Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems (Wiley Series Sponsored by IUPAC in Biophysico-Chemical Processes in Environmental Systems)

BIOPHYSICO-CHEMICAL PROCESSES OF ANTHROPOGENIC ORGANIC COMPOUNDS IN ENVIRONMENTAL SYSTEMS

Wiley-IUPAC Series in Biophysico-Chemical Processes In Environmental Systems

The Division of Chemistry and the Environment of the International Union of Pure and Applied Chemistry (IUPAC) is sponsoring this series which addresses the fundamentals of physical-chemical-biological interfacial interactions in the environment and the impacts on: (1) the transformation, transport, and fate of nutrients and pollutants, (2) food chain contamination and food quality and safety, and (3) ecosystem health, including human health. In contrast to classical books that focus largely on separate physical, chemical, and biological processes, this book series is unique in integrating the frontiers of knowledge of both fundamentals and impacts of interfacial interactions of these processes in the global environment. Books in the series: Biophysico-Chemical Processes of Heavy Metals and Metalloids in Soil Environments, edited by Antonio Violante, Pan Ming Huang, and Geoffrey M. Gadd Biophysico-Chemical Processes Involving Natural Nonliving Organic Matter in Environmental Systems, edited by Nicola Senesi, Baoshan Xing, and Pan Ming Huang Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems, edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang

BIOPHYSICO-CHEMICAL PROCESSES OF ANTHROPOGENIC ORGANIC COMPOUNDS IN ENVIRONMENTAL SYSTEMS

Edited by

BAOSHAN XING NICOLA SENESI PAN MING HUANG

Copyright Ó 2011 by John Wiley & Sons, Inc. All rights reserved Published by John Wiley & Sons, Inc., Hoboken, New Jersey Published simultaneously in Canada No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com. Library of Congress Cataloging-in-Publication Data: Biophysico-chemical processes of anthropogenic organic compounds in environmental systems / edited by Baoshan Xing, Nicola Senesi, Pan Ming Huang. p. cm. Includes index. ISBN 978-0-470-53963-7 (cloth) 1. Environmental chemistry. 2. Bioorganic chemistry. 3. Anthropogenic soils. I. Xing, Baoshan. II. Senesi, N. (Nicola) III. Huang, P. M. TD193.B5475 2011 628.5—dc22 2010033320 Printed in Singapore oBook ISBN: 978-0-470-94447-9 ePDF ISBN: 978-0-470-94446-2 ePub ISBN: 978-1-118-00211-7 10 9 8 7 6 5 4 3 2 1

CONTENTS

SERIES PREFACE

vii

PREFACE

ix

CONTRIBUTORS

xi

PART I FUNDAMENTAL BIOPHYSICO-CHEMICAL PROCESSES OF ANTHROPOGENIC ORGANIC COMPOUNDS IN THE ENVIRONMENT

1

1

Interactions of Anthropogenic Organic Chemicals with Natural Organic Matter and Black Carbon in Environmental Particles

3

Joseph J. Pignatello

2

Comprehensive Study of Organic Contaminant Adsorption by Clays: Methodologies, Mechanisms, and Environmental Implications

51

Stephen A. Boyd, Cliff T. Johnston, David A. Laird, Brian J. Teppen, and Hui Li

3

The Role of Organic Matter–Mineral Interactions in the Sorption of Organic Contaminants

73

Myrna J. Simpson and Andre J. Simpson

4

Photocatalytic Degradation of Organic Contaminants on Mineral Surfaces

91

Chuncheng Chen, Zhaohui Wang, Wanhong Ma, Hongwei Ji, and Jincai Zhao

PART II ANTHROPOGENIC ORGANIC COMPOUNDS IN AIR, WATER, AND SOIL, AND THEIR GLOBAL CYCLING 5

Sorption of Anthropogenic Organic Compounds to Airborne Particles

113 115

Hans Peter H. Arp and Kai-Uwe Goss

6

Measurement and Modeling of Semivolatile Organic Compounds in Local Atmospheres

149

Songyan Du and Lisa A. Rodenburg

7

Pharmaceuticals and Personal Care Products in Soils and Sediments

185

Bo Pan and Baoshan Xing

v

vi

8

CONTENTS

Fate and Transport of Organic Compounds in (to) the Subsurface Environment

215

Peter Grathwohl

9

Pharmaceuticals and Endocrine-Disrupting Compounds in Drinking Water

233

Daniel W. Gerrity, Mark J. Benotti, David A. Reckhow, and Shane A. Snyder

10

Intermedia Transfers and Global Cycling of Persistent Organic Pollutants

251

Claudia Moeckel and Kevin C. Jones

11

Emission of Polycyclic Aromatic Hydrocarbons in China

267

Shu Tao, Bengang Li, Yanxu X. Zhang, and Huishi Yuan

PART III ANALYTICAL TECHNIQUES 12

Principles and Guidelines of Sampling, Extraction, and Instrumental Analysis Techniques for Measurements of Organic Pollutants in Environmental Matrices

283

285

Eddy Yongping Zeng, Zhaohui Wang, and O. Samuel Sojinu

13

NMR Application in Environmental Research on Anthropogenic Organic Compounds

315

Robert L. Cook

14

Synchrotron-Based X-Ray and FTIR Absorption Spectromicroscopies of Organic Contaminants in the Environment

341

John R. Lawrence and Adam P. Hitchcock

15

Application of Solid-Phase Microextraction in Determination of Organic Compounds from Complex Environmental Matrices

369

Sanja Risticevic, Dajana Vuckovic, and Janusz Pawliszyn

16

Application of Biosensors for Environmental Analysis

413

Marinella Farre´, Sandra Pe´rez, Lina Kantiani, and Dami a Barcelo´

17

Analyses of Drugs and Pharmaceuticals in the Environment

439

Imran Ali, Hassan Y. Aboul-Enein, and Klaus Kummerer

PART IV RESTORATION OF NATURAL ENVIRONMENTS CONTAMINATED BY ORGANIC POLLUTANTS 18

Biochemistry of Environmental Contaminant Transformation: Nonylphenolic Compounds and Hexachlorocyclohexanes–Two Case Studies

463

465

Hans-Peter E. Kohler

19

Biodegradation of Anthropogenic Organic Compounds in Natural Environments

483

Jose-Luis Niqui-Arroyo, Marisa Bueno-Montes, and Jose-Julio Ortega-Calvo

20

Phytoremediation of Soils Contaminated with Organic Pollutants

503

Jason C. White and Lee A. Newman

21

Bioavailability of Hydrophobic Organic Contaminants in Soils and Sediments

517

Wesley H. Hunter, Jay Gan, and Rai S. Kookana

22

Abiotic and Biotic Factors Affecting the Fate of Organic Pollutants in Soils and Sediments

535

Richard E. Meggo and Jerald L. Schnoor

INDEX

559

SERIES PREFACE

Scientific progress is based ultimately on unification rather than fragmentation of knowledge. Environmental science is the fusion of physical and life sciences. Physical, chemical, and biological processes in the environment are not independent but rather interactive processes. Therefore, it is essential to address physical, chemical, and biological interfacial interactions in order to understand the composition, complexity, and dynamics of ecosystems. Keeping separate these domains, no matter how fruitful, one cannot hope to deliver on the full promise of modern environmental science. The time is upon us to recognize that the new frontier in environmental science is the interface, wherever it remains unexplored. The Division of Chemistry and the Environment of the International Union of Pure and Applied Chemistry (IUPAC) has approved the creation of an IUPAC-sponsored book series entitled Biophysico-Chemical Processes in Environmental Systems published by John Wiley & Sons, Hoboken, New Jersey. This series addresses the fundamentals of physical--chemical--biological interfacial interactions in the environment and the impacts on (1) the transformation, transport, and fate of nutrients and pollutants; (2) food chain contamination and food quality and safety; and (3) ecosystem health, including human health. In contrast to classical books that focus largely on separate physical, chemical, and biological processes, this book series is unique in integrating the frontiers of knowledge of both fundamentals and impacts of interfacial interactions of these processes in the global environment. With the rapid developments in environmental physics, chemistry, and biology, it is becoming much harder, if not impossible, for scientists to follow new progress outside their immediate area of research by reading the primary research literature. This book series will capture pertinent research

topics of significant current interest and will present to the environmental science community a distilled and integrated version of new developments in biological physical, and chemical processes in environmental systems. This book, Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems, is volume 3 of the Wiley-IUPAC series. This book comprises 22 chapters by renowned experts in the topic and is unique in integrating both fundamentals and impacts of interfacial interactions of physical, chemical, and biological processes pertaining to adsorption, transformation, bioavailability, toxicity, and transport processes of anthropogenic organic compounds in the air--water--soil environment and their global cycling. Further, the most modern techniques used for sampling, extraction, and instrumental analyses and various means for the restoration of natural environments contaminated by organic pollutants are treated. This book can be used as an advanced reference source on biological, physical, and chemical processes and performance, analytical techniques, and restoration of anthropogenic organic compounds in the global environment for senior undergraduate and graduate students in environmental sciences and engineering. It is an essential reference for chemists and biologists studying environmental systems, as well as for geochemists, environmental engineers; and soil, water, and atmosphere scientists. It will serve as a useful resource book for professors, instructors, research scientists, professional consultants, and other individuals working on environmental and ecological systems. P. MING HUANG NICOLA SENESI Series Editors vii

PREFACE

Anthropogenic organic compounds (AOCs) are synthetically made organic chemicals. They range from gasoline components (e.g., benzene, toluene, xylene) to emerging contaminants such as endocrine-disrupting chemicals. Because of their wide use and disposal, AOCs are commonly found in our environment such as the water we drink, the air we breathe, and the soil from which we obtain our food. These compounds are often toxic and can severely deteriorate an ecosystem. They can also bioaccumulate through food chains and cause various diseases (and even death) to organisms, including humans. AOCs behave differently in various environmental media which differ in their different physical, chemical, and biological components and processes. Therefore, an in-depth and more complete understanding of the biological, physical, and chemical processes of AOCs in environmental systems is essential for the development of innovative management strategies for sustaining the environment and ecosystem integrity. Physical, chemical, and biological interfacial interactions and processes govern the fate, transport, availability, exposure, and risk of AOCs. However, the fundamentals of many physicochemical and biological interfacial reactions of AOCs and their impacts on ecosystems remain largely unknown. As a result, predictive models for their fate, transport, and risk in different media are often off target. To advance the frontiers of knowledge on the subject matter would require a concerted and comprehensive effort of scientists in relevant physical and life sciences such as chemistry; mineralogy; geochemistry; microbiology; ecology; environmental engineering; and soil, atmospheric, and aquatic sciences. In addition, physical, chemical, and biological reactions and processes of AOCs in the environment are not independent but rather interactive and closely interrelated. Therefore, it is essential to systematically address

these interactive processes and interactions through an interdisciplinary approach. Scientific progress in advancing the understanding of environmental fate and behavior of AOCs is based ultimately on integration rather than separation of knowledge across scientific disciplines. To achieve the goal of knowledge integration, this book entitled, Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems, brings together world-renowned scientists on the subject matter across scientific disciplines to integrate the current stateof-the-art knowledge, especially the latest discoveries, development, and future prospects on the research of AOCs in the environment. By virtue of the complex nature of the interactions of AOCs with different environmental components and matrixes, no single available technique, instrument, or model is satisfactory yet for determining their fate, transport, availability, and risk in the environment. In order to fully understand the biological, physical, and chemical interactions and processes of AOCs in the environment, it is critical to know the chemical, physical, and biological properties of AOCs and their analytical techniques. This book is unique because of its multidisciplinary approach, which provides a comprehensive and integrated coverage of biological, physical, and chemical reactions and processes of AOCs in various environments, associated sampling and analytical techniques, and restoration of natural environments contaminated by AOCs. There are 22 chapters in this book, and these chapters are divided into four parts. Part I contains four chapters focusing on the fundamental biological, physical, and chemical processes of AOCs in the environment; Part II, with seven chapters, presents the occurrence and distribution of AOCs in air, water, and soil, and their global cycling; Part III, containing six chapters, discusses the state-of-the-art ix

x

PREFACE

sampling methods and current analytical, biological, spectroscopic, and microscopic techniques for monitoring and studying AOCs; and Part IV consists of five chapters emphasizing the restoration of natural environments contaminated by organic pollutants. This book is an informative and important reference book for scientists, engineers, and professionals who are interested in the biological, physical, and chemical processes and interactions of AOCs in environmental systems. This book is also a critical addition to the existing literature on the subject matter. Further, this book can be used by undergraduate and graduate students, instructors and professors in the disciplines of environmental science and engineering; aquatic, soil, marine, and atmospheric sciences; geosciences; and

ecological, biological, and chemical sciences. Again, the book chapter authors are leading authorities in their respective fields of research. Each chapter was rigorously reviewed externally as for refereed journal articles. We sincerely thank all chapter authors and reviewers who graciously volunteered their time and effort, and contributed their knowledge and wisdom to improve the quality and clarity of this book. We are also highly grateful to the staff of IUPAC and John Wiley & Sons for their strong support and great cooperation in the publication of the book. BAOSHAN XING NICOLA SENESI PAN MING HUANG

CONTRIBUTORS

Dr. Hassan Y. Aboul-Enein, Pharmaceutical and Medicinal Chemistry Department, Pharmaceutical and Drug Industries Research Division, National Research Center, Cairo, Egypt Dr. Imran Ali, Department of Chemistry, Jamia Millia Islamia (Central University), New Delhi, India Dr. Hans Peter H. Arp, Norwegian Geotechnical Institute, Department of Environmental Engineering, Oslo, Norway Dr. Dami
a Barcelo´, Department of Environmental Chemistry, IDAEA-CSIC c/Jordi Girona, Barcelona, Spain Dr. Mark J. Benotti, Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, Las Vegas, Nevada Dr. Stephen A. Boyd, Department of Crop and Soil Sciences, Michigan State University, East Lansing, Michigan Dr. Marisa Bueno-Montes, Institute of Natural Resources and Agrobiology of Seville—CSIC, Seville, Spain Dr. Chuncheng Chen, Beijing National Laboratory for Molecular Sciences, Key Laboratory of Photochemistry, Institute of Chemistry, Chinese Academy of Sciences, Beijing, China Dr. Robert L. Cook, Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana Dr. Songyan Du, Department of Environmental Sciences, Rutgers, the State University, New Brunswick, New Jersey Dr. Marinella Farre, Department of Environmental Chemistry, IDAEA-CSIC c/Jordi Girona, Barcelona, Spain

Dr. Jay Gan, Department of Environmental Sciences, University of California, Riverside, California Dr. Daniel W. Gerrity, Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, Las Vegas, Nevada Dr. Kai-Uwe Goss, Department Analytical Environmental Chemistry, Helmholtz-Center for Environmental Research—UFZ, Leipzig, Germany Dr. Peter Grathwohl, Center for Applied Geosciences, University of T€ubingen, T€ubingen, Germany Dr. Adam P. Hitchcock, Brockhouse Institute for Materials Research,McMasterUniversity,Hamilton,Ontario,Canada Dr. Pan Ming Huang (deceased), Department of Soil Science, University of Saskatchewan, Saskatoon, Saskatchewan, Canada Dr. Wesley H. Hunter, Department of Environmental Sciences, University of California, Riverside, California Dr. Hongwei Ji, Beijing National Laboratory for Molecular Sciences, Key Laboratory of Photochemistry, Institute of Chemistry, Chinese Academy of Sciences, Beijing, China Dr. Cliff T. Johnston, Department of Agronomy, Purdue University, West Lafayette, Indiana Dr. Kevin C. Jones, Lancaster Environment Centre, Lancaster University, Lancaster, United Kingdom Lina Kantiani, Department of Environmental Chemistry, IDAEA-CSIC c/Jordi Girona, Barcelona, Spain Dr. Hans-Peter E. Kohler, Department of Environmental Microbiology, Eawag, Swiss Federal Institute of Aquatic Science and Technology, D€ubendorf, ZH, Switzerland xi

xii

CONTRIBUTORS

Dr. Rai S. Kookana, CSIRO Land and Water, Glen Osmond, Australia Dr. Klaus Kummerer, Department of Health Sciences, University Medical Center, Freiburg, Germany Dr. David A. Laird, National Laboratory for Agriculture and the Environment, USDA-ARS, Ames, Iowa Dr. John R. Lawrence, Environment Canada, National Hydrology Research Centre, Saskatoon, Saskatchewan, Canada Dr. Bengang Li, College of Urban and Environmental Sciences, Peking University, Beijing, China Dr. Hui Li, Department of Crop and Soil Sciences, Michigan State University, East Lansing, Michigan Dr. Wanhong Ma, Beijing National Laboratory for Molecular Sciences, Key Laboratory of Photochemistry, Institute of Chemistry, Chinese Academy of Sciences, Beijing, China

Dr. Jerald L. Schnoor, Department of Civil and Environmental Engineering, University of Iowa, Iowa City, Iowa Dr. Nicola Senesi, Department of Biology and Environmental and Agroforestal Chemistry, University of Bari, Bari, Italy O. Samuel Sojinu, Department of Chemical Sciences, Redeemer’s University, Mowe, Ogun State, Nigeria Dr. Myrna J. Simpson, Department of Chemistry, University of Toronto, Toronto, Ontario, Canada Dr. Andre J. Simpson, Department of Chemistry, University of Toronto, Toronto, Ontario, Canada Dr. Shane A. Snyder, Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, Las Vegas, Nevada Dr. Shu Tao, College of Urban and Environmental Sciences, Peking University, Beijing, China

Richard E. Meggo, Department of Civil and Environmental Engineering, University of Iowa, Iowa City, Iowa

Dr. Brian J. Teppen, Department of Crop and Soil Sciences, Michigan State University, East Lansing, Michigan

Dr. Claudia Moeckel, Lancaster Environment Centre, Lancaster University, Lancaster, United Kingdom

Dr. Dajana Vuckovic, Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada

Dr. Lee A. Newman, Biology Department, Brookhaven National Laboratory, Upton, New York

Dr. Ji-Zhong Wang, School of Earth and Space Sciences, University of Science and Technology of China, Hefei, Anhui Province, China

Dr. Jose-Luis Niqui-Arroyo, Institute of Natural Resources and Agrobiology of Seville—CSIC, Seville, Spain Dr. Jose-Julio Ortega-Calvo, Institute of Natural Resources and Agrobiology of Seville—CSIC, Seville, Spain Dr. Bo Pan, College of Environmental Science and Engineering, Kunming University of Science and Technology, Kunming, China Dr. Janusz Pawliszyn, Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada Dr. Sandra Perez, Department of Environmental Chemistry, IDAEA-CSIC c/Jordi Girona, Barcelona, Spain Dr. Joseph. J. Pignatello, Department of Soil and Water, Connecticut Agricultural Experiment Station, New Haven, Connecticut Dr. David A. Reckhow, Department of Civil and Environmental Engineering, University of Massachusetts, Amherst, Massachusetts Sanja Risticevic, Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada Dr. Lisa A. Rodenburg, Department of Environmental Sciences, Rutgers, the State University of New Jersey, New Brunswick, New Jersey

Dr. Zhaohui Wang, Beijing National Laboratory for Molecular Sciences, Key Laboratory of Photochemistry, Institute of Chemistry, Chinese Academy of Sciences, Beijing, China Dr. Jason C. White, Department of Analytical Chemistry, Connecticut Agricultural Experiment Station, New Haven, Connecticut Dr. Baoshan Xing, Department of Plant, Soil, and Insect Sciences, University of Massachusetts, Amherst, Massachusetts Huishi Yuan, College of Urban and Environmental Sciences, Peking University, Beijing, China Dr. Eddy Yongping Zeng, State Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy Science, Guangzhou, China Yanxu X. Zhang, College of Urban and Environmental Sciences, Peking University, Beijing, China Dr. Jincai Zhao, Beijing National Laboratory for Molecular Sciences, Key Laboratory of Photochemistry, Institute of Chemistry, Chinese Academy of Sciences, Beijing, China

Figure 1.14. Equilibrium configurations for the adsorption of methane and water in a carbon slit pore of width 1 nm with varying active site densities at 300 K. See page 27 for text discussion of this figure.

Figure 2.14. Illustration of sorbed dinitro-o-cresol (DNOC) molecule laying flat on the siloxane surface showing the siloxane surface and water molecules surrounding the exchangeable cations. See page 64 for text discussion of this figure.

Figure 2.15. Molecular dynamics snapshot of dinitro-o-cresol (DNOC) in K-smectite (K-SWy-2) clay interlayer. See page 65 for text discussion of this figure.

Figure 5.1. A SEM scan of a particle sample indicating a large porous conglomerate of organic particles with some salts (yellow circle), a sharp-edged, smooth-surfaced salt crystal (blue circle), a pollen grain (red circle) and a smooth-edged, smooth-surfaced mineral particle (green circles). See page 120 for text discussion of this figure.

Figure 5.5. Illustration of nonspecific interactions. See page 126 for text discussion of this figure.

Figure 5.2. Simultaneous particle and air-phase chemical collectors used in sample-and-extract methods. See page 120 for text discussion of this figure.

Figure 5.6. Illustration of specific interactions. See page 127 for text discussion of this figure. Figure 5.7. Examples of adsorptive interactions at surfaces. See page 128 for text discussion of this figure.

Figure 5.8. The influence of relative humidity (RH) on adsorption to minerals and metal oxides, showing the increase of the water layer thickness with increasing RH. See page 131 for text discussion of this figure. Montmorillonite

Kaolinite

Mica

Water and exchangeable cations O

OH

Si, Al

K

Al, Mg, Fe

Figure 7.3. Structures for typical clay minerals. See page 196 for text discussion of this figure.

Oven dried montmorillonite Silica tetrahedra Alumina octahedra Silica tetrahedra Silica tetrahedra ~1.0 nm

Alumina octahedra Silica tetrahedra

Silicon Oxygen Hydrogen Aluminium

Wet montmorillonite Silica tetrahedra

Cations

Alumina octahedra Silica tetrahedra

Water molecules Hydrated exchangeable cations

3 nm or more

Silica tetrahedra

Oxytetracycline

Alumina octahedra Silica tetrahedra

Figure 7.4. Interlayer spacing of montmorillonite at different conditions. See page 197 for text discussion of this figure. 1628 1709

1580

Free CIP at pH 5 1385 CIP on HFO at pH 5 1360

1633

Absorbance

1619

CIP on HAO at pH 5

1354

1640

1627 1358

1800

1600

1400

1200

Wavenumber (cm-1)

Figure 7.5. ATR-FTIR evidence for ciprofloxacin sorption on hydrous oxides of Al (HAO) and Fe (HFO). See page 198 for text discussion of this figure.

Aqueous phase

+ + -

+

Electrostatic attraction + +

-

Adsorbent Cation exchange

+ - +

-

+

+

+ - + +

PPCP molecule Cation

Electrostatic attraction

+

-

+

+

-

+

Exchangeable cation -

-+

Ligand exchange

Negatively charged site

Cation bridging

Figure 7.7. Schematic illustration of adsorption of PPCPs on solid particles in the presence of cations. See page 200 for text discussion of this figure.

Figure 10.1. Global migration processes of POPs. See page 252 for text discussion of this figure.

Figure 10.2. Schematic representation of the key processes affecting concentrations of POPs in the air, vegetation and soil systems. See page 256 for text discussion of this figure.

Figure 10.3. Schematics of (a) different deposition types supplying airborne POPs to vegetation and soil surfaces and (b) important partitioning processes in the terrestrial environment. See page 257 for text discussion of this figure.

Figure 10.4. The Maximum Reservoir Capacity for HCB, January and July. See page 258 for text discussion of this figure.

Figure 11.8. Six zones for analysis of seasonal variations on different regions. See page 278 for text discussion of this figure.

Figure 11.10. PAH emission density maps for the major emission sources in different seasons. See page 279 for text discussion of this figure.

Figure 14.8. Color coded composite maps (in each case with rescaling within each color) assembled from the component maps presented in Figure 14.7. See page 352 for text discussion of this figure.

Figure 14.9. STXM analysis of a colloid sample from 100 m below Lake Brienz. See page 354 for text discussion of this figure.

Figure 14.13. Contour diagrams from infrared mapping obtained at the end of the experiment showing the spatial distribution of the infrared absorption peaks corresponding to (top) Mycobacterium sp. JLS bacteria, (middle) ESHA, and (bottom) pyrene completely degraded. See page 362 for text discussion of this figure.

Figure 15.15. The comparison of different SPME derivatization approaches for the determination of pesticides in rainwater samples with PDMS/DVB fiber coating and PFBBr derivatization reagent. See page 387 for text discussion of this figure.

Figure 22.4. Conceptual depiction of increasing sequestration of organic pollutants in soil pores with the passage of time. See page 542 for text discussion of this figure.

PART I FUNDAMENTAL BIOPHYSICO-CHEMICAL PROCESSES OF ANTHROPOGENIC ORGANIC COMPOUNDS IN THE ENVIRONMENT

1 INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS WITH NATURAL ORGANIC MATTER AND BLACK CARBON IN ENVIRONMENTAL PARTICLES JOSEPH. J. PIGNATELLO 1.1. Introduction 1.1.1. Nature of the Sorbents 1.1.2. The Sorption Process in General 1.1.3. Overview of Weak Intermolecular Forces in Physisorption 1.2. Sorption from the Perspective of the Sorbate 1.2.1. Thermodynamic Driving Forces in Sorption 1.2.2. Quantification of Driving Force Contributions 1.3. Sorption from the Perspective of the Sorbent 1.3.1. Shape of the Isotherm 1.3.2. Sorbent Properties of Natural Organic Matter 1.3.3. Sorbent Properties of Black Carbon 1.3.4. Competitive Sorption 1.3.5. Hysteresis 1.3.6. Apportionment of Sorption between NOM and BC in Environmental Samples 1.3.7. Mass Transfer Rates 1.4. Summary and Concluding Remarks

matter (NOM) and black carbon (BC) materials. These organic substances usually dominate the sorption of nonionic compounds except when present in very low levels or when water is scarce. The chapter does not cover sorption to anthropogenic organic waste liquids or semiliquids, such as weathered oil residues (Jonker and Barendregt 2006) and coal tar (Bayard et al. 2000; Khalil et al. 2006), which can play a significant role as sorbent when present in soil or sediment at high levels. The chapter emphasizes sorption from the aqueous phase, since sorption at the relative humidities existing in most environments approaches that under water-saturated conditions. The chapter is not meant to be an exhaustive review, but a personal perspective of the author that emphasizes the author’s own work and the more recent literature. It considers sorption separately from the perspective of the sorbate and the sorbent. Although this distinction is artificial (obviously, they form a complex), it is useful here as an organizational aid.

1.1.1. Nature of the Sorbents 1.1. INTRODUCTION Sorption refers to the process of molecular exchange between molecules in a gas or liquid phase and a solid phase (the sorbent). Sorption to natural solids typically plays a major, fundamental role in a compound’s transport, reactivity, and bioavailability in the environment. The topic of sorption to natural solids is huge and extends over many decades. This chapter emphasizes sorption to organic substances in soils and sediments, including natural organic

Natural organic matter (NOM) is defined as the organic substances remaining after advanced decomposition of biomass below temperatures where pyrolysis becomes important. It includes humic substances (humic and fulvic acids, humin) and geologically older organic matter present in kerogen and coal. Biomass sources of NOM derive mainly from photosynthetic processes of plants and algae and secondary processes of fungi and heterotrophic bacteria. Natural organic matter can exist in a variety of states: dissolved molecules or molecular aggregates, colloidal particles,

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang.
2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

3

4

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

surface patches or coatings on minerals, intimate complexes with clay-size minerals, and discrete particles. At the primary level NOM is a heterogeneous mixture of functional units within charged, polydisperse molecules that include nonpolar alkyl, carbohydrate-like, protein-like, lignin-like, heterocyclic, and polyaromatic moieties (Hayes and Clapp 2001; Schulten and Schnitzer 1997). This author hesitates to illustrate a hypothetical NOM structure for the reader, since consensus does not exist for one. Most of our knowledge about the structure and composition of NOM has been gained from studies on “dissolved” NOM (DNOM) from natural waters or soil extracts. The distinction between dissolved and colloidal is arbitrary but operationally it is defined by many researchers as the filtration membrane cutoff of 0.45 mm. Most DNOM molecules do not exist individually, but rather are self-associated by hydrogen-bonding and other weak forces in aggregates. Molecules of NOM complex strongly with trivalent metal cations (e.g., Fe and Al) and less strongly with divalent cations and other inorganic ions. The size of individual NOM molecules depends on the source, and may range from a few hundred to 105 daltons (Da) or larger. Most researchers believe that solid NOM can be characterized as a three-dimensional “phase,” consisting primarily of macromolecules with a fraction of molecules below 103 Da. The reader is referred to various treatises and reviews on humic substances (Aiken et al. 1985; Greenland and Hayes 1981; Hayes et al. 1989; Leenheer 2009; Schnitzer and Khan 1978; Stevenson 1994; Sutton and Sposito 2005). Valuable background on geologically aged NOM is given by Allen-King et al. (2002). Solid NOM may be characterized as a randomnetwork macromolecular organic solid. The closest analogy is lignin, the random–network “polymer” in plants derived from phenylpropane units substituted randomly with methoxyl, phenolic, carbonyl, and quinoid groups and giving plants their woody character (Glasser and Kelley 1987). Black carbon is the carbonaceous byproduct of the incomplete combustion of biomass or fuels (Goldberg 1985). It includes char, in which the original material is carbonized in the solid state, and soot, which is formed by the condensation of precursors from the gas phase. Black carbon is a small but important component of total organic carbon in undisturbed soils and sediments, but can be more prevalent in deep-sea and marine shelf sediments, in areas of frequent or recent fires, and in areas of high industrial activity (Masiello and Druffel 1998; Skjemstad et al. 1999). Soot is a component of atmospheric aerosols (Schmidt and Noack 2000). Black Carbon plays important roles in various geo- and bio-geochemical processes, the global carbon cycle, carbon sequestration, radiative heat balance of the planet, and pulmonary toxicity of aerosols (Lighty et al. 2000; Ramanathan and Carmichael 2008). The potential importance of BC as an adsorbent of environmental pollutants was first noted in the pioneering work of Gustafsson and Gschwend (Gustafsson and Gschwend 1997; Gustafsson et al. 1997).

Black carbon is not a single material, but a continuum of materials whose properties depend on source stock and formation conditions. The BC body is composed of single and short stacks of polyaromatic (graphene) platelets rimmed with ketone, ether, hydroxyl, quinoid, carboxyl, and other functional groups. The BC platelet—averaging 7–17 fused rings for biomass chars depending on formation conditions (Brewer et al. 2009; Kaneko et al. 1991)—is considered well represented by the terraced (0001) basal surface of graphite (Donnet et al. 1993). Thus, the BC surface may be considered micrographitic in character. The stacks are arranged in a highly disordered fashion creating a pore network (Boehm 1964; Goldberg 1985; Palot
as et al. 1996). The platelets may contain five- and seven-membered rings that introduce curvature, contributing to the disorder (Harris and Tsang 1997; Shibuya et al. 1999). Raw BC is typically highly porous and thus highly surface-active. The pores lie mainly in the micropore (<2 nm) and mesopore (2–50 nm) size ranges (IUPAC definitions). The high surface area and presence of abundant micro- and mesoporosity render raw BC a stronger sorbent of organic compounds than NOM, especially at low concentration. The pore structure of BCs has sometimes (Boehm 1964; Xia and Ball 1999) been likened to that of activated carbons (ACs), which are produced under controlled conditions of temperature and gas composition. However, some of the porosity of BC may be inaccessible because of poor connectivity, plugging by uncombusted residues, or plugging by natural substances in the environment. In most places the chapter considers the sorbent behavior of NOM and BC separately, although a certain degree of overlap in their natures is to be recognized. 1.1.2. The Sorption Process in General Sorption may be categorized by the type of intermolecular forces that characterize the sorbate-immobile phase interaction: chemisorption, physisorption, and ion exchange. Chemisorption is typified by strong bonds (e.g., covalent and ligand–metal coordination bonds) in which orbital mixing predominates. It is less frequently encountered in environmental systems than are physisorption and ion exchange. Unless the covalent bond breaks reversibly—in most relevant reactions it typically does not—chemisorption will lead to loss of identity of the starting molecule. Physisorption and ion exchange are characterized by the “weak” intermolecular forces discussed later, and the identity of the starting molecule is preserved on desorption. This chapter does not deal with chemisorption. Physisorption may be further categorized by how the sorbate mixes with the solid, whether by adsorption or absorption. Adsorption is the association of a molecule at the interface between a fluid and the surface of a solid whose atomic/molecular lattice cannot be penetrated by sorbate molecules. Absorption (sometimes called partitioning) is

INTRODUCTION

the intermingling (dissolution) of a molecule within the atomic/molecular lattice of a solid. The only solids allowing lattice penetration in this way commonly found in the environment are composed of organic matter. While the terms adsorption and absorption have utility, it is more at the phenomenological level than the mechanistic. Adsorption to a lattice-impenetrable solid may encompass several qualitatively different processes: (1) the resting of molecules on discrete surface sites, (2) the partitioning of molecules into an ordered microscopic hydration phase near the surface, (3) the condensation of the molecules into a liquid-like state in small pores, and (4) the layering of molecules on the surfaces of water films that coat particles. Processes (2) and (3) can be regarded mechanistically as more akin to dissolution than to association at a discrete surface. Likewise, absorption in natural organic matter may involve residence of some molecules within closed pores and/or at specific molecule-scale sites within the phase; it is not clear whether a distinction between adsorption and absorption is meaningful in such cases, at least at the molecular scale. Sorption of a chemical species x from the aqueous phase (xw) can be written as a series of steps in a thermodynamic cycle (Fig. 1.1): Step 1

xw > xg

Step 2

0

S>S 0

xg Step 1

xw Step 2

water-wet sorbent

xg Step 3

x

Figure 1.1. Sorption as a series of steps in a thermodynamic cycle: (1) transfer from water to gas phase; (2) preparation of the (waterwet) sorbent; (3) transfer from the gas phase to the prepared site.

DGg
w DGreorg 0

Step 3 xg þ S > x    S DGinteract Sum

5

xw þ S > x    S0 DGinteract ¼ DGg
w þ DGreorg þ DGinteract

ð1:1Þ

Step 1 is the transfer of x from water to the gas phase, encompassing desolvation of x and collapse of the preexisting solvation shell to the bulk water structure. Step 2 is the reorganization of the sorbent matrix needed to accommodate x, such as the formation of a cavity to fit x, or displacement of existing water molecules or other organic molecules or ions at the site. Step 3 establishes interactions of x with the nowprepared sorbent. Sorption from the vapor phase is the same except without step 1. The Gibbs free energy of sorption DGsorp is the net change in free energy summed over all the steps. The DGsorp is related to the dimensionless thermodynamic sorption coefficient Ksorp by 

DGsorp ¼
RTlnKsorp ;

Ksorp ¼

as =as  afl =afl

ð1:2Þ

where R is the universal gas constant, T is temperature [in Kelvins], a is the activity in the solid or fluid state, and a is the corresponding reference state activity. The value of DGsorp will depend on the choice of the reference state for the sorbed phase. The relationship between Ksorp and a conventional sorption coefficient defined by one of the

isotherm models (see Section 1.3.1) will likewise depend on the choice of the sorbed phase reference state. Note that if the sites are heterogeneous in energy, DGinteract and possibly DGreorg will be dependent on concentration; hence, Ksorp will be concentration-dependent. However, sorption always tends toward linearity (i.e., a constant ratio of sorbed to fluid-phase concentrations) as the concentration in the fluid phase tends toward zero. In the environment many organic contaminants can be present at concentrations in the fluid phase ranging from vanishingly small to the limit of solubility or vapor pressure. 1.1.3. Overview of Weak Intermolecular Forces in Physisorption Noncovalent intermolecular attractive forces are listed in Table 1.1. For physisorbing compounds these so-called weak forces act simultaneously and additively in combinations appropriate to the structures of the interacting species. 1.1.3.1. London–van der Waals and Coulombic Forces. The forces between a nonionic molecule and an uncharged site may include: dipole–dipole, the interaction between permanent dipoles (Type 1 in Table 1.1); dipole–induced dipole, the attraction of a permanent dipole with the dipole that it induces in its neighbor; (Type 2 in Table 1.1); and induced dipole-induced dipole (Type 3 in Table 1.1), the

6

Term (synonym)

Dipole–dipole (dipolar, “Keesom”)

Dipole–induced dipole, (induction, “Debye”)

Induced dipole–induced dipole (dispersion, “London”)

Charge–dipole

Charge–induced dipole (a form of induction)

Charge–charge (coulombic)

H bond

Type

1

2

3

4

5

6

7

TABLE 1.1. Weak Intermolecular Attractive Forces

AH    :B

r

Depiction

2ð4pe0 Þ2 e2 r6

Q2 a

6ð4pe0 Þ2 e2 kTr6

Q2 u2

3 a1 a2 I1 I2 2 ð4pe0 Þ2 e2 r6 ðI1 þ I2 Þ

ð4pe0 Þ2 e2 r6

u2 a

3ð4pe0 Þ2 e2 kTr6

u21 u22

Proportional to 1/r2; directional;  70 kJ/mol

Q1 Q2 ð4pe0 Þer
















Energy Averaged Over All Orientationsa

Charge–quadrupole

9 (+)

− + −

sandwich

− + −

parallel-displaced

quadraupole vector

+ − +

+ − +

T-shaped

− + −

attraction between like polarized rings

− + −

attraction between oppositely polarized rings

− + −

− + −

− + −

− + − − + −

3M n2
1  2D 4prNA nD þ 2

where M is molecular weight, r liquid density, and NA ¼ Avagadro’s number, 6.02  1023 mol
1; the index of refraction is the ratio of the velocity of light in a vacuum to that in the substance]; I—first ionization potential of molecule; (i.e., energy needed to remove the least tightly held electron to infinity: i ! i þ þ e
). Source: After Israelachvili (1992).

a ¼ ð4pe0 Þ

Notation: e0—permittivity (ability to transmit an electric field) of a vacuum, 8.85419  10
12C (coulomb) m
1 V
1; e—dielectric constant of the surroundings (ratio of permittivity in the medium to permittivity of a vacuum); r—intermolecular distance; k—Boltzmann constant, 1.380658  10
23 J K
1; T—temperature; Q—Coulombic charge ze, where z is formal charge ( þ 1, þ 2, etc.) and e is charge of an electron; u—dipole moment; a—distortion polarizability [response of electron cloud to an applied electric field, related to index of refraction, nD:

Quadrupole

8 − + −

7

8

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

mutual attraction of momentary dipoles produced by the synchronization of electronic motion in interacting neighbors. These are commonly called dipolar, induction, and dispersion forces, respectively. They are often referred to collectively as London–van der Waals, or simply van der Waals (vdW) forces. The strength of individual vdW forces depends on the separation distance to the inverse sixth power, how the molecules are oriented, and applicable molecular properties of the interacting species such as dipole moment, ionization potential, and polarizability (Table 1.1). Except for very small molecules the most important of the vdW interactions is thought to be dispersion (Hunter 2004). Placing a charge on either the sorbate or the site leads to charge–dipole (Type 4 in Table 1.1) and charge–induced dipole (Type 5 in Table 1.1) interactions, both proportional to the charge and the sixth power of the separation distance. When the sorbate and site are fully and oppositely charged, a Coulombic force (Type 6 in Table 1.1) exists between the charges that is proportional to the first power of the separation distance. While the Coulombic force usually dominates the energy of interaction between ions, it is important to realize that since ions are not ideal point charges, vdW forces contribute significantly. This is true even for simple inorganic ions, as witnessed in chromatographic separations (Fritz 2005). Moreover, the uncharged regions of the organic ion will undergo the other weak interactions permitted by structure and orientation. 1.1.3.2. Hydrogen (H-) Bonding. Hydrogenbonding(type7 in Table 1.1) is the interaction between an acidic proton of a donor and the lone pair electrons of an acceptor (AH    :B).It is included among the noncovalent forces; however, whereas weak H bonds are due mainly to dipolar interactions, the covalent nature of the H bond increases with its strength, such that very strong H bonds are essentially three-center, fourelectron covalent bonds (Gilli and Gilli 2000). The most important type of H bond in environmental systems occurs when the atoms A and B terminate in O, N, or S. These “ONS” H bonds are highly oriented (A--H--B angle, 180  15 , except in intramolecular H bonds), and in most relevant environmental systems range in enthalpy from 6 to 70 kJ/mol. Gilli et al. (2009) categorize ONS H bonds into ordinary (Ohb), charge-assisted ( þ /
CAhb, þ CAhb, and
CAhb), and resonance-assisted (RAhb) leitmotifs: A
H    : B ½Að1=2Þ
   H    : Bð1=2Þ þ

ð1:3Þ

Ohb þ =
CAhb

ð1:4Þ

½A    H    : B þ

þ CAhb

ð1:5Þ

½A    H    : B



CAhb

ð1:6Þ

R
AH    : B¼R () R¼A    HB
R

RAhb

ð1:7Þ

The Ohb and þ /
CAhb leitmotifs differ only in the degree of proton transfer between A and B. Formation of an RAhb takes place when H bonding is accompanied by p–conjugated bond delocalization, such as in the carboxylic acid dimer: H O

R2

H O

R1

O O O H

R1

O R2

O O H

The leitmotifs þ CAhb,
CAhb, or RAhb are the strongest H bonds. The strength of the H-bond increases as the difference in pKa values of AH and BH þ (DpKa ¼ pKa,AH – pKa,BH þ ) approaches zero (Gilli et al. 2009). Thus, water (DpKa ¼ 14) forms only weak H-bonds with itself [8 kJ/mol (Silverstein et al. 2000)] and with most ONS functional groups. Water forms stronger H bonds with carboxylic acids, phenols having strongly electron-withdrawing groups, and protonated nitrogen groups. Table 1.2 groups H bonds in terms of strength between organic functional groups relevant to the structures of NOM and many common contaminants. Particularly strong H bonds exist in carboxylic acid dimers, in amide dimers, in conjugate pairs (RCO2H/RCO2
, ArOH/ ArO
), between RCO2H or RCO2
and certain nitrogen or protonated nitrogen compounds, and between phenols and certain nitrogen compounds (Table 1.2). Hydrogen bonds are also possible between ONS donors and the p bond in alkenes or aromatic rings (the so-called p--H bond); between ONS acceptors and acidic C--H groups, TABLE 1.2. Estimated Strengths of Intermolecular H-Bonded Pairs for Some Relevant Functional Group Pairs Weak H Bonds (4–17 kJ/mol) Most pairwise combinations of aliph-OH, R3N (1 , 2 , and 3 ), R--O--R, R2C¼O, RNO2, RC(¼O)NR2, RCN H Bonds of Moderate Strength (17–29 kJ/mol) Pairs of RC(¼O)NHR or RC(¼O)NH2 ArNR2 with Ar--OH or RCO2H Ar--OH with RCO2
aliph-OH with ¼N-Strong H Bonds (29-67 kJ/mol) Dimers of RCO2H RCO2H with RCO2
ArOH with ArO
RCO2H with ¼N-RCO2
with ¼NH þ-- or R3NH þ ArOH with ¼N-ArO
with ¼NH þ-- or R3NH þ Notation: aliph—is aliphatic; Ar—is aromatic; R—is either aliphatic or aromatic group bonded through C; ¼N-- is N that is doubly bonded to C, such as in azine- and azole-type (i.e., heterocyclic aromatic N not in conjugation with the ring), or oxime (R2C¼N--OH or RCH¼N--OH) functional groups. Source: Gilli et al. (2009).

INTRODUCTION

and between acidic ONS protons and aliphatic halogen atoms. As these H bonds are, with some exceptions, much weaker than ONS-type H bonds, they may not be able to compete with water when water is abundant. 1.1.3.3. Interactions between p-Conjugated Systems. It has been realized for a long time that special weak interactions may occur between arene units. Arene is defined as aromatic and related cyclic p conjugated systems. Arene–arene interactions are believed to play a role in many chemical and biological phenomena (Hunter et al. 2001; Hunter and Sanders 1990; Janiak 2000; Meyer et al. 2003; SchmidtMende et al. 2001). It now appears likely that they play a role in environmental partitioning as well. The nature of arene–arene interactions is not fully understood. The strongest interactions occur when one unit is p-electron rich (donor, D) and the other is p-electron poor (acceptor, A). These interactions are known as p-stacking or p–p electron donor–acceptor (p–p EDA) interactions. The units undergoing p–p EDA interaction typically associate in a sandwich, but more often in a parallel displaced fashion (Table 1.1). According to current understanding (Cockroft et al. 2007; Gung and Amicangelo 2006; Hunter et al. 2001; Hunter and Sanders 1990; Sinnokrot and Sherrill 2006), the p–p EDA complex is a hybrid structure that can be written n
 o D   A ¼ ðD   AÞquad () ðD AÞvdW () D þ
A
CT

The 1 : 1 C6H6: C6F6 complex forms a solid that crystallizes in a parallel displaced arrangement and melts 20 K higher than either monomer (Williams 1993). While neither monomer has a permanent dipole moment, the 1 : 1 gas–phase complex has a dipole moment of 0.44 Debye, about one-eight of a charge (Steed et al. 1979). All VdW forces between rings and their substituents are believed to be minor because they largely cancel out with those previously existing between momomers and displaced solvent molecules (Cockroft et al. 2007; Hunter 2004). (However, they obviously would play an important role in gas-solid adsorption.) The CT structure is the dative bond formed by one electron transfer from the highest occupied molecular orbital of D to the lowest unoccupied molecular orbital of A. It contributes primarily to the excited state, often yielding an absorbance band in the near UV–visible region, but may contribute appreciably to the ground state energy in strong D    A complexes (Gung and Amicangelo 2006). The p–p EDA bond enthalpy approaches that of a strong H bond (Foster 1969; Perry et al. 2007). It scales with the respective donor and acceptor ability of the opposing units determined by substituents that donate or withdraw electron density, especially through the p system. The flow of electron density is illustrated below: a

ð1:8Þ representing quadrupolar, vdW, and charge transfer interactions between ring systems. In solution, ðD    AÞquad is believed to dominate the energy. In general, p systems have a quadrupole vector perpendicular to the plane of the nuclei, where the p cloud on either side of the plane is polarized oppositely to the s-bond framework (type 8 in Table 1.1). Complexation may result if interactions between quadrapoles is net attractive. A classic example is the 1 : 1 complex between benzene and hexafluorobenzene (Williams 1993). The monomers have quadrapole moments of opposite sign and magnitude (
33.3 and 31.7 C/m2, respectively (Vrbancich and Ritchie 1980). Benzene is electron rich in the p cloud and electron poor in

+

N

9

d

Multiple alkyl and/or --NR2 groups confer strong donor character. Donor strength also trends with p cloud polarizability; thus, polyaromatic hydrocarbons (PAHs) are p donors whose strength appears to increase with fused ring number, at least up to 4. Especially strong electron withdrawing groups include --NO2, --NR3 þ , --CF3, --C(¼O)R, --SO2R, and --CN. Halogen has opposing s-withdrawing and p-donating effects; yet a ring may become net p-accepting as the number of halogens exceeds a critical number, as occurs with F. Especially favorable complexation occurs when the acceptor is a charged heteroaromatic ring. In this case stability is aided by a charge–quadrupole (“p
cation”) interaction (Type 9, Table 1.1) (Qu et al. 2008). An example is the complex in

N

H

N H

the s bond system, while hexafluorobenzene is polarized in the opposite sense. Thus, attractive forces occur between the p clouds and to a lesser extent between the s systems.

H

N H

water between phenanthrene and o-phenanthroline mono- or dication (Wijnja et al. 2004), leading to a dramatic increase in the apparent solubility of phenanthrene.

10

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

The p–p EDA complexes form readily in water and in both polar and apolar organic solvents (Breault et al. 1998; Ferguson and Diederich 1986; Foster 1969), although the relation with solvent polarity is complex (Breault et al. 1998). Like-polarized arenes (D    D; A    A) associate much more weakly than do oppositely polarized arenes, and their complexes tend to orient T-shaped in order to maximize p–s and minimize p–p interactions. Thus, benzene and small PAHs orient roughly T–shaped in their crystals (Newcomb 1994), and benzene orients T-shaped in its most stable gas–phase complex (Iimori et al. 2002; Morimoto et al. 2007).

that raises an entropy and/or enthalpy penalty for dissolution in water. It should be noted that hydrophobicity may have a different connotation when speaking of the distribution of an organic molecule among qualitatively different microdomains within a solid (i.e., NOM phases, narrow pores, and near surfaces) because water exists in a different organizational state in those environments than it does in bulk water. The following are the salient features of the hydrophobic effect according to contemporary understanding. .

1.2. SORPTION FROM THE PERSPECTIVE OF THE SORBATE 1.2.1. Thermodynamic Driving Forces in Sorption The contribution of each force listed in Table 1.1 to the net driving force for sorption from aqueous solution depends on the difference in free energy of the interaction with water compared to that with the sorbent. Teasing apart the contributions of individual noncovalent forces in sorption has been a challenge. The following discusses the state of our knowledge about the contributions of individual noncovalent forces to the sorption free energy. Such a discussion must consider both positive and negative driving forces; examples of the latter are solvation and steric hindrance to sorption. 1.2.1.1. Desolvation and the “Hydrophobic Effect”. A major driving force for sorption from the aqueous phase of nonionic compounds, whether to minerals or organic matter, is the exclusion of the organic solute from water known as the hydrophobic effect. This is supported by a vast and relentless stream of data. The term hydrophobic effect refers to the forces that limit the solubility of apolar molecules—or parts thereof—in water. It is also terminology that refers to the clustering of hydrophobic units such as surfactants (surfaceactive agents) into micelles, the folding of biological molecules, and the behavior of solutes and water near apolar surfaces. Because of the central role of the hydrophobic effect in chemistry and biology, many studies have been devoted to a resolution of its underlying nature. For reviews, see Chandler (2005), Lazaridis (2001); and Southall et al. (2002). While concordance has not yet been reached on all its aspects, researchers are in general agreement that the hydrophobic effect arises from disruption of the cohesive energy of water, not from any special attraction between hydrophobic molecules nor of any special repulsion between apolar entities and water molecules. The disruption originates in the greater ordering, and fewer—if not stronger— water–water H bonds within the hydration shell surrounding the hydrophobic entity than in bulk water itself, a situation

.

.

.

.

There is no evidence for long-range forces between hydrophobic entities. The solvation free energy for alkanes in water is pairwise-additive (Wu and Prausnitz 2008); that is, the Henry law constant, which is proportional to the free energy of solvation, is linearly related to the number of C--C bonds. This means that the hydrophobic effect is primarily a local phenomenon limited to only a few angstroms (Chandler 2005; Wu and Prausnitz 2008). Aggregates of nonpolar molecules less than 1 nm in radius are unstable in water (Chandler 2005; Southall et al. 2002). This means that nonpolar entities such as alkanes (Wu and Prausnitz 2008) and PAHs (Wijnja et al. 2004) have no tendency to self-associate in water much below their solubility limit. This and the previous point are consistent with a solvent-centered, rather than solute-centered origin, of the hydrophobic effect. Since there appears to be no special driving force for association of hydrophobic entities, terms such as “hydrophobic bonding” and “hydrophobic interactions” are misleading. The hydrophobic effect is sometimes described as being entropy-driven. In fact, both entropy and enthalpy play a role, depending on the solute and temperature (Southall et al. 2002). The dissolution of hydrocarbons is unfavorable in cold water because of entropy, and in hot water because of enthalpy. Smaller solutes tend to force ordering of neighboring water molecules, resulting in an entropy penalty for dissolution, whereas larger solutes tend to be more effective at breaking water–water H bonds, resulting in an enthalpy penalty. For water at the interface of an organic liquid or an extended hydrophobic surface, each water molecule participates in about one fewer H bonds than in bulk water (Chandler 2005). However, the H bonds near a hydrophobic surface appear to be stronger; the enthalpy of the H bond of water in the first hydration shell of argon in water is 2.1 kJ/mol greater than in bulk water (Silverstein et al. 2000). Some OH groups of water at thewater–liquid interface of hydrocarbons or chlorinated hydrocarbons orient toward the organic phase in the fashion, --O--H    organic. That this interaction is attractive is indicated by the shift to

SORPTION FROM THE PERSPECTIVE OF THE SORBATE

.

.

lower energy in the stretching frequency of such OH groups relative to that of the O--H protruding into air at the water–water vapor interface (Moore and Richmond 2008). This is consistent with a generally attractive interaction between water and “hydrophobic” surfaces. Due to vdW interactions of water with the hydrophobic entity, the interfaces between water and organic liquids or between water and hydrophobic surfaces are sharp, with no evidence for a vacuum-like gap (Chandler 2005; Moore and Richmond 2008). While the hydration shell of apolar entities may be more ordered than bulk water, both the “iceberg” structure originally proposed by (Frank and Evans 1945) and the “crystalline water cage” structure that forms around the clathrate hydrates of noble gases and some small hydrocarbons at suitable temperature and pressure (Bontha and Kaplan 1999; Jeffrey 1984) are probably exaggerations of its true nature.

Any given functional unit of a molecule introduces opposing forces for sorption. On one hand, its size and polarizability tend to disrupt the structure of bulk water and drive the molecule to the solid; on the other hand, its permanent polarity tends to drive the molecule to the aqueous phase. This is best exemplified by the halogens. Halogen (except F) is large and polarizable, contributing to molecular hydrophobicity. By contrast, halogen polarizes the C--X bond because of its electronegativity. We know that bond polarization plays a role because hydrophobicity (as octanol–water partition coefficient, Kow) increases with halogen substitution more so for sp2- than sp3-substituted carbon (alkane < alkene < aromatic ring). The same is true for most functional groups bonded to carbon through O, N, or S. This is so for two main reasons. (1) , sp2 carbon is more electronegative and competes better for the s-bond electrons than does sp3 carbon; and (2) , in sp2 systems, delocalization of a halogen electron pair into the p system counteracts the bond dipole moment. δX δ+ sp3 halogen

δX δ+

+ X sp2 halogen

The question arises about the importance of dispersion as a driving force for sorption. Many will point to the general relationship between molecular size of a series of apolar compounds (or apolar functional groups) and sorption intensity as support for the importance of dispersion interactions with the sorbent. However, it is believed that dispersion forces between a hydrocarbon entity and its water solvation shell are similar in magnitude to those between the entity and its hydrocarbonaceous neighbors when dissolved in a

11

hydrocarbon solvent (Israelachvili 1992). It follows that, in the process of exchange between the aqueous phase and the sorbent, dispersion interactions roughly cancel out, and dispersion per se is not a major driving force in sorption, but rather the effect of molecular size is manifested in the hydrophobic effect. By contrast with solution–solid sorption, gas–solid sorption is highly driven by dispersion, since intermolecular forces in the gas phase are negligible. 1.2.1.2. Hydrogen Bonding and Dipolar Interactions. Hydrogen bonding favors sorption only to the extent that the net free energy of H bonding with the sorbent is greater than that with water. Hydrogen bonding opportunities are greater in water because of the sheer abundance of H-bond donor and acceptor groups—111 and 55.5 mol/L, respectively. On the other hand, NOM and BC both are rich in carboxylate and phenolate groups, that tend to form stronger H bonds than water, especially with solutes that have phenolic, aromatic amine, amide, and carboxylate groups (Table 1.2).

OH

H-bonding with water

H-bonding with sorbent

Given the abundance of H-bonding groups in NOM, H bonding of contaminants with H-bonding substituents almost certainly takes place in this phase. Nevertheless, direct evidence for this has been elusive. Dixon et al. (1999) observed a progressive downfield shift in the 19 F NMR signal of 4-fluoroacetophenone with increasing fulvic acid concentration in methanol–water, but whether this shift was due to H bonding is questionable since it was not reversed by addition of a six fold excess of acetophenone, which arguably should have outcompeted the fluoro derivative for H-bond acceptor sites interacting with the ketone O. Welhouse and Bleam (1993a, 1993b) showed by 1 H NMR that atrazine dimerizes in CCl4 and also forms H bonds (as donor or acceptor) of moderate strength with mono functional molecules. With bifunctional molecules such as carboxylic acids and amides, atrazine forms strongly H-bonded cyclic complexes by simultaneously accepting and donating a H bond, as in the structure below: Cl N (CH3)2CHNH

N NCH 2CH3

N

H H O

O R

12

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Working with the trifluoroethyl derivative of atrazine in 10% humic acid at pH 11.8, Chien and Bleam (1997) inferred H bonding of atrazine to DNOM by the apparent absence of 19 F NMR resonances characteristic of atrazine dimers. The contribution of H bonding to atrazine sorption, however, remains unclear. In order to gain insight into the driving forces for sorption to NOM, Borisover and Graber (2003) transformed regular soil–water isotherms to soil–hexadecane isotherms (Fig. 1.2) by converting aqueous concentration to the thermodynamically equivalent n-hexadecane concentration via Henry’s constant ratios or solubility ratios. This amounts to switching the reference state from the pure liquid/subcooled liquid to the dilute solution in an “inert” solvent (n-hexadecane) capable only of dispersion. The resulting “rebuilt” isotherms (Fig. 1.2) therefore represent the difference in free energy of interactions in the water-wet solid and the inert solvent, and thus highlight interactions with NOM of a more polar nature. Compounds showing exceptionally strong sorption to a peat soil (45% OC) were those capable of forming strong H bonds with carboxyl and phenolic groups on NOM: 3-nitrophenol, phenol, pyridine, benzoic acid, benzyl alcohol, atrazine, and 2,4-dichlorophenol. Compounds of intermediate sorption strength included trichloromethane and those capable of serving as H-bond acceptors to form weaker H bonds: acetophenone, anisole, 2-chloronitrobenzene, and

NPh Py Atr AP CNB

106 105

nitrobenzene. Compounds sorbing more weakly were apolar aliphatic and aromatic compounds. Interestingly, chlorinated aliphatic compounds sorbed more strongly than did chlorinated aromatic compounds. This suggests that dipolar interactions between the bond dipoles of chlorinated aliphatics, which are greater than those of chlorinated aromatics, engage in dipolar interactions with NOM to favor their transfer from the hydrocarbon phase. 1.2.1.3. p–p EDA Interactions. When permitted by structure, p–p EDA interactions can only add to the driving force of sorption, since water is incapable of p–p EDA interactions. Many pesticides, explosives, antibiotics and other classes of environmental contaminants contain arene groups with strong p–donor or p-acceptor character. In addition, certain natural compounds in soil can act as strong acceptors—there are many examples of quinones among plant signaling chemicals and allelochemicals, such as juglone (5-hydroxy-1,4naphthoquinone). Heteroaromatic amines have been detected in surface and subsurface waters contaminated by shale oil or coal liquification wastes (Sims and O’Loughlin 1989). Humic substances are rich in p-acceptor units, including quinones, charged heterocyclic amines, and aromatic rings having multiple electron-withdrawing carbonyl groups. It has been postulated that the CT bands from internal quinone–hydroquinone complexation partially account for absorbance of

Phe BA DCP NB Ari

Solution concentration, mg/kg

104 103 102 101 100 DBP TCM CH DCCH 1,2-DCB 1,3-DCB 1,4-DCB 1,2,4-TCB 1,2,3-TCB 1,3,5-TCB Bcnz Phcn Naphi Tol

10-1 10-2 10-3 10-2

10-1

100

101

102

103

104

105

106

Solution concentration in n-hexadecane, mg/L

3-nitrophenol NPh Phe phenol py pyridine BA benzyl alcohol Atr atrazine DCP 2,4-dichlorophenol AP acetophenone NB nitrobenzene CNB 2-chloronitrobenzene Ani anisole TCM trichloromethane DBP 1,3-dibromopropane CH cyclohexane trans-1,2-dichlorocyclohexane DCCH Phen phenanthrene Napht naphthalene Tol toluene Benz benzene 1,2-DCB 1,2-dichlorobenzene 1,3-DCB 1,3-dichlorobenzene 1,4-DCB 1,4-dichlorobenzene 1,2,4-TCB 1,2,4-trichlorobenzene 1,3,5-TCB 1,3,5-trichlorobenzene 1,2,3-TCB 1,2,3-trichlorobenzene

Figure 1.2. Pahokee peat-n-hexadecane sorption isotherms (Borisover and Graber 2003). The concentration in hexadecane is calculated by the concentration in water times the ratio of the solubility in hexadecane to that in water.

SORPTION FROM THE PERSPECTIVE OF THE SORBATE

humic substances in the visible region (Del Vecchio and Blough 2004). Humics may also contain strong p-donor units such as alkyl-substituted rings, polyaromatic rings, and pyrrole-type heterocyclic rings. The graphene platelets of black carbon, with their polyaromatic surfaces, may contain both electron-rich and electron-poor regions, depending on their size and the distribution of functional groups along their rims, which may attract p-acceptor and p-donor molecules, respectively (Zhu and Pignatello 2005a) (see Fig. 1.3). The term “p-p interactions” has been used rather loosely in the more recent environmental literature. Speculation is the norm, and the nature of the force is frequently misunderstood. Nevertheless, evidence is emerging for a contribution of p–p EDA interaction to sorption of some compounds by NOM, as well as by BC. Mixing of substituted pyridines and triazine herbicides with DNOM is reported to generate CT absorbance at 460 nm, suggestive of p–p EDA (M€ uller-Wegener 1987). Sorption of the p-donor compounds, pentamethylbenzene, naphthalene, and phenanthrene, to several soils increased with decreasing pH from 7 to 2.5 (Zhu et al. 2004), whereas no similar pH effect occurred for non-p-donor hydrophobic compounds. Other possible systematic affects varying with pH were ruled out. These p-donor solutes may interact with p-acceptor sites in NOM, such as aromatic rings with multiple carboxyl groups or charged aromatic and heteroaromatic amines, whose acceptor ability increases with degree of protonation (i.e., --CO2H and --NH þ ¼ are more electronegative than --CO2
and --N¼, respectively). Pairwise complexation in methanol–water between donors, pentamethylbenzene, naphthalene or phenanthrene, and each of the following model NOM acceptors was identified spectroscopically: 1,3,5-benzenetricarboxylic acid, 1,4,5,8-naphthalenetetracarboxylic acid, and pyridine. Donor complexes with these model acceptors was pH-dependent and gave a CT band in the UV/visible spectrum and upfield 1 H NMR chemical shifts indicative of face-to-face association. On the basis of the free-energy relationship to be discussed in Section 1.2.2, p-p EDA interactions made a small contribution (5%–8%) to the free energy of phenanthrene sorption in a soil (Zhu and Pignatello 2005b). Other, more indirect evidence for p–p EDA interactions of PAHs with NOM or DNOM exists. Polycyclic aromatic hydrocarbons (PAHs) consistently sorb more strongly to soils than do PCBs of comparable hydrophobicity (Kow) (AllenKing et al. 2002; Chiou et al. 1998; Cornelissen et al. 2004; van Noort 2003); PAHs are strong p-donors, while PCBs are not strong donors or acceptors. Isotherms of PAHs to DNOM are frequently nonlinear (Laor and Rebhun 2002; Polubesova et al. 2007), suggesting that specific interactions take place. In terms of the n-hexadecane reference state, aromatic hydrocarbons sorption to Pahokee peat (Borisover and Graber 2003) follows the order, phenanthrene naphthalene > benzene, the same order as their p-donor capablility.

13

A number of papers have postulated “p–p charge transfer” processes between pesticides (triazine, urea, and bipyridylium herbicides) and humic substances based on elevated free radical concentrations measured by ESR spectroscopy (e.g., Senesi et al. 1995; Sposito et al. 1996). Whatever the source of free radicals, their presence cannot be related to the reversible p–p EDA complex referred to here, as no electron transfer to produce free radicals takes place. Moreover, free radicals would likely react with O2 or couple with NOM, resulting in loss of solute identity. Evidence also exists for p–p EDA interactions with elemental carbonaceous materials. Zhu and Pignatello (2005a) found that adsorption of nitroaromatics on nonporous microcrystalline graphite and on BC (wood charcoal) is far greater than predicted by the hydrophobic effect, based on a calibration set of compounds, and in accord with their p-acceptor strength (mono- < di- < trinitrotoluene). Hydrogen bonding of the nitro groups was ruled out. Adsorption of PAHs on the same solids was likewise greater than predicted by the hydrophobic effect and followed the p-donor strength (naphthalene < phenanthrene). Complexation between the PAH donors and the nitroaromatic acceptors was observed in chloroform. These complexes displayed upfield shifts of NMR spectral frequencies due to ring current effects, which is indicative of face-to-face association. They also gave CT bands in the visible region often seen with p–p EDA complexes. The association constant followed the order in expected strength of D    A interaction; namely, mono- < di< tri-nitrotoluene with a given PAH, and naphthalene < phenanthrene < pyrene with a given nitroaromatic. Figure 1.3 shows the strong relationship between the free energy of molecular complexation in chloroform solution and the excess free energy of nitroaromatic adsorption on graphite based on the hexadecane reference state. Chen et al. (2007b) confirmed p–p EDA interactions between polynitroaromatic compounds and the graphene-like surface of carbon nanotubes using a similar approach as Zhu and Pignatello (2005a). Taken together, the results indicate that the graphene surface may be amphoteric with respect to p-interactive adsorbates; referring to Figure 1.3, electron rich regions of the surface attract strong p acceptors while electron-poor regions attract strong p donors. Polarization of the graphite surface near defects and edges is visible by scanning tunneling microscopy (McDermott and McCreery 1994). Figure 1.3 further presents a hypothetical structure of BC showing a potential donor region overlying the polarizable polyaromatic center of the sheet, and acceptor regions in the vicinity of p-acceptor moieties along the rims. 1.2.1.4. Steric Effects. Steric hindrance is possible both in partitioning and adsorption, but very little attention has been paid to it until relatively recently. Steric effects in partitioning to NOM have not been systematically investigated. Partitioning into a solid phase involves the opening of a cavity for

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

-Ge of adsorption on graphite relative to hexadecane reference state (J/mol)

14

26000 24000 TNT

22000 20000

DNT

18000 16000 14000 with D = NAPH with D = PHEN with D = PYR

MNT

12000 10000

-4000 -2000 0 2000 -ΔG of molecular complexation with model donor in chloroform solution (J/mol) (a)

step defect

⇑

⇓

δ−

⇑

δ+

δ−

edge defect

(edge-on view of graphite layers)

π-donor surface

CO2

O O

NH

O OH O

CO2

π -acceptor surface

Figure 1.3. (a) Free-energy relationship between excess adsorption on nonporous graphite and complexation with p donors, naphthalene, phenanthrene, or pyrene in chloroform measured by 1 H NMR for mononitrotoluene, 2,4-dinitrotoluene, and trinitrotoluene. The excess adsorption free energy is postulated to originate from p–p EDA interactions on the graphene surface. (bottom) Strong p acceptors (e.g., trinitrotoluene) interact with electron-rich regions of the surface, while strong p donors (e.g., phenanthrene) interact with electron-poor regions. Surface charge separation may be due to step/ edge defects (graphite) or rim functionality [black carbon (BC)]. A hypothetical BC platelet with potential donor and acceptor regions is indicated.

the incoming molecule if one does not already exist. The free energy of partitioning from water must include a term for the difference in free energy of cavitation in the solid and in the aqueous phase. The excess free energy required for opening a spherical cavity of radius r in a liquid is the sum of a volumetric term (4/3 pr3E), where E is the volumetric energy density, and a surface term (4pr2c), where c is surface tension (Hummer et al. 1998; Southall et al. 2002). The volumetric term

dominates for small molecules, whereas the surface term dominates for very large molecules or extended surfaces. Cavitation of a solid phase strongly depends on its viscoelastic properties and free volume distribution. For a flexible solid in its equilibrium state, the free-energy cost for forming a cavity (“cavity penalty”) increases systematically with penetrant size for a homologous series of penetrants; atoms or very small penetrants may fit into existing free volume, thus requiring little or no cavitation, whereas complete

SORPTION FROM THE PERSPECTIVE OF THE SORBATE

Cl

vs A

Cl

B

Cl

Cl Cl

Cl vs Cl

A

pore network

15

B

Cl Cl

Cl

A can achieve a flatter conformation on the surface and better fit into pore throats than B

Figure 1.4. Illustration of the contact area and size exclusion hypotheses for steric effects in adsorption.

exclusion from the internal phase would be reached at some very large size. Suppressed sorption on the basis of molecular size for large molecules has been reported to occur in phospholipid bilayers (Dulfer and Govers 1995; Gobas et al. 1988; Kwon et al. 2006; Yamamoto and Liljestrand 2004) ordinarily regarded as being a partition medium. Steric effects are known to play a role in the adsorption of organic compounds to matrix-impenetrable solids (K€arger and Ruthven 1992), both porous and nonporous, and have been claimed for natural solids. There are at least two sources of steric effects, illustrated in Figure 1.4: deviation from molecular planarity size exclusion. (1) Since interactions depend on close molecular approach, deviation from molecular planarity should reduce contact area with a flat surface, other molecular properties being equal (2) Steric effects in microporous solids can be manifested by size exclusion at pore throats. The contact area hypothesis for adsorption to a smooth surface has been tested for alkanes. As they are able to readily adopt a planar conformation, normal alkanes are expected to achieve closer contact with a surface than cycloalkanes, which are restricted by ring puckering. Closer contact favors vdW forces, whose magnitudes are inversely related to separation distance to the sixth power (Table 1.1). The difference in sorption coefficient between n- and c-alkanes, however, is small: the ratio Kn/Kc for gas-phase transfer to the surfaces of liquids and nonmicroporous inorganic solids is close to unity (0.83–1.62) and tends to decrease with the number of carbons (Endo et al. 2008b). However, the difference in interaction enthalpy may be greater than it would appear by the ratio Kn/Kc, since the entropy penalty for transfer to a surface or phase is greater for the n-alkane because it has more degrees of freedom in the gas phase than does the cycloalkane. The contact area hypothesis has also been invoked to explain sorption trends for BC materials and sediments with high levels of BC. For example, in sediment systems polychlorinated biphenyl (PCB) congeners in which coplanarity of the rings is impaired due to ortho chlorine substitution have a lower sorption coefficient (Barring et al. 2002; Bucheli

and Gustafsson 2001; Jonker and Smedes 2000; Jonker et al. 2004; van Noort et al. 2002), greater “fast-desorbing” fraction (van Noort et al. 2002), and higher bioavailability (Jonker et al. 2004) than do coplanar congeners with the same number of chlorines. However, noncoplanar congeners have an inherently weaker tendency to interact with themselves, as witnessed by their higher subcooled vapor pressures (Schwarzenbach et al. 2002). They also partition less favorably into octanol from water (Schwarzenbach et al. 2002) for reasons that are not clear, since close approach applies to both solvents. The coplanarity effect for PCB adsorption to soot and soot-like materials from water held after normalizing the sorption coefficient by Kow (Jonker and Koelmans 2002b). Since BC materials are highly porous, an alternative explanation for the coplanarity effect is size exclusion (molecular sieving) in pores; a noncoplanar congener has a larger critical diameter and therefore would be more restricted in the passages that it could enter. It is well known that diffusion in zeolites becomes severely hindered as the minimum critical molecular diameter approaches the pore diameter (K€arger and Ruthven 1992). Anthracene and phenanthrene are both planar and have nearly identical KOW values (Schwarzenbach et al. 2002), yet anthracene consistently sorbs more strongly than does phenanthrene to BC materials (Jonker and Koelmans 2002b) and BC present in sediments (Cornelissen et al. 2004). Size exclusion may also explain the decline in Langmuir adsorption capacity of PAHs on activated carbon with increasing size (Walters and Luthy 1984), although no mechanism was offered in that study. Adsorption of n-hexane in some activated carbons is much greater than is cyclohexane, said to reflect size exclusion (Endo et al. 2008b). Systematic evidence has now been presented for size exclusion in charcoal BC for a series of planar aromatic compounds, both polar and apolar (Pignatello et al. 2006a; Zhu and Pignatello 2005a). For this sample of BC about 80% of the porosity exists in pores up to 2 nm wide. To normalize for hydrophobic effects, nonporous graphite was used as the reference state; BC–graphite isotherms were constructed

16

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Mean Kch-gr, m2/m2

10 TOL XYL 124 TMB 1235 TeMB 1245 TeMB 12 DCB 124 TCB BNTL MNT DNT TNT BEN

1

0.1 0

1 2 3 Number of substituents

4

10 Mean Kch-gr, m2/m2

NAPH BEN PHEN

1

0.1

1

2 3 Number of fused rings

Figure 1.5. Char-graphite distribution coefficients normalized by surface area of aromatic compounds as a function of molecular size. Each point represents the mean and standard deviation of 11-18 data measured over a range of concentrations. [Data from (Zhu and Pignatello, 2005).] TOL ¼ toluene; XYL ¼ 1,4-dimethylbenzene; 124 TMB ¼ 1,2,4-trimethylbenzene; 124 TCB ¼ 1,2,4-trichlorobenzene; 1235 TeMB ¼ 1,2,3,5-tetramethylbenzene; 1245 TeMB ¼ 1,2,4,5-tetramethylbenzene; 12 DCB ¼ 1,2-dichlorobenzene; BNTL ¼ benzonitrile; MNT ¼ 4-nitrotoluene; DNT ¼ 2,4-dinitrotoluene; TNT ¼ 2,4,6-trinitrotoluene; BEN, benzene; NAPH ¼ naphthalene; PHEN ¼ phenanthrene.

from the experimental BC–water and graphite–water isotherms: xw > xBC xgr > xw xgr > xBC

KBC
w 1=Kgr
w KBC
w qBC KBC
gr ¼ ¼ Kgr
w qgr

ð1:9Þ

In this reaction, K is the distribution concentration ratio between the sorbed and the solution phases, and q is the sorbed concentration normalized by surface area. The BC–graphite isotherms are shown in Figure 1.5. Benzene and monosubstituted benzenes sorb somewhat more strongly on BC than on graphite, which could be true or merely an artifact

of the technique for measuring surface area (Braida et al. 2003). Nevertheless, in Figure 1.5a it can be seen, that the char-graphite distribution ratio decreases—that is sorption becomes weaker relative to graphite—increasing number of ring substituents, regardless of the compound’s polarity. The effect is considerable; for example, adsorption of tetramethylbenzene to BC is about an order of magnitude weaker than benzene at constant sorbed concentration on graphite. Likewise, the char-graphite distribution ratio decreases with increasing fused ring size (Fig. 1.5b), revealing a one-order-of-magnitude difference in benzene and phenanthrene affinities for BC at constant concentration on graphite. Since these compounds are all planar (although some of the nitro groups may orient in a nonplanar relationship with the benzene ring), these results conclusively show that a size exclusion effect is operative that restricts the internal pore network surface area available for adsorption as molecular size increases. Support for steric effects is also evident in the work of Nguyen et al. (2007), who found that in two chars the maximum sorption capacity increased in the order of decreasing molecular diameter among planar compounds: phenanthrene < naphthalene < 1,2-dichlorobenzene/1,2,4trichlorobenzene < 1,4-DCB (Nguyen et al., 2007). 1.2.1.5. Sorption of Ionic and Ionizable Compounds. Because of the deprotonation of carboxyl and phenoxyl groups NOM contains an abundance of charged sites where ion exchange of organocations with native cations (M þ ) may occur. It is often difficult to quantify the contribution of NOM to cation exhange in whole soils because cation exchange sites are abundant on mineral surfaces as well. Organic anions may also show a tendency to sorb, particularly if the nonionic parts of the molecule are large; the classic examples are tetra- and pentachlorophenoxide (Schellenberg et al. 1984), whose sorption may occur as an ion pair with an inorganic cation and is weaker than sorption of the neutral molecule. Perfluoroalkanoic acids and perfluoroakane sulfonic acids (Ahrens et al. 2009; Higgins and Luthy 2006) also undoubtedly sorb as the ion pair, since their pKa values are well below 1. For further information on ion sorption, refer to Schwarzenbach et al. (2002). Sorption, of ionic or ionizable molecules containing multiple functional groups presents a much more complex situation. This will be illustrated with two antibiotic compounds whose sorption to NOM has been studied in some detail—sulfamethazine and tetracycline and their analogs. Sulfamethazine (Fig. 1.6) may exist in water as a cation, neutral molecule, or anion. Therefore the possible modes of interaction depend strongly on pH. (The zwitterion is a small fraction of the net-zero charged molecule.) NOM and, to a lesser extent clay minerals, play a role in its sorption, whereas sesquioxides have not been studied. The mean log KOC of

17

SORPTION FROM THE PERSPECTIVE OF THE SORBATE

1.0 pKa3 = 9.7

8

O

N

9

7

6 6a

5 5a

4 OH 3

10a 10

11a 11

12a 12

1

OH

O

OH

O

OH NH2

2

S NH pKa1 = 2.46

O

Fraction

N

HO H pKa2 = 7.45

TCH22

0.8 0.6

TCH-

TCH2+

TC2-

0.4 0.2

N

O 0.0

H2N Sulfamethazine

pKa1 = 3.3

pKa2 = 7.7

Tetracycline

2

8

4

5

6

7

8

9

10 11

pH

Figure 1.6. Structures of the antibiotics and speciation diagram for tetracycline in solution in the absence of complexing metal ions (Gu et al. 2007). Metal ion coordination sites on tetracycline are as follows. Depending on pH and stoichiometry, Ca2 þ coordinates through the N4/O12a or O12/O1 atomic pairs, and Mg2 þ through the N4/O3 or O10/O12 atomic pairs. As the pH is raised from 2, Cu2 þ successively coordinates to O3, then to O10/O12, and finally to N4/OH-12a.

sulfamethazine in five soils is 2.11 (Carda-Broch and Berthod 2004). The KOC of a structural analog, sulfathiazole, in compost, manure, and solid humic acid follows the order KOCcation > KOCneutral > KOCanion (Kahle and Stamm 2007) and is hardly affected by K þ or Ca2 þ , suggesting that the interaction of sulfathiazole with NOM occurs by ordinary weak forces and the hydrophobic effect. This is consistent with the predominance of the neutral molecule over the zwitterion in solution. Sorption of the cation is enhanced relative to the neutral form due to electrostatic interactions with negatively charged sites on NOM, whereas sorption of the anion is suppressed for the same reason. Sorption of sulfamethazine to a charcoal BC followed the order KOCneutral > KOCcation > KOCanion (Teixido´ Planes and Pignatello unpublished). The KOCneutral was almost six orders of magnitude greater than the Kow of 0.27 for neutral sulfamethazine (Carda-Broch and Berthod 2004), a result that underscores the high affinity that polar compounds can have for BC. Charcoals have some cation exchange capacity owing to carboxyl groups (Chan and Xu 2009) that increases with weathering. However, sorption of the sulfamethazine cation was unaffected by NH4 þ , indicating that cation exchange was unimportant. Sorption of tetracyclines is even more complex than the sulfonamides because of the greater number of exchangeable protons and the tendency to coordinate with bound and free metal ions (Fig. 1.6). Sorption of the cationic form of tetracyclines may dominate sorption in most soils over the pH range 4–8, according to a more recent model (Sassman and Lee 2005). Sorption of tetracycline to dissolved humic acid is maximal for the zwitterion (Gu et al. 2007), which predominates between pH 4 and 7. The strong competition by NaCl indicates that the zwitterion sorbs primarily by ion exchange at carboxylate sites, with little contribution from hydrophobic effects. Sorption of the cation is weaker than sorption of the zwitterion, due to competition with H þ at lower pH values. Sorption of the anion is weaker still as a

result of charge repulsion and unfolding of humic macromolecules at higher pH values. In the presence of metal ions Al and Fe, metal ion bridging between the antibiotic and NOM (Tet–Mn þ –NOM) becomes important (MacKay and Canterbury 2005), and may be responsible for the strong binding of oxytetracycline to manure via the 5-OH (Loke et al. 2002). 1.2.2. Quantification of Driving Force Contributions A critical tool for separating the contributions of individual forces in sorption to natural solids is the free-energy relationship (FER). One obvious FER for testing hydrophobic effects is water solubility log KOC ¼
a log Sw þ b

ð1:10Þ

where a and b are fitting parameters. One problem with the solubility FER is that the reference state implied by the FER—the pure liquid or subcooled liquid—is different for each compound; hence, the relationship only works well for, (1) apolar compounds whose activity coefficient in each other’s liquid is typically not much different than in itself (unity) or (2) a series of compounds with a systematic change in the size of the nonpolar portion of the molecule (Allen-King et al. 2002). Other problems include the limited number of accurate solubility values and the necessity of converting the solubility of solids to that of the sub-cooled state, a conversion that rests on an estimation of the free energy of fusion (Schwarzenbach et al. 2002). A more popular predictive model for contaminant sorption is the FER with n-octanol–water partitioning, which takes the form log KOC ¼ c log Kow þ d

ð1:11Þ

where Kow is the octanol–water partition coefficient (L/L) and c and d are fitting parameters. Because of its importance in biology, the Kow is known for many thousands of

18

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

compounds, is relatively easy to measure, or can be calculated with available algorithms to a fairly high degree of accuracy (SRC). The KOC–Kow correlation has mechanistic significance in the sense that the free energy of octanol–water partitioning reflects the hydrophobic effect and to some degree polar forces (H bonding, dipolar), owing to the --OH group and the presence of a large amount of water in the octanol (21 mol% at equilibrium) in the test. Nevertheless, octanol partitioning cannot effectively represent all the driving forces for sorption of all types of compounds; indeed, while KOC–Kow correlation is reasonably good within a series of structurally related compounds, it deteriorates when compounds of different polarities are included in the dataset. [For tabulations of KOC–Kow FERs, see Allen-King et al. (2002) and Schwarzenbach et al. (2002).] In recent years researchers have tested polyparameter FERs (ppFERs) for various environmental partitioning phenomena, including sorption to NOM (Nguyen et al. 2005; Poole and Poole 1999; Endo et al. 2009b; Niederer et al. 2006b) and to activated carbon (Shih and Gschwend 2009) (see also Chapter 5 in this book for a fuller review). The ppFER takes the following form log KOC ¼ eE þ sS þ aA þ bB þ vV þ c

ð1:12Þ

where the uppercase letters are molecular descriptors and the lowercase letters are regression coefficients whose values reflect the net difference of interactions between the sorbed and dissolved states. The E descriptor combines dispersion and induction forces, S is the dipolarity/polarizability (D/P) descriptor encompassing dipolar and induction forces, A represents combined H-bond acidity (i.e; as A--H) and electron acceptor ability, B represents combined H-bond basicity and electron donor ability, and V represents the energy required to form a cavity to accommodate the molecule in the respective phase. The descriptors were calculated from physical constants or derived from extensive chromatographic datasets in the literature. While its main purpose has been predictive—the fiveparameter Equation (1.12) generally gives a better fit to KOC data than does the single-parameter OC–octanol FER—it is possible to draw some limited conclusions about the contributions of intermolecular forces in sorption to NOM through the application of Equation (1.12). The greatest influence on log KOC among apolar and weakly polar compounds appears to be the cavitation energy term V, followed by E. The contribution of the S term is variously reported to be positive (Endo et al. 2009b) or negative (Nguyen et al. 2005; Niederer et al. 2006b). Among the polar compounds, the H-bond donor/electron acceptor ability (A) term contributes little or nothing to log KOC, while the H-bond acceptor/electron donor ability term (B) contributes substantially, but negatively—that is, favoring water. For example (Nguyen et al. 2005), B contributed negatively by up to 28%

of log KOC, and especially for the ureas, the benzamides, and benzyl alcohol (Nguyen et al. 2005). At least the first two are strong H-bonders and this property favors water not NOM. Equation (1.12) applied to adsorption of a set of 14 diverse compounds from water to granular activated carbon (Shih and Gschwend 2009) showed adsorption depended positively on the sorbate’s V term, negatively on its B term, and weakly on its S term (the sign depended on concentration). A major obstacle to interpreting the ppFER in Equation (1.12) is the appreciable overlap of fundamental forces among the descriptors. Dispersion appears in both E and V descriptors. Dipolar and induction appear in S, A, B, and—to the extent that H-bonding and dipolar interactions play a role in water organization in the hydration shell—V. Another obstacle is the unclear meaning of A and B, which seems to have morphed from a pure H-bond acidity–basicity meaning to the dual H-bond and electron donor/acceptor meaning. Moreover, the contribution of those terms is sometimes incongruous with structural theory; for example, some PCBs and chlorinated alkanes and alkenes contribute several percent to log KOC via the B term (Nguyen et al. 2005), yet have negligible ability to accept a H bond, and are clearly not noted for their electron donor ability. Zhu and Pignatello (2005b) applied a multiparameter FER to sorption of apolar and polar compounds to polyethylene and three high organic soils in which NOM was presumed to be the dominant sorbent. Following the lead of others (Borisover and Graber 2003; Kleineidam et al. 1999a), they found it advantageous to switch the thermodynamic reference state from the pure liquid state to the dilute solution in n-hexadecane, as discussed above. The free energy of sorption to OC was partitioned into terms representing the hydrophobic effect (hyd), dipolarity/polarizability (D/P), H-bonding, and p–p EDA interactions, all in excess of the reference state (notated by the superscript e). The FER is written as:
RTln KOC ¼ DGOC ðhydÞ þ GeOC ðD=PÞ þ GeOC ðH
bondÞ þ GeOC ðp
pEDAÞ þ c

ð1:13Þ

where c is a constant. The DGOC(hyd) was taken to be a linear function of the free energy of hexadecane–water partitioning, so that, ln KOC ðhydÞ ¼ aln KHD þ b. The GeOC ðD=PÞ was taken to be proportional to the D/P descriptor S used in Equation (1.9). The isotherms, which were nonlinear, were constructed over a wide concentration range, which established the concentration- dependence of individual freeenergy contributions. Not surprisingly, sorption of all compounds to polyethylene was linear and could be modeled by considering hydrophobic effects only. For sorption of apolar compounds to the three natural sorbents, hydrophobic effects still predominated, but D/P effects contributed from 15%–40% of

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

Amherst soil

Beulah Zap lignite

Pahokee

1.0

1.0

}

0.8

DCCH Fractional contribution

Fractional contribution

BEN HYD

0.6

0.4

}

0.2

0.0 10 -4

10 -3

10 -2

10 -1

10 0

D/P

101

}

0.8

0.4

}

0.2

0.0 10 -5

10 2

10 -4

10 -3

10 -2

10 -1

D/P

10 0

10 1

Solute Concentration, mmol/L 1.0

1.0

MCB vs DCB vs TCB in BZL

}

0.8

Fractional contribution

BEN, NAPH, PHEN in AS and BZL Fractional contribution

HYD

0.6

Solute Concentration, mmol/L

HYD

0.6

0.4

0.2

19

}

D/P + EDA for PHEN in AS

D/P

0.8

}

12 DCB MCB

0.6

HYD

14 DCB 124 TCB

0.4

135 TCB

}

0.2

D/P

EDA for PHEN in AS

0.0

10 -5

0.0

10 -4

10 -3

10 -2

10 -1

10 0

10 1

Solute Concentration, mmol/L

10 -5

10 -4

10 -3

10 -2

10 -1

10 0

10 1

Solute Concentration, mmol/L

Figure 1.7. Contributions of hydrophobic effects (HYD), dipolar/polarizability (D/P), and p–p EDA to the sorption free energy for the natural sorbents in selected systems. Compounds include benzene, naphthalene, phenanthrene, trans-1,2-dichlorocyclohexane, chlorobenzene, 1,2- and 1,4-dichlorobenzenes, and 1,2,4- and 1,3,5-trichlorobenzenes.

log KOC. Some trends are summarized below, with selected examples given in Figure 1.7: .

.

.

. .

In most cases, with increasing sorbed concentration the percent contribution of the hydrophobic term (% hyd) increases, while % D/P decreases (Fig. 1.7). This indicates a transition to loading of sites that are poorer in D/ P influence—possibly less polar and less aromatic. These results are consistent with those of Endo et al. (2008a). The % D/P increases with dipole moment (CCl4 < CHCl3 < CH2Cl2) and polarizability (cyclohexane < benzene < naphthalene < phenanthrene) (Fig. 1.7). Both % hyd and % D/P for mono through trichlorinated benzenes were insensitive to the number and position of Cl atoms (Fig. 1.7). The % D/P was greater for 2,4-dichlorophenol than for the less polar, less polarizable n-nonanol. The absolute contribution of H-bonding was roughly equal for n-nonanol and 2,4-dichlorophenol, even through the latter was a better H-bond donor toward

carboxylate and phenolate groups on NOM (Table 1.1.) However, it is not possible to tell whether H bonding of these compounds with NOM is more favorable than with water, since the free energy of H bonding with water is embedded in the hydrophobic term.

1.3. SORPTION FROM THE PERSPECTIVE OF THE SORBENT 1.3.1. Shape of the Isotherm When constructed in sufficient detail and over sufficiently wide range in solute concentration, sorption isotherms on natural solids, as well as NOM and BC reference materials, are often found to be nonlinear in solute concentration. Since the activity coefficient in water cw [see Eq (1.2)] of solutes is typically independent of concentration under dilute conditions, nonlinearity implies a change in the activity coefficient in the sorbed state cs. That, in turn, can mean that either sorption sites are heterogeneous in energy

20

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

or the nature of the solid changes with the degree of loading. Both are possible. Heterogeneity can arise from a distribution in pore widths where condensation occurs. Heterogeneity may even be a property of pure nonporous materials due to surface irregularities, defects, and edge effects. Partitioning into a flexible solid can result in swelling that feeds back on the solid’s sorbent properties. In addition, solid phases may not be internally homogeneous with respect to sorption potential. How sorption varies with solute (or vapor-phase) concentration and the concentration of other solutes has important implications for modeling or predicting the behavior and bioavailability of chemicals, as well as for understanding the sorption process mechanistically. The variation in sorption distribution coefficient ranging in aqueous concentration from infinite dilution to water solubility can be as much as three orders of magnitude depending on the degree of nonlinearity in the isotherm. Since, in most cases, bioavailability depends on the concentration in the fluid phase, not the sorbed phase, nonlinearity can have a profound influence on bioavailabilty (Pignatello 2009). Nearly all isotherms models are based on the Langmuir, Freundlich, or Polanyi–Manes equations or variations thereof: 1. The Langmuir equation assumes a limited number of sites of a single energy (Adamson and Gast 1997) and is given by QL KL C q¼ ð1:14Þ 1 þ KL C where q is the equilibrium sorbed concentration, C the equilibrium solute concentration, and QL and KL are, respectively, the Langmuir maximum capacity and affinity parameters. 2. The Freundlich isotherm may be derived from the Langmuir equation by assuming a distribution of site energies (Adamson and Gast 1997). It is given by q ¼ KF C n

ð1:15Þ

where KF is the affinity coefficient and n the exponent reflecting the degree of linearity. The linear isotherm is a special case of the Freundlich isotherm where n ¼ 1. 3. The Polanyi–Manes isotherm is based on the concept of condensation in small pores and is given by q ¼ QP  10½aðe=N Þ b

ð1:16Þ

where e ¼ RT lnðCmax =C Þ and a and b are fitting parameters; QP is maximum capacity; N is a normalizing factor, often taken to be the adsorbate molar volume; e is adsorption potential corrected for adsorption potential of an equal volume of water displaced; Cmax is maximum water solubility of the pure liquid/ subcooled liquid; R is the thermodynamic gas constant;

and T is temperature [see Xia and Ball (2000) and references cited therein]. For vapor–solid isotherms, Cmax/C is replaced by p0/p, where p is partial pressure and p0 is the pure liquid or sub-cooled liquid vapor pressure. Developed originally for AC adsorbents, the Polanyi– Manes model assumes chemical condensation in a liquidlike state in the pore at a pressure governed by capillary forces, beginning in pores of the smallest diameter; water, if present, is displaced (Allen-King et al. 2002; Manes 1998; Xia and Ball 1999). Sorption is nonlinear because of the wide range of pore widths in the solid. In some cases sorption of compounds whose thermodynamic state is solid at the experimental temperature has been observed to be weaker than predicted on the basis of the Polanyi characteristic curve for liquids of similar structure (e.g., benzene vs. PAHs). To explain this, the solid solute is postulated to condense in a crystalline state, filling the pore less efficiently than would a liquid (Xia and Ball 2000). At this time there is no independent evidence for crystallization, and there are examples where the phenomenon of weaker sorption of the solid solute is not observed [e.g., for example, in the chlorinated benzene series (Nguyen et al. 2007; Xia and Pignatello 2001)]. While crystallization seems plausible in mesopores, which are filled at higher concentrations, the concept of crystallization in a micropore is not a compelling one. As we have seen, size exclusion in pores may limit the available adsorption space of the larger (solid) molecules and rationalize their observed weaker sorption. Cation exchange of simple inorganic ions on soils follows a Langmuir-like isotherm, leveling off at the maximum cation exchange capacity of the sample (CEC, measured in equivalents per unit mass). However, for organocations, one often finds continued sorption beyond the CEC related to molecular size. Since charge cannot accumulate on the solid, this additional sorption must be due to sorption of the ion pair. The following three equations (1.17 to 1.19) illustrate the simple monovalent counterion case: x þ þ M þ
S > x þ
S þ M þ x þ þ y
> ðx þ y
Þ
S qtot ¼ qx
S þ qðxyÞ
S ¼

Kex KIP

ð1:17Þ ð1:18Þ

CEC  Kex ½x þ þ KIP ½y
½x þ ½M þ þ Kex ½x þ ð1:19Þ

In these equations, y
is a counterion forming an ion pair with x þ , q refers to solid-phase concentrations, brackets refer to aqueous-phase concentrations, Kex is the ion exchange equilibrium constant; and KIP is the ion pair sorption equilibrium

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

constant. The value of KIP itself may be concentration-dependent, depending on the tendency of the organocation to form ordered layers (hemimicelles, admicelles) on the surface, as is the case for organocation surfactants, such as tetraalkylammonium salts. In theory as well as experimentally, sorption always tends toward linearity as solute concentration approaches infinite dilution. Note that the Freundlich and the Polanyi models are unrealistic at very low concentrations because they fail to linearize. Assigning mechanism predominantly on the basis of data fit to a particular model is not advised; for one thing, the above three models do not have the same number of fitting parameters. 1.3.2. Sorbent Properties of Natural Organic Matter Solid-phase dissolution (“partitioning”) is the historic paradigm for sorption to NOM. The basis of this paradigm is the conceptualization of NOM as a hydrated, “loosely-knit” gel phase capable of “dissolving” organic solutes and permitting them move about freely in the milieu, as if it were a liquid (Chiou 1989, 2002; Chiou et al. 1998). Since liquid–liquid partitioning is generally linear under dilute conditions (i.e., the activity coefficient in each liquid is concentration-independent), the frequent finding of sorption “linearity” has reinforced the solid-phase dissolution concept. However, in many of these studies, isotherms were constructed over a limited range in solute concentration, where curvature, if any, would have been difficult to discern, or where number and/or quality of data were inadequate to statistically distinguish linearity from nonlinearity. In many studies linearity had been merely assumed for the sake of simplicity in modeling, especially when sorption was not the main focus. More detailed studies show that sorption to soils when NOM is likely to be the predominant sorbent material (i.e., high OC content, nonpolar solute, abundant water) is usually nonlinear. The nonlinearity of sorption in Pahokee peat, a high-organic reference soil from the International Humic Substances Society, is such that the distribution coefficient can range over two orders of magnitude from very low to concentrations approaching water solubility (Endo et al. 2008a; Xia and Pignatello 2001). In addition, sorption is nonlinear and competitive even for nonpolar solutes in BC-free humic acid isolates (e.g., Huang et al. 1997; Pignatello et al. 2006b; Xing and Pignatello 1997; Zhao et al. 2001). Finally, isotherms constructed using a variety of experimental techniques of PAHs—mere hydrocarbons—in DNOM, arguably the simplest and most homogeneous form of NOM, are convincingly nonlinear and competitive (Laor and Rebhun 2002; Pan et al. 2007; Polubesova et al. 2007). Recognizing that the solid-phase dissolution paradigm is useful, but likely a simplification, a new paradigm for sorption to NOM has emerged—one in which SOM exists as a heterogeneous mixture of physical states that provides a

21

hierarchy of sites, each limited in capacity. Several postulates have been offered to explain this heterogeneity. They are based on (1) domain-based preferential sorption, (2) functional-group-based preferential sorption, and (3) the physical state in which NOM exists. A clear choice among these is hampered by a lack of molecule-scale evidence. 1.3.2.1. Domain-Based Preferential Sorption. This hypothesis holds that strands of NOM associate together on the basis of functional group identity to form domains large enough in scale to act independently as micropartition phases (a “carbohydrate-like domain,” an “aromatic domain,” etc.). Domain selectivity can clearly depend on domain chemistry—witness the wide range in solvent–water partition coefficient for a given compound in different solvents (Schwarzenbach et al. 2002). Partition domains in NOM would have to be limited in maximum capacity, however; otherwise, sorption would merely be a linear combination of linear terms—still linear overall. Evidence for functional unit homogeneity on a scale large enough to serve as separate partition domains is mixed. Several research groups have identified crystalline and amorphous polymethylene domains in humic acids (Gunasekara et al. 2003; Hu et al. 2000; Mao et al. 2002; Mao and SchmidtRohr 2006; Schaumann et al. 2005) and whole soils (Lattao et al. 2008) by nuclear magnetic resonance (NMR) spectroscopy. These domains originate from the remnants of plant cuticular waxes, cutan, cutin, and subarin, which are long preserved in NOM and often make up percent levels of NOM (Chen and Xing 2005). It is reasonable to postulate sorption in the amorphous polymethylene domains. Sorption in the crystalline domain, by definition, is not possible. On the other hand, carbohydrate-like and lignin-like moieties in humic acid (Mao and Schmidt-Rohr 2006) and whole soils (Lattao et al. 2008) appear to be intimately mixed and therefore not likely to form independent domains. Most studies testing domain-based preferential sorption rely on correlations between sorption and bulk functional group composition of NOM determined by solid-state NMR techniques. Overall, however, the results are conflicting or inconclusive in many respects: on whether sorbates prefer aromatic or aliphatic domains (Chefetz et al. 2000; Kile et al. 1999; Mao et al. 2002; Ran et al. 2007), or have no preference (Wen et al. 2007); on whether desorption rate correlates or not with aromaticity (Cornelissen et al. 2000; Lucht and Peppas 1987); and on whether sorption intensity correlates with an index of NOM polarity (Chen et al. 2005; Kile et al. 1999). Cook, in Chapter 13 of this book, discusses many of the technical factors that limit the success of the NMR approach. There are also a number of conceptual limitations: (1) a disregard for the role of local “physical organization” of NOM, (2) the importance of excess free volume (porosity) (see below), (3) the often-arbitrary defini-

22

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

tion of functional units (e.g., “aromatic” may be hydrophobic or hydrophilic depending on substituents), (4) a neglect of functional unit mixing, and (5) the possible influence of hydration on partitioning in a given domain. In a review of the literature (Chefetz and Xing 2009), no significant correlations existed between either aromaticity or aliphaticity of NOM and sorption affinity for hydrophobic compounds. However, the authors emphasize that hydrophobic compounds do partition readily into alkane phases and therefore the polymethylene domains are probably important sorbents. 1.3.2.2. Functional-Group-Based Preferential Sorption. According to this hypothesis, molecules may interact with individual functional groups capable of specific or directional interactions. Nonideal behavior would result if the number of sites of each discrete type were limited. The case that first comes to mind is organocation exchange at negatively charged sites, which are inherenetly limited in number. So, too, compounds capable of specific weak interactions such as H-bonding or p–p EDA interactions with NOM may encounter a matrix with a limited number of complementary sites, and therefore could potentially show nonlinearity and competition. The relatively few available data available lead tentatively to the conclusion that nonlinearity seems not to require the existence of functional groups in the molecule that can undergo specific interactions; rather, nonlinearity seems to be due to a change in the nature of sites in NOM with loading. In their study of many polar and nonpolar compounds on Pahokee peat (45% OC), Borisover and Graber (2003) found no basis for predicting nonlinearity by the presence of strongly-interacting functional groups. Using Pahokee peat, a second high-organic soil, and a lignite, Zhu and Pignatello (2005b) observed concentration-dependent contribution of driving forces to sorption. However, the concentration dependence was relatively insensitive to solute structure. They found a general decrease in the contribution of the “dipolarity/polarizability” (D/P) term to log KOC with loading, and attributed it to preferential affinity for sites that are rich in functional groups with D/P influence, such as polar or aromatic groups. Endo et al. (2008a) came to similar conclusions in their study of polar and apolar compounds on a lignite and Pahokee peat. In both sorbents, H bonding played no role in nonlinearity. For lignite, the cause of nonlinear sorption was fairly compound-independent. For Pahokee peat, n was inversely linearly dependent on the S descriptor of Equation (1.12); they attributed this to a transition with increasing sorbate loading to sites of less aromatic and/or less polar character. 1.3.2.3. Heterogeneity Due to the Physical State of the Matrix—the Glassy Polymer Concept. In principle, macromolecular organic solids can exist in four interconvertible physical states—melted, rubbery, glassy, and crystalline (Eisenberg 1993). Matrix structure and dynamics play an

important role in the sorbent properties of synthetic polymers, and researchers, mainly of the author’s group and the group of Weber and coworkers, have postulated that this is true for NOM, as well. The polymer analogy for NOM has been made by many investigators. The most relevant states of solid NOM are the amorphous rubbery and glassy states. A critical property of amorphous polymers is the glass-torubber transition temperature (Tg), which correlates with the rigidity of the matrix. The rubbery state (T > Tg) has relatively high segmental flexibility of the macromolecules, allowing equilibrium to be achieved quickly following changes in an applied external condition such as temperature. The glassy state is the opposite. As a rubbery solid cools through the Tg to a temperature below the Tg, macromolecular rearrangements necessary for reaching the thermodynamic state require more time than is available by the imposed cooling rate, ultimately becoming prohibitive. As a result, excess free volume is frozen into the matrix in the form of poorly interconnected micropores or “holes.” The glassy state is, thus, a perpetual nonequilibrium state. Hong et al. (1996) confirm the existence of holes in glassy bisphenol A polycarbonate using positron annihilation lifetime spectroscopy (PALS) during the sorption of CO2. In the PALS technique positrons (b þ ) emitted from 22 Na interact with matter to form positronium atoms (Ps, electron þ positron), whose triplet state lifetime correlates with hole size, and whose intensity is a measure of free volume. Growing evidence indicates that NOM solids are—at least in part—glassy, a property that significantly affects their behavior as sorbents. Thermal transitions in heat capacity or coefficient of expansion that are characteristic of glassy synthetic and biopolymers have been observed for humic and fulvic acids, whole soils, shales, and coals, as well as for lignin, a major precursor of humic substances (DeLapp and Leboeuf 2004; LeBoeuf and Weber 1997; Schaumann and Antelmann 2000; Schaumann and LeBoeuf 2005; Zhang et al. 2007;) Lucht et al. 1987; Yun and Suuberg 1993). Based on solid-state wide-line 1 H NMR spectroscopy of dry humic and fulvic acids, a soil NOM, and a whole soil, Mao and Schmidt-Rohr (2006) found that segments undergoing fast, large-amplitude motions are a minor component of the sample, consistent with the postulated glassy character of NOM (Xing and Pignatello 1997). Glassy polymer theory has long been used to describe sorption of chemicals by coal (Milewska-Duda 1993). One important property of polymers that bears on the behavior of NOM is that the glass-to-rubber conversion takes place, not only with increasing temperature but also with increasing concentration of a penetrant. Technically known as plasticization, this is due to the softening effect resulting from progressive replacement of polymer–polymer interactions with penetrant–polymer interactions. Plasticization and its influence on sorption properties of polymers is well documented by the close correspondence between sorption and dilation isotherms (Fleming and Koros 1990). Evidence

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

Sorbed concentration

rubbery state

glassy state

“plasticization” to the rubbery state at the Sg

Pressure or dissolved concentration

Figure 1.8. Hypothetical sorption isotherms of a penetrant in identical glassy (solid-line curve) and rubbery (dashed-line curve) polymers. The Sg indicates the glass-to-rubber transition concentration. The dashed line below the Sg would correspond also to the thermodynamic state of the glassy polymer.

for swelling of humic substances during sorption is based on sedimentation volumetric changes in cosolvent–water mixtures (Lyon 1995) and “conditioning effects” (see Section 1.3.5). Pyridine—a known swelling agent of coals—promotes rapid macromolecular motions and expansion of void volume as shown by 1 H NMR studies of Argonne premium coals (Xiong and Maciel 2002). Sorption of PAHs by plant cuticular waxes coated on montmorillonite causes transitions from a less flexible to a more flexible amorphous state (Chen and Xing 2005). Figure 1.8 schematically presents the sorption isotherms for a penetrant in hypothetically identical polymers, except that one is rubbery (T > Tg; curved line) and the other is glassy (T < Tg; solid line) at the experimental temperature T. Sorption in the rubbery state obeys the Flory-Huggins equation (Chiou 2002) h i C w¼ exp
ðwpoly þ xw2poly Þ ð1:20Þ Csat where w is the volume fraction of penetrant in the solid, wpoly is the volume fraction of polymer, and x is the–polymer-penetrant interaction parameter. Strictly speaking, linearity is not characteristic of the rubbery state; while appearing linear over most of the range, the Flory–Huggins curve becomes nonlinear, concave-up at sufficiently high concentrations dependent on the value of x. The concave-up trend occurs because (1) plasticization reduces the cavity penalty for partitioning and (2) loading makes the solid more “penetrant-like” (i.e., drives the penetrant activity coefficient in the solid toward that of the pure liquid penetrant). An isotherm of chlorobenzene in rubbery polyethylene showing the concave-up bending at high concentration has been published (Sander et al. 2006). For the glassy polymer, the isotherm initially takes on a nonlinear, concave-down shape at low concentration (Fig. 1.9). Fit to the Freundlich equation in this range would

23

give an exponent n less than 1. At sufficiently high concentration, the solid is converted to the rubbery state through plasticization, and the isotherm inflects at the point where it joins the rubbery isotherm—the glass transition concentration Sg in Figure 1.8—and from there on, the isotherm would follow the Flory–Huggins equation. Figure 1.9 shows the isotherms of compounds in polymers and in OC-rich soils that show the inflection point and the reverse-S shape. As a result of this characteristic shape, the Freundlich n varies depending on the experimental concentration range over which it is calculated—typically, starting out near unity at the lowest concentrations, reaching a minimum at an intermediate concentration, and then approaching unity again at high concentration (Xia and Pignatello 2001). The isotherm of trans-1,2-dichlorocyclohexane in Figure 1.10 shows the difficulty of observing the characteristic shape unless the isotherm is constructed with many data points over a wide range in concentration (Xia and Pignatello 2001):, when the number of data is reduced from 64 to 32 or to 16, the resulting isotherm could easily be accepted as being “linear.” The proposed cause of the concave-down curvature at low concentration is the presence of holes due to incomplete relaxation relative to the thermodynamic state. These holes act as adsorption sites toward solute molecules. (Note that adsorption is not exactly accurate here, as within such a pore there can be no real distinction between containing the wall and not containing the wall.) “Holes” are defined as semipermanent cavities of subnanometer size resulting from the incomplete relaxation of the solid to its thermodynamic (rubbery) state that may lie within folds of individual macromolecules, between macromolecules, or between the organic phase and a mineral surface. This is illustrated in Figure 1.11 and is applicable to macromolecular organic matter solids— ordinary humic substances and ancient organic matter, kerogen, and soft coals. The model often used to describe sorption to glassy polymers is the dual-mode model (DMM), which combines the linear Freundlich and Langmuir isotherms: q ¼ KD C þ

QH KH C 1 þ KH C

ð1:21Þ

Here, the subscripts D and H refer to dissolution domain and hole domain, respectively. When the solid is a composite of rubbery and glassy materials, KD represents the sum of the solid–water partition coefficient for the rubbery-phase and that for the dissolution domain of the glassy phase. A conceptually similar model has been introduced by Weber and co-workers called the dual-reactive-domain model (DRDM) (Huang et al. 1997; Johnson et al. 2001; Weber et al. 1992; Young and Weber 1995). The DRDM combines linear partitioning into a “soft” state with nonlinear partitioning into a “condensed” state of NOM. Sorption occurs preferentially in the holes because unlike the dissolution domain it requires little or no cavity penalty .

24

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

(a)

1,2-DCB sorbed to PVC (mg/kg)

(b) 2e+5 10000

q (µg/g)

2e+5

5000

0

1e+5

0

5

10

5e+4

0 0

200

400

600

800

6x10 4 5000

0 0

5

10

2x10 4

1000 1200 1400

0

C (µg/mL)

10

20

30

1,2-DCB in solution (mg/L)

(c)

(d)

15000

DCCH sorbed (mg/kg)

2500

4x10 4

dual-mode fit below nmin

dual-mode fit to data below nmin

linear fit above nmin, thru origin

linear fit to data above nmin, thru 0

150000

10000

100000

5000

50000

0

0 0

100

200

300

DCCH in solution (mg/L)

0

1000

2000

DCP in solution (mg/L)

Figure 1.9. Isotherms of organic compounds in polymers and NOM (Pahokee peat soil) showing reverse-S shape. (a) Nitrobenzene in Tenax [Tg (glass transition temperature) ¼ 227 C]; (b)1,2dichlorobenzene in polyvinylchloride (Tg ¼ 85 C); (c) trans-1,2-dichlorocyclohexane in NOM; (d) 2,4-dichlorophenol in NOM. The nmin is close to the inflection point. [Data for nitrobenzene from Zhao and Pignatello (2004); all others from (Xia and Pignatello 2001).

The free energy of cavitation for a molecule the size of 1,2,4-trichlorobenzene or naphthalene in the dissolution domain of soil organic matter has been estimated to be 15–20 kJ/mol (Lu and Pignatello 2004a). Since holes are finite in number, the preference for holes predicts concave-down nonlinearity and a competitive effect when multiple solutes are present, behavior that is quite commonly observed. Polymers in the glassy state that trap a large amount of interconnected free volume are said to behave like microporous materials in many respects (Budd et al. 2005). In whole soils the unrelaxed free volume should be considered the sum of unrelaxed free volume within the NOM solid phase and that between the mineral surface and NOM molecular strands (Fig. 1.11). Equation 1.21 represents a simplification of reality: it assumes two discrete types of domains, a single hole energy, and does not reflect plasticization. The extended dual-mode model (EDMM) of Kamiya (Kamiya et al. 1992, 1986, 1998;

Wang et al. 1998) based on sorption and dilation isotherms of gases and hydrocarbons in polymers includes terms for progressive plasticization. As penetrant is loaded, the extra hole free volume is gradually eliminated through plasticization and the solid changes toward an equilibrium (rubbery) configuration. The cavity penalty is inversely dependent on the sorbed concentration, consistent with a plasticization effect (Zhu and Pignatello 2005b). The Tg of macromolecular solids increases with molecular weight, chain branching, degree of unsaturation, ring content, and interstrand crosslinking—properties that introduce chain stiffness and/or strand interconnectedness to the solid, and therefore reduce its flexibility. Dissolved NOM (DNOM) studied in its natural state or isolated from waters (Leenheer 2009) tends to consist of smaller, more readily hydrated molecules than those that are immobilized on geosolids. As those properties favor flexibility, DNOM aggregates or colloids are expected to be on the rubbery end of

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

(a)

64 data

DCCH sorbed (mg/kg)

15000

dual-mode fit below nmin linear fit above nmin, thru origin

dissolution domain

10000

5000

25

unrelaxed free volume (“hole” domain)

NOM

0 0

100

200

300

mineral

DCCH in solution (mg/L) (b)

15000 DCCH sorbed (mg/kg)

Figure 1.11. Artist’s conception of holes in natural organic matter associated with a mineral surface.

32 data (every second one)

10000 Kp = 57.7 2

r = 0.9797

5000

320

0 0

100

200

300

DCCH in solution (mg/L)

DCCH sorbed (mg/kg)

(c)

16 data (every fourth one)

15000

10000 Kp = 60.6 r2 = 0.9847

5000

218 0 0

100 200 300 DCCH in solution (mg/L)

Figure 1.10. Effect of isotherm detail on appearance of linearity. (a) Isotherm of trans-1,2-dichlorocyclohexane, the same as in Figure 1.7, consisting of 64 data; (b,c) successive removal of data and fit to linear isotherm (number at bottom of graph is the intercept).

the scale. Likewise, humic and fulvic acids extracted from soil tend to be the smaller, more soluble molecules of SOM favored by the commonly used extraction techniques. Nevertheless, sorption isotherms of apolar compounds to solid soil humic acid particles (Lu and Pignatello 2004a,b; Pan et al. 2007) are nonlinear, show competition, and are hysteretic (Lu and Pignatello 2004b; Pignatello et al. 2006b; Sander et al. 2006)—all of which are consistent

with solid humic acid existing in the glassy state. Crosslinking of humic acid with trivalent metal ions increases the Tg slightly (Pignatello et al. 2006b) and increases sorption nonlinearity and the contribution of hole filling to total sorption (Lu and Pignatello 2004b). Figure 1.12 illustrates these effects for 1,2,4-trichlorobenzene sorption in H þ exchanged soil humic acid (H-HA), Al-exchanged HA (Al-HA), and a brown coal [Beulab–Zap lignite (BZL) (Lu and Pignatello 2004a). Figure 1.12 shows the isotherms [plotted as the sorption distribution ratio Kd(L/kg) vs.the reduced concentration] fit to the DMM [Eq (1.21)]; the ratio QHKH/KD, which corresponds to the ratio of concentrations in the hole domain to the dissolution domain in the limit of infinite dilution; and a plot of the fraction sorbed in the hole domain versus reduced concentration. Notice that fit to the DMM captures the inflection of the isotherm. The collective results of thermal and thermomechanical analysis of various samples (DeLapp and Leboeuf 2004; Schaumann and Antelmann 2000; Schaumann and LeBoeuf 2005; Zhang et al. 2007; Lucht et al. 1987; Yun and Suuberg 1993) indicates that Tg increases as a function of diagenetic or geothermal alteration (Zhang et al. 2007) in the following order (rough range in Tg): Fulvic acid particles < humic acid particles (30 C–70 C) 150 C) Soil organic matter (i.e., recently deposited material in nearsurface horizon soil samples) seems to fall between humic acid and types I and II kerogens (DeLapp and Leboeuf 2004; Schaumann and Antelmann 2000; Schaumann and LeBoeuf 2005). Sorption of apolar compounds in soft coal (BZL) is greater overall and has a larger hole-filling component than does sorption in soil humic acid (Fig. 1.12). It is known that moisture greatly suppresses adsorption of organic vapors to ordinary soils, since water competes strongly with organic compounds for mineral surfaces, which

26

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS 10 5

6

both donor and acceptor properties) give the opposite behavior, while monopolar compounds show nonuniform behavior (Niederer et al. 2006a). Using a concentrationspecific FER, Zhu and Pignatello (2005b) found that, with increasing loading from aqueous solution, organic molecules fill sites in NOM of progressively greater hydrophilic character. These observations may be explained as follows. Water initially has an anti-plasticizing effect on NOM (increase in Tg) due to crosslinking of polar functional groups, and then a plasticizing effect above 12% moisture (Schaumann and LeBoeuf 2005). Water competes for adsorption space of organic compounds within glassy matrices. Hence, the decrease in sorption with increasing moisture content may be due to the progressive transition from a more to a less glassy material, coupled with the competitive effect of water. In the case of bipolar compounds, the extra H bonding afforded by the presence of water molecules may overcome the competitive and plasticizing effects of water.

4

1.3.3. Sorbent Properties of Black Carbon

TCB

Kd, L/kg

10 4

10 3

10 2 10 -5

10 -4

10 -3

10 -2

10 -1

C/Sw 8

(QHKH)/KD

non-conditioned conditioned

2

0 H-HA

Al-HA

BZL

1.0 TCB

Fraction in Hole-domain

BZL 0.8 Al-HA 0.6 H-HA 0.4

0.2

0.0

-5

10

-4

10

-3

10

C/Sw

-2

10

-1

10

0

10

Figure 1.12. Sorption of 1,2,4-trichlorobenzene in H þ -exchanged soil humic acid (H-HA), the Al-exchanged HA (Al-HA), and a brown coal [Beulah–Zap lignite (BZL)] before and after “conditioning.” Fit is to the dual-mode model [see Eq. (1.21)]. Conditioning effect is discussed in Section 1.3.5.

are usually abundant in soils (Chiou 2002). Water, however, has a more modest effect on sorption to NOM, which can absorb as much as one-fourth its weight in water. Apolar compounds typically experience up to three fold suppression of sorption from the dry state to the fully humid (98% relative humidity) or water-saturated state (Niederer et al. 2006a; Rutherford and Chiou 1992). Bipolar compounds (those with

As mentioned, sorption to raw BC is highly nonlinear and competitive, more so than NOM. Nonlinearity can be so severe that the distribution coefficient can range over many orders of magnitude. For example, the Freundlich n of the neutral form of sulfamethazine is 0.27 and so the KBC varies by 3 orders of magnitude from 1 mg L
1 (near the limit of quantification) to 30 mg L
1 (about 6.7% of water solubility) (Teixido´ Planes and Pignatello unpublished). Nonlinearity is due to a combination of surface site heterogeneity and pore size heterogeneity. The distribution of adsorption potentials on a graphene surface can be pronounced even on what would appear to be a smooth, homogeneous surface; for example, the Freundlich n values for sorption of benzene and a number of other apolar aromatic compounds from water onto nonporous graphite (99.95% C; BET specific surface area, 4.5 m2/g) are below 0.60 (Zhu and Pignatello 2005a). Graphite has at least three different kinds of sites: basal plane, defect sites along the basal plane (pits), and edge sites along step elevations. At higher concentrations, adsorbate–adsorbate interactions are possible as is confirmed by atomic force microscopy for substituted benzoic acids (Martin 2003). A detailed isotherm of benzene sorption from water to a wood charcoal over the range in concentration from 1.8  10
7 times to 0.6 times water solubility is shown in Figure 1.13 (Braida et al. 2003). The isotherm is highly nonlinear, and, owing to its reverse-S shape, does not fit any of the conventional isotherm models in section 1.3.1. The log BC–water distribution ratio ranges from a maximum of 4 to a minimum of 2.7; these values are, respectively, about 200 and about 10 times greater than KOC for benzene calculated on the basis of an OC–octanol FER. Note that the Freundlich n for benzene approaches 1 at low concentrations but declines as the range over which it is calculated increases.

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

1e+6

q(μg/gsoot)

log K BC = 2.7

Experimental Freundlich Fit Langmuir Fit

1e+5

27

1e+4

1e+3

log KBC = 4

Freundlich n

1e+2

0.706 0.825

1e+1

0.759

0.903

1e+0 0.00010.001

0.01

0.1

1

10

100

1000

Benzene dissolve conc. (g/mL)

Figure 1.13. Sorption isotherm of benzene on charcoal powder suspended in water. The highest concentration is about 60% of water solubility. Horizontal arrows indicate the range over which the Freundlich n is calculated. KBC (L/kg) is the charcoal-water distribution ratio. [After Braida et al. (2003).]

Adsorption to BC is strongly affected by the polarity of the surface. Offgassing the polar C by catalytic reduction in a stream of H2 hardly affected the surface area, but enhanced adsorption on a surface area basis, regardless of the polar or nonpolar nature of the adsorbate (Zhu et al. 2005). An inverse linear relationship exists between the oxygen content of multiwalled carbon nanotubes and their maximum adsorption capacity for naphthalene (Cho et al. 2008). Black carbon undergoes oxidation as it weathers in soil. As proposed for AC (M€ uller and Gubbins 1998; M€ uller et al. 2000), polar groups on micrographitic sorbents apparently exert a “crowding out” effect resulting from the clustering of water molecules around polar sites that limits the available adsorption surface for the organic compound. Grand canonical Monte Carlo simulations of water competition with methane are consistent with this hypothesis (Fig. 1.14) and also show that competition diminishes as pore size increases (M€ uller et al. 2000). The surface activity of raw BC can be greatly attenuated by the presence of other substances, including, (1) incompletely carbonized biomass polymer fragments, (2) unburned liquid fuel or fuel byproducts, and (3) natural substances in the environment. For a given biomass source material, the specific surface area of the char varies with pyrolysis temperature, ramping steeply above 300 C (Fig. 1.15). Sorption trends with specific surface area, although this correlation is not very useful because it is complicated by pore size distribution and surface chemistry (Jonker and Koelmans 2002b), and there are conceptual issues with the customary use of N2 as at 77 K as the probe gas for surface

Figure 1.14. Equilibrium configurations for the adsorption of methane and water in a carbon slit pore of width 1 nm with varying active-site densities at 300 K (M€ uller et al. 2000). Methane, red balls; water, blue balls. Gray asterisks represent active sites on surface in foreground or background, depending on their size. (See insert for color representation of this figure.)

area (Kwon and Pignatello 2005; Pignatello et al. 2006a). The uncarbonized material is a weaker sorbent than the carbonized material. Semivolatile oil-like and tar-like materials can condense in the pores of soot particles during synthesis and/or cooling (Akhter et al. 1985b). Semivolatiles can also condense on charcoal if the offgases do not vent during charring. Aerosol BC particles can accumulate secondary photolysis products of semivolatile organic compounds in the atmosphere that subsequently affect sorption (Dachs and Eisenreich 2000; Kamens et al. 1995; Strommen and Kamens 1997). On weathering in the soil/sediment environment, the ability of BC particles to sorb organic compounds is attenuated by humic substances, metal ions, and possibly metal sesquioxides that may adsorb or coat their surfaces. Dissolved humic substances are well-known to foul activatedcarbon adsorbents used in the water treatment industry (Kilduff and Wigton 1999; Li et al. 2003; Newcombe et al. 1997). Field evidence supporting a weathering effect on BC include the finding of weaker sorption than expected on the basis of raw BC (up to nine fold) for PAHs and PCBs

28

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

N2B.E.T. Surface Area

(a)

100

400 qEC, mmol/kgEC

Specific surface area, m2/g

500

300 200 100

10 qEC,char = 103.72 Cw0.38 1 qEC,cofloc = 10 4.12 Cw0.66 0.1 char alone Al-HA-char co-floc

0

0.01 10-9 200

300

400 T, oC

500

600

700

Figure 1.15. Effect of pyrolysis temperature on the specific surface area of charred maple wood shavings. The SSA is based on the N2 probe gas adsorption isotherm at 77 K and calculated by the Brunauer–Emmett–Teller equation (Brunauer et al. 1938).

in sediment believed to be rich in BC (Cornelissen and Gustafsson 2004; Jonker et al. 2004). Direct evidence for attentuation of BC surface activity by natural substances has been obtained (Koelmans et al. 2009; Kwon and Pignatello 2005; Pignatello et al. 2006b; Wen et al. 2009). [Negative evidence is also reported (Cornelissen and Gustafsson 2006).] For example, wood char particles placed in a soil–water suspension experienced after a few weeks a 13-fold decline in the N2–BET-specific surface area and about a three-fold decline in benzene sorption coefficient. Loading of humic acid, fulvic acid or triglycerides (as surrogates for the humic lipid fraction) from the aqueous phase resulted in up to two-orders-of-magnitude reduction in char-specific surface area and a modest suppression of naphthalene and phenanthrene sorption (after correcting for partitioning to DNOM colloids) (Pignatello et al. 2006b). Sorption suppression increased with solute molecular size, consistent with competition between humic substances and the solute on external surfaces. The greatest suppression—an order of magnitude or more for phenanthrene—is found for samples of char that had been coflocculated with humic acid by Al3 þ (Fig. 1.16a) or merely coated with humic acid (Figure 1.16b). Tannic acid, a model for humic substances, decreased surface area and sorption of hydrophobic compounds (Qiu et al. 2009). It is also known that metal ions compete with nonionic compounds on the surfaces of both BC (Chen et al. 2007a) and carbon nanotubes (Chen et al. 2009); in these cases, it is believed that the adsorbed metal ion and its hydration sphere “crowd out” the organic. In summary, these results collectively suggest that NOM, mineral–NOM complexes, and metal ions play a role in sorption suppression by competition for surface area and/or by pore blockage.

10-7

10-6

10-5

10-4

C w, mmol/L

(b) 10000 Sorbed concentration (mg/kg)

100

10-8

100 1E-3

0.01

0.1

Equilibrium concentration (mg/L)

Figure 1.16. (a) Strong suppression of phenanthrene sorption by coating black carbon with humic substances: elemental carbon– normalized isotherms in wood charcoal and in the complex obtained by adding Al ion to a suspension of black carbon in a soil humic acid solution (Al-HA-char co-floc). The EC-normalized isotherm of the latter is obtained after subtracting out the OC contribution on the the basis of Freundlich parameters of the AlHA co-floc.[From Pignatello et al. (2006a)]. (b) Freundlich fit of phenanthrene isotherm by a char (&), HA (.), lipid (~), and chars coated with HA at low (&), medium (*), or high (~) levels, and char coated with lipid at low (!), medium (x), or high ( ) levels. [From Wen et al. (2009).]

1.3.4. Competitive Sorption Contaminants in the environment more often exist in mixtures than singly. Competition is treated here as a sorbentcentered phenomenon because it impacts the sorbent reorganization free-energy term (step 2, equation 1.1), creating a negative driving force for sorption of the principal solute. Put simply, competition between the principal solute and a cosolute will be experienced when sites are distributed in energy and sites of a given energy are limited in number. Nonlinearity, after all, is simply a manifestation of selfcompetition due to site energy distribution. Competitive

29

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

QL1 KL1 C1 P q1 ¼ 1 þ KL1 C1 þ n2 KLn Cn

ð1:22Þ

Sander and Pignatello (2005a) showed that benzene and toluene compete for the same sites on a charcoal, and that, when normalized for the hydrophobic effect via the hexadecane–water partition coefficient, the single-solute benzene and toluene data, together with the bisolute data (qben þ qtol, molar basis), fall onto a single isotherm (Fig. 1.17b). Hence, benzene and toluene are “ideal” competitors and replace each other on the charcoal surface in a 1 : 1 molar ratio. Nitrobenzene was also found to be an ideal competitor for both toluene and benzene on the same char (Sander and Pignatello 2005a). All three competition models mentioned above predict the greater the competitive effect, the more linear the isotherm of the principal solute will be [apparent by inspecting Eqs. (1.22) and (1.23)]. This is generally confirmed experi-

(a)

10 1

10 0 q [mol kg-1]

sorption in soils and isolated humic substances has been noted by many investigators. Competitive effects among PAHs for BC is argued to account for the lower-than-expected sorption of individual PAHs in sediments (Cornelissen and Gustafsson 2006). Competition has also been observed between contaminants and small natural molecules such as aromatic acid plant exudates in soils (Xing and Pignatello 1998), and between contaminants and dissolved humic acid on BC (Koelmans et al. 2009; Pignatello et al. 2006a). Competition has also been shown to increase the bioavailability of the principal solute in soils (White et al. 1999a) and is expected to play an even greater role in the bioavailability of chemicals sorbed to BC because nonlinearity is typically more pronounced than in NOM (Pignatello 2009). Competition may also occur between contaminants and unburned liquids accompanying soot particles; for example, phenanthrene sorption was weaker and more linear in a reference diesel soot having the greater amount of native extractable oils (20% vs. 2% by weight) (Nguyen and Ball 2006). Competition for sorption of a solute 1 is manifested in the Langmuir model by the sum of terms in the denominator corresponding to solutes 2 through n:

10 -1

-1

CW_0(TOL)= 0.00 mol L -1 CW_0(TOL)= 5.24E-6 mol L -1 CW_0(TOL)= 1.11E-5 mol L CW_0(TOL)= 2.30E-5 mol L-1 -1 CW_0(TOL)= 4.88E-5 mol L CW_0(TOL)= 1.07E-4 mol L-1

10 -2

Assuming that each site of a discrete energy is limited in number, a competitive Freundlich isotherm can be derived (Fritz and Schlunder 1981; Sheindorf et al. 1981), !n1
1 n X q1 ¼ KF1 C1 C1 þ a1n Cn ð1:23Þ

10 -3

benzene competitor: toluene

10

CW_0(TOL)= 4.23E-3 mol L

-4

10 -8

10 -7

10 -6

10 -5

10 -4

10 -3

-1

10 -2

Cw[mol L-1]

2

(b) 10 1 Benzene Toluene

q(BEN) + q(TOL) [mol kg-1]

where a1n represents the competition coefficient between solute 1 and solute n. In a bisolute system, the competition coefficients can be obtained by regressing a linearized form of equation 23 (not shown). In theory, a12 ¼ 1/a21 in a bisolute system. The model has been tested successfully on carbon adsorbents (Fritz and Schlunder 1981; Sheindorf et al. 1981). Another important competitive model is ideal adsorbed solution theory (IAST), which is derived from the Gibbs equation. This theory has been applied successfully to multisolute sorption on activated carbon (Crittenden et al. 1985; Radke and Prausnitz 1972) and soil (e.g., McGinley et al. 1989; Xing et al. 1996). The main assumption of IAST is that in dilute conditions the adsorbed phase forms a twodimensional ideal solution, such that the fugacity of each adsorbed component in a mixture is proportional to its adsorbed-phase mole fraction times its fugacity in the single-solute system at the same spreading pressure. Spreading pressure is defined as the difference in interfacial tension between pure solvent–solid and solution–solid systems. The advantage of IAST is that, in principle, it is adaptable to any isotherm model, although the mathematics get quite complicated for all except the simplest models.

CW_0(TOL)= 2.47E-4 mol L -1 -1 CW_0(TOL)= 6.18E-4 mol L -1 CW_0(TOL)= 1.57E-3 mol L

10 0

CW_0 (TOL)=8.24E-6 mol L

-1

CW_0 (TOL)=1.11E-5 mol L

-1

CW_0 (TOL)=2.30E-5 mol L

-1

CW_0 (TOL)=4.88E-5 mol L

10 -1 -1

CW_0 (TOL)=1.07E-4 mol L

-1

CW_0 (TOL)=2.47E-4 mol L

10 -2

-1

CW_0 (TOL)=6.18E-4 mol L

-1

CW_0 (TOL)=1.57E-3 mol L

-1

10 -3 10 -5

CW_0 (TOL)=4.23E-3 mol L

10 -4

10 -3

10 -2

10 -1

10 0

10 1

-1 CH(Ben)+CH'(TOL) [mol LH ]

Figure 1.17. (a) Sorption isotherm of benzene in the benzene–toluene bisolute system. CW_0(TOL) is the time zero molar concentration of toluene. (b) Sorption data plotted as the sum of sorbed concentrations of the two solutes versus the sum of the corrected solution concentrations. Corrected solution concentration is the equivalent concentration in n-hexadecane, which was used as a reference solvent to normalize for hydrophobic effects (isotherms also overlapped when are used benzene as the reference solvent).

30

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

mentally (e.g., McGinley et al. 1989; Sander and Pignatello 2005a, 2007) (see Fig. 1.17a). Competition-forced linearity has three major origins: 1. At a given co-solute concentration, the co-solute becomes less and less effective as a competitor with increasing principal solute concentration. This has the effect of increasing the slope of the principal solute’s isotherm on a log scale. 2. When both a dissolution phase and an adsorbing phase are present, suppression of adsorption occurs to the benefit of dissolution-type sorption. 3. When the sorbing phase is glassy, the cosolute competitively fills holes and, at sufficiently high concentration, plasticizes the solid to the rubbery state, which exhibits more linear behavior. An assumption common to all three competition models is that the respective sorption domains fully overlap; that is, all sites are available to all molecules, albeit with unequal affinity. If this condition is not met, the respective competition isotherms cannot be constructed from a simple combination of the single-solute isotherms. This hypothesis has not been fully tested, but some results are supportive. Using IAST, Xing et al. (1996) modeled competition between atrazine and either a triazine analog, prometon, or the structurally unrelated compound, trichloroethene, on a high-organic soil, humic acid particles, the commercial glassy polymer poly(2,6-diphenyl-p-phenylene oxide) known as Tenax, and amorphous silica gel as shown below. They found that, whereas IAST simulated competition by prometon satisfactorily, it greatly overpredicted competition by trichloroethene. Whether for steric or electronic reasons, the shared sorption domain of atrazine and trichloroethene on these sorbents is smaller than that of atrazine and promoton. The importance of steric effects in sorption to NOM is suggested by the inverse relationship between the competitive effect and the difference in molecular free surface area between competing halogenated aliphatic compounds (Pignatello 1991). Moreover, whereas benzene shares the same sorption domain with nitrobenzene and toluene on a charcoal, the size exclusion effect on this adsorbent (Section 1.2.2.6 and Fig. 1.5) clearly prevents

benzene and naphthalene or benzene and phenanthrene from fully sharing the same sorption domain. Schaefer et al. (2000) suggested that solute polarity and/or sorbent geometry may have contributed to the failure of IAST to simulate competition between trichloroethene and tetrachloroethene on natural solids. Adsorption of hydrophobic compounds by BC is competitively suppressed by water (Endo et al. 2009a). Water molecules compete for surface area by interacting with the graphene surface through dispersion forces and especially by clustering around polar functional groups along the platelet rims and defect sites. Finally, competition can lead to apparent hyteresis if the cosolute concentration is altered in the desorption step (see Section 1.3.5). 1.3.5. Hysteresis Hysteresis means the lagging of a physical effect on a body behind its cause (Merriam-Webster 2000). The phenomenon of sorption hysteresis is widely encountered. It results when chemical released by dilution of the solute in the fluid phase is less than predicted by the isotherm constructed in the “forward” (uptake) direction. Hysteresis may be true or artificial. True, and in some cases artificial hysteresis has important implications for environmental behavior. Among its consequences are lower bioavailability, greater “tailing” of a contaminant plume traveling through porous medium, and stronger resistance toward remediation than predicted assuming reversibility. 1.3.5.1. Artificial and True Hysteresis. Artifacts fall into three categories: degradation, insufficient time allowed for mass transfer to reach a static condition, and perturbation of a competitive situation. If in an experiment sorbed concentration is calculated by mass balance rather than measured directly, uncorrected chemical or biological transformation of the parent compound will lead to overestimation of the actual sorbed concentration. Mass transfer limitations can result in underestimation of the equilibrium sorbed concentration in the uptake step, overestimation of the equilibrium sorbed concentration in the release step, or both. Such artifact has been documented in experiments where the contact time was deliberately shortened (Altfelder et al. 2000), and underscores the importance of taking mass

OCH3

Cl N CH3CH2HN

N N

Atrazine

N NHCH(CH 3)2

(CH3)2CHNH

N N

Prometon

Cl

Cl

H

Cl

NHCH(CH3)2 Trichloroethene

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

transfer rates into account in models that include sorption terms. The plasticizing effect of moisture on NOM takes a number of days to complete (Schaumann and LeBoeuf 2005), so artificial hysteresis can occur if sufficient hydration time is not allowed. A third type of artificial hysteresis is known as the competitor dilution effect (Sander and Pignatello 2007). The competing substance may be a solute or a nonseparable microsorbent in the liquid phase. Examples of a competing microsorbent include colloidal NOM (Curl and Keolelan 1984) and surfactant micelles. In any case, because competitive pressure is partially relieved when the competitor concentration is reduced in the dilution step, the principal solute sorbs more strongly than predicted by the uptake branch of the isotherm. Sander and Pignatello (2007) showed that pairwise sorption to a charcoal of benzene, nitrobenzene, and toluene resulted in much greater hysteresis in the biosolute case than in the corresponding single-solute case owing to the competitor dilution effect. The competitor dilution effect—while artificial in the sense that it is a secondary effect and competitor concentration is a controlled variable—has real implications because competing solutes and nonseparable microsorbents are more often than not a part of natural and disturbed environments. The underlying cause of true hysteresis is the formation of a metastable state or states during the sorption process. True hysteresis may be distinguished from artificial hysteresis caused by mass transfer limitation because the metastable state persists indefinitely in the absence of a perturbing force, like the book balanced on its edge. Sorption in such cases is deemed “irreversible” in the thermodynamic context of the word in that sorption does not follow the same pathway in the forward and reverse directions. (“Irreversible” should not be confused with its more common meaning, that the chemical cannot be retrieved without drastic means.) The existence of different pathways in the forward and reverse directions implies that sorption is dependent on the history of the system. (Interestingly, the words hysteresis and history are not related etymologically.) At least two types of true hysteresis have been identified for organic compound sorption. One is capillary condensation hysteresis (Fig. 1.18a) (Rouquerol et al. 1999). This occurs when vapors condense in a fixed mesopore to form a metastable film that collapses at a threshold pressure to form the thermodynamic meniscus plug state, which has a lower vapor pressure. It gives rise to the classic hysteresis loop observed at high relative pressures. The desorption branch represents the true equilibrium path. The other (Fig. 1.18b) is referred to as pore deformation hysteresis and occurs with sorbents whose matrix is sufficiently flexible to undergo physical changes by the incoming solute. At ordinary temperatures deformable sorbents typically include organic materials such as polymers, NOM, and possibly carbons.

31

1.3.5.2. True Hysteresis in Natural Organic Matter. First applied to glassy polymers, the concept of pore deformation hysteresis holds that permeant molecules exert pressure on internal pores or proto-pores smaller than the adsorbate molecular volume, causing pore expansion or creation (Fig. 1.18b). At the same time, the local matrix is softened by interaction of permeant with segments, increasing their mobility. On desorption from the local matrix the process attempts to reverse, but is unable to do so fully because the matrix stiffens before relaxation of the expanded or created pores is completed. Thus, swelling and shrinking is inelastic (irreversible) and leaves the local matrix with greater excess free volume after the cycle than before that persists indefinitely. The greater excess free volume leads to higher affininity for the solute in a subsequent desorption or repeat sorption step, manifesting as sorption hysteresis. For uptake of CO2 by polycarbonate, positron annihilation lifetime spectroscopy showed that the polymer is left with a greater hole volume and hole size after a sorption–desorption cycle than in the original sample (Hong et al. 1996).) Swelling of flexible matrices like gels and rubbers, on the other hand, is fully elastic (reversible) and shows no sorption hysteresis. In thermodynamic terms, during pore deformation hystersis in a glassy material. part of the free energy of sorption in a glassy material goes into matrix expansion that is not fully recovered during the cycle. The thermodynamic index of irreversibility (TII) via pore deformation mechanism may be defined in terms of the hysteresis loss in solute chemical potential over the cycle (Sander et al. 2005): TII ¼

lnCc
lnCD lnC S
lnC D

ð1:24Þ

where, respectively, CS, CD, and Cc represent solute concentration at the observed sorption point, observed desorption point, and the hypothetical reversible desorption point corresponding to the same sorbed concentration as at the observed desorption point (i.e., where qc ¼ qD). Considered another way, TII is the ratio of the observed to the upper-limit loss of free energy due to irreversibility. The TII is 0 for completely reversible systems and approaches 1 as mass desorbed approaches zero. Pore deformation hysteresis does not occur in melts, gels, or rubbery solids, nor in rigid, fixed-pore solids whose pores are not strained beyond their elastic limit (Bailey et al. 1971). These endpoints suggest a trend, illustrated in Figure 1.18c, in which the hysteresis potential of a solid matrix capable hypothetically of varying continuously between the rubbery and fixed-pore states, while holding all else constant, would reach a peak somewhere in between. This trend predicts that solids lying on the rubber side of the apex would show decreasing hysteresis with increasing solute concentration, while solids lying on the fixed-pore side of the apex would

32

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

(a) q

p/p

0

sorption

desorption

(b)

(reversible if “rubbery”) +

swelling

sorbate carbonaceous solid

(irreversible if “glassy”) shrinking

Part of Gsorp goes into matrix expansion

+

hysteretic effect

(c)

increasing matrix stiffness rubbery solid

glassy solid

fixed-pore solid

Figure 1.18. True hysteresis of organic compounds. (a) Capillary condensation hysteresis in a mesopore and the resulting hysteresis loop. The condensate film is metastable and collapses at a threshold pressure to the meniscus plug. The desorption branch represents the thermodynamic isotherm. (b) Schematic illustrating pore deformation hysteresis during a sorption–desorption cycle in a glassy solid. Sorption causes swelling that is reversible in very flexible (rubbery) solids and incompletely reversible (irreversible) in stiffer solids. (c) Schematic showing that the magnitude of the hysteretic effect in a solid whose matrix is able to vary continuously between the rubbery and very rigid (fixed-pore) states reaches a peak.

show the opposite. Sander found trends consistent with this prediction (Sander and Pignatello 2009). True hysteresis has been confirmed for NOM-rich solids by isotope tracer exchange experiments performed on two sorbents and two compounds, 1,4-dichlorobenzene, and naphthalene (Sander and Pignatello 2005b, 2009). Following equilibration at each step of a normal–sorption-desorption cycle, a slight amount of the 14 C-labeled chemical (tracer) is added and equilibrated under the same conditions, keeping the bulk chemical concentration constant. Such experiments showed that, while sorption of the bulk chemical is hysteretic

(depending on its concentration), tracer exchange was complete and nonhysteretic under all conditions. Thus, hysteresis is due to changes in the sorbent induced by uptake by the bulk chemical. This is further supported by differences in rates between bulk chemical and tracer in both the uptake and release directions (Sander and Pignatello 2005b, 2009). Convincing evidence for the pore deformation mechanism in NOM has been obtained in the so-called conditioning effect experiments (Lu and Pignatello 2002, 2004a,b; Sander et al. 2006; Xia and Pignatello 2001), where a sample of humic acid, whole SOM, or coal is “conditioned” by

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

subjecting it to a swelling–shrinking cycle performed by uptake–release of an organic solute. The response isotherm of a test compound—either the same or a similar compound—is constructed before and after conditioning. Such experiments show enhanced second-time sorption, consistent with irreversible free volume expansion by the conditioning agent. Figure 1.12 shows overall sorption enhancement and an increase in the contribution of hole filling relative to dissolution sorption as a result of conditioning in three organic matter materials, including a humic acid. Figure 1.19 shows sorption enhancement in a peat soil. As mentioned, the incompletely relaxed system represents a persistent metastable condition. Sorption enhancement through conditioning persists for months if the solid is stored at room temperature (Lu and Pignatello 2002). If the solid is heated (annealed), the conditioning effect diminishes in relation to the annealing temperature and time (Sander et al. 2006) (see Fig. 1.19). As in polymer systems, relaxation at a given temperature follows a double exponential rate law with a nonzero constant term descriptive of the final state that itself varies inversely with temperature. This means that at temperatures below the glass transition temperature, annealing does not fully relax the solid even after long annealing times. 1.3.5.3. Hysteresis in Black Carbon. True hysteresis may also be characteristic of carbons, although more study is needed to confirm it. Bailey et al. (1971) observed hysteresis of nonpolar organic vapor sorption by activated carbon; terming it “low-pressure hysteresis” to distinguish it from capillary condensation hysteresis in mesopores at higher pressures, they attributed it to “intercalation of molecules of adsorbate in narrow pore spaces leading to irreversible

changes in the pore structure.” Environmental BC clearly can be swelled by organic solvents (Akhter et al. 1985b; Braida et al. 2003; Razouk et al. 1968), confirming that BC platelets are held together in part by noncovalent forces. Swelling of wood charcoal by uptake of benzene dissolved in water affected the shape of the isotherm, and was likely responsible for benzene adsorption hysteresis (Braida et al. 2003; Sander and Pignatello 2007). Swelling may open up new sectors previously not connected with the external fluid. On desorption, some molecules become trapped as the polyaromatic scaffold collapses during desorption. The ability of some solvents to extract PAHs and other compounds from BC has been attributed to swelling (Akhter et al. 1985b; Jonker and Koelmans 2002a). Although NMR studies show an increase in molecular motion of sorbed 13 C-benzene with its concentration in charcoal (Smernik et al. 2006), other causes are possible. Finally, LeBoeuf and co-workers (Zhang et al. 2007) report thermal transitions indicative of a glass transition for several chars at 140 C–160 C, but interestingly not for diesel and hexane soot up to 200 C. 1.3.5.4. Hysteresis in Related Model Systems. Studying plant cuticular waxes deposited on montmorillonite Chen and Xing (2005) found sorption hysteresis of naphthalene and phenanthrene above but not below the approximate point where the sorbate induces a phase transition from a more to a less flexible amorphous state. This behavior is the opposite that observed for NOM. These authors proposed that hysteresis is due to “trapping” of molecules caused by collapse of the flexible amorphous state to the less flexible state on desorption.

Distribution ratio (sorbed divided by solution concentrations, L/kg)

8000

104

6000

20 oC, 21 d o

50 C, 6h o 60 C, 6h 75 oC, 6h 90 oC, 6h

4000

103 2000

10 -3

10 -2

10 -1

33

10 0

10 1

0.01

0.02

-1 Solution concentration, mg L

Figure 1.19. Sorption isotherm of 1,2,4-trichlorobenzene at 20 C on Pahokee peat before (filled circles) and after (open triangles) conditioning with chlorobenzene. The dashed curves are fits to the dual-mode model [Eq. (1.21)]. The inset shows the effects of storage at 20 C for 21 days and heating (annealing) at different temperatures for 6 h. It also shows a tendency for the distribution ratio to “relax” to its original isotherm after annealing. [After Sander et al. (2006).]

34

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Sorption of organic compounds by expanding clays may exhibit sorption hysteresis if sorption induces hysteretic swelling or structural reorganization of the quasicrystals (Chatterjee et al. 2008).

1.3.6. Apportionment of Sorption between NOM and BC in Environmental Samples Soil and sediment organic matter may contain a mixture of NOM and BC. While BC may be a minor component (estimates place it at 1%–10% of total organic carbon in undisturbed soils, higher at industrial sites and in marine sediments), BC in its raw state is a much stronger sorbent than NOM by an order of magnitude or more, and therefore may play a role disproportionate to its mass fraction. Researchers have tried to partition sorption in whole soil or sediment by using multiterm equations that account for contributions by NOM and BC (Accardi-Dey and Gschwend 2002, 2003; Allen-King et al. 2002; Xia and Ball 1999). In some cases coal and NOM are distinguished (Cornelissen and Gustafson 2005). The models use a linear “partitioning” term for NOM and either Freundlich or Polanyi–Manes terms for the other carbonaceous sorbents, as, for example, the following two-compartment isotherm qT ¼ fOC KOC C þ fBC KBC CnBC

ð1:25Þ

where qT is the total observed mass-based sorption, OC is NOM carbon, BC is BC carbon, and f is the mass fraction of each type of carbon in the sample. It is important to understand, however, that this model rests a number of assumptions that are questionable: 1. As yet, there is no generally accepted method for quantifying fBC in soils and sediments (Hammes et al. 2007, 2008), especially at low BC levels. A fundamental analytical problem is that “BC” is not a single material but a continuum of materials with different properties, some of which overlap NOM. In fact, raw BC often contains a nontrivial amount of sp3 C and nonaromatic sp2 C from functional groups on graphene rims and from incompletely carbonized precursors in pores (Schmidt and Noack 2000). Thus, a single analytical method may sample only a window of the BC continuum. The most commonly used of the various methods proposed for quantifying BC is chemical–thermal–oxidation (CTO375) method—the sample is first acidified to remove inorganic carbon, then combusted at 375 C in air to volatize NOM, and finally the BC carbon that is assumed to have survived is then quantified by conventional hightemperature combustion (Gustafsson et al. 1997; Nguyen et al. 2004). The CTO-defined fBC must be regarded as operational, however, since studies show that low-temperature combustion also volatilizes most or all of the char and a large

portion of the soot (Nguyen et al. 2004). This means that fBC is likely to be underestimated in this test. Hawthorne et al. (2007) were unable to explain PAH partitioning in 114 historically contaminated and background sediments by a linear KOC–KBC dual model on the basis of the CTO-375 method. Another method that has been attempted is the acid–dichromate method [See Grossman and Ghosh (2009), Knicker et al. (2007), Pignatello et al. (2006b), and references cited therein]. However, acid–dichromate does not oxidize polymethylene groups (Knicker et al. 2007; Pignatello et al. 2006b), which can constitute several percent of NOM; thus, acid–dichromate will likely overestimate fBC. The contribution of BC to total sorption will be overestimated proportionate to the degree fBC is underestimated, and vice versa. Likewise, coal contains substances overlapping in properties with both “ordinary” NOM and BC. So, while there are petrographic methods to identify BC and coal (Karapanagioti et al. 2001, 2000), there are no unambiguous methods yet to quantify the “BC,” “NOM,” and “coal” contents of soil or sediment for purposes of predicting sorption and bioavailability. 2. The validity of using isotherm parameters obtained from existing reference standards for BC in the sample is questionable, as such parameters are not universal, and it is seldom possible to trace the source of BC in a sample. For example, the Freundlich KBC varied by two to three orders of magnitude, and nBC varied by several tenths of a unit (Jonker and Koelmans 2002b) on different raw BCs. Moreover, the value of nBC depends on the concentration range over which it is calculated (see Sections 1.3.2 and 1.3.3). Overestimating KBC would correspondingly overestimate the contribution of BC at any given concentration. Overestimating nBC would overestimate the contribution of BC, increasingly so as solute concentration declines. Some investigators have used the residual carbon remaining after CTO treatment as the reference standard; however, such treatment removes non-BC materials (unburned biomass, NOM, unburned liquids) that can strongly affect sorption (Endo et al. 2009a; Pignatello et al. 2006a). 3. As is usual in applications of Equation (1.25), all nonlinearity is attributed to BC. However, as we have seen, sorption even to BC-free humic substances can be nonlinear. When reasonable nonlinearity is ascribed to NOM, the contribution of BC declines compared to the case where all nonlinearity is assigned to BC. This is illustrated in Figure 1.20. The estimated contribution of BC to sorption in a fabricated dataset in which both the OC and BC terms in Equation (1.25) are nonlinear is considerably less than when all the nonlinearity is ascribed to the BC term (Fig. 1.20a), even as the data remain accurately predicted on forcing the OC term to be linear (Fig. 1.20b). 4. In most applications of Equation (1.25), KOC is estimated from historic OC–octanol FERs. These values are

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

Fraction sorbed to black carbon

(a) 1.0

(especially in the case of highly hydrophobic compounds). The contribution of BC will be overpredicted proportionate to the degree that KOC is underestimated. 5. As we have seen in Section 1.3.3, adsorption on BC is attenuated by weathering in the environment. Attenuation can be an order of magnitude or more.

0.8

0.6

1.3.7. Mass Transfer Rates

0.4 data, nOC = 0.80; nBC = 0.60 calc assuming linear OC term

0.2

data, nOC = 0.85; nBC = 0.60 calc assuming linear OC term

0.0 -6

-5

-4

-3

-2

-1

0

1

2

3

log solute concentration (b) 6

log sorbed concentration

data, nOC = 0.80; nBC = 0.60 calc assuming linear OC term

4

data, nOC = 0.85; nBC = 0.60 calc assuming linear OC term

2

0

-2 -8

35

-6

-4 -2 0 log solute concentration

2

4

Figure 1.20. Plots illustrating the effects of forcing sorption by OC to be linear on estimating the fraction sorbed to black carbon in a soil containing 90% OC and 10% BC relative to total organic carbon. (a) Fabricated data are the sum of a Freundlich isotherm to BC, where KBC ¼ 2000 and nBC ¼ 0.6, and a Freundlich isotherm to OC, where KFOC ¼ 300 and where nOC is either 0.80 (circles) or 0.85 (crosshairs). The lines represent calculated values after (1) a linear regression of the OC data forced through zero, which yielded KOC ¼ 96.8 (circles data) or 128 (crosshair data); and (2) regressing log (qT –KOCC) versus log C, which resulted in modified values K0 BC ¼ 3184 and n0 BC ¼ 0.625 (corresponding to circles data), or K0 BC ¼ 3007 and n0 BC ¼ 0.626 (corresponding to crosshairs data) (b) plot shows that the isotherms remain unchanged by linearizing the OC data.

taken to truly represent NOM, yet they were obtained on ordinary soils that contain unknown amounts of BC particles that contribute to sorption to an unknown degree due to the problems mentioned in paragraphs 1–3 and 5. Sorption to OC was generally assumed linear in the historic studies. Consequently, there is a certain degree of circular reasoning behind the acceptance of reference KOC values. Furthermore, too often sorption was underestimated in the historic studies because the equilibration times were insufficiently long

1.3.7.1. General Considerations. Thus far this chapter has dealt with sorption equilibria. However, sorption/desorption rates are often critical factors controlling the transport and bioavailability of contaminants. The rates of sorption and desorption of physisorbing compounds are governed primarily by molecular diffusion and secondarily by matrix flexing properties. Both of these processes are governed by the structure of the sorbent. Diffusion is the tendency of molecules to migrate in response to a gradient in their chemical potential so as to achieve maximum entropy. Diffusivity is a function of molecular structure, the nature and geometry of the diffusing medium, concentration gradients, interfacial boundary conditions, and temperature. Rate laws for soil systems inevitably involve simplifying assumptions because of the system heterogeneity. The simplest situation is the one in which the chemical is stable over the observational timeframe, and soil particles are well dispersed in a fluid such that diffusion and advection in the fluid phase is not limiting. The complexity increases dramatically as one moves to the soil column situation; as water flows or evaporates; and as biological, chemical, or physical processes act to remove the contaminant or change concentration gradients. Diffusive equilibrium in well-mixed water suspensions may require as short as a few hours or as long as several months, depending on the compound and the soil or soil isolate being tested (Pignatello and Xing 1996). Equilibrium after performing a dilution step can take even longer. Many studies have used physical stripping techniques to mimic desorption to an infinitely dilute fluid (sink). Commonly, the stripping technique employs a stream of gas (Werth and Hansen 2002; Werth and Reinhard 1997a,b), or a polymer resin, such as Tenax or XAD added in sufficient quantity to ensure theoretical quantitative mass transfer had equilibrium been achieved (Pignatello 1990a; Zhao and Pignatello 2004). Complete desorption of strongly sorbing contaminants under stripping conditions can require exceedingly long times. This is illustrated in Figure 1.21 for phenanthrene that was desorbed from six different sterilized soils after an uptake step lasting 180 days. Desorption was carried out for up to 606 days in the presence of a large excess of Tenax renewed at each timepoint. Desorption tended eventually toward completion in some soils, but in others was reduced to extremely slow rates and considerable levels of phenanthrene remained.

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

1

1 Seal Beach 240 µg/gOC 25000 µg/gOC

250 µg/g OC 4100 µg/g OC

Pahokee

0.1

Seal Beach

0.1 1

1

530 µg/g OC 25000 µg/g OC

280 µg/g OC 4900 µg/g OC

log q / q0

36

0.1

0.1 Mount Pleasant

Modified Seal Beach

0.01 1

1

980 µg/g OC 21000 µg/g OC

160 µg/g OC 2000 µg/g OC 13000 µg/g OC

Cheshire

Port Hueneme

0.1

0.1 0

100 200 300 400 500 600

0

100 200 300 400 500 600

Time, Days

Figure 1.21. Desorption of phenanthrene from different sterilized soils after preequilibration for 180 days. Lines are for visualization only. [Data are from Braida et al. (2002).] The kinitial and kfinal were calculated on the basis of the first three and last three data in each case.

q0, mg/gOC Pahokee peat Pahokee peat Mount Pleasant Mount Pleasant Cheshire Cheshire Cheshire Seal Beach Seal Beach Modified Seal Beach Modified Seal Beach Port Hueneme Port Hueneme

250 4,100 280 4,900 160 2,000 13,000 240 25,000 530 25,000 980 21,000

kinitial, d
1 0.12 0.20 0.35 0.96 0.29 0.30 0.35 1.0 0.85 0.81 0.97 0.28 0.29

kfinal, day
1 0.00073 0.00047 0.0027 0.15 0.00050 0.00050 0.00023 0.00053 0.00079 0.0012 0.00020 0.0012 0.00049

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

1.3.7.2. Nature and Geometry of the Diffusing Medium. Soils may contain mineral grains with patches and/or coatings of humic substances on their surfaces. These gains may be cemented together in aggregates having a wide range in both pore size and pore connectivity. The aggregates may include microscopic particles of NOM in varying stages of diagenesis, as well as particles of BC. Intraparticle mass transport may involve diffusion through pore fluids (pore diffusion), along pore walls (surface diffusion), or through solid matrices of organic matter (solid-phase, or matrix, diffusion). Pore and surface diffusion are conceptually difficult to distinguish in pores not greatly larger than the width of the diffusant, such as micropores (IUPAC definition, < 2 nm). Diffusion of molecules to or from the furthest reaches of a soil particle requires multiple “jumps” and the crossing of many grain–grain and grain–water interfaces. The length scale over which diffusion is rate-limiting may be much smaller than the macroscopic particle radius and will depend on the micromorphology of the particle (Pignatello 2000). Pore diffusion is retarded by the tortuosity of pore network pathways, sorption on pore walls, and steric hindrance. Steric hindrance begins to appear when the minimum critical molecular diameter is about 10% of the pore diameter and becomes severe as it approaches 100% (K€arger and Ruthven 1992). Steric effects for most molecules of interest will be important in micropores and the smaller mesopores. Matrix diffusion requires cooperative motions between the diffusant and matrix strands as the molecule “jumps” from site to site (Pignatello 2000). Relatively soft NOM represented by humic acid films impedes diffusion by three or four orders of magnitude compared to bulk water (Chang et al. 1997). Glassy solids can impede molecular diffusion by many more orders of magnitude, depending on diffusant diameter, the glass transition temperature Tg (Pignatello 2000), and the magnitude and interconnectivity of the free volume in the solid. The basis for intraparticle diffusion models is Fick’s second law, which is given in radial coordinates for a v-dimensional (v ¼ 1 for a slab, v ¼ 2 for a cylinder, v ¼ 3 for a sphere) particle that is isotropic and homogeneous, as follows   @s 1 @ @s ðv
1Þ ¼ r Deff ðT; sÞ ð1:26Þ @t rðv
1Þ @r @r where r is the thickness (slab) or radius (cylinder, sphere) of the diffusing medium; s (M/L3) is the total local volumetric concentrationin the diffusing medium, including sorbed chemical and chemical dissolved in internal pore water; and Deff(T,s) [L2/T] is the effective diffusion coefficient, or diffusivity, which may be concentration- and/or temperature-dependent.

37

The average sorbed concentration in the particle is obtained by integration; in the example of a sphere it is given by SðtÞ ¼ 3 r3

ðr sðr; tÞr2 dr

ð1:27Þ

0

The boundary condition at the particle–external solution interface is given by, sðtÞr¼r ¼

ds CðtÞ dC

ð1:28Þ

where ds/dC is the distribution coefficient usually presumed equivalent to Kd obtained from the isotherm. Analytical solutions to the diffusion equation when Deff is constant are available for uniform spherical particles applicable to various situations, such as finite external solution, infinite external solution of constant or variable concentration, and desorption to a “vacuum” (Crank 1975; Haws et al. 2006; K€arger and Ruthven 1992; Pignatello 2000) (see also Chapter 8). Numerical solutions have been worked out for less homogeneous media and situations of greater complexity. Typically, the output parameter of the diffusion model is Deff/reff2 (T
1) since the characteristic diffusion length scale reff is unknown. If the diffusing medium is an isotropic phase like NOM or a network of micropores, Deff is not broken down further. If the diffusing medium is a network of waterfilled meso- or macropores, then intraparticle diffusion can be regarded to occur only in the pore fluid with retardation due to sorption on or in the walls. The Deff may then be expressed as Deff ¼

wkt
1 Dw w þ ð1
wÞK

ð1:29Þ

where j is the intraparticle porosity; Dw is the diffusivity in water; K is the sorption distribution coefficient, usually taken to be the bulk Kd; t ( 1) is a tortuosity factor that reflects deviation from straight-line paths and pore interconnectedness (often taken to be proportional to j
1); and k (1) is a parameter that reflects steric hindrance by the pore walls. Steric effects become significant when the molecular diameter reaches about 10% of the pore diameter (K€arger and Ruthven 1992). In reality, pore surface diffusion occurs as well, but it is difficult to estimate out. If sorption sites are microdomains of NOM on internal pore surfaces, then Deff should correlate inversely with OC content of the soil, since K would be proportional to fOC [Eq. (1.29)]; so far, such a correlation seems not to exist (Birdwell et al. 2007; Kukkonen et al. 2003; Shor et al. 2003).

38

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Soil heterogeneity has an important influence on rates. Theoretically, the uptake or release rate is inversely related to the square of the particle radius. In some studies, the rate is found to increase with decreasing nominal particle radius (Ball and Roberts 1991b; Kleineidam et al. 1999b; Wu and Gschwend 1986), while in others no dependence is observed (e.g., Carroll et al. 1994; Farrell and Reinhard 1994; Steinberg et al. 1987). However, it is usually true that pulverization of the soil increases rates (Ball and Roberts 1991a; Pignatello 1990b; Steinberg et al. 1987). Examining coaly sedimentary sands and gravels separated on the basis of size, color, and porosity, Kleineidam (1999b) found that sorption rate decreased with increasing size, increasing OC content, and decreasing porosity of particles. As described by Cook in Chapter 13 of this book, water wetting of soil is a process that affects the rate of chemical uptake, yet may take days to reach equilibrium. Matrix flexing has an effect on rates. Sander and Pignatello (2005b; 2009) compared uptake and release curves of a chemical and its radiolabeled tracer. The experiments were performed under identical gradient conditions, except that the tracer was added after the bulk chemical had already equilibrated and in an amount sufficiently small so as to not perturb the bulk chemical equilibrium. In all cases (naphthalene and 1,4-dichlorobenzene; two solids; high and low concentrations), tracer rates were faster, especially at the higher bulk chemical concentration. There are three reasons for this result: (1) diffusion is faster for the tracer because the bulk chemical already occupied the higher energy (slower filling/emptying) sorption sites, (2) diffusion of the tracer takes place in an already filled solid that is more open and flexible than it was during diffusion of the bulk chemical because of plasticization (Section 1.3.2.3), and (3) sorption or desorption of the bulk (but much less so for the tracer) is accompanied by matrix swelling or shrinking that prolongs its uptake or release; this is reflected in a distinct biphasic shape of the curves for the bulk (but not tracer) at its higher (but not lower) concentration. 1.3.7.3. Influence of Solute Structure, Solute Concentration, and Competition. Typically, the characteristic rate parameter decreases with increasing molecular size or hydrophobicity. Such behavior is expected in all of the relevant diffusion media—in pores, on surfaces, and in a matrix—and is consistent with the findings of fundamental studies on polymers (Berens 1989; Rogers 1965) and porous inorganic reference materials (K€arger and Ruthven 1992). However, systematic studies of natural solids involving more than three compounds are rare (Carroll et al. 1994; Piatt and Brusseau 1998), as are studies contrasting polar and nonpolar molecules of similar size (Piatt and Brusseau 1998). Compounds with multiple points of interaction are expected to diffuse more slowly because when strong interactions occur

simultaneously at more than one place in the molecule, all such interactions must be broken before the molecule can jump to the next site. Diffusion in glassy materials is far more sensetive to molecular diameter than in rubbery materials. Take, for example, the diffusion of gases and hydrocarbons up to  10 A in diameter in polyvinylchloride (PVC; Tg ¼ 85 C) (Berens 1989): the diffusion coefficient declined at the rate of 3.4 log units per angstrom diffusant diameter in the asprepared (glassy) polymer, while in the plasticized polymer (with added phthalate ester compounds) it declined at the rate of only 0.46 log units per angstrom. The concentration dependence of the diffusion coefficient, D for a chemical species within a given diffusing medium (say, NOM or the interstices of a porous particle) is dictated by the gradient in chemical potential with respect to concentration (K€arger and Ruthven 1992; Pignatello 2000): D ¼ D0

dlnp dlns

ð1:30Þ

Here, p is the pressure in the external fluid, s is the concentration in the diffusing medium, and D0 is the self-diffusivity (also called corrected diffusivity). Thus, D ! D0 when s is a linear function of p, which is always true as s ! 0. Thus, sorption nonlinearity and competition affect sorption and desorption rates (Braida et al. 2001, 2002; Sander and Pignatello 2005b, 2009). Sorption–desorption by a compound exhibiting a linear isotherm with a particular sorbent is symmetric; that is, the normalized rate will be independent of the absolute concentration, and the normalized sorption curve (Mt/M1 vs. t) and the corresponding normalized desorption curve after dilution will coincide. However, the same is not true for a compound exhibiting a nonlinear isotherm. Recalling that a concavedown isotherm indicates weakening affinity with increasing concentration, the normalized sorption or desorption rate increases with absolute concentration because the solid provides progressively lower resistance to diffusion. Thus, for two identical particles placed in separate infinite liquid media, the one in the more dilute medium will take longer to reach equilibrium. Moreover, in an experiment in which desorption follows sorption by diluting the liquid phase, the system will take longer to reach equilibrium in the desorption than in the sorption step, in relation to the deviation from isotherm linearity. This happens because the strongest sites are filled from a relatively high concentration source, but emptied to a relatively low concentration sink. Competition also affects diffusion. A competing cosolute will increase the rate of sorption or desorption of a principal solute, proportionate to cosolute concentration, because the principal solute occupies weaker and weaker sites, and becomes progressively more labile (White et al. 1999c; Zhao et al. 2001).

SORPTION FROM THE PERSPECTIVE OF THE SORBENT

Diffusion in solids is temperature-dependent. The governing equation is D ¼ Ae
Ea =RT

ð1:31Þ

where Ea is the diffusion activation energy and A is a regression parameter. Diffusion is more temperature-dependent in glassy than rubbery solids and in microporous than mesoporous solids. 1.3.7.4. Strongly Resistant Desorption. Over the years many researchers have discovered that chemicals placed in contact with soil can generate a fraction that becomes highly resistant to desorption and biodegradation. Such fractions may be found in historically contaminated samples as well as in spiked samples of clean soil, and may be exhibited even by small, weakly sorbing molecules like C2 and C3 halogenated hydrocarbons. This topic is the subject of several reviews (Brusseau and Rao 1989; Luthy et al. 1997; Pignatello 1990b, 2000; Pignatello and Xing 1996) and will only be summarized here. Desorption resistance has important implications for environmental transport, natural attenuation, bioavailability, and bioremediation and physicochemical remediation strategies (Alexander 1995, 2000; Loehr and Webster 1997; Pignatello and Xing 1996; Pignatello, 2009). The term resistant—and the converse, labile—are not rigorously defined but depend on the experimental timeframe and methodology. Nevertheless, a fraction showing pronounced resistance often makes up a nontrivial fraction— up to several percent or more of the total concentration present. Historically contaminated samples may be enriched in the resistant fraction, due to the depletion of more labile fractions by dissipation and degradation during the lengthy time that passes before the sample is collected (Pignatello et al. 1993). It has been shown that a highly resistant fraction can be generated after a contact time as short as a few hours (Cornelissen et al. 1997; Kan et al. 1997; 1998; Pignatello 1990a,b; Ten Hulscher et al. 2005). Biodegradation of a chemical, previously allowed to equilibrate with a sterilized soil before inoculation, often tails off to leave a biodegradation-resistant fraction that correlates with the desorption-resistant fraction (Braida et al. 2004; White et al. 1999b) [see also citations in Pignatello and Xing (1996)]. An example of extreme resistance of an otherwise labile compound in a historically contaminated soil is 1,2-dibromoethane. This biodegradable, volatile, and moderately water-soluble compound persisted at microgram per kilogram (mg/kg) levels in topsoil up to at least 19 years after its last known application (Pignatello et al. 1990; Steinberg et al. 1987). Compared to similar concentrations of freshly added 14 C-1,2-dibromoethane, the field residues exhibited far greater soil–water distribution ratios (Kd), far slower rates of desorption, and greatly reduced degradation by native

39

microorganisms. An example of extreme resistance in a spiked system is trichloroethene (TCE); passage of a stream of N2 at 100% relative humidity through preequilibrated soil columns released most of the TCE with 10 min, but a small fraction was extrapolated to take months to years to desorb (Farrell and Reinhard 1994). Various nonexhaustive extraction techniques have been proposed for measuring physical availability or predicting bioavailability of contaminants. A popular approach is the use of polymer adsorbents such as Tenax or XAD resin added in large excess. Using Tenax, Cornelissen et al. (2000) applied an exponential desorption model that assumes “fast,” “slow,” and sometimes “very slow” compartments. Each compartment desorbs in a first-order manner. The twocompartment model is given by q ¼ Ffast e
kfast t þ Fslow e
kslow t q0

ð1:32Þ

where q0 is the initial sorbed concentration and F represents the mass fraction and k (T
1) the rate constant for contaminant in the designated compartment. This approach is equivalent to applying a driving force condition at the surface [i.e., rate is proportional to the difference in chemical activity in the two phases at the interface (near zero in the aqueous phase)] and assuming that the chemical in the particle is always well mixed. In the example of Figure 1.21, one can see that the final desorption rate constant (corresponding to the slope of log q/q0 vs. time) is very much smaller than the initial desorption rate constant—in most cases, three orders of magnitude smaller. Several studies report correspondence between the Tenax-desorbed fraction and the biodegraded fraction (Cornelissen et al. 2000; Braida et al. 2004; Lei et al. 2004; Li et al. 2005; Pignatello 2006). Possible causes of strongly resistant desorption include the following: .

The normal process of retarded diffusion in and out of remote domains . Occlusion of molecules in closed pores during particle genesis . Alteration of the soil matrix during a sorption– desorption cycle that results in highly hindered diffusion or occlusion. Desorption resistance is usually attributed to the limitations of molecular diffusion through highly tortuous and sterically hindered pore networks or through highly viscous organic matter phases. The progress of diffusion through the mesopores of mineral aggregates may be retarded by the presence of microparticles of organic matter within the aggregates (Kleineidam et al. 1999a), which provide “way stations.” NOM phases may contain domains at the microscopic scale that are rigid, and thus present great, although

40

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

not insurmountable, barriers to diffusion. For highly hydrophobic compounds such as PAHs in stagnant soil columns, sorption equilibrium is not easily reached because the fluid-phase mass (representing the actual diffusant) is a very small fraction of the total mass per unit volume. Sorption intensity increases and diffusivity decreases as soil moisture content decreases. Retarded diffusion can explain the “aging effect,” in which bioavailability (uptake or degradation) decreases with precontact time of the chemical (Alexander 2000). The greater the degree of equilibrium reached in the precontact step, the less will leak out after contact with the organism is initiated. Because of the “random walk” nature of diffusion, incomplete pre-equilibration will always lead to molecules migrating both “inward” and “outward” of the particle after it contacts the organism. Thus, even for very short precontact times, some of the contaminant will end up becoming “bioresistant” to the observer. Diffusion can also explain the “rebound” effect that occurs when a contaminant is seemingly exhaustively removed by a remediation technique, and then redistributes over time, being placed once again in a more available state. Occlusion during particle synthesis is exemplified by PAHs in soot. Soot condenses after a complex series of gas-phase free-radical reactions, in which PAHs are among the intermediates (Akhter et al. 1985b; Lahaye 1990; Smedley et al. 1992). It is possible that some unreacted PAH molecules become trapped in internal pores of the nascent soot condensate—pores that later may have no connection to external fluids. The evidence for occlusion of PAHs during BC synthesis is only circumstantial; it rests mainly on the desorption-resistance of some fraction of PAH content under extraordinary conditions—for example, high temperatures (Harmon et al. 2001) or supercritical carbon dioxide extraction (Jonker et al. 2005)—and is also implied by the ability of organic solvents to facilitate extraction (Akhter et al. 1985a; Jonker and Koelmans 2002a). Native PAHs in sediment thought to be associated primarily with soot particles equilibrated poorly with an isotope-labeled PAH spike (Jonker and Koelmans 2002b). Alteration of the soil matrix leading to occlusion has been little investigated but holds some credence. Farrell et al. (1999) proposed that mineral precipitation leads to blockage of intra-granular micropores in silica, entrapping TCE and PCE. Some have suggested that inelastic swelling of NOM (Sander and Pignatello 2009; Weber et al. 2002) and BC (Braida et al. 2003) may lead to immobilization via an antiplasticization mechanism occurring during the desorption step. In such a case, an abrupt release of sorbed chemical causes the matrix to collapse and stiffen around some molecules before they have a chance to escape. Release in this case may involve cooperative flexing of the matrix of a nature that requires a high activation energy. This hypothesis has not been rigorously tested.

1.4. SUMMARY AND CONCLUDING REMARKS Whether one concludes that we know a little or a lot about sorption to natural organic substances depends on one’s perspective. This author believes we clearly know a lot about individual processes and interactions in well-focused experimental situations (although much remains to be learned about the details); but when it comes to actually predicting in situ sorption–desorption behevior from sorbate and sorbent properties, we are a long way off. This is poignantly illustrated by a recent study (Arp et al. 2009) that found that established models could predict in situ KOC values for apolar compounds in impacted sediments (PAHs, PCBs, PCDD/Fs, and chlorinated benzenes; 55 compounds, 473 data) to within a range of only three orders of magnitude (i.e.,  a factor of 30) at best! The two best models were based on coal tar as a sorption surrogate for sediment organic matter, not NOM itself; while one was a ppFER, the other was simple singleparameter Raoult’s law model (i.e., KOC inversely proportional to subcooled water solubility). Sorption of uncharged molecules from the aqueous phase to NOM and BC will be driven primarily by the hydrophobic effect, with noncovalent forces directly with the sorbent contributing. Suitable free-energy models to assign contributions of individual driving forces are in an early stage of development. Poly-parameter FERs are better than singleparameter FERs, but are difficult to interpret unambiguously in terms of driving force contributions. Dispersion forces seem to be balanced between water and the sorbent. Hydrogen-bond donating by the sorbate seems to have little effect on sorption intensity, while H-bond accepting apparently disfavors sorption. It is likely that p-p electron donor–acceptor interactions will be found to play an important role in sorption involving moieties with especially strong donor/ acceptor character—those with strong electron-withdrawing substituents (e.g., polynitro, charged aromatic and heteroaromatic amines) and highly polarizable donor ability (e.g., PAHs, graphene surfaces on BC). Steric effects play an important role in sorption to microporous materials such as BC; however, they have not yet been quantified. The importance of steric effects in partitioning to NOM is unknown but could be substantial for very large molecules that require a large cavity. Sorption intensity of ionic and ionizable organic compounds to both NOM and BC is still poorly understood and predictable. Nonlinearity in the xisotherm reflects the heterogeneity of the sorption process. Depending on the degree of nonlinearity, the sorption distribution coefficient can vary considerably— by as much as three orders of magnitude—over the range in concentration from infinite dilution to maximum solubility in water. The implications of nonlinearity for bioavailability and environmental fate are, therefore, quite substantial but have not been fully appreciated. The causes of nonlinear sorption in NOM are still being debated, but appear at this point to be due

REFERENCES

to changes in the chemical and/or physical nature of sorption sites or domains with loading, rather than a saturation limitation of any particular functional group interaction. Evidence for preferential sorption based on domains segragated on the basis of functional unit identity is mixed. The glassy polymer model of NOM provides a rationale for nonlinearity of even hydrocarbons in NOM in regard to its postulate of progressive filling of internal microporosity frozen into the matrix. Competitive sorption is simply the “flip side” of nonlinearity. A competitor will both suppress sorption and increase the isotherm linearity of the chemical of interest. The electronic and steric factors controlling the degree of overlap in the sorption domains of competing sorbates are poorly understood. Competition can give rise to artificial hysteresis if the competitor is diluted. Sorption to BC is typically more highly nonlinear than sorption to NOM. Nonlinearity in BC originates from both nonuniform pore sizes and chemical inhomogeneity of the surface. Adsorption to BC is inversely related to polar atom (mainly O) content, due to competition for adsorption space from water molecules. Environmental BC lies on a continuum between partially charred biomass and graphitized carbon depending on formation conditions, which confounds its quantification in environmental samples. The surface activity of BC can be attenuated by incompletely carbonized biomass, unburned fuel products, and natural substances—organic and inorganic—that adsorb during weathering in the environment. All of these factors make prediction of sorption intensity from the supposed composition of soil organic matter tenuous. The composition of organic matter in soils is best regarded as a continuum of materials ranging from fully flexible macromolecular substances that behave as true partition phases, to fully rigid graphitic substances with extended pore networks that behave analogously to fixed-pore adsorbents. Except for the latter extreme, these materials are dynamic in the sense that they can undergo swelling or shrinking during uptake or release of sorbates. The swelling process becomes inelastic as the ambient temperature approaches the glass transition temperature of the substance. The inelasticity of expansion results in an increase in the excess free volume, which persists, and leads to sorption hysteresis. Demonstration of pore deformation hysteresis in NOM not only represents an example of true hysteresis but also provides convincing evidence for the glassy character of at least some fraction of NOM in soil. Sorption–desorption rates play a critical role in the bioavailability and mobility of contaminants, yet little progress has been made in our ability to predict rates from soil and contaminant properties and existing conditions. Diffusion is sensitive to molecular diameter, possibly polarity, concentration, temperature, competing sorbates, the nature of the organic matter, and how the organic matter is distributed within soil particles and aggregates. At high concentrations,

41

diffusion may be sensitive to matrix changes accompanying the uptake or release of contaminant. Diffusion in soil materials appears not to be symmetric—that is, a single diffusivity is unable to describe the entire uptake or desorption profile. The quantities in the “fast,” “slow,” and “very slow” compartments have been the focus of many papers, but this distinction, while useful for some experimental purposes, is artificial and rarely allows prediction or translation to another system. A small fraction of the total chemical present may be so resistant to desorption that for all practical purposes it is physically and biologically unavailable. The causes of highly resistant desorption are poorly understood, and so predicting this fraction from soil/chemical properties is not yet possible. Sorption hysteresis and resistant desorption are the “elephants in the room” of sorption research— common problems with serious implications that nobody wants to tackle.

REFERENCES Accardi-Dey, A. and Gschwend, P. M. (2002), Assessing the combined roles of natural organic matter and black carbon as sorbents in sediments, Environ. Sci. Technol. 36, 21–29. Accardi-Dey, A. and Gschwend, P. M. (2003), Reinterpreting literature sorption data considering both absorption into organic carbon and adsorption onto black carbon, Environ. Sci. Technol. 37, 99–106. Adamson, A. W. and Gast, A. P. (1997), Physical Chemistry of Surfaces, 6th ed., Wiley, New York. Ahrens, L., Yamashita, N., Yeung, L. W. Y., Taniyasu, S., Horii, Y., Lam, P. K. S., and Ebinghaus, R. (2009), Partitioning behavior of per- and polyfluoroalkyl compounds between pore water and sediment in two sediment cores from Tokyo Bay, Japan, Environ. Sci. Technol. 43, 6969–6975. Aiken, G. R., McKnight, D. M., Wershaw, R. L., and MacCarthy, P., eds. (1985), Humic Substances in Soil, Sediment, and Water, Wiley, New York. Akhter, M., Chughtai, A, and Smith, D. (1985a), The structure of hexane soot I: Spectroscopic studies, Appl. Spectrosc. 39, 143–153. Akhter, M.S.,Chughtai, A. R, and Smith, D. M. (1985b), The structure ofhexanesootII:Extractionstudies,Appl.Spectrosc.39,154–167. Alexander, M. (1995), How toxic are toxic chemicals in soil? Environ. Sci. Technol. 29, 2713–2717. Alexander, M. (2000), Aging, bioavailability, and overestimation of risk from environmental pollutants, Environ. Sci. Technol. 34, 4259–4265. Allen-King, R. M., Grathwohl, P., and Ball, W. P. (2002), New modeling paradigms for the sorption of hydrophobic organic chemicals to heterogeneous carbonaceous matter in soils, sediments, and rocks, Adv. Water Resour. 25, 985–1016. Altfelder, S., Streck, T, and Richter, J. (2000), Nonsingular sorption of organic compounds in soil: The role of slow kinetics, J. Environ. Qual. 29, 917–925.

42

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Arp, H. P. H., Breedveld, G. D, and Cornelissen, G. (2009), Estimating the in situ sediment–porewater distribution of PAHs and chlorinated aromatic hydrocarbons in anthropogenic impacted sediments, Environ. Sci. Technol. 43, 5576–5585. Bailey, A., Cadenhead, D. A., Davies,D. H., Everett, D. H., and Miles, A. J. (1971), Low pressure hysteresis in the adsorption of organic vapours by porous carbons, Trans. Faraday Soc. 67, 231–243. Ball, W. P. and Roberts, P. V. (1991a), Long-term sorption of halogenated organic chemicals by aquifer material. 1. Equilibrium, Environ. Sci. Technol. 25, 1223–1235. Ball, W. P. and Roberts, P. V. (1991b), Long-term sorption of halogenated organic chemicals by aquifer material. 2. Intraparticle diffusion, Environ. Sci. Technol. 25, 1237–1249. Barring, H., Bucheli, T. D., Broman, D., and Gustafsson, O. (2002), Soot-water distribution coefficients for polychlorinated dibenzop-dioxins, polychlorinated dibenzofurans and polybrominated diphenylethers determined with the soot cosolvency-column method, Chemosphere 49, 515–523. Bayard, R., Barna, L., Mahjoub, B., and Gourdon, R. (2000), Influence of the presence of PAHs and coal tar on naphthalene sorption in soils, J. Contam. Hydrol. 46, 61–80. Berens, A. (1989), Transport of organic vapors and liquids in poly (vinyl chloride), Makromol.Chem., Macromol.Symp.29, 95–108. Birdwell, J., Cook, R. L., and Thibodeaux, L. J. (2007), Desorption kinetics of hydrophobic organic chemicals from sediment to water: A review of data and models, Environ. Toxicol. Chem. 26, 424–434. Boehm, H. P. (1964), Some aspects of the surface-chemistry of carbon-blacks and other carbons, Carbon. 32, 759–769. Bontha, J. R. and Kaplan, D. I. (1999), Immobilization or recovery of chlorinated hydrocarbons from contaminated groundwater using clathrate hydrates: A proof-of-concept, Environ. Sci. Technol. 33, 1051–1056. Borisover, M. and Graber, E. R. (2003), Classifying NOM-organic sorbate interactions using compound transfer from an inert solvent to the hydrated sorbent, Environ. Sci. Technol. 37, 5657–5664. Braida, W. J., White, J. C., Ferrandino, F. J., and Pignatello, J. J. (2001), Effect of solute concentration on sorption of polyaromatic hydrocarbons in soil: Uptake rates, Environ. Sci. Technol. 35, 2765–2772. Braida, W., White, J. C., Zhao, D., Ferrandino, F. J., and Pignatello, J. J. (2002), Concentration-dependent kinetics of pollutant desorption from soils, Environ. Toxicol. Chem. 21, 2573–2580. Braida, W., Pignatello, J. J., Lu, Y., Ravikovitch, P. I., Neimark, A. V., and Xing, B. (2003), Sorption hysteresis of benzene in charcoal particles, Environ. Sci. Technol. 37, 409–417. Braida, W., White, J. L., and Pignatello, J. J. (2004), Indices for bioavailability and biotransformation potential of contaminants in soils, Environ. Toxicol. Chem. 23, 1585–1591. Breault, G. A., Hunter, C. A., and Mayers, P. C. (1998), Influence of solvent on aromatic interactions in metal tris-bipyridine complexes, J. Am. Chem. Soc. 120, 3402–3410. Brewer, C. E., Schmidt-Rohr, K., Satrio, J. A, and Brown, R. D. (2009), Characterization of biochar from fast pyrolysis and

gasification systems, Environ. Prog. Sustain. Energy 28, 386–396. Brunauer, S., Emmett, P. H., and Teller, E. (1938), Adsorption of gases in multimolecular layers, J. Am. Chem. Soc. 60, 309–319. Brusseau, M. L. and Rao, P. S. C. (1989), Sorption nonideality during organic contaminant transport in porous media, Crit. Rev. Environ. Control. 19, 33–99. Bucheli, T. D. and Gustafsson, O. (2001), Ubiquitous observations of enhanced solid affinities for aromatic organochlorines in field situations: Are in situ dissolved exposures overestimated by existing partitioning models? Environ. Toxicol. Chem. 20, 1450–1456. Budd, P. M., McKeown, N. B., and Fritsch, D. (2005), Free volume and intrinsic microporosity in polymers, J. Mater. Chem. 15, 1977–1986. Carda-Broch, S. and Berthod, A. (2004), Countercurrent chromatography for the measurement of the hydrophobicity of sulfonamide amphoteric compounds, Chromatographia 59, 79–87. Carroll, K. M., Harkness, M. R., Bracco, A. A., and Balcarcel, R.R. (1994), Application of a permeant/polymer diffusional model to the desorption of polychlorinated biphenyls from hudson river sediments, Environ. Sci. Technol. 28, 253–258. Chan, K. Y. and Xu, Z. (2009), Biochar: Nutrient properties and their enhancement, in Biochar for Environmental Management, Lehmann, J. and Joseph, S., Earthscan, London, UK or Sterling, VA pp. 67–81. Chandler, D. (2005), Interfaces and the driving force of hydrophobic assembly, Nature 437, 640–647. Chang, M., Wu, S., and Chen, C. (1997), Diffusion of volatile organic compounds in pressed humic acid disks, Environ. Sci. Technol. 31, 2307–2312. Chatterjee, R., Laird, D. A., and Thompson, M. L. (2008), Interactions among K þ - Ca2 þ exchange, sorption of m-dinitrobenzene, and smectite quasicrystal dynamics, Environ. Sci. Technol. 42, 9099–9103. Chefetz, B., Deshmukh, A. P., Hatcher, P. G., and Guthrie, E. A. (2000), Pyrene sorption by natural organic matter, Environ. Sci. Technol. 34, 2925–2930. Chefetz, B. and Xing, B. (2009), Relative role of aliphatic and aromatic moieties as sorption domains for organic compounds: A review, Environ. Sci. Technol. 43, 1680–1688. Chen, B., Johnson, E. J., Chefetz, B., Zhu, L. and Xing, B. (2005), Sorption of polar and nonpolar aromatic organic contaminants by plant cuticular materials: Role of polarity and accessibility, Environ. Sci. Technol. 39, 6138–6146. Chen, B. and Xing, B. (2005), Sorption and conformational characteristics of reconstituted plant cuticular waxes on montmorillonite, Environ. Sci. Technol. 39, 8315–8323. Chen, G., Shan, X., Wang, Y., Wen, B., Pei, Z., Xie, Y., Liu, T., and Pignatello, J. J. (2009), Adsorption of 2,4,6-trichlorophenol by multi-walled carbon nanotubes as affected by Cu(II), Water Res. 43, 2409–2418. Chen, J., Zhu, D., and Sun, C. (2007a), Effect of heavy metals on the sorption of hydrophobic organic compounds to wood charcoal, Environ. Sci. Technol. 41, 2536–2541.

REFERENCES

Chen, W., Duan, L. and Zhu, D. Q. (2007b), Adsorption of polar and nonpolar organic chemicals to carbon nanotubes, Environ. Sci. Technol. 41, 8295–8300. Chien, Y. and Bleam, W. F. (1997), Fluorine-19 nuclear magnetic resonance study of atrazine in humic and sodium dodecyl sulfate micelles swollen by polar and nonpolar solvents, Langmuir 13, 5283–5288. Chiou, C. T. (1989), Theoretical considerations of the partition uptake of nonionic organic compounds by soil organic matter, in Reactions and Movement of Organic Chemicals in Soil, Sawhney, B. L. and Brown, K., eds., Soil Science Society of America (special publication, Madison, WI, pp. 1–29. Chiou, C. T. (2002), Partition and Adsorption of Organic Contaminants in Environmental Systems, Wiley, Hoboken, NJ. Chiou, C. T., McGroddy, S. E., and Kile, D. E. (1998), Partition characteristics of polycyclic aromatic hydrocarbons on soils and sediments, Environ. Sci. Technol. 32, 264–269. Cho, H. H., Smith, B. A., Wnuk, J. D., Fairbrother, D. H., and Ball, W. P. (2008), Influence of surface oxides on the adsorption of naphthalene onto multiwalled carbon nanotubes, Environ. Sci. Technol. 42, 2899–2905. Cockroft, S. L., Perkins, J., Zonta, C., Adams, H., Spey, S. E., Low, C. M. R., Vinter, J. G., Lawson, K. R., Urch, C. J, and Hunter, C. A. (2007), Substituent effects on aromatic stacking interactions, Org. Biomol. Chem. 5, 1062–1080. ¨ . (2004), Cornelissen, G., Elmquist, M., Groth, I., and Gustafson, O Effect of sorbate planarity on environmental black carbon sorption, Environ. Sci. Technol. 38, 3574–3580. ¨ . (2005), Importance of unburned Cornelissen, G. and Gustafson, O coal carbon, black carbon, and amorphous organic carbon to phenanthrene sorption in sediments, Environ. Sci. Technol. 39, 764–769. Cornelissen, G. and Gustafsson, O. (2004), Sorption of phenanthrene to environmental black carbon in sediment with and without organic matter and native sorbates, Environ. Sci. Technol. 38, 148–155. Cornelissen, G. and Gustafsson, O. (2006), Effects of added PAHs and precipitated humic acid coatings on phenanthrene sorption to environmental black carbon, Environ. Pollut. 141, 526–531. Cornelissen, G., Hassell, K. A., van Noort, P. C. M., Kraaij, R., van Ekeren, P. J., Dijkema, C., de Jager, P. A., and Govers, H. A. J. (2000), Slow desorption of PCBs and chlorobenzenes from soils and sediments: Relations with sorbent and sorbate characteristics, Environ. Pollut. 108, 69–80. Cornelissen, G., van Noort, P. C. M., and Govers, H. A. J. (1997), Desorption kinetics of chlorobenzenes, polycyclic aromatic hydrocarbons, and polychlorinated biphenyls: Sediment extraction with Tenax and effects of contact time and solute hydrophobicity, Environ. Toxicol. Chem. 16, 1351–1357. Crank, J. (1975), The Mathematics of Diffusion, 2nd ed., Clarendon Press, Oxford, UK. Crittenden, J. C., Luft, P., Hand, D. W., Oravitz, J. L., Loper, S. W., and Arl, M. (1985), Prediction of multicomponent adsorption equilibria using ideal adsorbed solution theory, Environ. Sci. Technol. 19, 1037–1043.

43

Curl, R. L. and Keolelan, G. A. (1984), Implicit-adsorbate model for apparent anomalies with organic adsorption on natural adsorbants, Environ. Sci. Technol. 18, 916–922. Dachs, J. and Eisenreich, S. J. (2000), Adsorption onto aerosol soot carbon dominates gas-particle partitioning of polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 34, 3690–3697. Del Vecchio, R. and Blough, N. L. (2004), On the origin of the optical properties of humic substances, Environ. Sci. Technol. 38, 3885–3891. DeLapp, R. C. and Leboeuf, E. J. (2004), Thermal analysis of whole soils and sediment, J. Environ. Qual. 33, 330–337. Dixon, A. M., Mai, M. A, and Larive, C. K. (1999), NMR investigation of the interactions between 4’-fluoro-1’-acetonaphthone and the Suwannee River fulvic acid, Environ. Sci. Technol. 33, 958–964. Donnet, J.-B., Bansal, R. C., and Wang, M.-J. (1993), Carbon Black, 2nd ed., Marcel Dekker, New York. Dulfer, W. J. and Govers, H. A. J. (1995), Membrane-water partitioning of polychlorinated biphenyls in small unilamellar vesicles of four saturated phosphatidylcholines, Environ. Sci. Technol. 29, 2548–2554. Eisenberg, A. (1993), The glassy state and the glass transition, in Physical Properties of Polymers, Mark, J. E., Eisenberg, A., Graessley, W. W., Mandelkern, L., Samulski, E. T., Koenig, J. L., and Wignall, G. D. eds., American Chemical Society, Washington, DC, pp. 61–95. Endo, S., Grathwohl, P., Haderlein, S. B., and Schmidt, T. C. (2008a), Compound-specific factors influencing sorption nonlinearity in natural organic matter, Environ. Sci. Technol. 42, 5897–5903. Endo, S., Grathwohl, P., and Schmidt, T. C. (2008b), Absorption or adsorption? Insights from molecular probes n-alkanes and cycloalkanes into modes of sorption by environmental solid matrices, Environ. Sci. Technol. 42, 3989–3995. Endo, S., Grathwohl, P., Haderlein, S. B., and Schmidt, T. C. (2009a), Effects of native organic material and water on sorption properties of reference diesel soot, Environ. Sci. Technol. 43, 3187–3193. Endo, S., Grathwohl, P., Haderlein, S. B., and Schmidt, T. C. (2009b), LFERs for soil organic carbon-water distribution coefficients (Koc) at environmentally relevant sorbate concentration, Environ. Sci. Technol. 43, 3094–3100. Farrell, J., Grassian, D., and Jones, M. (1999), Investigation of mechanisms contributing to slow desorption of hydrophobic compounds from mineral solids, Environ. Sci. Technol. 33, 1237–1243. Farrell, J. and Reinhard, M. (1994), Desorption of halogenated organics from model solids, sediments, and soil under unsaturated conditions. 2. Kinetics, Environ. Sci. Technol. 28, 63–72. Ferguson, S. B. and Diederich, F. (1986), Electron donor-acceptor interactions in host-guest complexation in organic solutions, Angew. Chem. Int. Ed. Engl. 25, 1127–1129. Fleming, G. K. and Koros, W. J. (1990), Carbon dioxide conditioning effects on sorption and volume dilation behavior for biphenol a-polycarbonate, Macromolecules 23, 1353–1360.

44

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Foster, R. (1969), Organic Charge-Transfer Complexes, Academic Press, London. Frank, H. S. and Evans, M. W. (1945), Free volume and entropy in condensed systems. .3. Entropy in binary liquid mixtures— partial molal entropy in dilute solutions—structure and thermodynamics in aqueous electrolytes, J. Chem. Phys. 13, 507–532. Fritz, J. S. (2005), Factors affecting selectivity in ion chromatography, J. Chromatogr. A (Proc. 17th Int. Ion Chromatography Symp. 1085, 8–17. Fritz, W. and Schlunder, E. U. (1981), Competitive adsorption of two dissolved organics onto activated carbon-I, Chem. Eng. Sci. 36, 721–730. Gilli, G. and Gilli, P. (2000), Towards an unified hydrogen-bond theory, J. Molec. Struct. 552, 1–15. Gilli, P., Pretto, L., Bertolasi, V., and Gilli, G. (2009), Predicting hydrogen-bond strengths from acid-base molecular properties. The pK(a) slide rule: Toward the solution of a long-lasting problem, Acc. Chem. Res. 42, 33–44. Glasser, W. G. and Kelley, S. S. (1987), Lignin, in Encyclopedia of Polymer Science and Technology, Kroschwitz, J., ed., Wiley, New York, pp. 795–852. Gobas, F. A. P. C., Lahittete, J. M., Garofalo, G., Shiu, W. Y., and Mackay, D. (1988), A novel method for measuring membranewater partition coefficients of hydrophobic organic chemicals: Comparison with 1-octanol-water partitioning, J. Pharm. Sci. 77, 265–272. Goldberg, E. D. (1985), Black Carbon in the Environment, Wiley, New York. Greenland, D. J. and Hayes, M. H. B. (1981), Soil processes, in The Chemistry of Soil Processes ed. Greenland, D. J., and Hayes, M. H. B., eds., Wiley, Chichester, UK, pp. 1–35. Grossman, A. and Ghosh, U. (2009), Measurement of activated carbon and other black carbons in sediments, Chemosphere 75, 469–475. Gu, C., Karthikeyan, K. G., Sibley, S. D., and Pedersen, J. A. (2007), Complexation of the antibiotic tetracycline with humic acid, Chemosphere 66, 1494–1501. Gunasekara, A. S., Simpson, M. J., and Xing, B. (2003), Identification and characterization of sorption domains in soil organic matter using structurally modified humic acids, Environ. Sci. Technol. 37, 852–858. Gung, B. W. and Amicangelo, J. C. (2006), Substituent effects in C6F6-C6H5X stacking interactions, J. Org. Chem. 71, 9261–9270. ¨ . and Gschwend, P. M. (1997), Soot as a strong Gustafsson, O partition medium for polycyclic aromatic hydrocarbons in aquatic systems, in Molecular Markers in Environmental Geochemistry, Eganhouse, R. P., ed., American Chemical Society, Washington, DC, pp. 365–381. ¨ ., Haghseta, F., Chan, C., MacFarlane, J., and Gustafsson, O Gschwend, P. M. (1997), Quantification of the dilute sedimentary soot phase: Implications for PAH speciation and bioavailability. Environ. Sci. Technol. 31, 203–209. Hammes, K., Schmidt, M. W. I., Smernik, R. J., Currie, L. A., Ball, W. P., Nguyen, T. H., Louchouarn, P., Houel, S., Gustafsson, O.,

Elmquist, M., Cornelissen, G., Skjemstad, J. O., Masiello, C. A., Song, J., Peng, P., Mitra, S., Dunn, J. C., Hatcher, P. G., Hockaday, W. C., Smith, D. M., Hartkopf-Froeder, C., Boehmer, A., Luer, B., Huebert, B. J., Amelung, W., Brodowski, S., Huang, L., Zhang, W., Gschwend, P. M., Flores-Cervantes, D. X., Largeau, C., Rouzaud, J. N., Rumpel, C., Guggenberger, G., Kaiser, K., Rodionov, A., Gonzalez-Vila, F. J., Gonzalez-Perez, J. A., de la Rosa, J. M., Manning, D. A. C., Lopez-Capel, E., and Ding, L. (2007), Comparison of quantification methods to measure firederived (black/elemental) carbon in soils and sediments using reference materials from soil, water, sediment and the atmosphere, Global Biogeochem. Cycles 21. Hammes, K., Smernik, R. J., Skjemstad, J. O., and Schmidt, M. W. I. (2008), Characterisation and evaluation of reference materials for black carbon analysis using elemental composition, colour, BET surface area and C-13 NMR spectroscopy, J. Appl. Geochem. 23, 2113–2122. Harmon, T. C., Burks, G. A., Aycaguer, A.-C., and Jackson, K. (2001), Thermally enhanced vapor extraction for removing PAHs from lampblack-contaminated soil, J. Environ. Eng. 127, 986–993. Harris, P. J. F. and Tsang, S. C. (1997), High-resolution electron microscopy studies of non-graphitizing carbons, Philos. Mag. A 76, 667–677. Haws, N. W., Ball, W. P., and Bouwer, E. J. (2006), Modeling and interpreting bioavailability of organic contaminant mixtures in subsurface environments, J. Contam. Hydrol. 82, 255–292. Hawthorne, S. B., Grabanski, C. B., and Miller, D. J. (2007), Measured partition coefficients for parent and alkyl polycyclic aromatic hydrocarbons in 114 historically contaminated sediments: Part 2. Testing the k(oc)k(bc) two carbon-type model, Environ. Toxicol. Chem. 26, 2505–2516. Hayes, M. H. B. and Clapp, C. E. (2001), Humic substances: Considerations of compositions, aspects of structure, and environmental influences, Soil Sci. 166, 723–737. Hayes, M. H. B., MacCarthy, P., Malcolm, R. L.and Swift, R. S., eds. (1989), Humic Substances II. In Search of Structure. Wiley, Chichester, UK. Higgins, C. P. and Luthy, R. G. (2006), Sorption of perfluorinated surfactants on sediments., Environ. Sci. Technol. 40, 7251–7256. Hong, L., Jean, Y. C., Yang, H., Jordan, S. S., and Koros, W. J. (1996), Free-volume hole properties of gas-exposed polycarbonate studied by positron annihilation lifetime spectroscopy, Macromolecules 29, 7859–7864. Hu, W.-G., Mao, J., Xing, B., and Schmidt-Rohr (2000), Poly (methylene) crystallites in humic substances detected by nuclear magnetic resonance, Environ. Sci. Technol. 34, 530–534. Huang, W., Young, T., Schlautman, M. A., Yu, H., and Weber, W. J., Jr. (1997), A distributed reactivity model for sorption by soils and sediments. 9. General isotherm nonlinearity and applicability of the dual reactive domain model, Environ. Sci. Technol. 31, 1703–1710. Hummer, G., Garde, S., Garcia, A. E., Paulaitis, M. E., and Pratt, L. R. (1998), Hydrophobic effects on a molecular scale; doi:10.1021/jp982873 þ , J. Phys. Chem. B 102, 10469–10482.

REFERENCES

Hunter, C. A. (2004), Quantifying intermolecular interactions: Guidelines for the molecular recognition toolbox, Angew. Chem. Int. Ed. Engl. 43, 5310–5324. Hunter, C. A., Lawson, K. R., Perkins, J., and Urch, C. J. (2001), Aromatic interactions, J. Chem. Soc. Perkin Trans. 2, 651–669. Hunter, C. A. and Sanders, J. K. M. (1990), The nature of pi-pi interactions, J. Am. Chem. Soc. 112, 5525–5534. Iimori, T., Aoki, Y., and Ohshima, Y. (2002). S[sub 1]--S[sub 0] vibronic spectra of benzene clusters revisited. II. The trimer, J. Chem. Phys. 117, 3675–3686. Israelachvili, J. N. (1992), Intermolecular and Surface Forces, 2nd ed., Academic Press, London. Janiak, C. (2000), A critical account on p-p stacking in metal complexes with aromatic nitrogen-containing ligands, J. Chem. Soc. Dalton Trans. 3885–3896. Jeffrey, G. A. (1984), Inclusion Compounds, Academic Press, London. Johnson, M., KeinathII, T., and Weber, J. W. (2001), A distributed reactivity model from sorption by soils and sediments. 14. Characterization and modeling of phenanthrene desorption rates, Environ. Sci. Technol. 35, 1688–1695. Jonker, M. and Smedes, F. (2000), Preferential sorption of planar contaminants in sediments from Lake Ketelmeer, The Netherlands, Environ. Sci. Technol. 34, 1620–1626. Jonker, M. T. and Koelmans, A. A. (2002a), Extraction of polycyclic aromatic hydrocarbons from soot and sediment: Solvent evaluation and implications for sorption mechanism, Environ. Sci. Technol. 36, 4107–4113. Jonker, M. T. O. and Koelmans, A. A. (2002b), Sorption of polycyclic aromatic hydrocarbons and polychlorinated biphenyls to soot and soot-like materials in the aqueous environment: Mechanistic considerations, Environ. Sci. Technol. 36, 3725–3734. Jonker, M. T. O. and Barendregt, A. (2006), Oil is a sedimentary supersorbent for polychlorinated biphenyls, Environ. Sci. Technol. 40, 3829–3835. Jonker, M. T. O., Hawthorne, S. B., and Koelmans, A. A. (2005), Extremely slowly desorbing polycyclic aromatic hydrocarbons from soot and soot-like materials: evidence by supercritical fluid extraction, Environ. Sci. Technol. 39, 7885–7895. Jonker, M. T. O., Hoenderboom, A. M., and Koelmans, A. A. (2004), Effects of sedimentary sootlike materials on bioaccumulation and sorption of polychlorinated biphenyls, Environ. Toxicol. Chem. 23, 2563–2570. Kahle, M. and Stamm, C. (2007), Sorption of the veterinary antimicrobial sulfathiazole to organic materials of different origin. Environ. Sci. Technol. 41, 132–138. Kamens, R. M., Odum, J. R., and Fan, Z.-H. (1995), Some observations on times to equilibrium for semivolatile polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 29, 43–50. Kamiya, Y., Bourbon, D., Mizoguchi, K., and Naito, Y. (1992), Sorption, dilation, and isothermal glass transition of poly(ethyl methacrylate)-organic gas systems, Polymer 24, 443–449. Kamiya, Y., Hirose, T., Mizoguchi, K., and Naito, Y. (1986), Gravimetric study of high-pressure sorption of gases in polymers, J. Polym. Sci., Part B: Polym. Phys. 24, 1525–1539.

45

Kamiya, Y., Mizoguchi, K., Terada, K., Fujiwara, Y., and Wang, J.-S. (1998), CO2 sorption and dilation of poly(methyl methacrylate), Macromolecules 31, 472–478. Kan, A. T., Fu, G., Hunter, M., Chen, W., Ward, C. H., and Tomson, M. B. (1998), Irreversible sorption of neutral hydrocarbons to sediments: Experimental observations and model predictions, Environ. Sci. Technol. 32, 892–902. Kan, A. T., Fu, G., Hunter, M. A., and Tomson, M. B. (1997), Irreversible adsorption of naphthalene and tetrachlorobiphenyl to lula and surrogate sediments, Environ. Sci. Technol. 31, 2176–2186. Kaneko, K., Yamaguchi, K., Ishii, C., Ozeki, S., Hagiwara, S., and Suzuki, T. (1991), Size evaluation of graphitic crystallites in activated carbon fibers from diamagnetic susceptibility measurements, Chem. Phys. Lett. 176, 75–78. Karapanagioti, H., Childs, J., and Sabatini, D. A. (2001), Impacts of heterogeneous organic matter on phenanthrene sorption: Different soil and sediment samples, Environ. Sci. Technol. 35, 4684–4690. Karapanagioti, H. K., Kleineidam, S., Sabatini, D. A., Grathwohl, P., and Ligouis, B. (2000), Impacts of heterogeneous organic matter on phenanthrene sorption: Equilibrium and kinetic studies with aquifer material, Environ. Sci. Technol. 34, 406–414. K€arger, J. and Ruthven, D. M. (1992), Diffusion in Zeolites and Other Microporous Solids, Wiley, New York. Khalil, M. F., Ghosh, U., and Kreitinger, J. P. (2006), Role of weathered coal tar pitch in the partitioning of polycyclic aromatic hydrocarbons in manufactured gas plant site sediments, Environ. Sci. Technol. 40, 5681–5687. Kilduff, J. E. and Wigton, A. (1999), Sorption of TCE by humicpreloaded activated carbon: Incorporating size-exclusion and pore blockage phenomena in a competitive adsorption model, Environ. Sci. Technol. 33, 250–256. Kile, D. E., Wershaw, R. L., and Chiou, C. T. (1999), Correlation of soil and sediment organic matter polarity to aqueous sorption of nonionic compounds, Environ. Sci. Technol. 33, 2053–2056. Kleineidam, S., Rugner, H., and Grathwohl, P. (1999a), Influence of petrographic composition/organic matter distribution of fluvial aquifer sediments on the sorption of hydrophobic contaminants, Sediment. Geol. 129, 311–325. Kleineidam, S., R€ ugner, H, and Grathwohl, P. (1999b), The impact of grain scale heterogeneity on slow sorption kinetics, Environ. Toxicol. Chem. 18, 1673–1678. Knicker, H., Muffler, P., and Hilscher, A. (2007), How useful is chemical oxidation with dichromate for the determination of “black carbon” in fire-affected soils? Geoderma 142, 178–196. Koelmans, A. A., Meulman, B., Meijer, T., and Jonker, M. T. O. (2009), Attenuation of polychlorinated biphenyl sorption to charcoal by humic acids, Environ. Sci. Technol. 43, 736–742. Kukkonen, J. V. K., Landrum, P. F., Mitra, S., Gossiaux, D. C., Gunnarsson, J., and Weston, D. (2003), Sediment characteristics affecting desorption kinetics of select PAH and PCB congeners for seven laboratory spiked sediments, Environ. Sci. Technol. 37, 4656–4663.

46

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Kwon, J. H., Liljestrand, H. M., and Katz, L. E. (2006), Partitioning of moderately hydrophobic endocrine disruptors between water and synthetic membrane vesicles, Environ. Toxicol. Chem. 25, 1984–1992. Kwon, S. and Pignatello, J. J. (2005), Effect of natural organic substances on the surface and adsorptive properties of environmental black carbon (char): Pseudo pore blockage by model lipid components and its implications for N2-probed surface properties of natural sorbents, Environ. Sci. Technol. 39, 7932–7939. Lahaye, J. (1990), Mechanisms of soot formation, Polym. Degrad. Stab. 30, 111–121. Laor, Y. and Rebhun, M. (2002), Evidence for nonlinear binding of PAHs to dissolved humic acids, Environ. Sci. Technol. 36, 955–961. Lattao, C., Birdwell, J., Wang, J. J., and Cook, R. L. (2008), Studying organic matter molecular assemblage within a whole organic soil by nuclear magnetic resonance, J. Environ. Qual. 37, 1501–1509. Lazaridis, T. (2001), Solvent size vs cohesive energy as the origin of hydrophobicity, Acc. Chem. Res. 34, 931–937. LeBoeuf, E. J. and Weber, W. J., Jr. (1997), 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. 31, 1697–1702. Leenheer, J. A. (2009), Systematic approaches to comprehensive analysis of natural organic matter, Ann. Environ. Sci. 3, 1–130. Lei, L., Suidan, M. T., Khodadoust, A. P., and Tabak, H. H. (2004), Assessing the bioavailability of PAHs in field-contaminated sediment using XAD-2 assisted desorption, Environ. Sci. Technol. 38, 1786–1793. Li, J., Pignatello, J. J., Smets, B. F., Grasso, D., and Monserrate, E. (2005), Bench-scale evaluation of in situ bioremediation strategies for soil at a former manufactured gas plant site, Environ. Toxicol. Chem. 24, 741–749. Li, Q., Snoeyink, V. L., Mari~aas, B. J., and Campos, C. (2003), Elucidating competitive adsorption mechanisms of atrazine and NOM using model compounds, Water Resour. Res. 37, 773–784. Lighty, S. J., Veranth, J. M., and Sarofim, A. F. (2000), Combustion aerosols: Factors governing their size and composition and implications to human health, J. Air Waste Manage. Assoc. 50, 1565–1618. Loehr, R. C. and Webster, M. T. (1997), Environmentally Acceptable Endpoints in Soil, American Academy of Environmental Engineers, Annapolis, MD. Loke, M. L., Tjornelund, J., and Halling-Sorensen, B. (2002), Determination of the distribution coefficient (log K-d) of oxytetracycline, tylosin A, olaquindox and metronidazole in manure, Chemosphere 48, 351–361. Lu, Y. and Pignatello, J. J. (2002), Demonstration of the “conditioning effect” in soil organic matter in support of a pore deformation mechanism for sorption hysteresis, Environ. Sci. Technol. 36, 4553–4561. Lu, Y. and Pignatello, J. J. (2004a), History-dependent sorption in humic acids and a lignite in the context of a polymer model

for natural organic matter, Environ. Sci. Technol. 38, 5853– 5862. Lu, Y. and Pignatello, J. J. (2004b), Sorption of apolar aromatic compounds to soil humic acid particles affected by aluminum (III) ion cross-linking, J. Environ. Qual. 33, 1314–1321. Lucht, L. and Peppas, N. (1987), Macromolecular structure of coals. 2. Molecular weight between crosslinks from pyridine swelling experiments, Fuel 66, 803–809. Lucht, L. M., Larson, J. M., and Peppas, N. A. (1987), Macromolecular structure of coals.: IX. Molecular structure and glass transition temperature, Energy Fuels 1, 56–58. Luthy, R. G., Aiken, G. R., Brusseau, M. L., Cunningham, S. D., Gschwend, P. M., Pignatello, J. J., Reinhard, M., Traina, S. J., Weber, W. J., Jr., and Westall, J. C. (1997), Sequestration of hydrophobic organic contaminants by geosorbents, Environ. Sci. Technol. 31, 3341–3347. Lyon, W. G. (1995), Swelling of peats in liquid methyl, tetramethylene and propyl sulfoxides and in liquid propyl sulfone, Environ. Toxicol. Chem. 14, 229–236. MacKay, A. A. and Canterbury, B. (2005). Oxytetracycline sorption to organic matter by metal-bridging, J. Environ. Qual. 34, 1964–1971. Manes, M. (1998), Activated carbon adsorption fundamentals, in Encyclopedia of Environmental Analysis and Remediation, Meyers, R. A., ed., Wiley, New York, pp. 26–68. Mao, J.-D., Hundal, L. S., Thompson, M. L., and Schmidt-Rohr, K. (2002), Correlation of poly(methylene)-rich amorphous aliphatic domains in humic substances with sorption of a nonpolar organic contaminant, phenanthrene, Environ. Sci. Technol. 36, 929–936. Mao, J.-D. and Schmidt-Rohr, K. (2006), Absence of mobile carbohydrate domains in dry humic substances proven by NMR, and implications for organic-contaminant sorption models, Environ. Sci. Technol. 40, 1751–1756. Martin, D. S. (2003), The adsorption of aromatic acids onto the graphite basal surface, Surf. Sci. 536, 15–23. Masiello, C. A. and Druffel, E. R. M. (1998), Black carbon in deepsea sediments, Science 280, 1911–1913. McDermott, M. T. and McCreery, R. L. (1994), Scanning tunneling microscopy of ordered graphite and glassy carbon surfaces: Electronic control of quinone adsorption, Langmuir 10, 4307–4314. McGinley, P. M., Katz, L. E., and Weber, W. J. (1989), Multi-Solute Effects in the Sorption of Hydrophobic Organic Compounds by Aquifer Solutions, Presented before American Chemical Society, pp. 146–149. Merriam-Webster (2000), Unabridged Dictionary, Vol. 2.5. ed. Meyer, E. A., Castellano, R. K., and Diederich, F. (2003), Interactions with aromatic rings in chemical and biological recognition, Angew. Chem. Int. Ed. 42, 1210–1250. Milewska-Duda, J. (1993), The coal-sorbate system in the light of the theory of polymer solutions, Fuel 72, 419–425. Moore, F. G. and Richmond, G. L. (2008), Integration or segregation: How do molecules behave at oil/water interfaces? Acc. Chem. Res. 41, 739–748. Morimoto, T., Uno, H., and Furuta, H. (2007), Benzene ring trimer interactions modulate supramolecular structures, Angew. Chem. Int. Ed. 46, 3672–3675.

REFERENCES

M€ uller, E. A. and Gubbins, K. E. (1998), Molecular simulation study of hydrophilic and hydrophobic behavior of activated carbon surfaces, Carbon 36, 1433–1438. M€uller, E. A., Hung, F. R., and Gubbins, K. E. (2000), Adsorption of water vapor-methane mixtures on activated carbons, Langmuir 16, 5418–5424. M€uller-Wegener, U. (1987), Electron donor acceptor complexes between organic nitrogen heterocycles and humic acid, Sci. Total Environ. 62, 297–304. Newcomb, L. F. (1994), Aromatic stacking interactions in aqueous solution: Evidence that neither classical hydrophobic effects nor dispersion forces are important, J. Am. Chem. Soc. 116, 4993–4994. Newcombe, G., Drikas, M., and Hayes, R. (1997), Influence of characterised natural organic material on activated carbon adsorption: II. Effect on pore volume distribution and adsorption of 2-methylisoborneol, Water. Resour. Res. 31, 1065–1073. Nguyen, T. H. and Ball, W. P. (2006), Absorption and adsorption of hydrophobic organic contaminants to diesel and hexane soot, Environ. Sci. Technol. 40, 2958–2964. Nguyen, T. H., Brown, R. A., and Ball, W. P. (2004), An evaluation of thermal resistance as a measure of black carbon content in diesel soot, wood char, and sediment, Org. Geochem. 35, 217–234. Nguyen, T. H., Cho, H.-H., Poster, D. L., and Ball, W. P. (2007), Evidence for a pore-filling mechanism in the adsorption of aromatic hydrocarbons to a natural wood char, Environ. Sci. Technol. 41, 1212–1217. Nguyen, T. H., Goss, K.-U., and Ball, W. P. (2005), Polyparameter linear free energy relationships for estimating the equilibrium partition of organic compounds between water and the natural organic matter in soils and sediments, Environ. Sci. Technol. 39, 913–924. Niederer, C., Goss, K.-U. and Schwarzenbach, R. P. (2006a), Sorption equilibrium of a wide spectrum of organic vapors in leonardite humic acid: experimental setup and experimental data, Environ. Sci. Technol. 40, 5368–5373. Niederer, C., Goss, K.-U., and Schwarzenbach, R. P. (2006b), Sorption equilibrium of a wide spectrum of organic vapors in leonardite humic acid: Modeling of experimental data, Environ. Sci. Technol. 40, 5374–5379. ´ . B., Rainey, l. C., Feldermann, C. J., Sarofim, A. F., and Palot
as, A Vander Sande, J. B. (1996), Soot morphology: An application of image analysis in high-resolution transmission electron microscopy, Microsc. Res. Tech. 33, 266–278. Pan, B., Ghosh, S., and Xing, B. (2007), Nonideal binding between dissolved humic acids and polyaromatic hydrocarbons, Environ. Sci. Technol. 41, 6472–6478. Perry, M., Carra, C., Chretien, M. N., and Scaiano, J. C. (2007), Effect of hexafluorobenzene on the photophysics of pyrene, J. Phys. Chem. A 111, 4884–4889. Piatt, J. J. and Brusseau, M. L. (1998), Rate-limited sorption of hydrophobic organic compounds by soils with well-characterized organic matter, Environ. Sci. Technol. 32, 1604–1608. Pignatello, J. J. (1990a), Slowly reversible sorption of aliphatic halocarbons in soils. I. Formation of residual fractions, Environ. Toxicol. Chem. 9, 1107–1115.

47

Pignatello, J. J. (1990b), Slowly reversible sorption of aliphatic halocarbons in soils. II. Mechanistic aspects, Environ. Toxicol. Chem. 9, 1117–1126. Pignatello, J. J. (1991), Competitive effects in the sorption of nonpolar organic compounds by soils, in Organic Substances and Sediments in Water, Baker, R. A., ed., Vol. 1, Humics and Soils, Lewis Publishers, Chelsea, MI, pp. 291–307. Pignatello, J. J. (2000), The measurement and interpretation of sorption and desorption rates for organic compounds in soil media, in Advances in Agronomy, Sparks, D. L., ed., Vol. 69, Academic Press, San Diego, pp. 1–73. Pignatello, J. J. (2006), Fundamental issues in sorption related to physical and biological remediation of soils, in Soil and Water Pollution Monitoring, Protection and Remediation, Springer, New York, pp. 3–23. Pignatello, J. J. (2009), Bioavailability of contaminants in soil, in Advances in Applied Bioremediation, Soil Biology, Singh, A., ed., Springer-Verlag, Berlin Heidelberg, Vol. 17, Chap. 2, pp. 35–71. Pignatello, J. J., Ferrandino, F. J., and Huang, L. Q. (1993), Elution of aged and freshly added herbicides from a soil, Environ. Sci. Technol. 27, 1663–1671. Pignatello, J. J., Frink, C. R., Marin, P. A., and Droste, E. X. (1990), Field-observed ethylene dibromide in an aquifer after two decades, J. Contam. Hydrol. 5, 195–214. Pignatello, J. J., Kwon, S., and Lu, Y. (2006a), Effect of natural organic substances on the surface and adsorptive properties of environmental black carbon (char): Attenuation of surface activity by humic and fulvic acids, Environ. Sci. Technol. 40, 7757–7763. Pignatello, J. J., Lu, Y., LeBoeuf, E. J., Huang, W., Song, J., and Xing, B. (2006b), Nonlinear and competitive sorption of apolar compounds in black carbon-free natural organic materials, J. Environ. Qual. 35. 1049–1059. Pignatello, J. J. and Xing, B. (1996), Mechanisms of slow sorption of organic chemicals to natural particles, Environ. Sci. Technol. 30, 1–11. Polubesova, T., Sherman-Nakache, M., and Chefetz, B. (2007), Binding of pyrene to hydrophobic fractions of dissolved organic matter: Effect of polyvalent metal complexation, Environ. Sci. Technol. 41, 5389–5394. Poole, S. K. and Poole, C. F. (1999), Chromatographic models for the sorption of neutral organic compounds by soil from water and air, J. Chromatogr. A 845, 381–400. Qiu, Y., Xiao, X., Cheng, H., Zhou, Z., and Sheng, G. D. (2009), Influence of environmental factors on pesticide adsorption by black carbon: pH and model dissolved organic matter, Environ. Sci. Technol. 43, 4973–4978. Qu, X. L., Liu, P., and Zhu, D. Q. (2008), Enhanced sorption of polycyclic aromatic hydrocarbons to tetra-alkyl ammonium modified smectites via cation-pi interactions, Environ. Sci. Technol. 42, 1109–1116. Radke, C. J. and Prausnitz, J. M. (1972), Thermodynamics of multisolute adsorption from dilute liquid solutions, Am. Inst. Chem. Eng. J. 18, 761–768. Ramanathan, V. and Carmichael, G. (2008), Global and regional climate changes due to black carbon, 1, 221–227.

