Humic Substances. Nature's Most Versatile Materials

HUMIC SUBSTANCES

Nature’s Most Versatile Materials

HUMIC SUBSTANCES Nature’s Most Versatile Materials Edited by

Elham A.Ghabbour Northeastern University, Boston, USA and Soil, Water and Environmental Research Institute, Alexandria, Egypt Geoffrey Davies Northeastern University, Boston, USA

Based on the proceedings of the 11th Biennial Conference of the International Humic Substances Society and the 6th Humic Substances Seminar held on 21–27 July 2002 at Northeastern University, Boston, Massachusetts, USA

Taylor and Francis, Inc. New York

Denise T.Schanck, Vice President Robert L.Rogers, Editor Liliana Segura, Editorial Assistant Thomas Hastings, Marketing Manager Maria Corpuz, Marketing Assistant Dennis P.Teston, Production Director Anthony Mancini Jr., Production Manager Brandy Mui, STM Production Editor Mark Lerner, Art Manager Daniel Sierra, Cover Designer Published in 2004 by Taylor & Francis 29 West 35th Street New York, NY 10001 This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”

Published in Great Britain by Taylor & Francis 11 New Fetter Lane London EC4P 4EE Copyright © 2004 by Taylor & Francis Books, Inc. All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publisher. Library of Congress Cataloging-in-Publication Data Humic substances: nature’s most versatile materials/edited by Elham A.Ghabbour, Geoffrey Davies. p, cm. “Based on the proceedings of the 11th biennial conference of the International Humic Substances Society and the 6th Humic Substances Seminar held on 21–27 July 2002 at Northeastern University, Boston, Massachusetts, USA.” Includes bibliographical references and index. ISBN 0-59169-015-3 (alk. paper) 1. Humus—Congresses. 2. Soils—Humic contents—Congresses. I. Ghabbour, Elham A. II. Davies, Geoffrey. 1942–III. International Humic Substances Society. IV. Humic Substances Seminar (6th:2002: Northeastern University) S592.8.H92 2003 631.4'′17–dc21 2003053112 ISBN 0-203-48760-5 Master e-book ISBN

ISBN 0-203-59493-2 (Adobe eReader Format)

Contents

Preface

vi

Contributors

x

PART 1. FRACTIONATION AND CHARACTERIZATION: THE STATE-OF-THE-ART 1.

Use of Radioactive Tracers for the Characterization of Humic and Fulvic Acids in High Performance Size Exclusion Chromatography Karsten Franke, Doris Rössler and Hermann Kupsch

2

2.

Interpreting Capillary Electrophoresis—Electrospray/Mass Spectrometry (CZE-ESI/MS) of Suwannee River Natural Organic Matter (NOM) Philippe Schmitt-Kopplin

6

3.

Comparison of As-Delivered and AFFFF-Size-Fractionated Suwannee River Fulvic Acid by Time-ofFlight Mass Spectrometry Wilfried Szymczak, Manfred Wolf andKlaus Wittmaack

21

4.

Molecular Fingerprinting of Aquatic Fulvic Acids by Ultra-High Resolution ESI FT-ICR Mass Spectrometry William T.Cooper and Alexandra C.Stenson

27

5.

The Macromolecular or Supramolecular Nature of Humic Substances: A Dynamic Light Scattering Study Gustavo González-Gaitano and Josemaría García-Mina

35

6.

A Proposal for the Establishment of a Database of Thermodynamic Properties of Natural Organic Matter Rossane C.DeLapp and Eugene J.LeBoeuf

40

PART 2. HYDRATION, SWELLING AND SORPTION: CONTRIBUTING FACTORS 7.

Effect of Hydration/Solvation of Organic Matter on Sorption of Organic Compounds: Conception and Sorption Isotherm Model Ellen R.Graber and Mikhail Borisover

54

8.

Swelling of Organic Matter in Soil and Peat Samples: Insights from Proton Relaxation, Water Absorption and PAH Extraction Gabriele E.Schaumann, Julia Hurrass, Martin Müller and Wolfgang Rotard

65

9.

Sorption of PAHs to Natural Sorbents: Impacts of Humic and Lipid Fractions Luc Tremblay, Scott D.Kohl, James A.Rice and Jean-Pierre Gagné

76

10.

Interactions and Conversions of Polycyclic Aromatic Compounds in the Process of Humification Matthias Hübner, Kristoffer E.N.Jonassen and Torben Nielsen

88

11.

Pyrolytic Study of the Bound Residues of 13C-Atrazine in Soil Size Fractions and Soil Humin Marie-France Dignac, Yahya Zegouagh, Ludovic Loiseau, Gérard Bardoux, Enrique Barriuso, Sylvie Derenne, André Mariotti and Claude Largeau

101

12.

Phenanthrene Sorption by Clay-Humic Complexes Kaijun Wang, Elham A.Ghabbour, Geoffrey Davies and Baoshan Xing

109

13.

Kinetics of Desorption of 2,4-dichlorophenoxyacetic Acid from Humic Acid, Metal Oxides and Metal Oxide-Humic Complexes C.LiuP.M.Huang

115

v

PART 3. METAL BINDING AND MOBILITY: THEORY, DATA AND CONSEQUENCES 14.

Exploring the Molecular Character and Heterogeneity of Humic Substances via the Study of the IonBinding Process Using an Extended Polyelectrolyte Model Josemaría García-Mina

123

15.

Study of Fulvic-Aluminum(III) Ion Complexes by 27Al Solution NMR Norman C.Y.Lee and David K.Ryan

139

16.

Investigation of Colloidal Properties and Trace Metal Complexation Characteristics of Soil-Derived Fulvic Acids by Flow Field-Flow Fractionation-Inductively Coupled Plasma-Mass Spectrometry (Flow FFF-ICPMS) Jonathan Bell, Dula Amarasiriwardena, Atitaya Siripinyanond, Baoshan Xing and Ramon M.Barnes

145

17.

Comparison of Dialysis, Polarography and Fluorimetry for Quantification of Cobalt(II) Binding by Dissolved Humic Acid Fanny Monteil-Rivera, Jean-Paul Chopart and Jacques Dumonceau

154

18.

Diffusion of Metal Cations in Humic Gels Martina Klu áková and Miloslav Peka

167

19.

The Role of Humic Substances in Trace Element Mobility in Natural Environments and Applications to Radionuclides Valerie Moulin, Badia Amekraz, Nicole Barre, Gabriel Planque, Florence Mercier, Pascal Reillerand Christophe Moulin

175

20.

Influence of Humic Substances on the Migration of Actinides in Groundwater G.Buckau, M.Wolf, S.Geyer, R.Artinger and J.I.Kim

184

21.

Catalytic Effects of Ni-Humic Complexes on the Reductive Dehalogenation of C1 and C2 Chlorinated Hydrocarbons Edward J.O’Loughlin, Huizhong Ma and David R.Burris

191

PART 4. BIOGEOCHEMICAL EFFECTS: THE GOOD, THE BAD AND THE UGLY 22.

Humic Substances and Their Direct Effects on the Physiology of Aquatic Plants 209 Stephan Pflugmacher, Constanze Pietsch, Wiete Rieger, Andrea Paul, Torsten Preuer, Elke Zwirnmann and Christian E.W.Steinberg

23.

More Evidence for Humic Substances Acting as Biogeochemicals on Organisms C.Wiegand, N.Meems, M.Timoveyev, C.E.W.Steinberg and S.Pflug macher

224

Index

233

Preface

More people, less arable land and more pollution mean greater reliance on humic substances, Nature’s most versatile materials and the major regulators of plants, soils and water. Food and pollution are everyone’s problem and understanding humic substances is in everyone’s best interest. Fortunately, we are witnessing a remarkable growth of knowledge of natural organic matter (NOM) in general and its humic components (the relatively long-lived humic and fulvic acids and humin) in particular. The Humic Substances Seminars at Northeastern University promote and acknowledge this increased understanding. This book captures the spirit of the impressive progress being made. Humic Substances Seminar VI in July 2002 immediately followed the Twentieth Anniversary Biennial Conference of the International Humic Substances Society in Boston. As a result we have been able to invite contributions from Seminar and the Conference, and by doing so cover the major advances in the field. This book reports the best current research on humic substances from around the world. Its contributions follow a natural progression of humic substances separation, characterization, sorption (including all-important interactions with water), metal binding and transport, and increasingly apparent biogeochemical effects. Each contribution encourages us to combat drought and pollution with humic substances and save our planet for this and future generations. FRACTIONATION AND CHARACTERIZATION: THE STATE-OF-THE-ART Chromatography still heads the list of methods that fractionate humic substances (HSs). The workhorses are high performance size exclusion chromatography (HPSEC) and dialysis, whose calibration and detector sensitivity are troublesome. Franke and colleagues have labeled humic and fulvic acids with two radioisotopes and used a high sensitivity germanium detector to watch 131I and 111In exiting as HPSEC column. Close similarity of the radiochromatograms confirms covalent binding of iodine and tight binding of indium, a surrogate for aluminum, which is toxic but difficult to study. This contribution illustrates the analytical benefits of radiolabeling and complements descriptions of radionuclide binding and mobility later in this book. Schmitt-Kopplin is contributing strongly to improved HSs separation, detection and characterization. The main focus of this contribution is capillary electrophoresis, the workhorse of genome and proteome research and increasingly useful for humic substances characterization. Efficient fractionation of HSs by selective methods such as capillary zone electrophoresis is still not achieved. One sensible suggestion is to avoid reactive buffers, with carbonate recommended and ammonium as the counterion to get rid of the volatile buffer prior to analysis by electrospray mass spectrometry. The interplay between mass and charge in determining electrophoretic mobility is dominated by charge. Combined electrophoretic/mass spectral data in this contribution are a prelude to other chromatographies and successful applications of mass spectrometry in later chapters. A pervading theme is that humic substances from completely different sources exist as homologous families more similar than we thought just a year ago. Scymczak and colleagues have used the relatively new asymmetric flow field flow fractionation technique to separate Suwannee River fulvic acid into fractions. Comparison of the time-of-flight secondary ion mass spectra for m/z=200 to 2000 Da of the fractions and the unseparated fulvic sample indicates remarkable similarity. This supports conclusions that fulvic acid fractions contain the same or very similar low mass molecules that may associate to form aggregates. It wasn’t long ago that getting any kind of mass spectrum of a humic sample was a cause for celebration. Now Cooper and Stenson at Florida State University and the National High Magnetic Field Laboratory have identified individual molecular components of Suwannee River fulvic acid with electrospray ionization Fourier transform—ion cyclotron resonance mass spectrometry (ESI FT-ICR MS). Virtually all the 5500 ions between m/z=300–1100 Da have been identified by this ultrasensitive method and guess what: the components are closely similar members of 266 homologous molecular families! We can’t think of better use of an ESI FT-ICR MS instrument than to confirm humic substances similarity after centuries of thinking that NOM samples have little or nothing in common. Are humic substances macromolecules or supramolecular aggregates of small molecules held together by weak intermolecular forces? This history-making debate goes on. González-Gaitano and García-Mina contribute by using dynamic light scattering to investigate fulvic samples fractionated with HPSEC. They conclude that acidification and re-alkalization

vii

with hydrochloric, citric and oxalic acids gives the same result: no difference in mass distribution and no evidence for supramolecule formation from the lowest molar mass constituents. Thermal analysis techniques provide a means to quantify the thermodynamic properties of heat capacity, thermal expansion coefficient and glass transition temperatures, as described in detail by DeLapp and LeBoeuf. These parameters combined impose constraints on existing molecular simulations models and help to focus attention on models with closest similarities to real HSs systems. HYDRATION, SWELLING AND SORPTION: CONTRIBUTING FACTORS Humic substances are major importers, exporters and transporters of solutes in soils and natural waters, and they play a much greater role than clays and minerals in this respect. These HSs functions depend on sorption rates and equilibria. As of now, we can only speculate on the overall contributions of HSs behavior to the global ecosystem. But now their importance is receiving the close scrutiny it deserves. The seven contributions in this section represent excellent work on sorption being conducted around the world. Interactions of humic substances with water dominate all the processes investigated. Dissolved NOM can be analyzed as a system of natural ligands with the aid of well-tried models. Sorption by solid HSs makes a larger environmental contribution, but it is more difficult to conceptualize. One important effort is to test and re-test the ‘dual-mode” sorption model, which divides interactions of solid HSs with solutes into partition (as between an organic solvent and water) and adsorption, which implies either interaction with a rigid sorbent framework or with specific (though heterogeneous) adsorption sites. The test often uses the Freundlich model to explain isotherm non-linearity due to solute adsorption. New attention is being paid to effects on sorption of wetting, swelling and hydration of solid HSs, particularly as they affect solute sorption by hydrated solid HSs from water on the one hand and by dry solid HSs from “dry” solvents on the other. In essence, we are using solvent/hydration behavior to interrogate the active sites of HSs, a worthy cause that requires no definite molecular structure but still reveals interaction mechanisms, as for uncrystallized proteins and enzymes. Graber and Borisover’s review of their latest sorption work makes fascinating reading. They focus on a link-disruption (or link-solvation) model developed from the local to general to co-operative levels. This gives a general isotherm model that relates the differences between intra- and intermolecular interactions in the dry state to competition between the solvent and solute for interactions with sites released when the linkages are broken on hydration. The model simplifies to the phenomenological Freundlich description at low solute activities that correspond to the natural conditions used in the most revealing experimental studies. Schaumann and colleagues have studied dehydration effects on the extractability of polyaromatic hydrocarbons from soils. Peat and whole organic soil hydration/swelling have been investigated with measured 1H nuclear relaxation times and gravimetry. Starch and semolina are used as standards. The relaxation times depend on chemical composition and also on the particle size of the dry materials, which affects the pore size distributions. A difference between pore size distributions in the dry and hydrated states is necessary to monitor hydration. Pores in swollen semolina are larger than those in swollen peats and soils. The data depend strongly on the individual sample. Schaumann et al. have measured the low swelling rates (time constants of days instead of hours for minerals) and suggest a relation between PAHs extractability from soil and its state of hydration for PAH-laden samples of the same contaminant age. Slow PAHs sorption on sorbents from coastal sources is the focus of the contribution from Tremblay and colleagues, who demonstrate the importance of lipid content in sediments and humins on the extent, rate and character of PAHs sorption. Removing lipid components increases the non-linearity of the sorption isotherms, consistent with unblocking of lipidrestricted adsorption sites. Re-addition of lipids increases isotherm linearity, as expected. There is a strong correlation between the capacity for PAHs and the carbon content of the sorbent, even for non-linear isotherms. Huebner and colleagues have attached humic residues to chromatographic silica and used the products as solid phases to study the sorption of polycyclic aromatic compounds (PACs) by the attached humic substances. Complementary data from solid phase microextraction and liquid-liquid extraction are reported, along with an investigation of iron(III) catalyzed humification of PACs as a means of contaminated sites remediation in concert with current bioremediation strategies. Leftover pesticides become non-extractable (by water and methanol) or ‘bound’ residues in soils, and people worry about their long-term health effects. Identifying the residues is difficult, as illustrated by the contribution from Dignac and colleagues. Analytical pyrolysis results pesticide-SOM reactions that obscure bound residue identification by PY-GC-MS. Fresh soil forms the most bound residues and it seems that microbes ‘adapt’ to metabolize atrazine, lessening the percentage of ‘bound’ pesticide in previously exposed soils. Most of the pesticide detected is unchanged atrazine, which is another example of how humic substances are able to wrap around metals and other solutes to slow or stop their desorption. A sizeable fraction of solid humic substances exist in soils as clay-humic composities, and the question is how the sorptive properties of HSs are affected when they coat a mineral surface. Wang and colleagues report that a humic coating increases the sorptive capacity of two common clays for phenanthrene and that the isotherms are non-linear, indicating specific

viii

adsorptive interactions. Two possible reasons for this behavior are 1) a more rigid HS structure induced by the mineral surface template, and 2) disruption of functional group linkages on surface binding or hydration, as implied by Graber, Borisover and other researchers. This theme of HS-mineral surface effects also is the focus of Liu and Huang’s description of the kinetics of desorption of the common pesticide 2,4-dichlorophenoxyacetic acid (2,4-D) from humified catechol on Al, Fe and Mn oxide surfaces. 2,4-D was one of the first pesticides used in Canada and exists in enormous quantities in its soils. Lui and Huang report that 2,4-D desorption is assisted by citrate and obeys the rate law expected for overall parabolic diffusion. Solutes have to fight their way through humic matrices for desorption to occur. Metal cation diffusion through humic gels that model mineral coatings is explored later in this book. METAL BINDING AND MOBILITY: THEORY, DATA AND CONSEQUENCES Wanting to understand metal binding by humic substances has three main driving forces: nutrient supply to plants, toxic metal sequestration by soils and radionuclide transport from nuclear waste sites by water. As with any solute, we need to know the thermodynamics and kinetics of metal binding, and some of the best minds in the field are working on these issues. Garcia-Mina adds to the stock of metal binding tools by describing a model based on polyelectrolyte theory that incorporates the key factors of competitive proton and metal ion binding without requiring a pre-conceived humic geometry or structural pattern. A distinctive feature of the ‘EPM’ model is that electrostatic parameters related to molecular geometry are obtained by graphical analysis of the data. This removes the need to assume a specific molecular geometry. One intriguing conclusion is that ‘hidden’ carboxylic acid groups become accesssible for metal binding due to humic molecule expansion at higher pH. This is another way in which humic substances buffer natural systems: they bind metals more strongly at higher pHs where many metals precipitate. As mentioned earlier, aluminum binding by humic substances is difficult to measure and model because of aluminum’s complicated aqueous chemistry. Lee and Ryan have studied Al binding with 27Al nuclear magnetic resonance (I=5/2), which is simple in principle but difficult in practice. Al-FA complexes form at pH ~4 but Al hydrolysis takes over at higher pH. The Al-FA and Al-oxalate NMR spectral data compare well, suggesting oxalate as a useful model for Al-fulvic acid binding in the environment. The contribution from Bell et al. spotlights renewed attention to flow fleld flow fractionation, which in principle is an ideal way of studying humic aggregation and solute binding. Fractionation is coupled to ICP-MS to study the distribution of a range of metals complexed by fulvic acids from soils with different cropping systems. The results show that metals bind to a range of FA fractions with apparent molar masses of 900 Da and hydrodynamic diameters of 2 nm. Fractionation has no large effect on metal binding, as expected for similar molecular constituents. Agricultural management practices also have little effect on the molar masses or hydro-dynamic diameters of the FA samples investigated. Measuring weak metal binding by humic substances is challenging because these natural ligands have both strong and weak binding sites and high total metal concentrations are needed to detect the weakest ones. As an example, cobalt(II) binding by humic substances is difficult to measure because the binding is weak and different measurement methods work best over different total metal concentrations. The contribution by Monteil-Rivera et al. compares the requirements and limitations of three potential ways of measuring weak cobalt(II) binding: fluorescence quenching, polarography and dialysis. The preferred method is dialysis, which is the most cumbersome of the three but gives the most straightforward results. Transport of nutrient and toxic metals and of radionuclides as mobile humic substance complexes is a major environmental concern. Metal transport by HSs depends on the rates of dissociation of metals from predominantly carboxylate ligands. It is known that the dissociation rates are much lower than for dissociation of a given metal from the same donor atoms in simple mononuclear complexes and the question is why? Analysis of data with a model for diffusion of Cuaq2+, Coaq2+, and Niaq2+ through a lignite-derived humic substance gel at 25°C in the work by Klu′ áková and Peka′ includes a metal binding-dissociation component. However, the apparent diffusion coefficients are similar to those for the same cations in water. This is because humic substance gels typically are 95% water by mass because of the very high affinity of humic substances for water. Moulin et al. review methods to measure and predict the effects of humic substances on the solubility, speciation and transport of radionuclides and other environmental metals. Humic substances affect the bioavailability and toxicity of run-off from nuclear waste sites and determine the safety of such repositories in the short and long term. Time-resolved laser induced fluorescence is a very useful tool in this work. The results confirm the existence of tight and weaker metal binding sites in humic substances and the inclusion of hydroxo and carbonato ligands along with HSs in ternary complexes formed by radionuclides and lanthanides under environmental conditions. Also discussed is experimental evidence that iodine, another nuclear waste component, is covalently bonded to aromatic constituents in its reactions with fulvic acids, and the effects on radionuclide retention of humic coatings on mineral surfaces. This contribution contains very useful insight for researchers in environmental nuclear chemistry.

ix

The metal transport theme is continued in the contribution by Buckau et al., who focus on the origin, concentrations, stability and mobility of humic substances in radionuclide-contaminated ground waters and the effects of HSs interactions on radionuclide transport. For example, the transport of americium by humic colloids can be described with a model consisting of two consecutive binding steps, quasi-irreversible binding and no colloid retention. Precautions necessary to obtain reliable data for HSs system modeling are carefully described and the contribution emphasizes the need for kinetic data rather than reliance on equilibrium interactions that underestimate trace metal transport. The list of known humic substances functions would easily fill this page, which is why we designate them ‘Nature’s most versatile materials.’ Work in recent years has revealed one of humic substances’ least appreciated functions: redox catalysis associated with quinoid components in HSs structures. The contribution by O’Laughlin et al. describes the products and kinetics of homogeneously catalyzed trichloroethene (TCE) and other chlorinated hydrocarbon reductions with titanium(III) citrate in water. The catalysts are Ni-Aldrich humic acid complexes and the products are non-chlorinated hydrocarbons, which is the grail of chlorohydrocarbon-contaminated water remediation. The contribution discusses likely rate-determining steps and subsequent reactions leading to the desired products. As such, it will be of interest to anyone working on remediation of halocarbon-contaminated water. Of equal importance is heightened awareness of HSs catalysis of many biochemical and environmental processes. BIOGEOCHEMICAL EFFECTS: THE GOOD, THE BAD AND THE UGLY Humic substances mostly consist of plant metabolites, and the onset of humification is associated with senescence, the twilight zone between the life and death of a plant. The link between the death of one organism and the physiology of another is through the products of death, which are humic substances. Significant levels of humic substances were first discovered in a marine alga about eight years ago, and now we are seeing more studies of the direct effects of humic substances on plants and animals. The last two contributions of this book explore the effects of humic substances on the physiology of aquatic plants and water animals. As such, they are major contributions to our understanding of yet another role of humic substances in the environment and on human health. The contribution by Pflugmacher et al. demonstrates strong effects of natural HSs and a synthetic HS on the enzyme systems and inhibited photosynthetic activity of two common aquatic plants (Ceratophyllum demersum and Vesicularia dubyana). The effects are similar in nature and magnitude to those of anthroquinone, a known inhibitor of plant photosynthesis. The HS spin densities are related to their effects on photosynthetic oxygen production, which is thus shown to be a redox reaction mediated by humic substances. A plant either synthesizes humic substances during its own lifetime or is exposed to humic substances from another organism. Put another way, an organism can induce senescence in a plant. This collection of important new work in humic substances science is completed with the contribution from Wiegand et al. on humic substances as geochemicals. Specifically, the physiological effects of humic substances on the common carp (Cyprinus carpio), the water flea (Daphnia magna) and three amphipod species are enumerated. Animal stress and changes in enzymatic activity on exposure to humic substances from different sources are demonstrated. Possible mechanisms are HSs uptake by cells or adsorption on cell surfaces. The sources of HSs affect their structures and so a quantitative structure activity relationship (QSAR) is expected as more data become available. Many participants in Humic Substances Seminar VI and the IHSS Conference commented that the scientific level was the highest they have seen at any previous forum on humic substances. Hard work on HSs over many years is beginning to pay dividends, as demonstrated by the contributions in this book. We are standing on the shoulders of giants with the prospect that Nature’s most versatile materials will one day be fully understood and appreciated. In the meantime, we encourage you to make your own best contribution to this important goal. ACKNOWLEDGEMENTS We thank the authors for their enthusiasm and co-operation, the contributors to the 20th Anniversary Conference of the International Humic Substances Society and Humic Substances Seminar VI, the staff of Northeastern University who contributed so much to their success and the sponsors who make this progress possible. Daniel Mercuri kept our computers running as we edited this work. Robert Rogers and the staff at Taylor & Francis made our work as Editors feasible and enjoyable. We are grateful. Elham A.Ghabbour Geoffrey Davies Editors Boston, Massachusetts December, 2002

Contributors

Dula Amarasiriwardena Professor, School of Natural Science, Hampshire College, Amherst, MA 01002, USA Badia Amekraz Research Scientist, CEA, Nuclear Energy Division & UMR CEA-CNRS-UEVE 8587, Analysis and Environment Laboratory, 91191 Gif-sur-Yvette Cedex, France Robert Artinger Senior Scientist, Forschungszentrum Karlsruhe, Institut für Nuklerare Entsorgung, Postfach 3640, 76021 Karlsruhe, Germany Gérard Bardoux Technical Staff, Laboratoire de Biogéochime Isotopique, Université Pierre et Marie Curie, 4 place Jussieu, 75252 Paris Cedex 05, France Ramon M.Barnes Professor and Director, University Research Institute for Analytical Chemistry, 85 North Whitney St., Amherst, MA 01002–1869, USA Nicole Barre Research Scientist, CEA, Nuclear Energy Division & UMR CEA-CNRS-UEVE 8587, Analysis and Environment Laboratory, 91191 Gif-sur-Yvette Cedex, France Enrique Barriuso Research Scientist, UMR INRA, INAP-G Environnement et Grandes Cultures, 78850 ThivervalGrignon, France Jonathan Bell Undergraduate Student, School of Natural Science, Hampshire College, Amherst, MA 01002, USA Mikhail Borisover Research Scientist, Institute of Soil, Water and Environmental Sciences, The Volcani Center, Agricultural Research Organization, P.O.B. 6, Bet Dagan 50250, Israel Gunnar Buckau Senior Scientist, Forschungszentrum Karlsruhe, Institut für Nuklerare Entsorgung, Postfach 3640, 76021 Karlsruhe, Germany David R.Burris Principal Scientist, Integrated Science and Technology Inc., 433 Harrison Avenue, Panama City, FL 32401, USA Jean-Paul Chopart William T.Cooper Geoffrey Davies Rossane C.DeLapp Sylvie Derenne Marie-France Dignac Jacques Dumonceau Karsten Franke Jean-Pierre Gagné Josemaría García-Mina

Professor, University of Reims Champagne-Ardenne, Dynamique des Transferts aux Interfaces UMR 6107 SC CNRS, UFR Sciences, Moulin de la Housse, BP 1039, 51687 Reims Cedex 2, France Associate Professor, Department of Chemistry and Biochemistry, Florida State University, Tallahassee, FL 32306–4390, USA Professor, Department of Chemistry and Chemical Biology, Northeastern University, Boston, MA 02115–5000, USA Graduate Student, Department of Civil and Environmental Engineering, Vanderbilt University, Nashville, TN 37235, USA Research Scientist, Laboratoire de Chimie Bioorganique et Organique Physique UMR CNRS 7573, ENSCP, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France Research Scientist, Laboratoire de Biogéochime Isotopique, Université Pierre et Marie Curie, 4 place Jussieu, 75252 Paris Cedex 05, France Professor and Group Leader, Groupe de Recherche en Chimie Inorganique, Equipe de Chimie de Coordination aux Interfaces, University of Reims Champagne-Ardenne, UFR Sciences, Moulin de la Housse, BP 1039, 51687 Reims Cedex 2, France Institute of Interdisciplinary Isotope Research, Permoserstrasse 15, 04318 Leipzig, Germany Professor, Laboratoire d’analyses et d’études en géochimie organique, Institut des Sciences de la mer de Rimouski, Université du Québec à Rimouski, Rimouski, QC G5L 3A1, Canada Associate Professor, Department of Chemistry and Soil Science, University of Navarra, and Director, R&D Department, Inabonos-Roullier Group Poligono Arazuri-Orcoyen, C/C. n° 34, 31160 Orcoyen, Spain

xi

Stefan Geyer

Senior Scientist, Environmental Research Center Leipzig-Halle, Hydrogeology Section, 06120 Halle, Germany Elham A.Ghabbour Senior Scientist, Department of Chemistry and Chemical Biology, Northeastern University, Boston, MA 02115–5000, USA Gustavo González-Gaitano Associate Professor, Department of Chemistry and Soil Science, University of Navarra, Pamplona, Navarra, 31080, Spain Ellen R.Graber

Research Scientist, Institute of Soil, Water and Environmental Sciences, The Volcani Center, Agricultural Research Organization, P.O.B. 6, Bet Dagan 50250, Israel P.Ming Huang Professor, Department of Soil Science, University of Saskatchewan, 51 Campus Drive, Saskatoon, SK S7N 5A8, Canada Matthias Hübner Postdoctoral Fellow, Centro Ricerche Ambientali— Montecatini, Via Ciro Menotti 48, I-48023 Marina di Ravenna, Italy Julia Hurrass Doctoral Student, Technical University Berlin, Institute of Environmental Protection, Dept. Environmental Chemistry, Sekr. KF 3, Strasse des 17. Juni 135, D-10623 Berlin, Germany Kristoffer E.N.Jonassen Doctoral Student, Plant Research Department, Risoe National Laboratory, P.O.B. 49, DK-4000 Roskilde, Denmark Jae-II Kim Professor, Forschungszentrum Karlsruhe, Institut für Nuklerare Entsorgung, Postfach 3640, 76021 Karlsruhe, Germany Martina Klu áková Assistant Professor, Institute of Physical and Applied Chemistry, Faculty of Chemistry, Brno University of Technology, Purky′ ova 118, 612 00 Brno, Czech Republic Scott D.Kohl Ph.D., Box 323, Aurora, SD 57002, USA Hermann Kupsch Ph.D., Institute of Interdisciplinary Isotope Research, Permoserstrasse 15, 04318 Leipzig, Germany Claude Largeau Research Scientist, Laboratoire de Chimie Bioorganique et Organique Physique UMR CNRS 7573, ENSCP, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France Eugene J.LeBoeuf Assistant Professor, Department of Civil and Environmental Engineering, Vanderbilt University, Nashville, TN 37235, USA Norman C.Y.Lee Doctoral Student, Department of Chemistry, University of Massachusetts Lowell, Lowell, MA 01854, USA Chen Liu Research Scientist, Kuo Testing Labs, Inc., 337 S. 1st Ave, Othello, WA 99344, USA Ludovic Loiseau Doctoral Student, UMR INRA, INAP-G Environnement et Grandes Cultures, 78850 ThivervalGrignon, France Huizhong Ma Postdoctoral Fellow, Air Force Research Laboratory AFRL/MLQR, 139 Barnes Drive, Tyndall Air Force Base, Florida 32403–5323, USA André Mariotti

Professor, Laboratoire de Biogéochime Isotopique, Université Pierre et Marie Curie, 4 place Jussieu, 75252 Paris Cedex 05, France N.Meems Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/Metabolism, Müggelseedamm 301, 12561 Berlin, Germany Florence Mercier Research Scientist, UMR CEA-CNRS-UEVE 8587, Analysis and Environment Laboratory, 91191 Gif-sur-Yvette Cedex, France Fanny Monteil-Rivera Research Officer, Group of Analytical and Environmental Chemistry, National Research Council of Canada—Biotechnology Research Institute, 6100 Royalmount Avenue, Montreal, Quebec H4P 2R2, Canada Valerie Moulin Research Scientist, CEA, Nuclear Energy Division & UMR CEA-CNRS-UEVE 8587, Analysis and Environment Laboratory, 91191 Gif-sur-Yvette Cedex, France Christophe Moulin Professor and Head, Department of Physico-Chemistry, CEA, Nuclear Energy Division, 91191 Gif-sur-Yvette Cedex, France Martin Müller Postdoctoral Fellow, Technical University Berlin, Institute of Applied Geosciences, Dept. of Applied Geophysics, ACK 2, Ackerstr. 76, D-13355 Berlin, Germany Torben Nielsen Senior Scientist, Plant Research Department, Risoe National Laboratory, P.O.B. 49, DK-4000 Roskilde, Denmark Edward J.O’Loughlin Staff Scientist, Environmental Research Division, Argonne National Laboratory, 9700 South, Cass Ave, Argonne, IL 60439, USA

xii

Andrea Paul Miloslav Peka Stephan Pflugmacher Constanze Pietsch

Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/Metabolism, Müggelseedamm 301, 12561 Berlin, Germany Associate Professor, Institute of Physical and Applied Chemistry, Faculty of Chemistry, Brno University of Technology, Purky′ ova 118, 612 00 Brno, Czech Republic Assistant Director, Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/Metabolism, Müggelseedamm 301, 12561 Berlin, Germany Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/Metabolism, Müggelseedamm 301, 12561 Berlin, Germany

Gabriel Plancque

Research Scientist, CEA, Nuclear Energy Division, Department of Physico-Chemistry & UMR CEA-CNRS-UEVE 8587, Analysis and Environment Laboratory, 91191 Gif-sur-Yvette Cedex, France Torsten Preuer Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/Metabolism, Müggelseedamm 301, 12561 Berlin, Germany Pascal Reiller Research Scientist, CEA, Nuclear Energy Division, Department of Physico-Chemistry, 91191 Gif-sur-Yvette Cedex, France James A.Rice Professor and Head, Department of Chemistry and Biochemistry, South Dakota State University, Brookings, SD 57007–0896, USA Wiete Rieger Diploma Student, Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/Metabolism, Müggelseedamm 301, 12561 Berlin, Germany Doris Rössler Institute of Interdisciplinary Isotope Research, Permoserstrasse 15, 04318 Leipzig, Germany Wolfgang Rotard Professor, Technical University Berlin, Institute of Environmental Protection, Dept. Environmental Chemistry, Sekr. KF 3, Strasse des 17. Juni 135, D-10623 Berlin, Germany David K.Ryan Professor, Department of Chemistry, University of Massachusetts Lowell, Lowell, MA 01854, USA Gabriele E.Schaumann Postdoctoral Fellow, Technical University Berlin, Institute of Environmental Protection, Dept. Environmental Chemistry, Sekr. KF 3, Strasse des 17. Juni 135, D-10623 Berlin, Germany Philippe Schmitt-Kopplin Ph.D., GSF-National Center for Environment and Health, Institute of Ecological Chemistry, Ingoldstädter Landstraße 1, D-85764 Neuherberg, Germany Atitaya Siripinyanond Lecturer, Department of Chemistry, Faculty of Science, Mahidol University, Rama 6 Rd., Rajthevee, Bangkok 10400, Thailand Christian E.W.Steinberg Professor and Director, Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication-Metabolism, Müggelseedamm 301, 12561 Berlin, Germany Alexandra C.Stenson Research Scientist, Vertex Pharmaceuticals, Inc., 130 Waverly Street, Cambridge, MA 02139– 4242, USA Wilfried Szymczak Senior Scientist, Institute of Radiation Protection, GSF-National Research Center for Environment and Health, 85758 Neuherberg, Germany M.Timoveyev Luc Tremblay

Irkutsk State University, Karl Marx 1, 664003 Irkutsk, Russia Postdoctoral Fellow, Department of Biological Sciences and Marine Science Program, University of South Carolina, Columbia, SC 29208, USA Kaijun Wang Doctoral Student, Department of Plant and Soil Sciences, University of Massachusetts, Amherst, MA 01003, USA Claudia Wiegand Postdoctoral Fellow, Institute of Biology, Humboldt-University of Berlin, Chausseestr 117, 10115 Berlin, Germany Klaus Wittmaack Head of Radiation Physics Group, Institute of Radiation Protection, GSF-National Research Center for Environment and Health, 85758 Neuherberg, Germany Manfred Wolf Senior Scientist, Institute of Hydrology, GSF-National Research Center for Environment and Health, 85764 Neuherberg, Germany Baoshan Xing Associate Professor, Department of Plant and Soil Sciences, University of Massachusetts, Amherst, MA 01003, USA Yahya Zegouagh Postdoctoral Fellow, Laboratoire de Biogéochime Isotopique, Université Pierre et Marie Curie, 4 place Jussieu, 75252 Paris Cedex 05, France

xiii

Elke Zwirnmann

Staff Scientist, Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/ Metabolism, Müggelseedamm 301, 12561 Berlin, Germany

Part 1 FRACTIONATION AND CHARACTERIZATION: THE STATEOF-THE-ART

Chapter 1 USE OF RADIOACTIVE TRACERS FOR THE CHARACTERIZATION OF HUMIC AND FULVIC ACIDS IN HIGH PERFORMANCE SIZE EXCLUSION CHROMATOGRAPHY Karsten Franke, Doris Rössler and Hermann Kupsch Institute of Interdisciplinary Isotope Research, Permoserstrasse 15, 04318 Leipzig, Germany

1.1. INTRODUCTION Humic substances (HSs) influence many geochemical and environmental processes in soil [1–5]. As ubiquitous major components of soil organic matter, their regulation of metal interactions in soils is unquestioned. Depending on geochemical parameters, HSs can act as geochemical barriers or contribute to the non-retarded migration of metals. An important aspect in understanding these processes is the conformational nature of HSs [6,7]. Depending on circumstances, HSs can form refractory colloids [8], precipitate via aggregation or remain in solution as negatively charged complexes. Therefore, the effect of molecular size and shape of HSs must be taken into account [9]. Much effort has been devoted to the use of size exclusion chromatography (SEC) to answer some fundamental questions about HSs [1,10]. Overcoming artifacts caused by secondary interaction with the column material, high performance size exclusion chromatography (HPSEC) has become a predictive tool for investigation of humic substances [11–14]. Nevertheless, common detection methods (UV-absorption, fluorescence, ICP-MS) have different limitations, such as detection limits, signal quenching and problems caused by online coupling. This prompted us to investigate the possibility of using radioactive tracers in HPSEC. We adapted radiolabeling techniques to label HSs, leading to very good agreement of the radiochromatograms obtained with the results of classical detection methods. In addition to labeling of the carbon backbone of HSs (with 131I), our interest was focused on the interaction of humic substances with aluminum. In the absence of a suitable aluminum isotope, we used 111In as a surrogate for aluminum in our investigations. Table 1.1 Elemental compositions and amounts of humic and fulvic acids in soil Depth 0 cm–10 cm 10 cm–20 cm 20 cm–30 cm 30 cm–40 cm 40 cm–50 cm

Amount [mg/g]

HA Elemental composition

FA Elemental composition

HA

FA

%N

%C

%H

%S

%N

%C

%H

%S

12.3 15.1 12.1 8.4 2.2

2.0 2.3 2.2 2.3 1.0

4.5 3.8 3.8 3.4 3.1

41.4 33.8 34.5 33.5 32.5

5.0 4.5 4.5 4.2 3.8

1.3 0.8 0.8 0.7 0.8

3.4 3.6 3.3 3.5 3.3

41.0 40.9 42.7 41.2 41.4

4.3 4.2 4.4 4.3 4.4

0.5 0.5 2.0 0.7 0.5

1.2. MATERIALS AND METHODS 1.2.1. Humic Substances Humic substances were extracted from different soil profiles at depths of 0–50 cm. The sampling site is located in Schlema/ Alberoda, Saxony, Germany. The soil samples were taken with a 10 cm spacing. A detailed description of the sampling area can be found in [15]. The humic acids (HAs) and fulvic acids (FAs) were obtained following the IHSS procedure [16]. The elemental compositions and percent yields of the humic and fulvic acids are given in Table 1.1.

RESULTS AND DISCUSSION |

3

Figure 1.1 Reaction scheme of the 131I labeling of humic and fulvic acids

Figure 1.2 Decay scheme of 131I (a) and 111In (b) (simplified plot of the ′ -rays used)

1.2.2. Radiolabeling 131I (t =8.041d) was used for labeling of the HAs and FAs. In the reaction, the aromatic moieties of HS molecules are 1/2 labeled via electrophilic H J-substitution (Figure 1.1). Differing from the known iodination method for humic substances [17,18], 1,3,4,6-tetrachloro-3a,6a-diphenylglycouril was used as the oxidizing agent. This so-called Iodogen method is widely known in radiopharmaceutical chemistry [19]. The main advantage of the Iodogen method is the easy separation of the oxidizing agent from the sample. The radiochemical yields were measured and the stability of the labeling was proved by ultracentrifugation, dialysis and precipitation experiments. Different yields were observed depending on the labeling method and the source of HS. The yields of the labeling procedure varied between 65% and 80%. The known photo-susceptibility of iodination must be considered in following experiments. In addition to iodination, the complexing functional groups of HSs have been used for 111In labeling of the HA and FA. The labeling was performed with no-carrier-added [111In]InCl3 (t1/2=2. 81d) (Nycomed-Amersham) via complexation.

1.2.3. High Performance Size Exclusion Chromatography HPSEC was used to determine the molar mass distribution of the humic and fulvic acids (10 mg/L). The instrument was equipped with a TSK gel column (TSK-G3000PWXL, 300 mm×7.8 mm) with a size fraction range of 0.5–800 kDa. A flow of 0.5 mL/min and a sample volume of 25 µL were used for chromatography. The mobile phase consisted of 0.02 M KCl in MilliQ water with 0.05 M tris-buffer (pH 8). Sodium polystyrene sulfonates (Mn: 4.3, 8.6, 17.4 and 33.8 kDa), blue dextran (Mn: 100 kDa) and acetone were used as standards. A high purity Ge-detector was used to measure the radiochromatograms. Due to their excellent energy resolution, the following gamma rays were used for detection (131 I: ′ 1=364 keV, ′ 2=637 keV, 111In: ′ 1=171 keV, ′ 2=245 keV, Figure 1.2a,b). 1.3. RESULTS AND DISCUSSION The coincidence of the radiochromatogram of the [131I]HA and the UV absorbance chromatogram was used as a criterion for testing the iodination of HS. These signals are compared in Figure 1.3. The observed radiochromatogram is in clear correspondence with the UV signal, which means that all constituents of the HS are labeled in similar proportions. The sample is completely eluted in the size fraction range 0
4

| CHAPTER 1: THE USE OF RADIOACTIVE TRACERS

Figure 1.3 Comparison of the [131I]HA radiochromatogram and the UV-absorbance of the humic acid

Figure 1.4 HPSEC radiochromatograms of fulvic acids (FA) from various depths of soil profile; comparison of the simultaneously measured [131I]FA and [111In]FA radiochromatograms

Figure 1.5 HPSEC radiochromatograms of humic acids (HA) from various depths of soil profile; comparison of the simultaneously measured [131I]HA and [111In]HA radiochromatograms

chromatogram implies an increased selectivity for the humic and fulvic acid compounds with a lower nominal molar mass for the 111In labeling. This agrees with the results of investigations of the coagulation-flocculation behavior of natural organic matter (NOM) with Al and Fe [20–22]. These 27Al-NMR and EXAFS studies show a decrease of molecular size via coagulation/flocculation induced by metal interaction with NOM. Due to the use of tracer amounts of In (n.c.a. [111In]InCl3), no change in the [131I]FA/ HA chromatogram could be observed. The nature of the additional peaks at higher retention times is not clear. Possibly, [111In] In(OH)x species are formed during the chromatographic separation due to the high pH of the mobile phase.

RESULTS AND DISCUSSION |

5

1.4. CONCLUSIONS Radioactive double labeling can be a useful tool in HPSEC. This technique complements other online coupling techniques like HPSEC-ICP-MS/OES. The extremely low detection limits (online/offline) of the radiomarkers and their easy measurement offers alternatives in experiments focused on natural concentration levels and to overcome mobile phase problems caused by online coupling. The observed selectivity of low molar mass fractions of the humic and fulvic acids agrees with the results found by 27Al-NMR and EXAFS. Because of the interaction with trivalent ions, a shift of the nominal molar mass to lower values is observed. Further investigations that focus on the kinetics of metal interaction are required. Especially needed are studies of molecular interactions and competition reactions of humic substances. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20 21 22

Aiken GR, McKnight DM, Wershaw RL, MacCarthy P eds. Humic substances in soil, sediment, and water. New York: Wiley, 1985. Davies G, Ghabbour EA eds. Humic substances: Structures, properties and uses. Cambridge: Royal Society of Chemistry, 1998. Greenland DJ, Hayes MHB eds. The chemistry of soil constitutents. Chichester: Wiley, 1978. Senesi N, Miano TM eds. Humic substances in the global environment: Implications for human health. Amsterdam: Elsevier, 1994. Stevenson F. Humus chemistry: Genesis, composition, reactions. New York: Wiley, 1982. MacCarthy P. Soil Sci., 2001; 166:738–751. Piccolo A. Soil Sci., 2001; 166:810–832. Kim JI, Buckau G, Zhuang W. Mater. Res. Soc. Symp. Proc., 1987; 84:747ff. Tate R. Soil Sci., 1999; 164:775–776. Hayes MHB, MacCarthy P, Malcolm RL, Swift RS eds. Humic substances II. In search of structure. New York: Wiley, 1989. Perminova IV. Soil Sci., 1999; 164:834–840. Ceccanti B, Calcinai M, Bonmati-Pont M, Ciardi C, Tarsitano R. Sci. Total Environ., 1989; 81/82:471–479. Conte P, Piccolo A. Chemosphere, 1999; 38:517–528. Piccolo A, Conte P, Cozzolino A. Eur. J. Soil Sci., 1999; 50:687–694. Franke K, Rößler D, Gottschalch U, Kupsch H. Isotopes Environ. Health Stud., 2000; 36:223–239. Sparks DL, Methods of soil analysis. Part 3. Chemical methods. Madison: Soil Sci. Soc. Am., 1996. Warwick P, Mason I, Hall A, Holmes R. Radiochim. Acta, 1994; 66/67: 427–432. Warwick P, Carlsen L, Randall A, Lassen P. Chem. Ecol., 1993; 8:65–80. Fraker P, Spek J. Biophys. Biochem. Res. Commun., 1978; 80:849–856. . Masion A, Vilge-Ritter A, Rose J, Stone WEE, Teppen BJ, Raybacki D, Bottero JY. Environ. Sci. Technol., 2000; 34:3242–3246. . Vilge-Ritter A, Rose J, Masion A, Bottero JY, Laine YM. Coll. Surf., 1999; 147: 297–308. . Lippold H, Rößler D, Kupsch H. Appl. Radiation and Isotopes, submitted.

Chapter 2 INTERPRETING CAPILLARY ELECTROPHORESIS— ELECTROSPRAY/MASS SPECTROMETRY (CZE-ESI/MS) OF SUWANNEE RIVER NATURAL ORGANIC MATTER (NOM) Philippe Schmitt-Kopplin GSF-Research Center for Environment and Health, Institute for Ecological Chemistry, Ingolstädter Landstrasse 1, 85764 Neuherberg, Germany 2.1. INTRODUCTION Natural organic matter (NOM) is a highly complex, polydisperse mixture of naturally occurring polyelectrolytic organic compounds whose average residence time in the hydrosphere is substantially longer than that of ordinary biomolecules such as carbohydrates and proteins [1]. NOM includes, but is not limited to, aquatic humic and fulvic acids, and much that is known about NOM is based on studies of those humic substances. The chemical and biological properties (rates and mechanisms of transformation, bioavailabilities, toxicities) of other natural and xenobiotic constituents of natural waters are often substantially affected by their interactions with NOM. Those interactions, in turn, are influenced by the physicochemical properties of NOM. Most distinctly, NOM is acidic. Its acidic functional groups contribute to the acid-base balance of natural waters, form complexes with trace metal cations of environmental and biological importance, and dictate its ligand-based adsorption to metal (hydr)oxides. In natural waters, the net average charge of NOM is determined by dissociation of acidic functional groups and complexation of other cations by those functional groups, and all of these reactions are controlled by pH and ionic strength. The role of ionic strength, in particular, has been difficult to parameterize properly, mainly because the actual distribution of charge (the real proportions of singly-, doubly-, and multiply-charged ions) in NOM is not known. Capillary electrophoresis (CE) is a relatively non-perturbing method that allows the electrophoretic separation of small molecules, polymers, macromolecules, col loids up to nanoparticles, bacteria and viruses in various modes of separation using non-aqueous or aqueous buffers with coated or uncoated capillary columns. The combination of CE with mass spectrometry provides a very versatile analytical system. Furthermore, the performance of electrospray ionization MS (ESI-MS) has considerably improved, allowing detection limits down to low picomole for single components (the femtomole level is quite promising). Additionally, the small volume of the analyte used in CE experiments has made possible the analysis of cellular contents and even the analysis of single cells in proteomic studies. The present objectives were the hyphenation of a capillary electrophoresis system to an electrospray ionization mass spectrometer (CE-ESI/MS) and its use for the characterization of complex mixtures exemplified by Suwannee River natural organic matter (NOM). For polydisperse mixtures such as NOM, hyphenation could provide insight into the distribution of charge density at a specified pH and ionic strength. For the correct interpretation of the CE-ESI/MS signals of complex mixtures such as natural organic matter, an independent understanding of the behaviour in both CE and ESI/MS is needed. The different steps required to reach these goals are illustrated in Figure 2.1 and can unfortunately not all be presented in this paper: i. Electrophoretic mobility is the driving force in the selectivity of the separations. The electrophoretic mobility scale concept was presented elsewhere [2,3] with the resulting qualitative and quantitative improvements of data reduction, as illustrated with applications to structurally unknown mixtures [2,4]. ii–v. Semi-empirical models needed to be developed for the interpretation of the electrophoretic mobility and mobility distributions of small molecules and macromolecules, based on their charge and size distributions [5]. These models allow the prediction of electrophoretic mobility of molecules in capillary zone electrophoresis (CZE) from their chemical structure and constitute an easy and robust screening tool when developing new analytical separation methods. Reversing this approach, chemical and physical characteristics could be simulated from the changes in effective mobility of unknown molecules by systematically selecting and varying the separation buffers [6–8]. In such characterisations the capillary is considered as a reactor in which specific chemical reactions can be analyzed and modelled [9,10]. This approach is used for the characterization of colloidal NOM, giving information on their pH-dependant combined charge and size evolution (nanogel properties), using capillary zone electrophoresis, capillary gel electrophoresis and free flow electrophoresis (FFE). The latter is a very promising preparative separation tool based on scaling up of CZE and is used in proteomics. Free flow

INTRODUCTION |

7

Figure 2.1 Different experimental steps needed to be investigated to interpret the CE-ESI/MS characterisation of complex mixtures such as NOM

electrophoresis allows a fractionation based on charge density alone and further characterisation of structure and function [11]. vi. A micro-electrospray ionization (µESI) platform was set up prior to an ion-trap MS. This system was optimized in the flow-injection mode (FI-ESI/MS) to investigate the influence of experimental parameters on the measured m/z distributions of different bulk NOM factions. Results on the FI-ESI/MS characterization of selected FFE fractions are presented elsewhere [11]. vii. The last requirement is the hyphenation of capillary electrophoresis to mass spectrometry via the micro-ESI-platform and its optimization for routine use [12]. This instrumental setup is an online possibility and a miniaturization of the FFE separation followed offline by FI-ESI/MS to give complementary information on the polymer and colloidal behaviour of natural organic matter. This chapter shows the possible hyphenation of CZE and µESI/MS (CE-ESI/MS), recently developed and routinely used in our group for the purpose of identification and quantification of natural compounds and the characterization of inseparable mixtures such as NOM. To be able to validate and interpret the obtained electromassograms (combined electropherogram and mass spectral information in a contour plot), theoretical and experimental approaches to the CZE-ESI/MS of natural organic matter are considered. 2.2. MATERIALS AND METHODS CE-ESI/MS measurements were performed with a Beckman P/ACE 5510 capillary electrophoresis system (Beckman Instruments, Fullerton, CA), equipped with on-column diode array detection, an auto sampler and a power supply able to deliver up to 30 kV coupled to an ion trap mass spectrometer Finnigan LCQ duo (Thermo-Quest, San Jose, CA) with a selfoptimized micro-electrospray ionization mode (ESI) interface [5]. On-line coupling of the CE instrument to the mass spectrometer detector was achieved with a commercial coaxial sheath liquid interface (Thermo-Quest), which was positioned at the MS entrance (Figure 2.2). The CE column outlet was set at the same height as the CE-ESI-MS probe to prevent siphoning. To maintain stable electrospray, a 20 mm portion of the polyimide coating was removed from the outlet end of the capillary. This procedure was effective in providing better mixing characteristics at the probe tip. A CCD camera allowed controlling the formation of the electrospray jet (Figure 2.3).

8

| CHAPTER 2: INTERPRETING CAPILLARY ELECTROPHORESIS

Figure 2.2 (left) Sheath liquid interface chosen for the hyphenation of capillary electrophoresis with mass spectrometry via the electrospray device.

Figure 2.3 (right) Observed Taylor cone (CCD camera) and illustrated droplet formation

CE separations were performed with a non-coated, open fused-silica capillary from Polymicro Technologies Inc., Phoenix, AZ (50 m i.d., 80 cm total length, 20 cm to the detector). The front portion (about 60 cm) of the capillary was liquid cooled and set at 30°C, and the remaining portion (about 20 cm) was exposed to ambient air conditions. Instrument control, data acquisition and data processing were carried out using the GOLD software package for CE and the Xcalibur data system version 1.1 software for ESI-MS. CE-MS was carried out by applying 30 kV to the anode and 4.5 kV to the cathode. The cathode was the electrospray tip on the ESI interface. Hydrodynamic injection was 5 or 10 sec. The sample HS concentration had no significant influence on the peak-average electrophoretic mobility (AEM, µp). Day to day changes in migration times occurring because of relative changes in the electroos motic flow (different capillary surface conditions) could be limited by washing the capillary with 0.1 M NaOH for 2 min between each run. Buffers used were volatile and non-complexing (ammonium acetate mixed with ammonium carbonate) at pH 7. The raw electrophoretic CZE-UV data were treated by use of the self-designed program “GelTreat”, originally developed by Perminova and Kudryavtsev. To make it applicable for the treatment of electropherograms in mobility scale, the designer modified the software to allow transformation of the x-axis of the initial electropherogram from the time- into µ-scale [3]. The transformation in µ-scale is certainly possible with classical data processing programs, but this self-designed program allows the complete analysis of a data set much more rapidly. 2.3. CAPILLARY ZONE ELECTROPHORESIS OF NOM More than 60 papers on the use of CE with polydisperse and heterogeneous NOM and humic materials can be found in the literature [13]. Many pitfalls exist in the interpretation of the raw data because capillary electrophoresis is a completely different analytical way of thinking than classical chromatographic separation methods [4]. To be used as a structure characterization tool, it especially needs the use of non-reactive buffer systems (such as non-zwitterionic or non-borate buffers) in which the electrophoretic velocity of the NOM is governed only by the structures of the molecules [9,14,15]. The velocity per unit of field strength (effective mobility µ) of single ions in CZE can be linearly related to the charge to size ratio (Offord’s equation: µ= Z/M2/3) [16] of the analytes at the given separation buffer pH, and from a pH titration in CZE the corresponding pKa values can elegantly be experimentally derived [6]. Figure 2.4 gives an example of a CZE pH titration of the Suwannee River NOM in 25 mM buffer systems where the changes in effective mobility distributions are visualised in a contour plot of pH versus mobility. The eletropherograms can be related to the changes in both charge and size distributions under the chosen separation buffers conditions. The conversion of the time based acquisition data into effective mobility scaled domains is the prerequisite for a direct comparison of the distributions in mobility over the wide pH range [3].

INTRODUCTION |

9

Figure 2.4 CZE pH-titration of Suwannee river River NOM from pH 3 to 11.5 (UV, Abs. at 254 nm)

The mobility data can be put alongside the potentiometric titration data (charge) to follow the changes of the relative sizes with pH (polymer gels phase transitions). Depending on the studied NOM (origin, structure), single molecules as well as molecular associates and macromolecules can be observed and analyzed by varying the experimental conditions [5]. The separation capillary becomes comparable to a chemical reactor in which reactions can be induced between the constituents of the sample and the separation buffer, leading to information on charge (pKa) [6], hydrophobicities (Log Kow) [7], size [5], and reactivity towards organic (Log Koc) [8] or inorganic constituents (pollutants, essential elements…) [17]. The limitations of this one-dimensional CZE approach are in the interpretations of the distribution of the electrophoretic distributions. In CZE, a signal at a given mobility is only an indication of a given charge to size ratio, neither of the effective charge nor of the size alone. What are the correspondences between the observed distributions in electrophoretic mobility and the possible distributions in charge and in size (mass) within the mixtures? 2.4. A THEORETICAL SEMI-EMPIRICAL APPROACH TO CHARACTERIZING POLYDISPERSE ELECTROLYTE SYSTEMS Heterogeneity in charge and in size within a given system is the source of effective mobility distributions measured with NOM. The resulting electrophoretic heterogeneity was found to be a major source of observed peak broadening when analysing polymeric materials [18], and liposomes [19]. High-resolution CE analysis allows the analysis of oligomeric distribution of polydisperse polymers [20] but often reaches its limits in separation possibilities with the different classes of com ponents that are affected by mobility heterogeneity: 1) carbohydrate-based natural polymer mixtures (hyluronic acid mixtures) [21]; 2) cellulose based materials [22]; 3) fuel [23]; 4) ligninosulfonates [24–26 and refs. therein]; 5) natural colloids [3,4,18]; 6) ferrofluids [27]; 7) ionic and non-ionic surfactants [28–30]; and 8) ionic and non-ionic synthetic polymers [31–33]. An example of a mobility distribution pattern in a polydisperse mixture is shown in Figure 2.5, which shows a catechol oxidation polymer [18,34] separated in a 25 mM carbonate buffer at pH 11.4. At this pH phenolic groups are ionized and even the different oligomers are differentiated (…n−1, n, n+1…), showing mobility patterns like those observed with polymer oligomers. The total covered mobility range of the catechol polymer is narrow, probably resulting from a narrow charge range of the dissolved analytes. It is also evident that even though the distribution in mobility is gaussian like, the raw data are asymmetric (see papers on mobility transformation in CE [3,15]).

10

| CHAPTER 2: INTERPRETING CAPILLARY ELECTROPHORESIS

Figure 2.5 Catechol polymer (in time scaled electropherogram and corresponding mobility scale) showing the distribution of the ionized oligomers at pH 11.4

How can such a mobility profile be described in terms of distributions of charges and mass in the mixtures? Is it possible to differentiate different populations of individual molecules having different mass and charge distributions within polydisperse mixtures? Some possible answers to these questions will be presented below. 2.4.1. Simulation of the Electropherograms of Mixtures; Limitations of the Interpretation We approached the description of these polydisperse systems by considering populations comprising n2 individuals possessing n charge (Zi) and n mass (Mj) possibilities, both with their own gaussian or asymmetric distributions. Thus, for all individuals of mass (Mj) existing with a given distribution there are corresponding n charge cases with their own distribution. We considered different scenarios of distributions in charge and mass (monodisperse, polydisperse, asymmetric) and their impact on the respective shapes of mobility distribution using Offord’s equation [16], which relates the charge and size of each molecule to its effective mobility [5]. For each population, a frequency matrix was calculated from the gaussian distributions in charge Zi and mass Mj (also considered as frequencies) so that: (2.1) where AZi and BMj are the respective frequencies in charge and mass. The mobility window was restricted from 0 to −0.045 cm2/Vmin, corresponding to a mobility range of substances like NOM and fulvic and humic acids, which exhibit charge densities such as between these of benzoic acid up to pyromellitic acid as a function of pH [35]. The corresponding mass and charge range were chosen from 0 to 2000 and 0 to −10, respectively. The shapes of the distribution profiles of the mass M (upper horizontal row) and the charges Z (left vertical row) are shown with the crossing intersection corresponding to their resulting mobility distribution profile. We chose different possible distribution profiles in charge and mass and their resulting

INTRODUCTION |

11

Figure 2.6a (top) Resulting mobility distributions from charge and mass distributions with constant width and constantly increasing maximum values.

mobility distribution shapes (monodisperse, polydisperse, asymmetric). Here it is interesting to concentrate on the shape of the simulated mobilities presented in Figures 2.6a–c. All considered combinations in Figure 2.6a–c demonstrate that the distribution of the charge essentially governs the resulting distributions of mobility. An asymmetric mobility silhouette results from monodisperse charge combined with polydisperse mass or from the combined asymmetric distribution of both charge and mass, where the charge shape dominates the resulting mobility profile. Thus, a strongly pronounced mobility figure (asymmetric, thin or broad) within the real electrophoretic data would essentially originate from the respective identical charge shape. We could continue this process to calculate the mobility distributions generated by successively mixing different populations having contrasting individual mass and charge ranges (e.g., low molar versus high molar mass, low charged versus highly charged) combined with different distributions (symmetric, sharp, wide, asymmetric). However, when trying to access the experimental mobility distribution in terms of charges and sizes, one rapidly reaches interpretive limits: charge and size are not immediately differentiable. The added mass spectroscopic detection should add the required dimension to enable a distinction in size when analysing polydisperse samples. 2.4.2. Simulation as Mobility/Mass Distributions Contour Plots Adding the mass information at each mobility step (such as is done with MS detection) allows an additional dimension of the information that can be illustrated as contour plots containing the effective mobility dimension in the x-axis (Z/M2/3) with the corresponding MS dimension on the y-axis (m/z). The calculations above can be integrated with the advantage that individual molecules with different mass and mass distributions can be differentiated and described in the so-called electromassograms. For this, let us first select different populations of individual molecules possessing a given charge and mass distribution (like in Figure 2.6a–c). In the first example (Figure 2.7a), we consider a wide distribution in mass with narrow charge dispersity. In the mobility contour plot Z/M2/3=f(m/z), a narrow typical trace over a wide m/z range can be observed. A representation of the electropherograms using restricted mass ranges (<400, 400<<800, 800<<1200, 1200<<1600 and 1600<<2000Da) shows lower polydispersity in mobilities and the systematic position of high mass in the low mobility range and the small molecules in the high mobility area (Figure 2.7a). This mobility distribution is very realistic and is close to those known for humic substances that were separated by size with ultrafiltration and measured in CZE [3]. By taking the opposite extreme situation, which is a wide charge distribution with a narrow mass dispersity, the obtained mobility is symmetrical and very wide (Figure 2.7b). In this situation, the low dispersity in mass cannot be seen from the electropherogram. Again, in the electromassogram this difference can immediately be evaluated. As expected, no differences can be seen in the restricted mass electropherograms representation (Figure 2.7b). Asymmetric distribution in mass can be

12

| CHAPTER 2: INTERPRETING CAPILLARY ELECTROPHORESIS

Figure 2.6b (bottom) Resulting mobility distributions from charge and mass distributions with variable width and constant maximum values

Figure 2.6c Resulting mobility distributions from asymmetric charge and/or mass distributions.—asymmetric Asymmetric distributions were built by summing three gaussian distributions

evaluated from both the electromassograms and the mass-selective electropherograms (Figure 2.7c). In the electromassogram the form is again a concave oblong ellipsoid. To reach other distribution patterns, such as highly charged high molar mass and low charged low molar mass substances, a combination of different populations having their own mass and charge distributions has to be considered (Figure 2.7d).

INTRODUCTION |

13

Figure 2.7a (top) Resulting contour plot and electromassograms from wide mass and narrow charge distribution

Figure 2.7b (bottom) Resulting contour plot and electromassogram from narrow mass and wide charge distribution

2.5. EXPERIMENTAL APPROACH: INTERPRETING THE CZE-ESI/MS OF NOM 2.5.1. An Application Example in Negative Electrospray Ionization Mode with Model Compounds Some of the first samples investigated in our group with this technique were phenolic acids. As in classical CZE, they provide a means (external standards) to evaluate the size of the electrophoretic mobility window in series of measurements as a function of the buffer quality and capillary surface: benzoic acid, phthalic acid and trimellitic acid were used for this calibration. UV detection was monitored on the capillary after 20 cm. A continuous extrapolation of the time the sample takes

14

| CHAPTER 2: INTERPRETING CAPILLARY ELECTROPHORESIS

Figure 2.7c (top) Resulting contour plot and electromassogram from asymmetric mass and narrow charge distribution.

Figure 2.7d (bottom) Resulting contour plot and electromassograms by combining 4 populations of individuals having charge and mass distributions with different increasing charge dispersity and charge to mass ratio

to arrive at the capillary end and give a signal at the MS is thus possible. This was systematically done to check the total mobility window and eventual problems due to changes in electroosmotic flow (EOF) leading to higher migration times. Figure 2.8 illustrates a separation of these 3 standards in a 10 mM ammonium carbonate buffer at pH 9.1. The total ion current (TIC in negative ionization mode) and the different mass traces corresponding to the analyte mass [M-H]− are shown as relative abundances. The heated capillary was set to 170°C. Phthalic acid decomposes into benzoic acid due to known processes. Trimellitic acid is also subject to decarboxylation but to a lesser extent. The degradation of analytes in the ESI can thus be differentiated from their degradation in the sample solution or during the separation because the mass traces show the ions at the same migration times. An elegant way to represent the CE-ESI/MS signals as a function of time is the contour plot (Figure 2.9) as presented theoretically in section 2.4. The contour plot combines the charge to size information within the x-axis and the mass to charge

INTRODUCTION |

15

Figure 2.8 Negative Mode CE-ESI/MS of a mixture of benzoic acid (m/z 121), phthalic acid (m/z 165) and trimellitic acid (m/z 209); total ion current (top) and selective mass traces for single components

Figure 2.9 CE-ESI/MS contour plot representation of a separation of selected phenols and phenolic acids in negative mode.

information in the y-axis. The separation buffer, allowing an ionization of the analyte and thus governing its migration time, gives the first restriction for the position of the “cross peak” in the plot. The second restriction is given by the m/z ratio obtained during the ESI process, and it may or may not be related to the solution charge. The general rule is that small molecules will generally be singly charged and higher molar mass compounds will be multiply charged, keeping a charge through ESI as a function of their charge in the CZE separation buffer and the experimental ESI conditions [5]. This approach is certainly of greater advantage than UV-Vis detection when analyzing such phenolic acids from natural waters [36] or from beverages [37,38]. Our separation system was optimized from the migration times so that the total mobility range of natural organic matter could be analyzed in one run. Due to the high data acquisition rate, and the enormous number of data points from an electrophoretic run (about 20 min) over the mass range from 100 to 2000 Da, any transformation of the whole data set in mobility scale was not possible. The software would have to be changed to handle contour plots in the mobility scale. All further contour plots shown are thus expressed in time scale (raw data).

16

| CHAPTER 2: INTERPRETING CAPILLARY ELECTROPHORESIS

Figure 2.10: Experimental CE-ESI/MS of Suwannee River NOM at pH 7 and simulated M/Z2/3 versus m/z contour plot showing similarities with the real world

2.5.2. CZE-ESI/MS in Negative Electrospray Ionization of Suwannee River NOM When using CE-ESI/MS to characterize unknown mixtures such as humic materials, one should be aware of any artifactual signals that can originate from interactions during the separation (CZE), the ion formation (ESI) and the transfer to the mass detection system. We will not present any results on the effect of experimental ESI and MS conditions (temperature, adduct formation…) on the obtained m/z distributions, nor will we focus on possible artefacts in CZE that were discussed in the literature on many occasions [14]. Herein we want to present, in light of the theoretical approach above, some results with Suwannee River NOM, analyzed at pH 7 in optimized CE-ESI/MS conditions. The independent validation of the CE-ESI/MS method could be achieved previously with preparative free flow electrophoresis (FFE) combined with off-line flow injection µ-ESI/MS. The setup and optimization of the microelectrospray ionization and micro-ESI-interfacing conditions (temperature, distances, solution conditions) are not presented herein either [5]. Figure 2.10 shows the CE-ESI/MS electromassogram obtained and a corresponding interpretation possibility from the already presented theoretical contour plots. Within our instrumental limits (m/z from 100 to 2000), the total ion current trace obtained with CE-ESI/MS (superimposable on the mass distribution) shows three distinct zones corresponding to two low molar mass fractions and to an unde fined zone with a distribution in both Z/M and m/z. From theoretical models presented above the oblong shaped pattern obtained is mainly governed by a narrow distribution of charge within a given size (mass) distribution. Figure 2.11 illustrates in more detail the mass restricted electrophoretic trace of Suwannee River NOM analyzed with CZE-ESI/MS at pH 7 and the corresponding theoretical traces (from contour plot data in Figure 2.10). The low molar mass fraction can easily be distinguished from the higher molar mass one and is currently further analyzed from the MS pattern: some structural details could not be distinguished when analyzing the sample with CE-UV. The observed polydispersity in effective mobility decreases with higher molar mass. The same tendency is observed when separating in alkaline conditions where phenolic groups also contribute to the mobility profile. High molar mass fractions

INTRODUCTION |

17

Figure 2.11 CE-ESI/MS m/z traces for Suwannee River NOM at pH 7 and corresponding simulation of different mass ranges and their evolution within the humic hump

contribute mostly to the lower mobility fraction at this separation pH. The problem here, however, is still to be able to differentiate multiplicity in charge, which is not trivial to achieve in complex unresolved mixtures. Only one example was given here to show the possibilities of CE-ESI/MS for the characterization of natural organic matter, which already are well established in proteomic types of studies. As we deal with mixtures, we always should be aware that we only see what is possible to be seen from the experimental set up; influences of different separation pH, ESI and MS conditions on the electromassograms with humic material that differ in structure will be presented in the near future. Further studies are needed at different pH to understand the real charge distribution within the different molecular sizes present.

18

| CHAPTER 2: INTERPRETING CAPILLARY ELECTROPHORESIS

Figure 2.12 HumArt; Terragen V0.8.11 Landscape generated based on the data from the CE-ESI/MS contour plots of Suwannee River NOM (same conditions as in Figures 2.10 and 2.11)

2.6. CONCLUSIONS The highly complex, polydisperse and irregular structures of HSs pose specific and yet unresolved challenges in the field of structural and quantitative analysis as well as in raw data interpretation. The possible systematic approach in analytical capillary electrophoresis and preparative free flow electrophoresis combined with online/offline mass spectrometry opens new possibilities in understanding the chemical structure, solution behaviour and functions of natural organic matter. We still have a long way to go to understand humics, and during that quest we will certainly discover many new lands: just imagine having to climb and name each mountain-peak shown in the HumArt depicted in Figure 2.12! ACKNOWLEDGEMENTS Thanks are due to H.Neumeir, A.Wüst, B.Look and B.Kopplin of the Institute for Ecological Chemistry—GSF, for skilful assistance. Many persons contributed efforts and discussions to our successful projects and are thanked herein: J.Junkers, F.Menzinger, Dr. N.Hertkorn, Prof. A.Kettrup, Prof. M.Perdue and Prof. J.Chen. It was a special honour to exchange these concepts and results with Prof J.Hedges (†). Part of this work was done within the projects GIF Young Scientist N°1084–307. 2/2000. REFERENCES 1. 2.

3. 4. 5.

6.

Aiken GR, McKnight DM, Wershaw RL, MacCarthy P eds. Humic substances in soil, sediment, and water. New York: Wiley, 1985. Schmitt-Kopplin P, Garmash AV, Kudryavtsev AV, Menzinger F, Perminova IV, Hertkorn N, Freitag D, Petrosyan VS, Kettrup A. Quantitative and qualitative precision improvements by effective mobility-scale data transformation in capillary electrophoresis analysis. Electrophoresis, 2001; 22:77–87. Schmitt-Kopplin P, Garmash AV, Kudryavtsev AV, Perminova IV, Hertkorn N, Freitag D, Kettrup A. Mobility distribution description of synthetic and natural polyelectrolytes with capillary zone electrophoresis. J. AOAC Internat., 1999; 82:1594–1603. Schmitt-Kopplin P, Menzinger F, Freitag D, Kettrup A. Improving the use of CE in a chromatographer’s world. LC-GC Europe, 2001; 14:284–388. Schmitt-Kopplin P. Comprehensive approaches for the characterization of polydisperse natural organic matter (NOM) with capillary electrophoresis-electrospray ionization/mass spectrometry (CE-ESI/MS), in Habilitationschrift, Wissenschaftszentrum Weihenstephan für Ernährung, Landnutzung und Umwelt. Munich-Weihenstephan, Germany: Technical University of Munich, 2002:250. Schmitt-Kopplin P, Poiger T, Simon R, Garrison AW, Freitag D, Kettrup A. Simultaneous ionization constants and isoelectric points determination of 12 hydroxy-s-triazines by capillary zone electrophoresis (CZE) and capillary elec trophoresis isoelectric focusing (CIEF). Anal. Chem., 1997; 69:2559–2566.

INTRODUCTION |

7. 8.

9. 10. 11.

12.

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

27. 28. 29. 30. 31. 32. 33. 34. 35.

19

Freitag D, Schmitt-Kopplin P, Simon R, Kaune A, Kettrup A. Interactions of hydroxy-s-triazines with SDS-micelles by micellar electrokinetic capillary chromatography (MEKC). Electrophoresis, 1999; 20:1568–1577. Schmitt P, Trapp I, Garrison AW, Freitag D, Kettrup A. Binding of s-triazines to dissolved humic substances: Electrophoretic approaches using affinity capillary electrophoresis (ACE) and micellar electrokinetic chromatography (MEKC). Chemosphere, 1997; 35:55–75. Schmitt-Kopplin P, Hertkorn N, Garrison AW, Freitag D, Kettrup A. Influence of borate buffers on the electrophoretic behavior of humic substances in capillary zone electrophoresis. Anal. Chem., 1998; 70:3798–3808. Schmitt-Kopplin P, Fischer K, Freitag D, Kettrup A. Capillary electrophoresis for the simultaneous separation of selected carboxylated carbohydrates and their related 1,4-lactones. J. Chromatography A, 1998. 807:89–100. Junkers J, Schmitt-Kopplin P, Munch JC, Kettrup A. Up-scaling capillary zone electrophoresis (CZE) separations of polydisperse anionic polyelectrolytes with preparative free flow electrophoresis (FFE) exemplified with a soil fulvic acid. Electrophoresis, 2002; 23:2872–2879. Bianco G, Schmitt-Kopplin P, De Benedetto G, Cataldi TRI, Kettrup A. Determination of glycoalkaloids and relative ′ -glycones by non aqueous capillary electrophoresis (NACE) coupled with electrospray ion-trap mass spectrometry (ESI-ion trap-MS). Electrophoresis, 2002; 23:2904–2912. Schmitt-Kopplin P, Junkers J. Capillary electrophoresis of natural organic matter. J. Chromatography, review article, submitted. Schmitt-Kopplin P. Comment on Determination of electrophoretic mobilities and hydrodynamic radii of three humic substances as a function of pH and ionic strength. Environ. Sci. Technol, 2002; 36:3041–3042. Schmitt-Kopplin P, Garrison AW, Perdue EM, Freitag D, Kettrup A. Capillary electrophoresis in humic substances analysis, facts and artifacts. J. Chromatography A, 1998; 807:101–109. Offord RE, Electrophoretic mobilities of peptides on paper and their use in the determination of amide groups. Nature, 1966; 211: 591. Schmitt P, Garrison AW, Freitag D, Kettrup A. Flocculation of humic substances with metal ions as followed by capillary electrophoresis (CZE). Fresenius J. Anal. Chem., 1996; 354:915–920. Schmitt-Kopplin P, Freitag D, Kettrup A, Hertkorn N, Schoen U, Klöcking R, Helbig B, Andreux F, Garrison AW. Analysis of synthetic humic substances for medical and environmental applications by capillary zone electrophoresis. Analusis, 1999; 27:6–11. Radko SP, Stastna M, Chrambach A. Polydispersity of liposome preparations as a likely source of peak width in capillary zone electrophoresis. J. Chromatography B, 2001; 761:69–75. Bullock J, Application of capillary electrophoresis to the analysis of the oligomeric distribution of polydisperse polymer. J. Chromatography, 1993; 645:169–177 . Hong M, Sudor J, Steffansson M, Novotny MV. High-resolution studies of hyoluronic acid mixtures through capillary gel electrophoresis. Anal. Chem., 1998; 70:568–573. Stefansson M. Characterization of cellulose derivatives and their migration behavior in capillary electrophoresis. Carbohydrate Res., 1998; 312:45–52. Wright BW, Ross AG, Smith RD. Capillary zone electrophoresis of fuel materials. Energy & Fuels, 1989; 3:428–430. Sjöholm E, Nilvebrant N-O. Characterization of dissolved kraft lignin by capillary zone electrophoresis. J. Wood Chem. Technol., 1993; 13:529–544. Dahlman O, Mansson K. Analysis of low molecular weight lignin-derived sulphonates by capillary zone electrophoresis. J. Wood Chem. Technol., 1996; 16:47–60. Sjöholm E. Characterization of kraft pulps by size exclusion chromatography and Kraft lignin samples by capillary zone electrophoresis. In: Department of Pulp and Paper Chemistry and Technology, Division of Wood Chemistry. Stockholm: Royal Institute of Technology, 1999. Morneau A, Pillai V, Nigam S, Winnik FM, Ziolo RF. Analysis of ferrofluids by capillary electrophoresis. Colloid Surfaces A: Physicochem. Engineer. Aspects, 1999; 154:295–301. Heinig K, Vogt C, Werner G. Separation of ionic and neutral surfactants by capillary electrophoresis and high-performance liquid chromatography. J. Chromatography A, 1996; 745:281–292. Heinig K, Vogt C, Werner G. Separation of nonionic surfactants of the polyoxyethylene type by capillary electrophoresis. Fresenius J. Anal. Chem., 1997; 357:695–700. Heinig K, Vogt C, Werner G. Separation of nonionic surfactants by capillary electrophoresis and high-performance liquid chromatography. Anal. Chem., 1998; 70:1885–1892. Grosche O, Bohrisch J, Wendler U, Jaeger W, Engelhardt H. Characterization of synthetic polyelectrolytes by capillary electrophoresis. J. Chromatography A, 2000; 894:105–116. Gao JY, Dubin PL, Sato T, Morishima Y. Separation of polyelectrolytes of variable compositions by free-zone capillary electrophoresis. J. Chromatography A, 1997; 766:233–236. Clos HN, Engelhardt H. Separations of anionic and cationic synthetic polyelectrolytes by capillary gel electrophoresis. J. Chromatography A, 1998; 802: 149–157. Schoen U. Erklärungsansatz zum Phenomen der isoelektrischen Fokussierung (IEF) von Huminstoffen auf Polyacrylamid Gelen in der Flachbett-Elektrophorese, Doktorarbeit der Geowissenschaften. Eischstätt: Catolische Universität Eischstät, 1999:235. Schmitt-Kopplin P, Freitag D, Kettrup A, Schoen U, Egeberg P. Capillary zone electrophoretic studies on Norwegian surface water natural organic matter. Env iron. Internat., 1999; 25:259–274.

20

36. 37. 38.

| CHAPTER 2: INTERPRETING CAPILLARY ELECTROPHORESIS

Deng Y, Fan X, Delgado A, Nolan C, Furton K, Zuo Y, Jones RD. Separation and determination of aromatic acids in natural water with preconcentration by capillary zone electrophoresis. J. Chromatography A, 1998; 817:145–152. Horie H, Kohata K. Analysis of tea components by high-performance liquid chromatography and high-performance capillary electrophoresis. J. Chromatography A, 2000; 881:425–438. Bronze MR, Boas LFV, Belchior AP. Analysis of old brandy and oak extracts by capillary electrophoresis. J. Chromatography A, 1997; 768:143–152.

Chapter 3 COMPARISON OF AS-DELIVERED AND AFFFF-SIZEFRACTIONATED SUWANNEE RIVER FULVIC ACID BY TIME-OFFLIGHT MASS SPECTROMETRY Wilfried Szymczak,1 Manfred Wolf2 and Klaus Wittmaack1 GSF-National Research Center for Environment and Health, 1Institute of Radiation Protection, 2Institute of Hydrology, 85758 Neuherberg, Germany 3.1. INTRODUCTION The mass and size mass distributions of humic and fulvic acids have been a matter of controversy for many years. Depending on the analytical technique, the method of sample preparation and the conditions during the measurement, mean masses between low-end values of 300–600 Da and high-end values of up to about 100,000 Da have been reported [1]. Recent work on fulvic acids by different mass spectrometry techniques consistently revealed mass distributions peaking at m/z between 200 and 400 [2–5]. Taken as a whole, these results comply with the idea that humic substances have a macromolecular (or supramolecular) structure brought about by the association (or self-assembly) of relatively small heterogeneous components to form large or even very large aggregates in aqueous solution [6–9]. The small molecules are assumed to be held together by dispersive forces [8]. Recently, the mean molar mass of different size fractions was determined for a soil humic acid [10]. The measurements involved size fractionation by high performance size exclusion chromatography (HPSEC) in combination with electrospray ionization mass spectrometry (ESI-MS). The mean molar mass was found to decrease only slightly with decreasing nominal molecular size of the isolated fractions. The results were considered to support the idea that the mean molar mass of soil humic extracts (600–1200 Da) is much lower than traditionally believed. The aim of this study was to explore mass spectral features of size fractions of a fulvic acid using asymmetrical flow field-flow fractionation (AFFFF) in combination with timeof-flight secondary ion mass spectrometry (TOF-SIMS). 3.2. MATERIALS AND METHODS 3.2.1. Reagents and Substances All chemicals and organic solvents were obtained from commercial sources (Merck, Aldrich, Fluka) with the highest purity available and were used without further purification. Aqueous solutions were prepared with high purity water (Milli-QPLUS, Millipore). Suwannee River fulvic acid (SRFA) standard, FA(IHSS), was purchased from the International Humic Substances Society (IHSS). 3.2.2. Asymmetrical Flow Field-Flow Fractionation Size fractionation of the fulvic acid was achieved with field-flow fractionation. The technique was first proposed by Giddings for macromolecules and colloids [11] and later also applied for humic substances [12–14]. In this method the analyte, dissolved in a carrier medium, is pumped with constant velocity through a thin ribbon-like channel equipped with a membrane permeable to the carrier. The flow-field is established by a cross-flow perpendicular to the channel flow and the size fractionation is a function of the diffusion coefficient. There are two types of fractionation channels available, the symmetrical channel with constant width (FFFF) and the more recently developed asymmetrical channel with continuously reduced channel width (AFFFF) [15]. We have used a system based on the AFFFF concept (Wyatt Technology, USA). The fractionation channel (ConSenxus, Germany) had a length of 286 mm and a spacer thickness of 0.54 mm. The specified cut-off of the regenerated cellulose membrane (Wyatt Technology) was 1 kDa (determined with globular proteins). 10 mM ammonium acetate (NH4Ac) served

22

| CHAPTER 3: COMPARISON OF SUWANNEE RIVER FULVIC ACID

as the mobile phase, operated at a channel flow of 0.6 mL min−1 and a cross-flow of 3 mL min−1. The absorbance of the effluent was recorded with a UV/Vis detector (K-2500, Knauer, Germany) at 280 nm. For fractionation, 20 µL samples of the IHSS Suwannee River fulvic acid standard were injected at two different concentrations (0.48 and 1.86 g/L). Each fulvic acid sample was dissolved in the mobile phase and the pH of the solution was adjusted with aqueous ammonia to ~7. The individual fractions of the experiments were sampled with a fraction collector (FC 203B, Gilson, USA). 3.2.3. Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS) The time-of-flight secondary ion mass spectrometer used in this work has been described elsewhere [16] and updates of the system were reported recently [3,17]. Based on previous experience [3], the measurements were restricted to negative secsource terminal voltage 30 kV, target bias 8 kV (for negative secondary ions), ion imondary ions. The bombardment parameters were as follows: primary ions SF5+, ion pact energy 36 kV, stationary beam current 1–2 nA, pulse width about 3 ns, repetition rate 19 kHz, 1–3×107 pulses per spectrum. The nominal angle between the primary ion beam line and the surface normal of the sample is 60°. The area of ion bombard ment was ~0.3×0.4 mm2. Neutrals generated along the first half of the beam line were removed by tilting the second half of the drift tube by a few degrees (both sections straight). The secondary ions were directed to the detector electrostatically. Postacceleration by biasing the detector was not applied. Solid samples for TOF-SIMS analysis were prepared by spray-deposition on cleaned silicon substrates [3]. The amount of material contained in the different size fractions of the fulvic acid was estimated to be small compared to the samples analyzed previously [3]. Hence it was considered necessary to perform a series of “calibration” experiments as a function of the total mass of fulvic acid in the sprayed solution. This parameter was varied to produce samples with a mean coverage ranging from sub-monolayer thickness to about 100 monolayers. Also explored was the effect of the surface properties of silicon on the drying and crystallization behavior of the sample material. For this purpose substrates with either hydrophobic or hydrophilic character were prepared. Solutions of 2–200 mg/L of fulvic acid were dissolved in an aqueous solution of 10 mM NH4Ac (that is, the same solvent as the mobile phase in AFFFF). Typically, 200 µL of solution were sprayed at a nozzle-to-substrate spacing of 70 mm. The area covered by the spray was three to four times larger than the area spanned by the silicon substrates (8×8 mm2). The total amount of fulvic acid sprayed for the calibration experiments ranged from 0.4 to 40 µg. This converts to area densities of roughly 0.2 to 20 µg/cm2. 3.3. RESULTS AND DISCUSSION 3.3.1. As-Delivered Fulvic Acids First we consider aspects of contamination and background. Figure 3.1 compares the negative TOF-SIMS spectrum of a lowconcentration fulvic acid sample with the spectrum obtained from a sample prepared by depositing the AFFFF mobile medium alone, i.e., the solution of 10 mM NH4Ac in ultra-pure water. The spectra are normalized to 107 primary ion pulses at a dc beam current of 1 nA. The mass scale is shown at the top of the graphs. The spectrum of the NH4Ac sample exhibits a number of characteristic peaks in the mass region 100<m/z<600, which are attributed to the dried ammonium acetate and to signals from the silicon substrate. It is interesting to note that all of these peaks gradually disappear with increasing concentration of fulvic acid in the solution. The upper spectrum in Figure 3.1, which relates to a relatively low concentration of fulvic acid, already shows the characteristic broad, almost featureless, high-mass distribution known from our previous work [3]. Apparently, the NH4Ac does not cause spectral distortions provided the sample thickness exceeds a certain limit. Examples of TOF-SIMS spectra of samples prepared from fulvic acids on hydrophilic and hydrophobic silicon are depicted in Figure 3.2. The spectra relate to the same amount of sprayed fulvic acid. As discussed before [3], the compositional features of fulvic acids appear in the spectra at masses m/z>150. According to Figure 3.2, the surface properties of the substrate have only a minor effect on the shape of the mass spectrum. This is somewhat surprising because the surface morphology of these samples, explored by optical microscopy, revealed significant differences that deserve further investigation. The similarity in shape of the spectra implies that the mean mass calculated from such spectra (above a certain low-mass cut-off [3]) is the same within experimental uncertainty. The absolute secondary ion yields recorded with the two samples of Figure 3.2 differ by as much as a factor of 4 to 5. This is an extreme example of the yield differences observed for samples containing nominally the same amount of sample material. However, it was found that the yields measured with samples of the same preparation series exhibited very similar behavior.

RESULTS AND DISCUSSION |

23

Figure 3.1 Comparison of mass spectra of a fulvic acid with the background spectrum due to the solution used for size fractionation

Figure 3.2 Comparison of mass spectra of the same fulvic acid prepared on different silicon substrates

Figure 3.3 Partial secondary ion yields from three different sets of samples prepared from different quantities of sprayed fulvic acid. Two sets of data were normalized to the third by applying the same factor to all spectra of the same set. The normalization factor was derived from the fractional yields in the mass range 10<m/z<34 (full squares)

By multiplying all spectra for one set of samples by the same factor, we were able to obtain a consistent dependence of partial secondary ion yields on substrate coverage. Figure 3.3 shows a compilation of partial secondary ion yields derived by summation over all yields within certain lower and upper limits of m/z. For example, is the partial yield in the mass range from m/z 200 to 400. The data shown in Figure 3.3 contain results for three different sets of samples, two sets with hydrophilic surfaces and one set with hydrophobic surfaces. Within experimental uncertainty, the partial yield for the different samples can be represented by the same coverage dependence. The partial yield exhibits only a very small coverage dependence. This result is attributed to the fact that the respective signals are mostly due to organic surface contaminants that are not representative of the fulvic acid. The trend is still present with except at very low coverage, where the yields decreased significantly with decreasing mass of deposited fulvic acid. The presence of fulvic acid on the substrates was easily observable by considering different partial yields in the m/z range from 200 to 2000 Da, represented by , , and . To illustrate the similarity in the coverage dependence of these yields more clearly, the high mass partial yields and are plotted in Figure 3.4 as a function of . A linear relation is evident, implying that the shape of the spectra measured does not depend on coverage. However, at very high coverage the partial yields for individual spectra may deviate appreciably from the general trend. These deviations suggest that the sample preparation technique may still be improved.

24

| CHAPTER 3: COMPARISON OF SUWANNEE RIVER FULVIC ACID

Figure 3.4 High-mass partial yields versus the medium mass partial yield, for as-delivered fulvic acids deposited on differently prepared silicon substrates

Figure 3.5 AFFFF fractograms of two Suwannee River fulvic acids of different mass concentrations.

Figure 3.6 Mass spectra of two AFFFF fractions of fulvic acid (run “high”). Fraction 0 denotes the forerunner fraction

3.3.2. Size-Fractionated Fulvic Acids The fractograms of two different AFFFF runs are presented in Figure 3.5, one for an input of 0.48 g/L of fulvic acid, referred to as “low”, the other one for 1.86 g/L (“high”). The different fractions are labeled by consecutive numbers, from 1 though 6. In both cases a prominent peak in the absorbance is observed at the upper end of fraction 3. Examples of TOF-SIMS spectra for fraction 4 of series “high” is depicted in Figure 3.6 together with the spectrum obtained with the forerunner fraction 0 of the same series (which did not give a detectable change in absorbance). The latter spectrum reveals a number of distinct peaks in the range m/z between 100 and 500, similar to the “background” spectrum of ammonium acetate in Figure 3.1. Again, these mass peaks are attributed to signals from substrate contamination and the substrate itself. Some of these peaks are also present in the spectrum of fraction 4, but not in the preceding and the following fractions 3 and 5 (spectra not shown here). These differences may reflect problems with sample uniformity. However, the yields con tained in

RESULTS AND DISCUSSION |

25

Figure 3.7 High-mass partial yields versus the medium mass partial yield, for size fractionated fulvic acids. The numbers denote the fractions according to Figure 3.6. The dashed lines are the calibration lines from Figure 3.4

these individual mass peaks are negligible compared to the partial yield in the mass regions of interest. Hence an analysis of the mass spectra in terms of partial yields is well justified. Figure 3.7 shows the partial yield and for size fractionated fulvic acids as a function of . The dashed lines are the calibration lines from Figure 3.4. Two aspects are remarkable. First, the partial yields measured for size fractionated fulvic acids are very close to those measured for as-delivered, unfractionated fulvic acids. Note that for these two sets of data a calibration factor as for two of the three sets of data in Figure 3.3 was not applied. Second, the results obtained for the two AFFFF runs agree very well. This implies that the spectral features do not depend on the total mass used for fractionation. The results described above seem to lend support to the idea that fulvic acids are composed of small molecules associated with each other in larger aggregates. If we assume that AFFFF provides optimum size fractionation of the aggregates, the conclusion is that aggregates of different sizes contain molecules of rather similar composition so that the mass spectra of different fractions do not differ appreciably. 3.4. CONCLUSIONS We have shown that a careful evaluation of mass spectra of as-delivered fulvic acids provides a means to circumvent problems associated with sample contamination and yield variations. The procedure allows a comparison of the essential mass spectral features of size fractionated samples and as delivered fulvic acids. The results agree with the assumption that fulvic acids are supramolecular aggregates of relatively small constituents. The size distribution of the large aggregates still needs to be determined. ACKNOWLEDGEMENTS We thank Dr. G.Buckau (Forschungszentrum Karlsruhe, Institut für Nukleare Entsorgung, Germany) for helpful discussions concerning size fractionation and Mr. D.Jurrat and Mr. G.Teichmann for skillful laboratory work and technical support of this investigation. REFERENCES 1. 2. 3. 4. 5. 6.

Stevenson FJ. Humus chemistry: Genesis, composition, and reactions. 2nd Edn. New York: Wiley, 1994. Persson L, Alsberg T, Kiss G, Odham G. On-line size-exclusion chromatography/electrospray ionisation mass spectrometry of aquatic humic and fulvic acids. Rapid Commun. Mass Spectrom., 2000; 14:286–292. Szymczak W, Wolf M, Wittmaack K. Characterisation of fulvic acids and glycyrrhizic acid by time-of-flight secondary ion mass spectrometry. Acta hydrochim. hydrobiol., 2000; 28:350–358. Plancque G, Amekraz B, Moulin V, Toulhoat P, Moulin Ch. Molecular structure of fulvic acids by electrospray with quadrupole time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom., 2001; 15:827–835. Moulin V, Reiller P, Amekraz B, Moulin Ch. Direct characterization of iodine covalently bound to fulvic acids by electrospray mass spectrometry. Rapid Commun. Mass Spectrom., 2001; 15:2488–2496. Wershaw RL. Molecular aggregation of humic substances. Soil Sci., 1999; 164: 803–813.

26

7. 8. 9.

10.

11. 12. 13. 14. 15. 16. 17.

| CHAPTER 3: COMPARISON OF SUWANNEE RIVER FULVIC ACID

Conte P, Piccolo, A. Conformational arrangement of dissolved humic sub stances. Influence of solution composition on association of humic molecules. Environ. Sci. Technol., 1999; 33:1682–1690. Piccolo A. The supramolecular structure of humic substances. Soil Sci., 2001; 166:810–833. Simpson AJ, Kingery WL, Hayes MHB, Spraul M, Humpfer E, Dvortsak P, Kerssebaum R, Godejohann M, Hofmann M. Molecular structures and associations of humic substances in the terrestrial environment. In: Proceedings of the 20th Anniversary Conference of the International Humic Substances Society, Boston, 2002; 90–92. Piccolo A, Conte P, Spiteller M. Electrospray mass spectrometry determination of molecular weights of a soil humic substance and relative size-fractions isolated by preparative HPSEC. In: Proceedings of the 20th Anniversary Conference of the International Humic Substances Society, Boston, 2002; 63–65. Giddings JC. A new separation concept based on a coupling of concentration and flow uniformities. Separ. Sci., 1966; 1:123–125. Beckett R, Jue Z, Giddings JC, Determination of molecular weight distributions of fulvic and humic acids using flow field-flow fractionation. Environ. Sci. Technol., 1987; 21:289–295. Schimpf ME, Petteys MP. Characterization of humic materials by flow field-flow fractionation. Coll. Surf. A, 1997; 120:87–100. Thang NM, Geckeis H, Kim JI, Beck HP. Application of the flow field flow fractionation (FFFF) to the characterization of aquatic humic colloids: Evaluation and optimization of the method. Coll Surf. A, 2001; 181:289–301. Wahlund KG, Giddings JC. Properties of an asymmetrical flow field flow fractionation channel having one permeable wall. Anal. Chem., 1987; 59:1332– 1339. Szymczak W, Wittmaack K. Evidence for strongly enhanced yield of negative molecular secondary ions due to bombardment with SFn cluster ions. Nucl. Instrum. Methods in Physics Res. B, 1994; 149–153. Szymczak W, Wittmaack K. Effect of water treatment on analyte and matrix ion yields in matrix-assisted time-of-flight secondary ion mass spectrometry: the case insulin in and on hydroxycinnamic acid. Rapid Commun. Mass Spectrom., 2002; 16:2025–2033.

Chapter 4 MOLECULAR FINGERPRINTING OF AQUATIC FULVIC ACIDS BY ULTRA-HIGH RESOLUTION ESI FT-ICR MASS SPECTROMETRY William T.Cooper and Alexandra C.Stenson Department of Chemistry and Biochemistry, Florida State University, Tallahassee, FL 32306–4390, USA

4.1. INTRODUCTION Mass spectral characterization of humic materials has advanced rapidly in the past few years due in large part to the development and now routine availability of electrospray ionization (ESI). ESI combined with high-resolution double-focusing [1] and QTOF [2] mass spectrometers has revealed much about the molecular character of these complex mixtures. However, it is now apparent that humic substances, including aquatic fulvic acids, exist as such complex mixtures that ultrahigh resolution electrospray ionization Fourier transform—ion cyclotron resonance mass spectrometry (ESI FT-ICR MS) at high magnetic fields is currently the only technique capable of resolving individual molecules [3–7]. Recently, this technique was used to determine molecular formulas of the majority of odd-mass ions observed in spectra of Suwannee River fulvic acids (SRFA) [3]. Of the 5,550 ions between 300 and 1100 Da that were observed in those spectra, 4,626 could be assigned an unequivocal chemical formula [3]. Careful calculation of the small differences between peaks in the ultra-high resolution FT-ICR mass spectra also revealed what appeared to be regular variations in the compositions of individual fulvic acids, and the regularity in these variations were confirmed by sorting all formulas according to the Kendrick Mass Defect (KMD). This sorting led to the conclusion that all SRFA molecules observed in the ESI FT-ICR mass spectra belonged to one of 266 homologous series; that is, the molecules belonged to a family of compounds that differed from each other only in the number of -CH2- groups. The homologous series identified differed from each other in degree of saturation and through simple substitutions of O for CH4 or CH for N. The molecular formulas previously identified for SRFA and the high degree of order that relates them should provide a molecular fingerprint for comparisons between SRFA and other humic substances, or for SRFA samples that have undergone different treatments (e.g., biodegradation or metal complexation). However, for such comparisons it is necessary to be aware of the effects that instrument conditions and sample composition may have on the observable molecular fingerprint. Sodium is a ubiquitous element in aquatic and terrestrial environments. Its presence in the ESI solvent has been shown to alter ESI mass spectra of humic substances [7,8]. Those results were difficult to analyze because humic ions were not fully resolved and because chemical noise peaks (e.g., salt clusters, solvent-base adducts) obscured the humic signal at high Na concentrations. The presence of sodium has also been shown to significantly influence ESI mass spectra of certain high molar mass mixtures. For instance, high molar mass poly(ethylene glycol) (PEG; straight-chained, ether-linked polymers) dissociate in positive-ion mode ESI-MS unless a sodiated base is added to the spray solvent [6], suggesting that the presence of Na may affect the softness of the ionization process. Finally, ionic strength has been shown to influence the apparent size and/or shape of humic substances in solution [9]. What is not understood is whether differences between sodiated and protonated fulvic molecules in solution are retained in electrospray ionization, and whether these differences affect the molecular fingerprints observed in ESI FT-ICR MS. Here we describe how the incredibly complex spectra of an aquatic fulvic acid can be reduced to a manageable distribution of homologous series that differ from each other in the degree of saturation and substitution of O for CH4. The relative distributions of these homologous series were determined in protonated and sodiated SRFA mixtures, and the resulting molecular fingerprints were compared. Suwannee River fulvic acid obtained from the International Humic Substances Society (IHSS) was used because it has long served as an aquatic fulvic acid standard and is accessible to other researchers.

