Development of New Binding Phases for Speciation Measurements of Trace Metals with the Diffusive Gradients in Thin Films Technique
Author Li, Weijia
Published 2004
Thesis Type Thesis (PhD Doctorate)
School School of Environmental and Applied Science
DOI https://doi.org/10.25904/1912/2018
Copyright Statement The author owns the copyright in this thesis, unless stated otherwise.
Downloaded from http://hdl.handle.net/10072/367741
Griffith Research Online https://research-repository.griffith.edu.au Development of New Binding Phases for Speciation Measurements of Trace Metals with the Diffusive Gradients in Thin Films Technique
A thesis submitted in fulfilment of the requirements for the Degree of Doctor of Philosophy
By WEIJIA LI
School of Environmental and Applied Sciences Faculty of Environmental Sciences Griffith University Australia
March 2004 CERTIFICATE OF ORIGINALITY
I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at Griffith University or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at Griffith University or elsewhere, is explicitly acknowledged in the thesis.
I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project’s design and conception or in style, presentation and linguistic expression is acknowledged.
(signed)……………………….
i ACKNOWLEDGEMENTS
I wish to take this special opportunity to thank my supervisors, Dr. Huijun Zhao, Dr. Peter Teasdale and Dr. Richard John, for their support, patience and assistance through out the course of my period as a Ph.D. student. Thanks for invaluable ideas and guidance from Dr. Huijun Zhao.
I wish to express my gratitude to School of Environmental and Applied Sciences, Griffith University, Australia for providing me with scholarship to undertake this project.
Many thanks must also be given to head of the school, Clyde Wild, for his support; school secretary, Carmel Wild, for her English corrections of my thesis, and many other staff members in the School of Environmental and Applied Sciences for their help.
I also thank to my research group, especially, Dr. Shangqing Zhang, Mr. Dianlu Jiang, Mr. Calvin Gladman, Miss Kylie Catterall, and Miss Kristy Morris, who have helped me in various ways.
Thanks to staff members in the Chemistry Department, School of Molecular & Microbiological Sciences, University of Queensland for their help and assistance.
I am deeply grateful my mother and father for their support and encouragement during the course of this study.
Finally, I wish to thank my wife, Yali Qu, and my daughter, Mandy Li, for their great understanding, patience and assistance at any time.
ii LIST OF PUBLICATIONS [1-7] [1] Li, W.; Teasdale, P. R.; Zhang, S.; John, R.; Zhao, H. Application of a Poly(4-
styrenesulfonate) Liquid Binding Layer for Measurement of Cu2+ and Cd2+ with
the Diffusive Gradients in Thin-Films Technique, Analytical Chemistry, 2003, 75,
2578-2583.
[2] Li, W.; Zhao, H.; Teasdale, P. R.; John, R.; Zhang, S. Synthesis and
characterisation of a polyacrylamide-polyacrylic acid copolymer hydrogel for
environmental analysis of Cu and Cd, Reactive and Functional Polymers, 2002,
52, 31-41.
[3] Li, W.; Zhao, H.; Teasdale, P. R.; John, R. Preparation and characterization of a
poly(acrylamidoglycolic acid-co-acrylamide) hydrogel for selective binding of
Cu2+ and application to diffusive gradients in thin films measurements, Polymer,
2002, 43, 4803-4809.
[4] Li, W.; Zhao, H.; Teasdale, P. R.; John, R.; Zhang, S. Application of a cellulose
phosphate ion exchange membrane as a binding phase in the diffusive gradients in
thin films technique for measurement of trace metals, Analytica Chimica Acta,
2002, 464, 331-339.
[5] Li, W.; Zhao, H.; Teasdale, P. R.; John, R. Application of new solid membrane
diffusive layer/liquid binding phase DGT technique for environmental speciation,
Environ. Sci. Technol., 2003, Submitted.
[6] Li, W.; Zhao, H.; Teasdale, P. R.; John, R. Development of a new generation DGT
technique using a solid membrane diffusive layer with a liquid binding phase,
Analytica Chimica Acta, 2003, Submitted.
iii [7] Li, W.; Zhao, H.; Teasdale, P. R.; John, R. Evaluation of new binding phases
developed for use in diffusive gradients in thin films technique, Environ. Sci.
Technol., 2003, Submitted.
iv Table of Contents Certification………………………………………………………………………………...i
Acknowledgement………………………………………………………………………….ii
List of Publications………………………………………………………………………..iii
Table of Contents…………………………………………………………………………..v
Nomenclature……………………………………………...……………………………….x
Abstract…………………………………………………………………………………...xii
CHAPTER 1 General Introduction ...... 1 1.1. SIGNIFICANCE OF THIS RESEARCH...... 2 1.2. THE SPECIATION OF TRACE METALS IN NATURAL WATERS ...... 6 1.3. THE NEED FOR SPECIATION MEASUREMENTS OF TRACE METALS .....9 1.3.1. The Free-ion Activity Model ...... 11 1.3.2. The Biotic Ligand Model...... 13 1.4. ISSUES TO CONSIDER WHEN SAMPLING AND MEASURING TRACE METAL SPECIES ...... 16 1.4.1. Sampling Factors ...... 16 1.4.2. Measurement Factors ...... 18 1.5. TECHNIQUES USED FOR IN SITU MEASUREMENT AND SPECIATION OF TRACE METALS ...... 21 1.5.1. Diffusive Gradients in Thin Films (DGT) ...... 23 1.6. OBJECTIVES OF THIS STUDY...... 39 CHAPTER 2 Experimental and Methodology...... 43 2.1. INTRODUCTION ...... 44 2.2. REAGENTS AND SOLUTIONS...... 44 2.2.1. Chemicals and Materials...... 44 2.2.2. Solutions ...... 45 2.3. PROCEDURES ...... 48 2.3.1. Preparation of Diffusive Gel...... 48 2.3.2. Preparation of Chelex 100 Binding Gel...... 49 2.3.3. Characterisation of the Structure and Composition of Binding Hydrogels 50 2.3.4. Assembling and Disassembling the Gel Based DGT Devices ...... 50 2.3.5. Measurement of Diffusion Coefficient in Diffusive Layer ...... 51
v 2.4. INSTRUMENTATION ...... 53 2.4.1. Atomic Absorption Spectroscopy (AAS) ...... 53 2.4.2. Measurement of Metal Concentrations in a Solution Containing PSS ...... 54 2.4.3. Solution pH Measurement ...... 54 2.4.4. Solution Salinity Measurement...... 54 CHAPTER 3 Synthesis and Characterisation of a Poly(acrylamide-co-acrylic acid) Copolymer Hydrogel Based Binding Phase for the Diffusive Gradients in Thin Films (DGT) Technique ...... …………………………………………………………………………55 3.1 INTRODUCTION ...... 56 3.2 EXPERIMENTAL...... 59 3.2.1 Preparation of the Poly(acrylamide-co-acrylic acid) Copolymer Hydrogel 59 3.2.2 Characterisation of the Structure and Composition of the PAM-PAA Hydrogels...... 59 3.2.3 Swelling Properties of the PAM-PAA Hydrogel...... 60 3.2.4 Metal Binding Properties of the PAM-PAA Hydrogel...... 60 3.2.5 Elution and Analysis of the Metal Ions ...... 61 3.2.6 Validation of the PAM-PAA Hydrogel for Use with DGT ...... 62 3.3 RESULTS AND DISCUSSION ...... 62 3.3.1 Preparation of the Poly(acrylamide-co-acrylic acid) Copolymer Hydrogel ...... 62 3.3.2 Composition of the PAM-PAA Copolymer Hydrogel ...... 63 3.3.3 PAM-PAA Hydrogel Swelling Properties ...... 66 3.3.4 Metal Binding Properties of the PAM-PAA Hydrogel...... 69 3.3.5 Application of the PAM-PAA Hydrogel as a Binding Phase with DGT....74 3.4 CONCLUSIONS ...... 76 CHAPTER 4 Preparation and Characterisation of a Poly(acrylamidoglycolic acid- co-acrylamide) Hydrogel as a New DGT Binding Phase for Determination of Trace Metals ...... 78 4.1. INTRODUCTION ...... 79 4.2. EXPERIMENTAL...... 79 4.2.1. Preparation of Poly(acrylamidoglycolic acid-co-acrylamide) Hydrogel....79 4.2.2. Characterisation of the PAAG-PAM Hydrogel ...... 80 4.2.3. Swelling Properties of the PAAG-PAM Hydrogel...... 81
vi 4.2.4. Metal Binding Properties of the PAAG-PAM Hydrogel...... 81 4.2.5. DGT Performance ...... 82 4.2.6. Preparation of Polyacrylamide Hydrogel ...... 83 4.3. RESULTS AND DISCUSSION ...... 83 4.3.1. Structure and Composition of the PAAG-PAM Hydrogel ...... 83 4.3.2. Swelling Properties of the PAAG-PAM Gel ...... 85 4.3.3. Metal Binding Properties of the PAAG-PAM Hydrogel...... 87 4.3.4. Validation of Poly(AAGA-co AAm) as a Binding Phase for DGT Use ....91 4.4. CONCLUSIONS ...... 93 CHAPTER 5 Application of a Cellulose Phosphate Ion Exchange Membrane as a Binding Phase in the Diffusive Gradients in Thin Films Technique...... 94 5.1. INTRODUCTION ...... 95 5.2. EXPERIMENTAL...... 96 5.2.1. Cellulose Phosphate Membrane Pre-treatment...... 96 5.2.2. Preparation of the Polyacrylamide Hydrogel...... 96 5.2.3. Binding of Metal Ions to Cellulose Phosphate Membrane ...... 97 5.2.4. Elution and Analysis of Metal Ions ...... 97 5.2.5. Assembly of DGT Devices ...... 98 5.2.6. DGT Validation Experiments ...... 98 5.2.7. Reuse of Binding Phase ...... 99 5.3. RESULTS AND DISCUSSION ...... 99 5.3.1. Metal Ion Binding Properties...... 99 5.3.2. Elution and Regeneration...... 105 5.3.3. Evaluation for Use as a Binding Phase with DGT...... 106 5.4. CONCLUSIONS ...... 111 CHAPTER 6 Development of a New Generation DGT Device Using a Solid Membrane Diffusive Layer with a Liquid Binding Phase ...... 112 6.1. INTRODUCTION ...... 113 6.2. EXPERIMENTAL...... 114 6.2.1. The DGT Device Using a Solution Binding Phase...... 114 6.2.2. Preparation of the Dialysis Membrane ...... 115 6.2.3. Interaction of Cd2+ and Cu2+ with the Cellulose Dialysis Membrane ...... 115 6.2.4. Purification of Poly(4-styrenesulfonate)...... 116 6.2.5. Determination of Metal-PSS Concentrations ...... 116 6.2.6. Optimisation of PSS Solution Concentration ...... 116
vii 6.2.7. Metal Binding Properties of the Poly(4-styrenesulfonate) Solution...... 117 6.2.8. Determination of Stability Constant ...... 117 6.2.9. Measurement of Metal Diffusion Coefficients in the Dialysis Membrane...... 118 6.2.10. Effect of Stirring Conditions on the DBL Layer ...... 118 6.2.11. Validation of the New DGT Device ...... 119 6.3. RESULTS AND DISCUSSION ...... 119 6.3.1. Dialysis Membrane Diffusive Layer...... 119 6.3.2. Optimization of PSS Solution Concentration ...... 121 6.3.3. Metal Ion Binding Properties of Poly(4-styrenesulfonate)...... 123 6.3.4. Diffusion of Cd2+ and Cu2+ in the Cellulose Dialysis Membrane ...... 130 6.3.5. Effect of Stirring Conditions on the DBL Layer ...... 133 6.3.6. Validation of the PSS/dialysis DGT Device...... 139 6.4. CONCLUSIONS ...... 142 CHAPTER 7 ...... Characterisation of the Dialysis Membrane/PSS DGT Device for Trace Metal Speciation Measurements...... 143 7.1. INTRODUCTION ...... 144 7.2. EXPERIMENTAL...... 147 7.2.1. Measurement of Diffusion Coefficients of EDTA-Metal Complexes...... 147 7.2.2. Measurement of DGT-labile Fractions ...... 147 7.2.3. Theoretical Calculation of Free Cu and Cd Fractions ...... 148 7.2.4. Field Deployments of PSS DGT Devices...... 149 7.2.5. Measurement of PSS DGT-labile and 0.45-filtered Cu and Cd Concentrations ..……………………………………………………………………152 7.3. RESULTS AND DISCUSSION ...... 153 7.3.1. Diffusion of EDTA-Cu and EDTA-Cd in the Dialysis Membrane Diffusive Layer …………………………………………………………………………..153 7.3.2. Measurement of Labile Metal Ions in the Presence of Ligands ...... 155 7.3.3. Field Deployments ...... 166 7.4. CONCLUSIONS ...... 170 CHAPTER 8 Evaluation of the New Binding Phases Developed for Use in the Diffusive Gradients in Thin Films Technique ...... 171 8.1. INTRODUCTION ...... 172 8.2. EXPERIMENTAL...... 172 8.2.1. Diffusion Layer Preparation ...... 172
viii 8.2.2. Binding Phase Preparation...... 172 8.2.3. DGT Measurements in Laboratory ...... 173 8.2.4. DGT Field Deployment ...... 174 8.3. RESULTS AND DISCUSSION ...... 175 8.3.1. Measurement of DGT Labile Metal Ions in the Presence of Ligands ...... 175 8.3.2. Field Deployments ...... 181 8.4. COMPARISON OF IMPORTANT PROPERTIES OF THE NEW BINDING PHASES DEVELOPED IN THIS STUDY...... 190 8.4.1. Assembly of DGT Devices and the Interface between the Binding and Diffusive Layers ...... 191 8.4.2. Swelling Effects ...... 193 8.4.3. Biofouling Effects...... 194 8.4.4. Reusability ...... 195 8.4.5. Elution...... 196 8.4.6. Valid Deployment Conditions and Metal Binding Properties ...... 196 8.5. CONCLUSIONS ...... 198 CHAPTER 9 General Conclusions ...... 200 REFERENCES…………………………………………………………………………..207
ix NOMENCLATURE
A diffusive area AAGA acrylamidoglycolic acid monohydrate AAm acrylamide AAS atomic absorption spectroscopy ASV anodic stripping voltammetry C the concentration in sample solution C' solute concentrations at the interface of membrane and binding solution
Cb solute concentrations in the bulk solution
Cm solute concentrations at the interface of membrane and the DBL
Ce concentrations of ions in the elution
CMF concentration of free metal
CM’ concentration of free metal dissociated from metal complexes
CML concentration of metal complex D diffusion coefficient
Di diffusion coefficient of ion i
Dm diffusion coefficient in dialysis membrane
DMF diffusion coefficient of free metal
DML diffusion coefficient of metal complex DBL diffusive boundary layer DBS dodecylbenzenesulfonic acid, sodium salt DET diffusive equilibration in thin films DGT diffusive gradients in thin films DOC dissolved organic carbon EDTA ethylenediaminetetraacetic acid, disodium salt dihydrate F flux FTIR Fourier transform infrared spectroscopy GL glucose HA humic acid ICP-MS inductively coupled plasma-mass spectrometry IDA iminodiacetic acid ISE ion selective electrode K stability constant
x k-1 metal complex dissociation reaction rate constant M mass measured in binding phase md the weights of the hydrogel disks in dried state ms the weights of the hydrogel disks in the swollen/hydrated state PAAG-PAM poly(acrylamidoglycolic acid-co-acrylamide) hydrogel PAM polyacrylamide gel PAM-PAA poly(acrylamide-co-acrylic acid) PIXE proton induced x-ray emissions PSS poly(4-styrenesulfonate) qw swelling ratio, defined as qw = ms / md Sc schmidt number t time of deployment td time for transport of metal complex TA tannic acid TEMED tetramethylethylenediamine
Vb volume of PSS solution
Ve volume of HNO3 solution for elution
Vs volume of sample solution P81 cellulose phosphate membrane x the distance from the leading edge of the plate
Zi charge of ion i ∆g thickness of the diffusive gel δ thickness of the boundary layer σ fluid density η dynamic viscosity β ratio between free metal concentration and total metal concentration τ time for dissociation of metal complex
xi ABSTRACT
The recently developed technique of diffusive gradients in thin films (DGT) for speciation measurement of analytes in the environment was further developed through the development of series of new binding phases including poly(acrylamide-co-acrylic acid) copolymer hydrogel (PAM-PAA), poly(acrylamidoglycolic acid-co-acrylamide)
(PAAGA-PAM) hydrogel, the Whatman P81 cellulose phosphate ion exchange membrane
(P81) and a liquid binding phase of poly(4-styrenesulfonate) (PSS). A new diffusion layer, cellulose dialysis membrane, was also employed for the liquid binding phase DGT.
PAM-PAA copolymer hydrogel was prepared by the controlled hydrolysis of polyacrylamide (PAM) in an alkaline solution of 10% sodium hydroxide. The capacity of the copolymer hydrogel to bind various metal ions was tested under a range of uptake conditions. Ions such as Cu2+ and Cd2+ were bound more strongly to the copolymer hydrogel than the competing ions such as Na+, K+, Ca2+ and Mg2+. Metals bound to the copolymer hydrogel can be efficiently eluted in 2 M HNO3 solution (>94%). Application of this new binding material to DGT technique was validated in a synthetic lake water
(Windermere, Lake District, UK) with a recovery of 99.0% for Cu2+.
PAAGA-PAM hydrogel was prepared by copolymerising 2-acrylamidoglycolic acid with acrylamide. The metal ion binding properties of the hydrogel were characterised for Cu2+,
Cd2+ and competing ions under various experimental conditions. The hydrogel was shown to bind Cu2+ and Cd2+ strongly under non-competitive binding conditions, with binding capacities of 5.3 and 5.1 µmol cm-2, respectively. The binding capacity of each metal decreased, under competitive binding conditions (with a range of metal ions present at
17.8 µN), to 1.3 and 0.17 µmol cm-2, respectively, indicating a strong selective binding
2+ towards Cu . The metal ions were readily recovered (>94%) by eluting with 2 M HNO3.
xii Finally, the copolymer hydrogel was tested as a binding phase with the DGT technique. A linear mass vs. time relationship was observed for Cu2+ in Windermere water with a recovery of close to 100%.
The use of a commercially available solid ion exchange membrane (P81) as the binding phase in DGT analysis was demonstrated. P81 is a strong cation exchange membrane. Its performance characteristics as a new binding phase in DGT measurement of Cu2+ and
Cd2+ were systematically investigated. Several advantages over the conventional ion exchange resin-embedded hydrogel based binding phases used in DGT were observed.
These include: simple preparation, ease of handling, and reusability. The binding phase preferentially binds to transition metal ions rather than competing ions. Within the optimum pH range (pH 4.0 – 9.0), the maximum non-competitive binding capacities of the membrane for Cu2+ and Cd2+ were 3.22 and 3.07 µmol cm-2, respectively. The suitability of the new membrane–based binding phase for DGT applications was validated experimentally. The results demonstrated excellent agreement with theoretically predicted trends. The reusability of this binding phase was also investigated.
Application of a liquid binding phase and a dialysis membrane diffusive layer were proposed for the first time. The binding phase was a 0.020 M solution of poly(4- styrenesulfonate) (PSS) polyelectrolyte using a specially designed DGT device. The binding properties of Cd2+, Cu2+, and a range of alkali and alkaline earth metal ions to the
PSS solution were characterised. The PSS behaved like a cation exchanger with preferential binding to Cd2+ (6.0 µmole ml-1, log K = 9.0) and Cu2+ (2.5 µmole ml-1, log K
= 8.1) under competitive binding conditions. The DGT devices were successfully validated for Cd2+ and Cu2+ in Windermere water.
The speciation performance of the solid and liquid binding phases developed in this study was investigated in solutions containing ethylenediaminetetraacetic acid disodium salt
xiii (EDTA), humic acid (HA), glucose (GL), dodecylbenzenesulfonic acid (DBS) and tannic acid (TA) with Cu2+ and Cd2+. The ratios of labile metals over total metals were at good agreement with calculated theoretical values using Stability Constants Database. The results indicated that the DGT-labile concentration measured by DGT with these binding phases is essentially free metal ion concentration in the sample.
All newly developed DGT binding phases were successfully applied for environmental speciation. The field deployments were carried out in both freshwater and salt-water test sites.
xiv Chapter 1
Chapter 1 General Introduction
1 Chapter 1
1.1. SIGNIFICANCE OF THIS RESEARCH
This dissertation describes research and development of the diffusive gradients in thin films (DGT) technique for the in situ measurement and speciation of trace metals, particularly Cu and Cd. The speciation of trace metals is important for a number of areas in environmental research and management, including toxicological studies and water quality monitoring. The need to undertake in situ measurements of trace metal in natural waters has also been increasingly recognized over the last decade 1-3. This recognition is the result of a number of observations concerning the limitations of commonly used approaches to trace metal measurement and speciation of natural waters. These observations are described briefly below, and in more detail in the following review of relevant literature. It will become apparent that the DGT technique has potential to meet many of the needs highlighted below and therefore should be thoroughly investigated for the purpose of in situ speciation of trace metals in waters.
(1) It is widely recognised that most waterways have compositions, including that of the trace metal fraction, which vary over characteristic time scales 2. Marine waters change only slowly due to their massive volume. Changes in trace metal concentrations
(usually gradual increases) have been measured in coastal waters and in enclosed seas, due to increased anthropogenic inputs 4-7. Changes happen to trace metal concentrations much more rapidly in estuaries and rivers than in marine waters. In estuaries and rivers trace metal concentrations are influenced by a range of events, both anthropogenic and natural, some of which can occur over a time scale of hours (e.g. tidal processes) 8-12. For these more dynamic waterways, trace metal concentrations need to be measured at frequent and regular intervals, especially when using conventional grab sampling approaches 2, 13, and
2 Chapter 1 particularly after events such as high rainfall or the release of effluent. An alternative approach to monitoring such systems is to use devices, usually deployed in situ, that continually accumulate trace metal analytes over a deployment period, such as the DGT technique 14-16. A recent study has indicated that DGT measurements of trace metal concentrations were significantly and highly correlated with measurements of composite
0.45 µm-filtered estuarine samples over the DGT deployment period 17. Although DGT used in this way does not provide a continual measurement of the trace metal concentration, it does give an average concentration for the entire deployment period. It also requires much less effort and expense than an intensive collection of grab samples 17.
The DGT deployment time can be varied to investigate changes in the trace metal concentration for a dynamic waterway over various time scales. While this aspect of DGT measurements is not investigated specifically in this study the point is made to emphasise the importance of developing techniques, such as DGT, which are capable of making representative measurements of trace metals, even dynamic waterways.
(2) The difficulties of maintaining the integrity of water samples (in which trace metal concentrations are to be measured) after collection and before analysis have also been recognised widely. Improved sample preparation and handling approaches have now minimised contamination and losses of trace metals before analysis 2, 18. The use of quality control procedures have also improved the accuracy of the data obtained from the trace measurements 1. However, while these procedures have improved total measurements (on filtered and unfiltered water), there are still many difficulties in carrying out speciation measurements in samples removed from a waterway. Indeed maintaining the trace metal speciation of a water sample after collection has proven to be very difficult 13, 19, 20. Virtually all approaches used to preserve trace metal samples will lead to a change in the speciation. Therefore the best way to determine trace metal
3 Chapter 1 speciation is to use a number of in situ speciation techniques or to measure parameters that can be used in speciation models. The DGT technique, while it is deployed in situ, accumulates trace metal species that are able to pass through the diffusive layer and bind to the binding phase. In this way DGT is able to selectively measure a range of trace metal forms 21, 22. More importantly the trace metals are accumulated in a form that is stable during transport and storage, and can be measured using sensitive laboratory instrumentation after elution. This study seeks to develop a number of new DGT binding phases that will be capable of measuring different trace metal fractions and of maintaining the speciation between sampling and analysis.
(3) A better understanding has developed recently concerning the interpretation and limitations of measurement techniques used to speciate trace metals. Most techniques have attempted to fractionate or separate the various trace metals species. However, these attempts at fractionation have been confounded by the fact that virtually all trace metals exist in a variety of physico-chemical forms. These forms are often in dynamic equilibrium with one another and they span continuums of both size and reactivity 1, 13.
Therefore most so-called speciation methods, rather than attempting to fractionate perfectly between particular forms of trace metal species, should instead be reproducible with respect to the species that they measure. There has consequently been a recent preference for describing measurements in operational terms rather than in terms of particular species. Through research these operational speciation measurements could be compared with other measurements of important processes, such as biological uptake and toxic effects. Such operational measurements will prove to be almost as useful as if they were accurately and precisely able to determine trace metal speciation in natural waters.
4 Chapter 1
The DGT technique has been reported as being capable of speciation measurements 14, 23
25, but the potential for speciation measurements with DGT has not, as yet, been investigated fully. The nature of the trace metal species that are measured by DGT (i.e. are DGT-labile) have also not been investigated fully either. A complex is labile if the thermodynamic equilibrium of dissociation of the complex is maintained at all distances from the binding phase. These aspects of DGT are investigated in this dissertation through the development of new binding phases that have various functional groups and, therefore, various strengths of interaction with trace metal species (and may therefore have different DGT-labile fractions). This study is the first to investigate, in depth, the development of new binding phases for trace metal speciation of DGT.
The following literature review includes a more detailed description of the current thinking and research on each of these and other topics. The review begins by describing the range of trace metal species present in natural waters and describes important processes that lead to changes in speciation. The importance of speciation measurements is then discussed with respect to toxicological studies and models that describe the interaction between trace metals species and organisms. Various methods used for speciation measurements are described briefly, followed by a detailed description of DGT and of studies that have contributed important information to the development of DGT.
Other important speciation methods are also described. Many of these methods have used complexing functional groups and selective membranes. The field of membrane and separation science, which frequently uses complexing functional groups, is also reviewed with respect to their potential for use with a DGT binding phase. The purpose of this research was to determine whether these approaches could be utilised in the preparation of a range of new DGT binding phases that could be used for speciation measurements of trace metal in natural waters.
5 Chapter 1
1.2. THE SPECIATION OF TRACE METALS IN NATURAL WATERS
It is well known 26 that trace metals in waters exist in various chemical forms due to the formation of stable complexes with numerous inorganic and organic ligands, and the adsorption of many species onto colloid and particle surfaces. The reactivity of metals in biological or environmental processes is determined not by the total metal concentration in a water sample, but by the concentration of the most reactive or labile species present.
The distribution of metal species influences the bioavailability, toxicity and mobility of the metal 27-30. These distributions vary with aquatic conditions, regulated by salinity, redox conditions, suspended sediments, organic matter and biota. Table 1.1 shows the major forms of metal species in natural waters.
Organic matter content varies between 0.3 and 3 mg l-1 of carbon in open seas, and usually between 1 to 10 mg l-1 in rivers, lakes and estuaries 31. In some waters, like wetlands, organic matter content can go as high as 30 mg l-1 32. These organic compounds, released by living organisms, or resulting from their decomposition, can be classified into two main categories 28, 29: non-humic substances and humic substances. Non-humic substances generally have well defined structures (e.g. proteins, polypeptides, carbohydrates, fats, waxes, resin, pigments, amino acids and other low molecular weight compounds). Such compounds are generally rapidly degraded and utilised by microorganisms 32. Humic substances (HS) are formed by microbial activity on nonhumic substances, as well as abiotic polymerisation. Phenol groups, quinines, phenol carboxylic acid groups and related functional groups are common in humic substances. Humic substances are also quite resistant to further microbial degradation and consequently tend to persist in
6 Chapter 1 waterways 32. In some freshwaters humic substances consist of between 60-80% of the dissolved organic carbon 33-35. This percentage is usually lower in seawater (10-30%) 36.
Table 1.1 Physico-chemcal Forms of Metal Species in Natural Waters 37, 38 Physical states and Chemical forms Examples size ranges (nm)
3+ 2 Soluble Oxidation state Cr (aq), CrO4 (aq)
(<5) 2+ Simple hydrated metal ion Zn(H2O)6 (aq)
Simple inorganic complexes Zn(H2O)2Cl2(aq)
Stable ion pairs ZnCO3(aq), PbS(aq)
2+ Complexed to low molecular Cu -glycinate(aq) weight HS
2+ Complexed to high molecular Cu -fulvate(aq) weight HS
Organometallic complexes Hg(CH3)2(aq)
2+ 2+ Colloidal Adsorbed on inorganic colloids Cu - Fe2O3(s), Cd - MnO2(s)
(10-500) 2+ Adsorbed on organic colloids Pb - humic acid
2+ Adsorbed on mixed colloids, Cu -Fe2O3(s)/humic acid (inorganic/ organic)
Particulate matter Precipitates, co-precipitates PbCO3(s), Cd-FeOx(s)
(>500) Mineral particles PbS(s)
2+ Metals adsorbed on solids Cu -CuS, CuCO3 on clay minerals, MnIV oxides
Metals incorporated with organic Metals in plankton, detritus material
7 Chapter 1
One of the most important speciation interactions is due to the complexation of trace metals by organic matter. Natural waters contain both a vast range of organic matter of biological origin and organic pollutants, which have a range of complexing properties.
Due to the high concentrations of natural organic matter (NOM) relative to trace metals and the presence of complexing functional groups on the NOM, a large fraction of the trace metals in many natural waters are complexed to NOM, usually the humic substances
(e.g. Pb-HS 39, Zn-HS 7 and Cu-HS 40). About 50% of dissolved lead 41, 42 15-35% of dissolved cadmium 43 and > 90% of dissolved copper 44-46 and zinc in seawaters 47, 48 are usually found to be complexed with natural organic ligands that appear to be produced by organisms in the upper ocean 49.
Trace metals can also adsorb readily to particulate materials (mineral and organic).
Approximately 95% of trace metals transported from land to sea by surface waters are adsorbed on mineral particles directly, or are bound to organic matter coating these particles 50. During such transportation, sorbed species may be redistributed between the aqueous and solid phases as a result of changes in the physicochemical conditions of the water, leading to redistribution amongst various competitive equilibria, including the formation of soluble complexes with inorganic ions (e.g. Cl-or OH-) and molecules (e.g.
H2O or NOM).
The flocculation of colloids into larger particles occurs in estuaries due to the increase in ionic strength that partly neutralises the stabilising charge of the colloids and due to the presence of humic matter that induces their aggregation 51, 52. These particles then settle out of the water column and are incorporated into the bottom sediments. Over 50% of the metal ions in rivers are removed by estuaries in this manner 53, 54.
8 Chapter 1
Some metals can occur in various oxidation states (e.g. Cr, Fe and Mn). While redox- active metals usually exist in an oxidised form in waters, conversion to reduced ions can occur in the sediment at depths below the redox boundary or when waters become anoxic.
Biological and photochemical-catalysed reactions can also influence the oxidation state of a metal. The photochemically enhanced reduction of insoluble Fe (III) oxides provided a possible source of Fe (II) 55, 56. A hydroxide ion donates an electron to a photoexcited Fe
(III) surface atom resulting in surface bound Fe (II) to solution 56. While iron is not a trace metal, many trace metals that do not have various oxidation states have species that adsorb strongly to iron (III) oxyhydroxide particles and colloids and therefore have speciation indirectly dependent upon the oxidation-reduction conditions of a waterway.
Some trace metals can form organometallic species. Organometallic metals usually have a harmful effect, due to their high solubility in fatty tissues and organs relative to their water solubility. Through biological and chemical processes methyl-mercury is formed from inorganic Hg in sediment 57, water 58, soil 59 and other sites, such as the roots of floating aquatic macrophytes 60. Other organometallic species have been produced artificially, such as tributyltin, which has been used as an antifouling agent within paints 61, 62.
1.3. THE NEED FOR SPECIATION MEASUREMENTS OF TRACE METALS
Metal speciation studies are required to understand metal availability and thus potential toxicity to organisms 27, 63-67. Changes in environmental conditions, whether natural or anthropogenic, can strongly influence the behaviour of both essential and toxic elements by altering the forms in which they occur. Some of the more important controlling factors, as discussed above, include pH 68, 69, ionic strength 70, oxidation-reduction (redox) potential 71 and the availability of “reactive species”, such as complexing ligands (both
9 Chapter 1 organic and inorganic) 72, particle surfaces for adsorption, and colloidal matter 73.
Examples of changes in the speciation of an element, that occurs in response to a change in one or more of the above parameters, and which leads to an increase in toxicity or bioavailability are:
(1) A decrease in the pH of soil groundwater, from acid rain or acid sulfate soils, can
increase the leachability of aluminium from aluminosilicate minerals in the soils 74,
75, resulting in detrimental effects, including, in extreme cases, fish-kills in
receiving waters.
(2) Arsenic, an extremely toxic element in its inorganic forms, is relatively innocuous
as arsenobetaine (a common form in fish) 76, 77.
(3) Organotin compounds, of which perhaps the best known are the antifouling agents
of tributyltin 78 and triphenyltin 79, are generally more toxic than inorganic tin
species 80.
(4) Changes in the oxidation state of an element, in response to a change in the redox
status of the water, can also have a profound effect on bioavailability and toxicity.
For example, while chromium (III) is an essential element, chromium (IV) is
highly toxic; similarly arsenic (III) is generally much more toxic than arsenic (V)
54, 81.
The toxicity of a particular dissolved metal species towards an aquatic organism is closely related to its ability to react with a biological membrane 5. The penetration of the membrane by a metal ion, to react with the cell components, depends on the direct lipid- solubility of the metal species (usually only uncharged organometallic species), or the extent and rate of reaction of the metal ion with a membrane transport protein. Metal- protein interactions, which lead to carrier-mediated transport of the metal across a biomembrane, will, for bivalent ions, be thermodynamically favoured when the metal is in
2+ + + the simplest chemical form, e.g. Cu(H2O)4 , CuCl or Cu(OH) . For tervalent ions, such
10 Chapter 1 as Fe(III), however, the most bioavailable form may be an organic complex, as hydrolysis and polymerisation can render the free ion inactive 82.
In some cases, kinetics rather than thermodynamics may dictate the biologically active
+ chemical species. The toxic form of aluminium appears to be Al(OH)2 , which reacts with gill mucus to hinder the transport of oxygen, potassium and sodium 83. This species was previously shown to be the kinetically-favored species in the reaction between aluminium
(III) and a hydroxyazo compound 84. The reaction of metal ions with biological membranes is a particularly complex process, and cannot be explained by simple diffusion models 85. Most studies of the toxicity of heavy metals for fish have shown that the free
(hydrated) metal ion is the most toxic form 54. In the case of copper, hydroxy complexes are also believed to be toxic, although to a lesser extent 86. Strong complexes, and species associated with colloidal particles, are usually assumed to be non-toxic, due to low biological uptake where the exposure route is through contact with water.
Several models relating trace metal speciation with biological uptake through contact are described below. However, contact with water containing trace metal species is not the only mechanism of exposure. Some organisms, such as filter feeders and particle feeders, are likely to take up trace metals through ingestion of particles and colloids with metals adsorbed to the surface 87, 88. The bioavailable forms then depend upon the gut conditions of the organism 89.
1.3.1. The Free-ion Activity Model
Prior to about 1975 researchers tended to emphasise the target organism and the influence of biological variables (e.g. life stage, nutrition and age) rather than the exposure regime
(e.g. metal speciation, pH, hardness, alkalinity and ionic strength). In 1976, due to an 11 Chapter 1 improved understanding of metal speciation from environmental chemistry, aquatic toxicologists shifted their focus from the target organism to the chemistry of the exposure medium. Toxicological studies were performed in a defined media with synthetic ligands
(with known stability constants and hence 'known' metal speciation) 90. This approach was highly successful in synthetic media, with ligands forming soluble hydrophilic complexes. Eight years later, Morel 91 proposed that the bioavailability of a dissolved metal is related to its free ion activity. He suggested that the decrease in metal toxicity, observed in the presence of chelating agents, is simply the result of a decrease in the bioavailability of metals due to chelation of the metals in the medium, and not to a positive physiological effect of the chelating agents.
Significant correlations have been established between the toxicity of a metal and the chemical reactivity of the metal, as measured by ionic size, ionization potential, electronegativity, and its tendency to form bonds of a covalent nature 92. These correlations presumably reflect the fact that metals must exert their toxicological activity ultimately by reacting with surface functional groups on susceptible target molecules in cellular compartments, and that these reactions are governed by physicochemical laws.
An insight into the potential toxicity of a metal and the candidate target molecules affected may be gained by considering the relative ability of different metals to bind to organic ligands. An understanding of metal-ligand binding is also fundamental to studying the types of cellular macromolecules that may be involved in detoxifying metals by sequestration 93, 94.
Experiments with a variety of aquatic organisms have developed a convincing body of evidence to support the concept that the biological response elicited by a dissolved metal
z+ is usually a function of the free-ion concentration, M(H2O)n . The free-ion concentration
12 Chapter 1 is determined not only by the total dissolved metal concentration, but also by the concentration and nature of the ligands present in solution 91.
The interaction of a metal with an aquatic organism involves the following steps:
(1) advection or diffusion of the metal from the bulk solution to the biological
surface;
(2) diffusion of the metal through the outer 'protective layer', i.e. biomembrane;
(3) sorption/surface complexation of the metal at passive binding sites within the
protective layer, or at sites on the outer surface of the plasma membrane;
(4) uptake or 'internalisation' of the metal (transport across the plasma membrane)
95-97.
The possibility of a metal entering a cell by passing across the cell membrane is clearly dependent on the routes that are available and the forms in which the metal exists 98-100.
The following species may be involved in varying degrees for a metal to permeate the membrane:
2+ 2+ Metal ions (e.g. M ); Hydrated ions (e.g. M(H2O)6 );
+ Charged metal complexes (e.g. MCl(H2O)5 );
0 Uncharged inorganic complexes (e.g. MCl2 ); and
Organometallic complexes (e.g. CH3M).
1.3.2. The Biotic Ligand Model
During recent years the biotic ligand model (BLM) has been proposed as a tool to evaluate quantitatively the manner in which water chemistry affects the speciation and biological availability of metals in aquatic systems 101-104. The BLM model incorporates features of several detailed chemical equilibrium models, including the Gill Surface Interaction 13 Chapter 1
Model 105 and the Free Ion Activity Model 26, 91, into a unified framework that is used to calculate the distribution of a metal among the free ion, inorganic complexes and organic complexes 106, 107. In the context of the BLM framework, the tissue at the site of metal accumulation is defined as the biotic ligand. The concentration of metal that is associated with the biotic ligand is calculated in the same way as the concentration of metal that exists in association with any other organic or inorganic complexing ligands in the water.
The biotic ligand competes with the other complexing ligands (e.g. natural organic matter or organic ions) for binding of the available metal. The BLM framework provides a direct basis for predicting the reduction in copper bioavailability due to increasing levels of natural organic matter, carbonate alkalinity or pH.
The BLM also takes consideration of the interaction of the biotic ligand with other cations in solution, such as calcium or sodium. The major ions compete with the trace metal ion for binding at physiologically active sites at the biotic ligand. At sufficiently high levels, this competitive binding of major ions to the biotic ligand will effectively inhibit the accumulation of trace metals at the site of action. The explicit incorporation of this competitive effect in the BLM, in conjunction with a relationship of toxicity to the level of metal accumulation, provides a basis for predicting the reduction in metal toxicity associated with the presence of elevated calcium concentrations. It is in this manner that the BLM can be used to predict the well-recognized effect of decreasing toxicity of metals with increasing hardness.
The excellent agreement between measured copper and silver LC50s (lethal concentration associated with 50% mortality) (Figure 1.1) and BLM predicted LC50s demonstrates that the BLM predictions represent a viable alternative to conducting bioassays to evaluate
14 Chapter 1 metal bioavailability 102, 103. The data requirements for application of the BLM include chemical analysis of the receiving water or effluent and receiving water mixtures
Figure 1.1 BLM predicted LC50 vs observed LC50 for copper and silver (Cu data: Erickson et al. 108; Diamond et al. 109 and Ag data: Bury et al. 110; Bills et al. 111)
15 Chapter 1 associated with a discharge location. The required chemical analyses are generally of a routine nature and would include pH, DOC, alkalinity, major cations (Ca2+, Mg2+, Na+,
+ - 2- K ) and major anions (Cl , SO4 ).
