No title

COMMENT pubs.acs.org/est

Environmental Science & Technology Presents the 2011 Excellence in Review Awards

I

n recognition of the contribution that reviewers provide to the scientific community and the publication of scholarly research, ES&T Editor Jerald L. Schnoor and the associate editors of the journal are proud to recognize the following reviewers for their significant contributions to the journal over the past year. A journal exists on the goodwill and expertise of its reviewer corps. Our editors rely on high-quality, constructive, and timely reviews from the more than 5000 reviewers in our database to ensure the excellence of papers that appear in ES&T. We realize that reviewing is a time-consuming job with no tangible reward. In recognition of this, and in order to express our gratitude to some of our best reviewers, ES&T established an annual “Excellence in Review” award in 2003 to honor reviewers who have consistently provided both scholarly and timely reviews. This year, we introduce our first “Super Reviewer” award to acknowledge and express our appreciation to three previous winners who continue to provide large numbers of high quality reviews. We salute all of the individuals below.

Alison M. Cupples Michigan State University, East Lansing, Michigan http://www.egr.msu.edu/∼cupplesa/

’ SUPER REVIEWER AWARD

Arpad Horvath University of California, Berkeley, California http://www.ce.berkeley.edu/∼horvath/

David J. Ehresman 3M Company, St Paul, Minnesota Steven J. Eisenreich Joint Research Centre, European Commission, Brussels, Belgium http://jrc.ec.europa.eu Claudia Gunsch Duke University, Durham, North Carolina http://gunsch.pratt.duke.edu/ Stuart Harrad University of Birmingham, Birmingham, United Kingdom http://www.gees.bham.ac.uk/staff/harradsj.shtml

Bruce E. Logan Penn State University, University Park, Pennsylvania www.engr.psu.edu/ce/enve/logan/

Xia Huang Tsinghua University, Beijing, China http://www.tsinghua.edu.cn/publish/enven/index.html

Dionysios (Dion) D. Dionysiou University of Cincinnati, Cincinnati, Ohio http://www.eng.uc.edu/dept_cee/people/faculty/dionysiou

Arturo A. Keller University of California, Santa Barbara, California www.bren.ucsb.edu/∼keller

Jason C. White The Connecticut Agricultural Experiment Station, New Haven, Connecticut http://www.ct.gov/caes/cwp/view.asp?a = 2812&q = 345092

Tamar Kohn Ecole Polytechnique Federale de Lausanne, Switzerland http://lce.epfl.ch Jung-Hwan Kwon Ajou University, Suwon, Republic of Korea http://www.ajou.ac.kr/∼jhkwon

’ EXCELLENCE IN REVIEW AWARD Souhail Al-Abed USEPA, National Risk Management Research Laboratory, Cincinnati, Ohio www.epa.gov

Rainer Lohmann University of Rhode Island, Narragansett, Rhode Island http://www.gso.uri.edu/users/lohmann

Hans Peter Arp Norwegian Geotechnical Institute, Norway http://www.ngi.no/no/

Shaily Mahendra University of California, Los Angeles, California http://www.cee.ucla.edu/faculty/mahendra/profile

Laurie S. Balistrieri US Geological Survey, Seattle, Washington http://minerals.usgs.gov/west/spokane/laurie.htm

Jonathan W. Martin University of Alberta, Edmonton, Canada http://lmp.med.ualberta.ca/Admin/Faculty/Pages/default. aspx?P = 38

Patrick Brezonik University of Minnesota, Minneapolis, Minnesota http://www.ce.umn.edu/directory/faculty/brezonik.html r 2011 American Chemical Society

Published: October 28, 2011 9113

dx.doi.org/10.1021/es202956u | Environ. Sci. Technol. 2011, 45, 9113–9114

Environmental Science & Technology

COMMENT

Andrew A. Meharg University of Aberdeen, Aberdeen, Scotland http://www.abdn.ac.uk/biologicalsci/staff/details/a.meharg Hans W. Paerl University of North Carolina, Chapel Hill, North Carolina http://www.unc.edu/ims/paerllab/ Gene Parkin University of Iowa, Iowa City, Iowa http://www.engineering.uiowa.edu/faculty-staff/profiledirectory/cee/parkin_g.php Alessandro Piccolo Universita di Napoli Federico II, Portici, Italy http://www.suprahumic.unina.it Amy Pruden Virginia Tech, Blacksburg, Virginia http://www.cee.vt.edu/people/pruden.html Debra R. Reinhart University of Central Florida, Orlando, Florida http://www.cece.ucf.edu/people/reinhart/ John Sumpter Brunel University, London, United Kingdom http://www.brunel.ac.uk/about/acad/ife Hideshige Takada Tokyo University of Agriculture and Technology, Tokyo, Japan http://www.pelletwatch.org/ Ronald F. Turco Purdue University, West Lafayette, Indiana http://www.ag.purdue.edu/agry/Pages/rturco.aspx Clemens von Sonntag Max Planck Institute for Bioinorganic Chemistry, Muelheim an der Ruhr and University of Duisburg-Essen, Essen, Germany http://www.mpibac.mpg.de/bac/index_en.php Eddy Y. Zeng Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou, China http://sourcedb.cas.cn/sourcedb_gig_cas/yw/rckyw/ 200907/t20090710_2057308.html Jerald L. Schnoor Editor ES&T, University of Iowa

9114

dx.doi.org/10.1021/es202956u |Environ. Sci. Technol. 2011, 45, 9113–9114

COMMENT pubs.acs.org/est

Eawag at 75

F

ew institutions and none outside the U.S. have been associated with ES&T as closely as Eawag, the Swiss Federal Institute of Aquatic Science and Technology. This shared history includes the service of Prof. Alexander (Sascha) Zehnder (Eawag Director, 1993 2004) as Senior Associate Editor for Europe and that of numerous Eawag researchers (W. Giger, J. Hering, A. Johnson, L. Sigg, R. Schwarzenbach, U. von Gunten) as Associate Editors and/or members of the Editorial Advisory Board. This history reflects the common interests of ES&T and Eawag, which celebrates its 75th anniversary this year. Founded in 1936 as an Advisory Center for Wastewater Treatment and Drinking Water Supply at the Swiss Federal Institute of Technology (ETH) in Zurich, Eawag moved to its present location in D€ubendorf in 1970, concurrent with the arrival of its new Director, Professor Werner Stumm. ES&T began publishing in 1967, and its early years were also critical transition years for Eawag under Stumm’s leadership. Under Stumm, Eawag’s central focus on the fundamental understanding of the processes determining water quality in technical systems and in aquatic ecosystems was established. This focus on chemical, physical and biological processes was and is necessarily supported by advances in analytical and modeling capabilities. The strong science base developed under Stumm allowed Eawag to contribute to solving many of the acute and visible problems of environmental degradation that were widespread in industrialized countries, including Switzerland. In addition, Eawag initiated its 30-year tradition of research for development and capacity building in developing countries (see photo of Eawag researcher testing well water for arsenic with children in Sumatra).

natural sciences and engineering at Eawag. This was paralleled by the explicit inclusion of environmental policy analysis in ES&T beginning in 1995. The planning, design and construction of Eawag’s “zero-energy” building, the Forum Chriesbach, was initiated during Zehnder’s term and completed under his successor Ueli Bundi. Eawag’s research on No-Mix technology (featured as the cover article of ES&T issue 9 in 2001) is the basis of one of the many energy- and resource-conserving features incorporated in the Forum Chriesbach. Eawag’s emphasis on fundamental process understanding, the integration of engineering, natural and social sciences, and the implementation of research advances into practice are critical elements to solve urgent environmental problems worldwide. It is one of the great challenges of our time to meet societal needs for natural resources while preserving the capacity of the environment to provide essential ecosystem services. Research institutions, like Eawag, and prominent environmental journals, like ES&T, have an important role to play in ensuring a sustainable future. In addition to their primary roles in performing and disseminating research, such institutions can and should seek to build broader awareness of environmental issues, address the need to preserve environmental data and make them widely accessible, support collaboration and cooperation within the environmental science community, and help to identify and promote solutions to the pressing environmental problems of our time. Congratulations to Eawag for 75 years of such dedication and service to a quality environment. Janet G. Hering Director, Eawag Jerald L. Schnoor Editor

’ AUTHOR INFORMATION Corresponding Author

[email protected].

Lenny Winkel, Eawag Arsenic-sampling at a drinking fountain in Sumatra.

The overlapping interests of Eawag and ES&T provided the motivation for Stumm’s successor, Prof. Alexander Zehnder, to establish the ES&T European office at Eawag in 2000. Zehnder also emphasized the integration of the social sciences with the r 2011 American Chemical Society

Published: September 30, 2011 9115

dx.doi.org/10.1021/es203291e | Environ. Sci. Technol. 2011, 45, 9115–9115

VIEWPOINT pubs.acs.org/est

Legitimate Conditions for Climate Engineering Richard Owen University of Exeter, U.K. decades of knowledge and incorporated into trials of efficacy and safety. We do not have this for the emerging science of climate engineering and are therefore compelled to proceed under conditions of ignorance. The response is that we should establish strong research governance processes, developing and then employing tests of efficacy and safety before any decision to deploy (i.e., proceed with caution) in the same way we have built up understanding of pharmaceutical efficacy and safety over time and incorporated this into the tests required of medicines before use. This is to be recommended.

O

n September 13th scientists announced preparations were underway for the first UK field trial of climate engineering feasibility.1 The proposed trial will be modest: it will pump water through a 1 km high balloon-tethered hose, to assess the feasibility of reflective particle injection high into the atmosphere, mimicking the temperature-reducing effects of volcanic eruptions. But it has stimulated considerable debate about whether research in this controversial field should be undertaken at all, and if so the conditions under which it is acceptable to proceed. Responding, the President of the UK’s Royal Society, Paul Nurse, replied that there should be research on both the efficacy and safety of geoengineering:2 “One would not take a medicine that had not been rigorously tested to make sure that it worked and was safe. But, if there was a risk of disease, one would research possible treatments and, once the effects were established, one would take the medicine if needed and appropriate. Similarly we need controlled testing of any technologies that might be used in the future”. His comments, and specifically this analogy to pharmaceuticals, raise important questions concerning the conditions under which we decide to deploy controversial technologies such as solar radiation management. Pharmaceuticals indeed go through a rigorous testing process before they are authorized for use (“data before market”), but this is because we know the harmful effects to look for and there are well-established test methods to quantify these, built up over r 2011 American Chemical Society

’ THE LIMITS OF KNOWLEDGE There is however an important caveat to this approach. Despite our best intentions, it may only be once deployment has actually occurred that any nasty surprises surface. The history of nasty surprises is long, from CFCs to asbestos.3 Indeed surprises such as thalidomide were a major driver of regulation for pharmaceuticals, which in turn strives to ensure these effects do not occur again. But this happens after the fact. Regulation is often blind to that which it has not encountered before. Such unanticipated effects might not emerge for solar radiation management, but this will always be a gamble for which the probabilities can never be known, a point acknowledged by the Royal Society in 2009. The unintended side effects of many well-intentioned innovations have not been predicted. Here there is an analogy with pharmaceuticals: despite tests, who could have predicted that the birth control pill would cause environmental endocrine disruption?4 The argument is that research can help us rule out the technology, on the basis of efficacy, safety or both. But what happens if it is not ruled out? What if, after careful consideration of risks and feasibility, solar radiation management becomes a serious option? Who then would be prepared to place a bet for which the stakes can never be fully known? Perhaps the seriousness of climate change would make deployment a gamble worth taking. But who would make that decision? Who would have the authority to make a (possibly intergenerational) commitment to solar radiation management? Who would decide that conventional attempts at carbon management and climate change mitigation had proved insufficient or unsuccessful? Who would negotiate the distribution of impacts across the globe (beneficial or otherwise, known or unknown) that might result? Who would compensate those who suffer for the collective good? What are the conditions for such planetary technological gambles? Received: September 21, 2011 Accepted: September 28, 2011 Published: October 10, 2011 9116

dx.doi.org/10.1021/es2033185 | Environ. Sci. Technol. 2011, 45, 9116–9117

Environmental Science & Technology

VIEWPOINT

In his essay on the Imperative of Responsibility5 Jonas wrote “One would not deny the statesman the right to risk his nation’s existence for its future if really ultimate issues are at stake. It is in this manner that awesome but morally defensible decisions about war and peace come about when, for future’s sake, the stake is the future itself’. He added that ‘this should never happen because of the enticement of a wonderful future but only under the threat of a terrible future”. The supreme “malum” justifies a collective wager. This has been the catalyst for many technological wagers in the past, of which the push for mass production of penicillin in World War Two is arguably one. Would the prevention of a terrible future (a “climate change tipping point” for example) be a legitimate condition for a collective gamble on a geoengineering solution? Perhaps, but this presupposes that this condition has been collectively arrived at, and that there is a mechanism for this to be achieved. This does not currently exist. It is particularly important for approaches such as solar radiation management which may have impacts that may be trans-national and unequally distributed in nature. It is here that Nurse’s statement “one would take the medicine if needed and appropriate” becomes critical. Who will decide there is a need, that climate engineering is an “appropriate medicine”? The conditions for making such a technological wager legitimate, democratic, and equitable must be explicit. For we are naive to assume the decision to administer a medicine by a physician is based on efficacy and safety alone. And we will pay the price if we fail to acknowledge that there are limits to knowledge and ignore the lessons of history.

’ REFERENCES (1) http://www.nerc.ac.uk/press/releases/2011/22-spice.asp (accessed June 10, 2011). (2) http://www.guardian.co.uk/environment/2011/sep/08/geoengineering-research-royal-society (accessed June 10, 2011). (3) http://www.eea.europa.eu/publications/environmental_issue_ report_2001_22 (accessed June 10, 2011). (4) Jobling S., Owen R. Ethinyl oestradiol: Bitter pill for the precautionary principle. In: Late Lessons from Early Warnings II; European Environment Agency, (in press). (5) Jonas, H. The Imperative of Responsibility; University of Chicago Press, Chicago, 1984.

9117

dx.doi.org/10.1021/es2033185 |Environ. Sci. Technol. 2011, 45, 9116–9117

VIEWPOINT pubs.acs.org/est

Green Power A Modest Experiment Michael R. Piggott* Chemical Engineering and Applied Chemistry, University of Toronto

A

t the present time, the Ontario government is encouraging people to invest in solar and wind power generation. They do this by subsidizing any excess power produced and fed into the grid at a rate many times the cost of power generated from big conventional plants. The Ontario Power Authority is offering feed-in tariff contracts at between 45 and 80.2 cents to companies building new solar power generating facilities, 13.5 cents on landbased wind farms, 19 cents on off-shore wind farms, and between 10.4 and 19.5 cents on biogas projects.1 3 The experiment described herein is a test of the economics of domestic production from a small and easily affordable system, on the shore of Lake Huron, in the light of these figures. Two solar panels, with a rated capacity of 80 W each, were mounted on the roof of a shed, which was located in an exposed spot. They faced approximately south and were about 23° to the horizontal. Having only 100 foot wide lot, a tall tower was not practical for a wind generator, since it would have to be guyed sufficiently to withstand gale force winds from any direction. So a 900 W wind generator was mounted between two cedar tree trunks, about 35 feet tall, set two feet apart in an approximately east west direction, and another cedar trunk, about 20 feet tall, set at 12 feet north of the westmost 35 foot cedar. They were braced by connecting them to each other, and to the shed. The cedar posts were embedded in reinforced concrete, resting on the bedrock, about two feet beneath. The wind generator was attached to the top of a 2 1/2 in. diameter galvanized steel pipe, about 24 feet long. The pipe was in two halves joined at the middle by a specially shaped hollow rod. The assembly was pivoted in the middle by a one in. diameter hardened steel rod. This steel rod went right through the support cedars, with the 2 1/2 in. pipe assembly in the middle. The generator was fixed to one end of the assembly, and balanced by steel slugs affixed inside the other end. Thus the generator could be swung down to connect the electrical conductors, and swung up again for wind generation of electrical power. (And swung down again for service.) This structure withstood gale force winds (sustained winds of 63 84 km/h) on many occasions. A gust of 162 km/h was r 2011 American Chemical Society

recorded on the anemometer in May 2010 and another, of 165km/h, in November of that year. The three phase electrical power produced by the generator was regulated and rectified by a special “charge controller”, and was stored in deep cycle lead-acid batteries. The power produced was dissipated as heat. An automatic switching system (Schmidt trigger) was designed and constructed to turn the 115 V heater on when the battery voltage reached 12.8 V, and turn it off again when the voltage fell to 11.4 V. The time that the heater was turned on was logged. With the aid of an anemometer, the power output of the generator was plotted as a function of wind velocity, averaged over about 12 h. This showed that a 16km/h wind was required before any power was generated. Thereafter, the power increased approximately linearly to 340 W at 40km/h. By extrapolation of the straight line, it was predicted that a 50km/h wind was required to produce 500 W. But above 30km/h wind velocity, the generator was designed to fold, so it was avoiding the full force of the wind. The manufacturer’s design figure of 900 W at 45km/h thus appears to be very optimistic, as reported elsewhere for these devices.4 (Note that the power was fed through an inverter to produce 120 V AC from 12 V DC. This conversion was about 90% efficient. So 900 W from the generator produces about 810 W of useable AC power.) The power produced by the solar panels working in tandem with the wind generator was logged for 12 months, from January 1 2010 to January 31 2011, see Figure 1. (April 2010 did not yield reliable results, due to instrument problems.) While there are significant variations, the overall mean useable power produced was 28.3 W. The solar panels only contributed a significant amount of power in the summer. They were covered in snow in December, January, February, and part of March. The monthly averages conceal much day-to-day variation. For instance, in June 2010 the power varied between 2 and 120 W. The out-of pocket cost of the setup was about 7450 Canadian dollars. The wind generator cost $2933, the solar panels $1333, the batteries $575, the inverter $580, and the concrete and tower hardware about $1150. Heavy cable was needed for the wind generator charge controller battery connections, and 115/230 V cable was needed to connect the system to the house. Hidden costs include the cedar tree trunks, the design and building of the Schmidt trigger from its basic components, and the value of the excavation, etc. work associated with the construction of the wind generator support. The current (March 2011) cost of domestic electricity in the household concerned with the project was 14.82 cents per Received: September 21, 2011 Accepted: September 26, 2011 Published: October 05, 2011 9118

dx.doi.org/10.1021/es2033229 | Environ. Sci. Technol. 2011, 45, 9118–9119

Environmental Science & Technology

VIEWPOINT

Figure 1. Monthly average power produced by the combined efforts of the solar panels and wind generator. Overall average for 12 months was 28.3 W. (Allowing for the 90% efficiency of the inverter, that amounts to 31.4 W at 12 V DC.).

kilowatt hour. $7450 would, at that rate buy 50,247 kWh. The solar/wind power assembly, working at an average rate of 30 kW would take about 190 years to produce the same amount of electricity. While there is no figure given for the feed-in tariff for solar/ wind domestic generation, we could take an average between 80.2c (solar) and 13.5c (wind farm), that is, 46.85c, and see how long the power generated would have to be fed into the grid to recover $7450 at that rate. We thus make the same calculation as in the previous paragraph, replacing 14.82c with 46.85c. This comes to about 60 years. Thus wind/solar power is not a good investment. In addition to spending $7430, there was a lot of time spent designing and constructing the system. All for about 30 W.

’ AUTHOR INFORMATION Corresponding Author

*Phone 519-534-5592; e-mail [email protected].

’ REFERENCES (1) Corcoran, T.; National Post (Toronto) September 30th 2010. (2) Source Watch: Comparative Electrical Generation Costs; California Regulatory Agencies data for May 2008. (3) Ontario Power Authority FIT Power Price Schedule, August 13th 2010. (4) Consumer Reports, October 2011, page 30.

9119

dx.doi.org/10.1021/es2033229 |Environ. Sci. Technol. 2011, 45, 9118–9119

VIEWPOINT pubs.acs.org/est

The Challenge of Exposure Correction for Polar Passive Samplers— The PRC and the POCIS Christopher Harman,†,* Ian John Allan,† and Patrick Steven B€auerlein‡ †

Norwegian Institute for Water Research (NIVA), Oslo Centre for Interdisciplinary Environmental and Social Research (CIENS), Gaustadalleen 21, NO-0349, Oslo, Norway ‡ KWR Watercycle Research Institute, P.O. Box 1072, 3430 BB, Nieuwegein, The Netherlands

P

assive sampling devices such as the polar organic chemical integrative sampler (POCIS)1 may have much to offer in response to the challenge of measuring low and fluctuating concentrations of polar compounds of interest in aquatic environments. For example they have recently been used to obtain illicit drug monitoring data that would not have been practically and financially possible to achieve using other sampling methods.2 One of the biggest challenges facing the quantitative use of such samplers is the lack of a method to correct for in situ exposure conditions (water flow rates, temperature, pH etc.) which are known to affect uptake rates. This issue has been elegantly overcome for hydrophobic passive samplers by the use of so-called performance reference compounds (PRCs).3 These analytically noninterfering substances are spiked into samplers prior to deployment and as their dissipation follows first order kinetics analogous to uptake they can be used to estimate sampling rates (Rs) of target compounds in situ. This is not currently the case for polar passive samplers. Despite a comprehensive consideration of the processes involved in chemical uptake in the original description of POCIS,1 most of the subsequent work has concentrated on making measurements rather than trying to understanding the mechanisms involved. r 2011 American Chemical Society

Thus theories from hydrophobic passive sampling have been applied for the polar samplers despite the fundamentally different processes of two-way isotropic exchange in hydrophobic samplers and sorption phenomena in polar samplers. Such an approach might be risky, especially with the range of interactions that need to be taken into account when considering polar compounds, particularly those containing N and O; H-bonding between water and solute, between water and sorbent, between solute and sorbent, van der Waals interactions etc. The majority of published POCIS studies (including our own) have made measurements directly in the environment using Rs derived from laboratory calibrations. Of these the static renewal type experiment dominates, which although likely to be the most reproducible between studies, is the one most unlikely to be replicated under field conditions. This means that although results are often presented as time-weighted average concentrations, in reality, due to the absence of a dedicated PRC approach (or equivalent), they are at best semiquantitative. For exactly these reasons investigations have been undertaken to develop a PRC approach for POCIS, built around the observation that some compounds could be released again after uptake.4 This is perhaps unsurprising as it was earlier shown that OASIS HLB exhibits both the properties of an adsorbent, and a contribution of some partitioning mechanism.5 All compounds that are suitable to be POCIS PRCs, that is, have the ability to desorp (or poor sorption) are likely to have relatively similar retention mechanisms for the sorbent. This means finding suitable PRCs for compounds which do not behave in this way, that is, those which more strongly bound might be problematic. The question thus arises can one correct for the other? As one objective of the PRC approach is to use several compounds to correct for a suite of similar ones, this is an important question. This may be plausible if the change in Rs due to a reduction in the water boundary layer for example, is equally represented for weakly bound PRCs on the way out of the sampler and strongly bound analytes on the way in. However, the situation is further complicated by an apparent increasing interaction with the polyethersulphone membrane used in POCIS, with increasing target compound hydrophobicity and the importance of apolar moieties generally for retention on OASIS HLB.5 Thus (assumed) weakly bound and rapidly equilibrating compounds that appear suitable for use as POCIS PRCs are unlikely to be Received: September 26, 2011 Accepted: September 27, 2011 Published: October 06, 2011 9120

dx.doi.org/10.1021/es2033789 | Environ. Sci. Technol. 2011, 45, 9120–9121

Environmental Science & Technology

VIEWPOINT

able to be used to correct for those with different interactions. Additionally as sorption is a competitive process, it is plausible that compounds with strong interactions can replace those with weak ones including PRCs? If this is the case then higher PRC dissipations may be seen, and thus higher Rs estimated, with increasing concentrations of competitive substances in exposure water. As a result even if it can be shown in laboratory experiments that the PRC approach works for certain compounds and exposure scenarios, these may not necessarily be applied to all compounds and all exposures. A point Mazzella and co-workers4 are careful to point out, but one which appears to be increasingly overlooked, based on presentations at the recent passive sampling workshop and symposium (Krakow, Poland). Other external correction methods may be applicable, such as using hydrophobic samplers as a “surrogate PRC approach”, or in situ calibration.2 However these approaches are also not without their challenges. Additionally the use of different materials, for example silicone rubber, which appears to be able to sample several groups of medium polar compounds, is worth investigating. There are many gaps in our knowledge concerning polar samplers and the processes involved in accumulation. For example the first detailed studies considering transport through the PES membrane are only just emerging. In summary we need to address some of these basic questions before we can hope to use polar passive samplers in a truly quantitative way. Even accepting that OASIS HLB may exhibit partitioning properties for some compounds, we should not correct sorption phenomena which we understand poorly with desorption phenomena (PRCs) we understand even less. The use of PRCs for certain exposure situations may be possible, but an all-encompassing, robust approach for POCIS and similar types of samplers appears unlikely. We suggest that, currently, polar passive samplers are a useful screening tool, which may be used to estimate not calculate the order of water concentrations.

’ AUTHOR INFORMATION Corresponding Author

*Phone: +47 22 18 51 00; fax: +47 22 18 52 00; e-mail: [email protected].

’ REFERENCES (1) Alvarez, D.; Petty, J.; Huckins, J.; Jones-Lepp, T.; Getting, D.; Goddard, J.; Manahan, S. Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environ. Toxicol. Chem. 2004, 23, 1640–1648. (2) Harman, C.; Reid, M.; Thomas, K. V. In situ calibration of a passive sampling device for selected illicit drugs and their metabolites in wastewater, and subsequent year-long assessment of community drug usage. Environ. Sci. Technol. 2011, 45, 6233–6234. (3) Booij, K.; Sleiderink, H.; Smedes, F. Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environ. Toxicol. Chem. 1998, 17, 1236–1245. (4) Mazzella, N.; Lissalde, S.; Moreira, S.; Delmas, F.; Mazellier, P.; Huckins, J. N. Evaluation of the use of performance reference compounds in an oasis-HLB adsorbent based passive sampler for improving water concentration estimates of polar herbicides in freshwater. Environ. Sci. Technol. 2010, 44 (5), 1713–1719. (5) Dias, N. C.; Poole, C. F. Mechanistic study of the sorption properties of OASIS HLB and its use in solid phase extraction. Chromatographia 2002, 56, 269–275. 9121

dx.doi.org/10.1021/es2033789 |Environ. Sci. Technol. 2011, 45, 9120–9121

CRITICAL REVIEW pubs.acs.org/est

Connecting the Dots: Responses of Coastal Ecosystems to Changing Nutrient Concentrations Jacob Carstensen,*,† María S
anchez-Camacho,‡ Carlos M. Duarte,‡,§ Dorte Krause-Jensen,† and N
uria Marba‡ †

Department of Bioscience, Aarhus University, Frederiksborgvej 399, DK-4000 Roskilde, Denmark Department of Global Change Research, IMEDEA (CSIC-UIB), Instituto Mediterr
aneo de Estudios Avanzados, Miquel Marques 21, 07190 Esporles (Illes Balears), Spain § The UWA Oceans Institute, University of Western Australia, 35 Stirling Highway, Crawley 6009, Australia ‡

bS Supporting Information ABSTRACT: Empirical relationships between phytoplankton biomass and nutrient concentrations established across a wide range of different ecosystems constitute fundamental quantitative tools for predicting effects of nutrient management plans. Nutrient management plans based on such relationships, mostly established over trends of increasing rather than decreasing nutrient concentrations, assume full reversibility of coastal eutrophication. Monitoring data from 28 ecosystems located in four well-studied regions were analyzed to study the generality of chlorophyll a versus nutrient relationships and their applicability for ecosystem management. We demonstrate significant differences across regions as well as between specific coastal ecosystems within regions in the response of chlorophyll a to changing nitrogen concentrations. We also show that the chlorophyll a versus nitrogen relationships over time constitute convoluted trajectories rather than simple unique relationships. The ratio of chlorophyll a to total nitrogen almost doubled over the last 30
40 years across all regions. The uniformity of these trends, or shifting baselines, suggest they may result from large-scale changes, possibly associated with global climate change and increasing human stress on coastal ecosystems. Ecosystem management must, therefore, develop adaptation strategies to face shifting baselines and maintain ecosystem services at a sustainable level rather than striving to restore an ecosystem state of the past.

’ INTRODUCTION Increased nutrient inputs to coastal ecosystems, derived from the rapid rise in fertilizer use in agriculture, production of manure from farm animals, domestic sewage, and atmospheric deposition associated with fossil fuel combustion,1
4 have led to the global spread of coastal eutrophication since the late 1970s. Realization of the negative effects of eutrophication, involving the loss of value of coastal ecosystem services,1,5,6 prompted efforts, initiated in the late 1980s, to reduce nutrient inputs. The result was expected to be a phase of oligotrophication with decreasing primary production7 which would reverse the effects of eutrophication and return coastal ecosystems to an earlier state.4,8,9 However, recent analyses have provided evidence that reduced nutrient inputs often fail to fully reverse the trajectories of ecosystems during eutrophication and have challenged the assumption that oligotrophication drives coastal ecosystems back to their original condition.10 The expectation that reduced nutrient inputs would reverse eutrophication effects originated from predictions derived from broad-scale relationships between chlorophyll a concentration (Chla), as an indicator of algal biomass, and nutrient concentrations across coastal ecosystems.11
13 These relationships were r 2011 American Chemical Society

comparable to those developed in the 1970s for lake ecosystems14
16 and confirmed experimentally (e.g., 17
19). Yet, the empirical basis supporting use of the general relationship to predict oligotrophication responses was lacking, as all case studies and experimental tests to the 1990s reflected ecosystem responses to addition of nutrients, rather than to their removal. Hence, the use of relationships between Chla and nutrient concentrations to predict the response of coastal ecosystems to oligotrophication rests on the assumption that eutrophication is a fully reversible process involving a single path identical for eutrophication and oligotrophication trajectories. This fundamental tenet underlying coastal ecosystem management has not been sufficiently tested to date, but the current availability of dozens of cases of individual ecosystems undergoing eutrophication and subsequent oligotrophication now allows such tests to be conducted. Empirical relationships between Chla and nutrient concentrations, presented as general static relationships, have been based Received: December 1, 2010 Accepted: September 29, 2011 Revised: September 23, 2011 Published: September 29, 2011 9122

dx.doi.org/10.1021/es202351y | Environ. Sci. Technol. 2011, 45, 9122–9132

Environmental Science & Technology

CRITICAL REVIEW

Table 1. Surface Samples of Water Quality Data Used in the Analysisa no. of observations region

no. of coastal ecosystems

no. of annual means

years

TN

TP

Chla

TN

TP

Chla

source www.chesapeakebay.net

Chesapeake Bay

7

1984
2006

17133

17465

21427

149

149

159

Denmark coast

10

1977
2006

12409

12354

12211

256

257

257

www.dmu.dk

Tampa Bay

4

1977
2006b

12825

17889

17885

104

120

120

www.tbeptech.org

Wadden Sea (Dutch part)

7

1977
2006c

3828

3878

3323

203

200

204

www.waterbase.nl

a

Data were downloaded from public monitoring databases with long time series (>20 years) of coastal ecosystems from four regions. b No TN data before 1981. c TN and TP calculated as sum of measured particulate and filtered fractions between 1991 and 1996.

on data from many different ecosystems encompassing wide ranges of Chla and nutrient concentrations.12,13,20 However, the broad, order-of-magnitude variability characteristic of these relationships may not represent only random variability, but may partially derive from diverse and idiosyncratic responses within individual ecosystems and/or changes in the nature of these relationships over time. Indeed, our conceptual understanding of ecosystem functioning has evolved from the initial expectation of a uniform response to accommodate a diversity of responses to nutrient inputs.8 Increased availability of long-term time series describing the responses of coastal ecosystems to changes in nutrient concentrations now makes it possible to connect the dots in Chla
nutrient relationships to examine the trajectories of individual ecosystems over time10 as well as to examine variability in trajectories among ecosystems. Such analyses may provide an improved basis to derive expectations on the possible response of individual coastal ecosystems to increases as well as to decreases in nutrient inputs. Here we use long-term monitoring data to explore the existence of general patterns in the relationship between Chla and nutrient concentrations in 28 coastal ecosystems from four different regions where eutrophication has led to the implementation of management plans to reduce nutrient inputs. We aim to examine whether the relationship between Chla and nutrient concentration follows similar pathways through periods of eutrophication and oligotrophication. We do so by deconstructing general Chla
nutrient relationship to examine the variability among regions and individual ecosystems, as well as the variability in these relationships over time. Data Sources and Processing. Water quality data were obtained from four public monitoring databases in Europe and North America (Table 1), constituting some of the most comprehensive and longterm data sets in the world. Although some of the databases had observations prior to 1977, the analysis was restricted to 1977
2006 for consistency across regions. The compiled database included over 45 000 observations of nutrient as well as chlorophyll a concentrations from the surface layer. Nutrient and Chla concentrations, the most widely used indicators of nutrient-driven eutrophication in the literature, were used as proxies of organic enrichment in coastal ecosystems.1 Sampling stations with salinity 130

>2.1
2.9

19
21

Chernobyl accident

0.31
0.40

135
213

0.49
0.56

20, 22

Pacific proving ground

0.306
0.36

27a

0.001
0.014

13, 23

Nagasaki atomic bomb

0.028
0.037

1.21b

0.074

24, 25

Thule hydrogen bomb debris Semipalatinsk nuclear test site

0.0551 0.03
0.05

0.87c


0.0161 0.0064
0.085

26 27, 28

a 241 +240

Pu/239+240Pu activity ratio, reference date of 241Pu: 1952
1954. b 241Pu/239+240Pu activity ratio, reference date of 241Pu: August 1945. c 241Pu/239 Pu activity ratio, reference date of 241Pu: October 2001.

and activity ratio of 239+240Pu/137Cs are also useful for the source identifications and dating in the aquatic sediments, especially in lakes close to nuclear testing sites.6,13,14 In particular, the isotopic composition of Pu is characteristic for various Pu sources (Table 1),13,15
28 and therefore accepted as a good indicator for identifying Pu sources in the environment. However, information on the distribution, inventory, and sources of fallout radionuclides in China is very limited, which severely hampered the study of recent lake sediment dating needed in China for the reconstruction of the pollution history of organic pollutants and heavy metals. It is known that Chinese atmospheric nuclear tests were conducted at the Lop Nor nuclear test site in northwestern China since 1964. However, the early nuclear weapon research and development activities performed at Atomic City in Lake Qinghai’s watershed remain unknown.29,30 This research center, located in Haibei Prefecture, about 20 km northeast of Lake Qinghai was constructed in 1958 for China’s independent development and assembly of nuclear weapons.29 It was here that China’s first atomic and hydrogen bombs were designed and developed in the 1950s and 1960s. It was completely shut down in 1987 following an official environmental safety evaluation and was officially declassified in 1995. It is now open to the public as a National Patriotism Education Demonstration Base. Since studies are very limited, it is not clear whether there were sediment records of radionuclides resulting from the early nuclear activities in Lake Qinghai. If so, the deposited Pu may bear a unique isotope composition different from that of global fallout source Pu, thus provide a new tool for recent sediment dating in the area. In the present work, in order to illustrate the possible influence of global fallout, the early nuclear activities conducted in the watershed, and close-in fallout from Lop Nor in Lake Qinghai, three sediment cores in the lake are collected, and vertical profiles of 239+240Pu and 137Cs activities and 240Pu/239Pu isotopic ratios are investigated and compared with those of another three lakes (Lakes Bosten, Sugan, and Shuangta), the only existing ones closest to Lop Nor area, China’s nuclear weapon test site. For the first time, the Pu input from the early nuclear research and development activities in Atomic City with a 240Pu/239Pu atom ratio well below typical global atmospheric fallout levels was found. The unique 240Pu/239Pu isotopic ratios and the vertical profiles of 239+240Pu and 137Cs activities in the sediments were found to be important for the further understanding of possible environmental impact of regional nuclear activities and useful for recent sediment dating which is needed for investigation of pollution histories in the lake and its watershed.

Figure 1. Map showing the locations of Lake Qinghai, Lake Bosten, Lake Sugan, and Lake Shuangta, and sampling sites in Lake Qinghai and Atomic City.

’ EXPERIMENTAL SECTION Sample Collection. Lake Qinghai (99360
100160 E,

36320
37150 N) is the largest inland salt lake in China with a lake surface area of 4278 km2, a watershed area of 16 570 km2, a lake surface height of 3193 m above sea level, a maximum depth of 26.5 m, a water pH value of 9.2 on the average, a mineralization degree of 12.3 g L
1, and a salinity of 14.2 %.31,32 Lake Qinghai is situated on the Qinghai-Tibetan Plateau in Qinghai Province, China, and is about 1000 km away from Lop Nor, China’s nuclear weapons test site, together with locations of Lake Bosten, 9189

dx.doi.org/10.1021/es202315c |Environ. Sci. Technol. 2011, 45, 9188–9194

Environmental Science & Technology

ARTICLE

Figure 2. Vertical profiles of 239+240Pu and 137Cs activities and 240Pu/239Pu atom ratios in the sediments of Lake Qinghai.

Shuangta, and Sugan. China’s nuclear weapons research and development center is located in Haibei Prefecture, about 20 km northeast of Lake Qinghai in the watershed (Figure 1). Three sediment cores were sampled in Lake Qinghai in 2006 (Figure 1). Three sediment cores were collected in the center area of Lakes Bosten, Sugan, and Shuangta in 2006
2007, respectively. Each sediment core was sectioned at 0.5-cm intervals in the field. The sediment samples were weighed immediately after collection. They were kept cool until brought to a laboratory and were then freeze-dried, weighed, and ground to 120 mesh for further analyses of 137Cs and 239+240Pu activities and 240Pu/239Pu isotopic ratios. Analytical Procedure. 137Cs activity was determined using gamma-spectrometry on a Canberra S-100 multichannel spectrometer with a GC5019 H
P Ge coaxial detector (efficiency 50%) at the Institute of Geochemistry of the Chinese Academy of Sciences. The 137Cs peak used to determine its activity was 661.6 KeV. Liquid standard (Catalog No. 7137) supplied by the Institute of Atomic Energy of the Chinese Academy of Sciences, was used. For 137Cs activity measurements, the relative standard deviation was within 10%.33 Activities of 239Pu and 240Pu were determined at National Institute of Radiological Sciences, Japan. Pu isotope separation and analytical methods were described in our previous reports.34,35 Briefly, 0.5
2.0 g samples were digested using HNO3 to extract Pu isotopes. AG 1-X8 and AG MP1-M anion resins were used in the separation and purification process. The analytical procedure was characterized by 75% Pu chemical recovery and a U decontamination factor of 105. 239Pu and 240Pu were analyzed using a sector-field ICP-MS (Element 2,

Finngan, Germany) combined with a high efficiency sample introduction system (APEX-Q). Sediment standard reference materials IAEA-368 (marine sediment standards, International Atomic Energy Agency) and SRM-4354 (freshwater lake sediment standards, American National Standards Institute of Technology) were used for analytical method validation. The results obtained for the two reference materials clearly show the suitability of the analytical procedure for the analysis of sediment samples (Table 1 in Supporting Information).

’ RESULTS AND DISCUSSION Vertical Profiles of Pu Isotopes and

137

Cs in Sediments. The vertical profiles of Pu and Cs activities in sediments of Lake Qinghai are characterized by a single-peak distribution pattern (Figure 2). The peak values of 239+240Pu and 137Cs activities appear at the mass depths of 0.459 and 0.716 g cm
2 in 2006QH-2 and 2006QH-3, respectively. 239+240Pu activities are as high as 4.74 and 6.99 mBq g
1, and 137Cs activities are 94.62 and 112.58 mBq g
1, respectively. These vertical profiles are consistent with those found in other lakes in China3,36 and are also similar to those in other lacustrine and marine sediments.37
39 This suggests that the single peak of 239+240Pu and 137Cs activities in the sediments corresponds to the 1963
1964 global fallout. In the sediment from the surface to the layer of peak deposition in 1964, 240Pu/239Pu isotopic ratios range from 0.169 ( 0.007 to 0.228 ( 0.042 (Figure 2; Table 2 in Supporting Information); the inventory-weighted mean 240Pu/239Pu atom ratios are 0.181, 0.175, and 0.170 for 2006QH-1, 2006QH-2, and 239+240

9190

137

dx.doi.org/10.1021/es202315c |Environ. Sci. Technol. 2011, 45, 9188–9194

Environmental Science & Technology

ARTICLE

Figure 3. Vertical profiles of 239+240Pu and 137Cs activities and 240Pu/239Pu atom ratios in the sediments of Lake Bosten, Lake Sugan, and Lake Shuangta (redrawn from Wu et al.6,41).

2006QH-3, respectively. These values are close to 0.180 ( 0.014, the value reported for the global atmospheric fallout.16 This provides further isotopic evidence for the source of 239+240Pu and 137 Cs in the surface sediments, i.e., they originated from global atmospheric fallout. However, in the deep sediment layers below the 1964 peak (below mass depth of 1.60 g 3 cm
2 in 2006QH-1, 0.46 g 3 cm
2 in 2006QH-2, and 0.72 g 3 cm
2 in 2006QH-3, respectively), 240Pu/239Pu ratios range from 0.038 to 0.159, which are significantly lower than those in the surface sediments, and they also decrease as depth increases. Previous studies have shown that 240Pu/239Pu ratios from nuclear weapon-grade materials ranged from 0.01 to 0.07,17 those from nuclear reactors ranged from 0.24 to 0.80, and the global atmospheric fallout prior to 1963 was slightly higher than 0.18.13,21,40 Therefore, the low 240 Pu/239Pu atom ratios in the deep sediment layers may suggest the possible occurrence of weapon-grade Pu input in the lake, which may not originate from the global atmospheric fallout. We also analyzed 241Pu/239Pu activity ratio in sediment cores of 2006QH-2 and 2006QH-3. Although the obtained 241Pu/239Pu activity ratios have relatively high RSD (25
68%) due to the low activity of 241Pu, they ranged from 2.35 to 3.04, close to the value of 2.7 for global fallout,15,16 in the sediment layers above the 1964

maximum deposition peak while low values of 0.55 to 2.08 were seen in the layers below the 1964 peak (Table 2 in Supporting Information). The plot of 240Pu/239Pu atom ratio vs 241Pu/239Pu activity ratio for 2006QH-2 and 2006QH-3 sediment cores (Figure 1 in Supporting Information) indicated the mixing of local source Pu input and the global fallout source in Lake Qinghai. Another possibility for the low 240Pu/239Pu atom ratios in Lake Qinghai is the influence of China’s nuclear weapons tests. China’s nuclear weapons test site, where the first nuclear test took place on October 16, 1964, is Lop Nor which is 1000 km from Lake Qinghai.29 240Pu/239Pu atom ratios in the sediments of Lakes Bosten, Sugan, and Shuangta near Lop Nor are generally within the global atmospheric fallout range (Figure 3) except at two incident mass-depth layers of 2.83 g 3 cm
2 in 07Bosten 10
2 (Lake Bosten) and 0.90 g cm
2 in 2007SG-2 (Lake Sugan) with ratios of 0.080 ( 0.016 and 0.103 ( 0.009, respectively.6,41 Those two layers are located at or above the 239+240Pu peak layer. The ratios are significantly lower than the global fallout value of 0.180 but slightly higher than the nuclear weapon-grade characteristic value (0.01
0.07), showing the limited influence of China’s atmospheric nuclear weapon tests. As the observed 240 Pu/239Pu atom ratios in the deep layers of Lake Qinghai 9191

dx.doi.org/10.1021/es202315c |Environ. Sci. Technol. 2011, 45, 9188–9194

Environmental Science & Technology

Figure 4. Inventories of 239+240Pu and 137Cs in the sediments of Lake Qinghai (note: 137Cs activity was corrected back to July 1998; for comparison, the average values of global atmospheric fallout in 30
40N latitude bands are also provided).

sediments are even lower than those found in Lake Bosten and Lake Sugan, and the sediment layer of low 240Pu/239Pu atom ratios were below the 1964 peak layer, the possibility of the influence of China’s atmospheric nuclear tests can be excluded. This further suggests that the low Pu ratios observed in the deep layers in Lake Qinghai originated from the early nuclear weapon research and development activities, e.g., trial tests and/or waste discharges made within Atomic City in the watershed. The Net 239+240Pu Inventories from Atomic City. Inventories of 239+240Pu and 137Cs (decay corrected to July 1998) in the sediments of Lake Qinghai are shown in Figure 4. 239+240Pu inventories are 68.4 ( 2.7, 31.9 ( 0.9, and 42.8 ( 0.9 MBq km
2 in 2006QH-1, 2006QH-2, and 2006QH-3, respectively, with an average of 47.7 ( 18.7 MBq km
2, which is slightly higher than the average value of global atmospheric fallout expected at the same latitude (42 MBq km
2).15 Inventories of 137Cs in 2006QH-2 and 2006QH-3 are 933.9 ( 81.1 and 1112.0 ( 78.0 MBq km
2, respectively, significantly lower than the average value of global atmospheric fallout at the same latitude (1923 MBq km
2, decay corrected to July 1998).42 This may be related to the geographical location of the lake in the area with dry weather, low annual precipitation, and strong prevailing winds, which resulted in the low atmospheric particle settlement.43 A low 137Cs inventory in sediments is also observed in Lake Sugan, where 239+240Pu inventory is lower than the average value of global atmospheric fallout at this latitude (Figure 2 in Supporting Information). The 239+240Pu/137Cs inventory ratios in the sediments of the four studied lakes are higher than the value of global atmospheric fallout (0.029 ( 0.003, decay corrected back to July 1998) (Figure 3 in Supporting Information),44 and the highest value comes from Lake Qinghai although the other three lakes are closer to Lop Nor. In 2006QH-2 and 2006QH-3, 239+240Pu/137Cs ratios are 0.034 and 0.038, respectively. The differential deposition of Pu and 137Cs, the so-called fractionation effect that fallout particles bearing mainly Pu were removed from the nuclear cloud at earlier times or at shorter distances than fallout particles bearing mainly 137Cs, has been reported by Simon et al.,45 and they found that 239+240Pu/137Cs ratio decreased with increasing distance to the nuclear weapons test site in the Marshall Islands. Thus, if the four lakes under discussion were all influenced by nuclear weapons test activities in Lop Nor, 239+240Pu/137Cs ratios in the far-away Lake Qinghai should be lower than those in the near-by Lakes Bosten, Sugan, and

ARTICLE

Figure 5. The relative contributions of 239+240Pu inventories from both global fallout and local input origins.

Shuangta. However, the present observations are just the opposite. This observation further supports above discussion that Lake Qinghai did not receive direct close-in fallout Pu from Lop Nor. The higher 239+240Pu/137Cs ratios in the lake can be attributed to the trace, but unique, Pu input from activities in Atomic City. The two-component mixing model 11,46 was used to estimate contributions of 239+240Pu input from Atomic City to the total Pu inventory, taking 0.18 as the 240Pu/239Pu ratio of global fallout origin and the minimum ratio of 0.038 in the sediments as the local input origin. The contributions are 5.1%, 15.7%, and 9.2% in 2006QH-1, 2006QH-2, and 2006QH-3, respectively, and the 239+240Pu inventories are 3.38, 5.02, and 3.92 MBq 3 km2, with an average of 4.11 ( 0.84 MBq 3 km2 (Figure 5). Based on the lake area, the total 239+240Pu inventory is approximately estimated to be 17.6 ( 3.6 GBq in the lake. During the period from 1964 to 1980, 22 atmospheric nuclear weapons tests were conducted at Lop Nor. Because little information on the Chinese nuclear weapons tests and related nuclear activities is available, the environmental impact, in particular, the possible radioactive contamination in northwestern China, has been a great concern. This study for the first time reveals that Lake Qinghai did not receive significant direct close-in fallout Pu from the Chinese atmospheric nuclear weapons tests at Lop Nor. The Pu input from the early nuclear activities in Atomic City is only 5
16% of the total Pu inventory, which is close to the expected Pu inventory from global fallout; therefore, the radiation effect on the local population would be expected to be negligible. Dating Sediments. Although the local Pu input recorded in the lake is very small, the unique and significantly low 240Pu/239Pu ratios in the deep sediment layers provide a new indicator for the dating of recent sediments, making it possible to estimate more accurately the sedimentation rate. The lowest 240Pu/239Pu ratios are 0.125 ( 0.018, 0.051 ( 0.019, and 0.038 ( 0.018 in the vertical profiles of 2006QH-1, 2006QH-2, and 2006QH-3, respectively (Table 2 in Supporting Information). These low ratios appear at mass depths of 2.52, 1.25, and 1.77 g cm
2 in 2006QH-1, 2006QH-2, and 2006QH-3, respectively (Figure 2). These depths were assumed to correspond to the initial period of nuclear weapon research and development activities in Atomic City in 1958, i.e., the date for the initial local-origin Pu input in Lake Qinghai, and from them the sedimentation rates are calculated as 0.143, 0.099, and 0.125 g cm
2 a
1 in 2006QH-1, 2006QH-2, and 2006QH-3 between 1958 and 1964. They are almost 9 times the 9192

dx.doi.org/10.1021/es202315c |Environ. Sci. Technol. 2011, 45, 9188–9194

Environmental Science & Technology

ARTICLE

Table 2. Sedimentation rates for Sediment Cores in Different Time-Scales, Together with the Contribution of Pu from Atomic City sediment cores 2006QH-1

time-scale

sedimentation

Pu from Atomic

rates g cm‑2 a‑1

City (MBq km‑2)

1964
2007

0.032

0.0

1958
1964

0.143

3.49

2006QH-2

1964
2007

0.011

0.0

2006QH-3

1958
1964 1964
2007

0.099 0.017

5.02 0.0

1958
1964

0.125

3.92

sedimentation rate from 1964 to the present (0.032, 0.011, and 0.017 g cm
2 a
1), suggesting a dramatic impact from unprecedented large-scale human activities in the watershed during the 1950
60s (Table 2). Therefore, the present results suggest that the sole use of 137Cs or 239+240Pu activity profiles for the estimation of recent sedimentation in Lake Qinghai is not reliable, where strong human activities have occurred in the recent decades. If the vertical profiles of 240Pu/239Pu ratios are taken into consideration, more detailed chronological evidence can be obtained for investigating historical nuclear activities in modern lake sediments in China, for reconstructing the pollution history of organic pollutants and heavy metals and for studying environmental changes. It is difficult to resolve the local source from the global fallout and evaluate the influence of local nuclear activities in areas close to nuclear research and/or test sites if only activities of radionuclides were measured. In the present work, at least two sources of plutonium could be identified in Lake Qinghai by using the vertical profiles of 239+240Pu and 137Cs activities, and 240Pu/239Pu isotopic ratios in sediment cores, combined with those in sediments of another three lakes (Lakes Bosten, Sugan, and Shuangta). For the first time, the existence of trace Pu contamination from the early nuclear weapons research and development activities in Atomic City was found in Lake Qinghai. However, from the calculation of Pu inventory of this local source, the radiation effect on the local population can be considered to be negligible. Because of its unique Pu isotopic composition, the trace Pu input recorded in sediments provides important chronological information for further studies on the water eutrophication process and climatic change, and reconstruction of pollution history of organic contaminants and heavy metals in the watershed of Lake Qinghai.

’ ASSOCIATED CONTENT

bS

Supporting Information. Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Tel: 0081-43-206-4634; fax: 0081-43-255-0721; e-mail: [email protected] (J.Z.); [email protected] (F.W.). Author Contributions ^

F. C. Wu and J. Zheng contributed equally to this study.

’ ACKNOWLEDGMENT This work was jointly supported by the exploratory research fund, National Institute of Radiological Sciences, Japan, Chinese National Basic Research Program (2008CB418200), Natural Science Foundation of China (40903037), and the Committee of National Defense Science & Technology of China (2008-124) and has been partly supported by the Agency for Natural Resources and Energy, the Ministry of Economy, Trade and Industry (METI), Japan. ’ REFERENCES

(1) Goldberg, E. D. Geochronology with 210Pb. In radioactive dating; Vienna: IAEA, 1963; pp 121
131. (2) Ketterer, M. E.; Hafer, K. M.; Jones, V. J.; Appleby, P. G. Rapid dating of recent sediments in Loch Ness: inductively coupled plasma mass spectrometric measurements of global fallout plutonium. Sci. Total Environ. 2004, 322, 221–229. (3) Zheng, J.; Wu, F. C.; Yamada, M.; Liao, H. Q.; Liu, C. Q.; Wan, G. J. Global fallout Pu recorded in lacustrine sediments in Lake Hongfeng, SW China. Environ. Pollut. 2008, 152, 314–321. (4) Krishnaswamy, S.; Lal, D.; Martin, J. M.; Meybeck, D. M. Geochronology of lake sediments. Earth Planet. Sci. Lett. 1971, 15, 407–414. (5) Zheng, J.; Liao, H. Q.; Wu, F. C.; Yamada, M.; Fu, P. Q.; Liu, C. Q.; Wan, G. J. Vertical distributions of 239+240Pu activity and 240 Pu/239Pu atom ratio in sediment core of Lake Chenghai, SW China. J. Radioanal. Nucl. Chem. 2008, 275, 37–42. (6) Wu, F. C.; Zheng, J.; Liao, H. Q.; Yamada, M. Vertical distributions of plutonium and 137Cs in lacustrine sediments in northwestern China: quantifying sediment accumulation rates and source identifications. Environ. Sci. Technol. 2010, 44, 2911–2917. (7) Pennington, W.; Cambray, R. S.; Fisher, E. M. Observation on lake sediments using fallout 137Cs as a tracer. Nature 1973, 242, 324–326. (8) Robbins, J. A.; Edgington, D. N. Determination of recent sedimentation rates in Lake Michigan using Pb-210 and Cs-137. Geochim. Cosmochim. Acta 1975, 39, 285–304. (9) Santschi, P. H.; Bollhalder, S.; Farrenkothen, K.; Lueck, A.; Zingg, S.; Sturm, M. Chernobyl radionuclides in the environment: tracers for the tight coupling between atmospheric, terrestrial, and aquatic geochemical processes. Environ. Sci. Technol. 1988, 22, 510–516. (10) Olivier, S.; Bajo, S.; Fifield, L. K.; Gaggeler, H.; Papina, T.; Santschi, P. H.; Schotterer, U.; Schwikowski, M.; Wacker, L. Plutonium from global fallout recorded in an ice core from the Belukha Glacier, Siberian Altai. Environ. Sci. Technol. 2004, 38, 6507–6512. (11) Hardy, E. P.; Krey, P. W.; Volchok, H. L. Plutonium fallout in Utah. In USAEC-Report HASL-257, 1972, pp 1-95. (12) MacKenzie, A. B.; Stewart, A.; Cook, G. T.; Mitchell, L.; Ellet, D. J.; Griffiths, C. R. Manmade and natural radionuclides in north east Atlantic shelf and slope sediments: Implications for rates of sedimentary processes and for contaminant dispersion. Sci. Total Environ. 2006, 369, 256–272. (13) Koide, M.; Bertine, K. K.; Chow, T. J.; Goldberg, E. D. The 240 Pu/239Pu ratio, a potential geochronometer. Earth Planet. Sci. Lett. 1985, 72, 1–8. (14) Ketterer, M.; Watson, B.; Matisoff, G.; Wilson, C. Rapid dating of recent aquatic sediments using Pu activities and 240Pu/239Pu as determined by quadrupole inductively coupled plasma mass spectrometry. Environ. Sci. Technol. 2002, 36, 1307–1311. (15) UNSCEAR. Ionizing Radiation: Sources and biological effects. In United Nations Scientific Committee on the Effects of Atomic Radiation. 1982 Report to the General Assembly, New York, 1982, pp 223
238. (16) Kelley, J. M.; Bond, L. A.; Beasley, T. M. Global distribution of Pu isotopes and 237Np. Sci. Total Environ. 1999, 237/238, 483–500. (17) Micholas, N. J.; Coop, K. L.; Estep, R. J. Capability and Limitation Study of DDT Passive-Active Neutron Waste Assay Instrument, Los Alamos National Laboratory, LA-12237-MS, 1992.

9193

dx.doi.org/10.1021/es202315c |Environ. Sci. Technol. 2011, 45, 9188–9194

Environmental Science & Technology (18) Yamamoto, M.; Tsumura, A.; Katayama, Y.; Tsukatani, T. Plutonium isotopic composition in soil from the former Semipalatinsk nuclear test site. Radiochim. Acta 1996, 72, 209–215. (19) Boulyga, S. F.; Becker, J. S. Isotope analysis of uranium and plutonium using ICP-MS and estimation of burn-up of spent uranium in contaminated environmental samples. J. Anal. At. Spectrom. 2002, 17, 1143–1147. (20) Ketterer, M. E.; Hafer, K. M.; Mietelski, J. W. Resolving Chernobyl vs. global fallout contributions in soil from Poland using plutonium atom ratios measured by inductively coupled plasma mass spectrometry. J. Environ. Radioact. 2004, 73, 183–201. (21) Warneke, T.; Croudace, I. W.; Warwick, P. E.; Taylor, R. N. A new ground-level fallout record of uranium and plutonium isotopes for northern temperate latitudes. Earth Planet. Sci. Lett. 2002, 203, 1047–1057. (22) Varga, Z. Origin and release date assessment of environmental plutonium by isotopic composition. Anal. Bioanal. Chem. 2007, 389, 725–732. (23) Muramatsu, Y.; Hamilton, T.; Uchida, S.; Tagami, K.; Yoshida, S.; Robinson, W. Measurement of 240Pu/239Pu isotopic ratios in soils from the Marshall Islands using ICP-MS. Sci. Total Environ. 2001, 278, 151–159. (24) Saito-Kokubo, Y.; Yasuda, K.; Magara, M.; Miyamoto, Y.; Sakurai, S.; Usuda, S.; Yamazaki, H.; Yoshikawa, S.; Nagaoka, S.; Mitamura, M.; Inoue, J.; Murakami, A. Depositional records of plutonium and 137Cs released from Nagasaki atomic bomb in sediment of Nishiyama reservoir at Nagasaki. J. Environ. Radioact. 2008, 99, 211–217. (25) Yamamoto, M.; Komura, K.; Sakanoue, M. Discrimination of the plutonium due to atomic explosion in 1945 from global fallout plutonium in Nagasaki soil. J. Radiat. Res. 1983, 24, 250–258. (26) Eriksson, M.; Lindahl, P.; Roos, P.; Dahlgaard, H.; Holm, E. U, Pu, and Am nuclear signatures of the Thule hydrogen bomb debris. Environ. Sci. Technol. 2008, 42, 4717–4722. (27) Yamamoto, M.; Hoshi, M.; Tanaka, J.; Sekerbaev, A. K.; Gusev, B. I. Plutonium isotopes and 137Cs in the surrounding areas of the former Soviet Union’s Semipalatinsk nuclear test site. J. Radioanal. Nucl. Chem. 1999, 242, 63–74. (28) Yamamoto, M.; Hoshi, M.; Tanaka, J.; Sakaguchi, A.; Apsalikov, K. N.; Gusev, B. I. Distributions of Pu isotopes and 137Cs in soil from Semipalatinsk nuclear test site detonations throughout southern districts. J. Radioanal. Nucl. Chem. 2004, 261, 19–36. (29) Norris, R. S.; Burrows, A. S.; Fieldhouse, R. W. Nuclear Weapons Databook, Vol. V: British, French and Chinese Nuclear Weapons; Westview Press: Boulder, 1994; pp 333
336. (30) Wright, D.; Gronlund, L. Estimating China’s production of plutonium for weapons. Sci. Global Secur. 2003, 11, 61–80. (31) Lanzhou Institute of Geology. A comprehensive investigation report of Lake Qinghai; Beijing Scientific Press: Beijing, 1979; pp 1
270 (in Chinese). (32) Li, J. G.; Philp, R. P.; Pu, F.; Allen, J. Long-chain alkenones in Lake Qinghai sediments. Geochim. Cosmochim. Acta 1996, 60, 235–241. (33) Wan, G. J.; Chen, J. A.; Wu, F. C.; Xu, S. Q.; Bai, Z. G.; Wan, E. Y.; Huang, R. G.; Yeager, K. M.; Santschi, P. H. Coupling between 210 Pbex and organic matter in sediments of a nutrient-enriched lake: an example from Lake Chenghai, China. Chem. Geol. 2005, 224, 223–236. (34) Zheng, J.; Yamada, M. Inductively coupled plasma sector field mass spectrometry with a high-efficiency sample introduction system for the determination of Pu isotopes in settling particles at femtogram levels. Talanta 2006, 69, 1246–1253. (35) Liao, H. Q.; Zheng, J.; Wu, F. C.; Yamada, M.; Tan, M. G.; Chen, J. M. Precise determination of plutonium isotopes in freshwater lake sediment by ICP-SFMS after separation using ion-exchange chromatography. Appl. Radiat. Isot. 2008, 66, 1138–1145. (36) Huh, C. A.; Chu, K. S.; Wei, C. L.; Liew, P. M. Lead-210 and plutonium fallout in Taiwan as recorded at a subalpine lake. J. Asian Earth Sci. 1996, 14, 373–376. (37) Shen, J.; Zhang, E. L.; Xia, W. L. Records from lake sediments of the Lake Qinghai to mirror climatic and environmental changes of the past about 1000 years. Quat. Sci. 2001, 21, 508–513 (in Chinese) (http://www.dsjyj.com.cn).

ARTICLE

(38) Xu, H.; Ai, L.; Tan, L. C.; An, Z. S. Geochronology of a surface core in the northern basin of Lake Qinghai: Evidence from 210Pb and 137 Cs radionuclides. Chin. J. Geochem. 2006, 25, 301–306. (39) Zeng, Y.; Zhang, X. B.; Zhou, W. J.; Qi, Y. Q. On the source of radioisotope 137Cs in the surface sediments of Lake Qinghai. J. Lake Sci. 2007, 19, 516–521 (in Chinese). (40) Taylor, R. N.; Warneke, T.; Andrew, M. J.; Croudace, I. W.; Warwick, P. E.; Nesbitt, R. W. Plutonium isotope ratio analysis at femtogram to nanogram levels by multicollector ICP-MS. J. Anal. At. Spectrom. 2001, 16, 279–284. (41) Wu, F. C.; Zheng, J.; Liao, H. Q.; Yamada, M. Distribution of artificial radionuclides in lacustrine sediments in China. Radiat. Prot. Dosim. 2011, 146, 291–294. (42) UNSCEAR. Sources and effects of ionizing radiation. In United Nations Scientific Committee on the Effects of Atomic Radiation. 1977 Report to the General Assembly, New York, 1977, pp 39
44. (43) Ren, T.; Zhang, S.; Li, Y.; Zhong, Z.; Su, Q.; Xu, C.; Tang, X. Methodology of retrospective investigation on external dose of the downwind area in Jiuquan region, China. Radiat. Prot. Dosim. 1998, 77, 25–28. (44) Hodge, V.; Smith, C.; Whiting, J. Radiocesium and plutonium still together in “background” soils after more than thirty years. Chemosphere 1996, 32, 2067–2075. (45) Simon, S. L.; Graham, J. C.; Borchert, A. W. Concentrations and spatial distribution of plutonium in the terrestrial environment of the Marshall Islands. Sci. Total Environ. 1999, 229, 21–39. (46) Krey, P. W.; Hardy, E. P.; Pachucki, C.; Rourke, F.; Coluzza, J.; Benson, W. K. Mass isotopic composition of global fallout plutonium in soil. Transuranium Nuclides in the Environment; International Atomic Energy Agency: Vienna, Austria, 1976; pp 671
678.

9194

dx.doi.org/10.1021/es202315c |Environ. Sci. Technol. 2011, 45, 9188–9194

ARTICLE pubs.acs.org/est

Standard Gibbs Free Energies of Reactions of Ozone with Free Radicals in Aqueous Solution: Quantum-Chemical Calculations Sergej Naumov*,† and Clemens von Sonntag*,‡,§ †

Leibniz-Institut f€ur Oberfl€achenmodifizierung (IOM), Permoserstrasse 15, D-04318 Leipzig, Germany Max-Planck-Institut f€ur Bioanorganische Chemie, Stiftstrasse 34
36, P.O. Box 101365, D-45470 M€ulheim an der Ruhr, Germany § Instrumentelle Analytische Chemie, Universit€at Duisburg-Essen, Universit€atsstrasse 5, 45117 Essen, Germany ‡

bS Supporting Information ABSTRACT: Free radicals are common intermediates in the chemistry of ozone in aqueous solution. Their reactions with ozone have been probed by calculating the standard Gibbs free energies of such reactions using density functional theory (Jaguar 7.6 program). O2 reacts fast and irreversibly only with simple carbon-centered radicals. In contrast, ozone also reacts irreversibly with conjugated carbon-centered radicals such as bisallylic (hydroxycylohexadienyl) radicals, with conjugated carbon/ oxygen-centered radicals such as phenoxyl radicals, and even with nitrogen- oxygen-, sulfur-, and halogen-centered radicals. In these reactions, further ozone-reactive radicals are generated. Chain reactions may destroy ozone without giving rise to products other than O2. This may be of importance when ozonation is used in pollution control, and reactions of free radicals with ozone have to be taken into account in modeling such processes.

’ INTRODUCTION There is a fast growing interest in ozone reactions in aqueous solution, as ozone is increasingly used in drinking water treatment and for pollution control in wastewater. For a better mechanistic understanding of ongoing primary processes, the fundamentals of ozone reactions have to be understood. Here, quantum-chemical calculations can be of considerable help.1
7 In ozone reactions, free radicals are often generated, and their reactions play an important role. Formation and reactions of • OH and O2•
are commonly taken into account, but peroxyl radicals, nitrogen- and sulfur-centered radicals, phenoxyl radicals, nitroxyl radicals, and even halogen-derived radicals are also formed in these reactions, and their ozone reactions must also be considered. It will be shown that they all react with ozone. This is in contrast to the most common free radical scavenger, O2. The ground state of O2 is a triplet state [3O2, O2(3∑g
)], and interaction with other molecules that are typically in their singlet ground states are spin forbidden and thus very slow. However, reactions with free radicals are spin allowed. Yet this reaction is restricted to a rather small number of radical types, and exceptions dominate (see below). The ground state of ozone, in contrast, is a singlet state, and this allows ozone to react fast with a large number of organic and inorganic compounds. It also reacts with many more free radicals than O2 does. Here, we explore the reason for this with the help of quantumchemical calculations based on density functional theory (DFT). For quantum-chemical calculations, ozone is an especially difficult molecule which has long been considered to be a demanding test case for quantum-chemical methods due to multireference problems,8,9 notably with DFT.10 The quality of calculations of ozone interaction with other molecules such as its reaction with r 2011 American Chemical Society

olefins, which proceeds by a 1,3-cycloaddition reaction,11,12 must thus be considered with some doubt. In contrast, this problem falls away in the calculations of the structures and standard Gibbs free energies of formation of ozone adducts to closed-shell molecules and free radicals, since here the multireference problem no longer prevails. For example, the reaction of ozone with benzene and its derivatives leads to a zwitterionic adduct as an intermediate (potential minimum), and there is an excellent correlation of the standard Gibbs free energy of ozone adduct formation with the logarithm of the ozone rate constant.5 The latter varies by 10 orders of magnitude between nitrobenzene and phenolate ion. Moreover, standard Gibbs free energy calculations may lead one to realize unexpected rearrangements of ozone adducts as has been the case with the ozone adducts to hypobromite4 and bisulfide ions.7 Also in a multifaceted system such as the reaction of ozone with nitrite, such calculations substantiate mechanistic suggestions.6 Moreover, DFT calculations allowed correction of the pKa of HO3• (reported at 6.15,13 revised to 8.2, B€uhler, private communication) to a much lower value (pKa ≈
2.0).14 DFT calculations also contributed in reassessing the mechanism of the decay of O3•
, a primary free radical intermediate in many ozone reactions, into •OH.2,14 Standard Gibbs free energies are related to the equilibrium constant K of a given reaction [ΔG° =
RT ln(K)]. Thus, endergonic reactions still take place at a substantial rate as long as the reverse reaction is fast. In the linear free energy relationship concept, use is made of the observation that the more exergonic Received: May 27, 2010 Accepted: September 12, 2011 Revised: September 12, 2011 Published: September 13, 2011 9195

dx.doi.org/10.1021/es2018658 | Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology

ARTICLE

Table 1. Reactions of Reducing Free Radicals with Ozone in Aqueous Solution: Compilation of Products, Rate Constants (M
1 s
1), and Calculated Standard Gibbs Free Energies (ΔG°, kJ mol
1) of the Overall Reaction and of Radical
Ozone Adducts (if Formed) and Their Decay educts eaq
, •

O3

H, O3

CO2•
, O3 CO2•
, O3 O2•
, O3 a

final products O3

•


•

OH, O2 •


reaction

rate constant

overall reaction

adduct formation a

1

3.6  10

(
374; pH > 11)

2/3

2  1010 23


411


373 b

10 23

see ref 2

4


315

CO3•
, O2

5


370

b

O3•
, O2

6


176

b

CO2, O3

1.5  109 23 1.6  109 13

adduct decay
38 [
47 14]

Calculated from reported reduction potentials.22 b An adduct cannot be calculated as a potential minimum on the reaction path.

the reaction the higher its rate. For some ozone reactions this concept works perfectly.5 Yet this holds only for a series of similar reactions. With markedly different reactions, such as addition vs electron transfer or H-abstraction, kinetics may overrun thermodynamics and the thermodynamically favored reaction may be barely observed. An example is given in the Supporting Information. With this caveat in mind, it is hoped that standard Gibbs free energies calculations, properly taken into perspective, will provide us with useful information on the reactions of free radicals with ozone.

’ THEORETICAL SECTION DFT calculations were carried out using the B3LYP hybride functional15,16 (Jaguar 7.6 program).17 The molecular geometries of all calculated molecules were optimized in water at the B3LYP/LACV3P**+ level of theory. The LACV3P**+ basis set uses the standard 6-311G**+ basis set for light elements and the LAC pseudopotential18 for third-row and heavier elements (in our case Br). For a given compound (intermediate), various structures were explored at a lower level of theory, and then the best one was calculated at the highest level of theory. The interactions between the molecule and the solvent were evaluated by Jaguar’s Poisson
Boltzmann solver (PBF).19 Frequency calculations were done at the same level of theory to characterize the stationary points on the potential surface and to obtain standard Gibbs free energies (G°), which were calculated at a standard temperature of 298.15 K and a pressure of 1 atm using unscaled frequencies. Reaction parameters were calculated as the difference of the standard Gibbs free energies ΔG° between reactants and products. It is difficult to estimate the errors that may be involved in our calculations. In absolute terms, thermochemical data for the reaction CO2•
+ O3 f CO3•
+ O2 (ΔG° =
335.8 kJ mol
1) and our calculated value (ΔG° =
370 kJ mol
1) differ by 34 kJ mol
1. Yet, in relative terms, this difference is only 10%, and this agreement is quite good. When the above reaction is compared with an electron transfer reaction, the difference between thermochemical and calculated values is only 5 kJ mol
1 (see the Supporting Information). The standard Gibbs free energies of the reaction HO3• f •OH + O2 obtained by Jaguar and G1 differ by 9 kJ mol
1 (cf. Table 1). G1 is the most reliable method for calculating standard Gibbs free energies of free radical reactions; that is, the values come closest to those of experiments.1 It is, however, restricted to very small and uncharged molecules/radicals and thus cannot be used here. With circumneutral reactions it seems fair to estimate an error of (10 kJ mol
1, and this is, of course, much more than 10% on relative terms.