48

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Ran, Y., Sun, K., Yang, Y., Xing, B. S, and Zeng, E. (2007), Strong sorption of phenanthrene by condensed organic matter in soils and sediments, Environ. Sci. Technol. 41, 3952–3958. Razouk, R., Saleeb, E., and Said, E. (1968), The heat of wetting and immersional swelling of charcoal, J. Colloid Interface Sci. 28, 487–492. Rogers, C. E. (1965), Solubility and diffusivity, in Physics and Chemistry of the Organic Solid State, Fox, D., Labes, M. M., and Weissberger, A., eds., Interscience Publishers, New York, Vol. II, pp. 509–635. Rouquerol, F., Rouquerol, J., and Sing, K. (1999), Adsorption by Powders and Porous Solids, Academic Press, San Diego. Rutherford, D. W. and Chiou, C. T. (1992), Effect of water saturation in soil organic matter on the partition of organic compounds, Environ. Sci. Technol. 26, 965–970. Sander, M., Lu, Y., and Pignatello, J. J. (2005), A thermodynamically based method to quantify sorption hysteresis, J. Environ. Qual. 34, 1063–1072. Sander, M., Lu, Y, and Pignatello, J. J. (2006), Conditioning annealing studies of natural organic matter solids linking irreversible sorption to irreversible structural expansion, Environ. Sci. Technol. 40, 170–178. Sander, M. and Pignatello, J. J. (2005a), Charcterization of charcoal adsorption sites for aromatic compounds: insights drawn from single-solute and bi-solute competitive experiments, Environ. Sci. Technol. 39, 1606–1615. Sander, M. and Pignatello, J. J. (2005b), An isotope exchange technique to assess mechanisms of sorption hysteresis applied to naphthalene in kerogenous organic matter, Environ. Sci. Technol. 39, 7476–7484. Sander, M. and Pignatello, J. J. (2007), On the reversibility of sorption to black carbon: Distinguishing true hysteresis from artificial hysteresis caused by dilution of a competing adsorbate, Environ. Sci. Technol. 41, 843–849. Sander, M. and Pignatello, J. J. (2009), Sorption irreversibility of 1,4-dichlorobenzene in two natural organic matter-rich geosorbents, Environ. Toxicol. Chem. 28, 447–457. Sassman, S. A. and Lee, L. S. (2005), Sorption of three tetracyclines by several soils: Assessing the role of pH and cation exchange, Environ. Sci. Technol. 39, 7452–7459. Schaefer, C., Schuth, C., Werth, C., and Reinhard, M. (2000), Binary desorption isotherms of TCE and PCE from silicia gel and natural solids, Environ. Sci. Technol. 34, 4341– 4347. Schaumann, G. E. and Antelmann, O. (2000), Thermal characteristics of soil organic matter measured by DSC: A hint on a glass transition, J. Plant Nutr. Soil Sci. 163, 179–181. Schaumann, G. E., Hobley, E., Hurrass, J., and Rotard, W. (2005), H-NMR relaxometry to monitor wetting and swelling kinetics in high-organic matter soils, Plant Soil 275, 1–20. Schaumann, G. E. and LeBoeuf, E. J. (2005), Glass transitions in peat: Their relevance and the impact of water, Environ. Sci. Technol. 39, 800–806. Schellenberg, K., Leuenberger, C., and Schwarzenbach, R. P. (1984), Sorption of chlorinated phenols by natural sediments

and aquifer materials; doi:10.1021/es00127a005, Environ. Sci. Technol. 18, 652–657. Schmidt, M. W. I. and Noack, A. G. (2000), Black carbon in soils and sediments: Analysis, distribution, implications, and current challenges, Global Biogeochem. Cycles 14, 777–793. Schmidt-Mende, L., Fechtenkotter, A., Mullen, K., Moons, E., Friend, R. H., and MacKenzie, J. D. (2001), Self-organized discotic liquid crystals for high-efficiency organic photovoltaics, Science 293, 1119–1122. Schnitzer, M. and Khan, S. U. (1978), Soil Organic Matter, Elsevier, New York. Schulten, H. -R. and Schnitzer, M. (1997), Chemical model structures for soil organic matter and soils, Soil Sci. 162, 115–130. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (2002), Environmental Organic Chemistry, 2nd ed., Wiley, New York. Senesi, N., D’Orazio, V., and Miano, T. M. (1995), Adsorption mechanisms of s-triazine and bipyridylium herbicides on humic acids from hop field soils, Geoderma 66, 273–283. Sheindorf, C., Rebhun, M., and Sheintuch, M. (1981), A Freundlichtype multicomponent isotherm, J. Colloid Interface Sci. 79, 136–142. Shibuya, M., Kato, M., Ozawa, M., Fang, P. H., and Osawa, E. (1999), Detection of buckminsterfullerene in usual soots and commercial charcoals, Fullerene Sci. Technol. 7, 181–193. Shih, Y. and Gschwend, P. M. (2009), Evaluating activated carbonwater sorption coefficients of organic compounds using a linear solvation energy relationship approach and sorbate chemical activities, Environ. Sci. Technol. 43, 851–857. Shor, L. M., Rockne, K. J., Taghon, G. L., Young, L. Y., and Kosson, D. S. (2003), Desorption kinetics for field-aged polycyclic aromatic hydrocarbons from sediments, Environ. Sci. Technol. 37, 1535–1544. Silverstein, K. A. T., Haymet, A. D. J., and Dill, K. A. (2000), The strength of hydrogen bonds in liquid water and around nonpolar solutes, J. Am. Chem. Soc. 122, 8037–8041. Sims, G. K. and O’Loughlin, E. J. (1989), Degradation of pyridines in the environment, Crit. Rev. Environ. Control 19, 309–340. Sinnokrot, M. O. and Sherrill, C. D. (2006), High-accuracy quantum mechanical studies of pi-pi interactions in benzene dimers, J. Phys. Chem. A 110, 10656–10668. Skjemstad, J. O., Taylor, J. A., and Smernik, R. J. (1999), Estimation of charcoal (char) in soils, Commun. Soil Sci. Plant Anal. 30, 2283–2298. Smedley, J. M., Williams, A., and Bartle, K. D. (1992), A mechanism for the formation of soot particles and soot deposits, Combust. Flame 91, 71–82. Smernik, R. J., Kookana, R. S., and Skjemstad, J. O. (2006), NMR characterization of 13-C-benzene sorbed to natural and prepared charcoals, Environ. Sci. Technol. 40, 1764–1769. Southall, N. T., Dill, K. A., and Haymet, A. D. J. (2002), A view of the hydrophobic effect, J. Phys. Chem. B 106, 521–533. Sposito, G., Martin-Neto, L., and Yang, A. (1996), Atrazine complexation of soil humic acids, J. Environ. Qual. 25, 1203–1209.

REFERENCES

SRC Syracuse Research Corporation Interactive LogKow (KowWin) Demo; available at http://www.srcinc.com/what-we-do/ databaseforms.aspx?id¼385; accessed 4/14/09. Steed, J. M., Dixon, T. A., and Klemperer, W. (1979), Molecularbeam studies of benzene dimer, hexafluorobenzene dimer, and benzene-hexafluorobenzene, J. Chem. Phys. 70, 4940– 4946. Steinberg, S. M., Pignatello, J. J, and Sawhney, B. L. (1987), Persistence of 1,2-dibromoethane in soils: Entrapment in intraparticle micropores, Environ. Sci. Technol. 21, 1201–1208. Stevenson, F. (1994), Humus Chemistry: Genesis, Composition, Reactions, 2nd ed., Wiley, New York. Strommen, M. R. and Kamens, R. M. (1997), Development and application of a dual-impedance radial diffusion model to simulate the partitioning of semivolatile organic compounds in combustion aerosols, Environ. Sci. Technol. 31, 2983–2990. Sutton, R. and Sposito, G. (2005), Molecular structure in soil humic substances: The new view, Environ. Sci. Technol. 39, 9009–9015. Teixido´ Planes, M. and Pignatello, J. J. (unpublished), Sorption of the antibiotic sulfamethazine to (biochar) black carbon. Ten Hulscher, T. E. M., Vrind, B. A., Van den Heuvel, H., Van Noort, P. C. M., and Govers, H. A. J. (2005), Influence of long contact time on sediment sorption kinetics of spiked chlorinated compounds, Environ. Toxicol. Chem. 24, 2154–2159. van Noort, P. C. M. (2003), A thermodynamics-based estimation model for adsorption of organic compounds by carbonaceous materials in environmental sorbents, Environ. Toxicol. Chem. 22, 1179–1188. van Noort, P. C. M., Cornelissen, G., Ten Hulscher, T. E. M., and Belfoid, A. (2002), Influence of sorbate planarity on the magnitude of rapidly desorbing fractions of organic compounds in sediment, Environ. Toxicol. Chem. 21, 2326–2330. Vrbancich, J. and Ritchie, G. L. D. (1980), Quadrapole moments of benzene, hexafluorobenzene and other non-dipolar aromatic molecules, J. Chem. Soc. Faraday Trans. 2 76, 648–659. Walters, R. W. and Luthy, R. G. (1984), Equilibrium adsorption of polycyclic aromatic hydrocarbons from water onto activated carbon, Environ. Sci. Technol. 18, 395–403. Wang, J.-S., Kamiya, Y. and Naito, Y. (1998), Effects of CO2 conditioning on sorption, dilation, and transport properties of polysulfone, J. Polym. Sci., Part B: Polym. Phys. 36, 1695–1702. Weber, W. J. Jr., Kim, S. H., and Johnson, M. D. (2002), Distributed reactivity model for sorption by soils and sediments.15. Highconcentration co-contaminant effects on phenanthrene sorption and desorption, Environ. Sci. Technol. 36, 3625–3634. Weber, W. J. J., McGinley, P. M., and Katz, L. E. (1992), A distributed reactivity model for sorption by soils and sediments. 1. Conceptual basis and equilibrium assessments. Environ. Sci. Technol. 26, 1955–1962. Welhouse, G. J. and Bleam, W. F. (1993a), Atrazine hydrogenbonding potentials, Environ. Sci. Technol. 27, 494–500. Welhouse, G. J. and Bleam, W. F. (1993b), Cooperative hydrogen bonding of atrazine, Environ. Sci. Technol. 27, 500–505.

49

Wen, B., Huang, R. X., Li, R. J., Gong, P., Zhang, S., Pei, Z. G., Fang, J., Shan, X. Q., and Khan, S. U. (2009), Effects of humic acid and lipid on the sorption of phenanthrene on char, Geoderma 150, 202–208. Wen, B., Zhang, J. J., Zhang, S. Z., Shan, X. Q., Khan, S. U., and Xing, B. S. (2007), Phenanthrene sorption to soil humic acid and different humin fractions, Environ. Sci. Technol. 41, 3165–3171. Werth, C. J. and Hansen, K. M. (2002), Modeling the effects of concentration history on the slow desorption of trichloroethene from a soil at 100% relative humidity, J. Contam. Hydrol. 54, 307–327. Werth, C. J. and Reinhard, M. (1997a), Effects of temperature on trichloroethylene desorption from silica gel and natural sediments. 1. Isotherms, Environ. Sci. Technol. 31, 689–696. Werth, C. J. and Reinhard, M. (1997b), Effects of temperature on trichloroethylene desorption from silica gel and natural sediments. 2. Kinetics, Environ. Sci. Technol. 31, 697–703. White, J. C., Hunter, M., Nam, K., Pignatello, J. J., and Alexander, M. (1999a), Correlation between the biological and physical availabilities of phenanthrene in soils and soil humin in aging experiments, Environ. Toxicol. Chem. 18, 1720–1727. White, J. C., Hunter, M., Nam, K., Pignatello, J. J., and Alexander, M. (1999b), Correlation between the biologican and physical availabilities of phenanthrene in soils and soil humin in aging experiments, Environ. Toxicol. Chem. 18, 1720–1727. White, J. C., Hunter, M., Pignatello, J. J., and Alexander, M. (1999c), Increase in the bioavailability of aged phenanthrene in soils by competitive displacement with pyrene, Environ. Toxicol. Chem. 18, 1728–1732. Wijnja, H., Pignatello, J. J., and Malekani, K. (2004), Formation of pi-pi complexes with phenanthrene and model pi-acceptor humic subunits, J. Environ. Qual. 33, 265–275. Williams, J. H. (1993), The molecular quadrapole moment and solid-state architecture, Acc. Chem. Res. 26, 593–598. Wu, J. Z. and Prausnitz, J. M. (2008), Pairwise-additive hydrophobic effect for alkanes in water, Proc. Natl. Acad. Sci. USA 105, 9512–9515. Wu, S. and Gschwend, P. M. (1986), Sorption kinetics of hydrophobic organic compounds to natural sediments and soils, Environ. Sci. Technol. 20, 717–725. Xia, G. and Ball, W. (2000), Polanyi-based models for the competitive sorption of low-polarity organic contaminants on a natural sorbent, Environ. Sci. Technol. 34, 1246–1253. Xia, G. and Ball, W. P. (1999), Adsorption-partitioning uptake of nine low-polarity organic chemicals on a natural sorbent, Environ. Sci. Technol. 33, 262–269. Xia, G. and Pignatello, J. J. (2001), Detailed sorption isotherms of polar and apolar compounds in a high-organic soil, Environ. Sci. Technol. 35, 84–94. Xing, B., Gigliotti, B., and Pignatello, J. J. (1996), Competitive sorption between atrazine and other organic compounds in soils and model sorbents, Environ. Sci. Technol. 30, 2432–2440. Xing, B. and Pignatello, J. J. (1997), Dual-mode sorption of lowpolarity compounds in glassy poly(vinyl chloride) and soil organic matter, Environ. Sci. Technol. 31, 792–799.

50

INTERACTIONS OF ANTHROPOGENIC ORGANIC CHEMICALS

Xing, B. and Pignatello, J. J. (1998), Competitive sorption between 1,3-dichlorobenzene or 2,4-dichlorophenol and natural aromatic acids in soil organic matter, Environ. Sci. Technol. 32, 614–619. Xiong, J. C. and Maciel, G. E. (2002), Interactions between pyridine and coal at the molecular level: Insights from variable-temperature H-1 NMR studies of pyridine-saturated coal, Energy Fuels 16, 497–509. Yamamoto, H. and Liljestrand, H. M. (2004), Partitioning of selected estrogenic compounds between synthetic membrane vesicles and water: Effects of lipid components, Environ. Sci. Technol. 38, 1139–1147. Young, T. M. and Weber, W. J., Jr. (1995), A distributed reactivity model for sorption by soils and sediments. 3. Effects of diagenetic processes on sorption energetics, Environ. Sci. Technol. 29, 92–97. Yun, Y. and Suuberg, E. M. (1993), New applications of differential scanning calorimetry and solvent swelling for studies of coal structure: Prepyrolysis structural relaxation, Fuel 72, 1245–1254. Zhang, L., Leboeuf, E. J., and Xing, B. S. (2007), Thermal analytical investigation of biopolymers and humic- and carbonaceousbased soil and sediment organic matter, Environ. Sci. Technol. 41, 4888–4894.

Zhao, D. and Pignatello, J. J. (2004), Model-aided characterization of tenax-TA for aromatic compound uptake from water, Environ. Toxicol. Chem. 23, 1592–1599. Zhao, D., Pignatello, J. J., White, J. C., Braida, W., and Ferrandino, F. (2001), Dual-mode modeling of competitive and concentrationdependent sorption and desorption kinetics of polycyclic aromatic hydrocarbons in soils, Water Resour. Res. 37, 2205–2212. Zhu, D., Hyun, S., Pignatello, J. J., and Lee, L. S. (2004), Evidence for pi-pi electron donor-acceptor interactions between pi-donor aromatic compounds and pi-acceptor sites in soil organic matter, Environ. Sci. Technol. 38, 4361–4368. Zhu, D., Kwon, S., and Pignatello, J. J. (2005), Adsorption of singlering organic compounds to wood charcoals prepared under different thermochemical conditions, Environ. Sci. Technol. 39, 3990–3998. Zhu, D. and Pignatello, J. J. (2005a), Characterization of aromatic compound sorptive interactions with black carbon (charcoal) assisted by graphite as a model, Environ. Sci. Technol. 39, 2033–2041. Zhu, D. and Pignatello, J. J. (2005b), A concentration-dependent multi-term linear free energy relationship for sorption of organic compounds to soils based on the hexadecane dilute-solution reference state, Environ. Sci. Technol. 39, 8817–8828.

2 COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS: METHODOLOGIES, MECHANISMS, AND ENVIRONMENTAL IMPLICATIONS STEPHEN A. BOYD, CLIFF T. JOHNSTON, DAVID A. LAIRD, BRIAN J. TEPPEN, 2.1. Introduction 2.2. Contributions of Clays Versus Soil Organic Matter to Sorption 2.3. Macroscopic Sorption Studies 2.4. X-Ray Diffraction 2.5. Vibrational Spectroscopic Studies 2.6. Molecular and Quantum Mechanical Simulations 2.7. Environmental Implications 2.8. Summary

2.1. INTRODUCTION The publication of two papers in 1979 (Chiou et al. 1979; Karickhoff et al. 1979) marked the beginning of a major redirection of thinking regarding the sorption of neutral organic contaminants (NOCs) and pesticides in soil/sediment–water systems. Prior to these important papers, soil was generally viewed as an adsorbent containing two highsurface-area components, soil organic matter (SOM) and soil clays, that were responsible for the adsorption of NOCs and pesticides by a variety of mechanisms (e.g., H bonding, van der Waals forces) unique to the particular combination of soil and the organic solute of concern. The 1979 papers set forth a new conceptualization of NOC sorption by soils, which has come to be known as partition theory. With the development of partition theory, the idea that SOM played a dominant, if not singular, role in the sorption of NOCs and pesticides by soils also gained acceptance. Soil organic matter was viewed as an organic partition

AND

HUI LI

phase rather than a high-surface-area adsorbent (Pennell et al. 1995). Accordingly, organic solutes in water would essentially dissolve in amorphous SOM in much the same way that they partition from water into an immiscible organic phase such as octanol or hexane, or into organic polymers/solids such as polyurethane or rubber. The magnitude of sorption would depend directly on the amount of SOM as well as the solubility of the NOC or pesticide in water and its solubility in SOM. Solutes with low water solubility partition to a greater degree into SOM manifesting a larger sorption or partition coefficient (often denoted Kp or Kd). Sorption is quantified by this coefficient, which is in essence a simple ratio of the two concentrations: Kp ¼ Csoil/Cwater, where Csoil and Cwater are the solute concentrations in SOM and water, respectively. Furthermore, when the individual Kp values (for a given NOC) among a series of soils are normalized to the corresponding fractional SOM content (fom) of the soil, the normalized values tend to converge within a factor of 2 (Kile et al. 1995), especially for NOCs devoid of polar functional groups. The SOMnormalized value, Kom ¼ Kp/fom, represents sorption per unit mass of SOM, and can for all practical purposes be considered a “constant.” Once the Kom of a NOC is known, Kp for soils can be easily estimated by Kp ¼ Komfom, that is, simply by knowing the fom of a soil. For context, owing to the semipolar nature of SOM (which has an O content of 40%–50% derived from polar functional groups such as phenolic ---OH and ---COOH groups), the Kom value of a specific NOC/pesticide is generally smaller than the corresponding octanol–water partition coefficient (Kow) by approximately an order of magnitude. Today, Kom is routinely used to predict Kp in NOC/pesticide leaching models; Kp

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

51

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

defines site-specific soil–water distribution for a given NOC–soil combination. Solute partitioning, which has now gained broad acceptance as a sorption mechanism, manifests certain sorption characteristics that distinguish it from adsorption (Chiou 2002; Chiou et al. 1983). Partitioning is a process of dissolution into the partition phase, whereas adsorption involves condensation of a solute on the surface of the adsorbent. Solute partitioning manifests linear sorption isotherms, noncompetitive sorption effects in multisolute systems (two or more NOCs sorbing simultaneously), and low heat effects on sorption (i.e., low and constant enthalpies). Solute adsorption manifests (generally) nonlinear isotherms, competitive effects (negative) on sorption in multisolute systems, and comparatively large negative enthalpies (or large heat effects on sorption). Considerable evidence has been provided to demonstrate the characteristics of solute partitioning in numerous studies of sorption of NOCs from water by soils and sediments (Chiou 2002). Further examination of NOC sorption by soil has revealed some evidence of sorption isotherm nonlinearity, and competitive effects on sorption in multisolute systems, especially at low relative solute concentrations (the ratio of concentration of solute in water to aqueous solubility of the solute). These effects, which are typically of low magnitude, have been reconciled by several theories that augment the fundamental process of solute partitioning. One such theory suggests that small amounts of high-surface-area carbonaceous materials (e.g., chars, black carbon) are responsible for the observed deviations in sorption behavior predicted from partition theory (Chiou et al. 2000). Also, SOM has been viewed as a dual sorbent containing both “hard” and “soft” components with the former responsible for the deviations from sorption characteristics predicted by strict solute partitioning (e.g., Pignatello and Xing 1996; Leboeuf and Weber 1997; Xia and Ball 2000). These ideas are best viewed as further refinements that augment the concept of solute partitioning as it pertains to NOC sorption by soils and sediments rather than a challenge to the basic idea of NOC partitioning into SOM as the dominant sorptive mechanism. One very important manifestation of the widespread acceptance of partition theory to describe NOC sorption by soils is that most NOC/pesticide leaching models now use simple linear partition coefficients (often estimated by Kp ¼ Komfom) to predict soil–water distribution of NOCs and pesticides. This approach implicitly ignores or discounts any possible contribution of soil clays to sorption. In this context some proponents of partition theory have suggested that soil mineral surfaces are in general polar or hydrophilic in nature, and that in the presence of ambient moisture, water molecules are preferentially adsorbed on these surfaces (Chiou 2002; Chiou and Shoup 1985). Accordingly, since NOCs lack the ability to displace strongly adsorbed water molecules from the mineral surfaces, the mineral compo-

nents are in essence deactivated as adsorbent surfaces for NOCs and pesticides. Evidence has been provided to demonstrate that the addition of water to anhydrous soils results in a diminution in the overall uptake of certain NOCs and pesticides by soils. These data were interpreted to mean that mineral surfaces (e.g., clays) are active solid adsorbents for NOCs in the absence of water, and that under such conditions both mineral phase adsorption and organic matter partitioning are operative sorption processes. However, as water is added, it displaces adsorbed NOCs and pesticides from mineral surfaces (which preferentially bind water) hence sorption overall is reduced and partitioning into SOM is left as the dominant process responsible for the sequestration of NOCs and pesticides in soil–water systems. At this point in the development of our understanding of NOC/pesticide sorption by soils, dogma held that minor “unsuppressed” sorption of NOCs to clays may occur, but that it was dwarfed by sorption to SOM. Then, in 1992 another important paper was published (Laird et al. 1992) showing that relatively pure smectite clays could effectively adsorb atrazine from water. Interestingly, atrazine adsorption by these clays ranged widely, from essentially complete removal to comparatively minimal uptake (Fig. 2.1). This variability was observed by using many different members of the group of smectite clays. These clays possess subtle differences in composition owing to differences in isomorphic substitution in the tetrahedral Si---O and/or octahedral 200

150 x/m (µmol kg-1)

52

100

50

0 0

10

20

30

40

50

-1)

CE (µmol L

Figure 2.1. Adsorption of atrazine from water by several different reference and soil smectite clays [from Laird et al. (1992)]. Plots are of concentrations of sorbed atrazine (x/m) versus equlibrium aqueous atrazine concentration (CE). Sorbents (in descending order) are as follows: Panther Creek beidellite, hectorite, IMV saponite, Amory montmorillonite, Wyoming bentonite, Belle Fourche montmorillonite, Upton montmorillonite, Webster smectite, Polkville montmorillonite, IMV bentonite, Chambers montmorillonite, Camp Bertean montmorillonite, Carmeron smectite/illite, and Otay montmorillonite.

CONTRIBUTIONS OF CLAYS VERSUS SOIL ORGANIC MATTER TO SORPTION

53

TABLE 2.1. Structural Formulas for Dioctahedral and Trioctahedral Clays with either Tetrahedral or Octahedral Isomorphic Substitution Substitution

Dioctahedral

Trioctahedral

Tetrahedral

Beidellite: Na0:33 Al2 ðSi3:67 Al0:33 ÞO10 ðOHÞ2 Nontronite Na0:33 Fe2 ðSi3:67 Al0:33 ÞO10 ðOHÞ2 Montmorillonite Na0:33 ðAl1:67 Mg0:33 ÞSi4 O10 ðOHÞ2

Saponite: Na0:33 Mg3 ðSi3:67 Al0:33 ÞO10 ðOHÞ2

Octahedral

Hectorite Na0:33 ðMg2:67 Li0:33 ÞSi4 O10 ðOHÞ2

Source: Adapted from Gaines et al. (1997).

Al---O sheets of the 2 : 1 clays, and the type (dioctahedral vs. trioctahedral) of octahedral sheet. Specific structures of important members of the smectite group are given in Table 2.1. Remarkably, these subtle differences manifested very large differences in the affinity of particular smectite clays for aqueous-phase atrazine. Soon thereafter, an important series of papers (Haderlein and Schwarzenbach 1993; Haderlein et al. 1996; Weissmahr et al. 1997) similarly showed that clay minerals common to soils, namely, kaolinite, illite, and smectite, could also effectively adsorb nitroaromatic compounds (NACs) from water. Among these, smectite clays were by far the most effective NAC adsorbents. From these and similar studies several questions arise: whether (1) the role of soil clays as effective adsorbents of NOCs/pesticides in soil–water systems been erroneously neglected; (2) when and under what conditions clay minerals are effective adsorbents for NOCs and pesticides in soils; (3) the specific underlying forces and mechanisms by which clays, especially smectite clays, functioned as highly effective adsorbents for NOCs; and (4) how one goes about revealing these molecular-scale forces and mechanisms?

2.2. CONTRIBUTIONS OF CLAYS VERSUS SOIL ORGANIC MATTER TO SORPTION In an attempt to address the first question, we conducted a study that directly compared the potential contributions of SOM and montmorillonite clay as sorbents for several important examples of NOCs and pesticides (Sheng et al. 2001). We used an organic soil (Houghton muck) essentially devoid of clay minerals (having an organic carbon content of 49.5%) as a “surrogate” for pure SOM. This soil, where SOM was the singular (organic) sorbent for NOCs/pesticides, was compared to potassium-saturated montmorillonite, which represented a purely clay mineral adsorbent. On a unit mass basis, SOM was a more effective (to varying extents) sorbent than was K-montmorillonite for biphenyl, diuron, and parathion. However, for atrazine, carbaryl, dichlobenil, and dinitro-ocresol, K-montmorillonite was more effective than SOM as a sorbent phase. Since the clay content of most mineral soils significantly exceeds the SOM content, overall sorption of certain NOCs and pesticides could plausibly be controlled by their interactions with clays, especially the expandable smectites, which possess particularly high surface areas of

800 m2/g. Among these pesticides, specific interactions between pesticide substituents and exchangeable cations of clay, hydrophobic and/or electron donor–acceptor interactions between the aromatic rings of pesticide molecule and the siloxane surfaces of clays, and steric hindrance due to bulky substituents on the pesticide structure seemed to be important determinants of the extents of pesticide adsorption by K-montmorillonite. The occurrence of significant clay–organic interactions observed in the study by Sheng et al. (2001) was consistent with earlier studies that had shown appreciable adsorption of certain pesticides by clays (Bailey and White 1970; Green 1974; Mortland 1986). Also, previous investigations had sought to define conditions under which clay minerals contributed significantly to NOC/pesticide retention in soils, and concluded that this could occur when clay to SOM ratios in soil were greater than 5 to 30 (Hassett et al. 1981; Karickhoff 1984). In another study, the critical clay to organic carbon ratios at which mineral phase sorption accounted for 50% of overall sorption by soils and sediments were ca. 62 for atrazine and 84 for alachlor (Grundl and Small 1993). In retrospect, these seem like fairly gross estimates of soil properties that could be used to benchmark the important role of clays in pesticide retention by soils. More recent studies clearly indicate a much more specific set of conditions that manifest high-affinity clay adsorbents that are unique to the particular combination of solute properties and structure as well as clay type and type of exchangeable cation associated with the clay. Following demonstration that for several classes of important NOCs and pesticides smectite clays are equally or more effective adsorbents than SOM (compared as isolated components), the next logical line of investigation was to reveal the underlying operative mechanisms and forces responsible for adsorption by smectite clays. These clays have been the primary focus of this body of research because of their widespread occurrence, large surface areas, and reversible expandability. These clays consist of a basic 2 : 1 layer structure with an octahedrally coordinated AlO sheet sandwiched between two tetrahedrally coordinated SiO sheets. The specific members of this group of clays differ in being either dioctahedral or trioctahedral, and on the type, location, and extent of isomorphic substitution. For example, saponite is a trioctahedral clay (MgO central sheets) with nearly 100% of the isomorphic substitution

54

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

TABLE 2.2. Characteristics of Smectite Clay Minerals Used in More Recent Studies as Adsorbents for NOCs and Pesticides

Clay

Smectite Clay Type and Octahedral Sheet Type

SWy-2 SAz-1 Upton SHCa-1 SapCa-2

Montmorillonite, dioctahedral Montmorillonite, dioctahedral Montmorillonite, dioctahedral Hectorite, trioctahedral Saponite, trioctahedral

Tetrahedral Charge, %

Cation Exchange Capacity cmol/kg

Surface Area m2/g

BET N2 Surface Area, m2/g

OC %

Surface Charge Density, mmol/m2

3.6 12 1.6 14 100

83.6 130 113.3 86.4 97.4

766 768 730 743 750

31.82 — 39.8 63.19 —

0.07 — 0.06 <0.1 0.128

1.09 — 1.55 1.16 1.30

Source: Adapted from Liu et al. (2009).

(Al3 þ for Si4 þ) occurring in the tetrahedral sheets. Other smectites, such as montmorillonite, are dioctahedral with substitution (e.g., Mg2þ for Al3þ ) primarily in the octahedral layer. Tables 2.1 and 2.2 present some key structural properties and characteristics that differentiate members of the family of smectite clays, and that serve as determinants of their affinities for NOCs and pesticides. Nitroaromatics are an important class of compounds used widely as energetics, solvents, fragrances, and base structure for pesticide classes such as the dinitrophenols, and occur widely as environmental contaminants (e.g., in soils). They are also strongly adsorbed by smectite clays (Haderlein and Schwarzenbach 1993; Haderlein et al. 1996; Boyd et al. 2001; Sheng et al. 2002; Li et al. 2003), and for these reasons they have been subjected to several studies in an effort to understand the basis of their strong affinity for smectites. Initial studies of these compounds, which have revealed several interesting and important sorption phenomena, relied primarily on macroscopic sorption behavior. These studies revealed many important clues regarding factors that influence sorption affinity and have enabled the development of structural hypotheses for operative adsorption mechanisms, which can be subjected to further study using molecule-scale methodologies, including X-ray diffraction, vibrational spectroscopies, and molecular/quantum mechanical simulations. In the following discussion we will show what information can be gained by these complementary techniques and how, by integrating this knowledge, detailed molecular scale adsorption mechanisms can be revealed. This is the goal of our chapter, rather than a detailed review of the published literature on clay–organic interactions [for these, see Bailey and White (1970), Green (1974), Mortland (1970, 1986), and Theng 1974].

2.3. MACROSCOPIC SORPTION STUDIES Early studies of organic compound–clay interactions relied heavily on macroscopic sorption data to reveal interesting phenomenalogic behavior, and to begin to understand associated sorption mechanisms. Often the most dramatic change

in sorption affinity of smectite clays for NOCs is achieved by replacing naturally occurring inorganic exchangeable cations with organic cations such as quaternary ammonium and phosphonuim cations (Boyd et al. 1988a–, Jaynes and Boyd 1990, 1991a,b; Lee et al. 1989; Kukkadapu and Boyd 1995; Lawrence et al. 1998; Sheng and Boyd 2000). However, the focus of this discussion is on the sorption properties for NOCs of smectites exchanged with inorganic cations since these are found in smectites as they exist in nature. As mentioned above, such smectites display high affinities for NACs (Haderlein and Schwarzenbach 1993; Haderlein et al. 1996; Weissmahr et al. 1997; Boyd et al. 2001; Sheng et al. 2002; Li et al. 2004a), and other important compounds, including atrazine (Laird et al. 1992; Aggarwal et al. 2006a), trichloroethene (Aggarwal et al. 2006b), and dioxin (Liu et al. 2009). The affinity of smectites for NACs is strongly dependent on the nature of the inorganic cation occupying the cation exchange sites of smectites. In a study of the herbicide dinitro-o-cresol (DNOC), Sheng et al. (2002) showed that adsorption of DNOC from water by smectite exchanged with a series of inorganic cations decreased in the order Cs þ > K þ  Al3þ > Ba2þ > Na þ > Ca2þ (Fig. 2.2). Calculation based on sorption isotherms showed that at a relative concentration of 0.1, Cs-SWy-2 was 1.3 times more effective than K-SWy-2, 5.5 times more effective than Al-SWy-2, 12 times more effective than Ba-SWy-2, 28 times more effective than Na-SWy-2, and 197 times more effective than Ca-SWy-2 for DNOC adsorption from water. Haderlein and Schwarzenbach (1993) and Haderlein et al. (1996) observed qualitatively similar effects of exchangeable cations on adsorption of NACs. Haderlein and co-workers attributed the strong adsorption of NACs by smectites to the formation of electron donor (negative charge sites in clays)–electron acceptor (NACs where the ---NO2 groups attract electron density from the aromatic p-ring system) (EDA) complexes. In a study by Boyd et al. (2001) adsorption of a series of substituted nitrobenzenes by K-SAz-1 revealed that electron-withdrawing groups did enhance sorption in a predictable manner on the basis of the Hammett substituent constant s, and this appeared to be consistent with the proposed formation of

MACROSCOPIC SORPTION STUDIES

75 Cs-SWy-2

Adsorption (mg/g)

60 K-SWy-2 45

30

15

Al-SWy-2 Na-SWy-2

0 0

10

20 Concentration (mg/L)

30

40 Ca-SWy-2

Figure 2.2. Adsorption of dinitro-o-cresol (DNOC) from water by SWy2 montmorillonite saturated with different exchangeable cations [from Sheng et al. (2002)]. Plots are of concentration of adsorbed DNOC versus equilibrium aqueous concentration of DNOC.

EDA complexes. However, quantum calculations of the gas-phase NACs revealed that the electron density of the aromatic p–ring system was unchanged among the substituted nitrobenzenes. Rather, electron density donated by a second substituent on nitrobenzene seemed to be appropriated by the ---NO2 group, leaving the aromatic p ring relatively unaffected. Thus, the quantum calculations did not support the dominance of an EDA mechanism since there was no evidence that the aromatic ring was any more or less electrondeficient regardless of the nature of the second substituent. This observation suggested the predominance of other mechanisms and forces in the adsorption of NACs by smectite clays. It was clear, from a cursory look at exchangeable cation ordering relative to the affinity of DNOC and other NACs, that cations with lower hydration energies, namely, Cs þ and K þ , produced higher-NAC-affinity smectites than did those saturated with cations that had higher hydration energies. This led us to hypothesize that ---NO2 groups, which possess partial negative charge, might form complexes with the exchangeable cations, either directly or through the intermediation of water. Weaker cation hydration would logically favor such interactions. Furthermore, the magnitudes for adsorption of various substituted NACs by a given K-saturated smectite (Boyd et al. 2001) followed an ordering that seemed plausibly consistent with the ability of the functional groups to form complexes with interlayer K þ . In a more recent study (Liu et al. 2009), the adsorption of dibenzo-pdioxin (dioxin) by saponite exchanged with different inorganic cations revealed that Cs-saponite effectively adsorbed dioxin from water, reaching nearly 1% wt/wt. In this instance there was a much larger difference in adsorption (of dioxin) by Cs- versus K-saponite, as compared to sorption of NACs.

55

Adsorption appeared to involve one or both of the oxygens in the dioxin ring. Dioxin adopted at least two orientations on the saponite interlayer. In one, dioxin is essentially dehydrated as it interacts with the opposing siloxane sheets and with coplanar Cs þ via one of the dioxin ring oxygens, analogous to adsorption of NACs. At higher loadings dioxin is intercalated between Cs þ and the clay surface in a tilted orientation when both oxygens of the dioxin ring interact with Cs þ . Because the negative charge character of the dioxin ring oxygens is less than that of the oxygens of ---NO2 groups, the adsorptive requirement for a weakly hydrated cation, namely, Cs þ , is greater than that observed for NACs. Hence, dioxin adsorption by Cs-saponite is much higher than that by K-saponite, whereas this difference is rather smaller for NAC adsorption (Fig. 2.2). A second important finding from macroscopic sorption studies was that smectite clays with lower charge densities were more effective adsorbents for NACs and other NOCs compared to clays with higher charge densities. This had first been observed in studies of adsorption of vapor-phase aromatic hydrocarbons (e.g., benzene, toluene) by smectites exchanged with tetramethylammonium (Lee et al. 1990). Comparison of a relatively “low-charge” smectite (SWy-2, surface charge density  1.09 mmol/m2) to a “high charge” smectite (SAz, surface charge density 1.69 mmol/m2) revealed that the lower-charge SWy-2 clay was a more effective adsorbent for gas-phase aromatic hydrocarbons. Also, uptake of the NOC vapor by dry clay was higher than uptake of the corresponding solute from water. Subsequent studies with similar trimethylphenylammonium clays showed the same trends for adsorption from water of a larger group of aromatic hydrocarbons (Jaynes and Boyd 1990, 1991a). Also, when the charge density of the high-charge SAz clay was chemically reduced (by the Li reduction method), adsorption was directly related to the degree of charge reduction (Jaynes and Boyd 1991b). In our study of DNOC adsorption by K-exchanged smectites we found that the lower charged SWy-2 adsorbed more DNOC from water than the higher-charge SAz, and charge reduction of the SAz clay resulted in proportionately higher adsorption of DNOC (Sheng et al. 2002), (Fig. 2.3). So, from these macroscopic sorption studies, three key observations were made that led to the beginnings of hypotheses regarding the operative adsorption mechanisms: (1) adsorption was higher in smectites exchanged with weakly hydrated exchangeable cations, (2) adsorption of the organic vapor by the dry clay was somewhat higher than adsorption of the corresponding solute from water (i.e., water reduced uptake) and (3) adsorption by lower-charge-density clays was higher than adsorption by higher-charge-density clays. From these observations we proposed that (1) adsorption could involve the exchangeable cation via complexation with polar functional groups or structural units (e.g., with ---NO2 groups of adsorbed NACs), (2) siloxane surfaces between

56

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

60

(a)

45

K-SWy-2

30

Amount absorbed (mg/g)

K-SAz-1

15

0

(b) 0.40 CEC

60 0.55 CEC

45 0.70 CEC

30

15 0.85 CEC 1.0 CEC K-SAz-1

0 0

5

10

15

20

25

Equilibrium concentration (mg/L)

Figure 2.3. (a) Adsorption of dinitro-o-cresol (DNOC) by homoionic lower K-smectite (K-SWy-2) and higher (K-SAz-1) surface charge density; (b) sorption of dinitro-o-sec-butylphenol (Dinoseb) by charge-reduced K-SAz-1 (from 15% to 60% reduction in cation exchange capacity—denoted 0.85 CEC to 0.4 CEC) [from Sheng et al. (2002)].

exchangeable cations were nanoscale adsorption domains, and (3) water tended to decrease adsorption by hydrating the exchangeable cation—thereby reducing the potential for complex formation and by obscuring the siloxane surface; in both instances more water (i.e., more cation hydration) meant lower adsorption overall. The effects are illustrated schematically in Figure 2.4. 2.4. X-RAY DIFFRACTION X-ray diffraction (XRD) has provided molecule-scale information regarding the separation of the 2 : 1 smectite layers (i.e., the interlayer distance) in clays with and without adsorbed NOCs and pesticides. The interlayer distance of smectites is determined largely by the hydration propensities of the exchangeable cations, and to a leaser extent the layer charge. For example, homoionic K-smectites at 100% relative humidity (RH) typically exhibit interlayer spacings  of 12.5–15 A, with lower-charged smectites (e.g., SWy-2)

tending toward the higher spacings (MacEwan and Wilson 1980). Cesium-smectites have a strong tendency to  equilibrate with 12.5 A spacings, even in bulk water, due to the lower hydration energy of Cs þ compared to K þ . Smec- tites saturated with Ca2þ , Ba2þ ,or Al3þ always swell to 15 A (d001) in aqueous suspension. Hence, exchangeable cation hydration also influences NOC/pesticide adsorption by smectite clay through its influence on the interlayer spacing of clay sheets in the clay tactoids. In a particularly revealing experiment, Sheng et al. (2002) used XRD to quantify the swelling behavior of K-SWy-2 films in the presence and absence of water vapor, and with varying amounts of adsorbed DNOC (Fig. 2.5). As expected, exposure of previously air-dried K-SWy-2 films to 100% RH  caused an increase in the interlayer spacing from 11 to 15 A. Interestingly, the presence of adsorbed DNOC restricted swelling of the rewetted clay. At higher DNOC loadings the interlayer spacings for the air-dried and rewetted (100% RH)  clays were nearly equal (12.2 A). The presence of adsorbed DNOC apparently caused the monolayer (i.e., one layer of intercalated water) to be retained even though the same KSWy-2 smectite would swell further (i.e., to 15 A) in the  absence of DNOC. Thus, the 12 A spacing appeared optimal for DNOC adsorption. In this configuration, the DNOC molecules would be flat on, and parallel to, the siloxane surfaces of K-SWy-2. Furthermore, the 12 A spacing would allow adsorbed DNOC to interact simultaneously with the opposing clay siloxane surfaces, thereby minimizing its contact with water. Considering that the free energy of hydration of many small NOCs is in the range of 10–30 kJ/ mol1, removal of DNOC from bulk water on intercalation may provide sufficient energy to prevent K-SWy-2 smectite  from swelling beyond the measured spacing of 12–12.5 A.  This spacing provides an interlayer distance of 3 A, which corresponds to the approximate thickness  of DNOC. Since Cs-smectites tend to maintain a 12.5 A spacing in water, due to the lower hydration energy of Cs þ versus K þ , the Cs-smectite is particularly well suited for intercalation of DNOC, and is a somewhat more effective adsorbent for DNOC (Fig. 2.2). It should be noted that these experiments were conducted using self-supporting clay films rather than aqueous suspensions of the K-SWy-2 clay, which are assuredly present under ambient environmental conditions, along with the particles in lesser-hydrated states. From the data of Sheng et al. (2002), it was unknown whether the clays adopt the 12 A spacing in suspension as a result of DNOC adsorption. However, more recently developed, novel XRD techniques now enable XRD measurement of interlayer spacings in actual clay suspensions (Chappell et al. 2005; Li et al. 2007) as well as in air-dried forms. Smectites exist as quasicrystals, which are stacks of subparallel 2 : 1 phyllosilicate layers with parallel c axes and randomly oriented a- and b-axes. In aqueous suspensions, smectite quasicrystals are dynamic in the sense that

X-RAY DIFFRACTION

-

•Low hydration exchangeable cation

Clay layer

EC Adsorptive

-

•Low charged clay

EC

domain

-

•Highest NOC adsorption

•Low hydration exchangeable cation

-

EC EC

EC

•High charged clay

EC

-

-

•Intermediate NOC adsorption

-

•High hydration exchangeable cation

EC

EC

•Low charged clay

EC

57

•Intermediate NOC adsorption

EC

•High hydration exchangeable cation

EC

-

EC

•High charged clay

-

•Lowest NOC adsorption

Figure 2.4. Availability of smectite clay interlayer domains for neutral organic contaminant (NOC) adsorption. Circles around inorganic exchangeable cation (EC) represent the hydration sphere. Lower hydration manifests larger adsorptive domains and greater potential for direct NOC–EC interactions as well as NOC interactions with the siloxane surface.

they are capable of swelling by imbibing water and/or organic molecules between layers within quasicrystals (crystalline swelling) and in the sense that large quasicrystals may cleave, producing two or more smaller quasicrystals, and conversely several smaller quasicrystals may coalesce forming a single large quasicrystal (Laird 2006). In aqueous suspensions some smectites, principally those saturated with Na þ and Li þ in dilute electrolyte solutions, are capable of 15.0

14.5

d 001 (Å)

12.5 Air-dried and re-wetted

12.0 Air-dried

11.5

11.0 0

10

20

30

40

50

Amount of DNOC adsorbe (mg/g)

Figure 2.5. Basal spacings of K-smectite (K-SWy-2) clays with varying levels dinitro-o-cresol (DNOC) sorption in the presence and absence of water [from Sheng et al. (2001)].

60

complete delamination such that the individual lamella are  separated by distances 30 A and behave as semi-independent colloids (colloids are not truly independent until the suspension is so dilute that colloid concomitant volumes do not interact). Both crystalline swelling/shrinking and the breakup/reformation of quasicrystals are inherently hysteretic processes, as energy is necessary to effect the physical rearrangements of matter that are required for both directions of both process [see Laird (2006) for a more complete discussion of smectite swelling processes]. Any change in the physical state of a smectite quasicrystal in an aqueous system inherently changes the affinity of the smectite for an organic solute. Crystalline swelling, for example, can change the size of interlayer domains from  12.5 A spacing, which is optimal for retention of many organic molecules, to a more hydrated and hence less optimal  15 or 18 A spacings. Similarly the breakup of a large smectite quasicrystal into two or more smaller quasicrystals diminishes the number of interlayer adsorption sites and increases the number of external surface adsorption sites. Within interlayers organic molecules may interact simultaneously with the basal oxygens of two opposing siloxane surfaces, whereas organic molecules interact with only one siloxane surface on an exposed external surface. Furthermore, the permittivity of the interlayer water is lower than the permittivity of water adjacent to an external surface. Weakly polar organic molecules are partitioned from a high-permittivity aqueous phase into a low permittivity aqueous phase with the driving force being an increase in system entropy.

58

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

20 Å

Relative intensity

17.5 Å 15.0 Å

12.5 Å

Ca-ND Ca-AD

K-ND K-AD

3

4

5 6 7 Degrees two theta

8

9

Figure 2.6. Transmission X-ray diffraction patterns for Panther Creek smectite suspensions. The smectite was saturated with Kþ or Ca2þ and either air-dried and resuspended (AD) or never dried (ND). The patterns have been vertically offset. [The figure is adapted from data originally presented by Chappell et al. (2005)].

4000 Ca-ND Ca-AD

Sorbed Atrazine (µmol kg-1)

The impact of crystalline swelling on adsorption of organic molecules was clearly demonstrated by Chappell et al. (2005). For the study, two suspensions of K þ -saturated Panther Creek smectite were prepared by dialysis from an original highly swollen Na þ Panther Creek biochar suspension. The two K-smectite suspensions were identical in every way except that one had been air-dried and then resuspended while K þ -saturated and the other was maintained as an aqueous suspension the entire time it was in the K form. Quasicrystals in the never-dried K-smectite (K-ND) suspension were swollen (Fig. 2.6: broad XRD peak between 15 and  17.5 A indicating interstratification of lamellae with two or three layers of interlayer water molecules) to a greater extent than were the quasicrystals in the air-dried K-smectite (KAD) suspension (Fig. 2.6: broad XRD peak between 12.5 and 15 A indicating interstratification of layers with one or two layers of interlayer water molecules). This legacy in the extent of crystalline swelling from having been air-dried and then resuspended caused an order of magnitude increase in the affinity of the K-AD sample for atrazine relative to the K-ND sample (Fig. 2.7). By contrast, Ca-smectite in air-dried and never-dried (Ca-AD and Ca-ND, respectively) suspensions adsorbed similar amounts of atrazine and exhibited  similar broad 20-A XRD peaks indicating that most interlayers held four layers of interlayer water molecules regardless of the air-dried or never-dried treatments. The results (Chappell et al. 2005) demonstrate that the impact of saturating cation on the affinity of a smectite for organic molecules comes primarily from the impact of the cation on the extent of interlayer hydration, which for K-smectites is strongly influenced by the history of sample handling, or by extension, wetting–drying cycles in nature. Comparison of XRD patterns for randomly oriented K- and Ca-smectite quasicrystals in aqueous suspensions

K-ND

3000

K-AD

2000

1000

0 0.00

0.05

0.10

0.15

0.20

0.25

0.30

Reduced concentration

Figure 2.7. Isotherms for sorption of atrazine on Panther Creek smectite saturated with Kþ or Ca2þ and either air-dried and resuspended (AD) or never dried (ND). Reduced concentration refers to the ratio of the concentration of atrazine in water to aqueous solubility of atrazine. [The figure is adapted from data originally presented by Chappell et al. (2005)].

(Shang et al. 1995) with XRD patterns for air-dried- and oven-dried-oriented films of the same smectites provided insight into the impact of the breakup and reformation of smectite quasicrystals on sorption of DNOC (Pereira et al. 2007, 2008). The phenolate form of DNOC, which is the dominant form of DNOC in solutions with pH  4.4, was adsorbed primarily on external surfaces of K-smectite quasicrystals in aqueous suspensions and entered the interlayers as K-DNOC complexes when individual K-smectite layers and small quasicrystals coalesced to form large quasicrystals on drying. By contrast, in Ca-smectite suspensions (pH 4.4) the phenolate form of DNOC was adsorbed only on external surfaces quasicrystals and there was no evidence for the formation of Ca-DNOC complexes. However, some of the phenolate DNOC became entrapped between substacks within Ca-smectite quasicrystals as the suspensions were dried to form oriented clay films. The neutral form of DNOC (pH  4.4) was adsorbed in the interlayers of a low-charge-density Ca-smectite but not in the interlayers of a high-chargedensity Ca smectite, apparently due to steric restrictions. Studies of homoionic smectite suspensions provide insight into mechanisms controlling smectite–organic interactions. However, soils and sediments invariably contain multiple types of cations, with Ca2þ , Ma2þ , and K þ typically being the most abundant cations in temperate region soils. Thus, in natural systems, cation exchange reactions and the adsorption/desorption of organic solutes occur simultaneously. As discussed above, both the type of cation adsorbed on the exchange complex of a smectite and the physical state of the smectite quasicrystals have a large influence on the

VIBRATIONAL SPECTROSCOPIC STUDIES

affinity of smectites for organic solutes. Similarly, the extent of crystalline swelling of smectites influences cation ex change selectivity, such that interlayers with 12.5 A spacings have a distinct preference for weakly hydrated monovalent cations (e.g., K þ ) and more expanded interlayers (d spacings  of 15–20 A) prefer more strongly hydrated divalent cations such as Ca2þ and Ma2þ (Laird and Shang 1997). Thus quasicrystal dynamics regulates interactions between cation exchange reactions and the adsorption/desorption of organic solutes, and conversely the loading of inorganic cations and organic molecules in the interlayers regulates quasicrystal dynamics (Li et al. 2004b; Chatterjee et al. 2008). Furthermore, the hysteresis that is inherent in crystalline swelling/ collapse and quasicrystal formation/breakup (Laird et al. 1995) causes hysteresis in both cation exchange reactions (Laird and Shang 1997) and hysteresis in the adsorption/desorption of organic solutes (Li et al. 2004b; Chatterjee et al. 2008). Demixing of both inorganic cations and organic solutes, such that K þ and organic solutes are selectively coadsorbed in collapsed (d ¼ 12.5 A) interlayer domains while Ca2þ and water molecules are selectively coadsorbed in  expanded (d ¼ 15–20 A) interlayer domains of the same quasicrystal (Pils et al. 2007), is another implication of the interaction between quasicrystal dynamics, cation exchange reactions, and the adsorption/desorption of organic solutes.

2.5. VIBRATIONAL SPECTROSCOPIC STUDIES In order to achieve a mechanistic understanding of NOC interactions with soil constituents, adapted spectroscopic methods sensitive to molecular interactions are required. Sorption data are macroscopic in nature and therefore, are fundamentally insensitive to molecular phenomena (Johnston and Sposito 1987). In the case of NAC sorption to clay minerals, vibrational spectroscopy, in particular, has provided molecule-level insights about the sorption mechanisms on several different aspects. In general, NACs are well suited for clay sorption studies because selected NACs show a high affinity for smectite, and because NACs have strong IR-active vibrational modes that are sensitive to short-range intermolecular interactions (Saltzman and Yariv 1975; Weissmahr et al. 1997; Johnston et al. 2002; Sheng et al. 2002). In the case of 1,3,5-TNB on K-SWy-2 smectite, for example, sorption values approach 400 mmol/g (Johnston et al. 2001). The vibrational bands associated with the ---NO2 groups are of particular interest in NAC surface studies. In earlier work, Yariv and co-workers (Yariv et al. 1966; Saltzman and Yariv 1975) used IR spectroscopy to study the interactions of nitrophenol and nitrobenzene with smectites. The positions of the ---NO2 vibrational bands were affected by the clay surface and by the nature of the exchangeable cation when compared to the neat compound. Spectra in these

59

earlier studies were collected from air-dried and heated clay films to minimize the interference from water and utilized high surface loadings of the organic solute of interest. No clear spectral trends were observed from the air-dried films and evidence for inner-sphere complexation was observed for the heated films exchanged with different cations (Saltzman and Yariv 1975). The principal bands of interest are the nasym(NO) and nsym(NO) bands along with the ---NO2 deformation bands. The ---NO2 group is highly electronegative, and the vibrational motions associated with nasym(NO) and nsym(NO) modes induce a large change in the induced dipole moment, which translates to intense/strong IR-active modes. Of particular benefit to clay surface studies is the fact that these vibrational modes occur in accessible spectral regions that are not obfuscated by the intense clay bands or by the presence of water (Fig. 2.8) (Johnston et al. 2001). The NACs can be considered to function as molecular probes that have diagnostic properties that are sensitive to changes in their local environment (Johnston et al. 1993). In a related way, the chemical shift of both 15 N and 17 O nuclei have been used to probe changes in the chemical environment around the ---NO2 group resulting from intermolecular interactions. Sorption of NAC is favored on low-charge-density clay minerals that are exchanged with less hydrated exchangeable cations (Haderlein et al. 1996; Boyd et al. 2001). On the basis of earlier studies (Jaynes and Boyd 1991b; Laird et al. 1992, 1994; Barriuso et al. 1994), increased sorption was attributed to the presence of neutral siloxane surface sites, which are regions on the smectite surface not in the near vicinity of exchangeable cations, waters of hydration surrounding exchangeable cations, or the isomorphic substitution sites themselves (Boyd et al. 2001; Sheng et al. 2002; Johnston et al. 2004; Li et al. 2004a). These portions of the clay surface are less hydrated and are considered to provide favorable sorption domains for semi-polar organic solutes such as NACs, atrazine, and related compounds. Haderlein and coworkers were among the first to show that a wide range of NACs showed a high affinity for smectites in sorption studies from aqueous suspension (Haderlein and Schwarzenbach 1993; Haderlein et al. 1996). Nitroaromatic compounds that show high affinities for smectites were generally found to be planar structures containing more than one nitro group (Haderlein et al. 1996). This work was extended using sorption and molecular spectroscopy [NMR, UV–visible, and in situ attenuated total veflection (ATR)–FTIR spectroscopy] to the study of NAC sorption mechanisms on smectite (Weissmahr et al. 1997). Although direct coordination mechanisms had been proposed in earlier NAC-smectite sorption studies (Yariv et al. 1966; Saltzman and Yariv 1975; Fusi et al. 1982) on the basis of air-dried clay films, they observed similar in situ ATR-FTIR results between Cs þ - and K þ exchanged smectites. Based on the spectral similarities they concluded that the exchangeable cation did not play a

60

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

Figure 2.8. FTIR spectra of 1,3-dinitrobenzene sorbed to K-SWy-1 smectite. Individual band assignments are shown as well as spectral regions where IR absorption resulting from the smectite, sorbed water, and 1,3-dinitrobenzene occur [from Johnston et al. (2001)].

significant role and attributed the high affinity of certain NACs for smectites to the formation of a electron-donor– acceptor (EDA) complex discussed earlier between the p electrons of the NAC and the electrons of the siloxane oxygen atoms, which was not entirely consistent with the earlier 1550

ex situ IR studies (Yariv et al. 1966; Saltzman and Yariv 1975; Fusi et al. 1982). In order to address this apparent inconsistency further, we investigated the spectroscopic properties in the study of NACs sorption on three different smectites exchanged

(a)

Na+ K+

1548

νasymm (NO) Mg+2

Band position (cm–1)

1546

Ba+2 Ca+2

1544 SWy-1 SAz-1 SHCa-1

1358 (b) 1356 1354

νsymm (NO)

1352 1350 1348 -2500

-2000

-1500

-1000

-500

0

Enthalpy of Hydration (kJ/mol)

Figure 2.9. Positions of the nasymm(NO) (a) and nsymm(NO) (b) bands of 1,3,5-trinitrobenzene sorbed to SAz-1, SWy-1, and SHCa-1 smectite exchanged with Mg, Ca, Ba, Na, and Cs; the data are plotted as a function of the enthalpy of hydration of the exchangeable cation [from Johnston et al. (2001)].

VIBRATIONAL SPECTROSCOPIC STUDIES

61

exchanged with strongly hydrated cations (Mg2þ , Ca2þ , Ba2þ , Na þ). The splitting decreased to  188 cm-1 for the weakly hydrated cations (K þ and Cs þ ) (Fig. 2.9). A similar decrease in the splitting between the nasym(NO) and nsym(NO) bands was observed for 1,3-dinitrobenzene (Fig. 2.10). In order to investigate this splitting further, quantum mechanical calculations of gas-phase 1,3,5-TNB, 1,3-dinitrobenzene, and inner-sphere K complexes of both NACs were undertaken (Johnston et al. 2001), using the B3LYP level of theory and the 6–311G basis set (Frisch et al. 1998). Splittings between the computed vibrational frequencies for the K complexes relative to the uncomplexed NACs showed the same shifts in direction, although of somewhat larger magnitude, that had been observed experimentally. These computational results showed that the decrease in splitting resulted from strengthened interactions between the ---NO2 oxygen atoms and K þ , thus supporting the hypothesis that adsorbed NACs formed complexes with interlayer K þ ions. For more strongly hydrated exchangeable cations, the effective distance between the ---NO2 group

with cations whose enthalpies of hydration varied from 315 kJ/mol (Cs þ) to 1960 kJ/mol (Mg2þ ) (Evangelou 1998). Although Weissmahr et al. (1997) concluded, in light of spectral similarities between Cs- and K-exchanged smectites, that ---NO2 groups were not involved in direct coordination to surface sites or exchangeable cations, the difference in the enthalpy of hydration between Cs þ (315 kJ/mol) and K þ (360 kJ/mol) is almost negligible relative to the difference of more strongly hydrated alkali-metal cations (Li þ and Na þ ) or the alkaline-earth cations (Mg2þ , Ca2þ and Ba2þ ) (Friedman and Krishnan 1973). We found that the nature of the exchangeable cation influenced both the position and the relative intensities of the ---NO2 symmetric and asymmetric stretching bands compared to their spectra obtained in aqueous solution (Fig. 2.9, Table 2.3) (Johnston et al. 2001a, 2002, 2004; Sheng et al. 2002). In the case of 1,3,5-trinitrobenzene, for example, the splitting between the asymmetric and symmetric NO2 stretching bands remained relatively constant at 200 cm1 for 1,3,5-TNB in aqueous solution and when sorbed to smectites

TABLE 2.3. Influence of Exchangeable Metal Cations on Positions (n, cm-1) and Relative Intensities (I) of ---NO2 Stretching Bands of 1,3,5-Trinitrobenzene Sorbed by Different Smectite Clays (SWy-1, SAz-1, SHCa-1) Element(s)

nasym(NO)

Isym

Frequency Difference

Iasym/Isym

0.98 1.12 1.18 2.48 3.40 18.79 22.16

200 200 200 200 196 196 191

0.79 0.83 0.84 0.83 1.12 1.05 1.15

1350 1350 1349 1348 1352 1355 — —

— — — — — 6.17 8.56 21.65

— — — — — 187 197 193

— — — — — 1.00 1.05 1.27

1349 1349 1349 1349 1352 1353 1350 1356

2.96 3.49 2.94 2.56 2.51 18.23 12.65 10.63

200 200 200 200 197 190 196 188

0.83 0.89 0.83 0.73 0.94 1.21 1.29 1.40

nsym(NO)

Iasym

SWy-1 Mg Ca Ba Al Na NH4 K

1549 1549 1549 1549 1548 1545 1546

0.78 0.93 0.99 2.07 3.81 19.65 25.39

1349 1349 1349 1348 1352 1349 1355 SAz-1

Mg Ca Ba Al Na Cs NH4 K

— — — — — 1542 1547 1547

— — — — — 6.18 — —

Mg Ca Ba Al Na Cs NH4 K

1549 1549 1549 1549 1549 1544 1546 1544

2.47 3.09 2.43 1.88 2.36 22.10 16.36 14.87

SHCa-1

Source: Adapted from Johnston et al. (2001).

62

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

Figure 2.10. Comparison of the FTIR spectra of 1,3-dinitrobenzene sorbed to K-SWy-1 montmorillonite (upper spectrum) to that of 1,3-dinitrobenzene in aqueous solution (lower spectrum, dashed line); the splitting between the nasymm(NO) and nsymm(NO) bands is shown as well as the intensity ratio Iasym/Isym [from Johnston et al. (2001)].

and the positively charged cation is increased, essentially turning off the site-specific interaction. In addition to the cation-induced change in band positions of the nasym(NO) and nsym(NO) bands, the intensity ratio of Iasym(NO)/Insym(NO) was also perturbed and varied as a function of enthalpy of hydration of the exchangeable cation (Fig. 2.11). In prior NAC surface studies, this intensity ratio was sensitive to different types of intermolecular interactions resulting from changes in the O---N---O angle and hydrogen bonding to oxide surfaces (Ahmad et al. 1996). Saltzman and Yariv (1975) observed that the position of the nasym(NO) band decreased and its relative intensity

Intensity ratio iasymm/isymm

1.45 1.35 1.25

SWy-1 SAz-1 SHCa-1 K

+

1.15 1.05 +2

0.95 0.85 0.75 -2500

Ca Mg

+2

+2

Ba

Na+

-1500 -1000 -500 -2000 Enthalpy of hydration (kJ/mol)

0

Figure 2.11. Intensity ratio (integrated area) of the nasymm(NO) and nsymm(NO) bands as a function of the enthalpy of hydration of the exchangeable cation on three smectite clays (SWy-1, SAz-1 and SHCa-1) [from Johnston et al. (2001)].

increased for p-nitrophenol sorbed on smectite when heated, and this was attributed to direct complexation with the exchangeable cation. Furthermore, the spectroscopic results obtained in these studies agree with the results obtained by Weissmahr et al. (1997) in that the IR data obtained for K- and Cs-exchanged smectites are similar. The vibrational bands associated with the ---NO2 group are relatively sharp and well resolved with full-width at halfmaximum (FWHM) bandwidths of 10 cm-1, which is quite narrow for solutes sorbed on surfaces in aqueous suspension. Although highly resolved, the band positions of the NO2 groups do not experience large shifts in position resulting from intermolecular interactions (Nyquist and Settineri 1990; Ahmad et al. 1996). In the context of prior NAC surface studies, the shift in band positions of the n(NO) bands for the NACsmectite complexes are some of the largest shifts reported (Conduit1959; Urbanski and Dabrowska 1959; Jonathan 1960; Borek 1963; Baitinger et al. 1964; Green and Lauwers 1971; Nyquist and Settineri 1990; Ahmad et al. 1996). In spectroscopic studies of environmentally relevant solutes such as NACs, it is useful to combine sorption and spectroscopic methodologies such that the amount of NAC sorbed is known. This type of information provides a direct link between the macroscopic sorption data and the molecular insights gained from spectroscopy. We conducted a series of parallel sorption and spectroscopic measurements and found that intensity of the NAC vibrational bands [nsym(NO) of 1,3-dinitrobenzene is shown in Fig. 2.12) increased linearly with increasing surface coverage as determined using HPLC (Fig. 2.12a). This confirmed the identity of the surface species as 1,3-DNB and not some

VIBRATIONAL SPECTROSCOPIC STUDIES

(c)

63

1

Absorbance@ 1353 cm-1

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

5

10

15

20

25

30

35

Amount of 1,3-DNB sorbed (mg/g)

Figure 2.12. (a,b) Absorbance of the nsymm(NO) band of 1,3-dinitrobenzene (DNB) at 1353 cm1 at different surface coverages (a) Each spectrum shown in (a) corresponds to a point on the HPLCderived sorption isotherm shown in (b). Combinations of HPLC- and FTIR-derived sorption isotherms are shown in the (b). Adsorption isotherms (b) are plots of concentration of sorbed DNB (Cs) versus equilibrium aqueous concentration of DNB (Ce) [from Johnston et al. (2001)]. Plot of the absorbance of the nsymm(NO) band of 1,3-dinitrobenzene (1,3-DNB) at 1353 cm-1 as a function amount of 1,3DNB sorbed by K-smectite (K-SWy-1) montmorillonite determined using HPLC methods [from Johnston et al. 2001)].

degradation product. In addition, unlike traditional batch sorption methods, the surface solute is directly observed using FTIR as opposed to batch-sorption-derived data, which are based on difference in aqueous-phase concentrations measured using HPLC. In addition, the quantitative analysis of the spectroscopic data provides a measure of the limit of detection of NACs on smectite surfaces. For the data shown in Figure 2.12b, based on a minimum absorbance value of 1 milliabsorbance unit (mAU), the detection limit of 1,3-DNB would be approximately 150 nmol/g. Polarized infrared spectroscopy can provide a direct means of determining the molecular orientation of sorbed species on oriented self-supporting film clay films (Johnston et al. 2002; Ras et al. 2003, 2007). In the case of planar organic solutes, such as many NACs, the vibrational modes

can be divided into in-plane and out- of-plane vibrational modes. The atomic motions corresponding to the nasym(NO) and nsym(NO) modes of 1,3,5-TNT, 1,3-DNB, and DNOC are aligned within same plane as the aromatic ring. In contrast, the vibrational motion of the atoms involved in the out-ofplane ---NO2 deformation are oriented (as their name implies) out of plane. Because the smectite particles have a high aspect ratio, highly oriented self-supporting films can be made (Johnston and Premachandra 2001). When planar organic solutes are sorbed on smectites, linear dichroism techniques can be used to determine the molecular orientation of the sorbed species (Margulies et al. 1988). If the dichroic ratio (DR) for a particular vibrational mode is >1, the orientation of that particular vibrational mode is aligned parallel to the clay surface. The DR for the nsym(NO) band of

64

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

Figure 2.13. FTIR spectra of dinitro-o-cresol (DNOC) adsorbed by K-smectite (K-SWy-2) at 0 and 45 of beam incidence [from Sheng et al. 2002)].

DNOC on K-SWy-2 is 1.32. In support of this value, the measured DR value for the out-of-plane ---NO2 rock is 0.52 (Fig. 2.13). Together, these linear dichroism values indicate that the molecular plane of DNOC is parallel to the [001] plane of the clay surface (parallel to the siloxane surface as illustrated in Fig. 2.14). Interestingly, similar studies with

dioxin sorbed on Cs-saponite revealed that intercalated dioxin was present in orientations that were not parallel to the siloxane surface, in agreement with XRD measurements  showing expanded interlayer distance of 15.4 A at high loadings of 0.8% wt/wt (Liu et al. 2009).

2.6. MOLECULAR AND QUANTUM MECHANICAL SIMULATIONS

Figure 2.14. Illustration of sorbed dinitro-o-cresol (DNOC) molecule laying flat on the siloxane surface showing the siloxane surface and water molecules surrounding the exchangeable cations. The dimensions of the DNOC molecule and the distance between the exchangeable cations are drawn to scale. The DNOC is shown as a representative nitroaromatic compound (NAC). [From Johnston et al. 2002)]. (See insert for color representation of this figure.)

In addition to the quantum chemical studies described above, molecular simulations using classical dynamics have been used (Teppen et al. 1997, 1998) to integrate experimental data and to explore the molecular mechanisms of NOC interactions with clay mineral surfaces. Clay minerals with compositions similar to various smectites were constructed (Boyd et al. 2001; Sheng et al. 2002; Chappell et al. 2005; Aggarwal et al. 2006a). Our experimental adsorption isotherms were used to choose realistic loading rates for the NOCs, and interlayer water contents were derived by inference from X-ray diffraction patterns gathered for both airdried clay films (Sheng et al. 2002; Li et al. 2004a) and suspension-phase (Chappell et al. 2005) NOC–water–clay complexes. Water molecules and NOCs were inserted into the simulation cell at random initial positions and orientations, and the water contents were adjusted until the simulated equilibrium d(001)-spacing agreed with experimental values, so that the simulated interlayer environment corre-

ENVIRONMENTAL IMPLICATIONS

65

2.7. ENVIRONMENTAL IMPLICATIONS

sponded as closely as possible to that of the adsorption experiments. After lengthy equilibrations, the interlayer structures were sampled. An example is shown in Figure 2.15, for which the clay layers have been removed to reveal the NAC–cation–water interactions in the interlayer region. The results are consistent with our hypotheses derived from FTIR spectroscopy in that a variety of inner- and outersphere NAC-K þ complexes are evident. One general observation from the simulations is that such complexes are almost forced by the sterically crowded nature of the smectite interlayer region. Specifically, in a smectite interlayer of 12.5-A d001 spacing as in Figure 2.5, each monovalent cation occupies only 0.5–1 nm2, while single-ring aromatic NOC molecules are at least 0.5 nm2 in size. Thus, the adsorbed NOCs must be forced into close proximity to cations, and this situation will be energetically favorable only if the NOC contains O- or N-based functional groups that are at least polar enough to plausibly coordinate cations. Indeed, two or more polar functional groups are even more favorable since the NOC must fit between several cations (Fig. 2.15). For example, the close proximity of many monovalent cations can be used to rationalize why the magnitudes of nitrobenzene sorption by K-smectites follow the order 1,3,5-TNB  1,4-DNB 1,3-DNB  NB. Also, dibenzo-p-dioxin sorbs up to 8000 mg/kg on certain Cs-smectites (Liu et al. 2009), while similar-sized PAHs that lack the oxygen functionality sorb to a much smaller extent to the same clay.

Webster A (K)

p-NCB sorbed (mg g-1)

Figure 2.15. Molecular dynamics snapshot of dinitro-o-cresol (DNOC) in K-smectite (K-SWy-2) clay interlayer, (Kþ ¼ green, O ¼ red, C ¼ gray, N ¼ blue, H ¼ white). (See insert for color representation of this figure.)

The results of our studies and those others (referenced above) clearly demonstrate that NOCs and pesticides often display strong affinities for expandable 2 : 1–layer smectite clays, especially those saturated with weakly hydrated cation (e.g., K þ , Cs þ ). Most of these studies have been conducted using relatively pure reference clay specimens. In the environment, soil clay minerals and SOM are usually associated with each other. Soil organic matter might obscure clay surfaces, thereby reducing the availability and hence efficacy of soil mineral fractions for adsorption of organic compounds. Until 2005 or so, there were no studies that directly assessed the effectiveness of unisolated clay minerals in soils as adsorbents for NOCs, and specifically the extent to which mineral components are available for adsorption of NOCs. As an initial step to approach this question, Charles et al. (2006a,b) measured sorption of several NACs by K þ and Mg2þ -saturated soils and SOM-removed soils. The results showed that extraction of SOM caused an increased sorption by K þ -saturated soils (Fig. 2.16) demonstrating that SOM posed an overall negative effect on sorption of NACs (e.g. p-nitrocyanobenzene, p-NCB), due to obscuration of p-NCB binding sites on soil clays. In contrast, sorption of p-NCB by Mg2þ -saturated Webster soil was greater than that by the Mg2þ -saturated SOM-removed soil (Fig. 2.16). Removal of SOM might liberate some clay surface sites; however, Mg-clays have low affinities for aqueous-phase NACs. Because clays in the Mg2þ -saturated SOM-removed soil were relatively inefficient sorbents for p-NCB, SOM was left as the principal sorbent phase, and its

Webster A (Mg) Webster A (OM-removed) (K)

0.6

Webster A (OM removed) (Mg) a

0.4

b

0.2

c d

0.0 0

10

20

30

Aqueous p-NCB concentration (mg L-1)

Figure 2.16. Adsorption isotherms for p-nitrocyanobenzene (pNCB) sorption by Kþ - and Mg2þ -saturated Webster soil (whole soil, and soil from which SOM was removed); the four isotherms (labeled a–b) are statistically different at p < 0.05 [from Charles et al. (2006a)].

66

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

removal decreased sorption overall. Results such as these demonstrate that SOM and clay minerals can contribute to the sequestration of NOCs in soils, and that these contributions are interrelated. To estimate contributions of sorption by minerals, several previous studies utilized the simple additive product of isolated SOM (e.g., humic acids) and mineral components (Pusina et al. 1992; Celis et al. 1997; Onken and Traina 1997; Li et al. 2003). However, this approach is inadequate to describe sorption of organic species by soils because it unrealistically assumes that these soil components act independently even though it is known that they are intimately associated in soils. Recognizing this, Karickhoff (1984) proposed the use of a mineral phase availability factor ( fa) to assess the fraction of mineral surfaces (mostly clays) available for sorption of NOCs by soils, although the approach was not evaluated experimentally. This equation incorporated potential SOM blockage of sorption sites on clays by summing NOC sorption to clay and SOM: Qsoil ¼ fa Qmin fmin þ Qsom fsom

ð2:1Þ

where Qsoil is the NOC mass sorbed per unit mass of soil, Qsom and Qmin are the SOM-sorbed and mineral-sorbed NOC per unit mass of the respective sorbent phase, and fmin and fsom are the fractional mineral and SOM contents of soil. The term fa represents the fractional availability of sorption sites on the clay components of whole soil, that is, the fraction of mineral sorptive surfaces available in whole soil. The term fa is a plausible refinement of simple addition of individual soil component contributions. Fractional availability ranges from 0 (unavailable) to 1 (100% available). Karickhoff (1984) equated the mineral fraction to the clay fraction owing to the high surface area of clays in general and of smectites in particular. Using Equation (2.1) to experimentally estimate the fa values requires knowledge of sorption by isolated soil minerals and SOM, but it is unachievable to fully separate unaltered soil components for sorption measurements. To address this deficiency, Charles et al. (2006a) developed a novel approach to determine fa values that involves measuring the difference in sorption by soil whose cation exchange sites are saturated with different cations (i.e., K þ vs. Mg2þ ) that render the clay surfaces adsorptive or nonadsorptive for NOC probe molecules. Specifically, the equation is expressed as QK-soil QMg-soil ¼ fa ðQK-min fmin QMg-min fmin Þ þ ðQK-som fsom QMg-som fsom Þ ð2:2Þ where the terms are the same as those in Equation (2.1), and subscripts K- and Mg- are added to differentiate K þ - and Mg2 þ -saturated soils or soil components.

Since QK-som QMg-som for NOCs (Charles et al. 2006a), it follows that QK-soil QMg-soil fa ðQK-min fmin QMg-min fmin Þ

ð2:3Þ

Equation (2.3) eliminates the Qsom and fsom terms, which are difficult to obtain experimentally but needed in Equation (2.1). For instance, reliance on Qsom introduces significant errors in Equation (2.1) since Qsom cannot be obtained directly from sorption by isolated and unaltered SOM. Using several NAC probe molecules, Charles et al. (2008) estimated the soil mineral availabilities of several smectitic soils. The results showed that 46%–100% of mineral surfaces in the Webster soil horizons and 35%–96% of mineral surfaces in the Clarion soil horizons were available for NOC adsorption. The fa values were negatively correlated with the ratio of SOM/smectite contents in soils. Thus, SOM can reduce the availability of clay mineral surfaces for NOC adsorption. In soils SOM may coat clay surfaces, bridge clay packets resulting in soil aggregate formation, inhibit smectite shrinking and swelling, and partially block the entrance of NACs into interlayer regions. Our studies have revealed that, in fact, fa is not a fixed intrinsic value for a given soil. Rather, the value depends on the type of the probe molecule used. Probe molecules with higher affinity for smectite surface adsorption sites (e.g., NACs) are more effective in accessing these sites, thereby manifesting higher fa values. This is attributed to the more effective competition for mineral adsorptive sites by NACs than by SOM. It is apparent that a large portion of soil mineral fractions are available for adsorption of NOCs and pesticides in whole soils, particularly for adsorbates (e.g., NACs) with strong affinities for soil smectites whose cation exchange sites are partially or fully saturated with weakly hydrated cations (e.g., K þ , Cs þ , NH4 þ ). It is not necessary to create fully K þ -, Cs þ -, or NH4 þ saturated soils or sediments for significant clay mineral adsorption. The amount of weakly hydrated inorganic cations could be just sufficient to produce clay demixing of exchangeable cations. In other words, some (but not all) clay layers or regions of clay tactoids may be fully compensated with K þ , Cs þ or NH4 þ leading to high-affinity NOC adsorption sites, while other regions are saturated with different exchangeable cations (e.g. Ca2þ , Mg2þ , or Na þ ) that do not induce strong affinities for NOCs (Chatterjee et al. 2008; Li et al. 2004b). To demonstrate this principle, adsorption of pesticides with different polarities (i.e., dichlobenil, monuron, biphenyl) was measured by homoionic K- and Ca-SWy-2 in KCl/ CaCl2 aqueous solutions (Li et al. 2004b). The presence of different amounts of KCl and CaCl2 in solution resulted in varying populations of K þ and Ca2þ on the clay exchange sites as a result of cation exchange. Pesticide sorption coefficients were calculated at a relative concentration of

1

K-SWy-2 Ca-SWy-2

0.1

0.01 0.0

dichlobenil monuron biphenyl

0.2

0.4

0.6

0.8

1.0

Figure 2.17. Sorption coefficients of pesticides normalized to the sorption coefficient by K-SWy-2 at the aqueous relative concentration (ratio of the solute concentration in water to the water solubility of solute) of 0.1 as a function of fractional Kþ populations on mineral surfaces. The open (hollow) symbols represent the sorption in which Kþ on homoionic K-SWy-2 was replaced by Ca2þ , and the solid symbols indicate the sorption in which Ca2þ on homoionic Ca-SWy-2 was replaced with Kþ from aqueous solutions. [from Li et al. (2004b)].

0.1 (aqueous equilibrium concentration/aqueous solubility) and normalized to the corresponding sorption coefficients by homoionic K-SWy-2 (i.e., 580 L/kg for dichlobenil, 27 L/kg for monuron, and 6.4 L/kg for biphenyl). When the normalized sorption coefficients were plotted against K þ fractions (fK) on mineral surfaces (Fig. 2.17), sorption of the least polar biphenyl remained nearly constant (and low) across the variation of fK on minerals from zero to one. No apparent enhancement was observed for monuron sorption on CaSWy-2 exchanged with KCl up to fK ¼ 0.71, whereas when K-SWy-2 underwent exchange with CaCl2, sorption was reduced by about half when the fractional K þ saturation decreased from 1 to 0.81. For dichlobenil, sorption by sorbents derived from Ca-SWy-2 increased by approximately 4 times as fK increased from zero to 0.71. Replacement of K þ from K-SWy-2 by Ca2 þ (fK ¼ 1–0.66) manifested gradually diminishing sorption of dichlobenil to 40% of that by homoionic K-SWy-2. Interestingly, dichlobenil sorption for clay derived from Ca-SWy-2 with fK ¼ 0.71 was substantially lower than the corresponding clay derived from K-SWy-2 with fK ¼ 0.66. The reason for such phenomena is a tendency for preservation of the original clay structures, namely, the smectite interlayer spacings associated with K- versus Casmectite as cation exchange (K þ ! Ca2 þ vs. Ca2 þ ! K þ ) proceeds, which can be viewed as a type of cation exchange hysteresis (Laird and Shang 1997; Laird 2006). Dichlobenil adsorbed much more strongly to K þ - rather than Ca2 þ -saturated smectite, so that as the K þ $ Ca2 þ exchange process occurs more favorable adsorption domains persist at a given fK when starting from the K þ -saturated end member.

67

Just as cation exchange processes on smectites alter adsorption of NOCs, changes in aqueous solution conditions such as ionic strength could also influence adsorption via their effects on clay quasicrystal structures. In aqueous solution, smectite clays are often present as quasicrystals consisting of stacks of clay platelets separated by interlayers filled with exchangeable cations and water molecules. Sorption of NOCs must effectively compete with interlayer water molecules to access adsorption domains on clay siloxane surfaces. Increasing aqueous salt concentration can promote the formation of clay aggregates, and reduce the amount of water between clay layers, thereby manifesting smaller interlayer distances in the formed quasicrystals and facilitating intercalation of NOCs. For example, sorption of 1,3DNB by K-SWy-2 increased with increasing KCl aqueous concentration (Fig. 2.18). At a relative 1,3-DNB aqueous concentration of 0.05, sorption by K-SWy-2 increased approximately 1.4, 1.7, 2.0 and 2.2 times as the KCl concentration increased from 0.01 M to 0.05, 0.10, 0.20, and 0.30 M, respectively (Li et al. 2007). Similar results were also observed for pesticide adsorption (Li et al. 2006). The small reduction in 1,3-DNB solubility due to the “salting out” effect was shown to be incapable of causing such a large increase in 1,3-DNB sorption by K-SWy-2. X-ray diffraction patterns and light absorbance of K-clay suspensions indicated the aggregation of clay particles and the formation of quasicrystal structures as KCl ionic strength increased, which is believed to be responsible for the enhanced adsorption of NACs and pesticides at higher ionic strength. For adsorption of NOCs by smectites, the type of exchangeable cation is the primary determinant of the size of sorptive domains in the clay galleries, clay interlayer distances, and the formation of complexes with NOCs. Cation exchange reaction (K þ $ Ca2 þ ) on smectite generates a range of K þ -saturated fractions or domains in the clay, 50000 Sorbed Concentration (mg kg-1)

Relative Sorption Coefficient

ENVIRONMENTAL IMPLICATIONS

0.30 M KCl 0.20 M KCl

40000

0.10 M KCl 0.05 M KCl

30000 0.01 M KCl

20000

10000

0 0

10

20 30 40 50 60 70 Aqueous Equilibrium Concentration (mg L-1)

Figure 2.18. Sorption isotherms of 1,3-dinitrobenzene (1,3-DNB) by K-SWy-2 from 0.01, 0.05, 0.10, 0.20, and 0.30 M KCl aqueous solutions [from Li et al. (2007)].

68

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

resulting in enhanced or reduced NOC adsorption (Chatterjee et al. 2008; Li et al. 2004b). Such simple cation exchange processes could be used as geochemical modulator in the development of environmentally friendly protocols to control the sorption, mobility, and bioavailability of NOCs in smectitic soils or soils amended with smectite clays. For example, Roberts et al. (2007a,2007b) demonstrated the successful application of K þ /Ca2 þ exchange reactions on smectite clays as a means to modulate the retention and release of NACs, and hence their toxicity to plants. Ionic strength is another plausible way to control clay interlayer environment, thereby modulating the degree of NOCs sorption/desorption in environmental systems. These simple geochemical controls on the adsorption/desorption of toxic NOCs could be used in bioremediation/phytoremediation to modulate the bioavailability and toxicity of NOCs to microorganisms and plants. 2.8. SUMMARY Sorption of NOCs by smectites can assume values along a continuum from zero up to 100 g NOC per kg clay. The critical factors that control sorption are a complex interplay among the functional groups of the NOC, the layer charge of the clay mineral, and the hydration of interlayer cations (Fig. 2.14): 1. The clay layer charge plays an important role because it controls the lateral distance between strongly hydrated interlayer cations. While “hydrated radii” are poorly defined, typical estimates of hydrated radii are   9.6 A for Ca2þ , 5.3 A for K þ , and 3.6 A for Cs þ (Evangelou 1998). Thus the cross-sectional area of hydrated Ca2þ is about 2.9 nm2, more than 7 times that of Cs þ (0.4 nm2), so one hydrated Ca2þ takes up 3.5 times as much surface area in the clay interlayer as do the two Cs þ ions that Ca2þ replaces. Using these hydrated radii along with an idealized smectite basal surface area of 750 m2/g and a 95 cmol/kg CEC, we compute that the hydrated radii of Ca2þ or K þ must overlap in the interlayer regions, meaning that the lateral adsorption domains (Fig. 2.4) are very small and the NOC would have to compete with strongly bound water for interlayer sorption sites. On the other hand, Cs þ and its hydration shells are projected to occupy only about 67% of the interlayer space; that is, about one-third of the surface area in a Cs-smectite may consist of lateral sorption domains (Fig. 2.4), implying that NOCs need compete only with weakly bound water in the Cs-smectite interlayers. This may explain why the Cs-smectite is the more effective adsorbent for essentially all NOCs studied. This idea of lateral domains is supported by several observations

(Lee et al. 1990; Jaynes and Boyd 1991b; Laird et al. 1992; Sheng et al. 2001, 2002) of an inverse relationship between the CEC of clays and the amount of the organic solute adsorbed, since fewer interlayer cations per unit surface area mean larger lateral adsorption domains available for organic solutes (Fig. 2.4). When layer charge is quite large, adsorption of even polar NOCs by smectites can be sharply reduced, because hydration radii for the greater number of exchangeable cations begin to overlap. In contrast, even Ca-smectites with low layer charges exhibit a strong sorption of atrazine (Laird et al. 1992). 2. As the organic functional group becomes more polar, the ability of the NOC to displace water from clay interlayers becomes stronger, apparently due to enhanced inner- and/or outer-sphere complexation between the NOC and interlayer cations, resulting in increased adsorption. Thus, pesticides with multiple strongly polar functional groups like ---NO2, ---C¼O or ---C N exhibit a strong sorption to K-smectites (Sheng et al. 2001; Boyd et al. 2001), even though the hydrated radii of the K þ ions are expected to overlap. Such polar functional groups apparently allow the NOC to form pesticide–cation complexes that are strong enough to displace water from the hydration shells of interlayer cations. 3. Compounds with functional groups that are less polar (e.g., atrazine, TCE, dioxin) are less strongly bound by K-smectites but can still be strongly adsorbed by Cssmectites, apparently because these NOCs are still able to displace the more weakly bound interlayer water found in the adsorptive domains (Fig. 2.4) between the hydrated Cs ions. The propensity for less polar NOCs to occupy these sites is further enhanced when the clay basal spacing is optimized at approximately 12.5 A, as in the cases of Cs- and certain K-smectites, as discussed above. In these cases, the NOCs can be mostly dehydrated, and a hydrophobic component contributes to NOC adsorption by smectites (Li et al. 2004a; Chappell et al. 2005). In this context, the site of negative charge on the clay can influence the retention of pesticide, presumably because tetrahedrally substituted smectites swell less in water and thereby contain narrower-slit pores that favor hydrophobic sorption (Aggarwal et al. 2006a,b). Additionally, the negative charges are more localized in tetrahedrally substituted smectites affording more neutral (hydrophobic) siloxane surface area for NOC adsorption. Apparently, then, an optimal inorganic sorbent for NOCs should be a Cs þ -saturated smectite with a low layer charge resulting from tetrahedral substitution. These criteria maximize adsorption domains parallel to the clay surfaces while

REFERENCES 

optimizing (near 12.5 A) the adsorption domains perpendicular to the clay surfaces. Such clays may adsorb 10% of their weight for an NOC such as TNB with multiple, strongly complexing functional groups. In contrast, such clays adsorb some 1% by weight of more hydrophobic NOCs of lesser ability to form complexes with interlayer cations.

ACKNOWLEDGMENTS This project was supported by grant P42 ES004911 from the National Institute of Environmental Health Science (NIEHS), National Institute of Health (NIH), and National Research Initiative Competitive Grants from the USDA National Institute of Food and Agriculture. The contents are solely the responsibility of the authors and do not necessarily represent the official views of these federal agencies. REFERENCES Aggarwal, V., Li, H., and Teppen, B. J. (2006a), Triazine adsorption by saponite and beidellite clay minerals, Environ. Toxic. Chem. 25, 392–399. Aggarwal, V., Li, H., Boyd, S. A., and Teppen, B. J. (2006b), Enhanced sorption of trichloroethene by smectite clay exchanged with Cs þ , Environ. Sci. Technol. 40, 894–899. Ahmad, I., Dines, T. J., Rochester, C. H., and Anderson, J. A. (1996), IR study of nitrotoluene adsorption on oxide surfaces, J. Chem. Soc. Faraday Trans. 92, 3225–3231. Bailey, G. W. and White, J. L. (1970), Factors influencing the adsorption, desorption, and movement of pesticides in soil, in Residue Reviews: Residues of Pesticides and Other Foreign Chemicals in Foods and Feeds, Gunther, F. A. and Gunthereds J. D. eds., Vol. 32, Springer-Verlag, , New York, PP. 29–92. Baitinger, W., Schleyer, P. V. R., Murty, T. S. S. R., and Robinson, L. (1964), Nitro groups as proton acceptors in hydrogen bonding, Tetrahedron 20, 1635–1647. Barriuso, E., Laird, D. A., Koskinen, W. C., and Dowdy, R. H. (1994), Atrazine desorption from smectites, Soil Sci. Soc. Am. J. 58, 1632–1638. Borek, F. (1963), Effect of p-CH2X substituents on vibrational frequencies of the aromatic nitro group, Naturwissenschaften 50, 471–472. Boyd, S. A., Mortland, M. M., and Chiou, C. T. (1988b), Sorption characteristics of organic compounds on hexadecyltrimethylammonium smectite, Soil Sci. Soc. Am. J. 52, 652–657. Boyd, S. A., Lee, J. F., and Mortland, M. M. (1988c), Attenuating organic contaminant mobility by soil modification, Nature 333, 345–347. Boyd, S. A., Sun, S., Lee, J. F., and Mortland, M. M. (1988a), Pentachlorophenol sorption by organoclays, Clays Clay Miner. 36, 125–130. Boyd, S. A., Sheng, G., Teppen, B. J., and Johnston, C. T. (2001), Mechanisms for the adsorption of substituted nitro-

69

benzenes by smectite clays, Environ Sci. Technol. 35, 4227–4234. Celis, R., Cox, M. C., Hermosin, M. C., and Cornejo, J. (1997), Sorption of thiazafluron by iron - and humic acid-coated montmorillonite, J. Environ. Qual. 26, 472–479. Chappell, M. A., Laird, D. A., Thompson, M. L., Li, H., Teppen, B. J., Aggarwal, V., Johnston, C. T., and Boyd, S. A. (2005), Influence of smectite hydration and swelling on atrazine sorption behavior, Environ. Sci. Technol. 39, 3150–3156. Charles, S. M., Li, H., Teppen, B. J., and Boyd, S. A. (2006a), Quantifying the availability of clay surfaces in soils for adsorption of organic contaminants and pesticides, Environ. Sci. Technol. 40, 7751–7756. Charles, S. M., Teppen, B. J., Li, H., Laird, D. A., and Boyd, S. A. (2006b), Exchangeable cation hydration properties strongly influence soil sorption of nitroaromatic compounds, Soil Sci. Soc. Am. J. 70, 1470–1479. Charles, S. M., Teppen, B. J., Li, H., and Boyd, S. A. (2008), Fractional availability of smectite surfaces in soils for adsorption of nitroaromatic compounds in relation to soil and solute properties, Soil Sci. Soc. Am. J. 72, 586–594. Chatterjee, R., Laird, D. A., and Thompson, M. L. (2008), Interactions among K þ -Ca2 þ exchange, sorption of m-dinitrobenzene, and smectite quasicrystal dynamics, Environ. Sci. Technol. 42, 9099–9103. Chiou, C. T. (2002), Partition and Adsorption of Organic Contaminants in Environmental Systems, Wiley, Hoboken, NJ. Chiou, C. T. and Shoup, T. D. (1985), Soil sorption of organic vapors and effects of humidity on sorptive mechanism and capacity, Environ. Sci. Technol. 19, 1196–1200. Chiou, C. T., Peters, L. J., and Freed, V. H. (1979), A physical concept of soil-water equilibria for nonionic organic compounds, Science 206, 831–832. Chiou, C. T., Porter, P. E., and Schmedding, D. W. (1983), Partition equilibria of nonionic organic compounds between soil organic matter and water, Environ. Sci. Technol. 17, 227–231. Chiou, C. T., Kile, D. E., Rutherford, D. W., Sheng, G., and Boyd, S. A. (2000), 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. 34, 1254–1258. Conduit, C. P. (1959), Ultraviolet and infrared spectra of some aromatic nitro-compounds, J. Chem. Soc. 3273–3277. Evangelou, V. P. (1998). Environmental Soil and Water Chemistry: Principles and Applications, Wiley, New York. Friedman, H. L. and Krishnan, C. V. (1973), Thermodynamics of ionic hydration, in Water: A Comprehensive Treatise, Vol. 3, Aqueous Solutions of Simple Electrolytes, Franks, F., ed, Plenum, New York, PP. 1–118. Frisch, M. J., Trucks, G. W., et al. (1998), Gaussian 98, Revision A.9, Gaussian, Inc., Pittsburgh, PA. Fusi, P., Ristori, G. G., and Franci, M. (1982), Adsorption and catalytic decomposition of 4-nitrobenzenesulphonylmethlycarbamate by smectite, Clays Clay Miner. 30, 306–309.

70

COMPREHENSIVE STUDY OF ORGANIC CONTAMINANT ADSORPTION BY CLAYS

Gaines, R. V., Skinner, H. C. W., Foord, E. E., Mason, B., and Rosenzweig, A. (1997), Dana’s New Mineralogy, Wiley, New York. Green, J. H. S. and Lauwers, H. A. (1971), Vibrational spectra of benzene derivatives — XIII The nitrobenzenes, Spectrochim. Acta. 27A, 817–824. Green, R. E. (1974), Pesticide-clay-water interactions, in Pesticides in Soil and Water, Guenzi, W. D. ed., Soil Science Society of America, Madison, WI PP. 3–37. Grundl, T. and Small, G. (1993), Mineral contributions to atrazine and alachlor sorption in soil mixtures of variable organic carbon and clay content, J. Contam. Hydrol. 14, 117–128. Haderlein, S. B. and Schwarzenbach, R. P. (1993), Adsorption of substituted nitrobenzenes and nitrophenols to mineral surface, Environ. Sci. Technol. 27, 316–326. Haderlein, S. B., Weissmahr, K. W., and Schwarzenbach, R. P. (1996), Specific adsorption of nitroaromatic explosives and pesticides to clay minerals, Environ. Sci. Technol. 30, 612–622. Hassett, J. J., Banwart, W. L., Wood, S. G., and Means, J. C. (1981), Sorption of a-naphthol: Implications concerning the limits of hydrophobic sorption, Soil Sci. Soc. Am. J. 45, 38–42. Jaynes, W. F. and Boyd, S. A. (1990), Trimethylphenylammoniumsmectite as an effective adsorbent of water soluble aromatic hydrocarbons, J. Air Waste Manage. Assoc. 40, 1649–1653. Jaynes, W. F. and Boyd, S. A. (1991a), Clay mineral type and organic compound sorption by hexadecyltrimethylammoniumexchanged clays, Soil Sci. Soc. Am. J. 55, 43–48. Jaynes, W. F. and Boyd, S. A. (1991b), Hydrophobicity of siloxane surface in smectites as revealed by aromatic hydrocarbon adsorption from water, Clays Clay Miner. 39, 428–436. Johnston, C. T. and Sposito, G. (1987), Disorder and early sorrow: Progress in the chemical speciation of soil surfaces, in Future Developments in Soil Science Research. Boersma, L. L. ed., Soil Science Society of America, Madison, WI, PP. 89–100. Johnston, C. T. and Premachandra, G. S. (2001), Polarized ATRFTIR study of smectite in aqueous suspension, Langmuir 17, 3712–3718. Johnston, C. T., Sposito, G., and Earl, W. L. (1993), Surface spectroscopy of environmental particles by Fourier transform infrared and nuclear magnetic resonance spectroscopy, in Environmental Particles, Vol. 2 in Environmental Analytical and Physical Chemistry Series, Buffle J. and van Leeuwen, H. P., eds., Lewis, Boca Raton, FL, PP. 1–36. Johnston, C. T., Boyd, S. A., Teppen, B. J., and Sheng, G. (2004), Sorption of nitroaromatic compounds on clay surfaces, in Handbook of Layered Materials, Auerbach, S. M., Carrado, K. A., and Dutta, P. K., eds., Marcel Dekker, New York, PP. 155–189. Johnston, C. T., De Oliveira, M. F., Teppen, B. J., Sheng, G., and Boyd, S. A. (2001), Spectroscopic study of nitroaromatic-smectite sorption mechanisms, Environ. Sci. Technol. 35, 4767–4772. Johnston, C. T., Sheng, G., Teppen, B. J., Boyd, S. A., and de Oliveira, M. F. (2002), Spectroscopic study of dinitrophenol herbicide sorption on smectite, Environ. Sci. Technol. 36, 5067–5074. Jonathan, N. B. H. (1960), Relations between force constants, bond orders, bond lengths, and bond frequencies for some nitrogenoxygen bonds, J. Molec. Spectrosc. 4, 75–83.

Karickhoff, S. W. (1984), Organic pollutant sorption in aquatic systems, J. Hydraul. Eng. 110, 707–735. Karickhoff, S. W., Brown, D. S., and Scott, T. A. (1979), Sorption of hydrophobic pollutants on natural sediments, Water Resour. Res. 13, 241–248. Kile, D. E., Chiou, C. T., Zhou, H., Li., H., and Xu, O. (1995), Partition of nonpolar organic pollutants from water to soil and sediment organic matters, Environ. Sci. Technol. 29, 1401–1406. Kukkadapu, R. K. and Boyd, S. A. (1995), Tetramethylphosphonium- and tetramethylammonium-smectites as adsorbents of aromatic and chlorinated hydrocarbons: Effect of water on adsorption efficiency, Clays Clay Miner. 43, 318–323. Laird, D. A. (2006), Influence of layer charge on swelling of smectites, Appl. Clay Sci. 34, 74–87. Laird, D. A. and Shang, C. (1997), Relationship between cation exchange selectivity and crystalline swelling in expanding 2:1 phyllosilicates, Clays Clay Miner. 45, 681–689. Laird, D. A., Shang, C., and Thompson, M. L. (1995), Hysteresis in crystalline swelling of smectites, J. Colloid Interface Sci. 171, 240–245. Laird, D. A., Barriuso, E., Dowdy, R. H., and Koskinen, W. C. (1992), Adsorption of atrazine on smectites, Soil Sci. Soc. Am. J. 56, 62–67. Laird, D. A., Yen, P. Y., Koskinen, W. C., Steinheimer, T. R., and Dowdy, R. H. (1994), Sorption of atrazine on soil clay components, Environ. Sci. Technol. 28, 1054–1061. Lawrence, M. A. M., Kukkadapu, R. K., and Boyd, S. A. (1998), Adsorption of phenol and chlorophenols by tetramethylammonium- and tetramethylphosphonium-exchanged montmorillonite, Appl. Clay Sci. 13, 13–20. Leboeuf, E. J. and Weber, W. J. (1997), 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. 31, 1697–1702. Lee, J. F., Crum, J., and Boyd, S. A. (1989), Enhanced retention of organic contaminants by soils exchanged with organic cations, Environ. Sci. Technol. 23, 1365–1372. Lee, J. F., Mortland, M. M., Chiou, C. T., Kile, D. E., and Boyd, S. A. (1990), Adsorption of benzene, toluene, and xylene by 2 tetramethylammonium-smectites having different charge-densities, Clay Clay Miner. 38, 113–120. Li, H., Teppen, B. J., Johnston, C. T., and Boyd, S. A. (2004a), Thermodynamics of nitroaromatic compound adsorption from water by smectite clay, Environ. Sci. Technol. 38, 5433–5442. Li, H., Teppen, B. J., Laird, D. A., Johnston, C. T., and Boyd, S. A. (2004b), Geochemical modulation of pesticide sorption on smectite clay, Environ. Sci. Technol. 38, 5393–5399. Li, H., Sheng, G., Teppen, B. J., Johnston, C. T., and Boyd, S. A. (2003), Sorption and desorption of pesticides by clay minerals and humic acid-clay complexes, Soil Sci. Soc. Am. J. 67, 122–131. Li, H., Teppen, B. J., Laird, D. A., Johnston, C. T., and Boyd, S. A. (2006), Effects of increasing potassium chloride and calcium chloride ionic strength on pesticide sorption by K- and Casmectite, Soil Sci. Soc. Am. J. 70, 1889–1895.

REFERENCES

Li, H., Pereira, T. R., Teppen, B. J., Laird, D. A., Johnston, C. T., and Boyd, S. A. (2007), Ionic strength-induced formation of smectite quasicrystals enhances nitroaromatic compound sorption, Environ. Sci. Technol. 41, 1251–1256. Liu, C., Li, H., Teppen, B. J., Johnston, C. T., and Boyd, S. A. (2009), Mechanisms associated with the high adsorption of dibenzo-pdioxin from water by smectite clays, Environ. Sci. Technol. 43, 2777–2783. MacEwan, D. M. C. and Wilson, M. J. (1980), Interlayer and intercalation complexes of clay minerals, in Crystal Structures of Clay Minerals and Their X-Ray Identification, Brindley, G. W. and Brown, G., eds., Mineralogical Society, London, PP. 197–248. Margulies, L., Rozen, H., and Banin, A. (1988), Use of X-ray powder diffraction and linear dichroism methods to study the orientation of montmorillonite clay particles, Clays Clay Miner. 36, 476–479. Mortland, M. M. (1970), Clay-oragnic complexes and interactions, Adv. Agron. 22, 75–117. Mortland, M. M. (1986), Mechanisms of adsorption of non-humic organic species by clay, in Interactions of Soil Minerals with Natural Organics and Microbes (special publication 17.), Huang, P. M. and Schritzer, M., eds., Soil Science Society of America, Madison, WI, PP. 59–76. Nyquist, R. A. and Settineri, S. E. (1990), Infrared study of substituted nitrobenzenes in carbon tetrachloride and chloroform solutions, Appl. Spectrosc. 44, 1552–1557. Onken, B. M. and Traina, S. J. (1997), The sorption of pyrene and anthracene to humic acid-mineral complexes: Effect of cosolute, J. Environ. Qual. 26, 132–138. Pennell, K. D., Boyd, S. A., and Abriola, L. M. (1995), Surface area of soil organic matter reexamined, Soil Sci. Soc. Am. J. 59, 1012–1018. Pereira, T. R., Laird, D. A., Johnston, C. T., Teppen, B. J., Li, H., and Boyd, S. A. (2007), Mechanism of dinitrophenol herbicide sorption on smectites in aqueous suspensions at varying pH, Soil Sci. Soc. Am. J. 71, 1476–1481. Pereira, T. R., Laird, D. A., Thompson, M. L., Johnston, C. T., Teppen, B. J., Li, H., and Boyd, S. A. (2008), Role of smectite quasicrystal dynamics in adsorption of dinitrophenol, Soil Sci. Soc. Am. J. 72, 347–354. Pignatello, J. J. and Xing, B. (1996), Mechanisms of slow sorption of organic chemicals to natural particles, Environ. Sci. Technol. 30, 1–11. Pils, J. R. V., Laird, D. A., and Evangelou, V. P. (2007), Role of cation demixing and quasicrystal formation and breakup on the stability of smectitic colloids, Appl. Clay Sci. 35, 201–211. Pusina, A., Liu, W., and Gessa, C. (1992). Influence of organic matter and its clay complexes on metolachlor adsorption on soil, Pesticide Sci. 36, 283–286. Ras, R. H. A., Schoonheydt, R. A., and Johnston, C. T. (2007), Relation between s-polarized and p-polarized internal reflection spectra: Application for the spectral resolution of perpendicular vibrational modes, J. Phys. Chem. A 111, 8787–8791.

71

Ras, R. H. A., Johnston, C. T., Franses, E. I., Ramaekers, R., Maes, G., Foubert, P., de Schryver, F. C., and Schoonheydt, R. A. (2003), Polarized infrared study of hybrid Langmuir-Blodgett monolayers containing clay mineral nanoparticles, Langmuir 19, 4295–4302. Roberts, M. G., Rugh, C. L., Li, H., Teppen, B. J., and Boyd, S. A. (2007a), Reducing bioavailability and phytotoxicity of 2,4dinitrotoluene by sorption on K-smectite clay, Environ. Toxic. Chem. 26, 358–360. Roberts, M. G., Rugh, C. L., Li, H., Teppen, B. J., and Boyd, S. A. (2007b), Geochemical modulation of bioavailability and toxicity of nitroaromatic compounds to aquatic plants, Environ. Sci. Technol. 41, 1641–1645. Saltzman, S. and Yariv, S. (1975), Infrared study of the sorption of phenol and p-nitrophenol by montmorillonite, Soil Sci. Soc. Am. J. 39, 474–479. Shang, C., Laird, D. A., and Thompson, M. L. (1995), Transmission X-ray diffraction technique for measuring crystalline swelling of smectites in electrolyte solutions, Clays Clay Miner. 43, 128–130. Sheng, G. and Boyd, S. A. (2000), Polarity effect on dichlorobenzene sorption by hexadecyltrimethylammonium-clays, Clays Clay Miner. 48, 43–50. Sheng, G., Johnston, C. T., Teppen, B. J., and Boyd, S. A. (2001), Potential contributions of smectite clays and organic matter to pesticide retention in soils, J. Agric. Food Chem. 49, 2899–2907. Sheng, G., Johnston, C. T., Teppen, B. J., and Boyd, S. A. (2002), Adsorption of dinitrophenol herbicides from water by montmorillonites, Clays Clay Miner. 50, 25–34. Teppen, B. J., Rasmussen, K., Bertsch, P. M., Miller, D. M., and Sch€afer, L. (1997), Molecular dynamics modeling of clay minerals. 1. Gibbsite, kaolinite, pyrophyllite, and beidellite, J. Phys. Chem. B 101, 1579–1587. Teppen, B. J., Yu, C.-H., Miller, D. M., and Sch€afer, L. (1998), Molecular dynamics simulations of the sorption of organic compounds at the clay mineral/aqueous solution interface, J. Comput. Chem. 19, 144–153. Theng, B. K. G. (1974), The Chemistry of Clay-Organic Reactions, Wiley, New York. Urbanski, T. and Dabrowska, U. (1959), The influence of the conjugation on the position of the infra-red band of the nitro group in some aromatic nitro-compounds, Bull. Acad. Polonaise Sci. 7, 235–237. Weissmahr, K. W., Haderlein, S. B., Schwarzenbach, R. P., Hany, R. and Nuesch, R. (1997), In situ Spectroscopic investigations of adsorption mechanisms of nitroaromatic compounds at clay minerals, Environ. Sci. Technol. 31, 240–247. Xia, G. S. and Ball, W. P. (2000), Polanyi-based models for the competitive sorption of low-polarity organic contaminants on a natural sorbent, Environ. Sci. Technol. 34, 1246–1254. Yariv, S., Russell, J. D., and Farmer, V. C. (1966), Infrared study of the adsorption of benzoic acid and nitrobenzene in montmorillonite, Israel. J. Chem. 4, 201–213.

3 THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS MYRNA J. SIMPSON AND ANDRE J. SIMPSON 3.1. Introduction 3.2. Organic Matter–Mineral Interactions 3.3. The Role of Organic Matter–Mineral Interactions in Organic Contaminant Sorption 3.4. Summary and Synthesis of Future Research Directions

3.1. INTRODUCTION Organic matter (OM) is ubiquitously found in the environment and plays several, critical roles in environmental processes, such as OM cycling and turnover and the sorption of problematic organic chemicals (Feng et al. 2005; Kleber et al. 2007; Kang and Xing 2008). The role of OM in the sorption of organic chemicals has been an active area of research since the early 1960s. Early studies on the sorption of organic chemicals to soil recognized that OM is an important soil characteristic, especially for nonionic hydrophobic chemicals. For example, Lambert et al. (1965) proposed that there was an “active” fraction of soil OM that was responsible for sorption of chemicals to soil. Other notable studies suggested that the variation in diuron sorption coefficients was due to the accessibility of soil OM groups on colloidal surfaces (Hance 1965). Doherty and Warren (1969) later hypothesized that the relationship between herbicide binding and soil OM was governed by some other physical or chemical factor that soil OM itself was correlated. These studies (Doherty and Warren 1969; Hance 1965; Lambert et al. 1965), followed by subsequent studies (Bailey and White 1970; Hance 1969; Lambert 1967), concluded that soil

OM was playing a principal role in the sorption of chemicals to soil and defined future research directions (Fig. 3.1). Although these types of studies continued in the 1970s, studies on soil fractions became more prevalent (Fig. 3.1) with emphasis on defining the role of OM and later, OM structure in the sorption of organic chemicals. For example, Hayes (1970) demonstrated the value of employing chemical fractions in herbicide sorption studies and showed that the use of OM fractions had not yet been explored to its full potential. The work of Karickhoff et al. (1979) established a linear relationship between sorption partition coefficients (Kd values) with soil organic carbon contents. Hence, the practice arose of reporting organic carbon normalized (KOC) sorption coefficients in addition to other experimental parameters emerged. Empirical correlations between KOC values and the octanol–water partition coefficient (KOW) were also proposed (Chiou et al. 1979; 1982; 1983; 1985; Chiou 2002) and suggested that a soil or sediment Koc value could be predicted from the KOW value because sorption mechanisms were due mainly to phase partitioning. However, research showed that octanol is a poor surrogate for soil OM (Mingelgrin and Gerstl 1983; Xing et al. 1994a), and the emphasis to further understand how soil OM chemical structure governed the sorption of organic contaminants in soil was reinforced by these studies. Figure 3.1 depicts an overview of three main research “stages”; as mentioned previously, early studies with whole soils identified the importance of organic matter content in organic chemical sorption processes. Several studies showed that clay minerals, especially in the presence of water, did not sorb significant amounts of organic chemicals, thus, a strong focus on OM and OM fractions ensued. Researchers used

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

73

74

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

Figure 3.1. An illustration of the progression of organic chemical sorption research since 1965. Early studies focused on the sorption of chemicals to soils (stage I) and propose mechanisms of sorption. With the onset of analytical techniques, the focus of the research shifted to interactions with soil components (stage II), which enabled researchers to link sorption behavior to structure. More recent studies have emphasized the importance of the relationship between OM and minerals on chemical sorption (stage III).

soil OM fractions [fulvic acid (FA), humic acid (HA), and humin], which are operationally defined and isolated using chemical fractionation, to gain insight into soil OM and contaminant interactions with the main objective of elucidating sorption mechanisms. Analytical instrumentation that were typically designed for simple mixtures and small molecules, often limited their application to “whole” soils. Therefore, using soil OM fractions as sorbents facilitated the study of macroscopic sorption mechanisms to individual soil OM fractions. Consequently, macroscopic studies with only soil OM chemical fractions began to emerge with the specific goal to develop soil OM structure–contaminant relationships for nonionic hydrophobic organic contaminants (Fig. 3.1). Interaction studies with charged and/or water-soluble organic chemicals were also conducted such that mechanisms could be defined and eventually predicted. Although studies with isolated OM or model compounds were critical for identifying and confirming the soil characteristics that governed organic chemical sorption processes, reports began to emerge showing that sorption processes were not regulated by OM content alone. For example, Garbarini and Lion (1986) reported that toluene and trichloroethylene sorption to FA, HA, and humin could not be explained by the organic carbon content alone. Their results suggested that oxygen content in addition to carbon content provides a more accurate prediction of toluene and trichloroethylene sorption (Garbarini and Lion 1986). In the late 1980s, researchers began to focus more on soil OM structure. Grathwohl (1990) demonstrated that samples of varying degrees of digenesis, namely, coals versus HA, produced varying Koc values for a series of chlorinated aliphatic hydrocarbons. A linear

relationship between log Koc values and the H/O atomic ratio further emphasized the importance of OM composition in addition to OM content. Numerous other studies have provided strong evidence for relationships between soil OM fraction structural parameters and Koc values (Chen et al. 1996; Xing 2001; Kang and Xing 2005). Studies with constructed OM–mineral complexes suggested that the mineral phase plays an indirect role by governing organic matter accessibility at the soil–water interface (Murphy et al. 1994; Jones and Tiller 1999). The evidence from these results combined with studies with soil OM fractions, which could not be explained by OM structure alone, resulted in more recent studies (represented by stage III in Fig. 3.1) whose sole focus is on examining the precise and indirect role of clay minerals and other inorganic soil constituents on controlling OM accessibility and reactivity at the soil–water interface. A large portion of OM in soil is associated with inorganic mineral phases, but only since the 1990s has it been acknowledged that this relationship can alter the environmental reactivity of OM (Baldock and Skjemstad 2000; Feng et al. 2005; Kleber et al. 2007; Kang and Xing 2008). This chapter focuses on recent research on OM–mineral interactions and their role in the sorption of nonionic hydrophobic organic contaminants. As indicated previously, studies of OM–mineral interactions have increased considerably since the 1990s but have primarily focused primarily on interactions with nonionic hydrophobic organic contaminants. This increase in research activity is not limited to organic chemicals but is conducted in parallel with research that aims to define the role of OM–mineral associations that may limit or slow the turnover of easily degradable OM

75

ORGANIC MATTER–MINERAL INTERACTIONS

(Baldock and Skjemstad 2000; Chenu and Plante 2006; Mikutta et al. 2007). Consequently, OM–mineral interaction studies exist well beyond the context of organic contaminant interactions and have been studied within the framework of soil biogeochemistry and soil OM responses to climate change. This chapter will first review OM–mineral studies followed by the role that OM–mineral complexes play in organic chemical sorption processes. The emphasis of the review will be on the use of molecule-level techniques, such as nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS), which are both used to investigate OM–mineral and organic contaminant interactions in detail. Readers who are interested in OM–mineral studies and their role in soil OM turnover are referred to several review papers [Baldock and Skjemstad (2000); Chenu and Plante (2006); Kleber et al. (2007); K€ ogel-Knabner et al. (2008), and references cited therein].

3.2. ORGANIC MATTER–MINERAL INTERACTIONS Organic matter–mineral interactions have been studied intensively since the 1990s because of their role in soil OM turnover, preservation of labile OM in sediments, and role in contaminant sorption processes (Arnarson and Keil 2000; Feng et al. 2005; Kleber et al. 2007; Kang and Xing 2008; K€ ogel-Knabner et al. 2008). Organic matter associations with mineral phases have resulted in reduced biodegradation of OM, and it is hypothesized that minerals provide OM both physical and chemical protection from biological attack (Baldock and Skjemstad 2000; Chenu and Plante 2006; Mikutta et al. 2007). It is believed that OM may be adsorbed to mineral surfaces via six mechanisms: ligand exchange, cation bridges (including water bridges), anion exchange, cation exchange, van der Waals interactions, and hydrophobic bonding (Arnarson and Keil 2000; Tombacz et al. 2004; Feng et al. 2005). Several researchers have shown that solution chemistry (ionic strength, pH, and dominant cation) determines the conformation of the OM prior to sorption on the mineral surface. Murphy et al. (1994) illustrated how solution conditions can promote OM to adopt a coiled or stretched conformation prior to sorption onto mineral surfaces (Fig. 3.2). High ionic strength and low pH values encourage dissolved OM to adopt a coiled or condensed configuration because negative charges of OM functional groups are neutralized by solution cations. A stretched or open structure results from charge repulsion under conditions of low ionic strength and high pH solution. Consequently, dissolved OM can acquire an open or closed geometry on the basis of the solution conditions prior to sorption onto the mineral surface, which, in turn, governs the physical accessibility to OM sorption domains (Jones and Tiller 1999; Murphy et al. 1994; Schlautman and Morgan 1993). Other

Humic substance in solution

Ligandexchange? Humic substance sorbed to surface

pH 6.5 Low I

pH 4.5 High I Low I Ca Na Ca Na Ca

Ligandexchange?

Na

Ca Ca Ca

Ternary Ligandsurface exchange? complex?

Ca

Ca Ca

Al- or Fe-Oxide Surface

Figure 3.2. Illustration of organic matter conformation with varying solution conditions; solution conditions (ionic strength) have been employed to induce a coiled or stretched OM structure prior to sorption to mineral surfaces [Reprinted with permission from Murphy et al. (1994)].

studies, which have explored the extent of OM sorption on mineral surfaces, have found OM sorption to increase with increasing ionic strength and decrease with increasing pH (Arnarson and Keil 2000; Baham and Sposito 1994; Satterberg et al. 2003). It has also been reported that di- and trivalent cations, namely, Ca2þ and Al3þ , enhance OM sorption by increasing the number of cation bridges (Arnarson and Keil 2000; Murphy et al. 1994; Schlautman and Morgan 1993). The concentration of dissolved OM also determines the mechanism by which OM–mineral phases form (Feng et al. 2005; Reiller et al. 2006). Mineral charge, surface area, and morphology will determine points of contact for OM and promote varying modes of OM sorption (Hur and Schlautman 2003; Meier et al. 1999; Feng et al. 2005; Kleber et al. 2007). For example, dissolved OM sorption by iron and aluminum oxides is believed to occur via ligand exchange, whereas cation bridges are believed to dominate in OM sorption to aluminosilicates (Chorover and Amistadi 2001; Zhou et al. 1994). Collectively, the aforementioned studies on OM–mineral interactions concluded that (1) sorption of dissolved OM to mineral surfaces is competitive and high-molecular-weight compounds are preferentially sorbed over low-molecularweight compounds, (2) aromatic moieties are preferentially adsorbed over aliphatic groups, (3) adsorption of OM increases inversely with pH, (4) the thickness of OM coatings on minerals varies with concentration, (5) adsorption to mineral surfaces is reversible, and (6) OM sorption is dependent on mineralogy (Arnarson and Keil 2000; Collins et al. 1995; Gu et al. 1996; Meier et al. 1999; Fein et al. 1999; Ochs et al. 1994; Theng 1982; Vermeer and Koopal 1998; Vermeer et al. 1998; Wershaw et al. 1996a,b). These conclusions are based on quantitative descriptors, such as thermodynamic data and isotherm shape, which were used to yield empirical relationships from which mechanistic information was inferred. Consequently, more recent studies have

76

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

focused on OM–mineral interactions using advanced OM characterization methods in attempts to define the underlying fundamental chemical processes (Wattel-Koekkoek et al. 2001; Feng et al. 2005; 2006; Kang and Xing 2005; Wang and Xing 2005; Reiller et al. 2006, Simpson et al. 2006; Joo et al. 2008a; Kang and Xing 2008). For example, WattelKoekkoek et al. (2001) employed 13C NMR and pyrolysis– gas chromatography (GC)-MS to characterize OM associated with different mineral types and found that kaoliniteassociated OM was rich in polysaccharide products while smectite-associated OM contained more aromatic compounds, suggesting that binding mechanisms and the quality of sorbed OM are mineral-dependent. Advanced solution-state and high-resolution magic-angle spinning (HR-MAS) NMR methods have been used to study the competitive sorption of model OM mixtures (Simpson et al. 2006). High-resolution MAS NMR is a semisolid NMR technique that probes structures at the soil–water interface, and only structures in contact with the NMR solvent are observed (i.e., pure solid domains are not observed). The model compound mixture used was comprised of four main structural groups that are found in OM: 1-palmitoyl-3-stearoyl-rac-glycerol (to mimic large aliphatic molecules with functionalities similar to those present in plant cuticles), a small peptide (Arg–Pro–Leu–Glc–NH2), maltohexose (to represent carbohydrates), and the 500–1000-molecularweight fraction of commercially prepared lignin. The model compound mixture was prepared using an equal mass of each model compound and the mixture reacted with mont2-D NMR (TOCSY) of unsorbed compounds in solution

morillonite for 48 h. The unsorbed compounds in the mixture contained signals from all compounds; however, the HRMAS NMR results showed that only the aliphatic structures sorbed to the clay surface (Fig. 3.3). These results indicate that aliphatic structures, such as those found in plant cuticles, are preferentially sorbed to montmorillonite surfaces. The other OM model compounds displayed a low affinity for the montmorillonite surface and remained in solution (Fig. 3.3). A mixture of FA and HA was also sorbed to montmorillonite and similarly, it was reported that aliphatic components had a stronger affinity for the mineral surface than did other OM components. Although this study provided detailed, NMR evidence for the preferential sorption of OM components to montmorillonite, the emphasis on the novel development of NMR did not allow for other experimental variables (mineral type and/or solution conditions) to be explored. Feng et al. (2005) employed batch experiments with peat HA and kaolinite and montmorillonite under varying ionic strength, solution cation, and pH values. The authors employed both solution-state and HR-MAS NMR to study the composition of unsorbed and sorbed OM, respectively. Parameters of the OM sorption isotherms were also determined and used to quantify the variation in binding mechanisms (i.e., ligand exchange, cation bridging, van der Waals, and hydrophobic bonding) with varying solution conditions. Ionic strength, pH, and the dominance of either Ca2þ or Na þ determined the extent of OM sorption to mineral surfaces (kaolinite and montmorillonite) in this study. The sorption isotherms were modeled using the Freundlich HR-MAS NMR (TOCSY) of sorbed compounds on montmorillonite

Carbohydrate Aliphatic

Lignin Peptide Figure 3.3. Two-dimensional solution-state and HR-MAS NMR spectra of unsorbed and montmorillonite-sorbed OM components. The NMR results show that aliphatic structures are preferentially sorbed to the mineral surface. [Data from Simpson et al. (unpublished) with experimental details in Simpson et al. (2006)].

ORGANIC MATTER–MINERAL INTERACTIONS

2.50 Hydrophobic interactions

Freundlich Kf (mgC1-N lN g-1)

2.25 2.00

Ligand exchange van der Waals forces Cation bridging

1.75 1.50 1.25 1.00 0.75 0.50 0.25 0.00

4 7 4 7 4 7 7 4 pH pH pH pH pH pH a, pH a, pH a, a, a, a, a, a, C C N N C C N N 1 1 01 0.01 01 01 0.01 01 0.0 0.0 0.0 0.0 0.0 0.0

Equilibrium Solution Conditions

Figure 3.4. Freundlich coefficients and quantification of different binding mechanisms under varying equilibrium solution conditions (ionic strength, pH, and dominant cation) during the sorption of dissolved humic acid to mineral surfaces [modified from Feng et al. (2005)].

equation, and the resulting Kf values varied significantly with solution conditions (Fig. 3.4). In addition, solution conditions were also found to vary the mode of binding. For example, Na þ (I ¼ 0.001 M) weakens cation bridging and decreases in pH-promoted ligand exchange between dissolved OM and mineral surfaces (Fig. 3.4). As the ionic strength increases in the presence of Ca2þ , both cation bridging and van der Waals interactions were enhanced by the compression of the electric double layer. Ligand exchange was estimated to account for 32% of HA sorption on clay surfaces, van der Waals 22%, and cation bridges 41% when Ca2þ was the background electrolyte. Kaolinite displayed higher or similar adsorption for OM than montmorillonite when Na þ was the dominant cation present. Although it was expected that the higher-surface-area mineral (montmorillonite) would sorb more HA, kaolinite showed a higher affinity for HA when compared to montmorillonite (when normalized to external surface area only). However, it has been suggested that OM compounds do not intercalate with expanding minerals (Baham and Sposito 1994), although intercalation has been observed for a 625-Da surfactant and Ca-montmorillonite (Salloum et al. 2000). Zhou et al. (1994) also reported increased sorption of HA with kaolinite than montmorillonite in NaCl solutions. Montmorillonite has been observed to sorb highmolecular-weight OM compounds because of its high CEC and surface area (Satterberg et al. 2003), but Chorover and Amistadi (2001) showed that montmorillonite selectively sorbed low-molecular-weight OM in comparison with kao-

77

linite. The discrepancies in these results suggest that either different types of OM are sorbed under varying solution conditions and/or multiple layers of OM may be sorbed to varying extents, depending on the mineral type. Solution-state 1 H NMR spectra of the unbound HA and 1 H HR-MAS NMR spectra of HA sorbed to kaolinite and montmorillonite 0.01 M Na þ (pH 7) are shown in Figure 3.5 (Feng et al. 2005). Sorbed OM to both kaolinite and montmorillonite are rich in CH3 groups from short side-chains found in amino acids. In addition, the presence of aromatic signals (6.5–8 ppm) suggests the presence of peptide material in both samples, but the peptide signature is more prevalent on the surface of montmorillonite (Fig. 3.5). An estimate based on the intensity of the CH3 versus CH2 signals suggests that there is more peptide material sorbed to montmorillonite than to kaolinite. The CH2 groups in the 1 H HR-MAS spectra suggest that long-chain aliphatic structures preferentially bind to both minerals, but carbohydrates largely remained in the unbound fraction. The spectrum of HA-kaolinite is clearly dominated by CH2 signals (Fig. 3.5), suggesting that long-chain polymethylene structures are preferentially sorbed to kaolinite rather than montmorillonite. A study by Wang and Xing (2005) that also utilized NMR spectroscopy corroborated these findings, and the authors concluded that aliphatic components of HA are preferentially sorbed over aromatic structures on both kaolinite and montmorillonite surfaces. Although the studies by Feng et al. (2005) and Wang and Xing (2005) provided insight into the nature of sorbed OM on kaolinite and montmorillonite mineral surfaces, these studies did not address OM molecular-weight-fractionation with sorption. Moreover, neither of these studies addressed interactions with iron or aluminum oxides, which can dominate in some soils, and hence, the results with kaolinite and montmorillonite may not apply to other mineral types. As mentioned previously, some studies have shown that high-molecular-weight OM is preferentially sorbed to some mineral surfaces (Chorover and Amistadi 2001; Collins et al. 1995; Vermeer and Koopal 1998). For example, Chorover and Amistadi (2001) found that high-molecularweight OM preferentially sorbed onto goethite, whereas lowmolecular-weight OM preferentially sorbed to montmorillonite. Reiller et al. (2006) further examined HA sorption to hematite to test whether high- or low-molecular-weight OM constituents were preferentially sorbed. Compositional shifts of HA with sorption were determined using electrospray ionization (ESI) quadrupole time-of-flight (QTOF) MS, which provides accurate molecular mass information. The sorption of HA to hematite resulted in a decline of the lowmolecular-weight (<600 Da) fraction in ESI-MS suggesting that low molecular weight HA constituents were preferentially sorbed to hematite surfaces. A significant change in the composition of the intermediate (600–900 Da) and highmolecular-weight (900–1700 Da) ranges of HA was not

78

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

(a)

*

Unbound HA

Aliphatic

Amino Acids & Polysaccharides1

Amino Acids & Polysaccharides1

Aromatic Amide

*

Aliphatic (b) Unbound HA

Aromatic Amide

*

CH2

CH2

*

HA sorbed to kaolinite

HA sorbed to montmorillonite CH3

CH3

9

8

7 1H

6

5

4

3

2

ppm

Chemical Shift

9

8

7 1H

6

5

4

3

2

ppm

Chemical Shift

Figure 3.5. Comparison of 1 H liquid-state and HR-MAS NMR spectra of the unbound humic acid (HA) and HA–mineral complexes in 0.01 M Na þ , pH 7: (a) HA–kaolinite complex, (b) HA–montmorillonite complex. Asterisks ( ) denote DMSO-d6 (NMR solvent). (Note: Other moieties such as esters, ethers, and some amino acids may also contribute strongly to this region. The aromatic signals and intensive CH3 resonance in this case result from peptide structures.) [Reprinted with permission from Feng et al. (2006).]

observed. Figure 3.6 shows the 352–362 Da range of the ESIMS spectrum for the HA and sorbed HA at various concentrations. When higher concentrations of HA were used, decreases in the ESI-MS spectrum were observed, suggesting that the preferential sorption of low-molecular-weight constituents was consistent over the range of HA concentrations used [3.3–33 mgC/L], although the selectively is less pronounced at the highest concentration of HA [33 mgC/L; Fig. 3.6]. The low-molecular-weight fraction also had a low mass defect, which is the difference between the theoretical

calculated mass and experimentally measured mass of a nucleus, suggesting that the constituents sorbed to hematite were low-molecular-weight, oxygen-containing aromatic compounds with oxygen-containing functional groups such as hydroxyl, carboxyl, or methoxyl groups. However, sorption of the low-molecular-weight fraction resulted in broadening and shifts toward a higher mass defect, which the authors interpreted as the sorption of condensed aromatics that may occur gradually after sorption of the low-molecularweight fraction.

Figure 3.6. Low-molecular-weight region of an electrospray ionization quadrupole time-of-flight (ESI-QTOF) mass spectrum of humic acid (HA) and after HA sorption at three different starting concentrations [Reprinted with permission from Rellier et al. (2006)].

ORGANIC MATTER–MINERAL INTERACTIONS

79

Adsorbed amount of HA (x/m, mg C/g)

(a) 6.0

pH 5 Q0 = 5.61 mg C/g b = 0.65, R2 = 0.99

5.0 4.0

pH 7 Q0 = 4.23 mg C/g b = 0.23, R2 = 0.98

3.0

pH 9 Q0 = 3.75 mg C/g b = 0.06, R2 = 0.95

2.0 1.0 0.0 0

5

10

15

20

25

30

35

40

50

45

Equilibrium concentration (Ce, mg C/L)

Aromatic C (108 ~ 145)

(b)

Anomeric C (96 ~ 108) -CH2O(60 ~ 96) CH3O (50 ~ 60)

Phenoloc C (145 ~ 162) C=O COOH (162 ~ 220)

Aliphatic C (0 ~ 50)

Source HA Unbound HA1

Unbound HA4 250

200

150

100 ppm

50

0

-50

Figure 3.7. (a) Sorption isotherms for humic acid (HA) sorption to goethite at varying solution pH values. (b) Solid-state 13C NMR spectra for the HA before sorption (source HA) and two samples of HA components that did not sorb to goethite (unbound HA1 and HA4). The NMR spectra suggest the preferential binding of aliphatic and carbohydrate organic matter structures [Reprinted with permission from Kang and Xing (2008).]

Kang and Xing (2008) investigated the sorption of OM to goethite using a host of spectroscopic methods to study the selectivity of OM sorption to goethite surfaces. Sorption of HA to goethite increased with decreasing solution pH (Fig. 3.7a) and resulted in high-affinity sorption isotherms that were modeled using the Langmuir equation. It was hypothesized that HA sorption increased at low pH because ligand exchange would have been encouraged under these conditions. Furthermore, at high pH values, acidic functionalities of the HA and hydroxyl groups on the mineral surface would be negatively charged, therefore decreasing electrostatic interactions. The polarity index ([O þ N/C]) of the HA decreased with sorption, suggesting the preferential sorption

of polar, nonaromatic compounds. Solid-state NMR analysis (Fig. 3.7b) showed that heteroaliphatic carbons, aliphatic alcohols, and carbohydrate moieties were preferentially sorbed to goethite. Interestingly, a major change in polymethylene carbon distribution was not observed as reported by Feng et al. (2005) with kaolinite, which further exemplifies the role of different minerals in selectivity of OM binding. Because of this observation, Kang and Xing (2008) suggested that the sorbed HA compounds were likely high in acidic functional groups that would promote ligand exchange on the goethite surface. Once these compounds are sorbed, their presence on the surface promotes the secondary sorption of long aliphatic chains and/or larger

80

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

aromatic ring structures through hydrophobic interactions. This hypothesis is consistent with results from NamjesnikDejanovic et al. (2000), who suggested that low-molecularweight components of FA sorbed to goethite during the initial stages of sorption and are later replaced by higher-molecularweight compounds as sorption continues. Using diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy and size exclusion chromatography (SEC), Kang and Xing (2008) proposed a sequential process of HA sorption to goethite: (1) electrostatic attraction of polar functional groups from low- and intermediate-molecular-weight compounds are attracted to the surface initially; (2) this is followed by ligand exchange with carboxylic groups to form inner-sphere complexes; and (3) hydrophobic interactions with high-molecular-weight, long-chain aliphatic groups results in a hydrophobic coating. The competitive nature of dissolved OM sorption and desorption (i.e., replacement by other OM molecules) is difficult to establish because most researchers examine OM and OM–mineral associations once an apparent equilibrium is reached. Joo et al. (2008a) examined dissolved OM sorption–desorption on hydroxide-coated sand with emphasis on kinetic changes in comparison to dissolved OM– mineral composition at equilibrium. Spectroscopic (E4/E6) results suggested that a two-stage sorption process was occurring. The initial and fast sorption step was selective toward low-molecular-weight compounds. After 4 h, the replacement of low-molecular-weight compounds with high-molecular-weight compounds was observed. The authors postulated that the low-molecular-weight compounds initially modify the surface of the mineral, which then facilitates the sorption of more hydrophobic, highmolecular-weight compounds. The OM sorption–desorption process proposed by Joo et al. (2008a) is depicted in Figure 3.8. Ligand exchange and electrostatic interactions between the dissolved low-molecular-weight-components and the mineral surface result in fast sorption during the initial stages. As the equilibration time continues, some lowmolecular-weight compounds are replaced by high-molecular-weight compounds through hydrophobic interactions (as shown in Fig. 3.8). These results are consistent with those of Kang and Xing (2008) as well as conclusions drawn by other researchers (Avena and Koopal 1999; NamjesnikDejanovic et al 2000; Hur and Schlautman, 2003). These studies highlight the importance of studying the kinetic aspects of OM–mineral interactions. Kleber et al. (2007) proposed a zonal model of OM– mineral associations in soil that organizes OM into various layers (Fig. 3.9). The authors constructed this model on the basis of several experimental observations in soils; however, this model is also consistent with molecule-level observations from OM–mineral interaction studies. The zonal model (shown in Fig. 3.9) illustrates varying types of OM interactions on mineral surfaces and also conceptually illustrates

OM–OM interactions that may occur after the initial sorption of the first set of OM components [as observed in several OM–mineral interaction studies such as those by Joo et al. (2008a) and Kang and Xing (2008)]. The formation of OM coatings or layers on mineral surfaces has been used to explain several observations in soils and is also important for understanding organic contaminant sorption processes (discussed in detail in Section 3.3). For example, Kleber et al. (2007) postulate that the zonal model can explain observed trends such as decrease in the C/N (carbon/nitrogen) ratio with decreasing particle size and increasing particle density, the preferential sorption of high-molecular-weight hydrophobic OM compounds in both soils and sediments, varying rates of turnover in OM pools (i.e., rapid vs. slow turnover), accumulation of OM in A horizons that cannot be explained by monolayer coverage, and poor correlations between OM contents and mineral phases. The zonal model shows polar OM components interacting with mineral surfaces via electrostatic interactions. The model also suggests that nonpolar molecules such as phenanthrene can interact directly with the mineral surface via hydrophobic interactions. Although this is consistent with some published experimental data (Carmo et al. 2000; Joo et al. 2008b), several other studies have not been able to detect phenanthrene and other hydrophobic compound sorption to mineral surfaces due to competition with water (Salloum et al. 2001; Feng et al. 2005; Bonin and Simpson 2007a). The OM–OM interactions that occur in the zone of hydrophobic interactions are believed to result in OM layers. This is consistent with experimental observations that report the initial sorption of low-molecular-weight compounds followed by highmolecular-weight compounds (Joo et al. 2008a). The kinetic region represents the area of exchange of molecules, and the authors note that the thickness of this region and its properties are dependent on soil solution characteristics (pH, ionic strength, and dominant cations) and OM exchange kinetics. Indirect evidence was predominantly used to construct the zonal model; however, the model is consistent with some of the molecule-level evidence generated to date. Future molecule-level studies coupled with the continued advancement of OM characterization methods will help refine the model such that it can be used to explain OM turnover and dynamics as well as the fate and transport of organic chemicals.

3.3. THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN ORGANIC CONTAMINANT SORPTION The fundamental processes that govern organic contaminant sorption to soil and sedimentary OM have been studied for many decades. Arguably, the main goal of this body of research is to elucidate the property or properties of soils and sediments that control the sorption of organic chemicals

81

COO

H

E

L

DOM

DOM

COOH

E

HO

COO

COOH O

DOM O L

0.25

Quartz

Metal (Hydr)oxides

+

OH2

COO

DOM

O

O

HO

DOM

DOM

COOH

E

HO

COO

COOH O

DOM O L

Quartz

Metal (Hydr)oxides

+

OH2

COO

DOM

Equilibration Time (hours)

COO L

O

Metal

COO

H O

O

HO

DOM

DOM

COO L

O

Metal

COO

H

H

COOH

DOM

COOH

O

O

HO

O

COO

DOM

4

O L

Quartz

Metal (Hydr)oxides

O

H

HO

COO

DOM

Figure 3.8. Illustration of dissolved organic matter (DOM) sorption to modified sand surfaces with time. Initially, low-molecular-weight DOM is sorbed (0.25 h) followed by the sorption of high-molecular-weight DOM (4 h). Electrostatic interactions, hydrophobic bonding, and ligand exchange all contribute to DOM sorption processes. [Reprinted with permission from Joo et al. (2008a).]

0

Quartz

Metal

COOH

COOH

HO

O

Metal (Hydr)oxides

OH2+

DOM

O

OH2+

OH2+

COOH

HO

O

H

OH

HO

DOM

COOH

O

: Same size unit DOM : Ligand exchange : Electrostatic interaction : Hydrophobic bonding

COOH

DOM

L

DOM COO

Metal

COO

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

contact zone R R

O H H

H O H

O H

O

O

CH3

O

O

H O

O

H

M2+

O

H

OH H3C

N H

HO

O OH

O

M

O O

O

O

O

OH O

O

M2+

O

M2+

OH

H

CH3

CH3

O

H3C H

O

O

H O

O

H3C

O

M2+

O H

O 2+

CH3

O

M2+

O

O

H

OH

H

M2+

O

O

H

H3C

H H3C

O

O

H

H

O OH

Fe O

uncharged 1:1 siloxane (kaolinite)

O

H

CH3

Fe O Fe OH Fe O

M2+

M2+

H3C

O HO P Fe O O Fe OH Fe

O

M2+

O

NH R

O

O H

H

O

SH2+ O R

O

HN

O

O

H

O O

H O

OH

O

O NH

O

2:1 mineral with coating of hydrous iron oxide

2+

O

O M

O

kinetic zone (outer region)

H

low charge 2:1 mineral (smectite) with protein conditioning

H3N+

zone of hydrophobic interactions

H

82

= octahedral charge OH

= tetrahedral charge 2+

M

= metal cation

Figure 3.9. The zonal model of organic matter–mineral interactions proposed by Kleber and colleagues (2007). The three zones—contact zone, zone of hydrophobic interactions, and kinetic zone—play varying roles in soil organic matter turnover and contaminant sorption. [Reprinted with permission from Kleber et al. (2007).]

such that sorption can be better predicted and targeted remedial methods can be developed. As discussed in the introduction section, mainstream sorption studies of the past focused on developing correlations between Koc values and organic matter structural properties (Fig. 3.1). Numerous studies have provided evidence for proportional relationships between the amount of contaminant sorption and OM aromaticity (Xing 2001; Gauthier et al. 1987; Perminova et al. 1999; Chefetz et al. 2000; Chin et al. 1997), specific OM domains (Salloum et al. 2000; Gunasekara et al. 2003; Simpson et al. 2003; Mao et al. 2002), and/or OM polarity (Xing et al. 1994b; Gauthier et al. 1987; Grathwohl 1990; Kopinke et al. 2001). In addition to this, several reports suggest that the physical conformation (as illustrated in Fig. 3.2) is responsible for the variability in Koc values of organic chemicals (Engebretson and von Wandruszka 1994; Jones and Tiller 1999; Murphy et al. 1994; Xing et al. 1994b). Garbarini and Lion (1986) showed that toluene and trichloroethylene sorption to OM fractions could not be explained by structure and OM distribution alone. Laor et al. (1998)

observed that HA phenanthrene Koc values decreased when the HA was associated with a mineral phase, which is consistent with earlier reports that accessibility to OM sorption domains is indirectly governed by OM–mineral interactions (Jones and Tiller 1999; Murphy et al. 1994). Increases in contaminant Koc for OM fractions have also been reported elsewhere in the literature (Salloum et al. 2001; Chen et al. 1996; Garbarini and Lion 1986; Nearpass 1976) and suggest that contaminant accessibility to OM is altered during OM fractionation. These studies hypothesized that both OM structure and physical conformation are important considerations in contaminant sorption to OM, and collectively, these studies show that sorption to OM is higher than when the same OM is associated with minerals. Thus, it is believed that the association with minerals results in a reduction of accessibility to OM sorption domains. The conclusions of these earlier studies resulted from a variety of types of experimental evidence and provided a new hypothesis to be explored by other researchers. More recently, there have been two main approaches to study the role of

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN ORGANIC CONTAMINANT SORPTION

16 Montmorillonite

14

Kaolinite

Koc (l/g)

12 10 8 6 4 2 0 4 7 7 4 7 4 4 7 pH pH pH pH pH a, pH a, pH pH a, a, a, a, a, a, C C N N C C N N 1 1 1 1 01 01 01 01 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Ionic strength, cation & pH

Figure 3.10. Organic carbon normalized sorption coefficients (Koc) values for phenanthrene sorption to organic matter–mineral complexes constructed under varying solution conditions (ionic strength, pH, and dominant cation). The variation in Koc was attributed to changes in organic matter conformation on the mineral surfaces (as illustrated in Fig. 3.2). [Data from Feng et al. (2006).]

OM–mineral interactions in organic contaminant sorption using molecule-level techniques: studies with constructed OM-mineral complexes and studies with soil OM fractions. Feng et al. (2005, 2006) examined the sorption of phenanthrene to constructed OM-mineral complexes that were prepared using varied solution conditions and characterized in detail using advanced NMR methods (Fig. 3.4). Phenanthrene sorption behaviour to the constructed OM–mineral complexes was also measured (Feng et al. 2006) to establish relationships between phenanthrene Koc values and OM conformation on mineral surfaces. The observed sorption coefficients varied considerably with OM conformation (Fig. 3.10). The phenanthrene sorption coefficients for OM-mineral complexes in general increased with lower pH values. From the understanding of OM-mineral interactions, it was anticipated that fewer OM binding sites were available at lower pH values (coiled configuration) because some hydrophobic domains are “hidden” or “protected.” This hypothesis is consistent with the observation that OM loading on both montmorillonite and kaolinite was higher at pH 4 (Feng et al. 2005, 2006). In addition, the preferential sorption of different OM compounds was believed to play a role in the resulting sorption coefficients. The 1H HR-MAS NMR results suggested that polymethylene compounds are prevalent on the surface of kaolinite while montmorillonite sorbed more aromatic/peptide compounds (Fig. 3.4). Peptides on the OM-mineral surface can block sorbate access to polymethylene organic matter, which is reported to sorb

83

high amounts (i.e., high Koc values) of phenanthrene (Salloum et al. 2002; Mao et al. 2002). The dominance of polymethylene structures at the OM-kaolinte surface as observed by 1H HR-MAS NMR, supports the increased uptake of phenanthrene by kaolinite-HA complexes in comparison to montmorillonite-HA complexes (Fig. 3.10). Wang and Xing (2005) also found that OM sorbed to kaolinite surfaces are more aliphatic in nature than those sorbed to montmorillonite, resulting in a more hydrophobic and nonpolar OM coating that enhances phenanthrene uptake. The concept of sorption sites being blocked by OM components via OM–OM interactions on mineral surfaces has been the focus of several studies (Feng et al. 2006; Joo et al. 2008b; Kang and Xing 2005; Van Emmerik et al. 2007). Observations with soils and sediments have revealed reduced porosity and surface area due to OM–OM associations (Mikutta et al. 2004; Kleber et al. 2007; Wagai et al. 2009) which is consistent with the zonal model discussed in Section 3.2. Kang and Xing (2005) performed a detailed study where “layers” of OM were sequentially removed from soil and phenanthrene sorption was measured on the solid fraction. The OM remaining in the solid fraction (i.e., nonextractable fraction) was characterized using solid-state 13 C NMR spectroscopy. The aromaticity and polarity of OM in the soil decreased with sequential extraction (Fig. 3.11). The nonextractable OM in the humin fractions (HuH, which had a high organic carbon content and HuL, which had a low organic carbon content) were enriched in polymethylene carbon, which has been shown to sorb high amounts of phenanthrene (Salloum et al. 2002; Mao et al. 2002). Organic-matter-normalized Freundlich sorption coefficients (KFOC) increased with increasing removal of OM (Fig. 3.11), suggesting that the extraction of OM layers improved access of OM to more favorable OM structures (i.e., polymethylene carbon). Simpson and Johnson (2006) similarly reported that humin fractions from soils sorbed considerable amounts of 1-naphthol. Solid-state 13 C NMR analysis revealed that six humin samples were enriched with amorphous, polymethylene-rich domains. The authors hypothesized that through fractionation, mobile domains in soil OM become more accessible for contaminant interactions and suggested that the physical conformation and availability of mobile polymethylene domains at the soil–water interface may determine the extent of their sorptive capacity rather than their presence alone. This suggestion is consistent with other studies, such as that of Kohl and Rice (1999), who reported increased contaminant sorption after the free lipid fraction was extracted from the soil. The free lipids were believed to block mobile domain sorption sites that would otherwise actively participate in the sorption of organic chemicals. Simpson et al. (2003) also reported an increase in phenanthrene sorption to HA after the selective removal of carbohydrates and aromatic components. Gunasekara et al. (2003) found that these chemical extractions resulted in a more

84

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

KFOC = 8,400 KFOC = 12,900 KFOC = 15,900 KFOC = 19,400

Increased phenanthrene sorption with increased removal of OM “coatings”

KFOC = 24,700 KFOC = 64,200

Figure 3.11. Solid-state 13C NMR spectra of sequential humic extractions from the same soil. The sequential removal of organic matter coatings resulted in increased sorption of phenanthrene. [Modified from Kang and Xing (2005).]

expanded OM conformation with partition-type domains. A decrease in isotherm nonlinearity was also observed, suggesting that the removal of aromatic structures “opened up” mobile, liquid-like regions and rendered these domains more accessible to contaminants (Gunasekara et al. 2003). A mass balance approach has also been used to test the role of OM conformation in contaminant sorption processes (Salloum et al. 2001; Bonin and Simpson 2007a). Phenanthrene Koc values for HA and humin were measured for four soil samples with varying degrees of OM content (Bonin and Simpson 2007a) (Table 3.1). The OM in the varying fractions was characterized using solid-state 13 C NMR spectroscopy. The Koc values for phenanthrene (Table 3.1) followed the same trend as observed by other researchers and showed that the humin fraction sorbed more than the HA and whole soil. The observed trends could not be attributed to OM structure alone. For example, an increase in aliphatic carbon content with some samples was not observed, yet phenanthrene Koc values still increased. The authors suggested that this increase in phenanthrene sorption was due to improved accessibility of more favorable sorption domains in soil humin that became available with OM fractionation. The DKoc [DKoc ¼ Koc (fraction) Koc (source)] was greatest for the low-OM-content soils (brown, dark brown, and black; Table 3.1), which also suggests that extraction improved access to OM domains in mineral soils. The authors proposed that if phenanthrene sorption was truly dependent on OM content alone, then the following mass balance equation would be valid: Koc,sample ¼ Koc,huminXhumin þ Koc,HAXHA þ Koc,FAXFA, where X represents the proportion of carbon in that fraction (HA, FA, or humin) versus the whole soil sample. The Koc of FA was predicted from

the Kow value because batch equilibration methods cannot be used to measure sorption to the FA fraction in this study. The “reconstituted” Koc values calculated from the mass balance equation were greater than the measured phenanthrene values for the whole soil (Table 3.1). This result suggests that accessibility to OM domains, rather than the presence of the OM domains, governs the extent of phenanthrene sorption. The authors further tested this by deashing humin samples (from the mineral soil samples) with hydrofluoric acid to determine if clay minerals were blocking OM sorption sites. A further increase in phenanthrene Koc values (Table 3.1) was observed, suggesting that OM– mineral associations are critical for evaluating the sorption of organic chemicals in soil and that minerals may indeed block phenanthrene access to sorption domains. Ahanger et al. (2008) also reported increased phenanthrene sorption after deashing of soils but did not observe the same trend with diuron. These results exemplify the continued need to examine the role of OM–mineral associations and their impact on sorptive behaviour with other, established trends such as the physiochemical properties of the organic contaminant and OM structure. For example, Joo et al. (2008b) tested a sorption model that included a parameter for OM blocking using experimental data for the sorption of 1,2,4trichlorobenzene, 1,4-dichlorobenzene, chlorobenzene, m-xylene, toluene, and benzene on sand and OM-coated sand. However, the results suggested that conventional sorption models that take into account both the mineral and OM phases were not suitable for predicting sorptive behavior. The OM blocking parameter did not improve model fitting; however, this could be due to the low OM contents of the sorbents (0.024%–0.154%) in combination with the relatively high solubility of the sorbates.

SUMMARY AND SYNTHESIS OF FUTURE RESEARCH DIRECTIONS

85

TABLE 3.1. Comparison of Measured Phenanthrene Kd and Koc Values with Reconstituted Phenanthrene Sorption Values

Sample Brown soil Brown humin Brown humic acid Brown humin (deashed) Dark brown soil Dark brown humin Dark brown humic acid Dark brown humin (deashed) Black soil Black humin Black humic acid Black humin (deashed) Peat soil Peat humin Peat humic acid

Reconstituted Koc, mL/g, with Fulvic Acid Fractione

OCa, %

Koc, mL/g (SE)b

D Koc, mL/gc

r2

Reconstituted Koc, mL/g, Without Fulvic Acid Fractiond

102  4 84.0  2 1523  100 4547  155

2.08 0.49 18.55 19.34

4,890  221 17,100  449 8,210  541 23,500  801

N/A 12,210 3,320 230,110

0.998 0.981 0.993 0.998

9750 — — —

13,160 — — —

182  9 134  5 2019  70

2.77 0.93 23.48

5,410  332 14,400  626 8,600  300

N/A 8,990 3,190

0.994 0.997 0.983

9180 — —

11,340 — —

6629  219

22.97

28,900  955

23,490

0.994

—

—

465  37 380  7 2338  147 12,605  131

5.26 0.83 25.37 23.54

6,550  708 45,800  895 9,220  581 53,700  558

N/A 39,250 2,670 47,150

0.965 0.964 0.984 0.991

21080 — — —

22,670 — — —

4224  70 7123  326 4860  245

48.35 46.84 55.36

8,740  145 15,200  697 8,780  443

N/A 6460 40

0.996 0.985 0.986

11,150

12,510

Kd (SE), mL/g

OC ¼ organic carbon content. Koc¼Kd/(OC%/100%). c DKoc ¼ Koc (fraction) Koc (source) from Salloum et al. (2001); N/A ¼ not applicable. d Koc;sample ¼ Koc;humin Xhumin þ Koc;HA XHA , where X represents the proportion of carbon in that fraction [humic acid (HA) or humin]. e Koc;sample ¼ Koc;humin Xhumin þ Koc;HA XHA þ Koc;FA XFA , where X represents the proportion of carbon in that fraction [humic acid (HA) or humin], fulvic acid (FA), Koc calculated from log Koc FA ¼ 0:989log Kow  0:346 [data from Karickhoff et al. (1979)] and log Kow ¼ 4.45 [data from Chen and Xing (2005)]. Source: Data from Bonin and Simpson (2007a). a b

3.4. SUMMARY AND SYNTHESIS OF FUTURE RESEARCH DIRECTIONS The progression of understanding sorption mechanisms has grown from correlations between sorption and organic matter content (Karickhoff et al. 1979) to relationships with specific organic matter structures and domains (Bucheli and Gustafsson 2000; Chen et al. 1996; Xing et al. 1994a,b; Grathwohl 1990; Ahmed et al. 2001; Huang and Weber 1997; LeBoeuf and Weber 2000) (Fig. 3.1). Organic matter fractions have been valuable when applying molecule-level techniques to study contaminant and OM interactions. The use of several molecule-level methods has facilitated the evolution of our understanding of organic contaminant interactions with OM and the indirect and direct roles of minerals in sorption processes. The continued use and future development of new analytical techniques will facilitate further advances in our understanding of contaminant sorption processes in soils. Two main types of experiments on the role of OM–mineral interactions on the sorption of organic chemicals exist: those that use constructed OM–mineral complexes and others that use OM fractions. Studies that examine contaminant interactions with constructed

OM–mineral complexes have shown that the mineral phase, which in comparison to soil OM, sorbs significantly less amounts of hydrophobic organic contaminants, plays an indirect role by governing organic matter accessibility at the soil–water interface (Murphy et al. 1994; Jones and Tiller 1999; Feng et al. 2005, 2006). Mass balance approaches also revealed that the “parts” (chemical fractions) sorbed more than the “whole” soil (Salloum et al. 2001; Bonin and Simpson 2007a). Sequential fractionation methods also support the mass balance trends (Kang and Xing 2005). These studies suggest that through chemical fractionation, more favorable or more sorption sites are made available, which results in higher Koc values. Also, 1H HRMAS NMR studies show that some OM structures are “buried” and may not be surface accessible at the soil–water interface (Simpson et al. 2001; Feng et al. 2005). Therefore, not all OM structures measured by various characterization techniques may participate in short-term surface interactions (fast sorption) but may be more important when delineating long-term interactions (slow sorption and diffusion). Furthermore, by investigating sorption phenomena through fractionation, a great deal of information has been gained on OM chemistry and the emergence of new hypotheses

86

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

regarding OM physical conformation. These results as well as the conclusions from numerous studies reviewed in this chapter clearly suggest that soil OM physical conformation in addition to soil OM chemistry governs sorption processes. The current understanding of OM–mineral interactions and their role in organic contaminant sorption is in its infancy and studies conducted only since the 1990s employ advanced molecule-level techniques. Although these types of studies are likely to continue and improve the fundamental understanding of OM–mineral associations, the greatest challenge lies in quantifying the role of OM physical conformation. Currently, researchers are using various techniques and isotherm models to quantify sorption data for both OM and organic contaminants. But there has been little in terms of quantitative comparisons with the current emphasis focusing more on corroborating trends. This is complicated by the chemical heterogeneity of OM and the use of different OM and mineral samples and varying solution conditions used by various researchers. Furthermore, the bulk of the corroborating results are for studies with phenanthrene—other organic chemicals did not show the same trends (Joo et al. 2008b; Ahanger et al. 2008; Bonin and Simpson 2007b). Consequently, there is a need to examine the role of OM–mineral complexes with varying organic chemicals. In addition, as this area of research continues to evolve, the research community will need to make attempts to quantitatively compare observations and develop predictive models. However, given the more recent nature of the majority of the studies reviewed here, it is clear that the role of OM–mineral complexes on organic contaminant sorption processes will continue and grow into a major area of research. ACKNOWLEDGMENTS The authors gratefully acknowledge support from the Natural Science and Engineering Council (NSERC) of Canada Strategic Grant and a University Faculty Award to MJS. REFERENCES Ahanger, A. G., Smernik, R. J., Kookana, R. S., and Chittleborough D. J. (2008), Chemosphere 72, 886–890. Ahmed, R., Kookana, R. S., Alston, A. M., and Skjemstad, J. O. (2001), The nature of soil organic matter affects sorption of pesticides, 1. Relationships with carbon chemistry as determined by 13 C CPMAS NMR spectroscopy, Environ. Sci. Technol. 35, 878–884. Arnarson, T. S. and Keil, R. G. (2000), Mechanisms of pore water organic matter adsorption to montmorillonite, Mar. Chem. 71, 309–320.

Avena, M. J. and Koopal, L. K. (1998), Desorption of humic acids from an iron oxide surface, Environ. Sci. Technol. 32, 2572–2577. Baham, J. and Sposito, G. (1994), Adsorption of dissolved organic carbon extracted from sewage sludge on montmorillonite and kaolinite in the presence of metal ions, J. Environ. Qual. 23, 147–153. Bailey, G. W. and White, J. L. (1970), Factors influencing the adsorption, desorption and movement of pesticides in soil, Residue Rev. 32, 29–92. Baldock, J. A. and Skjemstad, J. O. (2000), Role of the soil matrix and minerals in protecting natural organic materials against biological attack, Org. Geochem. 31, 697–710. Bonin, J. L. and Simpson, M. J. (2007a), Variation in phenanthrene sorption coefficients with soil organic matter fractionation: The result of structure or conformation? Environ. Sci. Technol. 41, 153–159. Bonin, J. L. and Simpson, M. J. (2007b), Sorption of steroid estrogens to soil and soil constituents in singleand multi-sorbate systems, Environ. Toxicol. Chem. 26, 2604–2610. Bucheli, T. D. and Gustafsson, O. (2000), Qualification of the sootwater distribution coefficient of PAHs provides mechanistic basis for enhanced sorption observations, Environ. Sci. Technol. 34, 5144–5151. Carmo, A. M., Hundal, L. S., and Thompson, M. L. (2000), Sorption of hydrophobic organic compounds by soil materials: Application of unit equivalent freundlich coefficients, Environ. Sci. Technol. 34, 4363–4369. Chefetz B., Deshmukh, A. P., Hatcher, P. G., and Guthrie E. A. (2000), Pyrene sorption by natural organic matter, Environ. Sci. Technol. 34, 2925–2930. Chen, Z., Xing, B., McGill, W. B., and Dudas, M. J. (1996), a-Naphthol sorption as regulated by structure and composition of organic substances in soils and sediments, Can. J. Soil Sci. 76, 513–522. Chen, B. and Xing, B. (2005), Sorption and conformational characteristics of reconstituted plant cuticular waxes on montmorillonite, Environ. Sci. Technol. 39, 8315–8323. Chenu, C. and Plante, A. F. (2006), Clay-sized organo-mineral complexes in a cultivation chronosequence: Revisiting the concept of the ‘primary organic-mineral complex’, Eur. J. Soil Sci. 57, 596–607. Chin, Y. P., Aiken G. R., and Danielsen, K. M. (1997), Binding of pyrene to aquatic and commercial humic substances: The role of molecular weight and aromaticity, Environ. Sci. Technol. 31, 1630–1635. Chiou, C. T. (2002), Partition and Adsorption of Organic Contaminants in Environmental Systems, Wiley, Hoboken, NJ. Chiou, C. T., Peters, L. J., and Freed, V. H. (1979), A physical concept of soil-water Equilibriums for nonionic organic compounds, Science 206, 831–832. Chiou, C. T., Schmedding, D. W., and Manes, M. (1982), Partitioning of organic compounds in octanol-water systems, Environ. Sci. Technol. 16, 4–10.

REFERENCES

Chiou, C. T., Porter, P. E., and Schmedding, D. W. (1983), Partition equilibria of nonionic organic compounds between soil organic matter and water, Environ. Sci. Technol. 17, 227–231. Chiou, C. T., Shoup, T. D., and Porter, P. E. (1985), Mechanistic roles of soil humus and minerals in the sorption of nonionic organic compounds from aqueous and organic solutions, Org. Geochem. 8, 9–14. Chorover, J. and Amistadi, M. K. (2001), Reaction of forest floor organic matter at goethite, birnessite and smectite surfaces, Geochim. Cosmochim. Acta 65, 95–109. Collins, M. J., Bishop, A. N., and Farrimond, P. (1995), Sorption by mineral surfaces: Rebirth of the classical condensation pathway for kerogen formation? Geochim. Cosmochim. Acta 59, 2387–2391. Doherty, P. and Warren, G. (1969), The adsorption of four herbicides by different types of organic matter and a bentonite clay, Weed Res. 9, 20–26. Engebretson, R. R. and von Wandruszka, R. (1994), Micro-organization in dissolved humic acids, Environ. Sci. Technol. 28, 1934–1941. Fein J. B., Boily J. F., Guclu, K., and Kaulbach, E. (1999), Experimental study of humic acid adsorption onto bacteria and Al-oxide mineral surfaces, Chem. Geol. 162, 33–45. Feng, X., Simpson, A. J., and Simpson, M. J. (2005), Chemical and mineralogical controls on humic acid sorption to clay mineral surfaces, Org. Geochem. 36, 1553–1566. Feng, X., Simpson, A. J., and Simpson, M. J. (2006), Investigating the role of mineral-bound humic acid in phenanthrene sorption, Environ, Sci. Technol. 40, 3260–3266. Garbarini, D. R. and Lion, L. W. (1986), Influence of the nature of soil organics on the sorption of toluene and trichloroethylene, Environ. Sci. Technol. 20, 1263–1269. Gauthier, T. D., Seitz, W. R., and Grant, C. L. (1987), Effects of structural and compositional variations of dissolved humic materials on pyrene Koc values, Environ. Sci. Technol. 21, 243–248. Grathwohl, P. (1990), Influence of organic matter from soils and sediments from various origins on the sorption of some chlorinated aliphatic hydrocarbons: Implications on Koc correlations, Environ. Sci. Technol. 24, 1687–1692. Gu, B., Mehlhorn, T. L., Liang, L., and McCarthy, J. F. (1996), Competitive adsorption, displacement, and transport of organic matter on iron oxide: I. Competitive adsorption, Geochim. Cosmochim. Acta 60, 1943–1950. Gunasekara, A. S., Simpson, M. J., and Xing, B. (2003), Identification and characterization of sorption domains in soil organic matter using structurally modified humic acids, Environ. Sci. Technol. 37, 852–858. Hance, R. (1965), Observations on the relationship between the adsorption of diuron and the nature of the adsorbent, Weed Res. 5, 108–114. Hance, R. J. (1969), An empirical relationship between chemical structure and the sorption of some herbicides by soil, J. Agric. Food Chem. 17, 667–668.

87

Hayes, M. H. B. (1970), Adsorption of triazine herbicides on soil organic matter, including a short review on soil organic matter chemistry, Residue Rev. 32, 131–174. Huang, W. and Weber, W. J. (1997), 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. 31, 2562–2569. Hur, J. and Schlautman, M. A. (2003), Molecular weight fractionation of humic substances by adsorption onto minerals, J. Colloid Interface Sci. 264, 313–321. Jones, K. D. and Tiller, C. L. (1999), 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. 33, 580–587. Joo, J. C., Shackelford, C. D., and Reardon, K. F. (2008a), Association of humic acid with metal (hydr)oxide-coated sands at solid-water interfaces, J. Colloid Interface Sci. 317, 424–433. Joo, J. C., Shackelford, C. D., and Reardon, K. F. (2008b), Sorption of non polar neutral organic compounds to humic acid-coated sands: Contributions of organic and mineral components, Chemosphere 70, 1290–1297. Kang, S. and Xing, B. (2005), Phenanthrene sorption to sequentially extracted soil humic acids and humins, Environ. Sci. Technol. 39, 134–140. Kang, S. and Xing, B. (2008), Humic acid fractionation upon sequential adsorption onto goethite, Langmuir 24, 2525–2531. Karickhoff, S., Brown, D., and Scott, T. (1979), Sorption of hydrophobic pollutants on natural sediments, Water Resour. Res. 13, 241–248. Kleber, M., Sollins, P., and Sutton, R. (2007), A conceptual model of organo-mineral interactions in soils: Self-assembly of organic molecular fragments into zonal structures on mineral surfaces, Biogeochemistry 85, 9–24. K€ ogel-Knabner, I., Guggenberger, G., Klber, M., Kandeler, E., Kalbitz, K., Scheu, S., Eusterhues, K., and Leinweber, P. (2008), Organo-mineral associations in temperate soils: Integrating biology, mineralogy, and organic matter chemistry, J. Plant Nutr. Soil Sci. 171, 61–82. Kohl, S. D. and Rice, J. A. (1999), Contribution of lipids to the nonlinear sorption of polycyclic aromatic hydrocarbons to soil organic matter, Org. Geochem. 30, 929–936. Kopinke, F. D., Georgi, A., and Mackenzie. K. (2001), Sorption of pyrene to dissolved humic substances and related model polymers. 1. Structure-property correlation, Environ. Sci. Technol. 35, 2536–2542. Lambert, S. M. (1967), Functional relationship between sorption in soil and chemical structure, J. Agric. Food Chem. 15, 572–576. Lambert, S. M., Porter, P. E., and Schieferstein, R. H. (1965), Movement and sorption of chemicals applied to soil, Weeds 13, 185–190. Laor, Y., Farmer, W. J., Aochi, Y., and Storm, P. F. (1998), Phenanthrene binding and sorption to dissolved and to mineralassociated humic acid, Water Resour. Res. 32, 1923–1931.

88

THE ROLE OF ORGANIC MATTER–MINERAL INTERACTIONS IN THE SORPTION OF ORGANIC CONTAMINANTS

LeBoeuf, E. J. and Weber, W. J. (2000), Macromolecular characteristics of natural organic matter. 2. Sorption and desorption behavior, Environ. Sci. Technol. 34, 3632–3640. Mao, J. D., Hundal, L. S., Thompson, M. L., and Schmidt-Rohr, K. (2002), Correlation of poly(methylene)-rich amorphous aliphatic domains in humic substances with sorption of a non polar organic contaminant, phenanthrene, Environ. Sci. Technol. 36, 929–936. Meier, M., Namjesnik-Dejanovic, K., Maurice, P. A., Chin, Y. P., and Aiken, G. R. (1999), Fractionation of aquatic natural organic matter upon sorption to goethite and kaolinite, Chem. Geol. 157, 275–284. Mikutta, C., Lang, F., and Kaupenjohann, M. (2004), Soil organic matter clogs mineral pores: Evidence from 1H-NMR and N2 adsorption, Soil Sci. Soc. Am. J. 68, 1853–1862. Mikutta, R., Mikutta, C., Kalbitz, K., Scheel, T., Kaiser, K., and Jahn, R. (2007), Biodegradation of forest floor organic matter bound to minerals via different binding mechanisms, Geochim. Cosmochim. Acta 71, 2569–2590. Mingelgrin, U. and Gerstl, Z. (1983), Reevaluation of partitioning as a mechanism of nonionic chemicals adsorption in soils, J. Environ. Qual. 12, 1–11. Murphy, E. M., Zachara, J. M., Smith, S. C., Phillips, J. L., and Wietsma. T. W. (1994), Interaction of hydrophobic organic compounds with mineral-bound humic substances, Environ. Sci. Technol. 28, 1291–1299. Namjesnik-Dejanovic, K., Maurice, P. A., Aiken, G. R., Cabaniss, S., Chin, Y. P., and Pullin, M. J. (2000), Adsorption and fractionation of a muck fulvic acid on kaolinite and goethite at pH 3. 7, 6, and 8, Soil Sci. 165, 545–559. Nearpass, D. C. (1976), Absorption of picloram by humic acids and humin, Soil Sci. 121, 272–277. Ochs, M., Cosovic, B., and Stumm, W. (1994), Coordinative and hydrophobic interaction of humic substances with hydrophilic Al2O3 and hydrophobic mercury surfaces, Geochim. Cosmochim. Acta 58, 639–650. Perminova, I. V., Grechishcheva, N. Y., and Petrosyan, V. S. (1999), Relationships between structure and binding affinity of humic substances for polycyclic aromatic hydrocarbons: Relevance of molecular descriptors, Environ. Sci. Technol. 33, 3781–3787. Reiller, P., Amekraz, B., and Moulin, C. (2006), Sorption of Aldrich humic acid onto hematite: Insight into fractionation phenomena by electrospray ionization with quadrupole time-of-flight mass spectrometry, Environ. Sci. Technol. 40, 2235–2241. Salloum, M. J., Dudas, M. J., McGill, W. B., and Murphy, S. M. (2000), Surfactant sorption to soil and geologic samples with varying mineralogical and chemical properties, Environ. Toxicol. Chem. 19, 2436–2442. Salloum, M. J., Dudas, M. J., and McGill, W. B. (2001), Variation of 1-naphthol sorption with organic matter fractionation: The role of physical conformation, Org. Geochem. 32, 709–719. Salloum, M. J., Chefetz, B., and Hatcher, P. G. (2002), Phenanthrene sorption to aliphatic-rich natural organic matter, Environ. Sci. Technol. 36, 1953–1958.

Satterberg, J., Arnarson, T. S., and Lessard, E. J. (2003), Sorption of organic matter from four phytoplankton species to montmorillonite, chlorite and kaolinite in seawater, Mar. Chem. 81, 11–18. Schlautman, M. A. and Morgan, J. J. (1993), Effects of aqueous chemistry on the binding of polycyclic aromatic hydrocarbons by dissolved humic materials, Environ. Sci. Technol. 27, 961–969. Simpson, A. J., Kingery, W. L., Shaw, D. R., Spraul, M., Humpfer, E., and Dvortsak, P. (2001), The application of 1H HR-MAS NMR spectroscopy for the study of structures and associations of organic components at the solid-aqueous interface of a whole soil, Environ. Sci. Technol. 35, 3321–3325. Simpson, M. J., and Johnson, P. S. C. (2006), Identification of mobile and highly sorptive domains in soil humin by solid-state 13 C nuclear magnetic resonance, Environ. Toxicol. Chem. 25, 52–57. Simpson, M. J., Chefetz, B., and Hatcher, P. G. (2003), Phenanthrene sorption to structurally modified humic acids, J. Environ. Qual. 32, 1750–1758. Simpson, A. J., Simpson, M. J., Kingery, W. L., Lefebvre, B. A., Moser, A., Williams, A. J., Kvasha, M., and Kelleher, B. P. (2006), The application of 1H high resolution magic-angle spinning NMR for the study of clay-organic associations in natural and synthetic complexes, Langmuir 22, 4498–4503. Theng, B. K. G. (1982), Clay-polymer interactions; summary and perspectives, Clays Clay Miner. 30, 1–10. Tombacz, E., Libor, Z., Illes, E., Majzik, A., and Klumpp, E. (2004), The role of reactive surface sites and complexation by humic acids in the interaction of clay mineral and iron oxide particles, Org. Geochem. 35, 257–267. Van Emmerik, T. J., Angove, M. J., Johnson, B. B., and Wells, J. D. (2007), Sorption of chlorphyrifos to selected minerals and the effect of humic acid, J. Agric. Food Chem. 55, 7527–7533. Vermeer, A. W. P. and Koopal, L. K. (1998), Adsorption of humic acids to mineral particles. 2. Polydispersity effects with polyelectrolyte adsorption, Langmuir 14, 4210–4216. Vermeer, A. W. P., van Riemsdijk, W. H., and Koopal, L. K. (1998), Adsorption of humic acid to mineral particles. 1. Specific and electrostatic interactions, Langmuir 14, 2810–2819. Wagai, R., Mayer, L. M., and, Kitayama, K. (2009), Extent and nature of organic coverage of soil mienral surfaces assessed by gas sorption approach, Geoderma 149, 152–160. Wang, K. and Xing, B. (2005), Structural and sorption characteristics of adsorbed humic acid on clay minerals, J. Environ. Qual. 34, 342–349. Wattel-Koekkoek, E. J. W., van Genuchten, P. P. L., Buurman, P., and van Lagen, B. (2001), Amount and composition of clayassociated soil organic matter in a range of kaolinitic and smectitic soils, Geoderma 99, 27–49. Wershaw, R., Llaguno, E. C., and Leenheer, J. A. (1996a), Mechanism of formation of humus coatings on mineral surfaces 3. Composition of adsorbed organic acids from compost leachate on alumina by solid-state 13 C NMR, Colloid Surf. A 108, 213–223.

REFERENCES

Wershaw R., Llaguno, E. C., Leenheer, J. A., Sperline, R. P., and Song, Y. (1996b), Mechanism of formation of humus coatings on mineral surfaces 2. Attenuated total reflectance spectra of hydrophobic and hydrophilic fractions of organic acids from compost leachate on alumina, Colloid Surf. A 108, 199–211. Xing, B., McGill, W. B., and Dudas, M. J. (1994a), Cross-correlation of polarity curves to predict partition coefficients of nonionic organic contaminants, Environ. Sci. Technol. 28, 1929–1933.

89

Xing, B., McGill, W. B., and Dudas, M. J. (1994b), Sorption of a-naphthol onto organic sorbents varying in polarity and aromaticity, Chemosphere 28, 145–153. Xing, B. (2001), Sorption of naphthalene and phenanthrene by soil humic acids, Environ. Pollut. 111, 303–309. Zhou, J. L., Rowland, S., and Mantoura, R. F. C. (1994), The formation of humic coatings on mineral particles under simulated estuarine conditions-a mechanistic study, Water Resour. Res. 28, 571–579.

4 PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES CHUNCHENG CHEN, ZHAOHUI WANG, WANHONG MA, HONGWEI JI, 4.1. Introduction 4.2. Important Concepts and Processes of the Environmental Photochemistry of Organic Contaminants 4.2.1. Some Important Laws in Photochemistry 4.2.2. Absorption of Light in Environmental Systems 4.2.3. Basic Photophysical and Photochemical Processes 4.3. Photooxidation of Organic Compounds on Iron-Bearing Minerals 4.3.1. Iron Oxides 4.3.2. Iron-Bearing Layer Silicates 4.4. Photosensitized Degradation of Organic Pollutants on Titanium Dioxide Surface Under Visible Irradiation 4.4.1. Dye Adsorption 4.4.2. General Performance for Photosensitized Degradation of Dye Pollutants 4.5. Conclusions

4.1. INTRODUCTION Because of their worldwide synthesis and application during the last few decades, anthropogenic organic pollutants have been released into the environment in a dramatically increasing rate by various pathways, such as direct or indirect anthropogenic discharge of noxious substances by municipal, industrial, and agricultural sources. The ever-proven and potentially detrimental effects of these compounds on the ecosystems and human health have prompted the development of relevant science and technologies centered mainly on three fields: (1) to clarify the pathways of the transformation and fates of pollutants in the environment, (2) to understand their environmental effect on the biosphere and

AND JINCAI

ZHAO

other environmental media, and (3) to develop efficient methods for their removal before their discharge and for the remediation of the polluted environmental media. Biological and (photo)chemical processes are expected to play vital roles in all these fields. The Natural organic compounds, such as plant and animal matter and other substances originating from living organisms, can be degraded easily by the microorganisms in the environment. However, many of the emerging anthropogenic organic compounds, such as hydrocarbons from oil, polyaromatic hydrocarbons (PAHs), poly(chlorinated biphenyls) (PCBs), poly(brominated diphenyl ethers) (PBDEs), and dioxins, are biologically refractory, and are generally difficult or very time-consuming to biodegrade. For these compounds, photochemical reactions are able to play an important role in their overall environmental transformation and fate. In the environment, significant fractions of hydrophobic pollutants are physically sorbed to the surface of environmental particles, and polar organic pollutants tend to be adsorbed chemically by electrostatic or complex interaction (Huang 2004; Matsuzawa et al. 2001; Reyes et al. 2000; Shichi and Takagi 2000). Many of the mineral particles, such as manganese or iron (hydro)oxides, and titanium oxides are photocatalytically active (Andreozzi et al. 2003; Sherman 2005). The photoinduced electron and energy transfer between adsorbed pollutant molecules and mineral surfaces may markedly influence the rate and pathway of the pollutant transformation (Ahn et al. 2006; Mao and Thomas 1993). Therefore, the photocatalytic reaction on the surface of mineral is one of the most important processes for the transformation of these organic pollutants in the environment. Investigation of the interfacial reaction mechanisms is critical for a full understanding of the fate of organic contaminants in the natural systems. Compared to

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

91

92

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

homogeneous reactions, heterogeneous processes depend not only on the physical and chemical properties of organic substrates but also on the nature of the solid on which they are adsorbed. The complex topographies, ill-defined surface and electronic structures, and the impurity of the natural minerals complicate studies of the intrinsic properties of the mineral and its surface. Therefore, although the photoactive minerals are relatively abundant in soils and in the water-suspended or airborne particulate matters and the photocatalytic principles are well understood, the role of these minerals in nature is unresolved. The photochemical and photocatalytic reactions lead to the formation of byproducts that can exhibit toxicological properties that differ from those of the original compound. Also, the photochemical transformation of organic pollutants is accompanied by a change in the polarity or other physical and chemical changes in the precursor molecule, and greatly alters the mobility, bioavailability, and bioaccumulation of the organic compounds as well as the toxicity. Accordingly, photochemical reactions must be considered during assessment of the environmental effects of pollutants. The purpose of this chapter is to explore the photochemical interactions and reactions between organic pollutants and the surfaces of some photoactive mineral particles, since the environmental photochemistry of the homogenous solution and atmosphere has been well reviewed in a general way (Burrows et al. 2002; Wallington and Nielsen 2005; Zafiriou et al. 1984). The photochemical transformation, especially photodegradation assisted by the mineral surface, and mechanism are emphasized in an attempt to shed some light on the photochemical environmental fate of some organic pollutants and also to survey the potential for employing photocatalysis as an effective method for remediation of organic pollutants. Accordingly, this chapter is arranged as follows. Some basic photochemical concepts and principles that are of great importance for our discussion on the following subjects are introduced briefly. Then two selected specific examples are presented and discussed in some detail in Section 4.2, Followed by a description of the photochemical degradation of organic pollutants on iron bearing mineral particles in Section 4.3. In Section 4.4, the photocatalytic degradation of dye pollutants themselves or other coexisting organic pollutants assisted by a semiconductor (TiO2) are reviewed.

4.2. IMPORTANT CONCEPTS AND PROCESSES OF THE ENVIRONMENTAL PHOTOCHEMISTRY OF ORGANIC CONTAMINANTS 4.2.1. Some Important Laws in Photochemistry It is well known that light can be considered as a quantized electromagnetic wave, which means that light can be

absorbed or emitted in quantized energy packets when it interacts with matter. These packets are termed photons, and each photon has an energy of e ¼ hn ¼ hc/l (where h ¼ Planck’s constant, c ¼ the velocity of light, n ¼ the frequency of electromagnetic wave, l ¼ wavelength). The first law of photochemistry (the Grottus–Draper law) states that only light that is absorbed by the molecule can result in the photochemical reaction. This law imparts several restrictions on the photochemical reaction possible in the environment: 1. Only sunlight radiation reaching the Earth’s surface is available for inducing photochemical reactions. Because the stratospheric ozone layer screens out radiation with higher energy (<290 nm), only sunlight with wavelengths longer than this cutoff needs to be considered to assess environmental photochemical reactions. In addition, when using photochemical process as techniques for the removal of organic pollutants, the artificial light source can avoid this restriction. On the other hand, most of the electronic excitation of environmental species can be induced by light with wavelength < 800 nm. Thus, the wavelengths under study focus on the 290–800 nm range in environmental photochemistry. 2. Photochemical reactions may take place only in the photic zone, such as the surface layer of freshwaters, oceans, the atmosphere, and the very upper surface of soil. In particular, because of the poor light penetration of soil, photochemical reactions are restricted to the upper several micrometers of the soil surface (Ciani et al. 2005a,b). Nonetheless, photochemical loss of organic chemicals may be significant because of the diffusive transport of the organic compounds (especially those with high volatility) from the lower part of the soil to the surface. 3. Only species with absorption spectra overlapping the sunlight emission spectrum can undergo direct photochemical reaction, or participate indirectly in the environmental photochemical reactions by sensitization (see discussion below). The second low of photochemistry (the Stark–Einstein law) suggests that, for each photon of light absorbed by a chemical system, only one molecule is activated for subsequent reaction. The absorption ability for light of a species is defined by the Beer–Lambert law, which states that the fraction of incident radiation absorbed is a function of the thickness (pathlength) and the concentration of absorbing molecules in its path. These two laws enable the efficiency, or quantum yield, of the reaction to be calculated. The quantum yield can be defined as the ratio of the number of molecules undergoing photoreaction to the number of photons absorbed.

IMPORTANT CONCEPTS AND PROCESSES OF THE ENVIRONMENTAL PHOTOCHEMISTRY OF ORGANIC CONTAMINANTS

4.2.2. Absorption of Light in Environmental Systems Light absorption is inherently the key factor for accomplishing photoinduced reactions. In the environment, irradiated light can be absorbed by organic molecules, metal complexes, and semiconductors and other species. When these species are placed in the oscillating electromagnetic field exerted by radiation, their interactions will lead to a redistribution of energy between the absorbing species and radiation. The process can excite an electron from a lower energy level to a higher one when the energy difference in these two levels is equivalent to the energy of a photon of radiation. In this process, the molecule absorbs energy of a photon from the radiation; that is, the energy flows from the electromagnetic radiation to the absorbing species. The absorption of a photon by a molecule occurs at a very short timescale (usually femtoseconds) without any change of molecular geometry (the Franck–Condon principle). The absorption properties of these species can be investigated and interpreted from in terms of molecular orbital theory, ligand field theory, and band theory, respectively. 4.2.2.1. Absorption of Organic Compounds. According to molecular orbital theory, electrons in a molecule are located on a series of molecular orbitals with a certain electron density distribution in the molecular framework and over certain energy levels. Usually, the electron in the ground state of organic compounds is located at the s, p, or n orbital, and the orbitals to which the electron is excited usually are s  and p -antibonding orbitals (where the asterisk indicates excited state). Accordingly, the transition followed by the absorption can be classified into n ! p , p ! p , n ! s  , and s ! s . The most important excitations in the environmental photochemistry are those due to electronic transitions from a bonding p orbital or from a nonbonding n orbital into an antibonding p orbital (p ! p and n ! p transitions, respectively). Typical organic compounds that have p ! p adsorption are polyenes and cyclic conjugated compounds (aromatic compounds). The adsorption of linear conjugated p systems (polyenes) would be shifted to longer wavelengths as the length of the number of conjugated double bonds increases. Once a certain chain length is reached, the systems show absorption in the visible region and are therefore colored. The electronic spectra of cyclic conjugated p systems depend inherently on the number of p electrons. In addition, the substituents of carbon in the conjugated framework by heteroatoms (e.g., N, O, S, and P atoms) or the substituents of hydrogen of the conjugation system by functionalities can cause major changes in its UV–visible absorption. This is important because synthetic dye pollutants usually contain these heteroatoms and carry various substituents. The typical organic compounds that can be excited via n ! p transitions are those that contain carbonyl and nitrogen groups. These groups carry lone pairs of electrons.

93

Sometimes these lone pairs of electrons are of p symmetry and form a part of the conjugated system, as in aniline, but more commonly their symmetry is s. Molecules with this type of heteroatom or substituent may therefore exhibit n ! p transitions from a doubly occupied n orbital (centered on the substituted atom) into a p orbital (derived from the p atomic orbital of heteroatoms and carbons). The most important classes of compounds with n ! p transitions are carbonyl compounds, azo compounds, and nitrogen heterocycles, all of which are universal in the nature and synthetic compounds. 4.2.2.2. Absorption of Organic Metal Complexes. In the environment, the existence of metal complexes formed between the organic ligands and the transition metal ions is rather ubiquitous, and these metal complexes are widely used as dyes and photosensitizers. Compared to the pure organic molecules, these compounds tend to exhibit some different absorption characteristics due to interaction between the ligands and central metals. In metal complexes, the ligands, metal, or both can contribute to their absorption. The involved molecular orbital (MO) that can donate or accept electrons for transition in most octahedral transition metal complexes is shown schematically in Figure 4.1. A MO can be conveniently classified according to its predominant atomic orbital contributions, as (1) predominantly ligandcentered strong bonding orbital (s L), (2) predominantly ligand-centered bonding orbital (pL), (3) metal-centered nonbonding orbital of t2g symmetry (pM), (4) predominantly metal-centered antibonding orbital of eg symmetry (sM ), (5) predominantly ligand-centered antibonding orbital (pL ), and (6) predominantly metal-centered strong antibonding orbital (s M ). In the ground-state electronic configuration of an octahedral complex of a d n metal ion, binding orbitals (types 1 and 2) are completely filled, while n electrons reside in the metal-centered nonbonding and antibonding orbitals (types 3 and 4) (Wallington and Nielsen 2005). According to the localized molecular orbital configurations, the electron transition band can be classified as a metalcentered (MC), ligand-centered (LC), or charge transfer (CT) transition. Among these transitions, a ligand-to-metal charge transfer (LMCT) mechanism is environmentally important for photolysis of naturally occurring metal complexes such as the Fe(III)-oxalate complex (see Section 4.3.1.3.). 4.2.2.3. Absorption of Semiconductor Materials. The absorption of irradiation by synthetic or natural semiconductor materials is another important transition in initiating the photochemical or photocatalytic reaction of environmental interest. The electronic properties of semiconductors are usually described in terms of the band theory of solids. In solids, bands result from the overlap of the orbitals of the atoms, which consist of the solid, just as molecular orbitals originate from overlap of atomic orbitals in small

94

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

σ M*

p

π L*

s

π*

C

σ M*

D

πM A

d

π

B

πL

σ

p

σL

AO from metal

MO of complex

A : Metal centered transition (MC)

AO from ligand

B : Ligand-to-metal charge transfer (LMCT)

C : Metal-to-ligand charge transfer (MLCT) D : Ligand centered transition (LC)

Figure 4.1. Molecular orbital diagram for an octahedral complex of transition metal; the dashed arrows indicate the four types of transitions based on localized MO configurations [adapted from Balzani et al. (2007)].

molecules. The general idea of how bands arise is illustrated in Figure 4.2. As isolated atoms, which are characterized by filled and unoccupied molecular orbital (HOMO and LUMO, respectively), are assembled into a crystal lattice, new molecular orbitals form. These orbitals are so closely spaced that they fall in essentially continuous bands. The filled bonding orbitals form the valence band (VB), and the unoccupied antibonding orbitals form the conduction band (CB). For the semiconductor, these bands are separated by a forbidden region or bandgap with moderate energy with of Eg (the energy difference between the bottom of the CB and the top of the VB).

N=1 (Isolated atom)

N=2

N=10

(Molecule) (Cluster)

N=2000

N >> 2000

(Quantum dot)

(bulk)

LUMO CB Energy

Band gap

HOMO

The electrons in the filled VB can be promoted to the CB on excitation with photons carrying energy greater than Eg. Such a process yields CB electrons (eCB) and VB holes (hVB þ ) in equal amounts. Accordingly, the absorption of a semiconductor starts at the point where the photon energy is equal to that of the bandgap, and after that, rises steeply with increasing photon energies. This excitation process is known as band-to-band transition, and usually provides the most effective means of adsorption for the semiconductor. 4.2.2.4. Absorption of Other Species in the Environment 4.2.2.4.1. Color Centers. Most metal oxides (known as insulators) have very wide bandgaps. Their intrinsic band–band transition requires irradiation with far-UV light, and it is impossible for them to be excited by the incoming solar light. In many cases, however, the light can be absorbed by some intrinsic surface defects (Volodin 2000). Typically, primary light absorption in such systems is related to the O ! M charge transfer of an M¼O bond:

M

VB

Figure 4.2. A schematic representation of the formation of an energy band during the assembly of atoms into a crystal lattice, assuming that one atom contributes one orbital; N denotes the number of atoms (adapted from Hoffmann et al. (1995)].

O

hυ

δ−

M

O

δ+

ð4:1Þ

Because of the high energy necessary for the excitation of M¼O bonds, these systems exhibit absorption and photocatalytic activity only in the UV region. Many surface defects and coordinatively unsaturated structures can have primary light absorption in the visible region. It has been shown that irradiation of a wide-bandgap insulator material, (e.g., alkaline-earth metal oxides) with highly energetic radiation generates relatively stable surface

IMPORTANT CONCEPTS AND PROCESSES OF THE ENVIRONMENTAL PHOTOCHEMISTRY OF ORGANIC CONTAMINANTS

defects (the so-called electron F-type color centers and hole V-type color centers, corresponding to anion and cation vacancies, respectively), among which the F centers are more important for initiating the photochemical transformations. For example, on the loss of an O atom in a metal oxide, the 0 ~ 2 electrons that remain trapped in the oxygen vacancies can give rise to an F center (Serpone 2006). Many of the color centers are able to absorb visible light and initiate photochemical transformations of adsorbed molecules (Emeline et al. 1998a,b, 1999). 4.2.2.4.2. Charge Transfer Complex. Photoprocesses can be also induced from absorption of light by adsorption complexes formed on the surface of wide-bandgap oxides. These complexes are usually generated by the adsorption of donor molecules on strong surface acceptor sites, and the absorption of light by such complexes is attributed to the intermolecular electron transfer from donor to acceptor. This class of absorption can extend the photoresponse of the oxide in the visible or near-visible region (Volodin 2000). Frei and co-workers proposed that the hydrocarbon–O2 charge transfer complex generated inside the cavities of alkali or alkalineearth ion-exchanged zeolites can absorb the light in the region of red and near-IRlight [see Eq. (4.2)]. The highly polar charge transfer states from the hydrocarbon to the oxygen molecule, after stabilization by the large electrostatic field present in the zeolite channels, can lead to selective oxidation of olefins, alkyl-substituted benzenes and alkanes under visible light irradiation at room temperature (Blatter and Frei 1994; Blatter et al. 1998). Similar photoprocesses initiated by the visible light have been observed on Ba-exchanged zeolites of different structures (Myli et al. 1997). In close relation to this finding, Seo et al. (2005) revealed that polycyclic aromatic compounds also form visible CT complexes with dry TiO2: visible light

½RH--O2  Ð ½RH þ --O2  

ð4:2Þ

A common characteristic for absorption of the abovementioned direct charge transfer complex is the existence of an electron donor (reductant) and an acceptor (oxidant), which are electronically coupled by a close contact. This close contact is frequently facilitated by the electrostatic attraction within an ion pair. The formation of a charge transfer complex has been shown to constitute a new and very promising class of photochemical system under light illumination. 4.2.2.4.3. Metal-Metal Charge Transfer. When widebandgap oxides contain transition metal ions, which are good candidates for environmentally applicable photocatalysts, the oxides usually exhibit absorption bands of transition metal ions. However, when two transition metals with different donor and acceptor abilities are

95

codoped into these oxides, metal-to-metal charge transfer (MMCT) may occur under visible irradiation, which has been shown to be a new class of visible-light-absorbing chromophores for photochemical reactions (Nakamura and Frei 2006; Weare et al. 2008) The MMCT of interest for the photocatalytic reaction included Zr(IV)--O--Cu(I) (Lin and Frei 2005b), Ti(IV)--O--Cu(I), and Ti(IV)--O--Sn(II) (Lin and Frei 2005a), Ti(IV)--O--Fe(II) (Xie et al. 2008), and Ti(IV)-O--Ce(III) (Nakamura et al. 2007). The visible absorption of these systems originates from the charge transition from the reducing metal sites (donors) to the adjacent oxidizing sites (acceptors), where they exhibit strong electronic coupling. The use of oxobridged heterobimetallic assembly as absorbing chromophores is a promising strategy for developing the visible-light photocatalysts, not only because of its flexible design and control of oxidation/ reduction (redox) potential and absorption wavelength but also because of its high durability for complete inorganic molecule-based photocatalysis (Nakamura et al. 2007). 4.2.3. Basic Photophysical and Photochemical Processes The absorption of a single photon is the only initiation event in photochemistry. After that, most of the absorbed photon energy is released as heat and light by the photophysical processes in all the abovementioned absorbing systems. Only a very small fraction of excited species can undergo chemical reactions. Considering that the reader can find a detailed description about these basic photophysical and photochemical processes in textbooks on photochemistry (Coyle 1991; Klessinger and Michl 1995), we give only a very brief introduction to them, focusing mainly on these relative processes in the following sections, such as emission of fluorescence or phosphorescence, the intersystem crossing process. 4.2.3.1. Photophysical Processes. Photophysical processes do not result in a chemical transformation of the substrates. Normally, the first event after absorption is to reach the zerovibration electronically first excited state S1 by the vibrational relaxation and/or by internal conversion from one of the vibrational levels of a higher electronic state such as S2. Figure 4.3 is a schematic representation of subsequent processes after the zero-vibration first excited state (1 AB  ). The excited species (e.g., organic molecule, semiconductor, or metal complex) can return to its ground state AB by emission of fluorescence (EF), or it can reach the triplet state 3 AB by intersystem crossing (ISC), and after loss of excess vibrational energy it can return to the ground state AB by emission of phosphorescence (EP). Radiationless deactivation from 1 AB  to ground state can occur via internal conversion (IC). Radiationless deactivation from 3AB to ground states can occur by intersystem crossing (ISC). It should be noted that the abovementioned processes usually are followed by a vibrational relaxation to the corresponding zero-vibration

96

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

1

AB*

Photochemical process

Photophysical process EF

IC

C

ISC EP

AB

3

A++ B- (Heterolytic Cleavage)

AB+C* (Energy Transfer)

A•+B• (Homolytic Cleavage)

AB+•+ C-• or AB-• + C+• (Charge Transfer)

AB

ISC

AB++e- (Ionization) IC: Internal Conversion ISC: Intersystem Crossing EF: Emission of Fluorescence

ABC (Addition)

BA (Isomerization) AB-AB (Dimerization) Indirect Photoreaction

EP: Emission of Phosphorescence Direct Photoreaction

Figure 4.3. Primary photophysical and photochemical processes immediately after a species (AB) absorbs radiation to its first excited state (AB). The asterisk (‘ ’) indicates excited state. All physical and chemical processes are usually accompanied by vibrational relaxation, but are not labeled as such here.

states. Although the photophysical process cannot cause the degradation of the organic compounds, it provides many useful ways to observe and measure the excited-state species. For example, the most effective detection method of 1 O2 is to measure the phosphorescence at 1280 nm. 4.2.3.2. Primary Photochemical Process. Competing with the photophysical processes, the photochemical reaction may occur, in which new chemicals are formed. Only the first excited singlet states and the lowest triplet states have a chance of living long enough to participate in photochemical processes. The photochemical reaction can be divided into direct photolysis and indirect photolysis according to the species that absorb the light to initiate the reaction. Direct reaction occurs when the initially excited molecule undergoes a chemical reaction. In indirect photolysis a reaction is initiated through light absorption by some chromophores other than the substrate itself. In general, the photochemical reaction can be divided into two steps: 1. The first step includes photoinduced energy transfer, electron transfer, bond cleavage, and ionization. This step leads to the formation of highly reactive species (e.g., free radicals, singlet oxygen), which can initiate further reactions, which can be termed primary reactions. 2. The second step consists of secondary or “dark” reactions. In this process, a series of reactions (usually free-radical chain reactions) are involved, and these reactions end in the formation of stable products. It is worthwhile to bear in mind that primary reaction

products always undergo a rapid backward electron transfer. The formation of stable photoproducts depends on the competition between this backward electron transfer and secondary processes. A permanent chemical change takes place if these secondary processes are faster than the backward electron transfer. 4.2.3.2.1. Direct Photolysis. As mentioned above, the absorption of light by organic molecules usually is caused by the transition of the electron from the bonding orbital (such as p) or nonbinding orbital (such as n) to the antibonding orbital (such as p ). The result of this transition is that the molecule will become less stable, thus facilitating transformation into other compounds. From the energy perspective, absorption of photons at the UV–visible range increases the energy content of a molecule by 150–400 kJ/mol, which yields enough energy to produce homolytic or heterolytic breakages in the molecules. Depending on the properties of molecules in excited states (e.g., geometries, dipole moments, acidity and basicity), the excess energy can drive the molecule to undergo various chemical transformations, such as photodissociations (homolysis or heterolysis), photoionization, intramolecular rearrangement, isomerization, abstraction of a hydrogen atom, or dimerization (Fig. 4.3). All these reactions can be initiated for the singlet or triplet excited states of the molecule. The importance of each pathway in the photolysis of the organic pollutants depends on their capacity to absorb photons and to undergo chemical changes after light absorption (Edhlund et al. 2006). In the environment, the direct photodegradation of organic pollutants with short-wavelength UV radiation (e.g., most

IMPORTANT CONCEPTS AND PROCESSES OF THE ENVIRONMENTAL PHOTOCHEMISTRY OF ORGANIC CONTAMINANTS

pesticides) is expected to be, in general, of limited importance (Burrows et al. 2002). In the artificial system, the direct photolytic process has been employed for the degradation of organic contaminants in some studies (Legrini et al. 1993). 4.2.3.2.2. Indirect Photochemical Reactions. An indirect photochemical reaction occurs when energy or electron from the initially excited molecule (photosensitizer) is transferred to another molecule, and cause the latter to undergo a chemical reaction. Indirect processes are common in the natural environment and are especially important because they can alter molecules that resist photolysis or have poor absorption of sunlight (Zafiriou et al. 1984). The efficiency of these reactions depends on the number of reactive species produced by excitation of the chromophores as well as on the ability of these species reacting with the contaminant. Photosensitization can occur through either energy transfer or electron transfer pathways. The most widely studied and important energy transfer photosensitization pathway in environmental photochemistry is the activation of the triplet ground state 3 O2 to more reactive singlet oxygen (1 O2 ). Followed the adsorption of light by sensitizer, its triplet states may be generated by intersystem crossing. If the triplet sensitizer has enough energy and lifetime, it can interact with dissolved oxygen to form singlet oxygen. In natural waters, dissolved organic matter (i.e., humic substances or the synthetic or natural dyes) are the most dominant sensitizers to produce singlet oxygen (Boreen et al. 2004, 2005, 2008; Latch et al. 2003). The singlet oxygen, despite its rapid deactivation in water solution (due mostly to intersystem crossing to the triplet ground state), is able to degrade a variety of organic compounds. Photosensitization may also occur through an electron transfer pathway. In an excited state, a molecule can be a better oxidant or reductant than in its ground state. For instance, after highest occupied–lowest unoccupied molecular orbital (HOMO-LUMO) excitation, the half-occupied HOMO can readily accept another electron, while the single electron from the LUMO can readily be transferred to an acceptor, that is, the excited molecules become powerful electron donors or acceptors. The photoinduced electron transfer can lead to separation of the charge and can initiate a free-radical reaction. Therefore, it plays an important role in many fields such as photoelectric conversation and the transformation and removal of organic contaminants. 4.2.3.3. Introduction to Photocatalytic Reactions. Photocatalytic reaction is one kind of indirect photolysis. In such a process, a photocatalyst is introduced, which can accelerate the photochemical reaction, but remain chemically unchanged. The most interesting and widely studied photocatalytic systems are those using wideband semiconductors, particularly TiO2 as the photocatalyst, in which the

97

photochemical reaction takes place in the interfacial region between the semiconductor and the solution. According to the chromophores of light absorption in initiating the photocatalytic reaction, heterogeneous photocatalysis can be classified into the semiconductor-initiated photocatalysis and sensitizer-initiated photocatalysis. 4.2.3.3.1. Semiconductor-Initiated Photocatalysis. The initiation step in semiconductor-initiated photocatalytic reactions is absorption of photon with sufficient energy to induce band–band transition (Fig. 4.4a). After that, the photogenerated valence band hole (h þ ) and conduction band (electron) migrate to the semiconductor surface and are trapped in the subsurface and surface states of the semiconductor particle. Charge carrier trapping can suppress recombination and increase the lifetime of the separated electron and hole. The ability of a semiconductor to undergo photoinduced electron transfer to the adsorbed species on the surface adsorbate is thermodynamically governed by the band energy positions of the semiconductor and the redox potentials of the adsorbate. If a reduction of the species in the electrolyte is to be performed, the conduction band position of the semiconductor has to be positioned above the relevant redox level, and for the oxidation of adsorbate by holes, the potential level of the adsorbate must be above (more negative than) the valence band position of the semiconductor. The natural minerals with semiconductor properties usually are the metal oxide and metal sulfide minerals. Most metal oxide semiconductors have valence band edges 1–3 eV below the H2O oxidation potential, but the potentials of their conduction band edges are close to, or lower than, the H2O reduction potentials. Accordingly, these oxides are strong photooxidation catalysts in aqueous solutions, but are limited in their reducing power. For example, the valence band of the anatase TiO2 semiconductor has an oxidation potential of þ 2.7 V, which is a strong oxidizing agent, whereas its conduction band has a reduction potential of 0.5 V (a moderate reductant). For non-transition-metal sulfides, the valence band holes are less oxidative than the metal oxides, but their conduction band electrons are strongly reductive. Most transition metal sulfides, however, are characterized by small bandgaps (<1 eV), and the band edges are situated within or close to the stability potentials of H2O. Hence, both the oxidizing power of the valence band holes and the reducing power of the conduction band electrons are lower than those of non-transition-metal sulfides (Xu and Schoonen 2000). Under ambient conditions, molecular oxygen is usually present to act as the primary acceptor of photogenerated conduction band electrons, which leads to the formation of a superoxide radical anion. If oxidative organic pollutants (e.g., halogenated compounds) are present, direct electron transfer to an organic molecule is also possible and may result

98

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

in reduction of the compound (Sun et al. 2009). The holes can react with surface-adsorbed H2O to produce . OH radicals or directly oxidize the organic substrates into their radicals. The reaction of the valence band (VB) hole should accompany the conduction band (CB) electron transfer to maintain the electroneutrality of the catalyst particle. As a result of the subsequent free-radical reaction in the presence of organic pollutants, all elements in the pollutant are mineralized to inorganic species: carbon to CO2, hydrogen to H2O, halogens to halide ions, sulfurs to sulfates, and phosphorus to phosphates, respectively. 4.2.3.3.2. Photosensitizer-Initiated Photocatalysis. The surface-adsorbed photosensitizer, after being excited by visible light, can transfer an electron into the conduction band of the semiconductor, and thus initiate the photocatalytic reactions (Wu et al. 1999a, 2000) [Eqs. (4.3) and (4.4)]. In photosensitized electron injection, an electron in the highest occupied molecular orbital (HOMO) of the sensitizer is excited to its lowest unoccupied molecular orbital (LUMO) with visible light irradiation, and is subsequently injected into the conduction band of the semiconductor. To inject an electron to the conduction band of the semiconductor, the energy of the LUMO of the sensitizer must be higher than that of the conduction band edge of the semiconductor. Another electron transfer mechanism may operate when the organic compounds form electron transfer complexes with the surface Ti sites. In this case, the irradiation can directly promote one electron from the organic ligand to the conduction band of TiO2, and the excited states of the organic sensitizer will be bypassed [denoted by arrows in Fig. 4.4b and Eq. (4.5)]. The direct outcome of electron injection from the sensitizer into the conduction band is the formation of cationic sensitizer radicals (dye þ . ) and conduction band electron. The chemical behavior of CB electrons is expected to be similar to that generated in the semiconductor-initiated reaction; that is, CB electrons can be scavenged by O2 or by oxidative

O2

e

- CB

H2O2 e

Dye þ visible ! dye*

OH

-

O2

O2-

H2O2

e- CB e-

ð4:3Þ

Dye* þ TiO2 ! dye þ þ TiO2 ðeÞ

ð4:4Þ

½DyeTiO2  þ visible ! dye þ þ TiO2 ðeÞ

ð4:5Þ

.

.

TiO2 ðeÞ þ O2 ! TiO2 þ O2  .

.

2O2  þ 2H þ ! O2 þ H2 O2 .

O2  þ TiO2 ðeÞ þ 2H þ ! H2 O2 þ TiO2 .

H2 O2 þ TiO2 ðeÞ ! OH þ OH þ TiO2 .

H2O

Degrade O2-

OH Dye*

Dye

H2O2

+

O2

e- CB e-

OH Dye+

Dye*

Visible

UV OH OH-

h+ VB h+

(a)

Dye VB

Org

(b)

ð4:6Þ ð4:7Þ ð4:8Þ ð4:9Þ ð4:10Þ

H2 O2 þ TiðIVÞ þ hn ! decomposition of H2 O2 ð4:11Þ

Visible

Org+

.

O2  þ H þ ! OOH pKa ¼ 4:69

H2O

H2O O2-

organic pollutants (Chen et al. 2002a) [Eqs. (4.6)–(4.11)]. However, the dye þ . may react along different pathways. This radical cation can be destroyed by the free-radical reaction initiated from the electron injection, which provides an efficient method for degrading the dye pollutants (Fig. 4.4b). Alternatively, dye þ . can obtain an electron from the electron donor present in the solution (e.g., organic pollutants) to regenerate the sensitizer (Fig. 4.4c), as implemented in photosensitized electrochemical cells (the Gr€atzel cell). Realization of this pathway requires thermodynamically that the energy level of HOMO of dye be lower than the redox potential of the electron donor in the solution, and kinetically that the electron donors be available for the dye þ . in its lifespan. As a result, this process can be used to remove the organic pollutants by oxidation (to regenerate the dye þ . ). It is also noted that, although we have demonstrated semiconductor-mediated photosensitization by using the dye as sensitizer and under visible irradiation, this concept can also be applied to excitation of surface complexes by employing UVirradiation:

P+

Dye VB

P

(c)

Figure 4.4. Electron transfer processes at a semiconductor–electrolyte interface during photocatalysis: (a) semiconductor-initiated photocatalytic reaction; (b) sensitizer-initiated photocatalytic reaction for the degradation of sensitizer itself; (c) sensitizer-initiated photocatalytic reaction for the degradation of other pollutants (P).

PHOTOOXIDATION OF ORGANIC COMPOUNDS ON IRON-BEARING MINERALS

In the environment, both semiconductor and photosensitizer mechanisms can occur. Although the mechanisms of semiconductor-initiated reaction are well understood and semiconductor-type minerals are ubiquitous on the surface of the Earth, whether and to what extent the semiconductor pathway are involved in the environmental photochemistry of minerals are not conclusively resolved, partly because it is difficult to rule out completely the sensitizer mechanisms (e.g., LMCT; see discussion below). However, as an advanced oxidation technique, the photocatalysis initiated from the band–band absorption of semiconductor (particularly TiO2) have been extensively applied in the removal of the organic pollutants. Sensitization mechanism (especially those induced by visible light) is significantly relevant to the photochemical cleanup of organic pollutants in nature, since chromophoric dissolved organic matter (CDOM) and synthetic dyes are abundant in the environment. The following two sections present example applications of these two mechanisms by reviewing the most recent progress in two typical environmental systems.

4.3. PHOTOOXIDATION OF ORGANIC COMPOUNDS ON IRON-BEARING MINERALS Iron is the most abundant transition element in the Earth’s crust (5%–6%), which is originally released from magmatic rocks through aerobic weathering in both terrestrial and marine environments and reprecipitates as iron(III) (hydro)oxides or in clay minerals (Cornell and Schwertmann 2003). Iron oxides, hydroxides, and oxide-hydroxides compound are ubiquitous components of soils and sediments (Table 4.1), collectively referred to as iron oxides in this chapter. They play important roles in the fate and mobility of organic pollutants because of their wide occurrence, important redox capabilities, and relatively high photoreactivity. 4.3.1. Iron Oxides Monodentate/bidentate inner- or outer-sphere surface complexes of organic pollutants at iron oxide surfaces might be the precursors of the photoreaction steps that result in photooxidation of organic ligands, photodissolution of iron oxides,

99

or a combination of both. Since iron oxides such as hematite possess semiconducting properties, they can promote photodecomposition of various adsorbed organic pollutants through different reaction mechanisms: (1) direct oxidation of pollutants by the holes generated by band–band excitation of a semiconductor, (2) photodegradation through ligand-tometal charge transfer (LMCT) processes within the Fe(III)organic ligand surface complex, and (3) oxidation of adsorbed species by the . OH radical. The strongly oxidizing . OH radical may result from any of the following pathways: (1) photolysis of an iron–hydroxyl complex bound at the surface sites ( FeIIIHO þ hn ! FeII þ . OH), (2) scavenging holes by OH groups on the catalyst surface; (3) secondary reactions including capture of photogenerated electrons by adsorbed O2, or reaction of oxidized organic radicals from LMCT processes with O2 leading to formation of H2O2, which reacts further with adjacent FeII via a Fenton-type reaction: FeII þ H2O2 ! FeIII þ . OH þ OH. 4.3.1.1. Properties of Iron Oxides. Most stoichiometric iron oxides are n- or p-type semiconductors. The conduction band consists of an empty Fe3 þ 3d/4s orbital, and the valence band is composed of a full 2t2g Fe 3d ligand field orbital with some admixture from the oxygen antibonding 2p orbital (Cornell and Schwertmann 2003). A bandgap is defined as the energy difference between the highest occupied and lowest unoccupied orbital. For a specific iron oxide, it corresponds to an O(2p) ! Fe(3d) LMCT transition (Sherman 2005). Hematite (a-Fe2O3) is an n-type semiconductor with a 2.2-eV bandgap. However, the lowest-energy LMCT transition in a (FeO6)9 cluster for a-Fe2O3 is estimated to be 4.4 eV. In such a case, hematite should not efficiently utilize visible light. Sherman and Waite (1985) suggested that the visible-light adsorption of hematite results primarily from pair excitations (6 A1g þ 6 A1g ! 4 T 1g þ 4 T 1g , 2.6 eV) and Fe(3d) ! Fe(3d) ligand field transitions (2.81 eV) and is associated with the tail of a strong LMCT transition band. Magnetite can be either an n- and p-type semiconductor with a small bandgap (0.1 eV). The proximity effect of Fe3 þ and Fe2 þ in an edge-sharing octahedral will facilitate the emigration of holes on the octahedral sites, which accounts for its good conductivity (102–103 V1 cm1). Maghemite and wustite are n-type and p-type semiconductors, respectively.

TABLE 4.1. Iron Oxides and Their Semiconducting Properties Mineral

Formula

Eg, eV

Mineral

Formula

Eg, eV

Goethite Lepidocrocite Feroxyhyte Magnetite Wustite

a-FeOOH c-FeOOH d-FeOOH Fe3O4 FeO

2.10, 2.5 2.06, 2.4 1.94 0.1 2.3

Hematite Maghemite Akaganeite Ferrihydrite Bernalite

a-Fe2O3 c-Fe2O3 b-FeOOH Fe5HO84H2O Fe(OH)3

2.02, 2.2 2.03 2.12 — —

Sources: Data from Leland and Bard (1987) and Cornell and Schwertmann (2003).

100

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

Bandgaps for other iron oxides listed in Table 4.1 are estimated at 2.0 eV from photocurrent measurements (Leland and Bard 1987), modestly lower than those determined using oxygen K-edge spectra and optical absorption spectra (Sherman 2005). Photoexcitation of iron oxides by light of the appropriate wavelength leads to creation of transient valence band þ holes (hvb ) and conduction band electrons (e cb ) in semiconductors. Take a-Fe2O3 for example [Eq. (4.2)]. The conduction band electron e cb is often trapped by a surface Fe(III) lattice site, resulting in its reduction to Fe(II), while þ hvb is trapped at the surface to oxidize adsorbed electron donors, if present, such as organic acids, thereby avoiding electron–hole recombination at the oxide surface. 4.3.1.2. Photocatalytic Pollutants

Reactivity

Toward

Different

4.3.1.2.1. Halogenated Acids. Pehkonen et al. (1995) investigated the photochemical reactivity of a range of iron oxide polymorphs toward the oxidation of monohaloacetate. They found that surface area of these oxides is roughly parallel to the order of reactivity: am-Fe2O33H2O > cFe2O3 > c-FeOOH  a-Fe2O3  Feaerosol > a-FeOOH. The halogenated acetic acid acts as an electron donor, and the relative rates of photooxidation of these acids with ferrihydrite (am-Fe2O33H2O) follow the order FCH2COOH > ClCH2COOH > BrCH2COOH > ICH2 COOH, FCH2COOH > F2CHCOOH > F3CCOOH. In terms of the observed strong kinetic isotope effects, Pehkonen and colleagues proposed that the bulk iron oxides phase is the principal chromophore, namely, that iron oxides serve as a photocatalyst in the oxidation of halogenated acetic acids. Photogenerated surface-bound hydroxyl radicals abstract hydrogen atoms from mono- and disubstituted haloacetic acids to yield haloacetate radicals, which in turn produce the corresponding halide and glycolic acid. For fully halogenated haloacetic acids, they seem to be oxidized via a photo-Kolbe-type reaction to yield the corresponding halo acids (i.e., HF, HCl, HBr) and CO2. 4.3.1.2.2. Phenols and Substituted Phenols. Several attempts have been made to explain the observed photodegradation of phenols (Chatterjee et al. 1994), aminophenols (Andreozzi et al. 2003), methylphenols (Mazellier and Bolte 2000), and chlorophenols (Bandara et al. 2001) by a band model. Andreozzi et al. (2003) investigated the photooxidation of 2-aminophenol by goethite at different pH values (3.0–8.0) and catalyst loading (100–500 mg/L). A negligible mineralization in the presence of O2 has been observed during the oxidation process. They ruled out the involvement of hydroxyl radicals in oxidation of 2-aminophenol. Mazellier and Bolte (2000) studied the light-induced transformation of 2,6-

dimethylphenol (DMP) in the system of UV/goethite/O2 and identified 2,6-dimethylbenzoquinone and 4,40 dihydroxy-3,30 ,5,50 -tetramethylbiphenyl (dimmer of DMP) as main products. They concluded that positive holes, rather than hydroxyl radicals, are responsible for DMP degradation, whereas conduction band electrons are trapped by oxygen adsorbed at surfaces. Bandara et al. (2001) observed that mono-, di-, and trichlorophenols can be efficiently degraded but partially mineralized on a-Fe2O3, and that degradation follows with pseudo-first-order kinetics. The overall photocatalytic degradation increases in the order 2,4,6trichlorophenol (2,4,6-TCP) < 2,3-dichlorophenol (2,3DCP) < 2-chlorophenol (2-CP) < 2,4-DCP. However, a-FeOOH is found to be active only for 2,4dichlorophenol (2,4-DCP). Like the process with TiO2 photocatalyst, photodegradation of chlorophenols on a-Fe2O3 is initiated from the para-hydroxylation of the parent compounds as suggested by the identified intermediates. 4.3.1.2.3. Dyes. Despite intensive investigations, there is still controversy concerning the mechanisms of dye degradation on iron oxides under visible-light irradiation. For instance, Orange II (Org II), a common toxic azo dye, is thought to be degraded through photosensitized and semiconductor photocatalysis pathways. The inherent difference between the two pathways (as described above) should lie in their excited chromophores; the former are dyes or dye–Fe surface complexes, but bulk catalyst for the latter one. Bandara et al. (1999) presented a series of spectroscopic evidences to support their photosensitized degradation concept. Org II can form a bridged bidentate complex with a-Fe2O3 via its sulfonic group, especially at acidic pH values (Bandara et al. 1999a). The complex [Org II    a-Fe2O3] is the prerequisite for an efficient charge transfer process. The electron injection from excited dye to the conduction band of a-Fe2O3 proceeds within the duration of a laser pulse (<108 s). Sulfobenzoic acid, 1,2-naphthol, 1,2-naphthaquinone, and sulforanilic acid are the identified products of Org II (Bandara et al. 1999b). Du et al. (2008), however, argued that semiconductor photocatalysis should account for the degradation of Org II. Their indirect evidence included the observations that (1) H2O2 and AgNO3, as effective scavengers of e cb , can significantly enhance the rate of dye photodegradation; (2) a positive effect of fluoride on dye photodegradation has been observed since fluoride is capable of replacing surface hydroxyl groups, thus promoting oxidation of water to radicals by the valence band holes; and (3) the dye degradation is much slower under visible-light irradiation (l  450 nm) than under UV irradiation, but was also accelerated by the addition of NaF and AgNO3. Thus, they proposed that iron oxides possess visible-light photocatalytic activities that derive from

PHOTOOXIDATION OF ORGANIC COMPOUNDS ON IRON-BEARING MINERALS

A schematic representation of the various hypothetical steps involved in photodissolution of an iron oxide in the presence of a ligand such as oxalate is shown in Figure 4.5 (Siffert and Sulzberger 1991). An essential step is to form bidentate inner-sphere surface complexes, which can shift electron density toward the metal ion center and facilitate electron transfer to iron oxide. The absorption of a photon results in an electronically excited state. Intramolecular electron transfer of excited surface complexes leads to a reduced surface iron center and the oxidized oxalate radical. After dissociation from the surface, the oxalate radical undergoes a rapid decarboxylation reaction, yielding CO2 and a strong reducing CO2 .  radical. This CO2 .  radical can reduce a second ambient iron(III) in a thermal reaction. In aerated system, CO2 .  will react with dioxygen to produce superoxide radical (O2 .  ). Desorption of reduced Fe(II) ion is the rate-determining step and is assisted by protons and oxalate Photolysis of surface Fe(III) complexes appears to be dependent on a variety of factors, including the concentrations of the adsorbed organic ligands, structural nature, and photoactivity of the Fe(III)-ligand complex (Brown et al. 1999; Stumm and Sulzberger 1992). All of those are pH-relevant. Furthermore, Siffert and Sulzberger (1991) found that oxalate was oxidized 3–4 times more quickly in oxic compared to anoxic systems, indicating that O2, as an abundant electron acceptor in a natural system, is the predominant factor controlling the rates of oxidation of organic ligand.

separation of photogenerated holes/electrons. Nevertheless, Du et al. (2008) do not completely rule out the possibility of photosensitized reaction. 4.3.1.2.4. Polymers. Kuramoto (1996) reported that an epoxy-terminated disulfide oligomer [a,v-(2, 3epoxypropane)hexa [1,10 -methylenedioxybis (2-thioethane)] (EDS) can be partially photodegraded into disulfoxide and thiosulfone on a-FeOOH. The photogenerated holes or O ions at surface sites may lead to oxidation of sulfur atoms of EDS molecules. Although most iron oxides absorb light up to 600 nm, they exhibit relatively weak photocatalytic activities and poor photostabilities as compared to TiO2 (Mazellier and Bolte 2000). This semiconductor-type mechanism may become significant only in very small colloidal iron oxides particles, due to the short diffusion length of the photogenerated electron/hole pairs in iron oxide lattice (Kiwi and Gr€atzel 1987). Since iron oxides have more positive flat-band potentials (0.06–1.00 V vs. NHE at pH 6.5) (Leland and Bard 1987; Sherman 2005), they are photocatalytically reactive only in the presence of a strong reducing or complexing agent (Guillard et al. 1995). 4.3.1.3. LMCT Mechanism Involved in Photooxidation of Organics. As noted above, iron oxides are potential photochemical catalysts in environmental systems. However, their contribution to photooxidation is relatively small, at least compared to those resulting from LMCT process involving a surface metal ion and specifically absorbed ligand. In general, photooxidation of organic compounds by LMCT transition process is closely associated with the photodissolution of bulk iron oxides. Waite (2005), Hoffmann et al. (1995), and Stumm and Sulzberger (1992) have published a series of critical reviews about photoreductive dissolution of iron oxides in the presence of naturally occurring compounds. O

H O

O

4.3.1.4. Oxidation of Organic Compounds via LMCT Process. There have been intensive investigations of the photooxidation of naturally abundant organic ligands with iron oxides. Benzoate, succinate (Cunningham et al. 1988), citrate (Borer et al. 2007), formaldehyde, formate, acetate, butyrate (Pehkonen et al. 1993), malonate (Litter et al. 1994), H O

OH2 +

Fe III

+

O

Fe III

Fe III

O

Fe III

+ 2H2O

I HO

O H

O

Fe aq

IV

+ CO2- +

O

+ k-1

II

hv

k3 H O

H O CO2

O

O H

OH H

II

Fe III

Fe II

k2

Fe III

O

O

O

O

Fe II

III O H

101

O H

Figure 4.5. Schematic representation of the different steps involved in the photooxidation of oxalate adsorbed on an iron(III) oxide [modified from Siffert and Sulzberger (1991)].