1.4. ISSUES TO CONSIDER WHEN SAMPLING AND MEASURING TRACE METAL SPECIES
In order to comprehend the environmental chemistry of an element it would be necessary to characterise, in full, the proportions and concentrations of all the various forms under the diverse range of conditions possible in natural systems. Whilst this is clearly impracticable, it is important to measure concentrations of some important species of trace metals 112.
1.4.1. Sampling Factors
The determination of selected or consistent trace metal species is more challenging than the determination of total metal concentrations. Trace metal species distributions are very sensitive to physico-chemical changes, such as those that occur with sampling, storage and handling. Some of the processes that may modify trace metal speciation include 18, 113, 114:
(1) Release or loss of elements or complexants (especially macromolecules and colloids) by desorption/adsorption to any surface used during sample handling
(polymer/glassware, filtration apparatus, etc.) 115;
(2) Gaseous re-equilibration of the sample with the atmosphere due to pressure change. Re-equilibration of gases with acid-base properties (e.g. CO2) may cause significant variations in pH and thus modify compound speciation. When anoxic samples are equilibrated with the atmosphere, oxidation of some of the inorganic species (Mn2+, 16 Chapter 1
2+ 2- 0 Fe , S ) may produce colloidal particles (MnO2, Fe(OH)3, S ) which may dramatically change the species distribution of many trace metals, owing to their strong redox or adsorption reactions with these colloids 116;
(3) Coagulation of colloidal matter, followed by sedimentation of the aggregates and the associated trace compounds. Colloids are ubiquitous in natural waters and include a large fraction of trace compounds 117;
(4) Microbial activity, such as the continued metabolism of microbes during sample storage, may significantly alter the chemical composition of the sample. For example, the pH may vary because of continued respiration (pH decrease) or photosynthesis (pH increase). Dissolved concentrations of trace metals may also be changed as a result of their continued uptake or release by living micro-organisms. Complexation or enzymatic properties may also change owing to the release of biomolecules 118; and
(5) Virtually all methods used to preserve the trace metal concentration will dramatically affect the speciation with the addition of acid or the freezing of the sample. Therefore samples collected for speciation are often changed, which means that measurement should be undertaken immediately.
Some of these problems (1-3) may be minimised by special (often tedious) precautions, but problems (4) and (5) are natural processes which cannot be eliminated without dramatically perturbing the sample. Indeed, in their natural environment, aquatic samples are not at thermodynamic equilibrium; at best, they may be in steady state conditions due to the continuous inputs (e.g. soil leaching, atmospheric inputs, cell growth) and outputs
(e.g. coagulation/sedimentation, cell death) of colloids and microbes 113, 119, 120. While the
17 Chapter 1 sampling process stops most of the inputs, coagulation and microbia turnover may continue and any anti-coagulant or antibiotic may either induce drastic changes in the chemical speciation of the test compounds or cause analytical problems 119.
1.4.2. Measurement Factors
Other difficulties with speciation occur at the measurement step, including:
(1) difficulties associated with isolating the metal species of interest from complex
matrices;
(2) most speciation measurement techniques available disturb (to some extent) the
equilibria existing between the various chemical species present in the system
under study;
(3) for species present at ultra-trace levels, few analytical procedures possess the
degree of sensitivity required; and
(4) suitable standard reference materials are often unavailable.
The nature of the challenge varies with matrix type; seawater is particularly challenging due to the high concentrations of matrix ions.
Basically three general approaches have been used for measuring trace metal speciation in waters. It will be useful to consider the general strategies of sampling with respect to these various problems of speciation measurement. The conventional approach to water quality sampling and analysis (where grab samples are collected, usually filtered, and preserved by acidification before analysed in a laboratory) suffer from most of the above limitations. Furthermore, the approach provides a measurement of the concentration/speciation at only one time. In dynamic waters, such as rivers and estuaries, this type of measurement is not likely to be representative of the average condition 121.
Therefore a comprehensive sampling program is required, with hourly sampling across 18 Chapter 1 tidal and diurnal cycles, for each major season, as well as event sampling 122. This sampling approach is logistically complex, being time consuming and expensive, but does provide good data, if the problems inherent to speciation measurements can be overcome.
The advantage of this approach is that the most sophisticated measurement instruments can be used.
Another approach, on-site analysis, involves removal of a water sample followed by immediate analysis on-site. The process is usually automated and often uses laboratory procedures and instruments, which may have been adapted to suit the appropriate field conditions. The on-site analyses approach comes close to the ideal of real time measurements, minimising some artefacts that are associated with sample storage 123, 124.
While close to the ideal, the approach has not yet been widely utilised for environmental monitoring. Problems that limit such use include it being expensive to implement for the most sensitive instrumentation, which usually require controlled laboratory conditions.
The main exceptions to this are electrochemical methods. The automation required is also likely to be challenging as the sample will usually have to be filtered on-line; this process will need to be maintained to achieve accurate measurements. Another problem arises with this approach having to be deployed close to land or boats, both of which can influence the sample composition. If the sample has to be piped for long distances then many of the surface related and microbiological processes listed above can become problems.
Given these difficulties, alternate approaches to trace metal speciation have been sought; they have usually involved some type of sensor that can be deployed in situ. There are three ways in which in situ sensors can operate. They can continuously respond to a trace metal species that interacts with the sensor (termed the labile fraction); they may make
19 Chapter 1 discrete measurements in situ; or the labile trace metal species are accumulated continuously in situ and are stored in a stable form, while quantification takes place upon returning to the laboratory 125. These various approaches minimise many of the difficulties associated with trace metal speciation measurements concerning sampling, preservation and storage. The main problems with many of the in situ techniques currently reported include: lack of sensitivity; they are technologically complex and/or expensive; or they cannot be used for very long time periods.
There has been a lot of recent interest and development with in situ measurement approaches, for the reasons described above. The major advantages of in situ measurements for natural water monitoring, compared with conventional sampling and laboratory analysis, are:
(1) elimination of many of the artefacts due to sample handling, i.e. no or minimum
sample transformation;
(2) minimisation of the overall cost of data collection (in particular, due to a reduction
of analysis time);
(3) possibility of real-time analysis, allowing rapid detection of pollutant inputs (e.g.
monitoring of industrial wastes or water quality in water treatment plants);
(4) ability to accumulate detailed spatial and temporal data banks of complete
ecosystems (lakes, aquifers, etc.);
(5) possibility to perform measurements in locations which are difficult to access
(boreholes, deep lakes or oceans); and
(6) possibility of measuring concentration gradients and fluxes at environmental
interfaces (sediment-water; air-water), at high (sub-mm) spatial resolution 116.
These aspects are important both for studies of ecosystem functioning and for water quality monitoring.
20 Chapter 1
A number of criteria have been recommended for the development of in situ probes:
(1) reliable, automatic measurements, in the field, (measurements are often required at
a depth in which no visual control is possible);
(2) simple, compact, low cost apparatus;
(3) no or minimum sample transformation (minimisation of artefacts);
(4) high sensitivity for minor and trace compounds (10-7-10-15 mol l-1) 4;
(5) multi-elemental analysis capability for trace metals;
(6) selective speciation measurement or other information on the distributions of
species;
(7) physically and chemically non-perturbing for the system tested; and
(8) preferably measurement time faster than the time scale of the process studied 18.
A number of speciation measurements used for in situ trace metal speciation measurements are described in the following section. The diffusive gradients in thin films
(DGT) technique is then described and assessed in detail.
1.5. TECHNIQUES USED FOR IN SITU MEASUREMENT AND SPECIATION OF TRACE METALS
Various approaches have been developed to measure trace metal species in situ over the last decade including ion-selective electrodes (ISE), various voltammetric techniques, ultra-filtration, dialysis, diffusive equilibrium in thin films (DET) and permeation liquid membranes (PLM), as well as the DGT technique. ISEs involve the use of potentiometric measurements. They directly relate the measured potential to the logarithm of the concentration (or more specifically, the activity) of a specific hydrated ion 126, 127. The applicability of ISEs may, however, be restricted by their sensitivity (detection limit 10-6 –
10-7) 128, 129 and selectivity. Unfortunately interfering ions in the waters can be an
21 Chapter 1 important source of errors, which can lead, in general, to the overestimation of the ion concentration. ISEs have been developed for both Cd2+ (detection limit 10-7 mol l-1, Hg2+,
Ag+, Cu+ interfere) and Cu2+ (detection limit 10-8 mol l-1, Hg2+, Ag+, Cu+ interfere). To date potentiometric measurements have been limited for use in natural waters due to the low practical detection limits.
Voltammetric techniques, particularly those involving a stripping approach (anodic stripping voltammetry, cathodic stripping voltammetry or chrono-potentiometric stripping analysis) 130 provide the most direct method for the study of trace metal speciation at low concentration levels (10-7 to 10-12 mol l-1). These techniques do not normally require the pre-concentration of the water sample by physical methods 131, 132. However, many factors, such as pH, temperature and ionic strength, may influence the electrode processes and affect the signal 133, 134. Consequently calibration must be performed with great care, and with due regard to the physicochemical processes involved. These voltammetric techniques are highly operationally-defined and the trace metal fraction that is measured has been defined already 134, 135. Their approaches have been used for on-site measurements as well.
Dialysis 136, 137 is also used in water studies to separate high molecular weight and colloidal forms of trace metals from smaller species, which are often more labile 134, 138.
Dialysis can be accomplished over a range of nominal molecular weights from 210 to
300,000. The passage of species through a filter membrane depends on species geometry as well as on molecular weight 139. A similar technique to dialysis 140, diffusive equilibrium in thin films (DET), is based on the free exchange of ions between the water in a hydrogel and the sample solution (e.g. natural waters), supposing that there are no reactions between the hydrogel and analytes 141. DET also operates on a size-fractionation
22 Chapter 1 basis, although the actual sizes excluded have not been defined as yet. Dialysis and DET are deployed in situ and are equilibrium techniques, where it is assumed that the concentration collected is the same as that in the water sampled. All of these methods require laboratory-based instrumental measurement of the trace metal metals accumulated.
Permeation liquid membranes (PLM) are an emerging technique that is similar to DGT in some regards. PLMs use a water immiscible liquid membrane containing a complexing agent selective for the analyte of interest 142, 143. This layer is called the carrier phase because it is meant to selectively transport the trace metal analyte to an aqueous phase containing an even stronger complexing agent (the stripping phase). The PLM device is deployed in situ with the analytes accumulating within the stripping phase. As this phase is a solution, some attempts have been made to incorporate an in situ sensor but sensitive laboratory instrumentation can also be used 144, 145. The PLMs are selective for free and lipophilic metal species. Determination of concentrations down to 10-13 mol l-1 is possible.
1.5.1. Diffusive Gradients in Thin Films (DGT)
The diffusive gradients in thin films (DGT) technique was developed from the DET technique 14, 146. DGT added a binding phase to the diffusive hydrogel layer 14, 15. Analyte species diffuse through the hydrogel layer to the binding phase, which for trace metals is a hydrogel containing beads of Chelex 100 resin. These beads are situated along the hydrogel surface in contact with the diffusive layer when the DGT device is assembled.
As a result, labile trace metal ions (and cations), diffuse through the diffusive layer to bind to the binding phase. The solution concentration at the interface between the diffusive layer and the binding phase should ideally be zero. If this occurs a constant concentration gradient is maintained within the diffusive gel layer between this interface and the solution
23 Chapter 1 analyte concentration. A flux of labile trace metal ions occurs into the DGT device. This flux is able to be quantified using an equation derived from Fick’s law of diffusion, and can also be used to estimate the analyte solution in the sampled water. DGT thus has the potential to be used to measure trace metal concentrations and speciation in natural waters
14, 16.
1.5.1.1. DGT Principle and Theory
Figure 1.2 shows a conceptional view of DGT. The diffusive gel is usually covered with a membrane to protect the hydrogel surface from having particles adhering to it. The membrane behaves like an extension of the diffusive layer 15, 147. The diffusive layer thickness is ∆g (diffusive hydrogel + membrane thickness).
Between the diffusive layer and the bulk solution a diffusive boundary layer (DBL) forms.
The DBL thickness, δ, is determined by the velocity of the water across the surface; it is also a region where the transport of ions occurs solely by diffusion. Within a few minutes of immersion into a water body with analyte concentration, C, a steady state linear concentration gradient, is established between the solution and the resin gel, as described above. By exploiting this simple steady state condition, the DGT technique can be used to measure concentrations in situ. The flux, J, of an ion through the gel is given by Fick's first law of diffusion (equation 1.1), where D is the diffusion coefficient for the ion at the given temperature and dC/dx is the concentration gradient:
dC J = − D × (1.1) dx
24 Chapter 1
If the concentration gradient of ions in the diffusive gel is kept constant, the flux is given by equation 1.2, where C is the bulk solution concentration of an ion and C' is the analyte concentration at the boundary between the binding gel and the diffusive gel:
'C −C J = − D × (1.2) ∆g + δ
Diffusive Layer A Solution Binding Phase Binding
C
M D
C’ Concentration Concentration Membrane
0 ∆ g δ Relative Distance (cm)
Figure 1.2 Conceptual view of the steady state concentration gradient of a solute through a DGT device deployed in a well stirred solution with solution analyte concentration, C, diffusive layer thickness, ∆g , including 0.45 µm pore size cellulose nitrate membrane thickness, diffusion boundary layer, δ , analyte accumulated (M), diffusion coefficient (D), cross-sectional area (A).
If the analyte species are in rapid equilibra with the binding functional group and the interaction between the two is strong enough (i.e. the stability constant is high), C' will effectively be zero, providing the binding sites do not become saturated. In well stirred
25 Chapter 1 solutions, or natural waters with sufficient current, the boundary layer thickness, δ, is negligibly small compared to the thickness of the diffusive layer, (usually ∆g of 0.4-1 mm). Various estimates suggest that a range of 0.1-0.01 mm DBL thickness 148 may be typical of well-mixed waters. In a recent paper, Scally et al. 149 reported an average value of 0.024 ± 0.002 cm for δ using the following equation 150:
1 ∆gC δ = + (1.3) M DCtA DCtA
Given that C' equals zero and δ is negligible, equation 1.2 then simplifies to equation 1.4.
C J = D × (1.4) ∆g
DGT devices are deployed for a certain time, t. On retrieval, the binding gel phase is peeled off and the amount (mass or moles usually) of the accumulated trace metals are measured. Mass can be measured directly in the binding gel by drying it and using a beam technique, such as proton induced x-ray emissions (PIXE) 151, or in the case of radionuclides, indirectly measuring radiation 152. More commonly, ions in the binding gel are eluted with a known volume, Ve, of HNO3 solution (1 or 2 M) in the case of trace
16, 21, 153 metals bound to Chelex 100 resin . The concentrations of ions in the elution, Ce, are then measured by an appropriate analytical technique after appropriate dilution. Using these parameters, the accumulated mass (M) of analyte can be calculated, which in turn can be used to calculate the flux through the known area of the exposed diffusive layer, A:
M J = (1.5) At
26 Chapter 1
Combining equations 1.4 and 1.5, the rearrangement gives equation 1.6 (the DGT equation), which relates the concentration, in the bulk solution, to the known values of ∆g,
D, and A, the measured deployment time, t, and the measured accumulated mass, M 14, 16.
g M ∆g C = (1.6) DAt
This feature of DGT, whereby concentration is calculated from the measured mass and deployment time makes it ideal for in situ applications. The relationship of external concentration to measured mass is determined by the values of ∆g and A, which are fixed geometric quantities, and by the diffusion coefficient of analyte species in the gel, which can be measured under controlled conditions. These factors make DGT a kinetic technique, which can be deployed for varying times (t).
The basic principles of DGT have been verified repeatedly in the laboratory 16. It has been shown that the mass accumulated in the binding gel is proportional to deployment time (t) and inversely proportional to diffusion layer thickness (∆g); these two parameters are the ones most readily varied as part of a series of experiments. Experiments have confirmed that there is no interaction between metal ions and the diffusion gel 154, 155, which is an assumption of the DGT equation. DGT theory also required that the analyte concentration on the interface between the diffusion layer and the binding phase be maintained at zero through out the deployment. This experiment has since been used as a test to evaluate the use of DGT with analytes other than trace metals 147, 150, 156.
Of course there is a difference between deploying DGT devices in controlled laboratory experiments and in deploying them in the field. The main difference is that the trace metal speciation will be much more complex, as most laboratory experiments will only use free
27 Chapter 1 metal ions or simple inorganic species. The other trace metal species present in natural waters will have characteristic diffusion coefficients, which will be much lower than that for the free metal ion species. With association and dissociation of metal complexes continually occurring, membrane uptake and biological uptake of metals does not simply depend on the free metal ion activity 26. The depletion of the free metal ion at the membrane surface results in the dissociation of free metal ions from complexes 157, 158.
Anodic stripping voltammetry was used to obtain information on dynamic dissociation of metal complexes in natural waters 159, 160.
In DGT, the measured analyte species are the ones which can diffuse through the diffusive layer and be bound to the binding phase. Nevertheless, when metal ions are removed at the surface of the resin phase, a dissociation of metal complexes may be induced within the diffusive layer; the DGT measured mass will be the sum of contributions from both free metal ions in solution, MF, and free metal ion, M’, dissociated from the complexes,
ML, 16, 149.
(C D + C D )At M = MF MF M ' ML (1.7) ∆g where CMF is the concentration of free metal ion in the solution and DMF is the diffusion coefficient of the free metal ion. CM’ is the concentration of metal dissociated from ML and measured in the resin phase by DGT, and DML is the diffusion coefficient of the metal complex, ML.
Assuming the dissociation of ML is a first order reaction with a rate constant, k-1, then,
C M ' = C ML (1− exp( −k −1 τ )) (1.8) where τ is the time taken for the dissociation. 28 Chapter 1
This reaction can occur while ML is transported through the diffusion layer and the concentration of MF is lowered in this zone. As MF is consumed at the resin phase, the dissociation reaction shifts to produce more MF.
The characteristic time for transport of a complex through a diffusion layer of thickness
160 ∆g, td, is given by equation 1.9 ,
( ∆g )2 td = (1.9) 2DML
As ML can only be measured if it dissociates during time td, it is a reasonable approximation to set τ = td. Combining equation 1.7, 1.8 and 1.9 the total accumulated mass of metal measured can be expressed
C D (1− exp( −k ( ∆g ) 2 / 2D )) + C D M = ML ML −1 ML MF MF At (1.10) ∆g
2 When the dissociation of the ML is significant, k-1 >> DML/(∆g) , then,
C ML DML + C MF DMF M = At (1.11) ∆g
2 160 When k-1 << DML/(∆g) then ML can be considered to be inert and the DGT measurement is effectively only determined by the diffusion of MF,
C D M = MF MF At (1.12) ∆g
1.5.1.2. Diffusion Boundary Layer (DBL) and Biofouling Effect
Another important aspect of field deployment is the significance of the diffusion boundary layer which varies in thickness with the velocity of the water across the surface of the 29 Chapter 1
DGT device. The flux of metal diffusing to the resin gel of a DGT assembly depends on the thickness of the diffusion layer and the diffusion boundary layer (DBL) in solution. In well-stirred solutions the DBL must be negligibly small because the measurements obeyed the DGT equation. Free stream velocities of only a few cm s-1 are required for diffusive boundary layer thickness to approach zero 148. If there is no stirring in the solution, or if there is no or low flow (during in situ deployment, the DBL thickness can become significant when compared with the diffusion layer thickness and a low mass accumulation is obtained. This effect was observed to be important for still lake water 16,
23. In other waters, such as estuaries, the rate of mixing will vary over a day, with the tidal flows.
Webb and Keough 161 found, from a large number of comparisons, that estimates of metal levels were higher for DGT devices with 0.8 mm diffusive layers than for those with 0.4 mm diffusive layers, where the 0.8 mm and 0.4 mm units were deployed simultaneously at four sites: Westernport Marina, Hastings Jetty, St. Kilda Marina and St. Kilda Pier,
Melbourne, Australia, during 1999, from February 23 to March 25, and from August 25 to
September 24. It was explained that the biofouling of the membrane surface effectively increased the diffusive boundary layer relative to the known diffusive layer thickness, ∆g ,
16, 161. When deploying DGT in still water or when major biofouling of the membrane is likely, DGT devices with different ∆g values are deployed at the same location for the same time 150, 162 and equation 1.13 is used.
1 1 ∆g − ∆g − = 1 2 (1.13) M1 M 2 DCtA
With the two gel layer thicknesses, ∆g1 and ∆g1, C can be calculated from the measured values of M1 and M2. Whether due to poor mixing or due to biofouling growth, the effect of the DBL thickness, is effectively cancelled out through the use of this equation. Errors, 30 Chapter 1 however, may arise due to the variation in the extent of biofouling or the thickness of the
DBL between the DGT devices. The latter would be quite unusual if they were deployed at the same site at the same time.
1.5.1.3. Speciation Ability
Several field studies using DGT have reported that DGT-labile concentrations differ from
0.45-µm filterable concentrations in a range of waterways. In seawaters, in the Gold
Coast Broadwater, Australia, the DGT-labile concentrations as a percent of the 0.45 µm filterable fraction, were: 27 ± 12% for Ni, 29 ± 11% for Pb, 21 ± 2% for Cu and 28 ± 5% for Zn 163 and 44 ± 14% for Ni, 41 ± 12% for Zn and 23 ± 13% for Cu in two different seasons 164. Other fractions of DGT-labile concentrations of 0.45 or 0.2 µm-filterable concentrations reported from field studies on marine or coastal waters include 14-38% Cu in marine and estuarine waters around the USA 25 and 44-63% for Cu, and nearly 100% for Co and Cd in north Australian coastal water 165. In fresh waters, similar ranges of
DGT-labile metal concentrations (as a percentage of a filtered fraction) have been reported, and include: over 70% for both Cd and Cu 24 in both the Ring and Stitt Rivers,
Tasmania, Australia; 2.3% for Cd, 1.3% for Cu and 1.7% for Ni in Lake Tantare, Quebec,
Canada 22, and over 90% for Cd and Zn, 20-40% for Co, Ni and Pb, 5% for Cu in the
Water of Leith, an urban stream in Dunedin, New Zealand 166. Only one study has systematically compared field DGT measurements with other speciation measurements, which found that 10-35% of organically complexed Cu in seawater was DGT-labile 25.
The mechanism by which the DGT technique selectively measures trace metal species has been investigated in a preliminary manner only. In principle, four factors provide mechanisms that determine which species are measured by DGT. These mechanisms as well as the results of studies that have experimentally investigated these are listed below: 31 Chapter 1
(1) Selectivity by a size exclusion mechanism, where trace metal species that are larger than a particular size are not able to enter the DGT device. The DGT devices have two size exclusion components, the 0.45 µm membrane at the interface with the water and the diffusive gel itself. The membrane will effectively exclude all >0.45 µm material which are usually defined as particulates. Less is known about the size exclusion properties of the polyacrylamide hydrogel. One study used diffusive gels of different porosity 154 to demonstrate that diffusion coefficients decreased as the gel pore sizes decrease or the ion sizes increase.
(2) Differentiation of species on the basis of diffusional fluxes, where some species will diffuse more quickly through the diffusive layer and therefore accumulate more quickly in the binding phase. These species will have a larger influence on the DGT- labile concentration estimated compared with species that diffuse more slowly. The diffusional flux is dependent upon both the diffusion coefficient, which decreases with the mass of the species 154, and the concentration of each species in the waterway (equation
1.4). The soluble inorganic trace metal species, including free ionic forms, ion pairs and inorganic complexes will have higher diffusion coefficients compared with organic complexes and colloidal species. As mentioned previously (section 1.2) in many natural waters the dominant species are the organic complexes, but there is a great variety and number of complexes present so the concentration of any given species might well be comparable to those of the inorganic complexes. This is an area in which little information is available, but it may well be the most important mechanism of selectivity for DGT. For the most popularly used diffusive gel, agarose cross-linked polyacrylamide, the diffusion coefficients of Cu2+, fulvic acid (AFA), humic acid extracted from peat
32 Chapter 1
(PHA) or stream water (AHA), and AHA-Cu complex were determined as 6.28 × 10-6 cm2 s-1, 1.15 × 10-6 cm2 s-1, 0.35 × 10-6 cm2 s-1, 0.60 × 10-6 cm2 s-1, and 0.57 × 10-6 cm2 s-1 154 .
If the pore size of the gel is small enough to exclude more than 90% of the organic species and still allow the inorganic species to diffuse through, then based on the DGT equation
(equation 1.6), the DGT device can estimate the concentration of inorganic species directly 23. To measure inorganic and organic species separately, a series of DGT devices with different pore size diffusive gels can be used. When the devices are deployed in the same solution under the same conditions, the mass of metal accumulated on each DGT device ((MDGT) is the sum of contributions from both labile inorganic and organic species.
According to equation 1.6, we have 23
C D C + D C M = in in or or At (1.14) DGT ∆g
where Din and Dor are diffusion coefficients of inorganic and organic species respectively, and Cin and Cor are concentration of inorganic and organic species respectively.
Rearranging this equation,
M DGT ∆g Dor = Cin + Cor (1.15) Din At Din then concentrations of inorganic and organic species can be determined through the plots of (MDGT / Din) × ∆g / At versus Dor / Din. The intercept gives the concentration of labile inorganic species and the slope the concentration of labile organic species 23, 167.
33 Chapter 1
(3) Selectivity for organic complexes based on the capability of the binding layer to remove the trace metal ions. There are no published results investigating the importance of this mechanism. However, studies have been undertaken comparing the uptake of trace metals using the Chelex-100 binding phase with a synthetic ferrihydrite material 14, also impregnated into a polyacrylamide hydrogel. The two binding phases gave significantly different results for synthetic solutions without organic complexes being present 164. This suggests that the strength of the binding phase will also be an important mechanism.
Given the focus on developing new binding phases in this dissertation, the significance of this mechanism will be investigated. Chelex 100 has very strong binding groups, at high effective concentrations, which may out-compete most other ligands for metal ions, especially at the low concentrations of organic compounds present in most natural waters
168.
(4) Selectivity for strong complexes that have rapid ligand exchange kinetics. Some complexes will be strong enough that the DGT binding phase does not replace the binding ligands of the complexes. However, these complexes may still be DGT-labile if the complexes dissociate within the time it takes the species to diffuse across the diffusive layer. In this case, equation 1.10 applies (section 1.5.1). The measured mass by DGT is the mass contributed from free metal ions in sample solution and dissociated from weak complexes 149.
1.5.1.4. Binding Phase
Polyacrylamide hydrogel, embedded with Chelex 100 chelating resin, has been used in most previous studies of trace metal measurements with DGT. Clearly there are many other materials that could be used as binding agents for DGT. Given the discussion in the previous section on speciation, it is apparent that different fractions of trace metal species 34 Chapter 1 could be measured using binding materials with varying binding strengths. This section describes concepts from other research fields in which the binding of trace metals is required; such concepts are likely to provide ideas and materials appropriate for applications with DGT.
Affinity membranes and hydrogels with chelating groups have been used for metal ion separation and concentration 169, 170. They can be obtained by the copolymerisation of different functional monomers, such as: diallyldimethylamonium chloride 171, dimethylaminoethylmethacrylate 172, 2-acrylamido propanesulphonic acid 173, 3- acrylamido-3-methylbutanoic acid 174; or by the modification of functional groups in the polymer matrix, such as: Cibacron Blue F3GA-incorporated poly(2-hydroxyethyl methacrylate) 175, Alkali Blue 6B-attached poly(2-hydroxyethyl methacrylate) 176 and modification of poly(4-vinylpyridine) with dithizone 177. In addition, there are a number of commercially available solid-state membranes containing functional groups appropriate for binding trace metals.
Water soluble high molecular weight complexing agents have also been used for removal of metal ions in ultrafiltration processes 178, 179. The use of a semipermeable membrane allowed the retention of the polymeric materials, while maintaining diffusibility for the species to be chelated or complexed. A wide variety of water soluble polymers may be used, such as poly(ethylenimine) 180, poly(2-acrylamido-2-methyl-1-propanesulfonic acid)
181 and poly(methacrylic acid) 182. Water-soluble polymers are commercially available or can be synthesized by different routes. One universal route for the synthesis of different polymers is copolymerisation. With a good selection of both monomers it is possible to improve properties such as water solubility, metal ion binding capacity and selectivity 183.
35 Chapter 1
N N CH CH2 CH CH 2 n+ CH2 M CH2 CO O C Mn+ C O OC O O O O
i ii
CH CH2 CH2 CH CH CH2 CH CH2 N N Mn+
N N O O S Mn+ S CH CH2 CH CH2 O O O O
iii iv
Figure 1.3 Schematic diagram representing the nature of metal ion binding: (i) carboxylic type complexes; (ii) maleylglycine type complexes; (iii) amine type complexes; and (iv) sulfonate type complexes.
Most suitable binding materials will make use of acidic functional groups, which are
Lewis bases (electron donors) or cation exchangers. As shown in Figure 1.3 183, polymers interact with metal ions mainly through electrostatic forces 184, while selectivities arise through the formation of coordination bonds 185. Functional groups, with carboxylic groups, can act as mono and bidentate ligands 186, while polymers with sulfonate groups favour the electrostatic polymer-metal ion interaction 184.
1.5.1.5. Overview of the DGT Technique
The DGT technique has many useful properties as an in situ measurement technique:
(1) the device is easy to use and effectively minimises contamination;
(2) it concentrates metals in situ and selects against many interfering matrix
constituents;
(3) many trace metals can be measured simultaneously;
36 Chapter 1
(4) it allows estimation of time-averaged concentrations over the length of the
deployment period, i.e. DGT never sleeps;
(5) it directly measures a flux;
(6) it is a kinetic technique, that is dependent on the deployment time and can thus be
used for varying times; and
(7) it is capable of speciation measurements, which is one of the main aspect being
explored in this study.
However, there are some aspects of DGT that could still be improved or developed further.
Firstly, the Chelex 100 impregnated-polyacrylamide gel used, while proving to be satisfactory for many studies, has a number of limitations for inexperienced users. The
Chelex 100 beads are aligned against the face of the gel, but it is sometimes difficult to tell which face this is. If a binding gel is deployed upside down, then the diffusive path length is increased to include some of the binding gel. if this happens the calculated solution concentration will be underestimated 187. Furthermore this binding phase is not ideal according to the assumptions implicit in the DGT equation. The Chelex 100 beads are not continuous at the interface with the diffusive gel. Consequently not all binding will occur right at the interface; some will occur at depth within the binding phase. This aspect will produce a systematic error of about 5-10% underestimation of the solution concentration and may also decrease the reproducibility of the results. Finally, the pattern of Chelex 100
(or other resins) within the binding gel is not fully reproducible. This leads to either binding gel being thrown away or to a further increased variability of the results.
Each of these limitations could be overcome by development of new binding phases that have a continuum of binding functional groups at each face; as required by the DGT equation. This also removes the possibility of the binding phase being placed upside
37 Chapter 1 down. Some DGT studies have used colloidal sized micro-Chelex beads 151, which should effectively overcome these limitations. However, this material is very expensive and has not been used routinely with DGT deployments as yet. The current study will investigate the development of materials for binding phases that more effectively meet the conditions required by the DGT equation.
Another possible advancement of the DGT technique would involve the use of materials, other than polyacrylamide gel for both the diffusive and binding phases. Acrylamide, the monomer of polyacrylamide, is a suspected carcinogen and must be handled with care.
Polyacrylamide gel is also expensive to make, mainly due to the agarose-derived cross- linker required. There is also some evidence that this cross-linker varies somewhat in nature from batch to batch and there is currently only one supplier in the world. Because of these reasons other materials will be investigated for use with DGT as part of this study.
A final aspect of DGT that could be improved is to remove the need to use an estimated correction factor to compensate for the fact that not all of the bound metal is released during the elution step from the binding phase. To date only one study describes the elution efficiencies for a range of metals 16. This step is also likely to decrease the reproducibility of DGT given that 100% elution does not occur, and the correction factors used are likely to be mean values. It would be more effective to use elution conditions that were reproducible around 100%. Such an approach might also require the use of binding phases from which metals are more easily removed, compared with the Chelex
100 binding phase. Another alternative is the use of a binding phase in which elution is not necessary at all, i.e. a liquid binding phase. This approach is also considered in this study.
38 Chapter 1
1.6. OBJECTIVES OF THIS STUDY
The aim of this work was to further develop the diffusive gradients in thin films (DGT) technique capable of being used for the in situ measurements of trace elements in the environment. The research had two main foci: the development of new DGT devices and their evaluation. The following have been investigated in this thesis.
(1) Develop new DGT devices, particularly with new binding phases, to overcome the limitations with the previous DGT as described above. These new binding phases need to behave in a more ideal manner (i.e. with uniform binding occurring only at the interface with the diffusive gel) and should be easier to handle and make. Ideally, it would be better that the interface is between a solid and a liquid. This idea induces the use of a solution as a binding phase along with the use of a membrane as a diffusive layer, which makes it possible to eliminate the involvement of the polyacrylamide gel. To achieve these objectives, a number of strategies, including syntheses of new hydrogels and applications of new type of binding phases, were to be used in DGT sensor development and new concept evaluation. Most importantly, these new DGT sensors with varying binding properties of binding phases and diffusive properties of diffusive layer have differing capacities of measuring trace metal species in waters.
(2) Investigate properties of the new binding phases and the new diffusive layer to study the suitability of the phases for DGT use. These need to be done under varying conditions.
(3) Deploy the newly developed DGT sensors in controlled laboratory conditions to validate the agreement between the DGT response and theoretical prediction, and in a 39 Chapter 1 number of well selected natural water sites with compositions of varying metal ions and binding ligands to test their ability for speciation measurements of trace metals in natural waters.
Inevitably, this course of study was a multidisciplinary effort, with the various practical requirements being documented in Chapter 2. These requirements included the design of new DGT systems reflecting the DGT objectives; the design of new DGT devices suitable for the new DGT systems; the syntheses and characterisation of new hydrogels with properties for DGT use; the preparation of new binding phases and new diffusive layers; the fabrication of DGT devices and application in waters; and the fabrication of cells for diffusion coefficient measurements and dialysis purification. A synopsis of the research carried out in each of the remaining chapters is described below.
Chapter 3 discusses the development of a new homogeneous binding hydrogel, synthesised by converting the polyacrylamide (PAM) diffusive hydrogel used in DGT to a binding hydrogel of poly(acrylamide-co-acrylic acid) copolymer hydrogel (PAM-PAA).
The specific objectives for this part of work were:
(1) to optimise synthesis conditions of the binding gel;
(2) to investigate the binding properties of the new binding phase under varying
conditions of pH, ionic strength, and time;
(3) to demonstrate the suitability and advantages of the new binding phase for DGT
use;
40 Chapter 1
This chapter aims to develop a new range of binding phases for DGT, i.e. binding phases prepared by chemically immobilisation of binding functional groups on the hydrogel backbone.
In order to develop a method for synthesising the homogeneous binding hydrogels, the synthesis of the poly(acrylamidoglycolic acid-co-acrylamide) (PAAG-PAM) with carboxylic binding functional groups is described as an example of a methodology
(Chapter 4).
Chapter 5 discusses the employment of a commercially available cellulose phosphate ion exchange membrane (P81) as a new DGT binding phase. This development aims to demonstrate a possibility of using non-gel based DGT binding phase.
Followed with the development of the new DGT solid binding phase, to further improve the binding interface, the invention of application of a poly(4-styrenesulfonate) (PSS) aqueous solution as a binding phase is discussed in Chapter 6. The aims of this chapter are:
(1) to demonstrate the feasibility of use of a liquid (non-solid) binding phase for DGT;
(2) to study the application of cellulose dialysis membrane as a diffusive layer;
(3) to investigate the diffusion properties of the diffusion layer and binding properties
of the binding phase;
In Chapter 7, the performance of the liquid binding phase DGT is further discussed. To study the capability of this new DGT device for speciation measurement of metals, a computer program, stability constant database, were to be used to calculate free metal fractions for comparison with the DGT results. The objectives of this chapter include,
41 Chapter 1 validation of this DGT device for the measurement of DGT labile metal in solutions containing varying complexing ligands and its application in natural waters.
To evaluate and compare the new binding phases developed in previous chapters, Chapter
8 describes the use of the new binding phases and new diffusive layer in DGT application, which include the deployment of the new DGT devices developed in previous chapters in solutions containing varying complexing ligands at laboratory conditions and in natural waters with varying compositions.
In summary, the overall objectives of this study are to develop new range of DGT binding phases. These binding phases improve the binding interface which is important for applying DGT equation. The use of the new binding phases and new diffusive layer in
DGT improve DGT technique as a tool for metal detection in waters to provide important information on metal bioavailability.
42 Chapter 2
Chapter 2 Experimental and Methodology
43 Chapter 2
2.1. INTRODUCTION
This chapter gives a brief description of the general experimental methods used in the current study. More detailed descriptions of experimental procedures, including those developed for this study, will be given in the relevant chapters related to the research described in that chapter.
2.2. REAGENTS AND SOLUTIONS
2.2.1. Chemicals and Materials
All reagents used in the study were of AR grade purity, unless otherwise stated. The 40% acrylamide monomer aqueous solution, ammonium persulphate and N,N,N',N'- tetramethylethylenediamine (TEMED) were supplied by Bio-Rad, Australia. The 2% agarose-derived cross-linker solution was obtained from DGT Research Ltd., Lancaster
University, UK. The acrylamidoglycolic acid monohydrate and poly(4-styrenesulfonate) of average molecular weight 70,000 were supplied by Aldrich, Australia. The cellulose nitrate and cellulose phosphate (Whatman P81) membrane were supplied by Whatman International
Ltd., UK. The cellulose dialysis membrane with molecular weight cut-off (MWCO) 12,000
Daltons was purchased from Sigma, Australia. The piston designed gel based DGT holders were obtained from DGT Research Ltd., Lancaster University, UK.
44 Chapter 2
Suprapur HNO3 (Merck) was used for the preparation of DGT elution and standard solution.
AR grade HNO3 was used to prepare 1:10 HNO3 acid baths for the cleaning of all plasticware used.
2.2.2. Solutions
2.2.2.1. Solutions for Polyacrylamide Gel Preparation
All solutions were prepared with deionised water (18 MΩ cm), as follows:
(i) 10% ammonium persulphate solution was freshly made before being used for the
preparation of polyacrylamide gel.
(ii) The 2% agarose-derived cross-linker solution and 40% acrylamide solution were used
as purchased.
(iii) The polyacrylamide gel stock solution was made by thoroughly mixing 18.75 ml of
40% acrylamide monomer (Bio-Rad), 7.50 ml of 2% agarose-derived cross-linker
(DGT Research Ltd., Lancaster University, UK) and 23.75 ml of deionised water
(Milli-Q).
2.2.2.2. Standard Solutions for Calibration of Measurements
All 1000 ppm stock solutions of copper, cadmium, potassium, sodium, calcium, magnesium, nickel, zinc and manganese were prepared by weighing the required amounts of the appropriate salt (Aldrich) preserved by acidification to pH 2 with HNO3 (Suprapur) and dissolving it in deionised water. The stock solutions were all stored in polyethylene bottles
45 Chapter 2 and kept in the dark. The standard solutions were prepared from the stock solutions by serial dilutions.
2.2.2.3. Synthetic Lake Water for Laboratory Evaluation of DGT
Synthetic lake water (Windermere, Lake District, UK) 152 with composition of [Mg2+] = 40.5
2+ + + - - µM; [Ca ] = 157 µM; [Na ] = 202 µM; [K ] = 17 µM; [Cl ] = 242 µM; [NO3 ] = 25 µM;
2- [SO4 ] = 85.5 µM was prepared by weighing the required amounts of the appropriate salt and dissolving in 25.0 l of deionised water according to the procedure of Chang 152. The concentration of each salt in the final dilution and quantity required for the concentrated stock solutions are shown in Table 2.1. Composite stock solutions I and III were prepared at 1000 times the final concentrations required in the synthetic lake water, while solution II was made up at 10 times the final concentration required. Additionally 0.89 g of CaCO3 was added to
25.0 l of deionised water and bubbled with CO2 for 8 h to ensure dissolution. The appropriate volumes of solutions I and III were then added to solution II, e.g. for 25.0 l of synthetic lake water, 25.0 ml of stock solution I, 2.50 l of stock solution II and 25.0 ml of stock solution III were mixed with 25.0 l of Deionised water. Air was bubbled through the mixed solution for one day to equilibrate it with the atmosphere. The pH of the final solution was approximately
7. The calculated and analysed compositions of the synthetic lake water are shown in Table
2.2.