’ RESULTS AND DISCUSSION There are a number of radicals whose ozone rate constants have been determined. These vary by 7 orders of magnitude and range from 103 M
1 s
1 to diffusion-controlled (k ≈ 1010 M
1 s
1) (cf. Tables 1
5). As mentioned in the Introduction, there is increasing evidence, including experimental evidence,20,21 for ozone adducts as the first intermediate in ozone reactions. Such adducts may also play a role in reactions of ozone with free radicals (R• + O3 f ROOO•). For some of these, a potential minimum is reached when the structures of such adducts are optimized and the standard Gibbs free energy of their formation can be calculated. Such adducts are regarded as intermediates rather than as transition states. In other cases, no potential minimum is reached, and during energy optimization the reaction proceeds to the products. “Adducts” are then only bona fide transition states. In the following, the ozone reactions of the various groups of radicals are discussed. Reactions of Ozone with Reducing Radicals. The oneelectron reduction potential of ozone is +1010 mV (at pH > 11),22 and electron transfer reactions with reducing radicals such as eaq
(
2870 mV), •H (
2400 mV), CO2•
(
1900 mV), and O2•
(
330 mV) are feasible. Rate constants and standard Gibbs free energies for the formation of adducts, if formed, and products are compiled in Table 1. The hydrated electron (eaq
) adds to ozone, reaction 1, and subsequent decomposition of O3•
into O•
plus O2 is measurably slow (for the rate and equilibrium constants see ref 24). O3 þ eaq
f O3 •


ð1Þ

For the •H atom, an adduct (reaction 2), subsequently decaying into •OH and O2 (reaction 3), is formed (for equilibrium 3 see ref 14). In competition with reaction 3, HO3• may deprotonate since HO3• is a very strong acid (pKa(HO3•) =
2.0).14 O3 þ • H f HOOO• a • OH þ O2

ð2=3Þ CO2•
,

In the calculations of the reaction of ozone with an adduct is not observed, and two competing reactions may occur, an O-transfer (reaction 4) and an electron transfer (reaction 5). Electron transfer, although also markedly exergonic, is not energetically favored, the difference from reaction 4 being ΔG° = +55 kJ mol
1 (cf. Table 1). This is in good agreement with thermokinetic data that yield ΔG° = +50 kJ mol
1 (see the Supporting Information). CO2 •
þ O3 f CO3 •
þ O2

ð4Þ

CO2 •
þ O3 f CO2 þ O3 •


ð5Þ

It is re-emphasized that kinetics may determine the route that is taken and the thermodynamically favored product may be 9196

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology

ARTICLE

Table 2. Reactions of Carbon-Centered Free Radicals and Phenoxyl Radicals with Ozone in Aqueous Solution: Compilation of Products, Rate Constants (M
1 s
1), and Calculated Standard Gibbs Free Energies (ΔG°, kJ mol
1) of the Overall Reaction and of Radical
Ozone Adducts (if Formed) and Their Decay

a

An adduct cannot be calculated as a potential minimum on the reaction path.

disfavored when both reactions are of a different reaction type and both are thermodynamically feasible. Examples for free radical reactions have been discussed in ref 25 and for an ozone reaction in ref 1. A further example is given in the Supporting Information. Thus, additional information is often required for assessing the preferred route in competing exergonic reactions. In the reaction of CO2•
with ozone, no chain reaction of any importance is observed, and it has been concluded that the electron transfer reaction 5 can be neglected.26 The reduction potential of O2•
is the least negative of these four reducing radicals, and the rate constant is an order of magnitude slower than those of eaq
and H•. No adduct is observed; the standard Gibbs free energy for the overall reaction 6 is lower than that for the reaction with CO2•
(Table 1), but as there is no competing reaction, electron transfer is the only process. O2 •
þ O3 f O2 þ O3 •


very fast, 2  109 M
1 s
1 (Table 2), if the acetate radical, CH2C(O)O
, is a good guide.27 Calculations have been carried out for the methyl radical and the acetate radical. For the methyl radical an adduct with a CH3O
OO• bond length of 1.58 Å can be calculated, reaction 7, that subsequently decays into methoxyl and O2 (reaction 8).

•

•

ð7=8Þ

With the acetate radical, an adduct as a potential minimum along the reaction path could not be established. The reaction of the acetate radical is very fast (Table 2), and other alkyl radicals may react similarly fast. The high driving force for reaction 7 is compatible with this. In water, the methoxyl radical undergoes a rapid (water-catalyzed) 1,2-H shift (reaction 9, k ≈ 106 s
1, ΔG° =
34 kJ mol
1).28,29 CH3 O• f • CH2 OH

ð6Þ

Reactions of Ozone with Carbon-Centered Radicals. The reaction of ozone with simple carbon-centered radicals must be

CH3 þ O3 f CH3 OOO• f CH3 O• þ O2

ð9Þ

Moreover, strongly branched alkoxyl radicals undergo a similarly fast β-fragmentation.28 Thus, alkoxyl radicals are too shortlived in water for reactions with ozone to occur, and their reactions with ozone (see below) are not of importance in practice. 9197

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology

ARTICLE

An interesting group of radicals are the hydroxycylohexadienyl radicals that are formed upon •OH attack on benzene and its derivatives (note that in the ozonation of a wastewater 13% of ozone is converted into •OH 30 that reacts predominantly with the aromatic moieties of the organic matter). In contrast to O2 that reacts reversibly with these radicals (see below), ozone must react irreversibly (reaction 10) without a detectable adduct as an intermediate. In this respect, the hydroxycyclohexadienyl radicals are similar to the acetate radical but differ from the methyl radical, where an, albeit very labile, adduct has been calculated (cf. reactions 7/8).

Phenoxyl radicals are commonly written with the free spin at oxygen, but the overwhelming spin density is at carbon at the ortho and para positions of the benzene ring (see Figure S1, Supporting Information). Recombination of two phenoxyl radicals at oxygen is strongly endergonic, but recombination may occur by forming C
O and C
C linkages31
33 (see Figure S2, Supporting Information). Thus, phenoxyl radicals largely behave as carbon-centered radicals. Although ozone addition at oxygen and concomitant O2 loss is exergonic (reaction 11), an addition at carbon (cf. reaction 12) is even more so (Table 2).

Ozone is an electrophilic agent,34 and addition at carbon seems to be more likely. Reaction 12 may hence be preferred over reaction 11. Further O2 loss from the phenylperoxyl radical is markedly endergonic (Table 2), and such peroxyl radicals are indeed well documented.35,36 No adducts have been detected as intermediates in reactions 11 and 12. The high reactivity of phenoxyl radicals with ozone is in contrast to their reaction with O2 (see below), and as phenoxyl radicals are important intermediates in the reactions of ozone with phenols,37,38 their reaction with ozone may be of some relevance. Reactions of Ozone with Oxygen-Centered Radicals. The reaction of ozone with O2•
has been discussed above. The reaction with the •OH radical is fast,39 but below diffusion controlled (Table 3). The reaction has been formulated as an O-transfer (reaction 13). An adduct could not be calculated. •

OH þ O3 f HO2 • þ O2

ð13Þ

In water, alkoxyl radicals are too short-lived (see above) for reacting with ozone. Yet reaction 14 is of some interest in comparison with the analogous reaction of •OH (reaction 13). A somewhat lower standard Gibbs free energy has been calculated (Table 3). Again, no adduct is detectable. CH3 O• þ O3 f CH3 OO• þ O2

With the methylperoxyl radical, a short-lived adduct can be calculated (reaction 15) that subsequently releases O2 (reaction 16); cf. Table 3. CH3 OO• þ O3 a CH3 OOOOO• f CH3 O• þ 2O2

With all the other peroxyl radicals, adducts could not be established, and the reactions run through to final products. Nitroxyl radicals also belong to the group of oxygen-centered radicals. The stable nitroxyl radical TEMPO that is often used as an ESR standard reacts with ozone with an observed rate constant kobsd = 1  107 M
1 s
1.40 In the first step, one molecule of O2 is released (reaction 17) without a detectable adduct as an intermediate. The nitrogen-centered peroxyl radical thus formed is unstable and loses O2 (reaction 18). For standard Gibbs free energies see Table 3.

In its reaction with ozone, the morpholine-derived nitroxyl radical shows the same overall standard Gibbs free energy for the release of two molecules of O2 (Table 3). In contrast to the TEMPO system, an intermediate peroxyl radical cannot be calculated here and must be a transition state rather than a very short-lived intermediate. Reactions of Ozone with Nitrogen- and Sulfur-Centered Radicals. Nitrogen-centered radicals are likely intermediates in the reactions of ozone with primary and secondary amines, and their reactions may contribute to product formation in these systems. An example is shown for the reaction of ozone with morpholine (reactions 19 and 20), for which an •OH yield near 33% has been found (A. Tekle-R€ottering, private communication). The precursor of •OH is O3•
(for its decay into •OH, see above), and material balance considerations require that an equivalent amount of aminyl radicals must be formed as well.

Here, we have carried out calculations on the nitrogen-centered radical formed in the reaction of ozone with TEMPO (cf. reaction 18) and in the morpholine system (reactions 19/20). The reaction of ozone with these radicals (reaction 21) is highly exergonic (Table 4) and proceeds without an adduct that is sufficiently stable to be calculated as an intermediate. R 2 N• þ O3 R 2 NO• þ O2

ð14Þ

A number of peroxyl radicals have been generated radiolytically, and their reactions with ozone have been studied.26 From an evaluation of the data, rate constants have been calculated that are compiled in Table 3. The rate constants of the carboncentered peroxyl radicals with ozone center near 104 M
1 s
1, and that of
O3SOO• is an order of magnitude higher (Table 3).

ð15=16Þ

ð21Þ

Rate constants for the reaction of aminyl radicals with ozone are as yet not known, but the exergonicity of this reaction is remarkable (Table 4), and this reaction may be quite fast. Combined with reactions such as reactions 17/18, it would give rise to a chain reaction that consumes ozone without giving rise to products other than O2. Such a cycle has been suggested to play a 9198

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology

ARTICLE

Table 3. Reactions of Oxygen-Centered Free Radicals with Ozone in Aqueous Solution: Compilation of Products, Rate Constants (M
1 s
1), and Calculated Standard Gibbs Free Energies (ΔG0, kJ mol
1) of the Overall Reaction and of Radical
Ozone Adducts (if Formed) and Their Decay

a

An adduct cannot be calculated as a potential minimum on the reaction path.

role in the ozone chemistry of diclofenac for explaining why the diclofenac consumption by ozone is so surprisingly low.41 Yet kinetic data that would substantiate this suggestion are still missing. Aminyl radicals may rearrange into the corresponding carboncentered radicals, e.g., reaction 22 (ΔG° =
31 kJ mol
1). CH3
CH2
• NH f CH3
• CH
NH2

ð22Þ

Such reactions may be slower than the corresponding 1,2-H shift reaction of alkoxyl radicals (ΔG°(ethoxylfhydroxyethyl) =
36 kJ mol
1), as the driving force is lower by 5 kJ mol
1. Such rearrangements may shorten the above chain reaction.

Thiols react very fast with ozone without giving rise to thiyl radicals (an electron transfer would be endergonic). Yet the thiyl radical/ozone reaction is mechanistically of interest in comparison with the corresponding reaction of O2 discussed below. Reaction 23, without an adduct as a potential minimum on the reaction path, is highly exergonic (Table 4). CH3 S• þ O3 f CH3 SO• þ O2

ð23Þ

Reactions of Ozone with Halogen-Containing Radicals. The reactions of ozone with halide ions are complex (see the Supporting Information) and do not directly give rise to free radicals. Only when the oxidation proceeds to chlorite42 and 9199

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology

ARTICLE

Table 4. Reactions of Nitrogen- and Sulfur-Centered Free Radicals with Ozone in Aqueous Solution: Compilation of Products and Calculated Standard Gibbs Free Energies (ΔG0, kJ mol
1) of the Overall Reaction

bromite, 43 ozone-induced free radical formation sets in (cf. reaction 24). ClO2
þ O3 f ClO2 • þ O3 •


ð24Þ

Yet the reaction of •OH with Br
that gives rise to Br• is very high (reaction 25, k = 1.1  1010 M
1 s
1 44), and the •OH route in ozone reactions can contribute to bromate formation in drinking water ozonation.45 •

OH þ Br
f Br• þ OH


ð25Þ


•

In neutral solutions, the reaction of Cl with OH gives rise to a three-electron-bonded intermediate that only in acid solution is converted into Cl• (equilibrium 26, K = 0.7 M
1 46 and reaction 27). •

OH þ Cl
a HOCl•


HOCl•
þ Hþ a H2 O þ Cl•

ð26Þ

The Cl2•
reaction resembles the reaction of •OH with ozone. In both cases, rate constants are well below the diffusioncontrolled limit (Table 5) despite the high exergonicity of these reactions. The reasons for this are not yet fully understood, but the pronounced electrophilicity of ozone may play a role in preventing a favorable transition state in reactions of nitrogen-, oxygen- and halogen-centered radicals (note that the reactions of TEMPO and peroxyl radicals with ozone are also below diffusion controlled). For the subsequent oxidation of the ClO• radical by ozone, two exergonic reactions may be written. Reaction 32 gives rise to chlorine dioxide, OClO•, but a reaction to its isomer, the chlorineperoxyl radical, ClOO•, is even more exergonic (reaction 33, Table 5). Missing additional information, a decision as to which of them is kinetically favored cannot be made.

ð27Þ

In the presence of Cl
, equilibrium 28 (K = 6  104 M
1,47 1.4  10 M
1 48) has to be taken into account. 5

Cl• þ Cl
a Cl2 •


Br• þ Br
a Br2 •


ð29Þ

The potential importance of halogen-derived radicals under ozonation conditions justifies having a closer look at their reactions with ozone. For rate constants and calculated standard Gibbs free energies, see Table 5. The •Cl atom undergoes an O-transfer without an adduct as a potential minimum on the reaction path (reaction 30). A rate constant has not been measured. As expected, the reaction becomes less exergonic when Cl• is complexed to Cl
(reaction 31, Table 5). Cl• þ O3 f ClO• þ O2

ð30Þ

Cl2 •
þ O3 f ClO• þ O2 þ Cl


ð31Þ

ð32Þ

ClO• þ O3 f ClOO• þ O2

ð33Þ

The peroxyl radical ClOO• is highly unstable (equilibrium 34).

ð28Þ

A similar equilibrium can be written for Br• plus Br
(equilibrium 29, K = 3.9  105 49), and there must also be an analogous mixed three-electron-bonded intermediate, BrCl•
.

ClO• þ O3 f OClO• þ O2

ClOO• a Cl• þ O2

ð34Þ

The species that has been obtained with a potential minimum in the calculations has a Cl
OO• bond length of 2.331 Å and ClO
O• bond length of 1.888 Å. The OdO bond length is practically identical, 1.886 Å. This indicates that the ClOO• species is a kind of van der Waals complex of Cl• with O2. The value of ΔG° = +5 kJ mol
1 can thus most likely not be related to the equilibrium constant of reaction 34. This is in contrast to the analogous equilibrium HOOO• a •OH + O2 (ΔG° =
47 kJ mol
1).14 Here, HO
OO• (1.520 Å) and HOO
O• (1.243 Å) bond lengths are quite different from those of the final products, and the HOOO• species thus characterized is not a mere van der Waals complex. The oxidation of chlorine dioxide by ozone (reaction 35) is only moderately fast (Table 5), and this is compatible with the reaction being slightly endergonic. OClO• þ O3 f ClO3 • þ O2 9200

ð35Þ

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology

ARTICLE

Table 5. Reactions of Halogen-Containing Free Radicals with Ozone in Aqueous Solution: Compilation of Products, Rate Constants (M
1 s
1), and Calculated Standard Gibbs Free Energies (ΔG0, kJ mol
1) of the Overall Reaction and of Radical
Ozone Adducts (if Formed) and Their Decay educts •

•

reaction

rate constant

overall reaction

adduct formation


204

a


133

a

Cl, O3

ClO , O2

30

Cl2•
, O3

ClO•, O2, Cl


31

ClO•, O3

OClO•, O2

32


119

a

•

ClO , O3

ClOO•, O2

33/34


229

+5; see the text

OClO•, O3

ClO3•, O2

35

OClO•, O3

ClO•, 2 O2

2 ClO3

•

(9 ( 0.7)  107 50

(1.05 ( 0.1)  103 42 1.1  103 51

+9

a

36


311

a

adduct decay

O3Cl
ClO3

37

+109

2 ClO3•

ClO4
, ClO2+

38/39


40

OClO• + ClO3•

O3Cl
ClO2

41

+93

OClO• + ClO3•

O2Cl
OClO2

42

OClO• + ClO3•

ClO• + O2 + OClO•

43/44

ClO3• + ClO
ClO3• + ClO2


ClO3
+ ClO• ClO3
+ OClO•

Br•, O3

BrO•, O2

45


149

a

BrO , O3

BrOO•, O2

46


268

a

BrOO•

Br• + O2

47


11

a

BrO•, O3

OBrO•, O2

48


47

a

2 OBrO•

OBrOOBrO

49

+11

2 OBrO•

O2BrOBrO

50

+130

OBrOOBrO OBrO•, O3

2 •Br, 2 O2 BrO• + 2 O2

51 52


282
380

OBrO•, O3

O2BrO•, O2

53/54

+81

+102


21

O2BrO , O3

BrO4•, O2

55/56

+93

+114


21

BrO3•

BrO•, O2

57


506

BrO4•

OBrO•, O2

58


413

•

•

a

final products

a

+29
222

a


144
162

An adduct cannot be calculated as a potential minimum on the reaction path.

An ozone addition to one of the oxygens may give rise via a tetroxide as a transition state to OCl• plus 2O2 (reaction 36). OClO• þ O3 f OCl• þ 2O2

ð36Þ

The overall reaction is strongly exergonic (Table 5). Yet reaction 35 may compete successfully in case there is substantial activation energy for reaction 36. Formation of 1O2 in uncommonly high yields (150%) has been reported for CCl4 solutions.52 The energetics of some of the reactions discussed above would also allow the formation of 1O2, but yields above 100% are difficult to reconcile. The ClO3• radical cannot be further oxidized by ozone and has to decay bimolecularly. There are two potential combination reactions, 37 (Cl
Cl = 2.5 Å) and 38 followed by 39. The high endergonicity of reaction 37 (Table 5) prevents this reaction from taking place, and the bimolecular decay of ClO3• radical must occur via an asymmetrical recombination, reaction 38. Upon trying to calculate the standard Gibbs free energy of this reaction, the program proceeded automatically toward perchlorate ion and ClO2+ (reaction 39, Table 5). ClO2+ reacts with water, yielding chlorate (reaction 40). 2ClO3 • f O3 Cl
ClO3 2ClO3 • f ½O3 ClOClO2  f ClO4
þ ClO2 þ ClO2 þ þ H2 O f ClO3
þ 2Hþ

Perchlorate formation has been intriguing the scientific community for a long time, and this question has been recently addressed again.53 ClO3• has been envisaged as a precursor, but details have remained obscure. At this point it is important to note that perchlorate is not an important product in the reaction of ClO
/ClO2
with ozone, and perchlorate formation53 must be a minor side reaction (see the Supporting Information). Since the proposed O-transfer (ClO2
+ O3 f ClO3
+ O2) has been ruled out in favor of an electron transfer (ClO2
+ O3 f OClO• + O3•
),42 a route from OClO• to chlorate has to be found. In competition with a bimolecular decay of ClO3•, one may consider a termination with OClO• radicals in case there is a sufficiently high steady-state concentration of OClO• radicals present in equilibrium (note the relatively slow reaction of OClO• with ozone, Table 5). This reaction could eventually give rise to two molecules of chlorate ion in case the dimers were sufficiently long-lived for reacting with water. Yet reaction 41 is markedly endergonic with a long Cl
Cl bond (1.980 Å) (Table 5); that is, reaction 41 is most likely reversible. Reaction 42 is less endergonic but potentially also reversible.

ð37Þ ð38=39Þ ð40Þ

ClO3 • þ OClO• a O3 Cl
ClO2

ð41Þ

ClO3 • þ OClO• a O3 Cl
OClO

ð42Þ

If the (exergonic) hydrolysis of these dimers (Cl2O5 + H2O f 2 ClO3
+ 2 H+) were fast compared to the reverse reactions, 9201

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology

ARTICLE


41 and
42 (note the ready hydrolysis of Br2O4), preference of chlorate over perchlorate formation could be rationalized. Moreover, electron transfer from hyopchlorite and chlorite to ClO3• that would also give rise to chlorate is highly exergonic (Table 5). To which extent these reactions contribute to chlorate formation has still to be explored. There is also the potential and highly exergonic termination via the oxygens that proceeds directly, with the peroxidic arrangement only as a likely transition state (reaction 43), to OCl• and OClO• plus O2 (reaction 44). ClO3 • þ OClO• ½O3 ClO
OClOTS f OCl• þ O2 þ OClO•

reaction sequence 53/54 is markedly endergonic (Table 5). OBrO• þ O3 f BrO• þ 2O2 OBrO• þ O3 f O2 BrOOO• f O2 BrO• þ O2 •

O2 BrO• þ O3 f O3 BrOOO• f O3 BrO• þ O2 •

O2 BrO• f BrO• þ O2

The reaction of the Br atom with ozone is strongly exergonic (reaction 45, Table 5). For the ensuing BrO • one can envisage reactions 46/47 and 48. BrO• þ O3 f BrOO• þ O2

ð46Þ

BrOO• a Br• þ O2

ð47Þ

BrO• þ O3 f OBrO• þ O2

ð48Þ

All three reactions are exergonic (Table 5), and a preference will be determined by kinetics, that is, whether ozone addition at oxygen or at bromine is preferred. In the ozonation of Br
-containing waters, there is a second route to OBrO•. There is now strong evidence that the current concept of bromate formation45 has to be revised. The reaction of bromite with ozone does not give rise to bromate by O-transfer but is an electron transfer reaction that yields OBrO• and O3•
43 (for details see the Supporting Information). Formation and decay of OBrO• has been studied by pulse radiolysis,54 and it has been concluded that the bimolecular decay leading to a Br2O4 intermediate (i) is reversible, (ii) decays by reacting with water, and (iii) decays by reacting with OH
. There are two conceivable dimers (reactions 49 and 50). 2OBrO• a OBrOOBrO

ð49Þ

2OBrO• a O2 BrOBrO

ð50Þ

While the symmetrical dimerization is mildly endergonic (Table 5) and could account for the reported reversibility, the asymmetrical dimerization is strongly endergonic and is unlikely to occur. There is, however also the possibility that the symmetrical dimer decays according to reaction 51. •

OBrOOBrO f 2Br þ 2O2

ð51Þ

This is the route taken upon prolonged optimization. Yet in water, where a OH
-induced component of the decay of OBrO• radicals is observed, this reaction seems not to be kinetically favored despite the high exergonicity (Table 5). The bimolecular decay of OBrO• to Br2O4 is reversible, but Br2O4 reacts fast with water and OH
, and it is unlikely that sufficiently high OBrO• concentrations build up for reactions of OBrO• with ozone to become relevant. Yet in case it would, the energetically favored (Table 5) route is reaction 52, while

ð55=56Þ

•

If O2BrO and O3BrO were formed, they would undergo the strongly exergonic 3O2-releasing reactions 57 and 58 (Table 5).

•

ð45Þ

ð53=54Þ

•

An oxidation of O2BrO to O3BrO (reactions 55 and 56) is endergonic (Table 5).

ð43=44Þ

Br• þ O3 f BrO• þ O2

ð52Þ

ð57Þ

O3 BrO• f OBrO• þ 2O2

ð58Þ •

It follows from the above that Br is not oxidized by ozone beyond OBrO•. Reactions of Ozone and Oxygen with Free Radicals: A Comparison. Oxygen is commonly regarded as a good scavenger of free radicals. This is only correct insofar as most free radicals under study are carbon-centered radicals, and any free radical chemistry involving alkyl radicals (k(R3C• + O2) ≈ 2  109 M
1 s
1) turns into a peroxyl radical chemistry in the presence of even low O2 concentrations.29 However, there are exceptions. Bisallylic radicals such as the hydroxycyclohexadienyl radicals formed upon •OH attack on benzene and its derivatives react already reversibly with O2,36,55
57 and the phenoxyl radicals do not react with O2 (k < 103 M
1 s
1 33,58), unless activated by electrondonating substituents,59 despite the fact that there is a high spin density at carbon (see the Supporting Information). Under ozonation conditions, the O2 concentration typically exceeds the ozone concentration. As reactions of O2 and ozone with alkyl radicals are equally fast, competition is in favor of a reaction with O2. Corrections for a reaction of alkyl radicals with ozone are minor and rarely exceed 10%. The reaction of O2 with the hydroxycyclohexadienyl radicals is reversible.36,55
57 This may give them a kinetic advantage in their reaction with ozone. To what extent this is of consequence cannot be predicted, as their rate constant with ozone is not yet known. Oxygen-centered radicals such as •OH, O2•
, ROO•, and R2NO• do not react with O2, but they react readily with ozone (see above). This also holds for nitrogen-centered radicals that react with O2 very slowly or not at all.29 This is now confirmed by our calculations (reaction
18 is endergonic). Sulfur-centered radicals such as thiyl radicals react reversibly with O2.36,60 In agreement with this, a standard Gibbs free energy of +15 kJ mol
1 has now been calculated for reaction 59. CH3 S• þ O2 a CH3 SOO•

ð59Þ

The 1,2-H shift reaction into a carbon-centered radicals observed with oxygen- and nitrogen-centered radicals, cf. reactions 9 and 22, is very slow here, as the reaction is endergonic (cf. reaction 60, ΔG° = +26 kJ mol
1). CH3
CH2
S• f CH3
• CH
SH

ð60Þ

The reaction of thiyl radicals with ozone, however, is irreversible (Table 4), and scavenging of thiyl radicals by ozone would be much more efficient than scavenging by O2. The much higher efficiency of ozone in scavenging free radicals as compared to O2 can lead to a much higher ozone demand in oxidation reactions by ozone as is currently believed. 9202

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology Advanced modeling of ozone reactions will have to take such reactions into account. Although there are already a number of rate constants of ozone with free radicals known (cf. Tables 1
5), a larger set is certainly required for reliable simulations. Here, the most important ones seem to be the nitrogen-centered radicals.

’ ASSOCIATED CONTENT

bS

Supporting Information. Standard Gibbs free energies for the reactions of •OH with phenol and the phenolate ion, spin distribution in the phenoxyl radicals and the standard Gibbs free energies for their recombination, reactions of ozone with chloride and bromide ions and subsequent reactions, and thermochemical calculations. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected] (C.v.S.); sergej.naumov@ iom-leipzig.de (S.N.).

’ REFERENCES (1) Merenyi, G.; Lind, J.; Naumov, S.; von Sonntag, C. The reaction of ozone with the hydroxide ion. Mechanistic considerations based on thermokinetic and quantum-chemical calculations. The role of HO4
in superoxide dismutation. Chem.—Eur. J. 2010, 16, 1372–1377. (2) Merenyi, G.; Lind, J.; Naumov, S.; von Sonntag, C. The reaction of ozone with hydrogen peroxide (peroxone process). A revision of current mechanistic concept based on thermokinetic and quantummechanical considerations. Environ. Sci. Technol. 2010, 44, 3505–3507. (3) Mvula, E.; Naumov, S.; von Sonntag, C. Ozonolysis of lignin models in aqueous solution: Anisole, 1,2-dimethoxybenzene, 1,4-dimethoxybenzene and 1,3,5-trimethoxybenzene. Environ. Sci. Technol. 2009, 43, 6275–6282. (4) Naumov, S.; von Sonntag, C. The reactions of bromide with ozone towards bromate and the hypobromite puzzle: A density functional theory study. Ozone: Sci. Eng. 2008, 30, 339–343. (5) Naumov, S.; von Sonntag, C. Quantum chemical studies on the formation of ozone adducts to aromatic compounds in aqueous solution. Ozone: Sci. Eng. 2010, 32, 61–65. (6) Naumov, S.; Mark, G.; Jarocki, A.; von Sonntag, C. The reactions of nitrite ion with ozone in aqueous solution—New experimental data and quantum-chemical considerations. Ozone: Sci. Eng. 2010, 32, 430–434. (7) Mark, G.; Naumov, S.; von Sonntag, C. The reaction of ozone with bisulfde (HS
) in aqueous solution—Mechanistic aspects. Ozone: Sci. Eng. 2011, 33, 37–41. (8) Raghavachari, K.; Trucks, G. W.; Pople, J. A.; Replogle, E. S. Higly correlated systems: Structure, binding energy and harmonic vibrational frequencies of ozone. Chem. Phys. Lett. 1989, 158, 207–212. (9) Watts, J. D.; Bartlett, R. J. Coupled-cluster calculations of structure and vibrational frequencies of ozone: Are triple excitations enough?. J. Chem. Phys. 1998, 108, 2511–2514. (10) Foresman, J. B.; Frisch, A. Exploring Chemistry with Electronic Structure Methods; Gaussian, Inc.: Pittsburgh, PA, 1995. (11) Anglada, J. M.; Crehuet, R.; Bofill, J. M. The ozonolysis of ethylene: A theoretical study of the gas-phase reaction mechanism. Chem.—Eur. J. 1999, 5, 1809–1822. (12) Olzmann, M.; Kraka, E.; Cremer, D.; Gutbrod, R.; Andersson, S. Energetics, kinetics, and product distributions of the reactions of ozone with ethene and 2,3-dimethyl-2-butene. J. Phys. Chem. A 1997, 101, 9421–9429. (13) B€uhler, R. E.; Staehelin, J.; Hoigne, J. Ozone decomposition in water studied by pulse radiolysis. 1. HO2/O2
and HO3/O3
as intermediates. J. Phys. Chem. 1984, 88, 2560–2564.

ARTICLE

(14) Naumov, S.; von Sonntag, C. The reaction of •OH with O2, the decay of O3•
and the pKa of HO3•
interrrelated questions in aqueous free-radical chemistry. J. Phys. Org. Chem. 2011, 24, 600–602. (15) Becke, A. D. Density-functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 1993, 98, 5648–5652. (16) Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 1988, 37, 785–789. (17) Jaguar 7.6; Schrodinger LLC: New York, 2009. (18) Wadt, W. R.; Hay, P. J. Ab initio effective core potentials for molecular calculations—Potentials for main group elements Na to Bi. J. Chem. Phys. 1985, 82, 284–298. (19) Tannor, D. J. M.; Murphy, B.; Friesner, R.; Sitkoff, R. A.; Nicholls, D.; Ringnalda, A.; Goddard, W. A., III; Honig, B. Accurate first principles calculation of molecular charge distributions and solvation energies from ab initio quantum mechanics and continuum dielectric theory. J. Am. Chem. Soc. 1994, 116, 11875–11882. (20) Mu~ noz, F.; Mvula, E.; Braslavsky, S. E.; von Sonntag, C. Singlet dioxygen formation in ozone reactions in aqueous solution. J. Chem. Soc., Perkin Trans. 2 2001, 1109–1116. (21) Liu, Q.; Schurter, L. M.; Muller, C. E.; Aloisio, S.; Francisco, J. S.; Margerum, D. W. Kinetics and mechanisms of aqueous ozone reactions with bromide, sulfite, hydrogen sulfite, iodide, and nitrite ions. Inorg. Chem. 2001, 40, 4436–4442. (22) Wardman, P. Reduction potentials of one-electron couples involving free radicals in aqueous solution. J. Phys. Chem. Ref. Data 1989, 18, 1637–1755. (23) Sehested, K.; Holcman, J.; Hart, E. J. Rate constants and products of the reactions of e
, O2
and H with ozone in aqueous solutions. J. Phys. Chem. 1983, 87, 1951–1954. (24) Elliot, A. J.; McCracken, D. R. Effect of temperature on O•
reactions and equilibria: A pulse radiolysis study. Radiat. Phys. Chem. 1989, 33, 69–74. (25) Naumov, S.; von Sonntag, C. The energetics of rearrangement and water elimination reactions in the radiolysis of the DNA bases in aqueous solution (eaq
and •OH attack). DFT calculations. Radiat. Res. 2008, 169, 355–363. (26) Lind, J.; Merenyi, G.; Johansson, E.; Brinck, T. The reaction of peroxyl radicals with ozone in water. J. Phys. Chem. A 2003, 107, 676–681. (27) Sehested, K.; Holcman, J.; Bjergbakke, E.; Hart, E. J. Ozone decomposition in aqueous acetate solutions. J. Phys. Chem. 1987, 91, 2359–2361. (28) von Sonntag, C. Free-Radical-Induced DNA Damage and Its Repair. A Chemical Perspective; Springer Verlag: Berlin, Heidelberg, 2006. (29) von Sonntag, C.; Schuchmann, H.-P. Peroxyl radicals in aqueous solution. In Peroxyl Radicals; Alfassi, Z. B., Ed.; Wiley: Chichester, U.K., 1997; pp 173
234. (30) N€othe, T.; Fahlenkamp, H.; von Sonntag, C. Ozonation of wastewater: Rate of ozone consumption and hydroxyl radical yield. Environ. Sci. Technol. 2009, 43, 5590–5595. (31) Benn, R.; Dreeskamp, H.; Schuchmann, H.-P.; von Sonntag, C. Photolysis of aromatic ethers. Z. Naturforsch. 1979, 34b, 1002–1009. (32) Ye, M.; Schuler, R. H. The second order combination reactions of phenoxyl radicals. J. Phys. Chem. 1989, 93, 1898–1902. (33) Jin, F.; Leitich, J.; von Sonntag, C. The superoxide radical reacts with tyrosine-derived phenoxyl radicals by addition rather than by electron transfer. J. Chem. Soc., Perkin Trans. 2 1993, 1583–1588. (34) Hoigne, J.; Bader, H. Rate constants of reactions of ozone with organic and inorganic compounds in water—I. Non-dissociating organic compounds. Water Res. 1983, 17, 173–183. (35) Mertens, R.; von Sonntag, C. Determination of the kinetics of vinyl radical reactions by the characteristic visible absorption of vinyl peroxyl radicals. Angew. Chem., Int. Ed. Engl. 1994, 33, 1262–1264. (36) Naumov, S.; von Sonntag, C. UV-visible absorption spectra of alkyl-, vinyl-, aryl- and thiylperoxyl radicals and some related radicals in aqueous solution: A quantum-chemical study. J. Phys. Org. Chem. 2005, 18, 586–594. 9203

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

Environmental Science & Technology (37) Mvula, E.; von Sonntag, C. Ozonolysis of phenols in aqueous solution. Org. Biomol. Chem. 2003, 1, 1749–1756. (38) Ramseier, M. K.; von Gunten, U. Mechanism of phenol ozonization—Kinetics of formation of primary and secondary products. Ozone: Sci. Eng. 2009, 31, 201–215. (39) Sehested, K.; Holcman, J.; Bjergbakke, E.; Hart, E. J. A pulse radiolytic study of the reaction OH + O3 in aqueous medium. J. Phys. Chem. 1984, 88, 4144–4147. (40) Mu~noz, F.; von Sonntag, C. Determination of fast ozone reaction by competition kinetics. J. Chem. Soc., Perkin Trans. 2 2000, 661–664. (41) Sein, M. M.; Zedda, M.; Tuerk, J.; Schmidt, T. C.; Golloch, A.; von Sonntag, C. Oxidation of diclofenac with ozone in aqueous solution. Environ. Sci. Technol. 2008, 42, 6656–6662. (42) Kl€aning, U. K.; Sehested, K.; Holcman, J. Standard Gibbs energy of formation of the hydroxyl radical in aqueous solution. Rate constants for reaction ClO2
+ O3 a O3
+ ClO2. J. Phys. Chem. 1985, 89, 760–763. (43) Nicoson, J. S.; Wang, L.; Becker, R. H.; Huff, K. E.; Muller, C. E.; Margerum, D. W. Kinetics and mechanism of the ozone/bromite and ozone/chlorite reactions. Inorg. Chem. 2002, 41, 2975–2980. (44) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (•OH/•O
) in aqueous solution. J. Phys. Chem. Ref. Data 1988, 17, 513–886. (45) von Gunten, U. Ozonation of drinking water: Part II. Disinfection and by-product formation. Water Res. 2003, 37, 1469–1487. (46) Stanbury, D. M. Reduction potentials involving inorganic free radicals in aqueous solution. Adv. Inorg. Chem. 1989, 33, 69–138. (47) Buxton, G. V.; Bydder, M.; Salmon, G. A. Reactivity of chlorine atoms in aqueous solution. Part 1: The equilibrium Cl 3 + Cl
= Cl2•
. J. Chem. Soc., Faraday Trans. 1998, 94, 653–657. (48) Yu, X.-Y.; Barker, J. R. Hydrogen peroxide photolysis in acidic solutions containing chloride ions. I. Chemical mechanism. J. Phys. Chem. A 2003, 107, 1313–1324. (49) Liu, Y.; Pimentel, A. S.; Antoku, Y.; Barker, J. R. Temperaturedependent rate and equilibrium constants for Br•(aq) + Br
(aq) a Br2
•(aq). J. Phys. Chem. A 2002, 106, 11075–11082. (50) Bielski, B. H. J. A pulse radiolysis study of the reaction of ozone with Cl2•
in aqueous solutions. Radiat. Phys. Chem. 1993, 41, 527–530. (51) Hoigne, J.; Bader, H.; Haag, W. R.; Staehelin, J. Rate constants of reactions of ozone with organic and inorganic compounds in water III. Inorganic compounds and radicals. Water Res. 1985, 19, 993–1004. (52) Chertova, Yu. S.; Avzyyanova, E. V.; Timerganzin, K. K.; Khalizov, A. F.; Sherreshovets, V. V.; Imashev, U. B. The formation of singlet molecular oxygen in the interaction of chlorine dioxide with ozone. Russ. J. Phys. Chem. 2000, 74 (Suppl. 3), 473–475. (53) Rao, B.; Anderson, T. A.; Redder, A.; Jackson, W. A. Perchlorate formation by ozone oxidation of aqueous chlorine/oxy-chlorine species: Role of ClxOy radicals. Environ. Sci. Technol. 2010, 44, 2961–2967. (54) Buxton, G. V.; Dainton, F. The radiolysis of aqueous solutions of oxybromine compounds; the spectra and reactions of BrO and BrO2. Proc. R. Soc. London, A 1968, 304, 427–439. (55) Pan, X.-M.; von Sonntag, C. OH-radical-induced oxidation of benzene in the presence of oxygen: R• a RO2• equilibria in aqueous solution. A pulse radiolysis study. Z. Naturforsch. 1990, 45b, 1337–1340. (56) Pan, X.-M.; Schuchmann, M. N.; von Sonntag, C. Oxidation of benzene by the OH radical. A product and pulse radiolysis study in oxygenated aqueous solution. J. Chem. Soc., Perkin Trans. 2 1993, 289–297. (57) Fang, X.; Pan, X.; Rahmann, A.; Schuchmann, H.-P.; von Sonntag, C. Reversibility in the reaction of cyclohexadienyl radicals with oxygen in aqueous solution. Chem.—Eur. J. 1995, 1, 423–429. (58) Hunter, E. P. L.; Desrosiers, M. F.; Simic, M. G. The effect of oxygen, antioxidants and superoxide radical on tyrosine phenoxyl radical dimerization. Free Radical Biol. Med. 1989, 6, 581–585. (59) Wang, D.; Gy€orgy, G.; Hildenbrand, K.; von Sonntag, C. Freeradical-induced oxidation of phloroglucinol—A pulse radiolysis and EPR study. J. Chem. Soc., Perkin Trans. 2 1994, 45–55.

ARTICLE

(60) Zhang, X.; Zhang, N.; Schuchmann, H.-P.; von Sonntag, C. Pulse radiolysis of 2-mercaptoethanol in oxygenated aqueous solution. Generation and reactions of the thiylperoxyl radical. J. Phys. Chem. 1994, 98, 6541–6547.

9204

dx.doi.org/10.1021/es2018658 |Environ. Sci. Technol. 2011, 45, 9195–9204

ARTICLE pubs.acs.org/est

Enhanced Transport of Colloidal Oil Droplets in Saturated and Unsaturated Sand Columns Micheal J. Travis, Amit Gross, and Noam Weisbrod* Department of Environmental Hydrology & Microbiology, Zuckerberg Institute for Water Research, The Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede Boqer Campus, Midreshet Ben-Gurion, 84990 Israel

bS Supporting Information ABSTRACT: Colloidal-sized triacylglycerol droplets demonstrated enhanced transport compared to ideal latex colloid spheres in both saturated and unsaturated quartz sand columns. Oil droplets (mean diameter 0.74 ( 0.03 μm, density 0.92 g cm 3, ζ-potential 34 ( 1 mV) were injected simultaneously with latex microsphere colloids (FluoSpheres; density 1.055 g cm 3, diameters 0.02, 0.2, and 1.0 μm, ζ-potentials 16 ( 1, 30 ( 2, and 49 ( 1, respectively) and bromide into natural quartz sand (ζ-potential 63 ( 2 mV) via short-pulse column breakthrough experiments. Tests were conducted under both saturated and unsaturated conditions. Breakthrough of oil droplets preceded bromide and FluoSpheres. Recovery of oil droplets was 20% greater than similarly sized FluoSpheres in the saturated column, and 16% greater in the 0.18 ( 0.01 volumetric water content (VWC) unsaturated column. Higher variability was observed in the 0.14 ( 0.01 VWC column experiments with oil droplet recovery only slightly greater than similarly sized FluoSpheres. The research presents for the first time the direct comparison of colloidal oil droplet transport in porous media with that of other colloids, and demonstrates transport under unsaturated conditions. Based on experimental results and theoretical analyses, we discuss possible mechanisms that lead to the observed enhanced mobility of oil droplets compared to FluoSpheres with similar size and electrostatic properties.

’ INTRODUCTION Triacylglycerol (i.e., food oil) contamination of the environment may occur in various ways such as: soil-based waste disposal of materials from edible oil or other food processing operations;1 bioremediation of recalcitrant organic pollutants in which edible oil is introduced to the subsurface;2,3 and irrigation with wastewater.4 Edible oils are essentially insoluble in water, but may be dispersed in solution as colloidal-sized droplets from a few nanometers to several micrometers in diameter. Emulsion droplets in water may be stabilized by surfactants or finely divided solids,5 and present difficult treatment challenges.6 If solutions containing oil emulsion droplets are applied to, and move through the soil, they can enhance the transport of oil soluble compounds such as pesticides, pharmaceuticals, or environmental estrogens.7 The transport of edible oil emulsion droplets through saturated porous media has been previously published (e.g., refs 2,3,8 11). However, many of these studies focused on relatively large oil droplets (several μm8), or highly concentrated emulsion solutions (e.g., >10% oil3,9). To the best of our knowledge, the direct comparison of dilute oil colloid mobility in porous media to that of latex microsphere “ideal” colloids10 14 has never been reported. Furthermore, the transport of oil emulsion droplets in unsaturated media has not been documented. Transport characteristics of colloids in porous media may be influenced by water content,12 pore velocity,13 colloid r 2011 American Chemical Society

concentration,14 pH,15 and ionic strength.16 Colloid retention is influenced by factors at the interface, collector, and pore scales.17 Filtration theory predicts the transport of colloids through porous media.18 Derjaguin Landau Verwey Overbeck (DLVO) theory quantifies electrostatic energies to predict colloid and collector surfaces interactions.19,20 Colloid colloid and colloid grain surface interactions may be influenced by surface charge heterogeneity,21 the air water interface,22 Lewis acid base interactions,23 and steric contributions.24 The link between colloid transport in unsaturated versus saturated media has been explored mainly for latex microspheres as “ideal” colloids (e.g., refs 17,25 27). Nevertheless, the interactions of physical and chemical processes that govern unsaturated colloid transport and retention are still not well understood.17,28 Multiple interfaces in unsaturated media (e.g., solid solid, solid water, air water, air water solid) increase potential colloid retention through wedging/straining, bridging, or film straining.16,17,29 Furthermore, different colloids have been shown to possess unique transport qualities (e.g., biocolloids30,31 and clay13,32). Received: December 2, 2010 Accepted: September 28, 2011 Revised: September 6, 2011 Published: September 28, 2011 9205

dx.doi.org/10.1021/es104040k | Environ. Sci. Technol. 2011, 45, 9205–9211

Environmental Science & Technology The objective of this research was to determine the transport characteristics of colloidal-sized oil droplets under conditions in saturated and unsaturated natural sand, and to compare with the mobility of “ideal” latex microspheres.

’ MATERIALS AND METHODS Soil Columns and Porous Media. Short-pulse tracer experiments were conducted in soil columns (9.9 cm diameter, 30 cm long) equipped to control and monitor volumetric water content (VWC) and matric pressure. The setup was similar to systems used in previous studies.33,34 A detailed schematic is included in Figure SI-1 of the Supporting Information (SI). The columns were packed with washed and sieved (300 500 μm) natural beach sand from the coastal region of southern Israel, as detailed in SI S1. No clay-minerals or organic matter were detected in the sand. The sand was >99% quartz, with iron and other metals detected as potential oxide coatings by elemental analysis of the grain surfaces (SI Table SI-1). Three separate column packs were used, each for a set of four replicate experiments: (1) saturated; (2) unsaturated “high” water content (0.18 ( 0.01 VWC); and (3) unsaturated “low” water content (0.14 ( 0.01 VWC), so that porosity and pore structure were uniform within each set of experiments. Average pore diameter for the sand was 75 ( 2 μm calculated from capillary rise experiments,35 described in SI S2. Bulk density of the sand was 1.72 ( 0.02 g cm 3, and porosity 0.35 ( 0.007. Saturated hydraulic conductivity, measured by the constant head method, was 0.039 ( 0.001 cm s 1. ζ-potential of the sand grains was 63 ( 2 mV (pH 7), calculated from streaming potential (Anton Paar, Graz, Austria) using the Helmholtz-Smoluchowski equation and the Fairbrother-Mastin approach.36 Soil column and sand characteristics are summarized in SI Table SI-2. Solution was introduced to the upper surface of the column via a rain simulator. Background Solution and Pulse Preparation. All experiments used artificial rainwater (ARW)37 for the background and tracer solutions (SI S3) with ionic strength 0.021 mM, pH 7.2 ( 0.2, and electrical conductivity 183 ( 10 μS cm 1. Tracers included (1) oil droplets (100 mg L 1); (2) 1.0, 0.2, and 0.02 μm diameter FluoSpheres (Invitrogen Corporation, Eugene, OR, at concentrations 1, 5, and 10 mg L 1, respectively); and (3) lithium bromide (40 mg L 1). FluoSpheres are monodisperse, carboxylate-modified latex microspheres impregnated with fluorescent dye. FluoSpheres sizes were selected within the approximate range of oil droplet sizes (described below) in order to enable comparison. Oil droplet preparation is detailed in SI S4. Oil droplets were prepared first in a stable, concentrated emulsion consisting of 45% refined sunflower oil, 2.5% surfactant and 52.5% ARW. The concentrated emulsion was then diluted into the tracer solution. Final surfactant concentration in the tracer solution was 5 mg L 1, MnIV
OH þ H3 AsIII O3 ðaqÞ f Mn2þ ðaqÞ þ HAsV O4 2
ðaqÞ þ 3Hþ

ð1Þ

To briefly summarize Lafferty et al.,6,24 the reaction between AsIII and δ-MnO2 in a stirred-flow reactor, under the conditions used in this study, proceeds in two distinct phases. First, an initial reaction phase occurs from 0 to 6.4 h, which includes the period of fastest AsIII oxidation, highest AsV sorption, and no Mn2+ release into solution (Figure 1). A second reaction phase characterized by lower δ-MnO2 reactivity occurs beyond 6.4 h, which includes a second period of (decreased) As sorption, a decrease in AsIII oxidation rate, and the presence of Mn2+ in solution (Figure 1). Decreased δ-MnO2 reactivity in the second phase of this reaction has been attributed to Mn2+ sorption on the δ-MnO2 surface and the subsequent production of MnIII via Mn(II)/(IV) conproportionation at the δ-MnO2 surface.24 In this study, desorption experiments are conducted by stopping the initial reaction between AsIII and δ-MnO2 after 4, 10, or 24 h (Figure 1), and simultaneously beginning desorption by PO4, Ca2+, or background electrolyte alone. The first time point for beginning desorption is after 4 h of reaction between AsIII and δ-MnO2, which coincides with maximum AsV concentration in the stirred-flow reactor effluent, and the end of an initial period of AsV sorption (Figure 1). Between 0 and 4 h of AsIII oxidation, Mn2+ is expected to react primarily with vacancy sites and not edge sites. The second time point for beginning desorption is after 10 h of AsIII oxidation by δ-MnO2, which is near the end of a second period of lesser As sorption, and occurs early in the second, less reactive phase of AsIII oxidation (Figure 1). Between 4 and 10 h, Mn2+ is expected to begin reacting with edge sites, resulting in formation of some MnIII.24 A change in the sorption 9219

dx.doi.org/10.1021/es201281u |Environ. Sci. Technol. 2011, 45, 9218–9223

Environmental Science & Technology

ARTICLE

Table 1. Structural Parameters Derived from Least-Square Fits to Raw k3-Weighted As-EXAFS Spectra for δ-MnO2 after 24 h Desorption by Background Electrolyte (Elec), Calcium Solution (Ca), and Phosphate Solution (PO4)a sample

As
O b

b

As
Mn σ

b

r

b

As
Mn CN

rb

σ2b

0.005(1)

0.6(4)

3.50(5)

0.005(3)

0.005(2)

0.3(5)

3.51(9)

0.004(7)

time

CN

r

10 h-elec

4.1(2)

1.70(1)

0.003(0)

1.1(3)

3.12(2)

10 h-Ca

4.1(2)

1.70(1)

0.003(0)

0.7(4)

3.14(3)

10 h-PO4 24 h-elec

4.2(2) 4.1(1)

1.70(1) 1.70(1)

0.003(0) 0.003(0)

0.8(3) 0.8(3)

3.16(3) 3.15(2)

0.005(2) 0.005(2)

2b

CN

σ

2b

24 h-Ca

4.1(1)

1.69(0)

0.003(0)

0.9(3)

3.14(2)

0.005(2)

24 h-PO4

4.3(2)

1.69(0)

0.003(0)

0.7(3)

3.16(3)

0.006(3)

b

a Desorption data shown here followed 10 or 24 h of AsIII oxidation. b Coordination number (CN), interatomic distance (r), and Debye
Waller factor (σ2) were obtained by fitting data with theoretical phase and amplitude functions. Estimated errors at 95% confidence interval from the least-squares fit are given in parentheses.

complexes formed between AsV and δ-MnO2 also occurs between 4 and 10 h of AsIII oxidation.24 The last time point for desorption is after 24 h of reaction, when the system is stable within the less reactive phase of the reaction (Figure 1). AsIII Desorption. No AsIII is desorbed from the δ-MnO2 surface in any desorption experiments discussed here. Also, As EXAFS analysis indicates that all As associated with δ-MnO2 after desorption is present as AsV (Table 1 and Figures 2 and 3), which agrees with previous results indicating that As present on phyllomanganate surfaces only occurs as AsV.14,23,24,35 However, it should be noted that there is not a sufficient amount of As remaining on the surface of δ-MnO2 after 4 h of AsIII oxidation followed by 24 h of desorption to measure As using EXAFS analysis. AsV Desorption. Previous studies have shown that As reacts primarily with edge sites of phyllomanganates rather than vacancy sites,14,23,24 therefore AsV desorption in this study is expected to occur at δ-MnO2 edge sites. Of the desorptives used in this study, PO4 is expected to desorb AsV most readily because it is chemically similar to AsV and is known to compete with AsV for sorption sites on metal oxide minerals.12,36
38 However, Ca2+ has the potential to react with δ-MnO2 vacancy sites as well as edge sites, therefore, Ca2+ also has the potential to desorb AsV. The background electrolyte (MOPS and NaCl) used in these studies is expected to react weakly with δ-MnO2 edge sites, and thus should not desorb AsV to a large extent. When AsV is desorbed (for 24 h) after 4 h of AsIII oxidation, roughly 67% of AsV sorbed during the 4 h of AsIII oxidation is mobilized from the δ-MnO2 surface by all three desorptives (Figure 2). Because of this, one can infer that the majority of AsV sorbed on the δ-MnO2 surface during the initial phase of high δ-MnO2 reactivity is fairly labile. Conversely, there is a portion of AsV sorbed during the first 4 h of AsIII oxidation that remains immobile on the δ-MnO2 surface, even in the presence of PO4. Previous EXAFS analysis of δ-MnO2 reacted with AsIII under identical experimental conditions revealed that AsV is bound in mononuclear-monodentate as well as binuclear-bidentate complexes on the δ-MnO2 surface during the first 4 h of AsIII oxidation by δ-MnO2.24 Unfortunately, there is not a sufficient amount of As remaining on the surface of δ-MnO2 after 4 h of AsIII oxidation followed by 24 h of desorption to determine the stability of these two complexes by EXAFS analysis. After 10 and 24 h of AsIII oxidation, PO4 is a more efficient desorptive of AsV than Ca2+ or the background electrolyte (Figure 2). Also, the proportion of AsV desorbed by PO4

increases in the 10 and 24 h experiments (Figure 2). It should be noted that after 10 and 24 h of AsIII oxidation by δ-MnO2, two significant changes occur in the speciation of Mn associated with δ-MnO2. First, Mn2+ begins sorbing at edge sites after δ-MnO2 vacancy sites are occupied by sorbed Mn2+ (at ∼6.4 h).6,24 Also, MnIII begins to appear in Mn octahedral layers of δ-MnO2 between 4 and 10 h of AsIII oxidation, and increases between 10 and 24 h.24 An increase in the proportion of AsV desorbed by all desorptives after 10 and 24 h (compared to 4 h) happens concurrently with increased competition from Mn2+ for edge sites, and an increase in MnIII content within the δ-MnO2 structure. Thus, increased AsV desorption in the 10 and 24 h experiments could be the result of direct competition between AsV and Mn2+ for sorption sites or the formation of weaker bonds between AsV and MnIII.39 It is difficult to distinguish between the effects of increased Mn2+ sorption and increased MnIII content at δ-MnO2 edge sites as they occur simultaneously. EXAFS analysis of δ-MnO2 after AsIII oxidation for 10 h and subsequent desorption by Ca2+ and background electrolyte revealed As
Mn distances of ∼3.13 Å and ∼3.50 Å (Table 1). These distances correspond to AsV bound to the δ-MnO2 surface in bidentate-binuclear and monodentate-mononuclear complexes, respectively.22
24 However, for all other desorption experiments after 10 (PO4) and 24 (Ca2+, PO4, and background electrolyte) hours of AsIII oxidation by δ-MnO2 the only As
Mn distances present in EXAFS spectra was ∼3.15 Å (Table 1 and Figure 3), corresponding to a bidentate-binuclear complex between AsV and the δ-MnO2 surface. While, it is tenuous to attribute a specific desorption event with a single adsorption complex, EXAFS data of As sorption complexes before and after desorption seem to indicate that bidentate-mononuclear and monodentate-mononuclear complexes between AsV and the δ-MnO2 surface are less stable than AsV- δ-MnO2 bidentatebinuclear complexes. Mn Desorption. During AsIII oxidation by δ-MnO2, Mn2+ is produced and subsequently sorbed by δ-MnO2 (eq 1 and Figure 1).6,24 Mn2+ tends to initially sorb at δ-MnO2 layer vacancy sites under the conditions used in this study, followed by sorption at δ-MnO2 edge sites as vacancy sites become more occupied.24 Previous studies have indicated that some As sorbed on phyllomanganate surfaces could be bound through a bridging complex through sorbed Mn.12,25 Although As/δ-MnO2 bridging complexes were not seen in previous studies conducted under the experimental conditions used in the reactions described here, it is possible that Mn2+ on δ-MnO2 could facilitate As sorption. 9220

dx.doi.org/10.1021/es201281u |Environ. Sci. Technol. 2011, 45, 9218–9223

Environmental Science & Technology

ARTICLE

Figure 2. AsV (left) and Mn2+ (right) desorbed by Ca2+, PO4, and background electrolyte (10 mM NaCl, 5 mM MOPS) after AsIII oxidation by δ-MnO2. The initial data points on each graph (time = 0 h) correspond to the beginning of desorption (initial AsIII oxidation data not shown). Data shown are first 10 h of 24 h desorption experiments.

Of the desorptives used in this study, Ca2+ is expected to react with δ-MnO2 sorption sites most similarly to Mn2+.40 Some cations could potentially desorb Mn2+ more readily than Ca2+,32,41,42 however Ca2+ is ubiquitous in nature, and thus has a high probability of interacting with Mn2+ sorbed on Mn-oxide surfaces. Desorption of Mn2+ by Na+ (in background electrolyte) is predicted to be negligible because Na+ reacts with δ-MnO2

interlayers differently than Ca2+ or Mn2+, in that Na+ is not expected to bind in triple corner sharing complexes at vacancy sites as is the case with Mn2+ and Ca2+.40 When AsIII is reacted with δ-MnO2 for only 4 h, no Mn2+ is desorbed under the conditions used in this study (data not shown). It is important to note that no Mn2+ appears in the stirred-flow reactor effluent during the first 4 h of AsIII oxidation, and all Mn2+ 9221

dx.doi.org/10.1021/es201281u |Environ. Sci. Technol. 2011, 45, 9218–9223

Environmental Science & Technology

ARTICLE

is a certain amount of As that is not desorbed under any of the conditions studied here. Thus, if As comes in contact with Mnoxides in nature, these minerals could potentially decrease As availability and mobility both by oxidation of AsIII and sorption of AsV. It appears that AsV and Mn2+ desorption potential is intricately linked to the type of reaction site on the δ-MnO2 surface to which each is bound, as well as Mn speciation within the δ-MnO2 structure. This study emphasizes the importance of understanding mineral structures and temporal variability when predicting As mobility in the environment.

’ ASSOCIATED CONTENT

bS

Supporting Information. Supporting Information is provided which includes detailed information about EXAFS analysis, Mn EXAFS fitting results, further Mn desorption discussion, and an example of aqueous data from a full experiment. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: (601) 634-3589; fax: (601) 634-4017; e-mail: blafferty@ gmail.com. Figure 3. Fourier transformed As K-edge EXAFS of δ-MnO2 reacted with AsIII (100 μM) in a stirred-flow reactor for 10 and 24 h (10 h start and 24 h start) and desorbed by Ca2+, PO4, and background electrolyte for 24 h following AsIII reaction. XAS data are presented as solid lines and fits are presented as dashed lines (fit data provided in Table 1).

produced during this time is expected to sorb strongly at δ-MnO2 vacancy sites.24 However, after 10 and 24 h of AsIII oxidation, Mn2+ is desorbed by all desorptives studied (Figure 2), indicating that Mn2+ sorbed at δ-MnO2 edge sites is more labile than Mn2+ sorbed at δ-MnO2 vacancy sites. Mn EXAFS analysis of δ-MnO2 revealed no detectable changes in Mn speciation of the solid material after desorption which would appear as a broadening and decrease in the peak height of the 9.25 Å
1 peak in the EXAFS spectra 24 (Figures S2A and S2B and Table S1 of the Supporting Information, SI). As predicted, Ca2+ is the most efficient Mn2+ desorptive of those studied. The proportion of Mn2+ desorbed by Ca2+ is greatest after 10 h of AsIII oxidation and decreases slightly after 24 h of AsIII oxidation (Figure 2). This decrease in Mn2+ mobility with increased AsIII oxidation time could potentially be due to increased formation of less mobile MnIII after 10 h of reaction.6,24 Also, desorption by PO4 (with background electrolyte present) is nearly identical to desorption by background electrolyte alone in the 10 and 24 h samples (Figure 2), which suggests that PO4 does not desorb Mn2+ appreciably. Interestingly, increased Mn2+ desorption by Ca2+ compared to other desorptives does not result in an increase in AsV desorption, which provides some evidence that AsV is not bound to the δ-MnO2 surface via a bridging complex through Mn2+. Implications for As Mobility. Phyllomanganates are capable of sorbing AsV, especially during AsIII oxidation. However, in this study, AsV can be desorbed from the δ-MnO2 surface, to some extent, under all conditions studied. Even Na+ (present in background electrolyte) is able to desorb AsV, to some extent, under all conditions studied here, indicating that a portion of As sorbed by Mn-oxides is potentially quite mobile in the environment. Although some sorbed AsV can be desorbed from δ-MnO2, there

Present Addresses †

United States Army Corps of Engineers, Engineer Research and Development Center, 3909 Halls Ferry Rd, Vicksburg, MS 39180. ‡ Calera Corporation, 14600 Winchester Blvd., Los Gatos, CA 95030.

’ ACKNOWLEDGMENT The authors thank Gerald Hendricks and Caroline Golt for laboratory assistance. B.L. is grateful for funding provided by a University of Delaware graduate fellowship and the Donald L. and Joy G. Sparks Graduate Fellowship in Soil Science. This research was funded by United States Department of Agriculture Grant 2005
35107-16105, National Science Foundation Grant EAR-0544246, and Delaware National Science Foundation EPSCoR Grant EPS-0447610. Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886. ’ REFERENCES (1) Cullen, W. R.; Reimer, K. J. Arsenic speciation in the environment. Chem. Rev. 1989, 89, 713–764. (2) Sadiq, M. Arsenic chemistry in soils: An overview of thermodynamic predictions and field observations. Water, Air, Soil Pollut. 1997, 93, 117–136. (3) Petrick, J. S.; Ayala-Fierro, F.; Cullen, W. R.; Carter, D. E.; Aposthian, H. V. Monomethylarsonous acid (MMA(III)) is more toxic than arsenite in Chang human hepatocytes. Toxicol. Appl. Pharmacol. 2000, 163, 203–207. (4) Driehaus, W.; Seith, R.; Jekel, M. Oxidation of arsenate (III) with manganese oxides in water treatment. Water Res. 1995, 29, 297–305. (5) Ginder-Vogel, M.; Landrot, G.; Fischel, J. S.; Sparks, D. L. Quantification of rapid environmental redox processes with quickscanning x-ray absorption spectroscopy (Q-XAS). Proc. Natl. Acad. Sci. 2009, 106, 16124–16128. (6) Lafferty, B. J.; Ginder-Vogel, M.; Sparks, D. L. Arsenite oxidation by a poorly crystalline manganese-oxide 1. Stirred-flow experiments. Environ. Sci. Technol. 2010, 44, 8460–8466. 9222

dx.doi.org/10.1021/es201281u |Environ. Sci. Technol. 2011, 45, 9218–9223

Environmental Science & Technology (7) Moore, J. N.; Walker, J. R.; Hayes, T. H. Reaction scheme for the oxidation of As (III) to As (V) by birnessite. Clays Clay Miner. 1990, 38, 549–555. (8) Nesbitt, H.; Canning, G.; Bancroft, G. XPS study of reductive dissolution of 7Å-birnessite by H3AsO3, with constraints on reaction mechanism. Geochim. Cosmochim. Acta 1998, 62, 2097–2110. (9) Oscarson, D.; Huang, P.; Defosse, C.; Herbillon, A. Oxidative power of Mn (IV) and Fe (III) oxides with respect to As (III) in terrestrial and aquatic environments. Nature 1981, 291, 50–51. (10) Oscarson, D.; Huang, P.; Liaw, W. Role of manganese in the oxidation of arsenite by freshwater lake sediments. Clays Clay Miner 1981, 29, 219–225. (11) Oscarson, D.; Huang, P.; Hammer, U.; Liaw, W. Oxidation and sorption of arsenite by manganese dioxide as influenced by surface coatings of iron and aluminum oxides and calcium carbonate. Water, Air, Soil Pollut. 1983, 20, 233–244. (12) Parikh, S. J.; Lafferty, B. J.; Meade, T. G.; Sparks, D. L. Evaluating Environmental Influences on AsIII Oxidation Kinetics by a Poorly Crystalline Mn-Oxide. Environ. Sci. Technol. 2010, 44, 3772–3778. (13) Scott, M. J.; Morgan, J. J. Reactions at oxide surfaces. 1. Oxidation of As (III) by synthetic birnessite. Environ. Sci. Technol. 1995, 29, 1898–1905. (14) Tournassat, C.; Charlet, L.; Bosbach, D.; Manceau, A. Arsenic(III) oxidation by birnessite and precipitation of manganese(II) arsenate. Environ. Sci. Technol. 2002, 36, 493–500. (15) Arai, Y.; Elzinga, E. J.; Sparks, D. L. X-ray absorption spectroscopic investigation of arsenite and arsenate adsorption at the aluminum oxide-water interface. J. Colloid Interface Sci. 2001, 235, 80–88. (16) Dixit, S.; Hering, J. G. Comparison of arsenic(V) and arsenic(III) sorption onto iron oxide minerals: Implications for arsenic mobility. Environ. Sci. Technol. 2003, 37, 4182–4189. (17) Raven, K. P.; Jain, A.; Loeppert, R. H. Arsenite and arsenate adsorption on ferrihydrite: kinetics, equilibrium, and adsorption envelopes. Environ. Sci. Technol. 1998, 32, 344–349. (18) Masue, Y.; Loeppert, R. H.; Kramer, T. A. Arsenate and arsenite adsorption and desorption behavior on coprecipitated aluminum: Iron hydroxides. Environ. Sci. Technol. 2007, 41, 837–842. (19) Anderson, M. A.; Ferguson, J. F.; Gavis, J. Arsenate adsorption on amorphous aluminum hydroxide. J. Colloid Interface Sci. 1976, 54, 391–399. (20) Gupta, S. K.; Chen, K. Y. Arsenic removal by adsorption. J. Water Pollut. Control Fed. 1978, 50, 493–506. (21) Hingston, F. J. In Adsorption of Inorganics at Solid-Liquid Interfaces; Anderson, M. A.; Rubin, A. J., Eds.; Ann Arbor Science: Ann Arbor, MI, 1981; pp 51
90. (22) Foster, A. L.; Brown, G. E.; Parks, G. A. X-ray absorption fine structure study of As (V) and Se (IV) sorption complexes on hydrous Mn oxides. Geochim. Cosmochim. Acta 2003, 67, 1937–1953. (23) Manning, B. A.; Fendorf, S. E.; Bostick, B.; Suarez, D. L. Arsenic(III) oxidation and arsenic(V) adsorption reactions on synthetic birnessite. Environ. Sci. Technol. 2002, 36, 976–981. (24) Lafferty, B. J.; Ginder-Vogel, M.; Zhu, M.; Livi, K. J. T.; Sparks, D. L. Arsenite oxidation by a poorly crystalline manganese-oxide. 2. Results from X-ray absorption spectroscopy and X-ray diffraction. Environ. Sci. Technol. 2010, 44, 8467–8472. (25) Tani, Y.; Miyata, N.; Ohashi, M.; Ohnuki, T.; Seyama, H.; Iwahori, K.; Soma, M. Interaction of inorganic arsenic with biogenic manganese oxide produced by a Mn-oxidizing fungus, strain KR21
2. Environ. Sci. Technol. 2004, 38, 6618–6624. (26) Drits, V. A.; Silvester, E.; Gorshkov, A. I.; Manceau, A. Structure of synthetic monoclinic Na-rich birnessite and hexagonal birnessite: I. Results from X-ray diffraction and selected-area electron diffraction. Am. Mineral. 1997, 82, 946–961. (27) Silvester, E.; Manceau, M.; Drits, V. A. Structure of synthetic monoclinic Na-rich birnessite and hexagonal birnessite: II. Results from chemical studies and EXAFS spectroscopy. Am. Mineral. 1997, 82, 962–978. (28) Marcus, M. A.; Manceau, A.; Kersten, M. Mn, Fe, Zn and As speciation in a fast-growing ferrmanganese marine nodule. Geochim. Cosmochim. Acta 2004, 68, 3125–3136.

ARTICLE

(29) Peacock, C. L.; Sherman, D. M. Sorption of Ni by birnessite: Equilibrium controls on Ni in seawater. Chem. Geol. 2007, 238, 94–106. (30) Manceau, A.; Tommaseo, C.; Rihs, S.; Geoffroy, N.; Chateigner, D.; Schlegel, M.; Tisserand, D.; Marcus, M. A.; Tamura, N.; Chen, Z. S. Natural speciation of Mn, Ni, and Zn at the micrometer scale in a clayey paddy soil using X-ray fluorescence, absorption, and diffraction. Geochim. Cosmochim. Acta 2005, 69, 4007–4034. (31) Manceau, A.; Lanson, M.; Geoffroy, N. Natural speciation of Ni, Zn, Ba, and As in ferromanganese coatings on quartz using X-ray fluorescence, absorption, and diffraction. Geochim. Cosmochim. Acta 2007, 71, 95–128. (32) Toner, B.; Manceau, A.; Webb, S. M.; Sposito, G. Zinc sorption to biogenic hexagonal-birnessite particles within a hydrated bacterial biofilm. Geochim. Cosmochim. Acta 2006, 70, 27–43. (33) Pena, J.; Kwon, K. D.; Refson, K.; Bargar, J. R.; Sposito, G. Mechanisms of nickel sorption by a bacteriogenic birnessite. Geochim. Cosmochim. Acta 2010, 74, 3076–3089. (34) Villalobos, M.; Bargar, J.; Sposito, G. Mechanisms of Pb(II) sorption on a biogenic manganese oxide. Environ. Sci. Technol. 2005, 39, 569–576. (35) Parikh, S. J.; Lafferty, B. J.; Sparks, D. L. An ATR-FTIR spectroscopic approach for measuring rapid kinetics at the mineral/ water interface. J. Colloid Interface Sci. 2008, 320, 177–185. (36) Jackson, B. P.; Miller, W. P. Effectiveness of phosphate and hydroxide for desorption of arsenic and selenium species from iron oxides. Soil Sci. Soc. Am. J. 2000, 64, 1616–1622. (37) Lafferty, B. J.; Loeppert, R. H. Methyl arsenic adsorption and desorption behavior on iron oxides. Environ. Sci. Technol. 2005, 39, 2120–2127. (38) Liu, F.; De Cristofaro, A.; Violante, A. Effect of pH, phosphate and oxalate on the adsorption/desorption of arsenate on/from goethite. Soil Sci. 2001, 166, 197–208. (39) Zhu, M.; Paul, K. W.; Kubicki, J. D.; Sparks, D. L. Quantum chemical study of arsenic (III, V) adsorption on Mn-oxides: Implications for arsenic(III) oxidation. Environ. Sci. Technol. 2009, 43, 6655–6661. (40) Drits, V. A.; Lanson, B.; Gaillot, A. C. Birnessite polytype systematics and identification by powder X-ray diffraction. American mineralogist 2007, 92, 771. (41) Murray, J. W. The interaction of metal ions at the manganese dioxide-solution interface. Geochim. Cosmochim. Acta 1975, 39, 505–519. (42) Tonkin, J. W.; Balistrieri, L. S.; Murray, J. W. Modeling sorption of divalent metal cations on hydrous manganese oxide using the diffuse double layer model. Appl. Geochem. 2004, 19, 29–53.

9223

dx.doi.org/10.1021/es201281u |Environ. Sci. Technol. 2011, 45, 9218–9223

ARTICLE pubs.acs.org/est

Multiple Fluorescence Labeling and Two Dimensional FTIR
13C NMR Heterospectral Correlation Spectroscopy to Characterize Extracellular Polymeric Substances in Biofilms Produced during Composting Guang-Hui Yu,‡,† Zhu Tang,‡,† Yang-Chun Xu,† and Qi-Rong Shen*,† †

Jiangsu Key Lab for Organic Solid Waste Utilization, College of Resources and Environmental Sciences, Nanjing Agricultural University, Nanjing 210095, People's Republic of China

bS Supporting Information ABSTRACT: Knowledge on the structure and function of extracellular polymeric substances (EPS) in biofilms is essential for understanding biodegradation processes. Herein, a novel method based on multiple fluorescence labeling and two-dimensional (2D) FTIR
13C NMR heterospectral correlation spectroscopy was developed to gain insight on the composition, architecture, and function of EPS in biofilms during composting. Compared to other environmental biofilms, biofilms in the thermophilic (>55 °C) and cooling (mature) stage of composting have distinct characteristics. The results of multiple fluorescence labeling demonstrated that biofilms were distributed in clusters during the thermophilic stage (day 14), and dead cells were detected. In the mature stage (day 26), the biofilm formed a continuous layer with a thickness of approximately 20
100 μm around the compost, and recolonization of cells at the surface of the compost was easily observed. Through 2D FTIR
13C NMR correlation heterospectral spectroscopy, the following trend in the ease of the degradation of organic compounds was observed: heteropolysaccharides > cellulose > amide I in proteins. And proteins and cellulose showed significantly more degradation than heteropolysaccharides. In summary, the combination of multiple fluorescence labeling and 2D correlation spectroscopy is a promising approach for the characterization of EPS in biofilms.