28

| CHAPTER 4: MOLECULAR FINGERPRINTING OF AQUATIC FULVIC ACIDS

4.2. MATERIALS AND METHODS 4.2.1. Materials Fulvic Acids. Solid SRFA standard was acquired from the IHSS and was stored frozen in a dark box. Small quantities were dissolved in the appropriate electrospray solvent shortly before mass spectra were obtained. Fulvic acid concentrations were in the range 1–3 mg/mL. Electrospray Solvents. Electrospray solvent compositions used for each sample are listed in Table 4.1. The solvents used were as follows: Milli-Q water, methanol (MeOH; Fisher, Purge and Trap Grade), acetic acid (HAc; Fisher, TraceMetal Grade), isopropanol (IsProp; Fisher, OPTIMA Grade), aqueous ammonia (NH3(aq); Fisher, TraceMetal Grade), sodium chloride (NaCl; Fisher TraceMetal Grade) and Table 4.1 Electrospray solvent compositions Label

Fulvic Acid Concentration (mg/mL)

Spray Solvent Composition

Ionization Mode

HAc

1

+

1.6

HAcII

50% H2O 50% MeOH 0.25% HAc 1.65

50% H2O 50% MeOH 0.2% HAc −

+

NaCl

1

NaOH

2.8

NH3(aq)

1

65% H2O 33% IsProp 2% HAc 50% H2O 50% MeOH 0.3M NaCl 50% H2O 50% MeOH 7.5×10–4 M NaOH 50% H2O 50% MeOH 0.2% NH3(aq) 65% H2O 33% IsProp 2% NH3(aq)

+

+ and −

+

sodium hydroxide (NaOH; Fisher, TraceMetal Grade). To prevent contamination, solvents were stored in the bottles supplied by the manufacturer or in acid and base washed vials with removable Teflon caps. 4.2.2. Methods Mass Spectrometry. Mass spectra were acquired on a home-built, external accumulation [10], 9.4 Tesla ESI FT-ICR mass spectrometer [11] located at the National High Magnetic Field Laboratory, Tallahassee, FL. This instrument is now capable of quadrupole mass filtered accumulation [12]. The instrument is also equipped with a dual-spray ESI source [13], which was used in some experiments to introduce an internal standard (poly(ethylene glycol), PEG) ions for mass calibration. The heated metal portion of the source was cleaned every day before use with distilled water and methanol (both of high purity), and then dried with air and N2 gas before it was reinstalled. This is an important experimental precaution and is required for generating high signal-to-noise spectra with the resolution and mass accuracy provided by FT-ICR mass spectrometry at 9.4 T. Data Treatment. Time-domain data were baseline-zeroed, zero-filled once, Hamming-apodized, fast Fourier transformed, subjected to magnitude calculations and finally converted to m/z by quadrupolar electric field approximations [14]. Ion Sorting. Ions were sorted into homologous series based on Kendrick mass [15] as described previously [3,6,16,17]. Briefly, any given combination of nominal mass series (z*) value [16] and Kendrick Mass Defect (KMD) identifies a specific homologous series. Equations 4.1 through 4.3 provide the definitions of nominal mass series (z*), Kendrick Mass (KM), and Kendrick mass defect (KMD), respectively, and illustrate how these parameters are calculated from the experimental data.

MATERIALS AND METHODS |

29

Figure 4.1 Positive-ion mode ESI mass spectra of SRFA with various spray solvents. Solvent compositions listed in Table 4.1. Left panel: complete broadband mass spectra between 225 and 1400 Da. Right panel: expanded 1 Da mass spectral regions showing resolution of ions spaced 0.0364 Da apart

(4.1) (4.2) (4.3) This approach divides all observed peaks into 14 nominal mass series (z* value), followed by sorting into homologous series based on the Kendrick Mass Defect (KMD). 4.3. RESULTS To investigate the effect of sodium in the ESI spray solvent on SRFA mass spectra, low concentrations of NaCl and NaOH were added to SRFA in positive and negative ionization modes. Previously it was demonstrated that the addition of NaOH does not affect the charge-state of SRFA in positive-ion mode [6]. However, careful examination of ESI mass spectra of SRFA with and without Na in the spray solvent (Figure 4.1) reveals that the addition of sodium to positive-ion mode ESI spray solvents results in fewer ions overall (left panel) and slightly fewer ions per unit m/z (right panel). This effect is clearly more pronounced with the addition of NaOH than with NaCl. However, at concentrations significantly greater than 10–3 M Na, interference from chemical noise (most likely solvent-base adducts) [7] is pronounced. High concentrations of Na should therefore generally be avoided, especially in positive-ion mode. Kendrick plots (i.e., plots of KMD versus nominal mass) allow easy visual comparison of high-resolution mass spectra [3–5,17]. Figure 4.2 is a superimposition of such plots for SRFA containing no sodium (acetic acid (HAc) only) and those containing NaOH and NaCl. It is apparent that the spectra of SRFA sodiated with NaOH contain fewer ions than the usual positive-ion mode spectra of protonated SRFA. For any given homologous series (any horizontal line), ions that are present in the sodiated spectrum are centered at higher mass, and for any nominal mass (any vertical line), are centered at lower double bond equivalency (DBE) and oxygen content. To determine if the variations in the Kendrick plots in Figure 4.2 are really due to the presence of sodium and not just to the normal variations in electrospray ionization efficiency, molecular distributions with only HAc in the spray solvent were compared with those from previous experiments [3]. Since a Kendrick plot of all ions observed in a complex spectrum can be

30

| CHAPTER 4: MOLECULAR FINGERPRINTING OF AQUATIC FULVIC ACIDS

Figure 4.2 Kendrick plots integrated over all z* for positive-ion mode ESI mass spectra of SRFA obtained in protonating (HAc and HAcII, dark regions) and sodiating (NaCl and NaOH, light regions) solvents

Figure 4.3 Kendrick plots of nominal mass series z*=−2 for positive-ion mode ESI mass spectra of SRFA obtained in similar protonating solvents but at different times

Figure 4.4 Kendrick plots of nominal mass series z*=−2 for positive-ion mode ESI mass spectra of SRFA obtained in sodiating solvents

quite cluttered (see Figure 4.2), Figure 4.3 only contains plots for ions with z* scores of −2. This format allows the distributions of homologous series (horizontal lines) and ions of the same nominal mass but different DBE and oxygen content (vertical lines) to be compared. The overlayed Kendrick plots in Figure 4.3 suggest that the differences observed when NaOH is present are indeed real. However, NaCl does not appear to produce the same effect, as indicated by comparison of the Kendrick plots in Figure 4.4. This is an important result, given the significant amount of data that suggest the addition of salts such as NaCl

MATERIALS AND METHODS |

31

Figure 4.5 Expanded 1 Da mass spectral regions of negative-ion (left panel) and positive-ion mode (right panel) ESI mass spectra obtained with various spray solvents

can cause changes in molar mass distributions of fulvic acids in solution. If the addition of NaCl induced changes in the overall size of the fulvic acid molecules, these changes do not appear to survive the ESI process. Negative-ion mode spectra of SRFA were also measured in spray solvents with and without NaOH. It has been reported that negative-ion mode spectra are usually centered at slightly lower mass-to-charge ratios (m/z) than positive-mode spectra [7,18]. Qualitative comparison of individual spectra in both modes with different spray additives (Figure 4.5) indicates that the same ions are observed in either mode and regardless of solvent conditions. Note that in positive-ion mode the chargecarrying feature of the molecule (normally H+) must be subtracted from m/z of the observed molecular ion to get the actual molecular formula. In negative-ion mode, the negative charge normally is due to the loss of H+, and the mass of H must then be added to get the molecular formula. Thus, positive- and negative-ion mode mass scales of the same molecules are shifted by 2 Da relative to each other, as indicated in Figure 4.5. These results agree with comparisons of positive and negative mode spectra obtained without sodium [3]. Positive and negative ion mode spectra were also compared based on the homologous series present. It was discovered that the same homologous series are present in both ionization modes as long as the NaCl spectrum is included. A similar conclusion was drawn previously based on a small (~2 Da) mass window of SRFA ions [3]. Therefore, given that the ionization mode does not appear to affect the classes of ions present in the mass spectrum, comparison of the effect of Na addition to spray solvents in either mode can be made. Figure 4.5 reveals that the addition of NaOH to negative-ion mode spray solvents (middle spectrum, left panel) does not reduce the complexity of the spectrum (that is, the number of ions per unit m/z), but it does in positive-ion mode (right panel). The combined positive and negative-ion mode data therefore suggest that the addition of Na+ reduces ionization efficiency in positive-ion mode only. One possible explanation is that certain fulvic acids within the mixture are more completely cationized by H+ than by Na+, yet they appear to be anionized with equal efficiency in the presence of either cation. This explanation is consistent with the observation that sodiated mass spectra contain far fewer ions overall than protonated spectra. It is worth noting that fulvic acids in solution are generally anions at relevant pH values. It appears, furthermore, that cationization efficiency is higher for fulvic solutions containing NaCl than for those containing NaOH, even at similar solution pH. However, increasing the pH by addition of ammonia does not diminish the cationization efficiency. It therefore appears that pH does not account for the difference in mass spectra of SRFA solutions containing NaOH versus those containing NaCl. Furthermore, it is likely that NH4+ is one of the charge-inducing cations when ammonia is added to positiveion mode spray solvent, which may explain the large complexity of ions per unit mass under those conditions. The data suggest, therefore, that some characteristic of the counter-ion (i.e., OH− vs.Cl−) is also involved in controlling the fate of ions in the ESI source in positive-ion mode. The characteristic in question may be the extent to which the anion is removed through electroosmotic flow away from the capillary exit before analyte ionization takes place. Anionic reactivity may also

32

| CHAPTER 4: MOLECULAR FINGERPRINTING OF AQUATIC FULVIC ACIDS

affect the extent to which either counter-ion competes with fulvic acids for the charge-inducing sodium ions or protons. Full elucidation of the mechanisms involved is beyond the scope of this paper. It is also interesting to note that to prevent dissociation of high molar mass PEG standards, the addition of NaOH to the spray solvent had a far more significant effect than with NaCl, while the addition of ammonia had no effect [6]. Therefore, an alternative explanation of the trends observed for SRFA may be that SRFA molecules, like high molar mass PEG, dissociate in acidic spray solvents in positive-ion mode but less so in sodiated, non-acidic spray solvents. The reduction in the number of ions observed, especially at low mass, could be due to fewer low mass fragments rather than the more efficient ionization of low mass SRFA molecules. 4.4. DISCUSSION One of the most striking features of previously published ultra-high resolution FT-ICR mass spectra of humic samples is the regular mass spacings between peaks [3,6]. Because of the mass accuracy of FT-ICR spectrometry at 9.4 Tesla, these spacings could be assigned to specific molecular differences. Those mass spacings and the molecular substitutions responsible for each are listed in Table 4.2. Of the over 5000 individual SRFA molecules previously identified, ~4600 could be placed within one of 266 homologous series once these regular variations were identified. Somewhat surprisingly, these homologous series appeared to be highly correlated based on z*, double bond equivalency, and the number of oxygen atoms. This correlation can be expressed by Eq. 4.4: Table 4.2 Mass spacing patterns observed in ESI FT-ICR mass spectra of Suwannee River fulvic acids Mass Spacing (Dalton)

Origin

0.0364 0.9953 1.0034 2.0157 14.0156

CH4 vs. O 14NH vs. 12CH 2 13C vs. 12C double bond or ring (loss of H2) addition of CH2

(4.4) where DBE is the double bond equivalency, o is the number of oxygen atoms, z* is the nominal mass series and b is the parameter that differs for different “families” of homologous series. This b parameter reflects the “degree of saturation”, or more precisely the “hydrogen deficiency” after the DBE and number of oxygen atoms are accounted for. It can vary in increments of 7 because the nominal mass series parameter z*, which accounts for hydrogen deficiency only, can range from −14 to −2 in increments of two. This is due to the fact that each double bond produces a hydrogen deficiency of 2 Da for a given mass. Those homologous series that differ from each other in this hydrogen deficiency parameter b by seven degrees of saturation (or any multiple thereof) will be referred to as different “families” of homologous series because each family is uniquely characterized by the same mathematical relationship between double bond equivalence, number of heteroatoms and nominal mass series. This mathematical relationship is an attractive approach to comparing different humic materials because the molecular variations that are responsible for differences in b can be attributed to specific reac tions suspected of altering humic precursors in nature [3,19–21]. For example, demethylation of lignin will result in the loss of −CH2- groups, side-chain oxidation causes loss of –H2- groups, and aromatic ring opening results in addition of –O2-. For comparison of molecular fingerprints based on families of homologous series, ions in the positive-ion mode spectra of Figure 4.1 were assigned to homologous series. The process of formula assignment involved calculating z* and KMD (Eqs. 4. 3 and 4.4) values for each ion in each spectrum. Ions that had the same z*, KMD and neutral mass as SRFA ions previously identified [3] were assigned to the same homologous series reported therein. Based on this assignment, the double bond equivalence and number of oxygen atoms for each ion is known (see the Supplementary Information of ref. [3]). Individual molecular formulas can be calculated by simultaneously solving Eqs. 4.5 and 4.6, where c, h, n and o stand for the number of carbon, hydrogen, nitrogen, and oxygen atoms per formula unit. It should be noted, however, that individual formulas were not necessary for these molecular class comparisons; only z*, DBE and o are required. (4.5) (4.6) To compare z* and KMD values with those previously identified for SRFA molecules, it is necessary to calculate the neutral mass of each ion. For SRFA ions formed in positive-ion mode from sodium-free solutions, this was accomplished by

MATERIALS AND METHODS |

33

Figure 4.6 Total relative odd-mass abundances of SRFA homologous series “families” sorted according to saturation parameter b defined in Eq. 4.4. These families of homologous series differ from each other by 7 DBE, hence even spacing of 7 integers between b values

subtracting the mass of a proton from all measured masses. For equivalent spectra of SRFA solutions that contained Na, the mass of sodium was subtracted when necessary. For SRFA taken in negative ion mode, the mass of a proton was added to all measured masses, regardless of solution conditions. Finally, some molecular formula assignments were spot-checked by entering the measured peak location (ion mass) into the mass calculator. Spot-checking did not reveal any incorrectly assigned molecular formulas and indicated that ions produced in positive-ion mode from SRFA solutions containing Na were virtually exclusively sodiated and those produced in negative-ion mode were virtually exclusively deprotonated. Assignment of each ion to a homologous series revealed that the vast majority of ions identified in both positive- and negative-ion mode spectra belong to the same sets of homologous series as ions previously identified [3]. In Figure 4.6, ions in the spectra of Figure 4.1 together with ions from the spectrum characterized previously [3] are sorted into families of homologous series based on the saturation parameter b. The two dominant families are those for which b is either 1 or −6. Other sets of homologous series contain only a negligible number of ions and comprise a very small fraction of the overall SRFA ion abundance. It is noteworthy, however, that the difference in b between different families is always exactly seven, which results from an increase in DBE by 7 while o and z* remain constant. This observation confirms the previous conclusion that these families of homologous series are related to each other by seven degrees of saturation [3]. It obvious from Figure 4.6 that there is no trend in the abundances of these two families as a function of spray solvent composition. Indeed, the difference in distributions from spectra obtained with similar ionization conditions but at different times (HAc and HAcII) is greater than those obtained using different spray solutions. That is, there is no observable difference in the families of ions that can be ionized under the various ESI spray solvent conditions investigated. This observation is confirmed by close inspection of z*-sorted Kendrick plots. For example, the Kendrick plots in Figure 4.4 reveal that sodiation by NaOH does not lead to the loss of any homologous series, nor does NaCl introduce any new homologous series relative to NaOH. While all members of a series are not always present in spectra using both spray solvents, each series does have representative ions present in both spectra. In other words, while sodiation by NaOH may affect the ionization of ions within a homologous series, it does not appear to selectively suppress the ionization of any given series. 4.5. CONCLUSIONS The results presented here confirm previous observations that individual ion intensities in ESI FT-ICR mass spectra of fulvic acids can be highly variable even when high purity salt-free spray solutions are used [4]. Low concentrations of sodium, particularly when OH− is present, appear to affect the number of ions that are observed within a given mass window. Fortunately, Na in the ESI spray solvent does not appear to significantly alter the homologous series present in SRFA ESI FTICR mass spectra, only the number of ions within each series that are observed. Comparison of spectra can therefore be made based on the homologous series present, but not on individual ion intensities or intensities of families of homologous series. Comparisons between humic substances that require accurate and precise abundance measurements (e.g., average molar mass calculations) should therefore not be based on ESI FT-ICR mass spectral data alone. However, mass spacings and the homologous series present within each mass spectrum do not appear to depend strongly on spray solvent and instrument

34

| CHAPTER 4: MOLECULAR FINGERPRINTING OF AQUATIC FULVIC ACIDS

conditions. It also appears that negative-ion mode spectra are better suited for comparisons of mass spectra, since they do not appear to be sensitive to the addition of sodium to the spray solvent and only deprotonated ions are observed, regardless of the nature of the counter-ion (H3O+, Na+ or NH4+). Finally, the data demonstrate that there is no consistent difference in mass distributions observed in sodiated and protonated SRFA mass spectra. Therefore, if the addition of Na induces changes in the size distribution of SRFA in solution, these changes do not appear to survive the ESI process. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

14. 15. 16. 17. 18. 19. 20. 21.

McIntyre C, Daniel J. Electrospray mass spectrometry of aquatic fulvic acids. Rapid Commun. Mass Spectrom., 2001; 15: 1974–1975. Plancque G, Amekraz B, Moulin V, Toulhoatt P, Moulin C. Molecular structure of fulvic acids by electrospray with quadrupole timeof-flight mass spectrometry. Rapid Commun. Mass Spectrom., 2001; 15:827–835. Stenson AC, Marshall AG, Cooper WT. Exact masses and chemical formulas of individual Suwannee River fulvic acids from ultrahigh resolution ESI FT-ICR mass spectra. Anal. Chem., 2002; in review. Kujawinski EB, Hatcher PG, Freitas M A. High-resolution Fourier transform ion cyclotron resonance mass spectrometry of humic and fulvic acids: Improvements and comparisons. Anal. Chem., 2002; 74:413–419. Kujawinski EB, Freitas MA, Zang X, Hatcher PG, Green-Church KB, Jones RB. The application of electrospray ionization mass spectrometry (ESI MS) to the structural characterization of natural organic matter. Org. Geochem., 2002; 33:171–180. Stenson AC, Landing WM, Marshall AG, Cooper WT. Ionization and fragmentation of humic substances in electrospray ionization Fourier transform-ion cyclotron resonance mass spectrometry. Anal. Chem., 2002; 74:4397–4409. Brown TL, Rice JA. Effect of experimental parameters on the ESI FT-ICR mass spectrum of fulvic acid. Anal. Chem., 2000; 72: 384–390. Ikeda K, Arimura R, Echigo S, Shimizu Y, Minear RA, Matsul S. The fractionation-concentration of aquatic humic substances by the sequential membrane system and their characterization with mass spectrometry. Water Sci. Technol., 2000; 42:383–390. Chin Y-P. Gschwend PM. The abundance, distribution and configuration of porewater organic colloids in recent sediments. Geochim. Cosmochim. Acta, 1991; 55:1309–1317. Senko MW, Hendrickson CL, Emmett MR, Shi S-H, Marshall AG. External accumulation of ions for enhanced electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. J. Amer. Soc. Mass Spectrom., 1997; 8: 970–976. Senko MW, Hendrickson CL, Pasa-Tolic L, Marto JA, White FM, Guan S, Marshall AG. Electrospray ionization Fourier transform ion cyclotron resonance at 9.4T. Rapid Commun. Mass Spectrom., 1996; 10:1824–1828. Hendrickson CL, Quinn JP, Emmett MR, Marshall AG. Quadrupole mass filtered external accumulation for Fourier transform ion cyclotron resonance mass spectrometry. 48th American Society for Mass Spectrometry Annual Conference, 2000, Long Beach, CA. Hannis JC, Muddiman DC. A dual electrospray ionization source combined with hexapole accumulation to achieve high mass accuracy of biopolymers in Fourier transform ion cyclotron resonance mass spectrometry. J. Amer. Soc. Mass Spectrom., 2000; 11: 876–883. Ledford EB, Rempel DL, Gross ML. Space-charge effects in Fourier-transform mass-spectrometry-mass calibration. Anal. Chem., 1984; 56:2744–2748. Kendrick EA. Mass scale based on CH2=14.0000 for high resolution mass spectrometry of organic compounds. Anal. Chem., 1963; 35:2146–2154. Hsu CS, Qian K, Chen YC. An innovative approach to data analysis in hydro-carbon characterization by on-line liquid chromatography-mass spectrometry. Anal. Chim. Acta, 1992; 264:79–89. Hughey CA, Hendrickson CL, Rodgers RP, Marshall AG. Elemental composition analysis of processed and unprocessed diesel fuel by electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Energy and Fuels, 2001; 15:1186–1193. Klaus U, Pfeiffer T, Spiteller M. APCI-MS/MS: A powerful tool for the analysis of bound residues resulting from the interaction of pesticides with DOM and humic substances. Environ. Sci. Technol., 2000; 34:3514–3520. Filley TR, Cody GD, Goodell B, Jellison J, Noser C, Ostrofsky A. Lignin demethylation and polysaccharide decomposition in spruce sapwood degraded by brown rot fungi. Org. Geochem., 2002; 33:111–124. Leonowicz A, Cho N-S, Luterek J, Wilkolazka A, Wojtas W, Matuszewska A, Hofrichter M, Wesenberg D, Rogalski J. Fungal laccase: Properties and activity on lignin. J. Basic Microbiol., 2001; 41:185–227. Higuchi T. Biodegradation mechanism of lignin by white-rot basidiomycetes. J. Biotechnol., 1993; 30:1–8.

Chapter 5 THE MACROMOLECULAR OR SUPRAMOLECULAR NATURE OF HUMIC SUBSTANCES: A DYNAMIC LIGHT SCATTERING STUDY Gustavo González-Gaitano1 and Josemaría García-Mina1,2 1Department 2R&D

of Chemistry and Soil Chemistry, Universidad de Navarra, 31080, Pamplona, Spain

Department, Inabonos-Roullier Group, Polígono Arazuri-Orcoyen, 31160, Orcoyen, Spain

5.1. INTRODUCTION An adequate knowledge of the chemical nature and structure of humic substances (HSs) is fundamental to better understanding of the effects of these organic systems on the biological dynamics of natural ecosystems. However, despite the great number of investigations devoted to this subject, there still is no definitive answer to the question, which is due to the chemical complexity of HSs [1–4]. Whereas many studies founded on the macromolecular chemical behavior of HSs in solution have proposed that they fundamentally consist of complex mixtures of different macromolecules whose physicochemical properties are qualitatively similar [3], other recent studies, normally based on the HPSEC technique, propose that these systems are supramolecular assemblies of relatively small molecules [4]. Because of these different results, an interesting debate about the macromolecular or supramolecular nature of HSs is underway in the literature [2]. Recent studies have demonstrated that the particular experimental conditions of HPSEC may cause artifacts that could lead to erroneous conclusions [5]. Several papers have reported significant discrepancies between the molar mass of a humic system calculated using HPSEC and alternative techniques such us multi-angle light scattering (MALS) [6–8]. The aim of our work was to investigate whether chemical processes associated with significant changes in the average molar mass according to HPSEC techniques give similar results when an alternative method is used. Among the different possible techniques available, we have chosen dynamic light scattering (DLS). This tech nique has advantages in the study of the sizes of macromolecular systems or the associations of small molecules to form supramolecular assemblies [9]. The basis of DLS lies in the measurement of the autocorrelation function of the scattered intensity, that is, the convolution of the signal intensity as a function of time. The longer the correlation with time, the slower the movement of the particles through the solution, which is a property that can be quantified by means of the diffusion coefficient. Both variables, size and diffusion, are related by the Stokes-Einstein Eq. 5.1, (5.1) where T is the absolute temperature, k is the Boltzmann constant, 0 is the viscosity of the solvent and D0 is the diffusion coefficient at infinite dilution. The latter is obtained with a mathematical procedure known as regularised inverse Laplace transformation, which gives the size distribution of the particles responsible for the measured autocorrelation function. The strong point of this technique is its sensitivity both to the concentration and to the size of the particles, especially to the latter. This allows the detection of aggregation at an early stage. We have studied the validity of this experimental technique to adequately describe the structural changes in solution undergone by a humic acid (HA) as a function of pH, ionic strength and acidification-realkalization, following the methodology used by Piccolo et al [1], We also have investigated the effects that fractionation has on the molar mass distribution of this same humic system. 5.2. MATERIALS AND METHODS 5.2.1. Humic Substances Humic acid was obtained from Aldrich and HA solutions were prepared in distilled, deionised water. The final HA concentration in solution was 0.8 g/L in all the experiments. Alkaline pH was fixed by adding NaOH. For the acidification studies, hydrochloric, oxalic and citric acids of analytical grade were employed. The samples for DLS were not filtered except

36

| CHAPTER 5: THE MACROMOLECULAR OR SUPRAMOLECULAR NATURE OF HUMIC SUBSTANCES

in the fractionation experiments, where 0.1 µm filters (Anotop, inorganic membrane) and 0.2 µm (Albet, cellulose acetate) were used. 5.2.2. Instrumentation Dynamic light scattering measurements were made at a scattering angle of 90° and at different temperatures using a DynaProMS/X photon correlation spectrometer equipped with a 248 channel multi-tau correlator and a Peltier effect thermostat unit. The wavelength of the laser was 825.5 nm. The size distribution was obtained from the intensity autocorrelation function by regularization analysis implemented in the Dynamics™ software package, and the hydrodynamic radii were calculated from measured diffusion coefficients with Eq. 5.1. 5.3. RESULTS AND DISCUSSION 5.3.1. Effect of pH, Ionic Strength and Temperature Figure 5.1 shows the correlogram and the corresponding regularization analysis for HA at pH 10 and 0.015 mol/L NaCl. The results for other ionic strengths are collected in Table 5.1. At pH 2, concentrations of NaCl higher than 0.015 M led to precipitation of the HA, whereas at pH 10 solutions were stable up to 3 M NaCl. The size distribution is in all cases bimodal, indicating the presence of two different species with mean radii of 47 nm and 144 nm at 25°C. The latter species is polydisperse. This slow mode is due to some kind of association and, although it represents a considerable part of the overall scattered intensity, its contribution in mass is less than a 23% (Figure 5.1). To understand this, one must distinguish between mass distribution and intensity distribution. The mass distribution can be calculated approximately by considering that the scattering intensity of a particle i is proportional to both its molar mass, Mi and its weight concentration, ci, Eq. 5.2. (5.2) The molar mass of the i particle can be estimated from Eq. 5.3, where v is its partial specific volume, a is a shape parameter equal to 2 for coils and 3 for spherical particles, and N is Avogadro’s number. (5.3) Then Eq. 5.2 becomes Eq. 5.4. (5.4) From Eq. 5.4 it can be readily inferred that large particles will scatter much more radiation than small ones due to the a exponent, even if large particles are in low concentration. The mass distribution, expressed in terms of the concentration,* can be deduced from the intensity after substituting Eq. 5.5 in Eq. 5.4: Table 5.1 Mean hydrodynamic radii and polydispersity (nm) of HA, 0.8 g/L at 25°C. NaCl (mol/L)

pH 10

pH 2

after realkalization

59b±4;

22a,b; 134±50 15a,c; 147±60 22a,d; 146±60 – –

1141±90 0.015 47a; 144±13 59c±22; 194±52 33d±10; 168±68 a 0.05 32 ; 156±54 – 3 96±12; 2085±605 – apolidispersity very small; bacidified with HCl; cacidified with oxalic acid; dacidified with citric acid

(5.5)

*A more rigorous calculation should include a shape factor, P( ), that accounts for the angle dependence of the scattered light due to the shape of the particle. The assumption that P( )=1 is only valid for particles smaller than the wavelength of the incident radiation, where scattering from different parts of the same particle is absent.

RESULTS AND DISCUSSION |

37

Figure 5.1 Intensity autocorrelation function (left) and size distribution (right) of aqueous HA (0.8 g/L, I=0.015, 25°C) at pH 10, not filtered

Figure 5.2 Intensity autocorrelation function (left) and size distribution (right) of aqueous HA (0.8 g/L, I=0.015, 25°C) at pH 2 (HCl), not filtered

DLS directly gives the intensity distribution, whereas the mass distribution is estimated with a model. The mass distribution thus defined will be strongly dependent on the radius of the particle. The peak centered in 144 nm (Figure 5.1) does not appear when the solutions are filtered with 0.2 µm filters, and its small contribution in mass is confirmed by UV spectroscopy from the decreased absorbance measured at 400 nm after filtration. As expected, increased ionic strength results in growth of the radii of both modes and in the scattered intensity. Raising the temperature does not seem to modify the size or the relative proportion of the particles, at least in the interval between 25 and 55°C. At acidic pH, the size distribution is different according to the nature of the acid used to reach pH 2. Thus, two different kind of particles are observed with HCl (Figure 5.2). The largest value is 1.1 µm in hydrodynamic radius. This suggest the Table 5.2 Mean hydrodynamic radii and polydispersity (nm) of filtered HA solutions at 25 and 55°C Temperature (°C)

0.1 µm filter

0.2 µm filter

after realkalization and 0.2 µm filtration

13±2b; 67±7 15a,c; 77±25 15a,d; 66±8 a a 55 1.4 ; 3.7±0.7; 25±5 20 ; 79±28 20a,b; 65±7 20a,c; 71±17 16d±3; 61±6 a polydispersity very small; b acidified with HCl; c acidified with oxalic acid; d acidified with citric acid

25

1.3±0.2; 3.8±0.7; 26 ±5

15a;

78±36

growth of very large aggregates in conditions close to those that result in precipitation. On the other hand, these large aggregates were not observed when oxalic and citric acids were used. Instead, addition of both acids led to bimodal size distributions similar to those obtained at basic pH, although in this case the radius corresponding to the fast mode is slightly smaller than that measured at alkaline pH (Table 5.1). When the temperature increases, in the case of the acidification with HCl, the 1.1 µm particle diminishes in its percentage intensity and a new mode arises at ca. 99 nm. This suggests the breaking of the large aggregates and their conversion into the other two types of particles (data not shown). Realkalization of all the acidic solutions up to pH 10 leads in all cases to bimodal distributions and to the same sizes (Tables 5.1 and 5.2). This fact is more noticeable with the 0.2 µm filtered samples. The slight differences in the realkalizated unfiltered samples with respect to the normal alkaline samples may be due to interference of the slow mode, whose intensity contribution is much higher. This makes the size estimation less precise.

38

| CHAPTER 5: THE MACROMOLECULAR OR SUPRAMOLECULAR NATURE OF HUMIC SUBSTANCES

Figure 5.3 Intensity autocorrelation function (left) and size distribution (right) of aqueous HA (0.8 g/L, I=0.015, 25°C) at pH 10, 0.1 µm filtered

5.3.2. Effect of Fractionation The presence of large aggregates with substantial statistical weight in the overall intensity masks the presence of smaller particles and makes precise calculation of the particle dimensions difficult. In order to exclude or minimize this effect, the HA solutions at pH 10 and I=0.015, were fractionated with 0.1 µm and 0.2 µm syringe filters (Table 5.2). The size distribution at 25°C after filtering with 0.2 µm filters consists of two peaks centered at 15 and 78 nm radii. The larger one is due to the remaining smaller particles of the 144 nm peak that pass through the filter. When a 0.1 µm filter is used (Figure 5.3), the remaining peak is practically absent, and the first peak becomes two peaks with 1.3 and 3.8 nm radii. The first peak represents 75–95% of the mass (depending on the model, coils or spheres, respectively). Taking into account that the absorbance of the HA filtered through 0.1 µm represents 86% of the absorbance of the unfiltered sample, these results indicate that the most abundant species in solu tion must be relatively small. The absolute molar mass of the different species can be approximated by the “hydrodynamic” molar mass. By considering a sphere of hydrodynamic radius Rh and specific volume v, and taking a mean value of 0.69 cm3/g for v [10], a molecular mass of 8±3 kDa is obtained. In this calculation the molecular mass tends to be overestimated, since it includes the solvation shell that covers the particle. Moreover, the assumption of a spherical shape might be incorrect. Nevertheless, the molecular mass thus calculated agrees with literature values found for many HAs [10]. In the temperature range studied, no changes were observed in any of the filtered solutions (Table 5.2), indicating the stability of both classes of molecules as the aggregates. 5.4. CONCLUSIONS This investigation shows the validity of a non invasive technique, dynamic light scattering, to study the structural changes undergone by HSs as a result of variations in pH, ionic strength or temperature. In the system investigated, a significant molecular aggregation of a HA caused by acidification with HCl has been observed as a probable consequence of significant formation of hydrogen bonds and hydrophobic interactions. This molecular aggregation was not detected when the acidification was accomplished with organic acids (oxalic and citric). This fact, as some authors have suggested, may be the consequence of the effect of the molecular solvation of HA by molecules of these organic acids that prevents the formation of hydrogen bonds or hydrophobic interactions [1,4]. However, when the HA acidic extracts (oxalic or citric) were realkalized, the mass distribution found was very similar to that associated with the normal alkaline HA extract before acidification. Only a less significant aggregation associated with the small fraction of larger molecules seems to disappear after realkalization. On the face of it, this result seems to contradict results observed with HPSEC. In fact, our data suggest that the alkaline HA extract used in this work is principally formed by a fraction of relatively small macromolecules (molar mass ca. 8 kDa) together with a smaller proportion of some kinds of supramolecular assemblies. Once these aggregates are filtered, no significant differences have been detected when analyzing the distribution of masses associated with the alkaline extract with or without previous citric or oxalic acidification. This suggests that these smaller molecules do not form more complex supramolecular structures. In this sense, our results would at least in part be coherent with the model proposed by Simpson [11]. Further studies using more humic samples and integrating different analytical techniques are necessary to assess these conclusions.

RESULTS AND DISCUSSION |

39

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

Piccolo A, Nardi S, Concheri, G. Macromolecular changes of soil humic substances induced by interactions with organic acids. Eur. J. Soil Sci., 1996; 47: 319–328. Hayes MHB, Clapp CE. Humic substances: Considerations of compositions, aspects of structure, and environmental influences. Soil Sci., 2001; 166:723–737. Swift RS. Molecular weight, size, shape and charge characteristics of humic substances: Some basic considerations. In: Hayes MHB, MacCarthy P, Malcolm RL, Swift RS eds. Humic Substances II. In search of structure. New York: Wiley, 1989:449–466. Piccolo, A. The supramolecular structure of humic substances. Soil Sci., 2001; 166:810–832. Varga B, Kiss G, Galambos I, Gelencser A, Hlavay J, Krivacsy Z. Secondary structure of humic acids. Can micelle-like conformation be proved by aqueous size exclusion chromatography? Environ. Sci. Technol., 2000; 34:3303–3306. Wagoner DB, Christman RF, Gauchon G, Paulson R. Molar mass and size of Suwannee River natural organic matter using multi-angle laser light scattering. Environ. Sci. Technol., 1997; 31:937–941. Wagoner DB, Christman RF. Molar masses and radii of humic substances measured by light scattering. Acta Hydrochim. Hydrobiol., 1998; 26:191–195. Wandruska R, Schimpf M, Hill M, Engebretson R. Characterization of humic acid size fractions by SEC and MALS. Org. Geochem., 1999; 30:229–235. Pecora R. Dynamic light scattering: Applications of photon correlation spectroscopy. New York: Plenum, 1985; Brown W ed. Dynamic light scattering: The method and some applications. Oxford: Clarendon Press, 1993. Swift RS. Molecular weight, size, shape and charge characteristics of humic substances: some basic considerations. In: Hayes MHB, MacCarthy P, Malcolm RL, Swift RS eds. Humic Substances II. In search of structure. New York: Wiley, 1989:467–495. Simpson, AJ. The structural interpretation of humic substances isolated from podzols under varying vegetation. Ph.D. Dissertation, England: University of Birmingham, 1999.

Chapter 6 A PROPOSAL FOR THE ESTABLISHMENT OF A DATABASE OF THERMODYNAMIC PROPERTIES OF NATURAL ORGANIC MATTER Rossane C.DeLapp and Eugene J.LeBoeuf Department of Civil and Environmental Engineering, Vanderbilt University, Nashville, Tennessee 37325, USA 6.1. INTRODUCTION Soil and sediment-derived natural organic matter (NOM) are ubiquitous in the environment. The importance placed on understanding the functional and structural compositions of NOM derives from this ubiquity, and a quest to better understand how its presence affects many fundamental environmental processes, including carbon cycling [1] and sequestration [2,3], health of soils (and consequent agricultural productivity) as evidenced by the quality and quantity of NOM present [4], and pollutant fate and transport governed, in part, by how NOM (in dissolved and undissolved form) influence contaminant sorption and desorption patterns [5]. NOM is a complex mixture of newly deposited and variously aged biomaterials of plant and animal origin. While some biomaterials may remain intact over time (e.g., some lignins), other biomaterials (e.g., polysaccharides, carbohydrates, proteins and organic acids) may be altered through various physical and chemical means to form humic materials, including humic acids, fulvic acids, humin and kerogens. Their formation may be described through various degradation and condensation models [6]. Degradation models infer that biomaterials undergo partial degradation to form humic and fulvic acids and then complete remineralization. Condensation models assume that biomaterials experience complete degradation into small, reactive molecules to repolymerize into humic and fulvic acids. Through geologic transformations (e.g., diagenesis), these molecules continue to repolymerize into kerogen components. Although controversy surrounds discussion of the applicability of various degradation and formation pathways [7,8], suffice it to say that, regardless of the pathway, the resulting NOM is complex and heterogeneous in nature. The challenge, then, is to provide a means to characterize these materials, and then to use this information to construct a predictive tool to better understand how NOM behaves in the environment. Because of the heterogeneity of most NOM, physical and chemical methods are often employed to separate the various components of NOM in an effort to better homogenize the samples in preparation for physical and chemical characterization studies. Typically, these fractionation procedures attempt to separate constituents without adversely altering their physical or chemical structures. Such fractions may include carbohydrate, fulvic acid, humic acid and humin. Once homogenized, numerous characterization techniques may be employed, including physical-, chemical-, spectroscopic-, and thermal-based means. Emphasis here is placed on advanced spectroscopic and thermal analysis techniques and how they may be employed in conjunction with molecular simulation methodologies to assist in the development of improved NOM models. This paper provides a basis to propose the formation of a database of thermodynamic information for natural organic matter. Subsequent to this introduction, a brief review of spectroscopic and thermal analysis techniques is presented. The usefulness of molecular simulations in improving our understanding of NOM structure is then highlighted, including discussions of how advances in spectroscopic and thermal analysis techniques are lending further knowledge to the structural conformation of NOM. Emphasis is placed on how thermal characterization methods may be used to quantify thermodynamic properties of NOM, and how this information can also be employed as additional constraints in NOM molecular simulations. Lastly, the paucity of thermodynamic data for NOM engenders a proposal for the establishment of a database of thermodynamic properties, including the standardization of thermal analysis protocols as the kernel to ensure reproducibility of results across a wide spectrum of samples. 6.2. SPECTROSCOPIC TECHNIQUES Spectroscopic methods rely on measured electronic, vibrational and rotational motions induced in molecules by electromagnetic radiation (e.g., radio waves, microwave, infrared, ultraviolet light, visible light or X-rays), and are generally favored over wet chemistry and pyrolysis characterization methods due to their nondestructive nature [9]. Spectroscopic

SPECTROSCOPIC TECHNIQUES |

41

studies employ single spectral analyses to characterize NOM, including Fourier transform infrared (FTIR), ultraviolet-visible light (UV-Vis), Raman, fluorescence, electron paramagnetic resonance (EPR), and solid-state nuclear magnetic resonance (NMR) spectroscopies. Many spectroscopic studies combine several methods to give further structural information and to validate the respective data from each technique. For example, Chen et al. [10] applied several spectroscopic techniques, including FTIR, UV/Vis, EPR and 13C NMR, to characterize soil humic acid (SHA) and polyphenolic-rich (NOM-PP) and carbohydrate-rich (NOM-CH) fractions from an aquatic NOM. Specific functional group composition was discussed and compared among samples. Their results for the NMR analysis showed an abundance of aromatic C=C and methoxyl functional groups in the order SHA>NOM-PP>NOM-CH; carboxylic and alcoholic groups in the order NOM-CH>NOMPP>SHA; and ketonic and phenolic groups in the order NOM-PP>NOM-CH>SHA. The FTIR results indicated the presence of condensed aromatic rings and carbonyl and carboxyl groups in SHA and the presence of low aromatic content and hydroxyl and methoxyl groups in the NOM fractions. Similar differences in functional group content between aquatic- and terrestrialbased NOM have been found in several other studies [11–21]. NMR (e.g., 1H, 13C or 15N), as compared to other spectroscopic techniques, provides especially valuable insights into functional group assignments and their quantification as a means of advancing our understanding of the structural composition of humic materials (e.g., [22–29]). NMR-derived information includes the relative contribution of aliphatic, aromatic, carbonyl, carboxyl, phenolic, methoxyl, and nitrogenous groups. In particular, the development of two-dimensional (2-D) NMR has led to improved understanding of aromatic functionalities and structural connectivities since 2-D NMR techniques can reduce peak overlap problems that may occur in one-dimensional (1-D) NMR. Simpson et al. [30] applied 2-D Correlation Spectroscopy (COSY) and Heteronuclear Single Quantum Coherence (HSQC) NMR to describe the connectivities of a fulvic acid. Of interest was the dependence of peak resolution on the homogeneity of the samples. The well-fractionated fulvic acid samples were depicted as mixtures of free fatty acid chains of 6, 7 and 10 carbon units connected by way of ester or ether linkages to predominately aliphatic structures. Mao et al. [31] studied the chemical structure of peat humic acid using two-dimensional 1H-13C heteronuclear correlation solid-state NMR (HETCOR). Results showed that OCH3 groups are linked to aromatic rings. Consequently, one-third of the aromatic C-O groups is COCH3, not C-OH. C-CH3 groups were near alkyl and aliphatic groups, not aromatic groups. Protonated acetal O-CHR-O and unprotonated ketal O-CRR’-O were identified. COO groups were mostly in OCH, but some were bonded to aromatic rings and aliphatic groups. Additional NMR studies have provided improved insights into NOM functional group composition, structural conformation, and chain mobility as samples are subjected to changes in pH or temperature, or interact with organic compounds. Chien and Bleam [32] illustrated pH-dependent conformational changes of a humic acid using 2-D homonuclear 1H nuclear Overhauser effect spectroscopy (NOESY). Two domains separated as an aromatic structure and an aliphatic structure emerged as pH was decreased. The authors found that greater flexibility resulted from aliphatic structures as compared to aromatic structures, suggesting multiple-mobility domains similar to rubbery and glassy domains suggested by LeBoeuf and Weber [33]. Mobility differences in structural models of three humic acids were also evaluated using onedimensional proton wideline and two-dimensional WISE (wide-line separation) solid-state NMR [34]. Three humic acids were derived from a peat soil, a mineral soil, and a Mollisol. The 1-D proton measurements revealed the mobility of certain regions of the samples after exposure to increasing temperatures. The use of 2-D NMR aided in separating the respective mobility of these regions into corresponding functional groups, where aliphatic, carbohydrate, and aromatic groups exhibited three types of mobility, and carboxyl groups displayed two types of mobility. NMR is also providing significant insights to organic compound interaction with humic materials and soils. For example, Kacker et al. [35] studied the fate of phenanthrene in different humic acid fractions, revealing ester-bound derivatives of phenanthrene associated with carboxyl groups. DePaolis and Kukkonen [36] observed a greater binding capacity of humic acid over fulvic acid for benzo(a)pyrene and pentachlorophenol solutions, and acetone [37] interactions in soil fractions provided evidence of hydrogen bonding between the carbonyl group of acetone and the hydroxyl protons in clays. Advanced spectroscopic methods have enabled researchers to understand the structural composition of humic substances in more detail. In addition to revealing the presence of specific functional groups, the ability to locate group connectivity and mobility, to a certain degree, is now possible through use of advanced NMR techniques. This information greatly benefits the development of more accurate NOM models, allowing for adjustments of functional group composition and connectivity as a means to further constrain existing models. Molecular simulations of NOM, however, have additional characterization capabilities—namely thermodynamic properties, such as heat capacity and thermal expansion coefficients. This information can only be provided through thermal analysis techniques.

42

| CHAPTER 6: A PROPOSAL FOR THE ESTABLISHMENT OF A DATABASE

Figure 6.1 DSC trace of a typical synthetic polymer

6.3. THERMAL ANALYSIS Recent advances in the use of thermal analysis techniques have also provided understanding of the physical conformation of NOM, including influences of temperature on macromolecular mobility as evidenced by glass transition temperatures (Tg). Tg represents a demarcation in macromolecular mobility, where temperatures below a sample’s Tg coincide with a relatively rigid, or glass-like matrix with “frozen-in” microvoids, while temperatures above a sample’s Tg coincide with a relatively fluid, or rubber-like matrix. The differing mechanical behavior of the glassy and rubbery states leads to differences in how contaminants interact with the matrix. For example, sorption in glassy systems is typically non-Fickian in nature, with nonlinear sorption isotherms, while sorption in rubbery systems may display isotherm linearity [5]. Thermal analysis may be performed by several methods—thermogravimetry (TG), differential thermal analysis (DTA), differential scanning calorimetry (DSC), temperature-modulated differential scanning calorimetry (TMDSC), thermoelectrometry, thermomechanical analysis (TMA), dilatometry, and positron annihilation lifetime spectroscopy (PALS). These techniques provide measurements of thermodynamic functions of state, such as temperature, pressure, volume, heat capacity, enthalpy, and mass. For example, calorimetric instruments measure heat, which can be linked to functions of state such as heat capacity, total energy, and enthalpy as discussed in the following section on glass transition temperature. Other thermal techniques pertain to mechanical properties measured through thermomechanical analysis and dilatometry— techniques that measure the state functions of length and volume, respectively. These two techniques provide a means to measure the glass transition temperature. The thermodynamic basis for the glass transition originates from observed changes in the dissipation of energy (heat capacity) or in volume (expansion), as illustrated in Figures 6.1 and 6.2, respectively. Rubbery behavior results in either greater dissipation of energy or greater ease in volume expansion because of the larger molecular motions [38]. Figure 6.1 is a differential scanning calorimeter scan of a synthetic organic macromolecule. Transitions occur at marked changes, represented by either an increase (endotherm) or a decrease (exotherm) in heat flow, from the baseline. Scanning from lower to higher temperatures, the first transition encountered is a glass transition, representing the temperature at which a sample transcends from a glassy state to a rubbery state. This transition typically is associated with an abrupt change in heat flow, representing the change in constant-pressure specific heat capacity ( Cp) of the sample between the glassy and rubbery states. Following a glass transition, a sample may have enough mobility in the rubbery state for individual macromolecular segments to align and undergo a crystallization (exothermic response), followed by a crystalline melt (endothermic response with little change in heat capacity following the melt). Figure 6.2 is a volume versus temperature scan for a similar macromolecule. The slope of the line represents the thermal expansion coefficient, , and the glass transition occurs at the point at which the slope increases sharply, marking the transition from a lower associated with the glassy state, to a higher representative of the increased mobility of the rubbery state. The intersection of the tangents of the two lines representing the thermal expansion of the glassy and rubbery states defines the Tg. The abrupt change in heat flow or volume observed in Figures 6.1 and 6.2, re spectively, is illustrated mathematically using thermodynamic principles, beginning with the Gibbs energy Eq. 6.1, (6.1) where enthalpy, H, is defined as H=U+pV, T is temperature, and S is entropy. Rewriting the Gibbs energy using the substitution for enthalpy leads to Eq. 6.2. (6.2)

SPECTROSCOPIC TECHNIQUES |

43

Figure 6.2 TMA scan of a typical synthetic polymer

Taking the second derivative of the Gibbs energy with respect to temperature and pressure results in Eqs. 6.3 and 6.4 (6.3) (6.4) where (6.5) The coefficient of thermal expansion, , and heat capacity, Cp, exhibit a discontinuity at the Tg [39], indicative of a secondorder thermodynamic transition. Although the above argument is a thermodynamic explanation for glass transi tion behavior, experimental measurements of samples with differing thermal histories may produce a wide range of Tgs. For example, a sample heated at a higher rate may show a glass transition at a higher temperature than when the sample is heated at a lower rate [40]. This is due to a characteristic relaxation time defined as the rate at which a macromolecule can accommodate incoming energy [39]. If the experimental time is longer than the characteristic time (low rate of heating), the transition occurs at or near the “true” glass transition temperature. On the other hand, if the experimental time is shorter than the characteristic time (high rate of heating), the macromolecule experiences a lag in the rate that it can accommodate the extra energy, and thus there is a shift of the Tg to higher temperatures. Therefore, the Tg depends on the heating rate in the experiment. LeBoeuf and Weber [33] used differential scanning calorimetry (DSC) as a means to quantify glass transition behavior of Aldrich humic acid, and further related the differences in sorption behavior of phenanthrene within the matrix to the presence of glassy or rubbery domains. Subsequent work of LeBoeuf and Weber [40,41], Young and LeBoeuf [42], and Schaumann et al. [43] revealed glass transition behavior of additional NOM samples, while LeBoeuf and Weber [44] illustrated the effects of differing mechanical mobility on contaminant sorption and desorption behavior. Recently, DeLapp et al. [45] quantified thermodynamic properties of additional International Humic Substances Society aquatic- and terrestrial-based humic materials, including specific heat capacity, glass transition temperature, and thermal expansion coefficients. Yu et al. [46] subsequently compared glass transition and thermal expansion coefficient information derived from molecular dynamics simulations of a peat humic acid model to actual experimental data. This favorable correlation suggests that thermodynamic information may also be used to further constrain molecular simulations of NOM, potentially leading to improved structural models with enhanced predictive capabilities. 6.4. MOLECULAR MODELING Computer simulations of chemical structures use classical mechanics to describe the motions of atoms and molecules. The core of simulation calculations derives from quantum mechanics [47]. Briefly, quantum mechanics solves the Schrödinger equation (6.6)

44

| CHAPTER 6: A PROPOSAL FOR THE ESTABLISHMENT OF A DATABASE

where H is the Hamiltonian, is a wave function for the electrons, and E is the energy eigenvalue for the system. An understanding of transitional states and electron orbitals from model predictions of vibrational modes and spectral properties results from these calculations. Except for solving the Schrödinger equation for a small system of molecules, a quantum mechanical approach is not recommended for NOM systems due to the large computational requirements [47]. More practical models use first principles and semi-empirical methods. First principles simulations include Density Functional Theory, yielding information related to the total energy, electron density, and wave functions of a system [48]. A second technique uses the Hartree-Fock method to produce information on electronic charge distribution and infrared frequencies [49]. Again, computation time becomes a liability when applying these methods [50] to larger natural macromolecules. Semi-empirical methods require less computational time but lose accuracy in detailing electronic structure due to the empirically-derived parameters used in describing valence electrons [50]. Molecular mechanics (MM) and molecular dynamics (MD) are examples of semi-empirical atomistic models. These methods behave similarly to force field-based simulations. The Dreiding force field [51] expresses the interaction between atoms (V) as contributions from intermolecular (non-bond) and intramolecular energies. (6.7) The non-bond energy derives from interactions between atoms that are not bonded to each other. This interaction includes steric (van der Waals) and electrostatic (Coulombic) components. The intramolecular energy consists of a connectivity (bond stretching) and flexibility (angle and torsion) components [52]. Molecular mechanics simulations are used in optimizing structural information based on energy minimization. Energy minimization involves the calculation of the potential energy and its first derivative with respect to atomic coordination. A minimum energy shows a first derivative of zero. Molecular dynamics follow similar calculations, but with time and temperature added into the simulation to provide insights into molecular mobility, including diffusion, vibration, and conformational stability. 6.5. MODELS OF HUMIC MATERIALS Molecular models of humic materials have been derived with numerous methods— experimental data from pyrolysis and spectroscopic techniques and molecular simulations. Earlier structures included Fuchs’ model, representing a coal humic acid structure as a condensed ring of carboxylate and hydroxylate groups [1], and the models of Flaig [53] and Dragunov [54]. Flaig’s model [53] is comprised of linked aromatic, phenolic, or quinonic rings while Dragunov’s model [54] consists of diand tri-hydroxyphenol rings [31]. Steelink’s model [55] includes phenols, quinones, and aliphatic links with phenolic groups. More recent models reveal the use of additional functional group information provided by advanced spectroscopic techniques. For example, two-dimensional models of humic acid featuring a humic acid C-C skeleton, as well as COOH, ketone, and aliphatic and aromatic OH functional groups were derived from pyrolysismass spectrometry studies [56,57]. Using a twodimensional structure from prior work, Schulten and Schnitzer [58–60] constructed a three-dimensional humic acid structure with the aid of molecular simulation. Molecular mechanics simulations were employed, involving adjustments in forces on atoms based on their bond lengths, bond angles, torsion angles, non-bonded interactions, electrostatic interactions, and hydrogen bonds. The final structure reflected the most favorable, energy-minimized conformation, with a molecular weight of approximately 5,500 Da, and elemental analysis representing 66.69% carbon, 6.09% hydrogen, 25.96% oxygen, and 1.26% nitrogen. A particular feature of this structure is the presence of voids that can act as traps for organic compounds. Other models of humic materials were influenced by modeling sorbent and contaminant interactions. Leenheer [61] performed studies on a Suwanee River fulvic acid to evaluate metal binding functional groups using 13C NMR, 1H NMR, and FTIR. A modeling approach to observe metal binding was followed. Model constraints were acquired from quantitative elemental, molar mass, spectrometric, and titrimetric data. For example, carbons assigned were based on 13C NMR data and oxygen from both 13C NMR and FTIR spectroscopies. Additional studies using NMR data for the molecular modeling of contaminant transport included a model of organo-mineral interactions of pesticides on soil surfaces [62], the modeling of humic complexes of diethyl phthalate [60,63] to reveal a hole-filling mechanism describing the interactions, and various modeling techniques in predicting the role of natural organic matter and structure in the sequestration of organic contaminants [50]. Existing models of NOM may also be improved through employment of spectroscopic information and molecular simulations. Mao et al. [64] used 13C NMR data to resolve discrepancies between model and experimental functional group distribution for several different models. For example, Leenheer’s fulvic acid model [65,66] showed a discrepancy with sp2-C data. A modification that included decreasing the −COO− content and increasing the phenolic and −OCH3 contents in the model produced better agreement. Steelink’s model [55] was considered a good building block for peat humic acid upon

SPECTROSCOPIC TECHNIQUES |

45

addition of −OCH3 and -CH groups, while a model of Schulten and Schnitzer [57] showed a resemblance to Aldrich humic acid. Each of the aforementioned studies used spectroscopic information and molecular simulations either to create a new model or revise existing models. Conversely, molecular modeling can predict molecular structure and conformation to estimate experimental data. The following sections describe in more detail the role of molecular modeling in creating compositional information on macromolecules. In particular, the role of thermodynamic properties in structural modeling and predicting experimental data are emphasized. 6.6. THERMODYNAMIC PROPERTIES AS MOLECULAR CONSTRAINTS Molecular simulation has been used to better understand conformational behavior and observed mobilities of synthetic macromolecules, including Tgs. Prediction of Tgs using molecular dynamics simulation typically employs use of volumetemperature (V-T) curves, where the volume of a periodic cell of a macromolecule is evalu ated at several different temperatures following energy minimization. Conformational analyses of liquid crystal polymers (LCPs) were performed using molecular dynamics simulations [51]. Monotropic LCPs formed helicoidal structures, whereas enantiotropic LCPs formed nearly linear structures. Molecular simulations of LCPs also provided predicted glass transition and thermal decomposition temperatures. Pavel et al. [67] carried out molecular dynamics simulations and conformational analyses on aromatic polymers containing oxetane rings in the main chain. Predicted Tgs were comparable to experimental Tgs derived from differential scanning calorimetry. Similarly, molecular dynamics simulations of a wide variety of other materials show close agreement with experimental data (e.g., polyethylene [68–70] polyvinylchloride [71], PMMA and PMA [72], polyphosphazenes [73], disubstituted polysilanes [74], polyhedral oligomeric silsesquioxane [75], polypropylene, polyisobutylene, polyoxymethylene and polydimethylsiloxane [70], amorphous CaAl2Si2O8 [76], and DNA [77]). Evaluations of macromolecules with varying backbones and/or functional side groups can provide insights into the influences of varying chemical composition and structure on macromolecular mobility. For example, the glass transition behavior of five polymers with different main chain backbones (C-C, Si-O, and C-O) and different side groups (-H, one −CH3, and two −CH3) was simulated through molecular dynamics [70]. Results show that the Si-O bond has a lower Tg than the C-C main chain. However, comparisons of C-C and C-O backbones show inconclusive results. Chain conformations of the polymers may also be analyzed below and above the Tg to determine the role of torsional interactions. The analysis reveals that the glass transition is associated with the freezing of the torsional degrees of the polymer chains. A simulated dilatometry method from molecular dynamics simulations is used to calculate the glass transition of poly(methyl methacrylate) (PMMA) and poly(methyl acrylate) (PMA) [72]. As in experimental dilatometry, Tg is measured from a plot of the specific volume against temperature. The addition of the methyl group in PMMA results in a higher Tg because the intermolecular interactions and van der Waal forces are stronger. Structural models of humic acids have become more sophisticated due to advances in spectroscopic techniques and model simulations. NMR experimental data have revealed many functional groups, as well as their interconnectivities and mobilities. Molecular mechanics and dynamics simulations are currently used to model contaminant interactions in soil organic matter. As illustrated, experimental data have been used to assist in constraining molecular simulation models. Similarly, data from simulated models help predict molecular properties. Currently, models of humic substances derived from molecular modeling have not extended its capabilities beyond structural composition. Additional information, specifically those detailing thermodynamic properties, will give insight into molecular motions, as well as structural composition. The following section emphasizes the need to establish a database of thermodynamic properties as a means to improve our understanding of NOM structure and conformation. 6.7. A DATABASE OF THERMODYNAMIC PROPERTIES Currently, there exists no collection of thermodynamic properties of NOM. Quantifying Tgs, s, and Cps for a variety of NOM within a database of thermodynamic properties provides an opportunity to relate physical and chemical properties to thermodynamic parameters, similar to methods employed for biomaterials [78] and synthetic polymers [79,80]. For example, Young and LeBoeuf [42] suggested that the higher Tg observed for a peat humic acid than for a Suwanee River fulvic acid may be attributed to its higher molar mass (higher molar mass tends to result in higher Tgs due to decreased macromolecular mobility from a decreased number of free end groups) and its more condensed aromatic structure (more aromatic structure typically suggests stiffer bonds). ProTherm, a database of thermodynamic properties of protein materials, was developed to supply insights into the mechanism of protein stability [78]. The database includes structural information, experimental methods and conditions, and

46

| CHAPTER 6: A PROPOSAL FOR THE ESTABLISHMENT OF A DATABASE

thermodynamic data obtained from denaturation experiments. These data cover unfolding Gibbs free energy, enthalpy change, heat capacity change, and transition temperatures for 57 protein samples. The Advanced Thermal Analysis Laboratory (ATHAS) was established to investigate experimental and mathematical relationships between heat capacity and structural information for linear macromolecules [81]. An ATHAS data bank [79,80] furnishes equilibrium information on heat capacities, as well as vibrational spectra, transition parameters and free enthalpies, for over 200 synthetic polymers. Conformational motion is revealed from comparisons of experimental data and vibrational heat capacities. Beyond supplying structural information as illustrated by ProTherm and ATHAS, the database of thermodynamic properties for NOM can be used as constraints for molecular dynamics simulation models (e.g., [46]). To be able to utilize thermodynamic data from various sources, however, standard experimental protocols must be established to ensure continuity of results among different users. 6.8. STANDARDIZATION OF THERMAL ANALYSIS PROCEDURES A variety of factors should be considered to ensure standardization and reproducibility of thermal analytical results. Factors include sample preparation, heating and cooling rates, and variables that are specific to the thermal method applied (e.g., loading force for TMA). Thermal characterization of humic materials has been primarily performed on NOM powders using DSC [33,40,43,45], TMDSC [42,45], and TMA [42,45]. Therefore, protocols for DSC, TMDSC, and TMA used to analyze anhydrous powders are presented. Specifically, procedures for using TA Instruments DSC/TMDSC 2920 and TA Instruments TMA 2940 are outlined. 6.8.1. DSC/TMDSC Experimental Protocols Sample materials typically are in powder form, and are stored in desiccators until used. Aluminum DSC sample pans and lids are washed with HPLC-grade acetone to remove residual oils from the pan manufacturing process, heated in a 440°C furnace to ash organic residues, and stored in desiccators for future use. Sample pans and lids are handled with forceps to avoid contamination. For DSC/TMDSC analyses involving heat capacity measurements, a microbalance is used to ensure that the mass of the reference pan/lid and the mass of the sample pan/lid are within ±0.1 mg. The pan is filled with approximately 10 to 20 mg of sample. For the most effective thermal contact, an adequate amount of powder is placed in the pan so that it covers the bottom and fills the pan to the rim. The lid is placed on the pan and crimped with a specially designed die. Development of standard operating procedures for DSC and TMDSC involves the adjustment of three primary variables— amplitude of temperature (TMDSC only), period of modulation (TMDSC only) and heating rate. The temperature amplitude and modulation period produce a sinusoidal change in heating rate in order to separate the total heat flow into its heat capacity and kinetic components [82]. The magnitude of the amplitude follows a rule of thumb that larger amplitudes provide more sensitivity in measurements. The period of modulation is dependent on the heat capacity and thermal conductivity of the sample. At higher settings, heat capacity is independent of periods of modulation. How rapidly the sample is heated is dictated by the heating rate in °C per minute. As discussed, higher heating rates produce larger heat flow signals at the glass transition (enthalpic overshoot). Multiple runs, with adjustments to these factors, were performed on Leonardite humic acid to decide which combination produces the strongest signal and most reproducible results of heat flow versus temperature. Following placement of the reference and sample pans in their respective DSC cells, the sample is heated under nitrogen (99.998%) at 10°C per minute from room temperature up to 100°C and is held at this temperature for 30 minutes to evolve physi-sorbed water. The sample is then cooled at 10°C per minute down to −130°C, and held for 5 minutes while the heat flow stabilizes. Analysis runs for DSC involve triplicate heating and cooling cycles based on heating rates of 10–120°C per minute, followed by cooling rates of 10–130°C per minute. TMDSC runs follow similar initial procedures as the DSC runs. The pans are loaded and then heated at 2.5°C per minute to 100°C and held for 30 minutes. A period of modulation of 90 seconds and amplitude of temperature of ±2°C are set before the first run. After the initial heating, the sample is cooled to −130°C and then heated to 120°C at the 2.5°C per minute heating rate. Two heating cycles are run to check reproducibility of the results. 6.8.2. TMA Experimental Protocol Samples for TMA are stored in a desiccator and handled with a spatula (powder form) or forceps (pellet form). A high carbon steel die, similar to the die used to form pellets for FTIR analyses, except with a 6.30-mm inner diameter and a pair

SPECTROSCOPIC TECHNIQUES |

47

Figure 6.3 DSC scan of Leonardite humic acid

of punches with a 6.25-mm diameter, is used to shape the powder. The die and punches are cleaned with a swab dipped in acetone before use. Approximately 50 mg (±0.5 mg) is put into the die assembly. A load of 450 N is applied for one minute on the punch to form the pellet. The resulting pellet measures approximately 6.25-mm in diameter and is between 0.8 to 1 mm thick (dependent on the humic material). The sample is stored in a desiccator for future use. Development of standard operating procedures for TMA also involved the adjustment of three primary variables—probe, loading force, and heating rate. The probe records the dimensional change that occurs as a sample is heated. It is attached to a movable-core linear variable differential transformer (LVDT), which produces a signal proportional to the linear displacement caused by the sample [83]. Three probes—expansion, macroexpansion, and penetration—can be used depending on the size and shape of the sample. The expansion and macroexpansion probes allow the sample to expand during heating. The head of the macroexpansion probe covers a larger surface area of the sample for better contact. The penetration probe penetrates the sample as it softens during heating. The loading force determines how much pressure is applied on the sample. How rapidly or slowly the sample is heated is dictated by the heating rate in °C per minute. Multiple runs, with adjustments among these three factors, were performed on Leonardite humic acid to decide which combination produces the strongest signal and the most reproducible results of dimension change versus temperature. Since both the probe and pellet diameters are identical, a macroexpansion probe was selected to allow for maximum probe contact with the pellet. The stage and the probe are cleaned with acetone before the sample is loaded. Two pellets, one stacked on top of the other, are loaded on the stage. After the probe is placed on the samples and the thickness measured, the samples are heated under nitrogen (99.998%) at 2.5°C per minute from room temperature to 100°C and held at this temperature for 30 minutes to evolve physi-sorbed water. The samples are then cooled to −50°C, held for 5 minutes, and heated again at 2.5°C per minute to 120°C. Three cooling and heating cycles were used to ensure reproducibility in the thermal scans. 6.8.3. Example Thermal Analysis Results Measurement of thermodynamic transitions in NOM can be challenging. Observation of a transition depends on analytical techniques sensitive enough to detect small changes in heat flow or dimension. The larger the amount of sample contributing to the transition, the larger the resulting heat flow or dimension change. The inherent heterogeneity of NOM, however, may only provide for one portion of the sample to undergo a transition at a particular temperature—as such, more homogeneous samples often provide improved results. More challenges arise after transitions are detected. Reliance on only one of the three thermal analysis techniques described here may not provide sufficient information to determine if an observed transition is a glass transition or some other type of thermal response. For example, most glass transitions are reversible—cycling a sample through multiple heating and cooling rates provides opportunities to confirm this reversible behavior. It is also highly recommended that two or more independent techniques be employed. For example, our laboratories typically employ TMA, DSC, and TMDSC, each through multiple cycles to confirm transition behavior. Figures 6.3 and 6.4 illustrate thermal scans of Leonardite humic acid (LHA) using DSC and TMDSC, respectively. Both Figures suggest glass transitions near 72–73°C, similar to the Tg reported for LHA in a previous study [41]. The TMDSC plot in Figure 6.4 differs from the DSC plot of Figure 6.3 in that the reversible heat flow is plotted instead of heat flow against temperature. Heat flow is the raw data signal from the instrument, indi cating the heat flow into the sample cell relative to the reference cell. The reversible heat flow signal represents a deconvoluted signal that separates the thermal scan into its kinetic components (e.g.,

48

| CHAPTER 6: A PROPOSAL FOR THE ESTABLISHMENT OF A DATABASE

Figure 6.4 TMDSC scan of Leonardite humic acid

Figure 6.5 TMA scan of Leonardite humic acid

the so-called enthalpic overshoot [40]), which are primarily artifacts of the experimental conditions, and true thermodynamic properties, such as Tg and change in specific heat capacity. TMDSC has the advantage of removing underlying kinetic effects, so concerns about misinterpretation of transitions due to enthalpic overshoots are removed, and only the underlying thermal response of the sample is displayed. Nonetheless, one should not conclude that the transition is a glass transition without additional data.

SPECTROSCOPIC TECHNIQUES |

49

However, use of TMA as the third thermal analysis technique (Figure 6.5) paints a clearer picture of the origin of the transition. Here, a transition is again measured at 73°C. Also revealing is the expansion that follows the transition without an accompanying “collapse” (i.e., melt) of the sample under a constant force. As a result, glass transition behavior may be confirmed. The coefficient of thermal expansion, ′ , is measured before and after the transition to quantify the mobility of the molecules. Table 6.1 summarizes measured thermodynamic properties of LHA. Techniques such as X-ray diffraction may also be used to explore the origin of the transition. For example, weak glass transitions may look similar in form to crys Table 6.1 Sample thermodynamic properties of Leonardite humic acid (after [45]) Sorbent

DSC Tg

TMDSC Tg

′ Cp at 0°C

′ Cp at 25°C

TMA Tg

′ Before Tg

(°C)

(°C)

(J/g°C)

(J/g°C)

(°C)

(µm/m°C)

(µm/m°C)

LHA

73

72

1.29

1.40

73

25.4

After Tg 43.7

talline melts. Evaluation of samples at temperatures below and above the transition using X-ray diffraction, which can detect regions of crystallinity, would show no change in signals over the temperature range of interest for a glass transition, while crystalline melt transitions would be reflected in a change in the X-ray diffraction pattern. 6.9. CONCLUSIONS Current research needs an improved understanding of the role of NOM in various environmental processes. The development of advanced characterization techniques is assisting this effort by improving the accuracy of predictive tools such as molecular simulations of NOM structures. Thermal analysis techniques provide a means to quantify thermodynamic properties of heat capacity, thermal expansion coefficient, and glass transition temperatures. Combined, these parameters impose substantial new constraints for existing molecular simulations models. However, thermodynamic data for NOM are scarce. The establishment of a database of thermodynamic properties of NOM samples, including standardized experimental protocols, is thus proposed to assist in filling this information void. ACKNOWLEDGEMENT Funding for this research was provided by the Biological and Environmental Systems Division of the National Science Foundation under Grant No. 998159. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11.

Stevenson FJ. Humus chemistry: Genesis, composition, reaction, 2nd Edn.; Wiley: New York, 1994. Chiou CT. Theoretical considerations for the partition uptake of nonionic organic compounds by soil organic matter. In: Sawhney BL, Brown K eds. Reactions and movement of organic chemicals in soils. Madison: Soil Science Society of America, 1989:1–29. Luthy RG, Aiken GR, Brusseau ML, Cunningham SD, Gschwend PM, Pignatello JJ, Reinhard M, Traina SJ, Weber WJ, Westall JC. Sequestration of hydrophobic organic contaminants by geosorbents. Environ. Sci. Technol., 1997; 31:3341–3347. Mader BT, UweGoss K, Eisenreich SJ. Sorption of nonionic, hydrophobic organic chemicals to mineral surfaces. Environ. Sci. Technol, 1997; 31: 1079–1086. Weber WJ, Jr., LeBoeuf EJ, Young TM, Huang W. Contaminant interactions with geosorbent organic matter: insights drawn from polymer sciences. Water Res., 2001; 4:853–868. Hedges JI. Polymerization of humic substances in natural environments. In: Frimmel FH, Christman RF eds. Humic substances and their role in the environment. New York: Wiley, 1988:45–58. Aiken GR, McKnight DM, Wershaw RL, MacCarthy P. Humic substances in soil, sediment, and water. New York: Wiley, 1985. Sparks DL. Environmental soil chemistry. San Diego: Academic Press, 1995. Hayes MHB, Swift RS. Progress towards understanding aspects of composition and structure of humic substances. In: Swift RS, Spark KM eds. Understanding and managing organic matter in soils, sediments, and waters. St. Paul: International Humic Substances Society, 2001:3–16. Chen J, Gu BH, LeBoeuf EJ, Pan HJ, Dai S. Spectroscopic characterization of the structural and functional properties of natural organic matter fractions. Chemosphere, 2002; 48:59–68. Ricca G, Severini F. Structural investigations of humic substances by FT-IR, C-13 NMR spectroscopy and comparison with maleic oligomer of known structure. Geoderma, 1993; 58:233–244.

50

12. 13. 14.

15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

25. 26. 27. 28. 29. 30.

31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41.

| CHAPTER 6: A PROPOSAL FOR THE ESTABLISHMENT OF A DATABASE

Deiana S, Gessa C, Manunza B, Rausa R, Solinas V. Iron(III) reduction by natural humic acids: A potentiometric and spectroscopic study. Eur. J. Soil Sci., 1995; 46:103–108. Preston CM. Applications of NMR to soil organic matter analysis: History and prospects. Soil Sci., 1996; 161:144–166. Fabbri D, Mongardi M, Montanari L, Galletti GC, Chiavari G, Scotti R. Comparison between CP/MAS 13C-NMR and pyrolysis-GC/ MS in the structural characterization of humins and humic acids of soil and sediments. Fresenius J. Anal. Chem., 1998; 362: 299–306. Davis WM, Erickson CL, Johnston CT, Delfino JJ, Porter JE. Quantitative Fourier transform infrared spectroscopic investigation of humic substance functional group composition. Chemosphere, 1999; 38:2913–2928. Gu ZM, Wang XR, Gu XY, Cao XD. Characterization of humic acid extracted from different soils by Fourier transform infrared spectrometry and nuclear magnetic resonance spectroscopy. Chinese J. Anal. Chem., 2000; 28:314–317. Poirier N, Derenne S, Rouzard JN, Largeau C, Mariotti A, Balesdent J, Maquet J. Chemical structure and sources of macromolecular, resistant, organic fractions isolated from a forest soil (Lacadee, southwest France), Org. Geochem., 2000; 31:813–827. Joly HA, Li HB, Belzile N. Pyrolysis gas chromatography—mass spectrometry of humic substances extracted from Canadian lake sediments. Can. J. Chem., 2000; 78:51–63. Francioso O, Sanchez-Cortes S, Casarini D, Garcia-Ramos JV, Ciavatta C, Gessa C. Spectroscopic study of humic acids fractionated by means of tangential ultrafiltration. J. Mol. Struc., 2002; 609:137–147. Francioso O, Sanchez-Cortes S, Tugnoli V, Marzadori C, Ciavatta C. Spectroscopic study (DRIFT, SERS and H-1 NMR) of peat, leonardite and lignite humic substances. J. Mol. Struc., 2001; 565:481–485. Xing B, Chen ZQ. Spectroscopic evidence for condensed domains in soil organic matter. Soil Sci., 1999; 164:40–47. Thorn KA, Kennedy KR. N-15 NMR investigation of the covalent binding of reduced TNT amines to soil humic acid, model compounds, and lignocellulose. Environ. Sci. Technol., 2002; 36:3787–3796. Simpson AJ, Salloum MJ, Kingery WL, Hatcher PG. Improvements in the two-dimensional nuclear magnetic resonance spectroscopy of humic substances. J. Environ. Qual., 2002; 31:388–392. Hertkorn N, Permin A, Perminova I, Kovalevskii D, Yudov M, Petrosyan V, Kettrup A. Comparative analysis of partial structures of a peat humic and fulvic acid using one- and two-dimensional nuclear magnetic resonance spectroscopy. J. Environ. Qual., 2002; 31: 375–387. Hatcher PG, Dria KJ, Kim S, Frazier SW. Modern analytical studies of humic substances. Soil Sci., 2001; 166:770–794. Mathers NJ, Mao XA, Xu ZH, Saffigna PG, Berners-Price SJ, Perera MCS. Recent advances in the application of C-13 and N-15 NMR spectroscopy to soil organic matter studies. Austr. J. Soil Res., 2000; 38:769–787. Kingery WL, Simpson AJ, Hayes MHB, Locke MA, Hicks RP. The application of multidimensional NMR to the study of soil humic substances. Soil Sci., 2000; 165:483–494. Cook RL, Langford CH. Structural characterization of a fulvic acid and a humic acid using solid-state ramp-CP-MAS C-13 nuclear magnetic resonance. Environ. Sci. Technol., 1998; 32:719–725. Bortiatynski JM, Hatcher PG, Knicker H. NMR techniques (C, N, and H) in studies of humic substances. In: Gaffney JS, Marley NA, Clark SB eds. Humic and fulvic acids. Washington DC: ACS Symposium Series, 1996; 651:57–77. Simpson AJ, Burdon J, Graham CL, Hayes MHB. Humic substances from podzols under oak forest and a cleared forest site. II. Spectroscopic methods. In: Hayes MHB, Wilson WS eds. Humic substances, peats, and sludges: Health and environmental aspects. Cambridge: Royal Society of Chemistry, 1997: 83–92. Mao JD, Xing B, Schmidt-Rohr K. New structural information on a humic acid from two-dimensional H-1–C-13 correlation solidstate nuclear magnetic resonance. Environ. Sci. Technol., 2001; 35:1928–1934. Chien YY, Bleam WF. Two-dimensional NOESY nuclear magnetic resonance study of pH dependent changes in humic acid conformation in aqueous solution. Environ. Sci. Technol., 1998; 32:3653–3658. LeBoeuf EJ, Weber WJ, Jr.. A distributed reactivity model for sorption by soils and sediments. 8. Discovery of a humic glass transition and an argument for invoking polymer sorption theory. Environ. Sci. Technol., 1997; 31:1697–1702. Mao J, Ding G, Ning B. Domain mobility of humic acids investigated with one-and two-dimensional nuclear magnetic resonance: Support for dual-mode sorption model. Commun. Soil Sci. Plant Anal., 2002; 33:1679–1688. Kacker T, Haupt ETK, Garms C, Francke W, Steinhart H. Structural characterisation of humic acid-bound PAH residues in soil by C-13−CPMAS-NMR-spec troscopy: Evidence of covalent bonds. Chemosphere, 2002; 48:117–131. DePaolis F, Kukkonen J. Binding of organic pollutants to humic and fulvic acids: Influence of pH and the structure of humic material. Chemosphere, 1997; 34:1693–1704. Jurkiewicz A, Maciel GE. Solid-state C-13 NMR studies of the interaction of acetone, carbon tetrachloride and trichloroethylene with soil components. Sci. Total Environ., 1995; 164:195–202. Cowie JMG. Polymers: Chemistry and physics of modern materials, 2nd Edn. New York: Chapman and Hall, 1991. McKenna GB. Glass formation and glassy behavior. In: Booth C, Price C eds. Comprehensive polymer science: the synthesis, characterization, reactions, and applications of polymers, volume II: Polymer properties. New York: Pergamon, 1989:311–362. LeBoeuf EJ, Weber WJ, Jr.. Macromolecular characteristics of natural organic matter. 1. Insights from glass transition and enthalpic relaxation behavior. Environ. Sci. Technol., 2000; 34:3623–3631. LeBoeuf EJ, Weber WJ Jr.Characterisation of natural organic matter: a thermal analysis approach. In: Swift RS, Sparks KM eds. Understanding and managing organic matter in soils, sediments, and waters. St. Paul: International Humic Substances Society, 2001: 17–34.

SPECTROSCOPIC TECHNIQUES |

42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77.

51

Young KD, LeBoeuf EJ. Glass transition behavior of a peat humic acid and a stream fulvic acid. Environ. Sci. Technol., 2000; 34: 4549–4553. Schaumann GE, Antelmann O. Thermal characteristics of soil organic matter measured by DSC: A hint of a glass transition. J. Plant Nutrit. Soil Sci., 2000; 163:179–181. LeBoeuf EJ, Weber. WJ, Jr. Macromolecular characteristics of natural organic matter. 2. Sorption and desorption behavior. Environ. Sci. Technol., 2000; 34: 3632–3640. DeLapp RC, LeBoeuf EJ, Young KD. Characterization of thermodynamic properties of several soil- and sediment-derived natural organic materials. Environ. Sci. Technol., 2002; in preparation for submission. Yu W, LeBoeuf EJ, Kosson D. Molecular dynamics simulations of the coefficients of thermal expansion and glass transition temperatures in a model humic acid. Environ. Sci. Technol., 2002; in preparation for submission. Frenkel D, Smit B. Understanding molecular simulation from algorithms to applications. San Diego: Academic Press, 1996. Becke AD. Density functional thermochemistry. 3. The role of exact exchange. J. Chem. Phys., 1993; 98:5648–5652. Liu SX, Li Y, Cao H, Xie XG, Liu CQ. Comparison of quantum chemistry calculation methods of nucleic acid base-quartets. Chinese J. Struct. Chem., 2001; 20:313–318. Kubicki JD, Apitz SE. Models of natural organic matter and interactions with organic contaminants. Org. Geochem., 1999; 30: 911–927. Pavel D, Ball J, Bhattacharya S, Shanks R, Hurduc N, Catanescu O. Molecular simulation and experimental characterisation of monotropic and enantiotropic polymers containing azobenzene and diphenyl mesogens. Comput. Theor. Poly. Sci., 2001; 303–318. Soldera A, Grohens Y. Local dynamics of stereoregular PMMAs using molecular simulation. Macromolecules, 2002; 35:722–726. Flaig W. Chemistry of humic substances in relation to coalification. Landbauforschung Volkenrode, 1966; 16:155. Dragunov SS. Fractional composition of humic acids. Soviet Soil Sci., 1969; 3: 375. Steelink C. Electron paramagnetic resonance studies of humic acid and related model compounds. Adv. Chem. Ser., 1966; 55:80. Schulten HR, Schnitzer M. Structural studies on soil humic acids by Curie-point pyrolysis-gas chromatography/mass spectrometry. Soil Sci., 1992; 153:205– 224. Schulten HR, Schnitzer M. A state of the art structural concept for humic substances. Naturwiss., 1993; 80:29–30. Schulten HR. The 3-dimensional structure of humic substances and soil organic matter studied by computational analytical chemistry. Fresenius J. Anal. Chem., 1995; 351:62–73. Schulten HR, Schnitzer M. Chemical model structures for soil organic matter and soils. Soil Sci., 1997; 162:115–130. Schulten HR. Interactions of dissolved organic matter with xenobiotic compounds: Molecular modeling in water. Environ. Toxicol. Chem., 1999; 18: 1643–1655. Leenheer JA, Brown GK, MacCarthy P, Cabaniss SE. Models of metal binding structures in fulvic acid from the Suwannee River, Georgia. Environ. Sci. Technol., 1998; 32:2410–2416. Shevchenko SM, Bailey GW. Non-bonded organo-mineral interactions and sorption of organic compounds on soil surfaces: A model approach. Theochem- J. Mol. Struc., 1998; 422:259–270. Schulten HR, Thomsen M, Carlsen L. Humic complexes of diethyl phthalate: Molecular modelling of the sorption process. Chemosphere, 2001; 45:357–369. Mao JD, Hu WG, Schmidt-Rohr K, Davies G, Ghabbour EA, Xing B. Quantitative characterization of humic substances by solidstate carbon-13 nuclear magnetic resonance. Soil Sci. Soc. Amer. J., 2000; 64:873–884. Leenheer JA, Wershaw RL, Reddy MM. Strong-acid, carboxyl group structures in fulvic-acid from the Suwannee River, Georgia 1. Minor structures. Environ. Sci. Technol., 1995; 29:393–398. Leenheer JA, Wershaw RL, Reddy MM. Strong-acid, carboxyl group structures in fulvic-acid from the Suwannee River, Georgia 2. Major structures. Environ. Sci. Technol., 1995; 29:399–405. Pavel D, Ball J, Bhattacharya S, Shanks R, Hurduc N. J. Polym. Sci. Part B: Polym. Phy., 1999; 37:2334–2352. Roe RJ. Short time dynamics of polymer liquid and glass studied by molecular dynamics simulation. J. Chem. Phys., 1994; 100: 1610–1619. Gee RH, Boyd RH. The role of the torsional potential in relaxation dynamics: a molecular dynamics study of polyethylene. Comput. Theor. Polym. Sci., 1998; 8:93–98. Yu K, Li Z, Sun J. Polymer structures and glass transition: A molecular dynamics simulation study. Macromol. Theory Simulat., 2001; 10:624–633. Abu-Sharkh BF. Glass transition temperature of polyvinylchloride from molecular dynamics simulation: Explicit atom model versus rigid CH2 and CHCl groups model. Comput. Theor. Polym. Sci., 2001; 11:29–34. Soldera A. Energetic analysis of the two PMMA chain tacticities and PMA through molecular dynamics simulations. Polymer, 2002; 43:4269–4275. Fried JR, Ren P. Atomistic simulation of glass transition of polyphosphazenes. Comput. Theor. Polym. Sci., 1999; 9:111–116. Fried JR, Li B. Atomistic simulation of the glass transition of disubstituted polysilanes, Comput. Theor. Polym. Sci., 2001; 11: 273–281. Bharadwaja RK, Berrb RJ, Farmerb BL. Molecular dynamics simulation study of norbornene-POSS polymers. Polymer, 2000; 41: 7209–7221. Morgan, NA, Spera FJ. Glass transition, structural relaxation, and theories of viscosity: A molecular dynamics study of amorphous CaAl2Si2O8. Geochim. Cosmochim. Acta, 2001; 65:1087–1092. Norberg J, Nilsson L. Glass transition in DNA from molecular dynamics simulations. Proc. Nat. Acad. Sci., 1996; 93:10173–10176.

52

78. 79. 80. 81. 82. 83. 84.

| CHAPTER 6: A PROPOSAL FOR THE ESTABLISHMENT OF A DATABASE

Gromiha MM, Jianghing A, Kono H, Oobatake M, Uedaira H, Sarai A. ProTherm: Thermodynamic database for proteins and mutants. Nucl. Acids Res., 1999; 27:286–288. Wunderlich B. Thermal analysis. San Diego: Academic Press, 1990. Wunderlich B. The ATHAS database on heat capacities of polymers. Pure Appl. Chem., 1995; 67:1019–1026. Wunderlich B, Gaur U. The role of the heat capacity data bank in recent advances in the understanding of the glass transition. Bull. Amer. Phys. Soc., 1981; 26:325. Thomas LC. Use of multiple heating rate DSC and modulated temperature DSC to detect and analyze temperature-time dependent transitions in materials. Amer. Lab., 2001; 33:26–29. Hassel RL. Thermomechanical analysis instrumentation for the characterization of materials. Amer. Lab., 1991; 23:30–35. Grulke EA. Solubility parameter rules. In: Brandup J, Immergut EH eds. Polymer handbook. New York: Wiley-Interscience, 1989: 519–560.

Part 2 HYDRATION, SWELLING AND SORPTION: CONTRIBUTING FACTORS

Chapter 7 EFFECT OF HYDRATION/SOLVATION OF ORGANIC MATTER ON SORPTION OF ORGANIC COMPOUNDS: CONCEPTION AND SORPTION ISOTHERM MODEL Ellen R.Graber and Mikhail Borisover Institute of Soil, Water and Environmental Sciences, The Volcani Center, ARO, P.O.B. 6 Bet Dagan 50250, Israel 7.1. INTRODUCTION The effect of hydration of mineral surfaces on sorption of organic compounds is well known and documented [1–3]. In the dry state, minerals have high sorption capacity for organic compounds. Upon hydration however, water molecules are preferentially adsorbed to mineral surfaces, reducing mineral sorption capacity for organic compounds substantially. In fact, because of this effect, natural organic matter is by far the most important sorbent for organic compounds in hydrated systems [4]. Yet surprisingly, and in contrast to the substantial literature concerning the mineral hydration effect, there is relatively little information related to the hydration effect on sorption of organic compounds by natural organic matter in soils and sediments. Considering that natural organic matter (NOM) consists in large part of heterogeneous polyelectrolytic macromolecules, hydration can be expected to result in swelling, increased flexibility, changes in ionization status of polar functional groups, and conformational reorientation of the macromolecules [5–8]. These hydration-related changes in NOM structure may affect sorbate diffusion in NOM and the kinetics of sorption and binding at NOM sites. Hydration-related phenomena such as watersorbate competition, change of sorbate speciation in the NOM phase, direct interaction between complexed water molecules and sorbate molecules in the NOM phase, change in the total volume of the NOM phase, and change in NOM polarity all may be expected to contribute to the overall hydration effect on sorbate interactions in NOM. Although certain of these aspects of NOM hydration appear to be straightforward, the net effect of NOM hydration on organic compound sorption in teractions is far from self-evident due to the scarcity of experimental organic compound sorption data in both dry and wet NOM, and the complexity of NOM hydration phenomena [9–11]. Based on experimental data for sorption of organic compounds by dry and hydrated (or solvated by active organic solvents) NOM, we developed a concept of water (solvent)-disruptable non-covalent links in NOM [9,10]. According to this idea, certain polar moieties of dry NOM are unavailable for compound sorption due to strong interactions (e.g., H-bonding, proton transfer complexation, bridging via metal cations). By penetrating into the NOM structure, water or active solvent molecules solvate these polar moieties, creating new NOM sorption sites for sorbate molecules at the solvated (disrupted) contacts. Since a sorbate molecule is not able to interact with the moiety in the non-solvated state, it is clear that if the solvent disrupts a polar contact by solvating only one moiety of the contact, the sorbate molecule would not be able to replace the solvent molecule at that moiety. Mechanistically, this means that solvent molecules must solvate both moieties of the polar contact in order for the sorbate molecule to compete with a solvent molecule at one of the solvated moieties. The driving force for solvent-assisted sorption is solvation of the partner of the disrupted polar contact that does not directly interact with the sorbate. There will be a trade-off between solvent-assisted penetration of organic compound molecules into polar contacts, versus competition between sorbate and solvent molecules for new sites at those disrupted contacts. Based on this conceptual model, Borisover and Graber [12,13] introduced a new flexible sorption isotherm model accounting for solvent-assisted sorption, sorbate/solvent competition at disrupted inter- and intramolecular contacts in NOM, and cooperative sorption. The following is a review of the data leading to the development of the conceptual model and a summary of the derived sorption isotherm. Implications of the model for understanding sorption mechanisms of organic compounds in NOM are discussed.

EXPERIMENTAL FINDINGS AND INTERPRETATION |

55

7.2. GENERAL EXPERIMENTAL CONCEPT AND CONSIDERATIONS 7.2.1. Batch Experiment Equilibrium sorption uptake on reference organic matter (Pahokee peat from the International Humic Substances Society, 83% organic matter on a dry weight basis) representing an essentially mineral-free soil organic matter end-member was measured as a function of solute activity from water (hydrated system), active solvents (e.g., acetonitrile and acetone; solvated systems) and hydrocarbons (n-hexadecane or n-hexane; dry systems) [9–11]. Inert hydrocarbon systems were used rather than vapor phase sorption for dry systems because of the long sorption uptake kinetics involved. By comparing isotherms on an activity scale, solute-solvent interactions (such as hydrophobic repulsion from water) are eliminated, and the solvent effect on interactions in the sorbed phase can be seen. 7.2.2. Activity Calculation Compound activities (a) referred to the pure compound liquid state were calculated according , where Ce is the compound equilibrium concentration in solution (mg/L), is the concentration of saturated organic vapor over the pure organic liquid (mg/ L, calculated on the basis of the saturated vapor pressure data), and H is the dimensionless compound Henry’s constant. Activity-based comparison of sorption isotherms of a single compound in different solvents does not depend on the absolute value of the organic compound vapor pressure or the accuracy of such data. Solute association in solution is not expected to affect the activity-based comparison [9,11]. In certain cases, reduced solution concentration (i.e., ratio of equilibrium solution concentration to compound solubility) was used to replace the activity calculation. 7.2.3. Effect of Hydrocarbon Phase As sorption was determined by change in solution phase concentration, solvent sorption may mask the sorption of probe compounds. In such a case, the measured sorbate distribution coefficient Kd (mL/g) differs from the true Kd by the value of solvent sorption (mL/g of sorbent) [10]. The hydrocarbon sorption effect on Kd has been evaluated to be 10% for Kd equal to 0.5 and is negligible for Kd values above unity [10]. Hence, any masking effect of solvent sorption is considered to be negligible in all the experiments described. There is little chance of significant competition between a saturated hydrocarbon and more polar compounds on a dry NOM sorbent. Linear sorption isotherms obtained for non-polar compounds on dry peat up to activities of 0.6 [14] demonstrate that interactions of non-polar compounds with dry NOM are far from saturating sorption sites. Experiments demonstrate that nhexadecane did not affect sorption of pyridine or m-nitrophenol [10,11]. Furthermore, when comparing different binary acetonitrile/acetone or n-hexadecane mixtures (such as in [10]), any n-hexadecane effect on sorption is nullified. As such, sorption from a hydrocarbon as a dry system is considered equivalent to sorption from the vapor phase, with the added advantage that low sorption rates may be followed. 7.3. EXPERIMENTAL FINDINGS AND INTERPRETATION 7.3.1. Activity-Based Sorption Effect The available experimental data for a hydration or solvation effect on activity-based sorption of organic compounds in NOM can be grouped into three major observations: an increase in sorption, no change in sorption or a decrease in sorption upon hydration or solvation of the NOM sorbent as compared with sorption by dry NOM. For phenol, pyridine, m-nitrophenol and benzyl alcohol, activity-based sorption was found to increase upon hydration [9–11]. Activity-based sorption of pyridine was also found to increase upon sorption from the active solvents acetonitrile, acetone, and solvent mixtures of acetonitrile and acetone with n-hexadecane [10]. For nitrobenzene and acetophenone, no change in activity-based sorption on Pahokee peat was found in hydrated versus dry systems [11]. Benzene, trichloroethylene and carbon tetrachloride showed a small decrease in reduced concentration-based sorption by NOM from hydrated systems as compared with dry systems [14]. Atrazine also

56

| CHAPTER 7: EFFECT OF HYDRATION/SOLVATION OF ORGANIC MATTER

Figure 7.1 Activity-normalized isotherms measured on NOM for (A) pyridine, (B) phenol, (C) carbon tetrachloride, and (D) benzene. Data depicted in panels A and B are from [9], and data depicted in panels C and D are from [14].

Figure 7.2 (A) Pyridine sorption at fixed pyridine activity levels (0.001 and 0.003) plotted against initial acetonitrile activity in nhexadecane (% v/v). Dotted bands indicate confidence intervals (95%) on Freundlich model determination of pyridine sorption at fixed activity levels. After [10]. (B) Solvent effect on sorbate uptake for local and general isotherm models defined in Eqs. 7.7 and 7.9, respectively, plotted against equilibrium activity of the active solvent. After [12].

showed a general decrease in reduced concentration-based sorption from a hydrated system as compared with a dry system [9]. Results for pyridine, phenol, carbon tetrachloride and benzene are given in Figure 7.1. It was further observed that the increase in pyridine sorption upon solvation of NOM was dependent upon the activity of the active solvent when pyridine was sorbed from mixtures of active solvent and inert hydrocarbon solvent up to an activity of unity for the active solvent [10]. For example, the amount of pyridine sorbed as a function of acetonitrile activity is shown in Figure 7.2A. In this figure, an increase in pyridine sorption from acetonitrile-n-hexadecane mixtures with increasing acetonitrile activities in n-hexadecane up to 0.7 (0.72% v/v) is clearly seen for two levels of pyridine equilibrium activity. At greater initial acetonitrile activities, pyridine sorption decreased, leading to a maximum in pyridine sorption as a function of acetonitrile activity. The decrease in pyridine activity-based sorption denotes acetonitrile competition with pyridine for new sorption sites at high acetonitrile activi ties. At initial acetonitrile concentrations less than 0.72% v/v, the apparently sigmoidal nature of the pyridine uptake curves also suggests an element of cooperativity in the effect of acetonitrile on pyridine sorption. We were also able to conclude that sorption of pyridine was similarly assisted by additions of acetone to nhexadecane, even though we did not measure activities of pyridine in the studied mixtures (5, 10, and 15% v/v acetone in nhexadecane). Recalling that acetone is a much better solvating medium for pyridine than is n-hexadecane, increasing the

EXPERIMENTAL FINDINGS AND INTERPRETATION |

57

acetone content in n-hexadecane should result in increased pyridine solvation and decreased pyridine sorption by NOM. In fact, acetone additions to n-hexadecane did not affect the observed sorption uptake of pyridine at either of two studied intitial pyridine concentration levels (Table 7.1), such that it can be understood that sorption of pyridine from acetone-n-hexadecane mixtures also displayed a solvent-assisted sorption effect. Table 7.1 Sorption uptake (mg/kg) of pyridine on NOM from acetone-n-hexadecane mixtures at two different initial pyridine concentrationsa Acetone in n-hexadecane (v/v)

Initial pyridine concentration

100 mg/L

600 mg/L

5% 10 % 15 % a Standard deviation for sorption uptake is 4%. From [10].

478 453 432

2057 2158 2058

7.3.2. Mechanistic Scenarios for the Hydration Effect on NOM Sorption of Organic Compounds A number of scenarios for a hydration (solvation) effect on NOM sorption of organic compounds were suggested and illustrated by experimental data [10]. The first scenario is when a sorbate molecule lacks the ability to penetrate into polar contacts in a dehydrated system, but may interact with sites not requiring disruption. Upon hydration, water may create new sorption sites via disruption of polar contacts, but sorbate molecules are unable to compete with water for these new sites. If water competes for sites that were originally accessible to sorbate molecules in the dehydrated state, a hydration-suppressed sorption effect may be observed. The observation of reduced sorption upon hydration of organic matter-rich sorbents reported for the non-polar molecules lindane [15], and benzene, carbon tetrachloride, and trichloroethylene [14] (Figure 7.1C,D) may be interpreted according to this model. In the next major scenario, the sorbate molecule is not effective at penetrating into a polar contact alone. Penetration occurs together with hydration of all the NOM moieties that make up the contact, and the sorbate competes favorably with water for newly created sorption sites. An overall hydration-assisted sorption effect will result, as seen for phenol and pyridine [9]

58

| CHAPTER 7: EFFECT OF HYDRATION/SOLVATION OF ORGANIC MATTER

Figure 7.3 Activity-normalized isotherms measured on NOM for (A) acetophenone, (B) nitrobenzene, (C) benzyl alcohol, and (D) mnitrophenol. After [11].