102

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

fulvic acid, EDTA (Karametaxas et al. 1995), and glycolate (Cunningham et al. 1985) have thus far been selected as organic ligands. These ligands are transformed to corresponding organic radicals, which continue to be oxidized by reaction with O2 or other electron acceptors. Leland and Bard (1987) reported that the different iron oxides promote photodegradation of oxalate and sulfite at rates varying by about two orders of magnitude. Lepidocrocite (c-FeOOH) exhibits the best activity toward oxidation of both compounds among the polymorphs of FeOOH. In the case of oxalate oxidation, the order of relative photochemical reactivity is as follows: c-Fe2O3 > c-FeOOH > d-FeOOH > a-Fe2O3 > a-FeOOH > b-FeOOH. The authors attributed the rate differences to intrinsic diversities in crystal and surface structure rather than to the surface area or bandgap. This type of LMCT mechanism involving carboxylic groups has also been applied to explain photooxidation of sulfurbearing organic pollutants (Johansen and Key 2006). Methanesulfinic acid (MSIA, CH3SO2H) is a dimethylsulfide (DMS) oxidation intermediate. There exists structural similarity between carboxylic acid (--COOH) and the sulfinic acid (--SOOH) as center C and S atoms have almost identical electronegativities (2.55 and 2.58, respectively). Therefore, sulfinic group may be involved in the formation of a surface complex between Fe(III) and MSIA. Borer et al. (2007) reported the photodecomposition of citrate on c-FeOOH by in situ attenuated total reflection fourier transform infrared spectroscopy (ATR-FTIR). The primary photoproduct of citrate is acetonedicarboxylic acid generated from photodecarboxylation. The adsorbed acetonedicarboxylic acid is further decomposed to acetoacetate and even acetone as its final product at pH 4, but at pH 6, no further degradation of acetonedicarboxylic acid was observed. They attribute this type of pH-dependent selective photooxidation of the a-hydroxycarboxylic acid functional group of citrate to the different molar fractions of innersphere citrate surface complexes at pH 4 and pH 6 and their varied photochemical reactivities. Photolysis of the complexes of iron oxides with carboxylates on the addition or absence of H2O2 has been used to initiate the degradation of contaminant species (often via heterogeneous photo-Fenton processes). This method is particularly suitable for decomposition of some organic compounds that have low affinity to iron oxide such as diuron. Mazellier and Sulzberger (2001) observed the degradation of diuron in irradiated goethite/oxalate suspension at 3 pH 6. They attributed the degradation to attack by . OH radicals derived from photoreduced Fe(II) and H2O2. Concentrations of both oxalate and protons affected the rate of light-induced diuron transformation. The formation of Fe(II)(aq) is the rate-determining step for diuron degradation in these heterogeneous photo-Fenton systems. Liu et al. (2006), Li et al. (2007), and Lan et al. (2008) have studied a variety of synthetic iron oxides and organic

pollutants, including c-FeOOH, c-Fe2O3, a-Fe2O3, 2-mercaptobenzothiazole, bisphenol A, and pentachlorophenol (PCP). They reported an optimal initial concentration of oxalate (0.8 mmol L1) for degradation of PCP in UV/ c-Fe2O3/oxalate systems (Lan et al. 2008). Photoactive Fe(C2O4)2 and Fe(C2O4)33 are assigned to the dominant Fe(III)-oxalate species according to Fe(III) speciation calculation. Six intermediates are identified, including tetrachlorocatechol (TeCC), 2,3,5,6-tetrachloro-1,4-hydroquinone (TeCHQ), tetrachloro-o-benzoquinone (o-chloranil), and tetrachloro-p-benzoquinone (p-chloranil) formic and acetic acids. However, why higher concentrations of oxalate inhibited the rates of PCP degradation is still open to debate, although Li et al hypothesized that oxalate also react with hydroxyl radicals, thereby slowing down PCP transformation at higher oxalate concentration. Alternatively, there is another feasible explanation, namely, that Fe(III)-oxalate surface complexes convert from a bidentate, binuclear to a monodentate, binuclear structure at this threshold value of oxalate concentration (0.8 mmol/L) (Mazellier and Sulzberger 2001). He et al. (2002) observed that UV irradiation can significantly accelerate the degradation of an azo dye–Mardant Yellow 10 (MY10) on goethite. Photoreaction intermediates are confirmed, including acetic acid, nitrobenzene, and 4hydroxybenzenesulfonic acid by MS analysis and NO3 and SO42 by IC measurement. The authors proposed that photoinduced LMCT process of the surface complex of H2O2 with the oxide surface metal centers may be responsible for the MY10 degradation (Fig. 4.6). Briefly, the reaction is initiated by the formation of a precursor surface complex of H2O2 with surface Fe(III). It is followed by a cleavage of the O--O bond of the surface complex, leading to generation of a high-valence iron-oxo (ferryl) species and hydroxyl radical. This unstable ferryl species is able to react with H2O to give another . OH or directly oxidize adjacent pollutant molecules. They confirmed the . OH attack mechanism by an ESR trapping experiment. In a heterogeneous photo-Fenton system, the formation of a surface complex of Fe(III) with target pollutants, instead of H2O2, can also assist their decomposition and mineralization (He et al. 2005). Rates of degradation of aromatic compounds are related to these sorption capacities toward iron oxides: salicylic acid m-hydroxylbenzoic acid > p-hydroxylbenzoic acid benzoic acid > p-biphthalic acid > phenol > benzenesulfonic acid. 4.3.2. Iron-Bearing Layer Silicates Layer silicates are perhaps the most important and chemically active components of the clay mineral fraction. These natural minerals possess layered structures, large surface areas, and high cation exchange capacity (CEC). Smectites are hydrated 2 : 1-layer silicates composed of an (Fe, Al, Mg) octahedral sheet linked between two (Si, Al) tetrahedral

PHOTOSENSITIZED DEGRADATION OF ORGANIC POLLUTANTS ON TITANIUM DIOXIDE SURFACE UNDER VISIBLE IRRADIATION

FeIIIOOH UV H2O2

Degradation product

FeIIIOH

O=Fe IV Dye

HO

H2O

Figure 4.6. Cycling of Fe species coupled with dye degradation in the presence of hydrogen peroxide under UV light irradiation [stages: I—surface complex formation; II—light absorption; III— dissociation and decarboxylation of the oxidized oxalate; IV— detachment of surface Fe (II)] (He et al. 2002).

sheets by oxygen ligands (Thomas 2005). Most smectites contain appreciable amounts of octahedral iron in the ferric and/or ferrous form, and total iron content can be up to 25 wt% (Murad and Fischer 1988). On microbial or chemical reduction, Fe(II) complexes associated with clay minerals are of great significance for contaminant transformation. It is reported that the reduced smectites can act as a reducing agent for most of pesticides, chlorinated aliphatics, and nitroaromatics, eliminating either --Cl or --NO2. Stucki and his colleagues have examined the reactivities of reduced ferruginous smectites with respect to a range of pesticides, including atrazine, alachlor, trifluralin, oxamyl, chloropicrin, dicamba, and 2,4-D (Stucki 2005). These pesticides, with the exception of 2,4-D, can extensively react with clay surfaces. Their degradation products are observed but have not yet been fully identified. Dechlorination of chlorinated aliphatics (Lee and Batchelor 2004) and reduction of nitroaromatics (Hofstetter et al. 2003) have also been achieved by reaction with reduced smectites. Iron in clay minerals exists in very different chemical environments within the clay structure and at the mineral surface. Four different types of iron species may be present on clays (Hofstetter et al. 2003): (1) free iron oxides that distribute randomly on the clay surface, (2) structural iron that substitutes aluminum/silicon in the octahedral/tetrahedral lattice, (3) iron complexed by surface hydroxyl groups at edge surfaces, and (4) iron bound by ion exchange at basal siloxane surfaces. Free iron oxides in clays can be removed with citrate/bicarbonate/dithionite (CBD) extraction (Mehra and Jackson 1960). This CBD method makes it possible to compare the photoreactivities of different iron species

103

toward the same organic pollutant. Hofstetter et al. (2003) characterized the various Fe(II) species of reduced clays in terms of their accessibility to and reactivity for nitroaromatic compounds (NACs). Results reveal that interlayer exchanged Fe(II) does not contribute to the NAC reduction, but both edge-complexed and structural Fe(II) are effective in yielding the aniline product. However, on light irradiation, interlayer exchanged Fe(II) exhibit much higher reactivity for catalysis of the mineralization of malachite green (MG) than did structural Fe(II) (Cheng et al. 2008). Song et al. (2006) compared the photoreactivities of free oxides on clay surface and structural iron sandwiched in between two silica tetrahedral sheets. It was found that structural iron in the octahedral lattice inefficiently photocatalyze the decomposition of hydrogen peroxide under UV irradiation, but free oxides exhibit good reactivity. This difference results from the fact that direct photoexcitation cannot lead to the photoreduction of structural iron, which is unlike free oxides: .

 FeIII OH þ hn !  FeII þ OH

ð4:12Þ

When photoreactive compounds such as N,N-dimethylaniline (DMA), rhodamine B (RhB), or malachite green (MG) intercalate into the clay layer, structural iron is able to greatly promote the decomposition of H2O2. According to the low redox potential of excited states of these compounds (2.85, 1.09, and 1.08 V vs. NHE for DMA, RhB, and MG, respectively), they can donate electron to structural iron (0.44 V vs. NHE) under light irradiation. Organics without absorption above 300 nm (such as cetyltrimethylammonium bromide) do not accelerate the reduction of H2O2, although it also exhibits high affinity toward clay particles. Therefore, the reduction of clay iron(III) to iron(II) can be achieved either by a light-induced LMCT process (for iron oxides) or electron injection of organic matters on irradiation (for structural iron) (Fig. 4.7).

4.4. PHOTOSENSITIZED DEGRADATION OF ORGANIC POLLUTANTS ON TITANIUM DIOXIDE SURFACE UNDER VISIBLE IRRADIATION Textile dyes and other industrial dyestuffs constitute one of the largest groups of anthropogenic organic compounds that represent an increasing environmental risk. There are more than 100,000 commercially available dyes with over 7  105 tons of dyestuff produced annually (Zollinger 1987). It has been estimated that about 10%–20% of the total world production of dyes is released into the environment from textile, paper, printing industries, and dye houses (Claus et al. 2002). These dyes have not only caused aesthetic problems but also exhibited great biotoxicity and possible mutagenic and carcinogenic effects (Chang et al. 2001). Most

104

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

Cr(VI) Photoreactive species*

Structural

FeIII / FeIII oxides

hv Photoreactive

species

Photoreactive

species+

Cr(III) UV irradiation

Structural

FeII / Surface

OH

FeII

H2O2

Figure 4.7. Proposed route of H2O2 decomposition catalyzed by iron oxides and structural iron in the presence of photoreactive substances [adapted from Song et al. (2006)].

commercial dyes are designed to resist photodegradation, and few of them can be removed by the applied aerobic microbial process of wastewater (Brown et al. 1981; Meehan et al. 2000; Robinson et al. 2001). Other traditional processes such as adsorption, chlorination, ozonation, and flocculation have been proved insufficient for treatment of these effluents. Therefore, it is urgent and important to exploit efficient methods to remove these dye pollutants. In this context, the self-sensitized degradation of dye pollutants in the presence of TiO2 under visible-light irradiation provides an efficient method for removal of these pollutants by using the solar light. In addition, the presence of photosensitized processes on the surface of semiconductor minerals can induce the oxidative or reductive removal of other organic pollutants under visible light. In the environment, the photosensitization reaction is also one of the primary pathways for the transformation of natural and synthetic dyes in the mineral surface. Thus, a detailed understanding of the photosensitized mechanism is expected to present useful information on the environmental fate and effect of these dye pollutants. Accordingly, the main focus of this section is on the photodegradation of organic pollutants by dye-sensitized TiO2 photocatalysis under visible-light irradiation. The conventional TiO2 photocatalytic degradation of organic contaminants (including dye pollutants) under UV irradiation is not included here, since there have been many extensive and excellent reviews on it since the early 1990s (Chen and Mao 2007; Hoffmann et al. 1995; Legrini et al. 1993; Thompson and Yates 2006). 4.4.1. Dye Adsorption Adsorption of the dye to the semiconductor surface is a prerequisite for photosensitization degradation, since the rapid electron injection from the dye to TiO2 requires strong interaction between dye and surface of TiO2. The TiO2assisted photodegradation of dye pollutants is thus a typical interfacial reaction that occurs just on the surface of TiO2 particles, other than in the bulk solution. The interaction extent and mode of the dyes on the surface of TiO2 particles

are important factors that govern the degradation rate and degradation mechanism. The interaction extent can be reflected by an adsorption isotherm of the dye on the TiO2, which can be estimated by measuring the amount of equilibrium adsorption of the dyes at various concentrations on a certain amount of photocatalyst particles. As shown in Equation (4.13), adsorption data can usually be processed in the form of a plot of amount of adsorbed dye (mad) versus the equilibrium concentration (C) according to the Langmuir adsorption isotherm C C 1 ¼ þ max mad mmax K m ad ad ad

ð4:13Þ

where C is the concentration of dye in the supernatant liquid at equilibrium, mad is the amount of dye adsorbed per unit mass of photocatalyst, mmax is the maximum adsorbed ad quantity per unit mass of photocatalyst, and Kad is the association constant of dye on photocatalyst; mmax ad and Kad can be determined from the slope and intercept of the linearity fitting, respectively. The adsorption isotherm can bear information on the total and macroscopic interaction between the surface and the dyes. To understand the interaction between the dye and surface sites at a molecular level, one must consider the adsorption mode of dye on the specific surface sites of photocatalyst. It is more difficult to determine the adsorption modes, which always involved several combined surface-characteristic techniques, such as UV–visible, FT-IR, Raman, and XPS spectroscopies. Complex and electrostatic interactions are the two most important interaction modes between the dye and surface of a photocatalyst. Complex interaction is the strong chemically binding mode between the functional groups of dyes and the surface metal sites of photocatalyst. In many photosensitized systems, the photosensitizers are linked to the surface by forming chemical complex to achieve efficient electron injection. In the Gr€atzel cell, for example, the photosensitizer usually anchors on the surface of TiO2 via carboxyl or phosphatic groups. For self-sensitized degradation, the dyes with strong complex group such as alizarin red generally can

PHOTOSENSITIZED DEGRADATION OF ORGANIC POLLUTANTS ON TITANIUM DIOXIDE SURFACE UNDER VISIBLE IRRADIATION

undergo rapid degradation (Liu et al. 2000b). The surface oxygen sites of TiO2 in aqueous media can undergo a series of protonization and deprotonization (acid–base) equilibria. As a result, a point of zero (surface) charge (PZC) is present with the change of the solution pH, where the surface sites with positive charge are equal to those with negative charge. Accordingly, the electrostatic interaction between the dye and surface depends greatly on the pH, the properties of surface, and the dyes. The PZC of TiO2 particles is at about pH 6.8 (Zhao et al. 1993). In general, the anionic dyes with sulfonate and carboxyl group manifest strong adsorption on the metal oxide at low pH, due to electrostatic interactions between the positive TiO2 surface and dye anions. For example, Bourikas et al. (2005) found that adsorption of Acid Orange 7 (AO7) on the TiO2 surface occurs to a significant extent only at pH values lower than 7, via the sulfonic group of the azo dye. Another anion dye amaranth with two sulfonate groups has been reported to be adsorbed on TiO2 via the sulfonate group located in the ortho position with respect to the OH group at natural pH (5.7) (Karkmaz et al. 2004). Accordingly, the anionic dye can be more rapidly degraded at low pH, whereas the degradation of cationic dyes needs a pH larger than the PZC of the photocatalyst. At pH greater than PZC, the cationic dyes (such as methylene blue and malachite green) are facilely interaction with the catalyst. Apart from the adjustment of pH to control the dye adsorption, the addition of other surface modifiers can also markedly influence the extent of dye adsorption on the photocatalyst surface. For example, addition of anionic surfactant DBS (sodium dodecylbenzene sulfonate) into the acidic TiO2 colloid can significantly enhance the adsorption of the cationic dye RhB on the TiO2 surface, due to the strong mutual interaction among dyes, DBS, and TiO2. As a result,

N

photoinduced electron transfer efficiency from dye molecules to the TiO2 particles was greatly enhanced after modification of the TiO2 surface by an anionic surfactant (Qu et al. 1998). The surface fluorination of TiO2 (F-TiO2) particles was observed to enhance the adsorption of cationic dyes, such as methylene blue, malachite green, and rhodamine (Rhodamine 6G and Rhodamine B) on the surface of photocatalyst, and their photocatalytic degradation rates are consequently greatly promoted after fluorination (Wang et al. 2008). On the contrary, if the dye (phenosafranin) is separated from the surface of TiO2 by capping the TiO2 with poly styrenesulfonate, the dye can be photostabilized by the inhibition of the photoinduced electron injection from the excited dye molecule (Ziolkowski et al. 1997) 4.4.2. General Performance for Photosensitized Degradation of Dye Pollutants Most of the dyes are relatively stable under visible-light irradiation in the absence of TiO2. In the presence of TiO2, the visible-light-induced electron transfer can cause the dyes to undergo significant degradation. Various dyes with different structure and functional groups (see Fig. 4.8 for structures of some representative dyes) have been examined for their selfsensitized degradation in the presence of TiO2 under visible irradiation. They include azo dyes Acid Orange 7 (Chen et al. 2004; Stylidi et al. 2004; Wang et al. 2004), ethyl orange (EO) (Zhao et al. 2003), xanthene dyes (including Eosin Y, rose bengal, Erythrosine B, Rhodamine B (Li et al. 2002; Zhao et al. 1998), Sulforhodamine B (Liu and Zhao 2000), triphenylmethane dye malachite green (Chen et al. 2002a), crystal violet and fuchsin basic (Li et al. 1999), anthraquinone dye alizarin red (Liu et al. 2000b), squarylium cyanine

O

N

N

O

N

Br O

SO3H

Br O

OH

Br

Br

COOH

COOH SO3H

(a)

(b)

(c)

N

N

O

OH

OH OH

N N

SO3H

SO3H O

(d)

105

(e)

(f)

Figure 4.8. Several representative dyes with different functional groups: (a) rhodamine B (RhB); (b) sulforhodamine B (SRB); (c) eosin; (d) malachite green (MG); (e) alizarin red (AR); (f) Acid Orange 7 (AO7).

106

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

dye (Wu et al. 1999a, 2000), basic blue dyes (Stathatos et al. 2001), diazo dye (naphthol blue-black) (Nasr et al. 1996), phenosafranin dyes (Ziolkowski et al. 1997), Active Brilliant Red Dye X-3B (Xie et al. 2005; Xu and Langford 2001), and Reactive Red 198 (Kaur and Singh 2007). Wu et al. (1999b) investigated the photodegradation of a series of dyes (Rhodamine B, Orange II, Sulfurhodamine B, fluorescein, alizarin red, squarylium cyanine, and eosin) in the presence of TiO2 particles under airequilibrated controlled conditions and visible light illumination. Epling and Lin (2002) examined the photosensitized degradation of 15 dyes with different functionalities. However, the thiazine dye methylene blue has been reported to be resistant to sensitized degradation under visible irradiation. Besides self-sensitized degradation pathways as mentioned above, sensitization processes can also lead to the oxidative degradation of other coexisting colorless organic pollutants via cationic radical of the sensitizer and the activeoxygen species formed during dye sensitization. Bendig and co-workers (Ross et al. 1994) found that irradiation of an aqueous solution of pollutant in the presence of TiO2 and rose bengal with visible light leads to decomposition of the herbicide terbutylazine, along with the photosensitized degradation of rose bengal. They further showed that the sensitized degradation of terbutryne on the TiO2 particles by tris (4,40 -dicarboxy-2,20 -bipyridyl)ruthenium(II) chloride complexes is more efficient (Lobedank et al. 1997). The selfsensitized degradation of xanthene dyes such as Eosin Y, rose bengal, eruthrosine, and Rhodamine B as sensitizers can result in the concomitant degradation of a small organic compound (2,4-dichlorophenol) under visible-light irradiation (Li et al. 2002). The sensitizers can be regenerated if the produced radical cation of the dye obtains an electron from the electron donor (organic pollutants). The dye-sensitized TiO2 can be considered as photocatalysts, which can degrade other organic pollutants under visible irradiation without the dyes being destroyed. Hodak et al. (1996) have reported the degradation of phenols, thiophenols, 4-chlorophenols, hydroquinones, and salicylic acid by a phthalocyanine dye (hydroxyaluminumtricarboxymonoamide phthalocyanine) sensitization of the TiO2 semiconductor particles, while EDTA, oxalic acid, and benzoquinone did not show any changes after irradiation. They proposed that the radical cation of the phthalocyanine produced by the electron injection of the dye into the conduction band of the TiO2 is the species responsible for the oxidation of the substrates (Hodak et al. 1996). Investigation of the photocatalytic activity of polycrystalline TiO2 samples sensitized by Cu(II)- or metal-free porphyrin dyes showed that the presence of the sensitizers is beneficial for the photoactivity of 4-nitrophenol (4-NP) photodegradation in aqueous suspension. A comparison with similar samples modified by sensitizer Cu(II)- and metal-free phthalocyanines showed that the presence of porphyrin is more

beneficial for both the decomposition rate of 4-nitrophenol and the disappearance of nonpurgeable organic carbon (NPOC) (Mele et al. 2003). Photodegradation of organic pollutants, such as phenol, chlorophenol, 1,2-dichloroethane, trichloroethylene, and pesticide (atrazine), in water has been achieved on the surface of TiO2 semiconductor modified with sensitizers thionine, Eosin Y, Rhodamine B, methylene blue, Nile Blue A, and Safranine O by using visible light. After 5 h of irradiation with a 50-W tungsten lamp, over 55%–72% degradation of pollutants is achieved for the thionine- and Eosin Y–sensitized systems (Chatterjee and Mahata 2001, 2004; Chatterjee et al. 2006). The activation of TiO2 photocatalyst for photocatalysis under visible light using Acid Red 44 (Moon et al. 2003) or poly(fluorene-co-thiophene) (PFT) (Song et al. 2007) has been investigated, and the decomposition of phenol can be realized on the dye-sensitized photocatalysis under visible-light irradiation There are also reports on the reduction of other pollutants by dye-sensitized TiO2 under visible-light irradiation. Evidently, the reduction reaction is induced from the conduction band electron injected from the excited dyes. For example, Choi and colleagues have developed several organic-dyesensitized TiO2 systems for the reductive dechlorination of the perchlorinated compounds such as CCl4 and CCl3CO2 and for the reduction of heavy-metal ions under visible irradiation (Cho and Choi 2002; Cho et al. 2001, 2004). They have investigated the photoreductive decomposition of CCl4 using visible light on TiO2 sensitized by tris-(4,40 dicarboxy-2, 20 -bipyridyl) ruthenium(II) complexes. It was found that the sensitized TiO2 could degrade CCl4 under the irradiation of > 420 nm with a quantum yield of about 103. The rate decreases in the presence of O2, due to competition for the conduction band electrons. 2-Propanol was used as an efficient sacrificial donor of electron to regenerate the oxidized sensitizer. They further found that the deposition of platinum nanoparticles on the dye-sensitized TiO2 drastically enhanced the reductive degradation of trichloroacetate and CCl4. In Pt/TiO2/dye system, the water molecules can act as electron donors to regenerate the sensitizer, with a concurrent production of dioxygen. Finally, it should be pointed out that, in order to clarify the role of visible light in the photocatalytic degradation of dyes and to investigate the mechanism of photosensitized degradation, only some of the studies that have employed visible irradiation to photocatalytically degrade the dyes are mentioned here. In these studies, ether the UV irradiation light is filtered out from a mixed light source or a monochromatic visible-light source (with wavelength longer than 420 nm) is used to avoid direct excitation of the semiconductor. There have also been many more studies using UV or UV-visible mixed light sources for the photocatalytic degradation of dye pollutant; these studies are not described here. In these systems, the pathways of semiconductor-initiated degradation and photosensitizer-initiated degradation are operative

REFERENCES

simultaneously. In fact, during the degradation of dye pollutants in the environment or in the photocatalytic reactor, UV and visible-light radiation pathways are usually coexistent. It is difficult to distinguish these two pathways from each other. However, it is certain that the existence of the photosensitizer-initiated pathway may improve the overall efficiency of the degradation of dye pollutants and the transformation of colorless pollutants in both the natural and photoreactors.

107

ACKNOWLEDGMENT Parts of the work described in Sections 3.3 and 3.4 were financially supported by Project 973 (of the Ministry of Science and Technology of China Grants 2007CB613306 and 1010CB933503), the National Science Foundation of China (Grants 20920102034, 20877076, and 20907056), and the Chinese Academy of Sciences.

4.5. CONCLUSIONS REFERENCES This chapter emphasizes the importance of photochemistry (particularly photocatalysis) on the environmental effect, transformation, and fate of anthropogenic organic compounds. Two typical environmentally relevant processes occurring on the surface of natural minerals are reviewed in detail: 1. Organic compounds with carboxylic and/or hydroxyl/ phenolic functional groups can readily adsorb to ironbearing solids and be oxidized through the pathways of semiconductor photocatalysis and/or photoinduced LMCT processes. However, on the basis of the available data, it is not possible yet to unambiguously distinguish which mechanism operates in the photodegradation of organic compounds at the iron oxide surface. The exact effect of these photocatalytic oxidation processes on the environmental fate of organic pollutant still needs further study. Another ubiquitous iron species in layered smectites also plays a critical role in the transformation of organic pollutants; however, its photoredox cycling scheme requires additional investigation. 2. Photosensitized degradation of organic pollutants, especially for dyestuffs, represents an important transformation pathway of anthropogenic organic compounds under visible-light irradiation. Although the laboratory studies have attempted to clarify the detailed mechanism of photosensitized degradation of dyes, there is still a dramatic lack of field data to evaluate its contribution to dye transformation in polluted soils and sediments. Another aspect that warrants further investigation is the coupling process and mechanism between the photocatalytic degradation of dye pollutants (or the natural chromogenic matters) with the transformation of other environmental species (such as colorless organic pollutants, toxic metals). Although the importance of these coupling processes has been recognized for a long time, systemic evaluation has been insufficient. The toxicity for intermediates and products during the photocatalytic degradation of different dye pollutants should also be evaluated critically.

Ahn, M. Y., Filley, T. R., Jafvert, C. T., Nies, L., Hua, I., and BezaresCruz, J. (2006), Photodegradation of decabromodiphenyl ether adsorbed onto clay minerals, metal oxides, and sediment, Environ. Sci. Technol. 40, 215–220. Andreozzi, R., Caprio, V., and Marotta, R. (2003), Iron(III) (hydro) oxide-mediated photooxidation of 2-aminophenol in aqueous solution: A kinetic study, Water Res. 37, 3682–3688. Bakardjieva, S., Stengl, V., Subrt, J., Houskova, V., and Kalenda, P. (2007), Photocatalytic efficiency of iron oxides: Degradation of 4-chlorophenol, J. Phys. Chem. Solids 68, 721–724. Balzani, V., Bergamini, G., Campagna, S., and Puntoriero, F. (2007), Photochemistry and photophysics of coordination compounds: Overview and general concepts, in Photochemistry and Photophysics of Coordination Compounds I, Balzani, V., and Campagna, S., eds., Springer-Verlag, Berlin/Heidelberg, pp. 1–36. Bandara, J., Mielczarski, J. A., and Kiwi, J. (1999a), 1. Molecular mechanism of surface recognition. Azo dyes degradation on Fe, Ti, and Al oxides through metal sulfonate complexes, Langmuir 15, 7670–7679. Bandara, J., Mielczarski, J. A., and Kiwi, J. (1999b). 2. Photosensitized degradation of azo dyes on Fe, Ti, and Al oxides. Mechanism of charge transfer during the degradation, Langmuir 15, 7680–7687. Bandara, J., Mielczarski, J. A., Lopez, A., and Kiwi, J. (2001), 2. Sensitized degradation of chlorophenols on iron oxides induced by visible light comparison with titanium oxide, Appl. Catal. B Environ. 34, 321–333. Blatter, F. and Frei, H. (1994), Selective photooxidation of small alkenes by O2 with red light in zeolite Y, J. Am. Chem. Soc. 116, 1812–1820. Blatter, F., Sun, H., Vasenkov, S., and Frei, H. (1998), Photocatalyzed oxidation in zeolite cages, Catal. Today 41, 297–309. Boreen, A. L., Arnold, W. A., and McNeill, K. (2004), Photochemical fate of sulfa drugs in the aquatic environment: Sulfa drugs containing five-membered heterocyclic groups, Environ. Sci. Technol. 38, 3933–3940. Boreen, A. L., Arnold, W. A., and McNeill, K. (2005), Tripletsensitized photodegradation of sulfa drugs containing six-membered heterocyclic groups: Identification of an SO2 extrusion photoproduct, Environ. Sci. Technol. 39, 3630–3638. Boreen, A. L., Edhlund, B. L., Cotner, J. B., and McNeill, K. (2008), Indirect photodegradation of dissolved free amino acids: The contribution of singlet oxygen and the differential reactivity

108

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

of DOM from various sources, Environ. Sci. Technol. 42, 5492–5498. Borer, P., Hug, S. J., Sulzberger, B., Kraemer, S. M., and Kretzschmar, R. (2007), Photolysis of citrate on the surface of lepidocrocite: An in situ attenuated total reflection infrared spectroscopy study, J. Phys. Chem. C 111, 10560–10569. Bourikas, K., Stylidi, M., Kondarides, D. I., and Verykios, X. E. (2005), Adsorption of acid orange 7 on the surface of titanium dioxide, Langmuir 21, 9222–9230. Brown, D., Hitz, H. R., and Sch€aer, L. (1981), The assessment of the possible inhibitory effect of dyestuffs on aerobic waste-water bacteria experience with a screening test, Chemosphere 10, 245–261. Brown, G. E., Henrich, V. E., Casey, W. H., Clark, D. L., Eggleston, C., Felmy, A., Goodman, D. W., Gratzel, M., Maciel, G., McCarthy, M. I., Nealson, K. H., Sverjensky, D. A., Toney, M. F., and Zachara, J. M. (1999), Metal oxide surfaces and their interactions with aqueous solutions and microbial organisms, Chem. Rev. 99, 77–174. Burrows, H. D., Canle, L. M., Santaballa, J. A., and Steenken, S. (2002), Reaction pathways and mechanisms of photodegradation of pesticides, J. Photochem. Photobiol. B Biol. 67, 71–108. Chang, J. -S., Chou, C., Lin, Y. -C., Lin, P. -J., Ho, J. -Y., and Hu, T. L. (2001), Kinetic characteristics of bacterial azo-dye decolorization by Pseudomonas luteola, Water Res. 35, 2841–2850. Chatterjee, D. and Mahata, A. (2004), Evidence of superoxide radical formation in the photodegradation of pesticide on the dye modified TiO2 surface using visible light, J. Photochem. Photobiol. A Chem. 165, 19–23. Chatterjee, D., Dasgupta, S., and Rao, N. N. (2006), Visible light assisted photodegradation of halocarbons on the dye modified TiO2 surface using visible light, Solar Energy Mater. Solar Cell 90, 1013–1020. Chatterjee, S., Sarkar, S., and Bhattacharyya, S. N. (1994), Photodegradation of phenol by visible light in the presence of colloidal Fe2O3, J. Photochem. Photobiol. A Chem. 81, 199–203. Chen, C. C., Li, X. Z., Ma, W. H., Zhao, J. C., Hidaka, H., and Serpone, N. (2002a). Effect of transition metal ions on the TiO2assisted photodegradation of dyes under visible irradiation: A probe for the interfacial electron transfer process and reaction mechanism, J. Phys. Chem. B 106, 318–324. Chen, C. C., Zhao, W., Li, J. Y., and Zhao, J. C. (2002b), Formation and identification of intermediates visible-light-assisted photodegradation sulforhodamine-B dye in aqueous TiO2 dispersion, Environ. Sci. Technol. 36, 3604–3611. Chen, X. and Mao, S. S. (2007), Titanium dioxide nanomaterials: Synthesis, properties, modifications, and applications, Chem. Rev. 107, 2891–2959. Chen, Y., Wang, K., and Lou, L. (2004), Photodegradation of dye pollutants on silica gel supported TiO2 particles under visible light irradiation, J. Photochem. Photobiol. A Chem. 163, 281–287. Cheng, M. M., Song, W. J., Ma, W. H., Chen, C. C., Zhao, J. C., Lin, J., and Zhu, H. Y. (2008), Catalytic activity of iron species in layered clays for photodegradation of organic dyes under visible irradiation, Appl. Catal. B Environ. 77, 355–363.

Cho, Y., Choi, W., Lee, C. H., Hyeon, T., and Lee, H. I. (2001), Visible light-induced degradation of carbon tetrachloride on dye-sensitized TiO2, Environ. Sci. Technol. 35, 966–970. Cho, Y. and Choi, W. (2002). Visible light-induced reactions of humic acids on TiO2, J. Photochem. Photobiol. A: Chem. 148, 129–135. Cho, Y., Kyung, H., and Choi, W. (2004), Visible light activity of TiO2 for the photoreduction of CCl4 and Cr(VI) in the presence of nonionic surfactant (Brij), Appl. Catal. B Environ. 52, 23–32. Ciani, A., Goss, K. U., and Schwarzenbach, R. P. (2005a), Light penetration in soil and particulate minerals, Eur. J. Soil Sci. 56, 561–574. Ciani, A., Goss, K. U., and Schwarzenbach, R. P. (2005b), Photodegradation of organic compounds adsorbed in porous mineral layers: Determination of quantum yields, Environ. Sci. Technol. 39, 6712–6720. Claus, H., Faber, G., and Konig, H. (2002), Redox-mediated decolorization of synthetic dyes by fungal laccases, Appl. Microbiol. Biotechnol. 59, 672–678. Cornell, R. M. and Schwertmann, U. (2003), The Iron Oxides: Structure, Properties, Reactions, Occurences and Uses, WileyVCH, Weinheim. Coyle, H. D. (1991), Introduction to Organic Photochemistry, Wiley, New York. Cunningham, K. M., Goldberg, M. C., and Weiner, E. R. (1985), The aqueous photolysis of ethylene glycol adsorbed on goethite, Photochem. Photobiol. 41, 409–416. Cunningham, K. M., Goldberg, M. C., and Weiner, E. R. (1988), Mechanisms for aqueous photolysis of adsorbed benzoate, oxalate, and succinate on iron oxyhydroxide (goethite) surfaces, Environ. Sci. Technol. 22, 1090–1097. Du, W. P., Xu, Y. M., and Wang, Y. S. (2008), Photoinduced degradation of Orange II on different iron (hydro)oxides in aqueous suspension: Rate enhancement on addition of hydrogen peroxide, silver nitrate, and sodium fluoride, Langmuir 24, 175–181. Edhlund, B. L., Arnold, W. A., and McNeill, K. (2006), Aquatic photochemistry of nitrofuran antibiotics, Environ. Sci. Technol. 40, 5422–5427. Emeline, A. V., Kataeva, G. V., Ryabchuk, V. K., and Serpone, N. (1999), Photostimulated generation of defects and surface reactions on a series of wide band gap metal-oxide solids, J. Phys. Chem. B 103, 9190–9199. Emeline, A. V., Kataeva, G. V., Litke, A. S., Rudakova, A. V., Ryabchuk, V. K., and Serpone, N. (1998a), Spectroscopic and photoluminescence studies of a wide band gap insulating material: Powdered and colloidal ZrO2 sols, Langmuir 14, 5011–5022. Emeline, A. V., Petrova, S. V., Ryabchuk, V. K., and Serpone, N. (1998b), Photochemical and photophysical processes on the surface of wide band gap insulator particulates: Gas/solid system involving scandia (Sc2O3) particles, Chem. Mater. 10, 3484–3491. Epling, G. A. and Lin, C. (2002), Photoassisted bleaching of dyes utilizing TiO2 and visible light, Chemosphere 46, 561–570.

REFERENCES

Guillard, C., Hoang-Van, C. F., Pichat, P., and Marme, F. (1995), Laboratory study of the respective roles of ferric oxide and released or added ferric ions in the photodegradation of oxalic acid in aerated liquid water, J. Photochem. Photobiol. A Chem. 89, 221–227. He, J., Ma, W. H., He, J. J., Zhao, J. C., and Yu, J. C. (2002), Photooxidation of azo dye in aqueous dispersions of H2O2/ a-FeOOH, Appl. Catal. B Environ. 39, 211–220. He, J., Ma, W. H., Song, W. J., Zhao, J. C., Qian, X. H., Zhang, S. B., and Yu, J. C. (2005), Photoreaction of aromatic compounds at a-FeOOH/H2O interface in the presence of H2O2: Evidence for organic-goethite surface complex formation, Water Res. 39, 119–128. Hodak, J., Quinteros, C., Litter, M. I., and Roman, E. S. (1996), Sensitization of TiO2 with phthalocyanines. Part 1. Photo-oxidations using hydroxoaluminium tricarboxymonoamidephthalocyanine adsorbed on TiO2, J. Chem. Soc. Faraday Trans. 92, 5081–5088. Hoffmann, M. R., Martin, S. T., Choi, W. Y., and Bahnemann, D. W. (1995), Environmental applications of semiconductor photocatalysis, Chem. Rev. 95, 69–96. Hofstetter, T. B., Schwarzenbach, R. P., and Haderlein, S. B. (2003), Reactivity of Fe(II) species associated with clay minerals, Environ. Sci. Technol. 37, 519–528. Huang, P. (2004), Soil mineral-organic matter-microorganism interactions: Fundamental and impacts, Adv. Agron. 82, 391–472. Johansen, A. M. and Key, J. M. (2006), Photoreductive dissolution of ferrihydrite by methanesulfinic acid: Evidence of a direct link between dimethylsulfide and iron-bioavailability, Geophys. Res. Lett. 33, L14818. Karametaxas, G., Hug, S. J., and Sulzberger, B. (1995), Photodegradation of EDTA in the presence of lepidocrocite, Environ. Sci. Technol. 29, 2992–3000. Karkmaz, M., Puzenat, E., Guillard, C., and Herrmann, J. M. (2004), Photocatalytic degradation of the alimentary azo dye amaranth: Mineralization of the azo group to nitrogen, Appl. Catal. B Environ. 51, 183–194. Kaur, S. and Singh, V. (2007), Visible light induced sonophotocatalytic degradation of Reactive Red dye 198 using dye sensitized TiO2, Ultrason. Sonochem. 14, 531–537. Kiwi, J. and Gr€atzel, M. (1987), Light-induced hydrogen formation and photo-uptake of oxygen in colloidal suspensions of a-Fe2O3, J. Chem. Soc. Faraday Trans. 83, 1101–1108. Klessinger, M. and Michl, J. (1995), Excited States and Photochemisty of Organic Molecule, VCH, Weinheim. Kuramoto, H. (1996), Disulfide oligomer on a-FeOOH adsorption state and photoadsorption, Langmuir 11, 3417–3422. Lan, Q., Li, F. B., Liu, C. S., and Li, X. Z. (2008), Heterogeneous photodegradation of pentachlorophenol with maghemite and oxalate under UV illumination, Environ. Sci. Technol. 42, 7918–7923. Latch, D. E., Stender, B. L., Packer, J. L., Arnold, W. A., and McNeill, K. (2003), Photochemical fate of pharmaceuticals in the environment: Cimetidine and ranitidine, Environ. Sci. Technol. 37, 3342–3350.

109

Lee, W. J. and Batchelor, B. (2004), Abiotic reductive dechlorination of chlorinated ethylenes by iron-bearing phyllosilicates, Chemosphere 56, 999–1009. Legrini, O., Oliveros, E., and Braun, A. M. (1993), Photochemical processes for water treatment, Chem. Rev. 93, 671–698. Leland, J. K. and Bard, A. J. (1987), Photochemistry of colloidal semiconducting iron oxide polymorphs, J. Phys. Chem. 91, 5076–5083. Li, F. B., Li, X. Z., Li, X. M., Liu, T. X., and Dong, J. (2007), Heterogeneous photodegradation of bisphenol A with iron oxides and oxalate in aqueous solution, J. Colloid Interface Sci. 311, 481–490. Li, X. Z., Liu, G. M., and Zhao, J. C. (1999), Two competitive primary processes in the photodegradation of cationic triarylmethane dyes under visible irradiation in TiO2 dispersions, New J. Chem. 23, 1193–1196. Li, X. Z., Zhao, W., and Zhao, J. C. (2002), Visible light-sensitized semiconductor photocatalytic degradation of 2,4-dichlorophenol, Sci. China Ser. B Chem. 45, 421–425. Lin, W. and Frei, H. (2005a), Anchored metal-to-metal chargetransfer chromophores in a mesoporous silicate sieve for visiblelight activation of titanium centers, J. Phys. Chem. B 109, 4929–4935. Lin, W. and Frei, H. (2005b), Photochemical CO2 splitting by metalto-metal charge-transfer excitation in mesoporous ZrCu(I)MCM-41 silicate sieve, J. Am. Chem. Soc. 127, 1610–1611. Litter, M. I., Villegas, M., and Blesa M. A. (1994), Photodissolution of iron oxides in malonic acid, Can. J. Chem. 72, 2037–2043. Liu, C. S., Li, F. B., Li, X. M., Zhang, G., and Kuang, Y. Q. (2006), The effect of iron oxides and oxalate on the photodegradation of 2-mercaptobenzothiazole, J. Molec. Catal. A Chem. 252, 40–48. Liu, G. M. and Zhao, J. C. (2000), Photocatalytic degradation of dye sulforhodamine B: A comparative study of photocatalysis with photosensitization, New J. Chem. 24, 411–417. Liu, G. M., Li, X. Z., Zhao, J. C., Hidaka, H., and Serpone, N. (2000a), Photooxidation pathway of sulforhodamine-B. Dependence on the adsorption mode on TiO2 exposed to visible light radiation. Environ, Sci. Technol. 34, 3982–3990. Liu, G. M., Li, X. Z., Zhao, J. C., Horikoshi, S., and Hidaka, H. (2000b), Photooxidation mechanism of dye alizarin red in TiO2 dispersions under visible illumination: An experimental and theoretical examination, J. Molec. Catal. A Chem. 153, 221–229. Lobedank, J., Bellmann, E., and Bendig, J. (1997), Sensitized photocatalytic oxidation of herbicides using natural sunlight, J. Photochem. Photobiol. A Chem. 108, 89–93. Mao, Y. and Thomas, J. (1993), Photoinduced electron transfer and subsequent chemical reactions of adsorbed thianthrene on clay surfaces, J. Org. Chem. 58, 6641–6649. Matsuzawa, S., Nasser-Ali, L., and Garrigues, P. (2001), Photolytic behavior of polycyclic aromatic hydrocarbons in diesel particulate matter deposited on the ground, Environ. Sci. Technol. 35, 3139–3143. Mazellier, P. and Bolte, M. (2000), Heterogeneous light-induced transformation of 2,6-dimethylphenol in aqueous suspensions

110

PHOTOCATALYTIC DEGRADATION OF ORGANIC CONTAMINANTS ON MINERAL SURFACES

containing goethite, J. Photochem. Photobiol. A Chem. 132, 129–135. Mazellier, P. and Sulzberger, B. (2001), Diuron degradation in irradiated, heterogeneous iron/oxalate systems: The rate-determining step, Environ. Sci. Technol. 35, 3314–3320. Meehan, C., Banat, I. M., McMullan, G., Nigam, P., Smyth, F., and Marchant, R. (2000), Decolorization of Remazol Black-B using a thermotolerant yeast, Kluyveromyces marxianus IMB3, Environ. Int. 26, 75–79. Mehra, O. P. and Jackson, M. L. (1960), Iron oxides removal from soils and clays by a dithionite-citrate system buffered with sodium carbonate, Clays Clay Miner. 7, 317–327. Mele, G., Del Sole, R., Vasapollo, G., Garcıa-Lo´pez, E., Palmisano, L., and Schiavello, M. (2003), Photocatalytic degradation of 4-nitrophenol in aqueous suspension by using polycrystalline TiO2 impregnated with functionalized Cu(II)-porphyrin or Cu (II)-phthalocyanine, J. Catal. 217, 334–342. Moon, J., Yun, C. Y., Chung, K. -W., Kang, M. -S., and Yi, J. (2003), Photocatalytic activation of TiO2 under visible light using Acid Red 44, Catal. Today 87, 77–86. Murad, E. and Fischer, W. R. (1988), The geobiochemical cycle of iron, in Iron in Soils and Clay Minerals, Stucki, J. W., Goodman, B. A., and Schwertmann, U., eds., Reidel, Dordrecht. Myli, K. B., Larsen, S. C., and Grassian, V. H. (1997), Selective photooxidation reactions in zeolites X, Y and ZSM-5, Catal. Lett. 48, 199–202. Nakamura, R. and Frei, H. (2006), Visible light-driven water oxidation by Ir oxide clusters coupled to single Cr centers in mesoporous silica, J. Am. Chem. Soc. 128, 10668–10669. Nakamura, R., Okamoto, A., Osawa, H., Irie, H., and Hashimoto, K. (2007), Design of all-inorganic molecular-based photocatalysts sensitive to visible light: Ti (IV)-O-Ce(III) bimetallic assemblies on mesoporous silica, J. Am. Chem. Soc. 129, 9596–9597. Nasr, C., Vinodgopal, K., Fisher, L., Hotchandani, S., Chattopadhyay, A. K., and Kamat, P. V. (1996), Environmental photochemistry on semiconductor surfaces. Visible light induced degradation of a textile diazo dye, Naphthol Blue Black, on TiO2 nanoparticles. J. Phys. Chem. 100, 8436–8442. Pehkonen, S. O., Ron, S., Erel, Y., Webb, S., and Hoffmann, M. R. (1993), Photoreduction of iron oxyhydroxides in the presence of important atmospheric organic compounds, Environ. Sci. Technol. 27, 2056–2062. Pehkonen, S. O., Siefert, R. L., and Hoffmann, M. R. (1995), Photoreduction of iron oxyhydmxides and the photooxidation of halogenated acetic acids, Environ. Sci. Technol. 29, 1215–1222. Qu, P., Zhao, J. C., Zang, L., Shen, T., and Hidaka, H. (1998), Enhancement of the photoinduced electron transfer from cationic dyes to colloidal TiO2 particles by addition of an anionic surfactant in acidic media, Colloid Surf. A Physicochem. Eng. Asp. 138, 39–50. Reyes, C. A., Medina, M., Crespo-Hernandez, C., Cedeno, M. Z., Arce, R., Rosario, O., Steffenson, D. M., Ivanov, I. N., Sigman, M. E., and Dabestani, R. (2000), Photochemistry of pyrene on

unactivated and activated silica surfaces, Environ. Sci. Technol. 34, 415–421. Robinson, T., McMullan, G., Marchant, R., and Nigam, P. (2001), Remediation of dyes in textile effluent: A critical review on current treatment technologies with a proposed alternative, Bioresour. Technol. 77, 247–255. Ross, H., Bendig, J., and Hecht, S. (1994), Sensitized photocatalytical oxidation of terbutylazine, Solar Energy Mater. Solar Cell 33, 475–481. Seo, Y. S., Lee, C., Lee, K. H., and Yoon, K. B. (2005), 1:1 and 2:1 charge-transfer complexes between aromatic hydrocarbons and dry titanium dioxide, Angew. Chem. Int. Ed. 44, 910–913. Serpone, N. (2006), Is the band gap of pristine TiO2 narrowed by anion- and cation-doping of titanium dioxide in second-generation photocatalysts? J. Phys. Chem. B 110, 24287–24293. Sherman, D. M. (2005), Electronic structures of iron(III) and manganese(IV) (hydro)oxide minerals: Thermodynamics of photochemical reductive dissolution in aquatic environments, Geochim. Cosmochim. Acta 69, 3249–3255. Sherman, D. M. and Waite, T. D. (1985), Electronic spectra of Fe3 þ oxides and oxide hydroxides in the near IR to near UV, Am. Miner. 70, 1262–1269. Shichi, T. and Takagi, K. (2000), Clay minerals as photochemical reaction fields, J. Photochem. Photobiol. C Photochem. Rev. 1, 113–130. Siffert, C. and Sulzberger, B. (1991), Light-induced dissolution of hematite in the presence of oxalate. A case study, Langmuir 7, 1627–1634. Song, L., Qiu, R., Mo, Y., Zhang, D., Wei, H., and Xiong, Y. (2007), Photodegradation of phenol in a polymer-modified TiO2 semiconductor particulate system under the irradiation of visible light, Catal. Commun. 8, 429–433. Song, W. J., Cheng, M. M., Ma, J. H., Ma, W. H., Chen, C. C., and Zhao, J. C. (2006), Decomposition of hydrogen peroxide driven by photochemical cycling of iron species in clay, Environ. Sci. Technol. 40, 4782–4787. Stathatos, E., Petrova, T., and Lianos, P. (2001), Study of the efficiency of visible-light photocatalytic degradation of basic blue adsorbed on pure and doped mesoporous titania films, Langmuir 17, 5025–5030. Stucki, J. W. (2005), Properties and behaviour of iron in clay minerals, in Handbook of Clay Science, Bergaya, F., Theng, B. K. G., and Lagaly, G., eds., Elsevier, Amsterdam, pp. 429–482. Stumm, W. and Sulzberger, B. (1992), The cycling of iron in natural environments: Considerations based on laboratory studies of heterogeneous redox processes, Geochim. Cosmochim. Acta 56, 3233–3257. Stylidi, M., Kondarides, D. I., and Verykios, X. E. (2004), Visible light-induced photocatalytic degradation of Acid Orange 7 in aqueous TiO2 suspensions, Appl. Catal. B Environ. 47, 189–201. Sun, C., Zhao, D., Chen, C., Ma, W., and Zhao, J. (2009), TiO2mediated photocatalytic debromination of decabromodiphenyl ether: Kinetics and intermediates, Environ. Sci. Technol. 43, 157–162.

REFERENCES

Thomas, J. K. (2005), Physical aspects of radiation-induced processes on SiO2, c-Al2O3, zeolites, and clays. Chem. Rev. 105, 1683–1734. Thompson, T. L., and Yates, J. T. (2006), Surface science studies of the photoactivation of TiO2—new photochemical processes, Chem. Rev. 106, 4428–4453. Volodin, A. M. (2000), Photoinduced phenomena on the surface of wide-band-gap oxide catalysts, Catal. Today 58, 103–114. Wallington, T. J. and Nielsen, O. J. (2005), Atmospheric Photooxidation of Gas Phase Air Pollutants, Springer-Verlag, Berlin/ Heidelberg. Wang, K., Zhang, J., Lou, L., Yang, S., and Chen, Y. (2004), UV or visible light induced photodegradation of AO7 on TiO2 particles: The influence of inorganic anions, J. Photochem. Photobiol. A Chem. 165, 201–207. Wang, Q., Chen, C., Zhao, D., Ma, W., and Zhao, J. (2008), Change of adsorption modes of dyes on fluorinated TiO2 and its effect on photocatalytic degradation of dyes under visible irradiation, Langmuir 24, 7338–7345. Weare, W. W., Pushkar, Y., Yachandra, V. K., and Frei, H. (2008), Visible light-induced electron transfer from di-m-oxo-bridged dinuclear Mn complexes to Cr centers in silica nanopores, J. Am. Chem. Soc. 130, 11355–11363. Wu, T. X., Lin, T., Zhao, J. C., Hidaka, H., and Serpone, N. (1999a), TiO2-assisted photodegradation of dyes. 9. Photooxidation of a squarylium cyanine dye in aqueous dispersions under visible light irradiation, Environ. Sci. Technol. 33, 1379–1387. Wu, T. X., Liu, G. M., Zhao, J. C., Hidaka, H., and Serpone, N. (1999b), Evidence for H2O2 generation during the TiO2-assisted photodegradation of dyes in aqueous dispersions under visible light illumination, J. Phys. Chem. B 103, 4862–4867. Wu, T. X., Liu, G. M., Zhao, J. C., Hidaka, H., and Serpone, N. (2000), Mechanistic study of the TiO2-assisted photodegradation of squarylium cyanine dye in methanolic suspensions exposed to visible light, New J. Chem. 24, 93–98. Xie, T.-H., Sun, X., and Lin, J. (2008), Enhanced photocatalytic degradation of RhB driven by visible light-induced MMCT of Ti (IV)OFe(II) formed in Fe-doped SrTiO3, J. Phys. Chem. C 112, 9753–9759.

111

Xie, Y., Yuan, C., and Li, X. (2005), Photocatalytic degradation of X-3B dye by visible light using lanthanide ion modified titanium dioxide hydrosol system, Colloid Surf. A Physicochem. Eng. Asp. 252, 87–94. Xu, Y. and Langford, C. H. (2001), UV- or visible-light-induced degradation of X3B on TiO2 nanoparticles: The influence of adsorption, Langmuir 17, 897–902. Xu, Y. and Schoonen, M. A. A. (2000), The absolute energy position of conduction and valence bands of selected semiconducting minerals, Am. Miner. 85, 543–556. Xyla, A. G., Sulzberger, B., LutherIII, G. W., Hering, J. G., Van Cappellen, P., and Stumm, W. (1992), Reductive dissolution of manganese(III,IV) (hydro)oxides by oxalate: The effect of pH and light, Langmuir 8, 95–103. Zafiriou, O. C., Joussot-Dubien, J., Zepp, R. G., and Zika, R. G. (1984), Photochemistry of natural waters, Environ. Sci. Technol. 18, 358A–371A. Zhao, J. C., Hidaka, H., Takamura, A., Pelizzetti, E., and Serpone, N. (1993), Photodegradation of surfactants 11. Zeta-potential measurements in the photocatalytic oxidation of surfactants in aqueous TiO2 dispersions, Langmuir 9, 1646–1650. Zhao, J. C., Wu, T. X., Wu, K. Q., Oikawa, K., Hidaka, H., and Serpone, N. (1998), Photoassisted degradation of dye pollutants. 3. Degradation of the cationic dye rhodamine B in aqueous anionic surfactant/TiO2 dispersions under visible light irradiation: Evidence for the need of substrate adsorption on TiO2 particles, Environ. Sci. Technol. 32, 2394–2400. Zhao, W., Chen, C. C., Ma, W. H., Zhao, J. C., Wang, D. X., Hidaka, H., and Serpone, N. (2003), Efficient photoinduced conversion of an azo dye on hexachloroplatinate(IV)-modified TiO2 surfaces under visible light irradiation—a photosensitization pathway, Chem. Eur. J. 9, 3292–3299. Ziolkowski, L., Vinodgopal, K., and Kamat, P. V. (1997), Photostabilization of organic dyes on poly(styrenesulfonate)-capped TiO2 nanoparticles, Langmuir 13, 3124–3128. Zollinger, H. (1987), Color Chemistry: Synthesis, Properties and Applications of Organic Dyes and Pigments, VCH, New York.

PART II ANTHROPOGENIC ORGANIC COMPOUNDS IN AIR, WATER, AND SOIL, AND THEIR GLOBAL CYCLING

5 SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES HANS PETER H. ARP AND KAI-UWE GOSS 5.1. Introduction 5.1.1. Environmental Relevance of Ambient Gas/Particle Partitioning 5.1.2. Definition of Ambient Equilibrium Gas/Particle Partitioning 5.1.3. Aerosol Sources, Composition, and Size Fractions 5.2. Measuring Ambient Gas/Particle Partitioning 5.2.1. Sample-and-Extract Methods 5.2.2. Inverse Gas Chromatography 5.3. Traditional Gas/Particle Partitioning Models for Apolar Semivolatile Organic Compounds 5.3.1. Basic Partitioning Theory 5.3.2. The Junge–Pankow Model 5.3.3. Pankow Absorption–Adsorption Model 5.3.4. The Octanol Absorptive Model 5.3.5. The EC þ OC Model 5.3.6. Criticisms and Applicability of Traditional Models and SP-LFERs 5.4. Sorbent–Sorbate Interactions Involved in Partitioning and PP-LFERs 5.4.1. Overview of Sorbent-Sorbate Interactions 5.4.2. Describing Absorptive Partitioning with PP-LFERs 5.4.3. Describing Adsorptive Partitioning with PP-LFERs 5.4.4. Other Partitioning Models that Account for Sorbate-Sorbent Diversity 5.4.5. Temperature Dependence of Partitioning 5.5. Partitioning to Individual Aerosol Components 5.5.1. Pure Water 5.5.2. Snow and Ice 5.5.3. Minerals and Metal Oxides 5.5.4. Salts 5.5.5. Elemental Carbon 5.5.6. Water-Soluble Organic Matter 5.5.7. Water-Insoluble Organic Matter

5.6. Partitioning to Mixed Particle Phases 5.6.1. Terrestrial Aerosols 5.6.2. Other Ambient Particles 5.7. Conclusions

5.1. INTRODUCTION 5.1.1. Environmental Relevance of Ambient Gas/Particle Partitioning The air you are breathing contains several organic chemicals. Although the majority of these chemicals are benign, some are bioaccumulative and in sufficient concentrations are toxic. A fundamental aspect governing how you are exposed to these chemicals is if they are present as a vapor or sorbed to airborne particles. For instance, some toxic chemicals are highly volatile (such as carbon monoxide and hydrogen cyanide), and exist in the air completely as a vapor. Other toxic chemicals, such as the heavier polychlorinated dibenzodioxins (PCDDs), exhibit low volatility, and thus do not enter the air substantially as a vapor; however, they can enter the air by sorbing to airborne particles, like dust and smoke particulates. Once inhaled, larger particles will be trapped by the mouth and throat, where they can subsequently release chemicals into the body through mucus membranes. Many of the smaller particles (<1 mm), however, will make it to the lungs, where they can accumulate and release chemicals. The smallest of particles (<100 nm) can even penetrate cell membranes in the lungs and throat, enter the bloodstream, and translocate into organs such as the brain (Oberdorster et al. 2003). Thus, how we are exposed to chemical

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

115

116

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

and particle pollutants, from either cigarette smoke, traffic emissions, industrial emissions, chemical warfare, or other sources, is to a large degree governed by gas/particle partitioning behavior, or how organic chemicals partition between the air and particle phases. Gas/particle partitioning also plays a central role in the long-range transport of semivolatile organic compounds (SVOCs). It is highly unlikely back in the 1940s–1950s, when people were first applying semivolatile organochlorine pesticides (e.g., lindane, DDT) on fields and polychlorinated biphenyls (PCBs) in transformers, that anyone would have anticipated that these SVOCs would rapidly be distributed in the atmosphere to ultimately accumulate in arctic organisms. But they did! As the adjective “semivolatile” implies, SVOCs do not substantially volatilize on their own, and thus the fact that they are able to so efficiently disseminate in the atmosphere, where they can subsequently pollute pristine, arctic atmospheres far away from source zones, is largely attributable to their gas/ particle partitioning behavior (Bidleman 1988). Gas/particle partitioning also plays a pivotal role in atmospheric photochemical transformation processes, because chemicals sorbed to particles usually undergo different photochemical reactions than when they are a vapor. These transformations are crucial to better understand, because they directly influence creation and destruction of organic contaminants, as well as the formation of secondary organic aerosols (SOAs). For instance, polyaromatic hydrocarbons (PAHs) are less toxic than some nitro-PAHs (Zielinska 2005), which can be formed from atmospheric PAHs and NO2. The presence of SOA particles in the atmosphere, which are formed from a series of polymerization and partitioning processes, has been strongly linked to atmospheric visibility, climate, and climate change (Chan et al. 2005; Fuzzi et al. 2006). In this chapter, we will present many of the traditional aspects and state-of-the-art advances in gas/particle partitioning theory, with emphasis on how anthropogenic organic compounds (AOCs) sorb to airborne particulate matter, namely, aerosols. A particular focus of this chapter will be on the sorption of polar and ionizable organic molecules to aerosols, as traditional gas/particle partitioning models were developed mainly for apolar SVOCs, and thus are not necessarily suitable for many emerging AOCs of concern, such as surfactants, pharmaceuticals, polar pesticides, phototransformation products of organic contaminants, and many SOA precursors. 5.1.2. Definition of Ambient Equilibrium Gas/Particle Partitioning Given a mixture of airborne organic chemicals and particulate matter in a parcel of air, a portion of the organic chemicals will sorb to the particulate matter, and the remainder will remain as a vapor. When equilibrium is achieved,

both the particle-phase and air-phase concentrations will remain constant, even though the chemicals can freely transfer back and forth between the air and particle phases. To quantify this steady state, the equilibrium gas/particle partitioning coefficient Kip is used: Kip ðm3 =gÞ ¼

ci p ci air

ð5:1Þ

where cip is the equilibrium concentration of compound i sorbed to the particle phase (mol/g) and ci air is the equilibrium concentration of i in the air phase (mol/m3). The coefficient Kip is dependent on many particulate, compound, and environmental parameters. These include the physicochemical properties of compound i, ambient conditions such as relative humidity (RH) and temperature, the composition of the particle, accessible surface area in the case of adsorptive partitioning, and the accessible volume in the case of absorptive partitioning. Understanding in more detail how these and other factors can influence Kip values, particularly for diverse AOCs, is the primary focus of this chapter. 5.1.3. Aerosol Sources, Composition, and Size Fractions In any given ambient atmosphere, the aerosols present come in many shapes and sizes and from many different sources. Understanding the makeup, composition, and origin of atmospheric aerosols is quite a complex and broad field, and the field of aerosol science is an extensive branch of science in its own right. Here, we will focus briefly on particular aspects that are of direct relevance to ambient gas/particle partitioning. Readers interested in further details are referred to relevant textbooks as a good starting point, such as those by Hinds (1999) and Seinfeld and Pandis (2006). 5.1.3.1. Aerosol Sources. Particulate components in the atmosphere can be of primary or secondary origin. Primary refers to particles that are directly emitted into the atmosphere and secondary to particles that are formed within the atmosphere. Anthropogenic sources of primary aerosols include combustion processes (e.g., traffic, coal burning), industrial activities (e.g., smelting, mining), engineered nanoparticles, and degradation of manufactured materials. Natural primary sources include biogenically derived organic matter (pollen, spores, viruses, plant debris, resuspended humic acid, lipids, etc.), sand, dry-soil particles, natural forest fires, ocean spray, and volcanic ash. Secondary aerosols include salts that form from volatile inorganic gas precursors such as SO2, NH3, NOx, and SOA particles that formed from volatile organic compound (VOC) precursors. These VOC precursors can be anthropogenic (e.g., gasoline fumes) or natural in origin (e.g., pinene released from pine trees). Secondary aerosols are generally on the order of 1 mm or less in diameter, whereas most primary aerosols tend to be on the order of 3–10 mm or larger (Seinfeld and Pandis 2006).

INTRODUCTION

5.1.3.2. Aerosol Composition. The components of ambient aerosol can be classified into four broad categories, and several potential subcategories. The four broad categories are minerals and metal oxides, salts, carbonaceous materials, and mixed-aqueous condensates. Minerals and metal oxides are typically from primary sources (e.g., volcanoes, deserts, fossil fuel impurities). Their presence in an aerosol sample is typically determined by quantifying the specific elements present, such as by implementing inductively coupled plasma–mass spectroscopy or X-ray fluorescence techniques (Chow et al. 1994; Hueglin et al. 2005). To calculate the total mineral fraction from these data, mass fractions of elements commonly associated with minerals (i.e., Al, Mg, K, Ca, Fe, Si) are assumed to be present as oxides, and the mass fractions of these oxides are added up (Chow et al. 1994). Salts in aerosols are of both primary and secondary origin. Their mass fractions are generally quantified by extracting aerosol samples with nanopure water, or with mildly acidic water, and determining the amounts of anions present (e.g., NO3, SO42, CO32-, Na þ , NH4 þ ) with techniques such as ion exchange chromatography (e.g., Hueglin et al. 2005). The carbonaceous components of aerosol samples are commonly divided into the three major subfractions: organic carbon (OC), elemental carbon (EC), and carbonate carbon (CC). Loosely speaking, OC is all the primary organic matter and SOA present in an aerosol, EC refers to graphite-like nanostructures that form during combustion and thermal degradation processes (and are of primary origin), and CC refers to inorganic carbonates. Note that CC belongs to both the carbonaceous and salt fractions, and is more commonly associated with the latter. How best to separate, measure, and even define the EC and OC fractions has proved a challenge, and has been the focus of many international conferences and roundrobin panels [notably those described by Schauer et al. (2003) and Schmid et al. (2001)]. The vast majority of separation approaches are based on the different susceptibilities of the EC and OC fractions to solvent extraction and/or thermodegradation. Measured OC and EC fractions depend largely on the separation technique being used; thus it is best to consider reported fractions as operationally defined quantities. Arguably the most popular and reproducible approaches involve a thermooptical step (Birch and Cary 1996), where the sample is pyrolized under different temperatures and atmospheres, and the amount of evolved CO2 and CH4 is measured, while accounting for artifacts caused by the charring of OC into EC via laser absorption techniques (Schauer et al. 2003). The OC fraction per se does not represent the actual mass fraction of the organic matter (OM), but only of the organic matter’s carbon mass. The conversion of OC to OM data is not necessarily straightforward; factors ranging from 1.2 to 2 have been proposed (e.g., Mader and Pankow 2002). Two important subfractions of OM are the water-soluble organic

117

matter (WSOM) and water-insoluble organic matter (WIOM) fractions. The WSOM fractions are usually determined by extracting aerosol samples with nanopure water, followed by carbon analysis of the water extracts (e.g., Krivacsy et al. 2001; Wang et al. 2005). The WIOM fraction is then determined by mass balance (i.e. WIOM þ WSOM ¼ OM). Typically, WIOM/OM and WSOM/OM range from 30% to 70% (Gysel et al. 2004; Krivacsy et al. 2001; Wang et al. 2005; Yang et al. 2005). Other important OC subfractions are primary OM and SOA. Secondary organic aerosol contains both WIOM and WSOM components, and differentiating it quantitatively from primary OM in ambient aerosols remains a challenge. One promising emerging method is the quantification of OC/OM ratios using high-resolution time-of-flight aerosol mass spectrometry (Aiken et al. 2008). Finally, there are the mixed-aqueous condensates. Many WSOM and salt components are capable of deliquescence, which means that these fractions can condense liquid water below 100% RH (i.e., below the point where the air is saturated with water vapor). The RH at which condensation starts is referred to as the deliquescence RH. As RH increases above the deliquescence RH, the amount of particle-associated water grows exponentially. Even under moderately moist conditions (e.g., RH 70%), the mass of condensed water can be greater than the mass of the dry ambient aerosols (e.g., Khlystov et al. 2005). This process of deliquescence by ambient aerosols is one of the major cloud seeding pathways (Seinfeld and Pandis 2006). As the amount of condensed water is dependent on RH, it is best measured as a function of RH, either gravimetrically in the lab (e.g., Arp et al. 2008b) or in situ using spectrometers that can differentiate between wet and dry aerosol sizes (e.g., Khlystov et al. 2005). Differentiating these mass fractions is of direct relevance to gas/particle partitioning, as each component has totally unique sorption properties. Because of their rigid structure, minerals, metal oxides, crystallized salts, and EC are capable of adsorbing organic compounds only onto their surfaces. Because WSOM, WIOM, and the aqueous components lack this rigid structure, they can additionally absorb organic compounds into their bulk matrix. It is additionally important to consider that these various components can exist as purely separate particles in the atmosphere, but can also occur together in an internally mixed aerosol. Mixing can have various impacts on gas/ particle partitioning, such as in attenuation, which occurs when one phase blocks access to the sorption sites of another phase; or in alteration, which occurs when two phases merge to become a different phase with different sorption properties. An example of attenuation is WIOM coating a soot or mineral surface. An example of alteration is deliquesced salts and WSOM fusing to form an organic salt solution. Understanding which pure or mixed aerosol components dominate sorption in a given atmosphere is necessary for

118

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

characterizing both gas/particle partitioning and transformation behavior. For instance, compounds that are sorbed to exposed mineral surfaces would be more accessible to transformation by light and radicals than would compounds trapped deep in a soot pore or absorbed into the organic matter. Table 5.1 presents a rough overview of aerosol sorption components, their mixing phases, and typical mass fractions in different environments. In Figure 5.1, a scanning electron microscope (SEM) image is shown of urban particulate matter, indicating some of the different types of particles.

This involves the simultaneous sampling of particles and the surrounding air, followed by extractions of the sampling media to determine both cip and ci air. The second approach is the usage of inverse gas chromatography (IGC), in which the particles are collected beforehand and placed in a column that is attached to a gas chromatographic system, and Kip values are subsequently determined through chromatographic sorption experiments. Details of these two methodological approaches shall be presented here.

5.1.3.3. Aerosol Size Fractions. By convention, as particle mass is much easier to measure than particle volume, a particle’s size is classified according to its aerodynamic diameter Daer, which is the diameter a particle would have if it were a sphere and its density were 1 g/cm3. Particles with a Daer <100 nm are referred to as ultrafine; below 1 mm as submicrometer; below 2.5 mm as fine; and between 2.5 and 10 mm, as coarse. When sampling atmospheric particles, it is generally desired to collect specific particle size fractions. To achieve this, particulate matter (PM) “cutoffs” are used, and particle samples are classified according to the PM cutoff. For instance, a PM2.5 (or PM 2.5) sample denotes a sample in which particles with Daer < 2.5 mm are collected with at least 50% efficiency while particles Daer > 2.5 mm are excluded from the sample. Other common typical cutoff sizes are PM1, PM10, and TSP, were TSP is the total suspended particle. Both PM1 and PM2.5 particles are of special interest in health and epidemiological studies, as large portions of these size fractions can enter the lungs and exhibit the longest atmospheric residence times (in the order of days to weeks) (Seinfeld and Pandis 2006). However, all particle sizes are of interest for better understanding atmospheric processes (climate change, cloud formation) and the short-range transport of contaminants. The particle-bound fraction (wip) in a parcel of air can be related to Kip and the differing size fractions, using the equation

5.2.1.1. Overview of Sample-and-Extract Setups. Sampleand-extract setups have been used in closed systems, such as smog chambers (e.g., Chandramouli et al. 2003b) or attached to engine exhaust outlets (e.g., Schauer et al. 1999), but are more commonly used in open systems, such as indoor and outdoor air. The sampling step involves allowing air to pass through media that separately collect particles and air-phase chemicals. For environmental studies, air is generally collected using high-volume or low-volume air pumps, although innovative passive sampling strategies that rely on wind are being developed (e.g., Tao et al. 2007). Four basic setups based on this approach are depicted in Figure 5.2. With the filter–sorbent (FS) setup (Fig. 5.2.a), incoming particles are collected on a particle filter and the air-phase chemicals are presumed to quantitatively pass through the filter and accumulate on the downstream sorbent (Bidleman and Olney 1974). Typical filters include quartz fiber filters (QFFs), glass fiber filters (GFFs) and Teflon membrane filters, which are available in a wide variety of thicknesses and porosities (Lee and Makund 2001). Fiber filters can also contain “binding materials” to strengthen the filter. Choosing an appropriate filter is a critical decision to make, as different filters can have different sorbing properties for different types of compounds. For example, GFFs that contain an organic binder sorb PAHs much stronger than do GFFs with no organic binder (Arp et al. 2007); on the other hand, GFFs with no organic binder sorb other compounds, such as perfluorocarboxylic acids, much stronger than do GFFs with an organic binder (Arp and Goss 2008). Details on the sorption strength of various filters can be found in the literature (Arp et al. 2007; Mader and Pankow 2001). When selecting a downstream sorbent, it is crucial that the target analytes be completely captured and do not (or only negligibly) break through. The two most common sorbents are polyurethane foam (PUF) plugs and XADÔ adsorbent cartridges. After collection, the filter and sorbent are extracted separately, using laboratory-specific extraction and cleanup procedures. The FS system is inherently prone to sampling artifacts referred to as blow on and blow off. Blow on refers to airborne chemicals sorbing to the filter and collected particles, which causes the determined cip to be too high (a positive bias when determining Kip). Blow off refers particle-bound chemicals

wip ¼

c0 i p PM10 ¼ Ki p  ci air þ c0 i p 1 þ Kip PM10

ð5:2Þ

where c0 i p is the mass of particle-bound compound i per volume of air (gi m3 air ). Alternatively, a different PM size fraction can be used than PM10. Ideally, the PM fraction used should be the same as the fraction for which Ki p data is available.

5.2. MEASURING AMBIENT GAS/PARTICLE PARTITIONING There are two main approaches to measure Kip values. The first approach is referred to as “sample-and-extract.”

5.2.1. Sample-and-Extract Methods

119

Combustion, volcanoes, deserts, etc. Minor

5–17 1–18 1–2 0.1–0.2 31–75 8–32 1–7 2–8

30–50 17–72 22–37 92–100 10–18 N/A N/A N/A

5–30 4–20 16–23 1–2 N/A N/A N/A 3–10

Salts, water, soot >1 nm

VOC

Oceans

WSOM

5–30 4–20 36–64 1–3 5–36 50–84 49–85 3–10

Soot, minerals 1 nm–3 mma

Biological debris, combustion, oceans VOC

WIOM

2 > 200 2 > 200 2 > 200 2 > 200 N/A N/A N/A >1

Deliquescence salts, WSOM Salt, WSOM 1 nm–100 þ mm

Water spray

Aqueous

Permeable Ad/Absorbing Phases

Notation: N/A—not available; VOC—volatile organic compounds. a Excludes biological debris. b Mass fractions are based on dry weights and are presented as approximate ranges based on the cited references; note that the mass fractions given in this table are highly dependent on the sample’s particulate matter (PM) fraction. c Data from Hueglin et al. (2005) with maximum salts including salts þ unknown, along with Sillanpaa et al. (2005) and Yttri et al. (2007) for EC/OC; WSOM and WIOM values estimated from both WSOM: OM and WIOM ratios ranging from 30%–70% (Gysel et al. 2004; Krivacsy et al. 2001; Wang et al. 2005; Yang et al. 2005). d Estimates based on Tables 5.2 and 5.4 in Cavalli et al. (2004). e Ranges based on data from various sources (Burtscher 2005; Hildemann et al. 1991; Norbeck et al. 1998; Roth et al. 2005b; Schauer et al. 1999) with OM assumed to be WIOM. f Ranges based on data from various sources (Hildemann et al. 1991; Norbeck et al. 1998; Schauer et al. 2002b) with OM assumed to be WIOM. g Ranges based on data from various sources (Hildemann et al. 1991; Schauer et al. 2001), with OM assumed to be WIOM. h From Roth et al. (2005b).

Salt, OM 1 nm–1 mm

None

Combustion

EC

Water, WSOM 1 nm > 10

NOx, SO2, NH3, etc.

Oceans, combustion

Salt

Solid Adsorbing Phases Minerals/Metal Oxides

Mixing phases OM Size ranges 1 > 10 mm Mass Fractions (% dry weight)b Terrestrial PM 2.5c 7–12 Terrestrial PM 10c 13–21 Marine PM 1d N/A Marine PM > 1d N/A Diesel soote 8–15 Gasoline sootf N/A Wood smokeg N/A Road tunnel Aerosolsg N/A

Secondary precursors

Primary sources (examples)

Parameter

TABLE 5.1. Overview of Unique Sorbing Components Found in Aerosols

120

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

Figure 5.1. A SEM scan of a particle sample collected in Berlin, Germany, during spring 2006 indicating a large porous conglomerate of organic particles with some salts (yellow circle), a sharp-edged, smooth-surfaced salt crystal (blue circle), a pollen grain (red circle), and smooth-edged, smooth-surfaced mineral particles (green circles). Where do the contaminants sorb preferentially? (See insert for color representation of this figure.)

blowing off the filter and onto the downstream sorbent, which causes the determined cip to be too low (a negative bias when determining Kip). To minimize these artifacts the other setups depicted in Figure 5.2 were proposed. The principle of the filter–filter–sorbent (FFS) (Fig. 5.2b) system is quite similar to the FS system, as it simply contains two filters instead of one. The additional filter, referred to as the backup filter, is added to correct for blow on artifacts. If both filters have reached sorption equilibrium with the incoming air, then cip can be corrected for by subtracting the amount of sorbate on the backup filter from the front filter, and ci air can be corrected for by adding twice the amount of sorbate collected on the backup filter. Although FFS setups are recommended over FS, it is worthwhile estimating

beforehand whether equilibrium partitioning to the front and backup filters can actually be achieved under the actual sampling conditions and if corrections should be made (Arp et al. 2007; Mader and Pankow 2001; Volckens and Leith 2003b). Partially in an attempt to avoid blow on artifacts completely, the denuder-filter-sorbent (DFS) systems were developed (Fig. 5.2c). The purpose of the denuder is to strip the air of gaseous SVOCs before they land on the filter. However, as the particles on filter are exposed to SVOC-free air, blow off occurs, and a downstream sorbent is placed to correct for this. When using the DFS setup, it must be ensured that the right denuder is chosen for the target analytes, so that particles do not collect in the denuder, and that breakthrough of analytes through the denuder does not occur (Mader et al. 2001; Volckens and Leith 2003a,b). The electrostatic precipitator–sorbent system (EPS) (Fig. 5.2d) (Volckens and Leith 2003a,b) can be thought of as a filter-free FS setup. Here particles are charged and collected on a conducting surface while gaseous compounds pass by to accumulate on the downstream sorbent. One unique drawback of this approach is that certain molecules can be transformed when they pass through the precipitator, thus leading to losses of those molecules in both the particle and gas phases. 5.2.1.2. Considerations When Designing Sample-andExtract Procedures. The artifacts associated with sampleand-extract methods have been reviewed extensively (e.g., Arp et al. 2007; Arp and Goss 2008; Arp et al. 2008c; Galarneau and Bidleman 2006; Volckens and Leith 2003a,b). All methods carry unique biases that need to be accounted for when designing sample-and-extract setups, as there is no method that is suitable for all chemicals and sampling situations. Generic biases to all techniques can come from fluctuating ambient conditions, such as RH,

Figure 5.2. Simultaneous particle and air-phase chemical collectors used in sample-and-extract methods: (a) filter–sorbent; (b) filter–filter–sorbent; (c) denuder–filter–sorbent; (d) electrostatic precipitator–sorbent. (See insert for color representation of this figure.)

MEASURING AMBIENT GAS/PARTICLE PARTITIONING

temperature, PM levels, and compound concentrations, which influences partitioning to fiber filters, denuders, and particles alike. In general, it is best to sample during stable weather conditions, and use sampling strategies that minimize the influence of temperature fluctuations (Galarneau and Bidleman 2006). To minimize the influence of fluctuating ambient conditions, so-called controlled field experiments can be carried out, in which particle filters are loaded with aerosols and subsequently exposed to a stream of gas containing a constant concentration of analyte at a controlled RH and temperature (Mader and Pankow 2002). An additional consideration is that filter (or electrostatic precipitator) extracts do not distinguish between those compounds existing in particle components that are exchangeable with the air phase and those that are not. A compound class for which this issue was raised early on is PAHs (Pankow and Bidleman 1991), as these can be trapped within soot particles during combustion processes and be inaccessible to air and other environmental matrixes. Many of the initial studies extracted PAHs from filters using Soxhlet extractions with methylene chloride, which can readily remove PAHs from nonexchangeable (i.e., extremely slowly exchangeable) aerosol components (Jonker and Koelmans 2002; Jonker et al. 2005). As a result, many of the Kip values for PAHs reported in the literature may be erroneously high (Arp et al. 2008a), compared to what is occurring in the atmosphere over the 1–10-day course of an aerosol’s lifetime. Thus, in order to obtain Kip values specifically for the exchangeable components, we recommend gentler extraction procedures (e.g., air stripping, supercritical fluid extraction (Jonker et al. 2005) or IGC techniques (see below). 5.2.2. Inverse Gas Chromatography Inverse gas chromatography (IGC) uses the same equipment and principles as gas chromatography; the main difference is that IGC experiments are conducted to characterize the sorption behavior of the stationary phase, and not to characterize the injected sorbents (i.e., the column packing material itself is what is being studied). The basic design of an IGC suitable for experiments conducted at environmental conditions is shown in Figure 5.3. In IGC experiments the carrier gas (typically nitrogen or synthetic air for environmental studies) passes through a temperature-regulated water saturator to adjust the RH (Fig. 5.3b) before entering the chemical injection port (Fig. 5.3c). The carrier gas then proceeds along capillaries and enters the column packed with the sorbent of interest (Fig. 5.3d), and then finally passes through the detector (Fig. 5.3e). With this setup, both RH and temperature can be easily and separately controlled (Arp et al. 2006a; Conder and Young 1979; Dorris and Gray 1981). To determine the equilibrium sorption coefficient for the material present in the column (Ki sorbent air), the volume of carrier gas required

121

Figure 5.3. Inverse gas chromatography assembly for sorption experiments at environmental conditions, indicating (a) the carrier gas source (generally synthetic air), (b) the gas humidifier for adjustment of the relative humidity (RH), (c) chemical injection port, (d) column in a temperature-controlled water bath, and (e) chemical detector (generally a flame ionization detector is used at ambient RH).

to bring the sorbate through the column, referred to as the retention volume (Vi ret) is measured. The following equation can then be used to calculate Ki sorbent air values: Ki sorbent air ¼

Vi ret Vdead Msorbent

ð5:3Þ

Here, Vdead is the “dead volume” (i.e., empty volume), which is determined by measuring the retention volume of a nonsorptive tracer (typically methane), and Msorbent is the sorbent mass in the case of absorption or sorbent surface area in the case of adsorption. Note that ideally, instead of using Vdead, it is better to measure the retention volume of the compound of interest in the IGC when an empty column is installed, in case of substantial sorption to the capillaries, frits, or column walls. It is also advised to measure Ki sorbent air at varying carrier gas flow rates and injection volume of sorbent, to test whether sorption equilibrium is reached and if sorption is occurring in the linear part of the isotherm respectively (see Section 5.3.1) (Conder and Young 1979; Roth et al. 2005b). Finally, in the case of peak skewing (which is common with IGC techniques due to heterogeneities in the sorbent and column packing), column breakthrough is best quantified using the first moment of the sorbate peak in the detector (Conder and Young 1979). Column packing is the most critical step in designing an IGC experiment. Packing a column too loosely may result in preferential flow paths, which cause a negative Kip bias. On the other hand, packing the column too tightly will cause excess flow resistance or column blockage, and may even alter the packing material (e.g., via particle congealing). Depending on the media, it may be impractical to pack the

122

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

sorbate itself as a pure phase, and thus an additional support phase is needed. Examples of support phases include silica supports such as porous glass or Chromosorb [used for water surfaces (Dorris and Gray 1981; Roth et al. 2002)], glass beads [used with diesel soot (Roth et al. 2005b) and humic acid (Niederer et al. 2006)], and GFFs [used with terrestrial aerosol samples (Arp et al. 2008c)]. The support phase must be nonreactive and as low-sorbing as possible. When a supporting phase is completely attenuated, Ki sorbent air values can be determined as before [Eq. (5.3)]. If, on the other hand, the supporting phase also sorbs the incoming sorbate, then Ki sorbent air must be determined using the equation. Ki sorbent air ¼

Vi sorbate þ support Vsupport Msorbent

ð5:4Þ

where Vi sorbate þ support is the total retention volume and Vsupport is the retention to the support and capillaries in the system. In Arp et al. (2008c) an IGC method was developed to measure Kip values for terrestrial aerosols using surfacetreated GFFs as a support phase. Such IGC methods have advantages and disadvantages compared to sample-andextract methods. The advantages are: (1) when measuring Vsupport one inherently accounts for filter sorption artifacts, (2) RH and temperature are controlled independently of each other, (3) only partitioning to the air-exchangeable components of the aerosols occurs, and (4) the chemical data-set for which Kip values can be measured is not limited by the compounds present on ambient particles. The notable disadvantages are (1) the IGC can be used to measure only Kip values and not concentrations and (2) currently it remains infeasible to measure large SVOCs and surfactants at ambient conditions and environmentally relevant concentrations (Arp et al. 2008a; Arp and Goss 2009a).

5.3.1. Basic Partitioning Theory The chemical potential of a substance in the ideal gaseous state [mi(g)] and in a sorbed state [mi(sorbed)] can be expressed as in the following set of equations: mi ðgÞ ¼ moi þ RT ln

mi ðsorbedÞ ¼ moi þ RT lnðci sorbent  Xi sorbent Þ

Since the mid-1980s or so, the controlling gas/particle partitioning mechanisms relevant for ambient particles have been under debate, and correspondingly so have Kip predictive models. Thus, in the scientific literature, one currently encounters a large variety of partitioning models and postulates. In this section, the traditional models that were developed for apolar SVOCs are presented. Models developed for polar, ionizable and surfactant molecules will be presented in Sections 5.5 and 5.6. Before doing so, we will introduce basic partitioning theory at low sorbate concentrations (i.e., infinite dilution), as this theory forms the basis for all partitioning models.

ð5:5Þ ð5:6Þ

where moi is the chemical potential of species i in its pure liquid state (i.e., the reference chemical potential), R is the ideal gas constant, T is temperature in kelvins, pi is the partial pressure of species i in the gas phase, p*i L is the saturated subcooled liquid vapor pressure of i (Pa), ci sorbent is the activity coefficient of i in the sorbent (unitless), and Xi sorbent is the mole fraction of i relative to the sorbent (moli mol1 sorbent ) [for more information, see Schwarzenbach et al. (2003)]. Partitioning equilibrium between two phases implies that the chemical potential of i in both phases is equal, that is mi(g) ¼ mi(sorbed), and therefore the equilibrium condition is defined as pi ¼ ci sorbent  Xi sorbent  p*i L

ð5:7Þ

This equation can also be expressed in terms of volumetric air concentration at equilibrium by dividing by RT: ci air ðmol=m3 Þ ¼

pi c  Xi sorbent  p*i L ¼ i sorbent RT RT

ð5:8Þ

In the case of absorption, Xi sorbent can be converted to a mass concentration by dividing by the molecular weight of the sorbent, MWsorbent (gsorbent=molsorbent): ci sorbent ðmol=gÞ ¼

5.3. TRADITIONAL GAS/PARTICLE PARTITIONING MODELS FOR APOLAR SEMIVOLATILE ORGANIC COMPOUNDS

pi p*iL

Xi sorbent MWsorbent

ð5:9Þ

The absorptive partitioning constant Ki sorbent, air can thus be expressed in terms of either these concentrations or ci sorbent and p*iL : Ki sorbent;air ðm3air =gsorbent Þ ¼

ci sorbent RT ¼ ci air MWsorbent  ci sorbent  p*i L ð5:10Þ

An analogous approach can be taken for adsorptive partitioning. When we consider that there is just one type of adsorption site available, the adsorbed concentration at equilibrium csurface can be expressed as csurface ¼ Xi surface  Ns

ð5:11Þ

TRADITIONAL GAS/PARTICLE PARTITIONING MODELS FOR APOLAR SEMIVOLATILE ORGANIC COMPOUNDS

where Xi surface is the mole fraction of i per mole of suface site (moli/molsurface sites) and Ns is the surface site density (molsurface sites =m2surface ). The equilibrium adsorption constant, Ki surface, air (m3air =m2surface ), is therefore Ki surface; air

ci surface RT Ns ¼ ¼ ci air ci surface  p*i L

ð5:12Þ

Note from Equations (5.10) and (5.12) that the only parameter that is both compound- and sorbent-specific is the activity coefficient (ci sorbent or ci surface). Elucidating sorbent and sorbate parameters that govern these activity coefficients has proved challenging. However, one special case exists, and that is for sorbent–sorbate pairs when the sorbate interacts with the sorbent similarly to how the sorbate interacts with itself (e.g., nonane absorbed in decane would lead to similar interactions with nonane absorbed in nonane); in this case, ci sorbent  1. This ideal case is commonly referred to as Raoult’s law. Raoult’s law generally does not apply when the sorbent and sorbing phases have markedly different properties (e.g., with an apolar sorbate and polar sorbent). The sorbent and sorbate identities are not the only variables to influence ci sorbent and ci surface. They are often dependent on the sorbate concentration. This dependence is referred to as nonlinear partitioning. Nonlinear partitioning behavior can take many forms, from ci decreasing with increasing sorbate concentration, to ci increasing with sorbate concentration, to more complex changes. However, at low sorbate concentration ranges, typical for environmental concentrations of AOCs, ci sorbent and ci surface values are typically constant. However, it should be emphasized that under special environmental conditions nonlinear partitioning can occur, such as at high concentrations near contaminant source zones (like chemical spills), or when dealing with very heterogeneous sorbing phases that contain competing sorptive components. 5.3.2. The Junge–Pankow Model The first gas/particle partitioning model to be widely adopted was the Junge–Pankow model (Pankow 1987). In essence, the model assumed that partitioning of apolar SVOCs to aerosols was occurring with equal affinity to the total surface of the aerosol. Thus, to parameterize this model, it was necessary to know the amount of aerosol surface area per cubic meter of air, q (m2aerosol =m3air ) and the total suspended particle concentration TSP (gaerosol =m3air ). The model can be written as follows: Ki p ðm =gÞ ¼ Ki surface; air ðm =m Þ  q  TSP 3

3

2

ð5:13Þ

Substituting Equation (5.12) into Equation (5.13) above and taking the logarithm of both sides gives

log Ki p ðm3 =gÞ ¼ log p*i L log ci surface þ log ðNS RT q=TSPÞ

123

ð5:14Þ

The original Junge–Pankow model includes extra parameters to empirically estimate ci surface, which are not shown here (Pankow 1987). Indeed, not only ci surface but also many of the parameters in this equation are difficult to measure or model, thus for practical purposes a simplified version of this equation was recommended for use, and became quite popular: log Ki p ðm3 =gÞ ¼ m log p*i L þ b

ð5:15Þ

Here, m and are b are fitted linear regression constants, with m representing any proportionalities between p*i L and ci surface, and b representing any spillover from this proportionality and the sorbate independent terms [i.e., log (NSRT q/TSP)]. Equations like Equation (5.15), where an unknown log K value is linearly related to a known log K value (here log p*i L , which is commonly available for many compounds) are commonly referred to as single-parameter linear free-energy relationships (SP-LFERs). The term single parameter linear refers to the fitting of the unknown m and b values by a singleparameter linear regression, and free energy refers to the log K values, which are directly proportional to the corresponding free energy of phase transfer, DG (i.e., DG ¼ ln K/RT þ constant). An early, simple, version of this SPLFER that was commonly used in fate models was Ki p ¼ ð6  106 Þ=p*i L (Mackay et al. 1986). 5.3.3. Pankow Absorption–Adsorption Model In the early 1990s there were an increasing number of studies reporting correlations between the total organic carbon (TOC) content in aerosols and Kip values for apolar SVOCs. Thus, the hypothesis emerged that OM absorption might be more important than surface adsorption. Pankow (1994b) showed, using derivations from basic partitioning theory, that even if absorption were the dominating mechanism, the resulting SP-LFER would be similar, as can be derived as follows. Assuming that the entire weight fraction OM in the aerosol, fOM (gOM/gaerosol) is available for absorption and is the exclusive sorbing phase, then Ki p ðm3 =gÞ ¼ fOM  KOM

ð5:16Þ

Thus, applying the general equation for absorptive partitioning [Eq. (5.10)] for the OM phase, we obtain Ki OM ðm3 =gÞ ¼

RT MWOM  ci OM  p*i L

ð5:17Þ

124

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

Substituting this into Equation (5.16) and taking the logarithm gives

5.3.6. Criticisms and Applicability of Traditional Models and SP-LFERs

fOM  RT MWsorbent ð5:18Þ

The models presented in Sections 5.3.2–5.3.5 make several simplifying assumptions (e.g., that the homogeneity and availability of OM and EC are similar across all ambient aerosols samples), and their appropriateness for ambient aerosols has been validated only for low-polarity SVOCs (PAHs, PCBs, PBDE, PCDD/Fs). Therefore, these models are open to some scrutiny and criticism, especially when applied to a broad variety of compounds and particle types. In Section 5.1 we discussed how the OM fraction of ambient aerosols consists of WIOM and mixed WSOM–aqueous domains. By the very (inherent) nature of these domains, they must exhibit widely differing sorption properties (Arp et al. 2008b; Erdakos and Pankow 2004; Erdakos et al. 2006; Griffin et al. 2003; Pun et al. 2002), one favoring waterinsoluble molecules; the other water-soluble molecules. Accordingly, organic compounds unique to these individual fractions have been extracted (e.g., Ding et al. 2008; Yang et al. 2005). Thus, simply because of OM heterogeneity, there is no mechanistic basis for correlations between log Kip and fOM (i.e., the sum of fWIOM and fWSOM) to be consistent for all aerosols and compounds classes, despite its practicality. More details on the differentiating sorption behavior of WIOM and WSOM are presented in Sections 5.5 and 5.6. The assumption that adsorption controls gas/particle partitioning processes is also not robust enough to cover all compounds and particle types; for instance, it is unlikely that amorphous and liquid-like aerosol OM phases in the atmosphere form and sorb by adsorptive processes only. The “EC þ OC” model has so far been validated only for PAHs (Lohmann and Lammel 2004). Polycyclic aromatic hydrocarbons are a special type of SVOC, as they are formed and associate with soot/EC particles during combustion processes (Hart et al. 1993); thus, Kip models developed for PAHs are not necessarily applicable to other compound classes (Arp et al. 2008a). Validations of the EC þ OC model for PAHs in the literature have also been criticized because the Ki EC values used to derive them are highly elusive and variable (Galarneau et al. 2006). An important critical aspect about SP-LFERs in general is that the regression coefficients m and b are both sorbate- and sorbent-specific, because of their dependence on ci (Goss and Schwarzenbach 1998, 2001). To illustrate the sorbent-phase dependence, we note that Ki sorbent; air p*i L SP-LFER m values [Eq. (5.15)] for alkanes absorbed in octanol, ethylene glycol, and water are 0.92, 0.55, and 0.21, respectively. To illustrate sorbate dependence for nonpolar compounds, PCBs and alkanes sorbed in octanol give Ki oa p*i L SP-LFER m values are 0.67 and 0.92, respectively (Goss and Schwarzenbach 1998). A similar analogy can be made for Ki sorbent, air  Ki oa SP-LFERs, as different organic solvents can have widely different m values for diverse sorbents and

log Ki p ðm3 =gÞ ¼ log p*iL logci OM þ log

Ignoring the role that varying sorbate–sorbent interactions can play on the ci OM term as before, one can derive the same SP-LFER as before [Eq. (5.15)], although m and b are now functions of different parameters. 5.3.4. The Octanol Absorptive Model Finizio et al. (1997) expanded the previous absorption model by saying that not only is Kip proportional to fOM, but that absorption into aerosol OM was proportional to absorption in pure octanol, that is Ki OM ¼ ao  Ki oa

ð5:19Þ

where Ki oa is the octanol–air partition coefficient and ao is some proportionality factor. Substituting Equation (5.19) into Equation (5.16) and taking the logarithm gives log Ki p ðm3 =gÞ ¼ log Ki oa þ log fOM þ logao

ð5:20Þ

This can also be related to a SP-LFER log Ki p ðm3 =gÞ ¼ m log Ki oa þ b

ð5:21Þ

where the y intercept b accounts for log fOM þ log ao, and the slope m accounts for sorbate-dependent deviations between the proportionality of Ki OM and Ki oa. 5.3.5. The EC þ OC Model Dachs and Eisenreich (2000) reported that adsorption onto the soot or elemental carbon (EC) fraction of aerosols is the main sorption process for PAHs, and not absorption into the OM. They suggested for PAHs that a dual-phase sorption model should be used that accounts for both OM absorption and EC adsorption. This prompted many researchers (e.g., Mader and Pankow 2002; Lohmann and Lammel 2004) to use equations of the form Ki p ¼ aOM  fOM  log Ki oa þ aEC  fEC  Ki EC

ð5:22Þ

where aOM and aEC are proportionality constants, fEC is the fraction of elemental carbon in the aerosol, and Ki EC is the equilibrium partition coefficient between diesel soot and air.

TRADITIONAL GAS/PARTICLE PARTITIONING MODELS FOR APOLAR SEMIVOLATILE ORGANIC COMPOUNDS

sorbates [see Fig. 5.12 of Schwarzenbach et al. (2003)]. The robustness of a Ki sorbent, air  Ki oa SP-LFER depends on how similar the sorbent is to octanol (e.g., hexanol is very similar, while octane, water, and glyoxal are not). As there is little evidence to indicate that aerosol OM is structurally similar to octanol (e.g., aerosol OM contains diverse functional groups, heteroatoms, and aromatic structures), the assumption that the Ki oa SP-LFERs performs equally well for all sorbates should be viewed with scrutiny. It should be mentioned here that a widely held fallacy in the gas/particle partitioning literature is that SP-LFER m values should be generally near 1 if equilibrium partitioning is reached, regardless of the sorbent, sorbate, and SP-LFER, based on some early empirical evidence for SVOCs and ambient particles (e.g., Harner and Bidleman 1998; Pankow 1994b). The examples and theory given above and in the literature show that there is no thermodynamic basis to support this for all sorbents and sorbates (Goss and Schwarzenbach 1998; Pankow and Bidleman 1992). Thus, discussions in the literature based on this assumption should be viewed with skepticism. Despite SP-LFERs exhibiting limited sorbate–sorbent applicability domains, they remain highly practical and popular. They are generally suitable for apolar SVOCs, and perform sufficiently for applications in which high accuracy in predicted Kip values is not required. When dealing with a broad variety of sorbates, however, it is of interest to address which of these two SP-LFER paradigms, p*i L -based and Ki oabased, exhibits the most robust performance for Kip values. Therefore, in Figure 5.4a,b, these two types of SP-LFERs were calibrated to a diverse set of Kip data for one specific aerosol sample, covering several apolar and polar sorbates of low water solubility.

125

In this example, the p*i L SP-LFER (Fig. 5.4a) gives the weaker regression statistic (r2 ¼ 0.57) because the highly polar molecules (seen well above the regression line) exhibit different cip than do compounds with low polarity (below the regression line). This is a typical result for p*i L SP-LFER regressions when both polar and apolar molecules are used, which is why this model performs best when only apolar molecules are considered (e.g., Arp et al. 2006a; Goss and Schwarzenbach 1998). The Ki oa SP-LFERs, on the other hand, reduce this deviation from the regression line by likening the sorbing phase cip to that of octanol, which, although not a perfect surrogate, reduces the deviation of the polar and nonpolar compounds from the regression line (at least for the compound classes tended). Thus, the Ki oa SPLFER is more robust than the p*i L SP-LFER. However, as foreshadowed here in Figure 5.4c, even better performing types LFERs exist, which will be introduced later on. To conclude, of the traditional gas/particle partitioning models, calibrated Ki oa SP-LFERs are the best performing; however, when applying this model to broad sets of aerosol types and compounds, the following cautions need to accounted for: (1) the b value in Eqn. (5.22) being dependent on fOM is an arbitrary assumption with low plausibility, as it does not account for WIOM and WSOM domains; (2) Ki oa SPLFERs alone should not be used to predict Ki p values of highly water-soluble, ionizable, or surfactant molecules, as these sorbents prefer aqueous components (Section 5.6.3); (3) the Ki oa values used should be experimentally determined and be validated with as many consistency checks as possible, as Ki oa data, similar to octanol–water partitioning data, are challenging to measure and literature data can vary substantially (Renner 2002); (4) if no experimentally determined Ki oa is available, a more appropriate surrogate than

Figure 5.4. Comparison of three fitted LFERs using the same compound dataset and experimental Ki p (15 C) values for the same aerosol sample [“Duebendorf fall,” from Arp et al. (2008a)], showing (a) the SP-LFER Pankow adsorption/absorption approach [Eq. (5.15)] (n ¼ 29, RMSE ¼ 0.41), (b) the SP-LFER Ki oa approach [Eq. (5.20)], (n ¼ 51, RMSE ¼ 0.28), and (c) the PP-LFER approach [Eq. (5.25)], (n ¼ 65, RMSE ¼ 0.174). Note that the number of compounds changes for the different datasets as experimentally determined p*i L , Ki oa, or PP-LFER descriptors were not available for all 65 compounds (Arp et al. 2008a).

126

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

octanol can be used to base Kip predictions (Section 5.6.6); and (5) more accurate models exist (as will be presented below) that should be favored when high accuracy is required (as shown in Fig. 5.4c) or for screening large datasets (Section 5.6.6).

5.4. SORBENT–SORBATE INTERACTIONS INVOLVED IN PARTITIONING AND PP-LFERS 5.4.1. Overview of Sorbent-Sorbate Interactions In order to make gas/particle partitioning models that are more robust than those described in Section 5.3 for diverse organic compounds, it is necessary to look at the specific sorbent–sorbate interactions involved in partitioning, and how these influence ci. Sorbent–sorbate interactions can be classified as specific and nonspecific, which are synonymous for polar and apolar interactions. Nonspecific interactions are called such because they do not occur between specific locations of the molecular surfaces of the sorbent and sorbate. An important type of nonspecific interactions are dispersive interactions (i.e., London interactions, which are a type of van der Waals interaction), which occur between random and temporary electron-rich and electron-poor areas that arise from random electron delocalization. These dispersive interactions occur for all molecular surfaces, and can vary vastly in strength (e.g., from Teflon to activated carbon). An example of dispersive interactions between two butane molecules is depicted in Figure 5.5a. An additional nonspecific interaction is referred to as cavity formation, depicted in Figure 5.5b, which occurs in the case of absorptive partitioning only. When

a sorbate enters a sorbent, the sorbent–sorbent interactions must be broken to accommodate a cavity for the sorbate to occupy. The larger the sorbate, the larger the cavity will have to be, and thus the more energy is required. Similarly, the stronger the sorbent–sorbent interaction, the more energy is required to form such a cavity. Specific interactions occur between specific sites of the sorbent and sorbate’s molecular surface. An example is a hydrogen-bond, which occurs between a permanent electron donor region of one molecule and permanent electron acceptor region of another molecule. Permanent electron donor/acceptor regions can arise from covalently bonded atoms of substantially different electronegativity. As a classic example, the hydroxyl functional group contains a highly electronegative oxygen atom (electronegativity ¼ 3.44) bonded to a weakly electronegative hydrogen atom (electronegativity ¼ 2.20), this causes the oxygen to have an electron donor region and the hydrogen, an electron acceptor region. Alternatively, an ether group contains a substantial electron donor region on the oxygen, but no substantial electron acceptor region is present on the neighboring carbons. Two butanol molecules forming two simultaneous specific interactions with each other are illustrated in Figure 5.6a. Electron-donating regions can also arise from well-structured molecular orbitals. For instance the p-electron rings of aromatic molecules are negatively charged, and can thus interact with the positive electron acceptor regions of a neighboring molecule (Fig. 5.6b). 5.4.2. Describing Absorptive Partitioning with PP-LFERs The overall free energy of phase transfer during partitioning is the sum of the individual free energies for the nonspecific and specific interactions that are occurring, that is X X DGi sorbent=air ¼ DGnonspecific DGspecific i sorbent=air þ i sorbent=air ð5:23Þ

Figure 5.5. Illustration of nonspecific interactions showing (a) random dispersive interactions between temporary electron-rich and electron-poor areas and (b) cavity formation energy, in which sorbent–sorbent bonds (shown in blue) are broken in order to accommodate the sorbate. Random dispersive interactions are always thermodynamically favorable and occur between all molecules. Cavity formation is thermodynamically favorable only when the energy cost of breaking sorbent–sorbent interactions is overcompensated by sorbent–sorbate interactions (black lines). (See insert for color representation of this figure.)

where DGi sorbent=air is the free energy of partitioning, P DGnonspecific i sorbent=air is the contribution of all the nonspecific P sorbate–sorbent interactions and DGspecific i sorbent=air is the contribution of all the specific sorbate–sorbent interactions. As free energies are directly proportional to log K values (again, DG ¼ log K/2.303RT þ constant), Equation (5.23) can be equivalently written as X X logKi sorbent=air ¼ logKinonspecific logKispecific sorbent=air þ sorbent=air þ 2:303RT  constant ð5:24Þ Equation (5.24) can also be related to the sorbate and sorbent dependence of the ci sorbent. By substituting

127

SORBENT–SORBATE INTERACTIONS INVOLVED IN PARTITIONING AND PP-LFERS

Figure 5.6. Illustration of specific interactions showing (a) two butanol molecules undergoing two simultaneous specific interactions, where the O- atom acts as an electron donor (negative dipole) and the H atom as an electron acceptor (positive dipole), and (b) the p-electron ring of benzene (negative dipole) interacting with the H atom of a water molecule (positive dipole). (See insert for color representation of this figure.)

Equation (5.10) into Equation (5.24), it becomes evident that ci sorbent itself can be subdivided into individual activity coefficients, each responsible for the various sorbate–sorbent specific interactions: ci sorbent ¼ Pcnonspecific i sorbent  Pci sorbent . In order to develop a robust and practical approach to solve Equation (5.24) for all sorbent–sorbate combinations, several unique approaches have been developed [for a partial review, see Lei et al. (2008)]. Of these approaches, Cramer (1980), followed by others (Abraham et al. 1994a; Taft et al. 1985), found that the partition properties of a given compound in any absorptive partition system can be fully characterized by five molecular descriptors. This was later corroborated by quantum chemical modeling (Zissimos et al. 2002). The five-molecular-descriptor equation that we recommend for general use in environmental systems is (Goss 2005) log Ki sorbent=air ¼ s  Si þ a  Ai þ b  Bi þ l  Li þ v  Vi þ c ð5:25Þ where Si, Ai, Bi, Li and Vi are the sorbate–specific Abraham descriptors for the polarizability/dipolarizability, electron acceptor (i.e. H-bond donor) capability, electron donor (H-bond acceptor) capability, logarithm of the hexadecane/air partition coefficient, and McGowan volume, respectively. Corresponding to the sorbate-specific descriptors are the sorbent-specific descriptors s, a, b, l, and v, along with the fitting constant, c. Unlike the SP-LFER above, log K values of an unknown partition system are linearly related to several known descriptor values and not just one; therefore, equations such as (5.25) are referred to as Polyparameter linear free-energy relationships (PP-LFERs). In this PP-LFER, the lLi and vVi account collectively for the nonspecific contributions of dispersive interactions and cavity formation P ( logKinonspecific i, and bBi account collectsorbent ), and sSi, aAP ively for specific interactions ( logKispecific sorbent ). The fitting constant c depends on contributions unrelated to sorbent– sorbate interactions, such as the units of Ki sorbent, the purity of the sorbent, and the heterogenity of sorption components.

The sorbate-specific descriptors in Equation (5.25) can be measured independently in the lab, as outlined in Abraham et al. (2004), except for Vi, which can be calculated on the basis of molecular structure after Abraham and McGowan (1987). Large tables of sorbate descriptors can be found in the literature (Abraham 1993a; Abraham et al. 1994a; Abraham et al. 1994b; Abraham and Al-Hussaini 2001, 2005; Arp et al. 2008a; Goss et al. 2008; Mintz et al. 2007; Tulp et al. 2008). Example sorbate descriptors for selected SVOCs are included in Table 5.5 at the end of this chapter. The sorbent-specific descriptors are determined via multiparameter linear regression with known experimental log Ki sorbent values and known sorbate descriptors (e.g., Goss and Schwarzenbach 2001; Goss et al. 2005). It should be noted here for readers familiar with these PP-LFERs, that Equation (5.25) has some advantages over more traditional PP-LFER equations found in the literature; it does not require the use of the descriptor Ei (i.e., excess molar refraction), whose estimation for solid compounds introduces additional error (Atapattu and Poole 2008), and it works for partitioning between two condensed phases and for gas and condensed phases alike [for more details, see Goss (2005)].

5.4.3. Describing Adsorptive Partitioning with PP-LFERs Analogously to the case of absorptive partitioning, the logarithm of adsorptive partitioning constants, log Ki surface/air, can be subdivided into specific and nonspecific interactions: logKi surface=air ¼

X

logKinonspecific þ surface

X

þ 2:303RT  constant

logKispecific surface ð5:26Þ

Examples of such interactions are illustrated in Figure 5.7. The general PP-LFER to describe adsorption is (Goss 2004; Roth et al. 2002)

128

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

Figure 5.7. Examples of adsorptive interactions at surfaces, showing (a) nonspecific dispersive interactions, (b) a specific interaction where the surface is the electron donor and butanol is the electron acceptor, and (c) the surface hydroxyl group engaging in two simultaneous specific interactions with a water molecule. (See insert for color representation of this figure.)

logKi surface=air ðm3 =m2 Þ ¼ 0:135Li 

qffiffiffiffiffiffiffiffiffiffi cvdW surf þ5:11Bi EAsurf

þ3:60Ai EDsurf 8:47 where the adsorbent-specific descriptors

pffiffiffiffiffiffiffi vdW csurf

ð5:27Þ

, EAsurf, and

EDsurf represent the adsorbents’ dispersive force capability, electron-accepting capability, and electron- donating capapffiffiffiffiffiffiffi bility, respectively. The “0.135Li cvdW ” term accounts for surf P nonspecific log Ki surface (the vdW stands for van der Waals forces), and “5.11BiEAsurf þ 3.60AiEDsurf” collectively account P for log Kispecific surface. The constant 8.47 in Equation (5.27) is derived from de Boer for the standard state of adsorption (assuming the molecules appearing on the surface to be due to random collisions with the surface without any intermolecular forces) (de Boer 1968; Roth et al. 2002). The coefficients in front of EAsurf and EDsurf were derived such that EAsurf and EDsurf could equal 1 for the water surface at 15 C, thus facilitating comparisons of the polarity of various surfaces with that of the water surface (Roth et al. 2002). Note, in contrast for the absorption PP-LFER [Eq. (5.25)], that the adsorption PP-LFER does not require a cavity formation term (as this is irrelavent), nor, interestingly, does it require an Si descriptor (Goss et al. 2005). Although this model has been found to apply to a large number of compounds (Arp et al. 2006a) and adsorbents (Goss 2004), there is one identified outlying compound class: highly fluorinated compounds. For these compounds, a slightly different adsorption PP-LFER is needed, due to the failure of Li to describe the dispersive interaction for these compounds (Arp et al. 2006b). 5.4.4. Other Partitioning Models that Account for Sorbate-Sorbent Diversity Equation (4.3) is not the only PP-LFER that can be used to quantify specific and nonspecific sorbent–sorbate interactions; indeed, a variety of alternative and additional

descriptors can be used (Goss 2005; Liu and Oberg 2009). A particular shortcoming of all PP-LFERs is the statistical correlation between parameters used to describe dispersive interactions and cavity formation, such as the Li and Vi parameters (Goss and Bronner 2006), as both nonspecific interactions are highly dependent on the sorbate’s volume. For certain applications, it may be worthwhile or even necessary to use alternative descriptors. Saying this, however, we favor the PP-LFER in Equation (5.25), especially as a starting point, due to its proven robustness for a large variety of compounds and sorbing phases. Further, the thermodynamic basis and shortcoming of most of the descriptors are heavily documented and well understood and are established using several consistency checks (e.g., Abraham 1993b; Arey et al. 2005; Goss and Bronner 2006). Thus, when outliers from this PP-LFER occur, a sound mechanistic explanation for the outlying behavior is thereby facilitated, and novel insight into the occuring partitioning mechanism may result (e.g., Goss and Bronner 2006). Besides PP-LFER approaches, several other modeling approaches attempt to explicitly account for diverse specific and nonspecific interactions [see the review by Lei et al. (2008)], although only a few have been applied to gas/particle partitioning. Variations of UNIFAC (the universal functional activity coefficient), which correlates ci and p*i L with molecular fragment contributions, are popular for p*i L and water activity coefficients (Asher et al. 2002; Chandramouli et al. 2003a; Chang and Pankow 2006; Clegg et al. 2008a, 2008b). Quantum chemical approaches are becoming increasingly popular for predicting partitioning constants, such as the commercial software COSMOtherm (COSMOlogic GmbH, Leverkusen, Germany), which uses density functional quantum chemical continuum solvation calculations with statistical thermodynamics to determine activity coefficients (Eckert and Klamt 2002, 2005). Another popular computer program that has been found to give good predictions for diverse compound classes is SPARC (scalable processor architecture performs automated reasoning in chemistry), and is a free, online Web application (http://

129

PARTITIONING TO INDIVIDUAL AEROSOL COMPONENTS

ibmlc2.chem.uga.edu/sparc/) that explicitly calculates sorbate–sorbent interactions by using various empirical molecular descriptors that are derived from molecular structure (Hilal et al. 2003, 2004). Particular advantages of COSMOtherm and SPARC are that they can allow for any sorbent and sorbate as input, as well as account for the influence of temperature.

partitioning behavior of organic compounds to these individual phases are discussed. In Sections 5.6 and 5.7, partitioning to mixed-particle samples are discussed. To aid in this discussion it is useful to provide PP-LFERs for these individual phases, as they can be used as predictive tools. Further, these PP-LFERs inherently provide insight on the role of specific and nonspecific interactions in these diverse phases.

5.4.5. Temperature Dependence of Partitioning As with any other equilibrium constant, the temperature dependence of Ki sorbent/air can be expressed using the van’t Hoff equation, which has the following solution when the enthalpy of partitioning between a sorbent and the air (DHi sorbent/air) is assumed to be temperature-independent: log Ki sorbent=air ¼

DHi sorbent=air þ RTa 1 þ constant ð5:28Þ T 2:303R

Here, Ta is the average temperature of the temperature range of interest [note that the term DHi sorbent/air þ RTa is the internal energy of partitioning] (Atkinson and Curthoys 1978)]; DHi sorbent/air is dependent on the sorbate’s enthalpy of vaporization (DvapHi) and the enthalpy of sorbate–sorbent interactions in the condensed phase (HiEsorbent ): DHi sorbent=air ðkJ=molÞ ¼ HiEsorbent Dvap Hi

ð5:29Þ

Ideally, DHi sorbent/air should be experimentally derived using Ki sorbent/air values covering an ambient temperature range. When this is not possible, the following empirical estimation techniques may be appropriate. For absorptive partitioning, it is often (but not always) the case that HiEsorbent is negligible, and thus DHi sorbent/air  DvapHi. For predicting the enthalpy of adsorptive partitioning (DHi surface/air), HiEsorbent can no longer be considered negligible. The following empirical relationship is recommended, which has been validated for a diverse dataset of compounds and sorbents (Arp et al. 2006a; Goss 2004). DHi surface=air ðkJ=molÞ ¼ 9:83:log Ki surface=air ð15 CÞ90:5 ðn ¼ 182; r2 ¼ 0:89Þ

ð5:30Þ

5.5.1. Pure Water Condensed water, present in clouds, fog, and rain, is the most abundant condensed phase in the atmosphere. Atmospheric water droplets are not necessarily “pure” but often contain several organic compounds and ions. Here sorption to pure water droplets is discussed, and the influence of dissolved ions and organic cosolvents to water droplet sorption is discussed in Section 5.7.1. Pure water droplets can both adsorb and absorb organic chemicals. In general, large apolar compounds and surfactants prefer to adsorb to the surface of the water droplet, while polar and ionic compounds prefer to absorb into the bulk water matrix. Whether adsorption or absorption dominates depends on the compound and the size of the water droplet. Regarding adsorption, Roth et al. (2002) derived the following adsorptive PP-LFER for the water surface [Eq. (5.27)], in which EAsurf and EDsurf were chosen to be pffiffiffiffiffiffiffi 1 at 15 C as a reference state, and cvdW was calibrated at surf 4.69: log Ki water surf=air ðm3 =m2 ; 15 CÞ ¼ 0:635  Li þ 3:60  Ai þ 5:11  Bi 8:47 ðn ¼ 60; r2 ¼ 0:932Þ ð5:31Þ The PP-LFER for absorptive partitioning [Eq. (5.25)] in the water bulk matrix, Ki wa (Goss 2005) is log Ki wa ðm3air =m3water ; 25 CÞ ¼ 2:07  Si þ 3:67  Ai þ 4:87  Bi þ 0:48  Li 2:55  Vi 0:59 ðn ¼ 390; r2 ¼ 0:997Þ

Other estimation techniques for specific sorbents will be presented later on in this chapter.

5.5. PARTITIONING TO INDIVIDUAL AEROSOL COMPONENTS In Table 5.1, six types of sorption components found in the aerosols were introduced: the aqueous component, salts, minerals, EC, WIOM, and WSOM. In this section, the

ð5:32Þ Note that normally in the literature the term Ki aw is used, which is the inverse of Ki wa; Ki aw is also commonly referred to as the “dimensionless” Henry’s law constant, named after William Henry, who first studied the influence of pressure and temperature on water/air partitioning. The cavity formation term (v) in Equation (5.32) (of 2.55) is more negative than those of other organic liquids [e.g., see Goss (2005)], indicating that it costs a lot more energy for

130

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

a solvent to make a cavity in water than a cavity in organic liquids, which is attributable to the strong intermolecular H-bonding of water molecules. Large apolar molecules, or apolar moieties of surfactants, cannot overcome this cavity energy and, thus, preferentially adsorb on the water surface. However, the ability of water to make strong H-bonds with polar organic molecules (as indicated by the relatively larger a and b values) overcompensates this high cavity formation energy, and makes water a good solvent for small polar molecules (e.g., methanol is totally miscible). Several organic compounds can additionally ionize in the water phase. The degree of this ionization depends on the compound’s acid dissociation constant in water (pKi a), and the aqueous-phase pH. For the water/air partitioning of ionizable compounds, the distribution ratio of both the neutral and ionic form of the compound (Di wa), can be calculated as follows Di wa

Ki wa ¼ ai a

Di wa ¼

Ki wa 1ai a

for organic acids for organic bases

ð5:33Þ ð5:34Þ

where ai a is the fraction of the compound in the protonated form: ai a ¼ ð1 þ 10

pHpKi

a

1

Þ

DHi wa ðkJ=molÞ ¼2:84  Si þ 32:0  Ai þ 41:8  Bi ð5:36Þ

ðn ¼ 368; r ¼ 0:964Þ 2

Although the vast majority of compounds exhibit decreasing Ki wa with increasing temperature (i.e., positive DHi wa), many compounds can exhibit the opposite trend or negligible temperature dependence (Arp and Schmidt 2004), which depends mainly on the value of HiEwater [which can be either positive or negative (Schwarzenbach et al. 2003)] relative to DvapHi (which is always positive, i.e. endothermic). To quantify whether absorption or adsorption dominates for a given compound and water particles of a given size, the enrichment factor EFi can be calculated as follows EFi ¼

Ki water surf=air  ðA=VÞ þ Ki wa Ki wa

wi bulk water ¼

ci water Di wa  LWC ¼ ci air þ ci water 1 þ Di wa  LWC

ð5:37Þ

where A is the surface area of the water droplet and V is the volume of the droplet. From this equation, an EF < 2

ð5:38Þ

where LWC is the dimensionless liquid water content in a unit of air (m3water =m3air ). 5.5.2. Snow and Ice Snow and ice are important atmospheric sinks of organic contaminants in arctic regions and high altitudes. Roth et al. (2004) conducted low-temperature IGC sorption experiments to snow, and calibrated the following “adsorbentlike” PP-LFER: log Ki snow surf=air ðm3 =m2 ; 6:8 CÞ ¼ 0:635  Li þ 3:38  Bi þ 3:53  Ai 6:85 ðn ¼ 57; r2 ¼ 0:90Þ

ð5:35Þ

The temperature dependence of adsorptive partitioning can be estimated using Equation (5.30). To account for the temperature dependence of absorptive partitioning in water, Mintz et al. (2007) derived

þ 6:35  Li 9:91  Ei þ 13:31

indicates that absorption is dominating and > 2, that adsorption is dominating. The equations outlined here are all that are necessary to estimate how much of any given AOC is distributed between air and condensed pure water droplets in a given parcel of air. For instance, if absorption dominates (EF < 2), one can calculate the fraction present in the water droplets (wi bulk water) analogously to Equation (5.2)

ð5:39Þ This above PP-LFER is referred to as “adsorbent-like” because the fitted constant term was significantly different from the universal adsorption constant of 8.47. This could have been due to incorrect determination of the snow surface area or an additional partitioning mechanism, such as to a quasi-liquid layer. Nevertheless, despite this ambiguity, Equation (5.39) is considered a useful predicting tool, although it should be kept in mind that the sorption properties of a snow surface may change depending on the snow’s degree of crystallization (Roth et al. 2004). 5.5.3. Minerals and Metal Oxides Adsorption to minerals and metal oxides is highly dependent on RH. The reason is that water layers accumulate on mineral and metal oxide surfaces as RH increases. As the RH approaches 100%, the water layer thickens exponentially, until adsorption to the mineral surface is ultimately the same as that onto the water surface (Goss 2004). This is depicted in Figure 5.8 using butanol as an example. This water layer only slightly alters the specific interactions, quantified with EAsurf and EDsurf; which is partly attributable to the fact that the EAsurf and EDsurf values of most minerals at low RH are already similar to that of the bulk water surface (Table 5.2). The thickening water layer with

PARTITIONING TO INDIVIDUAL AEROSOL COMPONENTS

131

Figure 5.8. The influence of relative humidity (RH) on adsorption to minerals and metal oxides, showing the increase of the water-layer thickness with increasing RH. (See insert for color representation of this figure.)

increasing RH lowers Ki surf/air values by substantially reducing the nonspecific interactions, which is seen in a pffiffiffiffiffiffiffi decrease of cvdW with increasing RH (Table 5.2). More surf information on partitioning to diverse mineral surfaces can be found elsewhere (Goss and Schwarzenbach 2002; Goss et al. 2003; Goss 2004).

salts found in the atmosphere, NaCl and (NH4)2SO4, sorbent descriptors at RH below the deliquescence RH can be found in Table 5.2. Interestingly, beneath the deliquescence RH pffiffiffiffiffiffiffi neither the Ki surface/air, EAsurf EDsurf, or cvdW term is surf influenced by increasing RH (Table 5.2). Partitioning to salt-rich aqueous aerosol droplets, such as a fully deliquesced salt, is discussed in more detail in Section 5.7.1.

5.5.4. Salts After water, salts are the second most abundant particle phase in the atmosphere by mass. Unlike minerals, many salts exhibit a deliquescence RH (see Section 5.1.3.2). The deliquescence RH depends on the salt, impurities present, and temperature (Seinfeld and Pandis 2006). For two abundant

5.5.5. Elemental Carbon The EC fraction of combustion aerosols has been found to vary widely, and it can be expected that numerous EC structures exist in the atmosphere. Freshly emitted diesel particles (i.e., “diesel soot”) contain EC as the main

TABLE 5.2. Examples of Some PP-LFER Descriptors for Mineral and salt Surfaces qffiffiffiffiffiffiffiffiffiffi 2 0:5 EAsurf () EDsurf () Surface RH % cvdW surf ðmJ=m Þ

Quartz a-Al2O32

NaCla

(NH4)2SO4a

45 70 90 40 70 90 20 40 60 20 40 60

Minerals 6.79  0.11 6.04  0.09 5.31  0.08 5.38  0.08 5.02  0.08 4.84  0.08 Salts 6.41  0.09 6.30  0.09 6.19  0.08 6.63  0.08 6.55  0.08 6.47  0.07

Deliquescence RH (25 C) for NaCl is 75% and (NH4)2SO4 is 80%. [Tang and Munkelwitz (1993)]. a

Reference

1.06  0.04 0.96  0.03 0.85  0.03 1.02  0.03 0.92  0.03 0.89  0.03

0.89  0.10 0.88  0.09 0.88  0.07 1.13  0.08 1.08  0.07 1.05  0.07

Goss and Schwarzenbach (2002)

0.85  0.03 0.80  0.03 0.78  0.05 0.67  0.40 0.67  0.03 0.58  0.03

0.93  0.09 0.94  0.08 0.94  0.08 1.02  0.10 1.02  0.09 1.04  0.08

Goss et al. (2003)

Goss and Schwarzenbach (2002)

Goss et al. (2003)

132

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

component, whereas EC is a minor component of most other combustion particles compared to OM (see Table 5.1). Pure, unattenuated EC is expected to sorb similarly to graphite, pffiffiffiffiffiffiffi which has the highest known cvdW value of 10.7–11.5, and surf thus would be a very attractive adsorbent for AOCs, especially if there is a large exposed surface area. However, likely because of this strong sorption, EC surfaces become rapidly attenuated with OM during combustion processes and even further with OM and salts on immediate release into ambient air (Jacobson 2001; Johnson et al. 2005; Shiraiwa et al. 2007). As a rough indication of the degree of this attenuation, the surface area of urban aerosols [e.g., 0.2–2.2 m2/g (Roth et al., 2005a), 0.82–2.4 m2/g (Sheffield and Pankow 1994)] and even road tunnel aerosols [ 7.4 m2/g (Roth et al. 2005b)] are one or two orders of magnitude smaller than the values for diesel soot [e.g., 91 m2/g (Roth et al. 2005b)], indicating that exposed EC surfaces in ambient atmospheres could be rare. A sorption study on commercially available diesel soot with a relatively high EC content (National Institute of Standards and Technology, (NIST), standard reference material 2975, diesel particulate matter (industrial forklift)] calibrated the following PP-LFER pffiffiffiffiffiffiffi descriptors: cvdW ¼ 8.08  0.07, EAsurf ¼ 0.48  0.02, and surf EDsurf ¼ 0.75  0.06 (15 C) (Roth et al. 2005b). The fact that pffiffiffiffiffiffiffi cvdW for diesel soot is than for graphite further indicates surf that EC surfaces in diesel soot are attenuated. As these attenuating phases are strongly sorbed, they are unlikely to be displaced from EC surface by airborne organics during an atmospheric particle’s lifetime (1–10 days). Supporting this claim, experimental evidence is presented below that sorption to EC surfaces within terrestrial aerosols is likely negligible. Thus, partitioning to EC surfaces is likely occurring to a substantial degree during combustion and during the immediate release of combustion particles in the atmosphere, but only negligibly for nonfreshly emitted particles.

5.5.6. Water Soluble Organic Matter Water-soluble organic matter (WSOM) contains several polar functional groups, especially carboxylic acid, carbonyl, alcohol, and ester groups (Duarte et al. 2007, 2008). Kiss et al. (2002) reported a C:H:N:O ratio of 24:34:1:14, indicating high oxidation, and the presence of polymers and aromatic groups. The most common surrogates for WSOM in lab and theoretical studies are dicarboxylic acids and polysaccharides, as these compounds are typically found in WSOM (e.g., Koehler et al. 2006; Marcolli and Krieger 2006). Like salts, WSOM phases are hygroscopic, with deliquescence RH values ranging from 65% (e.g., malonic acid) to 99% (oxalic acid) (Koehler et al. 2006; Marcolli and Krieger 2006). Because of the very polar nature of WSOM, it likely consists mainly of secondary OM, formed by the photooxidation of volatile organic precursors (including

AOCs). To our knowledge, no systematic sorption studies of WSOM or surrogates are available in the literature. However, it can be expected, from the general low deliquescence point of many atmospheric particles (often < 50% RH) and from several laboratory and theoretical studies, that WSOM in the atmosphere is largely mixed with salts and water and therefore that partitioning of contaminants to WSOM as a nonmixed, pure phase is likely negligible compared to partitioning to mixed salt–WSOM–water phases (see Sections 5.6 and 5.7.1). 5.5.7. Water-Insoluble Organic Matter As presented in Section 5.1.2, water-insoluble organic matter (WIOM) comprises both primary and secondary OM. The primary sources can be biologically generated (e.g., pollen grains, viruses) or anthropogenically generated (e.g., combustion oil residues). Similarly, SOA components can derive from either biological or anthropogenic VOC precursors. Thus, WIOM comprises many unique phases, which exhibit differing sorption characteristics, ranging from adsorbing hard cellulose fragments to absorbing liquid oils. For partitioning to a given WIOM phase, the general absorptive or adsorptive partitioning equations apply [Eqs. (5.10) and (5.12), respectively]. For instance, for absorptive partitioning into the SOA-WIOM domain, the equation would be Ki SOA-WIOM ðm3air =gSOA-WIOM Þ ¼

RT MWSOA-WIOM  ci SOA-WIOM  p*iL ð5:40Þ

This equation for the SOA-WIOM subdomain was not chosen arbitrarily. In fact, it is fundamental for understanding and modeling SOA growth (Pankow 1994a; Seinfeld and Pankow 2003). Additionally, as will be discussed in more detail in Section 5.6, a WIOM subdomain, most likely SOAWIOM, was found to be the dominating sorption domain for low-water-soluble organic compounds to ambient terrestrial aerosols.

5.6. PARTITIONING TO MIXED PARTICLEPHASES 5.6.1. Terrestrial Aerosols Terrestrial aerosols refer to airborne particles found in outdoor, terrestrial atmospheres. Thus, this broad term can refer to aerosols found in rural, urban, forest, desert, and other terrestrial environments. Because there are a large variety of different primary and secondary sources of minerals, OM, and salts in different terrestrial atmospheres, the types of particles found can vary widely. From a human and environmental health perspective, understanding the gas/

PARTITIONING TO MIXED PARTICLE-PHASES

particle behavior of terrestrial aerosols is important, as it is terrestrial aerosols that humans breathe and it is in terrestrial environments that most contaminant AOCs are initially released into the atmosphere. A major criticism of previous ambient gas/particle partitioning models presented in Section 5.3 was that they did not explicitly account for the presence of WIOM and mixed WSOM–aqueous domains. The importance of diverse partitioning into these two domains in ambient atmospheres was originally recognized by researchers interested in SOA growth and hygroscopicity (Griffin et al. 2003; Kleeman et al. 1999; Pun et al. 2002). Since then, several conceptual and UNIFAC-based models have been developed to account for sorption into these two phases more robustly (e.g., Clegg et al. 2008a,b; Erdakos and Pankow 2004; Erdakos et al. 2006; Pankow 2003; Zuend et al. 2008). However, these models do not explicitly account for overall gas/particle partitioning to ambient terrestrial aerosols, in which other particle domains than just WSOM and WIOM are present (EC, minerals, etc.). To address this, the authors of this chapter recently conducted a series of IGC experiments to isolate the dominating sorption mechanism of terrestrial aerosols. This study involved the experimental determination of over 1300 Kip values covering diverse ranges of RH (50%–90%), temperature (15 C–55 C), aerosol samples (covering the four seasons, desert dust, rural, coastal, urban, and suburban particles), and compound classes (polar, apolar, ionizable) (Arp et al. 2008a,b; Arp and Goss 2009a,b), the main results of which are summarized below. 5.6.1.1. Identification of Water-Soluble and Insoluble Sorption Domains. Essentially all types of pure phases and mixed phases can be found in terrestrial aerosols (see Table 5.1 and Fig. 5.1). Gotz et al. (2007) reported a gas/ particle partitioning model that accounted for the various mass fractions of the pure sorption components presented in Section 5.5, which assumed that no internal mixing of the individual components occurs, and that mixed-aqueous condensates were absent. One of the conclusions from this study was that minerals (specifically nonattenuated quartz-like surfaces) should dominate sorption of highly polar compounds at low RH, and that because of this Kip values of polar compounds should decrease with increasing RH (as in Fig. 5.8). However, our IGC experiments did not show this. Instead, the measured Kip values for some polar compounds and ionizable compounds actually increased with increasing RH (see Fig. 5.9), whereas for the remaining polar compounds and all the apolar compounds, no substantial change with RH was observed. These results indicate that adsorption to minerals is of little significance for polar compounds, contrary to what is expected when no mixing is occurring. The increase in Kip with increasing RH for some polar and ionizable compounds, however, indicates that the uptake of water with increasing RH (in the mixed-aqueous

133

Figure 5.9. Comparison of Ki p (15 C, PM10) data at 50% and 90% RH for an urban aerosol sample sampled in Berlin, Germany (Arp et al. 2008b).

condensates) increases the overall sorption capacity for these compounds, while it has a negligible influence on the remaining compounds. To further characterize the importance of the terrestrial aerosol’s hygroscopic components (salts and WSOM), we investigated what changes in Kip occur when the water soluble phases are removed from the aerosol sample (by extraction with water). Representative results are presented in Figure 5.10.

Figure 5.10. Comparison of log Ki p values (15 C, PM10) before and after extraction of water soluble material from the same aerosol sample shown in Figure 5.9 (Arp et al. 2008b).

134

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

As is evident from Figure 5.10, after extraction of the water soluble components, the Kip values of most compounds remained constant, with the exception of some “small polar” compounds and ionizable compounds (“small polar” compounds here refers specifically to the ones tested: ethanol, npropanol, isopropanol, n-propanoic acid, and 1,4-dioxane). Thus, from these two experiments it appears evident that two sorbing domains are present; a water-soluble, hygroscopic phase that is attractive to highly water-soluble compounds (such as small polar and ionizable compounds), and a water-insoluble, nonhygroscopic phase that is attractive to low-water-soluble compounds. 5.6.1.2. Partitioning into the Water-Insoluble Domain. To understand partitioning into the water insoluble domain, it is best to focus on Kip data at dry conditions (50% RH, and water extracted aerosols) for low-water-soluble compounds. When we compared such Kip data for nine diverse aerosol samples, it was found that the aerosol sample that gave the highest Kip values was a rural sample rich in small organic particles (collected in Aspvreten, Sweden) and the sample that gave the lowest Kip values was a mineral-rich sample (collected in Duebendorf, Switzerland, during a Sahara-sand event). Log Kip values for these two aerosol samples are plotted against those of a reference sample (for which the largest Kip dataset is available) in Figure 5.11 below, along with scanning electron microscope images of these two “extreme” samples. As is evident from Figure 5.11, the differences in the log Kip values from the reference sample and the two “extreme” samples appears to be generally systematic. The mineral-rich sample exhibits log Kip values 0.28  0.20 less than the

reference sample, and the Aspverten sample exhibits log Kip values 0.58  0.20 more than the reference sample. All other tested samples also exhibited a similar systematically deviating behavior. This general, direct proportionality in log Kip for the different aerosol samples and polar and apolar organic compounds at 50% RH indicates that the sorbing phase is similar for these different aerosols, it is only the relative amount (or mass fraction) of this sorbing phase that differs. To validate further if the dominant sorbing phase was similar in all terrestrial aerosol samples, the absorbent PPLFER equation [Eq. (5.25)] was calibrated with sample specific Kip data. As is evident from Table 5.3, the sorbent descriptors are quite similar across all terrestrial aerosol samples (although with some exceptions, such as “Duebenordf Winter’s” s and l), indicating that they all exhibit similar sorbent–sorbate interactions. Note that the predictions made with the fitted PP-LFERs are in very close agreement with experimental data (r2 > 0.95, RMSE < 0.204), indicating that the sorbent descriptors are of good quality. Having established that sorption to the water insoluble domain is similar for all tested terrestrial aerosol samples, the question arises as to what its identity is and whether it is WIOM, minerals, EC, or some mixture thereof. A close examination of the data reveals that absorption is occurring, and therefor WIOM alone is the only plausible dominating sorption phase. Evidence against contributions from adsorbing phases can be summarized as follows: (1) even after removing the water-soluble components, which could have potentially cleaned off surface-attenuating phases and made surfaces available for adsorption, sorption did not increase but stayed the same or decreased (Fig. 5.10); (2) as stated above, Kip values of certain polar compounds and

Figure 5.11. Experimental log Ki p (15 C, 50% RH, PM10) values of a terrestrial aerosol sample rich in organic particles and another that is rich in mineral particles, compared to a reference sample [“Duebendorf fall,” from Arp et al. (2008a)], along with corresponding SEM images.

135

PARTITIONING TO MIXED PARTICLE-PHASES

TABLE 5.3. Particle-Specific Aerosol Descriptors from Fitting Ki p Data (15 C, 50%RH, PM10) to the PP-LFER [Eq. (5.25)] Sample

s

l

v

b

a

c

r2

RMSE

n

Zurich Berlin Spring Berlin Wintera Duebendorf Spring Duebendorf Summer Duebendorf Falla Duebendorf Winter Aspvreten Roost Solvent Octanolb

1.09  0.15 1.01  0.09 1.38  0.18 1.14  0.11 1.09  0.11 1.19  0.12 1.63  0.15 0.95  0.09 1.45  0.13 0.66

0.75  0.06 0.78  0.03 0.63  0.09 0.61  0.04 0.66  0.04 0.66  0.05 0.51  0.06 0.64  0.03 0.60  0.05 0.91

0.35  0.15 0.51  0.09 0.98  0.28 0.69  0.12 0.71  0.12 0.73  0.16 1.51  0.19 0.49  0.09 0.86  0.15 -0.14

0.64  0.22 0.30  0.15 0.42  0.20 0.45  0.16 0.48  0.15 0.03  0.15 0.28  0.21 0.55  0.14 0.37  0.18 1.42

2.99  0.14 3.17  0.12 3.21  0.15 2.75  0.12 2.91  0.12 3.37  0.11 3.20  0.16 2.52  0.11 3.12  0.14 3.49

6.84  0.20 7.42  0.15 7.24  0.18 6.48  0.17 7.28  0.15 7.08  0.14 7.33  0.20 5.95  0.12 6.99  0.17 0.25

0.974 0.970 0.967 0.957 0.966 0.972 0.963 0.975 0.965

0.172 0.139 0.165 0.164 0.161 0.174 0.186 0.132 0.204

38 53 50 57 55 65 51 57 59

Notation:  ¼ standard deviation; r2 ¼ correlation coefficient; RMSE ¼ root mean square error; n ¼ number of data; definition of sorbent descriptors can be found in the main text. a Recommended for generic use as aerosol descriptors for terrestrial aerosols. b For comparative purposes, the sorbent descriptors for octanol are also provided (Goss 2005). Source: From Arp et al. (2008a).

ionizable compounds increase with RH (Fig. 5.9), and did not decrease or remain the same as they would if adsorption to salts and mineral surfaces dominated; (3) the sample richest in minerals (Duebendorf-Sahara) revealed much smaller Kip values than did any other sample collected (Fig. 5.11); (4) in general, if absorption dominates over adsorption then the ratios of K values for n-alkanes: cycloalkanes and of fluorotelomer alcohols: n-alcohols are <1 (Endo et al. 2008; Goss and Bronner 2006), for all nine diverse aerosol samples both tests consistently resulted in ratios <1; (6) the fitted PPLFERs coefficients listed in Table 5.3 are consistent with absorbing polymers (Arp et al. 2008a); (7) the maximum sorption capacities of most salts and many minerals are insufficient to account for measured Kip values (Roth et al. 2005a); and (8) empirical arguments listed above and elsewhere collectively indicate that EC surfaces are too attenuated to contribute substantially to partitioning (Arp et al. 2008b). Finally, we would like to highlight the improved accuracy of calibrated PP-LFERs over Koa SP-LFERs for Kip values of low-water-soluble compounds, as evident in Figure 5.4. Calibrated PP-LFERs for terrestrial aerosol result in correlations with r2 > 0.96 (Table 5.3, Fig. 5.4), whereas calibrated Koa SP-LFERs result in correlations with r2 ranges of 0.8–0.9 (e.g., Fig. 5.4). Thus, when high accuracy is needed over diverse-ranging sorbates, it is advantageous to account for diverse sorbent–sorbate interactions via PP-LFERs. A comparison of the PP-LFER descriptors for terrestrial aerosols with those of octanol (Table 5.3) provides insight as to how the sorption behavior of terrestrial aerosols and octanol differ, and thus for which set of molecules Ki oa SP-LFERs are the most inappropriate. For instance, of the specific interactions, a is similar, b is larger, and s is smaller for terrestrial particles, compared to octanol. Thus, if log Kip are being regressed with log Ki oa values for compounds with very low and very high Bi values, and/or with very low and

very high Si values, and assuming the Bi and Si values in the dataset are not statistically correlated with each other, weaker Koa SP-LFER correlations will result than if we had chemicals only with low Bi and Si values. 5.6.1.3. Simultaneous Partitioning to the Water-Soluble and Insoluble Domains. Highly water-soluble compounds (ionizable and small polar), especially under moist conditions (RH > 50%), are expected to partition favorably to the mixed-aqueous components of aerosols, as indicated by the increase in Kip with RH in Figure 5.9. To relate Kip to both the WIOM and RH-dependent, aqueous phases, assuming that these two phases do not mix with each other and thus sorb additively, an equation of the following form is needed: Ki p ¼ sorption to WIOM ð5:41Þ þ sorption to RHdependent aqueous phase For compounds that only absorb into the RH dependent aqueous phase, Equation (5.41) can be parameterized as Ki p ¼ fWIOM  Ki p WIOM þ

VwRH ðDiaw  S  Mdry Þ

ð5:42Þ

where fWIOM is the fraction of sorbing WIOM (this does not include the low-sorbing WIOM, such as wood particles and pollen grains), Kip WIOM is the WIOM-air partitioning coefficient, VwRH is the RH-dependent volume of the mixedaqueous phase, Di aw is the inverse of Di wa, and S is an empirical factor to account for any salting/cosolvency effects occurring in the mixed-aqueous phase. 5.6.1.4. Modeling Kip for Surfactants. An increasing number of substances of interest are surfactant in nature (e.g., perfluorinated surfactants, alcohol ethoxylates, alkyl

136

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

phenols). These compounds have an affinity for the water surface, thus it is important to account for these compounds adsorbing to the RH-dependent aqueous phase. Additionally, many surfactants can form dimers, aggregates, and structures such as micelles once they exceed certain concentrations, which should be taken into consideration. Thus, a way to parameterize Equation (5.41) for surfactants would be

Ki p ¼ fWIOM Ki WIOM þ

SAwRH  Dm VwRH i ws;a þ Dm Mdry i aw  Mdry ð5:43Þ

where Dm i aw is the bulk water–air phase distribution coefficient for all species (i.e., monospecies, aggregates, micelles), SAwRH (m2water surface ) is the RH-dependent water surface area, 3 2 and Dm i ws;a (mair =mwater surface ) is the water surface–air phase distribution coefficient for all species. So far, attempts to quantitatively account for the gas/particle partitioning of surfactants has proved challenging, and more research in this direction is needed (Arp and Goss 2009a). 5.6.1.5. Estimating Terrestrial Aerosol Kip Values Using PP-LFERs. If PP-LFER sorbate descriptors for the compound of interest are available, the combined fWIOMKi p WIOM term in Equations (5.42) and (5.43) can be estimated using either the Duebendorf Fall or Berlin Winter PP-LFER sorbent descriptors in Table 5.3 (or the average of both). These two sets of sorbent descriptors were calibrated for PM10 aerosols exhibiting “average” sorption behavior. To estimate the fWIOMKipWIOM term for temperatures substantially different from 15 C, DHip can be estimated from (a)

DHi p ¼ ð0:97  0:15ÞDvap Hi DHi p ¼ ð0:91  0:15ÞDHi octanol=air

0

0

-1

-1

-2

-2

-2

-3

-3

-3

-1

-4

-4 0 1 -4 -3 -2 -1 Dual phase fitted PP-LFER log K ip

ðn ¼ 101Þ ð5:45Þ

(c) 1

0

ð5:44Þ

5.6.1.6. Estimating Kip on the Basis of Molecular Structure. One practical shortcoming of the above sorption model for terrestrial aerosols is that PP-LFER descriptors, as well as

(b) Polar and apolar Small polar Ionizable 1:1 line

ðn ¼ 73Þ

These equations should give reasonable results over an ambient temperature range (i.e., from 10 C to 50 C), which is sufficient for most environmental modeling purposes. To model the VwRH =ðDSi aw  Mdry Þ term in Equation (5.42), one can obtain experimental pKa and Ki aw values for many compounds from the literature, otherwise the PPLFER for Ki aw [Eq. (5.32)] can be used. If experimentally determined Mdry, VwRH, and aerosol pH are not available, predictions can be done with assumed maximum and minimum VwRH/Mdry ratios, typically from 0.1 to 1 (Khlystov et al. 2005), and realistic pH ranges, typically from 1 to 4 (Yao et al. 2006; Zhang et al. 2007). Conveniently we found that an S value of 1 (i.e., no salt/cosolvent correction) consistently results in good predictions for water-miscible polar and ionic compounds, which implies that the mixedaqueous phase sorbs similarly to pure water (Arp et al. 2008b). The authors recommended checking the influence of the various assumptions with a subsequent sensitivity analysis. A comparison of experimental Kip values at different RH for a specific aerosol sample with PP-LFER estimated fWIOMKipWIOM and literature pKa and Ki aw values is presented in Figure 5.12.

1

1 Experimental log Kip (m3/g)

experimental DvapHi values or the enthalpy of octanol/air partitioning, DHi octanol/air (Arp et al. 2008a):

-4 -4 -3 -2 -1 0 1 Dual phase fitted PP-LFER log K ip

-4

-3 -2 -1 0 1 Dual phase fitted PP-LFER log K ip

Figure 5.12. Comparison of experimental log Ki p values (m3/g, 15 C) for the Berlin Spring aerosol sample at various relative humidity levels [(a) 50% RH, (b) 70% RH, (c) 90 RH] with predictions made using the dual-phase equation [Eq. (5.42)]. The PP-LFER approach was used to account for sorption to the water-insoluble components (fWIOMKi pWIOM) and weight measurements, a measured pH value of 3.0, and an assumed S value of 1 to account for sorption to the water-soluble components [VwRH/(Di awSMdry)] [from Arp et al. (2008b)].

PARTITIONING TO MIXED PARTICLE-PHASES

H3C

O

CH3 O

O

O

O

CH3

CH3

Figure 5.13. Surrogate for WIOM phase in ambient aerosols, proposed by Kalberer et al. (2004) [SMILES string ¼ c(c(cc1C (OC2(OC3C(¼O)C)C)OC2O3)C)c(c1)C], molecular weight ¼ 278.29 g/mol).

pKa and Ki aw values, are not available for every conceivable sorbate, thus making predictions unfeasible for many emerging (or not yet existing) compounds. To address this, we recently validated the following alternative predictive method, which requires only the molecular structure of the compound of interest as input (Arp and Goss 2009b): (1) assume that fWIOM is 0.1 (which is a typical value, see Table 5.1), (2) model KipWIOM, pKa, and Ki aw using either the quantum chemical program COSMOtherm or the SPARC online calculator (see below), and (3) estimated VwRH, Mdry, and pH ranges as above. To predict Ki p WIOM, SPARC and COSMOtherm first need to predict p*i L and ciWIOM [Eq. (5.10)]. For this, a molecular surrogate is needed to represent the WIOM phase. Out of all the suitable molecular surrogates that we tested for this purpose, the molecule depicted in Figure 5.13 resulted in the best estimations (Arp and Goss 2009b). This molecule is a proposed SOA structural unit (Kalberer et al. 2004), and thus is a good representative for WIOM’s SOA–sorbate interactions. Figure 5.14 compares experimental Kip values at different RH levels for a specific aerosol sample with COSMOtherm estimated fWIOMKipWIOM and Ki aw values. It is evident from this figure that the calibrated PP-LFER approach (Fig. 5.12)

137

works better than the noncalibrated COSMOtherm approach. However, as the COSMOtherm approach used only molecular structure as input and was not fitted, the COSMOtherm model is more robust, especially considering that the accuracy of being within an order of magnitude is sufficient for many practical applications. COSMOtherm and SPARC can also predict Kip values at different temperature. In general, we found that COSMOtherm gives better correlations to experimentally determined DHip than does SPARC (Arp and Goss 2009b). Note that it may be attractive for octanol to be used as a surrogate instead of the WIOM structure presented in Figure 5.13. However, we do not recommend doing so, as it was earlier established that WIOM exhibits different sorbent–sorbate interactions than octanol (Section 5.6.1). Further, this would mean forfeiting one of the primary advantages of using estimation software such as COSMOtherm or SPARC: that the user is not limited in terms of which molecular surrogates can be used to represent the sorbing phase. Essentially, any molecular structure can be tested; thus it is worthwhile to try and find the best-performing molecular surrogate with the broadest possible chemical application domain. This is not only because of the potential of improved predictions, but also because the process of testing different surrogates can lead to additional useful information, such as the types of sorbent–sorbate interactions involved and structural information about the sorbing phase. As an example, the correlation presented here indicates that SOA composed of aromatic moieties (Fig. 5.13) is the dominating sorption component of terrestrial aerosols for sorbents with low water solubility, which supports the hypothesis that terrestrial SOA consists largely of polymers from the methyl glyoxal pathway with incorporated aromatic moieties (Kalberer et al. 2004).

Figure 5.14. Comparison of experimental log Ki p values (m3/g, 15 C) for the Berlin Spring aerosol sample at various relative humidities [(a) 50%RH, (b) 70%RH, (c) 90%RH] with predictions made using the dual-phase equation [Eq. (5.42)] based on COSMOtherm values to determine Ki pWIOM) and weight measurements and a measured pH value 3.0 to determine sorption to the water-soluble components [VwRH =ðDSi aw  Mdry ] [from Arp et al. (2008b)].

138

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

Nevertheless, there are some current shortcomings of using the tested versions of the COSMOtherm and SPARC approaches that should be taken into account. There are compound classes whose Kip values were consistent outliers (i.e., were not predicted within an order of magnitude compared to experimental data), specifically large halogenated SVOCs using COSMOtherm and highly fluorinated compounds using certain versions of SPARC (Arp et al. 2006b, Arp and Goss 2009b). COSMOtherm has known difficulties predicting the p*i L for large multichlorinated and large multibrominated species. Thus for such compounds we recommend combining COSMOtherm-derived activity coefficients with literature-derived p*i L . One notable shortcoming with SPARC compared to COSMOtherm is that SPARC cannot account for all the differences in specific and nonspecific interactions due to 3D conformation, thus, important differences in the partitioning of stereoisomers of AOCs (such as those of hexachlorohexanes and hexabromocyclododecanes) can be overlooked (Goss et al. 2008). In general, when testing emerging compound classes, we strongly advise performing checks regardless of whether the compound class of interest falls in the model’s application domain. Additionally, as predictive programs such as COSMOtherm and SPARC are updated, it is necessary to reestablish the chemical applicability domain as new versions are released. In Figure 5.15, COSMOtherm estimations (with literature corrected p*i L values for PBDEs and PCDD/Fs) and SPARC estimations are compared with the experimental Kip over nine orders of magnitude from various sources in the literature, which were measured for a wide variety of terrestrial aerosols and environmental conditions. Considering this variety, both the COSMOtherm and SPARC estimations

5.6.2. Other Ambient Particles 5.6.2.1. Mixed-Aqueous Droplets. As mentioned previously, the mixed-aqueous phase of individual aerosols can increase several times in volume with increasing RH, due to deliquescence of salts and WSOM. During cloud-seeding events, the air is thick with completely deliquesced mixedaqueous droplets. The influences that dissolved salts and WSOM can have on partitioning to mixed-aqueous droplets are, indeed, multifarious and complex. Here we will first present, with broad general strokes, the influence that salts and WSOM can have individually, before commenting on their combined influence. When certain salts are present in high concentrations, salting-out and salting-in effects can occur. “Salting out” occurs when a salt lowers the sorptive capacity of an aqueous phase for organic compounds, and “salting in” applies when a salt raises the sorptive capacity. Whether salting out or salting in occurs depends on the salts and the organic compound. Salting out is commonly observed for small ion pairs (such as NaCl and KNO3) and apolar compounds. This is explainable using the sorbent–sorbate interactions presented in Section 5.4, in that the presence of negative and positive ions increases the sorbate–sorbate interactions and thus the required cavity formation energy. Salting in can occur, for instance, when organic salts or large inorganic ions are dissolved in solution (such as carboxylates and surfactants), which can alternatively lower the sorbent–sorbent interactions at the air–water interface and also lower the cavity

8

PAHs PCDDS-P*iL corrected PCDDS-P*iL corrected Organochlorines PCBs Alkanes PBDEs-p*iL corrected 1:1 line

7 6 5 4

Average Literature Kip (m3/g, various T)

Average literature Kip (m3/g, various T)

8

show excellent potential as screening tools, especially as they were not calibrated to any experimental data.

3 2 1 0 -1

PAHs PCDDs PCDFs Organochlorines PCBs Alkanes PBDEs 1:1 line

7 6 5 4 3 2 1 0 -1 -2

-2 -2

-1

0

1

2

3

4

5

6

7

8

COSMO therm predicted log Kip (m3/g, 15ºC, fWIOM = 0.1)

-2

-1

0

1

2

3

4

5

6

7

SPARC predicted log Kip (m3/g, 15ºC, fWIOM = 0.1)

Figure 5.15. Comparison of average literature log Ki p values from various literature sources with predictions using COSMOtherm and SPARC v4.2 [figure from Arp and Goss (2009b)].

8

PARTITIONING TO MIXED PARTICLE-PHASES

formation energy within the bulk matrix. A popular empirical approach for quantifying the salting-out effect is expressed as ci water; salt ¼ ci water  10Ki ½salttot S

ð5:46Þ

where ci water, salt is the activity coefficient of species i in the salty water, ci water that in pure water, KiS ; the Setschenow or “salting constant” (units M-1); and [salt]tot, the molar concentration of the salt. Some example values of the salt and sorbate-dependent KiS are compiled in Schwarzenbach et al. (2003). One generalization that can be made from this relationship is that because KiS values generally range from 0.1 to 0.3, the salting-out effect is generally small until salt concentrations become quite high. For example, a theoretical salt present at 3 M and with a KiS of 0.2 M-1 will change ci water by a factor of only 4. Nonionic WSOCs and other organics can cause cosolvency effects in the aqueous phase. A somewhat more detailed introduction of this topic can be found in Schwarzenbach et al. (2003); here we will mention only some general trends. Completely water-miscible organic compounds (such as methanol) increase the water solubility exponentially with concentration; this is also true for many polar partially miscible organic compounds (such as pentanol and phenols), but not -necessarily for nonpolar partially miscible organic compounds. This cosolvency effect is attributable to changes in sorbent–sorbent interactions. For instance, the presence of a compound like methanol in the water matrix weakens the overall sorbent–sorbent interactions per volume, and thereby lowers cavity formation energy. Designing thermodynamically explicit approaches to deal with the sorption properties of mixed-aqueous phases, considering all salts and WSOM over broad, nonlinear concentration ranges, would be a complex task, indeed. To present an example of the complexity, the most common atmospheric salts in acidic urban aersols, (NH4)2SO4 can cause both salting-out and salting-in effects with polar compounds, and certain mixtures of salts and organics in solution even result in phase changes (Marcolli and Krieger 2006). A common approach to dealing with such mixtures is to use variations of UNIFAC-based methods (e.g., Erdakos et al. 2006; Tong et al. 2008; Zuend et al. 2008). Although UNIFAC-based methods are generally used to describe hygroscopicity and water activity, they are also being developed to predict partitioning of organic compounds (e.g., Clegg et al. 2008a,b). Currently, a UNIFAC-based approach is the most feasible accurate way to approach the gas/particle partitioning of organic compounds to a broad array of mixed-aqueous aerosols, particularly during cloud nuclei formation. However, this level of sophistication is likely not necessary for predicting Kip at trace sorbate concentrations for the aqueous components of terrestrial aerosols. As shown above

139

and in Arp et al. (2008b), predictions based on no substantial salting or cosolvency effects were reasonable. For the partitioning of surfactants, however, this assumption may not hold (Arp and Goss 2009a). 5.6.2.2. Combustion and Road Tunnel Aerosols. The composition of combustion particles depends on the combustion media, combustion system, and ambient conditions on release into the atmosphere. Most types of combustion particles contain either primary OM or EC as the main component [see Table 5.1 and Hildemann et al. (1991)]. Systematic PP-LFERs for such aerosols are available only for diesel particles (Section 5.5.5), although calibrated SP-LFERs are available for various combustion particle types [e.g., for combustion particles from wood and cooking oils (Schauer et al. 2001, 2002a)]. More research is needed to fully characterize the sorption behavior of primary WIOM in diverse combustion aerosols. Road tunnel and traffic aerosols contain a mixture of diesel and gasoline combustion particles, which, on release into the environment, condense, grow, and mix with each other as well as with the components in the surrounding atmosphere. This initial mixing is quite rapid, and profound effects on aerosol properties are observed within minutes after emission (Jacobson 2001; Johnson et al. 2005; Shiraiwa et al. 2007). Accordingly, the specific surface area of road tunnel aerosols is significantly reduced compared to diesel soot (see Section 5.5.5), although not as much as it is for terrestrial aerosols, indicating that road tunnel aerosols are in an intermediate stage of mixing. Roth et al. (2005b) measured Kip values for a commercial road tunnel aerosol standard [European Union’s Institute for Reference Materials and Measurements (IRMM), Geel, Belgium, sold as CRM 605]. Although the dominating sorption mechanism could not be definitively isolated, the available Kip data collectively pointed to an absorptive mechanism. The PP-LFER sorbent descriptors derived from these data are not significantly different from the range of descriptors that have been derived for terrestrial WIOM (Table 5.4). This gives some indication that the dominant sorption mechanism to road tunnels is WIOM absorption. 5.6.2.3. Marine Aerosols. Marine aerosols are formed by wave action or wind, which blows off small particles and aqueous droplets from the ocean’s surface. In addition to water and oceanic salts, these particles and droplets contain WIOM (such as lipids and apolar compounds), WSOM (dissolved compounds), and surfactants that reside on the surface microlayer (Oppo et al. 1999). The weight fractions of the WIOM, WSOM, and salt fractions are highly dependent on particle size fraction [see Table 5.1 and O’Dowd et al. (2004)]. Little is currently known about the gas/particle partitioning behavior of marine aerosols. From what is known of the

140

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

TABLE 5.4. Absorbent PP-LFER Sorbent Descriptors (15 C) for Road Tunnel Aerosols and a Rural Aerosol Sample Rich in WIOM Sample Ki Ki

a road tunnel b Aspvreten

s

l

v

b

a

c

r2

n

0.87  0.08 0.95  0.09

0.91  0.04 0.64  0.03

0.23  0.13 0.49  0.09

0.17  0.08 0.55  0.14

2.5  0.10 2.52  0.11

6.2  0.07 5.95  0.12

0.99 0.975

66 57

Notation:  ¼ standard deviation; r2 ¼ correlation coefficient; RMSE ¼ root mean square error; n ¼ number of data. a Data from Roth et al. (2005b). b From coastal, organic-rich aerosols sampled in Aspvreten, Sweden (Arp et al. 2008a).

contents of these aerosols, it is expected that submicrometer particles are more attractive to organic sorbents of low water solubility, due to their abundance of WIOM, and that coarse particle size fractions are better sorbents for highly watersoluble compounds and surfactants. An important consideration is that when marine aerosols enter the atmosphere, they can change size and sorbent composition, by either agglomeration or particle breakup (Oppo et al. 1999; Tabazadeh 2005). Thus, the degree of internal mixing in a sample of marine aerosols, as with the release of combustion aerosols, is likely in a constant state of flux near the site of aerosol generation. 5.6.2.4. Indoor Aerosols. Indoor aerosols consist of terrestrial aerosols (i.e., outdoor aerosols that have come indoors) and particles that are generated indoors (Jones et al. 2000). Sources of indoor-generated particles include cooking, heating, smoking, erosion of materials, shedding of skin flakes, and particle emissions from household and office equipment. Previous studies on the gas/particle partitioning of indoor aerosols (e.g., Kavouras and Stephanou 2002; Naumova et al. 2003; Weschler et al. 2008) and cigarette smoke (Pankow et al. 1994) have developed SP-LFERs for several compound classes. Because of the diversity of indoor environments and indoor aerosols, and the potential exposure to toxic chemicals in indoor atmospheres, more research is needed to better understand the diversity of gas/particle partitioning of indoor aerosols before any generalized statements can be made. Future research in this direction is important, particularly considering that in most societies more time is spent indoors than outdoors. Future research should focus on developing innovative, low-volume sampling strategies that prevent an unrepresentative amount of outside air being sampled, and which target particles and chemicals are of concern in indoor environments (e.g., engineered nanoparticles, particles emitted from modern appliences, pharmaceuticals, antibiotics, flame retardants).

5.7. CONCLUSIONS Although the field of gas/particle partitioning has progressed, more work is needed, particularly for indoor environments (e.g., office buildings, hospitals, underground subway

passages, smoking lobbies), for tropical and arctic environments, emerging contaminants (e.g., surfactants, pharmaceuticals), and transformation products. Progress in this area will involve the continuous evolution, modifications and improvements of existing Kip measurement and modeling techniques. An inherent sampling bias of all the Kip measurements mentioned in this chapter that should be addressed is that sampling itself alters particle morphology and composition, and thereby potentially alters the measured gas/particle distribution behavior. Thus, ideally, future measurement methods should aim toward directly measuring particle and vapor concentrations in the atmosphere in situ, without the need for sample collection. In this chapter, several different methods of estimating Kip values are presented. The best method to use ultimately depends on the required accuracy of model predictions, the availability of experimental data, and the properties of the compound of interest (particularly its water solubility/ surfactant nature). For molecules with a high affinity to water, such as surfactants and ionizable compounds, partitioning models need to explicitly account for sorption to the RH dependent aqueous component [Eqs. (5.42) and (5.43)]. Because condensed water is by far the most abundant condensed phase in the atmosphere, the atmospheric fate of highly water–soluble molecules and surfactants is more likely controlled by sequestration and transport by condensed water droplets (e.g., clouds) than by dry aerosols. Partitioning to water droplets can be more complex to account for, due to simultaneously occurring adsorptive, absorptive, transformation, and micellization processes, which are all influenced by a variety of parameters (e.g., temperature, ionic strength, presence of organics, etc.). Future work on understanding how these parameters and processes interplay is needed. In special situations when human health is directly related to exposure to air particles and chemicals, such as in industrial environments where there is a high risk of exposure to toxic chemicals (e.g., waste incinerators, toxic waste landfills) or with the use of drug inhalers, obtaining accurate Kip values may be a necessity. To assure accuracy, Kip values should be directly measured. If this is not possible for the compounds of interest, than the best way of estimating Kip values currently is to first measure Kip values for other compounds with the IGC method, followed by the derivation of particle-specific calibrated PP-LFERs.

CONCLUSIONS

141

TABLE 5.5. PP-LFER Descriptors for Various Apolar Semivolatile AOCs B

Va

L

S

Organochlorines 0.00b 0.00d 0.00d

0.00b 0.47d 0.50d

1.451 1.580 1.580

7.624c 7.340d 7.570d

0.99b 1.20d 1.28d

0.00e 0.00e 0.00e 0.00e

0.00e 0.00e 0.00e 0.00e

2.645 2.927 3.208 3.490

8.725f 9.743f 10.732f 11.759f

0.00e 0.00e 0.00e 0.00e

Acenaphthylene Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benzo[a]fluorene Benzo[b]fluorene Benz[a]anthracene Chrysene Benzo[a]pyrene Benzo[e]pyrene Benzo[k]fluoranthene

0.00c 0.00g 0.00b 0.00b 0.00h 0.00g 0.00h 0.00h 0.00b 0.00b 0.00h 0.00c 0.00c

0.20e 0.20g 0.29b 0.28b 0.20h 0.29g 0.20h 0.20h 0.35b 0.36b 0.44h 0.44e 0.44e

1.216 1.357 1.454 1.454 1.585 1.585 1.726 1.726 1.823 1.823 1.954 1.954 1.954

6.175c 6.922c 7.632c 7.568c 8.827c 8.833c 9.404c 9.520c 10.291c 10.334c 11.715c 11.656c 11.607c

1.14c 1.03g 1.29b 1.34b 1.55h 1.71g 1.59h 1.57h 1.70b 1.73b 1.98h 1.99c 1.91c

PCBs 2,20 ,4,50 -Tetrachloro 3,30 ,4,40 -Tetrachloro 2,20 ,4,5,50 -Pentachloro 2,3,30 ,4,40 -Pentachloro 2,30 ,4,40 ,5-Pentachloro 3,30 ,4,40 ,5-Pentachloro 2,20 ,3,4,40 ,50 -Hexachloro 2,20 ,4,40 ,5,50 -Hexachloro 2,20 ,3,30 ,4,40 ,6-Heptachloro

0.00i 0.00i 0.00i 0.00i 0.00i 0.00i 0.00i 0.00i 0.00i

0.15i 0.11i 0.13i 0.11i 0.11i 0.09i 0.11i 0.11i 0.09i

1.814 1.814 1.936 1.936 1.936 1.936 2.059 2.059 2.181

8.186i 9.205i 8.868i 9.594i 9.396i 9.884i 9.772i 9.587i 10.031i

1.48i 1.44i 1.61i 1.59i 1.59i 1.57i 1.74i 1.74i 1.87i

CAS or Congener Number (CN)

Compounds

118-74-1 319-84-6 58-89-9 593-45-3 112-95-8 629-97-0 646-31-1

HCB a-HCH c-HCH Alkanes n-Octadecane n-Eicosane n-Docosane n-Tetracosane

208-96-8 86-73-7 85-01-8 120-12-7 206-44-0 129-00-0 238-84-6 243-17-4 56-55-3 218-01-9 50-32-8 192-97-2 207-08-9 49 (CN) 77 (CN) 101 (CN) 105 (CN) 118 (CN) 126 (CN) 138 (CN) 153 (CN) 171 (CN)

A

PAHs

a

McGowan volume as calculated in Abraham and McGowan (1987). Torres-Lapasio et al. (2004). c Abraham (1993a). d Goss et al. (2008). e Ai, Bi, and Si values assumed identical to homolog chemicals that differ only by the amount of n-alkyl or cyclo --CH2-- groups; Li values extrapolated from homolog by adding 0.505 to per --CH2-- unit. f Mutelet and Rogalski (2001). g Abraham et al. (1994a). h Abraham et al. (1994b). i Abraham and Al-Hussaini (2005). b

For large-scale environmental models that cover months or even years, there are a wide variety of factors (e.g., weather, diverse particle sources) that cause a large heterogeneity in particle types and sorption properties, which cause variations in Kip values probably up to two orders of magnitude, or even more for certain compound classes. Here, estimated Kip values only have to be representative, such as the median or average Kip value of all measured values. Therefore, whether one predicts Kip values for molecules of low water solubility with the PP-LFER,

COSMOtherm, SPARC, or the Ki oa SP-LFER method is probably not critical. In fact, a more recent systematic study on the influence of using PP-LFERs and Ki oa SP-LFER for determining the fate of a large chemical dataset found little differences between using these two types of model paradigms, and thus concluded the “best” model to use depends on parameter availability (Brown and Wania 2009). Ideally, to better serve large-scale fate models, it would be desirable to have a better understanding on the temporal and geographical distribution of ambient ambient Ki p values. For

142

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

instance, once ambient Ki p for a given compound become available for hundreds of locations and dates, the median and standard deviations of these naturally occurring Ki p values could be better represented. It should be emphasized here that input parameters used in Kip predictive models need to be verified and of high quality. For instance, Pontolillo and Eganhouse (2001) reported that octanol–water partition coefficients, from which Kioa data are commonly derived, for DDT in the literature have ranged over four orders of magnitude, with recommended values ranging over two orders of magnitude. This range in Ki oa values may very well be larger than the range of naturally occurring Kip values for DDT, thus highly erroneous Ki oa-SP LFER calibrations and resulting Kip estimations can potentially result from employing inaccurate Ki oa values. An advantage with PP-LFER descriptors (listed in Table 5.5) in this regard is that, because these are determined over a large set of equations and experiments (and not just one experiment), large errors in reported PP-LFERs do not go unnoticed for very long. For the large-scale screening of compounds and for Kip predictions of transformation products and potential industrial products in which no experimental data are available, we recommend using the COSMOtherm and SPARC approaches presented here, which require only molecular structures as input. Ideally, both COSMOtherm and SPARC should be used, perhaps in combination with additional models, as comparing output datasets from several models assists in identifying outliers and the limits of chemical application domains. We recommend that future research continue to develop and utilize COSMOtherm/SPARC-type models. Ultimately, such developments will continue to branch and hold the link together between the more fundamental and applied sciences, not only conceptually but via interdisciplinary scientific collaborations. By applying fundamental theory involving quantum chemistry and thermodynamics, and combining this with microscale structural information from imaging and spectroscopic techniques, and further combining this with macroscale environmental observations, reliable and mechanistically satisfying models can be designed that shed simultaneous insight into microand macroscale processes that are of relevance to the environment and human health.

REFERENCES Abraham, M. H. and McGowan, J. C. (1987), The use of characteristic volumes to measure cavity terms in reversed phase liquid-chromatography, Chromatographia 23, 243–246. Abraham, M. H. (1993a), Hydrogen bonding XXVII. solvation parameters for functionally substituted aromatic compounds and heterocyclic compounds, from gas-liquid chromatographic data, J. Chromatogr. 644, 95–139.

Abraham, M. H. (1993b), Scales of solute hydrogen-bonding: Their construction and application to physicochemical and biochemical processes, Chem. Soc. Rev. 22, 73–83. Abraham, M. H., Andonian-Haftvan, J., Whiting, G. S., Leo, A., and Taft, R. S. (1994a), Hydrogen bonding. Part 34. The factors that influence the solubility of gases and vapours in water at 298K, and a new method for its determination, J. Chem. Soc. Perkin Trans. 28, 1777–1791. Abraham, M. H., Chada, H. S., Whiting, G. S., and Mitchell, R. C. (1994b), Hydrogen bonding. 32. An analysis of water-octanol and water-alkane partitioning and the Dlog P parameter of Seiler, J. Pharm. Sci. 83, 1085–1100. Abraham, M. H. and Al-Hussaini, A. J. M. (2001), Solvation descriptors for the polychloronaphthalenes: Estimation of some physicochemical properties, J. Environ. Monit. 3, 377–381. Abraham, M. H., Ibrahim, A., and Zissimos, A. M. (2004), Determination of sets of solute descriptors from chromatographic measurements, J. Chromatogr. A 1037, 29–47. Abraham, M. H. and Al-Hussaini, A. J. M. (2005), Solvation parameters for the 209 PCBs: Calculation of physicochemical properties, J. Environ. Monit. 7, 295–301. Aiken, A. C., DeCarlo, P. F., Kroll, J. H., Worsnop, D. R., Huffman, J. A., Docherty, K. S., Ulbrich, I. M., Mohr, C., Kimmel, J. R., Sueper, D., Sun, Y., Zhang, Q., Trimborn, A., Northway, M., Ziemann, P. J., Canagaratna, M. R., Onasch, T. B., Alfarra, M. R., Prevot, A. S. H., Dommen, J., Duplissy, J., Metzger, A., Baltensperger, U., and Jimenez, J. L. (2008), O/C and OM/OC ratios of primary, secondary, and ambient organic aerosols with highresolution time-of-flight aerosol mass spectrometry, Environ. Sci. Technol. 42, 4478–4485. Arey, J. S., Green, W. H., and Gschwend, P. M. (2005), The electrostatic origin of Abraham’s solute polarity parameter, J. Phys. Chem. B 109, 7564–7573. Arp, H. P. H. and Schmidt, T. C. (2004), Air-water transfer of MTBE, its degradation products, and alternative fuel oxygenates: The role of temperature, Environ. Sci. Technol. 38, 5405–5412. Arp, H. P. H., Goss, K.-U., and Schwarzenbach, R. P. (2006a), Evaluation of a predictive model for air/surface adsorption equilibrium constants and enthalpies, Environ. Toxicol. Chem. 25, 45–51. Arp, H. P. H., Niederer, C., and Goss, K.-U. (2006b), Predicting the partitioning behavior of various highly fluorinated compounds, Environ. Sci. Technol. 40, 7298–7304. Arp, H. P. H., Schwarzenbach, R. P., and Goss, K.-U. (2007), Equilibrium sorption of gaseous organic chemicals to fiber filters used for aerosol studies, Atmos. Environ. 41, 8241–8252. Arp, H. P. H. and Goss, K. U. (2008), Irreversible sorption of trace concentrations of perfluorocarboxylic acids to fiber filters used for air sampling, Atmos. Environ. 42, 6869–6972. Arp, H. P. H., Schwarzenbach, R. P., and Goss, K.-U. (2008a), Ambient gas/particle partitioning. 2: The influence of particle source and temperature on sorption to dry terrestrial aerosols, Environ. Sci. Technol. 42, 5951–5957. Arp, H. P. H., Schwarzenbach, R. P., and Goss, K.-U. (2008b), Ambient gas/particle partitioning. 1. Sorption mechanisms of

REFERENCES

apolar, polar and ionizable organic compounds, Environ. Sci. Technol. 42, 5541–5547. Arp, H. P. H., Schwarzenbach, R. P., and Goss, K.-U. (2008c), Determination of ambient gas-particle partitioning constants of non-polar and polar organic compounds using inverse gas chromatography, Atmos. Environ. 42, 303–312. Arp, H. P. H. and Goss, K.-U. (2009a), The gas/particle partitioning behavior of perfluorocarboxylic acids with terrestrial aerosols, Environ. Sci. Technol. 43, 8542–8547. Arp, H. P. H. and Goss, K.-U. (2009b), Ambient gas/particle partitioning. 3. Estimating partition coefficients of apolar, polar, and ionizable organic compounds by their molecular structure, Environ. Sci. Technol. 43, 1923–1929. Asher, W. E., Pankow, J. F., Erdakos, G. B., and Seinfeld, J. H. (2002), Estimating the vapor pressures of multi-functional oxygen-containing organic compounds using group contribution methods, Atmos. Environ. 36, 1483–1498. Atapattu, S. N. and Poole, C. F. (2008), Solute descriptors for characterizing retention properties of open-tubular columns of different selectivity in gas chromatography at intermediate temperatures, J. Chromatogr. A 1195, 136–145. Atkinson, D. and Curthoys, G. (1978), Determination of heats of adsorption by gas-solid chromatography, J. Chem. Educ. 55, 564–566. Bidleman, T. F. and Olney, C. E. (1974), High-volume collection of atmospheric polychlorinated biphenyls, Bull. Environ. Contam. Toxicol. 11, 442–450. Bidleman, T. F. (1988), Atmospheric processes—wet and dry deposition of organic-compounds are controlled by their vapor particle partitioning, Environ. Sci. Technol. 22, 361–367. Birch, M. E. and Cary, R. A. (1996), Elemental carbon-based method for monitoring occupational exposures to particulate diesel exhaust, Aerosol Sci. Technol. 25, 221–241. Brown, T. N. and Wania, F. (2009), Development and exploration of an organic contaminant fate model using poly-parameter linear free energy relationships, Environ. Sci. Technol. 43, 6676–6683. Burtscher, H. (2005), Physical characterization of particulate emissions from diesel engines: A review, J. Aerosol. Sci. 36, 896–932. Cavalli, F., Facchini, M. C., Decesari, S., Mircea, M., Emblico, L., Fuzzi, S., Ceburnis, D., Yoon, Y. J., O’Dowd, C. D., Putaud, J. P., and Dell’Acqua, A. (2004), Advances in characterization of sizeresolved organic matter in marine aerosol over the North Atlantic, J. Geophys. Res. Atmos. 109, D24215. Chan, M. N., Choi, M. Y., Ng, N. L., and Chan, C. K. (2005), Hygroscopicity of water-soluble organic compounds in atmospheric aerosols: Amino acids and biomass burning derived organic species, Environ. Sci. Technol. 39, 1555–1562. Chandramouli, B., Jang, M., and Kamens, R. M. (2003a), Gasparticle partitioning of semi-volatile organics on organic aerosols using a predictive activity coefficient model: Analysis of the effects of parameter choices on model performance, Atmos. Environ. 37, 853–864. Chandramouli, B., Jang, M. S., and Kamens, R. M. (2003b), Gasparticle partitioning of semivolatile organic compounds (SOCs) on mixtures of aerosols in a smog chamber, Environ. Sci. Technol. 37, 4113–4121.

143

Chang, E. I. and Pankow, J. F. (2006), Prediction of activity coefficients in liquid aerosol particles containing organic compounds, dissolved inorganic salts, and water—Part 2: Consideration of phase separation effects by an X-UNIFAC model, Atmos. Environ. 40, 6422–6436. Chow, J. C., Watson, J. G., Fujita, E. M., Lu, Z. Q., Lawson, D. R., and Ashbaugh, L. L. (1994), Temporal and spatial variations of PM(2.5) and PM(10) aerosol in the southern California airquality study, Atmos. Environ. 28, 2061–2080. Clegg, S. L., Kleeman, M. J., Griffin, R. J., and Seinfeld, J. H. (2008a), Effects of uncertainties in the thermodynamic properties of aerosol components in an air quality model—Part 1: Treatment of inorganic electrolytes and organic compounds in the condensed phase, Atmos. Chem. Phys. 8, 1057–1085. Clegg, S. L., Kleeman, M. J., Griffin, R. J., and Seinfeld, J. H. (2008b), Effects of uncertainties in the thermodynamic properties of aerosol components in an air quality model—Part 2: Predictions of the vapour pressures of organic compounds, Atmos. Chem. Phys. 8, 1087–1103. Conder, J. R. and Young, C. L. (1979), Physicochemical Measurement by Gas Chromatography, Wiley, New York. Cramer, R. D. (1980), Bc(Def) parameters. 1. Intrinsic dimensionality of intermolecular interactions in the liquid-state, J. Am. Chem. Soc. 102, 1837–1849. Dachs, J. and Eisenreich, S. J. (2000), Adsorption onto aerosol soot carbon dominates gas-particle partitioning of polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 34, 3690–3697. de Boer, J. H. (1968), The Dynamic Character of Adsorption, 2nd ed., Clarendon Press, Oxford. Ding, X., Zheng, M., Yu, L. P., Zhang, X. L., Weber, R. J., Yan, B., Russell, A. G., Edgerton, E. S., and Wang, X. M. (2008), Spatial and seasonal trends in biogenic secondary organic aerosol tracers and water-soluble organic carbon in the southeastern United States, Environ. Sci. Technol. 42, 5171–5176. Dorris, G. M. and Gray, D. G. (1981), Adsorption of hydrocarbons on silica-supported water surfaces, J. Phys. Chem. 85, 3628–3635. Duarte, R., Santos, E. B. H., Pio, C. A., and Duarte, A. C. (2007), Comparison of structural features of water-soluble organic matter from atmospheric aerosols with those of aquatic humic substances, Atmos. Environ. 41, 8100–8113. Duarte, R., Silva, A. M. S., and Duarte, A. C. (2008), Twodimensional NMR studies of water-soluble organic matter in atmospheric aerosols, Environ. Sci. Technol. 42, 8224–8230. Eckert, F. and Klamt, A. (2002), Fast solvent screening via quantum chemistry: COSMO-RS approach, Am. Inst. Chem. Eng. J. 48, 369. Eckert, F. and Klamt, A. (2005), COSMOtherm v 2.1, COSMOlogic GmbH KG, Leverkusen, Germany. Endo, S., Grathwohl, P., and Schmidt, T. C. (2008), Absorption or adsorption? Insights from molecular probes n-alkanes and cycloalkanes into modes of sorption by environmental solid matrices, Environ. Sci. Technol. 42, 3989–3995. Erdakos, G. B. and Pankow, J. F. (2004), Gas/particle partitioning of neutral and ionizing compounds to single- and multi-phase aerosol particles. 2. Phase separation in liquid particulate matter

144

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

containing both polar and low-polarity organic compounds, Atmos. Environ. 38, 1005–1013. Erdakos, G. B., Asher, W. E., Seinfeld, J. H. and Pankow, J. F. (2006), Prediction of activity coefficients in liquid aerosol particles containing organic compounds, dissolved inorganic salts, and water—Part 1: Organic compounds and water by consideration of short- and long-range effects using X-UNIFAC.1, Atmos. Environ. 40, 6410–6421. Finizio, A., Mackay, D., Bidleman, T., and Harner, T. (1997), Octanol-air partition coefficient as a predictor of partitioning of semi-volatile organic chemicals to aerosols, Atmos. Environ. 31, 2289–2296. Fuzzi, S., Andreae, M. O., Huebert, B. J., Kulmala, M., Bond, T. C., Boy, M., Doherty, S. J., Guenther, A., Kanakidou, M., Kawamura, K., Kerminen, V. M., Lohmann, U., Russell, L. M., and Poschl, U. (2006), Critical assessment of the current state of scientific knowledge, terminology, and research needs concerning the role of organic aerosols in the atmosphere, climate, and global change, Atmos. Chem. Phys. 6, 2017–2038. Galarneau, E. and Bidleman, T. F. (2006), Modelling the temperature-induced blow-off and blow-on artefacts in filter-sorbent measurements of semivolatile substances, Atmos. Environ. 40, 4258–4268. Galarneau, E., Bidleman, T. F., and Blanchard, P. (2006), Seasonality and interspecies differences in particle/gas partitioning of PAHs observed by the integrated atmospheric deposition network (IADN), Atmos. Environ. 40, 182–197. Goss, K.-U. and Schwarzenbach, R. P. (1998), Gas/solid and gas/liquid partitioning of organic compounds: Critical evaluation of the interpretation of equilibrium constants, Environ. Sci. Technol. 32, 2025–2032. Goss, K.-U. and Schwarzenbach, R. P. (2001), Linear free energy relatinships used to evaluate equilibrium partitioning of organic compounds, Environ. Sci. Technol. 35, 1–9. Goss, K.-U. and Schwarzenbach, R. P. (2002), Adsorption of a diverse det of organic vapors on quartz, CaCO3 and alpha-Al2O3 at different relative humidities, J. Colloid Interface Sci. 252, 31–41. Goss, K.-U., Buschmann, J., and Schwarzenbach, R. P. (2003), Determination of the surface sorption properties of talc, different salts, and clay minerals at various relative humidities using adsorption data of a diverse set of organic vapors, Environ. Toxicol. Chem. 22, 2667–2672. Goss, K.-U. (2004), The air/surface adsorption equilibrium of organic compounds under ambient conditions, Crit. Rev. Environ. Sci. Technol. 34, 339–389. Goss, K.-U. (2005), Predicting the equilibrium partitioning of organic compounds using just one linear solvation energy relationship (LSER), Fluid Phase Equilib. 233, 19–22. Goss, K.-U., Arp, H. P., and Roth, C. (2005), Comment on “Model for the adsorption of organic compounds at gas-water interfaces” by C. F. Poole, [J. Environ. Monit. 7, 577 (2005)], J. Environ. Monit. 7, 1105–1106. Goss, K.-U. and Bronner, G. (2006), What is so special about the sorption behavior of highly fluorinated compounds? J. Phys. Chem. A 110, 9518–9522.

Goss, K.-U., Arp, H. P. H., Bronner, G., and Niederer, C. (2008), Partition behavior of hexachlorocyclohexane-isomers, J. Chem. Eng. Data 53, 750–754. Gotz, C. W., Scheringer, M., Macleod, M., Roth, C. M., and Hungerbuhler, K. (2007), Alternative approaches for modeling gas-particle partitioning of semivolatile organic chemicals: Model development and comparison, Environ. Sci. Technol. 41, 1272–1278. Griffin, R. J., Nguyen, K., Dabdub, D., and Seinfeld, J. H. (2003), A coupled hydrophobic-hydrophilic model for predicting secondary organic aerosol formation, J. Atmos. Chem. 44, 171–190. Gysel, M., Weingartner, E., Nyeki, S., Paulsen, D., Baltensperger, U., Galambos, I., and Kiss, G. (2004), Hygroscopic properties of water-soluble matter and humic-like organics in atmospheric fine aerosol, Atmos. Chem. Phys. 4, 35–50. Harner, T. and Bidleman, T. F. (1998), Octanol-air partition coefficient for describing particle/gas partitioning of aromatic compounds in urban air, Environ. Sci. Technol. 32, 1494– 1502. Hart, K. M., McDow, S. R., Giger, W., Steiner, D., and Burtscher, H. (1993), The correlation between in-situ, real-time aerosol photoemission intensity and particulate polycyclic aromatic hydrocarbon concentration in combustion aerosols, Water Air Soil Pollut. 68, 75–90. Hilal, S. H., Karickhoff, S. W., and Carreira, L. A. (2003), Prediction of the vapor pressure boiling point, heat of vaporization and diffusion coefficient of organic compounds, Quant. Struct.–Act. Relat. Combin. Comb. Sci. 22, 565–574. Hilal, S. H., Karickhoff, S. W. and Carreira, L. A. (2004), Prediction of the solubility, activity coefficient and liquid/liquid partition coefficient of organic compounds. Quant. Struct.–Act. Relat. Combin. Comb. Sci. 23, 709–720. Hildemann, L. M., Markowski, G. R., and Cass, G. R. (1991), Chemical-composition of emissions from urban sources of fine organic aerosol, Environ. Sci. Technol. 25, 744–759. Hinds, W. C. (1999), Aerosol Technology: Properties, Behavior and Measurement of Airborne Particles, 2nd ed., WileyInterscience, New York. Hueglin, C., Gehrig, R., Baltensperger, U., Gysel, M., Monn, C., and Vonmont, H. (2005), Chemical characterisation of PM2.5, PM10 and coarse particles at urban, near-city and rural sites in Switzerland, Atmos. Environ. 39, 637–651. Jacobson, M. Z. (2001), Strong radiative heating due to the mixing state of black carbon in atmospheric aerosols, Nature 409, 695–697. Johnson, K. S., Zuberi, B., Molina, L. T., Molina, M. J., Iedema, M. J., Cowin, J. P., Gaspar, D. J., Wang, C., and Laskin, A. (2005), Processing of soot in an urban environment: Case study from the Mexico Citymetropolitanarea,Atmos.Chem. Phys.5, 3033–3043. Jones, N. C., Thornton, C. A., Mark, D., and Harrison, R. M. (2000), Indoor/outdoor relationships of particulate matter in domestic homes with roadside, urban and rural locations, Atmos. Environ. 34, 2603–2612. Jonker, M. T. O. and Koelmans, A. A. (2002), Extraction of polycyclic aromatic hydrocarbons from soot and sediment: Solvent evaluation and implications for sorption mechanism, Environ. Sci. Technol. 36, 4107–4113.

REFERENCES

Jonker, M. T. O., Hawthorne, S. B., and Koelmans, A. A. (2005), Extremely slowly desorbing polycyclic aromatic hydrocarbons from soot and soot-like materials: Evidence by supercritical fluid extraction, Environ. Sci. Technol. 39, 7889–7895. Kalberer, M., Paulsen, D., Sax, M., Steinbacher, M., Dommen, J., Prevot, A. S. H., Fisseha, R., Weingartner, E., Frankevich, V., Zenobi, R., and Baltensperger, U. (2004), Identification of polymers as major components of atmospheric organic aerosols, Science 303, 1659–1662. Kavouras, I. G. and Stephanou, E. G. (2002), Gas/particle partitioning and size distribution of primary and secondary carbonaceous aerosols in public buildings, Indoor Air—Int. J. Indoor Air Qual. Clim. 12, 17–32. Khlystov, A., Stanier, C. O., Takahama, S. and Pandis, S. N. (2005), Water content of ambient aerosol during the Pittsburgh air quality study, J. Geophys. Res. Atmos. 110. Kiss, G., Varga, B., Galambos, I., and Ganszky, I. (2002), Characterization of water-soluble organic matter isolated from atmospheric fine aerosol, J. Geophys. Res. Atmos. 107. Kleeman, M. J., Hughes, L. S., Allen, J. O., and Cass, G. R. (1999), Source contributions to the size and composition distribution of atmospheric particles: Southern California in September 1996, Environ. Sci. Technol. 33, 4331–4341. Koehler, K. A., Kreidenweis, S. M., DeMott, P. J., Prenni, A. J., Carrico, C. M., Ervens, B., and Feingold, G. (2006), Water activity and activation diameters from hygroscopicity data—Part II: Application to organic species, Atmos. Chem. Phys. 6, 795–809. Krivacsy, Z., Gelencser, A., Kiss, G., Meszaros, E., Molnar, A., Hoffer, A., Meszaros, T., Sarvari, Z., Temesi, D., Varga, B., Baltensperger, U., Nyeki, S., and Weingartner, E. (2001), Study on the chemical character of water soluble organic compounds in fine atmospheric aerosol at the Jungfraujoch, J. Atmos. Chem. 39, 235–259. Lee, K. W. and Makund, R. (2001), Filter collection, in Aerosol Measurement: Principles, Techniques and Applications, Baron P. A. and K., Willeke, eds., 2nd ed., Wiley, New York, PP. 197–228. Lei, Z. G., Chen, B. H., Li, C. Y., and Liu, H. (2008), Predictive molecular thermodynamic models for liquid solvents, solid salts, polymers, and ionic liquids, Chem. Rev. 108, 1419–1455. Liu, T. and Oberg, T. (2009), Modelling of partition constants: Linear solvation energy relationships or PLS regression? J. Chemom. 23, 254–262. Lohmann, R. and Lammel, G. (2004), Adsorptive and absorptive contributions to the gas-particle partitioning of polycyclic aromatic hydrocarbons: State of knowledge and recommended parametrization for modeling, Environ. Sci. Technol. 38, 3793–3803. Mackay, D., Paterson, S., and Schroeder, W. H. (1986), Model describing the rates of transfer processes of organic-chemicals between atmosphere and water, Environ. Sci. Technol. 20, 810–816. Mader, B. T., Flagan, R. C., and Seinfeld, J. H. (2001), Sampling atmospheric carbonaceous aerosols using a particle trap impactor/denuder sampler, Environ. Sci. Technol. 35, 4857–4867.

145

Mader, B. T. and Pankow, J. F. (2001), Gas/solid partitioning of semivolatile organic compounds (SOCs) to air filters. 3. An analysis of gas adsorption artifacts in measurements of atmospheric SOCs and organic carbon (OC) when using Teflon membrane filters and quartz fiber filters, Environ. Sci. Technol. 35, 3422–3432. Mader, B. T. and Pankow, J. F. (2002), Study of the effects of particle-phase carbon on the gas/particle partitioning of sernivolatile organic compounds in the atmosphere using controlled field experiments, Environ. Sci. Technol. 36, 5218–5228. Marcolli, C. and Krieger, U. K. (2006), Phase changes during hygroscopic cycles of mixed organic/inorganic model systems of tropospheric aerosols, J. Phys. Chem. A 110, 1881–1893. Mintz, C., Clark, M., Acree, W. E., and Abraham, M. H. (2007), Enthalpy of solvation correlations for gaseous solutes dissolved in water and in 1-octanol based on the Abraham model, J. Chem Inf. Model. 47, 115–121. Mutelet, F. and Rogalski, M. (2001), Experimental determination and prediction of the gas-liquid n-hexadecane partition coefficients, J. Chromatogr. A 923, 153–163. Naumova, Y. Y., Offenberg, J. H., Eisenreich, S. J., Meng, Q. Y., Polidori, A., Turpin, B. J., Weisel, C. P., Morandi, M. T., Colome, S. D., Stock, T. H., Winer, A. M., Alimokhtari, S., Kwon, J., Maberti, S., Shendell, D., Jones, J., and Farrar, C. (2003), Gas/particle distribution of polycyclic aromatic hydrocarbons in coupled outdoor/indoor atmospheres, Atmos. Environ. 37, 703–719. Niederer, C., Goss, K.-U., and Schwarzenbach, R. P. (2006), Sorption equilibrium of a wide spectrum of organic vapors in leonardite humic acid: Experimental setup and experimental data, Environ. Sci. Technol. 40, 5368–5373. Norbeck, J. M., Durbin, T. D., and Truex, T. J. (1998), Measurement of Primary Particulate Matter Emissions from Light-Duty Motor Vehicles, Final Report for CRC Project E-24-2. Oberdorster, G., Sharp, Z., Atudorei, V., Elder, A., Gelein, R., Kreyling, W., and Cox, C. (2003), Translocation of inhaled ultrafine particles to the brain, Proc. 4th Colloquium on PM and Human Health, Pittsburgh, PA, PP. 437–445. O’Dowd, C. D., Facchini, M. C., Cavalli, F., Ceburnis, D., Mircea, M., Decesari, S., Fuzzi, S., Yoon, Y. J., and Putaud, J. P. (2004), Biogenically driven organic contribution to marine aerosol, Nature 431, 676–680. Oppo, C., Bellandi, S., Innocenti, N. D., Stortini, A. M., Loglio, G., Schiavuta, E., and Cini, R. (1999), Surfactant components of marine organic matter as agents for biogeochemical fractionation and pollutant transport via marine aerosols, Mar. Chem. 63, 235–253. Pankow, J. F. (1987), Review and comparative-analysis of the theories on partitioning between the gas and aerosol particulate phases in the atmosphere, Atmos. Environ. 21, 2275–2283. Pankow, J. F. and Bidleman, T. F. (1991), Effects of temperature, tsp and per cent nonexchangeable material in determining the gas particle partitioning of organic compounds, Atmos. Environ. Pt. A-General. Topics. 25, 2241–2249. Pankow, J. F. and Bidleman, T. F. (1992), Interdependence of the slopes and intercepts from log-log correlations of measured gas

146

SORPTION OF ANTHROPOGENIC ORGANIC COMPOUNDS TO AIRBORNE PARTICLES

particle partitioning and vapor-pressure.1. Theory and analysis of available data, Atmos. Environ. Pt. A—General Topics. 26, 1071–1080. Pankow, J. F. (1994a), An absorption-model of the gas aerosol partitioning involved in the formation of secondary organic aerosol, Atmos. Environ. 28, 189–193. Pankow, J. F. (1994b), An absorption-model of gas-particle partitioning of organic-compounds in the atmosphere, Atmos. Environ. 28, 185–188. Pankow, J. F. (2003), Gas/particle partitioning of neutral and ionizing compounds to single- and multi-phase aerosol particles. 1. Unified modeling framework, Atmos. Environ. 37, 4993–4993. Pankow, J. F., Isabelle, L. M., Buchholz, D. A., Luo, W. T., and Reeves, B. D. (1994), Gas-particle partitioning of polycyclic aromatic-hydrocarbons and alkanes to environmental tobaccosmoke, Environ. Sci. Technol. 28, 363–365. Pontolillo, J. and Eganhouse, R. P. (2001), The search for reliable aqueous solubility (Sw) and octanol-water partition coefficient (Kow) data for hydrophobic organic compounds: DDT and DDE as a case study, in Water-Resources Investigations Report, U.S. Department of the Interior, U.S. Geological Survey, Reston, VA, p. 50. Pun, B. K., Griffin, R. J., Seigneur, C., and Seinfeld, J. H. (2002), Secondary organic aerosol—2. Thermodynamic model for gas/ particle partitioning of molecular constituents, J. Geophys. Res. Atmos. 107, 1–14. Renner, R. (2002), The K-ow controversy, Environ. Sci. Technol. 36, 410A–413A. Roth, C. M., Goss, K.-U., and Schwarzenbach, R. P. (2002), Adsorption of a diverse set of organic vapors on the bulk water surface, J. Colloid Interface Sci. 252, 21–30. Roth, C. M., Goss, K.-U., and Schwarzenbach, R. P. (2004), Sorption of diverse organic vapors to snow, Environ. Sci. Technol. 38, 4078–4084. Roth, C. M., Goss, K.-U., and Schwarzenbach, R. P. (2005a), Sorption of a diverse set of organic vapors to urban aerosols, Environ. Sci. Technol. 39, 6638–6643. Roth, C. M., Goss, K.-U., and Schwarzenbach, R. P. (2005b), Sorption of a diverse set of organic vapors to diesel soot and road tunnel aerosols, Environ. Sci. Technol. 39, 6632–6637. Schauer, J. J., Kleeman, M. J., Cass, G. R., and Simoneit, B. R. T. (1999), Measurement of emissions from air pollution sources. 2. C-1 through C-30 organic compounds from medium duty diesel trucks, Environ. Sci. Technol. 33, 1578–1587. Schauer, J. J., Kleeman, M. J., Cass, G. R., and Simoneit, B. R. T. (2001), Measurement of emissions from air pollution sources. 3. C-1-C-29 organic compounds from fireplace combustion of wood, Environ. Sci. Technol. 35, 1716–1728. Schauer, J. J., Kleeman, M. J., Cass, G. R., and Simoneit, B. R. T. (2002a), Measurement of emissions from air pollution sources. 4. C-1-C-27 organic compounds from cooking with seed oils, Environ. Sci. Technol. 36, 567–575. Schauer, J. J., Kleeman, M. J., Cass, G. R., and Simoneit, B. R. T. (2002b), Measurement of emissions from air pollution sources. 5. C-1-C-32 organic compounds from gasoline-powered motor vehicles, Environ. Sci. Technol. 36, 1169–1180.

Schauer, J. J., Mader, B. T., Deminter, J. T., Heidemann, G., Bae, M. S., Seinfeld, J. H., Flagan, R. C., Cary, R. A., Smith, D., Huebert, B. J., Bertram, T., Howell, S., Kline, J. T., Quinn, P., Bates, T., Turpin, B., Lim, H. J., Yu, J. Z., Yang, H., and Keywood, M. D. (2003), ACE-Asia intercomparison of a thermal-optical method for the determination of particle-phase organic and elemental carbon, Environ. Sci. Technol. 37, 993–1001. Schmid, H., Laskus, L., Abraham, H. J., Baltensperger, U., Lavanchy, V., Bizjak, M., Burba, P., Cachier, H., Crow, D., Chow, J., Gnauk, T., Even, A., ten Brink, H. M., Giesen, K. P., Hitzenberger, R., Hueglin, C., Maenhaut, W., Pio, C., Carvalho, A., Putaud, J. P., Toom-Sauntry, D., and Puxbaum, H. (2001), Results of the “carbon conference” international aerosol carbon round robin test stage I, Atmos. Environ. 35, 2111–2121. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (2003), Environmental Organic Chemistry, 2nd ed., Wiley, Hoboken, NJ. Seinfeld, J. H. and Pandis, S. N. (2006), Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, 2nd ed., Wiley-Interscience, Hoboken, NJ. Seinfeld, J. H. and Pankow, J. F. (2003), Organic atmospheric particulate material, Annu. Rev. Phys. Chem. 54, 121–140. Sheffield, A. E. and Pankow, J. F. (1994), Specific surface-area of urban atmospheric particulate matter in Portland, Oregon, Environ. Sci. Technol. 28, 1759–1766. Shiraiwa, M., Kondo, Y., Moteki, N., Takegawa, N., Miyazaki, Y., and Blake, D. R. (2007), Evolution of mixing state of black carbon in polluted air from Tokyo, Geophys. Res. Lett. 34, 5. Sillanpaa, M., Frey, A., Hillamo, R., Pennanen, A. S., and Salonen, R. O. (2005), Organic, elemental and inorganic carbon in particulate matter of six urban environments in Europe, Atmos. Chem. Phys. 5, 2869–2879. Tabazadeh, A. (2005), Organic aggregate formation in aerosols and its impact on the physicochemical properties of atmospheric particles, Atmos. Environ. 39, 5472–5480. Taft, R. W., Abboud, J. L. M., Kamlet, M. J., and Abraham, M. H. (1985), Linear solvation energy relations, J. Solution Chem. 14, 153–186. Tang, I. N. and Munkelwitz, H. R. (1993), Composition and temperature-dependence of the deliquescence properties of hygroscopic aerosols, Atmos. Environ. Pt. A—General Topics 27, 467–473. Tao, S., Liu, Y., Xu, W., Lang, C., Liu, S., Dou, H. and Liu, W. (2007), Calibration of a passive Sampler for both gaseous and particulate phase polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 41 568–573. Tong, C. H., Clegg, S. L., and Seinfeld, J. H. (2008), Comparison of activity coefficient models for atmospheric aerosols containing mixtures of electrolytes, organics, and water, Atmos. Environ. 42, 5459–5482. Torres-Lapasio, J. R., Garcia-Alvarez-Coque, M. C., Roses, M., Bosch, E., Zissimos, A. M., and Abraham, M. H. (2004), Analysis of a solute polarity parameter in reversed-phase liquid chromatography on a linear solvation relationship basis, Anal. Chim. Acta 515, 209–227.

REFERENCES

Tulp, H. C., Goss, K. U., Schwarzenbach, R. P., and Fenner, K. (2008), Experimental determination of LSER parameters for a set of 76 diverse pesticides and pharmaceuticals, Environ. Sci. Technol. 42, 2034–2040. Volckens, J. and Leith, D. (2003a), Comparison of methods for measuring gas-particle partitioning of semivolatile compounds, Atmos. Environ. 37, 3177–3188. Volckens, J. and Leith, D. (2003b), Effects of sampling bias on gasparticle partitioning of semi- volatile compounds, Atmos. Environ. 37, 3385–3393. Wang, H. B., Kawamura, K., and Shooter, D. (2005), Carbonaceous and ionic components in wintertime atmospheric aerosols from two New Zealand cities: Implications for solid fuel combustion, Atmos. Environ. 39, 5865–5875. Weschler, C. J., Salthammer, T., and Fromme, H. (2008), Partitioning of phthalates among the gas phase, airborne particles and settled dust in indoor environments, Atmos. Environ. 42, 1449–1460. Yang, H., Yu, J. Z., Ho, S. S. H., Xu, J. H., Wu, W. S., Wan, C. H., Wang, X. D., Wang, X. R., and Wang, L. S. (2005), The chemical composition of inorganic and carbonaceous materials in PM2.5 in Nanjing, China, Atmos. Environ. 39, 3735–3749.

147

Yao, X. H., Ling, T. Y., Fang, M., and Chan, C. K. (2006), Comparison of thermodynamic predictions for in situ pH in PM2.5, Atmos. Environ. 40, 2835–2844. Yttri, K. E., Aas, W., Bjerke, A., Cape, J. N., Cavalli, F., Ceburnis, D., Dye, C., Emblico, L., Facchini, M. C., Forster, C., Hanssen, J. E., Hansson, H. C., Jennings, S. G., Maenhaut, W., Putaud, J. P., and Torseth, K. (2007), Elemental and organic carbon in PM10: A one year measurement campaign within the European Monitoring and Evaluation Programme EMEP, Atmos. Chem. Phys. 7, 5711–5725. Zhang, Q., Jimenez, J. L., Worsnop, D. R., and Canagaratna, M. (2007), A case study of urban particle acidity and its influence on secondary organic aerosol, Environ. Sci. Technol. 41, 3213–3219. Zielinska, B. (2005), Atmospheric transformation of diesel emissions, Exp. Toxicol. Pathol. 57, 31–42. Zissimos, A. M., Abraham, M. H., Klamt, A., Eckert, F. and Wood, J. (2002), A comparison between the two general sets of linear free energy descriptors of Abraham and Klamt, J. Chem. Inf. Comput. Sci. 42, 1320–1331. Zuend, A., Marcolli, C., Luo, B. P., and Peter, T. (2008), A thermodynamic model of mixed organic-inorganic aerosols to predict activity coefficients, Atmos. Chem. Phys. 8, 4559–4593.

6 MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES SONGYAN DU AND LISA A. RODENBURG 6.1. Introduction 6.2. Some Important Semivolatile Organic Compound (SVOC) Classes 6.2.1. Persistent Organic Pollutants 6.2.2. Polycyclic Aromatic Hydrocarbons 6.3. Cycling in the Atmosphere 6.3.1. Atmospheric Deposition 6.3.2. Chemical Reaction of SVOCs in the Atmosphere 6.4. Monitoring Programs 6.5. Sampling and Analysis 6.5.1. Sampling 6.5.2. Analysis 6.5.3. Challenges 6.6. Source Identification 6.6.1. Diagnostic Ratios and Fingerprints 6.6.2. Other Techniques of Source Identification 6.6.3. Receptor Models 6.6.4. Transport Models 6.7. Conclusions

6.1. INTRODUCTION Semivolatile organic compounds (SVOCs) are defined as the substances with vapor pressures roughly between 10
4 and 10
11 atm (101–10
6 Pa) at ambient temperatures (Bidleman 1988). At these vapor pressures, significant fractions of their masses in the atmosphere are found in both the particle and the gas phases. This partitioning has implications

for the compounds’ transport and reactivity, as well as the techniques used to measure them. With such a broad definition, it is no surprise that SVOCs constitute a diverse class of chemicals. Semivolatile organic compounds include a wide variety of anthropogenic and naturally occurring chemicals; a comprehensive review of the major SVOC classes is provided by Lee and Nicholson (1994). Our discussion focuses mainly on the organochlorine compounds and aromatic hydrocarbons, more specifically, polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs), and polychlorinated dibenzo-p-dioxins and -furans (PCDD/Fs). Semivolatile organic compounds have many environmental impacts. Atmospheric chemists are concerned primarily with SVOCs as sinks for OH radicals and possible precursors for ozone production and particle nucleation (Finlayson-Pitts and Pitts Jr. 1997). Public health professionals are concerned with SVOCs because some of them, such as PAHs, have human health impacts (IARC 1994). Environmental chemists are concerned with SVOCs because many of them are persistent organic pollutants (POPs), or are subject to air or water quality standards (AQSs or WQSs) (Table 6.1). This chapter provides background on the importance of SVOCs, and then focuses on the methods used to understand their sources and fate on a local scale. Many of these methods rely on the analysis of large datasets, so that issues of data comparability across various monitoring networks and programs are important. For this reason, this chapter will also briefly discuss the methods used to measure and monitor SVOCs in the atmosphere.

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

149

150

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

6.2. SOME IMPORTANT SEMIVOLATILE ORGANIC COMPOUND (SVOC) CLASSES 6.2.1. Persistent Organic Pollutants International concerns over chlorinated organic pollutants culminated in the signing of the Stockholm Convention on persistent organic pollutants (POPs), a global agreement on POPs that entered force in 2004. In total, 12 substances were identified as POPs under the original Stockholm Convention, including several OCPs [aldrin, dieldrin, dichlorodiphenyltrichloroethane (DDT), endrin, chlordane, heptachlor, mirex, and toxaphene], as well as the industrial chemicals hexachlorobenzene (HCB), PCBs, and PCDD/Fs (UNEP 2001). The vapor pressures of these POPs range from about 10
0.3 Pa (monochlorobiphenyls) to as low as 10
10 Pa (octachlorodibenzo-p-dioxin), such that most of them qualify as “semivolatile.” The criteria that are generally applied in identifying a chemical as a POP are inherent toxicity, bioaccumulation potential, persistence, and susceptibility to longrange atmospheric transport (LRAT). The United Nations Environment Programme (UNEP) has called for actions to limit the production and release of these so-called “dirty dozen” POPs. Because of their LRAT potential, these chemicals may be transported from countries where they are still in use to countries where they are banned, potentially causing transboundary environmental damage and political disputes. In the European Union, the requirements of two international conventions (the 1979 Convention and the Stockholm Convention) have been implemented by Regulation (EC) 850/2004 on POPs. This regulation also amends the EU regulation on POPs and Directive 79/117/EEC, which prohibits the marketing and use of plant protection products containing certain active ingredients. In the United States, the only existing air standards for SVOCs are

typically set by the Occupational Safety and Health Administration (OSHA) and are designed to protect workers from high (generally indoor) acute exposures, so the standards are high. For example, the OSHA limit for the PCB formulation Aroclor 1254 is 500,000,000 pg/m3 (http:// www.osha.gov/pls/oshaweb/owadisp.show_document? p_table¼STANDARDS&p_id¼9992). Polychlorinated biphenyls have never been reported to exceed that level in ambient air. However, in the United States, all the dirty dozen except mirex are subject to water quality standards (WQSs) (http://www.epa.gov/waterscience/criteria/wqctable/index. html; Table 6.1). The federal WQS for the consumption of water and organisms for PCBs is 64 pg/L (http://www. epa.gov/waterscience/criteria/wqctable/index.html). According to Henry’s law constants for PCBs (Schenker et al. 2005; Bamford et al. 2002b), the concentration in air that would be at equilibrium with this water concentration is about 500 pg/m3, a level that is routinely exceeded in urban areas (Simcik et al. 1997; Totten et al. 2004; Sun et al. 2007). Similarly, heptachlor epoxide, dieldrin, and DDTs can sometimes approach or exceed atmospheric concentrations that are in equilibrium with the WQS (Gioia et al. 2005). In such circumstances, the WQS cannot be met without reducing atmospheric emissions, and studying the atmospheric fate and transport of these chemicals is crucial to managing them in aquatic systems. In addition, “new” or “emerging” substances were more recently added to the UNEP’s list of POPs. These new POPs are a- and b-hexachlorocyclohexane (HCH); chlordecone; tetra-, penta-, hexa-, and hepta- bromodiphenyl ethers (BDEs); lindane (c-HCH); pentachlorobenzene; and perfluorooctane sulfonic acid (and its salts and perfluorooctane sulfonyl fluoride). Many of these compounds have vapor pressures that fall in the semivolatile range, including BDEs

TABLE 6.1. National Air and Water Quality Standards and Calculated Air Concentrations in Equilibrium with Water Quality Standards for Some of the SVOCs

Pollutant PCBs Chlordane 4,4’-DDT Dieldrin a-Endosulfan b-Endosulfan Endrin Heptachlor Heptachlor epoxide Toxaphene a

Henry’s Law,a (Pa  m3)/mol 4–100 9.0 1.3 4.5 10. 1.9 0.76 150 3.3 490

Molecular Weight, g/mol

U.S. National Water Quality Standard Water þ Organismb mg/L

188.65–464.2 409.8 354.5 380.93 406.95 406.95 380.92 373.3 389.2 414

0.000064 0.00080 0.00022 0.000052 62 62 0.059 0.000079 0.000039 0.00028

Henry’s law constant from Mackay et al. (1999) except for PCBs from Schenker et al. (2005). http://www.epa.gov/waterscience/criteria/wqctable/index.html. c http://www.osha.gov/pls/oshaweb/owadisp.show_document?p_table¼STANDARDS&p_id¼9992. b

Concentration in Air at Equilibrium with WQS, pg/m3

U.S. OSHA Standards, pg/m3c

500 2,900 120 94 260,000,000 48,000,000 18,000 4,800 51 55,000

1,000,000,000 500,000,000 1,000,000,000 250,000,000 — — 100,000,000 500,000,000 — —

CYCLING IN THE ATMOSPHERE

(10
3–10
7 Pa) (Shoeib et al. 2004) and HCHs (0.1–0.01 Pa) (Monosmith and Hermanson 1996). Many of the POPs noted in the Stockholm Convention are considered potentially toxic for humans. The Agency for Toxic Substances and Disease Registry (ATSDR) provides toxicity profiles for most POPs (http://www.atsdr.cdc.gov). On the basis of these profiles, dioxins, including PCDDs and PCDFs, have been characterized as likely human carcinogens. Some of the non- or mono-ortho-substituted PCBs that share a similar chemical structure and a common mechanism of toxic action (Molina et al. 2000) are also considered to be “dioxin-like.” The health concerns for PCBs include carcinogenicity and a variety of adverse health effects on the immune, reproductive, nervous, and endocrine systems. The adverse health effects associated with the OCPs include the fact that they are probable carcinogens. There is also evidence that they may harm the endocrine, nervous, and digestive systems. Health effects caused by HCB include harm to the liver, immune system, kidneys, and blood. All BDEs are known to have neurological effects and are suspected to affect the immune system. 6.2.2. Polycyclic Aromatic Hydrocarbons Another important class of SVOCs is PAHs, which are produced mainly from combustion processes and petrogenic sources (e.g., coal, oils, fossil fuels). Some PAHs are produced naturally from forest fires and volcanic eruptions, but anthropogenic emissions of PAHs from fossil fuel burning tend to dominate in most areas (Wild and Jones 1995). Petrogenic sources are the main source of PAHs in areas impacted by oil spills and fossil fuel contamination (e.g., by unburned coal), such as shipping ports and areas around oil refineries. The major health concerns for PAHs and substituted PAHs (e.g., nitro-PAHs) are their mutagenicity and carcinogenicity (Atkinson and Arey 1994). Polycyclic aromatic hydrocarbons were the first class of atmospheric pollutants to have been identified as suspected carcinogens. Their carcinogenicity appears to increase with increasing molecular weight. In contrast, acute toxicity apparently decreases with increasing molecular weight (Ravindra et al. 2001, 2008). The International Agency for Research on Cancer (IARC) has classified the carcinogenicity of 49 PAH into a few groups. Three PAHs are considered as probable carcinogens (class 2A): benzo[a]anthracene, benzo[a]pyrene, and dibenzo[a,h] anthracene. Twelve other PAHs are considered as possible carcinogens (class 2B), including benzo[b]fluoranthene, benzo[j]fluoranthene, benzo[k]fluoranthene, and indeno[1,2,3cd]pyrene. Other PAH compounds that can neither be classified as nor excluded as carcinogenic are listed as group 3 (IARC 1994). Polycyclic aromatic hydrocarbons are also major constituents of soot particles and are involved in the production of secondary organic aerosols (Wild and Jones 1995; Dachs and Eisenreich 2000; Vione

151

et al. 2004). Both these types of particles also have human health impacts. For instance, soot can cause tissue irritation and the release of toxic chemicals from scavenger cells (Lighty et al. 2000). Secondary organic aerosols influence human health through the spread of microorganisms and also cause the enhancement of of mutagenicity and allergenicity by nitration of PAHs in soot and proteins in bioparticles, respectively (Fuzzi et al. 2006). Furthermore, both soot and secondary organic aerosols have important implications on climate and climate change, and thus influence human health in this way as well (Fuzzi et al. 2006). Many PAHs undergo oxidation and/or nitration in the atmosphere, generating toxic products (Vione et al. 2006). The PAHs differ from other classic POPs in their shorter atmospheric half-lives (due to faster reactions with OH radical), their strong affinity for sorption to soot carbon, and the influence of ongoing primary sources on their distribution (Nizzetto et al. 2008). They are therefore a complementary group of compounds for investigating the role of atmospheric persistence on environmental partitioning mechanisms of POPs (Nizzetto et al. 2008).

6.3. CYCLING IN THE ATMOSPHERE Atmospheric transport has been identified as the major mode of long-range transport and global dispersal of most legacy SVOCs, such as hexachlorobenzene (HCB), hexachlorocyclohexane (HCH), and PCBs. These chemicals are capable of being transported from source areas to extremely pristine and remote areas, such as the arctic (Risebrough et al. 1976; Bidleman et al. 1989; Simonich and Hites 1995). Chapter 10 deals with the global cycling of POPs in more detail. Localscale atmospheric transport of POPs is also extremely important to understand, and is the focus of this chapter. High atmospheric concentrations lead to large atmospheric deposition fluxes that can be important and sometimes dominant sources of POPs to aquatic systems (Offenberg and Baker 1997; Totten et al. 2004, 2006a). One of the most famous examples is from Chicago, Illinois, where atmospheric deposition of PCBs into the adjacent Lake Michigan is thought to be one of the largest sources of PCBs to the lake (Zhang et al. 1999; Offenberg et al. 2005). The case in Delaware River is another instance where the atmospheric deposition load exceeds the entire total maximum daily load (TMDL) recently established for PCBs (Fikslin and Suk 2003; Totten et al. 2006a). Semivolatile organic compounds that are emitted to ambient air are subsequently removed from the atmosphere by different processes. The atmospheric lifetimes of SVOCs determine their LRAT potential and overall persistence. Their major atmospheric removal processes include deposition and degradation. Even within the same chemical class, the relative importance of these removal processes can be

152

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

very compound-specific. Chapter 10 discusses these processes in more detail. This section therefore includes only a brief overview of these removal processes, with emphasis on their implications for measurement and modeling of SVOCs. 6.3.1. Atmospheric Deposition Atmospheric deposition occurs via three major processes: wet deposition, dry deposition of fine/coarse particles, and gaseous deposition. Wet deposition occurs via two mechanisms: the gaseous absorption of SVOCs into raindrops and the scavenging of particles containing SVOCs by rain and snow (Bidleman 1988). Semivolatile organic compounds can also ad/absorb to rain or fog droplets, which have a high surface area/volume ratio. Ad/absorption onto fog droplets (i.e., fog sequestration) is important near coastal areas. Typically, the particle scavenging mechanism dominates over gas absorption into the droplet due to the relatively high Henry’s law constants of many apolar to weakly polar SVOCs (Poster and Baker 1996; Offenberg and Baker 2002). Scavenging of particles by rain/fog droplets is non-size-discriminatory. In contrast, dry particle deposition preferentially removes large particles with high settling velocities. Unlike wet and dry particle deposition, gaseous deposition is a reversible process that is more correctly referred to as air–surface exchange. The gas/particle partitioning behavior of the SVOC is the primary factor that determines which mode of atmospheric deposition will dominate its overall removal from the atmosphere (see Chapter 5 for a comprehensive look at gas particle partitioning). Any SVOCs that reside primarily in the gas phase are most likely to be removed from the atmosphere via air–surface exchange, while dry particle deposition is typically more important for chemicals that have a strong affinity for aerosols. For example, the majority of PAHs (70%–90%) are bound to suspended particles at ambient temperature (Ravindra et al. 2008), therefore, particle-associated deposition tends to dominate their atmospheric removal. For more polar chemicals, wet deposition dominates their atmospheric removal (Gotz et al. 2008). Studies comparing the different removal mechanisms for major SVOCs have been conducted. For heavier PCBs, PCDD/Fs, and some of the OCPs such as DDT, their atmospheric fates are controlled by particle-associated deposition (Koester and Hites 1992; Lohmann and Jones 2005; Gotz et al. 2008). Wet deposition has been observed to be the most efficient process for the atmospheric removal of highly brominated BDE congeners and PCDD/Fs (Bidleman 1988; Hoff et al. 1996; Baker and Hites 2000; Raff and Hites 2007). In contrast, gaseous transfer is the dominant transfer process for some OCPs such as the HCHs, dieldrin, and lighter PCBs (Hoff et al. 1996). Atmospheric deposition always competes with atmospheric degradation. For example, the less volatile hexa-, hepta-, and octa-PCDDs are removed from the atmosphere primarily via deposition to terrestrial surfaces, but for

the more volatile tetra- and penta-PCDD/Fs, atmospheric degradation is predicted to outweigh deposition (Lohmann and Jones 2005). Quantifying atmospheric deposition is difficult. Much of the monitoring of SVOCs that has been conducted to date has been performed in order to quantify their atmospheric inputs to aquatic ecosystems for the purposes of building loads for water quality models. Therefore these monitoring programs must measure the target SVOCs in the gas, aerosol, and precipitation phases. The methodology must be able to accurately quantify SVOCs in the gas versus aerosol phases, which can be difficult since gas/particle partitioning can change during sampling (see Chapter 5). Similarly, measuring SVOCs in precipitation can be subject to artifacts related to accurately separating dry particle deposition from wet deposition. These artifacts are discussed in more detail in Section 6.5. Assuming that the monitoring program is able to accurately measure SVOC concentrations in the gas, aerosol, and precipitation phases, these concentrations must then be converted to deposition fluxes in order to estimate total atmospheric loads of SVOCs to water bodies or terrestrial surfaces. 6.3.1.1. Dry Particle Deposition. The dry particle deposition flux (Fdry ) is calculated by multiplying the particle-phase SVOC concentration (Cpart ) by the particle phase deposition velocity (vd): Fdry ¼ vd Cpart

ð6:1Þ

Since the settling rate and therefore deposition velocity of particles depends on their size, Equation (6.1) would be more correctly written as n X Fdry ¼ Ci vd;i ð6:2Þ i¼1

where the concentration of SVOC is measured on each particle size fraction (Ci ), and the appropriate deposition velocity (vd,i) for each size range is employed. In practice, this is seldom done. Measurement of SVOCs is typically expensive, and for long-term monitoring studies it is generally not feasible to measure SVOCs in several particle size ranges. Also, detection limits can be a problem when the total particle phase SVOC burden is split into many fractions. Therefore most studies use Equation (6.1) and attempt to identify a characteristic deposition velocity that will reflect an average of the deposition velocities of the various particle size fractions, weighted by the percentage of the total SVOC mass contained on each fraction. However, because different SVOCs have different sources and different gas/particle partitioning behavior, they may reside in different particle size fractions and therefore have different characteristic deposition velocities. Also, the particle size distribution can

CYCLING IN THE ATMOSPHERE

153

TABLE 6.2. Dry Deposition Velocity for SVOCs Associated with Particulate Phase Location

Method

Chicago Tainan City, Taiwan Rural Taiwan Lake Michigan San Francisco Bay

Dry deposition plates Back-calculated from flux Back-calculated from flux Mylar strips, Apezion L grease Estimation based on wind speed, particle size Dry deposition plates Dry deposition plates Bucket Water vessel Dry deposition plates Dry deposition plates Glass jars Dry deposition plates Inverted frisbees and flat-plate samplers Glass jars Derived from measured concentration in vegetation Dry deposition plates

Chicago Japan

Chicago Canadian deciduous forest Taiwan Bloomington and Indianapolis, IN Canadian deciduous forest Canadian spruce needles Turkey (industrial site)

vary in a systematic way, with urban areas having a greater preponderance of large particles, for example (Pirrone et al. 1995a,b; Franz and Eisenreich 1998). This also affects the characteristic deposition velocity. Many studies have attempted to measure this kind of characteristic dry particle deposition velocity for PCBs and PAHs, but the results vary over a wide range from 0.01 to nearly 10 cm/s (Table 6.2) (Lee 1991; Sheu et al. 1996; Franz and Eisenreich 1998; Odabasi et al. 1999; Vardar et al. 2002; Shannigrahi et al. 2005; Rowe 2006). The reported deposition velocities for different chemical classes vary significantly at different sampling locations and with different sampling methods. Despite many attempts to measure dry deposition directly, there is no generally accepted methodology for collecting dry particle deposition. Different kinds of surrogate surfaces including Teflon plates, Petri dishes, dry or diol-coated filters, buckets, pans filled with water, oil-coated glass plates, and greased strips have all been used to measure dry particle deposition (Eisenreich et al. 1981; Bidleman 1988; Lee 1991; Sheu et al. 1996; Franz and Eisenreich 1998; Odabasi et al. 1999; Vardar et al. 2002; Shannigrahi et al. 2005). Semivolatile organic compounds are difficult to sample because they not only partition between the gas and particle phases but can also can revolatilize from the collection surface. The types of surfaces typically used to investigate dry particle deposition are usually designed to be surrogates for water. Very little is known about dry particle deposition of SVOCs to other surfaces. Horstmann and McLachlan (1998) investigated dry deposition of a variety of SVOCs to forest

Deposition Velocity vd, cm/s

Compound

Reference

0.5 0.39 0.68 0.9 0.2

SPCBs SPCBs SPCBs SPCBs SPCBs

Holsen et al. (1991) Lee et al. (1996) Lee et al. (1996) Franz et al. (1998) Tsai et al. (2002)

5.2  2.9 0.98  0.59 1.39  0.82 1.38  1.23 6.7  2.8 0.4–3.7 0.11 0.42 0.2

SPCBs SPAHs SPAHs SPAHs SPAHs SPAHs PAHs Total PCDD/Fs Total PCDD/Fs

Tasdemir et al. (2004) Shannigrahi et al. (2005) Shannigrahi et al. (2005) Shannigrahi et al. (2005) Odabasi et al. (1999) Franz et al. (1998) Su et al. (2007) Shih et al. (2006) Koester and Hites, (1992)

0.8 0.11

PBDEs PBDEs

Su et al. (2007) St-Amand et al. (2007)

4.9  4.1

OCPs

Bozlaker et al. (2009)

canopies and observed that the dry particle deposition velocities were as much as 15 times higher in deciduous versus coniferous forests. Welsch-Pausch et al. (1995) investigated the deposition of PCDD/Fs to fields containing Welsh ray grass and concluded that dry gaseous deposition is the dominant pathway for the less volatile PCDD/Fs to this specific grassland and therefore probably governs the accumulation of these chemicals in the agricultural food chain. Rowe et al. (2007a,b) point out that the reported values of vd for PCBs are log-normally distributed, with most values falling below 0.5 cm/s. These researchers therefore calculated the geometric mean of the available literature values for PCBs, and used this value of 0.5 cm/s to estimate dry deposition loads to forested watersheds. Similarly, Rodenburg et al. (2010) calculated the geometric mean of a series of literature values for dry deposition of PAHs to water to be 0.3 cm/s. Since both the dry deposition velocity and concentrations of pollutants in the environment tend to be lognormally distributed, the uncertainties in the calculated fluxes and loads are not normally distributed. The uncertainties are therefore not symmetric about the mean, and therefore cannot be propagated by standard methods. Nor is the mean necessarily the most appropriate parameter for characterizing the load or flux. The treatment of uncertainty in fate models relying on estimates of dry particle (and as we shall see later, gaseous) deposition is therefore a challenge. Monte Carlo approaches have since been developed to assess uncertainty in these models (Venier and Hites 2008) and other environmental fate models (Schenker et al. 2009).