’ INTRODUCTION Biofilms are well-organized communities of microorganisms embedded in a matrix of extracellular polymeric substances (EPS).1
4 The composition of EPS is complex and is dependent on the bacterial species and the growth conditions. However, the main constituents of EPS are proteins, polysaccharides, cellulose, and lipids.1,3
5 In many bioprocesses, the growth of biofilm affects the degradation and conversion of organic compounds.6 Unfortunately, a complete biochemical profile of biofilms is difficult to obtain.5 Therefore, the structure and function of biofilms must be elucidated to obtain a deeper understanding of bioprocesses. Currently, one of the best approaches for the investigation of biofilms in situ is the use of fluorescently labeled lectins.5,7 Many investigators have shown that multiple fluorescence labeling and confocal laser scanning microscopy (CLSM) can be combined to obtain a powerful tool for studying the composition, architecture, and function of biofilm constituents at the microscale.3,4,6
10 Nevertheless, the identification and quantification of specific biofilm constituents is limited by the availability of fluorescently labeled probes.11 CLSM and various methods based on chemical structural analysis, such as Fourier transform infrared (FTIR) and nuclear magnetic resonance (NMR) spectroscopy, can be combined to provide a comprehensive understanding of biofilm development. In previous studies, changes in the structure of biofilm r 2011 American Chemical Society

constituents have been detected via traditional FTIR12
14 or NMR11,14,15 spectroscopy. However, the individual spectral features of FTIR or NMR often overlap because of the extreme heterogeneity of biofilm constituents.16 Two-dimensional (2D) correlation spectroscopy17 can be used to resolve the overlapped peaks problems of traditional FTIR or NMR spectroscopy. By distributing spectral intensity trends within a data set collected as a function of the perturbation sequence (e.g., time, temperature, pressure) over a second dimension, one can get 2D correlation spectroscopy. The main advantages of 2D correlation spectra are as follows: (i) simplification of complex spectra consisting of many overlapped peaks, and enhancement of spectral resolution by spreading peaks over the second dimension; (ii) establishment of unambiguous assignments through correlation of bands; (iii) probing the specific sequencing of spectral intensity changes through asynchronous analysis; (iv) so-called heterospectral correlation, i.e., the investigation of correlation among bands in two different types of spectroscopy; and (v) truly universal applicability of the technique, which is not limited to any type of spectroscopy, or even any form of analytical technique (e.g., chromatography, microscopy, etc).18,19 Received: April 30, 2011 Accepted: September 13, 2011 Revised: August 16, 2011 Published: September 13, 2011 9224

dx.doi.org/10.1021/es201483f | Environ. Sci. Technol. 2011, 45, 9224–9231

Environmental Science & Technology Although the second derivative and peak fitting analysis would also be used to solve the peak overlapping problem and enhance spectral resolution,14,20 both of them could not be applied to probe the specific sequencing of spectral intensity changes and investigate the heterospectral correlation. However, information on the distribution and architecture of biofilms cannot be obtained using this method. To our knowledge, 2D correlation spectroscopy has not previously been applied to investigate the function of biofilms in a bioprocess. Composting is a cheap, efficient, and sustainable treatment for solid organic materials.21
23 Until now, composting research has mainly focused on optimization of process parameters, degradtion of organic matter, and assessment of maturity.21
25 Few studies have been explored in the structure and function of EPS in biofilms of compost, which is essential for understanding biodegradation processes. Thus, the objectives of the present study were to combine multiple fluorescence labeling and 2D correlation spectroscopy to characterize the composition, architecture and function of biofilms. For this purpose, two piles in a full-scale compost facility were constructed and biofilms were allowed to grow. Compared to other environmental biofilms, those present in compost are expected to have distinct characteristics because of the presence of thermophilic and cooling (mature) stages.

’ MATERIALS AND METHODS Composting Process and Biofilm Sample Collection. Two windows with dimensions of 13  1  1.5 m (length  height  width) were constructed from a mixture of swine manure and wheat straw. The moisture content, pH, water extractable organic carbon (WSC), and water extractable total nitrogen (WSN) content of the feedstock in the two piles were 66.8 ( 0.1%, 8.0 ( 0.1, 13.6 ( 0.3 mg g
1, and 2.0 ( 0.1 mg g
1, respectively. Composting was performed under aerobic conditions for 26 days, and the piles were turned when a temperature of 60 °C was attained. During the composting process, 2 kg of representative material was collected on days 0, 2, 6, 10, 14, 18, 22, and 26 of composting and was then divided into two subsamples. The detailed description of sampling could be seen in Tang et al.24 Multiple fluorescent labeling and CLSM were conducted on one subsample of compost, and biofilm was extracted from the other subsample to determine its chemical structure and composition of the biofilm. Briefly, the biofilm was separated from the compost by shaking the samples in deionized water (solid to water ratio of 1:10 w/v) for 24 h on a horizontal shaker at room temperature. The separated biofilm from the fresh compost was filtered through a 0.45-μm polytetrafluoroethylene (PTFE) filter in dead-end membrane filtration tests controlling 30 cmHg vacuum before being freeze-dried at
50 °C for 48 h prior to performing a FTIR and NMR spectral analysis. Multiple Fluorescence Labeling and CLSM Observation. The hydrated compost samples were labeled with fluorescent stains possessing different excitation and emission spectra, and the distribution patterns of proteins, α-polysaccharides, cellulose, total cells, and dead cells were simultaneously visualized according to the method of Chen et al.8 In brief, fluoresceinisothiocyanate (FITC), concanavalin A (Con A), calcofluor white (CW), STYO 63, and SYTOX blue identify proteins, α-polysaccharides, cellulose, total cells, and dead cells, respectively. Specifically, SYTO 63 (20 μM, 100 μL) was added to the

ARTICLE

sample, and the resulting mixture was placed on a shaker table for 30 min. Subsequently, 0.1 mol of NaHCO3 buffer (100 μL) was added to maintain a pH of 9. A solution of FITC (10 g/L, 10 μL) was added, and the mixture was stirred for 1 h. Next, a solution of Con A (250 mg/L, 100 μL) was added to the sample for 30 min, followed by CW (300 mg/L, 100 μL) for 30 min. After each stage of the labeling process, the sample was washed twice with phosphate buffered saline (PBS) solution to remove the extra probe. Finally, a solution of SYTOX blue (2.5 μM, 100 μL) was incubated with the sample for 10 min. The labeled samples were embedded for cryosectioning and were then frozen at
20 °C. Subsequently, 30-μm sections were cut on a cryomicrotome (Cyrotome E, Thermo Shandon Limited, U.K.) and were mounted onto gelatin-coated (0.1% gelatin and 0.01% chromium potassium sulfate) microscopic slides for CLSM (Leica TCS SP2 confocal spectral microscope imaging system, Germany) observation. Four slides were made for each sample. In order to ensure the integrity of each slide, it was important to keep bubbles out of the samples when the labeled samples were embedded for cryosectioning. The samples were imaged using a  20 objective. Detailed information about the sample preparation for the CLSM slides is shown as Figure S1 of the Supporting Information, SI. Three-dimensional reconstructions were obtained with Leica confocal software, and movie files generated from the image stack were saved as uncompressed AVI files. Morphological parameters of the CLSM image were determined using Image J software (NIH, Bethesda, MD, U.S.A.). Analysis of FTIR and Solid-State 13C NMR Spectroscopy. Samples were prepared as a mixture of 1 mg of freeze-dried sample and 100 mg of potassium bromide (KBr, IR grade) and then ground and homogenized to reduce light scatter.26 A subsample was then compressed between two clean, polished iron anvils twice in a hydraulic press at 20 000 psi to form a KBr window. The FTIR spectra were obtained by collecting 200 scans with a Nicolet 370 FTIR spectrometer. Solid-state 13C
CPMAS-NMR spectroscopy was conducted on a Bruker AV-400, equipped with a 4-mm wide-bore MAS probe. NMR spectra were obtained by applying the following parameters: rotor spin rate of 13 000 Hz, 1 s recycle time, 1 ms contact time, 20 ms acquisition time, and 4000 scans. Samples were packed in 4-mm zirconia rotors with Kel-F caps. The pulse sequence was applied with a 1H ramp to account for nonhomogeneity of the Hartmann
Hahn condition at high spin rotor rates. Structural carbons determined include the following group shifts: 0
50 ppm (alkyl), 50
112 ppm (alcohol, amine, carbohydrate, ether, methoxyl and acetal), 110
145 ppm (aromatic), 145
163 ppm (phenolic), 163
215 ppm (carboxyl and carbonyl). Chemical shifts were calibrated with adamantine. Analysis of 2D Correlation Spectroscopy. The 2D correlation spectra were produced according to the method of Noda and Ozaki.18 In this study, the composting time was applied as an external perturbation, and a set of time-dependent FTIR or NMR spectra was obtained. Let us consider analytical spectrum I(x, t). The variable x is the index variable representing the FTIR or NMR spectra induced by the perturbation variable t. We intentionally use x instead of the general notation used in conventional 2D correlation equations based on spectral index v. Analytical spectrum I(x, t) at m evenly spaced points in t (between Tmin and Tmax) can be represented as follows: Ij ðxÞ ¼ Iðx, tj Þ, j ¼ 1, 2, 3 3 3 , m 9225

ð1Þ

dx.doi.org/10.1021/es201483f |Environ. Sci. Technol. 2011, 45, 9224–9231

Environmental Science & Technology

ARTICLE

Figure 1. Performance of the composting process.

A set of dynamic spectra is given by the following: ~I ðx, tÞ ¼ Iðx, tj Þ
̅ lðxÞ

ð2Þ

where ̅ l(x) denotes the reference spectrum, which is typically the average spectrum and is expressed as ̅ l(x) = 1/m∑m j = 1I(x, tj). The synchronous correlation intensity can be directly calculated from the following dynamic spectra: ϕðx1 , x2 Þ ¼

1 m ~I j ðx1 Þ~I j ðx2 Þ m
1 j¼1

∑

ð3Þ

Asynchronous correlation can be obtained by the following: Lðx1 , x2 Þ ¼

m 1 m ~I j ðx1 Þ Njk~I j ðx2 Þ m
1 j¼1 k¼1

∑

∑

ð4Þ

The term Njk corresponds to the jth column and the kth raw element of the discrete Hilbert
Noda transformation matrix, which is defined as follows: 8 > 0 < if j ¼ k 1 ð5Þ Njk ¼ > : πðk
jÞ otherwise The intensity of a synchronous correlation spectrum (L(x1, x2)) represents simultaneous changes in two spectral intensities measured at x1 and x2 during the interval between Tmin and Tmax. In contrast, an asynchronous correlation spectrum (j(x1, x2)) includes out-of-phase or sequential changes in spectral intensities measured at x1 and x2.

Figure 2. CLSM images of biofilms in pile 1 after 14 (A) and 26 (B) days of cultivation. The images were obtained with a 20 objective lens: (a) proteins (FITC), green; (b) α-polysaccharides (Con A), light blue; (c) cellulose (CW), blue; (d) total cells (SYTO 63), red; (e) dead cells (SYTO blue), violet; (f) merged image of (a)-(e). Bar = 100 μm.

Prior to 2D analysis, the FTIR or NMR spectra were normalized by summing the absorbance from 4000 to 400 cm
1 or 0
200 ppm, respectively, and multiplying by 1000. Subsequently, normalized FTIR or NMR spectra were analyzed using principal component analysis (PCA) to reduce the level of noise.27 Finally, 2D correlation spectroscopy was produced using 2Dshige software (Kwansei-Gakuin University, Japan).

’ RESULTS Performance of the Compost. The temperature, moisture content, and pH at various stages of the composting process are shown in Figure 1. Both of the piles attained a plateau value of 65 °C at the second day after composting, indicating that the piles rapidly reached the thermophilic phase. The temperature decreased to approximately 50 °C on the 18th day and then remained constant at approximately 60 °C. The moisture content abruptly declined from 70% on the first day to 30% on the 16th day. Evolution of temperature, moisture content, and fluorescence excitation
emission matrix contours of dissolved organic matter (Figure S2 of the SI) indicated that the compost was mature after 18 days, which is consistent with the results of our previous investigations.22
24 The pH of the piles climbed rapidly from 8.2 on the first day to 8.6 on the eighth day and remained constant over time. During composting, the pH of both 9226

dx.doi.org/10.1021/es201483f |Environ. Sci. Technol. 2011, 45, 9224–9231

Environmental Science & Technology piles ranged from 8.0 to 8.6, evidencing satisfactory microbial activity.25 The aforementioned results suggested that the characteristics of the piles were typical of those observed during composting. Moreover, the findings of the present study were consistent with those obtained from previous investigations.24,28,29 Architecture and Structure of Biofilms Observed by Multiple Fluorescence Labeling Combined with CLSM. Compared to biofilms observed during the cooling (mature) stage, biofilms in the thermophilic stage are expected to be distinct. Therefore, CLSM images of compost samples were obtained during both stages. For brevity, the CLSM images of pile 2 are provided in Figure S3 of the SI. Figure 2 displays the CLSM images of compost samples from pile 1 during the thermophilic (14th d) and mature stages (26th d), respectively. During the thermophilic stage, pig manure and wheat straw were visually apparent, with the former surrounding the latter. Bright images of the composts revealed that pig manure and wheat straw were present in the compost (Figure S4 of the SI). Proteins (FITC) were predominant in pig manure, whereas cellulose (CW) and αpolysaccharides (Con A) formed a continuous layer on the wheat straw. Total cells (SYTO 63) were primarily detected in wheat straw, while dead cells (SYTOX blue) were nearly ubiquitous in both pig manure and wheat straw. Three-dimensional reconstructions of the composts on the 14th day clearly demonstrated that biofilms in the thermophilic stage were highly dispersed throughout the material and were aggregated into clusters located along the outer of the compost (the movie documents generated from the image stack are provided in the Supporting Information). In addition, the wheat straw displayed the characteristics of lignocellulose, which suggested that most of the wheat straw was not completely degraded. During the mature stage, most of the pig manure was degraded, and only wheat straw was observed in the CLSM images. The fluorescence intensity of proteins, cellulose, and α-polysaccharides in the wheat straw during the mature stage was markedly lower that of the thermophilic stage. Moreover, the structure of wheat straw was loose, indicating that most of the wheat straw was degraded. Compared to the thermophilic stage, a quantity of cells in biofilm was observed during the mature stage. The total cell count in the mature stage was markedly greater than that of the thermophilic stage (14th d). Most of the total cells were distributed within the biofilm surrounding the wheat straw. Alternatively, dead cells were primarily observed in the interior of the wheat straw. These results revealed that cell recolonization occurs during the mature stage of composting. When composts were applied to soil, recolonized cells play an important role in the biological control of plant disease.31 In summary, multicolor fluorescence labeling provides information on the detailed architecture and distribution of biofilms in compost. The architecture and distribution patterns of biofilm constituents are closely related to the degradability of biofilms, which can be observed using 2D heterospectral correlation spectroscopy. Function of Biofilms Investigated by 2D Heterospectral Correlation Spectroscopy. The area-normalized FTIR and NMR spectra of the composts over time were noisy. Because the first two principal components accounted for 96% and 98% of the peaks in the FTIR and NMR spectra, respectively, the PCA noise reduction method was applied to reconstruct less noisy spectra. In the reconstructed spectra, the primary bands were maintained, and the level of noise was reduced (data not shown

ARTICLE

Figure 3. Synchronous and asynchronous 2D FTIR correlation maps generated from the 1800
900 cm
1 region of the spectra and 2D NMR correlation maps of dissolved organic matter in the two piles over time. Red represents positive correlations; a higher color intensity indicates a stronger positive correlation.

for brevity). All of the FTIR and NMR 2D correlation results presented below were generated from reduced-noise spectra. A synchronous spectrum is a symmetric spectrum with respect to a diagonal line. Correlation peaks include autopeak and crosspeak, which appear at both diagonal and off-diagonal positions, respectively. An autopeak represents the overall susceptibility of the corresponding spectral region to change in spectral intensity as an external perturbation is applied to the system. Crosspeaks represent simultaneous or coincidental changes of spectral intensities observed at two different spectral variables. Such a synchronized change, in turn, suggests the possible existence of a coupled or related origin of the spectral intensity variations. An asynchronous spectrum is antisymmetric with respect to the diagonal line, which has no autopeaks and consists exclusively of crosspeaks located at off-diagonal positions. The sign of an asynchronous cross peak can be either 9227

dx.doi.org/10.1021/es201483f |Environ. Sci. Technol. 2011, 45, 9224–9231

Environmental Science & Technology negative or positive. It provides useful information on the sequential order of events observed by the spectroscopic technique along the external variable. The 1800
900 cm
1 region of the 2D FTIR correlation spectra was evaluated because this region of the spectra contains bands corresponding to amides, carboxylic acids, esters, and carbohydrates.32 Time-dependent one-dimensional FTIR spectra during composting of the two piles are shown as Figure S5 of the SI for brevity. The synchronous maps of the biofilms (Figure 3) from the two piles were similar, and three major autopeaks were detected at 1650, 1380, and 1080 cm
1. The greatest change in intensity was observed in the band located at 1650 and 1380 cm
1, followed by the peak at 1080 cm
1. The band at 1650 cm
1 was attributed to amide I in proteinaceous compounds, the band at 1380 cm
1 was assigned to the OH bending vibration of cellulose, and the band at 1080 cm
1 was attributed to the C
O stretching of polysaccharides or polysaccharide-like substances.12,13,32,33 Polysaccharide-like substances are composed of cellulose and hemicellulose. Cellulose is a homopolysaccharide composed of D-glucose units linked to each other via β-1,4-glucosidic bonds; however, hemicellulose is a heteropolysaccharide composed of different sugar units, i.e, mannans, xylans, arabinans, and galactans.34 In this study, polysaccharide-like substances are referred to both homopolysaccharide and heteropolysaccharide, whereas cellulose is assigned to the homopolysaccharide. The above results suggested that proteins and cellulose degraded at a faster rate than polysaccharides during composting. Moreover, three crosspeaks at (1650 and 1380 cm
1), (1650 and 1080 cm
1), and (1380 and 1080 cm
1) were identified. These crosspeaks were positively correlated, suggesting that proteins, cellulose, and polysaccharides varied/degraded concurrently during composting. Compared to the synchronous maps, the asynchronous maps of the biofilms from the two piles displayed distinctive characteristics (Figure 3). In the map of pile 1, three positive crosspeaks were observed at (1690 and 1650 cm
1), (1550 and 1380 cm
1), and (1420 and 1380 cm
1). Moreover, five negative crosspeaks were observed at (1650 and 1550 cm
1), (1650 and 1380 cm
1), (1650 and 1110 cm
1), (1380 and 1110 cm
1), and (1380 and 1250 cm
1). However, in the map of pile 2, three positive crosspeaks were detected at (1650 and 1280 cm
1), (1400 and 1380 cm
1), and (1110 and 1080 cm
1). In addition, four negative crosspeaks were observed at (1650 and 1610 cm
1), (1650 and 1420 cm
1), (1650 and 1110 cm
1), and (1380 and 1110 cm
1) (Figure 3). According to Noda’s rule,18 the following trend in the degradation of peaks was observed during the composting of piles 1 and 2, respectively: 1550, 1420, 1110 cm
1 > 1380 cm
1 > 1650 cm
1 and 1110 cm
1 > 1080, 1420 cm
1 >1650 cm
1 > 1280 cm
1 for piles 1 and 2, respectively. Therefore, organic compounds in piles 1 and 2 were degraded in the following sequence: amide II, heteropolysaccharides > cellulose > amide I and heteropolysaccharides > cellulose > amide I for piles 1 and 2, respectively. In conclusion, cells embedded in the biofilms matrix of compost initially utilize easily degradable heteropolysaccharides. Subsequently, the cells degrade cellulose, followed by proteins. The synchronous map of 2D NMR spectra showed that during composting for the two piles, the greatest degradation of organic compounds was O-alkylated (HCOH) carbons (74 ppm), followed by long chain aliphatic carbons (38 ppm), mirroring the results of 2D FTIR spectra that heteropolysaccharides and

ARTICLE

Figure 4. Synchronous maps obtained via 2D heterospectral correlation analysis of the FTIR and 13C NMR spectra of dissoloved organic matter in the composting piles over time. Red represents positive correlations and blue represents negative correlations; a higher color intensity indicates a stronger positive or negative correlation.

cellulose degradated much more than proteins. Moreover, the asynchronous map of 2D NMR spectra further demonstrated that during composting, long chain aliphatic carbons (38 ppm) degraded prior to O-alkylated (HCOH) carbons (74 ppm). The 2D heterospectral correlation maps were used to examine the covariation between bands in the FTIR and 13C NMR spectra. As shown in Figure 4, the FTIR bands at 1650, 1380, 1080 cm
1 were positively correlated with the NMR band at 38 ppm. In addition, a negative correlation between the three FTIR bands and the NMR band at 74 ppm was observed. Lastly, the FTIR bands at 1650 and 1380 cm
1 were positively correlated with the NMR band at 168 ppm. These results revealed that proteins, cellulose, and heteropolysaccharides in the biofilms consisted of long chain aliphatic carbons rather than O-alkylated (HCOH) carbons. Moreover, proteins and cellulose in the biofilm also contained carboxyl groups, suggesting that O-alkyl carbons were produced during the degradation of long chain aliphatic compounds (i.e., proteins, cellulose, and heteropolysaccharides). These results are supported by those of previous investigations, which showed that the aromatic process performed by microorganisms occurs predominantly in the water-soluble phase.22,24 The fluorescence EEM data (Figure S2 of the SI) also suggested that the extent of aromatic polycondensation conjugated chromophores content and degree of humification increased with an increase in composting time.

’ DISCUSSION Heterogeneity of the Biofilms in Composts. Fluorescence labeling is a valuable tool for assessing the in situ detection of EPS glycoconjugates in undisturbed and fully hydrated and complex environmental biofilms. Although the previous investigations had shown the microscale heterogeneities of bacteria in biofilms,6 few studies are conducted in the compost system. Quantitative 9228

dx.doi.org/10.1021/es201483f |Environ. Sci. Technol. 2011, 45, 9224–9231

Environmental Science & Technology analysis with Image J software clearly demonstrated that the biofilms were approximately 20
100 μm thick (Figure S6 of the SI), suggesting that significant heterogeneity in the biofilm of composts was observed. This result will help to modify the modeling of compost degradation. The individual colonies were observed in the biofilm and penetrated the composts for significant depths. Since the performance of the compost system is closely connected with its characteristic, it is reasonable to suppose that a good biofilm may be developed by adjustment of oxygen and moisture content, which will be beneficial for achieving a good performance of composting. Moreover, the distribution of organic compounds observed by a fluorescence labeling approach could also be applied to explain their degradation patterns. The organic compounds of composts, i.e., proteins, α-polysacchrides, and cellulose, were found to have a distinct distribution pattern, determining that the degradation pattern of them may also be different.1 In this study, during the thermophilic stage, α-polysacchrides had the same distribution pattern with cells, whereas cellulose possessed a similar distribution pattern with cells (Figure 2). However, proteins had a distinct distribution with cells. As a consequence, the heteropolysacchrides and cellulose in compost were degraded prior to proteins (Figure 3). Cells observed during the thermophilic stage were associated with cellulose and α-polysaccharides; thus, these cells were attributed to cellulose- and α-polysaccharides-degrading bacteria rather than protein-degrading bacteria. The results of previous investigations also suggest that the majority of cellulose-degraded bacteria are thermophilic.30 Alternatively, the dead cells were attributable to the poor adaption of mesophilic bacteria to the thermophilic environment. The CLSM observations also revealed that cells were evenly distributed throughout the wheat straw, suggesting that degradation did not occur from the outside. It should be noted that a fluorescence labeling approach depends on the specificity of the selected probes or stains and limits by a lack of understanding of EPS composition and structure.15 Zippel and Neu35 showed that the fluorescence labeling approach is not completely free of uncertainties and the selected probes or stains interact with their target through multiple binding sites increases affinity and specificity, owing to the enormous variety of macromolecules in complex natural microbial biofilms. It has been suggested that determination of “dead cells” by SYTOX blue is questionable. Therefore, investigators should be cautious when they want to apply a fluorescence labeling approach. Degradation of Organic Compounds in Biofilms. Characterizing the chemical and biological changes of organic matter can improve the knowledge of organic matter transformations and maturity assessment during the composting process. In previous investigations, the degradation of organic compounds during composting was often studied by conventional methods. For example, Francou et al. 36 demonstrated that at the thermophilic phase, the hemicellulose fraction varied as cellulose, but with lower contents. Our results also support the codegradation of polysaccharides, cellulose, and proteins during composting by the conventional methods (Figure S7 of the SI), which is consistent with that proteins, cellulose, and heteropolysaccharides varied/degraded concurrently during composting by the synchronous map (Figure 3). Nevertherless, it is difficult to give the sequencing of organic compounds degradation by the conventional methods. Therefore, the findings in this study need to

ARTICLE

be further verified by independent methods in the future investigation. Through 2D FTIR—13C NMR correlation heterospectral spectroscopy, our results for the first time demonstrated that the degradation of organic compounds in biofilm followed the order: heteropolysaccharides > cellulose > amide I in proteins. The degradation sequencing is closely related to the nature of organic compounds. As we all know, cellulose is a semicrystalline polymeric material containing both crystalline and amorphous components, whereas hemicellulose is considered as an amorphous component.34,37 Himmel et al. 37 had shown that crystalline cellulose is resistant to degradation because of the strong interchain hydrogen-bonding network, whereas hemicellulose and amorphous cellulose are readily degradable. M€aki-Arvela et al. 34 also demonstrated that the crystalline structure in cellulose is very stable. Therefore, heteropolysaccharides were degraded prior to cellulose during composting. In this study, the thermophilic phase were attained at the second day for the two piles after composting (Figure 1), in which cellulose—rather than proteins—degraded bacteria should be predominant.30 As a consenquence, proteins were degraded at the last sequencing. However, another investigation (unpublished data) showed that the degradation of organic compounds during composting was related to the distribution of them in materials. Therefore, as for the different materials, the degradation sequencing of organic compounds may be different. Environmental Implications. Although 13C NMR is a powerful approach to investigating functional group variations, it suffers from ambiguities in the structural information it provides. For example, the carbonyl band (CdO) of carboxyl, amide, and aliphatic esters in biofilms all resonate at the same position (around 175 ppm). FTIR can be used to provide an additional view of the functional groups, and can help to resolve the carboxyl, amide, and aliphatic ester contributions to biofilms. Therefore, their complementarity in providing information on the distribution of biofilms functional groups could help to construct a more comprehensive picture of the change in biofilms. The novelty of this study is that we applied, for the first time, two-dimensional correlation spectroscopy to characterize the function of biofilms, which provide many advantages when compared with traditional FTIR or NMR spectroscopy. Knowledge on the composition, architecture, and function of biofilms is essential for understanding biodegradation processes. In the present study, a novel method for the characterization of the composition, architecture, and function of biofilms was developed by combining multiple fluorescence labeling and two-dimensional correlation spectroscopy. Multiple fluorescence labeling supplies structural information on the distribution of biofilm constituents in situ, while two-dimensional correlation spectroscopy provides detailed but locally unresolved information on biofilm constituents. The combination of multiple fluorescence labeling and 2D correlation spectroscopy is a promising approach for the characterization of biofilms. Knowledge on the constituents of biofilm contributes to our understanding of the composting process and provides novel information for engineering applications and scientific research.

’ ASSOCIATED CONTENT

bS

Supporting Information. Detailed descriptions of determination of fluorescence EEM, FTIR, and solid-state 13C NMR spectroscopy; one table listing evolution of Pi,n (%) during

9229

dx.doi.org/10.1021/es201483f |Environ. Sci. Technol. 2011, 45, 9224–9231

Environmental Science & Technology composting achieved from fluorescence regional integrity (FRI) analysis; one figure showing fluorescence EEM contours of composts; two figures showing the CLSM images of biofilms for compost from pile 2 and bright images of composts in pile; one figure showing image analysis results; one figure presenting the degradation of organic matter by the conventional method. This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: +86-25-8439 5212; fax: +86-21-8439 5212; e-mail: [email protected]. Author Contributions ‡

G.H.Y. and Z.T. contributed equally to this work

’ ACKNOWLEDGMENT The work was funded by the National Basic Research Program of China (No. 2011CB100503), the National Natural Science Foundation of China (No. 21007027), Specialized Research Fund for the Doctoral Program of Higher Education (No. 20100097120015), China Postdoctoral Science Foundation (No. 20100481156), the Agricultural Ministry of China (No. 2011-G27 and 201103004), and Key Agricultural Project of Jiangsu Province (SX(2010)220). We would also like to thank three anonymous reviewers for their helpful comments and Dr. David Chadwick from North Wyke Research, U.K. for his careful revision on this manuscript. ’ REFERENCES (1) Yu, G. H.; He, P. J.; Shao, L. M.; Zhu, Y. S. Extracellular proteins, polysaccharides and enzymes impact on sludge aerobic digestion after ultrasonic pretreatment. Water Res. 2008, 42, 1925–1934. (2) Wagner, M.; Ivleva, N. P.; Haisch, C.; Niessner, R. Combined use of confocal laser scanning microscopy (CLSM) and Raman microscopy (RM): Investigations on EPS-matrix. Water Res. 2009, 43, 63–76. (3) Adav, S. S.; Lin, J. C. T.; Yang, Z.; Whiteley, C. G.; Lee, D. J.; Peng, X. F.; Zhang, Z. P. Stereological assessment of extracellular polymeric substances, exo-enzymes, and specific bacterial strains in bioaggregates using fluorescence experiments. Biotechnol. Adv. 2010, 28, 255–280. (4) Dominik, D. M.; Nielsen, J. L.; Nielsen, P. H. Extracellular DNA is abundant and important for microcolony strength in mixed microbial biofilms. Environ. Microbiol. 2010, 13, 710–721. (5) Flemming, H. C.; Neu, T. R.; Wozniak, D. J. The EPS matrix: the “house of biofilm cells. J. Bacteriol. 2007, 189, 7945–7947. (6) Stewart, P. S.; Franklin, M. J. Physiological heterogeneity in biofilms. Nat. Rev. Microbiol. 2008, 6, 199–210. (7) Neu, T. R.; Kuhlicke, U.; Lawrence, J. R. Assessment of fluorochromes for two photon laser scanning microscopy of biofilms. Appl. Environ. Microbiol. 2002, 68, 901–909. (8) Chen, M. Y.; Lee, D. J.; Tay, J. H. Distribution of extracellular polymeric substances in aerobic granules. Appl. Microbiol. Biotechnol. 2007, 73, 1463–1469. (9) Yu, G. H.; Juang, Y. C.; Lee, D. J.; He, P. J.; Shao, L. M. Enhanced aerobic granulation with extracellular polymeric substances (EPS)-free pellets. Bioresour. Technol. 2009, 100, 4611–4615. (10) Yu, G. H.; Lee, D. J.; He, P. J.; Shao, L. M.; Lai, J. Y. Fouling layer with fractionated extracellular polymeric substances of activated sludge. Sep. Sci. Technol. 2010, 45, 993–1002. (11) Garny, K.; Neu, T. R.; Horn, H.; Volke, F.; Manz, B. Combined application of 13C NMR spectroscopy and confocal laser scanning

ARTICLE

microscopy-investigation on biofilm structure and physic-chemical properties. Chem. Eng. Sci. 2010, 65, 4691–4700. (12) Hu, Z. H.; Liu, S. Y.; Yue, Z. B.; Yan, L. F.; Yang, M. T.; Yu, H. Q. Microscale analysis of in vitro anaerobic degradation of lignocellulosic wastes by rumen microorganisms. Environ. Sci. Technol. 2008, 42, 276–281. (13) Cao, B.; Shi, L.; Brown, R. N.; Xiong, Y. J.; Fredrickson, J. K.; Romine, M. F.; Marshall, M. J.; Lipton, M. S.; Beyenal, H. Extracellular polymeric substances from Shewanella sp. HRCR-1 biofilms: Characterization by infrared spectroscopy and proteomics. Environ. Microbiol. 2011, 13, 1018–1031. (14) Yuan, S. J.; Sun, M.; Sheng, G. P.; Li, Y.; Li, W. W.; Yao, R. S.; Yu, H. Q. Identification of key constituents and structure of the extracellular polymeric substances excreted by Bacillus megaterium TF10 for their flocculation capacity. Environ. Sci. Technol. 2011, 45, 1152–1157. (15) Seviour, T.; Lambert, L. K.; Pijuan, M.; Yuan, Z. G. Structural determination of a key exopolysaccharide in mixed culture aerobic sludge granules using NMR spectroscopy. Environ. Sci. Technol. 2010, 44, 8964–8970. (16) Plaza, C.; Senesi, N.; Brunetti, G.; Mondelli, D. Evolution of the fulvic acid fractions during co-composting of olive oil mill wastewater sludge and tree cuttings. Bioresour. Technol. 2007, 98, 1964–1971. (17) Noda, I. Generalized two-dimensional correlation method applicable to infrared, Raman, and other types of spectroscopy. Appl. Spectrosc. 1993, 47, 1329–1336. (18) Noda, I., Ozaki, Y., Eds. Two-Dimensional Correlation Spectroscopy- Applications in Vibrational and Optical Spectroscopy; John Wiley & Sons: England, 2004. (19) Noda, I. Two-dimensional correlation spectroscopy-biannual survey 2007
2009. J. Mol. Struct. 2010, 974, 3–24. (20) Abdulla, H. A.; Minor, E. C.; Dias, R. F.; Hatcher, P. G. Changes in the compound classes of dissolved organic matter along an estuarine transect: A study using FTIR and 13C NMR. Geochim. Cosmochim. Acta 2010, 74, 3815–3838. (21) Gajalakshmi, S.; Abbasi, S. A. Solid waste management by composting: State of the art. Crit. Rev. Environ. Sci. Technol. 2008, 38, 311–400. (22) Yu, G. H.; Luo, Y. H.; Wu, M. J.; Tang, Z.; Liu, D. Y.; Yang, X. M.; Shen, Q. R. PARAFAC modeling of fluorescence excitationemission spectra for rapid assessment of compost maturity. Bioresour. Technol. 2010, 101, 8244–8251. (23) Yu, G. H.; Wu, M. J.; Luo, Y. H.; Yang, X. M.; Ran, W.; Shen, Q. R. Fluorescence excitation-emission spectroscopy with regional integration analysis for assessment of compost maturity. Waste Manage. 2011, 31, 1729–1736. (24) Tang, Z.; Yu, G. H.; Liu, D. Y.; Xu, D. B.; Shen, Q. R. Different analysis techniques for fluorescence excitation-emission matrix spectroscopy to assess compost maturity. Chemosphere 2010, 82, 1202–1208. (25) Bernal, M. P.; Alburquerque, J. A.; Moral, R. Composting of animal manures and chemical criteria for compost maturity assessment: A review. Bioresour. Technol. 2009, 100, 5444–5453. (26) Yu, G. H.; He, P. J.; Shao, L. M.; Lee, D. J. Enzyme activities in activated sludge flocs. Appl. Microbiol. Biotechnol. 2007, 77, 605–612. (27) Jung, Y. M.; Shin, H. S.; Kim, S. B.; Noda, I. New approach to generalized two-dimensional correlation spectroscopy. 1: Combination of principal component analysis and two-dimensional correlation spectroscopy. Appl. Spectrosc. 2002, 56, 1562–1567. (28) Wang, C. M.; Watson, P.; Michel, M. E.; Hoitink, H. A. J. Assessment of the reliability of the solvitaw maturity test for composted manures. Compost Sci. Util. 2003, 11, 125–143. (29) Bustamante, M. A.; Paredes, C.; Marhuenda-Egea, F. C.; PerezEspinosa, A.; Bernal, M. P.; Moral, R. Co-composting of distillery wastes with animal manures: carbon and nitrogen transformations in the evaluation of compost stability. Chemosphere 2008, 72, 551–557. (30) Bergquist, P. L.; Gibbs, M. D.; Morris, D. D.; Te’o, V. S. J.; Saul, D. J.; Morgan, H. W. Molecular diversity of thermophilic cellulolytic and hemicellulolytic bacteria. FEMS Microb. Ecol. 1999, 28, 99–110. 9230

dx.doi.org/10.1021/es201483f |Environ. Sci. Technol. 2011, 45, 9224–9231

Environmental Science & Technology

ARTICLE

(31) Hoitink, H. A. J.; Stone, A. G.; Han, D. Y. Suppression of plant diseases by composts. HortScience 1997, 32, 184–187. (32) Abdulla, H. A. N.; Minor, E. C.; Hatcher, P. G. Using twodimensional correlations of 13C NMR and FTIR to investigate changes in the chemical composition of dissolved organic matter along an estuarine transect. Environ. Sci. Technol. 2010, 44, 8044–8049. (33) Tandy, S.; Healey, J. R.; Nason, M. A.; Williamson, J. C.; Jones, D. L.; Thain, S. C. FT-IR as an alternative method for measuring chemical properties during composting. Bioresour. Technol. 2010, 101, 5431–5436. (34) M€aki-Arvela, P.; Salmi, T.; Holmbom, B.; Willf€or, S.; Murzin, D. Y. Synthesis of sugars by hydrolysis of hemicelluloses—A review. Chem. Rev. 2011, DOI: 10.1021/cr2000042. (35) Zippel, B.; Neu, T. R. Characterization of glycoconjugates of extracellular polymeric substances in tufa-associated biofilms by using fluorescence lectin-binding analysis. Appl. Environ. Microbiol. 2011, 77, 505–516. (36) Francou, C.; Lineres, M.; Derenne, S.; Le Villio-Poitrenaud, M.; Houot, S. Influence of green waste, biowaste and paper-cardboard initial ratios on organic matter transformations during composting. Bioresour. Technol. 2008, 99, 8926–8934. (37) Himmel, M. E.; Ding, S. Y.; Johnson, D. K.; Adney, W. S.; Nimlos, M. R.; Brady, J. W.; Foust, T. D. Biomass recalcitrance: Engineering plants and enzymes for biofuels production. Science 2007, 315, 804–807.

9231

dx.doi.org/10.1021/es201483f |Environ. Sci. Technol. 2011, 45, 9224–9231

ARTICLE pubs.acs.org/est

Environmentally Persistent Free Radicals (EPFRs)-2. Are Free Hydroxyl Radicals Generated in Aqueous Solutions? Lavrent Khachatryan and Barry Dellinger* Louisiana State University, Department of Chemistry, Baton Rouge, Louisiana 70803, United States

bS Supporting Information ABSTRACT: A chemical spin trap, 5,5-dimethyl-1-pyrroline-Noxide (DMPO), in conjunction with electron paramagnetic resonance (EPR) spectroscopy was employed to measure the production of hydroxyl radical ( 3 OH) in aqueous suspensions of 5% Cu(II)O/silica (3.9% Cu) particles containing environmentally persistent free radicals (EPFRs) of 2-monochlorophenol (2MCP). The results indicate: (1) a significant differences in accumulated DMPO
OH adducts between EPFR containing particles and non-EPFR control samples, (2) a strong correlation between the concentration of DMPO
OH adducts and EPFRs per gram of particles, and (3) a slow, constant growth of DMPO
OH concentration over a period of days in solution containing 50 μg/mL EPFRs particles + DMPO (150 mM) + reagent balanced by 200 μL phosphate buffered (pH = 7.4) saline. However, failure to form secondary radicals using standard scavengers, such as ethanol, dimethylsulfoxide, sodium formate, and sodium azide, suggests free hydroxyl radicals may not have been generated in solution. This suggests surface-bound, rather than free, hydroxyl radicals were generated by a surface catalyzed-redox cycle involving both the EPFRs and Cu(II)O. Toxicological studies clearly indicate these bound free radicals promote various types of cardiovascular and pulmonary disease normally attributed to unbound free radicals; however, the exact chemical mechanism deserves further study in light of the implication of formation of bound, rather than free, hydroxyl radicals.

’ INTRODUCTION Stable and relatively nonreactive ‘‘environmentally persistent free radicals (EPFRs)’’ have recently been demonstrated to form in the postflame and cool-zone regions of combustion systems and other thermal processes.1
3 These resonance-stabilized radicals, including semiquinones, phenoxyls, and cyclopentadienyls can be formed by the thermal decomposition of molecular precursors including catechols, hydroquinones, and phenols. Association with the surfaces of fine particles imparts additional stabilization to these radicals such that they can persist almost indefinitely in the environment.2,4 A mechanism of chemisorption and electron transfer from the molecular adsorbate to a redox-active transition metal or other receptor is shown through experiment, and supported by molecular orbital calculations, to result in EPFR formation.2,3,5 Both oxygen-centered and carboncentered EPFRs are possible, the exact structure of which can significantly affect their environmental and biological activity.6,7 An important question is whether EPFRs associated with transition metal oxide-containing nanoparticles can red-ox cycle to generate reactive oxygen species (ROS) such as hydroxyl radicals ( 3 OH), superoxide anion-radicals (O2 3 ‑), and hydrogen peroxide (H2O2) in aqueous media? Information about formation and identification of these ROS has been reported recently.8 A chemical spin trap 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) in conjunction with electron paramagnetic resonance (EPR) r 2011 American Chemical Society

spectroscopy was employed to measure the production of ROS in aqueous suspension of laboratory surrogates of particleassociated EPFRs derived from 2-monochlorophenol (2-MCP) by chemisorption on Cu(II)O/silica particles. The concentration of hydroxyl radicals was measured at ∼1 μM for a 140 min incubation of EPFR-containing solution.8 Hydroxyl radical is one of the most aggressive intermediate species responsible for critical tissue damage and oxidative stress.9
12 However, its high reactivity, and short lifetime may result it in not being able to reach some biological targets. Furthermore, its short half-life, makes direct detection of hydroxyl radical virtually impossible; therefore, indirect detection methods such as EPR, coupled with appropriate spin-trapping agents such as 5,5-dimethyl1-pyrroline-N-oxide (DMPO), has been used.13
17 We provide here evidence of in vitro generation of hydroxyl radical by EPFRs produced from adsorption of 2-monochlorophenol at 230 °C (2-MCP-230) on copper oxide catalyst supported by silica nanoparticles, 5% Cu(II)O/silica (3.9% Cu).3,18 Our results suggest hydroxyl radicals generated at the interface of the particle and solution remain associated with the surface of the particle. Received: May 18, 2011 Accepted: September 26, 2011 Revised: September 26, 2011 Published: September 26, 2011 9232

dx.doi.org/10.1021/es201702q | Environ. Sci. Technol. 2011, 45, 9232–9239

Environmental Science & Technology

’ EXPERIMENTAL SECTION Materials. High purity 5,5-dimethyl-1-pyrroline-N-oxide (DMPO, 99%+, GLC) was obtained from ENZO Life Sciences International and used without further purification. Desferrioxamine, (DFO, assay 92.5%+, TLS); diethylenetriaminepentacetic acid, (DETAPAC, 99%+); L-Ascorbic acid (99%+); β-Nicotinamide Adenine Dinucleotide phosphate, (NADPH, assay, g 95%); sodium formate (BioUltra, g 99%); sodium azide (BioXtra); dimethyl sulfoxide, (DMSO, 99.7+%); 2-monochlorophenol, (2-MCP, 99+%); copper nitrate hemipentahydrate (99.9+%), 0.01 M phosphate buffered saline, (PBS; NaCl 0.138M: KCl 0.0027M), were all obtained from Sigma-Aldrich. Hydrogen peroxide and Cab-O-Sil, as silica powder, were obtained from Fluka (assay, 30%) and Cabot (EH-5, 99+%), respectively. EPFR Surrogate Synthesis. Five % CuO/silica (3.9% Cu), particles were prepared by impregnation of Cab-O-Sil powder with 0.1 M solution of copper nitrate hemipentahydrate and calcinated at 450 °C for 12 h.19 The sample was then ground and sieved (mesh size 230, 63 μm). Prior to exposure, the particles were heated in situ in air to 450 °C for 1 h to pretreat the surface. They were then exposed to saturated vapors of 2-MCP at 230 °C using a custom-made vacuum exposure chamber for 5 min. Once exposure was completed, the temperature of the system was cooled to 150 °C for 1 h at 10
2 Torr. The EPR spectra were then acquired at ambient conditions to confirm the existence of EPFRs. EPR Measurements. EPR spectra were recorded using a Bruker EMX-20/2.7 EPR spectrometer (X-band) with dual cavities, modulation and microwave frequencies 100 kHz and 9.516 GHz, respectively. Typical parameters were as follows: sweep width of 100 G, EPR microwave power of 10 mW, modulation amplitude of 0.8 G, time constant of 40.96 ms, and sweep time of 167.77 s. Values of the g-tensor were calculated using Bruker’s WIN-EPR SimFonia 2.3 program, which allows control of the Bruker EPR spectrometer, data-acquisition, automation routines, tuning, and calibration programs on a Windows-based PC.20 The exact g-values for key spectra were determined by comparison with 2,2-diphenyl-1-picrylhydrazyl (DPPH) standard. ROS Generation Studies. Both control and sample solution suspensions, containing particles without EPFRs (CuO/Silica) and with EPFRs (EPFR/CuO/Silica), respectively, were prepared in similar manners. One mg/mL suspensions of the control, CuO/silica, and sample, EPFR/CuO/silica were prepared in water and saturated with air by bubbling for 5 min. Prior to adding DMPO, the surrogate solutions were sonicated 5 min (Fisher Scientific, FS-20) at 40 W. 0.01 M PBS was used to maintain the pH at 7.4 and balance the final volume at 200 μL. The order of introduction of final solution components to PBS was as follows: particle suspension (10 μL from solution of 1 mg/mL), DMPO (10 μL from a freshly prepared solution of 3 M), reagents (chelators, ascorbic acid, NADPH), and PBS to balance at 200 μL. The final composition of the suspension in most experiments was particles (50 μg/mL), DMPO (150 mM), reagent (200 μL). The solutions were stored in the dark and shaken in touch mode for 30 s using a Vortex Genie 2 (Scientific Industries). Twenty μL of solution was transferred to an EPR capillary tube (i.d. ∼1 mm, o.d. 1.55 mm) and sealed at one end with sealant (Fisherbrand). The capillary was inserted in a 4 mm EPR tube and placed into the EPR resonator.21 The intensities of the EPR spectra of DMPO
OH adducts were reported in arbitrary units, DI/N, (double integrated (DI) intensity of the EPR spectrum normalized (N) to account for the conversion time, receiver gain,

ARTICLE

Scheme 1. Hypothesized Red-Ox Cycle of EPFRs (2-Hydroxyphenoxyl) Originating from 2-MCP Molecule Adsorbed on Cu(II) Domain in Biological System8

number of data points and sweep width.20 Each experiment was performed at least twice, and the reported EPR intensities are an average of all spectra obtained for each experiment. Since the interaction chemistry of chelators with the surface of the model particles is unclear, we abstained from the use of chelators such as DFO, DETAPAC, which minimize the iron content in solution. Chelators have been reported to drastically change the reactivity of particles by affecting the redox potential of metals.22 Adsorbed EPFRs may also undergo enhanced extraction in the presence of metal chelators, based on the metal-chelate complex stability. The oxidizing species formed by the Fentontype reactions can also depend on the nature of the iron chelator.23
25 Nonuse of chelators in this work was also based on the fact that the buffer, prepared in deionized water and treated with Chelex 100 ion-exchange resin (Bio-Rad Laboratories, Hercules, LA) to remove trace heavy metal contaminants,26 did not significantly impact the spin trapping results.

’ RESULTS AND DISCUSSION Our model of formation of EPFRs, reduced metals, and ROS via chemisorption of molecular precursors on metal centers is summarized in Scheme 1.8 The EPFRs formed from 2-MCP adsorbed on CuO/Silica are o-semiquinone (2-hydroxyphenoxyl) and 2-chlorophenoxyl (latter not shown for clarity).3 These EPFRs may red-ox cycle to generate ROS as depicted.8 The average concentration of EPFRs on Cu(II)O/silica was ∼1017 spins/g and exhibited a singlet, structureless EPR spectrum (g = 2.0042, ΔHp-p = 6.5 G). Undosed Cu(II)O/silica particles, which did not contain EPFRs, were used as controls. To establish the optimal conditions for generating DMPO
OH adducts, a series of experiments were initially performed in different solvent solutions (water, dimethylsolfoxide (DMSO), and ethanol (EtOH)) in which the concentrations of spin traps and reductants were varied. Spin Trapping by DMPO. The appearance of the EPR spectrum of DMPO
OH adducts is depicted in Figure 1A. The time evaluation of the DMPO
OH adducts EPR intensity at different DMPO concentrations is represented in Figure 1B. Further incubation resulted in increasing intensity of the DMPO
OH adduct spectrum, and a prominent signal was detected at 180 min. A noisy spectrum was obtained in sample solutions of EPFRs (50 μg/mL) and DMPO (150 mM) in PBS at 2 min of incubation. The 4 lines marked with red asterisks at 2 min incubation correspond to the DMPO
OH adduct (hfsc αN = αH = 14.95 G, 9233

dx.doi.org/10.1021/es201702q |Environ. Sci. Technol. 2011, 45, 9232–9239

Environmental Science & Technology

ARTICLE

Figure 1. (A) Evaluation of EPR spectra of DMPO
OH adducts as a function of incubation time for a solution of EPFRs (50 (μg/mL)), DMPO (150 mM) in PBS. (B) Time dependence of EPR spectral intensity of DMPO
OH adducts as a function of DPMO concentration in solution.

literature data αN = αH = 14.90 G17,27). The other 6 lines are characteristic of a carbon centered species reported in literature and identified as aminoxyl radical formed from the hydroxylamine impurity in DMPO, or formed immediately as high purity DMPO is transferred in an oxygen reach environment.28,29 The EPR intensity of this impurity decreases slowly with time and does not interfere with the measurements of DMPO
OH adduct generation. To provide sufficient EPR intensity for convenient analysis, but avoid potential secondary reactions reactions (Decomposition by light, oxidation by dissolved oxygen, reducing/oxidation, dimerization etc.)16 which may occur at high DMPO concentrations, a final DMPO solution concentration of 150 mM was used in all further studies. The results of nonaeration/aeration on both non-EPFR (control) and EPFR particles solutions are presented in Figure 2A,B. The difference in the DMPO
OH adduct spectral intensity for the sample and control solutions (calculated from the second line at low magnetic field of their respective 4 line spectra) increased with time, most notably at incubation times >150 min. This difference was larger and occurred at earlier times for the aerated solution. The difference in the nonaerated solution was only ∼50% at 1055 min for the nonaerated solution but was ∼100% for the aerated solution at only 220 min. These results confirm involvement of O2 in the redox cycle generating 3 OH, and all further experiments were performed with aerated samples. Dependence of DMPO
OH Adduct Generation on EPFR Concentration. Figure 3A depicts three sets of experimental data for the sample solutions of EPFR-containing particles (50 μg/mL) and DMPO (150 mM) in PBS (total 200 μL) with different initial concentrations of EPFRs (spins/gram). The initial rate of hydroxyl radical generation increased proportionally to the EPFR concentration when it was doubled from 5.56  1016 spins/gram and 1.29  1017 spins/gram. When the EPFR concentration was again approximately doubled to 2.32  1017 spins/gram, there was no increase in the DMPO
OH adduct concentration and the concentration actually decreased at longer incubation times. This is probably due to radical
radical recombination at high EPFR concentrations. The dependence of DMPO
OH adduct concentration on particle concentration in solution (mg of particle/mL of solution) is depicted in Figure 3B. The intensity of DMPO
OH adduct signal exhibited a maximum at a particle concentration of 0.25 mg/mL for concentrations ranging from 0.05 to 1.0 mg/mL. For concentrations >0.25 mg/mL, the DMPO
OH concentration decreased, again, probably due to radical
radical annihilation

Figure 2. (A) The time evolution of the EPR signal intensity of DMPO
OH for the nonaerated control (red) and sample (black) solutions. (B) The time evaluation of the EPR signal intensity of DMPO
OH for the aerated control (red) and sample (black) solutions. The green line represents the time evaluation of the EPR signal intensity of DMPO
OH for pure DMPO (150 mM) in 200 μL PBS solution and indicates no statistical increase in DMPO
OH concentration.

reactions at higher concentrations. Consequently, very dilute suspensions of 50 μg/mL were used in all subsequent experiments. Dependence of DMPO
OH Adduct Generation on EPFR Aging. To determine the lifetimes of EPFRs in aqueous solution and their viability for hydroxyl radical generation, a time evaluation 9234

dx.doi.org/10.1021/es201702q |Environ. Sci. Technol. 2011, 45, 9232–9239

Environmental Science & Technology

ARTICLE

Figure 3. EPR spectral intensity of DMPO
OH as a function of EPFR concentration (spins/g) and incubation time. (A) The initial EPFR concentration on the particles was parametrically varied (radicals/g of particle). (B) The initial particle concentration in solution was varied (mg of particle/mL).

Figure 4. EPR spectral intensity of DMPO
OH adducts as a function of incubation time for solution allowed to age over 1 to 7 days while exposed to air.

of DMPO
OH spectral intensity was performed for particles allowed to age in room air over a period of a week (cf. Figure 4). A stock solution of EPFR-containing particles (50 μg/mL) and DMPO (150 mM) in PBS (total 200 μL) was prepared, and fresh sample solutions were subjected to the spin trapping procedure each day. A marked increase in the DMPO
OH adduct intensity was detected on the third day. Concomitantly, the concentration of the EPFRs on the catalyst surface was measured. For the latter, the solutions were subjected to agitation, decanting, and vacuum drying before the residue solid powder was subjected EPR examination. On the first day, the sample exhibited a singlet EPR spectrum (g = 2.0042. ΔHp-p = 6.5 G) which matched well with the initial spectra before aging. This signal remained strong through the third day, before decaying slowly to barely detectable quantities by the seventh day. This coincided with the maximum in hydroxyl radical generation observed on day three. Thus, the observed hydroxyl radical formation is thought to be mediated, rather than catalyzed, and EPFRs are consumed in the mediated process. However, as proposed in Scheme 1, a biological reducing equivalent is necessary to complete the truly catalytic hydroxyl radical generation cycle. Dependence of DMPO
OH Adduct Generation on Biological Reducing Equivalents. The generation of hydroxyl radical in biological systems is usually enhanced by the presence of H-donors, e.g., NADPH and ascorbates etc.30,31 Up to 1 mM of NADPH or ascorbic acid was added to the solution of EPFR’s

(50 ug/mL) and DMPO (150 mM) in PBS (cf. Supporting Information). The NADPH induced a small effect; while ascorbic acid significantly increased DMPO
OH adduct formation at its lower concentrations (100 μM). However, at higher ascorbic acid concentrations (500 μM), it acted as an effective antioxidant by formation of a characteristic doublet line of ascorbyl radicals. To avoid secondary reactions involving these reductants, their use was minimized in subsequent experiments. Radical Scavengers. Existence of free hydroxyl radicals in biological red-ox systems is usually confirmed using solution, radical scavengers and formation of secondary radicals.32,33 The kinetic competition between DMPO and hydroxyl radical scavengers (ethanol, DMSO, sodium formate, and sodium azide) is used to establish or rule out the presence of free hydroxyl radical. Otherwise, the hydroxyl radical may be bound, or associated with a surface or other molecular species. The effect of scavengers on radical formation were performed in the sample solutions used previously. Inhibition of DMPO
OH adduct formation was observed with addition of 10% (v/v) EtOH at 50 min incubation time (cf. Figure 5A). Thirty minutes later, the intensity was decreased by 20% (DI/N = 20) and then slightly increased. The effect of different ethanol concentrations on inhibition of DMPO
OH adduct formation is presented in Figure 5B. Hydroxyl radicals in the presence of DMPO and ethanol can undergo the competitive reactions: 3 OH

þ DMPO f DMPO
OH

ðreaction1Þ

3 OH

þ CH3 CH2 OH f CH3 3 CHOH þ H2 O ðreaction2Þ

and the new radical CH3 3 CHOH will be trapped by DMPO via: CH3 3 CHOH þ DMPO f DMPO
CHðCH3 ÞOH ðreaction3Þ The resulting DMPO-1-hydroxyethyl adduct exhibits a characteristic 6 line spectrum, vide infra.34,35 If this reaction occurs, then the DMPO
OH signal should decrease as the DMPO
CH(CH3)OH signal increased. This was not observed, suggesting the reaction of ethanol and hydroxyl radical did not occur. DMSO is also a good scavenger of free hydroxyl radicals via reactions similar to Rxns. reaction 1 and reaction 2 for ethanol.34 9235

dx.doi.org/10.1021/es201702q |Environ. Sci. Technol. 2011, 45, 9232–9239

Environmental Science & Technology

ARTICLE

Figure 5. (A) The inhibitory effect of EtOH 10% (v/v) on DMPO
OH EPR signal intensity (DI/N) added after 50 min incubation for a sample solution of EPFRs (50 ug/mL) + DMPO (150 mM) + PBS. (B) Time dependence of DMPO
OH adduct intensity as a function of ethanol concentration (percent, v/v).

Figure 6. Time evaluation of DMPO
OH adduct intensity (DI/N) as a function of DMSO concentration (percent, v/v) in water solution (EPFRs (50 μg/mL) and DMPO (150 mM) in PBS).

At low concentration (2%, v/v), DMSO was a promoter of DMPO
OH formation, while at >10% concentration, it completely inhibited formation, (cf. Figure 6). As in case of ethanol, no concomitant formation of the DMPO
CH3 adduct, 6-line EPR spectrum was detected. In an attempt to generate observable free hydroxyl radical, 20 mM of hydrogen peroxide was added to both the EtOH- and DMSO-containing solutions.36 Generation of hydroxyl radical by H2O2 has been reported in the presence of Cu(II) ions and CuO micrometer-sized catalyst particles.36,37 The Cu(I), present in our EPFR-containing particles (Scheme 1), should also react with the peroxide via Fenton-type reactions to produce hydroxyl radical. The results are summarized in Figure 7 for the EPR spectra of DMPO
OH adduct (blue line). The EPR spectra of DMPO
CH(CH3)
OH adduct (red line), marked by asterisks, was derived from the solution of EPFRs, DMPO, and ethanol (30%,v/v) + H2O2 (20 mM). The observed EPR spectral parameters (hfsc: αN = 15.1 G, αHβ = 23.1G) agreed well with the literature values (αN = 15.8 G, αHβ = 22.8G34). The EPR spectra of DMPO
CH3 adduct (black line), marked by asterisks, was generated from a solution of EPFRs, DMPO and DMSO (10%, v/v) + H2O2 (20 mM). The spectral parameters (hfsc of αN = 16.8 G, αHβ = 23.8G) also agreed well with the literature (αN = 16.4 G, αHβ = 23.4G34). These results suggest free hydroxyl radicals generated by the EPFR particle systems should have been detected in the previously tested solutions if they were present.

Figure 7. EPR spectra of spin adducts generated in different solutions. (1) DMPO
OH in solution of EPFRs (50 μg/mL) and DMPO(150 mM) in PBS + H2O2 (20 mM). (2) Mixture of DMPO
OH and DMPO
CH(CH3)OH in solution of EPFRs (50 μg/mL), DMPO (150 mM), and EtOH (30% v/) in PBS + H2O2 (20 mM). (3) Mixture of DMPO
OH and DMPO
CH3 in a solution of EPFRs (50 μg/mL), DMPO(150 mM), and DMSO (10% v/v) in PBS + H2O2 (20 mM). The background DMPO
OH EPR signal after 5 min without addition of H2O2 was extremely weak (data not shown).

Formate and sodium azide were also used in an attempt to scavenge free hydroxyl radical via the reactions: DMPO

OH þ HCOO
f H2 O þ 3 COO
sf DMPO
COO


ðreaction4Þ 3 OH

DMPO

þ N3
f OH
þ 3 N 3 sf DMPO
N3 ðreaction5Þ

Neither the characteristic 6-line spectrum of DMPO
COO
nor the 12-line spectrum of DMPO-N3 was detected.35 These lines were generated only following addition of H2O2 (20 mM in solution) as a source of free hydroxyl radicals (data not shown). Free versus Bound Hydroxyl Radical. The DMPO spin trapping results indicate hydroxyl radical is being generated. However, the failure of scavenging hydroxyl radicals to form secondary radicals in solution, suggests they are not truly free hydroxyl radicals. Several observations support this contention. 1 . The control solutions, containing Cu(II)O/siica, generate free hydroxyl radical, as evidenced by the scavenger results 9236

dx.doi.org/10.1021/es201702q |Environ. Sci. Technol. 2011, 45, 9232–9239

Environmental Science & Technology

ARTICLE

Figure 8. (A) Time evaluation of DMPO
OH EPR spectral intensity (DI/N) as a function of hydrogen peroxide concentration in a sample solution of EPFRs (50 μg/mL) and DMPO (150 mM) in PBS. (B) The minimum amount of hydrogen peroxide (0.25 mM) at which hydroxyl radical is scavenged by 1.7 M ethanol to result in a detectable 6-line spectrum of DMPO
CH(CH3)OH adduct (marked with asterisks).

(not shown), only when hydrogen peroxide is added. The detection of DMPO
OH adducts without the addition of hydrogen peroxide is likely due to the Cu(II) or Fe(III) impurity, catalyzed nucleophilic addition of water to DMPO.16,17,38 This has been proposed in the literature, but the issue is by no means settled.39
42 The nonradical nucleophilic reaction of water has been proposed to be a significant pathway to the formation of DMPO
OH radical adducts, even during a Fenton reaction,40,41 i.e., 80
90% of the total DMPO
OH in 17O-enriched water was due to irondependent nucleophilic addition of water.41 However, the same authors also discuss a water-independent mechanism of DMPO
OH formation,41 and how Fe or Cu ioninduced nucleophilic addition of water to DMPO may be significantly suppressed in experiments performed in most common buffers.40 2 . The observed DMPO
OH adducts may form due to a secondary mechanism not involving hydroxyl radical trapping.43 DMPO
OH adduct formation has been proposed to be the result of conversion of DMPO
superoxide adduct (DMPO-OOH).34,43
45 However, only 3% of DMPO
OOH adduct has been reported to be converted into DMPO
OH,34 and the concentration of DMPO
OH adduct may be affected only when there is a high concentration of superoxide radicals.46 Other researchers have reported that this conversion does not occur.47 Another secondary mechanism might be oxidation of DMPO through it is cation radical, by addition of water (and elimination of a proton) with ultimate formation of DMPO
OH adduct.48 This, as well as all other possible conversions of DMPO in aqueous solution (oxidation by dissolved oxygen, dimerization, reduction/oxidation, etc.), may occur and every specific case must be considered. The differences we have observed in accumulation of DMPO
OH adducts between the control and sample solutions (cf. Figure 2) and the direct dependency of the intensity of DMPO
OH adducts on EPFR concentration per gram of particle (cf. Figure 3A) are explained by of the activity of EPFRs. Other potential explanations appear to apply to both the samples and controls. 3 Scheme 1 depicts how hydroxyl radicals can be generated by a surface-catalyzed, redox cycle. Our results indicate hydroxyl radical is produced and forms adducts with DMPO, but

the concentration of free hydroxyl radical in solution is too low to be scavenged to form secondary radicals or the rate of reaction with the secondary radicals with DMPO is too slow to be easily detectable. The rate coefficient for reaction of hydroxyl radicals with organics has been reported to be (2.1
5.7)  109 M
1.s
1 for reaction 1 and 1.8  109 M
1. s
1 for reaction 2,49 while the reaction coefficient for secondary 3 CH(CH3)OH radicals, formed by the ethanol scavenger has been reported to be 2 orders of magnitude slower, viz. 4.1  107 M
1.s
1 for reaction 3.50 Using these rate constants it can be easily established that reaction 3 may compete with reaction 1 at a ratio of concentration of secondary to hydroxyl radicals of ∼100. Thus, it appears the secondary radicals cannot compete with hydroxyl radicals to be trapped by DMPO unless the concentration of 3 CH(CH3)OH is much greater than hydroxyl radical. To determine the minimum concentration of hydroxyl radicals that could be effectively detected using DMPO, additional experiments using hydrogen peroxide to generate hydroxyl radical were performed. The ability of sample solutions to generate DMPO
OH adducts at hydrogen peroxide concentrations of 1.5, 0.3, 0.06, and 0.006 mM was determined. The minimum concentration of hydrogen peroxide necessary to generate more DMPO
OH adduct than the EPFR-containing particles during the first 100 min is between 0.006 and 0.06 mM (cf. Figure 8A, blue, pink, and black lines). The question is then at what minimum concentration of hydrogen peroxide, assuming the hydroxyl radicals are free in solution, secondary 3 CH(CH3)OH might be generated in detectable quantities. Thus, the previous experiments were repeated in the presence of a large excess (1.7 M) of ethanol scavenger. The characteristic 6 lines of the DMPO
CH(CH3)OH adduct 34,35 were barely detectable and were only clearly identifiable when the hydrogen peroxide concentration was increased to 0.25 mM and above (cf. Figure 8B). At < 0.25 mM hydrogen peroxide, the scavenging efficiency of ethanol is not high enough to form detectable secondary radicals. Thus, a more effective spin trap for secondary radicals is needed with a scavenging rate coefficient >4.1  107 M
1 3 s
1,50 or a more sensitive EPR method, which can detect less than 10
6 M in solutions,51,52 must be employed. For instance, the detection limit of particle-generated hydroxyl 9237

dx.doi.org/10.1021/es201702q |Environ. Sci. Technol. 2011, 45, 9232–9239

Environmental Science & Technology radicals can be quantified at a concentration of only 50 nM by a fluorescence-based technique using 30 -(p-aminophenyl) fluorescein (APF).53 However, this technique may not have sufficient specificity.44,45,53,54 New EPR spin trapping techniques employing heteroaryl nitrones may be useful since they have been reported to be highly soluble in water, are less sensitive to nucleophilic attack, have long half-lives of the spin adducts, and exhibit high selectivity.55 On the basis of all of these experiments, we believe hydroxyl radicals are generated by the surface-mediated cycle in Scheme 1, with the resulting hydroxyl radicals remaining primarily on the surface such that they cannot be readily scavenged to form secondary organic radicals. This hypothesis is not without experimental or theoretical precedence.56
63 For example, oxidizing metal sites have been proposed to form and trap hydroxyl radicals with reactivities similar (but distinguishable) to those of free hydroxyl radical.64,65 The surface bound hydroxyl radicals have even been suggested of being capable of oxidizing of substrates which are oxidized by free hydroxyl radical.56 In our theory, the combination of the surface-bound hydroxyl radical and the reduced metal in the immediate vicinity are responsible for this enhanced activity of the particles.

’ ASSOCIATED CONTENT

bS Supporting Information. Details of “Dependence of DMPO
OH adduct generation on biological reducing equivalents”, NADPH and Ascorbic acid, Figures S1 and S2. This material is available free of charge via the Internet at http:// pubs.acs.org. ’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT The authors gratefully acknowledge the partial support of this research from NIEHS as part of the LSU Superfund Center under Superfund Research and Training Program Grant P42ES13648. ’ REFERENCES (1) Dellinger, B.; Pryor, W. A.; Ceuto, R.; Squadrito, G. L.; Hedge, V.; Deutsch, W. A. Role of free radicals in the toxicity of airborne fine particulate matter. Chem. Res. Toxicol. 2001, 14, 1371–1377. (2) Dellinger, B.; Lomnicki, S.; Khachatryan, L.; Maskos, Z.; Hall, R.; Adounkpe, J.; McFerrin., C.; Truong, H. Formation and stabilization of persistent free radicals. Proc. Combust. Inst. 2007, 31, 521–528. (3) Lomnicki, S.; Truong, H.; Vejerano, E.; Dellinger, B. Copper oxide-based model of persistent free radical formation on combustionderived particulate matter. Environ. Sci. Technol. 2008, 42 (13), 4982– 4988. (4) Valavanidis, A.; Iopoulos, N.; Gotsis, G.; Fiotakis, K. Persistent free radicals, heavy metals and PAHs generated in particulate soot emissions and residue ash from controlled combustion of common types of plastic. J. Hazard. Mater. 2008, 156 (1
3), 277–284. (5) McFerrin, C. A.; Hall, R. W.; Dellinger, B. Ab Initio study of the formation and degradation reactions of chlorinated phenols. J. Mol. Struct. 2009, 902 (1
3), 5–14. (6) Dellinger, B.; Pryor, W. A.; Cueto, R.; Squadrito, G. L.; Deutsch, W. A. The role of combustion-generated radicals in the toxicity of PM2.5. Proc. Combust. Inst. 2000, 28, 2675–2681.