(Figure 7.1A,B), and for m-nitrophenol and ben zyl alcohol [11] (Figure 7.3C,D). A maximum in sorption effect at some intermediate solvent activity may also be observed, due to competition between the active solvent with the sorbate for newly created sorption sites. This behavior was seen for sorption of pyridine from acetonitrile-n-hexadecane mixtures [10] (Figure 7.2A). Interplay between a hydration-assisted effect and hydration-competitive effect may also be interpreted from the maximum in vapor sorption of trichloroethylene on Aldrich humic acid salt as a function of humidity [16]. The third major scenario may be expected when a sorbate is effective at penetrating into dry NOM contacts, and hydration results in a decrease of sorption due to competition. Acetonitrile sorption from n-hexadecane as compared with sorption from water may be an example of this mechanism. Based on the acetonitrile sorption isotherm from n-hexadecane, it was calculated that 15– 20% of acetonitrile should be sorbed by the peat NOM in an aqueous sorption experiment, and even more if acetonitrile sorption would be hydration-assisted [10]. However, no acetonitrile sorption from water was observed, which can be interpreted as due to competition by water for acetonitrile sorption sites. 7.3.3. Classifying the Sorption Effect as a Function of Sorbate Characteristics Considering the above scenarios and sorbate molecular characteristics such as size, polarizability, hydrogen (H)-bond donating and accepting capability, sorption behavior may be classified and related to the intensity of intra-NOM interactions. This provides a basis for estimating the hydration effect on sorption of organic compounds by NOM. One may expect that the greater a compound’s ability to undergo specific interactions with NOM (i.e., H-bonding, charge complex formation, proton transfer reactions), the greater will be the hydration-assisted sorption effect. This is because penetration of organic compounds into polar contacts solvated by water must involve competition with water molecules. If a sorbate molecule only has a weak ability to undergo specific interactions, it will not be able to compete with water molecules for a newly solvated NOM site (swollen polar contact). Therefore, a clear indication of the role of swollen NOM polar contacts in hydration-assisted sorption would be found if sorption were assisted by hydration for compounds capable of strong specific interactions, and not for compounds of lesser ability. To test this, we compared the hydration effect on NOM sorption of a series of 4 selected organic compounds whose isotherms were measured from water (hydrated system) and a nonpolar saturated hydrocarbon (dry system [11]). Compounds (acetophenone, nitrobenzene, benzyl alcohol, m-nitrophenol) were selected according to their similar electronic polarizability (′ ), similar molar volume (V), and increasing ability to undergo strong interactions with NOM evaluated on the basis of our

EXPERIMENTAL FINDINGS AND INTERPRETATION |

59

earlier classification scheme [17]. Electronic polarizability (′ ) is replaced by molar refraction (MR) according to MR=(4/3) ′ N′ , where N is Avogadro’s number. MR can be estimated easily from the Lorenz-Lorentz equation MR=V(n2−1)/(n2+2), where n is the refractive index of the substance in the liquid state at 293 K for the sodium D-line. This scheme compares sorbate distribution coefficients for transfer from the gas phase to the hydrated NOM phase for the compound of interest and for a nonpolar compound (aromatic hydrocarbon or halogen-substituted hydrocarbon) having the same molar refraction [17]. According to this, acetophenone and nitrobenzene have the weakest ability to undergo specific interactions, benzyl alcohol has an intermediate ability, and m-nitrophenol has the greatest ability [17]. Of the 4 compounds, m-nitrophenol exhibited the greatest hydration-assisted sorption effect (4–8 times, depending on solution phase activity), benzyl alcohol showed an intermediate effect (4 times), and acetophenone and nitrobenzene showed no hydration-assisted sorption effect (Figure 7.3) [11]. The extent of the hydration-assisted sorption effect was correlated with a compound’s ability to undergo strong interactions. Like these results for acetophenone and nitrobenzene, it was also shown by others that activity-based or reduced concentration-based isotherms for compounds with low specific-interacting abilities are usually similar in aqueous and water-free systems (e.g., vapor phase: [14]; n-hexane: [15]). As such, it is worth emphasizing that for many compounds, the changes in NOM sorbent structure that accompany hydration (e.g., increased flexibility, change in ionization status of polar functional groups; conformational reorientation of macromolecules) do not significantly affect organic compound sorption by NOM. In contrast, m-nitrophenol and benzyl alcohol show clear hydrationassisted sorption (Figure 7.3). Such hydration-assisted sorption conforms to the hydration-assisted sorption observed for other strongly specifically interacting compounds phenol and pyridine [9], and to solvent-assisted sorption seen for pyridine in acetonitrile, acetonitrile-n-hexadecane mixtures, and acetone-n-hexadecane mixtures [10]. These results offer additional confirmation of the conceptual model of link solvation. 7.4. LINK SOLVATION MODEL (LSM) ISOTHERM Considering the conceptual model described above, we pursued the development of an isotherm equation that could explore how a sorbate molecule may interact with a solvated contact [12,13]. The development of the link solvation model (LSM) isotherm was intended to reinforce and clarify our mechanistic concept of solvent-assisted sorption, and thus to enhance our understanding of sorption processes in NOM. This model, presented in [12,13], is summarized below. 7.4.1. Model Development Sorbate-sorbent interactions are assumed to occur at sites that are distributed throughout the NOM sorbent. At the outset, only one type of site is considered (local isotherm), which is then extended to account for a diversity of sorption sites (general isotherm). The site is a certain non-covalently linked contact formed by chemical groups or fragments. This link is depicted as {-P…P′-}. The entire {-P…P′-} link may be a sorption site, as in the traditional Langmuir model. In our departure from the traditional models, we conceive that the {-P…P′-} link may be disrupted by a sorbate molecule (denoted by A), or by a sorbed solvent molecule (denoted by S), thus creating new potential sorption sites. The disruption may be slow, resulting in slow sorption kinetics. The link disruption mechanism is formalized in terms of chemical equilibrium equations: (7.1) (7.1′) (7.2) (7.3) (7.3′) (7.4)

60

| CHAPTER 7: EFFECT OF HYDRATION/SOLVATION OF ORGANIC MATTER

(7.5) (7.5′) These equilibrium equations cover all possibilities for A and S molecules to disrupt an asymmetric {-P…P′-} link. For all the above reactions we introduce the equilibrium constants KA, K’A, , KS, K’S, , KAS and K’AS (Eqs. 7.1–7.5′, respectively), where ′ denotes the fraction of differently solvated links, and aA and aS are equilibrium activities of A and S molecules. The pure liquid state is the reference state for compound activity. Considering equilibrium constants for the chemical equilibrium equations and a mass balance for “concentration” of links (), a local sorption isotherm was obtained, which, for the case of sorption from dilute solutions, reduces to a Langmuir-like expression for sorbate uptake SA Eq. 7.6: (7.6) where kAeff is the sum of equilibrium constants of reactions (1) and (1′). The affinity of this Langmuir-like expression includes the B term, which is dependent on solvent activity (as) and is defined, for sorption at asymmetric NOM links, according to Eq. 7.7, (7.7) where the ratio describes link solvation. When , adding a solvent can increase the B term, while when as increases, becomes important, and the B term decreases Figure 7.2B shows how B in Eq. 7.7 changes with solvent activity. It demonstrates that this simple model can predict solvent-assisted sorption of a compound with a maximum in the dependence of sorption on solvent activity (concentration). A maximum for pyridine sorption on NOM from mixtures of n-hexadecane with acetonitrile was observed experimentally (Figure 7.2A). It shows also that the driving force for solvent-assisted sorption is not sorbate-site interaction. Rather it is solvation of the partner of the disrupted contact that does not directly interact with the sorbate (i.e., as expressed by the term of Eq. 7.7). 7.4.2. General Sorption Isotherm In developing the general isotherm model it was assumed that interactions of the solute with a fragment of the disrupted link vary strongly in energy, These different energies contribute to the overall heterogeneity of sorption sites. To account for site energy heterogeneity, we exploited the classical concept of site distribution as an exponential function of energy. Such a function applied to the local Langmuir sorption model results in the Freundlich isotherm [18], the exponent (n) of which is a measure of site energy heterogeneity. If this exponent is constant in different solvent systems, then it is only necessary to consider sorbate-sorbent interactions in order to account for site energy heterogeneity. As seen earlier, in pyridine sorption experiments [10], the Freundlich exponent n was found to be similar in all studied solvent systems (n-hexadecane, water, acetone, acetonitrile, mixtures of acetonitrile in n-hexadecane [10]). The approximate constancy of the Freundlich exponent is also seen graphically by examining the “solvent effect” on pyridine sorption (Figure 7.4). The general sorption isotherm is given by Eq. 7.8,

(7.8)

where and n are parameters of the exponential heterogeneity of sorption sites, and kSeff is the sum of the equilibrium constants of reactions 7.(3) and (7.3′).

EXPERIMENTAL FINDINGS AND INTERPRETATION |

61

Figure 7.4 Solvent effect on pyridine sorption uptake at equilibrium pyridine activity 0.002 plotted against solvent effect on pyridine sorption uptake at equilibrium pyridine activity 0.0002 in different solvent media. Sorption uptakes were calculated by Freundlich equations, and pyridine activities were selected to cover one decade and to be in virtually all cases within the experimental activity range. After [12].

Figure 7.5 Activity-normalized isotherms measured on NOM for (A) phenol, and (B) m-nitrophenol. Data depicted in panel A are from [9], and data depicted in panel B are from [12].

Although derived by assuming an exponential distribution of site-energies, the isotherm in Eq. 7.8 was shown in general to have a non-Freundlich-shape, displaying a concave upward curve on a log-log plot. Eq. 7.8 reduces to a Freundlich isotherm at low solute activities where , and at high solute activities where (see Figure 4 of ref. [12]). Considering low sorbate activities, the solvent effect term defined as the ratio of sorbate uptake from a given solvent (SA) to sorbate uptake from n-hexadecane (SA(Y)) at a given sorbate activity may be derived from Eq. 7.9: (7.9) This solvent effect (Eq. 7.9) is directly related to the B term (Eq. 7.7), both of which are depicted in Figure 7.2B. From Eq. 7.9 it is clear that when is greater than KS, there may be an increase in pyridine sorption upon addition of the solvent S. At sufficiently high solvent activities (as), a maximum in the solvent effect-solvent activity dependence will result (Figure 7.2B). 7.4.3. General Observations and Predictions Derived from the General Isotherm Model The general isotherm model engenders a number of observations and predictions that have been borne out in experiments carried out to this time: 1) the Freundlich exponent is not expected to be affected by the particular solvent. Experimental data for pyridine sorption indeed show that the Freundlich exponent is not strongly different in different solvent systems [12]; 2) the solvent-assisted effect is not a function of solute activity (Eq. 7.9). This is borne out by data depicted in Figures 7.2A and 7.4; 3) the solvent effect will be similar for different organic compounds that successfully compete with a solvent for the same kind of link fragment (Eq. 7.9). This means that strongly interacting compounds such as phenols and substituted phenols will experience similar extents of solvent assisted sorption. This is borne out by data for phenol and m-nitrophenol sorption from water and n-hexadecane (Figure 7.5); and 4) for cases where a solute cannot successfully compete with a solvent for a given

62

| CHAPTER 7: EFFECT OF HYDRATION/SOLVATION OF ORGANIC MATTER

Figure 7.6 Solvent effect for pyridine sorbed by NOM from n-hexadecane-acetonitrile mixtures plotted against acetonitrile equilibrium activity. Data are fitted with Eq. 7.12 using a Freundlich exponent of 0.54 (averaged for data in all acetonitrile-hexadecane systems in Table 1 of ref. 12). Eq. 7.12 accounts for sorption cooperativity. R2 for model fit is 0.965 with a standard deviation of 1.1.

link fragment, no solvent-assisted sorption is expected. Indeed, sorption of acetophenone and nitrobenzene on fully hydrated NOM was not assisted by hydration (Figure 7.3A,B). 7.4.4. Extension to Account for Sorption Cooperativity Finally, the general isotherm model was extended to consider the cooperative nature of the hydration/solvation effect on sorption as depicted in Figure 7.2A, which shows uptake of pyridine as a function of acetonitrile activity in the solvent. Sorption cooperativity may be related to the aggregation of NOM by non-covalent links between different organic radicals and functional groups. Due to such aggregation, it may be necessary to disrupt a number of links simultaneously in order to achieve solvation [9]. When an NOM contact is disrupted, we consider there are m sites for sorption (as distinct from two sites assigned for disrupted contact {-P…P′-} in the original scheme, Eqs. 7.1–5′). Cooperativity may be accounted for by replacing Eqs. 7.4 and 7.5 with Eqs. 7.10 and 7.11. (7.10)

(7.11) The “solvent effect” for the cooperative extension defined similarly to Eq. 7.9 is given by Eq 7.12. (7.12) The solvent effect was evaluated for pyridine sorption from acetonitrile-n-hexadecane systems. A good fit of the data to the model was obtained (Figure 7.6), where the m parameter (number of sorption sites) was computed to be 5.8 with a standard error of ±1.4. This suggests that at least 4 molecules have to penetrate into a certain NOM region to disrupt contacts and to cooperatively swell a NOM moiety. The successful fit depicted in Figure 7.6 is gratifying for such a complicated heterogeneous sorbent as NOM, particularly considering that the model does not account for non-specific partitioning of solute molecules into the sorbent or sorption at sites that are available without recourse to disruption by a solvent. The model is expected to be very sensitive to sorption curves that demonstrate sigmoidality or curves exhibiting a maximum in sorption uptake versus solvent activity, thus aiding in model parameter evaluation. More sets of accurate experimental data are needed to evaluate the model and its extension in detail. Nevertheless, we consider that the successful data fit depicted in Figure 7.6 supports the link disruption concept in the development of a flexible sorption model.

EXPERIMENTAL FINDINGS AND INTERPRETATION |

63

7.5. CONCLUSIONS Our results indicate the importance of polar NOM non-covalent links in the organization of the NOM phase and in controlling the hydration effect on sorption of organic compounds. Altogether, our studies demonstrating hydration and solvation-assisted sorption for several different probe molecules [9–11], development of a conceptual model to explain this effect, and derivation of a sorption isotherm based on this conceptual model [12,13] help to relate organic sorbate structure, sorbate ability to interact with functional groups in NOM, and the resultant hydration effect on sorption. As NOM is a very important environmental sorbent for organic compounds in hydrated systems, these results and the conceptual model for the NOM hydration effect on organic compound sorption are relevant in understanding the effect of NOM aggregation on sorption. It is seen that for many compounds with only weak ability to interact specifically with NOM, the changes in NOM sorbent structure that accompany hydration (e.g., increased flexibility, change in ionization status of polar functional groups, conformational reorientation of macromolecules) do not significantly affect their sorption. This observation should have important ramifications for understanding the role of sorbent rigidity (as suggested for example as “glassiness”) for sorption of many compounds. If such major changes in NOM sorbent structure as occur upon hydration and wetting do not significantly affect activity-based sorption of many non-polar or weakly polar compounds, then it is not clear how a “glassy” versus “rubbery” structure may have such an effect. The strength of the derived sorption isotherm is its flexibility and potential to account for asymmetry of disruptable NOM links and diverse forces of sorbate-solvent interactions at fragments of disrupted contacts, thus providing a basis for the successful interpretation of the solvent effect on sorption of organic compounds by NOM. ACKNOWLEDGEMENTS This research was supported by a grant from the Israel Science Foundation (Grant no. 400/00–1) and a grant from the Ministry of Environmental Quality (No. 901). REFERENCES 1. 2. 3. 4. 5. 6. 7.

8.

9. 10. 11. 12. 13. 14. 15.

Yaron B, Saltzman S. Influence of water and temperature on adsorption of parathion by soils. Soil Sci. Soc. Am. Proc., 1972; 36: 583–586. Chiou CT, Shoup TD, Porter PE. Mechanistic roles of soil humus and minerals in the sorption of nonionic organic-compounds from aqueous and organic solutions. Org. Geochem., 1985; 8:9–14. Unger DR, Lam TT, Schaeffer CE, Kosson DS. Predicting the effect of moisture on vapor-phase sorption of volatile organic compounds to soils. Environ. Sci. Technol., 1996; 30:1081–1091. Schwarzenbach RP, Gschwend PM, Imboden DM. Environmental organic chemistry; New York: Wiley, 1993. Hayes MHB. Extraction of humic substances from soil. In Aiken GR, McKnight DM, Wershaw RL, MacCarthy P eds. Humic substances in soil, sediment and water. New York: Wiley, 1985:329–362. Swift RS. Molecular weight, size, shape, and charge characteristics of humic substances: Some basic considerations. In: Hayes MHB, MacCarthy P, Malcolm RL, Swift RS eds. Humic substances II: In search of structure, Chichester: Wiley, 1989:449–465. Clapp CE, Hayes MHB, Swift RS. Isolation, fractionation, functionalities, and concepts of structures of soil organic macromolecules. In: Beck AJ, Jones KC, Hayes MHB, Mingelgrin U eds. Organic substances in soil and water: Natural constituents and their influences on contaminant behavior. London: Royal Society of Chemistry, 1993:31–69. Mingelgrin U, Gerstl Z. A unified approach to the interaction of small molecules with macrospecies. In: Beck AJ, Jones KC, Hayes MHB, Mingelgrin U eds. Organic substances in soil and water: Natural constituents and their influences on contaminant behavior. London: Royal Society of Chemistry, 1993:102–127. Graber ER, Borisover MD. Hydration-facilitated sorption of specifically-interacting compounds by model soil organic matter. Environ. Sci. Technol. 1998; 32:258–263. Borisover M, Reddy M, Graber ER. Solvation effect on organic compound interactions in soil organic matter. Environ. Sci. Technol., 2001; 35:2518–2524. Borisover M, Graber ER. Relationship between strength of organic sorbate interactions in NOM and hydration effect on sorption. Environ. Sci. Technol., 2002; in press. Borisover M, Graber ER. Simplified link solvation model (LSM) for sorption in natural organic matter. Langmuir, 2002; 18: 4775–4782. Borisover M, Graber ER. Simplified link solvation model (LSM) for sorption in natural organic matter. Extended abstracts, Meeting of International Humic Substances Society (IHSS11), Boston, MA, 2002. Rutherford DW, Chiou CT. Effect of water saturation in soil organic-matter on the partition of organic compounds. Environ. Sci. Technol., 1992; 26:965–970. Mills AC, Biggar JW. Solubility-temperature effect on the adsorption of gamma- and beta-bhc from aqueous and hexane solutions by soil materials. Soil Sci. Soc. Am. Proc. 1969; 33:210–216.

64

16. 17. 18.

| CHAPTER 7: EFFECT OF HYDRATION/SOLVATION OF ORGANIC MATTER

Ong SK, Lion LW. Trichloroethylene vapor sorption onto soil. Soil Sci. Soc. Am. J. 1991; 55:1559–1568. Borisover MD, Graber ER. Specific interactions of organic compounds with soil organic matter. Chemosphere, 1997; 34:1761–1776. Sips R. On the structure of a catalyst surface. J. Chem. Phys., 1948; 16: 490–495.

Chapter 8 SWELLING OF ORGANIC MATTER IN SOIL AND PEAT SAMPLES: INSIGHTS FROM PROTON RELAXATION, WATER ABSORPTION AND PAH EXTRACTION

Gabriele E.Schaumann,1 Julia Hurrass,1 Martin Müller2 and Wolfgang Rotard1 of Environmental Protection, Dept. Environmental Chemistry, Technical University Berlin, Sekr. OE 1, Franklinstr, 28/29, D-10587 Berlin, Germany 2Institute of Applied Geosciences, Dept. of Applied Geophysics, Technical University Berlin, ACK 2, Ackerstr. 76, D-13355 Berlin, Germany

1Institute

8.1. INTRODUCTION 8.1.1. Relevance of the Interactions of SOM with Water Under field conditions, the upper soil layers are exposed to variations in moisture and temperature. In the course of moisture fluctuation, the soil organic matter (SOM) changes its water content and its state of swelling. Recent results show that with the state of swelling, SOM gradually changes its physicochemical properties, such as sorbent properties [1,2], macromolecular structure [3,4] thermal characteristics [5] or the binding of hydrophobic organic compounds. These changes are well-known to affect sorption and transport phenomena and thus have to be taken into consideration. Despite this relevance, very little is known about the process of swelling of SOM, and even less is known about its effect on SOM physicochemistry [6], This may be due to methodical challenges. In order to observe and describe swelling, SOM has to be maintained in its original state, which means that SOM has to be investigated within the whole soil sample. An extraction of humic fractions would change the composition and the macromolecular structure of SOM, both of which are important factors for the process of swelling. Methods applied to investigate swelling of homogeneous materials cannot easily be transferred to highly heterogeneous soil samples. Thus, the following questions arise: Which methods can be used to describe, quantify and understand swelling of SOM in soil samples containing organic matter? How fast is swelling and which swelling kinetics can be observed? Which physicochemical properties of soil and SOM are affected by swelling of SOM? To what extent does swelling affect sorption and transport phenomena? With special emphasis on 1H-NMR relaxation, the scope of this study was to investigate the swelling of organic matter in soils and peat and its effect on the extraction of PAH from soil. 8.1.2. Relaxation

1H-NMR

In the method of 1H-NMR relaxation, the protons of water molecules are excited and the relaxation kinetics observed. The relaxation time is the time constant of this first order relaxation process. In porous systems, the relaxation time is determined by the pore size, the surface relaxivity and the relaxation time in the bulk phase [7,8]. According to [8,9], the relationship is expressed by Eq. 8.1. (8.1) Here, T is the relaxation time of water in the porous system, S is the (internal) surface area of the porous system, ′ is the surface relaxivity and TB is the bulk relaxation time, which is the relaxation time at infinite distance from the walls. The correlation of T with the particle size distribution of natural and artificial samples has recently been investigated [10–12]. However, all the experiments were performed on mineral systems. As we expect a difference of the surface relaxivity in samples containing organic matter, it is not possible to make conclusions about the absolute pore size from the relaxation time. However, the variation of relaxation times observed during hydration allows conclusions on the development of the pore sizes and thus allows determination of relative pore sizes [13,14]. The relaxation time is also determined by the degree of mobility of the water molecules and by the degree of water binding. Relaxation times <5 to 30 ms usually are attributed to bound water [9]. Due to swelling, we would expect an increase in the degree of freedom of water molecules in the gel phase, parallel to a decrease of the surface relaxivity due to a growing

66

| CHAPTER 8: SWELLING OF ORGANIC MATTER IN SOIL AND PEAT SAMPLES

softness of the gel phase. Both factors would increase the relaxation time. If instead we observe the water in the pore system around the swelling particles, we would expect a decrease of the mean pore size and thus a decrease of the relaxation time. The application of 1H-NMR relaxation is new for investigating water in soil systems. The infiltration of water into a soil sample was observed by Amin [15]. As et al. [16] used NMR imaging to study slow and fast transport processes in biosystems. Assuming two groups of water molecules with two different relaxation times, Sauerbrey investigated the state of decomposition of peat samples using 1H-NMR relaxation [17]. Stallmach et al. used 1H-NMR relaxation to determine the fractal geometry of defined porous systems [18]. Using 1H-NMR relaxation and NMR imaging, Belliveau et al. were able to observe the absorption of pesticides from water into dry soil [19]. They also gave first hints on a possible swelling of organic matter and on changes of the relaxation times of water during the first days of swelling. The use of 1H-NMR relaxation to explicitly investigate swelling of soil organic matter has been recorded only recently [14,20], and shows the high potential of 1H-NMR relaxation for the investigation of swelling of organic matter in soil and peat samples. 8.2. MATERIALS AND METHODS 8.2.1. Materials Soil Samples. The soil samples investigated were taken from the Ah of sandy soils close to Berlin, Germany. The samples were analyzed either in the fresh state or after air drying and conditioning over a saturated NaCl solution to maintain a constant relative humidity of 76%. Soil A: The soil material used was taken from a sandy forest soil near Chorin, Germany (61% sand, 30% silt and 8% clay fraction). The organic matter content was 8.3%. The PAH content was below detectability. Results are presented for two representative samples of soil A, which are labeled A1 and A2. Soils N and G were samples of sandy soils at the edge of two German highways. Both samples were similar in mineral composition and soil chemistry. The mineral part consisted of>90% sand, with an organic matter content of 5.3% (soil N) and 4.4% (soil G). The PAH content (sum of 11 PAHs) was 13.1 mg/kg (soil N) and 6.4 mg/kg (soil G). The contamination in soil N was estimated to be 1–5 years old and at least 10–15 years old in soil G. Peat Samples. The peat samples investigated were taken from two wetlands of different origin. Peat S originated from a floodplain moor in the Spreewald near Berlin. The organic matter was in an advanced state of decomposition due to strong variations of the groundwater table accompanied by temporal floods and dry periods. More details on the wetlands and peat properties have been described [21]. The sample was taken from the top 15 cm of the organic horizon, which had been influenced by plowing. After sampling, the peat was air dried and sieved to particle sizes < 2mm. The organic content of the sample was 51% and the pH (0.01 M CaCl2) was 5.6. Results from two representative samples of S, labeled S1 and S2, are presented. Peat R: From the fen Rhinluch, 60 km northwest of Berlin, samples were taken from different depths. The fen and the peat properties have been described by Schwärzel [22]. The material was air dried directly after sampling. The dried aggregates were crushed with a mortar and then sieved to particle sizes <2mm. The organic content varied between 45 and 55%. In this paper, two representative samples, labeled R1 and R2, taken from the shrinkage horizon of the fen (see [22]) are described as examples. Swelling References. Swelling references were used to distinguish between effects resulting from the dynamics of the organic matrix and from minerals in the soil samples. To distinguish between the effects of the inorganic and organic matrices of the samples we prepared ashed soil samples by heating soil A to 500°C for 3 hours. Since semolina contains starch as swelling polymers and is less homogeneous than synthetic polymers, semolina and pure starch were used as swelling references. The chemical compositions of starch and semolina are similar, with semolina containing starch and additional compounds. One important difference between starch and semolina results from their different particle sizes. The investigated starch sample mainly consisted of small particles within the size of coarse clay/fine silt (<0.02 mm). Semolina consisted of particles about the size of fine sand.

MATERIALS AND METHODS |

67

8.2.2. Methods 1H−NMR

Relaxation. The air-dried soil samples and a mineral reference sample were placed into plastic containers and were then moistened with water. We chose a water content of 35% for the soil samples and of 52±2% of the wet mass for the peat samples. Swelling was allowed for three to four weeks at 20°C. Within this period of time, the 1H-NMR relaxation experiments were performed two to three times per week for every sample. The experiment was repeated twice under the same conditions. The measurements described in this paper were carried out with a 2 MHz relaxometer (Maran 2, Resonance Instruments, UK) using the inversion recovery pulse sequence to determine the relaxation time T1. The resulting relaxation decays of each sample at each selected point of time during swelling were fitted to a sum of exponential decay functions (multiexponential fit), using time constants between 0.001 and 10000 ms. For the fitting procedure, the inversion algorithm provided with the WINDXP software (Resonance Instruments) was used. With this procedure we obtained a distribution of relaxation times, which also is called a relaxation time spectrum in this paper. These spectra are characterised by the number and position of the peaks. Taking the sensitivity of the NMR instrument into account, peaks were considered significant if they accounted for more than 10% of the water. The peak data were plotted as a function of swelling time, and kinetic parameters were derived from these plots. Gravimetric Investigation of Water Uptake. The gravimetric method provides information about the amount of water “absorbed” by the soil sample up to a certain point of time at a certain matrix potential. Different grades of water binding can be distinguished by varying the matrix potential. The soil sample was placed in a vessel and put onto a ceramic plate, through which the water was brought to a defined matrix potential by a hanging water column. The experiments were carried out at 23° C (two replicates). The water uptake was measured by weighing the same samples at several points of time within two weeks of swelling. State of Swelling and Extractability of PAH. The objective of this method was to measure the effect of swelling on the amount of PAH extracted from soil samples by a well-known extraction method. Three replicates of the soil samples were moistened with water to defined final water contents of up to 30% of the wet mass. After moistening, swelling was allowed for up to three weeks. Three times within this period, a part of the sample was taken and extracted according to a modified VDLUFA method (shaking for 16 h with a mixture of acetone, dichloromethane and saturated NaCl solution and clean-up with a sodium sulfate/alumina/silica gel column using hexane as the mobile phase). PAH detection was performed by HPLC with fluorescence detection. 8.3. RESULTS 8.3.1. Swelling and 1H-NMR Relaxation Reference I: Ashed Soil Samples. Figure 8.1 shows the T1 spectra of the ashed soil sample A directly after moistening with water and after 21 days of hydration (resulting water content 30% of the wet mass). Directly after the addition of water, the spectrum shows a peak at 36 ms (Figure 8.1). The spectra do not differ significantly when comparing the situation directly after and 21 days after the addition of water. The slight shift of the peak from 36 ms to 32 ms and the slight decrease of the half width from 128 ms to 118 ms are not significant, as can be seen in Figure 8.2, which shows the development of the peak position of the ashed soil A within a period of 21 days after moistening. In Figure 8.2 it becomes obvious that the peak data changes are statistical and that the observed changes between days zero and 21 are smaller than the statistical changes during this period of time. From these observations, we conclude that 1) the ashed soil sample investigated did not show any observable development within three weeks of hydration; and 2) we may expect an uncertainty of the peak position of approx. 10 to 13%, estimated from the statistical changes of the peak positions during the hydration time. Thus, differences in peak positions smaller than 13% are not considered significant. Reference II: Starch and Semolina. Figure 8.3 shows the relaxation time spectra of a mixture of starch (20%) and water (80%) before and after swelling. Both spectra consist of one significant peak. Starting at 238 ms, the peak moves to 265 ms due to swelling. Parallel to movement of the peak to increasing relaxation times, the half width of the peak increases slightly from 812 to 841 ms. Figure 8.4 also shows the relaxation time spectrum of a mixture of semolina (20%) and water (80%) with one significant peak. Due to swelling, the peak moves towards decreasing relaxation times from 922 to 367 ms. The half widths change from 4790 to 1170 ms. Thus, as indicated by the peak position, the water is located in pores without significant binding forces.

68

| CHAPTER 8: SWELLING OF ORGANIC MATTER IN SOIL AND PEAT SAMPLES

Figure 8.1 Relaxation time spectra T1 of the ashed soil sample A directly after addition of water and after 21 days of hydration

Figure 8.2 Variation of the peak position and of the half width of the peaks in the ashed soil A as a function of hydration time

Figure 8.3 Relaxation time spectra T1 of a mixture of starch and water (80 %) before and after swelling

Figure 8.4 Relaxation time spectra T1 of a mixture of semolina and water (80 %) before and after swelling

Comparing the changes of the relaxation time spectra of semolina/water and starch/water results in the following observations: 1) before swelling, the spectra of starch and semolina differ strongly (Figure 8.5). However, after swelling they have approached each other, with still slightly longer relaxation times for semolina than for starch (Figure 8.6); 2) semolina has longer relaxation times than starch; 3) the half widths of the peaks decrease for semolina and increase for starch; 4) the changes due to swelling are dramatic for semolina but slight for starch. The different behaviour of these samples is surprising based only on the similar chemical composition of starch and semolina. A difference of chemical composition would account for the behaviour of the water bound to the functional groups and thus for the relaxation times of bound water. Also, due to their similar chemical composition, we would not expect significant differences in surface relaxivity between pure starch and semolina. The main difference between the two samples is attributed to the particle size of the dry samples. The investigated starch sample mainly consisted of small particles about the size of coarse clay/fine silt, while semolina consisted of particles about the size of fine sand. This is considered to cause

MATERIALS AND METHODS |

69

Figure 8.5 Comparison of the relaxation time spectra T1 of the mixture of semolina and water and of starch and water before swelling

Figure 8.6 Comparison of the relaxation time spectra T1 of starch and semolina pudding after swelling

the main differences in the swelling effects, as this results in different surface areas and especially in different pore size distributions. The main difference between starch and semolina is observed before swelling: due to the larger particle sizes at this point of time, the average pore size in the semolina was higher than that of starch. After swelling, the pore size distribution is similar in starch and semolina pudding, indicating a similar structure of the swollen matrix, regardless of the differences of the initial particle size distribution. Soil Samples. Figure 8.7 exemplarily shows T1 spectra of soil A (representative sample A1). The first spectrum was recorded 15 minutes after moistening to 35% water content of the wet mass. In sample A1, two peaks (peak Ia: 52 ms and peak I: 460 ms) are observed. Peaks I and Ia are not fully separable from each other. In sample A2, only one peak at 105 ms is observed (not shown). This peak is located between Peak Ia and Peak I of sample A1, approximately at their focal point. In most mineral organic soils, the spectra taken 15 minutes after moistening consist of one or two peaks with varying separability. They are found at relaxation times between 10 and 1000 ms, indicating that the water is free and positioned in the pore system. The spectra were reproducible for measurements of the same sample. Each sample kept its spectrum identity as long it was not shaken or taken out of the container. However, the position and the number of the peaks varied with the soils and also with the samples taken from one soil. The number, distribution and separability of peaks were determined by the individual sample. During the hydration time, the T1 spectrum of the samples changes gradually for at least two to three weeks. This is shown exemplarily in Figure 8.7 for sample A1. In spectra consisting of two peaks, peak Ia and I approach each other at different rates during swelling. Peak Ia is separable from peak I only in the first few days and then becomes fully integrated into peak I, which moves towards 34 ms. In the spectrum consisting of one peak only, this peak moves towards shorter relaxation times and thus is comparable to peak I. The gradual change of the spectra was not fully completed after 3 weeks of swelling, indicating that the process of swelling is quite slow and may last for more than 3 weeks. These observations are typical of most of the mineral organic soil samples investigated. However, there were a small number of samples for which we did not observe any significant changes. The movement of the peaks as a function of hydration time for samples A1 and A2 is shown in Figure 8.8. The movement of peak I can be fitted by the sum of two exponential decay functions, with time constants t1=(0.44±0.06) days and t2=(4.6 ±0. 7) days for sample A1 and t1=(0.5±0.4) days and t2=(9±7) days for sample A2. The error given here corresponds to the standard error of fitting. The movement of peak Ia of sample A1 could not be described mathematically because this peak was only identified at day 0 and day 1. Afterwards, it is fully integrated into peak I. We assume either a movement towards longer relaxation time or a constant position during the process of hydration. Peat Samples. Figure 8.9 shows T1 spectra of the two peat samples S and R, which contain 40 to 50% organic matter. The spectra were recorded 15 minutes after moistening to 52% water of the wet mass. The spectra consist of two peaks with varying separability. They are reproducible within several measurements, the position and the number of the peaks being

70

| CHAPTER 8: SWELLING OF ORGANIC MATTER IN SOIL AND PEAT SAMPLES

Figure 8.7 Relaxation time spectra T1 of the soil sample A1 at different points of time in the course of swelling

Figure 8.8 Movement of peak I of soil sample A1 and of the peak of the mineral reference sample as a function of hydration time

Figure 8.9 Relaxation time spectra T1 of peat samples S1, S2, R1 and R2 15 minutes after moistening

more reproducible for the peat samples than for the mineral soil samples. The distribution and separability of the peaks was mainly determined by the peat soil. Occurring at approximately 1 ms for all peat samples, peak II may indicate bound water or water in very fine pores. Depending on the peat sample, the ratios of the heights of Peak II to Peak I covers a wide range. For example, the height of peak II is lower than the height of peak I in peat S, but it is higher in peat R. This hints at different proportions of water being located in very fine pores or being bound to the solid matrix. Peak I is located at (720±20) ms in peat S and at (200±70) ms in peat R. In the peat samples not shown here, the initial position of peak I varies between 50 ms and 800 ms. During swelling, the relaxation time spectra change gradually, as shown in Figure 8.10 for the swelling of peat S. During swelling, peak I moves towards shorter relaxation times. It approaches peak II, which stays constant at 1 ms, the peaks fusing within 3 weeks of swelling in half of the samples more or less completely. Approaching peak II, peak I increase its width and decrease its intensity. At the same time, peak II increases in intensity. These observations are similar for all peat samples investigated. The gradual change in the spectra was not fully completed after 3 weeks of swelling, indicating that the process of swelling in peat is quite slow and may last for more than 3 weeks. In order to describe the kinetics of the peak movements quantitatively, the position of peak I was plotted against the hydration time as shown in Figure 8.11 for peat R and in Figure 8.12 for peat S. The time constants are (1.7±0.4) days for peat S and (4±2) days for peat R. The movement of this peak can be described by one first-order process for most peat samples investigated in this study. The time constants range from 0.4 days to 6 days with standard errors of about 50% and most time constants being longer than 1 day. The quality of the determination of the time constants depended strongly on the

MATERIALS AND METHODS |

71

Figure 8.10 Relaxation time spectra T1 of peat S at different times in the course of swelling

Figure 8.11 Movement of peaks I and II in two representative samples of peat R during swelling. The movement of peak I has been fitted by one first-order process

Figure 8.12 Movement of the peaks I and II in two representative samples of peat S during swelling. The movement of peak I has been fitted by one first-order process

separability of peaks I and II. Best results were found when the distance between the peaks was greater than 2.5 decades. With smaller distances, the scattering of the peak data was more significant. 8.3.2. Gravimetric Investigation of Water Uptake Figure 8.13 shows the water uptake of the air-dried and the ashed soil A at a matrix potential of 63 cm (hanging water column). In the mineral sample, the water uptake is rapid within the first few hours and reaches a constant value of 33% (w/w). The water uptake can be fitted by an exponential decay function with a time constant of 0.2 days. However, in the air-dried sample A, the water uptake is much slower. The final water content (12%) is reached after 14 days. Only data fitting a twoexponential decay function provides good results. The time constants of the fast process range from 0.1 to 0.5 days for all curves measured, whereas the time constants of the slow part varied between 2.5 and 6 days. Thus, the time constants of the fast process in the soil samples are similar to the time constants observed with the mineral sample. We assume that the fast process describes the water uptake of the mineral part of the sample, and the slow process describes water uptake of the organic part of the sample. The final water content of the air-dried sample (12%) was expected to be at least as high as the final water content of the ashed sample (33%). The low value might be due to only partial wetting

72

| CHAPTER 8: SWELLING OF ORGANIC MATTER IN SOIL AND PEAT SAMPLES

Figure 8.13 Water absorption of the air-dried and the ashed soil A from a ceramic plate at a matrix potential of 63 cm. The data have been fitted to two first order processes for soil A and one first-order process for the ashed sample

Figure 8.14 Extracted amounts of benzo(a)pyrene from two sandy soils in the course of swelling of SOM after moistening to 25% water content

of the Ah sample caused by hydrophobic areas or small, air-filled pores with hydrophobic surfaces. This phenomenon will be investigated in further experiments. 8.3.3. State of Swelling and Extractability of PAH Figure 8.14 shows the amount of benzo(a)pyrene (BAP) extracted from soil N and soil G as a function of swelling time, after bringing the field moist samples to 25% water content. The results with BAP are representative for most other PAH compounds investigated. For Soil N, an increase of BAP extractability is observed, whereas soil G shows a decrease of extractability with increasing swelling time. Also, the initial state before moistening and the final water content affected the extractability of BAP. The extractability after 9 days of hydration was lower when moistening the air dried sample than when moistening the field moist samples [23]. When taking the measurement error into account, the observed effects are at the limit of significance. However, they were observed on several replicates and showed a similar behaviour for different water contents. Also, the differences between soil N and soil G were reproduced under different conditions. We can speculate on the qualitative existence of effects of the initial state before moistening, the final water content and the hydration time. Also, we can consider a significant effect of the soil sample. However, more detailed studies are required for quantitative significance. 8.4. DISCUSSION 8.4.1. Effects of Swelling of Starch and Semolina As shown by the results (Figure 8.6), the differences between the spectra of starch and semolina pudding are at the limit of significance after swelling. Thus, we might conclude that the solid matrix structure of swollen starch and semolina pudding has become similar due to swelling. This indicates that the water in the semolina is no longer located mainly between the semolina particles, but within the particles, being absorbed and integrated into the gel structure, which is often described as a highly interpenetrated pore system [24,25]. We also assume that in the swollen state of starch and semolina, the water between and within the particles cannot be distinguished, because only one peak was observed after swelling. The slight

MATERIALS AND METHODS |

73

differences observed between semolina pudding and swollen starch may be due to either chemical or macromolecular differences of the starches or due to still remaining differences of pore sizes. The results also show that swelling is observable using 1H-NMR relaxation, due to its capability of distinguishing between pore sizes and changes in pore size distribution. If the initial pore size distribution is different from the final distribution, swelling can be observed by observing this change. If the initial particle size produces a pore size distribution similar to the final distribution, the effect of swelling is difficult to observe. This was probably the case for starch. Thus, it should be possible to increase the detectability of swelling processes by varying the initial pore size distribution. We also expect a strong dependence of both the initial relaxation time spectrum and its development during swelling on the initial sample packing. This may especially affect heterogeneous samples. 8.4.2. What Does 1H-NMR-Relaxation Tell Us About the Swelling of Organic Matter in Soil and Peat? As no significant change in the relaxation time spectra was observed with the ashed soil, it can be concluded that the effects observed in the soil and peat samples are not caused by their mineral components. This coincides with the lack of swelling of the minerals in the samples investigated in this study. Thus, the observed effects can be assigned to the organic matter in the samples. This conclusion is strengthened by the changes in the relaxation time spectra due to swelling in mixtures of semolina or starch with water, because in these samples the mineral part is negligible. This raises the question: what mechanisms are responsible for the shifts of relaxation times in peat and soil samples? In Table 8.1, peak data and their variations due to swelling are compared for the samples investigated. While for starch, semolina and the soil samples, the significant peaks were observed at relaxation times >10 ms, peat showed two peaks, peak II being located at approximately 1 ms and peak I being located at relaxation times >50 ms. Peak II accounts for more than 20% of the total intensity for all peat samples and increases during swelling. The value of the relaxation time of this peak indicates bound water or water located in very fine pores. Relatively speaking, this type of water increases significantly during the swelling of peat R. However, the actual results do not allow us to draw firm conclusions on the behaviour of peak II and its interpretation. More studies are needed to draw conclusions about this part of the mechanism of swelling. The peak of semolina has similar behaviour to peak I of soil and peat, which show a movement towards decreasing relaxation times due to swelling. This gives a hint that the mechanism of swelling observed by 1H-NMR relaxation is similar in semolina, soil and peat. However, the final position of the peak in semolina is at significantly longer relaxation times than that of peak I of peat and soil samples. For peat and soil, the relaxation times of peak I after swelling may hint at a certain degree of water binding. This means that the mobility of water, expressed by the effective pore size, is much higher in semolina pudding than in swollen peat or soil. This is not surprising, as semolina is an easily swollen compound, and it can be assumed that in the organic matter of peat and soil the macromolecular network is much more rigid, heterogeneous and probably more cross-linked, reducing pore size and flexibility of the organic matrix. Table 8.1 Position and share of the total intensity of the peaks in the relaxation time spectra T1 for the investigated soil, peat and reference samples Peak II position [ms] proportion Peak I position [ms]

starch

semolina

soil A

peat R

peat S

n.s.a

n.s.a

n.s.a

1.0′ 2.5

1′ 2d

n.s.a 240′ 265

n.s.a 920′ 370

n.s.a 53%′ 75% 23%′ 25% 100…500b 100…500c 800′ 100 ′ 20…35b ′ 30…50c a n.s.: the peak is not significant; b 100′ 20 ms for peak II in sample A2; 500′ 35 ms for peak IIa in sample A1; c this is the range for all peat samples from Rhinluch; d the movement was not significant.

From these arguments the following conclusions are drawn for the process of swelling. Two groups of water can be distinguished by NMR relaxation, which are bound water or water located in very fine pores (peak II) and pore water (peak I) in the unswollen and swollen sample. The pores are located rather between the particles before swelling and also within the particles in the course of swelling. In this model, the pores strongly decrease in size during swelling. This may be explained by the expansion of SOM due to water absorption. Based on these experiments, conclusions about the condition of binding of water or its degree or freedom within the swollen matrix are still difficult, whereas the changes due to swelling are characterized satisfactorily by the nature and rate of swelling.

74

| CHAPTER 8: SWELLING OF ORGANIC MATTER IN SOIL AND PEAT SAMPLES

8.4.3. Swelling Kinetics by 1H-NMR Relaxation The observations of the changes of the relaxation time spectra show that the process of swelling lasted at least two to three weeks, and probably even longer. This shows that swelling of organic matter must not be neglected for processes taking place under fluctuating water contents. However, in trying to quantify the swelling kinetics, we found differences between representative samples. As we assume that the initial pore size distribution determines the initial relaxation time spectrum, we explain the differences between representative samples with differences in the state of packing of each sample. These differences still will have an influence on the relaxation time spectra after swelling, but the more water enters the organic matter causing swelling, the less water is determined by the size distribution of the original pores. Thus, different initial pore size distributions in the same sample may cause different time constants. Therefore, more and systematic studies involving a large number of samples and sample repetitions will have to be made in order to quantify the kinetics of swelling. Nevertheless, we can conclude from the observations made with the investigated peat samples that swelling is a slow process lasting for at least 2 to 3 weeks, and with time constants varying between 1 and 6 days. 8.4.4. Further Evidence from the Ceramics Method and the Extractability of PAH Both the NMR and the ceramics method hint at a first order process during the hydration of soil organic matter and peat. The time constants of swelling as determined by 1H-NMR relaxation are similar to those determined by the ceramic method. Thus, we observe the same process from different points of view by both methods. While it is the change of the state of mobility of the water during swelling that is observed by 1H-NMR relaxation, the ceramics method indicates the development of the amount of water absorbed during swelling. Thus, with the 1H-NMR relaxation method it is possible to suggest the pore size distribution during swelling. The ceramics method gives information on the water absorption during swelling, providing boundary conditions for the water binding by regulating the matrix potential. Both methods supply important information on the swelling process. Further evidence of the process of swelling is provided by the observation that the extractability of PAH is determined by the initial state of the sample before moistening, the hydration time and the water content. Swelling of SOM is one probable factor to explain this observation, whereas microorganism activity is improbable as an influence factor (an increase in extractability cannot easily be explained by microorganism activity). The different behaviour of soil N and G may be explained by the different age of their contaminations, implying different forms of binding and different proportions of bound residues. This is also supported by several studies investigating the sorption of organic chemicals by soil during drying and rewetting cycles (see for example [26–28]). 8.5. CONCLUSIONS Swelling of soil organic matter was observed by NMR relaxation and by gravimetric methods. Both methods provided comparable information on the swelling kinetics and indicated a slow swelling process with time constants of 1 to 6 days. Based on measurements with the gravimetric method, swelling of SOM is considered to be significantly slower than water uptake by mineral components (time constant 0.5 days). From the NMR experiment we learned that the pore size distribution changes in the course of swelling. Further, the extractability and thus the binding of hydrophobic organic compounds to SOM seems to be determined by SOM swelling. Swelling may have various effects on the binding of hydrophobic organic compounds to SOM [1,26–28]. Studies not presented here also implied an influence of swelling on the thermal characteristics of SOM as measured by DSC [5]. Hence, swelling of SOM is an important factor for understanding most ecologically relevant processes in upper soil layers, especially when processes in field studies are the focus of interest. Swelling has to be considered when investigating sorption and transport phenomena as well as the structure and relevance of the solid SOM. Due to this relevance, swelling of SOM opens a wide, enigmatic and challenging field of research with high demand. ACKNOWLEDGEMENTS The authors thank Prof. Dr. U.Yaramaci for the opportunity to use the NMR relaxometer and the DFG (German Research Foundation) for supporting this study within the research group INTERURBAN. Also, we thank Eleanor Hobley for helping us with the English language.

MATERIALS AND METHODS |

75

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

14.

15. 16. 17.

18. 19. 20.

21.

22. 23.

24. 25. 26. 27. 28.

Altfelder S, Streck T, Richter J. Effect of air-drying on sorption kinetics of the herbicide chlortoluron in soil. J. Environ. Qual., 1999; 28:1154–1161. LeBoeuf EJ, Weber WJ Jr. Macromolecular characteristics of natural organic matter. 2. Sorption and desorption behavior. Environ. Sci. Technol., 2000; 34: 3632–3640. LeBoeuf EJ, Weber WJ Jr. Macromolecular characteristics of natural organic matter. 1. Insights from glass transition and enthalpic relaxation behavior. Environ. Sci. Technol, 2000; 34:3623–3631. Schaumann GE, Siewert C, Marschner B. Kinetics of the release of dissolved organic matter (DOM) from air-dried and premoistened soil material. J. Plant Nutr. Soil Sci., 2000; 163:1–5. Schaumann GE, Antelmann O. Thermal characteristics of soil organic matter measured by DSC: A hint on a glass transition. J. Plant Nutr. Soil Sci., 2000; 163:179–181. Schaumann GE, Hurrass J. Water uptake and swelling behaviour of soil organic matter (SOM). Mitteilgn. Dt. Bodenkundl. Gesellsch., 2000; 92:25–28. Kenyon WE. Nuclear magnetic resonance as a petrophysical measurement tool. Nuclear Geophysics, 1992; 6:153–171. Kenyon WE. Petrophysical principles of applications of NMR logging. Log Analyst, 1997; 38:21–46. Straley C, Rossini D, Vinegar H, Tutunjian P, Morris C. Core analysis by low field NMR. Log Analyst, 1997; 38:84–94. Schirov M, Legchenko A. A new direct non-invasive groundwater detection technology for Australia. Exploration Geophysics, 1991; 22:333–338. Yaramanci U, Lange G, Knödel K. Surface NMR within a geophysical study of an aquifer at Haldensleben (Germany). Geophysical Prospecting, 1999; 47: 923–943. Krüger U. Untersuchungen mit der nuklearmagnetischen Resonanz (NMR) an Gesteinsproben und synthetischen Materialien (NMR investigations with rock and synthetic materials). Thesis. TU Berlin, Inst. of Applied Geosciences, Sect. Applied Geophysics, 2001. Schaumann GE, Hurrass J. 1H-NMR relaxation zur untersuchung des quellungszustandes der organischen bodensubstanz? (1H-NMR relaxation: A method to investigate the state of swelling of soil organic matter?). Mitteilgn. Dt. Bodenkundl. Gesellsch., 2001; 96: 277–278. Schaumann GE, Hurrass J, Berger S, Sieg K, Weber J, Stoffregen H, Mueller M, Rotard W. Swelling of soil organic matter in humic soil samples: Methods and relevance. Proceedings of the 11th Conference of the International Humic Substances Society, Boston, MA (USA), 2002:437–439. Amin MHG, Chorley RJ, Richards KS, Hall LD, Carpenter TA, Cislerova M, Vogel T. Study of infiltration into a heterogeneous soil using magnetic resonance imaging. Hydrological Processes, 1997; 11:471–483. As HV, Lens P. Use of 1H-NMR to study transport processes in porous biosystems. J. Industr. Microbiol. Biotechnol., 2001; 26: 43–52. Sauerbrey R, Sobottka J, Kalähne R, Amin M. Die niedrigauflösende kernmagnetische Protonenresonanz (1H-LR-NMR), eine Meßtechnik zur Bestimmung des Wasserstatus und der Bodenentwicklung von Niedermoortorfböden (Low resolution nuclear proton resonance (1H-LR-NMR), a method to determine the state of water and the pedogenesis of peat soils). Mitteilgn. Dt. Bodenkundl. Gesellsch., 1999; 91:238–241. Stallmach F, Vogt C, Kärger J, Helbig K, Jacobs F. Fractal geometry of surface areas of sand grains probed by pulsed field gradient NMR. Phys. Rev. Lett., 2002; 88:105505/1–4. Belliveau SM, Henselwood TL, Langford CH. Soil wetting processes studied by magnetic resonance imaging: Correlated study of contaminant uptake. Environ. Sci. Technol., 2000; 34:2439–2445. Langford CH, Todoruk T, Litvina M, Kantzas A. Wetting cycles of air-dried and “hydrophobic” soils studied by NMR relaxometry— the role of humic substances. Proceedings of the 11th Conference of the International Humic Substances Society, Boston, MA (USA), 2002:434–436. Grossmann M, Dietrich O, Bangert U, Schwärzel K, Vater G, Hartje V, Kowarik I, Quast J, Wessolek G. Management strategies for regulated wetland ecosystems in the context of global change: case study Spreewald. Berlin: Project ID: 07 GWK 03 (GLOWA-Elbe Subproject 2.2), 2002:1–4. Schwärzel K, Renger M, Sauerbrey R, Wessolek G. Soil physical characteristics of peat soils. J. Plant Nutr. Soil Sci., 2002; 165: 479–486. Berger S, Sieg K, Weber J, Schaumann GE, Rotard W, Kocher B, Wessolek G. Extrahierbarkeit von PAK in abhängigkeit von wassergehalt und befeuchtungsdauer (Extractability of PAH as influenced by the water content and the hydration time). Mitteilgn. Dt. Bodenkundl. Gesellsch., 2001; 96:151–152. Voyutsky S. Colloid Chemistry. Moscow: MIR Publishers, 1978. Doerfler HD. Grenzflächen—und kolloidchemie. Weinheim: Verlag Chemie, 1994. Gaillardon P. Influence of soil moisture on long-term sorption of Diuron and Isoproturon by soil. Pesticide Sci., 1996; 47:347–354. White JC, Quinones-Rivera A, Alexander M. Effect of wetting and drying on the bioavailability of organic compounds sequestered in soil. Environ. Toxicol. Chem., 1998; 17:2378–2382. Zhang YQ, Frankenberger WT Jr, Moore JN. Effect of soil moisture on dimethylselenide transport and transformation to nonvolatile selenium. Environ. Sci. Technol., 1999; 33:3415–3420.

Chapter 9 SORPTION OF PAHs TO NATURAL SORBENTS: IMPACTS OF HUMIC AND LIPID FRACTIONS Luc Tremblay, 1,2 Scott D.Kohl, 1 James A.Rice1 and Jean-Pierre Gagné2 1Department

2Institut

of Chemistry and Biochemistry, South Dakota State University, Brookings, South Dakota 57007–0896, USA

des Sciences de la Mer de Rimouski (ISMER), Université du Québec à Rimouski, Rimouski, Québec G5L 3A1, Canada 9.1. INTRODUCTION

Hydrophobic organic compounds (HOCs) can interact with natural particles such as soils or sediments [1–19] and be removed from aqueous solution. These sorption interactions may result in strong binding and greatly affect the toxicity, transport, and fate of contaminants such as polycyclic aromatic hydrocarbons (PAHs) [16,20–22]. Soil or sediment organic matter (SOM) and their mineral matrices have very different contributions to the sorption. For HOCs, the adsorption on wet minerals is strongly suppressed by competition from water [4,16] and SOM becomes the dominant sorption medium. However, the complexity and heterogeneity of SOM represent a real challenge for the identification of molecular-scale sorption mechanisms [16]. Though the sorption/adsorption of HOCs in/on SOM has been extensively studied, the specific role of the different SOM fractions is still mostly unknown. Sorption to SOM, including humic substances (HSs), has commonly been treated as a hydrophobic solid-phase partitioning [1–5,23,24]. This model is the result of experimental data showing linear sorption isotherms, correlation between the partition coefficient and HOC aqueous solubility, and noncompetitive sorption of HOCs. In the partitioning model, SOM behaves like a homogeneous nonpolar liquid phase that dissolves HOCs. Thus, the uptake is mainly controlled by the water solubility of the sorbate and by the sorbent organic carbon fraction (fOC) [1–4,23]. The ability of SOM to partition highly hydrophobic molecules is relatively constant, no matter what its origin and nature [23]. Recently, many studies dealing with the sorption of HOCs have reported non linear isotherms [6–12] and results that cannot be explained solely by partitioning, such as solute-solute competition [6–8,13,14] or a nonlinear relationship between solute hydrophobicity and uptake by HSs [24]. These observations indicate that site-specific interactions occur between HOCs and at least a part of SOM. Multi-mode interaction models have been proposed to explain this departure from partitioning. The detailed description of these models is beyond the scope of this paper and can be found in references [15–17]. However, we can briefly generalize the characteristics of these models by depicting the two types of sorption domains they generally describe. The first domain is viewed as an amorphous, “soft”, rubbery phase responsible for the partitioning properties of the SOM. The second domain, more condensed and rigid, possesses a higher affinity for HOCs and may contain internal microvoids and adsorption sites. Because there is a limited number of adsorption sites and space in the rigid domain, as the sorbate concentration increases the adsorption and/or diffusion in this domain become negligible over a lower affinity sorption having partitioning characteristics [13,16,18]. This transition from sorption of varying affinities produces nonlinear isotherms. However, the chemical nature of these sorption domains and the causes of sorption nonlinearity are not well understood [16]. Work [7,12] has correlated sorption affinity at low phenanthrene concentrations and isotherm nonlinearity with a decrease in the O/C and H/C atomic ratios of the SOM, both of which decrease with increasing degree of diagenesis. The implication of various SOM fractions, including HSs, in HOCs sorption and their specific behavior were also investigated. However, results are often inconsistent and the effects of some fractions, such as lipids, were ignored. In a recent paper, Chiou et al. [11] show linear isotherms for a nonpolar solute sorbed onto a peat humic acid (HA), while nonlinear isotherms were observed for the whole peat sample and for its humin (HU) fraction. They presumed that HA was free of high-surface-area-carbonaceous material (e.g., charcoal or soot) that they consider to be responsible for the observed nonlinearity of the HU sorption isotherms. Linear sorption on HA [25] and nonlinear behavior for HU (more than for its parent peat) [26] were also noticed by others with different HOCs. Contrary to this view, LeBoeuf and Weber [18], Yuan and Xing [27], and Mao et al. [28] have shown nonlinear isotherms for the sorption of PAHs, including phenanthrene, on HA. They attributed this behavior to the heterogeneous nature of HA having an expanded (periphery) and a condensed (center) component. Moreover, HU can exhibit linear sorption [28], which again indicates that the nature of the sorbent can affect its sorption behavior. Another SOM component, the lipid fraction, may greatly affect the sorption of HOCs, considering its similar hydrophobic nature. However,

MATERIALS AND METHODS |

77

lipids were ignored or unintentionally discarded during sorbent treatments because of its relatively low abundance. Lipid content in most SOM is typically between 1 to 7% and it consists principally of a range of fatty acids, n-alkyl monoesters, nalkanes and sterols [29–31]. Recently, Kohl and Rice [9] reported that the removal of lipids from soils and peats may increase the PAH fraction sorbed. Tremblay et al. [32] confirmed this observation for sedimentary samples. However, the role of lipid content in SOM, including specific fractions, in sorption and the mechanisms involved remain unknown. The objectives of this work were to study the sorption of PAHs to natural sediments and their HSs fractions and to provide data showing the importance and the role of HSs and lipid contents. The sorption processes were studied by sorption kinetics experiments to estimate the equilibrium time and find the mechanisms involved. Some samples were treated to remove or restore lipid content before sorption experiments. The various factors are discussed in terms of sample characteristics, including HS and lipid contents. 9.2. MATERIALS AND METHODS 9.2.1. Materials Particulate Sorbents. Sediments used in this study were collected in the Saguenay Fjord (SAG-05 station, 99 m deep) and the St. Lawrence estuary (Lulu station, 290 m deep) in Québec, Canada, with a Hessler-type box corer. These samples include a batch of homogenized sediments from various locations in the Fjord (0–40 cm layer), called “reference sediment”, and from the first 4 cm layer at specific locations (i.e., SAG-05 and Lulu). Site descriptions can be found elsewhere [33,34]. Suspended particulate matter (SPM) was collected with sediment traps moored for 3 months at 80 m deep in SAG-05. Prior to mooring, the traps were filled with filtered (0.5 µm) dense seawater containing NaN3 to limit biological activity. SPM was recovered by decantation and washed with distilled water. Sediments and SPM were freeze-dried prior to analysis. HA and HU fractions were extracted from these natural particles using the method adopted by the International Humic Substances Society [35]. Briefly, this procedure uses a pretreatment in 0.1M HCl followed by two extractions in 0.1M NaOH under a nitrogen atmosphere. The alkaline supernatants for each extraction are combined, acidified, and centrifuged to collect the HA. The solid residue left after extractions contains the HU fraction. The HU was washed with water before freeze-drying. The ash content of HA is reduced by additional precipitations, treatment with 0.1M HCl/0.3M HF and water-washing prior to freezedrying. In sediment samples, HU represents the dominant SOM fraction (50–70% of its mass), while this proportion is 28% in the SPM. HA constitutes 10 to 14% of the SOM in sediments and 30% in the SPM. The elemental compositions of all the samples studied were obtained with a Perkin-Elmer 2400 Series II CHNS/O elemental analyzer. The results and sample descriptions are listed in Table 9.1. Reagents. Phenanthrene-9-14C (8.5 mCi/mmol, >98% purity) and fluoranthene-3-14C (45 mCi/mmol, >98% purity) used in PAH sorption experiments were obtained from Sigma. Individual PAH stock solutions were prepared in methanol and stored at −20°C in the dark. Phenanthrene (99.5+%) and 2,2,2-trifluoroethyl laurate (>98%) were obtained from Aldrich. Bis(4fluorophenyl)methane (99%), 4-fluoropentylbenzene (97%), and 1-fluorooctane (99%) were purchased from Fluorochem USA (West Columbia, SC). These F-labeled lipids are liquids at room temperature. All other solvents and chemicals were reagent grade or better and were used as received. Aqueous solutions were prepared using Type 1 reagent water (resistivity >18 megohms). Table 9.1 Description of the studied sorbentsa Sorbent

Sampling Location % Cb

% Hc

% Nb

% Ob

% Sd

SPMe SAG-05 HAg SPM SAG-05 Sediment SAG-05 HA Sediment SAG-05 Sediment Lulu HA Sediment Lulu Reference Sediment

48°24.527’N 70°49.515’W

4.7 41.5

0.95 5.1

0.39 4.1

NAf 27.8

0.48 1.3

48°24.566’N

1.5

0.36

0.07

NA

0.22

70°49.988’W

36.1

5.6

3.8

28.4

2.3

48°13.300’N 69°29.480’W

1.3 35.4

0.49 5.9

0.10 6.0

NA 28.0

0.30 2.8

Saguenay Fjord

1.8

NA

0.12

NA

NA

78

| CHAPTER 9: SORPTION OF PAHS TO NATURAL SORBENTS

Sorbent

Sampling Location % Cb

% Hc

% Nb

% Ob

% Sd

HUh Reference various locations 1.1 0.31 0.09 NA 0.13 Sediment a Before lipid extraction and re-addition; b Weight%±2% (relative value); c Weight%±5% (relative value); d Weight %±20% (relative value); e Suspended particulate matter; f Not available; g humic acids; h humin in its mineral matrix.

9.2.2. Methods Lipid Extraction. Lipids are operationally defined as organic compounds that are soluble in nonpolar organic solvents such as hexane, benzene or chloroform [36]. Here, lipid-extracted reference sediment and its HU were prepared by Soxhlet extraction for 72 hr with a benzene:methanol azeotrope (3:1, v/v) [37]. After extraction, the solid was air-dried for 3 days before sorption experiments were begun. Kohl and Rice [9] have indicated that this procedure does not leave significant benzene or methanol sorbed to the particles. The extractable-lipid organic content of the sediment and HU was determined by difference between the elemental analysis of the material before and after Soxhlet extraction. Based on these analyses, the Saguenay reference sediment and HU contained less than 5% of total organic carbon as extractable lipids, which is within the range of the analysis uncertainty. These values are similar to observations made in previous studies on Saguenay sediments, which show that less than 0.1% of the whole sediment consists of lipids (J.-P.Gagné, unpublished data). Because of the relatively small quantity of reference sediment HA available, lipids were extracted from HA in a 7 mL glass tube filled with a benzene:methanol (3:1, v/v) solution and shaken every 6 h. After 24 h, the tube was centrifuged and the solvent was exchanged with a clean benzene:methanol solution and the cycle repeated for a total of three extractions. Re-addition of Fluorinated Lipids. 2,2,2-trifluoroethyl laurate, 4-fluoropentylbenzene, bis(4-fluorophenyl)methane, and 1fluorooctane were sorbed individually on lipid-extracted reference sediment. Fluorine labeled lipids were chosen to allow detection by solid state 19F-nuclear magnetic resonance (NMR). The addition of F-lipids was performed in two different ways. First, the individual neat F-lipid was placed directly onto a dried sorbent sample. The solid was then vigorously stirred by hand for 60 s with a micro spatula for homogenization and stored in the dark at room temperature for 7 days. The proportion of F-lipids added to the total solid material by this technique was 2% of the sediment (w/w) and 5% of the HA (w/ w). The second technique consisted of placing solid particles in a 20 mL vial and adding a saturated F-lipid water solution until the vial was completely filled (i.e., no head space under the cap). The lipid quantity inside the vial was adjusted to 10% of the sorbent (w/w) added to the suspension. The suspensions were stored at room temperature in the dark for 7 days and hand-shaken for 30 seconds four times a day. Particles were collected by centrifugation and air-dried for three days before sorption experiments were begun. 14C-PAH Sorption Experiments. Sorption experiments were carried out as suspensions of solid particles (i.e., sediment, SPM, HU or HA) in a synthetic seawater solution [38] having a salinity of 30.0 ppt, close to that of the Saguenay Fjord and St. Lawrence marine estuary bottom water [33,34]. 200 µg/mL of HgCl2 was added as a bio-inhibitor. The effect of Hg2+ on on the sorption affinity of the sediment was evaluated. An increase of 8% of the phenanthrene fraction sorbed to sediment was observed when HgCl2 was added. Hg2+ forms strong complexes with SOM and can neutralize charges, resulting in a more hydrophobic sorbent [23,29]. Quantities of sorbent used in the vials (i.e., 4–12 mg for HA, 50–150 mg for sediment, SPM, and HU depending on the volume of the vial used) were such that between 50 to 85% of the 14C-compound would be sorbed by the end of the incubation. Five solutions having concentrations ranging over four orders of magnitude (i.e., 30% to 0.021% of PAH water solubility [39]) were prepared by dilution of the methanol stock solution in the artificial seawater. The final methanol concentration in all experiments was less than 0.2% v/v, a concentration insufficient to produce observable cosolvent effects on sorption [40]. PAH solutions (7.00 mL) were transferred to 8 mL glass vials fitted with PTFE-lined caps. For the most diluted concentrations, 50.0 mL of solution in 52 mL glass vials was used. Blank vials containing an individual PAH solution but no sorbent were also prepared at each concentration to account for sorption to vial walls. Comparison between blanks after 7 days of incubation and initial solutions showed that sorption to vials walls was always less than 7% for the 8 mL vial and less than 25% for the 52 mL vial. Three replicates of each suspension and blank were analyzed. The incubation vials were placed in the dark in a water bath at 2.0°C (±0.1°C) to reproduce environmental conditions and hand-shaken for 30 seconds four times a day. This mild agitation avoids breaking sedimentary aggregates but may decrease sorption kinetics compared to constant or strong agitation. Sorption characteristics were measured after 7 days of incubation as discussed later. A kinetic study was performed using only one phenanthrene or fluoranthene concentration in sorption experiments lasting from 5 hours to 49 days at 20°C. Vials were then centrifuged for 15 min at 800 g and the supernatant sampled according to the procedure described by Kohl [37]. For the 8 mL vials, the 14C activity of the supernatant was measured by liquid scintillation counting. The supernatant of the 52 mL vials was concentrated by hexane extraction using 5.

MATERIALS AND METHODS |

79

00 mL of hexane that was then collected and counted for 14C activity. All counting was performed with Ecolume® scintillation cocktail on a Packard Tri-Carb 2100TR liquid scintillation counter. The quantity of PAH sorbed to the particles was determined by activity difference between blank samples and those containing the sorbent. After sorption, we calculated for each vial the distribution ratio, Kd, and the organic carbon normalized distribution ratio, KOC with Eqs. 9.1 and 9.2, (9.1) (9.2) where S is the solid-phase equilibrium sorbate concentration (µg/g), C is the liquid-phase equilibrium sorbate concentration (µg/mL), and fOC represents the organic carbon fraction in the sorbent. For every sorbent, the sorption data form the entire concentration range were then fitted to the Freundlich isotherm. This model has the form of Eq.9. 3, (9.3) where the parameter KF is the Freundlich unit-capacity coefficient and N represents the energy distribution of the sorption process [41]. This model is theoretically related to one consisting of multiple Langmuir terms and is thus a logical choice for heterogeneous sorbents like soils or sediments where various sorption processes can occur [41]. Though a Freundlich isotherm does not provide detailed information about the sorption mechanism involved, it is a simple way to assess isotherm nonlinearity. If N<1, then a limited number of specific high affinity binding sites are available for the sorbate molecules. These sites are saturated at low sorbate concentrations leaving sites that are less attractive [16,23]. By comparison, an N value of unity can be conceptualized as occurring with a sorbent having sorption sites of homogeneous affinity, such as would be expected in a purely partitioning process. Because of the influence of N on KF dimensions (i.e. (µg/g)(µg/mL)-N), the modified Freundlich coefficient, K’F, was used to compare the sorption capacity of the different sorbents. The description of this coefficient can be found elsewhere [42]. Briefly, C in Eq. 9.3 is replaced by a dimensionless reduced concentration (Cr) given by the ratio of C to the supercooled liquid-state solubility (Sscl) of the solute. Thus, K’F represents the sorbent sorption capacity (S) when the aqueous concentration of the solute approaches saturation (i.e. Cr~1). Conversion of standard KF values to K’F values can be done with Eq. 9.4. (9.4) Sscl for phenanthrene and fluoranthene was estimated at 2°C and 30 ppt salinity to be 2.9 mg/L and 0.79 mg/L, respectively, according to published equations [42] using available melting point and heat of fusion data [43], and assuming that the aqueous solubility of both sorbates was a third of that at 25°C in distilled water (i.e., 1.29 mg/L and 0.239 mg/L respectively) [23,39,44]. K’F can be normalized by the organic carbon fraction in the sorbent (fOC) to give K’FOC. 9.3. RESULTS AND DISCUSSION 9.3.1. Sorption Kinetics The results of preliminary experiments on whole reference sediment and its HA fraction are shown in Figure 9.1, where the distribution ratio, Kd, is expressed as a function of incubation time. In all cases, the majority of sorption occurs in the first 7 days, although a small amount of additional sorption occurs at longer incubation times, especially for HA. Slow sorption is a known phenomenon during which Kd for PAHs can double compared to the fast sorption occurring in the first 3 days of contact time [15]. Even though true equilibrium is not reached after 7 days, triplicate measurements showed good sorption reproducibility with relative standard deviation always below 8%. As expected, the higher hydrophobicity of fluoranthene than of phenanthrene and the higher organic content of HA compared with sediment both increase Kd values [23] (see below). Although such experiments leading to macroscopic observations cannot tell much about sorption mechanisms, the observed kinetic curves and trends lead to several hypotheses. Under our experimental setup, the observed sorption process is relatively slow and in agreement with models describing diffusion limitation as the rate-limiting step of at least a part of the sorption phenomenon [12,13,15,16,45]. Though SOM can often be described as a medium showing predominantly partitioning behavior, sorbents can contain a rigid organic phase and internal meso- or micro pores (organic and mineral) that can sorb and/ or adsorb HOCs only after diffusion of the solute to those locations [15,16,45]. Therefore, pore diffusion and diffusion into spatially restricted organic matrices may be the limiting step responsible for the observed low sorption rates. The observed kinetics can reflect many of these diffusive constraints, which may act in series or parallel. In the case of HA in Figure 9.1b, the proportions of mineral pores are assumed to be negligible compared to sediment. Adsorption onto a readily accessible organic surface is unlikely to be a dominant process because it is generally not activated, and is very fast [15,16]. The slow

80

| CHAPTER 9: SORPTION OF PAHS TO NATURAL SORBENTS

Figure 9.1 Variations in distribution ratios with time for PAHs sorption on whole reference sediment (a) and reference sediment humic acids (b). Initial phenanthrene (′ ) and fluoranthene (′ ) concentration of 3.0% and 2.1% of water solubility, respectively. Salinity=30 ppt, temperature=20°C

Figure 9.2 Phenanthrene sorption isotherms for SAG-05 sediment and its humic acids fraction after 7 days of sorption at 2°C. t

sorption fraction observed after 3 days with HA, which is even more important than for whole sediment, can be caused by the highly aggregated nature of the organic structures in SOM, which also have condensed regions [18,27] that retard diffusive absorption, leading to linear or nonlinear sorption [16,45]. The bulk SOM seems more easily accessible to PAHs than HA solid grains when highly dispersed in a mineral matrix. Sediment and HA have larger affinity for fluoranthene than for phenanthrene. This results in a higher sorbed fraction and Kd in Figures 9.1a and 9.1b. However, these Figures show similar sorption curves for both sorbates that correspond to similar apparent equilibration times between the aqueous phase and the individual sorbent. Though the present kinetic experiments use only one PAH concentration and cannot discriminate small rate differences, the differences in PAH molecular sizes and hydrophobicities do not seem important enough to show substantial diffusion rate variations in micropores [45]. 9.3.2. Sorption Isotherms of Whole Sorbents and Humic Substances Figure 9.2 shows an example of phenanthrene sorption isotherms obtained with the SAG-05 sediment and its HA fraction. The results obtained with these sorbents, in the presence or absence of lipids, for phenanthrene and fluoranthene using Eqs. 9. 3 and 9.4 are listed in Table 9.2. The correlation coefficients approach 1 (0.993–0.999), indicating that the data are well described by the Freundlich model. Table 9.2 Freundlich sorption isotherm parameters obtained with the studied sorbents Sorbent

Sorbate

K Fa

K′F, µg/g

Nb

R2

SPMc SAG-05

Phenanthrene Fluoranthene Phenanthrene Fluoranthene Phenanthrene Fluoranthene

840±80d 1500±300 9700±900 17000±3000 310±20 730±60

2260 1200 29000 13500 830 590

0.93 0.93 1.01 1.00 0.92 0.91

0.996 0.993 0.998 0.996 0.998 0.999

HAe SPM SAG-05 Sediment SAG-05

MATERIALS AND METHODS |

Sorbent

Sorbate

K Fa

K′F, µg/g

Nb

R2

HA Sediment SAG-05

Phenanthrene

12100±800

35000

1.00

0.998

Fluoranthene Phenanthrene Fluoranthene Phenanthrene Phenanthrene Fluoranthene Phenanthrene

21000±3000 260±20 1000±100 12000±700 430±50 1100±160 250±20

16000 690 800 34500 1160 880 680

1.00 0.91 0.94 1.00 0.93 0.93 0.95

0.998 0.998 0.998 0.999 0.994 0.997 0.998

Sediment Lulu HA Sediment Lulu Reference Sediment HUf Reference Sediment

81

Fluoranthene 580±60 460 0.90 0.998 Sediment -lipidsg Phenanthrene 450±60 1010 0.76 0.993 HU -lipids 220±20 480 0.72 0.998 HA -lipids 7000±700 19000 0.92 0.996 + (4-fluorophenyl) Phenanthrene 1060±90 2880 0.94 0.999 2methaneh +2,2,2-trifluoroethyl 1800±180 4900 0.95 0.997 lauratei +1-fluorooctanei 780±80 2060 0.91 0.997 aUnits in (µg/g)/(µg/mL)N; bStandard deviation better than±0.02; cSuspended particulate matter; dStandard deviation; eHumic acids; fHumin in mineral matrix; g-lipids=lipid-extracted material; hLipid sorbed to extracted sediment using water solution technique; iLipid sorbed to extracted sediment using the dry sample technique.

For natural samples containing lipids, Figure 9.2 and Table 9.2 show that HA always displays linear sorption (i.e. N=1) while whole sediment and HU isotherms are slightly nonlinear with N values ranging between 0.90 and 0.95, respectively. A linear isotherm for HA agrees with recent reports [11,25]. The linear or nearly linear sorption is consistent with the model describing the sorption on SOM as dominated by a partitioning process [1–5,23,24]. However, lower N values for sediment and HU compared to HA fractions indicate a higher degree of sorbent heterogeneity [11,42]. These sorbents may be involved in more than one sorption process, such as site-specific adsorption, and not only partitioning. No trend in N values is observed among the different PAHs (Table 9.2). Yuan and Xing [27] have shown a decrease in N value with increasing sorbate size, which is believed to be caused by a more efficient trapping of larger molecules in HA adsorption sites. The size difference between phenanthrene and fluoranthene molecules may be too small to observe this steric effect. However, Table 9.2 shows that except for sediment from Lulu station, the modified Freundlich coefficient (K’F) is always larger for fluoranthene than for phenanthrene. In contrast, Carmo et al.[42] found similar K’F values with naphthalene and phenanthrene sorbed on a given sorbent. Because K’F equals S when the solute aqueous concentration approaches saturation (Cr~1), this maximum loading for the more soluble PAHs corresponds to a higher aqueous concentration. This may cancel the lowering effect on S of the sorbate’s lower affinity and produces constant K’F values, especially when the N value is higher (that is, less concave isotherms give higher K’F values) for the more soluble sorbate [42]. In the present case, the higher N value for fluoranthene sorbed on sediment from Lulu station is responsible for the higher K’F exception. For all other sorbents, even similar N values give higher K’F for phenanthrene, indicating a higher maximum sorption capacity of the sorbents for this smaller molecule. However, K’F for different sorbates refers to different solution concentrations. For a given concentration below solute saturation, the distribution ratios or the sorption affinity (and KF if similar N) of a sorbent is always higher for the more hydrophobic sorbate (Figure 9.1 and Table 9.2) [1,23,42]. Figure 9.2 also shows that the PAH sorbed fraction at a given solute concentration is higher for HA compared to parent material, leading to the higher KF and K’F values for HA in Table 9.2. For the HU having the lowest %C value (Table 9.1), these capacity coefficients are the lowest. In fact, as predicted by almost all sorption models, the sorption affinity of these materials increases with its organic carbon content. Because of the influence of N on KF dimensions, KF can only be used to compare the sorption capacity of sorbents having the same N value. Except for SPM SAG-05 and HA SPM SAG-05 that have a slightly lower relative affinity for PAHs, the correlation between KF with similar N values and %C is very good (Tables 9.1 and 9.2). Moreover, the K’F values for both phenanthrene and fluoranthene sorption are well correlated with organic carbon fraction. The correlation curves (not shown) give and for phenanthrene and fluoranthene, respectively. The correlation coefficient for both sorbates is 0.95, indicating that organic carbon content controls the sorbent maximum capacity for PAHs even for nonlinear isotherms. In a previous study, Carmo et al.[42] also found a correlation between K’F and fOC for phenanthrene but with a lower curve slope (i.e. 193). However, they used sorbents with lower range of organic carbon

82

| CHAPTER 9: SORPTION OF PAHS TO NATURAL SORBENTS

Figure 9.3 Organic carbon-normalized distribution ratio (KOC) and modified Freundlich coefficients (K’FOC at Cr~1) for phenanthrene sorption (7 days) on sorbents as a function of initial solution concentration. Concentrations are expressed in multiples of the phenanthrene water solubility. 1=Suspended Particulate Matter (SPM) SAG-05; 2=Sediment SAG-05; 3=Sediment Lulu; 4=Reference Sediment

content, they extrapolated the curve’s y-intercept to zero, and their correlation coefficient was 0.86. These may explain the differences with the present work. K’F denotes the sorbed concentration near solution saturation and thus we must select another parameter to describe the sorbent sorption affinity at lower concentrations. To accomplish this, Kd values at a given PAH concentration (Eq. 9.1) can be used. Correlations of sorption affinity with fOC have been presented in previous studies using the Kd value obtained by considering linear isotherms (i.e., Kd=slope of the isotherm) [e.g., 1,23]. In the present case, Kd also correlates well with foc at all PAH concentrations tested (R2>0.88, data not shown). When SPM SAG-05 and HA SPM SAG-05 of lower affinity are not considered, the correlation coefficient reaches 0.999 for phenanthrene at 3% and 0.03% of water saturation. Moreover, when fOC approaches zero, Kd is very small at intermediate and high PAHs concentrations, leading to correlation y-intercepts, like for K’F curves, statistically not different from zero (t-test, 90% confidence limit). Because the sorbents with low fOC have relatively higher sorption affinities at low PAH concentration (i.e., sorption nonlinearity), there is a deviation of the correlation y-intercept toward positive values at the lowest concentrations tested. Although a good correlation is obtained, sorbents with different N values have different relative affinity at this concentration and may correspond to more than one linear correlation with their fOC. Thus, while the nature of SOM can influence its sorption affinity (see also below), organic carbon content can be considered as the most determining factor for PAH uptake. Total SOM and HS fractions have a different impact on PAH sorption according to sorbate concentrations. The impact of SOM from its nature can be satisfactorily described with the organic carbon normalized distribution ratios (KOC) and the modified Freundlich coefficient (K’FOC). The results for phenanthrene are presented in Figure 9.3, where KOC and K’FOC are shown for the four natural sorbents studied and their HA or HU associated fractions. In this Figure, KOC for the lowest and an intermediate concentration were chosen. K’FOC reflects the sorption behavior at the maximum solute concentration. Although these two parameters are placed in the same Figure, allowing relative comparison between sorbents affinity at a given concentration, they represent different entities with different units and thus absolute values should not be compared. Figure 9.3 shows that the KOC values of all the sediments as well as their HA and HU fractions are almost equal at intermediate phenanthrene concentration. Because of this similarity, whole HS fractions can be viewed to be representative of the total SOM in terms of sorption affinity at this concentration. For SPM material, the slightly lower relative sorption affinity is confirmed. Consistent with our results at intermediate concentration, the partitioning model predicts a constant KOC with values between 20 000 and 40 000 for phenanthrene on whole soils and sediments [3,4,23,39]. This shows that the different origins (e.g., mostly marine at Lulu station vs more terrigeneous at SAG-05 station), structures (e.g., different C/N ratios) and matrixes (e.g., SPM vs sediment or HS) of the SOM studied here do not significantly influence its sorption affinity for phenanthrene at these concentrations. Because some studied sorbents show sorption linearity (i.e., are concentration-independent) while others exhibit nonlinear behavior, the relative sorption affinity varies with sorbate concentration. At low phenanthrene concentration, Figure 9.3 shows higher KOC values for the whole SOM associated with SPM and sediments, indicating a higher impact on sorption compared with their HSs fractions. For fluoranthene (data not shown), the HU fraction has a KOC value similar to those of the sediments because of its lower N value compared with phenanthrene (Table 9.2). Sorption nonlinearity is indeed caused by sorption at lower affinity sites when the sorbate concentration increases above saturation of high affinity sorption sites. Except for SPM SAG-05 and its HA fraction having slightly lower KOC values, the sorption affinity is nearly constant among the whole parent SOM group and the HSs group at this low PAH concentration. The opposite trend is observed at high PAH

MATERIALS AND METHODS |

83

Figure 9.4 Organic carbon-normalized distribution ratio for phenanthrene sorption (7 days) on whole and lipid-extracted reference sediment (a), humin (b), and humic acids (c) as a function of initial solution concentration. Concentrations are expressed in multiples of the phenanthrene water solubility

concentrations. For HA that do not have concave isotherms, the maximum mass of PAH per gram of organic carbon that can be sorbed is relatively larger than for the whole SOM. Because this part of the sorption isotherm does not represent the relatively small fraction of high affinity sites that may be present, the sorption capacities calculated here usually reflect the partitioning behavior to which HAs seem to make a large contribution [1–5,11,23]. 9.3.3. Sorption Isotherms of Lipid-extracted Sediment and Humic Substances The importance and the role of the lipid fraction in the sorption of PAHs were addressed by performing phenanthrene sorption experiments before and after lipid extraction. When lipid extraction is performed, the sorption behavior of all the tested sorbents is considerably modified. The lower part of Table 9.2 and Figure 9.4 clearly show these effects. The removal of lipids increases isotherm nonlinearity (Table 9.2) decreasing the N values for reference sediment, HU and HA to 0.76, 0.72, and 0.92 respectively. Thus, the HU fraction experiences the largest change in N value and displays the highest degree of nonlinearity after lipid extraction. Figure 9.4 presents KOC values at all phenenthrene concentrations used to facilitate direct comparisons between sorbents. Figure 9.4 shows that for whole sorbents the sorption affinity is almost concentration-independent. However, lipid extraction increases the fraction of phenanthrene sorbed by all samples, particularly at low sorbate concentrations. Though this increase is insignificant or very small (t-test, 90% confidence limit) at 30% of phenanthrene’s water solubility (Figure 9.4a–c), the distribution ratio of lipid-extracted sediment (Figure 9.4a) and HU (Figure 9.4b) is more than an order of magnitude larger than for the whole samples at the lowest phenanthrene concentration. For HA (Figure 9.4c), the effect of lipid extraction is less marked. The more important sorption as the phenanthrene concentration is reduced is responsible for the observed increase in isotherm nonlinearity. Using pentane extraction to remove oil in a contaminated soil, Braida et al. [45] also observed a decrease in N value similar to the HA studied here.

84

| CHAPTER 9: SORPTION OF PAHS TO NATURAL SORBENTS

Figure 9.5 Distribution ratio for phenanthrene sorption (7 days) on reference sediment before and after lipid extraction and after re-addition of bis-(4-fluorophenyl)methane as a function of initial solution phenanthrene concentration. Concentrations are expressed in multiples of the phenanthrene water solubility

In spite their low proportions in the sorbents (e.g., <0.1 % of Saguenay sediment), lipids have an important impact on the sorption properties of sediment and specific SOM fractions. The observed increase of the sorption affinity for all geosorbents after lipid extraction, particularly at low phenanthrene concentrations, can be explained by a competitive phenomenon. If we attribute the nonlinear sorption character to site-specific interactions in the particulate SOM (i.e., adsorption and diffusion/ absorption in matrix), it is possible that lipids are competing for those sorption sites in samples. Lipid extraction may open up additional sorption sites that were previously occupied by lipids with a higher affinity for phenanthrene (e.g., adsorp tion sites) compared to the partitioning domain. As the aqueous phenanthrene concentration increases, these sites become saturated and the impact of lipid extraction becomes less and less observable. These results are in agreement with current models that describe sorbent SOM as being composed of a small proportion of adsorption sites and a larger proportion of partition domain, especially for HA [10,11]. A competition phenomenon was observed in other studies using different HOC cosolutes and natural sorbents [8,11,13]. Acting like cosolutes, lipids in geosorbents are usually present at concentrations allowing for competition with HOCs. For instance, the phenanthrene fraction sorbed to sediment in the present study was always lower than 100 µg/g, about 10 times lower than the natural lipid concentration in Saguenay sediment (J.-P.Gagné, unpublished data). The absence of nonlinear sorption and the observation of partitioning characteristics in the presence of lipids in geosorbents may also be caused by a process similar to plasticization in synthetic polymers. After penetration into the “glassy” or rigid domain, lipids may act like plasticizers by weakening macromolecular interactions and disrupting adsorption sites. Plasticizers increase the sorbent’s free volume and the mobility of both chain segments and sorbates [46]. Because the solvents used for lipid extraction can possibly alter the unextracted SOM, it is important to perform additional experiments. BET-(N2) surface area and small angle x-ray scattering (SAXS) measurements of surface morphology in a series of soils by Kohl [376] showed no significant difference before and after extraction. However, N2 sorption may mostly reflect the inorganic surface of the sorbents [42] and be less sensitive to the lipid content. Though lipid extraction increases sediment affinity for phenanthrene sorption and the overall sorption rate, a previous study indicated similar sorption kinetics curves leading to a similar apparent equilibrium time with the aqueous phase [32]. Therefore, the removal of lipids does not seem to noticeably disturb the sorption rate-limiting step. Further evidence that the isotherm nonlinearity variations shown in the lower part of Table 9.2 are caused by the presence or absence of lipids is obtained from the interactions of the F-labeled lipids with lipid-extracted Saguenay sediment prior to sorption experiments with 14C-phenanthrene. The lowest part of Table 9.2 shows the impact of the re-addition of lipids on the Freundlich isotherm parameters. In all cases, re-addition of lipids reduces the nonlinear sorption character and restores the coefficient (N) of the whole natural sediment to approximately its original value (i.e., 0.93±0.02). Figure 9.5 presents the effect of lipid extraction and lipid re-addition on the phenanthrene sorption affinity of the sediment expressed by the distribution ratio (Kd). Only the results obtained with bis(4-fluorophenyl)methane are presented in Figure 9.5 although similar results were obtained with all the lipids tested. Figure 9.5 reveals higher sorbed phenanthrene concentrations after re-addition of lipid compared to the whole sediment containing natural lipids. The increase of Kd at all initial phenanthrene concentrations explains the higher Freundlich capacity parameters values after lipid re-addition (Table 9.2). This probably is caused by the addition of an excess of F-lipids (i.e. above saturation of higher affinity sites), leading to a lipid and OM content higher than in the natural sediment, and an increased sorption capacity for HOCs or higher K’F values. Other studies indicate that the Kd of trace HOCs sorbed onto natural particles are correlated and increase with lipid content [47,48]. Excess lipids may be absorbed or “dissolved” in the partition domain or may form a lipidic domain that promotes partitioning at all HOC concentrations, which in turn leads to nearly linear isotherms. Braida et al. [45] supported this view and believe that oils in contaminated soil offer a linear partition domain. The general trend of K’F and N values

MATERIALS AND METHODS |

85

decrease after lipid extraction and the opposite after re-addition suggest that lipids can constitute a partition domain, increasing sorption capacity at high sorbate concentrations. Figure 9.5 also shows that at low phenanthrene concentrations, the Kd values of the sediment never reach the value obtained with the lipid-extracted sediments that have a lower organic carbon content. This behavior at low sorbate concentrations is inconsistent with the partitioning model [1–4,23]. This observation again supports the concept that lipids compete with phenanthrene for a limited number of specific sites in a multi-mode sorption process or act like plasticizers, disrupting the SOM matrix. The presence in SOM of at least two sorption domains where lipids possess different mobility, such as rigid sites in competition with PAHs and a liquid-like partition domain, were studied at a molecular-scale in previous NMR studies [32,49]. 9.4. CONCLUSIONS The present work provides new data on the effects of different SOM fractions on the sorption behavior of PAHs. When lipids are present in natural sorbents, good correlation between K’F or Kd at various sorbate concentrations and the proportion of organic carbon indicates the importance of this last parameter for the uptake of a given PAH. Moreover, the constancy of the organic carbon normalized distribution ratios, especially at intermediate PAH concentrations, reveals that the quantity of SOM is much more significant for sorption than its nature, at least when lipids are present in the sorbent. Nevertheless, the organic matter of the SPM has a slightly lower relative affinity for PAHs than that of the sediments studied. Moreover, even if the HSs fractions can be described as fairly representative of their parent total SOM in terms of their sorption affinity, these two groups of sorbents have different impact on sorption as a function of the sorbate concentrations. HAs have a higher sorption capacity near PAH saturation that seems to be a dominant partitioning medium, while the rest of SOM has a higher impact on sorption at low PAH concentrations, leading to slightly nonlinear isotherms. Therefore, the notion of relative sorption affinity between sorbents can be concentration-dependent. As in the present study, Carmo et al.[42] found that sorbents with the lowest K’FOC can have the highest KOC at low PAHs concentrations. Contrary to HS, the lipid content greatly affects the SOM relative sorption affinity (KOC) for PAHs despite the low natural abundance of lipids in the sorbents studied. The extraction of lipids from SOM drastically increases its sorption affinity, especially for sediment and HU, at low phenanthrene concentrations, resulting in increased isotherm nonlinearity. An increase of sorption affinity while the organic carbon content is reduced is inconsistent with the partitioning model and a multimode sorption must be considered. Our results support a competitive-type mechanism between lipids and phenanthrene, or a plasticizer effect of lipids, for high affinity rigid domains. Sediment and HU seem to contain a higher proportion of these sites than HAs. Although HAs can also be a heterogeneous sorbent for HOCs as revealed by lipid extraction (N=0.92), only a small amount of lipids is necessary to occupy or disrupt all the high affinity sorption sites in HA and cause linear sorption in lower affinity sites. When lipids are present in the sorbent at a concentration above that needed to saturate the higher affinity sorption sites, they promote phenanthrene uptake in a manner having partitioning characteristics. There are different causes of sorption nonlinearity found in the literature, such as high-surface-area-carbonaceous material or rigid SOM domains [e.g., 7–17], increasing C/O ratio [7,12] and aromatic content [19]. In the present study, nonlinear sorption is only observed after lipid extraction. This work emphasizes the importance of the selection of appropriate sorbent treatments when attempting to explain the cause of HOC sorption isotherm nonlinearity or behavior. If treatments employed remove lipids, one can expect an increase of uptake at low sorbate concentrations for heterogeneous sorbents. This may explain why the relatively higher sorption affinity of HU compared to HA shown by Chiou et al. [11] with different sorbents is only significantly observable here after lipid extraction. Artificial ageing with hot water under pressure [12] or extensive extractions [11] may partially extract lipids and complicate the determination of the real cause of sorption nonlinearity variations or of the sorption behaviors observed. Taken together, these results support a multi-mode sorption of PAHs in SOM, including HS that involves partitioning within aggregated particles and site-specific interactions/diffusion in a small proportion of high affinity sites that become more available after lipid extraction. Considering the slow rates in Figure 9.1 and the absence of correlation between sorption nonlinearity and sorbent surface area [10,13], even after lipid extraction [37], adsorption onto readily accessible organic surfaces is unlikely to be important [15,16]. Diffusion processes in micropores and space-limited SOM matrix seem to be the rate-limiting steps for PAH uptake. ACKNOWLEDGEMENTS We acknowledge Yves Gélinas (Concordia University, Montréal) for his generous contribution in elemental analyses and the crew of the Alcide C.Horth and Martha L. Black for the sediment sampling. This research is supported by grants from the U.S. Office of Naval Research’s Harbor Processes Program to J.A.R. (contract no. N00014–99–0587), the Natural Science and

86

| CHAPTER 9: SORPTION OF PAHS TO NATURAL SORBENTS

Engineering Research Council of Canada (NSERC) to J.-P.G., and a scholarship from Le Fonds pour la formation de Chercheurs et l’Aide à la Recherche (FCAR) to L.T. REFERENCES 1. 2. 3. 4.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

Karickhoff SW, Brown DS, Scott TA. Sorption of hydrophobic pollutants on natural sediments. Water Res., 1979; 12:241–248. Chiou CT, Peters LJ, Freed VH. A physical concept of soil-water equilibria for nonionic organic compounds. Sci., 1979; 206: 831–832. Karickhoff SW. Semi-empirical estimation of sorption of hydrophobic pollutants on natural sediments and soils. Chemosphere, 1981; 10:833–846. Chiou CT. Roles of organic matter, minerals, and moisture in sorption of nonionic compounds and pesticides by soil. In: MacCarthy P, Clapp CE, Malcolm RL, Bloom PR eds. Humic substances in soil and crop sciences: Selected readings. Madison: American Society of Agronomy and Soil Science Society of America, 1990:111–160. Kile DE, Chiou CT, Zhou H, Li H, Xu O. Partition of nonpolar organic pollutants from water to soil and sediment organic matter. Environ. Sci. Technol., 1995; 29:1401–1406. McGinley PM, Katz LE, Weber WJ Jr. A distributed reactivity model for sorption by soils and sediments. 2. Multicomponent systems and competitive effects. Environ. Sci. Technol., 1993; 27:1524–1531. Huang W, Weber WJ Jr. A distributed reactivity model for sorption by soils and sediments. 10. Relationships between desorption, hysteresis and the chemical characteristics of organic domains. Environ. Sci. Technol., 1997; 31: 2562–2569. Chiou CT, Kile DE. Deviation from sorption linearity on soils of polar and non-polar organic compounds at low relative concentrations. Environ. Sci. Technol., 1998; 32:338–343. Kohl SD, Rice JA. Contribution of lipids to the nonlinear sorption of polycyclic aromatic hydrocarbons to soil organic matter. Org. Geochem., 1999; 30: 929–936. Xia G, Ball W. Adsorption-partitioning uptake of nine low-polarity organic chemicals on a natural sorbent. Environ. Sci. Technol., 1999; 33:262–269. Chiou CT, Kile DE, Rutherford DW, Sheng G, Boyd SA. Sorption of selected organic compounds from water to a peat soil and its humic-acid and humin fractions: potential sources of the sorption nonlinearity. Environ. Sci. Technol., 2000; 34:1254–1258. Johnson MD, Huang W, Weber WJ Jr. A distributed reactivity model for sorption by soils and sediments. 13. Simulated diagenesis of natural sediment organic matter and its impact on sorption/desorption equilibria. Environ. Sci. Technol., 2001; 35:1680–1687. Xing B, Pignatello JJ, Gigliotti B. Competitive sorption between atrazine and other organic compounds in soils and model sorbents. Environ. Sci. Technol., 1996; 30:2432–2440. Cornelissen G, Van Noort PCM, Govers HAJ. Mechanism of slow desorption of organic compounds from sediments—A study using model sorbents. Environ. Sci. Technol., 1998; 32:3124–3131. Pignatello JJ, Xing B. Mechanisms of slow sorption of organic chemical to natural particles. Environ. Sci. Technol., 1996; 30:1–11. Luthy RG, Aiken GR, Brusseau ML, Cunningham SD, Gschwend PM, Pignatello JJ, Reinhard M, Traina SJ, Weber WJ Jr, Westall JC. Sequestration of hydrophobic organic contaminants by geosorbents. Environ. Sci. Technol., 1997; 31:3341–3347. Weber WJ Jr. Sequestration of organic solutes by natural geosorbents: Equilibrium insight from polymer sciences. Abstracts of Papers, 220th National Meeting of the American Chemical Society, Washington, DC: American Chemical Society 2000; 40: 109–110. LeBoeuf EJ, Weber WJ Jr. A distributed reactivity model for sorption by soils and sediments. 8. Sorbent organic domains: Discovery of a humic acid glass transition and an argument for a polymer-based model. Environ. Sci. Technol., 1997; 31:1697–1702. Xing B, Chen Z. Spectroscopic evidence for condensed domains in soil organic matter. Soil Sci., 1999; 164:40–47. Alexander M. How toxic are toxic chemicals in soil? Environ. Sci. Technol., 1995; 29:2713–2717. Linz DG, Nakles DV. Environmentally acceptable endpoints in soil. Annapolis: American Academy of Environmental Engineers, 1997. Liu H, Amy G. Modeling partitioning and transport interactions between natural organic matter and polynuclear aromatic hydrocarbons in groundwater. Environ. Sci. Technol., 1993; 27:1553–1562. Schwarzenbach RP, Gschwend PM, Imboden DM. Environmental organic chemistry. New York: Wiley, 1993. Kopinke F-D, Pörschmann J, Stottmeister U. Sorption of organic pollutants on anthropogenic humic matter. Environ. Sci. Technol., 1995; 29:941–950. Boonsaner M, Hawker DW, Connell DW. Partitioning of higher molecular weight chlorohydrocarbons in a humic acid/water system. Toxicol. Environ. Chem., 1999; 70:129–147. Pignatello JJ. A revised physical concept of natural organic matter as a sorbent of organic compounds. In: Sparks DL, Grundl TJ eds. Kinetics and mechanisms of reactions at the mineral/water surface. Washington: American Chemical Society, 1998:204–211. Yuan G, Xing B. Effects of metal cations on sorption and desorption of organic compounds in humic acids. Soil Sci., 2001; 166: 107–115. Mao J-D, Hundal LS, Thompson ML, Schmidt-Rohr K. Correlation of poly(methylene)-rich amorphous aliphatic domains in humic substances with sorption of the nonpolar organic contaminant phenanthrene. Environ. Sci. Technol., 2002; 36:929–936. Stevenson FJ. Humus chemistry: Genesis, composition, reactions. 2nd Edn. New York: Wiley, 1994.