Solutions containing Cu2+ or Cd2+ were prepared by adding specific amounts of the salts in the synthetic lake water solution (as made above).
46 Chapter 2
Table 2.1 Synthetic lake water stock solution compositions Stock Chemicals Final dilution Concentration of salt Concentration -1 solution (µeq l ) used to make stock factor solutions (g l-1)
I MgCl2⋅6H2O 84 8.5 ×1000
CaCl2⋅6H2O 162 17.7
Ca(NO3)2⋅4H2O 34 4.1
II CaCO3 71 0.036 ×10
III Na2SO4 191 13.6 ×1000
KHCO3 20 2.0
NaHCO3 9 0.76
Although, the concentrations of metal ions in the solution can be calculated from the dilution of the known mass of salts added to the solution, this calculation does not take into account the proportions taken up by the DGT devices, nor any surface sorption or precipitation reactions, either of which could reduce the metal concentrations. Any evaporation during the course of the experiment can also increase the concentrations relative to the calculated values.
Because of these unknown factors, the metal concentrations in the solution were measured independently using inductively coupled plasma - mass spectrometry (ICPMS). Aliquots of
5.0 ml of solution were withdrawn for analysis at the same time as the devices were extracted.
47 Chapter 2
Table 2.2 The composition of Windermere Lake Water 188 Measured Concentrations Concentrations Expected Concentrations in Natural in Synthetic Lake Water from Preparing Synthetic -1 Ions -1 -1 Lake Water (µmol l ) (µmol l ) Lake Water (µmol l )
Mg2+ 41 42 40.5
Ca2+ 135 133.5 157
Na+ 202 200 202
K+ 21 20 17
Cl+ 246 246 242
- NO3 35 34 25
2- SO4 88 95.5 85.5
2.3. PROCEDURES
2.3.1. Preparation of Diffusive Gel
The diffusive gel – polyacrylamide (PAM) – was prepared with acrylamide, agarose derived cross-linker (DGT Research Ltd., UK), ammonium persulphate and N,N,N',N'- tetramethylethylenediamine (TEMED). A stock solution comprising 18.75 ml of 40% acrylamide monomer (Bio-Rad), 7.50 ml of 2% agarose-derived cross-linker (DGT Research
Ltd., Lancaster University, UK) and 23.75 ml of deionised water (Milli-Q) was mixed thoroughly.
48 Chapter 2
Polymerisation was induced by adding 70 µl of freshly prepared 10% (w/w) ammonium persulphate (Bio-Rad) solution and 25 µl of 99% N,N,N',N'-tetramethylethylenediamine
(TEMED) (Bio-Rad) to the 10.0 ml of monomer stock solution. The solution was gently mixed and immediately pipetted into a mould comprising two slightly offset, clean glass plates (12 cm × 12 cm) separated by an inert U-shaped plastic spacer of known thickness
(0.25 mm) and held firmly together with clips. The mould was then incubated at 40 ± 2oC for
1 h to allow the polymerisation to occur and the resulting PAM hydrogel to set. The hydrogel was then hydrated in deionised water for at least 24 h with the water being changed at least three times to remove any unreacted reagents. The PAM sheets were then stored in 0.1 M
NaNO3 prior to use.
2.3.2. Preparation of Chelex 100 Binding Gel
The Chelex 100 resin embedded polyacrylamide hydrogel was prepared according to Davison
14 . Firstly, the resin was soaked in Milli Q water, then the excessive water was removed with a clean tissue. Then 10 ml of the gel solution (acrylamide solution containing the cross- linker) used for preparing the diffusive gel was mixed thoroughly with 2 g (wet weight) of
Chelex-100 (100-200 mesh) to create a stable suspension. Sufficient resin (about 0.2 g wet weight per millilitre) should be used to ensure that the resin density on the gel surface is maximal without affecting the casting or the setting of the gel. Polymerisation and gelation were induced by the addition of 60 µl of ammonium persulphate solution and 20 µl of
TEMED. The casting and rehydration procedures described above (for the preparation of the diffusive gel) were then carried out. The typical resin-gel thickness was 0.4 mm after hydration. The binding gel was stored in pure water (milli-Q).
49 Chapter 2
2.3.3. Characterisation of the Structure and Composition of Binding Hydrogels
The FTIR spectra of hydrogels were obtained using a Perkin Elmer FTIR Series 1000FTIR spectrophotometer to determine the change in the functional groups present. Disks of the gels were stretched out to make very thin films (a few µm), dried in the air and placed directly in the instrument. Each spectrum was collected by accumulating five scans at a resolution of eight wave numbers in the wavelength range of 4000 – 400 cm-1.
The elemental compositions of the hydrogels were determined by elemental analysis on a
Carlo Erba 1106 Elemental Analyser. The samples (~ 5 mg) were air dried before analysis.
2.3.4. Assembling and Disassembling the Gel Based DGT Devices
The gel based DGT devices were designed based a simple piston design commercially available from DGT Research, Ltd, Lancaster University, UK (Figure 2.1) 16. It consisted of a backing cylinder and a front cap with a 2.0 cm diameter window. The gel sheets described above were cut into discs of 4.9 cm2 and assembled by placing the appropriate binding gel, the diffusion gel and a 100 µm thick 0.45 µm pore size cellulose nitrate membrane filter (0.45
µm, Whatman), in order, on the top of the backing cylinder before firmly pressing down the front cap to form a tight seal. The membrane filter effectively extends the diffusive layer and protects the gel from particles and biological films in natural waters 15. It is essential that there are no trapped air bubbles to impede the diffusion. The diffusive gels were conditioned in an electrolyte solution, such as NaNO3 (0.01-0.1 M) prior to deployment to prevent the possibility of a junction potential across the gel which would affect diffusion 154.
50 Chapter 2
After deployment, the diffusive gel was peeled off and the binding gel was soaked in 5 ml of
2 M HNO3 shaking well, overnight. The acid solution was then analysed for metal ion content by FAAS after the appropriate dilution.
Membrane Filter Diffusive Gel Binding Layer Outer Sleeve with 2.0 Piston cm Diameter Window
Figure 2.1 Schematic representation of plastic DGT holder based on a simple piston design.
2.3.5. Measurement of Diffusion Coefficient in Diffusive Layer
Two types of diffusive membranes were used, a commercial and well-characterized dialysis membrane and a carefully produced gel-membrane. In both cases, the diffusion of metal ions across the membranes was experimentally calibrated to correct any factors that may affect the diffusion. The diffusion coefficients, D, of Cd2+ and Cu2+ ions in the diffusive layer
(hydrogel or dialysis membrane) were determined using a specially designed diffusion cell
147, as shown in Figure 2.2.
The diffusion medium, a polyacrylamide hydrogel of known thickness (∆g) was sealed in between the two compartments of the diffusion cell. A 1.4 cm diameter area of the gel was exposed to the solution through the circular opening between the two compartments. One
51 Chapter 2 compartment (compartment A) initially contained a solution of 50.0 ml of 10 ppm Cd2+ or
Cu2+ prepared in synthetic lake water. The other compartment (compartment B) initially contained a 50.0 ml solution of the same matrix ions as compartment A. Each compartment was mixed well with an overhead stirrer. 4.50 ml aliquots of solution were taken from compartment B and 0.200 ml from compartment A, (accompanying the addition of the same volumes of the two initial solutions to both compartments) at 10 min. intervals for 70 or 80 min. Concentrations of Cd2+ and Cu2+ in the samples were measured using FAAS. The replacement of the withdrawn samples with the original matrix solution diluted the solution in both compartments. The effect of the sampling on the concentration of compartments A and
B was corrected for, based on the sample analysis for the gel case.
Stirring motors Openings for sampling
Spacers
Stirrers Diffusive layer Diffusive
A B
Clamps
Figure 2.2 Cross section through a diaphragm diffusion cell.
52 Chapter 2
The diffusion coefficients were calculated from the slope of a linear plot of the measured mass passing through the membrane, M, versus the product of time, t, and ∆C, the corrected concentration difference of Cd2+ or Cu2+ between the two compartments during the diffusing time period using the following equation:
DA M = ()∆Ct (2.1) ∆g
DA slope = , where A and ∆g are known. ∆g
2.4. INSTRUMENTATION
2.4.1. Atomic Absorption Spectroscopy (AAS)
The flame atomic absorption spectrometer (FAAS) (SpectrAA-200, Varian) was used for the measurement of the total metals in the samples 189. Analytical standards and blanks were prepared in the same matrix as the samples and run before each batch of the sample runs. A standard was run every 6 samples to ensure that there was no change in sensitivity of AAS and that no evaporation of the sample occurred. 5.0 ml samples were used in the measurement.
The analytical errors were determined by replication of the blanks 190 and the data were interpreted by the statistical methods given by Wilson 191. The detection limit, defined as
-6 4.653σw, where σw is the standard deviation of the blanks (n = 15), was found to be 6.3×10 mol l-1 for Cu, 1.8×10-6 mol l-1 for Cd, 7.5×10-6 mol l-1 for Na, 5.8×10-7 mol l-1 for K, 3.7×10-7 mol l-1 for Mg and 6.6×10-6 mol l-1 for Ca with an air/acetylene flame. 53 Chapter 2
2.4.2. Measurement of Metal Concentrations in a Solution Containing PSS
The concentrations of metal-PSS complexes were determined by a flame atomic absorption spectrometer (FAAS) (SpectrAA-200, Varian), according to the standard guidelines of the manufacturers. The calibration curves were obtained by plotting the apparent concentrations of the standard solutions containing PSS of the same amount as in the diluted samples 192.
The instrument sensitivities were 5.5×10-7 mole l-1 for Cu and 1.7×10-7 mole l-1 for Cd with air/acetylene flame.
2.4.3. Solution pH Measurement
The pH of solutions was measured using a pH meter from Ionode PTY LTD, Australia. It was calibrated with standard buffer solutions (Australian Chemical Reagents, Australia) before use.
2.4.4. Solution Salinity Measurement
The salinity of the natural waters was measured using MC-84 Conductivity-Salinity-Temp.
Meter (TPS Pty Ltd, Australia), calibrated by manufacturers specified standard solutions (TPS
Pty Ltd, Australia).
54 Chapter 3
Chapter 3 Synthesis and Characterisation of a Poly(acrylamide-co-acrylic acid) Copolymer Hydrogel Based Binding Phase for the Diffusive Gradients in Thin Films (DGT) Technique
55 Chapter 3
3.1. INTRODUCTION
The advantages and limitations of the DGT technique have been described and discussed in Chapter 1. Most of the limitations arise from the use of the Chelex-100 impregnated polyacrylamide gel binding phase. While most of these are overcome with user experience, some of the limitations are inherent to the binding phase. The interface between the binding phase containing Chelex 100 and the polyacrylamide diffusive layer is not an ideal one. As the diagrams below indicate, the Chelex 100 binding gel does not have a continuous interface with the diffusive gel. This discontinuance occurs, even if each of the beads are packed as close together as possible, which may not actually be the case (Figure 3.1).
Effective interface
Figure 3.1 Schematic diagram of Chlex 100 binding gel and its interface with diffusive gel.
56 Chapter 3
The influence of the discontinuous interface is to move the effective interface into the binding phase. Even if the beads are perfectly aligned, the average interface will move about 25 µm into the binding phase on average. This move will introduce a slight systematic error in the DGT measurements of about 3%, which then underestimates the concentration calculated. The interface will also not be truly two-dimensional, possibly increasing the variation of results obtained. Consequently, there is a need to develop new binding phases for use with the DGT technique. These new binding phases should overcome the limitations, described here and in the Chapter 1, whilest maintaining all of the advantages inherent to DGT. This chapter, and the next three, will describe studies into different strategies to develop new binding phases. These strategies include introducing functional groups to the surface of the polyacrylamide gel, which are then able to bind the trace metals. Such a binding phase is likely to be an ideal phase, according to the assumptions implicit in the use of the DGT equation. Two main approaches can be used to introduce these functional groups: derivatization of the polyacrylamide hydrogel; or use of a copolymer derived from acrylamide and another monomer that has a suitable functional group and which will form a copolymer with acrylamide. Both approaches are described in this and the following chapter, respectively.
Another strategy (Chapter 5) for introducing new binding phases is to use a non-gel based membrane as the binding phase. Today, there are many commercially available chelating or ion exchange membranes that can bind trace metals (Introduction to Chapter 5). These introduce new advantages to DGT, such as re-use of the binding phase.
A final strategy for introducing new binding phases is to remove the need to use polyacrylamide at all, either as a binding phase or diffusive layer. A range of other
57 Chapter 3 membranes is investigated as diffusive layers, along with soluble polymers with functional groups able to bind trace metals.
Each of these strategies to develop new binding phases is expected to lead to different functional groups with differing binding strengths. These binding phases can measure quite different trace metal species, and thus this aspect is investigated for each of the binding phases developed here.
This chapter focuses on the development of a polyacrylamide hydrogel with a homogeneous and dense distribution of the functional groups able to bind trace metals.
Through chemical reactions it is possible for the functional groups that are capable of binding metal ions, such as amine 193, amidoxime 194, dithizone 177, carboxylic acid 195, and phosphorus 196, to be covalently immobilised on a hydrogel network backbone. There are some well known polymer hydrogels, such as polyacrylamide, poly(ethylene glycol), poly(vinyl alcohol) and poly(2-hydroxyethyl methacrylate), which can be chosen as polymerisation reaction skeletons.
Also described in this chapter is the preparation for, and the characterisation of a poly(acrylamide-co-acrylic acid) copolymer hydrogel (PAM-PAA) by the conversion of a fraction of the amide groups on the polyacrylamide (PAM) gel backbone to acrylic acid.
Additionally, the ability of this copolymer hydrogel to bind various metal ions under a range of conditions is examined, while the swelling properties of the new hydrogel are also characterised. Finally the feasibility of using the new homogeneous copolymer hydrogel for environmental analysis with DGT is investigated.
58 Chapter 3
3.2. EXPERIMENTAL
3.2.1. Preparation of the Poly(acrylamide-co-acrylic acid) Copolymer Hydrogel
The poly(acrylamide-co-acrylic acid) (PAM-PAA) copolymer hydrogel was prepared directly from the PAM hydrogel with a partial hydrolysis reaction similar to that described previously 197. However, in this work, the hydrolysis process was controlled by equilibration reaction conditions and was undertaken on a different form of polyacrylamide.
PAM hydrogel sheets were placed in 40 ml of 10% (w/v) NaOH solution 198, within a sealed flask at 75 - 80oC for 5 h, during which the amide groups underwent a controlled hydrolysis reaction. The ammonium gas given off could not leave the flask, however, which meant that the reaction proceeded to equilibrium and did not need to be regulated by time in order to obtain a reproducible degree of hydrolysis. The PAM-PAA hydrogel was then thoroughly washed with deionised water and rehydrated for 24 h. The sheets were then soaked in 0.10 M HNO3 for another 24 h, followed by a thorough washing with
o deionised water and storage in 0.1 M NaNO3 at 4 C.
3.2.2. Characterisation of the Structure and Composition of the PAM-PAA Hydrogels
The FTIR spectra of the PAM and PAM-PAA hydrogels were obtained using a FTIR spectrophotometer (Perkin Elmer FTIR Series 1000) to determine the changes in the functional groups present (Chapter 2).
Further, the elemental composition of the PAM and PAM-PAA hydrogels were determined by elemental analysis on a Carlo Erba 1106 Elemental Analyser (Chapter 2). 59 Chapter 3
3.2.3. Swelling Properties of the PAM-PAA Hydrogel
The weight of each dried PAM-PAA hydrogel disk was measured before the swelling test experiment. These disks were then soaked in solutions with pH ranging from 1.8 to 9.0 and a constant ionic strength (equivalent to 0.010 M NaCl) at room temperature (23ºC) for
24 h. The solution pH was adjusted by adding hydrochloric acid or sodium hydroxide, while the constant ionic strength was maintained by the addition of an appropriate amount of NaCl. The weights of each of the hydrated (swelled) samples were measured after the hydrogels had equilibrated in the test solution. Each measurement was performed in triplicate. The equilibrium swelling ratios were calculated, based on equation 3.1 199:
qw = ms / md, (3.1)
where qw, is the equilibrium swelling ratio, and ms and md are the weights of the hydrogel disks in the swollen (hydrated) and the dry state, respectively. The effect of the ionic strength on the hydrogel swelling behaviour was tested in the same manner by varying the
-5 NaNO3 concentrations from 1.0 × 10 M to 1.0 M at pH 7.0.
3.2.4. Metal Binding Properties of the PAM-PAA Hydrogel
The binding properties of the PAM-PAA hydrogel to metals were investigated using individual (non-competitive binding) and mixed (competitive binding) metal ion solutions.
The binding capacities of a range of individual metal ions, including Cd2+, Cu2+, K+, Na+,
Ca2+ and Mg2+ were examined. The experiments were carried out by immersing a PAM
PAA hydrogel in a solution containing 7.5 µM of a metal ion at pH 7.0 for 24 h, with constant stirring. The metal solutions were sampled and preserved before and after the
60 Chapter 3 hydrogels were deployed. Each measurement was undertaken in triplicate, as were those described below.
The influence of the ionic strength on the binding capacity of the copolymer hydrogel was
-5 tested by varying the NaNO3 electrolytes from 1.0 × 10 M to 0.10 M for metal ion concentrations of 7.5 µM at pH 7.0 for 24 h. To evaluate the influence of pH on the hydrogel binding capacity, the pH of the uptake solution was varied from 1 to 12, as described above.
The competitive binding of metal ions was also investigated by immersing a PAM-PAA hydrogel disk in a solution containing 7.5 µM of each metal ion at pH 7.0 for 24 h.
The binding capacity is defined as the maximum metal ion uptake measured at the saturated part of the binding curve and is determined by measuring the disappearance of metal ion in the uptake solution.
3.2.5. Elution and Analysis of the Metal Ions
The elution efficiency for the PAM-PAA hydrogel was measured for all metals of interest.
The solutions, with the same concentrations as in the uptake experiments, were used. The initial and final concentrations were measured for the uptake solutions to determine the amount of metal ion removed. The elution efficiency was calculated by comparing this value with the amount of metal ion eluted and measured. The elution of the metals from the copolymer hydrogel was carried out in 5.0 ml of 2.0 M HNO3 for 24 h. The elution solution was then diluted to an appropriate concentration with deionised water. The metal ion concentrations were determined using a flame atomic absorption spectrometer
61 Chapter 3
(FAAS). The metal concentrations of the various test solutions were also measured by
FAAS after preservation with HNO3 to pH < 2.
3.2.6. Validation of the PAM-PAA Hydrogel for Use with DGT
The applicability of the new binding phase for DGT analysis was validated according to the DGT equation (equation 1.6) 14, 21, 24.
The PAM-PAA hydrogel binding phase and the PAM hydrogel diffusive layer were placed, in a layered manner, into the traditional DGT device and covered with a 0.45 µm pore size nitrocellulose membrane 24. The DGT assemblies were deployed in duplicate for different periods ranging from 25 to 150 h, in a well-stirred solution of 0.75 µM Cd2+ in synthetic lake water (Windermere, Lake District, UK) 152. These experiments were set up so that the Cd2+ concentration was maintained at the initial value throughout the deployment period. The mass of Cd2+ in the binding gel and the concentration of Cd2+ in the solution before and after exposure to the DGT devices were measured according to the procedure described above.
3.3. RESULTS AND DISCUSSION
3.3.1. Preparation of the Poly(acrylamide-co-acrylic acid) Copolymer Hydrogel
The poly(acrylamide-co-acrylic acid) copolymer hydrogel (PAM-PAA) was prepared from the polyacrylamide (PAM) hydrogel via a hydrolysis reaction. The reaction was carried out in a strong alkaline solution to convert the amide groups on the PAM gel backbone into carboxylic groups. As the hydrolysis reaction proceeded, the PAM units were converted into polyacrylic acid (PAA) units and ammonia was produced. Under the
62 Chapter 3 experimental conditions used, the hydrolysis of the amide groups could not proceed to completion because the released ammonia remained within the sealed flask, but rather proceeded to equilibrium instead. Thus, the end product of the reaction was the PAM
PAA copolymer:
C OH O O O H2N C H2NC - m + n OH CH2 CH CH CH2 CH CH2 q xr
As the hydrolysis reaction proceeded, the gel was observed to swell and become fragile.
Further swelling of the PAM-PAA occurred when the hydrogel was hydrated in deionised water. As the swollen hydrogel had poor mechanical strength, placing the PAM-PAA hydrogels in a 0.1 M solution of HNO3 for 24 h reduced the swelling and improved the mechanical strength. Treating the hydrogel with the HNO3 solution also removed all the un-reacted alkali and products, such as ammonia. The hydrogels produced in this manner were found to be quite sticky, which made handling and deployment difficult. This stickiness was removed by storing the hydrogel in 0.1 M NaNO3 prior to use. After this pre-treatment the hydrogel was easy to handle at pH values appropriate for deployment in natural waters. These observations suggested that the cross-linker was not significantly degraded during the hydrolysis reaction.
3.3.2. Composition of the PAM-PAA Copolymer Hydrogel
The FTIR and elemental microanalysis were carried out to ensure that the reaction had proceeded as expected and to elucidate the composition of the new copolymer hydrogel.
The FTIR spectrum of the PAM hydrogel had absorption peaks at 3365.73, 1654.55,
1451.21, 1325.44 and 634.31 cm-1 (Figure 3.2), which was consistent with the standard infrared spectrum of polyacrylamide 200. The spectrum of the PAM-PAA copolymer 63 Chapter 3 showed peaks at 3401.41, 2756.22, 1702.51, 1247.41 and 634.50 cm-1. The absorption peak at 1247.41 cm-1 (Figure 3.3) suggested a syndiotactic-rich structure of PAM-PAA 201
203, while peaks indicative of atactic-rich 201 and isotactic forms 203 of polyacrylic acid were not observed. The band at 1702.51 cm-1 for PAM-PAA gel was due to the carbonyl component of the carboxylic acid group. The band for commercial polyacrylic acid occured at 1715 cm-1 201. The carbonyl absorption bands of the remaining amide groups overlapped with the carboxylic absorption bands. The broad absorption bands from
3401.41 cm-1 to 2956.22 cm-1 were assigned to the –OH from the carboxylic group 200.
Based on the above information, the general composition of the PAM-PAA copolymer was proposed to be:
O C OH H2N C O
CH2 CH CH CH2 q xr
98.7
80 1 3 2 5 . 4 4
60 1 4 5 1 . 2 1
% 6 3 4 . 3 1 T 40 3 3 6 5 . 7 3 1 6 5 4 . 5 5 20
4000 3000 2000 1500 1000 400 cm-1
Figure 3.2 FTIR Spectrum of PAM hydrogel recorded as a film. 64 Chapter 3
107.6 1 0 0
9 0
8 0 T% 7 0 2956.22 6 0 3401.41 1247.41 1702.51 5 0 4 0 0 0 3 0 0 0 2 0 0 0 1 5 0 0 1 0 0 0 6 0 0 cm-1
Figure 3.3 FTIR Spectrum of PAM-PAA hydrogel recorded as a film.
Table 3.1 shows the results of the elemental microanalysis. The molar ratio of nitrogen to carbon in the PAM gel (1:3.04) was close to that of the acrylamide monomer (H2C =
CHCONH2, 1:3). The nitrogen to carbon ratio decreased to 1:8.38, after the hydrolysis reaction, to form the PAM-PAA copolymer. This meant that the molar ratio of acrylamide to acrylic acid in the copolymer repeat units was 1:1.8, suggesting that most of the polymer is made of repeat units of acrylic acid units in a ratio of approximately 2:1 with the acrylamide units, i.e. q ≈ 2r.
Table 3.1 Elemental microanalysis of the PAM and PAM-PAA hydrogels
% N % C % H
PAM-PAA 3.99±0.05 28.65±0.09 5.17±0.08
PAM 16.1±0.03 42.0±0.05 7.22±0.09
65 Chapter 3
3.3.3. PAM-PAA Hydrogel Swelling Properties
The swelling properties of hydrogels are of interest in many applications 204, 205. The characteristic hydrogel swelling behaviour depends on the functional groups present. For a given hydrogel, the degree of swelling usually depends on the pH and ionic strength of the solution 199, 206-210. In using the PAM-PAA hydrogel as a binding phase for DGT applications, the swelling properties of the gel have to be characterised because the pore size of the gel network, and the volume of the hydrogel, depend upon its swelling properties. These properties, in turn, affect the ease of handling of the hydrogel and other practical considerations for its use with the DGT technique.
The effect of pH on the swelling properties of the PAM-PAA hydrogel is shown in Figure
3.4. The swelling mostly occurred from pH 3 to 6. At pH > 6 the change in the equilibrium swelling ratio with pH was slight. A fully hydrated gel disk was 120 times heavier than a dried gel disk for pH > 6. At pH < 3 the degree of swelling was very low.
When the hydration was carried out in a solution of pH < 3, the carboxylic acid groups on the copolymer backbone were converted to the protonated acid form. A low qw (Figure
3.4) indicated that the water content for the acid form of the hydrogel was low. When the solution pH was above 6, the carboxylic groups on the copolymer backbone were converted to the salt (basic) form and the maximum degree of swelling was achieved.
Within the pH range 3 to 6, an almost linear relationship between the swelling ratio and pH was observed. Within this pH range, the acid and salt forms of the carboxylic groups on the copolymer backbone were both present. The exact ratio of the acid and salt forms of the carboxylic groups was determined by the equilibrium pH under constant ionic
66 Chapter 3 strength. The presence of both acid and salt forms of the carboxylic groups on the copolymer backbone constituted a “hydrogel buffer” system. Under such conditions the
Henderson-Hasselbalch equation 3.2 applies:
[base form of carbox group]ylic pH = pK + log (3.2) a [acid form of carbox group]ylic
Since the ratio of the acid and base forms of the carboxylic acid groups were determined by pH, and directly proportional to the swelling ratio, the concentration of the acid and the base forms of the carboxylic acid groups were equal, at the point where the swelling ratio equals half of the maximum value. Figure 3.4 shows that, at this point, pH ≈ 4.5.
According to equation 3.2, the pKa for the PAM-PAA copolymer was, therefore, 4.5. This number agreed with a previously reported pKa value for a polyacrylamide-polyacrylic acid copolymer 210.
Figure 3.5 shows the dependence of the swelling of the PAM-PAA copolymer on the ionic strength at a given pH. When the pH of the solution was fixed at 7.0, an increase in the ionic strength resulted in a decrease in the swelling ratio. The most significant effect on the swelling ratio of the gel was observed when the NaNO3 concentration ranged between
10-3 M and 10-1 M. The ionic strength affected the swelling ratio by changing the charge distribution on the surface of the gel network. At high ionic strength, such as with the
NaNO3 concentrations above 0.1 M, the solution produced a strong “charge screening effect” on the hydrogel network, in which the electrostatic repulsion between adjacent strands of polymer were minimised, causing the strands to move closer together and the polymer to have less capacity to absorb water 211. As a result, the degree of hydration and swelling was reduced 199, 211.
67 Chapter 3
120 ) w q 80
40 Swelling Ratio (
0 024 6810 pH
Figure 3.4 Effect of pH on the equilibrium swelling ratio (qw) of the PAM-PAA gel (n = 3). The experiments were carried out at room temperature (23oC) and with a constant ionic concentration equivalent to 0.010 M NaCl.
For application as a binding gel with DGT, the swelling properties of PAM-PAA need to be minimised. The DGT devices with PAM-PAA would ideally be deployed in environments with a pH greater than 6 and with a fairly stable ionic strength. Fortunately, these conditions are met by most natural waters. The PAM-PAA hydrogels would also need to be pretreated by storing them in a solution with similar ionic strength to the environment in which it was to be deployed, taking care not to introduce contamination.
This procedure is recommended for the diffusion gels currently used with DGT 14.
68 Chapter 3
300
200 ) w q (
100 Swelling Ratio Swelling Ratio
0 -7 -5 -3 -1 1
Log [NaNO3]
Figure 3.5 Effect of ionic strength on equilibrium swelling ratio (qw) of the PAM PAA gel (n = 3). The experiments were carried out at room temperature (23oC) and at pH 7.0.
3.3.4. Metal Binding Properties of the PAM-PAA Hydrogel
The solution forms of poly(acrylic acid-co-polyacryamide) copolymers have been investigated previously for their metal retention properties 212. The binding properties of the PAM-PAA copolymer hydrogel for metal ions were investigated, initially, in a non competitive manner. Metal uptake curves for Cu2+ and Cd2+ are given in Figure 3.6. In both cases, a linear relationship between the metal ion uptake and the uptake time were observed initially, before saturation at about 1.59 µmoles cm-2 for Cu2+ and at about 1.56
µmoles cm-2 for Cd2+. The linear rate of uptake was greater for Cu2+ (0.088 µmoles cm-2 h-1) than for Cd2+ (0.057 µmoles cm-2 h-1).
69 Chapter 3
2
1.5 ) -2
1 mole cm µ ( take take
p 0.5 Cd U Cu
0 0 1020304050 Time (h)
Figure 3.6 Effect of immersion time on Cu2+ and Cd2+ ion uptake (n = 3). The experiments were carried out at room temperature (23oC) using gel disks with 4.9 cm2 surface area and 0.040 cm thickness.
Table 3.2 Non-competitive metal Ion binding capacity of PAM-PAA gel (n = 3)
Cu2+ Cd2+ K+ Na+ Ca2+ Mg2+
Binding Capacity 1.59 1.56 0.550 0.670 0.490 0.530 (µmole cm-2 )
Elution 1.51 1.50 0.530 0.730 0.470 0.530 (µmole cm-2 )
Elution Efficiency 95.0 96.2 98.0 109 95.9 100 (%)
The individual metal ion binding capacity in these sample solutions were calculated from the maximum metal ion uptake (saturation) and the results are summarised in Table 3.2.
The binding capacities observed were in the order of Cu2+ > Cd2+ >> Na+ > K+ > Mg2+ >
Ca2+. This order indicates that the affinity of the PAM-PAA hydrogel toward the binding
70 Chapter 3 of the transition metal ions, such as Cu2+ and Cd2+, was stronger than that towards alkali or alkaline earth metal ions.
From a mechanistic viewpoint, the order of increasing binding capacities does not exhibit classical ion exchange behaviour, where the affinity is proportional to the ionic charge only. The copolymer does, however, exhibit an order of affinity in accord with the increasing acidity and polarizability of the divalent metal ions measured, similar to that described by the Irving-Williams theory 213. The monovalent alkali metal ions have a higher capacity than the divalent alkaline earth metal ions due to the influence of the ion exchange behaviour and the greater concentrations required to balance the charge with the copolymer hydrogel. It is likely that the Cd2+ and Cu2+ also have increased affinity, due to the coordination (formation of inner sphere complexes) with the carboxylic acid groups
214. Previous studies 154 have shown that the amide group in polyacrylamide does not interact significantly with the metal ions. The amide groups may, however, play a secondary role in the complexation of the transition metals alongside the carboxylic acid groups.
The most common binding phase previously used with the DGT technique was Chelex
100 resin encapsulated in polyacrylamide. The reported Cd2+ binding capacity for such a binding phase was 1.1 µmole cm-2 16, while the maximum binding capacity of the PAM
PAA hydrogel obtained was 1.59 µmole cm-2 (Table 3.2). This capacity suggests that the
PAM-PAA hydrogel has a density of binding sites comparable with the Chelex-100 binding gel.
In general, the pH affects the binding capacity by shifting the equilibrium of the coordination reaction and/or ion exchange ability in two ways: changing the concentration
71 Chapter 3 of the active ligands and/or the concentration of the soluble metal ions. The effect of the solution pH on the binding capacity is shown in Figure 3.7. A very low binding capacity was observed when the solution pH was below 5. This decrease of capacity was because the protonated forms of the carboxylic acid groups are much less capable of forming complexes than the salt form 215. When the solution pH was greater than 5, the salt form of the carboxylic groups dominated and the maximum binding capacity was achieved.
The binding capacity decrease, at very high pH, was due to a significant change in the speciation of the metal ions from the free metal ion to the metal hydroxide, which is much less soluble 216. The optimum solution pH range for the PAM-PAA binding gels was greater than 5 on this basis, but greater than 6 based on the swelling dependence on pH.
This ideal pH range for metal ion uptake by PAM-PAA made it suitable for DGT applications in natural waters.
2 )
-2 Cd 1.6 Cu mole cm µ
( 1.2 y acit p 0.8 Ca g
0.4 Bindin
0 051 0 pH
Figure 3.7 Effect of pH on the binding capacity of the PAM-PAA gel for Cu2+ and Cd2+ (n = 3). The experiments were carried out at room temperature (23oC) using gel disks with 4.9 cm2 surface area and 0.04 cm thickness and an immersion time of 24 h. 72 Chapter 3
An important requirement for a binding phase with the DGT technique was the efficient elution of bound metal ions from the binding phase. The weak binding of the protonated hydrogel implies that the bound metal ions can be readily eluted from the gel in acidic media. The high efficiency of the acid elution is demonstrated by the recovery data for a range of metals in Table 3.2.
The effect of the ionic strength on the binding capacities of Cd2+ and Cu2+ is shown in
Figure 3.8. It was found that the binding capacities decreased almost linearly with a logarithmic increase in the NaNO3 concentration. This decrease of binding strength with increasing ionic strength demonstrates that the complexes formed between the transition metal ions and the PAM-PAA copolymer were weak. Therefore, acid-base interactions would be the dominant interaction type with ion exchange behaviour also important for the non-transition metal ions. The binding capacities, even at these high ionic strengths, were sufficient for the gel to be used to bind Cu2+ and Cd2+ from natural waters.
The binding capacities of the PAM-PAA copolymer hydrogel for metals under conditions of direct competitive uptake were also investigated (Table 3.3). The binding capacities for
Cu2+ and Cd2+ dropped from 1.59 µmole cm-2 and 1.56 µmole cm-2 for non-competitive binding to 0.923 µmole cm-2 and 0.574 µmole cm-2, respectively, with competitive binding. When the solution contains more than one type of metal ion, the binding capacity for each metal ion depends on its ability to compete with others for the available binding sites. This is a greater decrease than expected for an ionic strength effect (Figure 3.8), indicating a direct competition between the two transition metals (and, to a lesser extent, the alkali and alkali earth metal ions). The results in Table 3.3 suggests that the PAM
PAA hydrogel binds with Cu2+ more selectively than with Cd2+.
73 Chapter 3
2 )
-2 1.6
mole cm 1.2 µ
0.8
0.4 Cd
Binding Capacity ( Cu
0 -6 -4 -2 0
Log [NaNO3]
Figure 3.8 Effect of ionic strength on the binding capacity of the PAM-PAA gel for Cd2+ and Cu2+ (n = 3). The experiments were carried out at room temperature (23oC) using gel disks with 4.9 cm2 surface area and 0.04 cm thickness and an immersion time of 24 h.
Table 3.3 Competitive metal Ion binding capacity of PAM-PAA gel (n = 3)
Cu2+ Cd2+ K+ Na+ Ca2+ Mg2+
Binding Capacity 0.923 0.574 0.019 0.018 0.024 0.024 (µmole cm-2)
3.3.5. Application of the PAM-PAA Hydrogel as a Binding Phase with DGT
The use of the PAM-PAA hydrogel as a binding phase for DGT, was validated using the linear relationship between M and t, derived from the DGT equation 15, as shown in Figure
3.9. The mass of Cd2+ accumulated on the hydrogel increased linearly with time over a
74 Chapter 3 deployment period of up to 150 h with r2 = 0.994. The theoretical line is also shown in
Figure 3.9 and was calculated using the known parameters associated with the DGT equation. The line of best fit for the correlation indicated a recovery of 99.0% when compared with the theoretical line. A similar r2 value and recovery percentage was obtained with Cu2+ indicating that the PAM-PAA copolymer is suitable for use as a binding phase with DGT. For the AAS instrument, used to measure the concentration in the eluents in this experiment with detection limits of 1.8×10-6 mol l-1 for Cd, the detection limits for DGT after 100 h of deployments, were 1.5×10-7 mol l-1 for Cd when 5 ml solutions were used for the AAS analysis.
5
4 R2 = 0.9938
3 g) µ
2 Mass (
1
0 0 50 100 150 Time (h)
Figure 3.9 Mass vs. time validation of the PAM-PAA hydrogel for use with DGT for Cd2+. The experiments were carried out in a well-stirred solution made of the synthetic lake water matrix containing 0.75 µM Cd2+, with ∆g = 0.040 cm and A = 3.1 cm2. The dash line is the theoretical line calculated using standard DGT equation.
75 Chapter 3
6
5 R2 = 0.9798
4 g) µ 3
Mass ( Mass 2
1
0 0 50 100 150 200
Time (h)
Figure 3.10 Mass vs. time validation of the PAM-PAA hydrogel for use with DGT for Cu2+. The experiments were carried out in a well-stirred solution made of the synthetic lake water matrix containing 0.80 µM Cu2+, with ∆g = 0.040 cm and A = 3.1 cm2. The dash line is the theoretical line calculated using standard DGT equation.
3.4. CONCLUSIONS
A novel poly(acrylamide-co-acrylic acid) copolymer hydrogel (PAM-PAA) was found to be selective for the transition metals Cu2+ and Cd2+, over alkali and alkaline earth metals, and also suitable for use with the diffusive gradients in thin-films technique (DGT) used for the environmental analysis of most natural waters. The PAM-PAA hydrogel is a single-phase homogeneous binding gel which, therefore, overcomes many of the limitations associated with the conventional Chelex 100 resin based gel. The PAM-PAA hydrogel was prepared by the controlled hydrolysis of polyacrylamide hydrogel in alkali solution at mild conditions. The gel can potentially be mass produced due to the facile synthesis process. The FTIR and elemental analysis indicated that the composition of the 76 Chapter 3 copolymer was approximately 2 units of acrylic acid for every unit of acrylamide in a syndiotactic-rich structure (evenly distribution of functional groups). The equilibrium degree of swelling (qw) of the PAM-PAA hydrogel increased to 120 times that of the dried state at pH > 6, and was relatively stable at higher pHs. This swelling property limited its application in DGT. A dramatic increase or decrease of the gel volume may cause the breakage of upper layer in DGT devices or the incomplete coverage of the diffusive layer.
The pKa was found to be about 4.5. At a pH > 5, most carboxylic acid groups were in the salt form and had a greater capacity to bind the transition metals Cu2+ (1.59 µmoles cm-2) and Cd2+ (1.56 µmoles cm-2). The degree of swelling and the binding capacity were also influenced by the ionic strength. With these swelling properties the gels needed to be stored in appropriate conditions (NaNO3 solution) before use with DGT to avoid swelling and stickiness during deployment. The optimum deployment conditions were at pH > 6 under conditions of relatively stable ionic strength.
77 Chapter 4
Chapter 4 Preparation and Characterisation of a Poly(acrylamidoglycolic acid-co-acrylamide) Hydrogel as a New DGT Binding Phase for Determination of Trace Metals
78 Chapter 4
4.1. INTRODUCTION
The previous chapter describes the preparation of a new binding phase for the measurement of trace metals with DGT, where the polyacrylamide hydrogel was converted to a poly(acrylamide-co-acrylic acid) copolymer. Another strategy to produce copolymers with suitable functional groups is to use different starting monomers. Such copolymers can be synthesised by photo-induced polymerisation 217, electrolytic polymerisation 218, catalyst induced polymerisation 219, interpolymer complex 220 and the modification of natural products 221. Theoretically, any monomers with binding functional groups, such as amine 193, amidoxime 194, dithizone 177, carboxylic acid 195 and phosphorus
196, can be combined to form a copolymer suitable for use as a binding gel for DGT.