ARTICLE

(7) Cormier, S. A.; Lomnicki, S.; Backes, W.; Dellinger, B. Origin and health impacts of emissions of toxic by-products and fine particles from combustion and thermal treatment of hazardous wastes and materials. Environ. Health Perspect. 2006, 114, 810–817. (8) Khachatryan, L.; Vejerano, E.; Lomnicki, S.; Dellinger, B. Environmentally Persistent Free Radicals (EPFRs). 1. Generation of Reactive Oxygen Species (ROS) in aqueous solutions. Environ. Sci. Technol. 2011, 45 (19), 8559–8566. (9) Halliwell, B.; Gutteridge, J. M. C. Free Radicals and Metal Ions in Human Disease in Methods in Enzymology. Methods Enzymol. 1990, 186, 1–85. (10) Fridovich, I. Superoxide radical—An endogenous toxicant. Ann. Rev. Pharmacol. Toxicol. 1983, 23, 239–257. (11) Halliwell, B. Oxidative Stress Dis. 2001, 7, 1–16. (12) Zweier, J. L.; Talukder, M. A. H. The role of oxidants and free radicals in reperfusion injury. Cardiovasc. Res. 2006, 70 (2), 181–190. (13) Forshult, S.; Lagercra, C Use of nitroso compounds as scavengers for study of short-lived free radicals in organic reactions. Acta Chem. Scand. 1969, 23 (2), 522–523. (14) Janzen, E. G.; Blackbur, Bj Detection and identification of short-lived free radicals by an electron spin resonance trapping technique. J. Am. Chem. Soc. 1968, 90 (21), 5909–5910. (15) Harbour, J. R.; Chow, V.; Bolton, J. R. Electron-spin resonance study of spin adducts of OH and HO2 radicals with nitrones in ultraviolet photolysis of aqueous hydrogen-peroxide solutions. Can. J. Chem. 1974, 52 (20), 3549–3553. (16) Finkelstein, E.; Rosen, G. M.; Rauckman, E. J. Spin trapping of superoxide and hydroxyl radical—Practical aspects. Arch. Biochem. Biophys. 1980, 200 (1), 1–16. (17) Makino, K.; Hagiwara, T.; Murakami, A. Fundamental-aspects of spin trapping with dmpo. Radiat. Phys. Chem. 1991, 37 (5
6), 657–665. (18) Truong, H., Copper (II) oxide mediated formation and stabilization of combustion generated persistent free radicals. Dissertation for Ph.D. degree 2007, LSU, LA. (19) Truong, H.; Lomnicki, S.; Dellinger, B. Potential for misidentification of environmentally persistent free radicals as molecular pollutants in particulate matter. Environ. Sci. Technol. 2010, 44 (6), 1933–1939. (20) Eaton, G. R.; Eaton, S. S.; Barr, D. P.; Weber, R. T. Quantitative EPR; Springer/Wien: NewYork, Germany,2010; p 185. (21) Nakagawa, K. Is quartz flat cell useful for the detection of superoxide radicals? J. Act. Oxyg. Free Rad. 1994, 5, 81–85. (22) Fubini, B.; Mollo, L.; Giamello, E. Free radical generation at the solid/liquid interface in iron containing minerals. Free Rad. Res. 1995, 23 (6), 593–614. (23) Tomita, M.; Okuyama, T.; Watanabe, S.; Watanabe, H. Quantitation of the hydroxyl radical adducts of salicylic-acid by micellar electrokinetic capillary chromatography-oxidizing species formed by a Fenton reaction. Arch. Toxicol. 1994, 68 (7), 428–433. (24) Yamazaki, I.; Piette, L. H. EPR spin-trapping study on the oxidizing species formed in the reaction of the ferrous ion with hydrogen-peroxide. J. Am. Chem. Soc. 1991, 113 (20), 7588–7593. (25) Zhu, B. Z.; Har-El, R.; Kitrossky, N.; Chevion, M. New modes of action of desferrioxamine: scavenging of semiquinone radical and stimulation of hydrolysis of tetrachlorohydroquinone. Free Radical Biol. Med. 1998, 24 (5), 880–880. (26) Finkelstein, E.; Rosen, G. M.; Rauckman, E. J. Spin trapping— Kinetics of the reaction of superoxide and hydroxyl radicals with nitrones. J. Am. Chem. Soc. 1980, 102 (15), 4994–4999. (27) Harbour, J. R.; Bolton, J. R. Involvement of hydroxyl radical in destructive photo-oxidation of chlorophylls in vivo and in vitro. Photochem. Photobiol. 1978, 28 (2), 231–234. (28) Buettner, G. R.; Oberley, L. W. Considerations in spin trapping of superoxide and hydroxyl radical in aqueous systems using 5,5-dimethyl1-pyrroline-1-oxide. Biochem. Biophys. Res. Commun. 1978, 83 (1), 69–74. (29) Makino, K.; Imaishi, H.; Morinishi, S.; Hagiwara, T.; Takeuchi, T.; Murakami, A.; Nishi, M. An artifact in the esr-spectrum obtained by spin trapping with dmpo. Free Radical Res. Commun. 1989, 6 (1), 19–28. 9238

dx.doi.org/10.1021/es201702q |Environ. Sci. Technol. 2011, 45, 9232–9239

Environmental Science & Technology (30) Alaghmand, M.; Blough, N. V. Source-dependent variation in hydroxyl radical production by airborne particulate matter. Environ. Sci. Technol. 2007, 41 (7), 2364–2370. (31) Briede, J. J.; De Kok, T. M. C. M.; Hogervorst, J. G. F.; Moonen, E. J. C.; Den Camp, C. L. B. O.; Kleinjans, J. C. S. Development and application of an electron spin resonance spectrometry method for the determination of oxygen free radical formation by particulate matter. Environ. Sci. Technol. 2005, 39 (21), 8420–8426. (32) Yim, M. B.; Chock, P. B.; Stadtman, E. R. Copper, zinc superoxide-dismutase catalyzes hydroxyl radical production from hydrogen-peroxide. Proc. Natl. Acad. Sci. U.S.A. 1990, 87 (13), 5006–5010. (33) Buettner, G. R.; Mason, R. P.Critical Reviewes of Oxidative Stress and Aging: Advances in Basic Science, Diagnostics and Intervention. Chapter 2; Cutler, R. G., Rodrigues, H., Eds.; 2003, 2738 (34) Finkelstein, E.; Rosen, G. M.; Rauckman, E. J. Production of Hydroxyl Radical by Decomposition of Superoxide Spin-Trapped Adducts. Mol. Pharmacol. 1982, 21 (2), 262–265. (35) Zhu, B. Z.; Zhao, H. T.; Kalyanaraman, B.; Frei, B. Metalindependent production of hydroxyl radicals by halogenated quinones and hydrogen peroxide: An ESR spin trapping study. Free Radical Biol. Med. 2002, 32 (5), 465–473. (36) Kim, J. K.; Metcalfe, I. S. Investigation of the generation of hydroxyl radicals and their oxidative role in the presence of heterogeneous copper catalysts. Chemosphere 2007, 69 (5), 689–696. (37) Ozawa, T.; Hanaki, A. The 1st ESR spin-trapping evidence for the formation of hydroxyl radical from the reaction of copper(II) complex with hydrogen-peroxide in aqueous-solution. J. Chem. Soc.-Chem. Commun. 1991, 5, 330–332. (38) Burkitt, M. J.; Tsang, S. Y.; Tam, S. C.; Bremner, I. Generation of 5,5-dimethyl-1-pyrroline N-oxide hydroxyl and scavenger radical adducts from copper/H2O2 mixtures—Effects of metal-ion chelation and the search for high-valent metal-oxygen intermediates. Arch. Biochem. Biophys. 1995, 323 (1), 63–70. (39) Makino, K.; Hagiwara, T.; Hagi, A.; Nishi, M.; Murakami, A. Cautionary note for dmpo spin trapping in the presence of iron-ion. Biochem. Biophys. Res. Commun. 1990, 172 (3), 1073–1080. (40) Hanna, P. M.; Chamulitrat, W.; Mason, R. P. When are metal ion-dependent hydroxyl and alkoxyl radical adducts of 5,5-dimethyl-1pyrroline N-Oxide artifacts. Arch. Biochem. Biophys. 1992, 296 (2), 640–644. (41) Chamulitrat, W.; Iwahashi, H.; Kelman, D. J.; Mason, R. P. Evidence against the 1
2-2
1 quartet dmpo spectrum as the radical adduct of the lipid alkoxyl radical. Arch. Biochem. Biophys. 1992, 296 (2), 645–649. (42) Alegria, A. E.; Ferrer, A.; Sepulveda, E. Photochemistry of water-soluble quinones. Production of a water-derived spin adduct. Photochem. Photobiol. 1997, 66 (4), 436–442. (43) Finkelstein, E.; Rosen, G. M.; Rauckman, E. J.; Paxton, J. Spin trapping of superoxide. Mol. Pharmacol. 1979, 16 (2), 676–685. (44) Roubaud, V.; Sankarapandi, S.; Kuppusamy, P.; Tordo, P.; Zweier, J. L. Quantitative measurement of superoxide generation using the spin trap 5-(diethoxyphosphoryl)-5-methyl-1-pyrroline-N-oxide. Anal. Biochem. 1997, 247 (2), 404–411. (45) Bacic, G.; Spasojevic, I.; Secerov, B.; Mojovic, M. Spin-trapping of oxygen free radicals in chemical and biological systems: New traps, radicals and possibilities. Spectrochim. Acta A- Mol. Biomol. Spectrosc. 2008, 69 (5), 1354–1366. (46) Pou, S.; Cohen, M. S.; Britigan, B. E.; Rosen, G. M. Spintrapping and human-neutrophils - limits of detection of hydroxyl radical. J. Biol. Chem. 1989, 264 (21), 12299–12302. (47) Shi, H. L.; Timmins, G.; Monske, M.; Burdick, A.; Kalyanaraman, B.; Liu, Y.; Clement, J. L.; Burchiel, S.; Liu, K. J. Evaluation of spin trapping agents and trapping conditions for detection of cell-generated reactive oxygen species. Arch. Biochem. Biophys. 2005, 437 (1), 59–68. (48) Eberson, L. Adv. Phys. Org. Chem. 1998, 31, 91–141. (49) Dorfman, L. M., and; Adams, G. E., Reactivity of the hydroxyl radical in aqueous solutions. Nat. Stand. Ref. Data Ser., Nat. Bur. Stand. 1973, NSRDS-NBS 46.

ARTICLE

(50) Brustolon, M.; Giamello, E. Electron Paramagnetic Resonance A Practitioner’s Toolkit; Wiley: New Jersey, 2009. (51) Buettner, G. R. Spin trapping—electron-spin-resonance parameters of spin adducts. Free Radical Biol. Med. 1987, 3 (4), 259–303. (52) Harbour, J. R.; Hair, M. L. Transient radicals in heterogeneous systems—detection by spin trapping. Adv. Colloid Interface Sci. 1986, 24 (2
3), 103–141. (53) Cohn, C. A.; Pedigo, C. E.; Hylton, S. N.; Simon, S. R.; Schoonen, M. A. A. Evaluating the use of 3 0 -(p-Aminophenyl) fluorescein for determining the formation of highly reactive oxygen species in particle suspensions. Geochem. Trans. 2009, 10 (8), doi:10.1186/14674866-10-8. (54) Li, B. B.; Gutierrez, P. L.; Blough, N. V. Trace determination of hydroxyl radical in biological systems. Anal. Chem. 1997, 69 (21), 4295–4302. (55) Barriga, G.; Olea-Azar, C.; Norambuena, E.; Castro, A.; Porcal, W.; Gerpe, A.; Gonzalez, M.; Cerecetto, H. New heteroaryl nitrones with spin trap properties: Identification of a 4-furoxanyl derivative with excellent properties to be used in biological systems. Bioorg. Med. Chem. 2010, 18 (2), 795–802. (56) Tojo, S.; Tachikawa, T.; Fujitsuka, M.; Majima, T. Oxidation processes of aromatic sulfides by hydroxyl radicals in colloidal solution of TiO2 during pulse radiolysis. Chem. Phys. Lett. 2004, 384 (4
6), 312–316. (57) Hodgson, E. K.; Fridovich, I. Interaction of Bovine Erythrocyte Superoxide-Dismutase with Hydrogen-Peroxide—Inactivation of Enzyme. Biochemistry 1975, 14 (24), 5294–5299. (58) Hodgson, E. K.; Fridovich, I. Interaction of bovine erythrocyte superoxide-dismutase with hydrogen-peroxide—Chemiluminescence and peroxidation. Biochemistry 1975, 14 (24), 5299–5303. (59) Lawless, D.; Serpone, N.; Meisel, D. Role of OH radicals and trapped holes in photocatalysis—A pulse-radiolysis study. J. Phys. Chem. 1991, 95 (13), 5166–5170. (60) Donaldson, K.; Brown, D. M.; Mitchell, C.; Dineva, M.; Beswick, P. H.; Gilmour, P.; MacNee, W. Free radical activity of PM10: Iron-mediated generation of hydroxyl radicals. Environ. Health Perspect. 1997, 105, 1285–1289. (61) Venkatachari, P.; Hopke, P. K.; Brune, W. H.; Ren, X. R.; Lesher, R.; Mao, J. Q.; Mitchel, M. Characterization of wintertime reactive oxygen species concentrations in Flushing, New York. Aerosol Sci. Technol. 2007, 41 (2), 97–111. (62) Chen, X.; Hopke, P. K.; Carter, W. P. L. Secondary organic aerosol from ozonolysis of biogenic volatile organic compounds: Chamber studies of particle and reactive oxygen species formation. Environ. Sci. Technol. 2011, 45 (1), 276–282. (63) Narayanasamy, J.; Kubicki, J. D. Mechanism of hydroxyl radical generation from a silica surface: Molecular orbital calculations. J. Phys. Chem. B 2005, 109 (46), 21796–21807. (64) Lloyd, R. V.; Hanna, P. M.; Mason, R. P. The origin of the hydroxyl radical oxygen in the Fenton reaction. Free Radical Biol. Med. 1997, 22 (5), 885–888. (65) Borda, M. J.; Elsetinow, A. R.; Strongin, D. R.; Schoonen, M. A. A mechanism for the production of hydroxyl radical at surface defect sites on pyrite. Geochim. Cosmochim. Acta 2003, 67 (5), 935–939.

9239

dx.doi.org/10.1021/es201702q |Environ. Sci. Technol. 2011, 45, 9232–9239

ARTICLE pubs.acs.org/est

Adsorption of Aromatic Carboxylate Ions to Black Carbon (Biochar) Is Accompanied by Proton Exchange with Water Jinzhi Ni,† Joseph J. Pignatello,*,‡ and Baoshan Xing§ †

College of Geographical Sciences, Fujian Normal University, Fuzhou 350007 China Department of Environmental Sciences, Connecticut Agricultural Experiment Station, 123 Huntington Street, P.O. Box 1106, New Haven, Connecticut 06504-1106, United States § Department of Plant, Soil and Insect Sciences, University of Massachusetts, Amherst, Massachusetts 01003, United States ‡

bS Supporting Information ABSTRACT: We examined the adsorption of the allelopathic aromatic acids (AA), cinnamic and coumaric, to different charcoals (biochars) as part of a study on bioavailability of natural signaling chemicals in soil. Sorption isotherms in pH 7 buffer, where the AAs are >99% dissociated, are highly nonlinear, give distribution ratios as high as 104.8 L/ kg, and are insensitive to Ca2+ or Mg2+. In unbuffered media, sorption becomes progressively suppressed with loading and is accompanied by release of OH
with a stoichiometry approaching 1 at low concentrations, declining to about 0.4
0.5 as the pH rises. Sorption of cinnamate on graphite as a model for charcoal was roughly comparable on a surface area basis, but released negligible OH
. A novel scheme is proposed that explains the pH dependence of adsorption and OH
stoichiometry and the graphite results. In a key step, AA
undergoes proton exchange with water. To overcome the unfavorable proton exchange free energy, we suggest AA engages in a type of hydrogen bond recognized to be of unusual strength with a surface carboxylate or phenolate group having a comparable pKa. This bond is depicted as [RCO2 3 3 3 H 3 3 3 O-surf]
. The same is possible for AA
, but results in increased surface charge. The proton exchange pathway appears open to other weak acid adsorbates, including humic substances, on carbonaceous materials.

’ INTRODUCTION The carboxylic acid functional group is abundant in natural soil organic matter and is present in the molecular structures of many natural and synthetic compounds released to soil, including plant exudates, natural signaling chemicals between rhizosphere species, pesticides, and environmental contaminants. Charcoal black carbon is a component of the soil carbon pool as a result of forest fires and deliberate burning practices.1 In addition, interest has emerged in the application of engineered charcoal from biomass waste, known as biochar, to agricultural and forest lands for its potential benefits to soil quality and for its carbon sequestration value.2 Contemplated levels of biochar to croplands and potting soils range from 1 to 10% or more by weight. The effects of natural or added charcoal on chemical and biological processes in the rhizosphere are mostly uncharacterized. A potentially critical property of charcoal with respect to these processes is its surface activity as an adsorbent. The adsorbent strength of charcoal toward organic compounds is a function of the biomass precursor, charring conditions (time and temperature profile, oxygen concentration), degree of postcharring weathering, and other factors that dictate specific surface area, microporosity, and surface chemistry of the final material. Depending on these factors and abundance in soil, charcoal may contribute substantially to sorption, and therefore reduce the physical mobility and biological availability of r 2011 American Chemical Society

contaminants, as well as the above-mentioned natural compound classes. The factors that govern interactions of neutral organic compounds with charcoal and soot are well-known and characterized.3
5 By contrast, the interactions of charcoals with weak organic acids that undergo dissociation within the normal pH range of most soils—most relevantly, carboxylic acids, phenols, and sulfonamides—have received little attention. Sorption of weak acids in soils is a function of pH, ionic strength, surface charge and charge density, type and concentration of metal ions, and in some cases the structural metal ion. Sorption of the neutral molecule is governed by the weak forces available to neutral compounds including van der Waals, hydrogen bonding, and solvophobic effects. Specific interactions of organoanions with minerals and whole soils that have been identified include (i) anion-exchange at positively charged sites; (ii) repulsion with the developing negative charge on the surface as the pH increases above the point of zero net charge (pzc); (iii) bridging by metal cations; and (iv) when chelation is possible, inner-sphere coordination to structural Received: May 31, 2011 Accepted: September 22, 2011 Revised: September 12, 2011 Published: October 14, 2011 9240

dx.doi.org/10.1021/es201859j | Environ. Sci. Technol. 2011, 45, 9240–9248

Environmental Science & Technology metal ions.6
10 Sorption of the organoanion may also involve the above-mentioned weak forces and solvophobic effects, depending on the structure of the rest of the molecule, but solvophobic effects are weaker because of the increased water solubility of the anion relative to the neutral molecule. Sorption of the organoanion in some studies is said to be “negligible”, while in others it is found to be appreciable; for example, polychlorophenolate ions sorb significantly to variable-charge soils even at high pH.6 Although soil organic matter (SOM) is known to be important in the binding of weak acids to whole soils, it has been difficult to separate the influence of SOM from the other components. Binding of carboxylic acids and their anions to SOM has also been studied computationally.11,12 The prior literature on adsorption of weak acids to carbonaceous materials is negligible except in regard to activated carbon.13
15 It is generally found that adsorption decreases with increasing ionization of the molecule as the pH increases above the pzc of the surface due to charge repulsion between the anion with the increasingly negatively charged surface, and to the reduced solvophobic effect of the anion relative to the molecule. However, the anion appears to have appreciable affinity for carbons even under strongly alkaline condition. M€uller et al.13,14 modeled adsorption of weak organic electrolytes (benzoic acid and p-nitrophenol) from aqueous solution by combining electrochemical, diffuse-double-layer, and normal adsorption thermodynamic models. Their model assumes that the affinity of the molecular and ionized forms for the surface are identical except for the charge attraction or repulsion term acting on the ionized form. Thus, at pH values where the surface is net negatively charged, the organoanion would be excluded from the surface unless the nonelectrostatic interaction energy outweighed the electrostatic repulsion energy. Our study was undertaken to characterize the adsorption of selected aromatic acid (AA) allelochemicals by black carbon as part of a broader study on the influence of biochar addition to agricultural fields on chemical signaling in the rhizosphere. We studied sorption of cinnamic and coumaric acids to commercial biochar prototypes. Allelochemicals are low molecular weight compounds secreted into soil by plant tissues and/or decay of plant residues that influence the interaction of plants with other individuals of the same species, other plant species, microbes, viruses, or insects. Allelochemicals play an important role in agricultural and ecological dynamics.16
20 An important class of allelochemicals is the single-ring “phenolic acids” released by many plants that include coumaric, ferulic, caffeic, p-hydroxybenzoic, phenylacetic, salicylic, trans-cinnamic, vanillic, gallic, and syringic acids, among others.19,20 We have identified an important and heretofore unrecognized mechanism of adsorption of organoanions of weak acids on black carbon—namely, proton exchange with water that results in a speciation change on the surface and concomitant release of hydroxide ion into solution. It should be noted that none of the studies above report any change in pH associated with sorption of organoanions.

’ EXPERIMENTAL SECTION

ARTICLE

CQuest) or gently broken up in a mortar and passed through a sieve to obtain the 18.2 MΩ-cm. Surface/pore analysis was conducted by gas porisimetry on an Autosorb-1 (Quantachrome Instruments., Boynton Beach, FL). The outgas temperature was 200 °C. Gas adsorption isotherms were evaluated with the Brunaur
Emmett
Teller (N2 isotherm at 77 K; 11 points) or Grand Canonical Monte Carlo Density Functional Theory (CO2 isotherm at 273 K) models using built-in software to calculate surface areas and pore size distribution. Potentiometric Titration of the Biochars. Biochar (0.4 g for Agrichar and 0.5 g for Soil Reef) was prewetted in 5 mL of nanopure water for 48 h at 20 ( 1 °C with end-over-end mixing at 40 rotations per minute (rpm). Then varying amounts of standard HCl or NaOH solution were added to each sample and to a corresponding blank vial containing the water but no biochar. Preboiled water was used for titration in the alkaline region and the vials were degassed with N2 prior to addition of the NaOH through the septum. The pH was measured after 48 h of mixing at 20 ( 1 °C. The nominal initial H+ or OH
concentration in the sample was calculated from the pH of its corresponding blank. Sorption Experiments. Sorption isotherms were constructed by placing 40 mg of Agrichar or 100 mg of Soil Reef into a 60-mL polytrifluoroethylene (PTFE)-lined screw cap glass vial, along with 50 mL of nanopure water or 0.05 M phosphate buffer (pH 7.0). A parallel set of controls without biochar was set up. Samples and controls without buffer were degassed with N2. After 48 h prewetting, the pH was measured in three sacrificed samples to establish initial pH, and a stock solution of the AA was adjusted to the average pH of the sacrificed samples. This stock solution was used to spike the samples and corresponding controls. The vials were mixed end-over-end at 40 rpm at 20 ( 1 °C for an additional 48 h. The aqueous phase was then sampled and microfiltered (0.45 μm) to remove any biochar. The AA concentration was determined by high-performance liquid chromatography on a C-18 column (S 5 ODS2; phase Sep, Clwyd, U.K.) eluted with 30:70 (v/v) CH3CN/water containing 20 mM acetic acid (pH 3.2) with monitoring at 270 nm for cinnamic acid and 314 nm for coumaric acid. The sorbed concentration was calculated by material balance. In preliminary experiments 48 h appeared sufficient to reach equilibrium. Whereas true equilibrium is difficult to judge, we make the reasonable assumption that trends in sorption observed over the 48-h contact period are representative of trends in any sorption occurring after that time. Isotherms were fit to the Freundlich model (eq 1) and the Langmuir model (eq 2)

Materials. Biochars were generously provided by different manufacturers: Soil Reef by EcoTechnologies Group, LLC, Berwyn, PA; CQuest by Dynamotive Energy Systems Corp., McLean, VA; and Agrichar by BEST Energies Australia, Somersby, Australia. The samples were used either as-received (Agrichar and

S ¼ K F CN S¼ 9241

ð1Þ

Smax L KL C 1 þ KL C

ð2Þ

dx.doi.org/10.1021/es201859j |Environ. Sci. Technol. 2011, 45, 9240–9248

Environmental Science & Technology

Figure 1. Isotherms of (A) cinnamate and (B) coumarate on Agrichar and Soil Reef in phosphate buffer (pH 6.9
7.0) and fits to two sorption models.

where S and C are the sorbed (mg/kg) and solution (mg/L) concentrations, respectively, N is the Freundlich exponent, is the KF is the Freundlich affinity-capacity parameter, Smax L Langmuir capacity parameter, and KL is the Langmuir affinity parameter. The Freundlich parameters were determined by linear regression of log-transformed data, while the Langmuir parameters were determined by nonlinear regression of untransformed data. In both cases the data were weighted by the dependent variable. The distribution ratio, Kd, is defined as S/C at a specified concentration. Sorption experiments to determine stoichiometry were carried out in the same way except using a higher biochar/water ratio (0.4 g for Agrichar, 1.0 g for Soil Reef, and 3.0 g for graphite per 10 mL). Experiments to determine the influence of metal ions on sorption of AA by Agrichar (40 mg of solids and 50 mL of liquid phase) were conduced in a similar manner except for the addition at the prewetting step of CaCl2 or MgCl2 and NaCl to keep ionic strength equal in all vials. A constant mass of AA was added to each vial.

’ RESULTS AND DISCUSSION In screening tests we measured the reduction in solutionphase concentration of AAs after equilibration with increasing

ARTICLE

biochar concentrations in water initially adjusted to pH 5 or 7 with HCl/NaOH (Figure S1, SI). At pH 5, the fraction of cinnamic and coumaric acids in dissociated form is 78.5% and 80.3%, respectively. At pH 7, cinnamic and coumaric acids are >99.7% dissociated. We found that sorption is greater at pH 5 than 7, follows the order Agrichar > Soil Reef . CQuest, and is slightly greater for cinnamic than coumaric acids in all cases at the tested concentrations. Undoubtedly the weak sorbent property of CQuest in comparison to the others is due to the fast pyrolysis method of production, which leaves the material with significant incompletely charred biopolymer and permeated with a greater amount of tarry residue. Sorption isotherms of cinnamic acid and coumaric acid for Agrichar and Soil Reef in phosphate buffer at pH 6.9 are shown in Figure 1 and the model parameters are listed in Table S2. Isotherms on CQuest were not constructed in view of its poor sorbent ability in the screening tests. The isotherms are highly nonlinear even on log scale. Neither the Freundlich nor the Langmuir models proved universally suitable. The order in sorption intensity regardless of liquid phase concentration is Agrichar > Soil Reef. Sorption intensity follows the order cinnamate > coumarate over most of the tested concentration range; the difference is more pronounced for Soil Reef than Agrichar. The trends displayed in the screening tests and the isotherms have conventional explanations. Sorption is greater at pH 5 due to the greater abundance of the molecular form and the lower negative charge of the surface (see below) compared to pH 7.14 Sorption trends qualitatively with the N2 BET of the biochars listed in Table S1: namely, Agrichar (427 m2/g) > Soil Reef (338 m2/g) . CQuest (0.1 m2/g). Sorption also trends with the CO2 GCMC surface area. The order in sorption intensity between the two AAs is plausibly related to solvophobic effects. The octanol
water partition coefficient (Kow) is a commonly used index of solvophobicity. According to SPARC calculator (http://sparc. chem.uga.edu/sparc/; accessed November 17, 2010) the log Kow of the molecular and anionic forms of cinnamic acid are 2.50 and
0.42, respectively, and those of coumaric acid are 1.78 and ∼
1, respectively, consistent with this conclusion. Sorption of the organoanions, reflected in the Kd at pH 6.9, is remarkably strong, however, a fact that is not well-explained by solvophobic effects alone. Depending on concentration, the log Kd for cinnamate on Agrichar ranges 3.7
4.2 and on Soil Reef ranges 3.1
3.8. Likewise, log Kd of coumarate on Agrichar ranges 3.5
4.8 and on Soil Reef ranges 2.6
3.9. The Kd values are thus many orders of magnitude greater than the estimated Kow value of the respective organoanion. This finding seems inconsistent with the sorbed species being the free organoanion. Rather, it implicates either a speciation change or a strong specific interaction of the organoanion on the surface. We next determined the effects of up to 0.1 M Ca2+ and Mg2+ on sorption of the AAs at constant mass of AA
added and ionic strength (Figure S2, SI). We expected that if the anionic form were sorbing, these metal ions would enhance sorption by serving as a cation bridge between the carboxylate group and a negatively charged surface group, such as a carboxylate or phenolate group (e.g., RCO2
3 3 3 M2+ 3 3 3
O2C-BC). The metal may interact with these anions either by contact or solvent-separated ion pairing.23 Cation bridging is an important mechanism triggering the aggregation of humic molecules into larger colloidal structures (NOM) according to molecular dynamics computations.23 Cation bridging also has been 9242

dx.doi.org/10.1021/es201859j |Environ. Sci. Technol. 2011, 45, 9240–9248

Environmental Science & Technology

ARTICLE

Figure 2. Sorption isotherms of cinnamate for biochars comparing buffered (phosphate pH 6.9) and nonbuffered conditions and the accompanying evolution of hydroxide ion concentration. The initial solution composition was 0.005 M CaCl2. The initial nonbuffered pH averaged 7.38 for Agrichar and 7.95 for Soil Reef.

proposed as a mechanism for sorption of carboxylate and phenolate compounds to whole soils,10,6 model soil minerals,9,24 and soil organic matter25 on the basis of physical experiments, as well as to model humic structures on the basis of computations.11,12 Figure S2, however, reveals little, if any, systematic change in sorption induced by Ca2+ and Mg2+. This finding implies that sorption of the AAs is not greatly affected by charged sites under the influence of these metal cations. Figure 2A shows linear-scale plots of the isotherms of cinnamate on Agrichar in phosphate-buffered vs nonbuffered suspensions. At zero concentration of cinnamate the buffered and nonbuffered suspensions had equilibrated during the prewetting stage to a similar pH (6.9 and 7.2, respectively). The isotherms are seen to deviate from one another as AA concentration increases—the nonbuffered samples giving reduced sorption relative to the buffered samples. Moreover, the OH
concentration of the nonbuffered solutions increases relative to the buffered solution as loading increases. Because the AA stock solution was adjusted to the approximate initial concentration of the biochar suspension, vials containing just the aqueous phase showed no significant increase in hydroxide ion concentration with increasing cinnamate concentration up to the same levels added (data not shown). Soil Reef showed results qualitatively similar to those for Agrichar, except the isotherm and [OH
] data are more scattered (Figure 2B). Taken together, the results show that sorption of AA
by biochar is accompanied by the release of hydroxide ion into solution (eq 3), which presumably is the cause of progressive sorption suppression. RCO2
þ BC h ðRCO2
Þ 3 3 3 BC þ OH


ð3Þ

To determine the magnitude of OH
release the buffering capacity of the biochar must be taken into account OH
þ BC h BC
þ H2 O

ð4Þ

At a given pH, the amount of OH
released by AA sorption is the observed amount appearing in solution plus the amount consumed by the biochar at the final pH as determined in an independent titration experiment using the same equilibration period (48 h) and temperature (20 °C) as the sorption experiment. The raw titration curves and the curves representing specific uptake of H+ or OH
versus pH calculated from the raw titration data are provided in Figures S3 and S4, respectively. The crossover pH—where the pH of the sample is equal to the pH of the blank (see Figure S3)—is 8.07 for Agrichar, 7.96 for Soil Reef, and 6.6 for CQuest. Consumption of OH
at any pH above the crossover pH, which represents the biochar’s buffering capacity, follows the order CQuest > Soil Reef > Agrichar. Consumption of H+ at any pH below the crossover pH follows the reverse order. The pH at the pzc is best determined by electrophoretic mobility. The pHpzc for Agrichar is 3.9
4.3 (Table S1), indicating that the net charge on the surface is negative under the conditions of all sorption experiments of this study. This is likely to be true also for Soil Reef because of the similarity in the crossover pH. Quantification of OH
released as a function of AA
sorbed (the stoichiometry) required separate experiments using higher biochar/water ratios than used for constructing the isotherms in Figure 2 in order to obtain greater accuracy in the pH change. Figure 3 shows the results of these experiments. Total moles OH
generated is the observed moles OH
in solution in these sorption experiments plus the moles OH
consumed by the biochar at the same pH in the titration experiments, both after 48 h. Moles OH
consumed by the biochar at each pH was estimated by curve fitting the titration curve in the alkaline region, shown as the curves in Figure S4, and using the fit for interpolation purposes in the sorption experiment. Figure 3 shows that the stoichiometry between OH
and cinnamate sorbed is not constant but decreases with increasing cinnamate loading and/or pH accompanying loading. At the lowest sorbed concentration the OH
/cinnamate molar ratio is 9243

dx.doi.org/10.1021/es201859j |Environ. Sci. Technol. 2011, 45, 9240–9248

Environmental Science & Technology

ARTICLE

Figure 3. Stoichiometry of hydroxide ion release versus moles cinnamic acid adsorbed for (A) Agrichar (per 0.4 g), (B) Soil Reef (per 1.0 g), and (C) graphite (per 3.0 g). The pH record of the blanks (without biochar) shows that adding AA does not contribute appreciably to the increase in [OH
] in solution in the absence of biochar.

approximately 1, while this ratio decreases to about 0.4 (Agrichar) or about 0.5 (Soil Reef). We also tested whether hydroxide is released on adsorption of AA
to nonporous powdered graphite, which we found previously to be a good model for black carbon with respect to adsorption of nonionic compounds.26 Sorption and titration experiments were conducted for cinnamate on graphite in the same way as for the biochars. Not surprisingly, sorption of cinnamate was much weaker on graphite than on Agrichar and Soil Reef on a sorbent mass basis (Figure S3, Table S2), the Kd (L/kg) being >300 times smaller than on Agrichar and >70 times smaller than on Soil Reef. However, on a N2
BET surface area

basis, adsorption of cinnamate was 345 nm). Two additional measures of model robustness, the sum of absolute errors from single peak back-tests and standard deviation of transmittance values among the seven filters were plotted as a function of wavelength (SI Figure S8). Error values were highest above 450 nm with a smaller peak at 290 nm, while transmittance variance was lowest below 300 nm. Irradiance-weighted photodamage spectra show the relative contributions of different wavelengths of light to inactivation under typical sunlight conditions. From SI Table S2 and Figure 4, it can be seen clearly that the sensitivity of PRD1 to longer wavelengths, which are present in much higher intensity in sunlight, results in an overall higher inactivation rate constant compared to MS2. The small peaks in photodamage coefficients observed at 380 nm for MS2 and approximately 420 nm for PRD1 were not found to be significantly different from the baseline in sensitivity analyses, and are estimated to account for less than 5% of inactivation under typical sunlight conditions. Thus, while it is not known whether these peaks are authentic or artifactual, they are likely to be negligible for most applications.

’ DISCUSSION Sensitivity of MS2 and PRD1 to Simulated Sunlight. PRD1 was found to be more sensitive to simulated sunlight than MS2 for all conditions studied, particularly to UVA light, which had

ARTICLE

little effect on MS2. The greater sensitivity of PRD1 is consistent with that phage’s larger genome, and is in agreement with most previous studies,22,25 although Hotze and colleagues26 found that MS2 was more sensitive than PRD1 to UVA light from a fluorescent UVA source. This discrepancy is puzzling, but may be due to differences in experimental methods and in the output spectra of the light sources used. Model-Derived Sensitivity Spectra. Published virus action spectra are typically characterized by peaks around 260 nm where DNA and RNA maximally absorb UV, followed by a steady decline in virus susceptibility up to approximately 300 nm, where most researchers stopped collecting data.27 We studied inactivation and sensitivity from 280 to 500 nm to explore the effects of all likely biocidal sunlight wavelengths on viruses. While the precise mechanisms of inactivation remain unknown, absorption of UVB and UVA photons by the two basic components of viruses, nucleic acids and proteins, may be a critical step in the inactivation of MS2 and PRD1 in PBS.27,28 These excited chromophores may undergo direct photolysis or react in aerobic solutions to form reactive oxygen species that damage other targets.27 A study on the inactivation of MS2 by UVC suggests that nucleic acids may photosensitize damage to proteins;14 such protein-genome interactions might play a similar role in UVBmediated damage. The observation that both MS2 and PRD1 were highly sensitive to the shortest simulated sunlight wavelengths (280
290 nm) is consistent with direct or indirect nucleic acid-sensitized damage. By contrast, the sensitivity peaks identified for both bacteriophages in the 305
310 nm region (Figure3), while similar to a ∼313 nm shoulder in the UV sensitivity of T4 bacteriophage,18do not correspond to known peaks for DNA or RNA absorbance or photodamage. These peaks may represent absorbance by and damage to aromatic amino acids (e.g., tryptophan) or other protein components. Light at 254 nm can damage amino acid residues in the protein capsid of MS214 and UVB light might produce similar damage, affecting viruses’ capsid integrity or their ability to attach to, infect, or replicate within a host. While previous MS2 absorbance spectra did not reveal a peak near 305
310 nm,29 nor did quantum yield data reveal a peak in that range for many viruses of interest,27 neither approach measured virus inactivation in the 305
310 nm region. However, circular dichroism (CD) spectroscopy (a technique that measures protein folding and stability under stress) showed aromatic amino acid activity at 305
310 nm for hepatitis C virus.30 Thus, spectra for photochemical activity and/or virus inactivation may differ from absorbance spectra.31 Sunlight absorption by and damage to viral nucleic acids and proteins should be measured in parallel with loss of infectivity to further elucidate the mechanisms of inactivation. The Role of Photosensitizers. We attempted to eliminate all sensitizers from our experimental solutions, whereas Sinton et al.’s work was performed in river water or seawater spiked with 2
3% (v/v) waste stabilization pond effluent or sewage,19,20 and thus very likely contained significant concentrations of photosensitizers. Our normalized MS2 inactivation rates were far lower than those of Sinton et al.’s F+ RNA coliphage (a family to which MS2 belongs), particularly at longer wavelengths (Figure 2B). Although biological differences may partly explain the variations in spectral response, a more likely explanation is that photons at longer wavelengths were absorbed by photosensitizers in Sinton et al.’s reactors, producing ROS such as singlet oxygen which subsequently damaged the coliphage.15,32,33 Interestingly, the 9253

dx.doi.org/10.1021/es201875x |Environ. Sci. Technol. 2011, 45, 9249–9255

Environmental Science & Technology normalized inactivation rates of PRD1 in our study were in good agreement with the rates for somatic coliphage reported by Sinton and colleagues (Figure 2A). This agreement may be coincidental, or may indicate similar spectral sensitivity of PRD1 and somatic coliphage to sunlight, and may likewise indicate that exogenous photosensitizers did not play a significant role in the inactivation of the latter variety of somatic coliphages. It should be noted that somatic coliphage are a diverse group, and variable response to sunlight has been documented in field isolates.22 Sensitivity Analysis. The results of the sensitivity analysis (SI Figure S3) and model back-testing (SI Figures S4
S8) suggest that the computational model produced reasonable estimates of virus sensitivity to simulated sunlight over the 285
345 nm range, and that the spectral sensitivity peaks observed at wavelengths 5000 would have been necessary (at the expense of considerable CPU time of ∼4 weeks) to achieve an accurate stationary solution, since the size fluctuation is large at high IS due to a significant degree of settling. According to the DLVO theory, agglomerate sizes would also increase with increasing magnitude of the Hamaker constant, as this would imply increasing van der Waals attraction energy (SI, Table S1, eq S1). For the above TiO2 example, the simulations reveal a linear dependence of agglomerate nanoparticle size on the Hamaker constant (AH) demonstrating an agglomerate size increase from 110 nm to 550 nm with AH increase from 10 zJ to 90 zJ (SI, Figure S7), corresponding to the range of literature reported AH values for the anatase62,63 and rutile63,64 forms of TiO2. The simulations revealed that the agglomerate size of nanoparticles in aqueous suspension increases with decreasing primary particle diameter (Figure 6). This result should not be surprising since van der Waals interactions increase with decreasing particle size (SI, eq S1), the collision frequency is more pronounced with smaller particles (eq 4) and electrostatic repulsion increases with increased particle size (SI, eqs S2 and S3). As a result, smaller primary nanoparticles will form larger agglomerates. It is noted, however, that DLS measurements are only indicative of the size of particles remaining in suspensions and do not provide a measure of the true distribution of all

Figure 5. Average agglomerate diameter of nanoparticle aggregates after 24 h as a function of (left) pH levels (at IS = 0.037 mM) and (right) ionic strength. Simulation conditions: AH = 42 zJ, dp = 21 nm, Co = 20 mg L
1, temperature = 23
C. Note: the vertical bars represent one standard deviation over 10 simulation replicates. 9289

dx.doi.org/10.1021/es202134p |Environ. Sci. Technol. 2011, 45, 9284–9292

Environmental Science & Technology

ARTICLE

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT This work was supported, in part, by the National Science Foundation and the Environmental Protection Agency under Cooperative Agreement Number DBI 0830117, the UCLA Water Technology Research Center and the California Department of Water Resources. Any opinions, findings, conclusions or recommendations expressed herein are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or the Environmental Protection Agency. This work has not been subjected to an EPA peer and policy review. We also thank Zhaoxia Ji for assistance in obtaining the TEM images. Figure 6. Dependence of average agglomerate diameter on the primary nanoparticle diameter. Simulation condition: IS = 0.5 mM, Co = 20 mg L
1, T = 23
C.

agglomerates that may have formed. Accordingly one would expect that, as a result of agglomeration and sedimentation, the nanoparticle size distribution in suspension (as determined by DLS) will reveal an increasing tail of smaller size agglomerates with increasing primary particle diameter (Figure 6; SI, Figure S8). The above behavior should not be taken as a universal representation of nanoparticle agglomeration as one must be cautious with the limitation of the DLVO theory to nanoparticles with kr , 1 (e.g., corresponding to ri , 10 nm at IS = 1 mM; note that k is the inverse Debye length, SI, Table S1). Moreover, it is also noted that the present application of the DLVO theory, as well as the simple application of gravitational settling, does not consider the impact of the details of agglomerate geometry and morphology. Nonetheless, the present analysis suggests that interpretation of nanoparticle behavior in environmental aquatic media and potential toxic outcomes due to exposure to nanoparticles must carefully consider not only the particle size distribution but also the experimental protocols used to determine such size distributions. In summary, the present constant-number DSMC approach of simulating nanoparticle agglomeration in aqueous suspensions demonstrated that classical DLVO theory can provide reasonably accurate predictions of the average nanoparticle agglomerate size as well as the particle size distribution over a wide range of solution pH (3
10) and ionic strength (0.01
156 mM; SI, Table S2). Extension of the present approach using the extended DLVO theory is presently underway in order to explore a wide range of nanoparticle types aqueous solution chemistries of environmental interest.

’ ASSOCIATED CONTENT

bS

Supporting Information. Additional information is included in the supporting information regarding figures of the box expansion approach, particle pair selection method, added figures of experimental and simulation results, calculation of sedimentation distances, TEM images of nanoparticles, list of DLVO working equations, and table of experimental and simulation conditions. This material is available free of charge via the Internet at http://pubs.acs.org/.

’ REFERENCES (1) Guo, Z.; Tan, L. Fundamentals and Applications of Nanomaterials, 1st ed.; Artech House Publishers: Norwood, MA, 2009. (2) Klaine, S. J.; Alvarez, P. J. J.; Batley, G. E.; Fernandes, T. F.; Handy, R. D.; Lyon, D. Y.; Mahendra, S.; McLaughlin, M. J.; Lead, J. R. Nanomaterials in the environment: Behavior, fate, bioavailability, and effects. Environ. Toxicol. Chem. 2008, 27 (9), 1825–1851. (3) Handy, R. D.; Owen, R.; Valsami-Jones, E. The ecotoxicology of nanoparticles and nanomaterials: Current status, knowledge gaps, challenges, and future needs. Ecotoxicology 2008, 17 (5), 315–325. (4) Wiesner, M. R.; Lowry, G. V.; Alvarez, P.; Dionysiou, D.; Biswas, P. Assessing the risks of manufactured nanomaterials. Environ. Sci. Technol. 2006, 40 (14), 4336–4345. (5) The Project on Emerging Nanotechnologies: Consumer Products Inventory (Woodrow Wilson International Center). http://www. nanotechproject.org/inventories/consumer/ (accessed 8/29/2011). (6) Farre, M.; Gajda-Schrantz, K.; Kantiani, L.; Barcelo, D. Ecotoxicity and analysis of nanomaterials in the aquatic environment. Anal. Bioanal. Chem. 2009, 393 (1), 81–95. (7) Barnard, A. S. Computational strategies for predicting the potential risks associated with nanotechnology. Nanoscale 2009, 1 (1), 89–95. (8) Stone, V.; Nowack, B.; Baun, A.; van den Brink, N.; von der Kammer, F.; Dusinska, M.; Handy, R.; Hankin, S.; Hassell€ov, M.; Joner, E.; Fernandes, T. F. Nanomaterials for environmental studies: Classification, reference material issues, and strategies for physico-chemical characterisation. Sci. Total Environ. 2010, 408 (7), 1745–1754. (9) Kahru, A.; Dubourguier, H.-C. From ecotoxicology to nanoecotoxicology. Toxicology 2010, 269 (2
3), 105–119. (10) Biswas, P.; Wu, C. Y. Critical Review: Nanoparticles and the environment. J. Air Waste Manage. Assoc. 2005, 55 (6), 708–746. (11) Colvin, V. L. The potential environmental impact of engineered nanomaterials. Nat. Biotechnol. 2003, 21 (10), 1166–1170. (12) Petosa, A. R.; Jaisi, D. P.; Quevedo, I. R.; Elimelech, M.; Tufenkji, N. Aggregation and deposition of engineered nanomaterials in aquatic environments: Role of physicochemical interactions. Environ. Sci. Technol. 2010, 44 (17), 6532–6549. (13) Long, T. C.; Saleh, N.; Tilton, R. D.; Lowry, G. V.; Veronesi, B. Titanium dioxide (P25) produces reactive oxygen species in immortalized brain microglia (BV2): Implications for nanoparticle neurotoxicity. Environ. Sci. Technol. 2006, 40 (14), 4346–4352. (14) Kajihara, M. Settling velocity and porosity of large suspended particle. J. Oceanogr. 1971, 27 (4), 158–162. (15) Elimelech, M. Particle Deposition and Aggregation: Measurement, Modelling, And Simulation. Butterworth-Heinemann: Boston, 1995; p xv. (16) Areepitak, T.; Ren, J. Model simulations of particle aggregation effect on colloid exchange between streams and streambeds. Environ. Sci. Technol. 2011null-null. 9290

dx.doi.org/10.1021/es202134p |Environ. Sci. Technol. 2011, 45, 9284–9292

Environmental Science & Technology (17) Sharma, V. K. Aggregation and toxicity of titanium dioxide nanoparticles in aquatic environment—A review. J. Environ. Sci. Health, Part A: Environ. Sci. Eng. 2009, 44 (14), 1485–1495. (18) Zhang, Y.; Chen, Y.; Westerhoff, P.; Hristovski, K.; Crittenden, J. C. Stability of commercial metal oxide nanoparticles in water. Water Res. 2008, 42 (8
9), 2204–2212. (19) Guzman, K. A. D.; Finnegan, M. P.; Banfield, J. F. Influence of sufface potential on aggregation and transport of titania nanoparticles. Environ. Sci. Technol. 2006, 40 (24), 7688–7693. (20) Keller, A. A.; Wang, H. T.; Zhou, D. X.; Lenihan, H. S.; Cherr, G.; Cardinale, B. J.; Miller, R.; Ji, Z. X. Stability and aggregation of metal oxide nanoparticles in natural aqueous matrices. Environ. Sci. Technol. 2010, 44 (6), 1962–1967. (21) Ji, Z.; Jin, X.; George, S.; Xia, T.; Meng, H.; Wang, X.; Suarez, E.; Zhang, H.; Hoek, E. M. V.; Godwin, H.; Nel, A. E.; Zink, J. I. Dispersion and stability optimization of TiO2 nanoparticles in cell culture media. Environ. Sci. Technol. 2010, 44 (19), 7309–7314. (22) Jiang, J. K.; Oberdorster, G.; Biswas, P. Characterization of size, surface charge, and agglomeration state of nanoparticle dispersions for toxicological studies. J. Nanopart. Res. 2009, 11 (1), 77–89. (23) Baveye, P.; Laba, M. Aggregation and toxicology of titanium dioxide nanoparticles. Environ. Health Perspect. 2008, 116 (4), A152–A152. (24) French, R. A.; Jacobson, A. R.; Kim, B.; Isley, S. L.; Penn, R. L.; Baveye, P. C. Influence of Ionic strength, pH, and cation valence on aggregation kinetics of titanium dioxide nanoparticles. Environ. Sci. Technol. 2009, 43 (5), 1354–1359. (25) Baalousha, M. Aggregation and disaggregation of iron oxide nanoparticles: Influence of particle concentration, pH and natural organic matter. Sci. Total Environ. 2009, 407 (6), 2093–2101. (26) Thio, B. J. R.; Zhou, D. X.; Keller, A. A. Influence of natural organic matter on the aggregation and deposition of titanium dioxide nanoparticles. J. Hazard. Mater. 2011, 189 (1
2), 556–563. (27) He, Y. T.; Wan, J. M.; Tokunaga, T. Kinetic stability of hematite nanoparticles: The effect of particle sizes. J. Nanopart. Res. 2008, 10 (2), 321–332. (28) Huynh, K. A.; Chen, K. L. Aggregation kinetics of citrate and polyvinylpyrrolidone coated silver nanoparticles in monovalent and divalent electrolyte solutions. Environ. Sci. Technol. 2011, 45 (13), 5564–5571. (29) Buettner, K. M.; Rinciog, C. I.; Mylon, S. E. Aggregation kinetics of cerium oxide nanoparticles in monovalent and divalent electrolytes. Colloids Surf., A 2010, 366 (1
3), 74–79. (30) Bos, R.; van der Mei, H. C.; Busscher, H. J. Physico-chemistry of initial microbial adhesive interactions
its mechanisms and methods for study. FEMS Microbiol. Rev. 1999, 23 (2), 179–230. (31) Grasso, D.; Subramaniam, K.; Butkus, M.; Strevett, K.; Bergendahl, J. A review of non-DLVO interactions in environmental colloidal systems. Rev. Environ. Sci. Bio/Technol. 2002, 1 (1), 17–38. (32) Hermansson, M. The DLVO theory in microbial adhesion. Colloids Surf., B 1999, 14 (1
4), 105–119. (33) Hong, Y. S.; Honda, R. J.; Myung, N. V.; Walker, S. L. Transport of iron-based nanoparticles: Role of magnetic properties. Environ. Sci. Technol. 2009, 43 (23), 8834–8839. (34) Honig, E. P.; Roeberse., Gj; Wiersema, P. H. Effect of hydrodynamic interaction on coagulation rate of hydrophobic colloids. J. Colloid Interface Sci. 1971, 36 (1), 97–109. (35) Alimohammadi, M.; Fichthorn, K. A. Molecular dynamics simulation of the aggregation of titanium dioxide nanocrystals: Preferential alignment. Nano Lett. 2009, 9 (12), 4198–4203. (36) Maisels, A.; Kruis, F. E.; Fissan, H. Direct simulation Monte Carlo for simultaneous nucleation, coagulation, and surface growth in dispersed systems. Chem. Eng. Sci. 2004, 59 (11), 2231–2239. (37) Markutsya, S.; Subramaniam, S.; Vigil, R. D.; Fox, R. O. On Brownian dynamics simulation of nanoparticle aggregation. Ind. Eng. Chem. Res. 2008, 47 (10), 3338–3345. (38) Zhao, H.; Maisels, A.; Matsoukas, T.; Zheng, C. Analysis of four Monte Carlo methods for the solution of population balances in dispersed systems. Powder Technology 2007, 173 (1), 38–50.

ARTICLE

(39) Peng, Z. B.; Doroodchi, E.; Evans, G. DEM simulation of aggregation of suspended nanoparticles. Powder Technol. 2010, 204 (1), 91–102. (40) Jiang, W. T.; Ding, G. L.; Peng, H.; Hu, H. T. Modeling of nanoparticles’ aggregation and sedimentation in nanofluid. Curr. Appl. Phys. 2010, 10 (3), 934–941. (41) Lin, Y. L.; Lee, K.; Matsoukas, T. Solution of the population balance equation using constant-number Monte Carlo. Chem. Eng. Sci. 2002, 57 (12), 2241–2252. (42) Smith, M.; Matsoukas, T. Constant-number Monte Carlo simulation of population balances. Chem. Eng. Sci. 1998, 53 (9), 1777–1786. (43) Lee, K.; Matsoukas, T. Simultaneous coagulation and break-up using constant-N Monte Carlo. Powder Technol. 2000, 110 (1
2), 82–89. (44) Kim, T.; Lee, C. H.; Joo, S. W.; Lee, K. Kinetics of gold nanoparticle aggregation: Experiments and modeling. J. Colloid Interface Sci. 2008, 318 (2), 238–243. (45) Kruis, F. E.; Maisels, A.; Fissan, H. Direct simulation Monte Carlo method for particle coagulation and aggregation. AIChE J. 2000, 46 (9), 1735–1742. (46) Zhao, H. B.; Zheng, C. G.; Xu, M. H. Multi-Monte Carlo method for particle coagulation: Description and validation. Appl. Math. Comput. 2005, 167 (2), 1383–1399. (47) Friedlander, S. K. Smoke, Dust, And Haze: Fundamentals of Aerosol Dynamics, 2nd ed.; Oxford University Press: New York, 2000; p xx. (48) Kruyt, H. R. Colloid Science; Elsevier: New York, 1949. (49) Liu, R.; Rallo, R.; George, S.; Ji, Z. X.; Nair, S.; Nel, A. E.; Cohen, Y. Classification nanoSAR development for cytotoxicity of metal oxide nanoparticles. Small 2011, 7 (8), 1118–1126. (50) Rallo, R.; France, B.; Liu, R.; Nair, S.; George, S.; Damoiseaux, R.; Giralt, F.; Nel, A.; Bradley, K.; Cohen, Y. Self-organizing map analysis of toxicity-related cell signaling pathways for metal and metal oxide nanoparticles. Environ. Sci. Technol. 2011, 45 (4), 1695–1702. (51) Allouni, Z. E.; Cimpan, M. R.; Høl, P. J.; Skodvin, T.; Gjerdet, N. R. Agglomeration and sedimentation of TiO2 nanoparticles in cell culture medium. Colloids Surf., B 2009, 68 (1), 83–87. (52) Tiraferri, A.; Chen, K. L.; Sethi, R.; Elimelech, M. Reduced aggregation and sedimentation of zero-valent iron nanoparticles in the presence of guar gum. J. Colloid Interface Sci. 2008, 324 (1
2), 71–79. (53) Fedele, L.; Colla, L.; Bobbo, S.; Barison, S.; Agresti, F. Experimental stability analysis of different water-based nanofluids. Nanoscale Res. Lett. 2011, 6 (1), 300. (54) Li, X. F.; Zhu, D. S.; Wang, X. J. Evaluation on dispersion behavior of the aqueous copper nano-suspensions. J. Colloid Interface Sci. 2007, 310 (2), 456–463. (55) Hunter, R. J. Foundations of Colloid Science, 2nd ed.; Oxford University Press: New York, 2001; p xii. (56) Filella, M.; Zhang, J. W.; Newman, M. E.; Buffle, J. Analytical applications of photon correlation spectroscopy for size distribution measurements of natural colloidal suspensions: Capabilities and limitations. Colloids Surf., A 1997, 120 (1
3), 27–46. (57) Chen, K. J.; Wolahan, S. M.; Wang, H.; Hsu, C. H.; Chang, H. W.; Durazo, A.; Hwang, L. P.; Garcia, M. A.; Jiang, Z. K.; Wu, L.; Lin, Y. Y.; Tseng, H. R. A small MRI contrast agent library of gadolinium(III)-encapsulated supramolecular nanoparticles for improved relaxivity and sensitivity. Biomaterials 2011, 32 (8), 2160–2165. (58) Vold, R. D.; Vold, M. J. Colloid and Interface Chemistry; Addison-Wesley: Reading, MA, 1983; p xxv. (59) Schwarzer, H. C.; Peukert, W. Prediction of aggregation kinetics based on surface properties of nanoparticles. Chem. Eng. Sci. 2005, 60 (1), 11–25. (60) Brant, J.; Lecoanet, H.; Wiesner, M. R. Aggregation and deposition characteristics of fullerene nanoparticles in aqueous systems. J. Nanopart. Res. 2005, 7 (4
5), 545–553. (61) Chen, K. L.; Elimelech, M. Relating colloidal stability of fullerene (C60) nanoparticles to nanoparticle charge and electrokinetic properties. Environ. Sci. Technol. 2009, 43 (19), 7270–7276. (62) Gomez-Merino, A. L.; Rubio-Hernandez, F. J.; VelazquezNavarro, J. F.; Galindo-Rosales, F. J.; Fortes-Quesada, P. The Hamaker 9291

dx.doi.org/10.1021/es202134p |Environ. Sci. Technol. 2011, 45, 9284–9292

Environmental Science & Technology

ARTICLE

constant of anatase aqueous suspensions. J. Colloid Interface Sci. 2007, 316 (2), 451–456. (63) Visser, J. On Hamaker constants: A comparison between Hamaker constants and Lifshitz-van der Waals constants. Adv. Colloid Interface Sci. 1972, 3 (4), 331–363. (64) Ackler, H. D.; French, R. H.; Chiang, Y. M. Comparisons of Hamaker constants for ceramic systems with intervening vacuum or water: From force laws and physical properties. J. Colloid Interface Sci. 1996, 179 (2), 460–469.

9292

dx.doi.org/10.1021/es202134p |Environ. Sci. Technol. 2011, 45, 9284–9292

ARTICLE pubs.acs.org/est

Impact Assessment of Ammonia Emissions on Inorganic Aerosols in East China Using Response Surface Modeling Technique Shuxiao Wang,*,† Jia Xing,† Carey Jang,‡ Yun Zhu,§ Joshua S. Fu,|| and Jiming Hao† †

)

School of Environment, and State Key Joint Laboratory of Environment Simulation and Pollution Control, Tsinghua University, Beijing 100084, P. R. China ‡ U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, United States § School of Environmental Science and Engineering, South China University of Technology, Guangzhou 510006, P. R. China Department of Civil and Environmental Engineering, University of Tennessee, Knoxville, Tennessee 37996, United States

bS Supporting Information ABSTRACT: Ammonia (NH3) is one important precursor of inorganic fine particles; however, knowledge of the impacts of NH3 emissions on aerosol formation in China is very limited. In this study, we have developed China’s NH3 emission inventory for 2005 and applied the Response Surface Modeling (RSM) technique upon a widely used regional air quality model, the Community Multi-Scale Air Quality Model (CMAQ). The purpose was to analyze the impacts of NH3 emissions on fine particles for January, April, July, and October over east China, especially those most developed regions including the North China Plain (NCP), Yangtze River delta (YRD), and the Pearl River delta (PRD). The results indicate that NH3 emissions contribute to 8
11% of PM2.5 concentrations in these three regions, comparable with the contributions of SO2 (9
11%) and NOx (5
11%) emissions. However, NH3, SO2, and NOx emissions present significant nonlinear impacts; the PM2.5 responses to their emissions increase when more control efforts are taken mainly because of the transition between NH3-rich and NH3-poor conditions. Nitrate aerosol (NO3
) concentration is more sensitive to NOx emissions in NCP and YRD because of the abundant NH3 emissions in the two regions, but it is equally or even more sensitive to NH3 emissions in the PRD. In high NO3
pollution areas such as NCP and YRD, NH3 is sufficiently abundant to neutralize extra nitric acid produced by an additional 25% of NOx emissions. The 90% increase of NH3 emissions during 1990
2005 resulted in about 50
60% increases of NO3
and SO42‑ aerosol concentrations. If no control measures are taken for NH3 emissions, NO3
will be further enhanced in the future. Control of NH3 emissions in winter, spring, and fall will benefit PM2.5 reduction for most regions. However, to improve regional air quality and avoid exacerbating the acidity of aerosols, a more effective pathway is to adopt a multipollutant strategy to control NH3 emissions in parallel with current SO2 and NOx controls in China.

’ INTRODUCTION The importance of ammonia (NH3) in contributing to secondary inorganic aerosols (SIA, i.e., sulfate (SO42
), nitrate (NO3
), and ammonium (NH4+)) has been well documented in recent studies. Excess NH3 provides a weak base, which allows a larger aqueous uptake of sulfur dioxide (SO2) to be oxidized and, at the same time, also affects the effective cloud SO2 oxidation rate due to strong pH-dependent oxidation by ozone (O3).1,2 In the presence of NH3, NO3
is formed by the gas-to-particle conversion process from nitric acid (HNO3) which was first produced through a photochemical reaction as nitrogen dioxide (NO2) and hydroxyl radical (•OH). Multisensitivity studies for European countries and the United States2
9 have been conducted using air quality models (AQMs) to explore the response of inorganic fine particles to emission changes of SO2, nitrogen oxides (NOx = NO + NO2), NH3, or nonmethane volatile r 2011 American Chemical Society

organic compounds (NMVOC). Derwent et al.9 used a moving air parcel trajectory model to estimate the mass concentrations of PM components for a rural location in the southern UK, and found that PM mass concentrations are nonlinear with PM precursor emissions, and suggested that abatement of NH3 emissions should be considered to obtain the largest PM2.5 reduction. Tsimpidi et al.2 applied a three-dimensional chemical transport model (PMCAMx) to investigate the changes in PM2.5 concentrations responding to changes of SO2 and NH3 emissions in the eastern United States, and indicated that coupled reductions of SO2 and NH3 emissions are more effective than the Received: June 30, 2011 Accepted: September 22, 2011 Revised: September 8, 2011 Published: September 22, 2011 9293

dx.doi.org/10.1021/es2022347 | Environ. Sci. Technol. 2011, 45, 9293–9300

Environmental Science & Technology

Figure 1. Map of the CMAQ/RSM modeling domain and spatial distributions of NH3 emissions.

control of individual pollutants. Pinder et al.6 conducted a series of PMCAMx simulations to estimate the cost-effectiveness and uncertainty of NH3 emission reductions on inorganic aerosols in the eastern United States and found that many currently available NH3 control technologies were cost-effective compared to SO2 and NOx. China, as the most populated country in the world, has significant agricultural activities which release large amounts of NH3 to the atmosphere. Enhanced concentrations of NH3 over the Beijing area in northeast China have been first detected in space-based nadir viewing measurements that penetrate into the lower atmosphere.10 The North China Plain (NCP), as shown in Figure 1, is one of the areas with the highest NH3 column density retrieved from infrared satellite observations.11 National NH3 emissions in China are estimated to be 12
14 Tg for year 2000 and 13
16 Tg for year 2005,12
14 and account for 30
55% of total Asia NH3 emissions.12,15,16 SO2 emissions have become better controlled in China.17 National emissions of SO2 were required by the government to be reduced 10% by 2010 compared to the level in 2005. However, such reduction of SO2 may adversely affect PM2.5, because it will lead to an increase in aerosol nitrate in the regions where air quality is more acidic.5,18,19 Additionally, in terms of acidification effects, Zhao et al.20 indicated that the benefits of SO2 reductions by 10% in China during 2005 to 2010 would almost be negated by the increase of NOx and NH3 emissions. Xing et al.18 suggested NH3 emission control should be considered to reduce the total nitrogen deposition in the future. Undoubtedly, NH3 is one of the most important pollutants which should receive attention; however, modeling studies to understand the impacts of NH3 emission on fine particles in China are quite limited. In this paper, we conducted 3-D air quality simulations in conjunction with the Response Surface Modeling (RSM) technique to investigate sensitivities of the PM components to changes of their precursor emissions, including SO2, NOx, NH3, NMVOC, and primary particles, in east China. Nonlinear impacts of NH3 emissions on SIA have been evaluated, and a more effective NH3 emission control pathway is recommended.

’ METHODOLOGY The processes involved are the establishment of emission inventories, selection of air quality modeling domain and

ARTICLE

configuration, development, and validation of the emission control/air quality response prediction using RSM methodology. Emission Inventory. Emissions of SO2, NOx, PM10, PM2.5, black carbon (BC), organic carbon (OC), NH3, and NMVOC were calculated based on the framework of the GAINS-Asia model.21 The general method and some improvements used to develop the China regional emission inventory are described in our previous papers.22 In 2005, NH3 emissions from livestock farming, N-fertilizer application, N-fertilizer production, and human excreta are estimated to be 7.16, 8.35, 0.17, and 0.87 Tg, respectively. The first two are the most important NH3 contributors; they account for 43% and 50% of total emissions, respectively. Urea, ammonium bicarbonate (ABC), and other fertilizers account for 56%, 22%, and 22% of the N-fertilizer used in China. The consumption of N-fertilizer has been increasing in the past 15 years. In 2010, the consumption of ammonia fertilizer was 26.4% higher than that in 2005. Large variations presented in the geographical distribution are shown in Figure 1. The North China Plain, including Henan, Shandong, Hebei, and Jiangsu Provinces, contribute approximately 33% of national emissions, with an emission intensity as high as 9.0 t km
2, 4 times above the national average level (i.e., 1.7 t km
2). NH3 emissions have strong seasonal variations since the related agricultural activities and emission factors (i.e., N-volatilization rate) are significantly affected by the meteorological conditions.12,14,23,24 Highest NH3 emissions occur during June
August because of more favorable meteorological conditions (i.e., higher temperature) for NH3 volatilization and intensive agricultural activities. In this study, the monthly NH3 emissions in January, April, July, and October are estimated as 2.9%, 4.2%, 18.3%, and 7.5% of annual emissions, respectively. MM5/CMAQ Modeling. The air quality model used in this study is the Model-3/Community Multiscale Air Quality (CMAQ) modeling system (ver. 4.7), developed by U.S. EPA,25 which has been tested, evaluated, and applied in China.26
31 A one-way nested technique is employed in this study. Modeling domain 1 covers almost all of China with a 36  36 km horizontal grid resolution and generates the boundary conditions for nested domain at 12  12 km resolution over East China. The three most developed regions, North China Plain (NCP), Yangtze River delta (YRD), and Pearl River delta (PRD), have been chosen as the target areas, as shown in Figure 1 and Table S1. The target period is January, April, July, and October in 2005. A complete description of CMAQ configuration, meteorological, emission, and initial and boundary condition inputs used for this analysis are described in Xing et al.18,33 The Aerosol Optical Depth (AOD), NO2 and SO2 column density, as well as the ground concentrations of SO2, NO2, PM10, PM2.5, and its chemical components simulated by this modeling system have been validated through comparison with observations of satellite retrievals and surface monitoring data. Response Surface Modeling (RSM) Technique. A real-time emission control/air quality response tool, i.e., RSM, was developed at the U.S. EPA and applied to a number of air quality policy analyses and assessments.32 RSM uses advanced statistical techniques to characterize the relationships between model outputs (i.e., air quality responses) and input parameters (i.e., emission changes) in a highly efficient manner. Table 1 gives the sampling method and numbers of simulations used in this RSM application. Following the principle of RSM development as discussed in our previous paper,33 the responses of PM concentrations to 9294

dx.doi.org/10.1021/es2022347 |Environ. Sci. Technol. 2011, 45, 9293–9300

Environmental Science & Technology

ARTICLE

Table 1. Sample Methods and Key Parameters Involved during PM RSM Development RSM case

variable number

sample method

sample number

LHS-30-a

total-NOx, total-SO2

Latin hypercube sampling

30

LHS-30-b

total-NOx, total-NH3

Latin hypercube sampling

30

LHS-30-c

total-SO2, total-NH3

Latin hypercube sampling

30

LHS-30-d

total-NOx, total-NMVOC

Latin hypercube sampling

HSS-100

total-NOx, SO2, NH3, NMVOC, and PM

Hammersley quasi-random sequence sample

the changes of the total emissions of SO2, NOx, NH3, NMVOC, and PM over east China have been calculated. We define “emission ratios” as the ratio of the changed emissions compared to the baseline emissions. The emissions of each pollutant change from 0 to 200%, which means the emission ratios are from 0 to 2. We used 100 random emission control scenarios generated by Hammersley quasi-random Sequence Sample (HSS) method to establish the emission-based prediction model (HSS-100). In this study, RSM surface (emissions control and corresponding concentration change) prediction system is statistically generalized by MPerK (MATLAB Parametric Empirical Kriging) program followed Maximum Likelihood Estimation
Experimental Best Linear Unbiased Predictors (MLE-EBLUPs).34 Such control/response prediction system (i.e., HSS-100) has been validated through “leave-one-out cross validation” (LOOCV) (see Table S2), “out of sample” validation (see Table S3) and 2-D isopleths validation (see Figures S1 and S2). These results indicate that the HSS-100 predictions have good accuracy compared with CMAQ simulations. The stability of RSM with high dimensions (i.e., HSS-100) has been confirmed through its comparison with the one with low dimensions (i.e., LHS-30).

’ RESULTS AND DISCUSSION PM2.5 Sensitivity to NH3 Emissions. Following other sensitivity studies,35,36 we defined the PM2.5 sensitivity as the change ratio of PM2.5 concentration change to the change ratio of emissions, to evaluate the control effects of each pollutant,

SXa ¼

ΔC=C ðC
Ca Þ=C ¼  ΔEX =EX 1
a

ð1Þ

where SXa is the PM2.5 sensitivity to pollutant X (i.e., SO2, NOx, NH3, NMVOC, and PM) at its emission ratio a; Ca is the concentration of PM2.5 when the emission ratio of X is a; C* is the baseline concentration of PM2.5 (when emission ratio of X is 1). Figure 2 gives the comparison of PM2.5 sensitivities to the emissions of each pollutant (i.e., SO2, NOx, NH3, NMVOC, and PM) in the three target regions. The SIA accounts for about 20
50% of PM2.5 concentrations in three regions, which is consistent with observations.37 The PM2.5 sensitivities to PM emissions are about the same in various control levels. However, NH3, SO2, and NOx present significant nonlinear impacts; the PM2.5 sensitivities to their emissions get larger when more control efforts are taken, because of the transition between NH3-rich and NH3-poor conditions, the transition between NOx-limited and VOC-limited for ozone chemistry regimes and other thermodynamic effect and etc. The PM2.5 response to NH3 emissions is comparable with that of SO2 and NOx, and it is larger under higher control levels. Under 50% control level, NH3, NOx, and SO2 emissions reduce 7.9%, 10.8%, and 10.4% of

30 100

PM2.5 concentrations in NCP; 10.7%, 7.7%, and 8.9% in YRD; 9.9%, 5.2%, and 10.8% in PRD; and 10.7%, 10.2%, and 11.4% in east China. Nonlinear Impacts of NH3 Emission on SO42
and NO3
Aerosol. The reaction mechanism of atmospheric chemistry is given in Figure S3. Using the Beijing urban site as an example, the nonlinear response of SO42
and NO3
aerosol concentrations to the emission changes of precursors, is given in Figure 3. For SO42
concentration, the dominating contributor is SO2 emissions (Figure 3a, c). NH3 emissions slightly enhance the SO42
concentrations under NH3-poor condition, because NH3 provides a weak base condition to uptake more SO2 and also enhances the cloud SO2 oxidation rate by O3. But no effects are found under NH3-rich condition for both January and July (Figure 3c). Lower NOx emissions (an emission ratio of 0.2
0.4 in January and 0.7
0.9 in July, higher in summer due to stronger atmospheric oxidation capacity than in winter) and suitable NOx/NMVOC ratios benefit SO42‑ generation (Figure 3b, d). The hydroxyl radical is the key reactive species in both homogeneous (SO2 + •OH) or aqueous-phase paths of SO42
formation. Both NOx and NMVOC could be involved in •OH removals during the generation of NO3
and RO2/ HO2, therefore suitable NOx/NMVOC ratios will enhance the generation of ozone, the major source of the hydroxyl radical. In NOx-rich conditions, the SO42
sensitivity to NOx emissions is negative. The results are opposite in NOx- poor conditions. For NO3
concentration, NOx emissions are the dominating contributor. However, NH3 emissions are very important under NH3-poor conditions (as shown in Figure 3b), because NH3 reacts preferentially with H2SO4 instead of HNO3. The sensitivities of NO3
concentration to NOx and NH3 emissions (under baseline, i.e., emission ratio =1) are relatively larger in summer than those in winter, because NO3
is very volatile in the summer (due to high temperature) and, thus, the equilibrium moves dominantly toward the gas-phase HNO3 instead of particle-phase NH4NO3. SO2 emissions slightly benefit NO3
formation under NH3-rich conditions, especially at lower SO2 emissions level (Figure 3c). This is caused by the thermodynamic effect.2 The increase of NH4+ and SO42
ions will decrease the NH4NO3 equilibrium constant, shifting its partitioning toward the particulate phase.38 However, when NH3 is insufficient, SO2 emissions inhibit NO3
formation due to its competition with NH3. NMVOC emission slightly contributes NO3
pollution under NOx-rich condition in both January and July, and NOx emission slightly inhibits NO3
formation under NOx-rich condition in January (Figure 3d). Identification of NH3-Rich/-Poor Condition. Indicators for PM chemistry such as the degree of sulfate neutralization (DSN), gas ratio (GR), and adjusted gas ratio (AdjGR) could be used to identify the NH3-poor, -neutral, or -rich condition, then to 9295

dx.doi.org/10.1021/es2022347 |Environ. Sci. Technol. 2011, 45, 9293–9300

Environmental Science & Technology

ARTICLE

Figure 2. PM2.5 concentration sensitivity to the stepped control of individual pollutants (PM2.5 sensitivity = change ratio of PM2.5/change ratio of emission; all values are monthly average in January, April, July, and October in 2005).

determine the sensitivity of NO3
to precursors’ emissions.39 Their definitions are given as follows: DSN ¼

GR ¼


½NHþ 4 
½NO3 
½SO4 

2
ð½NH3  þ ½NHþ ½TA
2  ½TS 4 Þ
2  ½SO4  ¼
½TN ½NO3  þ ½HNO3 

ð2Þ

ð3Þ

AdjGR ¼

½NH3  þ ½NO
½TA
DSN  ½TS 3 ¼ ½TN ½NO
 þ ½HNO 3 3

ð4Þ

where [TA], [TN], and [TS] are the total molar concentration of ammonia ([NH3] + [NH+4 ]), nitrate ([NO
3 ] + [HNO3]), and sulfate ([SO2
4 ]), respectively. From RSM results, not only the NH3-rich/-poor condition under baseline scenario but also that under certain emission 9296

dx.doi.org/10.1021/es2022347 |Environ. Sci. Technol. 2011, 45, 9293–9300

Environmental Science & Technology

ARTICLE

Figure 3. 2-D Isopleths of SO42
and NO3
response to the emission changes of NOx, SO2, NH3, and NMVOC in Beijing, monthly average, 2005 (μg/m3).

control scenarios can be determined.30 The NH3-poor condition means the total available ammonia (gaseous ammonia + aerosol phase ammonium) is insufficient to charge-balance difference the remaining of other anions and cations,40 with the result that small perturbations in the ammonia emissions may have a significant effect on particle mass.41 Based on this principle, we defined an indicator—“Flex Ratio (FR)”—to identify the NH3-poor/-rich condition. As shown in Figure S4, under baseline NOx emissions (i.e., NOx emission ratio = 1), along with the decrease of NH3 emission ratio from 2.0 to 0, the NO3
slightly increases at first, but it gets sharply increased after the transition point (i.e., Flex Ratio). In the isopleths of NO3
response to NOx/NH3 emission changes predicted by RSM, the Flex Ratio is defined as the NH3 emission ratio at the flex NO3
concentrations under baseline NOx emissions (see Figure S4). When the FR is larger than the current NH3 emission ratio (in baseline = 1), the sensitivity of the NO3
concentration to NH3 emissions is more than that to NOx emissions, which indicate NH3-poor condition (see Table S4). In contrast, when the FR is less than 1, the NO3
concentration is more sensitive to NOx emissions instead of NH3 emissions, which indicates a NH3-rich condition, and the value (1
FR) reflects the ratio of free NH3 which could neutralize extra nitric acid produced by additional increases of NOx emissions. The spatial distributions of NO3
concentrations and GR are given in Figure S5. NO3
concentrations are found higher in January and lower in July, since higher temperature benefits NO3
evaporation and stronger atmospheric oxidation capacity favors converting S(IV) to S(VI), then enhancing the NH3

competition between SO42
and NO3
in July. Values of GR indicate NH3-rich, neutral, and poor conditions.39 The spatial distributions of GR value suggest that most of the polluted areas are located in NH3-rich conditions in all months (i.e., GR > 1). The FR over east China is shown in Figure 4. The FR derived from RSM gives consistent results, and the FR values in heavy NO3
pollution areas are mainly below 0.8. On an average annual basis, NCP and YRD are mainly located in NH3-rich conditions (FR is 0.6
0.7 and 0.8
1.0, respectively), therefore NO3
is more sensitive to NOx emissions, but PRD is located in NH3-poor conditions (FR is 1.0
1.5) and NO3
in PRD is more sensitive to NH3 emissions. The FR is around 0.8 in high NO3
areas, indicating NH3 is sufficiently abundant to satisfy an additional 25% (= 1/0.8
1) increase of NOx emissions to generate NO3
. Impacts of NH3 Emission Increase on SO42
and NO3
Aerosols. Previous studies on the emission trends in China indicate the NH3 emissions have been growing along with other precursors. According to these results, the emission trends for each pollutant during 1990
2005 could be fitted by parameterized quadratic functions, as shown in Figure 5a. NOx is the fastest growing pollutant, increasing over 100% from 1990 to 2005. SO2 emissions have increased by 30% during the same period. The NH3 and NMVOC emissions in 2005 are about 90% increased from that in 1990. The growth trends of SO42
and NO3
concentrations driven by the increases of the emissions during 1990
2005 have been calculated by RSM. The results are given in Figure 5 9297

dx.doi.org/10.1021/es2022347 |Environ. Sci. Technol. 2011, 45, 9293–9300

Environmental Science & Technology

ARTICLE

Figure 4. Flex ratio in January, April, July, and October, 2005 (FR < 1 suggests NH3-rich condition; FR > 1 suggests NH3-poor condition).