154

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

6.3.1.2. Wet Deposition. Wet deposition is calculated as Fwet ¼ vr  Cr;VWM

ð6:3Þ

where Cr,VWM is the volume-weighted mean concentration of the SVOC in precipitation and vr is the rain depth. In contrast to vd and vg, vr can be measured relatively accurately, resulting in relatively low uncertainties in wet deposition fluxes and loads. Semivolatile organic compound concentrations in rain tend to vary over many orders of magnitude and are often a function of the amount of rain; small rain events tend to contain high SVOC concentrations due to the efficient scavenging of particles at the beginning of the rain event. For this reason, the volume-weighted mean concentration is used in equation 6.3 (Van Ry et al. 2002). 6.3.1.3. Gaseous Exchange. Gaseous exchange is reversible, and the overall gaseous air/surface exchange can be calculated via Fnet

  Cs ¼ vg  C g
KSA

ð6:4Þ

where Fgas is the gaseous absorption flux, Cg is the concentration of the chemical in ambient air, Cs is the concentration of the chemical on the surface (or in the deposition matrix), and KSA is the equilibrium constant for partitioning of the chemical between air and the surface media. Under extreme disequilibrium situation, specifically, strong fugacity gradient toward the surface such as in remote “uncontaminated” areas, gaseous exchange can be calculated as Fgas ¼ Cg vg

ð6:5Þ

where Fgas is the gas absorption flux, vg is the gaseous deposition velocity, and Cg is the concentration of the SVOC in the gas phase. Measurement of Cs can be simple (i.e., when the matrix is water) or complex, as when the matrix is the wax of a leaf cuticle. For this reason, most monitoring networks measure Cg only and apply Equation (6.5) to calculate only the deposition flux. Similar to dry particle deposition velocities, gaseous exchange velocities are highly uncertain (Bohme et al. 1999) and depend on the receptor surface (water, soil, vegetation). Air–water exchange is probably the best understood of the air–surface exchange processes, but even here there are many competing models that calculate vg as a function of meteorological variables such as wind speed and temperature, as well as the diffusivity and Henry’s law constant of the SVOC (Whitman 1923; Mackay and Yeun 1983; Liss and Merlivat 1986; Wanninkhof et al. 1987; Wanninkhof 1992; Achman et al. 1993; Zhang

et al. 1999; Totten et al. 2003; Frew et al. 2004; Blomquist et al. 2006). Other models seek to describe air–water exchange in flowing waters, where turbulence induced by flow of water over the rough river bottom provides energy for mass transfer (O’Connor and Dobbins 1958; Lamont and Scott 1970; Moog and Jirka 1999a, b). Water parameters such as salinity and the presence of surface films can also have an effect on vg (Downing and Truesdale 1955; Broecker et al. 1978; Schwarzenbach et al. 2003). For air–water exchange calculations, vg is typically calculated as a function of resistances to mass transfer in both the air (va) and water (vw) 1 1 1 ¼ þ vg vw va H 0

ð6:6Þ

where H0 is the dimensionless Henry’s law constant. The coefficients va and vw have been empirically defined on the basis of experimental studies using tracer gases such as CO2 and SF6 (Whitman 1923; Wanninkhof et al. 1987; Wanninkhof 1992; Schwarzenbach et al. 2003), because the mass transfer coefficients for air–water exchange have rarely been measured directly for SVOCs. Differences in diffusivity or Schmidt number between these gases and SVOCs are then used to estimate vg and vw for SVOCs. This approach implicitly assumes that differences in solubility between the various gases are unimportant, an assumption that has been called into question. For example, Blomquist et al. (2006) note that increased solubility can reduce the impact of bubbles on the exchange velocity. Thus even the most widely studied air–surface exchange process (air–water exchange) is not well understood for SVOCs. In addition, these models suggest that the dependence of vg on wind speed is nonlinear and increases dramatically at high wind speeds. Wind speed values are typically lognormally distributed, so vg values and fluxes are also lognormally distributed (Rodenburg et al. 2010). A relatively new approach is the use of the water surface sampler (WSS), which has been used to collect both particleand gas-phase deposition of SVOCs, including PAHs (Odabasi et al. 1999; Tasdemir and Esen 2007) and PCBs (Tasdemir and Holsen 2005). The WSS is used in combination with dry deposition plates, so that the dry particle deposition component of total deposition can be subtracted out and gaseous deposition to the water surface can be calculated and then used to derive vg. The WSS is assumed to capture the deposited particles with 100% efficiency since there is no possibility for particles to bounce off the water surface. The WSS has a surface area of about 0.5 m2, so it is unclear whether results from this approach are applicable in the field. Also, the design of the WSS presumably prevents it from being able to investigate the impact of factors such as breaking waves and natural organic matter on air/water exchange.

CYCLING IN THE ATMOSPHERE

Sampling artifacts associated with the WSS can include loss of particle-phase SVOCs from the WSS filter due to dissolution into the water, or an increase in the particle-phase SVOC flux caused by adsorption onto the filter itself or onto particles on the filter. For relatively soluble SVOCs such as lower-molecular-weight PCBs and PAHs, the first artifact is thought to dominate and the measured deposition flux will therefore be biased low. More recent measurements using the WSS suggest that the vg values derived by the two-film method described above may be too low by a factor of 2, although it is not clear whether this discrepancy is due to some artifact associated with the geometry of the WSS (Odabasi et al. 2001). Other air–surface exchange processes such as air–soil exchange and air–vegetation exchange also play important roles in the atmospheric fate and transport of SVOCs (Schroder et al. 1997; Horstmann and McLachlan 1998; Bohme et al. 1999; Wania and McLachlan 2001; Ould-Dada 2002). Airborne SVOCs can be trapped by forest canopies, which lead to reduced air concentrations and elevated contaminant levels in forest soils. This effect is also observed in global POP fate models, which predict that the world’s forests markedly decrease the long-range atmospheric transport (LRAT) of some SVOCs (Wegmann et al. 2004; Su and Wania 2005). The appropriate estimation of the deposition velocity to the forest is important for the simulation of LRAT of SVOCs. However, relatively few values for gaseous deposition velocities to surfaces other than water have been published [see Rowe et al. (2007b) for a summary]. The limited number of published values appear to be lognormally distributed. Horstmann and McLachlan (1998) have published gaseous deposition velocities for deciduous and coniferous forests. The average summer gaseous deposition velocity was nearly 5 times higher in the deciduous canopy than in the coniferous forest. Later, Su et al. (2007) further confirmed the significance of forest trapping effect for SVOCs despite differences in local climate, canopy composition, and structure. These longer-term studies did not investigate the effects of wind speed or temperature on deposition velocities. Rowe et al. (2007a) have argued that gaseous exchange between air and forested ecosystems is at or near equilibrium, suggesting that these mass transfer coefficients may be of minor importance in modeling. Thus, as with dry particle deposition, SVOC concentrations in the gas phase can be measured with relatively low uncertainty, while the large uncertainties inherent in gaseous deposition velocities introduce large uncertainties into the calculated atmospheric deposition fluxes and loads. Previous investigations of air–water exchange of PCBs (Nelson et al. 1998; Bamford et al. 2002a) have estimated that the inherent uncertainty in fluxes ranges from 40% to 900%, with the majority (88%) of this uncertainty attributed to the uncertainty in vg (Nelson et al. 1998). Many water quality models use a single value of vg for each chemical, despite its

155

dependence on wind speed. For example, the water quality model for PCBs, PAHs, and PCDD/Fs in the New York/ New Jersey Harbor uses air–water exchange mass transfer coefficients that are independent of wind speed (HydroQual 2007). In contrast, the Delaware River PCB TMDL model uses time-variable mass transfer coefficients that are a function of both wind speed and current velocity (Delaware River Basin Commission 2003). As with dry particle deposition velocities, gaseous deposition velocities are lognormally distributed. As noted above, standard methods of propagating uncertainty in gas absorption fluxes are not applicable, and treatment of uncertainty in fate models for SVOCs requires more sophisticated methods such as Monte Carlo analysis (Venier and Hites 2008; Schenker et al. 2009). 6.3.2. Chemical Reaction of SVOCs in the Atmosphere The extent to which SVOCs react in the atmosphere can dictate their long-term environment fate. If the atmospheric degradation reactions are slow compared to the rates of deposition, then a greater proportion of the emitted SVOCs could reach human and terrestrial food chains. Atmospheric SVOCs can be removed by photodegradation, including direct photolysis and indirect photolysis, that is, reaction with photochemically generated radicals. Direct photolysis in the atmosphere of a given compound is faster in the gas phase than when the compound is sorbed on particles (Koester and Hites 1992). Generally, the OH radical is the most important photochemically generated oxidant, and chemical reactions with the OH radical are the dominant loss processes for most organic compounds in the atmosphere (Atkinson 1990). Gas-phase reactions of OH radical with PCDD/Fs (Lohmann et al. 2006), PCBs (Anderson and Hites 1996; Totten et al. 2002; Mandalakis et al. 2003), and PAHs (Kwok et al. 1994) have been investigated in lab and field studies. The OH radical reaction is expected to be a significant removal pathway in the atmosphere for the lighter and intermediate PCB congeners and two- to four-ring PAHs, as well as PCDD/Fs with less than five chlorines because of their preferential partitioning into the gas phase (Anderson and Hites 1996; Brubaker and Hites 1998). The influence of OH radical reactions on the loss of SVOCs can be illustrated by the diurnal measurements of a given source strength as in studies on PCBs (Anderson and Hites 1996; Totten et al. 2002; Mandalakis et al. 2003; MacLeod et al. 2007) and PAHs (Simcik et al. 1997) or by laboratory chamber studies (Atkinson and Arey 1994; Anderson and Hites 1996). In addition, PAHs react with nitrate radical (NO3) at night (Arey et al. 1989; Atkinson and Arey 1994). This is not a major degradation pathway, but is notable because it leads to the formation of mutagenic nitro-PAHs and other nitropolycyclic aromatic compounds including nitrodibenzopyranones (Atkinson and Arey 1994).

156

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

Photolysis effectively occurs to PAHs only in the gas phase or when sorbed to particle surfaces. No evidence has been observed for the direct gas-phase photolysis of two- to four-ring PAHs. Photolysis of nitro-PAHs has been observed under ambient outdoor sunlight conditions and in indoor chambers with black light irradiation (Atkinson et al. 1989; Arey et al. 1990). Studies demonstrated that PAHs and PCDD/F sorbed to atmospheric particles had significantly slower rates of direct photolysis than did gas-phase compounds (Behymer and Hites 1985; Koester and Hites 1992). However, it has also been suggested that direct photolysis of PAHs on particles might play a significant role in their transformation, due to the longer residence time of particle bound PAHs, even though the kinetics of reaction are slower than in the gas phase (Vione et al. 2004) The rate of direct photolysis of gas- and particle-phase PBDEs in the atmosphere has also been studied. Raff and Hites (2007) estimated that in the lower troposphere, wet and dry deposition account for > 95% of the removal of BDE-209, while photolysis accounts for  90% of the removal of gas-phase congeners such as BDE-47. Atmospheric half-lives of low-molecular-weight PCBs (di-, tri-, and tetra-PCB) due to direct photolysis have been estimated to be on the order of weeks (Bunce et al. 1989), while the calculated atmospheric lifetimes of PCB congeners based on their gas-phase reactions with OH radicals range from 3 to 120 days (Sinkkonen and Paasivirta 2000). Photochemical reactions of SVOCs can cause their atmospheric concentrations to vary on a diel cycle. Diel variability in atmospheric SVOC concentrations has important implications for monitoring, as SVOCs are typically measured in 24-h integrated sampling to avoid any systematic bias related to the time of day when the samples are collected. Recent studies have studied the diel (24-h) variation of the concentration of SVOCs including PCBs, PBDEs, and chlordane (MacLeod et al. 2007; Moeckel et al. 2008; Gasic et al. 2009). These studies used a combination of carefully designed and timed experiments and modeling to predict the short-term behavior of SVOCs. These researchers argue that diel variations in atmospheric SVOC concentrations are a function of four factors: temperature-driven volatilization sources, atmospheric stability (boundary-layer dynamics), OH radical reactions, and source types. The type of source can affect the diel variability in SVOC concentrations due to diel variations in emissions. For example, emission rates of PAHs from vehicles will be higher during daytime when more vehicles are on the road.

6.4. MONITORING PROGRAMS Many programs measure atmospheric concentrations of POPs for the purpose of estimating atmospheric deposition fluxes and loads to aquatic ecosystems. As such, these

monitoring programs in the United States are motivated primarily by implementation of the Clean Water Act. Considerations in the design of monitoring networks are . . . . .

Where should monitoring sites be located? How often and for what duration should samples be collected? Which analytes should be measured? What types of sampling methods should be employed? What types of analytical methods should be employed?

Different monitoring networks have approached these issues in different ways. In terms of location, some networks have focused on remote areas to measure background atmospheric deposition signals, while others have targeted urban areas to capture the effects of urban or industrial activity on atmospheric SVOC concentrations. Sampling frequency and duration are often limited by cost and availability of funding. The analyte list is typically determined by the specific research goals, although the sampling and analysis methods used for many SVOCs (PCBs, OCPs, PAHs, and PBDEs) are basically the same. As a result, all these classes of SVOCs can typically be measured for low incremental cost once the samples have been processed for the target analytes. In contrast, measuring PCDD/Fs in the atmosphere often requires longer sample collection times and more rigorous cleanup methods. While sampling methodologies have been standardized largely for active air sampling for SVOCs, the analytical methodologies are constantly improving. In particular, many protocols are moving toward the use of high-resolution mass spectrometry. This technique provides greater sensitivity and selectivity, but at much higher cost. The first large SVOC monitoring network in the United States was the Integrated Atmospheric Deposition Network (IADN), which has been in operation since 1990. It includes five master stations located on remote shorelines of the Great Lakes in the United States and Canada (Hoff et al. 1996; Hillery et al. 1998). Urban sites within Chicago were later added to the network (Basu et al. 2004) (http://www.msc.ec. gc.ca/iadn/stations/chicago_e.html). After successfully fulfilling its first and second implementation plans, the third implementation plan is currently ongoing, covering the years 2005–2010. In addition to the monitoring of “core chemicals” (PCBs, OCPs, and PAHs), chemicals of emerging concern such as PBDEs as well as PCDD/Fs have been added to the list of analytes (Environment Canada and the United States Environmental Protection Agency 2005). The IADN operates on a 12-day sampling schedule, collecting one 24-h integrated sample for gaseous and particle-phase SVOCs every twelfth day. Precipitation is usually collected via integrated 24-day samples. This sampling schedule appears to be adequate for SVOCs, and generates about 100 samples per year per site for SVOC analysis [including blanks and

MONITORING PROGRAMS

QA (quality assurance) samples]. The objectives of the IADN study are to determine the loading of persistent toxic contaminants from the atmosphere to the Great Lakes basin from both urban and regional sources. The IADN has been continuously funded via an agreement between the United States and Canada. Results from the IADN network show that the concentrations of most SVOCs are much higher in the urban area of Chicago than at the remote sites (Basu et al. 2004; Sun et al. 2006). Concentrations of PCBs and OCPs display longterm declines at most IADN sites (Cortes et al. 1998; Buehler et al. 2002; Buehler and Hites 2002; Sun et al. 2006, 2007). There is no declining trend observed for PAHs because of the presence of ongoing sources (Buehler and Hites 2002). Concentrations of most SVOCs at most IADN sites are temperature-dependent, with higher concentrations occurring during warmer periods. This temperature dependence is typically modeled using the Clausius–Clapeyron equation (Simcik et al. 1999b; Carlson and Hites 2005). The Chesapeake Bay Atmospheric Deposition Study (CBADS) (Leister and Baker 1994) was conducted at three nonurban sites along the shoreline of the bay during 1990–1993. Ambient air samples were collected biweekly, and air sampling times ranged from 12 to 24 h. The primary objective of the CBADS study was to provide the best possible estimates of total annual atmospheric loading of a variety of trace elements and organic contaminants directly to the surface waters of the Chesapeake Bay. The CBADS study was not designed to capture long-term time trends in atmosphere contaminant concentrations, and thus the network shut down after sufficient short-term data had been collected. The measured atmospheric concentrations of PAHs and PCBs at the northern and southern regions of Chesapeake Bay are similar and exhibit seasonal variability (Dickhut and Gustafson 1995). Though wet deposition concentrations displayed little seasonal variation, spatial variation of the wet deposition fluxes was observed. Atmospheric deposition was found to be an important source of SVOCs to Chesapeake Bay (Dickhut and Gustafson 1995). On the basis of the IADN experience, the New Jersey Atmospheric Deposition Network (NJADN) was designed to capture both the urban and regional signals of SVOCs by locating monitoring sites in a variety of land use types (Gigliotti et al. 2000; Van Ry et al. 2002; Totten et al. 2004). The NJADN project was designed to monitor the loadings of organic pollutants to adjacent water bodies and to assess spatial and temporal trends in SVOC concentrations. The history of the NJADN illustrates one of the major problems confronting SVOC monitoring networks in the United States—these networks are typically designed to address a local or regional water quality issue, and once data to support that issue are generated, the network generally shuts down. The NJADN was originally (1997–1999) funded by the Hudson River Foundation and consisted of three sites focused on the New York/New Jersey Harbor

157

(Eisenreich 1998). Later (1999–2001) it was funded by the New Jersey Department of Environmental Protection and included nine sites throughout New Jersey and focused on water and air quality statewide, measuring SVOCs as well as trace metals (Reinfelder et al. 2004; Totten et al. 2004). Most recently (2002–present) it has been funded by the Delaware River Basin Commission and includes three sites focused on deposition of PCBs to the Delaware River. The NJADN sites have included varying land-use regimes, including urban/ industrial, suburban, coastal, and rural areas. The sampling schedule and methodology are similar to those of the IADN; gas and particle phase samples are collected for 24 h every twelfth day, and integrated precipitation samples are collected over 24-day periods. Analytes have at various times included PAHs, PCBs, OCPs, trace metals, and mercury. Data from the NJADN have been used to estimate atmospheric deposition loads of PCBs and PAHs to the New York/ New Jersey Harbor (Totten et al. 2004; Gigliotti et al. 2005) and loads of PCBs to the Delaware River (Totten et al. 2006a) in support of the TMDL for PCBs. Like the IADN network, data from the NJADN demonstrate that concentrations of most SVOCs are higher in urban areas, display significant temperature dependence via a Clausius–Clapeyron-type relationship, and are undergoing long-term declines (Totten et al. 2004; Gioia et al. 2005). In the United States, dioxin-like compounds are monitored via the National Dioxin Air Monitoring Network (NDAMN) (Cleverly et al. 2000), which is run by the U.S. Environmental Protection Agency (USEPA). The NDAMN now consists of 35 sites situated in rural and remote locations, including many national parks, across the United States. An ambient air sample is collected for a 4-week period, every 3 months, concurrently at each site (Cleverly et al. 2000). Other monitoring programs for SVOCs exist in the United States, but many are run by state governments and do not publish results in peer-reviewed sources. This makes obtaining data and comparing data between networks difficult. Table 6.3 presents a partial list of some other atmospheric SVOC monitoring programs. As with the networks described above, the expense of measuring SVOCs often means that monitoring networks are operated for only short periods. In Europe, the European Monitoring and Evaluation Program (EMEP) was established as a co-operative program for monitoring and evaluation of the long-range transmissions of air pollutants in Europe (http://tarantula.nilu. no/projects/ccc/index.html). Persistent organic pollutants (POPs) were added to the EMEP’s monitoring program in 1999 (Gusev et al. 2008). The measured POPs include PAHs, PCBs, OCPs, HCB, and HCHs. Currently there are 15 sites monitoring POPs in air and precipitation, with most sites located around the North and Baltic Seas in the Arctic and in the Czech Republic. The monitoring in individual countries is financially supported by national environmental agencies. The EMEP service provides a guidance manual on sampling

158

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

TABLE 6.3. Examples of SVOC Monitoring Programs Monitoring program

Description

California Ambient Dioxin Air Monitoring Program (CADAMP) (http://www.arb.ca.gov/aaqm/qmosopas/ dioxins/dioxins.htm) National Dioxin Air Monitoring Network (NDAMN) (http:// cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid¼54936)

Monitoring for dioxins and PBDEs

2001–2004

California EPA

Monitoring dioxins at rural and nonimpacted locations throughout the United States Heavy metals, POPs, and acidifying compounds from Arctic sites Atmospheric deposition of heavy metals, POPs, and sulfur and nitrogen compounds to the North Sea Investigating airborne contaminants in national park ecosystems using a network of eight primary parks in the western United States Identifying long-range transport and the load of POPs and other organic compounds in remote alpine regions

1998–2002

USEPA

1991–present

Arctic council

1987–present

OSPAR Commission

2002–2008

National Park Service

2005

European Union

Arctic Monitoring and Assessment Programme (AMAP) (http://www.amap.no) Comprehensive Atmospheric Monitoring Programme (CAMP) (http://www.ospar.org)

Western Airborne Contaminants Assessment Project (WACAP) (http://www.nature.nps.gov/air/Studies/air_toxics/wacap. cfm)

Monitoring Network in the Alpine Region for Persistent and other Organic Pollutants (MONARPOP) (http://www.monarpop.at)

and analysis methods. However, individual countries are free to use their own methods if they can be documented to be equally reliable.

6.5. SAMPLING AND ANALYSIS Chapter 12 provides more detail on methods of measuring SVOCs in air. This section therefore summarizes the main methods used in monitoring studies and highlights issues of quality assurance and quality control that can affect the usefulness of the data generated. 6.5.1. Sampling 6.5.1.1. Active Air Sampling. Currently, the high-volume air sampler is the most widely used approach for sampling SVOCs in air. This active sampling approach utilizes a pump that draws air through a filter to retain the particle-phase SVOCs and then through an ad/absorbent media (typically a foam or resin) to retain vapor-phase SVOCs. High-volume air samplers equipped with polyurethane foam (PUF) adsorbents have been used extensively and are recommended by USEPA to study the occurrence and speciation of SVOCs such as pesticides, PCBs, and PAHs in the atmosphere (USEPA 1999a). The EMEP, IADN, CBADS, and NJADN systems all used high-volume air samplers; EMEP, CBADS, and NJADN use PUF as the adsorbent. While IADN switched

Duration

Agency

from PUFs to XAD-2 in 1992 at U.S. sites, Canadian sites still use PUF only. In general, XAD-2 resin has a higher collection efficiency for volatile SVOCs than does PUF, as well as a higher retention efficiency; however, XAD-2 may be associated high blank levels of some compounds, especially PAHs (Franz and Eisenreich 1993). Polyurethane foam cartridges are easier to handle in the field and maintain better flow characteristics during sampling (USEPA 1999a). Some studies use a PUF/XAD/PUF sandwich system that is designed to combine the best features of PUF and XAD sorbents, and can be used in a PUF hi-vol without modifying the sampler (Yao et al. 2006; Primbs et al. 2007). Active sampling with a hi-vol sampler requires expensive equipment, a secure location for sampling, a power source, and a significant manpower investment. Active air sampling typically involves collection of a single sample for 24 h, or at most a few days. Longer sampling times introduce problems of breakthrough of analytes on the polyurethane foam sampling cartridge. This is especially a problem for PCDD/Fs, since their low concentrations in the atmosphere typically necessitate sampling intervals of 48 h or more (Cleverly et al. 2000). Acquiring long-term average SVOC concentrations requires collecting, analyzing, and averaging the data from many samples, thus adding to the cost. Therefore, active sampling is often an impractical and expensive approach for investigating the spatial variability of SVOCs. There are some sampling artifacts associated with the design of the high-volume sampler that can affect gas- versus

SAMPLING AND ANALYSIS

particle-phase measurements (Cotham and Bidleman 1991; Gundel et al. 1995; Peters et al. 2000) One of the notable sampling artifacts is known as “blowoff,” whereby SVOCs that have been collected on particulate matter from ambient air can volatilize off of the particles. This blowoff will lead to enhanced apparent vapor-phase concentrations. Another sampling artifact associated with the high-volume sampler is “blowon,” whereby vapor-phase SVOCs can sorb onto the filter media or the accumulated particulate matter and/or organic matter on the surface of the filter media. This can cause an apparent increase in the particle-phase concentration of SVOCs. A backup filter is sometimes installed in series with the first collection filter to determine the magnitude of the artifact due to gas-phase sorption to the filter collection media (Primbs et al. 2007). This approach is presented in more detail in Chapter 5. Additionally, highvolume samplers can be subject to contamination artifacts. Basu et al. (2000) reported that the foam gasket in the highvolume air samplers can become contaminated with PCBs from the air and then release PCBs back into the sampled airstream. Another active air sampler, the diffusion denuder sampler, has been developed to provide an alternative sampling method to minimize potential sampling artifacts (Gundel et al. 1995). In denuder samplers, vapor-phase SVOCs are removed from the airstream by diffusion onto an adsorbent coating prior to removal of particulate matter by filtration. An adsorbent downstream of the filter then collects any SVOCs that are volatilized from the collected particulate matter. Diffusion denuder samplers (e.g., the high-capacity integrated organic gas and particle sampler (IOGAPS)) can dramatically improve the recovery of lowermolecular-weight gas and particle PAHs such as naphthalene and anthracene (Poor et al. 2004). Potential sampling artifacts with diffusion denuders are loss of fine particulate matter to the surface of the denuder tube, which would result in an apparent higher vapor-phase measurement; desorption of SVOC from particles while in transit through the denuder tube; and breakthrough of volatile analytes to the downstream adsorbent, resulting in an apparent higher particlephase measurement (Peters et al. 2000). Peters et al. (2000) compared the performance of high-volume and diffusion denuder samplers for the measurement of SVOCs. Their results suggested that the difference in the measured gas/ particle partitioning of SVOCs in these two different types of samplers can be viewed as a function of vapor pressures and sampler geometry. The high-volume samplers are suitable for measuring SVOCs with subcooled liquid vapor pressures ( pL ) < 0.2 Pa. Although the diffusion denuder sampler does not appear to suffer from this limitation and could efficiently sample the more volatile SVOCs, it may underestimate the partitioning of the less volatile SVOCs to the particle phase. These issues are presented in more detail in Chapter 5.

159

6.5.1.2. Passive Air Sampling. Passive air sampling (PAS) techniques have been developed more recently to address the need for inexpensive and simple monitoring of SVOCs in the atmosphere. Passive air samplers are chemical accumulators that rely on air currents to deliver chemical to the sampler media, which may consist of semipermeable membrane devices (SPMDs), PUF disks, polymer-coated fibers, polymer-coated glass, or XAD-2 resin. Because passive air samplers do not need a power source, they are more suitable for remote locations away from the electric grid. In general, PAS is a more cost-effective approach, which can provide integrated atmospheric SVOC concentrations over a period of months and assess concentrations in air simultaneously at multiple sites at far lower cost. Polyurethane Foam passive samplers are the most widely used media for the investigation of SVOCs in the atmosphere in PAS studies. The PUF samplers were designed to sorb only gas-phase organics, but particles will also deposit to some extent in the passive sampler. The geometry of the sampler housing generally limits the amount of particles reaching the sampling matrix. The sampler housing can not only help prevent the particle deposition but also reduce the dependence of sampling rate on meterological conditions such as precipitation and wind speed (Pozo et al. 2004). The use of passive sampling methods to monitor atmospheric concentrations has greatly increased. Passive samplers have been used to investigate the vertical (Moreau-Guigon et al. 2007; Li et al. 2009), temporal (Meijer et al. 2003; Motelay-Massei et al. 2005; Moreau-Guigon et al. 2007), and spatial (Meijer et al. 2003; Harner et al. 2006a; Du et al. 2009) distribution of atmospheric POP concentrations. The utility of PASs has been demonstrated not only at the local scale but also at global scale (Jaward et al. 2004, 2005). The theory of passive air sampling is described in detail elsewhere (Muller et al. 2000; Shoeib and Harner 2002; Bartkow et al. 2005), and is briefly summarized here. The passive sampling medium (PSM) is a hydrophobic organic matrix that has a high capacity for organic chemicals. Hydrophobic organic contaminants such as PCBs in the gas phase will preferentially partition into this matrix. The extent to which the organic chemicals are enriched relative to air can be described by a passive sampler medium–air partition coefficient (KSV). It has been shown that KSV is related to the octanol–air partition coefficient (Koa) of the chemical, which is typically very high for SVOCs, leading to a strong driving force for uptake of SVOCs onto the PSM. For example, Koa values for PCBs are on the order of 106–1012 (Harner and Bidleman 1996), 108–1012 for PCDD/Fs (Harner et al. 2000), 109–1012 for BDEs (Harner and Shoeib 2002), and 105–1012 for PAHs (Ma et al. 2010). The exchange of gaseous SVOCs between the ambient air and the passive sampler matrix occurs via diffusion. Molecular diffusion is the exclusive transport mechanism

160

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

for mass transfer at the air–sampler interface and in the sampler. The exchange of gaseous SVOCs between the atmosphere and the passive sampler matrix occurring via diffusion can be described by the equation VS

dCS CS ¼ kO AS ðCV
Þ dt KSV

ð6:7Þ

where VS is the volume of the passive sampling medium, CS and CV are the chemical concentrations in the passive sampler medium and in the bulk air, respectively; AS is the surface area of the passive sampler; KSV is the sampler/air partition coefficient; kO is the overall mass transfer coefficient, and t is time. The mass transfer across the sampler–air interface is simply the net result of resistance in the air boundary layer and in the sampler matrix. According to the Whitman two-film approach, the overall mass transfer coefficient (kO ) can be deduced from the two mass transfer coefficients acting in series, namely, the air-side (kV ) and the passive-sampler-side (kS ) mass transfer coefficients: 1 1 1 ¼ þ kO kV kS KSV

ð6:8Þ

Therefore, substituting Equation (6.8) into Equation (6.7) yields VS

dCS 1 CS ¼ AS ðCV
Þ 1=kV þ 1=kS KSV dt KSV

becomes more significant and the uptake has moved into the curvilinear stage. Finally, the uptake will reach quasi-equilibrium as the fugacity of the chemicals in the sampling matrix and the surrounding air stabilize around the KSV value (i.e., CS/KSV ¼ CV), there is no net uptake and CS becomes constant. The time taken to attain equilibrium is influenced by sampler and analyte characteristics. The differential accumulation equation [Eq. (6.8)] can be solved analytically with certain boundary conditions to provide an exact description of the uptake profile. For a deployed sampler in the field, we can assume that the vaporphase concentration (CV ) is constant, the initial sampler concentration (CS0 ) is zero, and all other parameters with the exception of CS are constant with time. The solution of Equation (6.9) is then  CS ¼ KSV CV


AS 1
exp t KSV VS ð1=kV þ 1=kS KSV Þ ð6:10Þ

This equation is a complete uptake profile of chemicals onto the passive air sampler. The KSV coefficient can be interpreted as the equivalent volume of air that contains the same mass of analyte as one unit volume of passive sampling medium under equilibrium conditions (i.e., KSV ¼ Vair =VS ¼ CS =CV ). Therefore, by analogy and replacing terms in Eq. (6.10), the equivalent air sample volume is

ð6:9Þ

The exchange of chemicals between the passive sampler and the ambient air can be presented in three stages as depicted by Equation (6.9) and illustrated in Figure 6.1. In the beginning of the sampling period when the concentration of chemicals in the PAS is minimal, the elimination rate term (CS /KSV) is negligible and the uptake of organic chemicals is linear. When the chemical builds up in the PAS, CS /KSV



 Vair ¼ KSV VS

    As k v 1
exp
t Vs Ksv

ð6:11Þ

When the diffusion transport is limited by the resistance on the sampler side, kV  kS KSV , Equation (6.10) can be further simplified as follows:   
AS kS CS ¼ KSV CV 1
exp t VS

ð6:12Þ

Amount sequestered

On the other hand, if the transport is limited by the resistance on the air side, that is, kV kS KSV , Equation (6.10) can be expressed as Curvilinear

  
AS kV CS ¼ KSV CV 1
exp t KSV VS Equilibrium partitioning

Linear

time

Figure 6.1. Three-stage uptake curve for passive sampler.

ð6:13Þ

Air-side limitation to chemical exchange has been observed for a range of PCBs (Shoeib and Harner 2002) and for certain PAHs (Muller et al. 2000; Bartkow et al. 2005). As indicated in the Equation (6.11), the equivalent sample volume is a function of KSV . For heavier and more hydrophobic chemicals (i.e., high KSV values), higher equivalent sample volume is obtained because of longer time required to achieve equilibrium. The time required for a chemical that

SAMPLING AND ANALYSIS

has higher KSV to reach equilibrium can be quite long. For instance, for chemicals with KOA greater than 109, the deployment time to reach equilibrium will be greater than 450 days (Shoeib and Harner 2002). Thus it is not necessary, and not always desirable or possible, to deploy passive samplers long enough to achieve equilibrium. The main challenge associated with the use of PAS is the conversion of the mass of analyte on the sample to a concentration in the gas phase, which requires knowledge of the volume of air actually sampled. This volume can be calculated from Equation (6.11) provided that all the necessary parameters are available. The air volume can also be estimated by combining passive and active measurements and thereby calibrating the passive sample mass to the measured concentration from active monitoring. This partially negates the benefits of passive sampling, and it also assumes that all PASs have the same sampled volume, despite different locations and micrometeorology. Another approach to calculating the air concentrations when the passive sampler is not at equilibrium is to assume an arbitrary sampling rate. For instance, a sampling rate of 3–5 m3 of air per day has been assumed in some studies (Pozo et al. 2004; Harner et al. 2006b). Adding depuration compounds [i.e., performance reference compounds (PRCs)] is a more direct way to measure the sampling rate (Pozo et al. 2004). These compounds are typically either isotopically labeled chemicals or unlabeled chemicals that are not present in the atmosphere. These compounds are added to the PAS prior to its deployment to the field, and loss or depuration of the PRCs is then related to the uptake rates of target analytes. Depuration of the initially spiked PRC can be described as CPRC ¼ CPRC;0 expð
ke tÞ

ð6:14Þ

lnðCPRC =CPRC;0 Þ ¼
ke t

ð6:15Þ

or

where CPRC is the concentration of PRC remaining in the passive sampler, CPRC;0 is the initial concentration of PRC added to the passive sampler, and ke is the depuration rate constant. Because uptake of SVOCs is usually airside-controlled, the rate of chemical uptake will be the same as the rate of loss: kU ¼ ke . According to Equation (6.13), the uptake rate kU for chemicals under air-side control can be calculated as follows: kU ¼

A S kV KSV VS

ð6:16Þ

By combining Equation (6.14) and (6.15), we can calculate the air-side mass transfer coefficient (kV ) from the recovery of PRCs initially spiked into the PSM (i.e.,CPRC =CPRC;0 ):

kV ¼ ln

CPRC VS KSV * * CPRC;0 AS t

161

ð6:17Þ

The term kV AS represents the passive air sampling rate. This is a very useful term since it can be used to calculate how much air is sampled by the passive air sampler during the linear uptake stage. Thus, although there are several methods of converting the mass of analyte in the passive samples to gas-phase concentrations, there is still uncertainty involved in this calculation that can limit the utility of the PAS approach. For this reason, when passive air samplers are deployed simultaneously, they are an excellent method of obtaining relative concentrations, but may not be the best choice when absolute concentrations are needed. They are particularly useful for determining “fingerprints” of SVOCs, since the sampling volume is unimportant in this case. For this reason, PAS can be particularly useful for tracking down specific SVOC sources that have unique fingerprints, because such trackdown studies necessarily involve many samplers concentrated in a specific geographic area. An additional limitation of PAS is the relatively low sampling rates (typically a few cubic meters of air per day) of most PAS, requiring long sampling periods, which restricts the temporal resolution that can be achieved. Some researchers (Harner et al. 2003; Farrar et al. 2005a, b) have attempted to overcome this problem by developing rapidly equilibrating sampling media such as polymer-coated plates that can be deployed for as little at 7 days. Passive air samplers have has demonstrated that spatial variability in SVOC concentrations is significant. Passive sampling campaigns in Birmingham (UK), Toronto, and Philadelphia (Harner et al. 2004; Harrad and Hunter 2006; Du et al. 2009) have demonstrated that SVOC sources in urban areas are more patchy and localized than might be expected from IADN and NJADN data, suggesting that the mix of SVOC sources in urban areas is complex. Xiao et al. (2007, 2008) developed a flowthrough sampler (FTS) to collect gaseous and particle-bound SVOCs from large volumes of air by turning it toward the wind and having the wind blow through a porous sampling medium. This sampler provides substantially faster uptake rates than does the typical PAS, while requiring no access to reliable network power. The shorter sampling time resulted from a higher sampling rate compared to PAS, allowing higher temporal resolution. Quantitative relationships between the wind speed outside the sampler and after passage through the PUF were established with a battery-operated data logger and allow the accurate estimation of sampling volumes under conditions of low and high wind speed. However, this wind speed dependence of sampling rate also implies a sampling bias toward times with higher wind speed. This bias can be alleviated by including an annular bypass in the sampler design that helps stabilize the wind speed through the sampler

162

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

(Xiao et al. 2007). Additionally, it was confirmed that even relatively volatile SVOCs, such as naphthalene and anthracene, do not experience serious breakthrough (Xiao et al. 2008). Considering the absence of reliable power supply and lower atmospheric SVOC levels at remote sites, the application of FTS could allow measurement of SVOC concentrations at higher temporal resolution (seasonal or monthly basis). 6.5.1.3. Sampling of Precipitation. Sampling of precipitation can be conducted by the use of wet-only precipitation collectors (Poster and Baker 1996; Molina et al. 2000; Van Ry et al. 2002). The sampler is fitted with a moisture sensor that, when activated, causes the cover on the rain sampler to pivot open to reveal a square stainless-steel collection funnel. The rainwater then feeds through the funnel and then through a column containing XAD-2 resin, which extracts the organics. This method does not separate particle-bound versus dissolved SVOCs. The evaporating rainwater can leave SVOCs on the funnel that can be sampled via wiping it with an agent such as glass wool. This wipe can contain as much as half of all the SVOCs in the rain sample (Van Ry et al. 2002). 6.5.2. Analysis In general, analysis of SVOCs after sampling is commonly based on a laborious analytical sample treatment involving extraction and cleanup, followed by instrument analysis. 6.5.2.1. Extraction. Air samples are traditionally extracted by Soxhlet extraction, a robust and well-established solid– liquid extraction technique. The shortcomings associated with the use of Soxhlet extraction are the consumption of large amount of solvents (250 mL or more) and longer extraction time (16–24 h). A relatively new rapid and effective extraction technique is the accelerated solvent extraction (ASE) system, which extracts samples under elevated temperature and pressure conditions with less solvent (50 mL) and time (20 min) compared with Soxhlet extraction. The extracts remain liquid under high-pressure conditions. Accelerated solvent extraction systems have been successfully used for extraction of SVOCs from a variety of environmental matrices, including soils, sediment, and particulate matter (Bautz et al. 1998; Fitzpatrick et al. 2000; Hussen et al. 2006; Ryno et al. 2006). Because of the concerns that ASE will coextract or dissolve the polymers used for collecting gas-phase SVOCs, the use of ASE for air samples has been restricted until quite recently (Primbs et al. 2008; Genualdi et al. 2009). 6.5.2.2. Cleanup. Analysis of SVOCs almost always involves a cleanup step prior to instrumental analysis for the purpose of removing interfering chemicals and for fractionation. Atmospheric samples, especially gas-phase samples,

contain relatively low levels of interfering compounds, especially as compared to more challenging environmental matrices such as biota or sewage sludge, so cleanup of atmospheric samples is usually easier and less aggressive. The cleanup step is typically based on gravity flow solid– liquid chromatography with different stationary phases. The most commonly used stationary phases include silica, alumina, Florisil, and different types of carbon (Carbosphere, Carbopack, Amoco PX-21, Celite 545). Generally, acid, neutral, and basic silica gels are used to remove polar interferences, and Florisil or alumina columns are used to remove nonpolar interferences. Carbon columns are sometimes used to remove nonpolar interferences or to fractionate analytes on the basis of polarity. For instance, Carbopak/ Celite can be used to separate some of the coplanar PCBs from the mono- and di-ortho-substituted PCBs (USEPA 1999b). Under certain circumstances, pressurized flow [i.e., high-pressure liquid chromatography (HPLC)] can be used to provide specificity for certain congeners and congener groups. For example, USEPA Method 1613 (USEPA 1999b) uses HPLC equipped with two Zorbax-ODs columns in series (DuPont Instruments Division, Wilmington, DE, or equivalent) to separate 2,3,7,8-substituted CDD and CDF isomers. It should be noted that some stationary phases are incompatible with some analytes. For example, Florisil adsorbs PAHs and removes them from the sample, so it is not a good choice when PAHs are target analytes. Sample cleanups for the analysis of SVOCs from air samples typically utilize one of the following: multilayer silica (activated silica, sulfuric acid on activated silica, sodium hydroxide on activated silica), basic alumina, or Florisil. In general, more efficient and rigorous cleanup methods are needed for the analysis of PCDD/Fs compared to other SVOCs because of their ultralow concentrations in air samples and the coexistence of much higher levels of interferences such as PCBs (Lohmann and Jones 2005). The National Dioxin Air Monitoring Network uses a modified version of the USEPA Method 1613 cleanup, consisting of a two-step cleanup with Biosil A silica gel and Amoco PX-21 carbon columns as the stationary phases (Cleverly et al. 2007). The first step removes polar compounds, while the second step separates out the 2,3,7,8-substituted isomers from all the other PCDD/Fs. In contrast, IADN and NJADN use a one-step cleanup method with alumina as the stationary phase (Gigliotti et al. 2000; Buehler et al. 2002), since they do not measure PCDD/Fs. Additionally, chemical treatments are sometimes applied as part of the cleanup. For example, concentrated acid is used to destroy all nonchlorinated compounds for the measurement of PCBs at the EMEP sites in Norway (Eckhardt et al. 2009). Most cleanup methods separate the sample into multiple fractions containing different target analytes. For example, the cleanup method used by the NJADN network uses alumina and elutes two fractions (Gigliotti et al. 2000). The

SAMPLING AND ANALYSIS

first fraction is eluted in hexane and contains the PCBs. The second fraction is eluted in a combination of hexane and dichloromethane and contains the PAHs. Organochlorine pesticides (OCPs) and PBDEs are present in both fractions. This cleanup method therefore allows the measurement of all four of these compound classes from a single procedure. Surrogate compounds, which are typically added to the sample during extraction, are used to track the performance of the extraction and cleanup steps. There is some confusion in terminology here. In Europe, these may be called recovery standards, while the term surrogate standard refers to the standard used for the determination of chemical concentration. In the United States, the standards used in quantitation are typically called internal standards, and surrogate standards are the compounds used to track the losses of analyte that may occur during extraction and cleanup. Here we will use the U.S. terminology. The surrogate compounds may consist of isotopically labeled compounds or compounds that are not present in the environmental samples. For example, some PCB congeners that are not present in most commercial PCB formulations are often used as surrogate standards for the analysis of PCBs by electron-capture detection (ECD). In PCB analysis using USEPA Method 1668B (USEPA 1999b), 39 13 C-labeled PCB congeners are used as performance checks: 3 field standards that are added during sampling, 28 surrogates that are added before extraction, 3 cleanup standards that are added just prior to cleanup, and 5 internal standards added just prior to instrumental analysis. This number of performance standards allows a rigorous tracking of analytical performance, but comes at a price, since 13 C-labeled standards are typically quite expensive. 6.5.2.3. Instrumental Analysis. For PCBs, high-resolution capillary GC coupled with electron-capture detection (ECD) has been a standard method for their measurement in environmental matrixes for decades. A Method 8082 (USEPA 2000) is the standard method for measurement of PCBs by ECD. The ECD method achieves relatively high sensitivity while using a relatively low-cost, lowmaintenance instrument. These advantages led the ECD technique to become the most widely used detection method for the analysis of low levels of halogenated contaminants including PCBs and OCPs. Gas chromatographic ECD is recommended for the analysis of PCBs and OCPs excluding non-ortho-PCBs and toxaphene at part per billion (ppb) levels (Muir and Sverko 2006). Even though GC-ECD is one of the most widely used methods, USEPA Method 8082 is not followed exactly by any of the monitoring networks noted above (IADN, NJADN, CBADS) except for some EMEP sites. The disadvantage of the ECD method is that identification of analytes relies solely on their retention times. Thus coelution of interfering compound with the target analytes can prevent accurate quantification. In the 1990s, the USEPA developed Method 1668 (USEPA 1999b)

163

for measurement of PCBs by high-resolution GC (HRGC) with high-resolution mass spectrometry (HRMS). A minor revision of 1668A, Method 1668B, was published in November 2008 (USEPA 1999b). The HRGC/HRMS method is a more powerful alternative for the determination of SVOCs, including PCBs, organochlorine pesticides, and PCDD/Fs, especially at part per thousand (ppt) levels. High-resolution MS methods have become the preferred methods in many cases because of their very high sensitivity and powerful identification capability. HRGC and HRMS methods generally avoid the problems of ambiguous identification of PCB congeners inherent in the ECD method, although within homologs, many PCB congeners still coelute and are quantified as a sum in Method 1668B. A major obstacle to the adoption of Method 1668B is its cost. The high-resolution GC/MS instrument is at least 10 times more expensive than a standard ECD instrument and is therefore usually beyond the means of most academic labs. In addition, as noted above, 1668B uses a large number of 13 C-labeled PCB congeners as surrogates and internal standards, adding to the cost. As a result, the typical cost for analysis of a single sample by 1668B is approximately $800. For air-monitoring networks that generate as much as 100 samples per site per year, this cost is prohibitive. Thus there is a need for a more cost-effective approach to PCB measurement that still avoids the problem of coelution and ambiguous identification of PCB congeners. The standard analytical method approved by the USEPA for the analysis of PAHs in ambient air (TO-13A) uses capillary gas chromatography with low-resolution mass spectrometry (USEPA 1999b). This approach offers good sensitivity and selectivity. High-resolution MS is not necessary because the concentrations of PAHs in atmospheric samples are typically orders of magnitude higher than those of PCBs. Method TO-13A (USEPA 1999b) uses the isotope dilution approach, which utilizes isotopically labeled internal standards and surrogate compounds. Some of the sites of the EMEP use HPLC with fluorescence detection to measure PAHs (Mano and Schaug 2003). Analysis of PCDD/Fs in environmental matrixes represents one of the biggest analytical challenges owing to a series of problems, including their ultralow concentrations in the environment, the coexistence of a large number of interfering compounds in much higher concentrations, and the need for congener-specific determination to differentiate the toxic congeners from other structurally similar congeners (Eijarrat and Barcelo 2002). Method 1613 is an approved standard USEPA method for determination of the 2,3,7,8substituted PCDD/Fs, which are the most toxicologically important, using isotope dilution by HRGC/HRMS (USEPA 1999b). The air concentrations of 2,3,7,8-TCDD are typically less than 1 fg/m3. Currently, the most sensitive HRGC/HRMS can have sensitivities of <50 fg on column. Therefore, given the detection of  1 fg/m3 and an injection

164

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

volume of 1 mL, about 500 m3 of air has to be sampled to capture a detectable mass of PCDD/Fs and the volume of extraction has to be minimized to  15 mL (Lohmann and Jones 2005). Sensitive HRGC-HRMS is therefore needed to routinely quantify all the toxicologically relevant congeners, despite its high cost. The NDAMN technique uses a modified version of Method 1613 (Cleverly et al. 2007). Because of the high cost of HRMS methods used by USEPA for the standard analysis of PCBs and PCDD/Fs, alternative methods using ion trap mass spectrometry (ITMS) or tandem quadrupole mass spectrometry (MS/MS) have been investigated. Several researchers have used ITMS (Malavia et al. 2004; Barro et al. 2005; Verenitch et al. 2007) or MS/MS (Asher et al. 2007; Hu et al. 2008; Du et al. 2009) for measurement of PCBs. These techniques have been successfully applied to the determination of other halogenated compounds, such as PCBs, PCDDs/Fs, PBDEs, pesticides, and PAHs (Santos and Galceran 2003; Eppe et al. 2004; Malavia et al. 2004), but none of these methods has been standardized. 6.5.3. Challenges Monitoring for SVOCs presents many challenges that limit the amount and quality of data generated. In the United States, monitoring networks have been developed to address local or regional problems, typically related to water quality issues. This has led to a patchwork of monitoring networks, most of which operate for only a short time because of budget constraints. Indeed, the costs associated with monitoring for SVOCs are considerable due to the large amount of labor required for sample processing and the use of capillary GC and/or HRMS methods. These high costs impose limits on the number of monitoring sites, the frequency of sampling, and the number of analytes. Another major challenge for the monitoring of SVOCs is data comparability. Problems with data comparability between SVOC monitoring networks derive from differences in both the sampling process and the subsequent chemical anaysis of collected samples. Although most networks use the high-volume sampler with PUF adsorbent, IADN uses XAD-2 at U.S. sites and PUF at Canadian sites, which may be one reason for the issues of data comparability observed even within the IADN network (Wu et al. 2009). No studies have attempted to assess the level of comparability of data collected by different SVOC monitoring networks in the United States. The various EMEP sites use slightly different sampling, cleanup, and analysis methods (Aas and Breivik 2008). A laboratory intercomparison study conducted in 2000–2002 revealed concerns about data comparability within the EMEP POP network (Mano and Schaug 2003). Data comparability is an important issue that should be addressed in both the United States and Europe.

Data comparability between various passive sampling campaigns designed to measure SVOCs in the atmosphere has not been investigated. As described earlier, deriving absolute SVOC concentrations using passive methods can be problematic. Use of PRCs minimizes the error in the measurement of relative concentrations, which makes PAS a good approach for trackdown of atmospheric SVOC sources. Nevertheless, data comparability between PAS studies in which the samplers were deployed for different lengths of time, in different seasons, or in areas with widely varying meteorology (especially wind speed) is likely to be poor. Use of different adsorbents (e.g., PUF vs. XAD) is also likely to result in poor data comparability. Comparing active air monitoring data with passive data is also likely to be problematic, since the uptake rates of SVOCs in passive samplers are a function of their molecular weight while active air monitoring using high-volume samplers results, at least in theory, in the same uptake rate for all SVOCs. Thus comparing congener profiles measured in PAS to those measures by high-volume sampling, even when PUF is the adsorbent in both cases, is inappropriate unless the varying uptake rates have been taken into account for the PAS samples. For example, Du et al. (2009) attempted to compare PCB congener patterns from active and passive sampling in the Philadelphia area. They found that it was possible to directly compare congener patterns by calculating either an R2 or cos  similarity value. However, systematic differences in congener patterns were observed that could be attributed to the fact that heavier PCBs have higher sampling volumes in passive sampling. These results suggest that the comparison of active versus passive samples that were analyzed via the same methods (i.e., same sample preparation and instrumentation) may be less problematic than comparison of data acquired using different analytical methodologies (i.e., ECD vs. HRMS). Because of the significant analytical challenges associated with measuring SVOCs in atmospheric samples, data comparability between and within studies and/or labs is a serious concern. There is no national network and no consistent set of sampling and analysis guidelines for SVOC sampling. Different laboratories have developed different procedures and may have modified them over time. Furthermore, technical advances have enabled more sensitive analytical methods for the measurement of target analytes, so that networks in operation for many years, such as the IADN and NJADN, may experience data comparability issues between old and new data. All these factors lead to problems when comparing data collected at different points in time and analyzed by different laboratories using different methods. A national or international strategy is needed to address monitoring of SVOCs that fall under the Stockholm Convention, and/or are subject to water quality standards and for which atmospheric deposition is a major source for surface waters. In Europe, monitoring for SVOCs is being conducted

SOURCE IDENTIFICATION

under the umbrella of the EMEP, which was originally founded to measure species such as major ions (http:// www.emep.int/index.html). A similar structure, the National Atmospheric Deposition Program (NADP), exists in the United States. The NADP works with the states to set up monitoring sites for species such as major ions and mercury (http://nadp.sws.uiuc.edu). The NADP also standardizes measurement methods and thereby ensures data comparability between sites. The United States should therefore follow the European example and conduct SVOC monitoring under the umbrella of the NADP. Alternatively, SVOC sampling could be conducted by expanding the NDAMN to include other POPs. The EMEP approach of allowing each country to use its own methods if the methods can be documented to be equally reliable has advantages. It allows each country to use available equipment, keeping costs low, and allows for innovation in methodology. Thus the body overseeing SVOC sampling, whether it is EMEP, NADP, NDAMN or other organization, could provide some flexibility in methodology. At a very minimum, however, the organization overseeing atmospheric SVOC monitoring should provide a common set of calibration standards and reference materials, and should conduct periodic reviews of data comparability.

6.6. SOURCE IDENTIFICATION The results of monitoring studies, which have shown that atmospheric deposition is a significant source of SVOCs in many areas, have highlighted the need to identify and characterize the sources of SVOCs in order to appropriately manage these chemicals in the environment. In order to effectively decrease emissions of atmospheric pollutants, quantitative knowledge of the relative importance of all major sources is needed. Therefore, identifying potential source locations and source types, both quantitatively and qualitatively, can aid in the implementation of policies such as the Clean Air Act and the Clean Water Act. Two major approaches can be employed to evaluate source contributions: source-oriented models and receptor-oriented models. Source-oriented models use emission data and transport calculations to predict pollutant concentrations at specific receptor monitoring locations. This type of model is validated by comparing the predicted spatial and temporal (spatiotemporal) distribution of pollutant concentrations against measured concentrations. Receptor-oriented models infer source contributions by determining the best-fit linear combination of emission source profiles needed to reconstruct the measured chemical concentrations (Schauer et al. 1996). For most SVOCs, their source inventories and source locations are not known, and source-oriented models are not applicable. Therefore, the discussion here is limited to receptor-based models.

165

6.6.1. Diagnostic Ratios and Fingerprints The simplest approach to source identification is to match concentration ratios of compounds, congener profiles, or “fingerprints” to specific sources. This fingerprinting approach has been used extensively for the variety of sources associated with different SVOCs. Table 6.4 summarizes some of the different strategies used to identify atmospheric sources of SVOCs. 6.6.1.1. PCBs. The industrial use of PCBs in the United States was limited primarily to technical mixtures manufactured under the trade name “Aroclor” by Monsanto from 1930s through 1970s (Erickson 1997). The Aroclors are different complex mixtures of PCBs containing 50–80 congeners with overlapping composition. The list of Aroclors include Aroclors 1016, 1221, 1232, 1242, 1248, 1254, 1260, 1262, and 1268. The last two digits of each Aroclor identifier usually represent the chlorine weight percentage, with the exception of Aroclor 1016, which is 41% chlorine by weight and has a fingerprint very similar to that of Aroclor 1242. Aroclors 1242, 1254, 1248, and 1260 were the most widely used, accounting for more than 90% of the U.S. sales of PCBs in 1970 (www.cdc.gov/niosh/ 78127_7). In Europe, PCB formulations were manufactured in several countries under trade names such as Chlophen, Phenoclor, Pyralene, Kanechlor, Fenchlor, and Delor (Erickson 1997). Most of the PCB formulations sold under these trade names were roughly equivalent to one of the Aroclors. If environmental weathering processes have not substantially altered the composition of PCBs in an environmental sample, the observed congener profile in the sample should match the congener profile of a single Aroclor or mixture of Aroclors. Although weathering processes can alter the original Aroclor fingerprint, the PCB fingerprints are often stable enough to allow the determination of likely source Aroclor(s) (Sather et al. 2001). Previous studies have developed different strategies to determine the original Aroclor contamination in the environment. For example, Sather et al. (2001) used a least mean square approach for PCB pattern matching, which calculates the residuals between the amounts of each congener found in the sample and in the Aroclor. A characteristic ratio approach has been proposed by Newman et al. (1998) that utilizes the ratio of certain persistent congers present in specific Aroclors to identify the source Aroclor. Some of the non-Aroclor PCB congeners are also very informative for source identification. The PCB 11 congener is produced during the manufacture of diarylide yellow pigments (Litten et al. 2002) and has been detected in the air at several locations (Choi et al. 2008; Hu et al. 2008; Du et al. 2009). Since PCB 11 is present in only trace amounts in the Aroclors, its presence in the atmosphere is almost

166

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

TABLE 6.4. Different Strategies for Source Identification of Atmospheric SVOCs SOC

Source Apportionment Strategy

Reference

PCB

Individul or mixed Aroclors Non-Aroclor congeners Chiral signature Diagnostic ratio/source marker Radiocarbon analysis Congener/homolog profile of 2,3,7,8-PCDD/Fs Full congener fingerprint Indicatory PCDD/Fs Source markers DDT/DDE ratio TC/CC ratio Chiral signature

Du and Rodenburg (2007) Du et al. (2009) Robson and Harrad (2004); Jamshidi et al. (2007); Asher et al. (2007) Larsen and Baker (2003); Khalili et al. (1995); Nielsen (1996); Lim et al. (1999) Zencak et al. (2007) Bright et al. (1999); Lee et al. (2004)

PAHs PCDD/Fs

OCPs

Ogura et al. (2001); Masunaga et al. (2003); Sundqvist et al. (2009) Lee et al. (2004) Rappe et al. (1990a,b); Buekens et al. (2000); Persson et al. (2007) Sovocool et al. (1977) Qiu et al. (2004) Bidleman et al. 1998, 2004; Gouin et al. (2007)

certainly due to the manufacture and use of diarylide yellow or similar pigments. Since these pigments are the most commonly used yellow pigments in printing inks, PCB 11 sources are ubuitous in the environment. The PCB 209 congener is generated during the production of titanium chloride (Gamboa et al. 1999), but it is also present in the heavy Aroclors (Rushneck et al. 2004) and in foundry wax (Horton 1991). Few monitoring networks attempt to measure PCB 209 precisely because it was present in only small amounts in the Aroclors (Rushneck et al. 2004). Detection of PCB 209 and the related congeners PCBs 206 and 208 as a high proportion of SPCBs therefore may indicate an industrial source. It should also be noted that ferric chloride, produced as a byproduct during the production of titanium chloride, was sometimes sold to wastewater and drinking water treatment plants as a flocculant, carrying its PCB 206/208/209 contamination with it (Du et al. 2008). Thus detection of high proportions of these congeners in environmental samples could also be an indicator of treated wastewater as a PCB source. 6.6.1.2. PAHs. The relative abundance of the various PAH compounds are useful indictors of PAH sources, assuming that these ratios remain the same during their transport from source to receptor site. Yunker et al. (2002) provides a review of parent PAH diagnostic ratios that are most useful in distinguishing the variety of natural and anthropogenic PAH sources. For example, a dominance of chrysene and benzo[k] fluoranthene in PAH signatures can be used as an indicator of coal combustion as a source of PAHs (Khalili et al. 1995). Particulate-phase coronene has been hypothesized to be derived exclusively from traffic-related sources. Thus, regional specific ratios of coronene to benzo[e]pyrene can be used as indicators of traffic sources (Nielsen 1996; Lim et al. 1999). The ratio of 1,7-dimethylphenanthrene to 2,6-dimethylphenanthrene can be used to distinguish the

contribution of wood combustion from other sources (Benner et al. 1995). However, the use of these diagnostic ratios for the identification is not always valid since the observed ratios at receptor sites always deviate to some extent from those measured in source emissions (Schauer et al. 1996; Zhang et al. 2005; Sofowote et al. 2010), which is not a complete surprise considering that the physicochemical properties of these isomers are not always identical and thus that ratios may change during the transport process. Therefore, these ratios must be used with caution for source identification, or at least changes in ratios during transport must be accounted for. An alternative method is to determine a well-defined local source fingerprint as a reference datum for source apportionment of atmospheric PAHs rather than relying on literature values alone. 6.6.1.3. PCDD/Fs. Because of the difficulties associated with measuring PCDD/Fs, most published PCDD/F fingerprints are based either on the congener profiles of only the 17 2,3,7,8-substituted congeners or on homologs (Bright et al. 1999; Lee et al. 2004). Several studies have demonstrated that congener-specific information on all 136 tetra- through octachlorinated PCDDs/PCDFs is more effective for source identification (Ogura et al. 2001; Masunaga et al. 2003; Sundqvist et al. 2009). There have been reports of PCDD/F congener fingerprints associated with different sources (Rappe et al. 1990a,b; Buekens et al. 2000; Persson et al. 2007). The best PCDD/ F markers for some industrial processes, including chlorophenol use and pulp/paper production, are non-2,3,7,8substituted congeners. Therefore, measurement of more than just the 2,3,7,8-substituted congeners significantly increases the probability of success when attempting to identify PCDD/F sources, and relying on USEPA Method 1613 alone to measure PCDD/Fs may not be sufficient when source apportionment is a goal. Since combustion is typically a

SOURCE IDENTIFICATION

major contributor to the PCDD/Fs measured in the atmosphere, sometimes local PCDD/F profiles must be determined for the pattern matching to be successful. Some studies use indicatory PCDD/Fs for source identification. Indicatory PCDD/Fs are identified by calculating the following ratio: P ð Xi = X Þ j P Ratioji ¼ ð6:18Þ ðXi = X Þmin The numerator represents the mass fraction of the ith congener in the emission source j, and the denominator represents the minimum value of the mass fraction of the ith congener among all emission sources. Higher values of this ratio for the indicatory PCDD/Fs indicate that source j contributes more than other emission sources (Lee et al. 2004). Lee et al. (2004) chose the PCDD/F congeners with the three highest ratio values as the indicatory congeners for each type of source. For instance, the indicatory PCDD/Fs for municipal solid-waste incinerators were OCDD with a ratio of 12.6, 1,2,3,4,6,7,8-HpCDD (8.1) and OCDF (6.4). 6.6.1.4. OCPs. Diagnostic ratios are frequently used to identify atmospheric sources of OCPs. Technical chlordane is a mixture of components consisting mainly of transchlordane (TC), cis-chlordane (CC), and trans-nanochlor (TN) in the proportion 1.00/0.77/0.62, respectively (i.e., TC/CC ¼ 1.3) (Sovocool et al. 1977). In the environment, TC degrades more quickly than CC, resulting in a TC/CC ratio <1 for aged chlordanes. Similar to the chlordanes, technical DDT consists of p,p0 DDTand o,p0 -DDT in proportions of 1:0.15 (i.e., p,p0 -DDT/o, p0 -DDT ¼ 6.7). In the environment, p,p0 -DDT is transformed to p,p0 -DDE and p,p0 -DDD. In general, a DDT/DDE ratio of <1 is indicative of aged DDT, whereas a ratio of >1 suggests the presence of fresh sources (Iwata et al. 1993; McConnell et al. 1996; Qiu et al. 2004). 6.6.2. Other Techniques of Source Identification Techniques utilizing radiocarbon analysis and chiral signatures can also be used to indicate specific sources. Classspecific radiocarbon analysis has been applied to investigate the atmospheric sources of PAHs, revealing a significant contribution from the combustion of nonfossil material to atmospheric PAH pollution even in urban and industrialized areas (Zencak et al. 2007). Chiral signatures have been used to identify atmospheric OCP and PCB sources (Bidleman et al. 1998, 2004; Gouin et al. 2007). Chiral compounds are typically produced by chemical processes that generate racemic mixtures. For racemates, the enantiomeric ratios (ERs), which are the concentration of the first eluted enantiomer (A) over the

167

concentration of the second eluted enantiomer (B), are equal to 1. A more widely used metric is the enantiomeric fraction (EF) is calculated as follows: EF ¼

A AþB

ð6:19Þ

For racemates, EF equals 0.5. For pure single enantiomers, EF is either 0 or 1 (Wong et al. 2001a). Physical and chemical processes, such as abiotic degradation, deposition, and volatilization, affect each enantiomer equally, resulting in no change in their EFs. The physical processes of extraction, cleanup, and analysis of samples are also non-enantioselective, which greatly improves data comparability within and among studies. Microbial degradation of these compounds is typically enantioselective, however, causing their residues in soils and water to become nonracemic (EF 6¼ 0.5). Because of these properties, racemic EFs observed in the atmosphere likely indicate fresh primary sources, or those that have not been subjected to microbial degradation, while nonracemic EFs indicate an aged or secondary source. More recent studies have utilized the chiral properties of OCPs such as a-HCH, chlordanes, and heptachlor epoxide to elucidate the relative significance of emissions from secondary sources resulting from soil from that from primary sources (Bidleman et al. 1998, 2004; Gouin et al. 2007). For example, Bidleman et al. (2004) examined the chiral signatures of chlordane since the 1970s in the atmosphere and observed a shift in the chiral signature from racemic to nonracemic, indicating that the relative influence of volatilization from soils (a secondary source) has been growing over the past few decades. This chiral technique is also applicable to PCBs, as a number of individual PCB congeners are both chiral and present at measurable quantities in the environment. The chirality of PCB molecules arises from asymmetric substitution of the phenyl rings and restricted rotation (atropisomerism) around the central C--C biphenyl bond caused by the presence of three or four ortho chlorine atoms. Of the 209 possible PCB congeners, 19 are chiral (Kaiser 1974). Of these, only about half are typically present in environmental samples at measurable concentrations (Wong et al. 2001b). Robson and Harrad (2004) compared the chiral signatures of three PCBs in commercial formulations, outdoor air, and topsoil in the United Kingdom West Midlands and concluded that the measured racemic atmospheric PCBs arise from primary sources, rather than volatilization from soil. Jamshidi et al. (2007) compared the chiral signature of PCBs in outdoor and indoor air and surface soil samples. They concluded that the principal contemporary source of PCBs in the West Midlands is ventilation of indoor air and not volatilization from soil. Asher et al. (2007) examined the PCB chiral signatures in different environmental media in the NY/NJ Harbor Estuary and concluded that the predominant

168

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

atmospheric source of PCBs is likely unweathered local pollution rather than pollution due to volatilization from the estuary. In addition, these researchers found nonracemic PCB signals in phytoplankton, suggesting that the atmosphere is a less important source of PCBs to the food chain than contaminated sediment. 6.6.3. Receptor Models Receptor models seek to identify or locate sources on the basis of data from the receptor site. The simplest of these use wind direction or back trajectories to qualitatively examine the sources associated with individual samples collected at the receptor site. More complex, quantitative models based on back air trajectory analysis, also called hybrid receptor models, have been developed to locate the source regions of atmospheric SVOCs. These hybrid receptor models include the potential source contribution function (PSCF), the concentration weighted field (CWT) and the residence time weighted concentration (RTWC) models. The principles and applications of these hybrid receptor models are discussed below. One challenge encountered in using meteorology to determine SVOC source regions is that the results can be confounded by highly temperature-dependent air–surface exchange processes. This effect can be accounted for through the use of a model of equilibrium air–surface exchange to remove temperature dependence from the dataset prior to trajectory analysis (Subhash et al. 1999). 6.6.3.1. Receptor Models Combining Meteorology with Measured Chemical Data. Several receptor models have been developed to locate atmospheric SVOC sources by combining meteorology with long-term measurements of chemical concentrations. Source locations of SVOCs have been investigated using regression models with local wind speed and direction parameters (Hillery et al. 1997; Cortes et al. 2000) or back air trajectories (Subhash et al. 1999). In all these studies, the back trajectories were calculated using the NOAA (National Oceanic and Atmospheric Administration) HYSPLIT (hybrid single-particle Lagrangian integrated trajectory) model (http://ready.arl.noaa.gov/HYSPLIT.php), a free and user-friendly resource. The HYPSLIT trajectories describe the advection of a single pollutant particle. Although the HYSPLIT trajectories can be augmented with dispersion models, in practice this has not been done because these studies focused on long-range transport. On long length scales (>100 km), dispersion is negligible. The potential source contribution function (PSCF) can be interpreted as a conditional probability describing the spatial distribution of source locations inferred by using trajectories arriving at the sampling site (Ashbaugh et al. 1985; Polissar and Hopke 2001). Construction of the PSCF model is based on the idea that once SVOCs are emitted and incorporated into the air parcel, they can

be transported along the air trajectories to the receptor site. The trajectories are composed of a series of endpoints that are presented as paired latitude and longitude values for each specific time interval being modeled. The PSCF value for a single-grid cell is calculated as PSCFij ¼

mij nij

ð6:20Þ

where mij ¼ the number of trajectories resulting in high concentrations and nij ¼ the total number of trajectories passing through that grid cell. Concentration weighted trajectory (CWT) was developed as a solution to the problem where grid cells having the same PSCF values can result from samples of quite different concentrations, that is, larger sources cannot be differentiated from moderate sources. In CWT, each grid cell has a weighted concentration obtained by averaging sample concentrations that have associate trajectories that crossed that grid cell as follows 1 Cij ¼ PM

M X

l¼1 tijl l¼1

Cl tijl

ð6:21Þ

where Cij is the average weighted concentrations in the grid cell (i; j), Cl is the measured chemical concentration, pijl is the number of trajectory endpoints in the grid cell (i; j) associated with the Cl sample, and M is the number of samples that have trajectory endpoints in grid cell (i; j). This method can help to determine the relative significance of potential sources (Hsu et al. 2003b). One of the problems with PSCF is that each segment of the back trajectory receives an equal weighting in the model, even though pollutants picked up in distant sections of the trajectory will undergo dilution during travel. In order to address the problem of chemical concentrations being equally distributed in each segment of the related trajectory, residence time weighted concentration (RTWC) was developed by redistributing the concentrations via Xkl Nl Xkl Ckl ¼ Cl PNl ¼ Cl
; Xl X j¼1 jl

j ¼ 1; Nl

ð6:22Þ

where Xkl are the mean concentrations of the grid cells that
l is the average are hit by segments k ¼ 1, Nl of trajectory l, X of the mean concentrations of the grid cells transected by the Nl segments of trajectory l. The concentration input for receptor models is computed as 1 C
ij ¼ PM PNl l¼1

Nl M X X

k¼1 tijkl l¼1 k¼1

logðCkl Þtijkl

ð6:23Þ

where tijkl is the residence time of segment k of trajectory l in grid cell (i; j). Equation (6.23) is similar to the CWT method

SOURCE IDENTIFICATION

169

except that RTWC uses logarithmic concentrations. This procedure is repeated until the average difference between the concentration fields of two successive iterations is below 0.5%. The RTWC method is better at estimating the gradient of the concentration field. Although the results of these hybrid receptor models can help indicate the direction or approximate location of the potential sources, none of these models can give either the source profile (i.e. fingerprint) of the potential sources or a quantitative indication of the source strength, which can be achieved by the application of source apportionment techniques.

different sources by solving a system of equations in which the concentrations of specific constituents are described using a linear combination of source profiles (Watson et al. 1990). In contrast to other receptor models, which extract source compositions from the data, CMB models require a priori knowledge of the source signatures for a given area. Another difference between CMB and other receptor models is that CMB is applied separately to each observation, rather than operating on the dataset as a whole (Watson et al. 1990). The fundamental equation of the CMB model is

6.6.3.2. Source Apportionment. Source categories and the contribution of each source category to the measured atmospheric burden of SVOCs can be quantitatively assessed by receptor-based source apportionment methods including chemical mass balance (CMB) and principal-component(s) analysis (PCA), followed by multiple linear regression (MLR), Unmix, and positive matrix factorization (PMF). These multivariate source apportionment methods extract information about a source’s contribution to the total contaminant burden on the basis of the variability of contaminant fingerprints measured in a large number of samples. In general, receptor modeling relies on conservation of mass and conducts a mass balance analysis that can be applied to identify and apportion the sources (Hopke et al. 2006). Assuming that n air samples are analyzed for m chemical species emitted from p independent sources and that these species are conservative, the mass balance on each species can be expressed as

k¼1

xij ¼

P X

fik gkj þ eij

ði ¼ 1; 2; . . . ; m;

k ¼ 1; 2; . . . ; nÞ

k¼1

ð6:24Þ where xij is the measured concentration of the ith species in the jth sample, fik is the concentration of the jth species in material emitted by source p, gkj is the contribution of the pth source to the ith sample, and eij is the error item, caused by either analytical uncertainty or variations in the sources’ composition (Henry 1997) Equation (6.24) is the fundamental equation underlying the receptor models discussed below. The concentration Cij is subject to random error and fik to random variations. This equation can be solved by a variety of ways depending on what information is available. If the sources that contribute to the receptor concentrations can be identified and their compositional patterns measured, then only the contributions of the sources to each sample need to be determined. These calculations can be performed by a CMB model. Chemical mass balance models use the chemical compositions of receptor concentrations (i.e., chemical fingerprints) to estimate the contributions of

xij ¼

P X

fik gkj

ð6:25Þ

Chemical mass balance equations are solved by an effective variance least-squares estimate, in which the weighted sum of squared differences (x2k ) between the measured and modeled concentrations is minimized (Watson et al. 1984): x2j

" # P m ðxij
pk¼1 fij gjk Þ2 1 X ¼ ði
pÞ i¼1 Vij

ð6:26Þ

where Vij ¼ s2 Eij þ

p X ðSkj Þ2 s 2 Fik

ð6:27Þ

k¼1

where sEij is one standard deviation of the measured exposure concentration of compound i for sample j and sFik is one standard deviation of the measured fraction of compound i from source k. In many cases, the sources are either unknown or the compositions of sources have not been determined, so that it is desirable to identify the potential sources and estimate the contribution of the different type of sources to the receptor sites from the data. Factor analysis methods, including PCA, PMF, and Unmix, are generally employed to solve this kind of problem. These models can be summarized by writing Equation (6.24) in matrix notation X ¼ G FþE ðm nÞ

ðm pÞ

ð6:28Þ

ðp nÞ

where the data matrix (X) can be described as the product of the factor loading matrix (G) and the factor score matrix (F), with an error matrix (E). The G and F matrices represent source contributions and compositions, respectively. The elements of the residue matrix E, which can be derived from Equation (6.24), are defined as follows: eij ¼ xij


p X k¼1

fik gkj

ð6:29Þ

170

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

The goal of factor analysis is then to minimize the error with respect to G and F. Differences among the three factor analysis models include their treatment of errors and uncertainty in the data, and their ability to incorporate physical constraints on factor profiles and source contribution estimates. Principal-component(s) analysis is the most commonly used factor analysis method. The PCA results are generally determined using an eigenvector analysis of a correlation matrix (Hopke 1985). In PCA, the data are transformed into a dimensionless standardized form Zik ¼

Xik
X
i si

ð6:30Þ

where E
i is the arithmetic mean concentration for chemical i over all samples and si is its standard deviation. The PCA model can be written as Zik ¼

p X

Gij Hjk

ð6:31Þ

j¼1

or in matrix notation: Z ¼ GH

ð6:32Þ

The factor loading matrix (G) can be determined from an eigenvector decomposition of the matrix of pairwise correlation coefficients for the m compounds: R¼

1 ZZ T m
1

ð6:33Þ

If L represents an m m matrix whose columns are the eigenvectors of Rand l (a diagonal matrix of its eigenvalues), the matrix G is: G ¼ Ll2

ð6:34Þ

With G determined, H can be calculated by inverting Equation (6.32). The PCA method does not employ a nonnegativity constraint, so the determined source profiles and source contribution estimates can be negative, which is not physically possible and can cause problems when one tries to identify or interpret the resolved factors. Better physical interpretation can be achieved with the performance of orthogonal varimax rotation, which rotates the predetermined principal components while maintaining that the individual components remain orthogonal to each other (Yu and Chang 2001). Because factor scores given by PCA are correlated with, but not proportional to, source contributions, quantitative estimates of source contributions and profiles associated with each factor have to be derived

separately, by coupling with the multiple linear regression (MLR) analysis. The lack of a nonnegativity constraint is a significant limitation of both PCA/MLR and CMB. Newer factor-based approaches have been developed, including Unmix and PMF to address the disadvantages inherent in PCA. In contrast to PCA, these methods place restrictions on the possible source impact solutions to require them to meet certain physical constraints (e.g., nonnegative source impacts). Unmix was developed for the USEPA by Henry (Henry and Kim 1990; Henry et al. 1994; Henry 1997). The most recent version is EPA Unmix 6.0 (http://www.epa.gov/heasd/ products/unmix/unmix.html). Unmix was developed by incorporating the method of graphical ratio analysis for composition estimates (GRACE) into source apportionment by factors with explicit restrictions (SAFER). The GRACE method is a mathematical procedure used to generate pairwise scatterplots of measured chemical concentrations at receptor sites, which provide physical constraints for input to SAFER—a multivariate receptor model based on PCA. The GRACE/SAFER method is designed to estimate factor profiles by a combination of graphical analysis and multivariate receptor modeling. The unique features of the Unmix model include the ability to replace missing data and to estimate large numbers of sources [the current (as of 2010) limit is 15] using duality concepts applied to receptor modeling. Furthermore, Unmix can use additional constraints supplied by the user. Unmix provides a relatively coarse means of downweighting outliers, through two adjustable weighting parameters. Positive matrix factorization is an advanced factor analysis method, described in detail by Paatero and Tapper (1994). In contrast to PCA, which factors the standardized (dimensionless) data, PMF factors the data directly. It solves Equation (6.24) by minimizing the weighted sum of squares of the differences (i.e., eij) between the observations (X) and the model (GF), weighted by the measurement uncertainties (sij): Q¼

n X m  2 X eij i¼1 j¼1

sij

ð6:35Þ

Some special features inherent in the PMF technique distinguish it from other factor analysis methods. For instance, unlike PCA and CMB, the built-in nonnegativity constraints in PMF minimize the ambiguity caused by rotating the factors. Unlike that of Unmix, the PMF input includes a point-by-point estimate of uncertainty, allowing the user to downweight data points that are missing, close to, or below the detection limits, and outliers. For example, missing data can be replaced with the geometric mean concentration and then assigned a higher uncertainty, often 3 times the uncertainty of other, similar data points

SOURCE IDENTIFICATION

171

TABLE 6.5. Application of Different Receptor Models for Source Apportionment of Atmospheric SVOCs Chemical

Source Apportionment Models

PCB

PSCF PSCF, CWT, RTWC PMF/PSCF

CMB Factor analysis/CMB PAH

CMB PCA/MLR

PSCF PCA/MLR, PMF, Unmix PCA with orthogonal varimax rotation PCA/APCA PCA/PMF

OCPs

PSCF

PCDD/Fs

PCA

PCA/cluster analysis

Description of the Study

Reference(s)

Applying PSCF model to locate source of PCBs in Great Lakes and Chicago area Comparison of use of three hybrid receptor models to locate source of PCBs in Chicago air PMF used to identify source types of PCBs; PSCF used to identify geographic source regions of measured PCBs in Philadelphia and Chicago Elucidate Aroclor pattern for measured atmospheric PCB in New Jersey Using factor analysis to identify sources of particle-phase PCBs and contribution from each source as apportioned by CMB Applying CMB model to evaluate contribution of different sources for measured ambient PAHs in urban air Combination of PCA/MLR enables identification and quantitative evaluation of contribution from identified sources

Hafner and Hites (2003), Hsu et al. (2003a) Hsu et al. (2003b)

PSCF used to estimate source region of PAHs to Great Lakes area Comparison of application of three different source apportionment method for identifying source of PAHs in urban air Identification and quantification of main sources of PAHs in Brazil Identification and quantification of the PAHs source to air of Beijing Application of PCAwas abandoned because of excessively negative values and the application of PMF identify and quantify the main source of PAHs to an urban–industrial area in Canada PSCF is used to estimate the source regions of OCPs to Great Lakes and other parts of North America Application of PCA and cluster analysis to identify sources responsible for measured PCDD/Fs in different ambient air in Taiwan and Spain Identification of main sources affecting ambient PCDD/F levels in an industrial area in Spain

(Paatero 2003). Data points that are reported as below detection limits are often replaced with either one-half the detection limit or a random number between 0 and the detection limit, and assigned three times higher uncertainty (Paatero 2003). In addition, the PMF model output includes both source fingerprints and their quantitative contributions to each sample, allowing an estimation of source strength. 6.6.3.3. Results of Receptor Modeling. Table 6.5 summarizes the application of the receptor models discussed above for the source apportionment of SVOCs. One concern regarding the application of these techniques to SVOCs is their tendency to partition between gas and particle phases (Larsen and Baker 2003). This problem can be overcome by using the sum of the measured gas- and particle-phase concentrations as the model input. Even when this approach is used, the source apportionment technique may identify the particle phase as a source (Du and Rodenburg 2007). In

Liu et al. (2003), Du and Rodenburg (2007) Totten et al. (2006b) Cetin et al. (2007) Pistikopoulos et al. (1990), Yang and Chen (2004) Simcik et al. (1999a,b); Harrison et al. (1996); Larsen and Baker (2003) Hafner and Hites (2003) Larsen and Baker (2003) Dallarosa et al. (2005) Zhang et al. (2009) Sofowote et al. (2010)

Hafner and Hites (2003), Hoh and Hites (2004) Lee et al. (2004)

Mari et al. (2008)

addition, the loss of the source signatures by destruction of SVOCs via degradation reactions also complicates source apportionment studies (Larsen and Baker 2003). Some of the SVOCs, such as the more reactive PAHs, will degrade before being brought to the receptor site by air parcels since their atmospheric residence times are less than the duration of the back trajectory (Hafner and Hites 2003). Larsen and Baker (2003), however, argue that the effect of atmospheric degradation on the composition of the source profile is negligible compared with the more significant differences between the fingerprints of various sources. The PSCF approach has been used to identify source regions of atmospheric SVOCs in several studies (Hafner and Hites 2003; Hsu et al. 2003a; Hoh and Hites 2004; Du and Rodenburg 2007). Although PSCF has been quite successful in locating sources of particulate matter and ozone precursors (Ashbaugh et al. 1985; Cheng et al. 1993; Malm et al. 1994), the application of PSCF to SVOC sources has been less

172

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

successful. As mentioned above, SVOC concentrations display strong urban to rural gradients, implying that the sources are located in urban areas and that they are rapidly diluted and/or removed from the atmosphere. Therefore to effectively locate sources, models must have fine spatial resolution, (e.g., 1 km). On this spatial scale, dispersion is important, and PSCF models that rely on Lagrangian HYSPLIT back trajectories that do not incorporate dispersion are not effective. In addition, as a probabilistic model, the PSCF models require datasets with a convincing number of measurements to gain statistic significance, but for the many reasons noted above, SVOC datasets tend to be small, less than a few hundred samples at most. The largest datasets come from monitoring networks such as IADN and NJADN, but these networks collect integrated 24-h samples. This sampling scheme is not ideal for PSCF modeling because meteorology, and therefore back trajectories, can change dramatically on this timescale. Thus application of PSCF to SVOCs has met with limited success. Hafner and Hites (2003) were able to identify the Boston–Washington (DC) Corridor and Upper Hudson River as possible source regions for PCBs to the eastern Great Lakes, and Du and Rodenburg (2007) were able to demonstrate that different PCB fingerprints measured in Camden, New Jersey arise from different parts of nearby Philadelphia. However, PSCF is more effective in determining sources of SVOCs such as OCPs that tend to be dominated by long-range sources (Hafner and Hites 2003). The CMB approach (Pistikopoulos et al. 1990; Yang and Chen 2004) has been used to evaluate the contributions of different urban PAH sources, including coal combustion, diesel combustion, domestic heating, and municipal waste combustion. Totten et al. (2006b) applied a CMB-type model to PCB concentration data from Bayonne and Jersey City (NJ) to elucidate the Aroclor pattern of the atmospheric PCBs and to determine whether periodic spikes in PCB concentrations were due to emissions of an identifiable Aroclor. This model employed a least-squares fit of the congener pattern to a linear combination of the four main Aroclors, with a built-in nonnegativity constraint. The CMB approach is also useful in interpreting the factors extracted by other models, such as PCA and PMF (Du et al. 2008). The PCA approach is the most widely used multivariate statistical technique in atmospheric sciences (Table 6.5), and it has been successfully applied in the source apportionment studies of a variety of SVOCs. Ravindra et al. (2008) provide a review of the application of PCA to the source apportionment of PAHs. The combination of PCA with orthogonal varimax rotation provided better physical interpretation for the source identification of measured PAHs in Brazil (Dallarosa et al. 2005). The combination of PCAwith MRLA has been employed for the source apportionment of atmospheric PAHs (gas þ particle) in the urban and adjacent coastal atmosphere of Chicago/Lake Michigan (Simcik

et al. 1999a) and in Birmingham (UK) (Harrison et al. 1996). In these studies, the combination of PAH data with measurements of inorganic pollutants such as metals and anions provided more powerful tracers of emission sources than PAH data alone. Principal-component(s) has also been used for the source identification for PCDD/Fs (Lee et al. 2004; Mari et al. 2008). Other studies are highlighted in Table 6.3, but this is by no means a comprehensive list. As a more advanced factor analysis model, PMF has been successfully implemented in many source apportionment studies, including identification of atmospheric sources of particulate matter and volatile organics (Paterson et al. 1999; Anderson et al. 2001, 2002; Polissar and Hopke 2001). The application of PMF to source apportionment of SVOCs is still relatively new. Larsen and Baker (2003) compared PCA/MLRA, Unmix, and PMF for source apportionment of atmospheric PAHs. The PMF method demonstrated a better ability to resolve more complete information from the input data than did the other models, which is consistent with the conclusion drawn by Anderson et al. (2002) from a source apportionment study of personal exposure measurements for toxic volatile organic compounds. The combination of two receptor modeling methods, PMF and PSCF, provides an effective way to identify atmospheric SVOC sources and their likely locations (Liu et al. 2003; Du and Rodenburg 2007). One of the most serious constraints on the use of factor analysis methods to apportion SVOC sources is the size of the dataset. In general, the data matrix used for factor analysis should have more samples (n) than analytes (m), because a small number of samples cannot contain enough information to constrain the concentrations of a large number of analytes. The advent of more sensitive methods for measuring SVOCs has resulted in datasets containing large numbers of analytes, but the cost of those methods often constrains the number of samples. For example, it is now possible to measure all 209 PCB congeners by USEPA Method 1668B, but datasets containing more than 200 such samples are rare, indeed. Combination of several small datasets could dramatically improve the utility of the various factor analysis approaches, but issues of data comparability have made this approach largely unworkable. As datasets become larger, they have the potential to yield greater and more detailed information about sources, but to extract all that the dataset has to offer, more sophisticated factor analysis methods are needed. Thus PCA is starting to give way to more sophisticated methods such as PMF as datasets become larger and more detailed. 6.6.4. Transport Models The detection of many POPs in polar areas far from sources prompted the development of model-based approaches to assess the long-range transport potential (LRTP) of SVOCs,

CONCLUSIONS

including calculating a chemical’s characteristic travel distance (Scheringer 1996, 1997; Bennett et al. 1998; Beyer et al. 2000) or persistence and spatial range (Scheringer 1996). These models are discussed in more detail in Chapter 10, and therefore are only briefly mentioned here. Multimedia mass balance models such as Globo-POP (Wania and Daly 2002) and Berkeley-Trent (BETR) (MacLeod et al. 2001) take into account the influence of a chemical’s environmental phase distribution on its ability to be transported over long distances. The characteristic travel distance is calculated by estimating how atmospheric degradation and deposition limit the LRT of a substance in the atmosphere. The spatial range approach accounts for oceanic transport and for emissions into media other than air. An alternative approach to assess LRTP is Globo-POP, which is a zonally averaged multimedia model of the global fate of POPs (Wania and Mackay 1995). This model calculates the interphase and meridional transfer of SVOCs in the atmosphere and surface ocean with consideration of loss processes in all media. This dynamic model also allows for changes in emissions and environmental parameters over time. GloboPOP has been employed to simulate the long-term global fate of a-HCH and PCBs (Lakaschus et al. 2002; Wania and Daly 2002). The BETR model is fugacity-based, describing the kinetics of exchange of contaminants between different environment compartments (MacLeod et al. 2001). The EMEP project has developed the multicompartment POP transport model (MSCE-POP) (Gusev et al. 2008). It consists of nested hemispheric (50 50 km resolution) and regional (2.5 2.5 resolution) models and is, among other things, designed to estimate transboundary transport of POPs (Gusev et al. 2005). On a regional scale, Diamond et al. (2001) developed the multimedia urban model (MUM) to estimate the fate of SVOCs in urban environments by incorporating mechanisms of chemical transport and transformation that are important in urban systems. This model is based on the steady-state, level III fugacity model of Mackay and Peterson (1991). This model is unique in its treatment of impervious surfaces that are associated with urbanization. Some studies have applied a simple radial dilution model (McDonald and Hites 2003; Zhu and Hites 2006) or Gaussian diffusion models (Qiu and Hites 2008) to describe the spatial variations in SVOC concentrations. Both of these models assume a point source and no degradation of the chemical during transport. These models have been used to successfully explain the spatial distributions of toxaphene, BDEs, and dechlorane in North America. On this large spatial scale, the assumption of a point source may be justified, and degradation may be assumed to be negligible. Du et al. (2009) applied the Gaussian diffusion model to atmospheric PCB levels measured within 20 km of Philadelphia, and found that the model adequately described the spatial variability in an empirical sense, even though the assumption of a single point

173

source was not justified. The model was judged to be useful in interpolating PCB concentrations between monitoring sites, a task that may be necessary for modeling atmospheric deposition in water quality models. None of these SVOC transport models explicitly incorporates meteorology. As described above, MacLeod et al. (2007) included a limited treatment of meteorological conditions (including temperature, wind speed, atmospheric mixing height) into a mass balance model to interpret the diel variability in atmospheric SVOC concentrations. Many sophisticated mesoscale models are available to predict meteorological fields on local and regional scales, including the regional atmospheric model system (RAMS; http://rams. atmos.colostate.edu/), Mesoscale Model 5 (MM5; http:// www.mmm.ucar.edu/mm5/mm5-home.html), the advanced regional prediction system (ARPS; http://www.caps.ou.edu/ ARPS/), and the weather research and forecasting (WRF) model (http://www.wrf-model.org/index.php). The meteorology generated by these models can then be used in conjunction with a transport model to predict the movement of chemicals in the troposphere. These types of models have generally not been applied to SVOCs because of the long sampling times typically used in their measurement. Over these long timescales (e.g., 24-h integrated sampling), meteorological parameters such as wind speed and direction can change dramatically. Perhaps the only example of the use of a sophisticated atmospheric transport model to predict SVOC concentrations and examine source locations is the study by Totten et al. (2006b), who used RAMS to generate local meteorology simulations and then applied the hybrid particle and concentration transport (HYPACT) model to model transport of PCBs emitted from hypothetical sources. This study was able to rule out a confined disposal facility as a significant source of atmospheric PCBs to Bayonne, New Jersey, and estimated that the total atmospheric emissions of PCBs from New York City total at least 300 kg per year. Models such as RAMS can predict meteorological fields on a very fine spatial scale, down to a few meters. Their computational cost, however, limits their application on a temporal scale. Typical studies simulate meteorological fields for a week at most. As a result, typical data collected by air monitoring networks (i.e., one 24-h integrated sample every twelfth day) is particularly ill-suited for the application of mesoscale transport models.

6.7. CONCLUSIONS Contaminant SVOCs are ubiquitously emitted into the atmosphere, and are widely distributed, contaminating pristine areas. For this reason, various strategies have been devised to measure and model them in the environment. Because of the diverse physicochemical properties of SVOCs in the atmosphere, these various compounds must be dealt with

174

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

individually. Their physicochemical properties, combined with the types and locations of emissions and local meteorology lead to large spatial and temporal variability in SVOC concentrations. Temporally, concentrations vary according to diel and seasonal cycles, and also sometimes exhibit longterm decline. Spatially, SVOC concentrations can drop by a factor of 2 over distances of just a few kilometer and are strong functions of urbanization. Thus the choice of sampling location, frequency, and duration must take into account all of these factors in the context of the specific research goals of the sampling campaign. More research into the spatio temporal variation of SVOCs is needed. To characterize temporal variability, measurement of SVOCs on shorter timescales (<24 h) is needed. Shortening the length of sample collection using high-volume samplers lowers the mass collected and can lead to problems with detection limits. Nevertheless, shorterduration sampling is needed to better characterize diel variations and for input into sophisticated transport models that explicitly incorporate meteorology (see discussion below). To better characterize spatial variations in SVOC concentrations, especially in urban areas, PAS has proved to be an effective tool, but the uncertainty in the sampled volume must be reduced for these methods to yield credible absolute concentrations. Because of this uncertainty, PAS studies are most useful in trackdown studies that rely on fingerprinting techniques. The PAS studies done to date suggest that spatial variations in SVOC concentrations are extremely patchy, especially in urban areas (Harner et al. 2004; Harrad and Hunter 2006; Du et al. 2009). Passive air sampler studies with even higher spatial densities are needed to determine just how spatially variable SVOC concentrations are. More information is also needed on the vertical concentration gradients of SVOCs, which is an important parameter in transport models that are coupled with sophisticated meteorological models such as RAMS and WRF. In order to be useful in these types of high-resolution atmospheric models, concentrations obtained from PAS must be correct in an absolute sense, however, so uncertainty in the sampled volume must be reduced. Cost of analysis is a major impediment to the collection of large and comprehensive datasets, and high costs generally mean that studies must focus on characterizing either spatial or temporal variability, but not both. In this regard, combining active (for temporal resolution) and passive (for spatial resolution) sampling strategies can be very effective, but care must be taken to ensure data comparability between the two sampling approaches. To date, most sampling campaigns have been either short in duration and focused on answering specific research questions, or long-term monitoring programs in which data are acquired for environmental management goals, but are then stretched for use in studies aimed at source apportionment and/or trackdown, or to answer more fundamental

questions about cycling of SVOCs among environmental compartments. Monitoring networks generate large datasets, but care must be used in interpreting them, due to issues of data comparability between and even within studies. As data on atmospheric SVOC concentrations are increasingly collected to address issues related to the Stockholm Convention and the Clean Water Act in the United States, the need for standardized measurement methods and data comparability studies grows. These standardized methods should incorporate the development of lower-cost analytical methods for trace organics such as PCBs and PCDD/Fs. We suggest that SVOC monitoring should be conducted in the United States under the umbrella of either the NADP or the NDAMN. This would help to standardize protocols and therefore provide data comparability between sites. Many monitoring networks have been founded for the specific goal of providing information about the loads of SVOCs to nearby waters, and in many cases these networks have produced large high-quality datasets on atmospheric concentrations of SVOCs. What is lacking, however, are accurate deposition velocities that can convert concentrations to atmospheric deposition fluxes. There remains significant uncertainty in dry deposition velocities (vd) as well as gaseous deposition velocities (vg). New approaches to the measurement of these two parameters are needed. A variety of source apportionment models and transport models are available to predict SVOC sources and concentrations in the atmosphere, but application of these models is hampered by insufficient data (a problem that is related to the high cost of data acquisition) and issues of data comparability across labs and sampling approaches. Standardizing data collection methods would improve data quality and allow more confident application of these models. Source apportionment models perform best with very large datasets, so having comparable data collected across several networks would greatly enhance the effectiveness of these models in identifying atmospheric SVOC sources. As datasets grow larger and more detailed, the more sophisticated source apportionment models such as PMF will supplant PCA. In the future, it is clear that atmospheric modeling of POPs must move beyond fugacity-based models and crude source–receptor models such as PSCF to more sophisticated models that explicitly incorporate meteorology, such as those used to forecast ozone concentrations. An example of a widely used air quality model that predicts concentrations of ozone, particulates, toxics, and acid deposition is the community multiscale air quality (CMAQ) model (http:// www.cmaq-model.org/), which relies on WRF or MM5 to generate meteorology. This type of model could be adapted to predict SVOC concentrations. The main obstacle to use of these types of models is paucity of data. As mentioned above, better characterization of temporal and spatial variability is needed to provide enough data to constrain these models, which have spatial resolution that is finer by at least an order

REFERENCES

of magnitude than the current fugacity-based models. Meteorological models such as WRF can have resolution in both the horizontal and vertical directions as fine as a few meters, although this level of resolution is often neither necessary nor desirable, due to high computing costs. The resolution of the transport model that is coupled to the meteorological model can be much coarser, to better match the resolution of the measure SVOC concentrations. The vertical concentration profile is particularly important in understanding transport of the chemical across the atmospheric boundary layer, above which it is more likely to undergo long-range transport (Stenchikov et al. 2005, 2006). Especially because urban areas are major sources of atmospheric SVOCs, modeling the transport of these chemicals across the boundary layer on the 50 km horizontal “urban” spatial scale will further our understanding of LRAT of SVOCs to arctic regions. Vertical concentration gradients may also be useful in identifying source locations (Totten et al. 2006b). High-resolution transport models require an type of dataset entirely different from those gathered by monitoring networks. They are best applied to datasets with many short-duration samples collected over a short timescale, at most a few days. This chapter has focused primarily on the “old guard” SVOCs (PCBs, PAHs, and OCPs), because source apportionment and atmospheric modeling of these compounds is most advanced. The new list of POPs most recently approved under the Stockholm Convention includes compounds such as perfluorooctane sulfonic acid (and its salts and perfluorooctane sulfonyl fluoride) that are more polar than the “dirty dozen.” Other emerging environmental contaminants such as pharmaceuticals and personal care products (PPCPs) and hormones have vapor pressures that fall within the SVOC range, but are again more polar. These more polar compounds will require different sampling and analysis methods (Van Ry et al. 2002; Xie and Ebinghaus 2008), and will no doubt pose other, unidentified challenges. REFERENCES Aas, W. and Breivik, K. (2008), Heavy metals and POP measurements, 2006, in EMEP Co-operative Programme for Monitoring and Evaluation of the Long-Range Transmission of Air Pollutants in Europe, Norwegian Institute for Air Research, Kjeller, Norway. Achman, D. R., Hornbuckle, K. C., and Eisenreich, S. J. (1993), Volatilization of polychlorinated biphenyls from Green Bay, Lake Michigian, Environ. Sci. Technol. 27, 75–87. Anderson, M. J., Miller, S. L., and Milford, J. B. (2001), Source apportionment of exposure to toxic volatile organic compounds using positive matrix factorization, J. Exposure Anal. Environ. Epidemiol. 11, 295–307. Anderson, M. J., Daly, E. P., Miller, S. L., and Milford, J. B. (2002), Source apportionment of exposures to toxic volatile organic

175

compounds: II. Application of receptor models to team study data, Atmos. Environ. 36, 3643–3658. Anderson, P. H. and Hites, R. A. (1996), OH radical reactions: The major removal pathway for polychlorinated biphenyls from the atmosphere, Environ. Sci. Technol. 30, 1756–1763. Arey, J., Atkinson, R., Zielinska, B., and McElroy, P. A. (1989), Diurnal concentrations of volatile polycyclic aromatic hydrocarbons and nitroarenes during a photochemical air pollution episode in Glendora, California, Environ. Sci. Technol. 23, 321–327. Arey, J., Atkinson, R., Aschmann, S. M., and Schuetzle, D. (1990), Experimental investigation of the atmospheric chemistry of 2-methyl-1-nitro-naphthalene and a comparison of predicted nitroarene concentrations with ambient air data, Polycyclic Aromat. Compd. 1, 33–50. Ashbaugh, L. L., Malm, W. C., and Sadeh, W. D. (1985), A residence time probability analysis of sulfur concentrations at Grand Canyon National Park, Atmos. Environ. 19, 1263–1270. Asher, B. J., Wong, C. S., and Rodenburg, L. A. (2007), Chiral source apportionment of polychlorinated biphenyls to the Hudson River atmosphere and food web, Environ. Sci. Technol. 41, 6163–6169. Atkinson, R. (1990), Gas-phase tropospheric chemistry of organic compounds: A review, Atmos. Environ. 24A, 1–41. Atkinson, R. and Arey, J. (1994), Atmospheric chemistry of gas-phase polycyclic aromatic hydrocarbons: Formation of atmospheric mutagens, Environ. Health Perspect. 102, 117–126. Atkinson, R., Aschmann, S. M., Arey, J., Zielinska, B., and Schuetzle, D. (1989), Gas-phase atmospheric chemistry of 1and 2-nitronaphthalene and 1,4-naphthoquinone, Atmos. Environ. 23, 2679–2690. Baker, J. E. and Hites, R. A. (2000), Is combustion the major source of polychlorinated dibenzo-p-dioxins and dibenzofurans to the environment? A mass balance investigation, Environ. Sci. Technol. 34 2879–2886. Bamford, H. A., Ko, F. C., and Baker, J. E. (2002a), Seasonal and annual air-water exchange of polychlorinated biphenyls across Baltimore Harbor and the Northern Chesapeake Bay, Environ. Sci. Technol. 36, 4245–4252. Bamford, H. A., Poster, D. L., Huie, R. E., and Baker, J. E. (2002b), Using extrathermodynamic relationships to model the temperature dependence of Henry’s law constants of 209 PCB congeners, Environ. Sci. Technol. 36, 4395–4402. Barro, R., Ares, S., Garcia-Jares, C., Liompart, M., and Cela, R. (2005), Sampling and analysis of polychlorinated biphenyls in indoor air by sorbent enrichment followed by headspace solidphase microextraction and gas chromatography-tandem mass spectrometry, J. Chromatogr. A 1072, 99–106. Bartkow, M. E., Booij, K., Kennedy, K. E., Muller, J. F., and Hawker, D. W. (2005), Passive air sampling theory for semivolatile organic compounds, Chemosphere 60, 170–176. Basu, I., O’Dell, J. M., Arnold, K., and Hites, R. A. (2000), A source of PCB contamination in modified high-volume air samplers, Environ. Sci. Technol. 34, 527–529. Basu, I., Hafner, W. D., Mills, W. J., and Hites, R. A. (2004), Differences in atmospheric persistent organic pollutant

176

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

concentrations at two locations in Chicago, J. Great Lakes Res. 30, 310–315. Bautz, H., Polzer, J., and Stieglitz, L. (1998), Comparison of pressurized liquid extraction with Soxhlet extraction for the analysis of polychlorinated dibenzo-p-dioxins and dibenzofurans from fly ash and envrionmental matrices, J. Chromatogr. A 815, 231–241. Behymer, T. D. and Hites, R. A. (1985), Photolysis of polycyclic aromatic hydrocarbons adsorbed on simulated atmospheric particulates, Environ. Sci. Technol. 19, 1004–1006. Benner, B. A., Wise, S. A., Currie, L. A., Klouda, G. A., Klinedinst, D. B., Zweidinger, R. B., Stevens, R. K., and Lewis, C. W. (1995), Distinguishing the contributions of residential wood combustion and mobile source emissions using relative concentrations of dimethylphenanthrene isomers, Environ. Sci. Technol. 29, 2382–2389. Bennett, D. H., McKone, T. E., Matthies, M., and Kastenberg, W. E. (1998), General formulation of characteristic travel distance for semivolatile organic chemicals in a multimedia environment, Environ. Sci. Technol. 32, 4023–4030. Beyer, A., Mackay, D., Matthies, M., Wania, F., and Webster, E. (2000), Assessing long-range transport potential of persistent organic pollutants, Environ. Sci. Technol. 34, 699–703. Bidleman, T. F. (1988), Atmospheric processes: Wet and dry deposition of organic compounds are controlled by their vapor-particle partitioning, Environ. Sci. Technol. 22, 361–367. Bidleman, T. F., Wong, F., Cecilia, B., Sodergren, A., BrorstromLunden, E., Helm, P. A., and Stern, G. A. (2004), Chiral signatures of chlordanes indicate changing sources to the atmosphere over the past 30 years, Atmos. Environ. 38, 5963–5970. Bidleman, T. F., Patton, G. W., Walla, M. D., Hargrave, B. T., Vass, W. P., Erickson, P., Fowler, B., Scott, V., and Gregor, D. J. (1989), Toxaphene and other organochlorines in Arctic Ocean fauna— evidence for atmospheric delivery, Arctic 42, 307–313. Bidleman, T. F., Jantunen, L. M. M., Wiberg, K., Harner, T., Brice, K. A., Su, K. Falconer, R. L., Leone, A. D., Aigner, E. J., and Parkhurst, W. J. (1998), Soil as a source of atmospheric heptachlor epoxide, Environ. Sci. Technol. 32, 1546–1548. Blomquist, B. W., Fairall, C. W., Huebert, B. J., Kieber, D. J., and Westby, G. R. (2006), DMS sea-air transfer velocity: Direct measurements by eddy covariance and parameterization based on the NOAA/COARE gas transfer model, Geophys. Res. Lett. 33, L07601. Bohme, F., Welsch-Pausch, K., and Mclachlan, M. S. (1999), Uptake of airborne semivolatile organic compounds in agricultural plants: Field measurments of interspecies variability, Environ. Sci. Technol. 33, 1805–1813. Bozlaker, A., Muezzinoglu, A., Odabasi, M. (2009), Processes affecting the movement of organochlorine pesticides (OCPs) between soil and air in an industrial site in Turkey, Chemosphere 77, 1168–1176. Bright, D. A., Cretney, W. J., MacDonald, R. W., Ikonomou, M. G., and Grundy, S. L. (1999), Differentiation of polychlorinated dibenzo-p-dioxing and dibenzofuran sources in coastal

British Columbia, Can., Environ. Toxicol. Chem. 18, 1097–1108. Broecker, H., Petermann, J., and Siem, W. (1978), The influence of wind on CO2 exchange in a wind-wave tunnel, including the effects of monolayers, J. Mar. Res. 36, 595–610. Brubaker, W. W. and Hites, R. A. (1998), OH reaction kinetics of polycyclic aromatic hydrocarbons and polychlorinated dibenzo-p-dioxins and dibenzofurans, J. Phys. Chem. A 102, 915–921. Buehler, S. S. and Hites, R. A. (2002), Integrated atmospheric deposition network, Environ. Sci. Technol. 36, 355A–359A. Buehler, S. S., Basu, I., and Hites, R. A. (2002), Gas-phase polychlorinated biphenyl and hexachlorocyclohexane concentrations near the Great Lakes: A historical perspective, Environ. Sci. Technol. 36, 5051–5056. Buekens, A., Cornelis, E., Huang, H., and Dewettinck, T. (2000), Fingerprints of dioxins from thermal industrial processes, Chemosphere 40, 1021–1024. Bunce, N. J., Landers, J. P., Langshaw, J., and Nakal, J. S. (1989), An assessment of the importance of direct solar degradation of some simple chlorinated benzenes and biphenyls in the vapor phase, Environ. Sci. Technol. 23, 213–218. Carlson, D. L. and Hites, R. A. (2005), Temperature dependence of atmospheric PCB concentrations, Environ. Sci. Technol. 39, 740–747. Cetin, B., Yatkin, S., Bayram, A., Odabasi, M. (2007), Ambient concentrations and source apportionment of PCBs and trace elements around an industrial area in Izmir, Turkey, Chemosphere 69, 1267–1277. Cheng, M., Hopke, P. K., Barrle, L., Rippe, A., Olson, M., and Landsberger, S. (1993), Qualitative determination of source regions of aerosol in Canadian High Arctic, Environ. Sci. Technol. 27, 2063–2071. Choi, S., Baek, S., Chang, Y., Wania, F., Ikonomou, M. G., Yoon, Y., Park, B., and Hong, S. (2008), Passive air sampling of polychlorinated biphenyls and organochlorine pesticides at the Korean Arctic and antarctic research stations: Implications for long-range transport and local pollution, Environ. Sci. Technol. 42, 7125–7131. Cleverly, D. H., Ferrario, J., Byrne, C., Riggs, K., Joseph, D., and Hartford, P. (2007), A general indication of the contemporary background levels of PCDDs, PCDFs, and coplanar PCBs in the ambient air over rural and remote areas of the United States, Environ. Sci. Technol. 41, 1537–1544. Cleverly, D. H., Winters, D., Ferrario, J., Riggs, K., Hartford, P., Joseph, D., Wisbith, T., Dupuy, A., and Byrne, C. (2000), The national dioxin air monitoring network (NDAMN): Measurements of CDDs, CDFs and coplanar PCBs at 18 rural, 8 national parks, and 2 suburban areas of the United States: Results for the year 2000, Organohalogen Compd. 56, 437–440. Cortes, D. R., Basu, I., Sweet, C. W., and Hites, R. A. (2000), Temporal trends in and influence of wind on PAH concentrations measured near the Great Lakes, Environ. Sci. Technol. 34, 356–360. Cortes, D. R., Basu, I., Sweet, C. W., Brice, K. A., Hoff, R. M., and Hites, R. A. (1998), Temporal trends in gas-phase concentrations

REFERENCES

of chlorinated pesticides measured at the shore of the Great Lakes, Environ. Sci. Technol. 32, 1920–1927. Cotham, W. E. and Bidleman, T. F. (1991), Laboratory investigtions of the partitioning of organochlorine compounds between the gas phase and atmospheric serosols on glass fiber filters, Environ. Sci. Technol. 26, 469–478. Dachs, J. and Eisenreich, S. J. (2000), Adsorption onto aerosol soot carbon dominates gas-particle partitioning of polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 34, 3690–3697. Dallarosa, J. B., Teixeira, E. C., Pires, M., and Fachel, J. (2005), Study of the profile of polycyclic aromatic hydrocarbons in atmospheric particles (PM10) using multivariate methods, Atmos. Environ. 39, 6587–6596. Delaware River Basin Commission (2003), PCB Water Quality Model for Delaware Estuary (DELPCB), Delaware River Basin Commission, West Trenton, NJ. Diamond, M. L., Priemer, D. A., and Law, N. L. (2001), Developing a multimedia model of chemical dynamics in an urban area, Chemosphere 44, 1655–1667. Dickhut, R. M. and Gustafson, K. E. (1995), Atmospheric inputs of selected polycyclic aromatic hydrocarbons and polychlorinated biphenyls to Southern Chesapeake Bay, Mar. Pollut. Bull. 30, 385–396. Downing, A. L. and Truesdale, G. A. (1955), Some factors affecting the rate of solution of oxygen in water, J. Appl. Chem. 5, 570–581. Du, S. and Rodenburg, L. A. (2007), Source identification of atmospheric PCBs in Philadelphia/Camden using positive matrix factorization followed by the potential source contribution function, Atmos. Environ. 41, 8596–8608. Du, S., Belton, T. J., and Rodenburg, L. A. (2008), Source apportionment of polychlorinated biphenyls in the Tidal Delaware River, Environ. Sci. Technol. 42, 4044–4051. Du, S., Wall, S. J., Cacia, D., and Rodenburg, L. A. (2009), Passive air sampling for polychlorinated biphenyls in the Philadelphia metropolitan area, Environ. Sci. Technol. 43, 1287–1292. Eckhardt, S., Breivik, K., Li, Y. F., Mano, S., and Stohl, A. (2009), Source regions of some persistnet organic pollutants measured in the atmosphere at Birkenes, Norway, Atmos. Chem. Phys. Discuss. 9, 6597–6610. Eijarrat, E. and Barcelo, D. (2002), Congener-specific determination of dioxins and related compounds by gas chromatography coupled to LRMS, HRMS, MS/MS and TOFMS, J. Mass Spectrom. 37, 1105–1117. Eisenreich, S. J. (1998), Atmospheric Deposition of PCBs, PAHs, Trace Metals and Nitrogen to the Hudson River Estuary. Final Report to the Hudson River Foundation. Eisenreich, S. J., Hollod, G. J., and Johnson, T. C. (1981), Atmospheric pollutants in natural waters, in Eisenreich, S. J. ed., Ann Arbor Science, Ann Arbor, MI, pp. 425–444. Environment Canada and the United States Environmental Protection Agency (2005), Atmospheric Deposition of Toxic Substances to the Great Lakes: IADN Results Through 2005.

177

Eppe, G., Focant, J., Pirard, C., and Pauw, E. (2004), PTV-LV-GC/ MS/MS as screening and complementary method to HRMS for the monitoring of dioxin levels in food and feed, Talanta 63, 1135–1146. Erickson, M. D. (1997), Analytical Chemistry of PCBs, Lewis Publishers. Farrar, N. J., Harner, T. J., Sweetman, A. J., and Jones, K. C. (2005a), Field calibration of rapidly equilibrating thin-film passive air samplers and their potential application for low-volume air sampling studies, Environ. Sci. Technol. 39, 261–267. Farrar, N. J., Harner, T., Shoeib, M., Sweetman, A., and Jones, K. C. (2005b), Field deployment of thin film passive air samplers for persistent organic pollutants: A study in the urban atmospheric boundary layer, Environ. Sci. Technol. 39, 42–48. Fikslin, T. J. and Suk, N. (2003), Total Maximum Daily Loads for Polychlorinated Biphenyls (PCBs) for Zones 2–5 of the Tidal Delaware River, Report to USEPA regions II and III. Finlayson-Pitts, B. J. and Pitts J. N. Jr., (1997), Tropospheric air pollution: Ozone, airborne toxics, polycyclic aromatic hydrocarbons, and particles, Science 276, 1045–1052. Fitzpatrick, L. J., Zuloaga, O., Etxebarria, N., and Dean, J. R. (2000), Environmental applications of pressurized fluid extraction, Rev. Anal. Chem. 19, 75–122. Franz, T. P. and Eisenreich, S. J. (1993), Wet deposition of polychlorinated-biphenyls to Green Bay, Lake-Michingan, Chemosphere 26, 1767–1788. Franz, T. P. and Eisenreich, S. J. (1998), Snow scavenging of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in Minnesota, Environ. Sci. Technol. 32, 1771–1778. Franz, T. P., Eisenreich, S. J., Holsen, T. M. (1998), Dry deposition of particulate polychlorinated biphenyls and polycyclic aromatic hydrocarbons to lake michigan, Environ. Sci. Techno. 32, 3681–3688. Frew, N. M., Bock, E. J., Schimpf, U., Hera, T., HauAyecker, H., Edson, J. B., McKenna, S. P., Uz, B. M., and J€ahne, B. (2004), Air-sea gas transfer: Its dependence on wind stress, small-scale roughness, and surface films, J. Geophys. Res. 109, C08s17. Fuzzi, S., Andreae, M. O., Huebert, B. J., Kulmala, M., Bond, T. C., Boy, M. Doherty, S. J., Guenther, A., Kanakidou, M., Kawamura, K., Kerminen, V.-M., Lohmann, U., Russell, L. M., and P€ oschl, U. (2006), Critical assessment of the current state of scientific knowledge, terminology, and research needs concerning the role of organic aerosols in the atmosphere, climate, and global change, Atmos. Chem. Phys. 6, 2017–2038. Gamboa, J. A., Bohe, A. E., and Pasquevich, D. M. (1999), Carbochlorination of TiO2, Thermochim. Acta 334, 131–139. Gasic, B., Moeckel, C., MacLeod, M., Brunner, J., Scheringer, M., Jones, K. C., and Hungerbuhler, K. (2009), Measuring and modeling short-term variability of PCBs in air and characterization of urban source strength in Zurich, Switzerland, Environ. Sci. Technol. 43, 769–776. Genualdi, S. A., Killin, R. K., Woods, J., Wilson, G., Schmedding, D., and Massey Simonich, S. L. (2009), Trans-Pacific and regional atmospheric transport of polycyclic aromatic hydrocarbons and pesticides in biomass burning emissions to western North America, Environ. Sci. Technol. 43, 1061–1066.

178

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

Gigliotti, C. L., Dachs, J., Nelson, E. D., Brunciak, P. A., and Eisenreich, S. J. (2000), Polycyclic aromatic hydrocarbons in the New Jersey coastal atmosphere, Environ. Sci. Technol. 34, 3547–3554. Gigliotti, C. L., Totten, L. A., Offenberg, J. H., Dachs, J., Reinfelder, J. R., Nelson, E. D., Glenn, T. R., and Eisenreich, S. J. (2005), Atmospheric concentrations and deposition of polycyclic aromatic hydrocarbons to the mid-Atlantic east coast region, Environ. Sci. Technol. 39, 5550–5559. Gioia, R., Offenberg, H. J., Gigliotti, C. L., Totten, L. A., Du, S., and Eisenreich, S. J. (2005), Atmospheric concentrations and deposition of organochlorine pesticides in the US Mid-Atlantic region, Atmos. Environ. 39, 2309–2322. Gotz, C. W., Scheringer, M., MacLeod, M., Wegmann, F., and Hungerbuhler, K. (2008), Regional differences in gas-particle partitioning and deposition of semivolatile organic compounds on a global scale, Atmos. Environ. 42, 554–567. Gouin, T., Jantunen, L., Harner, T., Blanchard, P., and Bidleman, T. (2007), Spatial and temporal trends of chiral organochlorine signatures in Great Lakes air using passive air samplers, Envrion. Sci. Technol. 41, 3877–3883. Gundel, L. A., Lee, V. C., Mahanama, K. R. R., Stevens, R. K., and Daisey, J. M. (1995), Direct determination of the phase distributions of semi-volatile polycyclic aromatic hydrocarbons using annular denuders, Atmos. Environ. 29, 1719–1733. Gusev, A., Rozovskaya, O., Shatalov, V., Sokovyh, V., Aas, W., Breivik, K., and Halse, A. K. (2008), Persistent Organic Pollutants in the Environment, EMEP status report, 3/2008. Gusev, A., Mantseva, E., Rozovskaya, O., Shatalov, V., Strukov, B., Vulykh, N., Aas, W., and Breivik, K. (2005), Persistent Organic Pollutants in the Environment, EMEP status report, 3/2005. Hafner, W. D. and Hites, R. A. (2003), Potential sources of pesticides, PCBs, and PAHs to the atmosphere of the Great Lakes, Environ. Sci. Technol. 37, 3764–3773. Harner, T. and Bidleman, T. F. (1996), Measurements of octanol-air partition coefficients for polychlorinated biphenyls, J. Chem. Eng. Data 41, 895–899. Harner, T. and Shoeib, M. (2002), Measurements of octanol-air partition coefficients (KOA) for polybrominated diphenyl ethers (PBDEs): Predicting partitioning in the environment, J. Chem. Eng. Data 47, 228–232. Harner, T., Green, N. J. L., and Jones, K. C. (2000), Measurements of octanol-air partition coefficients for PCDD/Fs: A tool in assessing air-soil equilibrium status, Environ. Sci. Technol. 34, 3109–3114. Harner, T., Farrar, N. J., Shoeib, M., Jones, K. C., and Gobas, F. A. P. C. (2003), Characterization of polymer-coated glass as a passive air sampler for persistent organic pollutants, Environ. Sci. Technol. 37, 2486–2493. Harner, T., Shoeib, M., Diamond, M., Stern, G., and Rosenberg, B. (2004), Using passive air samplers to assess urban-rural trends for persistent organic pollutants. 1. Polychlorinated biphenyls and organochlorine pesticides, Environ. Sci. Technol. 38, 4474–4483. Harner, T., Shoeib, M., Gouin, T., and Blanchard, P. (2006a), Polychlorinated naphthalenes in Great Lakes air: Assessing

spatial trends and combustion inputs using PUF disk passive samplers, Environ. Sci. Technol. 40, 5333–5339. Harner, T., Shoeib, M., Diamond, M., Ikonomou, M., and Stern, G. (2006b), Passive sampler derived air concentrations of PBDEs along an urban-rural transect: Spatial and temporal trends, Chemosphere 64, 262–267. Harrad, S. and Hunter, S. (2006), Concentrations of polybrominated diphenyl ethers in air and soil on a rural-urban transect across a major UK conurbation, Environ. Sci. Technol. 40, 4548–4553. Harrison, R. M., Smith, D. J., and Luhana, L. (1996), Source apportionment of atmospheric polycyclic aromatic hydrocarbons collected from an urban location in Birmingham, U.K., Environ. Sci. Technol. 30, 825–832. Henry, R. C. (1997), History and fundamentals of multivariate air quality receptor models, Chemom. Intell. Lab. Syst. 37, 37–42. Henry, R. C. and Kim, B. M. (1990), Extension of self-modeling curve resolution to mixtures of more than three components. Part 1: Finding the basic feasible region, Chemom. Intell. Lab. Syst. 8, 205–216. Henry, R. C., Lewis, C. W., and Collins, J. F. (1994), Vehiclerelated hydrocarbon source compositions from ambient data: The GRACE/SAFER method, Environ. Sci. Technol. 28, 823–832. Hillery, B. R., Basu, I., Sweet, C. W., and Hites, R. A. (1997), Temporal and spatial trends in a long-term study of gas-phase PCB contaminations near the Great Lakes, Environ. Sci. Technol. 31, 1811–1816. Hillery, B. R., Simcik, M. F., Basu, I., Hoff, R. M., Strachan, W. M. J., Burniston, D., Chan, C. H., Kenneth, A. Brice, K. A., Sweet, C.W., and Hites, R. A. (1998), Atmospheric deposition of toxic pollutants to the Great Lakes as measured by the integrated atmospheric deposition network, Environ. Sci. Technol. 32, 2216–2221. Holsen, T. M., Noll, K. E., Liu, S. P., Lee, W. J. (1991), Dry deposition of polychlroinated biphenyls in urban areas, Environ. Sci. Technol. 25, 1075–1081. Hoff, R. M., Strachan, W. M. J., Sweet, C. W., Chan, C. H., Shackleton, M., Bidleman, T. F., Brice, K. A., Burniston, D. A., Cussion, S., Gatz, D. F., Harlin, K., and Schroeder, W. H. (1996), Atmospheric deposition of toxic chemicals to the Great Lakes: A review of data through 1994, Atmos. Environ. 30, 3505–3527. Hoh, E. and Hites, R. A. (2004), Sources of toxaphene and other organochlorine pesticides in North America as determined by air measurements and potential source contribution function analysis, Environ. Sci. Technol. 38, 4187–4194. Hopke, P. K. (1985), Receptor Modeling in Environmental Chemistry, Wiley, New York. Hopke, P. K., Ito, K., Mar, T., Christensen, W. F., Eatough, D. J., Henry, R. C., Kim, E., Laden, F., Lall, R., Larson, T. V., Liu, H., Neas, L., Pinto, J., St€ olzel, M., Suh, H., Paatero, P., and Thurston, G. D. (2006), PM source apportionment and health effects: 1. Intercomparison of source apportionment results, J. Exposure Sci. Environ. Epidemiol. 16, 275–286.

REFERENCES

Horstmann, M. and McLachlan, M. S. (1998), Atmospheric deposition of semivolatile organic compounds to two forest canopies, Atmos. Environ. 32, 1799–1809. Horton, R. A. (1991), Investment casting, in ASM Handbook, Stefanescu, D. M., ed., American Society of Metals International, p. 937. Hsu, Y. K., Holsen, T. M., and Hopke, P. K. (2003a), Locating and quantifying PCB sources in Chicago: Receptor modeling and field sampling, Environ. Sci. Technol. 37, 681–690. Hsu, Y. K., Holsen, T. M., and Hopke, P. K. (2003b), Comparison of hybrid receptor models to locate PCB sources in Chicago, Atmos. Environ. 37, 545–562. Hu, D., Martinez, A., and Hornbuckle, K. C. (2008), Discovery of non-aroclor PCB (3, 3’-dichlorobiphenyl) in Chicago air, Environ. Sci. Technol. 42, 7873–7877. Hussen, A., Westbom, R., Megersa, N., Mathiasson, L., and Bjorklund, E. (2006), Development of a pressurized liquid extraction and clean-up procedure for the determination of a-endosulfan, b-endosulfan, and endosulfan sulfate in aged contaminated Ethiopian soils, J. Chromatogr. A 1103, 202–210. HydroQual (2007), A Model for the Evaluation and Management of Contaminants of Concern in Water, Sediment, and Biota in the NY/NJ Harbor estuary. Contaminant Fate, Transport, and Bioaccumulation Sub-models, in report prepared for the Hudson River Foundation on behalf of the Contamination Assessment and Reduction Project (CARP). IARC (1994), Polynuclear Aromatic Compounds. Part 1: Chemical, Environmental, and Experimental Data. in: Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans (Series). Iwata, H., Tanabe, S., Sakai, N., and Tatsukawa, R. (1993), Distribution of persistent organochlorines in the oceanic air and surface seawater and the role of ocean on their global transport and fate, Environ. Sci. Technol. 27, 1080–1098. Jamshidi, A., Hunter, S., Hazrati, S., and Harrad, S. (2007), Concentrations and chiral signatures of polychlorinated biphenyls in outdoor and indoor air and soil in a major U.K. conurbation, Environ. Sci. Technol. 41, 2153–2158. Jaward, F. M., Farrar, N. J., Harner, T., Sweetman, A. J., and Jones, K. C. (2004), Passive air sampling of PCBs, PBDEs, and organochlorine pesticides across Europe, Environ. Sci. Technol. 38, 34–41. Jaward, F. M., Zhang, G., Nam, J. J., Sweetman, A. J., Obbard, J. P., Kobara, Y., and Jones, K. C. (2005), Passive air sampling of polychlorinated biphenyls, organochlorine compounds, and polybrominated diphenyl ethers across Asia, Environ. Sci. Technol. 39, 8638–8645. Kaiser, K. L. E. (1974), On the optical activity of polychlorinated biphenyls, Environ. Pollut. 7, 93–101. Khalili, N. R., Scheff, P. A., and Holsen, T. M. (1995), PAH source fingerprints for coke ovens, diesel and gasoline engines, highway tunnels, and wood combustion emissions, Atmos. Environ. 29, 533–542. Koester, C. J. and Hites, R. A. (1992), Photodegradation of polychlorinated dioxins and dibenzofurans adsorbed to fly ash, Environ. Sci. Technol. 26, 502–507.

179

Koester, C. J., Hites, R. A. (1992), Wet and dry deposition of chlorinated dioxins and furans, Environ. Sci. Technol. 26, 1375–1382. Kwok, E. S. C., Arey, J., and Atkinson, R. (1994), Gas-phase atmospheric chemistry of dibenzo-p-dioxin and dibenzofuran, Environ. Sci. Technol. 28, 528–533. Lakaschus, S., Weber, K., Wania, F., Bruhn, R., and Schrems, O. (2002), The air-sea equilibrium and time trend of hexachlorocyclohexanes in the Atlantic Ocean between the Arctic and Antarctica, Environ. Sci. Technol. 36, 138–145. Lamont, J. C. and Scott, D. S. (1970), An eddy cell model mass transfer into the surface of a turbulent liquid, Am. Inst. Chem. Eng. J. 16, 513–519. Larsen, R. K. and Baker, J. E. (2003), Source apportionment of polycyclic aromatic hydrocarbons in the urban atmosphere: A comparison of three models, Environ. Sci. Technol. 37, 1873–1881. Lee, D. S. and Nicholson, K. W. (1994), The measurement of atmospheric concentrations and deposition of semi-volatile organic compounds, Environ. Monit. Assess. 32, 51–91. Lee, W. J., Lin Lewis, S. J., Chen, Y. Y., Wang, Y. F., Shen, H. L., Su, C. C., Fan, Y. C. (1996), Polychlorinated biphenyls in the ambient air of petroleum refinery, urban, and rural areas, Atmos. Environ. 30, 2371–2378. Lee, W., Chang-Chien, G., Wang, L., Lee, W., Tsai, P., Wu, K., and Lin, C. (2004), Source identification of PCDD/Fs for various atmospheric environments in a highly industrialized city, Environ. Sci. Technol. 38, 4937–4944. Lee, W. J. (1991), The Determination of Dry Deposition Velocities for Ambient Gases and Particles, Illinois Institute of Technology, Chicago. Leister, D. L. and Baker, J. E. (1994), Atmospheric deposition of organic contaminants to the Chesapeake Bay, Atmos. Environ. 28, 1499–1520. Li, Y., Zhang, Q., Ji, D., Wang, T., Wang, Y., Wang, P., Ding, L., and Jiang, G. (2009), Levels and vertical distributions of PCBs, PBDEs, and OCPs in the atmospheric boundary layer: Observation from the Beijing 325-m meteorological tower, Environ. Sci. Technol. 43, 1030–1035. Lighty, J. S., Veranth, J. M., and Sarofim, A. F. (2000), Combustion aerosols: Factors govening their size and composition and implications to human health, J. Air Waste Manage. Assoc. 50, 1565–1618. Lim, L. H., Harrison, R. M., and Harrad, S. (1999), The contribution of traffic to atmospheric concentrations of polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 33, 3538–3542. Liss, P. S. and Merlivat, L. (1986), In The Role of Air-Sea Exchange in Geochemical Cycling, Reidel Publishing, Norwood, MA. Litten, S., Fowler, B. I., and Luszniak, D. (2002), Identification of a novel PCB source through analysis of 209 PCB congeners by US EPA modified method 1668, Chemosphere 46, 1457–1459. Liu, W., Hopke, P. K., Han, Y., Yi, S. M., Holsen, T. M., Cybart, S., Kozlowski, K., and Milligan, M. (2003), Application of receptor modeling to atmospheric constituents at potsdam and stockton, NY, Atmos. Environ. 37, 4997–5007.

180

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

Lohmann, R. and Jones, K. C. (2005), Dioxins and furans in air and deposition: A review of levels, behaviour and processes, Sci. Total Environ. 219, 53–81. Lohmann, R., Jurado, E., Dachs, J., Lohmann, U., and Jones, K. C. (2006), Quantifying the importance of the atmospheric sink for polychlorinated dioxin and furans relative to other global loss processes, J. Geophys. Res. 111, D21303. Ma, Y.-G., Lei, Y. D., Xiao, H., Wania, F., and Wang, W. -H. (2010), Critical review and recommended values for the physical-chemical property data of 15 polycyclic aromatic hydrocarbons at 25C, J. Chem Eng. Data 55, 819–825. Mackay, D. and Yeun, A. T. K. (1983), Mass transfer coefficients correlations for volatilization of organic solutes from water, Environ. Sci. Technol. 17, 211–233. Mackay, D. and Paterson, S. (1991), Evaluating the multimedia fate of organic chemicals: A level III fugacity model, Environ. Sci. Technol 25, 427–436. Mackay, B., Shiu, W. Y., Ma, K. C. (1999), Physical-chemical properties and environmental fate handbook, Chapman & Hall. MacLeod, M., Scheringer, M., Podey, H., Jones, K. C., and Hungerbuhler, K. (2007), The origin and significance of short-term variability of semivolatile contaminants in air, Environ. Sci. Technol. 41, 3249–3253. MacLeod, M., Woodfine, D. G., Mackay, D., Mckone, T., Bennett, D., and Maddalena, R. (2001), BETR North America: A regionally segmented multimedia contaminant fate model for North America, Environ. Sci. Pollut. Res. Int. 8, 156–163. Malavia, J., Santos, F. J., and Galceran, M. T. (2004), Gas chromatography-ion trap tandem mass spectrometry versus GC– high-resolution mass spectrometry for the determination of non-ortho-polychlorinated biphenyls in fish, J. Chromatogr. A 1056, 171–178. Malm, W. C., Sisler, J. F., Huffman, D., Eldred, R. A., and Cahill, T. A. (1994), Spatial and seasonal trends in particle concentration and optical extinction in the United States, J. Geophys. Res. 99, 1347–1370. Mandalakis, M., Berresheim, H., and Stephanou, E. G. (2003), Direct evidence for destruction of polychlorobiphenyls by OH radicals in the subtropical troposphere, Environ. Sci. Technol. 37, 542–547. Mano, S. and Schaug, J. (2003), EMEP POP laboratory comparison 2000–2002. Norwegian Institute for Air Research. Oslo, Norway. Available at: http://www.nilu.no/index.cfm?ac¼publications& folder _id¼4309&publication_id¼5189&view¼rep. Mari, M., Nadal, M., Schuhmacher, M., and Domingo, J. L. (2008), Monitoring PCDD/Fs, PCBs and metals in the ambient air of an industrial area of Catalonia, Spain. Chemosphere 73, 990–998. Masunaga, S., Yao, Y., Ogura, I., Sakurai, T., and Nakanishi, J. (2003), Source and behavior analyses of dioxins based on congener-specific information and their application to Tokyo Bay basin, Chemosphere 53, 315–324. McConnell, L. L., Kucklick, J. R., Bidleman, T. F., Ivanon, G. P., and Chernyak, S. M. (1996), Air-water gas exchange of organochlorine compounds in Lake Baikal, Russia, Environ. Sci. Technol. 30, 2975–2983.

McDonald, J. G. and Hites, R. A. (2003), Radial dilution model for the distribution of toxaphene in the United States and Canada on the basis of measured concentrations in tree bark, Environ. Sci. Technol. 37, 475–481. Meijer, S. N., Ockenden, W. A., Steinnes, E., Corrigan, B. P., and Jones, K. C. (2003), Spatial and temporal trends of POPs in Norwegian and U.K. background air: Implications for global cycling, Environ. Sci. Technol. 37, 454–461. Moeckel, C., MacLeod, M., Hungerbuhler, K., and Jones, K. C. (2008), Measurement and modeling of diel variability of polybrominated diphenyl ethers and chlordanes in air, Environ. Sci. Technol. 42, 3219–3225. Molina, L., Cabes, M., Diaz-Ferrero, J., Coll, M., Marti, R., BrotoPuig, F., Comellas, L. and Rodrıguez-Larena, M. C. (2000), Separation of non-ortho polychlorinated biphenyl congeners on pre-packed carbon tubes. Application to analysis in sewage sludge and soil samples, Chemosphere 40, 921–927. Monosmith, C. L. and Hermanson, M. H. (1996), Spatial and temporal trends of atmospheric organochlorine vapors in the central and upper Great Lakes, Environ. Sci. Technol. 30, 3464–3472. Moog, D. B. and Jirka, G. H. (1999a), Air-water gas transfer in uniform channel flow, J. Hydraul. Eng. 125, 3–10. Moog, D. B. and Jirka, G. H. (1999b), Stream reaeration in nonuniform flow: Macroroughness enhancement, J. Hydraul. Eng. 125, 11–16. Moreau-Guigon, E., Motelay-Massei, A., Harner, T., Pozo, K., Diamond, M., Chevreuil, M., and Blanchoud, H. (2007), Vertical and temporal distribution of persistent organic pollutants in Toronto. 1. Organochlorine pesticides, Environ. Sci. Technol. 41, 2171–2177. Motelay-Massei, A., Harner, T., Shoeib, M., Diamond, M., Stern, G., and Rosenberg, B. (2005), Using passive air samplers to assess urban-rural trends for persistent organic pollutants and polycyclic aromatic hydrocarbons. 2. Seasonal trends for PAHs, PCBs, and organochlorine pesticides, Environ. Sci. Technol. 39, 5763–5773. Muir, D. and Sverko, E. (2006), Analytical methods for PCBs and organochlorine pesticides in environmental monitoring and surveillance: A critical appraisal, Anal. Bioanal. Chem. 386, 769–789. Muller, J. F., Hawker, D. W., Connell, D. W., Komp, P., and McLachlan, M. S. (2000), Passive sampling of atmospheric SOCs using tritearin-coated fibreglass sheets, Atmos. Environ. 34, 3525–3534. Nelson, E. D., McConnell, L. L., and Baker, J. E. (1998), Diffusive exchange of gaseous polycyclic aromatic hydrocarbons and polychlorinated biphenyls across the air-water interface of the Chesapeake Bay, Environ. Sci. Technol. 32, 912–919. Newman, J., Becker, J., Blondina, G., and Tjeerdema, R. (1998), Quantitation of aroclors using congener-specific results, Environ. Toxicol. Chem. 17, 2159–2167. Nielsen, T. (1996), Traffic contribution of polycyclic aromatic hydrocarbons in the center of a large city, Atmos. Environ. 30, 3481–3490.

REFERENCES

Nizzetto, L., Lohmann, R., Gioia, R., Jahnke, A., Temme, C., Dachs, J. Herckes, P., Di Guardo, A., and Jones, K. C. (2008), PAHs in air and seawater along a North-South Atlantic transect: Trends, processes and possible sources, Environ. Sci. Technol. 42, 1580–1585. O’Connor, D. J. and Dobbins, W. E. (1958), Mechanisms of reaeration in natural streams, Trans. Am. Soc. Civ. Eng. 123, 641–684. Odabasi, M., Sofuoglu, A., Vardar, N., Tasdemir, Y., Holsen, T. M. (19990, Measurement of dry deposition and air-water exchange of polycyclic aromatic hydrocarbons with the water surface sampler, Environ. Sci. Technol. 33, 426–434. Odabasi, M., Sofuoglu, A., and Holsen, T. M. (2001), Mass transfer coefficients for polycyclic aromatic hydrocarbons (PAHs) to the water surface sampler: Comparison to modeled results, Atmos. Environ. 35, 1655–1662. Odabasi, M., Sofuoglu, A., Vardar, N., Tasdemir, Y., and Holsen, T. M. (1999), Measurement of dry deposition and air-water exchange of polycyclic aromatic hydrocarbons with the water surface sampler, Environ. Sci. Technol. 33, 426–434. Offenberg, J., Simcik, M., Baker, J., and Eisenreich, S. J. (2005), The impact of urban areas on the deposition of air toxics to adjacent surface waters: A mass budget of PCBs in Lake Michigan in 1994, Aquat. Sci. 67, 79–85. Offenberg, J. H. and Baker, J. (1997), Polychlorinated biphenyls in Chicago precipitation: Enhanced wet deposition to near-shore Lake Michigan, Environ. Sci. Technol. 31, 1534–1538. Offenberg, J. H. and Baker, J. E. (2002), Precipitation scavenging of polychlorinated biphenyls and polycyclic aromatic hydrocarbons along an urban to over-water transect, Environ. Sci. Technol. 36, 3763–3771. Ogura, I., Masunaga, S., and Nakanishi, J. (2001), Congenerspecific characterization of PCDDs/PCDFs in atmospheric deposition: Comparison of profiles among deposition, source, and environmental sink, Chemosphere 45, 173–183. Ould-Dada, Z. (2002), Dry deposition profile of small particles within a model spruce canopy, Sci. Total Environ. 286, 83–96. Paatero, P. (2003), User’s Guide for Positive Matrix Factorization Programs PMF2 and PMF3, Part 1: Tutorial. Paatero, P. and Tapper, U. (1994), Positive matrix factorization: A non-negative factor model with optimal utilization of error estimates of data values, Environmetrics 5, 111–126. Paterson, K. G., Sagady, J. L., Hooper, D. L., Bertman, S. B., Carroll, M. A., and Shepson, P. B. (1999), Analysis of air quality data using positive matrix factorization, Environ. Sci. Technol. 33, 635–641. Persson, Y., Lundstedt, S., Oberg, L., and Tysklind, M. (2007), Levels of chlorinated compounds (CPs, PCPPs, PCDEs, PCDFs and PCDDs) in soils at contaminated sawmill sites in Sweden, Chemosphere 66, 234–242. Peters, A. J., Lane, D. A., Gundel, L. A., Northcott, G. L., and Jones, K. C. (2000), A comparison of high volume and diffusion deduner samplers for measuring semivolatile organic compounds in the atmosphere, Environ. Sci. Technol. 34, 5001–5006.

181

Pirrone, N., Keeler, G. J., and Holsen, T. M. (1995a), Dry deposition of trace elements to Lake Michigan: A hybrid-receptor deposition modeling approach, Environ. Sci. Technol. 29, 2112–2122. Pirrone, N., Keeler, G. J., and Holsen, T. M. (1995b), Dry deposition of semivolatile organic compounds to Lake Michigan, Environ. Sci. Technol. 29, 2123–2132. Pistikopoulos, P., Masclet, P., and Mouvier, G. (1990), A receptor model adapted to reactive species: Polycyclic aromatic hydrocarbons: Evaluation of source contributions in an open urban site. I. Particle compounds, Atmos. Environ. 24A, 1189–1197. Polissar, A. V. and Hopke, P. K. (2001), Atmospheric aerosol over vermont: Chemical composition and sources, Environ. Sci. Technol. 35, 4604–4621. Poor, N., Tremblay, R., Kay, H., Bhethanabotla, V., Swartz, E., Luther, M., and Campbell, S. (2004), Atmospheric concentrations and dry deposition rates of polycyclic aromatic hydrocarbons (PAHs) for Tampa Bay, Florida, USA, Atmos. Environ. 38, 6005–6015. Poster, D. L. and Baker, J. E. (1996), Influence of submicron particles on hydrophobic organic contaminants in precipitation. 1. Concentrations and distributions of polycyclic aromatic hydrocarbons and polychlorinated biphenyls in rainwater, Environ. Sci. Technol. 30, 341–348. Pozo, K., Harner, T., Shoeib, M., Urrutia, R., Barra, R., Parra, O., and Focardi, S. (2004), Passive-sampler derived air concentrations of persistent organic pollutants on a north-south transect in Chile, Environ. Sci. Technol. 38, 6529–6537. Primbs, T., Genualdi, S., and Simonich, S. M. (2008), Solvent selection for pressurized liquid extraction of polymeric sorbents used in air sampling, Environ. Toxicol. Chem. 27, 1267–1272. Primbs, T., Simonich, S., Schmedding, D., Wilson, G., Jaffe, D., Takami, A., Kato, S., Hatakeyama, S., and Kajii, Y. (2007), Atmospheric outflow of anthropogenic semivolatile organic compounds from East Asia in spring 2004, Environ. Sci. Technol. 41, 3551–3558. Qiu, X. and Hites, R. A. (2008), Dechlorane Plus and other flame retardants in tree bark from the Northeastern United States, Environ. Sci. Technol. 42, 31–36. Qiu, X., Zhu, T., Li, J., Pan, H., Li, Q., Miao, G., and Gong, J. (2004), Organochlorine pesticides in the air around the Taihu Lake, China, Environ. Sci. Technol. 38, 1368–1374. Raff, J. D. and Hites, R. A. (2007), Deposition versus photochemical removal of PBDEs from Lake Superior air, Environ. Sci. Technol. 41, 6725–6731. Rappe, C., Glas, B., and Wiberg, K. (1990a), Solved and remaining PCDD and PCDF problems in the pulp industry, Organohalogen Compd. 3, 287–290. Rappe, C., Andersson, R., Lundstrom, K., and Wiberg, K. (1990b), Levels of polychlorinated dioxins and dibenzofurans in commercial detergents and related products, Chemosphere 21, 43–50. Ravindra, K., Mittal, A. K., and Grieken, R. V. (2001), Health risk assessment of urban suspended particulate matter with special reference to polycyclic aromatic hydrocarbons: A review, Rev. Environ. Health 16, 169–118.

182

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

Ravindra, K., Sokhi, R., and Grieken, R. V. (2008), Atmospheric polycyclic aromatic hydrocarbons: Source attribution, emission factors and regulation, Atmos. Environ. 42, 2895–2921. Reinfelder, J. R., Totten, L. A., and Eisenreich, S. J. (2004), The New Jersey Atmospheric Deposition Network (NJADN), Final Report to the New Jersey Department of Environmental Protection. Risebrough, R. W., Walker, W., Schmidt, T. T., Lappe, B. W., and Connors, C. W. (1976), Transfer of chlorinated biphenyls to Antarctica, Nature 264, 738–739. Robson, M. and Harrad, S. (2004), Chiral PCB signatures in air and soil: Implications for atmospheric source apportionment, Environ. Sci. Technol. 38, 1662–1666. Rodenburg, L. A., Valle, S. N., Panero, M. A., Munoz, G. R., and Shor, L. M. (2010), Mass balances on selected polycyclic aromatic hydrocarbons (PAHs) in the New York/New Jersey Harbor, J. Environ. Qual. 39, 642–653. Rowe, A. A. (2006), Interactions of Polychlorinated Biphenyls with the Air, Water, and Sediments of the Delaware River Estuary. Ph.D. dissertation, Department of Environmental Science, Rutgers University, New Brunswick, NJ. Rowe, A. A., Totten, L. A., Cavallo, G. J., and Yagecic, J. R. (2007a), Watershed processing of atmospheric polychlorinated biphenyl inputs, Environ. Sci. Technol. 41, 2331–2337. Rowe, A. A., Totten, L. A., Xie, M., Fikslin, T. J., and Eisenreich, S. J. (2007b), Air-water exchange of polychlorinated biphenyls in the Delaware River, Environ. Sci. Technol. 41, 1152–1158. Rushneck, D. R., Beliveau, A., Fowler, B., Hamilton, C., Hoover, D., Kaye, K. Berg, M., Smith, T., Telliard, W. A., Roman, H., Ruder, E., and Ryan, L. (2004), Concentrations of dioxin-like PCB congeners in Unweathered Aroclors by HRGC/HRMS using EPA Method 1668A, Chemosphere 54, 79–87. Ryno, M., Rantanen, L., Papaioannou, E., Konstandopoulos, A. G., Koskentalo, T., and Savela, K. (2006), Comparison of pressurized fluid extraction, Soxhlet extraction and sonication for the determination of polycyclic aromatic hydrocarbons in urban air and diesel exhaust particulate matter, J. Environ. Monit. 8, 488–493. Santos, F. J. and Galceran, M. T. (2003), Modern developments in gas chromatography-mass spectrometry based environmental analysis, J. Chromatogr. A 1000, 125–151. Sather, P. L., Ikonomou, M. G., Addison, R. F., He, T., Ross, P. S., and Fowler, B. (2001), Similarity of an aroclor-based and a full congener-based method in determining total PCBs and a modeling approach to estimate aroclor speciation from congener-specific PCB data, Environ. Sci. Technol. 35, 4874–4880. Schauer, J. J., Rogge, W. F., Hildemann, L. M., Mazurek, M. A., Cass, G. R., and Simoneit, B. R. T. (1996), Source apportionment of airborne particulate matter using organic compounds as tracers, Atmos. Environ. 30, 3837–3855. Schenker, U., Macelod, M., Scheringer, M., and Hungerbuhler, K. (2005), Improving data quality for environmental fate models: A least-squares adjustment procedure for harmonizing physicochemical properties or organic compounds, Environ. Sci. Technol. 39, 8434–8441.

Schenker, U., Scheringer, M., Sohn, M. D., Maddalena, R. L., Mckone, T. E., and Hungerbuhler, K. (2009), Using information on uncertainty to improve environmental fate modeling: A case study on DDT, Environ. Sci. Technol. 43, 128–134. Scheringer, M. (1996), Persistence and spatial range as endpoints of an exposure-based assessment of organic chemicals, Environ. Sci. Technol. 30, 1652–1659. Scheringer, M. (1997), Characterization of the environmental distribution behavior of organic chemicals by means of persistence and spatial range, Environ. Sci. Technol. 31, 2891–2897. Schroder, J., Welsch-Pausch, K., and Mclachlan, M. S. (1997), Measurement of atmospheric depostion of polychlorinated dibenzo-p-dioxins (PCDD) and dibenzofurans (PCDF) to a soil, Atmos. Environ. 31, 2983–2989. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (2003), Environmental Organic Chemistry, Wiley, Hoboken, NJ. Shannigrahi, A. S., Fukushima, T., and Ozaki, N. (2005), Comparison of different methods for measuring dry deposition fluxes of particulate matter and polycyclic aromatic hydrocarbons (PAHs) in the ambient air, Atmos. Environ. 39, 653–662. Sheu, H. L., Lee, W. J., Su, C. C., Chao, H. R., and Fan, Y. C. (1996), Dry deposition of polycyclic aromatic hydrocarbons in ambient air, J. Environ. Eng. 122, 1101–1109. Shih, M., Lee, W.-S., Chang-Chien, G.-P., Wang, L.-C., Hung, C.-Y., Lin, K.-C. (2006), Dry deposition of polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) in ambient air, Chemosphere 62, 411–416. Shoeib, M. and Harner, T. (2002), Characterization and comparison of three passive air samplers for persistent organic pollutants, Environ. Sci. Technol. 36, 4142–4151. Shoeib, M., Harner, T., Ikonomou, M., and Kannan, K. (2004), Indoor and outdoor air concentrations and phase partitioning of perfluoroalkyl sulfonamides and polybrominated diphenyl ethers, Environ. Sci. Technol. 38, 1313–1320. Simcik, M. F., Eisenreich, S. J., and Lioy, P. J. (1999a), Source apportionment and source/sink relationships of PAHs in the coastal atmosphere of Chicago and Lake Michigan, Atmos. Environ. 33, 5071–5079. Simcik, M. F., Basu, I., Sweet, G. W., and Hites, R. A. (1999b), Temperature dependence and temporal trends of polychlorinated biphenyl congeners in the Great Lake atmosphere, Environ. Sci. Technol. 33, 1991–1995. Simcik, M. F., Zhang, H., Eisenreich, S. J., and Franz, T. P. (1997), Urban contamination of the Chicago/coastal Lake Michigan atmosphere by PCBs and PAHs during AEOLOS, Environ. Sci. Technol. 31, 2141–2147. Simonich, S. L. and Hites, R. A. (1995), Global distribution of persistent organochlorine compounds, Science 269, 1851–1854. Sinkkonen, S. and Paasivirta, J. (2000), Degradation half-life times of PCDDs, PCDFs and PCBs for environmental fate and modeling, Chemosphere 40, 943–949. Sofowote, U. M., Allan, L. M., and McCarry, B. E. (2010), Evaluation of PAH diagnostic ratios as source apportionment tools for air particulates collected in an urban-industrial environment, J. Environ. Monit. 12, 417–424.

REFERENCES

Sovocool, G. W., Lewis, R. G., Harless, R. L., Wilson, N. K., and Zehr, R. D. (1977), Analysis of technical chlordane by gas chromatography/mass spectrometry, Anal. Chem. 49, 734–740. St-Amand, A. D., Mayer, P. M., Blais, J. M. (2007), Modeling atmospheric vegetation uptake of PBDEs using field measurements, Environ. Sci. Technol. 41, 4234–4239. Stenchikov, G., Lahoti, N., Lioy, P., Georgopoulos, P., Diner, D., and Kahn, R. (2006), Micrometeorological pollutant transport from the collapse of the World Trade Center on September 11, 2001, Environ. Fluid Dynam. 6, 425–450. Stenchikov, G., Pickering, K., Decaria, A., Tao, W. K., Scala, J., Ott, L. Bartels, D., and Matejka, T. (2005), Simulation of the fine structure of the 12 July 1996 Stratosphere-Troposphere Experiment: Radiation, Aerosols and Ozone (STERAO-A) storm accounting for effects of terrain and interaction with mesoscale flow, J. Geophys. Res. Atmos. 110, D14304. Su, Y. and Wania, F. (2005), Does the forest filter effect prevent semi-volatile organic compounds from reaching the Arctic?, Environ. Sci. Technol. 39, 7185–7193. Su, Y., Wania, F., Harner, T., and Lei, Y. D. (2007), Deposition of polybrominated diphenyl ethers, polychlorinated biphenyls, and polycyclic aromatic hydrocarbons to a boreal deciduous forest, Environ. Sci. Technol. 41, 534–540. Subhash, S., Honrath, R. E., and Kahl, J. D. W. (1999), Backtrajectory analysis of atmospheric polychlorinated bipheny concentrations over Lake Superior, Environ. Sci. Technol. 33, 1509–1515. Sun, P., Basu, I., and Hites, R. A. (2006), Temporal trends of polychlorinated biphenyls in precipitation and air at Chicago, Environ. Sci. Technol. 40, 1178–1183. Sun, P., Basu, I., Blanchard, P., Brice, K. A., and Hites, R. A. (2007), Temporal and spatial trends of atmospheric polychlorinated biphenyl concentrations near the Great Lakes, Environ. Sci. Technol. 41, 1131–1136. Sundqvist, K. L., Tysklind, M., Geladi, P., Cato, I., and Wiberg, K. (2009), Congener fingerprints of tetra- through octa-chlorinated dibenzo-p-dioxins and dibenzofurans in Baltic surface sediments and their relations to potential sources, Chemosphere 77, 612–620. Tasdemir, Y., Vardar, N., Odabasi, M., Holsen, T.M. (2004), Concentrations and gas/particle partitioning of PCBs in Chicago, Environ. Poll. 131, 35–44. Tasdemir, Y. and Holsen, T. M. (2005), Measurement of particle phase dry deposition fluxes of polychlorinated biphenyls (PCBs) with a water surface sampler, Atmos. Environ. 39, 1845–1854. Tasdemir, Y. and Esen, F. (2007), Dry deposition fluxes and deposition velocities of PAHs at an urban site in Turkey, Atmos. Environ. 41, 1288–1301. Totten, L. A., Eisenreich, S. J., and Brunciak, P. A. (2002), Evidence for destruction of PCBs by the OH radical in urban atmosphere, Chemosphere 47, 735–746. Totten, L. A., Gigliotti, C. L., Offenberg, J. H., Baker, J. E., and Eisenreich, S. J. (2003), Reevaluation of air-water exchange fluxes of PCBs in Green Bay and southern Lake Michigan, Environ. Sci. Technol. 37, 1739–1743.

183

Totten, L. A., Panangadan, M., Eisenreich, S. J., Cavallo, G. J., and Fikslin, T. J. (2006a), Direct and indirect atmospheric depositions of PCBs to the Delaware River watershed, Environ. Sci. Technol., 40, 2171–2176. Totten, L. A., Stenchikov, G., Gigliotti, C. L., Lahoti, N., and Eisenreich, S. J. (2006b), Measurement and modeling of urban atmospheric PCB concentrations on a small (8km) spatial scale, Atmos. Environ. 40, 7940–7952. Totten, L. A., Gigliotti, C. L., Vanry, D. A., Offenberg, J. H., Nelson, E. D., Dachs, J., Reinfelder, J. R., and Eisenreich, S. J. (2004), Atmospheric concentrations and deposition of polychlorinated biphenyls to the Hudson River estuary, Environ. Sci. Technol. 38, 2568–2573. Tsai, P., Hoenicke, R., Yee, D., Bamford, H. A., Baker, J. E. (2002), Atmospheric concentrations and fluxes of organic compounds in the Northern San Francisco Estuary, Environ. Sci. Technol. 36, 4741–4747. UNEP (2001), Final Act of the Conference of Plenipotentiaries on the Stockholm Convention on Persistent Organic Pollutants, United Nations Environment Program, Geneva, Switzerland, p. 44. USEPA (1999a), Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air, 2nd ed., EPA/625/ R-96/010b, United States Environmental Protection Agency, Cincinnati, OH. USEPA (1999b), USEPA Method 1668, Revision A: Chlorinated Biphenyl Congeners in Water, Soil, Sediment, and Tissue by HRGC/HRMS. USEPA (2000), USEPA Method 8082: Polychlorinated Biphenyls (PCBs) by Gas Chromatography. Van Ry, D. A., Gigliotti, C. L., Glenn, T. R., Nelson, E. D., Totten, L. A., and Eisenreich, S. J. (2002), Wet deposition of polychlorinated biphenyls in urban and background areas of the Mid-Atlantic states, Environ. Sci. Technol. 36, 3201–3209. Vardar, N., Odabasi, M., and Holsen, T. M. (2002), Particulate dry deposition velocities of polycyclic aromatic hydrocarbons (PAHs), J. Environ. Eng. 128, 269–274. Venier, M. and Hites, R. A. (2008), Atmospheric deposition of PBDEs to the Great Lakes featuring a Monte Carlo analysis of errors, Environ. Sci. Technol. 42, 9058–9064. Verenitch, S. S., deBruyn, A. M. H., Ikonomou, M. G., and Mazumder, A. (2007), Ion-trp tandem mass spectrometry-based analytical methodology for the determination of polychlorinated biphenyls in fish and shellfish performance comparison against electron-capture detection and high-resolution mass spectrometry detection, J. Chromatogr. A 1142, 199–208. Vione, D., Barra, S., Gennaro, G., De Rienzo, M., Gilardoni, S., Perrone, M. G., and Pozzoli, L. (2004), Polycyclic aromatic hydrocarbons in the atmosphere: Monitoring, sources, sinks and fate. II: Sinks and fate, Anal. Chim. 94, 257–268. Vione, D., Maurino, V., Minero, C., Pelizzetti, E., Harrison, M. A. J., Olariu, R., and Arsene, C. (2006), Photochemical reactions in the tropospheric aqueous phase and on particulate matter, Chem. Soc. Rev. 35, 441–453. Wania, F. and Mackay, D. (1995), A global distribution model for persistent organic chemicals, Sci. Total Environ. 160–161 211–232.

184

MEASUREMENT AND MODELING OF SEMIVOLATILE ORGANIC COMPOUNDS IN LOCAL ATMOSPHERES

Wania, F. and McLachlan, M. S. (2001), Estimating the influence of forests on the overall fate of semivolatile organic compounds using a multimedia fate model, Environ. Sci. Technol. 35, 582–590. Wania, F. and Daly, G. L. (2002), Estimating the contribution of degradation in air and deposition to the deep sea to the global loss of PCBs, Atmos. Environ. 36, 5581–5593. Wanninkhof, R. (1992), Relationship between wind speed and gas exchange over the ocean, J. Geophy. Res. 97, 7373–7382. Wanninkhof, R., Leswell, J. P., and Broecker, W. S. (1987), Gas exchange on Mono Lake and Crowley Lake, California, J. Geophys. Res. 92, 14567–14580. Watson, J. G., Cooper, J. A., and Huntzicker, J. J. (1984), The effective variance weighting for least squares calculations applied to the mass balance receptor model, Atmos. Environ. 18, 1347–1355. Watson, J. G., Robinson, N. F., Chow, J. C., Henry, R. C., Kim, B. M., Pace, T. G., Meyer, E. L., and Nguyen, Q. (1990), The USEPA/DRI chemical mass balance receptor model, Environ. Software 5, 38–49. Wegmann, F., Scheringer, M., and Moller, M. (2004), Influence of vegetation on the environmental partitioning of DDT in two global multimedia model, Environ. Sci. Technol. 38, 1505–1512. Welsch-Pausch, K., McLachlan, M. S., and Umlauf, G. (1995), Determination of the principal pathways of polychlorinated dibenzo-p-dioxins and dibenzofurans to Lolium multiflorum (Welsh Ray Grass), Environ. Sci. Technol. 29, 1090–1098. Whitman, W. G. (1923), The two-film theory of gas absorption, Chem. Metal. Eng. 29, 146–148. Wild, S. R. and Jones, K. C. (1995), Polynuclear aromatic hydrocarbons in the United Kingdom environment: A preliminary source inventory and budget, Envrion. Pollut. 88, 91–108. Wong, C. S., Garrison, A. W., and Foreman, W. T. (2001a), Enantiomeric composition of chiral polychlorinated biphenyl atropisomers in aquatic bed sediment, Environ. Sci. Technol. 35, 33–39. Wong, C. S., Garrison, A. W., Smith, P. D., and Foreman, W. T. (2001b), Enantiomeric composition of chiral polychlorinated biphenyl atropisomers in aquatic and riparian biota, Environ. Sci. Technol. 35, 2448–2454. Wu, R., Backus, S., Basu, I., Blanchard, P., Brice, K., DryfhoutClark, H., Fowlie, P., Hulting, M., and Hites, R. (2009), Findings from quality assurance activities in the integrated atmospheric deposition network, J. Environ. Eng. 11, 277–296.

Xiao, H., Hung, H., Harner, T., Lei, Y. D., and Wania, F. (2008), Field testing a flow-through sampler for semivolatile organic compounds in air, Environ. Sci. Technol. 42, 2970–2975. Xiao, H., Hung, H., Harner, T., Lei, Y. D., Johnston, G. W., and Wania, F. (2007), A flow-through sampler for semivolatile organic compounds in air, Environ. Sci. Technol. 41, 250–256. Xie, Z. and Ebinghaus, R. (2008), Analytical methods for the determination of emerging organic contaminants in the atmosphere, Anal. Chim. Acta 610, 156–178. Yang, H.-H. and Chen, C.-M. (2004), Emission inventory and sources of polycyclic aromatic hydrocarbons in the atmosphere at a suburban area in Taiwan, Chemosphere 56, 879–887. Yao, Y., Tuduri, L., Harner, T., Blanchard, P., Waite, D., Poissant, L. Murphy, C., Belzer, W., Aulagnier, F., Li, Y., and Sverko, E. (2006), Spatial and temporal distribution of pesticide air concentrations in Canadian agricultural regions, Atmos. Environ. 40, 4339–4351. Yu, T.-Y. and Chang, L.-F. W. (2001), Delineation of air-quality basins utilizing multivariate statistical methods in Taiwan, Atmos. Environ. 35, 3155–3166. Yunker, M. B., MacDonald, R. W., and Vingarzan, R. (2002), PAHs in the Fraser River basin: A critical appraisal of PAH ratios as indicators of PAH source and composition, Org. Geochem. 33, 489–515. Zencak, Z., Klanova, J., Holoubek, I., and Gustafsson, O. (2007), Source apportionment of atmospheric PAHs in the Western Balkans by natural abundance radiocarbon analysis, Environ. Sci. Technol. 41, 3850–3855. Zhang, H. X., Eisenreich, S. J., Franz, T. R., Baker, J. E., and Offenberg, J. H. (1999), Evidence of increased gaseous PCB fluxes to Lake Michigan from Chicago, Environ. Sci. Technol. 33, 2129–2137. Zhang, S., Zhang, W., Wang, K., Shen, Y., Hu, L., Wang, X. (2009), Concentration, distribution and source apportionment of atmospheric polycyclic aromatic hydrocarbons in the southeast suburb of Beijing, China, Environ. Monit. Assess. 151, 197–207. Zhang, X. L., Tao, S., Liu, W. X., Yang, Y., Zuo, Q., and Liu, S. Z. (2005), Source diagnostics of polycyclic aromatic hydrocarbons based on species ratios: A multimedia approach, Environ. Sci. Technol. 39, 9109–9114. Zhu, L. and Hites, R. A. (2006), Brominated flame retardants in tree bark from North America, Environ. Sci. Technol. 40, 3711–3716.

7 PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS BO PAN AND BAOSHAN XING 7.1. Introduction 7.2. Occurrence of Pharmaceuticals and Personal Care Products (PPCPs) in Soils and Sediments 7.2.1. Application and Sources 7.2.2. PPCP Distribution in Soils and Sediments 7.3. Interactions Between Soil/Sediment Organic Matter and PPCPs 7.3.1. Linear Free-Energy Relationship for Predicting PPCP Sorption Coefficient 7.3.2. Nonideal Interactions 7.3.3. Bound Residues 7.4. Importance of the Inorganic Fraction 7.4.1. Mineral Structures 7.4.2. Mechanisms Involved in PPCP Sorption on Soil/ Sediment Inorganic Fractions 7.5. Sorption as Affected by Solution Chemistry 7.5.1. pH 7.5.2. Ionic Strength 7.5.3. Different Types of Cations 7.5.4. Dissolved Organic Matter (DOM) 7.5.5. Other Coadsorbates 7.6. PPCP Degradation 7.7. PPCP Leaching 7.8. Quantitative Description of PPCP Environmental Fate 7.9. Summary and Perspectives

7.1. INTRODUCTION Pharmaceuticals and personal care products (PPCPs) include a wide range of chemicals, such as antibiotics and other compounds for humans and livestock, nutritional supple-

ments, fragrances (perfumes), and cosmetics (Daughton and Ternes 1999; McClellan and Halden 2010). Because of their frequent use, the presence of PPCPs has been widely detected in environmental media such as sewage treatment plant effluents, surface water, seawater, groundwater, soils, and sediments (Nikolaou et al. 2007; Kemper 2008; McClellan and Halden 2010). Unfortunately, some PPCPs may pose risks to organisms, such as disrupting endocrine systems of both humans and wildlife. In addition, microorganisms may become drug-resistant after contact with antibiotics (mutant) or adopt genes with drug resistance from other microorganism species (gene transfer) (Seveno et al. 2002; CheeSanford et al. 2009). The development of drug resistance results in the failure of antibiotics to function properly, which is a severe clinical problem. Human health is seriously threatened because of the wide application and presence of PPCPs in the environment. Drug resistance of three major bacterial respiratory pathogens is widely observed around the world (Alpuche et al. 2006), making acute respiratory tract infections the leading cause of global infectious deaths (WHO 2005). Therefore, concerns about PPCP environmental exposure and risk have attracted a great deal of research attention and interest. Many studies indicate that soils and sediments are important sinks for and control the environmental fate of PPCPs by inhibiting biodegradation, photodegradation, and hydrolysis (Thiele-Bruhn 2003; Monteiro and Boxall 2010); dissipating PPCPs because of the formation of bound residues (Loffler et al. 2005; Al-Rajab et al. 2009); and decreasing PPCP leaching into groundwater (Oppel et al. 2004; Xu et al. 2009). Soil/sediment–PPCP interactions and their mechanisms are the crucial factor for properly understanding and accurately predicting PPCP environmental fate and behavior. Studies have been conducted to investigate soil/

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang.
2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

185

186

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

sediment–PPCP interactions and associated mechanisms on different types of soils and sediments or isolated soil/sediment components. However, compared to nonpolar organic contaminants, PPCPs possess contrasting properties such as high water solubility, pH-dependent speciation, and fast degradation. Our current knowledge of PPCP behavior in soils/sediments is still relatively limited, and systematic studies are needed. General reviews on PPCP presence and behavior in soils are available for veterinary pharmaceuticals (Tolls 2001), psychiatric pharmaceuticals (Calisto and Esteves 2009), and antibiotics in soil (Thiele-Bruhn 2003). This chapter summarizes current research progress on PPCP behavior in soils/sediments with an emphasis on sorption-related topics. The occurrence of PPCPs in soils/sediments is summarized in Section 7.2. The interactions between PPCPs and soil/ sediment components (organic matter and inorganic fractions) are the main focuses of Sections 7.3 and 7.4. In Section 7.5, the effect of solution chemistry on PPCP sorption is presented. The effect of PPCP sorption on PPCP degradation and leaching are discussed in Sections 7.6 and 7.7. In Section 7.8 we present models for predicting PPCP sorption in soils/sediments with an emphasis on the departure from traditional models for the environmental behavior of hydrophobic organic contaminants. Finally, in Section 7.9 we summarize the overall approaches and perspectives regarding soil/sediment–PPCP interactions. 7.2. OCCURRENCE OF PHARMACEUTICALS AND PERSONAL CARE PRODUCTS (PPCPs) IN SOILS AND SEDIMENTS 7.2.1. Application and Sources The application of PPCPs has largely increased (Sarmah et al. 2006). Antibiotic application in animal feeding in-

creased 100 times (from 91,000 kg to 9.3 million kg) from 1950 to 1999 (AHI 2002). Take a single county, Poland, as an example, where the value of the pharmaceutical market has been growing systematically: by 10.7% in 2001, by 4.3% in 2002, and by 13.1% in 2003 (Willert 2007). The global pharmaceutical market reached $446 billion in 2003 with an increase of 15% compared to 2002 (Fig. 7.1). North America, Europe, and Japan accounted for 88% of the worldwide pharmaceutical market in 2003. The global pharmaceutical market reached $712 billion in 2007 and was forecast to grow to $929 billion in 2009 (Seget and Pharma 2008). Pharmaceuticals and personal care products are widely used for human and (nonhuman) animal health. Antibiotics are the main PPCPs used for animals. They are commonly used in livestock feed (mostly swine, poultry, and beef cattle) for the purpose of disease treatment and prevention, and growth promotion (Table 7.1). Humans use PPCPs for a broader range of purposes, including prescription and over-the-counter therapeutic drugs (e.g., pain relievers, cold medications, antibiotics), medical adjuvants (e.g., X-ray contrast media), fragrances, nutritional supplements, and cosmetics (Table 7.2). Some antibiotics are used for both humans and animals (e.g., sulfonamides and tetracyclines), while other antibiotics are specifically for human beings or animals. Thus in sites with different pollution sources, different chemicals should be considered. Some basic physicochemical properties of common PPCPs are listed in Table 7.3. The absorption efficiencies of PPCPs by humans and animals are very low, for example, 25%–75% for tetracyclines (TCs) (Kulshrestha et al. 2004). Generally, as much as 30%–90% of applied antibiotics could be excreted as original compounds or their metabolites via urine or feces (Hirsch et al. 1999). Some of the metabolites are active 2003 Total: $ 446 billion

2002 Total: $ 406 billion Latin America 8%

CIS* 1% Asia, Africa, Australia 8%

Asia, Africa, Australia 12% North America 41%

Japan 11% Europe 27%

Latin America 4%

Japan 11%

North America 49%

Europe 28%

*: Commonwealth of Independent States

Figure 7.1. Global pharmaceutical market in 2002 and 2003 data from Health (2004) and www.imsglobal.com. Global pharmaceutical market increased 15% in 2003 compared to 2002, and most of the pharmaceuticals are produced in North America, Europe, and Japan.

187

OCCURRENCE OF PHARMACEUTICALS AND PERSONAL CARE PRODUCTS (PPCPs) IN SOILS AND SEDIMENTS

TABLE 7.1. Veterinary Pharmaceuticals Industry

Purpose

Swine

Poultry

Beef Cattle

Disease Treatment

Growth Promotion

ü ü — — ü ü ü — ü ü ü

ü ü — ü ü ü ü ü ü ü ü

ü ü ü ü — ü — ü ü ü ü

ü ü — — ü ü ü — ü — ü

— ü — — — ü ü — ü — ü

ü ü — ü

ü — ü ü

ü — — —

Antibiotic Class Aminoglycosides (gentamycin, neomycin, streptomycin) b-Lactams (penicillins, ceftiofur) Chloramphenicol (florfenicol) Ionophores (monensin, salinomycin, semduramicin, lasalocid) Lincosamides (lincomycin) Macrolides (erythromycin, tilmicosin, tylosin) Polypeptides (bacitracin) Quinolones (fluoroquinolones, sarafloxacin, enrofloxacin) Streptogramins (virginiamycin) Sulfonamides (sulfadimethoxine, sulfamethazine, sulfisoxazole) Tetracyclines (chlortetracycline, oxytetracycline, tetracycline) Others Bambermycin Carbadox Novobiocin Spectinomycin Source: Modified From Chee-Sanford et al. (2009)

compounds or could be deconjugated to their parent compounds. Thereby, the major source for PPCP contamination is the PPCPs excreted from humans and animals. The possible pathways for PPCPs to enter the environment

and interact with soils/sediments are illustrated in Figure 7.2. The main PPCP inputs in soils include the irrigation with wastewater, land application of sludge from wastewater

TABLE 7.2. Human Pharmaceuticals and Personal Care Products Purpose

Names

Analgesic/anti-inflammatory

Phenazone; acetaminophen; acetylsalicylic acid; ketoprofen; meclofenamic acid; naproxen; propyphenazone; tolfenamic acid; carbamazepine; diclofenac-Na; dimethylaminophenazone; fenoprofen; ibuprofen; indomethacine Salicylic acid; o-hydroxyhippuric acid

Metabolite of acetylsalicylic acid, keratolytic, dermatice, food preservatve Antiepileptic Antibiotic Antidepressant Antineoplastic Antiseptic, fungicide b2-Sympathomimetic (bronchodilator) b-Blocking (antihypertensive, antianginal, antiarrhythmic) Cardiac drug, antihypertensive Lipid regulator Metabolite of lipid regulators, polar and active Oral contraceptive (in combination with progestogens) Psychiatric drug (anxiolytic; muscle relaxant) Sunscreen agent Sympathomimetic amine (anorexic) Synthetic musk Polycyclic musk X-ray contrast media Source: Modified from Daughton and Ternes (1999).

Carbamazepine Fluoroquinolone; sulfonamides Fluoxetine; fluvoxamine; Paroxetine Cyclophosphamide; ifosfamide 4-chloro-3,5-xylenol; chlorophene; triclosan; biphenylol; 3,4,5,6-ttetrabronio-o-cresol Clenbuterol; fenoterol; salbutamol; terbutaline Metoprolol; bisoprolol; nadolol; propranolol; timolol; carazolol; betaxolol Verapamil Bezafibrate; clofibrate; etofibrate; fenofibrate; gemfibrozil Clofibric acid; fenofibric acid 17a-Ethinyl estradiol Diazepam Methylbenzylidene camphor Fenfluramine Musk ambrette; musk xylene; musk ketone; musk moskene; musk tibetene Galaxolide; Tonalide; Celestolide Diatrizoate (Na); iohexol; iopamidol; iopromide; iotrolan

188

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

TABLE 7.3. Physicochemical Properties of Various PPCPs Chemical Name

Structure

pKa

OH CH3

C

H

log Kow

MW

CH

H

17a-Ethinyl, Estradiol

Sw, mg/L

10.4

11–44

4.15

296.4

N/Aa

3.6

3.1–4.0

272.3

N/A

0.01

3.19

875

9.59–10.2

120–300

3.4

228.3

N/A

N/A


1.40–0.15 262

N/A

2 5000

1.14

323.14

5.90; 8.89

30 000

0.4

331.35

H

HO

CH3

OH

H

17b-Estradiol H

H

HO

O H HO

O

O

O O

H

O

Avermectin

O

O

OH

H

O

H

O H OH

CH3

Bisphenol A

HO

OH CH3

ON+ O

Carbadox

N O

N+

N H

OH

Cl

OH H N

Cl

Chloramphenicol H O 2N

4

F

Ciprofloxacin

6 7

HN

O

OH

N

1

N

COOH 3

OCCURRENCE OF PHARMACEUTICALS AND PERSONAL CARE PRODUCTS (PPCPs) IN SOILS AND SEDIMENTS

189

TABLE 7.3. (Continued) Chemical Name

Structure O

pKa

Sw, mg/L

log Kow

MW

O F

HO

Enrofloxacin

N

N

6.27; 8.3

130 000

1.1

359.4

6.4

70

1.7

261.25

2.38

10 000

0.02

171.16

5.97; 8.28

N/A

0.35

361.38

10

N/A


2.34

263.25

6.9

4

0.68

261.23

N

O F

COOH

Flumequine N

N N+

N

Metronidazole

O-

O

OH

O F

COOH

Ofloxacin N

H3C-N

N O CH3

O-

O

N+

OH N H

Olaquindox

N+ O-

O

O O

HO

Oxolinic acid

O

N

O

O

OH

O

OH

OH

H2N

Oxytetracycline

3.27; 7.32; 9.11 1000

HO H N


1.22–1.97 460

H OH

OH

(continued )

190

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

Physicochemical Properties of Various PPCPs (Continued ) TABLE 7.3. (Continued) Chemical Name

Structure

pKa

log Kow

Sw, mg/L

MW

O OH

p-Aminobenzoic acid

4.9

6100

0.83

137.1

6.0; 8.6

100

N/A

385.37

2.3; 7.4

1500

0.8

278.33

8.4

270

0.35

249.3

2.4; 7.1

590

1

255.3

3.3; 7.7; 9.7

231–52,000
1.19–1.97 444.4

7.1

5000

H2N O

O

F OH HCl N

N

Sarafloxacin HN

F O

H N

N

S

Sulfamethazine

O

N

H2N

O

H N S

Sulfapyridine

O

N

H2N O

N

O S N H

Sulfathiazole

S

H2N

R1

R4

R3

R2

H

N H

OH

Tetracycline NH2 OH OH

O

OH

O

O

O CH3

O HC H3C

H3C H3C

Tylosin

N O

HO

O OCH3

O

H3C

O

CH3 O CH3

OH

OCH3 O CH3

a

HO O

Not available. Source: Modified from Tolls (2001) and Pan et al. (2009).

OH

CH3 OH

O CH3

3.5

917.14

OCCURRENCE OF PHARMACEUTICALS AND PERSONAL CARE PRODUCTS (PPCPs) IN SOILS AND SEDIMENTS

Humans

191

Livestock

Excretion of applied drugs and nutrients

Disease treatment

Flushing of expired prescriptions Wash-off of cosmetics and sun creams

Growth promotion Feed efficiency improvement

Medical wastes and hospital discharge Tap water

Manure Wastewater treatment plant

Field application Irrigation of treated wastewater on fields

Treated water

Purified water

Sewage sludge

Application of sludge or manure as fertilizers

Ground water recharging Discharge from wastewater treatment plant

Leakage or leaching from manure storage facilities

Flood, leaching, surface runoff Adsorption and degradation in soils/sediments

Uptake by plants

Figure 7.2. Applications and sources of PPCPs. In wastewater treatment plants, PPCPs are not efficiently removed using current technologies; thus the water and sewage sludge discharged from wastewater treatment plants, and tapwater purified from wastewater could contain considerable amounts of PPCPs. Application of treated water and sewage sludge for agricultural purposes may have environmental risks. Contamination by veterinary drugs is mainly from application of manure as fertilizer and/or leakage from manure storage facilities.

treatment plants and/or manure of livestock as fertilizers, and possible discharge from wastewater treatment plants (Scheytt et al. 2007; Chee-Sanford et al. 2009). It has been reported that the application of sludge slurries to soils poses little risk to adjacent water and crops because the soil retains PPCPs and field application increases PPCP biodegradation (Topp et al. 2008a). However, some of the PPCPs could leach through soil profiles and may contaminate groundwater. In addition, surface runoff may enhance the transport of PPCPs, thus spreading them to places other than application areas. Previous studies indicated that injection of sludge below the soil surface effectively reduced the runoff (Topp et al. 2008b). Another problem is that application of PPCP-containing water or sludge in farm areas could also induce the development of drug-resistant genes in the application area, and these drug-resistant genes could be transported in the environment (Chee-Sanford et al. 2009). Therefore, the environmental risk from land application of wastewater, sludge, or manure should be carefully considered

because of possible contamination from PPCPs and propagation of resistant genes. 7.2.2. PPCP Distribution in Soils and Sediments Soils/sediments are high-capacity sinks for PPCPs that are adsorbed and/or bound, then retained by soils/sediments. Table 7.4 summarizes the occurrence of PPCPs in soils/ sediments that have been detected in various countries. The most widely detected PPCPs are TCs, sulfonamides, and fluoroquinolones. Information on PPCP occurrence in water bodies is more widely available than are surveys or studies of PPCP occurrence in soils and sediments because of analytical difficulties and the limitations of current technologies (Wilga et al. 2008). The analytical challenges include low PPCP concentration levels, the association of PPCPs with the solid matrix, the presence of various metabolites, and problems with extraction and separation. Analysis of PPCP concentrations in the environment usually involves the application of

192

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

TABLE 7.4. Occurrence of PPCPs in Soils/Sediments in Various Countries Country

Matrix

PPCPs

Concentration, mg/kg

Reference

United States

South San Francisco Bay sediment

MX MK SMZ, STA TC CTC CTC TYL ENR CIP LCM SDZ TMP OTC TC CTC SMZ TML TC CTC CIP NOR OTC FMQ

0.034  0.021 0.038  0.026 0.10–12.4 86.2–171.7 4.6–7.3 10–15 20–55 13–204 53 8.5 1 0.5 305 35–295 4–39 0–2 0.7 86.2–198.7 4.6–7.3 270–2420 240–2370  246.3  578.8

Rubinfeld and Luthy (2008)

Soil and sediments

Denmark

Soil at research station

Turkey

Soils NW and central Turkey at two depths

UK

Soils around livestock feeding area

Germany

Soils and manure treated soils

Soils in Dortmund Soils and manure-treated soils Switzerland

Sludge and soil near Zurich

Italy

Sediments in fish farms

Kim and Carlson (2005)

Jacobsen et al. (2004) Uslu et al. (2008) Boxall et al. (2005)

Hamscher et al. (2005)

Schlusener et al. (2003) Hamscher et al. (2002) Golet et al. (2002) Lalumera et al. (2004)

Notation: MX—musk xylene; MK—musk ketone; SMZ—sulfamethazine; STA—sulfathiazole; TC—tetracycline; CTC—chlortetracycline; TYL—tylosin; ENR—enrofloxacin; CIP—ciprofloxacin; SDZ—sulfadiazine; TMP—trimethoprim; NOR—norfloxacin; FMQ—flumequine; TML—tiamulin; LCM— lincomycin.

expensive instruments, such as high-performance liquid chromatography coupled with mass spectrometry (HPLCMS) or tandem mass spectrometry (HPLC-MS/MS), which also restricts wide PPCP measurements of soil and sediment samples. Various mechanisms are proposed to explain soil/ sediment–PPCP (or soil/sediment components) interactions, such as hydrophobic interaction, ion exchange, H bonding, surface complexation, cation bridging, electron–donor– acceptor systems, inner/outer-sphere complexation, and/or electrostatic interactions (Table 7.5). soil/sediment–PPCP interactions should not be viewed as separate, individual mechanisms, but as several working together. The possible mechanisms involved are discussed in the following sections.

7.3. INTERACTIONS BETWEEN SOIL/SEDIMENT ORGANIC MATTER AND PPCPs 7.3.1. Linear Free-Energy Relationship for Predicting PPCP Sorption Coefficient The linear free-energy relationship (LFER) has been systematically discussed to predict hydrophobic organic chemical (HOC) adsorption. Soil/sediment organic matter (SOM) is believed to be the most important fraction controlling the

environmental behavior of HOCs. The organic carbon normalized adsorption coefficient (KOC) values are employed to describe the interaction between HOCs and soils/sediments. For HOCs, KOC was reported to be linearly related to KOW, and LFERs relationships have been proposed to predict KOC values based on HOC physicochemical properties, such as solubility and KOW (Chiou et al. 1979). The interactions between PPCPs and SOM also affect PPCP environmental fate, for example, enhancing sorption, reducing bioavailability (Jacobsen et al. 2005), and thus reducing PPCP environmental risk (Fan et al. 2007). Correlations between PPCP sorption parameters and soil/sediment properties have been widely discussed. Soil/sediment organic matter was found to be the most important factor that controls PPCP sorption in soils/sediments as indicated by the significantly positive relationship between the Freundlich sorption parameter KF or the distribution coefficient Kd of PPCPs and the organic carbon content of soils/sediments (During et al. 2002; Bowman et al. 2002; Holthaus et al. 2002; Loffredo and Senesi 2006; Maskaoui et al. 2007; Uslu et al. 2008; Ottmar et al. 2010). However, the application of KOC to describe soil/sediment–PPCP interactions has been unsuccessful (Tolls 2001; Holbrook et al. 2004; Liu et al. 2005; Patrolecco et al. 2006; Zhou et al. 2007). Several studies attributed this variation and failure to differences in SOM properties (Thiele 2000; Fan

INTERACTIONS BETWEEN SOIL/SEDIMENT ORGANIC MATTER AND PPCPs

193

TABLE 7.5. Mechanisms Involved in PPCP Sorption in Soils/Sediments (or Soil/Sediment Components) Mechanism

PPCP

Sorbent

Evidence

Reference

Hydrophobic interaction

TC

DHA

Refute: KOW predicted KDOC values substantially lower than experimental results; negligible hydrophobic interaction

Gu et al. (2007)

OTC

HDTMA-MMT

Confirm: At pH 5.0, increased adsorption compared to other pHs; FTIR spectrum showed disappearance of four aliphatic peaks at 2930, 2849, 1490, and 1473 cm
1 with appearance of small residual peaks at 2923 and 2848 cm
1 and fingerlike region from 1500 to 1300 cm
1; this feature is not observed in other pHs

Kulshrestha et al. (2004)

SMX etc.; CLA

DOM; DHS

Confirm: Correlation between KDOC and KOW

TC OTC

DHA HA

OTC

Clays

OTC

Clays

Competition between Na þ and TC At pH 5.5, higher sorbed Ca concentrations resulted in lower OTC sorption Decreased adsorption of cation species with increased ionic strength Adsorption decreases with increased pHs; when inorganic cations were replaced by HDTMA, montmorillonite showed negligible adsorption

Maskaoui et al. 2007; Sibley and Pedersen (2008) Gu et al. (2007) MacKay and Canterbury (2005) Figueroa et al. (2004)

TC

DHA

OTC0

Clays

H-bond

CLA

DHA

CLA-DHA showed strongest interaction at pH 6–7; at acidic pH, intra/inter molecular interactions collapse molecular conformation and diminish external H bonding; at alkaline pH, deprotonation of DHA functional groups increases repulsion of inter intramolecules, thus diminishing capacity to donate H bond

Sibley and Pedersen (2008)

Cation bridge

OTC

HA

MacKay and Canterbury (2005)

TC

DHA

TC

HAO and HFO

Sorption of OTC to HA increased greatly as HA-bound concentrations of Al3 þ or Fe3 þ increased Increased adsorption coefficient after Ca2 þ amendation at pH > 5 Increase of ionic strength lowered sorption only at high loadings; lack of competition from Na þ and Cl
at low loadings indicated initial comlexation via innersphere complexation; decreased surface charge with increasing sorption

Ion exchange

Surface complex

Inner-sphere complex

At pH <4.3, zwitterionic and cationic species compete with H þ ; at pH > 7, sorption decreases dramatically because of dominant anion species Surface protonation: (1) pH increase was proportional to OTC0 adsorption; (2) saturation of ions decreased adsorption; (3) net increase of pH was greater for metal-saturated montmorillonite than untreated montmorillonite

Kulshrestha et al. (2004)

Gu et al. (2007)

Figueroa et al. (2004)

Gu et al. (2007) Gu and Karthikeyan (2005a)

(continued )

194

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

TABLE 7.5. (Continued) Mechanism

PPCP

Sorbent

Evidence

Reference

CBX

Clays and soils

Much stronger sorption on K-dominated adsorbents than Ca-saturated ones; the large hydration shell of Ca2 þ hinders formation of inner-sphere complex in comparison with K þ

Strock et al. (2005)

Electron donor–acceptor

CBX and associated N-oxide reduced metabolites

Soils, sediments, and clays

N-oxide (electron acceptor) and siloxane (electron donator)

Strock et al. (2005)

Electrostatic interaction

SCP, TYL, OTC

Soil

Sorption decreases as pH increases; increased sorption of SCP with increased ionic strength; soil particles are generally negatively charged; presence of H þ or cations could neutralize soil charges and thus increase sorption

ter Laak et al. (2006b)

et al. 2006; Stumpe and Marschner 2007), but the literature relating SOM properties to PPCP sorption is very limited. Sun et al. (2007) suggested that PPCP sorption was the result of aromaticity of SOM, but other researchers have reported that PPCPs interact with both polar and nonpolar surfaces because of their amphiphilic properties (Suntisukaseam et al. 2007). Hence, it is understandable that PPCP sorption could not be predicted well from organic carbon content alone or a certain SOM fraction. Both organic and inorganic fractions contribute to PPCP sorption in soils/sediments (Casey et al. 2003; Strock et al. 2005; Pan et al. 2009). Investigators pointed out that none of the specific soil parameters such as pH, organic matter content, cation exchange capacity (CEC), or mineral fractions (silt, clay and sand) could explain the variation of PPCP sorption coefficients (Yeager and Halley 1990; ter Laak et al. 2006a). ter Laak et al. (2006a) applied partial least-squares models to integrate all the soil properties (pH, organic carbon content, clay content, cation exchange capacity, aluminum oxyhydroxide content, and iron oxyhydroxide content), and the variation in sorption coefficients was explained by as much as 78%. A concept of using artificial neural networks was proposed to relate PPCP adsorption coefficients with molecular descriptors such as log P, pKa, molar refractivity, aromatic ratio, hydrophilic factor, and topological surface area (Barron et al. 2009). The predicted Kd values seem to closely match the experimentally determined values. The general failure to relate KOC with KOW (or SW) can also be understood from the physicochemical properties of PPCPs. As summarized in Table 7.5, log KOW values vary from
2.34 to 4.15, indicating that PPCPs cover a wide range of chemicals, including very hydrophilic compounds to strongly hydrophobic chemicals. The common functional

groups found in PPCPs are –OH, –COOH, –C¼O, and –NH2, and the acid dissociation constants (pKa) are in the range of 2.3–10.4. As a result, in the range of pH values found in the environment, PPCPs can exist as different species. Since no common model can be adopted to include the variation of KOC values caused by the physicochemical properties of PPCPs. A possible method is to classify PPCPs according to their similarity in properties, and different models should be applied for different classes of PPCPs. 7.3.2. Nonideal Interactions Nonideal interaction is different from ideal interaction in terms of linear partitioning and lack of competition. Nonideal interactions were commonly observed for PPCPs in soils/ sediments, such as nonlinear sorption (Li and Lee 1999; Thiele-Bruhn et al. 2004; Yu et al. 2004,2006; Loffredo and Senesi 2006; Zeng et al. 2006; Gu et al. 2007) and desorption hysteresis (Li and Lee 1999; During et al. 2002; Zeng et al. 2006; Williams et al. 2006; Gu et al. 2007). Sorption kinetics of PPCP in soils/sediments showed nonideal behavior as indicated by concentration-dependent sorption kinetics (Yu et al. 2004), two-stage sorption (Loffredo and Senesi 2006; Zhou 2006), and three-site sorption (Wehrhan et al. 2007). Nonlinear sorption and desorption hysteresis have also been attributed to black carbon (Yu et al. 2006; Zeng et al. 2006). However, it is easily predicted that the heterogeneous nature of soils/sediments will contribute to nonideal PPCP sorption. A dual-mode model used for nonpolar organic chemical sorption was used to describe the sorption of bisphenol A on sediment samples (Zeng et al. 2006). According to the dual-mode model for nonpolar chemicals, slow sorption is controlled by diffusion in the

IMPORTANCE OF THE ORGANIC FRACTION

glassy (hard-carbon) domain (Pignatello and Xing 1996). Prolonged sorption processes are attributed to the diffusion and entrapment of PPCPs in SOM (Fan et al. 2006). However, the enlarged interlayer space in inorganic minerals could also be responsible for the kinetic process as discussed in the next section. Therefore, PPCP sorption kinetics is likely controlled by both organic matter diffusion and sorption-retarded pore diffusion (Pignatello and Xing 1996), and the dualmode model of SOM alone may be oversimplified to describe PPCP sorption in soils/sediments because of the contribution from the mineral particles. 7.3.3. Bound Residues As adopted by International Union of Pure and Applied Chemistry (IUPAC), pesticide non-extractable or bound residual fraction is defined as “chemical species originating from pesticides, used according to good agricultural practice, that are unextracted by methods which do not significantly change the chemical nature of these residues” (Roberts 1984). The formation mechanisms of bound residues include ionic bonding, hydrogen binding, van der Waals forces, ligand exchange, charge transfer complexes, covalent bonding, hydrophobic partitioning, and sequestration (Gevao et al. 2000). Non-extractable fractions were also observed for PPCPs in soils/sediments (Colucci et al. 2001; Colucci and Topp 2002; Ericson 2007). Ericson (2007) emphasized that nonextractable residues should be attributed to irreversibly bound pharmaceuticals or their metabolites rather than CO2 trapped as bicarbonate or reincorporated into biomass. These bound residues are believed to be retained in soils/ sediments and have limited toxic effects (Colucci and Topp 2002; Ericson 2007). However, it was documented that soil-bound antibiotics are still biologically active (Chander et al., 2005). In addition, bound residues can potentially be released during changes in environmental conditions (e.g., change of temperature, pH, and ionic strength; the presence of complexing agent EDTA, surfactant, or dissolved organic matter), then posing environmental risk (Eschenbach et al. 1998). How to consider and define the extent of PPCP-bound residue release will directly guide the establishment of soil/sediment environmental quality standards. To date, studies on PPCP-bound residues in soils/sediments are limited, but good reviews are available for pesticidebound residues (Gevao et al. 2000). The concepts and methods used to investigate pesticide-bound residues may be applied for PPCP studies because of the similarities between pesticides and PPCPs (e.g., functional groups, pH-dependent species, polarity). Covalent binding between SOM and PPCPs was reported in several studies. The active sites on SOM that could form covalent bonds with PPCPs include quinone-like moieties (Bialk et al. 2005, 2007), carboxylic groups, N-heterocyclic structures, and lignin decomposition products (Thiele-Bruhn

195

et al. 2004). The trapped PPCPs may be persistent in soils for several years, and the half-lives of PPCPs could be extended for about 100 times (Forster et al. 2009). Understanding interaction mechanisms requires identification of intermediate products, and thus extensive spectroscopic examination and characterization (e.g., mass spectrum, NMR, IR, XPS) as well as computational chemistry methods are needed to obtain relevant information. In addition to chemical binding, threedimensional (3D) conformation may also play a role in PPCP nonextractable residues. For example, sulfadiazine geometric structure and torsion angles showed that this compound is highly flexible (Huschek et al. 2008). Sulfadiazine could be easily reconformed and fit well with adsorbents, and thus nonextractable residues could be rapidly formed after contact with soils/sediments. Again, it should be noted that for PPCPs, strong binding is not limited to SOM. The interactions between PPCPs and inorganic fractions in soils/sediments contribute greatly to controlling PPCP environmental behavior, as discussed in the next section.

7.4. IMPORTANCE OF THE INORGANIC FRACTION 7.4.1. Mineral Structures Soil inorganic particles are usually classified according to size designations: sand (0.05–2.00 mm), silt (0.002– 0.05 mm), and clay (<0.002 mm). It should be noted that clay in size fractions is a general term consisting mostly of secondary minerals, such as aluminosilicate and oxides. The term clay mineral more specifically refers as to phyllosilicates, including kaolin, smectites, micas, and chlorite groups. To avoid any confusion, the term clay in this chapter refers to secondary mineral clays. Structures of typical clay minerals are illustrated in Figure 7.3. The kaolin group minerals have one octahedral sheet bonded with one tetrahedral sheet. This group of minerals includes dioctahedral minerals (kaolinite, dickite, nacrite, and halloysite) and the trioctahedral minerals (antigorite, chamosite, chrysotile, and cronstedite). The smectitic group includes dioctahedral smectites (i.e., montmorillonite and nontronite) and trioctahedral ones (i.e., saponite). These minerals have a structural unit of two inward-pointing tetrahedral sheets with a central octahedral sheet. These minerals are expandable because water molecules and exchangeable cations can diffuse into the interlayers of smectites. Mica group minerals are nonexpanding, clay-sized, micaceous minerals. Their basic unit is a layer composed of two inward-pointing silica tetragonal sheets with a central octahedral sheet. The chloritic group includes a wide variety of similar minerals with considerable chemical variation. This group of minerals has a 2 : 1 sandwich structure (tetrahedral– octahedral–tetrahedral), and their interlayer space is a brucitelike layer (Mg2þ , Fe3 þ )(OH)6, and thus is not expandable.

196

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

Montmorillonite

Kaolinite

Mica

Water and exchangeable cations O

OH

Si, Al

K

Al, Mg, Fe

Figure 7.3. Structures for typical clay minerals. Clay minerals are commonly denoted as 1 : 1 and 2 : 1 minerals according to their structural organization of tetrahedral and octahedral sheets. Kaolinite is denoted as a 1 : 1 clay (one tetrahedral sheet and one octahedral), while montmorillonite and mica are denoted as 2 : 1 clays (one octahedral sheet sandwiched between two tetrahedral sheets). (See insert for color representation of this figure.)

Another group of mineral particles that interacts strongly with PPCPs is metal oxides, such as gibbsite, goethite, and birnessite. This type of mineral particles includes hydroxides, oxyhydroxides, and hydrous oxides depending on their hydration state (Sparks 2003). In the middle of the oxides, the metals share OH
or O2
, and strong coordination and stable structure can be formed. However, at the edge of the oxides, the unshared OH
may protonate as a function of pH and thus provides active adsorption sites for PPCPs. 7.4.2. Mechanisms Involved in PPCP Sorption on Soil/Sediment Inorganic Fractions Various studies have indicated the importance of inorganic minerals for PPCP sorption in soils/sediments (Lai et al. 2000; Schafer et al. 2002; Jones et al. 2005; Strock et al. 2005). Pan et al. (2009) summarized PPCP sorption on different adsorbents. The calculated single-point sorption coefficient Kd (at Ce ¼ 10 mg/L) showed comparable values for mineral particles and soils/sediments. Some studies have indicated that coating of SOM on inorganic particle surfaces could mask sorption sites or inhibit interlayer diffusion in clays, thus reducing PPCP sorption [such as TC in Pils and Laird (2007)]. In addition, the adsorption of PPCPs and SOM on inorganic particles may share common adsorption sites due to their similar functional groups, most likely the polar functional groups (Kang and Xing 2008). Therefore, inorganic particles in soils/sediments play very important roles in PPCP adsorption. Understanding the relationship between the properties of inorganic particles and PPCP sorption is critical to revealing PPCP-soil/sediment interaction mechanisms. Mineral properties, including surface area, particle size, permanent charge, and CEC, could all be important factors

depending on the interaction mechanisms between PPCPs and solid particles. For absorption characterized predominantly by cation exchange and bridging (e.g., ciprofloxacin adsorption on minerals), higher permanent charge and CEC resulted in higher adsorption (Carrasquillo et al. 2008). However for adsorption controlled by surface complexation, soil metal oxides and edge sites on aluminosilicate clays are more important (Carrasquillo et al. 2008). Clay mineral–PPCP interaction mechanisms could also be discussed according to PPCP conformational properties. The tetrahedral –SO2– group in sulfamethazine prevents close, coplanar orientation of the aromatic moiety with the siloxane surface of smectites. Thus p complexes of basal plane oxygens on smectite surface with either the aniline or 4,6dimethylpyrimidine rings of sulfamethazine could be excluded. Water and cation bridging was proposed to explain sulfamethazine adsorption on smectite surfaces (Gao and Pedersen 2005). Hydrophobic interaction is also expected for PPCP adsorption on mineral surface. The pH-independent adsorption of 17b-estradiol (E2) on kaolinite, illite, and montmorillonites implicated hydrophobic interaction and adsorption contributed by edge hydroxyls is very limited (Van Emmerik et al. 2003). Siloxane surfaces are an active sorption domain for PPCPs. Because of the hydrophobic nature of the siloxane surface, organic chemicals could be adsorbed through hydrophobic interaction, especially when hydrophilic, inorganic cations are replaced by hydrophobic molecules. Jaynes and Boyd (1991) investigated the adsorption of aromatic hydrocarbons by organoclays. The hydrophilic, exchangeable cations of smectites were replaced by a small hydrophobic organic cation, trimethylphenylammonium (TMPA). Although TMPA cations could keep the smectite interlayer open, the authors observed decreased sorption of aromatic

IMPORTANCE OF THE ORGANIC FRACTION

hydrocarbons with increased TMPA content. Thus, the main sorption region is not the interlayer, but the hydrophobic siloxane surface. In addition, the siloxane surface could be viewed as electron donors to form electron–donor–acceptor systems with electron-accepting PPCPs. Weissmahr et al. (1998) observed that sorption of p-acceptor organic chemicals on some naturally occurring minerals was several orders of magnitude higher than expected from nonspecific interactions. These minerals were phyllosilicates that contain the electron–donor siloxane. The authors emphasized electron–donor–acceptor interactions between p-acceptor organic chemicals and the siloxane surface. Adsorption of substituted nitrobenzenes on K-smectite clay depended greatly on the abilities of substituents to complex additional interlayer cations and their water solubilities (Boyd et al. 2001). This adsorption, however, could be explained independently from electron donor–acceptor theory. Therefore, the role of the electron donor–acceptor mechanism in PPCP adsorption could be case-dependent, and extensive study is needed to provide a general view for the applicability of this theory. Other than the abovementioned mechanisms, discussion of PPCP–clay mineral interaction mechanisms should also consider the structural features of different mineral particles, namely, layer structures and nonexpanding lattice. Mineral particles with layer structures always show special sorption characteristics with organic chemicals. For example, mont-

morillonite showed slower sorption kinetics, higher sorption capacity, and stronger desorption hysteresis for estrogens (Van Emmerik et al. 2003; Bonin and Simpson 2007). The authors compared the sorption of E2 from aqueous solutions by goethite, an iron oxide, and by kaolinite, illite, and montmorillonite. Among these adsorbents, much higher adsorption of E2 was observed on montmorillonite. Although initial sorption to the clay minerals was similar, 65% of total E2 was taken up by montmorillonite after a period of 3 days, in contrast to 10%–15% by kaolinite and illite. Water or methanol extraction could easily and quickly recover sorbed E2 from goethite, kaolinite, and illite, but none was desorbed into either solvent from montmorillonite. The authors indicated that E2 is adsorbed on the surface of goethite, kaolinite, and illite, but diffused into the interlayers of montmorillonite. This strong interaction between E2 and montmorillonite may at least partly explain desorption resistance and bound residue in soils/sediments. Sorption properties similar to those of montmorillonite were also observed for bisphenol A, 17a-ethynylestradiol, and estrone (Shareef et al. 2006). For dry montmorillonite, the spacing (d spacing) of a layer unit is about 1.0 nm. Once it is wet, the interlayer spacing can expand up to 3 nm (van Olphen 1977) (Fig. 7.4). Most PPCPs could easily fit into montmorillonite expanded interlayers, and the sorbed molecules are resistant to desorption or extraction using organic solvents (Van Emmerik et al. 2003).

Oven dried montmorillonite Silica tetrahedra Alumina octahedra Silica tetrahedra Silica tetrahedra ~1.0 nm

Alumina octahedra Silica tetrahedra

Silicon Oxygen Hydrogen Aluminium

Wet montmorillonite Silica tetrahedra

Cations

Alumina octahedra Silica tetrahedra

Water molecules Hydrated exchangeable cations

3 nm or more

197

Silica tetrahedra

Oxytetracycline

Alumina octahedra Silica tetrahedra

Figure 7.4. Interlayer spacing of montmorillonite under different conditions. In dry conditions, montmorillonite interlayer is occupied mostly by exchangeable cations and the d spacing is around 1.0 nm; in conditions, wet, water molecules penetrate into the interlayers. The interlayer is filled with water and hydrated exchangeable cations. The spacing of a layer unit is increased to 3 nm, and most PPCP molecules can fit into these sites. (See insert for color representation of this figure.)

198

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

Kulshrestha et al. (2004) also observed that the high adsorption of oxytetracycline (OTC) on montmorillonite at pH 1.5 and 5.0 was accompanied by significant increase of interlayer space. The increase of montmorillonite c-axis spacing was about 0.9 nm, which is in good agreement with the dimension of OTC (0.83 nm). The authors proposed tilted orientation of OTC molecules in interlayer spaces. In some of the cases, because of the presence of interlayer cations, surface charges, and the associated water molecules, the interlayer space may not be easily available for all chemicals, especially for hydrophobic ones. However, exchangeable cations may complex with PPCPs via inner/outer-sphere complexation (Akyuz and Akyuz 2003). For minerals with a nonexpanding lattice (such as illite and kaolinite) and oxides, surface sorption and diffusion into voids or regions within mineral aggregates are expected (Van Emmerik et al. 2003). Spectroscopic analysis would provide useful information to reveal sorption mechanisms. Gu and Karthikeyan (2005b) compared the ATR-FTIR spectra of free ciprofloxacin and oxide–ciprofloxacin complexes at different pH levels and clearly demonstrated different sorption mechanisms of ciprofloxacin on hydrous oxides of Al (HAO) and Fe (HFO) (Fig. 7.5). For HAO, a monodentate mononuclear complex (with –COO–) was proposed. While for HFO, a mononuclear bidentate complex (i.e., a sixmembered ring) with an Fe atom on the HFO surface through the keto O and one of the oxygen in the carboxylate group was observed (Gu and Karthikeyan 2005b).

7.5. SORPTION AS AFFECTED BY SOLUTION CHEMISTRY 7.5.1. pH One of the most important properties for PPCPs is their various species at typical environmental pH levels. Because pKa values of some PPCPs are in the range of pH values in soils/sediments, these compounds can be protonated or deprotonated in soil/sediment solutions, and thus can exist as different species, such as cations, zwitterions, and anions (Figueroa et al. 2004; MacKay and Canterbury 2005; Gu et al. 2007). Different mechanisms for PPCP–soil/sediment interactions are involved for different species (Kulshrestha et al. 2004). Thus, investigation of PPCP adsorption at different pHs could reveal the contribution of different mechanisms. Van Emmerik et al. (2003) reported that at pH around 7, a maximum adsorption is observed for E2 on goethite, indicating that E2 is bound mostly with an uncharged surface. At lower pH, hydroxyl groups on goethite were protonated and the goethite surface polarity increased, E2 adsorption decreased. An increase of the pH above the pKa of E2 and the point of zero charge for goethite would result in negatively charged sorbent and sorbate; thus, electrostatic repulsion between E2 and goethite reduced the amount of E2 sorption. However, at alkaline pHs (pH 8.7–11.0), low but obvious adsorption of OTC was observed on native montmorillonite

1628 1709

1580

Free CIP at pH 5 1385 CIP on HFO at pH 5 1360

1619

Absorbance

1633

CIP on HAO at pH 5

1354

1640

1627 1358

1800

1600

1400

Wavenumber

1200

(cm-1)

Figure 7.5. ATR-FTIR evidence for ciprofloxacin sorption on hydrous oxides of Al (HAO) and Fe (HFO). The absence of the peak at 1709 cm
1 for HAO and HFO complexes compared to the spectrum of free ciprofloxacin in solution indicates that the carboxylic group is involved in the formation of the ciprofloxacin–oxide complex. The peak at 1628 cm
1 downshifted to 1619 cm
1 for sorption on HFO, but no shift on sorption on HAO. This spectral characteristic suggests that a keto group participates in surface complexation on HFO, but not on HAO. [Modified from Gu and Karthikeyan (2005b).] (See insert for color representation of this figure.)

SORPTION AS AFFECTED BY SOLUTION CHEMISTRY

(Kulshrestha et al. 2004). The authors attributed this sorption to some positive charges around the edges (Al3 þ cations). Figueroa et al. (2004) attributed OTC adsorption to the sum of cation species and zwitterionic species. They fitted the adsorption data at different pHs, and the contribution from different species was identified. As a result, they observed 20 times higher adsorption for cation species than zwitterionic species. On the other hand, decreased sorption of PPCP at decreased pH was also reported (ter Laak et al. 2006b), which was explained by the competition between cations (H þ ) and positively charged PPCP species or complexes at acidic pHs. In addition, cation bridging could also occur at pH higher than the pKa of PPCPs (Figueroa et al. 2004; Jia et al. 2008). Thus, the effect of pH on PPCP sorption needs to be considered case by case. Sorption mechanisms of PPCP at different pH levels are summarized in Figure 7.6. In general, cation exchange has been proved to be the most important mechanism for the sorption of PPCP cation species, while for the zwitterion species, surface complexation (Figueroa et al. 2004) and

hydrophobic interactions (Kulshrestha et al. 2004; Sibley and Pedersen 2008) are important. Therefore, species-specific sorption coefficients should be coupled in PPCP–soil/sediment modeling (ter Laak et al. 2006b). In addition, the increase in pH increases PPCP solubility (Campbell et al. 2006; Zeng et al. 2006), and decreases their sorption. Figure 7.6 indicates that various mechanisms may play roles simultaneously in PPCP sorption. Identifying the contribution of individual mechanisms may greatly facilitate the prediction of PPCP environmental behavior. For example, MacKay and Seremet (2008) used two compounds to probe the mechanisms of surface complexation and cation exchange, and the adsorption of ciprofloxacin to soil could be estimated from the contribution of these two mechanisms. Figure 7.6 also reminds us that sorption experiments conducted at various pH values could enable us to focus on one or a few mechanisms depending on PPCP species. Alternatively, hydrophobic interactions could be excluded when PPCPs are adsorbed from an organic solvent (e.g., hexadecane) to the solid phase. Extensive work on this topic

Hydrophobic Interaction

Cation exchange

H Bond

Electrostatic Attraction

Electrostatic repulsion

Adsorption Coefficient K

Electrostatic repulsion

pKa of PPCPs

199

pHzpc of adsorbents

pH

Figure 7.6. Potential PPCP sorption mechanisms in minerals with variable charge at different pH levels. At pH lower than pKa, the sorption will decrease with decreased pH because of the electrostatic repulsion between PPCPs and adsorbent (both of them are positively charged). At pH around pKa, the zwitterionic species could interact with minerals through hydrophobic interactions. In the pH range between pKa of PPCPs and pHzpc (zero point of charge) of adsorbents, PPCPs are negatively charged while adsorbents are positively charged. Thus electrostatic attraction and cation exchange are expected. As pH further increases, both PPCPs and adsorbents are negatively charged and the electrostatic repulsion results in a substantial decrease of PPCP sorption. [Modified from Zhang et al. (2010).]

200

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

would help establish comprehensive models to predict PPCP sorption based on PPCP functional groups and their classification. 7.5.2. Ionic Strength Both increased and decreased PPCP sorption in soils/sediments was observed with the increase in ionic strength. Because of metal bridging, ternary complexation of PPCP– metal–sorbent could be formed. Ternary complexation of PPCP–metal–SOM or metal-PPCP-SOM could enhance their sorption capacity and strength in soils/sediments (MacKay and Canterbury 2005; Gu et al. 2007). Sorption of PPCP is reported to increase with increasing salinity (Zhou and Liu 2000; Bowman et al. 2002; Zhou 2006). Salting- out of PPCP molecules is an important mechanism (Zhou 2006), but ternary complexation could also contribute greatly to this process. However, metals could also compete with PPCPs for soil/ sediment sorption sites depending on PPCP charges (Sibley and Pedersen 2008; Bai et al. 2008) (Fig. 7.7). A typical case is that PPCP sorption is dominated by cation exchange. Decreased pH or the presence of metal ions could significantly inhibit PPCP sorption because of their competition with H þ or metal ions (Kulshrestha et al. 2004; Gu et al. 2007). Metal ions could form stable complexes with PPCPs. For example, the active sites for sulfamethoxazole binding with cadmium are sulfonamide nitrogen and sulfonic oxygen (Kesimli and Topacli 2001). ter Laak et al. (2006b) stated that the formation of Ca-OTC complexation does not influence adsorption directly. However, if the complex is positively charged, competition between the complex and metal ions is

Aqueous phase

+ + -

+

expected (Fig. 7.7). Studies on the effect of PPCP-cation complex on PPCP sorption in soils/sediments are limited. As summarized in Table 7.6, PPCP sorption could increase or decrease in the presence of cations. Rational explanations were proposed to describe the opposite effects. However, consensus has not yet been reached on how to predict the effect of ionic strength on PPCP sorption, indicating that further study is needed. One way to promote our understanding on the ionic strength effect on PPCP sorption is to improve the experimental designs and procedures in this line of research. In previous studies, the effect of ionic strength was investigated in terms of total applied concentration. How the distribution of ions between aqueous and solid phase is unclear. Fundamentally, the effect of ionic strength is attributed to the fact that adsorption of ions on solid surface changes the properties of the surface. But if PPCPs are complexed with cations, chemical properties of the PPCP could be altered considerably. Therefore, it is essential to identify the ions complexed with adsorbent or PPCPs as well as freely dissolved ones in the aqueous phase (Fig. 7.8). The apparent PPCP sorption in the presence of cations cannot be completely explained until the distribution of cation species in the sorption system is understood. Different methods can be applied to distinguish cation species in sorption systems. For example, ion-selective electrodes can identify freely-dissolved ions in the aqueous phase, and atomic absorption spectrometry measures the total ion concentration. Combinations of these two techniques could reliably quantify different cation species. In studies of the environmental fate and toxicity of heavy metals, metal speciation (with an emphasis on freely dissolved or bound metals) is calculated using various models,

Electrostatic attraction + +

-

Adsorbent Cation exchange

+ - + +

-

+ +

-+

PPCP molecule

+

Cation Exchangeable cation

+ - +

-

Electrostatic attraction

+

+

+

-

Ligand exchange

Negatively charged site

Cation bridging

Figure 7.7. Schematic illustration of adsorption of PPCPs on solid particles in the presence of cations. Cations may compete with positively charged PPCPs or PPCP–cation complex for adsorption sites of cation exchange and electrostatic attraction. Cations could also promote PPCP sorption through cation bridging. Hydrogen bonding and hydrophobic interaction may not be dramatically affected by cations; thus they are not shown in this figure. (See insert for color representation of this figure.)

SORPTION AS AFFECTED BY SOLUTION CHEMISTRY

201

TABLE 7.6. PPCP Sorption Affected by Ionic Strength Sorbent

Sorbate

Ions

Effects

Explanation

Reference

Sediment

BPA

Cd, Pb

Increase

Li et al. (2007)

Soil

SCP, TYL, OTC

Ca

Decrease

Dissolved humic acid

CLA

K, Na

Decrease

Soil

TC

Cu(II)

Acid: decrease; Alkaline: increase

Manure and humic acid

STA

Ca

Increase

Oxides

TC

Na

Sediment

OP

Humic substances

CBZ

Salinity 0.5%–3.5% Cu(II)

At low sorption; no effect; at high sorption: decrease Increase

(1) Screen the negative charges of sediments and thus decrease the repulsion; (2) salting-out effect, decrease BPA solubility; (3) inhibit release of organic matter; (4) pH decreases with ionic strength Competition of electrolyte cations with positively charged TYL species and positively charged OTC complexes Competition; K þ impact is more significant;. K þ has smaller hydration shell and binds more strongly with humic acid than does Na þ Acid: competition of Cu2 þ with TC and TC-Cu complexes and to increased positive surface charge of soil by Cu (II) adsorption; alkaline: cation bridging Neutral cation-STA
pairs: neutral STA via hydrogen bonds but also via van der Waals forces to aromatic parts of the organic sorbents At low sorption, inner-sphere mechanism; higher sorption, outer-sphere mechanisms Salting-out of 4-tert-octylphenol in increased salinity Competition between Cu and CBZ

decrease

ter Laak et al. (2006b)

Sibley and Pedersen (2008)

Jia et al. (2008)

Kahle and Stamm (2007)

Gu and Karthikeyan (2005) Zhou (2006) Bai et al. (2008)

Notation: TC—tetracycline; OP—4-tert-Octylphenol; CBZ—carbamazepine; STA—sulfathiazole; CLA—clarithromycin; SCP—sulfachloropyridazine; TYL—tylosin; OTC—oxytetracycline; BPA—bisphenol A.

including MINTEQA and FITEQL. These models could be used in PPCP sorption systems to calculate metal speciation. In addition, differentiating PPCP species (free PPCP and metal-bound PPCP) could also facilitate our understanding of PPCP sorption mechanisms under the effect of ionic strength. Werner et al. (2006) measured pKa of TC species and equilibrium constants for the association of Ca2þ and Mg2þ with aqueous TC species. With these constants, TC speciation distribution could be easily calculated at different pHs and water hardness. The FITEQL model was then applied to calculate PPCP species (Gu et al. 2007). Alternatively, free PPCP and metal-bound PPCP could be identified from combining quantitative measurements using liquid scintillation counting (LSC) and high-performance liquid chromatography (HPLC) (Gu and Karthikeyan 2005a). HPLC is compound-specific and responds only to free forms, while LSC provides total concentration. If the methods to identify PPCP species as well as cation species could be integrated in a single study, undoubtedly, a better understanding on PPCP sorption mechanisms will be obtained.

For nonpolar organic chemicals, the effect of ionic strength generally focused on adsorbent conformational changes, specifically for SOM. Metal ions could extensively complex with SOM and alter its 3D structure. For example, Kazpard et al. (2006) characterized the hydrophobic microenvironment of organic matter using I1/I3 of pyrene, which is the fluorescence intensity ratio between the first (at emission wavelength 372 nm) and the third (at emission wavelength 383 nm) peaks. This parameter gives a lower value if pyrene is in a less polar environment. Their results indicated that Al3 þ could induce the formation of an intramolecular hydrophobic microenvironment, while further addition of Al3 þ results in an opposite effect (Kazpard et al. 2006). From a more general viewpoint, the effect of ionic strength on KDOC could be viewed as a three-stage conformation change, namely, variation of organic matter structural configuration, organic matter aggregation, and salting-out effect (Lee et al. 2003). This topic has not been studied for PPCP sorption mechanisms. Therefore, it is safe to say that limited information is available on PPCP–metal–

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

Metal adsorbed by sorbent

Metal complexed with PPCPs

Adsorbed/complexed metal concentration

PPCP adsorption coefficient (K)

202

Total cation concentration

Figure 7.8. Illustrative diagram for the effect of total cation concentration and species on PPCP sorption. If the PPCP sorption mechanism is dominated by metal bridging, an increased PPCP sorption may be expected with increased total cation concentration. However, if PPCP sorption is controlled by cation exchange, increased total cation concentration may result in decreased PPCP sorption. The cations may exist as free dissolved, sorbent-sorbed, and PPCP-complexed ions. Knowing the species of cations could greatly improve our understanding on the effect of ionic strength on PPCP sorption.

soil/sediment complexation and systematic investigations are warranted. 7.5.3. Different Types of Cations The adsorption of carbadox (CBX) and bisdesoxycarbadox (DCBX) showed significant cation-type-dependent behavior. Sorption from KCl solution was increased by 46%–119% for CBX and 11%–77% for DCBX more than from CaCl2 solution (Strock et al. 2005). The pKa values of acidic nitrogen atoms of CBX and DCBX are <1. Thus, in a normal pH range of soils, these chemicals exist as neutral compounds and sorption by ion exchange mechanisms are negligible. The CBX or DCBX adsorption mechanisms on soils are dominated by electron donor–acceptor interactions and

R

RS

inner-sphere complexation. The larger hydration shell of Ca2þ hinders the adsorption sites related to both mechanisms. Therefore, the size of the hydration shell is one of the most important metal ion properties affecting PPCP adsorption. Weissmahr et al. (1998) also indicated that the accessibility of siloxane hydrophobic sites is controlled by the steric effects of hydrated exchangeable cations, due to their different hydration shells. Chen et al. (2007) reported that in the presence of different cations, different trends of HOC sorption on a wood charcoal were observed. The presence of Cu2þ suppressed HOC sorption while Ag þ increased HOC sorption. The “hard–soft cation” concept is adopted to explain the sorption data—the smaller the ratio of charge to cation radius, the softer the cation. The existence of a water shell around hard cations (e.g., Ca2þ , Cu2þ , Mg2þ ) could intrude into the adjacent charcoal surface, occupying the adsorption sites of HOCs. The complexation of soft cations (such as K þ , Cs þ , and Ag þ ) with the sorbent could make the functional groups less soluble and decrease the hydration shells, and thus facilitate organic chemical adsorption. This concept is illustrated in Figure 7.9. 7.5.4. Dissolved Organic Matter (DOM) As estimated by Zhou et al. (2007), colloid-associated pharmaceuticals accounted for 10%–29% in the environment, and more than 70% in the aqueous phase. Interaction between PPCP and DOM is a significant factor in PPCP environmental fate prediction. Ideal partitioning-like DOM binding has been widely accepted for nonpolar organic chemicals. However, more recent evidence has been presented for nonideal HOC–DOM interactions. Site-specific interactions (Laor and Rebhun 2002) and relatively discrete hydrophobic regions (Pan et al. 2007) were proposed for nonideal HOC–DOM interaction. Using a cross-flow ultrafiltration system, Maskaoui et al. (2007) investigated the interactions between pharmaceuticals and aquatic colloids (mostly DOM) and observed a linear relationship between log KCOC (the partitioning coefficient normalized to colloidal organic carbon

RH

R Hydrophilic functional groups

Adsorbent surface

S Soft cation

Hydration shell

H Hard cation

Available sorption surface

Figure 7.9. Illustraton of the hydration shell size and relative availability of adsorption sites. The existence of hydrated shells around hard-metal cations could decrease the availability of adjacent sorption sites, while soft metals would decrease the hydration shell around the functional groups and increase PPCP sorption.