In this chapter the synthesis and characterisation of a poly(acrylamidoglycolic acid-co- acrylamide) (PAAG-PAM) copolymer hydrogel is described. This copolymer also contains carboxylic acid functional groups. The copolymer is characterised to determine its structure, metal binding properties under various conditions, and whether it is suitable for use as a binding phase for accumulating trace metals with DGT.
4.2. EXPERIMENTAL
4.2.1. Preparation of Poly(acrylamidoglycolic acid-co-acrylamide) Hydrogel
Poly(acrylamidoglycolic acid-co-acrylamide) (PAAG-PAM) hydrogel was prepared from a monomer solution containing 2.23 g of acrylamidoglycolic acid monohydrate (AAGA)
(Aldrich), added to 10.0 ml of deionised water (Milli-Q), followed the addition of 1.5 ml of 30% NaOH with constant stirring to bring the pH close to 7. The solution was then 79 Chapter 4 titrated with 1 M NaOH to pH = 7.0. The volume of the solution was made up to 15.0 ml with deionised water (Milli-Q) and thoroughly mixed. The PAAG-PAM hydrogel was prepared by adding 780 µl of 40% acrylamide (AAm), 800 µl of 2% agarose-derived cross-linker (DGT Research Ltd, Lancaster, UK), 70 µl of freshly made 10% (w/w) ammonium persulphate (Bio-Rad) solution and 30 µl of 99% N,N,N',N'- tetramethylethylenediamine (TEMED) (Bio-Rad) to the AAGA solution. This formula gave a molar ratio of 3:1 for AAGA and AAm, which was subsequently found to be the optimum ratio for handling purposes and for the metal binding capacity. The well-mixed solution was immediately pipetted into a suitable mould comprising two slightly offset, clean glass plates, separated by an inert plastic spacer of known thickness (0.040 cm) and held firmly together with clips. The mould was then incubated at 60 ± 2oC for 3 h. The set gel sheets were then hydrated in deionised water (Milli-Q) for 24 h, with the water being replenished several times. They were then soaked in 10% HNO3 for another 24 h
(to remove any metal contamination within the gel) prior to being washed again with the
2 water. The gel sheets were cut into disks of 4.9 cm and stored in 0.1 M NaNO3, which facilitated subsequent handling. This storage solution contained several grams of Chelex
20 resin to reduce trace metal impurities.
4.2.2. Characterisation of the PAAG-PAM Hydrogel
The FTIR spectra of the PAAG-PAM hydrogel were obtained using a Perkin Elmer Series
1000 FTIR spectrophotometer (see Chapter 2 for conditions used).
The elemental composition of the PAAG-PAM gel was determined by elemental microanalysis on a Carlo Erba 1106 Elemental Analyser (see Chapter 2 for conditions used). 80 Chapter 4
4.2.3. Swelling Properties of the PAAG-PAM Hydrogel
The hydrogel disks were soaked in solutions of differing pH at room temperature (23oC) for 24 h. The pH value of each solution was obtained by adding either hydrochloric acid or sodium hydroxide. After the hydrogel disks had been equilibrated in each solution, the degree of swelling was measured by accurately weighing each disk. For the measurement of the equilibrium swelling ratio, qw, the disks were weighed both in their hydrated and dried states. The dried gel disks were obtained by drying them in an oven at 40 °C to a constant weight over 48 hours. The equilibrium swelling ratio 199 was calculated from equation 3.1,
qw = ms / md where ms and md are the weights of the hydrogel disks in the swollen/hydrated state and dried state, respectively. The experiments on the swelling properties of the hydrogel were also carried out as a function of the electrolyte concentration, with NaNO3 ranging from
10 µM to 0.10 M.
4.2.4. Metal Binding Properties of the PAAG-PAM Hydrogel
The binding properties of the PAAG-PAM hydrogel for metal ions Cu2+, Cd2+, K+, Na+,
Ca2+ and Mg2+, were tested under non-competitive conditions by immersing a gel disk in a solution containing a metal ion concentration of 1.0 mM at pH 6 – 7, for several time periods up to 24 h. This experiment enabled the measurement of the binding capacity and also the rate of binding for each metal, provided that the saturation occurred within the 24 h period. The amount of metal ion bound in the gel was eluted by soaking the gel in 5.0 ml of 2 M HNO3 for 24 h before being diluted to an appropriate concentration with
81 Chapter 4 deionised water and measured with a flame atomic absorption spectrometer (FAAS). The metal concentrations of the tested solutions were also measured by FAAS before and after gel immersion.
To study the competitive binding of the gel, a disk was immersed in a stirred solution containing Cu2+, Cd2+, K+, Na+, Ca2+ and Mg2+, each at 17.8 µN concentration. Elution was performed as described above. The effects of various conditions on the gel binding for Cu2+ and Cd2+ were studied by changing the time of exposure (1 - 24 h), pH (from 0.2 to 12) and NaNO3 concentration (10 µM - 1 M).
4.2.5. DGT Performance
A validation test for the new DGT binding phase was undertaken according to procedures described previously 16, 147. The PAAG-PAM hydrogel binding phase, the PAM hydrogel diffusive layer (see Chapter 2 for preparation) and a wet 100 µm pore size cellulose nitrate membrane filter, as a protective layer, were placed, in that order, into a DGT assembly 16.
Eight DGT assemblies were exposed to a well-stirred solution of 0.79 µM of Cu2+ over various periods of time, up to 145 h, in synthetic lake water (Windermere, Lake District,
UK). A sufficient volume of solution was provided to ensure that the depletion of Cu2+ by the DGT assemblies was negligible. The mass of Cu2+ accumulated in the PAAG-PAM gel was measured by FAAS. The elution was carried out as described above.
The performance of the DGT measurement was evaluated in two ways. Firstly, from equation 3.1, a linear relationship was expected between measured M and deployment time, t, up to the capacity of the binding gel, if the DGT device was acting according to theory. Secondly, a line of best fit for M vs. t was compared with a theoretical line 82 Chapter 4 derived from the actual solution concentration, C and equation 3.1. A comparison of the gradients of these lines provided an estimation of the recovery of the measurement.
Recoveries of > 90% are desirable with DGT.
4.2.6. Preparation of Polyacrylamide Hydrogel
Polyacrylamide (PAM) hydrogels were prepared according to previously described procedures (Chapter 2) 153, 156.
4.3. RESULTS AND DISCUSSION
4.3.1. Structure and Composition of the PAAG-PAM Hydrogel
The PAAG-PAM hydrogel was prepared by copolymerising AAGA with AAm at a 3:1 molar ratio in the presence of the initiators ammonium persulphate and N,N,N',N'- tetramethylethylenediamine (TEMED) according to the following reaction:
OH OH TEMED HOOCCHNH C O CO NH2 (NH4)2S2O8 HOOCCHNH C O CO NH2 3 + o CH CH CH CH 60 C CH CH CH CH 2 2 2 a 2 b n The FTIR and elemental analysis were undertaken to determine the composition and structure (i.e. a and b values) of the copolymer gel. The FTIR spectrum of PAAG-PAM with the characteristic peaks is shown in Figure 4.1. The Peaks appear at 1651.67 cm-1
(C=O in the carboxylic and amide groups), 3443.37 cm-1 (NH and OH in carboxylic groups and amide groups), and 1096.28 cm-1 (the secondary alcohol groups), 2922.29 cm-1
(carboxylic groups) and 1384.07 cm-1 (aminoacid) 222. These data confirmed that the
PAAG-PAM copolymer hydrogel was formed with functional groups from both acrylamide and AAGA.
83 Chapter 4
65
62
58 % T 2352.64 650.96 54 2922.29 1096.28
50 3443.37 1651.67 1384.07
4000.0 3000 2000 15000 1000 400 cm-1
Figure 4.1 FTIR spectrum of PAAG-PAM hydrogel in KBr pellet with the main diagnostic peaks highlighted.
Table 4.1 Microelemental analysis results of PAAG-PAM C% H% N%
Experimental values 36.96±0.57 5.96±0.09 10.88±0.16
Theoretical* values 40.91 4.92 10.61
* Based on the monomer ratio of AAGA to AAm of 3:1.
The elemental analysis of PAAG-PAM gel was performed to determine the values of a and b. The elemental analysis indicated a C:N:H ratio of 4:1:0.5 (Table 4.1). The molar ratio of AAGA and AAm in the PAAG-PAM gel, calculated from the carbon and nitrogen stoichiometry, based on the data shown in Table 4.1, was approximately 3.53:1, or approximately 7:2. This result indicated that the copolymerisation reaction was not exactly complete, as some AAm monomers did not form part of the polymer and would
84 Chapter 4 have been removed during rinsing. The PAAG-PAM gel structure can therefore be written as follows:
OH
HOOCCHNH C O O C NH2
CH CH CH CH 2 7 2 2 n
4.3.2. Swelling Properties of the PAAG-PAM Gel
Since the PAAG-PAM gel had carboxylic acid groups on the network chain, the degree of swelling was expected to vary with the pH. Figure 4.2 shows the observed variation of the
o gel swelling ratio, qw, with the pH at 23 C. The qw was strongly dependent on the pH, due to the dissociation of the carboxylic group on the gel network. The highest degree of swelling was reached at around pH 5.4. A similar swelling phenomenon has been observed in other hydrogel work 207.
600
500 )
w 400 q
300
200 Swelling Ratio ( 100
0 04 81 2 pH
Figure 4.2 Effect of pH on the equilibrium swelling ratio, qw, of the PAAG-PAM hydrogel; temperature 23oC.
85 Chapter 4
As shown in Figure 4.3, the swelling ratio of the PAAG-PAM gel decreased as the NaNO3 concentration increased. The increase of NaNO3 concentration tends to screen the attraction between the polar water molecule and the polyelectrolyte gel, therefore, decreasing the water content of the hydrogel 211.
600
) 400 w q
200 Swelling Ratio (
0 -6 -5 -4 -3 -2 -1 0
Log [NaNO3]
Figure 4.3 Effect of ionic strength on the equilibrium swelling ratio, qw, of the PAAG PAM hydrogel; temperature 23oC, pH 7.0.
This dependence of the swelling of the hydrogel on the pH and ionic strength has implications for its use as a binding phase for DGT. The gel will thus need to be stored in a solution of similar pH and ionic strength to that in which the DGT assembly is to be deployed to ensure that the swelling does not interfere with the measurement. This swelling problem is also the case present with PAM/Chelex 100 resin-based binding gels.
86 Chapter 4
4.3.3. Metal Binding Properties of the PAAG-PAM Hydrogel
4.3.3.1. Non-competitive Binding Capacities
The binding properties of the PAAG-PAM hydrogel for metal ions were investigated initially in a non-competitive manner. The non-competitive binding capacity for each metal ion was calculated from the maximum metal ion uptake (saturation) within 24 h.
The results are summarised in Table 4.2. The binding capacities observed were in the order of Cu2+ ≈ Cd2+ >> Na+ ≈ Mg2+ ≈ K+ ≈ Ca2+. This order indicated that the affinity of the PAAG-PAM hydrogel towards the binding of the transition metal ions, such as Cu2+
(5.3 µmole cm-2) and Cd2+ (5.1 µmole cm-2), was stronger than that towards the alkali or alkaline earth metal ions. This difference was due to the coordination bonds that formed between these transition metals and the various ligands on the gel network 214.
Table 4.2 Non-competitive and competitive binding capacities (µmole cm-2) of various metals by PAAG-PAM hydrogel
Cu2+ Cd2+ K+ Na+ Ca2+ Mg2+
Non-competitive binding capacity 5.3 5.1 0.78 0.85 0.68 0.82
Competitive binding capacity 1.3 0.17 0.027 0.024 0.024 0.027
4.3.3.2. Competitive Binding Capacities
As there are many ions present in natural waters, it was necessary to collectively test the competitive binding of metal ions to the PAAG-PAM hydrogel before it was used for practical DGT analysis. Table 4.2 shows that the binding selectivity order was Cu2+ >>
87 Chapter 4
Cd2+ >> K+ ≈ Mg2+ ≈ Na+ ≈ Ca2+ when all the metal ions were present at the same normality. These results indicated a much higher selectivity of the gel for Cu2+ (1.3 µmole cm-2), compared to the other ions tested under competitive binding, including Cd2+ (0.17
µmole cm-2). This result was generally supported by the observations made of the change in the metal binding capacity with increasing ionic strength (section 4.3.3.5). However, it seems that the Cd2+ interaction is greatly reduced when Cu2+ is also present. The binding capacity for Cu2+ was comparable to the capacities reported for the Chelex 100 binding gel 16 previously used with the DGT technique.
4.3.3.3. Binding Rate
The metal uptake curves for the non-competitive binding of Cu2+ and Cd2+ are shown in
Figure 4.4. The initial concentrations of the metal ions in the aqueous phase were 1.0 mM. For both metals, a rapid initial rate of uptake was observed in the first two hours.
The linear rate of uptake during this time was 2.64 µmoles cm-2 h-1 for both the Cu2+ and
Cd2+. The binding capacities were effectively reached within 6 h for each metal ion. The binding rate observed here was approximately equivalent to the experimental data for the binding kinetics of heavy metal ions by various sorbent systems in membrane or microsphere forms (about 6 h) 216.
There were also several parameters which determined the binding rate, such as sorbent structural properties (e.g. size, porosity, surface area), amount of sorbent, metal ion properties (e.g. hydrated ionic radius), and initial concentration of metal ions 223. In the case of a single disk of gel immersed in a solution, the binding rate also depended upon how well the solution was stirred. However, the binding rate obtained with the PAAG
88 Chapter 4
PAM gel was deemed to be satisfactory for the application with DGT, as is confirmed in section 4.3.4.
6
5
) 4 -2
3 mole cm µ 2 Cu
Uptake ( 1 Cd
0 0 102 03 0 Time (h)
Figure 4.4 Uptake of Cu2+ and Cd2+ by the PAAG-PAM hydrogel at various times; temperature 23oC, pH 7.0. The metal uptake curves for the non-competitive binding of Cu2+ and Cd2+ are shown in Figure 4.4. The curve shown here represents one set of the experiments. The curves obtained from different experiments showed the same trend
4.3.3.4. Effect of pH on the Binding Capacity of Cu2+ and Cd2+
A change in the pH can influence the uptake of a metal by a complexing agent. In deed the pH influences the transition metal speciation and solubility and the charge of the binding functional groups 216, 224, 225. The proportion of the basic form of the glycolic acid groups, the main binding site, also increases with an increase of pH. Figure 4.5 shows that the binding capacity of ions first increased with increasing pH, due to a change in the ratio between the basic and acidic form of the glycolic acid groups. Interestingly the pH 89 Chapter 4 influenced the uptake of Cd2+ and Cu2+ in different ways, with the Cu2+ taken up at lower pH values (1.5-4.0) and the Cd2+ at slightly higher values (2.5-5.0). This binding preference is further evidence for the selectivity of the PAAG-PAM hydrogel for Cu2+.
The binding capacities increased slightly at pH values > 5, until the uptake decreased, due to metal hydroxide insolubility at pH > 9.
6 ) -2 4 mole cm µ
2
Uptake ( Uptake Cu Cd
0 0 510 pH
Figure 4.5 Effect of pH on the binding capacity of the PAAG-PAM hydrogel; temperature 23oC, time 24 h.
4.3.3.5. Effect of Electrolyte Concentration on the Binding Capacity of
Cu2+ and Cd2+
As natural waters have a range of ionic strengths, the binding behaviour of the PAAG
2+ 2+ PAM gel to Cu and Cd was studied in aqueous solutions with NaNO3 concentrations ranging from 10 µM to 0.10 M. Figure 4.6 shows that the binding of Cu2+ to the gel was slightly stronger than that of Cd2+ at all electrolyte concentrations. In both cases, as expected, the binding capacity decreased with an increase in ionic strength. Even at a
90 Chapter 4 concentration of 0.1 M NaNO3, the binding capacities obtained were still appropriate for
DGT applications.
6 ) -2 4 mole cm µ
2
Uptake ( Cu Cd
0 -6 -4 -2 0
Log [NaNO3]
Figure 4.6 Effect of electrolyte concentration on the binding capacity of the gel for Cu2+ and Cd2+. Temperature 23oC, pH 7.0, time 24 h.
4.3.4. Validation of Poly(AAGA-co AAm) as a Binding Phase for DGT Use
Figure 4.7 showed mass-time DGT curve. The data shown were values drawn from one set of experimental data. Determination of the slope of DGT curve in such way is highly reproducible and reliable The PAAG-PAM hydrogel was tested as a binding phase with
DGT for Cu2+, based on its selectivity for Cu2+ from the metal binding experiments above.
When the DGT assemblies were deployed, for time periods up to 150 h, the measured mass (M) of Cu2+ in the gel increased linearly (R2 = 0.975) with time (t).
91 Chapter 4
This linear relationship indicates that the PAAG-PAM binding phase was capable of reducing the Cu2+ ion concentration to zero at the interface between the binding and diffusive layers. In addition, the data agreed well with the theoretical line calculated from the DGT equation using the known concentration of Cu2+ in the experiment (0.79 µM in synthetic lake water (Windermere)). A recovery of close to 100% was measured this way, indicating that the diffusion coefficient (D) used was appropriate. For application in more complex systems, with a range of species present, for each analyte, and each with their own diffusion coefficient, the results can be interpreted as a flux or as an indicative concentration value. This interpretation is an area requiring additional research.
5
4 R2 = 0.9754
3 g) µ
2 Mass ( Mass
1
0 05 01 0015 0 Time (h)
Figure 4.7 Accumulation mass vs. time response of DGT uptake for Cu2+ ion and theoretical response calculated from standard DGT equation using the solution concentration and other known parameters; ∆g = 0.36 mm, [Cu2+] = 0.79 µM, A = 4.9 cm2, D = 2.2×10-6 cm2 s-1 at 23 oC.
92 Chapter 4
The two positive outcomes, above, confirm that the PAAG-PAM hydrogel is suitable for use as a binding phase for Cu2+ ions using the DGT technique. Indeed the results indicate that the analyte concentration on the interface between the diffusion gel and the binding gel phase was effectively reduced to zero during the deployment, a condition of applying the DGT equation. The fact that the gel binding functional groups were evenly distributed on the three dimensional hydrogel backbone, and the contact between the diffusion gels and the binding phase would have been close to an ideal two dimensional phase, also make it ideal for application to DGT.
4.4. CONCLUSIONS
A new copolymer PAAG-PAM was prepared with a 7:2 ratio of AAGA monomer units to
AAm monomer units. This polymer was found to bind Cu2+ ions selectively with a binding capacity of 5.3 µmole.cm-2 for non-competitive uptake and 1.30 µmole.cm-2 for competitive uptake with other metal ions. This binding capacity combined with rapid uptake kinetics made the polymer suitable for use as a binding phase with the DGT technique. This suitability was confirmed when a linear response was obtained for the accumulated mass vs. the uptake time of the Cu2+ and a 95 - 100% recovery with the DGT uptake experiment.
Similar swelling properties of this gel to PAM-PAA gel as described in Chapter 3 were observed. When the gel contains a high percentage of water, it becomes fragile. Thus the gel needs to be stored in a NaNO3 solution with similar ionic strength to the water solution in which the DGT devices are going to be deployed, to minimise the degree to which the gel size changes.
93 Chapter 5
Chapter 5 Application of a Cellulose Phosphate Ion Exchange Membrane as a Binding Phase in the Diffusive Gradients in Thin Films Technique
94 Chapter 5
5.1. INTRODUCTION
The development and evaluation of hydrogels, with homogeneous distribution of functional groups as binding phases for DGT, are described in the previous two chapters.
Although some advantages of the gels were identified, there are still some limitations in the use of these gel-based binding phases, due to the nature of the hydrogels. The gels can be fragile and there is a need for them to be stored in solutions, such as NaNO3, to minimise their tendency to swell. These features make handling the gels and the assembly of the DGT devices more difficult. In addition, the gel-based binding phases are not reusable, because the polyacrylamide degrades during the elution process.
In this chapter, a commercially available, solid-state ion-exchange membrane is proposed as an alternative to the binding phases currently used with the DGT for the measurement of trace metal species. There are many such membranes currently available, which made by the addition of electrophilic functional groups (such as phosphoric acid 226, carboxyl
227, amidoxime 228, hydroxamic acid 229 and sulphonate and triazine 175, 230), to a backbone membrane structure, such as cellulose. Cellulose phosphate membranes, in particular, have been used for the binding metal ions and for the separation of trace metals 231, 232.
This material has excellent ion exchange properties, combined with a desirable hydrophilic nature. The binding functional groups, which are chemically immobilised on the cellulose backbone, provide good chemical stability and uniformity of coverage on all surfaces of the membrane. The excellent mechanical strength and flexibility of the material also makes it convenient for the handling and the preparation of the DGT assembly. Furthermore, the ion-exchange properties of the membrane can be easily regenerated under acidic conditions to allow for the reuse of the material as a binding phase 233, 234 . 95 Chapter 5
A commercially available Whatman P81 cellulose phosphate ion exchange membrane
(P81) was selected as a test case to demonstrate the feasibility of the solid-state DGT binding phase concept. The binding properties of the Whatman P81 cellulose phosphate ion exchange membrane, for a range of metal ions under various conditions, were systematically investigated. The performance of this new solid-state binding phase for
DGT applications was also evaluated.
5.2. EXPERIMENTAL
5.2.1. Cellulose Phosphate Membrane Pre-treatment
The cellulose phosphate membranes (P81), with a 25 mm diameter and a 0.20 mm thickness, were purchased from Whatman International Ltd., UK and were used as the binding phase for DGT. The Whatman P81 membrane is a strong cation exchange membrane with a high ion exchange capacity of 18.0 µEq/cm2 (Whatman catalogue book).
The functional group responsible for binding metal ions is the ester-linked orthophosphoric acid group 233 with Na+ counterions. In order to minimise any trace metal ion contamination, the membranes used for all experiments were immersed in 10% HNO3 for 24 h before being thoroughly rinsed with, and stored in, deionised water (Milli-Q).
This process also served to pre-wet the membrane providing a desirable hydrophilic surface to facilitate the construction of the DGT assembly.
5.2.2. Preparation of the Polyacrylamide Hydrogel
Polyacrylamide (PAM) was employed as the hydrogel diffusive layer in the DGT assemblies and was prepared according to previously described procedures 147 (Chapter 2).
96 Chapter 5
5.2.3. Binding of Metal Ions to Cellulose Phosphate Membrane
The non-competitive binding capacities of a range of individual metal ions, including
Cu2+, Cd2+, Zn2+, Mn2+, Ni2+, K+, Na+, Ca2+ and Mg2+, were examined by immersing cellulose phosphate membranes in a solution containing 1.1 mN of metal ions at pH 7.0 for 24 h with a constant rate of stirring. The competitive binding capacities of the same ions were also measured by immersing the membranes in a solution containing 0.054 mN for each of the above metal ions for 24 h with a constant rate of stirring. In both cases, the metal ion concentrations of the solutions were measured against immersion time, using flame atomic absorption spectrometry (FAAS) (SpectrAA-200, Varian). The mass uptake by the membrane was also measured by FAAS after elution (as described below).
The effect of the initial metal ion concentration on the uptake was investigated by immersing membranes in solutions ranging from 0.018 to 1.1 mM for Cd2+, and from
0.016 to 0.94 mM for Cu2+, with an immersion time of 24 h with stirring. The influence of pH on the binding capacity was tested in the pH range of 0.5 to 12.1 for Cd2+ and 1.0 to
12.2 for Cu2+ respectively. The effect of ionic strength on the binding capacity was
-5 carried out in solutions containing various concentrations of NaNO3 ranging from 1.0×10 to 1.0 M.
5.2.4. Elution and Analysis of Metal Ions
The elution of the metals from the cellulose phosphate was carried out in 5.0 ml of 2.0 M
HNO3 for 24 h, after which the elution solution was diluted to an appropriate concentration with deionised water (Milli-Q). The metal ion concentrations were
97 Chapter 5 determined by FAAS, according to the guidelines of the manufacturer. The metal concentrations of the various test solutions were also measured by FAAS after preservation with HNO3 to pH < 2.
5.2.5. Assembly of DGT Devices
The DGT devices were assembled according to previously reported procedures 14, 16 with cellulose phosphate membranes (P81) used as the binding phase (Chapter 2). The DGT device consists of a backing support and a front cap with a 2.0 cm diameter window. The binding phase was placed on the support (it does not matter which side is facing up) and the polyacrylamide diffusive layer was placed on top of the membrane followed by the cellulose nitrate covering layer. The DGT device was then sealed.
5.2.6. DGT Validation Experiments
From the DGT equation 3.1, a linear relationship between M and t was expected for any given solution, with Cb and the other parameters kept constant. Confirmation of this relationship has been used as a test for the feasibility of the new DGT devices. Indeed the
DGT devices were deployed in well-stirred sample solutions containing 0.80 µM Cu2+ or
0.45 µM Cd2+. The matrix for these experiments was synthetic lake water (Windermere,
Lake District, UK) (Chapter 2) 152. The masses of Cu2+ or Cd2+, accumulated by the binding phase, were measured after elution, according to the procedure described above.
The concentrations of Cu2+ or Cd2+ in the deployment solution, before and after exposure to the DGT devices, were also measured.
98 Chapter 5
5.2.7. Reuse of Binding Phase
The DGT validation experiments for the Cu2+ and Cd2+ were repeated with the binding phases re-used up to four times. After each DGT experiment was completed, the binding membrane was washed thoroughly with deionised water and then immersed in 2.0 M
HNO3 for 24 h to remove the un-eluted metal ions (if any) from the previous experiment.
The membrane was then soaked in deionised water for 24 h, with three changes of water prior to reuse.
5.3. RESULTS AND DISCUSSION
5.3.1. Metal Ion Binding Properties
The metal ion uptake vs. time curves for Cd2+, Cu2+, Na+ and Mg2+ in Figure 5.1 show that the amount of metal ion uptake increased rapidly, initially, before levelling off, indicating that saturation had occurred. The saturation time observed was approximately 8 h for all four metal ions shown in Figure 5.1. Similar behaviour was also observed for other metal ions tested (i.e. Zn2+, Mn2+, Ni2+, K+ and Ca2+), with the maximum metal uptake values
(binding capacities) for all metals being derived from these curves (Table 5.1). These binding capacities are comparable to the binding capacities of the Chelex 100-impregnated polyacrylamide gel 16 widely used for metal measurements with DGT 14, 21, 22, 24. Indeed the binding capacities for the metal ions investigated were, in order of decreasing strength,
Zn2+ > Cu2+ > Cd2+ > Mn2+ > Ni2+ > Mg2+ > Ca2+ > K+ > Na+, indicating that the binding capacities of the Whatman P81 membrane, for transition metal ions, were higher than that for the alkali earth metal ions. It should also be noted that the binding capacities of all the doubly charged metal ions were greater than were those of the singly charged ions, such as
K+ and Na+. This result is particularly interesting since the binding capacity of an ion
99 Chapter 5 exchange membrane is usually related to the number of available ion-exchange sites. For this reason, the ion-exchange capacity of the membrane for a singly charged species should be higher than that of a doubly charged species. The capacity trend obtained here may be due to the fact that the affinity between the doubly charged ions and the membrane is better than the singly charged alkali ions. In other words, the membrane binds the doubly charged ions stronger than does the singly charged alkali ions. This preference was demonstrated when the competitive ions were introduced.
4
3 ) -2
2 mole cm µ
1
Uptake ( Uptake Cd Cu Na Mg 0 0 102 03 0 Time (h)
Figure 5.1 Effect of immersion time on the metal ion uptake by the Whatman P81 membrane binding phase at pH 7.0 for Cd2+, Cd2+, Na+ and Mg2+. Concentration of metal ions: 1.1 mN; Temperature: 23oC.
Table 5.2 shows the binding capacities obtained when all of the test ions were present in the solution with the same concentration of 0.054 mN. Despite the fact that the binding capacities of all ions were decreased under the competitive conditions, the selectivity between the transition metal and the alkali metal ions, or between the alkali earth and the alkali metal ions indicate that the binding strength was in the order of: transition metal 100 Chapter 5 ions > alkali earth ions > alkali ions. This order becomes more obvious with a comparison of the capacity ratios shown in Table 5.1 with the selectivity data shown in Table 5.2.
Table 5.1 Binding capacity of P81membrane to various ions Cu2+ Cd2+ Zn2+ Mn2+ Ni2+ Ca2+ Mg2+ K+ Na+
Uptake 3.22 3.07 4.21 2.73 2.58 1.75 1.93 1.37 1.11 (µmole cm -2)
Capacity Ratio 1.00 0.95 1.3 0.85 0.80 0.54 0.60 0.43 0.35
Note: Capacity ratio was calculated based on the uptake value of Cu.
Table 5.2 The binding capacity of the Whatman P81 membrane for various metal ions under the competitive conditions
Cu2+ Cd2+ Zn2+ Mn2+ Ni2+ Ca2+ Mg2+ K+ Na+
Capacity 0.88 0.88 0.86 0.82 0.35 0.20 0.20 0.071 0.069 (µmole cm-2 )
Selectivity 1.0 1.0 0.98 0.93 0.40 0.23 0.23 0.081 0.078
Note: Selectivity was calculated based on the binding capacity of Cu.
The order of the selectivity and binding capacities suggests that the coordination (inner sphere complexation) bonding between the transition metals and the membrane was the dominant interaction for the ion exchange membrane. Further, the order of selectivity for the transition metal ions did not conform exactly to the hard-soft acid-base theory or the
Irving-Williams series, suggesting a number of parameters were important, including the morphologies of the gel 235. All the doubly charged metal ions had greater selectivity than the singly charged metal ions (i.e. K+ and Na+) indicating that the classical ion exchange interactions, where a charge was the most important factor, were not significant 235.
101 Chapter 5
The uptake of Cu2+ and Cd2+ ions were also investigated further. Figure 5.2 shows the effect of the concentration on the uptake of Cu2+ and Cd2+. With a deployment time of 24 h, the amount of metal ion uptake increased, with an increase in concentration up to 0.50 mM. Above this concentration, the metal ion uptake levelled off and was almost independent of the metal ion concentration, indicating the maximum binding capacity of the membrane.
4 ) -2 3 mole cm µ 2
Capacity ( Capacity 1 Cd Cu
0 0 0.25 0.5 0.75 1 1.25 Concentration (mM)
Figure 5.2 Effect of Cd2+ and Cu2+ concentration on the binding capacity of the Whatman P81 membrane at pH 7.0. Deployment time: 24 h; Temperature: 23oC.
The pH of the solution also appeared to affect the chemical forms of both the test metal ions and the functional groups of the ion exchange membrane. Indeed the phosphate functional group can have a charge of 0, -1, -2 or -3, or some fractional value in between, depending upon the solution pH. Figure 5.3 shows that at very low pH (pH < 1.8 for Cu2+ and pH < 1.0 for Cd2+), the membrane lost its ability to bind these metal ions. This decrease of capacity was because at such a low pH the orthophosphoric acid (the major
102 Chapter 5 binding functional group of the membrane) existed predominantly in its acid form, which is less capable of binding metal ions. As the pH increased, the binding capacity increased and then saturated when the pH was greater than 4.0. This increase of capacity was due to the increase in the basic forms of orthophosphoric acid in the membrane. This is probably the reason behind the preference for binding doubly charged metal ions (Table 5.1) by complexation. A further increase to pH > 9 resulted in a decrease in the binding capacity for both Cd2+ and Cu2+. This decrease of capacity was due to significant changes in the speciation of the metal ions from the free metal ions to the metal hydroxide species, which are much less soluble 216. The optimum solution pH range of the Whatman P81 membrane for binding metal ions was 4.0 < pH < 9.0, making it suitable for DGT applications in natural waters.
3.5
) 2.5 -2
mole cm 1.5 µ
0.5 Capacity ( Capacity Cd Cu -0.5 0 2 4 681012 pH
Figure 5.3 Effect of pH on the binding capacity of the Whatman P81 membrane for Cd2+ and Cu2+. Concentration of metal ion: 1.1 mN; Deployment time: 24 h; Temperature: 23oC.
103 Chapter 5
4
) 3 -2
mole cm 2 µ (
1 Cd Capacity Cu
0 -6 -5 -4 -3 -2 -1 0 1
Log [NaNO3]
Figure 5.4 Effect of ionic strength on the binding capacity of the Whatman P81 membrane for Cd2+ and Cu2+ at pH 7.0. Concentration of metal ion: 1.1 mN; Deployment time: 24 h; Temperature: 23oC.
As natural waters have varying concentrations of major ions, the effect of the ionic strength on the binding capacity was investigated (Figure 5.4). The ionic strength of the
-5 solution was adjusted using NaNO3, with the concentrations ranging from 1.0 × 10 M to
1.0 M. The binding capacities of Cd2+ and Cu2+ decreased as the concentration of the
-4 NaNO3 increased. When the concentration of the NaNO3 was below 10 M, the effect of
-2 the ionic strength on the binding capacity was slight. At 10 M NaNO3, the binding capacities of Cd2+ and Cu2+ decreased to 2.52 and 2.76 µmole cm-2, respectively. At
2+ 2+ NaNO3 concentrations of 0.5 M, the binding capacities for Cu and Cd were 1.1 µmole cm-2 and 1.90 µmole cm-2, respectively. These capacity values were still adequate for the
DGT applications in natural waters, given the low analyte concentrations typically present.
The binding capacity for Cu2+ decreased to 0.31 µmole cm-2 at a very high ionic strength
(1.0 M), but more than 60% of its maximum binding capacity for Cd2+ (1.8 µmole cm-2)
104 Chapter 5 was maintained. This result suggests that the application of the binding phase in DGT for the measurement of Cd in seawater would be feasible. Consequently, the measurement of copper and other elements in seawater should be further investigated.
5.3.2. Elution and Regeneration
Elution is an important step in the DGT measurement. In order to ensure the accuracy of the measurement, high elution efficiency is required for the removal of analyte ions from the binding phase. As demonstrated in Figure 5.3, the Whatman P81 membrane lost all its binding capability at pH < 2. This property was utilised to elute the metal ions from the membrane in a 2.0 M HNO3 solution, with high elution efficiencies being obtained under these conditions (Table 5.3).
In order to evaluate whether the membrane binding phase was reusable, the effect of the membrane regeneration (Section 5.2.8) on the binding capacity was investigated (Figure
5.5). It was found that less than 15% capacity was lost for both Cu2+ and Cd2+ after five successive uses. The relatively high binding capacity observed after five successive uses was considerably above the binding capacity required in most DGT applications which, despite long deployment times, typically involve the determination of very low metal ion concentrations 14. These data therefore suggest the possibility of the reuse of the binding phase in DGT applications (Section 5.3.3).
Table 5.3 Elution efficiency of various metal ions
Cu2+ Cd2+ Zn2+ Mn2+ Ni2+ Ca2+ Mg2+ K+ Na+
Elution efficiency 97.8 101 97.9 100 101 97.1 96.9 98.5 97.3 (%)
105 Chapter 5
3.5 Cd 3 Cu
) 2.5 -2
2 mole cm µ 1.5
1 Capacity ( Capacity 0.5
0 1 2 3 4 5 Number of Uses
Figure 5.5 Effect of consecutive membrane regeneration on the binding capacity of Cd2+ and Cu2+ at pH 7.0. Concentration of metal ion: 1.1 mN; Deployment time: 24 h; Temperature: 23oC.
5.3.3. Evaluation for Use as a Binding Phase with DGT
The use of the cellulose phosphate membrane as a DGT binding phase was evaluated using Cu2+ and Cd2+ as the test species in a synthetic lake (Windermere, UK) water 152.
Figure 5.6 shows the relationship between the mass of metal ions accumulated by the binding phase (M) and the deployment time (t). Significant coefficients of the determination, r2, were obtained from the regression lines for Cu2+ (r2 = 1.00, p = 0.000) and Cd2+ (r2 = 0.985, p = 0.000). These results strongly validate the use of the Whatman
P81 membrane as a binding phase for DGT. Indeed the regression (solid) lines also indicate high recoveries (103% for Cu2+ and 97% for Cd2+) when compared with the theoretical (dashed) lines calculated from the DGT equation using the known experimental parameters.
106 Chapter 5
6
4 g) µ (
2 Mass Cd Mass
0 0 50 100 150 200
(a)
6
4 g) µ
2 Mass Cu ( Cu Mass
0 050 100 150 200
Time (h) (b) Figure 5.6 Accumulated mass vs. deployment time curves for (a) Cd2+ and (b) Cu2+ in a well-stirred synthetic lake water (Windermere) at 23oC. The solid and the dashed lines represent respectively the regression line for the experimental data and the theoretical line estimated from the following parameters: Cb = 0.45 µmol/L of Cd2+ or 0.80 µmol/L of Cu2+; ∆g = 0.040 cm; D (Cd) = 2.1×10-6 cm2 s-1; D (Cu) = 2.2×10-6 cm2 s-1; A = 3.14 cm2.
107 Chapter 5
The laboratory conditions, where free ionic species dominate, are also a desirable condition for the evaluation of DGT. The reproducibility was tested with nine replicate measurements (Table 5.4). A relative standard deviation (rsd) of 4.9% and an average recovery of 102% were found for Cu2+ and an rsd of 5.5% and an average recovery of
104% were found for Cd2+. The slightly high recoveries are likely due to a small error in the diffusion coefficients estimated for the calculations. Nevertheless these results confirm that the membrane meets the requirements of a DGT binding phase.
Table 5.4 Reproducibility and recovery data from DGT experiments Experimental Trial 1# 2# 3# 4# 5# 6# 7# 8# 9#
2+ Cu 0.88 0.84 0.82 0.79 0.84 0.76 0.80 0.82 0.80 (µM)
Average Concentration 0.82 (µM)
RSD 4.9%
Recovery 110 105 103 99 105 95 100 103 100 (%) 2+ Cd 0.46 0.51 0.50 0.48 0.47 0.47 0.47 0.46 0.43 (µM)
Average Concentration 0.47 (µM)
Relative Standard 5.5% Deviation
Recovery 102 113 111 107 104 104 104 102 96 (%) Note: The actual concentration (added concentration) was 0.80 µM for Cu2+ and 0.45 µM for Cd2+. The experiments were conducted at room temperature of 23°C.
The DGT performance, with the regenerated binding phases, was also investigated.
Figure 5.7 shows the M vs. t relationship for five consecutive uses of the binding phases at different deployment times for the measurements of Cu2+ and Cd2+. It is readily apparent that there was little or no degradation of the performance. The r2 values for Cu2+ were 108 Chapter 5 between 0.976 - 1.00 and for Cd2+ were between 0.979 - 0.993, all were significant (p <
0.05). A comparison of the regression with the theoretical lines gave average recoveries of about 101% for Cu2+ and 98% for Cd2+. Therefore, the capability to reuse the Whatman
P81 membrane binding phases for at least four deployments was established. This reuse of a DGT binding phase had not been reported before and this finding may lead to a reduction in the costs of the measurements.
Other advantages of the cellulose phosphate membrane were observed during this study.
The binding phase was easy to use even by inexperienced analysts, overcoming some of the problems highlighted in Chapter 1. The surface coverage of the orthophosphate binding groups appears to be both even and high, probably due to the chemical preparation of the membrane 233, when compared with the physical process used with DGT to date.
Consequently, analysts can obtain very good reproducibility of measurements (≤ 5%).
Additionally, there is no chance of gross errors of the sort outlined in Chapter 1 because both surfaces of the membrane contain the binding functional groups. Further, the membrane is mechanically more robust than the hydrogel based binding phases previously used in DGT, making the handling and the assembly of DGT devices much easier.
Apart from its application for routine monitoring, this new binding phase will be a great benefit to the application of DGT in sediment, especially where measurements are made at high resolution 21, 151 and the binding phase handling becomes very problematical.