MATERIALS AND METHODS |

30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48.

49.

87

Colombo JC, Silverberg N, Gearing JN. Biogeochemistry of organic matter in the Laurentian Trough, II. Bulk composition of the sediments and relative reactivity of major components during early diagenesis. Marine Chem., 1996; 51: 295–314. Wakeham SG, Lee C, Hedges JI, Hernes PJ, Peterson ML. Molecular indicators of diagenetic status in marine organic matter. Geochim. Cosmochim. Acta, 1997; 61:5363–5369. Tremblay L, Kohl SD, Rice JA, Gagné J-P. Geosorbent lipid content as a cause of nonlinear sorption of polycyclic aromatic hydrocarbons. Environ. Sci. Technol., submitted. Deflandre B, Mucci A, Gagné J-P, Grignard C, Sundby B. Early diagenetic processes in coastal marine sediments disturbed by a catastrophic sedimentation event. Geochim. Cosmochim. Acta. 2002; 66:2547–2558. El-Sabh MI, Silverberg N. Oceanography of a large-scale Estuarine system: The St. Lawrence. New York: Springer, 1990. Swift RS. Organic matter characterization. In: Methods of soil analysis. Madison: Soil Science Society of America and American Society of Agronomy, SSSA Book Series #5, 1996:1011–1069. Bergman W. Geochemistry of lipids. In: Breger IA ed. Organic geochemistry. New York: Pergamon, 1963:503–542. Kohl SD. Ph.D. Dissertation, South Dakota State University, 1999. Millero FJ. Chemical oceanography. New York: CRC Press, 1996. Mackay D, Shiou WY, Ma KS. Illustrated handbook of physical-chemical properties and environmental fate for organic chemicals. Chelsea, MI: Lewis, 1992. Rao PSC, Lee LS, Pinal R. Cosolvency and sorption of hydrophobic organic chemicals. Environ. Sci. Technol., 1990; 24:647–654. Weber WJ Jr, DiGiano FA. Process dynamics in environmental systems. New York: Wiley-Interscience, 1996. Carmo AM, Hundal LS, Thompson ML. Sorption of hydrophobic organic compounds by soil materials: Application of unit equivalent Freundlich coefficients. Environ. Sci. Technol., 2000; 34:4363–4369. Lide DR. CRC handbook of chemistry and physics. Boca Raton: CRC Press, 1991. Whitehouse BG. The effects of temperature and salinity on the aqueous solubility of polynuclear aromatic hydrocarbons. Marine Chem., 1984; 14:319–332. Braida WJ, White JC, Ferrandino FJ, Pignatello JJ. Effect of solute concentration on sorption of polyaromatic hydrocarbons in soil: Uptake rates. Environ. Sci. Technol., 2001; 35:2765–2772. Elias H-G. Macromolecules. New York: Plenum Press, 1977. Turner A, Tyler AO. Modelling adsorption and desorption processes in estuaries. In: Jickells TD, Rae JE eds. Biogeochemistry of intertidal sediments. Cambridge: Cambridge University Press, 1997:43–58. Broman D, Näf C, Rolff C, Zebühr Y. Occurence and dynamics of polychlorinated dibenzo-p-dioxins and dibenzofurans and polycyclic aromatic hydrocarbons in the mixed surface layer of remote coastal and offshore waters of the Baltic. Environ. Sci. Technol., 1991; 25:1850–1864. Kohl SD, Toscano PJ, Hou W, Rice JA. Solid-state 19F NMR investigation of hexafluorobenzene sorption to soil organic matter. Environ. Sci. Technol., 2000; 34:204–210.

Chapter 10 INTERACTIONS AND CONVERSIONS OF POLYCYCLIC AROMATIC COMPOUNDS IN THE PROCESS OF HUMIFICATION Matthias Hübner,1 Kristoffer E.N. Jonassen2 and Torben Nielsen2 1Centro 2Risø

Ricerche Ambientali, Montecatini, I-48023 Marina di Ravenna, Italy

National Laboratory, Plant Research Department, P.O.B. 49, DK-4000 Roskilde, Denmark

10.1. INTRODUCTION Numerous contaminated sites around the world bear witness to the fact that over the decades, polycyclic aromatic compounds (PACs) have been released into the environment on a large scale by human activity. PAC contaminations still constitute a source of eminent risks for human health. Such contaminations often coincide with natural organic matter in both soil and water. Natural organic matter is largely composed of humic substances (HSs), which together with non-HSs form a so-called system of HSs (SHS) [1]. HSs are presumed to account for a considerable part of PAC sorption [1]. The structural diversity found in HSs samples results in a variety of interactions between PACs and HSs, starting from sorption phenomena and ending with conversions of the PACs’ molecular skeleton in the process of humification. Humification is accomplished by microbial action and/or in a purely chemical way. Sorption may be the first step of chemical transformations in humification. Chemical PAC conversion and microbial degradation both are affected by sorption. Risk assessment of polluted sites thus requires knowledge of PAC sorption by HSs as much as knowledge of the composition of HSs and their capacity to assist PAC conversion. PAC sorption by HSs may have both positive and negative impact on PAC bioavailability: on the one hand it may decrease the direct risk of uptake by humans by diminution of the free concentration [2,3]. On the other hand there have been ambiguous reports on PAC degradability by microorganisms [4–6]. In some cases an increased rate of degradation was found and traced back to matrix-assisted co-metabolic effects [7–9]. In other cases HSs seem to have blocked microbial access to the contaminant [10], especially in soil with long-term immobilization, either based on physicochemical processes [11,12] or biologically mediated immobilization [13,14]. The types of interaction observed are largely determined by the properties of the HSs matrix. Understanding the fate of PACs in soil and water would help us to make predictions of the remobilization potential of PACs, especially if environmental conditions (pH, salinity, and temperature) change. Thus, risk assessment would be facilitated and strategies for the bioremediation of polluted sites could benefit significantly. Our investigations are aimed at confronting sorption properties of different PACs and at elucidating the influence of varying ambient conditions on the ‘remobilizability’ of PACs. Secondly, possible pathways of purely chemical (non-microbial) PAC conversion in the process of humification are contemplated and their applicability for alternative remediation procedures is tested. 10.2. MATERIALS AND METHODS 10.2.1. General Remarks Solutions of aqueous humic substances were obtained according to [15], allowing naturally present non-HSs to remain in the sample. Low OC laboratory water was produced with a MilliQ water purification system (Millipore). Glassware was cleaned with detergents and several rinses with water and acetone. Glassware in section 2.2. was in addition heated to 450°C for 18 h while flushing with air filtered through activated carbon. All reactions and preparations were carried out under standard conditions unless specified otherwise. For Teflon® tubing in HPLC, rinses with 0.01 M sodium hydroxide, water, 0.1 M hydrochloric acid, and again water were used. For sorption experiments, 250-mL Amberlite storage vessels (Duran-Schott, Germany) were used. For SPME measurements, 2-mL wide-neck screw cap vials equipped with a Teflon®-coated septum (Supelco) and a magnetic glass stir bar were used. Prior to use in SPME, all glassware was silanized by swirling with a solution of trimethylsilyl azide in ether. Elemental analyses were carried out on a LECO CS200 instrument.

MATERIALS AND METHODS |

89

10.2.2. HPLC Experiments with Immobilized HSs Material Chemicals. The test compounds [naphthalene (NA), flourene (FL), phenanthrene (PH), anthracene (AN), pyrene (PY), quinoline (QI), isoquinoline (IQ), acridine (AC), 5,6-benzoquinoline (5B), 7,8-benzoquinoline (7B) and 1,2-benzacridine (BA)] were obtained from diverse commercial sources. Respective solutions were prepared by dissolving in methanol (Lichrosolv 99.8%, Merck, Darmstadt) with a typical concentration of 0.04 g L−1. 0.1 M phosphate buffer was made from potassium dihydrogenphosphate and sodium hydroxide in water, adjusted to pH 7. Preparation of Chemically Bonded HS Silica Gel for HPLC Columns. The silica gel with chemically bonded HSs was prepared in a number of stages according to a procedure almost identical to Kollist-Siigur et al. [20] and developed by Nielsen et al. [16] and Szabo and Bulman [17]. A suspension of silica gel (Nucleosil−Si−300–20, Macherey-Nagel; 20 g, activated at 130°C) and 3-(triethoxisilyl)propylamine (Merck, 98%, 5 mL) in toluene (100 mL) was stirred at ambient temperature in a nitrogen atmosphere for 24 h. The crude product was separated from the supernatant by centrifugation and consecutively washed with toluene, methanol, water, and methanol (2×50 mL each, followed by centrifugation after each washing step). 3 h of drying under reduced pressure (2.0 kPa) at 50°C gave 19.0 g of aminopropyl silica gel, which was subsequently subjected to three different ways of modification: (a) Preparation of silica gel modified with aqueous HSs. 5 g of the aminopropyl modified silica gel were added to 500 mL of an aqueous HS filtrate [15] (nominal range of molar masses according to ultrafiltration: 1–100 kDa; DOC content: 82 mg prior to reaction, 69.5 mg following the reaction) and stirred for 21.5 h in a nitrogen atmosphere. After deactivation of amino groups and work up (see below), 3.105 g of a brown product (according to elemental analysis 0.84% C and 0.00% S) were obtained. (b) Preparation of silica gel modified with soil HSs. 4 g of the aminopropyl modified silica gel were added to 380 mL of a soil HS filtrate [15] (filtered through a G4 glass filter and subsequently with a 0.45 µm cellulose acetate membrane filter; DOC content: 99.54 mg prior to reaction and 26.65 mg following the reaction) and stirred for 21.5 h in a nitrogen atmosphere. After deactivation of amino groups and work up (see below), 4.562 g of a brown silica gel were obtained (according to elemental analysis 1.72% C and 0.00% S). (c) Preparation of reference (blank) column material. 2 g of the aminopropyl modified silica gel were added to 60 mL of water and spiked with a solution of formaldehyde (37%, 3 mL, 0.04 mol, stabilized with methanol) and stirred for 12.5 h at ambient temperature in a nitrogen atmosphere. After work up, 1.926 g of a white silica gel modified with formaldehyde were obtained (according to elemental analysis 0.54% C and 0.00% S). Deactivation of free amino groups. To 1 g aminopropyl-HS modified silica gel an excess of a solution of formaldehyde (37%, stabilized with methanol) were added and stirred for 12.5 h at ambient temperature in a nitrogen atmosphere. Work-up procedure. The reaction product was centrifuged and consecutively washed with 0.5 M phosphate buffer (pH 7.5, 10×10 mL, one minute treatment with ultrasound), and water (2×10 mL) with the solid reaction product retained after centrifugation. Following overnight treatment with water (10 mL) under agitation, subsequent washing with water (5×20 mL, one minute treatment with ultrasound) and centrifugation, the product was dried overnight at 60°C. The gels obtained with procedures (a)–(c) were suspended in methanol and packed into the HPLC columns (id 4.6 mm, length 12 cm) at 35 mPa. Mobile Phases. Mixtures of methanol (Merck Lichrosolv, 99.8%) and 0.1 M phosphate buffer were used as eluents. The content of methanol was adjusted to the indi vidual retention properties of each PAC on the different stationary phases (up to 40% of methanol was added for experiments on HSs columns (a and b) and up to 20% for experiments on the reference column (c)). In order to determine correlations between the capacity coefficient of PAC and the ionic strength of the eluent, phosphate buffer concentrations of 0.005, 0.01, 0.05, and 0.1 M were chosen (all at pH 7). HPLC Instrumentation. A low-pressure gradient Shimadzu LC-10 HPLC system with photodiode array detector, thermostated column oven and auto-injection unit was used. Appropriate mixtures of the compounds or single compounds were injected and the retention time of each compound was recorded in duplicate experiments with an average injection amount of about 2 µg. Correct peak identification was guaranteed by UV spectral control. The dead time (t0) of the system, used for calculating the capacity coefficient (k’=(tr−t0)/t0; tr=retention time) of each PAC was determined from the solvent peak as an internal standard in each HPLC spectrum. The standard deviation of the coefficient was below 2%. System Class LC 10 software was used for instrument control and data processing. With the exception of the temperature dependent measurements, all experiments were carried out at 25°C. In correlation with the flow rate adjusted between 0.4 and 2 mL min−1, applied pressure varied between 3 and 11 MPa; at 15°C pressure rates reached 14 MPa for aqueous humic substances and 18 MPa for soil humic substances (flow rate=2 mL min−1).

90

| CHAPTER 10: INTERACTIONS AND CONVERSIONS OF POLYCYCLIC AROMATIC COMPOUNDS

10.2.3. Sorption Experiments with SPME Chemicals. All reagents were of analytical grade and purchased from Carlo Erba, Italy unless specified otherwise. Anthracene (puriss.) was obtained from Fluka, Switzerland. Solutions were obtained by diluting or dissolving corresponding reagents with high purity water (Milli-Q system, Millipore) or 99.9% methanol (Baker). Sample Preparation. Aqueous HSs solution (126 mL) [15] was diluted with high purity water (266 mL). The resulting solution was spiked with a solution of anthracene in methanol (1 mg L−1, 8mL), yielding a sample of 400 mL with an anthracene concentration of 0.02 ppm and a HSs concentration (incl. non-HSs) of 970 mg L−1. The samples were light protected and agitated on a shaker at 100 rpm and room temperature for 22 days. Variations of the concentration were monitored by solid phase micro extraction (SPME) and subsequent GC analysis. Each experiment was run in triplicate. SPME-GC Measurements. For SPME measurements, Supelco equipment consisted of a holder and a fibre-syringe with a 100 µm polydimethylsiloxane non-bonded coating. The fibre was activated at 270°C for 12 h under nitrogen. Consecutively, the fibre was equilibrated with the matrix in a tenfold cycle of sample extraction and desorption with the procedure described below for the main analysis. Leaving a small head space, 1.6-mL samples were pipetted into screw cap vials with septum closure and extracted for 30 min with an SPME fibre under rapid stirring (maximum velocity) with a glass stir bar at 27°C (thermostated). The SPME fibre was fully immersed and placed at the half radius of the sample vial to guarantee maximum agitation around the fibre and maximum mass transfer rates. Analytical gas chromatography was carried out on a HP 5890 Series II instrument equipped with an SPME liner, an AT-5I column (30 m×0.25 mm, coated with a 1.0 µm film of cross-linked 5% PHME siloxane) and an FID detector (300°C). In all SPME experiments, anthracene was desorbed from the fibre over a 3-min period at an injector temperature of 270°C and a column temperature of 90°C in the splitless mode (no carry-over was observed under these conditions). A temperature program (3 min 90°C, 50°C/min to 250°C, 5°C/min to 270°C then 4 min at 270°C) at 0.6 mL min −1 helium flow gave a retention time of 9.40 min for anthracene. All SPME measurements were carried out in the exhaustive mode with triple extraction of one sample and extrapolation to the totally extractable amount of analyte. The results were referred to a blank sample of only the analyte in water treated in the same manner. 10.2.4. Experiments on pH Dependent PAC Sorption and Covalent PAC Interaction Chemicals. All reagents were of analytical grade and purchased from Carlo Erba, Italy unless specified otherwise. pH Dependent Variations of PAC Concentrations. A PAC solution was prepared from PAC Standard Mix 1-mL capsules (2 mg/mL in dichloromethane/benzene—1:1, AccuStandard) by diluting the content of one capsule to 200 mL with methanol (99.9%, Baker) yielding a concentration of 10 mg L−1. 6mL (0.06 mg PAC) of that solution was added to three samples with water (300 mL, reference samples), 54 mL (0.54 mg PAC) of the solution were added to an aqueous HS solution [15] (2 L), which was diluted with water (700 mL). The concentration of HS and non-HS corresponded to 2.2 g L−1 (freeze-dried product). The final PAC concentration was 200 ppb in each sample. The samples were stored in 0.5-L and 2.5-L glass bottles and maintained at room temperature for 15 d under light protection and with manual short-term agitation each day. After that period, the HS sample was divided into portions of 300 mL each and the pH was adjusted (Hanna HI 98150 GLP pH/ORP Meter) in each sample to one of nine values ranging from 1.9 to 12.2 by addition of conc. or 1 M hydrochloric acid or sodium hydroxide solution. The reference samples were adjusted to pH 1, 6.5, and 12. The titration volumes of acid/base added were supplemented with water to a total addition of 10 mL. PAC were subsequently extracted by LLE as described below. Fe(III) Mediated Transformations of PAC. A PAC solution was prepared from PAC Standard Mix 1-mL capsules (2 mg/ mL in dichloromethane/benzene—1:1, AccuStandard) by diluting the content of one capsule to 200 mL with methanol (99.9%, Baker) yielding a concentration of 10 mg/L. Eight mL (0.08 mg PAC) of that solution was added to water (Milli-Q, Millipore) or to diluted HS solution [15] (Table 10.1), Table 10.1 Fe(III) mediated conversions of PAC: experimental overviewa N° Name

PAC mix, mL HS, solution mL High purity water, mL Amount of reagent, final concentration (mol/l), additional water(mL).

Final pH, ca

1 2 3 4

8.0 8.0 8.0 8.0

6.5 1.0 3.0 2.0

Ref HCl 100 Fe 1 Fe 10

427.0 418.7 400.0 400.0

– 8.3 mL (0.25 M)Ab 0.27 g (0.0025 M) Bb, 27.0 2.7 g (0.025 M) Bb, 27.0

MATERIALS AND METHODS |

N° Name

PAC mix, mL HS, solution mL High purity water, mL Amount of reagent, final concentration (mol/l), additional water(mL).

5 Fe 100 8.0 400.0 27.0 g (0.25 M) Bb, 27.0 6 Fe 100 HS 8.0 280.0 120.0 27.0 g (0.25 M) Bb, 27.0 7 HS 8.0 280.0 147.0 a All experiments were conducted at 298±2 K, b A: conc. HCl, B: FeCl •6H O 3 2

91

Final pH, ca 1.0 1.0 7.5

respectively. The concentration of HSs and non-HS corresponded to 2.2 g L−1. The total PAC concentration was 200 ppb in each sample. The samples were agitated in 1-L Amberlite glass bottles for 24 hours at 120 rpm on a shaker under light protection and at room temperature. Subsequently, solutions of different amounts (Table 10.1) of reagent [FeCl3•6H2O (ICO, RPE-ACS, >99%, CarloErba, Italy) were prepared in water (27 mL); conc. HCl] was added to the reaction mixtures. Agitation was continued at 120 rpm for 20 d at room temperature. The reaction was stopped by liquid liquid extraction of the substrate. LLE (Liquid-Liquid Extraction) Procedure. After addition of sodium chloride (5 g) and water (100 mL), the reaction mixture was extracted with hexane (4×100 mL, Merck, for organic trace analysis). Since the presence of HSs material favored the formation of emulsions, the layers were allowed to separate for two hours in the first three steps and overnight in the fourth extraction step. In each extraction step, the roughly separated layers were slowly transferred into different bulbs so that further separation was facilitated and residues of the aqueous layer in the organic layer could be removed by pipetting and decanting into another bulb. Without drying, the combined organic layers were carefully concentrated under reduced pressure to give a 1.6-mL sample ready for GC analysis. GC Analyses. An internal standard (100 µL) containing a solution of 50 ppm o-terphenyl and 50 ppm triphenylbenzene in hexane (Aldrich) was added to each sample. Analytical gas chromatography was carried out on a Varian CP-3800 GC instrument equipped with a 8200 auto-sampler, GC liner, a Zebron ZB-5 7HG-G002– 11AT-5I column (30 m×0.25 mm ID, coated with a 0.25 µm film of FT S/N, 5% phenylpolysiloxane) and an FID detector. The injector was adjusted to 310°C ansd run in the split mode. A temperature program (2 min 40°C, 40°C/min to 140°C, 8°C/min to 240°C, 2°C/min to 280°C, 15 min 290°C) run at 0.8 mL min−1 nitrogen flow gave the following retention times (min): 6.86 (naphthalene), 9.58 (acenaph thylene), 9.97 (acenaphthene), 11.14 (flourene), 13.57 (phenanthrene), 13.70 (anthracene), 14.17 (carbazole), 14.75 (o-terphenyl—internal standard), 15.80 (anthraquinone), 16.90 (flouranthene), 17.58 (pyrene), 22.52 (benz[a]anthracene), 22.74 (chrysene), 28.83 (benzo[b] flouranthene), 29.01 (benzo[k]flouranthene), 30.83 (benzo[a]pyrene), 32.42 (triphenylbenzene—internal standard), 38.49 (indenopyrene), 38.81 (dibenz[a]anthracene) and 40.31 (benzo[g,h,i]perylene). Peak assignment was verified by GC-MS control (HP 5890 Series II). All results were referred to the reference sample with only the PAC standard in water. HPLC Analyses. Analytical HPL chromatography was carried out on a HP instrument consisting of an auto-injection unit, a HP Series 1050 Pumping System and a HP Series 1050 wavelength detector (adjusted to 210 nm) equipped with a Merck Chromalith Performance RP18e column (100×4.6 mm) and a guard column. The instruments were controlled by AGTOP software (HP Chem Station). The elution was run with acetonitrile (Carlo Erba, HPLC Grade) and acidic water, obtained by acidifying water with 250 µL of conc. sulfuric acid per 500 mL. The flow rate was maintained constant at 1.5 mL min−1. Agradient program with the following ratios of water/acetonitrile was applied: 1 min 75/25, 39 min gradient to 1/99, 20 min gradient to 0/100, 5 min gradient to 75/25 and 75/25 maintained for 20 min until the system was completely re-equilibrated. 15-µL injections (corresponding to ca. 0.4 µg) were made from GC samples after hexane had been evaporated from an 880 µL sample under nitrogen flow to almost dryness and the residue had been diluted with water (600 µL) and acetonitrile (600 µL). 10.3. RESULTS AND DISCUSSION 10.3.1. PAC Sorption. Retention on HPLC Modified Column Material General Remarks. HPLC methods that measure sorption coefficients of PAC on HSs modified column material were derived from bioscreening procedures [18,19] and have proven a powerful, precise and reproducible tool [20] for determining sorption parameters, supplementing a set of other analytical methods like dialysis [16] or SPME [21]. It was the scope of our experiments to develop and validate this novel HPLC method for further application in environmental monitoring and to obtain a better understanding of sorption processes. To that end, the sorption potential of HS material from soil and water was analyzed with respect to a series of selected PAC compounds and the sorption properties were characterized under various

92

| CHAPTER 10: INTERACTIONS AND CONVERSIONS OF POLYCYCLIC AROMATIC COMPOUNDS

Figure 10.1 Principle of the determination of capacity coefficients on modified HPLC columns

Figure 10.2 Extrapolation of log k’ to a methanol content of 0%—column material modified with soil HSs

conditions (Figure 10.1). The results not only permit estimation of the PAC remobilization potential, but also identify thermodynamic parameters of the sorption process and mechanisms of sorption. Preparation of HPLC Column Material. The preparation of the column material involved covalent bonding of HSs to amino modified silica, Eq. 10.1–3. (10.1) (10.2) (10.3) In the first step, silanol groups of the silica were modified with 3-aminopropyl residues, which in the second step were coupled with carbonyl functions in the HSs matrix. In the third step, unreacted amino functionalities were saturated with formaldehyde. This third step served to eliminate positive charges caused by ammonium cations. In this way potential alterations of the retention times can be avoided. However, comparison of retention times of test substances on column material after (10.2) and after (10.3) revealed negligible differences. A blank column was treated in the same manner, leaving out step 2. The PAC-capacity Coefficient k’ and its Correlation with the Sorption Coefficient KOC. The HPLC-capacity coefficient k’=(tr−t0)/t0 represents a relative measure of the sorption coefficient KOC. These parameters are linked in Eq. 10.4, (10.4) where k’HA is the estimated capacity coefficient of a HS modified column, k’blank is the estimated capacity coefficient of the blank column, v0 is the void volume of the column (flow×time of retention). cv is the total column volume (diameter×length), %OCHA is the wt% of organic carbon in the column material, ′ is the density of the column material, tr is the retention time of the compound and t0 is the retention time of the void volume The capacity coefficient k’ cannot be determined directly since the application of pure water would result in retention times too long for practical analyses and also in peak broadening. Therefore, methanol was added to the eluent in different proportions and afterwards the capacity coefficients were extrapolated to a water content of 100% so as to refer the capacity coefficients to only-water-based media conditions (Figure 10.2). Plotting the logarithms of the sorption coefficients KOC for bound HSs material from aqueous sources and from soil (Figure 10.3) revealed some important findings. N-PAC (N ring-substituted PAC) have sorption properties different from those of PAHs. A linear correlation results only for one group. The reason for that may be a different sorption mechanism. Sorption of N-PAC may chiefly occur through opposite charges on the N-PAC and the HS matrix carboxylic groups. For example, a different content of carboxylic acid functionalities in the aqueous and soil HS would result in different sorption coefficients. In relative terms, N-PAC have a

MATERIALS AND METHODS |

93

Figure 10.3 Correlation of sorption coefficients on HSs from different sources

Figure 10.5 Logarithms of capacity coefficients k’ plotted against 1/T

higher affin ity for soil HSs, whereas PAH show stronger sorption by aqueous HSs. Speaking in absolute terms, both N-PAC and PAH bind more strongly to soil HSs than to aqueous HS. Another observation is that PAH sorption to HSs generally increases with increasing molecular size (Figure 10.4). Similar observations have been made in precipitation batch experiments with linear and angular PAH [22], where a correlation was found between the decrease of ionization energies of linear PAH and their enhanced sorption. These findings were explained by formation of charge-transfer complexes. However, for certain types of HSs that were assumed to be in an earlier stage of humification, sorption properties of angular PAH showed abnormal behavior with respect to their ionization energies. Pyrene, having an ionization energy similar to that of anthracene, showed diminished sorption. This particular phenomenon was traced back to steric differences, leading to the conclusion that sorption of PAC occurs at least partly by means of pockets within the HS matrix. An angular PAC has higher steric demands than linear PAC. Thus, its sorption is partially hampered. In our investigation a corresponding observation was made: anthracene and phenanthrene have the same molar mass (178 Da). Nevertheless, the sorption of phenanthrene is considerably reduced with respect to anthracene. Also in this case, steric or stereoelectronic (electron density distributions/orbital symmetry) parameters seem to affect sorption to HSs. Impact of Temperature on PAC Sorption Coefficients—Thermodynamic Constants (Enthalpy and Entropy) in Sorption of PAC to HS. Temperature is a crucial parameter for the remobilizability of a contaminant. Varying the temperature resulted in a drastic effect on the logarithm of the capacity coefficient. Elevating the temperature from 15 to 30°C caused the sorption coefficient to decrease by about 50%. HPLC measurements of PAC sorption coefficients on HS-modified solid phases proved to be a straightforward method to determine thermodynamic properties of the sorption process. This method is useful with other HSs modified column materials [23]. ′ H and ′ S were calculated from a plot of In k’ versus 1/T (Figure 10.5) for HSs from soil. There is thermodynamic evidence that PAH sorption to the organic material investigated is not only controlled by molecular size and/or ionization energy, but also by steric/stereoelectronic parameters. Figures 10.6 and 10.7 show plots of ′ H versus ′ S for HSs isolated from soil. Whereas naphthalene (KOC=3.11), quinoline (3.04) and isoquinoline (2.90) have about the same sorption coefficient, the sorption enthalpy and entropy of naphthalene differs significantly despite steric similarities. Thus, the N-substituent must be involved in the sorption of the two N-PAC. Obviously, binding of the N-atom results in a large gain in enthalpic stability, but

94

| CHAPTER 10: INTERACTIONS AND CONVERSIONS OF POLYCYCLIC AROMATIC COMPOUNDS

Figure 10.6 Enthalpy change plotted against entropy change for soil HS—N-PAC

Figure 10.7 Enthalpy change plotted against entropy change for soil HSs— PAH

Figure 10.8 Variation of sorption (measured as k’) of some selected PAC in dependence of the phosphate concentration, column material modified with aqueous HS; PAHs with k’ below FL are not shown in this illustration

there also is a similarly large loss in entropic stability, such that degrees of rotation in the sorbed state become hindered (Figure 10.6). Further comparisons in the group of N-PAC show that the more sterically hindered the access to the N-atom is, the smaller is the gain of enthalpic stability and the smaller is the loss in entropic stability (lower degree of orientation). Both PAH and N-PAC show tendencies towards higher enthalpic stabilization with an increasing number of aromatic rings (Figure 10.7). However, steric effects superimpose this effect, when, for example, one molecule is broader than another. Typically in such a case, degrees of rotation are less hindered because the molecule is not forced into narrow pockets of the HS matrix. A molecule with a ‘slim’ design can enter into these pockets, gaining a lot of enthalpic stability but losing entropic

MATERIALS AND METHODS |

95

Figure 10.9 Variation of sorption (measured as k’) of some selected PAC in dependence of the phosphate concentration, column material modified with soil HS; PAHs with k’ below 5B are not shown in this illustration nor BA as it has k’=6000

stability at the same time. KOC values calculated indirectly from thermodynamic investigations agreed well with directly obtained KOC data. Impact of the Ion Concentration on PAC Capacity Coefficients. Not only variations of the temperature, but also changes of ion concentration may affect sorption of contaminants to HS material. Investigations have been carried out previously with other HS material [20]. A variation of the ion concentration in that case was accomplished by adding extra salt to the buffer solution, whereas in this example the concentration of the buffer itself was varied. Figures 10.8 and 10.9 show that the sorption of PAC remained practically unaffected over a broad range of ionic concentration, down to about 0.05 M phosphate buffer concentration, depending on the PAC. Below that level, the sorption tends to increase, passing in some cases through a maximum at a phosphate concentration of 0.01 M. The sorption properties of certain PAC on aqueous HSs differ significantly from others, which down to a concentration of 0.005 M show a steady increase of the sorption coefficient. Possibly, the sorption coefficients also would pass through a maximum for a lower concentration of the buffer. This topic has not been investigated because of possible interferences caused by pH variations at low buffer concentrations and the risk of observing artifacts. 10.3.2. PAC Sorption. Exhaustive SPME General Remarks. In order to determine the freely available concentration of a contaminant, SPME is generally carried out under equilibrium conditions such that the extraction does not disturb sorption equilibria [21]. For that purpose, the sample volume has to be much larger than the volume of the fiber film so that only a very small quantity of contaminant is transferred to the fiber. Frequently, however, equilibria in nature become disturbed so that more contaminant is released than is freely available. This results in a depletion of the sample. On a laboratory scale, depletion of a sample is usually achieved by liquid-liquid extractions (LLE). Yet, SPME also may serve to determine the totally extractable amount of a contaminant. Applying small sample volumes and extracting the same sample several times results in a depletion of the sample as well. Such a strategy of exhaustive extraction was applied in this work. Extrapolation of the extracted amounts (GC response) of a few extractions to an infinite number of extractions gave the totally extractable amount (Figure 10.10). The results for each sample were consecutively referenced to a blank sample of contaminant in water so that the amount of contaminant totally extractable is presented as a percentage of the total amount of contaminant. The procedure described requires only small amounts of a sample and helps to avoid excessive use of solvent necessary for LLE procedures. After equilibration of the fiber with the matrix by repetitive extractions and desorptions, reproducibility is also obtained for samples rich in organic material. Exhaustive SPME in Sorption Kinetics. After validation of the method, exhaustive SPME served to monitor the kinetics of the sorption of anthracene to aqueous HSs [15]. These investigations were aimed at determining the time required to reach the saturation point, that is, when the totally extractable amount of anthracene does not diminish any more. In return, this information gave a rough idea for investigations described in Section 10.3.3, where the effect of pH variations on anthracene remobilizability was the centre of interest. This required a system with stable concentrations, including those concentrations that are not freely available but are, however, extractable. It was found that after a period of about 15 d (Figure 10.11), the anthracene concentration remained almost constant and thus the system was sufficiently stabilized.

96

| CHAPTER 10: INTERACTIONS AND CONVERSIONS OF POLYCYCLIC AROMATIC COMPOUNDS

Figure 10.10 Exhaustive SPME— extrapolating to infinity. Integration of the area under the curve and dividing the result by the integration area of a reference sample gives the percentage of the totally extractable amount

Figure 10.11 Exhaustive SPME—kinetics of the sorption of anthracene by aqueous HSs; solid line: exhaustive SPME (integration after extrapolation), dotted line: only the first extraction, both referenced to a blank sample

10.3.3. PAC Sorption. Application of LLE to Determine pH Dependence of the Totally Extractable Amount General Remarks. In order to investigate the pH dependence of sorption, the totally extractable amount was not determined by SPME but by LLE because the stability of SPME fibres under extreme pH conditions is reported to be problematic by the manufacturer. The contaminant solutions were allowed to equilibrate over a period of 15 days which (according to the results of section 10.3.2) is adequate to guarantee stabilization of the system. To exclude the possibility that exposure of PAC to acidic or basic media would provoke a diminution of the relative concentrations by itself, additional control samples at pH 1 and 12 were run without the HS matrix. Experimental Findings. Conventional LLE with subsequent gas chromatographic analysis revealed significant variations of the totally extractable amount for certain PAH (Figure 10.12). The investigated PACs were classified into four different groups (as indicated in Figure 10.12 by differently formatted lines). So far, no correlations were found between PAC structural properties and the pH dependence of their sorption behavior. pH-dependent risk assessment of contaminated sites certainly will require indi vidual tests for each site, but our results show that there are strong variations of the totally extractable amount of PAC possible. Thus, changing environmental conditions, for example by removing a layer of soil in a contaminated site, may have an impact on the remobilization of the contaminant underneath. Further effort will have to be made to understand the fate of long term bound residues, which may behave differently from short term sorbed contaminants. 10.3.4. PAC Conversion. Impact of Iron(III) General Remarks. There is evidence in the literature [24] that radical like Fe(III) induce radical and/or electrocyclic reactions between anthracene and quinoid systems. Moreover, it has been reported [25] that silver(I) can convert polyphenolic substructures into quinoid systems, which readily react with olefins in [4+2] cycloadditions. A series of further reactions [26]

MATERIALS AND METHODS |

97

Figure 10.12 pH dependence of the totally extractable amount of some selected PAC, classified into groups with similar behavior (as shown by differently formatted curves); all measurements were referenced to a blank sample. Key; carbazol (CA); flouranthene (FA); pyrene (PY); benzo[b]flouranthene (BB); benz[a]anthracene (BN); benzo[k]flouranthene (BK); benzo[a]pyrene (BP); chrysene (CR); indeno [l,2,3-cd]pyrene (IP); dibenzoanthracene (DA); and benzo[g,h,i]perylene (BG)

Figure 10.13 (top) Fe(III) mediated reactions of benzo[a]anthracene, benzo[a]pyrene, benzo[g,h,i] perylene

Figure 10.14 (middle) Fe(III) mediated reactions of anthracene and phenanthrene

showed that [4+2] cycloadditions between anthracene and quinoid systems are also possible in the presence of water with catalysis by Fe(III) or compounds with more Lewis acid activity like the water stable tris-(pentafluorphenyl)borane. Since HSs are assumed to contain lignoid substructures, they are rich in polyphenols and should thus—after oxidation—be able to furnish reactive quinone units for [4+2] cycloadditions. In this manner, anthracene and other PAC may principally be linked covalently to the HS matrix. The addition of radicals may facilitate the process [24]. As the tendency to form radicals or to undergo electrocyclic reactions is considered to increase with an increase of the number of aromatic rings, HSs might assist covalent binding especially of large PAC and help in this way to transform them into insoluble and harmless co-polymeric products. Experimental Findings. Reactions carried out with iron(III) chloride in an aqueous solution of PAC showed that there is a diminution of PAC concentrations even in the absence of HS material (Figure 10.13–10.15). Depending on its concentration in water and the pH of the solution, Fe(III) forms different complexes (Figure 10.16). In our experiments, the medium Fe(III) concentration of 0.025 M proved most effective in diminishing PAC concentrations,

98

| CHAPTER 10: INTERACTIONS AND CONVERSIONS OF POLYCYCLIC AROMATIC COMPOUNDS

Figure 10.15 Fe(III) mediated reactions of pyrene and flouranthene

Figure 10.16 Fe(III) me diated reactions of Fe(III) complexes

Figure 10.17 Fe(III) mediated reactions of acenaphtheneand acenaphthylene

perhaps due to the highest concentration of active Fe(III) complexes. The most concentrated Fe(III) solution (0.25 M) was less effective because its high concentration obviously decreased the pH such that the equilibria were shifted to inactive forms. The medium Fe(III) concentration was found to be capable of almost completely removing all PAHs with molar mass greater than that of pyrene (Figure 10.13). The mere presence of acid (HCl) did not induce any significant variations of PAC concentrations. Apparently, a PAC’s capacity to form and stabilize radicals is crucial for these transformations. This finding is also underpinned by the fact that anthracene was highly susceptible to Fe(III), whereas its structural isomer phenanthrene persisted with almost no loss (Figure 10.14). Adding HSs to the reaction mixtures usually resulted in a reduced efficacy of Fe(III). Fe(III) is reduced to Fe(II) by the electron donating groups in the HS matrix. So far no evidence has been found that the HS matrix assists the chemical conversion by being a reaction partner for PAC. Instead, Fe(III) is consumed preferably by HSs and only after a certain progress of the reaction are PAC attacked. Flouranthene and pyrene seem to react with Fe(III) more easily in the presence of HSs (Figure 10.15). However, most probably HS activity only diminishes a higher Fe(III) concentration (0.25 M) to a lower level, which has a higher concentration of active Fe(III) species (Figure 10.16). One exception was found in acenaphthylene, which was preferentially eliminated from aqueous solution in the presence of HSs. This is obviously due to the presence of an isolated (real) double bond in the molecule (Figure 10.17), which accounts for its difference from all other PAC investigated and also from the otherwise very similar acenaphthene. The N-PAC carbazole (sterically similar to fluorene) shows strong sorption to HSs. Yet, addition of Fe(III) obviously favors the release of sorbed carbazole (Figure 10.18). That finding could either be due to blocking of sorption sites or their conversion into less sorptive functional groups.

MATERIALS AND METHODS |

99

Figure 10.18 Fe(III) mediated reactions of flourene and carbazole

HPLC Control of the Product Mixture. No new compounds (with respect to an untreated reference sample) were found in the PAC samples pre-treated with 0.025 M Fe(III) except for two, which were assigned to anthraquinone (5.54 min) and benz [a]anthraquinone (13.23 min). Signals of benz[a]anthracene, chrysene, benzo[b]flouranthene, benzo[k]flouranthene, benzo[a] pyrene, indenopyrene, dibenz[a]anthracene, benzo[g,h,i]perylene at retention times of 23.93, 26.36, 26.78, 27.26, 28.88, 29.84 min (no individual assignment due superimposed peaks) diminished completely, whereas signals at 12.39, 15.26, 17.66, 18.70, 20.88 and 21.48 min persisted, accounting for naphthalene, acenaphthylene, acenaphthene, flourene, carbazole, flouranthene and pyrene (superimposed peaks). The intensity of a peak at 19.44 min was reduced, thus probably being originally a combined signal of anthracene and phenanthrene. The sample treated with 0.25 M Fe(III) in the presence of HSs showed one signal less at 15.26 min, which belongs to acenaphthylene. 10.4. CONCLUSIONS HPLC experiments on HS modified solid phases served to determine thermodynamic sorption parameters and helped us to understand sorption mechanisms of PACs on HS material. Moreover, these investigations unraveled the significant impact of environmental parameters like temperature and ion concentration on sorption. pH variations affected sorption significantly as shown by LLE. For risk assessment of contaminated sites, variations of especially pH and temperature cannot be neglected. Some further investigations will help us to understand if diminution of concentrations is reversible by changing the pH to the original value. Long term bound PAC residues may behave differently from those affected only by short-term sorption. Further investigations are necessary. Exhaustive SPME was found to be a useful tool for determining the kinetics of anthracene sorption to HSs. A primary equilibrium state is reached after a period of about two weeks. It has to be assumed that the sorption process (thus, diminution of extractable concentrations) continues even after that period but with a significantly reduced rate. Fe(III) proved to be a highly potent agent for reducing PAH concentrations, especially for high molar mass PAH. HPLC investigation of the reaction product showed no soluble products that could be chromatographed. Only anthracene and benz[a] anthracene were transformed into their corresponding quinones. We assume that the other compounds end up as polymers of diverse structure as a result of radical recombination, that is, processes resembling natural paths of humification. Application of Fe(III) compounds in remediation of contaminated sites requires experiments on the pilot scale. Remaining Fe(III) concentration after a possible remediation could be removed quantitatively by adding an excess of HSs with subsequent precipitation of both Fe(III) ions and HSs. Fe(III)-mediated conversion of PAC can be a useful tool in supplementing existing remediation procedures based on microbial activity. Whereas the latter preferentially attacks low molar masses aromatic compounds, Fe(III) preferentially converts high molecular weight PAC. ACKNOWLEDGEMENTS Useful and fruitful help is kindly acknowledged from Stiftung Industrieforschung (Köln, Germany), BREGAU Institutes (Bremen, Germany), Gesellschaft für Angewandte Geologie und Biologie—Rainer Hartmann (Göttingen, Germany), Centre of Biological Processes in Contaminated Soil and Sediment under the Danish Environmental Research Program (Århus, Denmark), Sartorius AG (Göttingen, Germany), A.P.Guimarães, colleagues from the CRA Montecatini (especially D.Cam), the research groups of Prof. W.Thiemann and Prof. D.Leibfritz (University of Bremen, Germany), Wasserwirtschaftsamt

100

| CHAPTER 10: INTERACTIONS AND CONVERSIONS OF POLYCYCLIC AROMATIC COMPOUNDS

(Bremen, Germany), Niedersächsisches Landesamt Landesuntersuchungsanstalt (Bremen, Germany).

für

Ökologie

(Hannover,

Germany)

and

Bremer

REFERENCES 1. 2. 3. 4. 5. 6.

7. 8. 9.

10. 11. 12.

13.

14. 15.

16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

Ziechmann W. Humic substances. Mannheim: Wissenschaftsverlag, 1994. Means JG, Wood SG, Hassett JJ, Banwart WL. Sorption of polynuclear aromatic hydrocarbons by sediments and soils. Environ. Sci. Technol., 1980; 14: 1524–1528. Hassett JJ, Banwart BL. The sorption of nonpolar organics by soils and sediments. In: Sawhney BL, Brown K eds. Reactions and movement of organic chemicals in soil. Madison, Wisconsin: SSSA and ASA, 1989:31–44. Eschenbach A, Wienberg R, Mahro B. Einsatz von Kompost und Rindenmulch bei der biologischen Altlastensanierung PAKkontaminierter Bodenmaterialien. Mitteilungen der Deutschen Bodenkundlichen Gesellschaft, 1997; 83:279–282. Eschenbach A, Wienberg R, Mahro B. Bildung und Langzeitstabilität von nichtextrahierbaren PAK-Rückständen im Boden. Altlastenspektrum, 1997; 5:1–7. Mahro B. Untersuchung von Möglichkeiten zur gezielten Stimulation der biogenen Mineralisierung und Humifizierung von PAK in Böden. In: Verband deutscher Biologen c/o Froese W eds. Symposium “Umweltbiotechnologie” vom 2.-3. März 1994 in Bochum, Köln: Comwork, 1994:60–68. Cerniglia CE, Heitkamp MA. Microbial metabolism of polycyclic aromatic hydrocarbons (PAH) in the aquatic environment. In: Varanasi, U. ed. Metabolism of polycyclic aromatic hydrocarbons in the aquatic environment. Boca Raton, FL: CRC, 1989:41–68. Dhawale S-W, Dhawale S-S, Dean-Ross D. Degradation of phenanthrene by Phanerochaete chrysosporium occurs under lignolytic as well as nonlignolytic conditions. Appl. Environ. Microbiol., 1992; 58:3000–3006. Fritzsche W, Günther TH, Hofrichter M, Sack U. Metabolismus von polycyclischen aromatischen Kohlenwasserstoffen durch Pilze verschiedener ökologischer Gruppen. In: Weigert B ed. Biologischer abbau von polycyclischen aromatischen kohlenwasserstoffen, schriftenreihe biologische abwasserreinigung 4. Berlin: SFB 193, TU Berlin, 1994:167–182. Schulz-Berendt. Biologische bodensanierung. Spektrum der Wissenschaft, 1993; 10:96. Georgi A, Lebelt I. Sorptionsuntersuchungen an festen und gelösten humin stoffhaltigen Matrizes. In: Stegmann R ed. Neue techniken der bodenreinigung, Hamburger berichte abfallwirtschaft TUHH, Bonn: Economica Verlag, Volume 10,1996:87–99. Sinder C, Mann V, Pfeifer F, Klein J. Bildung von huminstoffen im zuge des mikrobiellen PAK-abbaus: Einfluß der bodenmatrix und ökotoxikologische bewertung. In: Stegmann R ed. Neue Techniken der Bodenreinigung, Hamburger Berichte Abfallwirtschaft TUHH, Bd. 10, Economica Verlag, Bonn, 1996: S. 229–240. Mahro B. Vergleichende untersuchungen zur wirkung der zugabe biologisch aktiver supplemente auf den mikrobiellen abbau polycyclischer aromatischer kohlenwasserstoffe (PAK) im Boden. Harburg. Eigendruck: Habilitationsschrift at the Technische Universität Hamburg, 1995. Mahro B, Kästner M. Der mikrobielle abbau polycyclischer aromatischer kohlenwasserstoffe (PAK) in Böden und sedimenten: Mineralisierung, metabolitenbildung und entstehung gebundener rückstände. Bioengineer., 1993; 9: 50–58. Ziechmann W, Hübner M, Jonassen KEN, Batsberg W, Nielsen T, Hahner S, Hansen PE, Gudmundson AL. Humic substances and humification. In: Ghabbour EA, Davies G eds. Humic substances—versatile components of plants, soils and water. Cambridge: Royal Society of Chemistry, 2000:9–20. Nielsen T, Siigur K, Helweg C, Jørgensen O, Hansen PE, Kirso U. Environ. Sci. Technol., 1997; 31:1102–1108. Szabó G, Bullmann RA. Comparison of adsorption coefficients (KOC) for soils and HPLC retention factors of aromatic hydrocarbons using a chemically immobilized humic acid column in RP-HPLC. J. Liq. Chromatogr., 1994; 17: 2593–2604. Suleiman AA, Villarta RL, Guilbault GG. Flow injection analysis of glucose by fiber optic chemiluminescence measurement. Anal. Lett., 1993; 26:1493–1503. Andres RT, Narayanaswamy R. Effect of the coupling reagent on the metal inhibition of immobilized urease in an optical biosensor. Analyst, 1995; 120: 1549–1554. Kollist-Siigur K, Nielsen T, Grøn C, C, Hansen PE, Helweg C, Jonassen KEN, Jør-gensen O, Kirso U. Sorption of polycyclic aromatic hydrocarbons to humic and fulvic acid HPLC column materials. J. Environ. Qual., 2001; 30:526–537. Pawliszyn J. Solid phase microextraction: Theory and practice. Weinheim: Wiley-VCH, 1997. Ziechmann W. Huminstoffe. Weinheim, Germany: Verlag Chemie, 1st Edn, 1980:276–283. Nielsen T, Jonassen KEN. unpublished results. Fillion H, Moeini L, Aurell-Piquer M-J, Luche J-L. Cycloadditions de dielsalder entre l’anthracene et l’anhydride maleique ou la benzoquinone, catalyses chimique et sonochimique. Bull. Soc. Chim. Fr., 1997; 134:375–378. Lee J, Shan Mei H, Snyder JK. Synthesis of miltirone by an ultrasound-promoted cycloaddition. J. Org. Chem., 1990; 55: 4995–4999. Hübner M. unpublished results.

Chapter 11 PYROLYTIC STUDY OF THE BOUND RESIDUES OF 13C-ATRAZINE IN SOIL SIZE FRACTIONS AND SOIL HUMIN Marie-France Dignac,1 Yahya Zegouagh,1 Ludovic Loiseau,2 Gérard Bardoux,1 Enrique Barriuso,2 Sylvie Derenne,3 André Mariotti1 and Claude Largeau3 1Laboratoire

2UMR

de Biogéochime Isotopique, Université Pierre et Marie Curie, 4 place Jussieu, 75252 Paris Cedex 05, France

INRA, INAP-G Environnement et Grandes Cultures, 78850 Thiverval-Grignon, France

3Laboratoire

de Chimie Bioorganique et Organique Physique, UMR CNRS 7573, ENSCP, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France 11.1. INTRODUCTION

Agricultural activities lead to the incorporation of organic compounds like pesticides in soils, where they undergo different processes of biodegradation, mineralization and stabilization. The association of pesticides with soil constituents may influence their transport, bioavailability and toxicity in natural environments. The ability of the soil to retain organic molecules is attributed to adsorption phenomena and chemical reactions with soil surfaces and soil humic substances [1]. The formation of bound pesticide residues can range up to 90% of the original amount applied [2]. Contrary to the other dissipation processes of pesticides in soils (mineralization, leaching, volatilization), the formation of bound residues leads to stabilization of the molecules that increases their persistence in the soil matrix. More information is needed on the binding processes and the chemical forms of the stabilized residues in order to evaluate the environmental risk associated with these residues and the time scale of their stabilization. Atrazine has been widely used during the last decades for the treatment of maize fields. Barriuso and Houot [3], using radiolabeled material, have shown that bound residues of atrazine represent 20 to 60% of the initial concentration applied to a soil. The relative amount of bound residues depends on soil properties and agricultural factors. A rapid mineralization of atrazine was observed for soils that had repeatedly received atrazine during several years, while for soils receiving atrazine for the first time the mineralization rate was lower and the formation of bound residues was higher [3]. The authors proposed the development of microflora able to degrade atrazine due to repeated atrazine application to the soil. The ability of soils to form bound residues with pesticides also depends on the humification level of organic matter (OM). The humin fraction generally represents more than 50% of organic carbon in soil and plays a major role in the retention of organic molecules, leading to relatively stable residues [4]. The formation of covalent bonds between pesticides and humic substances has been suggested to be a major process in the formation of bound residues [2]. A significant part (40–60%) of the bound residues of atrazine were found in the humin fraction, containing the more refractory OM, of four silty, loamy French agricultural soils [5,6]. To understand the processes leading to formation of bound residues, to evaluate their long-term fate in the soil and to establish if they represent a possible source of pollution, it is necessary to know the chemical structures of these residues. In this study, the chemical structures of the bound residues of atrazine were investigated with the use of 13C-labeled atrazine. After incubation with a soil and removal of the extractable forms of the pollutant, soil size fractions were investigated along with the more refractory humic constituents present in humin fractions. This study aimed at acquiring information on the molecular distribution of bound residues and at evaluating an original technique of investigation of bound residues. Bound residues were looked for by pyrolysis/gas chromatography/Combustion-stable isotope ratio mass spectrometry (Py/GC/C-SIRMS) and tentatively characterized by pyrolysis/gas chromatography/mass spectrometry (Py/GC/MS). Parallel analysis by these two spectrometric techniques coupled with pyrolysis was used by Gleixner et al. [7] to follow soil OM dynamics but Py/GC/CSIRMS was never used before for the study of bound residues of pesticides in soils. 11.2. MATERIALS AND METHODS 11.2.1. Materials Four agricultural soils were studied (Table 11.1). MG8 and WG8 are located in Grignon (Ile de France region, France) and cropped with maize and wheat, respectively. MV6 and MS5 are cropped with maize and situated in Versailles (Ile de France

102

| CHAPTER 11: PYROLYTIC STUDIES OF THE BOUND RESIDUES OF 13C-ATRAZINE

Figure 11.1 Protocol used to separate the different soil size andhumin fractions. The studied fractions are indicated in bold character

region, France) and Salinis (Pyrénées region, France), respectively. MG8 and MS5 had received treatment with atrazine for several years, while WG8 and MV6 had never been treated with atrazine before the incubation experiments. Table 11.1 General features of the soil samples studied Soil sample

MG8

WG8

MV6

MS5

Location Agricultural use Corg (g kg−1) pH

Grignon maize with atrazine 17.1 8.1

Grignon wheat without atrazine 19.9 8.0

Versailles maize without atrazine 12.3 6.4

Salinis maize with atrazine 45.5 5.5

11.2.2. Methods Incubation with 13C-Labeled Atrazine. The four soil samples (50 g dry weight) were incubated with 20 mg kg-1soil of 13C ringlabeled atrazine (6-chloro-N2-ethyl-N4-isopropyl-1,3,5-triazine-2,4-diamine, Isotopchim). During incubation, the soil samples spiked with 13C-labeled atrazine were placed in hermetic glass jars, with control of temperature (28°C) and water content (100% of the water holding capacity) for seven months. At the end of the incubation, the extractable atrazine residues were thoroughly extracted with an aqueous solution of 0.01 M CaCl2 followed by five extractions with methanol. Separation of the Soil Size Fractions, Humin and Refractory Humin Fractions. The protocol used was previously described by Loiseau and Barriuso [6] and is presented in Figure 11.1. Briefly, the extracted soil containing the bound residues were fractionated by wet sieving and centrifugation into three fractions: >50 µm, 20–50 µm and <20 µm. The mineral and organic fractions of the >50 µm fraction were separated by flotation. Humin fractions were isolated from the <20 µm fractions by removal of humic and fulvic acids with Na4P2O7/NaOH treatment. The residue (humin and mineral) was then treated with HF 2.5% to dissolve minerals and obtain (humin)after HF. The (humin)after HF was freed of remaining humic and fulvic acids by Na4P2O7/NaOH treatment to obtain a fraction called (humin) HF. This fraction was treated with 2M HCl at reflux conditions for 24 h. The solid residue, called refractory humin, was recovered by centrifugation. On-line Py/GC/MS. Approximately 1–2 mg of sample were loaded in tubular ferromagnetic cylinders inductively heated to their Curie temperature of 650°C in 0.15 s (10 s hold). The pyrolysis unit (GSG Curie-Point Pyrolyser 1040 PSC) was directly coupled to the GC/MS. Pyrolysis products introduced via a split/splitless injector (splitless mode for 1 min) were separated with a Hewlett Packard HP-5890 gas chromatograph equipped with a 50 m fused silica capillary column BPX5 (0.32 mm i.d., film thickness 0.25 µm) with helium as the carrier gas. The GC was coupled to a Hewlett Packard HP-5889 mass spectrometer (electron energy 70 eV, ion source temperature 250°C, scanning from m/z 25 to 400 Da, 1.4 scan s−1). Compounds were identified based on their mass spectra, GC retention times and comparison with library mass spectra. On-line Py/GC/C-SIRMS. The same pyrolysis and GC conditions as above were used for the Py/GC/C-SIRMS analyses. The HP5890 gas chromatograph operated under continuous helium flow was interfaced via a combustion CuO furnace (850° C) and a cryogenic trap (−100°C) to a high-precision SIRMS Micromass Isochrom III mass spectrometer continuously

MATERIALS AND METHODS |

103

Figure 11.2 Py/GC/MS of 13C-labeled atrazine deposited on a 650°C pyrolysis wire.

Figure 11.3 Mass spectra of 13C-labeled atrazine (A) and 13C-labeled OH-simazine (B)

monitoring ion currents at m/z=44, 45 and 46. The 13C/12C isotopic ratio of individual molecules transformed to CO2 are expressed in per mil (‰) relative to the Pee Dee Belemnite standard. 11.3 RESULTS 11.3.1. Preliminary Tests: Fate of Atrazine when Pyrolysed with Humic Substances Pyrolysis is a high temperature degradation process that can lead to major alterations of the primary pyrolysis products generated from OM [8]. These products can lose water molecules and rearrange before entering the GC column. In order to evaluate the products formed during pyrolysis of atrazine, preliminary experiments consisted of pyrolysing at 650°C pure 13C-atrazine and mixtures of atrazine with different soil fractions. These preliminary measurements were aimed at: 1) studying the fate of atrazine during pyrolysis with and without humic substances; 2) determining the retention time of the atrazine-derived products formed by pyrolysis, since these pyrolysis products could also be found in the pyrograms of incubated soils; and 3) acquiring the mass spectra of the atrazine-derived products. Pyrolysis of Pure 13C-atrazine. Atrazine labeled with 13C on the three C atoms of the triazine ring was analyzed by Py/GC/ MS. A single peak is observed (Figure 11.2), whose mass spectrum corresponds to labeled atrazine (m/z 218, Figure 11.3A). The mass spectrum of 13C-atrazine can be recognized from the molecular ion (218), a fragment characteristic of a labeled triazine ring (m/z 70) and a fragment originating from the N-isopropyl substituent (m/z 58). Pyrolysis of Atrazine with Soil Fractions. The reaction of atrazine with pyrolysis products originating from fresh OM of soils (>50 µm fractions) and more humified fractions (humin fractions separated from the <20 µm fractions) was studied. The MS5 and MG8 soils were chosen for these preliminary experiments because they have a high OM content, especially for MS5, and their pyrolysis products are thus more likely to form secondary products by reaction with atrazine. Soil fractions were

104

| CHAPTER 11: PYROLYTIC STUDIES OF THE BOUND RESIDUES OF 13C-ATRAZINE

Figure 11.4 Py/GC/C-SIRMS chromatograms of fractions of the MS5 soil spiked with 13C-labeled atrazine and immediately pyrolysed. The i) traces are the ratios of m/z=45 to m/z=44 (multiplied by 100) and the ii) traces represent total CO2. (A) >50 µm fraction; (B) Humin; (C) (Humin)after HF; (D) (Humin)HF; (E) Refractory humin. The numbers on peaks correspond to enriched- pyrolysis products, numbered according to increasing retention times.

Figure 11.5 Py/GC/C-SIRMS chromatograms of fractions of the MG8 soil spiked with 13C-labeled atrazine and immediately pyrolysed. The i) traces are the ratio of m/z=45 to m/z=44 (multiplied by 100) and the ii) traces represent total CO2. (A) >50 µm fraction; (B) Humin; (C) (Humin)after HF; (D) (Humin)HF; (E) Refractory humin. Numbering is te same as for Figure 11.4.

spiked after separation with a dilute solution of 13C-labeled atrazine and immediately analyzed with Py/GC/C-SIRMS to determine the retention times of the products formed from atrazine. In Figures 11.4 and 11.5, the enriched peaks are numbered according to their retention times. Peaks 14 and 15 are present in almost all the spiked fractions of soil. For these two peaks, the enriched signal observed on the 13C/12C trace can be related to a clearly observed peak in the total CO2 trace. This is not the case for the other enriched peaks: they can be observed on the 13C/12C trace because they are largely enriched compared to organic matter with natural 13C abundance, but their concentration in the pyrogram is so small that no peak can be observed on the total CO2 trace. The same spiked fractions as above were analyzed by Py/GC/MS to search for the chemical structures of the labeled products observed via Py/GC/C-SIRMS. Only compounds 14 and 15 that also display a “real” peak on the total CO2 trace (Py/ GC/C-SIRMS) and thus also on the Total Ion Current (TIC) trace of the GC/MS pyrograms could be identified. For the other peaks, only observed because of their high 13C enrichment, structure determination with Py/GC/MS was not possible. The chemical structures of compounds 14 and 15 were recognized with Py/GC/MS as 13C-labeled OH-simazine (m/z 214) and 13Clabeled atrazine (m/z 218), respectively (Figure 11.3). Their mass spectra both display a fragment at m/z 70, characteristic for

MATERIALS AND METHODS |

105

Figure 11.6 Py/GC/C-SIRMS chromatograms of size fractions of the MS5 soil incubated with 13C-labeled atrazine and extracted to remove the extractable forms of atrazine-derived residues. The i) traces are the ratio of m/z=45 to m/z=44 (multiplied by 100) and the ii) traces represent total CO2. (A) >50 µm fraction; (B) 20–50 µm fraction; (C) < 20 µm fraction. Same numbering as for Figure 11.4.

Figure 11.7 Py/GC/C-SIRMS chromatograms of >50 µm size fractions of the MV6, WG8 and MG8 soils incubated with 13C-labeled atrazine and extracted to remove the extractable forms of atrazine-derived residues. The i) traces are the ratio of m/z=45 to m/z=44 (multiplied by 100) and the ii) traces represent the total CO2. (A) MV6; (B) WG8; (C) MG8. Same numbering as for Figure 11.4.

the presence of the triazine ring (Figure 11.3). The presence of atrazine and OH-simazine was confirmed in the pyrograms of most of the spiked soil fractions by monitoring this characteristic mass fragment. 11.3.2. Pyrolysis of Incubated Soils The bound residues of atrazine were searched for in the size fractions of the four incubated soils (>50 µm, 20–50 µm, <20 µm), and in the different humin fractions (humin, (humin)after HF, (humin)HF and refractory humin). Only the pyrograms displaying 13C-enriched peaks are shown in Figures 11.6 and 11.7. Enriched products were detected in the <20 µm fraction of MS5, but not in the corresponding fractions of the three other soils. For MS5 soil, it was not possible to detect any enriched product after chemically fractionating the <20 µm fraction. Enriched molecules originating from 13C-atrazine were evident via Py/GC/C-SIRMS in some fractions of the incubated soils (Figures 11.6 and 11.7). Several enriched peaks were detected in the three size fractions of MS5 and in the larger size fractions of the three other soils. These peaks were similarly numbered by comparing the retention times of the enriched compounds in the incubated soils with those of the spiked fractions of soils (Figures 11.4 and 11.5). The presence of peak 15, identified as atrazine in the spiked fractions, suggests the presence of bound 13C-atrazine in some of the fractions of the incubated soils, along with other 13C enriched atrazine-derived compounds. The larger size fractions (>50 µm) of MV6 and WG8 soils (Figures 11.7A and 11.7B) display numerous enriched compounds. In contrast, the larger fractions of MS5 (Figure 11.6A) and MG8 (Figure 11.7C) display only a few enriched compounds apart from compound 15 potentially corresponding to atrazine. In the 20–50 µm fraction of the MS5 soil

106

| CHAPTER 11: PYROLYTIC STUDIES OF THE BOUND RESIDUES OF 13C-ATRAZINE

(Figure 11.6B), only a very weak enriched peak of compound 15 was observed, while in the <20 µm fraction (Figure 11.6C), compound 15 was not present. However, an enriched compound eluting earlier (compound 13) is observed. The same fractions of the incubated soils as above were analyzed via Py/GC/MS. The same problem of detection level was encountered as for the spiked fractions. The detection level of 13C enrichment is low with GC/C-SIRMS, but the enriched peaks do not necessarily correspond to detectable peaks in GC/MS. The bound residues in the incubated soils could thus be observed via Py/GC/C-SIRMS because of their 13C enrichment, but they do not correspond to clearly detected peaks either on the total CO2 trace of the GC/C-SIRMS pyrogram or on the TIC trace of the GC/MS pyrogram. By monitoring its characteristic mass fragments (m/z 58, 70 and 218), the presence of 13C-atrazine could be confirmed in the spiked fractions, but these fragments were not clearly detected in any Py/GC/MS traces of the incubated and extracted soils. 11.4. DISCUSSION 11.4.1. Formation of Secondary Products during Pyrolysis The presence of a single peak in the pyrogram of 13C-labeled atrazine shows the purity of the compound. It also appears that under these conditions atrazine undergoes simple thermovaporisation rather than pyrolytic degradation. The other enriched products observed in the pyrograms of the spiked fractions of MS5 and MG8 soils, like product 14 identified as 13C-labeled OH-simazine, are therefore formed by secondary reactions of atrazine. These products thus correspond to artifacts of the analytical method. The formation of such secondary products occurs in the pyrolysis unit by reaction of atrazine with organic compounds generated upon pyrolysis of the soil OM and/or with the mineral matrix. The MS5 and MG8 soils are both cropped with maize and treated each year with atrazine. The former soil from Pyrénées is an acidic soil with a high OM content, whereas MG8 from Grignon has a substantially lower OM content and a higher pH. Comparison of the chromatograms obtained from different spiked fractions of these soils (Figures 11.4 and 11.5) shows a higher relative abundance of the secondary products with low retention times for the MG8 fractions. These compounds observed at the beginning of the chromatograms correspond to low molar mass and/or low polarity derivatives of atrazine. In fact, in the case of the Humin and (Humin)after HF fractions, only such compounds were detected for the MG8 soil (Figures 11.5B and 11.5C). For both soils, the chromatograms of the refractory humin fractions show only a few enriched peaks with low intensities, including the atrazine peak (Figures 11.4E and 11.5E). In contrast, a wide range of enriched compounds is detected for the >50 µm fractions (Figures 11.4A and 11.5A) and (Humin)HF (Figures 11.4D and 11.5D) from the two soils. Removal of the mineral matrix by HF treatment does not cause extensive changes in the secondary reactions of pyrolysis products of the OM with atrazine. Similar distributions of the enriched atrazine-derived products are thus observed for the spiked Humin and (Humin)after HF fractions from the MS5 soil (Figures 11.4B and 11.4C) and, to a lesser extent, from the MG8 soil (Figures 11.5B and 11.5C). Accordingly, the above mentioned differences observed for the secondary products formed with the spiked >50 µm and the spiked humin fractions from the two soils should mostly be related to differences in OM abundance and/or nature rather than differences in mineral components. For a given soil, the large differences sometimes observed between the chromatogram of one spiked fraction and that of the next one in a chemical fractionation process should reflect the changes induced by the chemical treatment in the composition of the OM and hence its influence on the secondary reactions that occur upon pyrolysis. 11.4.2. Bound Residues in Incubated Soils Various enriched compounds can be observed in the pyrograms of some size fractions of the incubated soils (Figures 11.6 and 11.7). These compounds are different from one fraction to another. The main artifactual product (compound 14), identified in the pyrograms of the spiked fractions as 13C OH-simazine, was not present in the pyrograms of the fractions of the incubated soils. This suggests that the products observed in the latter are not artifacts but actually correspond to bound residues released during pyrolysis. The lack of products originating from secondary reactions between atrazine and soil OM, contrary to what took place in the spiked fractions, may be explained by the very low amounts of bound residues of atrazine present in the incubated soils compared to the much larger quantities of atrazine present in the spiked fractions. For the MS5 soil, compound 15 likely corresponds to atrazine according to its retention time. It is present in the larger size fractions (>50 µm and 20–50 µm) but was not found in the more humified fraction (Figure 11.6). This compound was also found in high relative abundance in the >50 µm fractions of the three other soils (Figure 11.7). This suggests that atrazine is the main form of bound residues in these soils.

MATERIALS AND METHODS |

107

The history of the soil influences the formation of bound residues. In the fractions of the soils adapted to atrazine (MS5 and MG8), that have developed a microflora able to degrade this pesticide, only one or two enriched compounds are pre sent apart from atrazine. For the “non adapted” soils atrazine is present in the pyrograms of the fractions of the incubated WG8 and MV6 soils, along with many other enriched bound products. An adaptation of the microflora therefore seems to decrease the number of chemical forms of the bound residues, which are mainly bound atrazine in that case. Barriuso and Houot [3] have shown that the total amount of bound residues formed in adapted soils is lower. Accordingly, adaptation of soil microflora appears to influence both the abundance and the nature of the bound residues of atrazine. The content of OM in the soil does not appear to influence the bound residues formed. Indeed both in the soil with the highest OM content (MS5) and in the soil with a low OM content (MG8), only a few bound residues are observed apart from atrazine. 11.4.3. Evaluation of the Analytical Technique In previous studies [3,5,6], the use of the 14C isotope to follow the bound residues of atrazine has proven to be very useful in determining the distribution of these bound residues in the different soil fractions. To examine the chemical structures of these residues with a chromatographic method, labeling of the atrazine with 13C isotope is required, as in the present work. Since the bound residues are chemically bound to OM [6], pyrolysis was supposed to be a good option to break the corresponding bonds and release the residues for molecular analyses. Analysis of the fractions of the incubated soils with Py/GC/C-SIRMS confirms that the bound enriched molecules can be released upon pyrolysis and several enriched peaks can be observed. Nevertheless, it was not possible to determine the chemical structures of the enriched compounds thus detected. The detection limit of SIRMS is very low because this method specifically detects the 13C isotope of enriched compounds. The 13C-labeled atrazine used to incubate the soils was highly enriched in 13C (99% of the triazine ring). Therefore, the 13C-enriched compounds formed from the bound residues can be detected, even when their peaks on the total CO2 trace of the GC/C-SIRMS analysis are not distinguishable from the noise or from the peaks of the pyrolysis products of soil OM. In contrast, MS was not sensitive enough to allow the detection of extremely low amounts of such compounds, especially when they are highly diluted by large quantities of the pyrolysis products arising from OM present in the soil fractions. Moreover, the enriched compounds detected via Py/GC/C-SIRMS often coeluted with much more intense peaks originating from the pyrolysis of soil OM. Even by monitoring the characteristic mass fragments of atrazine, it was not possible to confirm its presence as a bound residue by MS analysis. Thus, only the GC retention times of the detected enriched compounds could be used for tentative identification. The concentration of labeled atrazine used for incubation in the soil was 20 times the agronomic dose used for field treatments. Even this relatively large value leads to concentrations of labeled bound residues that are not high enough for GC/ MS analysis. The pyrolytic technique is satisfactory for releasing the bound residues, but problems are encountered for detection of the very low concentration of these released bound residues that eluted with much larger amounts of pyrolysis products from soil OM. New analytical techniques need to be developed to study the chemical structure of these bound residues. Such techniques should combine 1) an efficient release of the residues chemically bound to soil OM; 2) an efficient detection of the released products; and 3) a very sensitive identification technique to distinguish the chemical structures of the released residues from those of products originating from soil OM. The two first requirements were achieved in the present work by pyrolysis and SIRMS, respectively, but compound identification still has to be improved. 11.5. CONCLUSIONS 13C-labeled

atrazine was incubated with different soil samples to examine the bound residues of atrazine at a molecular level. This stable isotope labeling allows sensitive detection with SIRMS. Analytical pyrolysis was used to break the chemical bonds between the atrazine residues and the organic matrix of the soil fractions. Preliminary experiments showed the formation of secondary compounds by reaction of atrazine with products released upon pyrolysis of OM of soil fractions. Comparison of fractions from MS5 and MG8 soils spiked with 13C-labeled atrazine and immediately analyzed showed marked differences. These differences depend on the nature and relative abundance of the 13C-enriched compounds formed. Differences were observed both between the successive humic fractions obtained from a given soil by chemical fractionation and between the same type of fraction from the two soils. Among the 13C-enriched products generated upon pyrolysis of the spiked fractions of soils and detected by SIRMS, atrazine and OH-simazine could be identified by parallel Py/GC/MS analysis in almost all the spiked fractions. Analysis by Py/GC/C-SIRMS of size fractions of incubated soils for the bound residues of 13C-labeled atrazine indicated the presence of several enriched compounds with different chemical structures whose distribution changed from one fraction to the next. Comparison of the retention times of the enriched products suggested the presence of atrazine along with other

108

| CHAPTER 11: PYROLYTIC STUDIES OF THE BOUND RESIDUES OF 13C-ATRAZINE

labeled compounds. Atrazine was found to be the main form of bound residues in the studied soils. The content of OM in soil did not appear to influence the number of enriched peaks detected. However, the history of the soil seemed to be important as regards the number of chemical forms of bound residues formed. Adaptation of the microflora decreased the number of bound compounds, which were mainly bound atrazine, compared to “non-adapted” soils. Parallel analysis by on-line Py/GC/C-SIRMS and on-line Py/GC/MS was used for the first time to study bound residues in soil. This work showed the necessity of developing new tools in order to be able to detect and characterize these bound residues. It appears that 1) pyrolysis effectively releases bound residues; 2) the detection limits of the bound residues at a molecular level was clearly enhanced by the use of 13C labeled compounds, which allowed the detection of trace amounts of bound residues released upon pyrolysis; and 3) the analytical technique has to be improved for the GC/MS determination of the chemical structure of products that occur in extremely low amounts and moreover are highly diluted by the pyrolysis products generated from the OM of the soil. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

Hamaker JW, Thomson HB. Adsorption. In: Goring CAJ, Hamaker JW eds. Organic chemicals in the soil environment. New York: Dekker, 1972:49–143. Dec J, Bollag JM. Determination of covalent and non-covalent binding interactions between xenobiotic chemicals and soil. Soil Sci., 1997; 162:858–874. Barriuso E, Houot S. Rapid mineralization of the s-triazine ring of atrazine in soils in relation to soil management. Soil Biol. Biochem., 1996; 28:1341–1348. Rice JA. Humin. Soil Sci., 2001; 166:848–857. Loiseau L, Barriuso E, Zegouagh Y, Largeau C, Mariotti A. Release of the atrazine non-extractable (bound) residues of two soils using degradative techniques. Agronomie, 2000; 20:513–524. Loiseau L, Barriuso E. Characterization of atrazine’s bound (nonextractable) residues using fractionation techniques for soil organic matter. Environ. Sci. Technol., 2002; 36:683–689. Gleixner G, Poirier N, Bol R, Balesdent J. Molecular dynamics of organic matter in a cultivated soil. Org. Geochem., 2002; 33: 357–366. Hatcher PG, Dria K, Kim S, Frazier SW. Modern analytical studies of humic substances. Soil Sci., 2001; 166:770–794.

Chapter 12 PHENANTHRENE SORPTION BY CLAY-HUMIC COMPLEXES Kaijun Wang,1 Elham A.Ghabbour,2 Geoffrey Davies2 and Baoshan Xing1 1Department

of Plant and Soil Science, University of Massachusetts, Amherst, MA 01003, USA

2Chemistry

Department, Northeastern University, Boston, MA 02115, USA

12.1. INTRODUCTION Sorption of hydrophobic organic compounds (HOCs) in soils and geologic materials is of growing concern because of the strong affinity of these compounds for soil colloids and the potential risks associated with their long-term persistence in the environment. It is well accepted that soil organic matter (SOM) is the predominant sorbent of HOCs. The sorption of HOCs by humic substances has been extensively studied, and it is generally believed that partitioning and adsorption-like mechanisms are involved in the sorption process [1–5]. According to the dual-mode model [2,3], sorption in rubbery domains is anticipated to exhibit a linear (partitioning) isotherm contribution, while sorption in glassy domains is expected to represent the nonlinear (Langmuir) contribution. The heterogeneous nature of humic substances (HSs) and their colloidal aggregates give rise to numerous potential binding sites for a wide range of materials of diverse chemistry, such as metal ions, organic pollutants, and biocides used for agricultural purposes. Humic substances can also interact with clay particles, and ultimately enhance the colloidal stability of the clay particles and the structural stability of soils [6,7]. Another consequence of those interactions is to stabilize organic matter in soils as clay-humic complexes [8]. Previous studies have demonstrated that organic matter is likely to be associated with the clay fractions as clay-organic complexes in soil and sedimentary environments [9,10], most commonly in the form of coatings on solid surfaces [11,12]. In this respect, soil organic matter may play very important roles in controlling the fate and transport of organic or inorganic contaminants in natural environments. In surface soil containing as little as 1% organic mat ter, for instance, the surface chemistry was found to be controlled by organic components coated on phyllosillicate clay minerals, and on Al and iron oxides [13]. Humic substances may undergo conformational changes after associating with clay minerals [14,15], and such modification will affect their sorption behavior with HOCs. Therefore, it may be inappropriate to extrapolate the sorption behavior of HS coated on mineral particles from extracted HS alone. So far, a few studies have included sorption of HOCs to constructed clay-humic complexes [16–19]. The results show that the organic carbon normalized sorption coefficient (Koc) values are lower for mineral-bound humic materials compared with humic materials alone. It is proposed that humic acids might adopt a more condensed conformation once they were bound to a mineral surface, resulting in a decreased number of sites for HOCs sorption than for bulk humic acids [19]. In addition, Murphy et al. [17] observed nonlinearity and competitive sorption of three organic compounds by humic-coated minerals implying adsorption rather than partitioning into the surface organic phase. However, Onken and Traina [20] measured the sorption of pyrene and anthracene to three humic acid-mineral, complexes (HA-CaCO3, HA-Calcite, and HA-Na-Montmorillonite) and found that partitioning is the predominant process for the higher foc (fractional organic carbon content) system (>3×10–5), while adsorption and condensation are major processes for the lower foc system (<3×10–5). In recent years, a significant research effort has focused on the sorption of HOCs by either extracted humic substances [5,21,22] or pure minerals [23–25]. Information on sorption on clay-humic complexes is rather limited. Therefore, the objective of this study is to explore the sorption behavior of HOCs by clay-humic complexes at different loading levels of humic acid.

110

| CHAPTER 12: PHENANTHRENE SORPTION BY CLAY-HUMIC COMPLEXES

12.2. MATERIALS AND METHODS 12.2.1. Materials New York humic acid was used in this study. It was extracted from soil at an organic farm in Pine Island, NY and demineralized as described previously [26]. Kaolinite and montmorillonite, two commonly found clay minerals in soils, were used in this study. Kaolinite represents a non-expanding 1:1 layered silicate mineral, while montmorillonite represents an expanding 2:1 layered silicate mineral. They were purchased from Fluka. The carbon contents of the two minerals were measured by the Microanalysis Laboratory at the University of Massachusetts, Amherst, and the results were below the detection limit (<0.05 wt %). Kaolinite was purified as described previously [27] and montmorillonite was treated with 0.5 M CaCl2 solutions, washed with water many times by centrifugation, then freeze dried. 14C labeled phenanthrene, which was selected to represent HOCs, was purchased from Sigma. The unlabeled phenanthrene was purchased from Aldrich. Compounds were used as received without further purification because of their high purity. 12.2.2. Preparation of Clay-humic Complexes Clay-humic complexes were prepared by the method described by Ghabbour et al. [28]. Two solutions of HA in 0.050 M NaOH were prepared. Solution A and B had mass/volume ratios of 1:10 w/v and 1:20 w/v, respectively. Both solutions were adjusted to pH 5.0 with concentrated HCl. Then, freeze dried clay samples were treated with solution A or B at a mass/ volume ratio of 1:1. The suspensions were shaken for 48 h at room temperature. Then the clay-HA suspensions were centrifuged. Finally, the precipitates were washed repeatedly with water until there was no color in the supernatant, then freeze-dried, and stored for subsequent uses. 12.2.3. Sorption Experiments Sorption experiments were conducted using a batch equilibration method in 2-mL screw-cap vials (for clay-humic complexes) or in 8-mL screw-cap vials (for pure humic acid) with Teflon-lined septa. The background solution was 0.01 M CaCl2 with 200 µg/mL HgCl2 as a biocide. The 14C-labeled phenanthrene and its non-radioactive stock solutions were mixed with clay-humic complexes at different solid to solution ratios. The complex concentrations were adjusted to achieve 30–80% sorption of phenanthrene. Each isotherm consisted of 40 concentration points in the range from 0.006 to 0.8 µg/mL, and two blanks were prepared for each initial concentration. The suspensions were shaken for 3 days on hematology mixers to achieve apparent equilibrium. Then, vials were centrifuged at 1000 g for 30 min, and 1 mL supernatant was sampled for liquid scintillation counting (Beckman LS 3801). The amount of compound sorbed by clay-humic complexes and humic acid alone was calculated from the mass difference. The sorption of HOCs by kaolinite and mont-morillonite was performed following the same procedures. 12.2.4. Data Processing Sorption data were fitted using the modified Freundlich equation [29], , where Cs (µg g−1) is the sorbed concentration, K’f is the modified Freundlich coefficient, n is the exponent, and Cr is the ratio of the concentration in the equilibrium solution (Ce) to the supercooled liquid-state solubility of sorbate (Sscl), which can be estimated from its melting point, the aqueous solubility of the crystalline solute and the heat of fusion. Since Ce and Sscl have the same unit, Cr is dimensionless. In this way, K’f has same units as Cs and refers to sorption capacity. Therefore, the magnitude of K’f can be used to compare the sorption behavior of different sorbents. 12.3. RESULTS AND DISCUSSION The total carbon contents of the kaolinite-humic complexes were 0.30% and 0.19% (w/w) for high and low loadings, respectively. Montmorillonite sorbed more than twice as much HA as kaolinite at the same HA to clay ratio. The carbon

RESULTS AND DISCUSSSION |

111

Figure 12.1 Sorption isotherms of phenanthrene on montmorillonite (A) and kaolinite (B) at 20°C.

contents of montmorillonite-humic complexes at high and low loadings were 0.97% and 0.93%. We did not get the two different loading levels for montmorillonite-humic complexes that we expected. Although soil organic matter is the predominant sorbent of HOCs, it is generally accepted that mineral phase sorption is important when the clay to organic matter ratio is sufficiently high (i.e., >10~30) [24]. Figure 12.1 shows sorption isotherms of phenanthrene by Ca2+-saturated kaolinite and montmorillonite. Montmorillonite retained a larger amount of phenanthrene than kaolinite, as indicated by the K’f values, which were 120 and 50 µg/g for montmorillonite and kaolinite, respectively. Nearly linear sorptions were observed for both types of minerals with n values slightly greater than 1. The n value greater than 1 might have resulted from self-adsorption of the solute at high concentrations, leading to an inflection of the sorption isotherm [30]. Similar results have been reported [23,25,31,32]. All the sorption data were fitted well by the modified Freundlich equation and isotherm parameters are listed in Table 12.1. Sorption of phenanthrene on New York HA had a nonlinear isotherm with n equal to 0.82 (Figure 12.2), and its sorption capacity (K’f) is three orders of magnitude higher than those of the clay minerals themselves. The isotherms for kaolinite-humic and montmorillonite-humic complexes are shown in Figures 12.3 and 12.4, respectively. It is apparent that the modified sorption coefficients (K’f) of clay-humic complexes were higher than those of pure clay minerals but their magnitudes were within an order of magnitude (Table 12.1). However, the complexes had much lower K’f values than HA particles alone. This is attributed to the great difference of carbon content between mineral-HA complexes and HA itself. In this case, the organic carbon normalized sorption coefficient (K’oc) must be used for comparison. The results in Table 12.2 reveal that the montmoril-Ionite-humic complexes had a higher K’oc than HA alone, while the kaolinite-humic complexes had lower K’oc values. All the isotherms for clay-humic complexes were Table 12.1 Modified Freundlich isotherm parameters of New York HA and clay-humic complexes Montmorillonite-humic complexes High loading Low loading Kaolinite-humic complexes High loading Low loading New York HA Ca2+-montmorillonite Ca2+-kaolinite

K’f (µg/g)

K’OC (µg/g)

n

R2

541 532

55800 57200

0.921±0.008 0.909±0.009

0.998 0.998

99 68 25100 120 50

33100 35700 49600 – –

0.850±0.010 0.845±0.009 0.820±0.009 1.09±0.022 1.08±0.056

0.996 0.997 0.997 0.991 0.982

nonlinear, the n values ranging from 0.84 to 0.92 (Figures 12.3 and 12.4, Table 12.1). Obviously, the complexes had more linear isotherms than the HA particles alone. This result contradicts our hypothesis that HA may adopt a more condensed configuration once it is adsorbed on minerals, leading to more nonlinear sorption isotherms. Moreover, montmorillonitehumic complexes had higher linearity than kaolinite-humic complex, probably due to their different HA loading levels. For both types of complexes, the slight differences in organic carbon loading contents gave similar n values (Table 12.1). We believe that both HA and minerals contributed to the sorption of phenanthrene in the experiments with mineral-HA complexes and that the higher linearity of isotherms than HA alone might result from the sorption of phenanthrene by mineral.

112

| CHAPTER 12: PHENANTHRENE SORPTION BY CLAY-HUMIC COMPLEXES

Figure 12.2 Sorption isotherms of phenanthrene on New York humic acid particles at 20°C

In order to distinguish the contribution of HA from overall sorption results in complexes, the sorption portion due to the clay minerals should be defined and deducted. Because it is very difficult to determine the percentage of mineral surface Table 12.2 Modified Freundlich isotherm parameters of New York HA and clay-humic complexes after calibration Montmorillonite-humic complexes High loading Low loading Kaolinite-humic complexes High loading Low loading New York HA

K’f (µg/g)

K’OC (µg/g)

n

456 446

47000 48000

0.906±0.008 0.894±0.009

70 44 25100

23300 23200 49600

0.815±0.010 0.807±0.009 0.820±0.009

covered by HA, we assumed that the sorption of phenanthrene by minerals was not affected by the HA coating, and then subtracted this portion from the overall sorption amount. Obviously, this is the maximum adjustment and the sorption of phenanthrene by mineral-bound HA, to some extent, was underestimated. Table 12.2 illustrates the calibrated Freundlich parameters. The K’oc of montmorillonite-humic complexes was slightly lower than that of HA alone, while the K’oc Of kaolinite-humic complexes was only half that of HA alone. Kaolinite-humic complexes had similar n values to those of HA alone, while the n values of montmorillonite-humic complexes were higher than that of HA alone, indicating that HA on the montmorillonite surface might be less condensed (i.e., more extended and flexible) than solid HA alone. It is possible that the HA coated on minerals from an HA solution at pH 5 had a more flexible structure than solid HA particles, which were precipitated at pH 1.5, washed and freeze-dried. Another possibility is that since only a fraction of dissolved HA may be adsorbed by minerals, the sorption behavior of clay-humic complexes would be different from the whole HA particles. More work needs to be done to determine the effect of different loading levels of HA on the sorption capacity and isotherm linearity of clay-humic complexes. 12.4. CONCLUSION Study of the sorption of HOCs by clay-humic complexes is critical for better understanding of the fate and transport of HOCs in soils and sediments. Our results demonstrate that clay-humic complexes have higher sorption capacities than clay minerals alone. All the sorption isotherms for clay-humic complexes were nonlinear, while clay minerals alone exhibited nearly linear sorption of phenanthrene. The sorption of phenanthrene by clay-humic complexes is different from sorption by HA particles alone, which needs further investigation. ACKNOWLEDGEMENTS This work was supported by the US Department of Agriculture, National Research Initiative Competitive Grants Program (2002–35107–12544), and the Federal Hatch Program (Project No. MAS00860).

RESULTS AND DISCUSSSION |

113

Figure 12.3 Sorption isotherms of phenanthrene on montmorillonite coated with New York HA at 0.97% (A) and 0.93% (B) loading levels and 20°C

Figure 12.4 Sorption isotherms of phenanthrene on kaolinite coated with New York HA at 0.30% (A) and 0.19% (B) loading levels

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

Weber WJ Jr, Huang W. A distributed reactivity model for sorption by soils and sediments. 4. Intraparticle heterogeneity and phasedistribution relationships under nonequilibrium conditions. Environ. Sci. Technol., 1996; 30:881–888. Xing B, Pignatello JJ. Dual-mode sorption of low polarity compounds in glassy polyvinychloride and soil organic matter. Environ. Sci. Technol., 1997; 31: 792–799. Xing B, Pignatello JJ. Competitive sorption between 1,3-dichlorobenzene or 2,4-dichlorophenol and natural aromatic acids in soil organic matter. Environ. Sci. Technol., 1998; 32:614–619. Xing B, Chen Z. Spectroscopic evidence for condensed domains in soil organic matter. Soil Sci., 1999; 164:40–47. Salloum MJ, Dudas MJ, McGill WB. Variation of 1-naphthnol sorption with organic matter fractionation: The role of physical conformation. Org. Geochem., 2001; 32:709–719. Day GM, Hart BT, McKelvie ID, Beckett R. Adsorption of natural organic matter onto goethite. Coll. Surf. A: Physicochem. Engineer. Aspects, 1994; 89: 1–13. Jones MN, Bryan ND. Colloidal properties of humic substances. Adv. Coll. Interface Sci., 1998; 78:1–48. Laird DA, Martens DA, Kingery WL. Nature of clay-humic complexes in an agricultural soil: I. Chemical, biochemical, and spectroscopic analysis. Soil Sci. Soc. Am. J., 2001; 65:1413–1418. Schnitzer, M, Khan SU. Humic substances in the environment. New York: Dekker, 1972. Ransom B, Bennett RJ, Baerwald R, Shea K. TEM study of in situ organic matter on continental shelf margins: Occurrence and the monolayer hypothesis. Mar. Geol., 1997; 138:1–9. Maurice PA, Namjesnik-Dejaaovic K. Aggregate structures of sorbed humic substances observed in aqueous solution. Environ. Sci. Technol., 1999; 33: 1538–1541. Mayer LM, Xing B. Organic matter-surface area relationship in acid soils. Soil Sci. Soc. Am. J., 2001; 65:250–258. Bertsch PM, Seaman JC. Characterization of complex mineral assemblages: Implications for contaminant transport and environmental remediation. Proc. Natl. Acad. Sci., 1999; 96:3350–3357. Schevchenko S, Bailey GW. Non-bonded organo-mineral interactions and sorption of organic compounds on soil surfaces: A model approach. J. Mol. Structure, 1998; 422:259–270.

114

| CHAPTER 12: PHENANTHRENE SORPTION BY CLAY-HUMIC COMPLEXES

15.

Specht CH, Kumke MU, Frimmel FH. Characterization of NOM adsorption to clay minerals by size exclusion chromatography. Wat. Res., 2000; 34: 4063–4069. Murphy EM, Zachara JM, Smith SC. Influence of mineral-bound humic substances on the sorption of hydrophobic organic compounds. Environ. Sci. Technol., 1990; 24:1507–1516. Murphy EM, Zachara JM, Smith SC, Phillips JL, Wietsma TW. Interaction of hydrophobic compounds with mineral-bound humic substances. Environ. Sci. Technol., 1994; 28:1291–1299. Laor Y, Farmer WJ, Aochi Y, Strom PF. Phenanthrene binding and sorption to dissolved and to mineral-associated humic acid. Wat. Res., 1998; 32: 1923–1931. Jones KD, Tiller CL. Effect of solution chemistry on the extent of binding of phenanthrene by a soil humic acid: A comparison of dissolved and clay bound humic. Environ. Sci. Technol., 1999; 33:580–587. Onken BM, Traina SJ. The sorption of pyrene and anthracene to humic acid-mineral complexes: Effect of fractional organic carbon content. J. Environ. Qual., 1997; 26:126–132. Chefetz B, Deshmukh AP, Hatcher PG, Guthrie EA. Pyrene sorption by natural organic matter. Environ. Sci. Technol, 2000; 34: 2925–2930. Xing B. Sorption of anthropogenic organic compounds by soil organic matter: A mechanistic consideration. Can. J. Soil Sci., 2001; 81:317–323. Mader BT, Uwe-Goss K, Eisenreich SJ. Sorption of nonionic, hydrophobic organic chemicals to mineral surfaces. Environ. Sci. Technol., 1991; 31: 1079–1086. Boyd SA, Sheng G, Teppen BJ, Johnston CT. Mechanisms for the adsorption of substituted nitrobenzenes by smectite clays. Environ. Sci. Technol., 2001; 35: 4227–4234. Hundal LS, Thompson ML, Laird DA, Carmo AM. Sorption of phenanthrene by reference smectites. Environ. Sci. Technol., 2001; 35:3456–3461. Davies G, Ghabbour EA, Steelink C. Humic acids: Marvelous products of soil chemistry. J. Chem. Educ., 2001; 78:1609–1614. Jackson M. Soil chemical analysis-advanced course; Madison, Wisconsin: University of Wisconsin, 1956. Ghabbour EA, Davies G, O’Donaughy K, Smith TL, Goodwillie ME. Adsorption of a plant and a soil derived humic acid on the common clay kaolinite. In: Davies G, Ghabbour EA eds. Humic substances: Structures, properties and uses. Cambridge: Royal Society of Chemistry, 1998:185–194. Carmo AM, Hundal LS, Thompson ML. Sorption of hydrophobic organic compounds by soil materials: application of unit equivalent Freundlich coefficients. Environ. Sci. Technol, 2000; 34:4363–4369. Schwarzenbach RP, Gschwend PM, Imboden DM. Environmental organic chemistry. 2nd Edn. New Jersey: Wiley, 2002. Schlautman MA, Morgan JJ. Effects of aqueous chemistry on the binding of polycyclic aromatic hydrocarbons by dissolved humic materials. Environ. Sci. Technol., 1993; 27:961–969. Huang W, Schlautman MA, Weber WJ Jr. A distributed reactivity model for sorption by soils and sediments. 5. The influence of near-surface characteristics in mineral domains. Environ. Sci. Technol., 1996; 30:2993–3000.

16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

29. 30. 31. 32.

Chapter 13 KINETICS OF DESORPTION OF 2, 4-DICHLOROPHENOXYACETIC ACID FROM HUMIC ACID, METAL OXIDES AND METAL OXIDEHUMIC COMPLEXES C.Liu and P.M.Huang Department of Soil Science, University of Saskatchewan, 51 Campus Drive, Saskatoon, SK S7N 5A8, Canada 13.1. INTRODUCTION About a half century ago the introduction of 2, 4-dichlorophenoxyacetic acid (2, 4-D) and 4-chloro-2-methylphenoxyacetic acid (MCPA) started the widespread use of pesticides in Canada [1]. Since then, the increases in crop yields have largely been due to the use of enormous quantities of pesticides each year [2]. As the demand of yield and quality of food and fiber increases with the world’s increasing population, the application of pesticides in agricultural environments to control weeds, pests or diseases continues to grow. However, pesticide residues may constitute a significant source of contamination of air, water, soil and food. Attention has increasingly been paid to the adverse effects of certain pesticides on environmental quality and ecosystem health in recent years [3,4]. The fate of a pesticide in the environment is governed by transformation and transport processes and the interaction of these processes [5]. Regardless of the method of application, large amounts of pesticides ultimately reach the soil. As a result, soils are accumulating ever increasing amounts of residues of a wide variety of pesticides, which then volatilize into the air, move into water through leaching and surface run-off, are absorbed by organisms, and are broken down into other products. Therefore, the fate of pesticides in soils is of major interest to environmental scientists. Biotic and abiotic processes in soil can transform pesticides. Although it is assumed that biotic transformations predominantly control the degradation of pesticides in soil and sediment environments, biotic transformation processes undoubtedly are influenced by abiotic transformation processes, especially their interactions with soil minerals [6]. Sorption and desorption of pesticides by soils are key processes from the very beginning of pesticide use. The availability of pesticides for uptake by the target organisms, for chemical and biochemical transformations, and for transport in solution or in gaseous phases are all affected by sorption-desorption processes. The main factors affecting the sorption-desorption of pesticides in soils are the physicochemical characteristics of pesticides, the nature and properties of soil colloids and the soil environment [6,7]. Metal oxides and humic substances have large and physicochemically active surface areas and thus provide a major sink for pesticide retention. The ability of soils to bind 2, 4-D is significantly correlated with their contents of organic matter and pyrophosphate-extractable Al and Fe [8]. Organic and inorganic components are closely associated in soil [9]. However, little is known of the dynamics of release of pesticides adsorbed by metal oxide-humic complexes. Aluminum, iron, and manganese oxides are the most abundant metal oxides in natural environments. They are readily complexed with organic materials in soils and related environments. The objective of this work was to investigate the kinetics of 2, 4-D desorption from Al, Fe, and Mn oxide-humic complexes for comparison with 2, 4-D release from humic acid and pure Al, Fe and Mn oxides.