Figure 5. Historical and future growth of emissions impacts on SO42
and NO3
(average of 4 months, in east China).

(in a 4-month average). As seen in Figure 5, the base scenario reflects the impacts of all five pollutants emission simultaneous changes with SO42
and NO3
concentration. In addition, a series of hypothetical scenarios has been conducted to evaluate the impacts of each pollutant emission change on SO42
and NO3
concentrations. In each hypothetical scenario, one pollutant is held at the 1990 level (i.e., no increases during 1990
2005) and the rest are kept the same as the base scenario. In the baseline, the NO3
and SO42
concentrations increase by 150% and 20%, respectively. It is obvious that the growth of NOx

and SO2 emissions are the dominant factor to enhance NO3
and SO42
, respectively. Significant impacts could also be seen from the growth of NH3 emissions. About 50
60% increases of NO3
and SO42
are caused by the growth of NH3 emissions. The growths of NMVOC and SO2 emissions have no significant impacts on NO3
, while the growth of NOx hasnegative impacts on SO42
formation, possibly due to its influence on •OH as discussed in the previous section, especially during wintertime. Emissions of air pollutants and their projections have been changing significantly in recent years. The satellite data have 9298

dx.doi.org/10.1021/es2022347 |Environ. Sci. Technol. 2011, 45, 9293–9300

Environmental Science & Technology shown that NO2 increase in East Asia has been growing much faster than previous projections. Therefore, it is important to understand how China’s air pollutant emission change will affect the regional air quality in the future. Alternative scenarios for future SO2, NOx, and NMVOC emissions 18 were developed using forecasts of energy consumption and emission control strategies based on emissions in 2005, and on recent development plans for key industries in China, as shown in Figure 5b and c. In the reference scenario, which is based on the current control legislations and implementation status, i.e., REF scenario, the emissions of all pollutants are increasing from 2005 to 2030. In 2030, NO3
and SO42
will increase significantly, by 50% and 10%, respectively. In 2030, the NH3 emissions will increase by 20%, which may cause 15% and 4% increase of NO3
and SO42
, respectively. In the policy scenario, which is based on the improvement of energy efficiencies and strict environmental legislation, i.e., PC2 scenario, though NOx emissions will be better controlled in 2030, the increase of NH3 emissions will enhance NO3
by 10%. The decrease of SO2 emissions leads to significant reduction of SO42
, while the growth of NH3 will slightly improve SO42
by 2%. This implies future potential control of NH3 is important, especially for NO3
reduction. NH3 Impacts on the Acidity of Aerosols. The major concern about the potential negative impacts of NH3 control is the enhancement of aerosol acidity. In this study, we select the DSN as the indicator of the acidity of aerosols. When the DSN is less than 2, SO42‑ is insufficiently neutralized and the aerosol is more likely to be acid. The NH3 emissions level resulting in DSN less than 2 are calculated from RSM. Its spatial distributions over four months are given in Figure S6. High NH3 emissions are beneficial to the formation of NO3
. Over the polluted areas such as NCP and YRD which have the highest NH3 emission intensities,14 the values are 0.8
1 in January, April, and October, but higher than 1 in July. This indicates the acidity of aerosols is more sensitive to NH3 emissions in summer than in other seasons, mainly because of the high evaporation of NO3
in summer and the stronger atmospheric oxidation capacity which converts S(IV) to S(VI) and enhances the NH3 competition between SO42
and NO3
in July. Therefore, the acidity of aerosols is more sensitive to NH3 emissions in the summer than in other seasons.

’ ASSOCIATED CONTENT

bS

Supporting Information. This information is available free of charge via the Internet at http://pubs.acs.org/.

’ AUTHOR INFORMATION Corresponding Author

*Phone: +86-10-62771466; fax: +86-10-62773650; e-mail: shxwang@ tsinghua.edu.cn.

’ ACKNOWLEDGMENT The study was financially supported by Natural Science Foundation of China (20921140409), MEP’s Special Funds for Research on Public Welfares (201009001), and the U.S. EPA. We thank Dr. Thomas J. Santner and Dr. Gang Han at The Ohio State University for their help using the MperK program and Satoru Chatani from Toyota Central R&D Laboratories for aid

ARTICLE

with emission processing. We appreciate that Dr. Chuck Freed helped improve the language of the paper.

’ REFERENCES (1) Makar, P. A.; Moran, M. D.; Zheng, Q.; Cousineau, S.; Sassi, M.; Duhamel, A.; Besner, M.; Davignon, D.; Crevier, L.-P.; Bouchet, V. S. Modelling the impacts of ammonia emissions reductions on North American air quality. Atmos. Chem. Phys. 2009, 9, 7183–7212, DOI: 10.5194/acp-9-7183-2009. (2) Tsimpidi, A. P.; Karydis, V. A.; Pandis, S. N. Response of Inorganic Fine Particulate Matter to Emission Changes of Sulfur Dioxide and Ammonia: The Eastern United States as a Case Study. J. Air Waste Manage. Assoc. 2007, 57, 1489–1498, DOI: 10.3155/ 1047-3289.57.12.1489. (3) Nguyen, K.; Dabdub, D. NOx and VOC Control and Its Effects on the Formation of Aerosols. Aerosol Sci. Technol. 2002, 36, 560–572. (4) Mueller, S. F.; Bailey, E. M.; Kelsoe, J. J. Geographic Sensitivity of Fine Particle Mass to Emissions of SO2 and NOx. Environ. Sci. Technol. 2004, 38, 570–580. (5) Blanchard, C. L.; Tanenbaum, S.; Hidy, G. M. Effects of Sulfur Dioxide and Oxides of Nitrogen Emission Reductions on Fine Particulate Matter Mass Concentrations: Regional Comparisons. J. Air Waste Manage. Assoc. 2007, 57, 1337–1350, DOI: 10.3155/1047-3289.57.11.1337. (6) Pinder, R. W.; Adams, P. J.; Pandis, S. N. Ammonia Emission Controls as a Cost-Effective Strategy for Reducing Atmospheric Particulate Matter in the Eastern United States. Environ. Sci. Technol. 2007, 41, 380–386. (7) Tsimpidi, A. P.; Karydis, V. A.; Pandis, S. N. Response of Fine Particulate Matter to Emission Changes of Oxides of Nitrogen and Anthropogenic Volatile Organic Compounds in the Eastern United States. J. Air Waste Manage. Assoc. 2008, 58, 1463–1473, DOI: 10.3155/1047-3289.58.11.1463. (8) Redington, A. L.; Derwent, R. G.; Witham, C. S.; Manning, A. J. Sensitivity of modelled sulphate and nitrate aerosol to cloud, pH and ammonia emissions. Atmos. Environ. 2009, 43, 3227–3234. (9) Derwent, R.; Witham, C.; Redington, A.; Jenkin, M.; Stedman, J.; Yardley, R.; Hayman, G. Particulate matter at a rural location in southern England during 2006: Model sensitivities to precursor emissions. Atmos. Environ. 2009, 43, 689–696. (10) Beer, R.; Shephard, M. W.; Kulawik, S. S.; Clough, S. A.; Eldering, A.; Bowman, K. W.; Sander, S. P.; Fisher, B. M.; Payne, V. H.; Luo, M. Z.; Osterman, G. B.; Worden, J. R. First satellite observations of lower tropospheric ammonia and methanol. Geophys. Res. Lett. 2008, 35, L09801, DOI: 10.1029/2008GL033642. (11) Clarisse, L.; Clerbaux, C.; Dentener, F.; Hurtmans, D.; Coheur, P. F. Global ammonia distribution derived from infrared satellite observations. Nat. Geosci. 2009, 2, 479–483, DOI: 10.1038/ngeo551. (12) Streets, D. G.; Bond, T. C.; Carmichael, G. R.; Fernandes, S. D.; Fu, Q.; He, D.; Klimont, Z.; Nelson, S. M.; Tsai, N. Y.; Wang, M. Q.; Woo, J. H.; Yarber, K. F. An inventory of gaseous and primary aerosol emissions in Asia in the year 2000. J. Geophys. Res. 2003, 108, 8809, DOI: 10.1029/2002JD003093. (13) Wang, S. W.; Liao, Q. J. H.; Hu, Y. T.; Yan, X. Y. A Preliminary Inventory of NH3-N Emission and Its Temporal and Spatial Distribution of China. Chin. J. Agro-Environ. Sci. 2009, 28 (3), 619–629. (14) Dong, W. X.; Xing, J.; Wang, S. X. Temporal and Spatial Distribution of Anthropogenic Ammonia Emissions in China: 1994
2006. Chin. J. Environ. Sci. 2010, 31 (7), 1457–1463. (15) Zhao, D.; Wang, A. Emission of anthropogenic ammonia emission in Asia. Atmos. Environ. 1994, 28, 689–694. (16) Yamaji, K.; Ohara, T.; Akimoto, H. Regional-specific emission inventory for NH3, N2O, and CH4 via animal farming in South, Southeast, and East Asia. Atmos. Environ. 2004, 38, 7111–7121. (17) Lu, Z.; Streets, D. G.; Zhang, Q.; Wang, S.; Carmichael, G. R.; Cheng, Y. F.; Wei, C.; Chin, M.; Diehl, T.; Tan, Q. Sulfur dioxide emissions in China and sulfur trends in East Asia since 2000. Atmos. Chem. Phys. 2010, 10, 6311–6331. 9299

dx.doi.org/10.1021/es2022347 |Environ. Sci. Technol. 2011, 45, 9293–9300

Environmental Science & Technology (18) Xing, J.; Wang, S. X.; Chatani, S.; Zhang, C. Y.; Wei, W.; Klimont, Z.; Cofala, J.; Amann, M.; Hao, J. M. Projections of Air Pollutant Emissions and its Impacts on Regional Air Quality in China in 2020. Atmos. Chem. Phys. 2011, 11, 3119–3136, DOI: 10.5194/ acp-11-3119-2011. (19) West, J. J.; Ansari, A. S.; Pandis, S. N. Marginal PM2.5: Nonlinear aerosol mass response to sulfate reductions in the eastern United States. J. Air Waste Manage. Assoc. 1999, 49, 1415–1424. (20) Zhao, Y.; Duan, L.; Xing, J.; Larssen, T.; Nielsen, C. P.; Hao, J. M. Soil Acidification in China: Is Controlling SO2 Emissions Enough? Environ. Sci. Technol. 2009, 43, 8021–8026. (21) Amann, M.; Bertok, I.; Borken, J.; Chambers, A.; Cofala, J.; Dentener, F.; Heyes, C.; Hoglund, L.; Klimont, Z.; Purohit, P.; Rafaj, P.; Sch€opp, W.; Toth, G.; Wagner, F.; Winiwarter, W. A Tool to Combat Air Pollution and Climate Change Simultaneously; Methodology report; International Institute for Applied Systems Analysis (IIASA): Laxenburg, Austria, 2008. (22) Klimont, Z.; Cofala, J.; Xing, J.; Wei, W.; Zhang, C.; Wang, S.; Kejun, J.; Bhandari, P.; Mathur, R.; Purohit, P.; Rafaj, P.; Chambers, A.; Amann, M. Projections of SO2, NOx and carbonaceous aerosols emissions in Asia. Tellus, Ser. B 2009, 61, 602–617. (23) Zhang, M. S.; Luan, S. J. Application of NARSES in Evaluation of Ammonia Emission from Nitrogen Fertilizer Application in Planting System in China. Chin. J. Anhui Agric. Sci. 2009, 37 (8), 3583–3586. (24) Cao, G. L.; An, X. Q.; Zhou, C. H.; Ren, Y. Q.; Tu, J. Emission inventory of air pollutants in China. Chin. Environ. Sci. 2010, 30 (7), 900–906. (25) Byun, D. W.; Schere, L. K. Review of the governing equations, computational algorithms and other components of the models-3 Community Multiscale Air Quality (CMAQ) Modeling System. Appl. Mech. Rev. 2006, 59 (2), 51–77. (26) Zhang, M.; Uno, I.; Zhang, R.; Han, Z.; Wang, Z.; Pu, Y. Evaluation of the Models-3 Community Multi-scale Air Quality (CMAQ) modeling system with observations obtained during the TRACE-P experiment: comparison of ozone and its related species. Atmos. Environ. 2006, 40 (26), 4874–4882. (27) Streets, D. G.; Fu, J. S.; Jang, C.; Hao, J.; He, K.; Tang, X.; Zhang, Y.; Li, Z.; Zhang, Q.; Wang, L.; Wang, B.; Yu, C. Air quality during the 2008 Beijing Olympic games. Atmos. Environ. 2007, 41 (3), 480–492. (28) Fu, J. S.; Jang, C. J.; Streets, D. G.; Li, Z.; Kwok, R.; Park, R.; Hang, Z. MICS-Asia II: Evaluating gaseous pollutants in East Asia using an advanced modeling system: Models-3/CMAQ System. Atmos. Environ. 2008, 42 (15), 3571–3583. (29) Fu, J. S.; Streets, D. G.; Jang, C. J.; Hao, J.; He, K.; Wang, L.; Zhang, Q. Modeling Regional/Urban Ozone and Particulate Matter in Beijing, China. J. Air Waste Manage. Assoc. 2009, 59, 37–44. (30) Wang, L.; Carey, C. J.; Zhang, Y.; Wang, K.; Zhang, Q.; Streets, D. G.; Fu, J.; Lei, Y.; Schreifels, J.; He, K.; Hao, J.; Lam, Y. F.; Lin, J.; Meskhidze, N.; Voorhees, S.; Evarts, D.; Phillips, S. Assessment of air quality benefits from national air pollution control policies in China. Part II: Evaluation of air quality predictions and air quality benefits assessment. Atmos. Environ. 2010, 44, 3449–3457. (31) Li, L.; Chen, C. H.; Fu, J. S.; Huang, C.; Streets, D. G.; Huang, H. Y.; Zhang, G. F.; Wang, Y. J.; Jang, C. J.; Wang, H. L.; Chen, Y. R.; Fu, J. M. Air quality and emissions in the Yangtze River Delta, China. Atmos. Chem. Phys. 2011, 11, 1621–1639. (32) U.S. Environmental Protection Agency. Technical Support Document for the Proposed PM NAAQS Rule: Response Surface Modeling; Office of Air Quality Planning and Standards: Research Triangle Park, NC, 2006. (33) Xing, J.; Wang, S. X.; Jang, C.; Zhu, Y.; Hao, J. M. Nonlinear Response of Ozone to Precursor Emission Changes in China: a Modeling Study using Response Surface Methodology. Atmos. Chem. Phys. 2011, 11, 5027–5044, DOI: 10.5194/acp-11-5027-2011. (34) Santner, T. J.; Williams, B. J.; Notz, W. The Design and Analysis of Computer Experiments; Springer Verlag: New York, 2003. (35) Yarwood, G.; Wilson, G.; Morris, R. Development of The CAMx Particulate Source Apportionment Technology (Psat), final report; ENVIRON International Corporation, 2005.

ARTICLE

(36) Koo, B.; Wilson, G. M.; Morris, R. E.; Dunker, A. M.; Yarwood, G. Comparison of Source Apportionment and Sensitivity Analysis in a Particulate Matter Air Quality Model. Environ. Sci. Technol. 2009, 43, 6669–6675. (37) Chan, C. K.; Yao, X. H. Air pollution in mega cities in China. Atmos. Environ. 2008, 42, 1–42. (38) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics: From Air Pollution to Climate Change; John Wiley and Sons, Inc.: New York, 2006. (39) Zhang, Y.; Wen, X. Y.; Wang, K.; Vijayaraghavan, K.; Jacobson, M. Z. Probing into Regional O3 and PM Pollution in the U.S., Part II. An Examination of Formation Mechanisms through a Process Analysis Technique and Sensitivity Study. J. Geophys. Res. 2009, 114(D22305), DOI: 10.1029/2009JD011900. (40) Blanchard, C. L.; Roth, P. M.; Tanenbaum, S. J.; Ziman, S. D.; Seinfeld, J. H. The use of ambient measurements to identify which precursor species limit aerosol formation. J. Air Waste Manage. Assoc. 2000, 50, 2073–2084. (41) Makar, P. A.; Moran, M. D.; Zheng, Q.; Cousineau, S.; Sassi, M.; Duhamel, A.; Besner, M.; Davignon, D.; Crevier, L.-P.; Bouchet, V. S. Modelling the impacts of ammonia emissions reductions on North American air quality. Atmos. Chem. Phys. 2009, 9, 7183–7212, DOI: 10.5194/acp-9-7183-2009.

9300

dx.doi.org/10.1021/es2022347 |Environ. Sci. Technol. 2011, 45, 9293–9300

ARTICLE pubs.acs.org/est

Plasma-TiO2 Catalytic Method for High-Efficiency Remediation of p-Nitrophenol Contaminated Soil in Pulsed Discharge Tie Cheng Wang,† Na Lu,†,‡ Jie Li,†,‡,* and Yan Wu†,‡ † ‡

Institute of Electrostatics and Special Power, Dalian University of Technology, Dalian 116024, PR China Key Laboratory of Industrial Ecology and Environmental Engineering, Ministry of Education of the People’s Republic of China, Dalian 116024, PR China

bS Supporting Information ABSTRACT: Nonthermal discharge plasma and TiO2 photocatalysis are two techniques capable of organic pollutants removal in soil. In the present study, a pulsed discharge plasma-TiO2 catalytic (PDPTC) technique by combining the two means, where catalysis of TiO2 is driven by the pulsed discharge plasma, is proposed to investigate the remediation of p-nitrophenol (PNP) contaminated soil. The experimental results showed that 88.8% of PNP was removed within 10 min of treatment in PDPTC system and enhancing pulse discharge voltage was favorable for PNP degradation. The mineralization of PNP and intermediates generated during PDPTC treatment was followed by UV-vis spectra, denitrification, total organic carbon (TOC), and COx selectivity analyses. Compared with plasma alone system, the enhancement effects on PNP degradation and mineralization were attributed to more amounts of chemically active species (e.g., O3 and H2O2) produced in the PDPTC system. The main intermediates were identified as hydroquinone, benzoquinone, catechol, phenol, benzo[d][1, 2, 3]trioxole, acetic acid, formic acid, NO2
, NO3
, and oxalic acid. The evolution of the main intermediates with treatment time suggested the enhancement effect of the PDPTC system. A possible pathway of PNP degradation in soil in such a system was proposed.

’ INTRODUCTION Nitrophenols are toxic and biorefractory organic compounds, used extensively as raw materials and intermediates in the production of explosives, pharmaceuticals, pesticides, pigments, dyes, wood preservatives, and rubber chemicals.1 Three nitrophenols (2-nitrophenol, 4-nitrophenol, and 2, 4-dinitrophenol) have been listed in the USEPA list of priority pollutants.2 Nitrophenols were released to soil as fugitive emissions during their production and use, causing serious health hazards. Taking China as an example, some contaminated lands, where nitrophenols extensively exist, have been left at the center of the city after some chemical factories migrated to the suburb in the industrial rearrangement. These lands are of high economic values because of their locations, and hence, need to be remedied rapidly. Several technologies such as physical method,3 chemical method,4 bioremediation,5 electrokinetic remediation,6 thermal technology,7 and photocatalysis8 have been employed to remediate organic pollutants contaminated soils. With the increasing strictness of industrial standard and the increment of economic values of lands, high-efficient and rapid soil remediation method is becoming a necessity. In this case, the conventional remediation technologies would not meet the requirement of high-efficient and rapid r 2011 American Chemical Society

remediation due to the drawbacks such as second pollution and time-consuming. In our previous study, pulsed discharge plasma technology, one of the advanced oxidation processes, has been employed to remediate pentachlorophenol contaminated soil, and great performance of soil remediation was obtained in a short remediation period.9 In pulsed discharge plasma process, discharge energy is released in forms of high energy electrons, strong electric field, and UV light radiation, etc., some of which can excite gases in plasma region to generate chemically active species. However, the utilization efficiency of the discharge energy is still needed to be enhanced in order to satisfy the requirement of practical application. Catalysis is suggested to be viable by introducing the active catalyst into the discharge plasma soil remediation system. Anatase TiO2, an economic and photosensitive semiconductor material with a band gap of about 3.2 eV, can be excited by strong electric field and UV light radiation to generate pairs of electrons and holes, resulting in more numerous chemically Received: April 26, 2011 Accepted: September 16, 2011 Revised: September 15, 2011 Published: September 16, 2011 9301

dx.doi.org/10.1021/es2014314 | Environ. Sci. Technol. 2011, 45, 9301–9307

Environmental Science & Technology

ARTICLE

Figure 1. Effect of pulsed discharge voltage on PNP degradation in PDPTC system.

Figure 2. Evolution of UV-vis absorption spectra of PNP with treatment time.

active species generation.10 Recently, the combination of nonthermal discharge plasma with TiO2 photocatalysis for pollutant removal has drawn great attention.11
13 Previous studies presented that the introduction of TiO2 into discharge plasma system could enhance pollutants removal and promote energy efficiency.11 However, relevant research was mostly focused on wastewater treatment and little has been reported on soil remediation. In this work, a pulsed discharge plasma-TiO2 catalytic (PDPTC) technique is proposed to enhance the remediation of p-nitrophenol (PNP) contaminated soil. The study was focused on exploring the enhancement of PNP degradation in this PDPTC system, and the variances of the main intermediates between PDPTC system and plasma alone system were compared to evaluate the enhanced behavior. Possible mechanisms of such enhancement were discussed by analyzing the variances in the amounts of chemically active species. A possible pathway of PNP degradation in such a system was proposed.

Extraction and Analysis. After discharge treatment, PNP in soil was extracted immediately, and the extraction procedure was described in SI S4. The extractions produced average recoveries of 90.1
95.3%. PNP concentration, total organic carbon (TOC), intermediates, O3 and H2O2 concentrations, and CO2 and CO were analyzed and the details were shown in SI S5. The COx selectivity, CO2 conversion, and denitrification efficiency were defined in SI S6. All experiments were conducted in duplicates.

’ MATERIALS AND METHODS Materials. PNP was used in the study, and its detailed introduction was presented in S1 of the Supporting Information (SI). Soil samples were collected from a suburb of Dalian, China. The details were presented in SI S2. The original PNP concentration in the soil was 800 mg kg
1. TiO2 (Degussa, P25) (BET area =50 m2 g
1) was used as the catalyst. Treatment of Contaminated Soil. The schematic diagram of the experimental apparatus was illustrated in Figure S1 in the SI, which was similar with our previous work.9 The details of the reactor were showed in SI S3. The pulse frequency and pulseforming capacitance Cp were 100 Hz and 200 pF, respectively, and the input energies per pulse were 0.016, 0.020, and 0.023 J at pulse voltage of 16, 18, and 20 kV, respectively. The experiments were all conducted at 20 kV unless special illustration. In each experiment, a certain amount of TiO2 was added into PNP contaminated soil and then homogenized. TiO2 amount in the soil was 2 wt %. The soil sample (approximately 2.0 g) was spread on the ground electrode with a thickness of about 1.3 mm. Prior to discharge treatment, the moisture content of soil was adjusted to 10% with deionized water. Air was injected for one side of the reactor and out from the other side with the flow rate of 0.5 L min
1.

’ RESULTS AND DISCUSSION PNP Degradation in PDPTC System. Figure 1 showed the effect of pulse discharge voltage on PNP degradation in soil in PDPTC system. On the one hand, the introduction of TiO2 enhanced PNP removal in soil. At pulsed discharge voltage of 20 kV, PNP degradation efficiency reached 88.8% after 10 min of discharge treatment in the PDPTC system, which was 78.1% in plasma alone system. In the case of TiO2 catalyst, conduction band electrons and holes (h+) would be generated when TiO2 was irradiated with high energy input, and the photogenerated electrons and holes could react with PNP directly or indirectly, increasing PNP degradation. Hoffmann14 has reported that the conduction band electrons, holes (h+) and the reactive oxygen species such as •OH radicals and superoxide radicals generated on the illuminated catalyst could promote pollutant removal. On the other hand, increase in pulsed discharge voltage greatly enhanced PNP degradation efficiency in the PDPTC system. For example, at discharge voltage of 16 kV, only 65.2% of PNP was removed after 10 min of discharge treatment, while it increased to 88.8% at 20 kV. More energy is injected into the reactor when the discharge voltage increases, and then more plasma channels with strong energy are very effective to generate more amounts of chemically active species, and therefore PNP degradation efficiency is enhanced; meanwhile, the effects including high energy electrons, strong electric field and UV light radiation etc would also become stronger at higher discharge voltage. In this case, more conduction band electrons and holes (h+) would be generated, and then the formation of chemically active species was accelerated, which promoted PNP removal. In addition, the energy efficiencies for PNP removal in the PDPTC system after 10 min of discharge treatment were 3.90, 3.84, and 3.70 g kWh
1 at pulse voltages of 16, 18, and 20 kV, respectively, as presented in SI Table S1. Considering PNP degradation and energy efficiency comprehensively, 20 kV was used in the following experiments. 9302

dx.doi.org/10.1021/es2014314 |Environ. Sci. Technol. 2011, 45, 9301–9307

Environmental Science & Technology

Figure 3. Evolution of NO2
and NO3
with treatment time.

Control experiments with O3 and H2O2 addition in the absence of plasma were conducted, and the results were presented in SI Table S2. Herein, method of ozonation experiment was the same as our previous study.9 The results in SI Table S2 suggested that O3 played a decisive role for PNP degradation, and H2O2 also played a certain role. Mineralization of PNP in PDPTC System. Mineralization of PNP in the PDPTC system was studied by changes of UV
vis spectra, NO3
and NO2
formation, TOC removal, and CO2 and CO generation. UV-vis Absorption Spectra. The change of the UV
vis absorption spectra of PNP during degradation process in the PDPTC system was shown in Figure 2. The absorption peak at 400 nm disappeared quickly with the increase of treatment time, indicating that PNP was removed gradually. Formation of NO2
and NO3
. Since there is a nitro-group in PNP molecule, tracing changes of nitrogen forms is an approach to evaluate the degree of PNP degradation. To our knowledge, the nitro-group can be converted into NO3
and NO2
ions.15,16 Therefore, NO3
and NO2
ions were both monitored during PNP degradation in soil, and as control experiments, the formation of NO2
and NO3
were also analyzed in clean soil (uncontaminated soil) during pulsed discharge process. The concentrations of NO3
and NO2
released from PNP were calculated by subtracting the concentrations in clean soil from those in contaminated soil, respectively. The evolution of NO3
and NO2
concentrations released from PNP with treatment time was shown in Figure 3, and their concentrations generated in clean soil during pulsed discharge plasma were presented in SI Figure S2. The NO2
concentration in the PDPTC system declined after an initial increase in Figure 3, probably due to its further oxidation to other nitrogen forms, whereas the NO3
concentration increased gradually with treatment time. Similar trends were also presented in plasma alone system. Moreover, the NO3
concentration increased slowly in the initial 10 min, and then the rate increased gradually. These results indicated that the NO2
mainly resulted from PNP degradation, and NO2
was formed first when the nitrogen-tocarbon single bond (—N—C—) of the PNP was broken down, and then it was oxidized into NO3
. Active species reacted rapidly with nitrophenol to produce NO2
, and the NO2
concentration quickly reached a maximum and then decreased rapidly, and during the process it was oxidized into NO3
.15,16 More importantly, as shown in Figure 3, more amounts of NO3
and NO2
were formed in the PDPTC system than in plasma alone system in the initial 10 min, whereas higher NO3
and lower NO2
concentrations occurred in the PDPTC system after

ARTICLE

Figure 4. Evolution of TOC removal with treatment time.

Figure 5. Changes of COx selectivity with treatment time in PDPTC system.

30 min of treatment, compared with those in plasma alone system. These results suggested that more amounts of NO2
were oxidized into NO3
in the PDPTC system, which was attributed to the intense oxidation environment caused by TiO2 catalyst. The enhanced oxidation environment in the PDPTC system could also be further confirmed by denitrification efficiency. The evolution of denitrification efficiency of PNP with treatment time was presented in SI Figure S3. It was found that 81.3% of denitrification efficiency was achieved after 30 min of treatment in the PDPTC system, and there was a 20% rise as compared with that in plasma alone system. TOC Removal. The TOC values have been related to the total concentration of organic compounds, and the decrease of TOC with treatment time can reflect the degree of mineralization. Therefore, the TOC removal during PNP degradation process in the PDPTC system was shown in Figure 4. TOC removal efficiency achieved 55.1% in the PDPTC system after 30 min of treatment, compared with that of 42.9% in plasma alone system. These results demonstrated that more PNP and intermediates were mineralized to smaller organic molecules or to CO2 in the PDPTC system. Generation of CO2 and CO in Offgas. The changes of UV
vis absorption spectra, the formation of NO3
and NO2
, and the TOC removal only reflect indirectly the mineralization extent of PNP, whereas the generation of CO2 and CO can reflect directly its mineralization. Therefore the generation of CO2 and CO during PNP degradation was measured. Figure 5 presented the evolution of COx selectivity with treatment time during PNP degradation. With the treatment time continued, CO2 selectivity increased and CO selectivity decreased in the 9303

dx.doi.org/10.1021/es2014314 |Environ. Sci. Technol. 2011, 45, 9301–9307

Environmental Science & Technology

ARTICLE

Table 1. Comparison of O3 and H2O2 Concentrations and Their Energy Costs with/Without TiO2 in Plasma Process uncontaminated soil

contaminated soil

CO3 (mgL-1)

CH2O2 (mmol L
1)

O3 energy costs (mgkJ-1)

H2O2 energy costs (mgkJ-1)

CO3 (mgL-1)

CH2O2 (mmol L
1)

without TiO2

42

0.038

0.76

0.014

23

0.022

with TiO2

57

0.062

1.03

0.022

30

0.041

•OH þ •OH f H2 O2

ð4Þ

O2 þ ecb
f O2 •


ð5Þ

O2 •
þ 2H2 O f H2 O2 þ 2OH
þ O2

ð6Þ

On the other hand, O3 concentration in the PDPTC system was always higher than that in plasma alone system, as shown in Table 1. O2 can be cleaved into single atomic oxygen radical anion (O•) by gas phase electrical pulses, and it can also be converted to superoxide radical anion (O2•
) by the effect of highly energized electrons (e
) and ecb
on the surface of TiO2. Subsequently, O•
and O2•
are transformed into more chemically active species (O and O2) by hvb+ on the surface of TiO2, resulting in more amounts of O3 production through the following possible reaction pathways:10,20,21 Figure 6. Evolution of main intermediates with treatment time.

O2 þ e
f O2 •


ð7Þ

PDPTC system. 96.8% of CO2 selectivity was obtained in the PDPTC system after 20 min of treatment, with 3.2% of CO selectivity. These results suggested that CO formed in the reaction was further oxidized into CO2. More importantly, higher CO2 selectivity and lower CO selectivity occurred in the PDPTC system than in plasma alone system. The intense oxidative environment in the PDPTC system was the reason for the enhanced CO2 selectivity and suppressed CO selectivity. Formation of Active Species. To explore the enhancement mechanisms of PNP degradation in the PDPTC system, the variances of chemically active species such as O3 and H2O2 were analyzed in clean and contaminated soils, respectively, and the results were presented in Table 1. Herein the discharge time was 20 min. As shown in Table 1, on the one hand, more amounts of H2O2 were generated in the PDPTC system both in clean soil and contaminated one, compared with those in plasma alone system. When an electron on the valence band (VB) of TiO2 absorbs some energy higher than the band gap between the VB and the conduction band (CB), it will be promoted to the CB and thus an electron
hole pair (ecb

hvb+) is formed. The holes will oxidize either H2O molecule or OH
anions to form •OH. The electrons on the CB react with O2 to generate superoxide radical anion (O2
). Then, the O2
will react with H2O to produce •OH. •OH can react with each other to form H2O2. Therefore, more amounts of H2O2 were generated through the following possible reaction pathways:17
19

O2 •
þ hvb þ f O2

ð8Þ

O•
þ hvb þ f O

ð9Þ

TiO2 þ hv sf ecb
þ hvb þ

energy

ð1Þ

hvb þ þ H2 O f Hþ þ •OH

ð2Þ

hvb þ þ OH
f •OH

ð3Þ

O þ O2 f O3

ð10Þ

As mentioned above, it is believed that the PDPTC system is effective to generate more chemically active species (such as O3 and H2O2), leading to the enhancement of PNP degradation in soil. Possible Degradation Pathways. The intermediates of PNP degradation in soil in the PDPTC system were analyzed using HPLC, HPLC/MS and IC. They mainly included hydroquinone, benzoquinone, catechol, phenol, benzo[d][1, 2, 3]trioxole, acetic acid, formic acid, NO2
, NO3
, and oxalic acid. Similar results were also reported by Oturan,22 where hydroquinone and benzoquinone were two main intermediates during PNP degradation by Fenton method. Hydroquinone, benzoquinone, phenol, formic acid, and oxalic acid were also detected as intermediates during PNP degradation by ozonation, and
NO2 group could be easily removed from the aromatic ring in the process.23 The evolution of some aromatic intermediates with treatment time in the PDPTC system and plasma alone system were analyzed, as depicted in Figure 6. Maximum concentrations of hydroquinone, benzoquinone and catechol in the PDPTC system were all lower than those in the plasma alone system. Hydroquinone reached the maximum concentration earlier in the PDPTC system, whereas benzoquinone and catechol reached the maximum concentrations almost at the same time in the two discharge systems. Phenol was generated earlier in the PDPTC system. Besides that, these intermediates in each reaction system all encountered further degradation with the 9304

dx.doi.org/10.1021/es2014314 |Environ. Sci. Technol. 2011, 45, 9301–9307

Environmental Science & Technology

ARTICLE

Figure 7. Possible degradation pathways of PNP in soil in PDPTC system.

treatment process went on, and lower concentrations occurred in the PDPTC system finally. The results presented that greater degradation performance occurred in the PDPTC system, which was attributed to the enhanced generation of chemically active species by the effect of TiO2 catalyst. On the other hand, from the sequence of accumulation of aromatic intermediates, phenol appeared later than hydroquinone, benzoquinone and catechol. Therefore, it could be concluded that hydroquinone, benzoquinone and catechol were generated more easily than phenol during PNP degradation. The differences in evolution of intermediates were due to the intensive oxidation environment in the PDPTC system. On the one hand, more amounts of H2O2 and O3 generated in PDPTC system could increase the formation of •OH radicals through trapping of photogenerated electrons, and they also interfered in recombination of electrons and positive holes due to the electrophilic properties of H2O2 and O3.24 On the other hand, active species generated in discharge plasma could act directly on the active sites of TiO2, and then accelerate TiO2 to trigger reactions, and meanwhile, the strong electric field in discharge plasma could inhibit the recombination of electrons and holes on TiO2 surface.25,26 Based on the intermediates and their evolution with treatment time, and the roles of O3, H2O2, and •OH radicals played in the present study, possible degradation pathways of PNP in soil were proposed in Figure 7. The patterns of intermediates indicated

that hydroxylation was the main oxidation pathway. Hydroxylated intermediates could result from electrophilic attack on PNP by O3 and •OH radicals. Aromatic ring of PNP contains two substituents,
OH and
NO2. The
OH is electron-donating and an ortho- and para-director, while
NO2 is electron-withdrawing and meta-director. •OH radicals preferentially attack the ortho- or para-position with respect to the
OH group due to the electrophilic nature. The •OH radicals may eliminate nitrous acid from PNP to yield 1,4-benzosemiquinone as an intermediate, which subsequently disproportionates into hydroquinone and benzoquinone. Similar results were reported by Liu et al.27 and Suarez et al.28 The possibility of a direct attack of •OH radicals at the position carrying the
NO2 group also exists, with resultant hydroquinone formation.29 In addition, the •OH radicals may attack the
NO2 group due to the relatively long length of C—N bond in PNP molecule, which is the longest bond and would be potential to be attacked to form phenol,30 and then hydroquinone, benzoquinone and catechol would be generated by further oxidation of phenol. Further reactions of these intermediates with •OH radicals lead to ring cleavage and formation of aliphatic compounds. O3 reacts with organic pollutants through nucleophilic, electrophilic and cyclo-addition reactions.31,32 The nucleophilic and electrophilic attacks of O3 on PNP proceeds preferentially on the ortho- and para-positions with respect to the
OH group to yield 9305

dx.doi.org/10.1021/es2014314 |Environ. Sci. Technol. 2011, 45, 9301–9307

Environmental Science & Technology the same hydroxylated intermediates as in the case of •OH radicals attack.23 Furthermore, the cyclo-addition reactions of O3 may cause the addition of O3 to PNP molecule structure, and finally form cyclo-addition intermediates,33 such as benzo[d][1, 2, 3]trioxole in the present study. Further attacks to these intermediates by O3 would lead to ring-cleavage of the aromatic ring to form acetic acid, formic acid, and oxalic acid. The formation of NO2
is the result of the denitrification of PNP. In the process, the NO2
concentration quickly reaches a maximum and then decreases rapidly to be oxidized into NO3
. This study opens a possible way to improve soil remediation in pulsed discharge plasma through TiO2 catalyst.

’ ASSOCIATED CONTENT

bS Supporting Information. Text S1
S6 include introduction of PNP and other reagents, details of the soil sample, reactor introduction, extraction procedure, analysis methods, COx selectivity, and denitrification efficiency. Figures S1
S3 include the reactor system, nitrite and nitrate concentrations in clean soil, and denitrification efficiency. Table S1 presents the energy efficiency within 10 min of discharge treatment at different discharge voltages. Table S2 presents the comparison of PNP degradation in soil by pulsed discharge plasma, ozonation, and H2O2 oxidation. This material is available free of charge via the Internet at http://pubs.acs.org. ’ AUTHOR INFORMATION Corresponding Author

*Phone: +86-411-84708576; fax: +86-411-84709869; e-mail: [email protected].

’ ACKNOWLEDGMENT We thank the National Natural Science Foundation, P.R. China (Project No. 40901150), the Ministry of Science and Technology, P.R. China (Project No. 2008AA06Z308), and Program for Liaoning Excellent Talents in University, China (Project No. 2009R09) for their financial support to this research. ’ REFERENCES (1) Uberoi, V.; Bhattacharya, S. K. Toxicity and degradability of nitrophenols in anaerobic systems. Water Environ. Res. 1997, 69 (2), 146
156; DOI: 10.2175/106143097X125290. (2) USEPA; http://www.scorecard.org. 2002. (3) Khodadoust, A. P.; Bagchi, R.; Suidan, M. T.; Brenner, R. C.; Sellers, N. G. Removal of PAHs from highly contaminated soils found at prior manufactured gas operations. J. Hazard. Mater. 2000, 80 (1
3), 159
174; DOI: 10.1016/S0304-3894(00)00286-7. (4) Liao, C. J.; Chung, T. L.; Chen, W. L.; Kuo, S. L. Treatment of pentachlorophenol-contaminated soil using nano-scale zero-valent iron with hydrogen peroxide. J. Mol. Catal. A: Chem. 2007, 265 (1
2), 189
194; DOI: 10.1016/j.molcata.2006.09.050. (5) Lamar, R. T.; Evans, J. W.; Glaser, J. A. Solid-phase treatment of a pentachlorophenol-contaminated soil using lignin-degrading fungi. Environ. Sci. Technol. 1993, 27 (12), 2566
2571; DOI: 10.1021/ es00048a039. (6) Zhang, S. P.; Rusling, J. F. Dechlorination of polychlorinated biphenyls on soils and clay by electrolysis in a biocontinuous microemulsion. Environ. Sci. Technol. 1995, 29 (5), 1195
1199;DOI: 10.1021/es00005a009.

ARTICLE

(7) Acharya, P.; Ives, P. Incineration at bayou bounfouca remediation project. Waste Manage. 1994, 14 (1), 13
26; DOI: 10.1016/0956053X(94)90017-5. (8) Balmer, M. E.; Goss, K. U.; Schwarzenbach, R. P. Photolytic transformation of organic pollutants on soil surfaces-An experimental approach. Environ. Sci. Technol. 2000, 34 (7), 1240
1245; DOI: 10.1021/es990910k. (9) Wang, T. C.; Lu, N.; Li, J.; Wu, Y. Evaluation of the potential of pentachlorophenol degradation in soil by pulsed corona discharge plasma from soil characteristics. Environ. Sci. Technol. 2010, 44 (8), 3105
3110; DOI: 10.1021/es903527w. (10) Mills, A.; LeHunte, S. An overview of semiconductor photocatalysis. J. Photochem. Photobiol., A 1997, 108 (1), 1
35; DOI: 10.1016/S1010-6030(97)00118-4. (11) Hao, X. L.; Zhou, M. H.; Lei, L. C. Non-thermal plasma-induced photocatalytic degradation of 4-chlorophenol in water. J. Hazard. Mater. 2007, 141 (3), 475
482; DOI: 10.1016/j.jhazmat.2006.07.012. (12) Lukes, P.; Clupek, M.; Sunka, P.; Peterka, F.; Sano, T.; Negishi, N.; Matsuzawa, S.; Takeuchi, K. Degradation of phenol by underwater pulsed corona discharge in combination with TiO2 photocatalysis. Res. Chem. Intermed. 2005, 31 (4
6), 285
294; DOI: 10.1163/ 1568567053956734. (13) Maroulf-Khelifa, K.; Abdelmalek, F.; Khelifa, A.; Addou, A. TiO2-assisted degradation of a perfluorinated surfactant in aqueous solutions treated by gliding arc discharge. Chemosphere 2008, 70 (11), 1995
2001; DOI: 10.1016/j.chemosphere.2007.09.030. (14) Hoffmann, M. R.; Martin, S. T.; Choi, W. Y.; Bahnemann, D. W. Environmental applications of semiconductor photocatalysis. Chem. Rev. 1995, 95 (1), 69
96; DOI: 10.1021/cr00033a004. (15) Tang, Q.; Lin, S.; Jiang, W. J.; Lim, T. M. Gas phase dielectric barrier discharge induced reactive species degradation of 2, 4-dinitrophenol. Chem. Eng. J. 2009, 153 (1
3), 94
100; DOI: 10.1016/j. cej.2009.06.022. (16) Wang, K. H.; Hsieh, Y. H.; Chen, L. J. The heterogeneous photocatalytic degradation, intermediates and mineralization for the aqueous solution of cresols and nitrophenols. J. Hazard. Mater. 1998, 59 (2
3), 251
260; DOI: 10.1016/S0304-3894(97)00151-9. (17) Zhang, J. L.; Xu, H. S.; Chen, H. J.; Anpo, M. Study on the formation of H2O2 on TiO2 photocatalysts and their activity for the photocatalytic degradation of X-GL dye. Res. Chem. Intermed. 2003, 29 (7
9), 839
848;DOI: 10.1163/156856703322601843. (18) Sato, M.; Ohgiyama, T.; Clements, J. S. Formation of chemical species and their effects on microorganisms using a pulsed high-voltage discharge in water. IEEE Trans. Ind. Appl. 1996, 32 (1), 106
112; DOI: 10.1109/28.485820. (19) Pichat, P.; Disdier, J.; Hoang-Van, C.; Mas, D.; Goutailler, G.; Gaysse, C. Purification/deoderization of indoor air and gaseous effluents by TiO2 photocatalysis. Catal. Today 2000, 63 (9), 363
369; DOI: 10.1016/S0920-5861(00)00480-6. (20) Simek, M.; Clupek, M. Efficiency of ozone production by pulsed positive corona discharge in synthetic air. J. Phys. D: Appl. Phys. 2002, 35 (11), 1171
1175; DOI: 10.1088/0022-3727/35/11/311. (21) Ghezzar, M. R.; Abdelmalek, F.; Belhadj, M.; Benderdouche, N.; Addou, A. Gliding arc plasma assisted photocatalytic degradation of anthraquinonic acid green 25 in solution with TiO2. Appl. Catal., B 2007, 72 (3
4), 304
313; DOI: 10.1016/j.apcatb.2006.11.008. (22) Oturan, M. A.; Peiroten, J.; Chartrin, P.; Acher, A. J. Complete destruction of p-nitrophenol in aqueous medium by electro-Fenton method. Environ. Sci. Technol. 2000, 34 (16), 3474
3479; DOI: 10.1021/es990901b. (23) Shi, H. X.; Xu, X. W.; Xu, X. H.; Wang, D. H.; Wang, Q. D. Mechanistic study of ozonation of p-nitrophenol in aqueous solution. J. Environ. Sci. 2005, 17 (6), 926–929. (24) Logemann, F. P.; Annee, J. H. J. Water treatment with a fixed bed catalytic ozonation process. Water Sci. Technol. 1997, 35 (4), 353
360; DOI: 10.1016/S0273-1223(97)00045-0. (25) Sano, T.; Negishi, N.; Sakai, E.; Matsuzawa, S. Contributions of photocatalytic/catalytic activities of TiO2 and gamma-Al2O3 in 9306

dx.doi.org/10.1021/es2014314 |Environ. Sci. Technol. 2011, 45, 9301–9307

Environmental Science & Technology

ARTICLE

non-thermal plasma on oxidation of acetaldehyde and CO. J. Mol. Catal. A: Chem. 2006, 245 (1
2), 235
241; DOI: 10.1016/j.molcata. 2005.10.002. (26) Chavadej, S.; Kiatubolpaiboon, W.; Rangsunvigit, P.; Sreethawong, T. A Combined multistage corona discharge and catalytic system for gaseous benzene removal. J. Mol. Catal. A: Chem. 2007, 263 (1
2), 128
136; DOI: 10.1016/j.molcata.2006.08.061. (27) Liu, Y. J.; Wang, D. G.; Sun, B.; Zhu, X. M. Aqueous 4-nitrophenol decomposition and hydrogen peroxide formation induced by contact glow discharge electrolysis. J. Hazard. Mater. 2010, 181 (1
3), 1010
1015; DOI: 10.1061/(ASCE)WR.1943-5452.0000099. (28) Suarez, C.; Louys, F.; Gunther, K.; Eiben, K. OH-radical induced denitration of nitrophenols. Tetrahedron Lett. 1970, 11 (8), 575–578. (29) Di Paola, A.; Augugliaro, V.; Palmisano, L.; Pantaleo, G.; Savinov, E. Heterogeneous photocatalytic degradation of nitrophenols. J. Photochem. Photobiol., A 2003, 155 (1
3), 207
214; DOI: 10.1016/ S1010-6030(02)00390-8. (30) Dai, Q. Z.; Lei, L. C.; Zhang, X. W. Enhanced degradation of organic wastewater containing p-nitrophenol by a novel wet electrocatalytic oxidation process: Parameter optimization and degradation mechanism. Sep. Purif. Technol. 2008, 61 (2), 123
129; DOI: 10.1016/j. seppur.2007.10.006. (31) Hong, P. K. A.; Zeng, Y. Degradation of pentachlorophenol by ozonation and biodegradability of intermediates. Water Res. 2002, 36 (17), 4243
4254; DOI: 10.1016/S0043-1354(02)00144-6. (32) Benitez, F. J.; Acero, J. L.; Real, F. J.; Garcia, J. Kinetics of photodegradation and ozonation of pentachlorophenol. Chemosphere 2003, 51 (8), 651
662; DOI: 10.1016/S0045-6535(03)00153-X. (33) Lukes, P.; Locke, B. R. Degradation of substituted phenols in a hybrid gas-liquid electrical discharge reactor. Ind. Eng. Chem. Res. 2005, 44 (9), 2921
2930; DOI: 10.1021/ie0491342.

9307

dx.doi.org/10.1021/es2014314 |Environ. Sci. Technol. 2011, 45, 9301–9307

ARTICLE pubs.acs.org/est

Influence of pH on the Formation of Sulfate and Hydroxyl Radicals in the UV/Peroxymonosulfate System Ying-Hong Guan,† Jun Ma,*,†,‡ Xu-Chun Li,† Jing-Yun Fang,† and Li-Wei Chen† † ‡

State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, Harbin, People's Republic of China National Engineering Research Center of Urban Water Resources, Harbin Institute of Technology, People's Republic of China

bS Supporting Information ABSTRACT: The influence of pH on the degradation of refractory organics (benzoic acid, BA) in UV(254 nm)/Peroxymonosulfate (UV/PMS) system was investigated. The degradation of BA was significantly enhanced at the pH range of 8
11, which could not be explained only by the generally accepted theory that SO4•‑ was converted to HO• at higher pH. A hypothesis was proposed that the rate of PMS photolysis into HO• and SO4•‑ increased with pH. The hypothesis was evidenced by the measured increase of apparent-molar absorption coefficient of PMS (εPMS, 13.8
149.5 M
1 3 cm
1) and photolysis rate of PMS with pH, and further proved by the increased quasi-stationary concentrations of both HO• and SO4•‑ at the pH range of 8
10. The formation of HO• and SO4•‑ in the UV/PMS system was confirmed mainly from the cooperation of the photolysis of PMS, the decay of peroxomonosulfate radical (SO5•‑) and the conversion of SO4•‑ to HO• by simulation and experimental results. Additionally, the apparent quantum yield for SO4•‑ in the UV/PMS system was calculated as 0.52 ( 0.01 at pH 7. The conclusions above as well as the general kinetic expressions given might provide some references for the UV/PMS applications.

’ INTRODUCTION Increasing attention has been paid to the sulfate radical (SO4•‑) due to its high efficiency of mineralization of organic pollutants and its efficient removal of halogen-substituted pollutants.1,2 SO4•‑ is a strong oxidant with a redox potential of 2.5
3.1 V,3 which is similar to hydroxyl radical (HO•) with a redox potential of 1.8
2.7 V.4 Peroxodisulfate (PDS) activated by UV, heat, base, or transition metals is commonly used to generate SO4•‑ and has been widely studied.5
8 The activation of peroxymonosulfate (Oxone, PMS) is also an efficient source of SO4•‑. Recently, many studies related to the activation of PMS have focused on transition metals, among which Co(II) was found to be the best activator.8 But the adverse effects of Co(II) on human health need to be considered. Supported cobalt catalysts were used as heterogeneous activators of PMS to reduce the concentration of free cobalt ion.9 Meanwhile, PMS activated by iron or UV irradiation was considered as an environmentally friendly and applicable technology.10,11 PMS irradiated by UV was proposed to generate HO• and SO4•‑ through the cleavage of the peroxo bond.12 The production of SO4•‑ from PMS irradiated by a pulsed laser at λ = 248 nm was verified by its characteristic absorption at λ = 445 nm.13 The molar absorption coefficients of PMS at λ = 248 nm and at λ = 254 nm available in the form of HSO5
were 19.11 M
1 3 cm
1 and 12.5 or 14 M
1 3 cm
1, respectively.10,13 It was rational to expect that PMS irradiated by a low pressure Hg lamp (λ = 254 nm), could result in the production of SO4•‑ and HO•. r 2011 American Chemical Society

SO4•‑ reacted with HO
to form HO• with a rate constant of 6.5  107 M
1 3 s
1 at alkaline pH.14 It was reported that the degradation of nitrobenzene (NB) was enhanced as pH increased from 7 to 12 in the thermally activated PDS system, which was due to the conversion of SO4•‑ to HO•.15 The conversion also contributed to the increased rate of butylated hydroxyanisole decay as pH increased from 3 to 11 in the UV/ PDS system.5 Meanwhile, the degradation efficiencies of acetic acid and iopromide decreased at the pH range of 9
11 in the UV/PDS system, due to the conversion of SO4•‑ to HO•.1,16 Hence, controlling pH could be considered as one approach to manipulate the degradation of pollutant in the SO4•‑-existing system. However, the mechanism on the pH-dependent degradation of pollutant was rarely studied in UV/PMS system. The aim of this study was to investigate the formation of HO• and SO4•‑ versus pH, which was the fundamental reason for the variation of pollutant degradation rate versus pH, and to present the main factors affecting the formation of HO• and SO4•‑. Benzoic acid (BA) and nitrobenzene (NB) were selected as the probe compounds to investigate the formation of HO• and SO4•‑ versus pH in the UV/PMS system. Received: May 21, 2011 Accepted: September 23, 2011 Revised: September 15, 2011 Published: October 14, 2011 9308

dx.doi.org/10.1021/es2017363 | Environ. Sci. Technol. 2011, 45, 9308–9314

Environmental Science & Technology

ARTICLE

’ EXPERIMENTAL SECTION Materials. Potassium peroxymonosulfate (2KHSO5 3 KHSO4 3 K2SO4 available as Oxone, PMS), potassium peroxodisulfate (PDS), benzoic acid, nitrobenzene, sodium phosphate monobasic monohydrate, sodium phosphate dibasic, sodium tetraborate decahydrate, boric acid, and 5,5-dimethyl-1-pyrrolidine Noxide (DMPO) were of ACS reagent grade and purchased from Sigma-Aldrich, Inc. Phosphoric acid and methanol of HLPC grade were purchased from Dima-Tech Inc. and Thermo Fisher Scientific Inc., respectively. Tert-butyl-alcohol (TBA) was of guaranteed reagent grade and purchased from Tianjin Chemical Reagent Co., Ltd., China. Catalase (from bovine liver) was purchased from Tokyo Kasei Kogyo Co., Ltd. Other reagents used were of analytical-reagent grade and purchased from Sinopharm Chemical Reagent Co., Ltd., China. All of the chemicals were used as received without further purification. All solutions were prepared in 18.2 MΩ 3 cm Milli-Q-water produced on a Milli-Q Biocel water system. Experimental Procedure. All of the photochemical experiments were performed in a 1.3 L cylindrical borosilicate glass vessel (1.2 L samples, 3.5 cm path length). The optical path length (b) was determined to be 4.30 ( 0.05 cm by measuring the photolysis rate of H2O2.17 A low-pressure mercury UV lamp (Heraeus, GPH 135T5 L/4, 6 W nameplate output at 254 nm) was placed in the center of the cylindrical vessel axially along the length of the reactor. The incident radiation intensity of UV lamp (I0) was about 1.5  10
6 Einstein 3 s
1 measured by the method of iodide-iodate chemical actinometer.18 Phosphate/borate buffer with the concentration of 2 mM was used to adjust pH value from 6 to 12 and the pH value during the reaction was measured to be in the range of predetermined pH value (0.1. Samples were withdrawn at predetermined time intervals and quenched using excessive sodium nitrite. Most experiments were conducted in triplicate, at ambient temperature (20 ( 2 °C). Error bars were based on the results of replicate experiments. Details concerning the experimental procedure are provided in Text S1 of the Supporting Information, SI. Analytical Method. The concentrations of BA and NB were determined by high performance liquid chromatography 1525

RQ SC ¼

equipped with a Waters 717 autosampler and a Waters 2487 dual λ detector. Separate column used was a Waters symmetry C18 column (4.6 mm 150 mm, 5 μm particle size). An eluent of water (pH 3, adjusted by phosphoric acid) and methanol (55:45, v/v %) was used to separate BA and its products at a flow rate of 1.0 mL/min. The concentration of BA was quantified at λ = 227 nm. The concentration of NB was analyzed at λ = 263 nm with an eluent of water and methanol (55:45, v/v %). The pH value was measured by a pH meter (PHS-3C, Shanghai Precision & Scientific instrument Co., Ltd.). The solution of PMS was prepared as needed and standardized using iodometric titration.19 The absorption spectra of PMS solution at different pH values were carried out on a Varian Cary 300 spectrometer. EPR experiments were performed on a Bruker A200 spectrometer with DMPO as a spin-trapping agent. Detailed parameters can be seen in SI Text S2. Kinetic Rate Expressions. Table 1 summarizes the photochemical and chemical reactions in the UV/PMS system along with their rate constants. On the basis of the reactions in Table 1, radicals consumed by radical collision induced by HO• and SO4•‑ could be neglected in the UV/PMS system in the presence of about 10 μM BA. When 10 mM TBA was added, less than 1% of HO• reacted with BA. Hence, the degradation of BA by HO• was negligible. The kinetic expression of BA degradation in the UV/ PMS system with the addition of TBA could be expressed as eq 1, based on the pseudosteady state assumption:


dcBA ¼ k12 ½SO•
4 SS cBA þ kd, BA cBA dt ¼ kS0, BA cBA þ kd, BA cBA

ð1Þ

where [SO4•‑]ss is defined as the quasi-stationary concentration of SO4•‑, kS0,BA is defined as k12[SO4•‑]ss, kd,BA is the first-order rate constant of the direct photolysis of BA. Assuming that HO• and SO4•‑ were formed from the photolysis of PMS, the decay of peroxomonosulfate radical (SO5•‑) and the conversion of SO4•‑ to HO•, the relative quasi-stationary concentration of SO4•‑ (RQSC) could be derived as eq 2 (details can be seen in SI Text S3).

½SO•
1
10
A 10
pH 10
pKa1 4 SS ¼ cPMS b εHSO
5 þ
pKa1 ε 2
ϕI0 =V A 10
pKa1 þ 10
pH 10 þ 10
pH SO5

!

! 11 10
pH 10
pKa1 k2
pKa1 þ k cPMS 3 6 10 þ 10
pH 10
pKa1 þ 10
pH !   1 10
pH 10
pKa1 pH
14 k2
pKa1 cPMS þ k16 cTBA Þ þ k12 cBA þ k17 cTBA þ k5 10 þ k3
pKa1 6 10 þ 10
pH 10 þ 10
pH

k11 cBA þ k16 cTBA þ

 ðk12 cBA þ k17 cTBA þ k5 10pH
14 Þðk11 cBA

ð2Þ The variations of kS0,BA and RQSC (based on eqs 1 and 2) with pH were compared to testify the hypothesis that HO• and SO4•‑ were formed from the cooperation of the photolysis of PMS, the decay of SO5•‑ and the conversion of SO4•‑ to HO•. In the UV/PMS system, NB and BA were used simultaneously as probe compounds to indicate the variations of the formation rates of HO• and SO4•‑ (FHO 3 and FSO43 ‑) with pH. FHO 3 and FSO43 ‑ were defined as eqs 3 and 4. On the basis of the pseudosteady

state assumption that the quantity of radicals formed was equal to the quantity of radicals consumed, the sum of FHO 3 and FSO43 ‑ (Ftotal) in the UV/PMS system could be derived as eq 5 (details can be seen in SI Text S4).

9309

FHO• ¼ PPMS þ QHO•

ð3Þ

FSO4 •
¼ PPMS þ TSO4 •

QSO4 •


ð4Þ

dx.doi.org/10.1021/es2017363 |Environ. Sci. Technol. 2011, 45, 9308–9314

Environmental Science & Technology

ARTICLE

Table 1. Principal Reactions in the UV/PMS System no.

references

1

HSO5
=SO25
sf SO•4
þ HO•

r = ϕI0bεPMScPMS(1
10
A)/(AV)a

20

2

HO• þ HSO5
f SO•5
þ H2 O

k2 = 1.7  107

21

k3 = 2.1  10

21

hv

•

3

HO þ

4

SO•4


5

SO•4


SO25


þ

HO þ

f

HSO5


þ HO

•

6

a

rate constants (M
1 3 s
1)

reactions




HSO4


•

SO•5
f

þ HO

SO•5
•

þ

f HO þ SO•4


f




HSO4


9

k4 < 10

5

SO24


21

k5 = 6.5  10

þ H2 O

•

7

14

k6 = 6.9  105

4

7

HO þ HO f H2 O2

k7 = 5.5  109

4

8

SO•4
þ SO•4
f S2 O28


k8 = 3.1  10

3

•

SO•4


9

HO þ

10a

SO•5


10b

SO•5


11

in the presence of BA HO• þ C6 H5 COO
f product

k11 = 5.9  109

4

12

SO•4


k12 = 1.2  10

3

k13 ≈ 4  107

4

•


f

HSO5


8

þ

SO•5


þ

SO•5


f

SO•4


f

S2 O28


þ C6 H5 COO þ C6 H5 COO







k9 = 1  10

10

þ

SO•4


þ O2

k10 = 1  10

22

8

23

þ O2

f product

9

f product

13

O

14

in the presence of NB HO• þ C6 H5 NO2 f product

k14 = 3.9  109

4

15

SO•4


k15 e 106

3

16

in the presence of TBA HO• þ ðCH3 Þ3 COH f product

k16 = 6.0  108

4

17

SO•4


5

k17 = 4.0  10

3

18

in the presence of methanol HO• þ CH3 OH f product

k18 = 9.7  108

4

19

SO•4


k19 = 3.2  10

3

20

HSO5


pKa1 = 9.4

24

pKa2 = 11.9

4

pKa3 = 4.2

25

•

þ C6 H5 NO2 f product

þ ðCH3 Þ3 COH f product

þ CH3 OH f product S

SO25


•


þ H

þ H

6

þ

þ

21

HO S O

22

C6 H5 COOH S C6 H5 COO
þ H þ

A is the absorbance of solution.

Ftotal ¼ FHO• þ FSO4 •
¼ RBA þ RNB ! 10
pH 10
pKa1 k2 þ
pKa1 k3 cPMS 10
pKa1 þ 10
pH 10 þ 10
pH

þ

k14 cNB

RNB

ð5Þ PPMS ¼

þ

’ RESULTS AND DISCUSSION

1 1 RBA þ RNB 2 2

1 12

where FHO 3 and FSO43 ‑ are the formation rates of HO• and SO4•‑, PPMS is the rate of PMS photolysis into HO• and SO4•‑, QHO 3 and QSO43 ‑ are the production rate of HO• and the consumption rate of SO4•‑ through the conversion of SO4•‑ to HO•, TSO43 ‑ is the production rate of SO4•‑ from the decay of SO5•‑, Ftotal is the sum of the formation rates of HO• and SO4•‑, RNB and RBA are the consumption rates of NB and BA.

! 10
pH 10
pKa1 k2 þ
pKa1 k3 cPMS 10
pKa1 þ 10
pH 10 þ 10
pH k14 cNB

RNB

ð6Þ

Effect of pH on BA Degradation. BA was used as a recalcitrant organic compound to investigate the decontamination effect in the UV/PMS system at the pH range of 6
12. The degradation of BA was found to be enhanced significantly as pH increased from 8 to 11 (Figure 1a). Pseudofirst-order kinetic model fitted very well to the degradation of BA within the initial 9310

dx.doi.org/10.1021/es2017363 |Environ. Sci. Technol. 2011, 45, 9308–9314

Environmental Science & Technology

ARTICLE

Figure 2. The εPMS, speciation of PMS, and decomposition of PMS in the UV/PMS system in the presence of TBA at different pH (condition: [PMS] = 1 mM as 1/2 Oxone; [TBA] = 100 mM; irradiation time: 10 min; error bar represents a confidence interval with a confidence of 0.95).

Figure 1. (a) Degradation of BA in the UV/PMS system at different pH. Inset indicates kinetics of BA degradation versus pH. (b) Pseudofirst-order rate constants versus pH in the UV/PMS system. (conditions: [BA] = 9.90 μM; [PMS] = 100 μM as 1/2 Oxone; error bar represents a confidence interval with a confidence of 0.95).

1.5 min (Inset of Figure 1a), though the BA degradation in the UV/PMS system was complex and might not follow a pseudofirst-order kinetic model for longer experimental time in theory. By fitting each pH series for pseudofirst-order kinetics, pseudofirst-order rate constants (k0,BA) were obtained and plotted as the function of pH (Figure 1b). k0,BA kept almost invariant at the pH range from 6 to 8, increased significantly from 8 to 11, and dropped from 11 to 12. No obvious degradation of BA by PMS was observed in the investigated time scale. The degradation of BA by direct photolysis of UV, with kd,BA of (5.80 ( 0.31)  10
5 s
1, could be negligible compared with that achieved by UV/PMS (SI Figure S3). This indicated that UV and PMS should have a synergistic effect on the degradation of BA. PMS was reported to produce HO• and SO4•‑ by pulsed laser irradiation at λ = 248 nm.13 The formation of HO• and SO4•‑ was also checked in the present system (λ = 254 nm). Radical-scavenging (TBA and methanol) experimental results and EPR spectra (SI Figures S3 and S4) indicated that both HO• and SO4•‑ contributed to BA degradation (details can be seen in SI Text S5). SO4•‑ could react with HO
to form HO• at higher pH.14 In the present study, both HO• and SO4•‑ were mainly consumed by BA at neutral pH and HO• was also consumed by PMS forming SO5•‑ at basic pH. Meanwhile, SO5•‑ was reported to partly decay into SO4•‑.23 However, the conversion of SO4•‑ to HO• and the decay of SO5•‑ to SO4•‑ would not result in the increased efficiency of BA degradation as pH increased, which was not in accordance with the experimental results (Figure 1a,b) that a sharp increase was observed at the pH range of 8
11. Apparent-Molar Absorption Coefficient (εPMS) and Speciation of PMS versus pH. The absorption spectra curve of PMS shift to right as pH increased from 8 to 11 (SI Figure S5). The apparent-molar absorption coefficient (εPMS) was calculated

from the absorbance of PMS at λ = 254 nm and shown in Figure 2. The εPMS increased with pH from 13.8 to 149.5 M
1 3 cm
1 in the pH range of 6
12. It was similar to the variation of εH2O2 that increased from 19.6 to 229 M
1 3 cm
1 as H2O2 dissociated into HO2
.17 The εPMS of 13.8 M
1 3 cm
1 (existing in HSO5
form) obtained in this study was also in accordance with the value of 14 M
1 3 cm
1 reported previously.13 The parent acid of PMS, H2SO5, was reported to be a strong acid as sulfuric acid.26 Hence, the speciation of PMS at studied pH range was calculated based on its second pKa (Table 1) and shown in Figure 2. PMS mainly exists in its monoanion form (HSO5
) at the pH range of 6
8 and its dianion form (SO52‑) at pH g 11. The εPMS correlated closely with the speciation of PMS. Assuming the quantum yield was constant for dissociated and undissociated species, it was reasonable to infer that the photolysis rate of PMS should increase significantly with pH near its second pKa and the increased photolysis rate of PMS into HO• and SO4•‑ would contribute to the sharply enhanced degradation of BA at pH around 9.4. Decomposition of PMS versus pH. HO• was reported to induce the acceleration of PMS decomposition.21 Thus, the decomposition of PMS in the UV/PMS system might consist of three parts: the photolysis by UV irradiation, decomposition by HO• or other radicals attack, and spontaneous decomposition. PMS was unstable and decomposed to H2O2 at basic pH.7 Catalase was used to quench H2O2 if produced.27 The results showed that no observable quantity of H2O2 was produced and the spontaneous decomposition of PMS was found to be negligible (e3%) under the given condition (details can be seen in SI Text S6). TBA was reported to be a good radical scavenger to reduce the decomposition of H2O2 by radical attack.28 Also, TBA was selected to reduce the decomposition of PMS by radical attack rather than methanol in the present study (see SI Text S6 for details). Therefore, the decomposition of PMS in the UV/ PMS system with the addition of TBA was used to indicate the photolysis of PMS, although slight decomposition of PMS by radical attack might exist. Figure 2 shows that C/C0,PMS decreased as pH increased and a sharp decrease was observed at pH around 9.4. This indicated that a sharp increase of photolysis of PMS was obtained at pH around its second pKa, which strengthened the inference above. Quasi-stationary concentration of SO4•‑ (QSCSO43 ‑) versus pH. In order to investigate the variation of QSCSO43 ‑ with 9311

dx.doi.org/10.1021/es2017363 |Environ. Sci. Technol. 2011, 45, 9308–9314

Environmental Science & Technology

Figure 3. (a) kS0,BA and RQSC versus pH in the UV/PMS system with the addition of TBA (conditions: [BA] = 9.90 μM; [PMS ] = 100 μM as 1/2 Oxone; [TBA] = 10 mM; error bar represents a confidence interval with a confidence of 0.95). (b) k0,NB versus pH in the UV/PMS system (conditions: [NB] = 18.07 μM; [BA] = 9.90 μM; [PMS] = 100 μM as 1/2 Oxone; error bar represents a confidence interval with a confidence of 0.95).

pH, kS0,BA of BA degradation in the presence of TBA was obtained based on eq 1 by fitting each pH series for pseudofirst-order kinetics (SI Figure S7). Figure 3a shows that kS0,BA was almost invariant at the pH range of 6
8, increased sharply from 8 to 10, and decreased obviously from 10 to 12. Hence, QSCSO43 ‑ varied in the same way as kS0,BA did over the corresponding pH range. Quasi-stationary Concentration of HO• (QSCHO 3 ) versus pH. Reaction between SO4•‑ and NB was quite slow (Table 1), which was also proven in SI Figure S8 (detail can be seen in SI Text S7). The first-order rate constant of direct photolysis of NB (kd,
5
1 s and the degradation of NB NB) was (1.8531 ( 0.0992)  10 by UV could be negligible compared with that achieved by UV/ PMS. NB was thus selected as the probe compound to indicate the variation of QSCHO 3 with pH in the UV/PMS system in the presence of BA. Figure 3b shows that the pseudofirst-order rate constant (k0,NB) of NB degradation was almost invariant at the pH range of 6
8 and increased with pH from 8 to 12. The dissociation of HO• into O•‑ became obvious at pH g 11, which would result in the change of apparent second-order rate constant (the weighted average of the rate constants of HO• and O•‑ with NB). Assuming that the rate constant of the reaction between NB and O•‑ was no more than that of HO• and NB,4 the conclusion could be derived that the QSCHO 3 (the sum of quasi-stationary concentrations of HO• and O•‑) kept almost invariant at the pH range of 6
8 and increased with pH from 8 to 12. Role of the Photolysis of PMS, the Decay of SO5•‑ and the Conversion of SO4•‑ to HO• in the Formation of HO• and SO4•‑. The almost invariant QSCHO 3 and QSCSO43 ‑ with pH increasing from 6 to 8, were in accordance with the speciation of PMS, the variation of εPMS, and the almost unchanged C/C0, PMS at the same pH range (Figure 2). The increase of both QSCHO 3 and QSCSO43 ‑ with pH from 8 to 10 confirmed the

ARTICLE

conjecture that the rate of PMS photolysis into HO• and SO4•‑ increased with pH around its second pKa. Theoretically, the production of SO4•‑ from the photolysis of PMS should keep increasing with pH from 10 to 11 and remain stable at pH > 11 according to the variation of εPMS. Besides, the decay of SO5•‑ to SO4•‑ would also contribute to the increase of QSCSO43 ‑. However, QSCSO43 ‑ decreased as pH increased from 10 to 12 (Figure 3a). It might result from the conversion of SO4•‑ to HO•, which became apparent at pH > 9.3 (more than 10% SO4•‑ converted to HO•) when the initial concentration of BA was 10 μM. RQSC was then deduced from eq 2 and plotted in Figure 3a, based on a simplified hypothesis that HO• and SO4•‑ were formed mainly from the photolysis of PMS, the decay of SO5•‑ and the conversion of SO4•‑ to HO•. The simulation results of RQSC showed the same trends with the variation of kS0,BA as pH varied. But the ratios of maximum value to minimum value of kS0,BA and simulated RQSC were not consistent. It might be due to the omission of the competition of intermediate products for radicals in the simulation model for simplification. This would lead to a higher estimated value of RQSC than the actual value and the difference was especially obvious around pH 10 where BA was degraded quickly. Besides, some intermediate radical products such as superoxide radical and semiquinone might promote the decomposition of PMS and the production of radicals by one electron transfer,3,29
32 which made the reactions in the studied system more complicated. However, the simplified simulation results of RQSC showed the same trends with the experimental results as pH changed, which confirmed that the formation of HO• and SO4•‑ was mainly due to the photolysis of PMS, the decay of SO5•‑ and the conversion of SO4•‑ to HO• in the UV/PMS system. It could also be obtained from Figure 3a that the apparent quantum yield of SO4•‑ (ϕSO43 ‑) at pH 7 was 0.52 ( 0.01 in the present system (λ = 254 nm). The apparent quantum yield for both HO• and SO4•‑ was estimated to be 1.04 based on the assumption that HO• and SO4•‑ were produced equally by the photolysis of PMS, which was close to the value of 1.0 for HO• from UV/H2O2 at λ = 254 nm.17 The ϕSO43 ‑ from UV/PMS previously reported was available at λ = 248 nm and it was reported to be 0.12.13 The large difference of ϕSO43 ‑ in the UV/ PMS system between the present and previous study might be due to the SO4•‑ sink through the conversion of SO4•‑ to HO•, which was taken into consideration in the present study but omitted in the previous study. Furthermore, QSCHO 3 and QSCSO43 ‑ were calculated to be in the magnitude of 10
13
10
12 M for SO4•‑ and 10
14
10
13 M for HO•, which indicated that the omission of the radical combination reactions in the UV/PMS system in the presence of BA was reasonable (details see SI Text S8). PPMS versus εPMS. On the basis of the hypothesis that the formation of HO• and SO4•‑ was mainly due to the photolysis of PMS, the decay of SO5•‑ and the conversion of SO4•‑ to HO•, the rate of PMS photolysis into HO• and SO4•‑ (PPMS) could be derived as eq 6. Then, the calculation results of PPMS would change in proportion to εPMS in the UV/PMS system (εNB = 6200 M
1 3 cm
1 and εBA = 760 M
1 3 cm
1 in the form of benzoate ion). As shown in Figure 4, PPMS was in direct proportion to εPMS with R2 = 0.9992, which further strengthened the hypothesis that the formation of HO• and SO4•‑ mainly resulted from the photolysis of PMS, the decay of SO5•‑ and the conversion of SO4•‑ to HO• in the UV/PMS system. Furthermore, the ϕSO43 ‑ from the photolysis of PMS was estimated to be 9312

dx.doi.org/10.1021/es2017363 |Environ. Sci. Technol. 2011, 45, 9308–9314

Environmental Science & Technology

ARTICLE dcSO•
5 ¼ dt

! 10
pH 10
pKa1 k2 þ
pKa1 k3 cHO• cPMS
2k10 c2SO•
5 10
pKa1 þ 10
pH 10 þ 10
pH

ð9Þ
ðεPMS cPMS þ ∑ εi ci Þb

PPMS

Figure 4. PPMS and k0,BA versus εPMS in the UV/PMS system (error bar represents a confidence interval with a confidence of 0.95).