109 Chapter 5
6
5
4 g) µ 3 5th 4th 2 Mass Cd ( Cd Mass 3rd 2nd 1 1st
0 0 50 100 150 200 Time (h) (a)
7
6
5
g) 4 µ 5th 3 4th 2 3rd Mass Cu ( Cu Mass 2nd 1 1st
0 0 50 100 150 200 Time (h) (b)
Figure 5.7 Accumulated mass vs. deployment time curves with consecutively regenerated cellulose phosphate as the binding phase for (a) Cd2+ and (b) Cu2+ in a well- stirred synthetic lake water (Windermere) at 23oC. The solid lines represent the regression the theoretical lines estimated from the following parameters: 2+ 2+ Cb = 0.45 µmol/L of Cd or 0.80 µmol/L of Cu ; ∆g = 0.040 cm; D (Cd) = 2.1×10-6 cm2 s-1; D (Cu) = 2.2×10-6 cm2 s-1; A = 3.14 cm2.
110 Chapter 5
5.4. CONCLUSIONS
The Whatman P81 cellulose phosphate ion exchange membrane has been successfully used as the binding phase for DGT applications. The performance of this new DGT binding phase was demonstrated in the determination of Cu2+ and Cd2+ in a synthetic lake water matrix, with high recovery. The ion exchange activity of the new binding phase can be regenerated and, therefore, reuse of the binding phase in DGT applications is possible.
This reuse of the binding phase will lower the cost of DGT applications. The polyacrylamide gel is expensive to make, due to the expense of the agarose derived cross linker. The new binding phase exhibited excellent mechanical properties and overcame many of the problems of the hydrogel based binding phases. Those problems include fragility, caused handling difficulty; swelling or shrinking with the changes of pH or ionic strength conditions, caused breakage of the diffusive gel or the incomplete coverage of upper layers.
Perhaps the most significant aspect of this work is that it opens up the possibility of employing a new range of binding phases in DGT analysis, i.e. binding phases not limited to gel-based systems. There are also a myriad of other solid ion exchange membranes and other binding materials available. This work has shown the feasibility of employing such materials in the DGT technique. Minor limitations of this binding membrane may include: limited capacity in some applications; the roughness of the solid surfaces making an impact contact between the surfaces; and the need for elution procedures. In the next chapter, the deployment of a liquid binding phase is introduced, which overcomes these limitations.
111 Chapter 6
Chapter 6 Development of a New Generation DGT Device Using a Solid Membrane Diffusive Layer with a Liquid Binding Phase
112 Chapter 6
6.1. INTRODUCTION
This chapter describes the development of a binding phase, for use with DGT to measure trace metals, according to the fourth strategy outlined in Section 3.1: the development of a solution binding phase. Most binding phases described previously have used particles of a solid binding material dispersed throughout a polyacrylamide hydrogel to accumulate the analyte. In previous chapters (Chapters 3 and 4) the use of functionalised hydrogel copolymers were described while Chapter 5 demonstrated the use of a solid state membrane binding phase. Each of these binding phases was able to be used with the usual polyacrylamide diffusive layer 15. However, the use of a solution binding phase has required the development of a new diffusive layer.
A possible combination is the use of the water soluble poly(4-styrenesulfonate) (PSS) as the binding material and a dialysis membrane as the diffusive layer. These materials have been described previously for use in ultrafiltration to remove heavy metal ions from aqueous solutions 236, 237 and the separation of ionic metal species from metals associated with colloids 238. A dialysis membrane is capable of retaining the soluble PSS polymer as well as metals complexed to the PSS. Free metal ions, simple inorganic complexes and small complexes with soluble organic matter are able to pass through the dialysis membrane 239, 240 . These factors suggest that these materials would be able to be combined to form a new type of DGT device.
The metal binding properties of the PSS solution were investigated as functions of ionic strength, pH and electrolyte concentration. The stability constant for the binding of Cu2+ and Cd2+ was measured to determine the binding mechanism. The diffusional properties of the dialysis membrane were also investigated, particularly with respect to the diffusion 113 Chapter 6 coefficients of metal ions within the membrane under various conditions. The new DGT device was validated for use in DGT applications with particular emphasis on ensuring that the assumptions of the DGT equation (as described in Chapter 1) were met.
6.2. EXPERIMENTAL
6.2.1. The DGT Device Using a Solution Binding Phase
Figure 6.1 is a cross-sectional schematic of the DGT device developed for use with a solution-based binding phase. It consists of a polyethylene tube (1.4 cm diameter) containing 2.0 ml of 0.020 M PSS solution. The tube is covered and clamped with a 5.0 cm diameter pretreated dialysis membrane acting as the diffusion layer. The device is deployed with the dialysis membrane down to ensure appropriate contact with the binding solution. Sections 6.2.2 and 6.2.4 describe the preparation and pre-treatment of the PSS solution and the dialysis membrane.
Polypropylene Tube
PSS Aqueous Solution
Rubber Gasket
Clamp (Perspex)
Dialysis Membrane
Figure 6.1 Schematic representation of the new DGT device for use with the PSS solution binding phase.
114 Chapter 6
6.2.2. Preparation of the Dialysis Membrane
Cellulose acetate dialysis membranes (Sigma, Mw ca.12,000 or greater retain) were pre treated by soaking in deionised water (Milli-Q) for 4 h to remove glycerin. They were then washed with a 0.3% (w/v) solution of sodium sulfide (Sigma, Analytical grade,
98.0%) at 80oC for one minute, and washed again with hot deionised water (60oC) for 2 minutes; this process was followed by acidification with a 0.2% (v/v) solution of sulfuric acid (BDH, Analytical grade, 98.0%) to remove the sulfur compounds. After a final rinse with hot deionised water to remove the acid, the membranes were stored in deionised water. After this process, the thickness of the dialysis membrane increased reproducibly from 20 µm to 50 µm, as measured by an optical microscope.
6.2.3. Interaction of Cd2+ and Cu2+ with the Cellulose Dialysis Membrane
When considering whether the cellulose dialysis membrane was an appropriate material for use as a diffusive layer in DGT, there was a need to ensure that there was a minimal interaction between the ions of interest and the membrane itself. The interaction experiment was carried out according to the following procedures: a 100 cm2 sheet of 50
µm thick pre-treated cellulose dialysis membrane was exposed to a stirred solution containing 2.0 µM Cd2+ and Cu2+ for 24 h to accumulate the metal ions. The membrane was then equilibrated with 1.0 ml of 1.0 M HNO3 (Suprapur, Merck) for 12 h (under stirring) to elute the accumulated metal ions into solution. The concentration of the metal ions in the elution solution was determined by FAAS. The concentrations of Cd2+ and
Cu2+ in the membrane were calculated, based on the metal ion concentration in the elution solution. The experiments were performed in solutions of varying ionic strengths ranging from 10 µM to 1.0 M NaNO3.
115 Chapter 6
6.2.4. Purification of Poly(4-styrenesulfonate)
The poly(4-styrenesulfonate) (PSS) employed had an average Mw ca. 70,000 (Aldrich).
51.5 g of PSS was dissolved in 200 mL deionised water by ultrasonication. The solution was then transferred into a cellulose acetate dialysis bag (prepared as described in Section
6.2.2) and placed in deionised water for 72 h with the water frequently replenished. This process effectively removed all of the low molecular weight PSS that could pass through the dialysis membrane. The dialysed PSS was then filtered through a 0.45 µm pore size cellulose nitrate filter membrane to remove any undissolved particles. After purification the concentration of PSS was determined gravimetrically and a PSS stock solution of 0.50
M (concentration of sulfonate groups) was prepared.
6.2.5. Determination of Metal-PSS Concentrations
The concentrations of the metal and metal PSS complex solutions (after dilution) were determined by flame atomic absorption spectrometry (FAAS) (SpectrAA-200, Varian), after appropriate dilutions. The calibration solutions for the measurement of the metal-
PSS complexes were matrix-matched to the dilution factor 192. The method detection limits were 5.3 × 10-7 mole l-1 for Cu and 8.1 × 10-8 mole l-1 for Cd.
For the FAAS instrument used to measure the metal concentrations in the PSS solution the detection limit for DGT after 100 h of deployments was 3.1×10-9 mol l-1 for Cu and
9.0×10-10 mol l-1 for Cd when 5 ml PSS solutions were used for the FAAS analysis.
6.2.6. Optimisation of PSS Solution Concentration
The PSS concentration used in the binding phase was optimized using the same diffusion cell as that used for the measurement of the diffusion coefficients. The experimental 116 Chapter 6 procedures are the same as detailed in section 2.3.5. Compartment B initially contained
50.0 ml of PSS solution, with concentrations varying from 0.0050 to 0.050 M, in synthetic
Windermere Lake water matrix. Compartment A contained 50.0 ml of 10 ppm Cd2+ or
Cu2+ in synthetic Windermere Lake water solution. The mass (M) of the metal that diffused through the membrane of the exposed area (A), from compartment A to compartment B, was measured by FAAS, being sampled at certain time intervals (t) from
M both compartments. The fluxes (J) were calculated according to the equation J = 23. At
Masses were calculated from concentrations measured according to Section 6.2.5.
6.2.7. Metal Binding Properties of the Poly(4-styrenesulfonate) Solution
The metal binding properties of PSS were investigated using the new DGT device shown in Figure 6.1. A concentration series experiment was undertaken to determine that the binding phase reached its capacity within 24 h. Concentrations above 2.0 mM were required for both Cu2+ and Cd2+. Non-competitive binding experiments were carried out in 2.5 meq l-1 solutions of Cd2+, Cu2+, Ca2+, Mg2+, K+ or Na+ (individually) at pH 7.0 for
24 h. All the salts used were of an analytical grade and were supplied by Sigma.
Competitive binding studies were undertaken in a solution containing all of the above metal ions at 18 µeq l-1 concentration at pH = 7.0. The effects of the varying pH (0.2 to
2+ 11) and electrolyte concentration (10 µM to 1 M as NaNO3), on the binding of Cd and
Cu2+ to 0.020 M PSS, were studied.
6.2.8. Determination of Stability Constant
The stability constants of the complexation reaction between PSS and Cd2+ and Cu2+ were also estimated using the modified DGT devices. 2.0 mL of 0.10 mM PSS solutions were placed within the DGT device and suspended in solutions of 0.0010, 0.0020, 0.0050, 117 Chapter 6
2+ 2+ 0.010, 0.020, 0.040 mM Cd or Cu solutions in 1 mM NaNO3 for 48 h to allow the devices to reach equilibrium. The metal concentrations in the PSS solution, and those remaining in the original solution, were measured. The stability constants and the coordination numbers were calculated by the ultrafiltration approach, according to procedures used by Juang et al. 241 and Samadfam et al. 242.
6.2.9. Measurement of Metal Diffusion Coefficients in the Dialysis Membrane
2+ 2+ The diffusion coefficients, Dm, of Cd and Cu ions in the dialysis membrane were determined using the specially designed diffusion cell with a disc of pretreated cellulose dialysis membrane fitted between the compartments (Chapter 2, Figure 2.2).
Compartment B of the diffusion cell was filled with 50.0 ml receiving solution containing
0.020 M PSS in synthetic Windermere Lake water matrix. Compartment A of the diffusion cell was filled with 50.0 ml of source solution containing 10.0 ppm Cd2+ or Cu2+ in synthetic Windermere Lake water matrix. The samples were taken from both the compartments and were measured by FAAS (Section 6.2.5). The diffusion coefficients were to be calculated according to the method described in Chapter 2.
6.2.10. Effect of Stirring Conditions on the DBL Layer
The diffusion cell used was the same as that used for the determination of the diffusion coefficients. Compartment B of the cell was filled with 65.0 ml receiving solution containing 0.020 M PSS and 0.10 M NaNO3. Compartment A of the diffusion cell was
2+ filled with 65.0 ml source solution containing 10.0 ppm Cd and 0.10 M NaNO3. Both compartments were stirred using the same over head motor with the same speed (Figure
6.2). The samples were taken from both compartments within a 3 h time interval and the
Cd2+ concentrations were measured by FAAS (Section 6.2.5). 118 Chapter 6
+ -
φ =1.5 mm 12 mm φ =8 mm
Figure 6.2 Schematic diagram of the over head motor used.
6.2.11. Validation of the New DGT Device
The DGT devices (Figure 6.1) were deployed in triplicate in a well-stirred solution of Cd2+
(0.40 µM) and Cu2+ (0.70 µM) over periods of time from 3 to 200 h in synthetic lake water (Windermere) (Section 2.2.2.3). A sufficient volume of the sample solution was used to ensure that the depletion of Cd2+ and Cu2+, by the DGT devices, was negligible.
The Cu2+ and Cd2+ uptake by DGT was also validated, in this way, in solutions ranging from 0 to 1.0 M NaNO3. The devices were also deployed in a solution containing natural freshwater (Parkwood Pond) as the matrix, spiked with 0.70 µM Cu2+, for the same deployment times and replication.
6.3. RESULTS AND DISCUSSION
6.3.1. Dialysis Membrane Diffusive Layer
An important aspect of this work was to demonstrate that the new DGT devices, employing a dialysis membrane diffusive layer, met the same set of assumptions as the conventional DGT system and could be used within the existing theoretical framework 14,
16. In particular, the use of a dialysis membrane as the diffusive layer had to meet the 119 Chapter 6 assumptions that there were no detectable interactions between the analyte species and the diffusive layer membrane, and that the diffusion boundary layer (DBL), formed at the interface between the diffusive layer and the bulk solution, was not a significant influence on the mass transport into and across the diffusive layer (Figure 6.3). The DBL effects were investigated under well mixed conditions using the usual DGT validation experiments (Section 6.3.6) and under various flow conditions (Section 6.3.5).
Binding Diffusive Sample
Layer Layer DBL Solution
Cm
SS Cb P
M2+ 2+ Concentration M n
’
(PSS) C
0 ∆g δ
Distance
Figure 6.3 Schematic representation of the concentration gradients of free ionic species in a DGT device consists a solid membrane diffusive layer and a liquid binding phase.
An essential pre-requisite of the DGT analysis was that there was very little interaction between the dialysis membrane diffusive layer and the measured solute. If this condition was true, then, when a membrane was immersed in a solution, the concentration of the solute in the membrane should, at equilibrium, be the same as the concentration in the solution. Figure 6.4 shows the ratio of Cd2+ or Cu2+ concentration in the membrane
2 (Cmembrane) and in the solutions (Csolution) after a 100 cm sheet of 50 µm thick pre-treated cellulose dialysis membrane was exposed to a well-stirred solution containing 2.0 µM 120 Chapter 6
2+ 2+ –5 Cd and Cu for 24 h in solutions of varying ionic strength (10 M – 1.0 M NaNO3).
The ratio of the membrane metal concentration to the solution metal concentration was in the range 0.98-1.05 after 24 h exposure. This indicates that there was little or no interaction between the dialysis membrane and the Cd2+ and Cu2+ ions. This means that the concentration of these ions in the diffusive layer will not gradually increase and the assumption required by the DGT equation holds.
120
100
80 (%)
60 solution /C 40 Cd membrane 20 C Cu 0 -6 -4 -2 0
Log [NaNO3]
Figure 6.4 The interaction between the cellulose dialysis membrane (100 cm2 × 50 µm) and metal ions (2.0 µM) under different ionic strengths for 24 h. The ratio (%) was defined as the ratio of Cd2+ or Cu2+ concentration in the membrane (Cmembrane) to that in the solutions (Csolution).
6.3.2. Optimization of PSS Solution Concentration
One of the most important assumptions made for the DGT equation was that the free metal ion concentration, C’, at the internal membrane interface was zero (Figure 6.3) at which point the maximum flux can be achieved. In order to satisfy this condition, the metal ions must be taken up rapidly by the PSS binding phase. The PSS concentration thus needed to
121 Chapter 6 be as high as possible to make this rapid uptake likely, but as low as possibly to minimise matrix effects during analysis. Therefore, the PSS concentration was optimised.
Figure 6.5 shows the effect of the PSS concentration on the fluxes of Cd2+ or Cu2+ across the membrane (fluxes were calculated according to procedures in Section 6.2.6). Figure
6.5 shows that the maximum fluxes were observed for both Cd2+ and Cu2+ when the PSS concentration was 0.015 M or greater. 0.020 M of PSS was chosen as the binding solution to ensure that it was sufficient to reduce the interfacial concentration (C’) to, effectively, zero, even when dilution of the solution occurred due to osmotic pressure. Considering that the concentration of metal ions used in these experiments were very high in comparison with real deployment conditions, this PSS concentration would probably be sufficient for most, if not all, deployment conditions.
0.7
0.6
) 0.5 -1 s
-2 0.4
g cm 0.3 µ
0.2 Flux ( Cd 0.1 Cu 0 0 0.01 0.02 0.03 0.04 0.05 0.06
PSS Concentration (M)
Figure 6.5 Effect of PSS solution concentration on fluxes through the diffusive membrane (see Section 6.2.8 for experimental details).
122 Chapter 6
6.3.3. Metal Ion Binding Properties of Poly(4-styrenesulfonate)
Figure 6.6 shows the effect of concentration on the amount of metal ion uptake by the binding phase under non-competitive conditions. It was found that for both Cd2+ and
Cu2+, the amount of metal ion uptake increased with concentrations up to about 2 mM before saturation occurred. The maximum uptake (saturation) values for Cd2+ and Cu2+ were 13.5 µmole cm-2 and 13.0 µmole cm-2, respectively. The non-competitive binding capacities for other metal ions were obtained in the same way. The results are summarised in Table 6.1. The binding capacities of the monovalent ions doubled those of the divalent ones, suggesting that the interactions with the PSS were largely ion exchange interactions with the sulfonate groups of PSS.
16
14 )
-2 12
10 mole cm µ 8
6
Uptake ( Uptake 4 Cd 2 Cu
0 0 1 2 3 4 Concentration (mM)
Figure 6.6 Effect of initial metal ion concentrations in the sample solutions on the uptake by PSS liquid binding phase. The DGT devices were immersed in various concentrations of metal ion solutions for 24 h at pH~6.
The metal binding capacities under competitive conditions were also investigated (Table
6.1). The test solution contained Cd2+, Cu2+ and all the alkali and alkali earth metals used 123 Chapter 6 in the non-competitive study. The results showed that the binding capacities obtained under competitive conditions for all metal ions decreased compared with the non competitive conditions, as expected (Table 6.1). The competitive binding capacities were shown to be in the order Cd2+ > Cu2+ >> alkali earth metals > alkali metals. This order indicates that the PSS can be used to selectively bind heavy metal ions such as Cd2+ and
Cu2+. The results in Table 6.1 also show that the total metal binding capacities of PSS were approximately the same under both competitive and non-competitive conditions (the competitive capacities add up to 20.5 µmole ml-1 counting the divalent ions twice). This result further suggests that the interactions between PSS and the metal ions are dominated by cation exchange. The order of selectivity for competitive binding indicates a stronger preference for Cd2+ and Cu2+ than for the other metals that were used. This preference suggests that some complexation may occur (see next section).
Table 6.1 Non-competitive and competitive binding of various metals by a 2.0 ml 0.020 M PSS solution Cd2+ Cu2+ Ca2+ Mg2+ K+ Na+ a Non-competitive (µmole cm-2) 13.5 13.0 12.9 13.3 25.6 26.1 a Non-competitive (µmole ml-1) 10.4 10.0 9.9 10.2 19.7 20.1 b Competitive (µmole cm-2) 7.8 3.3 0.94 1.1 0.49 0.35 b Competitive (µmole ml-1) 6.0 2.5 0.72 0.82 0.37 0.27 a A liquid binding phase DGT device was immersed in a 50 ml solution containing individual metal ion of 2.5 mN. b A liquid binding phase DGT device was immersed in a 1000 ml solution containing all the above metal ions of individual metal ion of 18 µN. c.The binding capacities were expressed by number moles of metal bound in the binding solution per unit volume and per unit exposed area of membrane.
124 Chapter 6
The binding capacity of the 0.020 M PSS solution was given in terms of both µmole cm-2
(for comparison with other DGT binding phases) and µmole ml-1 (for comparison with the concentration of the PSS solution). This volume characteristic highlights an important advantage of this DGT approach, compared with the use of a solid state binding phase16 i.e. the capacity of the PSS solution was considerably higher than the solid state binding phases and could be increased further simply by increasing the volume of binding solution
(see section 8.3.2 for a detailed comparison).
6.3.3.1. Stability Constants of the Polymer Metal Complexes
Stability constant values were calculated to further characterise the PSS binding interaction. Several methods used to determine the stability constants and the corresponding average coordination numbers of metal ions in polymeric solutions have been reported previously 241-245. These methods include the potentiometric or pH titration methods 243, spectroscopic methods 244 and the equilibrium dialysis method 241, 245. In this work, the stability constants of PSS with Cd2+ and Cu2+ were determined according to the dialysis method proposed by Juang et al. 241 and Samadfam et al. 242. To calculate the stability constant, the following assumptions were made: (a) there was no interaction between the free metal ions and the membrane; (b) the rejection coefficient of PSS was the same as that of the metal-PSS complex; (c) the complex reaction was in equilibrium in the polymer phase; and (d) at equilibrium, the free ion concentration in the sample solution phase was the same as that in the polymer phase.
When a DGT device was immersed in the metal solutions of various concentrations, PSS and metal-PSS complex were accumulated within the binding phase of the DGT device until equilibrium was achieved. At equilibrium, the concentrations of free metal ions in
125 Chapter 6 solution were considered to be the same as those in the DGT devices. Once this assumption is made the following theory can be used.
The stability constant K, for the binding reaction, can be written as (charges are omitted for all equations below):
[ M( PSS ) ] K = n (6.1) [ M ] × [ PSS ] n
At equilibrium, the free metal concentration, [M ] , can be expressed by subtracting the concentration of the metal-PSS complex from the initial free metal ion concentration:
[M] = ([M]0⋅V s − [M(PSS)n]⋅Vb)/Vs, where [M]0 is the initial free metal ion concentration; [ M(PSS)n] is the concentration of metal PSS-complex; and Vs and Vb are the volumes of the sample and the binding phase solutions.
The concentration of PSS, [PSS], can be obtained according to: [PSS] = [PSS]0 −
[M(PSS)n], where [PSS]0 is the initial concentration of PSS.
Assuming that the formation of the metal hydroxides were negligible, at pH 7 for the given metal ion concentrations 246, and only two forms of metals, free metal ion and its
PSS complex, existed in the solution and polymer phase. Thus, equation 6.1 can be rewritten as (charges are omitted):
[ M( PSS ) ] K = n (6.2) [ M ] 0 × V s − [ M( PSS )n ] × Vb n × ([ PSS ] 0 − [ M( PSS )n ]) V s Rearranging equation 6.2, we have:
[ M ] 0 Vb − Log( − ) = LogK + nLog ([ PSS ] 0 − [ M( PSS )n ]) [ M( PSS )n ] V s
126 Chapter 6
[ M ] 0 Vb let Y = − Log( − ) ; X = Log ([ PSS ] 0 − [ M( PSS )n ]) , then: [ M( PSS )n ] V s
Y = LogK + nX (6.3)
16.8
y = 1.80x + 8.11 2 16.4 R = 0.980
16
15.6
15.2 4 4.2 4.4 4.6 4.8 X
Figure 6.7 Plot of Y against X for determination of the stability constant and coordination number of PSS-Cu2+ complex reaction at pH~6, in 1 mM NaNO3 solution.
2+ Plotting Y against X obtained for Cu gives Figure 6.7. The stability constant, LogKCu =
8.1, and the coordination number, n = 1.8 (in 1 mM NaNO3), were obtained from the Y intercept and the slope of the curve, respectively. The stability constant, logKCd = 9.0, and the coordination number, n = 2.2 for Cd2+ were obtained by the same means. The binding stoichiometry formed here is very close to that determined in Section 6.3.3. Therefore, cation exchange processes seem to dominate, although transition metals seem to exhibit the strongest interactions with the PSS, which makes it ideal for DGT applications. These logK stability constants also confirm that the stability constants for Cd2+ and Cu2+ were high enough for their strong binding to the PSS polyelectrolyte.
127 Chapter 6
6.3.3.2. Effect of pH on the Binding Capacity
The change in binding capacity for Cd2+ and Cu2+ with the 0.020 M PSS solution with solution pH is shown in Figure 6.8. The binding capacity increased rapidly from pH 1 to
3, probably due to the increase in the proportion of the base form of the sulfonic acid groups in the PSS solution. From pH 4 to 8 the binding capacity remained quite constant for both metal ions. At pH>8 the binding of Cu2+ decreased rapidly, due to the formation of insoluble hydroxides. The Cd2+ binding decreased rapidly at pH>10. The fact that the
247, 248 sulfonic acid groups have a low pKa value (in fact it may be a strong acid functional group) allows them to function as a binding phase for metal ions over a wide pH range
210 compared with weaker acid groups, such as carboxylic acid (pKa about 4.8) and Chelex
249 100 (pKa about 3.5) .
16
14
12 ) -2 10
8 mole cm mole µ 6 Cd 4
Uptake ( Uptake Cu 2
0 0 510 pH
Figure 6.8 Effect of sample solution pH on the metal ion uptake by the PSS liquid binding phase.
128 Chapter 6
6.3.3.3. Effect of Ionic Strength on the Binding Capacity
The change in the binding capacity of 0.020 M PSS for Cd2+ and Cu2+ with increasing ionic concentrations is shown in Figure 6.9. The binding capacity of PSS for Cd2+
-2 -2 decreased from 13.5 µmole cm to 3.3 µmole cm as the ionic concentration (as NaNO3) increased from 10-5 M to 1.0 M. The binding of Cu2+ decreased from 13.0 µmole cm-2 to
-2 2.5 µmole cm over the same increase in NaNO3 concentration. This decrease in binding capacity was due to increased competition from Na+, which was the least competitive ion tested here (but which is important in seawater deployments, Section 6.2.11).
16
14
) 12 -2 10
8 mole cm µ 6
4 Cd Uptake ( Uptake 2 Cu
0 -6 -4 -2 0
Log [NaNO3]
Figure 6.9 Effect of ionic strength presented as NaNO3 concentrations in the sample solutions on the metal ion uptake by the PSS liquid binding phase.
2+ 2+ Even at 1 M NaNO3 the binding capacities of Cd and Cu were comparable with those for the solid phase binding agents used previously for DGT at lower ionic strengths. At lower ionic strengths they were substantially higher: Chelex 100 resin gel, 1.1 µmole cm-2 for Cd 16; polyacrylamide/polyacrylic acid copolymer hydrogel, 1.56 µmole cm-2 for Cd2+,
129 Chapter 6
1.59 µmole cm-2 for Cu2+ (Chapter 3); poly(acrylamidoglycolic acid-co-acrylamide), 5.2
µmole cm-2 for Cd2+, 5.4 µmole cm-2 for Cu2+ (Chapter 4); cellulose phosphate membrane,
3.07 µmole cm-2 for Cd2+, 3.22 µmole cm-2 for Cu2+ (Chapter 5).
6.3.4. Diffusion of Cd2+ and Cu2+ in the Cellulose Dialysis Membrane
The diffusion coefficients of metal species within the diffusive layer need to be known (or measured) for DGT to be used to estimate analyte concentrations. This is how the DGT technique is calibrated. In order to ensure the applicability of the measured diffusion coefficients, the conditions employed for the measurements had to be similar to the DGT deployment conditions. The diffusion cell was set up as described in Section 2.3.5, with
0.020 M PSS in the receiving solution and 10.0 ppm free metal ions in the source solution.
The experimental procedures are explained in Chapter 2. The transport process involved in the diffusion coefficient measurement was an active one because the following reaction occurred at the interface with the diffusion cell (or binding phase in the case of DGT).
2+ M (aq) + nPSS(aq) M(PSS)n(aq)
The concentration of PSS used was sufficient to reduce the free metal ion concentration at the membrane/PSS solution interface to, efficiently, zero through the above reaction
(Sections 6.3.2 and 6.3.3 for supporting evidence). Under this condition, the concentration difference across the membrane equalled the bulk solution concentration provided that the effect of a DBL in the source solution is negligible. The measurement of diffusion coefficients can then be carried out according to the DGT equation,
AC Dm AC b M = t . Plotting M versus the product of Cb and t gives a slope value (ADm/∆g), ∆g which allows the calculation of Dm, since A and ∆g are known constants (Figure 6.10).
130 Chapter 6
40
y = 0.0535x - 0.406 2 30 R = 0.997 g) µ ( 20
Mass Cd Mass 10
0 0 200 400 600 800
40
y = 0.0400x + 0.649 30 2 g) R = 0.992 µ (
20 Mass Cu Mass
10
0 0 200 400 600 800
Ci×t (ppm min.)
Figure 6.10 Plots of mass vs C×t for Cd2+ and Cu2+ (10.0 ppm) in synthetic lake water solution (Windermere). The curves were drawn with mass (µg) transported to compartment B versus the product of time (t, min.) and average metal ion concentrations (Ci, ppm) of each diffusion period in compartment A.
Table 6.2 shows the diffusion coefficients of Cd2+ and Cu2+ in the cellulose dialysis membrane measured under different ionic strengths (NaNO3 concentrations). The results indicate that when the NaNO3 concentration was varied from 0 to 0.010 M, a dramatic
2+ 2+ -6 2 -1 -7 2 -1 decrease in the Dm values for both Cd and Cu (5.1×10 cm s to 7.8×10 cm s and 131 Chapter 6
-6 2 -1 -7 2 -1 4.2×10 cm s to 6.8×10 cm s ) was observed. However, the effect of the NaNO3
-7 2 -1 -7 2 -1 concentration on the Dm values was less significant (7.8×10 cm s to 2.8×10 cm s
-7 2 -1 -7 2 -1 and 6.8×10 cm s to 2.1×10 cm s ) in the more concentrated NaNO3 solutions (from
2+ 2+ 0.010 M to 1.0 M). When 90 µM Cd or 160 µM Cu was used, within the NaNO3 concentration range from 0.01 M to 1.0 M, non-linear curve fitting indicated that the relationships between the Dm and NaNO3 concentrations (CI) could be expressed as Dm =
-0.230 2+ -0.258 2+ 0.279 CI for Cd and Dm = 0.196 CI for Cu (Figure 6.11). These results are quite interesting in that they indicate that there is little change in the Dm at high ionic strengths. This relatively constant value of Dm will be an advantage in the DGT
2+ 2+ deployments in estuarine and other coastal waters. The Dm values for Cd and Cu in synthetic lake water matrix (Windermere, Lake District, UK) were determined as 2.5×10-6 cm2 s-1 and 1.9×10-6 cm2 s-1, respectively.
1
-0.230 Cd Dm = 0.279 CI 0.8 ) -0.258 Cu Dm = 0.196 CI s 2 0.6 cm -6 -1 10 ( 0.4 m D
0.2
0 0 0.2 0.4 0.6 0.8 1
[NaNO3] (M)
Figure 6.11 Diffusion coefficients of Cd2+ and Cu2+ in dialysis membrane in various concentrations of NaNO3 (0.010 M-1.0 M).
132 Chapter 6
Table 6.2 Diffusion coefficients of Cd2+ and Cu2+ (10-6 cm2 s-1) in cellulose dialysis membrane under various concentrations of NaNO3 (M) (pH~6)
[NaNO3] 1.0 0.10 0.050 0.010 0.0050 0.0023 0
D(Cd2+) 0.28 0.47 0.59 0.78 1.2 2.9 5.1
D(Cu2+) 0.21 0.32 0.43 0.68 0.94 2.2 4.2
As Torre 22 noted, in order to satisfy the condition of electroneutrality in a medium where ions are co-diffusing, the effective diffusion coefficient of a given ion (Di,eff), should be described by:
n ⎡ ∑ j =1 z j D j ( dC j / dx ) /( dC i / dx )⎤ Di , eff = Di − Di z i C i × ⎢ n 2 ⎥ (6.4) ⎣⎢ ∑ j =1 z jD j C j ⎦⎥ where Di and zi are the tracer diffusion coefficient and charge, respectively of ion "i".
Equation 6.4 shows that the diffusion coefficient of an ion is influenced by the concentrations, concentration gradients and diffusion coefficients of all "j" ions, including the ion "i" of interest; the second term on the right hand side of equation 6.4 refers to the coulombic or electrical component of Di,eff. When the DGT devices are deployed in natural waters, there are numerous cations present. Equation 6.4 therefore indicates that the nature of the matrix can also have a significant influence on the diffusion coefficient of analyte metals. Therefore, diffusion coefficient should be measured in water of similar composition (synthetic) to that in which DGT device is being deployed (Chapter 7).
6.3.5. Effect of Stirring Conditions on the DBL Layer
Another important assumption made in the derivation of the DGT equation was that the double boundary layer (DBL), developed during the DGT deployment, must be insignificant in comparison with the thickness of the diffusive layer.
133 Chapter 6
Zhang and Davison have described an approach to minimise the influence of a substantial
DBL being formed in poorly mixed waters, in which the DBL thickness becomes significant compared with the diffusive layer thickness 150. The need to consider DBL effects arises because the diffusion coefficients of metal ions in the bulk solution are similar to the diffusion coefficients in the polyacrylamide gels 154. Additionally, the diffusive gradient within the DBL can also limit the overall mass transport. The approach by Zhang and Davison to overcome DBL effects involved deployment of two DGT devices with hydrogel diffusive layers of different thicknesses, which effectively allows the value of an average DBL to be subtracted from the estimation of the accumulated metal ions, M, assuming that the DBL thicknesses in the two DGT devices were the same150.
With the dialysis membrane diffusive layer described here, it was thought that the DBL might not have a significant effect on the concentration measurement. This is because of the differences in the diffusion coefficients in solution with those in the membrane.
Diffusion coefficients of 5.55×10-6 cm2 s-1 for Cd2+ and 5.67×10-6 cm2 s-1 for Cu2+ in
250 seawater (Dw) at 16 °C have been previously reported . Dw is the diffusion coefficient of metal ion in water solution. These values increase slightly with a decrease in ionic strength; an 8% increase in deionised water compared with seawater was also reported 16.
On the other hand, diffusion coefficients of metal ions in the dialysis membrane (Dm) varied from 5.1×10-6 cm2 s-1 to 0.28×10-6 cm2 s-1 for Cd2+ and 4.2×10-6 cm2 s-1 to 0.21×10-
6 2 -1 2+ cm s for Cu , as the ionic concentration increased from 0 to 1.0 M (as NaNO3) (Table
6.2). The diffusion coefficients decreased dramatically in the lower ionic strength range
(<0.1 M) but changed little in the higher ionic strength range (0.1-1.0 M) (Figure 6.11).
These data indicate that the diffusion across the dialysis membrane in high ionic strength
134 Chapter 6 natural waters under most hydrodynamic conditions would be the rate-limiting step of the overall transport process (due to the diffusion coefficient in bulk solution being nearly 20
× greater than the diffusion coefficient in the membrane). In low ionic strength waters the
DBL could become much more significant as there is less difference between the diffusion coefficients in the water and in the membrane. As the DBL is dependent upon the solution hydrodynamics 148, experiments were undertaken to investigate this phenomenon.
Figure 6.12 shows the effect of the stirring rate on the mass of Cd2+ transported across the membrane. The negative Y-intercepts reflect the time taken for the concentration gradients to develop, which decreased as the stirring rates increased. For all cases, a linear relationship between the mass and the time were obtained (R2 = 0.994 - 1) indicating that the DBL was fixed during the deployments under each hydrodynamic condition (equation
6.5). It was also found that the rate of transport increased as the stirring rate increased from 0 to 200 RPM (the slopes of the lines in Figure 6.12 represents the rates of transport) due to the decrease of the DBL thickness (equation 6.5). A further increase in the stirring rate above 200 RPM showed no significant effect on the rate of transport because the minimum DBL thickness was achieved. Under these conditions the flux through the membrane is the rate-limiting step. At the RPM below 200, however, the flux in the DBL became rate limiting. As these experiments were carried out in 0.1 M NaNO3 solution, for which there will be a large difference between Dm and Dw, the DBL definitely needs to be considered.
The DGT equation has been modified to include the DBL contribution. Considering the
D ( C − C ) flux through the DBL ( J = m b m ) equals the flux through the diffusive layer δ
D ( C − 'C ) ( J = m m , Figure 6.3), we have: ∆g 135 Chapter 6
M D ( C − C ) D ( C − 'C ) J = = m b m = m m , At δ ∆g
2 -1 where Dm and Dw, are diffusion coefficients (cm s ) of the solute in the membrane and in the bulk solution, respectively; ∆g and δ represent the thickness of the diffusion layer and
-1 the DBL layer, respectively; C', Cm and Cb are the solute concentrations (mol l ) at the internal membrane surface (the membrane/binding phase interface), the external membrane surface (the membrane/DBL interface) and the bulk solution.
50
40
g) 30 µ
20 Mass ( Mass
10
0 0 5 10 15
Time (h) 1000 RPM M = 4.06t - 0.33 R2 = 0.994 200 RPM M = 4.08t - 0.33 R2 = 1 100 RPM M = 3.82t - 0.77 R2 = 1 50 RPM M = 3.49t - 0.61 R2 = 0.999 0 RPM M = 2.97t - 1.47 R2 = 1
Figure 6.12 Effect of stirring rate on the mass transport across the dialysis membrane in 0.10 M NaNO3 solution.
When the PSS concentration in the receiving solution is sufficient to reduce C’ to zero, the
DGT equation can be obtained:
C D AD D C M = w m b ⋅ t (6.5) δD m + ∆gD w
136 Chapter 6
Equation 6.5 gives the relationship between the mass (M), the DBL thickness (δ) or the deployment time (t) for a given DGT device and bulk solution concentration. Under the condition of δDm << ∆gDw, the maximum rate of transport, ADmCb/∆g, can be achieved
g M∆ g and Equation 6.5 becomes the normal DGT equation, C b = . More importantly, the AtDm effect of DBL can be experimentally quantified according to the relationship given by the equation.
Table 6.3 Effect of double boundary layer (δ) at stirring rate on the accumulated mass
Stirring Rate (RPM) Mass transport rate (µg/h) DBL increase (cm) ∆C/C (%)
1000 4.06 0 0
200 4.08 0 0
100 3.82 0.0038 6
50 3.49 0.010 17
0 2.97 0.022 34 Note: the membrane area is 1.2 cm2, thickness (∆g) 0.005 cm, diffusion coefficient in the –7 -1 2+ membrane (Dm) = 4.7 × 10 cm s , concentration of Cd 10 ppm.
The thickness of DBL and the measurement errors caused by DBL under different stirring conditions can be estimated based on the experimental data shown in Figure 6.12 using equation 6.5 (Table 6.3). It was found that the DBL thickness obtained under the quiescent condition had increased by 0.022 cm, which is more than 4 times the thickness of the diffusive layer used. Under such conditions, without correction for the DBL effect, the DGT measurement will underestimate the concentration by 34%. Actually, this also means that DBL effects at this ionic strength will lead to an error of no more than 34%.
The thickness of DBL decreased rapidly as the rate of stirring was increased. When the rotation rate was greater than 200 RPM, the thickness of DBL achieved its minimum value and maintained. At 100 RPM, the measurement error caused by the development of DBL 137 Chapter 6 was 6%, which is acceptable for most environmental analysis. These results became more significant for low ionic strength waters, which are poorly mixed.
DGT Device Diffusive Layer Vibrating Wave Action
DBL Development Direction
Figure 6.13 Schematic diagram of wave action effect on the dialysis membrane diffusive layer vibration.
It should be mentioned that the convective condition generated by the stirrer used for this experiment (Figure 6.2) should be quite gentle even with 200 RPM rotation rate.
Although we could not quantitatively relate the stirring rate used in this experiment to the hydrodynamic conditions in natural environments, we believe the convective condition generated by the wave action in the real field deployment should be much stronger than the convection generated under 200 RPM in the laboratory. This is due to the fact that the thickness of DBL depends on not only the surface flow rate but also the direction of the flow. The wave action can cause the diffusive layer vibrating in the direction against the direction of DBL developed (Figure 6.13). Therefore, it is unlikely that the significant thickness of DBL can be developed under such conditions. By considering this, and the observation from Figure 6.12 that a constant transport rate can be obtained under each given hydrodynamic condition, we have good reason to believe that the development of a
DBL would have only a minor impact on the accuracy of the DGT measurement in real
138 Chapter 6 environmental conditions. The DBL effect can be further minimised by measuring the diffusion coefficient under the imitated hydrodynamic conditions of natural waters where the DGT devices will be deployed.