116

| CHAPTER L3: KINETICS OF DESORPTION OF 2, 4-DICHLOROPHENOXYACETIC ACID

13.2. MATERIALS AND METHODS 13.2.1. Humic Acid, Metal Oxides and Metal Oxide-Humic Complex Samples The humic acid (HA) used in this study was the standard soil humic acid (1S102H) obtained from the International Humic Substances Society (IHSS). The Al, Fe, and Mn oxides and the metal oxide-humic complexes were synthesized with Alanar grade chemicals. Deionized distilled water used throughout this study was autoclaved. The synthesis of the metal oxides and their humic complexes followed the procedure reported in our previous study [10]. Aluminum oxide was synthesized by titrating 800 mL of 6.25×10−3 M AlCl3 solution with 0.2 M NaOH solution under Table 13.1 Basic properties of metal oxides, metal oxide-humic complexes and the humic acid Sample

N2-BET Specific Surface, m2 g−1

PZSE

Al oxide Fe oxide Mn oxide Al oxide-humic complex Fe oxide-humic complex Mn oxide-humic complex HA

36.12±1.31 191.8±5.5 49.78±1.62 18.50±0.26 16.17±0.63 17.11±0.78 22.40±0.55

7.5±0.1 7.8±0.2 3.2±0.1 5.8±0.1 4.6±0.1 4.2±0.1 NA

vigorous stirring until the pH was 8.2. The rate of titration was 60 mL h−1. The solutions were then made up to 1000 mL so that the final Al concentration was 5×10−3 M. The suspensions were aged for 5 days at 296.5±0.5 K in a polypropylene container. Iron oxide was synthesized by titrating 0.15 M Fe(NO3)3 solution with 13% aqueous ammonia to pH 5.5. The suspensions were aged for 2 days at 296.5±0.5 K. Manganese oxide was synthesized by the method of McKenzie [11]. The precipitates were dialyzed against water and then freeze-dried. The Al oxide product was a mixture of gibbsite and bayerite. The Fe oxide product was amorphous. The Mn oxide product was birnessite. The metal oxide-humic complexes were prepared by reaction of the metal oxides with catechol for 20 days [10]. The specific surface areas of the solid sample were measured from a 5-point BET N2 adsorption isotherm [12] obtained with a Quantachrome Autosorb-1 apparatus. Prior to N2 adsorption, 100 mg samples were outgassed for 24 h at 10 mTorr and 296.5±0.5 K. During N2 adsorption the solids were thermostated in liquid N2 (77– 78 K). The points of zero salt effect (PZSE) of the solid samples were determined in 0.01, 0.1, and 1 M NaCl solutions with the potentiometric method of Parks and de-Bruyn [13] as modified by Atkinson et al. [14]. The automatic titration described by Sakurai et al. [15] was conducted with a Metrohm Model 682 titroprocessor. The properties of the metal oxides, the corresponding metal oxide-humic complexes, and the humic acid are given in Table 13.1. 13.2.2. Desorption of 2, 4-Dichlorophenoxyacetic Acid Desorption of 2, 4-D from the solid samples was conducted following one-day equilibration of 2, 4-D with the solid samples. One hundred milligrams of each sample, i.e., Al, Fe, and Mn oxides, the metal oxide-humic complexes, and HA, were suspended in 10 mL water and shaken for 1 h. The pH of the suspension was adjusted to 6.0. Ten milliliters of 14C-labeled 2, 4-D stock solution (0.494 µgmL−1, pH 6.0) in 0.02 M NaNO3 were added to the suspension. Four mg of thimerosal (sodium ethyl-mercuri-thiosalicylate) were added to the mixture to inhibit the growth of microorganisms. The final volume was 20 mL. The 2, 4-D concentration was 0.247 µgmL−1 and the solid concentration was 5 gL−1. The mixture was shaken for 24 h at 298 K. The suspension was filtered with a 0.1 -µm Millipore membrane. The concentration of 2,4-D remaining in the solution was determined by liquid scintillation counting with a Beckman LS3108. The amounts of 2,4-D adsorbed by the solid phases were calculated by the difference in 2,4-D concentrations before and after equilibration with the solid phases. The solid phase on the filter membrane was washed 5 times with water. The solid sample was transferred to a plastic vial and the volume of the suspension was brought to 10 mL. Ten milliliters of 0.02 M sodium nitrate or sodium citrate at pH 6.0 with 4 mg of thimerosal were added to the vial. The mixed suspension was shaken for 0.5, 1, 2, 4, 8, 16, and 24 h at 298 K. The suspensions at the end of each reaction period were filtered through a 0.1 -µm Millipore membrane. The concentration of 2,4-D in the solution was also determined by liquid scintillation counting. Each experiment was carried out in duplicate.

MATERIALS AND METHODS |

117

Figure 13.1 Amounts of 2,4-D sorbed by metal oxides, metal oxide-humic complexes and humic acid at 298 K after 24h

Figure 13.2 Percentage of 2, 4-D desorbed after 24h at 298 K from metal oxides, metal oxide-humic complexes, and humic acid

13.3. RESULTS AND DISCUSSION 13.3.1. Amounts and Percentages of 2, 4-D Desorbed The amounts of 2,4-D sorbed by various metal oxides, metal oxide-humic complexes and HA after 24h are given in Figure 13.1. The metal oxide-humic complexes sorbed substantially less 2,4-D than the metal oxides and sorbed a similar amount of 2,4-D compared with the HA. This is due to metal oxides’ specific surface area (SA) and surface charge variations (Table 13.1) caused by catechol humification [16]. The SA of the Al, Fe, and Mn oxide-humic complexes decreased by 49%, 92%, and 66% compared with the respective metal oxides (Table 13.1). Thus, the reactive sites on the surface of the metal oxide-humic complexes that sorb 2,4-D were decreased. Compared with the respective metal oxides, the Al and Fe oxide-humic complexes had lower PZSE values (Table 13.1) and thus had less net positive surface charges at the same pH. This should decrease the sorption of 2,4-D, which is an anion at pH 6.0. However, humification of catechol on the surface increased the PZSE value of the Mn oxide, which thus had more net positive surface charges at the same pH. However, this effect was counteracted by the large decrease in the SA of the Mn oxide after humification. The data indicate that both SA and surface charge affect the sorption of 2,4-D by the metal oxides and their humic complexes. It should be noted that the different metal oxide-humic complexes had similar SA (Table 13.1). Therefore, the amounts of 2,4-D sorbed were similar. Following a 24h period of 2,4-D sorption, the addition of nitrate and citrate to the systems resulted in desorption of 2,4-D from the solid samples. The percentage of 2,4-D desorbed after 24-h over the 2,4-D sorbed ranged from 11% to 56% (Figure 13.2). Compared with nitrate, citrate caused much more desorption of 2,4-D from the same solid sample (Figures 13.2 and 13.3). Pesticides can be divided into cationic, basic, anionic, acidic, and nonionic classes. 2,4-D is an acidic and anionic pesticide and could thus be sorbed by metal oxides and metal oxide-humic complexes through electrostatic attraction with the positively charged reaction sites on the metal oxides and metal oxide-humic complex surfaces and/or through ligand exchange reactions. Citrate had a higher ability to exchange 2,4-D from the surfaces of metal oxides and the metal oxidehumic complexes, resulting in more 2,4-D desorption. Compared with the respective metal oxides, the metal oxide-humic complexes desorbed less 2,4-D with both nitrate and citrate (Figures 13.4a, 13.5a and 13.6a). This is because the metal oxides sorbed more 2,4-D (Figure 13.1) and had more 2,4-D

118

| CHAPTER L3: KINETICS OF DESORPTION OF 2, 4-DICHLOROPHENOXYACETIC ACID

Figure 13.3 Amounts of 2,4-D released into solution by 0.01 M nitrate and citrate from (a) metas oxides and (b) metal oxide-humic complexes and humic acid. Al stands for Al oxide, Fe for Fe oxide, and Mn for Mn oxide. Al-H, Fe-H, and Mn-H stand for Al oxide-, Fe oxide-, and Mn oxide-humic complexes, respectively

available to be substituted with nitrate or citrate anions. Although the same trend was observed for the percentage of 2,4-D desorbed over the 2,4-D sorbed on the Al and Fe oxides and their humic complexes, relatively more 2,4-D was desorbed by nitrate in the Mn oxide-humic complex system compared with the Mn oxide system (Figures 13.4b, 13.5b and 13.6b). The Mn oxide-humic complex had a higher PZSE than the Mn oxide (Table 13.1). The opposite trend was observed for the Al and Fe oxides and their humic complexes. The higher PZSE value indicates more net positive surface charges on the Mn oxidehumic complex than on the Mn oxide at pH 6.0. For this reason the Mn oxide-humic complex has a higher affinity for anions such as nitrate and citrate. This effect is particularly pronounced for the weak ligand nitrate. 13.3.2. Kinetics of 2, 4-D Desorption The amounts of 2,4-D desorbed from the metal oxides, the metal oxide-humic complexes and humic acid increased with reaction time (Figure 13.3). Desorption of 2,4-D reached apparent equilibrium after 8h, as indicated by the similar amounts of 2,4-D released after 8h, 16h and 24h. Kinetic and empirical equations, including zero-order, first-order and second-order, the parabolic diffusion equation, the modified Freundlich equation and the Elovich equation, were used to fit the 2,4-D desorption data at reaction times from 0.5 to 8 h. To compare the relative desorption rates in various systems, the percentage of 2,4-D

MATERIALS AND METHODS |

119

Figure 13.4 (a) Amounts and (b) percentages of 2,4-D released into solution by 0.01 M nitrate and citrate from Al oxide and its humic complex. Al stands for Al oxide and Al-H for the Al oxide-humic complex

Figure 13.5 (a) Amounts and (b) percentages of 2,4-D released into solution by 0.01 M nitrate and citrate from Fe oxide and its humic complex. Fe stands for Fe oxide and Fe-H for the Fe oxide-humic complex

Figure 13.6 (a) Amounts and (b) percentages of 2,4-D released into solution by 0.01 M nitrate and citrate from Mn oxide and its humic complex. Mn stands for Mn oxide and Mn-H for the Mn oxide-humic complex

released was used for data fitting. The degree of fit of the rate equations to the data was examined with the correlation coefficient (r2), probability (p), and standard error (SE) of linear regression analysis. An example of the fitting of various equations to the 2, 4-d desorption data in the Al oxide system is given in Table 13.2. Similar behavior was observed in other systems. The data show that 2,4-D desorption from the metal oxides, metal oxide-humic complexes, and HA is best described by the parabolic diffusion Eq. 13.1 (Table 13.2), where qt is the percentage of 2,4-D desorbed from the metal oxides, their humic complexes and the humic acid at reaction time t, k is the apparent diffusion coefficient and B is a constant. Eq. 13.1 is often used to indicate rate-limiting diffusion-controlled phenomena [17]. (13.1) Figure 13.7 shows the fit of the parabolic diffusion equation to the data of 2, 4-D desorption from Al oxide. The apparent diffusion (rate) coefficients of 2,4-D desorption from the metal oxides, the metal oxide-humic complexes, and humic acid caused by citrate were substantially longer than those of 2,4-D desorption from var Table 13.2 Fitting of various equations to 2, 4-D desorption from Al oxide Equation r2 0-order 1st-order 2nd-order Overall diffusion Elovich Freundlich a Standard error (µg/g)

Nitrate

Citrate

p

SEa

r2

p

SE

0.934 0.839 0.717 0.997 0.988 0.991

7.52×10−3

0.06 0.53 1.09 0.05 0.07 0.13

0.939 0.865 0.761 0.987 0.955 0.984

6.43×10−3 2.19×10−2 5.36×10−2 6.42×10−4 4.07×10−3 8.45×10−4

2.88×10−2 7.02×10−2 6.86×10−5 5.30×10−4 3.47×10−4

0.08 0.54 0.88 0.04 0.08 0.04

120

| CHAPTER L3: KINETICS OF DESORPTION OF 2, 4-DICHLOROPHENOXYACETIC ACID

Figure 13.7 Fitting of the overall parabolic diffusion equation to 2, 4-D desorption from Al oxide at 298 K

ious systems caused by nitrate (Table 13.3). As discussed above, citrate has a greater ability to replace 2,4-D from the surfaces of metal oxides, their humic complexes, and humic acid, resulting in faster release of 2,4-D into solution. Compared to Al and Fe oxides, the rate coefficients of 2,4-D desorption from the Al and Fe oxide-humic complexes were smaller with both nitrate and citrate. This is attributed to the decrease in PZSE of Al and Fe oxides after catechol humification. Lower PZSE would increase the activation energy barrier due to stronger repulsion between metal oxidehumic complexes and nitrate or citrate. However, the rate coefficient of 2,4-D desorption from the Mn oxide-humic complex by nitrate was larger than that of 2,4-D desorption from the Mn oxide. This is similar to the amounts of 2,4-D desorbed by nitrate in the Mn oxide and Mn oxide-humic complex systems. Compared with the Mn oxide, the Mn oxide-humic complex has a higher PZSE, indicating more net positive charges on the surface of the latter at the same pH. Therefore, the Mn oxide-humic complex would have a higher affinity for anions, leading to a faster release of 2,4-D through their exchange with 2,4-D. This effect was observed even for the weak ligand nitrate. It should be noted that HA had the smallest rate coefficient of Table 13.3 Apparent diffusion (rate) coefficients of desorption for 2, 4-D from various samples in Eq. 13.1 at 298 K Sample

Rate coefficients (percent released h−1)a

Nitrate

Citrate

Al oxide Fe oxide Mn oxide Al oxide-humic complex Fe oxide-humic complex Mn oxide-humic complex HA a The standard errors were less than 5%

9.0 11.2 3.9 6.8 5.8 4.6 3.7

16.6 17.6 14.5 11.7 10.8 8.2 7.8

2,4-D release caused by both nitrate and citrate. This indicates that, compared to metal oxides and metal oxide-humic complexes, the rate of 2,4-D desorption from HA was relatively low, apparently due to its stronger binding of 2,4-D. 4. CONCLUSIONS Desorption of 2,4-D from the investigated metal oxides, metal oxide-humic complexes and HA obeys the parabolic diffusion equation. The amounts and percentages of 2,4-D released and the reaction rates depend on the type and surface properties of the solid sorbents. Compared with nitrate, citrate generally induced more release of 2,4-D. The percentage of 2,4-D desorbed ranged from 11% to 56%. The amounts, percentages and rates of 2,4-D desorption from the metal oxide-humic complexes were much lower than for desorption from the pure metal oxides due to alteration of the SAs and surface charges of the metal oxides caused by catechol humification. However, the amounts, percentages and rates of 2,4-D desorption from the metal oxide-humic complexes were slightly larger than for desorption from the HA system. Our results are of fundamental significance in understanding the release of pesticides such as 2,4-D from metal oxides, metal oxide-humic complexes and HAs and the impact of their effects on the transformations and degradation of this pesticide in soil and related environments.

MATERIALS AND METHODS |

121

ACKNOWLEDGEMENTS This research was supported by Research Grant GP2383-Huang from the Natural Sciences and Engineering Research Council of Canada. REFERENCES 1. 2. 3. 4. 5. 6.

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

Smith, AE. Herbicides and the soil environment in Canada. Can. J. Soil Sci., 1982; 62:433–460. Khan SU. Pesticides in the soil environment. New York: Elsevier, 1980. Hodgson E, Roe RM, Motoyama N. Pesticides and the future: Toxicological studies of risks and benefits. Raleigh, North Carolina: North Carolina State University, 1991. Hoar SK, Blair A, Holmes FF, Boysan OD, Robel RJ, Hover R, Fraumeni JF Jr. Agricultural herbicide use and risk of lymphoma and soft-tissue sarcoma. J. Am. Med. Assoc., 1986; 266:1141–1147. Cheng HH. Pesticides in the soil environment: Processes, impacts, and modeling. Madison, Wisconsin: Soil Science Society of America, 1990. Huang PM. Transformations of pesticides as influenced by soil minerals in the environment. In: Agricultural health and safety: Workplace, environment, sustainability. Saskatoon, SK Canada: Centre for Agricultural Medicine, University of Saskatchewan, 1994: 165–175. Sparks DL. Kinetics of soil chemical processes. In: Kinetics of pesticide and organic pollutant reaction. New York: Academic Press, 1989:128–145. Huang PM, Grover R. Kinetics and components involved in the adsorption of 2, 4-D by soils. In: Principles of health and safety in agriculture. Boca Raton, FL: CRC Press, 1990:228–230. Schnitzer M. Organic-inorganic interactions in soils and their effects on soil quality. In: Environmental impact of soil component interactions. Vol. 1. Natural and anthropogenic organics. Boca Raton: CRC/Lewis Publishers, 1995: 3–19. Liu C, Huang PM. The influence of catechol humification on surface properties of metal oxides. In: Ghabbour EA, Davies G eds., Humic substances. Structures, models, and functions. Cambridge: Royal Society of Chemistry, 2001: 253–270. McKenzie RM. The synthesis of cryptomelene and some oxides and hydroxides of manganese. Mineral Mag., 1971; 38:493–502. Gregg SJ, Sing KSW. Adsorption, surface area and porosity, 2nd Edn. London: Academic Press, 1982. Parks GA, de Bruyn PL. The zero point of charge of oxides. J. Phys. Chem., 1962; 66:967–973. Atkinson RJ, Posner AM, Quirk JP. Adsorption of potential determining ions at the ferric oxide-aqueous electrolyte interface. J. Phys. Chem., 1961; 71: 550–558. Sakurai K, Ohdate Y, Kyuma K. Potentiometric automatic titration (PAT) method to evaluate zero point of charge (ZPC) of variable charge soils. Soil Sci. Plant Nutr., 1989; 35:89–100. Liu C, Huang PM. Kinetics of 2,4-dichlorophenoxyacetic acid adsorption by metal oxides, metal oxide-humic complexes and humic acid. Proceedings of the 11th International Humic Substances Society Conference. Boston: IHSS, 2002: 405–407. Sparks DL. Kinetics of sorption/release reactions on natural particles. In: Structure and surface reactions of soil particles. IUPAC Series on Analytical and Physical Chemistry of Environmental Systems. Chichester, UK: Wiley, 1999: 419–454.

Part 3 METAL BINDING AND MOBILITY: THEORY, DATA AND CONSEQUENCES

Chapter 14 EXPLORING THE MOLECULAR CHARACTER AND HETEROGENEITY OF HUMIC SUBSTANCES VIA THE STUDY OF THE ION-BINDING PROCESS USING AN EXTENDED POLYELECTROLYTE MODEL Josemaría García-Mina Department of Chemistry and Soil Chemistry, University of Navarra and R&D Department, Inabonos/ Roullier Group, Poligono Arazuri-Orcoyen, C/C n° 34, 31160 Orcoyen, Spain 14.1. INTRODUCTION At least two crucial issues must be addressed in order to achieve an adequate knowledge of the dynamics of metals and some organic substances in natural ecosystems: 1) the molecular nature and structure of humic substances (HSs); and 2) the physicochemical parameters that define the ion-binding process in humic molecular systems. Both questions are closely related, so that new knowledge about one sheds light on the other. As regards the first question, a very interesting debate about the real chemical nature and structure of HSs is ongoing in the literature [1]. Whereas some authors propose that HSs are supramolecular associations of relatively simple and small molecules through weak attraction forces [2–4], others assign a real macromolecular character to HSs [5–7]. However, despite the lively terms in which this debate is conducted, if we analyze the experimental data supporting both points of view it becomes clear that they are perfectly compatible. Thus, while studies defending a supramolecular approach do not permit us to neglect a possible macromolecular character of the molecules involved in a possible supramolecular structure (or of that supramolecular structure in itself), neither do those who support a macromolecular approach neglect possible formation of supramolecular aggregates under specific environmental conditions. In this framework, the study of the chemical or biological functions of HSs may provide valuable information about their behavior and, hence, about their chemical nature. Thus, through “how they work” we might know “what they are.” Among the different properties or functions to be studied, the ion-binding process is especially important. In fact, a great number of environmental problems or situations would be better understood if a deeper knowledge about the ion-binding processes involved in them were achieved. However, the development of complete models, that is, those that explicitly consider 1) the possible macromolecular character and dynamic electrostatic effects; 2) the possible binding site heterogeneity; and 3) possible ion competition that can correctly describe in chemical terms these types of interactions in HSs, is complicated due to both the molecular complexity of HSs and the lack of knowledge of HS chemical structure as a whole. Some semi-empirical models, such as the NICA-Donnan and the NICCA-Donnan models, have proven excellent at describing experimental data, thereby providing some useful empirical parameters to gain information regarding the molecular nature of HSs [8–10]. However, the values of some structurally related parameters that are obtained from the iterative adjusting process do not always agree with those obtained from direct chemical analysis [11,12]. This fact suggests that significant factors involved in the ion-binding process in HS are not, or are only partially, explicitly considered in this approach. In these circumstances, the polyelectrolytic approach might provide valuable and complementary information, since the chemical meaning of the results obtained is based on the physicochemical grounds of the equations used to analyze the experimental data [13]. In fact, different authors have employed this model (or related models, such as the electrostatic model) to analyze the chemical behavior and chemical nature of different HSs, thereby obtaining valuable information [14,15]. However, it is necessary to note that the interpretation of this information in relation to the possible molecular character of HSs is limited by the need to have a previously assumed, specific molecular geometry or molecular pattern [14,15]. In this context, the aim of this work is: 1) to present a complete model based on the polyelectrolytic approach—the “extended polyelectrolytic model” (EPM)—that is able to analyze the proton- and metal-binding process in HSs and other complex molecular systems without the need to assume a molecular geometry or molecular pattern [16]; and 2) to detail the application of the model, considering the conceptual consequences with reference to our current knowledge of the chemical nature of HSs. As will be discussed, the introduction into the model of specific parameters and functions directly related to either the macromolecular character or the heterogeneity or possible ion-competition will permit us to explore the real implication of these characteristics in the whole nature and chemical behavior of the HSs under study.

124

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

14.2. THEORETICAL BASIS In order to develop the model we assume that the ion-binding process is principally governed by electrostatic forces. Likewise we assume ideal behavior of both the molecules and the solution. 14.2.1. Proton Binding Analysis We consider HSs as a molecular system composed of a number of identical ideal molecules with i main classes of acid groups involving sites of different or equal intrinsic acidity (possible heterogeneity) that can be influenced by variation of the electrostatic potential in the structural environment of the binding sites and associated with conformational changes that the molecules may undergo (possible macomolecular character). From acid-base titration of a given molecular system, we can obtain for each mean class i of acidic groups a set of values of Kapp associated with the variation of ′ , in which each value of Kapp, corresponding to each site i that is dissociated as a function of pH, is defined in the classical Eq. 14.1 derived directly from the equilibrium. (14.1) Equally, the influence of the variation of charge in the structural environment of each site i on the real value of K (Kint) can be considered using Eq.14. 2 for the dissociation process [13]. (14.2) Here, the expression of the electrostatic potential ′ (Z) around each site i may also be expressed by Eq. 14.3 (we assume that all surface groups experience the same average electrostatic potential) [13], (14.3) where w is an electrostatic function depending on the molecular conformation (intercharges distance), z is the charge of the ion and Z is the net charge of the molecule. If, as proposed by Tanford [13], we let Z=−n′ , where n is the total concentration of acid groups belonging to the class i (as Tanford proposed, this assumption will be closer to reality for low values of ′ , Z being lower than−n′ as a increases due to the binding of counterions [13]), from Eqs. (14.2) and (14.3) we obtain the well known Eq. 14.4. (14.4) Eq. (14.4) can also be expressed as Eq. 14.5, (14.5) where pKappi and pKinti define each site i that is dissociated at a certain value of pH (′ i), and wi is the value of w that corresponds to the structural conformation of the macromolecule associated with ′ i. Once w has been calculated, we can obtain the value of pKinti from Eq. 14.5 that is independent of ′ i. This corresponds to the site i that is dissociated at a certain pH and that defines ′ i. Conceptually, we obtain a set of values of a as a function of pKint. Calculation of w ( ). To calculate w (′ ) we have two possible alternatives: 1) to use one of the equations that define w as a function of a previously assumed molecular geometry [13]; or 2) to develop a graphical approach derived from the analysis of the plot of pKapp vs ′ . In our model, we use the graphical approach to avoid the need to assume a specific molecular geometry or molecular pattern. Considering Eq.14.5 and analyzing the pKapp vs ′ plot, we observe that the values of both the expressions (0.868win and pKinti) for each site i correspond to the value of the slope at the point (pKappi, ′ i) and the intersection with the pKapp axis of the tangent line of pKapp (′ ) in (pKappi, ′ i), respectively (Figure 14.1). Thus, the following equations are obtained: (14.6) (14.7) For practical purposes we can use a mathematical function to describe pKapp(′ ), thus facilitating the application of Eqs. 14.6 and 14.7. Among the different possibilities, a polynomial function seems to be the best since it is mathematically coherent with the condition of obtaining a finite pKapp for ′ =0. Likewise, the use of different polynomial functions for describing the different parts of pKapp(′ ) is also possible. The factor dpKapp(′ )/d(′ ) may also be directly obtained from the experimental data by using the expression ′ [pKapp(′ )]/′ (′ ).

THEORETICAL BASIS |

125

Figure 14.1 Graphical representation of the tangent line distribution for the curve pKapp(′ ).

In this way, we have described the proton-binding process for each site i belonging to a specific class i of acidic groups by means of values of w, pKapp and pKint. From the values set of pKint an affinity spectrum may be obtained using a density function (Sips, Gauss, etc…). 14.2.2. Metal Binding Analysis As in the case of the proton-binding approach, in order to develop the mathematical formalism of the model we consider HSs as composed of a number of identical ideal molecules that have different complexation sites with equal or different intrinsic stability (possible heterogeneity), whose M(H)-binding process (possible ion-competition) can be influenced by variation of the electrostatic surface potential associated with the conformational changes that are undergone by the molecules as a result of the influence of experimental conditions, such as the pH and ionic strength (I), and the complexation process itself (possible macromolecular nature). In principle, we can define the complexation process of a metal M by a specific HS with different complexation sites by means of the following parameters for a site: 1) characterizes the Si-M interaction without considering either H-M competition or electrostatic effects on Si-M(H); 2) that characterizes the Si-M interaction considering H-M competition but not the electrostatic effects on Si-M(H); 3) that characterizes the Si-M interaction considering the electrostatic effects on Si-M(H) but not H-M competition; 4) that characterizes the Si-M interaction considering both the H-M competition and the electrostatic effects on Si-M(H); 5) that characterizes the Si-H interaction without considering the electrostatic effects on Si-M (H); and 6) that characterizes the Si-H interaction considering the electrostatic effects on Si-M(H). By assuming that 1) the stoichiometry Site:M is 1:1; 2) the nature of the sites may be polydentate but only one of the binding centres (the functional groups that form the binding site) experiences a significant dissociation process; and 3) there can exist a Mz+: H+ exchange ratio (′ ) in binding sites whose value can be obtained from the experimental study. Parameter ′ would be the ratio between the bound metal and the concentration of base that must be used to keep a constant pH during the titration. The equilibrium that governs the binding process is defined by Eqs. 14.8 and 14.9. (14.8) From Eq. 14.8 we have for Si: (14.9) As pointed out in the analysis of the proton-binding process, the influence of the conformational changes that the molecule can undergo as a consequence of both experimental conditions and the complexation process and on the electrostatic charge in the structural environment of Si-M(H) can be explicitly considered by introduction of an electrostatic factor. This parameter can be defined as a function of the electrostatic surface potential, e−′ (Z,z (′ ,′ ,D)) (D is the intercharge distance z-Z, z is the ion charge, and Z the net molecular charge). Thereby, may be defined as: (14.10)

126

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

Thus, the different binding constants described above that may be used to define the whole binding process are related to one another by Eq. 14.11, 4.11) where e−′ (′ ,′ ) and e−′ ’(′ ,′ ) define the electrostatic factor e−′ (Z,z(′ ,′ ,D)) for the M-and H- binding process, respectively, and ′ is defined as ([MB]/MBA). [MB] is the concentration of complexed metal (also expressed as bound metal) and MBA A is the maximun metal binding capacity that can be calculated from the complexation study. From Eq. 14.11 we obtain: (14.12) and from Eq. 14.12 we obtain: (14.13) (e−′ (′ ,′ )

e−′ ’(′ ,′ ))

If calculations of the electrostatic factors and and the values are carried out, the values may be obtained as follows: 1) directly from Eq. 14.13 if the value of is known from a previous study of the proton-binding process; and 2) if the value is unknown, the calculation can be performed from the complexation study but using two differing values of one of the specific experimental variables, such as pH or I, that affect the binding process but not the structural identity and concentration of the binding sites. Thus, with values (I) and (II) of a certain experimental condition we have as follows. For (I) we have: (14.14) For (II) we have: (14.15) If, for example, we consider two different I at fixed pH, from Eqs. 14.13 and 14.14 we obtain Eq. 14.16 (14.16) and knowing , can be obtained from Eqs. 14.14 or 14.15. In our opinion, a complete study of an M-HS interaction should include a proton-binding analysis before the metal-binding analysis. This strategy, besides facilitating the whole application of the model, also provides more complete information about the intrinsic characteristics of the functional groups probably involved in the metal binding process. Calculation of e- ( , ) and e- ( , ); and [(I) and (II)]. From Eq. 14.11 we have Eq. 14.17. (14.17) If, as in the proton-binding study, we define ′ (′ ,′ )=2zWZ, where z is the charge of M, Z is the net charge of the molecule and W is an electrostatic function that depends on the inter-charge distance, and we express Eq. 14.16 in a logarithmic form, we obtain Eq. 14.18. (14.18) As an approximalion, we define , where nM=MBA and n is the concentration of acidic groups to obtain Eq. 14.19. (14.19) From analysis of the plot of vs MB we can observe that the values of both the expressions and for each site i correspond to the value of the slope of the point ( MBi) and the intersection of the tangent line of (MB) in (, MBi) with the axis, respectively. Thus, we can obtain Eqs. 14.20 and 14.21. (14.20)

THEORETICAL BASIS |

127

(14.21)

As in the case of the proton-binding analysis, for operational calculations we can express (MB) by means of a mathematical function. Again, among the different functions that can be employed, polynomial functions would be more coherent with the condition of obtaining a finite value for MB=0. Evidently, the calculation of the expression may also be directly carried out from the experimental data by means of . Likewise, we can calculate e−′ (′ i,′ ) (the electrostatic factor associated with the metal binding process) from the expression or from each value of Wi, and e−′ ’(′ i,′ ) (the electrostatic factor associated with the proton binding process) from the expression where z is the charge of M. In this way we have described the metal-binding process by the calculation of the differently defined stability constants: , , , , and , and specific values for the electrostatic factors (e−′ (′ i,′ ), e−′ ’(′ i,′ )) and W. From these value sets of stability constants an affinity spectrum may obtained using some density function (Sips, Gauss, etc…). 14.2.3. Some Relationships between EPM Results and the Chemical Nature and Structure of the Molecule Regarding the possible macromolecular nature of HSs, we have introduced into the model two inter-related and equivalent parameters that must reflect the influence of the conformational changes that the molecule can undergo as a result of the electrostatic charge in the structural environment of a given binding site: the electrostatic functions, W for metal-binding and w for proton-binding, and the electrostatic factors (e−′ (′ i,′ ), e−′ ’(′ i,′ )). Of these parameters, W (or w) may be more useful in practical terms to analyze the HSs molecular nature since its value is easily related to the molecular conformation. Thus, if we analyze the expression of the electrostatic potential (′ =2W(or w)zZ), we observe that the contributions of the charges and the inter-change distance (R) in the expression of ′ have been separated. Therefore, the electrostatic function W (or w) must be inversely related to the inter-charge distance (W must be proportional to 1/R, where R is the inter-charge distance). Thus, a continuous decrease in W (or w) values would signify corresponding increases in R values and therefore more expanded molecular conformations. Conversely, less expanded molecular conformations will be related to an increase in W (or w) values associated with a decrease in R values. Thus, the magnitude of the variation of W (or w) values may be a good means of evaluating the macromolecular character of HS. In fact, by using known macromolecular systems with different structural patterns, these W (or w) values could be standardized. On the other hand, once the W (or w) values have been calculated and hence the macromolecular character has been evaluated, more information on the structural characteristics, for example the hydrodynamic radius, of the molecule may be obtained from specific mathematical expressions for W according to different molecular patterns [13]. It is important to note that significant but constant values of W (or w) do not necessarily signify a macromolecular character. It is the variation of W (or w) that will be directly related to significant macromolecular character. Thus, the existence of a relatively constant value of W (or w) may only be the consequence of electrostatic effects among different acidic groups in simple molecules with a high density of acidic groups [17]. With respect to the chemical HSs structure, the analysis of the variation of the apparent constants compared to those of the intrinsic constants may also give valuable information. Thus, in the case of the proton-binding studies, considering a given point of the plot of pK vs (′ ) as a reference, higher intrinsic acidic constants that correspond to lower apparent acidic constants could reflect the presence of certain acidic groups surrounded by a significant number of clusters with a high density of acidic groups. In this case, the structural confinement of certain acidic groups along with the increase of the negative charge derived from the ionization of the other adjacent acidic groups constrain their ionization (lower acidity and higher values of the apparent association constant). On the other hand, the effects of the different adjacent acidic groups on the electronic distribution in either the cluster or in the confined acidic groups cause an increase in the intrinsic acidity of the latter (higher intrinsic acidity and lower values of the intrinsic association constant). This same type of situation may also be found in the context of metal binding: lower values of the apparent association constants corresponding to higher values of the intrinsic association constants may be related to the real accessibility of the binding sites in the structure, which also will be associated with the conformational changes that the molecules undergo as a consequence of the binding process. As to possible metal

128

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

competition, we can observe in analyzing Eqs. 14.13 and 14.16 that the importance of this process may be directly evaluated by studying the ratio of to . Finally, regarding evaluation of the heterogeneity of both the intrinsic acidity and the intrinsic metal-binding stability, the calculation of specific affinity spectra may be a valuable tool. Thus, the parameter related to the width of the peak (or peaks) representing the formation function (′ ) as a function of Kint can be used to evaluate the degree of heterogeneity. It may be useful for comparative purposes as well. In order to present a practical description of the use of EPM, we describe in the following sections a study of the proton binding process followed by an analysis of the Zn-binding process for both the whole distribution of binding sites and the binding sites of high affinity. 14.3. MATERIALS AND METHODS 14.3.1. Materials HS was extracted from a compost of solid wastes (HSA) from wineries using the methods described by Stevenson [18] with some operational modifications. In short, 10 g of the moist organic material were weighed in a 250 mL flask to which 0.1 M NaOH was added until all the air had been displaced. After 48 hours stirring at 25°C in darkness, the supernatant containing the extracted humic system was separated from the solid fraction by centrifugation at 7650 g for 30 minutes. Subsequently, the supernatant was treated with the necessary amount of cation exchange resin (Amberlite IRA−118H+) until pH values were in the 3.5–4 range. The resin was separated by centrifugation (15 minutes at 5000 g) followed by filtration (Whatman 42 filter paper), and the supernatant containing the whole humic extract (HSA) was freeze-dried. The elemental composition of HSA was 48.22% C, 3.70% H, 43.52% O and 4.56% N. E465/E665:6.35 and the degree of aromaticity was 33.35%. 14.3.2. Methods Proton-Binding Studies. A stock solution of 0.1 % (w/w) HSA was prepared by dissolving freeze-dried HSA in 0.1 M NaOH. Once HSA was dissolved, adequate amounts of an H+-cationic exchange resin (Amberlite IRA−118H+) were added to the stock solution to attain a final pH of 3.5. After 6 h of treatment, the resin was separated by centrifugation (15 minutes at 5000 g) followed by filtration (Whatman 42). No coagulation of HSA occurred, as was confirmed by measuring the light absorption of the solution at 400 nm and pH 10 before and after the resin treatment. To carry out the titration studies, an aliquot of the stock solution containing 9.5 mg of HSA was added to water containing 0.5 mL of 0.1 M HClO4 and 3.5 mL of 1M KNO3. The final volume was 35 mL, I was 0.1 M and the initial pH was 2.9. The solution was titrated with 0.05 mL increments of 0.1 M NaOH using a Crison Compact titrator and the pH was measured by means of an Ingold combined pH glass electrode. To ensure that equilibrium between measurements had been reached, no increment in the base addition was made until the measured pH did not vary by ±0.01 pH units over 5 min. The titrations were carried out under N2. The experimental data were analyzed for functional group contents by the procedure of Takamatsu and Yoshida [19] using the Gran plot method to calculate the different equivalence points corresponding to each mean class of acidic groups (strong acid (SA), weak acid carboxylic groups (CA) and very weak acidity phenol groups (PA)) [20]. The dissociation degree (′ ) for each acidic class is obtained from the experimental data using Eqs.14.22 and 14.23. (14.22) (14.23) Metal-Binding Studies. The whole Zn-HSA binding distribution was studied with the following methodology. To study the stability of the Zn(II)-HSA binding process, the reaction between a constant concentration of HSA and different concentrations of the metal was carried out at constant value ′ H=8.5 and I 0.003M. The HS (mg): metal (mg) ratios in the reaction were: 2:0.10, 2:0.25, 2:0.40, 2:0.55, 2:0.70, 2:0.85, 2:1.00, 2:1.50, and 2:2.00. The concentration of HSA in the reaction was 90 mg/L. All the reactions were carried out in darkness, with constant stirring, at 25°C for 24h, and with air displaced by the liquid. Adequate amounts of Zn were added very slowly as sulfate solutions (0.1%), adjusting the pH with 0.1 M NaOH after each addition when equilibrium was reached. In order to measure the solution fractions of bound metal (MB), free metal (M) and HSA, the following methodology was used: 1) Once the reaction had finished, the precipitated fraction containing metal hydroxides and insoluble humic complexes was separated by centrifugation at 10,000 g for 30 min; 2) Ultrafiltration was used

THEORETICAL BASIS |

129

to separate bound and unbound metal in the solution obtained from step 1. A 1 kDa ultrafiltration cell (Filtron, polyethersulfone type) was used and ultrafiltration plots were first obtained with known concentrations of each metal in similar conditions of pH and I. In all cases, the retention coefficient of the ultrafiltration cell for both Zn-HSA complexes and HSA was 1 (full retention). From the supernatant, the concentration of HSA in solution was determined by measuring light absorption at 400 nm and pH 8.5 using a HSA standard curve obtained under the same experimental conditions. Metal concentrations in solution, before and after ultrafiltration, were analyzed by atomic absorption spectrometry (AAS). Values of HSA, MB and M considered for stability studies were obtained from reactions in which there was no precipitation of HSA. Differential pulse anodic stripping voltametry (DPASV) was employed for study of the binding process for the high affinity binding sites. Voltammetric mea surements were performed with an Autolab PGSTAT12 system (Eco Chemie, Utrech, The Netherlands) connected to a Metrohm (Herisau, Switzerland) 663 VA stand equipped with a static mercury drop electrode (SMDE) of the minimum size. A conventional three-electrode arrangement consisting of a glassy carbon counter electrode, an Ag-AgCl (3M KCl) reference electrode and a Metrohm multimode mercury electrode was used. To this end an aliquot of the stock solution of HSA used for titration studies corresponding to 0.11 mg was added to water containing 1.5 mL of 1M NaClO4 and 1 mL of 1 M NH4Cl (pH 8.5). The final volume before the binding study was 25 mL. The pH and I were 8.5 and 0.15 M, respectively. The solution was titrated with 0.05 mL increments of a solution of 5 mg L−1 of ZnNO3. To avoid potential interference of oxygen with the voltammetric measurements, a stream of nitrogen was used for 5 min before obtaining each data point (each final measure was the mean of three different measures). A delay of 10 min between each metal addition and the measurements was required to ensure that equilibrium in the binding process had been reached. The measurement conditions were a deposition potential of −1.2 V was applied for 60 s with stirring followed by a potential scan (from −1.2 V to −0.2 V) from which the labile Zn was analyzed by measuring the peak intensity at −1 V. From the plot of the peak intensity vs total added Zn with and without HSA, a measure of the labile Zn (lab Zn) can be obtained. This lab Zn includes both the free Zn ([Zn]) and the Zn forming inorganic complexes (in our case Zn(NH4)2+ is considered the main inorganic complex in solution). From the concentration of labile Zn and the equilibrium constant (K) of Zn (NH4)42+ the concentration of free Zn is calculated, Eq. 14.24. (14.24) 14.4. RESULTS 14.4.1. Functional Group Analysis of HSA Analysis of the Gran plot obtained from the titration study (Figure 14.2) allowed us to differentiate three acidic classes. A class I corresponding. to the strongest acidity (1.375 meq. g−1), aclass II corresponding to the weak acidity carboxylic acids (2. 433 meq. g−1) and a class III corresponding to the very weak acidity phenol groups (2.945 meq. g−1). Inour analysis we explicitly consider the classes II and III. 14.4.2. Proton Binding Analysis 1. From the values of a for each acidic class and the corresponding values of pH we obtain a set of values of pKapp as a function of a using Eq. 14.1 (Table 14.1). 2. In order to facilitate the further analysis we obtain specific polynomial functions to describe both pKapp(′ ) (Figures 14.3 and 14.4). 3. From these polynomial functions describing the pKapp(′ ) (Classes I and II) and applying Eqs. 14.6 and 14.7 we obtain a set of values of w(′ ), e−′ (′ )(′ ) and spe cific values of pKint for each ′ i but independent of it (Figures 14.5–14.8). 4. From the values of pKint and ′ (conceptually ′ (pKint)) for each mean acidic class an affinity spectrum may be obtained using the Gauss density function (Figures 14.9 and 14.10). Table 14.1 pH, ′ and pKapp values obtained from the base-titration of HSA COOH

pH

′

log(′ /(1−′ ))

pKapp

2.958 3.014

0.058 0.071

−1.211 −1.114

4.169 4.128

130

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

Figure 14.2 Gran plot analysis of the base titration of HSA.(′ )HClO4; (′ ) HClO4+HAS

Figure 14.3 pKapp(′ ) plot for carboxylic groups COOH

ph-OH

pH

′

log(′ /(1−′ ))

pKapp

3.082 3.160 3.249 3.356 3.490 3.668 3.912 4.255 4.734 5.517 6.245 7.076 7.550 8.730 9.400 9.780 10.035 10.200 10.350 10.480

0.074 0.082 0.102 0.127 0.161 0.207 0.275 0.379 0.526 0.703 0.899 0.078 0.238 0.392 0.528 0.646 0.749 0.849 0.931 0.998

−1.096 −1.046 −0.947 −0.836 −0.716 −0.584 −0.421 −0.214 0.045 0.375 0.947 −1.074 −0.506 −0.191 0.049 0.262 0.475 0.751 1.133 2.796

4.178 4.206 4.196 4.192 4.206 4.252 4.333 4.469 4.689 5.142 5.298 8.150 8.056 8.921 9.351 9.518 9.560 9.449 9.217 9.684

THEORETICAL BASIS |

Figure 14.4 pKapp(′ ) plot for phenolic groups

Figure 14.5 pKapp(′ ) (′ ) and pKint(′ ) (′ ) plots for carboxylic groups

Figure 14.6 pKapp(′ ) (′ ) and pKint(′ ) (′ ) plots for phenolic groups Table 14.2 Total added Zn ([Zn]T), bound Zn ([Zn]B), free Zn ([Zn]), ′ ([Zn]B/MBA) and Kapp values for the whole Zn-HSA binding process (Zn concentrations are expressed in mM) [Zn]T

[Zn]B

[Zn]

[Zn]B/[Zn]

′

Kapp

log Kapp

0.086 0.213 0.343 0.471 0.599 0.729 0.857 1.285 1.713

0.086 0.212 0.321 0.412 0.450 0.535 0.531 0.717 0.867

0.000 0.000 0.021 0.060 0.149 0.194 0.326 0.567 0.846

56000 1390 15.000 6.897 3.011 2.764 1.629 1.264 1.026

0.090 0.224 0.338 0.433 0.473 0.564 0.559 0.755 0.913

6580 1883 23.86 12.81 6.018 6.666 3.886 5.438 12.41

4.811 3.275 1.378 1.108 0.779 0.824 0.590 0.735 1.094

131

132

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

Figure 14.7 w(′ ) plots for carboxylic (COOH) and phenolic (ph-OH) groups

Figure 14.8 e-′

(′ )

(exp -EP) plots for carboxylic (′ ) and phenolic groups (′ )

Figure 14.9 Affinity spectrum (Gauss density function) for ′ (COOH) (pKint)

14.4.3. Metal-Binding Analysis Whole Zn-HSA Binding Study. 1) From the complexing study we obtain a set of values corresponding to the bound Zn (ZnB) and the free Zn (Zn) associated with each point of the complexation titration (Table 14.2). The maximum binding ability (MBA) may be calculated from that values set using the Langmuir representation ZnB vs (ZnB/Zn) (Figure 14.11). In our case, the MBA value obtained is 0.95 mM (this MBA value is referred to 90 mg of HSA). From this MBA value a set of values of ′ (ZnB/MBA) related to the ZnB and Zn can be obtained (Table 14.2). 2) From the values of ′ and Zn a set of values of is obtained from Eq. 14.17 (Table 14.2). 3) To facilitate application of Eqs. 14.20 and 14.21 we use a polynomial function to describe (ZnB) (Figure 14.12). 4) By using the polynomial function describing (ZnB), Eqs. 14.20 and 14.21 can be applied to obtain a set of values of both W(ZnB), e−′ (′ ,′ )(ZnB) and (the intrinsic constant that does not consider H-Zn competition) corresponding to each MB but independent of it (conceptually a set of values of MB or ′ , or Z as a function of ) (Figures 14.13–14.15). 5) Bearing in mind that the value of the corresponding to the pH at which the binding study was made is known from the proton-binding study (in this case the value associated with pH 8.5 is 6.99), a set of values of can be obtained from Eq. 14.13 (Figure 14.16). In this way we have characterized the Zn-HSA whole binding process by means of the calculation of a set of

THEORETICAL BASIS |

133

Figure 14.10 Affinity spectrum (Gauss density function) for a (ph-OH) (pKint)

Figure 14.11 Langmuir plot for the calculation of MBA

Figure 14.12 log KMapp (ZnB) plot from experimental data

values of W, e−′ (′ ,′ ), , and corresponding to each point of the complexation process operationally defined by MB (equivalent to ′ or Z). From the set an affinity spectrum can be obtained using the Gauss density function (Figure 14.17). The High Affinity Binding Study. 1) As explained above, from the titration study we obtain a set of correlative values of ZnB, Zn and ′ (Table 14.3). 2) By working as has been detailed in the whole binding study, we obtain different values sets of W, e−′ (′ ,′ ), , and that permit us to characterize the Zn-HSA binding process for the high affinity binding sites (Figures 14.18–14.21). Table 14.3 Total added Zn ([Zn]T), bound Zn ([Zn]B), free Zn ([Zn]), ′ ([Zn]B/MBA) and Kapp values for the Zn-HSA binding process associated with the higher affinity sites (Zn concentrations are expressed in mM) 103 [Zn]T

103 [Zn]B

[Zn]a

′

Kapp

log Kapp

0.15 0.31 0.46 0.61 0.76

0.13 0.28 0.41 0.55 0.68

1.9869E-10 3.1489E-10 4.7771E-10 6.6564E-10 8.3922E-10

0.0008 0.0017 0.0026 0.0034 0.0043

4204623.65 5471082.74 5407219.62 5152093.96 5107175.63

6.62 6.74 6.73 6.71 6.71

134

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

Figure 14.13 log KMapp (′ ) and log K’Mint (′ ) vs MB plots for the whole binding process

Figure 14.14 W (ZnB) plot for the whole binding process 103 [Zn]T

103 [Zn]B

[Zn]a

0.92 0.81 1.0774E-09 1.07 0.95 1.2775E-09 1.22 1.06 1.6368E-09 1.38 1.19 1.8657E-09 1.53 1.31 2.2386E-09 1.68 1.43 2.5751E-09 1.84 1.55 2.9617E-09 1.99 1.65 3.4602E-09 2.14 1.77 3.7801E-09 2.29 1.86 4.4938E-09 2.45 1.97 4.8969E-09 2.60 2.05 5.5841E-09 2.75 2.10 6.6736E-09 2.91 2.12 8.0479E-09 3.06 2.09 9.9559E-09 3.21 2.11 1.1254E-08 3.37 2.21 1.1792E-08 a[Zn] =labile [Zn]—inorganic complexed [Zn] (Zn[NH] ++) 4

′

Kapp

log Kapp

0.0051 0.0059 0.0066 0.0075 0.0082 0.0089 0.0097 0.0103 0.0111 0.0116 0.0123 0.0128 0.0131 0.0132 0.0130 0.0132 0.0138

4737775.96 4655535.61 4089127.58 4031400.43 3690166.59 3504619.72 3294785.16 3012215.84 2962892.56 2610730.01 2544294.68 2329701.31 1994021.78 1668372.14 1327028.25 1188806.74 1189248.70

6.68 6.67 6.61 6.61 6.57 6.54 6.52 6.48 6.47 6.42 6.41 6.37 6.30 6.22 6.12 6.08 6.08

THEORETICAL BASIS |

Figure 14.15 e −′

(′ )

135

(exp −EP) vs MB plot for the whole binding process

Figure 14.16 log KM vs ZnB plots for the whole binding process (′ log K’Mapp; ′ log KMint; ′ log KMint)

Figure 14.17 Affinity spectrum (Gauss density function) for ′ (log KMint)

14.5. DISCUSSION 14.5.1. Proton Binding Study As can be seen in Figure 14.5, considering as reference the first set of carboxylic groups with higher pKapp there exists a second set of carboxylic groups that presents lower pKapp values but higher values of pKint. This fact suggests the existence of regions in the HSA structure with a high density of carboxylic groups in which the dissociation of those acidic groups located in the inner part of these clusters is constrained by the dissociation of the groups located in the outside (electrostatic effects), causing lower pKapp values. However, the electronic effects of the outside groups on the electronic distribution in the structural location of inside groups would cause an increase of the intrinsic acidity of the latter, thus explaining the higher values of pKint. In the case of phenolic groups, this behavior was not observed (Figure 14.6). On the other hand, the variation of w values during the titration process indicates that among the different effects that could explain the differences between the pKapp and pKint values, specific conformational effects are also involved (Figure 14.7). Thus, in the section of the titration curve associated with the dissociation of carboxylic groups a molecular contraction

136

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

Figure 14.18 log KM vs ZnB plots for the high affinity binding process (′ log KMapp; ′ log K’Mint;′ log KMint)

Figure 14.19 W(ZnB) plot for the high affinity binding process

Figure 14.20 e−′

(′ )

(exp -EP) vs MB plot for the high affinity binding process

corresponding to the dissociation of the outside set is followed by a molecular expansion as the inner carboxylic groups are dissociated, due to the repulsion forces among negative charges [21]. However, in the case of the dissociation of the phenolic groups a clear and more significant molecular expansion is observed. These facts indicate that the phenolic groups are probably located inside the molecular conformation that HSA adopts at acid pH (before starting the titration process), whereas the carboxylic groups would be located in the outer surface region. In this sense, these results suggest a micelle-type molecular conformation at acid pH with the hydrophilic groups located outside and the less hydrophilic groups located inside, as has been proposed by some authors [6]. By contrast, the affinity spectrum obtained for both families of acidic groups (Figures 14.9 and 14.10) shows a significant heterogeneity that seems to be more important in the case of the phenolic groups. This fact could reflect a significant and varied substitution in the phenolic rings. In any event, these results clearly show a macromolecular chemical behavior of HSA as also reflected in the variation of the surface electrostatic potential during the titration process (Figure 14.8).

THEORETICAL BASIS |

137

Figure 14.21 Affinity spectrum (Gauss density function) for ′ (log KMint)

14.5.2. Metal Binding Study As for the metal-binding process considered as a whole, the clear differences between the values of the intrinsic and apparent stability constants indicate that significant electrostatic interactions among binding sites evolve during the complexation process (Figure 14.13). In this sense, if we analyze the variation of W as a function of the bound metal, clear conformational changes are observed (Figure 14.14). Thus, after a slight molecular expansion (a decrease in W values), a rapid and important molecular contraction (an increase in W values) is observed, followed by a similarly significant molecular expansion to reach finally a stable molecular conformation similar to that corresponding to the molecular system before starting the complexation process. In principle, the significant molecular contraction noted before may correspond to a decrease in the repulsive forces among negatively charged groups as a result of the complexation process. However, the final and stable molecular expansion leading to a molecular conformation similar to that of the native state at the titration pH (8.5) could indicate that the global balance of negative charges is only slightly modified. This fact also suggests that the percentage of acidic groups that intervene in the truly stable complexation is relatively small. These results also agree with the values of the intrinsic stability constants, which are very low in the medium and final part of the complexation process (for MB values up to 0.35 mM). In any event, these results confirm the clear macromolecular character of HSA already deduced from the proton-binding analysis. Likewise, this study manifests the suitability of W(MB) as an efficient tool to analyze the macromolecular behavior of a molecular system even when the analysis of e−′ (′ ,′ ) (MB) does not allow us to reach clear conclusions (Figure 14.15). On the other hand, significant differences between the values of the intrinsic stability constant obtained with or without consideration of the possible H-M competition process are not observed (Figure 14.16). This was expected on taking into account the pH value at which the complexation study was carried out (pH 8.5). Regarding the heterogeneity of the binding sites, the affinity spectrum indicates clear differences among the degrees of stability associated with the different binding sites. Likewise, the values of these intrinsic stability constants indicate that more stable inner-sphere metal complexes as well as less stable outer metal-complexes are probably formed during the whole binding study. If we analyze the results obtained in relation to the binding sites with higher affinity, we observe that as in the case of the proton-binding study there exist two specific binding-site sets that present higher intrinsic stability and lower apparent stability (Figure 14.18). This could be the consequence of both the involvement in these binding sites of those carboxylic groups situated in the inner part of specific regions of the molecule with a high density of carboxylic groups and the location of these binding sites in less accessible structural locations that are liberated, becoming more accessible as a result of the first molecular expansion indicated earlier. This molecular expansion is also reflected in the variation of both W(MB) and e−′ (′ ,′ ) (MB) as the complexation degree increases (Figures 14.19 and 14.20). As expected, the degree of heterogeneity in this selected group of binding sites is less important than in the case of the whole group of binding sites, but it is also significant (Figure 14.21). 14.6. CONCLUSIONS The results show the suitability of the EPM model to analyze the ion-binding process in humic systems. They explicitly explore the macromolecular character or functional heterogeneity and possible ion competition as well from a systematic study of the experimental data. In this sense, the fact that the complete analysis is carried out without assuming a specific

138

| CHAPTER 14: EXPLORING THE MOLECULAR CHARACTER OF HUMIC SUBSTANCES

molecular geometry or structural pattern makes EPM an objective and powerful tool to gain valuable, complementary information about the real structural nature and chemical behavior of HSs. ACKNOWLEDGEMENTS The author thanks M.David Rhymes for his help during the preparation of the English version of the manuscript, and Drs. Carolina Santamaria and Rodrigo G-Cantera for their help in conducting some experiments. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11. 12. 13. 14. 15. 16. 17.

18. 19. 20. 21.

Hayes MHB, Clapp CE. Humic substances: Considerations of compositions, as pects of structure, and environmental influences. Soil Sci., 2001; 166:723–737. Piccolo A. The supramolecular structure of humic substances: A novel understanding of humus chemistry and implications in soil science. Adv. Agronomy, 2002; 75:57–134. Simpson AJ. The structural interpretation of humic substances isolated from podzols under varying vegetation. Ph.D. dissertation, Birmingham: University of Birmingham, 1999. Burdon J. Are the traditional concepts of the structures of humic substances realistic? Soil Sci., 2001; 166:752–769. Swift RS. Macromolecular properties of soil humic substances: Fact, fiction, and opinion. Soil Sci., 1999; 164:790–802. Wershaw RL. Molecular aggregation of humic substances. Soil Sci., 1999; 164: 803–813. Chen Y, Senesi N, Schnitzer M. Information provided on humic substances by E4/E6 ratios. Soil. Sci. Soc. Am. J., 1977; 41:352–358. Benedetti MF, Van Riemsdijk WH, Koopal JK. Humic substances considered as a heterogeneous Donnan gel model. Environ. Sci. Technol., 1996; 30: 1805–1813. Kinniburgh DG, Van Riemsdijk WH, Koopal JK, Borkovec M, Benedetti MF, Avena MJ. Ion binding to natural organic matter: Competition, heterogeneity, stoichiometry and thermodynamic consistency. Colloids Surfaces A: Physicochem. Eng. Aspects, 1999; 151:147–166. Milne CHJ, Kinniburgh DG, Tipping E. Generic NICA-Donnan model parameters for proton binding by humic acids. Environ. Sci. Technol., 2001; 35: 2049–2059. Christl I, Kretzschmar R. Relating ion binding by fulvic and humic acids to chemical composition and molecular size. 1. Proton Binding. Environ. Sci. Technol., 2001; 35:2505–2511. Christl I, Milne CHJ, Kinniburgh DG, Kretzschmar R. Relating ion binding by fulvic and humic acids to chemical composition and molecular size. 2. Metal Binding. Environ. Sci. Technol., 2001; 35:2512–2517. Tanford CH. Physical chemistry of macromolecular. New York: Wiley, 1961. De Witt JCM, Van Riemsdijk WH, Nederlof MM, Kinniburgh DG, Koopal JK. Analysis of ion binding on humic substances and the determination of intrinsic affinity distributions. Anal. Chim. Acta, 1990; 232:189–207. Barak P, Chen Y. Equivalent radii of humic macromolecules from acid-base titration. Soil Sci., 1992; 154:184–195. García-Mina JM. Physicochemical description of the ion binding process in complex macromolecular systems with unknown or poorly characterized structural features. I. Theoretical development of the model. J. Phys. Chem. B, in in press. Perdue EM. Acidic functional groups of humic substances. In: Aiken GR, McKnight D, Wershaw RL, MacCarthy P eds. Humic substances in soil, sediment, and water: geochemistry, isolation, and characterization. New York: Wiley-Interscience, 1985: 493–526. Stevenson FJ. Humus chemistry: Genesis, composition, reactions. 2nd Edn. New York: Wiley-Interscience, 1994. Takamatsu T, Yoshida T. Determination of stability constants of metal-humic acid complexes by potentiometric titration and ionselective electrodes. Soil Sci., 1978; 125:377–386. Aplincourt M, Bee-Debras A, Prudhomme JC. Complexes du cuivre (II) avec des molecules. Modeles de synthese et implications pour l’interaction du cuivre avec la matiere organique des sols. Sci. du Sol., 1988; 26:157–168. Swift RS. Molecular weight, size, shape, and charge characteristics of humic substances: Some basic considerations. In: Hayes, MHB, MacCarthy, P, Malcolm, RL, Swift, RS eds. Humic substances II. In search of structure. New York: Wiley, 1989:449–466.

Chapter 15 STUDY OF FULVIC-ALUMINUM(III) ION COMPLEXES B Y 27AL SOLUTION NMR Norman C.Y.Lee and David K.Ryan Department of Chemistry, University of Massachusetts Lowell, Lowell, Massachusetts 01854, USA

15.1. INTRODUCTION Aluminum is the most abundant metallic element in the earth’s lithosphere and is a very strong neurotoxic agent, especially to children [1,2]. Aluminum also has been linked to Alzheimer’s disease in several studies [3–6]. The sources of aluminum in the environment can be categorized as natural and anthropogenic. The natural sources include volcanic activity and the weathering and erosion of rocks, while the major anthropogenic aluminum source is water-treatment. Alum (aluminum sulfate) is used extensively in water treatment plants as a flocculation agent to remove particulate matter and certain dissolved substances in drinking water. Aluminum’s ability to flocculate and coagulate has put alum into an essential role in the treatment of drinking water [7]. In addition to these sources of aluminum in the environment, there has long been concern about the effects of acid rain (~pH 5 and lower) on the mobility and toxicity of metals to organisms and ecosystems, since toxicity depends heavily on the form of the metal species present in the environment [8]. Metals like aluminum undergo strong hydrolysis that depends on the pH of the aqueous system [9]. Our previous work [10] indicates that aluminum exists as free and labile hexaquoaluminum(III), Al(H2O)63+ when the pH is below 5. AsthepH approaches 5, aluminum hydrolysis products start to form, namely Al(OH)aq2+ and Al(OH)2aq+. At about pH 7, a white precipitate of aluminum hydroxide, Al(OH)3 forms and upon further increase of the pH to 8, the aluminum hydroxide redissolves to form soluble aluminate anions Al(OH)4− inaqueous systems. Hence, this study is mainly concerned with systems at pH′ 5. Humic substances (HSs) greatly alter the mobility of metals under environmental conditions [11,12]. The toxicity of aluminum is determined by the species of aluminum present, and the most toxic form of aluminum is Alaq3+ ion, the free and labile species of aluminum. However, when aluminum is present with humic materials it is bound by the humics. This can reduce both the toxicity and availability of aluminum in the environment. Nuclear magnetic resonance (NMR) spectroscopy is a powerful tool when applied to humic material characterization [12]. It is non-destructive and is by far the least likely method to disturb the binding equilibria between humic material and metals [13,14]. Traditionally, 1H and 13C are the elements of choice when conducting NMR experiments on HSs. The data can give binding information from the humic material or ligand point of view. It is also possible to use multinuclear NMR probes to study humic-metal binding from the perspective of the metal. Therefore, this study focuses on using 27Al solution phase NMR to investigate binding between aluminum ions and humic material under acidic aqueous conditions. 15.2. MATERIALS AND METHODS The humic material used in this study was the fulvic acid (FA) isolated by Weber and Wilson [15] from an organic-rich B2 horizon Podzol soil from Conway, New Hampshire, USA. Its elemental composition is summarized in Table 15.1. Complexes of aluminum and fulvic acid were prepared by dissolving A1C13•6H2O (Baker) and fulvic acid in deionized water in polypropylene volumetric flasks. The mixture was stirred for 12–24 hrs after adjusting the pH to 3,4 and 5 with NaOH (0.01 M). Oxalic and salicylic acids were purchased from Fisher. NMR samples were prepared by filtering Al-FA solutions with a syringe filter (Corning, Cellulose Acetate Membrane 0.20 µm) before adding them to the NMR tube. Suprasil synthetic quartz NMR tubes were purchased from Wilmad. Two drops of D2O were added to provide a signal lock. Two NMR instruments were used in this study. Initially, NMR spectra were generated on a 250 MHz Bruker ARX, but the majority of the spectra were generated on a 500 MHz Bruker DRX solution instrument equipped with a 5 mm multinuclear broadband probe. The typical 27Al NMR parameters for the 500 MHz instrument are RG=30,720 and NS=15k at 300K. To minimize instrumental contributions to the aluminum background peak, a blank spectrum containing D2O was subtracted from the sample spectrum. By convention, all spectra generated in this study were referenced to AlCl3•6H2O at pH 2 in 20% D2O as an external

140

| CHAPTER 15: STUDY OF FULVIC-ALUMINUM(III) ION COMPLEXES

Figure 15.1 Al3+(135 mg/L)-FA(25,000 mg/L) at pH 3 generated with 15k scans on a 250 MHz NMR instrument

reference set to 0 ppm. Aluminum is a good candidate for NMR studies, because although the nuclear spin of aluminum is 5/2, which gives broad peaks, its gyromagnetic ratio (6.9704 radT−1s−1), Table 15.1 Elemental composition of soil fulvic acid (from Wilson and Weber [15]) FA

%C

%H

%N

% Ash

53.1

3.24

0.90

0.8

magnetic moment (3.6415 µ) and natural abundance (100%) make aluminum sensitive to magnetic fields. This allows signal detection to be possible with a reasonable number of scans. 15.3. RESULTS There are a number of potential problems implicit in 27Al NMR studies, namely background peaks and signal to noise ratios [16]. Figure 15.1 shows a spectrum generated on the 250 MHz instrument with an Al(III) ion concentration of 135 mg/L and a soil fulvic acid concentration of 25,000 mg/L. The broad peak centered at 76 ppm is a contribution from the NMR probe, coils, and tube. Besides the reference peak at 0 ppm, there are two Al-FA peaks that are barely visible in the spectrum between 0 and 20 ppm. These peaks are small and essentially useless for the measurement of aluminum species. Therefore, in order to work with low concentrations of aluminum that are more environmentally relevant, samples have to be run using an NMR instrument with a higher magnetic field strength than 250 MHz. Figure 15.2 shows NMR spectra for three different Al-FA concentrations at pH 3, 4 and 5. The chemical shifts are summarized in Table 15.2. Figure 15.2A shows that three peaks are observed at pH 3. At pH 4 and 5, peak intensities are shifted downfield and a new peak at 17 ppm is observed. At pH 3, two peaks are observed (Figure 15.2B). Upon increasing the pH to 4 and 5, a peak at 12 ppm is observed, while the intensities of peaks at 0 and 7 ppm diminish. Very similar chemical shift and peak behavior, despite the noisy background, are also observed for a lower concentration Al-FA system in Figure 15.2C. Using oxalic acid (Ox) as the aluminum binding ligand gives similar results to those observed for FA. Figure 15.3 shows 27Al NMR spectra for Al-Ox complexes at two metal-to-ligand molar ratios. The chemical shift data are summarized in Table 15.3. Figure 15.3A shows three peaks at pH 3. When the pH is increased to 4, the peak at 0 ppm decreases in intensity and upon further increase of the pH to 4.8 the peak disappears completely while the downfield peak at 12 ppm grows in intensity. As the concentration of the oxalic acid ligand increases, more bound aluminum species are expected. This is what is evident in Figure 15.3B: peaks at 7 ppm and 12 Table 15.2 Chemical shifts for three different concentrations of Al-FA complexes at pH 3, 4 and 5 pH 3 6.7 ppm 12.2 ppm pH 4 6.8 ppm 12.1 ppm 16.8 ppm pH 5 12.0 ppm 16.7 ppm

Al3+(135 mg/L)-FA(12,500 mg/L)

Al3+(135 mg/L)-FA(5,000 mg/L)

Al3+(27 mg/L)-FA(l,000 mg/L)

0.8 ppm 6.9 ppm

0.7 ppm 6.9 ppm

0.6 ppm

0.4 ppm 7.1 ppm 12.4 ppm

1.0 ppm 6.6 ppm

−0.7 ppm

−1.2 ppm 7.0 ppm 12.2 ppm

1.0 ppm 12.1 ppm

−1.0 ppm

RESULTS |

141

Figure 15.2 500 MHz 27Al NMR spectra of (A) Al3+(135 mg/L)-FA(12,500 mg/L), (B) Al3+(135 mg/L)-FA(5,000 mg/L) and (C) Al3+(27 mg/L)-FA(1,000 ,000 mg/L) at pH 3, 4 and 5

ppm are observed and the 0 ppm peak is essentially non-existent at pH 3. At pH 5 a new downfield peak appears at 17 ppm. Table 15.3 Chemical shifts of two molar ratios of Al-Ox complexes at 1 mM: 1mM and 1mM: 2mM respectively pH 3 6.6 ppm 12.2 ppm pH 4 6.6 ppm 12.3 ppm pH 5 12.3 ppm

Al3+:Ox=1:1

Al3+:Ox=1:2

0.3 ppm 12.0 ppm

6.6 ppm

0.9 ppm 12.0 ppm

6.5 ppm

7.2 ppm 12.0 ppm 16.7 ppm

6.8 ppm

15.4. DISCUSSION The 250 MHz NMR spectrum of Al3+-FA shown in Figure 15.1 has a broad and intense peak at 76 ppm, which is contributed by the NMR probe, coils and tubes [17]. All commercially available probes and coils are made of metal alloys that contain a significant amount of aluminum metal and aluminum oxides. Three other peaks are barely visible in the spectrum: the reference peak is observed at 0 ppm and the other two peaks are the Al-FA peaks of interest. The broad instrumental background peak is much more intense than the Al-FA peaks. Three methods have been adopted to minimize the interference from the background. One method is to study samples with high aluminum concentration, which is important theoretically but is not environmentally relevant and limits the applications of the technique. Another method is to employ a pulse sequence to filter the broad peak, but it is very difficult to filter out a peak as broad as the one shown in Figure 15.1, which is more than 100 ppm in width. The last method, which is used most frequently by researchers, is to subtract a blank spectrum from the sample spectrum. This method works well if the peaks are relatively intense. However, this method can introduce errors in the peak intensity when incomplete subtraction occurs. For the rest of the discussion we shall concern ourselves with spectra generated on the 500 MHz instrument, which lowered the detection limit for aluminum. Figure 15.2 shows spectra for three different concentrations of Al-FA samples at pH 3, 4 and 5. The chemical shifts are summarized in Table 15.2. Figure 15.2A shows spectra of Al3+(135 mg/L)-FA(12,500 mg/L)

142

| CHAPTER 15: STUDY OF FULVIC-ALUMINUM(III) ION COMPLEXES

Figure 15.3 500 MHz 27Al NMR spectra of Al-oxalic acid complexes, (A) Al3+(1 mM)-Ox(1 mM) and (B) Al3+(1 mM)-Ox(2 mM)

complexes. As pH is increased from 3 to 5 a shift of peak intensity is observed. Peaks at 0 and 7 ppm decrease in intensity since there is less and less free Al3+ ion in the system, while the 17 ppm peak gains intensity and the peak at 12 ppm stays relatively constant. Also note the broadening of the peak at 0 ppm that resulted from the formation of hydrolysis products as pH is increased from 3 to 4 to 5. Figure 15.2B shows spectra of Al3+(135 mg/L)-FA(5,000 mg/L) complexes. Again, as pH is increased from 3 to 5, a broadening and a decrease in intensity is observed for peaks at 0 and 7 ppm, while the downfield peak grows in intensity. Figure 15.2C shows spectra of Al3+(27 mg/L)-FA(1,000 mg/L) complexes. Since the aluminum concentration is low, the noise level is higher than in the other two spectra. However, the peaks at 0, 7 and 12 ppm can still be observed at pH 4. Similarly, 0 and 7 ppm peaks decrease in intensity and the peak at 12 ppm grows as pH is increased. Note that the sample used to obtain the data in Figure 15.2C was 5 times more dilute than the sample used for Figure 15.2B. The spectra show very similar peak shape and relative intensity between peaks. This suggests a consistency in the binding environment of the fulvic acid. Numerous aromatic and aliphatic carboxylic acid compounds have been studied by researchers as model compounds for humic materials [18]. When salicylic acid was studied with aluminum, only two peaks (0 and 3 ppm) were observed (not shown). However, oxalic acid showed four peaks with aluminum. Figure 15.3 shows Al-Ox complexes at pH 3, 4 and 5 at two different molar ratios. The chemical shifts are summarized in Table 15.3. In Figure 15.3A, as pH was increased from 3 to 4, the peak at 0 ppm decreased in intensity and broadened. At pH 4.8, the peak at 0 ppm disappeared completely and the peak at 12 ppm gained intensity. In Figure 15.3B ox alic acid was in excess, so that no free Al3+ was observed, as indicated by the complete absence of a 0 ppm peak. Similarly, as pH increased from 3 to 4 to 5, the peak at 7 ppm decreased in intensity while the 17 ppm peak grew in intensity. These chemical shifts and the peak intensity behavior are analogous to the Al-FA system, which suggests that the Al3+ binding site is best described by an oxalic acid type of environment in this fulvic acid sample [19]. It has been reported [20] that aluminum (III) ions form octahedral complexes with oxalic acid. Since oxalic acid is a dicarboxylic acid and a bidendate ligand, one oxalic acid replaces two water molecules on the aquo aluminum ion and a total of three oxalic acids can bind to one aluminum (III) ion. At this time, the data are too inconclusive to determine if the oxalate type of environment around the aluminum(III) ion is created by a single fulvic acid molecule that rearranges its functional groups to bind with aluminum or if the aluminum ion is bound with two or three separate fulvic acid molecules. According to NMR theory, deshielded nuclei are under a stronger influence of the applied magnetic field compared to more shielded nuclei. Hence, they appear in the downfield region of the spectrum. The degree of shielding is a direct indication of the electron density around a nucleus [21]. Since oxalate groups are electrophilic, the electron density on an aluminum nucleus decreases as the number of oxalate ligands attached to the aluminum nucleus increases. Therefore, the peaks shown in Figure 15.3 at 7, 12 and 17 ppm are believed to be AlOx+, AlOx2− and AlOx33− respectively. The computer speciation model MINTEQ [22] was employed to calculate aluminum speciation with the oxalate ligand and compare results with data generated by integrating the NMR peaks in Figure 15.3. The results of the comparison are summarized in Tables 15.4 and 15.5 (MINTEQ values are in parentheses and stability constants are from Sitler and Martell [23]). Table 15.4 shows the data

RESULTS |

143

for equimolar aluminum ion and oxalate. At the these pH values MINTEQ (with over-saturation precipitation allowed) gave results that are in reasonable agreement with the measured NMR results at pH 3 and 4 for free Al3+, AlOx+ and AlOx2− complexes. AtpH 4.8, MINTEQ predicted 10%, 64% and 19% for Al3+, AlOx+ and AlOx2− species, respectively, while the NMR data showed no free Al3+, 53% for the AlOx+ complex and 47% for the AlOx2− complex. However, if the precipitation option is disallowed in MINTEQ, for the pH 4.8 case, instead of having AlOx+ as the dominating species, MINTEQ predicts 4%, 86% and 10% for AlOx+, AlOx2− andAlOx33− complexes, respectively. The NMR data in Table 15.4 at pH 4.8 fall between the over-saturation allowed case and the disallowed case, which suggests that the solid form of Al(OH)3 was present in our sample although no visible white precipitate was observed. Trace amounts of Al(OH)3 mayhave formed and may have been measured, appearing under the broad background peak around 76 ppm, but were undetectable in the presence of the large instrumental background. Similar results were also obtained using another computer speciation model, GEOCHEM [24]. The predicted results also agreed with experimental NMR data when the oxalate is twice as concentrated as the aluminum ion. As shown in Table 15.5, MINTEQ predicted very similar percentages at all three pH values (pH 3, 4 and 5) shown on the NMR spectra. These results not only reconfirmed the assignments of the Al-Ox peaks, but also provided another piece of evidence that NMR is a powerful tool in conducting metal speciation studies. Table 15.4 Percentage of observed and predicted Al-Ox speciation (MINTEQ values) Al3+:Ox=1:1

Al3+

AlOx+

AlOx2−

pH 3.0 pH 4.0 pH 4.8

14 (15) 12 (14) 0(10)

81 (69) 77 (69) 53 (64)

5 (16) 12 (17) 47 (19)

AlOx33−

Table 15.5 Percentage of observed and predicted Al-Ox speciation (MINTEQ values) Al3+:Ox=1:2 pH 3.0 pH 4.0 pH 5.0

Al3+

AlOx+

AlOx2−

AlOx33−

22(16) 21 (7) 7 (2)

76 (81) 77 (87) 86 (76)

2 (3) 2(6) 7 (22)

15.5. CONCLUSIONS Nuclear magnetic resonance instrument hardware interferes with 27Al experiments. The metal alloy probe and coil produce a broad and intense peak around 76 ppm that limits our ability to study aluminum at a concentration lower than the mg/L range. The soil fulvic acid sample that was used has a binding site for aluminum ion that is best described by an oxalate type of environment. Three aluminum-soil fulvic acid species were observed. They were concluded to be aluminum ion bound by one oxalate, two oxalate and three oxalates type ligands with configurations that give rise to peaks at 7, 12 and 17 ppm, respectively. Currently, we are quantifying each of the peaks using an external standard so that we can determine the quantity of each aluminum species in the fulvic acid system. Also, we have conducted NMR studies at lower temperature to ensure the identity of each peak. We are still interpreting our data. In the future we would like to compute binding constants and be able to use these parameters in computer speciation models, such as MINTEQ to better predict speciation of aluminum. We would also like to investigate a variety of different fulvic acids isolated from different parts of the world to determine if aluminum binds similarly or differently with other fulvic acids. ACKNOWLEDGEMENT The authors thank the University of Massachusetts Lowell Research Foundation for financial support and give special thanks to Xiao-Dong Wu and Jeffery Njus for helpful discussions. REFERENCES 1. 2. 3. 4.

Klein GL. Metabolic bone disease of total parenteral nutrition. Nutrition, 1998; 14:149–152. Harris WR, Berthon G, Day JP, Exley C, Flaten TP, Forbes WF, Kiss T, Orvig C, Zatta PF. Speciation of aluminum in biological systems. J. Toxicol. Environ. Health, 1996; 48:543–568. Exley C. A molecular mechanism of aluminum-induced Alzheimer’s disease? J. Inorg. Biochem., 1999; 76:133–140. Makjanic J, McDonald B, Chen CP, Watt F. Absence of aluminium in neurofibrillary tangles in Alzheimer’s disease. Neurosci., 1998; 240:123–126.

144

| CHAPTER 15: STUDY OF FULVIC-ALUMINUM(III) ION COMPLEXES

5.

Solomon B, Koppel R, Jossiphov J. Immunostaining of calmodulin and aluminium in Alzheimer’s disease-affected brains. Brain Res. Bull., 2000; 55: 253–256. Flaten TP. Aluminum as a risk in Alzheimer’s disease, with emphasis on drinking water. Brain Res. Bull., 2001; 55:187–198. Dempsey BA. Reaction between fulvic acid and aluminum: Effects on the coagulation process. In: Suffet IH, MacCarthy P eds. Aquatic humic substances: Influence on fate and treatment of pollutants. Washington DC: Library of Congress Cataloging-in-Publication Data, 1989:409–424. Lambert J, Buddrus J, Burba P. Evaluation of conditional stability constants of dissolved aluminum/humic substance complexes by means of 27Al nuclear magnetic resonance. Fres. J. Anal. Chem., 1995; 351:83–87. Bertsch PM, Barnhisel RI, Thomas GW, Layton WJ, Smith SL. Quantitative determination of aluminum-27 by high-resolution nuclear magnetic resonance spectrometry. Anal. Chem., 1986; 58:2583–2585. Ryan DK, Shia CP, O’Conner DV. Fluorescence spectroscopic studies of Al-fulvic acid complexation in acidic solutions. In: Gaffney JS, Marley NA, Clark SB eds. Humic and fulvic acids: Isolation, structure, and environmental role, Washington DC: American Chemical Society, 1996:125–139. Karweer SB, Mhatre SN, Pillai BP, Lyer RK, Moorthy PN. Studies on speciation of aluminum complexes with malic acid by high field 27Al NMR spectroscopy. Ind. J. Chem., 1993; 32A:502–505. Francioso O, Sanchez-Cortes S, Tugnoli V, Ciavatta C, Gessa C. Characterization of peat fulvic acid fractions by means of FT-IR, SERS, and 1H, 13C NMR spectroscopy. Appl. Spectrosc., 1998; 52:270–277. Yokoyama T, Abe H, Kurisaki T, Wakita H. 13C and 27Al NMR study on the interaction between aluminum ion and iminodiacetic acid in acidic aqueous solutions. Anal. Sci., 1999; 15:393–395. Bi SP, Yang XD, Zhang FP, Wang XL, Zou GW. Analytical methodologies for aluminum speciation in environmental and biological samples—a review. Fres. J. Anal Chem., 2001; 370:984–996. Weber JH, Wilson SA. The isolation and characterization of fulvic acid and humic acid from river water. Water Res., 1975; 9: 1079–1084. MacFall JS, Ribeiro AA, Cofer GP, Dai KH, Labiosa W, Faust BC, Richter DD. Design and use of background-reduced 27Al NMR probes for the study of dilute samples from the environment. Appl. Spectros., 1995; 49:156–162. Faust BC, Labiosa WB, Dai KH, MacFall JS, Browne BA, Ribeiro AA, Richter DD. Speciation of aqueous mononuclear Al(III)hydroxo and other Al(III) complexes at concentrations of geochemical relevance by aluminum-27 nuclear magnetic resonance spectroscopy. Geochim. Cosmochim. Acta, 1995; 59: 2651–2661. Hook J, Lu XQ, Howe RF. A 27Al NMR study of aluminum: Humic substances interactions. Bull. Mag. Reson., 1995; 18:186–187. Howe RF, Lu XQ, Hook J, Johnson WD. Reaction of aquatic humic substances with aluminum: A 27Al NMR study. Mar. Freshwater Res., 1997; 48:377–383. Kerven GL, Larsen PL, Bell LC, Edwards DG. Quantitative 27Al NMR spectroscopic studies of Al(III) complexes with organic acid ligands and their comparison with GEOCHEM predicted values. Plant Soil, 1995; 171:35–39. . Silverstein RM, Webster FX. Spectrometric identification of organic compounds. New York: Wiley, 1998. Allison JD, Brown DS, Novo-Gradac KJ. MINTEQA2/PRODEFA2, A geochemical assessment model for environmental systems: Version 3.0 user’s manual. U.S. Environmental Protection Agency, Athens, GA. EPA/600/3–91/021. Sitler LG, Martell AE. Stability constants. London: Metcalfe & Cooper Ltd, 1964. Sposito G, Mattigood SV. GEOCHEM: A computer program for the calculation of chemical equilibria in soil solutions and other natural water systems. The Kearney Foundation of Soil Sciences, University of California, Riverside, USA, 1980.

6. 7.

8. 9. 10.

11. 12. 13. 14. 15. 16. 17.

18. 19. 20. 21 22. 23. 24.

Chapter 16 INVESTIGATION OF COLLOIDAL PROPERTIES AND TRACE METAL COMPLEXATION CHARACTERISTICS OF SOIL-DERIVED FULVIC ACIDS BY FLOW FIELD-FLOW FRACTIONATIONINDUCTIVELY COUPLED PLASMA-MASS SPECTROMETRY (FLOW FFF-ICP-MS) Jonathan Bell,1 Dula Amarasiriwardena,1 Atitaya Siripinyanond,2 Baoshan Xing3 and Ramon M.Barnes2, School of Natural Science, Hampshire College, Amherst, MA 01002, USA 2Department of Chemistry, University of Massachusetts, Amherst, MA 01003–9336, USA 1 Department of Plant and Soil Sciences, University of Massachusetts, Amherst, MA 01003, USA University Research Institute for Analytical Chemistry, 85 North Whitney St., Amherst, MA 01002–1869, USA 16.1. INTRODUCTION Fulvic3 acids (FAs) play an important role in the sequestration and transport of metals in the environment by complexing with metal ions [1]. Physico-chemical information about FA-bound metal ions and their mobility is therefore important in 4environmental investigations. Fulvic acids are low molar mass, heterogeneous organic molecules present in aqueous and subsurface soil environments. They are soluble in water under all pH conditions and known to be rich in metal binding, oxygencontaining functional groups like carboxylic, phenolic and aliphatic alcoholic groups [1– 4]. Among these functional groups, the carboxylate group is particularly responsible for FA-metal binding either by ion exchange, thus forming fulvates (RCOO− M+), or by the formation of chelated metal complexes using active bidentate oxygen ligand sites in carboxylate anions, RCOO − [3]. High performance size exclusion chromatography (HPSEC) [5], capillary electrophoresis (CE) [6] and methods based on colligative properties of humic substances (for example, viscometry [1,3]) are often used to determine the molar masses of fulvic acids. Different methods give a range of molar masses depending on the analytical approach and the molar mass markers used for method calibration. Each analytical approach has advantages and limitations. Trace metals associated with dif ferent aqueous humic acid (HA) molar mass fractions were previously investigated by on-line HPSEC coupled to inductively coupled plasma-mass spectrometry (ICP-MS) [7,8]. Application of HPSEC-ICP-MS for qualitative determination of “tightly” bound trace metals in soil derived humic acids molecular fractions was demonstrated earlier [9,10]. Flow field-flow fractionation (flow FFF) is a mild separation technique capable of fractionating macromolecules and colloids in the range from 1 to 0.001 µm. Detailed FFF theory and the principles of fractionation, FFF sub-techniques [11–16] and instrumentation are discussed elsewhere [14–18]. Macromolecular size fractionation is based on the molecular diffusion coefficients in the carrier medium in a long, ribbon-like hollow channel, the floor of which is a semi-permeable membrane. Two forces are applied perpendicular to each other: a cross flow that enters through the channel ceiling and exits through the membrane floor and the second flow, called channel flow, runs parallel to the membrane axis. Typically, a small aliquot of a sample such as the fulvic acids under investigation is injected into the fractionation channel. Channel flow is momentarily stopped, and the cross flow is introduced. This cross flow facilitates the accumulation of macromolecules against the membrane, which is impermeable to macromolecules larger than the membrane molar mass cutoff but lets the cross flow fluid pass through. During this relaxation period, steady state equilibrium is achieved between the cross flow driving force and macromolecular back diffusion corresponding to the field induced velocity and the molecular diffusion coefficients. This leads to a diffusion gradient in which the small macromolecules are located away from the membrane wall while large macromolecules are positioned close to the wall. When the parabolic linear channel flow is reintroduced, the smaller molecules caught up in the high velocity region of the laminar channel flow are flushed out of the channel first. Increasingly larger molecules are driven out at longer emergence times, and the largest mass fraction exits the channel last. Often, the emergence time and composition of these various size fractions are determined by placing a UV-visible detector at the downstream end of the channel flow. As a result, a UV absorbance vs. emergence time plot can be obtained, and this plot is commonly called a fractogram. Experimentally, the diffusion coefficient is directly proportional to the emergence time. As discussed previously, the separation is based on macromolecular diffusion away from the accumulation wall, and the diffusion coefficient (D) is inversely related to the molar masses (M) of the macromolecules being fractionated [11–16], Eq. 16.1, (16.1) where A is a constant for a given molecule in the carrier solution, and c is a constant that depends on the conformation of the

146

| CHAPTER 16: INVESTIGATION OF COLLOIDAL PROPERTIES AND TRACE METAL COMPLEXATION

macromolecule in the carrier solution. Experimentally, the emergence time (tr) of the analyte species is directly proportional to the diffusion coefficient (D). Thus, a calibration function obtained using the emergence times of a set of molar mass standards can be used to calculate unknown molar masses. In contrast, macromolecular hydrodynamic diameter (dh) information can be obtained directly from the Stokes-Einstein relationship using the retention time (tr) for well-retained fractions in normal mode flow FFF, Eq. 16.2, (16.2) where k is Boltzman’s constant, T is the absolute temperature, is the viscosity of the carrier liquid, w is the channel thickness, Vc is the cross flow rate and V is the channel flow rate. Elimination of the solid supports used in chromatography and with mild physical forces encountered in flow, the flow FFF technique minimizes solute adsorption, charge repulsion effects and potential denaturation of large macromolecules observed in SEC. This relatively new separation technique offers information about colloidal particle diffusion characteristics, hydrodynamic diameters and molar mass information [14–17]. Flow-FFF has been used to determine molar mass distributions of soil and aqueous humic [15,16,18,19] and aqueous fulvic acids [15,20]. Information from Flow-FFF studies can provide valuable insights into transport and mobilization of humic substancebound, nutrient and toxic trace metals in the environment. Inductively coupled plasma-mass spectrometry (ICP-MS) is a sensitive element detector capable of trace metal quantification and ideal for identification of trace metals bound to fractionated macromolecules like FAs when combined with flow FFF. The concept and first experiments demonstrating the interfacing of FFF with ICP-MS were described by Beckett [14] and Taylor et al. [21] for the characterization trace elements bound to river sediments and clay minerals like goethite. Later, the method was applied to analysis of trace metals bound to biological macro molecules [17] and environmental colloids [22,23]. Multielement trace metal analyses by flow FFF-ICP-MS of soil-derived humic acids, colloids in natural waters, sediments, and simultaneous determination of their size distributions were demonstrated recently [18,19,24]. The purpose of this investigation was to identify trace metals (i.e., Al, Cu, Fe, Ti, Pb and Zn) complexed to soil-derived fulvic acids by flow FFF-ICP-MS. 16.2. MATERIAL AND METHODS 16.2.1. Materials Fulvic Acids (FAs). Soil derived FAs were extracted from soil plots at the Massachusetts Agriculture Experimental Station Farm, South Deerfield, MA. The fine sandy loam soil is low in soil organic matter (~2%) and typical of Connecticut River Valley soils in Massachusetts, which are extensively used in growing a variety of cash crops [25,26]. Cover crops rye or hairy vetch were grown with or without nitrogen fertilizer in different soil test plots. The site description and experimental design of these cover crop experiments were discussed previously [25,26] and are summarized in Table 16.1. A detailed description of the extraction, fractionation and purification of FAs from these soils is given by Ding et al. [26]. Table 16.1 Fulvic acid samples origins FA sample

Nitrogen fertilizer added to soil

Type of cover crop

FAVR1 FAVR4 FARA1 FARA4 FAC1 FAC4

None Used (202 kg N/ha) None Used (202 kg N/ha) None Used (202 kg N/ha)

Hairy Vetch+Rye (46+65 kg/ha) Hairy Vetch+Rye (46+65 kg/ha) Rye (125 kg/ha) Rye (125 kg/ha) None None

16.2.2. Methods Fulvic Acid Characterization. Five agricultural soil-derived FA samples were characterized by UV-visible spectroscopy for the computation of E4/E6 ratios. Each sample was prepared by dissolving 0.2 to 0.4 mg of FA in 10 mL of 0.05 M sodium bicarbonate solution. A diode array spectrophotometer (HP 8542, Hewlett Packard Instruments, Fall River, DE) was used to measure UV-visible spectra. The ratios E4/E6 of absorbances at 465 and 665 nm were calculated by taking the average of absorbance at 464 and 466 nm and at 664 and 666 nm, respectively. Diffuse reflectance Fourier transform infrared spectroscopy (DRIFTS) was used to identify the major metal complexing functional groups. Both the sample and KBr were ground to a fine powder using an agate mortar and pestle. A mixture of approximately 3 mg of FA and 97 mg KBr was

MATERIALS AND METHODS |

147

prepared and placed in the sample compartment. A blank of spectroscopic grade powdered KBr was likewise prepared and placed in the blank position. The sample compartment was flushed with dry N2 for five minutes to remove CO2. The DRIFTS spectra were obtained with a Midac 2000 M series FTIR spectrometer (Midac Instruments, Irvine, CA). Normally, 100 scans at a resolution of 16 cm−1 were collected for each spectrum. Spectra were analyzed using the Grams-32 software package (Galactic Corporation, Salem, NH) and Microsoft Excel spreadsheet software (Excel-2002, Redmond, Washington). Flow Field-Flow Fractionation-Inductively Coupled Plasma-Mass Spectrometry (Flow FFF-ICP-MS). Poly(styrene sulfonate) (PSS) molecular standards of 2,4 and 7 kDa (American Polymer Standards, Mentor, OH) and 30 mM tris (hydroxymethyl)aminomethane-nitric acid (TRIS/HNO3) buffer mobile phase (pH 7.4) were used to calibrate the flow FFF channel. Fulvic acid samples (10 mg) were completely dissolved in 30 mM TRIS/HNO3 buffer and a 40-µL aliquot of sample was injected into the flow-FFF system (Model F-1000-FO, FFFractionation LLC, Salt Lake City, UT) coupled to an ICP-MS system (Sciex/Elan 5000a, Perkin-Elmer Instruments, Shelton, CT). Details of this instrument were given previously [17,18,24]. The FFF channel dimensions were 27.7 cm long, 2.0 cm wide and 0.0254 cm thick and fitted with a 1 kDa molar mass cut-off (MWCO) polyregenerated cellulose membrane. The channel flow rate and the cross flow rate were controlled by HPLC pumps (Model L-6010, Hitachi Instruments, Stoughton, MA and Model 300, Scientific Systems, State College, PA, respectively). The absorbances of fractionated FAs were measured at 254 nm with a Hitachi Model L-4000 UV-detector (Hitachi Instruments, Stoughton, MA). The flow-FFF-ICP-MS element detector measured the intensities of 27Al, 63Cu, 54Fe, 208Pb, 48Ti and 64Zn associated with fractionated FA fractions. Details of the flow-FFF and ICP-MS interface were described previously [17,18,24]. The UV-fractogram and ion intensity signals for analyte elements vs. time of the ion fractograms were converted to spreadsheet format with Microsoft Excel® for further analysis. Table 16.2 summarizes the operational conditions used for flow FFF-ICP-MS. Acid Digestion of Fulvic Acids for Elemental Analysis. Fulvic acid samples were digested with nitric acid for total trace metal analysis. The FA samples (50–100 mg) were transferred to the acid cleaned quartz vessels of a high pressure asher (HPA) (Anton PAAR KG, Graz, Austria) and 2.0 mL of sub-boiled nitric acid (Optima grade, Fisher Scientific, Fairlawn, NJ) was added in a clean room. The quartz vessels containing FA and nitric acid were digested in the asher at 230°C and 133 bar for 90 minutes. The digested fulvic acids were transferred to polypropylene tubes and subsequently analyzed by ICP-AES (Model Optima 2000 DV Perkin- Elmer Instru Table 16.2 Fractionation conditions and ICP-MS instrument operating parameters for flow FFF-ICP-MS Flow FFF conditions (Model F-1000-FO, Salt Lake City, UT) FFF Normal Mode (FFFractionation, Salt Lake City, UT) FFF channel dimensions/cm 27.7×2.0×0.0254 Carrier liquid 30 mM TRIS/HNO3 (pH 7.4) Cross flow rate/ml min−1 2.0 −1 Channel flow rate/ml min 1.0 Equilibration time/min 1.5 UV wavelength/nm 254 Membrane 1k Da MW cut-off polyregenerated cellulose mem brane ICP-MS instrument settings and operating parameters (Perkin Elmer Instruments, PE-Sciex/Elan -5000a) RF generator Frequency/MHz 40 RF forward power/W 1000 Torch Sciex, short Spray chamber Ryton® Scott-type Nebulizer Cross-flow (Perkin Elmer) Nebulizer, gas flow rate/ L min−1 0.98 −1 Intermediate gas flow rate/ L min 0.96 Outer gas flow rate/ L min−1 14.5 Resolution 1±0.1 at 10% peak max. Measurements per peak 1 Dwell time /ms 250 27Al, 63Cu, 57Fe, 48Ti, 208Pb and 64Zn Isotope monitored (m/z)

148

| CHAPTER 16: INVESTIGATION OF COLLOIDAL PROPERTIES AND TRACE METAL COMPLEXATION

Figure 16.1 DRIFT spectra of soil derived fulvic acid. Refer to the description of the FA sample codes in Table 16.1

ments) for Cu (327.393 nm), Fe (238.204 nm), Zn (206.200 nm) and Al (396.153 nm) and by ICP-MS for lead (208Pb) and arsenic (75As). The calibration functions were obtained for ICP-AES and ICP-MS with 10, 5, 2, 1, 0.5 mg L−1 Al, Cu, Fe and Zn, and 100, 10, 1, 0.1 µg L−1 As and Pb multielement solutions in 2% (v/v) sub-boiled nitric acid prepared by diluting individual elemental standards (Spex Certiprep, Metuchen, NJ). 16.3. RESULTS 16.3.1. Spectroscopic Data for Fulvic Acids The E4/E6 values of all the investigated FAs ranged from 8.1 (FAB4—Rye cover crop with fertilizer) to 8.9 (FAB1-Rye without fertilizer), which are typical values for soil FAs [27]. The DRIFT spectra of the FA samples (Figure 16.1) were similar to those reported in the literature [28–30]. The broad peak in the 3330–3000 cm−1 region is due to phenolic-OH stretching or to OH functional groups from undissociated water and aromatic CH stretches. A small hump at~2600 cm−1 results from the OH stretch of H-bonded –COOH groups. A prominent peak for all FAs is the>C=O stretch of -COOH appears at 1716–1719 cm−1. Inaddition, the 1610 cm−1 band of aromatic >C=C< stretch or asymmetric -COO- stretch, the 1415 cm−1 CHdeformation of CH3 and -CH bending of CH3 groups, the 1212 cm−1 band (due to>C=O stretch and OH deformation of -COOH), and the 1070 cm−1 C-C stretch of aliphatic groups were prominent in all the fulvic acid DRIFT spectra. The peaks in the low wavenumber region are mostly due to aromatic CH out of plane bending or to organically bound mineral phases [28–30]. Table 16.3 Hydrodynamic diameters and molar masses determined by flow FFF FA Sample

Hydrodynamic diameter (nm) dh at peak maxa

FAVR1 2.01 FAVR4 2.01 FAR1 2.01 FAR4 2.01 FAC1 2.01 FAC4 2.01 a Average obtained from duplicate flow FFF runs.

Molar Mass (Da), (Mp)a 920 920 920 920 870 870

16.3.2. Flow-Field Flow Fractionation Analyses A flow FFF fractogram is a plot of UV signal as a function of emergence time (tr), as shown in Figure 16.2 for sample FAC1. The molar masses and hydrodynamic diameters at peak maximum obtained for the soil derived FAs are listed in Table 16.3.

MATERIALS AND METHODS |

149

Figure 16.2 Typical UV-fractogram of a fulvic acid sample (FACl)

Figure 16.3 Fractograms of FAVR4 (with cover crops and fertilizer treatments) fulvic acid. (top left) frequency distribution of hydrodynamic diameters; (top right) Frequency distribution of molar masses; (bottom left) ion fractograms: ion intensity of 63Cu, 64Zn, 208Pb vs. hydrodynamic diameter; (bottom right) ion intensity of 63Cu, 64Zn’ 208Pb vs. calculated molar mass

Transformation of raw FFF fractograms into molar mass or hydrodynamic diameter distributions is described in the literature [22,31,32]. The time axis of the fractogram can be transformed into hydrodynamic diameter with Eq. 16.2. With the linear calibration function obtained from logarithmic tr vs. logarithmic molar mass plots, the retention time of each sample was converted to molar mass. The UV signal is directly proportional to fulvic acid concentration. The analyte signals were measured at small time interval slices (dm/dt) and transformed into a frequency function of hydrodynamic diameter (dm/dd) or molar mass distributions (dm/dM) at fixed digitized intervals with Eqs. 16.3 and 16.4, where dt is the emergence time difference for successive (16.3) (16.4) digitized points and dd and dM are the corresponding differences in the hydrody namic diameter and molar masses for those points, respectively [22,31,32]. Fractograms of molar mass and hydrodynamic diameter distributions of fulvic acid extracted from vetch and rye grown soil with nitrogen fertilizer (FAVR4) and fulvic acid derived from the control soil (FAC1, no fertilizer or cover crops used) are shown in Figures 16.3a,b and 16.4a,b, respectively.