SORPTION AS AFFECTED BY SOLUTION CHEMISTRY

content) and log KOW. They thus emphasized that hydrophobicity of the chemicals is the controlling factor for PPCP– DOM interactions. Their data, however, seems to be too scattered to extensively discuss nonlinearity of the sorption isotherms. Nonlinear sorption and sorption/desorption hysteresis have been observed in TC-DOM (Gu et al. 2007) and clarithromycin–DOM interactions (Sibley and Pedersen 2008) using a dialysis system. The nonlinear factor n was 0.85  0.01 for clarithromycin–DOM, and the thermodynamic index of irreversibility (TII) was calculated to be 0.22  0.06 (TII ¼ 0 indicates completely reversible sorption) (Sibley and Pedersen 2008). The nonideal behavior suggests site-specific interactions between PPCPs and DOM, which could not be explained by hydrophobic interactions alone. In addition to equilibrium dialysis systems and cross-flow ultrafiltration, various methods have been applied in studying interactions between organic chemicals and DOM, such as fluorescence and solubility enhancement (Yamamoto et al. 2003). All these methods could be applied to investigate PPCP–DOM interaction mechanisms. Pan et al. (2009) has systematically discussed all these methods in revealing HOC–DOM interaction mechanisms. It seems that dialysis equilibrium systems are better choices for understanding the actual organic chemical–DOM interaction mechanisms. The improved data quality and reliability will enable us to more clearly discuss PPCP–DOM interaction mechanisms. Because of the strong interaction between PPCPs and DOM, PPCP solubility could be increased in the presence of DOM, and thus PPCP mobility is also enhanced. On the other hand, the interaction between PPCP and particle-sorbed

203

DOM could promote overall sorption (Fig. 7.10). In this case, PPCP mobility is decreased. Pan et al. (2009) discussed the apparent HOC sorption in the presence of DOM. The apparent sorption coefficient of HOCs on solid particles could be increased, decreased, or further decreased to a value below sorption in the absence of DOM depending on DOM concentrations. This discussion also applies in PPCP–DOM interactions. Kulshrestha et al. (2004) investigated OTC sorption on clay in the presence of DOM. The experiment was conducted at pH 11.0 to exclude cation exchange mechanisms, and the discussion focused on hydrophobic interactions. At lower DOM concentration (1 mg/L), increased adsorption was observed, while at a higher DOM concentration (10 mg/L), an opposite effect was observed. The authors discussed only DOM nonlinear sorption. However, other processes, such as DOM fractionation and reconformation after sorption on the solid particles (Pan et al. 2009), may also greatly control PPCP–DOM interactions. This line of research is still not available for PPCP– DOM interaction studies. Dissolved organic matter could also be operationally separated into hydrophobic and hydrophilic fractions. Limited studies with these DOM fractions showed that hydrophobic fractions had higher sorption of HOCs (Ilani et al. 2005; Chefetz et al. 2006; Polubesova et al. 2007). Thus hydrophobicity is the controlling factor for HOC–DOM interactions. However, because of the presence of functional groups, various species, and metabolites, various mechanisms need to be examined to understand PPCP–DOM interaction, such as covalent binding and amorphous inclusion complex (Agarwal et al. 2008).

Figure 7.10. Schematic illustration for the effect of DOM concentration on PPCP sorption on soils and sediments. At low DOM concentrations, DOM is mostly adsorbed on solid particles and thus PPCP sorption could be increased. As DOM concentration increases, their sorption on particles reaches saturation and more DOM molecules are in the aqueous phase. Sorption of PPCP on particles could be decreased because of the association of PPCPs with DOM molecules in aqueous phase or the competition between PPCPs and DOM on particle surface.

204

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

7.5.5. Other Coadsorbates Pharmaceuticals and personal care products could be easily transformed to other compounds (Hildebrand et al. 2006). The presence of metabolites may compete with parent compounds for sorption sites (Strock et al. 2005; Casey et al. 2004) and hence increase the mobility of parent compounds. Therefore, the potential environmental risk of those metabolites should be taken into consideration for PPCP fate modeling and risk assessment. The competition between PPCPs and their metabolites and the behavior of the inactive metabolites should also be incorporated in future environmental fate studies. For some PPCPs that are moderately hydrophobic compounds, hydrophobic interactions are an important mechanism of PPCP sorption in soils/sediments, such as on humic substances and siloxane surface. Therefore, the presence of HOCs may also compete with PPCPs. For example, the coexistence of 17a-ethinylestradiol (EE2) and phenanthrene (PHE) or naphthalene (NAPH) exhibited significant competition (Yu and Huang 2005). These authors observed increased sorption linearity and reduced single-point sorption coefficients for NAPH with increased EE2 concentrations; PHE sorption was also affected by EE2 in the same manner, but to a much lesser extent, which is consistent with the higher hydrophobicity of PHE. For the same reason, EE2 sorption was also depressed as much as 35% by PHE, but insignificantly by NAPH. These data indicate obvious competition between PPCPs and HOCs. Still, our understanding of the extent and quantification of this competition is limited. Moreover, because of the charged properties of PPCPs, the presence of charged chemicals could significantly alter PPCP sorption. The effects of cations were discussed previously. Anions may also compete with negatively charged PPCP species, and decreased PPCP apparent adsorption would be expected. On the other hand, if the adsorbent is positively charged, anions may act like a bridging agent between positively charged PPCPs and the adsorbent, and thus PPCP adsorption could be increased. Because of the wide presence of anions in the environment, their interactions with PPCPs could be of significance. However, research along this line is not available yet.

7.6. PPCP DEGRADATION Pharmaceuticals and personal care products can quickly degrade in the environment, with half-lives generally within one month (Cousins et al. 2002; Lucas and Jones 2006; Accinelli et al. 2007), especially in an aerobic environments (Ying et al. 2004; Ying and Kookana 2005). Therefore, PPCPs have been reported to be nonpersistent in the environment (Klecka et al. 2001; Lucas and Jones 2006). It should be mentioned that because PPCPs are continuously entering

the environment, their concentrations are maintained in a certain range. In this case, PPCP behavior could be viewed as the same as that of persistent organic pollutants. Degradation of PPCPs is driven by many processes, including photodegradation, biodegradation, and chemical degradation. Photodegradation was reported to be a major pathway for PPCP degradation in the aqueous phase, but once adsorbed by soils/sediments, the latter two are more important. For biodegradation, aeration is the controlling factor. Kunkel and Radke (2008) observed shorter half-lives of various PPCPs with increased exchange of surface and pore water, indicating the importance of oxygen content in PPCP degradation. Interestingly, the metabolites are different when PPCP degradation occurs at different aeration conditions. This phenomenon could be adopted to understand PPCP degradation pathways. For example, aerobic degradation of ibuprofen produces both carboxyibuprofen and hydroxylibuprofen, and anaerobic degradation produces carboxyibuprofen only. Thus the presence of hydroxylibuprofen indicates that aerobic pathways were predominant in ibuprofen removal (Matamoros et al. 2008). The interaction between PPCPs and soils/sediments could either promote or decrease PPCP degradation. Soil/sediment organic matter could form complexes with PPCPs through Michael addition with PPCP active functional groups and thus deactivate PPCPs (Bialk et al. 2007; Bialk and Pedersen 2008). In aqueous phase, DOM may act like a photosensitizer and promote PPCP photodegradation (Andreozzi et al. 2003; Lam and Mabury 2005; Werner et al. 2005; Kwon and Armbrust 2005). Inorganic fractions of soils/sediments could also promote PPCP degradation. For example, in a OTC-MnO2 system, OTC was degraded rapidly and the degradation was accompanied by generation of Mn2þ ions (Rubert and Pedersen 2006). This observation indicated that OTC was oxidized. Because MnO2 is widely present in soils/ sediments, their roles in PPCP degradation should not be overlooked. On the other hand, PPCPs may be protected from photodegradation after adsorption on soils/sediments (Elmund et al. 1971; Gonsalves and Tucker 1977; Belden et al. 2007). Therefore, the dual roles of soils/sediments on PPCP degradation should be borne in mind in understanding PPCP environmental behavior. As indicated in the schematic drawing (Fig. 7.11), biodegradation generally decreases because of PPCP sorption to DOM or soil/sediment particles. However, photodegradation and chemical degradation could be either enhanced or depressed after sorption. Systematic studies are needed to examine the direction and extent of different degradation pathways. Generally speaking, the half-lives of PPCPs in soils/ sediments are relatively longer than those in the aqueous phase (Samuelsen et al. 1992), but PPCPs may still have environmental risks associated with their presence. For PPCPs protected from degradation by adsorbing to soils/ sediments, they may release back into the environment under

PPCP LEACHING

205

Aqueous phase Photodegradation

Soil/sediment

decreasing Biodegradation decreasing

PPCP molecule

Degraded PPCP

Chemical degradation DOM

Photosensitizer

Reaction with SOM or minerals Protected from degradation

Adsorbed on soil/sediment

Figure 7.11. Different pathways of PPCP degradation. Photodegrdation is restricted to the area accessible by light, biodegradation occurs mainly in aerobic conditions, and chemical degradation may occur in all the locations. DOM could act like a photosensitizer and promote PPCP photodegradation. On the other hand, DOM may also protect PPCP from photodegradation. Chemical reaction of PPCPs with SOM or mineral components is expected, whereas the sorption of PPCPs on particles or complexation with DOM may protect them from biodegradation.

the change of water chemistry conditions. For example, the adsorption of TC on clays is controlled by surface complexation and cation exchange (Figueroa et al. 2004). The adsorption of TC on clays is minimal at pH > 9 because of electrostatic repulsion between TC and the clay surface, meaning that the adsorbed TC may release when pH values are higher than 9. The extent of release of adsorbed PPCP is still unknown and further study is required. Study in this area will provide valuable information for environmental risk assessment of PPCPs in soils/sediments. In addition, soilbound PPCPs may still retain their activity. For example, Chander et al. (2005) reported that even though TC and tylosin were tightly bound with clay particles, they show antimicrobial effects that result in the development of antibiotic resistance in the terrestrial environment. During batch sorption experiments, calculation of the chemical concentration in the solid phase is often based on mass balance (Strock et al. 2005). The fast degradation of PPCPs limits the use of traditional mass balance calculation methods. In addition, the contact of PPCPs with mineral particles (such as MnO2) may promote PPCP degradation (Rubert and Pedersen 2006). Therefore, an effective way to establish a reliable sorption isotherm is to determine the concentration in the aqueous phase as well as in the solid phase (Stein et al. 2008). However, because the extraction of PPCPs from solid phases always have low recovery (Figueroa et al. 2004), or the extraction procedure includes substantial errors, the isotherms established via determination of both

aqueous- and solid-phase concentrations could show significant data variation. Thus it would be difficult to discuss adsorption mechanisms on the basis of data with huge variation. In order to properly discuss PPCP sorption in soils/ sediments, the following issue needs to be addressed: how to control/calibrate the recovery in PPCP sorption experiments and thus properly derive solid-phase concentration.

7.7. PPCP LEACHING Examination of groundwater and leachate from manureapplied fields showed no or small amounts of PPCPs (Hirsch et al. 1999, Kay et al. 2005, Lindberg et al. 2007), because of the strong interaction between PPCPs and soils/sediments as well as rapid degradation (Table 7.7). The leaching behavior of PPCPs through soil columns is greatly dependent on PPCP chemical properties and their sorption characteristics in soils. For example, the leaching potential of diazepam, ibuprofen, ivermectin, and carbamazepine could be rated as low (Oppel et al. 2004). Therefore, these compounds are unlikely to contaminate groundwater if the water table is sufficiently deep. Clofibric acid and iopromide showed higher mobility and their mobility is consistent with their environmental occurrences (Oppel et al. 2004). The preferential leaching of clofibric acid was also observed by Scheytt et al. (2007). Clofibric acid showed no transformation and no retardation as indicated by breakthrough curves from soil column

206

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

TABLE 7.7. Leaching Behavior of PPCPs Experimental Description

Source

Leaching

Reference

Sewage treatment plant, field investigation

Dehydrated erythromycin, roxithromycin, clarithromycin, sulfamethoxazole, trimethoprim Sulfonamide, tetracycline, and macrolide groups Six selected pharmaceuticals including clofibric acid, iopromide, diazepam, ibuprofen, ivermectin, carbamazepine Pharmaceutically active compounds, including primidone, propyphenazone, clofibric acid, diclofenac, ibuprofen

Detected in groundwater—sulfamethoxazole, sulfamethazine

Hirsch et al. (1999)

Detected in leachate— sulfachloropyridazine Mobile—clofibric acid and iopromide; Low mobility—diazepam, ibuprofen, ivermectin, carbamazepine Highly mobile and persistent— primidone, propyphenazone, clofibric acid; transformed, low risk — diclofenac and ibuprofen Highly mobile compounds— metronidazole, olaquindox; weakly mobile—oxytetracycline, tylosin

Kay et al. (2005)

Clay soil, lysimeter experiment Leaching behavior in different soils (e.g., different grain size distribution, pH, organic carbon) Sediment column experiments; leaching from sewage irrigation farms

Three sandy loam soils and a loamy sand soil

Metronidazole, olaquindox, oxytetracycline, tylosin

experiments, whereas transformation of diclofenac was so high (79%) that no retardation factor could be calculated. Rabolle and Spliid (2000) also reported different leaching behaviors of various PPCPs. Olaquindox was leached through the columns, while tylosin was retained at different depths. No leaching was observed for OTC because of their strong interaction with soil particles. The leaching behavior of PPCP is also reviewed by Kemper (2008). Therefore, according to PPCP leaching studies using different methods (field investigation, lysimeter experiments, and column study), leaching through soil horizons does not seem to significantly pollute groundwater. However, rain events shortly after manure application may cause PPCP transport through surface runoff. Transport of PPCP through this route to water resources could be much higher than transport through other routes. Previous studies have indicated that the presence of DOM or surfactants in soils may enhance the leaching of HOCs (Petruzzelli et al. 2002; Makkar and Rockne 2003). Strong interactions between DOM and PPCPs are also expected. However, how PPCP leaching behavior is affected by DOM or surfactants is still unknown.

7.8. QUANTITATIVE DESCRIPTION OF PPCP ENVIRONMENTAL FATE Owing to the presence of different PPCP species and the multiple interaction mechanisms associated with PPCPs, it will be necessary to consider PPCP models different from those used to model HOC sorption. Several typical models describing PPCP sorption in soils/sediments are summarized in Table 7.8. Because different species of certain PPCPs have different sorption coefficients, the overall sorption coefficient

Oppel et al. (2004)

Scheytt et al. (2007)

Rabolle and Spliid (2000)

could be viewed as contributions of the different species. This idea could be conceptualized by attributing the total sorption to the sum of the species-specific sorption coefficient weighted with the fraction of the respective species present in the aqueous phase. This model was successfully applied to describe the variation of OTC sorption coefficients with pH (ter Laak et al. 2006b). Clearly, this proposed model does not provide descriptions of different mechanisms, and the contributions of different mechanisms were viewed as blackboxes. Thus, this model is inadequate for discussing PPCP sorption behavior, which is controlled by various adsorption mechanisms. A complex conceptual model considering the contribution of individual mechanisms is also presented in Table 7.8 (Tolls 2001). The specific processes considered in soil/sediment–PPCP interactions include sorption to organic matter, surface adsorption to mineral fractions, ion exchange, and other reactions (such as complexation and H bonding). For chemicals with more than one polar functional group, the role of individual functional groups in PPCP sorption needs to be understood. Carrasquillo et al. (2008) compared the adsorption of a series of chemicals with similar structure units. These authors proposed that a structure-based descriptor could be employed to predict PPCP adsorption. This idea indicates that methods similar to the fragment constant approach to predict KOC are possible. However, the current data are still too limited, and thus this method is not applicable yet. The different sorption mechanisms of a complex compound with multiple functional groups may be differentiated through observation of sorption of simpler compounds. MacKay and Seremet (2008) used two probe compounds to estimate the adsorption of ciprofloxacin to soil. The probe flumequine characterizes the sorption mechanism of surface complexation, while phenylpiperazine

QUANTITATIVE DESCRIPTION OF PPCP ENVIRONMENTAL FATE

207

TABLE 7.8. Different Concepts in PPCP Sorption Modeling Characteristic Species-specific sorption

Contribution of different mechanisms

Model Kd1 Kd2 þ 1 þ 10pH
pKa 1 þ 10pKa
pH where Kd0 ¼ apparent sorption coefficient; a ¼ fraction of a specific species; Kd1, Kd2 ¼ sorption coefficient of different species 1 Kd ¼ ðCOM fOM þ Cmin Amin þ Cie sie Aie þ Crxn srxn Arxn Þ Caq neutral þ SCaq ionic Kd0 ¼ Kd1  a þ Kd2  ð1
aÞYKd0 ¼

Reference ter Laak et al. (2006b)

Tolls (2001)

Where A ¼ specific surface area available for adsorption or reaction; C ¼ concentration of chemical interacting via respective mode; s ¼ surface concentration of an interaction site on a given adsorbent; fOM ¼ fraction of organic matter on adsorbent; subscript OM ¼ sorption to organic matter; subscript min ¼ surface adsorption to mineral constituents; subscript ie ¼ ion exchange; subscript rxn ¼ reactions (e.g., complexation and H bonding, with subscript rxn) Individual mechanisms as probed using probe compounds

pH-dependent KOW

CSIE ¼

ðCEC þ Vvic ½co-ionKxs ÞKxs a  Cw ½counterion þ Kxs aCw

CSSC ¼

ssurf AKxs Krxn a  Cw ½competing ion þ Kxs Krxn aCw

where CSIE ¼ concentrations of compound sorbed by cation exchange; CSSC ¼ concentrations of compound sorbed by surface complexation; CEC ¼ cation exchange capacity of sorbent; Vvic ¼ the volume of vicinal water coating sorbent surface; [co-ion] ¼ concentration of ions with charge opposite that of sorbate ion; [counterion] ¼ concentration of competing ions; a ¼ fraction of total dissolved compound present as the sorbing species ¼[1/(1 þ 10
pH þ pKa)]; Cw ¼ aqueous concentration; ssurf ¼ concentration of surface sites; A ¼ specific surface area of sorbent; [competing ion] ¼ concentration of competing ions; Krxn ¼ electrostatic interaction of particular ligand group of an organic molecule; Kxs ¼ empirical exchange parameter, which can be estimated as Kxs ¼ expð
DGhydrophobic =RTÞ ¼ 2Csat ðLÞ
0:2 Neutral molecules—logDOW ¼ logKOW ; acidic molecules (OH, COOH)—   1 logDOW ¼ logKOW þ log ; ð1 þ 10pH-pKa Þ

MacKay and Seremet (2008)

Carballa et al. (2008)

basic molecules (RNH2, R2NH, R3N) ¼   1 logDOW ¼ logKOW þ log , where DOW ¼ pH-dependent ð1 þ 10pKa
pH Þ octanol–water distribution coefficient; KOW ¼octanol–water distribution coefficient obtained from literature

characterizes the mechanisms of cation exchange. The proposed models capture the mechanism of surface complexation and cation exchange, and the overall sorption is the sum of these two mechanisms. Hydrophobic properties of certain compounds, which are described using KOW, are key factors in modeling PPCP environmental behavior. Many studies indicated a failure of linear free-energy modeling to relate KOW and KOC. However, it should be noted that the application of KOW needs to be improved. Because of the effect of pH, KOW of PPCPs may not be properly applied in previous studies relating KOW and KOC. A given PPCP may have different KOWvalues at different pH values because of different species. In environ-

mentally related pH ranges, PPCPs may not exist as a single species. Therefore, KOW needs to be corrected with respect to species composition in sorption systems. Carballa et al. (2008) calibrated KOW on the basis of pH and pKa. This method could remarkably improve the accuracy of estimating the contribution of hydrophobic interactions. Fugacity models are widely applied to investigate the environmental behavior of organic chemicals. This model describes the distribution of organic chemicals in different environmental media on the basis of the parameter, fugacity, which is an expression of the tendency of a compound to escape from a compartment (MacKay and Paterson 1991). Fugacity models have shown great potential in environmen-

208

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

tal fate modeling (Tao et al. 2003). This model is also adopted to calculate overall persistence, concentrations, and intermedia fluxes of PPCPs between air, water, soil, and sediments at steady-state (Zukowska et al. 2006). However, current application of this model still employs the concept of LFERs. As discussed earlier in this chapter, LFERs are unable to adequately predict PPCP environmental behavior. Further modeling investigation is needed to incorporate sorption models combining different mechanisms into fugacity models.

7.9. SUMMARY AND PERSPECTIVES The widespread application and occurrence of PPCPs in the environment has raised concerns of PPCP environmental and health risks. The data available on the occurrence of PPCPs in soils/sediments are very limited because of the analytical challenges for solid matrix samples. An extensive survey of PPCP distribution in solid environmental media is needed. The environmental behavior of PPCPs is controlled mostly by their interactions with soils and sediments. Further, PPCP sorption in soils/sediments cannot be described simply using linear free-energy relationships, and various mechanisms (including hydrophobic interaction, ion exchange, H bonding, surface complex, cation bridging, electron–donor– acceptor systems, inner/outer-sphere complexation, and electrostatic interactions) need to be considered. Because PPCPs can be tightly bound by soils/sediments, their activities are reduced, but may not completely vanish. The antimicrobial effect of adsorbed PPCPs as well as the release of bound PPCPs should be incorporated in PPCP risk assessment. Unlike hydrophobic organic contaminants, PPCPs bind strongly with mineral particles. The adsorption coefficient of PPCPs on inorganic particles is comparable to those on whole soils/sediments. Also, PPCP molecules could diffuse into the layer structures of minerals and expand the interlayer space. Nonexpandable mineral particles could also bind with PPCPs through their edge and surface functional groups and hydrophobic surfaces. Thus, structural features of soil minerals should be factored in to understand PPCP sorption mechanisms in soils/sediments. Therefore, to predict PPCP–soil/ sediment interactions, detailed characterization of soil/ sediment components as well as organic–inorganic conformational organization is needed. Soil/sediment–PPCP interactions change remarkably with water chemistry conditions. At different pHs, PPCPs exist as different species, and thus soil/sediment–PPCP interaction can result from different mechanisms. For the same reason, KOW values for PPCPs vary at different pH levels. This concept should be widely employed in PPCP sorption studies. Mechanisms for the effects of both cation concentration and type on PPCP sorption were discussed, but examination of the effect of anions on PPCP sorption is

lacking. Ionic strength is usually expressed as the total concentration of the ions. However, in soil/sediment–PPCP interaction systems, cations can exist as different species. The influence of ionic strength cannot be explicitly observed if cationic species are not examined. Therefore, techniques distinguishing cation species should be included in studies on the effect of ionic strength. Nonideal interaction between PPCPs and DOM has been reported in limited studies, and PPCP–DOM interaction mechanisms are unclear. Extended research is required to systematically investigate PPCP– DOM interactions while accounting for the shortage of different experimental procedures and designs. Photodegradation, biodegradation, and chemical degradation are all involved in PPCP degradation. The interaction between PPCPs and soils/sediments could either promote or decrease PPCP degradation. Thus, different influences of soils/sediments on PPCP degradation should be taken into account in understanding PPCP environmental behavior. The leaching properties of PPCPs vary depending on PPCP chemical properties and their sorption characteristics in soils. Generally speaking, PPCP leaching does not seem to be a major contamination source for groundwater. However, how the presence of DOM or surfactants affects PPCP leaching through soil column has not been studied. Modeling methods for soil/sediment–PPCP interaction have been proposed with an emphasis on the difference from hydrophobic organic contaminant sorption. The contributions of different species and mechanisms have been incorporated in modeling PPCP sorption in soils/sediments. However, this new modeling concept has not been adopted in PPCP multimedia environmental fate modeling and their risk assessment. In addition, because PPCP sorption is greatly dependent on soil/sediment properties, detailed characterization of soil/sediment properties and how these properties are related to PPCP sorption must be incorporated into future PPCP fate modeling and risk assessment.

ACKNOWLEDGMENTS The authors gratefully acknowledge financial support from the Massachusetts Water Resource Center (2007MA73B, 2009MA177B) and National Scientific Foundation of China (40973081, 40803034).

REFERENCES Accinelli, C., Koskinen, W. C., Becker, J. M., and Sadowsky, M. J. (2007), Environmental fate of two sulfonamide antimicrobial agents in soil, J. Agric. Food Chem. 55, 2677–2682. Agarwal, S. P., Anwer, M. K., and Aqil, M. (2008), Complexation of furosemide with fulvic acid extracted from shilajit: A novel approach, Drug Dev. Ind. Pharm. 34, 506–511.

REFERENCES

AHI (2002), Animal Health Institute; available from< http://www. ahi.org > Akyuz, S. and Akyuz, T. (2003), FT-IR spectroscopic investigation of adsorption of pyrimidine on sepiolite and montmorillonite from Anatolia, J. Inclusion. Phenom. Macrocycl. Chem. 46, 51–55. Al-Rajab, A. J., Sabourin, L., Scott, A., Lapen, D. R., and Topp, E. (2009), Impact of biosolids on the persistence and dissipation pathways of triclosan and triclocarban in an agricultural soil, Sci. Total Environ. 407, 5978–5985. Alpuche, C., Garau, J., and Lim, V. (2006), Global and local variations in antimicrobial susceptibilities and resistance development in the major respiratory pathogens, Proc. Symp. Augmentin Heritage in Evolving Antibacterial Therapy, London, pp. S135–S138. Andreozzi, R., Raffaele, M., and Nicklas, P. (2003), Pharmaceuticals in STP effluents and their solar photodegradation in aquatic environment, Chemosphere 50, 1319–1330. Bai, Y., Wu, F., Liu, C., Guo, J., Fu, P., Li, W., and Xing, B. (2008), Interaction between carbamazepine and humic substances: A fluorescence spectroscopy study, Environ. Toxicol. Chem. 27, 95–102. Barron, L., Havel, J., Purcell, M., Szpak, M., Kellehera, B., and Paull, B. (2009), Predicting sorption of pharmaceuticals and personal care products onto soil and digested sludge using artificial neural networks, Analyst 134, 663–670. Belden, J. B., Maul, J. D., and Lydy, M. J. (2007), Partitioning and photo degradation of ciprofloxacin in aqueous systems in the presence of organic matter, Chemosphere 66, 1390–1395. Bialk, H. M., Simpson, A. J., and Pedersen, J. A. (2005), Crosscoupling of sulfonamide antimicrobial agents with model humic constituents, Environ. Sci. Technol. 39, 4463–4473. Bialk, H. M., Hedman, C., Castillo, A., and Pedersen, J. A. (2007), Laccase-mediated Michael addition of N-15-sulfapyridine to a model humic constituent, Environ. Sci. Technol. 41, 3593–3600. Bialk, H. M. and Pedersen, J. A. (2008), NMR investigation of enzymatic coupling of sulfonamide antimicrobials with humic substances, Environ. Sci. Technol. 42, 106–112. Bonin, J. L. and Simpson, M. J. (2007), Sorption of steroid estrogens to soil and soil constituents in single- and multi-sorbate systems, Environ. Toxicol. Chem. 26, 2604–2610. Bowman, J. C., Zhou, J. L., and Readman, J. W. (2002), Sedimentwater interactions of natural oestrogens under estuarine conditions, Mar. Chem. 77, 263–276. Boxall, A. B. A., Fogg, L. A., Baird, D. J., Lewis, C., Telfer, T. C., Kolpin, D., and Gravell, A. (2005), Targeted Monitoring Study for Veterinary Medicines in the UK Environment, Final Report to the UK Environmental Agency Boyd, S. A., Sheng, G. Y., Teppen, B. J., and Johnston, C. J. (2001), Mechanisms for the adsorption of substituted nitrobenzenes by smectite clays, Environ. Sci. Technol. 35, 4227–4234. Calisto, V. and Esteves, V. I. (2009), Psychiatric pharmaceuticals in the environment, Chemosphere 77, 1257–1274. Campbell, C. G., Borglin, S. E., Green, F. B., Grayson, A., Wozei, E., and Stringfellow, W. T. (2006), Biologically directed environmental monitoring, fate, and transport of estrogenic

209

endocrine disrupting compounds in water: A review, Chemosphere 65, 1265–1280. Carballa, M., Fink, G., Omil, F., Lema, J. M., and Ternes, T. (2008), Determination of the solid-water distribution coefficient (K-d) for pharmaceuticals, estrogens and musk fragrances in digested sludge, Water Res. 42, 287–295. Carrasquillo, A. J., Bruland, G. L., Mackay, A. A., and Vasudevan, D. (2008), Sorption of ciprofloxacin and oxytetracycline zwitterions to soils and soil minerals: Influence of compound structure. Environ. Sci. Technol. 42, 7634–7642. Casey, F. X. M, Larsen, G. L., Hakk, H., and Simunek, J. (2003), Fate and transport of 17 beta-estradiol in soil-water systems, Environ. Sci. Technol. 37, 2400–2409. Casey, F. X. M., Hakk, H., Simunek, J., and Larsen, G. L. (2004), Fate and transport of testosterone in agricultural soils, Environ. Sci. Technol. 38, 790–798. Chefetz, B., Illani, T., Schulz, E., Chorover, J. (2006), Wastewater dissolved organic matter: characteristics and sorptive capabilities, Water Sci. Technol. 53, 51–57. Chander, Y., Kumar, K., Goyal, S. M., and Gupta, S. C. (2005), Antibacterial activity of soil-bound antibiotics, J. Environ. Qual. 34, 1952–1957. Chee-Sanford, J. C., Mackie, R. I., Koike, S., Krapac, I. J., Lin, Y. F., Yannarell, A. C., Maxwell, S., and Aminov, R. I. (2009), Fate and transport of antibiotic residues and antibiotic resistance genes following land application of manure waste, J. Environ. Qual. 38, 1086–1108. Chen, J. Y., Zhu, D. Q., and Sun, C. (2007), Effect of heavy metals on the sorption of hydrophobic organic compounds to wood charcoal, Environ. Sci. Technol. 41, 2536–2541. Chiou, C. T., Peters, L. J., and Freed, V. H. (1979), Physical concept of soil-water equilibria from non-ionic organic-compounds, Science 206, 831–832. Colucci, M. S., Bork, H., and Topp, E. (2001), Persistence of estrogenic hormones in agricultural soils: I. 17 beta-estradiol and estrone, J. Environ. Qual. 30, 2070–2076. Colucci, M. S. and Topp, E. (2002), Dissipation of part-per-trillion concentrations of estrogenic hormones from agricultural soils, Can. J. Soil Sci. 82, 335–340. Cousins, I. T., Staples, C. A., Klecka, G. M., and MacKay, D. (2002), A multimedia assessment of the environmental fate of bisphenol A, Hum. Ecol. Risk Assess. 8, 1107–1135. Daughton, C. G. and Ternes, T. A. (1999), Pharmaceuticals and personal care products in the environment: Agents of subtle change? Environ. Health Perspect. 107, 907–938. During, R. A., Krahe, S., and Gath, S. (2002), Sorption behavior of nonylphenol in terrestrial soils, Environ. Sci. Technol. 36, 4052– 4057. Elmund, G. K., Morrison, S. M., Grant, D. W., and Nevins, M. P. (1971), Role of excreted chlortetracycline in modifying decomposition process in feedlot waste, Bull. Environ. Contam. Toxicol. 6, 129–132. Ericson, J. F. (2007), An evaluation of the OECD 308 water/ sediment systems for investigating the biodegradation of pharmaceuticals, Environ. Sci. Technol. 41, 5803–5811.

210

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

Eschenbach, A., Wienberg, R., and Mahro, B. (1998), Fate and stability of nonextractable residues of [C-14]PAH in contaminated soils under environmental stress conditions, Environ. Sci. Technol. 32, 2585–2590. Fan, Z. S., Casey, F. X. M., Larsen, G. L., and Hakk, H. (2006), Fate and transport of 1278-TCDD, 1378-TCDD, and 1478-TCDD in soil-water systems, Sci. Total Environ. 371, 323–333. Fan, Z. S., Casey, F. X. M., Hakk, H., and Larsen, G. L. (2007), Persistence and fate of 17 beta-estradiol and testosterone in agricultural soils, Chemosphere 67, 886–895. Figueroa, R. A., Leonard, A., and Mackay, A. A. (2004), Modeling tetracycline antibiotic sorption to clays, Environ. Sci. Technol. 38, 476–483. Forster, M., Laabs, V., Lamshft, M., Groeneweg, J., Zhlke, S., Spiteller, M., Krauss, M., Kaupenjohann, M., and Amelung, W. (2009), Sequestration of manure-applied sulfadiazine residues in soils, Environ. Sci. Technol. 43, 1824–1830. Gao, J. A. and Pedersen, J. A. (2005), Adsorption of sulfonamide antimicrobial agents to clay minerals, Environ. Sci. Technol. 39, 9509–9516. Gevao, B., Semple, K. T., and Jones, K. C. (2000), Bound pesticide residues in soils: A review, Environ. Pollut. 108, 3–14. Golet, E. M., Strehler, A., Alder, A. C., and Giger, W. (2002), Determination of fluoroquinolone antibacterial agents in sewage sludge and sludge-treated soil using accelerated solvent extraction followed by solid-phase extraction, Anal. Chem. 74, 5455–5462. Gonsalves, D. and Tucker, D. P. H. (1977), Behavior of oxytetracycline in Florida citrus and soils, Arch. Environ. Contam. Toxicol. 6, 515–523. Gu, C. and Karthikeyan, K. G. (2005a) , Interaction of tetracycline with aluminum and iron hydrous oxides, Environ. Sci. Technol. 39, 2660–2667. Gu, C. and Karthikeyan, K. G. (2005b) , Sorption of the antimicrobial ciprofloxacin to aluminum and iron hydrous oxides, Environ. Sci. Technol. 39, 9166–9173. Gu, C. Karthikeyan, K. G., Sibley, S. D., and Pedersen, J. A. (2007), Complexation of the antibiotic tetracycline with humic acid, Chemosphere 66, 1494–1501. Hamscher, G., Sczesny, S., Hoper, H., and Nau, H. (2002), Determination of persistent tetracycline residues in soil fertilized with liquid manure by high-performance liquid chromatography with electrospray ionization tandem mass spectrometry, Anal. Chem. 74, 1509–1518. Hamscher, G., Pawelzick, H. T., Hoper, H., and Nau, H. (2005), Different behavior of tetracyclines and sulfonamides in sandy soils after repeated fertilization with liquid manure, Environ. Toxicol. Chem. 24, 861–868. Health, I. (2004), IMS World Review, IMS, Fairfield, CT. Hildebrand, C., Londry, K. L., and Farenhorst, A. (2006), Sorption and desorption of three endocrine disrupters in soils, J. Environ. Sci. Health Pt. B—Pesticide Contam. Agric. Wastes 41, 907– 921. Hirsch, R., Ternes, T., Haberer, K., and Kratz, K. L. (1999), Occurrence of antibiotics in the aquatic environment, Sci. Total Environ. 225, 109–118.

Holbrook, R. D., Love, N. G., and Novak, J. T. (2004), Sorption of 17-beta-estradiol and 17 alpha-ethinylestradiol by colloidal organic carbon derived from biological wastewater treatment systems, Environ. Sci. Technol. 38, 3322–3329. Holthaus, K. I. E., Johnson, A. C., Jurgens, M. D., Williams, R. J., Smith, J. J. L., and Carter, J. E. (2002), The potential for estradiol and ethinylestradiol to sorb to suspended and bed sediments in some English rivers, Environ. Toxicol. Chem. 21, 2526–2535. Huschek, G., Hollmann, D., Kurowski, N., Kaupenjohann, M., and Vereecken, H. (2008), Re-evaluation of the conformational structure of sulfadiazine species using NMR and ab initio DFT studies and its implication on sorption and degradation, Chemosphere 72, 1448–1454. Ilani, T., Schulz, E., Chefetz, B. (2005), Interactions of organic compounds with wastewater dissolved organic matter: Role of hydrophohic fractions, J. Environ. Qual. 34, 552–562. Jacobsen, A. M., Halling-Sorensen, B., Ingerslev, F., and Hansen, S. H. (2004), Simultaneous extraction of tetracycline, macrolide and sulfonamide antibiotics from agricultural soils using pressurised liquid extraction, followed by solid-phase extraction and liquid chromatography-tandem mass spectrometry, J. Chromatogr. A 1038, 157–170. Jacobsen, A. M., Lorenzen, A., Chapman, R., and Topp, E. (2005), Persistence of testosterone and 17 beta-estradiol in soils receiving swine manure or municipal biosolids, J. Environ. Qual. 34, 861–871. Jaynes, W. F. and Boyd, S. A. (1991), Clay mineral type and organiccompound sorption by hexadecyltrimethlyammonium-exchanged clays, Soil Sci. Soc. Am. J. 55, 43–48. Jia, D. A., Zhou, D. M., Wang, Y. J., Zhu, H. W., and Chen, J. L. (2008), Adsorption and cosorption of Cu(II) and tetracycline on two soils with different characteristics, Geoderma 146, 224– 230. Jones, A. D., Bruland, G. L., Agrawal, S. G., and Vasudevan, D. (2005), Factors influencing the sorption of oxytetracycline to soils, Environ. Toxicol. Chem. 24, 761–770. Kahle, M. and Stamm, C. (2007), Sorption of the veterinary antimicrobial sulfathiazole to organic materials of different origin, Environ. Sci. Technol. 41, 132–138. Kang, S. H. and Xing, B. S. (2008), Humic acid fractionation upon sequential adsorption onto goethite, Langmuir 24, 2525–2531. Kay, P., Blackwell, P. A., and Boxall, A. B. A. (2005), A lysimeter experiment to investigate the leaching of veterinary antibiotics through a clay soil and comparison with field data, Environ. Pollut. 134, 333–341. Kazpard, V., Lartiges, B. S., Frochot, C., de la Caillerie, J. B. D., Viriot, M. L., Portal, J. M., Gorner, T., and Bersillon, J. L. (2006), Fate of coagulant species and conformational effects during the aggregation of a model of a humic substance with Al-13 polycations, Water Res. 40, 1965–1974. Kemper, N. (2008), Veterinary antibiotics in the aquatic and terrestrial environment, Ecol. Indic. 8, 1–13. Kesimli, B. and Topacli, A. (2001), Infrared studies on Co and Cd complexes of sulfamethoxazole, Spectrochim. Acta Pt. A—Molec. Biomolec. Spectrosc 57, 1031–1036.

REFERENCES

Kim, S. C. and Carlson, K. (2005), LC-MS2 for quantifying trace amounts of pharmaceutical compounds in soil and sediment matrices, Trends Anal. Chem. 24, 635–644. Klecka, G. M., Gonsior, S. J., West, R. J., Goodwin, P. A., and Markham, D. A. (2001), Biodegradation of bisphenol a in aquatic environments: River die-away, Environ. Toxicol. Chem. 20, 2725–2735. Kulshrestha, P., Giese, R. F., and Aga, D. S. (2004), Investigating the molecular interactions of oxytetracycline in clay and organic matter: Insights on factors affecting its mobility in soil, Environ. Sci. Technol. 38, 4097–4105. Kunkel, U. and Radke, M. (2008), Biodegradation of acidic pharmaceuticals in bed sediments: Insight from a laboratory experiment, Environ. Sci. Technol. 42, 7273–7279. Kwon, J. W. and Armbrust, K. L. (2005), Degradation of citalopram by simulated sunlight, Environ. Toxicol. Chem. 24, 1618–1623. Lai, K. M., Johnson, K. L., Scrimshaw, M. D., and Lester, J. N. (2000), Binding of waterborne steroid estrogens to solid phases in river and estuarine systems, Environ. Sci. Technol. 34, 3890–3894. Lalumera, G. M., Calamari, D., Galli, P., Castiglioni, S., Crosa, G., and Fanelli, R. (2004), Preliminary investigation on the environmental occurrence and effects of antibiotics used in aquaculture in Italy, Chemosphere 54, 661–668. Lam, M. W. and Mabury, S. A. (2005), Photodegradation of the pharmaceuticals atorvastatin, carbamazepine, levofloxacin, and sulfamethoxazole in natural waters, Aquatic Sci. 67, 177– 188. Laor, Y. and Rebhun, M. (2002), Evidence for nonlinear binding of PAHs to dissolved humic acids, Environ. Sci. Technol. 36, 955– 961. Lee, C. L., Kuo, L. J., Wang, H. L., and Hsieh, P. C. (2003), Effects of ionic strength on the binding of phenanthrene and pyrene to humic substances: Three-stage variation model, Water Res. 37, 4250–4258. Li, H. and Lee, L. S. (1999), Sorption and abiotic transformation of aniline and alpha-naphthylamine by surface soils, Environ. Sci. Technol. 33, 1864–1870. Li, J. H., Zhou, B. X., Shao, J. H., Yang, Q. F., Liu, Y. Q., and Cai, W. M. (2007), Influence of the presence of heavy metals and surface-active compounds on the sorption of bisphenol A to sediment, Chemosphere 68, 1298–1303. Lindberg, R. H., Bjorklund, K., Rendahl, P., Johansson, M. I., Tysklind, M., and Andersson, B. A. V. (2007), Environmental risk assessment of antibiotics in the Swedish environment with emphasis on sewage treatment plants, Water Res. 41, 613–619. Liu, R. X., Wilding, A., Hibberd, A., and Zhou, J. L. (2005), Partition of endocrine-disrupting chemicals between colloids and dissolved phase as determined by cross-flow ultrafiltration, Environ. Sci. Technol. 39, 2753–2761. Loffler, D., Rombke, J., Meller, M., and Ternes, T. A. (2005), Environmental fate of pharmaceuticals in water/sediment systems, Environ. Sci. Technol. 39, 5209–5218. Loffredo, E. and Senesi, N. (2006), Fate of anthropogenic organic pollutants in soils with emphasis on adsorption/desorption pro-

211

cesses of endocrine disruptor compounds, Pure Appl. Chem. 78, 947–961. Lucas, S. D. and Jones, D. L. (2006), Biodegradation of estrone and 17 beta-estradiol in grassland soils amended with animal wastes, Soil Biol. Biochem. 38, 2803–2815. MacKay, A. A. and Canterbury, B. (2005), Oxytetracycline sorption to organic matter by metal-bridging, J. Environ. Qual. 34, 1964– 1971. MacKay, A. A. and Seremet, D. E. (2008), Probe compounds to quantify cation exchange and complexation interactions of ciprofloxacin with soils, Environ. Sci. Technol. 42, 8270–8276. MacKay, D. and Paterson, S. (1991), Evaluating the multimedia fate of organic-chemicals. A level-III fugacity model, Environ. Sci. Technol. 25, 427–436. Makkar, R. S. and Rockne, K. J. (2003), Comparison of synthetic surfactants and biosurfactants in enhancing biodegradation of polycyclic aromatic hydrocarbons, Environ. Toxicol. Chem. 22, 2280–2292. Maskaoui, K., Hibberd., A., and Zhou, J. L. (2007), Assessment of the interaction between aquatic colloids and pharmaceuticals facilitated by cross-flow ultrafiltration, Environ. Sci. Technol. 41, 8038–8043. Matamoros, V., Caselles-Osorio, A., Garcia, J., and Bayona, J. M. (2008), Behaviour of pharmaceutical products and biodegradation intermediates in horizontal subsurface flow constructed wetland. A microcosm experiment, Sci. Total Environ. 394, 171–176. McClellan, K. and Halden, R. U. (2010), Pharmaceuticals and personal care products in archived US biosolids from the 2001 EPA national sewage sludge survey, Water Res. 44, 658–668. Monteiro, S. C. and Boxall, A. B. A. (2010), Occurrence and fate of human pharmaceuticals in the environment, Rev. Environ. Contam. Toxicol. 202, 53–154. Nikolaou, A., Meric, S., and Fatta, D. (2007), Occurrence patterns of pharmaceuticals in water and wastewater environments, Anal. Bioanal. Chem. 387, 1225–1234. Oppel, J., Broll, G., Loffler, D., Meller, M., Rombke, J., and Ternes, T. (2004), Leaching behaviour of pharmaceuticals in soil-testing-systems: A part of an environmental risk assessment for groundwater protection, Sci. Total Environ. 328, 265– 273. Ottmar, K. J., Colosi, L. M., and Smith, J. A. (2010), Sorption of statin pharmaceuticals to wastewater-treatment biosolids, terrestrial soils, and freshwater sediment. J. Environ. Eng. [Am. Soc. Civil Eng. (ASCE)] 136, 256–264. Pan, B., Ghosh, S., and Xing, B. S. (2007), Nonideal binding between dissolved humic acids and polyaromatic hydrocarbons, Environ. Sci. Technol. 41, 6472–6478. Pan, B., Ning, P., and Xing, B. S. (2009), Humic substances. Part V: Sorption of pharmaceuticals and personal care products, Environ. Sci. Pollut. Res. 16, 106–116. Patrolecco, L., Capri, S., De Angelis, S., Pagnotta, R., Polesello, S., and Valsecchi, S. (2006), Partition of nonylphenol and related compounds among different aquatic compartments in Tiber River (central Italy), Water Air Soil Pollut. 172, 151–166.

212

PHARMACEUTICALS AND PERSONAL CARE PRODUCTS IN SOILS AND SEDIMENTS

Petruzzelli, L., Celi, L., Cignetti, A., and Marsan, F. A. (2002), Influence of soil organic matter on the leaching of polycyclic aromatic hydrocarbons in soil, J. Environ. Sci. Health Pt. B— Pesticide Contam. Agric. Wastes 37, 187–199. Pignatello, J. J. and Xing, B. S. (1996), Mechanisms of slow sorption of organic chemicals to natural particles, Environ. Sci. Technol. 30, 1–11. Pils, J. R. V. and Laird, D. A. (2007), Sorption of tetracycline and chlortetracycline on K- and Ca-saturated soil clays, humic substances, and clay-humic complexes, Environ. Sci. Technol. 41, 1928–1933. Polubesova, T., Sherman-Nakache, M., and Chefetz, B. (2007), Binding of pyrene to hydrophobic fractions of dissolved organic matter: Effect of polyvalent metal complexation, Environ. Sci. Technol. 41, 5389–5394. Rabolle, M. and Spliid, N. H. (2000), Sorption and mobility of metronidazole, olaquindox, oxytetracycline and tylosin in soil, Chemosphere 40, 715–722. Roberts, T. R. (1984), IUPAC reports on pesticides. 17. Nonextractable pesticide-residues in soils and plants, Pure Appl. Chem. 56, 945–956. Rubert, K. F. and Pedersen, J. A. (2006), Kinetics of oxytetracycline reaction with a hydrous manganese oxide, Environ. Sci. Technol. 40, 7216–7221. Rubinfeld, S. A. and Luthy, R. G. (2008), Nitromusk compounds in San Francisco Bay sediments, Chemosphere 73, 873–879. Samuelsen, O. B., Torsvik, V., and Ervik, A. (1992), Long-range changes in oxytetracycline concentration and bacterial-resistance towards oxytetracycline in a fish farm sediment after medication, Sci. Total Environ. 114, 25–36. Sarmah, A. K., Meyer, M. T., and Boxall, A. B. A. (2006), A global perspective on the use, sales, exposure pathways, occurrence, fate and effects of veterinary antibiotics (VAs) in the environment, Chemosphere 65, 725–759. Schafer, A. I., Mastrup, M., and Jensen, R. L. (2002), Particle interactions and removal of trace contaminants from water and wastewaters, Desalination 147, 243–250. Scheytt, T. J., Mersmann, P., Rejman-Rasinski, E., and These, A. (2007), Tracing pharmaceuticals in the unsaturated zone, J. Soils Sediments 7, 75–84. Schlusener, M. P., Spiteller, M., and Bester, K. (2003), Determination of antibiotics from soil by pressurized liquid extraction and liquid chromatography-tandem mass spectrometry, J. Chromatogr. A 1003, 21–28. Seget, S. and Pharma, D. (2008), Pharmaceutical Market Trends, 2008–2012: Key Market Forecasts and Growth Opportunities, URCH Publishing. Seveno, N. A., Kallifidas, D., Smalla, K., van Elsas, J. D., Collard, J. M., Karagouni, A. D., and Wellington, E. M. H. (2002), Occurrence and reservoirs of antibiotic resistance genes in the environment, Rev. Med. Microbiol. 13, 15–27. Shareef, A., Angove, M. J., Wells, J. D., and Johnson, B. B. (2006), Sorption of bisphenol A, 17 alpha-ethynylestradiol and estrone to mineral surfaces, J. Colloid Interface Sci. 297, 62–69.

Sibley, S. D. and Pedersen, J. A. (2008), Interaction of the macrolide antimicrobial clarithromycin with dissolved humic acid, Environ. Sci. Technol. 42, 422–428. Sparks, D. L. (2003), Environmental Soil Chemistry, Elsevier, San Diego. Stein, K., Ramil, M., Fink, G., Sander, M., and Ternes, T. A. (2008), Analysis and sorption of psychoactive drugs onto sediment, Environ. Sci. Technol. 42, 6415–6423. Strock, T. J., Sassman, S. A., and Lee, L. S. (2005), Sorption and related properties of the swine antibiotic carbadox and associated N-oxide reduced metabolites, Environ. Sci. Technol. 39, 3134–3142. Stumpe, B. and Marschner, B. (2007), Long-term sewage sludge application and wastewater irrigation on the mineralization and sorption of 17 beta-estradiol and testosterone in soils, Sci. Total Environ. 374, 282–291. Sun, W. L., Ni, J. R., Xu, N., and Sun, L. Y. (2007), Fluorescence of sediment humic substance and its effect on the sorption of selected endocrine disruptors, Chemosphere 66, 700–707. Suntisukaseam, U., Weschayanwiwat, P., and Sabatini, D. A. (2007), Sorption of amphiphile pharmaceutical compounds onto polar and nonpolar adsorbents, Environ. Eng. Sci. 24, 1457– 1465. Tao, S., Cao, H. Y., Liu, W. X., Li, B. G., Cao, J., Xu, F. L., Wang, X. J., Coveney, R. M., Shen, W. R., Qin, B. P., and Sun, R. (2003), Fate modeling of phenanthrene with regional variation in Tianjin, China, Environ. Sci. Technol. 37, 2453–2459. ter Laak, T. L., Gebbink, W. A., and Tolls, J. (2006a) , Estimation of soil sorption coefficients of veterinary pharmaceuticals from soil properties, Environ. Toxicol. Chem. 25, 933–941. ter Laak, T. L., Gebbink, W. A., and Tolls, J. (2006b) , The effect of pH and ionic strength on the sorption of sulfachloropyridazine, tylosin, and oxytetracycline to soil, Environ. Toxicol. Chem. 25, 904–911. Thiele, S. (2000), Adsorption of the antibiotic pharmaceutical compound sulfapyridine by a long-term differently fertilized loess Chernozem, J. Plant Nutr. Soil Sci.—Z. Pflanzenernahr. Bodenkd. 163, 589–594. Thiele-Bruhn, S. (2003), Pharmaceutical antibiotic compounds in soils—a review, J. Plant Nutr. Soil Sci.—Z. Pflanzenernahr. Bodenkd. 166, 145–167. Thiele-Bruhn, S., Seibicke, T., Schulten, H. R., and Leinweber, P. (2004), Sorption of sulfonamide pharmaceutical antibiotics on whole soils and particle-size fractions, J. Environ. Qual. 33, 1331–1342. Tolls, J. (2001), Sorption of veterinary pharmaceuticals in soils: A review, Environ. Sci. Technol. 35, 3397–3406. Topp, E., Hendel, J. G., Lapen, D. R., and Chapman, R. (2008a) , Fate of the nonsteroidal anti-inflammatory drug naproxen in agricultural soil receiving liquid municipal biosolids, Environ. Toxicol. Chem. 27, 2005–2010. Topp, E., Monteiro, S. C., Beck, A., Coelho, B. B., Boxall, A. B. A., Duenk, P. W., Kleywegt, S., Lapen, D. R., Payne, M., Sabourin, L., Li, H. X., and Metcalfe, C. D. (2008b) , Runoff of pharmaceuticals and personal care products following application of biosolids to an agricultural field, Sci. Total Environ. 396, 52–59.

REFERENCES

Uslu, M. O., Yediler, A., Balcioglu, I. A., and Schulte-Hostede, S. (2008), Analysis and sorption behavior of fluoroquinolones in solid matrices, Water Air Soil Pollut. 190, 55–63. Van Emmerik, T., Angove, M. J., Johnson, B. B., Wells, J. D., and Fernandes, M. B. (2003), Sorption of 17 beta-estradiol onto selected soil minerals, J. Colloid Interface Sci. 266, 33–39. van Olphen, H., ed. (1977), An Introduction to Clay Colloid Chemistry. Wiley-Interscience, New York. Wehrhan, A., Kasteel, R., Simunek, J., Groeneweg, J., and Vereecken, H. (2007), Transport of sulfadiazine in soil columns— Experiments and modelling approaches, J. Contam. Hydrol. 89, 107–135. Weissmahr, K. W., Haderlein, S. B., and Schwarzenbach, R. P. (1998), Complex formation of soil minerals with nitroaromatic explosives and other pi-acceptors, Soil Sci. Soc. Am. J. 62, 369– 378. Werner, J. J., McNeill, K., and Arnold, W. A. (2005), Environmental photodegradation of mefenamic acid, Chemosphere 58, 1339– 1346. Werner, J. J., Arnold, W. A., and McNeill, K. (2006), Water hardness as a photochemical parameter: Tetracycline photolysis as a function of calcium concentration, magnesium concentration, and pH, Environ. Sci. Technol. 40, 7236–7241. WHO (2005), World Health Report. available at www.who.int/whr/ 2005/annexesen.pdf. Wilga, J., Kot-Wasik, A., and Namiesnik, J. (2008), Studies of human and veterinary drugs’ fate in environmental solid samples—Analytical problems. J. Chromatogr. Sci. 46, 601–608. Willert, P. L. (2007), Assessment of the pharmaceutical market in Poland after accession to the European Union, Eur. J. Health Econ. 8, 347–357. Williams, C. F., Williams, C. E., and Adamsen, E. J. (2006), Sorption-desorption of carbamazepine from irrigated soils, J. Environ. Qual. 35, 1779–1783. Xu, J., Chen, W. P., Wu, L. S., Green, R., and Chang, A. C. (2009), Leachability of some emerging contaminants in reclaimed municipal wastewater-irrigated turf grass fields, Environ. Toxicol. Chem. 28, 1842–1850. Yamamoto, H., Liljestrand, H. M., Shimizu, Y., and Morita, M. (2003), Effects of physical-chemical characteristics on the sorption of selected endocrine disrnptors by dissolved

213

organic matter surrogates, Environ. Sci. Technol. 37, 2646– 2657. Yeager, R. L. and Halley, B. A. (1990), Sorption-desorption of [C-14] efrotomycin with soils, J. Agric. Food Chem. 38, 883–886. Ying, G. G., Kookana, R. S., and Dillon, P. (2004), Attenuation of two estrogen compounds in aquifer materials supplemented with sewage effluent, Ground Water Monit. Remediat. 24, 102–107. Ying, G. G., and Kookana, R. S. (2005), Sorption and degradation of estrogen-like-endocrine disrupting chemicals in soil, Environ. Toxicol. Chem. 24, 2640–2645. Yu, X. Y., Ying, G. G., and Kookana, R. S. (2006), Sorption and desorption behaviors of diuron in soils amended with charcoal, J. Agric. Food Chem. 54, 8545–8550. Yu, Z. Q., Xiao, B. H., Huang, W. L., and Peng, P. (2004), Sorption of steroid estrogens to soils and sediments, Environ. Toxicol. Chem. 23, 531–539. Yu, Z. Q. and Huang, W. L. (2005), Competitive sorption between 17a-ethinyl estradiol and naphthalene/phenanthrene by sediments, Environ. Sci. Technol. 39, 4878–4885. Zeng, G. M., Zhang, C., Huang, G. H., Yu, J., Wang, Q., Li, J. B., Xi, B. D., and Liu, H. L. (2006), Adsorption behavior of bisphenol A on sediments in Xiangjiang River, central-south China, Chemosphere 65, 1490–1499. Zhang, D., Pan, B., Zhang, H., Ning, P., and Xing, B. (2010), Contribution of different sulfamethoxazole species to their overall adsorption on functionalized carbon nanotubes, Environ. Sci. Technol. 44, 3806–3811. Zhou, J. L. and Liu, Y. P. (2000), Kinetics and equilibria of the interactions between diethylhexyl phthalate and sediment particles in simulated estuarine systems, Mar. Chem. 71, 165–176. Zhou, J. L. (2006), Sorption and remobilization behavior of 4-tertoctylphenol in aquatic systems, Environ. Sci. Technol. 40, 2225– 2234. Zhou, J. L., Liu, R., Wilding, A., and Hibberd, A. (2007), Sorption of selected endocrine disrupting chemicals to different aquatic colloids, Environ. Sci. Technol. 41, 206–213. Zukowska, B., Breivik, K., and Wania, F. (2006), Evaluating the environmental fate of pharmaceuticals using a level III model based on poly-parameter linear free energy relationships, Sci. Total Environ. 359, 177–187.

8 FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT PETER GRATHWOHL 8.1. Introduction: Subsurface Pollution 8.1.1. Overview of Relevant Organic Compounds 8.1.2. Persistence in the Subsurface Environment 8.1.3. Equilibrium Distribution in the Three-Phase System: Soil Solids, Air, and Water 8.1.4. Geosorbents 8.1.5. Partitioning 8.1.6. Nonlinear Sorption 8.2. Sorption/Desorption Kinetics 8.2.1. Intraparticle Diffusion 8.2.2. Hysteresis in Diffusion-Limited Sorption/Desorption Kinetics 8.2.3. Sorption and Biodegradation 8.3. Reactive Transport in the Subsurface Environment 8.3.1. Vapor-Phase Diffusion in the Unsaturated Soil Zone 8.3.2. Diffusion and Reaction of Biodegradable Organic Compounds in the Vapor Phase 8.3.3. Mass Transfer Across the Capillary Fringe 8.3.4. Natural Attenuation in Groundwater: Steady-State Plumes

This chapter introduces sorption and mass transfer principles of organic compounds in soils, sediments, and groundwater. Reaction and biodegradation are limited to simple models and steady-state scenarios. For contaminant fate and transport in soils, sediments, and groundwater extensive literature exists of which a (noncomprehensive) selection is listed in the references at the end of the chapter for more detailed information.

8.1. INTRODUCTION: SUBSURFACE POLLUTION 8.1.1. Overview of Relevant Organic Compounds Humankind produces an unprecedented chemical footprint on the whole planet that is an often overlooked but important aspect of global change. After onset of the industrial revolution, emission of pyrogenic compounds [e.g., polycyclic aromatic hydrocarbons (PAHs)] into the environment increased by a factor of
100 above natural background due to increased demand for fossil fuels (coals and oil). In the twentieth century, many manufactured chemicals for various purposes were released into the environment in large volumes. The most prominent groups were halogenated organic compounds used as pesticides (e.g., DDT, lindane, PCP), flame retardants (e.g., PCBs or later PBDE), chlorinated solvents, and plastic additives (e.g., phthalates, bisphenol A), among others. Many of these anthropogenic compounds are inherently stable in the environment, and because of transport in the atmosphere as gases or associated with particles, they occur now as “diffuse pollutants” from a regional to global scale (Simonich and Hites 1995; Wilcke 2007). In terms of absolute concentrations, the most prominent group are PAHs followed by pesticides, PCBs, and more recently emerging pollutants used in personal care products, flame retardants, and other applications. In industrial countries, PAHs typically occur in concentrations 100--1000 times higher than do PCBs, and exceed health standards set for soil and water in many environmental matrices such as urban soils, road dust, deposition on glass surfaces, urban runoff, and snowmelt water (Gocht et al. 2005, 2007a; Johnsen and Karlson 2007).

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang.
2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

215

216

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

PAHs in concentrations up to 80 mg/kg originating likely from steel industry emissions from the 1920s to the 1970s in Saarland, Germany (Yang et al. 2008b). The most detailed evidence so far is provided by Gocht et al. (2007a,b) based on 2 years0 input--output mass balances at catchment scale in southern Germany. Here historical trends of atmospheric deposition obtained from sediment cores were compared with the current burden of PAHs in soils. Even semivolatile and easily degradable PAHs such as phenanthrene exceeded the expected concentrations in topsoils by a factor of >2 indicating, that if biodegradation occurs at all, it is much slower than the deposition rates (timescales may be centuries). Soils and sediments in Europe often have higher concentrations of four-or five- ring PAHs than expected from current concentrations in the air reflecting nonequilibrium conditions and higher input in former time periods (nineteenth--twentieth century). In the tropics the opposite is observed (Daly et al. 2007) — there soils still act as sinks for atmospheric pollution. Natural soils and sediments are able to store chemical signals over many decades, centuries, if not millennia. Processes that appear to be slow in laboratory investigations may be relatively fast at the field scale. Sorption/desorption kinetics in laboratory experiments may be considered slow and very slow at a timescale of 100 days (“master’s thesis”) and 1000 days (“PhD thesis”), respectively. R€ ugner et al. (1999), for example, showed that sorption kinetics are limited by diffusion processes typically related to the inverse of the particle size squared; that is, very small particles equilibrate very quickly because of the large surface-tovolume ratio. Biodegradation experiments often rely on spiked batches of well-mixed aqueous suspensions, and half-life times are typically on the order of days (easily degradable compounds) up to months (slowly degrading

Organic compounds as separate liquid phase [nonaqueous-phase liquids (NAPLS)] also infiltrated locally into the ground from leaking pipelines, underground storage tanks and landfills. Such massive “oil” infiltrations are the reason for the most severe groundwater contamination in many industrialized countries. Organic liquids denser than water [Dense NAPL (DNAPL)] typically penetrate into aquifers to large depths and accumulate on top of low-permeability strata as so-called pools, which dissolve only extremely slowly [on the timescale of centuries (Eberhardt and Grathwohl 2002)]. Figure 8.1 shows that persistent compounds such as the chlorinated aliphatic hydrocarbons are most frequently above legal standards in groundwater, whereas other hydrocarbons such as fuel constituents, such as benzene, toluene, ethylbenzene and xylenes (BTEX) — although more frequently occurring in spills — represent a much smaller share of groundwater contamination. This is due to natural attenuation (mostly biodegradation) of hydrocarbons in the subsurface environment. 8.1.2. Persistence in the Subsurface Environment Sea, lake, or estuary sediments can record and preserve environmental signals (“geologic record”). Sediments are often anaerobic, which limits aerobic biodegradation of many organic compounds. Topsoils, however, are mostly aerobic, and high bioactivity is inherent. Nevertheless, evidence that organic compounds are very persistent even in topsoils is increasing. For example, even in agricultural soils chemicals are found decades after the last application such as DDT in Canada in the 1970s (Kurt-Karakus et al. 2006), fumigants in Connecticut (Steinberg et al. 1987), polycyclic musks (personal care products) in southern Germany after sewage sludge application in the 1980s (LUBW 2003) or

Petroleum hydrocarbons: gasoline stations, refineries, underground storage tanks 20.3%

64.8%

Benzene, Toluene, Ethylbenzene, Xylene, 8.0%

4.5% Chlorinated hydrocarbons: workshops, metal/ electronic industries, dry cleaners

Polycyclic Aromatic Hydrocarbons: Gasworks, cokeries Inorganic compounds 0.4% Heavy metals 1.0% Other Other organic compounds 1.0%

Figure 8.1. Frequent groundwater pollutants in southern germany [data from Environmental Protection Agency (LfU), Baden-W€ urttemberg 1996)].

INTRODUCTION: SUBSURFACE POLLUTION

compounds). From such laboratory findings, timescales of native pollutant persistence of more than a decade can hardly be expected and other mechanisms or spatiotemporal scales have to be considered to explain the persistence of organic compounds in the field.

KFr and 1/n are the Freundlich sorption coefficient [mg/kg (mg/L)1/n] and the Freundlich exponent [], respectively. It should be noted that values for KFr depend on the concentration units chosen. More convenient is the unit equivalent form of the Freundlich equation, where Cw is normalized by the compounds’ water solubility S:

8.1.3. Equilibrium Distribution in the Three-Phase System: Soil Solids, Air, and Water Depending on their physicochemical properties, organic compounds distribute in the soil system. Highly volatile compounds occur preferentially in the gas phase of soils, highly soluble compounds dissolve in the aqueous phase, and strongly sorbing compounds accumulate within the solid phase. Fundamental parameters for fate and transport models are distribution coefficients, defined as concentration ratios between the different phases that are established under equilibrium conditions as shown in Figure 8.2. Most important are the Henry’s law constant (¼ the air/water distribution coefficient: H) and the solids/water distribution coefficient Kd. In humid climate, solid surfaces in the subsurface environment are covered by a water layer and no direct contact of the vapor phase on dry surfaces occurs. In simple (i.e., linear) cases distribution coefficients are independent of concentration, and are termed partitioning coefficients (e.g., H). If sorption is nonlinear, as is often the case in soils if large concentration ranges are considered, then Kd becomes a function of the concentration. Nonlinear sorption is best described by the Freundlich sorption isotherm Cs ¼ KFr Cw1=n

ð8:1Þ

where Cs (e.g., in mg/kg) and Cw (e.g. in mg/L) denote the concentrations in the solid and aqueous phases, respectively;

217

Cs ¼

* KFr


1=n Cw S

ð8:2Þ

* (¼ KFr S1/n) is independent of the concentration and Here, KFr carries the same units as Cs; for Cw/S ¼ 1, it may be envisioned as the maximum loading capacity of a sorbent if the solute concentration approaches the solubility limit. Traditionally, nonlinear sorption is explained by a superposition of, for instance, Langmuir sorption isotherms (see Chapter 1 by J. Pignatello). Xia and Ball (1999) and later Kleineidam et al. (2002) showed that actually a combination of nonlinear adsorption and linear partitioning (absorption) explains sorption isotherms of organic compound in soils very well (see Section 8.1.6). An example of the combination is shown in Figure 8.4. Mechanistically, nonlinearity is due to adsorption onto surfaces or filling of molecule-sized pores, which predominates sorption at low concentrations. Typically, nonlinear sorption is characterized by high Kd values at low concentrations, whereas at high concentrations the partitioning component predominates with lower Kd values (see Fig. 8.4).

8.1.4. Geosorbents Geosorbents, especially natural organic matter and carbonaceous particles, are of crucial importance for the fate and transport of organic compounds in soils, sediments, and

Figure 8.2. Distribution of pollutants between soil solids, soil air, and soil water.

218

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

aquifers (Yang et al. 2008a,b; Jeong et al. 2008). Carbonaceous particles comprise thermally altered organic matter and can be of natural or anthropogenic origin, including coals and charcoal, coke, chars, and soot (commonly called “black carbon”). For a review of organic matter and carbonaceous materials and their properties, see Allen-King et al. (2002), and for methods of identification of the various types of organic or carbon particles, see Ligouis et al. (2005) as well as the contribution of J. Pignatello to this book (Chapter 1). The role of carbon nanoparticles in sorption has been intensively discussed [see review by Pan and Xing (2009)]. Since natural sediment deposits and soils always contain significant amounts of organic carbon, this can be considered the controlling geosorbent in the case of nonpolar organic compounds. Specific interactions with mineral surfaces (e.g., iron hydroxides, clay minerals) become additionally important if polar compounds are concerned. A special case is dry mineral surfaces, which may occur occasionally in topsoils at the soil--atmosphere interface (see K.-U. Goss, Chapter 5); such largely water-depleted mineral surfaces may adsorb organic compounds strongly (e.g., semivolatile compounds form the atmosphere: naphthalene, PCBs, etc.). If rewetted, these compounds are rapidly displaced from the surfaces by water, which is a strong competitor for sorption sites on mineral surfaces. This causes the release of a flash of pollutants, which either revolatilize to the atmosphere or leach toward the groundwater with infiltrating water (e.g., during a rain event). 8.1.5. Partitioning Since sorption of organic compounds often depends on the organic matter in soils, Kd values are normalized to the soil organic carbon fraction (foc), which allows calculation of the partitioning coefficient Koc ¼ Kd/foc. Many empirical correlations exist that relate Koc to single parameters such as the water solubility (S) or the octanol/water partition coefficient (KOW) of organic compounds [for an overview, see AllenKing et al. (2002) or Razzaque and Grathwohl (2008)]. Such empirical relationships work well for nonpolar or only slightly polar organic compounds (PAHs, PCB, BTEX, etc.), but they strictly apply to linear sorption, such as partitioning of the compounds between soil water and soil organic matter as derived from Raoult’s law (Schwarzenbach et al. 2002; Chiou et al. 1979, 1983). Following Chiou et al. (2005) and Razzaque and Grathwohl (2008), the partitioning coefficient may be defined as Koc ¼

Coc fo fo b ¼   Cw xo go S fo ðMo =Mi Þgo S Sðkg=LÞ

ð8:3Þ

where Coc and Cw denote the equilibrium concentration of a compound i normalized to the organic carbon content and its

equilibrium concentration in water, respectively. The parameter fo denotes the mass fraction of i in the organic matter (e.g., in g/g) [and drops out of Eq. (8.3)]. The term (foMo/Mi) represents the mole fraction (xo) of i in the organic matter [Mo and Mi represent the molecular weight of the organic phase and the compound (g/mol)]. Under ideal conditions the activity coefficient go is 1 and an inverse linear with the water solubility is expected. The parameter b in Equation (8.3) equals 1 for partitioning of a pure compound between itself and water. Note that if S in Equation 8.3 has the units kg/L, it accounts for the molecular weight Mi [S (kg/ L)] ¼ S (mol/L)  Mi [(kg/mol)], which improves the fit of Koc versus 1/S as shown by Razzaque and Grathwohl (2008). In reality, such empirical relationships are valid only for a certain group of similar compounds. Additionally, they are often slightly nonlinear, indicating compound-dependent solute--sorbent interactions as shown in Figure 8.3; b is a fitting factor usually significantly lower than 1 (e.g., 0.054 for the data shown in Fig. 8.3). Empirical relationships based on Koc are preferred to organic matter partition coefficients because the soil organic carbon content can be more precisely determined than the content of organic matter (often an organic carbon content of 58% is assumed for soil organic matter). Single-parameter relationships as discussed above (i.e., Koc vs. 1/S or Koc vs. Kow) do not consider the specific interactions that may occur between the compound and the soil solids. Polyparameter linear free energy relationships (PP-LFER) have been developed that apply for a wider range of compounds, including polar organic compounds (Goss and Schwarzenbach 2001; Nguyen et al. 2005; Endo et al. 2008a, 2009). Again, such correlations apply for linear sorption or partitioning; however, a first attempt exists to extend polyparameter approaches to nonlinear sorption (Endo et al. 2008b). 8.1.6. Nonlinear Sorption Nonlinear sorption of organic compounds is frequently observed in soils, especially if large concentration ranges (several orders of magnitude) are considered. Reasons for isotherm nonlinearity are adsorption mechanisms involving pore filling or surface covering of rigid particles. Since very heterogeneous mixtures of many geosorbents occur in soils, a superposition of sorption models is often applied to describe sorption of organic pollutants. Different models can be combined as described in the literature [see, e.g., the series of papers on “distributed reactivity models” by Weber et al. (1992) and Huang et al. (1997) and on “dual-mode sorption” by Xing and Pignatello (1997); see also Cornelissen and Gustafsson (2004)]. Figure 8.4 shows as an example of the combination of partitioning [Eq. (8.3)] and pore filling, which fits the unit equivalent Freundlich sorption isotherm reasonably well. Equation 8.4 combines Raoult’s

INTRODUCTION: SUBSURFACE POLLUTION

219

Figure 8.3. Regressions between log Koc and water solubility (log S) for different groups of organic compounds (solid lines with slopes of approximately 1; for all compounds: log Koc ¼ 0.85 log S 0.55) and Koc* ¼ Koc S (horizontal dotted line: average Koc* for all 64 compounds ¼ 0.054); symbols denote compound groups: &—monoaromatic hydrocarbons, &—chlorinated aromatic compounds; }—PCBs; ¤—chlorinated aliphatics; &—PAHs [data from Razzaque and Grathwohl (2008)].

law [partitioning; see Eq. 8.3)] and the Polanyi--Manes model for pore-filling  ! bCw RT ½ln ðCw =SÞ 2 þ Cs;max exp  Cs ¼ foc bEo S *  KFr


1=n Cw S

ð8:4Þ

where Cs,max and foc denote maximum adsorbed concentration due to pore filling, for example, in carbonaceous particles, and the fraction of the “partitioning” organic carbon, respectively; ßEo represents an energy term (kJ/mol), R is the ideal gas constant in appropriate units, and T is the temperature in Kelvin. For more details see Allen-King et al. (2002) and Manes (1998). The superposition of adsorption (nonlinear) and absorption (linear) may explain Freundlich-type sorption models (Xia and Ball 1999). The advantage of Equation (8.4) is that both the partitioning and the adsorption parts can be defined conveniently using solubility-normalized aqueous concentrations (Cw/S), which lead to the unit equivalent Freundlich sorption isotherm shown above (Kleineidam et al. 2002). This allows one to predict sorption coefficients of a variety of organic pollutants based on single or few measurements with a chemical probe that is representative of the group of compounds of interest as described by Allen-King et al. (2002). The unit equivalent Freundlich equation allows one to express Kd independent of the sorption isotherm linearity Kd ¼

Figure 8.4. Nonlinear Freundlich sorption isotherm (dashed line) fitted to a superposition ( þ ) of the Polanyi sorption model (triangles) and linear partitioning (squares); see Equation (8.4).

* * Cs KFr ðCw =SÞ1=n KFr ðCw =SÞ1=n1 ¼ ¼ Cw Cw S

ð8:5Þ

which indicates an inverse linear relationship between Kd and S at a given ratio of Cw/S, independent of the nonlinearity of the sorption isotherms as shown in Figure 8.5. The fit over very different compounds is remarkable; nevertheless, this single-parameter approach is, of course, valid only for strongly hydrophobic compounds that do not undergo specific interactions with the soil solids. For polar compounds, polyparameter approaches still have to be developed for a

220

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

Figure 8.5. Inverse linear relationship of Kd values of different organic compounds calculated at Cw/ S ¼ 0.1 for highly nonlinear sorption to activated carbon (top, squares), bituminous coals, lignite coke (diamonds) and linear sorption to peat (circles) [data from Kleineidam et al. (2002)].

variety of natural adsorbents (Endo et al. 2008a,b) (see also Chapters 1 and 5, below). For linear sorption (1/n ¼ 1), Eq. 8.5 reduces to Kd ¼ KFr /S.

8.2. SORPTION/DESORPTION KINETICS 8.2.1. Intraparticle Diffusion For fate and transport of pollutants in the subsurface environment, the sorption capacity (or the storage capacity) as well as the kinetics of sorption/desorption are important. Sorption and desorption kinetics control not only the transport velocity of a compound in the gas phase, seepage water, and groundwater but also the bioavailability (e.g., in the sense that compounds can be used by microorganisms as carbon sources). On the particle scale, desorption (as well as sorption) kinetics may involve different physical processes such as (1) film diffusion connecting the mobile phase with the immobile phase and (2) intrasorbent or intraparticle diffusion. In the long term, mostly process 2 becomes controlling, and spherical diffusion models are appropriate for describing sorptive uptake and desorption kinetics on the particle scale  2  @Cw @ Cw 2 @Cw ¼ Da þ r @r @t @r2

ð8:6Þ

where Da, r, and t denote the apparent diffusion coefficient (m2/s), for example, the intrasorbent or polymer diffusion coefficient for diffusion in organic matter or the retarded pore diffusion coefficient in porous particles (e.g., charcoal, coals, clay aggregates, rock fragments), the radial distance from the center of the spherical particle (m), and time (s), respectively; Da can be derived from the transport equation as follows

  @Cw Daq @ 2 Cw 2 @Cw @Cs ¼ e þ R e 2 r @r @t tf @r @t

ð8:7Þ

where Daq/tf denotes the pore diffusion coefficient (¼ Dp), which accounts for diffusion in water (Daq) and the tortuosity factor of the pores (tf); e is the intraparticle porosity (), accounting for the diffusion effective area as well as for the pore volume; R is the bulk density of the particle (kg/L); Kd (¼ Cs/Cw) can be used to simplify Equation (8.7): @Cw @t

2 3 Dp e 4@ 2 Cw 2 @Cw 5 ¼ þ r @r ðe þ Kd RÞ @r2 2 3 2 3 2 Dp 4@ 2 Cw 2 @Cw 5 @ C 2 @C w w 5 ¼ Da 4 2 þ ¼ þ r @r r @r R @r2 @r ð8:8Þ

where R is the retardation factor of pore diffusion (R ¼ 1 þ Kd R/e). The term Dp e denotes the effective diffusion coefficient (De), which can be estimated from the aqueous diffusion coefficient (Daq) by empirical relationships based on Archie’s law: De ¼

Daq e e De ¼ Daq em Yem ¼ ¼ tf Daq tf

ð8:9Þ

Here, tf is the tortuosity factor accounting for the pore geometry. The empirical exponent m lies between 1.5 (soft sediments) and 2.5 (hard rocks); Boving and Grathwohl (2001) report a value of 2.2 for limestones and sandstones [see also Grathwohl (1998); and R€ugner et al. (1999)]. The spherical diffusion equation has the following analytical solution for diffusion in a spherical particle of radius a

SORPTION/DESORPTION KINETICS

221

under the boundary conditions of an infinite bath (¼ constant concentration at the particle surface) 2 0 1 0 1 M 6 4 @ 2 Da A 1 D a ¼ 1 2 exp p 2 t þ 2 exp@22 p2 2 tA Meq p a a 2 0 1 # 1 D a þ 2 exp@32 p2 2 tA þ . . . a 3 0 1 ¥ M 6X 1 D a ! ¼ 1 2 exp@n2 p2 2 tA Meq p n¼1 n2 a

ð8:10Þ

where Meq is the solute mass in the particle under equilibrium conditions and M is the mass that has diffused into (sorptive uptake) or out (desorption) of a particle. This analytical solution holds only for linear sorption (Kd ¼ constant). For nonlinear sorption, numerical solutions of Equation (8.7) are needed. Since numerical models and the analytical solution [Eq. (8.10)] are laborious, in many sorption/desorption kinetic studies first-order approximations are used to simplify diffusion limited sorption/ desorption kinetics: M ¼ 1 expðl tÞ Meq

ð8:11Þ

This corresponds to just one term in the series expansion of Equation (8.10) and the first-order rate constant l [in reciprocal seconds (s1)] replaces p2Da/a2. For long-term approximations of diffusion rates, the first term in Equation (8.10) is sufficient, whereas the short-term approximation is given by a “square root of time” relationship that is typical for transient diffusion processes: M ¼6 Meq

rffiffiffiffiffiffiffi Da t pa2

ð8:12Þ

If first order approximations are used to fit early time sorptive uptake or desorption [e.g., Eq. (8.11)], then it fits only part of the kinetic profile and therefore second- or third first-order terms have to be added to achieve a reasonable fit as shown in Figure 8.6. In such cases, l in Equation (8.11) becomes a function of the time range used for fitting. The term Dat/a2 in Figure 8.6 denotes dimensionless time (also called Fourier number), which allows one to calculate the time needed to achieve a certain degree of sorption or desorption (for infinite bath boundary conditions). For M/Meq of 0.9 (90% sorbed or desorbed) and 0.99, the respective time periods needed are:

Figure 8.6. Comparison of desorption rates calculated from time derivatives of Equation (8.11) (30 terms, solid line); squares are the sums of three first-order approximations shown as dashed lines (rate constants of the ninth, third, and first terms with fractions of 0.77, 0.23, and 0.077, respectively).

t90 ¼ 0:183

t99

a2 Da

ð8:13Þ

a2 ¼ 0:416 Da

Thus, timescales for sorption or desorption can be estimated, provided that sorption capacity (e.g., Kd) and diffusion coefficients are known; note that t in Equations (8.13) increases with the size of the particles squared. 8.2.2. Hysteresis in Diffusion-Limited Sorption/ Desorption Kinetics J. Pignatello already discussed potential reasons for sorption/ desorption hysteresis comprehensively in Chapter 1. Diffusion-limited sorption/desorption kinetics do not show any hysteresis provided that (1) sorption isotherms are linear, (2) initial conditions for sorption/desorption are the same [e.g., complete sorption equilibrium was reached before desorption starts again (Altfelder et al. 2000)], and (3) no structural changes of the sorbent occur during sorption/ desorption. In contrast to reason 1 and 2, reason 3 relates to diffusion coefficients that account for changes in physicochemical properties, which may further affect the activation energy of diffusion (i.e., the slower the diffusion, the higher the activation energy). The activation energy for desorption accounts for both the enthalpy of sorption/desorption and the activation energy of molecular diffusion. Kleineidam et al. (2004) showed that no real hysteresis occurs if sorption/desorption kinetics is limited by aqueous diffusion in the pore space (i.e., if diffusion is independent of the direction of transport). Wang et al. (2007, 2009) measured

222

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

enthalpies from sorption/desorption isotherms of phenanthrene as well as the activation energies for desorption kinetics as a function of concentration. The results show that for the investigated fine-grained carbonaceous sorbents (peat, lignite, bituminous coal, and two soils), no significant sorption/desorption hysteresis occurred. Furthermore, no significant increase of activation energies for diffusion-limited desorption of phenanthrene from coals was observed. This indicates that with decreasing concentration no slow diffusion mechanisms appear that would require higher activation energies (Wang et al. 2009). Nevertheless, with sorption enthalpies in the range of 20--35 kJ/mol, the activation energies of desorption from lignite and bituminous coal (60--70 kJ/mol) were significantly higher than expected from free diffusion in water-filled pores, indicating that diffusion occurs in micropores or in the organic matrix. Generally, desorption of a “labile” fraction would be expected to be associated with lower activation energies compared to desorption of strongly or recalcitrant sorbed compounds. For hydrophobic organic compounds such as PAHs, the literature reports activation energies in the range of 60--70 kJ/mol (Cornelissen et al. 1997) for sediments and 40--70 kJ/mol for soils, clays, and aquifer materials (Johnson and Weber 2001; Kleineidam et al. 2004; Ghosh et al. 2001). Only for strongly contaminated coal particles, Ghosh et al. (2001) report activation energies above 100 kJ/mol, which might indicate some specific trapping mechanism of the sorbate. 8.2.3. Sorption and Biodegradation Slow biodegradation of organic pollutants in soils is often attributed to slow desorption kinetics from soil particles. However, as shown in Section 8.2.1, diffusion-limited desorption form small particles (e.g., black carbon) is fairly rapid and does not explain the persistence of organic pollutants for decades or centuries. Objective of the following to show how equilibrium sorption alone can slow down biodegradation rates. Sorption decreases the concentration of organic compounds in the aqueous phase of soil and thus their bioavailability. This may reduce toxic effects, and led to efforts to mix contaminated sediments with strong adsorbents for organic compounds such as activated carbon (Cho et al. 2009). However, strong sorption also appears to slow biodegradation, which may lead to increased persistence of organic compounds. The effect of sorption on the kinetics of biodegradation can be derived from a mass balance equation that assumes the aqueous concentration as the rate-limiting factor for biodegradation (first-order kinetics): n@Cw R@Cs þ ¼ lnCw @t @t

ð8:14Þ

Here, the left term denotes the mass of organic compound in the water and the solids (sorbed); n () and r (kg/L)

represent the volume of water and the dry mass of the solids in a unit volume (e.g., a batch reactor) or the porosity and the dry bulk density in a water-saturated porous medium, respectively. The parameter l is the first-order rate constant for biodegradation (1/s). Incorporating the definition of Kd ( ¼ Cs/Cw) allows simplification of Equation (8.14) to @Cw l ¼ Cw 1 þ Kd ðR=nÞ @t

ð8:15Þ

which has the following analytical solution:   C lt ¼ exp C0 1 þ Kd ðR=nÞ

ð8:16Þ

In an equilibrated system C/C0 represents the relative concentration in aqueous and solid phases (or both phases together), and C0 is the initial concentration. The term (1 þ Kd r/n) is known as the retardation factor (), which slows down transport and biodegradation in porous media. This shows that biodegradation rates are slowed down by sorption but also depend on the solid : liquid ratio (here r/n), that is, decrease with decreasing water content. Sorption decreases the aqueous concentration Cw with increasing Kd and an increasing portion of available sorbent (r/n). Low solid : liquid ratios are typically used in laboratory batch experiments. Typical values for r/n in the field may range, for example, from 10 (unsaturated) to 5 (water-saturated), whereas in laboratory batch experiments they can be as low as 0.001. Figure 8.7 shows how half-lives of first-order

Figure 8.7. Half-lives of first-order biodegradation as a function of liquid : solid ratios (LS
0.1 is representative of field condition; LS 10--100 commonly used in batch experiments) for hypothetical compounds with different Kd values ranging from 1, 10, 100, and 1000 to 10000 L/kg (inverse to water solubility).

REACTIVE TRANSPORT IN THE SUBSURFACE ENVIRONMENT

biodegradation depend on sorption, and thus water solubility for hydrophobic compounds [see Eq. (8.5)] at different water : solid ratios. Half-lives of strongly sorbing (lowsolubility) compounds may differ by two orders of magnitude depending on the solid : water ratios. For other biodegradation kinetics (zero order, Monod, etc.) and nonequilibrium sorption, see Haws et al. (2006) and Mittal and Rockne (2008).

This chapter introduces the principles of transport of organic compounds in the subsurface environment. Since substantial literature exists about reactive solute transport in seepage water and groundwater [for a compilation, see, e.g., Rolle et al. (2010)], this chapter focuses on mixing processes, which are relevant for transport and biodegradation of organic compounds, mostly relying on our own work. Transport including reaction is treated for vapor-phase diffusion in the unsaturated zone, mass transfer across the capillary fringe into groundwater, and finally the length of steady-state plumes of degradable organic compounds. Emphasis is on derivation of the relevant transport parameters, and simple calculations that allow a rough estimate of typical transport distances and timescales. 8.3.1. Vapor-Phase Diffusion in the Unsaturated Soil Zone Many organic compounds such as gasoline constituents (e.g., benzene, toluene, xylenes) and chlorinated solvents have relatively high vapor pressures and can be transported in vapor phase in the subsurface environment. Since diffusion coefficients in the gas phase are much higher than in a liquid (by a factor of
10,000), the diffusive flux in the water unsaturated zone depends mainly on the air-filled pore space, which, in turn, is a function of the water content. Diffusive transport of a compound in the gas phase is retarded by partitioning into water as described by Henry’s law and sorption to the soil solids ng

@Cg @ 2 Cg @Cw @Cs ¼ De;g R nw @t @x2 @t @t

simplified if the distribution coefficients between gas and water (Henry’s law constant: H ¼ Cg/Cw) and between water and soil solids (Kd ¼ Cs/Cw) are introduced ng

@Cg @t

ð8:17Þ

where De,g denotes the effective gas diffusion coefficient, which can be estimated with empirical relationships similar to the definition of the diffusion coefficients in the watersaturated zone [Eq. (8.19)]. The parameter ng denotes the porosity filled by soil air (gas phase), nw is the water-filled pore volume (¼ n S , where S is the degree of water saturation of the pore space), and R is the dry bulk density; Cg, Cw, and Cs denote the concentrations in the gas, water, and solid phases, respectively. Again, equation (8.17) can be

¼ De;g

@ 2 Cg nw @Cg Kd @Cg R  @x2 H @t H @t

Y @Cg @t

8.3. REACTIVE TRANSPORT IN THE SUBSURFACE ENVIRONMENT

223

¼

De;g @ 2 Cg De;g @ 2 Cg ¼ ng þ ðnw =HÞ þ RðKd =HÞ @x2 a @x2

¼ Da;g

@ 2 Cg @x2

ð8:18Þ

where Da,g denotes the apparent diffusion coefficient in the gas phase (¼ De,g/a) and a is the capacity factor. Multiplication of a by Cg gives the total concentration per unit volume of the porous medium. The effective gas diffusion coefficient (De,g) in the water-unsaturated zone can be calculated using on empirical correlations of the general form (Currie 1960) De;g ng ¼ ¼ f ðng Þ  f nm g Dg tf

ð8:19Þ

where Dg, De,g, and ng are the molecular gas diffusion coefficient, the effective diffusion coefficient in the vadose zone, and the air-filled porosity, respectively. The parameters f and m are empirical constants. Millington and Quirk (1960) proposed including additionally the overall porosity (n): 10=3

De;g ng ¼ 2 Dg n

ð8:20Þ

Sallam et al. (1984) extended these correlations to low airfilled porosities by replacing the exponent of ng by 3.1. More recently, a new relationship was developed by Moldrup et al. (2000), which gives very good predictions for De,g in sandy materials (Wang et al. 2003): De;g n2:5 g ¼ Dg n

ð8:21Þ

Figure 8.8 shows a comparison of some selected empirical correlations. Experimental data can be found in Wang et al. (2003) and field data, in Werner et al. (2004). Note, that Equation (8.20) is similar to the empirical description of relative hydraulic conductivity in unsaturated flow. For spreading of organic vapors in the unsaturated soil zone, analytic solutions of Equation (8.17) can be used if geometry and boundary conditions are simple. For realworld geometries, numerical models are often needed. A simple way to estimate how far organic vapors spread by

224

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

x calculated using Equation (8.22) denotes the distance from the center, where 78% of the maximum concentration is reached. In this case, the absolute concentrations depend on the dimensionality of diffusive mass transfer; diffusion in 1D space results in less dilution than in 2D or 3D cases as illustrated in Figure 8.9. 8.3.2. Diffusion and Reaction of Biodegradable Organic Compounds in the Vapor Phase

Figure 8.8. Comparison of empirical relationships to estimate effective gas diffusion coefficients in the unsaturated soil zone (filled symbols—toluene; open symbols—MTBE) [data from Wang et al. (2003)].

diffusion after a given time t is to use the mean-square relationship x¼

pffiffiffiffiffiffiffiffiffiffi Da;g t

ð8:22Þ

where Da,g denotes the apparent diffusion coefficient in the gas phase as derived in Equation (8.18). The diffusion distance x represents the distance in which about half of the concentration of a constant concentration boundary is reached. For an instantaneous (¼ Dirac-type) source, a Gaussian-type distribution of the compound is achieved and

Many volatile organic compounds in soils undergo biodegradation; for instance, many fuel constituents are easily degraded under aerobic conditions. Here, the biodegradation rates depend on diffusion rates of oxygen and hydrocarbons and on the kinetics of the microbial processes (Pasteris et al. 2001). Biodegradation limits the spreading of vaporphase plumes of organic contaminants because at a certain distance from the contaminant source, the biodegradation rates equal the diffusion rates of the contaminants and steadystate conditions are reached. In a one-dimensional case (e.g., floating gasoline on the groundwater table and oxygen diffusion from the soil--atmosphere interface) and quasiinstantaneous biodegradation, the depth z of the reaction zone is z¼

h ½ðDeHC =DeO2 ÞðgCHC =CO2 Þ þ 1

ð8:23Þ

where DeO2 =DeHC and CO2 =CHC denote the ratio of the effective diffusion coefficients of oxygen and hydrocarbons

Figure 8.9. Diffusion from an instantaneous source in one, two, and three dimensions (1D, 2D, 3D); M0 is the initial mass and r is the radial distance from the center (horizontal axis).

REACTIVE TRANSPORT IN THE SUBSURFACE ENVIRONMENT

and the respective concentrations, c denotes the stoichiometric ratio for the reaction of oxygen and hydrocarbons, and h is the distance between the constant source of hydrocarbons and the boundary with a constant oxygen concentration (e.g., the atmosphere) as shown in Figure 8.10. For an analytical solution of a radial source of fuel and a corresponding radial reaction zone, see Klenk (2000). For more complicated cases, numerical models are needed. Such codes also allow one to account for vapor-phase diffusion, seepage water flow and multiple reactions of organic contaminant mixtures, as well as metabolic pathways and microbial growth (Broholm et al. 2005; Molins and Mayer 2007; Maier and Grathwohl 2005; Grathwohl et al. 2002; Mayer et al. 2002). Figure 8.11 shows an example case for radial diffusion with and without biodegradation. Note that for multicomponent organic mixtures such as fuels, Raoult’s law has to be applied to calculate the respective vapor or aqueous-phase concentrations of the hydrocarbons (Cg,HC, Cw,HC): Pi  Mi RT ¼ xi;o gi;o Si

Cg;HC ¼ xi;o gi;o Cw;HC

ð8:24Þ

Here, xi,o and ci,o denote the mole fraction of a compound i in the organic mixture (e.g., fuel) and the respective activity coefficient, which is 1 for ideal cases and > 1 if i behaves nonideally in the organic mixture.The terms Pi and Si denote the saturation vapor pressure [e.g., in (kilopascals)] and the water solubility of i (e.g., in g/L), respectively; Mi, R, and T are the molecular weight of i, the gas law constant (e.g., 8.3144 L kPa mol1 K1 or J mol1 K1), and the tempera-

ture in Kelvin (e.g. 283 K in the subsurface environment), respectively. The value of xi,o can be calculated by the weight fraction of i in the organic mixture multiplied by the ratio of the molecular weights of the mixture and compound i (¼ fi,o Mo /Mi). For typical fuel constituents (e.g., aromatic and aliphatic hydrocarbons), ci,o close to 1 is reasonable (Broholm et al. 2005; Grathwohl et al. 2002; Reckhorn et al. 2001). This even holds for polycyclic aromatic hydrocarbons in more complex organic mixtures such as coal tars (Eberhardt and Grathwohl 2002), as long as the organic phase is liquid and the subcooled liquid solubility is used for compounds that are solid at ambient temperatures. For solidified coal tars, however, this no longer seems to hold (Liu et al. 2009 a,b). 8.3.3. Mass Transfer Across the Capillary Fringe Organic compounds can reach groundwater by transport in seepage water and by vapor-phase diffusion. Horizontal flow of water, which already occurs in the water-saturated capillary fringe above the groundwater table, causes vertical transverse dispersion and thus mixing of the solutes into the groundwater. Figure 8.12 summarizes the possible behavior of organic contaminants during transport into groundwater in a simplified scheme. The vertical flux of a contaminant into groundwater thus consists of two components: (1) solute transport by groundwater recharge and (2) transverse vertical mixing. The first can be calculated simply from the groundwater recharge rate and the solute concentration: FGWR ¼ GWR Cw xy

Concentrations Depth to groundwater table: h

225

Electron acceptor (EA) (e.g. O2 ca. 300 gm-3)

“reaction front” at depth z

CO FO 2 DeO 2 z 2 γ ÀCHC FHC DeHC h–z

Electron donor (ED) (e.g. hydrocarbons, benzene ca. 400 gm-3)

Figure 8.10. Reaction front for steady-state biodegradation; 1D case, such as oxygen coming from the atmosphere (9.4 mol/m3) is consumed by microorganisms degrading volatile hydrocarbons coming from a layer of organic liquid floating on the groundwater table, for example, benzene (5 mol/L); degradation of 1 mol of benzene requires 7.5 mol O2 (C6H6 þ 7.5 O2 ¼ 6 CO2 þ 3 H2O); the concentration-based stoichiometric factor is then around 3 (7:5  MO2 =Mbenzene ), the ratio of the diffusion coefficients benzene/O2 is
0.4; then z ¼ h/2.56 (Eq. (8.23)].

ð8:25Þ

226

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

(a) 30

Depth[m]

25 20 15 10 s

C /Co [-]

5 0 0

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

1.0E+000

25

5.0E-001

20

4.0E-001

15

3.0E-001

10

2.0E-001

5 0 0

[m]

7.5E-001

Depth

(b) 30

1.0E-001

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

5.0E-002 1.0E-002 1.0E-003

(c) 30

1.0E-004

Depth[m]

25

1.0E-005

20 15 10 5 0 0

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Horizontal Distance [m]

Figure 8.11. Reaction zones around a quasi-steady-state source of hydrocarbons subject of aerobic degradation in the unsaturated soil zone with an unsealed surface, that is, open to the atmosphere (Klenk 2000): (a) no biodegradation; (b, c) with biodegradation—distribution under steady-state conditions; (c) additionally accounting for with seepage water infiltration that has little influence because spreading is dominated by vapor-phase diffusion.

where GWR is the groundwater recharge rate, for example, as represented by the long-term annual mean, often between 0.5 and 1 mm per day (or 0.183 -- 0.365 m3 m2 year1), and Cw is the solute concentration in the water arriving at the groundwater table. The coordinates x and y denote longitudinal and transverse horizontal distance. In a first approximation, this can be envisioned as a layer (or wedge) of contaminated water that increases in height with travel distance x as shown in Figure 8.13. Vertical fluxes due to dispersion and diffusion from a boundary of constant concentration (e.g., a stable vapor-phase contamination at the top of the capillary fringe) can be assessed by introducing a boundary layer Fdisp ¼ Cw xz0 va n

ð8:26Þ

where z0 , va, and n denote the thickness of the vertical boundary layer (m), the average horizontal flow velocity of the groundwater (m/day), and the porosity (), respectively. In a first approximation, estimation of z0 may be based on the mean square displacement:

rffiffiffiffiffiffiffiffiffiffiffiffi x4 z ¼ D va p 0

ð8:27Þ

The term x/va represents the transport time in the x direction (at constant flow velocity); D denotes the hydrodynamic dispersion coefficient (m2/s) commonly defined as D ¼ D p þ a t va

ð8:28Þ

Here, Dp is the pore diffusion coefficient, often empirically described as the product of the aqueous diffusion coefficient and the porosity (Daq n). For many organic compounds, Dp typically is in the range of 1--3  10-10 m2/s. The parameter at denotes the transverse vertical dispersivity (m). At sufficiently high flow velocities the dispersion term in Equation (8.28) predominates and Equation (8.27) approaches z0 

pffiffiffiffiffiffiffi at x

ð8:29Þ

REACTIVE TRANSPORT IN THE SUBSURFACE ENVIRONMENT

227

Groundwater Risk Assessment in Case of Contaminated Soil / Contaminated Soil Air Source zone: The contaminant concentration in a soil leaching test or in soil air is above the legal limit !!!

Groundwater recharge (A)

D) n (A+

Is the concentration in the groundwater above the legal limit ???

Soil air sampling CO2, CH2 Column test

Possible scenarios: (A) (B)

Residual oil phase petroleum hydrocarbon B) No recharge (immobile Contaminants)

adatio

O2 - Diffusion

Volatile compounds, no recharge: Persistent compounds can reach the groundwater by vapor diffusion

(D)

Biodegradable compounds may reach the groundwater in very low concentrations (below legal limit ?) Further attenuation in the shallow plume due to volatilization and biodegradation ???

(E)

D) Biodegradation ???

with B

Groundwater recharge: Persistent contaminants can reach the groundwater No recharge: Nonvolatile compounds are immobile

(C)

A) Contaminant transport by seepage water C) no recharge: vapor diffusion only

iodegr

Depth (unsahurated zone)

Contaminant Concentration

E) Volatilization

CO2 from biodegradation

O2 - Supply Capillary fringe

Shallow plume E) Biodegradation ???

Aqueous concentration

Contaminant transport across the capillary fringe by diffusion and transverse dispersion

Contaminant volationzation across the capillary fring downgradient from the source zone

Groundwater

Oxygen transport across the capillary fringe acrobic biodegradation of hydrocarbons

Downgradient attenuation zone

Contaminant source zone

Distance from source zone

Figure 8.12. Overview on transfer mechanisms of organic compounds from the unsaturated zone across the capillary fringe into groundwater.

The term at is not well known for natural, especially heterogeneous, porous media. Laboratory tests with sandy porous media often show values that are far below 1 mm (Olsson and Grathwohl 2007). Field observations, however, yield values that are in the range of 3--10 cm (Maier

et al. 2005, 2007; R€ugner et al. 2004). These elevated values are probably due to heterogeneities that cause flow focusing (Werth et al. 2006; Rolle et al. 2009) or chromatographic mixing (Janssen et al. 2006). Transient conditions alone do not seem to cause much net mixing, as shown by numerical

Figure 8.13. Contaminant transport into the groundwater by groundwater recharge (solid line) and by dispersion (dotted line).

228

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

studies (Prommer et al. 1998, 2002; Schirmer et al. 2001). More recently, experimental evidence and theoretical considerations showed that, in contrast to textbook knowledge, dispersivity is compound-specific (Chiogna et al. 2010). The preceding equations show that FGWR increases linearly with x [Equation (8.25)] while Fdisp grows with the square root of x [Eqs. (8.26) and (8.29)]. Thus, at sufficiently large distance x, FGWR will predominate as shown in Figure 8.13. The same considerations apply to the transport of oxygen into groundwater, which is of major concern for aerobic biodegradation of organic compounds in groundwater as discussed in Section 8.3.4. 8.3.4. Natural Attenuation in Groundwater: Steady-State Plumes As outlined in the introduction, most groundwater contaminations are caused by organic compounds. Many hydrocarbons but also other frequent groundwater contaminants such as ammonium can be degraded aerobically by microorganisms. For groundwater risk assessment, not only the local contamination at a contaminated site but also the maximum spreading of a contaminant plume are relevant. Non- or very slowly degrading compounds cause kilometerlong plumes in groundwater, which, after many decades may finally discharge to surface waters (rivers, lakes, oceans). Plumes with biodegradable organic compounds reach steady state if the release rate from a continuous source is balanced by the overall biodegradation rate. If the source becomes depleted, then the plume starts to shrink. Several studies suggest that not only local biodegradation rate constants limit the overall biodegradation rates but also the mixing of the electron acceptors with the hydrocarbons (electron donors) by transverse dispersion (Cirpka et al. 1999, 2006; Grathwohl et al. 2000; Maier et al. 2005; Maier and Grathwohl 2006). Suitable electron acceptors are typically oxygen, and also nitrate and sulfate. Intrinsic electron acceptors that do not occur as solutes in water (and thus are not limited by mixing) are iron and manganese oxides, which often occur as coatings to sediment particles, or as precipitates in the pore space. The latter case is not considered here [see discussion in Rolle et al. (2010)]. For reactive transport, such as of oxygen, into a reduced contaminant plume in groundwater, the vertical depth of the reaction front (z ) is of interest [following Liu et al. (2009a)]   rffiffiffiffiffiffiffiffiffi x 1 * 1 z ¼ 2 D erfc þ1 va ðCEA =gCEDÞ

ð8:30Þ

where erfc1 denotes the inverse complementary error function. Its argument accounts for initial concentrations (CEA, CED) and the stoichiometric ratio of the bimolecular reaction between electron donor (ED) and acceptor (EA).

It should be noted that Equation (8.30) is valid only for a quasi-instantaneous reaction of electron donor and acceptor. To solve Equation (8.30) explicitly, an approximation can be used [for arguments b between 0.25 and 1, erfc1(b) approaches 1  b]:  rffiffiffiffiffiffiffiffiffi
x 1 D 1 va ðCEA =gCED þ 1
 pffiffiffiffiffiffiffi 1 ¼ 2 at x ðgCED =CEA Þ þ 1

z* ¼ 2

ð8:31Þ

If D is dominated by transverse dispersion, as discussed above, then the term under the square root becomes simply atx. It should be noted that the concentration of the reaction product depends only on the concentration of an electron acceptor and thus builds up in the region z < z as long as the electron donor is present in groundwater. If the term in parentheses becomes 12, which is the case for a stoichiometric ratio of 1 and equal concentrations of electron acceptor and donor, then Equation (8.31) equals Equation (8.29) for the non-reactive case. The treatment shown above is valid for infinite vertical dimensions—or the progress of the reaction front into a plume far away from the lower boundary. For a contaminated aquifer of a given thickness (h), the length (L) of the plume until the organic pollutant is degraded completely can be calculated as follows (Liedl et al. 2005):

 4h2 4 gCED L ¼ 2 ln þ1 p CEA p at

ð8:32Þ

Longitudinal dispersion does not contribute much to mixing under steady-state conditions and thus is neglected here. Again, major uncertainties exist in estimating appropriate values for at for natural porous media with heterogeneous groundwater flow (Werth et al. 2006). Equation (8.32) is valid only for a quasi-instantaneous reaction of electron donor and acceptor and thus gives the minimum length of the plume. Shorter plumes are possible only if intrinsic electron acceptors such as iron hydroxides are additionally consumed. Most important for the plume length is the source strength, that is, the height of the contaminated zone. REFERENCES Allen-King, R. M., Grathwohl, P., and Ball, W. P. (2002), New modeling paradigms for the sorption of hydrophobic organic chemicals to heterogeneous carbonaceous matter in soils, sediments, and rocks, Adv. Water Res. 25, 985--1016. Altfelder, S., Streck, T., and Richter, J. (2000), Nonsingular sorption of organic compounds in soils: The role of slow sorption kinetics, J. Environ. Qual. 29, 917--925.

REFERENCES

Boving, T. and Grathwohl, P. (2001), Matrix diffusion coefficients in sandstones and limestones: Relationship to permeability and porosity, J. Contam. Hydrol. 53, 85--100. Broholm, M., Christophersen, M., Maier, U., Stenby, E., Hoehener, P., and Kjeldsen, P. (2005), Compositional evolution of the emplaced fuel source in the vadose zone field experiment at airbase Værløse, Denmark, Environ. Sci. Technol. 39, 8251--8263. Chiogna, G., Eberhardt, C., Grathwohl, P., and Rolle, M. (2010), Evidence of compound-dependent hydrodynamic and (hydro) mechanical transverse dispersion with multi-tracer laboratory experiments, Environ. Sci. Technol. 44, 688--693. Chiou, C. T., Peters, L. J., and Freed, V. H. (1979), A physical concept of soil--water equilibria for nonionic organic compounds, Science 206, 831--832. Chiou, C. T., Porter, P. E., and Schmedding, D. W. (1983), Partition equilibria of nonionic organic compounds between soil organic matter and water, Environ. Sci. Technol. 17, 227--231. Chiou, C. T., Schmedding, D. W., and Manes, M. (2005), Improved prediction of octanol-water partition coefficients from liquidsolute water solubilities and molar volumes, Environ. Sci. Technol. 39, 8840--8846. Cho, Y.-M., Ghosh, U., Kennedy, A. J., Grossman, A., Ray, G., Tomaszewski, J. E., Smithenry, D. W., Bridges, T. S., and Luthy, R. G. (2009), Field application of activated carbon amendment for in-situ stabilization of polychlorinated biphenyls in marine sediment, Environ. Sci. Technol. 34, 3815--3823. Cirpka, O. A., Frind, E. O., and Helmig, R. (1999), Numerical simulation of biodegradation controlled by transverse mixing, J. Contam. Hydrol. 40, 159--182.  Ju, Q., Rahman, A., and Grathwohl, Cirpka, O. A., Olsson, A., P. (2006), Determination of transverse dispersion coefficients from reactive plumes lengths, Ground Water 44, 212--221. ¨ . (2004), Sorption of phenanCornelissen, G. and Gustafsson, O threne to environmental black carbon in sediment with and without organic matter and native sorbates, Environ. Sci. Technol. 38, 148--155. Cornelissen, G., van Noort, P. M., Parsons, J. R., and Govers, H. A. J. (1997), Temperature dependence of slow adsorption and desorption kinetics on organic compounds in sediments, Environ. Sci. Technol. 31, 454--460. Currie, J. A. (1960), Gaseous diffusion in porous media. Part I. A non-steady state method. Part II. Dry granular materials, Br. J. Appl. Phys. 12, 314--324. Daly, G. L., Lei, Y. D., Castillo, L. E., Muir, D. C. G., and Wania, F. (2007), Polycyclic aromatic hydrocarbons in Costa Rican air and soil: A tropical/temperate comparison, Atmos. Environ. 41, 7339--7350. Eberhardt, C. and Grathwohl, P. (2002), Time scales of pollutants dissolution from complex organic mixtures: Blobs and pools, J. Contam. Hydrol. 59, 45--66. Endo, S., Grathwohl, P., and Schmidt, T. (2008a), Adsorption or absorption? Insights from molecular probes into modes of sorption by environmental solid matrices, Environ. Sci. Technol. 42, 3989--3995.

229

Endo, S., Grathwohl, P., Haderlein, S., and Schmidt, T. (2008b), Compound-specific factors influencing sorption nonlinearities in organic matter, Environ. Sci. Technol. 42, 5897--5903. Endo, S., Grathwohl, P., Haderlein, S., and Schmidt, T. (2009), Characterization of sorbent properties of soil organic matter and carbonaceous materials using n-alkanes and cycloalkanes as molecular probes, Environ. Sci. Technol. 43, 393--400. Ghosh, U., Talley, J. W. and Luthy, R. G. (2001), Particle-scale investigation of PAH desorption kinetics and thermodynamics from sediment, Environ. Sci. Technol. 35, 3468--3475. Gocht, T., Klemm, O., and Grathwohl, P. (2007a), Atmospheric bulk deposition of polycyclic aromatic hydrocarbons in rural areas of southern Germany, Atmos. Environ. 41, 1315--1327. Gocht, T., Ligouis, B., Hinderer, M., and Grathwohl, P. (2007b), Accumulation of polycyclic aromatic hydrocarbons in rural soils based on mass balances at the catchment scale, Environ. Toxicol. Chem. 26, 591--600. Gocht, T., Steidle, D. K., and Grathwohl, P. (2005), Polyzyklische aromatische Kohlenwasserstoffe in st€adtischen Umweltkompartimenten, Wasser und Abfall 7, 10--15. Goss, K.-U. and Schwarzenbach, R. P. (2001), Linear free energy relationships used to evaluate equilibrium partitioning of organic compounds, Environ. Sci. Technol. 35, 1--9. Grathwohl, P. (1998), Diffusion in Natural Porous Media: Contaminant Transport, Sorption/Desorption and Dissolution Kinetics, Kluwer Academic Publishers, Boston. Grathwohl, P., Klenk, I. D., Eberhardt, C., and Maier, U. (2000), Steady state plumes: Mechanisms of transverse mixing in aquifers, Proc. Contaminated Site Remediation: From Source Zones to Ecosystems Conf., Johnston, C. D., ed. CSRC, Melbourne, Australia, Dec. 4--8, 2000, pp. 459--466. Grathwohl, P., Klenk, I. D., Maier, U., and Reckhorn, S. B. F. (2002), Natural attenuation of volatile hydrocarbons in the unsaturated zone and shallow groundwater plumes: Scenario-specific modeling and laboratory experiments, Proc. Groundwater Quality: Natural and Enhanced Restoration of Groundwater Pollution Conf. Sheffield, UK, Thornton S.F., Oswald, S.E. ed. IAHS Publ. 275, 141--146. Haws, N. W., Ball, W. P., and Bouwer, E. J. (2006), Modeling and interpreting bioavailability of organic contaminant mixtures in subsurface environments, J. Contam. Hydrol. 82, 255--292. Huang, W., Young, T. M., Schlautman, M. A., Yu, H., and Weber Jr., W. J. (1997), A distributed reactivity model for sorption by soils and sediments. 9. General isotherm nonlinearity and applicability of the dual reactive domain model, Environ. Sci. Technol. 31, 1703--1710. Janssen, G., Cirpka, O. A., and van der Zee, S. E. A. T. M. (2006), Stochastic analysis of nonlinear biodegradation in regimes controlled by both chromatographic and dispersive mixing, Water Resour. Res. 42, W01417. Jeong, S., Werth, C. J., Wander, M. W., Kleineidam, S., Ligouis, B., and Grathwohl, P. (2008), The role of condensed carbonaceous materials on the sorption of hydrophobic organic contaminants in subsurface systems, Environ. Sci. Technol. 42, 1458--1464. Johnsen, A. R. and Karlson, U. (2007), Diffuse PAH contamination of surface soils: Environmental occurrence, bioavailability,

230

FATE AND TRANSPORT OF ORGANIC COMPOUNDS IN(TO) THE SUBSURFACE ENVIRONMENT

and microbial degradation, Appl, Microbiol. Biotechnol. 76, 533--543. Johnson, M. D. and WeberJr., W. J. (2001), Rapid prediction of longterm rates of contaminant desorption from soils and sediments, Environ. Sci. Technol. 35, 427--433. Kleineidam, S., Rugner, H., and Grathwohl, P. (2004), Desorption kinetics of phenanthrene in aquifer material lacks hysteresis, Environ. Sci. Technol. 38, 4169--4175. Kleineidam, S., Sch€uth, C., and Grathwohl, P. (2002), Solubilitynormalized combined pore-filling-partitioning sorption isotherms for organic pollutants, Environ. Sci. Technol. 36, 4689--4697. Klenk, I. D. (2000), Transport of Volatile Organic Compounds (VOC’s) from Soil-Gas to Groundwater, Ph.D. dissertation, TGA C55, Center for Applied Geosciences, T€ubingen, Germany. Kurt-Karakus, P., Bidleman, T. F., Staebler, R., and Jones, K. C. (2006), Measurement of DDT fluxes from a historically treated agricultural soil in Canada, Environ. Sci. Technol. 40, 4578--4585. Liedl, R., Valocchi, A. J., Dietrich, P., and Grathwohl, P. (2005), Finiteness of steady-state plumes, Water Resour. Res. 41, W12501. Ligouis, B., Kleineidam, S., Karapanagioti, H. K., Kiem, R., Grathwohl, P., and Niemz, C. (2005), Organic petrology: A new tool to study contaminants in soils and sediments, in Environmental Chemistry. Green Chemistry and Pollutants in Ecosystems, Lichtfouse, E., Dudd, S., and Robert, D., eds., SpringerVerlag, Berlin, pp. 89--98. Liu, S., Liedl, R., and Grathwohl, P. (2009a), Simple analytical solutions for oxygen transfer across the capillary fringe into anaerobic groundwater, Water Resour. Res. 46, W10542, doi:10.1029/2009WR008434. Liu, L., Endo, S., Eberhardt, C., Grathwohl, P., and Schmidt, T. (2009b), Partitioning behavior of polycyclic aromatic hydrocarbons between aged coal tar and water, Environ. Toxicol. Chem. 28, 1578--1584. LUBW (State Institute for Environmental Protection BadenW€urttemberg) (2003), Contaminants in Arable Soils in Baden-W€urttemberg Fertilized with Sewage Sludge, Bodenschutz. 16. Maier, U. and Grathwohl, P. (2005), Natural attenuation in the unsaturated zone and shallow groundwater: Coupled modeling of vapor phase diffusion, biogeochemical processes and transport across the capillary fringe, in Reactive Transport in Soil and Groundwater, N€utzmann, G., Viotti, P., and Aagard, P., eds., Springer-Verlag, pp. 141--155. Maier, U. and Grathwohl, P. (2006), Numerical experiments and field results on the size of steady state plumes, J. Contam. Hydrol. 85, 33--52. Maier, U., R€ugner, H., and Grathwohl, P. (2005), Steady state plumes: Transverse mixing in aquifers, numerical experiments and field scale predictions for natural attenuation, Proc. Groundwater Quality: Bringing Groundwater Quality Research to the Watershed Scale, Conf. Waterloo, Canada, Thomson N.R., ed. IAHS Publ. 297, 296--304.

Maier, U., R€ ugner, H., and Grathwohl, P. (2007), Gradients controlling natural attenuation of ammonium, Appl. Geochem. 22, 2606--2617. Manes, M. (1998), Activated carbon adsorption fundamentals, in Encyclopedia of Environmental Analysis and Remediation, R. A., Meyers, ed., Wiley, New York, pp. 26--67. Mayer, K. U., Frind, E. O., and Blowes, D. W. (2002), Multicomponent reactive transport modeling in variably saturated porous media using a generalized formulation for kinetically controlled reactions, Water Resour. Res. 38, 1174--1195. Millington, R. J. and Quirk, J. P. (1960), Transport in Porous Media, Proc. 7th Int. Congress of Soil Science, Madison, WI, pp 97--106. Mittal, M. and Rockne, K. J. (2008), Naphthalene and phenanthrene sorption to very low organic content diatomaceous earth: Modeling implications for microbial bioavailability, Chemosphere 74, 1134--1144. Moldrup, P., Oleson, T., Schjønning, P., Yamaguchi, T., and Rolston, D. E. (2000), Predicting the gas diffusion coefficient in repacked soil: Water-induced linear reduction model, Soil Sci. Soc. Am. J. 64, 1588--1594. Molins, S. and Mayer, K. U. (2007), Coupling between geochemical reactions and multicomponent gas and solute transport in unsaturated media: A reactive transport modeling study, Water Resour. Res. 43, W05435. Nguyen, T. H., Goss, K. U., and Ball, W. P. (2005), Polyparameter linear free energy relationships for estimating the equilibrium partition of organic compounds between water and the natural organic matter in soils and sediments, Environ. Sci. Technol. 39, 913--924. Olsson, A. H. and Grathwohl, P. (2007), Transverse dispersion of non-reactive tracers in porous media: A new nonlinear relationship to predict dispersion coefficients, J. Contam. Hydrol. 92, 149--161. Pan, B. and Xing, B. (2009), Adsorption mechanisms of organic chemicals on carbon nanotubes, Environ. Sci. Technol. 42, 9005--9013. Pasteris, G., Werner, D., Kaufmann, K., and H€ ohener, P. (2001), Vapor phase transport and biodegradation of volatile fuel compounds in the unsaturated zone: a large scale lysimeter study, Environ. Sci. Technol. 36, 30--39. Prommer, H., Barry, D. A., and Davis, G. B. (1998), The effect of seasonal variability on intrinsic biodegradation of a BTEX plume, Proc. Groundwater Quality: Remediation and Protection Conf. (GQ’98), T€ ubingen, Germany, Sept. 21--25 1998, Herbert, M. and Kovar, K., eds., IAHS Publ. 250, 213--220. Prommer, H., Barry, D. A., and Davis, G. B. (2002), Modelling of physical and reactive processes during biodegradation of a hydrocarbon plume under transient groundwater flow conditions, J. Contam. Hydrol. 59, 113--131. Razzaque, M. M. and Grathwohl, P. (2008), Predicting organic carbon-water partitioning of hydrophobic organic chemicals in soils and sediments based on water solubility, Water Research 42, 3775--3780.

REFERENCES

Reckhorn, F. S. B., Zuquette, L. V., and Grathwohl, P. (2001), Experimental investigations of oxygenated gasoline dissolution, J. Environ. Eng. 127, 208--216. Rolle, M., Eberhardt, C., Chiogna, G., Cirpka, O. A., and Grathwohl, P. (2009), Enhancement of dilution and transverse reactive mixing in porous media: Experiments and model-based interpretation, J. Contam. Hydrol. 110, 130--142. Rolle, M., Maier, U., and Grathwohl, P. (2010), Contaminant fate and reactive transport in groundwater, in Dealing with Contaminated Sites (from Theory Towards Practical Application), Swartjes, F., ed., Springer, Chap. 19 (in press). R€ ugner, H., Holder, T., Maier, U., Bayer-Raich, M., Grathwohl, P., and Teutsch, G. (2004), Natural attenuation untersuchungen, ehemalige Abfalldeponie Osterhofen, Grundwasser 9, 98--108. R€ ugner, H., Kleineidam, S., and Grathwohl, P. (1999), Long-term sorption kinetics of phenanthrene in aquifer materials, Environ. Sci. Technol. 33, 1645--1651. Sallam, A., Jury, W. A., and Letey, J. (1984), Measurements of gas diffusion coefficient under relatively low air-filled porosity, Soil Sci. Soc. Am. J. 48, 3--6. Schirmer, M., Durrant, G. C., Molson, J. W., and Frind, E. O. (2001), Influence of transient flow on contaminant biodegradation, Ground Water 39, 276--293. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (2002), Environmental Organic Chemistry, Wiley-Interscience, New York. Simonich, S. L. and Hites, R. A. (1995), Global distribution of persistent organochlorine compounds, Science 269, 1851--1854. Steinberg, S. M., Pignatello, J. J., and Sawhney, B. L. (1987), Persistence of 1,2-dibromethane in soils: Entrapment in intraparticle micropores, Environ. Sci. Technol. 21, 1201--1208. Wang, G., Reckhorn, S. B. F., and Grathwohl, P. (2003), Volatilization of VOC from multicomponent mixtures in unsaturated porous media, Vadose Zone J. 2, 692--701.

231

Wang, G., Kleineidam, S., and Grathwohl, P. (2007), Sorption/ desorption reversibility of phenanthrene in soils and carbonaceous materials, Environ. Sci. Technol. 41, 1186-1193. Wang, G., Kleineidam, S., and Grathwohl, P. (2009), Activation energies of phenanthrene desorption from carbonaceous materials: Column studies, J. Hydrol. 369, 234--240. Weber, Jr., W. J., McGinley, P. M., and Katz, L. E. (1992), A distributed reactivity model for sorption by soils and sediments. 1. Conceptual basis and equilibrium assessments, Environ. Sci. Technol. 26, 1955--1962. Werner, D., Grathwohl, P., and H€ ohener, P. (2004), Review of field methods for the determination of the tortuosity and effective gasphase diffusivity in the vadose zone, Vadose Zone J. 3, 1240--1248. Werth, C. J., Cirpka, O. A., and Grathwohl P. (2006), Enhanced mixing and reaction through flow focusing in heterogeneous porous media, Water Resour. Res. 42, W12414.1--W12414.10. Wilcke, W. (2007), Global patterns of polycyclic aromatic hydrocarbons (PAHs) in soil, Geoderma 141, 157--166. Xia, G. and Ball, W. P. (1999), Adsorption-partitioning uptake of nine low-polarity organic chemicals on a natural sorbent, Environ. Sci. Technol. 33, 262--269. Xing, B. and Pignatello, J. J. (1997), Dual-mode sorption of lowpolarity compounds in glassy poly(vinyl chloride) and soil organic matter, Environ. Sci. Technol. 31, 792--799. Yang, Y., Hofmann, T., Pies, C., and Grathwohl, P. (2008a), Sorption of polycyclic aromatic hydrocarbons (PAHs) to carbonaceous materials in a river floodplain soil, Environ. Pollut. 156, 1357--1363. Yang, Y., Ligouis, B., Pies, C., Grathwohl, P., and Hofmann, T. (2008b), Occurrence of coal and coal-derived particle-bound polycyclic aromatic hydrocarbons (PAHs) in a river floodplain soil, Environ. Pollut. 151, 121--129.

9 PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER DANIEL W. GERRITY, MARK J. BENOTTI, DAVID A. RECKHOW,

AND

SHANE A. SNYDER

9.1. Introduction 9.2. Occurrence of Pharmaceuticals and Endocrine-Disrupting Compounds (EDCs) in Drinking Water 9.3. Pharmaceutical and EDC Removal During Wastewater Treatment 9.3.1. Conventional Wastewater Treatment 9.3.2. Advanced Wastewater Treatment 9.4. Pharmaceutical and EDC Removal During Drinking Water Treatment 9.4.1. Coagulation, Flocculation, and Sedimentation 9.4.2. Activated Carbon Adsorption 9.4.3. Ultraviolet Light (Photolysis) 9.4.4. Free Chlorine/Chloramine 9.4.5. Ozone 9.5. Conclusions

The most common route by which these compounds enter the environment is via wastewater, as they are incompletely removed during wastewater treatment (Ternes 1998; Snyder et al. 2007). Selected pharmaceuticals and EDCs may also enter the environment via other routes, such as urban or agricultural runoff (Standley et al. 2000). Once in the environment, pharmaceuticals and EDC concentrations attenuate by processes such as dilution, adsorption to solids, microbial degradation, photolysis, or other forms of abiotic transformation. For those compounds that are not easily removed, they can persist in drinking water supplies and ultimately contaminate finished drinking water. This chapter outlines some of the issues associated with the presence of pharmaceuticals and EDCs in drinking water. Specifically, this chapter discusses their occurrence in drinking water and their removal or transformation through different wastewater and drinking water treatment processes.

9.1. INTRODUCTION

9.2. OCCURRENCE OF PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS (EDCs) IN DRINKING WATER

Pharmaceuticals and endocrine-disrupting compounds (EDCs) are a structurally diverse class of organic contaminants that have been detected throughout the world in wastewater, surface water, groundwater, estuarine water, and drinking water (Snyder et al. 1999; Ternes et al. 1999; Heberer 2002; Kolpin et al. 2002; Benotti and Brownawell 2007; Barnes et al. 2008; Benotti et al. 2009a). These compounds include, but are not limited to, prescription pharmaceuticals, over-the-counter medications, naturally occurring compounds that elicit a physiological effect (e.g., estrogens), and compounds used in consumables for the benefit of human health and safety (e.g., benzophenone).

The presence of pharmaceuticals and EDCs in the environment and drinking water is not a new topic despite the recent increase in attention. Although these compounds are often referred to as “emerging contaminants,” researchers have detected certain pharmaceuticals and EDCs in wastewater for almost five decades (Stumm-Zollinger and Fair 1965; Tabak and Bunch 1970; Garrison et al. 1976; Hignite and Azarnoff 1977). Moreover, Stumm-Zollinger and Fair (1965) noted that steroid hormones “may occur in drinking water under the most unfavorable conditions.” The more recent

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

233

234

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

TABLE 9.1. Concentrations (in ng/L) of Pharmaceuticals and EDCs in U.S. Source and Finished Drinking Water Source (n ¼ 20) Contaminant Acetaminophen Androstenedione Atrazine Caffeine Carbamazepine DEET Erythromycin Estrone Fluorene Galaxolide Gemfibrozil Hydrocodone Ibuprofen Iopromide Meprobamate Metalochlor Musk ketone Naproxen Oxybenzone Phenytoin Progesterone Sulfamethoxazole TCEP Triclosan Trimethoprim

Finished (n ¼ 20)

MRLa

Maximum

Median

Number

Maximum

Median

Number

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 — 1.0 1.0

9.5 1.9 570 87 39 28 3.5 1.4 13 30 11 1.9 24 46 16 170 17 16 7.4 13 1.1 44 66 30 2.3

1.6 1.9 28 27 3.1 6.9 2.2 1.2 13 28 4.8 1.9 4.2 7.6 5.9 15 16 2.2 2.9 3.2 1.1 8.1 13 1.9 2.2

7 1 17 14 18 20 8 2 1 3 13 1 16 14 16 7 3 10 4 18 1 17 15 6 3

<MRL <MRL 430 83 5.7 30 1.3 2.3 <MRL <MRL 6.5 <MRL 32 31 13 160 17 8 1.1 6.7 1.1 <MRL 19 43 1.3

<MRL <MRL 29 23 2.8 5.1 1.3 1.7 <MRL <MRL 4.2 <MRL 3.8 6.5 3.8 86 17 8 1.1 2.3 1.1 <MRL 5.5 43 1.3

— — 15 12 11 18 1 2 — — 5 — 13 13 15 4 1 1 1 14 2 — 7 1 1

Method reporting limit. Source: Adapted from Snyder et al. (2008a).

boom in environmental pharmaceutical and EDC research can be attributed to improvements in analytical instrumentation rather than pharmaceuticals and EDCs “emerging” as environmental contaminants. For example, it is probable that caffeine has been a surface water contaminant for as long as people have been drinking coffee or tea and ibuprofen for as long as people have been using this pain medication. Regardless, the presence of pharmaceuticals and EDCs in drinking water has received attention from scientists, utilities, the media, and the general public (Daughton and Ternes 1999; Snyder et al. 2003; Kuehn 2008; Donn et al. 2008) because is it unclear what toxicological effects, if any, are posed by the presence of low (ng/L) concentrations of pharmaceuticals and EDCs in drinking water. There is a small but growing understanding of pharmaceutical and EDC occurrence in drinking water. In 1997, a German research group measured ng/L concentrations of the lipid regulator clofibric acid in Berlin tapwater (Heberer and Stan 1997), thereby highlighting the impact that wastewater can have on drinking water quality. This study, among many others, contributed significantly to the more recent attention of pharmaceuticals and EDCs in the environment and in drinking water. Although clofibric acid and other trace

organic compounds have been detected in drinking water, the concentrations are relatively low. In one study, concentrations of five compounds in finished drinking water were found to be less than 10 ng/L (Ternes et al. 2002). Subsequent studies attributed the elimination of pharmaceuticals at German drinking water treatment plants (DWTPs) primarily to ozone oxidation or adsorption onto granular activated carbon (GAC). Stackelberg et al. (2004) documented the occurrence of 106 organic wastewater contaminants, including some pharmaceuticals and EDCs, at different stages of a U.S. DWTP. Eighteen compounds were measured in finished drinking water at concentrations of
258 ng/L. Bruchet et al. (2005) investigated the occurrence of 21 antibiotics and X-ray contrast agents in the Seine River, through groundwater recharge, and in finished drinking water. In this case, only four X-ray contrast agents persisted into finished drinking water but at concentrations of <60 ng/L. Selected antibiotics were measured in the finished water of three U.S. DWTPs at concentrations of
5 ng/L (Ye et al. 2007). A more comprehensive discussion of the occurrence of pharmaceuticals and EDCs in surface water and in drinking water is presented elsewhere (Snyder et al. 2008a).

235

OCCURRENCE OF PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS (EDCs) IN DRINKING WATER

TABLE 9.2. Concentrations (in ng/L) of Pharmaceuticals and EDCs in U.S. Source Water, Finished Drinking Water, and Distribution Systems Source (n ¼ 19)

Finished (n ¼ 18)

Contaminant

MRL

Maximum

Median

Number

Estradiol Ethynylestradiol Atenolol Atorvastatin Atrazine BHT Bisphenol A Butylbenzylphthalate Carbamazepine DEET Diazepam Diclofenac Diethylhexylphthalate Estrone Fluoxetine Galaxolide Gemfibrozil Linuron Meprobamate Metolachlor Naproxen Nonylphenol Norfluoxetine o-Hydroxyatorvastatin Phenytoin p-Hydroxyatorvastatin Progesterone Risperidone Sulfamethoxazole TCEP TCPP Testosterone Triclosan Trimethoprim

0.50 1.0 0.25 0.25 0.25 25 5.0 50 0.50 25 0.25 0.25 120 0.20 0.50 25 0.25 0.50 0.25 10 0.50 80 0.50 0.50 1.0 0.50 0.50 2.5 0.25 50 50 0.50 1.0 0.25

17 1.4 36 1.4 870 49 14 54 51 110 0.47 1.2 170 0.90 3.0 48 24 9.3 73 81 32 130 <MRL 1.2 29 2.0 3.1 <MRL 110 530 720 1.2 6.4 11

17 1.4 2.3 0.80 32 49 6.1 53 4.1 85 0.43 1.1 150 0.30 0.80 3 2.2 4.1 8.2 17 0.90 100 <MRL 0.70 5.1 1.0 2.2 <MRL 12 120 180 1.1 3.0 0.80

1 1 12 3 15 1 3 2 15 6 2 4 2 15 3 4 11 5 16 7 11 8 — 3 14 3 4 — 17 10 8 2 6 11

Distribution (n ¼ 15)

Maximum

Median

Number

<MRL <MRL 18 <MRL 870 26 25 <MRL 18 93 0.33 <MRL <MRL <MRL 0.82 33 2.1 6.2 42 27 <MRL 100 <MRL <MRL 19 <MRL 0.57 <MRL 3.0 470 510 <MRL 1.2 <MRL

<MRL <MRL 1.2 <MRL 49 26 25 <MRL 6.0 63 0.33 <MRL <MRL <MRL 0.71 31 0.48 6.1 5.7 16 <MRL 93 <MRL <MRL 6.2 <MRL 0.57 <MRL 0.39 120 210 <MRL 1.2 <MRL

— — 8 — 15 1 1 — 8 6 1 — — — 2 2 7 2 14 6 — 2 — — 10 — 1 — 4 7 5 — 1 —

Maximum

Median

Number

<MRL <MRL 0.84 <MRL 930 <MRL <MRL <MRL 10 63 <MRL <MRL <MRL <MRL 0.64 <MRL 1.2 <MRL 40 22 <MRL 110 0.77 <MRL 16 <MRL <MRL 2.9 0.32 200 240 <MRL <MRL <MRL

<MRL <MRL 0.47 <MRL 50 <MRL <MRL <MRL 6.8 49 <MRL <MRL <MRL <MRL 0.64 <MRL 0.43 <MRL 5.2 18 <MRL 97 0.77 <MRL 3.6 <MRL <MRL 2.9 0.32 150 220 <MRL <MRL <MRL

— — 8 — 12 — — — 6 4 — — — — 1 — 4 — 11 3 — 2 1 — 10 — — 1 1 6 6 — — —

Source: Adapted from Benotti et al. (2009a).

Snyder et al. (2008a) and Benotti et al. (2009a) present a more comprehensive reconnaissance of pharmaceuticals and EDCs in U.S. drinking water, although both studies targeted DWTPs susceptible to wastewater contamination, and therefore resultant detection frequencies and concentrations do not necessarily represent typical concentrations in drinking waters across the United States. Snyder et al. (2008a) surveyed the occurrence of 36 pharmaceuticals and EDCs in the source and finished drinking water of 20 U.S. DWTPs (Table 9.1). The 12 compounds that were detected in at least half of the source water samples were atrazine, caffeine, carbamazepine, DEET, gemfibrozil, ibuprofen, iopromide, meprobamate, naproxen, phenytoin, sulfamethoxazole, and TCEP. Median concentrations of detected pharmaceuticals and EDCs in source drinking water were

low. Median concentrations of these compounds were usually less than 10 ng/L, except for atrazine (28 ng/L), caffeine (27 ng/L), fluorene (13 ng/L), galaxolide (28 ng/L), metalochlor (15 ng/L), musk ketone (16 ng/L), and TCEP (13 ng/L), although values for fluorene, galaxolide, and musk ketone were biased by low frequencies of detection. The eight compounds that were detected in at least half of the finished drinking water samples were atrazine, caffeine, carbamazepine, DEET, ibuprofen, iopromide, meprobamate, and phenytoin. Median concentrations of detected pharmaceuticals and EDCs in finished drinking water were also low. Median concentrations were usually less than 10 ng/L, except for atrazine (29 ng/L), caffeine (23 ng/L), metalochor (86 ng/L), and triclosan (43 ng/L), although values for metalochlor and triclosan were biased by low frequencies of detection.

236

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

Benotti et al. (2009a) surveyed the occurrence of 51 pharmaceuticals and EDCs in 19 source waters, 18 finished drinking waters, and 15 distribution systems from utilities throughout the United States (Table 9.2). The 11 compounds that were detected in at least half of source water samples were atenolol, atrazine, carbamazepine, estrone, gemfibrozil, meprobamate, naproxen, phenytoin, sulfamethoxazole, TCEP, and trimethoprim. As with the previous study, median concentrations of detected pharmaceuticals and EDCs in source water were low. Median concentration were usually less than 10 ng/L, except for atrazine (32 ng/L), butylbenzyl phthalate (53 ng/L), BHT (49 ng/L), diehtylhexylphthalate (150 ng/L), DEET (85 ng/L), estradiol (17 ng/L), metolachlor (17 ng/L), nonylphenol (100 ng/L), sulfamethoxazole (12 ng/L), TCEP (120 ng/L), and TCPP (180 ng/L), although values for estradiol, BHT, butylbenzylphthalate, and diehtylhexylphthalate were biased by low frequencies of detection. Only three compounds (atrazine, meprobamate, and phenytoin) were detected in at least half of the finished drinking water samples. Median concentrations of detected pharmaceuticals and EDCs in finished drinking water were generally less than 10 ng/L, except for atrazine (49 ng/L), bisphenol A (25 ng/L), galaxolide (31 ng/L), nonylphenol (93 ng/L), BHT (26 ng/L), metolachlor (16 ng/L), DEET (63 ng/L), TCEP (120 ng/L), and TCPP (210 ng/L). Some of these median concentrations were biased by low frequencies of detection. In both of these studies, the occurrence of pharmaceuticals and EDCs in the finished drinking waters was governed by (1) pharmaceutical and EDC occurrence in source water and (2) removal during treatment by either chlorine or ozone (see following sections discussing chlorine and ozone treatment). Finally, the four compounds that were detected in at least half of the sampled distribution systems were atrazine, atenolol, meprobamate, and phenytoin. Median concentrations of detected pharmaceuticals and EDCs in distribution systems were generally less than 10 ng/L, except for atrazine (50 ng/L), DEET (49 ng/L), metolachlor (18 ng/L), nonylphenol (97 ng/L), TCEP (150 ng/L), and TCPP (220 ng/L), although values for metolachlor and nonylphenol were biased by low frequencies of detection.

9.3. PHARMACEUTICAL AND EDC REMOVAL DURING WASTEWATER TREATMENT The presence of pharmaceuticals and EDCs in drinking water is directly related to their presence in wastewater discharge, natural attenuation in the environment, and removal during drinking water treatment. An understanding of pharmaceutical and EDC implications for the drinking water industry requires a comprehensive knowledge of these processes and the biophysicochemical mechanisms that attenuate contaminant concentrations. While a discussion of fate and transport

is outside the scope of this chapter, a discussion of contaminant removal during wastewater and drinking water treatment is provided. Raw wastewater quality varies tremendously depending on the contributing sources (small residential communities, large urban areas, industrial discharge, etc.), and the extent of treatment ultimately depends on the intended use or effluent discharge location. For example, wastewater permitted for ocean discharge does not have the same water quality requirements as that permitted for indirect potable reuse. Conventional wastewater treatment has evolved over time but generally includes the following unit operations and processes (Tchobanoglous et al. 2004): preliminary solids removal, primary gravity settling, secondary biological treatment, filtration, and disinfection. Depending on the specific requirements of the discharge permit, conventional treatment may be supplemented with nutrient removal (i.e., for nitrogen or phosphorus removal) or other advanced processes in order to achieve a higher quality effluent. This may be required for discharge to a sensitive ecosystem (e.g., areas susceptible to algal blooms and eutrophication) or for indirect potable reuse. Advanced treatment may include membrane treatment and advanced oxidation processes (AOPs) such as UV peroxide or ozone peroxide. Traditionally, wastewater treatment trains have not been designed for the removal of trace organic compounds. However, the growing body of occurrence data for wastewater-derived contaminants (including pharmaceuticals and EDCs) in surface waters (Kolpin et al. 2002; Snyder et al. 2008a,b), the recognition that wastewater effluents are impacting natural waters, and the potential adverse effects on aquatic ecosystems (Snyder et al. 2001) have brought these issues to the forefront. Since wastewater discharge is the primary source of pharmaceuticals and EDCs in the environment (Glassmeyer et al. 2005), optimization of wastewater treatment processes may be the most efficient strategy to mitigate the adverse effects of these compounds. The following describes the general efficacy of both conventional and advanced wastewater treatment processes for pharmaceutical and EDC removal. Disinfection will be described in relation to drinking water treatment. 9.3.1. Conventional Wastewater Treatment The effectiveness of secondary wastewater treatment, which involves both adsorption and biotransformation/biodegradation by microorganisms, is highly dependent on the target contaminant (Kim et al. 2007). Biotransformation generally refers to microbial metabolism or cometabolism resulting in organic byproduct formation, whereas biodegradation refers to complete mineralization. Activated-sludge processes, whether in conventional activated-sludge (CAS) configurations or membrane bioreactors (MBRs), may achieve high

PHARMACEUTICAL AND EDC REMOVAL DURING WASTEWATER TREATMENT

removals (up to 99%) of hormones and certain pharmaceuticals (e.g., acetaminophen and ibuprofen), but biological treatment may be insufficient to remove other compounds (e.g., trimethoprim and carbamazepine) (Kim et al. 2007; Snyder et al. 2008b; Radjenovic et al. 2009). Limited removal efficiencies have been observed for certain antibiotics and antimicrobial compounds, such as erythromycin (10%), sulfamethoxazole (64%), and triclosan (68%) (Kim et al. 2007). It is important to note that MBR systems contain microfiltration (MF) or ultrafiltration (UF) membranes, but it is generally the biological processes that are responsible for pharmaceutical and EDC removal (see Section 9.3.2.1 for a discussion of membrane efficacy). Joss et al. (2005) did not observe any relationships between structural characteristics of the compounds and treatment efficacy, but the study did identify microbial transformation, rather than sludge partitioning, as the dominant mechanism for all of the target compounds. For the most susceptible compounds, CAS and MBRs achieve comparable removals, but the longer solids retention times (SRTs) associated with MBRs provide significant benefits with respect to the removal of recalcitrant compounds (Radjenovic et al. 2009). Membrane bioreactors can be operated with longer SRTs because of their high microbial loads and more concentrated return activated sludge. Conventional activated sludge (CAS) would require excessive return flows in order to achieve comparable SRTs. Clara et al. (2005) observed a positive correlation between pharmaceutical and EDC removal and longer SRTs. For most of the target compounds, a critical SRTof 10 days was observed, but for a small number of compounds (e.g., anticonvulsants), pharmaceutical and EDC removal was poor regardless of SRT. Geographic factors, such as climate, can also influence the removal of pharmaceuticals and EDCs. For example, Ternes et al. (1999) observed 80% to greater than 99% removals of estrogens in a Brazilian wastewater treatment plant (WWTP), but the removal efficiencies of those same compounds were lower (0% – 70%) in a German WWTP. This difference was primarily attributed to the higher water temperature of the Brazilian WWTP. Therefore, the efficacy of biological treatment is dependent on a variety of factors, including the compound of interest, process configuration, operational parameters, and geographic location. Regardless, pharmaceuticals and EDCs are never completely removed, and they are typically detected in secondary WWTP effluent at ng/L to mg/L concentrations (Heberer 2002). 9.3.2. Advanced Wastewater Treatment 9.3.2.1. Membranes. The efficacy of membranes for pharmaceutical and EDC removal varies with membrane pore size. Microfiltration (MF) and ultrafiltration (UF) are generally ineffective alternatives for the removal of trace contaminants (Snyder et al. 2007) because their pore sizes are

237

relatively large and the molecular weight cutoff is approximately 100,000 and 2,000 Da, respectively. Thus, pharmaceuticals and EDCs, which are usually less than 500 Da, have the potential to easily pass through the pores. Indirect pharmaceutical and EDC removal by MF and UF membranes is affected by physiochemical parameters. Hydrophobic compounds adsorbed onto particulates or colloids that will not pass through the membrane pores are readily rejected. Nanofiltration (NF) and reverse-osmosis (RO) membranes have much tighter pores (the molecular weight cutoff for these membranes is approximately 250 and 100 Da, respectively) so pharmaceuticals and EDCs are generally rejected by these membranes. There is an abundance of information pertaining to pharmaceutical and EDC removal by membranes. Although MF/UF treatment is largely ineffective at removing pharmaceuticals and EDCs, the concentrations of these target contaminants are generally below the method reporting limit (MRL; 1.0–10 ng/L) after RO and NF treatment (Kim et al. 2007; Bellona et al. 2008). Snyder et al. (2007) studied a variety of pilot and full-scale membrane processes and reported similar results, which are summarized in Table 9.3. They concluded that hydrophobic compounds with aliphatic substituted aromatic ring structures and high pKa values were removed by low-pressure MF and UF membranes. This can be attributed to adsorption onto larger material that is readily rejected by the membrane or to electrostatic repulsion from the membrane surface. Neutrally charged or hydrophilic compounds were not removed by MF or UF membrane treatment. Effective removal of all pharmaceuticals and EDCs was observed following treatment with NF and RO membranes. 9.3.2.2. Advanced Oxidation Process (AOPs). These processes utilize highly reactive chemical species such as free radicals to oxidize chemical contaminants in water (Singer and Reckhow 1999). The most common AOPs, such as UV peroxide and ozone peroxide, involve hydroxyl radical . ( OH)-dominated reactions (Snyder et al. 2006, 2007). Other AOP technologies, such as UV titanium dioxide photocatalysis and nonthermal plasma (NTP), may be viable alternatives in the future (Benotti et al. 2009b; Gerrity et al. 2009). Although AOPs provide some level of treatment with their base mechanisms (e.g., direct photolysis of chemical contaminants from UV peroxide), the dominant treatment pathway generally involves oxidation by highly reactive, . nonspecific OH. In general, AOPs are very effective treatment technologies for removing pharmaceuticals and EDCs from water, although the processes are usually energy-intensive. Moreover, the process is very fast, given the short-lived and highly . reactive nature of OH. Huber et al. (2003) reported second. order OH rate constants for a suite of pharmaceuticals and EDCs ranging from 3.3  109  9.8  109 M1 s1. Snyder

238

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

TABLE 9.3. Pharmaceutical and EDC Removal by Membrane and MBR Processes Percent Removal Contaminant

MF (3)

UF (5)

UF/MBR (4)

NF (3)

RO (9)

Acetaminophen Androstenedione Atrazine Benzo(a)pyrene Caffeine Carbamazepine DDT DEET Diazepam Diclofenac Dilantin Erythromycin Estradiol Estriol Estrone Ethinyl estradiol Fluorene Fluoxetine Galaxolide Gemfibrozil Hydrocodone Ibuprofen Iopromide Lindane Meprobamate Metolachlor Musk ketone Naproxen Oxybenzone Pentoxifylline Progesterone Sulfamethoxazole TCEP Testosterone Triclosan Trimethoprim

<20 <20 NDa ND <20 <20 ND <20 ND <20 <20 <20 <20 ND <20 ND ND 20–50 <20 <20 <20 <20 <20 ND <20 ND <20 <20 <20 <20 ND <20 <20 ND 20–50 <20

<20 20–50 <20 >80 <20 <20 >80 <20 20–50 <20 <20 20–50 20–50 <20 20–50 20–50 >80 >80 20–50 <20 <20 <20 <20 20–50 <20 20–50 20–50 <20 50–80 <20 50–80 20–50 <20 20–50 >80 <20

>80 >80 ND ND >80 20–50 50–80 50–80 <20 <20 <20 20–50 50–80 >80 >80 >80 ND 20–50 ND 20–50 20–50 50–80 <20 ND <20 ND ND >80 >80 >80 >80 20–50 <20 >80 50–80 20–50

20–50 50–80 50–80 >80 50–80 50–80 >80 50–80 50–80 50–80 50–80 >80 50–80 50–80 50–80 50–80 >80 >80 50–80 50–80 50–80 50–80 >80 50–80 50–80 50–80 >80 20–50 >80 50–80 50–80 50–80 50–80 50–80 >80 50–80

>80 >80 ND ND >80 >80 ND >80 >80 >80 >80 >80 >80 >80 >80 >80 ND >80 >80 >80 >80 >80 >80 ND >80 ND >80 >80 >80 >80 >80 >80 >80 ND >80 >80

Notation: In column headings, symbols MF, UF, UF/MBR, NF, and RO (microfiltration, ultrafiltration, ultrafiltration/membrane bioreactor, nanofiltration, and reverse osmosis, respectively) indicate relative pore; numbers in parentheses indicate number of systems tested. a Not detected. Source: Adapted from Snyder et al. (2007).

et al. (2007) reported that treatment with ozone versus ozone peroxide were similar in terms of overall degradation of most pharmaceuticals and EDCs, particularly in wastewater where . ozone is readily converted to OH, but the AOP process yielded faster reaction rates. In drinking water treatment, a limited number of compounds (e.g., clofibric acid and ibuprofen) were not removed by ozone alone (<10% removal), but were effectively removed by ozone peroxide (>90% removal). For UV peroxide, pharmaceutical and EDC removal was generally not attributed to direct photolysis. The . addition of peroxide was necessary to generate OH, which

were responsible for oxidation of trace contaminants. Rosenfeldt and Linden (2004) reported small reductions in EDC concentrations with a UV dose of 1000 mJ/cm2, but those EDCs were removed by more than 90% with the same UV dose and 15 mg/L hydrogen peroxide. The authors also calculated second-order rate constants on the order of 1010 M1 s1. Snyder et al. (2007) reported greater than 80% removal for 19 of 29 pharmaceuticals and EDCs following UV peroxide treatment (372 mJ/cm2 and 5 mg/L hydrogen peroxide). Of the remaining 10 compounds, 8 were between 50% and 80% removed, and only meprobamate and

PHARMACEUTICAL AND EDC REMOVAL DURING WASTEWATER TREATMENT

TCEP, which are both highly resistant to oxidation, were less than 50% removed. . Because of the highly reactive nature of OH, scavengers such as organic matter and alkalinity reduce the efficacy of AOPs (Beltran 2004; Rosario-Ortiz et al. 2008; Wert et al. 2009). UV peroxide is also affected by water with high turbidity and high levels of UV absorbance, both of which reduce UV transmissivity. Therefore, it is important to understand the target water matrix when selecting the most appropriate AOP. The UV peroxide and ozone peroxide AOPs also require peroxide addition and subsequent quenching, which is a significant cost over the life of the system. Ozone and other alternative AOP treatment technologies are not limited by these issues. For example, UV titanium . dioxide photocatalysis, which generates OH by irradiating a titanium dioxide slurry with UV light, and nonthermal plas. ma, which generates UV light, ozone, and OH with highvoltage pulses across two electrodes, are not limited by lightattenuating substances. Additionally, these processes do not require peroxide, so chemical addition and quenching is not necessary; peroxide may increase reaction rates, however. Benotti et al. (2009b) and Gerrity et al. (2009) evaluated the degradation of a suite of pharmaceuticals and EDCs in surface waters with direct UV photolysis, UV peroxide, UV titanium dioxide photocatalysis, and nonthermal plasma. Table 9.4 provides a summary of the electrical energy per order (EEO) of magnitude destruction values for each of the processes. EEO values are a basis of comparison for many treatment options as they standardize energy consumption to the volume of water treated and the extent of treatment (i.e., kWh/m3 per log contaminant removal). Results from these studies indicate that of these four AOP technologies, UV peroxide is the most efficient process, although UV TiO2 photocatalysis and nonthermal plasma provide viable, chemical-free alternatives. 9.3.2.3. Residual Management. Advanced water and wastewater treatment technologies, such as AOPs and NF/

TABLE 9.4. Summary of AOP EEO Values (kWh/m3 per log contaminant removal) for Seven Pharmaceuticals and EDCs Contaminant

UVa

UV Peroxidea,b

UV Titanium Dioxidea,c

NTPd

Atenolol Atrazine Carbamazepine Dilantin Meprobamate Primidone Trimethoprim

1.4 3.3 2.3 2.1 6.6 3.7 0.8

0.5 1.2 0.4 1.0 1.0 0.6 0.4

2.0 4.7 2.1 2.2 6.8 3.9 1.5

1.0 3.7 <0.7 2.0 3.5 2.2 <0.7

a

Benotti et al. (2009b). UV peroxide with 10 mg/L of peroxide. c UV TiO2 with approximately 500 mg/L of TiO2. d Nonthermal plasma (Gerrity et al. 2009). b

239

RO membranes, are particularly effective for removing pharmaceuticals and EDCs. However, the viability of these processes is tempered by residual management issues, including transformation products or the discharge of concentrated brine streams. With any type of oxidation process, including more conventional forms such as chlorination and ozonation, it is impractical to achieve complete mineralization (i.e., conversion of organic molecules to water, mineral acids, and carbon dioxide). Short of complete mineralization, oxidation processes will convert target compounds into transformation products that may or may not bear toxicological significance. For example, Vanderford et al. (2008) studied the chlorination of the antimicrobial compound triclosan and noted the formation of mono- and dichlorinated byproducts within minutes of chlorine addition. Furthermore, Canosa et al. (2005) noted that the chlorinated byproducts of triclosan are more toxic than the parent compound. Researchers have begun to develop an understanding of the reaction pathways and/or transformation products that are produced following advanced treatment of waters containing pharmaceuticals and EDCs. For example, the . OH-induced destruction of several compounds or classes of compounds, including DEET (Song et al. 2009), fibrate pharmaceuticals (Razavi et al. 2009), fluoroquinolone antibiotics (Santoke et al. 2009), and b-lactam antibiotics (Song et al. 2008) has been documented. Research is currently underway to develop models that can predict these transformation products, as well as their reactivity and toxicity (Lei and Snyder 2007). There is a balance that must be achieved between oxidation of trace organic compounds, including pharmaceuticals and EDCs, and the formation of transformation products. It is possible that some transformation products may carry toxicological significance, thereby requiring utilities to (1) avoid their formation or (2) implement additional treatment to remove them or further convert them into a benign form. Membrane treatment is also affected by residual management issues. Considering that reverse osmosis typically requires a pressure of 500 psi (lb/in2), it is evident that a substantial amount of energy is required to drive these processes. As the membranes foul, additional energy is required to maintain sufficient water production, and periodic chemical treatments may be required to clean the membranes. Given that typical RO membrane systems generally recover 80%–90% of water (Crittenden et al. 2005), 10%–20% of the water coming into the RO system is rejected. Thus, a system producing 8–9 MGD (million gallons per day) of treated waters would also generate 1–2 MGD of a concentrated waste stream (commonly referred to as the “brine” stream). In addition to the loss of this water volume, the brine stream contains five- to ten-fold greater concentrations of pharmaceuticals and EDCs, salts, organic matter, and other contaminants. Capital costs, operational costs, and responsible disposal of the brine stream are the

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

major limitations facing widespread use of RO membranes for wastewater treatment and water reuse. At present, disposal of brine streams is often restricted to coastal environments or to other wastewater treatment plants, limiting areas in which this technology can be implemented.

9.4. PHARMACEUTICAL AND EDC REMOVAL DURING DRINKING WATER TREATMENT Despite the reductions in pharmaceutical and EDC concentrations during conventional and advanced wastewater treatment, absolute removal of these target compounds is generally not practical. Although additional reductions in concentration will be achieved through natural attenuation in the environment (e.g., dilution, photolysis, biotransformation/ biodegradation, sorption to sediments), many pharmaceuticals and EDCs persist in the environment for great distances away from the point of wastewater discharge. Thus, a better understanding of their removal through different drinking water treatment processes is necessary, as these compounds may be present at minute concentrations (e.g., low level ng/L) in drinking water sources. The following section discusses pharmaceutical and EDC removal by conventional drinking water treatment processes. 9.4.1. Coagulation, Flocculation, and Sedimentation Coagulation involves the addition of treatment chemicals such as aluminum sulfate [alum, Al2(SO4)3], ferric chloride (FeCl3), and ferric sulfate [Fe2(SO4)3] to promote the destabilization of small suspended particles and colloidal material (Letterman and Amirtharajah 1999). During this process the metal salts hydrolyze, form complexes with organic solutes, and ultimately precipitate as amorphous metal hydroxides. Conventional coagulation was originally intended for turbidity removal via destabilization of existing particles. Modern coagulation processes are also intended for removal of dissolved organic compounds, frequently employing higher coagulant doses and/or pH reduction. A slower mixing condition known as flocculation follows the rapid mix phase and is often used to promote the aggregation of smaller particulates and organic matter into larger settleable flocs. These can be removed by granular media filtration either with or without prior gravity settling or dissolved air flotation (Cleasby and Logsdon 1999; Gregory and Zebel 1999). Westerhoff et al. (2007) evaluated the efficacy of alum and ferric chloride coagulation for pharmaceutical and EDC removal in bench-scale experiments. In four different surface waters, 34 of the 36 pharmaceuticals and EDCs were removed by less than 15%. The two remaining compounds [DDT and benzo(a)pyrene] were removed by 31% and 70%, respectively, due to their greater hydrophobicity [log

100

Average Percent Removal (%)

240

10

1 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00

log Kow

Figure 9.1. Correlation between log KOW and contaminant removal by coagulation [Adapted from Snyder et al. 2007].

octanol–water partition coefficients (log KOW) > 6.0]. Results from the bench-scale tests suggested that (1) removal efficiency and KOW were linearly correlated (Fig. 9.1), (2) there was no added benefit with enhanced coagulation, and (3) removal efficiencies were similar between the two coagulants. Coagulation was also deemed ineffective for pharmaceutical and EDC removal in another study (Ternes et al. 2002), which noted no significant difference in pharmaceutical concentrations before and after coagulation. A summary of these results is provided in Table 9.5. 9.4.2. Activated Carbon Adsorption Activated carbon has typically been used for control of taste and odor problems, although its applicability is expanding as a result of changes in disinfection byproduct (DBP) regulation and the ability for activated carbon to remove DBP precursors (Crittenden et al. 2005). The two main forms of activated carbon are utilized in different ways. Application of powder activated carbon (PAC) is similar to that of a coagulation process and can be used as an additional coagulant during seasonal contaminant spikes. Granular activated carbon (GAC) requires permanent contactors configured as media filters or fixed-bed adsorbers, which can also allow for microbial growth and significant biotransformation/biodegradation. As with coagulation or any adsorption process, the efficacy of PAC and GAC is highly dependent on the hydrophobicity and size of the target compounds. Many researchers have reported on the efficacy of GAC and PAC for the removal of pharmaceuticals and EDCs (Ternes et al. 2002; Snyder et al. 2007; Westerhoff et al. 2007). Either GAC or PAC treatment for trace contaminants can be hindered by the presence of high concentrations of NOM, as they compete for the same adsorption sites on the substrate. Thus, the effectiveness and lifespan of activated

PHARMACEUTICAL AND EDC REMOVAL DURING DRINKING WATER TREATMENT

241

TABLE 9.5. Pharmaceutical and EDC Removal by Coagulationa <20% Removal

20%–50% Removal

50%–80% Removal

Acetaminophen Androstenedione Atrazine Caffeine Carbamazepine DEET Diazepam Diclofenac Dilantin Erythromycin Estradiol Estriol Estrone Ethinyl estradiol Fluorene Fluoxetine Galaxolide Gemfibrozil Hydrocodone Ibuprofen Iopromide Lindane Meprobamate Metolachlor Musk ketone Naproxen Oxybenzone Pentoxifylline Progesterone Sulfamethoxazole TCEP Testosterone Triclosan Trimethoprim

DDT

Benzo(a)pyrene

a

>80% Removal

Rate: 10 mg alum per mg total organic carbon or equivalent dose ([Fe3 þ ]/[Al3 þ ] ¼ 1) of FeCl3.

Source: Adapted from Snyder et al. (2007).

carbon is highly dependent on the characteristics of the target water matrix (Snyder et al. 2007). In contrast to coagulation, the overall octanol–water partition coefficient (DOW), rather than KOW, is a better indicator of performance for many of the compounds (Snyder et al. 2007). Removal can be correlated with KOW for neutral compounds, however. In general, increasing PAC concentrations lead to increased removal of most pharmaceuticals and EDCs. Protonated bases are very susceptible to removal by PAC, because they are electrostatically attracted to negatively charged moieties on the substrate’s surface. Conversely, deprotonated acids are electrostatically repelled from the surface-bound negatively charged moieties and do not adsorb. The removal of neutrally charged molecules is controlled by the hydrophobicity of a particular compound given that the mechanism for adsorption is hydrophobic exclusion from the aqueous phase;

compounds with low KOW values are less likely to adsorb to activated carbon (Westerhoff et al. 2007). Adsorption of target contaminants is often modeled with batch isotherm testing and the Freundlich isotherm model, as described below (Crittenden et al. 2005) 1=n

qA ¼ KA CA

ð9:1Þ

where qA ¼ equilibrium adsorbent-phase concentration of contaminant (mg contaminant/g adsorbent), KA ¼ Freundlich adsorption capacity parameter [(mg/g)(L/mg)1/n], CA ¼ equilibrium concentration of contaminant in solution (mg/L1), and n ¼ Freundlich adsorption intensity parameter (unitless). Empirical determination of the KA and n parameters allows engineers to calculate the expected removals of certain compounds in addition to design criteria specific to

242

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

TABLE 9.6. Freundlich Parameters for Several Trace Organic Compounds Deionized Water Contaminant Bezafibrate Carbamazepine Clofibric acid Diclofenac

Groundwater

N

KA

n

KA

0.19 0.38 0.25 0.19

141 430 71 141

0.22 0.22 0.54 0.21

77 90 63 36

Source: Adapted from Ternes et al. (2002).

the activated-carbon reactor. Ternes et al. (2002) published KA and n values for four pharmaceuticals in deionized water and groundwater (Table 9.6). With respect to general removal trends, Table 9.7 categorizes pharmaceutical and EDC removal for 5 mg/L and 4–5 h of contact time with PAC (Snyder et al. 2007). In contrast to coagulation, only two of the target compounds are removed by less than 20% with PAC, and a majority of the compounds are removed by more than 50%. Activated carbon is generally superior to coagulation, but there are still compounds that are resistant to removal. Again, increased removal of target contaminants must be balanced with the additional operational costs (infrastructure, virgin material, regeneration, disposal, etc.) associated with PAC and GAC. 9.4.3. Ultraviolet Light (Photolysis) Treatment of water with ultraviolet (UV) light has become more common since the discovery in the late 1990s that it is

highly effective for Cryptosporidium inactivation. Although typical disinfection doses are in the range of 20–100 mJ/cm2, much higher doses (e.g., 500–1000 mJ/cm2) are usually employed for contaminant oxidation. Most UV reactors can be divided into two categories according to lamp characteristics and resulting output: (1) monochromatic/low pressure and (2) polychromatic/medium pressure. Both types of lamps contain mercury gas that emits ultraviolet light when excited by electrons. Low-pressure lamps produce a monochromatic output at 254 nm, which is extremely effective for UV disinfection, and medium-pressure bulbs produce a polychromatic output at a higher intensity that induces reactions in a broader range of compounds. Both types of reactors are susceptible to fouling due to the lower solubility of many natural constituents of water (e.g., CaCO3) at higher temperatures found at the surface of the bulb. High turbidity and high levels of organic matter also reduce the effectiveness of photolysis. Photolysis can modify or destroy organic contaminants by direct bond cleavage. However, the extent of photolysis at typical UV disinfection doses is quite small, and thus UV disinfection is not a viable option for controlling most pharmaceuticals and EDCs (Snyder et al. 2007). In benchtop and pilot-scale experiments, only 4 of the 29 detected compounds were degraded by more than 20% (Table 9.8). Direct photolysis using higher UV doses provided significantly increased removals (Table 9.9), and the addition of peroxide provided further improvements to the process (Snyder et al. 2007). Structural properties of individual compounds play a role in how effectively a compound may be removed by photolysis. For example, aromatic compounds absorb light in the UV spectrum. Therefore, compounds with aromatic centers

TABLE 9.7. Pharmaceutical and EDC Removal by PAC Adsorptiona <20% Removal

20%–50% Removal

50%–80% Removal

>80% Removal

Ibuprofen Iopromide

DEET Diclofenac Dilantin Erythromycin Estriol Gemfibrozil Meprobamate Metolachlor Naproxen Sulfamethoxazole TCEP

Acetaminophen Androstenedione Atrazine Caffeine Carbamazepine DDT Diazepam Estradiol Estrone Ethinyl estradiol Galaxolide Hydrocodone Lindane Musk ketone Pentoxifylline Testosterone Trimethoprim

Benzo(a)pyrene Fluorene Fluoxetine Oxybenzone Progesterone Triclosan

PAC dose ¼ 5 mg/L and contact time ¼ 4–5 h. Source: Adapted from Snyder et al. (2007). a

PHARMACEUTICAL AND EDC REMOVAL DURING DRINKING WATER TREATMENT

243

TABLE 9.8. Pharmaceutical and EDC Removal by Medium-Pressure UV Photolysis (40 mJ/cm2) <20% Degradation

20%–50% Degradation

50%–80% Degradation

Androstenedione Atrazine Caffeine Carbamazepine DEET Diazepam Dilantin Erythromycin Estradiol Estriol Estrone Ethinyl estradiol Fluoxetine Gemfibrozil Hydrocodone Ibuprofen Iopromide Meprobamate Naproxen Oxybenzone Pentoxifylline Progesterone TCEP Testosterone Trimethoprim

Acetaminophen

Diclofenac Sulfamethoxazole Triclosan

>80% Degradation

Source: Adapted from Snyder et al. (2007).

are more susceptible to photolysis. Of the pharmaceuticals and EDCs investigated, diclofenac, sulfamethoxazole, and triclosan were most susceptible to removal by photolysis. All have absorption spectra that overlap with the wavelengthspecific peaks generated by medium-pressure lamps. Conversely, aliphatic compounds that lack conjugated double bonds and the appropriate absorption bands (e.g., lindane) are very resistant to UV photolysis. Although UV photolysis may be effective at removing some pharmaceuticals and EDCs, it

is seldom viable as a standalone treatment as many compounds have structures that are not amenable to UV photolysis. 9.4.4. Free Chlorine/Chloramine Chlorination is the most common form of disinfection because of its effectiveness against a variety of pathogens and the ease with which a residual can be maintained throughout a

TABLE 9.9. Pharmaceutical and EDC Removal by Medium-Pressure UV Photolysis (450 mJ/cm2) <20% Degradation

20%–50% Degradation

50%–80% Degradation

>80% Degradation

Androstenedione Caffeine DEET Diazepam Meprobamate TCEP

Carbamazepine Gemfibrozil Ibuprofen Pentoxifylline Progesterone Testosterone Trimethoprim

Atrazine Dilantin Erythromycin Iopromide

Acetaminophen Diclofenac Estradiol Estriol Estrone Ethinyl estradiol Fluoxetine Hydrocodone Naproxen Oxybenzone Sulfamethoxazole Triclosan

Source: Adapted from Snyder et al. (2007).

244

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

TABLE 9.10. Pharmaceutical and EDC Oxidation by Chlorination <20% Degradation

20%–50% Degradation

50%–80% Degradation

>80% Degradation

Androstenedione Atrazine Caffeine Carbamazepine DDT DEET Dilantin Fluorene Fluoxetine Ibuprofen Iopromide Lindane Meprobamate Metolachlor Progesterone TCEP Testosterone

Diazepam Galaxolide Pentoxifylline

Gemfibrozil

Acetaminophen Benzo(a)pyrene Diclofenac Erythromycin Estradiol Estriol Estrone Ethinyl estradiol Hydrocodone Musk ketone Naproxen Oxybenzone Sulfamethoxazole Triclosan Trimethoprim

Chlorine concentration ¼ 3 mg/L, contact time ¼ 24 h, pH ¼ 7.9–8.5. Source: Adapted from Snyder et al. (2007).

a

distribution system. However, many utilities are currently turning toward chloramination for residual disinfection (Routt et al. 2007) because of its greater stability in distribution systems and lower potential to form halogenated disinfection byproducts (Hua and Reckhow 2008). The amount of chlorine or chloramine utilized in drinking water applications is usually reported as units of concentration  time (CT). Chlorine and chloramine doses of 3 mg/L for 24 h [CT ¼ (4320 mg min)/L] were evaluated for pharmaceutical and EDC oxidation (Snyder et al. 2007). These results are illustrated in Tables 9.10 and 9.11, respectively. Compounds most susceptible to removal by chlorine or chloramine often contain aromatic structures with electrondonating functional groups (e.g., hydroxyl, amine, and to a lesser extent, methoxy groups) (Reckhow et al. 1990; Deborde and von Gunten 2008). For example, steroid hormones containing phenolic groups were removed by more than 95%. Other compounds susceptible to chlorine or chloramine oxidation may contain primary amines attached to conjugated rings (e.g., trimethoprim, sulfamethoxazole), highly alkylated benzenes (e.g., gemfibrozil, hydrocodone), and polycyclic aromatic hydrocarbons [e.g., carbamazepine, benzo(a)pyrene, diclofenac, naproxen]. The most resistant compounds often lack carbon–carbon double bonds and contain carboxyl groups, ketones, heterocyclic nitrogen, or primary amides (e.g., iopromide, meprobamate) (Snyder et al. 2007). Given that some compounds are resistant to chlorine or chloramine oxidation, complete mineralization is not possible. As with any treatment technology, the potential effects of molecular [e.g., chlorinated triclosan (Vanderford

et al. 2008)] or bulk [e.g., total organic halogens (Hua and Reckhow 2008)] transformation products must be considered. 9.4.5. Ozone The use of ozone for chemical disinfection, although relatively energy-intensive, is highly effective for pathogen inactivation (including Cryptosporidium oocysts). Ozone reacts either directly with organic molecules or indirectly through the formation of radical species (Langlais et al. 1991). For direct reactions, ozone reacts rapidly with amines, phenols, and double bonds in aliphatic compounds. In contrast to photolysis, many pharmaceuticals and EDCs are degraded rapidly (Snyder et al. 2006; Wert et al. 2007, 2009) with ozone CTs commonly used for disinfection applications [(<20 mgmin)/L). Since molecular ozone is very effective for pharmaceutical and EDC treatment, modifying the process with peroxide is not always necessary, although it does increase the reaction rate (Snyder et al. 2007). However, some recalcitrant compounds (e.g., clofibric acid, ibuprofen) may necessitate augmentation with peroxide to achieve higher levels of treatment (Snyder et al. 2007). Table 9.12 lists the relative removal results of a suite of pharmaceuticals and EDCs by ozonation. Huber et al. (2003) calculated second-order rate constants for the ozonation of pharmaceuticals and EDCs (Table 9.13). The authors noted that the aromatic and tertiary amine moieties found in sulfonamide and macrolide antibiotics are reactive with ozone, and all compounds within these classes should have similar reaction rates. Furthermore, the authors indicated

PHARMACEUTICAL AND EDC REMOVAL DURING DRINKING WATER TREATMENT

245

TABLE 9.11. Pharmaceutical and EDC Oxidation by Chloramination <20% Degradation

20%–50% Degradation

50%–80% Degradation

>80% Degradation

Androstenedione Atrazine Caffeine Carbamazepine DDT DEET Diazepam Dilantin Erythromycin Fluorene Fluoxetine Gemfibrozil Ibuprofen Iopromide Lindane Meprobamate Metolachlor Musk ketone Naproxen Pentoxifylline Progesterone Sulfamethoxazole TCEP Testosterone Trimethoprim

Hydrocodone Galaxolide

Benzo(a)pyrene Diclofenac Oxybenzone

Acetaminophen Estradiol Estriol Estrone Ethinyl estradiol Triclosan

a

Chloramine concentration ¼ 3 mg/L, contact time ¼ 24 h, pH ¼ 8.0.

Source: Adapted from Snyder et al. (2007).

TABLE 9.12. Pharmaceutical and EDC Oxidation by Ozonation <20% Degradation

20%–50% Degradation

50%–80% Degradation

>80% Degradation

TCEP

Atrazine Iopromide Meprobamate

DEET Diazepam Dilantin Ibuprofen

Acetaminophen Androstenedione Caffeine Carbamazepine Diclofenac Erythromycin Estradiol Estriol Estrone Ethinyl estradiol Fluoxetine Gemfibrozil Hydrocodone Naproxen Oxybenzone Pentoxifylline Progesterone Sulfamethoxazole Testosterone Triclosan Trimethoprim

a

Ozone concentration ¼ 2.5 mg/L and contact time ¼ 24 min.

Source: Adapted from Snyder et al. (2007).

246

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

TABLE 9.13. Second-Order Rate Constants for Ozonation of Pharmaceuticals and EDCs Contaminant

pKa

kozone, M1 s-1a

Reactive Species

Bezafibrate Carbamazepine Diazepam Diclofenac Ethinyl estradiol Ibuprofen Iopromide Sulfamethoxazole Roxithromycin

3.6 N/A N/A 4.2 10.4 4.9 N/A 5.7 8.8

590  50 3  105 0.75  0.15 1  106 7  109 9.6  1 <0.8 2.5  106 5  106

Dissociated Neutral Neutral Dissociated Dissociated Dissociated Neutral Dissociated Neutral

a

Reaction rates are specific to dominant species commonly found at pH 5–10.

Source: Adapted from Huber et al. (2003).

that ketone-containing steroid hormones are likely to have rate constants that are approximately one order of magnitude less than those of the phenolic steroid hormones. The compounds experiencing the least amount of degradation are generally characterized by extensive branching (e.g., meprobamate, iopromide) and are sometimes designed specifically to resist oxidation (e.g., the flame retardant TCEP). As with chlorine and other oxidation processes, complete mineralization with ozone is impractical given the energy requirement and the potential to form disinfection byproducts (e.g., bromate). Thus, the potential effects of ozone transformation products must be considered.

. . . . . . . . . . . .

9.5. CONCLUSIONS In general, pharmaceuticals and EDCs are detected in source and finished drinking water, albeit at extremely low concentrations (generally <10 ng/L). Their presence in source water stems from their ubiquity in wastewater effluent and their resistance to natural attenuation following environmental discharge. Their occurrence in finished drinking water stems from their concentrations in source drinking water and their resistance to conventional and advanced treatment. Thus, their concentrations likely vary by location. In areas downstream of wastewater discharges, which have been treated with less effective forms of oxidation (e.g., chlorine), one would expect greater frequencies of detection and higher concentrations. In areas far removed from wastewater discharge or areas affected by ozone-treated wastewater effluent, for example, one would expect lower frequencies of detection and lower concentrations. This chapter summarized pharmaceutical and EDC occurrence from two studies of drinking water treatment plants throughout the United States. The most commonly detected compounds in source drinking water were as follows: . .

Atenolol Atrazine

.

Caffeine Carbamazepine Estrone DEET Gemfibrozil Ibuprofen Iopromide Meprobamate Naproxen Phenytoin Sulfamethoxazole TCEP Trimethoprim

The most commonly detected compounds in finished drinking water were as follows: . . . . . . . .

Atrazine Caffeine Carbamazepine DEET Ibuprofen Iopromide Meprobamate Phenytoin

In most cases median concentrations of these compounds were less than 10 ng/L1. Of these compounds, only atrazine exhibited median concentrations of 10–50 ng/L. Median concentrations of some other some compounds were greater than 10 ng/L, although these values were biased by very low frequencies of detection. Ultimately the impact of pharmaceuticals and EDCs in drinking water will be based on whether these compounds pose any risk to human health. Should some of these compounds be regulated, there are wastewater or drinking water

REFERENCES

treatment technologies capable of drastically reducing concentrations of pharmaceuticals and EDCs, although the most effective technologies bear significant costs. For example, the use of NF/RO membranes, ozone, ozone peroxide, and UV peroxide have been shown to be particularly effective, yet costly, treatment options. In fact, these treatment processes lower the concentrations of most pharmaceuticals and EDCs to below current detection limits. However, achieving “total” removal is impossible. As analytical technologies improve, scientists will someday be able to routinely measure picogram per liter—or possibly even femtogram per liter— concentrations. With such analytical capabilities, pharmaceuticals and EDCs will undoubtedly be detected in drinking water even if they are removed to greater than 99.99% by treatment. REFERENCES Barnes, K. K., Kolpin, D. W., Furlong, E. T., Zaugg, S. D., Meyer, M. T., and Barber, L. B. (2008), A national reconnaissance of pharmaceuticals and other organic wastewater contaminants in the United States. I. Groundwater, Sci. Total Environ. 402(2–3), 192–200. Bellona, C., Drewes, J. E., Oelker, G., Luna, J., Filteau, G., and Amy, G. (2008), Comparing nanofiltration and reverse osmosis for drinking water augmentation. J. Am. Water Works Assoc. 100(9), 102–116. Beltran, F. J. (2004), Ozone Reaction Kinetics for Water and Wastewater Systems, Lewis Publishers, CRC Press, Boca Raton, FL. Benotti, M. J. and Brownawell, B. J. (2007), Distributions of pharmaceuticals in an urban estuary during both dry- and wet-weather conditions, Environ. Sci. Technol. 41(16), 5795– 5802. Benotti, M. J., Trenholm, R. A., Vanderford, B. J., Holaday, J. C., Stanford, B. D., and Snyder, S. A. (2009a) , Pharmaceuticals and endocrine disrupting compounds in U.S. drinking water, Environ. Sci. Technol. 43(3), 597–603. Benotti, M. J., Stanford, B. D., Wert, E. C., and Snyder, S. A. (2009b), Evaluation of a photocatalytic reactor membrane pilot system for the removal of pharmaceuticals and endocrine disrupting compounds from water, Water Res. 43, 1513–1522. Bruchet, A., Hochereau, C., Picard, C., Decottignies, V., Rodrigues, J. M., and Janex-Habibi, M. L. (2005), Analysis of drugs and personal care products in French source and drinking waters: The analytical challenge and examples of application, Water Sci. Technol. 52(8), 53–61. Canosa, P., Morales, S., Rodriguez, I., Rubi, E., Cela, R., and Gomez, M. (2005), Aquatic degradation of triclosan and formation of toxic chlorophenols in presence of low concentrations of free chlorine. Anal. Bioanal. Chem. 383(7–8), 1119–1126. Clara, M., Kreuzinger, N., Strenn, B., Gans, O., and Kroiss, H. (2005), The solids retention time—a suitable design parameter

247

to evaluate the capacity of wastewater treatment plants to remove micropollutants, Water Res. 39(1), 97–106. Cleasby, J. L. and Logsdon, G. S. (1999), Granular bed and precoat filtration, in Water Quality and Treatment: A Handbook of Community Water Supplies, Letterman. R. D., ed., McGrawHill, New York. Crittenden, J. C., Trussell, R. R., Hand, D. W., Howe, K. J., and Tchobanoglous, G. (2005), Water Treatment: Principles and Design, 2nd ed., Wiley, Hoboken, NJ. Daughton, C. G. and Ternes, T. A. (1999), Pharmaceuticals and personal care products in the environment: Agents of subtle change? Environ. Health Perspect. 107, 907–938. Deborde, M. and von Gunten, U. (2008), Reactions of chlorine with inorganic and organic compounds during water treatment— kinetics and mechanisms: A critical review, Water Res. 42(1–2), 13–51. Donn, J., Mendoza, M., and Pritchard, J. (2008), Pharmaceuticals Found in Drinking Water, Affecting Wildlife and Maybe Humans, The Associated Press. Garrison, A. W., Pope, J. D., and Allen, F. R. (1976), GC/MS analysis of organic compounds in domestic wastewaters, in Identification and Analysis of Organic Pollutants in Water, Keith, C. H., ed., Ann Arbor Science Publishers, Ann Arbor, MI, pp. 516–556. Gerrity, D., Stanford, B. D., Trenholm, R. A., and Snyder, S. A. (2009), An evaluation of a pilot-scale nonthermal plasma advanced oxidation process for trace organic compound degradation, Water Res. 44, 493–504. Glassmeyer, S. T., Furlong, E. T., Kolpin, D. W., Cahill, J. D., Zaugg, S. D., Werner, S. L., Meyer, M. T., and Kryak, D. D. (2005), Transport of chemical and microbial compounds from known wastewater discharges: Potential for use as indicators of human fecal contamination, Environ. Sci. Technol. 39(14), 5157–5169. Gregory, R. and Zebel, T. F. (1999), Sedimentation and flotation, in Water Quality and Treatment: A Handbook of Community Water Supplies, Letterman, R. D., ed., McGraw-Hill, New York. Heberer, T. and Stan, H. J. (1997), Determination of clofibric acid and N-(phenylsulfonyl)-sarcosine in sewage, river and drinking water, Int. J. Environ. Anal. Chem. 67(1–4), 113–123. Heberer, T. (2002), Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: A review of recent research data, Toxicol. Lett. 131(1–2), 5–17. Hignite, C. and Azarnoff, D. L. (1977), Drugs and drug metabolites as environmental contaminants—chlorophenoxyisobutyrate and salicylic acid in sewage water effluent, Life Sci. 20(2), 337–341. Hua, G. and Reckhow, D. A. (2008), DBP formation during chlorination and chloramination: Effect of reaction time, pH, dosage, and temperature, J. Am. Water Works Assoc. 100(8), 82–95. Huber, M. M., Canonica, S., Park, G. Y., and Von Gunten, U. (2003), Oxidation of pharmaceuticals during ozonation and advanced oxidation processes, Environ. Sci. Technol. 37(5), 1016–1024. Joss, A., Keller, E., Alder, A. C., Gobel, A., McArdell, C. S., Ternes, T., and Siegrist, H. (2005), Removal of pharmaceuticals and fragrances in biological wastewater treatment, Water Res. 39, 3137–3152.

248

PHARMACEUTICALS AND ENDOCRINE-DISRUPTING COMPOUNDS IN DRINKING WATER

Kim, S. D., Cho, J., Kim, I. S., Vanderford, B. J., and Snyder, S. A. (2007), Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters, Water Res. 41(5), 1013–1021. Kolpin, D. W., Furlong, E. T., Meyer, M. T., Thurman, E. M., Zaugg, S. D., Barber, L. B., and Buxton, H. T. (2002), Pharmaceuticals, hormones and other organic wastewater contaminants in US streams, 1999–2000: A national reconnaissance, Environ. Sci. Technol. 36(6), 1202–1211. Kuehn, B. M. (2008), Traces of drugs found in drinking water: Health effects unknown, safer disposal urged, J. Am. Med. Assoc. 299(17), 2011–2013. Langlais, R. T., Reckhow, D. A., and Brink, D. R. (1991), Ozone in Water Treatment: Application and Engineering. Lewis Publishers, Chelsea, MI. Lei, H. X. and Snyder, S. A. (2007), 3D QSPR models for the removal of trace organic contaminants by ozone and free chlorine, Water Res. 41(18), 4051–4060. Letterman, R. D. and Amirtharajah, A. (1999), Coagulation and flocculation, in Water Quality and Treatment: A Handbook of Community Water Supplies, Letterman, R. D., ed., McGraw-Hill, New York. Radjenovic, J., Petrovic, M., and Barcelo, D. (2009), Fate and distribution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Res. 43, 831– 841. Razavi, B., Song, W. H., Cooper, W. J., Greaves, J., and Jeong, J. (2009), Free-radical-induced oxidative and reductive degradation of fibrate pharmaceuticals: Kinetic studies and degradation mechanisms, J. Phys. Chem. A 113(7), 1287–1294. Reckhow, D. A., Singer, P. C., and Malcolm, R. L. (1990), Chlorination of humic materials—by-product formation and chemical interpretations, Environ. Sci. Technol. 24(11), 1655–1664. Rosario-Ortiz, F. L., Mezyk, S. P., Wert, E. C., Doud, D. F. R., Singh, M. K., Xin, M., Baik, S., and Snyder, S. A. (2008), Effect of ozone oxidation on the molecular and kinetic properties of effluent organic matter, J. Adv. Oxid. Technol. 11(3), 529–535. Rosenfeldt, E. J. and Linden, K. G. (2004), Degradation of endocrine disrupting chemicals bisphenol A, ethynyl estradiol, and estradiol during UV photolysis and advanced oxidation processes, Environ. Sci. Technol. 38(20), 5476–5483. Routt, J. C., Mackey, E., Whitby, E., Connell, G., Passantine, L., and Noak, R. (2007), Tracking Utility Disinfection over Four Decades of Change: Disinfection Survey 2007—Preliminary Summary, American Water Works Association (AWWA), Charlotte, NC. Santoke, H., Song, W. H., Cooper, W. J., Greaves, J., and Miller, G. E. (2009), Free-radical-induced oxidative and reductive degradation of fluoroquinolone pharmaceuticals: Kinetic studies and degradation mechanisms, J. Phys. Chem. A 113(27), 7846– 7851. Singer, P. C. and Reckhow, D. A. (1999), Chemical Oxidation, in Water Quality and Treatment: A Handbook of Community Water Supplies, Letterman, R. D., ed., McGraw-Hill, New York.

Snyder, S. A., Keith, T. L., Verbrugge, D. A., Snyder, E. M., Gross, T. S., Kannan, K., and Giesy, J. P. (1999), Analytical methods for detection of selected estrogenic compounds in aqueous mixtures, Environ. Sci. Technol. 33(16), 2814–2820. Snyder, S. A., Villeneuve, D. L., Snyder, E. M., and Giesy, J. P. (2001), Identification and quantification of estrogen receptor agonists in wastewater effluents, Environ. Sci. Technol. 35(18), 3620–3625. Snyder, S. A., Westerhoff, P., Yoon, Y., and Sedlak, D. L. (2003), Pharmaceuticals, personal care products, and endocrine disruptors in water: Implications for the water industry, Environ. Eng. Sci. 20(5), 449–469. Snyder, S. A., Wert, E. C., Rexing, D. J., Zegers, R. E., and Drury, D. D. (2006), Ozone oxidation of endocrine disruptors and pharmaceuticals in surface water and wastewater, Ozone Sci. Eng. 28(6), 445–460. Snyder, S. A., Wert, E. C., Lei, H. X., Westerhoff, P., and Yoon, Y. (2007), Removal of EDCs and Pharmaceuticals in Drinking and Reuse Treatment Processes, AWWA Research Foundation, Denver. Snyder, S. A., Trenholm, R. A., Snyder, E. M., Bruce, G. M., Pleus, R. C., and Hemming, J. D. C. (2008a), Toxicological Relevance of EDCs and Pharmaceuticals in Drinking Water, AWWA Research Foundation, Denver. Snyder, S. A., Vanderford, B. J., Drewes, J. E., Dickenson, E., Snyder, E. M., Bruce, G. M., and Pleus, R. C. (2008b), State of Knowledge of Endocrine Disruptors and Pharmaceuticals in Drinking Water, AWWA Research Foundation, Denver. Song, W. H., Chen, W. S., Cooper, W. J., Greaves, J., and Miller, G. E. (2008), Free-radical destruction of beta-lactam antibiotics in aqueous solution, J. Phys. Chem. A 112(32), 7411–7417. Song, W. H., Cooper, W. J., Peake, B. M., Mezyk, S. P., Nickelsen, M. G., and O’Shea, K. E. (2009), Free-radical-induced oxidative and reductive degradation of N,N0 -diethyl-m-toluamide (DEET): Kinetic studies and degradation pathway, Water Res. 43(3), 635–642. Stackelberg, P. E., Furlong, E. T., Meyer, M. T., Zaugg, S. D., Henderson, A. K., and Reissman, D. B. (2004), Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-water treatment plant, Sci. Total Environ. 329(1–3), 99–113. Standley, L. J., Kaplan, L. A., and Smith, D. (2000), Molecular tracers of organic matter sources to surface water resources, Environ. Sci. Technol. 34(15), 3124–3130. Stumm-Zollinger, E. and Fair, G. M. (1965), Biodegradation of steroid hormones, J. Water Pollut. Control. Fed. 37, 1506–1510. Tabak, H. H. and Bunch, R. L. (1970), Steroid hormones as water pollutants. I. Metabolism of natural and synthetic ovulationinhibiting hormones by microorganisms of activated sludge and primary settled sewage, Dev. Ind. Microbiol. 11, 367–376. Tchobanoglous, G., Burton, F. L., and Stensel, H. D. (2004), Wastewater Engineering: Treatment and Reuse, 4th ed., McGraw-Hill, New York. Ternes, T. A. (1998), Occurrence of drugs in German sewage treatment plants and rivers, Water Res. 32(11), 3245–3260. Ternes, T. A., Stumpf, M., Mueller, J., Haberer, K., Wilken, R. D., and Servos, M. (1999), Behavior and occurrence of estrogens in

REFERENCES

municipal sewage treatment plants. I. Investigations in Germany, Canada and Brazil, Sci. Total Environ. 225(1–2), 81–90. Ternes, T. A., Meisenheimer, M., McDowell, D., Sacher, F., Brauch, H. J., Gulde, B. H., Preuss, G., Wilme, U., and Seibert, N. Z. (2002), Removal of pharmaceuticals during drinking water treatment, Environ. Sci. Technol. 36(17), 3855–3863. Vanderford, B. J., Mawhinney, D. B., Rosario-Ortiz F. L., and Snyder, S. A. (2008), Real-time detection and identification of aqueous chlorine transformation products using QTOF MS, Anal. Chem. 80(11), 4193–4199. Wert, E. C., Rosario-Ortiz, F. L., Drury, D. D., and Snyder, S. A. (2007), Formation of oxidation byproducts from ozonation of wastewater, Water Res. 41(7), 1481–1490.

249

Wert, E. C., Rosario-Ortiz, F. L., and Snyder, S. A. (2009), Effect of ozone exposure on the oxidation of trace organic contaminants in wastewater, Water Res. 43(4), 1005–1014. Westerhoff, P., Mezyk, S. P., Cooper, W. J., and Minakata, D. (2007), Electron pulse radiolysis determination of hydroxyl radical rate constants with Suwannee river fulvic acid and other dissolved organic matter isolates, Environ. Sci. Technol. 41(13), 4640–4646. Ye, Z. Q., Weinberg, H. S., and Meyer, M. T. (2007), Trace analysis of trimethoprim and sulfonamide, macrolide, quinolone, and tetracycline antibiotics in chlorinated drinking water using liquid chromatography electrospray tandem mass spectrometry, Anal. Chem. 79(3), 1135–1144.

10 INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS CLAUDIA MOECKEL

AND

KAVIN C. JONES

10.1. Introduction 10.2. Persistent Organic Pollutants (POPs) and How They Enter the Environment 10.3. Primary and Secondary POP Sources and the Role of Recycling 10.4. Key Intermedia Transfers 10.4.1. Atmospheric Transport and Transfers 10.4.2. POPs in Vegetation and Soil 10.4.3. Exchange of POPs Between Air and Vegetation or Soil 10.4.4. Partitioning of POPs Between Environmental Compartments 10.4.5. Cold Condensation and Global Fractionation 10.5. POPs in the Global Environment: Some Further Observations 10.5.1. Elimiation 10.5.2. Processes Influencing Availability of POPs for Surface–Air Exchange 10.6. POP Modeling in the Environment 10.6.1. Introductory Remarks 10.6.2. Box Models 10.6.3. Transport Models 10.7. Conclusions

environmental chemists. This relates to their multimedia exchanges and persistence—features that give them a truly global distribution. This has led to political and regulatory controls at the international level, intended to reduce emissions and thus environmental concentrations in order to protect human health and the environment. They provide excellent case studies as to why research is needed to better understand and predict (model) the complex behavior of trace substances in our global environment.

10.2. PERSISTENT ORGANIC POLLUTANTS (POPs) AND HOW THEY ENTER THE ENVIRONMENT The term persistent organic pollutants (POPs) refers to a diverse range of compound classes, which meet the following criteria as established by the United Nations Economic Commission for Europe (UNECE) and the United Nations Environment Programme (UNEP) (UNECE 1998; UNEP 2001): . . .

10.1. INTRODUCTION .

This chapter focuses on a general group of chemicals, known as persistent organic pollutants. They are considered from the perspective of their sources and significance in the global environment, and aspects of their environmental fate and behavior that make them particularly interesting to

Pollutants that are longlived in the environment Pollutants that are prone to long-range transboundary atmospheric transport and deposition Pollutants that can bioaccumulate in food chains (i.e., the uptake rate of a toxic substance by organisms exceeds the elimination rate) Pollutants that can cause adverse human health or environmental effects near and distant from their sources

Persistent organic pollutants have been subject to international controls and restrictions on their production and use, and there are regulatory approaches designed to allow other

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

251

252

INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS

TABLE 10.1. UNECE Criteria Defining Organic Chemicals as POPs Property

Criteria Defining Organic Chemicals as POPs

Potential for long-range atmospheric transport (LRAT) Persistence Potential for bioaccumulation Toxicity

Vapor pressure < 1000 Pa or t1/2a in air > 2 days and chemical present in remote regions t1/2 in water > 2 months, t1/2 in soils > 6 months, or t1/2 in sediments > 6 months log KOWb > 5 or BCFc > 5000 Potential of adverse effects on human health and/or the environment

t1/2 ¼ half-life Octanol–water partition coefficient. c Bioconcentration factor. Source: UNECE (1998). a b

compounds with similar traits (see Table 10.1) to be added to the list. The two most important international legally binding treaties addressing POPs are the Aarhus Protocol on POPs as part of the UNECE Convention on Long-Range Transboundary Air Pollution (CLRTAP) and the Stockholm Convention under UNEP. Although the amounts of POPs produced globally are small compared to many other chemicals and some of them had been restricted or even banned decades ago, these chemicals can still be detected in the environment close to sources but also in very remote areas. This is partly a function of the high sensitivity of analytical methods but also because of the high persistence of these chemicals, and the fact that many can exist in the atmosphere in the gas phase. The combination of persistence and semivolatility means that they can undergo repated air–surface exchange, to transfer far from sources. Hence, they can reach the Arctic, for

example, where, because of their tendency to bioaccumulate, they occur at high concentrations in fatty tissues of animals at the end of the food chain such as polar bears (e.g., Bentzen et al. 2008) or birds of prey (Newton et al. 1993) (Fig. 10.1). This is of concern because many POPs have been found to be toxic to the immune and nervous systems, to disrupt the endocrine system, impair the reproductive performance, and cause or promote cancer (Godduhn and Duffy 2003; Howsam et al. 2004; Mariussen and Fonnum 2006). Acute effects at high doses of individual contaminants can be studied relatively easily. However, it is much more difficult to investigate the long-term impact of a complex variety of low-dosed and possibly interacting substances (Godduhn and Duffy 2003) that are of higher relevance for most environmental situations. Hence, to try to protect humans and the environment from risks associated with POPs, the UNECE and the Governing Council of the UNEP decided on priority

Figure 10.1. Global migration processes of POPs [adapted from Wania and Mackay (1996)]. (See insert for color representation of this figure.)

PERSISTENT ORGANIC POLLUTANTS (POPs) AND HOW THEY ENTER THE ENVIRONMENT

253

TABLE 10.2. Chemicals Included in CLRTAP, Banned or Restricted by Stockholm Convention to Date, and Compounds Proposed for Inclusion in Stockholm Convention Compounds Included in Stockholm Conventiona Aldrin Chlordane Chlordecone Dieldrin Dichlorodiphenyltrichloroethane (DDT) Endrin Heptachlor Hexabromobiphenyl Hexachlorobenzene (HCB) a-, b-, c-Hexachlorocyclohexanes (HCHs) Mirex Pentachlorobenzene Perfluorooctane sulfonate (PFOS) Perfluorooctanoic acid (PFOA) Polybrominated diphenyl ethers (PBDEs) Pentabromodiphenyl formulation Octabromodiphenyl formulation Polychlorinated biphenyls (PCBs) Polychlorinated dibenzo-p-dioxins (PCDDs) Polychlorinated dibenzofurans (PCDFs) Toxaphene

Compounds in CLRTAP Protocol Not Included in Stockholm Convention to Dateb

Compounds Proposed for Inclusion in Stockholm Convention by Nongovernmental Organizations

Polycyclic aromatic hydrocarbons (PAHs)

Dicofolc Endosulfansd Hexabromocyclododecanee Hexachlorobutadienec Octachlorostyrenec Decabromodiphenyl formulationc Polychlorinated naphthalenes (PCNs)c Polycyclic aromatic hydrocarbons (PAHs)c Short-chained chlorinated paraffinsf

a

UNEP (2001, 2009). UNEP (2008a, 2009) and UNECE (1998). c WWF (2005). d UNEP (2008a). e UNEP (2008b) f UNEP (2008c). b

compound groups to be included in the Aarhus Protocol and the Stockholm Convention on Persistent Organic Pollutants, respectively (see Table 10.2). The Stockholm Convention is an international treaty that requires parties to take measures to: .

. . .

Eliminate or reduce the production, use, import, and export of intentionally produced and used POPs (e.g., pesticides, PCBs) Reduce releases of unintentionally produced POPs [e.g., poly(chlorinated dibenzodioxin)s and -furans] Promote the best practices and techniques available to reduce POP emissions Dispose of old stockpiles of POPs in an environmentally responsible way (UNEP 2001)

Considering all possible routes for POPs to enter the environment, this is a more comprehensive approach to regulate these chemicals than the UNECE process, which only addresses emissions of POPs to the atmosphere. Both treaties established processes that allow for inclusion of additional chemicals. Some products suggested for inclusion

into the Stockholm convention by nongovernmental organisations and those proposed by governmental representatives to date are also listed in Table 10.2. Under the Stockholm Convention each party needs to submit a national implementation plan (NIP) to the Governing Council of the UNEP. The NIP serves as a framework for developing regulations and measures in order to comply with the convention. This plan must be reviewed and updated periodically (UNEP 2001). Under the CLRTAP, national laws should be amended in such a way that the requirements of this treaty are met. To demonstrate the sufficiency and effectiveness of such laws, the parties have to submit reports on strategies and policies adopted to achieve this goal as well as POP usage/emission monitoring reports to the implementation committee on a regular basis (UNECE 1998). The idea that persistent, semivolatile compounds can undergo repeated air–surface exchange (intermedia transfers) led to the notion of a global redistribution of POPs, driven by temperature. This was best articulated by Wania and Mackay (1996) and is conceptualized in Figure 10.1. This framework, embodying the ideas of cold condensation, global distillation and fractionation (see later) has raised

254

INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS

some fascinating questions about the global fate and behavior of POPs, which are the motivations behind this chapter.

10.3. PRIMARY AND SECONDARY POP SOURCES AND THE ROLE OF RECYCLING Persistent organic pollutants (POPs) can be divided into intentionally and unintentionally produced compounds, with the former representing pesticides such as DDT or HCHs that are usually applied openly to the environment, and industrial chemicals (e.g., PCBs) that are partly used in closed systems. The latter are byproducts, for example of incomplete burning processes where PAHs, PCDDs, and PCDFs may form. These sources, including releases due to accidents or inappropriate disposal, are referred to as primary sources. After entering the atmosphere or water bodies, POPs can be trapped or held by or in compartments for which they have a high affinity, such as vegetation, soils, sediments, or transferred elsewhere (e.g., the deep ocean). These compartments can become secondary sources of POPs if the chemicals revolatilize or dissolve. Where and how POPs are released into the environment affects their fate, transport, and behavior. If they are released from high temperature combustion to the atmosphere, they can be spread far and wide, whereas if they are placed in a landfill site, for example, their mobility is restricted. Given the focus on global dispersion, early attention focused on atmospheric emissions and inventories of these compounds, which lead to early and rapid spreading of a fraction of these compounds. However, as the years and decades go by, transfers to the Arctic via aquatic discharges, rivers, and marine currents presumably become more important. One difficulty for researchers is that it can be very difficult to obtain reliable information on the amounts of a chemical produced (now and in the past), on where and how it was used, and the consequent potential for release to the environment. Breivik et al. (2002a, 2002b, 2007) have undertaken a detailed case study of polychlorinated biphenyls (PCBs) in this regard. The industrial production of PCBs started in 1929 and peaked in the late 1960s. Their production and use were banned in the United States and Europe in the late 1970s because of the serious risks that they pose to the human health and the evidence found for the wide environmental distribution of these chemicals. Polychlorinated biphenyls have been used extensively not only in closed systems as cooling liquids in transformers, dielectric liquids in capacitors, fireresistant hydraulic fluids in mining equipment, and vacuum pumps but also in open systems as plasticizers, paints, adhesives, lubricants, sealing liquids, fire retardants, and pesticides. Release of PCBs to the environment is not limited to open systems but also occurs from closed systems in the case of leakage, improper disposal, and fires. According to

estimates (Breivik et al. 2007), only a small percentage of the 1.3  106 metric tons of PCBs that were produced globally, predominantly in the Northern Hemisphere between 30 N and 60 N (Breivik et al. 2002a), has been released to the atmosphere to date. There is evidence from monitoring studies that the releases and ambient levels of PCBs have been declining since the 1970s–1980s. Hence, PCBs have been introduced as a long “pulse” into the environment, peaking in the 1960–1970s. However, given their persistence in the environment, questions arise as to how much of what is currently measured in air, humans, and biota reflects ongoing diffusive (primary) emissions, and how much reflects the “old stock” of PCBs that may have been released over the last decades. This is an interesting question for scientists to resolve, but it has a very practical significance for regulators and policymakers as well. If ongoing primary sources still dominate ambient levels and exposure, then further measures to reduce sources could be considered and effective; if not, then current and future exposures will essentially be controlled by the rates of “global degradation and clearance” of these persistent compounds. These issues have driven much research effort to study and understand the processes of intermedia transfers of POPs.

10.4. KEY INTERMEDIA TRANSFERS As noted above, POPs are ubiquitous in the environment. They can be found in various compartments such as the atmosphere, oceans, lakes, glaciers, sediments, soils, vegetation, and animals all over the world, including very remote areas such as the Arctic and Antarctic. This is explained by their physicochemical properties: .

.

.

Their subcooled liquid vapor pressures at 298 K range approximately from 4 Pa (naphthalene) to 1.3  106 Pa (e.g., octachlorodibenzo-p-dioxin) with PCBs, PBDEs, and organochlorine pesticides showing intermediate values. This enables most of them to exist in the vapor phase of the air and thus be relatively mobile in the environment. On the other hand, they are also of sufficiently low volatility to easily partition from the air to surfaces, such as of airborne particles, vegetation, or soil. The very low polarity of POPs causes their aversion to water (hydrophobicity) and high affinity to other nonpolar substances such as fatty tissues of plants and animals or humic substances in soils (lipophilicity), rendering them prone to accumulation in biota and soil. The high chemical inertness of POPs prevents them from being metabolized by (micro)organisms or degraded by other processes quickly. Together with the lipophilicity, this property results in increasing

KEY INTERMEDIA TRANSFERS

accumulation of POPs along the food chain (biomagnification) with highest burdens found in top predators.

10.4.1. Atmospheric Transport and Transfers The air has been important as an environmental transport medium for POPs on a global scale as it can carry gas-phase compounds by advection over hundreds of kilometers within a few hours. However, the Intertropical Convergence Zone (ITCZ) forms a barrier between air masses from the Northern Hemisphere (NH) and the Southern Hemisphere (SH), resulting in much lower POP burdens in the air of the SH compared to the NH where most of the sources of POPs can be found. Concentrations of POPs in the air are determined by the travel routes of air masses and can vary considerably between air that has passed relatively “clean” regions such as oceans or the Arctic and air that has “seen” polluted places such as large cities or large agricultural areas. Since POP emissions from sources are temperature-dependent, seasonal changes of concentrations can even be observed for air masses with similar origin. Depending on their compound-specific vapor pressure, POPs can be found in both the vapor and the particulate phases of the air. The most volatile substances (e.g., HCB, HCHs, low chlorinated PCBs) are almost exclusively present in the vapor phase, whereas less volatile compounds (e.g., highly brominated PBDEs, dioxins, furans) show considerable or even dominating particle bound fractions. Because of the temperature dependence of the compounds’ vapor pressure, the vapor-phase : particle-bound ratio varies between climatic zones, seasons, and even over the course of the day. The high affinity of some POPs to nonpolar particles suggests that partitioning between gas and particle phases may also be affected by their lipophobicity. Longrange transport of POPs bound to airborne particulates is less efficient than for gas-phase compounds but still highly significant for their transfer from sources to more remote areas. Binding to particles may also further reduce the reactivity of these already very persistent chemicals (see Section 10.5.1). 10.4.2. POPs in Vegetation and Soil Away from urban areas, vegetation and soil are the most important compartments in the terrestrial environment that expose surfaces to the air (except for areas with permanent snow or ice cover). Airborne POPs can partition onto these surfaces and may therefore be trapped temporarily. This will slow down their long-range transport if they revolatilize back to the atmosphere later, but it may also prevent further transport entirely if the chemicals are not released again. All POPs have a high affinity to the nonpolar components that can be found in both vegetation and soil (e.g., waxy cuticles

255

of plants, lipids, and humic substances in the organic matter fraction of soils). Leaves and needles—the parts of plants that expose the largest surface to the atmosphere—are relatively short-lived. When they fall to the ground and become part of the soil they also carry over POPs associated with them. This means that (1) abundant vegetation can enhance the supply of POPs from the atmosphere to the soil compared to bare or sparsely covered soil, a process also known as the “forest filter effect” (Horstmann and McLachlan 1998; Nizzetto et al. 2005) and (2) soils can potentially act as a final sink for these pollutants. 10.4.3. Exchange of POPs Between Air and Vegetation or Soil Persistent organic pollutants can be transferred from the atmosphere to vegetation and soil surfaces by dry or wet deposition (see Fig. 10.2). The first includes dry gaseous deposition, that is, adsorption of vapor phase chemicals to surfaces; and dry particle deposition, which is driven by particulate matter, POPs are bound to, settling onto surfaces (Fig. 10.3). Which of these two processes dominates depends on the chemical’s affinity to particles, which is affected by the POP’s physicochemical properties such as vapor pressure and lipophilicity (see discussion later). Other influencing parameters are the size and chemical composition of particulates, and their concentration in the air. Dry gaseous deposition velocities may be limited on the plant side, that is, by the transport of the chemical within the plant away from the surface immediately exposed to the air (if air–surface equilibrium is reached relatively quickly) or on the air side, that is, transport of the chemical through the laminar air layer (LAL) that surrounds surfaces and can be passed only by diffusion (if air–surface equilibrium is reached relatively slowly) (McLachlan 1999). Therefore wind speed also has an impact on dry deposition as it affects both particle settling velocities and the thickness of the LAL (Barber et al 2002). Wet deposition refers to the washout of airborne chemicals due to precipitation, including rain, snow, dew, mist, and fog (Fig. 10.3). This process can be very efficient in reducing concentrations of particle-associated POPs in the air whereas more volatile compounds that are present predominantly in the gaseous phase will be removed to a much lower extent. Because of their very low water solubility, wet deposition only plays a minor role in cleansing the air from vapor-phase POPs (Poster and Baker 1996). However, at low temperatures, snow that exposes large surfaces to the vapor phase of the air and thus encourages adsorption of gaseous compounds (Fig. 10.3) may scavenge more significant fractions of gas phase POPs (Agrell et al 2002; Lei and Wania 2004). (Re)volatilization, in contrast to deposition, delivers POPs from anthropogenic media but potentially also from

256

INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS

Figure 10.2. Schematic representation of the key processes affecting concentrations of POPs in the air, vegetation, and soil systems. (See insert for color representation of this figure.)

environmental compartments (mainly soil and vegetation) to the air (Fig. 10.2). This emission process involves evaporation of the chemical from the surface that it is bound to, followed by transport to the atmosphere. The first step occurs if a compound’s concentration in the air is below its equilibrium concentration with the surface. Among other factors that are presented in the next section, this depends strongly on the chemical’s temperature-driven vapor pressure. The second step requires diffusive transport through the LAL and may therefore also be controlled by wind speed. 10.4.4. Partitioning of POPs Between Environmental Compartments As mentioned earlier, partitioning of POPs between different environmental media has an important impact on the

surface–air exchange of these compounds. Chemicals will move toward equilibrium between media that are in contact with each other. Therefore, the direction of the exchange depends on the current distribution of the chemical between these media compared to the equilibrium distribution. The state of equilibrium is influenced by many parameters, including the compound’s temperature-dependent vapor pressure; its polarity and resulting water solubility, hydrophobicity, and lipophilicity; and the storage capacity that the exchanging media can provide. Given the high lipophilicity of POPs, a compartment’s capacity is usually the higher, the more lipophilic constituents it contains, following the principle of “like dissolves like.” For example, epidermal plant cells, containing cuticular waxes embedded in or covering a cutin matrix, or humic substances in soils are components POPs show a high affinity to, whereas parenchyma cells or

KEY INTERMEDIA TRANSFERS

257

Figure 10.3. Schematics of (a) the different deposition types supplying airborne POPs to vegetation and soil surfaces and (b) important partitioning processes in the terrestrial environment. (See insert for color representation of this figure.)

the mineral fraction of soils attract such chemicals much less (Karickhoff 1981; Simonich and Hites 1994; Ahmad et al. 2001; Wang et al. 2008). To describe the state of equilibrium between air and other terrestrial compartments, a number of dimensionless partition coefficients can be used. They are derived from the ratio between the equilibrium concentrations in both compartments, such as the plant–air partition coefficient KPA, the cuticle–air partition coefficient KCA, the soil–air partition

coefficient KSA, the organic matter–water partition coefficient KOM, the organic carbon–water partition coefficient KOC, the dissolved organic matter–water partition coefficient KDOM (see Fig. 10.3b). These coefficients were shown to correlate well with the octanol–air partition coefficient KOA or the octanol–water partition coefficient KOW (Kerler and Sch€onherr 1988; Tolls and McLachlan 1994; Burkhard 2000), which are easier to access experimentally for a large number of chemicals. The nondimensionless particle–gas partition

258

INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS

coefficient KP (mg/m3) (Fig. 10.3) also correlates well with KOA (Falconer and Harner 2000). The coefficient KP is defined as (F/TSP)/A, where F and A are the concentrations (ng/m3) of the chemical associated with the particulate and the vapor phase, respectively, and TSP is the total suspended particle concentration (mg/m3). These findings explain the high importance of KOA and KOW values in model approaches to describe and predict the surface–air exchange of POPs. 10.4.5. Cold Condensation and Global Fractionation Provided that POPs are not subject to any loss or generation and are readily available for partitioning between different environmental compartments, they should eventually approach a global equilibrium in the environment. As they

approach this state, an important fraction is transported from the temperate source areas to remote, colder regions due to long-range atmospheric movements of air masses (see Fig. 10.1). Some of these cold environments, such as boreal zones, contain soils with very high organic carbon contents and hence large storage capacity for POPs. Two phenomena, namely, cold condensation and global fractionation, are believed to affect both the compound-specific time needed to equilibrate globally and the concentrations in different environmental media at a given time during this process (Wania and Mackay 1993). Cold condensation refers to the effect of decreasing vapor pressure as temperatures fall during a compound’s journey from south to north (in the Northern Hemisphere) and the resulting increasing ambition of these chemicals to leave the vapor phase of the air and

Figure 10.4. The maximum reservoir capacity of soils for HCB at January and July average temperatures [see Dalla Valle et al. (2005)]. (See insert for color representation of this figure.)

POPs IN THE GLOBAL ENVIRONMENT: SOME FURTHER OBSERVATIONS

adsorb to surfaces (Wania and Mackay 1993). As a consequence, POPs that are present predominantly in the vapor phase of the air in temperate regions will presumably preferentially accumulate in soils at higher latitudes. As shown in Figure 10.1, the lightest, that is, most volatile and thus most mobile, compounds will be able to travel farthest northward while large fractions of heavier, less volatile ones partition onto surfaces further south already (in the Northern Hemisphere). Thus the transport of POPs toward cold regions is assumed to be more restricted the lower their compoundspecific vapor pressure at a given temperature. Wania and Mackay (1993) described this effect in the global fractionation hypothesis since it is like the fractionation of chemicals in a chromatographic column (Fig. 10.1). Evidence for such a process is believed to arise from spatiotemporal monitoring data. For example, Agrell et al. (1999) found a growing contribution of the most volatile PCB congeners to the total atmospheric PCB concentration with increasing latitude along the Baltic Sea coast. Meijer et al. (2003) obtained similar results from topsoil samples, collected along a latitudinal United Kingdom–Norway transects. Muir et al. (1996) reported a shift toward more volatile PCB congeners with depth (i.e., age) in sediment cores from high-latitude lakes, and Blais and Muir (2001) found a delayed PCB concentration peak in lake sediments from the far north compared to more southern lakes. Wania and Mackay (1996) proposed two different scenarios as to how global fractionation may proceed: as a single step, that is, once a compound has deposited to surfaces, it will be retained there “indefinitely,” or by repeated deposition and revolatilization (“the grasshopper” effect) as illustrated in Figure 10.1, for example, driven by seasonal temperature variations or release of POPs during decomposition of the organic matter to which they are bound (see Fig. 10.2). As a POP approaches global air–surface equilibrium, the “capacity” of the surface environment will influence how readily POPs are retained. Jurado et al. (2004) and Dalla Valle et al. (2004, 2005) calculated the maximum reservoir capacity of surface soils, oceans, and vegetation, on the basis of theoretical air–surface partitioning conditions, as a function of temperature. Figure 10.4 illustrates this for HCB and soils. It shows that, rather than the far-northern Arctic being the projected “sink” for POPs, soils of high organic matter content (such as peats and forest litter) are large reservoirs, capable of storing POPs.

10.5.1. Elimiation Despite the high chemical persistence of POPs, there are several degradation processes that result in their elimination from the environment. In the atmosphere photodegradation can cause destruction of POPs, depending on their susceptibility to reaction with OH radicals or direct photolysis. A number of studies reported significant photodegradation of PBDEs (e.g., Eriksson et al. 2004) and PAHs (Arey and Atkinson 2003), for example. However, binding to atmospheric particles can reduce the chemicals’ susceptibility to photochemical breakdown significantly, resulting in comparably low reaction rates for many chemicals that are present mainly in the particulate phase (Fioressi and Arce 2005; Raff and Hites 2007). In soil POPs are subject to biodegradation, that is, to both aerobic and anaerobic metabolism by microorganisms (Fig. 10.2). This has been investigated in a number of studies under laboratory conditions, often with high concentrations of the contaminant (Meharg and Cairney 2000; Borja et al. 2005), but rarely using environmentally background concentrations (Abramowicz 1995). However, extrapolation of laboratory results to environmental background conditions introduces problems because biodegradation rates observed in laboratory systems may deviate from those under environmental conditions (Spain et al. 1980; Abramowicz 1995). The lack of field data on this important process is due mainly to the very low degradation rates of POPs, requiring long-term experiments, and other loss processes (see next section) that complicate the interpretation of results. 10.5.2. Processes Influencing Availability of POPs for Surface–Air Exchange Although still present in the environment, POPs may not be readily available for surface–air exchange, as could be expected from their volatility or partition coefficients. Particularly in soil, several processes can reduce the chemical’s accessibility: .

.

10.5. POPs IN THE GLOBAL ENVIRONMENT: SOME FURTHER OBSERVATIONS Despite the ideas introduced earlier regarding the approach to equilibrium over time, several confounding factors and processes may restrict establishment of a global equilibrium distribution of POPs in the environment.

259

Burial of Surface Layers by Fresh Litter (Humus Buildup; see Fig. 10.2). Depending on the soil structure, only a superficial layer of the order of micrometers to a few millimeters can directly participate in the most effective exchange between soil and the atmosphere. When new litter covers that layer, the exchange may be kinetically limited (Mackay 2001; Harner et al. 2001). Transport to Deeper Layers (see Fig. 10.2). Similar to burial, downward transport of POPs within the soil profile will limit their exchange with the atmosphere by removing them from the surface. Translocation may be caused by leaching, that is, water-mediated transport of chemicals that are either truly dissolved or associated with dissolved organic matter (DOM) (Krauss et al. 2000). The former will affect the least hydrophobic,

260

.

.

INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS

most water-soluble POPs most, whereas the latter may also be important for very hydrophobic compounds, provided a sufficient amount of DOM is present in the soil. Bioturbation, that is, mixing of the soil by macroorganisms, mainly earthworms, is another process that can carry POPs away from the soil surface to deeper layers, resulting in limited exchange with the atmosphere. On the contrary, it may also bring buried material of higher contaminated layers up to the surface and therefore possibly encourage soil–air exchange. The abundance of earthworms and thus the extent of bioturbation depend strongly on soil conditions, particularly on the acidity. Under favorable conditions (pH 5–6) that can be found in many grassland and agricultural soils the earthworm biomass can exceed 200 g/m2 (M€ullerLemans and van Dorp 1996), whereas virtually no earthworms inhabit very acidic coniferous forest soils (Lavelle et al. 1995). Formation of Bound Residues. A number of studies reported that fractions of the POP load in soils become strongly bound to soil particles as they age, that is, as they remain in contact with the soil matrix for a long time. This is believed to be due to diffusion of the chemicals deep into micropores of soil particles and will depend on the physicochemical properties of the chemical, meteorological parameters, and also soil properties such as organic matter content and quality, particle size, and pore size (Doick et al. 2005). “Bound” or “nonextractable” fractions of POPs may be too immobile to participate in an exchange with the atmosphere. Permanent Retention of POPs in the Aquatic Environment. The deep ocean (Dachs et al. 2002), marine shelf sediments (J€ onsson et al. 2003), and lake sediments (Muir et al. 1996; Blais and Muir 2001) are believed to act as final sinks for POPs in the global environment. The chemicals can be supplied to these compartments by (1) direct (mainly anthropogenic) contamination; (2) surface waters, following air-water exchange; or (3) runoff of polluted material originating from the terrestrial environment (Fig. 10.2). Compounds trapped there may be permanently removed from the atmosphere and thus can no longer participate in the exchange with terrestrial surfaces anymore.

10.6. POP MODELING IN THE ENVIRONMENT 10.6.1. Introductory Remarks Models are used to describe and predict the potential environmental distribution and fate of chemicals (Mackay 2001). Examples include the assessment of their atmospheric transport potential (e.g., Stroebe et al. 2004) or their persistence in the environment, also for new chemicals before their commercial production begins (Fenner et al. 2005;

Klasmeier et al. 2006). Models can help to interpret monitoring results and provide additional information, such as assessing the distance from the thermodynamic equilibrium, thus predicting the net flux direction among the compartments, or studying the respective input of pollutants released from various distant sources to a remote site. Other uses of models applied to POPs include the testing of hypotheses on their behavior in the environment, for example, to verify the importance of a specific process. Models are a conceptual representation of the real world, a simplification of the reality used to interpret the functioning of environmental processes. They can be built with different degrees of complexity, depending on the detail of the available information (about the relevant environmental characteristics and about the properties of the studied compound) and on the objectives of the study (Cahill and Mackay 2003). The level of detail of models is a compromise between simplicity and accuracy in describing real-world conditions. A complex model is expected to be a more accurate representation of the environment than a simple one, and should produce a more accurate output. However, the more complex the model, the more information is needed and the longer is the computational time required to generate the model output. In addition, results may be difficult to interpret, due to excessive complexity of the model. The two main approaches adopted so far to study the environmental distribution of POPs make use of mass balance models (or box models) or atmospheric transport models: (1) classic trajectory models that use calculated or measured wind fields to describe the time-dependent advection-driven pathway of a pollutant emitted at a known location, (2) atmospheric dispersion models that are usually used to calculate the concentration distribution of chemicals downwind of a point source, or (3) approaches based on spatiotemporily highly resolved meteorological or climate models that have been adapted to POP modeling (e.g., Hansen et al. 2004; Lammel et al. 2007). Box models are usually of multicompartmental structure as this best meets the requirements of investigating the fate and behavior of POPs, substances that partition and can be transferred between different environmental media due to their semivolatile nature. Box models describe the environment in a simple way, dividing it into a certain number of compartments (boxes) that are assumed to be internally homogeneous and well mixed. The environmental properties within each box are homogeneous, and so are the concentrations. The concentration of a chemical in each compartment is determined by its physicochemical properties, the properties of the compartment, and the inter- and intracompartmental mass exchanges. As opposed to this, most transport models applied to POPs and chemicals with similar physicochemical properties focus on the atmospheric compartment, although other compartments such as oceans, groundwater, or soil may be addressed

POP MODELING IN THE ENVIRONMENT

as well (e.g., Pang et al. 2000; Ilyina et al. 2006). They describe processes in these compartments (e.g., air or water circulation) in much greater detail than box models do. Highly resolved atmospheric dispersion models, for instance, not only represent temporally averaged advective flows but also account for mechanical or convective turbulence, resulting in a much more detailed picture of the atmosphere in space and time than box models can usually achieve (e.g., Draxler and Hess 1998; Stohl et al. 1998). Meteorological or climate model (e.g., general circulation models of the atmosphere)–based approaches are able to include a more detailed picture of the atmosphere with several layers and further processes such as the formation and breakup of clouds or the occurrence of various types of precipitation (Hansen et al. 2004; Lammel et al. 2007). However, such highly complex models may be limited by the computational power needed. Box models, on the other hand, usually have a low spatial resolution and use simplifying assumptions of equilibrium or steady state. However, for several applications this is not a limitation, as POPs are, by definition, persistent and have been present in the environment for decades, allowing enough time to reach relatively uniform concentrations (Wania and Mackay 1999a). In addition, in most modeling exercises, the capability of the model to provide a satisfactory prediction is limited by several uncertainties in emission estimations, environmental characteristics, degradation rates, physicochemical properties of the chemical, and in describing some multimedia fluxes. In many cases these uncertainties limit model performances more than the low spatial and temporal resolution does. In order to benefit from the advantages of both model types, they have been combined, for instance, by modeling the atmosphere with a dispersion model that is linked to a box model describing other environmental compartments (Ma et al. 2003; Hansen et al. 2004; Malanichev et al. 2004; Semeena et al. 2006). Of all the different models that are available for describing and predicting the environmental fate and behavior of POPs, there is probably no “all-purpose” one, but the decision as to which type and degree of complexity needs to be chosen should always be based on the requirements arising from the goal of a specific study. 10.6.2. Box Models The most common box models currently used are based on the concept of fugacity, and are usually referred to as “Mackay-type models,” after Professor Don Mackay (Trent University, Canada), who developed the multimedia environmental modeling approach [see Mackay (2001) for a detailed summary of his modeling approaches]. Fugacity is a thermodynamic function that expresses the tendency of a chemical to escape from one phase to another. This means

261

that a given chemical tends to move from one phase to another with a lower fugacity, until the fugacity in all the phases is equal (i.e., equilibrium is reached). Fugacity has units of pressure (Pa) and is logarithmically related to the chemical potential m. As a consequence, fugacity is linearly related to the concentration, so that it is more convenient to use than the chemical potential. Fugacity-based models have four degrees of complexity depending on the scope of the models: .