0.35 from the slope of the fitted line in Figure 4, which was smaller than the value of 0.52 ( 0.01 for ϕSO43 ‑ obtained above at pH 7. The faster decomposition of PMS and the more quantity of intermediates produced at higher pH might lead to the lower value of ϕSO43 ‑ obtained by the fitted line in Figure 4. k0,BA versus εPMS. On the basis of the pseudosteady state assumption, k0,BA should increase in proportion to εPMS, since absorbed quanta of irradiation could be simplified to be proportional to εPMS (the maximum absorbance of solution was about 0.06 and the maximum error induced by simplification was less than 8%). But k0,BA did not increase as assumed. As shown in Figure 4, the increase of k0,BA slowed down obviously as εPMS g 118 M
1 3 cm
1 (Figure 4). It might be due to the increased consumption of HO• by PMS as εPMS increased because the rate constant of the reaction between HO• and PMS was reported to increase as HSO5
dissociated to SO52‑ (Table 1). Meanwhile, the decay of SO5•‑ to SO4•‑ might contributed to the increase of k0,BA. Considering the two factors simultaneously, the termination reaction of SO5•‑ (reaction 10b) might be the main reason for the nonproportional increase of k0,BA with εPMS.33 Besides, the lower rate constant of the reaction between O•‑ and BA (Table 1) was an important reason for the significant decrease of k0,BA when εPMS was more than 146 M
1 3 cm
1 (pH g 11). The formation of HO• and SO4•‑ in the UV/PMS system was mainly due to the photolysis of PMS, the decay of SO5•‑ and the conversion of SO4•‑ to HO•. However, the latter was actually influenced by probe compounds (BA in the present system). Thus, it was necessary to give general kinetic expressions (eqs 7
10) to depict the variation of the concentrations of HO• and SO4•‑ under neutral or basic pH (pH e 11) in the application to natural water. The general kinetic expressions were examined by simulation results of the pseudofirst-order rate constant of BA degradation (k0,BA) in the UV/PMS system. The simulation result were basically coincident with the experimental results as pH changed as shown in SI Figure S11. dcHO• ¼ PPMS þ k5 cSO•
10pH
14 4 dt

∑i ki, HO ci cHO •

•

dcSO•
10 4 ¼ PPMS þ k10 c2SO•

k5 cSO•
10pH
14 4 5 6 dt
ki, SO•
ci cSO•
4 4

∑i

i

∑i εi ci Þ

Þ

ð10Þ

where PPMS is the rate of PMS photolysis into HO• and SO4•‑, I0 is the incident radiation intensity, ϕ is the apparent quantum yield of PMS photolysis into HO• and SO4•‑ (0.435 is used in the simulation of BA degradation for it is the average of 0.52 and 0.35 obtained in this study), ci is the concentration of probe compound i, εi is the molar absorption coefficient of probe compound i, ki,HO 3 and ki,SO43 ‑ are the second-order rate constants of probe compound i with HO• and SO4•‑. In summary, the formation of HO• and SO4•‑ in the UV/PMS system was increased with pH at the pH range of 8
10. It was quite different from the increased formation of HO• and the decreased formation of SO4•‑ with the increase of pH at the corresponding pH range in the UV/PDS system. Therefore, PMS could be more suitable for the application to enhance the degradation of organic matter under basic condition. Comparing with H2O2, PMS is easier to be transported and more effective in the degradation of some kinds of organics,2 although not as environmentally friendly as H2O2 and a bit more expensive than H2O2.33 The conclusions derived from the present study, as well as the general kinetic expressions (eqs 7
10) would provide some references for practical applications.

’ ASSOCIATED CONTENT

bS Supporting Information. Additional texts and figures. This material is available free of charge via the Internet at http:// pubs.acs.org. ’ AUTHOR INFORMATION Corresponding Author

*Phone: 86-451-86283010; fax: 86-451-86282292; e-mail: majun@ hit.edu.cn.

’ ACKNOWLEDGMENT The support from the Natural Science Foundation of China (No. 50821002), the 863 high tech. scheme (No. 2009AA06Z310), the special S&T project on water treatment and control of pollution (2009ZX07424-005, 2009ZX07424006), and SKLUWRE (No. 2010DX10) are greatly appreciated. ’ REFERENCES

!

10
pH 10
pKa1
k2 þ
pKa1 k3 cHO• cPMS 10
pKa1 þ 10
pH 10 þ 10
pH


I0 ϕεPMS cPMS ð1
10 ¼ V ðεPMS cPMS þ

ð7Þ

ð8Þ

(1) Criquet, J.; Leitner, N. K. V. Degradation of acetic acid with sulfate radical generated by persulfate ions photolysis. Chemosphere 2009, 77 (2), 194–200, DOI: 10.1016/j.chemosphere.2009.07.040. (2) Hori, H.; Yamamoto, A.; Hayakawa, E.; Taniyasu, S.; Yamashita, N.; Kutsuna, S. Efficient decomposition of environmentally persistent perfluorocarboxylic acids by use of persulfate as a photochemical oxidant. Environ. Sci. Technol. 2005, 39 (7), 2383–2388, DOI: 10.1021/es0484754. (3) Neta, P.; Huie, R. E.; Ross, A. B. Rate constants for reactions of inorganic radicals in aqueous solution. J. Phys. Chem. Ref. Data 1988, 17 (3), 1027–1284, DOI: 10.1063/1.555808. 9313

dx.doi.org/10.1021/es2017363 |Environ. Sci. Technol. 2011, 45, 9308–9314

Environmental Science & Technology (4) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (•OH/•O
) in aqueous solution. J. Phys. Chem. Ref. Data 1988, 17 (2), 513–886, DOI: 10.1063/1.555805. (5) Lau, T. K.; Chu, W.; Graham, N. J. D. The aqueous degradation of butylated hydroxyanisole by UV/S2O82‑: study of reaction mechanisms via dimerization and mineralization. Environ. Sci. Technol. 2007, 41 (2), 613–619, DOI: 10.1021/es061385a. (6) Waldemer, R. H.; Tratnyek, P. G.; Johnson, R. L.; Nurmi, J. T. Oxidation of chlorinated ethenes by heat-activated persulfate: kinetics and products. Environ. Sci. Technol. 2007, 41 (3), 1010–1015, DOI: 10.1021/es062237m. (7) Furman, O. S.; Teel, A. L.; Watts, R. J. Mechanism of base activation of persulfate. Environ. Sci. Technol. 2010, 44 (16), 6423–6428, DOI: 10.1021/es1013714. (8) Anipsitakis, G. P.; Dionysiou, D. D. Radical generation by the interaction of transition metals with common oxidants. Environ. Sci. Technol. 2004, 38 (13), 3705–3712, DOI: 10.1021/es035121o. (9) Yang, Q. J.; Choi, H.; Chen, Y. J.; Dionysiou, D. D. Heterogeneous activation of peroxymonosulfate by supported cobalt catalysts for the degradation of 2,4-dichlorophenol in water: The effect of support, cobalt precursor, and UV radiation. Appl. Catal., B 2008, 77 (3
4), 300–307, DOI: 10.1016/j.apcatb.2007.07.020. (10) Rivas, J.; Gimeno, O.; Borralho, T.; Carbajo, M. UV-C photolysis of endocrine disruptors. The influence of inorganic peroxides. J. Hazard. Mater. 2010, 174 (1
3), 393–397, DOI: 10.1016/j. jhazmat.2009.09.065. (11) Rastogi, A.; Al-Abed, S. R.; Dionysiou, D. D. Effect of inorganic, synthetic and naturally occurring chelating agents on Fe(II) mediated advanced oxidation of chlorophenols. Water Res. 2009, 43 (3), 684–694, DOI: 10.1016/j.watres.2008.10.045. (12) Anipsitakis, G. P.; Dionysiou, D. D. Transition metal/UVbased advanced oxidation technologies for water decontamination. Appl. Catal., B 2004, 54 (3), 155–163, DOI: 10.1016/j.apcatb.2004.05.025. (13) Herrmann, H. On the photolysis of simple anions and neutral molecules as sources of O
/OH, SOx
and Cl in aqueous solution. Phys. Chem. Chem. Phys. 2007, 9 (30), 3935–3964, DOI: 10.1039/b618565g. (14) Hayon, E.; Treinin, A.; Wilf, J. Electronic spectra, photochemistry, and autoxidation mechanism of the sulfite-bisulfite-pyrosulfite systems. SO2
, SO3
, SO4
, and SO5
radicals. J. Am. Chem. Soc. 1972, 94 (1), 47–57, DOI: 10.1021/ja00756a009. (15) Liang, C. J.; Su, H. W. Identification of sulfate and hydroxyl radicals in thermally activated persulfate. Ind. Eng. Chem. Res. 2009, 48 (11), 5558–5562, DOI: 10.1021/Ie9002848. (16) Chan, T. W.; Graham, N. J. D.; Chu, W. Degradation of iopromide by combined UV irradiation and peroxydisulfate. J. Hazard. Mater. 2010, 181 (1
3), 508–513, DOI: 10.1016/j.jhazmat.2010.05.043. (17) Baxendale, J. H.; Wilson, J. A. The photolysis of hydrogen peroxide at high light intensities. Trans. Faraday Soc. 1957, 53, 344–356, DOI: 10.1039/TF9575300344. (18) Rahn, R. O.; Stefan, M. I.; Bolton, J. R.; Goren, E.; Shaw, P. S.; Lykke, K. R. Quantum yield of the iodide-iodate chemical actinometer: dependence on wavelength and concentration. Photochem. Photobiol. 2003, 78 (2), 146–152, DOI: 10.1562/0031-8655(2003). (19) Ball, R. E.; Edwards, J. O.; Haggett, M. L.; Jones, P. A kinetic and isotopic study of the decomposition of monoperoxyphthalic acid. J. Am. Chem. Soc. 1967, 89 (10), 2331–2333, DOI: 10.1021/ja00986a015. (20) Crittenden, J. C.; Hu, S. M.; Hand, D. W.; Green, S. A. A kinetic model for H2O2/UV process in a completely mixed batch reactor. Water Res. 1999, 33 (10), 2315–2328, DOI: 10.1016/S0043-1354(98) 00448-5. (21) Maruthamuthu, P.; Neta, P. Radiolytic chain decomposition of peroxomonophosphoric and peroxomonosulfuric acids. J. Phys. Chem. 1977, 81 (10), 937–940, DOI: 10.1021/j100525a001. (22) Klaning, U. K.; Sehested, K.; Appelman, E. H. Laser flash photolysis and pulse radiolysis of aqueous solutions of the fluoroxysulfate ion, SO4F
. Inorg. Chem. 1991, 30 (18), 3582–3584, DOI: 10.1021/ ic00018a040.

ARTICLE

(23) Huie, R. E.; Clifton, C. L.; Altstein, N. A pulse radiolysis and flash photolysis study of the radicals SO2
, SO3
, SO4
, and SO5
. Int. J. Radiat. Appl. Instrum. C Radiat. Phys. Chem. 1989, 33 (4), 361–370, DOI: 10.1016/1359-0197(89)90034-9. (24) Rani, S. K.; Easwaramoorthy, D.; Bilal, I. M.; Palanichamy, M. Studies on Mn(II)-catalyzed oxidation of alpha-amino acids by peroxomonosulphate in alkaline medium-deamination and decarboxylation: a kinetic approach. Appl. Catal., A 2009, 369 (1
2), 1–7, DOI: 10.1016/j. apcata.2009.07.048. (25) Tao, L.; Han, J.; Tao, F. M. Correlations and predictions of carboxylic acid pka values using intermolecular structure and properties of hydrogen-bonded complexes. J. Phys. Chem. A 2008, 112 (4), 775–782, DOI: 10.1021/jp710291c. (26) Lawrance, G. A.; Ward, C. B. Kinetics of oxidation of manganese(II) by peroxomonosulfuric acid in aqueous acidic solution. Transition Met. Chem. 1985, 10 (7), 258–261, DOI: 10.1007/ bf00621082. (27) Liu, W. J.; Andrews, S. A.; Stefan, M. I.; Bolton, J. R. Optimal methods for quenching H2O2 residuals prior to UFC testing. Water Res. 2003, 37 (15), 3697–3703, DOI: 10.1016/s0043-1354(03)00264-1. (28) Popov, E.; Mametkuliyev, M.; Santoro, D.; Liberti, L.; Eloranta, J. Kinetics of UV-H2O2 advanced oxidation in the presence of alcohols: The role of carbon centered radicals. Environ. Sci. Technol. 2010, 44 (20), 7827–7832, DOI: 10.1021/es101959y. (29) Peiro, A. M.; Ayllon, J. A.; Peral, J.; Domenech, X. TIO2photocatalyzed degradation of phenol and ortho-substituted phenolic compounds. Appl. Catal., B 2001, 30 (3
4), 359–373, DOI: 10.1016/ S0926-3373(00)00248-4. (30) Anipsitakis, G. P.; Dionysiou, D. D.; Gonzalez, M. A. Cobaltmediated activation of peroxymonosulfate and sulfate radical attack on phenolic compounds. Implications of chloride ions. Environ. Sci. Technol. 2006, 40 (3), 1000–1007, DOI: 10.1021/Es050634b. (31) Weinstein, J.; Bielski, B. H. J. Kinetics of the interaction of perhydroxyl and superoxide radicals with hydrogen peroxide. The Haber-Weiss reaction. J. Am. Chem. Soc. 1979, 101 (1), 58–62, DOI: 10.1021/ja00495a010. (32) Koppenol, W. H.; Butler, J. Energetics of interconversion reactions of oxyradicals. Adv. Free Radic. Biol. Med. 1985, 1 (1), 91–131, DOI: 10.1016/8755-9668(85)90005-5. (33) Anipsitakis, G. P.; Dionysiou, D. D. Degradation of organic contaminants in water with sulfate radicals generated by the conjunction of peroxymonosulfate with cobalt. Environ. Sci. Technol. 2003, 37 (20), 4790–4797, DOI: 10.1021/Es0263792.

9314

dx.doi.org/10.1021/es2017363 |Environ. Sci. Technol. 2011, 45, 9308–9314

ARTICLE pubs.acs.org/est

Randomized Intervention Study of Solar Disinfection of Drinking Water in the Prevention of Dysentery in Kenyan Children Aged under 5 Years Martella du Preez,† Ronan M. Conroy,‡ Sophie Ligondo,§ James Hennessy,§ Michael Elmore-Meegan,§ Allan Soita,§ and Kevin G. McGuigan*,|| †

Natural Resources and the Environment, CSIR, P.O. Box 395, Pretoria, South Africa Division of Population Health Sciences, Royal College of Surgeons in Ireland, 123 St Stephens Green, Dublin 2, Ireland § ICROSS, P.O. Box 507, Kenya, Ngong Hills, Kenya Department of Physiology & Medical Physics, Royal College of Surgeons in Ireland, 123 St Stephens Green, Dublin 2, Ireland

)

‡

bS Supporting Information ABSTRACT: We report the results of a randomized controlled intervention study (September 2007 to March 2009) investigating the effect of solar disinfection (SODIS) of drinking water on the incidence of dysentery, nondysentery diarrhea, and anthropometric measurements of height and weight among children of age 6 months to 5 years living in peri-urban and rural communities in Nakuru, Kenya. We compared 555 children in 404 households using SODIS with 534 children in 361 households with no intervention. Dysentery was recorded using a pictorial diary. Incidence rate ratios (IRR) for both number of days and episodes of dysentery and nondysentery diarrhea were significantly (P < 0.001) reduced by use of solar disinfection: dysentery days IRR = 0.56 (95% CI 0.40 to 0.79); dysentery episodes IRR = 0.55 (95% CI 0.42 to 0.73); nondysentery days IRR = 0.70 (95% CI 0.59 to 0.84); nondysentery episodes IRR = 0.73 (95% CI 0.63 to 0.84). Anthropometry measurements of weight and height showed median height-for-age was significantly increased in those on SODIS, corresponding to an average of 0.8 cm over a 1-year period over the group as a whole (95% CI 0.7 to 1.6 cm, P = 0.031). Median weight-for-age was higher in those on SODIS, corresponding to a 0.23 kg difference in weight over the same period; however, the confidence interval spanned zero and the effect fell short of statistical significance (95% CI 0.02 to 0.47 kg, P = 0.068). SODIS and control households did not differ in the microbial quality of their untreated household water over the follow-up period (P = 0.119), but E. coli concentrations in SODIS bottles were significantly lower than those in storage containers over all follow-up visits (P < 0.001). This is the first trial to show evidence of the effect of SODIS on childhood anthropometry, compared with children in the control group and should alleviate concerns expressed by some commentators that the lower rates of dysentery associated with SODIS are the product of biased reporting rather than reflective of genuinely decreased incidence.

’ INTRODUCTION Although a preventable and treatable disease, nearly 1.8 million children under 5 years of age die from diarrhea each year.1 The World Health Organization estimates that in 94% of cases diarrhea is preventable by increasing the availability of clean water and improving sanitation and hygiene.1 Diarrheal disease is strongly linked to fecal contamination. Contamination can occur at source or within the storage container during transport or storage.2 Recontamination may also occur if the drinking utensils are not subject to a regular hygiene regimen.3,4 The prohibitive cost of universally supplying piped water has made household water treatment (HWT) an attractive alternative worldwide. Reviews of the effectiveness of HWT methods5 7 have confirmed that in-home interventions, such as filtration 8,9 chlorination,10,11 a combination of flocculation and chlorination,12,13 and solar disinfection 14 18 can reduce the incidence of diarrhea substantially. r 2011 American Chemical Society

The fundamental principles of one of the simplest and cheapest HWT, solar disinfection (SODIS), were first discussed in 1877 by Downes and Blunt.19 Acra and his colleagues from the American University of Beirut laid the foundations of current research on SODIS with their work on solar irradiation of water and oral rehydration solutions in 1980.14,20,21 More recent laboratory studies have consistently shown that exposing water to sunlight results in significant reduction in bacterial contamination.22 26 However, there are still relatively few controlled field trials to show that this reduction in bacterial levels translates into a reduction in risk of disease in people. Initial trials in Received: June 2, 2011 Accepted: September 21, 2011 Revised: September 13, 2011 Published: September 21, 2011 9315

dx.doi.org/10.1021/es2018835 | Environ. Sci. Technol. 2011, 45, 9315–9323

Environmental Science & Technology Kenyan children reported that solar disinfection was associated with a significant reduction in the risk of diarrheal disease in children aged 5 and under16 and in older children,15 and a further study reported a significant reduction of risk of cholera in children.17 A study by Rose and his colleagues in India in children under 527 showed a significant reduction in risk which occurred despite 86% of the children drinking water other than the solar disinfected water. Rai and co-workers showed a reduction of childhood diarrhea by approximately 76% in an urban population of 136 children under age 5 in North Eastern India.28 A recent trial of solar disinfection in Bolivia by M€ausezahl and colleagues in a setting of very low compliance (32%) failed to show a statistically significant reduction in diarrheal disease, although a reduction in diarrhea was observed for both the test and control communities.29 A study of SODIS in a South African peri-urban environment by du Preez and co-workers in 200930 also reported low compliance levels. Dysentery incidence rates were, however, lower in those drinking solar disinfected water (incidence rate ratio 0.64, 95% CI 0.39 1.0, P = 0.071) but not statistically significantly so. Compared with the control, only participants with higher motivation (defined as adhering to the study protocol at least 75% of the time) achieved a significant reduction in dysentery (incidence rate ratio 0.36, 95% CI 0.16 0.81, P = 0.014). There was no significant reduction in risk at lower levels of motivation. These two studies underline the importance of participant motivation in translating the bactericidal effects of SODIS into health gains. The published research has also some deficiencies. All published trials to date have been carried out on children; there are no trials of the effect of solar disinfection in populations of adults at high risk of water-borne diseases, such as the elderly or those with compromised immune function. Previous Kenyan trials were all carried out in populations drinking heavily contaminated water with high levels of disease risk.31 Furthermore, since the control group participants in these three studies stored their SODIS water indoors in lidded SODIS bottles and refrained from consuming drinking water normally stored in-house, the effect of this improved storage may have caused an underestimation of the true benefit of solar disinfection. Importantly, the previous trial methodology did not allow for the differentiation between dysentery, which has serious health consequences, and nondysentery diarrhea. This is an important weakness, as Wright and his colleagues reported that dysentery in children in rural South Africa and Zimbabwe is associated with faecal contamination of source water, while nondysentery diarrhea was uncorrelated with water quality.32,33 The present trial was one of a series of trials which were carried out in South Africa, Zimbabwe, Kenya, and Cambodia as part of the EU funded SODISWATER project.34 It aimed to address some of the deficiencies of earlier research by distinguishing between dysentery and nondysentery diarrhea in a setting of moderate, rather than severe fecal contamination of drinking water. In the SODISWATER randomized intervention study of 12 month duration among a large population (n = 927) of children under age 5 years in rural Cambodia, McGuigan and coworkers35 have reported high compliance (>90%) and reduced incidence of dysentery, with an incidence rate ratio (IRR) of 0.50 (95% CI 0.27 to 0.93, p = 0.029). SODIS also had a protective effect against nondysentery diarrhea, with an IRR of 0.37 (95% CI 0.29 to 0.48, p < 0.001).

ARTICLE

’ METHODS Participant Selection. Participants were recruited in August and September 2007 from six areas in the Nakuru District of Kenya. Three of these areas (Bondeni, Lanet, and Kaptembwa) are urban slum townships in the city of Nakuru, while three (Mogotio, Salgaa, and Wanyororo) are poor rural areas. The urban locations were supplied almost exclusively by standpipes provided by the Nakuru Water Sanitation Services Company. The Company uses conventional water treatment methods to treat ground- and surface water (personal communication, ICROSS, 2010). In the rural locations, water sources were more variable. Only Salgaa was partly supplied by standpipes (54 of 97 households), while the other rural areas used a mix of river (20.7% of households) borehole water, both protected (4.7%) and unprotected (9.1%) and a small number of miscellaneous sources (see Table 1 of the Supporting Information). Sample Size. Sample size was estimated based on comparison of two Poisson event rates in the presence of significant clustering. Since neither the underlying rates of dysentery nor the strength of clustering effects within households were known, we carried out a series of calculations based on rates of 1 to 10 days of dysentery per year and on different degrees of clustering effects. The projected sample of 1000 children was chosen as offering a 90% power to detect a 10% reduction in risk where the underlying rate was 5 episodes per child per year and clustering effects were strong (rho = 0.2). The sample provided more than 90% power to detect a 20% reduction in incidence for all rates of 2 episodes per child per year or greater Randomization. After obtaining ethical approval from the Kenya Medical Research Institute households were identified using local information provided by health workers operating in the areas. Eligible households stored water in containers in-house, did not have a drinking water tap in the house or yard, and had at least one child (but not more than 5) between 6 months and 5 years old residing in the house. Field workers located the households on foot and recorded their addresses. A study area acronym and house number, linked to the Global Positioning System (GPS) coordinates of the household, was allocated to each household. The addresses and coordinates constituted the sample frame of households. Random numbers between zero and one were generated and allocated to the households. If the random number allocated to a household was less than 0.5 the household was randomized to the test group. If the allocated number was above 0.5 the house was randomized to the control group. Field workers were unaware of how the numbers were allocated. Sampling Issues. The decision to use multistage (cluster) sampling method used in the study was a pragmatic one. No regional sample frame exists which would have allowed the identification of eligible households. The identification of eligible households within villages thus entailed a sampling procedure which selected villages and, within these villages, recruited households. There are two significant sources of clustering within the data: at village level, shared environmental factors such as water and sanitation as well as sociodemographic factors will cause households from each village to resemble each other. Furthermore, recruitment of more than one child per household generates further clustering within the data, since children within the same household will share environmental factors affecting health to a greater extent than children from different households within the same village. This required the use of robust (Huber-White) 9316

dx.doi.org/10.1021/es2018835 |Environ. Sci. Technol. 2011, 45, 9315–9323

Environmental Science & Technology variance estimation in order to correct for the statistical effects of clustering on estimates of precision. Presurvey. Details of the study and what would be expected of each household and the children during the study were provided verbally and in writing in the local language to parents or carers. Written informed consent was obtained from the head of the household or the carer. Household selection, during which participants were trained at home to complete diarrhea diaries and the use of SODIS was undertaken by field workers that are well trained in aspects of community work and data collection. In addition a field manual provided clear instructions on all the procedures executed during the field study. The presurvey was completed three months prior to the start of the main survey. Household information with regard basic hygiene and water use practices and sanitation were also collected (see Table 1 of the Supporting Information). Field data were captured using handheld computers and scanned barcodes to link records. The data were downloaded into a database and checked for completeness and consistency before analysis. Two 2-L PET bottles were provided for each child in the intervention group. Carers of children in the intervention group were instructed to fill one bottle and place it in full, unobscured sunlight for a minimum of 6 h every day. In practice most bottles were exposed for longer than 6 h since parents or guardians usually placed the bottle outdoors early in the morning and brought it in at the end of the day. Consequently children in the intervention group drank from a bottle which had been exposed to sunlight on the previous day. Treated water was consumed on the day after exposure. To minimize the possibility of regrowth of partially inactivated bacteria carers were instructed to store the water for a maximum of 48 h. Carers were advised that, where possible, children in the intervention group should drink disinfected water directly from the SODIS bottle rather than from a cup or other container which might have presented a risk of recontamination of the water. Children in the control group were not provided with SODIS bottles and instead were instructed to maintain their usual practices. Anthropometry. Formal anthropometric standardization to determine the precision and accuracy of each person taking height and weight measurements was not conducted because anthropometry was not the main focus of the study. However, field supervisors, who took the measurements, attended a weeklong training session in South Africa during which the use of the equipment (standard adult digital battery operated weighing scales, stature meter and rollameter) was demonstrated and practiced. Special attention was given to the basic anthropometric principles such as calibration of the scales, accuracy when taking measurements, measuring techniques, and ensuring that correct data were recorded. Babies weighing less than 10 to 15 kg were weighed in the arms of the mother or carer. The weight of the babies was calculated in the laboratory. Older children were weighed standing unsupported on the adult scale. In either case the child was shoeless, wearing only a minimum of light clothing. A plank was used as a smooth horizontal position for the scales, stature meter, and the rollameter. The stature meter was always set up against a sturdy vertical wall or door frame. Attention was given to the position of the feet, knees, and position of the head of the subjects when using either the stature or rollameter (see Figures 1 and 2 of the Supporting Information). A manual provided detailed illustrated information and instructions on conducting anthropometry. An initial pilot scale study in South

ARTICLE

Africa and second pilot scale study undertaken in Kenya provided further practical sessions. Health Outcome. The primary health outcome of the study was days on which the child had dysentery, defined according to Baqui, as any loose stool which contained blood or mucus.36 A dysentery diarrhea day was defined as a single day in which one or more stools contained either blood or mucus. One or more consecutive dysentery diarrhea days occurring followed by three consecutive days on which neither dysentery nor nondysentery diarrhea occurred constituted a dysentery episode. Nondysentery diarrhea was defined as three or more loose or watery stools on the same day without blood or mucus, while an episode was characterized by three consecutive days on which neither dysentery nor nondysentery diarrhea occurred. Nondysentery diarrhea days and dysentery diarrhea days were recorded daily using pictorial diaries developed by Gundry and colleagues,33 which record the number and consistency of the child’s stools. Diarrheal incidence was recorded daily for both control and test children for 17 months. Monitoring. Three monitoring visits to determine the microbial water quality and anthropometry were undertaken (July 2008, October 2008, and January 2009). Each visit included all households, but carers and children were often absent from their homes. This was particularly so during holidays when children are sent away to live with relatives or grandparents. Attempts to obtain data from these households included additional follow-up visits, but the long distances between the study areas made this an expensive and unfeasible procedure. As a result we were unable to collect data from every household for each of the three visits. Between monitoring visits trained field staff visited participating households every two weeks to collect diarrheal diaries. The diaries were checked for discrepancies and corrected when possible. Problems raised by the participants were resolved during these visits. Compliance was measured from the collection of the pictorial diarrhea diaries and by recording the responses of household caregivers during monitoring visits, every three months. On these occasions caregivers were asked (i) whether they were using SODIS and (ii) whether it was possible to collect a water sample from the SODIS bottle that was in use. Between monitoring visits field staff regularly reminded the SODIS group about the technique and inquired if they were still using it. Water from the storage containers and SODIS bottles were collected in commercially available 100 mL sample bottles containing sodium thiosulfate to neutralize any residual chlorine in the water. Samples were transported on ice and analyzed on the same day using the Colilert-18 Quantitray, most probable number (MPN) method37 to quantify E. coli. The maximum possible count obtainable using the 0 200 cell forming units per 100 mL Quanti-tray is >200.5 and the minimum 420 nm) irradiation.

ARTICLE

Figure 7. Photocatalytic performances of SnS2-(c) in the first five reuse cycles.

(Eg) of SnS2-(a
f) were determined based on the theory of optical absorption for direct band gap semiconductors: αhν ¼ Bðhν
Eg Þ1=2 where hν and B are discrete photon energy and a constant related to the material, respectively. The value of α can be calculated from the diffuse reflectance data using the Kubelka
Munk function. But, for the diffused reflectance spectra, the Kubelka
Munk function can be used instead of α for estimating the optical absorption edge energy.32
35 So, the curves of (F(R∞)hν)2 versus (hν) for SnS2-(a
f) are plotted in Figure 5. By extrapolating the linear portion of the (F(R∞)hν)2 versus (hν) curves to F(R∞) = 0, the Eg values of SnS2-(a
f) were estimated to be 2.22
2.32 eV (Table 1). Photocatalytic Tests. Photocatalytic Activities. Figure 6 shows the photocatalytic activities of SnS2-(a
f) and P25 TiO2 in the reduction of aqueous Cr(VI) under visible light (λ > 420 nm) irradiation. As can be seen from Figure 6, in the presence of P25 TiO2 or in the absence of any photocatalyst, the reduction of Cr(VI) hardly occurred under visible light (λ > 420 nm) irradiation for 120 min. Instead, the reduction of Cr(VI) proceeded quite rapidly in the presence of SnS2-(a
f). Nevertheless, the photocatalytic activities of SnS2-(a
f) differed greatly, indicating that the synthesis conditions of SnS2 nanocrystals played an important role in their photocatalytic activities. The photocatalytic activities of SnS2-(a
f) followed the order of SnS2-(c) > SnS2-(d) > SnS2-(f) > SnS2-(e) > SnS2-(b) > SnS2-(a). For instance, when irradiated for 120 min, the reduction ratios of Cr(VI) over SnS2-(c), SnS2-(d), SnS2-(f), SnS2-(e), SnS2-(b), and SnS2-(a) were 99.6%, 97.5%, 96.0%, 91.2%, 90.4%, and 77.9%, respectively. Moreover, the photocatalytic reaction rate constants (k) in the presence of SnS2-(c), SnS2-(d), SnS2-(f), SnS2-(e), SnS2-(b), and SnS2-(a) were in turn 0.0394, 0.0301, 0.0252, 0.0205, 0.0198, and 0.0129 min
1 (Supporting Information Figure S2 and Table 1), obtained using the pseudo-first-order model as expressed by36
41 lnðC0 =CÞ ¼ kt The difference in the photocatalytic activities of SnS2-(a
f) was most likely a result of the combined action of many factors, such as particle size, specific surface area, adsorption capacity for Cr(VI), band gap, morphology, composition, crystallinity, crystal defects, and dispersibility, etc. Since almost all of the aforementioned factors were strongly coupled, it was difficult to characterize the specific function and influence of a single parameter in the

Figure 8. HRTEM image of SnS2-AP. The fringe interval of 0.316 nm in this image is consistent with the interplanar spacing of (100) crystal planes of hexagonal phase SnS2.

photocatalytic activity of SnS2 nanocrystals. But, there was a direct correlation between the dark adsorption amounts for Cr(VI) and the photocatalytic activities of SnS2-(a
f) (that is, both the photocatalytic activities and the dark adsorption amounts for Cr(VI) of SnS2-(a
f) followed the same order of SnS2-(c) > SnS2-(d) > SnS2-(f) > SnS2-(e) > SnS2-(b) > SnS2-(a), as shown in Table 1), suggesting that the adsorption capacities for Cr(VI) of SnS2 nanocrystals should play a predominant role in their photocatalytic activities. Because the photocatalytic reactions are commonly believed to occur on the surface of the photocatalyst, the larger adsorption amounts of Cr(VI) onto SnS2 nanocrystals may contribute to the faster reduction rate of Cr(VI).41,42 Photocatalytic Stability. Since the stability of sulfide photocatalysts has always been a concern, it is important to investigate the stability and reusability of the as-synthesized SnS2 nanocrystals in photocatalytic reduction of aqueous Cr(VI). So, in the current work, SnS2-(c) was recycled for five times in the same photocatalytic reactions. After each reuse cycle which lasted for 120 min, the photocatalyst was separated from the aqueous suspension by filtration, washed with 1 mol/L HNO3 aqueous solution (to reduce the amount of greenish Cr(OH)3 deposited on the surface of SnS2-(c)) and deionized water, dried in vacuum at 100 °C for 4 h, and weighed for the next reuse cycle. Taking into account the mass loss of photocatalyst during each reuse cycle, the fourth reuse cycle must be conducted twice in order to accumulate enough sample for the fifth reuse cycle, the third reuse cycle must be conducted twice in order to accumulate enough sample for the fourth reuse cycle, and so on. Figure 7 shows the photocatalytic performance of SnS2-(c) in the first five 9328

dx.doi.org/10.1021/es202012b |Environ. Sci. Technol. 2011, 45, 9324–9331

Environmental Science & Technology

ARTICLE

Figure 9. XPS spectra of SnS2-AP and SnS2-(c).

reuse cycles. Apparently, the photocatalytic activity of SnS2-(c) deteriorated with the increase in the number of reuse cycle, but only very slightly. Even in the fifth reuse cycle of SnS2-(c), the reduction ratio of Cr(VI) can still reach 97% after visible light irradiation for 120 min. The product collected after the fifth reuse cycle of SnS2-(c) in photocatalysis (which was hereinafter called SnS2-AP for the convenience of description) was further characterized by means of XRD, HRTEM, and XPS. Both the XRD pattern (Figure 1 SnS2-AP) and the HRTEM image (Figure 8) of SnS2-AP demonstrated that it was still hexagonal phase SnS2. The survey XPS spectrum of SnS2-AP (Figure 9) showed the presence of Sn and S components, as well as Cr, C, and O contaminants. From the high resolution XPS spectra of Sn 3d and S 2p core levels (Figure 9), it can be seen that the binding energies of Sn 3d and S 2p of SnS2-AP were nearly the same as those of SnS2-(c), for instance, the binding energies of Sn 3d5/2 and S 2p3/2 of SnS2-AP and SnS2-(c) were 486.61 and 486.65 eV, 161.68, and 161.74 eV, respectively. Furthermore, the binding energies of Sn 3d5/2 and S 2p3/2 of SnS2-AP and SnS2-(c) were all consistent with the reference data of Sn4+ and S2
in SnS2.11,43,44 Besides, the binding energy of Cr 2p3/2 was observed at 577.36 eV (Figure 9), which corresponded to Cr(III) in Cr(OH)3.45,46 The formation of Cr(OH)3 on the surface of SnS2-AP can be due to the hydrolysis-precipitation of Cr(III) cations, which were generated from the photocatalytic reduction of adsorbed Cr(VI). Unfortunately, the deposition of Cr(OH)3 on the surface of SnS2-(c) was likely to occupy some photocatalytic active sites of the latter, and accordingly decreased slightly the photocatalytic activity of SnS2-(c) during its reuse.46

’ ASSOCIATED CONTENT

bS

Supporting Information. Figure S1 and Figure S2. This material is available free of charge via the Internet at http://pubs. acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: 086 0514 87962581; fax: 086 0514 87975244; e-mail: [email protected].

’ ACKNOWLEDGMENT Thanks to the China Postdoctoral Science Foundation funded project, Jiangsu Planned Projects for Postdoctoral Research Funds, National Natural Science Foundation of China (50873085), and Natural Science Foundation of Jiangsu Province (08KJB150019). ’ REFERENCES (1) Testa, J. J.; Grela, M. A.; Litter, M. I. Heterogeneous photocatalytic reduction of chromium(VI) over TiO2 particles in the presence of oxalate: Involvement of Cr(V) species. Environ. Sci. Technol. 2004, 38 (5), 1589
1594; DOI: 10.1021/es0346532. (2) Kleiman, A.; Marquez, A.; Vera, M. L.; Meichtry, J. M.; Litter, M. I. Photocatalytic activity of TiO2 thin films deposited by cathodic arc. Appl. Catal., B 2011, 101 (3
4), 676
681; DOI: 10.1016/j.apcatb.2010.11.009. (3) Vinu, R.; Madras, G. Kinetics of simultaneous photocatalytic degradation of phenolic compounds and reduction of metal ions with nano-TiO2. Environ. Sci. Technol. 2008, 42 (3), 913
919; DOI: 10.1021/es0720457. (4) Mu, R.; Xu, Z.; Li, L.; Shao, Y.; Wan, H.; Zheng, S. On the photocatalytic properties of elongated TiO2 nanoparticles for phenol degradation and Cr(VI) reduction. J. Hazard. Mater. 2010, 176 (1
3), 495
502; DOI: 10.1016/j.jhazmat.2009.11.057. (5) Yu, H.; Chen, S.; Quan, X.; Zhao, H.; Zhang, Y. Fabrication of a TiO2
BDD heterojunction and its application as a photocatalyst for the simultaneous oxidation of an azo dye and reduction of Cr(VI). Environ. Sci. Technol. 2008, 42 (10), 3791
3796; DOI: 10.1021/es702948e. (6) Luo, S.; Xiao, Y.; Yang, L.; Liu, C.; Su, F.; Li, Y. Simultaneous detoxification of hexavalent chromium and acid orange 7 by a novel Au/ TiO2 heterojunction composite nanotube arrays. Sep. Purif. Technol. 2011, 79 (1), 85
91; DOI: 10.1016/j.seppur.2011.03.019. (7) Yang, L.; Xiao, Y.; Liu, S.; Li, Y.; Cai, Q.; Luo, S. Photocatalytic reduction of Cr(VI) on WO3 doped long TiO2 nanotube arrays in the 9329

dx.doi.org/10.1021/es202012b |Environ. Sci. Technol. 2011, 45, 9324–9331

Environmental Science & Technology presence of citric acid. Appl. Catal., B 2010, 94 (1
2), 142
149; DO:I 10.1016/j.apcatb.2009.11.002. (8) Sun, B.; Reddy, E. P.; Smirniotis, P. G. Visible light Cr(VI) reduction and organic chemical oxidation by TiO2 photocatalysis. Environ. Sci. Technol. 2005, 39 (16), 6251
6259; DOI: 10.1021/es0480872. (9) Kim, G.; Choi, W. Charge-transfer surface complex of EDTATiO2 and its effect on photocatalysis under visible light. Appl. Catal., B 2010, 100 (1
2), 77
83; DOI: 10.1016/j.apcatb.2010.07.014. (10) Wang, N.; Zhu, L.; Deng, K.; She, Y.; Yu, Y.; Tang, H. Visible light photocatalytic reduction of Cr(VI) on TiO2 in situ modified with small molecular weight organic acids. Appl. Catal., B 2010, 95 (3
4), 400
407; DOI: 10.1016/j.apcatb.2010.01.019. (11) Zhang, Y. C.; Du, Z. N.; Li, K. W.; Zhang, M.; Dionysiou, D. D. High-performance visible-light-driven SnS2/SnO2 nanocomposite photocatalyst prepared via in situ hydrothermal oxidation of SnS2 nanoparticles. ACS Appl. Mater. Interfaces 2011, 3 (5), 1528
1537; DOI: 10.1021/am200102y. (12) Yang, C.; Wang, W.; Shan, Z.; Huang, F. Preparation and photocatalytic activity of high-efficiency visible-light-responsive photocatalyst SnSx/TiO2. J. Solid State Chem. 2009, 182 (4), 807
812; DOI: 10.1016/j.jssc.2008.12.018. (13) Xie, Y.; Ali, G.; Yoo, S. H.; Cho, S. O. Sonication-assisted synthesis of CdS quantum-dot-sensitized TiO2 nanotube arrays with enhanced photoelectrochemical and photocatalytic activity. ACS Appl. Mater. Interfaces 2010, 2 (10), 2910
2914; DOI: 10.1021/am100605a. (14) Li, W.; Li, D.; Meng, S.; Chen, W.; Fu, X.; Shao, Y. Novel approach to enhance photosensitized degradation of rhodamine B under visible light irradiation by the ZnxCd1‑xS/TiO2 nanocomposites. Environ. Sci. Technol. 2011, 45 (7), 2987
2993; DOI: 10.1021/es103041f. (15) Guo, Y.; Wang, L.; Yang, L.; Zhang, J.; Jiang, L.; Ma, X. Optical and photocatalytic properties of arginine-stabilized cadmium sulfide quantum dots. Mater. Lett. 2011, 65 (3), 486
489; DOI: 10.1016/j. matlet.2010.10.057. (16) Zhang, Y. C.; Du, Z. N.; Li, S. Y.; Zhang, M. Novel synthesis and high visible light photocatalytic activity of SnS2 nanoflakes from SnCl2 3 2H2O and S powders. Appl. Catal., B 2010, 95 (1
2), 153
159; DOI: 10.1016/j.apcatb.2009.12.022. (17) Zhang, Y. C.; Du, Z. N.; Li, K. W.; Zhang, M. Size-controlled hydrothermal synthesis of SnS2 nanoparticles with high performance in visible light-driven photocatalytic degradation of aqueous methyl orange. Sep. Purif. Technol. 2011, 81(1), 101
107; DOI: 10.1016/j. seppur.2011.07.016. (18) Lei, Y.; Song, S.; Fan, W.; Xing, Y.; Zhang, H. Facile synthesis and assemblies of flowerlike SnS2 and In3+-doped SnS2: Hierarchical structures and their enhanced photocatalytic property. J. Phys. Chem. C 2009, 113 (4), 1280
1285; DOI: 10.1021/jp8079974. (19) Wang, C.; Zhang, H.; Li, F.; Zhu, L. Degradation and mineralization of bisphenol A by mesoporous Bi2WO6 under simulated solar light irradiation. Environ. Sci. Technol. 2010, 44 (17), 6843
6848; DOI: 10.1021/es101890w. (20) Idris, A.; Hassan, N.; Rashid, R.; Ngomsik, A. F. Kinetic and regeneration studies of photocatalytic magnetic separable beads for chromium (VI) reduction under sunlight. J. Hazard. Mater. 2011, 186 (1), 629
635; DOI: 10.1016/j:jhazmat.2010.11.101. (21) Uchiyama, H.; Shirai, Y.; Kozuka, H. Formation of spherical SnO2 particles consisting of nanocrystals from aqueous solution of SnCl4 containing citric acid via hydrothermal process. J. Cryst. Growth 2011, 319 (1), 70
78; DOI: 10.1016/j.jcrysgro.2011.02.002. (22) Reddy, N. K.; Devika, M.; Ahsanulhaq, Q. Growth of orthorhombic SnS nanobox structures on seeded substrates. Cryst. Growth Des. 2010, 10 (11), 4769
4772; DOI: 10.1021/cg100621d. (23) Tsukimura, K.; Suzuki, M.; Suzuki, Y.; Murakami, T. Kinetic theory of crystallization of nanoparticles. Cryst. Growth Des. 2010, 10 (8), 3596
3607; DOI: 10.1021/cg100488t. (24) Vekilov, P. G. Nucleation. Cryst. Growth Des. 2010, 10 (12), 5007
5019; DOI: 10.1021/cg1011633. (25) He, H.; Gao, C. Supraparamagnetic, conductive, and processable multifunctional graphene nanosheets coated with high-density

ARTICLE

Fe3O4 nanoparticles. ACS Appl. Mater. Interfaces 2010, 2 (11), 3201
3210; DOI: 10.1021/am.100673g. (26) Jimenez, J. A.; Sendova, M.; Sendova-Vassileva, M. Real-time monitoring of plasmonic evolution in thick Ag:SiO2 films: Nanocomposite optical tuning. ACS Appl. Mater. Interfaces 2011, 3 (2), 447
454; DOI: 10.1021/am101021a. (27) Guo, C.; Ge, M.; Liu, L.; Gao, G.; Feng, Y.; Wang, Y. Directed synthesis of mesoporous TiO2 microspheres: Catalysts and their photocatalysis for bisphenol A degradation. Environ. Sci. Technol. 2010, 44 (1), 419
425; DOI: 10.1021/est9019854. (28) Wang, Y.; Shi, J. C.; Cao, J. L.; Sun, G.; Zhang, Z. Y. Synthesis of Co3O4 nanoparticles via the CTAB-assisted method. Mater. Lett. 2011, 65 (2), 222
224; DOI: 10.1016/j.matlet.2010.09.090. (29) Guo, G.; Huang, J. Preparation of mesoporous tantalum oxide and its enhanced photocatalytic activity. Mater. Lett. 2011, 65 (1), 64
66; DOI: 10.1016/j.matlet.2010.09.027. (30) Araujo, P. Z.; Luca, V.; Bozzano, P. B.; Bianchi, H. L.; Soler-Illia, G. J. A. A. Aerosol-assisted production of mesoporous titania microspheres with enhanced photocatalytic activity: The basis of an improved process. ACS Appl. Mater. Interfaces 2010, 2 (6), 1663
1673; DOI: 10.1021/am100188q. (31) Feng, Y.; Li, L.; Ge, M.; Guo, C.; Wang, J.; Liu, L. Improved catalytic capability of mesoporous TiO2 microspheres and photodecomposition of toluene. ACS Appl. Mater. Interfaces 2010, 2 (11), 3134
3140; DOI: 10.1021/am100620f. (32) Wodka, D.; Biela nska, E.; Socha, R. P.; Wodka, M. E.; Gurgul, J.; Nowak, P. Photocatalytic activity of titanium dioxide modified by silver nanoparticles. ACS Appl. Mater. Interfaces 2010, 2 (7), 1945
1953; DOI: 10.1021/am1002684. (33) Zhang, L.; Cao, X. F.; Chen, X. T.; Xue, Z. L. BiOBr hierarchical microspheres: Microwave-assisted solvothermal synthesis, strong adsorption and excellent photocatalytic properties. J. Colloid Interface Sci. 2011, 354 (2), 630
636; DOI: 10.1016/j.jcis2010.11.042. (34) Spadavecchia, F.; Cappelletti, G.; Ardizzone, S.; Bianchi, C. L.; Cappelli, S.; Oliva, C. Solar photoactivity of nano-N-TiO2 from tertiary amine: role of defects and paramagnetic species. Appl. Catal., B 2010, 96 (3
4), 314
322; DOI: 10.1016/j.apcatb.2010.02.027. (35) Liao, Y.; Que, W.; Zhong, P.; Zhang, J.; He, Y. A facile method to crystallize amorphous anodized TiO2 nanotubes at low temperature. ACS Appl. Mater. Interfaces 2011, 3 (7), 2800
2804; DOI: 10.1021/ am200685s. (36) Wang, L.; Wang, N.; Zhu, L.; Yu, H.; Tang, H. Photocatalytic reduction of Cr(VI) over different TiO2 photocatalysts and the effects of dissolved organic species. J. Hazard. Mater. 2008, 152 (1), 93
99; DOI: 10.1016/j.jhazmat.2007.06.063. (37) Waldmann, N. S.; Paz, Y. Photocatalytic reduction of Cr(VI) by titanium dioxide coupled to functionalized CNTs: An example of counterproductive charge separation. J. Phys. Chem. C 2010, 114 (44), 18946
18952; DOI: 10.1021/jp105925g. (38) Parida, K. M.; Sahu, N. Visible light induced photocatalytic activity of rare earth titania nanocomposites. J. Mol. Catal. A: Chem. 2008, 287 (1
2), 151
158; DOI: 10.1016/j.molcata.2008.02.028. (39) Nasrallah, N.; Kebir, M.; Koudri, Z.; Trari, M. Photocatalytic reduction of Cr(VI) on the novel hetero-system CuFe2O4/CdS. J. Hazard. Mater. 2011, 185 (2
3), 1398
1404; DOI: 10.1016/j. jhazmat.2010.10.061. (40) Qamar, M.; Gondal, M. A.; Yamani, Z. H. Laser-induced efficient reduction of Cr(VI) catalyzed by ZnO nanoparticles. J. Hazard. Mater. 2011, 187 (1
3) 258
263; DOI: 10.1016/j.jhazmat.2011.01.007. (41) Jiang, F.; Zheng, Z.; Xu, Z.; Zheng, S.; Guo, Z.; Chen, L. Aqueous Cr(VI) photo-reduction catalyzed by TiO2 and sulfated TiO2. J. Hazard. Mater. 2006, B134 (1
3), 94
103; DOI: 10.1016/j.jhazmat.2005.10.041. (42) Gherbi, R.; Nasrallah, N.; Amrane, A.; Maachi, R.; Trari, M. Photocatalytic reduction of Cr(VI) on the new hetero-system CuAl2O4/ TiO2. J. Hazard. Mater. 2011, 186 (2
3), 1124
1130; DOI: 10.1016/j. jhazmat.2010.11.105. (43) Ma, D.; Zhou, H.; Zhang, J.; Qian, Y. Controlled synthesis and possible formation mechanism of leaf-shaped SnS2 nanocrystals. Mater. 9330

dx.doi.org/10.1021/es202012b |Environ. Sci. Technol. 2011, 45, 9324–9331

Environmental Science & Technology

ARTICLE

Chem. Phys. 2008, 111 (2
3), 391
395; DOI: 10.1016/j.matchemphys. 2008.04.035. (44) Yang, Q.; Tang, K.; Wang, C.; Zhang, D.; Qian, Y. The synthesis of SnS2 nanoflakes from tetrabutyltin precursor. J. Solid State Chem. 2002, 164 (1), 106
109; DOI: 10.1006/jssc.2001.9453. (45) Chenthamarakshan, C. R.; Rajeshwar, K.; Wolfrum, E. J. Heterogeneous photocatalytic reduction of Cr(VI) in UV-irradiated titania suspensions: Effect of protons, ammonium ions, and other interfacial aspects. Langmuir 2000, 16 (6), 2715
2721; DOI: 10.1021/la9911483. (46) Wang, N.; Xu, Y.; Zhu, L.; Shen, X.; Tang, H. Reconsideration to the deactivation of TiO2 catalyst during simultaneous photocatalytic reduction of Cr(VI) and oxidation of salicylic acid. J. Photochem. Photobiol. A. 2009, 201 (2
3), 121
127; DOI: 10.1016/j.photochem. 2008.10.002.

9331

dx.doi.org/10.1021/es202012b |Environ. Sci. Technol. 2011, 45, 9324–9331

ARTICLE pubs.acs.org/est

Advanced Oxidation Process Based on the Cr(III)/Cr(VI) Redox Cycle Alok D. Bokare and Wonyong Choi* School of Environmental Science and Engineering, Pohang University of Science and Technology (POSTECH), Pohang 790-784, Korea

bS Supporting Information ABSTRACT: Oxidative degradation of aqueous organic pollutants, using 4-chlorophenol (4-CP) as a main model substrate, was achieved with the concurrent H2O2-mediated transformation of Cr(III) to Cr(VI). The Fenton-like oxidation of 4-CP is initiated by the reaction between the aquo-complex of Cr(III) and H2O2, which generates HO• along with the stepwise oxidation of Cr(III) to Cr(VI). The Cr(III)/H2O2 system is inactive in acidic condition, but exhibits maximum oxidative capacity at neutral and near-alkaline pH. Since we previously reported that Cr(VI) can also activate H2O2 to efficiently generate HO•, the dual role of H2O2 as an oxidant of Cr(III) and a reductant of Cr(VI) can be utilized to establish a redox cycle of Cr(III)
Cr(VI)
Cr(III). As a result, HO• can be generated using both Cr(III)/H2O2 and Cr(VI)/H2O2 reactions, either concurrently or sequentially. The formation of HO• was confirmed by monitoring the production of p-hydroxybenzoic acid from [benzoic acid + HO•] as a probe reaction and by quenching the degradation of 4-CP in the presence of methanol as a HO• scavenger. The oxidation rate of 4-CP in the Cr(III)/H2O2 solution was highly influenced by pH, which is ascribed to the hydrolysis of CrIII(H2O)n into CrIII(H2O)n‑m(OH)m and the subsequent condensation to oligomers. The present study proposes that the Cr(III)/H2O2 combined with Cr(VI)/H2O2 process is a viable advanced oxidation process that operates over a wide pH range using the reusable redox cycle of Cr(III) and Cr(VI).

’ INTRODUCTION Advanced oxidation processes (AOPs) using hydrogen peroxide (H2O2) as a precursor of hydroxyl radical (HO•) have emerged as efficient technologies for the rapid destruction of recalcitrant organic pollutants.1 The nonselective reactivity of HO• toward most organic pollutants with near diffusion-limited bimolecular rate constants (108
109 M
1 s
1), combined with the easy availability (million metric ton-scale), low price (ca. 1.0 $/kg of 100% H2O2), and environmentally benign nature of H2O2, facilitates large-scale applications.2 The success of H2O2based oxidation processes depends critically on the choice of reagent used to enhance the formation of HO• from H2O2 decomposition. Transition metal ions (Fe2+, Fe3+, Cu2+) have been extensively used in classical or modified Fenton (photoFenton and electro-Fenton) and Fenton-like reactions for the oxidation of various organic contaminants.3
8 However, the fact that the active metal species is consumed as a reagent and lost through precipitation severely limits the process efficiency. As a result, the continuous supply of metal reagent is needed to sustain the activation of H2O2, which causes the problem of metal sludge. Heterogeneous transition metal catalysts may provide an alternative solution for such problems but suffer from mass transfer limitation and metal leaching.9 Therefore, the ideal process of the metal-induced decomposition of H2O2 should require the regeneration of the active metal species through a redox cycle. To achieve this objective, the redox states of the involved metal species should be stable over a wide r 2011 American Chemical Society

pH range. Compared to iron and copper, chromium exists in a wider range of oxidation states (from
2 to +6), with the trivalent [Cr(III)] and hexavalent [Cr(VI) or chromate] species commonly found in water. Being an oxyanion, chromate is completely soluble over the entire pH range.10 However, due to its extreme toxicity and carcinogenecity, any deliberate addition of Cr(VI) as a reagent into wastewaters is not sensible, even if various physicochemical or biological post-treatments11 can easily remove it from aqueous solution. Trivalent chromium [Cr(III)], on the other hand, is the most thermodynamically stable oxidation state of chromium, kinetically inert, and significantly less toxic. Although the two chromium species (CrVI vs CrIII) are characterized by different chemical behavior, bioavailability, and toxicity,12 they are readily interconverted in aqueous solution. Cr(VI) is a strong oxidant [E0(HCrO4
/Cr3+ = 1.35 VNHE)]13 and reacts rapidly with numerous reducing agents (like Fe0, Fe2+, S2‑, and natural organic matter) to form Cr(III).14 On the other hand, Cr(III) is thermodynamically stable under reducing conditions and is oxidized to Cr(VI) by Mn(III,IV) (hydr)oxides15 or photo-oxidized by FeOH2+.16 H2O2 alone can interconvert Cr(III) and Cr(VI) into each other because of its Received: June 25, 2011 Accepted: September 22, 2011 Revised: September 21, 2011 Published: October 11, 2011 9332

dx.doi.org/10.1021/es2021704 | Environ. Sci. Technol. 2011, 45, 9332–9338

Environmental Science & Technology

ARTICLE

Scheme 1. Schematic Illustration of HO• Generation from H2O2 using Cr(III)
Cr(VI) Redox Cyclea

a

The numbered paths indicate the following: (1) hydrolysis of Cr(III)
aquocomplex; (2) Fenton-like oxidation of Cr(III) to Cr(VI) by H2O2; (3) Cr(VI)-mediated decomposition of H2O2 via dissociation of Cr(V)
peroxo complex (demonstrated in the previous work20); (4) regeneration of Cr(III) by H2O2-mediated reduction of Cr(VI).

ability to act as both an oxidizing agent [E0(H2O2/H2O) = 1.77 V] and a reducing agent [E0(O2/H2O2) = 0.68 V].17 The pe-pH relationship of Cr(VI)/Cr(III) and O2/H2O2 couples indicates that H2O2 can oxidize Cr(III) at pH > 8 and reduce Cr(VI) at lower pH.18 Since the reducing strength of H2O2 strongly increases with decreasing pH, the H2O2-induced reduction of Cr(VI) to Cr(III) at pH < 3 is used for removing chromate from wastewaters.19 In our previous work,20 we demonstrated that Cr(VI) can also activate H2O2 and generate HO• for the oxidative degradation of aqueous organic pollutants although the toxicity of Cr(VI) limits practical applications only to the degradation of organics in chromate-contaminated wastewaters. The oxidation mechanism involves the formation of tetraperoxochromate(V) complex and works over a wide range of pH 3
11. However, the previous method of H2O2 activation can utilize only Cr(VI), not Cr(III). The present study successfully demonstrates that Cr(III) can also activate H2O2 to generate HO• along with the stepwise oxidation of Cr(III) to Cr(VI). This enables a redox cycling of Cr(III)/Cr(VI) by H2O2 that serves as both an oxidant of Cr(III) and a reductant of Cr(VI). As a result, a new AOP that generates HO• repeatedly based on the Cr(III)/Cr(VI) redox cycle is developed (see Scheme 1). We also demonstrate that the Cr(III)/Cr(VI) redox transformation can be easily manipulated by H2O2 in pH-controlled reactions and H2O2 serves the dual roles of a precursor of HO• and an oxidant/reductant of Cr(III)/ Cr(VI). Through this reversible chromium catalytic cycle coupled with the decomposition of H2O2, the hydroxyl radicalmediated degradation of organic compounds can be achieved in repeated cycles in a single batch reactor.

’ EXPERIMENTAL SECTION Chemicals and Materials. Chemicals that were used as received in this study included chromium(III) nitrate (Sigma), sodium chromate (Sigma), hydrogen peroxide (30%, Kanto), 4-chlorophenol (4-CP, Sigma), phenol (Aldrich), aniline (Aldrich), nitrobenzene (Aldrich), benzoic acid (Aldrich), p-hydroxy

Figure 1. (a) Effect of initial pH on the degradation of 4-CP in the Cr(III)/H2O2 system. [4-CP]0 = 100 μM, [Cr(III)]0 = 2 mM, and [H2O2]0 = 20 mM. (b) Degradation of 4-CP and the concurrent generation of chloride and chromate (Cr(VI)) under the condition of (a) and pHi = 7. The degradation of 4-CP in the presence of Cr(III) only (no H2O2) or H2O2 only (no Cr(III)) is denoted by open triangle (r) and closed triangle (2), respectively.

benzoic acid (p-HBA, Aldrich), β-cyclodextrin hydrate (Aldrich), methanol (Daejung), and acetonitrile (Merck). All solutions were prepared in ultrapure water (18 MΩ cm) obtained from a Barnstead purification system. Procedure and Analytical Methods. The reactions were carried out in 50-mL glass beakers stirred on a magnetic stirrer. An aliquot of stock solution of 4-CP (or other substrate, 1 mM) was added to make a desired concentration (typically 0.1 mM), and then the reaction was initiated by the sequential addition of Cr(III) and H2O2. Unless otherwise mentioned, [Cr(III)] was fixed at 2 mM. The initial pH (pHi) of the solution was adjusted to a desired value with 1 N NaOH standard solution. Sample aliquots (1 mL) were withdrawn at regular time intervals from the reactor and injected into 4-mL glass vials containing 50 μL of sodium sulfite (Na2SO3, 2 M) to quench residual H2O2. All experiments were carried out in triplicate for a given condition. Quantitative analysis of substrates was done using a highperformance liquid chromatograph (HPLC Agilent 1100) equipped with a C-18 column (Agilent Zorbax 300SB) and a diode-array detector. The eluent compositions were as follows: (a) 0.1% phosphoric acid aqueous solution and acetonitrile 9333

dx.doi.org/10.1021/es2021704 |Environ. Sci. Technol. 2011, 45, 9332–9338

Environmental Science & Technology

ARTICLE

Figure 2. Degradation of phenol, nitrobenzene, and aniline (separate single-component experiments) in the presence of Cr(III) and H2O2 ([Cr(III)]0 = 2 mM, [H2O2]0 = 20 mM, [substrate]0 = 100 μM, and pHi = 7).

(80:20 v/v) for 4-CP, (b) water, acetonitrile, and acetic acid (78:20:2 v/v) for phenol, (c) water and methanol (50:50 v/v) for nitrobenzene and aniline, and (d) 0.1% phosphoric acid aqueous solution and acetonitrile (85:15 v/v) for benzoic acid. Quantification of ionic intermediates/products was performed using an ion chromatograph (IC, Dionex DX-120) equipped with Dionex IonPac AS-14 column and a conductivity detector. The eluent composition was 3.5 mM Na2CO3 + 1 mM NaHCO3. Cr(VI) concentration was determined using a modified diphenylcarbazide (DPC) method. H2O2 interferes in the Cr(VI) determination by the standard DPC method21 due to its ability to rapidly reduce Cr(VI) to Cr(III) in acidic solution. In the modified method,22 2-mL sample aliqouts were quenched by sequential addition of 0.5 mL of DPC in acetone (20 g/L) followed by the addition of 0.2 mL of 9 M H2SO4. The absorbance of the colored Cr-DPC complex was analyzed spectrophotometrically at 540 nm within 5 min of the color development. Total organic carbon (TOC) was measured using a TOC analyzer (TOCVCSH, Shimadzu). Various species of Cr(III) aquo-complexes were analyzed by forming their inclusion complexes with β-cyclodextrin (β-CD), which were then determined by matrix-assisted laser desorption and ionization time-of-flight mass spectrometry (MALDI-TOF MS, Bruker REFLEX III). The inclusion complexes were prepared by mixing Cr(III) and β-CD at 1:1 molar ratio and adjusting the pH to the desired value with 1 N NaOH. The matrix used for the MALDI-TOF experiments was α-cyano-4hydroxycinnamic acid (CHCA, Aldrich) dissolved in acetone at 80 mg/mL. The CHCA and Cr(III) + β-CD solutions were mixed at 4:1 ratio (matrix:analyte, v/v), and the mixed solution was dropped onto the MALDI plate and air-dried.