150

| CHAPTER 16: INVESTIGATION OF COLLOIDAL PROPERTIES AND TRACE METAL COMPLEXATION

Figure 16.4 Fractograms of FAC1 (without treatments) fulvic acid. (top left) frequency distribution of hydrodynamic diameters; (top right) frequency distribution of molar masses; (bottom left) ion fractograms: ion intensity of 63Cu, 64Zn, 208Pb vs. hydrodynamic diameter; (bottom right) ion intensity of 63Cu, 64Zn, 208Pb vs. calculated molar mass

16.3.3. Elemental Analyses Total trace metal concentrations in the digested fulvic acid samples are shown in the Table 16.4. 16.3.4. Investigation of Trace Metals Complexed to Fulvic Acids by Flow FFF-ICP-MS Ion fractograms of soil-derived fulvic acid were obtained for 27Al, 63Cu, 54Fe, 48Ti, Table 16.4 Elemental concentrations in soil derived fulvic acids Elemental Concentrations (µg/g) FA Sample

Cu

Zn

Fe

FAVR1 407 7 525 FAVR4 143 8 788 FAR4 387 3 542 FAC1 289 1 511 FAC4 315 3 475 Note: FAR1 was not analyzed due to limited sample size available 208Pb

Mn

Pb

Al

As

40 129 48 38 49

11 2.3 9.9 4.5 6.2

1477 1094 1534 1500 1287

1.7 .3 1.9 1.6 2.0

and 64Zn using flow FFF-ICP-MS. The ion fractograms (63Cu, shown in Figures 16.3c,d (FAVR1) and 16.4c,d (FAC1).

208Pb, 64Zn)

of size and molar mass distributions are

MATERIALS AND METHODS |

151

16.4. DISCUSSION 16.4.1. Spectroscopic Characterization of Fulvic Acids The E4/E6 ratios obtained from UV-visible spectra are consistent within the values reported by Stevenson (6.0–8.5) [1] and by Chen et al., (8.4–8.9) [27] for soil fulvic acids at various concentrations (100–500 mg/L). The DRIFTS peak at ~1720 cm−1 in all fulvic acid spectra is quite intense and indicates an abundance of -COOH groups in the fulvic acids investigated. Carboxylic groups are important cation exchange and ligand sites for trace metal complexation. 16.4.2. Flow Field-Flow Fractionation of Fulvic Acids All fulvic acid samples exhibited monomodal fractograms with a skewed distribution toward large hydrodynamic diameters or molar masses. The average apparent molar mass at peak (Mp) for these fulvic acids was 920 Da except for the FAs derived from soils without cover crops (870 Da). The hydrodynamic particle diameters (dh) of all FA were ~2 nm. Identical Mp were obtained for treated (cover crops and nitrogen fertilizer) FA, and the magnitude of the molar mass is consistent with literature data obtained with flow FFF and PSS standards [15,33]. As a word of caution, the comparison of apparent molar mass obtained for fulvic or humic acids often depends on the type of calibration standards, the separation method and experimental conditions such as pH and ionic strength [33,34]. In addition, molecular aggregation or association by bridging with complexed cations, hydrogen bonding and van der Waals forces may lead to apparent large molecular size and hence large molar masses [33,34]. On the other hand, the dh values for FAs were calculated directly from Eq. 16.2 without the need for any standards. Identical apparent molar masses and hydrodynamic diameters were obtained for FAs even though they were derived from soils with various cover crops and fertilizer applications. 16.4.3. Elemental Analyses Interestingly, no discernable pattern appears in trace metals concentrations in FA that are extracted from soils treated with various soil management practices (Table 16.1). As expected, elevated levels (that is, in the upper ppm range) of Al, Fe and Cu concentrations were found in all the FAs investigated (Table 16.3). Lead concentrations in the FAs were between 2.3 and 11.0 µg/g and arsenic ranged from 1.6 to 3.3 µg/g. These results demonstrate the strong metal sequestering capability of soil fulvic acids. 16.4.4. Flow FFF-ICP-MS Investigation of Trace Metals Complexed to Fulvic Acids Our results also demonstrate that metals are complexed to a wide range of molar mass fractions, consistent with the data obtained from total elemental analyses (Table 16.4). Trace metals Zn, Cu and Pb complexed by FA molecular fractions were qualitatively identified by flow FFF-ICP-MS. The corresponding ion fractograms of 63Cu, 208Pb and 64Zn for soil-derived FAVR4 (soils treated with fertilizer and hairy vetch and rye cover crops) and FAC1 (control soil without any cover crops or fertilizer) are illustrated in Figures 16.3c,d and 16.4c,d for hydrodynamic distributions and molar mass distributions, respectively. No obvious differences appear in binding with fulvic acid size fractions among the various fulvic acids studied. All of these metals are bound to 1 to 5 nm diameter size FA fractions with maximum complexation with ~2 nm fractions (~1 kDa). As expected, intense signals were obtained for 27Al and 54Fe, indicating strong binding of these elements to soil fulvic acids. 16.5. CONCLUSIONS Our results demonstrate that flow FFF-ICP-MS can be used for characterization of trace metals complexed to various soilderived FA molecular fractions. Monomodal elemental fractograms showed that trace metals Al, Cu, Fe, Ti, Pb and Zn are bound to a broad range of FA molecular sizes. DRIFTS spectra demonstrated the presence of ligand sites, particularly the intense carboxylic sites that play an important part in metal sequestration by fulvic acids. The total trace metals present in these fulvic acids revealed the presence of iron and aluminum along with all other cations. The apparent molar mass (Mp) of

152

| CHAPTER 16: INVESTIGATION OF COLLOIDAL PROPERTIES AND TRACE METAL COMPLEXATION

fulvic acid obtained by flow FFF ranged from 870–920 Da. All treated soil-derived FA samples had very similar apparent molar masses Mp of 920 Da, except for the FA sample extracted from untreated soil. The hydrodynamic particle diameters (dh) of all FA studied was 2 nm. The data indicate that various soil management practices have no apparent effect on the molar masses or hydrodynamic diameters of these fulvic acid samples. Particle size and molar mass information for metalbound FA is useful in assessing the environmental behavior of biologically and toxicologically important trace elements in soil and water bodies. The small size and low molar mass of these FA molecules and their strong association with nutrient (Fe, Zn) and toxic (Pb, Al) cations is consistent with the great mobility and rapid binding [35] of these elements in the soil environment. ACKNOWLEDGEMENTS Dula Amarasiriwardena gratefully acknowledges the financial and instrument support of the National Science Foundation (BIR 951270) and the Kresge Foundation. This research was supported in part by ICP Information Newsletter, Inc., Hadley, Massachusetts. Jonathan Bell is grateful for the undergraduate research grant awarded by the Howard Hughes Medical Institute. REFERENCES 1. 2.

3. 4. 5. 6. 7. 8. 9.

10.

11. 12. 13. 14. 15. 16. 17. 18.

19.

Stevenson, FJ. Humus chemistry: Genesis, composition and reactions. 2nd Edn. New York: Wiley, 1994. Davies G, Fataftah A, Cherkasskiy A, Ghabbour EA, Radwan A, Jansen SA, Kolla S, Paciolla MD, Sein LT Jr, Buermann W, Balasubramanian M, Budnick J, Xing B. Tight metal binding by humic acids and its role in biomineralization. J. Chem. Soc., Dalton Trans., 1997; 21:4047–4060. Steelink C. Investigating humic acids in soils. Anal. Chem., 2002; 74:326A– 333A. Davies G, Ghabbour EA, Steelink C. Humic acids: Marvelous products of soil chemistry. J. Chem. Ed., 2001; 78:1609–1614. Chin YP, Aiken G, O’Loughlin E. Molecular weight, polydispersity and spectroscopic properties of aquatic humic substances. Environ. Sci. Technol., 1994; 28:1853–1858. Schmitt P, Kettrup A, Freitag D, Garrison AW. Flocculation of humic substances with metal ions as followed by capillary zone electrophoresis. Fresenius J. Anal. Chem., 1996; 354:915–920. Rottmann L, Heumann KG. Development of an on-line isotope dilution technique with HPLC/ICP-MS for the accurate determination of elemental species. Fresenius J. Anal. Chem., 1994; 350:221–227. Vogl J, Heumann KG. Determination of heavy metal complexes with humic substances by HPLC/ICP-MS coupling using an on-line isotope dilution technique. J. Anal. At. Spectrom., 1994; 359:438–441. Ruiz-Haas P, Amarasiriwardena D, Xing B. Determination of trace metals bound to soil humic acids species by size exclusion chromatography and inductively coupled plasma mass spectrometry. In: Davies G, Ghabbour EA eds. Humic substances: Structures, properties and uses. Cambridge: Royal Society of Chemistry, 1998:147–163 Bhandari SA, Amarasiriwardena D, Xing B. Application of high performance size exclusion chromatography (HPSEC) with detection by inductively coupled plasma-mass spectrometry (ICP-MS) for the study of metal complexation properties of soil derived humic acid molecular fractions. In: Ghabbour EA, Davies G eds. Understanding humic substances: Advanced methods, properties and applications. Cambridge: Royal Society of Chemistry, 1999:203–221 Giddings JC. A new separation concept based on a coupling of concentration and flow non-uniformities. Sep. Sci., 1966; 1:123–125. Giddings JC. Field-flow fractionation of macromolecules. J. Chromatog., 1989; 470:327–335. Giddings JC, Yang FJ, Myers MN. Theoretical and experimental characterization of flow field-flow fractionation. Anal. Chem., 1976; 48:1126–132. Beckett R. Field-flow fractionation-ICP-MS: A powerful new analytical tool for characterizing macromolecules and particles. At. Spectrosc., 1991; 12:228–232. Beckett R. Zhang J, Giddings JC. Determination of molecular weight distributions of fulvic and humic acids using flow field-flow fractionation. Environ. Sci. Technol, 1987; 21:289–295. Dycus PJM, Healy KD, Stearman GK, Wells MJM. Diffusion coefficients and molecular weight distributions of humic and fulvic acids determined by flow-field-flow fractionation. Sep. Sci. Technol., 1995; 30:1435–1453. Siripinyanond A, Barnes RM. Flow-field-flow fractionation-inductively coupled plasma mass spectrometry and metal speciation in proteins: A feasibility study. J. Anal. At. Spectrom., 1999; 14:1527–1531. Amarasiriwardena D, Siripinyanond A, Barnes RM. Trace elemental distribution in soil and compost-derived humic acid molecular fractions and colloidal organic matter in municipal wastewater by flow field-flow fractionation-inductively coupled plasma mass spectrometry. J. Anal. At. Spectrom., 2001; 16: 978–986. Anderson T, Shifley L, Amarasiriwardena D, Xing, B, Siripinyanond A, Barnes RM. Characterization of trace metals complexed to humic acids derived from agricultural soils, annelid composts and sediment by flow field-flow fractionation-inductively coupled plasma mass spectrometry. In: Ghabbour EA, Davies G eds. Humic substances: Structures, models and functions. Cambridge: Royal Society of Chemistry, 2001:165–177.

MATERIALS AND METHODS |

20.

21. 22.

23. 24. 25.

26.

27. 28. 29. 30. 31. 32. 33.

34. 35.

153

Lead JR, Wilkinson KJ, Balnois E, Cutak BJ, Larive CK, Assemi S, Beckett R. Diffusion coefficients and polydispersities of the Suwannee river fulvic acid: Comparison of fluorescence correlation spectroscopy, pulsed-field gradient nuclear magnetic resonance, and flow field-flow fractionation. Environ. Sci. Technol., 2000; 34:3508–3513. Taylor HE, Garbarino JR, Murphy DM, Beckett R. Inductively coupled plasma mass spectrometry as an elemental detector for fieldflow fractionation particle speciation. Anal. Chem., 1992; 64:2036–2041. Ranville JF, Chittleborough DJ, Shanks F, Morrison RJS, Harris T, Doss F, Beckett R. Development of sedimentation field-flow fractionation-inductively coupled plasma mass spectrometry for the characterization of environmental colloids. Anal. Chim. Acta, 1999; 381:315–329. Hassellöv M, Lyvén B, Haraldsson C, Sirinawin W. Determination of continuous size and trace element distribution of colloidal material in natural water by on-line coupling of flow field-flow fractionation with ICP-MS. Anal. Chem., 1999; 71:3497–3502. Siripinyanond A, Barnes RM, Amarasiriwardena D. Flow field-flow fractionation-inductively coupled plasma mass spectrometry for sediment bound trace metal characterization. J. Anal. At. Spectrom., 2002; 17:1055–1064. Ding G, Mao J, Herbert S, Amarasiriwardena D, Xing B. Spectroscopic evaluation of humin changes in response to soil managements. In: Ghabbour EA, Davies G eds. Humic substances: Structures, models and functions. Cambridge: Royal Society of Chemistry, 2001:271–279. Ding G, Amarasiriwardena D, Herbert S, Novak J, Xing B. Effects of cover crops systems on the characteristics of soil humic substances. In: Ghabbour EA, Davies G eds. Humic substances: Versatile components of plants, soil and water. Cambridge: Royal Society of Chemistry, 2000:53–61. Chen Y, Senesi N, Schnitzer M. Information provided on humic substances by E4/E6 ratios. Soil Sci. Soc. Am. J., 1977; 41:352–358. Baes AU, Bloom PR. Diffuse reflectance and transmission Fourier transform infrared (DRIFT) spectroscopy of humic and fulvic acids. Soil Sci. Soc. Am. J., 1989; 53:695–700. Niemeyer J, Chen Y, Bollag J-M. Characterization of humic acids, composts, and peat by diffuse reflectance Fourier-transform infrared spectroscopy. Soil Sci. Soc. Am. J., 1992; 56:135–140. Wander MM, Traina SJ. Organic fractions from organically and conventionally managed soils: II. Characterization of composition. Soil Sci. Soc. Am. J., 1996; 60:1087–1094. Becket R, Hotchin DM, Hart BL. Use of field-flow fractionation to study pollutant-colloid interactions. J. Chromatogr., 1990; 517: 435–448. Chen B, Shand CA, Beckett R. Determination of total and EDTA extractable metal distributions in the colloidal fraction of contaminated soils using SdFFF-ICP-MS. J. Environ. Monit., 2001; 3:7–14. Wolf M, Buckau G, Geckeis H, Thang NM, Hoque E, Szymczak W, Kim J-I. Aspects of measurement of the hydrodynamic size and molecular mass distribution of humic and fulvic acids. In: Ghabbour EA, Davies G eds. Humic substances: Structures, models and functions. Cambridge: Royal Society of Chemistry, 2001:51–61. Schimpf ME, Wahlund K-G. Asymmetrical flow filed-flow fractionation as a method to study the behavior of humic acids in solution. J. Microcol. Sep., 1977; 9:535–543. Cabaniss SE, Zhou Q, Maurice PA, Chin Y-P, Aiken GR. A log-normal distribution model for the molecular weight of aquatic fulvic acids. Environ. Sci. Technol., 2000; 34:1103–1109.

Chapter 17 COMPARISON OF DIALYSIS, POLAROGRAPHY AND FLUORIMETRY FOR QUANTIFICATION OF COBALT(II) BINDING BY DISSOLVED HUMIC ACID Fanny Monteil-Rivera,1 Jean-Paul Chopart2 and Jacques Dumonceau1 1Université 2Université

de Reims Champagne-Ardenne, GRECI—BP 1039, 51687 Reims Cedex 2, France

de Reims Champagne-Ardenne, DTI UMR CNRS 6107, BP 1039, 51687 Reims Cedex 2, France 17.1. INTRODUCTION

The bioavailability and migration of pollutant ions (for example, radionuclides or heavy metals) in soils is closely related to the presence of humic substances (HSs). Because they strongly bind cationic species, humic substances when solid or adsorbed on mineral surfaces can retard the transport of metal ions; or conversely, dissolved HSs can enhance the mobility of these ions [1–10]. To predict speciation of metal ions in soils one must first evaluate the binding properties of both the minerals and the organic matter towards the studied metal species. Binding properties of dissolved organic matter (DOM) have been studied with numerous analytical methods including potentiometric, voltammetric, spectroscopic or physical separation based methods [2]. Among the different techniques employed, potentiometry using ion selective electrodes (ISE) is, when applicable, a method of choice due to its capacity to reach, in a direct way, the concentration of free metal ions over a wide concentration range. Most attention has been given to the complexation of divalent trace metals that can be analyzed by ISE (Cu2+, Cd2+, Pb2+ and Ca2+) and far fewer studies have been devoted to the interaction of DOM with ions like Co2+, for which there is no existing ISE. The complexation of Co(II) by HS has been studied with techniques such as fluorescence quenching [11,12], equilibrium dialysis [3], ion exchange [13,14], voltam-metry coupled with ligand exchange [15] and solvent extraction [16]. Each of these techniques may present sources of errors and difficulties of interpretation, so a com parison of results obtained by several of them would help to validate speciation tools for cobalt. In dialysis, adsorption of metal on the membrane and diffusion of organic matter through the membrane might occur and result in experimental errors. In polarography, reversibility, lability, adsorption of organic macromolecules on the mercury electrode and the coexistence of molecules of different sizes all may make polarogram interpretation difficult. In fluorescence, the interpretation of quenching curves, and more specifically the extent of complexation at the maximum quenching and the correlation between quenching and complexation (which generally is assumed to be linear) are main difficulties of the method. The aim of the present work was to evaluate the strengths, weaknesses, and applicability of dialysis, polarography and fluorimetry to study the speciation of cobalt in the presence of dissolved humic acid. Dialysis and polarography experiments were performed with similar concentrations of humic acid (~60 mgc L−1) while fluorimetry involved a more diluted medium (~3 mgc L−1). 17.2. MATERIALS AND METHODS 17.2.1. Chemicals Reagents used were Co(NO3)2.6H2O (Acros Organics), HNO3 (Normadose Prolabo 1N), KOH (Normadose Prolabo 1N), and KNO3 (Fluka). Deionized, distilled and filtered (Millipore: 0.2 µm) water was used to prepare all solutions. Leonardite humic acid (LHA) was obtained from the International Humic Substances Society (IHSS) and used as received. The elemental composition determined with a C, H, N, S Analyser (LECO CHNS-932) was C, 57.74; H, 4.17; N, 1.16; S, 0. 69 and O, 33.89%, with the latter deduced by difference after subtracting the ash content (2.35%). LHA stock solutions of about 500 mg L−1 (pH′ 7) were prepared by stirring 500 mg of LHA with 0.1 M KOH (20.8 mL) and 50 mL of water for 24 h, making up the volume with water and centrifuging twice at 26300 g for 30 minutes to remove the insoluble fraction. The total organic carbon (TOC) concentration of the supernatant determined with a Shimadzu TOC-5050 Analyzer was 271 mgc/L−1.

MATERIALS AND METHODS |

155

The total acidity and carboxylic acid groups were determined by Ba(OH)2 and Ca(Ac)2 exchange [17], respectively, and found to be 12.0 and 7.1 mmol gc−1, respectively. In the present study, the molarity of LHA is deduced from the total acidity and the TOC concentration. 17.2.2. Instrumentation pH measurements were performed under argon with a Metrohm automatic system equipped with a combination Ag/AgCl glass electrode (Metrohm) calibrated with a solution of 10−3 M HNO3 in 0.099 M KNO3. For dialysis experiments, Co(II) concentrations were determined by inductively coupled plasma atomic emission spectrometry (ICP-AES). Differential pulse (dp) and Tast direct current (dc) polarography experiments were carried out at 25° C with a Metrohm 626 Polarecord attached to a Metrohm 663 VA Stand with a hanging mercury drop electrode (HMDE). Drop times of 1 s and scan rates of 2 mV s−1 were used. The polarograms were performed from −0.8 to −1.6 V (relative to SCE) and a 10 mV pulse amplitude was applied for dp polarography. Fluorescence measurements were performed with a Perkin-Elmer Model 50 B luminescence spectrometer with both slits set at 10 nm. An excitation wavelength of 340 nm was used for all the emission measurements, while the synchronous fluorescence measurements were made with two different offsets (′ ′ 1=20 nm and ′ ′ 2=80 nm) between the excitation and emission monochromators. 17.2.3. Procedures Dialysis Experiments. The complexation of Co2+ by LHA was determined in 0.1 M KNO3. Solutions of LHA (20 mL) were added to 1000 Da molecular weight cutoff Spectra/Por 6 dialysis tubing immersed in 150 mL of solutions containing 0.113 M KNO3 and the required amount of cobalt(II) nitrate. The pH was adjusted to steady-state values by adding small aliquots of HNO3 or KOH to the outside solution. After a 4-day equilibrium period at 25°C, TOC concentration and total concentration of Co(II) were determined inside and outside the dialysis tubing. Two different concentrations of LHA inside the tubing were tested under different sets of conditions that are listed in Table 17.1. The above procedure was preceded by preliminary experiments to either minimize or evaluate experimental artifacts associated with equilibrium dialysis. First, preservation agents (sodium azide), trace metals and sulfides in the dialysis tubing were removed by extensive washing with EDTA and hot water (80°C). Second, the extent of Co(II) binding to the tubing walls was evaluated at pH 7 by checking the mass balance. It was found to be 2% of total Co2+ added and therefore was neglected in the calculations. Third, the extent of LHA transfer through the tubing walls was measured at pH 7. After a 4-day equilibrium period, the concentration of LHA in the external compartment was found to be about 10 % of the total amount of LHA. The effect of LHA transfer on the calculation of equilibrium constants will be discussed later. Polarography Experiments. The complexation of Co2+ by LHA was determined in 0.1 M KNO3. Solutions containing the amounts of LHA and Co(II) given in Table 17.1 were prepared. The pH was adjusted to 5.0, 6.0 or 7.0 by adding small aliquots of KOH or HNO3, the solutions were then stirred for 12 h, and the pH was readjusted to the desired value (deviations of 0.5 unit pH from the initial value were observed). Measurements were made at 25°C after removing O2 by bubbling N2 through the solutions for 15 min. Although a “first-order” maximum was observed on the current voltage curves of cobalt, the experiments were performed in the absence of a maximum suppressor because the latter (gelatin) was found to react with humic acid and cause precipitation. Table 17.1 Experimental conditions used with the three different techniques Dialysis pH 4.4−7.5 4.4–6.4 ~6 Polarography 5.0 6.0

[LHA]initial, inside mgC 43.4 65.0 65.0

L−1

65.0 130.0 65.0

[Co(II)]total (M)

R=[LHA]/[Co]

M 5.21×10−4 7.80×10−4 7.80×10−4

1×10−5 1×10−5 5×10−6−5×10−4

52.1 78.0 1.56−156.0

7.80×10−4 1.56×10−3 7.80×10−4

1×10−5−2×10−4 8×10−5−4×10−4 1×10−5−2×10−4

3.90−78.0 3.90−19.5 3.90−78.0

156

| CHAPTER 17: COMPARISON OF DIALYSIS, POLAROGRAPHY, AND FLUORIMETRY

Dialysis pH 7.0 Fluorescence 5.0 6.0 7.0

[LHA]initial, inside

[Co(II)]total (M)

R=[LHA]/[Co]

130.0 65.0

1.56×10−3

3×10−5

7.80×10−4

1×10−5−2×10−4

52.0 3.90−78.0

3.3 3.3 3.3

3.96×10−5 3.96×10−5 3.96×10−5

1×10−6−1.6×10−3 1×10−6−1.6×10−3 1×l0−6−1.6×10−3

0.025−39.6 0.025−39.6 0.025−39.6

17.2.3.3. Fluorescence Experiments. An LHA concentration of 3.3 mgC L−1 was selected as a good compromise between significant intensity and a linear relationship between intensity and concentration. The ionic strength was adjusted to 0.1 with KNO3. Solutions containing LHA and Co(II) at the concentrations in Table 17.1 were adjusted to pH 5.0, 6.0 or 7.0 with either HNO3 or KOH, stirred for 12 h, and readjusted to the desired pH. Emission or synchronous fluorescence spectra were collected at 25°C. 17.3. THEORY 17.3.1. Definition of Complexing Parameter K To be compared, the results obtained with the three techniques have to be expressed in a common way. Because experimental conditions (pH, ionic strength) were kept similar, the conditional stability constant, K, which is valid only for a given set of experimental conditions, was used to compare results. Humic acids are complex mixtures of acidic functional groups i, each of which can react in the presence of cobalt (II) according to Eqs. 17.1 and 17.2. (17.1) (17.2) The average stability constant, K, may be calculated from the free metal concentration [M], the sum of all individual metalhumic acid complexes concentrations [MLi] and the sum of all individual free binding sites concentrations [Li], Eq. 17.3, (17.3) where is the concentration of bound metal, [M] is the concentration of free metal and is the concentration of ligand not bound to the metal. 17.3.2. Evaluation of K from Dialysis Since hydrolysis is negligible in our experiments (pH 7, [Co(OH)+]/[Co2+]=1/1000 [18]), the [Co(II)] inside and outside the dialysis tubing were assumed to be identical. [Co(II)] also corresponds to the free [Co2+] inside the tubing at equilibrium. Knowing the free cobalt concentration ([Co2+]inside= [Co2+]outside=[Co(II)]outside), the total concentration of cobalt ([Co(II)]inside) and the total concentration of LHA ([LHA]inside) inside the tubing, one can calculate K from Eq. 17.4, where all concentraions are molar. (17.4)

17.3.3. Evaluation of K from Polarograms Both [M] and [ML] have to be known to determine K with Eq. 17.3. These concentrations can be determined for a reversible process from the values of either the potential or the current obtained for reducing waves in the presence and absence of ligand. When plotting log [(id–i)/i] against E for the reduction of Co(II) (1×10−5 M in 0.1 M KNO3), a straight line was

MATERIALS AND METHODS |

157

obtained but with a slope a of 11.20, giving rise to a value of n=0.66 (0.0591×′ ), which is not a whole number. This shows that the reduction of free Co2+ ions on the electrode is non-reversible [19]. Irreversible cobalt reduction affects polarogram interpretation. First, for differential pulse (dp) polarography it is not possible to get quantitative information on [Co(II)] from the peak currents. Dillard et al. [20] showed that the peak current measured by dp polarography for irreversible systems, which is much less sensitive than for reversible systems, is subject to residual current from charging of the electrical double layer. Second, the half wave potentials (E1/2) for direct current (dc) polarograms, which are a function of the charge transfer rate constant, k0, vary not only because of complexation but also because of changes in the double layer [2]. For these reasons, neither dp polarograms nor E1/2 values of dc polarograms were used in the present work. Instead, [M] and [ML] were exclusively determined from diffusion currents, which is applicable to both irreversible and reversible systems. Humic acids are macromolecules, so the diffusion constant of ML, DML, should be smaller than the value for M, DM. This slower diffusion favors labile complex behavior, that is complexes that dissociate faster than they diffuse. The lability of the metal/LHA systems implies that free and bound metal ions with unequal diffusion coefficients are present in the sample solutions, including at the electrode. In this case, the limiting current, il, can be expressed by Eq. 17.5, (17.5) where A is a constant that depends on the number of electrons transferred and on experimental conditions, CM is the total bulk concentration of metal an is the mean diffusion coefficient given by Eq. 17.6. (17.6) It should be noted that Eqs. 17.5 and 17.6 only apply if is constant in the diffusion layer, that is if an excess of ligand is present. Parameter A=417 was calculated after measuring the mercury flow rate under our conditions. The slope of the calibration plot in the absence of ligand allowed determining the diffusion coefficient of free cobalt (DM) in 0.1 M KNO3. A value of 9. 9×10−6 cm2 s−1 was obtained, in relatively good agreement with the literature value 7.3×10−6 cm2 s−1 for Co2+ in water [21]. The experimental value of can then be determined from Eq. 17.5 for different values of CL and CM. Provided DML is known, the values of [ML] and [M] can then be calculated from Eq. 17.6 and the mass balance of the metal. The value of [L] is deduced from the mass balance of the ligand, and the formation constant K can be calculated. 17.3.4. Evaluation of K from Fluorescence Spectra Three methods were used to extract conditional constants from fluorescence spectra: 1) a non-linear adjustment initially applied to humic substances [11,22], 2) Stern-Volmer plots [23], and 3) a discrete log K spectrum model describing metal binding by humic matter [24]. The two first methods give a unique average conditional constant for the whole range of concentrations covered by the titration curve, while the last method gives thermodynamic constants that allow calculation of K at any desired concentration. Non Linear Adjustment Procedure. Ryan and Weber have described the fluorescence quenching of a ligand by complexation with metal ions [22]. The total intensity for a given concentration of metal, I, is expressed in Eq. 17.7, (17.7) where xL is the mole fraction of the free ligand, xML is the mole fraction of bound ligand and IL and IML are limiting fluorescence intensities at the beginning and end of the titration, respectively. As the ligand goes from the unbound to the bound form in the titration, the fraction of total ligand bound is given by Eq. 17.8, (17.8) where [MLi] is the concentration of metal bound to the ith site, and CL is the maximum capacity of the ligand under the conditions examined. Combining Eqs. 17.3 and 17.8 gives Eq. 17.9. (17.9) The best values of K, CL and IML can then be obtained by a non-linear adjustment of the experimental data in the form of relative fluorescence versus CM. In the present work, this adjustment was performed with the non-linear fitting function of Microcal Origin software, with a minimum of 10 iterations that were repeated until convergence. Stern-Volmer Plots. When a fluorescent ligand, L, forms no fluorescent complexes with a metal (static quenching) in Eq. 17.2, the following Stern-Volmer Eq. 17.10 is expected to hold [23]. (17.10)

158

| CHAPTER 17: COMPARISON OF DIALYSIS, POLAROGRAPHY, AND FLUORIMETRY

Fluorophores in HSs with no metal ion complexation properties but that contribute a constant factor to the overall quenching profiles cause deviation from linearity of Eq. 17.10 towards the x-axis. If f is the fraction of the initial fluorescence that corresponds to the fluorescent structures accessible for complexation, the following modified Stern-Volmer Eq. 17.11 is obtained [23]. (17.11) If the plot of IL/(IL-I) vs. 1/[M] is linear, K and f can be readily estimated from the slope and intercept of the straight line. However, only the total metal ion concentration, CM, is known experimentally, and its use instead of [M] in Eqs. 17.10 and 17. 11 requires that CM′ [M]. This condition can only be satisfied with high concentrations of metal and/or low values of K. Only the experiments involving concentrations of cobalt′ 4×10−4 M(CM>10× [LHA]) were used for the calculations of K from modified Stern-Volmer plots. Discrete log K Spectrum Model [24]. After converting the quenching to the concentration of bound metal, [ML] with Eq. 17.8 (see below for the value of CL), the free metal concentration can be calculated from the mass balance equation for the total metal, Eq. 17.12. (17.12) Knowing the concentrations of the bound and free metal species at any point in the titration, one can then apply a model specifically designed to represent metal/HS binding. From the different existing models (NICA-Donnan [25], Tipping [26]), we chose the simplest to use model of Westall et al. [24] to analyze the data. In this model, the HA is represented by an assembly of monoprotic acids with assumed pKa values, the anions of which bind cations in a 1:1 complex. Each site of LHA is thus expected to react according to Eqs. 17.1, 17.2 and 17.13. (17.13) Ka(i), KK(i) and sites concentrations are calculated from acid-base titrations whereas the stability constants for cobalt can be calculated from experiments involving Co(II). In the present study, acid-base titrations were performed in 0.1 M KNO3. Parameters were fitted with the computer program FITEQL 3.2 [27], which iteratively optimizes adjustable parameters by minimizing the differences between calculated and experimental data using a nonlinear least-squares optimization routine. Consistent with the discrete log K spectrum approach, the values of four pKa(i) were set at 4,6,8 and 10 to cover the pH range of the data. The concentration of each site as well as the values of KK(i) were adjusted. The resulting parameters are given in Table 17.2. The overall variance, WSOS/DF (weighted sum of squares of residuals divided by the degree of freedom), indicates good fitting (WSOS/DF<10 [27]). A unique value of KK(i) was sufficient to account for the background electrolyte effect. The parameters in Table 17.2 were then kept constant and values of log KCo(i) were determined from fluorescence data at pH 7.0 by introducing the resulting set of ([ML], [M]) data in the computer program FITEQL 3.2. The model was then applied to predict data at pH 5.0 and 6.0. Table 17.2 Discrete log K spectrum modela [24] obtained with the FITEQL3.2 program [27]. For H-binding, parameters were determined by modeling acid-base titrations of LHA. For Co-binding, parameters were determined by modeling fluorescence data while keeping Hbinding parameters constant. H-Binding (WSOS/DF=2.57) Site

pKa(i)b

Co-Binding (WSOS/ DF=0.12)

Log KK(i)c

[L(i)] (mmol gC−1)c

Log KCo(i)

1 4.00 1.55±0.04 4.15±0.07 4.3±0.2 2 6.00 1.55±0.04 2.82±0.09 3.76±0.2 3 8.00 1.55±0.04 0.85±0.11 7.32±0.4 4 10.00 1.55±0.04 2.48±0.10 7.67±0.4 a Values of other parameters in model: ; WSOS/DF stands for Weighted Sum of squares/degrees of freedom); b Fixed values; c Values obtained while fitting potentiometric data (H-bonding) that were kept constant

17.4. RESULTS AND DISCUSSION 17.4.1. Dialysis The complexation of Co(II) by LHA was studied under two sets of conditions: 1) varying pH at constant total Co(II) concentration (10−5 M), and 2) varying Co(II) concentration at constant pH (~6.0). The results are presented in Figures 17.1

MATERIALS AND METHODS |

159

Figure 17.1 Conditional stability constant, K, for Co2+/LHA binding as a function of pH and LHA concentration. Total Co(II) concentration was 10−5 M. Background electrolyte was 0.1 M KNO3

Figure 17.2 Conditional stability constant, K, for Co2+/LHA binding as a function of concentration of free Co(II). Background electrolyte was 0.1 M KNO3

Figure 17.3 Size distribution of LHA as determined by size exclusion chromatography (column: SEPHADEX 25; Standards: polyethylene glycol from 4000 to 23000 Da; [KNO3]=0.1 M; pH 6)

and 17.2, respectively. Cobalt complexation increased with pH due to greater ionization of LHA (Eq. 17.1). This behavior is commonly observed for metal complexation by humic substances. Complexation tends to increase with increasing LHA concentration for a given concentration of Co(II) (Figure 17.1) or with decreasing Co(II) concentration for a given concentration of LHA (Figure 17.2). Again, this is a common tendency for humic substances because of their heterogeneity. At low ratios of metal to ligand, the strongest binding sites coordinate metal ions while weaker binding sites only start reacting at higher metal concentrations. The stability constant, K, which is an average of binding power for all sites, appears smaller at larger metal loadings. As mentioned above, about 10 % of LHA transfers from the internal to the external solution during dialysis equilibration. The LHA concentration was measured in both the internal and external solutions after each experiment so that the effective concentration of LHA at equilibrium could be used in the calculations of K. However, an error could be caused by the presence of LHA outside the tubing: the concentration of cobalt measured outside does not correspond to free cobalt, as assumed in the calculations, but to the sum of free cobalt and cobalt bound to the fraction of LHA that diffused to the external solution. This causes the concentration of free cobalt to be overestimated and the calculated stability constants to be too low.

160

| CHAPTER 17: COMPARISON OF DIALYSIS, POLAROGRAPHY, AND FLUORIMETRY

There were two possible ways to correct the data: either to use pre-dialyzed humic acid that would not diffuse through the tubing walls or to estimate the proportions of free and bound cobalt in the external solution to accomplish more precise calculations. The first possibility was investigated. Humic acid at 65.0 mgC L−1 was dialyzed for 16 days by renewing the external solution with fresh 0.1 M KNO3 every 4 days. The total fraction of extracted LHA rose to 17.8% of total LHA, in good agreement with the 20% that was obtained by size exclusion chromatography (Figure 17.3). The resulting pre-dialyzed acid was then used to evaluate its binding capacity towards Co(II), repeating the general procedure described above. The results are presented as stars in Figure 17.1. Because of the lower external [Co(II)], higher constants should be observed than the values obtained with 43.4 mgC L−1, but the opposite was observed. This shows that the preliminary dialysis retained a material that is less complexing than the initial one. The small particles that transferred from inside to outside the bag were the most complexing. This is not surprising since it is well known that fulvic acids contain more binding sites per gram than their larger homologues [28]. Since the present experiments are part of a project that involves other experiments (potentiometric titrations, sorption studies) that have all been performed with the nonpredialyzed LHA, it was not worthwhile to assess the Co/LHA binding properties with a chemically different material. Therefore, we decided to work with the untreated LHA and to account for the fraction of bound cobalt in the external solution. To do so, the amounts of bound and free Co(II) were evaluated outside the bag with the total measured Co and LHA concentrations and the stability constant initially determined. The constant was then recalculated from the corrected value for free Co(II). These two operations were reiterated until ′ log K<0.01. Table 17.3 gives the original and amended constants for the experiments with varying pH or Co(II) concentration. The correction is more useful at higher pH or for smaller Co concentrations where the bound fraction of Co(II) is more important. For the rest of the study, corrected constants will be considered when discussing dialysis results. Table 17.3 Dialysis experiments. Original and corrected values of stability constants after taking into account the amount of bound Co(II) outside the dialysis bag. pH

[Co(II)] (M)

[LHA] (mgC L−1)

Log K (original)

Log K (corrected)

4.37 5.36 6.32 6.98 7.52 6.03 5.95 6.02 6.06 5.98

1×10−5

43.4 43.4 43.4 43.4 43.4 65.0 65.0 65.0 65.0 65.0

2.5287 3.0592 3.403 3.7905 4.0641 3.5232 2.9582 2.7812 2.5516 2.3758

2.5642 3.0977 3.4947 3.8776 4.2722 3.5946 3.0089 2.8321 2.6164 2.3758

1×10−5 1×10−5 1×10−5 1×10−5 5×10−6 5×10−6 1×10−4 3×10−4 5×10−4

17.4.2. Polarography Determination of DML. As established above, DML must be known to deduce [M] and [ML] from the value of with Eq. 17.6. Theoretically, DML could be determined from the slope of the plot of il vs. CM using the first points of the titration of LHA with cobalt, where DM[M]<
MATERIALS AND METHODS |

161

For comparison, a value of 5×10−8 cm2 s−1 was estimated for a peat humic acid by conducting anodic stripping voltammetry with lead in the presence of humic acid [31]. The smaller value than ours agrees with the larger molecules (50 nm
162

| CHAPTER 17: COMPARISON OF DIALYSIS, POLAROGRAPHY, AND FLUORIMETRY

Figure 17.4 Conditional stability constant, K, for Co2+/LHA binding as a function of pH and concentration of free Co(II) at equilibrium ([KNO3]=0.1 M; [LHA]=65.0 mgC L−1)

Figure 17.5 Fluorescence quenching curves of LHA with Co(II) at different pH after subtraction of the background spectrum ([KNO3]=0.1 M; [LHA]=3.3 mgC L−1). Filled symbols: fluorescence emission (′ exc= 340 nm; ′ mes=465 nm). Open symbols: synchronous fluorescence (′ ′ =80 nm; ′ mes=427 nm); continuous line: modeling with the log K spectrum model; dashed lines: predictions with the log K spectrum model

Determination of K. Knowing DM and DML (′ DL), the concentrations of bound and free cobalt were deduced with Eqs. 17. 5 and 17.6 from the diffusion current mea sured for the wave at −1.1 V. Figure 17.4 shows the resulting binding constants for the lower LHA concentration. The effects of pH and cobalt concentration were similar to those observed by dialysis. However, the values of K determined by polarography are higher than those determined by dialysis. 17.4.3. Fluorimetry Quenching Curves. Complexation of Co(II) by LHA was studied by varying the concentration of Co(II) at three different pH values (5.0, 6.0, 7.0). Self-absorption of LHA allows work with TOC concentrations as high as that employed for the two other techniques and a concentration of 3.3 mgC L−1 was used in the experiments. A detailed analysis of spectra has been published elsewhere [38]. Briefly, among the three procedures employed (emission and synchronous modes with an offset of 20 or 80 nm), the synchronous spectra measured with an offset of 20 nm gave better resolution but also produced an enhanced fluorescence intensity induced by the different chemical environment of the fluorophores, which overlaps with the quenching phenomenon. Quantitative study was only possible from quenching curves obtained in the emission mode and the synchronous mode with an offset of 80 nm (Figure 17.5). The decrease of fluorescence with increasing amounts of Co(II) suggests that complexes are formed between LHA and cobalt. However, high ratios of metal to binding sites (68.4 at the end of titrations carried out at pH 5.0 and 6.0, based on the LHA complexing capacity of 12.0 mmol gC−1) had to be introduced to approach equilibrium, whereas a ratio of ~0.5 was sufficient for Cu(II) [39]. This behavior verifies the weaker binding of Co to humic substances than for copper [11,12]. As observed with the two other techniques, complexation is stronger at higher pH. Modeling according to [11,22]. A first attempt to estimate the three unknown parameters (K, CL and IML) of Eq. 17.9 led to negative CL values. Similar findings were reported by Ryan et al. when fitting experiments involving Co2+ ions [12]. The

MATERIALS AND METHODS |

163

Figure 17.6 (a) Original and (b) modified Stern-Volmer plots for different pH. For clearness, only data obtained by emission fluorescence (′ exc=340 nm) are shown

authors attributed this failure to the lack of data at high concentrations of metal, but by analyzing the dependence of the fitting on the CL value, it can be seen that CL has hardly any effect on the fitting, as already pointed out [40 and refs. therein]. For example, varying the values of CL from 5×10−6 to 4×10−5 M had no effect on the values of log K and IML or on the final overall adjustment. Nevertheless, as seen in Eq. 17.8, parameter CL is a key factor to deduce [ML] values from a set of given I and IML values and although it does not have any effect on the values of both other parameters, it greatly affects the speciation of cobalt. The fact that parameter CL cannot be estimated is a real weakness of the present model. One has to bear in mind that good fitting of fluorescence quenching data is not a goal in itself and that the fitted parameters should allow predicting metal speciation. Another attempt was made to try to determine CL: the number of fitting parameters for Eq. 17.9 was reduced to two by first deducing IML from nonlinear regression analysis of a plot of |I/IL−1| vs. CM from Eq. 17.15 with fitting parameters (17.15) |IML/IL−1| and α. However, this did not give more consistent CL values, with thenew ones exceeding the maximum sites concentration. Values of log K and IML were calculated for the three values of pH after fixing CL to the maximum concentration of sites (3. 96×10−5 M). Results are shown in Table 17.4. The parameter IML is the residual fluorescence of the sample when all the Table 17.4 Fluorescence data. Equilibrium parameters calculated at different pH with different models. I=[(IML−100)/(2KCL)] {(KCL+KCM+1)−(KCL+KCM+1)−4K2CLCM]1/2}+100 Syn.a pH Log K 5.0 3.35 6.0 3.47 7.0 3.99 IL/(IL−I)=1/(fK[M])+1/f

IML(%) 47.20 34.04 36.40 Syn.a f 0.57 0.71 0.74

Em.a r2 0.991 0.985 0.982

r2

Log K 3.43 3.47 3.96

pH Log K Log K 5.0 3.27 0.983 3.29 6.0 3.34 0.985 3.30 7.0 3.64 0.974 3.58 Log K spectrum model pH Log K ([LHA]/[Co(II)]) 5.0 3.33−2.88(129−0.021) 6.0 4.06−3.23 (100−0.021) 7.0 4.83−3.34 (34−0.021) a Syn.: synchronous fluorescence (′ ′ =80 nm); Em.: emission fluorescence (′ exc=340 nm)

IML (%) 51.24 39.21 42.54

r2 0.989 0.988 0.980

Em.a f 0.53 0.67 0.68

r2 0.971 0.969 0.990

164

| CHAPTER 17: COMPARISON OF DIALYSIS, POLAROGRAPHY, AND FLUORIMETRY

available ligand is bound. The fact that IML values differ from zero indicates that a fluorescent fraction of LHA does not complex Co(II). This is due to the presence of non-complexing fluorophores in LHA or to the presence of cations previously complexed by LHA that have low quenching power but higher affinity for the binding sites (Al3+, Pb2+). Stern-Volmer Plots. Figure 17.6 shows the Stern-Volmer plots of the quenching profiles. The raw data (Figure 17.6a) curve towards the x-axis at higher concentrations. As discussed in the theory section, a deviation towards the x-axis suggests the quenching of fluorophores not involved in the complexation reaction. Quasi-linear plots were obtained when data were treated with the modified Stern-Volmer Eq. 17.11. The quantitative information obtained from these plots (log K and f) is shown in Table 17.4. The values of IML and f (or more precisely (100-f) with/expressed in %) are in good agreement. However the values of log K calculated by the non-linear adjustment are larger than those determined by Stern-Volmer plots, especially for measurements at pH 7. This is consistent with the fact that Stern-Volmer parameters were calculated with only the data at higher concentrations of cobalt (CM>2×10−4 M), where smaller stability constants are expected. As seen from the values of r2, the fitting was poor for both models. It is well known for HAs that, because they are heterogeneous, the conditional metal binding constant decreases with the metal loading level, with the weaker binding sites starting to react when the stronger ones are already occupied. Thus, fitting the data with a single constant over a wide Co(II) concentration range can only be an approximation. A different approach was then investigated. Modeling with a Discrete Log K Spectrum Model [24]. The Discrete Log K Spectrum Model fitting procedure has been presented elsewhere [38]. Parameter CL is the amount of ligand bound to metal at the end of a titration, when I equals IML, and it varies with pH. The first step in the modeling consists of assigning a value to CL. Several CL values were tried. Based on the observation of convergence when introducing the corresponding sets of ([ML], [M]) in the program FITEQL 3.2 to calculate KCo(i) constants, CL was fixed at 2.77×10−5 M at pH 7.0 (70% of the total amount of sites). The values of log KCo(i) resulting from the calculation at pH 7.0 are shown in Table 17.2 with the goodness of fit. The latter indicates very smooth fitting, as confirmed by the theoretical curve (continuous line in Figure 17.5). Predictions with these constants are the dashed lines in Figure 17.5. The theoretical quenching curves were calculated with Eq. 17.2, where [MLi] values are the output data of the fit and parameter CL corresponds to the sum of [MLi] at the end of a titration. The calculated curves agree well with the experimental data. From these predictions, maximum capacities of 2.55×10−5 and 1.85×10−5 M were found at pH 6.0 and 5.0 that correspond to 64 and 47 % of the total concentration of sites, respectively. It is confirmed that CL varies with pH. It is emphasized that the set of constants calculated from the titration monitored by synchronous fluorescence (′ ′ 2) at pH 7.0 not only fits the titrations performed at different pH but also predicts well the data monitored by emission fluorescence. To compare the three models, the conditional constants K as defined in Eq. 17.3 were calculated over the whole added cobalt (II) range with the speciation resulting from the 4-sites model. Results are presented in Table 17.4. As expected, the value of log K varies with the degree of binding: it decreases with increasing added cobalt concentration from 1.0×10−6 to 1.6×10−3 M. Except for pH 5.0, which corresponds to the weakest binding and consequently to the largest calculation error, the calculated intervals include the constants from the two first models. This demonstrates good consistency between the three models employed. It is easy to understand the poor fitting obtained with the first two models: the average stability constants underestimate the binding for the low concentrations of cobalt and overestimate it at higher Co(II) concentrations. 17.5. COMPARISON OF TECHNIQUES AND CONCLUSIONS Figure 17.7 gathers the values of log K vs. R obtained at pH 5.0, 6.0 and 7.0 with the three different techniques. The complexity of this figure shows the large impact that the choice of analytical technique has on the results. Figure 17.7 shows that fluorescence gives higher stability constants than the two other techniques. Dialysis generates lower constants than those from polarography but the difference is not that large. Taking into account the applied technique, our results are in very good agreement with literature data. Saar and Weber studied Co(II)/FA complexation by fluorescence and reported log K values of 2.8 and 3.8 at pH 4.0 and 6.0, respectively, for 0.0625
MATERIALS AND METHODS |

165

Figure 17.7 Log K as a function of the ratio of ligand to metal and pH (open symbols: polarography experiments; filled symbols: fluorescence experiments; grey symbols: dialysis experiments)

diffusion constant for Co/LHA complexes (DML). This was estimated with the Stoke-Einstein equation and the gyration radius of LHA, itself estimated from the molar mass. The Stoke-Einstein equation assumes the molecule is spherical, which may not be valid for LHA. Consequently, the coefficient DML used in the calculations may be inexact. The impact of this coefficient on the calculations is important and constants K easily can be doubled or tripled by modifying DML. The need to maintain a constant value for in the diffusion layer does not allow investigating high concentrations (>2×10−4 M) of cobalt. Moreover, the detection limit also prevents interpreting results at [Co(II)] <10−5 M. As a result, only a narrow concentration range can be analyzed by polarography. These restrictions make polarography an unreliable method to quantify cobalt complexation by humic matter. Several conclusions emerge from our fluorescence study. First, the use of oversimplified models is not appropriate for materials as heterogeneous as HSs. The conditional constant actually varies over the whole concentration range. A better option is to convert the fluorescence data into sets of [ML] and [M] values and use them in models designed to speciate metals in the presence of HSs. Second, the larger constants obtained with this technique than from polarography and dialysis highlight the problem of converting fluorescence quenching to bound metal concentration. The difficulty in interpreting fluorescence quenching curves arises from analysis of the ligand instead of the metal. The uncertainty due to residual fluorescence makes it hard to establish the metal status. A linear relationship between quenching and bound metal is commonly assumed but it may not be valid. This eventuality has been exposed by comparing fluorescence and ISE data for Cu2+/fulvic acid binding [41]. A suggestion to solve the problem was to calibrate fluorescence measurements with a technique more directly related to metal concentrations [41]. This means that fluorescence should be measured and a second technique (useful in the same concentration range) used to relate the quenching to the concentration of free or bound metal. This requirement makes fluorescence less useful especially for cobalt, which cannot be analyzed by ISE. In summary, interpretations of polarography and fluorescence data have inherent uncertainty. Dialysis is tedious and requires long equilibrium times, but it gives results in a straightforward way. It will be interesting to apply another analytical method such as ion exchange and compare the results with those obtained by dialysis. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

Davis JA, Leckie JO. Effect of adsorbed complexing ligands on trace metal uptake by hydrous oxides. Environl Sci. Technol., 1978; 12:1309–1315. Buffle J. Complexation reactions in aquatic systems: An analytical approach. Chichester: Ellis Horwood, 1988. Zachara JM, Resch CT, Smith SC. Influence of humic substances on Co2+ sorption by a subsurface mineral separate and its mineralogic components. Geochim. Cosmochim. Acta, 1994; 58:553–566. Ticknor KV, Vilks P, Vandergraaf TT. The effect of fulvic acid on the sorption of actinides and fission products on granite and selected minerals. Appl. Geochem., 1996; 11:555–565. Spark KM, Wells JD, Johnson BB. Sorption of heavy metals by mineral-humic acid substrates. Austral. J. Soil Res., 1997; 35: 113–122. Fairhurst AJ, Warwick P. The influence of humic acid on europium/mineral interactions. Coll. Surf. A: Physicochem. Engineer. Aspects, 1998; 145:229–234. Vermeer AWP, McCulloch JK, van Riemsdijk WH, Koopal LK. Metal ion adsorption to complexes of humic acid and metal oxides: deviations from the additivity rule. Environ. Sci. Technol., 1999; 33:3892–3897. Petrovic M, Kastelan-Macan M, Horvat AJM. Interactive sorption of metal ions and humic acids onto mineral particles. Water, Air, Soil Pollut., 1999; 111: 41–56.

166

| CHAPTER 17: COMPARISON OF DIALYSIS, POLAROGRAPHY, AND FLUORIMETRY

9.

Czerwinski KR, Cerefice GS, Buckau G, Kim JI, Milcent MC, Barbot C, Pieri J. Interaction of europium with humic acid covalently bound to silica beads. Radiochim. Acta, 2000; 88:417–424. Masset S, Monteil-Rivera F, Dupont L, Dumonceau J, Aplincourt M. Influence of humic acid on sorption of Co(II), Sr(II), and Se(IV) on goethite. Agronomie, 2000; 20:525–535. Saar RA, Weber JH. Comparison of spectrofluorometry and ion-selective electrode potentiometry for determination of complexes between fulvic acid and heavy-metal ions. Anal. Chem., 1980; 52:2095–2100. Ryan DK,Thompson CP,Weber JH. Comparison of Mn2+, Co2+, and Cu2+ binding to fulvic acid as measured by fluorescence quenching. Can. J. Chem., 1983; 61:1505–1509. Schnitzer M, Skinner SIM. Organo-metallic interactions in soils: 7. Stability constants of Pb2+, Ni2+, Mn2+, Co2+, Ca2+, and Mg2+fulvic acid complexes. Soil Sci., 1967; 103:247–252. Higgo JJW, Kinniburgh D, Smith B, Tipping E. Complexation of Co2+, Ni2+, UO22+ and Ca2+ by humic substances in groundwaters. Radiochim. Acta, 1993; 61:91–103. Qian J, Xue HB, Sigg L, Albrecht A. Complexation of cobalt by natural ligands in freshwater. Environ. Sci. Technol., 1998; 32: 2043–2050. Kurk DN, Choppin GR. Determination of Co(II) and Ni(II)-humate stability constants at high ionic strength NaCl solutions. Radiochim. Acta, 2000; 88: 583–586. Schnitzer M, Kahn SU. Humic substances in the environment. New York: Dekker, 1972. Martell AE, Smith RE. Critical selected stability constants of metal complexes database, Version 2.0. College Station: Texas A & M University, 1995. Crow DR. Polarography of metal complexes. London: Academic Press, 1969. Dillard JW, O’Dea JJ, Osteryoung RA. Analytical implications of differential pulse polarography of irreversible reactions from digital simulation. Anal. Chem., 1979; 51:115–119. Trémillon B. Électrochimie analytique et reactions en solution, Tome 2. Paris: Masson, 1993. Ryan DK, Weber JH. Fluorescence quenching titration for determination of complexing capacities and stability constants of fulvic acid. Anal. Chem., 1982; 54:986–990. Lakowicz JR. Principles of fluorescence spectroscopy. New York: Plenum Press, 1983. Westall JC, Jones JD, Turner GD, Zachara JM. Models for association of metal ions with heterogeneous environmental sorbents. 1. Complexation of Co(II) by Leonardite humic acid as a functions of pH and NaClO4 concentration. Environ. Sci. Technol., 1995; 29: 951–959. Benedetti MF, Milne CJ, Kinniburgh DG, Van Riemsdijk WH, Koopal LK. Metal ion binding to humic substances: Application of the non-ideal competitive adsorption model. Environ. Sci. Technol., 1995; 29:446–457. Tipping E. Humic ion-binding model VI: An improved description of the interactions of protons and metal Ions with humic substances. Aquatic Geochem., 1998; 4:3–47. Westall JC, Herbelin A. FITEQL 3.2: A user’s manual. Corvallis: Oregon State University, 1996. Stevenson FJ. Humus chemistry, genesis, composition, reactions. New York: Wiley-Interscience, 1982. Thurman EM, Wershaw RL, Malcolm RL, Pinckney, DJ. Molecular size of aquatic humic substances. Org. Geochem., 1982; 4: 27–35. Lide DR ed. Handbook of chemistry and physics. 78th Edn. New York: CRC Press, 1997. Pineiro JP, Mota AM, Gonçalves MLS. Complexation study of humic acids with cadmium(II) and lead(II). Anal. Chim. Acta, 1994; 284:525–537. Cleven RFM, van Leeuwen HP. Electrochemical analysis of the heavy metal/humic acid interaction. Internat. J. Environ. Anal. Chem., 1986; 27: 11–28. Buffle J, Cominoli A. Voltammetric study of humic and fulvic substances. Part IV. Behaviour of fulvic substances at the mercury-water interface. J. Electroanal. Chem., 1981; 121:273–299. Ritchie GSP, Posner AM, Ritchie IM. The pzc of mercury in the presence of humic acids and their complexes with aluminium. J. Electroanal. Chem., 1981; 123:397–407. Mota AM, Pinheiro JP, Gonçalves ML. Adsorption of humic acid on a mercury/ aqueous solution interface. Water Res., 1994; 28: 1285–1296. Van Leeuwen HP, Buffle J, Lovric M. Reactant adsorption in .analytical pulse voltammetry: Methodology and recommendations. Pure App. Chem., 1992; 64: 1015–1028. Galceran J, Rene D, Salvador J, Puy J, Esteban M, Mas F. Reverse pulse polarography of labile metal+macromolecule systems with induced reactant adsorption: Theoretical analysis and determination of complexation and adsorption parameters. J. Electroanal. Chem., 1994; 375:307–318. Monteil-Rivera F, Dumonceau J. Fluorescence spectrometry for quantitative characterization of cobalt(II) complexation by Leonardite humic acid. Anal. Bioanal. Chem., in press. Cook RL, Langford CH. Metal ion quenching of fulvic acid fluorescence intensities and lifetimes: Nonlinearities and a possible threecomponent model. Anal. Chem., 1995; 67:174–180. Esteves da Silva JCG, Machado AASC, Oliveira CJS, Pinto MSSDS. Fluorescence quenching of anthropogenic fulvic acids by Co (II), Fe(III) and UO22+. Talanta, 1998; 5:1155–1165. Cabaniss SE, Shuman MS. Fluorescence quenching measurements of copper-fulvic acid binding. Anal. Chem., 1988; 60:2418–2421.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37.

38. 39. 40. 41.

Chapter 18 DIFFUSION OF METAL CATIONS IN HUMIC GELS Martina Klu′ áková and Miloslav Peka′ Institute of Physical and Applied Chemistry, Faculty of Chemistry, Brno University of Technology, Purky ova 118, 612 00 Brno, Czech Republic

18.1. INTRODUCTION Transport phenomena including diffusion are important for the solution of various problems in science and technology. The interpretation of experimental data is a general problem. The theory of diffusion was developed as an independent discipline for the description of processes, at first in binary systems (Ficks’ laws). Its thermodynamic interpretation led to generalization and deduction of diffusion laws for multicomponent systems [1–4]. Our diffusion studies were conducted with several objectives: 1) we are interested in low energy applications of lignite including agriculture as a carbofertilizer with controlled release of plant nutrients that is slow and diffusion-controlled rather than dissolution-controlled; 2) the gel environment supports controlled-release action as compared with the water environment; and 3) coal, including lignite, is known to be able to swell, so some gel-like systems can work under natural conditions. Our first results were obtained with Cu2+, Co2+ and Ni2+ ions because of the well-known high affinity of humic acids for these ions. This was also confirmed for our lignitic HA. Because of the simplicity of analysis, the method and its implementation could thus be evaluated as well. Analytical solutions of diffusion equations, which give the concentration distribution as a function of time, are known only for some model systems with defined parameters. There are methods of determination of diffusion coefficients based on knowledge of the initial and boundary conditions. In this study, the method of one-dimensional diffusion from a medium with a constant analyte concentration into a plane sheet is used. 18.2. THEORY To solve equations for the diffusion of metal ions in a humic gel, the system was modeled as a two-component incompressible mixture. The balance of component 1 (metal ions) in this system under isothermal-isobaric conditions is: (18.1) where c1 is the concentration of component 1, D is its diffusion coefficient and t is time. If the diffusion coefficient is not dependent on concentration (D f(c1), we can write (18.2) and for one-dimensional non-stationary diffusion Eq. 18.3 is valid, (18.3) where x is the diffusion distance. To solve Eq. 18.3 analytically, the studied system has to comply with the conditions shown in Table 18.1. Initial condition: before the experiment, there would be homogenous distribution of concentration of the diffusing substance. In this case, the concentration of metal ions in the gel in t=0 is zero. Boundary conditions: at the solution-gel interface, the concentration of metal ions would be constant during diffusion (c1,s) and symmetrically exposed to the gel. Using these conditions, Eq. 18.3 can be Laplace transformed and the time development of the concentration profile of metal ions in the experimental gel can be calculated from Eq. 18.4, where (18.4)

168

| CHAPTER 18: DIFFUSION OF METAL CATIONS IN HUMIC GELS

c1,s is the concentration of component 1 at the solution-gel interface. Eq. 18.4 can be used only if the relation is valid. In the opposite case, Eq. 18.5 has to be used for the calculation of the concentration profile at given times [1–3]: (18.5) Table 18.1 Initial and boundary conditions for simple diffusion from saturated metal ion solutions t=0 t>0 x=0 t>0 x=L/2a a L is the length of the gel specimen

c1=0 c1=const. ′ c1/′ x=0

If the intensity of diffusion flux J at x=0 is given by Eq. 18.6, (18.6) the derivative of Eq. 18.4 and its incorporation into Eq. 18.6 will give the expression for the total amount of metal ions transported through the solution-gel interface (m), Eq. 18.7. (18.7) If we know the total amount of diffusion flux through the surface planes of the experimental gel, we can determine the value of the diffusion coefficient from Eq. 18.7. It is known that transition elements form complexes with HA, mainly with its carboxylic and phenolic groups [5–8]. The studied process thus involves sorption and complexation, and for one-dimensional non-stationary diffusion Eq. 18.3 has to be transformed to Eq. 18.8, (18.8) where r is the reaction rate. The solution of Eq. 18.8 is conditional on knowledge of the reaction rate. The following rate law is valid for the first-order complexation, (18.9) where k is the reaction rate constant. From Eq. 18.9 the well-known dependence of concentration of component 1 on time is Eq. 18.10, (18.10) where c1,0 is the initial concentration of component 1. Combination of Eqs. 18.8 and 18.9 gives Eq. 18.11. (18.11) The initial and boundary conditions for solution of Eq. 18.11 are given in Table 18.2. Knowledge of the value of the diffusion coefficient D and the use of some numerical methods are necessary to solve Eq. 18.11 [1–3]. Eq. 18.8 is not directly soluble if the chemical reaction has a more complicated mechanism. Table 18.2 Initial and boundary conditions for diffusion with first-order analyte chemical reaction t=0 t>0 x=0 t>0 x=L/2a aL is the length of the gel specimen

c1=0 c1=c1,0 exp−kt ′ c1/′ x=0

18.3. MATERIALS AND METHODS 18.3.1. Materials Humic Acids. HA samples were obtained from South-Moravia lignite (SML) by means of alkaline extraction. The finely ground lignite was mixed with 0.25 M NaOH solution and heated for 180 min at 80°C with a reflux condenser. The

THEORY |

169

Figure 18.1 Scheme of experimental arrangement

Figure 18.2 Experimentally determined values of total diffusion flux for Co2+ fitted to Eq. 18.7

alkalisoluble portion was separated with a centrifuge. HA was precipitated from solution with HCl to pH=1, centrifuged and washed repeatedly with deionized water until free of Cl-. HA samples prepared in this way were dried and finely milled. The yield of HA was about 45% of the starting material. Humic Gel. Solid HA samples were dissolved in 0.5 M NaOH. This solution was acidified with concentrated HCl to pH=1 −2 and centrifuged. The supernatant was discarded. The resulting humic gel contained about 90% of liquid phase. Other Materials. Co(II), Ni(II) and Cu(II) chlorides were used to prepare saturated solutions in water. 18.3.2. Methods Diffusion Experiment. The humic gel was gently pressed into a silicone rubber tube. The length of the tube was chosen to simulate a semi-infinite system. The tube with gel was placed in a saturated solution of a metal ion (Figure 18.1). After a certain time the humic gel was isolated and sliced. The concentrations of studied metal ions in separate slices were determined with UV/VIS spectrophotometry (Hitachi U3300). On the basis of the obtained data the total diffusion flux and concentration profiles in humic gel bodies were determined. Study of Reaction Kinetics. The kinetics of reaction of metal ions with humic gel were determined by conductimetry (Hanna Instruments HI 8820N) and potentiometry (Sentron Titan K185–016). The metal ion solution was mixed with humic gel at ratio of 50 cm3: 1 g and its solution pH and conductivity were measured with stirring until stabilization of conductance or potential. On the basis of the measured time dependences, the rate constants of complexation of Co2+, Ni2+ and Cu2+ with humic gel were determined. Measurements were carried out at laboratory temperature (about 25°C). 18.4. RESULTS AND DISCUSSION 18.4.1. Simple Diffusion First, we focused on simple metal ion diffusion. The values of diffusion coefficients were determined on the basis of the total diffusion flux from the saturated metal ion solutions. We presumed that metal ions could diffuse only through circular planes of the cylinder (Figure 18.1) and that the diffusion rate is much lower than the rate of solid salt dissolution in the bottom of the vessel. Thus the concentration of the metal ion solution is constant and the concentration of metal ions at the solution/humic gel interface is also constant [1,3]. Therefore, Eq. 18.7 is the appropriate equation and the dependence of the total diffusion flux m on the square root of time has to be linear. An example of experimental data for Co2+ ions fitted by Eq. 18.7 is shown

170

| CHAPTER 18: DIFFUSION OF METAL CATIONS IN HUMIC GELS

Figure 18.3 Dependence of concentration of Ni-HA complexes on the initial concentration of Ni2+

in Figure 18.2. Good agreement between Eq. 18.7 and experimental data was observed for all studied metal ions. Diffusion coefficients calculated from the slope of fitted lines are compared with published values for diffusion in water [9] in Table 18.3. 17.4.2. Complexation of HA with Metal Ions Obtained values of diffusion coefficients of metal ions in humic gels are composite. They may also include the influence of chemical reactions such as complexation. As mentioned above, the interaction of transition metals with HA leads to the formation Table 18.3 Diffusion coefficients of metal ions into humic gels obtained from Eq. 18.7 (Dnum) and for diffusion in water (Dpub) [9] Co2+ Ni2+ Cu2+

Dnum, m2/s

Dpub,m2/s

7.98×10−10 8.18×10−10 1.07×10−9

1.46×10−9 1.32×10−9 1.43×10−9

of complexes, as envisioned in Eq. 18.12. (18.12) Here, H2R is the diprotic binding site for Me2+ ion and MeR is the complex formed. The liberation of H+ ions was confirmed by an increase of acidity of the solution during complexation. According to Eq. 18.12, the reaction of HA with metal ions is a second-order reaction and its kinetics can be described using Eq. 18.13, (18.13) where a and b are initial concentrations of H2R and Me2+, y is the concentration of complex MeR, k is the rate constant and t is time. The integrated form of Eq. 18.13 is Eq. 18.14. (18.14) Eqs. 18.13 and 18.14 are valid provided that the rate of dissociation of complex MeR is very low compared to the rate of its formation. This assumption was tested experimentally by measurement of pH as . The amount of occupied binding sites increases with the initial concentration of Me2+ ions (b) and gradually approaches the amount of binding sites in the system. For the initial concentration of Me2+ ions c0=0.2 mol/dm3 the amount of occupied binding sites is equal to the total acidity of HA in this system (total acidity=3.06×10−2 mol/dm3), Figure 18.3. Therefore, solutions with this initial concentration (c0=0.2 mol/dm3) were used for the determination of complexation rate constants. The material balance in Table 18.4 is valid for our reaction system. We assume that the concentration of MeR at t= 0 is zero. The kinetics of complexation of HA by metal ions was studied using conductimetry and potentiometry. Figure 18.4 shows that the conductivity in the system increases with time and after several minutes it is in a steady state.

THEORY |

171

Figure 18.4 Kinetics of complexation of Co2+ with HA for initial Co(II) concentrations 0.01 mol/dm3 (′ ), 0.10 mol/dm3 (•) and 0.20 mol/ dm3 (′ )

Figure 18.5 Experimental reaction rates for Ni2+ fitted with Eq. 18.17 (conductimetry) Table 18.4 Material balance for complexation of HA with transition metal ions (all the binding sites complexed) [H2R] t=0 a t>0 a−x t′ ′ 0 a w is the concentration of liberated H+ ions

[Me2+]

[MeR]

[H+]a

b b−x b−a

0

W0 2x+wt 2a+W′

a

If the reaction proceeds until all the HA binding sites are occupied by Me2+ ions. Because the mobility of H+ ions is much higher than that of other ions in the system, the increase of conductivity during complexation is caused mainly by the growth of acidity of the solution (the influence of the decrease of concentration of Me2+ ions can be neglected because b>>a). At time t=′ all binding sites are occupied and the total increase of conductivity is proportional to [H+]′ −[H+]0=2a. Then we can write (18.15) and (18.16) where Go, Gt and G′ are the initial conductivity of the solution, and its conductivity at time t>0 and t′ ′ , respectively. If y is taken from Eqs. 18.15 and 18.16 and incorporated in Eq. 14, after rearrangement we obtain the linear Eq. 18.17. (18.17) An example of fitting experimental data to Eq. 18.17 is shown in Figure 18.5. Similar results were obtained using potentiometry. Recall that according to Eq. 18.12 we can write that the amount of MeR formed is directly proportional to the increase of the concentration of H+ ions:

172

| CHAPTER 18: DIFFUSION OF METAL CATIONS IN HUMIC GELS

Figure 18.6 Experimental reaction rates for Ni2+ fitted with Eq. 8.19 (potentiometry)

(18.18) Thus y can be calculated from potentiometric measurements. Incorporating Eq. 18.18 into Eq. 18.14 we obtain (18.19) An example of fitting experimental data to Eq. 18.19 is shown in Figure 18.6. Rate constants determined from conductimetry and potentiometry are listed in Table 18.5. Table 18.5 Comparison of rate constants determined using conductimetry and potentiometry k, m3/mol s Co2+ Ni2+ Cu2+

conductimetry

potentiometry

9.29×10−5 5.23×10−5 4.48×10−5

6.93×10−5 1.47×10−5 7.58×10−5

We can see that all obtained values have magnitudes of 10−5−10−4 m3/mol s but the rate constants obtained from conductimetry and potentiometry are different and should be viewed as apparent rate constants. They include the influence of the reaction in the reverse direction and other interactions in the studied system. It is also possible that the complexes of HA gel with metal ions can be formed by more complicated mechanisms or that other functional groups of HA take an active part in these reactions. For a more complicated complexation mechanism the analytical solution of Eq. 18.8 is not possible. The application of numerical methods is necessary to solve the corresponding equations [1–3]. 18.4.3. Diffusion with Chemical Reaction For other than a first-order complexation reaction, Eq. 18.8 cannot be solved simply. However, we can solve the problem of diffusion with chemical reaction in another way. If the reaction by which the immobilized reactants in a humic gel is formed proceeds very rapidly compared to the diffusion process, local equilibrium can be assumed to exist between the free and immobilized metal ions. The rates of diffusion and chemical reaction can be compared on the basis of experimental results. In Figure 18.4, the time dependences of conductivity for various initial concentrations of Co2+ ions are shown. We can see that the studied system attained the steady state in 3–5 minutes (depending on the initial concentration). Analogous results also were obtained from measurement of pH. On the other hand, the experimental concentration profiles in the humic gel (Figure 18.7) and measured values of diffusion flux (Figure 18.2) show that the diffusion is slower because its duration is at least several hours. The comparison of reaction and diffusion time scales confirms that our assumption about local equilibrium is plausible. In the simplest case, the concentration of immobilized ions c1,im is directly proportional to the concentration c1 of free ions

THEORY |

173

Figure 18.7 Comparison of experimentally measured concentration profiles of Cu2+ ions in humic gel after 1 h (′ ) and 24 h (′ ) and those calculated with Eq. 18.7 (solid lines)

(18.20) and Eq. 18.8 can be modified to Eq. 18.21. (18.21) Substituting for c1,im from Eq. 218.0 we have (18.22) which is seen to be the usual form of the equation for diffusion governed by a diffusion coefficient given by Eq. 18.23. (18.23) Therefore, the values obtained from Eq. 18.7 and Figure 18.2 should be called the effective diffusion coefficients [2] that include the influence of a chemical reaction. This simplification was justified by calculation of concentration profiles in the humic gel using Eq. 18.4 and values of diffusion coefficients from Table 18.4. Calculated profiles were compared with experimental ones and very good agreement was observed. An example of the comparison for Cu2+ ions and two different diffusion times is shown in Figure 18.7. From the concentration evolution at some fixed point, the effect of the gel medium on the rate of diffusional transport is explicit. A comparison of metal ion transport in water and the humic gel is given in Figure 18.8. Deceleration of ion transport by complexation in the gel is evident. 18.5. CONCLUSIONS A simple method for determination of diffusion coefficients in humic gels was developed and tested. Values of diffusion coefficients were determined from the total diffusion flux. Profiles computed from the simple one-dimensional diffusion equation using these values were in good accord with measured ones, which is another verification of the experimental procedure. In spite of its simplicity, the method may be successfully employed for the intended purpose. Diffusion coefficients measured for there metal cations are close to those reported for water solutions, which can be attributed to the high water content in the gels and may indicate the low organic network density. Nevertheless, metal ion diffusion may include the effects of chemical reactions, namely complexation of ions with humic acids. Therefore, the parameters are best described as apparent diffusion coefficients. REFERENCES 1. 2. 3. 4. 5. 6.

Crank J. The mathematics of diffusion. Oxford: Clarendon Press, 1956. Hlavá′ J. Technology of glass and ceramics (in Czech). Prague: SNTL, 1988. Havrda J. Silicate engineering I (in Czech). Prague: SNTL, 1987. Moore WJ. Physical chemistry (in Czech translation). Prague: SNTL, 1981. Aiken GA, McKnight DM, Wershaw RL, Mac Carthy P eds. Humic substances in soil, sediment and water. New York: Wiley, 1985. Frimmel FH, Christmann RF eds. Humic substances and their role in the enviroment. Chichester: Wiley, 1988.

174

| CHAPTER 18: DIFFUSION OF METAL CATIONS IN HUMIC GELS

Figure 18.8 Concentration evolution of Co2+ ions at x=1 mm computed with diffusion coefficients published for water (solid line) [9] and measured for the humic gel (dotted line) 7. 8. 9.

Hayes MBH, MacCarthy P, Swift RS, Malcom RL eds. Humic substances in soil, sediment and water: In search of structure. Chichester: Wiley-Interscience, 1989. Schnitzer M, Kahn SU. Humic substances in the environment. New York: Dekker, 1972. Lide DR. Handbook of chemistry and physics. 76th Edn. New York: CRC Press, 1995.

Chapter 19 THE ROLE OF HUMIC SUBSTANCES IN TRACE ELEMENT MOBILITY IN NATURAL ENVIRONMENTS AND APPLICATIONS TO RADIONUCLIDES Valérie Moulin,1,2 BadiaAmekraz,1,2 NicoleBarre,2 Gabriel Planque,1,2 FlorenceMercier,2 PascalReiller1 and Christophe Moulin1 1Commissariat

2UMR

à l’Energie Atomique (CEA-Saclay), Nuclear Energy Division, 91191 Gif-sur-Yvette, France

8587, Analysis & Environment (CEA/CNRS/Université d’Evry Val-d’Essonne), CEA-Saclay, 91191 Gif-sur-Yvette Cedex, France 19.1. INTRODUCTION

Nuclear activities that relate to waste disposal and storage, dismantling or new processes may produce toxic chemicals and radionuclides. Transfer into the geosphere and biosphere is a key issue of research that addresses the mobility, toxicity and risk of trace elements in natural systems. Knowledge of trace element speciation is essential for a better understanding of their environmental retention, transport and bioavailability [1–4]. As water is a most important vector for trace element dissemination, and particularly for radionuclides, the impact of humic substances (HSs), which are ubiquitous in all environments, is important to study, since HSs react strongly with trace elements [4–7]. The questions to be addressed are whether HSs increase or decrease trace element solubility, how trace element speciation is changed by HSs and how trace metal bioavailability and toxicity are affected. The purpose of the present work is to obtain thermodynamic, chemical and spectroscopic data to determine the speciation of a given radionuclide in the presence of HSs. Complexation studies were performed to characterize the stoichiometry of the complexes formed as well as the associated interaction constants. Data on actinides (thorium, uranium), on fission products (iodine) and activation products (lanthanides) will be presented with a focus on the possibilities of analytical techniques to study the systems. Moreover, the influence of HSs on trace element retention on mineral surfaces will also be illustrated. 19.2. CHARACTERIZATION OF ORGANIC COMPLEXES The formation of complexes between humic or fulvic acids (HAs, FAs) and trace metals will modify their solubility and their speciation, depending both on the ligand concentration and the interaction constants. Many studies have investigated complexation reactions of radioelements with humic substances in pH ranges where competing reactions (carbonate complexation or hydrolysis) are insignificant [5,7–10 and references therein]. Characterization of these simple complexes mostly has been conducted with spectroscopic methods (laser induced fluorescence or spectrophotometry) for M(III), M(V), M(VI) actinide ions of Am, Cm, Np, U, and for lanthanide ions (Dy, Eu, Tb). 19.2.1. Trivalent elements For trivalent elements Eu3+, Am3+, Dy3+ and Cm3+, the use of time-resolved laser-induced fluorescence (TRLIF) has allowed determination of apparent interaction constants through titrations of the metal solution with the organic ligand [5, 6, 11]. Like the different models that describe metal-humate interactions, namely the neutralization charge model [12], the polyelectrolyte model [13], the “conservative roof” model [14] and other models [14–16], the model used here considers a 1:1 stoichiometry and equivalent sites treated as the ligand. More details may be found in previous papers [5–9]. The influence of different parameters, in particular pH and metal concentration, shows no effect of pH on the interaction constants in the pH range of non-hydrolysis, whereas a dependence of the interaction constants on the metal concentration is observed, particularly with trivalent ions (Figure 19.1). This effect (observed for Tb, Eu, Am, Cm [6,17]) is attributed to the presence of different HA and FA sites (strong and weak sites). Different sites have also been characterized by luminescence spectroscopy in the case of Eu(III) [18]. Nevertheless, the influence of metal concentration is controversial since no effect was ob served by Kim and collaborators [7,12; see also 19]. Moreover, analysis of lifetimes obtained by TRLIF indicates the formation of outer-sphere complexes between Eu3+ and humic acids [20].

176

| CHAPTER 19: THE ROLE OF HUMIC SUBSTANCES IN TRACE ELEMENT MOBILITY

Figure 19.1 Effect of metal concentration on the interaction constant (log ′ ) for the system of humic acids and trivalent elements (actinides/ lanthanides) at I=0.1 M NaClO4

Figure 19.2 Impact of ternary complexes on U(VI) speciation in the presence of humic acids. [U]=1 mg/L, [HA]=1 mg/L, I=0.1 M NaClO4, pCO2=10−3.5 atm, log ′ used for the simple complex=5.4, log ′ used for the mixed complex=6.7 (′ in L/eq)

The question that emerges is the extrapolation of these interaction constants to higher pH (the pH of natural waters 6
The approach with TRLIF has been applied to hexavalent elements, and especially uranium (UO22+). The different binary systems have been characterized (hydroxo complexes, humate complexes) [24–26], and study of the global system has allowed us to characterize spectroscopically and thermodymanically a hydroxo-metal-humate complex [4,10]. From these data, the formation of ternary complexes drastically changes U speciation, as shown in Figure 19.2. This underlines the fact that, at neutral pH, U(VI) speciation is dominated by ternary organic complexes, the fate of which will differ from the behavior of inorganic species if no ternary complexes are considered. These results are interesting in relation to data obtained for U distribution in humic acids extracted from granitic waters [27] and having high U content (Figure 19.3). Figure 19.3 shows that the size distribution of HAs (detected through the UV absorbance at 254 nm) strongly depends on ionic strength due to conformational changes, and the U content in the filtrates (analyzed by TRLIF) follows the UV signal as a tracer of humic acids in the filtrate. This correlation supports the absence of inorganic complexes of U (whatever the size

CHARACTERIZATION OF ORGANIC COMPLEXES |

177

Figure 19.3 Filtration of Fanay-Augères humic acids at pH 7 (I=0.1 M, I=0.001 M) at 100 mg/L. U content and UV absorbance (at 254 nm) as a function of size cut-off

cut-off of 100 % uranium in the filtrates), and has to be due to formation of mixed hydroxo-uranium-humate complexes characterized by TRLIF. 19.2.3. Tetravalent Elements In the case of tetravalent elements, and especially Th4+, the formation of organic complexes was studied with an extraction technique based on competition between the organic ligand (HA) and silica colloids [9,10,28]. As shown in Figure 19.4,

178

| CHAPTER 19: THE ROLE OF HUMIC SUBSTANCES IN TRACE ELEMENT MOBILITY

Figure 19.4 log Kd (distribution coefficient expressed in mL/g) as a function of Aldrich humic acid concentration at pH 7.2 (dots) and 7.9 (squares). [SiO2]=250 mg/l, [Th]=10−12 M, [NaClO4]=0.1 M

Figure 19.5 Percent of Th-humate complexes as a function of humic acid concentration at pH 7.2: full line with log ′ (determined in [28]) =17.3±1.1, dotted line with log ′ (determined in [13])=15.8 (same Th hydrolysis data set), (′ in L/eq)

which plots isotherms at neutral pH 7.2 and slightly basic pH 7.9, the retention of Th strongly decreases with increasing humic acid concentrations, indicating the formation of organic species that govern Th behavior in solution. Since Th chemistry is relatively complex (strong hydrolysis, low solubility, easy formation of polynuclear or colloidal species leading to an extremely narrow domain of free thorium ions), the formation of mixed or ternary species should be seriously taken into consideration when organic ligands such as humic substances are present. The formation of hydroxothorium-humate complexes is assumed [9,10] bolstered by the fact that high apparent global interaction constants (′ ) are found for the Th(IV)-HA system, according to the Th hydrolysis data set chosen, with ′ values (expressed in L/eq) ranging from 1017–1019 at pH7.2 and 1019–1021 at pH 7.9 [28]. Figure 19.5 illustrates the impact of such complexes on Th mobility in natural waters [28]. Even at low HA concentrations (1 mg/L), 50% of Th can be present as organic complexes. This result is of prime importance since thorium and other tetravalent actinides (U, Np, Pu), are very insoluble elements in natural waters. 19.2.4. Summary of Actinide Data Table 19.1 summarizes information obtained for the HA-actinide systems. Interaction constants have been determined for trivalent elements, namely lanthanides and trivalent actinides [5–7 and references therein, 8], pentavalent elements, namely neptunium [6,15], tetravalent elements namely thorium [28] and hexavalent elements namely uranium [4,7,10,24,29] according to the model used to describe these organic-metal systems. These studies lead to the following affinity order of actinides (An) for humic acid ligands: An(V)
CHARACTERIZATION OF ORGANIC COMPLEXES |

179

Figure 19.6 Structure of fulvic acids obtained by ES-MS-MS [33] Table 19.1 Summary of data obtained on actinides Actinides

Formation of simple complexes ML

Formation of ternary complexes MLL

Trivalent

Strong complexes; strong & weak sites; determination of log Evidence by TRLIF for carbonatohumate complexes

Tetravalent Strong complexes; possible formation for hydroxo-humate complexes; determination of global interaction constants Pentavalent Weak complexes; determination of log Hexavalent Relatively strong complexes; determination of log Evidence by TRLIF of hydroxohumate complexes; determination of global interaction constants

performed [15] indicates which radionuclides among actinides (Th, Pa, U, Np, Pu, Am, Cm), and activation or fission products (Ni, Co, Cd), are most affected in their speciation by humic ligands, and for which conditions (pH, presence of competing reactions, humic concentrations) HSs may govern their speciation under natural water conditions (6
180

| CHAPTER 19: THE ROLE OF HUMIC SUBSTANCES IN TRACE ELEMENT MOBILITY

Figure 19.7 Negative-ion ESI mass spectrum of 150 mg/l FA solution with 1.5×10−4 M iodine at pH 7. Inset: Deconvolution using Grams® of FA without [33] and with iodine [34]

19.3. CONSEQUENCES OF TRACE ELEMENT RETENTION ON SURFACES Humic substances can retard trace element migration when sorbed on mineral surfaces or trapped in the pores of a filtering medium. In the former case, studies performed at the Mol underground laboratory [35] have shown that humic acids are filtered by the clay medium. The behavior of HSs with mineral phases is important, and particularly when an organic film is formed. The mechanisms governing the formation of this coating are presently unknown [7]. Knowledge of the functional groups involved in such processes is important since it will regulate the complexing behavior of the free residual sites and hence the behavior of trace elements in solution. The strong influence of humic acids on the sorption of tri- and tetravalent actinides and iodine, and to a lesser extent on sorption of hexavalent actinides on oxides has been demonstrated for silica [32,36–37] and iron oxides [38–40]. Each binary system (radionuclide/surface and humic acids/surface) has been thoroughly characterized. Figure 19.8 illustrates the cases of americium on silica and thorium on hematite. From nuclear microprobe analysis, iodine appears to be fixed onto silica only when humic acids are present, due to the formation of an organic coating that reacts with iodine [32,37]. This sorption (though low in quantity) occurs even in the absence of an oxidizing agent (and also in the absence of an enzyme), indicating a role of oxygen in the air. Under controlled reducing conditions (presence of thiosulfate ions), iodide does not react with humic acids. From the ternary system study, it appears that dissolved humic acids may increase radionuclide availability since they are oversaturated and have organic complexes of high binding strength [36,38–40]. On the contrary, surfaces that form an organic film could exhibit higher retention of radionuclides. In such systems, the sequence order of constituents binding is of prime importance in determining the reversibility of such interactions. 19.4. CONCLUSIONS AND PERSPECTIVES Humic substances, which are ubiquitous in many environments, may strongly affect radionuclide speciation due to the formation of simple or mixed complexes, and hence may increase their solubility. This is particularly important in the environment. Since speciation will affect the fate of a toxic element (predicted through mod eling) in the geosphere and biosphere, these particular HS ligands, the structure of which has recently been elucidated by direct measurements, can not be neglected in speciation codes or in geochemical modeling. Further investigations in different environments are needed for us to have a more complete understanding of the role of humic substances in radionuclide retention and mobility. Among factors to be understood are the role of competing natural cations, the rates and reversibility, the redox behavior and the properties of humic substances adsorbed on mineral phases. Moreover, the consequences of HSs effects on trace element speciation, namely the presence of organic complexes in solution, on bioavailability must also be taken into account. The impact of such organic complexes on soil-plant transfer or on the transfer in the food chain has to be investigated.

CHARACTERIZATION OF ORGANIC COMPLEXES |

181

Figure 19.8 Percent of Am (a) or Th (b) sorbed on silica (a) or hematite (b) as a function of pH in the presence or not of Aldrich humic acids: (a): [Am]=10−8.5 M, [silica]=21 g/L; I=0.1 M; (b): [Th]=10−12 M, [hematite]=50 mg/L; I=0.1 M

Speciation is a key factor in the distribution of elements in aquatic and terrestrial systems as well as in trace element bioavailability, Its determination and prediction is of prime importance for migration and toxicity purposes. Developments in analytical methods (ES-MS-MS, EXAFS) as well as molecular modeling are indirectly determining the speciation of the element of interest under conditions relevant to natural systems. ACKNOWLEDGEMENTS Dr. Touhoat (CEA and UMR 8587) is acknowledged for helpful discussion. We also thank Drs. N.Labonne, I.Laszak and F.Casanova for their contribution to this work. The results are part of the work performed within the European Commission project HUMICS: “Effects of humic substances on the migration of radionuclides: Complexation and transport of actinides”, contract FI4W-CT96-0027, as well as with the CEA project 98T13 “Transport, transfer of radionuclides in the life”. REFERENCES 1. 2. 3. 4. 5.

Tessier A, Turner DR. Metal speciation and bioavailability in aquatic systems. New York: Wiley, 1995. Contamination des sols par Les elements en traces: les risques et leur gestion. Rapport N°42 (Août 1998). Académie des Sciences (Lavoisier TECDOC). OECD Proceedings of the Workshop on “Evaluation of speciation technology,” Japan, October 1999. Code 662001041P1 OECD, May 2001. Moulin V, Moulin C. Radionuclide speciation in the environment: A review. Radiochim. Acta, 2001; 89:773–778. Moulin V, Moulin C. Fate of actinides in the presence of humic substances under conditions relevant to nuclear waste disposal. Appl. Geochem., 1995; 10: 573–580.

182

| CHAPTER 19: THE ROLE OF HUMIC SUBSTANCES IN TRACE ELEMENT MOBILITY

6.

Moulin V, Moulin C, Dran JC. Role of humic substances and colloids on the behaviour of radiotoxic elements in relation with nuclear waste disposal: Confinement or enhancement of migration? In: Gaffney JS, Marley NA, Clark SB eds. Humic and fulvic acids and organic colloidal materials in the environment. Symposium Series 651, Washington, DC: American Chemical Society, 1996: 259–271. Buckau G ed. Effects of humic substances on the migration of radionuclides: complexation and transport of actinides, EUR Report 19610 EN, 2000. Plancque G, Moulin C, Moulin V, Toulhoat P. Complexation of Eu(III) by humic substances: Eu speciation determined by timeresolved laser-induced fluorescence. In: Buckau G ed. Third Technical Progress Report FZKA 6524, October 2000. Reiller P, Moulin V, Dautel, C, Casanova F. Complexation of Th(IV) by humic substances. In: Buckau G ed. Third Technical Progress Report FZKA 6524, October 2000. Moulin V, Reiller P, Dautel C, Plancque G, Laszak I, Moulin C. Complexation of Eu(III), Th(IV) and U(VI) by humic substances. In: Buckau G ed. Second Technical Progress Report FZKA 6324, June 1999. Glaus MA, Hummel W, van Loon L. Trace metal-humate interactions. I. Experimental determination of conditional stability constants. Appl. Geochem., 2000; 15:953–973. Kim JI, Czerwinski K. Complexation of metal ions with humic acid: Metal ion neutralization model. Radiochim. Acta, 1996; 73: 5–10. Choppin GR, Allard B. Complexes of actinides with naturally occurring organic coumpounds. In: Freeman AJ, Keller C eds. Handbook of the physics and chemistry of the actinides. Amsterdam: Elsevier, 1985: Chapter 11. Hummel W, Glaus MA, van Loon LR. Trace metal-humate interactions. II. The conservative roof model and its application. Appl. Geochem., 2000; 15: 975–1001. Reiller P, Moulin V, Giffaut E. On the influence of humic substances upon radionuclide speciation. A sensitivity study. Appl. Geochem., submitted Choppin GR, Labonne-Wall N. Comparison of two models for metal-humic interactions J. Radioanal. Nucl. Chem., 1997; 221: 67–71. Bidoglio G, Grenthe I, Robouch P, Omenetto N. Complexation of Eu and Tb with humic substances by time-resolved laser-induced fluorescence. Talanta, 1991; 9:999–1003. Shin HS, Choppin GR. A study of Eu(III)-humate complexation using Eu(III) luminescence spectroscopy. Radiochim. Acta, 1999; 86:167–174. Hummel W, Glaus MA, van Loon LR. Complexation of radionuclides with humic substances: The metal concentration effect. Radiochim. Acta, 1999; 84: 111–114. Moulin C, Wei J, Van Iseghem P, Laszak I, Plancque G, Moulin V. Europium complexes investigations in natural waters by timeresolved laser-induced fluorescence. Anal. Chim. Acta, 1999; 396:253–261. Panak P, Klenze R, Kim JI. A study of ternary complexes of Cm(III) with humic acid and hydroxide or carbonate in neutral pH range by time-resolved laser-induced fluorescence spectroscopy. Radiochim. Acta, 1996; 74:141–146. Plancque G, Moulin V, Toulhoat P, Moulin C. Europium speciation by time-resolved laser induced fluorescence. Anal. Chim. Acta, 2002, in press. Diercks A. Complexation of Europium with humic acids—Influence of cations and competing ligands. Ph D Disseration, Leuven University, 1995. Laszak I. Etude des interactions entre colloïdes naturels et elements radiotoxiques par spectrofluorimétrie laser à résolution temporelle. Etude spectroscopique et chimique. These de l’Université Pierre et Marie Curie-Paris VI, 1997. Moulin C, Decambox P, Moulin V, Decaillon JG. Uranium speciation in solution by Time-Resolved Laser-Induced Fluorescence. Anal. Chem., 1995; 34: 348–353. Moulin C, Laszak I, Moulin V, Tondre C. Time-resolved laser-induced fluorescence as a unique tool for low-level uranium speciation. Appl. Spectrosc., 1998; 52:528–535. Moulin V, Billon A, Theyssier M, Dellis T. Study of the interactions between organic matter and transuranic elements. EUR Report 13651, 1991. Reiller P, Moulin V, Casanova F, Dautel C. On the study of Th(IV)-humic acids interactions by competition towards sorption onto silica and determination of global interaction constants. Radiochim. Acta, submitted. Lenhart J, Cabaniss SE, MacCarthy P, Honeyman B. Uranium(VI) complexation with citric, humic and fulvic acids. Radiochim. Acta, 2000; 88:345–354. Moulin V, Tits J, Ouzounian G. Actinide speciation in the presence of humic substances in natural water conditions. Radiochim. Acta, 1992; 58/59:179–190. Mercier F, Moulin V, Guittet MJ, Barré N, Toulhoat N, Gautier-Soyer M, Toulhoat P. Applications of different analytical techniques such as NAA, PIXE and XPS for the evidence and characterization of the humic substances/iodine associations. Radiochim. Acta, 2000; 88:779–785. Mercier F, Moulin V, Guittet MJ, Barré N, Gautier-Soyer M, Trocellier P, Toulhoat P. Applications of new surface analysis techniques (NMA and XPS) to humic substances. Org. Geochem., 2002; 33:247–255. Plancque G, Amekraz B, Moulin V, Toulhoat P, Moulin C. Molecular structure of fulvic acids by electrospray with quadrupole/timeof-flight mass spectrometry. Rapid Commun. Mass Spectrom., 2001; 15:827–835 Moulin V, Reiller P, Amekraz B, Moulin C. Direct characterization of covalently bound iodine to fulvic acids by electrospray mass spectrometry. Rapid Commun. Mass Spectrom., 2001; 15:2488–2496.

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31.

32. 33. 34.

CHARACTERIZATION OF ORGANIC COMPLEXES |

35. 36. 37. 38. 39. 40.

183

Diercks A, Put M, De Cannière P, Wang L, Maes N, Aertsens M, Maes A, Van-cluysen J, Verdickt W, Gielen R, Christiaens M, Warwick P, Hall A, Van der Lee J. TRANCOM Clay. Final Report. EUR 19135 EN, 2000. Labonne-Wall N, Moulin V, Vilarem JP. Retention properties of humic substances onto amorphous silica: Consequences for the sorption of cations. Radiochim. Acta, 1997; 79:37–49. Mercier F, Moulin V, Barré N, Trocellier P, Toulhoat P. Study of a ternary system silica/humic acids/iodine: Capabilities of the nuclear microprobe. Nucl. Methods, 2001; B181:628–633. Reiller P, Moulin V, Casanova F. Sorption behaviour of humic substances towards iron oxides. In: Buckau G ed. Second Technical Progress Report FZKA 6324, 1999. Reiller P, Moulin V, Dautel C. Sorption behavior of humic substances towards hematite: Consequences on thorium availability. In: Buckau G ed. Third Technical Progress Report, FZKA 6524, October 2000. Reiller P, Moulin V, Casanova F, Dautel C. Retention behaviour of humic substances onto mineral surfaces and consequences upon Th(IV) mobility: Case of iron oxides. Appl. Geochem. in press.