.

.

.

Level 1, which assumes that equilibrium between the compartments is used to predict the potential distribution of a chemical among the individual environmental compartments. Level 2, which is still an equilibrium model, but advective inputs and outputs are considered. In this approach inputs and outputs are allowed to vary over time. A typical application is the study of the decay of concentration of a chemical introduced in the model environment at zero time, where only removal processes are considered. Level 3, which considers the more realistic case of environmental compartments with associated different fugacity. Steady-state condition is assumed, and the fugacity of the individual environmental compartments is constant over time. Level 4, which assumes dynamic conditions; therefore, the mass of the chemical and its fugacity in the compartments are allowed to change over time, as a result of inputs, outputs, and loss processes.

The applications range of fugacity-based models is wide, from testing the behavior of a chemical in a specific environment (e.g., Mackay 1989), to studying fluxes at the interface (e.g., air–water exchange and air–soil exchange), predicting bioaccumulation (e.g., Campfens and Mackay 1997), calculating indoor exposure to air pollutants (e.g., Bennett and Furtaw 2004), estimating their persistence and long-range transport potential (e.g., Prevedouros et al. 2004; Scheringer 1996; Scheringer et al. 2004), or removal rates from different compartments (Gong et al. 2003), or for source apportionment (Wania and Mackay 1999b; Wania et al. 2006). The original “Mackay-type models” can also be extended, for instance, by georeferencing compartments (Suzuki et al. 2004). The architecture and level of detail of a model are strictly related to the scale at which the model is applied. The environmental characterization of the model is a function of the scale and depends on the aims of the model. For example, one of the main applications of multimedia fate models is that of studying in detail the distribution and fluxes of POPs within a specific region, often a contaminated area where there are particular reasons for concern about human and wildlife exposure to POPs. Such models are used with

262

INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS

different purposes, such as estimating multimedia fluxes or concentrations in some compartments or reconstructing temporal trends, and they require a good level of detail in their parameterization in order to achieve sufficient accuracy.

10.6.3. Transport Models 10.6.3.1. Trajectory Models. From a Lagrangian perspective on air movement, trajectories represent the path that infinitesimally small particles take in the atmosphere. Thus advective transport pathways of air pollutants can be modeled, allowing for studying where chemicals released from a known source are carried to or from where those reaching a certain area actually originated (e.g., Seibert et al. 1994; Stohl 1996; Hafner and Hites 2005). Numerous methods are used to compute trajectories, that is, to calculate stepwise the position of a particle with known initial coordinates, called Lagrangian coordinates, at later (forward trajectories) or earlier (backward trajectories) times, using wind data from diagnostic or prognostic wind field models or numerical weather predictions. For more information on trajectory calculations, the reader is referred to Stohl (1998). 10.6.3.2. Dispersion Models. In order to more realistically simulate the transport of pollutants within the planetary boundary layer where turbulences play a significant role, trajectory models have been extended to Lagrangian dispersion models such as HYSPLIT (Draxler and Hess 1998) or FLEXPART (Stohl et al. 1998). Three different methods are commonly used to describe and predict the dispersion of pollutants in the atmosphere or aquatic environments: 1. Plume models, which are used to calculate the concentration of a chemical released from a point source depending on the position downwind or downstream from this source under steady-state conditions (Pasquill 1961; Wagner et al. 1985). 2. Puff models, which can be used under non-steady-state conditions to provide information on concentrations of a pollutant that has been carried to a certain position from a source where the release is modeled as a series of puffs (Draxler and Hess 1998). 3. Particle dispersion models, where the transport and diffusion of a passive tracer can be shown by deriving trajectories for a large number of “tagged particles” (Draxler and Hess 1998; Stohl et al. 1998). This is the most sophisticated of these three types of Lagrangian dispersion models. Apart from modeling the dispersion of chemicals in the atmosphere, it also allows for involving processes such as dry and wet deposition or emission (Stohl 1998). This type of model has been used to identify emission sources of POPs and estimate their contribution to the amount of these chemicals

found in remote regions such as the Arctic (Kallenborn et al. 2007; Eckhardt et al. 2009).

10.6.3.3. Meteorological and Climate-Model-Based Approaches. These models, describing the transport of chemicals in air or other fluids (e.g., groundwater or ocean water), are driven by meteorological data provided by Eulerian models, such as ASIMD (Pekar and van Pul 1999) or DEHM (Hansen et al. 2004) or by general circulation models such as ECHAM (Lammel et al. 2007). They are combined with the description of the chemistry or partitioning of POPs in the environment in order to represent substance-specific multimedia processes involved, such as phase exchange and degradation (e.g., Lammel et al. 2007). These 3D atmospheric transport models take large-scale weather patterns and their seasonality into account, which includes, for instance, a more detailed picture of the global temperature distribution, cloudiness, or precipitation. This allows for a detailed prediction of a compound’s spatiotemporal distribution within the area studied. Depending on the meteorological input data, the scale of such models ranges from regional to hemispheric and global (Ma et al. 2003; Hansen et al. 2004; Malanichev et al. 2004, Koziol and Pudykiewicz 2001). Most of these models that were applied to POPs to date focus on the atmosphere. However, they can also be used to describe transport processes in other compartments such as oceans (Ilyina et al. 2006) or may be combined with less complex models (usually box models, as mentioned above) to include other compartments, for instance, soil, vegetation, or seawater and ice (e.g., Ma et al. 2003; Hansen et al. 2004; Malanichev et al. 2004; Semeena et al. 2006).

10.7. CONCLUSIONS Persistent organic pollutants provide an excellent case study, where environmental chemistry (at the large scale), policy, regulation, and controls come together. Scientists have had an important role to play in improving our understanding of the complex behavior of these substances. They have shown how POPs interact and exchange with the Earth’s surfaces and compartments, in a dynamic fashion, which has changed over time. Their persistence and potential toxicity raise special concerns about their possible chronic effects, highlighting the need for testing and regulatory frameworks to be watchful for similar substances in the future.

REFERENCES Abramowicz, D. A. (1995), Aerobic and anaerobic PCB biodegradation in the environment, Environ. Health Perspect. 103, 97–99.

REFERENCES

Agrell, C., Okla, L., Larsson, P., Backe, C., and Wania, F. (1999), Evidence of latitudinal fractionation of polychlorinated biphenyl congeners along the Baltic Sea region, Environ. Sci. Technol. 33, 1149–1156. Agrell, C., Larsson, P., Okla, L., and Agrell, J. (2002), PCB congeners in precipitation, wash out ratios and depositional fluxes within the Baltic Sea region, Europe, Atmos. Environ. 36, 371–383. Ahmad, R., Kookana, R. S., Alston, A. M., and Skjemstad, J. O. (2001), The nature of soil organic matter affects sorption of pesticides. 1. Relationships with carbon chemistry as determined by 13 C CPMAS NMR spectroscopy, Environ. Sci. Technol. 35, 878–884. Arey, J. and Atkinson, R. (2003), Photochemical reactions of PAHs in the atmosphere, in PAHs: An Ecotoxicological Perspective, Douben, P. E. T. ed., Wiley Chichester, UK pp. 47–63. Barber, J. L., Thomas, G. O., Kerstiens, G., and Jones, K. C. (2002), Air-side and plant-side resistances influence the uptake of airborn PCBs by evergreen plants, Environ. Sci. Technol. 36, 3224–3229. Bennett, D. H. and Furtaw, E. J. (2004), Fugacity-based indoor residential pesticide fate model, Environ. Sci. Technol. 38, 2142–2152. Bentzen, T. W., Follmann, E. H., Amstrup, S. C., York, G. S., Wooller, M. J., Muir, D. C. G., and O’Hara, T. M. (2008), Dietary biomagnification of organochlorine contaminants in Alaskan polar bears, Can. J. Zool. 86, 177–191. Blais, J. M. and Muir, D. C. G. (2001), Paleolimnological methods and applications for persistent organic pollutants, In Tracking Environmental Change Using Lake Sediments, Vol. 2, Physical and Geochemical Methods, Last, W. M. and Smol, J. P., eds., Kluwer, Dordrecht, pp. 271–298. Borja, J., Taleon, D. M., Auresenia, J., and Gallardo, S. (2005), Polychlorinated biphenyls and their biodegradation, Process Biochem. 40, 1999–2013. Breivik, K., Sweetman, A., Pacyna, J. M., and Jones, K. C. (2002a) Towards a global historical emission inventory of selected PCB congeners—a mass balance approach. 1. Global production and consumption, Sci. Total Environ. 290, 181–189. Breivik, K., Sweetman, A., Pacyna, J. M., and Jones, K. C. (2002b) Towards a global historical emission inventory for selected PCB congeners—a mass balance approach. 2. Emissions, Sci. Total Environ. 290, 199–224. Breivik, K., Sweetman, A., Pacyna, J. M., and Jones, K. C. (2007), Towards a global historical emission inventory for selected PCB congeners—a mass balance approach, 3. An update, Sci. Total Environ. 377, 296–307. Burkhard, L. P. (2000), Estimating dissolved organic carbon partition coefficients for nonionic organic chemicals, Environ. Sci. Technol. 34, 4663–4668. Cahill, T. M. and Mackay, D. (2003), Complexity in multimedia mass balance models: When are simple models adequate and when are more complex models necessary? Environ. Toxicol. Chem. 22, 1404–1412. Campfens, J. and Mackay, D. (1997), Fugacity-based model of PCB bioaccumulation in complex aquatic food webs, Environ. Sci. Technol. 31, 577–583.

263

Dachs, J., Lohmann, R., Ockenden W. A., Mejanelle, L., Eisenreich, S. J., and Jones, K. C. (2002), Oceanic biogeochemical controls on global dynamics of persistent organic pollutants, Environ. Sci. Technol. 36, 4229–4237. Dalla Valle, M., Dachs, J., Sweetman, A. J., and Jones, K. C. (2004), Maximum reservoir capacity of vegetation for persistent organic pollutants: Implications for global cycling, Global Biogeochem. Cycles 18, Art. No. GC4032. Dalla Valle, M., Jurado, E., Dachs, J., Sweetman, A. J., and Jones, K. C. (2005), The maximum reservoir capacity of soils for persistent organic pollutants and implications for their global cycling, Environ. Pollut. 134, 153–164. Doick, K. J., Klingelmann, E., Burauel, P., Jones, K. C., and Semple, K. T. (2005), Long-term fate of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in an agricultural soil, Environ. Sci. Technol. 39, 3663–3670. Draxler, R. R. and Hess, G. D. (1998), An overview of the HYSPLIT_4 modelling system for trajectories, dispersion and deposition, Austral. Meteorol. Mag. 47, 295–308. Eckhardt, S., Breivik, K., Li, Y. F., Manø, S., and Stohl, A. (2009), Source regions of some persistent organic pollutants measure in the atmosphere at Birkenes, Norway, Atmos. Chem. Phys. Discuss. 9, 12345–12383.  Eriksson, J., Green, N., Marsh, G., and Bergman, A. (2004), Photochemical decomposition of 15 polybrominated diphenyl ether congeners in methanol/water, Environ. Sci. Technol. 38, 3119–3125. Falconer, R. and Harner, T. (2000), Comparison of the octanol-air partition coefficient and liquid-phase vapor pressure as descriptors for particle/gas partitioning using laboratory and field data for PCBs and PCNs, Atmos. Environ. 34, 4043–4046. Fenner, K., Scheringer, M., MacLeod, M., Matthies, M., McKone, T., Stroebe, M., Beyer, A., Bonnell, M., Le Gall, A. C., Klasmeier, J., Mackay, D., Van De Meent, D., Pennington, D., Scharenberg, B., Suzuki, N., and Wania, F. (2005), Comparing estimates of persistence and long-range transport potential among multimedia models, Environ. Sci. Technol. 39, 1932–1942. Fioressi, S. and Arce, R. (2005), Photochemical transformations of benzo[e]pyrene in solution and adsorbed on silica gel and alumina surfaces, Environ. Sci. Technol. 39, 3646–3655. Godduhn, A. and Duffy, L. K. (2003), Multi-generation health risks of persistent organic pollution in the far north: Use of the precautionary approach in the Stockholm Convention, Environ. Sci. Policy 6, 341–353. Gong, S. L., Barrie, L. A., Blanchet, J. P., von Salzen, K., Lohmann, U., Lesins, G., Spacek, L., Zhang, L. M., Girard, E., Lin, H., Leaitch, R., Leighton, H., Chylek, P., and Huang, P. (2003), Canadian aerosol module: A size-segregated simulation of atmospheric aerosol processes for climate and air quality models 1. Module development, J. Geophys. Res. Atmos. 108, 4007. Hafner, W. and Hites, R. A. (2005), Effects of wind and air trajectory directions on atmospheric concentrations of persistens organic polltuants near the Great Lakes, Environ. Sci. Technol. 39, 7817–7825.

264

INTERMEDIA TRANSFERS AND GLOBAL CYCLING OF PERSISTENT ORGANIC POLLUTANTS

Hansen, K. M., Christensen, J. H., Brandt, J., Frohn, L. M., and Geels, C. (2004), Modelling atmospheric transport of alphahexachlorocyclohexane in the Northern Hemisphere with a 3-D dynamical model: DEHM-POP, Atmos. Chem. Phys. 4, 1125–1137. Harner T., Bidleman, T. F., Jantunen, L. M. M., and Mackay, D. (2001), Soil-air exchange model of persistent pesticides in the United States cotton belt, Environ. Toxicol. Chem. 20, 1612–1621. Horstmann, M. and McLachlan, M. S. (1998), Atmospheric deposition of semivolatile organic compounds to two forest canopies, Atmos. Environ. 32, 1799–1809. Howsam, M., Grimalt, J. O., Guino, E., Navarro, M., Marti-Rague, J., Peinado, M. A., Capella, G., and Moreno, V. (2004), Organochlorine exposure and colorectal cancer risk, Environ. Health Perspect. 112, 1460–1466. Ilyina, T., Pohlmann, T., Lammel, G., and S€undermann, J. (2006), A fate and transport ocean model for persistent organic pollutants and its application to the North Sea, J. Mar. Syst. 63, 1–19. ¨ ., Axelman, J., and Sundberg, H. (2003), J€ onsson, A., Gustafsson, O Global accounting of PCBs in the continental shelf sediments, Environ. Sci. Technol. 37, 245–255. Jurado, E., Lohmann, R., Meijer, S. N., Jones, K. C., and Dachs, J. (2004), Latitudinal and seasonal capacity of the surface oceans as a reservoir of polychlorinated biphenyls, Environ. Pollut. 128, 149–162. Kallenborn, R., Christensen, G., Evenset, A., Schlabach, M., and Stohl, A. (2007), Atmospheric transport of persistent organic pollutants (POPs) to Bjørnøya (Bear Island), J. Environ. Monit. 9, 1082–1091. Karickhoff, S. W. (1981), Semi-empirical estimation of sorption of hydrophobic pollutants on natural sediments and soils, Chemosphere 10, 833–846. Kerler, F. and Sch€ onherr, J. (1988), Accumulation of lipophilic chemicals in plant cuticles: Prediction from octanol/water partition coefficients, Arch. Environ. Contam. Toxicol. 17, 1–6. Klasmeier, J., Matthies, M., MacLeod, M., Fenner, K., Scheringer, M., Stroebe, M., Le Gall, A. C., Mckone, T., Van De Meent, D., and Wania, F. (2006), Application of multimedia models for screening assessment of long-range transport potential and overall persistence, Environ. Sci. Technol. 40, 53–60. Koziol, A. S. and Pudykiewicz, J. A. (2001), Global-scale environmental transport of persistent organic pollutants, Chemosphere 45, 1181–1200. Krauss, M., Wilcke, W., and Zech, W. (2000), Polycyclic aromatic hydrocarbons and polychlorinated biphenyls in forest soils: Depth distribution as indicator of different fate, Environ. Pollut. 110, 79–88. Lammel, G., Klopffer, W., Semeena, V. S., Schmidt, E., and Leip, A. (2007), Multicompartmental fate of persistent substances— comparison of predictions from multi-media box models and a multicompartment chemistry-atmospheric transport model, Environ. Sci. Pollut. Res. 14, 153–165. Lavelle, P., Chauvel, A., and Fragoso, C. (1995), Faunal activity in acid soils, in Plant Soil Interactions at low pH, Date, R. A. ed., Kluwer Academic Publishers, Dordrecht, pp. 201–211.

Lei, Y. D. and Wania, F. (2004), Is rain or snow a more efficient scavenger of organic chemicals? Atmos. Environ. 38, 3557–3571. Ma, J., Daggupaty, S., Harner, T., and Li, Y.-F. (2003), Impacts of Lindane usage in the Canadian prairies on the Great Lakes ecosystem. 1. Coupled atmospheric transport model and modeled concentrations in air and soil, Environ. Sci. Technol. 37, 3774–3781. Mackay, D. (1989), An approach to modelling the long term behavior of an organic contaminant in a large lake: Application to PCBs in Lake Ontario, J. Great Lakes Res. 15, 283–297. Mackay, D. (2001), Multimedia Environmental Models: The Fugacity Approach, Lewis/CRC, Boca Raton, FL. Malanichev, A., Mantseva, E., Shatalov, V., Strukov, B., and Vulykh, N. (2004), Numerical evaluation of the PCBs transport over the Northern Hemisphere, Environ. Pollut. 128, 279–289. Mariussen, E. and Fonnum, F. (2006), Neurochemical targets and behavioral effects of organohalogen compounds: An update, Crit. Rev. Toxicol. 36, 253–289. McLachlan, M. (1999), Framework for the interpretation of measurements of SOCs in plants, Environ. Sci. Technol. 33, 1799–1804. Meharg, A. A., and Cairney, J. W. G. (2000), Ectomycorrhizas— extending the capabilities of rhizosphere remediation? Soil Biol. Biochem. 32, 1475–1484. Meijer, S. N., Ockenden, W. A., Sweetman, A., Breivik, K., Grimalt, J. O., and Jones, K. C. (2003), Global distribution and budget of PCBs and HCB in background surface soils: Implications for sources and environmental processes, Environ. Sci. Technol. 37, 667–672. Muir, D. G. C., Omelchenko, A., Grift, N. P., Savoie, D. A., Lockhart, W. L., Wilkinson, P., and Brunskill, G. J. (1996), Spatial trends and historical deposition of polychlorinated biphenyls in Canadian midlatitude and arctic lake sediments, Environ. Sci. Technol. 30, 3609–3617. M€ uller-Lemans, H. and van Dorp, F. (1996), Bioturbation as a mechanism for radionuclide transport in soil: Relevance of earthworms, J. Environ. Radioact. 31, 7–20. Newton, I., Wyllie, I., and Asher, A. (1993), Long-term trends in organochlorine and mercury residues in some predatory birds in Britain, Environ. Pollut. 79, 143–151. Nizzetto, L., Cassani, C., and Di Guardo A. (2005), Deposition of PCBs in mountains: The forest filter effect of different forest ecosystem types, Ecotoxicol. Environ. Safety 63, 75–83. Pang, L. P., Close, M. E., Watt, J. P. C., and Vincent, K. W. (2000), Simulation of picloram, atrazine, and simazine leaching through two New Zealand soils and into groundwater using HYDRUS2D, J. Contam. Hydrol. 44, 19–46. Pasquill, F. (1961), The estimation of the dispersion of windborne material, Meteorol. Mag. 90, 33–49. Pekar, M. and van Pul, W. A. J. (1999), Modelling of lindane and PCB transport in the European region, Proc. EUROTRAC Symp. ’98 Garmisch-Partenkirchen, Borrell P. M. and Borrell, P., eds., Witpress, Southampton, UK pp. 380–384.

REFERENCES

Poster, D. L. and Baker, J. E. (1996), Influence of submicron particles on hydrophobic organic contaminants in precipitation. 2. Scavenging of polycyclic aromatic hydrocarbons by rain, Environ. Sci. Technol. 30, 349–354. Prevedouros, K., MacLeod, M., Jones, K. C., and Sweetman, A. J. (2004), Modelling the fate of persistent organic pollutants in Europe: Parameterisation of a gridded distribution model, Environ. Pollut. 128, 251–261. Raff, J. D. and Hites, R. A. (2007), Deposition versus photochemical removal of PBDEs from Lake Superior air, Environ. Sci. Technol. 41, 6725–6731. Scheringer, M. (1996), Persistence and spatial range as end-points of an exposure-based assessment of organic chemicals, Environ. Sci. Technol. 30, 1652–1659. Scheringer, M., Salzmann, M., Stoebe, M., Wegmann, F., Fenner, K., and Hungerb€uhler, K. (2004), Long-range transport and global fractionation of POPs: Insights from multimedia modeling studies, Environ. Pollut. 128, 177–188. Seibert, P., Kromp-Kolb, H., Baltensperger, U., Jost, D. T., and Schwikowski, M. (1994), Trajectory analysis of high-alpine air pollution data, in Air Pollution Modelling and Its Application, Vol. X, Gryning, S.-E. and Millan, M. M., eds., Plenum Press, New York, pp. 595–596. Semeena, V. S., Feichter, J., and Lammel, G. (2006), Impact of the regional climate and substance properties on the fate and atmospheric long-range transport of persistent organic pollutants— examples of DDT and c-HCH, Atmos. Chem. Phys. 6, 1231–1248. Simonich, S. L. and Hites, R. A. (1994), Vegetation—atmosphere partitioning of polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 28, 939–943. Spain, J. C., Pritchard, P. H., and Bourquin, A. W. (1980), Effects of adaptation on biodegradation rates in sediments/water cores from estuarine and freshwater environments, Appl. Environ. Microbiol. 40, 726–734. Stohl, A. (1996), Trajectory statistics—a new method to establish source–receptor relationships of air pollutants and its application to the transport of particulate sulfate in Europe, Atmos. Environ. 30, 579–587. Stohl, A. (1998), Computation, accuracy and applications of trajectories—a review and bibliography, Atmos. Environ. 32, 947–966. Stohl, A., Hittenberger, M., and Wotawa, G. (1998), Validation of the Lagrangian particle dispersion model FLEXPART against large-scale traces experiment data, Atmos. Environ. 32, 4245–4264. Stroebe, M., Scheringer, M., Held, H., and Hungerb€uhler, K. (2004), Inter-comparison of multimedia modelling approaches: Modes of transport, measures of long range transport potential and the spatial remote state, Sci. Total Environ. 321, 1–20. Suzuki, N., Murasawa, K., Sakurai, T., Nansai, K., Matsuhashi, K., Moriguchi, Y., Tanabe, K., Nakasugi, O., and Morita, M. (2004), Geo-referenced multimedia environmental fate model (GCIEMS): Model formulation and comparison to the generic model and monitoring approaches, Environ. Sci. Technol. 38, 5682–5693.

265

Tolls, J. and McLachlan, M. (1994), Partitioning of semivolatile organic compounds between air and Lolium multiflorum (Welsh ray grass), Environ. Sci. Technol. 28, 156–166. UNECE (1998), Protocol on Persistent Organic Pollutants Under the 1979 Convention on Long-Range Transboundary Air Pollution, United Nations Economic Commission for Europe (ECE/EB.Air/60). UNEP (2001), Stockholm Convention on Persistent Organic Pollutants, United Nations; available at http://chm.pops.int/Portals/0/ Repository/convention_text/UNEP-POPS-COP-CONVTEXTFULL.English.pdf UNEP (2008a) Endosulfan Proposal, United Nations; available at http://chm.pops.int/Portals/0/docs/from_old_website/documents/meetings/poprc/POPRC4/doc_e/POPRC4_doc_14_e.pdf UNEP (2008b) Hexabromocyclododecane as a Possible Global POP, United Nations; available at http://chm.pops.int/Portals/0/ Repository/poprc4/UNEP-POPS-POPRC.4-INF-15.English.pdf. UNEP, (2008c) Summary of Short-Chained Chlorinated Paraffins Proposal, United Nations; available at http://chm.pops.int/Portals/0/docs/from_old_website/documents/meetings/poprc/ POPRC2/POPRC2_doc_e/POPRC2_doc_14_e.pdf. UNEP (2009), Report of the Conference of the Parties of the Stockholm Convention on Persistent Organic Pollutants on the work of its fourth meeting, United Nations; available at http:// chm.pops.int/Convention/COP/hrMeetings/COP4/COP4Documents/tabid/531/language/en-US/Default.aspx. Wagner, J., Watts, S. A. and Kent, D. C. (1985), Plume 3D: ThreeDimensional Plumes in Uniform Ground Water Flow, Project Report CR81 1142, EPA 600/2–85–067, PB85–214443, Robert S. Kerr Environmental Research Laboratory, USEPA, Ada, OK. Wang, Y. Q., Tao, S., Jiao, X. C., Coveney, R. M., Wu, S. P., and Xing, B. S. (2008), Polycyclic aromatic hydrocarbons in leaf cuticles and inner tissues of six species of trees in urban Beijing, Environ. Pollut. 151, 158–164. Wania, F., Breivik, K., Persson, N. J., and McLachlan, M. S. (2006), CoZMo-POP 2—a fugacity-based dynamic multi-compartmental mass balance model of the fate of persistent organic pollutants, Environ. Model. Software 21, 868–884. Wania, F. and Mackay, D. (1993), Global fractionation and cold condensation of low volatility organochlorine compounds in polar regions, Ambio 22, 10–18. Wania, F. and Mackay, D. (1996), Tracking the distribution of persistent organic pollutants, Environ. Sci. Technol. 30, A390–A396. Wania, F. and Mackay, D. (1999a) The evolution of mass balance models of persistent organic pollutant fate in the environment, Environ. Pollut. 100, 223–240. Wania, F. and Mackay, D. (1999b) Global chemical fate of a-hexachlorocyclohexane. 2. Use of a global distribution model for mass balancing, source apportionment, and trend prediction, Environ. Toxicol. Chem. 18, 1400–1407. WWF (World Wildlife Foundation) (2005), Stockholm Convention: “New POPs.” Screening Additional POPs Candidates; available at http://assets.panda.org/downloads/newpopsfinal.pdf.

11 EMISSION OF POLYCYCLIC AROMATIC HYDROCARBONS IN CHINA SHU TAO, BENGANG LI, YANXU X. ZHANG,

AND

HUISHI YUAN

11.1. Introduction 11.2. Energy Consumption Modeling 11.2.1. Estimated Sources for PAH Emissions 11.2.2. Modeling the Geographic Distributions of Energy Consumption 11.2.3. Temporal Variations of Energy Consumption 11.2.4. Modeling Temporal Variations of Energy Consumption 11.3. Emission Inventory of PAHs 11.3.1. Emission Factors 11.3.2. Emission Inventory 11.3.3. Geographic Distribution 11.3.4. Temporal Change 11.4. Summary

11.1. INTRODUCTION Polycyclic aromatic hydrocarbons (PAHs) are chemical compounds that consist of two or more benzene rings that are fused together by sharing a pair of carbon atoms. They are ubiquitous in the environmental media of atmosphere, soil, water, sediment, and animals. Because of their persistence, carcinogenicity, and durability for long-range transport, much concern was raised regarding the emission and fate of PAHs (Keith and Telliard 1979; UNECE 1998). Sixteen PAH compounds (PAH16) were specified by the United States Environmental Protection Agency (USEPA) as priority monitoring PAHs, including naphthalene (NAP), acenaphthylene (ACY), acenaphthene (ACE), fluorene (FLO), phenanthrene (PHE), anthracene (ANT), fluoranthene (FLA), pyrene (PYR), benz[a]anthracene (BaA),

chrysene (CHR), benzo[b]fluoranthene (BbF), benzo[k] fluoranthene (BkF), benzo[a]pyrene (BaP), dibenz[a,h]anthracene (DahA), benzo[g,h,i]perylene (BghiP), and indeno [1,2,3-cd]pyrene (IcdP). Their molecular structures are shown in Figure 11.1. The PAH7 are the sum of seven PAH compounds, including BaA, CHR, BbF, BkF, BaP, IcdP, and DahA, which have more benzene rings among the 16 PAH compounds. Carcinogenicity increases with the number of benzene rings in the compound. Polycyclic aromatic hydrocarbons are produced primarily by incomplete combustion by natural and anthropogenic processes. The natural sources for PAHs include volcanic activity and forest fires, while anthropogenic sources include vehicular emissions, coal combustion, coke production, and biomass open burning (Xu et al. 2006). Polycyclic aromatic hydrocarbons can exist in the environment by gaseous phase and solid phase with particles. The PAHs in urban areas have gained more recent attention (Narvaez et al. 2008). Concentrations of PAH in urban areas are higher than in suburban areas because of the intensive carbon-emitting human activities, and thus there is a higher risk of PAH exposure in densely populated areas. Land-based and atmosphere-based PAHs can be transferred into aquatic environments by rainfall, contaminating the sediments, rivers, and estuaries. Fluxes of PAHs from the atmosphere to the estuary were also observed (Tsai et al. 2002). The abundant globally collected data on PAHs in rivers, lakes, estuaries, harbors, and coastal areas reveal sedimentary PAH concentrations ranging from 1 to 760,000 ng/g with 1000–10,000 ng/g as typical levels (Zakaria et al. 2002). Most PAHs are directly emitted into the atmosphere. Because of their robustness to long-range transport and deposition, PAHs are found in the environmental media of various pristine regions. A study by Ohkouchi and colleagues

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

267

268

EMISSION OF POLYCYCLIC AROMATIC HYDROCARBONS IN CHINA

Naphthalene

Acenaphthylene

Acenaphthene

Fluorene

Phenanthrene

Anthracene

Fluoranthene

Pyrene

Benz[a]anthracene

Chrysene

Benzo[b]fluoranthene

Benzo[k]fluoranthene

Benzo[a]pyrene

Indeno[1,2,3-cd]pyrene

Dibenz[a,h]anthracene

Benzo[ghi]perylene

Figure 11.1. Molecular structures for PAH16.

suggested that most of the PAHs that they found in deep-sea surface sediments were traced from fossil fuel combustion and/or biomass burning, indicating the long-range transport of PAHs (Ohkouchi et al. 1999). Efforts have been made to estimate and reduce the amount of PAH emission in various countries. Shrinking PAH emissions were observed in developed countries. The total emissions of BaP for 33 European countries were estimated to have decreased from 1300 tons in 1970 to 590 tons in 1995 (Pacyna et al. 2003). Seven carcinogenic PAHs in the United States were estimated to have decreased by 600 tons from 1990 to 1996 (USEPA 1998). With growth in the economies of and consequently growth in energy consumption in developing countries, PAH emissions in those areas are anticipated to increase. Polycyclic aromatic hydrocarbons are chiefly byproducts of incomplete combustion of fossil fuels and biomass and pyrosynthesis of organic materials. Increased population and rapid economic growth since 1980s have led to a rapid increase in energy consumption in China. Since the 1980s, the demand of energy in China has expanded dramatically. It has been reported that China became the second largest energy consumer in the world in 2003, following the United States (Crompton and Wu 2005). The consumption of coal and petroleum in China was 1660 million and 287 million tons per year in 2003 (NBSC 2004a,b). Otherwise, biomass is burned widely in

rural regions in China. About 537 million tons of biomass was burned in homes and other buildings and in fields in 2003 (NBSC 2004b). High PAH concentrations were observed in China (Mai et al. 2002). Because of the possible emission of a large amount of PAHs and potential long-range atmospheric transport, PAH emission in China is of both regional and global significance. It is necessary to estimate the amount of PAH emissions in China to better understand the fate and effects of PAHs in that country. For emission estimation of a pollutant, an emission inventory of a particular activity can be compiled by multiplying the activity rate by the corresponding emission factor (EF). Such an approach was used for various pollutants, including PAHs (Tsibulsky et al. 2001; Pacyna et al. 2003; Breivik et al. 2004). In our study, an emission inventory of PAHs in China was established, based on the rates of major PAHs emission activities and the EF values collected from the literature. The major emission sources of PAHs in China are burning of biomass, including straw and firewood, combustion for heating in homes and businesses, industrial combustion of fossil fuels, petroleum used for transportation, and production of coke and aluminum (Xu et al. 2006). Energy consumption, together with its spatiotemporal variations, was modeled for major PAHs emission sources. Then emission rates were calculated from consumption data for all counties in China for the 16 PAH compounds mentioned above.

ENERGY CONSUMPTION MODELING

11.2. ENERGY CONSUMPTION MODELING The emission of PAHs can be estimated by multiplying the emission activity rate by an emission factor. This approach was widely used in the pollutant emission estimation, including PAHs (Breivik et al. 2004; Pacyna et al. 2003; Tsibulsky et al. 2001). 11.2.1. Estimated Sources for PAH Emissions Incomplete combustion is the primary source of PAH. A preliminary investigation was conducted to determine the potential emission sources in China. Potential candidates were evaluated on the basis of data availability and significance. The emission sources for PAHs that have been evaluated include large- and small-scale coke production, firewood burning, open and indoor straw burning, domestic and industrial coal combustion, traffic oil combustion, aluminum electrolysis, consumer product usage, gasoline distribution, and industrial oil and petroleum refinery operation (Zhang et al. 2008). 11.2.2. Modeling the Geographic Distributions of Energy Consumption High-resolution data on the geographic distribution of emission and energy consumption are not yet available in China. Energy consumption is a reflection of socioeconomic activities. Consequently, quantitative relationships between energy consumption and some socioeconomic parameters are expected. These kinds of relationships have also been reported in the literature (Nathwani et al. 1992). In our study, by modeling the quantitative relationships between energy consumption and socioeconomic parameters on the provincial level in China, energy consumption at the resolution of 1
1 km2 was computed. This high-resolution energy consumption pattern, together with emission factors, provides a basis for constructing an inventory of high-spatial-resolution PAH emissions (see Section 11.3). The model has been described in detail in our previous study (Zhang et al. 2007). 11.2.3. Temporal Variations of Energy Consumption Seasonal variability of atmospheric PAH concentrations were widely reported in China. For example, PAH concentrations in Beijing during autumn and winter seasons were much higher than those in the other two seasons (Zhou et al. 2005). Except for lower temperature, decreased wet deposition, lower photolysis and radical degradation, high emissions during winter in northern China are other important factors for high PAH concentrations (Guo et al. 2003b; Zhou et al. 2005; Tan et al. 2006). The total residential energy consumption in Beijing in the summer half-year of 2006 was

269

3.5
106 tce (tons of coal equivalent), compared with 5.4
106 tce in the winter half-year (Beijing Statistical Bureau, http://www.bjstats.gov.cn). In addition, rural family statistics data showed that straw consumption increased from 200 during summer to 400 kg per month per family in winter in Jilin Province (Qin et al. 2007). Similar seasonal variations were reported in regions of Shenyang, Dalian, Qingdao, Nanjing, Guangzhou, and Hong Kong (Guo et al. 2003a,b; Wan et al. 2006; Wang et al. 2007; Tan et al. 2006; Tang et al. 2005). Seasonal variation for energy consumption was an important factor in the PAH emission inventory. An approximately five-fold variance for the hottest months in comparison to the coldest months in Asia was assumed by Streets et al. (2003). Similar high seasonal variation was assumed in other emission inventory research (Liousse et al. 1996). Some of the emission sources in China, exhibit wide seasonal variations, such as residential combustion of biofuel and coal, open burning of agricultural wastes, and wildfires. The variations for residential combustion can be attributed primarily to space heating. Because of the need for heating in winter in northern China, energy for residential consumption increases in winter. Open burning of straw was more concentrated in May, June, and July in China, especially in June over the North China Plain (Fu et al. 2007). Wildfires, as natural emission sources of PAHs, are subject to strong influence of seasonal weather conditions. For example, in 2002, 54,000 ha (hectares) of land in China was burned in September, much higher than the 180 ha burned in December (Giglio et al. 2003). The PAH emissions in 2003 were used as an example to further evaluate the seasonal variations for different sources for PAH emission in China, which could be regarded as evidence of the environmental impact of PAHs, emphasizing the need for a PAH control strategy, which could be achieved by obtaining basic information on PAHs through transport modeling with temporal resolution. 11.2.4. Modeling Temporal Variations of Energy Consumption Among the emission sources described above, industrial sources were relatively stable and were assumed to remain constant throughout the year. Biofuel and coal combustion in the residential sector, coal consumption for residential space heating in centralized heat-generating facilities, open burning of agricultural wastes, and wildfires were considered as factors for the seasonal variations for PAH emissions. Given that the available statistical data on energy consumption in China were obtained only on an annual basis, there was a need to develop additional models simulating the monthly or seasonal consumption of some of the energy. For the residential sector in China, energy is used mainly for cooking and heating. Energy used for cooking was

270

EMISSION OF POLYCYCLIC AROMATIC HYDROCARBONS IN CHINA

assumed to be constant within the year. Whether space heating is conducted depends on the location, which is temperature-dependent. Space heating is used mainly in northern rather than southern China. Firewood, straw, and coal are the conventional energy sources for cooking and heating in rural areas in China (Edwards et al. 2007). A regression model was developed to model the variation in per capita residential energy consumption with temperature as an independent parameter. The details for the model are presented in Section 11.24.1. In compliance with governmental regulations, heat would be provided only when the ambient temperature fell below 5 C. For most urban areas, heat is provided by central heat-generating facilities. To simplify, the amount of coal used for heat-generating facilities in the urban areas was evenly distributed for each month during the central heat-delivering period. The open burning of agricultural wastes occurs during the harvest seasons. The emission amounts from open burning of agricultural wastes are assumed to be zero during the remainder of the year. The harvest periods for different regions of China are different. Data on harvest periods in China for different areas were collected from the literature. Only the open burning of wheat, rice, and corn straw was considered in our study, as they contributed most to the cultivated areas (NBSC 2005). The Global Fire Emission Database was used to create a model of the monthly amount of biomass consumed by forest and grassland fires in China (Giglio et al. 2003). The monthly emission for each of these sources can be calculated by multiplying the monthly energy consumption strengths by the emission factors. These emission factors were considered to be constant throughout the year. 11.2.4.1. Model for Rural Residential Energy Consumption. The rural per capita energy consumption for residential use including coal, firewood, and straw for all the provinces in China during 2003 is listed in Table 11.1. On the basis of the data in Table 11.1, a linear regression model can be constructed from the provincial data after deleting several outliers. The following linear relationship was used to predict per capita residential energy consumption by annual mean temperature: Eav ¼ 8:29
103 Tav þ 0:406

ðr2 ¼ 0:32; n ¼ 22Þ

TABLE 11.1. Rural Per Capita Energy Consumption for Residential Use Including Coal, Firewood, and Straw with Annual Average Temperature for Each Province in China During 2003 Province

Average Annual Temperature,  C

0.30 0.78 0.35 0.46 0.30 0.22 0.23 0.57 0.40 0.39 0.44 0.17 0.32 0.19 0.53 0.22 0.20 0.40 0.39 0.34 0.27 0.26 0.24 0.04 0.36 0.50 0.27 0.25 0.31 0.29 0.11

16.93 11.93 18.04 19.77 8.76 22.62 19.79 15.42 25.17 14.73 5.62 15.25 17.84 18.24 7.14 16.63 18.23 7.07 9.46 10.14 6.22 14.52 14.00 17.51 10.59 17.22 13.47 5.17 3.65 14.82 17.71

Anhui Beijing Chongqing Fujian Gansu Guangdong Guangxi Guizhou Hainan Hebei Heilongjiang Henan Hubei Hunan Inner Mongolia Jiangsu Jiangxi Jilin Liaoning Ningxia Qinghai Shaanxi Shandong Shanghai Shanxi Sichuan Tianjin Tibet Xinjiang Yunnan Zhejiang a

Ton of coal equivalence.

to evaluate the monthly energy consumption. For month i, the energy consumption for the residential sector can be calculated by Ei ¼ (8.29
103Ti þ 0.406)/12, where Ei is the energy consumption for month i and Ti is the mean temperature in month i. The proportion of energy consumed during month i (fi) was calculated by the following equation:

ð11:1Þ Here, Eav is the average annual rural residential energy consumption per capita for a province [ton of coal equivalence (tec) per capita per year], and Tav is the corresponding average annual mean temperature of the province (in  C). Details on how the model was established and validated can be found in Zhang and Tao (2008). Equation (11.1), which was originally intended to describe the quantitative relationship on an annual basis, is used

Annual Per Capita Energy Consumption, tcea

fi ¼

ð8:29
103 Ti þ 0:406Þ=12 8:29
103 Tav þ 0:406

ð11:2Þ

When Ti was higher than 20 C, a value of 20 C was assigned in the calculation instead of the actual temperature. Furthermore, the separate monthly proportions for the three fuels of coal, firewood, and straw were assumed to be the same. Using Equation (11.2), the monthly proportions of residential energy consumption for each province are calculated

ENERGY CONSUMPTION MODELING

271

TABLE 11.2. Monthly Proportion of Residential Energy Consumption in Each Province Calculated from Equation (11.2) Monthlya Proportions Residential Energy Consumption/% Province

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

STD

Anhui Beijing Chongqing Fujian Gansu Guangdong Guangxi Guizhou Hainan Hebei Heilongjiang Henan Hubei Hunan Inner Mogolia Jiangsu Jiangxi Jilin Liaoning Ningxia Qinghai Shaanxi Shandong Shanghai Shanxi Sichuan Tianjin Tibet Xinjiang Yunnan Zhejiang

11.4 11.9 10.6 10.1 11.7 9.8 10.9 11 8.5 11.7 13 11.5 11.3 11.3 12.3 11.3 11.3 12.9 12.6 12.1 11.5 11.5 11.7 11 11.9 10.7 11.9 9.8 12.1 9.8 11.1

10.2 10.5 10 9.9 10.2 9.3 9.9 10.2 8.3 10.4 10.8 10.3 10.1 10 10.6 10.3 10 10.7 10.7 10.4 9.9 10.2 10.6 10.3 10.5 10.1 10.6 9.6 10.7 9.5 10.2

9.6 9.3 9.3 9.4 9.5 8.3 9.3 9.4 8.3 9.3 9.4 9.4 9.6 9.6 9.6 9.7 9.7 9.5 9.4 9.6 9.4 9.7 9.5 9.8 9.5 9.4 9.3 9.2 9.7 8.5 9.7

7.4 7.5 7.3 7.6 7.3 7.9 7.5 7.4 8.3 7.4 7.2 7.5 7.2 7.2 7.4 7.5 7.2 7.1 7.3 7.3 7.4 7.4 7.6 7.8 7.4 7.2 7.6 8.5 7.9 8 7.4

7 6.6 7.3 7.6 7.2 7.9 7.5 7.1 8.3 6.6 6.2 6.9 7.2 7.2 6.9 7 7.2 6.3 6.5 7 7.2 7.2 7 7.2 7 7.2 6.5 7.5 6.9 7.1 7.2

7 6.3 7.3 7.6 6.2 7.9 7.5 7 8.3 6.6 5.5 6.8 7.2 7.2 5.8 7 7.2 5.7 6 6.1 6.5 6.7 6.6 7.2 6.2 7.2 6.5 6.7 5.6 7.2 7.2

7 6.3 7.3 7.6 6 7.9 7.5 7 8.3 6.6 5.5 6.8 7.2 7.2 5.8 7 7.2 5.7 6 6.1 6.2 6.7 6.6 7.2 6.2 7.2 6.5 7.1 5.6 7.2 7.2

7 6.3 7.3 7.6 6.4 7.9 7.5 7 8.3 6.6 5.5 6.8 7.2 7.2 6 7 7.2 5.7 6 6.1 6.5 6.7 6.6 7.2 6.2 7.2 6.5 7.2 5.7 7.4 7.2

7 6.3 7.3 7.6 6.8 7.9 7.5 7 8.3 6.6 6.3 6.8 7.2 7.2 6.4 7 7.2 6.3 6.2 6.4 7 6.7 6.6 7.2 6.6 7.2 6.5 7.3 6.7 7.7 7.2

7.3 7.7 7.5 7.6 8.4 7.9 7.5 7.8 8.3 7.5 7.7 7.4 7.2 7.2 8.2 7.3 7.2 7.7 7.6 8.2 8.3 7.8 7.4 7.2 8.1 7.7 7.4 7.9 8.2 8.3 7.2

8.5 10 8.6 8.2 9.5 7.9 8.2 8.9 8.3 9.5 11 9.2 8.4 8.4 9.9 8.6 8.3 10.7 10.3 9.6 9.5 9 9.1 8.3 9.6 8.6 9.7 9 9.8 9.2 8.4

10.4 11.1 10.1 9.3 10.7 8.9 9.4 10 8.4 10.9 11.9 10.6 10.2 10 11.2 10.3 10 11.7 11.5 11 10.6 10.4 10.8 9.9 10.9 10.1 11 10.1 11.2 10 10.1

1.6 2.1 1.3 1.0 1.9 0.7 1.2 1.5 0.1 1.9 2.8 1.8 1.5 1.5 2.3 1.6 1.5 2.6 2.4 2.1 1.8 1.7 1.9 1.5 2.1 1.4 2.0 1.2 2.3 1.1 1.5

and are shown in Table 11.2. The temporal variations in southern China were lower than those for northern China. For example, the lowest standard deviations were observed in the provinces of Guangdong (0.7) and Hainan (0.1), located in the southern most part of China. In contrast, the highest standard deviation (2.8) was observed in Heilongjiang Province, located in the northern most part of China. The proportion of residential energy consumption for January in Heilongjiang province is 13%, about 2.4 times the amount for summer. The proportion of January is 1.8 times the amount for July in Beijing. The ratio represents a typical scenario for northern China. The proportion of residential energy consumption for the other provinces were somewhere between the two extremes of most northern and southern parts. To validate Equation (11.1) some statistical datasets in Jilin Province were collected for comparison with the calculated data. Qin et al. (2007) conducted an energy consumption survey in the rural area of Jilin Province. They investigated some energy sources in five regions in Jilin.

Electricity and straw were found to be the major energy sources in rural Jilin. However, the electricity consumption remained stable throughout the year, while straw consumption in summer was much smaller than that in winter. The average per capita consumption of straw in winter is 80 kg/month, which was 2 times the amount for summer. The surveyed monthly proportions of straw consumption in Jilin were compared with the calculated proportions in Figure 11.2. The two datasets agree with each other well. Residential energy consumption statistical data for Beijing were also used here to validate the model, which was available on the Beijing Statistical Bureau Website (http://www.bjstats.gov.cn). As data for 2003 in Beijing were not available, the data for 2006 were adopted for comparison instead. The surveyed proportions for the four time periods of 2006 (i.e., Jan.–March, April–June, July–Sept., Oct.–Dec.) in Beijing were 30.77%, 20.25%, 18.91%, and 30.07%, compared with the model predicted data of 34.08%, 19.93%, 16.57%, and 29.43%, respectively. Again, the two datasets

272

EMISSION OF POLYCYCLIC AROMATIC HYDROCARBONS IN CHINA

16% 14% 12% 10% 8%

Surveyed Calculated

6% 4% 2% 0%

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 11.2. The surveyed and modeled monthly proportions to the annual volume in Jilin Province.

agreed well. The agreement between the modeled and the surveyed proportions demonstrated that the model can represent seasonal variations of residential energy consumption. To further validate the model, emission inventories from the literature were also introduced to highlight the seasonal variations. Cao et al. (2006) estimated the ratio of largest and lowest monthly emission to be 1.4-fold lower than other estimations. Streets et al. (2003) estimated an approximately five-fold variation between hottest and coldest months in all of Asia. The data from the Statistical Yearbook of China, suggest that the five-fold variance may not apply in China. Our estimated value lies between these two estimated levels, ranging from 1.0 to 2.4.

11.3. EMISSION INVENTORY OF PAHs 11.3.1. Emission Factors Because of the variability in origin of fuel type, combustion facilities and conditions, and measuring procedures, emission factors were often reported in different values. For each emission source there were 16 emission factors for PAH16. A great effort has been made to collect the emission factors for individual emission source of PAHs in our estimate (Lee et al. 2005; USEPA 1996; Jenkins et al. 1996; Yang et al. 1998; Oanh et al. 2002, 1999; Nielsen et al. 2001; Friedrich et al. 2002; Witherington et al. ; Gullett et al. 2003). After the outlier was deleted, the geometric mean was used. The emission factors for the emission sources, except for indoor straw combustion and small-scale coke production, were selected on the basis of the collection of data from the literature. The emission factors for indoor straw combustion were derived from the experiments in our lab. A preliminary investigation revealed straw to be one of the most significant PAH emission sources in China (Xu et al. 2006). Straw is usually burned in an open field as agricultural waste and in indoor cooking stoves in China (Zeng et al. 2007). Of the two major methods of straw burning, straw burning using in indoor cooking stoves was more significant than open burning. In 2004, 340 million tons of straw was burned in China

by conventional household stoves for cooking and indoor heating, which was about 60% higher than that for residential firewood burning (NBSC 2005). For an accurate evaluation, an experiment was conducted to determine the emission factors of straw in China. Cereal straw collected from the neighboring rural areas of Beijing was used in the experiment. Wheat was used to represent cereal straw, owing to its large planting area and output. Wheat straw was cut into small (5-cm-length) pieces and burned in the burning–sampling system (Fig. 11.3) in a separate room to simulate indoor straw combustion (Zhang et al. 2008). The analysis process and parameters were described by Zhang et al. (2008). The straw was burned on a plate, and a balance scale was used to continuously measure the fuel burn rate. The hood and the blower induced the yielded smoke into a pipe where the smoke was mixed with ambient air with turbidity. The PAH sampler (consisting of a prepositioned glass fiber filter and two polyurethane foam plugs in tandem) collected both the gaseous and particulate phases of the combustion gas stream from this pipe. (Zhang et al. 2008). For small-scale coke production, only the emission factor for BaP was found in the literature (Zhong 1999). For the other PAH compounds, emission factors for uncontrolled coke production were adopted as substitutes. This substitution was based on the assumption that the emission of the small-scale coke production under controlled conditions in China are similar to that using “beehive” coke ovens in the United States in the 1990s (Kirton et al. 1991; Bond et al. 2004). Similarity between the calculated and the reported emission factors for BaP partially confirmed this assumption (Zhong 1999; Kirton et al. 1991). The emission factors for each source are listed in Table 11.3. 11.3.2. Emission Inventory The emission inventory in 2003 was selected as a representative year for analysis. Using the parameters presented above, the emissions of PAH16 and PAH7 in China during 2003 were estimated to be 116,000 and 11,800 tons, respectively.

EMISSION INVENTORY OF PAHs

273

Exhaust 1.3 × 104 L.min−1 350 cm

Dilution Tunnel 13 cm

120 cm

Blower 260 cm

Hood

GFF GFF & PUF Sampling 122 cm

0.93 L.min−1

200 cm

TMP−1500 Air Sampler

PUF

Wheat Straw Plate

Platform Balance

Figure 11.3. Setup of the facilities for measuring PAH emission factors from indoor straw combustion. The straw is burned on a plate, and a balance is continuously measuring the fuel burn rate. The hood and the blower induced the yielded smoke into a pipe, where the smoke was mixed with ambient air with turbidity. The PAH sampler (consisting of a prepositioned glass fiber filter and two polyurethane foam plugs in tandem) collected both the gaseous and particulate phases of the combustion gas stream from this pipe (Zhang et al. 2008).

The PAH emissions from each source are shown in Figure 11.4. Indoor straw, small-scale coke production, firewood burning, and domestic coal combustion yield the largest PAH emissions, with contributions of 34.6%, 27.2%, 21.2%, and 6.8%, respectively. The remaining sources contributed merely 10.2% to the total PAH emissions, demonstrating a highly uneven pattern. Straw and firewood are the main energy resources for rural areas in China. Of the residential fuel in rural areas, 33%–45% was supplied by straw (NBSC 2005). In total, 620 million tons of straw was produced in China during 2002, and 45% of the straw was used as fuel (Zeng et al. 2007). Moreover, in rural areas the combustion of straw is conducted in conventional stoves, which have poor ventilation and lack of oxygen, thereby yielding relative high emission factors for indoor straw burning (Table 11.3) (Mastral and Callen 2000). The high emissions from these two sources demonstrated a risk of PAH exposure in rural areas. In contrast, traffic emissions which were reported to be the most significant source of pollution in metropolitan areas, contributed only 2.5% to the total emissions in China.

However, traffic still resulted in an emission density of PAH emissions in cities at a factor of 30 higher than that in rural areas. The proportion of small-scale coke production to the total emission is about 25 times the proportion for large-scale coke production. The PAH emissions from coal combustion and consumer products usage were low. 11.3.3. Geographic Distribution On the basis of the regression models for energy consumption, the provincial-level emission was broken down into a 1
1 km2 resolution emission density was distribution. High emission density was concentrated in North China Plain and southwestern China (Fig. 11.5). The major cities were emission centers with high emission densities compared with rural areas, even for western China, most of which are rural regions. Major cities, such as Harbin, Changchun, and Shenyang in northeastern China; Tianjin, Shijiazhuang, Hohhot, Urumqi, and Xining in the north and west; Guiyang and Kunming; and even Lhasa in Tibet, which is a low-emission province, are distinguished

274

a

1.9 0.6 0.5 0.1 1.3 0.2 0.6 0.6 0.1 0.2 0 0.1 0 0 0 0 6.1

28.8 1.2 0.1 0.3 1.3 0.1 0.6 0.5 0.2 0.1 0.1 0.1 0 0 0 0.1 33.5

Traffic

37.5 0.3 0.8 0.2 4.1 1 6.6 2.5 2.3 2.3 0.7 0.5 0.3 0 0.5 1 60.6

Wheat 7.4 — 0.6 0.3 1.3 0.2 0.4 0.3 0.1 0.1 0.1 0 0 0 0 0 11

Rice 109 33.2 11 22.6 40.7 15.2 26.4 26.3 15.8 17.9 5.4 7.8 6.6 2.5 0.4 2.6 343

Wheat

0.5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.6

Industrial

10.2 3.2 2.7 0.4 6.9 1.1 3.1 3 0.6 0.9 0.2 0.7 0.1 0 0 0 33.1

Crop

Indoor

Oil

11 2.2 1.8 0.7 2.3 0.4 1.6 1.1 0.2 0.2 0.1 0.1 0.1 0.3 0.2 0.1 22.3

Open Burning

Firewood

2.6 0 0 0 0.8 0 0 0.1 0 0 0 0 0 0 0 0 3.5

Petroleum Refinery

<0.3 0.2 3.2 1.6 7 1.1 2.1 1.6 0.5 0.6 0.5 0.2 0.2 0 0.2 0.2 59.5

Rice

Compound abbreviations are defined in the first paragraph of Section 11.1.

NAP ACY ACE FLO PHE ANT FLA PYR BaA CHR BbF BkF BaP IcdP DahA BghiP P16

NAP ACY ACE FLO PHE ANT FLA PYR BaA CHR BbF BkF BaP IcdP DahA BghiP P16

Crop

Open Burning

Straw

Biomass

TABLE 11.3. PAH Emission Factors for Individual PAH Emission Sourcesa

47.8 24.9 8.9 3.8 15 3.7 6.6 4.3 1.5 1.2 0.8 0.6 0.8 0.2 0.3 0.4 120.7

Indoor

2.6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2.6

15.7 6.1 11.3 4.3 11.3 1.9 4.2 2.5 1.9 3.7 1.6 1.6 1.1 1 1.9 2.1 72.1

Non Anthracite

Gasoline Distribution

0.1 0 0 0 0.1 0 0 0 0 0 0 0 0 0 0 0 0.4

Anthracite

Domestic

0 0.1 0 0.4 7.5 0.6 9.1 2.7 0.4 1.6 0.4 0.8 0.4 0.3 0.1 0.6 25.3

3.1 0.8 0.1 0.4 1.5 0.1 0.7 0.5 0.2 0.1 0.1 0.1 0.2 0.1 0 0.1 8.2

Large-Scale

63.8 3.5 60.4 40.5 185.5 57 117.5 44.5 6 12.2 14.7 9.8 15.1 2.5 1.2 4.9 639.1

Synchronous Anode

192.6 39.9 9.6 25.7 91.1 3.9 41.1 34.5 11.7 8.4 9.3 5.2 10.1 4.3 1 5.9 494.3

Small-Scale

Coke Production

Aluminum Electrolysis

1 0 0.1 0 0 0 0.1 0 0 0 0 0 0 0 0 0 1.4

Other

Prebaked Anode

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Power

Industrial

Coal

EMISSION INVENTORY OF PAHs

275

Large-scale coke production 1.1% Small-scale coke production 27.2%

Domestic coal combustion 6.8% Industrial coal combustion 1.5% Traffic oil combustion 2.5% Aluminum electrolysis 0.9% Consumer products usage 0.9% Others 0.9%

Firewood burning 21.2%

Open straw burning 2.4%

Indoor straw burning 34.6%

Figure 11.4. Relative contributions of various sources to the total emission of PAH16 in China in 2003 (Zhang et al. 2008).

Figure 11.5. Emission density for PAH16 in China in 2003 (Zhang et al. 2008).

as significant sources. High PAH emissions are also obvious in the two important industrial zones of the Yangtze and Pearl River Delta. The spatial pattern implied a relationship between emission densities and population, which has been demonstrated in North America with a good agreement between atmospheric PAH concentrations and population (Hafner et al. 2005). The industrial area with high PAH emission is also shown on the map, in Figure 11.5, such as Guizhou and Shanxi. Both provinces are rich in coal, and a

large quantity of PAHs were emitted during the coke production process. 11.3.4. Temporal Change 11.3.4.1. Annual Variations. The increase in PAH16 emission and energy consumption from 1950 to 2005 is shown in Figure 11.6. Before 1990 the energy consumption and PAH emission increased stably. Straw, firewood, domestically

276

EMISSION OF POLYCYCLIC AROMATIC HYDROCARBONS IN CHINA 140000 Gasoline distribution Open straw burning Petroleum refinery Aluminum electrolysis Industrial oil combustion Traffic oil combustion Industrial coal combustion Small-scale coke production Large-scale coke production Domestic coal combustion Indoor straw burning Firewood burning

3000 2500 2000 1500 1000

100000 80000 60000 40000

2005

2000

1995

1990

1985

1980

1975

1970

1965

1960

1955

2005

2000

1995

1990

1985

1980

1975

1970

1965

1960

1955

0

1950

20000

500 0

Others Small-scale coke production Domestic coal combustion Indoor straw burning Firewood burning

120000

PAH16 emission, tons

3500

1950

Source strength, million tons

4000

Figure 11.6. Temporal changes in the annual emission activities (left) and PAH16 emission (right) in China from 1950 to 2005 (Zhang et al. 2008).

combusted coal, and coal for small-scale coke ovens, which were the major sources of PAH in 1990, were 3.0, 2.9, 6.0, 102.3 times the amounts in 1950. Meanwhile, the PAH emission increased from 19,000 tons in 1950 to 89,000 tons in 1990. The fluctuation in emission around 1995 can be attributed to the change in small-scale coke production during the same period, due to the regulatory influence of the coal industry (China Law Press 2001). Although industrial coal combustion became dominant in the energy consumption structure with the rapid economic development during that period, its emission was not enormous because its emission factor was low. The predictions from an authoritative organization indicate that the energy consumed in China is expected to increase in the next two decades, with growth rates of coal and biomass consumptions in China at 2.3% and 0.3% per year, respectively (The World Energy Outlook 2004, http://www.iea.org). Given that the two energy types are now the major emission sources, the increment in both two types of energy suggested an increasing scenario for PAH emissions in the future. Meanwhile, with the requirement for low carbon emissions and a clean environment for the whole Chinese society, some regulations have been imposed to improve the energy structure and thereby to lower the PAH emissions. For instance, the criterion for small-scale coke production is stricter now than that in the past, and 40% of these operations have been shut down (Huang 2006). Small-scale coke production has been phased out since 2010, which largely reduces the PAH emissions in China. The government also launched the National Improved Stove Program in the 1990s, in an attempt to improve energy efficiency. By the end of 2000, around 80% of the total stoves in rural areas were substituted by improved ones (Zeng et al. 2007). The improved stoves save about 20% fuel (1 ton per stove annually), which conserves biomass energy used in rural areas and reduces PAH emissions at the same time. As the number of conventional stoves decreases and only a small number of conventional stoves remain, the effect of the

improved stove approach seems to be limited (Zeng et al. 2007). There are some methods for transferring straw to other kinds of energy to reduce the amount of straw being burned directly; examples are straw gasification, anaerobic digestion, straw briquette, straw liquefaction, straw carbonization, and biocoal. Some of these approaches have been adopted in China. By 2000, 388 straw gasification facilities for central gas supply and 7.6 million household biogas digesters had been built in China (Lin 1998; NBSC 2005; Zeng et al. 2007). With these approaches, PAH emissions have been reduced significantly. 11.3.4.2. Seasonal Variations. The PAH emissions in 2003 in China was studied up for seasonal variation analysis. The monthly PAH emissions from different sources during 2003 are illustrated in Figure 11.7. In China, winter seasons include December, January, and February, while summer includes June, July, and August. For the total emissions (Fig. 11.7a), higher emissions occurred in winter, with a peak value in January. In the warmer seasons the PAH emissions were much lower; the lowest months were May, June, and September. However, there is still a minor peak in summer seasons in July and August. This high-in-winter/ low-in-summer temporal distribution pattern was caused primarily by residential emissions, which were closely related to the change in air temperature. The residential PAH emissions include the sources from domestic biofuel, coal, and industrial coal burning for space heating. These sources were the major contributors to the total PAH emissions and also exhibited strong seasonal variations (Zhang et al. 2008). As shown in Figure 11.7b, biofuel combustion is an important source of seasonal variation, followed by biomass burning. Biomass burning peaks in July and August. Biomass burning includes the open burning of agricultural residues and forest wildfires and grassland fires. The agriculture residues are usually burned in an open field in order to conserve time and labor for planting for the next season in

EMISSION INVENTORY OF PAHs

277

Figure 11.7. Monthly PAH emissions (Zhang and Tao 2008).

China (Cao et al. 2006). The harvest time in different regions across China depends on the type of plant and location. For example, corn straw is burned from September to October in northern China and from July to August throughout southern China, respectively. There are several harvest times for rice straw for the semitropical and tropical climate zones during the year. In the Guangdong and Hainan Provinces of southern China, rice straw is burned in March and April, whereas it is burned in July and August in other parts of southern China, and in August and September in northern China (EBCAY 2005). The PAH emitted by open straw burning only accounts for 2.4% of the total emissions in China during 2003. However, the burning of agricultural wastes occurred in short periods, making it a significant contributor to the seasonal variations of PAH emissions. Wildfires took place in different regions, whose emissions are strongly influenced by precipitation, wind, and other climatic factors. Precipitation and snowcover will depress the occurrence of wildfires. There are strong winds in the spring and large accumulated biomass during the summer and the fall, which tend to yield outbursts of wildfires (Sinomaps Press 2005). Consequently, enormous wildfires have occurred during spring and autumn (Yuan et al. 2008). However, with the small contribution to the total emissions, wildfires contributed less to the seasonal variations than did open straw burning.

a monthly temperature in January as low as 20 C. The low temperature in these areas resulted in the use of a large part of the total residential energy for space heating. Consequently, significant seasonal variations for residential emissions in these areas were observed (Fig. 11.9). There are large areas of forest in NE China. The forests cover a total area of over 35
106 ha, with a forest coverage rate as high as 37%, which is more than twice that of the average forest cover ratio of China (NFBC 2005). Forest fire PAH emissions contributed to 15% of the total emissions in NE China in the spring and more than 25% in May. Xinjiang in NW China is the major crop production region, due to the largest provincial yield of cotton and sugar beets (EBCAY 2005). The PAH emission from burning of agricultural wastes was a significant contributor in Xinjiang, as this region has a relatively low population density. In both NW China and the Tibetan Plateau, PAH emissions peaked during July and August. Northern China has the largest small-scale coke production (Zhang et al. 2008), which makes a significant contribution to PAH emissions with little seasonal variation, and masks the seasonal variations of other emission sources. For southern and southwestern (SW) China, biofuel consumption is the major reason for the seasonal variation since the air temperatures in these areas are high throughout the year and there is no need for space heating.

11.3.4.3. Seasonal Variations in Different Regions. Climate, landcover, and agricultural and residential activities vary markedly in China. The climate types in China include tropical and semitropical in the south, frigid -temperate in the north, monsoon in the east, and continental climate in the west (Sinomaps Press 2005). To evaluate the seasonal variations in China, the territory was divided into six regions (northeastern, northern, northwestern, southern, southwestern, China and the Tibetan Plateau; see Fig. 11.8) according to geographic location and socioeconomic conditions. Northeastern (NE) and northwestern (NW) China are located at the highest latitude among the six regions with

11.3.4.4. Spatial Distribution During Different Seasons. The spatial distributions for domestic biofuel combustion, domestic coal combustion, and open biomass burning in different seasons are shown in Figure 11.10. The PAH emission density for domestic biofuel and domestic coal, shown in the two columns on the left, exhibited similar spatial patterns. The burning of biofuel is generally concentrated in rural areas, while coal is commonly used in both urban and rural areas (NBSC 2005). Urban areas cover only 0.15% of the total area of China. In 2003, of the total coal consumption used for domestic heating, 36% was used in urban areas (NBSC 2005). Therefore, a much higher

278

EMISSION OF POLYCYCLIC AROMATIC HYDROCARBONS IN CHINA

Figure 11.8. Six zones for analysis of seasonal variations on different regions (zones shown in blue, green, yellow, purple, brown, and orange represent northwestern, northeastern, Tibetan Plateau, northern, southwestern and southern China, respectively) (Zhang and Tao 2008). (See insert for color representation of this figure.)

emission density from domestic burning was observed in urban areas (National Construction Agency of China, http:// www.cin.gov.cn/). The areas of high emission density in urban areas can be clearly observed in the domestic coal maps (Fig. 11.10, middle column) but not in the combustion biofuel maps (Fig. 11.10, left column). The spatial distribution for the emission density of open biomass burning (Fig. 11.10, right column) changed drastically from spring to winter. Among the various kinds of crops, wheat straw has 140

600

7000

300

N China

NW China

NE China

PAHs Emission, Mg/y

a higher PAH emission factor. Consequently, the regions where wheat was planted, such as Northern China Plain and the Sichuan Basin, have a high emission density. The Northern China Plain and the Sichuan Basin accounted for 64% of the wheat yield in China in 2003 (EBCAY 2005). In southern China large quantities of rice straw are burned in the fields, but the emission density for this area is relatively low because rice straw has a low PAH emission factor (Zhang et al. 2008). For the open biomass burning, its emission density during the

3500

70

Total emission Biofuel combustion 0

J F M A M J J A S O N D

0

J F M A M J J A S O N D

J F M A M J J A S O N D

S China

SW China

Domestic combustion Biomass buring

250

1800

3000

0

Tibetan Plateau

Industrial coal Other

0

125

900

1500

J F M A M J J A S O N D

0

J

F M A M J J A S O N D

0

J F M A M J J A S O N D

Figure 11.9. The monthly PAHs emission rate of several major sources in different regions of China (Zhang and Tao 2008).

SUMMARY

279

Figure 11.10. PAH emission density maps for the major emission sources in different seasons (Zhang and Tao 2008). (See insert for color representation of this figure.)

colder half-year is much lower than that in the warmer halfyear. On the PAH emission density maps for open biomass burning (Fig. 10.11, right column), high emission density appears in NE China in spring. This can be attributed to the wildfires during that period.

11.4. SUMMARY Quantitative relationships among social, economic, and climate parameters and energy consumption for Chinese provinces provide data for PAHs emission calculation at a resolution of 1
1 km2. A linear regression model was

established for calculation of rural residential energy consumption. Models were validated using available energy data of Jilin and Beijing from the literature. On the basis of these regression models, spatiotemporal variations of energy consumption were analyzed and discussed. Emission factors were carefully generated for each emission source and for 16 PAH compounds, based on literature data and experiment results. Then, the emission inventory of PAH16 was established. The emissions of PAH16 and PAH7 in China during 2003 were estimated to be 116,000 and 11,800 tons, respectively. The PAH emissions from indoor straw, small-scale coke production, firewood burning, and domestic coal combustion yield the largest PAH emissions,

280

EMISSION OF POLYCYCLIC AROMATIC HYDROCARBONS IN CHINA

with contributions of 34.6%, 27.2%, 21.2%, and 6.8%, respectively. The results showed that the magnitude of seasonal variation of PAH emission in China was estimated at about 1.3fold between the months with the largest and the smallest emissions. Residential solid fuel combustion, including straw, firewood, and coal, dominated the pattern of the overall seasonal variation, with a peak in the wintertime. Both PAH emissions and their temporal variations showed a clear distribution in space, depending on factors such as local climate, population density, and agricultural activities. This emission inventory of PAHs established in our study provides basic information for policymakers to improve control of PAH emission and help researchers better understand the magnitude of the seasonal variation in PAH emission in China.

REFERENCES Bond, T. C., Streets, D. G., Yarber, K. F., Nelson, S. M., Woo, J., and Klimont Z. (2004), A technology-based global inventory of black and organic carbon emissions from combustion, J. Geophys. Res. 109, D14203. Breivik, K., Alcock, R., Li, Y. F., Bailey, R. E., Fiedler, H., and Pacyna, J. M. (2004), Primary sources of selected POPs: Regional and global scale emission inventories, Environ. Pollut. 128, 3–16. Cao, G. L., Zhang, X. Y., and Zheng, F. C. (2006), Inventory of black carbon and organic carbon emissions from China, Atmos. Environ. 40, 6516–6527. Chen, Y. J., Sheng, G. Y., Bi, X. H., Feng, Y. L., Mai, B. X., and Fu, J. M. (2005), Emission factors for carbonaceous particles and polycyclic aromatic hydrocarbons from residential coal combustion in China, Environ. Sci. Technol. 39, 1861–1867. China Law Press (2001), Law of the People’s Republic of China on the Coal Industry, China Law Press, Beijing. Crompton, P. and Wu, Y. (2005), Energy consumption in China: Past trends and future directions, Energy Econ. 27, 195–208. EBCAY (2005), China Agricultural Yearbook 2004, China Agricultural Press, Beijing. Edwards, R. D., Liu, Y., He, G., Yin, Z., Sinton, J., Peabody, J., and Smith, K. R. (2007), Household CO and PM measured as part of a review of China’s National Improved Stove Program, Indoor Air 17, 189–203. Friedrich, R., Wickert, B., Blank, P., Emeis, S., Engewald, W., Hassel, D., Hoffmann, H., Michael, H., Obermeier, A., Schafer, K., Schmitz, T., Sedlmaier, A., Stockhause, M., Theloke, J., and Weber, F. J. (2002), Development of emission models and improvement of emission data for Germany, J. Atmos. Chem. 42, 179–206. Fu, T. M., Jacob, D. J., Palmer, P. I., Chance, K., Wang, Y. X. X., Barletta, B., Blake, D. R., Stanton, J. C., and Pilling, M. J. (2007), Space-based formaldehyde measurements as constraints on volatile organic compound emissions in east and south Asia

and implications for ozone, J. Geophys. Res. Atmos. 112, D06312. Giglio, L., Descloitres, J., Justice, C. O., and Kaufman, Y. J. (2003), An enhanced contextual fire detection algorithm for MODIS, Remote Sens. Environ. 87, 273–282. Gullett, B. K., Touati, A., and Hays, M. D. (2003), PCDD/F, PCB, HxCBz, PAH, and PM emission factors for fireplace and woodstove combustion in the San Francisco Bay region, Environ. Sci. Technol. 37, 1758–1765. Guo, H., Lee, S. C., Ho, K. F., Wang, X. M., and Zou, S. C. (2003a), Particle-associated polycyclic aromatic hydrocarbons in urban air of Hong Kong, Atmos. Environ. 37, 5307–5317. Guo, Z. G., Sheng, L. F., Feng, J. L., and Fang, M. (2003b), Seasonal variation of solvent extractable organic compounds in the aerosols in Qingdao, China, Atmos. Environ. 37, 1825–1834. Hafner, W. D., Carlson, D. L., and Hites, R. A. (2005), Influence of local human population on atmospheric polycyclic aromatic hydrocarbon concentrations, Environ. Sci. Technol. 39, 7374–7379. Huang, J. G. (2006), Analysis for production and management for China coking industry in 2005 and a vista for 2006, China Steel 4, 11–15. Jenkins, B. M., Jones, A. D., Turn, S. Q., and Williams, R. B. (1996), Emission factors for polycyclic aromatic hydrocarbons from biomass burning, Environ. Sci. Technol. 30, 2462–2469. Keith, L. H. and Telliard, W. A. (1979), Priority pollutants: I—a perspective view, Environ. Sci. Technol. 13, 416–423. Kirton, P. J., Ellis, J., and Crisp, P. T. (1991), The analysis of organic matter in coke oven emissions, Fuel 70, 1383–1389. Lee, R. G. M., Coleman, P., Jones, J. L., and Lohmann, R. (2005), Emission factors and importance of PCDD/Fs, PCBs, PCNs, PAHs and PM10 from the domestic burning of coal and wood in the U.K., Environ. Sci. Technol. 39, 1436–1447. Lin, D. (1998), The development and prospective of bioenergy technology in China, Biomass Bioenerg. 15, 181–186. Liousse, C., Penner, J. E., Chuang, C., Walton, J. J., Eddleman, H., and Cachier, H. (1996), A global three-dimensional model study of carbonaceous aerosols, J. Geophys. Res. -Atmos. 101, 19411–19432. Mai, B. X., Fu, J. M., Sheng, G. Y., Kang, Y. H., Lin, Z., Zhang, G., Min, Y. S., and Zeng, E. Y. (2002), Chlorinated and polycyclic aromatic hydrocarbons in riverine and esturine sediments from Pearl River Delta, China, Environ. Pollut. 117, 457–474. Mastral, A. M. and Callen, M. S. (2000), A review on polycyclic aromatic hydrocarbon (PAH) emissions from energy generation, Environ. Sci. Technol. 34, 3051–3057. Narvaez, R. F., Hoepner, L., Chillrud, S. N., Yan, B., Garfinkel, R., Whyatt, R., Camann, D., Perera, F. P., Kinney, P. L., and Miller, R. L. (2008), Spatial and temporal trends of polycyclic aromatic hydrocarbons and other traffic-related airborne pollutants in New York City, Environ. Sci. Technol. 42, 7330–7335. Nathwani, J. S., Siddall, E., and Lind, N. C. (1992), Energy for 300 Years: Benefits and Risks, Institute for Risk Research, University of Waterloo, Waterloo, Canada. NBSC (National Bureau of Statistics of China) (2004a), China Energy Statistical Yearbook 2003, China Statistics Press, Beijing.

REFERENCES

NBSC (2004b), China Statistical Yearbook 2003, China Statistics Press, Beijing. NBSC (2005), China Energy Statistical Yearbook 2004, China Statistics Press, Beijing. Nielsen, M., Illerup, J. B., Kristensen, P. G., Jensen, J., Jacobsen, H. H., and Johansen, L. P. (2001), Emission Factors for CHP Plants< 25Mwe, National Environmental Research Institute, Denmark. Oanh, N. T. K., Nghiem, L., and Phyu, Y. L. (2002), Emission of polycyclic aromatic hydrocarbons, toxicity, and mutagenicity from domestic cooking using sawdust briquettes, wood, and kerosene, Environ. Sci. Technol. 36, 833–839. Oanh, N. T. K., Reutergardh, L. B., and Dung, N. T. (1999), Emission of polycyclic aromatic hydrocarbons and particulate matter from domestic combustion of selected fuels, Environ. Sci. Technol. 33, 2703–2709. Ohkouchi, N., Kawamura, K., and Kawahata, H. (1999), Distributions of three- to seven-ring polynuclear aromatic hydrocarbons on the deep sea floor in the central pacific, Environ. Sci. Technol. 33, 3086–3090. Pacyna, J. M., Breivik, K., Munch, J., and Fudala, J. (2003), European atmospheric emissions of selected persistent organic pollutants, 1970–1995, Atmos. Environ. 37, S119–S131. Qin, F., Liu, H. Y., Jin, L., Liu, W., Yin, J., and Wei, X. D. (2007), The investigation of energy consumption in the village of Jilin Province, J. Jilin Architect. Civil Eng. Inst. 24, 37–40. Sinomaps Press (2005), Atlas of Physical Geography of China, Sinomaps Press, Beijing. State Forestry Administration of China (2005), China Forestry Statistical Yearbook, China Forestry Press, Beijing. Streets, D. G., Bond, T. C., Carmichael, G. R., Fernandes, S. D., Fu, Q., He, D., Klimont, Z., Nelson, S. M., Tsai, N. Y., Wang, M. Q., Woo, J. H., and Yarber, K. F. (2003), An inventory of gaseous and primary aerosol emissions in Asia in the year 2000, J. Geophys. Res. Atmos. 108, 8809. Tan, J. H., Bi, X. H., Duan, J. C., Rahn, K. A., Sheng, G. Y., and Fu, J. M. (2006), Seasonal variation of particulate polycyclic aromatic hydrocarbons associated with PM10 in Guangzhou, China, Atmos. Res. 80, 250–262. Tang, N., Hattori, T., Taga, R., Igarashi, K., Yang, X. Y., Tamura, K., Kakimoto, H., Mishukov, V. F., Toriba, A., Kizu, R., and Hayakawa, K. (2005), Polycyclic aromatic hydrocarbons and nitropolycyclic aromatic hydrocarbons in urban air particulates and their relationship to emission sources in the Pan-Japan Sea countries, Atmos. Environ. 39, 5817–5826. Tsai, P., Hoenicke, R., and Yee, D. (2002), Atmospheric concentrations and fluxes of organic compounds in the northern San Francisco Estuary, Environ. Sci. Technol. 36, 4741–4747. Tsibulsky, V., Sokolovsky, V., and Dutchak, S. (2001), MSC-E Contribution to the HM and POP Emission Inventories, Technical Note 7/2001, Meteorological Synthesizing Centre-East, Moscow. UNECE (1998), Protocol to 1979 Convention on Long-Range Transboundary Air Pollution on Persistent Organic Pollutants. United Nations Economic Commission for Europe.

281

USEPA (1998), 1990 Emissions Inventory of Section 112(c)(6) Pollutants: Polycyclic Organic Matter (POM), TCDD, TCDF, PCBs, Hexachlorobenzene, Mercury, and Alkylated Lead: Final Report, Research Triangle Park, NC. USEPA (1996), Compilation of Air Pollutant Emission Factors, AP-42, 5th ed., Vol. I, Stationary Point and Area Sources. USEPA, Washington, DC. Wan, X. L., Chen, J. W., Tian, F. L., Sun, W. J., Yang, F. L., and Saiki, K. (2006), Source apportionment of PAHs in atmospheric particulates of Dalian: Factor analysis with nonnegative constraints and emission inventory analysis, Atmos. Environ. 40, 6666–6675. Wang, G. H., Kawamura, K., Zhao, X., Li, Q. G., Dai, Z. X., and Niu, H. Y. (2007), Identification, abundance and seasonal variation of anthropogenic organic aerosols from a mega-city in China, Atmos. Environ. 41, 407–416. Witherington, P., Dhammapala, R., Claiborn, C., and Corkill, J. (2002), Particulate and Gaseous Pollutants from Wheat Stubble in Eastern Washington, Center for Multiphase Environmental, Washington State University, Pullman WA. Xu, S. S., Liu, W. X., and Tao, S. (2006), Emission of polycyclic aromatic hydrocarbons in China, Environ. Sci. Technol. 40, 702–708. Yang, H. H., Lee, W. J., Chen, S. J., and Lai, S. O. (1998), PAH emission from various industrial stacks, J. Hazard. Mater. 60, 159–174. Yuan, H. S., Tao, S., Li, B. G., Lang, C., Cao, J., and Coveney, R. M. (2008), Emission and outflow of polycyclic aromatic hydrocarbons from wildfires in China, Atmos. Environ. 42, 6828–6835. Zakaria, M. P., Takada, H., Tsutsumi, S., Ohno, K., Yamada, J., Kouno, E., and Kumata, H. (2002), Distribution of polycyclic aromatic hydrocarbons (PAHs) in rivers and estuaries in Malaysia: A widespread input of petrogenic PAHs, Environ. Sci. Technol. 36, 1907–1918. Zeng, X. Y., Ma, Y. T., and Ma, L. R. (2007), Utilization of straw in biomass energy in China, Renew. Sustain. Energy. Rev. 11, 976–987. Zhang, Y., Dou, H., Chang, B., Wei, Z., Qiu, W., Liu, S., Liu, W., and Tao, S. (2008), Emission of polycyclic aromatic hydrocarbons from indoor straw burning and emission inventory updating in China, Ann. NY Acad. Sci. 1140, 218–227. Zhang, Y. and Tao, S. (2008), Seasonal variation of polycyclic aromatic hydrocarbons (PAHs) emissions in China, Environ. Pollut. 156, 657–663. Zhang, Y. X., Tao, S., Cao, J., and Coveney, R. M. (2007), Emission of polycyclic aromatic hydrocarbons in China by county, Environ. Sci. Technol. 41, 683–687. Zhong, Y. F. (1999), Control small-scale coke production, promoting the healthy development of China coking industry, Fuel Chem. Process. 9, 216–208. Zhou, J. B., Wang, T. G., Huang, Y. B., Mao, T., and Zhong, N. N. (2005), Seasonal variation and spatial distribution of polycyclic aromatic hydrocarbons in atmospheric PM10 of Beijing, People’s Republic of China, Bull. Environ. Contam. Toxicol. 74, 660–666.

PART III ANALYTICAL TECHNIQUES

12 PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES FOR MEASUREMENTS OF ORGANIC POLLUTANTS IN ENVIRONMENTAL MATRICES EDDY YONGPING ZENG, ZHAOHUI WANG,

AND

O. SAMUEL SOJINU

12.1. Introduction 12.2. Active Sample Collection Techniques 12.2.1. Introduction 12.2.2. Large-Volume Air Sampling 12.2.3. Soil/Sediment Grab and Core Sampling 12.2.4. Water Sampling Strategy 12.3. Passive Sampling 12.3.1. Introduction 12.3.2. Theoretical Background 12.3.3. Applications 12.3.4. Representative Passive Sampling Techniques: Semipermeable Membrane Device and Solid-Phase Microextraction 12.4. Sample Extraction Protocols 12.4.1. Liquid–Liquid Extraction 12.4.2. Soxhlet Extraction 12.4.3. Accelerated Solvent Extraction 12.4.4. Microwave-Assisted Solvent Extraction 12.4.5. Supercritical Fluid Extraction 12.4.6. Solid-Phase Extraction 12.4.7. Extract Concentration and Solvent Exchange 12.4.8. Cleanup and Fractionation Procedures 12.4.9. Quality Assurance and Quality Control 12.5. Instrumentation 12.5.1. Chromatography Techniques 12.5.2. Mass Spectrometry Techniques 12.6. Conclusions

12.1. INTRODUCTION Sample processing, broadly including collection, extraction, and instrumental analysis herein, is the most critical component in the determination and assessment of the occurrence, fate, and effects of anthropogenic organic contaminants in the environment. Any precise description of the environmental behavior of these contaminants is impossible without high-quality field and laboratory data. Failure to acquire high-quality data implies failure of the entire process. Environmental research involves rigorous and stepwise processes, and each step must be dutifully approached to achieve quality and reproducible results. Within this context, the sample processing techniques that will be the focal point of this chapter are the key part of any environmental research program and should be paid the utmost attention. Of course, this is not to undermine the importance of other analytical stages. The first component in sample processing is sample collection, which is obviously the basis of any further analytical effort. Effective and adequate sampling is the most important element in environmental analysis, as without it all the data subsequently generated are deemed of little value. “Bad” samples can result from the wrong choice of sampling location, inappropriate sampling protocol, and preservation errors, among others. In any case, when such samples are analyzed, results are generated but are not reproducible or representative, and as such are ineligible for scientific assertion. Likewise, sample extraction, the

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

285

286

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

next component in sample processing, is an integral part of the analytical effort. Selection of an appropriate extraction protocol depends on the sample type and quantity, possible matrix interferences, and the need to meet any deadlines that may exist, and other factors. In many cases, predetermined extraction procedures need to be adjusted to allow for the occurrence of unexpected issues before being finalized for a given set of samples. The last component in sample processing, instrumental analysis, directly dictates the quality of the data generated. As a result, selection of proper instrument and detector (resulting in required detection sensitivity), as well as accurate calibration of the quantitation method, is the key factor in ensuring the highest quality possible of the analytical results. For example, gas chromatography (GC) is normally applicable to measurements of volatile and moderately volatile organic compounds, while liquid chromatography (LC) is often employed for analysis of water-soluble and thermally liable chemicals. Finally, it is worth noting that none of these components can be satisfactorily executed without an adequate and rigorous quality assurance/quality control (QA/QC) program in place. This chapter attempts to cover many aspects of the established and commonly used sampling techniques, sample preparation prior to extraction, extraction protocols for different sample matrices and analytical instruments. The potential target analytes relevant to this chapter, as indicated in the title, are organic pollutants; but more specifically, semivolatile organic compounds (SVOCs) will be the main targets under consideration, including largely persistent organic pollutants (POPs) mandated by the Stockholm Convention on POPs (http://chm.pops.int/), such as polychlorinated biphenyls (PCBs), organochlorine pesticides [OCPs; especially dichlorodiphenyltrichloroethane (DDT)], and polybrominated diphenyl ethers, as well as other classes of SVOCs that are not explicitly targeted by the Stockholm Convention but have been widely investigated, such as polycyclic aromatic hydrocarbons (PAHs), nalkanes, and linear alkylbenzenes. Volatile organic compounds (VOCs) will also be dealt with, but mainly in places involving sampling techniques. It is the hope of the authors that the contents covered in this chapter will allow the readers to decipher many problems encountered during field sampling and sample extraction while offering holistic assistance in the selection of an appropriate analytical instrument. It should be noted that given the hugely broad scope of subject matters addressed by this chapter, it is inevitable that the level of detail has to be somewhat compromised. In addition, slightly more space will be devoted to sample extraction procedures, as these protocols are relatively more comparable among various analytical laboratories than are procedures for sampling and instrumental analysis; they are more likely to be standardized. Hence, a practical,

step-by-step working protocol is provided for each sample extraction or cleanup method. Also, this chapter is intentionally confined to dealing with so-called legacy organic pollutants only, largely because analytical methods for “emerging” chemicals (e.g., perfluorinated compounds and pharmaceuticals) are still evolving. Nevertheless, some of the techniques discussed in this chapter can still be applied to measuring the environmental occurrence of emerging chemicals on some modifications.

12.2. ACTIVE SAMPLE COLLECTION TECHNIQUES 12.2.1. Introduction Needless to say, sample collection is the first step toward a larger plan to investigate organic pollutants. Success or failure of the entire analytical process largely hinges on whether representative and sufficient samples can be collected with confidence. In carrying out sample collection, the following questions must be answered correctly: .

.

.

When to Sample. When to sample is driven by many reasons, including perceived environmental problems, seasonal variability, climatic patterns, availability of resources, and research hypotheses and objectives. Among these, the study objective determines, to a large extent, the time of sampling. For instance, sampling should be conducted to investigate the cause if an environmental issue becomes urgent, while for a seasonal effect, samples should be collected during all the seasons. In any case, sample collection should be conducted as needed, not at convenience. Where to Sample. Experience has shown that not all areas in a given location are appropriate for sampling— or better stated, not all areas can provide representative samples of the location. Therefore it is crucial to identify and map out the sampling sites, especially for a heterogeneous location impacted by different anthropogenic activities. It is also essential to have at least a control site so that comparative sample analyses can be carried out. Several samples may also be taken from a location and homogenously mixed together to create a composite sample. How to Sample. How a sample is collected depends largely on the research objective and availability of sampling techniques. The details on a variety of sampling strategies that have been commonly used by environmental scientists are presented in the following sections. One important component with field sampling is to always carry field blanks during the course of sampling to monitor whether there is any contamination to the field samples from external sources.

ACTIVE SAMPLE COLLECTION TECHNIQUES

12.2.2. Large-Volume Air Sampling The large-volume sampling strategy can be used to collect both SVOCs and VOCs in air. Because VOCs are normally present in the vapor phase at room temperature [with vapor pressure > 0.1 mmHg (0.0133 kPa) at 298.95 K (Kelly et al. 1994], they are often collected with evacuated canisters and less frequently sampled by solid sorbents. On the other hand, SVOCs, by default, are less volatile, but may still be present in the vapor phase in affiliation with aerosols, as either dusts or liquid droplets, in the atmosphere. Hence, atmospheric SVOCs in the particulate and vapor phases are normally trapped with different types of adsorbents. Throughout the sampling period, samples must be preserved within the trapping materials. 12.2.2.1. Solid Sorbent Methods. The application of solid sorbent sampling techniques includes the capture of organic constituents using activated charcoal (for VOCs) or glass fiber filters (for SVOCs in the particulate phase) and polyurethane foam (PUF; for SVOCs in the vapor phase), desorption (for VOCs) or extraction (for SVOCs), and analysis. This approach is capable of processing a large volume of air (therefore collecting sufficient amounts of trace organics) within a relatively short time duration, and particularly useful for sensing a snapshot of the instantaneous levels of target analytes in the atmosphere. A typical solid sorbent air sampler (Fig. 12.1) for SVOCs consists of a high-volume air sampler that houses a glass fiber filter (GFF; e.g., 20.3
25.4 cm2, 0.6 mm nominal pore size) to collect particles over the pore size of the filter and a PUF plug (e.g., 6.5 cm in diameter and 8.0 cm in thickness with a density of 0.030 g/cm3) to trap gaseous components. In a typical field operational scheme, particulates are collected by drawing air through a GFF at a preset flow rate (e.g., 1.0 m3/min), and gaseous constituents are trapped with a PUF plug serially connected to the GFF holder. Depending on the sampling duration and concentrations of particles and gaseous constituents, more than one GFF and PUF plug may be needed for a sample. On sampling, loaded GFFs can be wrapped with prebaked aluminum foils or other clean media free of organics. Meanwhile, PUF plugs are stored in glass jars previously rinsed with a solvent (e.g., acetone) and capped with aluminum foil–lined lids (or anything free of organics). The GFFs and PUFs are transported to the laboratory and stored at 20 C until extraction. Prior to field sampling, GFFs should be baked at 450 C for 4–6 h, whrereas PUF plugs should be Soxhletextracted for 48 h with methanol and for another 48 h with dichloromethane. 12.2.2.2. Evacuated Air Container Method. The use of evacuated containers is generally painless, because samples

287

Acceleration region

Air inlet

Greased plate filte Protective screen

polyurethane foam Flow controller Suction pump with controller

manometer

Figure 12.1. A schematic depicting a typical large-volume air sampler.

can be stored in evacuated containers for up to a year without significant losses (Harper 2000). Whereas solid sorbent methods are limited in capturing chemicals, evacuated containers retain all organics, irrespective of chemical characteristics. Typical evacuated containers (Fig. 12.2) include canister (e.g., SUMMAÒ canister), bag (e.g., integrated bag sampler), a can (e.g., MSA Evacuated Can), and a test tube (e.g., Texas Research Institute IAQ Sampler) (Harper 2000). A typical sampling procedure is as follows. Evacuated canisters are thermally treated under vacuum. A certain amount of air is introduced into the canister by opening a valve for a predesignated period of time. The canister is usually made of stainless steel and will not easily collapse during transportation. Generally, one canister can be used both for preliminary sample screening and subsequent collection. After sampling, the valve opening should be secured, the pertinent information (e.g., sample number and location) is recorded, and the sample is shipped to the laboratory for analysis. Evacuated cans can be used only for screening. Ambient air test tubes are “test tubes” with a PTFE cover and a screw-on cap. To use an ambient air test tube, unscrew the cap and leave it open in the sample area for 15–30 min. Waving the tube through the air can reduce required exposure time. The test tube is capped, and is then ready for laboratory analysis.

288

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES .

Valv

. . .

Gauge

. . .

Canister

In order to collect samples of surface soil, the following steps should be followed: . .

Figure 12.2. A schematic depicting a typical canister for air sampling.

.

12.2.3. Soil/Sediment Grab and Core Sampling

.

12.2.3.1. Soil Sampling. In collecting soil samples, it is necessary to delineate the sampling area so that representative samples can be collected. The delineation process is done using different parameters. The first sampling point can be chosen arbitrarily with subsequent points based on a random number generator, and the grid rows start building from this point. Sampling grids can be square, rectangular, or hexagonal (triangular) with the grab samples collected from each grid nodule. However, the best way to lay out an accurate sampling grid is to employ the service of a land surveyor who will identify the sampling points and record the various coordinates using a Global Positioning System. Samples of surface soil are usually collected with disposable handheld gadgets and stored in vials with sufficiently thick PTFE/silicon septa that are commonly used for storage of water samples, to prevent VOC losses over the sample holding time [U.S. Environmental Protection Agency (USEPA) 2005]. Samples of subsurface soil, on the other hand, are retrieved with hand-powered augers, with directpush techniques, or by placing them in boreholes. Soil is brought to the surface from discrete known depths in liners placed inside the augers, direct-push rods, or split-spoon samplers used in borehole drilling with hollow stem augers. The materials listed below are essential for sampling soil for determination of SVOCs, and must be certified acceptable prior to sampling:

. .

. . .

.

.

Containers for surface soil samples: precleaned (e.g., kilnfired at 450oC for 4 h) jars and lids Containers for subsurface soil samples: stainless steel or brass core barrel liners 6 in. in length with PFTElined caps

Identify the sampling points in the field with respect to the delineation. Wear disposable gloves, safety glasses, and other appropriate PPE while sampling. Use a decontaminated shovel or trowel to remove the turf at the proposed sampling location, if necessary. Remove loose debris and exposed soil from the top 2.5–5 cm. Collect a sample for VOC analysis first. Collect a soil sample into a glass jar using a disposable scoop or a decontaminated stainless-steel spatula. Remove stones, twigs, grass, and other extraneous material. Fill the jar to the brim with firmly packed soil. Tightly replace the cap of the jar, label it, and place it on ice. Collect a field duplicate sample as close as possible to the location of the primary sample. Document the sampling locations in the field logbook, and decontaminate non-disposable equipment between sampling points. Ship the samples to the laboratory as soon as possible and by the fastest available delivery service.

These steps are not all-inclusive; the requirements of the project or analysis may necessitate additional inputs into the sampling protocol. To sample subsurface soil with a split-spoon sampler, the following steps should be followed: .

. .

Sample labels PTFE sheets (if liners are used) Stainless-steel spatulas, disposable plastic scoops, or wooden tongue depressors Steel trowels or shovels Disposable gloves, paper towels, and pens Safety glasses and other appropriate personal protective equipment (PPE) Waterproofed field logbook

. .

Wear disposable gloves, safety glasses, and other appropriate PPE while decontaminating sampling materials and conducting sampling. Decontaminate the liners and split spoon. Identify the sampling points. Drill to the desired depth and retrieve the auger. Attach the split spoon sampler with three liners inside to the sampling rod and insert into the boring. Drive the splitspoon sampler further down the borehole according to

ACTIVE SAMPLE COLLECTION TECHNIQUES

. .

.

.

. . .

ASTM Method D1586-08 (American Society for Testing and Materials). Retrieve the split-spoon sampler from the borehole and open it. If sampling for VOCs is required, use an airtight coring device and quickly collect samples from the ends of the middle liner. Cap the lower liner with PTFE sheets and plastic caps, label it, and place it on ice. If sampling for VOCs is not required, select a liner with the best recovery efficacy or from a specific depth, replace the cap and label, and place it on ice. Use only the middle or lower liner. Discard the soil in the upper liner containing debris. [Note: For samples that are to be analyzed for organics, the spatula and container must not be plastic (the container must be glass bottles provided by the laboratory). For samples that are to be analyzed for metals, the spatula must not be metallic.] Document the sampling locations in the field logbook. Decontaminate the split-spoon sampler and other non disposable equipment between sampling points. Ship the samples to the laboratory as soon as possible and by the fastest available delivery service.

For sampling of VOCs, a different approach may be used, as losses of VOCs from soil can occur during sampling, transport to the laboratory, storage, and actual analysis. A veritable method for sampling VOCs is the updated USEPA Method 5035A, referred to as “closed-system purge-and-trap and extraction for volatile organics in soil and waste samples” (USEPA 2005). This method describes the procedures for closed-system soil collection into special airtight coring devices or autosampler vials that are never opened after the samples have been collected. This method also proposes the use of methanol and sodium bisulfate for soil sample preservation in the field, and addresses the soil sample holding time requirements. A combination of the airtight device sampling technique, good preservation, and holding time restrictions have been proven to produce accurate data on VOC concentrations in soil samples (Hewitt 1995a,b; 1999). Sampling soils for VOCs can be done using the following stepwise process (Popek 2003): 1. Wear disposable gloves, safety glasses, and other appropriate PPE while sampling. 2. Identify the predetermined sampling points. 3. With a scoop remove the upper 5–8 cm of exposed surface soil. 4. Holding the coring device with the T-handle up and the cartridge down, insert the sampling device into the freshly exposed soil. Look down the viewing hole to

5.

6.

7.

8.

289

make sure the plunger O-ring is visible, indicating that the cartridge is full. If the O-ring is not visible, apply more pressure to fill up the cartridge. Withdraw the sampling device from soil, and with a spatula, remove excess soil from the bottom of the barrel to ensure that the soil surface is flush with the walls of the barrel. Wipe the barrel with a paper towel to remove soil from the external surface. Cap the cartridge while it is still on the T-handle. Push and twist the cap until the grooves are seated over the ridge of the coring body. Disconnect the capped sampler by pushing the locking lever down on T-handle, and twist and pull the sampler from T-handle. Lock the plunger by rotating the plunger rod until the wings rest against the tabs. Fill out the label with the sample information on the original sampler bag and place the cartridge into the bag. Collect three samples from each sampling point as close as possible to each other. Likewise, collect a field duplicate sample as unhomogenized split sample at a site as close as possible to the location of the primary sample. Clearly identify these samples on the label, and store the sample in a cooler with ice. Collect 10–20 g of soil into a clean, labeled VOC vial or a jar for dry weight and carbonate presence determination. This step is, however, optional. Document the sampling points in the field logbook.

12.2.3.2. Sediment Sampling. Sediment samples can be collected from oceans, lakes, rivers, streams, canals, lagoons, ponds, and other water bodies, and they are good indicators of current and historical contamination from anthropogenic activities. Generally, there are two types of samplers used for collecting bottom sediments: (1) grab samplers (Fig. 12.3) for collecting surface sediments, which provides samples for characterizing the horizontal distribution of the analytes, and (2) core samplers (Fig. 12.4) for collecting a depth profile of sediments, thereby providing samples for profiling the vertical distribution of the analytes in the area. Grab samplers, because of their ease of use and large quantity of sample obtained, are ideal for assessing recent inputs of pollutants, while core samplers are better suited for assessing long-term (temporal) trends. The sampler type used will depend on the objective of a specific study and will be dictated by the project design and cost. There are different types of grab and core samplers; however, it is necessary to know the water depth at each sampling site prior to sampling regardless of what sampler is chosen. If water depth information is unavailable, it is recommended that it first be taken. Measurement equipment can range from a weighted rope to an electronic depth sounder. The aim is to ensure adequate cable (rope) length

290

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

Top thieves Wire

Trigger mechanism Hoist wires

Valve Upper sheave block Gear segments

Closing steel pole Snap

Steady arm

Core barrel

Gear segments

Lower sheave block

Figure 12.3. A schematic depicting a typical sediment grab sampler.

for operation of the equipment and to control the speed of entry of the sampler into the sediment. The speed of deployment of the sampler can be vital to good operation and sample recovery. A deployment that is too rapid generates and increases the shockwave advancing in front of the equipment. This shockwave can displace the soft unconsolidated surface sediments. Rapid deployment may also cause equipment malfunction, such as activating the trigger mechanism before the device reaches the sediment. In the case of core samplers, if the deployment is too slow, an insufficient quantity of sediment is obtained (Popek 2003). 12.2.3.2.1. Sampling from a Boat. The collection of deepwater sediment samples requires that at least one member of the sampling team be very familiar with boat operation and safety. If the sampling trip involves the use of a boat, then the weather forecast or marine conditions should be obtained prior to departure. The sampling trip should be postponed if weather conditions are too poor. All members of the crew should wear life jackets and other PPE to ensure maximum safety. The boat to be used should be inspected and certified prior to use, and large enough to accommodate the sampling crew and equipment without invoking any risk thereof.

Core tube

Figure 12.4. A schematic depicting a typical sediment gravity core sampler.

The following procedure is typically used for sampling from a boat with a grab sampler (Fig. 12.3): 1. Set the grab sampling device with the jaw cocked open. Great care should be taken while handling the device as it is set; accidental closure can cause serious injuries. 2. Ensure that the rope is securely fastened to the sampler and that the other end is tied to the boat. 3. Lower the sampler until it is resting on the sediment (its own weight is adequate to penetrate soft sediments). At this point the slackening of the line activates the mechanism to close the jaws of the Ponar and Petersen grabs. 4. For the Ekman grab (a type of grab sampler), send the messenger down to “trip” the release mechanism. 5. Retrieve the sampler slowly to minimize the effect of turbulence that may result in loss or disturbance of surface sediments.

ACTIVE SAMPLE COLLECTION TECHNIQUES

6. Place a container (i.e., a shallow pan) beneath the sampler just as it breaks the water surface. (Note: If the jaws were not closed completely, the sample must be discarded.) Discard the sample into a bucket if the second collection attempt is made from the same general area. Dump the unwanted sample only after a sample has been successfully collected. For samples that are to be analyzed for organics, the spatula and container must not be plastic (the container must be a glass bottle provided by the laboratory). For samples that are to be analyzed for metals, the spatula must not be metallic. 7. Immediately record (in the field logbook) the appearance of the sediment (i.e., texture, color, odor, presence of biota and detritus, and the depth of sediment sampled). 8. With a clean spatula, carefully stir the sediment to homogenize, then scoop an aliquot into a prelabeled sediment sample bottle for dry weight and carbonater determination, this is also optional. 9. Place the samples in a cooler with ice packs as soon as they are transferred to the bottles. The following procedure is typically used for sampling from a boat with a core sampler: 1. Open the valve and set the trigger mechanism. Ensure that the rope is securely fastened to the corer and attach the other end of the rope to the boat. 2. Lower the corer to just above an area of undisturbed sediments and then allow it to penetrate the sediments with its own weight. Specialized corers and different types of sediment may require different techniques, but what seems to be important is to avoid the disturbance caused by impact. 3. Send the messenger down to release the trigger mechanism. 4. Carefully retrieve the sampler and place a stopper into the bottom opening before removing from the water to prevent loss of the sample. 5. Remove the liner from the corer and stopper the upper end. Store erect. Repeat this procedure to obtain replicate cores, each at least 20–30 cm in length. 6. Once on shore, carefully siphon off most of the water overlying the sediments in the core tube (leave a small amount at the sediment–water interface). Do not disturb the sediment–water interface. 7. Carefully measure the total length of the core and precise points (to nearest mm) of any layers of sediment that appear to be different. Note any changes in stratigraphy, such as color and texture. 8. Select a rubber stopper of a size sufficient to fit inside the liner tube tightly to form a watertight seal and

291

mounted in an extruder. Then gently and slowly force the core upward to the top of the tube. Most extruders allow the increment of sediment slices to be adjusted. 9. As the sediment core is extruded, carefully cut slices (from 2 mm to 1 cm thick) with clean spatulas and place into labeled sample bottles. An extruder greatly assists this operation, but good samples can be obtained by mounting the core tube in a vice and manually extruding the sediment if done carefully. It should be noted that the edges of a core should be discarded, as smearing with the inner surface of the core liner can take place, confounding the results. For samples that are to be analyzed for organics, the spatula and container must not be plastic (the container must be glass bottles provided by the laboratory). For samples that are to be analyzed for metals, the spatula must not be metallic. 10. Place the samples in a cooler with ice packs as soon as they are transferred to the labeled bottles. Many lake and oceanic sediments are anoxic, and there is the likelihood of chemical transformations to take place if sediment samples are exposed to ambient air. If samples are to be retained at as low an oxygen level as possible, they will need to be packed inside multiple airtight containers and frozen to minimize the chemical and microbial transformations. If samples are frozen, allow sufficient headspace for expansion. Otherwise, the container will split or break when the sample freezes. It is necessary to state here that sampling in winter presents extra elements of risk. It is advisable to always proceed with absolute caution over ice and to not jeopardize your safety. Check the ice for thickness with a rod or ice chisel every few steps as you proceed (ice should be a minimum of 10 cm thick, and at least 30 cm thick if a vehicle is used). Always have someone follow you, and carry a length of rope (with a harness tied around your waist) to use as a lifeline. If the ice is unsafe, do not take a sample. Never take unnecessary risks, with absolute safety considerations in mind. Clear loose ice and snow from the sampling location, and drill through the ice with a hand or motorized auger. Keep the area around the hole clear of potential contamination (dirt, fuel, oil, etc.). At least one member of the sampling team should be familiar with the operation and safety of both motorized and handoperated augers. Follow sample collection procedures outlined above for either grab or core samplers. Sediment sampling in deep sections of rivers and streams rarely involves the use of core samplers, as these devices require that flow be minimal (very few rivers worldwide have sufficiently low flow) (Resources Information Standards Committee 1997). Alternatively, core samples can be collected in shallow, flowing waters by physically pushing the corer into the sediment by hand. It is useful to have some understanding of the currents at the sampling site.

292

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

Strong near-bottom currents can lead to poor equipment deployment, deflect a grab sampler, or require a long cable or wire to be deployed. Care should be taken to ensure that the weight of the sampler is adequate for working in the particular current conditions and that the sampler collects sediment at or very near the desired sampling site. In locations where there are bridges, these structures provide a good channel for collecting sediment samples without the need for a boat, the same protocols can be utilized as enumerated above.

.

.

. .

12.2.4. Water Sampling Strategy There are different types of water, ranging from underground water, surface water (e.g., lakes, rivers, oceans), and atmospheric water (fogs, rain, mists, etc.). The approach adopted in sampling each of these water bodies may vary with water type. It is apparent that the main challenge for any water sampling plan is to collect representative samples. An effective sampling protocol would minimize any variability at the sampling point. Use of appropriate sampling kits will also further ensure the representativeness of the sample collected. Furthermore, when a sampling design is developed, some vital factors should be taken into consideration, such as the seasonal variation in the sampling location and the known contaminant stratification in the water medium. The frequency and quantity of samples (including QC samples) needed can be estimated on the project objectives, the funds available for the project, and the perceived level of pollution. Collection of composite water samples can be based on the spatial or seasonal distribution of the water medium under investigation. Rivers, streams, lakes, lagoons, ponds, and water-containing diteches are surface waters; as such, the same sampling procedure can be employed for these water bodies. The following principles are common to sampling of all types of surface water (Popek 2003). .

.

.

.

Start sampling in the areas suspected to have the least contamination and finish in the areas suspected of being the most contaminated; this can be deduced from proximities to known or suspected pollution sources. Start sampling a location farthest downstream and proceed upstream in order to avoid contaminating the downstream areas while disturbing the upstream water. If a point source of discharge is present, start sampling at the discharge point, and proceed with sampling the receiving water body on a radial sampling grid that starts at the discharge point. Field parameter measurements (pH, temperature, conductivity, etc.) may be a part of the sampling procedure as indicators of surface water quality (not stabilization indicators as in groundwater sampling).

. .

.

.

Whenever possible, take field measurements directly in the water body by lowering the field instrument probes into the water at the point where a sample will be taken. Collect samples directly into the sample containers with a pumping device or by lowering the containers under the water surface (sample containers in this case should not contain preservatives). Preserve samples as appropriate after they have been collected. When sampling directly with a container, approach the sampling point from downstream, placing the container in the water with its neck facing upstream and away from yourself. When sampling from a boat, collect samples upstream from the boat. If a water body has no current, entering it would potentially introduce extraneous contamination. In this case, sample from the shore using a long-handled dipper whenever possible. When wading in the water, do not needlessly disturb the bottom sediment. Wait until the sediment settles prior to sampling. The supplies and equipment for surface water sampling vary depending on the type of water body to be sampled and on the intended use of the data. If sampling personnel are planning on entering the water, personal protection gear, such as rubber boots, should be worn.

In addition to the principles mentioned above, other factors may also need to be considered depending on the geographic settings of the sampling sites. As an example to demonstrate the complexity that could result from specific sampling sites, the sampling design to collect riverine runoff samples from the Pearl River Delta, China (Ni et al. 2008), where surface water flows into the coastal ocean via eight major runoff outlets, is briefly described here. Because the main objective of that sampling program was to determine the riverine fluxes of organic contaminants to the coastal ocean, it became obvious that tidal influences had to be minimized, and cross-sectionally well-mixed water samples were desirable. As a result, sampling was conducted approximately one hour before the intraday lower tides. Furthermore, horizontal sampling points were three (<1500 m) or five (>1500 m) depending on the river width and were evenly distributed along the river cross section. Vertically, three sampling points were placed at the upper (1 m from the air–water surface), middle (at the middle of the water depth), and bottom (1 m from the riverbed) parts of the water column. The individual samples from the 9 or 15 sampling points were mixed to generate a representative composite sample, which was so designed to ensure cost-effectiveness.

PASSIVE SAMPLING

293

Measuring water quality parameters in situ is essential for effective evaluation of a water body, since some physicochemical parameters (oxidation–reduction potential, pH, temperature, conductivity, dissolved oxygen, turbidity) vary with exposure to oxygen in air and daylight and changes in temperature and barometric pressure, which will cause irreversible chemical changes in a water sample, which, in turn, will affect some of the water quality parameters. This thus necessitates the need to measure these parameters right in the natural state of the water. Quite a number of portable gadgets capable of taking these measurements are available.

mass balance considerations balancing all analyte amounts in all interacting phases before and after sampling, whereas kinetically controlled passive sampling can be characterized by Fick’s first law of diffusion, a simplified description of the sorption process. In general, equilibrium passive sampling retains larger sorption capacity than does kinetically controlled passive sampling, and thereby better sensitivity. Since 1927, when passive sampling was first used for semiquantitative determination of CO (Gordon and Lowe 1927), numerous techniques have been developed and popularly used in environmental field sampling, as evidenced by the gigantic number of published articles available in the literature.

12.3. PASSIVE SAMPLING

12.3.2. Theoretical Background

12.3.1. Introduction For a water body of reasonable size, a large number of samples (and large sample size as well) are often required to provide adequate information for assessment of the water quality. In this case, active sampling and subsequent laboratory analysis can be very costly. Even so, active sampling may only acquire snapshots of the analyte concentrations in the sampling area, which are considerably variable with time and therefore may not be representative of the actual contamination levels. In contrast, most passive sampling techniques vastly simplify the procedures of sampling and sample preparation, eliminate power requirements and other inherent logistics involved in active sampling, and thereby significantly reduce operational cost. The use of a pump and electricity with active sampling can be very inconvenient in field cruises. In some cases, passive sampling techniques, such as solid-phase microextraction (SPME) (Arthur and Pawliszyn 1990), are able to combine sample collection, extraction, and concentration into one single step, greatly minimizing cross contamination and reducing labor cost. More importantly, passive sampling, if designed properly, can be used to obtain the time-weighted average (TWA) concentration of a particular analyte, which is extremely difficult to accomplish with an active sampling approach. On the other hand, passive sampling typically requires prolonged field deployment time, thereby requiring extra resources to ensure safe retrieval of sampling devices, but this does not undermine its advantages Depending on the physicochemical characteristics of the adsorbent used, passive sampling methods can be generally classified as absorptive or adsorptive. The adsorptive methods take advantage of the physical or chemical retention by sorbent surfaces, and key parameters involve surface binding on surface area. The absorptive method, on the other hand, is clearly dependent on the partitioning of analytes into the interceding material. Passive sampling can be conducted either at equilibrium or within the kinetically controlled range. Equilibrium passive sampling can be quantified with

For kinetically controlled sampling, two distinct processes can be identified: diffusion and permeation. Diffusion results from the difference in concentrations of the sorbate in the opposite sides of a cross section; sorbates move from the side with higher concentration to the side with lower concentration. On the other hand, permeation means the movement of sorbates across a semipermeable membrane driven again by the different concentrations on two sides of the membrane. Both processes are governed by Fick’s first law of diffusion as shown here dm A ¼ D ðCa Csorbent Þ dt Z

ð12:1Þ

where dm/dt is the change of the mass (m) of the sorbate sorbed on the sorbent (mg) with time (t), D is the diffusion coefficient of the analyte (m2/min), A is the diffusion surface area (m2), Z is the length of the diffusion zone (m), Ca is the analyte concentration in the sample (mg/m3), and Csorb is the analyte concentration in the sorbent phase (mg/m3). As shown in Equation (12.1), passive sampling is driven by the difference in chemical potentials from an environmental medium to a collecting medium, which is an organic liquid or a polymer material. The amount of the analyte collected by the sampler depends on the concentration of that analyte in the environmental medium and exposure time. As noted above, one of the unique features of passive sampling within the kinetically controlled range, which active sampling does not have, is its ability to determine the TWA concentration of an analyte. Figure 12.5 illustrates the schematic of a typical passive sampler with the parameters that have been used in Equation (12.1). In this case, Ca is equivalent to Cface. The sampler parameters and sampling time can be so adjusted that the so-called “zero sink” condition (Ouyang and Pawliszyn 2006) can be satisfied, at which Csorbent ¼ 0, and Equation (12.1) can be integrated to become ð t2 M ¼ R Cface dt ð12:2Þ t1

294

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

The mass balance after SPME has become:

Z Cface

N0 ¼ Ns þ Ndom þ Nw þ Na þ Nf

ð12:6Þ

Concentration gradient when

sorbent

Csorbent ≠ 0 A

Concentration gradient when Csorbent = 0

Csorbent

Figure 12.5. A schematic depicting the “zero sink” condition for a diffusive sampler.

where Ns, Ndom, Nw, Na, and Nf are the amounts of the analyte in the suspended solid, DOM, aqueous and air phases, and the SPME fiber after SPME, respectively. The terms moc and mdoc are the mass of OC in the solid phase and DOC in the DOM phase, respectively; and Vw, Va, and Vf are the volumes of the aqueous and air phases, and the sorbing fraction of the SPME fiber, respectively. Again by definition, Ns ¼ Coc moc, Ndom ¼ Cdoc mdoc, Nw ¼ CwVw, Na ¼ CaVa, and Nf ¼ CfVf, where Coc, Cdoc, Cw, Ca, and Cf are the analyte concentrations in the solid (OC normalized), DOM (DOC normalized), aqueous and air phases, and the SPME fiber (normalized to the polymer phase) after SPME, respectively. At equilibrium, the conventional partition coefficients can be expressed as Koc ¼

where M is the total amount of the analyte sorbed on the sorbent (R ¼ DA/Z) and t1 and t2 are the initial and final timepoints for the sampling activity. The TWA concentration is defined as in accordance with the parameters used in Equation (12.2): ð t2 Cface dt  t ¼ t1 C ð12:3Þ t2 t1

Kdoc ¼

M Rðt2 t1 Þ

ð12:4Þ

Clearly, Equation (12.4) can be used to estimate the TWA concentration of the analyte only under zero-sink conditions. Compared to kinetically controlled sampling, equilibrium passive sampling, by definition, requires that the sample (including all individual separable phases), sample container (if conducted in laboratory), and sorbent material be at equilibrium. Taking an SPME sampling in a closed container (typically a glass type) as an example, the mass balance before SPME can be expressed as 0 N0 ¼ Ns0 þ Ndom þ Nw0 þ Na0

ð12:5Þ

0 , Nw0 , and Na0 are the amounts of the analyte where Ns0 , Ndom in the suspended solid, DOM, aqueous (truly dissolved), 0 and air phases, respectively. By definition, Ns0 ¼ Coc moc , 0 0 0 0 0 0 0 Ndom ¼ Cdoc mdoc , Nw ¼ Cw Vw , Na ¼ Ca Va , where Coc , 0 0 0 Cdoc , Cw , and Ca are the analyte concentrations in the suspended solid (OC normalized), DOM (DOC normalized), aqueous, and air phases, respectively, before SPME.

ð12:7Þ

0 Cdoc Cdoc ¼ Cw0 Cw

ð12:8Þ

Cf Cw

ð12:9Þ

KH Ca0 Ca ¼ 0 ¼ RT Cw Cw

ð12:10Þ

Kf ¼ KH0 ¼

Combining Equations (12.2) and (12.3) yields t ¼ C

0 Coc Coc ¼ 0 Cw Cw

where Koc, Kdoc, and Kf are the equilibrium partition coefficients of the analyte (solid–aqueous, DOM–aqueous, and SPME fiber–aqueous); KH and KH0 are the Henry’s law constant and the dimensionless Henry’s law constant, respectively; R is the universal gas constant; and T is the absolute temperature. Combining Equations (12.5–12.10) yields Nf ¼

Kf Vf ðVw þ  þ KH0 Va Þ 0 C Kf Vf þ Vw þ  þ KH0 Va w

ð12:11Þ

where  ¼ Kocmoc þ Kdocmdoc, which is a matrix sorption term proposed by Zeng and Noblet (2002) to reflect the effects on SPME from suspended solids and DOM. Equation (12.11) can be used to estimate the initial concentration of the sorbate from the amount of the sorbate sorbed on the SPME fiber. The matrix term  can be generalized to include any number of heterogeneous solid and DOM phases. Pn Ini generi alized form  can be expressed as  ¼ i¼1 Koc moc þ Pn0 j j 0 j¼1 Kdoc m doc , where n and n are the total numbers of solid and DOM phases, respectively.

PASSIVE SAMPLING

12.3.3. Applications 12.3.3.1. Air and Water Sampling. Many types of passive samplers have been deployed for air sampling, ranging from SUMMA canisters to SPME, semipermeable membrane device (SPMD), and Polyurethane foam (PUF) samplers. Passive sampling has been effectively used to assess workplace contaminants in air such as nitrogen monoxide (NO), sulfur dioxide (SO2), ammonia, VOCs, formaldehyde, and ozone. The applications of passive sampling involve, for example, screening for the presence or absence of contaminants, assessing the temporal trends in levels of waterborne contaminants, monitoring spatial contaminants distribution, examining the distribution and fate of pollutants among environmental compartments, measuring TWA concentrations of waterborne pollutants, and comparing contaminant patterns in biota and passive samplers. A large number of target analytes have been assessed through the use of various passive sampling techniques, including organochlorine pesticides (Herve et al. 1995; Bergqvist et al. 1998; Zeng et al. 2005) and PAHs (Axelman et al. 1999) in water. 12.3.3.2. Sediment/Soil Porewater Sampling. The use of passive sampling in sediment/soil porewater has been gradually adopted. King et al. (2004) successfully assessed the PAH concentrations in sediment porewater using SPME. They asserted that the four-ring PAHs were found in the highest concentration, while the five- and six-ring PAHs were generally low and below detection limit; they attributed the low five- and six-ring PAHs concentrations to the high dissolved organic carbon (DOC) in the porewater. They also discovered that the PAH concentrations varied with the depth of the sediment core. Likewise, Sijm et al. (2000) reviewed biomimetic passive sampling methods to study the bioavailability of chemicals in soil or sediment. They compared two different sampling approaches, biomimetic equilibrium sampling approaches using SPME and EmporeÔ disks; both methods could effectively mimic partitioning of contaminants between the porewater and the organism. Both approaches likewise reflected that the freely dissolved contaminant concentrations represent bioavailability. However, they suggested that for substances that may be biotransformed in the organism, the methods would overestimate the concentration in the organism (Sijm et al. 2000) but that for organisms that have several routes of contaminant uptake, the biomimetic method would underestimate the concentration in the organisms. Matrix SPME was the main focus of the work carried out by Mayer et al. (2000); the method was applied to measure the porewater concentrations of spiked as well as field sediments, and they reported that several hydrophobic organic substances (log Kow ¼ 5.2–7.5) were measured with a high precision in the pg–ng/L range. In a more recent study, Maruya et al. developed a copper-caged (copper resists biofouling) SPME device (with a polydimethylsiloxane-

295

coated fiber as the probe) and obtained a nice correlation between the concentrations (from 0.009 to 2400 ng/L) of 12 PAH, PCB, and OCP model compounds in sediment porewater measured with the SPME device (at equilibrium) and liquid–liquid extraction. 12.3.4. Representative Passive Sampling Techniques: Semipermeable Membrane Device and Solid-Phase Microextraction 12.3.4.1. Semipermeable Membrane Device (SPMD). This device, first developed by Huckins et al. (1990), has been widely used as a bioaccumulation indicator of lipophilic organic contaminants in water, sediment, and atmosphere throughout the world. This device, which can be constructed in different forms, with a typical one displayed in Figure 12.6, mimics biological systems to provide a measure of bioavailable organic pollutants in relevant environmental matrices. The passive transport mechanism with SPMD is similar to that of fish gills and human lungs. The SPMD, however— unlike typical biota used in pollution testing—does not metabolize or excrete the sequestered compounds, is site-specific, is much easier to extract, and will not overdose on and die from the contamination that it is supposed to be monitoring. When placed in an environmental matrix, SPMDs accumulate hydrophobic organic pollutants such as PCBs, PAHs, and OCPs from the surrounding phase. Semipermeable membrane devices provide an inexpensive method of collecting organic compounds from multiple locations simultaneously. After a typical deployment period of approximately 30 days, the SPMDs are removed from the sampling matrix and dialytically recovered with a nonpolar solvent such as hexane. This extract is then concentrated, cleaned up, and enriched. The cleanup procedure typically includes gel permeation chromatography. This process removes any lipid and polyethylene waxes that might have carried over during the dialysis extraction. Further cleanup details can be found in Section 12.4. The enriched extract can then be analyzed for target analytes using chromatographic techniques. The SPMD method has been successfully applied to the sampling of air and water. Several different types of SPMD for water sampling have been described. Unfilled polyethylene membranes have been used to collect low concentrations of PAH (<17 ng/L) from water; these collectors were found to perform as well as triolein-filled samplers, and losses of accumulated PAHs were low (Carls et al. 2004). To sample PCBs, PAHs and hexachlorobenzene in porewater and surface water, SPMDs consisting of lowdensity polyethylene (LDPE) strips were used and an equilibrium time of 1–6 days was determined for compounds with Kow < 7 (Booij et al. 2003). Samplers consisting of solid poly(dimethylsiloxane) rods enclosed in water-filled or air-filled LDPE membrane tubing were used to sample 20 persistent organic pollutants (Koester 2005).

296

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

Water or air

Low density polyethylene Triolein Holes (about 10 Å) Pollutant

Heat seal

Figure 12.6. A schematic depicting a typical sampling system based on a semipermeable membrane device.

12.3.4.2. Solid-Phase Microextraction. Sampling techniques based on SPME were developed to address the need for rapid sampling and sample preparation, both in the laboratory and on -site (Pawliszyn 1997). Since its introduction in the early 1990s (Arthur and Pawliszyn 1990), SPME has been widely applied to the sampling and analysis of environmental samples, food, and pharmaceuticals (Pawliszyn 1999; Ouyang and Pawliszyn 2006). It is

basically a solvent-free sample preparation technique and combines sampling, isolation, and enrichment in one step (Fig. 12.7). It can be used to sample air, water, and solids. Once sampling is complete, the SPME fiber (or sorbent) containing the analytes of interest can be directly introduced into either a gas chromatograph (GC) or a liquid chromatograph (LC) inlet. In addition, the commercial availability of several different polymer coatings has

Figure 12.7. A schematic depicting a typical solid-phase microextraction sampling system.

SAMPLE EXTRACTION PROTOCOLS

increased the range of compounds that can be sampled with SPME methods. The obvious advantages of SPME-based sampling techniques are the low cost and ease of use as compared to other conventional large-volume water sampling methods (Zeng et al. 2004). The disadvantages of SPME samplers are their vulnerability to damage in rough environments and the fact that the detection limits they afford are restricted by the effective volume of the polymer sorbent (and are also dependent on the hydrophobicity of the analyte of interest or Kow). There have been many applications of SPME (Pawliszyn 1999); for instance, a sol-gel-derived silicone divinylbenzene (DVB) copolymer has been found useful for sampling organic phosphonates (Liu et al. 2003). Solidphase microextraction has been used extensively to sample different air samples. A DVB/Carboxen/poly(dimethylsiloxane) fiber was used to sample air from landfill sites; likewise, SPME fiber has been used to sample the combination of benzene, toluene, ethylbenzene, and xylenes in the air of a gas station (Xiong et al. 2003). The use of SPME is not limited to air alone; it has found applications in water as well. An 85-mm polyacrylate fiber, placed in a steel mesh envelope and buried in sediment, was used to sample 2,4,6-trinitrotoluene (TNT) and its degradation products from sediment waters; recommended sampling times required reaching equilibrium within 48 h at room temperature and up to 7 days at temperatures of <5 C. Detection limits for TNT and its degradation products were 10–30 ng on fiber (Conder et al. 2003). Of course, the past and recent applications of SPME are much wider than what can be described in this short section, and more detailed information on the subject can be found in Chapter 15 of this book.

12.4. SAMPLE EXTRACTION PROTOCOLS Sample extraction is an integral part of the analytical process. In general, there are two types of extraction methods: solvent extraction and nonsolvent extraction. A solvent extraction method attempts to recover as much of the analytes as possible within the shortest possible time, and can be characterized as an exhausted extraction method. The basic principle of solvent extraction is the distribution of the analyte(s) between a selected solvent or solvent mixture and the sample being extracted. As a result, repeated extractions should increase the efficiency of recovering the target analytes from the sample matrix. Typical solvent and exhausted extraction techniques include liquid–liquid extraction (LLE), Soxhlet extraction, accelerated solvent extraction (ASE), microwave-assisted solvent extraction (MASE), and supercritical fluid extraction (SFE). Although solid-phase extraction (SPE) is not exactly a solvent extraction method, it

297

is an exhausted extraction technique and still requires certain amounts of solvent on the solid–liquid exchange process. Therefore, SPE will be discussed along with other solvent extraction techniques in the following section. Nonsolvent extraction methods can be best represented by SPMD and SPME (and the like), which have been described in Section 12.3. They are not covered here.

12.4.1. Liquid–Liquid Extraction Liquid–liquid extraction (Fig. 12.8a) has long been used to recover target analytes from an aqueous phase into a solvent phase. The fundamental principle for LLE lies in the partitioning of the target analytes in two immiscible liquid phases. Liquid–liquid extraction can be performed continuously or separatorily; separatory LLE is more widely used. Typically, sample solution is placed in a separatory funnel and then thoroughly mixed with an immiscible solvent of a different density. The funnel is rigorously shaken manually or with a mechanical device, which facilitates the partitioning of the target analytes between the two immiscible phases. Phase separation is achieved by gravity or by centrifugal force depending on the type of extractor selected. The efficacy of LLE is largely dictated by the type of solvent or solvent mixture; extraction parameters such as temperature, pH, and residence time; how effectively emulsion can be minimized; and to a lesser extent, the number of extractions performed. In particular, the selection of an appropriate solvent or solvent mixture appears to be the utmost factor in achieving the maximum recoveries of target analytes. Robbins (1980) summarized the organic group interactions that can be used to identify the desired functional group(s) in the solvent for any given solute and screen possible solvents. A high distribution coefficient, good selectivity, and little or no miscibility with sample matrix are the desired properties of solvents among others. Other factors for solvent selection are boiling point, density, interfacial tension, viscosity, corrosiveness, flammability, toxicity, stability, compatibility with product, availability, and cost. The following is a typical protocol for performing separatory LLE for hydrophobic organic pollutants: 1. Measure a certain amount (typically 1 L) of liquid sample and transfer it to a 2-L (or an appropriate size) separatory funnel with PTFE stopcock. 2. Add appropriate amounts of surrogate standards and then solvent or solvent mixture (typically 60 mL; methylene chloride is widely used for a variety of hydrophobic organic chemicals) to the funnel. 3. Shake the separatory funnel vigorously for 1–2 min with periodic venting to release excess pressure and allow the organic layer to separate from the aqueous phase. If emulsion forms, stir or rotate the separatory

298

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES Pump valve

Liquid with lower density Input water Liquid with higher density Funnel tap

Extraction cell

Output water

Separating funnel

Pump

Static valve

Pump valve Solve

Sample Collection vial

Container

Nitrogen

Heat setting (a)

(c)

(b) Cooler Supercritical fluid extractor Heat exchange 1

Digestion vessel Convection current Wave-guide Sample

High pressure pump

Heat exchange 2

Reservoir glass or plastic column

vapor Sorbent bed

Heat exchange 4

Extract Glass Wool CO2 bomb

Heat exchange 3 Magnetron

Gas liquid separator

(d)

Luer tip

(e)

(f)

Figure 12.8. Schematic diagrams showing various extraction techniques: (a) liquid–liquid extraction; (b) Soxhlet extraction apparatus; (c) accelerated solvent extraction; (d) microwave-assisted solvent extraction; (e) supercritical fluid extraction; (f) solid-phase extraction.

4.

5. 6. 7.

8.

9.

10. 11.

funnel gently. If this does not work, centrifugation or filtration may be needed. Place a flat-bottomed flask (e.g., 500 mL) with a shortstemmed glass funnel sitting on top, plugged with glass wool and with approximately 20 g of anhydrous sodium sulfate beneath the separatory funnel. Collect the solvent extract in the flat-bottomed flask. Repeat the extraction twice and combine the three extracts. Transfer the combined extract to a concentration tube and evaporate the solvent using an evaporator (e.g., Zymark TurboVap500, Zymark, Hopkinton, MA, USA) with optimized operational parameters. Add 10 mL of pesticide-grade hexane to the flask and concentrate further until the volume is reduced to 1 mL (solvent exchange). Transfer the concentrated extract from the flask to a 1-dram vial with PTFE-lined cap. Rinse the walls of the flask with hexane and transfer the rinsate into the 1-dram vial. Concentrate the sample to approximately 1 mL under a gentle nitrogen flow. Store the extract in a freezer at 20 C for subsequent cleanup and separation into different fractions.

12.4.2. Soxhlet Extraction A typical Soxhlet solvent extraction system (displayed in Fig. 12.8b) involves the placement of a solid sample in a tube and continuous passage of a heated organic solvent or solvent mixture through the sample. This method has two advantages: (1) the heated solvent or solvent mixture enhances the diffusion of the target analytes out of the sample; and (2) the system can be automated so that a reasonable number of samples can be extracted simultaneously. The drawbacks of Soxhlet extraction include the prolonged processing time and use of large amounts of organic solvent. In general, Soxhlet extraction can be conducted within the timeframe of 8–72 h, depending on the types of sample and solvent, and more than 200 mL of organic solvent may be needed for each sample. The heating temperature should be set to just above the boiling point of the solvent or solvent mixture selected. If the temperature is too high, significant volatilization and thermal degradation may occur, resulting in substantial losses of the target analytes. On the other hand, if the heating temperature is too low, solvent flushing may be inefficient, leading to incomplete extraction. From our experience, 5–6 drops of the solvent per minute is considered suitable. All Soxhlet extraction setups using organic solvent should be placed in a fume hood. The following is a typical protocol for Soxhlet extraction of filtered suspended

SAMPLE EXTRACTION PROTOCOLS

particles (also applicable to other solid samples with slight modifications): 1. Dry the samples with a suitable method such as freeze drying, and weigh the masses of the samples before and after drying (the amount of solid samples typically weighs 10–20 g). 2. Shred the filters into small pieces using solvent-sonicated stainless-steel forceps and pack the shredded filters with a piece of porous extraction thimble or preextracted filter paper. 3. Spike the samples with desired amounts of the surrogate standards selected and place them into the tubes of the Soxhlet system. 4. Add a suitable amount (typically 200 mL) of a solvent or solvent mixture into the flat-bottomed flask. 5. Connect the tube, condenser, and the flat bottomed flask, and place them onto a heater. 6. Connect the condenser with a circulating water flow system, and run the water flow. The flow should be constant, steady, and gentle to avoid pressure buildup in the tubing, which can lead to a disconnect of the tubing. Use of a cooling system can achieve greater efficiency and minimal solvent loss. 7. Turn the heater to start extraction with an appropriate temperature during the desired timeframe (it is advisable to hold on for a while to ensure that there is a smooth reflux). 8. After extraction, transfer each extract into a clean flask and rinse the flat-bottomed flask with 3–5 mL of the solvent or solvent mixture 3 times and combine all extracts for cleanup and concentration. 12.4.3. Accelerated Solvent Extraction An ASE system (Fig 12.8c) employs increased temperature and pressure to enhance and optimize the kinetics of the extraction process. In general, the solubility of most organic solvents will increase when extraction temperature increases. Pitzer (1995) found that the solubility of anthracene increased by about 13 times when the extraction temperature increased from 25 C to 150 C. The advantages of ASE over traditional solvent extraction methods are shorter time and much less solvent required. For example, a 1–30-g soil sample can be extracted in 15 min using solvent volumes 1.2–1.5 times that of the extraction cells (typically in the range of 3.5–32 mL) (Richter et al. 1996). Accelerated solvent extraction can also be automated, allowing up to 24 samples to be extracted unattended and simultaneously. However, two drawbacks of ASE should be noted: 1. A large number of pores on the surface of a solid sample may be surrounded by water, causing sequestration of

299

the target analytes. This sequestered portion is seldom considered to be bioavailable and cannot be extracted with conventional extraction methods. With ASE, the synergism of high extraction temperature and pressure greatly enhance the solubility and mass transfer efficacy of the target analytes, thus increasing the release of sequestered chemicals and probably overestimating the bioavailability of the target analytes. 2. Thermal decomposition of the target analytes may become possible because of the intensified temperature and pressure employed in ASE. For example, DDT could be degraded to DDD and DDE and aldrin to endrin aldehyde and endrin ketone during ASE extraction (Richter et al. 1996). It is therefore important to optimize the extraction parameters. The following steps are typically employed in extraction of a solid sample with ASE: 1. Dry the sample using a suitable method such as freeze drying. 2. Weigh an appropriate amount of the sample and place it into the extraction cell. 3. Spike desired amounts of surrogate standards into the sample cell. 4. If the sample is a PUF loaded with particles, use a clean spatula to depress the PUF in the middle and roll it up in the cell. 5. Slide out the spatula and twist closed the open ends of the cell. 6. Twist the rolled PUF into a 22-mL ASE cell and cap. 7. Fill with desired volume of solvent in 0.5–1 min. 8. Heat and pressure for 5 min. 9. Static extraction for 5 min. 10. Flush with fresh solvent for 0.5 min. 11. Purge with nitrogen for 1–2 min. 12.4.4. Microwave-Assisted Solvent Extraction In this procedure, the solvent or solvent mixture used is heated via absorption of microwave energy, and in turn the heated solvent or solvent mixture can more easily penetrate the sample placed in a glass liner. Therefore, the capacity of the solvent to absorb microwave energy is the controlling factor. A solvent must possess dielectric polarizability to be able to absorb microwave energy. There are several types of dielectric polarization depending on the chemical nature of the solvent. Microwave absorption results from reorientation of permanent dipoles by the electric field. The amount of microwave energy absorbed is proportional to the dielectric constant (e0 ) of the solvent and is, in most cases, proportional to its polarity. For example, the dielectric constants determined at a frequency of 3 GHz and a temperature of 25 C are

300

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

76.7, 23.9, and 1.9 for water, methanol, and heptane, respectively (Zlotorzynski 1995). Besides, the ability of the solvent to convert microwave energy into heat is also crucial. The conversion efficiency is governed by the dielectric loss factor (e00 ). For the solvents mentioned above, the dielectric loss factors are 12, 15, and 0.00019, respectively. Evidently, the dielectric constant and the dielectric loss factor of a certain matrix can differ greatly. Therefore, the overall heating efficiency using microwave energy is usually expressed by the dissipation factor (tan @), which is the ratio of the dielectric loss factor and the dielectric constant of the matrix involved (Zlotorzynski 1995). The resulting dissipation factors are 1570, 6400, and 4 for water, methanol, and heptane, respectively. In comparison to conventional extraction procedures, the benefits of MASE include shortened extraction time, low solvent consumption, and improved extraction efficiencies. The following is a typical protocol for MASE extraction of filtered suspended particles (also applicable to other solid samples with slight modifications) with a schematic of the experimental setup displayed in Figure 12.8d: 1. Dry the sample (take one glass fiber filter loaded with particulates as an example) with a suitable method such as freeze drying and weigh the mass of the sample before and after drying. 2. Shred the filter into small pieces using solventsonicated stainless-steel forceps. 3. Place the shredded filter pieces in the microwave glass vessel that has been spiked with desired amounts of surrogate standards and capped with a PTFE lid. 4. Add a desired amount of solvent or solvent mixture into the vessel. 5. Assemble the extraction unit by sliding the vessel into a sleeve and placing it into a support module. Tighten the nut at 108 inlb. A temperature control vessel must be used in each batch for monitoring. (Important note: Samples high in organics/water are capable of generating more heat and pressure than are samples low in organics/water. The temperature control vessel determines the amount of microwave emitted to all the other samples in the batch. If a sample with high organics/ water is used as a control vessel, it prevents higher temperatures and overpressurization in other samples with similar characteristics.) 6. Place the turntable containing the sample into the instrument cavity on the drive lug. Insert the fiberoptic temperature probe into the thermowell of the vessel. Note that the temperature control vessel must be in a particular location on the turntable. 7. Close the door and initiate the extraction process by pressing the appropriate buttons to recall and run the desired method.

8. On completion of the extraction cycle, allow the vessels to cool (inside the microwave cavity) for approximately 1 h until the temperature drops below 35 C. 9. Open the door. Press the rotation button to position vessels for removal. The turntable may be left in place and vessels removed individually. 10. Remove the fiberoptic temperature probe from the control vessel glass thermowell. 11. Carefully loosen the vessels from the support module with a wrench and remove the vessels from the composite sleeves. 12. Collect the solvent layer into a 250-mL flat-bottomed flask (with a 24/40 ground-glass joint) and rinse each vessel with 3–5 mL of solvent 3 times and combine the extracts for cleanup and concentration.

12.4.5. Supercritical Fluid Extraction Supercritical fluid extraction is not yet a widely used method. However, as new technologies are emerging, it will likely become an extraction method with the features of high purity, low solvent usage, and being environmentally friendly. Figure 12.8e sketches a flowchart of typical SFE procedures. The basic principle of SFE lies in the interaction between the feed material and a supercritical fluid (SCF), allowing the volatile substances to partition into the supercritical phase. On dissolution of soluble materials, the SCF containing the dissolved substances is removed from the feed material. The extracted target analytes are then completely separated from the SCF by altering temperature and/or pressure. Some advantages of SFE techniques when compared to conventional methods include: (1) the salvation power of SCF is controlled by pressure and/or temperature, (2) SCF is easily recoverable from the extract because of its volatility, (3) the use of nontoxic solvents leaves no harmful residues, (4) components of high boiling points are extracted at relatively low temperatures, (5) separation previously impossible using more traditional processes can sometimes be effected with SFE, and (6) thermally labile compounds can be extracted with minimal damage as low temperatures can be employed by the extraction. However, SFE is also laden with some disadvantages such as an elevated pressure required; as a result, the compression of solvent requires recycling measures to reduce energy cost and high capital investment for equipment. In addition, SFE often can extract only a specific class of chemicals with similar physicochemical properties in each extraction cycle; thus it may not be desirable for simultaneous extraction of multiple components. The choice of SFE solvent is similar to that of regular extraction. Carbon dioxide is the most commonly used SCF, due primarily to its low critical points (temperature 31.1 C and pressure 73.8 bar), low cost, and nontoxicity. Modifier

SAMPLE EXTRACTION PROTOCOLS

solvents are sometimes needed to alter the optimal extraction conditions to achieve better extraction efficacy. The following is a typical protocol for SFE extraction of filtered suspended particles (also applicable to other solid samples with slight modifications): 1. Dry the sample (take one glass fiber filter loaded with particulates as an example) with a suitable method such as freeze drying and weigh the mass of the sample before and after drying. 2. Open the fluid valve, open the valve on the SFE instrument, and turn on all components, including the coolant system. 3. Clean out the system by starting a run in the chamber without adding any solvents, and then clean off restrictors by using 5 mL each of the following solvents: hexane, methanol, acetone, and methylene chloride. 4. Add the sample into the sample cartridge and spike it with desired amounts of surrogate standards. 5. Add a suitable amount of modifier such as methylene chloride for PCBs (Tong and Imagawa 1995) and methanol for PAHs (Lutermann et al. 1998). 6. Seal the sample cartridge and install it in the extraction chamber by snapping it into the bottom of the chamber cap and insert it into the extraction chamber. Handtighten the chamber cap. 7. Fill the collection vial with a suitable solvent and tighten its cap. 8. Press the key to start the extraction. 9. On equilibration, note that fluid is passed through the chamber and collected in the collection vessel. 10. Once the extraction is complete, open the vent and depressurize the system. 11. Rinse off the restrictor with 3–5 mL of solvent and add the rinsate to the extract. 12. Transfer the extract from the collection vial into a flask for cleanup and concentration. 12.4.6. Solid-Phase Extraction Solid-phase extraction continues to be a leading technology for the extraction of both organic and inorganic species from aqueous samples. In fact, the term “solid- phase” or “sorbent extraction,” frequently abbreviated “SPE,” simply implies a physical extraction process involving a liquid phase and a solid phase. In practice, it means the use of commercial prepacked columns (or disks) containing stationary phases that may be adsorbents such as silica gel, reversed-phase materials, or ion exchange media (Fig. 12.8f). Solid-phase extraction uses many of the same types of stationary phases as used in LC columns. The stationary phase is contained in a glass or plastic column (glass is preferred for organics) above a frit or glass wool. In actual applications, the stationary phase

301

is selected depending on the type of samples and major target analytes. In addition, the adsorptive capacity of the stationary phase must be considered, and an appropriate volume of stationary phase is selected according to the adsorptive capacity and the sample volume. The SPE method is attractive because it affords relatively easy concentration of the analytes of interest, requires the addition of minimal amounts of solvent, and can be tailored to extract either a broad range of analytes or to provide specific extraction of a pollutant or compound class. It is often used to clean up a sample before a chromatographic or other analytical method is used to quantify the amount of analyte(s) in aqueous samples. It should be noted that although almost no solvent is needed during the extraction process, conditioning of SPE columns or disks and eluation of the target analytes from the sorbent phase do require fair amounts of solvents. The following is a typical protocol for SPE extraction of aqueous samples: 1. Remove suspended particulate matter from the aqueous sample using glass fiber filters with a vacuum system. 2. Assemble an apparatus for filtering the aqueous sample through an SPE column (an SPE disk can also be used), and connect the apparatus to the vacuum pump. 3. Spike the aqueous sample with desired amounts of surrogate standards. 4. Condition the SPE system with appropriate amounts of desired solvents. Allow the solvents to stand for a few minutes and turn on the vacuum pump to drain the solvents completely; repeat this action 3 times. 5. Add the aqueous sample to the reservoir, turn on the vacuum pump, and begin to filter. 6. Rinse sample bottle with deionized water and add to the reservoir. 7. Drain as much water from the sorbent bed as possible with the vacuum pump. 8. Place a drip tube to collect extract. 9. Rinse reservoir with a desired amount of known solvent. 10. Elute the SPE sorbent bed with a desired amount of known solvent. 11. Collect the extract for subsequent concentration and cleanup. 12.4.7. Extract Concentration and Solvent Exchange 12.4.7.1. Extract Concentration. All extracts collected from the extraction procedures discussed above or those not discussed here may be subject to further concentration. Extract concentration is a necessary step in sample processing, typically performed prior to extract clean-up/ chromatographic fractionation and/or before instrumental analysis. Before instrumental analysis, the extract volume

302

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

Heating tape Valve

Rotovap Control dial Distillation flask Extract

Vacuum pump system

Circulating condensed water system

Heated bath

Extract

(a) Manifold N2 bottle (b)

Figure 12.9. Schematic diagrams showing (a) a typical rotary evaporator and (b) a nitrogen evaporator.

is often reduced to 10–1000 mL depending on the sample type and the expected levels of target analytes. The extract volume is generally decreased to 1–2 mL or less prior to cleanup/ chromatographic fractionation. Two devices have been widely used to perform extract concentration: .

.

Rotary Evaporator. A rotary evaporator (Fig. 12.9a) is an instrument designed to distill a liquid under reduced pressure. It consists of a heated rotating vessel (usually a large flask), which is maintained under a vacuum through a tube connecting it to a condenser. The rotating flask is heated in a hot-water bath. The rotation provides enhanced heat transfer to the liquid inside, and also strongly reduces the occurrence of bumps caused by superheating of the liquid. The solvent vapors leave the flask by the connecting tube and are condensed in the condenser section. The condenser section is so arranged that condensed vapors drain into another flask, where they are collected. It is an efficient way to rapidly remove large quantities of solvent. Rotary evaporation is mainly used to recover non- or semivolatile solutes in preparative chromatography and solvents for recycling purposes. Nitrogen Evaporator. The second device is the nitrogen (N2) evaporator (Fig. 12.9b) designed to concentrate solvent by nitrogen stream supplied by a highpressure nitrogen container. A commercial N2 evaporator often contains a heating system. Solvent volatilizes quickly under desired temperature and N2 stream. In general, large solvent volume and solutes with high

boiling points are often processed with rotary evaporators, but rotary evaporators are not designed to reduce solvent to a desired volume. On the other hand, the N2 evaporator is effective in reducing a small amount of solvent to a metered volume. Very often an extract is reduced to 1–2 mL by rotary evaporation, and the extract is transferred to a vial and further concentrated with a N2 evaporator. By comparison, a rotary evaporator can process only one sample at a time, whereas an N2 evaporator can concentrate a large number of samples simultaneously. In any event, the efficacy of extract concentration can be assessed by the recovery data of the surrogate standards from spiked samples. 12.4.7.2. Solvent Exchange. The goal of solvent exchange is to replace one solvent with another appropriate for chromatographic separation. In general, a solvent with the weakest polarity possible should be used so that the solutes are strongly adhered to the stationary phase (not eluated with solvent as the extract is applied to the separation column), which subsequently ensures effective fractionation of the target analytes on the basis of their polarities. The polarities of several commonly used solvents are as follows: hexane < petroleum ether < toluene < benzene < methylene chloride < acetone < acetonitrile < methanol. Therefore, extract solvents are normally exchanged to hexane before chromatographic separation. In practice, a desired amount (often 5 mL is used) of hexane is added to the extract and its volume is reduced to 1–2 mL with a rotary evaporator. As the extract is further concentrated, the original solvent or solvent

SAMPLE EXTRACTION PROTOCOLS

303

mixture is preferably evaporated, leaving hexane as the lone solvent in the extract.

12.4.8. Cleanup and Fractionation Procedures 12.4.8.1. Removal of Elemental Sulfur. Elemental sulfur is widely distributed in sediment rocks and samples contaminated with crude oil. A cleanup procedure to remove sulfur is necessary because of the strong interference of elemental sulfur in qualitative and quantitative determination. Treatment with activated copper for removal of elemental sulfur is the most widely used approach. On extraction, small pieces of activated copper are added to the extract to react with elemental sulfur to form copper sulfide. Usually, cupric oxide covers the surface of copper; hence copper should be activated before use. A simple procedure for activation of copper is as follows. 1. Shred copper into small pieces using stainless-steel forceps (this step is not necessary if granulated copper is used). 2. Place the copper pieces into an ion exchange column with a fritted disk. 3. Add adequate volume of 6 N HCl over the copper pieces and discard the rinsate; repeat this step 3 times. 4. Add adequate volume of purified water over the copper pieces and discard the rinsate; repeat 3 times. 5. Add adequate volume of methanol/acetone over the copper pieces and discard the rinsate; repeat 3 times. 6. Add adequate volume of methylene chloride over the copper pieces and discard the rinsate; repeat 3 times. 7. Transfer the copper pieces to a glass container and add suitable volume of methylene chloride.

12.4.8.2. Removal of Lipid from Biological Specimen. Lipid is an important fraction extracted from an environmental sample, especially for biological specimens. Size exclusion chromatography, also called gel filtration or gel permeation chromatography (GPC), uses porous particles to separate molecules of different sizes and has been widely used in processing nonliquid matrices such as fruit, vegetables, and animal products. Gel permeation chromatography is a column chromatography technique employing swollen gel made from polymerizing and crosslinking styrene as the stationary phase (Fig. 12.10). As the sample extract is introduced at the top of the column and eluated with a solvent, molecules diffuse through the gel at rates depending on their sizes; molecules that are smaller than the pore size can enter the gel and therefore have a longer path and longer transit time than do larger molecules that cannot enter the gel (Fig. 12.10). With the same mechanism, lipid can pass

Small molecules

Large molecules

Figure 12.10. A schematic showing a typical gel permeation chromatograph.

through the GPC column in a shorter time than the target analytes can and is therefore separated from the sample extract. The following is a brief summary of a typical GPC procedure. .

.

Swelling of Beads. Before use, GPC beads should be swelled in a suitable solvent such as aromatics, methylene chloride, benzene, and ketones. The chosen solvent should be the one used for the separation, as well as the solvent in which the sample is dissolved. Any solvents used for GPC should be of highest quality available. On swelling, beads are packed into a chromatographic column and washed with the solvent used for swelling. Column Packing. Glass columns are often used in GPC, because they are visible while packing. Set up a clean column assembly, and place a small amount of solvent in the column to prevent bubble formation at the base of the poured column packing. The solvent reservoir may be connected to the column by any method, but it must be air-tight and maintained clean. Never allow the packed beads to become dry, because this will cause air pockets and channeling within the bed, resulting in poor efficiency and low resolution. If any portion within the column becomes dry, add extra solvent, cap both ends, and invert several times until the resin is fully slurred. Then allow the resin bed to settle, and begin flow as normal.

304 .

.

.

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

Flow Rate. Beads can be used at different flow rates depending on the crosslinkage and style. For example, the 1%–2% crosslinked resins for Bio-Beads S-X1 beads are quite soft when fully swollen and should be used only in gravity flow procedures. Bio-Beads S-X3 beads can withstand 5 mL/min with a backpressure of 300 psi. Fraction Collection. Various fractions containing different analytes can be collected for further analysis. A detector can be used to determine when the analytes are eluting from the column, and ultraviolet detector is a common choice. Regeneration. Regeneration of beads is necessary if the target analytes have become trapped within the pores of the resins. The beads are swelled with methylene chloride, toluene, or tetrahydrofuran to their maximum. After a long period of storage, small, loose polystyrene can leach out. This polystyrene, even with trace amounts, has high ultraviolet absorption because of its high extinction coefficient. One way to eliminate this problem is to wash the beads with several bed volumes of solvent.

12.4.8.3. Cleanup and Fractionation: Silica Gel/Alumina Column Chromatography. The preceding section presents a general procedure based on GPC to remove lipid from biological specimens. For nonbiological samples, GPC removal of lipid may be unnecessary, but sample cleanup is still a critical step in removing unwanted constituents and minimizing interference with subsequent instrumental analysis. In many cases, separation of chemical fractions of various polarities will facilitate qualitative and/or quantitative analyses, which can be applicable to both biological and nonbiological samples. In practicc, a method capable of performing both sample cleanup and fractionation is desirable for achieving cost-saving and fast turnaround time. Silica gel/ alumina column chromatography seems to be able to fulfill this objective. Silica gel/alumina column chromatography is a technique that has been widely used for extract purification in environmental analysis because of its simple packing procedure, low operating pressure and low operating cost. In a typical setup, silica gel/alumina serves as the stationary phase and is packed into a vertical glass column, while the mobile phase, a solvent or solvent mixture, is slowly flushed from the top of the column and allowed to flow down through the column by either gravity or external pressure. A sample extract is applied to the top of the column. Separation is achieved because of the different interactive strengths of different target analytes in the sample extract between the stationary and mobile phases, resulting in different rates for different target analytes dripping out of the silica gel/alumina column. In practice, a good application of silica gel/alumina column

chromatography involves the selection of appropriate adsorbents (stationary phase) and eluating solvent or solvent mixture (mobile phase), as well as the preparation of robust and consistent silica gel/alumina columns. Commercially available silica gel and alumina are sold in various mesh sizes (larger mesh size means smaller particles) and in acidic, neutral, and basic forms. In addition, alumina is often characterized with different activities, I–III; activity I is the most active. Furthermore, the polarity of a solvent or solvent mixture used as the mobile phase determines the relative rates at which the target analytes move through the silica gel/alumina column. Therefore, a series of increasingly polar solvents or solvent mixtures are often used to elute a column. The cleanup and subsequent fractionation steps mark the end of the sample preparation process before instrumental analysis. However, depending on the sample type and target analytes under investigation, various degrees of cleanup may be needed to achieve cost-effectiveness while maintaining sufficient quality. The fractions resulting from column chromatography are also a subject of interest, and these are achieved through the use of solvents with different polarities. Because of the versatility in applications of silica gel/alumina column chromatography, the readers are encouraged to check out the literature for detailed protocols using this technique. The following is a summary of how typical column chromatography is prepared and operated. .

.

.

.

Adsorbent Cleaning. To a 1-L beaker add approximately 350 g of silica gel and 900 mL of redistilled methanol. To another 1-L beaker add 500 g of alumina and 500 mL of redistilled methanol. Sonicate the mixtures for 30 min. Decant the methanol. Rinse the adsorbents 3 times with 100–200 mL of methylene chloride. Decant the methylene chloride. Add 500 mL of methylene chloride to each beaker and sonicate for 30 min. Decant the methylene chloride. Adsorbent Activation. Allow the adsorbents to dry in a fume hood overnight. Cover the beakers with aluminum foil with punctures to permit escape of vapors but prevent entry of particulate matter. Activate silica gel at 180 C and alumina at 250 C overnight. Transfer the adsorbents into 500-mL Erlenmeyer flasks and allow them to cool at room temperature. Adsorbent Deactivation. Weigh a preset amount of the adsorbents. Add an appropriate amount (e.g., 3% by weight) of purified water to the adsorbents and shake the flasks vigorously. Allow the flasks (covered with a ground-glass joint) to sit overnight to equilibrate. Storage. Add dry hexane (treated with anhydrated sodium sulfate prior to use; thus designated as dry hexane thereafter) to the flasks to prevent the adsorbents from hydrating. The adsorbents are now ready for use.

SAMPLE EXTRACTION PROTOCOLS .

.

Column Packing. Plug a glass column with a PTFE stopcock with a small amount of cotton to prevent the adsorbents from leaking out. Rinse the inner walls of the column with 10 mL of dry hexane to remove glass fibers and any organic residues in the apparatus. Allow the solvent to run through the stopcock. Close the stopcock and add 5 mL of dry hexane. Add silica gel to the column as a slurry to a desired height (e.g., 12 cm). Occasionally tap the column with a plastic rod to pack the column and remove any bubbles. Continue rinsing the walls of the column with dry hexane to prevent buildup of dry silica gel during packing (silica gel is always maintained wet with solvent). Using a separate pipette, add a desired height of alumina slurry over the silica gel layer. Lower the solvent level to just above the alumina surface and close the stopcock. Chromatographic Separation 1. Place a 35-mL pear-shaped flask beneath the column and run it through the silica gel/alumina column with the sample extract (usually in 1 mL). Rinse the vial and column walls with three 250-uL portions of dry hexane. Eluate the first fraction (e.g., aliphatic hydrocarbons; labeled as F1) with 10 mL of dry hexane. 2. Place another 50-mL pear-shaped flask beneath the column and eluate with 5 mL of dry hexane and 30 mL of a 30/70 mixture of methylene chloride and hexane to collect the second fraction (F2), often containing PAHs, PCBs, and DDTs. 3. Continue with 30 mL of methylene chloride to extract more polar components if necessary, with a 125-mL boiling flask. 4. Place another 50-mL pear-shaped flask beneath the column and eluate with 25 mL of methanol to collect the third fraction (F3) if necessary. Add F3 to a 500-mL separatory funnel. 5. Rinse the pear-shaped flask with methanol, followed by deionized water, and add rinsate to the separatory funnel. Add 25 mL of deionized water and 50 mL of methylene chloride to the separatory funnel. 6. Shake for 1 min to release vapor pressure through the stopper. Drain the methylene chloride layer into a 250-mL boiling flask. 7. Add another 50 mL of methylene chloride to the separatory funnel and repeat step 6. Repeat once more for a total of three extractions. 8. The three fractions collected using steps 1–7 are separately subject to further concentration to a desirable volume using an evaporator. Each fraction is then spiked with appropriate amounts of internal standards and ready for instrumental analysis.

305

12.4.9. Quality Assurance and Quality Control 12.4.9.1. Basic Concept. Quality assurance (QA) and quality control (QC) are closely related, but carry different features. The definitions of both QA and QC are multifold from different sources for different purposes. Readers are encouraged to search the literature for various forms of definitions for QA and QC to meet a specific objective. Within the context of environmental analysis, QA contains a series of measures that are established to ensure that chemical analysis is done in compliance with the preset criteria, whereas QC involves a set of actions that are used to maintain the quality of analytical results within the preset criteria. In other words, QA ensures that the process is well defined and appropriate, while QC focuses on finding defects in specific deliverables. Both QA and QC are requisite steps in environmental analysis. From the results of QA and QC (conventionally designated as QA/QC), we are able to identify the sources of the problems, diagnose, and/or troubleshot to determine the nature and scope of the problems and make appropriate corrections and adjustments. These steps may need to be repeated if necessary. As an example, the QA/QC activities from sample collection to instrumental analysis can be organized in three stages: 1. To ensure the success of a sampling cruise, the key QA/QC measures include selection of an appropriate sampling technique, collection of representative samples, prevention and minimization of external (cross-) contamination during sampling, and safe transportation of the samples to the laboratory. For example, when sampling on a boat, the sampling point should be chosen upstream of the boat to minimize any possible impact due to pollutant discharge from the boat. Another example is to use field blanks to monitor external contamination during sampling and take corrective actions if external contamination becomes severe. 2. In the laboratory, QA/QC measures for sample processing include, but are not limited to, the selection of an appropriate extraction method, use of contaminantfree glassware and supplies, and use of proper internal and surrogate standards. In addition, procedural blanks, spiked blanks, matrix-spiked samples, sample replicates, and standard reference materials (SRMs) are often processed along with a certain number (usually 10–20) of field samples to monitor the performance of the laboratory procedures and analysts and identify the source(s) of any possible problems. 3. During the course of instrumental analysis, the key components of a rigorous QA/QC program include, but are not limited to, the selection of a suitable instrument and a chromatographic column, proper instrument

306

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

tuning and calibration, establishment of one standard calibration range for each analyte and evaluation of method detection limit (MDL; see Section 12.4.9.6 for more details on MDL).

. . .

12.4.9.2. Prevention of Background Interference. The following actions should be taken to eliminate and/or gauge any background interferences resulting from the sample preparation process: .

.

.

.

A procedural blank should be processed to demonstrate the absence of interference for the target analytes. The level of interference must be below the MDL or other preset detection limits before this method can be performed on actual samples. All glassware should be washed with soap and water, rinsed with distilled water, and kilned at 450oC for 4 h (or cleaned with an appropriate strong-acid solution). Flush glassware immediately before use with the same solvent that is to be used in the analysis. Microsyringes used for any purpose during the sample processing should be cleaned thoroughly by rinsing them with an appropriate solvent (e.g., methylene chloride) and final rinsing with another solvent (e.g., hexane). Silica/alumina column saturation occurs when the content of extractable organic matter exceeds the capacity of the column, resulting in reduced analyte recoveries. To avoid column saturation, divide 25,000 by the theoretical lipid weight (which is determined by gravimetric analysis) of the sample. This will give you the sample volume in microliters that will saturate the column. Adjust the number of columns needed for cleanup/ fractionation accordingly.

12.4.9.3. Use of Surrogate and Internal Standards. The use of surrogate standards is designed to gauge the performance of the entire sample preparation process. This is determined on the basis of the percent recoveries of the spiked surrogate standards. It is intended to compensate for various aberrations in extraction, cleanup, transverses, and detectors. It is highly appropriate and also recommended that the surrogate standards be isotope-labeled (2 H or 13 C) so as not to interfere with the determination of analyte concentrations. On the other hand, the use of internal standards is intended to compensate for the variability of instrument responses to the masses of target analytes during instrumental analysis. The list of surrogate and internal standards for each analysis differs and could be confirmed from published data and established protocols. Successful surrogate and internal standard candidates usually possess the following attributes:

They are unlikely to exist in most, if not all, environmental matrices. They can sustain chemical breakdown during the course of sample extraction and instrumental analysis. Their physicochemical properties are similar to those of the target compounds. This attribute guarantees that the surrogate and internal standards are eluted closely with the target analytes in cleanup/fractionation and instrumental analysis.

12.4.9.4. Use of QA/QC Samples. To monitor and/or assess the performance of various analytical steps, QA/QC samples are often processed simultaneously with each set of field samples (typically 10–20). Typical QA/QC sample type includes field blank, procedural blank, spiked blank, matrix-spiked sample, sample replicate, and SRM. Sometimes a matrix-spiked replicate can be used in place of sample replicate or vice versa. Field blank samples are used to detect any external contamination during sample collection and transport processes. In general, a clean matrix similar to the sample type to be collected is brought to the field, exposed to the ambient environment, transported to the laboratory on sample collection, and then processed with the field samples. Solvent blank samples are the solvent (used in sample processing) spiked with surrogate standards; they undergo the entire procedure to detect any external contamination during the laboratory process. Spiked blank samples are similar to solvent blank samples, except that target analyte standards are also added to the solvent besides surrogate standards. Recoveries of the target analytes from spiked blank samples are indicators of the efficacy of the sample processing steps, such as whether the solvent selected and the heating temperature are appropriate. Matrix-spiked samples are prepared from selected field samples spiked with known amounts of the target analytes and surrogate standards. The spiking concentrations of the target analytes should be 5–10 times those naturally occurred in the field samples. In case such field samples are difficult to obtain, the field samples may be extracted lightly to deplete the analyte concentrations and used as matrix-spiked samples. The recoveries of the target analytes from the matrix spiked samples are used to gauge any potential interference with the analytical results from the sample matrix. Sample replicate (or matrix-spiked replicate), as indicated by its name, is a second portion of the same larger sample designated as “sample” and is used to examine the precision of the analytical method employed. Finally, SRMs often refer to the commercial standards supplied by the U.S. National Institute of Standards and Technology (NIST), which provides certified values for concentrations of the target analytes in SRMs, which are determined with several analytical approaches. The accuracy of an analytical method can be assessed through analyses of SRMs.

INSTRUMENTATION

12.4.9.5. Instrument Calibration. An analytical instrument needs to be calibrated prior to initial use and after each set of samples is analyzed or after each preset time period has passed. Initial calibration is to establish calibration curves for the target analytes for subsequent quantitative analyses. Generally, five concentration points are sufficient to define a calibration range with adequate confidence. Because the analyte concentrations may vary widely in environmental samples, the concentration range for the calibration curves should be as wide as possible, and in some cases more than five concentration points may be used to increase the flexibility of the quantitation method. The correlation coefficient (r2) and standard deviation (SD) are often used to evaluate the quality of calibration curves, with typical acceptable values of r2 > 0.99 and SD < 20%. In practice, though, it is unrealistic to require r2 > 0.99 and SD < 20% for a large number of target analytes. A practical guideline is that the calibration be regarded as acceptable if the calibration curves for >80% of the target analytes satisfy r2 > 0.99 and SD < 20%. Continuing calibration is to determine whether the calibration curves established in initial calibration remain valid prior to the analysis of a new set of samples or after the instrument parameters have been adjusted. Usually a solution containing all target analytes at the midpoint concentrations on the calibration curves is injected into the instrument and the target analytes are quantified with the calibration curves. The acceptance criteria may vary depending on the requirement of a specific project or program, but a recovery of 85% relative to the injected concentration is often taken to validate the calibration curves. A new initial calibration should be conducted if > 20% of the recoveries are < 85%. 12.4.9.6. Detection Limit. A detection limit is defined as the predetermined threshold against which the concentration of an analyte is compared to determine whether the analyte is present with a positive amount or is “not detected” (also called “less than”) in the sample. A positive value apparently carries a significant consequence for data analysis and interpretation. However, a less than value is equally important for regulatory agencies to reach legally binding decisions and for instrument manufacturers and analytical laboratories to demonstrate superior performance on an analytical report. As a result, it is as much a data point as a positive value and should be determined with equal precision and accuracy. Quite a few detection limits have been established to determine this less-than value; including quantitation limit, practical quantitation limit, reporting limit, instrument detection limit, and MDL, among others. The widely quoted approach to determine the less-than value is perhaps the MDL defined by the U.S. Environmental Protection Agency (USEPA), which is the smallest concentration of an analyte in a given matrix and is determined by a given method with a 99% confidence that this value is not zero. The detailed

307

procedures for acquiring MDLs for a given method can be found in a publication by Kimbrough and Wakakuwa (1993). Readers are encouraged to consult literature for the definitions of all other types of detection limits.

12.5. INSTRUMENTATION The quality, accuracy, and reproducibility of the entire sample preparation process will be revealed and assessed by the output from the instrumental analysis; as such, a high premium must be paid to the instrument precision. Numerous instruments and instrumental techniques have been developed and utilized in analytical research and applications; however, for the purpose of this section we will highlight the techniques that are most pertinent to instrumental analysis of organic pollutants in environmental matrices. Because actual instrumental parameters and procedures vary vastly from application to application, this section focuses on the fundamental principles and general features of the commonly used instruments and related technologies. Readers should consult the available literature for step-by-step procedures. 12.5.1. Chromatography Techniques Chromatography is a powerful and versatile technique for separating mixtures. A typical chromatography system contains a stationary phase (solid or liquid) and a mobile phase (liquid or gas, which defines LC or GC). Because sample extract is technically dissolved in the mobile phase as it is passed through the stationary phase, the two phases are so chosen such that the target analytes in the extract will have differential solubilities in each phase. The difference in solubility, which turns into a difference in mobility between the two phases, is the driving force in separating the analytes from each other as they transverse the stationary phase. Different types of chromatography, such as paper, thin-layer, and column chromatography, have been developed and veritably tested to be effective. Notably, column chromatography can be used to determine the number of analytes in a mixture, as well as to separate and purify substantial amounts of various analytes for subsequent analysis. Column chromatography, because of its superiority over most other chromatographic techniques, is commonly used in analysis of organic pollutants and is taken here as the example illustrating the mechanism of chromatography. In column chromatography, the stationary phase is packed in a column. The sample is passed through the column with continuous addition of the mobile phase, and this process is called elution. The average rate at which an analyte moves through the column is equivalent to the time duration that it stays in the mobile phase. Therefore, the distribution of the analytes between the mobile and

308

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

stationary phases is dictated by the relative strength for affiliation of the analytes between the two phases. The time interval between sample injection and the appearance of the analyte concentration peak at the end (usually connected to a detector) of the column is defined as retention time (tR). To obtain optimal separation of a mixture, sharp, symmetric chromatographic peaks are desirable. The column efficiency, related to the efficiency of separation, can be evaluated by the theoretical plate model of chromatography (Moretti et al. 2006) in terms of the number of theoretical plates. Depending on the type of the mobile phase, column chromatography can be characterized as GC or high-performance liquid chromatography (HPLC).

.

12.5.1.1. Gas Chromatography. A GC system generally consist of a carrier gas, an injector, a column housing a liquid stationary phase adsorbed onto the surface of an inert solid, and a detector (Fig. 12.11). An aliquot of the extract is usually injected into the GC system via an injector port manually or with an autosampler. As the injector temperature is normally higher than the boiling points of the analytes, the injected undegraded extract evaporates into the injector chamber and is subsequently flushed into the GC column by a carrier gas (mobile phase) from a pressurized gas cylinder. Separation of the analytes occurs within the GC column via the same mechanisms as described for column chromatography. Finally, a detector connected to the end of the column collects the mass counts of the target analytes at various retention times and transmits the signals to a recorder that generates a series of peaks on a data system. The area of each peak is proportional to the amount of an analyte injected into the GC system. The following is a brief description of the key components of a typical GC system: .

.

Carrier Gas. An appropriate carrier gas must be chemically inert and thermally stable. Commonly used carrier gases are helium, hydrogen, and nitrogen; argon and

carbon dioxide are used occasionally. The choice of carrier gas is often dependent on the type of detector used. Injection Port. The injection port, typically containing a heated chamber with a glass liner, is where the extract is injected onto, vaporized, and flushed into the chromatographic column. The injection port temperature is usually set at 50 C higher than the boiling point of the least volatile analyte in the sample to achieve optimal peak resolution. The injector temperature may also be programmed as an option. Aside from on-column injection, two injection modes are commonly used: split and splitless/split. In the split mode, the sample vaporizes to form a mixture of carrier gas, vaporized solvent, and vaporized solutes. A portion of this mixture passes onto the column, but most exits through the split outlet. In the splitless/split mode, the split outlet is closed for a preset time interval starting from when sample is injected. This time interval is defined as split time that can be programmed with other chromatographic parameters. The split outlet is turned on after the split time and remains open until the next injection. On-column injection, on the other hand, delivers the sample directly onto the top of the column, which is maintained at low temperature during injection, and is particularly suited for large molecules. Column. Two types of columns, packed and capillary (also known as open tubular), have been used; the capillary column is currently the more popular one. A packed column contains a fine, inert, and solid support material (commonly based on diatomaceous earth) coated with a liquid stationary phase. Most packed columns are 1.5–10 m in length with an internal diameter of 2–4 mm. On the other hand, a typical capillary column has an internal diameter of a few tenths of a millimeter. Two types of capillary columns have been made: wall-coated open tubular or support-coated open tubular. A wall-coated column consists of a

Injector

Recorder

Column Carrier gas

Detector Column oven Figure 12.11. A schematic showing a typical gas chromatograph.

INSTRUMENTATION

.

capillary tube coated with a liquid stationary phase, whereas the inner wall of a support-coated column is lined with a thin layer of support material such as diatomaceous earth, onto which the stationary phase is adsorbed. Both capillary column types are more efficient than packed columns. Detector. Quite a few detector types can be used in GC systems. Each detector type may be selectively sensitive to a certain group of analytes. A nonselective detector responds to all compounds except the carrier gas, while a selective detector responds to a range of compounds with similar physical or chemical properties. Detectors can also be categorized as concentration or mass-flowdependent. The signal from a concentration-dependent detector is related to the concentration of a solute detected, and the sample is rarely destroyed. With this detector type, dilution with make-up gas lowers the response. Conversely, mass-flow-dependent detectors usually destroy the sample, and the signal is related to the rate at which solute molecules are detected. The types of detectors commonly used in GC systems include, but are not limited to, mass spectrometer (MS), flame ionization detector (FID), electron-capture detector (ECD), thermal conductivity detector (TCD), nitrogen–phosphorus detector (NPD), flame photometric detector (FPD), photoionization detector (PID), and Hall electrolytic conductivity detector (HECD). Among these detector types, MS, FID, and TCD are non-selective detectors, while ECD, NPD, FPD, PID, and HECD are selective detectors. In addition, TCD and PID are concentration dependent and non-destructive detectors whereas MS, FID, ECD, NPD FPD, and HECD are both mass-flow-dependent and destructive. Because of its continuously improved sensitivity and selectivity, MS

has become increasingly popular as a detector in chromatography. Additional technical specifics on MS are provided in Section 12.5.2. For the other detector types, readers can refer to the literature and other relevant books for more information.

12.5.1.2. Liquid Chromatography. High-performance (or -pressure) liquid chromatography is a form of column chromatography used frequently to separate, identify, and quantify compounds with relatively high molecular weights (with corresponding high boiling points) or high water solubility. A typical HPLC system contains a column housing a chromatographic packing material (stationary phase), a pump used to push the mobile phase through the column, and a detector recording the retention times and response intensities of the target analytes (Fig. 12.12). The column is deemed the most important component in a HPLC system, as it separates individual analytes from a mixture. In many cases, the analytical column is protected with a guard column that removes debris and adsorbs undesirable materials that otherwise may irreversibly bind and possibly change the stationary phase. Several types of analytical columns have been utilized in HPLC and can be classified as follow: .

Normal Phase. The retention of an analyte on a normal phase is achieved by the interaction of the polar components in the stationary phase and the analyte. Consequently, packing materials in the stationary phase must be more polar than the mobile phase with respect to the analytes under consideration. A commonly used stationary phase is coated silica, while typical mobile phases include hexane, methylene chloride, chloroform, diethyl ether, and various combinations of these solvents.

Vacuum Reservoirs Solvent 1

Sample

Solvent 2

injection port Pressure gauge Pressure gauge

Flow splitter Degasser

309

Degasser Analytical column

Pre-column Mixing Vessel

Detector

High pressure pump To waste

To waste or fraction collector

Figure 12.12. A schematic showing a typical liquid chromatograph.

310 .

.

.

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

Reverse Phase. The packing material of a reverse phase is relatively nonpolar, and the solvent is polar with respect to the analytes. In this case, typical stationary phases include nonpolar hydrocarbons, waxy liquids, or bonded hydrocarbons (C18, C8, etc.) and mobile phases are polar aqueous–organic mixtures such as methanol–water or acetonitrile–water. Size Exclusion. As indicated by its name, a size exclusion column separates target analytes according to their molecular sizes. Analytes of small sizes are able to penetrate into the pores within the packing material, while larger analytes can only partially access the pores. As a result, large analytes elute before smaller analytes. Ion Exchange. The stationary phase of an ion exchange column contains charged groups that react with those with opposite charge on the analytes, and the analytes are separated according to the strength of this charge interaction. A typical mobile phase is water containing salts or small amounts of alcohol, while a stationary phase contains either acidic or basic fixed sites.

An analyte in the column effluent is detected as a change in refractive index, UV–visible absorption at a set wavelength, fluorescence after excitation with a suitable wavelength, or electrochemical response. A mass spectrometer can also be interfaced with a LC to construct a LC-MS system, which is a powerful tool for qualitative and quantitative analyses. 12.5.2. Mass Spectrometry Techniques A mass spectrometer uses the difference in mass:charge ratio (m/z) of ionized analytes (atoms or molecules) to separate a mixture. The schematic diagram in Figure 12.13 depicts a typical MS configuration with the main components, which may vary with needs, including an ion source, a massselective analyzer, and an ion detector. High vacuum inside the MS system is required to allow ions to travel from the ion Repellor electrode

source to the detector without any obstruction from air molecules, which justifies the need to monitor the air : water ratio in the vacuum system in instrument tuning. When a sample is introduced into the ionization source, the sample molecules are ionized and the ions are extracted into the analyzer, where they are separated according to their m/z values. A detector at the end of the analyzer area counts the number of ions passing by at various times and transmits the recorded data to a data system. Usually, ionization can be conducted in different modes, resulting in various MS applications. Numerous ionization modes have been developed and used in MS, such as electron ionization, chemical ionization, atmospheric-pressure chemical ionization, atmospheric-pressure photoionization, field ionization, electrospray ionization, and matrix-assisted laser desorption ionization, fast-atom bombardment, ion bombardment or secondary-ion mass spectrometry, plasma desorption, and field desorption. Detailed descriptions of these ionization modes can be found in many standard textbooks. Depending on the mechanisms of sample introduction, ionization, and ion selection and detection, a number of MS systems can be classified into the categories discussed below. 12.5.2.1. Quadrupole Mass Spectrometer. A typical quadrupole mass analyzer (Fig. 12.14a), a quite popular mass analyzer used in MS, consists of four circular metal rods, which are set parallel to each other. Sample ions are filtered and separated according to their m/z values and the stability of their trajectories in the oscillating electric fields applied to the rods. Each opposing rod pair is connected together electrically, and a radiofrequency (RF) voltage is applied between one pair of rods and the other. A direct-current voltage is then superimposed on the RF voltage. Ions travel down the quadrupole between the rods, and only the ions with a specific m/z value can reach the detector at a given ratio of voltages, whereas other ions with unstable trajectories will collide with the rods. This arrangement allows selection of an ion with particular m/z, or scanning ions with a range of m/z values by continuously tuning the RF voltage.

Accelerating slits Filament

Magnetic Unfocused heavy ions field

Sample

Detector Ion source

Unfocused light ions

Focused ions beam

Figure 12.13. A schematic showing a typical mass spectrometer.

311

O2 He Ref.

El source

Isotope Ratio MS

Faraday cups

(d)

Ref.

IRMS

(a)

44 45 46 detection system

magnetic sector

Endcap

Ion source Ions in Ions out Ions detector

Endcap

(e)

(b)

Ring

ro

Zo

Input

(f)

Output

Beam splitter

Mirror

(c)

d/2

Moving mirror

Partially transmissive mirror

Vacuum pump

Detector To mass conversion

Pulsed laser beam

Acceleration lenses

Samples

Figure 12.14. Schematic diagrams showing typical mass analyzers: (a) quadrupole mass analyzer; (b) ion trap mass spectrometer; (c) time-offlight spectrometer: (d) isotope ratio mass spectrometer; (e) magnetic mass spectrometer; (f) Fourier transform mass spectrometer.

GC

LN2

nafion dryer

reduction reactor Cu 600ºC

Backflush (remove solvent)

oxidation reactor NiO, CuO, Pt/940ºC

Ionizer

Ion beam

RF+beam RF-beam

Mass separated ion beam

Faraday cup

312

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

12.5.2.2. Ion Trap Mass Spectrometry. The quadrupole ion trap is the three-dimensional analog of a linear quadrupole mass analyzer. In an ion trap, sample ions are subject to forces exerted by a RF field that occurs in all three dimensions. The ion trap itself generally consists of two hyperbolic metal electrodes with their foci facing each other and a hyperbolic ring electrode halfway between the other two electrodes (Fig. 12.14b). The sample ions are trapped in the space between these three electrodes by alternating-current (AC; oscillating, nonstatic) and direct current (nonoscillating, static) electric fields. The AC RF voltage oscillates between the two hyperbolic metal endcap electrodes if ion excitation is desired; the driving AC voltage is applied to the ring electrode. The ions are first pulled up and down axially while being pushed in radially The ions are then pulled out radially and pushed in axially (from the top and bottom). As a result, the ions move in a complex motion that generally involves the cloud of ions being long and narrow and then short and wide, back and forth, oscillating between the two states. Because an ion trap analyzer can generate secondary ions on trapping and releasing primary-ion fragments within the ion trap, an ion trap MS can be regarded as an tandem MS/MS system. 12.5.2.3. Time-of-Flight Spectrometry. The time-of-flight analyzer (Fig. 12.14c) separates ions on the basis of their velocity difference, that is, the difference in the time required to traverse the length (L) of the flight tube from the ionization source to the ion detector. As the ions are accelerated from the ionization source, they possess the same kinetic energy from the potential difference applied. Equal kinetic energy ions arrive at the drift tube at the same time and are further accelerated with a drift velocity, which makes arrival time at the detector dependent on ion mass. Mathematically, the flight time (t) is related to the ion mass (m) and charge (z) by t ¼ L(m/2zV)1/2, where V is the potential of the electric field applied to the flight tube. As a result, the lighter ions travel faster than the heavier ones, and measuring the arrival time of the ions determines their masses. 12.5.2.4. Isotope Ratio Mass Spectrometry. Isotope ratio mass spectrometry (IRMS) has been around for a while and is probably ranked among the oldest types of MS used in analytical research (Newman 1996). However, only until recently has IRMS received the attention from other areas of applied analytical chemistry when IRMS instruments coupled to a GC via a combustion interface (Fig. 12.14d) in 1990 were made commercially available. Isotope ratio MS achieves highly precise measurements of isotopic abundances at the expense of the flexibility of scanning MS. Prior to isotope ratio measurements, the target analyte must be converted into a simple gas, isotopically representative of the original sample, before entering the ion source. Consequently, continuous-flow isotope ratio measurements of 2 H=1 H, 13 C=12 C, and 18 O=16 O are performed

on H2, CO2, and CO, respectively, with 13 C abundance measurements accounting for almost 70% of all gas isotope ratio analyses made (Meier-Augenstein, 2002). It is also important to note that IRMS, in fact, determines differences in isotope ratios with great precision and accuracy rather than the absolute isotope ratios. Isotope ratio MS measurements yield the relative isotopic abundance of the analyte gas, calibrated relative to the measured isotope ratio of a standard or reference gas. To achieve accurate and highly precise and reproducible measurements of isotope ratios, great care must be taken to ensure that no part of the analyte data is lost. In the case of CO2, the data comprise three ion traces for the different isotopomers 12 C 16 O2 , 13 C 16 O2 , and 12 C 18 O 16 O with their corresponding masses at m/z 44, 45, and 46, respectively. The three ion beams are registered simultaneously by a multiple Faraday cup arrangement with a dedicated Faraday cup for each isotopomer (Fig. 12.14d). The resulting ion currents are continuously monitored and subsequently digitized and the resulting peak data (points) transferred to the host computer. Here, the peak area for each isotopomer is integrated quantitatively, and the corresponding ratios are calculated. Since the small variations of the heavier isotope habitually measured by IRMS range from 0.07 to þ 1.09 APE (atom percentage excess), the d notation in per mil units (‰) has been adopted to report changes in isotopic abundance as a per mil deviation compared to a designated isotopic standard: ds ‰ðper milÞ ¼ 1000


Rs Rstd Rstd

ð12:12Þ

where Rs is the measured isotope ratio for the sample and Rstd is the measured isotope ratio for the standard. To give a convenient rule-of-thumb approximation in the d notation, a 13 C enrichment in the range of 0.033 to þ 0.0549 APE corresponds to a d13C value range of 30‰ to þ 50‰. 12.5.2.5. Magnetic Sector Mass Spectrometry. Magnetic sector mass spectrometry is one of the most common types of mass analyzer (the first mass spectrometers made in the 1950s were the magnetic sector types). It consists of a large electromagnet and some electrostatic focusing devices (Fig. 12.14e). Ions enter the instrument from the source where they are initially focused, and enter the magnetic sector through the source slit where they are deflected according to the left-hand rule. Higher-mass ions are deflected less than are lower-mass ions. Scanning the magnet enables ions of different masses to be focused on the monitor slit. At this stage, the ions have been separated only by their masses. To obtain a spectrum of good resolution, where all ions with the same m/z appear coincident as one peak in the spectrum, ions have to be filtered by their kinetic energies. After another stage of focusing, the ions enter the electrostatic sector where the energy distributions of ions of the

REFERENCES

same m/z are corrected for and are focused at the doublefocusing point on the detector slit. In general, magnetic sector mass spectrometry can provide considerably higher mass resolution than can quadrupole MS. One area of applications of magnetic sector mass spectrometry is where high mass resolution power is demanded, such as in isotope ratio mass spectrometry, described above. 12.5.2.6. Fourier Transform Mass Spectrometry. This is a mass analyzer for determining the m/z values based on the image current derived from ion cyclotroning in a fixed magnetic field (Fig. 12.14f). The ions are injected into a Penning trap (a magnetic field with electric trapping plates) where they are efficiently excited to a larger cyclotron circuit by an oscillating electric field perpendicular to the magnetic field. The electrical signal of ions will be measured by a detector at a fixed position, and a periodic signal will be recorded. Since the frequency of an ion’s cycling is determined by its m/z ratio, the resulting signal is a free induction decay, transient, or interferogram consisting of the superposition of a sine wave. The useful signal is extracted from this information by performing a Fourier transform to provide a mass spectrum.

12.6. CONCLUSIONS This chapter has described the main features of sampling, sample extraction, and instrumental analysis commonly involved in the determination of anthropogenic organic compounds (mainly persistent organic pollutants) in environmental samples. Sampling techniques reviewed include those often used in collection of air, soil, and sediment samples, which can also be divided into active and passive sampling approaches. Sample extraction is the focal point of this chapter, as it is a critical part of environmental analyses. Furthermore, sample extraction methods have been considerably comparable among different analytical laboratories, and can be easily integrated for large-scale monitoring programs. Therefore, each extraction method reviewed was accompanied by a detailed working protocol. On the other hand, instrumental procedures can be substantially variable from application to application; as a result, only general descriptions of the most widely used instrument types have been presented. Apparently, it is impossible to cover every aspect of environmental sample collection and processing with a short chapter like this, which should be regarded as a snapshot of the collection of environmental analytical methods, rather than a comprehensive practical handbook. Readers are encouraged to use more specific, step-by-step working manuals in actual situations. Fortunately, these practical manuals are readily available in the literature and can be easily located with a few clicks of a mouse or on a computer keyboard.

313

ACKNOWLEDGMENTS Financial support from the National Natural Science Foundation of China (Grants 40821003 and 40588001) and the Earmarked Fund of the State Key Laboratory of Organic Geochemistry (SKLOG2008A05) are gratefully appreciated.

REFERENCES American Society for Testing and Materials (2009), ASTM D158608a: Standard Test Method for Standard Penetration Test (SPT) and Split-Barrel Sampling of Soils; available at http://astm.nufu. eu/std/ASTM%20D1586%20-%2008a (accessed Jan. 2009). Arthur, C. L. and Pawliszyn, J. (1990), Solid phase microextraction with thermal desorption using fused silica optical fibers, Anal. Chem. 62, 2145–2148. Axelman, J., Naes, K., Naf, C., and Broman, D. (1999), Accumulation of polycyclic aromatic hydrocarbons in semipermeable membrane devices and caged mussels (mytilus edulis L.) in relation to water column phase distribution, Environ. Toxicol. Chem. 18, 2454–2461. Bergqvist, P. A., Strandberg, B., Ekelund, R., Rappe, C., and Granmo, A. (1998), Temporal monitoring of organochlorine compounds in seawater by semipermeable membranes following a flooding episode in western Europe, Environ. Sci. Technol. 32, 3887–3892. Booij, K., Hoedemaker, J. R., and Bakker, J. F. (2003), Dissolved PCBs, PAHs, and HCB in pore waters and overlying waters of contaminated harbor sediments, Environ. Sci. Technol. 37, 4213–4220. Carls, M. G., Holland, L. G., Short, J. W., Heintz, R. A., and Rice, S. D. (2004), Monitoring polynuclear aromatic hydrocarbons in aqueous environments with passive low-density polyethylene membrane devices, Environ. Toxicol. Chem. 23, 1416–1424. Conder, J. M., La Point, T. W., Lotufo, G. R., and Steevens, J. A. (2003), Nondestructive, minimal-disturbance, direct-burial solid-phase microextraction fiber technique for measuring TNT in sediment, Environ. Sci. Technol. 37, 1625–1632. Gordon, C. S. and Lowe, J. T. (1927), Carbon-Monoxide Detector, U.S. Patent 1,644,014. Harper, M. (2000), Sorbent trapping of volatile organic compounds from air, J. Chromatogr. A 885, 129–151. Herve, S., Prest, H. F., Heinonen, P., Hyotylainen, T., Koistinen, J., and Paasivirta, J. (1995), Lipid-filled semipermeable membrane devices and mussels as samples of organochlorine compounds in lake water, Environ. Sci. Pollut. Res. 2, 24–30. Hewitt, A. D. (1995a), Chemical Preservation of Volatile Organic Compounds in Soil Subsamples, A578292, USA Cold Regions Research and Engineering Laboratory, U.S. Army Corps of Engineers. Hewitt, A. D. (1995b), Enhanced Preservation of Volatile Organic Compounds in Soil with Sodium Bisulfate, A039203. USA Cold Regions Research and Engineering Laboratory, U.S. Army Corps of Engineers.

314

PRINCIPLES AND GUIDELINES OF SAMPLING, EXTRACTION, AND INSTRUMENTAL ANALYSIS TECHNIQUES

Hewitt, A. D. (1999), Storage and Preservation of Soil Samples for Volatile Compound Analysis, A106363, USA Cold Regions Research and Engineering Laboratory, U.S. Army Corps of Engineers. Huckins, J. N., Tubergen, M. W., and Manuweera, G. K. (1990), Semipermeable membrane devices containing model lipid: A new approach to monitoring the bioavailability of lipophilic contaminants and estimating their bioconcentration potential, Chemosphere 20, 533–552. Kelly, T. J., Mukund, R., Gordon, S. M., and Hays, M. J. (1994), Ambient Measurement Methods and Properties of the 189 Clean Air Act Hazardous Air Pollutants, Final Report, PB-95-123923/ XAB, U.S. Environmental Protection Agency (USEPA), Research Triangle Park, NC. Kimbrough, D. E. and Wakakuwa, J. (1993), Method detection limits in solid waste analysis, Environ. Sci. Technol. 27, 2692–2699. King, A. J., Readman, J. W., and Zhou, J. L. (2004), Determination of polycyclic aromatic hydrocarbons in water by solid-phase microextraction-gas chromatography-mass spectrometry, Anal. Chim. Acta 523, 259–267. Koester, C. J. (2005), Trends in environmental analysis, Anal. Chem. 77, 3737–3754. Liu, M. M., Zeng, Z. R., Wang, C. L., Tan, Y. J., and Liu, H. (2003), Solid-phase microextraction of phosphate and methylphosphonate using novel fibers coated with a sol-gel-derived silicone divinyl benzene co-polymer, Chromatographia 58, 597–605. Lutermann, C., Dott, W., and Hollender, J. (1998), Combined modifier/in situ derivatization effects on supercritical fluid extraction of polycyclic aromatic hydrocarbons from soil, J. Chromatogr. A 811, 151–156. Mayer, P., Vaes, W. H. J., Wijnker, F., Legierse, K. C. H. M., Kraaj, R., Tolls, J., and Hermens, J. L. M. (2000), Sensing dissolved sediment porewater concetrations of persistent and bioaccumulative pollutants using disposable solid-phase microextraction fibers, Environ. Sci. Technol. 34, 5177–5183. Meier-Augenstein, W. (2002), Stable isotope analysis of fatty acids by gas chromatography-isotope ratio mass spectrometry, Anal. Chim. Acta 465, 63–79. Moretti, P., Vezzani, S., and Castello, G. (2006), Prediction of theoretical plate number in isothermal gas chromatographic analysis on capillary columns, J. Chromatogr. A 1133, 305–314. Newman, A. (1996), EPA plans new approach to analytical methods, Anal. Chem. 68, A733–A737. Ni, H.-G., Lu, F.-H., Luo, X.-L., Tian, Y.-H., Wang, J.-Z., Guan, Y.-F., Chen, S.-J., Luo, X.-J., and Zeng, E. Y. (2008), Assessment of sampling designs to measure riverine fluxes from the Pearl River Delta, China to the South China Sea, Environ. Monit. Assess. 143, 291–301.

Ouyang, G. and Pawliszyn, J. (2006), SPME in environmental analysis, Anal. Bioanal. Chem. 386, 1059–1073. Pawliszyn, J. (1997), Solid Phase Microextraction: Theory and Practice, Wiley-VHC, New York. Pawliszyn, J. (1999), Applications of Solid Phase Microextraction, Royal Society of Chemistry, Cornwall, UK. Pitzer, K. S. (1995), Thermodynamics, 3rd ed., McGraw-Hill, New York. Popek, E. P. (2003), Sampling and Analysis of Environmental Chemical Pollutants: A Complete Guide. Academic Press, San Diego. Resources Information Standards Committee (1997), Lake and Stream Sediment Bottom Sampling Manual; available at http:// ilmbwww.gov.bc.ca/risc/pubs/aquatic/lake-stream/index.htm (accessed Sept. 2009). Richter, B. E., Jones, B. A., Ezzell, J. L., Porter, N. L., Avdalovic, N., and Pohl, C. (1996), Accelerated solvent extraction: A technique for sample preparation, Anal. Chem. 68, 1033–1039. Robbins, L. A. (1980), Liquid-liquid extraction: a pretreatment process for wastewater, Chem. Eng. Proc. 76, 58–61. Sijm, D., Kraaij, R., and Belfroid, A. (2000), Bioavailability in soil or sediment: Exposure of differerent organisms and approaches to study it, Environ. Pollut. 108, 113–119. Tong, P. and Imagawa, T. (1995), Optimization of supercritical fluid extraction for polychlorinated biphenyls from sediments, Anal. Chim. Acta 310, 93–100. U.S. Environmental Protection Agency (USEPA) (2005), USEPA Region 9, Technical Guidelines for Accurately Determining Volatile Organic Compound (VOC) Concentrations in Soil and Solid Matrices, R9QA/05.2, USEPA Region 9, San Francisco. Xiong, G., Chen, Y., and Pawliszyn, J. (2003), On-site calibration method based on stepwise solid-phase microextraction, J. Chromatogr. A 999, 43–50. Zeng, E. Y. and Noblet, J. A. (2002), Theoretical considerations on the use of solid-phase microextraction with complex environmental samples, Environ. Sci. Technol. 36, 3385–3392. Zeng, E. Y., Tsukada, D., and Diehl, D. W. (2004), Development of a solid-phase microextraction-based method for sampling of persistent chlorinated hydrocarbons in an urbanized coastal environment, Environ. Sci. Technol. 38, 5737–5743. Zeng, E. Y., Tsukada, D., Diehl, D. W., Peng, J., Schiff, K., Noblet, J. A., and Maruya, K. A. (2005), Distribution and mass inventory of total dichlorodiphenyldichloroethylene in the water column of the Southern California Bight, Environ. Sci. Technol. 39, 8170–8176. Zlotorzynski, A. (1995), The application of microwave-radiation to analytical and environmental chemistry, Crit. Rev. Anal. Chem. 25, 43–76.

13 NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS ROBERT L. COOK 13.1. Introduction 13.2. Background 13.2.1. Soil 13.2.2. Soil Degradation and Chemical Farming 13.2.3. The Guiding Hand of Mass Balancing Batch-Based Sorption Studies 13.3. Basics of the NMR Toolbox 13.3.1. Signal Intensity 13.3.2. Chemical Shift 13.3.3. Spin Relaxation 13.3.4. Relaxation Mechansims 13.4. A Review of NMR as Applied to the Association Between Anthropogenic Organic Compounds (AOCs) and Natural Organic Matter (NOM) 13.4.1. 1H (Protons) 13.4.2. 13C (Carbon) 13.4.3. 15N (Nitrogen) 13.4.4. 19F (Fluorine) 13.4.5. 2H (Deuterium) 13.5. A Highly Critical Evaluation of NMR as Applied to AOC Association with NOM 13.5.1. Environmentally Relevant Conditions 13.5.2. “Real World” Samples 13.5.3. Molecular Level Information 13.5.4. Thermodynamic and Kinetic Data 13.6. Conclusions, Final Thoughts, and Paths Forward

13.1. INTRODUCTION As a global collective, the human ecological footprint has been increasing almost linearly from 1960 until the present

(WWF 2008). In raw numbers, since 1960 our demand on this planet has more than doubled, to the point where we are withdrawing 30% more than planet Earth can sustainably deliver. If it remains business as usual, then by the year 2050 it is projected that we will require two planet Earths to sustain us. These numbers are rather shocking, to say the least, and result from of a number of forces, which include global population growth and increasing individual consumption. This overdraft (unsustainable activity) is leading to a degradation of the Earth’s ecosystems and the services that they provide. Soils are a major example of this and, as such, are a major focus of this chapter. Our ability to overdraft soils is a result of the use of anthropogenic chemicals, such as fertilizers and pesticides. However, the same practices degrade the soil as well as other resources in the surrounding and downstream ecosystems. Pesticides and other anthropogenic organic compounds (AOCs) are usually hydrophobic in nature, and this property means that they sorb to soils, leading to a range of environmental issues collectively termed as “fate and transport,” including bioavailability. Although the understanding of AOC sorption to soil is essential to our understanding and modeling of the fate and transport of these very important chemicals, it has been greatly wanting. This is especially true in regard to a molecular level understanding, as illustrated by the advances in the field of biochemistry. Nuclear magnetic resonance (NMR) has been one of the most prominent techniques that allowed our current molecular level insight into the workings of complex biological systems. Because of the complex and amorphous nature of soils, NMR is viewed as a highly promising method to yield molecular level understanding of the mechanisms that determine the fate and transport of AOCs within the environment. This leads us to the aims of this chapter, which are to provide the

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

315

316

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

reader with (1) a background understanding of the situation (Section 13.2) as well as the applicable basic NMR theory (Section 13.3), (2) a focused review on the application of NMR to AOCs from an environmental perspective (Section 13.4), and (3) a critical analysis of (a) the current state of NMR as applied to AOCs and (b) the insights that the NMR approach has provided to date (Section 13.5).

13.2. BACKGROUND 13.2.1. Soil This section very briefly introduces the reader to the central role that soils play in our lives, including a discussion of their complexity and the analytical challenge that this complexity represents. Soil (the pedosphere) links all major spheres of our planet, from the lithosphere (Earth’s surface) to the hydrosphere (water) to the atmosphere (air) as well as to the biosphere (living entities). Life and soil have always been entangled and, in fact, have evolved symbiotically. This evolution has taken place from the initial breakdown of the Earth’s crust into clays and then the coexistence of bacteria and clays to create primitive soils. This primitive combination began to support higher life in the form of plants. Respiration from these soil biota increased the Earth’s CO2 levels, allowing for further breakdown of the mineral phases, and hence, further evolution of soil and associated biota, which, in turn, led to more coevolution and the formation of soil organic matter, and finally, to the formation of the modern soils that we now simply call soil. Although much maligned, as evidenced by the commonly used words—such as soiled and dirty—soil is essential to life and should be viewed as a part of the ecosystem of utmost importance. Soil plays a multitude of services on planet Earth, including (1) food and other biomass production; (2) storing, filtering, and transformation; (3) habitat and gene pool; (4) physical and cultural environment for humankind; and (5) source of raw materials. Consequently, soil is essential to life, but it is a limited and nonrenewable resource that is much more difficult to recover once damaged compared to air and water (Van-Camp et al. 2004; UNEP 2007). Of particular interest to this work is soil contamination and damage caused by AOCs as well as the underlying sorption process affecting the fate and transport of these compounds within the environment. A molecular level understanding at the atomic scale of the sorption process is essential to our stewardship role in soil management, sustainability, and recovery. Opposing this is the complex nature of soil itself. This complexity means that analytical methods developed in other areas of scientific investigation, such as biomedical, polymer, and materials, may not be directly applicable to soils, and thus, great caution is required in their application and modification, and the interpretation of the yielded data.

The complexity of a soil is multilevel (and multiscale, from nanometers to hectares). If one breaks a soil down into its simplest component, then one indentifies air, water (about 70% of the Earth’s surface freshwater is held in soils), and solids. Another—and possibly more applicable for this work—approach identifies biological, organic, and inorganic constituents of soil. The biological component consists of all the living biota within the soil, the organic constituents include all dead and decomposed biological materials, while the inorganic components include all the mineral phases. Traditionally, when evaluating the sorption behavior of AOCs to soils, the biological component is removed by HgCl2 or NaN3, and thus, for the sake of brevity this will be assumed to be the case for the remainder of the discussion. This, then, leaves the organic and inorganic constituents. The inorganic constituents are usually in the form of minerals, including clays, which, in turn, means that their constituent elements are in a highly ordered environment. From an analytical perspective, this is highly desired and has allowed for a detailed characterization and study of this soil constituent. On the other hand, the organic constituent, known as soil organic matter (SOM), is a complex mixture with an ill-defined makeup, thus making it highly problematic for analytical characterization. The complex heterogeneous polydisperse nature of SOM means that all forms of long-range order are absent, and hence, the vast majority of molecular level analytical techniques are not amenable to its characterization. In addition, SOM can also be, and usually is, associated with the mineral phases within a natural soil. This adds another layer of complexity to the associations of SOM and what this does to the assemblage of the molecular moieties within the SOM. This association of SOM with mineral interfaces complicates the nuclear magnetic resonance (NMR) characterization of SOM on at least two accounts. The first is that minerals dilute the SOM, causing lower signal to noise ratios when characterizing SOM “in situ.” The second, and more important in terms of analytical characterization, is that paramagnetic centers within the mineral phases (e.g., Fe and Mn) can render nearby SOM invisible and thus warp one’s viewpoint of the SOM within a soil when carrying out in situ characterization. For brevity, the reader is referred to a review that addresses these issues directly (Cook 2004). In fact, a number of very good references on the application of NMR for the characterization of natural organic matter—including SOM—are available (Conte et al. 2004; Cardoza et al. 2004; Mopper et al. 2007). 13.2.2. Soil Degradation and Chemical Farming Since the 1980s the growth of land converted to cropland has decreased; however, the yields per hectare have increased from 1.8 to 2.5 metric tons. This has been achieved by a combination of both better farming practices and industrialized farming practices. Industrial farming practices include

BACKGROUND

the use of anthropogenic compounds, which increase growth (i.e., fertilizers), remove competition (i.e., herbicides), and control insects (i.e., pesticides; from here on, this term encompasses herbicides as well), and thus increase yield, albeit at the expense of the ecosystem. The use of pesticides may increase short-term crop yields, however, in the long term, it may cause quite an opposite result. An example of this is that different plants require different nutrients; thus the dominance of one plant species will lead to a much larger demand on a certain set of nutrients from the soil. The loss of competition also removes biodiversity from the ecosystem, and hence, can affect the naturally occurring rejuvenation of soils by destroying the relationship between the biota, organic components, and inorganic components—not to mention the effects of the mechanization of farming and its impact on soil quality via tilling and compaction, emission of greenhouse gases, spills of associated chemicals, the use of nonnative plants cultivated for mechanical farming, and the damage to the surrounding and downstream ecosystem. The points above emphasize the role of soil and land management in the next decade or two as it is anticipated that food demand will close to triple compared to that of 1997 (Penning de Vries 1997). This demand for food has already led to land change, which, in turn, has led to chemical contamination, soil erosion, nutrient depletion, water scarcity, and salinization. All these factors are major issues in terms of the stewardship of our planet, the human population, and the survival of this and future generations, especially when compounded with further soil losses due to urban expansion and climate change (UNEP 2007). In other words, soil is essential to our survival and we are living on borrowed resources. This means that our future stewardship must be based on understanding of soil processes, including those involving AOCs. 13.2.3. The Guiding Hand of Mass Balancing Batch-Based Sorption Studies Nuclear magnetic resonance holds a great deal of promise in gaining insight into how AOCs behave in the environment; however, the required experiments are very time- and resource-intensive. This means that it is beneficial to view NMR studies as a component of an overall approach, ranging from field-level studies all the way down to molecular level studies by NMR and other analytical techniques. Field-level studies are time-, labor- and resource-intensive, and thus, are usually modeled in the laboratory by a range of batch-type experiments. For the purpose of this discussion, a batch-type experiment will consist of experiments in which small aliquots of soil are placed in contact with a known amount of AOCs and, after determination of the soil: solute ratio as well as applicable kinetics, (typically) fitted by a sorption isotherm. The most popular fitting model is the Freundlich

317

isotherm as it does not assume homogeneity of sorption sites or monolayer coverage, as neither of these can be assumed for a system as complex and heterogeneous as soil. Traditionally, these experiments have been carried out in a mass balance approach, in which the amount of AOC sorbed is determined from the amount remaining in solution (and the amount added). Because of this mass balance approach, only bulk data—illustrative of macroscopic behavior observed in the field—are revealed by batch sorption experiments; however, the bulk data lack the molecular insight required for detailed chemical understanding, and hence, modeling. In view of these types of experiments, it was proposed initially that the sorption of AOCs to soils could be described in terms of simple linear partitioning models (Chiou et al. 1979,1983,1998; Karickhoff et al. 1979; Schwartzenbach and Westall 1981). This was then extended to an even simpler model based on the AOCs octanol–water partition coefficients Kow (Karickhoff ; Seth et al. 1999). Advances in analytical methods, especially in terms of detection level, led to lower and more environmentally relevant concentrations of sorbate (i.e., AOC) to be studied and to the observation of nonlinear sorption isotherms (Weber and Huang 1996, 1999; Huang et al. 1997; Xing et al. 1996; Xing 1997; Xing and Pignatello 1997; Luthy et al. 1997; Xing and Chen 1999; Chiou and Kile 1998; Chiou et al. 2000; Kang and Xing 2005). This nonlinearity may possibly be due to high-surface-area carbonaceous materials, such as charcoal and soot, collectively called black carbon (R€ugner et al. 1999; Accardi-Dey and Gschwend 2002; Kleineidam et al. 2002; Karapanagioti et al. 2003). Although plausible in some cases, this explanation does not hold for all cases in which a nonlinear isotherm has been observed, such as (1) humic substances and kerogen— playing a dominant role in nonlinear sorption by a peat sample (Ran et al. 2002), (2) other char-free NOM samples (Gunasekara and Xing 2003; Wang and Xing 2005a,b Pignatello et al. 2006a) and (3) dissolved humic acids (Laor and Rebhun 2002; Pan et al. 2007a,b). In addition, NOM will block or compete for AOCs, especially hydrophobic organic compounds (HOCs), sorption sites in black carbon (Matsui et al. 2002a,b; Pignatello et al. 2006b). A second explanation (and model) stemming from the observed nonlinearity comes from the polymer literature and consists of both “loose” (soft, mobile, or rubbery) and “condensed” (hard, immobile or glassy) sorption domains within the organic matter (Weber and Huang 1996; Xing et al. 1996; Huang et al. 1997; Huang and Weber 1997; Luthy et al. 1997; Xing 1997; Xing and Pignatello 1997). In this explanation, the “loose” domains are where partitioning (linear isotherm) occurs, while the “condensed” domains are where the nonlinear and competitive sorption takes place. The discovery of a phase transition for a number of NOM samples by both differential scanning calorimetry (DSC) techniques (LeBoeuf and Weber 2000a,b; Young and LeBoeuf 2000;

318

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

Zhang et al. 2007) and solid-state NMR (Hu et al. 2000; Gunasekara et al. 2003) add further support to the loose/ condensed-phase model. This model does raise an important question as to the role played by the NOM assemblage within soil in the sorption of AOCs (Lattao et al. 2008). 13.2.3.1. Molecular Assemblage of SOM Within a Soil. One contentious issue in the area of AOC sorption to soils is the role of interfaces, more specifically (1) the NOM–aqueous interface, (2) the NOM–particle interface, and (3) the “in between” NOM phase. This interface model can be combined with the loose/condensed-phase model by considering hydrophilic/hydrophobic and energy minimization principles. Simply stated, hydrophobic moieties within NOM wish to be as far away from the aqueous interface, while the opposite is true of the hydrophilic moieties. This, in turn, means that the hydrophobic moieties have much less room to occupy, and hence, occur in a highly condensed state. The hydrophilic moieties, having a large amount of wet space, and are much looser in the packing and experience dynamic hydrogen bonding with water. The space occupied by the interchangeable water enhances this looseness. The gradient between hydrophobic and hydrophilic moieties also correlates well with a gradient between condensed and loose phases. It is composed of moieties having both hydrophobic and hydrophilic character, and the balance between these two would dictate how condensed (or loose) the interface region is. The more recent paradigm shift from polymers to weak assemblies (WAS) in how NOM is viewed, suggests that these interfaces are dynamic and dependent on the conditions under which they are probed, and so is the condensed and loose character of soil-associated NOM. In fact, from the WAS perspective, a very similar three-layer model has been proposed by Kleber et al. (2007). It should be noted that, in absolute terms, for systems as complex and heterogeneous as NOM, one expects a continuum between the condensed and loose phases as well as between the hydrophobic and hydrophilic nature of the NOM. This continuum is in a state of perpetual dynamic flux. The dynamic flux—even when in the form of in situ SOM—can be seen in more recent work on nonwetting soils. This work shows clearly that nonwetting soils will slowly wet over the course of days to weeks. This slow kinetic process has been explained by a rearrangement of the SOM so that there is a relocation of the hydrophobic moieties from the surface to the interior, and vice versa, for the hydrophilic moieties as the soil wets (Ma’shum and Farmer 1985; Valat et al. 1991; Roy et al. 2000; Todoruk et al. 2003; Hurrass and Schaumann 2006; Diehl and Schaumann 2007; Bayer and Schaumann 2007). This work and that on the rearrangement of the in situ SOM is also in line with the findings that the sorption kinetics of two pesticides— which are substantially different in terms of their chemical makeup and properties—followed the uptake (wetting) of

water (Belliveau et al., 2000). Thus, great care must be taken while designing the introduction step of AOCs to any form of NOM, whether this is dissolved organic matter, isolated SOM, or in situ SOM. Some AOC sorption experiments have also revealed that the assemblage of NOM is an important parameter in determining the sorption behavior of the NOM. An elegant study on whole soils showed that sorption of different SOM fractions did not equal that of the whole soil. In fact, it was found that the sum was greater than the whole (Salloum et al. 2001). This work has been further supported by the extraction work of Wang and Xing (2005a, b), who showed that different humin extraction methods yielded humin fractions that “varied markedly” in their sorption capacity. This variation in sorption capacity was linked to both the modifications in the chemical makeup of the fractions and the physical conformation of the natural organic matter. Model systems have allowed even deeper insight into the role of NOM conformation in the sorption process, as illustrated in the work of Xing and co-workers (Chen and Xing 2005; Wang et al. 2007; Wang and Xing 2007). It should be stated that SOM’s chemical composition will strongly influence its sorption capacity for AOCs as well. 13.2.3.2. Chemical Composition of Sorption Domains. So far, the discussion has focused mainly on the physical state of organic matter; however, from a more chemical perspective, conformation may be a more appropriate word than state. Also, in dictating the observed AOC sorption/desorption behavior of soils, of at least equal importance to conformation is the chemical makeup of the SOM sorption domains. The scientific community has known this for quite some time, which has resulted in the creation of an area of intensive research. The many approaches applied include 1. Characterization of the sorbing SOM, and then correlatation of the observed sorption behavior to a select characterizable parameter. These studies correlated the amount of SOM present in the soil samples and their sorption capacity for AOCs (Karickhoff et al. 1979); however, little chemical insight was yielded. 2. Relating sorption to bulk or macroscopic characteristics or properties of a sorbent. This angle led to a correlation between sorption capacity of AOCs and oxygen content (Grathwohl 1990) or polarity indices of the SOM (Rutherford et al. 1992; Xing et al. 1994a), but this is not without controversy, as Kile and coworkers (Kile et al. 1999) found a very poor correlation between polarity indices and sorption behavior for 28 samples. 3. Chemical insight into sorption/desorption, especially by solid-state NMR and mainly via the 13 C-detected cross-polarization (CP) magic-angle spinning (MAS). From this type of analysis many groups have found

BASICS OF THE NMR TOOLBOX

convincing evidence that aromatic content strongly correlates with sorption capacity for AOCs (Xing et al. 1994b; Chen et al. 1996; Chin et al. 1997; Xing 1997; Chiou et al. 1998; Huang and Weber 1997; Perminova et al. 1999; Ahmad et al. 2001). More recently, it has been suggested that aliphatic domains play a significant role in AOC sorption (Chefetz et al. 2000; Mao et al. 2002; Salloum et al. 2002; Chen and Xing 2005; Chefetz 2007). The same conflicting data are also found for the role of aromatic and aliphatic domains in regard to nonlinear and competitive sorption (Chefetz et al. 2000; Xing 2001; Mao et al. 2002; Gunasekara and Xing 2003; Gunasekara et al. 2003; Kang and Xing 2005; Chefetz 2007). In both cases it is a bulk characterization of SOM, and hence, the relationships obtained are highly empirical. Chemically edited SOM samples, for aliphatic, aromatic, and carbohydrate domains, have yielded results that are not highly conclusive (Gunasekara et al. 2003; Simpson et al. 2003) in terms of the chemical nature of SOM AOC sorption sites. Accordingly, batch-type mass balance sorption experiments have shown that both chemical makeup and physical conformation are highly important (Ahmad et al. 2001; Salloum et al. 2001; Litvina et al. 2003; Todoruk et al. 2003; Wang et al. 2007); they also have left a number of important questions pertaining to these findings, especially at the molecular level. Hence, further research is warranted to elucidate how the chemical makeup and conformation of SOM affect its sorption behavior of AOCs. Stated from a molecular perspective, especially at the atomic scale, there is still a great deal of uncertainty in terms of AOC sorption/ desorption to/from SOM that needs to be resolved in order to allow for accurate predictive modeling. For further reading one is directed to Chapter 1 of this book by Joseph J. Pignatello.

13.3. BASICS OF THE NMR TOOLBOX Nuclear magnetic resonance is best known for its ability to study chemical and spatial structure, thus making it seem like an ideal method for addressing the chemical makeup and conformational issues raised in the preceding section. An understanding of the fundamentals of NMR is essential to understanding how this technique has been so successfully applied in the fields of chemistry, polymer science, and structural biology and the underlying assumptions and manipulations in the application of NMR. The following section highlights the aspects of NMR theory and application that are appropriate to the aim of this chapter; for a more comprehensive review of the NMR technique, the reader is referred to a number of excellent monographs on the

319

subject (Abragam 1961; Ernst et al. 1987; Harris 1987; Schmidt-Rohr and Spiess 1994; Claridge 1999; Deur, 2004; Jacobsen 2007; Levitt 2008), from which the discussion below is derived. 13.3.1. Signal Intensity The signal-to-noise ratio (SNR) of any modern NMR experiment is a complex concept, thus for this discussion, the following proportional relationship will suffice: SNR ¼

3=2 1=2 NQc5=2 B0 T2 T ð3=2Þ


 tmax 1=2 Tc

ð13:1Þ

where N is the amount of sample, Q is the probe quality factor, c is the nuclear magnetic strength (gyromagnetic ratio), B0 of laboratory magnetic field is the strength, T2 is longitudinal relaxation (see discussion below), T is the absolute temperature, tmax is the data acquisition time for each transient, and Tc is the recycle time between transients. This equation demonstrates three important issues involved in obtaining an acceptable SNR from an NMR experiment, due to the inherently low sensitivity of NMR: (1) the higher the gyromagnetic ratio of the nucleus under study, the higher the SNR; (2) the higher the static field, the better the SNR (unless other factors, such as anisotropy, are at play, as can be the case for 19 F); and (3) the higher the concentration, and hence natural abundance of the nucleus under investigation, the higher the SNR. 13.3.2. Chemical Shift The electrons that shield the nuclei also induce their own magnetic field around the nuclei, and this, in turn, affects the frequency at which these nuclei are observed. Although initially viewed as a nuisance, this phenomenon has become one of the cornerstones of modern NMR spectroscopy. The resonance frequency of a nucleus is always proportional to c and B0; however, the shielding electrons may cause the nuclei to experience a magnetic field different from B0. This phenomenon arises from the chemical environment around the nucleus, as it is this environment that determines how electrons are distributed around the nucleus, and hence, how the nucleus is shielded. The result is a shift in exact resonance frequency of the nucleus, giving rise to the phenomenon of chemical shift. This phenomenon can be mathematically expressed as n0 ¼ cB0 ð1sÞ

ð13:2Þ

where s is the shielding constant, which is related to the electron density (c2, for the purposes of this discussion) and distance of from the nucleus r, via Lamb’s equation (Ladd 2008)

320

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

s¼

e m0 3me 2

ð1

rc2 ðrÞdr

ð13:3Þ

0

where e, m0, and me have their usual physical meanings of elemental charge, vacuum permeability, and resting mass of an electron, respectively. The magnitude of this phenomenon is very small compared to the absolute resonance, so much so that it is expressed in terms of parts per million (ppm) of the resonance frequency. Thus, a 5 ppm difference due to chemical shift for a 1 H spectrum collected at 600 MHz results in a signal at either 599.997 or 600.003 MHz. The ability for the NMR to detect such small differences illustrates the advantage of the technique in monitoring small variations that other spectroscopic techniques may not be able to detect.

13.3.3. Spin Relaxation As with any other form of spectroscopy, an entity in an excited state must return to the ground state. The same holds for NMR; however, the full description of the relaxation picture is a little more complex than in the case of optical spectroscopy. On the whole, relaxation processes in NMR spectroscopy are much longer than in other spectroscopic techniques, and serve as another illustration of how isolated nuclei are from their surrounding environment. Long relaxation processes allow one to manipulate nuclear spin, as illustrated by the sophisticated pulse sequences used in NMR spectroscopy. It should also be stated that, when discussing rates in the context of spin relaxation, by convention, we commonly use spin relaxation times rather than rates.

13.3.3.2. Transverse, Spin–Spin, or T2 Relaxation. This form of relaxation occurs as a result of the nuclear spins experiencing different magnetic fields, causing the spins—as a collective—to lose their phase coherence. This loss of phase coherence results in a radial spreading of the spin vectors within the X–Y plane, resulting in cancellation and a loss of net magnetization within the transverse plane. The loss of magnetization within the transverse plane can occur in three different ways. The first is via the return of the magnetization to the Z-axis equilibrium orientation, as discussed above. If this is the only mechanism at work, then T1 ¼ T2. This is seldom the case. A second mechanism that can lead to T2 relaxation is due to the inhomogeneity in the external magnetic field, which can be minimized by making the applied magnetic field as homogenous as possible, and can be eliminated in the measurement of T2 via spin-echotype pulse sequence. The third mechanism, and most interesting to this discussion, is due to inhomogeneous local magnetic fields. These local inhomogeneities result from intra - and intermolecular interactions. The faster a molecule tumbles in a solution, the faster these local inhomogeneities are averaged, and thus eliminated. In the first approximation, the larger the molecule under investigation, the faster the T2 (molecular flexibility does affect this and enables liquid-state NMR experiments on high-molecular-weight polymers). For small molecules T2 values approach their T1 values; T2 relaxation can be viewed as a loss of spin order—an entropic process—, in contrast to T1 relaxation, which can be viewed as an enthalpic process. This loss of order takes place via a flip-flop mechanism between spins, and hence, can be viewed as a spin–spin process. 13.3.4. Relaxation Mechansims

13.3.3.1. Longitudinal, Spin Lattice, or T1 Relaxation. This type of relaxation, in essence, is a return to the equilibrium condition imposed by the external magnetic field. Typically, for the NMR experiments pertaining to this discussion, the external magnetic field is along the Z axis and is produced by the NMR instrument’s magnetic bore (i.e., a vertically oriented superconducting magnet) and can be expressed mathematically as h i MZ ¼ M0 1eðt=T1 Þ ð13:4Þ where t is time after an initial excitement [e.g., 90 pulse via a RF pulse of the correct frequency, strength, and duration to cause the spins’ vector to precess in the X–Y (transverse or detection plane)]. As the spins lose the energy resulting from the 90 pulse, they return to the Z-axis of their equilibrium state. The spins lose their energy to their surroundings (e.g., lattice for solids); however, the energy lost is so small in comparison to the energy of the surroundings that the surroundings can be regarded at as an infinite heat sink.