’ RESULTS AND DISCUSSION Oxidation in Cr(III)/H2O2 System. To evaluate the oxidative capacity of the Cr(III)/H2O2 system, 4-CP degradation in aqueous solution was investigated at different pHi under airequilibrated conditions. As shown in Figure 1a, 4-CP degradation was completely inhibited at pH 3, but increased with increasing

Figure 3. (a) Comparison of the time profiles of p-HBA formation during the oxidation of benzoic acid (BA) in the Cr(III)/H2O2 system. [BA]0 = 10 mM, [Cr(III)]0 = 2 mM, and [H2O2]0 = 20 mM. (b) Effect of methanol (OH radical scavenger) on the degradation of 4-CP in the Cr(III)/H2O2 system. [4-CP]0 = 100 μM, [Cr(III)]0 = 2 mM, [H2O2]0 = 20 mM, [CH3OH]0 = 100 mM, and pHi = 7.

pH leading to complete degradation in 6 h at pH 7. However, with further increase in pHi, the 4-CP removal rate decreased. The complete absence of 4-CP oxidation in acidic condition (pH < 4) can be attributed to the kinetic inertness of the Cr(III) aquocompexes. At pH < 4, Cr(III) exists as the hexaaquo complex [Cr(H2O)6]3+ (pKa = 4, see Supporting Information Figure S1) and its reaction with any organic or inorganic species (H2O2 in the present case) involves the substitution of the coordinated water molecules by the reactant. However, the extremely low frequency of water exchange within Cr(III)-aquocomplex (∼10
6.3 s
1 or half-life ∼40 h)23 makes the aquo-complex substitutionally inert and unreactive toward H2O2 in the present experimental time scale. For comparison, the corresponding water exchange frequencies for Al(III) and Fe(III) are 10
0.8 s
1 and 103.5 s
1, respectively.23 However, when pHi increased to 7, a complete removal of 4-CP was obtained. Cr(III) at neutral pH is neither an oxidant nor a reductant and exists solely as insoluble Cr(OH)3(s). To confirm whether 4-CP was removed from aqueous solution 9334

dx.doi.org/10.1021/es2021704 |Environ. Sci. Technol. 2011, 45, 9332–9338

Environmental Science & Technology

ARTICLE

through simple adsorption on Cr(OH)3, control degradation experiments were carried out only in the presence of Cr(III) at pH 7 (without H2O2). The removal of 4-CP was not observed at all in the absence of H2O2 (Figure 1b), which ruled out the possibility of 4-CP removal by adsorption. Moreover, the concurrent production of chloride ions (Figure 1b) accounted for 90% of the removed 4-CP. This implies that a reaction between Cr(III) and H2O2 generates a reactive species that is responsible for the degradation of 4-CP. TOC was also reduced by 33((4)% in 6 h reaction, which indicates that some fraction of 4-CP was actually mineralized. The minor deficit in chloride mass balance (∼10%) may be ascribed to the generation of chlorinated intermediates that were not determined in this work. This clearly demonstrates that 4-CP can be oxidatively degraded at neutral pH in the Cr(III)/H2O2 system. Oxidation of other organic pollutants such as phenol [34((2)% TOC reduction], nitrobenzene [35((1)% TOC reduction], and aniline [25((4)% TOC reduction] was also successfully achieved at neutral pH (Figure 2). To ascertain whether 4-CP is oxidized by hydroxyl radicals generated through Cr(III)-mediated decomposition of H2O2, the oxidative conversion of benzoic acid (BA) to p-hydroxybenzoic acid (p-HBA) was used as a probe reaction. The formation of p-HBA from the reaction of (BA + HO•) has been used as an indirect method to detect HO• formation.20,24 Figure 3a shows the production of p-HBA from BA in the Cr(III)/H2O2 system at different pHi. The formation of p-HBA is completely inhibited at pHi 3 but increases with increasing pH, which is similar to the pH-dependent behavior of 4-CP degradation (see Figure 1). This indicates that HO• generated from the reaction of Cr(III) and H2O2 is initiated only at pH g 5, which corroborates the electron paramagnetic resonance (EPR) study of Shi et al.25 who reported the formation of HO• in neutral (pH = 7.2) solution of Cr(III)/H2O2 but not under acidic (pH = 3) condition. Furthermore, the addition of methanol as a hydroxyl radical scavenger completely inhibited the degradation of 4-CP at neutral pH (Figure 3b). This confirms that H2O2 acts as a precursor of hydroxyl radicals, which are primarily responsible for 4-CP oxidation in the presence of Cr(III). The reaction between Cr(III) and H2O2 system generates HO• through a Fenton-like reaction with the simultaneous formation of intermediate Cr(IV) species.26 CrðIIIÞ þ H2 O2 f CrðIVÞ þ HO• þ OH


ð1Þ

•

Cr(IV) immediately generates another HO from H2O2 (reaction 2)27 or undergoes disproportionation to generate Cr(V) species (reaction 3).27 Cr(V) induces another Fentonlike reaction (reaction 4)28 with further generation of HO•. CrðIVÞ þ H2 O2 f CrðVÞ þ HO• þ OH


ð2Þ

2CrðIVÞ f CrðVÞ þ CrðIIIÞ

ð3Þ

CrðVÞ þ H2 O2 f CrðVIÞ þ HO• þ OH


ð4Þ

The generation of Cr(VI) during 4-CP oxidation via stepwise oxidation of Cr(III) is shown in Figure 1b. Thus, the Cr(III)/ H2O2 system generates HO• via a series of Fenton-like reactions involving intermediate Cr(IV) and Cr(V) species,29 leading to the transformation of Cr(III) into Cr(VI). However, it should be realized that the actual oxidation chemistry can be more complex. It may be possible that the intermediate Cr(IV) and

Table 1. Proposed Chemical Structures of Cr(III)
β-CD Inclusion Complexes Identified by MALDI-TOF at Different pHi m/z

7

1272.6

[Cr2(μ
OH)2(H2O)8]4+
β-CD

1311.7

[Cr2(μ
OH)2(H2O)7(OH)]3+
β-CD
Na+

1273.8

[Cr2(μ
OH)2(H2O)8]4+
β-CD

1312.2

[Cr2(μ
OH)2(H2O)7(OH)]3+
β-CD
Na+

1332.6 1358.4

[Cr2(μ
OH)2(H2O)6(OH)2]2+
β-CD
Na+ [Cr3(μ
OH)4(H2O)9]5+
β-CD

1359.2

[Cr3(μ
OH)4(H2O)9]5+
β-CD

9

11 a

proposed complex compositiona

pHi

Na
β-CD (m/z = 1158) adducts were observed in all samples. At pHi e 5, only the Na+ adducts were detected. +

Cr(V) species take part in the direct oxidation of 4-CP and its intermediates, which should complicate the overall redox chemistry. The formation of a stable Cr(V)-complex is possible in the presence of organic substrates with ligand groups (e.g., hydroxycarboxylate and 1,2-diol moieties in natural organic matters).30 Moreover, EPR studies have also suggested the formation of stable Cr(V)-peroxo complexes in the Cr(VI)/H2O2 system.31,32 Because these intermediate chromium complexes should influence the oxidation kinetics and mechanisms, the proposed path in Scheme 1 should be taken as a simplified representation of the complex redox process in the Cr(III)
Cr(VI)
H2O2 system. pH-Dependent Speciation of Cr(III) and the Reactivity for H2O2 Activation. The degradation rate of 4-CP increases with pH above 3 but subsequently decreases beyond neutral pH condition (see Figure 1). When pH increases toward near neutral and alkaline values, the hexaaquo ions are hydrolyzed to hydroxocomplexes (reaction 5). ½CrðH2 OÞ6 3þ f ½CrðH2 OÞ5 ðOHÞ2þ þ Hþ

ð5Þ

This Cr(III) monomeric hydroxo-complex subsequently undergoes hydrolytic condensation to form polynuclear complexes like dimer, trimer, and higher oligomers containing μ-hydroxo bridges between adjacent chromium atoms (see Supporting Information Figure S2). These soluble hydroxo-complexes are eventually polymerized and precipitated as Cr(OH)3(s). However, the hydrolytic conversion of Cr(III)-oligomers into solid Cr(OH)3 occurs sufficiently slowly (>1 yr) to permit separation and isolation of a series of individual oligomers up to a hexamer.33,34 This means that oligomers formed through Cr(III) hydrolysis should be long-lived enough to react with H2O2. The pH-dependent formation of Cr(III) oligomers, however, strongly influences the reaction kinetics with H2O2. Rao et al.35 demonstrated that the reaction rate constant (k) obtained with isolated oligomers decreases as oligomerization proceeds (kdimer > ktrimer). The rate constant for unseparated oligomers in solution (analogous to the present study) was 2 orders of magnitude slower than for the isolated dimer. Thus, the increase in 4-CP oxidation with increasing pH from 3 to 7 indicates that Cr(III) hydrolysis (or oligomerization) initiates the decomposition of H2O2 to generate HO•. On the other hand, the decrease in oxidation efficiency in alkaline condition suggests the formation of higher oligomers, which are less reactive toward H2O2 decomposition.35 To confirm the presence of Cr(III) oligomers and determine the degree of oligomerization at different pH, β-CD was used as a complexing ligand to form inclusion complexes with the 9335

dx.doi.org/10.1021/es2021704 |Environ. Sci. Technol. 2011, 45, 9332–9338

Environmental Science & Technology oligomers. β-CD is a torus-shaped cyclic oligosaccharide with an internal hydrophobic cavity and contains seven α-D-glucopyranose units linked together by α-1,4-glycosidic linkages. It has been extensively used for molecular recognition and as metal ion receptor in host
guest systems.36 Using the oxygen atoms on adjacent pyranose rings as a diol ligand (see Supporting Information, Figure S3), β-CD can form highly stable inclusion complexes with binuclear hydroxy-bridged metal structures (structurally similar to Cr(III) oligomers).37,38 This complexation property of β-CD can be used to isolate different Cr(III) oligomers by using the internal cavity as an efficient trapping site. Table 1 shows the MALDI-TOF analysis of Cr(III) aqueous solutions containing β-CD at different pH. At pH e 5, only the Na+ and K+ adducts of β-CD were detected at m/z = 1158 and 1174, respectively. The absence of β-CD complexes with Cr(III) at pH e 5 can be attributed to the weak interaction between unmodified β-CD and the metal ion. Transition metal cations can form stable complexes only with surface functionalized-βCD, wherein, the metal ion is complexed with ligands situated outside the CD cavity.37 Thus, Cr(III) monomeric species cannot form inclusion complexes with unmodified β-CD and, hence, were not detected in the MALDI-TOF analysis. However, with increasing pH, the degree of oligomerization increased along with the hydrolysis of Cr(III), which is evident from the detection of β-CD inclusion complexes with dimer [Cr2(μ
OH)2(H2O)8]4+, trimer [Cr3(μ
OH)4(H2O)9]5+, and intermediate hydroxo-bridged species (see Table 1). At pH 11, only the trimer complex was detected, which indicates that the dimer was involved in formation of higher oligomers (most probably tetramer).39 The absence of MALDI-TOF peaks corresponding to tetramer and/or higher oligomers may be attributed to their larger molecular size, which cannot fit inside the β-CD internal cavity. The MALDI-TOF analysis confirms that Cr(III) oligomerization is initiated at pH > 5 and the subsequent pH increase leads to the formation of higher oligomers. The reactivity of these different Cr(III) hydrolytic species can be correlated to the pH-dependent 4-CP oxidation behavior. In Fenton-like reactions involving H2O2 and transition metal complexes, the oxidation of the metal center is promoted by the formation of a metal
hydroperoxo complex intermediate (via ligand exchange), followed by the homolytic cleavage of the peroxo bond to generate HO•.40 In the case of Cr(III)-mediated Fenton-like reaction, the rate of water exchange in Cr(H2O)63+ is too slow (∼10
6.3 s
1) to form a hydroperoxo complex, which explains the absence of 4-CP oxidation at pH 3 (see Figure 1). However, at pH 5, the monohydroxy complex (H2O)5CrOH2+ (pKa = 6.1)34 is the dominant Cr(III) species (see Supporting Information, Figure S1, S2). The complex of (H2O)5CrOH2+ is 75 times more reactive in water exchange reaction41,42 and 60
5500 times more reactive in anion complexation reaction,43 compared to Cr(H2O)63+. Thus, the hydrolysis of the Cr(III) aquocomplex at pH 5 can facilitate the peroxo ligand substitution, and initiate HO• generation for the oxidation of 4-CP. At pH > 5, the (H2O)5CrOH2+ species can be sequentially deprotonated, and then oligomerized. The rapid oligomerization competes with and mostly dominates stepwise deprotonation reactions.42 At pH 7, the deprotonated dimer [Cr2(μ
OH)2(H2O)6(OH)]3+ was isolated and identified by MALDI-TOF analysis (see Table 1). The water-exchange rate of this deprotonated dimer is 27 times (for cis form) or 70 times (for trans form) higher compared to the (H2O)5CrOH2+ species.44 Thus, the

ARTICLE

Figure 4. (a) Repeated cycles of 4-CP degradation and the concurrent generation of Cr(VI) in the presence of Cr(III) and H2O2 (pHi = 7). At the end of each cycle, an aliquot of HClO4 (1 N, 1 mL) was added to regenerate Cr(III) (at the point of 1), then 4-CP (3 mM, 0.9 mL) and H2O2 (20 mM) were replenished (at the point of 2), and finally the pH of the solution was readjusted to 7 before initiating the next degradation cycle. (b) The initial cycle of 4-CP degradation ([Cr(III)]0 = 2 mM, [H2O2]0 = 20 mM, pHi = 7) and the subsequent cycle of 4-CP degradation without the regeneration of Cr(III) and without pH readjustment. The time profile of pH change is shown together.

peroxo ligand exchange reaction (hence the generation of HO•) will be enhanced when raising pH from 5 to 7, which is consistent with the faster oxidation of 4-CP at pH 7 than pH 5 (see Figure 1a and Figure 3a). When pH is further increased to alkaline values (pH 9), the OH-ligands are bridged through condensation. As a result, the concentration of oligomers containing nonbridging OH groups (species A and B in Supporting Information, Figure S4) will decrease with retarding the water exchange reaction,41 which subsequently leads to retardation of peroxo complexation and suppression of OH-radical mediated oxidation (see Figure 1). Furthermore, at pH 11, the complete absence of any species with nonbridging OH groups combined with the formation of higher oligomers (trimer and possibly tetramer) lowers the Cr(III) reactivity toward H2O2. Thus, Cr(III) species coordinated with OH ligands are required to catalyze the peroxo complexation mechanism for the generation of HO•. Cr(III)/Cr(VI) Redox Process as AOP. All practical applications of metal-catalyzed Fenton and Fenton-like AOPs are severely 9336

dx.doi.org/10.1021/es2021704 |Environ. Sci. Technol. 2011, 45, 9332–9338

Environmental Science & Technology

Figure 5. Effect of solution aging time on the degradation of 4-CP in the Cr(III)/H2O2 system. [Cr(III)]0 = 2 mM, [H2O2]0 = 20 mM, and pHi = 7.

limited by the fact that the precipitation of metal ions limits working conditions to acidic region,9 and prevents the reuse of the catalyst. The Cr(III)-mediated activation of H2O2 shows the maximum oxidation capacity at neutral and near-alkaline pH. Some hydrolytic oligomers are formed at neutral and alkaline pH and they remain soluble without precipitation and can activate H2O2 with less efficiency. This will reduce the need of adding a large amount of metal salt to compensate for the catalyst loss, and subsequently prevent the problem of sludge disposal. Furthermore, the resulting oxidation product, Cr(VI), is soluble over the entire pH range.10 However, the extreme toxicity of Cr(VI) is a major concern and its complete removal from the treated wastewater is essential. Cr(VI) can be reduced to Cr(III) by using H2O2 as a reductant in acidic condition with the concurrent generation of OH radicals.19,20 Therefore, Cr(III) species can be easily regenerated by simply decreasing pH to acidic values in the presence of H2O2. In this way, we can exploit the pH-dependent dual role of H2O2 as Cr(III) oxidant and Cr(VI) reductant to establish a cyclic redox transformation of chromium along with the generation of HO radicals. This makes the Cr(III)/ Cr(VI)/H2O2 system a new AOP based on the redox cycle of chromium species without the loss of active metal species. We have successfully established the process viability by sustaining the repeated cycles of 4-CP removal at neutral pH using Cr(III) regenerated from Cr(VI) prior to oxidation (Figure 4a). However, the inhibition of 4-CP oxidation under acidic condition requires the pH to be raised back to neutral before each successive oxidation cycle. This repeated addition of acid and base will increase the total ionic strength of the solution and the overall treatment cost. To minimize the salinity increase induced by repeated pH adjustments, Cr(VI)/H2O220 as well as Cr(III)/H2O2 process should be concurrently utilized to generate HO•. That is, as Cr(III) is depleted along with the generation of HO• in the Cr(III)/H2O2 process, the accompanying Cr(VI) can also generate HO• at the same time through the Cr(VI)/H2O2 process. It should be noted that the Cr(VI)/H2O2 and Cr(III)/H2O2 processes have complementary pH-dependence: the former favored at acidic pH but the latter inhibited in the acidic condition. We verified this dual process by achieving consecutive cycles of 4-CP oxidation without regenerating Cr(III) and without pH

ARTICLE

readjustment (Figure 4b). Because the pH decreased to around 4 after the first cycle, the Cr(III)/H2O2 reaction is inhibited in the second cycle. However, the Cr(VI)-induced activation of H2O2 is more efficient at acidic pH20 and hence 4-CP oxidation was achieved through the Cr(VI)/H2O2 process in the second cycle. That is, both Cr(III)/H2O2 and Cr(VI)/H2O2 can be utilized as an AOP that generates HO• and both processes can work either concurrently or sequentially (Scheme 1). The combination of Cr(VI)/H2O2 and Cr(III)/H2O2 processes should reduce the cost for pH adjustments. However, such sequential processes without additional pH adjustments cannot be efficiently sustained for further cycles since the pH gradually converges to a relatively higher pH (>5), at which the Cr(VI)/H2O2 process is very slow. Anyway, despite the additional cost needed for sequential pH adjustments, the Cr(III)/Cr(VI)/H2O2 process works over a wide pH range by recycling the active metal species and provides an advantage over the classical Fenton or Fentonlike process, which needs a strict acidic condition to prevent the iron loss by precipitation. Finally, it should be mentioned that the aging of Cr(III) solution is also an important factor to be considered. Figure 5 shows that the degradation of 4-CP was significantly retarded when using Cr(III) solutions aged at ambient temperature for 1 month. The oligomerization process is dependent on not only pH but also the aging time. Generally, the reactivity of Cr(III) aqueous solutions decreases with the aging time, which is attributed to the transformation of soluble Cr(III) hydrolytic species (monomer and oligomer) into insoluble polynuclear species and/or amorphous chromium oxyhydroxides.45 In the present study, although the 4-CP oxidation efficiency decreased as expected, a significant concentration of 4-CP could be still removed even after 30-day aging period. This AOP based on the redox cycle of Cr(III)/Cr(VI) with H2O2 can be versatilely applied to the degradation of organics through the generation of HO• in chromium-contaminated wastewaters regardless of the oxidation state of the chromium species. However, the practical implementation of this chromium-based AOP and its effectiveness will be influenced by various constituents in wastewaters, which may interfere with the catalytic cycle of Cr(III)/Cr(VI). More thorough studies are required to understand such interfering effects.

’ ASSOCIATED CONTENT

bS

Supporting Information. The pH-dependent speciation of aqueous Cr(III) and H2O2, schematic representation of Cr(III) hydrolysis and oligomerization reactions, chemical structure of β-cyclodextrin and comparative MALDI-TOF spectra at pHi = 7 and pHi = 9. This information is available free of charge via the Internet at http://pubs.acs.org/.

’ AUTHOR INFORMATION Corresponding Author

*Phone: +82-54-279-2283; fax: +82-54-279-8299; e-mail: wchoi@ postech.edu.

’ ACKNOWLEDGMENT This work was supported by KOSEF NRL program (R0A-2008000-20068-0), KOSEF EPB center (Grant R11-2008-052-02002), 9337

dx.doi.org/10.1021/es2021704 |Environ. Sci. Technol. 2011, 45, 9332–9338

Environmental Science & Technology and KCAP (Sogang Univ.) funded by MEST through NRF (NRF-2009-C1AAA001-2009-0093879).

’ REFERENCES (1) Gogate, P. R.; Pandit, A. B. A review of imperative technologies for wastewater treatment. I: Oxidation technologies at ambient conditions. Adv. Environ. Res. 2004, 8, 501–551. (2) M€agerlein, W.; Dreisbach, C.; Hugl, H.; Tse, M. K.; Klawonn, M.; Bhor, S.; Beller, M. Homogeneous and heterogeneous ruthenium catalysts in the synthesis of fine chemicals. Catal. Today 2007, 121, 140–150. (3) Pignatello, J. J.; Oliveros, E.; MacKay, A. Advanced oxidation processes for organic contaminant destruction based on the Fenton reaction and related chemistry. Crit. Rev. Environ. Sci. Technol. 2006, 36, 1–84. (4) Laine, D. F.; Cheng, I. F. The destruction of organic pollutants under mild reaction conditions: A review. Microchem. J. 2007, 85, 183–193. (5) Nam, S.; Renganathan, V.; Tratnyek, P. G. Substituent effects on azo dye oxidation by the FeIII-EDTA-H2O2 system. Chemosphere 2001, 45, 59–65. (6) Collins, T. J. TAML oxidant activators: A new approach to the activation of hydrogen peroxide for environmentally significant problems. Acc. Chem. Res. 2002, 35, 782–90. (7) Lin, T. Y.; Wu, C. H. Activation of hydrogen peroxide in copper(II)/amino acid/H2O2 systems: Effects of pH and copper speciation. J. Catal. 2005, 232, 117–126. (8) Shah, V.; Verma, P.; Stopka, P.; Gabriel, J.; Baldrian, P.; Nerud, F. Decolorization of dyes with copper(II)/organic acid/hydrogen peroxide systems. Appl. Catal., B 2003, 46, 287–292. (9) Pestunova, O. P.; Ogorodnikova, O. L.; Parmon, V. N. Studies on the phenol wet peroxide oxidation in the presence of solid catalysts. Chem. Sustain. Dev. 2003, 11, 227–232. (10) Beverskog, B.; Puigdomenech, I. Revised Pourbaix diagrams for chromium at 25
300 °C. Corros. Sci. 1997, 39, 43–57. (11) Owlad, M.; Aroua, M. K.; Daud, W. A. W.; Baroutian, S. Removal of hexavalent chromium-contaminated water and wastewater: A review. Water, Air Soil Pollut. 2009, 200, 59–77. (12) Wang, W. X.; Griscom, S. B.; Fisher, N. S. Bioavailability of Cr(III) and Cr(VI) to marine mussels from solute and particulate pathways. Environ. Sci. Technol. 1997, 31, 603–611. (13) CRC Handbook of Chemistry and Physics, 85th ed.; CRC Press: Boca Raton, FL, 2004. (14) Rai, D.; Eary, L. E.; Zacharia, J. M. Environmental chemistry of chromium. Sci. Total Environ. 1989, 86, 15–23. (15) Fendorf, S. E.; Zasoski, R. J. Chromium(III) oxidation by deltaMnO2 (I) Characterization. Environ. Sci. Technol. 1992, 26, 79–85. (16) Zhang, H.; Bartlett, R. J. Light-induced oxidation of aqueous chromium(III) in the presence of iron(III). Environ. Sci. Technol. 1999, 33, 588–594. (17) Cotton, F. A.; Wilkinson, G.; Murillo, C. A.; Bochmann, M. Advanced Inorganic Chemistry; Wiley-Interscience: New York, 1999. (18) Richard, F. C.; Bourg, A. C. M. Aqueous geochemistry of chromium: A review. Water Res. 1991, 25, 807–816. (19) Pettine, M.; Campanella, L.; Millero, F. Reduction of hexavalent chromium by H2O2 in acidic solutions. Environ. Sci. Technol. 2002, 36, 901–907. (20) Bokare, A. D.; Choi, W. Chromate-induced activation of hydrogen peroxide for oxidative degradation of aqueous organic pollutants. Environ. Sci. Technol. 2010, 44, 7232–7237. (21) APHA, AWWA, WEF. Standard Methods for the Examination of Water and Wastewater, 20th ed; APHA: Washington, DC, 1988; pp 3-65
3-68. (22) Pettine, M.; La Noce, T.; Liberatori, A.; Loreti, L. Hydrogen peroxide interference in the determination of chromium(VI) by the diphenylcarbazide method. Anal. Chim. Acta 1988, 209, 315–319. (23) Hunt, J. P. Metal Ions in Aqueous Solutions; Benjamin, Inc: New York, 1963. (24) Joo, S. H.; Feitz, A. Z.; Sedlak, D. L.; Waite, T. D. Quantification of the oxidizing capacity of nanoparticulate zero-valent iron. Environ. Sci. Technol. 2005, 39, 1263–1268.

ARTICLE

(25) Shi, X.; Mao, Y.; Knapton, A. D.; Ding, M.; Rojanasakul, Y.; Gannett, P. M.; Dalal, N. S.; Liu, K. Reaction of Cr(VI) with ascorbate and hydrogen peroxide generates hydroxyl radicals and causes DNA damage: Role of a Cr(IV)-mediated Fenton-like reaction. Carcinogenesis 1994, 15, 2475–2478. (26) Tsou, T.-C.; Yang, J.-L. Formation of reactive oxygen species and DNA strand breakage during interaction of chromium(III) and hydrogen peroxide in vitro: Evidence for a chromium(III)-mediated Fenton-like reaction. Chem.-Biol. Interact. 1996, 102, 133–153. (27) Haight, G. P., Jr.; Huang, T. J.; Shakhashiri, B. Z. Reactions of Cr(IV). J. Inorg. Nucl. Chem. 1971, 33, 2169–2176. (28) Shi, X.; Dalal, N. S. ESR spin trapping detection of hydroxyl radicals in the reactions of Cr(V) complexes with hydrogen peroxide. Free Radical Res. Commun. 1990, 10, 17–26. (29) Shi, X.; Dalal, N. S.; Kasprzak, K. S. Generation of free radicals from hydrogen peroxide and lipid hydroperoxides in the presence of Cr(III). Arch. Biochem. Biophys. 1993, 302, 294–299. (30) Codd, R.; Dillon, C. T.; Levina, A.; Lay, P. A. Studies on the genotoxicity of chromium: From the test tube to the cell. Coord. Chem. Rev. 2001, 216, 537–582. (31) Zhang, L.; Lay, P. A. EPR spectroscopic studies on the formation of chromium(V) peroxo complexes in the reaction of chromium(VI) with hydrogen peroxide. Inorg. Chem. 1998, 37, 1729–1733. (32) Perez-Benito, J. F.; Arias, C. A kinetic study of the chromium(VI)-hydrogen peroxide reaction. Role of the diperoxochromate(VI) intermediates. J. Phys. Chem. A 1997, 101, 4726–4733. (33) St€unzi, H.; Spiccia, L.; Rotzinger, F. P.; Marty, W. Early stages of the hydrolysis of chromium(III) in aqueous solution. 4. Stability constants of the hydrolytic dimer, trimer, and tetramer at 25 °C and I = 1.0 M. Inorg. Chem. 1989, 28, 66–71. (34) St€unzi, H.; Marty, W. Early stages of the hydrolysis of chromium(III) in aqueous solution. 1. Characterization of a tetrameric species. Inorg. Chem. 1983, 22, 2145–2150. (35) Rao, L.; Zhang, Z.; Friese, J. I.; Ritherdon, B.; Clark, S. B.; Hess, N. J.; Rai, D. Oligomerization of chromium(III) and its impact on the oxidation of chromium(III) by hydrogen peroxide in alkaline solutions. J. Chem. Soc., Dalton Trans. 2002, 267–274. (36) Szejtli, J. Introduction and general overview of cyclodextrin chemistry. Chem. Rev. 1998, 98, 1743–1753. (37) Norkus, E. Metal ion complexes with native cyclodextrins. J. Inclusion Phenom. Macrocyclic Chem. 2009, 65, 237–248. (38) McNamara, M.; Russell, N. R. FT-IR and Raman spectra of a series of metallo-β-cyclodextrin complexes. J. Inclusion Phenom. Mol. Recogn. Chem. 1991, 10, 485–495. (39) St€unzi, H.; Rotzinger, F. P.; Marty, W. Early stages of the hydrolysis of chromium(III) in aqueous solution. 2. Kinetics and mechanism of the interconversion between two tetrameric species. Inorg. Chem. 1984, 23, 2160–2164. (40) Ensing, B.; Buda, F.; Baerends, E. J. Fenton-like chemistry in water: Oxidation catalysis by Fe(III) and H2O2. J. Phys. Chem. A 2003, 107, 5722–5731. (41) Lay, P. A. Recent developments on the mechanisms of substitution reactions of octahedral coordination complexes. Coord. Chem. Rev. 1991, 110, 213–233. (42) Xu, F.-C.; Krouse, R.; Swaddle, T. W. Conjugate base pathway for water exchange on aqueous chromium(III): Variable-pressure and temperature kinetic study. Inorg. Chem. 1985, 24, 267–270. (43) Espenson, J. H. Formation rates of monosubstituted chromium(III) complexes in aqueous solution. Inorg. Chem. 1969, 8, 1554–1556. (44) Crimp, S. J.; Spiccia, L.; Krouse, H. R.; Swaddle, T. W. Early stages of the hydrolysis of chromium(III) in aqueous solution. 9. Kinetics of water exchange on the hydrolytic dimer. Inorg. Chem. 1994, 33, 465–470. (45) Pettine, M.; Gennari, F.; Campanella, L.; Millero, F. J. The effect of organic compounds in the oxidation kinetics of Cr(III) by H2O2. Geochim. Cosmochim. Acta 2008, 72, 5692–5707.

9338

dx.doi.org/10.1021/es2021704 |Environ. Sci. Technol. 2011, 45, 9332–9338

ARTICLE pubs.acs.org/est

Catalytic Ozonation of Oxalate with a Cerium Supported Palladium Oxide: An Efficient Degradation Not Relying on Hydroxyl Radical Oxidation Tao Zhang,† Weiwei Li,‡ and Jean-Philippe Croue*,† †

Water Desalination and Reuse Center (WDRC), King Abdullah University of Science and Technology (KAUST), Thuwal 4700, Kingdom of Saudi Arabia ‡ Research Center for Eco-Environmental Sciences (RCEES), Chinese Academy of Sciences, Beijing 100085, China

bS Supporting Information ABSTRACT: The cerium supported palladium oxide (PdO/ CeO2) at a low palladium loading was found very effective in catalytic ozonation of oxalate, a probe compound that is difficult to be efficiently degraded in water with hydroxyl radical oxidation and one of the major byproducts in ozonation of organic matter. The oxalate was degraded into CO2 during the catalytic ozonation. The molar ratio of oxalate degraded to ozone consumption increased with increasing catalyst dose and decreasing ozone dosage and pH under the conditions of this study. The maximum molar ratio reached around 1, meaning that the catalyst was highly active and selective for oxalate degradation in water. The catalytic ozonation, which showed relatively stable activity, does not promote hydroxyl radical generation from ozone. Analysis with ATR-FTIR and in situ Raman spectroscopy revealed that 1) oxalate was adsorbed on CeO2 of the catalyst forming surface complexes, and 2) O3 was adsorbed on PdO of the catalyst and further decomposed to surface atomic oxygen (*O), surface peroxide (*O2), and O2 gas in sequence. The results indicate that the high activity of the catalyst is related to the synergetic function of PdO and CeO2 in that the surface atomic oxygen readily reacts with the surface cerium-oxalate complex. This kind of catalytic ozonation would be potentially effective for the degradation of polar refractory organic pollutants and hydrophilic natural organic matter.

’ INTRODUCTION Catalytic ozonation with metal oxides is a potential process to enhance the degradation of recalcitrant organics in water. It is generally accepted that catalytic ozonation with metal oxides can be ascribed to either of the two pathways: ozone decomposition on the catalyst surface generating hydroxyl radicals and direct ozone oxidation of surface metal
organic complexes.1 In the past decade, many studies were conducted on catalytic ozonation that follows the hydroxyl radical pathway. Successful works were accomplished on the preparation of efficient catalyst accelerating hydroxyl radical production from ozone and the identification of active sites involved in this process.2
4 With regard to catalytic ozonation relying on surface complexes, a lot of publications refer to the degradation of organic acids with mono- or hybrid- metal oxides as catalysts.5
7 However, minor achievements were made in this domain of research because of the lack of highly efficient catalysts as well as the difficulties in identifying active sites and active oxidant species, information that is essential to understand the mechanism and optimize the catalyst preparation.1 Hydroxyl radical has high reaction rate constants with almost all organics in water. However, hydroxyl radical oxidation is not effective to degrade aliphatic hydrophilic compounds that contain carbonyl or carboxylic groups, e.g., the products formed during the ozonation of natural organic matter.8 This is because the consumption of hydroxyl radical by bicarbonates/carbonates r 2011 American Chemical Society

(k = 8.5  106/3.9  108 M
1 s
1) and ozone (k = 1  108
2  109 M
1 s
1) are usually faster than its reactions with the saturated compounds.8 From this point, catalytic ozonation through a complexation pathway would be more selective and efficient for the degradation of saturated hydrophilic organics, if highly active catalyst can be prepared. Oxalate is one of the major byproducts in ozonation or advanced oxidation of natural organic matter and organic pollutants.9 Its degradation in reaction with both molecular ozone (k e 0.04 M
1 s
1) and hydroxyl radical (k = 7.7  106 M
1 s
1) are relatively low.10 Therefore, it is usually used as a probe compound to study catalytic ozonation that follows surface complexation pathway. MnO2, Fe2O3, Co3O4, and NiO had been found effective in the catalytic ozonation of oxalic acid but the removal rates at normal pHs in a reasonable reaction time were too low and the ozone doses were too high for practical uses.5,7,11,12 There are still no data on ozone consumption for oxalate degradation which is critical to evaluate the catalysis efficiency. No evidence shows active sites and active oxidant species that are responsible for oxalate degradation, the information Received: June 28, 2011 Accepted: October 4, 2011 Revised: October 1, 2011 Published: October 04, 2011 9339

dx.doi.org/10.1021/es202209j | Environ. Sci. Technol. 2011, 45, 9339–9346

Environmental Science & Technology that is needed to better understand the reaction pathway and tailor cost-effective catalysts. PdO supported on ceria-zirconium oxide has been used successfully as a three-way catalyst for automobile-exhausted gases to reduce NO and oxidize CO and hydrocarbons.13 With regard to catalytic ozonation of organic compounds in water with supported PdO, there is still no report. PdO had been found effective in decomposing aqueous ozone,14 but intermediates in this process are unknown. CeO2 can reduce bromate formation during ozonation, while it is inactive in promoting ozone decomposition and organic compound degradation.15 In this study, the effectiveness and efficiency of a cerium supported palladium oxide (PdO/CeO2) in catalytic ozonation was tested using oxalate as a probe compound. In order to elucidate the efficient pathway for catalytic ozonation not relying on hydroxyl radical, surface active sites of the catalyst and surface oxygen intermediates formed from ozone decomposition were investigated with ATR-FTIR and in situ Raman spectroscopy.

’ EXPERIMENTAL SECTION Metal Oxides Preparation. The support CeO2 was synthesized with a urea-hydrothermal method. Ce(NO3)3 3 6H2O and urea were dissolved in distilled water with a molar ratio of 1:3. The mixture was heated at 140 °C for 5 h. After filtration and repeated washing, the precipitate was dried at 120 °C for 2 h and calcined at 450 °C for 4 h. PdO/CeO2 was prepared by impregnating the CeO2 with Pd(NO3)2 aqueous solution with incipient wetness. The impregnated oxide was dried at 60 °C and finally calcined in air at 550 °C for 2 h. Palladium mass proportion of the PdO/CeO2 was measured to be 3.8%. PdO particle was prepared by direct calcination of the dried Pd(NO3)2 at 550 °C for 2 h. Characterization. BET surface area of the metal oxides was determined on a Micromeritics ASAP2000 analyzer. pHpzc (pH at which the surface is zero-charged) was determined with acid
base titration. Average particle size was measured on a Mastersizer 2000 laser particle size analyzer. STEM (scanning transmission electron microscopy) pictures taken on a Titan 80
300 transmission electronic microscope were used to characterize the dispersion of PdO on CeO2. The major characteristics of the metal oxides used in this study are listed in Table S1 (Supporting Information). Experimental Procedure. Batch Reaction. Milli-Q water at room temperature (21°C) was continuously bubbled with gaseous ozone produced with an ozone generator (3S-A5, Tonglin Technology) from dried oxygen gas. The aqueous ozone concentration was analyzed continuously with an ultraviolet spectrometer (Hach 500) at 258 nm (molar absorbance coefficient = 3000 M
1 cm
1) until it reached a steady state. The steady ozone concentration in water was controlled by adjusting the electric current of the ozone generator. Predetermined volume of ozone stock solution was quickly mixed with tetraborate buffered oxalate solution in a glass reactor. In the case of catalytic ozonation, the catalyst (mostly at a dose of 150 mg L
1 unless specified) was also instantly introduced into the reaction solution. Then, the reactor was sealed and magnetically stirred. Samples were taken at each time-point and then filtered through 0.45 μm acetate-fiber syringe filters and purged with pure N2 to remove residual ozone in water. The filtration had no impact on the oxalate concentration. In order to reduce the impact of filtration on aqueous ozone, 50 mL of the ozone stock solution was pressed through the filter before sample filtration for ozone

ARTICLE

analysis. Then, the filtration had nearly no impact on the aqueous ozone concentration when the filter had been pretreated in this way. Semicontinuous Reaction. Gaseous ozone (concentration = 21 mg O3 L
1, flow rate = 3.0 L min
1) was continuously introduced into a 500 mL reaction solution (0.2 mM oxalate in 10 mM tetra-borate buffer) in a glass vessel. After the first 10 min, 10 mL of reaction solution was withdrawn from the reaction vessel with a syringe. Subsequently, the same volume of oxalate stock solution (10 mM, buffered with 10 mM tetraborate) was immediately added into the reactor. The same operation was repeated 30 times. The samples were filtered with 0.45 μm acetate fiber filters and then purged with pure N2 to remove residual ozone. Analysis. Ozone concentration in reaction solution was directly determined on the UV spectrometer at 258 nm after the filtration, using the tetraborate buffer (10 mM and pH 6.5) as zero background, because oxalate had no detectable absorbance at this wavelength. Oxalate was analyzed on a Dionex ICS-1600 IC equipped with an AS-9 column. The mobile phase was 9 mM Na2CO3 at a flow rate of 1.0 mL min
1. TOC was measured on a Teledyne Tekmar TOC Fusion analyzer. Atrazine used as a probe compound of hydroxyl radical in this study was determined on a Waters HPLC equipped with a Symmetry C-18 column at a UV wavelength of 220 nm. The mobile phase was isocratic H2O/acetonitrile at a volume ratio of 3/7 and a flow rate of 1.0 mL min
1. Dissolved palladium and cerium ions were determined on an ICP-MS (Agilent 7500) with detection limits of 9.4  10
4 and 1.7  10
2 μg L
1 for the two cations, respectively. Palladium content of the catalyst was also determined by the ICP-MS after digestion of the catalyst with HCl + HNO3 (v/v = 3:1) and HF in sequence. A Perkin Elmer FTIR spectrometer (Spectrum 100) equipped with a Universal ATR accessory was used to characterize the catalyst surface in presence or absence of ozone and oxalate. In the study of aqueous ozone adsorption, the suspension of metal oxide (750 mg L
1) was continuously bubbled with gaseous ozone (concentration = 21 mg L
1, flow rate = 3.0 L min
1) for over 10 min in a 25 mL glass tube which was cooled with ice. The suspension was quickly dropped on the ZnSe crystal of the ATR accessory with a glass pipet, covered with a stainless lid, and scanned in the range of 800
4000 cm
1 at a resolution of 4 cm
1. In the study of oxalate adsorption, 0.2 mM oxalate in 10 mM tetraborate buffer of pH 6.5 was mixed with metal oxide particles (150 mg L
1) and stirred for over 2 h. After settling, the particles were dropped on the ATR crystal and analyzed at the same conditions as that described above. In situ Raman spectra were taken on a confocal microscopic Raman spectrometer (Aramis, Horiba Jobin Yvon) with a 9 mW 633 nm laser light irradiation. Before analysis, the metal oxides were pressed into slices of 1
2 mm in thickness and 13 mm in diameter. The slice was then stuck on a microscope slide with a double-sided tape, wetted with Milli-Q water, and then blown with gaseous ozone (concentration = 1.5 mg L
1, flow rate = 1.4 L min
1) which was humidified through a gas washer. The slice was scanned from 300 to 1200 cm
1 at a resolution of 1 cm
1 and a duration time of 100 s under the humidified ozone gas.

’ RESULTS AND DISCUSSION Effectiveness. Figure 1A shows the decrease in oxalate concentration in batch reaction mode during catalytic ozonation 9340

dx.doi.org/10.1021/es202209j |Environ. Sci. Technol. 2011, 45, 9339–9346

Environmental Science & Technology

ARTICLE

Figure 1. Oxalate removal in catalytic ozonation and adsorption alone with the single and the binary oxides (A) and ozone decomposition in the catalytic ozonation (B). Experimental conditions: ozone dose = 0.18 mM, initial oxalate concentration = 0.2 mM, oxide dose = 150 mg L
1, T = 21 °C, 10 mM tetraborate buffered pH = 6.5.

Figure 2. Effect of catalyst dose (a- 0 mg L
1, b- 70 mg L
1, c- 150 mg L
1, and d- 250 mg L
1) on oxalate removal (A) and the effective ozone consumption ratio (B) in the catalytic ozonation. Experimental conditions: ozone dose = 0.18 mM, initial oxalate concentration = 0.2 mM, T = 21 °C, 10 mM tetraborate buffered pH = 6.5.

and adsorption alone with CeO2, PdO, and PdO/CeO2. Oxalate was quite stable during ozonation alone (absence of catalyst). Its concentration decreased less than 0.014 mM in 12 min at the ozone dose of 0.18 mM. Oxalate loss in O3/CeO2 approximated the sum of that removed during ozonation alone and that in CeO2 adsorption alone, meaning that CeO2 has no activity in oxalate degradation. PdO showed nearly no adsorption for oxalate. However, it promoted oxalate removal during catalytic ozonation to 0.037 mM. Even PdO mass on the PdO/CeO2 was less than 4%, the binary oxide showed a much higher efficiency than PdO. Oxalate loss reached to 0.08 mM in O3/(PdO/CeO2), which can largely be attributed to degradation because the oxalate loss in PdO/CeO2 adsorption alone was only 0.014 mM. After catalytic ozonation with PdO/CeO2, the presence of cerium and palladium ions in the reaction solution were examined. The cerium ion concentration was below the detection limit. The concentration of palladium ion was 0.24 μg L
1, which was 4.2  10
5 times the one of the solid PdO dose. The suspension was filtered with a 0.45 μm filter and ozonated again without the presence of PdO/CeO2. No further oxalate degradation was observed (not shown). Therefore, the oxalate degradation is due to heterogeneous catalysis but not due to catalysis effect of trace palladium ion in water. It was reported that effective catalytic oxalate degradation with several typical metal oxides were achieved only at much lower pHs (e.g., 4.1 and 3.2 for MnO2, 2.5 for Fe2O3 and Co2O3, and 2.4 for NiO) and continuous ozone introduction during the reaction over 30 min.5,7,11,12 This binary oxide seems to be more efficient for practical catalytic ozonation reaction.

The ozone decay recorded during ozonation in the presence or in the absence of the oxides exhibited a similar pattern as the oxalate removal profile (Figure 1B). Consistent with previous results,15 CeO2 nearly did not improve ozone decomposition. The ozone depletion in water was promoted by PdO and to a higher rate by PdO/CeO2. PdO seems to be an active metal oxide in oxalate degradation and ozone decomposition. The activity was significantly improved by CeO2 support. The disperse effect of CeO2 for small PdO crystals on the surface (i.e., the effect against PdO particle agglomeration) (Figure S1, Supporting Information) as well as the adsorption of oxalate on CeO2 possibly contributed to the high activity of the PdO/CeO2. Effect of Catalyst Dose. The increase of PdO/CeO2 dose from 0 to 250 mg L
1 significantly accelerated oxalate removal (Figure 2) as well as ozone decomposition (Figure S2A, Supporting Information). The oxalate removal in the presence of ozone (i.e., catalytic ozonation) was much more significant than that in the absence of ozone (i.e., adsorption alone) (Figure S2B, Supporting Information). Moreover, ozonation did not improve the absorbability of the catalyst for oxalate (Figure S2B, Supporting Information). Therefore, the high oxalate removal in the catalytic ozonation was due to degradation but not adsorption. It is interesting that a further loss of oxalate concentration by 0.02 mM was observed during the catalytic ozonation with 250 mg L
1 of PdO/CeO2 when ozone in water had been completely consumed after 12 min. Since oxalate concentration in adsorption alone reached equilibrium within 3 min at this PdO/CeO2 dose, the further oxalate loss might be due to the degradation of adsorbed 9341

dx.doi.org/10.1021/es202209j |Environ. Sci. Technol. 2011, 45, 9339–9346

Environmental Science & Technology

Figure 3. Effect of ozone dose (a- 0.05 mM, b- 0.13 mM, c- 0.18 mM) on oxalate removal (A) and the effective ozone consumption ratio (B) in the catalytic ozonation. Experimental conditions: initial oxalate concentration = 0.2 mM, catalyst dose = 150 mg L
1, T = 21 °C, 10 mM tetraborate buffered pH = 6.5.

oxalate by adsorbed ozone or new oxidant species and a further adsorption. The loss of oxalate during the catalytic ozonation with different catalyst doses compensated well for TOC loss (Figure S3, Supporting Information). It means that the oxalate was oxidized directly into CO2 with no formation of other stable organic intermediates, consistent with the results found in homogeneous catalytic ozonation of oxalate with Co2+.10 The effective ozone consumption ratio for oxalate degradation was calculated as the oxalate degradation (total removal subtracted by the removal in adsorption alone) per mole of ozone consumed. Figure 2B shows that this ratio substantially increased from 0.05 to 0.5 in 25 min reaction as the catalyst dose increased from 0 to 250 mg L
1. Effect of Ozone Dose. Over half of oxalate removal always occurred within the first 3 min of reaction in O3/(PdO/CeO2) (shown in Figure 1A and 2A). It was expected that the increase of ozone dose will improve oxalate removal in this initial reaction phase. However, the increase of ozone dose from 0.05 to 0.18 mM reduced oxalate removal rate (Figure 3A). The positive effect of increasing ozone dose on oxalate degradation rate was observed only after the initial phase. Because aqueous ozone concentration was relatively high in the initial phase compared with the following phase (Figure S4, Supporting Information), the result suggests that molecular ozone might not be the oxidant species directly responsible for the oxalate degradation. It is likely that several oxygen species can be formed in sequence from ozone decomposition on the catalyst surface. Ozone might compete with oxalate to react with some of the new oxygen species that are effective for oxalate degradation, thus leading to

ARTICLE

Figure 4. Effect of pH on oxalate removal in ozonation alone (solid symbols) and catalytic ozonation (open symbols) (A) and the effective ozone consumption ratio (B). Experimental conditions: ozone dose = 0.09 mM, initial oxalate concentration = 0.1 mM, catalyst dose = 150 mg L
1, T = 21 °C, 10 mM tetraborate buffer and diluted HNO3 adjusted pH, 2.5 mM bicarbonate was also used for pH 7.8.

the negative effect on oxalate degradation in the initial phase when ozone dose was raised. At the dose of 0.05 mM, ozone in water was totally consumed in 5 min (Figure S4, Supporting Information). However, oxalate was further removed by 0.007 mM thereafter, which is similar to that observed in Figure 2A. The effective ozone consumption ratio for oxalate degradation increased from 0.5 to around 1.0 as the ozone dose decreased from 0.18 to 0.05 mM (Figure 3B). It may indicate that ozone reacts with some of the new oxidant species that are effective for oxalate degradation, thus reducing the effective ozone consumption ratio at high ozone doses. Effect of pH. Owing to ozone reactions with OH
and some dissociating organics, ozone decomposition in water and the hydroxyl radical generation can be significantly accelerated at elevated pHs.8,16,17 Ozone decomposition at pH 9.2 was so fast that there was nearly no difference in residual ozone between ozonation alone and catalytic ozonation (Figure S5, Supporting Information). The fast ozone decomposition at this pH definitely accelerated hydroxyl radical generation. However, only about 5% of oxalate was removed during ozonation alone and catalytic ozonation (Figure 4A). The low efficiency of hydroxyl radical oxidation can be ascribed to its much higher reaction rate with molecular ozone (1  108
2  109 M
1 s
1)8 than oxalate oxidation (7.7  106 M
1 s
1).10 The disappearance of catalytic effect at this pH can be ascribed to the fast ozone decomposition in water, which reduced significantly the chances of ozonecatalyst surface interaction. As the pH was decreased, oxalate was removed by 25%, 40%, 70%, and 98% during catalytic ozonation at pH 7.8, 7.0, 6.5, and 4.2, respectively (Figure 4A). In parallel, the ozone decomposition 9342

dx.doi.org/10.1021/es202209j |Environ. Sci. Technol. 2011, 45, 9339–9346

Environmental Science & Technology rate was also enhanced by the catalyst with that at pH 4.2 was slightly higher than that at pH 6.5 (Figure S5, Supporting Information). A similar observation was described during the catalytic ozonation of oxalate with Co2+.10 Oxalate adsorption was promoted by the decrease of pH. About 2%, 9%, and 28% oxalate was removed in adsorption alone at pH 7.8, 6.5 and 4.2, respectively (not shown). As the pHpzc of PdO/CeO2 is 5.4, it became more positively charged at lower pH, improving oxalate adsorption through electrostatic attraction. Therefore, the improvement of oxalate removal at a lower pH is related to enhanced oxalate adsorption on the catalyst which might further promote surface oxalate degradation during catalytic ozonation. The effective ozone consumption ratio in catalytic ozonation also increased from 0.06 to 0.78 as the pH decreased from 9.2 to 4.2 (Figure 4B). The low ratio at alkaline pH possibly is due to 1) fast ozone depletion in water which reduced the chances for ozone-catalyst interaction and 2) activity decrease of the catalyst caused by the negative effect of high pH on oxalate adsorption. No Hydroxyl Radical Generation. The introduction of t-BuOH into the reaction solution showed nearly no influence on ozone decomposition and oxalate degradation in the catalytic ozonation (Figure S6A and S6B, Supporting Information). It is clear that hydroxyl radical is not involved in the catalytic ozone decomposition and oxalate degradation. Atrazine (kO3 = 6 M
1 s
1, k•OH = 3  109 M
1 s
1)8 was used as an additional probe compound for the catalytic ozonation (Figure S7, Supporting Information). Degradation rates of atrazine at trace level during catalytic ozonation in the presence or the absence of oxalate were similar. Pines and Reckhow observed that hydroxyl radical was a byproduct of oxalate degradation in catalytic ozonation with cobalt ion.10 However, no hydroxyl radical was generated from the catalytic oxalate degradation here, indicating that the degradation pathways involved in the two processes are different. The atrazine degradation rate during catalytic ozonation was much lower than during ozonation alone, proving again that hydroxyl radical was not generated from the catalytic ozone decomposition. Stable Activity. Oxalate stock solution was intermittently added into the reactor by 30 times in the semicontinuous catalytic ozonation to test the stability of the catalyst. The amount of oxalate added into the reaction solution at the end of each 10 min reaction can increase oxalate concentration by 0.2 mM. In most cases, nearly all of the oxalate added was degraded within the 10 min reaction (Figure S8, Supporting Information). No decrease of the oxalate removal was observed with the increase of reaction runs in the semicontinuous reaction. Therefore, the activity of PdO/CeO2 remained relatively stable for our experimental conditions. Active Sites, Oxidant Species, and Reaction Pathway. Figure 5A-C shows ATR-FTIR spectra obtained from PdO/ CeO2, CeO2, and PdO in water with and without the presence of ozone and oxalate, respectively. New double reflectance bands with weak intensities appeared for PdO/CeO2 (2086 and 2048 cm
1) and PdO (2040 and 2014 cm
1) in contact with aqueous ozone. According to refs 18
20, these bands can be assigned to distorted molecular ozone vibrations with interaction of ozone with surface metal sites. There was no adsorbed ozone features for CeO2 in contact with aqueous ozone. These results indicate that PdO is the component that adsorbs ozone. The reflectance of adsorbed ozone on the catalyst shifted to higher wavenumbers as compared with PdO alone, suggesting that the CeO2-supported PdO has a stronger binding or distortion effect

ARTICLE

Figure 5. ATR-FTIR spectra of PdO/CeO2 (A), CeO2 (B), and PdO (C) in contact with water (curve a), aqueous ozone (curve b), and aqueous oxalate (curve c).

for adsorbed ozone molecule. A stronger distortion can make the adsorbed ozone more unstable against dissociation.18 Therefore, the faster ozone decomposition on the catalyst than that on PdO alone (shown in Figure 1B) can be related to the higher affinity of the catalyst for ozone molecule. New reflectance bands at 1426 and 1432 cm
1 appeared for PdO/CeO2 and CeO2 in contact with oxalate solution, respectively. Oxalate was reported to have two characteristic IR bands in the region of 1650
1550 and 1400 cm
1.21 However, oxalate in water only showed the apparent 1400 cm
1 peak with the ATR-FTIR (Figure S9, Supporting Information), probably because pure water that was used for a background scan also had a reflectance around 1600 cm
1. The 1400 cm
1 band which is due to CdO stretch shifted to higher wavenumbers in the presence of CeO2 and PdO/CeO2. According to Marley et al.,21 this result indicates that oxalate was adsorbed onto cerium sites of the oxides forming surface complexes. Although oxalate can be adsorbed on CeO2 through surface complexation, CeO2 alone had nearly no activity for oxalate 9343

dx.doi.org/10.1021/es202209j |Environ. Sci. Technol. 2011, 45, 9339–9346

Environmental Science & Technology

ARTICLE

The lattice CeIII content increase can stabilize surface metal ions when their oxidation states increase.28 It is likely that here the lattice oxygen vacancy formation in the CeO2 support is due to the stabilization of the surface palladium ions at the strongly oxidative atmosphere. Two new broad peaks appeared at 828 and 937 cm
1. They are characteristic features of surface peroxide (*O2) and surface atomic oxygen (*O), respectively.19,22
24 When the ozone flow was cut off, the peaks slowly decreased (Figure S10, Supporting Information), indicating the catalyst can recover itself. The Raman spectra of CeO2 with and without the presence of ozone were nearly the same (Figure 6B). No intensity increase at 556 cm
1 was observed in the presence of ozone, meaning that the oxygen vacancies cannot be formed when ozone contacts with CeO2 alone. The features of surface palladium oxide, surface peroxide, and atomic oxygen were all observed for PdO in contact with ozone (Figure 6C). However, the peak wavenumbers of surface PdO (630 cm
1) and atomic oxygen (912 cm
1) were lower than the values observed for PdO/CeO2 and higher for peroxide (837 cm
1). This result indicates that there is a strong interaction between the CeO2 support and the palladium ion which led to the changes of binding strength between palladium and these surface oxygen-containing species. Our results clearly showed that PdO on the catalyst surface is the active site in inducing the decomposition of ozone into the intermediate oxygen species. Since ozone can adsorb onto PdO (shown in Figure 5A and 5C), it is possible that the adsorbed ozone further decomposed to surface atomic oxygen and gaseous O2 (eq 1). Another ozone molecule would react with the surface atomic oxygen forming a surface peroxide species and a gaseous O2 (eq 2). Because the peak of the surface peroxide disappeared when ozone was removed, the surface peroxide would further decompose to gaseous O2 (eq 3). Such a catalytic ozone decomposition process under the mimic aqueous condition is consistent with gaseous ozone decomposition on manganese oxides19,22 Figure 6. Raman spectra of PdO/CeO2 (A), CeO2 (B), and PdO (C) with and without the presence of ozone.

degradation during ozonation (shown in Figure 1A). Therefore, direct oxidation of the cerium-oxalate complex by molecular ozone is still not the catalytic ozonation pathway. It is likely that the adsorbed ozone molecule can produce more active surface oxygen species that are quite selective for the complex degradation. In situ Raman spectroscopy had been used to characterize intermediate oxygen species formed on manganese oxides in contact with gaseous ozone.19,22 It was applied here to get insights on ozone decomposition on the catalyst. Figure 6A-C shows Raman spectra of PdO/CeO2, CeO2, and PdO respectively with and without the presence of ozone. Ozone itself had no Raman signal in this spectrum range. The intensities of the peak at 556 cm
1 and the shoulder at 644 cm
1 increased significantly for PdO/CeO2 in contact with ozone (Figure 6A). The 556 cm
1 feature arises from lattice oxygen vacancies in CeO2.23,24 The 644 cm
1 is ascribed to new palladium oxide species.25,26 CeO2 has a special property of changing oxidation state of lattice Ce between CeIII and CeIV through oxygen release and storage.27 The intensity increase of the 556 cm
1 peak means that lattice CeIII content in the CeO2 support increased.

Pd þ O3 f PdO þ O2

ð1Þ

PdO þ O3 f PdO2 þ O2

ð2Þ

PdO2 f Pd þ O2

ð3Þ

The Raman feature of surface atomic oxygen is attributed to the stretches of metal
oxygen double bond MedO.29 The higher PddO wavenumber observed on the PdO/CeO2 indicates that the CeO2 support led to stronger affinity of palladium for the atomic oxygen. Such a kind of palladium-atomic oxygen interaction on the catalyst in comparison with that on PdO probably would increase the stability of the atomic oxygen and consequently reduce the rate of its reaction with ozone which forms surface peroxide. Because the oxidation potential of atomic oxygen (2.43 V in water) is much higher than that of peroxide (1.35 V in protonated form),30 the surface atomic oxygen is likely the active oxygen species in the catalytic ozonation. The maximum effective ozone consumption ratio of the catalyst was observed to be around 1 for oxalate degradation under our experimental conditions (shown in Figure 3B). This result reinforces the hypothesis that the surface atomic oxygen is the major effective oxidant in the catalytic ozonation. If surface peroxide was effective for oxalate degradation, the maximum effective ozone consumption ratio would be around 0.5 (2 mols 9344

dx.doi.org/10.1021/es202209j |Environ. Sci. Technol. 2011, 45, 9339–9346

Environmental Science & Technology

ARTICLE

of O3 needed to produce surface peroxide). This point is also supported by the fact that high ozone concentration showed negative effect on oxalate degradation in the initial phase (shown in Figure 3A). Ozone in excess may react with surface atomic oxygen to produce surface peroxide. Our reaction mechanism hypothesis agrees with the fact that further oxalate degradation is observed after complete consumption of aqueous ozone (Figure 2A and 3A) because the adsorbed ozone on the catalyst decomposes to form atomic oxygen and it can survive for some time in absence of aqueous ozone (Figure S10, Supporting Information). It has been reported that oxidative decarboxylation of CeIVcarboxylates occurs under heating conditions in water through one-electron transfer forming alkyl radical and two-electron transfer forming carbonium ion (eqs 4 and 5)31,32 CeIV ðO2 CRÞ3 þ f CeIII ðO2 CRÞ2 þ þ CO2 þ R •

ð4Þ

R • þ CeIV f CeIII þ R þ

ð5Þ

In the surface cerium-oxalate complex, oxalate partially donates its electron density to CeIV. Because atomic oxygen has a higher oxidation potential than CeIV (1.70 V), it is likely that the surface atomic oxygen extracts one or two electrons from the surface cerium-oxalate complex, thus initiating oxidative decarboxylation under mild conditions without reducing surface CeIV. The high efficiency of the catalytic ozonation with PdO/CeO2 in oxalate degradation can then be related to a synergetic effect between PdO and the support: 1) PdO acts as the active site to decompose ozone to surface atomic oxygen, and 2) the support CeO2 activates oxalate in reaction with the surface atomic oxygen through forming surface complexes. In addition, the CeO2 support also promotes the adhesion and decomposition of ozone on PdO. The strong affinity probably increased the stability of the surface atomic oxygen against the fast reaction with ozone forming peroxide. This work shows potentially effective degradation for polar refractory organics with composite metal oxide-assisted ozonation. It would be also promising for the degradation of hydrophilic natural organic matter (NOM) with high carboxylic contents known as refractory to conventional water treatments and one of the precursors of disinfection byproduct.33 Future research in following aspects might be necessary for the preparation of active catalysts and the application: 1) the influence of coordination affinity for target compounds on its degradation efficiency when another metal oxide is combined with CeO2 as a hybrid support, 2) activity changes when other metal oxides that can decompose ozone are supported on CeO2 as ozone active sites, and 3) degradation of polar refractory compounds and hydrophilic NOM components that cannot be effectively removed in hydroxyl radical oxidation.

’ ASSOCIATED CONTENT

bS

Supporting Information. One table and ten figures. This material is available free of charge via the Internet at http://pubs. acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: + 966 (0) 2 808 2984. E-mail: [email protected].

’ ACKNOWLEDGMENT We want to thank Dr. Yang Yang and Mr. Qingxiao Wang of Imaging and Characterization Laboratory of KAUST for their help in performing Raman and STEM analysis and Dr. Cyril Aubry, Ms. Tong Zhan, and Dr. Min Yoon of WDRC of KAUST in TEM, ICP-MS, and IC analysis. We also want to thank our anonymous reviewers for their valuable comments to improve this work. ’ REFERENCES (1) Nawrocki, J.; Kasprzyk-Horden, B. The efficiency and mechanisms of catalytic ozonation. Appl. Catal., B 2010, 99, 27–42. (2) Zhang, T.; Li, C.; Ma, J.; Tian, H.; Qiang, Z. Surface hydroxyl groups of synthetic a-FeOOH in promoting •OH generation from aqueous ozone: Property and activity relationship. Appl. Catal., B 2008, 82, 131–137. (3) Yang, L.; Hu, C.; Nie, Y.; Qu, J. Catalytic ozonation of selected pharmaceuticals over mesoporous alumina-supported manganese oxide. Environ. Sci. Technol. 2009, 43, 2525–2529. (4) Zhao, L.; Sun, Z.; Ma, J. Novel relationship between hydroxyl radical initiation and surface group of ceramic honeycomb supported metals for the catalytic ozonation of nitrobenzene in aqueous solution. Environ. Sci. Technol. 2009, 43, 4157–4163. (5) Andreozzi, R.; Insola, A.; Caprio, V.; Marotta, R.; Tufano, V. The use of manganese dioxide as a heterogeneous catalyst for oxalic acid ozonation in aqueous solution. Appl. Catal., A 1996, 138, 75–81. (6) Delanoe, F.; Acedo, B.; Vel Leitner, N. K.; Legube, B. Relationship between the structure of Ru/CeO2 catalysts and their activity in the catalytic ozonation of succinic acid aqueous solutions. Appl. Catal., B 2001, 29, 315–325. (7) Beltran, F. J.; Rivas, F. J.; Montero-de-Espinosa, R. Iron type catalysts for the ozonation of oxalic acid in water. Water Res. 2005, 39, 3553–3564. (8) von Gunten, U. Ozonation of drinking water: part I.Oxidation kinetics and product formation. Water Res. 2003, 37, 1443–1467. (9) Hammes, F.; Salhi, E.; Koster, O.; Kaiser, H. P.; Egli, T.; von Gunten, U. Mechanistic and kinetic evaluation of organic disinfection by-product and assimilable organic carbon (AOC) formation during the ozonation of drinking water. Water Res. 2006, 40, 2275–2286. (10) Pines, D. S.; Reckhow, D. A. Effect of dissolved cobalt(II) on the ozonation of oxalic acid. Environ. Sci. Technol. 2002, 36, 4046–4051. (11) Beltran, F. J.; Rivas, F. J.; Montero-de-Espinosa, R. Ozoneenhanced oxidation of oxalic acid in water with cobalt catalysts. 2. Heterogeneous catalytic ozonation. Ind. Eng. Chem. Res. 2003, 42, 3218– 3224. (12) Avramescu, S. M.; Bradu, C.; Udrea, I.; Mihalache, N.; Ruta, F. Degradation of oxalic acid from aqueous solutions by ozonation in presence of Ni/Al2O3 catalysts. Catal. Commun. 2008, 9, 2386–2391. (13) Jen, H. W.; Graham, G. W.; Chun, W.; McCabe, R. W.; Cuif, J. P.; Deutsch, S. E.; Touret, O. Characterization of model automotive exhaust catalysts: Pd on ceria and ceria
zirconia supports. Catal. Today 1999, 50, 309–328. (14) Lin, J.; Kawai, A.; Nakajima, T. Effective catalysts for decomposition of aqueous ozone. Appl. Catal., B 2002, 39, 157–165. (15) Zhang, T.; Chen, W.; Ma, J.; Qiang, Z. Minimizing bromate formation with cerium dioxide during ozonation of bromide-containing water. Water Res. 2008, 42, 3651–3658. (16) Hoigne, J.; Bader, H. Rate constants of reactions of ozone with organic and inorganic compounds in water. II. Dissociating organic compounds. Water Res. 1983, 17, 185–94. (17) Xiong, F.; Croue, J. P.; Legube, B. Long-term ozone consumption by aquatic fulvic acids acting as precursors of radical chain reactions. Environ. Sci. Technol. 1992, 26, 1059–1064. (18) Bulanin, K. M.; Lavalley, J. C.; Tsyganenko, A. A. IR spectra of adsorbed ozone. Colloids Surf., A 1995, 101, 153–158. 9345

dx.doi.org/10.1021/es202209j |Environ. Sci. Technol. 2011, 45, 9339–9346

Environmental Science & Technology

ARTICLE

(19) Radhakrishman, R.; Oyama, S. T. Ozone decomposition over manganese oxide supported on ZrO2 and TiO2: A kinetic study using in situ laser Raman spectroscopy. J. Catal. 2001, 199, 282–290. (20) Zeng, Y.; Liu, Z.; Qin, Z.; Liu, H. Infrared study on adsorption of O3 at SnO2 surface. Spectr. Spectral Anal. 2008, 28, 1035–1038. (21) Marley, N. A.; Bennett, P.; Janecky, D. R.; Gaffney, J. S. Spectroscopic evidence for organic diacid complexation with dissolved silica in aqueous systems I. Oxalic acid. Org. Geochem. 1989, 14, 525–528. (22) Li, W.; Gibbs, G. V.; Ted Oyama, S. Mechanism of ozone decomposition on a manganese oxide catalyst. 1. In situ Raman spectroscopy and Ab initio molecular orbital calculations. J. Am. Chem. Soc. 1998, 120, 9041–9046. (23) Wu, Z.; Li, M.; Howe, J.; Meyer, H. M.; Overbury, S. H. Probing defect sites on CeO2 nanocrystals with well-defined surface planes by Raman spectroscopy and O2 adsorption. Langmuir 2010, 26, 16595– 16606. (24) Vindigni, F.; Manzoli, M.; Damin, A.; Tabakova, T.; Zecchina, A. Surface and inner defects in Au/CeO2 WGS catalysts: Relation between Raman properties, reactivity and morphology. Chem.—Eur. J. 2011, 17, 4356–4361. (25) Otto, K.; Hubbard, C. P.; Weber, W. H.; Graham, G. W. Raman spectroscopy of palladium oxide on r-alumina applicable to automotive catalysts. Appl. Catal., B 1992, 1, 317–327. (26) Demoulin, O.; Navez, M.; Gaigneaux, E. M.; Ruiz, P.; Mamede, A. S.; Grangerb, P.; Payen, E. Operando resonance Raman spectroscopic characterization of the oxidation state of palladium in Pd/g-Al2O3 catalysts during the combustion of methane. Phys. Chem. Chem. Phys. 2003, 5, 4394–4401. (27) Yao, H. C.; Yu Yao, Y. F. Ceria in automotive exhaust catalysis I. Oxygen storage. J. Catal. 1984, 86, 254–265. (28) Mayernick, A. D.; Janik, M. J. Methane oxidation on Pd-Ceria: A DFT study of the mechanism over PdxCe1‑xO2, Pd, and PdO. J. Catal. 2011, 278, 16–25. (29) Che, M.; Tench, A. J. Characterization and reactivity of mononuclear oxygen species on oxide surfaces. Adv. Catal. 1982, 31, 78–128. (30) Bharara, M. S.; Atwood, D. A. Oxygen: Inorganic Chemistry. Encyclopedia of Inorganic Chemistry; John Wiley & Sons: New York, 2006. (31) Sheldon, R. A.; Kochi, J. K. Photochemical and thermal reduction of cerium (IV) carboxylates: Formation and oxidation of alkyl radicals. J. Am. Chem. Soc. 1968, 90, 6688–6698. (32) Serguchev, Y. A.; Beletskaya, I. P. Oxidative decarboxylation of carboxylic acids. Russ. Chem. Rev. 1980, 49, 1119–1134. (33) Dickenson, E. R. V.; Summers, R. S.; Croue, J. P.; Gallard, H. Haloacetic acid and trihalomethane formation from the chlorination and bromination of aliphatic β-dicarbonyl acid model compounds. Environ. Sci. Technol. 2008, 42, 3226–3233.

9346

dx.doi.org/10.1021/es202209j |Environ. Sci. Technol. 2011, 45, 9339–9346

ARTICLE pubs.acs.org/est

Odorous Compounds in Municipal Wastewater Effluent and Potable Water Reuse Systems Eva Agus,† Mong Hoo Lim,‡ Lifeng Zhang,‡ and David L. Sedlak†,* † ‡

Department of Civil and Environmental Engineering, University of California, Berkeley, California 94720, United States PUB, Singapore’s National Water Agency, 228231, Singapore

bS Supporting Information ABSTRACT: The presence of effluent-derived compounds with low odor thresholds can compromise the aesthetics of drinking water. The potent odorants 2,4,6-trichloroanisole and geosmin dominated the profile of odorous compounds in wastewater effluent with concentrations up to 2 orders of magnitude above their threshold values. Additional odorous compounds (e.g., vanillin, methylnaphthalenes, 2-pyrrolidone) also were identified in wastewater effluent by gas chromatography coupled with mass-spectrometry and olfactometry detection. Full-scale advanced treatment plants equipped with reverse osmosis membranes decreased odorant concentrations considerably, but several compounds were still present at concentrations above their odor thresholds after treatment. Other advanced treatment processes, including ozonation followed by biological activated carbon and UV/H2O2 also removed effluentderived odorants. However, no single treatment technology alone was able to reduce all odorant concentrations below their odor threshold values. To avoid the presence of odorous compounds in drinking water derived from wastewater effluent, it is necessary to apply multiple barriers during advanced treatment or to dilute wastewater effluent with water from other sources.

’ INTRODUCTION In many regions facing freshwater scarcity, municipal wastewater effluent constitutes a considerable part of the potable water supply. Over the past two decades, the practice of subjecting wastewater effluent to advanced treatment—including reverse osmosis, activated carbon adsorption and chemical oxidation— has become more commonplace. The even more widespread practice of obtaining potable water supplies from effluentimpacted surface waters is also growing as population pressures place further stress on freshwater supplies. Despite the increasing importance of potable water reuse and intensified attention being given to wastewater-derived trace organic contaminants, little effort has been directed at compounds that could cause taste and odor problems in drinking water. Previous research has demonstrated that potent odorants in lakes, rivers and water distribution systems 1 6 frequently result in consumer complaints. Odorous compounds in drinking water have often been attributed to algae or bacteria in the source water or fungi in biofilms on pipe surfaces (see Supporting Information (SI) Table S1). For example, geosmin and 2-methylisoborneol have been identified as the sources of earthy odors in numerous surface waters 6 8 while the musty odor of 2,4,6trichloroanisole has been detected in rivers and water distribution systems.3,4,7 Due to the potency of these odorants, sensitive r 2011 American Chemical Society

analytical methods with gas chromatography coupled with mass spectrometry or olfactometry are often needed to identify 9 11 and quantify these compounds in drinking water supplies.12,13 Municipal wastewater effluent also contains odorants but most previous studies on wastewater-derived odors have focused on nuisance air pollution produced by wastewater treatment processes (e.g., reduced sulfides in sludge thickening).14 16 These studies have been useful in the assessment of commonly applied control measures, such as biofilters, activated carbon, and chemical oxidants,17 but they have not provided insight into the potential for wastewater-derived odorants to compromise potable water supplies. Through experience, engineers have learned that it is often necessary to use activated carbon during drinking water treatment to minimize taste and odor issues in effluent-impacted sources but few attempts have been made to quantify the wastewater-derived compounds responsible for taste and odors. To assess the occurrence and fate of odorants in potable water reuse systems, analytical techniques developed by researchers Received: July 26, 2011 Accepted: September 27, 2011 Revised: September 19, 2011 Published: October 11, 2011 9347

dx.doi.org/10.1021/es202594z | Environ. Sci. Technol. 2011, 45, 9347–9355

Environmental Science & Technology studying taste and odors in drinking water and the food and beverage industry were applied to reclaimed water systems. Quantitative analysis of known potent odorants was accomplished by gas chromatography/mass spectrometry (GC/MS) while other compounds were analyzed by GC/MS-Olfactometry (GC/MS-Olf) and flavor profile analysis (FPA). To characterize the occurrence and fate of odorants, samples were collected at different stages of treatment from six full-scale advanced treatment plants. The removal of the most potent odorants was then evaluated in pilot- and bench-scale studies of different treatment processes under controlled conditions.

’ MATERIALS AND METHODS Chemical Standards. 2-Methylisoborneol, 2,3,4-trichloroanisole and 2,4,6-tribromoanisole were purchased from Dr. Ehrenstorfer Gmbh (Augsburg, Germany). 2-Bromophenol, 2,6dibromophenol, 2,4,6-tribromophenol, 2,4,6-trichlorophenol, 2,4,6-trichloroanisole, 2,3,6-trichloroanisole, β-ionone, and iodoform were purchased from Aldrich (St Quentin Fallavier, France) and Sigma-Aldrich (Saint Louis, MI). Deuterated surrogate standards (d5-geosmin and d5 2,4,6-trichloroanisole) were purchased from Cambridge Isotopes (Andover, MA). All other solvents and reagents were purchased at the highest level of purity available from Sigma-Aldrich and Merck KGaA (Darmstadt, Germany). Ultrapure deionized water (R g 18.2 MΩ-cm) was produced in-house with a Milli-Q purification system. Sample Collection. Samples were collected from six full-scale potable water reuse systems between September 2009 and February 2011 (SI Table S2). The plants had design capacities ranging from 60 to 200 ML d 1. Five rounds of bimonthly samples were collected at Plants A D while Plants E and F were sampled twice. All six advanced treatment plants received effluent from municipal wastewater treatment plants employing secondary biological treatment. In full-scale Plants A-D, incoming nitrified effluent was chlorinated with an initial concentration of approximately 2 mg/L Cl2 prior to microfiltration and reverse osmosis. The chlorine contact time between oxidant addition and the dechlorination point upstream of the reverse osmosis membrane was approximately 30 min. Plants E and F employed similar pretreatment trains except the wastewater entering the advanced treatment plants was not nitrified. After reverse osmosis, ultraviolet (UV) disinfection was employed at Plants A D at fluence values of approximately 80 mJ/cm2. UV/H2O2 was employed at Plants E and F with a fluence of approximately 500 mJ/cm2 and an initial H2O2 concentration of approximately 5 mg/L. In Plant A, ozonation (2 mg/L dose, 10 min contact time) was applied to a portion of the water after UV disinfection. Samples were also collected at a pilot plant treating denitrified municipal wastewater effluent with biological activated carbon filter (BAC) as detailed in Reungoat (2010).18 Pilot plant samples were collected during February and April 2010 before and after passage of the water through three different treatment columns: BAC without ozonation, ozonation followed by BAC, and ozonation followed by sand filtration. Before it was applied to the columns, wastewater effluent was ozonated (2 mg/L initial concentration) and subjected to coagulation, flocculation and aeration. For the two columns employing ozonation, an initial concentration of 5 mg/L O3 and a 15 min contact time was employed.

ARTICLE

All samples were collected in 1 L amber glass bottles with minimal headspace, shipped in iced coolers with overnight express service and extracted within 48 h of receipt. Samples were stored at 4 °C and were filtered (0.45 μm) prior to extraction. Field blanks, matrix spike samples and duplicates were included for analysis in all sampling rounds. Benchscale Experiments. Benchscale experiments were performed to assess the treatment efficacy of UV, UV/H2O2, chlorination, and chloramination. Secondary wastewater effluent or reverse osmosis permeate samples collected from Plants A and C were amended with target odorants at concentration approximately ten times higher than their lowest reported odor thresholds. Concentrated spiking solutions contained methanol because a number of commercial standards were only available in this solvent. Less than 50 μL of methanol was added to each 4 L sample prepared for the bench-scale experiments. Under these conditions, the steady-state concentrations of OH• are estimated to be reduced by methanol by approximately 90% and 20% in reverse osmosis permeate and secondary effluent, respectively (see SI). UV and UV/H2O2 treatments were assessed in a tubular stainless steel flow reactor (2.6 L, 15 cm o.d.) with helical internal baffles. Other than a 10-cm segment of Tygon tubing attached to the peristaltic pump, steel tubing was used to minimize losses of odorants via sorption. No loss of compounds was observed in control experiments without UV light. The reactor was equipped with two Puritec immersible low-pressure UV lamps (OSRAM, Munich, Germany) installed laterally in the center of the reactor. UV fluence was estimated from the average hydraulic residence time and photometer reading taken at quartz portholes located along the reactor. H2O2 was quantified in water flowing in and out of the reactor by KMnO4 titration.19 For chlorination and chloramination experiments, secondary effluent samples were dosed in 1-L amber glass bottles at initial concentrations of 5 and 15 mg/L as Cl2 typically applied in effluent chlorination with contact times up to 120 min. Free chlorine was added from a standardized stock solution of sodium hypochlorite. Premixed chloramine dosing solutions were made fresh daily by slowly adding sodium hypochlorite with NH4Cl at elevated pH.20 Free chlorine and monochloramine were determined using DPD colorimetric kits with a Hach DR 3800 spectrophotometer (Loveland, CO). Controls without free chlorine and chloramine indicated negligible losses of compounds. Experiments were carried out in triplicate. At the end of the experiments, excess oxidant was quenched by sodium bisulfite. Analytical Methods. Solid phase extraction of 0.45 μmfiltered samples was perfomed using a hydrophobic/hydrophilic polymeric resin (Oasis-HLB by Waters) conditioned with 5 mL methanol, 5 mL dichloromethane and 10 mL Milli-Q water. Sample pH values were adjusted to 4 5 with HCl to ensure that the weakly acidic bromophenols (pKa 7 9) and weakly basic methoxypyrazines (pKa ∼3) were present in their neutral forms. Samples were amended with 5 ng of d5-geosmin and d5 2,4,6trichloroanisole prior to extraction. Analytes were eluted from the cartridge with 10 mL dichloromethane. A sample preconcentration factor of 1000 yielded optimal instrument sensitivity while minimizing loss of the most volatile analytes. Sample extracts were concentrated to a final volume of 500 μL using a 40 °C circulating water bath and a gentle stream of ultrapure N2. Analysis was carried out with an Agilent 7890A series GC system with flow equally split between a mass spectrometer and 9348

dx.doi.org/10.1021/es202594z |Environ. Sci. Technol. 2011, 45, 9347–9355

Environmental Science & Technology an olfactory detector port (ODP). The 5975C series mass spectral detector (Agilent, Santa Clara, CA) was operated in selected ion monitoring (SIM) mode with chromatographic conditions as described in Zhang et al. (2006).12 Olfactometry was conducted with a Gerstel ODP3 (M€ulheim an der Ruhr, Germany). Sample from Plants E and F were analyzed using a Quattro micro GC triple quadrupole tandem mass spectrometer (Waters, Milford, MA) under similar chromatographic conditions. Olfactometry and flavor profile analysis (FPA) were also employed to identify other odorous compounds as described elsewhere.21 Briefly, olfactory analysis was carried out for 15 min beginning one minute after the solvent peak while, simultaneously, mass spectra were collected in full-scan mode between m/z 40 to 550. Each sample was analyzed by three members of a team of eight analysts who had been trained using reference standards and blind testing. Peak intensities of odorous compounds were classified on a scale of 0 to 4, with 4 being the strongest odor intensity. Only peaks eliciting a response of 3 (moderate intensity) or greater in 75% of the secondary effluent samples were evaluated further. Odor descriptors were categorized according to the wastewater odor wheel.22 Compounds associated with the most frequently detected odors were identified using several tools. Mass spectra were compared with the NIST mass spectral library (Agilent, Santa Clara, CA). Odor descriptions and retention times also were compared with data for compounds reported in peer-reviewed publications and public databases. Finally, compounds identified by these screening methods were compared with mass spectra, reference times and olfactometry data obtained from reference standards. Whole sample odor was assessed by sensory panels taken from the eight trained analysts using the flavor profile analysis method described in Standard Method 2170B.23

’ RESULTS AND DISCUSSION Odorous Compounds in Municipal Wastewater Effluent.