Chapter 20 INFLUENCE OF HUMIC SUBSTANCES ON THE MIGRATION OF ACTINIDES IN GROUNDWATER G.Buckau,1 M.Wolf,2 S.Geyer,3 R.Artinger1 and J.I.Kim1 1Forschungszentrum

Karlsruhe, Institut für Nukleare Entsorgung, P.O. Box 3640, 76021 Karlsruhe, Germany 2GSF-National Research Center for Environment and Health, Institute of Hydrology, 85764 Neuherberg, Germany Environmental Research Center Leipzig-Halle, Hydrogeology Section, 06120 Halle, Germany 3 20.1. INTRODUCTION The transport of radionuclides in groundwater is a function of their stability in solution and their retention. In solution, radionuclides may be transported by diffusion or with groundwater flow in ionic or particulate/colloidal form. Retention may be the result of reversible sorption on sediment, inclusion in pre-existing minerals or by inclusion during mineral formation. Microorganisms may be important, especially in retention and chemical conversion in biofilms. Humic substances will influence the mobility of radionuclides by 1) modification of sediment and inorganic colloid surface interactions with radionuclides, including the overall charge; 2) stabilization of inorganic/mineral colloids in solution; 3) modification of mineral growth/dissolution with respect to inclusion and re-dissolution of participating radionuclides; and 4) humic colloid mediated transport. Humic colloids denote aquatic humic substances in their form in groundwater, that is, the organic humic backbone and associated inorganic constituents, which also implies colloidal properties under the physico-chemical conditions of natural groundwaters. To assess the impact of humic substances, data, process understanding and implementation tools for these components are required. For a specific site, the concentration of humic substances (in solution and on sediments) is needed. This will be a function of time, depending on the different sources that contribute to the inventory. The state-of-the-art understanding of these different aspects may be summarized as follows. The modification of interaction properties of mineral surfaces has been demonstrated and partly quantified. The overall charge of mineral surfaces becomes modified and active sorption sites become blocked by sorption of humic substances. The stabilization of inorganic colloids by humic substances has been demonstrated in laboratory experiments. Modification of mineral dissolution/growth is known for simple organic substances. For all these aspects, mechanistic understanding allowing for trustworthy long-term predictions is missing and further research is required. The relevance of humic colloid mediated actinide transport in natural groundwater relies on three basic issues [1]. These are 1) the presence of humic substances in groundwater; 2) their long-term stability, that is, resistance to sorption/retardation and decomposition; and 3) the strength of radionuclide interaction with humic colloids. More recent analysis shows that the strength based on radionuclide-humic colloid interaction equilibrium is less important than account of the complex metal ion-humic colloid interaction kinetics [2]. Part of natural trace metal ions is bound with basically zero dissociation rates [3]. In addition, a small fraction of trivalent americium and europium binds with basically irreversible dissociation kinetics under laboratory conditions [4]. This has considerable impact on the approach to treatment of humic colloid mediated radionuclide transport in groundwater. In this paper, the state-of-the-art of three key issues identified from the ongoing developments in this field are discussed. These are 1) origin of humic colloids; 2) their stability/mobility; and 3) the kinetics of actinide-humic colloid interaction. Consequences of assessing the impact of humic colloid mediated actinide transport are also discussed. 20.2. ORIGIN OF HUMIC COLLOIDS Humic colloids in natural groundwater originate from inflow with recharge or in-situ generation. Humic colloids from inflow with recharge originate from the land surface or from lakes, rivers and other open waters. The conditions at the land surface will vary with time through climatic changes as well as changes in land-use, which include deforestation, agriculture and wetland drainage. Determination of the origin of humic acids (HAs) and fulvic acids (FAs) from different sources requires sufficient hydrological understanding in combination with isotope-geochemical data [5– 10]. The dissolved organic carbon (DOC) concentration of groundwaters from the Franconian Albvorland aquifer system has concentrations between 0.1 and 0.5 mgC/L, with around 10 % being fulvic acid. These groundwaters originate from recharge

ORIGIN OF HUMIC COLLOIDS |

185

Figure 20.1 Concentration of humic and fulvic acids (HAs and FAs) in a near surface groundwater from the Gorleben aquifer system with different land-use and vegetation. For comparison, data are shown for a surface near marsh and deep groundwaters originating from recharge from a previous wetland. In addition, data are shown from surface near groundwater under agricultural land of the Fuhrberg aquifer system

Figure 20.2 Generation of dissolved organic carbon (DOC), including aquatic humic and fulvic acids, and dissolved inorganic carbon (DIC) by microbiological processes from soil or sedimentary organic carbon (from [9], modified and extended from [8,10])

in a forest area at around 500 m altitude above sea level. Over the flow path of 22.5 km, the age of groundwaters range from relatively modern (<1000 years) to the end of the latest glacial maximum (around 15,000 years ago). This shows that the inflow of humic colloids has remained relatively stable over about 15,000 years in this aquifer system. The Gorleben aquifer system (Lower Saxony, Germany) is an example where the situation is very different and humic colloid recharge concentrations have varied considerably within the past few hundred years. Figure 20.1 shows the concentra tions of humic and fulvic acid for near surface recharge groundwater with different land-use. For comparison, deep groundwater originating from recharge in a previous wetland are shown together with the concentration from a near surface marsh groundwater. The wetland was drained between the years 1710 and 1750. Furthermore, data are shown for the Fuhrberg aquifer system with recent groundwaters (containing tritium from nuclear atmospheric testing, i.e. <40 years) at 4 to 14 m depth under agricultural land. The lowest concentrations of HAs and FAs are found in groundwater under conifer forest with a poor humus layer on fluviatile sand. Other types of forest show considerably higher inflow of humic and fulvic acids. The same is true for inflow from agricultural land, with about an order of magnitude higher inflow concentrations. Extremely high values are found near the surface in a marsh as well as for deep groundwaters originating from a former wetland. Aquatic humic substances may also be generated directly in deep groundwater. In-situ generation requires a source of organic matter and conditions for microbiological processes in which part of the sedimentary organic matter is released as humic matter. In the case of the Gorleben aquifer system, this is shown to be the case where microbiological activity in biofilms consume sulfate released from the underlying salt dome in conjunction with release of phosphate into the groundwater [8,10]. A principal scheme of this process is shown in Figure 20.2. For the in-situ generation in deep Gorleben groundwater, the process is verified by the isotopic shift of residual sulfate following preferential microbiological consumption of 32SO42− (Figure 20.3) and from 13C of the dissolved inorganic carbon (DIC). The scheme shown in Figure 20.2 can be generalized to the formation of aquatic humic substances from different sources. The concentration of dissolved inorganic carbon of biogenic origin correlates with the concentration of humic and fulvic acids for groundwaters where the aquatic humic substances originate from the soil zone, from recharge from a previous wetland, and from in-situ generation in deep groundwaters (Figure 20.4, [9]). In summary, the concentrations of humic and fulvic acids with recharge will vary with the type of land and changes in land-use. The latter includes changes in vegetation due to climatic changes. In addition, in-situ generation may contribute where appropriate conditions exist.

186

| CHAPTER 20: INFLUENCE OF HUMIC SUBSTANCES ON THE MIGRATION OF ACTINIDES

Figure 20.3 34S values of sulfate as a function of chloride concentration in Gorleben groundwater. Remaining sulfate becomes enriched in 34S due to microbiological preference for reduction of 32SO [9]. CDT is Canyon Diablo Troilite 4

Figure 20.4 The total concentration of aquatic humic and fulvic acids (HAs and FAs) vs. the concentration of dissolved inorganic carbon (DIC) of biogenic origin (evaluated via 13C) for different groundwaters where the humic substances originate from different sources [9]

Examples of additional sources due to disposal of waste and associated engineering activities are 1) release from clay backfill, release from clay host rock; and 2) conversion of clay and waste organic matter. Leaching of clay back-fill resulted in 6±1 mgC/L total concentration of humic and fulvic acids [11]. Typical concentrations of organic matter in Boom clay (clay formation under investigation for nuclear waste disposal in Belgium) interstitial pore-water range from 40–144 mgC/L, with exceptional values reaching up to a few hundred mgC/L [12]. Changes in chemical conditions may lead to conversion of hydrophobic clay organic matter into humic and fulvic acids. The Callovo-Oxfordian clay at the Meuse Haute Marne site (France) is presently under investigation for nuclear waste disposal. The original clay contains about 1.3 wt% of organic carbon of low maturity level characterized by the presence of unsaturated biomarkers [13,14]. The samples were contacted with “solid young fluid” mimicking the conditions from dissolution of cement used as a waste form. The pH of such solutions is about 13.2. After about 540 days the concentration of dissolved organic carbon was between 243 and 355 mgC/L. It was shown that this hydrophilic organic carbon has typical physicochemical and spectroscopic properties of humic and fulvic acids [14]. In groundwater, microbiological turnover of natural organic matter is the source of humic and fulvic acids of different origin. Disposal of organic material may provide conditions appropriate for microbiological activities, resulting in considerable generation of humic and fulvic acids. Presently, turnover of organic material is assumed to lead to carbon dioxide and/or methane, depending on the redox conditions. Information on the possible generation of humic matter is incomplete. In summary, the origin of humic and fulvic acids in natural groundwaters can be assessed with hydrological and isotopegeochemical data. A number of different sources contribute to a similar process for the generation of aquatic humic substances. Information on additional sources introduced from leaching of clay back-fill or host rock is available. The

ORIGIN OF HUMIC COLLOIDS |

187

Figure 20.5 Evaluation of groundwater flow velocity by hydrological evaluation and transition between Pleistocene and Holocene (18O data). The fulvic acid is shown to have the same transport velocity (from 14C dating) as the groundwater and thus migrates like an ideal tracer over 15,000 years and 22.5 km [2,15]. Hydrological modeling from [20]

generation of humic and fulvic acids from conversion of hydrophobic clay organic matter is a potential source under unusual high pH conditions. Reliable information is still missing with respect to microbiological conversion of waste organic matter. 20.3. STABILITY AND MOBILITY OF AQUATIC HUMIC SUBSTANCES The stability and mobility of humic colloids can be assessed with different approaches. One approach is analysis of isotopegeochemical data in conjunction with sufficient hydrological knowledge of given natural aquifer systems. Another approach is carrying out column experiments under near-natural conditions. In the latter case, concern will always exist with respect to the actual relevance for natural conditions. In the above cited hydrological and isotope-geochemical studies of four different aquifer systems [5–9], conclusions were drawn with respect to the longterm stability and mobility of aquatic humic substances. In the Franconian Albvorland aquifer system fulvic acid migrates like an ideal tracer for 15,000 years over a distance of 22.5 km (Figure 20.5, [15]). In the Gorleben aquifer system, the stability and unhindered mobility of humic colloids was shown from comparison of humic colloid concentrations with the concentrations of co-generated dissolved inorganic carbon [6] and by analysis of humic colloid content in groundwater originating from a previous wetland drained between the years 1710 and 1750 [9]. The results are verified from evaluation of the humic colloid mediated actinide migration from column experiments under near-natural conditions [4]. Humic colloid mediated americium transport was evaluated under the basic assumption of humic colloids migrating without sorption or decomposition. This assumption is verified by the correct description of the americium transport behavior as solely dependent on the kinetics of americium-humic colloid interaction. Different behavior can be found under conditions where the columns have not been adequately conditioned with the respective groundwater. In such cases, a dynamic sorption of FA between solution and sediment can be deduced [2]. This raises the question of the expected behavior under perturbed conditions with a geochemical gradient between, for example, the near-field of a repository and the far-field. In the near-field, chemical conditions will be perturbed by dissolution of waste, packages, back-fill and engineered structures as well as due to oxidizing conditions during construction and operation. If an additional source is introduced in the near-field, the mobility of these aquatic humic substances into the far-field remains an open question. 20.4. TRACE METAL ION INTERACTIONS WITH HUMIC COLLOIDS The interactions of trace metal ions with humic substances have been the subject of a large number of studies over the past decades. Various approaches are used for rationalization of experimental observations or fitting of parameters to describe the results with a proposed mechanism. None of the published approaches rely on proven mechanisms and the complex kinetics are treated separately, mainly through fitting of the observed sorption or desorption kinetics by a number of kinetic steps, each with a specific rate constant and an associated frequency (see for example refs [16,17]). Some general observations can be made with reference to studies conducted on laboratory time scales. With increasing ionic charge, the strength of the interaction increases and the dissociation rates decrease [18]. This indicates a contribution to the kinetics from the coordination of metal ion complexes. This is also verified by spectroscopic study of Eu(III) and Cm(III) humate complexation, which shows slight differences in coordination between different kinetic dissociation modes [19]. The dissociation rates show a complex dependence on physico-chemical parameters [18]. The kinetics show the appearance of an additional much slower dissociation mode in the neutral pH range below the limit of hydrolysis of the studied Ni(II) ion

188

| CHAPTER 20: INFLUENCE OF HUMIC SUBSTANCES ON THE MIGRATION OF ACTINIDES

Figure 20.6 Comparison of experimental data and modeling (extended KICAM approach) for the transport of Am(III) through sand columns under near natural conditions [4]

Figure 20.7 Time dependence of dissociation of trace metal ions as natural constituents in humic colloids of the Gorleben groundwater Gohy-2227. Metal ion concentrations in solution where dissociated metal ions are scavenged by Chelex 100 cation exchanger [3]

compared to acidic conditions [17]. This indicates that changes in humic and fulvic acids contribute to the dissociation kinetics in addition to the complex coordination. In addition, the metal ion concentration has a strong influence on the dissociation rates, being much lower for tetravalent trace metal ions in the pico-molar concentration range compared to higher concentrations, especially approaching saturation of the humate ligand concentration [20]. Based on two rate constants for consecutive binding of americium(III) to humic colloids, a fraction of americium binding quasi-irreversibly and no retention of humic colloids, the transport of americium through sand columns under near-natural conditions can be described (Figure 20.6) [4]. The distribution of the different ionic americium species and complexes in solution was not taken into account. Prior to experiments, the sand columns were conditioned with the groundwater for a minimum of three months under anoxic conditions (Ar with 1 % CO2). Such thorough conditioning is required to obtain stable conditions. Experiments were conducted with conditioning times (preconditioning of groundwater with americium prior to introduction into and migration through the column) reaching up to almost 70 hours and migration times (residence times in the column) between 5.7 and 34 hours. The capability to describe experimental data with just a few parameters is encouraging. However, the possibility of extrapolating to time-spans relevant for natural conditions is questionable. Furthermore, the use of operational constants without regard of distribution of dissolved ionic species is unsatisfactory. The main problem, however, arises from comparison of trace metal ion behavior observed in experiments under laboratory conditions and time scales with the behavior of natural trace metal ions in groundwater. Contrary to the behavior of metal ions added under laboratory conditions, trace metal ions bound to natural humic colloids show a considerable portion with basically irreversible association (Figure 20.7) [3]. The portion of metal ions bound in such a way increases with increasing metal ion charge. In addition to the lack in process understanding of the dissociation kinetics under laboratory conditions, the reason for basically irreversible binding of part of natural humic colloid constituents is unknown. Several different processes are helpful in explaining the complex kinetics of humic colloid trace metal ion interaction, such as 1) variation in access to exchange with solution via continuous rearrangement of small entities/building blocks; 2) variation in access to exchange with solution via metal ion complexation-induced humic-humic association; 3) a spectrum of preexisting coordination sites with different complex coordinations, including the generation of chelates; 4) rearrangement of the humate backbone, amendment of acidity of residual functional groups and dehydration of oxygen containing functional groups as a consequence of metal ion complexation; 5) generation of polynuclear species eventually leading to associated

ORIGIN OF HUMIC COLLOIDS |

189

mineral structures in the nanometer size range; 6) association of pre-existing inorganic structures in the nanometer size range; 7) migration over a pre-existing humic colloid size spectrum with differences in the interaction kinetics; 8) migration between the surface and inner volume of humics and minerals; 9) coiling and stretching of assumed long linear humic molecules, which is not very likely given recent findings on the structure of humic substances; and 10) binding of natural trace metal ions by different mechanisms (as compared to later addition via solution) through inclusion during microbiological generation processes. Presumably, a combination of various processes will contribute, depending on the part of the kinetic spectrum considered. For the behavior of actinide ions in groundwater, the key question is to what extent they will behave as found in laboratory studies or gradually transfer into the behavior of natural humic colloid-bound trace metal ions. This is the key issue for application to long-term predictions of humic colloid mediated actinide transport. 20.5. CONCLUSIONS Humic substances play an important role in the mobility of trace metals, including actinides, in groundwater. Both the stabilization in the mobile water phase and retention processes are affected. There is great progress on a number of issues, whereas developments on other questions are still at a level of problem recognition. With respect to humic colloid mediated metal transport, humic colloids in natural aquifer systems under unperturbed conditions are transported with the groundwater flow with no sign of retention or decomposition. The situation in geochemical gradients from perturbed geochemical conditions is still an open question. The interaction of trace metal ions with aquatic humic substances is governed by complex kinetics. Interaction rates are found from milliseconds to basically irreversible binding. For reliable assessment of the impact of humic colloid mediated actinide transport one needs to 1) assess the humic colloid concentration over the entire time-span considered; 2) assume ideal tracer behavior of the humic substances, unless otherwise proven for a specific condition/site; and 3) ensure that trustworthy kinetic data are available for the actinide-humic colloid interaction. Application of equilibrium data underestimates trace metal transport. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

11. 12. 13. 14. 15.

OECD-NEA. Binding models concerning natural organic substances in performance assessment. Proceedings of an OECD-NEA Workshop organized in cooperation with the Paul Scherrer Institute, Bad Zurzach, Switzerland, 1995. Buckau G ed. Effects of humic substances on the migration of radionuclides: Complexation and transport of actinides (Final Report). Report EUR 19610 EN, 2000. Geckeis H, Rabung Th, Ngo Manh T, Kim JI, Beck HP. Humic colloid-borne natural polyvalent metal ions: Dissociation experiment. Environ. Sci. Technol., 2002; 36:2946–52. Artinger R, Schuessler W, Schaefer T, Kim JI. A kinetic study of Am(III)/humic colloid interactions. Environ. Sci. Technol., in press. Buckau G, Artinger R, Kim JI, Geyer S, Fritz P, Wolf M, Frenzel B. Development of climatic and vegetation conditions and the geochemical and isotopic composition in the Franconian Albvorland aquifer system. Appl. Geochem., 2000; 15:1191–1201. Buckau G, Artinger R, Fritz P, Geyer S, Kim JI, Wolf M. Origin and mobility of humic colloids in the Gorleben aquifer system. Appl. Geochem., 2000; 15: 171–9. Buckau G, Artinger R, Geyer S, Wolf M, Kim JI, Fritz P. 14C dating of Gorleben groundwater. Appl. Geochem., 2000; 15:583–97. Buckau G, Artinger R, Geyer S, Wolf M, Kim JI, Fritz P. Groundwater in-situ generation of aquatic humic and fulvic acids and the mineralization of sedimentary organic carbon. Appl. Geochem., 2000; 15:819–32. Buckau G, Wolf M, Geyer S, Artinger R, Kim JI. Impact of changes in land-use on groundwater recharge and concentration and origin of aquatic humic substances in the Gorleben SE aquifer system. Submitted. Buckau G, Hooker P, Moulin V, Schmeide K, Maes A, Warwick P, Moulin Ch, Pieri J, Bryan N, Carlsen L, Klotz D,Trautmann N. Main conclusions of the EC-humics Project: Effects of humic substances on the migration of radionuclides: Complexation and transport of actinides. In: Ghabbour EA, Davies G eds. Humic substances. Versatile components of plants, soils and water. Cambridge: Royal Society of Chemistry, 2000:235–60. Vilks P, Stroes-Gascoyne S, Goulard M, Haveman SA, Bachinski, DB. The release of organic material from clay-based buffer materials and its potential implications for radionuclide transport. Radiochim. Acta, 1998; 82:385–91. Dierckx A. Boom clay in-situ pore-water chemistry, Report: BLG-734, SCK/CEN, Mol, Belgium, 1997. Elie M, Faure P, Michels R, Landais P, Griffault L. Natural and laboratory oxidation of low-organic-carbon-content sediments: Comparison of chemical changes in hydrocarbons. Energy & Fuels, 2000; 14:854–61. Claret F, Schaefer T, Bauer A, Buckau G. Generation of humic and fulvic acid from Callovo-Oxfordian clay under high alkaline conditions. Submitted. Eichinger L. Bestimmung des alters von grundwasser mit kohlenstoff-14: Messung und interpretation der grundwässer des fränkischen albvorlandes. Dissertation, Ludwig-Maximilians-Universitat München, 1981.

190

| CHAPTER 20: INFLUENCE OF HUMIC SUBSTANCES ON THE MIGRATION OF ACTINIDES

16. 17. 18.

Choppin GR, Clark SB. The kinetic interactions of metal ions with humic acids. Mar. Chem., 1991; 36:27–38. Langford CH, Gutzman DW. Kinetic studies of metal ion speciation. Anal. Chim. Acta, 1992; 256:183–201. King SJ, Warwick P, Hall A, Bryan N. The dissociation kinetics of dissolved metal-humate complexes. Phys. Chem. Chem. Phys., 2001; 3:2080–85. Monsallier JM, Artinger R, Denecke M, Scherbaum FJ, Buckau G, Kim JI. Kinetics of complexation of An(III) by humic substances by spectroscopic investigations (XAFS and TRLFS). In preparation. Artinger R, Buckau G, Zeh P, Geraedts K, Vancluysen J, Maes A, Kim JI. Humic colloid mediated transport of tetravalent actinides and technetium. In preparation.

19. 20.

Chapter 21 CATALYTIC EFFECTS OF Ni-HUMIC COMPLEXES ON THE REDUCTIVE DEHALOGENATION OF C1 AND C2 CHLORINATED HYDROCARBONS Edward J.O’Loughlin,1,2 Huizhong Ma1 and David R.Burris1,3 1Air

Force Research Laboratory, AFRL/MLQR, Tyndall Air Force Base, Florida 32403–5323, USA

2Environmental

Research Division, Argonne National Laboratory, Argonne, IL 60439–4843, USA

3Integrated

Science and Technology, Inc., Panama City, FL 32401–2731, USA 21.1. INTRODUCTION

The redox chemistry of humic substances (HSs) is important in many processes relevant to the fate and transport of contaminants in terrestrial and aquatic environments. In particular, the role of HSs in the abiotic and microbially mediated transformation of organic and inorganic contaminants is of great current interest. Humic substances can serve as electron donors and terminal electron acceptors for microbial anaerobic respiration [1–5], as reductants for organic and inorganic contaminants [6–8], and as electron transfer mediators [9–11]. Although anoxic environments often are sufficiently reducing for the reductive transformation of many contaminants to be thermodynamically favorable, the transfer of electrons from reduced species to a given contaminant is often kinetically limited, resulting in slow transformation rates in situ. However, in the presence of materials that can act as electron transfer mediators, such as bacterial transition-metal coenzymes (e.g., vitamin B12 [Co], coenzyme F430 [Ni], and hematin [Fe]), rates for reductive dehalogenation reactions can be greatly enhanced [12–20]. In addition, several low-molar mass quinones, dissolved organic matter (DOM), and HSs are known to be effective electron mediators for the reduction of nitroaromatics, dioxins and chlorinated aliphatic hydrocarbons by reduced sulfur species, TiIII citrate, or FeIIaq [9–11,19,21,22]. The ability of HSs to act as electron transfer mediators has generally been attributed to the presence of quinone groups within their structures [9,10,23–25]. However, HSs complex a wide range of transition metals, and recently DOM- and HS-transition metal complexes (specifically Ni and Cu complexes) have been shown to facilitate the reductive dechlorination of trichloroethene (TCE) by TiIII citrate [26–28], suggesting that HS-transition metal complexes may also contribute to the electron transfer capacity of HSs. This paper examines the ability of HS-Ni complexes to mediate the reduction of chlorinated methanes, ethanes, ethenes and ethynes in homogeneous aqueous solutions with TiIII citrate as the bulk reductant. Reaction rates, products, and potential pathways are presented. 21.2. MATERIALS AND METHODS 21.2.1. Chemicals and Reagents Hexachloroethane (HCA, 99%), pentachloroethane (PCA, 95%), 1,1,2,2-tetrachloroethane (1122TeCA, 99%), 1,1,2trichloroethane (112TCA), tetrachloroethene (or perchloroethene (PCE), 99.9+%), TCE (99.5%), 1,1-dichloroethene (11DCE, 99%), cis-l,2-dichloroethene (c12DCE, 97%), trans-1,2-dichloroethene (t12DCE, 98%), chloromethane (CM, 99.5%), sodium citrate dihydrate (99%), nickel(II) chloride hexahydrate (99.9999%), Aldrich humic acid (AHA), propene (99+%), 1-butene (99+%), and a mixture of cis- and trans-2-butene (38.1% cis and 61.6% trans) were obtained from Aldrich. Tetrachloromethane (or carbon tetrachloride (CT), 99%) and 1,1,1,2-tetrachloroethane (1112TeCA, 99%) were purchased from Chem Service. Titanium(III) chloride (15% in HCl) and 1,1-dichloroethane (11DCA, >97%) were obtained from Fluka. Dichloromethane (DCM, spectroscopic grade), trichloromethane (or chloroform (CF), HPLC grade), and 1,1,1trichloroethane (111TCA, purified) were purchased from Fisher. A multi-component gas mix containing methane, ethane, ethene, and acetylene (each at 1.0 mole percent in N2); chloroethene (or vinyl chloride (VC), 0.1% in N2); propane (1.002% in N2); and n-butane (1.0% in N2) were purchased from Scott Specialty Gases. Tris (Trizma® base, 99.9%), and 1,2dichloroethane (12DCA, reagent grade) were obtained from Sigma and J.T.Baker, respectively. Chloroacetylene (CAc), synthesized as described by Arnold and Roberts [29], was generously provided by Dr. Lynn Roberts’ laboratory.

192

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

Aldrich HA was treated extensively to remove residual fulvic acids and inorganic impurities including metals [30]. Briefly, AHA was dissolved in 10−2 M NaOH and then treated with HF/HCl and Na saturated Na-Chelex 100 (Bio-Rad Laboratories). Excess Na was removed by placing the AHA solution in 1000 Da molar mass cutoff (MWCO) cellulose ester dialysis tubing and dialyzing against 18 M′ -cm water. Titanium(III) citrate (250 mM) was prepared in a glove box containing an atmosphere of 4–6% H2 in N2 with a modified literature method [31]. Briefly, 7.35 g of trisodium citrate and 4.0 g of Trizma® base were dissolved in 25 mL of deoxygenated 18 M′ -cm water. The solution was placed in an ice bath with continuous stirring, and 15 mL of 15% TiCl3 in HCl was added, followed by 3.2 mL of 50% NaOH (w/w) to raise the pH to ~8.0 at 25°C. This stock solution oxidizes over a period of several hours and thus was prepared fresh for each experimental run. 21.2.2. Methods Experimental Setup. The experimental system consisted of 160-mL serum vials, each with 90 mL of aqueous phase consisting of AHA at a concentration of 11.1 mg of organic carbon (OC) L−1 or 18 M′ -cm water (for controls without AHA). The solutions were spiked with 1.0 mL of 10 mM Ni2+ in 0.1 M HCl (or 1.0 mL of 0.1 M HCl for controls without Ni) and 0.1 mL of 1 M NaOH. After 24 h, 3 mL of 2.0 M Tris (adjusted to pH 8.0) and 6 mL of 250 mM TiIII citrate were added and the vials were sealed with aluminum crimp caps with Teflon-lined rubber septa. The Eh of systems containing TiIII citrate was initially −700 mV as measured with a Pt redox electrode. Vials were spiked with 5 µL of a 1.0 M methanolic solution of n-heptane (added as an internal standard). Unless otherwise indicated, all preparative work was performed in a glove box with an atmosphere of 4–6% H2 in N2 or under continuous Ar sparging. Reactions were initiated by spiking the vials with 5 µL of a 1.0 M methanolic solution of a given chlorinated hydrocarbon with the exception of PCA and HCA systems, which were spiked with 50 µL of 0.1 M methanolic solutions, and CM systems, which were spiked with 120 µL of CM gas at STP. Experimental systems were prepared in duplicate. Initial solution concentrations were AHA at 10 mg OC L−1, 100 µM Ni2+, 100 mM Tris, and 15 mM TiIII citrate; the total mass of chlorinated hydrocarbon in each vial was 5 µmol. Experiments were conducted at 20°C, and the systems were kept well mixed either by placement on a roller drum that rotated vertically while the bottle axis remained horizontal, or by vigorous shaking by hand (for reactions with half-lives of less than 5 min). At selected intervals, 100-µL headspace samples were removed from the serum vials and analyzed for the parent compound and transformation product concentrations. Analytical Methods. Nonchlorinated hydrocarbons; CAc; chloroethenes; mono-, di-, and trichloroethanes; and chloromethanes (except CT) were analyzed with a Hewlett-Packard 5890 Series II gas chromatograph equipped with a GSQ column (0.53 mm id by 30 m) and a flame ionization detector (FID, 200°C). Samples were injected in split mode at 180°C. Helium was used as the carrier gas at a flow rate of 4.7 mL min−1, with the split vent at 9.7 mL min−1. The oven temperature was held at 50°C for 2 min, ramped at 25°C min−1 to 200°C, and held for 10 min at 200°C. Carbon tetrachloride and tetra-, penta-, and hexachloroethane were analyzed with a DB-1 column (0.53 mm id by 30 m) and an FID (200°C). Samples were injected in split mode at 200°C. The carrier gas was He at 7.7 mL min−1 with the split vent at 5.2 mL min−1. Carbon tetrachloride was analyzed under isothermal conditions at 50°C. The oven temperature for 1112TeCA and 1122TeCA was held at 60°C for 1.2 min, ramped at 30°C min−1 to100°C, and held for 2.5 min at 100°C; for PCA the oven temperature was held at 60°C for 1 min, ramped at 30°C min−1 to 150°C, and held at 150 °C for 1 min; and for HCA the oven temperature was ramped from 90°C to 180°C at 30°C min−1 then held at 180°C for 1 min. The system was calibrated by equilibrating known masses of analytes and internal standard n-heptane in serum vials containing the same ratio of aqueous phase to vapor phase as in the experimental systems, thus accounting for water-vapor partitioning. Experiments to detect low-yield intermediates in the reduction of TCE and PCE were performed as described above, except that TCE and PCE loadings were higher (150 µmol). Headspace samples for these experiments were analyzed with a Hewlett Packard 5890 gas chromatograph equipped with a Hewlett Packard 5971 mass-selective (MS) detector. Samples were injected splitless at 150°C onto a 0.32 mm id by 30 m GSQ-PLOT column. The oven temperature was held at 40°C for 3 min, ramped at 25°C min−1 to 200°C, and held at 200°C for 8 min. The detector temperature was 280°C. To avoid swamping the detector, the acquisition parameters were set with a 3-min initial solvent delay, and the MS was placed offline for the time interval corresponding to the elution window of the internal standard (n-heptane). Selected ion monitoring (SIM) methods were used for CAc and dichloroacetylene with target ions of M+, M+2 and M+4. Analysis of carbon monoxide in headspace samples was performed with a Trace Analytical (Menlo Park, CA) RGA3 Reduction Gas Analyzer with an oven temperature of 266°C. Formate was analyzed with a Dionex LC20 ion chromatograph with self-regenerating suppressor control. An IONPAC AG11 guard column (4×50 mm) and an IONPAC AG11 1 analytical column (4×250 mm) were used with a weakly basic eluent (NaOH, 350 µmol L−1) at a flow rate of 1.0 mL min−1. Kinetic Analysis. Apparent pseudo-first-order rate constants (kobs) were obtained by fitting the data for the mass of parent chlorinated hydrocarbon in the system over time to Eq. 21.1,

MATERIALS AND METHODS |

193

Figure 21.1 Reduction of VC in systems containing 10 mg OC L−1 AHA, 100 µM Ni2+, 100mM Tris, 15 mM TiIII citrate, and 5 µmol VC as indicated. Plots are of data from duplicate vials. Curves represent first-order fits to the data.

(21.1) where t is time and MCHt and MCHO are the masses of chlorinated hydrocarbon in the system at times t and zero, respectively. For a reaction taking place in the solution phase only, the apparent pseudo-first-order rate constant that pertains to the reaction of a volatile constituent capable of rapid partitioning between the solution phase and the headspace (kobs) is equivalent to the rate constant that would be obtained in a headspace-free system (k’obs), as described by Burris et al. [12]. Values of k’obs were calculated with “dimensionless” Henry’s law constants for 11DCA, 111TCA, VC, 11DCE, c12DCE, t12DCE, TCE, PCE, CF, and CT from Gossett [32]; 112TCA, 1112TeCA, and 1122TeCA from Tse et al. [33]; PCA from Mackay and Shiu [34]; HCA from Pankow et al. [35]; acetylene and ethene from Burris et al. [13]; and CAc from Semadeni et al. [36]. 21.3. RESULTS AND DISCUSSION 21.3.1. Reaction Kinetics The reduction of chlorinated ethynes, ethenes, and ethanes was markedly enhanced in the presence of Ni-AHA complexes (as shown for VC in Figure 21.1), an observation that is consistent with earlier reports that Ni-DOM complexes (including Ni-HS complexes) are effective electron transfer mediators for the reductive dehalo genation of TCE with TiIII citrate as the bulk electron source [26–28]. The reduction of chlorinated hydrocarbons (i.e., those for which reduction was observed) in our experimental systems was well described by pseudo-first-order kinetics, as shown for VC (Figure 21.1). Apparent pseudofirst-order rate constants (k’obs), the coefficients of determination, adjusted for the number of degrees of freedom (adj r2), and reaction half-lives (t1/2) are listed in Table 21.1. Although most chlorinated hydrocarbons were reduced in the presence of TiIII citrate alone, the reduction rate was much higher in the presence of Ni-AHA complexes, while a more moderate enhancement was observed in systems containing Ni without AHA. The reduction of chlorinated hydrocarbons in the presence of TiIII citrate and AHA was the same as with TiIII citrate alone (data not shown), indicating that AHA in the absence of added Ni was not an effective electron transfer mediator for these reactions. 21.3.2. Reaction Products and Pathways General Observations. The reduction of the parent chlorinated hydrocarbons typically resulted in fully dechlorinated products (Table 21.2). Loss of analytes due to sorption on the Teflon-lined septa and volatilization from the serum vials was minimal (<5%) over the time scale of these experiments (typically less than 2 d, but no longer than 14 d); thus, the incomplete carbon recovery observed for many of the compounds suggests the possible formation of products that are not identifiable by our analytical methods (e.g., compounds with relatively low volatility). The results for individual compounds are discussed in the following sections. Chlorinated Methanes. The reduction of CT initially resulted in the formation of CF, which was subsequently reduced to DCM, CM, carbon monoxide, methane, and a suite of C2–C4 nonchlorinated alkanes and alkenes (ethane, ethene, propane, propene, n-butane, 1-butene, and cis- and trans-2-butene) (Table 21.2). Volatile products accounted for approximately 71% and 69% of the carbon introduced as CT and CF, respectively (per mole C), two-thirds of which were nonchlorinated

194

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

hydrocarbons (primarily methane). Approximately 29% and 31% of the carbon added to the system as CT and CF was not accounted for in the mass balance of the identified products; sorption of the identified products to components of the reaction vessel and Table 21.1. Apparent pseudo-first-order rate constants (k’obs) and half-lives for the reductive transformation of chlorinated methanes, ethynes, ethenes, and ethanes. Compound Methanes Chloromethane (CM) Dichloromethane (DCM) Chloroform (CF) Carbon Tetrachloride (CT) Ethynes Acetylene

Chloroacetylene (CAc) Ethenes Ethene

Vinyl Chloride (VC) 1, 1 -Dichloroethene (11DCE) cis-1,2-Dichloroethene (c12DCE) trans-1,2-Dichloroethene (t12DCE)

Compound Trichloroethene (TCE) Tetrachloroethene (PCE) Ethanes Chloroethane (CA) 1,1 -Dichloroethane (11DCA) 1,2-Dichloroethane (12DCA) 1,1,1 -Trichloroethane (111TCA)

System

k’obs(s−1)

adj r2

Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti

NRa

t½ (min)

NR NR NR NR NR 3.74(±0.17)b×10–4 7.98(±0.28)×10–5 6.30(±1.01)×10–6 2.23(±0.08)×10–2 6.44(±0.35)×10–3 5.44(±0.21)×10–3 4.51(±0.07)×10–3 8.51(±0.19)×10–4 1.12(±0.11)×10–4 5.48(±0.25)×10–3 8.97(±0.44)×10–4 1.12(±0.11)×10–4 7.88(±0.63)×10–5 4.63(±0.09)×10–5 4.06(±0.41)×10–7 1.27(±0.046)×10–3 2.16(±0.l2)×10–4 3.13(±0.52)×10–6 7.45(±0.11)×10–4 2.74(±0.14)×10–5 6.49(±0.80)×10–6 6.11(±0.28)×10–4 1.70(±0.023)×10–4 5.43(±0.51)×10–6 4.13(±0.16)×10–4 2.30(±0.01)×10–4 1.47(±0.22)×10–5

0.950 0.994 0.794 0.992 0.983 0.991 0.997 0.998 0.970 0.997 0.991 0.991 0.969 0.999 0.998 0.968 0.997 0.893 0.996 0.993 0.957 0.971 0.999 0.958 0.976 0.999 0.958

41 145 1830 0.52 1.8 2.1 2.6 14 103 2.1 13 26 146 249 28400 9.1 53 3690 16 422 1780 18 68 2130 28 50 784

System

k’obs(s−1)

adj r2

t½ (min)

Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni

6.34(±0.46)×10−4

0.949 0.949 0.949 0.946 0.986 0.981

18 292 1230 107 1420 6800

0.865

32200

0.884 0.992

8.7 19

3.96(±0.39)×10−5 9.40(±0.89)×10−6 1.08(±0.038)×10−4 8.12(±0.48)×10−6 1.70(±0.11)×10−6 NR NR NR 3.58(±0.42)×10−7 NR NR NR NR NR 1.33(±0.13)×10−3 6.08(±0.37)×10−4

MATERIALS AND METHODS |

Compound

System Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti Ti+Ni+AHA Ti+Ni Ti

1,1,2-Trichloroethane (112TCA) 1,1,1 ,2-Tetrachloroethane (1112TeCA) 1,1,2,2-Tetrachloroethane (1122TeCA) Pentachloroethane (PCA) Hexachloroethane (HCA) aNon-reactive; bRate

k’obs(s−1) 8.07(±0.76)×10−5 2.51(±0.18)×10−5 7.63(±0.25)×10−6 NR 1.25(±0.85)×10−2 5.74(±0.26)×10−3 5.71(±0.18)×10−3 1.90(±0.075)×10−4 4.64(±0.23)×10−5 8.23(±0.70)×10−6 2.44(±0.31)×10−2 9.54(±1.08)×10−3 7.58(±0.38)×10−3 3.75(±0.47)×10−2 1.50(±0.17)×10−2 2.01(±0.19)×10−2

adj r2 0.969 0.965 0.990

t½ (min) 143 460 1510

0.981 0.992 0.996 0.972 0.978 0.939 0.983 0.967 0.991 0.991 0.985 0.994

0.93 1.9 2.1 61 249 1400 0.47 1.2 1.5 0.31 0.77 0.57

195

constant±standard error.

Table 21.2. Initial transformation reactions and products, observed intermediates, terminal products, and carbon recovery from the Ni-AHAmediated reductive transformation of chlorinated methanes, ethynes, ethenes, and ethanes. Compound Initial reduction reaction and product Observed intermediates Terminal productsa

Carbon recoveryb

Methanes CF

Unknown

unknown

69%

CT

hydrogenolysis′ CF

CF

methane (38%), COc (2%), DCM (16%), CM(6%), C2–C4 hydrocarbons (7%) methane (34%), CO (2%), DCM (10%), CM (3%), C2–C4 hydrocarbons (22%)

Ethynes Acd

bond reduction′ EEe

EE

94%

CAc

hydrogenolysis′ Ac

Ac, VC, EE

EE & EAf (79%), C3–C6 hydrocarbons (15%) EE & EA (77%), C3–C6 hydrocarbons (15%)

Ethenes Ethene VC 11DCE

bond reduction′ EA hydrogenolysis′ EE ′ -elimination′ EE

EE (4%), EA (94%) EE & EA (95%) EE & EA (89%), C3–C6 hydrocarbons (9%) EE & EA (92%), C3–C6 hydrocarbons (5%) EE & EA (89%), C3–C6 hydrocarbons (9%) EE & EA (72%), C3–C6 hydrocarbons (14%)

98% 95% 98%

EE & EA (78%), C3–C6 hydrocarbons (11%)

89%

EA (5%), 11DCA (92%), CA (1%)

98%

c12DCE t12DCE TCE

PCE Ethanes 11DCA

hydrogenolysis′ VC ′ -elimination′ Ac hydrogenolysis′ VC ′ -elimination′ Ac hydrogenolysis′ c/t12DCE hydrogenolysis′ 11DCE ′ -elimination′ CAc hydrogenolysis′ TCE ′ -elimination′ DCAc

EE EE VC, Ac, EE VC, Ac, EE c/t12DCE, 11DCE, CAc, Ac, VC, EE TCE, DCAc,g EE

′ -elimination′ EA hydrogenolysis′ CA

Compound

Initial reduction reaction and product

Observed intermediates

Terminal productsa

111TCA

hydrogenolysis′ 11DCA ′ -elimination′ CA

11DCA, EE

EE & EA (30%) 98% C3–C6 hydrocarbons (2%) 11DCA (14%) CA (52%)

71%

92%

97% 98% 86%

Carbon recoveryb

196

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

Compound

Initial reduction reaction and product

Observed intermediates

112TCA 1112TeCA

′ -elimination′ VC ′ -elimination′ 11DCE

VC, EE 11DCE, EE

Terminal productsa

Carbon recoveryb

EE & EA (90%) 90% EE & EA (87%), C3–C6 93% hydrocarbons (6%) 1122TeCA ′ -elimination′ c/ c/t12DCE, VC, EE EE & EA (89%), C3–C6 92% t12DCE hydrocarbons (3%) PCA ′ -elimination′ TCE TCE, EE EE & EA (76%), C3–C6 89% hydrocarbons (13%) HCA ′ -elimination′ PCE PCE, TCE, EE EE & EA (79%), C3–C6 90% hydrocarbons (11%) aIncluding, where appropriate, unreacted parent compound; bPercent of C derived from parent compound recovered as identified products and unreacted parent compound; cCarbon monoxide; dAcetylene; eEthene (EE); fEthane (EA); gSuspected but not confirmed.

loss due to leakage from the reaction vessels were minimal, suggesting that there are other significant products that are not detected by our analytical methods. The initial step in the reduction of halogenated hydrocarbons is commonly reported to be a single electron transfer with the concerted or stepwise cleavage of the C-Cl bond resulting in the formation of a radical [37–42]. For CT, this process results in the formation of a trichloromethane radical [43–45] that is unstable and can react by several mechanisms, including 1) accepting a second electron and abstracting a proton to form CF; 2) coupling with a second trichloromethane radical, forming HCA; and 3) accepting a second electron with the subsequent loss of an additional chlorine, forming dichlorocarbene. Chloroform was observed as the principal initial product of CT reduction; however, traces of carbon monoxide were also observed, which may indicate the formation of a dichlorocarbene (or carbenoid) intermediate; dichlorocarbene reacts readily with water to form carbon monoxide and formic acid [46]. Formate was not present at detectable levels (i.e., >2 µM) in any of our experiments. However, Robinson [47] showed that carbon monoxide is the primary product of the reaction of dichlorocarbene with water and that formate is produced by a subsequent (and slower) reaction of carbon monoxide with hydroxide. In addition to reduction of trichloromethyl radicals, dichlorocarbene may also be produced by the hydrolysis of CF [46], which likely accounts for the carbon monoxide formed during the reduction of CF. The reduction of CT and CF to methane may result from a series of sequential replacements of hydrogen for chlorine, such that methane. However, experiments independently examining the reduction of DCM and CM showed that neither DCM nor CM was reduced over the time scale of the experiments. Moreover, once formed from the reduction of CT or CF, DCM and CM concentrations remained stable, indicating that these compounds are not intermediates in the Ni-AHA mediated reduction of CT and CF to methane. Thus, the formation of methane from the reductive dehalogenation of CT and CF must result from processes other than sequential hydrogenolysis. Castro and Kray [40] have proposed a pathway for the reduction of CT to methane by CrII inaqueous solution that does not require DCM and CM intermediates. The proposed pathway involves a series of single electron transfers in which chloromethyl radicals accept an electron with the subsequent loss of a chlorine, forming the corresponding carbene/carbenoid, which can then be reduced to a methyl radical containing one less chlorine atom (Figure 21.2). The process can be repeated until the last chlorine is removed, resulting in carbene, which can be reduced to a methane radical and finally to methane via the stepwise transfer of two electrons and two protons. Although such a reaction scheme does not involve DCM or CM as an intermediate, these species may be formed by terminal side reactions in which dichloromethane radicals or chloromethane radicals accept an additional electron while abstracting a proton. Chlorinated Ethynes. The reduction of CAc in our experimental systems was extremely rapid. Acetylene was the dominant product of the reduction of CAc mediated by Ni-AHA complexes, indicating that reduction of CAc occurs primarily via hydrogenolysis. Chloroacetylene was also reduced to VC, though to a much smaller extent. The terminal products of CAc reduction mediated by Ni-AHA complexes (that is, ethane, ethene, and C3–C6 alkenes and alkanes) are consistent with the products observed for the reduction of acetylene and VC (as described in the following section). Acetylene and VC have been reported as products of CAc reduction by both Zn0 [29] and vitamin B12 with TiIII citrate [36]. However, VC was not observed as an intermediate in the reduction of CAc by Fe0 [48]. The formation of C3–C6 hydrocarbons was observed during the reduction of CAc by Fe0 [48], but not in CAc reduction by Zn0 [29] or cobalamin [36]. Acetylene was rapidly reduced to ethene, which was subsequently reduced to ethane, though at a much lower rate. In addition to ethene and ethane, the reduction of acetylene was accompanied by formation of longer-chained (C3–C6) alkenes and alkanes, primarily n-butane, 1-butene, cis-2-butene, and trans-2-butene. Independent experiments examining the reduction of ethene revealed that ethane is the only product of ethene reduction in our experimental system, indicating that the formation of C3–C6 alkene and alkane products from reduction of acetylene and CAc occurs prior to the reduction of ethene.

MATERIALS AND METHODS |

197

Figure 21.2 Potential reaction scheme for the reduction of CT and CF to CO, DCM, CM, and methane. Boxes indicate observed intermediates/products. The reduction of MC to CM or methane and CM to methane was not observed. For simplicity, reactions of carbenes with water and pathways leading to the formation of coupling products (C2–C4 alkanes and alkenes) are not shown.

The formation of C3–C6 hydrocarbons also has been observed during the reduction of acetylene by Fe0 in aqueous sys tems [48]; however, coupling products were not observed during the reduction of acetylene by Zn0 [29] or vitamin B12 with TiIII citrate [36]. Chlorinated Ethenes. Vinyl chloride was readily reduced via hydrogenolysis to ethene, which was further reduced to ethane. No coupling products were observed, and ethene and ethane accounted for nearly all of the carbon introduced as VC. The vicinal dichloroethenes c12DCE and t12DCE were reduced to ethene, ethane, and minor amounts of C3–C6 hydrocarbons. Acetylene and VC were observed as intermediates, indicating that the initial reduction of both 12DCE isomers occurs by both hydrogenolysis (forming VC) and ′ -elimination (forming acetylene); however, the relative significance of each process was isomer specific. Although the reaction rates for c12DCE and t12DCE were quite similar, the maximum yield of VC was 12% for c12DCE but only 0.3% for t12DCE. Moreover, the yields of the coupling products (which were observed during the reduction of acetylene but not VC) were 1.8% for c12DCE and 3.9% for t12DCE, values consistent with the maximum

198

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

Figure 21.3 Potential reaction scheme for the reduction of 11DCE to ethene and ethane. Acetylene and VC were not observed in our experimental system as intermediates in the reduction of 11DCE mediated by Ni-AHA complexes. The coupling of radical and/or carbene intermediates is not shown. Likewise, this pathway does not show reaction of ethenylidene with the solvent (water).

acetylene yields (0.3% and 0.6%, respectively). Thus, reductive ′ -elimination appears to be more significant for t12DCE than for c12DCE (for which hydrogenolysis was dominant) in our experiments as well as for reduction of these compounds by Fe0 and Zn0 [29,48,49]. Reduction of 11DCE resulted in the formation of ethene, ethane, and minor amounts of C3–C6 coupling products. Acetylene and VC have been observed as intermediates in 11DCE reduction by vitamin B12 (with TiIII citrate as the bulk reductant) [16,36], and VC (but not acetylene or coupling products) is an intermediate in 11DCE reduction by both Fe0 and Zn0 [29,48]; however, neither acetylene nor VC was observed as an intermediate in our system. Although the reductive dechlorination of 11DCE via hydrogenolysis (resulting in the formation of VC) could be considered a likely initial reaction in the reduction of 11DCE to ethene, this does not appear to occur in our system. Figure 21.3 shows a potential reaction scheme for the reduction of 11DCE to ethene (and ultimately ethane). As previously discussed, the initial step in the reduction of halogenated hydrocarbons is commonly reported to be a single electron transfer, with the concerted or stepwise cleavage of the C-Cl bond resulting in the formation of a radical. For 11DCE, this process results in the formation of a VC radical, which is unstable and can react by a number of mechanisms, including 1) accepting a second electron and abstracting a proton to form VC; and/or 2) accepting a second electron with loss of another chlorine to form ethenylidene. Given the rate constant for VC reduction relative to that for 11DCE reduction (Table 21.1), if VC were a significant intermediate in the reduction of 11DCE in our experiments, it would have accumulated to detectable levels. Thus, reduction of 11DCE to ethene via VC is not a significant pathway in our system. A more likely route involves the formation of ethenylidene (which likely would be present as a carbenoid, not a free carbene). Ethenylidene could undergo rearrangement to form acetylene; however, acetylene was not observed as an intermediate. Alternatively, with the transfer of an electron and a proton, ethenylidene could be reduced to an ethenyl radical, that, with an additional transfer of an electron and proton, would form ethene. The net result, the reduction of 11DCE to ethene via a reductive ′ -elimination reaction, does not require VC or acetylene as an intermediate. A mechanism of this type was proposed by Arnold and Roberts [48] to explain the kinetics of ethene formation during the reduction of 11DCE by Fe0. The reaction scheme in Figure 21.3 does not indicate pathways for the formation of the observed coupling products or hydrolysis of ethenylidene to acetaldehyde (which was not measured). Acetylene, ethene, ethane, and C3–C6 hydrocarbons were the only intermediates and products of TCE reduction observed in the experiments with low (5 µmol) initial TCE loading. However, in experiments with initial TCE loadings of 150 µmol, minor amounts of c12DCE, 11DCE, and VC were observed, as well as trace amounts of t12DCE and CAc (in addition to the

MATERIALS AND METHODS |

199

Figure 21.4 Proposed pathways for the reduction of chlorinated ethynes, ethenes, and ethanes mediated by Ni-AHA complexes, with TiIII citrate as the bulk reductant, based on observed products/intermediates. Dashed arrows designate reactions that were not directly observed. Pathways leading to the formation of coupling products (C3–C6 alkanes and alkenes) are not shown.

products mentioned above). These results are consistent with the intermediates and products previously reported for reduction of 16 µmol of TCE mediated by Ni-AHA complexes under similar experimental conditions [28]. The formation of the three DCE isomers as well as CAc indicates that TCE reduction occurs via both hydrogenolysis and reductive ′ -elimination. Overall, the intermediates and product distribution for TCE reduction mediated by Ni-AHA complexes are similar to those reported for the reduction of TCE by Zn0 [29], Fe0 [48], and cobalamin with TiIII citrate [12]. Trichloroethene and ethene were the only intermediates observed during the reduction of PCE at both PCE loadings (5 and 150 µmol), and the final products were ethene, ethane, and C3–C6 hydrocarbons. Although TCE was the major chlorinated intermediate observed during the reduction of PCE, no chlorinated intermediates of TCE reduction were detected. However, given the relatively low rate of PCE reduction compared to reduction of TCE and its reduction products (Table 21.1), it is plausible that the intermediates did not accumulate to detectable levels. As with the formation of acetylene from c12DCE and t12DCE, and CAc from TCE, reductive ′ -elimination of PCE would result in the formation of dichloroacetylene (DCAc). Although DCAc has not been directly observed as an intermediate in PCE reduction, kinetic modeling of the reduction of PCE by Zn0 and Fe0 provides indirect evidence for DCAc as an intermediate in PCE reduction by these metals [29,48]. The reduction of DCAc by zero-valent metals (e.g., Zn0 and Fe0) is extremely rapid and results primarily in the formation of CAc and t12DCE (and subsequently their daughter products), consistent with its implied role as a highly transient intermediate. In our experiments, a minor peak whose retention time was consistent with the theoretical retention time of DCAc (as predicted on the basis of its boiling point), was observed during PCE reduction in the control systems (TiIII citrate alone and with Ni (but without AHA)). The peak was below an acceptable S/N ratio for reliable confirmation; however, the mass spectrum of this peak was consistent with the expected spectrum for DCAc. Therefore, although the formation of DCAc from the Ni-AHA-me diated reduction of PCE was not directly observed, the tentative identification of DCAc in the controls and the highly transitory nature of DCAc under our experimental conditions suggest that the reduction of PCE to DCAc is likely. As is evident from the preceding discussion, the reduction of polychlorinated ethenes involves combinations of parallel and serial reactions. Figure 21.4 shows a pathway depicting the relevant reactions for the reduction of chlorinated ethenes. Pathways for formation of coupling products (C3–C6 hydrocarbons) are not shown in Figure 21.4 but are discussed later. Chlorinated Ethanes. No measurable reduction of CA or 12DCA occurred in any of the systems examined over the duration of the experiments (14 d). Reduction of 11DCA was observed only in the presence of Ni-AHA; however, after 72 h, only 8% was reduced, resulting in the formation of ethane and a minor amount of CA. The formation of CA and ethane from the reduction of 11DCA is consistent with a series of sequential hydrogenolysis reactions . However, as mentioned above, CA was not reduced under our experimental conditions, and therefore it is unlikely to be an intermediate in the reduction of 11DCA to ethane. A potential reaction scheme for the reduction of 11DCA to ethane (without CA as an intermediate) is shown in Figure 21.5. As with the reaction schemes proposed for CT and 11DCE, the initial step in 11DCA reduction involves a single electron transfer with the consecutive or concerted cleavage of the C-Cl bond, resulting in the formation of a

200

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

Figure 21.5 Potential reaction scheme for the reduction of 11DCA and 111TCA to ethene and ethane. Chloroethane was not reduced, and VC was not observed as an intermediate in the reduction of 111TCA mediated by Ni-AHA complexes in our experimental system. The coupling of radical and/or carbene intermediates is not shown. Likewise, this pathway does not show reaction of carbenes with the solvent (water).

chloroethane radical. This radical can accept a second electron and a proton, form ing CA and terminating the reaction, or the radical can accept another electron with the release of a second chloride, forming ethanylidene (or the corresponding carbenoid). The rearrangement of this carbene results in ethene. Trace levels of ethene were observed only during the initial hours of 11DCA reduction; however, because the reduction of ethene to ethane is over 100 times faster than the reduction of 11DCA, ethene might indeed be a significant intermediate. In addition to undergoing rearrangement, the carbene might form ethane (without ethene as an intermediate) by the sequential (Figure 21.5) or concerted transfer of two electrons and two protons. Chloroethane, 11DCA, and ethane were the major products of the reduction of 111TCA, along with minor amounts of ethene and C3–C6 hydrocarbons. As with 11DCA, the major products of 111TCA reduction can be accounted for by a series of sequential hydrogenolysis reactions . However, the rates of 11DCA and CA reduction observed independently do not support such a reaction sequence; specifically, CA is not reduced, and the kinetics of 11DCA reduction do not account for the observed rates of CA A and ethane accumulation. A reaction scheme consistent with the observed products and kinetics of 111TCA reduction in our system is shown in Figure 21.5. For 111TCA, a single electron transfer with the stepwise or concerted cleavage of a C–Cl bond results in the formation of a dichloroethane radical. The transfer of another electron and a proton to this radical forms 11DCA, which can react further, as described previously. Transfer of a single electron to the dichloroethane radical and loss of chloride results in chloroethanylidene (or its corresponding carbenoid). Rearrangement of chloroethanylidene gives VC, which subsequently is reduced to ethene and ultimately to ethane. However, we did not observe VC during the reduction of 111TCA, suggesting that this is not a significant pathway. Alternatively, chloroethanylidene can accept an electron and a proton, forming a chloroethane radical. As previously discussed, the chloroethane radical might react further, forming CA or the corresponding carbene/carbenoid (which ultimately results in ethene and ethane). The reaction scheme in Figure 21.5 is based primarily on a series of single electron transfers in which an ′ -haloethyl radical is reduced (with the loss of chloride) to the corresponding carbene. The carbene undergoes rearrangement or is reduced (with a proton transfer) to an ethyl radical containing one less chlorine. If the radical is chlorinated, the cycle can continue. Castro and Kray [40] proposed this type of process to explain the kinetics and product distributions observed during the reduction of geminal halides (including 111TCA) by CrII in aqueous solution. This reaction scheme was also invoked by Fennelly and Roberts [50] to explain the kinetics and product formation observed with the reduction of 111TCA by Fe0. The reaction scheme in Figure 21.5 does not depict reactions for the hydrolysis of chloroethanylidene and ethanylidene to acetaldehyde and ethanol (which were not measured), respectively.

MATERIALS AND METHODS |

201

The remaining chlorinated ethanes are vicinal halides and as such are susceptible to reductive ′ -elimination, forming the corresponding chlorinated ethene. Indeed, the only products observed from the initial reduction of 112TCA, 1112TeCA, 1122TeCA, PCA, and HCA were VC, 11DCE, c12DCE/t12DCE, TCE, and PCE, respectively. Moreover, hydrogenolysis, ′ -elimination, and dehydrohalogenation reactions, which have been observed as pathways for the dechlorination of vicinal chloroethanes under certain conditions [51–53], were not significant for these compounds in our system. The observed intermediates and final products were consistent with the daughter products observed from chlorinated ethene reduction, as described in the previous section. A pathway showing the various parallel and serial reactions involved in the reductive dechlorination of chlorinated ethanes and daughter products is given in Figure 21.4. Coupling Products. Compounds with carbon chains longer than the parent material are often identified as products of halogenated hydrocarbon reduction by reduced transition metal species [28,50,54–57], and they are generally attributed to radical coupling reactions [50,54,56,57]. Coupling products accounting for 2% to 22% of the carbon introduced as parent compounds were observed during the reduction of Ac, CAc, and all reducible polychlorinated hydrocarbons examined in this study with the exception of 11DCA. Ethene and ethane were the dominant coupling products observed from the reduction of CF and CT, with lesser amounts of propane, propene, n-butane, 1-butene, and cis and trans-2-butene. For the C2 hydrocarbons the majority (87% to 97%) of the coupling products were the C4 hydrocarbons n-butane, 1-butene, and cis- and trans-2-butene, with the remainder consisting of methane, propane, propene, and C5 and C6 alkanes and alkenes. Coupling products were observed only in systems containing both a reductant (TiIII citrate) and an appropriate parent compound (i.e., CF, CT, acetylene, CAc, or polyhalogenated ethene or ethane). Conversely, coupling products were not observed in systems containing TiIII, TiIII+Ni, TiIII+AHA, or TiIII+Ni+AHA to which an appropriate parent compound was not added, suggesting that the coupling products are derived from intermediates in the reduction of the parent compound and not from other Ccontaining compounds in these systems [58,59]. Formation of the proposed radical and carbene intermediates in Figures 21.2, 21.3 and 21.5 suggests a number of processes for the formation of C2–C6 hydrocarbons, including the coupling of radicals, the coupling of carbenes, and carbene insertion into C-H bonds. Although alkyl radicals have very short lifetimes, they may be stabilized by complexation with transition metals, thereby increasing the probability of radical coupling reactions [60]. The coupling of potential radical intermediates resulting from the reduction of CT and CF (Figure 21.2) could result in the formation of ethane and a suite of chlorinated ethanes ranging from CA to HCA. Coupling of radical intermediates likely to form during the degradation of polychlorinated ethenes would result in polyhalogenated 1,3-butadienes, while coupling of 11DCA and CA radicals from the reduction of 111TCA would form 2,2,3,3-tetrachloro-, 2,2,3-trichloro- and 2,3-dichlorobutane. Although no chlorinated coupling products were measured in our study, the observed C2–C4 hydrocarbons are plausible products of their partial reduction or from coupling of nonchlorinated radicals (i.e., methyl, ethyl, and ethenyl radicals). Coupling of carbenes typically is considered to be unlikely, largely because of their highly reactive nature and the correspondingly low probability of two carbene molecules interacting, as well as rapid dimer dissociation resulting from high excess internal energy [61,62]. Although these limitations are consistent with conditions in vapor-phase reactions, in our experimental system the proposed carbene intermediates would likely exist not as free carbenes but as transition metal organocomplexes (carbenoids), which may increase their relative stability. Moreover, the presence of solvent molecules also offers a means of dissipating the excess internal energy of potential coupling products, lowering the likelihood of dissociation and increasing the opportunity for carbene-coupling reactions. Indeed, several studies have suggested that carbene coupling does occur in systems that promote the stability and/or physical proximity of carbenes/carbenoids [63–66]. The proposed reaction schemes for CF, CT, 11DCE, and 111TCA (Figures 21.2, 21.3 and 21.5) involve carbene/carbenoid intermediates. The coupling of these species could form products that, on further reduction, yield C2 and C4 hydrocarbons, consistent with the observed products. Note that coupling products were not observed during the reduction of 11DCA, suggesting that coupling of ethanylidene was not significant. The formation of C2–C4 hydrocarbons from the reduction of chlorinated methanes and C3– C6 hydrocarbons from chlorinated C2 hydrocarbons might also have been the result of carbene insertion into C-H bonds of smaller-chain hydrocarbons (for example, ethane+carbene ′ propane). Several transition metal complexes (including Ni complexes) are known to pro mote the coupling of acetylene [67], resulting in the formation of vinyl acetylene (VAc). Vinyl acetylene was not measured in our experiments; however, the C4 coupling products observed during the reduction of acetylene (butene isomers and n-butane) are consistent with expected daughter products from the incomplete reduction of VAc. The C4 coupling products produced during acetylene reduction were also observed as products of CAc and polychlorinated ethene reduction. Moreover, acetylene was an observed or theoretical intermediate in the reduction of CAc and the polychlorinated ethenes. However, the differences in the C4 product profiles of acetylene, CAc, and the polychlorinated ethenes (Figure 21.6) suggest that they are not solely the products of acetylene coupling and that other mechanisms must be involved (particularly in the case of 111TCA, for which acetylene is not a plausible intermediate). This point is best illustrated by the differences in the ratios of cis- and trans-2-butene. Independent experiments showed that 1-butene is reduced to n-butane over the time scale of these experiments, while the 2-butene isomers are not reduced.

202

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

Figure 21.6 Profiles of C4 coupling products observed during the reduction of acetylene (Ac), CAc, DCE isomers, TCE, and PCE, with the respective ratios of cis-2-butene to trans-2-butene. The fractions of individual C4 products are relative to the total mass of C4 products. The C4 distribution for 111TCE is not shown because of an inability to resolve the peaks for cis-2-butene and CA.

Figure 21.7 Proposed role of organo-Ni complexes as electron mediators in the reduction of chlorinated hydrocarbons.

Although additional experiments are needed to identify specific intermediates and pathways for coupling products observed in this study, the C4 profiles for acetylene, CAc, and the polychlorinated ethenes (Figure 21.6) might suggest the relative significance of specific intermediates in the reduction of PCE, TCE, and the DCE isomers. In particular, the high cis- to trans-2-butene ratios (c:t) for PCE and TCE, relative to c12DCE and t12DCE, are consistent with the previously noted roles of CAc and 11DCE (which have comparatively high c:t ratios). Likewise, the c:t ratios for c12DCE and t12DCE are similar to that of acetylene, an intermediate in their reduction to ethane. 21.3.3. Mechanistic Considerations A number of organo-Ni complexes are effective agents for the reductive dehalogenation of halogenated hydrocarbons, including 1,4,8,11-tetramethyl-1,4,8,11-tetraaza-cyclotetra-decane nickel (Ni(tmc)) [38,54], nickel octaethylisobacteriochlorin (NiOEiBC) [57], and coenzyme F430 [15,17], a Ni porphinoid found in all methanogenic bacteria. Indeed, it has been suggested that coenzyme F430 is involved in the reductive dehalogenation of chlorinated hydrocarbons by methanogenic bacteria [68–71]. The reduction of halogenated hydrocarbons by organo-Ni complexes proceeds by reduction of the NiII complex to the corresponding NiI complex by a suitable bulk reductant. Typically, the NiI complex reduces a halogenated aliphatic hydrocarbon through a single electron transfer and the stepwise or concerted loss of a halide, resulting in the formation of the corresponding NiII complex and an alkyl or alkenyl radical (as either a free radical or a radical organo-Ni complex) that is further reduced by a second NiI complex. In the presence of additional bulk reductant, the NiI complex can be regenerated, thus allowing the Ni complex to act as an electron shuttle or mediator for the reaction (Figure 21.7); such is the case for the reduction of chlorinated hydrocarbons mediated by coenzyme F430 with TiIII citrate as the bulk reductant [15,17]. The enhanced reduction of chlorinated hydrocarbons observed in our systems containing Ni2+ and AHA, in which Ni-AHA complexes have been identified by Ma et al. [26] as the probable active mediator species, may involve a redox cycling of Ni analogous to that discussed above for coenzyme F430. However, attempts to identify NiI species in our system by electron paramagnetic resonance (EPR) spectroscopy were inconclusive because of the masking of any potential NiI signal by a broad, intense TiIII peak. The initial step in the reduction of halogenated hydrocarbons by transition metal species is commonly reported to be a single electron transfer with the stepwise or concerted loss of chloride, resulting in the formation of a radical [39–42]. Indeed, this appears to be the case for chlorinated hydrocarbon reduction by NiI organocomplexes [38,57]. For reduction reactions that occur by means of an outer-sphere single electron transfer mechanism, the initial electron transfer is often rate-limiting [72]. Thus, one might reasonably expect good correlation between the reaction rates and one-electron reduction potentials (E1). These potentials can be determined experimentally or calculated from thermodynamic data; however, significant

MATERIALS AND METHODS |

203

complications are associated with direct experimental determination of E1 values, and E1 values estimated from thermodynamic parameters are limited by the quality of the thermodynamic database [73]. Many researchers have used the homolytic C-Cl bond dissociation energy (BDE) or the energy of the lowest unoccupied molecular orbital (ELUMO) as surrogates for E1 [73–77]. For reductive dechlorination reactions, the homolytic BDE refers to the energy required to dissociate C-Cl to C· and Cl·, and as such it reflects the ease of radical formation. The LUMO is the frontier molecular orbital into which electron transfer takes place; consequently, ELUMO is directly related to the electron affinity of the molecule, and thus the energy of this orbital is a significant component of the overall driving force of the reaction. The relationships between the observed pseudo-first-order rate constants for the reduction of chlorinated methanes and ethanes mediated by Ni-AHA complexes and E1, BDE, and ELUMO show clear differences between the chlorinated alkanes and alkenes (Figure 21.8; values for E1, BDE, and ELUMO are from Totten and Roberts [73]). The data show distinct differences between the linear free-energy relationships (LFERs) for chlorinated alkanes and alkenes. The correlation between k’obs and E1 is fairly robust for the chlorinated alkanes and consistent with their expected relationship; that is, there is a general trend of increasing reaction rate with increasing one-electron reduction potential. Similarly, the reaction rates for the chlorinated alkanes decrease as the energy for homolytic C-Cl bond dissociation increases. However, the best correlation is observed between k’obs and ELUMO for the chlorinated ethanes. Carbon tetrachloride and CF are outliers, an observation that might indicate subtle differences in the relationship between k’obs and ELUMO with regard to chlorinated methanes and ethanes. Though they are not proof of a particular reaction mechanism, the results of these LFERs are consistent with a single electron transfer as the rate-limiting step for the Ni-AHA-mediated reduction of chlorinated ethanes. The correlations between k’obs and E1, BDE, and ELUMO are far less robust for the chlorinated ethenes (Figure 21.8). Moreover, the correlations between k’obs and E1, BDE, and ELUMO are of substantially lower magnitude and are the inverse of those observed for the chlorinated ethanes; that is, reaction rates decrease with increasing E1 and increase with increasing BDE, ELUMO, and extent of chlorination. For example, the reduction of VC was faster than that of PCE. This observation is counter to the commonly observed trend of faster reduction of chlorinated hydrocarbons with increasing extent of chlorination [13,29,77, and many others]. However, the reverse trend (increasing reduction rate with decreasing chlorination) has been observed for the reduction of chlorinated ethenes by Fe0 [48] and FeS [78]. The reatively poor correlation between k’obs and E1, BDE, and ELUMO, and the decrease in reaction rate with increasing extent of chlorination suggest that the rate-limiting step for the Ni-AHA-mediated reduction of chlorinated ethenes involves factors other than single electron transfer. Additional research is required to fully elucidate the reaction mechanism(s) in this system. 21.4. CONCLUSIONS Nickel-AHA complexes are capable of acting as electron mediators in the reductive dechlorination of a wide range of C1 and C2 chlorinated hydrocarbons, nearly all of which are U.S. Environmental Protection Agency Priority Pollutants. Chlorinated solvents such as CT, CF, PCE, and TCE have been used in many industrial and commercial applications, and the improper use and disposal of these materials has resulted in widespread surface and subsurface contamination. The reduction of chlorinated hydrocarbons in situ (and in some engineered systems) is often incomplete and may result in the accumulation of less chlorinated daughter products that pose a greater environmental hazard than the parent compound. For example, TCE may be reduced to VC, a known carcinogen. The Ni-AHA-mediated reduction of most of the chlorinated hydrocarbons examined in this study was relatively rapid (with half-lives typically less than 1 h under our experimental conditions) and generally resulted in nonchlorinated products; thus, HS-transition metal complexes might have potential applications for the remediation of chlorinated hydrocarbons and other contaminants. Although TiIII citrate is not representative of the reductants typically found in natural suboxic and anoxic environments (for example, reduced iron and sulfur species, among others) and nickel generally is not present in most natural systems at the concentrations used in our experiments, the ability of Ni-AHA complexes to act as electron mediators for the reduction of chlorinated hydrocarbons by TiIII citrate suggests that HS-transition metal complexes are effective electron transfer agents in natural systems. Indeed, DOM-metal complexes have been suggested to contribute to the reduction of substituted nitrobenzenes by hydrogen sulfide under acidic conditions [10]. However, additional research is needed to evaluate the overall significance of the activity of HS-metal complexes as electron mediators for redox reactions in natural systems. ACKNOWLEDGEMENTS The authors thank Eila Burr and Marlene Cantrell of Applied Research Associates, Inc., for their assistance in the laboratory. We also thank Karlin Danielsen at the University of Michigan for her assistance with the CO analysis, Carrie Delcomyn of Applied Research Associates, Inc., for performing the GC-MS analysis, and Alexander Angerhofer at the University of Florida for conducting the EPR studies. Financial support for this project, provided in part by the Air Force Office of

204

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

Figure 21.8 Observed pseudo-first-order rate constants (k’obsd) for the reduction of chlorinated methanes, ethanes, and ethenes mediated by Ni-AHA complexes versus one-electron reduction potentials (E1), C-Cl bond dissociation energies (BDE), and the energy of the lowest unoccupied molecular orbitals (ELUMO). E1 values for the most thermodynamically favorable product were used. Values for E1, BDE, and ELUMOare from Totten and Roberts [73]. Dotted lines indicate 95% confidence intervals. Numbers in parentheses are±the standard error of the preceding term.

Scientific Research (AFOSR) and the Environmental Security Technology Certification Program (ESTCP) of the U.S. Department of Defense, is gratefully acknowledged. Funding for manuscript preparation was provided in part by the U.S. Department of Energy Office of Science, under contract W-31-109-Eng-38.

MATERIALS AND METHODS |

205

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

Cervantes FJ, Dijksma W, Duong-Dac T, Ivanova A, Lettinga G, Field JA. Anaerobic mineralization of toluene by enriched sediments with quinones and humus as terminal electron acceptors. Appl. Environ. Microbiol., 2001; 67:4471–4478. Coates JD, Ellis DJ, Blunt-Harris EL, Gaw CV, Roden EE, Lovley DR. Recovery of humic-reducing bacteria from a diversity of environments. Appl. Environ. Microbiol., 1998; 64:1504–1509. Lovley DR, Coates JD, Blunt-Harris EL, Phillips EJP, Woodward JC. Humic-substances as electron acceptors for microbial respiration. Nature, 1996; 382: 445–448. Lovley DR, Fraga JL, Coates JD, Blunt-Harris EL. Humics as an electron donor for anaerobic respiration. Environ. Microbiol., 1999; 1:89–98. Lovley DR, Kashefi K, Vargas M, Tor JM, Blunt-Harris EL. Reduction of humic substances and Fe(III) by hyperthermophylic microorganisms. Chem. Geol., 2000; 169:289–298. Collins R, Picardal F. Enhanced anaerobic transformations of carbon tetrachlo ride by soil organic matter. Environ. Toxicol. Chem., 1999; 18:2703–2710. Wittbrodt PR, Palmer CD. Reduction of Cr(VI) in the presence of excess soil fulvic acid. Environ. Sci. Technol., 1995; 29:255–263. Allard B, Arsenie I. Abiotic reduction of mercury by humic substances in aquatic system—an important process for the mercury cycle. Water, Air, Soil Pollut., 1991; 56:457–464. Curtis GP, Reinhard M. Reductive dehalogenation of hexachloroethane, carbon tetrachloride and bromoform by anthrahydroquinone disulfonate and humic acid. Environ. Sci. Technol., 1994; 28:2393–2401. Dunnivant FM, Schwarzenbach RP, Macalady DL. Reduction of substituted nitrobenzenes in aqueous solutions containing natural organic matter. Environ. Sci. Technol., 1992; 26:2133–2141. Fu QS, Barkovskii AL, Adriaens P. Reductive transformation of dioxins: An assessment of the contribution of dissolved organic matter to dechlorination reactions. Environ. Sci. Technol., 1999; 33:3837–3842. Burris DR, Delcolymn CA, Smith MH, Roberts AL. Reductive dechlorination of tetrachloroethylene and trichloroethylene catalyzed by vitamin B12 in homogeneous and heterogeneous systems. Environ. Sci. Technol., 1996; 30: 3047–3052. Burris DR, Delcomyn CA, Deng B, Buck LE, Hatfield K. Kinetics of tetrachloroethylene-reductive dechlorination catalyzed by vitamin B12. Environ. Toxicol. Chem., 1998; 17:1681–1688. Chiu P-C, Reinhard M. Metallocoenzyme-mediated reductive transformation of carbon tetrachloride in titanium(III) citrate aqueous solution. Environ. Sci. Technol., 1995; 29:595–603. Gantzer CJ, Wackett LP. Reductive dechlorination catalyzed by bacterial transition-metal coenzymes. Environ. Sci. Technol., 1991; 25:715–722. Glod G, Brodmann U, Angst W, Holliger C, Schwarzenbach R. Cobalamin-mediated reduction of cis- and trans-dichloroethene, 1,1dichloroethene, and vinyl chloride in homogeneous aqueous solution: Reaction kinetics and mechanistic considerations. Environ. Sci. Technol., 1997; 31:3154–3160. Krone UE, Laufer K, Thauer RK, Hogenkamp HPC. Coenzyme F430 as a possible catalyst for the reductive dehalogenation of chlorinated C1 hydrocarbons in methanogenic bacteria. Biochem., 1989; 28:10061–10065. Krone UE, Thauer RK, Hogenkamp HPC. Reductive dehalogenation of chlorinated C1-hydrocarbons mediated by corrinoids. Biochem., 1989; 28: 4908–4914. Perlinger JA, Buschmann J, Angst W, Schwarzenbach R. Iron porphyrin and mercaptojuglone mediated reduction of polyhalogenated methanes and ethanes in homogeneous aqueous solution. Environ. Sci. Technol., 1998; 32: 2431–2437. Schanke CA, Wackett LP. Environmental reductive elimination reactions of polychlorianted ethanes mimicked by transition metal coenzymes. Environ. Sci. Technol., 1992; 26:830–833. Perlinger JA, Angst W, Schwarzenbach RP. Kinetics of the reduction of hexa chloroethane by juglone in solutions containing hydrogen sulfide. Environ. Sci. Technol., 1996; 30:3408–3417. Schwarzenbach RP, Stierli R, Lanz K, Zeyer J. Quinone and iron porphyrin mediated reduction of nitroaromatic compounds in homogeneous aqueous solution. Environ. Sci. Technol., 1990; 24:1566–1574. Cervantes FJ, de Bok FAM, Duong-Dac T, Stams AJM, Lettinga G, Field JA. Reduction of humic substances by halorespiring, sulphate-reducing and methanogenic microorganisms. Environ. Microbiol., 2002; 4:51–57. Nurmi JT, Tratnyek PG. Electrochemical properties of natural organic matter (NOM), fractions of NOM, and model biogeochemical electron shuttles. Environ. Sci. Technol., 2002; 36:617–624. Scott DT, McKnight DM, Blunt-Harris EL, Kolesar SE, Lovley DR. Quinone moieties act as electron acceptors in the reduction of humic substances by humics-reducing microorganisms. Environ. Sci. Technol., 1998; 32:2984–2989. Ma H, O’Loughlin EJ, Burris DR. Factors affecting humic-nickel complex mediated reduction of trichloroethene in homogeneous aqueous solution. Environ. Sci. Technol., 2001; 35:717–724. O’Loughlin EJ, Burris DR. Reductive dehalogenation of trichloroethene mediated by wetland DOC-transition metal complexes. In: Means JL, Hinchee RE eds. Wetlands and remediation. Columbus,OH: Battelle Press, 2000:1–8. O’Loughlin EJ, Burris DR, Delcomyn CA. Reductive dechlorination of trichloroethene mediated by humic-metal complexes. Environ. Sci. Technol., 1999; 33:1145–1147. Arnold WA, Roberts AL. Pathways of chlorinated ethylene and chlorinated acetylene reaction with Zn(0). Environ. Sci. Technol, 1998, 32:3017–3025.

206

| CHAPTER 21: CATALYTIC EFFECTS OF NI-HUMIC COMPLEXES

30.

Schnitzer M. Organic matter characterization. In: Page AL, Miller RH, Keeney DR eds. Methods of soil analysis, Part 2. Chemical and microbiological properties. Madison: American Society of Agronomy and Soil Science Society of America, 1982:581–594. Smith MH, Woods SL. Comparison of reactors for oxygen-sensitive reactions: Reductive dechlorination of chlorophenols by vitamin B12. Appl. Environ. Microbiol., 1994; 60:4107–4110. Gossett JM. Measurement of Henry’s law constants for C1 and C2 chlorinated hydrocarbons. Environ. Sci. Technol, 1987; 21: 202–208. Tse G, Orbey H, Sandler SI. Infinite dilution activity coefficients and Henry’s law coefficients of some priority water pollutants determined by a relative gas chromatographic method. Environ. Sci. Technol., 1992; 26:2017–2022. Mackay D, Shiu WY. A critical review of Henry’s law constants for chemicals of environmental interest. J. Phys. Chem. Ref. Data, 1981; 10:1175–1199. Pankow JF, Isabelle LM, Asher WE. Trace organic compounds in rain. 1. Sampler design and analysis by adsorption/thermal desorption (ATD). Environ. Sci. Technol., 1984; 18:310–318. Semadeni M, Chiu P-C, Reinhard M. Reductive transformation of trichloroethene by cobalamin: Reactivities of the intermediates acetylene, chloroacetylene, and the DCE isomers. Environ. Sci. Technol., 1998; 32: 1207–1213. Andrieux CP, Gallardo I, Savéant J-M, Su K-B. Dissociative electron transfer. Homogeneous and heterogeneous reductive cleavage of the carbon-halogen bond in simple aliphatic halides. J. Amer. Chem. Soc., 1986; 108:638–647. Bakac A, Espenson JH. Kinetics and mechanism of the alkylnickel formation in one-electron reductions of alkyl halides and hydroperoxides by a macrocyclic nickel(I) complex. J. Amer. Chem. Soc., 1986; 108:713–719. Buschmann J, Angst W, Schwarzenbach RP. Iron porphyrin and cysteine mediated reduction of ten polyhalogenated methanes in homogeneous aqueous solution: Product analyses and mechanistic considerations. Environ. Sci. Technol., 1999; 33:1015–1020. Castro CE, Kray WC, Jr. Carbenoid intermediates from polyhalomethanes and chromium(II). The homogeneous reduction of geminal halides by chromous sulfate. J. Amer. Chem. Soc., 1966; 88:4447–4455. Glod G, Angst W, Holliger C, Schwarzenbach R. Corrinoid-mediated reduction of tetrachloroethene, trichloroethene, and trichlorofluoroethene in homogeneous aqueous solution: Reaction kinetics and mechanistic considerations. Environ. Sci. Technol., 1997; 31:253–260. Wade RS, Castro CE. Oxidation of iron(II) porphyrins by alkyl halides. J. Amer. Chem. Soc., 1973; 95:226–230. Assaf-Anid N, Hayes KF, Vogel TM. Reductive dechlorination of carbon tetrachloride by cobalamin(II) in the presence of dithiothreitol: Mechanistic study, effect of redox potential and pH. Environ. Sci. Technol., 1994; 28:246–252. Lewis TA, Paszczynski A, Gordon-Wylie SW, Jeedigunta S, Lee C-H, Crawford RL. Carbon tetrachloride dechlorination by the bacterial transition metal chelator pyridine-2,6-bis(thiocarboxylic acid). Environ. Sci. Technol., 2001; 35: 552–559. Poyer JL, Floyd RA, McCay PB, Janzen EG, Davis ER. Spin-trapping of the trichloromethyl radical produced during enzymatic NADPH oxidation in the presence of carbon tetrachloride or bromotrichloromethane. Biochim. Biophys. Acta, 1978; 539:402–409. Hine J. Carbon dichloride as an intermediate in the basic hydrolysis of chloroform. A mechanism for substitution reactions at a saturated carbon atom. J. Amer. Chem. Soc., 1950; 72:2438–2445. Robinson EA. The reaction of dichloromethylene with water and with phenoxide ions (Reimer-Tiemann reaction). J. Chem. Soc., 1961; 1663–1671. Arnold WA, Roberts AL. Pathways and kinetics of chlorinated ethylene and chlorinated acetylene reaction with Fe(0) particles. Environ. Sci. Technol., 2000; 34:1794–1805. Roberts AL, Totten LA, Arnold WA, Burris DR, Campbell TJ. Reductive elimination of chlorinated ethylenes, by zero-valent metals. Environ. Sci. Technol., 1996; 30:2654–2659. Fennelly JP, Roberts AL. Reaction of 1,1,1-trichloroethane with zero-valent metals and bimetallic reductants. Environ. Sci. Technol., 1998; 32:1980–1988. Kle′ ka GM, Gonsior SJ. Reductive dechlorination of chlorinated methanes and ethanes by reduced iron(II) porphyrins. Chemosphere, 1984; 13:391–402. Miller PL, Vasudevan D, Gschwend PM, Roberts AL. Transformation of hexachloroethane in a sulfidic natural water. Environ. Sci. Technol., 1998; 32: 1269–1275. Roberts LA, Gschwend PM. Mechanism of pentachloroethane dehydrochlorination to tetrachloroethylene. Environ. Sci. Technol., 1991; 25:76–86. Bakac A, Espenson JH. Mechanistic investigation of carbon-carbon bond formation in the reduction of alkyl halides by organonickel complexes in aqueous solution. J. Amer. Chem. Soc., 1986; 108:719–723. Campbell TJ, Burris DR, Roberts AL, Wells JR. Trichloroethylene and tetrachloroethylene in a metallic iron-water vapor batch system. Environ. Toxicol. Chem., 1997; 16:625–630. Castro CE, Kray WC, Jr. The cleavage of bonds by low valent transition metal ions. The homogeneous reduction of alkyl halides by chromous sulfate. J. Amer. Chem. Soc., 1963; 85:2768–2773. Helvenston MC, Castro CE. Nickel(I) octaethylisobacteriochlorin anion. An exceptional nucleophile. Reduction and coupling of alkyl halides by anionic and radical processes. A model for factor F-430. J. Amer. Chem. Soc., 1992; 114: 8490–8496. Deng B, Campbell TJ, Burris DR. Hydrocarbon formation in metallic iron/water systems. Environ. Sci. Technol., 1997; 31: 1185–1190. Hardy LI, Gillham RW. Formation of hydrocarbons from the reduction of aqueous CO2 by zero-valent iron. Environ. Sci. Technol., 1996; 30:57–65. Kochi JK, Rust FF. Metal ion-free radical reactions: Coupling of free radicals. J. Amer. Chem. Soc., 1961; 83:2017–2018.

31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41.

42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.

MATERIALS AND METHODS |

61. 62. 63. 64. 65. 66. 67. 68. 69. 70.

71. 72. 73. 74. 75. 76. 77. 78.

207

Harris JM, Wamser CC. Fundamentals of organic reaction mechanisms. New York: Wiley, 1976. March J. Advanced organic chemistry: Reactions, mechanisms, and structure. New York: Wiley, 1992. Baratta W, Del Zotto A, Rigo P. Half-sandwich ruthenium(II) complexes as catalysts for stereoselective carbene-carbene coupling reactions. Organometal., 1999; 18:5091–5096. Beckhaus R. C2 building blocks in the co-ordination sphere of electron-poor transition metals. Aspects of the chemistry of earlytransition-metal carbenoide complexes. J. Chem. Soc., Dalton Trans., 1997; 1991–2001. Borer M, Loosli T, Minger A, Neuenschwander M, Engel P. Die Reaktion von 1-Halogen-1-Lithiocyclopropanen mit CuCl2: Konkurrenz zwischen “Carben-Dimerisierung” und oxidativer Kupplung. Helv. Chim. Acta, 1995; 78: 1311–1324. Dötz KH, Siemoneit S, Hohmann F, Nieger M. Reactions of complex ligands 74. Alkyne carbene chromium chelates: Synthesis, structure and carbene dimerization. J. Organometall. Chem., 1997; 541:285–290. Cadiot P, Chodkiewicz W. Coupling of acetylenes. In: Viehe HG ed. Chemistry of acetylenes. New York: Dekker, 1969:597–647. Castro CE, Helvenston MC, Belser NO. Biodehalogenation, reductive dehalo genation by Methanobacterium thermoautrophicum. Comparison with nickel(I)octaethylisobacteriochlorin anion. An F-430 model. Environ. Toxicol. Chem., 1994; 13:429–433. Fathepure BZ, Boyd SA. Dependence of tetrachloroethylene dechlorination on methanogenic substrate consumption by Methanosarcina sp. strain DCM. Appl. Environ. Microbiol., 1988; 54:2976–2980. Hollinger C, Kengen SW, Schraa G, Stams AJM, Zehnder AJB. Methyl-coenzyme M reductase of Methanobacterium thermoautotrophicum ′ H catalyzes the reductive dechlorination of 1,2-dichloroethane to ethylene and chloroethane. J. Bacteriol., 1992; 174:4435–4443. Hollinger C, Schraa G, Stupperich E, Stams AJM, Zehnder AJB. Evidence for the involvement of corrinoids and factor F430 in the reductive dechlorination of 1,2-dichloroethane by Methanosarcina barkeri. J. Bacteriol., 1992; 174: 4427–4434. Eberson L. Electron transfer reactions in organic chemistry. Berlin: Springer-Verlag, 1987. Totten LA, Roberts AL. Calculated one- and two-electron reduction potentials and related molecular descriptors for reduction of alkyl and vinyl halides in water. Crit. Rev. Environ. Sci. Technol., 2001; 31:175–221. Butler EC, Hayes KF. Kinetics of the transformation of halogenated aliphatic compounds by iron sulfide. Environ. Sci. Technol., 2000; 34:422–429. Liu Z, Betterton EA, Arnold RG. Electrolytic reduction of low molecular weight chlorinated aliphatic compounds: Structural and thermodynamic effects on process kinetics. Environ. Sci. Technol., 2000; 34:804–811. Perlinger JA, Venkatapathy R, Harrison JF. Linear free energy relationships for polyhalogenated alkane transformation by electrontransfer mediators in model aqueous systems. J. Phys. Chem. A, 2000; 104:2752–2763. Scherer MM, Balko BA, Gallagher DA, Tratnyek PG. Correlation analysis of rate constants for dechlorination by zero-valent iron. Environ. Sci. Technol., 1998; 32:3026–3033. Butler EC, Hayes KF. Kinetics of the transformation of trichloroethylene and tetrachloroethylene by iron sulfide. Environ. Sci. Technol., 1999; 33:2021–2027

Part 4 BIOGEOCHEMICAL EFFECTS: THE GOOD, THE BAD AND THE UGLY

Chapter 22 HUMIC SUBSTANCES AND THEIR DIRECT EFFECTS ON THE PHYSIOLOGY OF AQUATIC PLANTS Stephan Pflugmacher, Constanze Pietsch, Wiete Rieger, Andrea Paul, Torsten Preuer, Elke Zwirnmann and Christian E.W.Steinberg Leibniz Institute of Freshwater Ecology and Inland Fisheries, AG Detoxication/ Metabolism, Müggelseedamm 301, 12561 Berlin, Germany 22.1. INTRODUCTION The aquatic environment contains a wide range of dissolved or colloidal organic matter (DOM), which consists mainly of the products of decomposition of plant organic material. A major part of DOM is represented by humic substances (50–75% of total dissolved organic carbon (DOC)), of which 90% usually are fulvic acids (FAs) and 10% are humic acids [1]. DOM is a mixture of biogenic, mostly heterogeneous and refractory organic compounds. DOM interacts with many xenobiotics like PAH e.g., anthracene, pyrene, benzo(a)pyrene [2–4], PCB or DDT [5,6] by various modes of binding and adsorption such as ion exchange, hydrogen bonding, charge transfer, covalent binding or hydrophobic adsorption and partitioning [7,8]. Humic substances are not well defined because of their complex structure, but progress is being made in characterizing these substances according to their molar mass [9]. Little information is available concerning the effects of DOM on exposed aquatic organisms, and most of it is disregarded in studies of the influence of DOM on the bioconcentration of inorganic or organic chemicals [10–12]. From a chemical point of view, HSs are by no means dead in the sense of being unreactive. Indeed, they are natural chemicals with a variety of functional groups such as alcoholic, phenolic, methoxy, quinoid, keto, aldehyde and carboxylic groups, and a variety of molar masses ranging from 100 to several thousand Daltons [13]. For instance, on the average carboxyl groups comprise some 13% of the functional groups [14]. As chemical compounds, HSs may have an impact on exposed organisms. However, our knowledge of how HSs affect organisms is incomplete [15]. In many instances, only modulations of nutrient bioavailability, or the suppression of pathogens have been the focus of studies [16–18]. Thus, the effects of HSs on organisms are thought to be more or less indirect. One reason for this paradigm may be that HSs, due to their presumed high molar mass, are too large to penetrate biomembranes. This study focused on direct effects of HSs on different physiological parameters like photosynthetic oxygen production and the activity of detoxication enzymes like glutathione S-transferase as well as several ROS-enzymes in the aquatic plants Ceratophyllum demersum and Vesicularia dubyana. In previous papers we reported inhibitory effects on photosynthetic oxygen production in the aquatic macrophyte Ceratophyllum demersum [19], and on selected detoxication enzymes in the same plant [20]. Following these papers we suggested the following hypothesis about plant-DOM interactions (Figure 22.1): 1) DOM or fragments of DOM are taken up by plant cells; 2) Reactive groups in DOM are recognized by cell internal defence systems, which are activated detoxication enzymes; 3) Beginning detoxication processes create reactive oxygen species (ROS), which themselves activate ROS-en zymes for detoxication; and 4) DOM or fragments of DOM are able to cross membranes and reactive groups in DOM like quinoid structures can therefore act as inhibitors of photosynthesis. 22.2. MATERIAL AND METHODS 22.2.1. Materials Plant Material. Ceratophyllum demersum was collected during the summer from different sites around Berlin. Identification of the plant species was according to [21]. Vesicularia dubyana was a gift from Dr. T.Meinelt (IGB Berlin). Both aquatic plants were cultivated non-axenically prior to the experiment for 6 months in Provasoli’s medium (ESISI, 15ml/L) in 100 L tanks. Supplementary light was provided by daylight lamps with an irradiance of 12 µE/m2 s at a light/dark cycle of 14:10 h. Temperature was maintained at 22–24°C.

210

| CHAPTER 22: HUMIC SUBSTANCES AND THEIR DIRECT EFFECT ON AQUATIC PLANTS

Figure 22.1 Proposed reaction scheme of the interaction of DOM with aquatic plants, showing the uptake of DOM in plant cells, primary and secondary reaction in the plant cell including enzyme activation and effects on plant photosynthesis.

DOM Isolation. DOM samples used for this study had been isolated from a range of different sources in Europe (Norway) and USA (International Humic Substances Society). Concentrations of DOM (mg/L DOC) in the stock solutions were analysed as carbon dioxide (IR-detection) after high temperature combustion (Shimadzu TC 5000) [22]. Plant Exposure to DOM and 1-Aminoanthraquinone. Five gram fresh weight (FW) of C. demersum and V. dubyana was exposed to DOM from different sources in a volume of 500 mL each for 24 h under constant conditions of light and temperature. The pH was maintained at 7.5–8.0. All exposures were done in five replicates. The concentrations of DOM used were mostly in environmentally relevant loads from 0.5 µg L−1 to 50 mg L−1 DOC and various concentrations of 1aminoanthraquinone. DOM and the quinone were left out of the medium in control experiments. Measurement of Photosynthetic Oxygen Production. Photosynthetic oxygen production of the plant was measured with a Phosy-Mess 4000 (InnoConcept, Straussberg), 100% light intensity (approximately 2000 lux) and a dark/light/dark cycle of 10/12/10 min at a constant temperature of 20°C. Measurements were taken with a Clark electrode (WTW EO 196–1,5). The rates were calculated in µmoles O2h−1g FW−1. Measurement of photosynthetic oxygen production was performed with 12 replicates of 0.5 g FW of each plant. Enzyme Extraction and Measurement. Enzyme extraction from the plant tissue was as described in [23]. Measurement of soluble glutathione S-transferase (sGST) activity with the substrates 1-chloro-2,4-dinitrobenzene (CDNB) was according to [24]. Measurement of the peroxidase in the soluble protein fraction using guajacol as substrate was performed as described by [25]. All enzyme measurements were made in five-fold repeats of three independent samples. Protein determination in the microsomal and soluble fractions was done according to [26] using bovine serum albumin as standard. ESR Spectroscopy. Cw X-band ESR spectra of freeze-dried samples were recorded at room temperature with a ESR300 spectrometer (Center of Construction of Scientific Devices, Berlin-Adlershof, Germany) at a microwave power PMW=2 mW. The samples were placed into commercial quartz glass sample tubes (id 3 mm). For the determination of g-values, MgO/Cr3+ (g=1.9796) was used as reference. Spin contents of the samples were determined by the parallel recording of sample and reference, and calculated from the integrated areas normalized to the reference. To minimize errors, all quartz glass tubes

MATERIALS AND METHODS |

211

were adjusted in a resonator outside of the cavity before inserting into the equipment. Absolute spin densities were corrected to a value of 0.65×1017 spins g−1 for the Suwannee River humic acid according to the IHSS data. Chemical Analysis (LC-OCD). Size-exclusion chromatography with UV and organic carbon detection was used for DOC analysis of the humic sample (Svartberget NOM, HS 1500), 1-aminoanthraquinone and a mixture of both according to the method described [27,28]. Different DOC compounds were fractionated by size-exclusion chromatography (SEC) on a Toyopearl HW-50S resin column (250×20 mm) eluted with 29 mM, pH 6.5 phosphate-buffer at a flow rate of 1 mL min−1. Characterization and quantification of three different groups of DOC (polysaccharides, humic substances and low molar mass acids) were done by UV-detection at 254 nm. Quantification was performed by IR-detection after UV oxidation in a cylindrical thin-film UV reactor at 185 nm as described by [29]. Uptake Studies with the 14C-labeled Oxidation Product of Caffeic Acid. The caffeic acid oxidation product was dissolved in artificial tank water to a concentration of 44.000 dpm in 100 mL. Three parallels of Ceratophyllum demersum plant pieces (FW range: 0.318–0.351 g) were exposed for 24 h and rinsed 3 times with 5 mL water to remove adherent caffeic acid oxidation product after the exposure time. The treated samples were homogenized in liquid nitrogen to a fine powder, which was extracted with 100% methanol (2mL) for 24 h. Three subsamples of 10 µL were mixed with 4 mL scintillation cocktail (Ultima-gold-XR, Packard Instruments) and radioactivity in the samples was determined by liquid scintillation counting (LSC) on a TRI-Carb 1900 TR (Canberra-Packard, Germany). Statistics. Significance testing was done with one-way analysis of variances (ANOVA) and the Students T-test and Newman Keuls test both at ′ 0.05.

212

| CHAPTER 22: HUMIC SUBSTANCES AND THEIR DIRECT EFFECT ON AQUATIC PLANTS

Figure 22.2 Activity of soluble glutathione S-transferase in Ceratophyllum demersum after 24 h exposure to DOM from different sources at a concentration of 0.5 mg L−1 C and with the synthetic humic substance HS 1500 and several pure quinones. Solid column is control.

Figure 22.3 Activity of soluble glutathione S-transferase in Vesicularia dubyana after 24 h exposure to DOM from different sources at a concentration of 0.5 mg L−1 C and with the synthetic humic substance HS 1500 and several pure quinones. Solid column is control.

22.3. RESULTS 22.3.1. Uptake Studies Uptake of DOM was simulated by an oxidation product of 14C-labelled caffeic acid. After exposure of C. demersum for 24 h to this oxidation product, radioactivity was monitored by LSC and exhibited an uptake of 7.3±1.4 % in this plant species. 22.3.2. Activation of Detoxication Enzyme Activity Exposure of Ceratophyllum demersum to 0.5 mg/L DOC from different sources resulted in an activation of soluble glutathione S-transferase (sGST) using the substrate CDNB (Figure 22.2). This activation was significant as proven by the Newman-Keuls test (p=0.05) for all tested humic substances and for the synthetic HS 1500 as well as for all quinones used. In most cases the NOM-type humic substances elevated this enzyme system more strongly than the fractionated humic substances (humic acids or fulvic acids). Elevation of sGST activity by the different quinones was in most cases a factor 2 higher than with the natural humic substances. Exposure of the water moss Vesicularia dubyana exhibited a very similar elevation pattern using the humic substances from different sources at a concentration of 0.5 mg L−1 (Figure 22.3). All tested DOM showed significant elevation of sGST in V. dubyana and, as seen in C. demersum, the fractionated DOM showed lower elevation potential. Exposure to the synthetic humic substance 1500 and to the different quinones led to a strong elevation of sGST activity also by factor 2 compared to the natural HSs.

MATERIALS AND METHODS |

213

Figure 22.4 Activity of guajacol peroxidase in Ceratophyllum demersum after 24 h exposure to DOM from different sources at a concentration of 0.5 mg L−1 C and with the synthetic humic substance HS 1500 and several pure quinones

Figure 22.5 Kinetics of the elevation and inhibition of POD activity in C. demersum during 48 h of exposure compared with untreated controls

22.3.3. Activity of Guajacol Peroxidase (POD) Activity of guajacol peroxidase (POD) in C. demersum after exposure to the different DOM samples and quinones showed a strong and significant elevation of enzyme activity after 24 h of exposure (Figure 22.4). Enzyme activation elevation factors were between 3 and 15 compared to an untreated control. The highest elevation of this enzyme was detected with BS1 FA and Fuhrberg FA. For this enzyme it seems that the fractionated DOM, especially the fulvic acid samples, are the strongest activators. Exposure to the different quinones for 24 h showed in the case if 1-aminoanthraquinone a complete loss of enzyme activity and for the other two quinones (p-benzohydroquinone and hydroquinone) very low enzyme activity below the control level. A time-dependence (Figure 22.5) with one depicted DOM sample (Svartberget NOM) and 1-aminoanthraquinone showed very high activation of POD in the first 4 h and then a constant decrease during 48 h of exposure time. POD activity in V. dubyana after exposure to the different DOM samples also showed very high elevation levels by factors of 3–9 (Figure 22.6). In this case no clear preference of elevation potential by fractionated DOM samples could be seen. From the data for quinone used with the water moss, the activity of the POD was significantly decreased. A time-dependence (Figure 22.7) similar to those found with C. demersum showed a clear and high elevation of POD activity during the first 4 h of exposure time with Svartberget NOM as well as with 1-aminoanthraquinone, and then a sharp decrease of enzyme activity till the end of exposure after 48 h.