6.3.6. Validation of the PSS/dialysis DGT Device
The test of whether a new DGT method meets the assumptions required by the DGT equation is to investigate the mass vs. time relationship. If this plot is linear and passes through the origin then the assumptions are likely to hold. The new DGT assemblies were validated by testing this relationship between the mass of analyte accumulated in the binding phase (M) and the deployment time (t) with a solution of known concentration 16.
Figure 6.14 shows the M vs. t relationship for Cd2+ and Cu2+ in a well stirred synthetic
Windermere lake water. The mass of metal ions (µg) increased linearly with time over the deployment periods used for Cd2+ (r2 = 0.969) and Cu2+ (r2 = 0.980). The solid lines in
Figure 6.14 are lines of best fit for the experimental data; the dashed lines are the theoretical lines calculated using the DGT equation.
The theoretical lines are virtually identical to the lines of best fit for both Cd2+ and Cu2+.
These two results, along with those described above, confirm that the new DGT device meets the criteria for application of the DGT equation. Similar coefficients of determination for the experimental results and recoveries (when compared with the theoretical lines) were obtained for the validation experiments in a NaNO3 solutions ranging from 0 to 1.0 M. These results support the conceptual model described above in which diffusion through the membrane is the rate-limiting step under the conditions investigated. If this model were not true, then such high coefficients of determination and recoveries of about 100% would not have been obtained.
139 Chapter 6
30
20 g) µ
10 Mass Cd ( Cd Mass
0 0 50 100 150 30
20 g) µ
10 Mass Cu ( Cu Mass
0 0 50 100 150 200
Time (h)
Figure 6.14 The mass of the metals accumulated by the PSS liquid phase in DGT devices as a function of time. DGT devices were suspended in a well-stirred solution containing known concentrations for different time periods. The solid lines are the lines of best fit for the experimental data. The dashed lines are predicted relationships calculated from known deployment conditions and the DGT equation. For ∆: C = 0.40 µM Cd2+ in synthetic Windermere lake water, D = 2.5×10-6 cm2 s-1; for O: C = 0.70 µM Cu2+ in synthetic Windermere lake water, D = 1.9×10-6 cm2 s-1; and for : C = 0.70 µM Cu2+ added to Parkwood Pond water, D = 0.94×10-6 cm2 s-1 ; ∆g = 50 µm, A = 1.54 cm2.
140 Chapter 6
The reproducibility of these DGT devices was also tested and found to be satisfactory
(5.4% rsd, n = 9, for Cd2+ and 5.6% rsd, n = 15 for Cu2+). These data suggest that the configuration of this DGT device is close to ideal, as described by the DGT equation. In addition, the nature of the interface between the diffusive layer and binding phase (with a
100%, even coverage of the binding sites at this interface) combined with a well-defined and reproducible diffusive layer are ideal for DGT applications. The ease of handling means only minimal user experience is required. Furthermore, there is no requirement to elute the metals from the binding phase, thus removing the need for estimation and correction of the elution efficiency. However, the matrix does need to be matched during analysis.
The pre-concentration factor is defined as the ratio between the concentration of metal ion bound in the PSS solution and in the synthetic bulk solution 16. This factor increased as deployment time increased (indicated in the bracket): 94 (50 h), 188 (100 h), 281 (150 h) for Cd2+ and 71 (50 h), 140 (100 h) 209 (150 h) and 278 (200 h) for Cu2+. For a 30 day field deployment, a concentration factor of 1120 was observed for Cu.
Validation of the new DGT device was also carried out by spiking a natural freshwater
(Parkwood Pond) with Cu2+. The results are also shown in Figure 6.14. A linear relationship was observed between the accumulated mass and time (r2 = 0.981), indicating that the assumptions required for DGT were valid. The recovery was only 46%. This low recovery resulted from the presence of humic substances, which would have complexed a significant fraction of the added Cu2+. The work presented in the next chapter focuses on metal speciation information such as this when employing this new DGT device.
141 Chapter 6
6.4. CONCLUSIONS
This chapter demonstrates the potential for a new type of DGT device, using a poly(4- styrenesulfonate) (PSS) solution binding phase, and a dialysis membrane diffusive layer.
The diffusion properties of the dialysis membrane and the binding properties of the PSS solution were characterised and found to be suitable for use with DGT. The double boundary layer effect was investigated in detail with results suggesting that the effect may be minimal. A new DGT device was designed and validated by demonstrating a linear mass vs. time relationship for Cd2+ and Cu2+ in synthetic waters and in the Cu2+ spiked
Parkwood Pond matrix solution. The lower recovery in the Parkwood Pond solution indicated that the measurement of metals by this DGT device did not include the humic substances complexed fractions of metals, i.e. only inorganic metals were measured.
The advantages of this approach to DGT include theoretically ideal mass transport and accumulation due to the mobility of the binding solution and the ideal interface with the diffusive layer, with consequent good reproducibility; a well-defined, reproducible diffusive layer (commercially available), which overcomes the fragility and swelling problems of the gel based binding phases. In addition, there is no need for elution corrections, which are required for all solid binding phases that do not elute 100%. The only drawback was the need to dilute and matrix match standards to the PSS solution for instrumental analysis. This drawback could be overcome through further method development.
142 Chapter 7
Chapter 7 Characterisation of the Dialysis Membrane/PSS DGT Device for Trace Metal Speciation Measurements
143 Chapter 7
7.1. INTRODUCTION
In Chapter 6, a new DGT device, employing poly(4-styrenesulfonate) (PSS) aqueous solution as a binding phase and a dialysis membrane as a diffusive layer, was validated under laboratory and natural water conditions. This chapter describes a detailed investigation into the trace metal speciation characteristics of this new DGT device.
Natural waters contain various ligands that can form complexes with trace metal ions, as described in Section 1.2 184, 251. Consequently, free metal ions are usually a minor component of the total metal species present. These numerous complexes have very different physical properties, such as charge, size and diffusion coefficient, compared with free metal ions. A DGT device employing a small pore size diffusive layer, such as the dialysis membrane 202 and an appropriate binding phase (i.e. PSS), has the ability to make use of these physical differences to achieve selective measurement of species. This process occurs because the mass transport in DGT analysis is an active process in which a diffusive flux is maintained by continual binding of the labile species. Only species which are small enough (not excluded by the diffusive membrane), and which can be bound strongly by PSS, are measured. Species can also be differentiated on the basis of their diffusion coefficients, which influence the fluxes through the diffusion layer.
The purpose of this chapter is to investigate the effect of ligands of various types on the amount of trace metals measured by the PSS DGT device. In order to correctly interpret the DGT speciation measurement it is vital to characterise the difference between the diffusion coefficient of the organic complexes and those of the inorganic metal species.
One ligand for which the complexing properties have been well established is EDTA.
EDTA forms strong complexes with Cu2+ and Cd2+ in a 1.0:1.0 ratio, so will be an ideal 144 Chapter 7 case study to measure the diffusion coefficient of a metal complex. The experiment can be set up so that there is effectively no free or inorganic metal ions present in solution; in other words the diffusion coefficient of the complex can be measured without interference from the free metal ions. Being able to measure the diffusion coefficient of one complex will allow a full characterisation of the DGT speciation measurement of a known solution containing that complex. General trends may then also be made concerning the measurability of other metal complexes, for which diffusion coefficients are less readily characterised.
The fraction of metal species measured by DGT was then compared with a theoretical free metal ion fraction, where available, for various ligands commonly found in natural waters, including ethylenediaminetetraacetic acid (EDTA), humic acid (HA), tannic acid (TA), glucose (GL) and dodecylbenzenesulfonic acid (DBS). Several ratios of the ligand to the metal ion were investigated to see how the DGT-labile measurement varied in response.
This comparison will help determine the general speciation properties of the PSS DGT device. In addition, the DGT-labile fraction of metal ions was measured by the DGT device in various natural water sites with varying levels of organic contents.
Description of the Properties of the Ligands Chosen
The ligands chosen all have some environmental significance. Ethylenediaminetetraacetic acid (EDTA), which can form very stable complexes with a wide range of metal ions 246, is one of the examples of aminopolycarboxylic acids. Aminopolycarboxylic acids are widely used in industrial processes, and in particular, are used as substitutes for phosphates in detergents. After their release into the environment, these chelating agents may affect the speciation distribution of metals within the aquatic ecosystems 252.
145 Chapter 7
Humic acids (HA) are a major class of humic substances present in natural waters. Humic acids are defined operationally as the fraction of humic substances that precipitate upon acidification. These substances are present in soil, water and sediments, in both soluble and insoluble forms 253. Humic acids have complex structures including a substantial proportion of condensed aromatic rings with a large number of –OH and –COOH functional groups 134, 254, 255. HA are often the ligands present at the highest levels and therefore play a crucial role in the speciation, transportation and deposition of metal ions
256, 257 .
Tannic acid (TA), a type of tannin polyphenolic substance 258, resulting from the leaching of bark and leaf litter 259, was also selected for the study. It has properties similar to those of humic substances in regard to complex formation, absorbability and colour, but it has a smaller molecular weight 213.
Dissolved carbohydrates are ca. 10 - 40% of the dissolved organic matter in natural water and play an important role in the aquatic ecosystem 260. Polyhydroxy compounds have long been known to act as ligands in the formation of coordination complexes of metal ions 261, 262. For these reasons, D-glucose (GL) was selected as an example to represent the polyhydroxy compounds. GL compound with –OH function groups on the molecular structure interacts with metal ions in a complicated way, such as forming a strong sugar
H-bonding network or other interactions 263. However, the interaction was weaker than the complexes formation of HA or TA 264.
The wide use of synthetic surfactants in domestic and industrial applications and the inflow of heavy metal ions into aquatic systems have intensified investigations concerning ecological problems of natural and waste waters 265. Dodecylbenzenesulfonic acid (DBS)
146 Chapter 7 was selected for this study to represent this group of ligands. It is well known that DBS is able to form complexes with a wide range of metal ions by means of ion-exchange or electrostatic interaction, however these complexes are often unstable 184, 266.
7.2. EXPERIMENTAL
7.2.1. Measurement of Diffusion Coefficients of EDTA-Metal Complexes
Diffusion coefficients of metal-EDTA complexes in the dialysis membrane were measured using the diffusion cell described in Chapter 2. One compartment contained
0.40 mM EDTA (Aldrich) and 0.090 mM Cd2+ or 0.16 mM Cu2+ in a 50 ml synthetic lake water matrix (Windermere, UK). The other compartment was filled with 0.020 M PSS in a 50 ml solution of the same matrix. Samples were taken from both compartments and measured by FAAS at specific time intervals for 16 h. The diffusion coefficients were calculated according to the method described in Chapter 2.
7.2.2. Measurement of DGT-labile Fractions
Two solutions of synthetic Windermere water (Chapter 2, pH = 6.8) containing: (i) 0.70
µM Cu2+ and (ii) 0.40 µM Cd2+ were spiked with various ligands at molar ratios of
1.8:1.0, 1.0:1.0 and 1.0:1.8 with the metal ions. These solutions were mixed over night at
23°C, to allow equilibration of the complexation reactions, before nine DGT devices were deployed. Three devices were taken for FAAS measurement every twelve hours, while the solution metal concentrations were measured at the same time by FAAS. The ligands used were humic acid (HA) (Aldrich), tannic acid (TA) (Aldrich), glucose (GL) (Chem-
Supply, Australia), EDTA (Aldrich), and dodecylbenzenesulfonic acid (DBS) (Sigma).
The DGT-labile metal ion concentration was measured and calculated in the same way as the normal DGT deployment according to the DGT equation. The total metal ion
147 Chapter 7 concentration was obtained using ICP-MS. The DGT-labile fraction of metal ions (β) was calculated using the following equation:
labile DGT labile metal ion concentration β = . ion metal Total metal ion concentration
7.2.3. Theoretical Calculation of Free Cu and Cd Fractions
βtheoretical is defined as free metal ion concentration over total metal concentration, which was calculated, for a given ligand, using the speciation function of the IUPAC Stability
Constants Database (SC-database model, Academic Software) 267. This calculation requires inputs of known parameters, e.g. stability constants of the metal-ligand complexes, dissociation constants of the ligands, concentrations of the ligands and metal ion, pH of the solution, as outlined in Table 7.1.
Table 7.1 Constants used for theoretical calculation 246 268 258 - - 2- EDTA HA TA Cl NO3 SO4
269 270 271 LogβCu1 18.9 7.99 5.4 0.98 0.1 1.54
LogβCu2 - - 9.1 0.69
272 273 274 LogβCd1 16.5 7.18 - 1.57 0.46 0.72
LogβCd2 - - - 2.42 0.17 0.84 pKa1 1.99 - 8.68 -
246 pKa2 2.67 --- 1.92 pKa3 6.16 --- pKa4 10.26 ---
Molar mass 336.2 13,000 1701 35.5 62.0 96.1 Note: All constants are for 25°C; the solution pH was 6.8; the total concentrations of Cd2+ and Cu2+ were 0.40 µM and 0.70 µM respectively. Logβ: logarithm cumulative stability constant; pKa: negative logarithm acid dissociation constant. EDTA: ethylenediaminetetraacetic acid; HA: humic acid; TA: tannic acid.
148 Chapter 7
7.2.4. Field Deployments of PSS DGT Devices
Each deployment consisted of nine PSS DGT devices mounted on a foam buoy (Figure
7.1). The devices were deployed in various natural water sites (Figure 7.2) by being anchored to a jetty for specific periods of time.
Foam buoy O-ring
DGT devices
Anchor
Figure 7.1 Solution based DGT holders were fixed on a square shape foam buoy, floating on waters.
Two seawater sites were chosen: the first was a jetty within Runaway Bay Marina (Figure
7.3) (which had previously been shown to have high trace metal concentrations, particularly Cu) 275; the second was on a canal site in Biggera Waters (with relatively lower boat traffic) (Figure 7.4). Two freshwater sites were also chosen: one with relatively high concentrations of natural organic matter (NOM), Parkwood Pond (Figure
7.5); and the other with less NOM, Loders Creek (Figure 7.5). All the DGT devices were rinsed thoroughly with deionised water after collection, to minimise contamination.
149 Chapter 7
Grab water samples were also collected for each site, at the beginning, middle and end of the DGT deployment period, in polyethylene sample containers (Nalgene), pre-cleaned with 10% nitric acid solution. The samples were filtered immediately on site through 0.45
µm pore size cellulose nitrate (Whatman) membranes and acidified with 65% suprapur nitric acid (Merck) (2 ml acid per litre of sample) to pH < 2.
Temperature, pH and salinity were also measured on site for the purpose of estimating diffusion coefficients. The dissolved organic carbon concentrations (DOC) were measured using a Dohrmann DC-190 TOC analyser. Cu and Cd concentrations were measured by
ICPMS (Agilent Technologies, 7500 Series, Germany) to obtain the dissolved trace metal concentrations (Section 7.2.5).
N
Runaway Bay Marina
Gold Coast
Biggera Waters
Parkwood Pond
Loder Creek
Figure 7.2 DGT deployment site locations on the Gold Coast in Australia.
150 Chapter 7
N
Figure 7.3 The red circle shows the DGT deployment site at the Runaway Bay Marina (Runaway Bay Marina).
N
Figure 7.4 The red circle shows the DGT deployment site on Back Street in Biggera Waters.
151 Chapter 7
N
A
B
Figure 7.5 The red circle shows the DGT deployment site on Parkwood Pond (A) and Loders Creek (B).
7.2.5. Measurement of PSS DGT-labile and 0.45-filtered Cu and Cd
Concentrations
The concentrations of Cu and Cd accumulated by the PSS in the DGT devices were measured by ICP-MS after a five-fold dilution. Standards were matrix-matched with PSS diluted accordingly (as per Chapter 2). The detection limits of this method were 5.5×10-9 mole l-1 for Cu and 1.7×10-9 mole l-1 for Cd.
The 0.45 µm filtered Cu and Cd concentrations of grab samples were analysed by ICPMS
(Agilent Technologies, 7500 Series, Germany). Standard solutions were matrix-matched with NaCl and MgCl2 diluted accordingly. The detection limits of this method were 1.4 ×
10-8 mol l-1 for Cu and 5.7 × 10-8 mol l-1 for Cd in saline waters; 4.3 × 10-9 mol l-1 for Cu and 3.2 × 10-9 mol l-1 for Cd in fresh waters.
152 Chapter 7
7.3. RESULTS AND DISCUSSION
7.3.1. Diffusion of EDTA-Cu and EDTA-Cd in the Dialysis Membrane Diffusive
Layer
The DGT analysis is dependent on the rate of diffusion of trace metal species across the diffusive layer. The accumulation of metal ions in the binding phase arises from all labile metal ions, including free metal ions, inorganic complexes and organic complexes. The mass deriving from each species is proportional to the flux of that species across the diffusive layer, as described by equation 1.5. According to Fick’s law of diffusion
(equation 1.1), the flux depends upon both the diffusive coefficient and the concentration gradient of that species in the measured water.
The diffusion coefficients of Cd2+ and Cu2+ in the dialysis membrane were determined in
Chapter 6. It would be very useful to compare these diffusion coefficients values with those of metal-complexes. To measure the diffusion coefficient of a complex requires the complex to be the only form of metal ion present. Many natural ligands do not bind metal ions strongly enough or at a stoichiometric ratio sufficient to ensure this condition. As described earlier, one ligand that does meet this condition is EDTA, which binds Cu2+ and
Cd2+ strongly at a 1:1 molar ratio. The diffusion of the metal-complex can therefore be measured in isolation from free metal ion diffusion.
Since an excess amount of EDTA was used in the source solution, all metal ions were complexed. In addition, during the experiment (up to 16 h), the source solution concentration change, caused by transport of EDTA-metal complex to the receiving
153 Chapter 7 solution, was negligible (<< 1%). The concentration measured in the receiving solution was below the detection limit of FAAS within 16 h. The diffusion coefficients of EDTA metal complexes can be estimated to be less than 4.6 × 10-9 cm2 s-1 for Cd and 8.5 × 10-9 cm2 s-1 for Cu respectively (calculated based on the method in Chapter 2). These diffusion coefficients were over two orders of magnitude lower compared with the inorganic metal ion diffusion coefficients under the same conditions (2.5 × 10-6 cm2 s-1 for Cd2+ and 1.9 ×
10-6 cm2 s-1 for Cu2+). These results indicate that, with this dialysis membrane/PSS DGT device, the accumulation of free metal ions could be more than 200 times faster than the accumulation of their complexes, assuming they were present at similar concentrations.
Even if the metal-EDTA complex was present at concentrations 10 times higher than the inorganic metal ions, it would be underestimated considerably.
This result allows conclusions to be made concerning the likely role of diffusion coefficients of other metal complexes, which can not be measured so readily, on the PSS
DGT measurement. EDTA has either a comparable or a smaller molecular mass compared with other likely ligands, such as those discussed in section 7.1. As diffusion coefficients generally decrease in proportion to molecular mass 134 (although other factors such as shape are also important) metal complexes that form with these other ligands are likely to have diffusion coefficients that are similar or even lower when compared with the metal-EDTA complexes. Therefore, it is likely that DGT will measure largely inorganic and free metal ions, in preference to even small complexes with organic ligands.
However, in natural waters the speciation of trace metal ions is very complex, with a great range of inorganic and organic ligands present. The complexes that form from each of these ligands will have their own concentrations and diffusion coefficients, and therefore their own fluxes. In many natural waters, most (sometimes > 90%) of the trace metals are
154 Chapter 7 complexed to natural organic matter (as described in Chapter 1). The next section discusses experiments that investigated the amount of a trace metal solution measured by the PSS-dialysis membrane DGT device in the presence of different complexing agents at various proportions.
7.3.2. Measurement of Labile Metal Ions in the Presence of Ligands
The mass of a particular metal ion, measured in the PSS solution, is the sum of all forms of the metal that are DGT-labile. This can occur via several mechanisms:
(1) Exclusion of species larger than the pore size or molecular weight cut-off
(MWCO).
(2) Retardation of species flux (i.e. low diffusion coefficient).
(3) The ability of the binding phase functional group to remove the metal ions from a
complex. If this does not happen then the DGT equation does not apply and the
accumulation in the binding phase occurs by equilibration instead of a maintained
concentration gradient. The former will be a much slower process and
concentration of the metal ions will not occur.
(4) Exchange of the ligand within the time frame required for the species to diffuse
across the diffusive layer. This means that inorganic and organic species can be
interchangeable. This is possible for ligands that form weak complexes or
occasionally even for strong ligands 149.
The following sections describe the results of experiments in which the effect of various ligands, present at differing ratios with the metal ions, on the DGT-labile measurement was studied. The DGT labile metal fractions measured reflected the mechanisms as described in the earlier section of this chapter: 1. the membrane effectively distinguishes
155 Chapter 7 the metal species through size and diffusion differences; 2. different binding properties of the binding phases also distinguish the metal species; 3. some weak bound metal species can dissociate in the diffusion layer due to the concentration gradients. This comparison was used to indicate which speciation mechanism was dominant. Mechanism (1) was not relevant in these laboratory solutions. Mechanism (2) was likely to always be occurring to some extent. If the DGT-labile fraction was found to be the same as the theoretical estimation then it was likely that diffusional selectivity was the dominant mechanism.
This is because the complexed species would have diffused through the dialysis membrane much more slowly and effectively be insignificant to the DGT measurement compared with the inorganic metal species which diffuse through much more quickly. In other words only the inorganic fraction is DGT-labile.
Mechanism (3) will only become important for species which are not selected against significantly by mechanism (2) (i.e. have fluxes greater than 1% of the fluxes of the inorganic metal species) and which also bind the metal ions more strongly than the PSS solution. The laboratory experiment will not be able to examine this mechanism. It is possible that no such species exist when using the dialysis membrane diffusive layer, for which mechanism (2) becomes more important than for the polyacrylamide diffusive gels.
Zhang and Davison 154 have measured the diffusion coefficients of fulvic and humic acids in water and in several types of diffusive gels. In their work the diffusion coefficient was retarded by one order of magnitude at most, whereas the diffusion coefficients of large molecules/complexes through the dialysis membrane is likely to be retarded to a much greater extent. It has been reported that even 1000 Dalton molecules can not pass through the 10,000 MWCO membrane 276.
156 Chapter 7
The final potential mechanism, based on dissociation of complexes at the time scale of diffusion through the dialysis membrane, can be considered to be a modifier to both mechanism (2) and mechanism (3), with the former being more important. If a significant fraction of metal ions dissociates this will effectively speed up the mass transport slightly and act to maintain the concentration gradient. This would have the effect of increasing the DGT-labile fraction when compared with the theoretical estimation of the inorganic metal fraction.
These predictions will be discussed with respect to the experimental results obtained.
However it is important to remember that other factors could be responsible for a difference in the DGT-labile fraction and the theoretically estimated inorganic fraction.
The main one is an actual difference in the stability constant of the complexes in the experimental solution and that used in the speciation software. The humic substance ligands have a great variety of forms that occur in nature and any particular sample will have binding sites with a range of stability constants present 134, 277. For some ligands stability constants may not have been calculated while for others such as EDTA there will be a high degree of certainty. The precision of the DGT technique will also be a factor in establishing which mechanism is predominant. If the DGT-labile fraction is statistically equivalent to the theoretical inorganic fraction (α = 0.05) then mechanism (2) should be considered to dominate.
7.3.2.1. Speciation Measurement of Labile Metals in the Presence of EDTA
EDTA was the first ligand investigated. Figure 7.6 shows the effect of metal ion/EDTA molar ratio on the DGT labile metal ion concentration. It was found that, as the molar ratio of metal ion/EDTA decreased, the DGT-labile metal fractions dramatically
157 Chapter 7 decreased. Both βCd and βCu became virtually 0 at 1:1 molar ratio and lower. Therefore, as expected, with EDTA as the complexing ligand, the free metal ion concentration is effectively equal to the metal ion concentration minus the EDTA concentration. The fact that no Cu and Cd were measured in solutions with metal/EDTA ratios ≤ 1 indicates that the complexes are being selected against due to mechanism (2); i.e. slow diffusion through the dialysis membrane.
100 3.0:1.0 90 1.8:1.0 1.0:1.0 (%) 80 1.0:1.8 70 60 50 40 30 20 10 DGT Labile Metal Fraction β 0
Cd Cu Molar Ratios of Metal Ions and EDTA
Figure 7.6 Ratios of measured DGT labile metal ions (β) in solutions containing various molar ratios of metal ion and EDTA. Molar ratios between metal ions and EDTA are: 3.0:1.0, 1.8:1.0, 1.0:1.0 and 1.0:1.8, as shown in the Figure.
The theoretical free metal ion fractions at a different molar ratio of EDTA/metal ions were calculated (Table 7.2). The results revealed that both the DGT labile metal fractions and the theoretical free metal ion fractions were very similar and followed the same trend when the molar ratio of metal ion/EDTA was changed. The βCd values were 55.0% and
36.1%, the same as the theoretical value for 3:1 and 1.8:1 Cd/EDTA molar ratios 158 Chapter 7 respectively. The βCu values were 66.7% and 43.5%, the same as the theoretical value for the 3:1 and 1.8:1 Cu/EDTA molar ratios respectively (Table 7.2). The agreements between DGT measurement and theoretical calculation indicate that DGT measured free metal ion concentration in the EDTA solution.
Table 7.2 Effect of EDTA to metal ion molar ratio on the DGT labile metal ion fraction and free metal ion fraction calculated by SC-database model Molar Ratio of Metal Ion/EDTA
3.0:1.0 1.8:1.0 1.0:1.0 1.0:1.8
βCd (%) 55.0±3 36.1±2 0 0
βCd,theoretical (%) 56.1 37.4 0 0
βCu (%) 66.7±2 43.5±1 0 0
βCu,theoretical (%) 65.3 44.4 0 0
Note: βCd and βCd are the DGT labile fraction of Cd and Cu. The values shown here were the averages of 7 replicate experiments. βCd,theoretical and βCu,theoretical are the free metal ion fraction of Cd and Cu calculated by SC-database model 267.
7.3.2.2. Speciation Measurement of Labile Metal in the Presence of Humic Acid
Figure 7.7 shows the effect of varying the ratio between the metal ion and the HA in the sample solution on the DGT-labile fraction for Cd and Cu (βcd and βcu). It was found that both βCd and βCu decreased as the ligand to the metal ion ratio increased. When the metal ion/HA molar ratio increased from 1.0:1.8 to 1.0:1.0, to 1.8:1.0, the percentages of the
DGT labile metal ion, measured, increased from 18.7% to 32.8% to 51.0% for Cd and
1.2% to 10.1% to 42.4% for Cu. The DGT-labile fractions are compared to the theoretical inorganic fractions in Table 7.3.
From Table 7.3 it is apparent that the values of βCd and βCu obtained were very close to the theoretical values. The βCd values were 51.0%, 32.8 and 18.7%, close to the theoretical values for the 1.8:1.0, 1.0:1.0 and 1.0:1.8 Cd/HA molar ratio respectively. The βCu values 159 Chapter 7 were 42.4%, 10.1% and 1.2%, the same as the theoretical value for the 1.8:1.0, 1.0:1.0 and
1.0:1.8 Cu/HA molar ratios respectively (Table 7.3). It is more likely that, at 1.0:1.8 metal/HA molar ratio, the measured DGT labile metal ion fractions may be from the dissociation of the complexed metals (mechanism 4).
100 1.8:1.0 90 1.0:1.0 (%) 80 1.0:1.8 70 60 50 40 30 20 10 DGT Labile Metal Fraction β 0
Cd Cu Molar Ratios of Metal Ions and HA
Figure 7.7 Ratios of measured DGT labile metal ions (β) in solutions containing various molar ratios of metal ion and humic acid (HA). β is defined as the concentration of DGT labile metal ions over total metal ion concentration (%) in the solution. Molar ratios between metal ions and HA are: 1.8:1.0, 1.0:1.0 and 1.0:1.8, as shown in the Figure.
Table 7.3 Effect of humic acid to metal ion molar ratio on the DGT labile metal ion fraction and free metal ion fraction calculated by SC-database model Molar Ratio of Metal Ion/Humic Acid 1.8:1.0 1.0:1.0 1.0:1.8 βCd (%) 51.0±2 32.8±4 18.7±3 βCd,theoretical (%) 56.5 33.2 14.8 βCu (%) 42.4±4 10.1±3 1.2±1 βCu,theoretical (%) 46.1 11.4 1.81 Note: βCd and βCu are the DGT labile fraction of Cd and Cu. The values shown here were the averages of 7 replicate experiments. βCd,theoretical and βCu,theoretical are the free metal ion fraction of Cd and Cu calculated by SC-database model 267.
160 Chapter 7
It was also found that, for a given HA/metal ion molar ratio, the DGT labile metal fractions obtained for Cd2+ were always greater than that for Cu2+. This result can be explained by the stability constant for each metal complex (Table 7.1). The stability constant of HA-Cd (LogK = 7.18) was smaller than that of HA-Cu (LogK = 7.99).
7.3.2.3. Speciation Measurement of Labile Metals in the Presence of Tannic Acid
Figure 7.8 shows the effect of the molar ratio of TA/metal ion on the ratio of DGT-labile concentration to the total concentration for Cd and Cu (βcd and βcu). It was found that both
βcd and βcu decreased as the molar ratio of TA/metal ion increased, which was expected.
When the metal ion/TA molar ratio increased from 1.0:1.8 to 1.0:1.0 to 1.8:1.0, the percentages of the DGT labile metal ion, measured, increased from 38.8% to 47.5% to
64.3% for Cd and 58.4% to 66.7% to 70.1% to for Cu.
Table 7.4 shows the DGT-labile metal fractions compared with the theoretical inorganic fractions. It was found that the DGT-labile Cu fractions, βCu, were substantially less than the theoretical values for all molar ratios used. The most likely reason for this was a difference between the stability constant used in the speciation software and that of the sample of tannic acid used. However, these results may be explained by the complicated interactions between the metal ions and the tannic acid. Possible physical and chemical absorptions of the metal ions to the tannic acid may keep some metal ions as “bound” ions258. It is not possible to suggest a likely mechanism for speciation of the TA complexes based on this data. However, given the results obtained for the HA and EDTA complexes in the previous sections, the mechanism based on retarded diffusion of the complexes compared with the inorganic metal ions is most likely.
161 Chapter 7
100 1.8:1.0 90 1.0:1.0
(%) 1.0:1.8 β 80 70 60 50 40 30 20 10 DGT Labile Metal Fraction 0
Cd Cu Molar Ratios of Metal Ions and TA
Figure 7.8 Ratios of measured DGT labile metal ions (β) in solutions containing various molar ratios of metal ion and tannic acid (TA). β is defined as the concentration of DGT labile metal ions over total metal ion concentration (%) in the solution. Molar ratios between metal ions and TA are: 1.8:1.0, 1.0:1.0 and 1.0:1.8, as shown in the Figure.
Table 7.4 Effect of tannic acid to metal ion molar ratio on the DGT labile metal ion fraction and free metal ion fraction calculated by SC-database model Molar Ratio of Metal Ion/Tannic Acid
1.8:1.0 1.0:1.0 1.0:1.8
βCd (%) 64.3±2 47.5±3 38.8±2
βCd,theoretical (%) - - -
βCu (%) 70.1±1 66.7±3 58.4±4
βCu,theoretical (%) 92.2 86.7 78.9
Note: βCd and βCd are the DGT labile fraction of Cd and Cu. The values shown here were the averages of 7 replicate experiments. βCd,theoretical and βCu,theoretical are the free metal ion fraction of Cd and Cu calculated by SC-database model 267.
162 Chapter 7
No βCd,theoretical was given in Table 7.4 because of the lack of a stability constant for TA-
Cd. It has been reported previously that no significant complexation occurs between
2+ 258 between Cd and TA . However, our results do not support this, as the βCd, values obtained were even smaller than that of βCu, for each given molar ratio of TA/metal ion.
This means that Cd2+ was bound more tightly to our sample of tannic acid than Cu2+ was.
Comparing the results shown in Tables 7.2 and 7.3, it can be seen that the DGT labile metal fractions, in the presence of TA, were higher than that in the presence of HA, especially for Cu. This result indicates that the interactions between TA and metal ions were not as strong as that of HA. This result agrees with Ross’s results 278 where it was reported that TA belongs to the category of weaker natural binding ligands compared to the HA ligand, which is rich in both primary amines and carbohydrates.
7.3.2.4. Speciation Measurement of Labile Metals in the Presence of Glucose
Figure 7.9 shows that the DGT-labile metal fractions for both Cd and Cu (βCd and βCu) decreased as the molar ratios of metal ion/glucose decreased. The 1.8:1.0 ratio for both metals and the 1.0:1.0 ratio for Cu gave DGT-labile fractions close to 100%. The strongest binding, 1.0:1.8 for Cd, gave a DGT-labile fraction of only about 80%. These observations demonstrate that glucose weakly binds the metal ions in a way to make them non-DGT-labile. This effect was larger for Cd2+ than for Cu2+. For a given molar ratio, the DGT-labile metal fractions, obtained for both Cd and Cu in the presence of GL, were much higher than that obtained in the presence of HA, TA or EDTA, due to the weaker interactions between the metal ions and the GL 264, 279. Although there is no theoretical calculation for comparison, given that the metal glucose species are likely to diffuse much more slowly through the dialysis membrane than the inorganic species, it is again likely
163 Chapter 7 that the DGT-labile fractions measured here largely represents the inorganic metal ion fraction 263. It is possible that mechanism (4), dissociation of the complex during diffusion through the dialysis membrane, is important here given the weak complex formed with the glucose ligands.
1.8:1.0 100 1.0:1.0 90 1.0:1.8
(%) 80 70 60 50 40 30 20 10
DGT Labile Metal Fraction β 0
Cd Cu Molar Ratios of Metal Ions and GL
Figure 7.9 Ratios of measured DGT labile metal ions (β) in solutions containing various molar ratios of metal ion and glucose (GL). β is defined as the concentration of DGT labile metal ions over total metal ion concentration (%) in the solution. Molar ratios between metal ions and GL are: 1.8:1.0, 1.0:1.0 and 1.0:1.8, as shown in the Figure.
7.3.2.5. Speciation Measurement of Labile Metals in the Presence of
Dodecylbenzenesulfonic Acid
The effect of the molar ratio of metal ion/DBS on the DGT-labile metal fraction is given in Figure 7.10. It can be seen that the DGT-labile metal fractions for both Cd (βcd values are 81.5%, 64.8% and 52.8% for Cd/DBS molar ratios of 1.8:1.0, 1.0:1.0 and 1.0:1.8) and
Cu (βcu values are 96.0%, 84.3% and 75.7% for Cu/DBS molar ratios of 1.8:1.0, 1.0:1.0 164 Chapter 7 and 1.0:1.8) decreased as the molar ratios of metal ion/DBS decreased, particularly for Cd.
This indicated mid-strength binding for Cd2+ and somewhat weaker binding for Cu2+. The binding strength for both Cd and Cu seemed to be between that for the TA and glucose ligands. Like the TA case, it seems that the most likely mechanism responsible for speciation of the metal-DBS complex is slow diffusion through the dialysis membrane, although there may be some dissociation occurring as well.
100 1.8:1.0 1.0:1.0
(%) 1.0:1.8 β 80
60
40
20
DGT Labile Metal Fraction 0
Cd Cu Molar Ratios of Metal Ions and DBS
Figure 7.10 Ratios of measured DGT labile metal ions (β) in solutions containing various molar ratios of metal ion and (DBS). β is defined as the concentration of DGT labile metal ions over total metal ion concentration (%) in the solution. Molar ratios between metal ions and DBS are: 1.8:1.0, 1.0:1.0 and 1.0:1.8, as shown in the Figure.
Based on this work, the PSS-DGT device seems to largely measure the inorganic fractions of metal ions in solutions only. Metal complexes with EDTA, HA and TA are non-DGT- labile while complexes with glucose and DBS also seem likely to not be measured, although this has not been confirmed here. For binding ligands with greater stability constants, such as EDTA, the non-complexed fraction was a lot lower and therefore so
165 Chapter 7 was the DGT-labile fraction. The order of increasing DGT-lability of the complexes was
EDTA < HA < TA < DBS < glucose for both Cu and Cd. This speciation is influenced largely by retardation of the flux of the complex through the dialysis membrane.
The following section describes the results of field deployments of the PSS DGT device in natural waters, where a complicated mixture of ligands can potentially influence the DGT- labile measurements. The only previous field study to investigate the speciation of the
DGT-measurements used the conventional Chelex 100 DGT devices and found that 55.2% of the Cu complexes with organic matter were DGT-labile 23. These results suggest that with the PSS DGT device this is likely to be much lower, if not zero.
7.3.3. Field Deployments
The newly developed PSS DGT devices were deployed at several sites for in situ trace metal ion speciation. Both fresh (Parkwood Pond and Loders Creek) and saline water
(Runaway Bay Marina and Biggera Waters) sites were chosen. Detail experiment procedures were given in the experimental section 7.2.4 of this chapter.
In order to obtain the diffusion coefficients for inorganic Cu2+ and Cd2+, the matrices of each site were first determined. The composition of these matrices is given in Table 7.5.
Synthetic solutions were then prepared matching the major cation composition of the water at each site. The diffusion coefficients of Cd2+ and Cu2+ for each test site were measured in these synthetic solutions according to the previously described method
(Section 2.3.5) and are shown in Table 7.6. The magnitude of diffusion coefficients obtained was dependent on total ionic strength. The value of diffusion coefficients obtained from the salty waters (Runaway Bay Marina and Biggera Waters) were much
166 Chapter 7 smaller than those obtained from fresh waters (Parkwood Pond and Loders Creek) due to the high ionic strength of the former.
Table 7.5 Major cation concentrations (mM), dissolved organic carbon (DOC, mgC l-1), salinity, pH and water temperature (oC) on the DGT deployment sites Measured Parameters
Test Sites [K+] [Na+] [Ca2+] [Mg2+] DOC pH Salinity Temp (mM) (mM) (mM) (mM) (mgC l-1) (ppm) (oC)
Runaway Bay 13.2 530 32.1 43.6 0.93 8.1 34.3 26.2 Marina
Biggera 14.7 524 26.3 72.2 1.1 7.9 35.7 25.7 Waters
Parkwood 0.56 1.1 0.41 1.9 12 5.2 0.23 28.3 Pond
Loder Creek 0.080 0.87 1.4 0.18 7.5 6.5 0.31 25.3 Note: cation concentrations were measured by AAS after appropriate dilutions. Other parameters were measured at the beginning, middle and the end of the DGT deployment and averaged.
Table 7.6 Diffusion coefficient of metal ions in different test sites
Diffusion Coefficient
2+ 2+ Test Sites Dm (Cd ) Dm (Cu ) (cm2s-1) (cm2s-1)
Runaway Bay Marina 0.30×10-6 0.22×10-6
Biggera Waters 0.30×10-6 0.21×10-6
Parkwood Pond 1.0×10-6 0.89×10-6
Loders Creek 1.4×10-6 1.0×10-6
Note: diffusion coefficients of Cd2+ and Cu2+ in the membrane in the tested sites were determined by the measurements of mass diffused through the membrane in solutions containing NaCl and MgCl2 of concentrations equivalent to the ionic strengths of the sites using the diffusion cell in Chapter 2.
167 Chapter 7
The results obtained from the field deployment are summarised in Table 7.7 for Cu and
Table 7.8 for Cd. For the seawater sites, the 0.45 µm-filterable Cu components determined by ICP-MS were found to be very similar: 1.20 ppb for the Biggera Waters site and 1.30 ppb for the Runaway Bay Marina site. The DGT-labile Cu concentrations were 0.55 ppb for both the Biggera Waters site and the Runaway Bay Marina site. These data represent 46% and 42% of the 0.45 µm-filterable Cu, respectively. These percentages are effectively the same when experimental uncertainties are taken into consideration, probably because the sites are part of the one well-flushed and mixed coastal lagoon (Gold Coast Broad Water).