Twelve of the 15 target odorants were detected at least once in secondary effluent at concentrations up to approximately 100 ng/L (SI Table S3). The median concentrations of 2-methylisoborneol (2MIB, 11 ng/L), geosmin (27 ng/L), 2,6-dibromophenol (26DBP, 2.8 ng/L) and 2,4,6-trichloroanisole (246TCA, 9.5 ng/L) in secondary effluent were between 2 and 100 times higher than their respective odor thresholds. Another notable odorant, 2,4,6-tribromoanisole (246TBA) was detected in 40% of the secondary effluent samples at concentrations up to 6.6 ng/L. To express the concentration of odorants relative to their odor intensity, the measured concentrations were divided by the lowest reported odor thresholds (SI Table S1). This ratio, referred to as the relative odor intensity, indicates that the compounds of greatest concern detected in secondary effluent were 2,4,6-trichloroanisole and geosmin (Figure 1). The characteristic earthy and musty odors of these compounds were repeatedly detected during flavor profile analysis of secondary effluent. 2,4,6-trichloroanisole and geosmin were detected during olfactometry as strong odors—consistently scoring between 3 (moderate) and 4 (strong) during olfactometry runs—at retention times corresponding to those observed for authentic standards. The relative concentrations of the dominant target odorants in secondary effluent exhibited considerable intraplant variability

ARTICLE

Figure 1. Relative odor intensity (ROI) of common odor compounds detected in secondary effluent from municipal wastewater treatment plants.

Figure 2. Intraplant variability of common odor compounds in secondary effluent. Standard deviation was not calculated for locations E, F, and G because only two rounds of sampling were performed.

(Figure 2). 2,4,6-trichloroanisole was the dominant odorant at Plants A, B, F, and G while geosmin contributed significantly to the overall odor at Plants B, C, and D. Geosmin was the dominant odorant at Plant E, which was the only treatment plant employing a trickling filter. The intraplant variability may have been influenced by precursor concentrations in the raw sewage or by the microbial community in the biological treatment systems. Primary effluent samples collected between November 2009 and June 2010 indicated that biological wastewater treatment was a potential source for geosmin and 2,4,6-trichloroanisole (SI Table S3). In surface water supplies, geosmin is produced by a wide variety of microbes which also are commonly found in activated sludge, including cyanobacteria, actinomycetes,7 actinobacteria,24 and anabaena.25 Odors attributed to 2-methylisoborneol and geosmin have been reported in effluent from activated sludge plants treating wastes from pulp mills.2 Biological wastewater treatment was the main source of 2,4,6trichloroanisole. While primary effluent samples rarely contained the odorant (median concentration 104 M 1s 1].38 Geosmin, 2-methylisoborneol, and haloanisoles are transformed during ozonation mostly by OH•, making the process less effective in wastewater effluent where more OH• scavengers are present. Ozonation at Plant A (initial O3 concentration 2 mg/L, contact time 10 min) was applied on reverse osmosis permeate

Figure 3. UV treatment of odor compounds observed during benchscale experiment of spiked secondary effluent and reverse osmosis permeate at fluence 0 2000 mJ/cm2. Initial concentration Co = 50 ng/L. 9351

dx.doi.org/10.1021/es202594z |Environ. Sci. Technol. 2011, 45, 9347–9355

Environmental Science & Technology

ARTICLE

Figure 4. UV/H2O2 treatment of odor compounds observed during benchscale experiment of spiked secondary effluent and reverse osmosis permeate at UV fluence 0 2000 mJ/cm2 and 10 mg/L H2O2 dose.

containing geosmin, 2,4,6-trichloroanisole and 2,6-dibromophenol at concentrations up to 50 times the respective odor thresholds. Under these conditions, ozonation decreased the concentrations of odorants to levels below their GC-MS detection limits. The strong earthy/musty odors present in the permeate (intensity >3) were not reported by panelists in flavor profile analysis or GC-Olfactometry with the exception of 2-pyrrolidinone, which was present at a weak intensity (∼1). At the biofilter pilot plant (Plant G), preozonation (5 mg/L, 15 min) was applied to wastewater effluent that contained geosmin, 2-methylisoborneol, 2,4,6-trichloroanisole, 2,3,4-trichloroanisole, 2,4,6-tribromoanisole at concentrations up to 50 times higher than the respective odor thresholds. Under these conditions, the concentration of 2-methylisoborneol decreased by between 60 and 90% and the haloanisole concentrations decreased by approximately 40%. The odors of geosmin, 2-pyrrolidinone and lactones were still detected by the panelists during GC-olfactometry of the ozonated effluent. Fate of Odorous Compounds during Activated Carbon Treatment. Historically, granular and powder activated carbon have been used to eliminate taste and odor caused by geosmin and 2-methylisoborneol.42,43 Other odorous compounds identified in wastewater effluent generally have a similar or higher affinity for activated carbon to geosmin and 2-methylisoborneol, indicating a high potential for removal. BAC has previously been shown to remove a variety of pharmaceuticals with log Kow values above 318 with better removal observed for more readily biodegradable and hydrophobic compounds. At the BAC pilot treatment system, 2,4,6-trichloroanisole, 2-methylisoborneol and geosmin as well as 10 other odorants were detected by olfactometry in the column influent. Without ozone pretreatment (SI Table S6), BAC treatment reduced the concentration of geosmin (51 and 61%) and 2-methylisoborneol (60 and 53%). It also reduced the concentration of 2,4,6trichloroanisole from about 4 ng/L to below the method detection limit (95%) was observed. No significant odor was detected during GC-olfactometry of samples from the outlet of biofilter pretreated with ozone, while at least eight odorants (including 2-pyrrolidone, methylnaphthalene isomers, and alkyl acids) were still detected at weak intensity in BAC samples without ozonation.

Dilution and Volatilization of Odorous Compounds in Surface Waters. In many situations, secondary effluent is

discharged to surface waters that serve as potable water supplies. As indicated previously, at least 15 odorants are typically present in secondary effluent at concentrations above their odor thresholds. The dilution of secondary effluent with water free from odorous compounds could eliminate aesthetic problems downstream of the outfalls. For example, effluent containing 10 ng/L of 2,4,6-trichloroanisole (i.e., the median concentration detected in effluent samples) would need to be diluted until effluent accounted for less than 1% of the total flow before the concenontration of the compound in the source water would no longer exceed the odor threshold. Application of flavor profile analysis to diluted wastewater effluent from Plants A and C (11 and 27 ng/L 2,4,6-trichloroanisole, respectively) indicated that a weak earthy/musty odor could still be detected by panelists when effluent accounted for 3% of the sample volume. At this dilution factor, odors of 2,4,6-trichloroanisole and geosmin (intensity 2.0 3.0) were confirmed by GC-Olfactometry. In addition, weak odors at retention times corresponding to those of 2-pyrrolidinone and vanillin were detected in the diluted effluents. Assuming little removal downstream of treatment plant, the odorous compounds could pose aesthetic problems for many downstream water supplies. Volatilization of odorants during storage or downstream transport could reduce the concentrations of odorous compounds. Previous research has yielded predictive models for the fate of volatile organic compounds in rivers based on a twofilm model with or without turbulence.44 Similarly, a fugacitybased model has been developed to predict volatilization potential in reservoirs.45 In both models, the Henry’s Law constant (KH) is an indicator of volatilization potential (SI Table S5) with actual volatilization rates dependent on site-specific characteristics such as water and wind velocity, depth, temperature,44 hydraulic residence time, surface area and mixing.45 Assuming conditions typically encountered in rivers, compounds with KH > 101 Pa m3/mol are predicted to exhibit a decrease of approximately an order of magnitude during 25 km flow downstream in a river and a decrease of approximately 2 orders of magnitude during an 18-month storage period in a reservoir. Among the odorous compounds detected in wastewater effluent, the haloanisoles, crotyl mercaptan and 2,6-dibromophenol have the 9352

dx.doi.org/10.1021/es202594z |Environ. Sci. Technol. 2011, 45, 9347–9355

Environmental Science & Technology

ARTICLE

Table 1. Key GC-MS/Olfactometry Odor Peaks Detected in RO-Ozone, RO, UV/Peroxide and Ozone-BAC Treatment Trains

potential to undergo substantial losses through volatilization in surface waters (i.e., KH > 101 Pa m3/mol). However, 2-MIB, geosmin, 2-pyrrolidinone, vanillin, and hydroxyvanillin are unlikely to be substantially affected by volatilization. There are other potential mechanisms through which odorants might be attenuated in surface waters. For example, biotransformation and phototransformation of pharmaceuticals occurred with half-lives of approximately one week in the Trinity River.46 Limited information is available on the potential for odorants identified in wastewater effluent to undergo attenuation under similar mechanisms. For geosmin and 2-methylisoborneol, microbial transformation has been observed in reservoirs.8 Additional research is needed to make accurate predictions of

the potential for these compounds to undergo biotransformation and photolysis in surface waters.

’ IMPLICATIONS A suite of odorous compounds are present in wastewater effluent at concentrations well above their odor thresholds. While the presence of these compounds does not imply a health risk, their presence has the potential to pose challenges to potable water supplies. For surface waters that receive municipal wastewater effluent, substantial dilution coupled with long residence times are needed to reduce odorant concentrations to values below odor thresholds. Volatilization during storage or transit 9353

dx.doi.org/10.1021/es202594z |Environ. Sci. Technol. 2011, 45, 9347–9355

Environmental Science & Technology might be sufficient to remove haloanisoles but it will not remove less volatile odorants, such as geosmin, 2-pyrrolidone and hydroxyvanillin. To remove these odorants, downstream drinking water treatment plants may need to use activated carbon or an advanced oxidation process. Advanced treatment of secondary effluent with multiple treatment barriers—as practiced in most potable water reuse systems—is needed to reduce the concentrations of odorants to values below threshold levels. Reverse osmosis is effective in removing odorants but several may be present at concentrations above their odor thresholds in the permeate. Ozonation or UV/H2O2 can eliminate these odors from the permeate. Advanced oxidation processes (i.e., UV/H2O2) or ozonation coupled with biological activated carbon also may provide a means for removing odorous compounds even in systems that do not employ reverse osmosis. A summary of data from two full-scale advanced wastewater treatment plants and one pilot plant (Table 1) illustrates the ways in which GC-MS/Olfactometry of effluent coupled with GC/MS quantification of specific contaminants can be used to study the fate of odorants. As indicate by the olfactometry intensity scores, 2,4,6-trichloroanisole (RT = 17.0 min) and geosmin (RT = 18.5 min) are among the most persistent odorants in advanced treatment systems and can be used as indicators47 of other odors thereby avoiding the need for labor-intensive olfactometry studies. After advanced treatment is completed, any remaining compounds can be identified and quantified using the approach described above.

’ ASSOCIATED CONTENT

bS

Supporting Information. Additional figures, tables, calculations and method details are provided. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: (510) 643-0256; e-mail: [email protected].

’ ACKNOWLEDGMENT We thank the PUB, Singapore’s National Water Agency for financial support. We are also grateful to PUB staff—especially Mr. Qinglin Lu and Ms. Xiaoqing Qian—for their sampling, quantitative and sensory analysis assistance. We thank Dr. Julien Reungoat at University of Queensland (Australia), Mr. Patrick Versluis at Orange County Water District and Mr. Gregg Oelker at West Basin Water Management District for field sample collection. ’ REFERENCES (1) Izaguirre, G.; Hwang, J.; Krasner, S. W. Geosmin and 2-methylisoborneol from cyanobacteria in three water supply systems. App. Environ. Microbiol. 1982, 43, 708–714. (2) Brownlee, B. G.; MacInnis, G. A.; Noton, L. R. Chlorinated anisoles and veratroles in a Canadian river receiving bleached kraft pulp mill effluent: Identification, distribution and olfactory evaluation. Environ. Sci. Technol. 1993, 27, 2450–2455. (3) Karlsson, S.; Kaugare, S.; Grimvall, A.; Boren, H.; Savenhed, R. Formation of 2,4,6-trichlorophenol and 2,4,6-trichloroanisole during treatment and distribution of drinking water. Water Sci. Technol. 1995, 31, 99–103.

ARTICLE

(4) Piriou, P.; Malleret, L.; Bruchet, A.; Kiene, L. Trichloroanisole kinetics and musty tastes in drinking water distribution systems. Water Sci. Technol.: Water Supply 2001, 1, 11–18. (5) Watson, S. Aquatic taste and odor: A primary signal of drinking water integrity. J. Toxicol. Environ. Health, Part A 2004, 67, 1779–1795. (6) Peter, A.; K€ oster, O.; Schildknecht, A.; Von Gunten, U. Occurrence of dissolved and particle-bound taste and odor compounds in Swiss lake waters. Water Res. 2009, 43, 2191–2200. (7) Jensen, S. E.; Anders, C. L.; Goatcher, L. J.; Perley, T.; Kenefick, S.; Hrudey, S. E. Actinomycetes as a factor in odor problems affecting drinking water from the North Saskatchewan River. Water Res. 1994, 28, 1393–1401. (8) Westerhoff, P.; Rodriguez-Hernandez, M.; Baker, L.; Sommerfeld, M. Seasonal occurrence and degradation of 2-methylisoborneol in water supply reservoirs. Water Res. 2005, 39, 4899–4912. (9) Young, W. H.; Horth, H.; Crane, R.; Ogden, T.; Arnott, M. Taste and odour threshold concentrations of potential potable water contaminants. Water Res. 1996, 30, 331–340. (10) Whitfield, F. B. Chemistry of off-flavours in marine organisms. Water Sci. Technol. 1988, 20, 63–74. (11) Díaz, A.; Ventura, F.; Galceran, M. T. Determination of odorous mixed chloro-bromoanisoles in water by solid-phase microextraction and gas chromatography mass detection. J Chromatogr, A 2005, 1064, 97–106. (12) Zhang, L.; Hu, R.; Yang, Z. Routine analysis of off-flavor compounds in water at sub-part-per-trillion level by large-volume injection GC/MS with programmable temperature vaporizing inlet. Water Res. 2006, 40, 699–709. (13) Salemi, A.; Lacorte, S.; Bagheri, H.; Barcel o, D. Automated trace determination of earthy-musty odorous compounds in water samples by on-line purge-and-trap gas chromatography mass spectrometry. J. Chromatogr., A 2006, 1136, 170–175. (14) Lambert, D. D.; Beaman, A. L.; Winter, P. Olfactometric characterisation of sludge odours. Water Sci. Technol. 2000, 41, 49–55. (15) Gostelow, P.; Parsons, S. A.; Stuetz, R. M. Odour measurements for sewage treatment works. Water Res. 2001, 35, 579–597. (16) Kim, K. H.; Park, S. Y. A comparative analysis of malodor samples between direct (olfactometry) and indirect (instrumental) methods. Atmos. Environ. 2008, 42, 5061–5070. (17) Harshman, V.; Barnette, T. Wastewater Odor Control: An Evaluation of Technologies. Water Eng. Manage. 2000, 147, 34–46. (18) Reungoat, J.; Macova, M.; Escher, B. I.; Carswell, S.; Mueller, J. F.; Keller, J. Removal of micropollutants and reduction of biological activity in a full-scale reclamation plant using ozonation and activated carbon filtration. Water Res. 2010, 44, 625–637. (19) Klassen, N.; Marchington, D; McGowan, H. H2O2 determination by the I3-method and by KMnO4 titration. Anal. Chem. 1994, 66, 2921–2925. (20) Mitch, W. A.; Sedlak, D. L. Formation of N-nitrosodimethylamine (NDMA) from dimethylamine during chlorination. Environ. Sci. Technol. 2002, 36, 588–595. (21) Agus, E.; Sedlak, D. L. Application of gas chromatography with mass spectrometer and olfactory detectors (GC-MS/Olfactometry) to identify odor compounds in municipal wastewater effluent and advanced water treatment. In Preparation. (22) Burlingame, G. A.; Suffet, I. H.; Khiari, D.; Bruchet, A. L. Development of an odor wheel classification scheme for wastewater. Water Sci. Technol. 2004, 49, 201–209. (23) APHA, WEF. Standard Methods for the Examination of Water and Wastewater, 19th ed.; American Public Health Association: Washington, DC, 1997 (24) Klausen, C.; Nicolaisen, M. H.; Strobel, B. W.; Warnecke, F.; Nielsen, J. L.; Jørgensen, N. O. Abundance of actinobacteria and production of geosmin and 2-methylisoborneol in Danish streams and fish ponds. FEMS Microbiol. Ecol. 2005, 52, 265–278. (25) Saadoun, I. M. K.; Schrader, K. K.; Blevins, W. T. Environmental and nutritional factors affecting geosmin synthesis by Anabaena SP. Water Res. 2001, 35, 1209–1218. 9354

dx.doi.org/10.1021/es202594z |Environ. Sci. Technol. 2011, 45, 9347–9355

Environmental Science & Technology

ARTICLE

(26) Bux, F.; Kasan, H. C. A microbiological survey of 10 activatedsludge plants. Water SA 1994, 20, 61–72. (27) Sun, Y. X.; Wu, Q. Y.; Hu, H. Y.; Tian, J. Effect of bromide on the formation of disinfection by-products during wastewater chlorination. Water Res. 2009, 43, 2391–2398. (28) Hua, G. H.; Reckhow, D. A.; Kim, J. S. Effect of bromide and iodide ions on the formation and speciation of disinfection byproducts during chlorination. Environ. Sci. Technol. 2006, 40, 3050–3056. (29) Schor, E. Hydrocarbons in cereal stoke new debate over food safety. In New York Times. Published: July 13, 2010. (30) Trenholm, R. A.; Vanderford, B. J.; Drewes, J. E.; Snyder, S. A. Determination of household chemicals using gas chromatography and liquid chromatography with tandem mass spectrometry. J Chromatogr., A 2008, 1190, 253–262. (31) Bellona, C.; Drewes, J. E.; Xu, P.; Amy, G. Factors affecting the rejection of organic solutes during NF/RO treatment—A literature review. Water Res. 2004, 38, 2795–2809. (32) Schafer, A. I.; Nghiem, L. D.; Waite, T. D. Removal of the natural hormone estrone from aqueous solutions using nanofiltration and reverse osmosis. Environ. Sci. Technol. 2003, 37, 182–188. (33) Agus, E.; Sedlak, D. L. Formation and fate of chlorination byproducts in reverse osmosis desalination systems. Water Res. 2010, 44, 1616–1626. (34) West Basin Municipal Water District. Investigation of N-itrosodimethylamine (NDMA) Fate and Transport; WateReuse Foundation: Alexandria, VA2006 (35) Plumlee, M. H.; Lopez-Mesas, M.; Heidlberger, A.; Ishida, K. P.; Reinhard, M. N-nitrosodimethylamine (NDMA) removal by reverse osmosis and UV treatment and analysis via LC-MS/MS. Wat. Res. 2008, 42, 347–355. (36) Rosenfeldt, E.; Melcher, B.; Linden, K. UV and UV/H2O2 treatment of methylisoborneol (MIB) and geosmin in water. J. Water Supply: Res. Technol. 2005, 54, 423–434. (37) Pei, P.; Westerhoff, P.; Nalinakumari, B. Kinetics of MIB and geosmin during ozonation. Ozone: Sci. Eng. 2006, 28, 277–286. (38) Peter, A.; Von Gunten, U. Oxidation kinetics of selected taste and odor compounds during ozonation of drinking water. Environ. Sci. Technol. 2007, 41, 626–631. (39) Deborde, M.; Von Gunten, U. Reactions of chlorine with inorganic and organic compounds during water treatment—Kinetics and mechanisms: A critical review. Water Res. 2008, 42, 13–51. (40) Pereira, V. J.; Weinberg, H. S.; Linden, K. G.; Singer, P. C. UV Degradation kinetics and modeling of pharmaceutical compounds in laboratory grade and surface water via direct and indirect photolysis at 254 nm. Environ. Sci. Technol. 2007, 41, 1682–1688. (41) Benitez, F. J.; Beltran-Heredia, J.; Acero, J. L.; Rubio, F. J. Rate constants for the reactions of ozone with chlorophenols in aqueous solutions. J. Hazard Mater. 2000, 79, 271–285. (42) Chen, G.; Dussert, B.; Suffet, I. Evaluation of granular activated carbons for removal of methylisoborneol to below odor threshold concentration in drinking water. Water Res. 1997, 31, 1155–1163. (43) Cook, D.; Newcombe, G.; Sztajnbok, P. The application of powdered activated carbon for MIB and geosmin removal: Predicting PAC doses in four raw waters. Water Res. 2001, 35, 1325–1333. (44) Rathbun, R. E. Transport, Behavior and Fate of Volatile Organic Compounds in Streams, Professional Paper 1589; United States Geological Survey: Washington, DC, 1998 (45) Hawker, D. W.; Cumming, J. L.; Neale, P. A.; Bartkow, M. E.; Escher, B. I. A screening level fate model of organic contaminants from advanced water treatment in a potable water supply reservoir. Water Res. 2011, 45, 768–780. (46) Fono, L. J.; Kolodziej, E. P.; Sedlak, D. L. Attenuation of wastewater-derived contaminants in an effluent-dominated river. Environ. Sci. Technol. 2006, 40, 7257–7262. (47) Dickenson, E. R. V.; Drewes, J. E.; Sedlak, D. L.; Wert, E. C.; Snyder, S. A. Applying surrogates and indicators to assess removal efficiency of trace organic chemicals during chemical oxidation of wastewaters. Environ. Sci. Technol. 2009, 43, 6242–6247. 9355

dx.doi.org/10.1021/es202594z |Environ. Sci. Technol. 2011, 45, 9347–9355

ARTICLE pubs.acs.org/est

Effects of Copper Nanoparticles Exposure in the Mussel Mytilus galloprovincialis T^ania Gomes,† Jose P. Pinheiro,‡ Ibon Cancio,§ Catarina G. Pereira,† Catia Cardoso,† and Maria Jo~ao Bebianno†,* †

CIMA, Faculty of Science and Technology, University of Algarve, Campus de Gambelas, 8005-139 Faro, Portugal CBME, Faculty of Science and Technology, University of Algarve, Campus de Gambelas, 8005-139 Faro, Portugal § Dept. Zoology & Animal Cell Biology, Scholl of Science and Technology, University of the Basque Country, E-48080 Bilbao, Spain ‡

ABSTRACT: CuO NPs are widely used in various industrial and commercial applications. However, little is known about their potential toxicity or fate in the environment. In this study the effects of copper nanoparticles were investigated in the gills of mussels Mytilus galloprovincialis, comparative to Cu2+. Mussels were exposed to 10 μgCu 3 L
1 of CuO NPs and Cu2+ for 15 days, and biomarkers of oxidative stress, metal exposure and neurotoxicity evaluated. Results show that mussels accumulated copper in gills and responded differently to CuO NPs and Cu2+, suggesting distinct modes of action. CuO NPs induced oxidative stress in mussels by overwhelming gills antioxidant defense system, while for Cu2+ enzymatic activities remained unchanged or increased. CuO NPs and Cu2+ originated lipid peroxidation in mussels despite different antioxidant efficiency. Moreover, an induction of MT was detected throughout the exposure in mussels exposed to nano and ionic Cu, more evident in CuO NPs exposure. Neurotoxic effects reflected as AChE inhibition were only detected at the end of the exposure period for both forms of copper. In overall, these findings show that filter-feeding organisms are significant targets for nanoparticle exposure and need to be included when evaluating the overall toxicological impact of nanoparticles in the aquatic environment.

’ INTRODUCTION Nanotechnology is a rapid growing field that comprises the research and development of particles 0.05). The induction of GPX after a week of NPs exposure

ARTICLE

Figure 3. Metallothionein concentrations (A), inhibition of acetycholinesterase activity (B) and lipid peroxidation (C) in gills of mussels M. galloprovincialis from controls and exposed to CuO NPs and Cu2+ for 15 days (average ( Std). Capital and lower letters represent statistical differences between treatments in each exposure day and for each treatment during the exposure duration, respectively (p < 0.05). Asterisks represent statistical differences between control and exposed mussels (p < 0.05).

(15.8 ( 3.1 to 21.3 ( 1.7 nmol 3 min 3 mg
1prot) suggests the detoxification of hydroperoxides possibly associated with increased levels of hydroxyl radicals originated by CuO NPs, whereas at the beginning SOD and CAT levels may have been sufficient to counteract the overproduction of ROS. The SOD and CAT similar antioxidant efficiencies were supported by the PCA analysis (Figure 4A) that shows a significant correlation in the first week of exposure. After two weeks, both SOD and CAT activities decreased (38 and 33% of inhibition, p < 0.05) in mussels exposed to CuO NPs, whereas GPX continued to increase. These inhibitory effects suggest an overproduction of ROS that could have led to the degeneration of the enzymes. These ROS can be available to react with Cu2+ from CuO NPs dissolution, leading to the formation of hydroxyl radicals generated from H2O2 under Cu+ exposure through the Fenton and Haber Weiss reactions, possibly leading to SOD and CAT inactivation.5,6 These data are in line with recent observations that show that CuO NPs cytotoxicity is mediated by oxidative stress, altering the antioxidant capacity of cells against ROS. In human lung epithelial cells, CuO NPs (80 μg 3 cm
2, 4 h, 30 nm) blocked the 9359

dx.doi.org/10.1021/es200955s |Environ. Sci. Technol. 2011, 45, 9356–9362

Environmental Science & Technology

Figure 4.
Principal component analysis (PCA) of copper accumulation and the battery of biomarkers in gills of mussels M. galloprovincialis from controls and exposed to CuO NPs and Cu2+ for 15 days. A, PC1 vs PC2; B, PC1 vs PC3.

antioxidant defenses by inhibiting CAT and GR activities and increasing GPX or SOD and CAT activities after exposure to 10, 25, and 50 μg 3 mL
1 for 24 h (52.5 ( 10.2 nm).11,28,32 In bivalves, the only existing data on antioxidant efficiency are of Cu2+ exposure. Mussels exposed to Cu2+ showed different antioxidant responses (Figure 2) with the enzymatic activities unchanged or increased (Figure 2). SOD activity was activated during the whole experiment (171% increase by day 15) resulting in the formation of superoxide radicals. CAT activity only increased after 3 days of exposure (36%) and remained unchanged from day 7 until the end of the experiment, at levels similar to controls (p > 0.05). As mentioned above, this result can be associated with the involvement of Cu in Fenton and Haber Weiss reactions, leaving no substrate available for CAT activation, or to the induction of other components of the antioxidant defense system.5,6 Like for CAT, GPX activity was induced in the first 3 days of exposure (25.0 ( 1.7 nmol 3 min 3 mg
1prot, p < 0.05) remaining unchanged until the end of the experiment, always higher than that in control. This increase in GPX activity suggests a further detoxification of ROS combined with the action of MT; either by ROS scavenging (day 7) or Cu detoxification (day 15), justifying CAT unaltered activities. The PCA analysis shows a clear association between GPX activity and Cu2+-exposed mussels, validating the enhancement of this enzyme activity to neutralize ROS (Figure 4A). Similar results were detected in mussels exposed to 60 μgCu 3 L
17 for 3 weeks and in the clam R. decussatus exposed to 0.5 and 2.5 μgCu 3 L
1 Cu for 3 days.5

ARTICLE

Metallothioneins are low-molecular weight cysteine-rich proteins induced by metals that can also act as an oxygen species scavengers, participating in antioxidant processes protecting cells from oxidative stress.19,20,31 Although information on MT behavior upon exposure to CuO NPs is nonexistent, the role of MT in ionic/soluble Cu detoxification mechanisms is well understood in bivalves, either by controlling its intracellular availability or by detoxifying excessive metal concentrations.4,5,20,31,33 In mussels exposed to CuO NPs, MT increased linearly with time of exposure, with an induction rate of 0.3 mg 3 g
1prot 3 d
1 (r = 0.99, p < 0.05), reflecting not only the role of this protein in Cu homeostasis and detoxification (Figure 3A), but also a possible involvement in gills antioxidant defense system that can explain the absence of SOD and CAT responses (day 15). Only two studies addressed the role of MT in bivalve species: in C. virginica exposed to silver nanoparticles (16 μg 3 L
1-1.6 ng 3 L
1, 15 ( 6 nm) an increase in MT expression was associated with silver metabolism or to the increase of oxyradicals and in C. fluminea exposed to gold nanoparticles (1.6  103
1.6  105 Au NP/cell, 10 nm) to protect cells against gold-induced oxidative stress.34,35 In mussels exposed to Cu2+, MT levels also increased in the first week of exposure with a lower induction rate (0.2 mg 3 g
1prot 3 d
1, r = 0.99, p < 0.05) when compared to CuO NPs (Figure 3A), denoting its importance in Cu metabolism, as also seen by the close association between Cu concentrations and MT in the PCA (Figure 4). Contrarily to the response for CuO NPs, MT decreased in the gills of mussels exposed to Cu2+ at the end of the experiment (6.7 mg 3 g
1prot), suggesting a role of MT in copper detoxification, which is in agreement with the copper accumulation results in mussel gills (Figure 1C). Cu can bind to MT to form insoluble Cu
MT complexes that precipitate into lysosomes and are eliminated by exocytosis.3,20,31 Similar results were detected in R. decussatus31 and Crassostrea gigas20 exposed to 50 μgCu 3 L
1 and 0.5
5 μgCu 3 L
1, respectively. Acetylcholinesterase is a biomarker of exposure to organophosphorus pesticides that can also be inhibited by a diverse range of metals, including copper.4,5,33 A dose-dependent decrease of this enzyme after Cu2+exposure is well established in bivalve species, as in R. decussatus (75 μgCu 3 L
1, 5 days)5 and mussels (40 μg 3 L
1 and 60 μg 3 L
1, 1 and 3 weeks).33,4 In this study, inhibition of AChE was observed in CuO NPs and Cu2+ exposed mussels (Figure 3B) only at the end of the experiment, with a 34% and 53% inhibition, respectively (p < 0.05), also confirmed by the PCA (Figure 4). The high affinity of Cu to sulfur donor groups can cause AChE inhibition by binding to its thiol residues, as in MT.5 These results confirm the specificity of AChE response to Cu exposure, either in the nano or ionic form. The neurotoxic effects of nanoparticles in M. edulis exposed to 1 mg 3 L
1 Fe NPs (5
90 nm, 12 h) showed no significant differences in AChE activity.36 Nevertheless, one study showed that AChE has the potential to be used as a biomarker for CuO NPs (25 nm), because of its strong AChE inhibition (76%) and low median inhibitory concentration (4 mg 3 L
1).37 Significant variations of enzymatic activities exist between control and Cu-exposed mussels throughout the experimental period suggesting that gills responded differently to both forms of copper (Figure 2). The overall PCA analysis (Figure 4) indicates a clear separation between control and Cu-exposed mussels. Unexposed mussels, as well as those exposed to Cu2+ are closely associated at different times of exposure (day 3, 7, and 15) showing similar biomarker tendency. As for CuO NPs exposed mussels, a clear separation of the sampling periods 9360

dx.doi.org/10.1021/es200955s |Environ. Sci. Technol. 2011, 45, 9356–9362

Environmental Science & Technology occurred, suggesting a marked different behavior between mussel gills response with time of exposure. Failure of antioxidant defenses to counteract ROS produced by both forms of Cu either by being inhibited or overwhelmed can interrupt the balance between the antioxidant/prooxidant system in mussels leading to oxidative damage of biomolecules.4
6 One of the best known effects of excess Cu is the peroxidative damage to membrane lipids, triggered by the reaction of lipid radicals and oxygen to form peroxyl radicals that can alter membrane fluidity and permeability or attack other intracellular molecules.4
6 Despite different antioxidant efficiency, LPO increased linearly with time in mussels exposed to CuO NPs and Cu2+ (Figure 3C), with induction rates of 36.8 nmol 3 g
1 prot 3 d
1 (r = 0.99; p < 0.05) and 49.7 nmol 3 g
1prot 3 d
1 (r = 0.97, p < 0.05), respectively. In the first three days of CuO NPs exposure, SOD and CAT activities proved to be antioxidant efficient and prevent deleterious effects in lipids of cellular membranes, confirmed by the relative proximity of these mussels to the control group in the PCA analysis (Figure 4A). In the remaining period, CuO NPs seems to continuously increase ROS production activating the combined action of antioxidant defenses (SOD, CAT, GPX, and MT) until a point where the antioxidant capacity was overwhelmed causing SOD and CAT inactivation and a continuous MT and GPX increase. Although GPX and MT can remove most of the ROS by increasing its activities, they cannot compete with hydroxyl radicals’ generation via the Fenton reaction thereby causing an increase in LPO levels. In mussels exposed to Cu2+, antioxidant enzymes were activated during the whole exposure period (except CAT) along with an increase in MT levels leading to a detoxification process by the end of the exposure, nevertheless, not enough to prevent LPO. These results are in agreement with the PCA that shows a clear association between copper concentrations in gills and LPO levels, as well as with MT and GPX (Figure 4). In human cells and E. coli exposed to CuO NPs (30
50 nm) their toxicity was related to oxidative stress, mediated by lipid peroxidation, oxidative lesions and increase of intracellular ROS11,13,15,28,32 Evidence that LPO occurs after Cu exposure was also observed in several bivalve species, as clams and mussels.4
6 Altogether, our results support the conclusion that oxidative stress is a significant mechanism of toxicity for CuO NPs7,9,11,15,32,38 and that its mode of action appears distinct from Cu2+. In other aquatic organisms (V. fisheri, D. magna, T. platyurus, P. subcapitata, T. thermophila) CuO NPs (∼30 nm) showed a higher toxicity when compared to its ionic/soluble form, associated with Cu ions dissolution.9,10,12 Nevertheless, the dissolution of Cu ions do not fully explain the toxicity of CuO NPs in zebrafish,7,14 human cell cultures,11,29,32 or daphnids15 exposed to particles with similar size (30
50 nm), where other mechanisms derived from the particle effect had to be considered (e.g., oxidative stress due to ROS formation). In our study, a combination of the particle effect and ions dissolution can account for the differences in the toxic effects exerted by CuO NPs along the exposure period. Mussel gills can be taking up dissolved Cu released from the particles combined with a cellular uptake of nanoparticles aggregates. CuO NPs can pass the cellular membrane, enter inside the cell, dissolve rapidly and release high concentrations of ions sufficient to disrupt Cu homeostasis and generate radicals.1,14,28,38 This NPs mechanism of toxicity named “Trojan horse-type mechanism” was identified in cell cultures.38,39 The increasing copper concentrations in mussel gills can be indicative of an increasing rate of exposure that leads to a continuous release of Cu from the NPs. The reaction time of gills cells is slower than the particle dissolution and uptake leading to enzymatic

ARTICLE

breakdown and to a continuous increase in MT levels, whereas in mussels exposed to Cu2+, this metal is eliminated more rapidly via MT detoxification pathway. Another fraction of the CuO NPs can be taken up by endocytosis and their toxicological response controlled by surface processes (ROS, adsorption).1,2,38 The presence of CuO NPs aggregates in suspension (as seen by DLS) facilitate a continuous source of NPs that can either be dissolved or incorporated, leading to a continuous ROS generation (intra and/or extracellular), that increases with time of exposure. A correlation between formation of larger aggregates and biomarker responses with increasing time of exposure was suggested in M. galloprovincialis exposed to nano carbon black, C60 fullerenes, nano-TiO2 and nano-SiO2.40 A more efficient and rapid capture and ingestion of NPs in aggregated form was also observed in mussels and oysters exposed to polystyrene NPs when compared to those in suspension.29 As for M. edulis, NPs from glass wool and Fe are taken up by gills epithelial cells as pathways of uptake by diffusion or by endocytosis, independently of the size of the aggregates.37,41 Aggregation has a crucial role in nanoparticles toxicity, and the cumulative effects of the dissociation of metal ions, size and surface-area properties of these particles cannot be discarded and need further clarification in CuO NPs mechanisms.1,2,29,35,40 Despite the information given by acute experiments, they do not provide complete information about the interactions of nanomaterials with classical test species and there is a need to direct research toward invertebrate tests using long-term exposure to better understand NPs toxicity mechanisms.2,9,13,15 As for CuO NPs, most of the data available concerns acute toxicity across a wide spectrum of aquatic species,7
10,12
16 and this study is one of the first to address long-term effects of these NPs in this species. Overall our results show that mussels represent a target for environmental exposure to nanoparticles where exposure duration may be a contributing factor in NPs mediated toxicity. In summary, long-term exposure to CuO NPs cause oxidative stress in gills of mussels as evidenced by the breakdown of the antioxidant defense system and lipid peroxidation, as well as acetylcholinesterase inhibition and metallothionein induction. Nevertheless the underlying mechanisms associated with biomarkers responses are still uncertain, and the observed oxidative stress may due to an association between the nanoparticle effect and the dissociation of copper ions from the nanoparticles. Future research is required to understand the mechanisms of CuO NPs toxicity in aquatic organisms, where the uptake and accumulation of CuO NPs in other mussel tissues should be considered, as well as the importance of bioavailability and particle aggregation for long periods of time.

’ AUTHOR INFORMATION Corresponding Author

*Phone: (+351) 289800100; fax: (+351) 289800069; e-mail: [email protected].

’ ACKNOWLEDGMENT This research was supported by a Foundation of Science and Technology PhD Grant (SFRH/BD/41605/2007). ’ REFERENCES (1) Moore, M. N. Do nanoparticles present ecotoxicological risks for the health of the aquatic environment? Environ Int. 2006, 32, 967–976. 9361

dx.doi.org/10.1021/es200955s |Environ. Sci. Technol. 2011, 45, 9356–9362

Environmental Science & Technology (2) Baun, A.; Hartmann, N. B.; Grieger, K.; Kusk, K. O. Ecotoxicity of engineered nanoparticles to aquatic invertebrates: A brief review and recommendations for future toxicity testing. Ecotoxicol. 2008, 17, 387–395. (3) da Silva, J. J. R. F.; Williams, R. J. P. The Biological Chemistry of the Elements: The Inorganic Chemistry of Live, 2nd ed.; Oxford University Press: New York, 2001. (4) Regoli, F.; Principato, G. Glutathione, glutathione-dependent and antioxidant enzymes in mussel, Mytilus galloprovincialis, exposed to metals under field and laboratory conditions: Implications for the use of biochemical biomarkers. Aquat. Toxicol. 1995, 31, 143–164. (5) Bebianno, M. J.; Geret, F.; Hoarau, P.; Serafim, M. A.; Coelho, M. R.; Gnassia-Barelli, M.; Romeo, M. Biomarkers in Ruditapes decussatus: A potential biondicator species. Biomarkers 2004, 9 (4
5), 305–330. (6) Maria, V. L.; M. J. Bebianno, M. J. Antioxidant and lipid peroxidation responses in Mytilus galloprovincialis exposed to mixtures of benzo(a)pyrene and copper. Comp. Biochem. Pharmacol. C 2011, 154 (1), 56–63. (7) Griffitt, R. J.; Weil, R.; Hyndman, K. A.; Denslow, N. D.; Powers, K.; Taylor, D.; Barber, S. D. Exposure to copper nanoparticles causes gill injury and acute lethality in zebrafish (Danio rerio). Environ. Sci. Technol. 2007, 41, 8178–8186. (8) Yoon, K. Y.; Byeon, J. H.; Park, J. H.; Hwang, J. Susceptibility constants of Escherichia coli and Bacillus subtilis to silver and copper nanoparticles. Sci. Total Environ. 2007, 373, 572–575. (9) Heinlaan, M.; Ivask, A.; Blinova, I.; Dubourguier, H. C.; Kahru, A. Toxicity of nanosized and bulk ZnO, CuO and TiO2 to bacteria Vibrio fischeri and crustaceans Daphnia magna and Thamnocephalus platyurus. Chemosphere 2008, 71, 1308–1316. (10) Aruoja, V.; Dubourguier, H.
C.; Kasemets, K.; Kahru, A. Toxicity of nanoparticles of CuO, ZnO and TiO2 to microalgae Pseudokirchneriella subcapitata. Sci. Total Environ. 2009, 407 (4), 1461–1468. (11) Fahmy, B.; Cormier, S. A. Copper oxide nanoparticles induce oxidative stress and cytotoxicity in airway epithelial cells. Toxicol. In Vitro 2009, 23, 1365–1371. (12) Mortimer, M.; Kasemets, K.; Kahru, A. Toxicity of ZnO and CuO nanoparticles to ciliated protozoa Tetrahymena thermophila. Toxicol. 2010, 269 (2
3), 182–189. (13) Griffitt, R. J.; Luo, J.; Gao, J.; Bonzongo, J. C.; Barber, D. S. Effects of particle composition and species on toxicity of metallic nanomaterials in aquatic organisms. Environ. Toxicol. Chem. 2008, 27 (9), 1972–1978. (14) Griffitt, R. J.; Hyndman, K.; Denslow, N. D.; Barber, D. S. Comparison of molecular and histological changes in zebrafish gills exposed to metallic nanoparticles. Toxicol. Sci. 2009, 107 (2), 404–415. (15) Heinlaan, M.; Kahru, A.; Kasemets, K.; Arbeille, B.; Prensier, G.; Dubourgier, H.
C. Changes in the Daphnia magna midgut upon ingestion of copper oxide nanoparticles: A transmission electron microscopy study. Water Res. 2011, 45, 179–190. (16) Ruparelia, J. P.; Chatterjee, A. K.; Duttagupta, S. P.; Mukherji, S. Strain specificity in antimicrobial activity of silver and copper nanoparticles. Acta Biomat. 2008, 4 (3), 707–716. (17) Unfried, K.; Albrecht, C.; Klotz, L.; Mikecz, A. V.; GretherBeck, S.; Schins, R. P. F. Cellular responses to nanoparticles: Target structures and mechanisms. Nanotoxicol. 2007, 1 (1), 52–71. (18) Xia, T.; Kovochich, M.; Brant, J.; Hotze, M.; Sempf, J.; Oberley, T.; Sioutas, C.; Yeh, J. I.; Wiesner, M. R.; Nel, A. E. Comparison of the abilities of ambient and manufactured nanoparticles to induce cellular toxicity according to an oxidative stress paradigm. Nano Lett. 2006, 6 (8), 1794–1807. (19) Langston, W. J.; et al. Metal handling strategies in molluscs. In Metal metabolism in Aquatic Environments; Langston, W. J., Bebianno, M. J., Eds.; Kluwer Academic Publishers: 1998; 219
283. (20) Damiens, G.; Mouneyrac, C.; Quiniou, F.; His, E.; Gnassia-Barelli, M.; Romeo, M. Metal bioaccumulation and metallothionein concentrations in larvae of Crassostrea gigas. Environ. Pollut. 2006, 140, 492–499. (21) McCord, J. M.; Fridovich, I. Superoxide dismutase: An enzymatic function for erythrocuprein (hemocuprein). J. Biol. Chem. 1969, 244 (22), 6049–6955. (22) Greenwald, R. A. Handbook of Methods for Oxygen Radical Research; CRC Press: Boca Raton, FL 1985.

ARTICLE

(23) Lawrence, R. A.; Burk, R. F. Glutathione peroxidase activity in selenium-deficient rat liver. Biochem. Biophys. Res. Commun. 1976, 71, 952–958. (24) Bebianno, M. J.; Langston, W. J. Quantification of metallothioneins in marine invertebrates using differential pulse polarography. Port. Electrochim. Acta 1989, 7, 59–64. (25) Ellman, G. L.; Courtney, K. O.; Anders, V.; Featherstone, R. M. A new and rapid colorimetric determination of acetylcholinesterase activity. Biochem. Pharmacol. 1961, 7, 88–95. (26) Erdelmeier, I.; Gerard-Monnier, D.; Yadan, J. C.; J. Acudiere, J. Reactions of N-methyl-2-phenylindole with malondialdehyde and 4-hydroxyalkenals. Mechanistic aspects of the colorimetric assay of lipid peroxidation. Chem. Res. Toxicol. 1998, 11, 1184–1194. (27) Lowry, O. H.; Rosenbrough, N. J.; Farr, A. L.; Randall, R. J. Protein measurement with the Folin phenol reagent. J. Biol. Chem. 1951, 193, 265–275. (28) Karlsson, H. L.; Cronholm, P.; Gustafsson, J.; M€ oller, L. Copper oxide nanoparticles are highly toxic: A comparison between metal oxide nanoparticles and carbon nanotubes. Chem. Res. Toxicol. 2008, 21, 1726–1732. (29) Ward, J. E.; Kach, D. J. Marine aggregates facilitate ingestion of nanoparticles by suspension-feeding bivalves. Mar. Environ. Res. 2009, 68, 137–142. (30) Peyrot, C.; Gagnon, C.; Gagne, F.; Willkinson, K. J.; Turcotte, P.; Sauve, S. Effects of cadmium telluride quantum dots on cadmium bioaccumulation and metallothionein production to the freshwater mussel, Elliptio complanata. Comp. Biochem. Physiol. C 2009, 150, 246–251. (31) Serafim, A.; Bebianno, M. J. Metallothionein role in the kinetic model of copper accumulation and elimination in the clam Ruditapes decussatus. Environ. Res. 2009, 109, 390–399. (32) Ahamed, M.; Siddiqui, M. A.; Akhtar, M. J.; Ahmad, I.; Pant, A. B. Genotoxic potential of copper oxide nanoparticles in human lung epithelial cells. Biochem. Biophys. Res. Co. 2010, 396, 578–583. (33) Lehtonen, K. K.; Leini€o, S. Effects of exposure to copper and malathion on metallothionein levels and acetylcholinesterase activity of the mussel Mytilus edulis and the clam Macoma balthica from the Northern Baltic Sea. Bull. Environ. Contam. Toxicol. 2003, 71, 489–496. (34) Renault, S.; Baudrimont, M.; Mesmer- Dudons, N.; Gonzalez, P.; Mornet, S.; Brisson, A. Impacts of gold nanoparticle exposure on two freshwater species: A phytoplanktonic alga (Scenedesmus subspicatus) and a benthic bivalve (Corbicula fluminea). Gold Bullet. 2008, 41 (2), 116–126. (35) Ringwood, A. H.; McCarthy, M.; Bates, T. C.; Carroll, D. L. The effects of silver nanoparticles on oyster embryos. Mar. Environ. Res. 2009, 69 (1), 549–551. (36) Kadar, E.; Lowe, D. M.; Sole, M.; Fisher, A. S.; Jha, A. N.; Readman, J. W.; Hutchinson, T. H. Uptake and biological responses to nano-Fe versus soluble FeCl3 in excised mussel gills. Anal. Bioanal. Chem. 2010, 396, 657–666. (37) Wang, Z.; Zhao, J.; Li, F.; Gao, D.; Xing, B. Adsoprtion and inhibition of acetylcholinesterase by different nanoparticles. Chemosphere 2009, 77 (1), 67–73. (38) Studer, A. M.; Limbach, L. K.; Duc, L. V.; Krumeich, F.; Athanassiou, E. K.; Gerber, L. C.; Moch, H.; Stark, W. J. Nanoparticle cytotoxicity depends on intracellular solubility: Comparison of stabilized copper metal and degradable copper oxide nanoparticles. Toxicol. Lett. 1995, 197, 169–174. (39) Limbach, L. K.; Wick, P.; Manser, P.; Grass, R. N.; Bruinink, A.; Stark, W. J. Exposure of engineered nanoparticles to human lung epithelial cells: Influence of chemical composition and catalytic activity on oxidative stress. Environ. Sci. Technol. 2007, 41, 4158–4163. (40) Canesi, L.; Fabbri, R.; Vallotto, D.; Marcomini, A.; Pojana, G. Biomarkers in Mytilus galloprovincialis exposed to suspensions of selected nanoparticles (Nano carbon black, C60 fullerene, Nano-TiO2, Nano-SiO2). Aquat. Toxicol. 2010, 100, 168–177. (41) Koehler, A.; Marx, U.; Broeg, K.; Bahns, S.; Bressling, J. Effects of nanoparticles in Mytilus edulis gills and hepatopancreas—A new threat to marine life? Mar Environ. Res. 2008, 66 (1), 12–14. 9362

dx.doi.org/10.1021/es200955s |Environ. Sci. Technol. 2011, 45, 9356–9362

ARTICLE pubs.acs.org/est

Perchlorate Uptake in Spinach As Related to Perchlorate, Nitrate, And Chloride Concentrations in Irrigation Water Wonsook Ha,*,†,§ Donald L. Suarez,† and Scott M. Lesch‡ † ‡

U.S. Salinity Laboratory, USDA-ARS, 450 W. Big Springs Rd., Riverside, California 92507, United States Statistical Consulting Collaboratory, University of California-Riverside, Riverside, California 92507, United States ABSTRACT: Several studies have reported on the detection of perchlorate (ClO4
) in edible leafy vegetables irrigated with Colorado River water. However, there is no information on spinach as related to ClO4
in irrigation water nor on the effect of other anions on ClO4
uptake. A greenhouse ClO4
uptake experiment using spinach was conducted to investigate the impact of presence of chloride (Cl
) and nitrate (NO3
) on ClO4
uptake under controlled conditions. We examined three concentrations of ClO4
, 40, 220, and 400 nmolc/L (nanomoles of charge per liter of solution), three concentrations of Cl
, 2.5, 13.75, and 25 mmolc/L, and NO3
at 2, 11, and 20 mmolc/L. The results revealed that ClO4
was taken up the most when NO3
and Cl
were lowest in concentration in irrigation water. More ClO4
was detected in spinach leaves than that in the root tissue. Relative to lettuces, spinach accumulated more ClO4
in the plant tissue. Perchlorate was accumulated in spinach leaves more than reported for outer leaves of lettuce at 40 nmolc/L of ClO4
in irrigation water. The results also provided evidence that spinach selectively took up ClO4
relative to Cl
. We developed a predictive model to describe the ClO4
concentration in spinach as related to the Cl
, NO3
, and ClO4
concentration in irrigation water.

’ INTRODUCTION Perchlorate (ClO4
) salt is used as an oxidizing agent in rocket propellants and explosives.1 Perchlorate has been detected in various water sources, both surface and groundwater, as well as in wine, beverages, baby formula, breast milk, and leafy vegetables.2
6 Perchlorate has been found in ground and surface water in 35 states in the U.S. Currently, a drinking water standard for ClO4
has been set by U.S. Environmental Protection Agency7 and a few states have established advisory levels (for example, 5 μg/L in New York, 6 μg/L as a maximum contaminant level or public health goal in California, and 14 μg/L in Arizona, ref 8 and 9). Perchlorate salts are very soluble in water. Once dissolved, the ClO4
anion is chemically very stable having a +7 oxidation state and persisting in the environment because of high activation energy necessary for reduction.10 The main health concern for ClO4
ion is that it substitutes iodine (similar charge and ionic radius) and thus interrupts thyroid iodine uptake in human beings11 resulting in subsequent hormone disruption and potential perturbations of metabolic activities.12 Perchlorate in water is of concern due to impact on ecosystems13,14 and an additional pathway for humans’ intake via accumulation in vegetables from irrigation water.15
17 Elevated concentrations of ClO4
have been detected in various groundwater sources, related to the release of ammonium ClO4
by military operations, the aerospace industry, and among others. Perchlorate in Colorado River water has been related to ClO4
contamination by the ClO4
salt manufacturing plant previously located near the Las Vegas wash in Nevada.10,18
22 The fresh vegetable industry relies on Colorado River water for r 2011 American Chemical Society

irrigation in the lower Colorado River regions of California and Arizona. Use of Colorado River water thus caused elevated ClO4
concentrations in vegetables.20,23 More recently, installation of a treatment plant on Las Vegas wash has subsequently reduced the ClO4
concentration of the Colorado River.24 Spinach has above-ground parts consumed by humans; which makes it a good choice to study ClO4
uptake. The interaction between salts and ClO4
in edible plants when ClO4
is taken up by plant is not fully investigated. Leaf chloride (Cl
) declined from 4.37 to 2.43% Cl
as NO3
increased from 3 to 15 mmolc/ L in wheat (Triticum aestivum) leaves.25 Net uptake rate of NO3
in Plantago maritima L. was reduced by 23, 33, and 51% at 50, 100, and 200 mol/m3 NaCl, respectively.26 Tan and others14 reported that the uptake of ClO4
in smartweed (Polygonum spp.) was not greatly affected by the presence of NO3
-N, SO42‑, PO43‑, or Cl
in 500 mg L
1 solution. However, ClO4
uptake in three different types of lettuce as independently affected by NO3
, SO42‑, Cl
, pH, and HCO3
was evaluated by Seyfferth et al.27 They concluded that increasing solution NO3
markedly decreased ClO4
uptake but observed no severe effect of Cl
on ClO4
uptake in lettuce leaves. The combined effect of NO3
and Cl
ions on ClO4
uptake has not been examined although both NO3
and Cl
are present at varying concentrations in the soil
water during the crop growing season. Also, the accumulation pattern of ClO4
in root Received: March 25, 2011 Accepted: September 22, 2011 Revised: August 23, 2011 Published: September 22, 2011 9363

dx.doi.org/10.1021/es2010094 | Environ. Sci. Technol. 2011, 45, 9363–9371

Environmental Science & Technology tissues has not been thoroughly investigated under the presence of Cl
and NO3
salts in irrigation water. To date, the research focus has been on lettuce and there are almost no data on ClO4
accumulation of other leafy vegetables such as spinach, as related to ClO4
concentration in irrigation water, and no information on the effects of ions that may potentially inhibit ClO4
uptake in these other leafy vegetables. The objectives of this study are (1) investigate the uptake of ClO4
by spinach as related to ClO4
in irrigation water, (2) evaluate the effect of NO3
and Cl
anions on the uptake of ClO4
by spinach, (3) examine the physiological effect of ClO4
uptake by measuring ClO4
concentration in both leaf and root parts and determine the pattern of translocation of ClO4
within plant materials, and (4) develop predictive equations to represent ClO4
uptake in spinach as related to ClO4
, NO3
, and Cl
in irrigation water.

’ MATERIALS AND METHODS 1. Greenhouse Experiment. The experiment was conducted in 30 sand tanks at the greenhouse facility in U.S. Salinity Laboratory, Riverside, CA. Washed sand (average bulk density: 1.4 Mg/m3) was contained in sand tanks (1.2  0.6  0.5 m depth each). After filling up the water reservoir with deionized (DI) water, the water was circulated through the sand several times to ensure equilibration of the saturation within the sand media. The electric conductivity (EC) was monitored before initiating irrigation to ensure low EC in the DI water. Sorption of ClO4
onto the surface of the sand particles and container was determined to be negligible and thus is not considered further. There were 10 different combinations of irrigation water treatments. Each treatment was replicated three times (Table 1). Three randomly selected sand tanks were irrigated with each water composition during the experiment. We utilized three different concentrations of ClO4
, 40, 220, and 400 nmolc/L, (nanomoles of charge per liter of solution; approximately 4, 22, and 40 μg/L, respectively), Cl
at 2.5, 13.75, and 25 mmolc/L, and NO3
at 2, 11, and 20 mmolc/L. Half Hoagland’s solution (plant nutrient solution) was prepared and the concentration in the irrigation reservoir in mmolc/L was: 0.17 KH2PO4, 0.75 MgSO4 3 7H2O, 2.0 KNO3, 0.25 CaSO4. We utilized DI water and prepared the solutions with reagent grade salts. The pH of the irrigation water ranged between 7.7 and 8.5. The experiment was designed as a classic 23 factorial design, with a center point and one additional nonstandard point set to high ClO4
, medium NO3
, and medium Cl
level. Seeds of hybrid spinach (Spinacia oleracea L., cv. “Space”) were purchased from Johnny’s seeds (Winslow, ME) and planted in November 2007. The plants were irrigated in the sand tanks twice a day at 0900 and 1300 h, saturating the sand, and ensuring a uniform root zone solution composition. Irrigation time was 45 min per event. After each irrigation event, the nutrient solution drained back into the 890 L reservoirs below the sand tanks for a subsequent reuse. The ion concentrations in the reservoirs were constantly maintained by supplementing the nutrients every other week, back to the initial nutrient levels. Water loss by evapotranspiration (ET) was replenished by adding DI water back to the reservoirs to maintain constant volumes and osmotic potentials in each reservoir. Spinach was grown for 71 days under controlled greenhouse conditions which are 41% of relative humidity and 18 and 15 °C of day and nighttime temperatures, respectively. Spinach leaves

ARTICLE

Table 1. Initial Concentrations of ClO4
(nmolc/L), Cl
(mmolc/L), and NO3
(mmolc/L) in Irrigation Water reservoir number

ClO4


Cl


NO3


number of replications

1

40

2.5

2

3

2 3

400 40

2.5 25

2 2

3 3

4

400

25

2

3

5

40

2.5

20

3

6

400

2.5

20

3

7

40

25

20

3

8

400

25

20

3

9

220

13.75

11

3

10

400

13.75

11

3

and root tissue samples were harvested from each sand tank at the end of the experiment. 2. Plant Tissue Processing and Perchlorate Extraction. The plant tissue extraction procedure from Seyfferth and Parker28 was utilized, but weight of plant and volume of water added were modified. The leaves were frozen right after harvesting and stored in the freezer. Approximately 25.0 g of frozen plant sample was weighed and 80.0 mL of DI water was added to plant material for grinding. All standard solutions were made with at least 17.8 MΩ water. After grinding samples, plant material was transferred in 250.0 mL HDPE Nalgene bottle (Nalge Nunc International, Rochester, NY) for 4 h of shaking to release any remaining ClO4
to solution. Samples were centrifuged at 5400 RCF (relative centrifugal force) for 1 h. We filtered approximately 30.0 mL of supernatant using 0.2 μm cellulose NO3
membrane filters (Whatman International Ltd., Maidstone, England). We took approximately 3.0 mL of filtered aliquot and passed it through a preconditioned ENVI-18 SPE cartridge, discarding the first 1.0 mL of sample and collecting 2.0 mL of liquid sample in the glass tube for ClO4
analysis. Perchlorate standard solutions were made from reagent grade sodium perchlorate (NaClO4, Aldrich Chemical Co., Inc., Milwaukee, WI) having density of 2.0 g/cm3 and molecular weight of 122.44 g/mol.

3. ANALYSIS 3.1. Perchlorate Analysis of Plant Samples and Irrigation Water. Perchlorate was analyzed using an Agilent 1100 series

high performance liquid chromatography/mass spectrometry (HPLC/MS). Detailed method for ClO4
analysis utilized for this analysis can be obtained by Snyder et al.29 and U.S. EPA Method 6850. The HPLC settings are briefly discussed as follows. Analytical column: M.IX.MSD1 for LC/MS by MetrohmPeak; autosampler injection volume: 10.0 μL; HPLC pump— flow rate: 0.7 mL/min, mobile phase: 30% of 50.0 mM ammonium formate (NH4COOH), and 25 mM ammonium carbonate ((NH4)2CO3) mixture +70% of acetonitrile (CH3CN). These parameter combinations resulted in elution of the perchorate in approximately 16 min, with a total run time of 18 min. Mass spectrometer parameters are, ionization mode: Electrospray (API-ES); polarity: Negative; Spray chamber—drying gas flow: 12.0 L/min, nebulizer pressure: 35 psig, drying gas temperature: 250 °C; SIM parameters—SIM ion: 99.0, fragmentor: 70 V, gain: 1.0 EMV, dwell time: 290 ms, % relative dwell: 100.0; capillary voltage; 3500 Vcap. The response variable of interest in this study is the concentration of ClO4
in the fresh weight 9364

dx.doi.org/10.1021/es2010094 |Environ. Sci. Technol. 2011, 45, 9363–9371

Environmental Science & Technology

ARTICLE

plant tissue samples (μg/kg FW). The method detection limit (MDL) of HPLC/MS for ClO4
was determined to be 0.5 μg/L in plant extract, which was equivalent to 1.6 μg of ClO4
/kg FW of plant tissue. As shown in Table 2, some of spinach root samples were below the 1.6 μg/kg of detection limit (left-censored in the statistical analyses). All ClO4
concentrations in spinach leaves were above the detection limit. 3.2. Nitrate and Chloride Analysis of Plant Samples and Irrigation Water. The filtered samples after centrifugation (approximately 30.0 mL) were utilized for NO3
and Cl
analysis of plant extracts. Nitrate in plant slurry was measured by UV spectrometry method30 and Cl
was determined by coulometric-amperometric titration method.31 4. Statistical Methodology. The following statistical analysis was conducted to examine the factors controlling ClO4
uptake in spinach and to develop equations relating ClO4
, NO3
, and Cl
concentrations in irrigation water to ClO4
plant tissue concentration. The following linear factorial model (with 2-way interaction) was fit to both the natural log transformed leaf and root tissue data:
lnðClO
4 : accumÞ ¼ β0 þ β1 lnðClO4 Þ
þ β2 lnðNO3 Þ þ β3 lnðCl
Þ
þ β12 lnðClO
4 Þ  lnðNO3 Þ

þ β13 lnðClO4 Þ  lnðCl Þ
þ β23 lnðNO
3 Þ  lnðCl Þ þ ε

ð1Þ

In eq 1, the ε error term represents an independently, identically and normally distributed error component and the various β parameters quantify the primary (first order) and two-way interaction terms.32 Positive parameter estimates in this model imply that the log ClO4
concentrations in the plant tissue increase as the log transformed ClO4
, NO3
, and/or Cl
water concentrations Table 2. Number of Left-Censored Spinach Root Tissue for Perchlorate Measurements (I.E., Measurements Below the 1.6μg/kg FW Method Detection Limit of ClO4
) treatmenta





low ClO4 , low NO3 , low Cl

spinach roots


0

high ClO4
, low NO3
, low Cl


0

low ClO4
, low NO3
, high Cl


1

high ClO4
, low NO3
, high Cl


0

low ClO4
, high NO3
, low Cl
high ClO4
, high NO3
, low Cl


1 0

low ClO4
, high NO3
, high Cl


3

high ClO4
, high NO3
, high Cl


0

mid ClO4
, mid NO3
, mid Cl


0

high ClO4
, mid NO3
, mid Cl


0

Low, mid, and high ClO4
[ppb] represent 4, 22, and 40, respectively; Low, mid, and high NO3
[mmolc/L] represent 2, 11, and 20, respectively.; Low, mid, and high Cl
[mmolc/L] represent 2.5, 13.75, and 25, respectively. a

increase, while negative estimates imply that the log ClO4
concentrations decrease as these water concentration levels increase. For the spinach leaves data (where all samples were above the detection limit and thus no censoring occurred), eq 1 was estimated using standard linear modeling techniques.32 For the left-censored spinach root data, eq 1 was estimated using maximum likelihood techniques.33 All model estimation was performed using the GLM and LIFETEST procedures in SAS.34 Based on the p-values associated with the estimated parameters (Table 3), reduced forms of eq 1 were also fit to each plant tissue data set. These reduced models were estimated by removing all nonsignificant parameter estimates from the linear factorial model (at the 0.05 significance level). Goodness-of-fit (GOF) tests were calculated to assess the adequacy of each fitted equation. For the complete (i.e., noncensored) leaves data sets, traditional lack-of-fit (LOF) F-tests were computed.32,35 For the left-censored root data sets, asymptotic GOF tests were computed by calculating the log-likelihood (LL) score differences between the reduced and saturated models and then comparing these
2 LL scores to Chi-square distributions with the appropriate degrees of freedom. The primary goal in each analysis was to identify a parsimonious linear factorial model that fully described how the changing irrigation water ClO4
, NO3
, and Cl
concentrations influenced the plant tissue ClO4
concentrations.

’ RESULTS AND DISCUSSION The sand tank environment was hypothesized to potentially cause ClO4
degradation by bacteria in the root zone.14 Because of this consideration, various researchers utilized an aerated hydroponic system for the laboratory scale plant uptake experiment to minimize the rhizosphere degradation effect on ClO4
uptake (refs 14,27,36, etc.). However, commercial leafy vegetables have been grown primarily in soil under field environments. Our experiment was designed to evaluate the combined effect of NO3
and Cl
on ClO4
uptake in spinach in a controlled sand tank environment which both more closely reflects field conditions and yet enables accurate monitoring of root zone ClO4
concentrations. The concentration of ClO4
in reservoirs was monitored and maintained at the constant concentrations (4, 22, and 40 μg/L) throughout the experiment as indicated in the Materials and Methods section. We found no evidence of decrease in ClO4
related to a soil process, suggesting that, as expected, our rhizosphere was highly aerobic and ClO4
degradation did not need to be further considered in our experiment. ET losses were approximately equal within the range of 1.9 and 2.8 cm of water for all treatments. Low Cl
treatments had ET loss of 2.23 (average of four treatments) and high Cl
reservoirs showed 2.55 cm of water, while water losses by ET in low and high NO3
treatment reservoirs were 2.38 and 2.4 cm of water, respectively. The interactive effects of the three independent variables on ClO4
in plant parts can best be evaluated by a multivariate

Table 3. Summary Statistics of RMSE and Parameter P-Values: Full Factorial Models type III p-values for individual parameter estimates data set

RMSE

β1

β2

β3

β12

β13

β23

spinach: leaves

0.373

98%) of the exposures.

’ INTRODUCTION Bisphenol A (BPA) is an endocrine-disrupting chemical and has been implicated in a wide variety of adverse health outcomes in humans.1
6 Due to its toxicity and widespread human exposure, BPA has received the attention of regulatory agencies across the globe.1 An oral reference dose (RfD) for BPA of 50 μg/kg body weight (bw)/day has been established by the United States Environmental Protection Agency (USEPA) and the European Food Safety Authority (EFSA).7,8 Nevertheless, some studies have reported that BPA can stimulate cellular responses and toxic effects at exposure doses below the currently recommended RfD.1
3 The low-dose exposures and toxicity of BPA are subjects of debate.6,9 Produced in quantities of over 8 billion pounds each year worldwide, BPA is one of the most widely used chemicals, as the base chemical in the manufacture of polycarbonate plastics and the resin lining of food and beverage cans.10,11 BPA can leach out of products, through the hydrolysis of ester bonds linking BPA monomers, under acidic or basic conditions.12 BPA has been reported to occur in various environmental matrices, including air, water, sewage sludge, soil, dust, foodstuffs, and soft drinks.2,3,13
15 r 2011 American Chemical Society

Biomonitoring studies have reported widespread occurrence of BPA at ng/g or ng/mL levels in human tissues and fluids, including urine, blood (maternal blood, cord blood, and fetal blood), and other bodily fluids (amniotic fluid, follicle fluid, saliva, and breast milk).1
3,16
21 There are multiple sources that contribute to human exposure to BPA.3,9 Diet, however, appears to be a major source of human exposure to BPA.18,22
24 The U.S. National Toxicology Program reported that oral exposure from foods of adults in the USA to BPA ranged from 0.008 to 1.5 μg/kg bw/day.22 On the basis of the leaching levels from consumer products and consumption of canned foods, BPA exposures have been estimated to range from 212 for BPA (97%; Sigma-Aldrich, St. Louis, MO) and 239 > 224 for 13C12
BPA (99%; Cambridge Isotope Laboratories, Andover, MA). Quality Assurance and Quality Control (QA/QC). For each batch of 20 samples analyzed, a procedural blank, a spiked blank, a pair of matrix spiked samples, and duplicate samples were processed. The procedural blank, containing water in place of paper sample, was analyzed by passage through the entire analytical procedure as a check for interferences or laboratory contamination. BPA was not detected in procedural blanks or sample containers (i.e., polypropylene tube). The recoveries of BPA from spiked blanks and spiked matrices were 100 ( 15% and 101 ( 17% (mean ( SD), respectively. The relative standard deviation (RSD) of replicate analysis of samples was 0.99, n = 10). The limit of quantitation (LOQ) was 1 ng/g. Quantification was made using the isotope-dilution method. Prior to the analysis of samples, recovery and reproducibility of the method were verified by spiking known concentrations of BPA into selected paper samples. Because thermal receipt papers contained high concentrations of BPA, these samples were stored and analyzed in separate batches from other paper types. Estimation of Daily Intake. Skin uptake/dermal absorption from handling of paper products (especially thermal paper receipts) can be a pathway of human exposure to BPA. In thermal papers, BPA exists as a free monomer that is mobile and transferable to objects with which it comes in contact.29,35 The extent of human exposure to BPA through handling of paper products remains unknown.29,35,39,40 Biedermann et al. 29 tested the transfer of BPA from thermal receipt paper to human skin and reported that holding a thermal paper with 15.2 g BPA/kg (mean value) for 5 s resulted in the transfer of 1636 ng BPA to the surface of the hand. On the basis of this information, we calculated a paper-to-skin transfer rate of BPA as k = 1636 ng/(15.2 g/kg 3 5s) = 21 522 ng/s. It has been reported that 27% of BPA found on skin surface penetrates and reaches the bloodstream within 2 h.29 We estimated human exposure to BPA based on the measured concentrations and frequency of handling of paper products. We assumed that the general population handles thermal receipts twice a day. The frequency of use and handling time are probably different for various paper types. Some paper types, including magazines, newspapers, napkins, paper towels (or kitchen roll), and toilet papers may be handled more frequently. We assumed that the general population handles these five paper types ten times a day and the remaining paper types five times a day. Individuals who work at cash registers (e.g., cashiers, bank tellers), however, can handle receipts more frequently on a daily basis. For occupational exposures, we assumed that individuals handle thermal receipts 150 times a day, and handle other paper types at rates similar to the general population. Based on the geometric mean, median, fifth and 95th percentile concentrations of BPA measured in our paper samples, we estimated the daily intake (EDI; ng/day) of BPA as shown in eq 1: EDI ¼ k  C  HF  HT  AF=106

ð1Þ

where k is paper-to-skin transfer coefficient of BPA (calculated as 21522.4 ng/s); C is the concentration of BPA in paper samples (μg/g); HF is handling frequency (times/day; for thermal receipt paper, 2 and 150 times/day for the general population and occupationally exposed individuals, respectively; for other paper types, 5 times/day for flyer, ticket, mailing envelope, food

contact paper, food carton, airplane boarding pass, airplane luggage tag, printing paper, and business card, and 10 times/ day for magazine, newspapers, napkin, paper towel, and toilet paper for both the general population and occupationally exposed individuals); HT is handling time of paper and is assumed to be 5 s for each handling; and AF is the absorption fraction of BPA by skin, which is 27%.29 For example, if the median BPA concentration in thermal receipt is 0.299 mg/g (= 299 μg/g), the estimated median daily intakes of BPA, from handling of thermal receipts, by the occupationally exposed individual can be calculated as follows: EDI ¼ 21522:4ng=s  299μg=g  150times=day  5s=time  27%=106 ¼ 21:5224ng=s  0:299  150times=day  5s=time27% ¼ 1303ng=day

Statistical Analysis. Geometric mean (GM), median, and concentration ranges were used to describe the results. Concentrations below the LOQ were substituted with a value equal to the LOQ divided by the square root of 2 for the calculation of GM. Differences between groups were tested by a one-way ANOVA with the Tukey test.

’ RESULTS AND DISCUSSION Thermal Receipt Papers. BPA was detected in 94% of thermal receipt paper samples (n = 103) at concentrations ranging from below LOQ to 13.9 mg/g with a GM of 0.211 mg/g (Table 1). Of the 103 receipt samples analyzed, 73 (71%) were collected from Albany, New York. BPA concentrations in receipts from Albany were in the range of 0.005 to 9.38 mg/g with a GM of 0.341 mg/g (Figure 1). These values are slightly lower than those found in samples collected from other cities in the USA, including New York City, Buffalo, Boston, Chicago, Weston, and Charlotte (GM: 0.496 mg/g; one-way ANOVA, p > 0.05), but significantly lower than the concentrations found in samples from Incheon, Korea (GM: 1.56 mg/g; p < 0.01) and Hanoi, Vietnam (GM: 6.32 mg/g; p < 0.05). No significant difference was found in the concentrations of BPA in thermal receipt papers collected between Korea and Vietnam (Figure 1). Interestingly, BPA was not detected in any of the six thermal receipt paper samples collected from several stores in Matsuyama and Tokyo, Japan (Table 1). This is attributed to a phase-out of BPA usage in thermal receipt papers in Japan in 2001.41 Although BPA was reported to have been replaced with Bisphenol S by a major manufacturer of thermal receipt papers in the USA, BPA is still widely present in thermal receipt papers from the USA in 2010
2011. Some receipt papers claimed to be “BPA-free” (specifically printed on the receipt papers), but all of these receipt papers contained hundreds of μg/g levels of BPA (GM: 217 μg/g). A few earlier studies have reported BPA concentrations in thermal receipt papers.28,29 Eight of 10 blank (unprinted) thermal receipt papers collected from retail stores in Wilmington, Massachusetts, USA, contained BPA at concentrations ranging from