214

| CHAPTER 22: HUMIC SUBSTANCES AND THEIR DIRECT EFFECT ON AQUATIC PLANTS

Figure 22.6 Activity of guajacol peroxidase in Vesicularia dubyana after 24 h exposure to DOM from different source at a concentration of 0.5 mg L−1 C and with the synthetic humic substance HS 1500 and several pure quinones

Figure 22.7 Kinetics of the elevation and inhibition of POD activity in V. dubyana during 48 h of exposure compared with untreated controls

22.3.4. Photosynthetic Oxygen Production Direct effects of humic substances from different sources at a concentration of 0.5 mg/L DOC on the photosynthetic oxygen production of the macrophyte C. demersum showed in most cases a significant (p=0.05, Newman-Keuls test) reduction (Figure 22.8). The highest reductions were detected with fractionated humic substances like BS1 FA and Fuhrberg FA as well as with Laurentian FA. Inhibition factors for these fulvic acids were 2–7. The same inhibition was detected using synthetic HS 1500 and with the pure quinones. Photosynthetic oxygen production was nearly zero (with 1-amino-anthraquinone it was zero). Only the humic substance isolated at Hellerudmyra exhibited a significant elevation of photosynthetic oxygen production. This inhibitory effect after exposure to different DOM samples in V. dubyana was only observed for a few tested samples from Suwannee River HA, Hellerudmyra NOM, BS1 FA, Fuhrberg FA and Laurentian FA by factors of 2–3 (Figure 22.9). The DOM sample from Hellerudmyra exhibited an elevation of the photosynthetic oxygen production by factor 2. All other samples were not significantly differ ent from the untreated control. However, with the synthetic HS 1500 as well as the different quinones, the inhibition of the photosynthetic oxygen production was very sharp and significant. The inhibition pattern was similar to those for C. demersum. A dose-dependent exposure in the range of 0.5–50 mg L−1 with the humic sample from Svartberget NOM and 1-aminoanthraquinone showed a decrease in photosynthetic oxygen production with C. demersum (Figure 22.10).

MATERIALS AND METHODS |

215

Figure 22.8 Photosynthetic oxygen production in Ceratophyllum demersum after 24 h exposure to DOM from different sources at a concentration of 0.5 mg L−1 C and with the synthetic humic substance HS 1500 and several pure quinones

Figure 22.9 Photosynthetic oxygen production in Vesicularia dubyana after 24 h exposure to DOM from different sources at a concentration of 0.5 mg L−1 C and with the synthetic humic substance HS 1500 and several pure quinones

Figure 22.10 Dose-dependent inhibition of photosynthetic oxygen production in Ceratophyllum demersum after 24 h exposure to Svartberget NOM and 1-aminoanthraquinone

Above a concentration of 50 µg L−1 1-aminoanthraquinone, no oxygen production was observed. A comparable decrease using natural DOM showed that with 50 mg L−1 or greater a very low amount of oxygen was still produced by the plant. But this amount was a factor 12 lower than in the untreated control. After exposure to 1-aminoanthraquinone, inhibition of photosynthetic oxygen production in the water moss V. dubyana was not as sharp and only above a concentration of 50 mg L−1 was oxygen production completely inhibited (Figure 22.11). Also, a significant inhibition of oxygen production was only achieved with 50 mg L−1 Svartberget NOM. Size-exclusion-chromatography was used to show parallels and/or differences between different DOM samples, and may give hints on structural parallelism of the substances used. To find out whether quinoid sites are the main reason for the observed inhibition of photosynthetic oxygen production, Svartberget NOM, 1-aminoanthraquinone, a mixture of both and for comparison the synthetic HS 1500 showed typical fingerprint patterns (Figure 22.12), with at least two distinct peaks. Peak 1 is defined as HSs and peak 2 are low molar mass substances in the samples. The last peak probably is mono- and disaccharides or amino acids. Two distinct peaks were visible for 1-aminoanthraquinone under the conditions applied. Peak 1 corresponds to LMMA from humic substances and might be due to some aggregated 1-aminoanthraquinones, whereas peak 2 resembles

216

| CHAPTER 22: HUMIC SUBSTANCES AND THEIR DIRECT EFFECT ON AQUATIC PLANTS

Figure 22.11 Dose-dependent inhibition of photosynthetic oxygen production in Vesicularia dubyana after 24 h exposure to Svartberget NOM and 1-aminoanthraquinone

Figure 22.12 A–D. LC-OCD fingerprint pattern of A) Svartberget NOM, B) 1-aminoanthraquinone, C) a mixture of both and, for comparison, D) the synthetic humic substance HS 1500

the parent compound 1-aminoanthraquinone. Anthraquinones are organic dyes that are well known to form dimers and higher oligomers in solution [30,31]. We suppose that aggregates were formed as indicated by peak 1 in the chromatogram. A mixture of Svartberget NOM and 1-aminoanthraquinone showed an additive chromatogram with no change in the humic substance peak. The 1-aminoanthraquinone peak was still well separated. Under our experimental conditions the intensity of ESR absorption is proportional to the number of unpaired spins in the sample. Larger amounts of semiquinone free radicals correlate with a higher percentage of aromatic structures in humic substances. The plot of relative photosynthetic oxygen production versus relative spin content of the tested humic samples gave positive correlations for both plants. Correlation coefficients of r2=0.91 for C. demersum (Figure 22.13) and r2=0.81 for V. dubyana (Figure 22.14) were obtained. The highest spin content was measured was for synthetic HS 1500. 22.4. DISCUSSION Recent work concerning humic substances has dealt with the ability of DOM to reduce the bioconcentration factor (BCF) and bioavailability of xenobiotics in the aquatic ecosystem [32]. Many studies have addressed the relationship between DOM concentration and BCF of xenobiotics [5,33], which in organisms seems to be related to freely dissolved xenobiotics [34–35]. Thus, the hypothesis was developed that only the free dissolved fraction of xenobiotics is bioavailable. Studies of Caenorhabditis elegans (nematode) with 3H-benzo(a)pyrene, pyrene and samples from different DOM sources led to the assumption that DOM-bound xenobiotics are not taken up by the organisms [32]. To complement these studies, our study was set up to fill a gap, namely the direct effects of DOM on organisms starting with aquatic plants like Ceratophyllum demersum and Vesicularia dubyana. Other plant studies have shown that there are

MATERIALS AND METHODS |

217

Figure 22.13 Correlation of spin density and photosynthetic oxygen production in Ceratophyllum demersum (r2 =0.91). This shows the correlation of quinoid moieties in humic substances to the inhibition of photosynthetic oxygen production

Figure 22.14 Correlation of spin density and photosynthetic oxygen production in Vesicularia dubyana (r2 =0.81). This shows the correlation of quinoide moieties in humic substances to the inhibition of photosynthetic oxygen production

direct effects of humic substances obtained from decomposed leaves of the common cattail (significantly decreased chlorophyll a and b content in Salvinia minima [1]). This decrease might be due to increased chlorophyllase activity [36]. The concentrations of the different DOM samples used here were environmentally relevant, as DOM concentrations generally range from 0.5 to 4 mg L−1 DOC in aquatic ecosystems [37]. Furthermore, it is known that leachate from leaf litter and straw can inhibit algal development in circum-neutral water. For example, barley straw has been tested as a treatment for undesirable algal growth [38]. A humic acid-like isolate of a sewage-sludge-amended soil caused extensive alterations of tomato morphology and a significant reduction of the length and dry weight of both roots and shoots [39]. 22.4.1. Uptake Studies The uptake of HSs or fragments of HSs by organisms has been considered for many years. Looking at balneotherapeutic treatments there are hints that at least fragments of HSs can penetrate human skin. In the skin, the changes in enzymes like oxidoreductases, aminotransferases and phosphomonoesterhydrolases support these findings [40]. Recent studies in rice cell cultures showed that HSs or at least parts thereof can be taken up [41]. In our study we used synthetic HS, which is an oxidation product of caffeic acid with sodium metaperiodate [42]. This oxidation product had a molar mass of 6 kDa. The results showed that the coontail Ceratophyllum demersum could bioconcentrate the 14C-labelled substance in cells. The percentage of uptake was 7.3±1.4% during 24 h of exposure. Studies with human skin showed a penetration of this caffeic acid oxidation product within 30 min in the epidermis and dermis at a concentration of 1–3% [43]. Uptake studies using gammarids (Gammarus pulex) and the same concentration of the oxidation product exhibited similar uptake in this organism (Wiegand et al., this volume). It may still be argued that the whole intact oxidation product of the caffeic acid was not taken up and that smaller fragments are bioconcentrated. Nevertheless, considering the possible photodegradation of humic substances in the environment, it is obvious that at least low molar mass photodegradation products could be taken up by aquatic organisms and might be responsible for the several effects addressed below, which are plausible only if HSs are taken up by the organisms. Also, some HSs precursors might be

218

| CHAPTER 22: HUMIC SUBSTANCES AND THEIR DIRECT EFFECT ON AQUATIC PLANTS

able to pass membranes more easily than the whole HS [44]. Taking the work of [45] into account, some HSs regulate membrane permeability in microorganisms, so even the penetration of larger HSs structures seems likely. Another possibility takes the models of [40] into account: humic precursors might establish equilibrium with HSs in the cell after the smaller fragments have entered. Other studies [46] have described direct stimulatory effects on the green alga Ankistrodesmus bibraianus in a medium with an NOM isolate from Lake Skjervatjern. These results are indirect evidence for the uptake of humic substances by organisms. 22.4.2. Activation of Detoxication Enzymes Uptake of substances by an organism could activate the cell internal detoxication system to prevent the organism from further damage. An enzyme of the immune system of cells is the glutathione S-transferase system (GST). GSTs are ubiquitous multifunctional enzymes with key roles in cellular detoxication. The enzymes protect the cells against toxicants by conjugating them to glutathione. Thus, modulation of GST activity upon exposure may be considered as a signal of chemical stress. As glutathione is conjugated to electrophilic compounds, it is likely that it is also combined with humic material due to the relatively high concentration of functional groups. Elevation of the soluble GST measured with CDNB in both plant species indicates that HSs are recognized as foreign substances by GST itself or by specific Ah-receptors. It is well known that binding to specific Ah-receptors induces GST activity via the dioxin responsible element (DRE) and the antioxidant responsible element (ARE) in the cells, leading to increased enzyme activity. This goes hand in hand with beginning detoxication processes in the cell. From the results obtained, the NOM samples show a stronger effect on the GST than the fractionated samples. Thus, during preparation and fractionation of the humic samples from lake water it seems that some activators are lost. Pure quinones have a much stronger effect on the GST activity of both plants. Comparing both plants, C. demersum (a higher macrophyte) and V. dubyana (a water moss), it can be seen that the more primitive moss reacts less strongly to exposure to HSs than the higher macrophyte. Activation of enzyme activity as results of exposure to HSs was also found for phosphatase activity in roots of Lepidium sativum. This enzyme is essential for the synthesis of cell substance. It is assumed that HSs penetrate the plant cells and are reactive in various internal cell compartments. These reactions damage the protoplast and induce a defence mechanism in which the plant tries to elevate the synthetic processes in its cells [41]. Activation of GST was also shown previously [19,20] in C.demersum after exposure to HSs from Lake Hohlosee (Germany) and also by Wiegand et al. (this volume) in Daphnids. 22.4.3. Activity of Guajacol Peroxidase (POD) Peroxidases of plant origin normally reduce H2O2 to water at the expense of aromatic reductants, for which they are relatively non-specific. Located both membrane-bound and in the soluble fraction, there are several peroxidase isoforms and both constitutive as well as inducible forms are known. Elevation of POD under the influence of xenobiotic stress was observed in organisms in relation to water pollution [47]. In the aquatic macrophyte Hydrilla verticillata, POD activity was significantly increased after treatment with anthracene in concentrations higher than 0.01 mg L−1 [48]. POD activity also is elevated by plant growth regulators like gibberellic acid [49], stress-related plant hormones [50], and by aromatic ring structures [51,52]. Because of the elevation of reactive-oxygen (RO) enzyme, it can be presumed that the tested DOM samples contain structures similar to the known structures that elevate POD activity. Furthermore, it has been demonstrated that HSs contain stable semiquinone free radicals [53], which may also lead to activation of RO-enzymes. In addition, DOM itself can undergo photodegradation processes and release a variety of more or less stable forms of organic C and N by-products such as low molar mass organic acids and formaldehydes [52,54]. On the other hand, DOM can act as a photostabilizer, producing a number of reactive molecular species like superoxide anion, hydrogen peroxide or hydroxyl radicals. These reactive species are well known to elevate peroxidase activity in plants. The elevation found in our in vivo experiments contrasts with in vitro experiments with POD from Triticum aestivum roots or horseradish, where an inhibition of enzyme activity was detected in the presence of HSs. The decrease of POD activity was dose-dependent, but was interpreted as reactions with guajacol and HSs [55,56]. As shown for the aquatic macrophyte C. demersum [19] and also in this study with HSs from different sources, a high elevation of POD activity was detected after 24 h exposure to humic substances. For this enzyme system the fractionated samples seem to have more elevation potential than the NOM samples themselves. Using the pure quinones for exposure, a complete loss of enzyme activity was detected that could have two possible explanations: 1) the enzyme system does not react with these substances; or 2) the exposure time was too long and the enzyme system was damaged.

MATERIALS AND METHODS |

219

Figure 22.15 Different interactions of humic substances with enzymes and enzymatic reactions extended and modified according to [44]. There are several ways HSs could interact directly via modification of the active site of an enzyme or by formation of an enzyme-humic substance complex. Indirect ways might include the reduction of essential cofactors, interactions with the substrate itself or interaction with intracellular enzyme inhibitors

The first point seems unlikely because quinones are known to be metabolically activated, leading in the formation of highly reactive free radicals (superoxide, H2O2) [57] that activate RO enzymes like the peroxidases or superoxide dismutase to prevent further cell damage. The second point tested by a kinetic study with Svartberget NOM and 1-aminoanthraquinone showed that there is a clear time-dependence using the pure quinone: POD elevation during the first 4 h with a maximum elevation factor of 8 and then a sharp decrease in enzyme activity that indicates possible damage to this enzyme system. This behaviour was observed in both plants. As shown by [58–60], HSs may interact directly with isolated enzymes like carboxypeptidase A or pronase B by inhibiting their enzyme activity. Possible interactions of HSs within the cell are summarized in Figure 22.15. The direct effects of humic substances on enzymes were first summarized by [44] and are supported by the results of this study. Two main effects that can be seen in the interactions of enzymes and HSs. Firstly, there is an inhibition of enzyme activity. This can be due to binding of essential cofactors to HSs so that they are not available for the enzymatic reaction, modification of the enzyme structure by HSs acting on the active site of an enzyme or simply by binding of the substrate of the enzymatic reaction. Secondly, there could be an elevation of enzyme activity by, for example, binding of interacellular enzyme inhibitors or by HSs binding to special receptors as mentioned above. Inhibition and/or activation of enzymes ultimately will affect the metabolism of plants. 22.4.4. Photosynthetic Oxygen Production Another sensible biomarker for the xenobiotic influence of substances on plants is photosynthetic oxygen production. Previous studies on this topic [19,29] showed that after exposure of the aquatic plant C. demersum to HSs, significant inhibition of photosynthetic oxygen production occurred with HSs isolates from the German DFG priority ROSE project. This study is supported now with different humic samples from Norway and with IHSS standards. For most of our humic samples we found an inhibition of photosynthesis in both plants. At the moment there are two hypotheses concerning how photosynthesis inhibition could occur. They are mostly based on the quinoid structures in HSs (Figure 22.1). Quinoid structures can rapidly form radicals, which are electron acceptors and can therefore interfere in the electron transport chain between photosystem II (PS II) and photosystem I (PS I) in the chloroplasts of plants. The first hypothesis is that quinoid moieties act as electron acceptors that inhibit the photosynthesis of plants. The second hypothesis is that quinoid fragments of HSs bind to or substitute the primary (QA) or secondary (QB) electron acceptors in PS II electron transport, which are plastoquinone molecules. As known for several herbicides acting on the photosynthesis

220

| CHAPTER 22: HUMIC SUBSTANCES AND THEIR DIRECT EFFECT ON AQUATIC PLANTS

of plants, the mode of action is competition with plastoquinones for binding at the secondary quinone acceptor site (QB site) to the D1-protein. For quinones, especially 1-aminoanthraquinone, it is well known that they can have inhibitory effects on the electron transfer chain in photosystem II [61]. Using 1-aminoanthraquinone and comparing the effects of this quinone with those of Svartberget NOM, the inhibition of photosynthetic oxygen production seems to take place at PS II, which is downstream of the photosynthetic process. Thus, the hypothesis that quinoid structures of humic substances can inhibit the photosynthesis of plants by interruption of electron transport between photosystem II and photosystem I is supported. The results presented here strengthen the hypothesis that quinoid moieties in the structures of HSs are electron scavengers and thereby inhibit photosynthetic oxygen production in aquatic plants. Using size-exclusion chromatography, three fractions of DOC (polysaccharides (PS), humic substances (HSs) and low molar mass acids (LMMA)), can be separated, quantified and characterized by organic-carbon and UV-detection in samples Svartberget NOM, 1-aminoanthraquinone, a mixture of both and the synthetic humic substance HS 1500. The structural parallelelism of these samples and the mixture of Svartberget NOM with 1-aminoanthraquinone was shown by at least one peak (P1) that is prominent in every sample. Pure 1-aminoanthraquinone gave a clear signal in the low molar mass range. Taking electron spin resonance as an indirect but significant measure of quinoid structures [62] it is evident that the reduction of photosynthetic oxygen production is related to the quinoid structural units of HSs. The concentration of paramagnetic centres (free radicals) is correlated with the occurrence of electron donor-acceptor processes and the formation of charge-transfer complexes, and is mainly related to semiquinone-like C-radicals in aromatic units of HSs [62]. These semiquinone-type carriers are generated whenever an electron donor interacts with an electron acceptor and are stabilized for a long time in the resulting charge-transfer system. Based on the assumption that the number of semiquinone radicals is proportional to the number of quinoid moieties of HSs, ESR spectroscopy is a useful tool to quantify the amount of these structural components by investigation of spin densities [63]. Furthermore, some ESR parameters of HSs or fulvic acids are very close to those for hydroxyanthraquinones [64]. Also, catechol moieties in HSs were detected and considered to be responsible for HS free radical properties [62]. From the correlation of photosynthesis and free radical concentrations, this first approach towards a Quantitative-structure-relationship (QSAR) between the inhibition of photosynthetic oxygen production in aquatic plants and the quinoid moieties in HSs acting as electron acceptors seems to be correct. The potential of QSAR studies on effects of HSs was shown previously [65]. The presence of Fe(III), Cu(II) and Mn(II) in humic substances was observed by several authors [66–68]. Because Fe(III) is well known to participate in various redox reactions, Fe(III) could also account for the observed inhibition of photosynthesis. This explanation is not supported by our work because the synthetic HS1500, which only consists of aromatic moieties linked by alkyl spacers [69], does not contain Fe(III) but showed the highest inhibition of photosynthesis. Interference of humic substances in photosynthesis in aquatic plants appears to be an intrinsic property of these substances. However, there must be an additional mechanism to explain how HSs are involved directly or indirectly with this main function of plants, as the example of Hellerudmyra (Norway) NOM elevating photosynthetic oxygen production showed. The exposure of V. dubyana to this NOM led to at least a doubling of oxygen production. The mechanism of this is still unclear. Comparing the data for the two aquatic plants investigated, it is evident that the water moss is less sensitive to HSs than the higher macrophyte. Taking this as a general difference between mosses and angiosperms, a frequent field observation might be explained. Mosses commonly are the dominant species in humic waters and especially in running waters [70]. From our results it may be that mosses are always less sensitive to HSs because enzyme elevation of GST and POD showed opposite effects. But if the coupled land/water system develops simultaneously, organisms in the water will be less sensitive HSs from additional input of allochthonous material. 22.5. CONCLUSIONS The results presented reveal a new feature of humic substances. It is their possible xenobiotic mode of action on aquatic organisms, especially on aquatic macrophytes. Various metabolic pathways seem to be involved and affected, and it appears that plants recognize HSs as xenobiotics. The ecological significance of our findings is the different susceptibilities of aquatic plants or organisms towards humic substances. Acclimatization of particular plants to the “home” humic substance might occur. Therefore, humic substances seem to have the potential to structure and regulate whole ecosystems. We can imagine that HSs entering an ecosystem from natural or human influences like major flooding can have severe effects on the whole ecosystem community. Only those organisms that are able to deal with the new and for them “xenobiotic” HSs can survive. Humic substances with different qualities may be very important for niche-determination in an aquatic ecosystem.

MATERIALS AND METHODS |

221

ACKNOWLEDGEMENTS The authors thank R.Klöcking (Institut für Medizinische Mikrobiologie der Medizinischen Akademie, Erfurt, Germany) for the generous gift of 14C-labelled caffeic acid. R.Vogt (University of Oslo, Norway) and E.Gjessing (Agder College, Norway) provided humic substances for the experiments. Furthermore, E.Klose and G.Specht (InnoConcept, Straussberg, Germany) helped with the PhosyMess 4000 measurements and E.Giubileo (IGB Berlin) contributed excellent technical assistance in the laboratory. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

17. 18.

19.

20. 21. 22. 23. 24. 25.

Gardner JL, Al-Hamdani SH. Interactive effects of aluminium and humic substances on Salvinia. J. Aquat. Plant Managemt., 1998; 35:30–34. Spacie A, Landrum PF, Leversee GJ. Uptake, depuration and biotransformation of anthracene and benzo(a)pyrene in bluegill sunfish. Ecotoxicol. Environ. Safety, 1983; 7:330–341. Onken BM, Traina SJ. The sorption of pyrene and anthracene to humic acid mineral complexes: Effects of fractional organic carbon content. J. Environ. Qual., 1997; 26:126–132. Duxbury JL, Dixon DG, Greenberg BM. The effects of simulated solar radiation on the bioaccumulation of polycyclic aromatic hydrocarbons by the duckweed Lemna gibba. Environ. Toxicol Chem., 1997; 16:1739–1748. Landrum PF, Reinhold MD, Nihart SR, Eadie BJ. Predicting the bioavailability of organic xenobiotics to Pontoporeia hoyi in the presence of humic and fulvic materials and natural dissolved organic matter. Environ. Toxicol. Chem., 1985; 4:459–467. Landrum PF, Nihart SR, Eadie BJ, Herche LR. Reduction in bioavailibility of organic contaminants to the amphipod Pontoporeia hoyi by dissolved organic matter of sediment interstital waters. Environ. Toxicol. Chem., 1987; 6:11–20. Chen Y, Senesi N. Interaction of toxic organic chemicals with humic substances. In: Gerstl Z, Chen Y, Mingelgrin U, Yaron B eds. Toxic organic chemicals in porous media. Berlin: Springer, 1997:37–90. Senesi N. Binding mechanisms of pesticides to soil humic substances. Sci. Total Environ., 1992; 123/124:63–76. Perminova I, Frimmel FH, Kovalevskii DV, Abbt-Braun G, Kudryavtsev AV, Hesse S. Development of a predictive model for calculation of molecular weight of humic substances. Water Res., 1998; 32:872–881. Lee SK, Freitag D, Steinberg CEW, Kettrup A, Kim YH. Effects of dissolved humic materials on acute toxicity of some organic chemicals to aquatic organisms. Water Res., 1993; 27:199–204. Ahtiainen J, Wagman N, Oberg LG, Jorgensen KS. Fate and toxicity of chlorophenols, polychlorinated dibenzo-p-dioxins and dibenzofurans during composting of contaminated sawmill soil. Environ. Sci. Technol., 1997; 31: 3244–3250. Pavlikova D, Tlustos P, Szakova J, Balik J. The effect of application of potassium humate on the content of cadmium, zinc and arsenic in plants. Rostlinna Vyroba, 1997; 43:481–486. Senesi N, Loffredo E. Soil humic substances. In: Hofrichter M, Steinbüchel A eds. Biopolymers. Volume 1: Lignin, humic substances and coal. Weinheim: Wiley-VCH, 2001:247–299. Perdue EM. Chemical composition, structure and metal binding properties. In: Hessen DO, Tranvik LJ eds. Aquatic humic substances —ecology and biogeochemistry. New York: Springer, 1998, 133. Flaig W. Effects of humic substances on plant metabolism. Proc. 2nd Int. Peat Congress, Leningrad, 1970:579–606. Chen Y, Aviad T. Effects of humic substances on plant growth. In: MacCarthy P, Clapp CE, Malcolm RL, Bloom PR eds. Humic substances in soil and crop sciences; Selected readings. Madison, WI: American Society of Agronomy and Soil Science Society of America, 1990:161–186. Chen Y, Hoitink HAJ, Madden LV. Microbial activity and biomass in container media predicting suppressiveness to damping-off caused by Pythium ultimum. Phytopathology, 1988; 78:1447–1450. Chen Y, Clapp CE, Magen H, Cline VW. Stimulation of plant growth by humic substances: Effects of iron availability. In: Ghabbour EA, Davies G eds. Understanding humic substances—Advanced methods, properties and applications. Cambridge: Royal Society of Chemistry, 1999:255–263 Pflugmacher S, Spangenberg M, Steinberg CEW. Dissolved organic matter (DOM) and effects on the aquatic macrophyte Ceratophyllum demersum in relation to photosynthesis, pigment pattern and activity of detoxication enzymes. J. Appl. Botany, 1999; 73:184–190. Pflugmacher S, Tidwell LF, Steinberg CEW. Dissolved humic substances can directly affect freshwater organisms. Acta Hydrochim. Hydrobiol., 2001; 29: 34–40. Casper SJ, Krausch HD. Pteridophyta and anthophyta, parts 1 and 2. In: Ettl H, Gerloff J, Heynig H eds. Sü wasserflora von mitteleuropa, 23 and 24. Stuttgart: Gustav Fischer Verlag, 1980. DIN EN 1484: Anleitungen zur bestimmung des gesamten organischen kohlenstoffs (TOC) und des gelösten organischen kohlenstoffs, DEV, H3, 40. Lieferung 1998. Pflugmacher S, Steinberg CEW. Activity of phase I and phase II detoxication enzymes in aquatic macrophytes. J. Appl. Botany, 1997; 71:144–146. Habig W, Pabst MJ, Jacoby WB. Glutathione S-transferase: The first step in mercapturic acid formation. J. Biol. Chem., 1974; 249: 1730–1739. Drotar A, Phelps P, Fall R. Evidence for glutathione peroxidase activities in cultured plant cells. Plant Sci., 1985; 42:35–40.

222

| CHAPTER 22: HUMIC SUBSTANCES AND THEIR DIRECT EFFECT ON AQUATIC PLANTS

26.

Bradford M. A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of proteindye-binding. Anal. Biochem., 1976; 72:248–254. Huber SA, Frimmel FH. Flow injection analysis of organic carbon in the low-ppb range. Anal. Chem., 1991; 63:2123–2130. Huber SA, Frimmel FH. Size-exclusion chromatography with organic carbon detection (LC-OCD): A fast and reliable method for the characterization of hydrophilic organic matter in natural waters. Vom Wasser, 1996; 86:277–290. Sachse A, Babenzien D, Ginzel G, Gelbrecht J, Steinberg CEW. Characterization of dissolved organic carbon (DOC) in a dystrophic lake and an adjacent fen. Biogeochem., 2001; 54:279–296. Gorduza VM, Wagner L, Comanita, E. Studies of the aggregation of a new benzimidazolon-amino-anthraquinone dye in waterdioxane mixtures. Rev. Roum. Chim., 1995; 40:759–763. Becker HGO. Einführung in die Photochemie, Dt. Verl. der Wiss., Berlin 1991. Haitzer M, Höss S, Traunspurger W, Steinberg CEW. Effects of dissolved organic matter (DOM) on the bioconcentration of organic chemicals in aquatic organisms—a review. Chemosphere, 1998; 37:1335–1362. Day KE. Effects of dissolved organic carbon on accumulation and acute toxicity of fenvalerate, deltamethrin and cyhalothrin to Daphnia magna (Straus). Environ. Toxicol. Chem., 1991; 10:91–101. McCarthy JF, Jimenez BD, Barbee T. Effect of dissolved humic material on accumulation of polycyclic aromatic hydrocarbons: Structure-activity relationship. Aquatic Toxicol., 1985; 7:15–24. Kukkonen J, Oikari A. Bioavailability of organic pollutants in boreal waters with varying levels of dissolved organic matter. Water Res., 1991; 25:455–463. Sen AK, Mondal NG, Mandal S. Studies of uptake and toxic effects of Cr(VI) on Pista Stratiotes. Wat. Sci. Tech., 1987; 19:119–127. Thurman EM. Organic geochemistry of natural waters. The Netherlands, Dordrecht: Martinus Nijhof/Dr. W.Junk Publishers, 1985. Ridge I, Walters J, Street M. Algal growth control by terrestrial leaf litter: a realistic tool? Hydrobiologica, 1999; 385/386:173–180. Loffredo E, Senesi N, Dorazio V. Effects of humic acids and herbicides and their combinations on the growth of tomato seedlings in hydroponics. Zeitschrift für Pflanzenernährung und Bodenkunde, 1997; 160:455–461. Ziechmann W. Huminstoffe. Mannheim: Spektrum Akademischer Verlag, 1996. Wang WH, Bray CM, Jones MN. The fate of 14C-labelled humic substances in rice cells in culture. J. Plant Physiol., 1999; 154: 203–211. Helbig B, Klöcking R. Darstellung und charakterisierung von huminsäure-modellsubstanzen. Z. Physiother., 1983; 33:31–37. Wohlrab W, Helbig B, Klöcking R, Sprössig M. Penetrationskinetik eines potentiellen Virustaticums in die menschliche Haut. Pharmazie, 1984; 39: 562–564. Müller-Wegener U. Interaction of humic substances with biota. In: Frimmel FH, Christman RF eds. Humic substances and their role in the environment, New York: Wiley, 1988:179–192. Chaminade R, Blanchet R. Mécanisme de l’action stimulante de l′ humus sur la nutrition minérale des végétaux. C.R. Acad. Sci Paris, 1953; 237:1768–1770. Gjessing ET, Alberts JJ, Bruchet A, Egeberg PK, Lydrsen E, McGown LB, Mobed JJ, Münster U, Pempkowika J, Perdue M, Ratnawerra H, Rybacki D, Takacs M, Abbt-Braun G. Multi-method characterisation of natural organic matter isolated from water: Characterisation of reverse osmosis-isolates from water of two semi-identical dystrophic lakes basins in Norway. Water Res., 1998; 32: 3108–3124. Roy S, Ihantola R, Hänninen O. Peroxidase activity in lake macrophytes and in relation to pollution tolerance. Environ. Experi. Bot., 1992; 32:457–464. Boyle TD, Sutton HD, Klaine SJ. Evaluation of peroxidase as a biochemical indicator of toxic chemical exposure in the aquatic plant Hydrilla verticillata, Royle. Environ. Toxicol. Chem., 1994; 13:509–515. Catafamo JL, Feinberg JH, Smith GW, Birecka H. Effects of gibberellic acid and ethylene on peroxidase in pea internodes. J. Exp. Bot., 1978; 29:347–358. Kwak SS, Kim SK, Park IH, Liu JR. Enhancement of peroxidase activity by stress-related chemicals in sweet potato. Phytochem., 1996; 43:565–568. Wetzel A, Werner D. Quinoline increases ascorbate peroxidases and dehydrascorbate reductase activity in Vicia faba nodules. Bull. Environ. Contam. Toxicol., 1990; 45:619–626. Oberg LG, Glas B, Swanson SE, Rappe C, Paul KG. Peroxidase-catalyzed oxidation of chlorophenols to polychlorinated dibenzo-pdioxins and dibenzofurans. Arch. Environ. Contam. Toxicol., 1990; 19:930–938. Senesi N, Schnitzer M. Effects of pH, reaction time, chemical reduction and irradiation on ESR spectra of fulvic acid. Soil Sci, 1977; 123:224–234 Zepp RG. Environmental photoprocesses involving natural organic matter. In: Frimmel FH, Christman RF eds. Humic substances and their role in the environment. New York: Wiley, 1988:193–214. Ladd JN, Butler JHA. The effect of methylation of humic acids on their influence on proteolytic enzyme activity. Austr. J. Soil Res., 1969; 7:263. Vaughan D, Malcolm RE. Effect of soil organic matter on peroxidase activity of wheat roots. Soil Biol. Biochem., 1979; 11:57–63. Metodiewa D, Jaiswal AK, Narimantas C, Dickancaite E, Segua-Aguilar J. Quercetin may act as a cytotoxic prooxidant after its metabolic activation to semiquinone and quinoidal product. Free Radical Biol. Med., 1996; 26: 107–116. Ladd JN, Butler JHA. Humus-enzyme systems and synthetic organic polymerenzyme analogs. In: Paul EA, McLaren AD eds. Soil biochemistry, New York: Dekker, 1975:143–194.

27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.

MATERIALS AND METHODS |

59. 60. 61. 62. 63. 64. 65.

66. 67. 68. 69. 70.

223

Bukova M, Tichy V. The effects of humus fractions on the phosphorylase activity of wheat (Triticum aestivum L.). Biol. Plant., 1967; 9:401–406. Pflug W. Über aktivitätsänderungen ausgewählter hydrolytischer Enzyme durch Huminstoffe. Dissertation, Göttingen, Germany: Georg August Universität, 1978. Oettmeier W, Masson K, Donner A. Anthraquinone inhibitors of photosystem II electron transport. FEBS Letters, 1988; 231: 259–262. Senesi N, Steelink C. Application of ESR spectroscopy to the study of humic substances. In: Hayes MHB, MacCarthy P, Malcolm RL, Swift RS eds. Humic substances II. Chichester: Wiley, 1989. Scott DT, McKnight DM, Blunt-Harris EL. Kolesar SE, Lovley DR. Quinone moieties act as electron acceptors in the reduction of humic substances by humic-reducing microorganisms. Environ. Sci. Technol., 1998; 32:2984–2989. Steelink C, Tollin G. Biological polymers related to catechol: electron paramagnetic resonance and infrared studies of melanin, tannin, lignin, humic acid and hydroquinones. Biochim. Biophys. Acta, 1966; 112:377–379. Steinberg CEW, Haitzer M, Brüggemann R, Perminova IV, Yashchenko NY, Petrosyan VS. Towards a quantitative structure activity relationship (QSAR) of dissolved humic substances as detoxifying agents in freshwaters. Internat. Rev. Hydrobiol, 2000; 85: 253–266. Olk DC, Brunetti G, Senesi N. Organic matter in double-cropped lowland rice soils: Chemical and spectroscopic properties. Soil Sci., 1999; 164:633–649. Jerzykiewicz M, Drozd J, Jezierski A. Organic radicals and paramagnetic metal complexes in municipal solid waste composts. An EPR and chemical study. Chemosphere, 1999; 39:253–268. Jezierski A, Czechowski H., Jerzykiewicz M, Drozd J. EPR investigations of structure of humic acids from compost, soil, peat and soft brown coal upon oxidation and metal uptake. Appl. Magn. Reson., 2000; 18:127–136. Weyl GmbH. Vorläufiges Datenblatt RHS1500. Ausgabe 10 (1990). EP-B 0281678 v. 10.02.1987, EP 0537427 v. 31.07.1992. Vuori KM, Muotka T. Benthic communities in humic streams. In: Keskitalo J, Eloranta P eds. Limnology of Humic Waters, Leiden: Backhuys Publishers, 1999:193–207.

Chapter 23 MORE EVIDENCE FOR HUMIC SUBSTANCES ACTING AS BIOGEOCHEMICALS ON ORGANISMS C.Wiegand,1,2 N.Meems,1 M.Timoveyev,3C.E.W.Steinberg1 and S.Pflugmacher1 1Leibniz-Institute

2Institute

of Freshwater Ecology and Inland Fisheries, Mueggelseedamm 301, 12587 Berlin, Germany

of Biology, Humboldt-University of Berlin, Chausseestr 117, 10115 Berlin, Germany 3Irkutsk

State University, Karl Marx 1, 664003, Irkutsk, Russia 23.1. INTRODUCTION

Humic substances (HSs) are the main component of dissolved organic matter (DOM) in aquatic ecosystems [1,2]. Depending on the source of organic material involved in the humification process, HSs have a variety of molecular structures that differ with their source. Microbial and abiotic degradation of source materials results in a variety of HSs molecular structures, such as acylaromatic, quinoid and aliphatic structures in the core, and amino acid or carbohydrate-like structures and carbonyl-, carboxyl-, phenyl-, and hydroxyl-groups at the periphery [3,4]. The ecochemical relevance of HSs is mostly discussed in terms of their ability to bind or integrate pollutants like organic xenobiotics and heavy metals, consequently decreasing the bioavailability and the toxicity of these pollutants [5,6]. The same functional groups may also engage in direct interactions with biological systems, such as the following: membrane alteration by either accumulation [7] or tensid-like action of HSs [8–10]; alteration of growth and reproduction of the invertebrates Caenorhabditis elegans [11] and Daphnia magna [12] as well as of plants [13]; changing enzyme activities [14]; and antimicrobial and antimycotic activity [15]. In addition, photochemical reactions can initiate a chain reaction, leading to reactive oxygen species (ROS), e.g., 1O2, O2− and H2O2 [16,17], which can provoke oxidative damage to the cell structures of organisms. Thus, altered oxygen-stress enzymes were caused by HSs in the plant Ceratophyllum demersum [18]. In this study, direct effects of HSs on biological systems were investigated using the following parameters. Uptake of a Humic Acid-like Substance. First, uptake of a 14C-radiolabelled oxidation product of caffeic acid with a molar mass of approximately 600 Da was tested as a substitute for HSs. Uptake or membrane accumulation of HSs or their breakdown products is a requirement for direct interaction with cellular metabolism, as described below. Immobility and Lethality as General Toxicological Parameters. Immobility and lethality describe the consequences of exposure to a stressor for the single organism. They are generally accepted parameters for determining toxicity of unknown substances and calculation of risk assessment. However, the sensitivity of these parameters usually is low. Expression of Heat Shock Protein 70 as a Parameter for General Physiological Stress. Heat shock proteins are a chaperone family that protects proteins in cells during stress phases. They are named after the stress leading to their detection and their specific molar mass. Their cellular content increases after stress as caused by heat, but also by heavy metals and xenobiotics, with denaturated proteins being the causative principle [19]. Heat shock protein 70 (Hsp 70) binds to de novo synthesized proteins to prevent mistakes in folding. In this form the protein is transported to its target, where it achieves its final form and function. Hsp 70 also binds to a partly denatured protein to either refold it correctly and retain its function, or, if damage is too severe, guide it to locations for controlled lysis. Activity of the Oxidative Stress Enzymes Peroxidase and Glutathione Peroxidase. Peroxidase (POD) and glutathioneperoxidase (GPx) are enzymes that protect cells from oxidative damage caused by oxygen itself or by reactive oxygen species (ROS) like 1O2, O2− and H2O2 [20]. These ROS may cause lipid-, DNA-, or protein oxidation, with disfunction of membranes, gene regulation and enzyme function being the consequences. POD mainly reduces hydrogen peroxide to water and oxygen, while GPx also reduces organic hydroperoxides (mainly lipid hydroperoxides), thereby protecting membranes from damage or destruction. Activity of the Detoxication Enzyme System Microsomal and Soluble Glutathione S-Transferase. Microsomal and soluble glutathione S-transferases (m-, sGST) are ubiquitous conjugation enzymes of the detoxication pathway. They react with moderate hydrophilic xenobiotica, which include electrophilic groups. Activation by another enzyme system like cytochrome P-450 monooxygenases is not required if the xenobiotic already has a functional group for the GSTs. Conjugation to glutathione increases water solubility of the substances and supports their excretion. Broad substrate specificity is attained by several soluble GST isoenzymes, and one microsomal form [21]. The common carp (Cyprinus carpio), the waterflea (Daphnia magna) and three amphipod species (Gammarus tigrinus, Gammarus ischnus and Eulimnogammarus cyaneus) were used for the investigations with the following HSs: Suwannee

MATERIALS AND METHODS |

225

River FA, HA and NOM (XAD isolated International Humic Substances Society (IHSS) standards), Svartberget, Humex A, Humex B, Nordic Reference and Sanctuary Pond (all reverse osmosis (RO) isolated NOMs), and synthetic HS 1500, a hydroquinone radical polymer linked by alkyl spacers [22]. Investigations were conducted to answer the following questions: 1) are different organisms affected by HSs? 2) are similar effects caused by HSs from different sources? 3) are the intensities of these effects concentration dependent? and 4) are the responsible HSs structures recognizable? 23.2. MATERIALS AND METHODS 23.2.1. Rearing of the Organisms Carp and Daphnids were reared at 20°C and a light regime of 14 h light and 10 h dark in artificial tank water (reverse osmosis water containing 100 mgL−1 sea salt, 200 mgL−1 CaCl2•2H2O and 103 mgL−1 NaHCO3). The water was exchanged weekly. Amphipods were reared separately by species at a temperature of 6–8°C in aerated tanks. Carp were fed daily with commercial fish pellets. Daphnids were fed with algal powder every second day, and gammarids with commercial fish flakes twice per week. Preculture was 2 months for the carp and 4 weeks for the Daphnids and the gammarids. Juveniles and adult female Daphnids were kept for the culture. The other Daphnids, except for freshly shed animals, were used for the tests. 23.2.2. Humic Substances The radiolabelling level of the caffeic acid oxidation product was 53.28 kBq/mg, with molar mass 600 Da according to [23]. The isolated humic substances used were as follows: IHSS standard purified from Suwannee River (Georgia) by filtration and XAD-8 chromatography. Suwannee River FA and HA, as well as NOM were used. Svartberget, Humex A, Humex B, and Nordic Reference, all of which were purified by reverse osmosis (R.Vogt, E.Gjessing, Oslo, Norway). Sanctuary Pond NOM, purified by reverse osmosis (K.Burnison, Ontario, Canada). HS 1500, a synthetic HS was made from hydroquinone radicals and linked via alkyl spacers, leading to an HS with molar mass 1500 (Rüttgers Organics GmbH, Mannheim, Germany). 23.2.3. Uptake of the Radiolabelled Caffeic Acid Oxidation Product The caffeic acid oxidation product was dissolved in artificial tank water to a concentration of 125 µgL−1 (40.000 dpm in 100 mL). Three samples of Gammarus tigrinus (0.272, 0.318, 0.304 g fresh wight (FW)) were exposed for 24 h and rinsed 3 times with 5 mL water to remove adherent caffeic acid oxidation product. After homogenization, 3 subsamples of 10 µL were mixed with 4 mL scintillation cocktail (U1-tima-Gold-XR, Packard Instruments) and radioactivity in the samples was determined by liquid scintillation counting (Tri-Carb 1900TR, Canberra Packard, Dreieich, Germany). 23.2.4. Analysis of Silver and Mercury Suwannee River humic samples were tested for contamination with silver and mercury. Svartberget NOM was tested for contamination with mercury. Silver was analysed with a flame AAS (Perkin-Elmer 3300, ′ =328, 1 nm, air/acetylene-flame). Mercury was measured as total Hg (Hg(tot)) with a TMA 254 analyzer (Biologisch-Chemisches Institut, Hoppegarten, Germany). 23.2.5. Animal Exposure to Determine Physiological Reactions For exposure of the carp, Suwannee River FA, HA and NOM were dissolved at concentrations of 5 mgL−1. Svartberget NOM, Humex A, Humex B, and Nordic Reference were dissolved at concentrations of 0.5 mgL−1. All solutions were aerated for 12 h before 3 carps were exposed for 24 h. For exposure of Daphnids, Suwannee River NOM, Svartberget NOM and HS 1500 were dissolved at concentrations from 0.5– 50.0 mgL−1 (Suwannee River NOM up to 100 mgL−1), and Daphnids were exposed for 24 h. Amphipods were exposed to 50 mg/L Sanctuary Pond NOM for 24 h for the heat shock protein (Hsp 70)

226

| CHAPTER 23: MORE EVIDENCE FOR HUMIC SUBSTANCES ACTING AS BIOGEOCHEMICALS

determination and for up to 9 days for POD activity measurement. After exposure, Carp gills, livers and white muscles, respectively, and the Daphnids and amphipods were immediately frozen in liquid nitrogen and stored at −80°C. 23.2.6. Toxicological Parameters Immobility and lethality of Daphnids were observed and semiquantified after 24 h before termination of the experiment. Immobility was semiquantified in six groups (0–10, 10–20, 20–30, 30–40, 40–60, 60–80%) and dead Daphnids were counted to calculate absolute as well as relative lethality and the factor increase relative to the control group, which was not exposed to HSs. 23.2.7. Hsp 70 preparation Hsp 70 samples were prepared from 0.1 g FW organism, separated by SDS-PAGE and blotted to PVDF-membrane. The Hsp 70-protein bands were identified by mouse monoclonal antibody ′ -Hsp 70 (clone BRM-22, Sigma®) via the alkaline phosphatase reaction [24]. The blots show either the same tissue of three control carp (lines 1–3) and three exposed carp (lines 4–6) to equalize differences between individuals exposed to the same treatment, or one control carp versus one treated carp. Hsp 70 samples of the amphipods were prepared from one animal each. Blots show control and treated animals. 23.2.8. Enzyme Preparation and Measurement Enzymes were extracted according to [25] by homogenization, removal of cell debris, centrifugation at 100,000 g to obtain the microsomal fraction, the ammonium sulfate precipitation cut between 35% and 80% saturation, followed by centrifugation and desalting to isolate the soluble proteins. Activities of POD in the soluble protein fraction were measured according to [26] using guajacol as the substrate; GPx was determined according to [27]. M- and sGST were measured according to [28] using chlorodinitrobenzene (CDNB) as the substrate. The protein contents of the samples were determined according to [29]. 23.3. RESULTS 23.3.1. Uptake of the 14C-Radiolabelled Caffeic Acid Oxidation Product Radioactivity was clearly confirmed in G. trigrinus after 24 h exposure to the 14C-radiolabelled caffeic acid oxidation product. The uptake was 82845±9323 dpm per gram fresh weight of the animals, corresponding to 6.9±1.0% of the applied radioactivity. This radioactivity originates from the caffeic acid oxidation product, although at that stage of the experiments we could not differentiate between accumulation of the whole molecule or breakdown products and accumulation at cell surfaces or into the cell. 23.3.2. Contamination by Silver and Mercury One of the tested HSs samples (the Suwannee River HA) showed very high contamination by Ag and Hg(tot) (Table 23.1) due to the application of silver filters during the purification process. The FA and the NOM exhibited only weak Ag signals and low Hg contamination. Svartberget NOM showed minimal contamination by Ag and Hg. Table 23.1 Metal contaminations of Suwannee River FA, HA, and NOM, and of Svartberget NOM, and the calculated metal concentrations in the media during exposure of the carp Silver (Ag) aSRFA aSRHA

Mercury (Hg(tot)) µg/g dry wt

conc.

3 250

0.015 1.25

[µg/L]d

µg/g dry wt

conc. [µg/L]

0.272 10.39

0.0014 0.052

MATERIALS AND METHODS |

227

Figure 23.1 Western Blot detection of expression of the heat shock protein Hsp 70 in gills of carp, and in gammarids after exposure to HSs. A: Carp gills, lane 1–3: control carp, lane 4–6 exposure to 5 mg L−1 isolated Suwannee River HS fractions, as indicated. Each line represents a sample of one single carp to show individual differences. B: Carp gills of control versus exposed carp. Exposure to 0.5 mg L−1 of the named NOMs from the Nordic countries, from left to right: HUMEX A, HUMEX B, Nordic Reference, Svartberget, all NOM. C: Amphipods exposed to 50 mg L−1 of Sanctuary Pond NOM lines 1, Gammarus tigrinus control, 2, G. tigrinus exposed, 3, G. ischnus control, 4, G. ischnus exposed. Detection of Hsp 70 in Eulimnogammarus cyaneus failed (data not shown) Silver (Ag)

Mercury (Hg(tot)) µg/g dry wt

aSR

conc.

[µg/L]d

µg/g dry wt

conc. [µg/L]

0.0003e 0.00003−0.006f bSvartb NOM c ND – 0.112 0.000056−0.0056f a SR: Suwannee River; b Svartb: Svartberget; c ND, not measured; d calculated concentrations during exposure; e carp; f Daphnids NOM

3

0.015e

0.053

0.0015−0.3f

23.3.3. Acute Toxicity to Daphnia Magna Svartberget NOM up to 50 mgL−1 caused no toxicological effects. Suwannee River at concentrations of 50 and 100 mgL−1 caused immobility of 20% of the Daphnids versus 10% in the control. HS 1500 increased the immobility to 30%, and lethality to 6%, whereas the control exhibited a maximum at 3%. 23.3.4. Hsp 70 Expression The expression of Hsp 70 in carp showed different background intensities in the control fish, depending on the tissue: high expression in the liver but low to very weak expression in the gills and muscles. In the gills, which are the main contact organs during exposure via the medium, the moderate expression of Hsp 70 was markedly increased by all fractions of Suwannee River HS (Figure 23.1A). The intensity of Hsp 70 expression in the liver due to the high metabolic activity was not increased by HSs (data not shown). The muscles showed nearly no reaction to the HSs (data not shown). The different Suwannee River HSs fractions (FA, HA and NOM) caused almost no different reactions, except that FA provoked a slightly stronger reaction in the gills. Expression of Hsp 70 after exposure of the carp to Svartberget NOM, Humex A, Humex B and Nordic Reference NOMs was only investigated in the gills. All control carp showed weak or no Hsp 70 expression, but Hsp 70 expressions were slightly increased by Svartberget and Nordic Reference NOMs and clearly increased by Humex A and Humex B NOMs (Figure 23.1B). The amphipods already showed a significant basic Hsp 70 expression that was increased by a high concentration (50 mgL−1) of Sanctuary Pond NOM (Figure 23.1C).

228

| CHAPTER 23: MORE EVIDENCE FOR HUMIC SUBSTANCES ACTING AS BIOGEOCHEMICALS

Figure 23.2A Activity of glutathione peroxidase in Daphnids exposed to increasing concentrations of HSs: HS 1500, a synthetic polymer with alkyl spacers from hydroquinone radicals; SR NOM, Suwannee River natural organic matter (IHSS), SB, Svartberget NOM, reverse osmosis isolated by Vogt et al. HS 1500 provoked a minor concentration dependent elevation of GPx in Daphnids

23.3.5. Activities of Oxygen-Stress Enzymes POD and GPx In Daphnids, activity of GPx was elevated by all the NOMs tested. For HS 1500, a concentration dependence may be suggested but it was not statistically significant (Figure 23.2A). The time dependence showed that the POD activity of all the amphipod species was elevated by Sanctuary Pond NOM even after 30 min exposure time (Figure 23.2B). Only G. tigrinus showed a markedly basic POD. In the European species, POD activity markedly increased within 30 minutes of exposure and was maintained at that higher level during the whole exposure time. POD activity of G. ischnus was decreased by the end of the experiment but not down to the control level. In contrast, E. cyaneus POD activity increased uniformly to its maximum, which was approximately 5 times the control value, after 3 days of exposure and then decreased during succeeding days to the control level.

MATERIALS AND METHODS |

229

Figure 23.2B Time dependence of peroxidase activity in three different amphipod species exposed to 50 mgL−1 of Sanctuary Pond NOM. Highest increase of POD was found for Gammarus ischnus throughout the whole exposure, whereas the POD activity in G. tigrinus was already high in controls. The POD activity of Eulimnogammarus cyaneus was by almost a factor 100 lower than for the other two species and showed very weak reaction towards the HSs (note the gap in the x-axis).

Figure 23.3 Concentration dependent increase of soluble glutathione S-transferase activity of Daphnia magna exposed to different HSs for 24 h (for abbreviations see Figure 23.2A). Depending on the source of the HS, elevation was significant from different concentrations upwards, with HS 1500 provoking the steepest elevation.

23.3.6. Activities of Glutathione S-Transferases Detoxication Enzymes The soluble GST of Daphnids showed a concentration dependent increase of activity caused by all three HSs. The strongest reaction was provoked by HS 1500. Suwannee River NOM caused significant elevation from 10.0 mgL−1, Svartberget NOM from 1.0 mgL−1, and HS 1500 from 0.5 mgL−1 upwards (Figure23.3). Microsomal GST showed no effect (data not shown). 23.4. DISCUSSION This study shows that HSs cause direct effects in different aquatic organisms such as carp, Daphnids and amphipods. After exposure to HSs, expression of Hsp 70, a gen eral physiological stress parameter, was increased in the gills of carp and in amphipods. Also, different oxygen stress related enzymes (POD, GPx) and detoxication enzymes (m- and sGST) were modulated by HSs in Daphnids and amphipods. Furthermore, toxicological parameters such as immobility and lethality were increased in Daphnids. For these physiological interactions obviously caused by HSs, a necessary requirement is that the HSs come into contact with cell surfaces or cell internal structures. In contrast with biomedical research where such interactions are generally accepted within a distinct molecular size of HSs [30], reports concerning uptake and direct effects of HSs are

230

| CHAPTER 23: MORE EVIDENCE FOR HUMIC SUBSTANCES ACTING AS BIOGEOCHEMICALS

hard to find in the ecological field. We proved that some of the 14C-radiolabelled product was accumulated by an amphipod species as well as in an aquatic plant (Ceratophyllum demersum), the latter shown by Pflugmacher et al. (this volume). Even though we cannot decide if the whole HS molecule or some smaller breakdown products were absorbed into the cell, effects such as general stress, toxicity, and alteration of oxygen stress related and detoxication enzymes are provoked. The mechanism could be either on the cell membrane by adsorption or a tensid-like action, both of which alter membrane permeability [7,9] and membrane structures [31], or interaction with cell receptors, as shown by [30]. The other option is that HSs or breakdown products penetrate the cell, causing disturbances of intracellular structures, as indicated by the increase of Hsp 70 expression. Expression of Hsp 70 increases if denaturated proteins occur in the cell, no matter if physical stress, like heat, or chemical stress, like xenobiotica, lead to their denaturation [32]. Thus, the HSs used apparently caused protein denaturation in the gills of the carp and in the amphipods. Different direct interactions of HSs with proteins are possible, but a mechanism leading to assumed protein denaturation has not been investigated yet. A further possibility is HS-mediated generation of reactive compounds (as mentioned below), which then will cause damage to the proteins. HSs from different sources caused similar effects but they varied in intensity. The strength of the Hsp 70 bands in carp gills increased to different extents depending on the HSs used. For example, Humex A and B as well as Nordic Reference NOMs caused stronger Hsp 70 expression than Svartberget NOM. Elevations of POD and GPx occur on exposure to increasing levels of reactive oxygen species like O2−, H2O2, the GPx and organic peroxides, especially lipid peroxides [20]. ROS-like compounds can be generated from HSs during photolysis by sunlight, although the main breakdown products are short chain fatty acid-like compounds [16]. Photoreactions can also lead to singlet oxygen [33] and the formation of superoxide, which leads to hydrogen peroxide [34,35], hydroxyl radicals and organic peroxides [36]. All these ROS will lead to an increased oxygen stress situation for organisms, which, to avoid damage to biological structures such as proteins, lipids or DNA, will result in increased levels of antioxidative enzymes. Our data show that the amphipods used for POD determination had to resist oxygen stress caused mainly by hydrogen peroxide, the main substrate for the POD, whereas the GPx, which responds to organic peroxides, was not affected as much, as shown in Daphnids. Lipid peroxidation products probably were not caused by the HSs used in Daphnids. A very large increase of POD was also found in the aquatic plant Ceratophyllum demersum after exposure to isolated HSs from Lake Holosee [18]. The functional groups of the HSs obviously are recognized either by the detoxication enzymes of the GST itself in the cells or their specific Ah-receptor, which induces both P450 enzymes and GST. The sGST is induced via the DRE (dioxin responsible element) as well as via the ARE (antioxidant responsible element). The latter also induces the GPx [37]. Both reponses lead to the result that, like xenobiotica, HSs are considered foreign substances by the cell. To prevent further damage, the resulting reaction of the cell is to metabolise these compounds to an excretable form and hence detoxify them. Activation of HS by P450 enzymes is not necessary if electrophilic groups exist. An electrophilic group for the GST in HSs molecules can, for example, expose double bonds, epoxides, quinoid structures or ether groups. The GST conjugates glutathione to such groups to increase water solubility, which promotes excretion processes. In Daphnids, the elevation of the sGST was clearly concentration dependent in addition to the specificity caused by the HSs source. Svartberget NOM provoked effects in relatively high concentrations from 10 mgL−1 upwards, whereas 1 mgL−1 Suwannee River NOM and HS 1500 from 0.5 mgL−1 upwards provoked significant sGST elevation. The relatively high concentration of hydrophobic acids in Svartberget NOM and Suwannee River NOM could be responsible for this xenobiotic-like action. The HS 1500, which mainly consists of quinoid structures, caused the strongest effects in both sGST and GPx of the Daphnids. This gives first hints that the quinoids in HSs might be one of the responsible structures. The metal contamination found in the HSs used was in no case the causative agent, because the concentrations of silver, (0. 0015–1.25 µ gL−1) and mercury (0.0003–0.052 µ gL−1) were too low. Minimal effect concentrations of ionic Ag+ were 3.7 µgL −1 reducing the Na+ influx, inhibiting the Na+/Ka+-ATPase, and accumulating gradually in gills [38]. Up to an exposure concentration of 0.3 µgL−1 mercury as methylmercuric chloride, no effects were observable till the 3rd generation of brook trout [39]. Metal contamination would have been expected to contribute to the Hsp expression, but it did not cause much additional effect. However, the other HSs fractions caused comparable increases of Hsp 70. Mercury in both organic and inorganic form is spontaneously conjugated to glutathione or cysteine in a strictly chemical reaction and has a high affinity for SH-groups of enzymes such as the GST, which are altered or blocked after Hg binding [40]. If mercury contamination in Suwannee River HA had contributed, a decrease of GST activity compared to both the NOM and control should have occurred, which was not the case. Both metals might be bound to HS by chelating reactions, covering their harmful effects. 23.5. CONCLUSIONS From our data we conclude that aquatic organisms can accumulate HSs or at least HSs breakdown products. They are adsorbed on the cell surface or taken up by the cells and affect different cellular reactions. The observed effects were a

MATERIALS AND METHODS |

231

general stress together with a toxicological and a physiological level. Effects differed with respect to the concentrations of the HSs used and the HSs sources. The HSs also influence the amount and reactions of the different structures. Thus, variation in ob served effects should lead to a quantitative structure relationship (QSAR) of the HSs. Such QSAR has already been demonstrated for the binding of polyaromatic hydrocarbons by humic substances [41]. Comparative studies with the synthetic HS 1500 revealed quinoid structures as a causative component, but further investigations are necessary. ACKNOWLEDGMENTS We are grateful for the valuable gifts of the HSs samples: R.Klöcking (Medizinische Akademie, Erfurt, Germany) for the radiolabelled caffeic acid oxidation product; Egil Gjessing (Agder College, Kristiansand, Norway) for Humex A, Humex B, and Nordic Reference; Rolf Vogt (University of Oslo, Norway) for Svartberget NOM; and Kent Burnison (Ontario Canada) for Sanctuary Pond. Thanks are also expressed to H.J.Exner (IGB, Berlin, Germany) for metals analysis and to K.Greulich (IGB, Berlin, Germany) for conducting the uptake experiments. REFERENCES 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

14. 15. 16. 17. 18.

19. 20. 21. 22. 23.

Steinberg C, Muenster U. Geochemistry and ecological role of humic substances in lakewater. In: Aiken GR, McKnight DM, Wershaw RL, MacCarthy P eds. Humic substances in soil, sediment, and water—geochemistry, isolation, and characterization. New York: Wiley, 1985:105–145. Thurman EM. Organic geochemistry of natural waters. Dordrecht: Martinus Nijhof/Dr. W.Junk Publishers, 1985. Schulten HR. The three-dimensional structure of humic substances and soil organic matter studied by computative analytical chemistry. Fres. J. Anal. Chem., 1995; 351:62–73. Hayes MHB, MacCarthy P, Malcolm RL, Swift, RS eds. Humic substances II: In search of structure. New York: Wiley-Interscience, 1989. Haitzer M, Höss S, Traunspurger W, Steinberg C. Effects of dissolved organic matter (DOM) on the bioconcentration of organic chemicals in aquatic organisms—a review. Chemosphere, 1998; 37:1335–1362. Anisimova MA, Perminova IV, Lebedeva GF. Detoxifying ability of humic acids toward the trifluralin herbicide. Eur. J. Soil Sci., 1998; 31:973–978. Campbell JH, Evans RD. Inorganic and organic ligand binding of lead and cadmium and resultant implications for bioavailabilty. Sci. Total Environ., 1987; 62:219–224. Visser SA. Physiological action of humic substances on microbial cells. Soil Biol. Biochem., 1985; 17:457–462. Tranvik LJ. Bacterioplankton growth on fractions of dissolved organic carbon of different molecular weights from humic and clear water. Appl. Environ. Microbiol., 1990; 56:1672–1677. Vigneault B, Percot A, Lafleur M, Campbell PGC. Permeability changes in model and phytoplankton membranes in the presence of aquatic humic substances. Environ. Sci. Technol, 2000; 34:3907–3913. Höss S, Haitzer M, Traunspurger W, Steinberg C. Refractory dissolved organic matter can influence the reproduction of Caenorhabditis elegans (Nematoda). Freshwater Biol., 2001; 46:1–10. Petersen RC Jr, Persson U. Comparison of the biological effects of humic materials under acidified conditions. Sci. Total Environ., 1987; 62:387–398. Chen Y, Magen H, Riov J. Humic substances originating from rapidly decomposing organic matter: Properties and effects on plant growth. In: Senesi N, Miano TM eds. Humic substances in the global environment and implication on human health. Amsterdam: Elsevier, 1994:427–443. Ziechmann, W ed. Huminstoffe und ihre Wirkungen. Berlin: Spektrum akademischer Verlag, 1996:122–147. Bärlocher F. Effects of drying and freezing autumn leaves on leaching and colonization by aquatic hyphomycetes. Freshwater Biol., 1992; 28:1–7. Frimmel FH. Impact of light on the properties of aquatic natural organic matter. Environ. Intern., 1998; 24:559–571. Scully NM, McQueen DJ, Lean DR, Cooper WJ. Hydrogen peroxide formation: The interaction of ultraviolet radiation and dissolved organic carbon in lake waters along a 43–75°N gradient. Limnol. Oceanog., 1996; 41:540–548. Pflugmacher S, Spangenberg M, Steinberg CEW. Dissolved organic matter (DOM) and effects on the aquatic macrophyte Ceratophyllum demersum in relation to photosynthesis, pigment pattern and activity of detoxication enzymes. J. Appl. Botany, 1999; 73:184–190. Sanders BM. Stress proteins in aquatic organisms: An environmental perspective. Crit. Rev. Toxicol., 1993; 23:49–75. DiGiulio RT, Washburn PC, Wenning RJ, Wingston GW, Jewell WC. Biochemical responses in aquatic animals: A review of determinants of oxidative stress. Environ. Toxicol. Chem., 1989; 8:1103–1123. Ketterer B, Meyer DJ, Clark AG. Soluble glutathione transferase isoenzymes. In: Sies H, Ketterer B eds. Glutathione conjugation, mechanisms and biological significance. London: Academic Press, 1988. Weyl GmbH: Vorläufiges Datenblatt RHS1500. Ausgabe 10 (1990). EP-B 0281678 v. 10.02.1987, EP 0537427 v. 31.07.1992. Helbig B, Klöcking R. Darstellung und Charakterisierung von Huminsaeure-Modellsubstanzen. Zeit. Physiotherapie, 1983; 33, 31–37.

232

| CHAPTER 23: MORE EVIDENCE FOR HUMIC SUBSTANCES ACTING AS BIOGEOCHEMICALS

24.

Dyer S, Brooks L, Dickson KL, Sanders BM, Zimmermann EG. Synthesis and accumulation of stress proteins in tissues of arseniteexposed fathead minnows (Pimephales promelas). Environ. Toxicol. Chem., 1993; 12:913–924. Wiegand C, Pflugmacher S, Oberemm A, Steinberg CEW.Activity development of selected detoxication enzymes during the ontogenesis of the Zebra fish (Danio rerio). Intern. Rev. Hydrobiol, 2000; 85:413–422. Bergmeyer HU ed. Methods of enzymatic analysis. Weinheim: VCH, 1983; Vol 2:267–268. Peters LD, Livingstone DR. Antioxidant enzyme activities in embryologic and early life stages of turbot. J. Fish Biol., 1996; 49: 986–997. Habig W, Pabst MJ, Jacoby WB. Glutathione S-transferase: The first step in mercapturic acid formation. J. Biol. Chem., 1974; 249: 1730–1739. Bradford M. A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of proteindye-binding. Anal. Biochem., 1976; 72:248–254. Beer AM, Lukanov J, Sargorchev P. The influence of fulvic and humic acids from peat on the spontaneous contractile activity of smooth muscles. Phytore-mediation, 2000; 7:407–415. Thamm J, Jansen W, Rahmann, H. Effect of humic material on aquatic invertebrates in streams of a raised bog complex. In: Drozd J, Gonet SS, Senesi N, Weber J eds. The role of humic substances in the ecosystems and in environmental protection. Wroclaw: PTSHPolish. Soc. Humic Substances, 1997:929–935. Hightower LE. Heat shock, stress proteins, chaperones, and proteotoxicity. Cell, 1991; 66:191–197. Zepp RG, Wolfe NL, Baughman GL, Hollis RC. Singlet oxygen in natural waters. Nature, 1977; 207:421–423. Horth H, Frimmel FH, Hargitai L, Hennes EC, Huc AY, Mueller-Wegener U, Niemeyer J, Nissembaum A, Sekulov I, Tipping E, Weber JH, Zepp RG. Environmental reactions and functions, group report. In: Frimmel FH, Christman RF eds. Humic substances and their role in the environment. New York: Wiley, 1988:245–256. Blough NV, Zepp RG. Reactive oxygen species in natural waters. In: Foote CS, Valentine JS, Greenberg A, Liebmann, JF eds. Active oxygen in chemistry. New York: Chapman and Hall, 1995:280–333. Larson RA, Smykowski K, Hunt LL. Occurrence and determination of organic oxidants in rivers and wastewaters. Chemosphere, 1981; 10:1335–1338. Fernandes CL, Chavan SJ, Dong JH, Bornmann WG, Polski B, Chisari FV, Montali JA, Schmidi DE, Prochaska HJ. Regulation of glutathione S-transferases: Clues from a worm, a virus and a mouse with hepatitis. In: Vermeulen NPE, Mulder GJ, Nieuwenhuyse H, Peters WHM, van Bladeren PJ eds. Glutathione S-transferases: Structure, function and clinical implications. London: Taylor & Francis, 1996:97–109. Bury NR, McGeer JC, Wood CM. Effects of altering freshwater chemistry on physiological responses of rainbow trout to silver exposure. Environ. Toxicol. Chem., 1999; 18:49–55. McKim JM, Olson GF, Holcombe GW, Hunt EP. Long term effects of methylmercuric chloride on three generations of brook trout (Salvenilus fontinalis): Toxicity, accumulation, distribution, and elimination. J. Fish. Res. Board Canada, 1976; 33:2726–2739. Dierix PJ. In vitro interaction of organic mercury compounds with soluble glutathione S-transferases from rat liver. Pharmacol. Res. Commun., 1985; 17: 489–500. Perminova IV, Yashchenko N, Yu N, Petrosyan VS. Relationship between struc ture and binding affinity of humic substances for polyaromatic hydrocarbons: Relevance of molecular descriptors. Environ. Sci. Technol., 1999; 33: 3781–3787.

25. 26. 27. 28. 29. 30. 31.

32. 33 34.

35 36. 37.

38. 39. 40. 41.

INDEX

A

Bacterial transition-metal coenzymes, 297, 320 Benz[a]anthracene, 145, 153, 156–157 Benzene, 88, 90, 122, 138, 143, 353 Benzo[a]pyrene, 145, 153–154, 156 Benzo[b]flouranthene, 145, 153, 156 Benzo[g,h,i]perylene, 145, 153, 156 Benzoic acid, 16, 21–22 Benzyl alcohol, 87, 90–92 Binding sites, 124, 173, 197, 199, 201, 204–205, 210, 215–216, 247, 251– 252, 255, 257, 268–269 strongest, 251 weaker, 251, 257 Bioavailability, 116, 139, 161, 243, 275, 283–284, 328, 338, 344, 346, 349 Bioconcentration factor (BCF), 338 Biodegradation, 40, 51, 161 Biofilms, 287, 289 Biogenic origin, 290 Biogeochemicals, 349 Biomarkers, 291 bis-(4-fluorophenyl)methane, 121–122, 132 Blue dextran, 5 Bound residues, 51, 114, 153, 161–163, 167–171

111In,

3–6 131I, 3–6 1-Aminoanthraquinone, 329–330, 334– 337, 340, 342 1-Chloro-2,4-dinitrobenzene (CDNB), 329–330, 339, 353 2,2,2-Trifluoroethyl laurate, 121–122 2,4-Dichlorophenoxyacetic acid (2,4-D), 183–192 4-Chloro-2-methylphenoxyacetic acid, 183 Acenaphthene, 145, 155–156 Acenaphthylene, 144, 155–156 Adsorption, 9, 98–99, 119–120, 126– 127, 131–132, 134–135, 138, 159, 161, 172–174, 176, 180–181, 185, 192, 231, 244, 254, 258– 262, 321, 327, 357 sites, 120, 127, 131–132 Agriculture, 61, 161–162, 173, 179, 180, 183, 191–192, 231, 232, 241, 263, 288, 289 Aldrich, 32, 54, 67, 69, 91, 121, 144, 174, 278, 283, 298 Alga Ankistrodesmus bibraianus, 339 Alkaline phosphatase, 352 Aluminum, 3, 6, 72, 174–175, 184–190, 219–228, 233–234, 237– 239, 299 Americium, 282, 288, 292–293 Am3+, 276 Ammonium acetate, 13, 32–33, 35 Ammonium carbonate, 13, 21 Anaerobic respiration, 297, 319 Anoxic environments, 297, 319 Anthracene, 140, 142–143, 145, 148, 152–157, 159, 174, 181, 327, 340, 344 Anthraquinone, 145, 156, 335, 346–347 Aquatic humic substances, 9, 37–38, 50, 227–228, 240, 262, 287, 289–292, 295–296, 345, 360 Aquifer, 115, 288–289, 291–292, 295– 296 Aromatic, 4, 30, 47, 63–64, 68, 70–71, 92, 119, 134–139, 150, 153, 157– 159, 180–181, 224, 234, 282, 337, 340, 342–344, 346 aromatic hydrocarbons, 119, 135–138, 158–159, 181, 344, 346 extractability, 104, 111, 114, 116 polycyclic aromatic compounds (PAC), 139–140, 142–153, 155–157 polycyclic aromatic hydrocarbons (PAH), 78, 101–104, 111, 114, 116, 119–121, 123–130, 133–135, 138, 147–150, 152, 155, 157–159, 181, 327, 344, 346 uptake, 129, 134 Atrazine, 88, 136, 161–172 residues, 163, 171 Autocorrelation function, 54, 56, 58

C Caenorhabditis elegans, 338, 349, 360 Caffeic acid, 330, 338, 344, 350–351, 353, 359 Capacity coefficient, 124, 142, 146, 148 Capillary column, 164 Capillary electrophoresis (CE), 9–15, 22–30, 59, 87, 98–99, 135, 158, 175, 216, 229, 322–324, 345 CE-ESI/MS, 10–12, 22–27 Capillary gel electrophoresis, 11, 29 Capillary zone electrophoresis (CZE), 9, 11, 13–14, 21–24, 27–30, 240 Carbazole, 145, 156 Carbon cycling, 61 Carbonate, 13, 15, 21, 276–277, 285 Carbonato-metal-humate complexes, 277 Carboxypeptidase A, 340 Catechol, 15, 185–186, 190–192, 343, 348 Cellular contents, 10 Ceratophyllum demersum, 328–332, 334–335, 337–341, 345, 349, 357, 360 Charge transfer complexes, 148, 247, 254, 327, 342 Chrysene, 145, 153, 156 Citrate, 186–191, 297–299, 301, 307– 308, 310, 313, 316, 319–320 Citric acid, 54–55, 57–59, 286 Clays, 103–104, 107, 173–181, 231, 282, 286, 291, 296

B Bacteria, 10, 315, 319–320 233

234

| INDEX

Co(II), 243–247, 251–255, 257–258, 261 Coal, 68, 263, 345, 348 Coaxial sheath liquid interface, 12 Colloids, 3, 9, 15, 29, 32, 38, 50, 116, 173, 184, 217, 230–231, 241– 242, 278, 284, 287–288, 291–296 aggregates, 11, 173, 180, 229, 231, 241–242, 279, 284, 287, 327 Conductimetry, 266, 268, 270–271 Conformational nature, 3, 85, 92, 98 Contaminants, 61, 67, 69–70, 76, 99, 116, 137, 140, 148, 151–153, 180, 297 Contaminated sites, 139, 152, 157 Cooperative sorption, 86 Cr(II), 306, 312 Crops, 231, 236, 238–239, 241 Cu, 231, 233–234, 237–239, 255, 266, 297, 343 Cycloadditions, 153, 159 Cyprinus carpio, 350 D Daphnia magna, 346, 349–350, 354, 356 Daphnids, 339, 351–358 Databases, 61–62, 70–71, 76, 81, 261, 316 DDT, 327 Defence mechanism, 339 Desorption, 61, 67, 79, 115, 135–138, 142, 183–185, 187, 189–191, 292, 321 Detection limits, 3, 7, 10, 171 Detoxication enzymes, 328, 339, 345, 356–358, 360 Diagenesis, 61, 120, 136–137 Dialysis, 4, 145, 243–247, 251–255, 258–260, 298 Dibenz[a]anthracene, 145, 156 Diethyl phthalate, 69, 80 Differential pulse anodic stripping voltametry, 205 Differential scanning calorimetry (DSC), 64–65, 67, 70–72, 74–75, 79, 81, 114–115 Differential thermal analysis, 64 Diffuse reflectance Fourier transform infrared spectroscopy (DRIFTS), 232, 238–239 Diffusion, 32, 54, 68, 85, 120, 125–126, 131, 134, 189–191, 230– 231, 241, 244, 248, 253–254, 258–259, 263– 267, 271–273, 287 coefficient, 32, 54, 189, 230, 248, 253, 264–265, 27 controlled, 189, 263 limitation, 125 flux, 265–267, 271–272 Dilatometry, 64, 70 Dioxins, 138, 297, 320, 344, 347 Disrupted contacts, 86, 98 Dissolved inorganic carbon, 289–290, 292 Dissolved organic carbon (DOC), 141, 288–289, 291, 321, 327, 329–330, 334, 338, 342, 346, 359–360 Dissolved organic matter (DOM), 80, 115, 243, 297, 300, 319, 320, 327– 336, 338, 340, 344–346, 349, 359– 360 Distribution ratio, 124–125, 129–133 DNA, 70, 81, 350, 357 Double labeling, 5, 7 Dy3+, 276 Dynamic light scattering, 53–54, 58–59 E Ecosystems, 183, 338, 343

Electromassograms, 11, 18–21, 26 Electron acceptors, 297, 319, 321, 342– 343, 348 donors, 297 mediators, 297, 315, 318–319 Electron paramagnetic resonance (EPR, ESR), 62, 80, 316, 319, 329, 337, 342, 347, 348 Electron transfer mediators, 297, 300 quinones, 68, 116, 157, 297, 319, 331– 335, 339–340, 342 Electrophoretic separation, 9 Electrospray ionization, 9–12, 21–28, 31, 39–47, 49–51, 280–281 ESI-MS, 10, 12, 31, 40 ES-MS, 280–281, 284 Elemental analysis, 69, 122, 141, 233, 281 Endotherm, 65 Enthalpy, 64, 66, 71, 148–149 Enthalpic overshoot 72, 75 stability 150 Entropy 66, 148–149 stability 150 Environment, 7, 9, 31, 38, 61–62, 77, 139, 158, 172–173, 180, 183– 184, 191–192, 197, 200, 203, 219–220, 224–226, 228–229, 231, 239, 255, 261, 263, 273, 275, 280, 282, 284, 287, 327, 338, 346– 347, 360–361 Ethanes, 298, 300, 302–304, 310–313, 316–318, 320–321, 323 Ethenes, 298, 300, 302, 304, 308, 310– 311, 313, 315, 317–318 Ethynes, 298, 300, 302, 304, 306, 310 Eu3+, 277 Eulimnogammarus cyaneus, 350, 354– 355 EXAFS, 6–7, 284 F Fe, 6, 143–144, 153–157, 184–188, 190, 231, 233–234, 237–239, 262, 297, 307–310, 312, 318, 320, 322, 343 Fe(III), 143–144, 153–157, 262, 320, 343 Fertilizer applications, 238 Field-flow fractionation, 31–32, 38, 229– 230, 232, 240–242 asymmetrical, 31–32, 38, 242 flow, 231–233 FI-ESI/MS, 11 Fission products, 260, 275, 280 Flame ionization detector, 299 Flouranthene, 145, 153, 155–156 Flourene, 140, 145, 156 Flow-injection mode, 11 Fluka, 32, 142, 174, 244, 253, 298 Fluoranthene, 121, 123–128, 130 Fluorescence, 3, 62, 104, 227, 241, 243– 246, 248–250, 255–262, 276, 284– 285 quenching, 3, 243–244, 248–249, 255–262 time-resolved laser-induced fluorescence (TRLIF), 276–278, 280, 284–285 Fluorimetry, 243–244, 255 Fourier transform, 39, 50–51, 62, 77, 232, 242 Fourier transform—ion cyclotron resonance, 39 Fractogram, 35–36, 230, 233, 235–239 Free flow electrophoresis, 11, 23, 27– 28 FTIR, 62–63, 69, 72, 232 Fulvic acids (FAs), 3–7, 9, 31, 33, 35– 37, 39–40, 44, 46–47, 49– 50, 61, 78, 163, 226–227, 229–231–239, 241–242, 252, 262, 276, 281–282, 284, 286, 288–291, 293, 295, 298, 327, 331, 335, 342

INDEX |

Functional groups, 4, 9, 63–64, 68–70, 85, 92, 96, 98, 106, 156, 199, 201, 217, 225, 229, 232, 234, 246, 271, 280, 282, 294, 327, 339, 349, 358 acidic, 197, 199, 202–203, 205, 214–215 carboxylic, 63–64, 68, 147, 205– 206, 208–209, 214–216, 224, 229, 238–239, 244, 265, 327 carboxylic sites, 239 double bonds, 358 electrophilic, 350, 358 epoxides, 358 ether, 358 methoxyl, 63 quinone, 153, 297, 321, 329, 334, 340, 342, 348, 358–359 G Gammarus ischnus, 350, 355 Gammarus pulex, 338 Gammarus tigrinus, 350–351, 354 Ge-detector, 5 Gel, 5, 11, 29, 102, 104, 111, 141, 217, 253, 263–267, 271–273 GelTreat, 13 Geochemical barriers, 3 Gibbs free energy, 71 Glass transition, 64–67, 70, 72–73, 75– 76, 78–79, 81, 115, 136 Glassy domains, 63, 173 Globular proteins, 32 Glutathione S-transferase, 339 Groundwater, 103, 115, 136, 261, 287– 296 Guajacol peroxidase, 331–333, 340 H Heat capacity, 64–67, 71–72, 75–76, 81 Heat shock protein (Hsp 70), 350, 352, 354–358 Heavy metals 243, 260, 349–350 Hematin, 297 Henry’s constant, 87 Heteronuclear single quantum coherence NMR, 63 High performance size exclusion chromatography (HPSEC) 3, 5–7, 31, 38, 53, 59, 229–230, 240 Homologous series, 39–40, 42, 44–49 HPLC, 72, 104, 140–142, 145–146, 148, 156–157, 159, 232, 240, 298 Human health, 7, 139, 360 Human skin, 338 Humic acids (HAs), 4–6, 16, 31, 38, 54, 59, 61–64, 67–69, 70–75, 77–80, 91, 120, 125–126, 131, 136, 137, 159, 174–175, 177, 180– 181, 183–187, 189–190, 192, 217, 218, 227, 230–231, 238, 240– 248, 251– 254, 260–263, 266, 273, 276–279, 282–283, 285, 288, 296, 298, 320, 327, 330, 331, 338, 344, 346–348, 350, 359, 361 uptake by organisms, 339 Humification, 139–140, 148, 157, 159, 162, 186–187, 190–192, 349 Humin, 61–62, 120, 131, 135, 158, 161– 163, 165–167, 169, 172, 241 Hydration, 85–87, 90–92, 96, 98–99, 102, 105, 107–109, 111, 114, 116 effect, 85, 90–91, 98–99 Hydrilla verticillata, 340, 347 Hydrocarbons, 51, 86–88, 91–92, 119, 135–138, 158–159, 181, 296–297, 299–301, 313, 304–316, 318–321, 323, 344, 346, 359, 362

235

chlorinated, 297, 301, 315–316, 318– 319, 321 Hydrodynamic diameter, 230, 235–237 radius, 28, 54–58, 203 Hydrogen bonding, 64, 238, 327 deficiency, 47 peroxide, 340, 350, 357, 360 Hydrogenolysis, 304–309, 311–313 Hydrophobic organic compounds, 101, 114, 119, 137, 173, 180–181 Hydroquinone, 334, 350–351, 355 IHSS, 4, 32, 38, 40, 67, 77, 79, 86, 99, 116, 121, 184, 192, 244, 329, 330, 342, 350–351, 355 I Indenopyrene 145, 156 Inductively coupled plasma 229–232, 240–241 Inductively coupled plasma atomic emission spectrometry (ICPAES), 233–234, 244 Inductively coupled plasma-mass spectrometry (ICP-MS), 3, 7, 229–234, 236–242 Inorganic contaminants, 173, 297 in-situ generation, 288–290, 295 Iodine, 37, 275, 280–282, 286 Iodogen method, 4 Ion cyclotron resonance, 39, 50–51 Ionic strength, 9–10, 28, 40, 54–56, 58, 142, 199, 238, 246, 254, 261, 278 Ionization modes, 42, 45 Ion-trap MS, 11 Isotherm, 64, 85–88, 89, 90–96, 98, 119– 120, 124, 126–128, 130– 134, 173, 175–179, 185, 278 nonlinearity, 120, 124, 130–132, 134 parameters, 132, 176–178 K Kendrick mass defect (KMD), 39, 42, 47–48 Kinetics, 7, 85–86, 93, 101–102, 109, 113–115, 121, 123, 125–126, 132, 136, 152, 157, 183–184, 189, 192, 266, 268–269, 288, 292– 296, 300– 301, 309, 312, 320–322, 324, 332– 333 first-order complexation, 265, 271 pseudo-first-order, 300–302, 316–317 reaction rate, 191, 265, 270, 282, 298, 308, 316, 318 L Lanthanides, 275–276, 279–280 Leonardite, 72–75, 77, 244, 261–262 Lignite, 77, 263, 266 Lindane, 90 Linear sorption, 87, 119–120, 127, 134, 179 Link solvation model, 92, 99 Lipid, 119–123, 126–127, 130–135, 137, 350, 357 extraction, 122, 130–134 fractions, 119, 120, 130 Liposomes, 14 Liquid scintillation counting, 123, 175, 186, 330, 351 Luminescence spectroscopy, 276, 285 M Macromolecular, 31, 53–54, 59, 64–65, 70–71, 77, 79, 101, 111– 112, 115, 132, 195–196, 199, 203, 215–217, 230 mobility, 64, 70–71

236

| INDEX

structure, 101 Macromolecules, 9, 11, 13, 32, 53, 59, 68–71, 80, 85, 92, 98–99, 137, 217, 230–231, 240, 244, 248 Marsh, 289 Mass distributions, 17–18, 20, 31, 44, 49, 231, 235, 237, 239 Mass spacings, 46, 49 Mass spectrometry, 9–12, 27–28, 31–32, 37–39, 41, 50–51, 68, 77, 80, 162, 229–232, 240–241, 280, 286 Matrix potential, 104, 110, 114 Membrane accumulation, 350 permeability, 339, 357 Mercury, 206, 244–245, 248, 254, 259, 262, 320, 352–353, 358, 361 Metal binding, 3, 69, 80, 199–205, 210, 215–220, 229, 240, 248, 257, 345 migration, 3 oxide-humic complexes, 183–192 oxides, 183–192, 260 sequestering, 238 Methanes, 298, 301–302, 304, 314, 316–318, 320, 322–323 Microbiological, 289–291, 294, 321 Micro-electrospray ionization, 11 Micropores, 64, 120, 126, 134 Mineral, 63, 69, 76, 80, 85–86, 98, 102– 104, 107–108, 110–112, 114, 119, 126, 135, 136, 163, 168–169, 174, 176–181, 184, 192, 231, 234, 243, 260, 275, 282–283, 286–288, 294, 344 goethite, 180, 231, 260 hematite, 282–283, 286 humic complexes, 175–178 iron oxides, 174, 282, 286 kaolinite, 174–179, 181 montmorillonite, 174–178 silica, 12, 104, 141, 146, 164, 260, 278, 282–283, 286 surface, 174, 177 Mineralization, 161, 172, 295, 319 Mn, 5, 180, 184–190, 237, 240, 343, 346 Mn(II), 343 m-Nitrophenol, 87, 90–92, 96 Mobility, 11–18, 21–22, 26–27, 63–65, 67–68, 70–71, 75, 78, 102, 112, 114, 132–133, 219–220, 229, 239, 243, 269, 275, 279, 283, 286–288, 291–292, 295 contour plot, 21 distribution profile, 16 Models, 11, 24, 21, 56–57, 59, 61–64, 67–71, 76, 78–81, 85–86, 89– 90, 92, 94–99, 113, 119, 120, 124–129, 132, 133–136, 173, 179– 181, 185, 192, 195–196, 198–199, 201, 203, 216–217, 224–226, 228, 232–233, 241–242, 245, 248–250, 255–259, 261–263, 276, 279–280, 285, 295, 321, 323–324, 339, 344, 360 conservative roof, 276, 285 FITEQL, 250, 257, 261 GEOCHEM 50–51, 59, 77, 79, 98, 135, 172, 180, 225, 228, 260–262, 283–286, 295 log K spectrum, 248–250, 255–257 MINTEQ, 225–226 neutralization charge, 276 polyelectrolyte, 195, 276 sorption, 128 Moisture, 98, 101, 116–117, 135 Molar mass, 5–7, 18, 21–23, 26, 31, 40, 44, 46, 49, 53–55, 58–59, 69, 71, 148, 155, 169, 229–231, 235–239, 253, 259, 297–298, 327–328, 330, 336, 338, 340, 342, 350–351 Molecular

contraction, 215 Dragunov, 68,80 dynamics simulations, 67, 70, 79, 81 expansion, 215–216 fingerprint, 40 Flaig, 68, 80, 345 formulas, 39–40, 47–48 Fuchs, 68 Leenheer, 69, 80 modeling, 67, 69–70, 80, 284 Steelink, 68–69, 80, 181, 240, 347– 348 shape, 3, 16, 34–35, 40, 55, 58–59, 73, 95, 99, 218, 224, 254 simulation, 62, 68–70, 79–80 site distribution, 94 size, 3, 6, 31, 148–149, 217, 238, 262, 357 structures, 38, 349 Multi-angle light scattering, 53 N n-Alkyl monoesters, 120 Nanogel, 11 Nanoparticles, 10 Naphthalene, 128, 140, 144, 149, 156 Natural organic matter (NOM), 6, 9–11, 13–14, 16, 21–28, 30, 50, 59, 61– 64, 67, 69–71, 73, 76–77, 79, 85– 94, 96–99, 115, 136, 139, 180–181, 217, 291, 320–321, 330–331, 334– 337, 339–340, 342–343, 347, 350– 360 Negative ionization, 21, 42 NH4Ac, 32–33 Ni(II), 266, 293 Nitrate, 186–191, 245 Nuclear magnetic resonance (NMR), 6– 7, 62–64, 69–70, 77–80, 115, 102– 105, 111–116, 122, 133, 138, 219– 228, 241 27Al NMR, 6–7, 220–222, 227–228 13C NMR, 62, 69, 227 2-D correlation spectroscopy, 63 19F-nuclear magnetic resonance, 122 1H-NMR, 102–105, 111–116 exponential decay functions, 104, 108 inversion algorithm, 104 relaxation, 102–105, 111–115 NOESY, 63, 78 relaxation time, 67, 102, 104–109, 112–113 surface relaxivity, 102, 106 time constants, 104, 108–111, 113– 114 Nominal molar mass, 5–7 Non-reactive buffer, 13 Nordic reference, 350–352, 354–355, 357, 359 Nuclear activities, 275 waste, 284, 291 Nutrients, 231, 239, 328 O Offord’s equation, 13, 16 Organic carbon, 119, 122, 124–125, 128–131, 133–134, 146, 162, 174, 176–177, 181, 244, 288–289, 291, 295, 299, 327, 330, 344, 346, 359– 360 film, 282 surface contaminants, 34 o-Terphenyl, 144–145

INDEX |

Oxalic acid, 54–55, 57–59, 220–221, 223–225 P Partitioning, 97, 119–120, 124–125, 127, 129–130, 132–136, 173– 174, 300, 327 Pb, 231, 234, 237, 239, 322 p-Benzohydroquinone, 334 PCB, 327 Peak-average electrophoretic mobility, 12 Peats, 63, 67, 69, 71, 77–79, 86–88, 91, 101–104, 108–110, 112– 114, 116, 120, 135, 227, 242, 253, 345, 348, 361 Pentachlorophenol, 64 Peroxidase, 329, 331–333, 340, 345, 347, 350, 355 Pesticides, 51, 69, 102, 135, 161–162, 183–184, 187, 191–192, 344 Phenanthrene, 64, 67, 120–121, 123– 134, 137, 140, 145, 148, 154– 156, 158, 173–181 Phenol, 87–88, 90, 92, 96, 205–206, 214 Photodegradation, 338, 340 Photon correlation spectrometer, 54 Photosynthesis, 328, 342–343, 345 oxygen production, 328–329, 334–337, 341–343 photosystem I, 342 photosystem II, 342, 347 Phthalic acid, 21–22 Plasticizers, 132–133 Points of zero salt effect (PZSE), 185– 186, 188, 190 Polar contact, 86, 90–91 Polarography, 243–247, 252–254, 258– 262 Polyelectrolyte, 9, 85, 195–196, 276 Polynuclear, 136–137, 158, 279, 294 Pore size, 102, 107, 111–114 Positron annihilation lifetime spectroscopy, 64 Potentiometry, 243, 261, 266, 268, 270–271 Pronase B, 340 Proton binding, 197, 199–204, 206, 210, 214–217 Py/GC/C-SIRMS, 162, 164–168, 170– 171 Py/GC/MS, 162–165, 167–168, 171 Pyrene, 64, 110–111, 140, 145, 148, 153–156, 174, 181, 327, 338, 344 Pyridine, 87–92, 94–97, 322 Pyrolysis, 62, 68, 77, 80, 162–172 Q Q-TOF mass spectrometers, 39, 280 R Radicals, 96, 153, 155, 306, 313, 323, 337, 340, 342, 348, 351, 355, 357 hydroxyl, 340, 357 semiquinone, 337, 340 Radioactivity, 3, 5, 7, 175, 202, 332, 334, 336, 338, 340, 342, 344, 346, 348 Radioactive tracers 3, 202, 332, 334, 336, 338, 340, 342, 344, 346, 348 Radiochromatogram, 3, 5–6 Radionuclides, 243, 275, 280, 282, 284–285, 287, 295–296 Reductive dehalogenation, 297, 300, 306, 308, 313, 315, 316, 318, 320– 321, 324 Refractory colloids, 3 Reverse osmosis, 347, 350–351, 355

237

ROS-enzymes, 328 Rubbery domains, 64, 173 S Salvinia minima, 338 Sample contamination, 37 Secondary ion yields, 34–35 Sediments, 50, 77–79, 85, 119, 121–122, 124, 129–130, 133, 135– 138, 158, 179, 181, 231, 287, 296, 319 Selected ion monitoring (SIMS), 300 Self-assembly, 31 Semolina, 103–107, 111–113 Separation buffer, 11, 13, 22 Silicon substrates, 33–35 Silver, 153, 352–353, 358, 361 Site-specific interactions, 120, 131, 134 Size distribution, 11, 13, 37, 49, 54–58, 102, 107, 112–114, 231, 252, 278 Size exclusion chromatography, 3, 5, 29, 31, 59, 180, 229–230, 240, 252–253 Sodiated, 40, 42, 46, 48–49 Sodium polystyrene sulfonates, 5 Soils, 3, 61, 64, 76–80, 85, 98, 102–103, 107, 110, 116, 119–120, 124, 129, 132, 135–136, 158–159, 161–162, 164–165, 167–174, 179–181, 183– 184, 192, 231, 238–243, 261, 296, 348 ashed, 103, 105, 110, 112 fractions, 119–120, 131, 133 layers, 101, 114 organic matter, 3, 70, 76–80, 86, 99, 101–102, 110, 113–116, 119–121, 123, 125–127, 129–136, 138, 172– 173, 176, 180– 181, 231, 320, 347, 359 Solubility, 81, 87, 99, 119, 123–125, 129–131, 133, 137, 175, 275– 276, 279, 282, 350, 358 Soluble glutathione S-transferase, 329– 331, 350, 356 Solvent-assisted sorption, 86, 89, 92, 94, 96 Sorbate interactions, 85, 99 Sorption affinity, 120, 123, 128–132, 134 coefficients, 145, 147–148, 151, 176 Freundlich, 89, 94–95, 97, 124, 126– 127, 129, 132, 137, 175– 178, 181, 189 isotherm, 85–86, 91, 93–95, 98, 130, 134, 176 kinetics, 93, 115, 121, 123, 125, 132, 152 Langmuir, 92–94, 99, 124, 173, 210– 211 mechanisms, 145 properties, 131, 140, 145, 147–148, 150 sites, 86–88, 90–92, 94–95, 97, 124, 130–131, 134, 156, 288 Speciation, 85, 225–228, 241, 243–244, 256, 258, 275–277, 279– 280, 282– 286, 296 Spin contents, 330 SPME, 122, 140, 142–143, 145, 151– 152, 157 Stability constants, 216, 246–247, 251– 252, 254 Starch, 103–107, 111–113 Stern-Volmer equation, 248–249, 257 Sterols, 120 Stokes-Einstein equation, 54, 231, 253 Sulfur species, 297, 319 Superoxide anion, 340 Superoxide dismutase, 340 Supramolecular aggregates, 31, 37–38, 53–54, 59, 195, 217, 281 Surface area, 73, 102, 132, 134, 180, 185–186, 192 Suwannee River fulvic acid, 31–32, 40, 69, 71, 241

238

| INDEX

Swelling, 85, 101–116 mechanism, 112 T Th, 275–280, 282–286 Thermal analysis, 62, 64, 71, 73, 75–76, 79, 81 expansion, 64–67, 75–76, 79 Thermodynamic properties, 61–62, 64, 67, 69–71, 75–76, 79, 148 Thermogravimetry, 64 Ti, 231, 239 Ti(III), 297–299, 301, 307–308, 310, 313, 316, 319 Time-of-flight, 31–32, 37–38, 50, 286 TMA, 64, 66, 71–75, 352 TMDSC, 64, 71–72, 74–75 TOF-SIMS, 31–33, 35 Total ion current, 21–23, 165 Toxicity, 119, 161, 219–220, 275, 284, 344, 346, 349–350, 354, 357, 361 Trace elements, 231, 239, 241, 275, 280, 282–283, 229, 231, 236, 238, 241 Trace metals, 9, 229–231, 233, 236, 238–241, 243, 245, 260, 275, 276, 285, 288, 292–295 Transition metals, 267, 269, 297–298, 313–314, 316, 318–319, 321– 323 Transport phenomena, 101, 114, 263 Trichloroethane, 298, 303, 323 Trichloroethylene, 79, 88, 90–91, 99, 320, 323–324 Trimellitic acid, 21–22 Triphenylbenzene, 144–145 Triticum aestivum, 340, 347 Two-dimensional NMR, 63 U Ultrafiltration, 21, 77, 141, 205 Uranium, 275, 277–279, 285–286 UV, 3, 5–6, 13–14, 21–22, 26, 32, 56, 62, 142, 230, 232–233, 235, 238, 266, 278, 330, 342. V van der Waals, 68, 238 Vesicularia dubyana, 328–329, 331, 333–335, 337–338 Vinyl chloride 298, 302, 308, 320 Viruses, 10 Vitamin B12, 297, 307–308, 320–321 W Water uptake, 104, 110–111, 114–115 Wetland, 116, 288–290, 292, 321 X Xenobiotic, 9, 80, 172, 340–341, 343, 350, 358 Z Zn, 204–206, 210, 212, 231, 233–234, 237, 239, 307–308, 310, 321