This section discusses mechanisms by which relaxation takes place. 13.3.4.1. Dipole–Dipole Relaxation. As hinted in the discussion above on T2 relaxation, dipolar interactions provide an important relaxation mechanism. This mechanism involves the communication of different spins through space via their dipoles. As molecules tumble in solution, the orientation of the nuclear dipoles changes with respect to the external magnetic field. This reorientation influences how these dipoles associate with surrounding dipoles, causing fluctuations within the local magnetic environment, inducing alternative pathways for relaxation, and hence, increasing relaxation. The more remote the spin is from other spins, the less of an effect dipole–dipole interactions will have, and the longer it will take for this more isolated spin to relax. In addition, as mentioned above, the slower a molecule tumbles in solution, the less chance there is to average the influence of the external magnetic field, and hence, the larger an effect dipole–dipole interactions will have on the

A REVIEW OF NMR AS APPLIED TO THE ASSOCIATION BETWEEN ANTHROPOGENIC ORGANIC COMPOUNDS

relaxation of the spin systems within the molecules, and thereby, the faster the spin systems will relax. This means that the dipole–dipole relaxation time is inversely proportional to the tumbling rate (correlation time tc). It should be noted that the dipole–dipole relaxation time is frequency-dependent. 13.3.4.2. Chemical Shift Anisotropy Relaxation. The electron distribution around the nuclei in a molecule is usually anisotropic; thus, how these electrons shield the nuclei will depend on their orientation with respect to the magnetic field. This phenomenon, known as chemical shift anisotropy (CSA), is partially averaged out by molecular tumbling [for solids, macroscopic sample spinning at a 54.7 angle to the applied static field is required and known as magic-angle spinning (MAS)]; however, the stronger the applied static field, the greater the effect CSA will have. In fact, CSA increases as a square of the applied static magnetic field. It induces different local magnetic fields depending on the molecular orientation with respect to the vector of the static magnetic field, thus inducing relaxation, and hence, yielding short relaxation times. This means that for nuclei with large CSA effects, applying larger static magnetic fields may not be beneficial, and hence, a compromise between increased sensitivity and CSA relaxation must be found when one is studying nuclei with large CSA effects. The effects of CSA are usually minimal for commonly studied nuclei, such as 1 H and 13 C; however, they can be rather large for nuclei such as 19 F and 31 P. It should be noted that nuclei with large CSA effects also have wide chemical shift ranges, thus making these nuclei very amenable to environmental probing. 13.3.4.3. Quadrupolar Relaxation. This mechanism applies only to nuclei with spins greater than 12 (e.g. quadrupolar nuclei such as 2 H, 14 N, and 17 O) and arises from the ellipsoidal charge distribution around these nuclei. This ellipsoidal charge distribution gives rise to an electric quadrupole moment in addition to a magnetic dipole moment. The electric quadrupole moment induces non-symmetric electric field gradients around the nuclei that will be dependent on the molecule’s orientation with respect to the applied static magnetic field and thus induces a new relaxation mechanism. Appropriately, this mechanism is known as quadrupolar relaxation and, because of its efficiency, it is usually the dominant relaxation mechanism for quadrupolar nuclei. Quadrupolar relaxation is influenced primarily by two factors: quadrupole moment of the nucleus and the molecular geometry. Intuitively, the larger the quadrupole moment, the more dominant the quadrupolar relaxation mechanism. Similarly, the less symmetric the local molecular geometry (or symmetry)–and hence, electron symmetry surrounding the quadrupolar nuclei, the more dominant the quadrupolar relaxation mechanism. 13.3.4.4. Electron Nuclear or Paramagnetic Relaxation. The importance of electrons and local electronic environments

321

has been a recurring theme in our discussion of relaxation mechanisms, but the role of unpaired electrons has been ignored thus far. Arguably, while the unpaired electrons provide a unique relaxation mechanism, in reality, paramagnetic relaxation can be (or tends to be) viewed as a special case of the dipole–dipole relaxation mechanism. The reason for this is the fact that the magnitude of the magnetic moment associated with unpaired electrons is over 600-fold greater than that for protons, making paramagnetic relaxation an extremely efficient relaxation mechanism. 13.3.4.5. Spin Rotation Relaxation. For highly mobile molecules and groups, another relaxation mechanism can exist as a result of their rotational freedom, resulting, in turn, in the formation of a molecular magnetic moment due to the ratios of the electronic and nuclear charges. This mechanism can be prevalent for methyl groups and is dependent on molecular tumbling and the local steric environment, thus it is sensitive to molecular collisions and associations.

13.4. A REVIEW OF NMR AS APPLIED TO THE ASSOCIATION BETWEEN ANTHROPOGENIC ORGANIC COMPOUNDS (AOCs) AND NATURAL ORGANIC MATTER (NOM) This section reviews how NMR has been used to study the association of AOCs with NOM, as it has become one of the, if not the most, preeminent spectroscopic technique for such studies. One purpose of this section is to highlight the advances and information that such NMR experiments provide. What makes NMR such a powerful technique is the fact that it interrogates nuclear spin, and hence, the nucleus and its environment. This environment is shielded from the external world to a much greater extent than the electrons are, and hence, the nuclear signal reports on much more localized effects. Although initially this may seem like a disadvantage for complex systems, it is in fact a major advantage; an example of this is chemical shift, which will be discussed shortly. This isolation of nuclei does mean that NMR can be—as it is in today’s modern equipment—implemented as an element-specific method due to the marked differences in gyromagnetic ratios of the different NMR active nuclei. With this in mind, the following section will be broken down by nuclei studied and then, in some cases, further into liquid- and solid-state subsections. It should be noted that for this section SOM has been expanded to include NOM so that a range of studies on dissolved organic matter (DOM) can be included as well. 13.4.1. 1H (Protons) Protons are the most NMR-receptive of all naturally occurring nuclei, giving rise to the strong wish to develop 1 H NMR techniques, as in the field of biophysical NMR (Evans 1995;

322

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

Cavanagh et al. 1996; Teng 2005). This high receptivity is counteracted by the limited chemical shift of protons (1 H have a chemical shift range of 10 ppm for the work under discussion here, in comparison to a chemical shift of 220 ppm for 13 C), which can lead to peak overlap in complex mixtures. This, along with the solid-state linebroadening mechanism, have made solid-state 1 H NMR an inappropriate analytical method for NOM research as a whole. In liquidstate experiments, the overlap issue has traditionally been overcome by 2D NMR spectroscopic methods; however, the vast majority of the 2D NMR methods applied to the study of NOM has been based on scalar (J) coupling (Cook et al. 2002; Cook 2004), making them rather inapplicable to the study of AOC association, unless a chemical bond is formed. These limitations mean that the usefulness of 1 H in the study of AOC association with NOM is limited, but some very creative work has been carried out using 1 H NMR methods. One application of 1 H NMR has been in studying the role of p–p complexes in the association of AOCs with NOM by the use of model compounds. Although the use of model compounds does not allow for the complexity of NOM to be studied, it does allow for a much simpler system to be studied, and thus, to yield highly detailed information. The choice of model compounds must be done with care so as to allow for the system under study to mimic the property of NOM that one wishes to study. The work of Pignatello and co-workers (Wijnja et al. 2004; Zhu et al. 2004) demonstrates this approach. This work shows, by monitoring 1 H chemical shifts of the compounds under study, that there are indeed p–p interactions between model p-electron donor compounds [viz., pentamethylbenzene (PMB), naphthalene (Naph), and phenanthrene (PHEN)] and a range of model NOM p acceptors, such as 1,3,5-benzenetricarboxylic acid (BTA); 1,4,5,8-naphthalenetetracarboxylic acid (NTA); pyridine (PY); 9,10-anthraquinone-2,6-disulfonate; and ophenanthroline in its protonated forms (OPH þ and OPH22 þ ). The results from these studies demonstrated that p–p associations are important for the system studied, and that the more polar the solvent system, the more important these associations become. Although the majority of these studies were not carried out in pure aqueous (<99% H2O or D2O) solvent systems, it can be argued that this was not essential, as the NOM environment being modeled is more in line with the solution composition used rather than a pure aqueous solvent system. The results also showed that protons involved in these interactions could be determined by the magnitude of the signal of each different proton as it changed under different experimental conditions. The existence of a p–p complex was also confirmed by other spectroscopic methods, and these findings were further supported by sorption batch experiments. Another study in which the importance of p–p complexes has been studied from an environmental perspective is the

work of Viel and co-workers (Viel et al. 2002). However, in addition to using chemical shift, these researchers also used the diffusion rate of the molecules under study. The molecule of interest was the well-known herbicide 2-chloro-N-(2ethyl-6-methylphenyl)-N-(2-methoxy-1-methyethyl)acetamide (Metolachlor). In the presence of the aromatic ring, this compound can be expected to give rise to p–p interactions; however, it should be noted that a wide range of pesticides used in agriculture can be expected to also give rise to p–p interactions. The study showed that the expected p–p interaction induced upfield chemical shifts of the Metolachlor molecules in solutions with a higher Metolachlor concentration. The p–p interactions were evidenced by the formation of a polymeric form (due to p–p stacking) of Metolachlor with a diffusion rate of more than two orders of magnitude slower than that for the monomer Metolachlor. The spread in the diffusion rate is indicative of some polydispersivity in the polymeric form of Metolachlor. The polymeric form of Metolachlor was further confirmed by intermolecular 1 H nuclear Overhauser effect spectroscopy (NOESY) cross-peaks between Metolachlor monomer units. (Note: NOESY cross-peaks arise as a result of through–space dipole-dipole interactions.) This study also concluded that great caution should be exercised when using herbicides at high concentrations and that the absence of p–p interactions should be confirmed either by NMR methods or, preferably, by simpler techniques, such as UV–visible spectroscopy (via the appearance of a charge transfer band) or dynamic light scattering. Jayasundera et al. (2003) utilized 1 H T1 analysis to investigate the association of Metolachlor with three model NOMs: celloluse, chitin, and lignin. The 1 H T1 values are shorter then the corresponding 13 C T1 values for the same molecules; thus the use of 1 H T1 allows for more rapid collection of data, as NMR, which is a signal averaging technique, allows for a better SNR per unit time and facilitates analyses that are otherwise too time-consuming by 13 C T1 to be carried out. The results clearly showed that the aromatic protons in Metolachlor gave T1 values that decreased with increasing concentration of lignin. This finding is consistent with the aromatic domain of the Metolachlor interacting with aromatic domains of lignin and shift consistent with p–p interactions. In regard to the interaction of Metolachlor with chitin, it was found that CH3 and CH2Cl groups adjacent to the C¼O carbon had the largest decrease in T1 relaxation times, indicating a H bonding or dipole–dipole interactions; this was further supported by the downfield shift of the aromatic protons. This study showed the power of NMR to gain information at the molecular level at the atomic scale, which, in turn, allows for a more chemical viewpoint of how AOCs interact with model compounds and, possibly, NOM. Even by using 1 H T1 measurements, the researchers still needed to use a mixed-solvent system of 60% D2O and 40% deuterated mathanol. Along similar lines, Simpson

A REVIEW OF NMR AS APPLIED TO THE ASSOCIATION BETWEEN ANTHROPOGENIC ORGANIC COMPOUNDS

et al. (2004) studied the interaction between 1-naphthol and quinoline with an isolated soil A-horizon humic acid at varying concentration by 1 H T1 measurements. This work showed that there was an across-the-board reduction in the T1 values for all protons within both molecules. Also no changes in the chemical shifts of any of the protons of 1-naphthol were found after the addition of NOM; however, substantial linebroadening, consistent with reduced T2, was shown. The authors interpreted the results as an indication of noncovalent interactions between 1-naphthol and NOM. The last two studies to be discussed took advantage of the high resolution of NMR and allowed the investigators to view atom-by-atom interactions of the pollutant molecule with the humic acid. On the pollutant molecule side of the interaction, this approach is akin to the studies of drug interactions with biological molecules such as proteins. Within the biophysical field, this type of viewing the binding event is known as epitome mapping when applied to studying interactions of an antibody with protein antigens (specific regions of the protein). The use of the term epitome mapping has been loosely translated into the field of environmental chemistry in the more recent work of Simpson and co-workers (Shirzadi et al. 2008a, b) (Note: The work cited in the previous paragraph and above by Pignatello and co-workers can be viewed as initial steps in environmental “epitome” mapping.) This work utilizes the saturation transfer double-difference (STDD) NMR method, initially developed to monitor the weak binding of ligands to protein binding sites, to study the association of a range of organic pollutants with a peat soil and its isolated humic acid fraction. The results from these studies indicate that different protons in a range of organic ligands (model pollutant molecules) interact with the NOM at different levels of intimacy. This finding contradicts the nonselective associations observed by the 1 H T1 experiments discussed above; it is interesting to note that this is also the case for 1-naphthol—the only organic ligand that the 1 H T1 and STDD studies had in common. A direct comparison between the two STDD experiments is very difficult because very different sample preparation protocols were used between these two studies. 13.4.2.

13

C (Carbon)

In terms of NMR, 13 C nucleus is 1.76  104 as receptive (suitable for study by NMR) as the 1 H nucleus. A much larger chemical shift range, as well as its chemical, biological, and environmental importance, offset the low 13 C receptivity. Traditionally, 13 C has been the nucleus of choice when studying NOM, due the complex nature of these substances and the importance of unprotonated carbons. Also, 13 C NMR is a widely used NMR technique for study of the association of 13 C-labeled AOCs with NOM in both liquid and solid states.

323

13.4.2.1. Liquid State. A pivotal study in the application of 13 C NMR to study the association of pollutants with NOM was carried out by Hatcher et al. (1993). In this study the authors used 13 C-labeled 2,4-dichlorophenol as the AOC of interest and monitored its enzymatically directed covalent bonding to a peat humic acid. The results showed a range of new peaks in the aromatic region of the 13 C NMR spectrum, providing direct evidence—absent prior to this point—of covalent bonding. Whereas, detailed assignments of the peaks seen in the 13 C spectrum were impossible because of complex nature of NOM, tentative assignments were attempted based on chemical plausibility. This work was followed by the work of Wais et al. (1995) in which 13 C-labeled anilazine was incubated with four different soils and a NOM derived from 13 C-depleted maize straw. After an incubation period the soils were mixed with organic solvents to remove extractable or nonbound 13 C-labeled anilazine, after which the SOM (humic acid) was extracted, utilizing the NaOH extraction method. The SOMs were then freeze-dried. For the NMR analysis, the organic solvent extracts were dissolved in CDCl3, while the humic acids were dissolved in 0.5 mol/L deuterated sodium hydroxide. The results indicated that the nonextractable 13 C-labeled anilazine was bound via proposed dialkoxy bonds. This study also showed that rather high concentrations of 13 C-labeled AOCs were required and that 13 C-depleted NOM aided in reducing the amount of 13 C-labeled AOCs required to obtain an acceptable signal-to-noise ratio. The concept of using extracts was further utilized by Castro et al. (1996) to monitor the interactions of chloroacetic acid, chloroacetamide, and chloroacetonitrile with a range of environmental matrices, including soils, sediments, and an activated sludge; however, the extracts that were studied consisted of the supernatants only. These studies showed the clear promise of 13 C NMR in terms of monitoring environmental transformation of organic pollutants within the environment; however, the focus on the supernatants did not allow for insight into the binding mechanisms or yield information on how the organic pollutants interacted with the soil from a mechanistic perspective. Two of the studies discussed above in the section on 1 H NMR also used 13 C NMR relaxation methods. The first was that of Simpson et al. (2004). In this work the C1 and C4 sites of both naphthalene and 1-napththol and the C2 site of quinoline were 13 C-labeled and utilized for T113 C NMR analysis. The data indicated that increasing concentrations of humic acid decreased the T1 of the 13 C-labeled sites; however, on close inspection, 13 C T1 for naphthalene results gave trends different from those for 1-naphthol and quinoline (the two compounds for which both 1 H and 13 C T1 relaxation results were shown). The data also showed that, while there was no difference between the C1 and C4 carbons in naphthalene, this was not the case for the C1 and C4 carbons in 1naphthol. In fact, the results for the C1 carbon in 1-naphthol

324

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

closely resembled the trends seen for the C2 carbon in quinoline. For both the 1-naphthol C1 carbon and quinoline C2 carbon, it appears that there is a two-stage mechanism in how the pollutant interacts with the NOM versus NOM concentration. This dual behavior may indicate either two different binding modes or two different environments. The first, as discussed by the authors, is not supported as the chemical shift of the C1 of 1-naphthol and C2 of quinoline do not measurably change as a function of NOM concentration. This means that two different environments may be a possibility, where the two different environments could be more and less mobile domains within the NOM as dictated by NOM concentration; however, the C1 for the 1-naphthol does not follow this trend. This discussion shows the difficultly in interpreting relaxation data in these types of experiments; see text below for more discussion on this issue. The work of Jayasundera et al. (2003) monitored the association of 13 C-labeled (the Aromatic ring was fully labeled, as was the C¼O group) Acetanilide with a range of biolpolymers (lignin, chitin, cellulose, and collagen) over a range of biopolymer concentrations. The data obtained suggested that different entities of the acetanilide associated with different functional groups within the biopolymers depending on the biopolymer’s concentration. 13.4.2.2. Solid State. One major advantage of NMR in the area of environmental studies is that it is suitable for both liquid- and solid-state samples. For a majority of other fields (exceptions, such as polymer and materials, do exist) in which NMR is applied, liquid-state measurements dominate. The reasons for this are historic, economic, and practical. However, as solid matrices are an essential part of our environment, their analysis is essential and solid state NMR has arisen as one of the most—if not the most—important if not, the preeminent analytical tools in such studies. Some of the initial solid-state NMR work examined the association of 13 C-labeled pollutants with model soil components. An example of this early work was the work of Jurkiewicz and Maciel (1995), in which the pollutants (sorbates) were acetone, trichloroethylene (TCE) and carbon tetrachloride, while soil components (sorbents) under study were kaolinite, bentonite, humic acid, and a whole soil. It was found that the more heterogeneous the sorbent, the wider the 13 C label signal of the sorbate. However, it was also found that acid treatment with HCl did result in signal narrowing, which could be due to the removal of paramagnetic centers. It can also be argued that the acid treatment could also have caused a perturbation in the NOM, resulting in less of a heterogeneous environment being sampled by the sorbate under study. Little difference in chemical shift was observed, and, in fact, no difference in chemical shift was found for carbon tetrachloride between sorbents. This work was followed by the work of Tao et al. (1999) in which the photodecomposition of TCE was monitored in the presence of Ca,

Cu, and Zn montmorillonites, kaolinite, silica gel, and a whole soil by a combination of solid-state and liquid-state 13 C NMR techniques. Solid-state NMR showed similar spectral signatures for the photodecomposition of TCE on the whole soil and the Ca-montmorillonite; however, a very similar signal could also have been constructed if the signatures of the TCE photodecomposition in the presence of Cu-montmorillonite and silica gel were combined. These results did, however, hint that the NOM within the soil surprisingly played little to no role in the photodecomposition of TCE. Witte et al. (1998, 2002) used 13 C NMR to monitor the interactions of 13 C-labeled 2-aminobenzothiazole (ABT) and 2-(methylamino)benzothiazole (MABT) with NOM. The first of these studies found that oxygen promoted the uptake of ABT and MABT. The authors then isolated the labeled compound’s signal from the NOM’s background signal by subtracting the NOM NMR data from the ABT or MABT-loaded NOM NMR data. In doing so, the authors were able to obtain a clear difference spectrum of the associated ABT and MABT. The spectra of ABT and MABT molecules associated with NOM showed a major peak in the vicinity of 169–170 ppm and a second smaller peak around 10 ppm upfield, in the vicinity of 157–158 ppm. The first larger peak had the same chemical shift as that of the unassociated ABT and MABT molecules. The lack of chemical shift could mean that this fraction of ABTand MABTwas not covalently bound to the NOM. In contrast, the peak shifted 10 ppm upfield, which strongly indicated that this fraction of ABT and MABT was covalently bound to the NOM. This assertion was further supported by their findings that oxygen strongly promoted the association of ABT and MABT with the NOM under study and that the much smaller fraction of ABT and MABT was found to be represented by the 10 ppm upfield peak when the experiment was done in argon rather than oxygen. This work was followed by that of Witte et al. (2002) in which the brown coal NOM sample used in the first study was replaced with a soil-extracted NOM, and the association of ABT with NOM was done in both the presence and absence of the oxidative enzyme laccase. The association experiments carried out in the absence of laccase yielded results very similar to those found for the coalextracted NOM. This result would indicate, as pointed out by the authors, that the two NOMs bound ABT in a very similar manner; however, significantly more ABT associated with the soil-extracted NOM. The association of ABT in the presence of laccase yielded results in which the 10 ppm upfield peak was a more dominant feature than in the absence of laccase, thus indicating that, while the laccase promoted the covalent bonding mechanism, it did not affect it when already in place. The work discussed above by Witte et al. (1998, 2002) relied heavily on differential spectra in order to correct for background signal (e.g., in this case, ordered noise) of the

A REVIEW OF NMR AS APPLIED TO THE ASSOCIATION BETWEEN ANTHROPOGENIC ORGANIC COMPOUNDS

NOM. This method—used by a number of other researchers as a way of studying the sorbate—assumes that the spectrum of the neat NOM used to create the difference spectrum is sufficiently a close representation of the spectrum of NOM containing the sorbates so that the subtraction procedure does not perturb the sorbate signal. K€acker et al. (2002) utilized 13 C-labeled fluoranthene and phenanthrene as well as differential spectra to study the association of these sorbates with three different soils in the form of an Ah/composite and artificial soil. The results showed that 13 C labeling combined with differential spectra is a promising method for analyzing the association of pollutants with NOM; however, it should be noted that this technique yielded satisfactory results only for the Ah/composite and artificial soil. The results on these two soils showed the emergence of a new chemical shift, interpreted as evidence for the formation of covalent bonding between the sorbate and the sorbent. These findings were further confirmed by mass spectroscopy, liquid-state spectroscopy, database analysis, and molecular modeling. The authors also analyzed how much 13 C from the labeled compound compared to that found naturally in the NOM and came up with a ratio of approximately 1.5/1.0 for 13 C-activitysorbate/13 C-activityNOM. Smernik (2005) has further evaluated the use of 13 C labeling and differential spectra by combining chemical shift data with proton spin relaxation–edited data. (Note: Relaxation editing refers to using relaxation as a filter so that one can eliminate quickly or slowly relaxing components from the spectrum.) The advance in this work was the separation of organic matter within these soils into fast and slow relaxation components and then associating the amount of sorbate in these two fractions of the sorbent. This was accomplished by subtracting scaled versions of the slowly and rapidly relaxing subfractions of the pure sorbent. Carbon-13-labeled sorbates have also been used to utilize 13 C NMR in the study of pyrene in sediments, in conjuction with pyrolysis–gas chromatography–mass spectroscopy (Guthrie et al. 1999). The results from this study showed that sorbed pyrene preserved its chemical shift, thus indicating a noncovalent interaction. The results from the humin fraction of the sediment NOM match the findings on the whole sediment, indicating that pyrene was noncovalently associated with the humin fraction of the sediment. Pyrolysis–gas chromatography–mass spectroscopy results in the study confirmed the NMR finding. Although the use of 13 C-labeled sorbates allowed new insights into the association of these sorbates with NOM, very little direct data were found on the type of moieties within the NOM with which the sorbate was interacting. Sachleben et al. (2004) pursued this issue by monitoring the sorption of 13 C-labeled pyrene to both cutan and cutin. Taking advantage of spin diffusion and 2D 1 H–13 C heteronuclear correlation experiments, the authors were able to clearly show that pyrene was sorbed to the aliphatic moieties

325

for both the cutin and cutan samples. Spinning-sideband analysis also showed, as expected, that the sorbed pyrene was more mobile than crystalline pyrene; however, this analysis also showed that there was a large increase in anisotropy, which led to an incomplete magic-angle spinning (MAS) cancellation. 13.4.3.

15

N (Nitrogen)

Carbon-13 labeling is not “the only game in town.” In some cases, 15 N labeling may be more appropriate from a chemical perspective. Such is the case in the studies of the association of trinitrotoluene (TNT) and its metabolites within soils. This area of research has attracted a large amount of interest for two reasons: (1) the large amount of TNT residues left by military action, threatening local ecosystems and water supplies and (2) TNT and its metabolites covalently bond to NOM, which was shown to provide highly informative chemical shift information. The work of Thorn and co-workers (Thorn et al. 1992, 1996, 2002, 2004, 2008; Thorn and Kennedy 2002) highlights the difficulties as well as the systematic patience that is needed in studying the association of pollutants with environmental matrices. This work clearly illustrates the usefulness and importance of model sorbates and sorbents in aiding the interpretation of data from systems composed of actual sorbates and sorbents. These studies also showed the importance of combining solid- and liquid-state NMR methods in order to correlate what was being seen in the high-resolution spectra of the extract with what was taking place in the whole soil, as monitored by the lower-resolution solid-state spectra. This combination, in concert with 13 C NMR characterization of the sorbate, allowed for molecular level insight into the association of TNT with soils. The initial studies took advantage of the ACOUSTIC and refocused INEPT 15 N NMR methods to provide “evidence of the reaction of hydroxylamine with ketone, quinone and ester functional groups present in humic and fulvic acid samples” (Thorn et al. 1992). [Note: The ACOUSTIC (alternate compound oneeigthies used to suppress transients), pulse sequence is used with low-resonance nuclei, 15 N here, to overcome probe ringing and the associated baseline distortion. The INEPT (insensitive nuclei enhanced by polarization transfer) is a method used to allow nonselective polarization transfer between 1 H and 15 N spins in this case.] This work was followed by a more detailed study on the association of aniline with a range of humic acids, revealing that approximately 50% of the aniline was incorporated into heterocyclic linkages (Thorn et al. 1996). The results from this study suggested that a range of carbonyl groups were involved in the binding of aniline, thus meaning that a simple chemical mechanism seemed unlikely for the binding of reduced TNT amines with soil matrices and, in particular, the soil organic matter. To further investigate this possibility, Thorn and

326

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

Kennedy (2002) studied the association of a range of possible aromatic amines—reductive degradation products of TNT— with a humic acid and sawdust, in the presence and absence of horseradish peroxide, by a combination of solid-state 13 C CP-MAS spectra and 15 N inverse gated decouples, DEPTand ACOUSTIC liquid state spectra as well as 15 N CP-MAS spectra. [Note: DEPT (distortionless enhancement by polarization transfer) is another polarization transfer method that allows one to determine multiplicity in terms of bonded protons.] The results from this study built on the results from the studies by Thorn’s group on aniline, and revealed that all of the reduced TNT amines reacted with a range of carboxyl groups, including quinone-type entities, to form nonheterocyclic and heterocyclic condensation products. It was also found that neither enzymes nor metal catalysts were required for nucleophilic attack; however, catalyzed reaction did yield an increase in imine formation and an evolution of a new peak with possible assignment to imidazole, oxazole, and/or pyrazole. Finally, it was noted that diamines did bind to lignocellulose. This was then further built on by investigating the reduction and binding of TNT with the aid of an aerobic bench-scale reactor experiment (Thorn et al. 2002). Once again NMR, in the form of solid-state 15 Ndirect polarization (DP)-MAS and CP-MAS, as well as ACOUSTICS liquidstate analyses, were critical in the evaluation of what took place in these experiments. The 15 N NMR results showed the reduction of TNT on the bench scale, and that during the composting stage, amines covalently bound to the organic matter present. As the vast majority of the organic matter was lignocellulose, these results confirmed the previous results that TNT degradation products could bind to this form of NOM. However, the study did raise questions regarding the fate of these TNT degradation products during the natural humification processes. Thorn et al. (2008) have studied the association between 2,4- and 2,6-dinitrotoluene (DNT), both of which are intermediates and impurities in the production of military-grade TNT, and 4-methyl-3-nitroaniline (4M3NA, 2,4-DNT, a major reduction product) with NOM in the presence and absence of horseradish peroxidase via aerobic composting. As usual, the analytical tools used were a range of solid- and liquid-state 15 N techniques, as presented above. The finding of this study was that covalent bonds could be formed between either of the two studied DNT isomers and NOM, and occurred with the participation of the amine groups on the DNT molecules. It was also found that 4M3NA covalently bound (via nucleophilic addition) with soil-derived NOM utilizing quinone and other carbonyl entities within the NOM. The work by Knicker and co-workers (Knicker et al. 1999, 2001; Knicker 2003) on monitoring of the association of TNT with NOM via 15 N NMR was done in parallel to the work of Thorn and co-workers discussed above. A range of solidstate 15 N NMR techniques were used in all three studies. The initial study was aimed at optimizing the relevant NMR

parameters to obtain the most informative spectra possible. This work revealed that anaerobic/aerobic composting of TNT in soil led to the reduction of TNT to aromatic amines, which then bound covalently to a range of carbonyl groups— including that in quinone—associated with the NOM within the soil. Knicker et al. (2001) also looked into the possibility of HF treatment to increase of the SNR in regard to organic matter associated TNT. The results from this study did show an increase in SNR in the HF-treated samples; however, it was also shown that the HF treatment extracted between 6% and 29% of the spiked TNT. It was also concluded that the low-resolution nature of the solid-state 15 N spectra obtained was a major issue that prevented the assignment of specific chemicals. This led Knicker (2003) to examine the possibilities of two-dimensional solid-state NMR in the form of 13 C–15 N heteronuclear experiments to probe the association of TNT with NOM. The results clearly indicated that TNT was covalently bound to NOM; however, the low sensitivity of the technique required that the NOM be 13 C-labeled and demanded high TNT loading. The requirement for 13 C labeled NOM limits the application of this technique to model surrogate systems. 13.4.4.

19

F (Fluorine)

Up to this point, all the nuclei discussed have been natural constituents of NOM, which is the sorbent of interest. The same cannot be said about fluorine. This means that the only organic (there are natural forms of inorganic fluorine; e.g., fluorite and fluorapatite) fluorine signal within an environmental matrix should be from human input. Stated another way, there should be no natural organic fluorine, and hence, no fluorine background signal. In addition, 19 F is 100% naturally abundant, is highly receptive (in NMR terms), and has a wide chemical shift range. Because of these advantages, 19 F-labeled molecules have been used as model pollutants in 19 F-NMR-based experiments aimed at monitoring the association of a range of pollutants with a range of environmental sorbents. This work can be divided along the solid- and liquid state NMR lines, similar to the 13 C NMR studies discussed above. 13.4.4.1. Liquid State. Bleam and co-workers (Chien and Bleam 1997; Chien et al. 1997) monitored the association of CF3-labeled atrazine [2-chloro-4-(trifluoroethylamino)-6(isopropylamino-s-(triazine)] association with humic acid. Both studies were focused on determining the environment the associated atrazine within the dissolved humic acid. The first study approached this by comparing the environment that atrazine finds itself in when associated with SDS [sodium n-dodecyl(lauryl)sulfate] micelles and dissolved NOM. The authors concluded that atrazine experienced a hydrophobic domain within the dissolved NOM very different from that found in SDS micelles. The following observations contributed to that conclusion:

A REVIEW OF NMR AS APPLIED TO THE ASSOCIATION BETWEEN ANTHROPOGENIC ORGANIC COMPOUNDS

1. Chemical shift data, indicated that only one conformer of atrazine was observed when associated with the dissolved NOM as compared to the observation of multiple conformers of atrazine in the SDS micelle system. 2. Increasing atrazine concentration did not result in chemical shift of the 19 F signal for the NOM system; instead, there was a decrease in the chemical shift in the SDS system. 3. Aprotic nonpolar solvents failed to induce an increase in the number of conformers seen in the dissolved NOM solution, and while polar solvent stabilized atrazine dimers after a threshold point was crossed, in all cases in the SDS system the addition of aprotic and polar organic solvent resulted in alteration of the chemical shifts of the atrazine conformers. These results, when combined, would indicate that hydrogen bonding is involved in atrazine association with NOM and that this hydrogen bonding favors the association of atrazine with NOM, overriding hydrophobic partitioning into a range of organic solvents. The second study was aimed at determining if atrazine associated with NOM was wateraccessible or buried within the NOM assemblage. It was found that the aqueous soluble Gd relaxation agent yielded no change in linewidth and T2, while the hydrophobic relaxation agent TEMPO (2,2,6,6-tetramethyl-1-piperidinyloxy) yielded a large increase in linewidth and hence a profound decrease in T2 of the fluorine-labeled atrazine. These results strongly implied that atrazine associated with NOM interacts with deeply buried (in regard to the aqueous interface) hydrophobic domains within the humic acid. However, even within these hydrophobic pockets, hydrogen bonding between atrazine and the NOM is taking place, as implied by the previously discussed study. Dixon et al. (1999) further illustrated the utility of 19 F NMR in monitoring the association of organic compounds with NOM. This work consisted of monitoring the association of 4-fluoro-1-acetonaphthone with Suwannee river NOM by a range of 19 F NMR methods, including chemical shift, relaxation and correlation times and diffusion analysis. This combination of methods illustrated the level of information that can be obtained using NMR, and also showed the importance of combining a range of NMR techniques when studying systems as complex as NOM and its association with pollutants. The relaxation (T1 and T2), correlation, and diffusion data implied that if the concentration of 4-fluoro-1acetonaphthone was held constant, while the NOM concentration was increased, the 4-fluoro-1-acetonaphthone would become less mobile. Specifically, both the T1 and T2 values as well as the diffusion coefficient decreased with increasing fulvic acid concentration, while the correlation time increased. The chemical shift data showed a downfield shift with increasing NOM concentration and that 70% of the 4-

327

fluoro-1-acetonaphthone being bound in a system composed of 0.9 mmol/L 4-fluoro-1-acetonaphthone and 15 mmol/L NOM (assuming a molecular weight of 900 for the fulvic acid). Such a shift is strongly indicative of hydrogen bonding between the 4-fluoro-1-acetonaphthone and NOM, which can easily be envisioned. The hydrogen bonding hypothesis is also consistent with the relaxation time, correlation time, and diffusion coefficient data. The hydrogen bonding hypothesis was also further supported by pH dependent data, which showed a strong variation in both correlation time and diffusion coefficient in the vicinity of pH 6. These data also suggested that both hydrogen bonding and hydrophobic interactions are both at work in the association of 4-fluoro1-acetonaphthone with the NOM under study. Finally, the authors used a heteronuclear Overhauser enhancement (HOE) method to gain a molecular level understanding of the moieties within the NOM that the 4-fluoro-1-acetonaphthone was associating with. The results from these experiments revealed that 4-fluoro-1-acetonaphthone was associating with both aliphatic and aromatic moieties within the NOM. This multitechnique 19 F NMR study of the association of 4-fluoro-1-acetonaphthone with Suwannee River NOM clearly illustrated that the association of pollutant molecules with NOM is highly complex and can consist of a range of association mechanisms with a range of moieties within the NOM. The complexity of pollutant association with NOM and what it means to possible degradation products, including metabolites and the utility of NMR, was illustrated in the publication by Strynar et al. (2004) using 19 F NMR revealing that the chemical shifts of soil extracts did not match up with a range of predicted metabolites. The T1 relaxation times were consistent with trifluralin metabolites being associated with NOM. In addition, a range of different solvent extractions, followed by 19 F NMR analysis of the extracts, implied that trifluralin was covalently bound to soil NOM, and on the basis of chemical shift data, it was hypothesized that this binding utilized aminotrifluralin metabolites. 13.4.4.2. Solid State. One disadvantage of 19 F as an NMR nucleus is its inherent chemical shift anisotropy (CSA), which can lead to linebroadening, especially at higher magnetic fields, and reduce SNR, and hence, counteract its naturally high NMR receptivity. In solid-state 19 F NMR spectroscopy, CSA is a much larger problem than in the liquid state. This may help explain the relatively few reports of monitoring pollutant interactions with solid environmental matrices by 19 F NMR. Cornelissen et al. (2000) overcame the CSA problem by high sample spinning rates of 20–23 kHz. These high sample spinning rates help overcome CSA linebroadening as well as remove spinning sidebands from overlapping spectral areas of interest, while at the same time, focusing the signal into the central peak, rather than the sidebands. The downside to such

328

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

high sample spinning rates is a smaller sample volume, hence less sample within the probe’s coil for detection, and hence— all else being equal—lower SNR. Using this technique, the authors were able to show that the associated hexafluorinated benzene (HFB) was a chemical environment more closely modeled by polystyrene polymer than by activated carbon within two sediments. In addition, it was found that HFB sorbed into two different sorption domains within the sediment, giving rise to a downfield signal likely to be associated with a strongly electron-withdrawing environment. The width of this signal was substantially narrower, thus indicating that HFB within this environment was more mobile in comparison with other HFB sorption domains. Finally, it was found that this downfield signal was associated with rapidly desorbing HFB. On the other hand, the slowly desorbing HFB was associated with a much less electron-withdrawing environment, in which—on the basis of linewidth analysis— HFB was substantially less mobile. Rice and co-workers (Kohl et al. 2000; Khalaf et al. 2003) used solid-state 19 F NMR to study the sorption of HFB to soil NOM and fractionated NOM. It should be stated that these studies used solvent swollen NOM, and hence, the experimental method was akin to the high-resolution magic-angle spinning (HR-MAS) methodology. The solvent swelling makes the system under study more mobile, and hence provides the sorbed HFB with more mobility. This mobile HFB reduces the effect of the inherent CSA of the 19 F nuclei, with the MAS part of the experiment further reducing 19 F CSA. Consequently, high-resolution data can be obtained and compared to those possible in a traditional pure solidstate methodology. In fact, the approach by Rice and coworkers can be viewed as a nonlocked 19 F HR-MAS NMR methodology. The initial paper showed that, as HFB sorbed to either a whole or lipid-extracted peat, three distinct forms of HFB could be detected: free, mobile, and immobile. As the sorption processes were allowed to evolve, the free HFB disappeared after 60 min in both peats. In the whole peat it appeared that the amount of mobile HFB increased during the initial 60 min, before decreasing for the next 23 h. On the other hand, the amount of immobile HFB increased through the entire 24 h monitoring time. The lipid-extracted peat, however, showed a different trend, where the amount of mobile HFB continued to increase throughout the 24 h study period, while the amount of immobile HFB appeared to have increased over the first 60 min and then decreased over the remaining 23 h observation period. The high resolution spectra obtained by the HR-MAS technique also showed that there were two distinct peaks for the mobile HFB; these two peaks were more dominant by a factor of 1.8 in whole peat versus the lipid-extracted peat. However, further interpretation was difficult due to the complexity of sorbent under study. A much wider peak—likely indicating a shorter T2, and hence, less mobile HFB (however, it could also indicate a large heterogeneity in the local environments associated with

this fraction of the sorbed HFB)—was also found downfield of the other HFB forms, which would suggest that the immobile HFB was in a more electron-withdrawing environment. Rice and co-workers (Khalaf et al. 2003) extended the use of their methodology in order to study the association of HFB with seven different-size fractions of a soil NOM, ranging in size from 0.2 mm to 1 kDa. For the whole NOM and its high-molecular-weight fraction, two different types of associated HFB, namely, mobile (narrow-peak) and immobile (broad-peak) were found. This picture changed when the medium-weight (30–100-kDa) NOM fractions were analyzed; the broad peak transitioned to two sharper peaks encompassing the same chemical shift range. The narrowing of the peaks may indicate higher mobility for the HFB within these chemical environments with these NOM fractions and a less homogenous nature to these chemical environments in the medium-weight NOM fractions. The lightweight humic acid fractions gave spectra with even narrower peaks; however, these peaks resulted in spinning sidebands and were absent in static state experiments, thus it appears that the HFB with the previously labeled rigid domains is rigid but the responsible chemical environment in the NOM becomes more homogeneous at lower molecular weights. It was also found that the amount of HFB associated with the immobile domains was correlated with the percentage of aliphatic carbons within the humic acid fractions, thus implying that the immobile HFB was associated with aliphatic domains within the humic acid fractions. Simpson et al. (2007) studied the association of HFB within a whole soil column using 19 F NMR imaging. The results showed that pure HFB, when applied to the top of the soil column, quickly penetrated the soil column; however, this penetration slowed down dramatically over time, which was unexpected given that HFB is substantially more dense (1.60 g/cm3) than water, and gravity forcing of the HFB into the column is expected. It was found that, while it was difficult to monitor the distribution of sodium fluoride, it was possible to monitor its dissipation by signal loss as a function of time. It was stated that similar studies were carried out on trifluralin, but the signal completely disappeared in the course of the 24 h experiment. 13.4.5. 2H (Deuterium) Unlike the nuclei discussed so far, deuterium has a nuclear spin of 1 rather than 12. It is therefore a quadrupolar nucleus, meaning a nucleus with a spin greater than 12. As a whole, the dominant relaxation mechanism for such nuclei is via quadrupolar relaxation, which, in turn, simplifies the interpretation of relaxation data for deuterium. It should also be noted that deuterium has a relatively small quadrupolar moment, and hence, a small linewidth factor. It also has a rather low receptivity in comparison to other quadrupolar nuclei (Harris 1987).

A REVIEW OF NMR AS APPLIED TO THE ASSOCIATION BETWEEN ANTHROPOGENIC ORGANIC COMPOUNDS

13.4.5.1. Liquid State. Nanny and co-workers (Nanny 1999; Nanny and Maza 2001) studied the association of deuteriumlabeled monoaromatic compounds with dissolved NOM, taking advantage of the fact that the quadrupolar relaxation mechanism dominates the relaxation of deuterium nucleus under their experimental conditions. The results from the first study (Nanny 1999) revealed no differences in the T1 values for phenol in the presence and absence of NOM, which is strongly indicative of the lack of association between phenol sorbate and fulvic acid sorbent. Pyrine, on the other hand, showed evidence of noncovalent interaction with one of two NOMs examined. The noncovalent association appeared to involve the lone pair of electrons on the pyrine molecule and hydrogen atoms belonging to moieties within the NOM, possibly phenolic moieties, as postulated on the basis of the pH dependence of the deuterium T1 times. In addition, the T1 values of the a and b versus c deuterium nuclei are consistent with anisotropic C2 axis rotation. Benzene, on the other hand, showed a very different trend in the fact that its T1 values increased in the presence of both examined NOMs. This would indicate that the addition of NOM made benzene more mobile. Initially, this seemed counterintuitive; however, the author explained that the less polar environment of NOM changed the solvent system enough for benzene to disaggregate and tumble more freely. This would then decrease its correlation time, and hence, increase its T1 times, as observed. This work was followed by a study of the association of monoaromatic compounds with three NOMs by Nanny and Maza (2001). Within this latter study the possible impact of viscosity was examined. An increase in phenol T1 values was found for all three NOMs at acidic pH (here, pH <5). The magnitudes of these T1 values indicated that their increase was due to an increase in phenol’s correlation time, so that it was no longer in the extreme narrowing region. Such a shift indicated that phenol may have associated with the more compact, less mobile, and more hydrophobic molecular arrangement of a NOM at these lower pH conditions. At basic pH (here, pH >9), a decrease in T1 values was noted, although the magnitude did not indicate an exit from the extreme narrowing condition. This decrease was consistent with the association of phenol with the NOMs, possibly via hydrogen bonding between the unprotonated phenol and hydrogen-containing moieties within the NOM. The association of pyrine with the humic acids showed an anisotropic interaction. The pH dependence and decrease in T1 values indicated that at least two association mechanisms were at play, namely, through the lone pair of electrons on pyrine and via p–p interactions. The p–p interactions were more dominant under acidic pH conditions as well as for NOMs higher in aromatic content. Benzene gave results for its association with humic acids exactly opposite those found for its association with fulvic acids indicated by the fact that T1 decreased in the presence of NOMs. This decrease was consistent with a decrease in

329

benzene’s mobility, but when combined with the NOM findings discussed above, it indicated that there was a much stronger association of benzene with different NOMs based on hydrophobicity. Zhu et al. (2003a,b) have carried out studies involving the association of deuterated benzene, fluorobenzene, and toluene with suspended soil particles. A quick note as to why these experiments are being discussed as liquid state while the swollen soil studies of Rice and co-workers were discussed as solid-state experiments is in order here. The differentiation comes from what was the bulk phase, and hence, the equipment and type of experiments run. In the case of the work by Zhu and co-workers, water was the bulk phase and liquid-state NMR equipment and experimental techniques were used. The first of the two studies cited above focused on benzene, fluorobenzene, and toluene as the sorbates and a whole soil as well as a NaOH-extracted and H2O2-treated version of the same soil as the sorbent. The results showed that the T2 values of toluene were affected most when toluene was allowed to associate with the untreated soil particles. Although the monitored T2 values are, as pointed out by the authors, an average of the free and associated T2 values, they do allow insight into how restricted the associated molecule actually is. This result would indicate that the hydrophobic effect is the major driver behind the association of toluene, as well as those of fluorobenzene and benzene (association strengths are in the order given), with the suspended soil particles studied. The effect of pH was most prevalent in terms of effect on T2 for benzene and the aromatic ring of toluene, and showed decreased mobility with decreasing pH. This finding was consistent with the NOM component of the suspended particles becoming protonated, less mobile, and more hydrophobic. This was not the case for the methyl group of toluene or for fluorobenzene. This led the authors to conclude that the association of the studied pollutants occurred by a combination of hydrogen bonding, cation–p interactions (between charged NOM amino groups and the aromatic ring of the pollutant), and/or aromatic–aromatic (favored by low pH) interactions. For the treated soil particle experiments, it was found that the T2 value for benzene decreased and increased when it associated with the NaOH- and H2O2 treated soil-particles, respectively. In addition, the same pH dependence was found for the NaOH-treated as for the non-NaOH-treated soil particles. These observations were consistent with (1) the humin fractions (remaining after the NaOH treatment) being more rigid than NOM as a whole, (2) the NOM fraction of the soil particles being the major medium with which the pollutants under study associated, and (3) the humin fraction having some functionality (e.g., COOH groups). This study was complemented by another study by Zhu et al. (2003b), in which the association of pyrine with soil particles was investigated by deuterium relaxation and chemical shift measurements. Once again, it was found that the pollutant

330

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

associated with the organic matter component of the suspended soil particles. This work did raise a question as to the importance of mineral surfaces to the association of aromatic pollutants as well as pyrine; however, as only one soil was studied, further research will need to be carried out verify the findings of these studies as general. 13.4.5.2. Solid State. In solid-state experiments, the mobility of deuterium is somewhat restricted. Therefore, a range of information can be obtained within the timescale of NMR methods. A particular example of the usefulness of this method has emerged in 2 H NMR lineshape analysis. An explanation of the theory and mathematical implementation of this theory for analysis of experimental data is beyond the scope of this work; suffice it to say, the calculations are based on well-accepted density matrix methodology. For the purpose of environmentally relevant work usually a “powder summed” approach is adopted, due to the random orientation that one will encounter when analyzing amorphous samples. The reader is referred to the classic monograph by SchmidtRohr and Spiess (1994) for further discussion. One of the first environmental studies to employ 2 H lineshape analysis as an analysis tool was by Janusa et al. (1993), who analyzed the reorientation of d5-phenol within white Portland cement in order to gain insight into the solidification and stabilization of this pollutant within the cement. The results revealed that 50% of the phenol remained highly mobile, and liquid-like, even after a year of curing. It is interesting to note, however, that after 1 year it appeared that, on a percentage basis, 10% d5-phenol loaded cement yielded a greater amount of solid-like d5-phenol than the 1% loaded sample. The analysis did show that 2 H lineshape analysis (1) was a promising method even for complex solidstate matrixes, (2) could yield information in on mobility and types of reorientation—such as rotation and flip, and (3) could provide binding energy under certain conditions. This style of analysis was expanded on by Xiong et al. (1999) to soil components, including a soil NOM. The two pollutants that were sorbed to the soil NOM were ethylene glycol and acetone. It was found that a large fraction of ethylene glycol was essentially immobilized at temperatures as high as 25 C; however, at temperatures as low as 125 C a sizable mobile fraction of acetone was detected. An explanation for this observation involved hydrogen bonding, as ethylene glycol was able to hydrogen-bond more effectively than acetone, due to its two OH groups. It was also noted that acetone showed binding behavior with soil NOM that was very different from that found with Ca-montmorillonite; however, no whole soil data were shown to allow for a comparison of the soil NOM and Ca-montmorillonite data. This above study showed that the complex heterogeneous nature of soil NOM made it very difficult to apply 2 H analysis to the sorbed d-labeled pollutants. Difficulties stemmed from a multitude of different environments, and hence, pollutant

sorption sites, and consequently, many possible orientations within a soil NOM. This complexity is beyond both experimental resolution and the computational power currently available. The results of these two studies as well as others by Maciel and co-workers (Xiong and Maciel 1999) revealed, however, that 2 H NMR in combination with 2 H lineshape analysis is a powerful tool. It allows unique insight into the binding mode of 2 H-labeled pollutants to model soil components, thus revealing the power of model systems. However, obtaining detailed information at the molecular level is severely limited by the complexity of real soils and NOM. A final study of interest for this section was carried out by Butler and co-workers (Emery et al. 2001) in which 2 H spinning-sideband pattern analysis was applied to studying the sorption of deuterium-labeled TNT to quartz, two montmorillonite clays, and a soil. Once again, it was found that the quartz and clay samples modeled well, while the soil sample was much more difficult to model. Nevertheless, it appeared that the soil data were similar to those found for K10-montmorillonite but the spinning sidebands were broader (due, in all likelihood, to the heterogeneity of the soil). It was found, however, that the spinning sideband pattern for the soil was consistent with a fitting for C3 ring jump, free-ring rotation, and stationary rings, thus showing some promise for 2 H NMR in providing molecular level information of the bound state of deuteriumlabeled pollutants within soils with the aid of model systems.

13.5. A HIGHLY CRITICAL EVALUATION OF NMR AS APPLIED TO AOC ASSOCIATION WITH NOM This section provides an evaluation of how NMR has been used to -date to study the association of AOCs with NOM in terms of the absolute demand that these experiments 1. Be carried out under environmentally relevant conditions, including concentration, ionic strength and pH, temperature, and pressure 2. Be carried out on real samples that have undergone as little treatment as possible, where the optimum is in situ field monitoring 3. Yield molecular level information in regard to moieties within the NOM that are involved with its association with the AOCs under investigation 4. Provide both thermodynamic and kinetic data These demands cannot be met with the currently available NMR methodologies and the state of our understanding of NOM or soils; thus the highly critical nature of this chapter should not be viewed as a condemnation of any of the work discussed above. In fact, all examples chosen for discussion above in Section 13.6 demonstrate very creative

A HIGHLY CRITICAL EVALUATION OF NMR AS APPLIED TO AOC ASSOCIATION WITH NOM

experimental design, expert execution, and brilliant interpretation considering the complexity of the system under study. With that said, any obtained data need to be discussed along with a full disclosure of limitations to their respective experimental designs in mind, to ensure that the community is not led down a path of false hope, and causing skepticism by other scientific fields, public distrust, and hence, limiting resources being applied to this much-needed research. 13.5.1. Environmentally Relevant Conditions At present, the phrase “environmentally relevant” is treated rather loosely and is very open to interpretation. In order to better focus environmental NMR studies, there is a need to establish some ground rules in terms of exactly what “environmentally relevant” conditions mean. 13.5.1.1. Physical State. As discussed above, solution-state NMR has traditionally yielded higher-resolution spectra compared to solid-state NMR; however, it has also prevented the solid-state matrices from being investigated. This means that, if one wished to study the association of a pollutant with solid matrix natural organic matter (SMNOM), or more precisely a fraction of SMNOM, it must be extracted from the solid matrix and subsequently dissolved in a solvent system appropriate for liquid-state NMR studies to be carried out. This extraction–isolation process constitutes a large step away from the environmental conditions and raises a range of concerns as to whether the conditions under which studies on these extracted and isolated SMNOM fractions are carried out are environmentally relevant, including 1. Editing of the SMNOM by the extraction/isolation step. The first type of editing is that NOM that is not soluble within the extraction solution will not be present in the extracted NOM. An example of this is the loss of the humin fraction in the vast majority of aqueous extraction methods. 2. Loss of the SMNOM–mineral interface. As the objective of the extraction–isolation procedure is to separate SMNOM from its solid matrix, any structural features resulting from the native SMNOM’s association with a solid interface is lost. 3. Degeneration of conformational arrangements within the native SMNOM. The extraction–isolation step destroys a range of noncovalent interactions, including hydrogen-bonding networks as well as dipole–dipole interactions. This step may also induce noncovalent interactions normally absent in the native SMNOM, such as hydrophobic interactions occurring during the precipitation and resin adsorption–desorption steps in the isolation of the humic and fulvic acid fractions, respectively. Also of concern are chemical

331

alterations, such as loss of ester linkages that can be envisioned to occur during extraction–isolation procedures due to the large pH swings that take place. Each of these perturbations affects the manner in which the extracted and isolated NOMS associate with AOCs. The first such example involves the missing humin fraction in the extracted–isolated NOM. As time and research on the association of AOCs with soil and sediments have progressed, the importance of the humin fraction has begun to emerge (Kohl and Rice 1998; Rice 2001, Salloum et al. 2001; Borisover and Graber 2004; Kang and Xing 2005; Wang and Xing 2005a; Bonin and Simpson 2007; Wen et al. 2007). The data from these studies have clearly shown that humin plays a major role, up to 80%, in the sorption of AOCs. Consequently, the omission of humin from sorption studies biases the results of such studies enough to raise questions regarding their exact environmental relevance and the validity of extrapolating their findings to whole solid matrices. Compounding the omission of the humin fraction is the loss of the mineral interface. Mineral-associated SMNOM is motionally restricted as it is tethered to a solid surface. This motional restriction can be envisioned to create a more rigid fraction of SMNOM, that exhibits sorption behavior for AOCs different from that of the more mobile non-surface tethered SMNOM. A number of studies showed the importance of surfacetethered SMNOM, especially focusing on to the role it plays in sorption of AOCs (Murphy et al. 1994; Jones and Tiller 1999; Salloum 2001; Chen and Xing 2005; Wang and Xing 2005b; Bonin and Simpson 2007). Loss of the native SMNOM conformation is also of great concern. It may be argued that any rearrangement within the NOM sample caused by the extraction–isolation procedure is of no consequence as any conformational changes are lost in the complexity and heterogeneity of the system. This argument falls apart when one considers that order is actually imposed by the biogeochemical processes that create the SMNOM, and is reflected in the conformation and molecular arrangements within SMNOM, as shown by Lattao et al. (2008). It is this inherent order that is lost by the isolation–extraction procedure. Accordingly, studying extracted and isolated NOM samples, especially SMNOM, provides information that, by design, is incomplete as well as altered from that obtained on whole SMNOM. Because of the analytical difficulties encountered while working with whole SMNOM, the extracted and isolated “partial” samples are still considered to be useful and yield highly insightful data, as highlighted in Section 13.4 above. Nevertheless, great caution must be exercised when extrapolating findings from these “partial” samples to real samples. 13.5.1.2. Concentration. Extracted–isolated NOM samples have allowed for the use of liquid-state NMR techniques and resulted in advances in our understanding of how AOCs

332

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

associate with NOM. These advances, however, are in the shadow of high-concentration issues in terms of both NOM and AOC concentration. Because of the inherently low sensitivity of NMR, especially when compared with other analytical techniques such as fluorescence, high NOM concentrations have traditionally been used. These high concentrations lead to a range of concerns, in particular for dissolved organic matter (DOM; e.g., NOM found in the dissolved state in the environment). Concentration issues are not as critical for the extracted–isolated SMNOM samples, as it is expected that SMNOM is not evenly distributed within a soil, creating small pools of very high concentration of SMNOM in the directly surrounding liquid phase. This leniency does, however, have limits as the loss of the solid phase leads to increased mobility and allows for interactions not possible when the SMNOM is constrained by its association with a solid surface, and it can be argued that this uneven distribution is lost as well as homogenized in the extraction–isolation processes. This leads to two major concerns, the first of which has been discussed in Section 13.5.1.1. The second is related to different intra- and intermolecular interactions and associations, including hydrogen bonding, as well as electrostatic, hydrophobic, and van der Waals interactions (for a discussion of these interactions in complex system, the reader is referred to the monograph by Karshikoff (2006)). These associations, in turn, can be envisioned to impact the manner in which AOCs can associate with NOM (Chilom and Rice 2005; Sander et al. 2006; Yang et al. 2006; Wang and Xing 2007; Lattao et al. 2008). This means that these types of associations would magnify NOM concentration, reflecting only very specialized (quite displaced from) rather than common environmental situations (Pan et al. 2008; Marwani et al. 2009) The concentration of sorbate (AOCs) required for NMR studies is also of concern. As with NOM concentration, there is large leniency in regard to the concentration of AOCs that can be studied, especially in relation to the NOM concentration. High concentration of AOCs, especially when compared to NOM, can be envisioned in a number of cases—such as spills and within the uppermost soil and immediately after the application of a pesticide. However, both of these cases are complex and difficult to mimic in the laboratory. In the spill scenario one must consider multiple phases, as in the vast majority of spill cases, the AOC will be dissolved in a solvent other than water. In the case of pesticide application, the application solution is a complex mixture consisting of the pesticide and, in all probability, a range of other constituents, including surfactants. This means that studies that utilize high concentrations of AOCs (1) are studying highly specialized situations and (2) must consider other factors, such as other phases or chemical components beyond the AOC, NOM, and water. Except for some specialized cases, the AOC is usually at very low concentrations and, for the purposes of this

discussion, an upper limit for its concentration will be set at 1000 ppm in terms of NOM (by weight). This raises a question in regard to the background signal of NOM versus the signal of a specifically labeled pollutant. The background signals for the NMR active nuclei discussed above, namely, 13 C, 15 N, 19 F, and 2 H, within NOM are approximately 5900, 85, 0, and 6 ppm, respectively [as determined by multiplying average elemental composition of the IHSS standard humic and fulvic acid samples (53.38%, 2.29%, 0%, and 4% for C, N, F, and H, respectively, with a possible slight bias away from C and H as humin is not included) by the natural abundancies of 13 C, 15 N, 19 F, and 2 H, which are 1.1%, 0.37%, 100%, and 0.015%, respectively (Harris 1987)]. Nuclear magnetic resonance is not a bulk method; as a spectroscopic method, it spreads out the carbon signal in terms of chemical shift. If one assumes, for simplicity, that 10% of NOM is composed of aromatic or aliphatic moieties, and that hydrophobic AOCs associate with only one of these moieties, then our upper limit of 1000 ppm versus total NOM is increased to 10,000 ppm in regard to either the aromatic or aliphatic moieties. Assuming that 50% of the elemental composition of NOM is carbon, this number further increases by a factor of 2 to 20,000 ppm. It can then be argued that the aromatic and aliphatic signals are further diluted as they are spread over their respective chemical shift ranges. On the other hand, the AOC’s 13 C signal will also be spread out as a result of a range of association sites within the NOM—each with a unique electronic environment, and T2 relaxation occurring as a result of the reduced mobility of AOC as well as the presence of unpaired electrons within the NOM. Thus, our upper limit for environmentally relevant AOC concentration in terms of NMR measurements versus aliphatic and aromatic moieties within an NMR experiment is 20,000 ppm, or approximately 2%, if the AOC consists only of 13 C, with is very close for completely 13 C-labeled benzene, but this an exceptional case. Usually a pollutant is labeled in only one or two positions, each of which has a unique chemical shift. If one assumes an average molecular weight of 275 Da (derived from the average molecular weight of 2,4-D, Acetochlor, Alachlor, Atrazine, Carbaryl, Chlordane, Chlorpyrifosi, Cyanazine, DDT, Deethylatrazine, Diazinon, Dieldrin, Diuron, Heptachlor epoxide, Metolachlor, Prometon, Simazine, and Tebuthiuron), then it can be approximated that one-twentieth of the AOC’s molecular weight is due to 13 C, and hence, an upper environmentally relevant limit of 1000 ppm in terms of 13 C can be set. This means that the NOM itself has close to 6 times more signal due to the background 13 C than that due to a 13 C-labeled AOC, provided each gives unique nonoverlapping 13 C NMR signals. This amounts to saying that, on an organic matter basis, in order to use 13 C-labeled compounds a concentration of 12,000 ppm is required for each unique 13 C environment within the labeled AOC to overcome the 13 C background signal of the NOM, which exceeds our

A HIGHLY CRITICAL EVALUATION OF NMR AS APPLIED TO AOC ASSOCIATION WITH NOM

upper environmentally relevant concentration by more than one order of magnitude. Similar analysis sets environmentally relevant concentrations of approximately 1090 and 145 ppm, respectively, for 15 N- and 2 H-labeled AOCs. Although these numbers are very approximate, they allow for the 15 N- and 2 H-labeled compounds to stand out against the NOM background and to be theoretically studied at the 1000-ppm upper limit of the environmentally relevant concentration. In terms of “real world” applicability, both 15 N and 2 H are limited by their low relative (vs. proton) NMR receptivity of 3.85  106 and 1.45  106, respectively, in comparison to 1.76  104 for 13 C. Neither 1 H nor 19 F have been include in the preceding discussion for very different reasons. The natural abundance of 1 H is 99.985%, which essentially means that in order to overcome the natural 1 H background signal of NOM, the AOC concentration must exceed that of NOM by at least 2, meaning that 1 H-based detection experiments are for very limited cases. In regard to 19 F, there is essentially no natural NOM background signal, hence the discussion above is of no concern. In addition, 19 F NMR has the highest receptivity (0.834) of the all the nuclei discussed, except for 1 H. These advantages of 19 F are offset by the fact that, if a AOC does not inherently contain 19 F, labeling with 19 F can cause a major perturbation in the AOC electron distribution, due to the very high electronegativity of fluorine. An example of this is hexafluorinated benzene (HFB) versus benzene. The center of the benzene ring is partially negatively charged, while the center of the HFB ring is partially positively charged as the fluorine atoms in the HFB pull electrons away from the ring’s center. This means that the radial charge distribution from the center of the ring in HFB is reverse that in benzene. In terms of associations, this means that HFB will be able to ring-stack much more effectively with the aromatic moieties within NOM compared to benzene, and hence, caution must be taken when translating results found for HFB to other aromatic pollutants. 13.5.1.3. Solvent Matrix. The limited water solubility of the vast majority of AOCs is a challenge for NMR studies because of the low sensitivity of the NMR technique. To circumvent these issues, the aqueous solvent can be augmented with organic solvents. This co-solvent system results in a new solvent matrix, which can result in very different properties. The subject of co-solvents has been addressed thoroughly within the literature, and the reader is referred to the monograph by Schwarzenbach et al. (2003). As a general rule of thumb, up to 0.1% of the solvent matrix can be organic solvent without major deviations from the anticipated properties of a purely aqueous solvent. (Chefetz et al. 2000; Kang and Xing 2005; Wang et al. 2005). Because of the high concentration demands of NMR spectroscopic techniques, a large number of studies have been done in cosolvent systems where the organic phase exceeds that 0.1% guideline by at

333

least one order of magnitude. These studies need to be viewed as representing special cases, such as a spill; however, caution must be taken when extrapolating the results to more typical environmental conditions and situations. 13.5.2. “Real World” Samples There is a desire to study sorbates that are as real-world as possible when studying association of AOCs with NOM; however, the reality is that such sorbates are extremely complex. In fact, the removal of a sorbate from the field means that it loses its real-world sorbate status. In the discussion that follows, a sample that has been removed from the field but not treated in any harsh manner, for example, that has been only air-dried, will be referred to as a “real world” sample. On the other hand, any sample that has been treated harshly, such as via a HF treatment, including isolated SMNOM, will be referred to as a “natural” surrogate. For isolated dissolved NOMs, the situation is intermediate, because of the different conditions under which this isolation can take place; however, due to the concentration step discussed above and the possible introduction of artifacts, isolated DOMs will also be referred to as real-world samples. Real-world solid matrix sorbates are complex mixtures. In SMNOM-oriented studies the remainder of the solid matrix plays a support role literally (as discussed in Section 13.5.1.1. and figuratively. The focus of this discussion will be on the figurative role and its elimination. Solid matrix sorbates consist of a range of components, which can be broken down into biological, inorganic and organic, as discussed in Section 13.2.1. As a whole, the living biological component is viewed as a nuisance in association-based studies and is eliminated with either HgCl2 or NaN3. Either treatment affects the SMNOM; however, this alteration is considered acceptable. Another nuisance within the matrix is due to its inorganic components. These components are problematic for two reasons: (1) they dilute the SMNOM, and thus, limit AOCs loading, and hence, lower the SNR of the NMR experiments; and (2) there are inorganic paramagnetic centers, which are of great concern from the NMR perspective as they contain unpaired electrons, which induce rapid relaxation pathways for nuclei in close proximity. This rapid relaxation results in major linebroadening, which also leads to a loss of SNR. The traditional method to circumvent this issue has been an HF treatment of the solid matrix sorbate (Skjemstad et al. 1994). It has been claimed that this type of treatment does not significantly modify the nature of SMNOM, however, a range of studies have questioned this assertion in terms of both the SMNOM chemical makeup (Dai and Johnson 1999; Mathers et al. 2002; Keeler and Maciel 2003; Schilling and Copper 2004; Schmidt and Gleixner 2005) and physical state (Edgebretson and Von Wandruszka, 1999; Wang and Xing 2005a). In fact, these

334

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

studies raised great concern about using HF treatments, and have been reviewed in detail by Birdwell and Cook (2008). The take-home message from this review was that HF treatment should not be used except to provide chemical shift and integration information to describe the bulk chemical character of the SMNOM within a soil by one-dimensional NMR methods, and even then, these results should be approached with caution. Considering limitations of the HF treatment-based methods, as well as any other method that removes inorganic paramagnetic centers, it maybe more appropriate to choose real-world samples with inherently low mineral content and lone electron pairs. An example of such real-world samples would be peat soil samples. This means that only a very limited number of solid matrix sorbates that can be viewed as “real world” are amenable to NMR studies, as nearly any process that increases the SMNOM content, and hence the SNR, also affects the chemical and physical nature of the NOM, and hence, the association of AOCs with it. 13.5.3. Molecular Level Information The inherent complexity of NOM also makes it very difficult to gain molecular level information when studying the AOC/NOM associations at environmentally relevant conditions. In fact, the studies discussed above that revealed some molecular level information demanded creativity in experimental design, expert execution, and brilliant interpretation. The fact remains, however, that under the definition of environmentally relevant conditions advanced above, no such study has been reported, except in highly specialized cases where very high AOC concentrations and/or cosolvent system were used. The domain-level information currently achievable for NOM is significantly behind what is considered molecular level information in the fields of biochemistry and biophysics. The more recent atomic mapping information on AOCs is a major step forward; however, it still leaves us in the dark regarding the all-important NOM side of the association. This lack of molecular level information is not due to lack of effort, originality, or technical competence, but rather a result of the complex heterogenous polydisperse nature of NOM. 13.5.4. Thermodynamic and Kinetic Data Few timecourse data have been reported in the NMR study of the AOC–NOM association. Fitting data to kinetic models reported in the literature is even more elusive (e.g., Kohl et al. 2000). The reason for this is the difficulty of obtaining sufficiently high SNR in such studies, especially at environmentally relevant conditions. Once again, the major limitations are due to the complexity of the system under study and the importance of environmentally relevant conditions.

13.6. CONCLUSIONS, FINAL THOUGHTS, AND PATHS FORWARD From the discussion above it could be argued that the study of AOC association with geosorbents by NMR has reached its limit and that no further progress can be envisioned. This is simply not true; however, new approaches as well as concepts must be adopted, while dated approaches that have become transformed into beliefs must be reevaluated and heavily modified, if not abandoned. In fact, only such an approach could have permitted the paradigm shift from the large polymer to weak molecular associations view of NOM (Sutton and Sposito 2005). This new view of NOM has allowed the community to view old data from well-executed experiments in a new light and new insights into NOM. The introduction of NMR characterization methods contributed greatly to our understanding of the complexity of NOM. In fact, NMR played a central role in displacing the predominantly aromatic viewpoint of NOM that emerged from early wet chemical analyses. These examples illustrate that change can take place and that it is sometimes required to allow us to step out of confining ideologies. One such confining ideology is the insistence on the use of “real world” or “partial” samples as the only samples complex enough to yield any useful information. This approach assumes that data from these real-world and partial samples can be extrapolated to real-world systems, but it also places a great deal of constraint on experimental design and severely limits the interpretation of the yielded data. The complexity of NOM that remains in the real-world and even in highly simplified partial samples is also the major reason why detailed molecular information, such as the exact chemical structures of the components within the NOM, is lacking. Consequently, a range of analytical techniques that have been developed for chemical and biochemical analysis (e.g., spatial mapping by NMR) cannot be applied at the moment or in the foreseeable future. Instead of prematurely attacking exceedingly complex AOC–NOM associations with sophisticated techniques—which then leads to unduly far-reaching conclusions and misconceptions—an approach of studying first well-defined (i.e., well-understood) systems with those methods seems much more prudent. The biological community has taken a range of approaches in order to address the complexity of biological systems. One major approach has been the use of surrogates, such as artificially grown and labeled proteins, synthetic peptides, and phospholipid vesicles. The use of surrogates enabled researchers to obtain a fundamental understanding apply advanced analytical methods, including NMR studies of artificially grown labeled proteins. This approach has also led to great advances in analytical instrumentation and techniques, as exemplified by advances in NMR from 1D to 2D to 3D and beyond, in order to gain an atom-level view of proteins in three-dimensional space and subsequently,

ACKNOWLEDGMENTS

obtain atom-level maps of drug associations with proteins for use in drug development. It is this success that has led to the view of NMR as such a promising method in the study of NOM, despite great differences between NOM and proteins. These differences originate from proteins having their order dictated by genetic coding, compared to NOM being much more chaotic in nature. Consequently (1) unlike for proteins, a homogenous single molecular component cannot be isolated from NOM, and even if it could, it would be of little use, due to the complex and heterogeneous molecular assembly nature of NOM; (2) it is much more difficult to label NOM with 13 C, 15 N, and/or 2 H and impossible to label specific atoms; and (3) NOM does not have a highly ordered secondary, tertiary, and quaternary structure akin to that in genetically programmed proteins. On aggregate, when leveraging the advances and methods of biophysical NMR, it must be kept in mind that the majority of biophysical NMR was designed around the ability to isolate pure proteins, isotopically label precise atomic sites within proteins, and/ or take advantage of the higher-order nature of proteins. This does not mean that one cannot apply and learn from biophysical NMR studies when designing NMR approaches for the study of NOM; however, it must be done with caution. One such lesson is in regard to the value of surrogates as model systems. The shift from a polymer to a molecular assembly perspective for NOM is a major step forward from which a range of chemically engineered surrogates could be developed. A possibility would be to assemble model systems from combinations of single—but not overly simple— alkyl, aromatic, and O-alkyl moieties, both free and tethered, to various mineral surfaces, such as silica and aluminum minerals, as well as clays. Kleber et al. (2007) have argued in the form of a thought experiment that SMNOM might assemble in such a manner. The work of Lattao et al. (2008) would suggest that such an argument may be valid for whole soils, as does the work of Lehmann et al. (2007). This means that these types of samples would be viewed as chemically engineered surrogates that echo our current understanding of SMNOM, and as models on which to develop new or modify existing analytical approaches, especially in terms of gaining insight at the molecular and atomic levels. This type of information is essential to the advancement of our understanding of how AOCs interact with SMNOM, and NOM as a whole, and hence, how they impact the environment. Because of the complexity of solid matrix sorbates and NOM, working with model systems is required in order to gain fundamental insights. This demands the use of surrogate systems, including artificial chemically well-defined systems, as discussed above. Without such an approach we will not be able to progress to the level of understanding necessary to address environmental problems analogous to our current understanding of biological complex systems. It can be argued that the use of such chemically well-defined engineered soil surrogates constitutes an essential step toward

335

developing and applying advanced molecular level analytical methods to solid-state sorbates and under more environmentally relevant conditions than those afforded by altering the native samples to make them amenable to the analysis by the molecular level analytical methods. Moreover, the potential insights gained on the physical properties of organic matter from studies utilizing soil surrogates would be useful in studies, at the molecular and atomic levels of binding of a range of chemicals in soils, including trace elements. It should be pointed out that, although batch and other non-molecular level sorption-based approaches have allowed for a great amount of insight into the association of AOCs with SMNOM, they provided little molecular information. Nuclear magnetic resonance has been viewed as an analytical method that has the potential to deliver this highly elusive molecular information; thus this chapter has presented a critical evaluation of the application of NMR methods to the association of AOCs to SMNOM. From this assessment it can be seen that, while the chosen examples of such studies demonstrate very creative experimental design, expert execution and brilliant interpretation considering the complexity of the system under study, the following underlying issues and problems persist: 1. The complex heterogeneous polydisperse nature of environmental sorbates 2. The environmentally relevant low concentrations of the sorbent As steps are taken to mitigate their impacts, the studies probe systems quite displaced from typical environmental conditions. This then leads to questions as to (1) the exact environmental relevance of the data obtained and (2) what data, and hence what information are being lost or are not attainable. Similar issues plagued biochemistry and molecular biology fields prior to the use of chemically well-defined model systems to understand complex living systems. It would only seem logical that we draw from the that experience, cast aside our desire to work only on “real world” or partial samples and start to investigate, and leverage chemically well-defined surrogates systems. The use of such systems has the potential to address and overcome the range of problematic issues discussed in this chapter and create new opportunities for gaining molecular level information at the atomic level—including kinetic. ACKNOWLEDGMENTS This material is based on work supported by the National Science Foundation under Grant CHE-0547982 and the United Sates Department of Agriculture under Grants CSRESS 2004-03674, 2009-35201-05819, and 200965107-05926. Dr. Elzbieta Cook is also thanked for fresh eyes whenever needed.

336

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

REFERENCES Abragam, A. (1961), Principles of Nuclear Magnetism, Oxford University Press, New York. Accardi-Dey, A. and Gschwend, P. M. (2002), Assessing the combined roles of natural organic matter and black carbon as sorbents in sediments, Environ. Sci. Technol. 36, 21–29. Ahmad, R., Kookana, R. S., Alston, A. M., and Skjemstad, J. O. (2001), The nature of soil organic matter affects sorption of pesticides. 1. Relationships with carbon chemistry as determined by 13 C CPMAS NMR spectroscopy, Environ. Sci. Technol. 35, 878–884. Bayer, J. V. and Schaumann, G. E. (2007), Development of soil water repellency in the course of isothermal drying and upon pH change in two urban soils, Hydrol. Process. 21, 2266–2275. Belliveau, S. M., Henselwood, T. L., and Langford, C. H. (2000), Soil wetting processes studied by magnetic resonance imaging: Correlated study of contaminant uptake, Environ. Sci. Technol. 34, 2439–2445. Birdwell, J. E. and Cook, R. L. (2008), Studying organic matter and water in soils by NMR: Cautions and insights, in Natural Organic Matter and Its Significance in the Environment, Wu, F. and Xing, B., eds., Science Press, Beijing, 105–132. Bonin, J. L. and Simpson, M. J. (2007), Variation in phenanthrene sorption coefficients with soil organic matter fractionation: The result of structure or conformation? Environ. Sci. Technol. 41, 153–159. Borisover, M. and Graber, E. R. (2004), Hydration of natural organic matter: Effect on sorption of organic compounds by humin and humic fractions vs original peat, Environ. Sci. Technol. 38, 4120–4129. Cardoza, L. A., Korir, A. K., Otto, W. H., Wurrey, C. J., and Larive, C. K. (2004), Applications of NMR spectroscopy in environmental science, Prog. Nucl. Magn. Reson. Spectrosc. 45, 209–238. Castro, C. E., O’Shea, S. K., Wang, W., and Bartnicki, E. W. (1996), 13 C-NMR reactivity probes for the environment, Environ. Sci. Technol. 30, 1185–1191. Cavanagh, J., Fairbrother, W. J., Palmer, A. G., and Skelton, N. J. (1996), Protein NMR Spectroscopy: Principles and Practice, Academic Press. San Diego. Chen, Z., Xing, B., McGill, W. B., and Dudas, M. J. (1996), a-Naphthol sorption as regulated by structure and composition of organic substances in soils and sediments, Can. J. Soil Sci. 76, 513–522. Chen, L. and Xing, B. (2005), Sorption and conformational characteristics of reconstituted plant cuticular waxes on montmorillonite, Environ. Sci. Technol. 39, 8315–8323. Chefetz, B., Deshmukh, A. P., Hatcher, P. G., and Guthrie, E. A. (2000), Pyrene sorption by natural organic matter, Environ. Sci. Technol. 34, 2925–2930. Chefetz, B. (2007), Decomposition and sorption characterization of plant cuticles in soil, Plant Soil 298, 21–30. Chien, Y.-Y. and Bleam, W. F. (1997), Fluorine-19 nuclear magnetic resonance study of atrazine in humic and sodium dodecyl sulfate

micelles swollen by polar and nonpolar solvents, Langmuir 13, 5283–5288. Chien, Y.-Y, Kim, E.-U., and Bleam, W. F. (1997), Paramagnetic relaxation of atrazine solubilized by humic micellar solutions, Environ. Sci. Technol. 31, 3204–3208. Chilom, G. and Rice, J. A. (2005), Glass transition and crystallite melting in natural organic matter, Org. Geochem. 36, 1339–1346. Chin, Y.-P., Aiken, G. R., and Danielsen, K. M. (1997), Binding of pyrene to aquatic and commercial humic substances: The role of molecular weight and aromaticity, Environ. Sci. Technol. 31, 1630–1635. Chiou, C. T., Peters, L. J., and Freed, V. H. (1979), A physical concept of soil-water equilibriums for nonionic organic compounds, Science 206, 831–832. Chiou, C. T., Porter, P. E., and Schmedding, D. W. (1983), Partition equilibriums of nonionic organic compounds between soil organic matter and water, Environ. Sci. Technol. 17, 227–231. Chiou, C. T., McGroddy, S. E., and Kile, D. E. (1998), Partition characteristics of polycyclic aromatic hydrocarbons on soils and sediments, Environ. Sci. Technol. 32, 264–269. Chiou, C. T. and Kile, D. E. (1998), Deviations from sorption linearity on soils of polar and nonpolar organic compounds at low relative concentrations, Environ. Sci. Technol. 32, 338–343. Chiou, C. T., Kile, D. E., Rutherford, D. W., Sheng, G., and Boyd, S. A. (2000), 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. 34, 1254–1258. Claridge, T. D. W. (1999), High-Resolution NMR Techniques in Organic Chemistry, Elsevier, San Diego. Conte, P., Spacchi, R., and Piccolo, A. (2004), State of the art of CPMAS 13 C-NMR spectroscopy applied to natural organic matter, Prog. Nucl. Magn. Reson. Spectrosc. 44, 215–223. Cook, R. L. (2004), Coupling NMR to NOM, Anal. Biochem. 378, 1484–1503. Cornelissen, G. P. van Noort, P. C. M., Nachtegaal, G., and Kentgens, A. P. M. (2000), A solid-state fluorine-NMR study on hexafluorobenzene sorbed by sediments, polymers, and active carbon, Environ. Sci. Technol. 34, 645–649. Dai, K. H. and Johnson, C. E. (1999), Applicability of solid-state 13 C CP/MAS NMR analysis in spodosols: Chemical removal of magnetic materials, Geoderma 93, 289–310. Deur, M. J. (2004), Introduction to Solid-State NMR Spectroscopy, Blackwell Publishing, Oxford. Diehl, D. and Schaumann, G. E. (2007), The nature and wetting on urban soil samples: Wetting kinetics and evaporation assessed from sessile drop shape, Hydrol. Process. 21, 2255–2265. DiVerdi, J. A., Kobayashi, T., and Maciel, G. E. (2007), Molecular dynamics of pyridine adsorbed on the silica surface, J. Phys. Chem. 111, 5982–5989. Dixon, A. M., Mai, M. A., and Larive, C. K. (1999), NMR investigation of the interactions between 40 -fluoro-10 -acetonaphthone and the Suwannee River fulvic acid, Environ. Sci. Technol. 33, 958–964.

REFERENCES

Edgebretson, R. R. and Von Wandruszka, R. (1999), Effects of humic acid purification on interactions with hydrophobic organic matter: Evidence from fluorescence behavior, Environ. Sci. Technol. 33, 4299–4303. Emery, E. F., Junk, T., Ferrell, R. E., De Hon, R., and Butler, L. G. (2001), Solid-state 2 H MAS NMR studies of TNT absorption in soil and clays, Environ. Sci. Technol. 35, 2973–2978. Ernst, R. R., Bodenhausen, G., and Wokaun, A. (1987), Principles of Nuclear Magnetic Resonance in One and Two Dimensions, Oxford University Press, New York. Evans, J. N. S. (1995), Biomolecular NMR Spectroscopy, Oxford University Press, New York. Grathwohl, P. (1990), Influence of organic matter from soils and sediments from various origins on the sorption of some chlorinated aliphatic hydrocarbons: Implications on Koc correlations, Environ. Sci. Technol. 24, 1687–1693. Gunasekara, A. S. and Xing, B. (2003), Sorption and desorption of naphthalene by soil organic matter: Importance of aromatic and aliphatic components, J. Environ. Qual. 32, 240–246. Gunasekara, A. S., Simpson, M. J., and Xing, B. (2003), Identification and characterization of sorption domains in soil organic matter using structurally modified humic acids, Environ. Sci. Technol. 37, 852–858. Guthrie, E. A., Bortiatynski, J. M., van Heemst, J. D. H., Richman, J. E., Hardy, K. S., Kovach, M., and Hatcher, P. G. (1999), Determination of [13 C] pyrene sequestration in sediment microcosms using flash pyrolysis-GC-MS and 13 C NMR, Environ. Sci. Technol. 33, 119–125. Harris, R. K. (1987), Nuclear Magnetic Resonance Spectroscopy, Longman, Harlow. Hatcher, P. G., Boriatynski, J. M., Minard, R. D., Dec, J., and Bollag, J.-M. (1993), Use of high-resolution carbon-13 NMR to examine the enzymatic covalent binding of carbon-13-labeled 2,4-dichlorophenol to humic substances, Environ. Sci. Technol. 27, 2098–2103. Hu, W.-G., Mao, J., Xing, B., and Schmidt-Rohr, K. (2000), Poly (methylene) crystallites in humic substances detected by nuclear magnetic resonance, Environ. Sci. Technol. 34, 530–534. Huang, W., Young, T. M., Schlautman, M. A., Yu, H., and Weber Jr., W. J. (1997), A distributed reactivity model for sorption by soils and sediments. 9. General isotherm nonlinearity and applicability of the dual reactive domain model, Environ. Sci. Technol. 31, 1703–1710. Huang, W. and WeberJr., W. J. (1997), 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. 31, 2562–2569. Hurrass, J. and Schaumann, G. E. (2006), Properties of soil organic matter and aqueous extracts of actual water repellent and wettable soil samples, Geoderma 132, 222–239. Jacobsen, N. E. (2007), NMR Spectroscopy Explained; Simplified Theory, Applications and Examples for Organic Chemistry and Structural Biology, Wiley-Interscience, New York. Janusa, M. A., Xiao, W., Cartledge, F. K., and Butler, L. G. (1993), Solid-state deuterium NMR spectroscopy of d5-phenol in white

337

Portland cement: A new method for assessing solidification/ stabilization, Environ. Sci. Technol. 27, 1426–1433. Jayasundera, S., Schmidt, W. F., Hapeman, C. J., and Torrents, A. (2003), Examination of molecular interaction sites of acetanilides with organic matter surrogates using nuclear magnetic resonance techniques, J. Agric. Food Chem. 51, 3829–3835. Jones, K. D. and Tiller, C. L. (1999), 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. 33, 580–587. Jurkiewicz, A. and Maciel, G. E. (1995), Solid-state 13 C NMR of the interaction of acetone, carbon tetrachlorine and trichloroethylene with soil components, Sci. Total Environ. 164, 195–202. K€acker, T., E. Haupt, T. K., Garms, C., Francke, W., and Steinhart, H. (2002), Structural characterization of humic acid-bound PAH residues in soil by 13 C-CPMAS-NMR-spectroscopy: Evidence of covalent bonds, Chemosphere 48, 117–131. Kang, S. and Xing, B. (2005), Phenanthrene sorption to sequentially extracted soil humic acids and humin, Environ. Sci. Technol. 39, 134–140. Kang, S. and Xing, B. (2008), Humic acid fractionation upon sequential adsorption onto geothite, Langmuir 24, 2525–2531. Karapanagioti, H. K., Kleineidam, K. S., Sabatini, D. A., Grathwohl, P., and Ligouis, B. (2003), Impacts of heterogeneous organic matter on phenanthrene sorption: Equilibrium and kinetic studies with aquifer material, Environ. Sci. Technol. 34, 406–414. Karickhoff, S.W., Brown, D. S., and Scott, T. A. (1979), Sorption of hydrophobic pollutants on natural sediments, Water Res. 13, 241–248. Karickhoff, S. W. (1984), Organic pollutants sorption in aquaticsystems, J. Hydraul. Eng. ASCE 110, 707–735. Karshikoff, A. (2006), Non-covalent Interactions in Proteins, Imperial College Press, London. Keeler, C. and Maciel, G. E. (2003), Quantitation in the solid-state 13 C NMR analysis of soil and organic soil fractions, Anal. Chem. 75, 2421–2432. Khalaf, M., Kohl, S. D., Klumpp, E., Rice, J. A., and Tombacz, E. (2003), Comparison of sorption domains in molecular weight fractions of a soil humic acid using solid-state 19 F NMR, Environ. Sci. Technol. 37, 2855–2860. Kile, D. E., Wershaw, R. L., and Chiou, C. T. (1999), Correlation of soil and sediment organic matter polarity to aqueous aorption of nonionic compounds, Environ. Sci. Technol. 33, 2053–2056. Kleber, M., Sollins, P., and Sutton, R. (2007), A conceptual model of organo-mineral interactions in soils; self-assembly of organic molecular fragments into zonal structures on mineral surfaces, Biogechemistry 85, 9–24. Kleineidam, S., Sch€ uth, C., and Grathwohl, P. (2002), Solubilitynormalized combined adsorption-partitioning sorption isotherms for organic pollutants, Environ. Sci. Technol. 36, 4689–4697. Knicker, H., Bruns-Nagel, D., Drzyzga, O., von L€ ow, E., and Steinbach, K. (1999), Characterization of 15 N-TNT residues after an anaerobic/aerobic treatment of soil/molasses mixtures

338

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

by solid-state 15 N NMR spectroscopy. 1. Determination and optimization of relevant NMR spectroscopic parameters, Environ. Sci. Technol. 33, 343–349. Knicker, H., Achtich, C., and Lenke, H. (2001), Solid-state nitrogen-15 nuclear magnetic resonance analysis of biologically reduced 2,4,6-trinitrotoluene in a soil slurry remediation, J. Environ. Qual. 30, 403–410. Knicker, H. (2003), Incorporation of 15 N-TNT transformation products into humifying plant organic matter as revealed by one- ad two-dimensional solid state NMR spectroscopy, Sci. Total Environ. 308, 211–220. Kohl, S. D. and Rice, J. A. (1998), The binding of contaminants to humin: A mass balance, Chemosphere 36, 251–261. Kohl, S. D., Toscano, P. J., Hou, W., and Rice, J. A. (2000), Solidstate 19 F NMR investigation of hexafluorobenzene sorption to soil organic matter, Environ. Sci. Technol. 34, 204–210. Ladd, C. (2008), Introduction to Physical Chemistry, 3rd ed., Cambridge University Press, Cambridge, UK. Laor, Y., and Rebhun, M. (2002), Evidence for nonlinear binding of PAHs [polycyclic aromatic hydrocarbons] to dissolved humic acids, Environ. Sci. Technol. 36, 955–961. Lattao, C. V., Birdwell, J. E., Wang, J. J., and Cook, R. L. (2008), Studying organic matter molecular assemblage within a whole soil by NMR, J. Environ. Qual. 37, 1501–1509. LeBoeuf, E. J. and Weber Jr., W. J. (2000a), Macromolecular characteristics of natural organic matter. 1. Insights from glass transition and enthalpic relaxation behavior, Environ. Sci. Technol. 34, 3623–3631. LeBoeuf, E. J. and Weber Jr., W. J. (2000b), Macromolecular characteristics of natural organic matter. 2. Sorption and desorption behavior, Environ. Sci. Technol. 34, 3632–3640. Lehmann, J., Kinyangi, J., and Solomon, D. (2007), Organic matter stabilization in soil microaggregates: Implications from spatial heterogeneity of organic carbon contents and carbon forms, Biogeochemistry 85, 45–57. Levitt, M. H. (2008), Spin Dynamics: Basics of Nuclear Magnetic Resonance, 2nd ed., Wiley, Chichester, UK. Litvina, M., Todoruk, T. R., and Langford, C. H. (2003), Composition and structure of agents responsible for development of water repellency in soils following oil contamination, Environ. Sci. Technol. 37, 2883–2888. Luthy, R. G., Aiken, G. R., Brusseau, M. L., Cunningham, S. D., Gschwend, P. M., Pignatello, J. J., Reinhard, M., Traina, S., Weber Jr., W. J., and Westall, J. C. (1997), Sequestration of hydrophobic organic contaminants by geosorbents, Environ. Sci. Technol. 31, 3341–3347. Mao, J.-D., Hundal, L. S., Thompson, M. L., and Schmidt-Rohr, K. (2002), Correlation of poly(methylene)-rich amorphous aliphatic domains in humic substances with sorption of a nonpolar organic contaminant, phenanthrene, Environ. Sci. Technol. 36, 929–936. Marwani, H. M., Lowry, M., Xing, B., Warner, I. M., and Cook, R. L. (2009), Frequency-domain fluorescence lifetime measurements via frequency segmentation and recombination as applied to pyrene with dissolved humic materials, J. Fluor. 19, 41–51.

Ma’shum, M. and Farmer, V. C. (1985), Origin and assessment of water repellency of a sandy South Australian soil, Austral. J Soil Sci. 23, 623–626. Mathers, N. J., Xu, Z. H., Berners-Price, S. J., Perera, M. C. S., and Saffigna, P. G. (2002), Hydrofluoric acid pre-treatment for improving 13 C CPMAS NMR spectra quality of forest soils in southeast Queensland, Australia, Austral. J. Soil Res. 40, 655–674. Matsui, Y., Knappe, D. R. U., and Takagi, R. (2002a), Pesticide adsorption by granular activated carbon adsorbers. 1. Effect of natural organic matter preloading on removal rates and model simplification, Environ. Sci. Technol. 36, 3426–3431. Matsui, Y., Knappe, D. R. U., Iwaki, K., and Ohira, H. (2002b), Pesticide adsorption by granular activated carbon adsorbers. 2. Effects of pesticide and natural organic matter characteristics on pesticide breakthrough curves, Environ. Sci. Technol. 36, 3432–3438. Mopper, K., Stubbins, A., Ritchie, J. D., Bialk, H. M., and Hatcher, P. G. (2007), Advanced instrumental approaches for characterization of marine dissolved organic matter: Extraction techniques, mass spectrometry, and nuclear magnetic resonance spectroscopy, Chem. Rev. 107, 419–442. Murphy, E. M., Zachara, J. M., Smith, S. C., Phillips, J. L., and Wietsma, T. W. (1994), Interaction of hydrophobic organic compounds with mineral bound humic substances, Environ. Sci. Technol. 28, 1291–1299. Nanny, M. A. (1999), Deuterium NMR characterization of noncovalent interactions between monoaromatic compounds and fulvic acids, Org. Geochem. 30, 901–909. Nanny, M. A. and Maza, J. P. (2001), Noncovalent interactions between monoaromatic compounds and dissolved humic acids: A deuterium NMR T1 relaxation study, Environ. Sci. Technol. 35, 379–384. Pan, B., Xing, B., Liu, W., Xing, G., and Tao, S. (2007), Investigating interactions of phenanthrene with dissolved organic matter: Limitations of Stern-Volmer plot, Chemosphere 69, 1555–1562. Pan, B., Ghosh, S., and Xing, B. S. (2007), Nonideal binding between dissolved humic acids and polyaromatic hydrocarbons, Environ. Sci. Technol. 41, 6472–6478. Pan, B., Ghosh, S., and Xing, B. (2008), Dissolved organic matter conformation and its interaction with pyrene as affected by water chemistry and concentration, Environ. Sci. Technol. 42, 1594–1599. Penning de Vries, F. W. T., Rabbinge, R., and Groot, J. J. R. (1997), Potential and attainable food production and food security in different regions, Philos. Trans. Roy. Soc. Lond. B 352, 917–928. Perminova, I. V., Grechishcheva, N. Y., and Petrosyan, V. S. (1999), Relationships between structure and binding affinity of humic substances for polycyclic aromatic hydrocarbons: Relevance of molecular descriptors, Environ. Sci. Technol. 33, 3781–3787. Pignatello, J. J., Lu, Y., LeBoeuf, E. J., Huang, W., Song, J., and Xing, B. (2006a), Nonlinear and competitive sorption of apolar compounds in black carbon-free natural organic materials, J. Environ. Qual. 35, 1049–1059. Pignatello, J. J., Kwon, S., and Lu, Y. (2006b), Effect of natural organic substances on the surface and adsorptive properties of

REFERENCES

environmental black carbon (char): Attenuation of surface activity by humic and fulvic acids, Environ. Sci. Technol. 40, 7757–7763. Ran, Y., Huang, W., Rao, P. S. C., Liu, D., Sheng, G., and Fu, J. (2002), The role of condensed organic matter in the nonlinear sorption of hydrophobic organic contaminants by a peat and sediments, J. Environ. Qual. 31, 1953–1962. Rice, J. A. (2001), Humin, Soil Sci. 166, 848–857. Roy, C., Gaillardon, P., and Montfort, F. (2000), The effects of soil moisture content on the sorption of five sterol biosynthesis inhibiting fungicides as a function of their physicochemical properties, Pest Manage. Sci. 56, 795–803. R€ ugner, H., Kleineidam, S., and Grathwohl, P. (1999), Long term sorption kinetics of phenanthrene in aquifer materials, Environ. Sci. Technol. 33, 1645–1651. Rutherford, D. W., Chiou, C. T., and Kile, D. E. (1992), Influence of soil organic matter composition on the partition of organic compounds, Environ. Sci. Technol. 26, 336–340. Sachleben, J. R., Chefetz, B., Deshmukh, A., and Hatcher, P. G. (2004), Solid-state NMR characterization of pyrenecuticular matter interactions, Environ. Sci. Technol. 38, 4369–4376. Salloum, M. J., Dudas, M. J., and McGill, W. B. (2001), Variation of 1-naphthol sorption with organic fraction: The role of physical conformation, Org. Geochem. 32, 709–719. Salloum, M. J., Chefetz, B., and Hatcher, P. G. (2002), Phenanthrene sorption by aliphatic-rich natural organic matter, Environ. Sci. Technol. 36, 1953–1958. Sander, M., Lu, Y., and Pignatello, J. J. (2006), Conditioningannealing studies of natural organic matter solids linking irreversible sorption to irreversible structural expansion, Environ. Sci. Technol. 40, 170–178. Schilling, M. and Cooper, W. T. (2004), Effects of chemical treatments on the quality and quantitative reliability of solidstate C-13 NMR spectroscopy of mineral soils, Anal. Chim. Acta. 508, 207–216. Schmidt, M. W. I., and Gleixner, G. (2005), Carbon and nitrogen isotope composition of bulk soils, particle fractions and organic material after treatment with hydrofluoric acid, Eur. J. Soil Sci. 56, 407–416. Schmidt-Rohr, K. and Spiess, H. W. (1994), Multidimensional Solid-State NMR and Polymers, Academic Press, San Diego. Schwartzenbach, R. P. and Westall, J. (1981), Transport of nonpolar organic compounds from surface water to groundwater. Laboratory sorption studies, Environ. Sci. Technol. 15, 1360–1367. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (2003), Environmental Organic Chemistry, 2nd Ed., Wiley -Interscience, New York. Seth, R., Mackay, D., and Muncke, J. (1999), Estimating the organic carbon partition coefficient and its variability for hydrophobic chemicals, Environ. Sci. Technol. 33, 2390–2394. Shirzadi, A., Simpson, M. J., Xu, Y., and Simpson, A. J. (2008a), Application of saturated transfer double difference NMR to

339

elucidate the mechanistic interactions of pesticides with humic acid, Environ. Sci. Technol. 42, 1084–1090. Shirzadi, A., Simpson, M. J., Kumar, R., Baer, A. J., Xu, Y., and Simpson, A. J. (2008b), Molecular interactions of pesticides at the soil-water interface, Environ. Sci. Technol. 42, 5514–5520. Simpson, M. J., Chefetz, B., and Hatcher, P. G. (2003), Phenanthrene sorption to structurally modified humic acids, J. Environ. Qual. 32, 1750–1758. Simpson, M. J., Simpson, A. J., and Hatcher, P. G. (2004), Noncovalent interactions between aromatic compounds and dissolved humic acid examined by nuclear magnetic resonance spectroscopy, Environ. Toxicol. Chem. 23, 355–362. Simpson, M. J., Simpson, A. J., Gross, D., Spraul, M., and Kingerly, W. L. (2007), 1 H and 19 F nuclear magnetic resonance microimaging of water and chemical distribution in soil columns, Environ. Toxicol. Chem. 26, 1340–1348. Skjemstad, J. O., Clark, P., Taylor, J. A., Oades, J. M., and Newman, R. H. (1994), The removal of magnetic materials from surface soils; a solid state 13 C CP/MAS NMR study, Austral. J. Soil Res. 32, 1215–1229. Smernik, R. J. (2005), A new way to use solid-state carbon-13 nuclear magnetic resonance spectroscopy to study the sorption of organic compounds in soil organic matter, J. Environ. Qual. 43, 1194–1204. Strynar, M., Dec, J., Benesi, A., Jones, A. D., and Bollag, J.-M. (2004), Using 19 F NMR spectroscopy to determine trifluralin binding to soil, Environ. Sci. Technol. 38, 6645–6655. Sutton, R. and Sposito, G. (2005), Molecular structure in soil humic substances: The new view, Environ. Sci. Technol. 39, 9009–9015. Tao, T., Yang, J., and Maciel, G. E. (1999), Photoinduced decomposition of trichloroethylene on soil components, Environ. Sci. Technol. 33, 74–80. Teng, Q. (2005), Structural Biology: Practical NMR Applications, Springer, New York. Thorn, K. A., Arterburn, J. B., and Mikita, M. A. (1992), Nitrogen15 and carbon-13 NMR investigation of hydroxylamine-derivatized humic substances, Environ. Sci. Technol. 26, 107–116. Thorn, K. A., Pettigrew, P. J., Goldenberg, W. S., and Weber, E. J. (1996), Covalent binding of analine to humic substances. 2. 15 N NMR studies of nucleophilic additions reaction, Environ. Sci. Technol. 30, 2764–2775. Thorn, K. A. and Kennedy, K. R. (2002), 15 N NMR investigation of the covalent binding of reduced TNT amines to soil humic acid, model compounds, and lignocellulose, Environ. Sci. Technol. 36, 3787–3796. Thorn, K. A., Pennington, J. C., and Hayes, C. A. (2002), 15 N NMR investigation of the reduction and binding of TNT in an aerobic bench scale reactor simulating windrow composting, Environ. Sci. Technol. 36, 3797–3805. Thorn, K. A., Thorne, P. G., and Cox, L. G. (2004), Alkaline hydrolysis/polymerization of 2,4,6-trinitrotoluene: Characterization of products by 13 C and 15 N NMR, Environ. Sci. Technol. 38, 2224–2231.

340

NMR APPLICATION IN ENVIRONMENTAL RESEARCH ON ANTHROPOGENIC ORGANIC COMPOUNDS

Thorn, K. A., Pennington, J. C., Kennedy, K. R., Cox, L. G., Hayes, C. A., and Porter, B. E. (2008), N-15 NMR study of the immobilization of 2, 4- and 2, 6-dinitrotoluene in aerobic compost, Environ. Sci. Technol. 42, 2542–2550. Todoruk, T. R., Langford, C. H., and Kantzas, A. (2003), Pore-scale redistribution of water during wetting of air-dried soils as studied by low-field NMR relaxometry, Environ. Sci. Technol. 37, 2707–2713. UNEP (2007), Global Environment Outlook (GEO-4): Environment for Development, United Nations Environment Programme, Nairobi. Valat, B., Jouany, C., and Riviere, P. L. M. (1991), Characterization of the wetting properties of air-dried peats and composts, Soil Sci. 152, 100–107. Van-Camp, L., Bujarrabal, B., Gentile, A.-R., Jones, R. J. A., Montanarella, L., Olazabal, C., and Selvaradjou, S.-K. (2004), Reports of the Technical Working Groups Established Under the Thematic Strategy for Soil Protection, EUR 21319 EN/4, Office for Official Publications of the European Communities, Luxembourg. Viel, S., Mannina, L., and Segre, A. (2002), Detection of a p–p complex by diffusion-ordered spectroscopy (DOSY), Tetrahedron Lett. 43, 2515–2519. Wais, A., Haider, K., Spiteller, M., de Graaf, A. A., Burauel, P., and Fuehr, F. (1995), Using 13 C-NMR spectroscopy to evaluate the binding mechanism of bound pesticide residues in soils. I. Solution high resolution NMR spectroscopy, J. Environ. Sci. Health B B30 1–25. Wang, X., Sato, T., and Xing, B. (2005), Sorption and displacement of pyrene in soils and sediments, Environ. Sci. Technol. 39, 8712–8718. Wang, K. and Xing, B. (2005a), Chemical extractions affect the structure and phenanthrene sorption of soil humin, Environ. Sci. Technol. 39, 8333–8340. Wang, K. and Xing, B. (2005b), Structural and sorption characteristics of adsorbed humic acids on clay, J. Environ. Qual. 34, 342–349. Wang, X. L., Cook, R. L., Tao, S., and Xing, B. (2007), Sorption of organic contaminants by biopolymers: Role of polarity, structure and domain spatial arrangement, Chemosphere 66, 1476–1484. Wang, X. L. and Xing, B. (2007), Roles of acetone-conditioning and lipid in sorption of organic contaminants, Environ. Sci. Technol. 41, 5731–5737. Weber Jr., W. J. and Huang, W. (1996), A distributed reactivity model for sorption by soils and sediments. 4. Intraparticle heterogeneity and phase-distribution relationships under nonequilibrium conditions, Environ. Sci. Technol. 30, 881–888. Weber Jr., W. J. and Huang, W. (1999), Weber and Huang’s comment on “evaluation of the glassy/rubbery model for soil organic matter,” Environ. Sci. Technol. 33, 2829–2830. Wen, B., Zhang, J.-J., Zhang, S.-Z., Shan, Z.-Q., Khan, S. U., and Xing, B. (2007), Phenanthrene sorption to soil humic acid and different humin fractions, Environ. Sci. Technol. 41, 3165–3171. Wijnja, H., Pignatello, J. J., and Kalumbu, M. (2004), Formation of p–p complexes between phenanthrene and model p-acceptor humic subunits, J. Environ. Qual. 33, 265–275.

Witte, E. G., Philipp, H., and Vereecken, H. (1998), Binding of 13 C-labelled 2-aminobenzothiazole to humic acid as derived from 13 C NMR spectroscopy, Org. Geochem. 29, 1829–1835. Witte, E. G., Philipp, H., and Vereecken, H. (2002), Study of enzyme-catalysed and noncatalysed interactions between soil humic acid and 13 C-labelled 2-aminobenzothiazole using solidstate 13 C NMR spectroscopy, Org. Geochem. 33, 1727–1735. WWF -(World Wide Fund for Nature) (2008), Living Plant Report 2008, WWF, Gland, Switzerland. Xing, B., McGill, W. B., and Dudas, M. J. (1994a), Sorption of a-naphthol onto organic sorbents varying in polarity and aromaticity, Chemosphere 28, 145–153. Xing, B., McGill, W. B., and Dudas, M. J. (1994b), Cross-correlation of polarity curves to predict partition coefficients of nonionic organic contaminants, Environ. Sci. Technol. 28, 1929–1933. Xing, B., Pignatello, J. J., and Gigliotti, B. (1996), Competitive sorption between atrazine and other organic compounds in soils and model sorbents, Environ. Sci. Technol. 30, 2432–2440. Xing, B. (1997), The effect of the quality of soil organic matter on sorption of naphthalene, Chemosphere 35, 633–642. Xing, B. and Pignatello, J. J. (1997), Dual-mode sorption of lowpolarity compounds in glassy polyvinychloride and soil organic matter, Environ. Sci. Technol. 31, 792–799. Xing, B. and Chen, Z. (1999), Spectroscopic evidence for condensed domains in soil organic matter, Soil Sci. 164, 40–47. Xing, B. S. (2001) Sorption of naphthalene and phenanthrene by soil humic acids, Environ. Pollut. 111, 303–309. Xiong, J. and Maciel, G. E. (1999), Deuterium NMR studies of local motions of benzene adsorbed on Ca-montmorillonite, J. Phys. Chem. 103, 5543–5549. Xiong, J., Lock, H., Chuan, I. -S, Keeler, C., and Maciel, G. E. (1999), Local motions of organic pollutants in soil components, as studied by 2 H NMR, Environ. Sci. Technol. 33, 2224–2233. Yang, K., Zhu, Z., and Xing, B. (2006), Enhanced soil washing of phenanthrene by mixed solutions of TX100 and SDBS, Environ. Sci. Technol. 40, 4274–4280. Young, K. D. and Leboeuf, E. J. (2000), Glass transition behavior in a peat humic acid and an aquatic fulvic acid, Environ. Sci. Technol. 34, 4549–4553. Zhang, L., LeBoeuf, E. J., and Xing, B. (2007), Thermal analytical investigation of biopolymers and humic- and carbonaceousbased soil and sediment organic matter, Environ. Sci. Technol. 41, 4888–4894. Zhu, D., Herbert, B. E., and Schlautman, S. A. (2003a), Molecularlevel investigation of monoaromatic compound sorption to suspended soil particles by deuterium nuclear magnetic resonance, J. Environ. Qual. 32, 232–239. Zhu, D., Herbert, B. E., and Schlautman, M. A. (2003b), Sorption of pyrine to suspended soil particles studied by deuterium nuclear magnetic resonance, Soil Sci. Soc. Am. J. 67, 1370–1377. Zhu, D., Hyun, S., Pignatello, J. J., and Lee, L. S. (2004), Evidence for p-p electron donor-acceptor interactions between p-donor aromatic compounds and p-acceptor sites in soil organic matter through pH effects on sorption, Environ. Sci. Technol. 38, 4361–4368.

14 SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES OF ORGANIC CONTAMINANTS IN THE ENVIRONMENT JOHN R. LAWRENCE

AND

ADAM P. HITCHCOCK

14.1. Introduction 14.2. Scanning Transmission X-ray Microscopy (STXM) 14.2.1. Instrumentation and Access to STXM 14.2.2. Sample Preparation 14.2.3. Imaging and Data Collection 14.2.4. Quantification 14.2.5. Correlative Microscopy 14.2.6. Examples of STXM Applied to Organic Environmental Contaminants 14.2.7. Future Developments in STXM 14.2.8. Summary and Conclusions Related to STXM 14.3. Synchrotron-Based Fourier Transform Infrared Microspectroscopy 14.3.1. Infrared Spectroscopy 14.3.2. Examples of the Application of IR Microspectroscopy 14.3.3. Future Developments in Synchrotron-Based FTIR Microscopy 14.3.4. Summary and Conclusions Related to SynchrotronBased FTIR Microscopy

14.1. INTRODUCTION Synchrotron-based imaging and analyses represent important and increasingly accessible research tools for the analyses of complex environmental materials and their interactions with contaminants in natural systems (Bluhm et al. 2005; Dynes et al. 2006a,b; Thieme et al. 2007; Hitchcock et al. 2008b; Neu et al. 2010). Environmentally relevant processes may involve microscopically variable chemical

and biological interactions that must be analysed at high spatial resolution. Synchrotrons provide many different photon energies—ranging from IR to hard X-ray—that may be applied to environmental samples. The interaction of the light with the sample may be monitored as transmission, electron, or X-ray fluorescence yield or by a wide range of elastic and inelastic scattering processes (Bertsch and Hunter 2001; Brown and Sturchio 2002). More recent studies in the soft X-ray regime have demonstrated application to environmental matrices, including isolated bacterial cells (Obst et al. 2009a,b) and biofilms (Gilbert 1999; Lawrence et al. 2003; Dynes et al. 2006a,b, 2009; Hunter et al. 2008). Application of these techniques can significantly advance our understanding of the processes controlling the chemodynamics of contaminants in the environment, including speciation, distribution, reactivity, transformations, mobility, and their potential bioavailability in the environment. This chapter introduces synchrotron-based imaging and spectroscopies and how they are used for the analyses of complex materials and their interactions with contaminants in natural systems. We will discuss X-ray absorption theory and methodology, including instrumentation, how to prepare, measure, and quantitatively analyze environmental samples, focusing on the spectroscopy, microscopy, detection and mapping of organic contaminants. The same approach will be used to cover synchrotron-based Fourier transformed infrared spectroscopy (FTIR). Examples of both techniques applied to studies of organic contaminants in the environment will be given. Of particular interest for the environmental and biological sciences is soft X-ray scanning transmission X-ray microscopy (STXM), which was developed first by Kirz and Rarback (1985). Soft X-ray spectromicroscopy is increas-

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

341

342

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

ingly available with the rise in the number of high brightness, third-generation light sources that are equipped with STXMs, and improvements in instrumentation and data analysis (Bluhm et al. 2005; Hitchcock et al. 2008a). The STXM method uses the intrinsic properties of the sample based on the application of the near-edge X-ray absorption fine structure (NEXAFS) spectral signal (St€ ohr 1992) as the analytical contrast mechanism that allows mapping of chemical species based not only on elemental composition but also on bonding structure (Ade and Urquhart 2002; Ade and Hitchcock 2008). Further, since it is a photon-in/photon-out technique, STXM may be applied to fully hydrated biological materials—soft X rays penetrate water, particularly in the critical C1s spectral range (280–340 eV), where up to 3 mm of water can be tolerated. Although it has much lower spatial resolution, STXM provides significantly better analytical information than does electron energy loss spectroscopy in transmission electron microscopes (TEMs) because of its superior energy resolution, greater ability to measure thicker samples without spectral distortion, and much greater analytical signal per unit radiation damage (Rightor et al. 1997; Wang et al. 2009a,b). In the wider energy range provided by the undulator-based STXMs a core edge is available for nearly all elements. Many environmental samples, particularly in the fully hydrated state, were previously difficult or impossible to study by other means, because of inadequate spatial resolution (NMR, MRI, optical techniques), sample distortions and artifacts associated with dehydration, radiation effects (electronbeam based techniques), or a lack of suitable chemical information (scanning probe and elemental analysis techniques). In addition, STXM combines the chemical speciation sensitivity of NEXAFS spectroscopy with high spatial resolution (< 30 nm). Thieme et al. (2007, 2008) provide an overview of some potential applications of X-ray microscopy in the environmental sciences. Brown and Sturchio (2002) and Lombi and Susini (2009) provide excellent overviews of synchrotron techniques, including STXM, hard and soft X-ray NEXAFS spectroscopy, and hard X-ray EXAFS spectroscopy, as well as their applications in environmental science. Fourier transform infrared (FTIR) absorption spectroscopy is well established; however, with synchrotron light as the source, versus conventional light sources, both the detection limit and spatial resolution have been substantially improved, resulting in the development of improved spatially resolved FTIR spectromicroscopy. Although FTIR spectromicroscopy is well suited to research on environmental topics, given its excellent capacity to detect organic molecules, its spatial resolution is only in the micrometer range. Excellent overviews of the essentials of IR and IR microspectroscopy have been prepared by a number of authors covering biomedical application (Miller et al. 2003; Dumas and Miller 2003), bacterial cells, (Naumann et al. 1996),

applications of geochemistry and environment sciences (Hirschmugl 2002a,b; Rash and Vogel 2004; Holman and Martin 2006; Lombi and Susini 2009). In addition, an article in Research: Science and Education by M. Stern (Stern 2008) in combination with material by Miller (2010) provides an excellent entry point to synchrotron FTIR spectromicroscopy. In the following review the reader will be provided with information regarding the beamlines available world wide for carrying out these techniques. We will explain how these techniques are rapidly developing as powerful analytical tools for studies in the biological and environmental sciences because they offer chemically sensitive imaging with which to interrogate samples and, in the case of STXM, high spatial resolution. Infrared microscopy can directly map organics in a sample but is much more limited by spatial resolution. Discussion of applications to the detection and mapping of the major biomolecules (protein, lipids, polysaccharides, carbonate, and nucleic acids) and their characteristic spectra will be provided. Software available for the handling and processing of data collected will be described. We will present a critical review of the literature, providing examples of the successful application of these techniques to visualization and mapping of organic contaminants in environmental matrices. The chapter will also highlight future research needs and directions.

14.2. SCANNING TRANSMISSION X-RAY MICROSCOPY (STXM) In STXM, images are obtained in a pixel-by-pixel fashion by mechanically raster-scanning the sample through the focal point of a zone plate X-ray lens (or in some instruments, scanning the zone plate with the sample stationary) while detecting the intensity of the X rays transmitted through the sample (see Fig. 14.1). The quality of the zone plate is critical since the spatial resolution (Rayleigh criterion) of a perfect amplitude contrast Fresnel zone plate is 1.22 times the outermost zone width (Howells et al. 2007). The best zone plates in present-day soft X-ray STXMs have an outermost zone width of 15 nm, which provides a demonstrated spatial resolution of 18 nm (Vila-Comamala et al. 2009). Zone plates with 25 nm outer zone widths have been used in third order, achieving better than 15 nm spatial resolution (Tyliszczak 2010, private communication). There have been great improvements in zone plate fabrication methods and thus quality over the past few years, giving rise to expectations of ultimately achieving a spatial resolution in first order of > 10 nm. 14.2.1. Instrumentation and Access to STXM Soft X-ray imaging and spectromicroscopy with STXM has been carried out at a number of facilities, including twomicroscopes at the National Synchrotron Light Source

SCANNING TRANSMISSION X-RAY MICROSCOPY (STXM)

EPU

343

PGM M1

exit slit zone plate

OSA detector raster scanned sample stage

Figure 14.1. Schematic (not to scale) of 10ID1 beamline and STXM at the Canadian Light Source (CLS) (EPU—elliptically polarizing undulator; M1,M3—mirrors; PGM—plane-grating monochromator; OSA ¼ order-sorting aperture; exit slit also acts as a spatial coherence filter.

(NSLS), two (soon to be three) at the Advanced Light Source (ALS) (Kilcoyne et al. 2003), the Canadian Light Source (CLS), Elettra (Italy), and the Swiss Light Source (SLS). In addition, beamlines are under construction at the Stanford Synchrotron Light Source (SSRL, United States), Bessy (Berlin, Germany), SSRF (Shanghai, China), Soleil (Paris, France), Alba (Barcelona, Spain), and Diamond (Oxford, UK). In addition, STXMs covering the tender X-ray regime (2–8 keV) are available at each of the three third-generation high-energy rings: European Synchrotron Facility (ESRF), Argonne (APS), and Spring8. Some of these STXM installations have been the subject of detailed presentations with examples of their applications. For example, the STXM on the molecular environmental science beamline 11.02 at ALS has been reviewed by Bluhm et al. (2005), while Hitchcock et al. (2008a) provided an overview of the spectromicroscopy facility at the Canadian Light Source (CLS) in Saskatoon, Canada. The source point may be a bending magnet, linear undulator, or an elliptically polarized undulator (EPU) such as at the CLS. In the case of the CLS (Fig. 14.1), the X-rays are directed by a sagittal cylindrical mirror to a plane grating monochromator that allows performance over an energy range from 150 to 2500 eV (Hitchcock et al. 2008a); this instrument currently has the broadest energy range available for soft X-ray STXM applications. Most STXM analyses use the information provided by the NEXAFS of elements with absorption edges in the soft X-ray energy range, in particular the carbon, nitrogen, and oxygen K edges. The K-shell absorption edges of Na, Mg, Al, and Si and the L edges of important transition metals including Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Zn, as well

as the L edges of P, S, Cl, K, and Ca, are all available. For heavier elements there are other absorption edges, which can be accessed at STXM beamlines. The ability to clearly resolve hydrated samples at high magnification, with a resolution of 30 nm and with minimal sample preparation, are powerful features of STXM. Limitations to STXM include radiation damage, a requirement for very thin samples (<200 nm equivalent thickness of dry organic components), the presence of less <5 mm of water when wet, use of fragile silicon nitride windows (other options, such as formvar or polyimide, are available as described below), challenging sample preparation (i.e., encapsulation in a wet cell and absorption saturation distortion of analysis in thick regions of a specimen) and suitability of model compounds used to create reference spectra for quantitative mapping. Despite these limitations, STXM has been used for a wide variety of sample matrices, including those that are polymeric, geochemical, magnetic, extraterrestrial, biological, and environmental. 14.2.2. Sample Preparation All STXM samples must be prepared either freestanding (typically draped across a 3-mm Cu grid as used in electron microscopy), or on an X-ray transparent holder; STXM measurements can be performed on a wet cell constructed by sandwiching the sample between two thin (50–100-nm) silicon nitride windows (Norcada Inc, Edmonton AB, Canada; Silson Inc, Northampton, UK). In the case of biological materials, samples may be grown directly on the window (Lawrence et al. 2003; Dynes et al. 2009).

344

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Figure 14.2. (a) Photograph of a wet cell mounted on a STXM sample plate of the type used at the ALS, CLS, and SLS. Up to six samples can be mounted on a single STXM plate. (b) STXM image (transmission) at 288 eVof the whole 1-mm silicon nitride pane (2  75 nm) and the enclosed natural river biofilm sample. (c) STXM image (optical density, OD) of an 80  80 mm area of the biofilm at 280 eV, below the C1s absorption onset, where inorganic species (silica, calcium, etc.) are highlighted. (d) STXM image (OD) of the same 80  80 mm area as in (c) at 288.2 eV, the peak of the C1s ! p amide transition of proteins that highlights biological species. The white box indicates the region subjected to detailed C1s image sequence, with analysis results presented in Figure 14.4.

Figure 14.2 shows a typical completed silicon nitride window wet cell with enclosed sample mounted on an aluminum holder placed in the microscope. It is also possible to construct wet cells using polyimide windows (Luxel Corporation, Friday Harbor, WA, http://www.luxel.com/). Relative to silicon nitride, polyimde is less brittle, making it an easierto-handle window system that may be more amenable to environmental samples (Hitchcock et al. 2008a). Although polyimide windows have C1s signals, it is characteristically and radiation-stable. Also, for the 30-nm polyimide used, the absorbance in the C1s region (280–300 eV) is much lower that that of Si3N4 windows. This allows analysis of thicker

samples before absorption saturation becomes a limiting factor. This is of particular advantage for use with biological and environmental samples such as microbial biofilms. Although a wet cell may be preferred for many applications, the samples may also be mounted on a suitable substrate (window or coated grid) and air-dried prior to being examined dry. In many cases this is a straightforward process and allows imaging and data collection without the complications of maintaining a wet cell, although there can be concerns about modification of the sample on dehydration. For example, Obst et al. (2009a) simply deposited 1–2 mL of a cyanobacterial cell suspension onto a Si3N4 window (1 

SCANNING TRANSMISSION X-RAY MICROSCOPY (STXM)

1 mm, thickness 100 nm on a 200-mm-thick Si chip, 5  5 mm, Norcada Inc., Edmonton, Canada) and then used a filter paper to withdraw free water leaving the cells on the window surface. Use of bibulous papers (available from a number of suppliers, e.g., Fisher Scientific, Wards Natural Science) may also be considered as they provide greater control over the rate of water removal. Although observation of unaltered, fully hydrated materials is highly desirable, the necessity for special handling and preparation may also prevent it. In many cases the sample may be too thick for STXM, thus requiring embedding in epoxy (Spurr’s resin), or nanoplast (Lawrence et al. 2003). However, fixation, alcohol dehydration and embedding in epoxy resins removes most of the organic biological compound, and the spectrum of the embedding material masks the C1s spectral information of the sample as shown by Lawrence et al. (2003). Therefore approaches such as cryosectioning, in which the sample is frozen and sectioned, may be preferable to conventional electron microscopy sample preparation for environmental applications (Lawrence et al. 2003). Thin sections or ultrathin sections may be placed on a silicon nitride or polyimide window or on electron microscope grids for mounting and observation in STXM. Figure 14.2b illustrates a typical arrangement of samples on windows attached to an aluminum sample plate or holder. In another useful approach for specific sample types, Loo et al. (2001) describe the use of glycolmethacrylateembedded and sectioned materials for STXM analyses. Sulfur embedding has been highly successful for embedding interplanetary dust samples (Bradley et al. 1993); it has also been used in conjunction with soil aggregates (Lehmann et al. 2005). Essentially, particles are embedded in preheated (220 C) elemental sulfur and supercooled (in liquid N2) before sectioning, imaging, and spectroscopic analysis. Lehmann et al. (2005) and Kinyangi et al. (2006) provide a detailed description of the technique. This approach has the advantage of being carbon-free; however, sublimation at room temperature limits longer-term storage or examination under vacuum. Another option for environmental samples used to observe mineral interfaces is that of focused ion beam (FIB) milling, which allows for careful selection of the slice plane. Obst et al. (2005, 2009a) illustrates the preparation of ultrathin sections of CaCO3 crystals attached to cyanobacteria by FIB milling using the liftout approach. Essentially, the material is air-dried, coated with platinum to protect the organic material, and then sectioned using an optimized approach to minimize damage to the specimen (Obst et al. 2005). The prepared ultra thin sections may then be transferred to carbon-coated 200-mesh copper TEM grids using a micromanipulator. In addition to conventional 2D imaging of samples, it is possible to perform 3D imaging using STXM, employing either serial section (Hitchcock et al. 2003) or angle scan

345

techniques. Johansson et al. (2007) described an apparatus that enables rotation of samples at the focus of the X-ray beam in a STXM, allowing measurement of 3D chemical distributions with 50 nm spatial resolution in three dimensions. Samples can be placed in thin-walled glass capillaries similar to those typically used as micropipettes in intra- and extracellular physiology, or on strips of conventional copper grids. Prior to loading of a sample, the capillaries must be treated by heating and pulling using a micropipette puller to reduce the initial diameter of 1 mm to 2–3 mm with a uniform wall thickness in the range of a few hundred nanometers over the length to be scanned by STXM of approximately 10–20 mm. These capillaries may then be filled with the sample, either from the top, or from the end of the thin section, using capillarity. The method has been used to study yeast cells, bacterial cells, biofilm communities, and synthetic polymer samples (Johansson et al. 2007). After filling with the wet sample, the capillaries are centrifuged for 5–10 min at 6000g to ensure that the water and the sample move into the fine tip. The capillary is then sealed with silicone (RTV silicone, WPI, Sarasota, FL). However, imaging may be best done using O1s edge rather than C1s. In another option, Hitchcock et al. (2008b) described the use of a carbon tube with 45-nm walls and an internal diameter of <1 mm in the region of the sample. Use of these sample containers allowed imaging at the C1s. In this case, rather than centrifuging the samples, they were able to add the sample via capillarity. The capillary end was sealed with silicon grease prior to imaging in the STXM. Cell biologists have demonstrated that 3D reconstructions of data measured using the closely related full-field transmission X-ray microscopy (TXM) from individual cells have excellent natural contrast, eliminate the need for staining, and reveal additional details of intact cells in a near native state (Gu et al. 2007; Larabell and Le Gros 2004; Parkinson et al. 2008). Some of the best examples of this application are the imaging and reconstruction of whole, fully hydrated yeast cells Schizosaccharomyces pombe (Parkinson et al. 2008) and the 3D imaging and mapping of 1-mm acrylate-filled spheres (Johansson et al. 2007; Hitchcock et al. 2008b). Figure 14.3 illustrates five of a series of STXM projection images of a biofilm sample at different angles. A set of 26 such images was used to create a 3D reconstruction of a river biofilm community containing a diatom mounted in a capillary (Johansson et al. 2006). Thieme et al. (2003) present tomographic reconstructions of bacteria and colloids revealing detailed information on spatial arrangements. This approach holds considerable promise for 3D chemical mapping of biological and environmental samples. 14.2.3. Imaging and Data Collection After preparation, the wet cell, grid, or window is mounted on a aluminum holder plate, allowing it to be placed directly in

346

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Figure 14.3. Chemical maps of “protein” in a 3D volume—selected slices from a 3D reconstruction of the protein component of a wet river biofilm, derived from the difference of 2D projection images recorded at 532.2 and 530.0 eV.

the beamline for imaging (Lawrence et al. 2003; Hitchcock et al. 2002) (see Fig. 14.2). The sample is placed in the STXM chamber connected to the beamline, and moved to the focus of the zone plate. The sample chamber is then evacuated (if the sample is dry) or purged with dry helium (for sealed wet cells) prior to making measurements. Modern STXMs (Feser et al. 2000; Kilcoyne et al. 2003) have high-precision mechanical and piezoelectric scanning systems with interferometry-based control, allowing ranges of sample positioning and scanning from < 1 m to several centimeters, with nanometer precision. For scans made with the fine piezo stage, dwell times can be as fast as 100 ms/pixel, although 1 ms/ pixel is more typical for analytical results. Data acquisition of images larger than the range of the fast piezo stage use stepping motor-driven stages, which are much slower than the piezo stages. Use of an indexed optical microscope can help to rapidly identify areas of interest and thus use beam time most efficiently. Between the zone plate and the sample there is the order selection aperture (OSA), a 30–100-mmdiameter aperture placed so as to block zeroth-order light (in conjunction with the central stop of the zone plate) but not to clip the focused first-order diffracted light. Proper alignment of the OSA and central stop is required to achieve highspatial-resolution images and valid spectroscopy at the ultimate spatial resolution. Depending on the instrument, the transmitted X rays are detected by a proportional counter (NSLS), a Si photodiode (CLS), a segmented photodiode [NSLS (Feser

et al. 2000)], an avalanche photodiode (ALS), or a phosphor scintillator, and a photomultiplier tube [ALS, CLS (Kilcoyne et al. 2003)]. It is necessary to check the energy scale daily, particularly at the undulator beamlines at ALS and CLS, which do not have entrance slits and thus rely on reproducibility of the storage ring beam position. The energy scale can be established to a precision of 0.05 eV using sharp gasphase signals, typically the Rydberg peaks of gases such as N2, CO2, O2, and Ne. Calibration may also be carried out using other standard samples, including various solids such as Si, SiO2, and Al2O3. Additional details and explanations of beamline design, construction, and operation are provided in Kilcoyne et al. (2003), and Bluhm et al. (2005), while soft X-ray optics are dealt with by Jacobsen et al. (1991, 1992). A number of excellent reviews (Ade 1998; Ade and Urquhart 2002; Kirz et al. 1995, Howells et al. 2007; Ade and Hitchcock 2008) can be consulted for more technical details on the optics and mechanics of transmission X-ray microscopes, as well as comparisons of STXM to the three other common soft X-ray synchrotron spectromicroscopies—full-field transmission X-ray microscopy (TXM), scanning photoelectron microscope, and photoemission electron microscopy. The STXM technique can be used analytically by acquiring NEXAFS spectra at a single point, using a spectral line scan, or through collection of a sequence of images at different energies, often referred to as a “stack” (Jacobsen et al. 2000). The range and point spacing of the energy scale

SCANNING TRANSMISSION X-RAY MICROSCOPY (STXM)

are selected in order to differentiate the chemical components of interest in the sample. Monochromatic soft X rays are absorbed by the sample at specific energies at which core electrons are excited to unoccupied states, with intensities and linear dichroic properties characteristic of the chemical species being excited (St€ ohr 1992). The acquisition of a full STXM dataset consisting of an image sequence of 4  4 um2 area, sampled at 40 nm (100  100 pixels) with 1 ms/pixel dwell and 120 energy steps, takes 30 min, with perhaps another 30 min spent in initial focusing, navigating to a suitable area, and making higher-quality single energy images. If one wants very good statistics, higher spatial sampling (for some projects, pixels as small as 5 nm are used to oversample particular areas) or if multiple core levels are studied, a single sample can take 4 h or more to measure. Prescreening conventional light or laser light microscopy, or even transmission electron microscopy under low-dose conditions (for dry samples) is very useful for make a strategic selection of areas for STXM analysis. With this prescreening, the user will collect singleenergy images at selected energies for navigation within the sample and ultimate selection of the sites of interest for detailed STXM imaging. In some cases it is useful to employ a full-spectrum sampling strategy, particularly with samples that are poorly characterized and may contain unknowns. Figure 14.4 shows a selection of images from a 52-image sequence of a wet river biofilm, recorded using 233  233 pixels and 1 ms/pixel dwell. In order to improve statistics, images in regions with similar spectral characteristics are summed, and some differences are taken to emphasize the chemical contrast. A more complete, quantitative analysis of this image sequence using spectral fitting techniques has been presented elsewhere (Hitchcock et al. 2005). These types of STXM image sequences contain much analytical information, but they must be processed correctly to extract valid results. The measured transmitted signals (I) are first converted to optical densities [absorbance, OD ¼ -ln(I I0-1)] using incident flux (I0) measured through regions of the wet cell devoid of biofilm/sample, to correct for the absorbance by the silicon nitride membranes and the water in the wet cell. Following this conversion, a number of strategies may be used to extract quantitative chemical maps and to interrogate the reliability of the results. Some of these are discussed below. Although STXM provides much more information per unit damage as compared to TEM-EELS (Rightor et al. 1997; Wang et al. 2009a), it uses ionizing radiation, which does result in radiation damage to samples, particularly if long dwell times or many energies are used, or if a number of core edges are measured at the same location. Various organic compounds may be modified by ionization and dissociation of bonds (Cody et al. 2009; Wang et al. 2009b). A useful strategy to monitor the level of damage in any given mea-

347

surement is to collect a postsequence image over a region centered on, but a few micrometers larger than the main stack area. Since the raster scan starts at zero velocity, an additional dose is given to a 1-mm section just to the left of the area where the stack is measured, which forms a clear vertical stripe, if the damage is extensive (see Fig. 14.5, which shows damage visualization in a radiation-sensitive PMMA sample). In examining larger scale images it is possible to focus on two regions where damage is most likely to have occurred, the STXM operates in a line-at-a-time mode; therefore, the beam sits 1 mm to the left of the start position of each horizontal scan line for about 10 times longer than in any pixel in the stack; similarly, there is a deceleration region at the right side of the stack. Damage will be clearly visible within a vertical zone outside the area in which the stack was collected. Typically for each system there is one component more damage sensitive than others. Post image sequence imaging at the energy where the damage to that component is most visible is a good way to check whether any given measurement was made within acceptable limits of sample damage. For example, in microbial biofilm systems a damage check scan is typically measured at 289 eV to detect changes to polysaccharides, the most easily damaged chemical component. We are willing to accept data for further workup if the extracellular matrix polysaccharide signal is reduced by less than 20% (Dynes et al. 2006b). Toner et al. (2005) demonstrated that radiation damage in STXM can modify oxidation-state speciation (e.g., Fe 2p and Mn 2p). In addition to monitoring for damage, it is useful to establish the potential for damage in the STXM beamline by assessing the stability of the material under examination. Wang et al. (2009b), Cody et al. (2009), and others have discussed in detail radiation damage mechanisms in polymer and biological samples in a special issue of Journal of Electron Spectroscopy and Related Phenomena (February 2009). There are also strategies to minimize radiation damage, including defocusing the X-ray beam (with appropriately matched pixel spacing) or short dwell times and very restricted energies (which works well once a good understanding of the spectral–speciation relationship for a class of samples has been achieved). In addition, a strategy of extrapolating back to the unaltered spectrum at zero time may also be possible. However, one must note that, even if there is not evidence of damage at a particular energy due to destruction or formation of a bond absorbing at the energy or from mass loss (ablation), this does not mean that damage has not occurred. 14.2.4. Quantification After collection of the image sequences, additional processing is required before quantitative analyses can proceed. This includes alignment of images (if needed), conversion to

348

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Figure 14.4. STXM images of a wet river biofilm: (a) sum of 12 images (282.0–286.0 eV)—inorganic species including SiO2 of a diatom frustule and CaCO3; (b) sum of two images (287.5–288.3 eV)— protein amide p ; (c) sum of three images (288.53–288.9 eV)—lipids and protein; (d) image (c) minus 0.5 image (b)—lipids; (e) sum of 10 images (290.0–294.0 eV)—carbohydrates, lipids, and proteins; (f) sum of six images (296.5–301.0 eV) minus image (e)—potassium. Scale bar is 5 mm. Further results on this sample were presented in Hitchcock et al. (2005).

optical density, and removal of any single images where the spectral content statistically deviates from the local trend (glitches). The images can be aligned using 2D Fourier autocorrelation techniques or by manual alignment procedures. There should be minimal drift from image to image. To check alignment, images may be summed to assess whether objects shift or change size within the stack. The processed image sequence data can then be converted to

component maps (distributions of distinct chemical species, e.g., proteins, polysaccharides, nucleic acids, organic or metallic components) by spectral fitting using linear regression procedures. Each of the spectro-microscopic datasets consists of many thousands of NEXAFS spectra (one per pixel of the analyzed area). These spectra represent the sum of the spectral contributions of all the individual chemical species that are present at this spot of the sample (i.e.,

SCANNING TRANSMISSION X-RAY MICROSCOPY (STXM)

Figure 14.5. Postanalysis imaging of a polymethylmethacrylate (PMMA) thin film over an area larger than the scan used (white dashed lines) reveals vertical lines to left and sometimes to right of the imaged area, which are due to the greater exposure in the region of each line used for accelerating and decelerating the sample during each line scan.

typically 104–105 spectra). These datasets are best analyzed using efficient, matrix-based least-squares techniques such as singular value decomposition (Koprinarov et al. 2001, 2002) or by multivariate statistical methods such as principle components and cluster analysis (Jacobsen et al. 2003; Lerotic et al. 2005, 2004; Mitrea et al. 2008). A graphical user interface implementing many different data analysis protocols and with ability to read in most spectromicroscopy data formats is freely available (Hitchcock 2009). In a relatively typical case involving the application of singular value decomposition (SVD) an image sequence was recorded, which consisted of 160 images between 280 and 320 eV, each of dimension 300  125 pixels, collected with a dwell time of 1 ms/pixel (total acquisition time, including “dead” time to return the piezo scanner to the star to the next line was 2.5 h) (Dynes et al. 2006a). Quantitative maps of chlorhexidine, protein, lipid, polysaccharide, silica (SiO2), CO32, and K þ were then derived from this image sequences by using SVD (Dynes et al. 2006a; Pecher et al. 2003: Koprinarov et al. 2002; Jacobsen et al. 2000) to fit the spectrum at each pixel to a linear combination of reference spectra of the components suspected to be present. The reference spectra used in this case are displayed in Figure 14.6 (Dynes et al. 2006a). Quantitative reference spectra were determined using human serum albumin (protein), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (lipid),

349

xanthan gum (polysaccharide), and K þ (K2CO3 with the signal of CaCO3 subtracted to remove the C1s spectrum of the carbonate anion) (Dynes et al. 2006a,b). Other reference spectra may be found in the literature. As described by Dynes et al. (2006a,b), the reference spectra recorded from pure materials, are placed on absolute linear absorbance scales by matching them to the predicted response for the compound according to its elemental composition and density, using tabulated continuum absorption coefficients (Henke et al. 1993). Each component selected for analysis requires specific data treatment. In the case of potassium K þ it was necessary to subtract the appropriately weighted spectrum of KCO3, from that of K2CO3, allowing independent derivation of a reference spectrum for pure K þ . Silica (SiO2) presents a spectrum that is a slowly varying and featureless signal in the C1s region as taken from tabulated elemental absorption coefficients (Henke et al. 1993) and the composition and density of SiO2. The frustules in this diatom sample were nearly pure ( 97%) hydrated silica (SiO2) (Noll et al. 2002) and were the dominant nonorganic component of this sample type. Therefore the component maps derived from fitting the silica component represented the spatial distribution of SiO2. Then Ca2 þ was mapped from the difference in optical density (OD) images at 352.6 eV (Ca 2p1/2 ! Ca 3d resonance) and at 350.3 eV (in the dip between the 2p3/2 and 2p1/2 resonances). The OD scale was rendered quantitative by scaling by 0.10(2) nm1, the difference in the linear absorbance of nm of CaCO3 at these two photon energies. The image and spectral processing may be carried out using a number of software packages, with two of them, aXis2000 (Hitchcock 2009) and the suite of IDL programs provided by Chris Jacobsen (Jacobsen 2009), written explicitly to analyze soft X-ray spectromicroscopy datasets. For most confidence in the analysis result, it is prudent to explore the reliability of any given component mapping using several approaches. The first check is a residual image, which is the difference between fit and data at each pixel, averaged over the image sequence energy range. Typically a good fit has a low magnitude and absence of structure in the residual image. A second check is to examine the fit of the spectrum of “hot spots” identified by applying a threshold masking technique to each component map, extracting the spectrum from these selected component-rich pixels, then examining the quality of the spectral fit using the same reference spectra used to generate the components. A third check is to use principal component analysis (PCA) to assesses the number of chemical components that are reasonably identifiable from a given image sequence, based solely on the variance and the statistical precision of the data. In some instances, although rarely in environmental samples, all chemical species are known and therefore may be detected and quantified in the dataset. However, this

350

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

(a)

(b) K+ CaCO3

0.01 µm-1

1 µm-1

protein benzalkonium

polysaccharide

chlorhexidine digluconate

chlorhexidine dihydrochloride

chloride Linear absorption coefficient

Linear absorption

lipid

triclosan

chlorhexidine dihydrochloride

chlorhexidine-internal

285

290

295

300

305

Energy (eV)

285

290

295

300

305

Energy (eV)

Figure 14.6. (a) C1s reference spectra used for analysis of chlorhexidine in an environmental river biofilm sample from [Dynes et al. (2006a) reproduced with permission]; (b) comparison of C1s spectra of three common antimicrobial compounds.

approach is seldom possible. Lerotic et al. (2004, 2005) have developed multivariate statistical analysis (MSA) procedures to analyze spectromicroscopy data. The image sequence was first subjected to PCA in order to orthogonalize and noisefilter the data. They then applied cluster analysis to classify the principal components to obtain rotated principal components that should correspond to the spectra of the dominant chemical components in the region studied. The cluster analysis classifies pixels according to their spectral similarity. The classification can be improved by the applying of an angle distance measure rather than the Euclidian distance measure first used (Lerotic et al. 2005). These derived spectra are then used in a target analysis (similar to SVD) to obtain thickness maps of the components. Essentially, this approach allows the user to extract representative, cluster-averaged spectra and then perform a spectral–spatial decomposition using those internally generated reference spectra. While this method is very attractive in that it is “unsupervised” (i.e., it makes no assumptions about the chemistry or spectroscopy of the sample) it has its own limitations; (1) the cluster analysis is not unique, but depends on various parameters in the implementation; and (2) whenever two chemical components have very similar spatial distributions, the cluster procedure cannot differentiate them. In such cases, if one chemical species is known to be present in the sample, its

introduction at the target analysis set can frequently improve the final result. A simple application of PCA is to determine how many components may be meaningfully extracted from any given dataset. Dynes et al. (2006a) used this approach to find that up to eight components could be meaningfully extracted from the stack for which the analysis results are displayed in Figure 14.6. Meaningful spectral fits were achieved with six (protein, polysaccharide, lipid, chlorhexidine, SiO2, and CaCO3), seven (same as for six, plus K þ ), and for eight components (same as for seven, plus water) for each image sequence. Dynes et al. (2006a) also examined the application of several alternate model compound spectra for detection and mapping of the protein, polysaccharide, lipid, and chlorhexidine components in the stacks. Figure 14.7 illustrates the results of successful fitting of the reference spectra plotted in Figure 14.6 to the image stack of the diatom and bacterial colony. It shows the maps of protein, lipid, chlorhexidine, polysaccharide, CO32, K þ , silica and Ca2 þ as derived by SVD. Figure 14.8 presents several falsecolor composites that show the distributions of the chlorhexidine in relation to other components such as lipid and polysaccharides. Similarly, Obst et al. (2009a) analyzed C-1s image sequences measured from samples of Synecoccus leopoliensis by using SVD to fit reference spectra for

SCANNING TRANSMISSION X-RAY MICROSCOPY (STXM)

351

Figure 14.7. Quantitative component maps for protein, polysaccharide, lipid, chlorhexidine, K þ , SiO2, and CaCO3, plus Ca2 þ (from the Ca 2p edge) for a river biofilm in the region of a pennate diatom cyanobacterium. [From Dynes et al. (2006a), reproduced with permission.] The gray scale gives thickness in nanometers.

protein, polysaccharides, lipids, carbonate, and the spectrum of H2O, which has no spectral features at the C1s edge. This type of multistep evaluation supports the conclusion that the final result reported is relatively independent of the analysis used. There remains the question of reproducibility and sensitivity in measurements of this type. Typically the spatial distributions derived in SVD analyses are surprisingly insensitive to small changes in the data range used or the reference spectra chosen (as long as they have the characteristic peaks at the correct energies). Although the precision of any single image is typically only 2–5%, the point-to-point precision of the derived component maps is much better, due effectively to integration of the signal contributions from many images over the whole spectral range measured. However, the magnitudes of the thickness scales are considerably more variable (for weaker components, fluctuating by factors of 2). An evaluation of the variability (systematic error as opposed to random error) with different choices of reference spectra (all plausible) would assign approximately 20% uncertainty for the majority of components (maximum contributions > 50 nm), and up to 50% uncertainty in the scales for the

minority components (maximum contributions < 50 nm). The major contribution to this uncertainly stems from systematic errors in knowing the reference spectra appropriate for a specific sample. Another major concern with any analytical approach is its sensitivity. In general, STXM can provide quantitative maps of chemical species at environmentally relevant total concentrations (i.e., mg/kg or lower), as long as the minority components are spatially localized. (Rightor et al. 2002; Hitchcock et al. 2002; Lawrence et al. 2003). When examining biological samples, it is relatively straightforward to discriminate the major biomacromolecules, including protein, lipid, and polysaccharides. Discriminating within these categories has proved more complex. Obst et al. (2009a) noted that the chemical sensitivity of C1s spectromicroscopy did not allow detection of any major differences between the spectra obtained from EPS of cells cultured under varying nutrient conditions. In contrast, Stewart-Ornstein et al. (2006) found that it was possible to discriminate an antimicrobial peptide in a protein (albumin) background using STXM. This was based on the distinctive nature of the NEXAFS spectra of proteins that are richer in cysteine or methionine relative to other proteins. Differentiation of

352

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Figure 14.8. Color-coded composite maps (in each case with rescaling within each color) assembled from the component maps presented in Figure 14.7. [From Dynes et al. (2006a), reproduced with permission.] (See insert for color representation of this figure.)

specific peptides from a protein background has been used in studies of competitive absorption to candidate biomaterials (Leung et al. 2008). Dynes et al. (2009) estimated the detection limit for measuring chlorhexidine by STXM by adding the spectra of chlorhexidine and the major biomacromolecules over a range of compositions and evaluating the detection limit by factoring in the visibility of the characteristic sharp features at 285.1 eV (C1s ! p*ring ) and 286.4 eV (C1s ! p*C¼N ) transitions in the phenyl and imide groups of chlorhexidine against the background spectrum of the other species. Chlorhexidine was detected at 10 nm in the context of 50 nm of protein or 100 nm of lipid, and at a level of 20 nm against a background spectrum of 50 nm protein plus 100 nm of lipid. In this case the absolute sensitivity was impressive; corresponding to  1017 mol, assuming that the density of pure CHX was 1.2 g/cm3. Thus, provided accurate reference spectra are available, there is exceptional potential for the detection of organic contaminants in a relatively complex matrix using STXM.

14.2.5. Correlative Microscopy The use of STXM in conjunction with other spectromicroscopic imaging approaches, and with light and electron microscopies, is also possible. Lawrence et al. (2003) demonstrated the application of confocal laser scanning microscopy, TEM, and STXM to the same biological samples. Dynes et al. (2006a) demonstrated the combination of CLSM and STXM to map biomacromolecules and an organic contaminant in river biofilms. The development of beamlines with “universal” mounting systems and cross-referenced sample locations for ease of navigation to regions of interest also supports the notion of the application of correlative and multimicroscopies to all types of samples, including those of environmental origin. These combinations will allow extraction of the maximum in information from an environmental sample. Schafer et al. (2007) provides a useful example of the correlative application of three different methods, TEM, STXM, and m-FTIR microscopy, to characterize and map colloidal and particulate organic matter obtained from a lake

SCANNING TRANSMISSION X-RAY MICROSCOPY (STXM)

and two rivers. They identified the sources of colloidal material in the receiving lake environment and examined aspects of their stability, interactions, and fate. Some examples of those results are presented below. 14.2.6. Examples of STXM Applied to Organic Environmental Contaminants Although the full potential for applications of STXM in the environmental sciences is only now emerging, there are already good examples across a range of relevant disciplines, including geochemistry, hydrology, microbiology, and atmospheric and soil sciences. We have noted that STXM has already made significant contributions to soil biogeochemistry (Lehmann et al. 2005), environmental microbiology (Lawrence et al. 2003; Dynes et al. 2006a,b, 2009), biogeochemistry (Toner et al. 2005), and colloid chemistry (Schumacher et al. 2005; Mitrea et al. 2008), in atmospheric (Braun et al. 2005, 2008), marine/aquatic (Claret et al. 2008; Dynes et al. 2006a, b; Obst et al. 2009a, b), and soil remediation (Yoon et al. 2006). Carbonaceous airborne particulate matter (CAPM) or soot has considerable deleterious effects on human health and climate; however, it is relatively poorly defined. Braun et al. (2008) applied STXM to characterize subgroups of CAPM, including wood smoke, diesel soot, and urban air. They applied the STXMs at the ALS and NSLS to record C1s NEXAFS spectra of single particles of CPM. Their results point to significant differences in diesel exhaust particulates and those of wood smoke; the former have a semi-graphitic solid core, whereas the latter have none or less of this core material. The results help apportion relative contributions of these two air pollutants. This study builds on an earlier STXM study of coal by Cody et al. (1995) and by this same group on diesel exhaust particulates (Braun et al. 2004). Russell et al. (2002) also examined the carbon coatings on atmospheric dust particles. Another critical issue related to carbonaceous particulate materials is their interaction with hydrophobic organic compounds (HOCs), including PCBs and PAHs in the environment. The sorption of these compounds by carbonaceous particulate materials significantly influences their mobility and bioavailability and our capacity to recover and remediate sites contaminated with PCBs and PAHs. As part of the studies assessing these phenomena, Yoon et al. (2006) applied STXM to assess nanoscale chemical heterogeneities in these carbon materials and how they influence sorption of PCBs. They found that the functional groups of the carbonaceous materials vary on a 25-nm scale corresponding to the abundance of the HOCs. In general these interactions reduce the availability of PCBs in the environment acting to sequester them. Schumacher et al. (2005) also investigated the chemical heterogeneity of organic colloids at the particle scale, ap-

353

plying STXM spectromicroscopy to 49 individual particles isolated from the surface horizons of three forest soils. They found very large interparticle variation, particularly in the aromatic and carbonyl contents of the particles. Mitrea et al. (2008) used the scanning transmission X-ray microscope at BESSY II to examine colloidal structures from a Chernozem soil at a spatial resolution near 60 nm and a spectral resolution of 1700 at the C1s edge to determine the distribution of organic matter in these structures. Natural organic matter also plays an important role in the transport and stability of organic contaminants, and STXM based studies have allowed very fine scale examination of these interactions (Claret et al. 2008). Dynes et al. (2006a) reported on the use of STXM to examine the spatial distribution of chlorhexidine (1,10 hexamethylenebis[5-(p-chlorophenyl)biguanide]), a widely used antimicrobial agent, in river biofilm communities grown in the presence of 100 mg/L of chlorhexidine digluconate. Similarly, STXM was successfully applied to map the distribution of chlorhexidine in pure bacterial cultures showing its penetration of the cell and accumulation in the lipid fraction of the bacterial cells (Dynes et al. 2009). With STXM, it was possible to show unambiguously that chlorhexidine was sorbed or otherwise chemically associated with the lipids in the diatoms and bacteria. Some of these results are presented in Figures 14.4, 14.6–14.8. It is important to note that attempts to map the ubiquitous environmental contaminant triclosan (see spectra in Fig. 14.7) met with failure as it was not possible to differentiate the contaminant from the complex protein background of the sample; this is an issue in the detection of organic contaminants in many environmental matrices. Schafer et al. (2007) used a combination of STXM, FTIR, and electron microscopy to study organic colloidal/particulate material sampled as a function of depth in Lake Brienz, an ultraoligotrophic lake in Switzerland. Their study focused especially on organic functionality. Figure 14.9 presents STXM results from a sample at 100 m. The C1s spectra clearly show that organic material is associated with potassium rich inorganic colloids present in both the surface and deep water, which indicated a vertical transfer of aggregates by sedimentation. Figure 14.10 presents FTIR microscopy results from the same 100-m-depth sample. There is good correlation of the conclusions reached from both measurements and the NEXAFs and IR spectra of the deep organic matter was very similar to that found in two tributary rivers feeding the lake. 14.2.7. Future Developments in STXM Although three-dimensional visualization and quantitative analyses are at a reasonably advanced level, there is a need for additional improvements in sample handling, as well as data collection and data analysis. Instrumentation would

354

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Figure 14.9. STXM analysis of a colloid sample from 100 m below Lake Brienz: (a) absorption image taken at 280 eV below the C1s edge showing inorganic colloids and particles; (b) PCA and cluster analysis showing distributions of two distinctive clusters (yellow and red) and the background region (blue); (c) corresponding C1s spectra of yellow and red clusters. [From Schafer et al. (2007), reproduced with permission]. (See insert for color representation of this figure.)

benefit from improving resolution to 10 nm, availability of X-ray fluorescence yield detection methods, more rapid raster scanning, improved bright-field detectors, and faster throughput for tomography studies. Many novel experiments could be performed if there were greater control over the sample environment (aerobic, anaerobic, temperature, and capacity to change the environment for experimental purposes). In general, there is a desire or need to obtain conditions as close as possible to the environmental system of interest. Additional correlative approaches are also being facilitated within the synchrotron research community and with other microspectroscopic and imaging techniques that will allow a much more effective application of these approaches and attendant improvements in our understanding of both the fate and effects of contaminants. However, this will require more attention to sampler holders and navigation to study sites with high precision. There is also a considerable need for improvements in the postcollection processing and analyses of STXM data stacks to improve both resolution and sensitivity. Different approaches to spectral fitting may allow discrimination between organic contaminants that hitherto were hidden within the predominant backgrounds of organic matter proteins, and other biomolecules.

Three-dimensional visualization is already at a reasonably advanced level. However, new procedures for extraction of quantitative and structural information need to be established in order to automatically analyze multichannel datasets with collocalized signals. Finally, there remains the challenge of all microscale studies: relating these findings to the much larger scale.

14.2.8. Summary and Conclusions Related to STXM Soft X-ray scanning transmission X-ray microscopy is both a promising tool for advancing the study of environmental contaminants and one with a high degree of realized promise. Nevertheless, STXM will continue to increase our understanding of processes in the environmental dynamics or organic contaminants. The area of threedimensional chemical mapping at high spatial resolution using angle scan computed STXM tomography (Johansson et al. 2007; Hitchcock et al. 2008b; Obst et al. 2009b) has great potential with regard to environmental samples. In the near future these approaches likely will increase our understanding of the structures and processes in microbial communities and their role in the fate of environmental contaminants.

SYNCHROTRON-BASED FOURIER TRANSFORM INFRARED MICROSPECTROSCOPY

355

Figure 14.10. FTIR results for a colloid sample from 100 m below Lake Brienz: (a) visual light microscopy image of the TEM grid square subjected to FTIR spectromicroscopy; (b) functional group mappings of aromatics and clay associated OH vibrations at 3620 cm1; (c) aliphatics as well as correlation maps using target spectra from aquatic organic colloids from tributary rivers (scale bar is ¼ 5 mm) [from Schafer et al. (2007), reproduced with permission].

14.3. SYNCHROTRON-BASED FOURIER TRANSFORM INFRARED MICROSPECTROSCOPY 14.3.1. Infrared Spectroscopy The infrared part of the electromagnetic spectrum is divided into three regions on the basis of energy: the near (14,000–4000 cm1), the mid- (4000–400 cm1) and the far infrared (400-10 cm1). Far-infrared spectromicroscopy is of interest and, as noted by Miller et al. (2003), has considerable potential, but less is known about the spectra generated and, because of the long wavelength, spatial resolution is very limited (100 mm or worse). The midinfrared region is the most useful for examining organic contaminants in the environment, so our discussion will focus on this approach. The absorption of midinfrared

radiation by molecules excites vibrational modes, which have frequencies determined by atomic masses, bond strengths, and molecular symmetry and structure, and thus are very characteristic of the molecule (Hirschmugl 2002b). Hence, IR spectroscopy can be used to identify functional groups within a molecule (e.g., carboxylic acid), and because each compound has a unique molecular “fingerprint,” it can also be used to identify the molecule itself. Indeed, IR spectroscopy is frequently used for the identification of chemical compounds in unknown matrices. Biochemical molecules such as proteins, lipids, nucleic acids, and carbohyrates have unique IR spectra. For example, proteins have characteristic features labeled amide I bands (1600–1700 cm1)—[a carbonyl stretching mode and amide II bands (1500–1560 cm1)—a combined N---H bending–

356

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Figure 14.11. Various cellular components have dramatically different IR spectra as demonstrated by IR spectra of (a) a lipid (palmitic acid), (b) a protein (myoglobin), (c) a poly (nucleic acid), and (d) a carbohydrate (sucrose). For all spectra, films were prepared on BaF2 disks and 128 scans were collected at 4 cm1 resolution using a Globar source. [From Miller et al. (2003), reproduced with permission.]

C---N stretching vibration mode, along with more complex amide III, IV, and further bands. Note that to be IR-active, vibrational modes must involve changes in the permanent dipole moment. Thus, if there are symmetric structures around a site of interest, there will often be little dipole moment, and thus IR will be ineffective in such systems. Stem (2008) provides the following as an example, The nC---H, the C---H stretching mode, is silent in ethane and has nonactive bonds, but compounds with pendant methyl groups have a strong nC---H signal (Stem 2008). Figure 14.11 presents the IR spectra for a number of biologically relevant macromolecular species. Infrared microscopy achieves contrast through intramolecular vibrations and the intrinsic IR absorption bands. Jackson and Mantsch (2000) provide an overview of many relevant IR spectra. These spectra can, however, be influenced by sample history, including preparation techniques resulting in significant shifts from expected absorption peaks. [Nevertheless, these shifts can be a rich source of information regarding the sample in question. In addition to qualitative analysis, IR spectroscopy can also be used quantitatively to determine the amount of material present in a sample. (Hirschmugl 2002a,b; Stem 2008). 14.3.1.1. Instrumentation for and Access to FTIR Spectroscopy. Instruments for FTIR spectroscopy, imaging, and spectromicroscopy are commercially available, and they have been implemented at many synchrotron facilities worldwide. Infrared spectromicroscopy facilities in operation or

under construction include the National Synchrotron Light Source (NSLS, Brookhaven National Laboratory, USA), the Synchrotron Radiation Center (SRC, University of Wisconsin-Madison, USA), the Center for Advanced Microstructures and Devices (CAMD, Louisiana State University, USA), the Advanced Light Source (ALS, Lawrence Berkeley National Laboratory, USA), the Canadian Light Source (CLS, University of Saskatchewan, Canada), Diamond Light Source (Diamond, Rutherford Appleton Laboratory, UK), Australian Synchrotron (Melbourne, Victoria, Australia), ANKA Synchrotron Strahlunqsquelle (Karlsruhe, Germany), Berliner  r SynchrotronstrahElektronenspeicherring-Gesellschaft fu lung (BESSY, Berlin, Germany), Elettra Synchrotron Light Source (Trieste, Italy), MAX-lab (Lund, Sweden), National Synchrotron Radiation Center (NSRRC, Hsinchu, Taiwan), Singapore Synchrotron Light Source (SSLS, Singapore), Super Photon Ring (Spring8, Nishi-Harima, Japan), Swiss Light Source (SLS, Villigen, Switzerland), SOLEIL (SaintAubin, France), and Synchrotron Light Research Institute (SRLI) (Nakhon Ratchasima, Thailand). A complete list of synchrotrons around the world is available at http://www.als. lbl.gov/als/synchrotron_sources.html. In most cases there is a competitive, peer-reviewed access procedure that regulates the access to IR beamlines. In general, beamline scientists in charge of operations will provide assistance in setup and data collection—collaborative activities are also a common entry point to synchrotron research. 14.3.1.2. IR Spectromicroscopy. Infrared spectrophotometers consist of three main components: the light source, the optical path, and the detector. In the case of synchrotronbased IR, the conventional light source (Globar), is replaced by the high-brightness light emitted by a synchrotron or storage ring, and brought to the microscope and spectrophotometer via a beamline. Relative to Globar radiation, synchrotron IR has lower flux (integrated over all angles), but significantly higher brightness, by a factor of 100–1000 (Duncan and Williams 1983). Thus the main advantage of using synchrotron rather than lab-based FTIR microscopy is seen when the problem under study requires the highest possible spatial resolution. Typically an IR beamline operates under low vacuum ranging from 103 to 104 Torr with a diamond window (i.e., 0.5-mm-thick synthetic diamond) isolating the beamline from the high vacuum of the ring. Within the light path starting from the ring itself are a series of mirrors that act to direct the light to the microscope. Mirrors are the functional elements in an IR microscope because of the broad wavelength of IR light and the lack of IR-transparent materials from which to create lenses. The mirrors may be aluminized surfaces. The CLS, BESSY, and ALS the IR beamlines share similar designs and use combinations of planar, ellipsoidal, and cylindrical mirrors to direct the light to a 6-mm-thick KBr, CsI, or polyethylene window.

SYNCHROTRON-BASED FOURIER TRANSFORM INFRARED MICROSPECTROSCOPY

357

Figure 14.12. Schematic of the Spectra-Tech IrmsÔ scanning infrared microspectrometer. The system is a combined FTIR spectrometer and microscope, with an upper aperture to define the area being illuminated by the IR, and a lower aperture to limit the detector’s field of view onto the sample. The term confocal is often used to describe the optical configuration when both apertures are used. [After Miller et al. (2003), reproduced with permission.]

This light is then directed to a spectrometer, and then to piezoelectrically operated mirrors that direct the light through a microscope (May et al. 2007). Since IR light is influenced by water vapor and carbon dioxide, these are most often removed and the microscope is operated under a dry nitrogen or dry air environment. Figure 14.12 illustrates the typical arrangement of an IR beamline set up for FTIR microscopy. Commercially available IR microscopes from a number of companies (e.g., Thermo Nicolet, Bruker, Bio-Rad, PerkinElmer) can be adapted to a synchrotron IR source with very little modification. These microscopes are usually equipped with Schwarzschild objective/condenser pairs for imaging. The objective/condenser pair is arranged in a confocal configuration; one element focuses light on the specimen, while the other collects the light and relays it to the detector. Schwarzschild lenses range in magnification from 6 to 50 and have numerical apertures in the range of 0.3–0.7. In addition, the microscope has upper and lower apertures located between the specimen and the Schwarzschild objective and condenser; these act to limit the area of the specimen illuminated by the beam and the signal being detected. The combination of confocal lenses and apertures results in illumination free of chromatic aberration- and diffractionlimited imaging. The spatial resolution of FTIR spectromicroscopy is diffraction-limited, as it is dependent on the wavelength of light and the numerical aperture of the

focusing optic. The aperture controls the region that is illuminated by confining the beam to the sample’s area of interest. With a single aperture before the sample, the diffraction-limited spatial resolution is approximately 2l/3; thus, for the mid-IR range, it is 1.7 mm (at 4000 cm1) to 13 mm (at 500 cm1) (Carr 2001; Miller and Dumas, 2006). The addition of the second aperture after the sample further improves the spatial resolution of the microscope by up to l/2. Dumas and Miller (2003) describe a conventional synchrotron-based setup for IR microspectroscopy using a Nic-Plan IR microscope coupled to a Magna 560 FTIR spectrometer and a Continuum IR microscope coupled to a Nicolet Magna 860 FTIR spectrometer. As indicated above, a confocal mode was used with a focusing Schwarzschild objective 32 (NA ¼ 0.85) and a collection objective with 10 [numerical aperture (NA) ¼ 0.71] in combination with adjustable upper and lower apertures in the range of 3  3 mm2. The reader is referred to the literature for more details on instrument specifics (Smith 1979; Carr 1999; Carr 2001; Miller et al. 2003; Miller and Dumas 2006). The microscopes also provide a range of imaging options allowing the user to observe the sample with differential interference contrast, fluorescence, and polarized light. Typically the specimen is mounted on a motorized stage allowing the specimen to be moved precisely in the x–y for imaging, and data collection, using a mercury–cadmium– telluride (MCT-A) detector. Mercury–cadmium–telluride

358

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

(MCT) arrays [charge-coupled device (CCD) detectors] are available for the infrared region. They may be nitrogen cooled to increase sensitivity and improve the signal to noise ratio of the system (Mills et al. 2005). However, MCT detectors, do have issues with nonlinear response over the full range of brightness. It has been suggested that germanium photoconductor detectors offer a solution to this issue (Miller and Dumas 2006). Miller and Dumas (2006) discuss the application of focal plane array detectors in what is termed Fourier transform infrared microspectrometry imaging. In this approach there are an array of IR detectors that allow spectra from various regions of the sample to be collected simultaneously, thereby increasing the speed of collection. In conventional FTIR with a point detector there is either one or two apertures that act to limit the area illuminated and seen by the detector. With a focal plane array (FPA) there are no apertures; thus there is a loss of spatial resolution since the instrument cannot be operated as a confocal microscope. Further details regarding the schematics and operation of a confocal microscope as well as the use of FPA with synchrotron radiation may be obtained from Carr et al. (2005). A multielement detector (IRMSI-MED) for diffraction-limited chemical imaging using parallel acquisition of an array of spectra for microspectroscopic maps has been successfully used at the Synchrotron Radiation Centre (Nasse et al. 2007), greatly increasing the speed of data collection. Quantum structure infrared photodetector (QSIP) technology has been proposed to offer further advantages (LeVan and Beecken 2009). 14.3.1.3. Advantages of Synchrotron IR Source. Conventional light sources for IR spectroscopy have high signal-tonoise ratios (SNRs) that limit their practical resolution to 20 mm; however, this is due to the low intrinsic brightness of Globar sources rather than limitations of the optics of the microscope itself (Dumas and Miller 2003). Indeed, conventional FTIR has a high level of molecular information but lacks the desired spatial resolution, particularly for biological and some environmental applications. A synchrotron IR source is 100–1000 times brighter than a conventional thermal source (Duncan and Williams 1983; May et al. 2007). While a variety of high-brightness sources exist, only synchrotron light provides the full range of wavelengths in the infrared combined with a very small source size and narrow angles of emission (Miller and Dumas 2006). Miller and Tague (2002), state that the use of a high-brightness synchrotron light source dramatically improves spatial resolution, allowing collection of highquality data at the diffraction limit [3–25 mm in the mid-IR (400–4000 cm1)]. It is important to note, however, that if large aperture settings (> 20 mm) are used, then synchrotron-based IR has no advantage. In the final analysis, the major advantage of synchrotron-based IR spectromicroscopy over many other bioanalytical methods is its ability

to distinguish between subtle changes in overall biochemical composition, even in the absence of morphological differences (Jilkine et al. 2008). 14.3.1.4. Sample Preparation. In many cases, depending on the mode of data collection, there is relatively minimal sample preparation, allowing noninvasive in situ types of analyses with IR. However, special requirements may be necessary for transmission, reflection, grazing incidence, attenuated total reflectance, or photoacoustic spectroscopy. If the sample does not transmit adequate light for transmission, lacks reflectivity, or may not allow sufficient penetration of the sample, then special handling may be required. The most common approach to IR analysis is either transmission or reflectance, although the option exists to carry out a combination of the two methods. Transmission requires thin samples (1–10 mm). In many cases this can be achieved by spotting an appropriate aliquot of material onto an IR transmissive material and drying or maintaining the sample in a fully hydrated state prior to and even during examination. Some type of embedding (paraffin, epoxy, Nanoplast, polymethacrylate) and subsequent sectioning is frequently required. However, cryosectioning is a simpler, less transformative, and generally preferred method of sectioning for many microscopy-based studies. Miller et al. (2003), Miller and Dumas (2006), and Dumas and Miller (2003), in overviews of FTIR and its applications, present a number of sample types and their preparation methods, including bone that was embedded in polymethylmethacrylate and microtomed to 5 mm. The prepared samples were mounted in a compression cell (Thermo Spectra Tech, Shelton, CT) for data collection. They also embedded hair and skin in Tissue Tek (Reichert-Jung, Heidelberg, Germany) and cryosectioned at 5–7 mm. Alzheimer’s diseased brain material was snap-frozen in liquid nitrogen, and 15-mm-thick sections were cut with a cryomicrotome and placed on BaF2 disks. In most instances the prepared materials are placed on an IR transparent window or holder made from ZnS, BaF2, CaF2, KBr, ZnSe, or even diamond. Sections prepared for synchrotron IR do not exceed 30 mm in thickness. In reflection mode prepared sections may also be examined, although they are then mounted on an IR reflective surface such as a low-E glass slide (Kevley Technologies), and the signal collected is that reflected from the support surface after passing through the sample. Since reflection imaging requires extremely smooth surfaces to prevent degradation of the signal, polished sections of mineral, bone, or other materials are well suited to this approach. Miller (2010) provides useful suggestions regarding the mounting of samples on putty or a goniometer to allow optimizing the angle of incidence for reflection analyses of uneven samples. While reflectance is suitable for sections or polished materials at about 10 mm thickness, the application of grazing incidence is best for samples with < 1 mm thickness and

SYNCHROTRON-BASED FOURIER TRANSFORM INFRARED MICROSPECTROSCOPY

that are highly polished and conducive to high polish such as metals. An additional complication is that grazing incidence requires the use of specially designed Schwartzchild objectives that aid in limiting the penetration of the beam into the surface of material under investigation. Attenuated total reflection (ATR) infrared spectroscopy uses a high refractive index, infrared-transparent internal reflection element such as Ge, ZnSe, or diamond. This element provides a thin sampling region at its surface where it is in contact with the sample; thus the ATR material has to touch the sample to allow the IR to probe the sample and thereby determine its spectrum. This approach has been used in environmental studies to examine the chemical nature of sorption films by incubating the Ge element in water (Baier 1973), while Geesey and Suci (2000) describe this approach to study biofilms, biocorrosion, and the behavior of molecules at interfaces. Thomasson et al. (2000) described the measurement of IR microspectra utilizing a ATR objective containing a small crystal about 100 mm in diameter. This provided spectroscopic analyses of standard polished thin sections. Certainly there are limitations in terms of sample type, the spectral qualities of the element selected, the need for contact, potential damage to the element, cleaning issues, and so on as noted by Miller (2010). To date ATR does not appear to have been applied with synchrotron radiation but may offer potential. Photoacoustic spectroscopy offers many advantages and requires minimal preparation to obtain spectra for a range of sample types. It does not require transmission, is relatively insensitive to surface conditions of the sample, and can probe a sample with depths from a few micrometers >100 mm (Michaelian 2003). This approach has been applied to a number of solids and liquids, including coal, coke, bitumen, metal powders, wood products, and clays. However, some processing may be required such as grinding or powdering for highly variable samples or solvent extraction and drying as noted by McClelland et al. (2002). Photoacoustic infrared spectroscopy has also been implemented using synchrotron radiation instead of a thermal infrared source (Michaelian 2003; Michaelian et al. 2008). A number of specific sample preparation approaches that may have particular use for environmentally relevant studies have been outlined by various authors. Jilkine et al. (2008) used low-E microscope slides or gold-coated silicon wafers as substrates for fungal spores which were germinated in humidity chambers. The emerging hyphae grew out across the FTIR slides, and the authors froze and lyophilized the samples before data collection. As in other techniques (see Section 14.2), sulfur embedding and sectioning may be applied to environmental samples to obtain thin sections for FTIR studies (Schafer et al. 2009). Untreated rock samples embedded in sulfur, thin-sectioned (ultramicrotomed to 100 nm thickness) and placed on formvar grids can be used for both STXM and FTIR studies.

359

Many investigators transfer cultured cells, including human, cancer, Euglena, fungi, bacteria to gold-coated surfaces in water or culture media and maintain them in a living hydrated state during timecourses over several days (Hirschmugl 2002a; Dumas and Miller 2003; Holman et al., 2000, 2006; Szeghalmi et al. 2007). Holman et al. (1999), in a classic study, utilized a magnetite surface to carry Arthrobacter oxydans cells in a solution containing both chromium and ultimately toluene that was monitored via IR spectromicroscopy for several days. Marinkovic et al. (2000) describe a special flow cell with mixing properties that was specifically designed for FTIR studies. In a more recent study, using high-resolution FTIR, Holman et al. (2009) constructed a special high-humidity chamber to maintain essentially a 1-mm-thick biofilm of Desulfovibrio vulgaris and used synchrotron radiation–based Fourier transform infrared (FTIR) spectromicroscopy to determine the cellular chemical environment by continuously monitoring the dynamics of hydrogen bonding in cellular water in vivo. Holman and Martin (2006) refer to the development of automated microfluidic devices that allow manipulation of the microenvironment along with sensors to monitor relevant parameters and allow IR studies of biogeochemical processes in aqueous environments. Fourier transform infrared– compatible humidity cells for live cell visualization that enable real-time biochemical imaging at better than 5 mm spatial resolution are described by Jilkine et al. (2008). 14.3.1.5. Data Collection. Traditionally, infrared beamlines based at synchrotrons use a confocal configuration with raster scanning and a point detection system. However, as noted by a number of authors (Miller and Smith 2005), although the images are of high quality, the collection times are extremely long. While focal plane arrays offer improved collection speed, there are tradeoffs in terms of resolution (see discussion above). Nasse et al. (2007) indicated that a typical microspectroscopic map of 15  15 spectra with 128 scans each, at a resolution of 4 cm1 and a mirror velocity of 3.2 cm/s may be acquired in approximately 2.5 h using a point detection system. In contrast, using a multielement detector allows one to obtain chemical images with diffraction-limited resolution of the illuminated area in under a minute (Nasse et al. 2007). When combined with the fact that the intense IR synchrotron infrared beam does not result in heating of the samples and has no effect on living biological systems (Holman et al. 2002b; Martin et al. 2001), this allows real-time analyses and timecourses to be carried out. The user may start with collection of a reference spectrum in a sample-free area of the specimen holder, such as an IRreflective slide. Initial observations may be made at larger scale using an aperture open to 30 mm to survey the sample and select regions of interest for detailed mapping and taking specific spectra. Once a region is determined, the aperture

360

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

may be reduced to, for example, 12 mm or less for generating IR maps in 10 mm steps along the sample under study. 14.3.1.6. Analysis, Mapping, and Quantification. As is typical in spectromicroscopy, many hundreds of spectra are collected in each measurement, and thus it is a challenge to extract, analyze, and quantify the contained information. There can be considerable difficulty in extracting the signal signature from background/matrix responses in environmental samples, particularly if there is partial absorption saturation. Infrared spectroscopy allows the generation of images representing the chemical composition and distribution occurring with complex environmental samples. Most environmental samples are heterogeneous and complex; therefore, successful analysis, mapping, and quantification are highly dependent on the quality of the spectra acquired, in particular the signal-to-noise ratio, as well as the lateral resolution. Hirschmugl et al. (2006) describe a prescreening of spectra to ensure that intensities are within acceptable limits. Success may also be dependent on providing an effective blank, subtraction of culture medium/liquid water absorbance, or eliminating interferences [i.e., residual water vapor as in the case described by Holman et al. (2009)] and controlling for variations in source strength, due to the fill cycle of most synchrotrons. In this last regard, the recent trend to implementation of so-called top-off modes, where the current in the storage ring is maintained constant within a fraction of a percent, is particularly beneficial to synchrotronbased FTIR microscopy. Most IR microscopes are equipped with basic analytical packages, although typically these are not sufficient for the task at hand. Additional software is available from suppliers such as Thermo-Nicolet, Bruker, Spectral Dimensions, Cytospec, and Galactic. For example, Holman et al. (2009) performed all data processing with Thermo Electron’s Omnic 7.2 (http://www.thermo.com/) and Origin 6.0 (http:// www.originlab.com/). Application of Omnic software allows the user a range of options, such as subtraction of the substrate background (IR-reflective glass, etc.), automatic baseline correction, normalization, peak identification, and calculation of peak heights and areas. In their simplest application, IR spectra are used to evaluate the difference between two measurements in a controlled experiment; in this instance the user is assessing changes in lineshape and strength in a difference spectrum (Hirschmugl 2002a,b). The basic approaches to data analyses have been summarized by Miller (2010) and include the following: 1. Functional group mapping is the most common method for chemical imaging of samples. This is based on measuring peak height, peak area, or the peak height/area ratio. This provides a very basic approach to determining locations and quantity for identifiable compounds.

2. Second derivatives and Fourier self-deconvolution are used to identify peak frequencies. 3. Correlation methods are used to compare standard spectra to unknowns obtained from the sample wherein a perfect match has r ¼ 1.0. 4. Cluster analyses, which employ statistical methods, are used to evaluate spectral similarity in order to group spectra that may then be extracted or averaged and interpreted relative to known spectra. Spectral analyses may also benefit from the application of multivariate statistical analysis techniques such as linear discriminate and principal-component analyses (PCAs). The applications of cluster and PCA data treatments are driven in large measure by the collection of spatial datasets. For example, Schafer et al. (2009) applied both PCA and cluster analysis. Their results showed that humic acids were best correlated with organics found in smectite-rich regions, while fulvic acids grouped with the organics found in the illite mixed-layer materials. Hirschmugl et al. (2006) describe the application of agglomerative hierarchical clustering to FTIR datasets. This required prescreening, and conversion to first derivatives using a Savitzky–Golay algorithm, followed by normalization so that the sum-squared deviation equaled unity. The resulting cluster analysis calculates a distance matrix of the similarity of the spectra and uses these data to produce a dendrogram wherein the spectral sets are grouped according to their similarity. This allowed discrimination of Euglena gracilis cells on the basis of their nutritional history. Similarly, Yu (2007) found that they could use IR spectra and agglomerative hierarchical cluster analysis to distinguish inherent differences between maize and barley feed quality, as well as detecting the aleurone (protein, starch) and pericarp (lignin, cellulose) structural layers of seeds. Artificial neural network (ANN) methods are also being applied. General information on these methods may be obtained from a number of references (Jackson and Mantsch 1999; Naumann 2002; Chalmers and Griffiths 2002). It is felt that statistical pattern recognition techniques offer a simpler and unbiased approach to analyses of large spectral datasets (Hirschmugl et al. 2006; Lasch and Naumann 1988) relative to fitting or inspection techniques. Chemical maps are created from spectra by determining the intensity of an absorption band at each measuring position such that different bands create different maps. The peak height in the absorption spectra is proportional to the concentration of that functional group at a specific location. The user may obtain a quantitative impression of the map if object thickness can be obtained for each x–y location or by using a ratio method between peaks. Each peak or band may be assigned to a functional group. Timecourses may also be obtained with IR microspectroscopy to allow assessment of

SYNCHROTRON-BASED FOURIER TRANSFORM INFRARED MICROSPECTROSCOPY

changes with treatment or exposure, see for example (Holman et al. 2009) who created time-difference spectra using Thermo Electron’s Omnic 7.2 software (http://www.thermo. com/)] from experimental spectra after subtraction of culture medium/liquid water absorbance. 14.3.2. Examples of the Application of IR Microspectroscopy Although there are a variety of “environmental” applications of synchrotron-based infrared microspectroscopy, the series of publications by Holman in conjunction with various authors. (Several articles by Holman et al. 1997, 1998a,b, 1999, 2002a,b, 2009) represent an excellent series of examples. These studies illustrate the use of incubation methods and variations of the synchrotron-based IR approach to investigate inorganic–organic interactions at bacterial–mineral interfaces, localization of bacteria on geologic surfaces, evaluation of the effects of changing humidity on microbial activity, biogeochemical reduction of Cr(VI) on basalt surfaces, and catalysis of PAH biodegradation by humic acids. In addition to the development of the technique, these publications are also basic studies generating important insights into biogeochemical processes. 14.3.2.1. Real-Time Characterization of Biogeochemical Reduction of Cr(VI) on Basalt Surfaces by SR-FTIR Imaging. Holman et al. (1999) provide an excellent example of applying IR synchrotron imaging to monitor the reduction of Cr(VI) (oxidation state of Cr(III) compounds and the metabolism of toluene by bacterial colonies on a basalt surface, providing results on the oxidation state of Cr that were confirmed by micro-X-ray absorption fine-structure spectroscopy. Fourier transform IR spectromicroscopy was carried out on a model surface, magnetite, with the organism Arthrobacter oxydans, with and without toluene vapour as a carbon source and cocontaminant. Their biomass determinations and collocalization studies showed that endolithic bacteria were necessary to achieve a reduction of Cr(VI) to Cr(III) and that the presence of toluene as a carbon source enhanced the process. They observed a decrease in the concentration of the organic molecules at the same location as the living cells that were mapped relative to the toluene, based on the detection of amide absorption maxima (Holman et al. 1999). Although cell- or microcolony-level resolution remains an issue, this study demonstrated the capacity of IR microscopy to identify and quantitate molecules, including the mapping of metals, biological macromolecules, and chlorinated chemicals in a biofilm. In a related study, Holman et al. (2002a) used IR spectromicroscopy to examine the role of humic acids in the biodegradation of polycyclic aromatic hydrocarbons (PAHs) by a Mycobacterium species. The bacterium was

361

grown on magnetite surfaces in a specially constructed stage-mounted miniatureincubator. To initiate the experiments, they exposed the inoculated magnetite surfaces to pyrene, which sorbed and diffused into the magnetite. In some cases 300 ppm of filter-sterilized humic acid was applied on top of the pyrene. The degradation of pyrene was monitored in situ over time by returning to a series of selected sample locations and collecting a time series of spectra. Humic acid dramatically shortened the onset of pyrene biodegradation from 168 to 2 h. There was an increase in biomass absorption during the latter stage of pyrene degradation, implying that biomass formation was concurrent with the consumption of pyrene. The role of HA is to speed up the solubilization of the pyrene, making it bioavailable for degradation. It appears that the waterinsoluble pyrene is solubilized into the cores of HA pseudomicelles, thereby becoming directly available for bacterial uptake and consumption. Figure 14.13 illustrates the results obtained by IR mapping obtained at the end of the experiment with IR absorption peaks corresponding to the Mycobacterium sp., elliott soil humic acid (ESHA) and pyrene. The images show a region with a high density of bacteria and ESHA at locations where pyrene has been degraded by the bacteria. The study demonstrated that nontoxic humic acid may be a useful alternative to other surfactants in enhancing in situ remediation of poorly soluble contaminants. From the perspective of synchrotron-based FTIR microspectrometry, this study demonstrated the capacity to monitor a number of factors in real time under in situ conditions without damage from the IR data collection. However, there are also clear limitations in that only highly selected areas, that is areas that are very flat and within the focal limits, could be included in the monitoring, and the maps produced had 5 mm resolution. 14.3.2.2. Synchrotron Fourier Transform IR Microspectroscopy: A New Tool for Monitoring the Fate of Organic Contaminants in Plants. Phytoremediation is a process wherein plants absorb and biodegrade antheopogenic pollutants from contaminated soil. Dokken et al. (2005a) explored the use of IR spectromicroscopy to examine the fate of contaminant directly in plant material. The authors used FTIR spectromicroscopy to monitor the fate and effects of 2,6-dinotrotoluene (2,6-DNT) in corn (Zea mays) roots. Seedlings were grown hydroponically and exposed to 0, 5, 10, and 15 mg/L of 2,6-dinotrotoluene. Root material was frozen in Tissue-Tek OCT (Sakura Finetek USA, Inc., Torrance, CA) at 40 C, and 4-mm sections were prepared and mounted on IR reflecting low-E glass microscope slides. The IR imaging indicated that 2, 6-DNT may be incorporated in the lignin in corn plants. The detection of the nitro peak at 1530 cm1, CH3 peaks, and bands typical of aromatic ring structures are indicative of a lack of transformation of the 2,6-DNT prior to incorporation.

362

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Figure 14.13. Contour diagrams from infrared mapping obtained at the end of the experiment showing the spatial distribution of the infrared absorption peaks corresponding to (a) Mycobacterium sp. JLS bacteria, (b) elliott soil humic acid (ESHA), and (c) pyrene. Appropriate spectral regions were integrated for each point on the maps. The color scales for each contour plot are red for highly integrated IR peak area (high concentration of the corresponding component) and blue for low peak area (low concentration); black is an out-of-focus region of the sample. The center of the map shows a region with high density of bacteria and high concentration of ESHA where pyrene has been completely degraded. [Reprinted from Holman et al. (2002a).] (See insert for color representation of this figure.)

SYNCHROTRON-BASED FOURIER TRANSFORM INFRARED MICROSPECTROSCOPY

Similar results were obtained for 1 H-benzotriazole in the root tips of sunflowers. Here spectral analyses using principal component methods again suggested that the aromatic ring structure remained intact during absorption, transport and sequestration in plant tissues (Dokken et al. 2005b). Given that modifications to the surrounding plant tissue could also be monitored, this approach was concluded to be a useful option for direct detection of the fate and effects of contaminants in plant tissues. 14.3.2.3. Organic Matter Stabilization in Soil Microaggregates: Implications from Spatial Heterogeneity of Organic Carbon Contents and Carbon Forms. Lehman et al. (2005) used a combination of FTIR and STXM (NEXAFS) microscopies to map the distribution of carbon and its chemical forms in intact soil microaggregates (20–250 mm). The microaggregates were saturated with water, frozen, and sectioned (300–600 nm thickness) without embedding before being placed on copper grids and air-dried. Then STXM analyses were carried out using the Stony Brook STXM at NSLS BL X1A at Brookhaven National Laboratory by collecting stacks between 280 and 310 eV at increments varying between 0.3 and 0.1 eV. The authors then used these stacks to map the carbon amounts in the aggregate sections. Fourier transform IR was also carried out at the NSLS on the U108 beamline using the Spectra Tech Continuum IR microscope with a mercury–cadmium–telluride detector with 500–4000 cm1 wavenumber range and 1.0 cm1 spectral resolution. Mapping was carried out with a 7 mm aperture and a step size of 6 mm or 4000–650 cm1 at intervals of 4 cm1; 256 scans were added prior to Fourier transformation. They then assessed the spectra for peaks corresponding to stretching vibrations indicative of kaolinite, aliphatic biopolymers, aromatic carbon, and polysaccharides. This allowed them to examine the distribution of total carbon as well as the specific carbon forms in these microaggregates. Although the distribution of carbon based on C1s NEXAFS at 0.05 mm resolution appeared random, the types of carbon mapped using FTIR spectromicroscopy mapped at 5 mm resolution exhibited spatial patterns. Although the authors applied this approach to determine the developmental patterns and stabilization of soil aggregates, it could be easily extended to include mapping of associated metals, or organic contaminants providing information important to fate, effects and remediation schemes. Fourier transform IR spectromicroscopy also has tremendous potential in the field of toxicology, where it has seen direct applications, as well as studies suggestive of this use. See, for example, Holman et al. (2002b), who demonstrated the use of synchrotron IR spectromicroscopy to examine individual living cells. Holman et al. (2000) applied this approach to examine the response of single cells to a toxic organic contaminant Jilkine et al. (2008)

363

used FTIR spectromicroscopy to study differences between optimal and mildly stressed fungi (pH 5.0–8.5), while Szeghalmi et al. (2007) detected subcellular changes accompanying thermal stress in a variety of fungi. In this latter case the authors used specific gene mutations. They noted that FTIR spectromicroscopy facilitated interpretation of the biochemical effect of the gene in the context of the fungal hyphae. In a related study Hirschmugl et al. (2006) were able to spatially resolve the green flagellate Euglena gracilis and collect complete spectra for individual cells, which allowed them to discriminate cells on the basis of their nutritional status. Earlier studies had demonstrated the efficacy of FTIR spectromicroscopy in studying the biochemical composition of algae (Giordano et al. 2001). Fourier transform infrared has also proved useful in the characterization of wax deposits formed in oil-producing wells. Comparison of spectra from nonpolar and predominantly polar aggregates indicated that the polar aggregates contain higher relative concentrations of an inorganic material, most likely CaCO 3 (Marinkovic et al. 2002). 14.3.3. Future Developments in Synchrotron-Based FTIR Microscopy Lombi and Susini (2009) provide a fairly exhaustive review of the literature relevant in the main to X-ray microscopy techniques, including STXM, but also with reference to FTIR and other synchrotron-based techniques. In addition, they consider future directions and required developments quite extensively. In general, it is anticipated that there will continue to be considerable improvement in optics, detectors, and software related to FTIR spectromicroscopy and other synchrotron techniques, including STXM. New beamlines and microscopes are able to provide improved signal-tonoise ratios and increased spatial resolution, as well as improved acquisition times allowing real-time imaging of biological and biogeochemical processes of interest to environmental sciences. Developments such as the ID21 beamline at the ESRF, which is a combination of STXM and FTIR microscopes, is an excellent step in the necessary process of integrating FTIR with other synchrotron-based microspectrometric techniques (Lombi and Susini 2009; Dumas and Miller 2003). In addition to technical developments in beamlines and microscopes, further developments in sample handling and preparation as well as functional model systems and incubation systems are required. Another trend is that of integration of synchrotron analytical microscopies with genomic and other emerging experimental and analytical approaches. Owing to the inherent interdisciplinary nature of environmental science, only a combination of different techniques will provide a complete picture of environmental processes.

364

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

14.3.4. Summary and Conclusions Related to Synchrotron-Based FTIR Microscopy Fourier transform infrared spectroscopy (FTIR) is an analytical technique that allows examination of a variety of sample types, including solids, liquids, gases, and their mixtures. In addition, FTIR facilitates the analyses of a wide range of biological materials providing structural and chemical analyses. The method allows identification of most organic compounds. Indeed, outside of magnetic resonance techniques, FTIR is perhaps the most powerful tool for identifying types of chemical bonds. State-of-the-art FTIR spectromicroscopy in the mid-IR region (0.01–1 eV) allows the production of 2D spectral maps of IR absorption, with a spatial resolution on the order of a few micrometers. Analysis of the absorption signatures permits identification and mapping of the chemical compounds present. Infrared spectroscopy is especially sensitive to vibrations of polar bonds such as O---H, C---H, C---O, N---H, and C---N. Fourier transform IR spectromicroscopy can identify and quantitatively map molecules such as metals, biological macromolecules, and chlorinated chemicals, in environmental samples. Infrared spectroscopy is able to detect subtle biochemical and biogeochemical changes within a range of environmental samples and is able to deal with the relatively high heterogeneity present in environmentally relevant materials and model systems. However, the spatial resolution remains limited. Developments since circa 1990, particularly the use of synchrotron radiation, as a high brightness, full spectral source, have revolutionized FTIR and X-ray spectromicroscopy, creating techniques suitable for high-resolution imaging and spectroscopy and powerful research tools for studying organic contaminants in the environment.

REFERENCES Ade, H. (1998), X-ray spectromicroscopy, in Experimental Methods in the Physical Sciences, Vol. 32, Samson, J. A. R. and Ederer, D. L., eds., Academic Press, New York, pp. 225–262. Ade, H. and Urquhart, S. G. (2002), NEXAFS spectroscopy and microscopy of natural and synthetic polymers, in Chemical Applications of Synchrotron Radiation, Sham, T. K. ed., World Scientific Publishing, Singapore, pp. 285–355. Ade, H. and Hitchcock, A. P. (2008), NEXAFS microscopy and resonant scattering: Composition and orientation probed in real and reciprocal space, Polymer 49, 643–675. Baier, R. E. (1973), Influence of the initial surface condition of materials on bioadhesion, in Acker, R. F. ed., Proc. 3rd Int. Congress Marine Corrosion and Fouling, Northwestern University Press, Evanston, IL, pp. 633–639. Bertsch, P. M. and Hunter, D. B. (2001), Applications of synchrotron-based X-ray microprobes, Chem. Rev. 101(6), 1809–1842.

Bluhm, H., Andersson, K., Araki, T., Benzerara, K., Brown, G. E., Dynes, J. J., Ghosal, S., Gilles, M. K., Hansen, H.-Ch., Hemminger, J. C., Hitchcock, A. P., Ketteler, G., Kilcoyne, A. L. D., Kneedler, E., Lawrence, J. R., Leppard, G. G., Majzlam, J., Mun, B. S., Myneni, S. C. B., Nilsson, A., Ogasawara, H., Ogletree, D. F., Pecher, K., Salmeron, M., Shuh, D. K., Tonner, B., Tyliszczak, T., Warwick, T., and Yoon, T. H. (2005), Soft Xray microscopy and spectroscopy at the molecular environmental science beamline at the advanced light source, J. Electron. Spectrosc. Relat. Phenom. 150, 86–104. Bradley, J. P., Keller, L., Thomas, K. L., Vander Wood, T. B., and Brownlee, D. E. (1993), Carbon analyses of IDPs sectioned in sulfur and supported on beryllium films (abstract), Proc. 24th Lunar and Planetary Science Conf. p. 173. Braun, A. (2005), Carbon speciation in airborne particulate matter with C (1s) NEXAFS spectroscopy, J. Environ. Monit. 7, 1059–1065. Braun, A., Huggins, F. E., Kubatova, A., Wirick, S., Maricq, M. M., Mun, B. S., McDonald, J. D., Kelly, K. E., Shah, N., and Huffman, G. P. (2008), Toward distinguishing woodsmoke and diesel exhaust in ambient particulate matter, Environ. Sci. Technol. 42, 374–380. Braun, A., Shah, N., Huggins, F. E., Huffman, G. P., Wirick, S., Jacobsen, C., Kelly, K. E., and Sarofim, A. (2004), Study of fine diesel particulate matter with scanning transmission X-ray spectroscopy, Fuel 10(7/8), 997–1000. Brown, G. E. Jr. and Sturchio, N. C. (2002), An overview of synchrotron radiation applications to low temperature geochemistry and environmental science, in Applications of Synchrotron Radiation in Low-Temperature Geochemistry and Environmental Science, Fenter, P., Rivers, M., Sturchio, N. C., and Sutton, S. eds., Reviews in Mineralogy and Geochemistry (series) Vol. 49, Geochemical Society, pp. 149–220. Carr, G. L. (2001), Resolution limits for infrared microscopy explored with synchrotron radiation, Rev. Sci. Instrum. 72, 1613–1619. Carr, G. L. (1999), High-resolution microspectroscopy and subnanosecond time-resolved spectroscopy with the synchrotron infrared source, Vlb. Spectrosc. 19, 53–60. Carr, G. L., Chubar, O., and Dumas P. (2005), Multichannel detection with a synchrotron light source: Design and potential, in Spectrochemical Analysis Using Infrared Multichannel Detectors, Bhargava, R. and Levin, I. eds., Blackwell Publishing, pp. 56–83. Chalmers, J. M. and Griffiths, P. R. (2002), Handbook of Vibrational Spectroscopy, Wiley, Hoboken, NJ. Claret, F., Sch€afer, T., Brevet, J., and Reiller, P. E. (2008), Fractionation of Suwannee River fulvic acid and Aldrich humic acid on alpha-Al2O3: Spectroscopic evidence, Environ. Sci. Technol. 42, 8809–8815. Cody, G. D, Brandes, J., Jacobsen, C., and Wirick, S. (2009), Soft Xray induced chemical modification of polysaccharides in vascular plant cell walls, J. Electron Spectrosc. Relat. Phenom. 170, 57–64. Cody, G. D., Botto, R. E., Ade, H., Behal, S., Disko, M., and Wirick, S. (1995), Microanalysis and scanning X-ray microscopy of

REFERENCES

microheterogeneities in a high-volatile bituminous coal, Energy Fuels 9, 75–83. Dokken, K. M., Davis, L. C., and Marinkovic, N. S. (2005a), Using SR-IMS to study the fate and transport of organic contaminants in plants, Spectroscopy 20, 14–20. Dokken, K. M., Davis, L. C., Erickson, L. E., Castro-Diaz, S., and Marinkovic, N. S. (2005b), Synchrotron Fourier transform infrared microspectroscopy: A new tool to monitor the fate of organic contaminants in plants, Microchem. J. 81, 86–91. Dumas, P. and Miller, L. M. (2003), The use of synchrotron infrared microspectroscopy in biological and biomedical investigations, Vibrat. Spectrosc. 32, 3–21. Duncan, W. and Williams, G. P. (1983), Infrared synchrotron radiation from electron storage rings, Appl. Opt. 22, 2914–2923. Dynes, J. J., Lawrence, J. R., Korber, D. R., Swerhone, G. D. W., Leppard, G. G., and Hitchcock, A. P. (2006a), Quantitative mapping of chlorhexidine in natural river biofilms, Sci. Total Environ. 369, 369–383. Dynes, J. J., Tyliszczak, T., Araki, T, Lawrence, J. R., Swerhone, G. D. W., Leppard, G. G., and Hitchcock, A. P. (2006b), Quantitative mapping of metal species in bacterial biofilms using scanning transmission X-ray microscopy, Environ. Sci. Technol. 40, 1556–1565. Dynes, J. J., Lawrence, J. R., Obst, M., Swerhone, G. D. W., Korber, D. R., Leppard, G. G., Tyliszczak, T., and Hitchcock, A. P. (2009), Soft X-ray spectromicroscopy of nickel sorption in a natural river biofilm, Geobiology 7, 432–453. Feser, M., Beetz, T., Carlucci-Dayton, M., and Jacobsen, C. (2000), Instrumentation advances and detector development with the Stony Brook scanning transmission X-ray microscope, AIP Conf. Proc. 507, 367–372. Geesey, G. G. and Suci, P. A. (2000), Monitoring biofilms by Fourier transform infrared spectroscopy, in Biofilms: Recent Advances in Their Study and Control, Evans, L.V. ed., Harwood Academic Publishers, Chur, Switzerland, pp. 253–277. Gilbert, E. S., Khlebnikov, A., Meyer-Ilse, W. and Keasling, J. D. (1999), Use of soft X-ray microscopy for the analysis of early-stage biofilm formation, Water Sci. Technol. 39, 269–272. Giordano, M., Kansiz, M., Haeraud, R., Beardall, J., Wood, B., and McNaughton, D. (2001), Fourier transform infrared spectroscopy as a novel tool to investigate changes in intracellular macromolecular pools in the marine microalgae Chaetorceros muellerii (Bacillariophyceae), J. Phycol. 37, 271–279. Gu, W. W., Etkin, L. D., Le Gros, M. A., and Larabell, C. A. (2007), X-ray tomography of Schizosaccharomyces pombe, Differentiation 75, 529–535. Henke, B. L., Gullikson, E. M., and Davis, J. C. (1993), At Data Nucl Data Tables, 54, 181–297. Hirschmugl, C. J. (2002a), Applications of storage ring infrared spectromicroscopy and reflection-absorption spectroscopy to geochemistry and environmental science, Rev. Miner. Geochem. 49, 317–339. Hirschmugl, C. J. (2002b), Frontiers in infrared spectroscopy at surfaces and interfaces, Surf. Sci. 500, 577–604.

365

Hirschmugl, C. J., Bayarri, Z.-E, Bunta, M., Holt, J. B., and Giordano, M. (2006), Analysis of the nutritional status of algae by Fourier transform infrared chemical imaging, Infrared Phys. Technol. 49, 57–63. Hitchcock, A. P., Araki, T., Ikeura-Sekiguchi, H., Iwata, N., and Tani, K. (2003), 3d chemical mapping of toners by serial section scanning transmission X-ray microscopy, J. Phys. IV France 104, 509–512. Hitchcock, A. P., Dynes, J. J., Johansson, G. A., Wang, J., and Botton, G. (2008a), Comparison of NEXAFS microscopy and TEM-EELS for studies of soft matter, Micron 39, 741–748. Hitchcock, A. P., Johansson, G. A., Mitchell, G. E., Keefe, M., and Tyliszcak, T. (2008b) , 3D chemical imaging using angle-scan tomography in a soft X-ray scanning transmission X-ray microscope, Appl. Phys. A 92, 447–452. Hitchcock, A. P., Morin, C., Zhang, X., Araki, T., Dynes, J. J., St€ over, H., Brash, J. L., Lawrence, J. R., and Leppard, G. G. (2005), Soft X-ray spectromicroscopy of biological and synthetic polymer systems, J. Electron Spectrosc. Relat. Phenom. 144–147 259–269. Hitchcock, A. P. (2009), aXis2000 Is an IDL-Based Analytical Package; available at http://unicorn.mcmaster.ca/aXis2000. html. Hitchcock, A. P., Morin, C., Heng, Y. M., Cornelius, R. M., and Brash, J. L. (2002), Towards practical soft X-ray spectromicroscopy of biomaterials, J. Biomater. Sci. Polym. Educ. 13, 919–938. Holman, H.-Y. N., Perry, D. L., and Hunter-Cevera, J. C. (1997), Use of infrared microspectroscopy to assess effects of relative humidity on microbial population and their activity in fractured rocks, Proc. Gordon Conf. Applied Environmental Microbiology, Aug. 17–22 1997. Holman, H.-Y. N., Perry, D. L., and Hunter-Cevera, J. C. (1998a), Surface-enhanced infrared absorption reflectance SEIRA.microspectroscopy for bacteria localization on geologic material surfaces, J. Microbiol. Meth. 34, 59–71. Holman, H.-Y. N., Perry, D. L., Martin, M. C., Lamble, G. M., McKinney, W. R., and Hunter-Cevera, J. C. (1999), Real time characterization of biogeochemical reduction of Cr VI.on basalt surfaces by SR-FTIR imaging, Geomicrobiol. J. 16, 307–323. Holman, H.-Y. N., Perry, D. L., Martin, M. C., and McKinney, W. R. (1998b), Applications of synchrotron infrared microspectroscopy to the study of inorganic–organic interactions at the bacterial–mineral interface, Appl. Synchrotron Radiat. Tech. Mater. Sci. MRS Symp. Ser. 54, 17–24. Holman, H.-Y. N. and Martin, M. C. (2006), Synchrotron radiation infrared spectromicroscopy: A non-invasive molecular probe for biogeochemical processes, Adv. Agron. 90, 79–127. Holman, H.-Y. N., Nieman, K., Sorensen, D. L., Miller, C. D., Martin, M. C., Borch, T., Mckinney W. R., and Sims, R. C. (2002a), Catalysis of PAH biodegradation by humic acid shown in synchrotron infrared studies, Environ. Sci. Technol. 36, 1276–1280. Holman, H.-Y. N., Bjornstad, K. A., McNamara, M. P., Martin, M. C., McKinney, W. R., and Blakely, E. A. (2002b), Synchrotron infrared spectromicroscopy as a novel bioanalytical microprobe

366

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

for individual living cells: Cytotoxicity considerations, J. Biomed. Opt. 7, 417–424. Holman, H.-Y. N., Goth-Goldstein, R., Martin, M. C., Russell, M. L., and McKinney, W. R. (2000), Low-dose responses to 2,3,7,8tetrachlorodibenzo-p-dioxin in single living human cells measured by synchrotron infrared spectromicroscopy, Environ. Sci. Technol. 34, 2513–2517. Holman, H-Y. N., Wozei, E., Lin, Z., Comolli, L. R., Ball, D. A., Borglin, S., Fields, M. W., Hazen. T. C., and Downing, K. H. (2009), Real-time molecular monitoring of chemical environment in obligate anaerobes during oxygen adaptive response. Proceedings of the National Academy of Sciences of the United States of America, 106(31),12599–12604. Howells, M., Jacobsen, C., and Warwick, T. (2007), Principles and applications of zone plate X-ray microscopes, in The Science of Microscopy, Hawkes, P. and Spence, J. eds., Kluwer Press, pp. 835–926. Hunter, R. C., Hitchcock, A. P., Dynes, J. J., Obst, M., and Beveridge, T. J. (2008), Mapping the speciation of iron minerals in Pseudomonas aeruginosa biofilms using scanning transmission x-ray microscopy, Environ. Sci. Technol. 42, 8766–8772. Jackson, M. and Mantsch, H. H. (1999), Ex vivo tissue analysis by infrared microspectroscopy, in Handbook of Vibrational Spectroscopy, Chalmers, J. and Griffiths, P., eds., Wiley, New York. Jackson, M. and Mantsch, H. H. (2000), Encyclopedia of Analytical Chemistry, Wiley, Chichester, UK. Jacobsen, C. (2009), Analysis software and manuals; available at http://xray1.physics.sunysb.edu/. Jacobsen, C., Feser, M., Lerotic, M., Vogt, S., Maser, J., and T. Sch€afer. (2003), Cluster analysis of soft x-ray spectromicroscopy data, J. Phys. IV 104, 623–626. Jacobsen, C., Kirz, J., and Williams, S. (1992), Resolution in soft X-ray microscopes, Ultramicroscopy, 47 55–79. Jacobsen, C., Wirick, S., Flynn, G., and Zimba, C. (2000), Soft Xray spectroscopy from image sequences with sub-100 nm spatial resolution, J. Microsc. 197, 173–184. Jacobsen, C., Williams, S., Anderson, E., Browne, M. T., Buckley, C. J., Kern, D., Kirz, J., Rivers, M., and Zhang, X. (1991), Diffraction-limited imaging in a scanning transmission x-ray microscope, Optics Communications, 86, 351–364. Jilkine, K., Gough, K. M., Julian, R., and Kaminskyj, S. G. W. (2008), A sensitive method for examining whole-cell biochemical composition in single cells of filamentous fungi using synchrotron FTIR spectromicroscopy, J. Inorg. Biochem. 102, 540–546. Johansson, G. A., Dynes, J. J., Hitchcock, A. P., Tyliszczak, T., Swerhone, G. D. W., and Lawrence, J. R. (2006), Chemically sensitive 3D imaging at sub 100 nm spatial resolution using tomography in a scanning transmission x-ray microscope, in Developments in X-Ray Tomography V, Bonse, U. ed., Proc. SPIE, 6318, 11. Johansson, G. A., Tyliszczak, T., Mitchell, G. E., Keefe, M., and Hitchcock, A. P. (2007), Three dimensional chemical mapping by scanning transmission X-ray spectromicroscopy, J. Synchrotron Radiat. 14, 395–412.

Kilcoyne, A. L. D., Steele, W. F., Fakra, S, Hitchcock, P., Franck, K., Anderson, E., Harteneck, B., Rightor, E. G., Mitchell, G. E., Hitchcock, A. P., Yang, L., Warwick, T., and Ade, H. (2003), Interferometrically controlled scanning transmission microscopes at the Advanced Light Source, J. Synchrotron Radiat. 10, 125–136. Kinyangi, J., Solomon, D., Liang, B., Lerotic, M., Wirick, S., and Lehmann, J. (2006), Nanoscale biogeocomplexity of the organomineral assemblage in soil: Application of STXM microscopy and C 1s-NEXAFS spectroscopy, Soil Sci. Soc. Am. J. 70, 1708–1718. Kirz, J., Jacobsen, C., and Howells, M. (1995), Soft X-ray microscopes and their biological applications, Q. Rev. Biophys. 28, 33–130. Kirz, J. and Rarback, H. (1985), Soft X-ray microscopes, Rev. Sci. Instrum. 56, 1–13. Koprinarov, I. N., Hitchcock, A. P., McCrory, C., and Childs, R. F. (2002), Quantitative mapping of structured polymeric systems using singular value decomposition analysis of soft X-ray images, J. Phys. Chem. B 106, 5358–5364. Koprinarov, I. N., Hitchcock, A. P., Li, W. H., Heng, Y. M., and St€ over, H. D. H. (2001), Quantitative compositional mapping of core-shell polymer microspheres by soft X-ray spectromicroscopy, Macromolecules 34, 4424–4429. Larabell, C. A. and Le Gros, M. A. (2004), X-ray tomography generates 3D reconstructions of the yeast, Saccharomyces cerevisiae, at 60-nm resolution, Molec. Biol. Cell 15, 957–962. Lasch, P. and Naumann, D. (1988), FT-IR microspectroscopic imaging of human carcinoma thin sections based on pattern recognition techniques, Cell. Molec. Biol. 44, 189–202. Lawrence, J. R, Swerhone, G. D. W., Leppard, G. G., Araki, T., Zhang, X., West, M. M., and Hitchcock, A. P. (2003), Scanning transmission x-ray, laser scanning and transmission electron microscopy mapping of the exopolymeric matrix of microbial biofilms, Appl. Environ. Microbiol. 69, 5543–5554. Lehmann, J., Liang, B., Solomon, D., Lerotic, M., Luizao, F., Kinyangi, J., Schafer, T., Wirick, S., and Jacobsen, C. (2005), Near-edge X-ray absorption fine structure (NEXAFS) spectroscopy for mapping nano-scale distribution of organic carbon forms in soil: Application to black carbon particles, Global Biogeochem. Cycles 19, 1013–1025. Lerotic, M., Jacobsen, C., Gillow, J. B., Francis, A. J., Wirick, S., Vogt, S., and Maser, J. (2005), Cluster analysis in soft X-ray spectromicroscopy: Finding the patterns in complex specimens, J. Electron Spectrosc. Relat. Phenom. 144–147C, 1137–1143. Lerotic, M., Jacobsen, C., Sch€afer, T., and Vogt, S. (2004), Cluster analysis of soft x-ray spectromicroscopy data, Ultramicroscopy 100, 35–57. Leung, B. O., Hitchcock, A. P., Brash, J. L., Scholl, A., Doran, A., Henklein, P., Overhage, J., Hilpert, K., Hale, J. D., and Hancock, R. E. W. (2008), An X-ray spectromicroscopy study of competitive adsorption of protein and peptide onto polystyrene-poly (methyl methacrylate), Biointerphases 3, F27–F35. LeVan, P. D. and Beecken B. P. (2009), Hyperspectral opportunities with next-generation QSIP arrays, Infrared Phys. Technol. 52, 361–363.

REFERENCES

Lombi, E. and Susini, J. (2009), Synchrotron-based techniques for plant and soil science: Opportunities, challenges and future perspectives, Plant Soil 320, 1–35. Loo, Jr. B. W., Sauerwald, I. M., Hitchcock, A. P., and Rothman, S. S. (2001), A new sample preparation method for soft X-ray microscopy: Nitrogen based contrast and radiation tolerance properties of glycol methacrylate-embedded and sectioned tissue, J. Microsc. 204, 69–86. Marinkovic, N. S., Huang, R., Bromberg, P., Sullivan, M., Toomey, J., Miller, L. M., Sperber, E., Moshe, S., Jones, K. W., Chouparova, E., Lappi, S., Franzen, S., and Chance, M. R. (2002), Center for synchrotron biosciences’ U2B beamline: An international resource for biological infrared spectroscopy, J. Synchrotron Radiat. 9, 189–197. Marinkovic, N. S., Adzic, A. R., Sullivan, M., Kovacs, K., Miller, L. M., Rousseau, D. L., Yeh, S. R., and Chance, M. R. (2000), Design and implementation of a rapid-mix flow cell for timeresolved infrared microspectroscopy, Rev. Sci. Instrum. 71, 4057–4060. Martin, M. C., Tsvetkova, N. M., Vrowe, J. H., and McKinney, W. R. (2001), Negligible sample heating from synchrotron infrared beam, Appl. Spectrosc. 55, 111–113. May, T., Ellis, T., and Reininger, R. (2007), Mid-infrared spectromicroscopy beamline at the Canadian Light Source. Nucl. Instrum. Meth. Phys. Res. A 582, 111–113. McClelland, J. F., Jones, R. W., and Bajic, S. J. (2002), FT-IR photoacoustic spectroscopy, in Handbook of Vibrational Spectroscopy., Chalmers, J. and Griffiths, P. eds., Wiley, Hoboken, NJ. Michaelian, K. H. (2003), Photoacoustic Infrared Spectroscopy, Wiley-Interscience, Hoboken, NJ. Michaelian, K. H., May, T. E., and Hyett, C. (2008), Photoacoustic infrared spectroscopy at the Canadian Light Source: Commissioning experiments, Rev. Sci. Instrum. 79, 1–5. Miller, L. M. (2010), available at http://www.nsls.bnl.gov/newsroom/ publications/otherpubs/imaging/workshopmiller.pdf. Miller, L. M. and Dumas, P. (2006), Chemical imaging of biological tissue with synchrotron infrared light, Biochim. Biophys. Acta 1758, 846–857. Miller, L. M., Smith, G. D., and Carr, G. L. (2003), Synchrotronbased biological microspectroscopy: From the mid- to the farinfrared regimes, J. Biol. Phys. 29, 219–230. Miller, L. M. and Tague, T. J. (2002), Development and Biomedical Applications of Fluorescence assisted Synchrotron Infrared Micro-spectroscopy, Vibrational Spectroscopy 849, 1–7. Miller, L. M. and Smith, R. J. (2005), Synchrotrons vs. Globars, Point-Detectors vs. Focal Plane Arrays: Selecting the Best Source and Detector for Specific Infrared Microspectroscopy and Imaging Applications, Vibrational Spectroscopy, 38, 237–240. Mills, E. N. C., Parker, M. L., Wellner, N., Toole, G., Feeney, K., and Shewry, P. R. (2005), Chemical imaging: The distribution of ions and molecules in developing and mature wheat grain, J. Cereal Sci. 41, 193–201. Mitrea, G., Thieme, J., Guttmann, P., Heim, S., and Gleber, S. (2008), X-ray spectromicroscopy with the scanning transmission X-ray microscope at BESSY II, J. Synchrotron Radiat. 15, 26–35.

367

Nasse, M. J., Reininger, R., Kubala, T., Janowski, S., and Hirschmugl, C. (2007), Synchrotron infrared microspectroscopy imaging using a multi-element detector (IRMSI-MED) for diffraction-limited chemical imaging, Nucl. Instrum. Meth. Phys. Res., Sect. A 58, 107–110. Naumann, D. (2002), Infrared spectroscopy in microbiology, in: Mayers, R. A. ed., Encyclopedia of Analytical Chemistry, Wiley, Chickesster, UK, pp. 102–131. Naumann, D., Schultz, C.P., and Helm, D. (1996), What can infrared spectroscopy tell us about the structure and composition of intact bacterial cells? In Infrared spectroscopy of biomolecules, H.H. Mantsch and D. Chapman, eds., pp. 279–310. Wiley-Liss, New York. Neu, T. R., Manz, B., Volke, F., Dynes, J. J., Hitchcock, A. P., and Lawrence, J. R. (2010), Advanced imaging techniques for assessment of structure, composition and function in biofilm systems, FEMS Microbiol. Ecol. 72, 1–21. Noll, F., Sumper, M., and Hampp, N. (2002), Nanostructure of diatom silica surfaces and of biomimetic analogues, Nano Let. 2, 91–5. Obst, M., Gasser, P., Mavrocordatos, D., and Dittrich, M. (2005), TEM-specimen preparation of cell/mineral interfaces by focused ion beam milling, Am. Miner. 90, 1270–1277. Obst, M., Dynes, J. J., Lawrence, J. R., Swerhone, G. D. W., Karunakaran, C., Kaznatcheev K. V., Bertwistle, D., Benzerara, K., Tyliszczak, T., and Hitchcock, A. P. (2009a), Precipitation of amorphous CaCO3 (aragonite) controlled by cyanobacteria: A multi-technique study of the influence of EPS on the nucleation process, Geochim. Cosmochim. Acta 73, 4180–4198. Obst, M., Wang, J., and Hitchcock, A. P. (2009b), Soft X-ray spectro-tomography study of cyanobacterial biomineral nucleation, Geobiology 7, 577–591. Parkinson, D. Y., McDermott, G., Etkin, L. D, Le Gros, M. A., and Larabell, C. A. (2008), Quantitative 3-D imaging of eukaryotic cells using soft X-ray tomography, J. Struct. Biol. 162, 380–386. Pecher, K., Mccubbery, D., Kneedler, E., Rothe, J., Bargar, J., Meigs, G., Cox, L., Nealson, K., and Tonner, B. (2003), Quantitative charge state analysis of manganese biominerals in aqueous suspension using Scanning Transmission X-ray Microscopy (STXM). Geochim, Cosmochim. Acta, 67, 1089–1098. Rash, T. K. and Vogel, J. P. (2004), Ecological and agricultural applications of synchrotron IR microscopy, Infrared Phys. Technol. 45, 393–341. Rightor, E. G., Hitchcock, A. P., Ade, H., Leapman, R. D., Urquhart, S. G., Smith, A. P., Mitchell, G., Fischer, D., Shin, H. J., and Warwick, T. (1997), Spectromicroscopy of poly (ethylene terephthalate): Comparison of spectra and radiation damage rates in X-ray absorption and electron energy loss, J. Phys. Chem. B 101, 1950–1961. Rightor, E. G., Urquhart, S. G., Hitchcock, A. P., Ade, H., Smith, A. P., Mitchell, G. E., Priester, R. D., Aneja, R. D., Appel, G., Wilkes, G., and Lidy, W. E. (2002), Identification and quantitation of urea precipitates in flexible polyurethanes by X-ray spectromicroscopy, Macromolecules 35, 5873–5882.

368

SYNCHROTRON-BASED X-RAY AND FTIR ABSORPTION SPECTROMICROSCOPIES

Russell, L. M., Maria, S. F., and Myneni, S. C. B. (2002), Mapping organic coatings on atmospheric particles, Geophys. Res. Lett. 29, 26/21–26/ 24. Schafer, T., Chanudet, V., Claret, F., and Filella, M. (2007), Spectromicroscopy mapping of colloidal/particulate organic matter in Lake Brienz, Switzerland, Environ. Sci. Technol. 41, 7864–7869. Schafer, T. Michel, P., Claret, F., Beetz, T., Wirick, S., and Jacobsen, C. (2009), Radiation sensitivity of natural organic matter: Clay mineral association effects in the Callovo-Oxfordian argillite, J. Electron Spectrosc. Relat. Phenom. 170, 49–56. Schumacher, M., Christl, I., Scheinost, A. C., Jacobsen, C., and Kretzschmar, R. (2005), Chemical heterogeneity of organic soil colloids investigated by scanning transmission X-ray microscopy and C-1s NEXAFS microspectroscopy, Environ. Sci. Technol. 39, 9094–9100. Smith, A. L. (1979), Applied Infrared Spectroscopy: Fundamentals, Techniques and Analytical Problem-Solving, Wiley, New York. Stem, M. (2008), Understanding why researchers should use synchrotron-enhanced FTIR instead of traditional FTIR, J. Chem. Educ. 85, 983–989. Stewart-Ornstein, J., Hernandez Cruz, D., Hitchcock, A. P., Hale, J, Overhage, J., Hilpert, K., Hancock, R, Dynes, J. J., Lawrence, J. R., Korber, D. R., and Leppard, G. G. (2006), Predicting and Verifying NEXAFS Spectral Standards for STXM Mapping of Antimicrobial Peptides, abstract, presented at Advanced Light Source Users Meeting, Berkeley, CA, Oct. 9–11, 2006. St€ ohr, J. (1992), NEXAFS Spectroscopy, Springer-Verlag, Berlin. Szeghalmi, A., Kaminskyj, S., and Gough, K. M. (2007), A synchrotron FTIR microspectroscopy investigation of fungal hyphae grown under optimal and stressed conditions, Anal. Bioanal. Chem. 387, 1729–1789. Thieme, J, Gleber, S. C., Guttmann, P., Prietzel, J., McNulty, I., and Coates, J. (2008), Microscopy and spectroscopy with X-rays for studies in the environmental sciences, Miner. Mag. 72, 211–216.

Thieme, J., McNult, I., Vogt, S., and Paterson, D. (2007), X-ray spectromicroscopy—a tool for environmental sciences, Environ. Sci. Technol. 41, 6885–6889. Thieme, J., Schneider, G., and Knochel, C. (2003), X-ray tomography of a microhabitat of bacteria and other soil colloids with sub-100 nm resolution, Micron 34, 339–344. Thomasson, J., Coin, C., Kahraman, H., and Fredericks, P. M. (2000), Attenuated total reflectance infrared microspectroscopy of coal, Fuel 79, 591–718. Toner, B., Fakra, S., Villalobos, N., Warwick, T., and Sposito, G. (2005), Spatially resolved characterization of biogenic manganese oxide production within a bacterial biofilm, Appl. Environ. Mlicrobiol. 71. 1300–1310. Vila-Comamala, J., Jefimovs, K., Pilvi, T., Ritala, M., Sarkar, S. S., Solak, H. H., Guzenko, V. A., Stampanoni, M., Marone, F, Raabe, J., Tzvetkov, G., Fink, R. H., Grolimund, D., Borca, C. N., Kaulich, B., and David, C. (2009), Advanced X-ray diffractive optics, J. Phys. Conf. Ser. 186, 1–3. Wang, J., Botton, G. A., West, M. M., and Hitchcock, A. P. (2009a), Quantitative evaluation of radiation damage to polyethylene terephthalate by soft X-rays and high energy electrons, J. Phys. Chem. B 113, 1869–1876. Wang, J., Morin, C., Li, L., Hitchcock, A. P., Zhang, X., Araki, T., Doran, A., and Scholl, A. (2009b), Radiation damage in soft Xray microscopy, J. Electron Spectrosc. Relat. Phenom. 170, 25–36. Yoon, T. H., Benzerara, K., Ahn, S., Luthy, R. G., Tyliszczak, T., and Brown G. E. (2006), Nanometer-scale chemical heterogeneities of black carbon materials and their impacts on PCB sorption properties: Soft X-ray spectromicroscopy study, Environ. Sci. Technol. 40, 5923–5929. Yu, P. (2007), Microlocalization and distribution of digestionresistant aromatic lignin and cellulosic compounds in feeds at cellular and subcellular levels: A novel approach, J. Animal Feed Sci. 16, 505–525.