Table 7.7 DGT labile and 0.45 µm-filterable concentrations of Cu
* * ** Test Sites CDGT (ppb) CF (ppb) β (%)
Biggera Waters 0.55±0.04 1.20 46
Runaway Bay Marina 0.55±0.03 1.30 42
Parkwood Pond 0.050±0.002 0.61 8.2
Loder Creek 0.025±0.004 0.51 4.9
* Note: CDGT and CTF are the DGT labile metal ion concentration and the total filterable metal ion concentration measured by ICPMS respectively. Data presented here are the mean values of three replicates. ** β is defined as the DGT labile metal ion concentration (CDGT ) divided by the total filterable metal ion concentration (CF).
For the seawater sites, both the 0.45 µm-filterable and DGT-labile Cd concentrations were much lower when compared to the Cu. The 0.45 µm-filterable Cd components for the
Biggera Waters and the Runaway Bay Marina sites were found to be below the detection
168 Chapter 7 limits. The DGT labile Cd ion concentrations were 0.043 ppb for the Biggera Waters site and 0.039 ppb for the Runaway Bay Marina.
Table 7.8 DGT labile and 0.45 µm-filterable concentrations of Cd
* * ** Test Sites CDGT (ppb) CF (ppb) β (%)
Biggera Waters 0.043±0.01 <0.6 -
Runaway Bay Marina 0.039±0.004 <0.6 -
Parkwood Pond 0.019±0.008 <0.4 -
Loder Creek 0.014±0.002 <0.4 - * Note: CDGT and CTF are the DGT labile metal ion concentration and the total filterable metal ion concentration measured by ICPMS respectively. Data presented here are the mean values of three replicates. ** β is defined as the DGT labile metal ion concentration (CDGT ) divided by the total filterable metal ion concentration (CF).
For the freshwater sites, the 0.45 µm-filterable Cu and Cd were much lower than in the seawater sites. The DGT-labile Cu concentrations obtained from the fresh water sites were 0.050 ppb for Parkwood Pond and 0.025 ppb for Loders Creek. The DGT-labile Cd concentrations obtained from the freshwater sites were 0.019 ppb for Parkwood Pond and
0.014 ppb for Loders Creek. These DGT-labile metal concentrations were much lower than those at the seawater sites.
The freshwater sites had higher DOC levels and also had lower DGT-labile concentrations compared with the seawater sites (Tables 7.5 and 7.7). DOC has been used as an indicator of the amount of complexation with organic matter that is likely to occur 280, 281. Given the results in the previous sections of this chapter, it is likely that the low DGT-labile measurements are due to a higher proportion of the Cu and Cd being complexed to organic matter and therefore becoming non-labile to the PSS DGT device. Of course this should be studied in much more detail as part of future studies. 169 Chapter 7
7.4. CONCLUSIONS
It was demonstrated that the dialysis membrane/PSS DGT device was capable of measuring DGT labile fraction of metal ions. The measurement in solutions containing various complexing ligands showed good agreement with SC-database model calculations.
The measured DGT labile metal ion fractions were determined by the four mechanisms as described in Section 7.3.2.
The application of the DGT technique in natural waters validated its speciation ability.
The site at the Runaway Bay Marina showed higher concentrations of copper because of the release of antifouling paints from boats berthing at the site. At the site with high lever of DOC, the Parkwood Pond site, low DGT labile metal ion fraction was measured due to the complexation of metals to humic substances. Both the experiments in laboratory and in natural waters showed the potential of using this DGT device as a tool for speciation analysis of trace metals in waters.
170 Chapter 8
Chapter 8 Evaluation of the New Binding Phases Developed for Use in the Diffusive Gradients in Thin Films Technique
171 Chapter 8
8.1. INTRODUCTION
In the previous chapters, various new DGT binding phases were developed and validated for trace metal measurements. These are the first new DGT binding phases reported for trace metals since the Chelex 100 polyacrylamide gel was first described 14, 15. This chapter will compare the trace metal speciation characteristics of DGT using these new binding phases with Chelex 100 DGT devices. The methods used to characterise the PSS binding solution in Chapter 7 will be used. Comparison of the different binding phases was also carried out in field deployments. The known properties of each of the binding phases developed in this thesis are also compared.
8.2. EXPERIMENTAL
8.2.1. Diffusion Layer Preparation
Both polyacrylamide diffusive gel and the cellulose dialysis membrane diffusive layer were prepared according to the methods described in Chapters 2 and 6.
8.2.2. Binding Phase Preparation
Poly(acrylamide-co-acrylic acid) (PAM-PAA) gel, poly(acrylamidoglycolic acid-co- acrylamide) (PAAG-PAM) gel, Whatman P81 cellulose phosphate ion exchange membrane (P81), poly(4-styrenesulfonate) (PSS) solution and Chelex 100 polyacrylamide gel (Chelex 100) were employed as binding phases for this study. These binding phases were prepared in the same manner as described in previous chapters.
172 Chapter 8
8.2.3. DGT Measurements in Laboratory
DGT devices employing solid binding phases (Chelex 100, PAM-PAA, PAAG-PAM and
P81) were assembled using the conventional piston design 16 (Figure 2.1) by placing polyacrylamide gel diffusive layers (thickness, 0.08 cm and exposure area, 4.9 cm2) on top of each binding phases. The solution binding phase DGT devices were prepared using a special design by clamping a diffusive membrane on a plastic tube filled with 2.0 ml 0.020
M PSS solution and deployed in solutions in the way of the membrane side facing down
(Figure 6.1). The thickness and exposure area of the diffusive membrane were 0.0050 cm and 4.5 cm2 respectively.
Nine replicates of each of the above DGT devices were immersed in solutions containing
0.40 µM Cd2+ with varying EDTA concentrations (0.13 µM, 0.22 µM and 0.40 µM, making metal:ligand ratios of 3.0:1.0, 1.8:1.0 and 1.0:1.0) or varying humic acid concentrations (0.22 µM, 0.40 µM and 0.70 µM, making ratios of 1.8:1.0, 1.0:1.0 and
1.0:1.8). The replicates were deployed in groups of three with each group removed after
10, 24 and 36 hours. The same measurements were made on solutions of 0.70 µM Cu2+ with EDTA or humic acid concentrations varied to produce the same ratios. These solutions were prepared in 25 L volume to ensure that the depletion of metal ions by DGT devices was negligible and appropriate percentages of bound and unbound metal ions were close to equilibrium.
Metals accumulated in the solid/gel binding phases were measured by FAAS after elution of the solid binding phases in 10% HNO3 solution, whereas metals in PSS solution were directly measured by FAAS with PSS matrix matched calibration standards (as described in Section 6.2.7). The accumulated mass was calculated and then plotted vs time with the 173 Chapter 8 slope used to calculate C, given that all the other parameters were known. β was calculated as described in Chapter 7.
8.2.4. DGT Field Deployment
Each of the DGT devices were deployed in triplicate, using the field deployment apparatus shown in Figure 8.1, for 24 hours on each day over a 6 day period at Runaway Bay
Marina (Figure 7.3) and Parkwood Pond (Figure 7.5). All DGT devices were rinsed with deionised water thoroughly before disassembly to prevent contamination. Blanks were also analysed together with the deployed DGT devices as in Chapters 6 and 7.
Foam buoy
DGT devices
Anchor
Figure 8.1 Gel based DGT holders were fixed on a square shape foam buoy, floating on waters.
174 Chapter 8
Water samples were also collected from each site in acid-washed LDPE containers
(Nalgene) at the beginning, middle, and end of the DGT deployment. These samples were filtered immediately on-site through 0.45 µm cellulose nitrate membranes, acidified with
65% suprapur nitric acid (Merck) (2.0 ml acid per litre of sample) to pH ≈ 2, and analysed by ICP-MS (Agilent Technologies, 7500 Series, Germany) to obtain total dissolved trace metal concentrations (Chapter 2). Samples taken from seawater were analysed after appropriate dilution with matrix matching to minimise background interference. The detection limits were as described in 7.2.5.
To measure the diffusion coefficients of the Cu and Cd in the diffusive layers (i.e. polyacrylamide gel and cellulose nitrate dialysis membranes) in the natural waters investigated, the major ion compositions were measured and an artificial solution created.
For Runaway Bay Marina water 1.16 g KCl, 33.3 g NaCl, 2.8 g CaCO3 8.3 g and MgCl2
6H2O were dissolved in 800 ml of deionised water. For the Parkwood Pond water 0.036 g
KCl, 0.088 g NaCl, 0.071 g CaCO3 and 0.35 g MgCl2 6H2O were dissolved in 800 ml deionised water. 10.0 ml of 1000 ppm Cd2+ or Cu2+ was added to the solutions before they were diluted to one litre total volume. The diffusion coefficients of the metal ions in the two diffusive layer for each water were measured using these synthetic solutions according to the previously described method in Chapter 2.
8.3. RESULTS AND DISCUSSION
8.3.1. Measurement of DGT Labile Metal Ions in the Presence of Ligands
In this section, the effect of the presence of ligands in the sample solution on the measurement of the DGT labile metal fraction (β) by different DGT binding phases was investigated.
175 Chapter 8
8.3.1.1. Measurement of DGT Labile Metal Ions in the Presence of EDTA
The effect of the presence of EDTA on the DGT-labile metal fraction for the different binding phases is described here. EDTA is a good ligand to study because the thermodynamic properties of EDTA-metal ion complexes are well characterised and they are very strong complexes (Chapter 7). This makes it possible for us to accurately calculate the inorganic metal concentrations with varying metal/EDTA ratios. The DGT- labile metal fractions experimentally obtained from different binding phases were compared with the theoretical free metal ion fractions calculated by SC-database model
267. The results are given in Tables 8.1 and 8.2 for Cd and Cu, respectively.
Table 8.1 Effect of molar ratio Cd2+ to EDTA on the DGT labile metal ion fraction measured by binding phases and free metal ion fraction calculated by SC- database model
Molar Ratio of Cd2+ to EDTA
Fraction (%) 3.0:1.0 1.8:1.0 1.0:1.0
βtheoretical 56.1 37.4 0
βChelex 100 54.5±3 35.6±3 0
βPAM-PAA 52.6±2 34.1±4 0
βPAAG-PAM - - -
βP81 53.8±4 35.9±5 0
βPSS 55.0±3 36.1±2 0 See note below.
Table 8.1 shows the comparison of the theoretical free metal ion fractions (βtheoretical) with
DGT-labile fractions of Cd2+ for each binding phase. Except for PAAG-PAM binding
176 Chapter 8 phase, which does not bind Cd2+, as mentioned previously, the DGT-labile fractions obtained from all the binding phases were not significantly different (α=0.125, n=3, t-test) to the theoretical free metal ion fraction, βtheoretical for each ratio investigated. This indicates that the DGT-labile fractions obtained were the same as the inorganic metal fraction. The contribution to the DGT-labile fraction by the metal ions that result from dissociation of the EDTA-Cd2+ complex were negligible, as demonstrated by the results for the 1:1 experiment.
Table 8.2 Effect of molar ratio Cu2+ to EDTA on the DGT labile metal ion fraction measured by binding phases and free metal ion fraction calculated by SC- database model
Molar Ratio of Cu2+ to EDTA
Fraction (%) 3.0:1.0 1.8:1.0 1.0:1.0
66.7 44.4 65.1±5 43.0±6 62.2±3 42.0±4 60.6±4 41.5±2 64.3±2 42.3±3 66.1±2 43.5±1 Note: βChelex 100, βP81, βPAM-PAA, and βPSS are the DGT-labile fractions of Cu and Cd measured by binding phases as described. The values shown here were the averages of 3 replicate experiments. 2+ 2+ 267 βtheoretical is the free Cu or Cd fraction of calculated by SC-database model . 2+ Table 8.2 shows the comparison of βtheoretical to DGT-labile fractions of Cu for each binding phase. Once again, the DGT-labile fractions measured by each binding phase, including PAAG-PAM, were not significantly different to βtheoretical for each molar ratio 177 Chapter 8 (p<0.125, n=3 means only, t-test). This again suggests that the DGT-labile fractions obtained were the inorganic Cu fractions. These results were interesting. Two factors largely determine the measurement of metal- EDTA complexes: diffusion coefficients relative to the inorganic metal species and the ability of the binding functional group to remove metal ions from the complex with EDTA. It has been established, in the previous chapter, that metal-EDTA complexes are not labile with the PSS DGT device because the complexes diffuse so slowly through the dialysis membrane (>500x more slowly than inorganic metal species). However, previous measurements of diffusion coefficients of metal complexes through the polyacrylamide diffusive gel have found the diffusion of even metal-humic acid complexes occur at only 10x less than the free ionic species 149. Given that metal-EDTA complexes should have a higher diffusion coefficient than the HA complexes (as they have much lower molecular masses) it might be expected that diffusion retardation would not be a large enough discriminating effect (perhaps 2-3x only) to explain the results obtained. However, this flux retardation effect in combination with the apparent inability of binding agents to remove metal ions from the EDTA complex could explain these results. Only the Chelex and PSS binding layers have stability constants likely to allow competition with the EDTA ligands, however, as no β value is higher than the theoretical value (which would happen if the complexes were labile), it seems as if this removal of the metal ions does not occur to a significant extent at all. If the binding layer can not remove the metal ions from the complex then the diffusive gradient will not be maintained. Consequently, the technique will not operate in a kinetic mode but in an equilibrium mode instead, which will reduce the flux even further. A combination of these effects is likely to be the reason why no difference was observed between the DGT-labile fractions for any of the binding layers used, given the range of RSD values observed for the DGT measurements (up to 14%). 178 Chapter 8 Consequently, only inorganic metal species were DGT-labile for each binding layer investigated. 8.3.1.2. Measurement of DGT Labile Metal Ions in the Presence of Humic Acid The effect of the presence of humic acid on the DGT-labile metal fraction for the different DGT binding phases was investigated and compared with the theoretical free metal ion fractions calculated by SC-database model 267. The results for Cd with the PSS binding phase are taken from Chapter 7. Humic acids are a major category of natural organic matter and have been reported to readily bind trace metals like Cd and Cu. The results are given in Tables 8.3 and 8.4 for Cd and Cu, respectively. Table 8.3 Effect of molar ratio Cd2+ to humic acid on the DGT-labile metal fraction measured by various binding phases and the theoretical free metal fraction. Molar Ratio of Cd2+ to Humic Acid Fraction (%) 1.8:1.0 1.0:1.0 1.0:1.8 βtheoretical 56.5 33.2 14.8 βChelex 100 47.2±6 32.7±4 13.8±3 βPAM-PAA 50.6±3+ 31.1±4 11.1±3 βPAAG-PAM - - - βP81 48.6±3 30.8±4 12.9±2 βPSS 51.0±2 32.8±4 18.7±3 See note below. 2+ 2+ An increase in the molar ratio of Cd /HA and Cu /HA lead to an increase in βtheoretical, calculated by the SC-database model, as expected (see Chapter 7). A very similar trend was observed for the DGT-labile metal fractions for all binding phases, except for PAAG 179 Chapter 8 PAM with Cd which was previously found to be incapable of binding Cd2+ (Chapter 4). Furthermore, the DGT-labile fractions were not significantly different (p<0.125, n=3 means only, t-test) to the theoretical values. However, in some cases the DGT measurements were all either higher or lower than the theoretical value. Therefore, this should be evaluated again with more replication so that a more powerful statistical comparison can be utilised. The fact that the DGT-labile fractions are not significantly higher than the theoretical fractions suggests that the metal-HA complexes are not measured significantly for any of the DGT devices, with either the polyacrylamide or dialysis membrane diffusive layer. Table 8.4 Effect of molar ratio Cu2+ to humic acid on the DGT-labile metal fraction measured by various binding phases and the theoretical free metal fraction. Molar Ratio of Cu2+ to Humic Acid Fraction (%) 1.8:1.0 1.0:1.0 1.0:1.8 βtheoretical 46.1 11.8 1.8 βChelex 100 44.3±4 15.2±5 <2.6 βPAM-PAA 42.5±3 9.2±4 <1.4 βPAAG-PAM 40.2±6 11.0±2 <1.4 βP81 41.1±3 10.6±4 <1.9 βPSS 45.4±4 14.1±3 <1.0 Note: βChelex 100, βP81, βPAM-PAA, and βPSS are the DGT-labile fractions of Cu and Cd measured by binding phases as described. The values shown here were the averages of 3 replicate experiments. 2+ 2+ 267 βtheoretical is the free Cu or Cd fraction of calculated by SC-database model . These results are likely to come about largely due to the flux of humic acid complexes being considerably slower than that of the inorganic metal species (>10x for the 180 Chapter 8 polyacrylamide diffusive layer and >500x for the dialysis membrane diffusive layer). Given the RSD values typical of the DGT measurements (>6%) it is unlikely that a 1 in 10 increase will result in a significantly different measurement. This will be the case even if the binding agents are able to remove the metal ions from the HA complexes, which should happen to some extent. It remains to be seen whether these different binding phases measure different fractions in natural waters, which have a much more complex speciation characteristics than the laboratory solutions. 8.3.2. Field Deployments In order to compare the performance characteristics of DGT devices with different binding phases all measurements were performed at the same time for each site. In practice, for a given test site, all DGT deployments were carried out within the same time period of 6 days. For each binding phase, three DGT devices were removed every 24 hours for the purpose of constructing a mass vs. time curve, for which the average concentration for the deployment period was calculated by using the known values of A and ∆g (from slope = CDA/∆g). However, the diffusion coefficients for Cu2+ and Cd2+ through each diffusion layer were measured in synthetic solutions similar to those at each deployment site (see section below). The concentrations obtained are described in section 8.3.2.2. 8.3.2.1. Measurement of Diffusion Coefficients The diffusion coefficients used for DGT calculation depend on the type of diffusive layer employed and the sample matrices. The composition of the waters at both the sites used were analysed with the results summarised in Table 8.5. Synthetic solutions were created to match the major ion compositions (i.e. no organic complexes). The diffusion 181 Chapter 8 coefficients of the metal ions in each diffusive layer were measured using the synthetic matrices (Table 8.6). Table 8.5 Concentrations of major cations (mM), dissolved organic carbon (DOC, mgC l-1), pH and water temperature (oC) at Runaway Bay Marina and Parkwood Pond on the Gold Coast. Measured Parameters K+ Na+ Ca2+ Mg2+ DOC Salinity Temp. Test Sites pH (mM) (mM) (mM) (mM) (mgC l-1) (ppm) (oC) Runaway Bay Marina 15.5 570 28.0 41.1 0.84 8.2 35.0 22 Parkwood Pond 0.49 1.5 0.71 1.7 9.2 6.1 0.25 25 Table 8.6 Diffusion coefficients of Cd2+ and Cu2+ of the polyacrylamide and the dialysis membrane diffusive layers at different test sites. Diffusion Coefficient 2+ 2+ 2+ 2+ Test Sites Dg (Cd ) Dg (Cu ) Dm (Cd ) Dm (Cu ) (cm2s-1) (cm2s-1) (cm2s-1) (cm2s-1) Runaway Bay Marina 1.4×10-6 1.6×10-6 0.30×10-6 0.21×10-6 Parkwood Pond 2.0×10-6 2.1×10-6 0.92×10-6 0.81×10-6 2+ 2+ Note: D g and Dm are the diffusion coefficients of Cd and Cu in the polyacrylamide gel and the dialysis membrane respectively. It was found that the magnitude of diffusion coefficients decreased as the total ionic strength increased for both the polyacrylamide gel and the dialysis membrane diffusive layers. Therefore, the diffusion coefficients obtained for the Runaway Bay Marina were smaller than those obtained for Parkwood Pond. For a given matrix, it was found that the 182 Chapter 8 diffusion coefficients in dialysis membrane were significantly smaller than those in polyacrylamide gel, although the difference was much greater in the saline waters. Given these diffusion coefficients, the average DGT-labile concentrations for Cu and Cd for each binding phase at both sites were able to be calculated. 8.3.2.2. Measurement of DGT-Labile Metal Concentration and Fraction Runaway Bay Marina Table 8.7 contains a summary of the DGT-labile concentrations and fractions (if possible) of Cd2+ and Cu2+ obtained using the various binding phases in Runaway Bay Marina. 95% confidence interval (95% CI) values and detection limits for each binding phase were also shown, as well as the 0.45 µm-filterable concentration. For Cd the DGT-labile concentrations measured using the various binding phases, except PAAG-PAM, which does not bind Cd2+, were from 0.015 to 0.026 µg l-1. A Oneway ANOVA with a Least Squares Difference post-hoc analysis (df = 84) determined that the measurements obtained with the various binding layers were not all the same. Chelex was significantly different from the P81 binding layer (p=0.046) and was highly significantly different from the PAM-PAA binding layer (p<0.000). The P81 layer was also significantly different from the PAM-PAA (p=0.048). The PAM-PAA binding layer was also significantly different from the PSS binding phase (p=0.004). This means that there were essentially three different fractions of DGT-labile Cd: Chelex, P81 and PAM-PAA. The PSS binding phase was not significantly different from the Chelex and P81 binding layers, meaning that it was an intermediate DGT-labile fraction between the different Chelex and P81 fractions. The 0.45 µm-filterable Cd2+ concentration was not able to be 183 Chapter 8 estimated because it was below the detection limit. Therefore, the DGT-labile fractions were not able to be estimated as a percentage of the 0.45 µm-filterable fraction. Table 8.7 DGT labile Cd2+ and Cu2+ fractions measured using various DGT binding phases and their total filterable concentrations (TFC) measured by ICP-MS in Runaway Bay Marina on the Gold Coast. Binding Phases Specifications Chelex 100 PAM-PAA PAAG-PAM P81 PSS CCd (ppb) 0.026±0.010 0.015±0.010 - 0.020±0.010* 0.023±0.011* 95% CI 0.021-0.030 0.013-0.018 - 0.017-0.024 0.018-0.028 # LODCd,(ppb) 0.010 0.010 - 0.010 0.010 -1 FCCd (ppb) <0.6 µg l βCd (%) nc nc nc nc nc CCu (ppb) 0.25±0.11 0.18±0.09 0.14±0.08 0.23±0.09 0.23±0.07 95% CI 0.20-0.30 0.14-0.21 0.11-0.18 0.19-0.27 0.20-0.26 # LODCu (ppb) 0.05 0.05 0.05 0.05 0.05 -1 FCCu (ppb) <0.9 µg l βCu (%) nc nc nc nc nc 2+ 2+ Note: CCd,and CCu,are DGT labile concentration of Cd and Cu . FCCd and FCCu are 2+ 2+ 0.45 µm membrane filterable concentration of Cd and Cu (ppb). βCd and βCd are the DGT labile fraction of Cd2+ and Cu2+ measured. The values shown here were the averages of 3 replicate experiments. 95% CI is 95% confidence interval. nc means not able to be calculated # the detection limits were conservatively set at these levels even though some were actually lower * one extreme outlier was excluded from analysis and attributed to contamination 184 Chapter 8 Interpretation of these results is quite difficult, given the knowledge currently available. There was a very high power (df = 84) in this statistical analysis, so these results are likely to be accurate. The discussion of the mechanisms responsible for producing the laboratory results (in section 8.3.1) is not sufficient to explain the field results. It was concluded previously that the PSS DGT device was unlikely to measure organic complexes because the diffusion coefficients are considerably lower than the diffusion coefficients for the inorganic species. A consequence of this conclusion is that PSS DGT devices will only measure inorganic species. If this conclusion is correct then the Chelex DGT devices may be measuring some of the metal complexed by organic matter (given that the mean was slightly higher although not significantly so) which has been reported previously 282. However, this also means that the PAM-PAA binding layer (which gave a significantly different and lower concentration) does not measure all of the inorganic Cd species present. This would be a surprising result, if correct. One possible explanation is that the PAM-PAA functional groups do not bind some inorganic Cd species, present in + seawater, strongly enough. The SC-database calculates 1.57% CdCl and 6.35% CdCl2 as the main inorganic complexes present apart from aqua complexes. No certain conclusion can be made based on these results; further research is clearly required to fully determine the reasons for the different DGT-labile fractions measured. For Cu, the DGT-labile concentrations measured using the different binding phases were found to be in the range of 0.144 to 0.246 µg l-1. The 0.45 µm-filterable concentrations were below the detection limit, therefore the DGT-labile fractions were not able to be estimated. A Oneway ANOVA with a Least Squares Difference post-hoc analysis (df = 106) determined that the measurements obtained with the various binding layers were again not all the same. The Chelex binding layer was highly significantly different from the PAAG-PAM layer (p<0.000) and significantly different from PAM-PAA (p=0.008). 185 Chapter 8 Likewise, the P81 binding layer was significantly different from the PAAG-PAM (p=0.001) and PAM-PAA (p=0.039) layers. Finally, the PSS binding phase was also significantly different from the PAAG-PAM (p=0.002) and PAM-PAA (p=0.045) binding layers. Therefore there were two categories of results: the DGT-labile concentrations obtained with the Chelex, P81 and PSS binding phases were effectively the same and different from the DGT-labile concentrations of the PAAG-PAM and PAA binding layers, while the latter two gave the same results. Therefore, we have observed the same issues noted with the Cd measurements. The Chelex, PSS and P81 were once again similar, even more so than observed for the Cd measurements. The PAAG-PAM and PAM-PAA DGT devices give lower DGT-labile concentrations than do the PSS DGT devices. This means that either the PSS DGT devices do not measure only inorganic species, at least in seawaters, or the PAAG-PAM and PAM-PAA DGT devices do not measure all of the inorganic fractions. This phenomenon may be restricted to seawater measurements where the concentrations of inorganic ligands are very high. More research is also required for Cu DGT measurements. Parkwood Pond Table 8.8 summarises the DGT-labile concentrations and fractions (if possible) of Cd2+ and Cu2+ obtained using the various binding phases at Parkwood Pond. 95% confidence interval (95% CI) values and detection limits for each binding phase were also shown. The 0.45 µm-filterable concentration of Cd was below the detection limit but Cu was measured. 186 Chapter 8 For Cd, the DGT-labile concentrations measured using the various binding phases, except PAAG-PAM, which does not bind Cd2+, were from 0.014 to 0.023 µg l-1. A Oneway ANOVA with a Least Squares Difference post-hoc analysis (df = 85) determined that the measurements obtained with the various binding layers were not all the same. The PSS DGT devices gave the highest concentration and were significantly greater than the concentration measured by the Chelex (p=0.034) and PAM-PAA (0.001) binding layers. The P81 binding layers were significantly different to only the PAA (p= 0.003) binding layers. This means that the binding layers were not clearly delineated according to the results obtained. The PSS binding layer formed one category and the PAM-PAA binding layer was very different. However, the P81 and Chelex binding layers appeared to be intermediate, with P81 not being significantly different from the PSS or Chelex layers, but was significantly different to the PAM-PAA layer, whereas the Chelex layer was only significantly different to the PSS layer. For Cu, the DGT-labile concentrations measured using the various binding phases ranged from 0.023 to 0.030 µg l-1. A Oneway ANOVA with a Least Squares Difference post-hoc analysis (df = 106) determined that the measurements obtained with the various binding layers were not all the same. The PAAG-PAM DGT device gave the highest concentration and were significantly greater than the concentration measured by the PAM PAA (0.012) binding layers. The P81 binding layers were also significantly greater than the PAA (p= 0.025) binding layers. The PSS and Chelex binding layers formed an intermediate group that were not significantly different from any other binding phases. Once again these results are difficult to explain given our current understanding of the speciation mechanisms. 187 Chapter 8 Table 8.8 DGT labile Cd2+ and Cu2+ fractions measured using various DGT binding phases and their total filterable concentrations (TFC) measured by ICP-MS in Parkwood Pond on the Gold Coast. Binding Phases Specifications Chelex 100 PAM-PAA PAAG- P81 PSS PAM CCd (ppb) 0.018±0.010 0.014±0.010 - 0.022±0.010 0.023±0.011 95% CI 0.013-0.022 0.011-0.017 - 0.019-0.025 0.018-0.028 # LODCd,(ppb) 0.010 0.010 - 0.010 0.010 -1 FCCd (ppb) <0.4 µg l βCd (%) nc nc - nc nc CCu (ppb) 0.025±0.011* 0.023±0.011 0.030±0.010 0.030±0.010 0.027±0.011 95% CI 0.0193-0.030 0.018-0.027 0.026-0.035 0.026-0.035 0.0224-0.0322 # LODCu (ppb) 0.015 0.015 0.015 0.015 0.015 -1 -1 FCCu (ppb) 0.4±0.0 µg l (LOD = 0.3 µg l ) βCu (%) 6.3 6.3 7.5 7.5 6.8 2+ 2+ Note: CCd,and CCu,are DGT labile concentration of Cd and Cu . FCCd and FCCu are 2+ 2+ 0.45 µm membrane filterable concentration of Cd and Cu (ppb). βCd and βCd are the DGT labile fraction of Cd2+ and Cu2+ measured. The values shown here were the averages of 3 replicate experiments. 95% CI is 95% confidence interval. nc means not able to be calculated # the detection limits were conservatively set at these levels even though some were actually lower * one extreme outlier was excluded from analysis and attributed to contamination The DGT-labile fraction of the 0.45 µm-filterable concentration was only able to be estimated for Cu in Parkwood Pond and was found to be 6.3-7.5%. This is lower than 188 Chapter 8 many of the fractions reported in the literature. This is probably because Parkwood Pond has a dissolved organic carbon concentration of 9.2 mg l-1, which is quite high. Given that the DOC concentration is about 23,000 times higher than the 0.45 µm-filterable concentration it is obvious that most of the Cu has been complexed by natural organic matter. This supports the notion that most organically complexed metals are not DGT- labile. Overall Assessment There are few trends apparent in the results discussed, even when compared with the results from the Runaway Bay Marina deployment. About the only clear one is that the PAM-PAA DGT devices generally give lower concentrations than the other binding layers; although it does not always give the numerically lowest concentration, it has consistently been part of a statistically equivalent group that has given the lowest concentrations. The conclusion made in the previous chapter, concerning the PSS DGT devices measurement of inorganic species only, has not been supported by these results in this chapter. All organic complexes will diffuse through the polyacrylamide layer more quickly than through the dialysis membrane, so some other explanation needs to be sought to explain data where the other binding agents provide a concentration that is lower than the PSS value. The most likely explanation is based on the strength of the binding phase (stability constant). Only the stability constant of the PSS for Cd and Cu is known so we can not predict the order or binding strength a priori. Nor do the experimental results described here provide a clear comparison. We do know that Chelex 100 binds strongly to transition metals, so it would seem very improbable that any of the other binding agents that have a polyacrylamide diffusive layer would give concentrations higher than Chelex DGT devices. 189 Chapter 8 The most likely explanation is therefore that this statistical analysis has produced Type I errors, where a difference between data has been found where none actually exists. The probability of this is equal to α, which was 5%. If α is set at the much more stringent value of 0.001 instead, many of the differences found no longer become significant. Even where these highly significant differences do exist, there are usually intermediate groups that are not significantly different from the groups with the highest and lowest values. The only result where a clear trend is apparent is Cu at Runaway Bay Marina. However, this clear delineation is not apparent at α= 0.001. The Cu measurements at Parkwood Pond also have no significant relationships at α= 0.001. The small range of the DGT- labile concentrations as a percentage of the 0.45 µm-filterable concentration (6.3-7.5%) obtained for this site, suggest that Type I errors have occurred here. Clearly further studies are required to determine whether these various binding phases do give significantly different results in different waterways. In particular, the possibility that the PSS DGT devices are capable of measuring metal complexes with natural organic matter needs to be explored. On the other hand, investigation of the possibility that apparently weaker binding agents like PAM-PAA do not measure all of the inorganic metal species may be a simpler experiment to perform using synthetic solutions. 8.4. COMPARISON OF IMPORTANT PROPERTIES OF THE NEW BINDING PHASES DEVELOPED IN THIS STUDY This chapter has been devoted to comparison of the binding phases developed in this study with the Chelex 100 binding phase. Therefore, a summary and general comparison is included here of all of the major findings and properties of the new binding phases developed for measurement of Cu and Cd by the DGT technique. Table 8.9 contains a simple comparison of all the binding phases for important properties that influence the 190 Chapter 8 operation of DGT. These properties are described in much more detail in the following sections. Table 8.9 Properties of binding phases that can adversely effect DGT measurements: means there are problems with this binding layer, means that this binding layer is close to ideal, ) means that there are some minor problems only. Binding Phase Interface Swelling Biofouling Elution Reusability between binding and diffusive layer Chelex 100 ) PAM-PAA PAAG-PAM P81 PSS ) * *There is no elution with PSS, but there are matrix issues for analysis. 8.4.1. Assembly of DGT Devices and the Interface between the Binding and Diffusive Layers Assembly of the DGT device should be undertaken in a convenient, reproducible manner. Likewise, the results obtained with the method should also be reproducible and accurate. The conventional DGT binding phase is Chelex 100 ion-exchange resin embedded polyacrylamide gel 14, 15. This binding gel was the first developed DGT binding phase and has been widely used since 14, 22, 165, 283-285. Assembly of a DGT device with Chelex 100 binding gel is not a trivial task. Users have to identify which surface of the binding gel is that which the Chelex resin has settled towards during polymerization, and that needs to be in contact with the diffusive gel 16. If the binding gel is assembled up-side down in the 191 Chapter 8 DGT device then the results will have a large bias towards underestimation of the concentration. The Chelex binding gel also does not form an ideal interface with the diffusive layer because the binding functional groups are not aligned continuously at the surface; the interface has an actual thickness, which in theory should produce a small bias towards underestimation. These problems will be more important for inexperienced users. The newly developed gel-based DGT binding phases, PAM-PAA gel and PAAG-PAM gel, are homogeneous binding phases. The binding sites are spread evenly through out the binding phases, unlike the Chelex 100 binding gel. Orientation in a particular direction is therefore not necessary. Furthermore, the interface between the diffusive and binding phase is ideal, with a continuous coverage of binding functional groups, which is what is assumed with calculation using the DGT equation. However, other problems concerning the assembly of DGT device that Chelex 100 binding gel experienced also exist for other gel-based binding phases (see below). The cellulose phosphate P81 ion-exchange membrane DGT binding phase is a commercially available membrane. This membrane has much better mechanical strength than that of gel based binding phases described above. The strong mechanical strength of the membrane makes assembly of the DGT device much easier than for gel based binding phases. Since the binding functional groups are homogenously distributed across the membrane, orientation is not something that has to be considered. The PSS liquid binding phase provides a perfectly homogeneous interface with the diffusive layer. It overcomes the swelling and fragility problems of gel-based binding phase and eliminates elution steps needed for all solid binding phases. It also minimises the biofouling effects during DGT deployments. However, this DGT device requires 192 Chapter 8 special preparation of the PSS polymer and the dialysis membrane. It also requires a unique DGT probe. 8.4.2. Swelling Effects Swelling is another important property of binding phases that affects the handling and ease of use of the DGT technique. The P81 membrane binding phase does not swell or shrink under any experimental conditions investigated. The PSS liquid binding phase increases its volume slightly when the DGT device is deployed in waters due to osmotic pressure between the binding solution and the sample solution. This change of the binding phase volume does not affect its application because the PSS solution is diluted to an appropriate volume after deployment and the binding solution concentration change does not affect its binding property (the actually used PSS solution concentration is higher than its optimum binding concentration, Chapter 6). The volume of Chelex 100 gel binding phase changes a little (10-20%) for the pH range from 2 to 9 where the swelling is constant at ionic -5 concentrations ranging from 10 M - 1.0 M (as NaNO3). However, the volume of PAM PAA gel and PAAG-PAM gel binding phases depends on the solution pH and ionic strength. PAM-PAA and PAAG-PAM gel binding phases swell rapidly at pH ~ 3 and 2 respectively. This was due to the transformation of the acidic form of the carboxylate functional groups to the basic form, which interact with water differently due to a change in the surface charges. The basic form has a negative charge and thereby interacts with water more strongly which produces the swelling. The highest degree of swelling was achieved at pH 6 for PAM-PAA gel and pH 5.4 PAAG-PAM gel. For PAM-PAA gel, the degree of swelling was essentially constant when the pH > 6 due to the completion of the conversion of acidic form of the gel to basic 193 Chapter 8 form, while PAAG-PAM gel shrunk rapidly when pH > 5.4 due to the increase of ionic strength attributed by the addition of NaOH. The degree of swelling for PAM-PAA gel and PAAG-PAM gel decreased rapidly as NaNO3 concentration increased due to a strong charge screening effect on the gel network, in which the electrostatic repulsion between adjacent strands of polymer are minimised, causing the strands to move closer together resulting in a polymer that has less capacity to absorb water 211. The degree of swelling of these two gel based binding phases needs to be stabilised before DGT assembly by keeping the gels in NaNO3 solutions whose ionic strength and pH are similar to those of the deployment sites. This equilibration needs to be undertaken for at least 24 h 154. 