purified proteins synthesized in Escherichia coli. Amino Acids 38 (5), 1377e1385. Oliveira, R., Melo, L., Oliveira, A., Salgueiro, R., 1994. Polysaccharide production and biofilm formation by Pseudomonas fluorescens: effects of pH and surface material. Colloids and Surfaces B-Biointerfaces 2 (1e3), 41e46. Park, K.D., Kim, Y.S., Han, D.K., Kim, Y.H., Lee, E.H.B., Suh, H., Choi, K.S., 1998. Bacterial adhesion on PEG modified polyurethane surfaces. Biomaterials 19 (7e9), 851e859. Prats, R., De Pedro, M.A., 1989. Normal growth and division of Escherichia coli with a reduced amount of murein. Journal of Bacteriology 171 (7), 3740e3745.
Richmond, M.H., 1962. The effect of amino acid analogues on growth and protein synthesis in microorganisms. Bacteriological Reviews 26 (4), 398e420. Rickard, A.H., Campagna, S.R., Kolenbrander, P.E., 2008. Autoinducer-2 is produced in saliva-fed flow conditions relevant to natural oral biofilms. Journal of Applied Microbiology 105 (6), 2096e2103. Steinberger, R.E., Holden, P.A., 2005. Extracellular DNA in singleand multiple-species unsaturated biofilms. Applied and Environmental Microbiology 71 (9), 5404e5410. Surette, M.G., Bassler, B.L., 1998. Quorum sensing in Escherichia coli and Salmonella typhimurium. Proceedings of the National Academy of Sciences of the United States of America 95 (12), 7046e7050. Teeri, A.E., Josselyn, D., 1953. Effect of excess amino acids on growth of certain Lactobacilli. Journal of Bacteriology 66 (1), 72e73. Tsuruoka, T., Tamura, A., Miyata, A., Takei, T., Iwamatsu, K., Inouye, S., Matsuhashi, M., 1984. Penicillin-insensitive incorporation of D-amino acids into cell-wall peptidoglycan influences the amount of bound lipoprotein in Escherichia-coli. Journal of Bacteriology 160 (3), 889e894. Wilen, B.M., Jin, B., Lant, P., 2003. The influence of key chemical constituents in activated sludge on surface and flocculating properties. Water Research 37 (9), 2127e2139. Wood, T.K., Hong, S.H., Ma, Q., 2011. Engineering biofilm formation and dispersal. Trends in Biotechnology 29 (2), 87e94. Zita, A., Hermansson, M., 1994. Effects of ionic strength on bacterial adhesion and stability of flocs in a wastewater activated sludge system. Applied and Environmental Microbiology 60 (9), 3041e3048.
water research 45 (2011) 5805
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Erratum
Erratum to “Bioassays as a tool for evaluating advanced oxidation processes in water and wastewater treatment” [Water Research 45 (2011) 4311e4340] Luigi Rizzo Department of Civil Engineering, University of Salerno, via Ponte don Melillo 1, 84084 Fisciano (SA), Italy
On page 4319, left column, the sentence starting in line 11 from below should read: “According to the results available in scientific literature, AOPs were found to decrease and increase toxicity”. The subsequent sentence, “A decreased toxicity was . ., 2010).”, should be removed.
DOI of original article: 10.1016/j.watres.2011.05.035. E-mail address: [email protected]. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.08.013