15 APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS FROM COMPLEX ENVIRONMENTAL MATRICES SANJA RISTICEVIC, DAJANA VUCKOVIC,

AND JANUSZ

PAWLISZYN

15.1. Introduction 15.1.1. Importance of Sampling and Sample Preparation 15.1.2. Development of a Solid-Phase Microextraction (SPME) Device 15.2. SPME Theory and Principles 15.2.1. Thermodynamic Parameters Affecting SPME Extraction Efficiency 15.2.2. Kinetic Parameters Affecting SPME Extraction Efficiency 15.3. Optimization of SPME Methods 15.3.1. SPME Method Development Strategy 15.3.2. SPME Method Development: Factors Affecting Precision 15.4. Calibration in SPME 15.4.1. Traditional Calibration Approaches 15.4.2. Equilibrium Extraction Calibration Method 15.4.3. Kinetic Calibration Methods 15.5. Automated and High-Throughput SPME Approaches 15.5.1. SPME-GC Autosamplers 15.5.2. Metal Fiber Assemblies and Septumless Injection Systems 15.6. SPME Devices Other than Fiber-SPME 15.6.1. Thin Film Microextraction (TFME) 15.6.2. Cold-Fiber HS-SPME Devices 15.6.3. Needle Trap Devices 15.7. SPME Applications and Performance Characteristics 15.8. Future Perspectives of SPME 15.9. Conclusions

15.1. INTRODUCTION 15.1.1. Importance of Sampling and Sample Preparation In recent years, the methods of conducting quantitative and qualitative chemical analyses have advanced and more emphasis has been placed on generation of high-throughput separation and detection methods. Nevertheless, the analysis of complex samples consists of several steps, typically including sampling, sample preparation, separation, quantitation, statistical evaluation, and decision making. Each of these steps contributes to the overall length of analysis time with analysis speed depending mainly on the slowest step in the sequence (Pawliszyn 1997). However, the routine use of high-throughput methods requires faster sample preparation and injection methods, since the combination of any fast separation and detection method with long and tedious sample preparation procedure is not practical. Therefore, the development of a fast sample preparation procedure is a critical requirement to allow for full implementation of highthroughput methods. Furthermore, the choice of appropriate sampling procedure has the highest impact on selectivity, sensitivity, accuracy, and precision of an analytical method (Donato et al. 2007). The use of traditional sample preparation methods, including purge-and-trap, liquid–liquid extraction (LLE), and many others may introduce significant drawbacks such as, for

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang.
2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

369

370

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

example, the consumption of significant amounts of organic solvents, large sample amount requirements, and long sample preparation times (Kataoka et al. 2000). Nevertheless, while some sample preparation methods require large amounts of environmentally unfriendly organic solvents and thus, directly oppose U.S. Environmental Protection Agency (US EPA) requirements associated with the elimination or minimization of organic solvent consumption in residue analysis, the most important point is that most of the traditional sample preparation methods are not easily hyphenated with separation/detection systems (for the purposes of automation) (Risticevic et al. 2009). Furthermore, sample preparation time-costs and invasive nature of exhaustive sample preparation methods hinder their on-site implementation and direct investigation of chemical composition in living organisms (Pawliszyn 2007). 15.1.2. Development of a Solid-Phase Microextraction (SPME) Device Solid-phase microextraction (SPME) was developed by Pawliszyn and co-workers in 1989 to address the need for rapid laboratory and on-site sample preparation and provide an efficient way toward integration of sample preparation with separation/detection systems (Arthur and Pawliszyn 1990). Since its development the technique has demonstrated numerous advantages for the quantitative and qualitative analyses of organic compounds from various sample matrices, including (1) minimum or eliminated solvent consumption; (2) small sample amount requirements; (3) short sample preparation times; (4) easy automation leading to highthroughput analysis; (5) extraction and preconcentration of analytes from solid, liquid, and gaseous sample matrices; and (6) easy implementation for on-site and in vivo investigations (Risticevic et al. 2009; Pawliszyn 2007). Historically, the technique was developed through the early work associated with laser desorption/gas chromatograph (GC), where a rapid sample preparation method was needed to take full advantage of brief laser pulses and high-speed separation system. In order to coat the fiber tip with the sample, one end of the optical fiber was dipped into the solvent extract, after which volatile solvents were removed through evaporation. Subsequently, the fiber tip was placed into the GC injector, after which the analytes were volatilized onto the front of the column through the application of a laser pulse (Pawliszyn and Liu 1987). In a similar fashion, early work associated with development of SPME involved the use of fused-silica optical fibers coated and uncoated with polymeric phases that were dipped into aqueous samples containing target analytes and placed into a GC injector (Belardi and Pawliszyn 1989). As both polar and nonpolar species were efficiently extracted from aqueous samples and since chromatographers had good knowledge of fused-silica coating methods, the development of SPME accelerated very rapidly with the incorporation of

coated fibers into a microsyringe. This led to the development of the manual SPME device, which today in its most advanced form resembles the one presented in Figure 15.1. The most important part of this device is a fiber solid support coated with a thin layer of a polymeric stationary phase that is used to extract the analytes by concentrating them from the sample matrix (Pawliszyn 1997; Risticevic et al. 2010b). The fiber is housed inside the needle, which serves to protect the fiber/fiber coating from damage during vial/injector septa penetrations (Pawliszyn 1997; Risticevic et al. 2010b). While the combination of fiber and the needle represents the SPME fiber assembly, the SPME device is also composed of a fiber holder that is available in two different formats to allow for the performance of manual and automated SPME processes. Those early developments led to important discoveries on which theoretical principles and method development strategies of SPME were built. However, the development of new SPME devices and automation and calibration approaches is still a very active research area. Traditionally, SPME has been used routinely in combination with gas chromatography and gas chromatography–mass spectrometry (GC-MS). However, in order to analyze nonvolatile or thermally labile compounds not amenable to GC or GC-MS, significant improvements were made in direct coupling of SPME with highperformance liquid chromatography (HPLC) and liquid chromatography–mass spectrometry (LC-MS). Today, SPME is widely applicable in both targeted and nontargeted qualitative and quantitative analyses of organic compounds from various gaseous, solid, and liquid environmental, biological and food matrices. The objective of this chapter is to outline important aspects of SPME theory, method development, and calibration strategies in order to facilitate the development of SPME applications of environmental interest, including on-site and in vivo determinations. In addition to traditional method development approaches that can be used as a starting point for all sample matrices (regardless of complexity), a particular emphasis will be placed on the method development and calibration strategies that need to be undertaken when handling complex samples in which matrix effects are likely to occur. Selected applications of SPME for the analysis of organic compounds in various complex matrices are presented to illustrate the main points.

15.2. SPME THEORY AND PRINCIPLES The SPME principle relies on placement of a thin polymeric coating coated on the outside of fused-silica fiber directly to the sample matrix or to the headspace above it for a predetermined period of time (Fig. 15.2) (Pawliszyn 1997). As soon as the coated fiber is placed in contact with the sample, analytes partition by adsorption or absorption from the sample matrix to the extracting phase (Pawliszyn 1997; Mills and Walker 2000). As soon as the analyte concentration

SPME THEORY AND PRINCIPLES

371

Plunger Barrel Plunger Retaining Screw Z-slot Hub-Viewing Window Adjustable Needle Guide/Depth Gauge Tensioning Spring Septum Piercing Needle Sealing Septum

Fiber Attachment Tubing

Coated Fused Silica Fiber

Figure 15.1. Commercial fiber-SPME device for manual operations available from Supelco [Figure reprinted with permission from Lord and Pawliszyn (2000)].

reaches the distribution equilibrium between the sample matrix and the fiber coating, SPME extraction is considered to be complete. After reaching equilibrium state, the amount of analyte extracted by the SPME coating does not increase with further extraction time increments within the limits of experimental error. This means that sampling under equilibrium conditions provides the maximum sensitivity achievable with SPME (Pawliszyn 2007). However, SPME extraction can be interrupted at any instance before equilibrium is reached, provided that sufficient sensitivity is achieved for a

particular application of interest. Once the extraction process is completed, concentrated extracts are transferred onto the separation system either via thermal desorption in GC injection port or by solvent desorption in the case of HPLC (Pawliszyn 2007; Ner
ın et al. 2009). A basic understanding of kinetic and thermodynamic theories can guide SPME users toward appropriate and efficient method development. While only a brief overview of theoretical principles is given in the current contribution, the readers are strongly encouraged to consult publications where theory of SPME extraction, mass transfer, and kinetic and thermodynamic parameters and processes are addressed in more detail (Pawliszyn and Liu 1987; Pawliszyn 2007; Lord and Pawliszyn 2000; Pawliszyn 2003; Baltussen et al. 2002; Prosen and Zupancic-Kralj 1999; Pawliszyn 2001; Nongonierma et al. 2006). 15.2.1. Thermodynamic Parameters Affecting SPME Extraction Efficiency

Figure 15.2. The basic principle of SPME extraction. Fiber coating extracting phase is immersed in the sample matrix for a predetermined extraction time during which analytes partition via absorption or adsorption processes. After the completion of extraction process, the coating is introduced into the analytical instrument for a subsequent desorption of analytes, separation, and detection.

The main parameter affecting SPME extraction efficiency is fiber coating (extracting phase)–sample matrix distribution constant of the target analyte (Kfs) as it reflects the chemical composition of the extracting phase and determines the magnitude of enrichment factors possible through use of the SPME technique (Pawliszyn and Liu 1987; Pawliszyn 2003, 2007). The term Kfs is a physicochemical constant that has been discussed in detail in fundamental chromatographic literature, because it determines the retention and selectivity of a separation column (Pawliszyn 2003). Similar to the determination of octanol–water distribution constants (Kow),

372

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Kfs (or Kes when extraction phase geometries different from “fiber-configuration” are used) can be estimated by considering a variety of properties characteristic of a particular target analyte–matrix combination. For example, if appropriate mobile and stationary phases are employed so that they correspond to extracting phase and sample matrix, Kes can be estimated on the basis of chromatographic retention times obtained. In addition, Pawliszyn et al. have reported the correlation between the octanol–water distribution constants and SPME extraction phase–water distribution constants for compounds within the same chemical class and having similar structures when nonpolar polydimethylsiloxane (PDMS) fiber coating is used (Pawliszyn 1997, 2007; Risticevic et al. 2010b; Shurmer and Pawliszyn 2000). The distribution constant (Kes) between the liquid extraction medium and the sample matrix is expressed by the following equation: Kes ¼ Ce1

Ce1 Cs1

ð15:1Þ

Cs1

where and are the equilibrium concentrations of the target analyte in the fiber coating and sample matrix, respectively, (Pawliszyn 1997). Equation (15.1) is also applicable when solid is used as an extraction medium, provided that Ce is replaced by the solid extraction phase surface concentration of adsorbed analytes (Se) (Pawliszyn 2001). As will be emphasized later, the parameters that affect the magnitude of Kes are sample temperature and sample matrix conditions such as salt, pH, and organic solvent composition (Pawliszyn 2003). The number of moles of analyte extracted at equilibrium (ne) in a two-phase system (sample matrix and extraction phase) can be determined through the following equation ne ¼

Kes Ve Vs C0 Kes Ve þ Vs

ð15:2Þ

where Ve is the fiber coating volume, Vs is the sample volume, and C0 is the initial concentration of a target analyte in the sample matrix (Pawliszyn 1997; Lord and Pawliszyn 2000). However, the extraction process becomes more complicated in multiphase heterogeneous systems, where more than two phases are present such as, for example, headspace, immiscible liquids, and solids. This is a frequent phenomenon that takes place during the analysis of complex environmental samples. Two potential implications may arise when analytes are extracted from such complex samples: (1) competition among different phases for the target analyte and (2) fouling of the extraction phase (Pawliszyn 2003). While fouling of the extraction phase will be discussed in more detail later in the chapter, partitioning of target analyte between different phases in heterogeneous samples implies that the lower proportion of analyte is

available for extraction. This phenomenon depends on the analyte affinity toward the particular competing phase as well as competing phase volumes (Pawliszyn 2003). In such circumstances, the number of moles of analyte extracted at equilibrium by an extraction phase in contact with a multiphase sample matrix can be calculated using a different form of Equation (15.2): ne ¼

Kes Ve Vs C0 P Kes Ve þ i¼m i¼1 Kis Vi þ Vs

ð15:3Þ

Here, Vi is the competing phase volume and Kis ¼ Ci1 =Cs1 is the distribution constant of the analyte between the ith competing phase and the sample matrix (Pawliszyn 2007). Equation (15.3) would result in Equation (15.4) when the investigated system involves the extraction phase, gas phase (or headspace), and a homogeneous matrix such as pure water ne ¼

Kes Ve Vs C0 Kes Ve þ Khs Vh þ Vs

ð15:4Þ

where Khs is the headspace–sample matrix distribution constant and Vh is headspace volume (Pawliszyn 2007).

15.2.2. Kinetic Parameters Affecting SPME Extraction Efficiency Kinetic parameters affect the rate of extraction, and how the overall speed and times of extraction can be increased becomes clear when these parameters are understood. In order to enhance mass transfer of analytes from the sample matrix to the vicinity of the fiber, some level of agitation is usually required. With aqueous samples, independent of the agitation efficiency employed during a particular extraction, the fluid contacting fiber surface is always stationary and as the distance from the fiber surface increases, fluid movement increases as well, until it corresponds to bulk flow in the sample (Pawliszyn 2007; Lord and Pawliszyn 2000). In order to model such mass transport, a zone termed the Prandtl boundary layer was introduced and considered as a region whose thickness is dependent on both the rate of agitation and analyte diffusivity (Lord and Pawliszyn 2000). Therefore, in a single sample, the thickness of the boundary layer varies for different analytes. However, since the fluid contacting fiber surface is always stationary, in a boundary layer region, analyte flux is increasingly more dependent on analyte diffusion and less on agitation, as the extraction phase is approached. Agitation (convection) conditions are critical in reducing the thickness of the boundary layer and thus increasing the rate of mass transfer from the sample matrix to the fiber coating. This, in turn, leads to shorter equilibration times and increased overall speed of analysis.

OPTIMIZATION OF SPME METHODS

15.3. OPTIMIZATION OF SPME METHODS In the practical application of SPME, a variety of experimental factors need to be considered and addressed for a particular system under investigation. Selection of the parameters that affect SPME extraction efficiency is dependent mainly on the target analytes of interest, sample matrix, and objectives of analysis (Risticevic et al. 2010b). The parameters affecting sensitivity and reproducibility of SPME methods are displayed in Figure 15.3 and are thoroughly discussed in the following text. In addition, the current section also addresses important SPME method development strategies that must be critically considered when handling complex heterogeneous sample matrices. In such circumstances, certain SPME parameters (fiber coating, sampling mode, extraction time, ionic strength, sample dilution, organic solvent addition, and sample temperature) can be manipulated to account for matrix effects and at the same time improve extraction efficiencies. The suggestions associated with adjustment of these parameters are illustrated in Figure 15.4. 15.3.1. SPME Method Development Strategy 15.3.1.1. Selection of Fiber Coating. Selection of fiber coating is usually the first stage in SPME method optimization. As demonstrated in Equation (15.2), the sensitivity of the SPME method is proportional to the fiber coating–sample matrix distribution constant (or, in other words, the chemical composition of the extraction phase), as this parameter determines coating selectivity toward target analyte of interest versus other components present in the sample matrix (Pawliszyn 1997; Risticevic et al. 2010b; Buldini et al. 2002). Thus far, the only producer of commercial fiber assemblies has been Supelco (Bellefonte, PA, USA), which offers single-

373

polymer and mixed-polymer coatings of different polarities, thicknesses (7–100 mm range), and lengths (1 and 2 cm lengths available) (Table 15.1) (Pragst 2007). These polymers are commercially coated on solid supports consisting of fused-silica, StableFlex (flexible and less breakable fusedsilica), or more recently introduced metal/nickel–titanium alloy cores (Pragst 2007; Setkova et al. 2007c). Fiber selectivity for the particular analytes of interest is determined on the basis of the principle “like dissolves like” (Pawliszyn 2007). Therefore, single-polymer coatings such as nonpolar polydimethylsiloxane (PDMS) and polar polyacrylate (PA) provide high efficiency for the extraction of nonpolar and polar target analytes, respectively. Supelco has also introduced Carbowax–polyethylene glycol (PEG) singlepolymer fiber coating to offer more enhanced extraction capacity for highly polar analytes. In addition to singlepolymer coatings, mixed-polymer coatings, including carboxen/polydimethylsiloxane (CAR/PDMS), polydimethylsiloxane/divinylbenzene (PDMS/DVB), and divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS), are also available to provide better efficiency for volatile/ low-molecular mass compounds, more polar analytes, as well as target analytes possessing a wider range of physicochemical properties (Kataoka et al. 2000; Pawliszyn 2007). Figure 15.5 demonstrates that the type of fiber coating influences both the selectivity and the sensitivity of extraction (Vas and V
ekey 2004; Nakamura and Daishima 2005; Roberts et al. 2000). As can be seen from Figure 15.5, the chromatographic profiles corresponding to the extraction of volatile and semivolatile compounds from coffee brew obtained using PDMS/DVB, CAR/PDMS, and PDMS fiber coatings are markedly different (Roberts et al. 2000). The PDMS coating provided the lowest extraction efficiency. The CAR/PDMS coating provided the best sensitivity toward low molecular weight compounds such as 2-methylpropanol

Figure 15.3. Typical parameters that require optimization during general SPME method development.

374

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Figure 15.4. Suggestions on optimizing SPME parameters for analysis of complex and heterogeneous samples in which matrix effects are likely to occur.

and acetaldehyde. On the other hand, PDMS/DVB demonstrated the best overall sensitivity and best extraction efficiency for more polar compounds such as guaiacol, 4-ethylguaiacol, and 4-vinylguaiacol (Roberts et al. 2000). Even though Kes [Eq. (15.2)] determines method sensitivity, it must be emphasized at this point that compounds with high Kes values need more time to reach equilibrium (Pawliszyn 2007; Louch et al. 1992). This is because compounds having high Kes values require longer times for distribution equilibrium to be reached because more material must diffuse through the static layer. While the type of fiber coating and Kes can be manipulated to increase both the method sensitivity and selectivity, one efficient way to increase method sensitivity is to increase the volume of extraction phase and/or extraction phase thickness

[Eq. (15.2)]. This effect is demonstrated in Figure 15.6 for the application focused on determination of organophosphorous insecticides and their metabolites in olive oil samples using SPME (Tsoutsi et al. 2006). However, increasing the thickness of the extraction phase also results in longer equilibration times, since the extraction rate is controlled by diffusion from the sample matrix through the boundary layer to the extraction phase (Louch et al. 1992; Bruheim et al. 2003). Therefore, it is strongly recommended that the thinnest coating providing satisfactory method sensitivity should be used. Although some of the commercial coatings discussed in Table 15.1 are suitable for use with LC applications, it is important to mention that swelling of the extraction phase may occur on exposure to some organic solvents during the solvent desorption step. This may cause inadvertent extraction phase from

375

200–270

7

85

60 65

Polydimethylsiloxane (PDMS)

Polydimethylsiloxane (PDMS)

Polyacrylate (PA)

Carbowax–polyethylene glycol (CW-PEG) Polydimethylsiloxane/ divinylbenzene (PDMS/DVB)

2–11

250–310 250–310 230–270

75 85 50/30

2–11

2–11

2–11

2–11

2–9

2–11

2–11

2–11

60

220–300

220–320

200–280

2–10

pH Range Nonpolar

Nonpolar

Nonpolar

Polar

Polar Bipolar

30 min, 250 C 30 min, 250 C

60 min, 320 C

60 min, 280 C 30 min, 240 C 30 min, 250 C

Bipolar Bipolar Bipolar

60 min, 300 C 60 min, 300 C 60 min, 270 C

Bipolar

Polarity

Conditioning Procedure for GC Applications

Adsorption

Adsorption

Adsorption

Adsorption

Adsorption

Absorption

Absorption

Absorption

Absorption

Absorption

Mechanism of Extraction

Gases and volatiles, MW range 30–225 Gases and volatiles, MW range 30–225 Volatiles and semivolatiles, broad range of analyte properties (C2–C20 range), flavors and fragrances, MW range 40–275

Nonpolar analytes, volatiles, MW range 60–275 Nonpolar analytes, volatiles and semivolatiles, MW range 60–275 Nonpolar analytes, volatiles and semivolatiles, MW range 60–275 Polar semivolatiles, ideal for phenols, MW range 80–300 Polar analytes, good for alcohols Polar volatiles, amines, nitroaromatic compounds, MW range 50–300 General purpose

Analytical Application

GC

GC

GC

HPLC

GC

GC/HPLC

GC/HPLC

GC/HPLC

GC/HPLC

GC/HPLC

Analytical System

Coatings supplied by Supelco (Bellefonte, PA). Fiber assemblies consist of fused-silica, StableFlex, or metal coating supports coated with one of the polymeric phases listed in this table. Fiber assemblies are available in manual and automated formats and consist of 23- or 24-gauge needle sizes. Conditioning procedure for HPLC applications consists of immersing the coating in the mobile phase or solvent for a minimum of 30 min. Further details on conditioning, thermal cleaning, and solvent cleaning procedures are available from Supelco SPME fiber assembly product sheets. Coating for DVB/CAR/PDMS available in 1 and 2 cm lengths.

a

Polydimethylsiloxane/ divinylbenzene (PDMS/DVB) Carboxen/polydimethylsiloxane (CAR/PDMS) Carboxen/polydimethylsiloxane (CAR/PDMS) Divinylbenzene/carboxen/ polydimethylsiloxane (DVB/CAR/PDMS)

200–250

30

Polydimethylsiloxane (PDMS)

200–280

100

SPME Fiber Coating

Temperature Range ( C)

Coating Thickness (mm)

TABLE 15.1. Commercially Available SPME Fiber Coatingsa

376

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Figure 15.5. The effect of SPME fiber coating on extraction efficiency of volatile and semivolatile compounds from coffee brew (1 mL of sample placed in a 16-mL vial, SPME extraction performed in headspace above coffee brew for 10 min). The figure demonstrates the importance of choosing the appropriate fiber coating for a particular application of interest. [Figure reprinted with permission from Roberts et al. (2000).]

the core. To address this issue, a new line of coatings suitable for SPME-LC applications based on nonswelling biocompatible polymer is currently under development by Supelco and is expected to become commercially available shortly (Vuckovic et al. 2009).

15.3.1.1.1. Fiber Coating Considerations for Analysis of Complex Samples. Selection of the most appropriate fiber coating increases the selectivity of SPME determinations in complex samples. However, complex samples consist of multiple components, many of which might have sufficiently

Figure 15.6. The effect of SPME coating thickness on SPME method sensitivity obtained in the analysis of organophosphorus insecticides and their metabolites in olive oil samples (olive oil samples were spiked at a level of 0.10 mg/kg, 5 g of sample placed in 10-mL vial, extraction was performed from headspace for 60 min at 75 C) [Figure reprinted with permission from Tsoutsi et al. (2006)].

OPTIMIZATION OF SPME METHODS

high Kes values to be coextracted with target analytes of interest. In such circumstances, the selectivity can be further enhanced by addressing more carefully the separation and detection selectivity and efficiency. The mechanisms of extraction of analytes with liquid (single-polymer or single-phase) and solid (mixed-polymer or mixed-phase) SPME coatings are different (Pawliszyn 2007; Lord and Pawliszyn 2000; Pragst 2007). Mixedpolymer coatings include materials such as copolymers of PDMS with divinylbenzene (PDMS/DVB) and physical mixtures of PDMS with adsorbents such as carboxen (CAR/PDMS) (Baltussen et al. 2002). For example, CAR/ PDMS is a special case comprising a mixed-carbon (carboxen 1006 adsorbent) phase with small micropores (Mills and Walker 2000). With liquid SPME coatings, the analytes partition into the extraction phase, where the analyte molecules are solvated by the coating molecules (Pawliszyn 2007). Furthermore, since the diffusion coefficient of analyte is sufficiently high in the liquid coating, analyte molecules can penetrate the whole volume of the coating within a reasonable extraction time. Thus, the mechanism of analyte extraction with liquid coatings is based on absorption process (Lord and Pawliszyn 2000). On the other hand, solid coatings possess a well-defined crystalline structure that reduces the analyte diffusion coefficient significantly. Thus, during the extraction time employed for a particular application of interest, the extraction occurs on the surface of the coating via an adsorption process (Lord and Pawliszyn 2000). One advantage of such a mechanism is associated with shorter extraction times, since diffusion of analytes into adsorptiontype coatings does not occur. However, while the introduction of solid coatings has significantly increased the trapping capacity, an important drawback that needs to be dealt with is the loss of the true sorption mechanism (Baltussen et al. 2002). With solid coatings, limited surface area is available for adsorption. If this area becomes substantially occupied, which is likely to occur at long extraction times and with high concentration samples, analytes and matrix interferences (salts, humic acids, proteins, etc.) compete for the adsorption sites (Baltussen et al. 2002; Roberts et al. 2000). Therefore, it is quite likely that in such circumstances, displacement effects take place and the analytes with weaker affinities to the coating are extracted in significant amounts only at shorter extraction times and are displaced with analytes having stronger affinities at long extraction times (Risticevic et al. 2010b; Pawliszyn 1997). This leads to a smaller linear dynamic range, and the equilibrium amount extracted for “displaced” analyte dependent not only on its initial concentration in the sample matrix but also on the concentrations of other competing compounds. For example, in a study performed by Nakamura and colleagues focusing on determination of 22 volatile organic compounds (VOCs), methyl-tert-butyl ether, 1,4dioxane, 2-methylisoborneol, and geosmin in water samples,

377

the authors concluded that the linear dynamic range determined in a multicomponent system was narrow for CAR/ PDMS and PDMS/DVB fiber coatings (Nakamura and Daishima 2005). On the other hand, PDMS, with 100 mm coating, provided a wide range of linearity and was selected by the authors for future experiments. Similarly, Gorecki and co-workers studied the displacement effect under equilibrium conditions for a system composed of i-propanol (lower affinity for PDMS/DVB fiber coating) and methyl-isobutyl ketone (MIBK) (higher affinity for PDMS/DVB coating) (Gorecki et al. 1999). It was determined that as long as MIBK concentration remained 10 times lower than the concentration of i-propanol, the calibration curve remained linear up to 75 mg/L, and the deviation from linearity at higher concentrations was not very significant (Fig. 15.7). On the other hand, when MIBK concentration was equal to that of i-propanol, the calibration curve was linear only up to 25 mg/L (Gorecki et al. 1999). One way to deal with these drawbacks is to perform the actual extraction time profile curve in the sample matrix and determine the exact time point when the displacement effects start taking place (usually preequilibrium conditions). In addition, linearity of the method applied in the analysis of multicomponent systems needs to be assessed for every type of sample matrix in which the composition of matrix components might have been altered (Gorecki et al. 1999). 15.3.1.2. Selection of Extraction Mode. The two most commonly used extraction modes in combination with SPME are direct-immersion (DI-SPME) and headspace (HS-SPME) (Pawliszyn 1997). With DI-SPME, the fiber coating is exposed and completely immersed inside the sample matrix, whereas with HS-SPME, the fiber coating is placed in the vapor phase (headspace) above the sample matrix. As expected, DI-SPME is suitable for the extraction of analytes having low to medium volatility and high to medium polarity (Pawliszyn 2007). On the other hand, HSSPME represents an attractive alternative when target analytes of interest possess high to medium volatility and low to medium polarity. With HS-SPME, the limiting step affecting the overall mass transport to the fiber coating is the rate of mass transport from the sample to headspace (Pawliszyn 1997; Risticevic et al. 2010b). This represents a significant advantage for sampling of highly volatile analytes, which are extracted much more rapidly than are the less volatile analytes. First, more volatile analytes are present at a higher concentration in the headspace as compared to the analytes of lower volatility. Therefore, for high-sensitivity headspace extraction, headspace volume should be minimized (Pawliszyn 1997; Lord Pawliszyn 2000). Also, diffusion coefficients in the gaseous phase are four orders of magnitude higher than in the aqueous phase, and it was demonstrated that the time required to reach adsorption equilibrium with the coating is an order of magnitude shorter

378

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Figure 15.7. Calibration curves of i-propanol in the presence of methyl-isobutyl ketone (MIBK) obtained by HS-SPME sampling using PDMS/DVB fiber (&—calibration curve when MIBK concentration is 10 times lower than isopropanol concentration; *—calibration curve when MIBK concentration is equal to isopropanol concentration) [Figure reproduced by permission of The Royal Society of Chemistry from Go´recki et al. (1999)].

in the gaseous phase as compared to the liquid phase (Louch et al. 1992). These effects contribute to shorter equilibration times for volatile compounds with HS-SPME as compared to DI-SPME (Lord and Pawliszyn 2000; Mills et al. 2000). 15.3.1.2.1. Extraction Mode Considerations for Analysis of Complex Samples. As mentioned in Section 15.2.1., one of the potential complications that may arise when analytes are extracted from complex samples is associated with fouling of the extraction phase (Pawliszyn 2003). The presence of high-molecular weight and nonvolatile components in heterogeneous samples, including proteins, humic materials, and other macromolecules, complicates the DI-SPME process, due to the possibility of coextracting these interferences together with the analytes of interest (Pawliszyn 2003). In addition, high levels of nonvolatile and high-molecular weight interferences present in complex samples may adsorb and accumulate on the surface of the fiber, and may not be removed during the desorption process (Pawliszyn 2003; De Jager et al. 2008). These effects decrease SPME method reproducibility, extraction efficiency, and fiber coating lifetime (De Jager et al. 2008). Therefore, the analysis of complex heterogeneous samples with DI-SPME is usually carried out simultaneously with sample dilution (Zambonin et al. 2004; Fernandez-Alvarez et al. 2008a). Alternatively, a washing step may be introduced after extraction and before

the desorption process to remove the accumulated interfering compounds from fiber coating before potential artifact formation in the injector port and generation of unrepresentative chromatographic profiles (Verhoeven et al. 1997). The issue of fouling can also be addressed through development of coatings that are not susceptible to adhesion of macromolecules. For example, the use of SPME coatings that consist of biocompatible polymers can prevent or minimize adhesion of proteins to the surface (Musteata et al. 2007). Another way to deal with such complications is to introduce barriers between the sample matrix and the extraction phase in order to restrict transport of high-molecular weight interferences (Pawliszyn 2003). Such barriers may be composed of porous membranes (with pores smaller than the size of the interfering molecules) surrounding the extraction phase, an approach that is usually referred to as membrane-protected SPME (Pawliszyn 1997; Nongonierma et al. 2006; Musteata et al. 2007). Alternatively, a barrier composed of a gap consisting of gas or the use of HS-SPME is another very efficient approach to restrict the transport of non-volatile components to the coating (Pawliszyn 2003; Nongonierma et al. 2006). Headspace SPME not only improves the selectivity of SPME determinations in gaseous and dirty samples but also represents a very efficient approach toward the analysis of solid samples (Baltussen et al. 2002; Fromberg et al. 1996; Fernandez-Alvarez et al. 2008b). It provides good sensitivity for the extraction of highly volatile compounds; however, it is also a good alternative for semivolatiles, provided that extraction conditions are adjusted accordingly in order to promote the release of analytes into the vapour phase and accelerate mass transfer processes (Pawliszyn 2001, 2003; Llompart et al. 1999b). 15.3.1.3. Selection of Agitation Method. As mentioned in Section 15.2.2., agitation efficiency during the extraction process is very critical as it allows faster mass transport of the analytes from the sample matrix to the fiber coating (Pawliszyn 2007). Agitation reduces the thickness of the boundary layer, which in real-life analysis implies shorter equilibration times. In addition, the initial rate of SPME extraction is inversely proportional to the thickness of the boundary layer surrounding the extraction phase, which implies that more efficient agitation allows for higher extraction rate and higher mass of analyte extracted in preequilibrium conditions (Bruheim et al. 2003). This effect is illustrated in Figure 15.8 for the extraction of polycyclic aromatic hydrocarbons (PAHs) from aqueous samples (Lord and Pawliszyn 2000). Agitation efficiency is especially critical for analytes having high sample matrix–fiber coating distribution constants (Louch et al. 1992). If the agitation is perfect and stationary liquid around the extraction phase is very thin, even high-molecular weight compounds can be extracted

OPTIMIZATION OF SPME METHODS

379

Figure 15.8. Effects of different agitation efficiencies: (a) 75% stirring rate; (b) 100% stirring rate on extraction time profiles of PAHs in aqueous samples (HS-SPME mode of extraction employed). Curve A—naphthalene; curve B—acenaphthene; curve C—phenanthrene; curve D—chrysene. [Figure reprinted with permission from Lord and Pawliszyn (2000)].

with sufficient sensitivity by applying only short extraction times. Different agitation methods for aqueous samples that can be used in combination with SPME have been comprehensively described elsewhere (Pawliszyn 1997, 2007; Louch et al. 1992; Motlagh and Pawliszyn 1993) and their characteristics are summarized in Table 15.2. Agitation for on-site and in-field water sampling can be achieved by using a handheld battery-operated drill which is capable of holding SPME fiber assembly as well as other SPME extraction formats and maintaining constant agitation speed (Qin et al. 2008; Pawliszyn 2007). As far as the agitation of air samples is concerned, it has previously been observed that increasing wind speed (up

to  5 cm/s) enhances extraction efficiency in preequilibrium conditions (Pawliszyn 2007). The wind speed or air bulk movement significantly affects the mass transfer process of volatile organic compounds (VOCs) from bulk air to the extraction phase. As expected, the thickness of the boundary layer between the extraction phase and bulk air decreases on increasing the windspeed, which, in turn, leads to an increase in the mass transfer process. After decreasing of the boundary layer thickness to a certain degree, no further increase in extraction efficiency is observed when wind speed is increased and uptake of analytes becomes diffusion dependent. In practice, when using SPME for field sampling, it would be beneficial to use a fan (such as modified hairdryer fan with a mounting for the SPME device) to move air across

TABLE 15.2. Performance Characteristics of Various Agitation Methods for Aqueous Samples that can be used in Combination with SPME Agitation Method

Advantages

Disadvantages

Static Magnetic stirring

Simple, good performance for gaseous phase Simple equipment, good performance

Intrusive stirring Vortex/moving vial

Good performance Good performance, use of stirring bar not required

Fiber movement flow through

Good performance for small sample volumes, use of stirring bar not required Good performance at rapid flows

Sonication

Short extraction times

Limited to volatile analytes, limited to HS-SPME Use of stirring bar required, might not be constant, possible carryover problems, manual recovery of stirring bars required, heating of the stirring plate might occur above the optimum set sample temperature Difficulty in sealing the sample Stress placed on fiber and needle, suitable for automated sample preparation Stress placed on fiber and needle, limited to small vial sizes Constant flows are required, possible crosscontamination Sample heating occurs

380

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

the extraction phase during sampling in order to eliminate possible imprecision in extraction due to variations in wind speed (Pawliszyn 2007; Ouyang and Pawliszyn 2006b). 15.3.1.4. Selection of Extraction Time. Extraction time is one of the most crucial steps in SPME method development. As mentioned previously, maximum sensitivity is achievable with SPME methods when equilibrium conditions are applied. Therefore, the choice of optimum extraction time depends on the particular application under consideration and is always a compromise between sensitivity, speed, and precision required (Pawliszyn 2007). In order to optimize extraction time, usually an extraction time profile curve is constructed showing the dependence of mass of analyte extracted on time (Prosen and Zupancic-Kralj, 1999). Fig. 15.9 demonstrates a typical extraction time profile that can be obtained by varying extraction times for samples having fixed concentration of target analytes. At this point it must be emphasized that equilibrium time is not affected by concentration, so any concentration of target analyte can be employed to construct the extraction time profile curve. However, the time required to reach equilibrium is a function of the extraction phase-sample matrix distribution constant of the analyte, diffusion coefficient of the analyte, thickness of the extraction phase, and thickness of the boundary layer (Pawliszyn 1997). Returning to Figure 15.9, three objectives need to be defined before extraction time can be optimized. If the main objective of analysis is directed toward achieving maximum sensitivity, equilibrium extraction conditions should be employed (Risticevic et al. 2010b). However, from practical perspective, even if extraction conditions are manipulated (increasing extraction temperature to increase diffusion coefficient of target analyte, utilizing more efficient agitation

methods, using thinner coatings) so that equilibrium state is reached faster, sometimes, equilibration times are on the order of hours, days, or even weeks, so that equilibrium extraction is not feasible (Pawliszyn 2007). This is typically the case in SPME-LC applications, but SPME-GC applications are not excluded from such effects either. For example, Fromberg et al. published a study focused on the determination of a range of chlorinated and nitrated aromatic compounds in soil and determined that equilibration times were very long (on the order of 10 h) for high-molecular weight compounds (Fromberg et al. 1996). While these analytes partition very strongly into the SPME coating, their diffusion process from the sample matrix, and through the gaseous phase to the extraction phase is very slow. The equilibration times for hydrophobic organic chemicals (HOCs) of environmental significance including brominated flame retardants (BFRs) such as polybrominated diphenyl ethers (PBDEs) and polybrominated biphenyls (PBBs) analyzed in water samples (water samples spiked at pg mL1 level, PDMS fiber coating employed) ranged from 2 to 4 h at 100 C sample temperature, whereas at 25 C they ranged from 8 days to several weeks (Polo et al. 2006). In those circumstances when the objective of analysis is good method precision, equilibrium conditions should be used as well unless perfectly repeatable extraction times are guaranteed in nonequilibrium conditions (e.g., with the use of an autosampler). Figure 15.9 demonstrates that a small error in extraction time causes much higher errors in the amount of analyte extracted in preequilibrium conditions, especially when very short extraction times in the linear region of the extraction time profile curve are used. Therefore, the steeper the slope of the extraction time profile curve, the larger the relative errors that occur (Pawliszyn 2007). Having this in mind, time control is critical when analyzing

Figure 15.9. Typical extraction time profile curve indicating the dependence of amount of analyte extracted on extraction time. The curve indicates the importance of precise extraction timing for preequilibrium SPME methods, especially when very short extraction times in the linear portion of the curve are used.

OPTIMIZATION OF SPME METHODS

large numbers of samples in all SPME arrangements, especially in preequilibrium ones. More recently, analysts have been faced with challenges of generating high-quality data to many real-life problems in a high-throughput manner. In such circumstances, and because of the potential alteration of the experimental system over a period of time, the utilization of short extraction times is mandatory. Provided that extraction conditions are carefully adjusted to ensure optimum sensitivity, the use of short exposure preequilibrium times is an attractive alternative toward achieving reliable results and increasing the throughput of analytical determinations. 15.3.1.4.1. Extraction Time Considerations for Analysis of Complex Samples. As was already emphasized in Section 15.3.1.1.1, a limited number of sites available for adsorption on the surface of solid coatings leads to competition of analytes for the sorption sites and displacement effects in highly concentrated samples and during long extraction times. However, the use of solid coatings also decreases the length of extraction time and provides increased extraction efficiency and retention capability, which are obviously advantages appreciated in the analysis of complex samples (Pawliszyn 2007; Baltussen et al. 2002; Buldini et al. 2002). However, analyte competition and displacement still need to be accounted for. This is usually accomplished by using short extraction times to reduce the possibility of overloading the capacity of fiber coating (Roberts et al. 2000). Also, it is strongly advisable to perform the actual extraction time profile curves for target analytes of interest in all types of analyzed sample matrices, including those having variable composition. 15.3.1.5. Optimization of Sample Volume. Equation (15.2) illustrates that the number of moles of analyte extracted is proportional to the sample volume, indicating the possibility of enhancing the method sensitivity by increasing the sample amount. However, the amount of analyte extracted increases with the sample volume until a point at which Vs becomes much larger than the product of Kes and Ve which leads to a significant simplification of Equation (15.2) to Equation (15.5): ne ¼ Kes Ve C0

ð15:5Þ

This equation indicates that under certain conditions, the amount of analyte extracted by SPME coating is independent of sample volume, a finding that opens the doors to many alternatives of modern sample preparation. First, Equation (15.5) is valid when the Kes  Ve product becomes insignificant as compared to the sample volume, which is typically the case in many on-site and in vivo determinations. If such a condition is fulfilled, SPME sample preparation step does not require a defined sample volume to be collected prior to analysis, and analysts are able to collect real-time data

381

directly on the site where the investigated system is located (Lord and Pawliszyn 2000; Risticevic et al. 2010b). In addition, together with elimination of the sampling step, the errors associated with analyte losses through decomposition, volatilization, and adsorption during sampling and transportation are eliminated and the whole analytical procedure is accelerated. Furthermore, Equation 15.5 is usually applicable when the amount of analyte extracted is insignificant compared to the original amount present in the sample matrix that is obtainable when analyzing large sample volumes and compounds of small Kes, the latter effect resulting in negligible depletion of analyte concentration from the original sample (Pawliszyn 2007). At this point it must be emphasized that in practical laboratory arrangements, the use of an autosampler dictates the size of sample vials and thus limits the range of sample volumes that can be tested for optimum sensitivity. In such circumstances, the amount of analyte extracted may become dependent on sample volume and Equations 15.2, 15.3 are applicable for calculating the amount extracted. 15.3.1.6. Selection of Sample pH. Most coatings employed in SPME extract the undissociated/neutral species of target analyte only (Pawliszyn 2007). Therefore, one of the parameters that can be manipulated in order to increase method sensitivity is pH adjustment, provided that it ensures a full conversion of analytes into neutral form. While low pH values improve the extraction efficiency for acidic compounds, high pH values are more suitable for the extraction of basic compounds. For compounds possessing both the acidic and basic functionalities, optimum pH must be determined experimentally (Pawliszyn 2007; Risticevic et al. 2010b). Figure 15.10 demonstrates the effect of pH on the extraction efficiency of ochratoxin A (OTA) from wine samples (Aresta et al. 2006). The amount of OTA extracted was enhanced three fold on varying pH from 4.5 to 3.0. In this particular case, a pH value of 3.0 was selected for future experiments. At this point, it must be emphasized that pH adjustments are preferred with HS-SPME extraction mode, especially if the optimum pH value for a particular target analyte is close to the extremes of recommended pH ranges for a particular fiber coating used (Lord and Pawliszyn 2000). For the analysis of complex heterogeneous semisolid samples, where pH adjustment may not be feasible, a more recent study demonstrated that acidification of the extraction phase prior to extraction can yield equivalent results and improve extraction efficiency of ionizable compounds (Zhang et al. 2009a). 15.3.1.7. Optimization of Ionic Strength. The addition of soluble salts increases the ionic strength of the sample solution and may, in turn, decrease the solubility of organic compounds in aqueous solution (Pawliszyn 2007). Consequently, decreased aqueous solubilities of target analytes

382

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Figure 15.10. Dependence of the amount of Ochratoxin A (OTA) extracted by PDMS/DVB fiber coating from wine samples (the wine sample was spiked with OTA at 2 ng/mL level and diluted 20-fold with a 10 mM phosphate buffer) on sample pH. The optimum pH value that was used for future experiments is 3.0. [Figure reprinted with permission from Aresta et al. (2006)].

lead to increased fiber coating-sample matrix distribution constant and therefore, increased method sensitivity. The salts that are most commonly employed for the adjustment of ionic strength are NaCl, Na2SO4, K2CO3, and (NH4)2SO4 (Mills and Walker 2000). The effect of salt addition on extraction efficiency, as well as the optimum salt amount, is strongly dependent on the physicochemical properties of the target analyte of interest, and both of these effects should be determined experimentally (Pawliszyn 2007). For example, Deger and co-

workers determined that increasing salt concentration from 20%–30%, while increasing the extraction efficiency for endosulfan lactone, decreased the extraction efficiencies for a-endosulfan, b-endosulfan, endosulfan sulfate, and endosulfan ether from environmental water samples (Deger et al. 2003). The authors attributed such observations for a-endosulfan, b-endosulfan, endosulfan sulfate, and endosulfan ether to the fact that increased salt amounts lead to decreased solubilities in aqueous media and consequently increased surface adsorption (Deger et al. 2003). Another example is provided by Tsimeli and colleagues, who studied the effect of salt addition on the extraction efficiency of 1,2-dibromoethane, 1,4-dichlorobenzene, and naphthalene in honey samples (Tsimeli et al. 2008). They determined that for more hydrophobic compounds, namely, naphthalene (log Kow ¼ 3.3) and 1,4-dichlorobenzene (log Kow ¼ 3.44), the extraction efficiency was not significantly affected by salt addition (Fig. 15.11). However, for a more polar target analyte, 1,2-dibromoethane (log Kow ¼ 1.96), a positive salt addition effect was observed, and adjusting the concentration of sodium chloride to the saturation level improved the method detection limit drastically (Tsimeli et al. 2008). It was reported that the addition of salts in addition to improving the extraction efficiency for target analytes of interest may also improve the extraction efficiency for interfering compounds, which could cause complications when solid coatings are used (Pawliszyn 2007). 15.3.1.8. The Effects of Sample Dilution and Relevant Considerations for Analysis of Complex Samples. One parameter influencing SPME extraction efficiency and

Figure 15.11. The effect of sodium chloride concentration on the extraction efficiency of naphthalene (&), 1,4-dichlorobenzene (.), and 1,2-dibromoethane (~) from honey samples (2 mL of spiked honey sample, extraction performed at 40 C and 900 rotations per minute (rpm) for 30 min with PDMS100mm fiber coating) [Figure reprinted with permission from Tsimeli et al. (2008)].

OPTIMIZATION OF SPME METHODS

requiring optimization is sample dilution. The optimization of this parameter is especially crucial when complex samples are being handled. Equation (15.3) shows that components of heterogeneous samples affect the sensitivity of SPME methods; thus, SPME parameters need to be adjusted accordingly to account for potential decreased extraction efficiencies that may arise in complex samples. For example, Simplicio and co-workers have demonstrated that suspended matter as well as dissolved matter were responsible for reduced extraction efficiencies of organophosphorous pesticides in fruits and fruit juices by forming micelles, adsorbing target analytes, and slowing down their diffusion process toward the extraction phase (Simplicio and Boas 1999). Liu and others demonstrated the difference in the amount of bisphenol A extracted by SPME when extraction time profiles were performed in water, diluted milk, and pure milk samples (Liu et al. 2008). Similarly, Moeder and colleagues detected loss in extraction efficiency of pharmaceuticals in the presence of the organic matter when they analyzed water samples free of organic matter and water samples spiked with humic acids to simulate wastewater conditions (Moeder et al. 2000). In the presence of matrix effects in complex samples, it can be assumed that the analyte is distributed in the sample between a free form and a bound form with matrix components (Zambonin et al. 2004). Therefore, simple sample dilution with water implies a lower analyte concentration in the sample but at the same time, it may increase extraction recoveries as a consequence of the equilibrium displacement toward the free form of the analyte and consequently increased analyte transfer from the sample to the extraction phase (Zambonin et al. 2004; Fernandez-Alvarez et al. 2008a; Cajka et al. 2009; Liu et al. 2008). This was illustrated by Zambonin and co-workers in an application involving the determination of organophosphorous pesticides in wine and fruit juices, as they clearly demonstrated that relative recovery (extraction efficiency in fruit juices vs. pure water) increased with increased water dilution factor (Zambonin et al. 2004). The addition of water to liquid samples in which strong matrix activity effects are present, besides reducing the matrix effect, also makes the use of DISPME possible (Zambonin et al. 2004; Fernandez-Alvarez et al. 2008a; Risticevic et al. 2010b). For example, Fernandez-Alvarez and colleagues published a study on the determination of pesticides in bovine milk in which they intended to use HS-SPME mode, due to the complexity and lipophilicity of milk to avoid damage and contamination in either the fiber coating or the GC-MS instrument (Fernandez-Alvarez et al. 2008a). They did not observe increased extraction efficiency when sample dilution was applied with HS-SPME; however, sample dilution and application of DI-SPME mode markedly increased the extraction recoveries for some target analytes (cyfluthrin and deltamethrin) that were undetectable with HS-SPME.

383

Relative recoveries of analytes are also strongly dependent on the amount of water added to semisolid samples such as, for example, fruits and vegetables (Zambonin et al. 2004; Lambropoulou and Albanis 2003; V
azquez et al. 2008) and solid samples, such as soil and sediment (Llompart et al. 1999a,b; Fromberg et al. 1996; Fernandez-Alvarez et al. 2008b; Cam et al. 2004; Cam and Gagni 2001; Salgado-Petinal et al. 2006). The determination of volatile and semivolatile constituents in soil samples necessitates the use of HS-SPME, but the degree of partitioning of volatiles and semivolatiles between the soil and headspace is generally low (Llompart et al. 1999b; Fernandez-Alvarez et al. 2008b). Therefore, in order to enhance the release of analytes from the sample matrix and increase their concentration in the headspace, water can be added to soil (Fromberg et al. 1996; Fernandez-Alvarez et al. 2008b; SalgadoPetinal et al. 2006). The addition of small amounts of water to solid samples facilitates the desorption of analytes, since they are released from their active sorption sites (usually polar functional groups) in the solid matrix by polar water molecules (Cam and Gagni 2001; Llompart et al. 1999b). For example, it has been reported that the addition of 2 mL of water to 0.5 g of sediment resulted in an important increase in HS-SPME (PDMS fiber coating, 100 C sample temperature, 30 min extraction time) extraction efficiency (three- to four fold higher responses) for BFRs, namely, PBBs and PBDEs (Salgado-Petinal et al. 2006). However, since sample dilution implies both lower analyte concentration and reduction of matrix effects, optimum dilution factor needs to be determined for every particular type of sample matrix (Fernandez-Alvarez et al. 2008a; Zambonin et al. 2004; Lambropoulou and Albanis 2002; Llompart et al. 1999b). For example, Llompart and co-workers demonstrated that the optimum water amount to be added to 1 g of soil sample is 1 mL (Fig. 15.12) for the determination of volatile and semivolatile pollutants in soils (Llompart et al. 1999b). 15.3.1.9. Organic Solvent Content. According to the thermodynamic theory of SPME, the amount of organic solvents in a sample should be maintained at a minimum value, so that fiber coating-sample matrix distribution constant (Kes) does not decrease (Pawliszyn 2007). Typically, in order to obtain optimum extraction efficiencies and avoid change in Kes, natural organic solvent composition and/or sample composition after spiking/fortification should not exceed 1% of the sample volume. Fern
andez-Gonz
alez and others studied the effect of organic modifier content for the application involving the determination of polycyclic aromatic hydrocarbons (PAHs) in water samples (tapwater, seawater, well and superficial waters) using PDMS/DVB fiber coating (Fern
andezGonz
alez et al. 2007). The results of this study are demonstrated in Figure 15.13, indicating that organic

384

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Figure 15.12. The effect of water dilution factor on the extraction recovery of volatile and semivolatile pollutants in soils (1 g of soil, variable volumes of water added, 22-mL vial, 30 min HS-SPME extraction with 100-mm PDMS coating, desorption 3 min at 260 C) [Figure reprinted with permission from Llompart et al. (1999b)].

modifier percentages of 5% caused a dramatic decrease in extraction efficiency (Fern
andez-Gonz
alez et al. 2007). On the other hand, several authors indicated that the addition of small amounts of organic modifiers compensates for the adsorption of hydrocarbon compounds by the glass walls when water is the only solvent involved (Cam et al. 2004). The reason for such an effect is likely due to the increased water solubility in the presence of added organic solvents, such that the effect of analyte adsorption onto vessel walls is eliminated or minimized (Fern
andez-Gonz
alez et al. 2007;

1050000

naphthalene

900000

fluonene phenanthrene

Area

750000

pyrene

600000

benzo(a)pyrene benzo(ghi)pyrene

450000 300000 150000 0

0

5

10 15 20 % methanol

25

30

Figure 15.13. The effect of methanol content on the extraction efficiency of polycyclic aromatic hydrocarbons (PAHs) from water samples (extraction performed with PDMS/DVB fiber for 60 min at 60 C) [Figure reprinted with permission from Fern
andez-Gonz
alez et al. (2007)].

Cam et al. 2004). The solubility of hydrophobic compounds in aqueous media can also be increased with the addition of surfactants at concentrations higher than their critical micellar concentrations (CMCs) (Cam et al. 2004; Pino et al. 2003). These examples provide excellent demonstrations of multiple effects that may arise when organic solvents are added to the sample: (1) decreased Kes lowers the amount of analyte extracted and (2) increased analyte solubility in the aqueous medium eliminates or minimizes analyte adsorption onto extraction vessel walls. 15.3.1.9.1. Organic Solvent Content Considerations for Analysis of Complex Samples. Similar to the effects of sample dilution with water, it has been reported that the addition of small amounts of organic solvents to the sample may liberate the analytes from the active sites in the heterogeneous sample matrices and therefore enhance their release from the sample matrix into the fiber coating (Cam and Gagni 2001; Llompart and co-workers. 1999b; Fernandez-Alvarez et al. 2008b). For example, Cam et al. reported that the addition of some organic solvents having high dielectric constants (e.g., acetone, methanol, hydrogen dioxide) to the soil before extraction led to an increase in HS-SPME sensitivity (Cam and Gagni 2001). Llompart and coworkers compared the extraction recoveries of volatile and semivolatile pollutants in soil when different organic modifiers, (acetone, acetonitrile, methanol, water) were added (100 mL of organic modifier added) to 1 g of soil (Llompart et al. 1999b). The extraction efficiencies increased in all cases in the presence of organic modifiers,

OPTIMIZATION OF SPME METHODS

and water proved to be most effective for desorption of analytes from active soil sites. This group also reported that polar solvent molecules when added to the sample can substitute analyte molecules on the active sites also having mostly polar functional groups, thus leading to increased HSSPME extraction efficiencies for the analytes of interest (Llompart et al. 1999b). The degree of substitution depends on physicochemical properties of both the target analytes and solvent, as well as functional groups present on the active sites in sample matrix. For example, water was the most effective organic modifier for the extraction of organochlorine pesticides and their meta-bolites in soil samples (Doong and Liao 2001), whereas optimum results for determination of organophosphorus insecticide residues in strawberries and cherries were obtained after the addition of methanol (Lambropoulou and Albanis 2003). 15.3.1.10. Analyte Derivatization. Similar to the traditional sample preparation methods, analyte derivatization with SPME is employed to allow the chemical transformation of analyte into a form that is more suitable for extraction and/or analysis (Vas and V
ekey 2004). Derivatization is usually utilized to increase extraction recoveries of polar and/or thermally unstable target analytes by converting them into more easily extractable, thermally stable, and more volatile components with better chromatographic behaviour (Mills and Walker 2000; Ner
ın et al. 2009). Therefore, derivatization may improve the extraction efficiency and selectivity

385

(higher distribution constants of derivatives toward the fiber coating), chromatographic behavior, and detection sensitivity (Pawliszyn 2007; Mills and Walker 2000). With the wider range of available derivatization reagents [pentafluorobenzylhydroxylamine (PFBHA), 2,4-dinitrofluorobenzene (DNFB), N-ethylmaleimide (NEM), acetic anhydride, Nmethyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA)], derivatization is more commonly employed in SPME-GC applications (Quintana and Rodriguez 2006). The most common derivatization processes were summarized by Ner
ın and colleagues: (1) esterification of acids, which reduces their polarity; (2) transformation of lowmolecular-weight carbonyl compounds, such as aldehydes and ketones, into their more stable oximes; (3) transformation of phenols into their acetates, which improves chromatographic behavior; and (4) transformation of amines, amphetamines, and cocaine to derivatives of lower polarity and higher volatility (Ner
ın et al. 2009). Several derivatization approaches are possible with SPME (Fig. 15.14): (1) pre-extraction derivatization, (2) post-extraction derivatization, and (3) simultaneous extraction and derivatization. The choice of the most appropriate derivatization strategy is strongly dependent on the physicochemical properties of target analytes, derivatization reagents, and the type of sample matrix under consideration (Pawliszyn 2007). The pre-extraction derivatization approach, also referred to as direct derivatization in the sample matrix, is analogous

Figure 15.14. Different approaches to derivatization with SPME.

386

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

to well-established approaches used in solvent extraction (Lord et al. 2000). Briefly, this process involves adding derivatization reagent directly to the sample matrix containing target analytes of interest, allowing some time for derivatization reaction to take place, and exposing the SPME fiber coating to either a solution or headspace to extract the derivatized analytes (Mills and Walker 2000; Lord and Pawliszyn 2000; Vas and V
ekey 2004). Pre-extraction derivatization is implemented in those circumstances when the underivatized target analytes of interest (usually highly polar compounds) have low affinity (low Kes) toward commercially available polar SPME phases (Risticevic et al. 2010b). In such circumstances, the two possible approaches that can be implemented consist of: (1) use of polar coatings for extraction of underivatized analytes so that Kes is enhanced and (2) preextraction derivatization. For example, pre-extraction derivatization has been applied in the analysis of phenols (by converting them to acetates with acetic anhydride) in water samples using nonpolar PDMS fiber coating (Buchholz and Pawliszyn 1994). Pre-extraction derivatization improves extraction efficiency, chromatographic behavior, and detection sensitivity (Pawliszyn 2007). Because of the availability of polar extraction phases (PA and CW), extraction efficiency for polar underivatized target analytes of interest is frequently sufficient to reach the desired sensitivity (Lord and Pawliszyn 2000). However, problems associated with chromatographic separation and detection sensitivity are occasionally encountered (Lord and Pawliszyn 2000). In such circumstances, post-extraction derivatization is a good alternative, as this approach improves the chromatographic and detection attributes. With postextraction derivatization, SPME fiber coating is first introduced into the sample matrix to extract underivatized target analyte. This is followed by exposure of the coating to the derivatization reagent vial to allow for in-fiber derivatization (Pawliszyn 2007). Subsequently, fiber coating containing derivatized target analytes is transferred to the analytical instrument for desorption, separation, and detection. For example, high molecular mass carboxylic acids were extracted with SPME fiber coating, and the coating containing extracted analytes was exposed to diazomethane to produce ester derivatives, in order to obtain much better chromatographic behavior and detection sensitivity/selectivity (Lord and Pawliszyn 2000). An alternative method for performing post-extraction derivatization involves derivatization in the GC injector port (Lord and Pawliszyn 2000). In this less popular and less commonly used approach, SPME fiber coating is exposed to the sample matrix containing underivatized target analytes, which after extraction are derivatized through the use of a derivatizing reagent injected into the GC injector port just prior to introduction of SPME fiber for desorption (Nagasawa et al. 1996). In this case, derivatization approach takes place at high injector temperatures (Lord and Pawliszyn 2000).

Pan and co-workers investigated this approach for the analysis of C10–C22 fatty acids with ionpair reagents tetramethylammonium hydroxide (TMA-OH) and tetramethylammonium hydrogen sulfate (TMA-HSO4) in aqueous samples using polar PA fiber (Pan and Pawliszyn 1997; Stashenko and Martinez 2004). A very attractive approach to performing derivatization with SPME is the use of the simultaneous extraction derivatization option. This approach first involves the exposure of SPME coating to derivatization reagent vial to load the derivatization reagent onto the fiber coating prior to extraction (Pawliszyn 2007). Subsequently, the coating preloaded with the derivatization reagent is exposed to the vial containing the underivatized target analytes so that simultaneous extraction and derivatization processes can occur. This is followed by coating introduction into the GC injector port so that derivatized analytes can be desorbed and transferred into the analytical instrument for subsequent separation and detection. It has been reported that for most of the applications, involving, for example, determination of aliphatic amines, pesticides, and carboxylic acids, simultaneous extraction and derivatization approach provided methods with improved analytical sensitivity (Ner
ın et al. 2009). This derivatization strategy improves the extraction efficiency, chromatographic behavior, and detection properties (Risticevic et al. 2010b). Scheyer and colleagues recently reported a study on the determination of trace levels of pesticides in rainwater by SPME derivatization with pentafluorobenzyl bromide (PFBBr) (Scheyer et al. 2007). The study involved comparison of different SPME derivatization approaches: preextraction derivatization directly in the aqueous sample, simultaneous extraction derivatization (10 min headspace loading of PFBBr followed by SPME extraction), and post-extraction derivatization of extracted analytes in GC injector port. Figure 15.15 demonstrates the results of their study: the best overall results were obtained with simultaneous extraction and derivatization approach. More detailed comparison and suitability of different SPME derivatization approaches has been presented elsewhere (Pan et al. 1997; Herr
aez-Hern
andez et al. 2006; Stashenko and Martinez 2004; Ouyang and Pawliszyn 2006b). 15.3.1.11. Optimization of Sample Temperature. Sample temperature is another parameter that requires careful consideration during development of SPME methods. From the kinetic perspective, increased sample temperatures lead to higher diffusion coefficients and larger headspace capacity (when HS-SPME mode is applied) (Pawliszyn 2007). Consequently, faster extraction rates, faster mass transfer from the sample matrix into the fiber coating and shorter equilibration times are obtained. On the other hand, thermodynamic theory predicts a decrease in the fiber coating-sample matrix distribution constant when the extraction temperature is increased. In such circumstances, the amount of analyte

OPTIMIZATION OF SPME METHODS

387

Figure 15.15. Comparison of different SPME derivatization approaches for the determination of pesticides in rainwater samples with PDMS/DVB fiber coating and PFBBr derivatization reagent [Figure reprinted with permission from Scheyer et al. (2007)]. (See insert for color representation of this figure.)

extracted at equilibrium and hence the equilibrium method sensitivity decrease (Pawliszyn 1997; Risticevic et al. 2010b). Such an effect is illustrated in Figure 15.16 for the study involving the HS-SPME determination of methamphetamine (Lord and Pawliszyn 2000). While the lowest temperature employed (22 C) produces very long equilibration times, it also results in highest amount of analyte extracted at equilibrium. On the other hand, the highest temperature employed (73 C) results in shortest equilibration times and lowest amount of analyte extracted

Figure 15.16. The effect of sample temperature on the extraction time profile and mass extracted for methamphetamine (2-mL sample placed in a 4-mL vial, 0.5 M KOH, saturated NaCl, 15 min extraction with HS-SPME and PDMS 100-mm fiber coating, 15 min desorption) (symbols: ¤ 22 C; ~ 40 C; & 60 C; . 73 C) [Figure reprinted with permission from Lord and Pawliszyn (2000)].

at equilibrium. However, an important point to consider is that even though the amount of analyte extracted at equilibrium conditions will be lower at high extraction temperatures and higher at low extraction temperatures, in preequilibrium conditions, the amount of analyte extracted will still be higher at higher temperatures than what it would be at lower temperatures (Pawliszyn 2007). On the basis of these examples, the choice of optimum sample temperature will be significanlty affected by the objectives of analysis. Lower extraction temperatures should be used in those circumstances in which the main objective of analysis is maximum sensitivity under equilibrium conditions. On the other hand, if high sample throughput is the main objective of analysis and for improved sensitivity in preequilibrium conditions, higher sample temperatures should be employed (Pawliszyn 2007). The choice of the optimum extraction temperature is also affected by physicochemical properties of target analytes of interest. This is clearly demonstrated in Figure 15.17 for the study performed by Vichi and coworkers and involving the determination of volatile and semivolatile aromatic hydrocarbons in olive oil samples (Vichi et al. 2005). Hence, the recovery of more volatile analytes decreased after 40 C and 60 C (depending on the volatility) because for these compounds, the effect of decreasing distribution constant on increasing sample temperature predominates. On the other hand, the uptake of higher molecular weight and less volatile compounds was favored with higher extraction temperatures, since for these analytes, the effect of increasing diffusion coefficient and mass transfer processes was more important. At the end, the working temperature chosen by the authors

388

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Figure 15.17. The effect of sample temperature on the extraction efficiency for compounds having different physicochemical properties in olive oil sample matrix (2 g of 10 mg/kg olive oil sample placed into 10 mL vial, DVB/CAR/PDMS fiber coating used for HS-SPME for 30 min) (symbols: & C1benzene; ~ C2-benzenes; ¤ C3-benzenes; & C4-benzene; * naphthalenes;acenaphthene and acenaphthylene; þ phenanthrene and anthracene and fluorene; x fluoranthene and pyrene) [Figure reprinted with permission from Vichi et al. (2005)].

was 100 C, which is a typical example of the necessity to make compromise decisions during method development for applications when analytes of interest have a wide range of physicochemical properties (Vichi et al. 2005). 15.3.1.11.1. Sample Temperature Considerations for Analysis of Complex Samples. In heterogeneous samples, target analytes are frequently bound to the other interfering molecules present, and as such, the extraction rate is frequently controlled by the release of solid-bound analytes from the sample matrix (Pawliszyn 2001; Pawliszyn 2003). One of the parameters that can be manipulated to increase extraction rate is extraction temperature, since the dissociation of chemisorbed analytes can be successfully accomplished by the application of high temperatures (Pawliszyn 2001, 2003). For example, Fromberg and co-workers used an elevated extraction temperature of 80 C to enhance both the diffusion velocity and analyte liberation from active sorption sites in the soil when determining chloro- and nitroanilines and -benzenes by HS-SPME (Fromberg et al. 1996). Using an elevated temperature of 100 C also had a positive effect on the extraction recovery of pyrethroids, organochlorines, and other plant protection agents in agricultural soil analysis by HS-SPME (Fernandez-Alvarez et al. 2008b). 15.3.1.12. Selection of Separation and Detection Systems. As a microextraction technique, SPME represents an important contribution to the improvement of sample preparation performance and addresses the issues of miniaturization,

automation, on-site analysis, and time efficiency (Pawliszyn and Pedersen-Bjergaard 2006). In particular, because of its solvent-free nature and the fact that it eliminates the introduction of large volumes of extracts that typically exceed the required amount for chromatography and electrophoresis, SPME can be introduced to analytical instruments of various types (Pawliszyn 2007, 1997; Risticevic et al. 2010b). Solidphase microextraction is easily integrated to major separation systems, including GC, LC, and capillary electrophoresis (CE) (O’Reilly et al. 2005; Zhou et al. 2006; Fang et al. 2006; Pen˜alver et al. 2002; Sarrio´n et al. 2002; Pawliszyn 2007; Kataoka et al. 2000; Lord 2007). Thus, the choice of appropriate separation system will depend on the chromatographic behavior and physicochemical properties of the analytes of interest. The use of SPME can also be integrated with both traditional [flame ionization detector (FID), nitrogen–phosphorus detector (NPD), ultraviolet (UV), detector etc.] and mass selective detectors (Pen˜alver et al. 2002; Sarrio´n et al. 2002; Vas et al. 2004). 15.3.1.13. SPME Interfaces to Analytical Instrumentation 15.3.1.13.1. SPME-GC Interface. The gas chromatograph has been the most frequently used instrument in combination with SPME. Since the extracting phase is nonvolatile and only extracted analytes are introduced into the instrument, there is no need for complex injectors designed to deal with large amounts of solvent vapour (Pawliszyn 2007). Standard GC injectors, such as split/splitless models, are most conveniently applied as SPME interfaces, provided that

OPTIMIZATION OF SPME METHODS

desorption conditions are optimized as described in this Section. Alternatively, one way to achieve sharper injection zones and faster separation times is to utilize rapid injection autosampler devices, which will be discussed in Section 15.5.1. Alternatively, a dedicated injector [such as a programmed temperature vaporizer (PTV)], which should be cold during fiber coating introduction, but which heats up very rapidly after exposure of the coating to the carrier gas stream can be used (Pawliszyn 2007). 15.3.1.13.2. SPME-LC Interface. Four main approaches of coupling SPME to LC have been described in the literature to date. These include the manual desorption interface, in-tube SPME, manual offline desorption, and high-throughput automated offline desorption using the Concept 96 autosampler (PAS Technology, Magdala, Germany). For more extensive discussion, the reader is referred to the review by Lord (2007). Briefly, the manual desorption interface consists of a standard six-port injection valve and desorption chamber to enable static and/or dynamic desorption of analytes directly into the LC system, by direct insertion of the fiber into the mobile-phase path or other appropriate solvent (Chen and Pawliszyn 1995). A manual interface for SPME-LC can be easily built in the lab, although a commercial version from Supelco is also available. The main advantage of this desorption approach is high sensitivity because all of the analyte extracted is introduced into the LC instrument, similar to what is achieved with thermal desorption in GC injection port. The disadvantages include possible damage of the coating, leaks and associated loss of sample, and low throughput. Intube SPME was subsequently developed to address these limitations and provides an automated way of coupling SPME to any commonly available commercial HPLC autosampler (Eisert and Pawliszyn 1997). However, because of the difficulty of interfacing the commercial fiber format with the LC instrument, in-tube SPME relies on the extraction phase, which is coated inside of a capillary, rather than relying on fiber geometry. This coated capillary can then be inserted instead of an injection loop, flush loop or transfer line depending on the exact autosampler design (Lord 2007). In comparison to manual interface desorption,

389

in-tube SPME provides a high degree of automation, comparable sensitivity, and improved method performance in terms of precision and accuracy. However, this approach is limited to particulate-free samples and it is not applicable to in vivo and on-site analysis because of the limitations of its geometry. Manual offline desorption is the simplest, most cost-effective, and most flexible way to couple SPME to LC analysis. In this approach, desorption is achieved by immersing the entire length of the fiber in a vial containing appropriate desorption solvent volume. This means that no additional equipment is required and a wide selection of desorption solvents and agitation methods can be employed to achieve the desorption in the minimum amount of time. The main disadvantages of this approach are poorer sensitivity because 100% of analyte is not injected, possible adsorptive losses, and lack of automation. To date, the highest sample throughput with full automation can be achieved using the Concept 96 robotic station (Vuckovic et al. 2008). Concept 96 relies on fully automated parallel offline desorption of 96 fibers in a 96well plate format. The selection of the most appropriate SPME-LC interfacing strategy will depend primarily on the sample throughput and degree of automation required by a particular application as well as the availability of the necessary equipment in the lab. Table 15.3 compares the available interfacing options to aid the user in the selection of the most appropriate strategy. 15.3.1.14. Optimization of Desorption Efficiency. The optimization of conditions for SPME desorption is very critical, as this step ensures maximum transfer efficiency of extracted analytes into the analytical system. This issue will be discussed separately for SPME-GC and SPME-LC applications as it requires the optimization of different parameters due to the differences in the desorption processes in gaseous versus aqueous phases. 15.3.1.14.1. Desorption Efficiency in SPME-GC Applications. The factors that influence desorption efficiency in SPME-GC applications are carrier gas flow rate, desorption temperature, and desorption time (Risticevic et al. 2010b).

TABLE 15.3. Comparison of Performance Characteristics of Various SPME-LC Interfacing Approaches

Approach

Full Automation

Sample Throughput

Manual interface In-tube SPME Manual offline desorption Concept 96

No Yes No Yes

Low Medium Medium High

Sensitivity

Compatible with In Vivo and on-Site SPME

Additional Equipment Required

High High Medium Medium

Yes No Yes Yes

Yes No No Yes

390

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

During the thermal desorption of analytes from the SPME fiber coating in the GC injector port, high carrier gas linear flow rates around the fiber coating are needed. This condition accelerates transfer of analytes onto the column and allows for maximum desorption efficiency (Pawliszyn 2007). High carrier gas linear flow rates around the fiber coating can be achieved by simply (1) using narrow bore (reduced internal diameter) liners that are commercially available and (2) deactivating the split flows during desorption (Risticevic et al. 2010b). Since SPME injection does not introduce organic solvents into the system, large-volume liners typically used to accommodate the expansion of the evaporated solvent and split-valve opening employed to ensure the removal of remaining solvent vapors are not required. Besides carrier gas flow rate, desorption temperature is an additional parameter allowing for good desorption efficiency because high desorption temperatures produce a dramatic decrease in the coating/gas distribution constant and an increase in the diffusion coefficient of desorbed analytes (Pawliszyn 1997). Desorption temperature optimization for a particular SPME fiber coating is usually followed by optimization of desorption time and carryover tests. When an optimum desorption temperature has been fixed, it is strongly recommended to perform the desorption time profile at that particular temperature. Simultaneously with the performance of desorption time profile, each desorption time point needs to be verified for the absence of memory (carryover) effects. This is usually performed by running the traditional SPME extraction–desorption cycle for each desorption time point and sealing the needle of the SPME fiber assembly

immediately after completion of the desorption process with a septum/Teflon needle guide. The blank fiber coating is then reinjected as soon as the instrument becomes ready for the next injection (Pawliszyn 2007). Adjustment of parameters required for optimization of SPME-GC desorption efficiency is graphically illustrated in Figure 15.18. 15.3.1.14.2. Desorption Efficiency in SPME-LC Applications. The main factors that affect the desorption efficiency in SPME-LC applications are the composition of desorption solvent, desorption time, and desorption solvent volume. Because mass transfer processes in liquid phase are much slower, than in gaseous phase, the desorption process for SPME-LC applications is much slower, which translates into desorption times much longer than those encountered in SPME-GC methods, which rely on thermal desorption. Also, it is more difficult to achieve complete desorption of the analyte from SPME coating, so, during method development, it is crucial to examine analyte carryover in the fiber coating by performing a second desorption using a fresh portion of solvent immediately following the initial desorption. Carryover is caused by incomplete desorption of analyte from the coating, so the desorption conditions should be chosen to minimize carryover to <1%–2%. However, although desorption efficiency of 99% is suitable for quantitative analysis, if coatings are reusable and any carryover is detected, this carryover must be removed using appropriate cleaning procedures prior to subsequent analysis. Among the different interfacing methods discussed in Section 15.3.1.13.2, in-tube SPME is the least susceptible to the carryover effect, because very thin coatings are used

Figure 15.18. Parameters that affect desorption efficiency and require optimization in SPME-GC applications.

CALIBRATION IN SPME

and the capillary is washed with mobile phase continuously throughout the analytical run. Desorption solvent usually consists of an aqueous– organic solvent mixture (e.g., methanol–water or acetonitrile–water), and the eluent strength of the desorption solvent should not exceed the eluent strength of mobile phase at the analyte retention time for direct desorption methods. For example, using a manual interface, fenitrothion and its metabolites were effectively desorbed using 5-min dynamic desorption with mobile phase from PDMS/DVB fiber (Sanchez-Ortega et al. 2005). However, efficient desorption of phenylurea and propanil herbicides from PDMS/ DVB fiber coating required stronger eluent composition and longer times, so the combination of 4-min static desorption in 50% acetonitrile and 5-min dynamic desorption in mobile phase was employed (Mughari et al. 2007). Regarding the selection of desorption solvent volume, the minimum amount of solvent sufficient to immerse the entire length of the coating should be employed. Obviously, optimum sensitivity will be achieved when the smallest amount of desorption solvent is used, so the use of small volume HPLC inserts (100–300 mL) or microwell plates is preferred for offline desorption, with typical desorption volumes ranging from 20 to 300 mL. For example, a 50 mL desorption solvent volume was sufficient for effectively desorbing benzimidazole fungicides from CAR/PDMS fiber coating within 10 min. (Lopez-Monzon et al. 2007).

391

The rate of desorption can be enhanced by employing agitation during the offline desorption step in order to decrease the desorption times.

15.3.2. SPME Method Development: Factors Affecting Precision One of the main requirements that ensures good precision in SPME is associated with keeping the operating conditions such as time, temperature, and agitation consistent for all samples and standards to be analyzed in one batch. A detailed description of all factors affecting precision of SPME methods is given in Table 15.4.

15.4. CALIBRATION IN SPME In contrast to some traditional sample preparation methods, SPME is a nonexhaustive sample preparation method in which only a small proportion of target analyte is removed from the sample matrix. Therefore, the selection and optimization of calibration methods for quantitative analysis must be given thorough consideration. The purpose of the following section is to present a general overview of both the traditional and more recently developed novel and highthroughput calibration approaches. More detailed discussion

TABLE 15.4. Factors Affecting Precision in SPME Methods Factor Affecting Precision in SPME

Troubleshooting Suggestions

Variable content of organic solvent in spiked samples

Keep organic solvent content at same level for all samples when optimizing the method and constructing a calibration curve Use same type of vial for all samples in a batch Keep constant sample agitation during extraction of one sample or multiple samples in a batch Isolate vial from stirring plate by placing a septum between a vial and stirrer surface Keep fiber coating sampling depth in vial constant for all samples; avoid splashing of coating in HS-SPME during agitation; immerse coating entirely in sample for DI-SPME Condition fibers according to manufacturer’s specifications (see Table 15.1); observe fiber coating under microscope regularly; perform daily quality control extractions Ensure repeatable extraction timing; use autosampler if available Use silanized vials; test performance of different septa and choose the one where analyte losses are minimized Minimize time between extraction and analysis; cool fiber with dry ice and seal needle with a Teflon cap or septum; use portable devices for on-site sampling Add salt and/or adjust pH to compensate for small sample-to-sample variations in ionic strength and pH Replace injector septa frequently; use prepunctured septa or septumless sealing systems

Variable vial shape and consequently headspace volumes Agitation speed not constant during extraction of one sample or multiple samples in a batch Heating of stirring plate when using magnetic stirrers Inappropriate and irreproducible sampling depth inside a vial

Fiber coating does not exhibit reproducible performance; coating damaged by high operating temperatures and presence of matrix components Irreproducible timing in preequilibrium extraction Adsorption of target analytes onto the walls; analyte losses through punctured septa of vials Analyte losses due to prolonged storage/transportation times after sampling Intrinsic sample variations in ionic strength and/or pH Septum coring inside liner Source: Modified from Risticevic et al. (2010b).

392

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

of SPME calibration methods can be accessed elsewhere (Ouyang and Pawliszyn 2006b, 2008; Pawliszyn 2007). 15.4.1. Traditional Calibration Approaches Traditional calibration approaches such as external calibration curve, standard addition, and internal standard addition methods are most commonly used in combination with SPME (Ner
ın et al. 2009). These methods can be employed for both the preequilibrium and equilibrium SPME determinations (Ouyang and Pawliszyn 2008). 15.4.1.1. External Calibration Curve. The external calibration curve is one of the most widely implemented calibration approaches. This approach involves the preparation of several calibration standards in the sample matrix. These calibration solutions are then processed using the same sample preparation, separation, and detection conditions as those applied in the analysis of unknown samples. Subsequently, the relationship between the peak area responses and the spiked analyte concentrations in the calibration standards is obtained, and the resulting linear regression relationship is used to determine the analyte concentration in unknown samples. However, external calibration curve approach might not be always suitable for SPME because the sample matrix can have a very significant impact on the fundamentals of analyte extraction, as discussed in Section 15.3.1.8. Therefore, calibration samples must have the same matrix composition as those of unknown samples, and the use of matrix-matched analyte-free standards is mandatory (Pawliszyn 2007). If this requirement is not met, partitioning in calibration samples and unknown samples will not be the same and calibrations will not be valid (Pawliszyn 2007). For example, Regueiro and colleagues performed quantification of parabens, triclosan, and related chlorophenols in river, waste, and swimming pool water samples with external calibration using ultrapure water samples after they verified the absence of matrix effects (Regueiro et al. 2009a). Similarly, external calibration with ultrapure water standards was performed for the quantification of chlorinated toluenes in river and sewage water samples, since no matrix effects were observed (Regueiro et al. 2009b). On the other hand, it was impossible to use external calibration in ultrapure water for the quantification of pesticides in bovine milk, the reason of which, Fernandez-Alvarez and colleagues used pesticide-free milk standards for external calibration (Fernandez-Alvarez et al. 2008a). Schurek and co-workers also used matrix-matched external standard calibration for quantification of pesticides in tea samples using HSSPME analysis with PDMS-100 mm coating (Schurek et al. 2008). However, external calibration is not usable in those circumstances in which analyte-free matrix-matched

standards are not available for samples in which matrix effects are present, as well as in those cases in which there is significant variability in unknown sample composition (Pawliszyn 2007). 15.4.1.2. Standard Addition. The standard addition approach involves the addition of known quantities of the target analytes to a sample matrix that already contains an unknown concentration of target analyte. The samples are subsequently submitted to the entire SPME procedure, and the plot of peak area responses against target analyte concentrations is produced. The original concentration in the unspiked sample is subsequently determined by extrapolating the response of the plot to zero. As expected, since this calibration approach involves spiking the actual sample matrix with known concentrations of target analyte, its main advantage is compensation for matrix effects. Therefore, the use of standard addition should be considered for the analysis of heterogeneous complex samples as well as samples of unknown composition and variable composition. For example, Fernandez-Alvarez et al. verified that the organic matter content in soil greatly influences HS-SPME extraction efficiency for pesticides and their degradation products (Fig. 15.19) (Fernandez-Alvarez et al. 2008b). Therefore, rather than performing the external calibration curve approach in several types of soil samples to possibly match the content of organic matter, the authors decided to use the standard addition approach. Standard addition protocols were also applied toward the quantification of PAHs in soil samples with HS-SPME and brominated flame retardants in various solid samples (sandy–marine sediment, river sediment, soil, and sludge) (Llompart et al. 1999b; SalgadoPetinal et al. 2006). One important consideration that needs to be addressed when using the standard addition approach for quantitative determinations in solid samples is that the mass transfer mechanism may be different for spiked and native analytes, thus requiring the use of equilibrium conditions (Ouyang and Pawliszyn 2008). Other than that, the standard addition approach is feasible for quantification of a small number of samples since it requires extensive sample preparation. 15.4.1.3. Internal Standard. The internal standard addition approach involves the addition of known quantities of the internal standard to both the calibration solutions and unknown samples. The compound pre-selected as internal standard must be different from target analytes and chromatographically resolved from the target analytes, but it also should have physicochemical properties similar to those of the target analyte. A calibration plot is generated using ratios of the analyte to the internal standard peak areas against ratios of the analyte to the internal standard concentrations for calibration solutions having fixed concentration of internal standard and variable concentrations of target analytes.

CALIBRATION IN SPME

393

40000 35000

Area (counts)

30000 25000 20000 15000 10000 5000 0

γlindane

heptachlor

endosulfan I

endrin

4,4′-DDD

2000

Area (counts)

1600 1200 800 400 0

permethrin

cyfluthrin A

cypermethrin B

C

D

E

flucythrinate F

deltamethrin

G

Figure 15.19. Influence of matrix effects on the extraction efficiency of pesticides from soil (0.5 g of soil and 0.5 mL water placed in 10-mL vial, HS-SPME extraction with PA coating) (bar A—organic matter content 7.4%; bar B—organic matter content 13.7%; bars C-G—organic matter content 8.7%– 11.2%) [Figure reprinted with permission from Fernandez-Alvarez et al. (2008b)].

This calibration approach demonstrates many advantages, namely, compensation for matrix activity affects, losses of analytes during sample preparation, and factors contributing to irreproducibility in sample preparation and injection (Pawliszyn 2007). For example, Tsoutsi and colleagues concluded that acidity and total amount of sterols in olive oil samples were the main factors influencing HS-SPME extraction efficiency of organophosphorous insecticides (Tsoutsi et al. 2006). Therefore, since each olive oil sample had to be considered as a unique matrix, the external calibration approach was not feasible and they proceeded with internal standard (ethyl bromophos) addition calibration strategy. As opposed to the external calibration curve, which in this particular case would require a new calibration curve each time a new olive oil matrix was evaluated, the internal standard addition approach increased throughput and speed of quantitative analysis (Tsoutsi et al. 2006). However, sometimes suitable internal standards and especially deuterated analogues of target analytes are not readily available and in addition, might be quite expensive.

15.4.2. Equilibrium Extraction Calibration Method From Equations (15.2, 15.4) and (15.5) it can be seen that the amount of analyte extracted at equilibrium is linearly proportional to the initial concentration of the analyte in the sample matrix. This relationship represents the analytical basis for quantification with SPME. The equilibrium extraction method is based on exposure of a small amount of extraction phase into the sample matrix to extract target analytes until the analyte concentration reaches a distribution equilibrium between the sample matrix and the extraction phase (Pawliszyn 1997). Equations (15.2, 15.4) and (15.5) also illustrate that the initial concentration of target analyte in the sample matrix can be determined by knowing the amount of analyte extracted at equilibrium (calibration of SPME responses by liquid injection) and distribution constant of the analyte between the extraction phase and the sample matrix (Ouyang and Pawliszyn 2008; Ouyang et al. 2005b). Besides direct performance of equilibrium partitioning experiments, extraction phase/sample matrix distribution

394

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

constants can also be estimated from physicochemical data and chromatographic parameters. Distribution constants for PAHs, polychlorinated biphenyls (PCBs), (PBDEs), (PBBs), pesticides, and phthalates were reported in literature (Ouyang and Pawliszyn 2008; Valor et al. 2001).

Q ¼ exp ðatÞ q0

15.4.3. Kinetic Calibration Methods The internal standardization method described in Section 15.4.1.3., while compensating for matrix effects in complex samples, it demonstrates one potential disadvantage: manual spiking of internal standards into the sample matrix. The manual delivery process for internal standards is an additional step that adds to sample preparation and as such decreases the throughput of SPME methods and also becomes incompatible with on-site and in vivo determinations. Therefore, significant efforts have been directed toward development of high-throughput automated calibration methods as well as calibration strategies that are compatible with on-site sampling, so that the full potential of SPME can be exploited. Full implementation of SPME for high-throughput applications involves the use of shortexposure preequilibrium conditions for which it was theoretically proved that the amount of analyte extracted is related to both the initial concentration of analyte in the sample matrix and time, and as such quantification can be performed using time accumulation of analytes in the coating, provided that agitation conditions are kept constant (Ai 1997b). In 1997, Ai proposed a theoretical model based on a diffusion-controlled mass transfer process to fully describe kinetics involved in the entire SPME process (both preequilibrium kinetic and equilibrium regimes) (Ai 1997a,b). Subsequently, Chen, Zhou, and co-workers demonstrated the symmetry of absorption and desorption processes of analytes from the sample matrix to the extraction phase and of preloaded internal standards from the extraction phase to the sample matrix, respectively, for both liquid and solid SPME extraction phases (Chen and Pawliszyn 2004; Chen et al. 2004; Zhou et al. 2007). These theoretical demonstrations led to the development of kinetic calibration method, also referred to as in-fiber standardization (Ouyang and Pawliszyn 2008). Extraction kinetics of target analyte from the sample matrix into SPME fiber coating is described as n ¼ 1exp ðatÞ ne

cients, and the surface area of the extraction phase (Ouyang and Pawliszyn 2008; Chen et al. 2004). On the other hand, desorption kinetics of a preloaded internal standard from the SPME fiber coating to the sample matrix is described as

ð15:6Þ

where n is the amount of target analyte extracted at preequilibrium time t, ne is the amount of target analyte extracted at equilibrium, and a is the extraction rate constant that is dependent on the extraction phase, distribution constant, headspace and sample volumes, mass transfer coeffi-

ð15:7Þ

where Q is the amount of preloaded internal standard remaining in the coating after sampling with preequilibrium time t, q0 is the amount of internal standard that was preloaded in the SPME coating, and a is the desorption rate constant (Ouyang and Pawliszyn 2008). If the preloaded internal standard and target analyte possess similar physicochemical properties and thus, the same mass transfer SPME processes are applicable to both of them, and if the same sampling conditions are employed for both extraction and desorption processes, then the value of rate constant a is same for both extraction and desorption kinetics (Chen and Pawliszyn 2004). This condition leads to Equation (15.8) which indicates that the sum of extraction n=ne and desorption Q=q0 should be equal to 1 at any extraction/desorption time (Fig. 15.20). n Q þ ¼1 ne q0

ð15:8Þ

This equation represents the basis for quantification with SPME in preequilibrium conditions. Therefore, by preloading a certain amount of internal standard (q0) onto the SPME fiber coating and exposing the preloaded coating to the sample matrix containing the target analyte for preequilibrium time t, one can determine the amount of analyte extracted by the coating (n) and the amount of internal standard remaining in the coating (Q) for that particular time, provided that calibration of SPME by liquid injection is performed (Ouyang et al. 2005b). Equation (15.8) is subsequently used for the calculation of ne, after which the initial concentration of analyte in the sample matrix can be calculated with the use of Equation (15.2) (direct sampling), Equation (15.4) (headspace sampling), or Equation (15.5) (direct sampling with amount of analyte extracted independent of sample volume) (Zhao et al. 2007). A detailed description of steps involved in the in-fiber standardization procedure is presented in Figure 15.21. In addition to the optimization of the loading procedure for internal standards used in kinetic calibration, calibration of SPME by the injection of liquid standard solution must be critically considered here. With regard to this issue, one important assumption is made when attempting to calibrate SPME by the injection of target analytes/internal standards dissolved in organic solvent—sample transfer efficiencies for both SPME and liquid injection are the same (Pawliszyn 2007; Ouyang et al. 2005b; Ouyang and Pawliszyn 2008). While the

CALIBRATION IN SPME

Figure 15.20. Symmetric relationship between extraction and desorption of analytes from the sample matrix into the extraction phase and preloaded internal standards from the extraction phase into the sample matrix, respectively (100-mm PDMS fiber coating preloaded with deuterated toluene and exposed to a flowing aqueous toluene standard solution for different times at 25 C). (key: x— absorption of toluene; o—desorption of deuterated (d8) toluene; ~—sum of absorption and desorption; n—amount of analyte extracted at time t; n0—amount of analyte extracted at equilibrium; Q—amount of internal standard remaining in fiber coating at time t; q0—amount of internal standard preloaded in fiber coating. [reproduced by permission of The Royal Society of Chemistry from Chen et al. (2004)].

strategies used to increase SPME-GC desorption and sample transfer efficiencies were discussed in Section 15.3.1.14.1, it must be noted here that for liquid injection, the rate of sample transfer into the column is affected by many factors, including the dimensions of the liner, the presence of wool, and injector temperature, (Ouyang et al. 2005b). Ouyang and co-workers published an excellent and thorough study in which, among other suggestions, the main conclusions consisted of the following: (1) high injector temperature liquid injection may not be suitable for SPME calibration since the expansion associated with the solvent vaporization may cause analyte losses, and this is why programmed temperature vaporizing (PTV) injections are recommended; (2) complete transfer of analytes into the column minimizes the deviation for SPME quantitative analysis caused by the liquid injection calibration, and this is achieved by using a direct injection (DI) liner (SPI, Uniliner, etc.). However, the authors emphasized that in those circumstances in which DI liners are not available for a particular injector, narrow-bore SPME liners can be used as an alternative (provided that PTV liquid injection inside a wool-equipped liner is performed), resulting in high sample transfer efficiencies and the feasibility of calibrating SPME injection by liquid injection (Ouyang et al. 2005b). In-fiber standardization has been used successfully to quantify benzene, toluene, ethylbenzene, and o-xylene (BTEX) in milk and wine

395

samples; pesticides in river water; and pesticides in white wine (Zhao et al. 2007; Chen et al. 2004; Wang et al. 2005b; Zhou et al. 2007). The use of in-fiber standardization SPME calibration method increases the throughput of quantitative SPME methods since it allows quantitative determinations in preequilibrium (short-exposure) conditions as well as internal-standard delivery with the extraction phase. Besides its applicability in high-throughput and automated laboratory experimentation, this technique is especially important for the calibration of on-site and in vivo determinations, where control of the agitation condition of the matrix is sometimes difficult and direct spiking of standards into the matrix is typically not possible (Chen and Pawliszyn 2004). The change in environmental variables during sampling affects the extraction of analyte and desorption of the preloaded standard in the same way simultaneously, therefore, matrix effects as well as environmental factors (turbulence, temperature, etc.) can be accounted for (Ouyang and Pawliszyn 2008). For example, kinetic calibration has been employed for the on-site quantification of PAHs in environmental waters with thin-film microextraction samplers as well as determination of time-weighted average (TWA) concentrations of PAHs in lake waters (Bragg et al. 2006; Ouyang et al. 2007). As can be expected, one important requirement that should be given critical consideration in development of the in-fiber standardization SPME calibration approach for a particular application of interest is thorough optimization of loading procedure for the internal standards in terms of reproducibility, speed, and the degree of automation (Ouyang and Pawliszyn 2008; Zhao et al. 2007). Zhao, Wang, and co-workers have evaluated a variety of loading approaches for a wide range of GC-amenable analytes: (1) BTEX components, (2) PAHs, and (3) decachlorobiphenyl, which were used as model compounds for representing highly volatile analytes, semivolatiles, and analytes of low volatility, respectively (Zhao et al. 2007; Wang et al. 2005b). They determined that the best approach for loading highly volatile internal standards is to load from the headspace above the pump oil fortified with standards. This loading procedure provided reproducible results and the possibility of easily adjusting loaded amounts by varying the concentration in pump oil and loading times. In addition, since each loading cycle removes an insignificant amount of standard as compared to the amount originally present in the pump oil, it is possible to reuse one loading vial for more than 100 loading cycles and thus, completely automate the whole kinetic calibration procedure (Zhao et al. 2007; Wang et al. 2005b). For semivolatiles, direct loading from solution containing appropriate organic solvent spiked with standards proved to be a most universal approach. Similarly, loaded amounts can be adjusted by varying the concentration and exposure time and the automation of kinetic calibration procedure is easily achievable. Loading from solution containing appropriate organic

396

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Figure 15.21. Detailed procedure for in-fiber standardization SPME calibration approach. For relevant equations and definitions, refer to Sections 15.2.1, 15.3.1.5, and 15.4.3.

solvent spiked with standards is also applicable to compounds of low volatility. Low-volatility standards can also be reproducibly loaded by transferring directly 1 mL of organic solvent mixture fortified with standards to the extraction phase via liquid syringe (Zhao et al. 2007). Over the years, different formats of kinetic calibration were proposed. Zhou and co-workers proposed another variant of kinetic calibration using dominant preequilibrium desorption in which target analytes are used as internal standards (Zhou et al. 2008b). Because this approach utilizes dominant desorption of target analyte that is used as internal standard to calibrate extraction, it is very useful in those circumstances in which suitable internal standards that are different from target analyte (e.g., deuterated analogues) are not available or are too expensive. The method requires several modifications as compared to original in-fiber standardization, such as a larger number of fiber assemblies (for separate extraction and desorption processes), sufficient spatial resolution in sampling positions (so that the dominant desorption process does not influence extraction process), and adjustment of loaded amounts (the amount of analyte loaded must be at least 4 times higher than the amount of analyte extracted) (Zhou et al. 2008b). Kinetic calibration

using dominant preequilibrium desorption was applied toward quantitative determination of PAHs in water samples, pesticides in a jade plant (in vivo), and pharmaceuticals in fish (in vivo) (Zhou et al. 2008a,b). More recently, Ouyang and colleagues proposed a more universal form of kinetic calibration, termed one-calibrant kinetic calibration, which uses the desorption of a single standard to calibrate all extracted analytes (Ouyang et al. 2009). The technique was used for rapid on-site sampling of PAHs in water using rotated SPME fiber assemblies (Ouyang et al. 2009). Another variant of kinetic calibration referred to as standard-free kinetic calibration, was proposed by Ouyang and co-workers enabling the calibration of all target analytes with only two samplings and therefore, making it possible to obtain the amounts of analytes extracted at two different sampling times (Ouyang and Pawliszyn 2008; Ouyang et al. 2008). Ouyang and coworkers were able to shorten extraction times for PAHs (model compounds naphthalene, acenaphthalene, fluorene, anthracene, and pyrene at a concentration range of 4.1–96.3 ng/mL in the flowthrough system) in water samples from 2–24 h to 2–5 min and for BTEX in air from 5–10 min to 5–10 s with this approach (Ouyang et al. 2008).

AUTOMATED AND HIGH-THROUGHPUT SPME APPROACHES

15.5. AUTOMATED AND HIGH-THROUGHPUT SPME APPROACHES 15.5.1. SPME-GC Autosamplers Soon after development of the fiber-SPME device, it was realized that fiber arrangement of SPME is suitable for automation because of its similarity to traditional GC syringe for liquid injection (O’Reilly et al. 2005). Several adjustments were needed to allow for the complete automation of the whole SPME sample preparation procedure: (1) protection of fiber and fiber coating inside the needle during penetration of vial and injector septa; (2) adjustment of sampling depth during extraction and injection depth during desorption; (3) control of plunger movements to allow for exposure and retraction of fiber coating; (4) control of incubation, extraction, desorption, and fiber conditioning times; (5) control of extraction and fiber conditioning temperatures; and (6) efficient agitation during the extraction process. The first commercially available autosampler that was capable of performing all of these operations was the CombiPAL autosampler introduced by CTC Analytics (Zwingen, Switzerland) in 1999 (O’Reilly et al. 2005; Pawliszyn 2007). A photo of this device is presented in Figure 15.22. After introduction of the CombiPAL autosampler, several variants of SPME-compatible autosamplers for GC applications became available. For example, Gerstel GmbH (Mulheim an der Ruhr, Germany) introduced the MultiPurpose Sampler (MPS2) suitable for SPME-GC automation. Automated SPME operations with MPS2 are controlled via Maestro software, specifically designed to offer a user-friendly and simplified approach to SPME method development and optimization (Risticevic

Figure 15.22. Graphical illustration of commercially available CombiPAL autosampler (CTC Analytics, Zwingen, Switzerland) for the performance of automated SPME processes: (a) sample trays; (b) sample preparation station/agitator tray; (c) fiber conditioning station/needle heater; (d) sample preparation/injection arm.

397

et al. 2010a; Pawliszyn 2007). PAS Technology (Magdala, Germany) also commercialized the Concept autosampler for SPME-GC automation. All of these autosamplers provide an orbital agitation mechanism during incubation and extraction processes. On the other hand, Thermo Fisher Scientific (Milan, Italy) launched their variant of the SPME-compatible autosampler, which utilizes rocking agitation mechanism during incubation and extraction. These commercially available SPME-GC autosamplers do slightly differ in available temperature ranges and type of agitation mechanism (Risticevic et al. 2010a). However, all of them also provide common sequence of automated events: (1) samples are loaded onto sample trays, (2) sample preparation/injection arm transports the sample from the sample tray into sample preparation station/agitator, (3) agitation is activated under temperature-controlled conditions for a prespecified incubation period, (4) sample preparation/injection arm moves to agitator so that fiber coating can be exposed to sample matrix under temperature- and agitation-controlled conditions for a predefined extraction period, (5) sample preparation/injection arm moves to GC injector port so that fiber coating can be exposed for a predefined desorption period, and (6) sample preparation/injection arm moves to temperature-controlled fiber conditioning station/needle heater so that coating can be cleaned from potential memory effects before the beginning of next extraction cycle. In addition, the use of these autosamplers increases sample throughput, especially because all of the commercially available autosamplers are capable of starting sample preparation of the next sample in sequence, while the previously prepared sample is still undergoing GC analysis (Risticevic et al. 2010a). The use of autosamplers also ensures constant timing events, which is especially crucial to achieve good method precision in preequilibrium applications of SPME. In addition, the use of stirring bars is eliminated, which reduces potential sources of sample contamination and increases the throughput of analysis since stirring bars do not need to be added manually into the sample (Pawliszyn 2007; O’Reilly et al. 2005). CombiPAL, MPS2 and Concept autosamplers also allow adjustment of agitation speed during extraction processes. The higher the agitation speed, the more efficient the agitation process, and consequently, the higher the amount of analyte extracted in preequilibrium conditions, and the shorter the equilibration times (Risticevic et al. 2010a). However, even though agitation ranges of 250–750 rotations per minute (rpm) are typical, commonly employed agitation speeds are often lower than 750 rpm since agitation process causes stress on the needle of SPME fiber assembly. Introduction of the 23-gauge needle size was able to partially overcome this limitation. Over the years, significant improvements in the design of SPME-GC autosamplers and the software controlling their operations have been performed. For example, with the CombiPAL autosampler, users are allowed to use standard

398

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

SPME procedures provided in Cycle Composer and modify the parameters according to their applications of interest. In addition, Macro Editor software can be utilized to develop more sophisticated custom programs. The MPS2 autosampler and its Maestro software provide a special “derivatization” option. Through the use of Cycle Composer and “derivatization” option features for CombiPAL and MPS2, respectively, complete automation of internal standard and derivatization reagent loading procedures is possible. With these options, in-fiber internal standardization, post-extraction derivatization (on fiber coating), and simultaneous extraction and derivatization procedures can be completely automated (Pawliszyn 2007; Risticevic et al. 2010a; O’Reilly et al. 2005). In addition, more recent developments in SPME-GC automation have led to the possibility of automating SPME formats other than fiberSPME (described in Section 15.6). For example, the thermal desorption procedure for thin-film microextraction has been successfully automated with MPS2 autosampler by placing thin films (containing extracted analytes) into glass liners, which are subsequently introduced into a TDS-2 thermal desorption system for desorption process (see also Section 15.6.1) (Qin et al. 2008; Pawliszyn 2007). Miniaturization of a cold-fiber HS-SPME device was followed by its automation with the CombiPAL autosampler, and current research trends are also directed toward automation of coldfiber HS-SPME processes with the MPS2 autosampler (Chen and Pawliszyn 2006; Risticevic et al. 2009; Ghiasvand et al. 2006, 2007; Carasek and Pawliszyn 2006; Carasek et al. 2007). 15.5.2. Metal Fiber Assemblies and Septumless Injection Systems As higher sample throughput requirements become more important in quantitative and qualitative chemical analyses, the need for more robust SPME fiber assemblies arises. One potential limitation of the SPME technique is short lifetime of fused-silica SPME fiber assemblies (Setkova et al. 2007c). Under typical conditions, the lifetime of fused-silica fibers allows the performance of 50–100 extraction and desorption cycles; however, fiber lifetime is dependent on many factors, including, sample matrix complexity and extraction mode applied. On the basis of these observations, fused-silica fiber assemblies are perfectly suitable for batches containing up to 100 samples, provided that internal standard is added to account for potential aging of the fiber coating. However, current research trends require the analysis of a large number of samples. This is particularly important in food analysis and metabolomics studies, where large datasets and chromatographic profiles need to be compared in order to establish correlation between samples and subsequently verify the quality of products and/or to study biochemical processes that a particular system undergoes. In such applications, the

potential breakage of fused-silica fiber assembly during automated analysis would require its replacement by a new one, followed by systematic proofs that a new fiber assembly possesses performance characteristics similar to those of the old one and as such, that the precision of analytical determinations will not be adversely affected. Breakage of the fiber assembly may occur anytime during automated analysis, thus requiring analyst supervision and decreasing the throughput of analysis. In order to address the need for more durable fiber assemblies, Supelco has commercialized a new generation of superelastic metal fiber assemblies (Setkova et al. 2007c; Pawliszyn 2007; O’Reilly et al. 2005). The fiber needle, plunger and fiber core are made of a special type of inert, flexible nickel–titanium alloy (NiTinol) that possesses excellent shape memory and tensile strength (Setkova et al. 2007c; O’Reilly et al. 2005). As compared to silica-based assemblies, metal fiber assemblies can withstand higher agitation speeds and longer agitation periods without risk of bending or breaking the needle during automated analysis. These fiber assemblies are of beveled needle design, which also helps to pierce vial septa more easily. Overall, these fiber assemblies were designed to improve the robustness and sample throughput of automated SPME methods. Durability, intrafiber reproducibility, and interfiber reproducibility of these assemblies have been evaluated for highly volatile analytes, including benzene, 2-pentanone, 1-nitropropane, pyridine, and toluene spiked into pumping oil solution (Setkova et al. 2007c). A single DVB/CAR/ PDMS metal fiber assembly demonstrated excellent durability properties (lifetime of > 600 extraction/desorption cycles) when utilized in the application involving nontargeted determination of volatile and semivolatile constituents in ice wine samples and aiming at the verification of ice wine quality (Setkova et al. 2007a,b). It must be noted, however, that needle size of new metal fiber assemblies is 23-gauge, whereas fused-silica assemblies are available in both 23-gauge and 24-gauge sizes. With SPME, one potential problem that can be encountered when a large number of samples is submitted to analysis is septum coring, since the size of SPME needle is larger than the size of traditional liquid syringe for GC injection. Accumulation of septa pieces in the liner of the GC injector port results in reduced method sensitivity and reduced desorption efficiency, due to the adsorption of target analytes onto septum pieces; however, in extreme cases, it also may result in breakage of the SPME fiber assembly. With beveled tip needles of metal fiber assemblies, septum coring is even more prominent. The recent introduction of septumless injection systems for GC has overcome this problem. Septumless injection systems such as Merlin Microseal (Merlin Instrument, Half Moon Bay, CA) and a septumless sampling head (SLH) on a cooled injection system (CIS) PTV-type inlet (Gerstel GmbH, Mulheim an der Ruhr, Germany) have

SPME DEVICES OTHER THAN FIBER-SPME

by SPME is proportional to the volume of extraction phase; however, increasing the volume of the extraction phase by using thicker extraction phases also results in longer equilibration times (Bruheim et al. 2003). On the other hand, it has also been emphasized that the initial extraction rate is directly proportional to the surface area of the extraction phase (Pawliszyn 2007). Therefore, an ideal way to improve SPME method sensitivity by increasing extraction phase volume is to use the extraction phase with a large surface area-tovolume ratio, rather than increasing the thickness of the extraction phase and consequently the analysis time (Bruheim et al. 2003). These principles represent the underlying basis of the development of thin-film microextraction (TFME). The TFME technique utilizes a thin piece of polydimethylsiloxane film that is cut into a house-like shape having dimensions of 2  2 cm square and a 1-cm-high triangle on top (Qin et al. 2008). This thin film is attached to a stainlesssteel wire, and as such it can be coiled and fitted inside a glass liner after extraction (Pawliszyn 2007). Desorption is then carried out by an automated liner exchange procedure in TDS-2 thermal desorption system and with the use of a MPS2 autosampler (Fig. 15.23a).

made it possible to perform SPME analyses in a highthroughput manner with minimum analyst supervision.

15.6. SPME DEVICES OTHER THAN FIBER-SPME Increased understanding of theoretical principles of SPME has also led to the development of new and improved SPME devices. The following section briefly outlines descriptions and applications of some of the more recently developed SPME devices that have demonstrated promising performance in the laboratory and on-site analysis of environmental samples. Detailed information on the use of SPME devices in environmental analysis has been presented elsewhere (Pawliszyn 2007; Ouyang et al. 2005a, Ouyang and Pawliszyn 2006a,b, 2007a,b; Risticevic et al. 2009).

15.6.1. Thin-Film Microextraction (TFME) More recently, significant efforts have been directed toward obtaining increased extraction efficiencies with SPME. According to Equation (15.2), the amount of analyte extracted

Step 1. Rotate thin-film as it is inserted into the liner

(a)

Temperature controller

(b)

Power line

Step 2. Place cap onto the liner

CO2

Thin-film

Solenoid

Step 3. Introduce liner for thermal desorption

CO2 Hollow plunger housing the CO2 tubing and thermocouple

Syringe barrel

Coiled thin-film

Desorption liner

CO2 cylinder PDMS coating

(c)

12 Cm 10 Cm 3 Cm SB SP

399

SH NH

PS

Figure 15.23. Illustration of SPME devices other than fiber-SPME. (a) depiction of thin-film microextraction (TFME) principle; (b) experimental setup for cold-fiber HS-SPME procedure; (c) side-hole design needle trap device (NTD). (where SB—sorbent, SP—spiral plug, SH—side hole, NH—needle head, PS—PTFE sealer) [Figures reprinted with permission from Sanchez-Prado et al. (2010) and Eom et al. (2008b)].

400

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Bruheim and co-workers compared the performance of fiber SPME and TFME for the extraction of semivolatile compounds (PAHs) in water samples using direct and headspace extraction modes and concluded that higher extraction rates were obtainable with TFME (Bruheim et al. 2003). Bragg and colleagues later published a study in which PDMS thin films were used as passive samplers for on-site determination of PAHs in environmental waters (Bragg et al. 2006). Thin films were placed into small copper cages for easy deployment in the field and for preventing algal buildup, and as such were left in water for a sampling period of one month. The authors commented that the technique allowed quantification of fluoranthene and pyrene and proved to be a low-cost alternative to passive sampling. Ouyang and co-workers also deployed thin films as passive samplers in order to determine TWA concentrations of PAHs in lake waters and compared their performance to fiber-retracted device and PDMS-rod passive samplers. The authors concluded that among the three types of passive samplers tested, TFME was able to achieve the highest sensitivity (Ouyang et al. 2007). Qin and coworkers compared the performance of thin-film microextraction and stir bar sorptive extraction (SBSE) for analysis of PAHs in aqueous samples with controlled agitation conditions (active sampling) (Qin et al. 2008). Even though SBSE incorporates a magnetic stirring bar coated with a 0.5-mm-thick layer of PDMS (Twister) such that it possesses a higher amount of extraction phase and therefore higher sensitivity than does fiber SPME, the authors of this study reported that TFME manifested shorter equilibration times and higher extraction rates as compared to SBSE. 15.6.2. Cold-Fiber HS-SPME Devices As mentioned previously, HS-SPME extraction mode offers many possibilities in the analysis of complex environmental samples because the fiber coating is not in direct contact with the sample matrix. The problems that can be potentially encountered are associated with low volatility of some analytes and possible trapping and chemisorption of analytes in solid matrices (Zhang and Pawliszyn 1995). Therefore, as explained in Section 15.3, frequently, adjustment of SPME parameters in the form of increasing sample temperature and/ or adding organic modifiers is required in order to increase headspace (HS) capacity of analytes. Increased sample temperature enhances desorption of analytes from their native matrix and increases their concentration in headspace. However, heating the sample while improving HS capacity also causes decreased distribution constants and lowers the amount of analyte extracted at equilibrium. Since the extraction of analytes by the fiber coating is an exothermic process, increasing sample temperature may initially improve HS-SPME method sensitivity, and usually a further increase in sample temperature causes decreased distribution

constants, which can no longer be offset by increased HS capacity (Pawliszyn 2007; Zhang and Pawliszyn 1995). In such circumstances, the selected sample temperature condition usually involves the use of a temperature that is not high enough to allow efficient chemisorption and volatilization. A cold-fiber HS-SPME device was introduced in 1995 to deal with these drawbacks (Zhang and Pawliszyn 1995). The technique involves heating the sample matrix to very high temperatures while simultaneously cooling the fiber coating (with liquid CO2) to create a temperature gap, which significantly increases the distribution constants (Chen and Pawliszyn 2006). The originally developed cold-fiber device was further miniaturized by Chen and colleagues to render it compatible with the automated procedure (Chen and Pawliszyn 2006). The experimental setup of the cold-fiber HSSPME device is illustrated in Figure 15.23b (Sanchez-Prado et al. 2010). Chen and co-workers were able to achieve exhaustive extraction of hydrocarbons from air with this device after properly optimizing extraction conditions (Chen and Pawliszyn 2006). The cold-fiber device was also utilized for the extraction of flavor compounds from water and personal care products (Chen et al. 2007). Carasek and coworkers compared the extraction efficiency of commercially available SPME fiber coatings to the one obtained by the cold-fiber technique for the study involving nontargeted screening of tropical fruit volatile components and determined that the cold-fiber SPME device had the best performance characteristics (Carasek and Pawliszyn 2006). Quantitative extraction (absolute extraction recoveries 90%) of 2,4-dichloroanisole, 2,6-dichloroanisole, and 2,4,6-trichloroanisole from cork samples with the cold-fiber HS-SPME technique was also reported (Carasek et al. 2007). Ghiasvand and colleagues proposed the use of the cold-fiber HS-SPME technique for qualitative and nontargeted screening of solid food commodities such as rice, since premanipulation of solid rice samples was unnecessary before extraction (Ghiasvand et al. 2007). Cold-fiber HS-SPME was also tested in the application involving determination of PAHs in sediment, and the results obtained were in good agreement with certified reference values (Ghiasvand et al. 2006). 15.6.3. Needle Trap Devices As an alternative to the coated fiber in the needle SPME device, the sorbent can fill the whole diameter of the needle to form a needle trap device (NTD), a device that was developed and introduced in response to a demand for more robust SPME systems (Koziel et al. 2001). Tenax-filled needles for sampling and analysis of airborne volatile organic compounds (VOCs) were introduced in 1970; therefore, the concept of packed needles is not new (Koziel et al. 2001). However, in order to desorb analytes from these needles, a dedicated carrier gas purge line was required. Unlike these devices, the NTD device combines the idea of active

SPME APPLICATIONS AND PERFORMANCE CHARACTERISTICS

sampling and SPME, and the device can also be introduced into the GC injector port in a manner similar to that for conventional SPME (Koziel et al. 2001; Wang et al. 2005a). Similar to SPME, the use of NTDs represents a new approach to one-step, solvent-free sample preparation and injection. In contrast to SPME, NTD is an exhaustive technique, but it can work very effectively in combination with a fiber SPME device to better characterize the sample directly on site. Briefly, blunt-type hypodermic needles (having internal diameter of < 400 mm) packed with an appropriate sorbent are used for extraction of organic pollutants, and trapping is achieved by passing the gaseous sample through the needle (Eom et al. 2008a; Niri et al. 2009). Different types of packing materials have been tested, including quartz wool packing for sampling of particulate matter and aerosols in an inhaler-administered drug, spray insect repellant and tailpipe diesel exhaust, and a single layer of carboxen and multiple layers of polydimethylsiloxane, divinylbenzene, and carboxen for sampling of VOCs (Koziel et al. 2001; Wang et al. 2005a; Eom et al. 2008a,b). Wang and co-workers published a study on the optimization of desorption efficiency with needle traps (Wang et al. 2005a). With the direct syringe desorption technique, after withdrawing a known volume of VOCs from the standard gas generator through a NTD device with a syringe, 1.5 mL of post-extraction volume was kept in the plunger and injected when NTD was introduced in the GC injector port to aid in the desorption of analytes (Wang et al. 2005a). Strong memory effects were observed with this technique. With this in mind, a side-hole design needle trap device was introduced for use in combination with a narrow-necked GC insert (Fig. 15.23c) (Wang et al. 2005a; Eom et al. 2008b). With this design, desorption efficiency was significantly improved, as the side hole was opened during desorption, thus allowing carrier gas to enter the needle, pass through the sorbent, and facilitate the delivery of analytes onto the front of the column (Wang et al. 2005a; Eom et al. 2008b). Eom and colleagues published a study on the preparation, performance evaluation, and application of needle trap devices packed with divinylbenzene and carboxen particles for BTEX sampling in air from permanent marker fumes, mosquito coil smoke, and the interior inside a house (Eom et al. 2008b). The authors used a manual sampling pump, which requires no power source for controlling sampling flow rate in on-site sampling trials (Eom et al. 2008b). Niri and co-workers simultaneously used SPME and NTD to differentiate between free gaseous and particle-bound compounds in air samples such as mosquito coil smoke (Niri et al. 2009). SPME fiber coating was capable of extracting free molecules, and NTD collected both free and particlebound analytes, including less volatile allethrin, a higher boiling point compound that is present mostly as particlebound species (Niri et al. 2009). Therefore, the combination of the two needle-based devices can provide useful infor-

401

mation as to the extent of binding of the analytes to the particulate matter in air. Needle trap devices packed with carboxen were also used as TWA diffusive samplers for the collection of VOCs by molecular diffusion and adsorption to the packed sorbent (Gong et al. 2008).

15.7. SPME APPLICATIONS AND PERFORMANCE CHARACTERISTICS Several more recently published SPME applications in analysis of environmental pollutants in various air, water, solid, and other samples of environmental relevance and/or studies demonstrating the principles that were highlighted in this contribution are illustrated in Table 15.5. Reviews by Krutz, Wardencki, Ouyang and others discuss additional applications of SPME for environmental analysis (Krutz et al. 2003; Wardencki et al. 2007; Ouyang and Pawliszyn 2006b). At this point, comparison between SPME and several traditional sample preparation methods as well as a summary of interesting SPME applications in the preparation of atypical matrices in the environmental analysis area would be useful. Doong and colleagues published a study on the determination of organochlorine pesticides and their metabolites in soil samples using HS-SPME (see Table 15.5 for SPME conditions) (Doong and Liao 2001). The authors compared the results obtained using SPME with those of Soxhlet extraction and found that method precision was comparable [relative standard deviations (RSDs; n ¼ 6) ranged between 3.2% and 26.3% for SPME and between 2.6% and 19.6% for Soxhlet extraction] and the limits of detection (LODs) obtained were in good agreement (0.06–0.65 ng/g for SPME, 1.03–5.91 ng/g for Soxhlet extraction, and 1.1–5.7 ng/g for US EPA Method 8081). The analysis of certified reference material (CRM) 804–050 soil was also used for SPME method validation, and excellent correlations were detected between certified reference values (18.04–1530.6 mg/kg) and average concentration values obtained using SPME (16.5–1459.6 mg/kg) (Doong and Liao 2001). Schurek and colleagues applied the HS-SPME method (see specifications in Table 15.5) [hyphenated to comprehensive two-dimensional gas chromatography–time-offlight mass spectrometry (GC  GC-TOFMS)] toward the determination of pesticide residues in tea samples (Schurek et al. 2008). The authors compared the method to the conventional ethyl acetate extraction followed by highperformance gel permeation chromatography (HPGPC) and reached the following conclusions: (1) the repeatability of SPME measurements expressed in terms of RSD ranged between 2% and 24% at the test level of 50 mg/kg (RSD values lower than 22% are required for the concentration range 0.01–0.1 mg/kg, according to Document SANCO/ 10232/2006); (2) limits of quantification (LOQs) ranged

402 18 mL water

18 mL water

Tapwater, river water

Drinking water, surface water

Tapwater, seawater, well water, superficial water River water, sewage water

VOCs; methyl-tert-butylether; 1,4-dioxane; 2-methylisoborneol; geosmin Insecticides; herbicides; fungicides (organochlorine, organophosphorous, triazines, pyrethroids) PAHs

PCB congeners

Fragrance allergens

Baby bathwater, swimming pool water Sediment porewater samples

Fern
andez-Gonz
ales et al. (2007)

Extraction—DI-SPME with 65 mm PDMS/ DVB for 60 min at 60 C and 500 rpm; desorption—10 min, 300 C Extraction—HS-SPME with 65 mm PDMS/ DVB for 30 min at room temperature; desorption—2 min, 300 C Extraction—HS-SPME with 65 mm PDMS/ DVB for 20 min at 100 C; desorption— 220 C Extraction—7 mm PDMS for 30 min; desorption—5 min, 320 C

3 g NaCl, 10 mL water

Ozone

Oximes

Chlorinated toluenes

Beceiro-Gonz
ales et al. (2007)

Extraction—DI-SPME with 60 mm PDMS/ DVB for 45 min at 60 C and 500 rpm; desorption—5 min, 250 C

Standard gaseous mixtures, glass sampling bulb used in lab and field Standard gas generation system

Strawberries, greenhouse

Atmospheric pesticides

1.5 mL water

20% NaCl, 10 mL water

10 mL water

Extraction—100 mm PDMS for 140 min at room temperature; desorption—3 min, 295 C Extraction—100 mm PDMS for 30 min at temperature 16  C–32 C; desorption— 4 min, 250 C Derivatization reagent loading—HS above 2 mL of 17 mg/mL PFBHA solution for 1 min, 30 s immersion in 0.2 mg/mL DPE solution; simultaneous extraction derivatization—PDMS/DVB; desorption—7 min, 250 C Extraction—HS-SPME with 100 mm PDMS for 30 min at 60 C; desorption—1 min, 270 C

Air in laboratory and lecture hall

Organophosphate esters

Glass sampling bulbs (volume 0.5 L); for field sampling air sampling pump was connected to glass bulb 3.7 L/min volumetric airflow applied to the field sampler

Hawthorne et al. (2009)

Lamas et al. (2009)

Regueiro et al. (2009b)

Nakamura and Daishima (2005)

Lee and Tsai (2008)

Wang et al. (2009)

Tollb€ack et al. (2010)

Ras et al. (2008)

Air at sewage treatment plants

Tumbiolo et al. (2004)

Extraction—DI-SPME with 75 mm CAR/ PDMS for 30 min; desorption—2 min, 260 C Extraction—DI-SPME with CAR/PDMS for 45 min at room temperature; desorption— 2 min, 200 C

Volatile organic sulfur compounds

Standard gaseous mixtures, ambient air

Indoor and outdoor air

Reference

SPME Procedure

BTEX

Sample Preparation

Sample Matrix

Target Analytes

TABLE 15.5. Selected SPME Applications and Description of Conditions Employed in Environmental Analysis and/or Analysis of Environmental Pollutants

403

River water, wastewater, swimming pool water Rainwater

Water

Water containing product residues from explosive and pharmaceutical industries Wine, fruit juices

Olive oil

Olive oil

Bovine milk

Parabens, triclosan, chlorophenols

Phenoxy acid herbicides, phenyl urea herbicides, bromoxynil

Organometallic compounds of mercury, lead, and tin

Nitroaromatic compounds

Organophosphorus insecticides and metabolites

Volatile and semivolatile aromatic hydrocarbons

Pesticides

Organophosphorous pesticides

Surface water, sewage water

Alkyl and benzyl esters of p-hydroxybenzoic acid (parabens)

1 mL milk diluted with 9 mL water (1:10 dilution ratio)

2 g olive oil

Wines—5 mL of wine; fruit juices—centrifuged for 60 s at 5000 rpm, 1:25 dilution with water, 5 mL final volume 5 g olive oil

1 mL buffer solution (pH 5.3), fixed volume of 100 mg/L organometallic standard solution in methanol and water (total volume 5 mL); pre-extraction derivatization—100 mL of 2% NaBEt4 solution, derivatization time 150 min 4 mL water

3.5 g NaCl, 10 mL water; preextraction derivatization—100 mL acetic anhydride added pH 2, 75% NaCl, 3 mL water

pH 6, 150 mg/mL NaCl, 10 mL water

Zambonin et al. (2004)

Tsoutsi et al. (2006)

Extraction—DI-SPME with 85 mm PA for 30 min at room temperature; desorption— 5 min, 250 C

Extraction—HS-SPME with 100 mm PDMS for 60 min at 75 C; desorption—7 min, 250 C Extraction—HS-SPME with 50/30 mm, 2 cm DVB/CAR/PDMS for 60 min at 100 C; desorption—5 min, 265 C Extraction—DI-SPME with 65 mm PDMS/ DVB for 30 min at 100 C; desorption— 5 min, 270 C

(continued)

Fernandez-Alvarez et al. (2008a)

Vichi et al. (2005)

J€onsson et al. (2007)

Beceiro-Gonz
alez et al. (2009)

Scheyer et al. (2007)

Regueiro et al. (2009a)

Canosa et al. (2006)

Extraction—DI-SPME with 65 mm PDMS/ DVB for 60 min at room temperature; desorption—3 min, 250 C

Derivatization reagent loading—HS above 10 mL PFBBr for 10 min; simultaneous extraction and derivatization—65 mm PDMS/DVB for 60 min at 68 C; desorption—5 min, 250 C Extraction—HS-SPME with 50/30 mm DVB/ CAR/PDMS for 15 min at 40 C; desorption—2 min, 260 C

Extraction—DI-SPME with 85 mm PA for 40 min at room temperature; in-fiber postextraction derivatization—HS above 20 mL MTBSTFA for 10 min at room temperature; desorption—3 min, 280 C Extraction—HS-SPME with 50/30 mm DVB/ CAR/PDMS for 15 min at 100 C; desorption—2 min, 240 C

404 0.5 g sample, 5 mL of water

Rat serum and brain tissue Sediment, soil

Soil

Hexanal

PCB 126 and 153 congeners

brominated flame retardants (BFRs) - (polybrominated biphenyls (PBBs), polybrominated diphenyl ethers (PBDEs) Organochlorine pesticides and metabolites Soil

0.5 g sample, 2 mL water

Canned tuna, clam, anchovy, mussel; fresh salmon, smoked salmon Beef bouillons

PAHs

Pyrethroids, organochlorines, and similar plant protection agents

Tissue: 100 mg, 10% KCl; serum: 150 mL

Green, black, and fruit teas

Pesticides

0.5 g sample, 0.5 mL water

0.5 g bouillon, 5 mL water

5 mL of 24% w/v NaCl solution, 4 g sample

1 g honey sample, 2 g NaCl, 6.7 mL of 0.75% w/v potassium carbonate (pH 11); pre-extraction derivatization—280 mL acetic anhydride added 2 g tea, 2 mL water

Honey

Chlorophenols

Sample Preparation

Sample Matrix

Target Analytes

TABLE 15.5 (Continued )

Schurek et al. (2008)

Extraction—HS-SPME with 100 mm PDMS for 60 min at 70 C; desorption—2 min, 270 C Extraction—HS-SPME with 65 mm PDMS/ DVB for 60 min at 75 C; desorption— 5 min, 270 C

Extraction—HS-SPME with 100 mm PDMS for 60 min at 70 C; desorption—5 min, 270 C Extraction—HS-SPME with 85 mm PA for 30 min at 100 C; desorption—5 min, 290 C

Extraction—HS-SPME with 50/30 mm DVB/ CAR/PDMS for 40 min at 37 C; desorption—1 min Extraction—HS-SPME with 85 mm PA for 40 min at 100 C; desorption—10 min, 280 C Extraction—HS-SPME with 100 mm PDMS for 60 min at 100 C; desorption—3 min, 280 C

Campillo et al. (2006)

Extraction—HS-SPME with 65 mm PDMS/ DVB for 15 min at 75 C; desorption— 1 min, 200 C

Fernandez-Alvarez et al. (2008b)

Doong and Liao (2001)

Salgado-Petinal et al. (2006)

Poli et al. (2009)

Giuffrida et al. (2005)

Aguinaga et al. (2008)

Reference

SPME Procedure

FUTURE PERSPECTIVES OF SPME

from 1 to 28 mg/kg, and as such, SPME demonstrated potential to detect and quantify most of the target pesticides in tea samples at low ppb levels [levels detectable and quantifiable by SPME were lower than most maximum residue limits required by EU regulations for this particular matrix], and (3) the HS-SPME procedure generated much simpler and cleaner chromatograms as compared to injection of solvent extracts, as the latter approach hindered detection of four target pesticides and caused buildup of non-volatile deposits in the injector and front part of a separation capillary (Schurek et al. 2008). Aguinaga and co-workers evaluated, the SPME method for determination of PAHs in aquatic species such as canned tuna, clam, anchovy, mussel, and fresh and smoked salmon (Aguinaga et al. 2008). While the repeatability of the optimized SPME method (see conditions in Table 15.5) expressed in terms of RSD ranged from 5% to 15% and detection limits from 8 to 450 pg/g, the validation of the proposed method was assessed by analyzing standard reference mussel tissue (SRM 1974b) and obtained SPME results were in excellent agreement with the reference values (Aguinaga et al. 2008). Halasz and colleagues critically reviewed the current progress on the utilization of SPME-GC and SPME-HPLC in the determination of frequently encountered environmental chemicals and their (bio)transformation pathways (Halasz and Hawari 2006). The authors provided three examples: biodesulfurization pathway of dibenzothiophene, biotransformation routes of hydrocarbons, and biotransformation of explosives (Halasz and Hawari 2006). The authors emphasized that SPME providing rapid and sensitive determinations of organic compounds is perfectly suitable for the detection of trace amounts of relatively shortlived intermediate products (metabolites) in chemical and biochemical reactions. This is in contrast to traditional techniques that have been used so far to isolate and extract intermediates produced during (bio)transformation processes such as liquid–liquid extraction (LLE) and solid-phase extraction (SPE), which do not provide rapid sample preparation, thus leading to the loss of valuable information on the transformation pathways (Halasz and Hawari 2006). SanchezPrado and co-workers also recently reported the use of cold-fiber HS-SPME to investigate photodegradation of volatile analytes (Sanchez-Prado et al. 2010). Briefly, hexachlorobenzene (HCB) used as a model volatile compound after being sorbed by cooled PDMS coating was exposed to ultraviolet (UV) irradiation (254 nm), and photolysis took place directly in the PDMS coating, leading to in situ generation of photoproducts (pentachlorobenzene, tetrachlorobenzenes, and trichlorobenzene). In this particular case, cooled PDMS coating was used as a reaction medium, and this approach minimized analyte losses due to volatilization (Sanchez-Prado et al. 2010).

405

15.8. FUTURE PERSPECTIVES OF SPME In addition to various SPME applications for in vitro analysis of environmental contaminants, SPME also offers some unique possibilities that might provide solutions to many future challenges and questions in the field of environmental analysis. For example, as shown in Equation (15.5) and discussed in Section 15.3.1.5, under certain conditions the amount of analyte extracted by SPME is independent of sample volume. This feature of SPME was exploited for direct in vivo monitoring of environmental pollutants in muscle and adipose tissue of fish (Zhou et al. 2008a; Zhang et al. 2009b). The main steps of in vivo SPME sampling were: (1) induction of brief anesthesia, (2) insertion of 20-gauge needle into the dorsal–epaxial muscle to a depth of approximately 12–13 mm, (3) removal of the needle, (4) insertion of SPME probe with PDMS coating into the muscle for 20 min, (5) induction of brief anesthesia, and (6) removal of SPME probe. During a 20-min SPME sampling interval the fish regained consciousness and moved freely within the enclosure. This procedure was subsequently also extended to field sampling of different wild fish species by temporarily stunning the fish using an electroshocking unit (Zhou et al. 2008a). In addition to quantitative determination of target analytes (pharmaceuticals and other bioactive compounds), these studies also provided information regarding the distribution of these species in muscle versus adipose tissue. The development of this nonlethal sampling methodology, which does not require tissue removal prior to sampling, provides a new, rapid, and simple alternative for bioaccumulation and toxicity studies in fish. In vivo SPME was also successfully applied for in vivo monitoring of pesticides in plants (Lord et al. 2004; Zhou et al. 2008c). An additional feature of microextraction methods such as SPME that might be valuable in future studies is that the amount extracted is proportional to free concentration. By employing two calibration methods, SPME can be used for the simultaneous measurement of free and total concentrations in a single sample. This is in contrast to traditional sample preparation methods, which yield only free or total concentration, thus requiring two sets of methods if both concentrations are of interest. Thus, SPME can be used as a rapid and time- and cost-effective method to estimate bioavailability. For example, Bondarenko and Gan have used SPME to determine the dissolved organic carbon/water partitioning coefficient (KDOC) and ecotoxicologically relevant free concentration for eight pyrethroids in sediment porewater and field-contaminated samples (Bondarenko and Gan 2009). The proposed method was also validated against liquid–liquid extraction, and the main advantages of SPME versus LLE were cited as (1) single-step sample preparation minimizes sample losses, (2) smaller sample volumes are needed for SPME, and (3) it is possible to reanalyze samples when negligible depletion is used. Also, Hawthorne et al.

406

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

published a study associated with determination of 62 polychlorinated biphenyl congeners in sediment porewater samples (see Table 15.5 for SPME conditions) using SPME and GC-MS (Hawthorne et al. 2009). The authors summarized the following practical advantages of the SPME method: (1) the use of small water sample volumes (a few milliliters of water), (2) simplified separation and preparation of porewater, and (3) the ability to transfer all the extracted analytes to the GC injection port (Hawthorne et al. 2009). The authors assessed these advantages from comparison of SPME with conventional solvent extraction, as the latter requied liter quantities of porewater and transfer of only a tiny fraction of solvent extract into GC injection port. Besides good sensitivity (the detection limits for PCB congeners obtained in this study ranged from 0.7 to 3.3 pg/mL) and good correlation of SPME results with those of hexane extraction (hexane extract concentrations averaged 104  9% of SPME values), the authors were also able to distinguish between “freely dissolved” (in the free water phase only) and “total dissolved” (in the free water phase and associated dissolved organic matter) analytes (Hawthorne et al. 2009).

15.9. CONCLUSIONS This contribution provided an introduction to theoretical principles of SPME and SPME method development approaches, with particular emphasis on the analysis of complex real-life environmental samples and/or samples in which environmental contaminants are likely to be present. Calibration approaches, including traditional ones and those that can allow high-throughput and on-site determinations, were thoroughly discussed while being referenced to examples and challenges that analytical chemists are faced with in everyday protocols. More recent developments in automation, high-throughput approaches, and production of new devices that demonstrated promising performance in environmental analysis were addressed. As was illustrated, SPME represents a modern solvent-free and high-throughput alternative to traditional sample preparation techniques, and with this trend the number of SPME applications in the future is likely to increase.

REFERENCES Aguinaga, N., Campillo, N., Vin˜as, P., and Hern
andez-Cordoba, M. (2008), Evaluation of solid-phase microextraction conditions for the determination of polycyclic aromatic hydrocarbons in aquatic species using gas chromatography, Anal. Bioanal. Chem. 391, 1419–1424. Ai, J. (1997a) , Headspace solid phase microextraction. Dynamics and quantitative analysis before reaching a partition equilibrium, Anal. Chem. 69, 3260–3266.

Ai, J. (1997b) , Solid-phase microextraction for quantitative analysis in nonequilibrium situations, Anal. Chem. 69, 1230–1236. Aresta, A., Vatinno, R., Palmisano, F., and Zambonin, C. G. (2006), Determination of Ochratoxin A in wine at sub ng/mL levels by solid-phase microextraction coupled to liquid chromatography with fluorescence detection, J. Chromatogr. A 1115, 196–201. Arthur, C. L. and Pawliszyn, J. (1990), Solid phase microextraction with thermal desorption using fused silica optical fibers, Anal. Chem. 62, 2145–2148. Baltussen, E., Cramers, C. A., and Sandra, P. J. F. (2002), Sorptive sample preparation—a review, Anal. Bioanal. Chem. 373, 3–22. Beceiro-Gonz
ales, E., Concha-Gran˜a, E., Guimaraes, A., Gon¸calves. C., Muniategui-Lorenzo, S., and Alpendurada, M. F. (2007), Optimisation and validation of a solid-phase microextraction method for simultaneous determination of different types of pesticides in water by gas-chromatography-mass spectrometry, J. Chromatogr. A 1141, 165–173. Beceiro-Gonz
ales, E., Guimaraes, A., and Alpendurada, M. F. (2009), Optimisation of a headspace-solid-phase micro-extraction method for simultaneous determination of organometallic compounds of mercury, lead and tin in water by gas chromatography–tandem mass spectrometry, J. Chromatogr. A 1216, 5563–5569. Belardi, R. G. and Pawliszyn, J. (1989), The application of chemically modified fused silica fibers in the extraction of organics from water matrix samples and their rapid transfer to capillary columns, Water Pollut. Res. J. Can. 24, 179–191. Bondarenko, S. and Gan, J. (2009), Simultaneous measurement of free and total concentrations of hydrophobic compounds, Environ. Sci. Technol. 43, 3772–3777. Bragg, L., Qin, Z., Alaee, M., and Pawliszyn, J. (2006), Field sampling with a polydimethylsiloxane thin-film, J. Chromatogr. Sci. 44, 317–323. Bruheim, I., Liu, X., and Pawliszyn, J. (2003), Thin-film microextraction, Anal. Chem. 75, 1002–1010. Buchholz, K. D. and Pawliszyn, J. (1994), Optimization of solidphase microextraction conditions for determination of phenols, Anal. Chem. 66, 160–167. Buldini, P. L., Ricci, L., and Sharma, J. L. (2002), Recent applications of sample preparation techniques in food analysis, J. Chromatogr. A 975, 47–70. Cajka, T., Hajslova, J., Pudil, F., and Riddellova, K. (2009), Traceability of honey origin based on volatiles pattern processing by artificial neural networks, J. Chromatogr. A 1216, 1458– 1462. Cam, D., Gagni, S., Lombardi, N., and Punin, M. O. (2004), Solidphase microextraction and gas chromatography–mass spectrometry for the determination of polycyclic aromatic hydrocarbons in environmental solid matrices, J. Chromatogr. Sci. 42, 329– 335. Cam, D. and Gagni, S. (2001), Determination of petroleum hydrocarbons in contaminated soils using solid-phase microextraction with gas chromatography-mass spectrometry, J. Chromatogr. Sci. 39, 481–486. Campillo, N., Pen˜alver, R., and Hern
andez-Co´rdoba, M. (2006), Evaluation of solid-phase microextraction conditions for the

REFERENCES

determination of chlorophenols in honey samples using gas chromatography, J. Chromatogr. A 1125, 31–37. Canosa, P., Rodr
ıguez, I., Rub
ı, E., Bolla
ın, M. H., and Cela, R. (2006), Optimisation of a solid-phase microextraction method for the determination of parabens in water samples at the low ng per litre level, J. Chromatogr. A 1124, 3–10. Carasek, E. and Pawliszyn, J. (2006), Screening of tropical fruit volatile compounds using solid-phase microextraction (SPME) fibers and internally cooled SPME fiber, J. Agric. Food Chem. 54, 8688–8696. Carasek, E., Cudjoe, E., and Pawliszyn, J. (2007), Fast and sensitive method to determine chloroanisoles in cork using an internally cooled solid-phase microextraction fiber, J. Chromatogr. A 1138, 10–17. Chen, J. and Pawliszyn, J. (1995), Solid phase microextraction coupled to high-performance liquid chromatography, Anal. Chem. 67, 2530–2533. Chen, Y. and Pawliszyn, J. (2004). Kinetics and the on-site application of standards in a solid-phase microextraction fiber, Anal. Chem. 76, 5807–5815. Chen, Y. and Pawliszyn, J. (2006), Miniaturization and automation of an internally cooled coated fiber device, Anal. Chem. 78, 5222–5226. Chen, Y., Begnaud, F., Chaintreau, A., and Pawliszyn, J. (2007), Analysis of flavor and perfume using an internally cooled coated fiber device, J. Separ. Sci. 30, 1037–1043. Chen, Y., O’Reilly, J., Wang, Y., and Pawliszyn, J. (2004). Standards in the extraction phase, a new approach to calibration of microextraction processes, Analyst 129, 702–703. Deger, A. B., Gremm, T. J., Frimmel, F. H., and Mendez, L. (2003), Optimization and application of SPME for the gas chromatographic determination of endosulfan and its major metabolites in the ng L1 range in aqueous solutions, Anal. Bioanal. Chem. 376, 61–68. De Jager, L. S., Perfetti, G. A., and Diachenko, G. W. (2008), Analysis of tetramethylene disulfotetramine in foods using solid-phase microextraction-gas chromatography-mass spectrometry, J. Chromatogr. A 1192, 36–40. Donato, P., Tranchida, P. Q., Dugo, P., Dugo, G., and Mondello, L. (2007), Rapid analysis of food products by means of high speed gas chromatography, J. Separ. Sci. 30, 508–526. Doong, R.-A. and Liao, P.-L. (2001), Determination of organochlorine pesticides and their metabolites in soil samples using headspace solid-phase microextraction, J. Chromatogr. A 918, 177–188. Eisert, R. and Pawliszyn, J. (1997), Automated in-tube solid-phase microextraction coupled to high-performance liquid chromatography, Anal. Chem. 69 3140–3147. Eom, I.-Y., Niri, V. H., and Pawliszyn, J. (2008a) , Development of a syringe pump assisted dynamic headspace sampling technique for needle trap device, J. Chromatogr. A 1196–1197 10–14. Eom, I.-Y., Tugulea, A.-M., and Pawliszyn, J. (2008b) , Development and application of needle trap devices, J. Chromatogr. A 1196–1197 3–9. Fang, H., Liu, M., and Zeng, Z. (2006), Solid-phase microextraction coupled with capillary electrophoresis to determine ephedrine

407

derivatives in water and urine using a sol-gel derived butyl methacrylate/silicone fiber, Talanta 68, 979–986. Fernandez-Alvarez, M., Llompart, M., Lamas, J. P., Lores, M., Garcia-Jares, C., Cela, R., and Dagnac, T. (2008a) , Development of a solid-phase microextraction gas chromatography with microelectron-capture detection method for a multiresidue analysis of pesticides in bovine milk, Anal. Chim. Acta 617, 37–50. Fernandez-Alvarez, M., Llompart, M., Lamas, J. P., Lores, M., Garcia-Jares, C., Cela, R., and Dagnac, T. (2008b) , Simultaneous determination of traces of pyrethroids, organochlorines and other main plant protection agents in agricultural soils by headspace solid-phase microextraction-gas chromatography, J. Chromatogr, A 1188, 154–163. Fern
andez-Gonz
alez, V., Concha-Gran˜a, E., Muniategui-Lorenzo, S., Lo´pez-Mah
ıa, P., and Prada-Rodr
ıguez, D. (2007), Solidphase microextraction-gas chromatographic-tandem mass spectrometric analysis of polycyclic aromatic hydrocarbons. Towards the European Union water directive 2006/0129 EC, J. Chromatogr. A 1176, 48–56. Fromberg, A., Nilsson, T., Larsen, B. R., Montanarella, L., Facchetti, S., and Madsen, J. Ø. (1996), Analysis of chloro- and nitroanilines and -benzenes in soils by headspace solid-phase microextraction, J. Chromatogr. A 746, 71–81. Ghiasvand, A. R., Hosseinzadeh, S., and Pawliszyn, J. (2006), New cold-fiber headspace solid-phase microextraction device for quantitative extraction of polycyclic aromatic hydrocarbons in sediment, J. Chromatogr. A 1124, 35–42. Ghiasvand, A. R., Setkova, L., and Pawliszyn, J. (2007), Determination of flavour profile in Iranian fragrant rice samples using cold-fiber SPME–GC–TOF–MS, Flavour Fragr. J. 22, 377–391. Giuffrida, F., Golay, P.-A., Destaillats, F., Hug, B., and Dionisi, F. (2005), Accurate determination of hexanal in beef bouillons by headspace solid-phase microextraction gas-chromatography mass-spectrometry, Eur. J. Lipid Sci. Technol. 107, 792–798. Gong, Y., Eom, I.-Y., Lou, D. -W, Hein, D., and Pawliszyn, J. (2008), Development and application of a needle trap device for timeweighted average diffusive sampling, Anal. Chem. 80, 7275– 7282. Gorecki, T., Yu, X., and Pawliszyn, J. (1999), Theory of analyte extraction by selected porous polymer SPME fibers, Analyst 124, 643–649. Halasz, A. and Hawari, J. (2006), SPME in environmental analysis: Biotransformation pathways, J. Chromatogr. Sci. 44, 379–386. Hawthorne, S. B., Grabanski, C. B., and Miller, D. J. (2009), Solidphase-microextraction measurement of 62 polychlorinated biphenyl congeners in milliliter sediment pore water samples and determination of KDOC values, Anal. Chem. 81, 6936–6943. Herr
aez-Hern
andez, R., Ch
afer-Peric
as, C., Verdu´-Andr
es, J., and Camp
ıns-Falco´, P. (2006), An evaluation of solid phase microextraction for aliphatic amines using derivatization with 9fluorenylmethyl chloroformate and liquid chromatography, J. Chromatogr. A 1104, 40–46. J€ onsson, S., Gustavsson, L., and van Bavel, B. (2007), Analysis of nitroaromatic compounds in complex samples using solid-phase

408

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

microextraction and isotope diluation quantification gas chromatography–electron-capture negative ionisation mass spectroemtry, J. Chromatogr. A 1164, 65–73. Kataoka, H., Lord, H. L., and Pawliszyn, J. (2000), Applications of solid-phase microextraction in food analysis, J. Chromatogr. A 880, 35–62. Koziel, J. A., Odziemkowski, M., and Pawliszyn, J. (2001), Sampling and analysis of airborne particulate matter and aerosols using in-needle trap and SPME fiber devices, Anal. Chem. 73, 47–54. Krutz, L. J., Senseman, S. A., and Sciumbato, A. S. (2003), Solidphase microextraction for herbicide determination in environmental samples, J. Chromatogr. A 999, 103–121. Lamas, J. P., Sanchez-Prado, L., Garcia-Jares, C., and Llompart, M. (2009), Solid-phase microextraction gas chromatography-mass spectrometry determination of fragrance allergens in baby bathwater, Anal. Bioanal. Chem. 394, 1399–1411. Lambropoulou, D. A. and Albanis, T. A. (2002), Headspace solid phase microextraction applied to the analysis of organophosphorus insecticides in strawberry and cherry juices, J. Agric. Food Chem. 50, 3359–3365. Lambropoulou, D. A. and Albanis, T. A. (2003), Headspace solidphase microextraction in combination with gas chromatography-mass spectrometry for the rapid screening of organophosphorus insecticide residues in strawberries and cherries, J. Chromatogr. A 993, 197–203. Lee, I.-S. and Tsai, S.-W. (2008), Passive sampling of ambient ozone by solid phase microextraction with on-fiber derivatization, Anal. Chim. Acta 610, 149–155. Liu, X., Ji, Y., Zhang, H., and Liu, M. (2008), Elimination of matrix effects in the determination of bisphenol A in milk by solidphase microextraction–high performance liquid chromatography, Food Addit. Contam., A 25, 772–778. Llompart, M., Li, K., and Fingas, M. (1999a) , Headspace solidphase microextraction for the determination of polychlorinated biphenyls in soils and sediments, J. Microcolumn Sep. 11, 397– 402. Llompart, M., Li, K., and Fingas, M. (1999b), Headspace solid phase microextraction (HSSPME) for the determination of volatile and semivolatile pollutants in soils, Talanta 48, 451– 459. Lopez Monzon A., Vega Moreno, D., Torres Padron, M. E., Sosa Ferrera, Z., and Santana Rodriguez, J. J. (2007), Solid-phase microextraction of benzimidazole fungicides in environmental liquid samples and HPLC-fluorescence determination, Anal. Bioanal. Chem. 387 1957–1963. Lord, H. and Pawliszyn, J. (2000), Evolution of solid-phase microextraction technology, J. Chromatogr. A 885, 153–193. Lord, H. L., Moder, M., Popp, P., and Pawliszyn, J. B. (2004), In vivo study of triazine herbicides by SPME, Analyst 129, 107–108. Lord, H. (2007), Strategies for interfacing solid-phase microextraction with liquid chromatography, J. Chromatogr. A 1152, 2–13. Louch, D., Motlagh, S., and Pawliszyn, J. (1992), Dynamics of organic compound extraction from water using liquid-coated fused silica fibers, Anal. Chem. 64, 1187–1199.

Mills, G. A. and Walker, V. (2000), Headspace solid-phase microextraction procedurers for gas chromatographic analysis of biological fluids and materials, J. Chromatogr. A 902, 267–287. Moeder, M., Schrader, S., Winkler, M., and Popp, P. (2000), Solidphase microextraction–gas chromatography-mass spectrometry of biologically active substances in water samples, J. Chromatogr. A 873, 95–106. Motlagh, S. and Pawliszyn, J. (1993), On-line monitoring of flowing samples using solid phase microextraction-gas chromatography, Anal. Chim. Acta 284, 265–273. Mughari, A. R., Vazquez, P. P., and Galera, M. M. (2007), Analysis of phenylurea and propanil herbicides by solid-phase microextraction and liquid chromatography combined with post-column photochemically induced fluorimetry derivatization and fluorescence detection, Anal. Chim. Acta 593, 157–163. Musteata, M. L., Musteata, F. M., and Pawliszyn, J. (2007), Biocompatible solid-phase microextraction coatings based on polyacrylonitrile and solid-phase extraction phases, Anal. Chem. 79, 6903–6911. Nagasawa, N., Yashiki, M., Iwasaki, Y., Hara, K., and Kojima, T. (1996), Rapid analysis of amphetamines in blood using head space-solid phase microextraction and selected ion monitoring, Forens. Sci. Int. 78, 95–102. Nakamura, S. and Daishima, S. (2005), Simultaneous determination of 22 volatile organic compounds, methyl-tert-butyl ether, 1, 4dioxane, 2-methylisoborneol and geosmin in water by headspace solid phase microextraction-gas chromatography-mass spectrometry, Anal. Chim. Acta 548, 79–85. Ner
ın, C., Salafranca, J., Aznar, M., and Batlle, R. (2009), Critical review on recent developments in solventless techniques for extraction of analytes, Anal. Bioanal. Chem. 393, 809–833. Niri, V. H., Eom, I. -Y., Kermani, F. R., and Pawliszyn, J. (2009), Sampling free and particle-bound chemicals using solid-phase microextraction and needle trap device simultaneously, J. Separ. Sci. 32, 1075–1080. Nongonierma, A., Cayot, P., Le Qu
er
e, J. -L., Springett, M., and Voilley, A. (2006), Mechanisms of extraction of aroma compounds from foods, using adsorbents: Effect of various parameters, Food Rev. Int. 22, 51–94. O’Reilly, J., Wang, Q., Setkova, L., Hutchinson, J. P., Chen, Y., Lord, H. L., Linton, C. M., and Pawliszyn, J. (2005), Automation of solid-phase microextraction, J. Separ. Sci. 28, 2010–2022. Ouyang, G. and Pawliszyn, J. (2006a), Recent developments in SPME for on-site analysis and monitoring, Trends Anal. Chem. 25, 692–703. Ouyang, G. and Pawliszyn, J. (2006b), SPME in environmental analysis, Anal. Bioanal. Chem. 386, 1059–1073. Ouyang, G. and Pawliszyn, J. (2007a), Configurations and calibration methods for passive sampling techniques, J. Chromatogr. A 1168, 226–235. Ouyang, G. and Pawliszyn, J. (2007b), Passive sampling devices for measuring organic compounds in soils and sediments, Compr. Anal. Chem. 48, 379–390. Ouyang, G. and Pawliszyn, J. (2008), A critical review in calibration methods for solid-phase microexraction, Anal. Chim. Acta 627, 184–197.

REFERENCES

Ouyang, G., Cai, J., Zhang, X., Li, X., and Pawliszyn, J. (2008), Standard-free kinetic calibration for rapid on-site analysis by solid-phase microextraction, J. Separ. Sci. 31, 1167–1172. Ouyang, G., Chen, Y., and Pawliszyn, J. (2005a) , Time-weighted average water sampling with a solid-phase microextraction device, Anal. Chem. 77, 7319–7325. Ouyang, G., Chen, Y., Setkova, L., and Pawliszyn, J. (2005b) , Calibration of solid-phase microextraction for quantitative analysis by gas chromatography, J. Chromatogr. A 1097, 9– 16. Ouyang, G., Cui, S., Qin, Z., and Pawliszyn, J. (2009), One-calibrant kinetic calibration for on-site water sampling with solid-phase microextraction, Anal. Chem. 81, 5629–5636. Ouyang, G., Zhao, W., Bragg, L., Qin, Z., Alaee, M., and Pawliszyn, J. (2007). Time-weighted average water sampling in Lake Ontario with solid-phase microextraction passive samplers, Environ. Sci. Technol. 41, 4026–4031. Pan, L. and Pawliszyn, J. (1997), Derivatization/solid-phase microextraction: new approach to polar analytes, Anal. Chem. 69, 196–205. Pawliszyn, J. (2007), Handbook of Solid Phase Microextraction, University of Waterloo, Waterloo, Canada. Pawliszyn, J. (2001), 2000 Maxxam Award Lecture: Unified theory of extraction, Can. J. Chem. 79, 1403–1414. Pawliszyn, J. (2003), Sample preparation: Quo Vadis? Anal. Chem. 75, 2543–2558. Pawliszyn, J. (1997), Solid Phase Microextraction: Theory and Practice, Wiley-VCH, New York. Pawliszyn, J. and Liu, S. (1987), Sample introduction for capillary gas chromatography with laser desorption and optical fibers, Anal. Chem. 59, 1475–1478. Pawliszyn, J. and Pedersen-Bjergaard, S. (2006), Analytical microextraction: Current status and future trends, J. Chromatogr. Sci. 44, 291–307. ˜ Penalver, A., Pocurull, E., Borrull, F., and Marc
e, R. M. (2002), Method based on solid-phase microextraction–high-performance liquid chromatography with UV and electrochemical detection to determine estrogenic compounds in water samples, J. Chromatogr. A 964, 153–160. Pino, V., Ayala, J. H., Afonso, A. M., and Gonz
alez, V. (2003), Micellar microwave-assisted extraction combined with solidphase microextraction for the determination of polycyclic aromatic hydrocarbons in a certified marine sediment, Anal. Chim. Acta 477, 81–91. Poli, D., Caglieri, D., Goldoni, M., Castoldi, A. F., Coccini, T., Roda, E., Vitalone, A., Ceccatelli, S., and Mutti, A. (2009), Single step determination of PCB 126 and 153 in rat tissues by using solid phase microextraction/gas chromatography-mass spectrometry: Comparison with solid phase extraction and liquid/liquid extraction, J. Chromatogr. B 877, 773–783. Polo, M., Casas, V., Llompart, M., Garc
ıa-Jares, C., and Cela, R. (2006), New approach based on solid-phase microextraction to estimate polydimethylsiloxane fiber coating—water distribution coefficients for brominated flame reterdants, J. Chromatogr. A 1124, 121–129.

409

Pragst, F. (2007), Application of solid-phase microextraction in analytical toxicology, Anal. Bioanal. Chem. 388, 1393–1414. Prosen, H. and Zupancic-Kralj, L. (1999), Solid-phase microextraction, Trends Anal. Chem. 18, 272–282. Qin, Z., Bragg, L., Ouyang, G., and Pawliszyn, J. (2008), Comparison of thin-film microextraction and stir bar sorptive extraction for the analysis of polycyclic aromatic hydrocarbons in aqueous samples with controlled agitation conditions, J. Chromatogr. A 1196–1197 89–95. Quintana, J. B. and Rodriguez, I. (2006), Strategies for the microextraction of polar organic contaminants in water samples, Anal. Bioanal. Chem. 384, 1447–1461. Ras, M. R., Marc
e, R. M., and Borrull, F. (2008), Solid-phase microextraction–gas chromatography to determine volatile organic sulfur compounds in the air at sewage treatment plants, Talanta 77, 774–778. Regueiro, J., Becerril, E., Garcia-Jares, C., and Llompart, M. (2009a) , Trace analysis of parabens, triclosan and related chlorophenols in water by headspace solid-phase microextraction with in situ derivatization and gas chromatography-tandem mass spectrometry, J. Chromatogr. A 1216, 4693–4702. Regueiro, J., Llompart, M., Garcia-Jares, C., and Cela, R. (2009b) , Development of a solid-phase microextraction-gas chromatogprahy-tandem mass spectrometry method for the analysis of chlorinated toluenes in environmental waters, J. Chromatogr. A 1216, 2816–2824. Risticevic, S., Chen, Y., Kudlejova, L., Vatinno, R., Baltensperger, B., Stuff, J. R., Hein, D., and Pawliszyn, J. (2010a), Protocol for the development of automated high-throughput SPME-GC methods for the analysis of volatile and semivolatile constituents in wine samples, Nat. Protoc. 5, 162–176. Risticevic, S., Lord, H., Go´recki, T., Arthur, C. L., and Pawliszyn, J. (2010b), Protocol for solid-phase microextraction method development, Nat. Protoc. 5, 122–139. Risticevic, S., Niri, V. H., Vuckovic, D., and Pawliszyn, J. (2009), Recent developments in solid-phase microextraction, Anal. Bioanal. Chem. 393, 781–795. Roberts, D. D., Pollien, P., and Milo, C. (2000), Solid-phase microextraction method development for headspace analysis of volatile flavor compounds, J. Agric. Food Chem. 48, 2430–2437. Salgado-Petinal, C., Garcia-Chao, M., Llompart, M., Garcia-Jares, C., and Cela, R. (2006), Headspace solid-phase microextraction gas chromatography tandem mass spectrometry for the determination of brominated flame retardants in environmental solid samples, Anal. Bioanal. Chem. 385, 637–644. Sanchez-Ortega, A., Sampedro, M. C., Unceta, N., Goicolea, M. A., and Barrio, R. J. (2005), Solid-phase microextraction coupled with high performance liquid chromatography using on-line diode-array and electrochemical detection for the determination of fenitrothion and its main metabolites in environmental water samples, J. Chromatogr. A 1094, 70–76. Sanchez-Prado, L., Risticevic, S., Pawliszyn, J., and Psillakis, E. (2010), Low temperature SPME device: A convenient and effective tool for investigating photodegradation of volatile analytes, J. Photochem. Photobiol., A 206, 227–230.

410

APPLICATION OF SOLID-PHASE MICROEXTRACTION IN DETERMINATION OF ORGANIC COMPOUNDS

Sarrio´n, M. N., Santos, F. J., and Galceran, M. T. (2002), Determination of chlorophenols by solid-phase microextraction and liquid chromatography with electrochemical detection, J. Chromatogr. A 947, 155–165. Scheyer, A., Briand, O., Morville, S., Mirabel, P., and Millet, M. (2007), Analysis of trace levels of pesticides in rainwater by SPME and GC-tandem mass spectrometry after derivatization with PFFBr, Anal. Bioanal. Chem. 387, 359–368. Schurek, J., Portol
es, T., Hajslova, J., Riddellova, K., and Hern
andez, F. (2008), Application of head-space solid-phase microextraction coupled to comprehensive two-dimensional gas chromatography–time-of-flight mass spectrometry for the determination of multiple pesticide residues in tea samples, Anal. Chim. Acta 611, 163–172. Setkova, L., Risticevic, S., and Pawliszyn, J. (2007a), Rapid headspace solid-phase microextraction–gas chromatographic–timeof-flight mass spectrometric method for qualitative profiling of ice wine volatile fraction: I. Method development and optimization, J. Chromatogr. A 1147, 213–223. Setkova, L., Risticevic, S., and Pawliszyn, J. (2007b), Rapid headspace solid-phase microextraction-gas chromatographic–timeof-flight mass spectrometric method for qualitative profiling of ice wine volatile fraction: II: Classification of Canadian and Czech ice wines using statistical evaluation of the data, J. Chromatogr. A 1147, 224–240. Setkova, L., Risticevic, S., Linton, C. M., Ouyang, G., Bragg, L. M., and Pawliszyn, J. (2007c) , Solid-phase microextraction-gas chromatography-time-of-flight mass spectrometry utilized for the evaluation of the new-generation super elastic fiber assemblies, Anal. Chim. Acta 581, 221–231. Shurmer, B. and Pawliszyn, J. (2000), Determination of distribution constants between a liquid polymeric coating and water by a solid-phase microextraction technique with a flow-through standard water system, Anal. Chem. 72, 3660–3664. Simpl
ıcio, A. L. and Boas, L. V. (1999), Validation of solid-phase microextraction method for the determination of organophosphorus pesticides in fruits and fruit juice, J. Chromatogr. A 833, 35–42. Stashenko, E. E. and Mart
ınez, J. R. (2004), Derivatization and solid-phase microextraction, Trends Anal. Chem. 23, 553–561. Tollb€ack, J., Isetun, S., Colmsj€o, A., and Nilsson, U. (2010), Dynamic non-equilibrium SPME combined with GC, PICI, and ion trap MS for determination of organophosphate esters in air, Anal. Bioanal. Chem. 396, 839–844. Tsimeli, K., Triantis, T. M., Dimotikali, D., and Hiskia, A. (2008), Development of a rapid and sensitive method for the simultaneous determination of 1, 2-dibromoethane, 1, 4-dichlorobenzene and naphthalene residues in honey using HS-SPME coupled with GC-MS, Anal. Chim. Acta 617, 64–71. Tsoutsi, C., Konstantinou, I., Hela, D., and Albanis, T. (2006), Screening method for organophosphorus insecticides and their metabolites in olive oil samples based on headspace solid-phase microextraction coupled with gas chromatography, Anal. Chim. Acta 573–574 216–222. Tumbiolo, S., Gal, J.-F., Maria, P.-C., and Zerbinati, O. (2004), Determination of benzene, toluene, ethylbenzene and xylenes in

air by solid phase micro-extraction/gas chromatography/mass spectrometry, Anal. Bioanal. Chem. 380, 824–830. Valor, I., P
erez, M., Cortada, C., Apraiz, D., Molto´, J. C., and Font, G. (2001), SPME of 52 pesticides and polychlorinated biphenyls: Extraction efficiencies of the SPME coatings poly (dimethylsiloxane), polyacrylate, poly(dimethylsiloxane)-divinylbenzene, Carboxen-poly(dimethylsiloxane), and Carbowaxdivinylbenzene, J. Separ. Sci. 24, 39–48. Vas, G. and V
ekey, K. (2004), Solid-phase microextraction: A powerful sample preparation tool prior to mass spectrometric analysis, J. Mass Spectrom. 39, 233–254. V
azquez, P. P., Mughari, A. R., and Galera, M. M. (2008), Solidphase microextraction (SPME) for the determination of pyrethroids in cucumber and watermelon using liquid chromatography combined with post-column photochemically induced fluorimetry derivatization and fluorescence detection, Anal. Chim. Acta 607, 74–82. Verhoeven, H., Beuerle, T., and Schwab, W. (1997), Solid-phase microextraction: Artefact formation and its avoidance, Chromatographia 46, 63–66. Vichi, S., Pizzale, L., Conte, L. S., Buxaderas, S., and Lo´pez-Tamames, E. (2005), Simultaneous determination of volatile and semi-volatile aromatic hydrocarbons in virgin olive oil by headspace solid-phase microextraction coupled to gas chromatography/mass spectrometry, J. Chromatogr. A 1090, 146–154. Vuckovic, D., Cudjoe, E., Hein, D., and Pawliszyn, J. (2008), Automation of solid-phase microextraction in high-throughput format and applications to drug analysis, Anal. Chem. 80, 6870– 6880. Vuckovic, D., Shirey, B., Chen, Y., Sidisky, L., Aurand, C., Stenerson, K., and Pawliszyn, J., (2009), In vitro evaluation of new biocompatible coatings for solid-phase microextraction: Implications for drug analysis and in vivo sampling applications, Anal. Chim. Acta 638, 175–185. Wang, A., Fang, F., and Pawliszyn, J. (2005a), Sampling and determination of volatile organic compounds with needle trap devices, J. Chromatogr. A 1072, 127–135. Wang, Y., O’Reilly, J., Chen, Y., and Pawliszyn, J. (2005b), Equilibrium in-fiber standardisation technique for solid-phase microextraction, J. Chromatogr. A 1072, 13–17. Wang, J., Tuduri, L., Mercury, M., Millet, M., Briand, O., and Montury, M. (2009), Sampling atmospheric pesticides with SPME: Laboratory developments and field study, Environ. Pollut. 157, 365–370. Wardencki, W., Curylo, J., and Namies´nik, J. (2007), Trends in solventless sample preparation techniques for environmental analysis, J. Biochem. Biophys. Meth. 70, 275–288. Zambonin, C. G., Quinto, M., De Vietro, N., and Palmisano, F. (2004), Solid-phase microextraction-gas chromatography mass spectrometry: A fast and simple screening method for the assessment of organophosphorus pesticides residues in wine and fruit juices, Food Chem. 86, 269–274. Zhang, Z. and Pawliszyn, J. (1995), Quantitative extraction using an internally cooled solid phase microextraction device, Anal. Chem. 67, 34–43.

REFERENCES

Zhang, X., Cudjoe, E., Vuckovic, D., and Pawliszyn, J. (2009a), Direct monitoring of ochratoxin A in cheese with solid-phase microextraction coupled to liquid chromatography-tandem mass spectrometry, J. Chromatogr. A 1216, 7505–7509. Zhang, X., Cai, J., Oakes, K. D., Breton, F., Servos, M. R., and Pawliszyn, J. (2009b) , Development of the space-resolved solidphase microextraction technique and its application to biological matrices, Anal. Chem. 81, 7349–7354. Zhao, W., Ouyang, G., and Pawliszyn, J. (2007), Preparation and application of in-fiber internal standardization solid-phase microextraction, Analyst 132, 256–261. Zhou, S. N., Oakes, K. D., Servos, M. R., and Pawliszyn, J. (2008a) , Application of solid-phase microextraction for in vivo laboratory and field sampling of pharmaceuticals in fish, Environ. Sci. Technol. 42, 6073–6079.

411

Zhou, S. N., Zhao, W., and Pawliszyn, J. (2008b) , Kinetic calibration using dominant pre-equilibrium desorption for on-site and in vivo sampling by solid-phase microextraction, Anal. Chem. 80, 481–490. Zhou, S. N., Ouyang, G., and Pawliszyn, J. (2008c) , Comparison of microdialysis with solid-phase microextraction for in vitro and in vivo studies, J. Chromatogr. A 1196–1197 46–56. Zhou, S. N., Zhang, X., Ouyang, G., Es-haghi, A., and Pawliszyn, J. (2007), On-fiber standardization technique for solid-coated solid-phase microextraction, Anal. Chem. 79, 1221–1230. Zhou, X., Li, X., and Zeng, Z. (2006), Solid-phase microextraction coupled with capillary electrophoresis for the determination of propranolol enantiomers in urine using a sol-gel derived calix[4] arene fiber, J. Chromatogr. A 1104, 359–365.

16 APPLICATION OF BIOSENSORS FOR ENVIRONMENTAL ANALYSIS MARINELLA FARRE´, SANDRA PE´REZ, LINA KANTIANI, 16.1. Introduction 16.2. Classes and Fundamentals 16.2.1. Electrochemical Transduction 16.2.2. Optical Transducers 16.2.3. Mass Sensitive Sensors 16.2.4. Thermometric Sensors 16.3. Biosensors for Environmental Monitoring 16.3.1. Enzyme Biosensors 16.3.2. Immunosensors 16.3.3. Nucleic Acid Biosensors 16.3.4. Nuclear Receptors 16.3.5. Whole-Cell Biosensors 16.4. Autonomous Biosensor Wireless Networks 16.5. Future Trends and Conclusions

16.1. INTRODUCTION The increasing number of potentially harmful pollutants in the environment calls for fast and cost-effective analytical techniques to be used in extensive monitoring programs. Stricter regulations and a greater public awareness of environmental issues bring requirements to monitor an increasingly wide range of analytes in the environment, and to do so with greater frequency and accuracy. Meanwhile, operators are searching for ways to contain the costs of increasingly complex monitoring regimes. In response to these issues, environmental analysts have sought improvements in laboratory-based analytical methods as well as portable solutions that allow sampling and analysis to be undertaken reliably on site. A biosensor is defined by IUPAC as a self-contained integrated device that is capable of providing specific

AND

DAMIA` BARCELO´ quantitative or semiquantitative analytical information using a biological recognition element (biochemical receptor), which is retained in direct spatial contact with a transduction element. The main advantages offered by biosensors over conventional analytical techniques are the possibility of miniaturization, portability, the ability to measure pollutants in complex matrices with minimal sample preparation, online and at site measuring and recently the possibility of remote sensing. After many years of development, biosensors have begun to move out of the laboratory and into commercial applications. Despite the slow and limited commercialization, which could be attributed to mass production costs and some key technical barriers, the combination of advances in biotechnology, nanotechnology, and information processing, biosensors promise to open the door to many exciting new environmental monitoring solutions. Environmental biosensors can be used for the analysis of target compounds or to assess whole biological effects produced by complex mixtures of pollutants in receptor environments. In addition, biosensors appear well suited to complement standard analytical methods for a number of environmental monitoring applications. The implementation of safety programs calls for combined environmental analysis consisting of two parts: (1) screening methods based on high-throughput analysis and capable of continuous monitoring online at site with low costs and (2) analysis of positive samples with confirmatory analytical techniques. At present, biomonitoring is an essential tool for the complete implementation of EU directives, such as the Water Framework Directive (WFD) and the Marine Strategy Framework directive. Nevertheless, many of these new

Biophysico-Chemical Processes of Anthropogenic Organic Compounds in Environmental Systems. Edited by Baoshan Xing, Nicola Senesi, and Pan Ming Huang. Ó 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

413

414

APPLICATION OF BIOSENSORS FOR ENVIRONMENTAL ANALYSIS

technologies are still being developed and rely on combined efforts from diverse scientific fields. Several reviews on biosensors and biological techniques for environmental analysis have been published more recently (Farr
e et al. 2005; Farr
e and Barcelo´, 2003; Kurosawa et al. 2006; Gonz
alez et al. 2007; Rodriguez-Mozaz et al. 2006). This chapter aims to present the principles, advantages, and limitations of biosensor technology for environmental analysis.

16.2. CLASSES AND FUNDAMENTALS

environmental analysis of organic pollution are enzymes, antibodies, DNA, and whole cells. Figure 16.1 summarizes main classes of biosensors applied in environmental analysis. On the other hand, there are four basic groups of transduction elements: electrochemical, optical, mass-sensitive, and thermal sensors; electrochemical transducers are the biosensors mostly described in the literature. The following sections discuss the main classes of biosensors according to the transduction element. 16.2.1. Electrochemical Transduction

Biosensors are composed of two main parts: the transduction element and the biological receptor. They can be classified according to the bioreceptor elements involved in the recognition and according to the physicochemical transduction elements. The main classes of bioreceptors applied in

Initially biosensors applied to environmental analysis were based on electrochemical configurations. The best developed class of transducers utilize various electrochemical responses to measure changes in the electrical properties of the biological recognition element. In amperometric biosensors

BIOSENSOR

BIOLOGICAL RECEPTORS TRANSDUCERS

Enzymes

Immunoreagents Antibodies Antigens

Electrochemical: Voltammetry Potentiometry Amperometry

Electrochemical: Voltammetry, potentiometry, amperometry Optical: Evanescent wave SPR Piezoelectric

Nucleic acids

Electrochemical: Voltammetry, potentiometry, amperometry Optical: Evanescent wave SPR

Whole cells

Electrochemical: Voltammetry, potentiometry, amperometry Optical: Evanescent wave SPR

Figure 16.1. Summary of the main biosensors applied in environmental analysis.

CLASSES AND FUNDAMENTALS

these changes describe the movement of electrons produced in a redox reaction (electrical current), whereas potentiometric biosensors respond to changes in the distribution of charges, causing an electrical potential (voltage) to be produced. Conductance biosensors respond to changes in the ionic conditions (conductivity). Such effects may be stimulated electrically or may result in a spontaneous interaction at the zero-current condition. Voltammetric, amperometric, potentiometric, potentiometric solid electrolyte gas sensor, and chemically sensitized field effect transistor (CHEMFET) are subgroups that may be distinguished. The principal operation of amperometric biosensors is defined by a constant potential applied between a working and a reference electrode. The imposed potential promotes a redox reaction, which produces a current. The magnitude of this current is proportional to the concentration of electroactive species present in solution. Oxidase enzymes have been the most frequently investigated and applied biosensors. A number of amperometric biosensors are based on the monitoring of oxygen consumption, or hydrogen peroxide generation. Both are electrochemically active; oxygen can be electrochemically reduced, and hydrogen peroxide can be oxidized. The current generated is proportional to the concentration of the enzyme substrate present in a sample. The use of mediators should permit replacement of oxygen as an electron acceptor and operation at a much lower potential, reducing the effects of other electrochemically active species found in complex matrices. Potentiometric biosensors are based on monitoring of the potential produced at a working electrode, with respect to a reference electrode. The development of biosensors based on electrochemical transduction is rising again, due to the advances that can be offered by nanotechnology, which permit the technology to overcome some of the major limitations encountered in the past. One key step in the development of biosensors, not just under electrochemical configurations, is the immobilization of the biological component at the transducer surface. The immobilization requires both stabilization of the biomaterial and proximity and communication between the biomaterial and the transducer. The immobilization methods generally employed are physical adsorption at a solid surface, crosslinking between molecules, covalent binding to a surface, and entrapment within a membrane, surfactant matrix, polymer, or microcapsule. In addition, sol–gel entrapment, Langmuir–Blodgett (LB) deposition, electropolymerization, self-assembled biomembranes, and bulk modification have been widely used more recently. In this sense, a great effort of development is underway out to obtain more robust and more sensitive electrochemical biosensors. More recent research has led to powerful nanomaterialbased electrical biosensing devices and examines future

415

prospects and challenges. New nanoparticle-based signal amplification and coding strategies for bioaffinity assays are in use, along with carbon nanotube molecular wires for achieving efficient electrical communication with redox enzyme and nanowire-based label-free DNA sensors. 16.2.2. Optical Transducers Optical transducers are based on various technologies of optical phenomena, including adsorption, fluorescence, phosphorescence, polarization, rotation, and interference, or nonlinear phenomena, such as second-harmonic generation. The choice of a particular optical method depends on the nature of the application and desired sensitivities. In practice, fiberoptics can be coupled with all optical techniques, thus increasing their versatility. The optical biosensor formats may involve direct detection of the analyte of interest or indirect detection through optically labeled probes. 16.2.2.1. Techniques Based on Reflectometry. When a white incident light passes the interface between different refractive indices, which will be reflected in part these reflected beams superimpose and build a characteristic interference spectrum. The binding of biological receptor such as an antibody to the surface changes the thickness of the toggling layer, which results in a change in the reflectance spectrum. Thus, the interaction process between the bioreceptor and the analyte can be detected as time-resolved (Proll et al. 2004). 16.2.2.2. Other Techniques. Fluorescence/luminescence occurs when a valence electron is excited from its ground state to an excited singlet state. The excitation is produced by the absorption of light of sufficient energy (Lazcka et al. 2007). The common principle of luminescence immunosensors is that an indicator or chemical reagent placed inside or on an immmunoreactor is used as a mediator to produce an observable optical signal. Typically, conventional techniques, such as spectrometers, are employed to measure changes in the optical signal. This approach has been reviewed by Schobel et al. (2000). Interferometry has been also exploited for biosensor development. When an immunoreaction takes place on the waveguide surface, it produces a change in the refractiveindex profile within the evanescent field volume; correspondingly, the effective refractive index of the waveguide system is changed. In the Mach–Zehnder interferometry, an optical waveguide is split into two arms and after a certain distance, they are recombined again. The sensor arm will be exposed to a variation of the refractive index in response to a biorecognition reaction such as an immunoreaction in the sensor channel. During this process, light traveling in the sensing arm will experience a phase shift in comparison with guided light in the reference arm (Prieto et al. 2003).

416

APPLICATION OF BIOSENSORS FOR ENVIRONMENTAL ANALYSIS

Total internal reflection fluorescence (TIRF) has been used with planar and fiber-optic waveguides as signal transducers in a number of reported biosensors. In these transducers, light is propagated down a waveguide that generates an electromagnetic wave (evanescent wave) at the surface of the optically denser medium of the waveguide and the adjacent less optically dense medium. The amplitude of the standing wave decreases exponentially with distance into the lowerrefractive-index material. The fluorescence of a fluorophore excited within the evanescent field can be collected. At low angles, total internal reflection results when light propagating within a dense medium (e.g., quartz) reaches an interface with a less dense medium (e.g., aqueous solution). Although the light is fully reflected, an evanescent field is generated that extends beyond the interface and into the aqueous solution. Typically, the penetration depth (or thickness of the evanescent field) is in the range of half the wavelength of the light. The evanescent field provides the surface selectivity of TIRF. Only fluorophores adsorbed, adhered, or bound to the surface will be excited and, therefore, will fluoresce. Conversely, fluorophores in bulk solution will not be excited. Therefore, if the surface is made biologically active so that one may “trap” fluorescently labeled compounds of interest, one can detect analytes within complex sample solutions. Because the excitation light is totally reflected away from the detection, one can easily

discriminate the fluorescence signal from the excitation light and achieve high sensitivities and low detection limits. Total internal reflection fluorescence systems provide measurement of real-time kinetics of a bioanalyte’s binding to a surface-immobilized sensor molecule. The TIRF technique is fast, non-destructive, sensitive, and versatile and that is well suited for monitoring biomolecular interactions. It allows monitoring of conformational changes, orientation changes, and lateral mobility of biomolecules. Although each of these methods has its individual strengths and weaknesses, a strong case has been made that optical sensors, in particular those based on evanescent electromagnetic fields such as propagating surface plasmon polaritons (SPPs) at planar gold surfaces (Fig. 16.2), are rapidly becoming the methods of choice in many affinity biosensing applications. The SPP method or more commonly, surface plasmon resonance (SPR) spectroscopy, has been widely used to monitor a broad range of analyte-surface binding interactions. The sensing mechanism of SPR spectroscopy is based on the measurement of small changes in refractive index that occur in response to analyte binding at or near the surface of a noble metal (Au, Ag, Cu) thin film. Biosensors based on SPR spectroscopy possess many desirable characteristics including the following: (1) a refractive index sensitivity on the order of 1 part in 105
106 corresponding to an areal mass sensitivity of 10
1 pg/mm2

Buffer T

SPR angles

T

Light source

CCD Polarizer

Figure 16.2. Scheme of surface plasmon polaritons (SPPs) at planar gold surfaces, surface plasmon resonance (SPR) spectroscopy. SPR detects changes in the refractive index in the immediate vicinity of the surface layer of a sensor chip. SPR is observed as a sharp shadow in the reflected light from the surface at an angle that is dependent on the mass of material at the surface. The SPR angle shifts (from T1 to T2) when biomolecules bind to the surface and change the mass of the surface layer. This change in resonant angle can be monitored noninvasively in real time as a plot of resonance signal (proportional to mass change) versus time.

CLASSES AND FUNDAMENTALS

noble metal nanoparticles. Noble metal nanoparticles exhibit a strong UV–visible absorption band that is not present in the spectrum of the bulk metal. This absorption band results when the incident photon frequency is resonant with the collective oscillation of the conduction electrons and is known as the localized surface plasmon resonance (LSPR) (Fig. 16.3). Local surface plasman resonance excitation results in wavelength-selective absorption with extremely large molar extinction coefficients of approximately 3  1011 M
1/cm (Jensen et al. 2000) resonant Rayleigh scattering (Schultz et al. 2000) with an efficiency equivalent to that of 106 fluorophors (Yguerabide and Yguerabide 1998) and the enhanced local electromagnetic fields near the surface of the nanoparticle, which are responsible for the intense signals observed in all surface enhanced spectroscopies. It is well established that the peak extinction wavelength lmax, of the LSPR spectrum is dependent on the size, shape, and interparticle spacing of the nanoparticle as well as its dielectric properties and those of the local environment. Consequently, there are at least four different nanoparticle-based sensing mechanisms that enable the transduction of macromolecular or chemical binding events into optical signals, based on changes in the LSPR extinction or scattering intensity, shifts in LSPR lmax, or both: (1) resonant Rayleigh scattering from nanoparticle labels in a manner analogous to fluorescent dye labels, (2) nanoparticle aggregation, (3) charge transfer interactions at nanoparticle surfaces, and (4) local refractive index changes.

(Hall 2001; Jung et al. 1998), (2) multiple instrumental modes of detection (angle shift, wavelength shift, and imaging) (Brockman et al. 2000), (3) real-time detection on the 10
1–103 s timescale for measurement of binding kinetics (Jung and Campbell 2000; Knoll et al. 1997), and (4) lateral spatial resolution on the order of 10 mm enabling multiplexing and miniaturization, especially using the SPR imaging mode of detection (Brockman et al. 2000). Although SPR spectroscopy is a totally nonselective sensor platform, a high degree of analyte selectivity can be conferred using the specificity of surface-attached ligands and passivation of the sensor surface to nonspecific binding (Brockman et al. 2000). In addition, it is label-free (Haake et al. 2000), capable of probing complex mixtures, such as environmental samples, without prior purification (Brockman et al. 2000; Haake et al. 2000), and benefits from the availability of commercial instrumentation with advanced microfluidic sample handling. The development of large-scale biosensor arrays composed of highly miniaturized signal transducer elements that enable real-time, parallel monitoring of multiple species is an important driving force in biosensor research. This is particularly significant in high-throughput screening applications where many thousands of ligand– receptor or protein–protein interactions must be rapidly examined. More recently, several research groups have begun to explore alternative strategies for the development of optical biosensors based on the extraordinary optical properties of

I II 0.37

Extinction 0.36 0.35 0.34 540

417

580

580

620

Wavelength (nm)

Figure 16.3. Scheme of localized surface plasmon resonance. For example, the reflectance spectra of the Ag nanostructured film indicate that the shift in the LSPR wavelength follows a linear dependence on the refractive index of the surrounding medium. Silver nanostructured films functionalized with biological receptors can sensitively detect binding events on silver nanoparticles.

418

APPLICATION OF BIOSENSORS FOR ENVIRONMENTAL ANALYSIS

It has been demonstrated that nanoscale biosensors can be realized through shifts in the LSPR lmax of triangular silver nanoparticles (Haes and Van Duyne 2002). These wavelength shifts are caused by adsorbate-induced local refractive index changes in competition with chargetransfer interactions at the nanoparticle surface. Triangular silver nanoparticles have been shown to be unexpectedly sensitive to nanoparticle size, shape, and local dielectric environment (Anker et al. 2008). Optical waveguide light mode spectroscopy (OWLS) is a sensing technique using an evanescent field for the in situ and label-free study of surface processes at molecular levels. It is based on the precise measurement of the resonance angle of linearly polarized laser light, diffracted by a grating and incoupled into a thin waveguide layer. The incoupling is a resonance phenomenon that occurs at a defined angle of incidence that depends on the refractive index of the medium covering the surface of the waveguide. In the waveguide layer, light is guided by total internal reflection to the edges where it is detected by photodiodes. By varying the angle of incidence of the light, one obtains the spectrum from which the effective refractive indexes are calculated for both electric and magnetic field waves (Luppa et al. 2001). Optical biosensors offer advantages different from those of electrochemical biosensors, such as low limits of detection and robustness and, are, in general, suitable for screening a large number of samples simultaneously; however, a main limitation is that they cannot be easily miniaturized. 16.2.3. Mass Sensitive Sensors Measurements of small changes in mass is a transduction form that has been used for biosensor development. Piezoelectric devices and surface acoustic wave devices can be grouped under this category. Although this principle has been less frequently reported for environmental applications, due to their capabilities such as the possibility of miniaturizartion and the high sensitivity and specificity achieved when coupled to the proper bioreceptor, this is one of the most promising approaches. The vibration of piezoelectric crystals produces an oscillating electric field in which the resonant frequency of the crystal depends on its chemical nature, size, shape, and mass. These crystals can be made to vibrate at a specific frequency of oscillation that is dependent on the electric frequency. The frequency of oscillation is therefore dependent on the electrical frequency applied to the crystal as well as the crystal’s mass. When the mass increases, due to the binding of analytes, the oscillation frequency of the crystal changes, and this change can be measured. The general equation of crystal microbalances can be summarized as follows when the change in mass (Dm) is very small compared to the total

mass of the crystal Df ¼ Cf 2

Dm A

where f is the vibration frequency of the crystal in the circuit, A is the area of the electrode, and C is a constant determined in part by the crystal material and thickness. Piezoelectric crystals, sometimes referred to as quartz crystal microbalances (QCMs), are typically made of quartz and operate at frequencies between 1 and 10 MHz. These devices can operate in liquids with a frequency determination limit of 0.1 Hz; the detection limit of mass bound to the electrode surface is about 10
10–10
11 g. Acoustic wave devices, made of piezoelectric materials, are the most common sensors, which bend when a voltage is applied to the crystal. Acoustic wave sensors are operated by applying an oscillating voltage at the resonant frequency of the crystal, and measuring the change in resonant frequency when the target analyte interacts with the sensing surface. Limitations for this transduction method involve format and calibration requirements, which are time-consuming. 16.2.4. Thermometric Sensors Thermometric sensors are based on the measurement of the heat effects of a specific chemical reaction or adsorption that involves the analyte. In this group, heat effects may be measured in various ways, for example, in the catalytic sensors the heat of a combustion reaction, or an enzymatic reaction is measured by use of a thermistor. Calorimetric biosensors detect variations of heat during a biological reaction.

16.3. BIOSENSORS FOR ENVIRONMENTAL MONITORING The following sections present and discuss the development of biosensors for environmental applications classified according to the main biological recognition elements used in biosensors for the analysis of target compounds or the assessment of whole biological effects. 16.3.1. Enzyme Biosensors The first biosensor, described in the literature by Clark and Lyons (1962), was based on the use of glucose oxidase with electrochemical detection. Also, the first applications developed for environmental analysis were based on the use of enzymes. Many examples of applications have been reported since 1985, especially using oxidoreductases (as tyrosinase, peroxidase, and lactase) (Kulys et al. 2006) and hydrolases (choline esterases) (Andreescu and Marty 2006). Most of the transduction elements associated with enzyme-based biosensors are electrochemical: amperometric

BIOSENSORS FOR ENVIRONMENTAL MONITORING

and potentiometric. Some examples of environmental applications of electrochemical biosensors are given in Table 16.1. Neurotoxin detection has been accomplished using acetylcholinesterase (Wang et al. 2007), acetylcholine receptors, and butylcholinesterase (BChE) (Ng and Ilag ). The combined inhibition effects in mixtures of organophosphates and carbamates [e.g., paraoxon/carbaryl, diisopropylfluorophosphate (DFP)/carbaryl, paraoxon/DFP/carbaryl] were studied by Simonian et al. (2001). Mutual interactions of various neurotoxins did lead to competition for acetylcholinesterase (AChE) binding sites, and the overall inhibition effects were not additive but dependent on the types of chemicals present. Neuropathy target esterase is the target protein for neuropathic organophosphorous compounds that produce organophosphorous (compound)-induced delayed neurotoxicity (OPIDN). Inhibition and/or aging of brain neuropathy target esterase within hours of exposure predicts the potential for development of OPIDN in susceptible animal models. Lymphocyte NTE has also found limited use as a biomarker of human exposure to neuropathic organophosphorous compounds. Makhaeva et al. (2003) developed a highly sensitive biosensor for neuropathy target esterase activity using a tyrosinase carbon paste electrode for amperometric detection of phenol produced by hydrolysis of the substrate, phenylvalerate. Several examples of enzyme biosensors have been reported using optical transduction (Choi et al. 2001; Doong and Tsai 2001). More recently, optodes, such as fiberoptic biosensors, have generated interest because they provide some advantages, such as no direct electric connection, ease of miniaturization, possibility of remote sensing, and in situ monitoring, through different examples that have been reported such as a sol-gel acetyl cholinesterase fiberoptic biosensor for organophosphorous neurotoxins (Choi et al. 2001). In order to improve the storage stability of enzyme-based biosensors, different immobilizations and electrodes have been assayed, such as carbon paste electrodes (CPEs), solid graphite electrodes (Choi et al. 2008), and surface-modified electrodes (Nistor et al. 1999). Oxidases have been widely used in amperometric biosensors. A sensitive flow enzymatic biosensor for the detection of 2,4-D, atrazine, and ziram has been reported by Kim et al. (2008). Screen-printed electrodes (SPEs) containing immobilized acetylcholinesterase (AChE) were used for the electrochemical determination of organophosphorous and carbamate pesticides (Dutta et al. 2008). The intensely increased interest in nanomaterials is driven by their many desirable properties. In particular, the ability to tailor the size and structure and hence the properties of nanomaterials offers excellent prospects for designing novel sensing systems and enhancing the performance of the bioanalytical assay.

419

An important challenge in amperometric enzyme electrodes is the establishment of satisfactory electrical communication between the active site of the enzyme and the electrode surface. The redox center of most oxidoreductases is electrically insulated by a protein shell. Because of this shell, the enzyme cannot be oxidized or reduced at an electrode at any potential. In general, there is a need for cosubstrates or mediators to achieve efficient transduction of the biorecognition event. Aligned carbon nanotubes (CNTs), prepared by self-assembly, can act as molecular wires to allow electrical communication between the underlying electrode and redox proteins (covalently attached to the ends of the SWNT) (Yu et al. 2003). The deposition of platinum nanoparticles onto CNTs has led to further improvements in the detection of the enzymatically liberated peroxide species (Hrapovic et al. 2004). In addition to CNT films, it is possible to use CNT-based inks (Wang et al. 2005). The excellent electrocatalytic properties of metal nanoparticles (compared to bulk metal electrodes) can also benefit amperometric enzyme electrodes. For example, You et al. (2004) dispersed iridium nanoparticles in graphite-like carbon for improved amperometric biosensing of glutamate. Another promising and controllable route for preparing conductingpolymer nanowire enzyme sensors involves electrodeposition within the channel between electrodes (Ramanathan et al. 2004). The catalytic properties of metal nanoparticles have also facilitated the electrical contact of redox centers of proteins with electrode surfaces. For example, gold nanoparticles were shown to be extremely useful as electron relays for the alignment of glucose oxidase on conducting supports and wiring its redox center (Xiao et al. 2003). The development of electrochemical DNA and protein sensors on SPEs, based on the catalytic activity of hydrazine, has been described by Shiddiky et al. (2008). A wide range of enzyme electrodes based on dehydrogenase or oxidase enzymes rely on amperometric monitoring of the liberated NADH or hydrogen peroxide products. The anodic detection of these species is often hampered by the large overvoltage encountered for their oxidation. It has been proved that enhancement of the redox activity of hydrogen peroxide (Wang et al. 2003a) and NADH (Musameh et al. 2002) using CNT-modified electrodes helps overcome voltage limitations. The ability of CNT to promote electron transfer reactions is due to the presence of edge plane defects at their end caps. Yang et al. (2008) reported a lactate electrochemical biosensor with a titanate nanotube as direct electron transfer promoter. The nanotubes offer the pathway for direct electron transfer between the electrode surface and the active redox centers of LOx, which enables the biosensor to operate at a low working potential and to avoid the influence of the presence of O2 on the amperometric current response.

420 River water Water

Tyrosinase/amperometric detection AChE inhibition/amperometric detection Whole-cell biosensor/amperometric detection

Catechol, p-cresol, phenol, p-chlorophenol, p-methylcatechol Dichlovos Fenitrothion, ethyl-pnitrophenolthiobenzenephosphonate

Water

Parathion hidrolase/amperometric detection DNA/impedance measurement Tyrosinase/amperometric detection Antibodies/amperometric detection

Parathion PAT gene Phenol, o-cresol, p-cresol, m-cresol, catechol, dopamine, ephinephrine Polycyclic aromatic hydrocarbons

Surfactants Bacteria/amperometric detection Toxicity Bacteria/amperometric detection Acetylcholinesterase (AChE) and butyrylcholinesterase (BChE).

River water Water Aqueous solution

Whole-cell biosensor/amperometric detection

Organophosphates

Water Water and wastewater

Water

Water

Heavy metals

Wastewater Water

Aqueous solution Drinking and surface water Water Water

DNA/amperometric detection DNA electrochemical biosensors using four types of DNA (calf thymus ssDNA, calf thymus dsDNA, salmon testis ssDNA salmon testis dsDNA)/amperometric detection DNA/amperometric detection

Genotoxicity Genotoxicity

Spring and surface water

Tyrosinase/amperometric detection Tyrosinase/amperometric detection Antibody/potentiometric Tyrosinase/amperometric detection

Dimethyl-, diethyldithiocarbamates Bisphenol A Bisphenol A Bisphenol A, 17-b-estradiol

Aqueous synthetic samples

AChE, BChE/potentiometric detection

Carbamates

Matrix

Biological Receptor/Transducer

Analite/Effect

TABLE 16.1. Examples of Electrochemical Biosensors Reported for Environmental Analysis

Phenanthrene LOD of 2 mg/L 0.25 mg/L

4.0  10
11, 1.0  10
10, 1.0  10
9, and 5.0  10
9 M for Cu(II), Pb(II), Cd(II), and Fe (III), respectively Phosphate levels in the marine system were 0.04 mg/L 10 mg L
1 1.0  10
11 M 303 mA/mM

70 nM 1.4 mg/L of fenitrothion, 1.6 mg/ L of EPN

1.5  10 – 2.5  10
3 M 0.2 mM 0.02 mM 0.6 mg/L 10
6 M for bisphenol A 9  10
8 M


5

Limit of Detection

Taranova et al. (2004) Farr
e et al. (2001)

Fahnrich et al. (2003)

Sacks et al. (2000) Ma et al. (2008) Tsai and Chiu (2007)

Liu et al. (2007)

Babkina and Ulakhovich (2005)

Lucarelli et al. (2002a) Del Carlo et al. (2008)

Rippeth et al. (1997) Lei et al. (2007)

Kochana et al. (2008)

Perez Pita et al. (1997) Mita et al. (2007) Piao et al. (2008) Notsu et al. (2002)

Ivanov et al. (2000)

Reference

BIOSENSORS FOR ENVIRONMENTAL MONITORING

16.3.2. Immunosensors Antibody–antigen interactions have been exploited in many biosensors for environmental analysis. Electrochemical immunosensors have been used for environmental analysis in amperometric, potentiometric, and conductimetric configurations. Amperometric immunosensors measure the current generated by oxidation or reduction of redox substances at the electrode surface, which is held at an appropriate electrical potential. Wilmer et al. (1997) determined 2,4-dichlorophenoxyacetic acid (2,4-D) in water by an amperometric immunosensor with a limit of detection of 0.1 mg/L, and the same group developed also a flow injection system (Trau et al. 1997). Single-use SPEs have been developed, some examples are the detection of polycyclic aromatic hydrocarbons (PAHs) (Fahnrich et al. 2003) and the quantitative detection of 2,4,6-trichloroanisole (Moore et al. 2003). Other approaches explored the use of recombinant single-chain antibody (scAb) fragments; this approach was used for atrazine determination (Grennan et al. 2003). A nanostructured progesterone immunosensor using a tyrosinase colloidal gold–graphite–teflon biosensor as amperometric transducer has been developed by Pingarron’s group (Carralero et al. 2007). In another application, immunosensors with acoustic transduction have been applied for water analysis (O’Connell and Guilbault 2001; O’Sullivan and Guilbault 1999). Quartz crystal microbalance (QCM) has been used for dioxins detection (Kurosawa et al. 2006). Total internal reflection fluorescence (TIRF) has been used with planar and fiber-optic waveguides as signal transducers in a number of biosensors that have been reported. In these devices, light is propagated down a waveguide, which generates an electromagnetic wave (evanescent wave) at the surface of the optically denser medium of the waveguide and the adjacent less optically dense medium. The amplitude of the standing wave decreases exponentially with distance into the lower-refractive-index material. The fluorescence of a fluorophore excited within the evanescent field can be collected either outside the waveguide, or by coupling the emission frequencies back into the waveguide. The biological sensing element is immobilized on the side rather than the end of the waveguide. On the basis of the evanescent wave transducing principle, atrazine was detected at concentrations around 0.1 mg/L (Schipper et al. 1998) and cyclodiene insecticides in the mg/L range (Brummel et al. 1997). A large number of surface plasmon resonance (SPR) immunoassays have been developed for environmental analysis. A simple and rapid method for dioxins, poly (chlorinated biphenyll)s (PCBs) and atrazine was reported by Shimomura et al. (2001a,b) exhibiting limits of detection ranging from 0.1 to 5 ng/mL. Mauriz et al. (2006a) presented a binding inhibition immunoassay for the analysis of

421

the organophosphate pesticide chlorpyrifos in real water samples with limits of detection ranging from 45 to 64 ng/L. In this case, sensor reusability was ensured through the formation of alkanethiol self-assembled monolayers (SAMs). Using the same principle, Mauriz et al. (2006b) developed single- and multianalyte surface plasmon resonance assays for simultaneous detection of cholinesterase inhibiting pesticides. Farr
e et al. (2007) developed an ultrasensitive immunosensor able to detect a part per trillion (ppt) immunoassay for atrazine determination in natural waters without sample pretreatment. An immunosensor chip utilizing SPR was fabricated for detecting carcinoembryonic antigen based on protein A conjugation (Tang et al. 2006). In another example, a SPR immunosensor was developed for the detection of 2,4-dinitrophenol at ultralow concentrations (ranging between 1 ng/L and 1 mg/L). The sensor strategy was based on a competitive immunoreaction between 2,4-dinitrophenol and a 2,4-dinitrophenol–protein conjugate, namely, DNP–bovine serum albumin conjugate (DNP-BSA). Anti-2,4-dinitrophenol monoclonal antibody was immobilized on a gold thin-film coated SPR sensor chip by means of a chemical coupling process (Aizawa et al. 2007). Immunosensors have been reported for ecotoxicity assessment such as estrogenicity. Biomarkers, such as vitellogenin (Vtg), can be used to show the effects of endocrine substances on whole organisms (Oosterkamp et al. 1997). Vitellogenin is a serum phospholipoglycoprotein precursor to egg yolk, which is produced in high amounts by fish exposed to estrogenic compounds. Thus, the presence of Vtg in male fish is a useful biomarker for identifying estrogenic activity of natural or anthropogenic substances. By using a biosensor to detect vitellogenin in fish, Kroger et al. (2002) determined the estrogenicity of surface waters. Darain et al. (2005) demonstrated a disposable amperometric immunosensor for the rapid detection of carp (Carassius auratus) vitellogenin (Vtg). The sensor was fabricated utilizing screen-printed carbon arrays (SPCAs) containing eight carbon working electrodes and one integrated carbon counterelectrode. The sensor arrays exhibit a linear range of the Vtg concentration from 0.25 to 7.8 ng/mL and the detection limit was determined to be 0.09 ng/mL. Furthermore, the performance of the immunosensor for the determination of Vtg was evaluated by a standard addition method performed in fish serum samples. 16.3.3. Nucleic Acid Biosensors Because of their wide range of physical, chemical, and biological properties, nucleic acids have been incorporated into a wide range of biosensors and bioanalytical assays, many of which possess interesting features for environmental applications.

422

APPLICATION OF BIOSENSORS FOR ENVIRONMENTAL ANALYSIS

Several DNA biosensors and bioassays have been reported for the detection of chemically induced DNA damage. The structure of DNA is very sensitive to the influence of environmental pollutants, such as heavy metals (Drevensek et al. 2005), PCBs (Marrazza et al. 1999a,b), or polyaromatic compounds (PAHs) (Doong et al. 2005). These substances are characterized by a great affinity to DNA, causing mutagenesis and carcinogenesis. Therefore, it is very convenient to use DNA-containing systems, such as DNA-based biosensors (Lazarides et al. 2000; Lucarelli et al. 2002a,b; Chiorcea Paquim et al. 2004; Bakker 2006), to perform genotoxic assays, or for rapid testing of pollutants for mutagenic and carcinogenic activity. Several researchers developed different electrochemical DNA systems for environmental analysis. The determination of genotoxic compounds was measured by their effect on the oxidation signal of the guanine peak of calf thymus DNA immobilized on the surface of disposable electrodes and investigated by chronopotentiometric analysis by Mascinı´s group (Lucarelli et al. 2008). This type of DNA biosensor was able to detect known intercalating compounds, such as aromatic amines, and applicability to river and wastewater samples was also demonstrated (Babkina and Ulakhovich 2005). However, the rapid screening of genotoxic compounds using the molecular interaction between surface-linked DNA and the target pollutants or drugs have been applied in many different configurations using electrochemical as well as optical and mass transducers (Palchetti and Mascini 2008). There is also an ongoing effort in the area of biosensor technology for measuring nucleic acid (NA) hybridization. In this case a sequence-specific probe (usually a short synthetic oligonucleotide) is integrated within a signal transducer. The probe, immobilized onto the transducer surface, acts as the biorecognition molecule and recognizes the target DNA or RNA. Optical, electrochemical, and mass detectors are used mainly for transducing the biorecognition event in an analytical signal. However, such NA sensing applications require high sensitivity, and substantial efforts have been devoted to the amplification of the transduction of the oligonucleotide interaction (Gooding 2002; Lucarelli et al. 2003). Nanoparticle-based amplification schemes have led to improved sensitivity of bioelectronic assays by several orders of magnitude. Wang´s group (Wang et al. 2001b) reported the use of colloidal gold tags for electronic detection of DNA hybridization. This protocol relies on capturing the nanoparticles to the hybridized target, followed by highly sensitive anodic stripping electrochemical measurement of the metal tracer. Electronic DNA hybridization assays have been extended to other metal tracers, including silver (Cai et al. 2002) and iron (Wang et al. 2003b). Coupling of the biorecognition element to surfaces of magnetic beads offers an effective minimization of nonspecific binding. The hybridization of probe-coated magnetic beads with the metal-tagged targets results in three-dimensional network

structures of magnetic beads, crosslinked together through the DNA and gold nanoparticles. The packing of such magnetic bead/DNA/metal label assemblies onto the electrode surface leads to direct contact of the metal label to the surface and enables chronopotentiometric measurements without dissolving the metal tag (Wang et al. 2001a, 2002). Wang et al. (2001a) described a triple-amplification bioassay, coupling the carrier sphere amplifying units (loaded with numerous gold nanoparticle tags) with the “built in” preconcentration of the electrochemical stripping detection and a catalytic enlargement of the multiple gold particle tags. The success of these and other nanoparticlebased amplification strategies depends on maintaining a low background response. Several amplification processes can be used for dramatically enhancing the sensitivity of particlebased bioelectronic assays. The use of inorganic nanocrystal tracers for a multitarget electronic detection of DNA (Kawde and Wang 2004) or proteins has been proposed (Liu et al. 2004). Three encoding nanoparticles (zinc sulfide, cadmium sulfide, and lead sulfide) have thus been used to differentiate the signals of three protein targets in connection with a “sandwich” immunoassay and stripping voltammetry of the corresponding metals. Each binding thus yields a distinct voltammetric peak corresponding to a different antigen. This concept can be scaled up and multiplexed. Nanoparticle-induced changes in the conductivity across a microelectrode gap can also be exploited for highly sensitive and selective electronic detection of DNA hybridization (Park et al. 2002). One-dimensional nanowires can also be used for bridging two closely spaced electrodes for label-free DNA detection. For example, a p-type silicon nanowire functionalized with peptide nucleic acid (PNA) probes has been shown to be extremely useful for real-time label-free conductometric monitoring of the hybridization event (Hahm and Lieber 2004). This relies on the binding of the negatively charged DNA target that leads to an increase in conductance, reflecting the increased surface charge. Carbon nanotubes can also lead to ultrasensitive bioelectronic detection of DNA hybridization (Wang et al. 2003a). For example, carbon nanotubes can be used as carriers for several 1000 enzyme tags and for accumulating the a-naphthol product of the enzymatic reaction. Such a CNT-derived double-step amplification pathway (of both recognition and transduction events) allows the detection of DNA down to the 1.3 zmol level. Feng et al. (2008) reported a gold nanoparticle/polyaniline nanotube membrane on glassy carbon electrode (Au/nanoPAN/GCE) for DNA sensing. The properties of the Au/nanoPAN/GCE, the characteristics of the immobilization and hybridization of DNA, were studied by cyclic voltammetry, differential pulse voltammetry, and electrochemical impedance spectroscopy. The synergistic effect of the two kinds of nanomaterials, nanogold and nanoPAN, could dramatically enhance the sensitivity for the DNA hybridization recognition. A simple and sensitive electrochemical DNA

BIOSENSORS FOR ENVIRONMENTAL MONITORING

biosensor based on in situ DNA amplification with nanosilver as label and horseradish peroxide (HRP) as enhancer has been designed by Fu (2008). The electrochemical detection of harmful algae and other microbial contaminants in coastal waters using handheld biosensors has been reported by LaGier et al. (2007). The highly sensitive and sequence-specific detection of single-stranded oligonucleotides using nonoxidized silicon nanowires (SiNWs) has been demostrated by Zhang et al. (2008). To maximize device sensitivity, the surface of the SiNWs was functionalized with a densely packed organic monolayer via hydrosilylation, subsequently immobilized with PNA capable of recognizing the label-free complementary target DNA. Because of the selective functionalization of the SiNWs, binding competition between the nanowire and the underlying oxide is avoided. Thus, as reported above, the promising performance of NA biosensors for hybridization studies has been demonstrated in several studies. Moreover, NAs have found new roles in biotechnology. In particular, selected NAs have been proposed as highly specific receptors for biological and/or chemical species, such as target proteins, pollutants, or drugs. These selected sequences of oligonucleotides are known to possess the generic term of functional nucleic acid that constitutes aptamer, NAzyme, and aptazyme. Aptamers are singlestranded nucleic acid (DNA or RNA) originating from in vitro selection, which, starting from random-sequence libraries, optimize the nucleic acids for high-affinity binding to a given target. Generally unstructured in solution, aptamers, on association with their target, fold into complex threedimensional shapes in which the target becomes an intrinsic part of the NA structure. The term aptamer derives from the Latin aptus, “to fit,” and emphasizes the relationship between the aptamer and its target. The term NAzymes referes to NA-based catalysts that are not known in nature and are generated only by in vitro selection. The ligandbinding and catalytic features of nucleic acids can be combined to generate allosteric ribozymes or aptazymes. When ligands bind to an aptazyme, conformational changes in the ligand-binding domain are transduced through a change in the catalytic core and a concomitant modulation of catalytic activity. With respect to their application, functional nucleic acids were selected in the past mainly for their use as therapeutic agents; actually for the first time, an aptamer has been approved by the U.S. Food and Drug Administration for the clinical treatment of age-related ocular vascular disease (Ng and Ilag ). In addition to the therapeutic field, aptamers have been then used in some analytical methodologies, such as affinity chromatography, capillary electrophoresis, mass spectrometry, or biosensors. These aptamer-based methods have been employed mainly in the clinical area for the development of diagnostic assays, whereas to date only a

423

limited number of studies have demonstrated the possibility of using aptamers in assays for the analysis of environmental or food samples, even if aptamers specific for anthrax spores, cholera toxin, staphylococcal enterotoxin B, ricin and abrin toxin, and waterborne pathogens (giardia and criptospotidium), as well as 4-chloroaniline (4-CA), 2,4,6-trichloroaniline (TCA) and pentachlorophenol (PCP), and atrazine have been selected in more recent years. 16.3.4. Nuclear Receptors In concrete, the estrogen receptor (ER) has served as the basis for several biosensors to assess the estrogenic potency of complex matrices polluted with endocrine-disrupting compounds (EDCs). Steroid hormones induce different effects in mammalian cells after binding to specific intercellular receptors, which are ligand-dependent transcription factors. Many endocrine disruptors can bind to estrogen receptors as either agonists or antagonists. Thus, the binding ability of the chemicals toward the ER is used to test their potential environmental toxicity. The advantage of receptor assays is that they are quite simple to perform and allow the identification of all endocrine disrupters that act through the estrogen receptor (Oosterkamp et al. 1997). The natural sensing element most commonly used is the human ER (Kroeger et al. 2002). Some of these ER-based biosensors are reported in Table 16.2. By using the surface plasmon resonance (SPR), Usami et al. (2002) developed a simple competitive assay for the evaluation of different chemicals using human recombinant ER. The system measured the binding between estradiol immobilized on the sensor chip and an injected human recombinant ER. In another example using an optical SPR sensor, and the human estrogen a receptor, Hock et al. (2002) performed binding studies with estradiol and xenoestrogens. A rapid enzyme-linked receptor assay (ELRA) using the BIAcore system was developed and validated by Seifert et al. (1999) for detection of estrogens and xenoestrogens. The ELRA method can be used to estimate the estrogenic potential of chemicals, for drinking water control, and in environmental monitoring. Sesay and Cullen (2001) described the detection of hormone mimics by a portable SPR. Butala and Sadana (2003) described the use of a fractal analysis to model the binding and dissociation kinetics between analytes in solution and the ER immobilized on a SPR sensor chip. A novel evanescent biosensor (Wittliff et al. 2008) was developed with laser-based fiberoptics using fluorescent dye–labeled recombinant human estrogen receptor-alpha (rhER-a) and hER-b as probes. A three-tiered approach evaluating various steps in the formation of the estrogen– receptor complex and its subsequent activity was developed for instrument calibration to detect estrogen mimics in

424

APPLICATION OF BIOSENSORS FOR ENVIRONMENTAL ANALYSIS

TABLE 16.2. ER-Based Biosensors Transduction

Analytes

Effect Measured

Reference

Nanomechanical

Diethylstilbestrol, 17-b-estradiol, 17-a-estradiol, 2-OH-estrone, bisphenol A, p,p0 dichlorodiphenyldichloroethylene 17-b-Estradiol

Estrogenicity

Dutta et al. (2007)

Estrogenicity

Andres et al. (2008)

17-b-Testosterone 17-b-Estradiol Estrogens Ligans of nuclear hormone receptors

Androgenicity Estrogenicity Estrogenicity Estrogenicity, andogenicity Estrogenicity Estrogenicity Estrogenicity

Bovee et al. (2008) Wozei et al. (2006) Ramakrishnan et al. (2005) Muddana and Peterson (2003)

Estrogenicity Estrogenicity

Hock et al. (2002) Seifert et al. (1999)

Electrophoretic mobility shift assay Fluorescence Fluorescence SPR Fluorescence SPR Cyclic voltametry SPR SPR SPR

Estrogens 17b-Estradiol Estrogens, progestogens, bisphenol A, 4-nonylphenol, tamoxifen 17b-Estradiol, synthetic estrogens, xenostrogens Estrogens, xenoestrogens

biological samples, water, and soil. Using this approach, binding affinities and activities of certain known estrogen mimics were determined for their use as calibrator molecules. Results indicated that rhERalpha and rhERbeta may be employed as probes to distinguish estrogen mimics with a broad range of affinities. In another example, Dutta et al. (2007) developed a nanomechanical transducer to detect endocrine-disrupting chemicals (EDCs) by combining fluidic sample injection and delivery with bioreceptor protein functionalized microcantilevers. The combination of protein receptors, which include estrogen receptor alpha (ER-a) and estrogen receptor beta (ER-b), as well as monoclonal antibodies (Ab), with MC systems employing modified nanostructured surfaces provides for excellent nanomechanical response sensitivity and the inherent selectivity of biospecific receptor–EDC interactions. Murata et al. (2001) proposed a bioaffinity sensor based on the specific binding of estrogens to the receptor immobilized on a gold disk with cyclic voltammetry detection. The biosensor was applied to the detection of estradiol. Finally, Zhihong et al. (1999) developed a new sandwichtype assay for estrogens using a piezoelectric biosensor. The principle of the detection is that the ER captures the estrogen and then the complex is bound with an estrogen response element (ERE) immobilised on the sensor. 16.3.5. Whole-Cell Biosensors The main classes of whole-cell biosensors are . .

Bacterial biosensors Algal biosensors

. .

Butala and Sadana (2003) Murata et al. (2001) Usami et al. (2002)

Fungal and yeast biosensors Cell-based biosensors

Fabrication of a whole-cell biosensor requires immobilization of microorganisms on transducers. Since whole-cell biosensor response, operational stability, and long-term use are, to some extent, a function of the immobilization strategy used, immobilization technology plays a very important role and the choice of immobilization technique is critical. Microorganisms can be immobilized on transducer or support matrices by chemical or physical methods (Lei et al. 2006). Chemical methods of bacteria immobilization include covalent binding and cross-linking (D’Souza 2001b). Covalent binding methods are based on a covalent bond between functional groups of the microorganism cell wall components such as amine, carboxylic, or sulfydryl and the transducer such as amine, carboxylic, epoxy, or tosyl. To achieve this goal, whole cells are exposed to harmful chemicals and harsh reaction conditions, which may damage the cell membrane and decrease the biological activity. For that reason, this method has not been successful for immobilization of viable microbial cells (Lei et al. 2006). Crosslinking involves bridging between functional groups on the outer membrane of the cells by multifunctional reagents such as glutaraldehyde and cyanuric chloride, to form a network. This method has found a wide acceptance for immobilization of microorganisms. The cells may be crosslinked directly onto the transducer surface or on a removable support membrane, which can then be placed on the transducer. The ability to replace the membrane with the immobilized cells is an advantage of the latter approach. While crosslinking has advantages over covalent binding, the cell viability and/or the cell membrane biomolecules can be affected by the

BIOSENSORS FOR ENVIRONMENTAL MONITORING

crosslinking agents. Thus crosslinking is suitable in constructing microbial biosensors where cell viability is not important and only the intracellular enzymes are involved in the detection (D’Souza 2001a). Adsorption and entrapment are the two most widely used physical methods for microbial immobilization, because they produce a relatively small perturbation of microorganism native structure and function (Cosnier 2007; Lei et al. 2006). Physical adsorption is the simplest one. Typically, a microbial suspension is incubated with the electrode or an immobilization matrix, such as alumina and glass bead (D’Souza 2001a,b; Mart
ınez et al. 2007; Nanduri et al. 2007), followed by rinsing with buffer to remove unabsorbed cells. The microbes are immobilized by to adsorptive interactions such as ionic, polar, or hydrogen bonding and hydrophobic interaction. However, immobilization using adsorption alone generally leads to poor longterm stability because of microbes desorption. The immobilization of microorganisms by entrapment can be achieved by either retention of the cells in close proximity to the transducer surface using dialysis or filter membrane or in chemicobiological polymers and gels such as alginate, carrageenan, agarose, chitosan, collagen, polyacrylamide, polyvinylachohol, poly(ethylene glycol), and polyurethane. (Lei et al. 2006; Odaci et al. 2008). A major disadvantage of entrapment immobilization is the additional diffusion resistance offered by the entrapment material, which will result in lower sensitivity and detection limit. 16.3.5.1. Bacterial Biosensors. One of the most relevant applications of whole cell biosensors has been the determination of the biochemical oxygen demand (BOD) (Chan et al. 2000), and in toxicity, genotoxicity, and estrogenicity biosensors. Several publications have reviewed technology based on microbial biosensors for ecotoxicity assessment (Sorensen et al. 2006). Earlier microbial biosensors used the respiratory and metabolic functions of the microorganisms to detect a substance that was either a substrate or an inhibitor of these processes. Most BOD sensors rely on measurement of the bacterial respiration rate in close proximity to a transducer, commonly the Clark and Lyons (1962) type for measuring dissolved oxygen (Liu and Mattiasson 2002). Some BOD sensors have been marketed by various manufacturers in both biofilm and bioreactor-type configurations. Instrument information about many BOD commercial biosensors was provided by Liu and Mattiason (2002). However, BOD biosensor systems still present a series of limitations that restrict their applications: the lack of standardization and legislation in most countries, complicated maintenance requirements, and insufficient resistance to various toxic compounds. It is possible to eliminate the toxic effects of heavy-metal ions by using a chelating agent that complexes the ions, such as

425

ethylenediaminetetraacetate (EDTA) or sodiumdiethyldithiocarbamate (DDTC) (Tan and Qian 1997). Prevention of contamination by other microbes is also important. In addition to BOD biosensors, a large number of bacterial biosensor have been developed for toxicity assessment. Amperometric biosensors based on genetically engineered Moraxella sp. and Pseudomonas putida with surface-expressed organophosphorus hydrolase (OPH) have been developed (Lei et al. 2005, 2007) for the detection of organophosphorous pesticides. The biosensor consisted of recombinant p-nitrophenol-degrading/oxidizing bacteria P. putida JS444 anchoring and displaying OPH on its cell surface as a biological sensing element and a dissolved oxygen electrode as the transducer. Surface-expressed OPH catalyzed the hydrolysis of fenitrothion and EPN (o-ethyl-op-nitrophenylphenylthiophosphonate) to release 3-methyl-4nitrophenol and p-nitrophenol, respectively, which were oxidized by the enzymatic machinery of P. putida JS444 to carbon dioxide while consuming oxygen, which was measured and correlated to the concentration of organophosphates. Under the optimum operating conditions, the biosensor was able to measure as low as 277 mg/L of fenitrothion. A single-walled (carbon) nanotube (SWNT)-based biosensor for real-time detection of organophosphate has been developed by Liu et al. (2007). Horizontally aligned SWNTs were assembled to desirable electrodes using an AC dielectrophoresis technique. Organophosphorus hydrolase (OPH) immobilized on the SWNTs by nonspecific binding triggers enzymatic hydrolysis of organophosphates (OPs), such as paraoxon, consequently causing a detectable change in the conductance of the SWNTs. The conductance change is found to be correlated with the concentration of organophosphate. Amperometric bacterial biosensors based on monitoring of cell respiration have been reported in different studies for surfactants, assessment. Some examples are the use of surfactant-degrading bacteria (Taranova et al. 2002, 2004) and hydrogen peroxide by coupling immobilized living Acetobacter peroxydans (Rajasekar et al. 2000). The characterization of wastewater toxicity by means of a whole-cell bacterial biosensor using screen-printed electrodes (SPEs) with E. coli was reported by Farr
e et al. (2001). In another study, the same group determined the acute toxicity of wastewaters using P. putida electrodes, in conjunction with chemical analysis (Farr
e and Barcelo´ 2001). Many microbe-catalyzed reactions involve a change in ionic species. Associated with this change is a net change in the conductivity of the reaction solution. Even though the detection of solution conductance is nonspecific, conductance measurements are extremely sensitive. Nevertheless, a few applications have been developed. An example is single-use conductivity and microbial sensor developed by Bhatia et al. (2003) to investigate the metabolic activity of

426

APPLICATION OF BIOSENSORS FOR ENVIRONMENTAL ANALYSIS

Escherichia coli. This sensing system combines physicochemical and biological sensing and greatly increases the ease with which comparative data could be assimilated. The modulation in optical properties such as UV–Visible absorption, bio- and chemi-luminescence, reflectance, and fluorescence brought by the interaction of the biocatalyst with the target analyte is the basis for optical microbial biosensor technology. Optical biosensors offer advantages of compactness, flexibility, resistance to electrical noise, and a small probe size. A systematic review of the methods for immobilizing the different luminescent enzymes and microorganisms (natural and genetically engineered), for use in biosensors, is that by Kratasyuk et al. (2001). An immobilized film of Photobacterium phosphoreum has been used to prepare a sensor based on the determination of the acute toxicity effects of molecules that are difficult or even impossible to measure by traditional analytical chemistry methods (Yin et al. 2005). A biodetector based on a bioluminescence test with the bacterium Vibrio fischeri performed in a liquid continuous flowthrough system has been designed. This system consists of a new flow cell holder and a new case including a top cover to connect the flow cell with the waste and the incubation capillary in a lightproof manner (Stolper et al. 2008). Genotoxic agents, such as mitomycin C, have been also studied by a stable dark variant of P. phosphoreum (A2), fixed in agar-gel membranes immobilized onto the exposed end of optic fibers, linked with a bioluminometer (Sun et al. 2004). A biological electromocrochemical system (MEMS)-based cell chip can be fabricated by immobilizing bioluminescent bacteria in a microfluidic chip. The pattern of the recombinant E. coli strain GC2 was successfully generated by photolithography, utilizing the PVA-SbQ (polyvinylalkyl-styrylpyridinium) polymer, as the immobilization material (Yoo et al. 2007). A hydrogen peroxide detection system was obtained simply immobilizing bioluminescent bacteria, the DK1 strain, which increased bioluminescence in the presence of oxidative damage in the cells (Lee et al. 2003). Pseudomonas fluorescens HK44, which emits light when in contact with naphthalene and its metabolites, has been immobilized into a silica matrix, on glass slides, by the sol-gel technique (Troegl et al. 2005). The test slides could be used for multiple determinations, since the bacteria responded to the inductor at least 8 months after immobilization, and to more than 50 induction cycles. A series of two-stage bioreactor systems, connected by a fiberoptic probe to a luminometer, have been assembled to set up a multichannel system for continuous monitoring and classification of toxicants. Each channel was used for cultivating different recombinant bacterial strains: TV1061 (grpE::luxCDABE), DPD2794 (recA::luxCDABE), and DPD2540 (fabA::luxCDABE), which is induced by protein-, DNA-, and cell-membrane-damaging agents, respectively. The strain GC2 (lac::luxCDABE) is a bacteri-

um expressing bioluminescence constitutively (Gu and Choi 2001). Each channel showed a specific bioluminescent response according to the chemicals contained in wastewater samples, while GC2 showed a general response to cellular toxicity. A portable format of the previously described biosensor consists of three parts: a freezedried biosensing strain within a vial, a small lightproof test chamber, and an optical-fiber connecting the sample chamber to a luminometer. It can be used for field sample analyses and on-site monitoring of various water systems. A different arrangement of genetically engineered luminescent organisms has been used in a bacterial cell chip, making it possible to emit light of different colors from each well, by using quantum dots. Quantum dots have several advantages, such as broad absorption spectrum, narrow emission spectra, and stability. Each well gives off the specific signals (light) according to the recombinant strain exciting the quantum dots. A multistrain bacterial cell array chip has been developed by Lee et al. (2007) as a revealing test for the presence of reactiveoxygen species (superoxide radical and hydrogen peroxide), generated by different chemicals in the sample. The previously described multichannel system has been successfully implemented in the form of computer-based data acquisition. The bioluminescent signatures are delivered from four channels by switching one at a time, while the data are automatically logged to a personal computer (Lee et al. 2006). Improvements of the system have been the manipulation of the dilution rate and the use of thermo-lux fusion strains. The system has been implemented to a drinking water reservoir and river for remote sensing as an early-warning system (Lee et al. 2006). Another example is an electrogenerated precursor that has been developed for green synthesis of highly luminescent aqueous CdTe quantum dots with unique quantum yield and strong electrogenerated luminescence, which can access cellular targets via specific binding and have potential application as biolabels in highly sensitive biosensing and cell imaging (Ge et al. 2008). The studies cited above and additional papers concerning the biosensors based on luminescent and fluorescent bacteria are listed in Table 16.3. Different examples of the application of biosensors based on bioluminescent bacteria for wastewater monitoring have been reported. The construction of whole-cell, genetically modified, bioluminescent biosensors and their immobilization on thin films of poly(vinyl alcohol) cryogels has been carried out by Philp et al. (2003). The biosensor was designed for use in monitoring the toxicity of industrial wastewaters containing phenolic materials. It has been proved that they operate predictably with pure toxicants, within the wastewater treatment plant (Philp et al. 2003). Several recombinant bioluminescent bacteria have been employed to set up a multichannel, continuous, water toxicity monitoring system. The one developed by the research group of Kim was based

BIOSENSORS FOR ENVIRONMENTAL MONITORING

427

TABLE 16.3. Bioluminescence and Fluorescence Bacterial Biosensors Target

Bacteria

Type of luminescence

Transducer

References

Acute toxicity Acute toxicity Genotoxicity Genotoxicity Acute toxicity Acute toxicity Acute toxicity Acute toxicity Acute toxicity Toxicity, genotoxicity Genotoxicity Acute toxicity Genotoxicity Genotoxicity Genotoxicity Genotoxicity

Photobacterium phosphoreum Vibrio fischeri Photobacterium phosphoreum Escherichia coli Pseudomonas fluorescens Vibrio fischeri Escherichia coli Pseudomonas putida Escherichia coli Escherichia coli

Natural Natural Dark variant GC2 10568 Natural HB101 TVA8 DH5a DPD2511, DPD2540, DPD2794, TV1061 HK44 Natural

Bioluminescence Bioluminescence Bioluminescence Luminescence Fluorescence Bioluminescence Luminescence Luminescence Luminescence Luminescence

Yoo et al. (2007) Lee and Gu (2005) Troegl et al. (2005) Gu and Choi (2001) Bhattacharyya et al. (2005) Bhattacharyya et al. (2005) Bhattacharyya et al. (2005) Bhattacharyya et al. (2005) Bhattacharyya et al. (2005) Bhattacharyya et al. (2005)

Luminescence Bioluminescence Luminescence Luminescence Luminescence Luminescence

Lee et al. (2007) Pooley et al. (2004) Premkumar et al. (2002) Rosen et al. (2000) Polyak et al. (2000) Taguchi et al. (2004)

Pseudomonas fluorescens Photobacterium phosphoreum Escherichia coli Escherichia coli Escherichia coli Salmonella typhmurium TA1535

recA’::lux V. fischeri DPD1718 (recA’::lux) TL210 and TL210ctl

on channels, each one hosting a different recombinant bacterial strain, and was composed of two small-bioreactors, to enable a continuous operation, without system interruption due to highly toxic samples (Kim and Gu 2005). The luminescent s