8.4.3. Biofouling Effects Biofouling will occur in all waters, where organisms adhere to and grow on the membrane that covers the diffusive layer, or is the diffusive layer. Figure 8.2 shows the biofouling growth on filter membrane after 1, 3 and 7 weeks 275. After a three week deployment in Runaway Bay Marina serious biofouling had occurred. Biofouling has two detrimental effects. Firstly, the presence of layers of organisms changes the diffusional properties of the DGT device by increasing the diffusive path-length. Secondly, some of the organisms will adsorb some of the metal ions, effectively removing them from solution and making them non-measurable by DGT. Figure 8.2 Filter membrane biofouling in Runaway Bay Marina. 1, 3 and 7 week deployments from left to right, respectively. 194 Chapter 8 Figure 8.3 shows the biofouling that occurs on the cellulose nitrate dialysis membrane used in the PSS DGT device. Clearly after 3 weeks the biofouling is much less severe compared with that shown in Figure 8.2. This is because the dialysis membrane is more hydrophilic 286 and has been treated to become resistant to biofouling 287. Furthermore, in long deployments, this problem was solved by replacing the membrane weekly by a fresh membrane. This can be done without disturbing the PSS solution. This allows considerable extension of the deployment period, which will be useful for ultra-trace measurements. Figure 8.3 Dialysis membrane biofouling. 1, 3 and 7 week deployments from left to right, respectively. 8.4.4. Reusability Theoretically, all gel based binding phases and P81 membrane binding phase can be regenerated and reused. However, only the P81 membrane binding phase has been demonstrated to be practically reusable. This is because gel based binding phases are too fragile and physical damage can not be avoided during the disassembly and elution processes. The P81 membrane binding phase can be practically regenerated and reused due to its excellent mechanical strength and sharply pH dependent metal binding properties (Section 5.3.1). 195 Chapter 8 8.4.5. Elution All solid binding phase DGT devices require elution steps before the measurement of analyte mass accumulated in the phases, which can be achieved by immersing the binding phases in a strong acidic solution. The elution factor is defined as the ratio of the eluted metal mass to the mass bound in the binding phase. The value was used for the calculation of metal concentrations in a tested sample solution 16. Therefore, the accuracy of DGT measurements depends on the elution efficiency. Unlike the solid binding phases, the liquid binding phase does not require an elution step since the metal in the liquid can be analysed directly by FAAS or ICP-MS methods, which meets the requirement of rapid and accurate in situ analysis. However, to measure the metal-PSS concentrations in the binding phase, a series of PSS matrix matched standard metal solutions are required for calibration. The solutions also have to be diluted to an appropriate level as the matrix has not been diluted by an elution step. 8.4.6. Valid Deployment Conditions and Metal Binding Properties This section compares the important properties of the binding phases developed as part of this study with those of the Chelex 100 gel at pH ~7 and low ionic strength (NaNO3 concentration of 1.0 × 10-5 M). The maximum binding capacities (µmole cm-2) of the binding phases under non-competitive conditions are listed in Table 8.10. The order of decreasing binding capacity is: PSS (13.5 for Cd and 13.0 for Cu) > PAAG-PAM gel (5.1 for Cd and 5.3 for Cu) > P81 (3.1 for Cd and 3.2 for Cu) > PAM-PAA gel (1.6 for Cd and Cu) > Chelex 100 gel (1.1 for Cd). This same order of capacity between the various phases was observed under different conditions of pH and ionic strength. The PSS has the 196 Chapter 8 highest binding capacities because it is a solution phase which mobile, and so a functional group at the interface with the diffusive layer can bind a metal ion and then diffuse away to be replaced with another functional group. The Chelex 100 binding layer has the lowest capacity because the functional groups that accumulate the metal ion are not continuous at the surface. The other three solid binding phases do have continuous layer of functional groups at their surfaces. Table 8.10 Binding capacities of binding phases Binding Phase Chelex 100 P81 PAM-PAA PAAG-PAM PSS* Maximum Capacity 16 1.1 3.1 1.6 5.1 13.5* (Cd, µmole cm-2) Maximum Capacity - 3.2 1.6 5.3 13.0* (Cu, µmole cm-2) *Binding capacity of a 2.0 ml 0.020 M PSS solution It is required that DGT binding phases accumulate metals over a wide range of conditions of pH and ionic strength in natural waters. Ionic strength affects the binding of metal ions to the binding phases. At higher ionic strength solutions, the binding metal capacities of the binding phases are lower due to the competition accumulation of matrix ions. Table 8.11 shows that the studied binding phases are applicable in ionic strength -5 equivalent to NaNO3 concentration range of 1.0 × 10 to 1.0 M. Most natural waters are within this range. The pH of solutions also affects the accumulation properties of the binding phases. At lower pH range, the major binding functional groups of the binding phases are in acidic forms, which are neutral and thus will not bind metals. As reported by Zhang 16 and investigated in previous chapters, the optimal working pH ranges of the binding phases for Cd measurement were 5.0 – 8.3 for Chelex 100 gel, 5.0 – 10 for PAM 197 Chapter 8 PAA, 4.2 – 10 for PAAG-PAM, 3.9 – 8.2 for P81 and 2.0 – 10 for PSS; pH ranges for Cu measurement were 5.0 – 9.0 for PAM-PAA, 3.0 – 9.0 for PAAG-PAM, 3.9 – 8.2 for P81 and 3.0 – 8.7 for PSS. Table 8.11 Valid deployment conditions Binding Phase pH Range for Cd pH Range for Cu Ionic Strength (NaNO3, M) Chelex 100 5.0-8.3 - 10-5-1.0 PAM-PAA 5.0-10 5.0-9.0 10-5-1.0 PAAG-PAM 4.2-10 3.0-9.0 10-5-1.0 P81 3.9-8.2 3.9-8.2 10-5-1.0 PSS 2.0-10 3.0-8.7 10-5-1.0 8.5. CONCLUSIONS The speciation characteristics of the varying binding phase DGT devices were evaluated. This evaluation provided important directions for their applications. With comparison to SC-database model it was found that the measured DGT labile fractions of metals by varying DGT systems in solutions containing EDTA were inorganic fractions. In solutions containing humic acid, it was suggested that the metal-HA complexes were not significantly measured by the various DGT systems with either polyacrylamide or dialysis membrane diffusive layer. The results from field deployment of the DGT devices were explained by comparing the various binding phase DGT measurements with the PSS measurements. Other properties of the binding phases were also evaluated. The advantages of using P81 ion exchange membrane over gel based binding phases were its reusability without significant lost of its binding metal properties (five times) and its 198 Chapter 8 constant volume and mechanical strength irrespective of pH and ionic strength. The use of PSS liquid binding phase provided a mobile, high binding capacity and broad pH working range (3.0 – 8.7) for DGT. There is no need for elution steps for PSS binding phase DGT system which is required for solid binding phase DGT. This simplification increased the accuracy of DGT analysis. The use of dialysis membrane diffusive layer and PSS binding phase minimised biofouling effects on DGT, which was a serious problem for solid binding phase DGT 161, 275 when using 0.45 µm filter membrane as a protect layer and polyacrylamide gel as a diffusive layer. The applicable pH and ionic strength range of the binding phases were compared, which gave information on deployment in natural waters. 199 Chapter 9 Chapter 9 General Conclusions 200 Chapter 9 This thesis describes the development and applications of the diffusive gradients in thin films (DGT) technique for in situ measurements of labile metal species in waters. The key elements of the technique are the efficient and selective binding of the analytes to a binding phase and effective diffusion in a well-defined diffusion layer (Chapter 1). Throughout the course of this work, we have demonstrated that for a DGT sensor, the performance of the system can be further improved by employing the series of homogeneous binding phases: binding hydrogels, binding membrane, and polymeric binding solution. The use of these new binding phases has lead to the developments of the DGT technique in terms of speciation ability, reproducibility, and processibility. Firstly, a novel poly(acrylamide-co-acrylic acid) copolymer hydrogel (PAM-PAA) was synthesised and found to be capable of selective binding of the transition metals such as Cu2+ and Cd2+ over alkali and alkaline earth metals (Chapter 3). This gel based binding phase was used with the diffusive gradients in thin-films technique (DGT) for trace metal analysis in natural waters. The gel was prepared by the controlled hydrolysis of polyacrylamide hydrogel in alkali solution. FTIR and elemental analysis indicated that the composition of the copolymer was approximately two units of acrylic acid for every unit of acrylamide in a syndiotactic-rich structure. The pKa was found to be 4.5. At pH above 5, the carboxylic acid groups were in the salt form and had a greater capacity to bind the transition metals Cu2+ (1.59 µmoles cm-2) and Cd2+ (1.56 µmoles cm-2). The degree of swelling (qw) of the gel under equilibrium conditions increased to 120 times that of its dried state at pH > 6, and was relatively stable when pH > 6. The degree of swelling was also found to be influenced by ionic strength. This swelling property limited its application in DGT. Dramatic increase or decrease of the gel volume may cause the breakage of upper layer of the DGT device, or incomplete coverage of the diffusive layer. Because of these swelling properties the gels were needed to be stored in appropriate 201 Chapter 9 conditions, similar to those of deployment, before being used with DGT. The optimum deployment conditions were at pH above 6 and under conditions of relatively stable ionic strength. Secondly, a new copolymer poly(acrylamidoglycolic acid-co-acrylamide) (PAAG-PAM) hydrogel was prepared with a 3:1 ratio of AAGA monomer units to AAm monomer units (Chapter 4). This gel was found to bind Cu2+ ions selectively with a binding capacity of 5.3 µmole.cm-2 for non-competitive uptake and 1.30 µmole.cm-2 for competitive uptake with other metal ions. The suitability of using this gel as a binding phase with the DGT technique was confirmed by obtaining a linear response between the accumulated mass and the metal uptake time. 95 - 100% recovery with a DGT uptake experiment was also observed. Similar swelling properties of this gel to PAM-PAA gel as described in Chapter 3 were observed. When the gel contains high percentage of water it becomes fragile. It also needs to be stored in a NaNO3 solution with similar ionic strength to the water solution in which the DGT devices are going to be deployed, to minimise the degree of the gel size changing. To overcome the limitations of this gel, binding phase with more stable shape to the change of pH and ionic strength are required. Thirdly, the Whatman P81 cellulose phosphate ion exchange membrane (P81) has been successfully used as the binding phase for DGT applications (Chapter 5). The performance of this new DGT binding phase was investigated for analysis of Cu2+ and Cd2+ in a synthetic lake water matrix. The ion exchange activity of the new binding phase can be regenerated and, therefore, reused in DGT application. This reuse of the phase lowered the cost of DGT analysis since the conventional binding phase in DGT, Chelex 100/polyacrylamide gel, was expensive to make due to the agarose derived cross linker expense. The new binding phase exhibited excellent mechanical properties and overcame 202 Chapter 9 many of the problems of the hydrogel based binding phases. Those problems included fragility caused handling difficulty, swelling or shrinking with the changes of pH or ionic strength conditions caused breakage of the diffusive gel or incomplete coverage of the upper layer. Perhaps the most significant aspect of this work is that it opens up the possibility of employing a new range of binding phases in DGT analysis, i.e. binding phases not limited to gel-based systems. There are a myriad of other solid ion exchange membranes and other binding materials available. This work has shown the feasibility of employing such materials in DGT technique. The limitations of this binding membrane may be the roughness of the solid surfaces causing that the contact between the diffusive and binding phases is not perfect, and the need for elution procedures. Fourthly, a new style of DGT device has been designed using a poly(4-styrenesulfonate) (PSS) solution binding phase, and a dialysis membrane diffusive layer (Chapter 6). The diffusion properties of the dialysis membrane and the binding properties of the PSS solution were characterised and found to be suitable for use with DGT. A new DGT device was designed and validated by demonstrating a linear mass vs. time relationship for Cd and Cu in synthetic waters and spiked natural waters (Cu only). The major advantages of this DGT device include a theoretically ideal mass transport by the well-defined diffusive layer available from massive production and mass accumulation by the mobile binding solution. The fragility and swelling problems of gel based binding phases were overcome. In addition, the DGT procedures were simplified by removal of elution steps, which were required for all solid binding phases. This simplification resulted in the increase of analytical accuracy. One of the important features of DGT, long-term integrative measurement, was implemented using this device, due to its better 203 Chapter 9 antifouling property, allowing it to be deployed for longer time. The only drawback is the need for matrix-matched calibrations for instrumental analysis. The speciation capability and field deployment of this new DGT device were investigated in details in Chapter 7. The comparison of DGT results with computer modelling calculations showed that both methods give similar trends of information on labile fractions of metal concentrations. These results confirmed the validation of using the PSS liquid binding phase and cellulose dialysis membrane diffusive layer. The deployment of the DGT device in natural waters further validated its speciation ability. The site on Marine Stadium showed higher concentration of copper because of the release of antifouling paints from boats berthing. On the site with much abundant humic substances, Parkwood Pond, less percentage of DGT labile metal was measured due to the complexing reactions between the humic substances and metals. Finally, the newly developed binding phases and the Chelex 100 gel were evaluated for trace metal speciation measurement both in laboratory and in natural water conditions (Chapter 8). With comparison to SC-database model it was found that the measured DGT labile fractions of metals by varying DGT systems in solutions containing EDTA were inorganic fractions. In solutions containing humic acid, it was suggested that the metal- HA complexes were not significantly measured by the various DGT systems with either polyacrylamide or dialysis membrane diffusive layer. The results from field deployment of the DGT devices were explained by comparing the various binding phase DGT measurements with the PSS measurements. Other properties of the binding phases were also evaluated. The advantages of using P81 ion exchange membrane over gel based binding phases were its reusability without significant lost of its binding metal properties (five times) and its constant volume and mechanical strength irrespective of pH and ionic 204 Chapter 9 strength. The use of PSS liquid binding phase provided a mobile, high binding capacity and broad pH working range (3.0 – 8.7) for DGT. There is no need for elution steps for PSS binding phase DGT system which is required for solid binding phase DGT. This simplification increased the accuracy of DGT analysis. The use of dialysis membrane diffusive layer and PSS binding phase minimised biofouling effects on DGT, which was a serious problem for solid binding phase DGT when using 0.45 µm filter membrane as a protect layer and polyacrylamide gel as a diffusive layer. The applicable pH and ionic strength range of the binding phases were compared, which gave information on deployment in natural waters. There is a wide area of continuing research in this study. The syntheses of new polymers for selective and sensitive binding of metals for DGT applications may result in various fractions of DGT measurements. The solution binding phase, in particular, provided opportunities to develop the DGT technique in a wider view, due to its inherent features. Clearly, the development of these series of binding phases has opened doors to further development of the DGT technique. Although extensive investigations on the development of new generation DGT technique have been carried out in this work, some aspects of the newly developed DGT systems need to be explored further. This research has highlighted a number of areas for future studies. (i) Application of these newly developed DGT systems to different types of environmental samples under various conditions, which will gain better understanding on the applicability and usefulness of the developed method. 205 Chapter 9 (ii) Investigation of more binding phases, especially, the investigation of other types of liquid binding phases, which will lead to the improvement of DGT performance in terms of number of detectable species, selectivity, speciation capacity and applicability. (iii) Current DGT system has to be coupled with an appropriate analytical method to provide the analytical data. In this regard, it is essentially an in situ sampling device. More effort should be devoted to develop the third generation DGT system that capable of in situ detection by incorporating an analytical method into the DGT device. With our newly developed liquid binding phase DGT system we believe this is highly feasible. 206 References REFERENCES (1) Ure, A. M.; Davidson, C. M.; (Eds) Chemical Speciation in the Environment; Blackie Academic & Proffessional: London, 1995. (2) Buffle, J.; Horvai, G. In Situ Monitoring of Aquatic Systems Chemical Analysis and Speciation; John Wiley & Sons Ltd: Chichester, 2000. (3) Tessier, A.; Turner, D. R.; (Eds) Metal Speciation and Bioavailability in Aquatic Systems; Wiley: New York, 1995. (4) Buffle, J.; De Vitre, R. R.; (Eds) Chemical and Biological Regulation of Aquatic Systems; Lews: London, 1994. (5) Tessier, A.; Turner, D. R.; (Eds) Metal Speciation and Bioavailability in Aquatic Systems; Wiley: Chichester, 1995. (6) Tercier, M.-L.; Buffle, J.; Graziottin, F. Electroanalysis 1998, 10, 355-363. (7) Donat, J. R.; Bruland, K. W. Trace Elem. Nat. Waters 1995, 247-281. (8) Salbu, B.; Steinnes, E.; (Eds) Trace Elements in Natural Waters; CRC Press: London, 1995. (9) Braungardt, C.; Achterberg, E. P.; Nimmo, M. Analytica Chimica Acta 1998, 377, 205-215. (10) Paucot, H.; Wollast, R. Marine Chemistry 1997, 58, 229-244. (11) Khan, H.; Brush, G. S. Estuaries 1994, 17, 345-360. (12) Vale, C. Science of the Total Environment 1990, 97-98, 137-154. (13) Florence, T. M. Talanta 1982, 29, 345-364. (14) Davison, W.; Fones, G.; Harper, M.; Teasdale, P.; Zhang, H. In Situ Monitoring of Aquatic Systems Chemical Analysis and Speciation; John Wiley & Sons Ltd: Chichester, 2000. (15) Davison, W.; Zhang, H. Nature (London) 1994, 367, 546-548. (16) Zhang, H.; Davison, W. Anal. Chem. 1995, 67, 3391-3400. (17) Dunn, R. J. K. Evaluation of DGT (Diffusive Gradients in Thin Films) against Traditional Methods for Heavy Metal Measurements: Case Studies in the Gold Coast Broadwater, Honours Thesis; Griffith University: Gold Coast, 2002. (18) Buffle, J.; Wilkinson, K. J.; Tercier, M.-L.; Parthasarathy, N. Ann. Chim. (Rome) 1997, 87, 67-82. 207 References (19) Struempler, A. W. Analytical Chemistry 1973, 45, 2251-2254. (20) Truitt, R. E.; Weber, J. H. Analytical Chemistry 1979, 51, 2057-2059. (21) Zhang, H.; Davison, W.; Miller, S.; Tych, W. Geochim. Cosmochim. Acta 1995, 59, 4181-4192. (22) Torre, M. C. A.-D. l.; Beaulieu, P.-Y.; Tessier, A. Analytica Chimica Acta 2000, 418, 53-68. (23) Zhang, H.; Davison, W. Anal. Chem. 2000, 72, 4447-4457. (24) Denney, S.; Sherwood, J.; Leyden, J. Sci. Total Environ. 1999, 239, 71-80. (25) Twiss, M. R.; Moffett, J. W. Environmental Science & Technology 2002, 36, 1061 1068. (26) Campell, P. G. C. Metal Speciation and Bioavailability in Aquatic Systems; Wiley: New York, 1995. (27) Stumm, W.; Morgan, J. J. Aquatic Chemistry; John Wiley & Sons: New York, 1981. (28) Buffle, J. Complexation Reactions in Aquatic Chemistry; John Wiley & Sons: Chichester, UK, 1988. (29) Mantoura, R. F. C. Elsevier Oceanogr. Ser. (Amsterdam) 1981, 31, 179-223. (30) Lund, W. NATO ASI Ser., Ser. G 1990, 23, 43-56. (31) Cosovic, B. Aqueous surface chemistry: assessment of adsorption characteristics of organic solutes by electrochemical methods; John Wiley & Sons: New York, 1985. (32) Wetzel, R. G. Limnology; Saunders: Philadelphia, 1983. (33) Hering, J. G.; Morel, F. M. M. Environ. Sci. Technol. 1988, 22, 1234-1237. (34) Beveridge, A.; Pickering, W. F. Water, Air, Soil Pollut. 1980, 14, 171-185. (35) Ortego, L. S.; Benson, W. H. Environ. Toxicol. Chem. 1992, 11, 261-265. (36) Ma, H.; Kim, S. D.; Cha, D. K.; Allen, H. E. Environ. Toxicol. Chem. 1999, 18, 828 837. (37) Florence, T. M.; Batley, G. E. CRC Crit. Rev. Anal. Chem. 1980, 9, 219-296. (38) Noller, B. N. Chemistry in Australia 1992, 59, 403-405. (39) Gordon, A. S.; Dyer, B. J.; Kango, R. A.; Donat, J. R. Marine Chemistry 1996, 53, 163-172. (40) Donat, J. R.; Lao, K. A.; Bruland, K. W. Analytica Chimica Acta 1994, 284, 547-571. (41) Capodaglio, G.; Coale, K. H.; Bruland, K. W. Mar. Chem. 1990, 29, 221-233. 208 References (42) Capodaglio, G.; Toscano, G.; Scarponi, G.; Cescon, P. Ann. Chim. (Rome) 1989, 79, 543-559. (43) Batley, G. E.; Gardner, D. Estuarine and Coastal Marine Science 1978, 7, 59-70. (44) Apte, S. C.; Gardner, M. J.; Ravenscroft, J. E.; Turrell, J. A. Anal. Chim. Acta 1990, 235, 287-297. (45) Hanson, A. K., Jr.; Sakamoto-Arnold, C. M.; Huizenga, D. L.; Kester, D. R. Mar. Chem. 1988, 23, 181-203. (46) Van den Berg, C. M. G.; Rebello, A. d. L. Sci. Total Environ. 1986, 58, 37-45. (47) Donat, J. R.; Bruland, K. W. Mar. Chem. 1990, 28, 301-323. (48) Bruland, K. W. Limnol. Oceanogr. 1989, 34, 269-285. (49) Coale, K. H.; Bruland, K. W. Limnology and Oceanography 1988, 33, 1084-1101. (50) Salomons, W.; Foerstner, U. Metals in the Hydrocycle; Springer-Verlag: Berlin, Fed. Rep. Ger., 1984. (51) Horne, R. A. Marine Chemistry, the Structure of Water and the Chemistry of the Hydrosphere; Wiley-Interscience: New York, 1969. (52) Aster, S. R. Chemical Oceanography; Academic Press: London, 1978. (53) Hammond, L.; Synnot, R.; (Eds) Marine Biology; longman Cheshire Pty Ltd: Sydney, 1994. (54) Florence, T. M.; Batley, G. E. Critical Reviews in Analytical Chemistry 1980, 9, 219 296. (55) Waite, T. D.; Morel, F. M. M. Environmental Science and Technology 1984, 18, 860 868. (56) Cunningham, K. M.; Goldberg, M. C.; Weiner, E. R. Environmental Science and Technology 1988, 22, 1090-1097. (57) Gilmour, C. C.; Riedel, G. S. Water, Air, and Soil Pollution 1995, 80, 747-756. (58) Watras, C. J.; Bloom, N. S.; Claas, S. A.; Morrison, K. A.; Gilmour, C. C.; Craig, S. R. Water, Air, and Soil Pollution 1995, 80, 735-745. (59) Rudd, J. W. M. Water, Air, and Soil Pollution 1995, 80, 697-713. (60) Guimaraes, J. R. D.; Meili, M.; Malm, O.; de Souza Brito, E. M. Science of the Total Environment 1998, 213, 165-175. (61) Clark, E. A.; Sterritt, R. M.; Lester, J. N. Environmental Science and Technology 1988, 22, 600-604. 209 References (62) Seligman, P. F.; Valkirs, A. O.; Lee, R. F. Environmental Science and Technology 1986, 20, 1229-1235. (63) Kumaresan, M.; Riyazuddin, P. Research Journal of Chemistry and Environment 1999, 3, 59-79. (64) Williams, D. R. Coordination Chemistry Reviews 1999, 185-186, 177-188. (65) Kramer, J. R.; Allen, H. E.; Davison, W.; Godtfredsen, K. L.; Meyer, J. S.; Perdue, E. M.; Tipping, E.; Van de Meent, D.; Westall, J. C. Reassessment of Metals Criteria for Aquatic Life Protection: Priorities for Research and Implementation, Proceedings of the Pellston Workshop on Reassessment of Metals Criteria for Aquatic Life Protection, 26th, Pensacola, Fla., Feb. 10-14, 1996 1997, 57-70. (66) Koplik, R.; Curdova, E.; Mestek, O. Chemicke Listy 1997, 91, 38-47. (67) Donard, O. F. X.; Caruso, J. A. Spectrochimica Acta, Part B: Atomic Spectroscopy 1998, 53B, 157-163. (68) Witters, H. E. Ecotoxicology and Environmental Safety 1998, 41, 90-95. (69) Nelson, H.; Benoit, D.; Erickson, R.; Mattson, V.; Lindberg, J.; Environ. Res. Lab.,Duluth,MN,USA. FIELD URL:, 1986, pp 121. (70) Roast, S. D.; Widdows, J.; Jones, M. B. Environmental Toxicology and Chemistry 2001, 20, 1078-1084. (71) Banaszak, J. E.; Rittmann, B. E.; Reed, D. T. Journal of Radioanalytical and Nuclear Chemistry 1999, 241, 385-435. (72) Mandal, R.; Hassan, N. M.; Murimboh, J.; Chakrabarti, C. L.; Back, M. H.; Rahayu, U.; Lean, D. R. S. Environmental Science and Technology 2002, 36, 1477-1484. (73) Gustafsson, O.; Gschwend, P. M. Limnology and Oceanography 1997, 42, 519-528. (74) Garruto, R. M.; Flaten, T. P.; Wakayama, I. Mineral and Metal Neurotoxicology 1997, 27-34. (75) Robinson, J. W.; Deano, P. M. Am. Lab. (Fairfield, Conn.) 1986, 18, 17-20, 22, 24 16. (76) Mandal, B. K.; Suzuki, K. T. Talanta 2002, 58, 201-235. (77) Kaise, T.; Watanabe, S.; Itoh, K. Chemosphere 1985, 14, 1327-1332. (78) Cannavale, V.; Chiusano, A. Giornale di Medicina Militare 1992, 142, 161-163. (79) Leeuwangh, P.; Nijman, W.; Visser, H.; Kolar, Z.; De Goey, J. J. M. Meded. Fac. Landbouwwet., Rijksuniv. Gent 1976, 41, 1483-1490. 210 References (80) Zielinski, W.; Brzezinski, J. Rocz. Panstw. Zakl. Hig. 1983, 34, 95-102. (81) Mertz, W.; Cornatzer, W. E. Newer Trace Elements in Nutrition; Dekker: New York, 1971. (82) Cooley, T. N.; Martin, D. F. Journal of Inorganic and Nuclear Chemistry 1980, 42, 151-153. (83) Driscoll, C. T., Jr.; Baker, J. P.; Bisogni, J. J., Jr.; Schofield, C. L. Nature (London, United Kingdom) 1980, 284, 161-164. (84) Florence, T. M.; Belew, W. L. Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 1969, 21, 157-167. (85) Whitfield, M.; Turner, D. R. Chemical Modelling in Aqueous Systems; American Chemical Society: Washington D. C., 1979. (86) Magnuson, V. R.; Harriss, D. K.; Sun, M. S.; Taylor, D. K.; Glass, G. E. Chemical Modelling in Aqueous Systems; American Chemical Society: Washington D.C., 1979. (87) Naimo, T. J. Ecotoxicology 1995, 4, 341-362. (88) Santschi, P. H.; Adler, D. M.; Amdurer, M. NATO Conference Series IV: Marine Sciences 1983, 9, 331-349. (89) Chen, Z.; Mayer, L. M. Marine Ecology: Progress Series 1999, 176, 139-151. (90) Canterford, G. S.; Canterford, D. R. J. Mar. Biol. Assoc. U. K. 1980, 60, 227-242. (91) Morel, F. M. M. Principles of Aquatic Chemistry; Wiley-Interscience: New York, 1983. (92) Kaiser, K. L. E. Can. J. Fish. Aquat. Sci 1980, 37, 211-218. (93) Urbanowski, J. L.; Piper, R. C. Journal of Biological Chemistry 1999, 274, 38061 38070. (94) Satofuka, H.; Fukui, T.; Takagi, M.; Atomi, H.; Imanaka, T. Journal of Inorganic Biochemistry 2001, 86, 595-602. (95) Williams, R. J. Philos. Trans. R. Soc. London B, 1981, 294, 57-74. (96) Luoma, S. N. Sci. Total Environ. 1983, 28, 1-22. (97) Turner, D. R. Elsevier Oceanogr. Ser. (Amsterdam) 1986, 43, 191-200. (98) Simkiss, K.; Taylor, M. G. Rev. Aquat. Sci. 1989, 1, 173-188. (99) Rudakova, E. V.; Karakis, K. D.; Sidorshina, T. N. Fiziol. Biokhim. Kul't. Rast. 1988, 20, 3-12. 211 References (100) Kinter, W. B.; Pritchard, J. B., Bethesda, Md 1977. (101) Di Toro, D. M.; Allen, H. E.; Bergman, H. L.; Meyer, J. S.; Paquin, P. R.; Santore, R. C. Environmental Toxicology and Chemistry 2001, 20, 2383-2396. (102) Paquin, P. R.; Zoltay, V.; Winfield, R. P.; Wu, K. B.; Mathew, R.; Santore, R. C.; Di Toro, D. M. Comparative Biochemistry and Physiology, Part C: Toxicology & Pharmacology 2002, 133C, 305-343. (103) Santore, R. C.; Di Toro, D. M.; Paquin, P. R.; Allen, H. E.; Meyer, J. S. Environmental Toxicology and Chemistry 2001, 20, 2397-2402. (104) Janssen, C. R.; Heijerick, D. G.; De Schamphelaere, K. A. C.; Allen, H. E. Environment International 2003, 28, 793-800. (105) Pagenkopf, G. K. Environmental Science and Technology 1983, 17, 342-347. (106) Tipping, E. Computers & Geosciences 1994, 20, 973-1023. (107) Santore, R. C.; Driscoll, C. T. SSSA Special Publication 1995, 42, 357-375. (108) Erickson, R. J.; Benoit, D. A.; Mattson, V. R.; Nelson, H. P., Jr.; Leonard, E. N. Environmental Toxicology and Chemistry 1996, 15, 181-193. (109) Diamond, J. M.; Gerardi, C.; Leppo, E.; Miorelli, T. Environmental Toxicology and Chemistry 1997, 16, 1480-1487. (110) Bury, N. R.; Galvez, F.; Wood, C. M. Environmental Toxicology and Chemistry 1999, 18, 56-62. (111) Karen, D. J.; Ownby, D. R.; Forsythe, B. L.; Bills, T. P.; La Point, T. W.; Cobb, G. B.; Klaine, S. J. Environmental Toxicology and Chemistry 1999, 18, 63-70. (112) Karapanagiotis, N. K.; Sterritt, R. M.; Lester, J. N. Environmental Pollution (Oxford, United Kingdom) 1990, 67, 259-278. (113) Buffle, J.; Leppard, G. G. Environ. Sci. Technol. 1995, 29, 2169-2175. (114) Buffle, J.; Leppard, G. G. Environ. Sci. Technol. 1995, 29, 2176-2184. (115) Andersen, M. K.; Raulund-Rasmussen, K.; Strobel, B. W.; Hansen, H. C. B. Journal of Environmental Quality 2002, 31, 168-175. (116) Buffle, J.; Wilkinson, K. J.; Tercier, M.-L.; Parthasarathy, N. Annali di Chimica (Rome) 1997, 87, 67-82. (117) Branica, M. Kemija u Industriji 2001, 50, 493-499. (118) Mevius, W. Neue DELIWA-Zeitschrift 1988, 39, 82-87. (119) Chen, Y.-W.; Buffle, J. Water Res. 1996, 30, 2185-2192. 212 References (120) Chen, Y.-W.; Buffle, J. Water Res. 1996, 30, 2178-2184. (121) Mota, A. M.; Correia dos, M. M. Metal Speciation and Bioavailability in Aquatic Systems; Wiley: New York, 1999. (122) Kramer, K. J. M. Tidal Estuaries : Manual of Sampling and Analytical Procedures; A.A. Balkema for the European Commission: Rotterdam, 1994. (123) Gamo, T. Bunseki 1998, 832-839. (124) Beichert, J. K.; Kanitz, J. Gewaesserschutz, Wasser, Abwasser 1993, 135, 647-660. (125) Bufflap, S. E.; Allen, H. E. Water Res. 1995, 29, 165-177. (126) Koryta, J.; Stulik, K. Ion-Selective Electrodes. 2nd Ed; Cambridge University Press: Cambridge, UK, 1983. (127) Moody, G. J.; Thomas, J. D. R. Selective Ion-Sensitive Electrodes (Merrow Technical Library; Practical Science); Merrow: Watford, Engl., 1971. (128) Pineros, M. A.; Shaff, J. E.; Kochian, L. V. Plant Physiology 1998, 116, 1393-1401. (129) Midorikawa, T.; Tanoue, E.; Sugimura, Y. Analytical Chemistry 1990, 62, 1737 1746. (130) Buffle, J.; Tercier-Waeber, M.-L. In Situ Monitoring of Aquatic Systems Chemical Analysis and Speciation; John Wiley & Sons Ltd: Chichester, 2000. (131) Henze, G. NATO ASI Ser., Ser. G 1990, 23, 391-408. (132) Mizuike, A. Pure Appl. Chem. 1987, 59, 555-564. (133) Heyrowsky, J.; Kuta, J. Principles of Polarography; Academic Press: London, 1966. (134) Buffle, J. Complexation Reactions in Aquatic Systems: An Analytical Approach; Ellis Horwood Ltd.: Chichester, UK, 1988. (135) Buffle, J.; Wilkinson, K. J.; Stoll, S.; Filella, M.; Zhang, J. Environmental Science and Technology 1998, 32, 2887-2899. (136) Pavillon, J. F.; Menesria, R.; Lacaze, J. C. J. Rech. Oceanogr. 1992, 17, 12-16. (137) Mayer, L. M. Limnol. Oceanogr. 1976, 21, 909-912. (138) Batley, G. E. Trace Element Speciation: Analytical Methods and Problems; CRC Press: Boca Raton, Fla., 1989. (139) Buffle, J.; Perret, D.; Newman, M. Environ. Part. 1992, 1, 171-230. (140) Carignan, R.; Rapin, F.; Tessier, A. Geochim. Cosmochim. Acta 1985, 49, 2493 2497. 213 References (141) Shuttleworth, S. M.; Davison, W.; Hamilton-Taylor, J. Environ. Sci. Technol. 1999, 33, 4169-4175. (142) Cox, J. A.; Bhatnagar, A.; Francis Robert W, Jr. Talanta 1986, 33, 713-716. (143) Danesi, P. R. Separation Science and Technology 1984, 19, 857-894. (144) Knutsson, M.; Nilve, G.; Mathiasson, L.; Joensson, J. A. Journal of Agricultural and Food Chemistry 1992, 40, 2413-2417. (145) Malcus, F.; Djane, N. K.; Mathiasson, L.; Johansson, G. Analytica Chimica Acta 1996, 327, 295-300. (146) Davison, W.; Grime, G. W.; Morgan, J. A. W.; Clarke, K. Nature (London) 1991, 352, 323-325. (147) Chang, L.-Y.; Davison, W.; Zhang, H.; Kelly, M. Anal. Chim. Acta 1998, 368, 243 253. (148) Snodgrass, W. J. Sediments and Water Interactions; Springer: New York, 1986. (149) Scally, S.; Davison, W.; Zhang, H. Environmental Science and Technology 2003, 37, 1379-1384. (150) Zhang, H.; Davison, W.; Gadi, R.; Kobayashi, T. Anal. Chim. Acta 1998, 370, 29-38. (151) Davison, W.; Fones, G. R.; Grime, G. W. Nature (London) 1997, 387, 885-888. (152) Chang, L.-Y. Development and Application of Diffusive Gradients in Thin-films (DGT) for the Measurement of Stable and Radioactive Caesium and Strontium in Surface Waters, Ph.D Thesis; Lancaster University: Lancaster, UK, 1998. (153) Zhang, H.; Davison, W.; Grime, G. W. ASTM Spec. Tech. Publ. 1995, STP 1293, 170-181. (154) Zhang, H.; Davison, W. Anal. Chim. Acta 1999, 398, 329-340. (155) Davison, W.; Zhang, H.; Grime, G. W. Environ. Sci. Technol. 1994, 28, 1623-1632. (156) Teasdale, P. R.; Hayward, S.; Davison, W. Anal. Chem. 1999, 71, 2186-2191. (157) Hudson, R. J. M. Science of the Total Environment 1998, 219, 95-115. (158) Mirimanoff, N.; Wilkinson, K. J. Environmental Science and Technology 2000, 34, 616-622. (159) Van Leeuwen, H. P. Environmental Science and Technology 1999, 33, 3743-3748. (160) Davison, W. J. Electroanal. Chem. Interfacial Electrochem. 1978, 87, 395-404. (161) Webb, J. A.; Keough, M. J. Marine Pollution Bulletin 2002, 44, 222-229. 214 References (162) Beaulieu, P. Y. Verification d'une Methode DGT Pour Mesurer Les Metaux Traces et Son Utilisation In Situ Comparee a La Dialyse In Situ, M.Sc Thesis; Univerite du Quebec: Sainte-Foy, Canada, 1999. (163) Dunn, R. J. K.; Teasdale, P. R.; Warnken, J.; Schleich, R. R. Environmental Science and Technology 2003, 37, 2794-2800. (164) Teasdale, P. Unpublished results. (165) Munksgaard, N. C.; Parry, D. L. Journal of Environmental Monitoring 2003, 5, 145 149. (166) Sangi, M. R.; Halstead, M. J.; Hunter, K. A. Analytica Chimica Acta 2002, 456, 241 251. (167) Fones, G. R.; Moffett, J. W. Abstracts of Papers, 224th ACS National Meeting, Boston, MA, United States, August 18-22, 2002 2002, GEOC-018. (168) Pesavento, M.; Biesuz, R.; Gnecco, C.; Magi, E. Analytica Chimica Acta 2001, 449, 23-33. (169) Warshawsky, A. Synthesis and Separation Using Functional Polymers; Wiley: New York, 1988. (170) Klein, E. J. Membr. Sci. 2000, 179, 1-27. (171) Kremer, M.; Pothmann, E.; Roessler, T.; Baker, J.; Yee, A.; Blanch, H.; Prausnitz, J. M. Macromolecules 1994, 27, 2965-2973. (172) Dusek, K.; Janacek, J. Journal of Applied Polymer Science 1975, 19, 3061-3075. (173) Yao, K.-J.; Zhou, W.-J. Journal of Applied Polymer Science 1994, 53, 1533-1538. (174) Baker, J. P.; Hong, L. H.; Blanch, H. W.; Prausnitz, J. M. Macromolecules 1994, 27, 1446-1454. (175) Denizli, A.; Salih, B.; Arica, M. Y.; Kesenci, K.; Hasirci, V.; Piskin, E. J. Chromatogr., A 1997, 758, 217-226. (176) Denizli, A.; Kesenci, K.; Arica, M. Y.; Salih, B.; Hasirci, V.; Piskin, E. Talanta 1998, 46, 551-558. (177) Shah, R.; Devi, S. React. Funct. Polym. 1996, 31, 1-9. (178) Uludag, Y.; Ozbelge, H. O.; Yilmaz, L. Journal of Membrane Science 1997, 129, 93 99. (179) Sasaki, K. J.; Burnett, S. L.; Christian, S. D.; Tucker, E. E.; Scamehorn, J. F. Langmuir 1989, 5, 363-369. 215 References (180) Muslehiddinoglu, J.; Uludag, Y.; Ozbelge, H. O.; Yilmaz, L. Journal of Membrane Science 1998, 140, 251-266. (181) Rivas, B. L.; Martinez, E.; Pereira, E.; Geckeler, K. E. Polymer International 2001, 50, 456-462. (182) Rivas, B. L.; Pooley, S. A.; Luna, M. Macromolecular Rapid Communications 2001, 22, 418-421. (183) Rivas, B. L.; Pereira, E. D.; Moreno-Villoslada, I. Progress in Polymer Science 2002, 28, 173-208. (184) Rivas, B. L.; Moreno-Villoslada, I. J. Phys. Chem. B 1998, 102, 6994-6999. (185) Yin, X.-q.; Zhang, Q.; Yu, W.-x.; Yang, L.-c.; Lin, Q. Wuji Huaxue Xuebao 2002, 18, 87-90. (186) Kurpiel-Gorgol, R.; Brzyska, W. Journal of Thermal Analysis and Calorimetry 2003, 71, 539-548. (187) Teasdale, P. Personal Communication. (188) Sutcliffe, D. W.; Carrick, T. R. Freshwater Biol. 1983, 13, 323-352. (189) Harris, D. C. Quantitative Chemical Analysis. 2nd Ed; W. H. Freeman and Co.: New York, N. Y., 1995. (190) Skoog, D. A.; West, D. M. Analytical Chemistry: An Introduction. 7th Ed; Saunders College Publishing: Fort Worth, 1996. (191) Wilson, A. L. The Chemical Analysis of Water. General Principles and Techniques; Chem. Soc.: London, Engl., 1978. (192) Muslehiddinoglu, J.; Uludag, Y.; Ozbelge, H. O.; Yilmaz, L. Talanta 1998, 46, 1557 1565. (193) Chanda, M.; Rempel, G. L. React. Polym. 1993, 19, 213-223. (194) Kabay, N.; Katakai, A.; Sugo, T.; Egawa, H. J. Appl. Polym. Sci. 1993, 49, 599-597. (195) Montembault, V.; Folliot, V.; Soutif, J. C.; Brosse, J. C. React. Polym. 1994, 22, 81 89. (196) Alexandratos, S. D.; Trochimczuk, A. W.; Crick, D. W.; Horwitz, E. P.; Gatrone, R. C.; Chiarizia, R. Macromolecules 1996, 29, 1021-1026. (197) Kulkarni, R. A.; Gundiah, S. Makromol. Chem. 1984, 185, 957-967. (198) March, J. Advanced Organic Chemistry: Reactions, Mechanisms, and Structure, Fourth Edition; Wiley: New York, N. Y., 1992. 216 References (199) Durmaz, S.; Okay, O. Polymer 2000, 41, 3693-3704. (200) Hummel, D. O. Infrared Analysis of Polymers, Resins and Additives, An Atlas; Wiley Interscience: New York, 1969. (201) Ahn, J. S.; Choi, H. K.; Cho, C. S. Biomaterials 2001, 22, 923-928. (202) Nikaido, H.; Rosenberg, E. Y. J. Gen. Physiol. 1981, 77, 121-135. (203) Molyneux, P. Water-Soluble Synthetic Polymers: Properties and Behavior; CRC Press: Boca Raton, Fla., 1984. (204) Osada, Y.; Okuzaki, H.; Gong, J. P. Trends Polym. Sci. (Cambridge, U. K.) 1994, 2, 61-66. (205) Brazel, C. S.; Peppas, N. A. Macromolecules 1995, 28, 8016-8020. (206) Cohen, Y.; Ramon, O.; Kopelman, I. J.; Mizrahi, S. J. Polym. Sci., Part B: Polym. Phys. 1992, 30, 1055-1067. (207) Shibayama, M.; Mizutani, S.-y.; Nomura, S. Macromolecules 1996, 29, 2019-2024. (208) Shibayama, M.; Uesaka, M.; Shiwa, Y. J. Chem. Phys. 1996, 105, 4350-4357. (209) Ravichandran, P.; Shantha, K. L.; Rao, K. P. International Journal of Pharmaceutics 1997, 154, 89-94. (210) Yoo, M. K.; Sung, Y. K.; Lee, Y. M.; Cho, C. S. Polymer 2000, 41, 5713-5719. (211) Baker, J. P.; Blanch, H. W.; Prausnitz, J. M. Polymer 1995, 36, 1061-1069. (212) Rivas, B. L.; Pooley, S. A.; Soto, M.; Geckeler, K. E. J. Polym. Sci., Part A: Polym. Chem. 1997, 35, 2461-2467. (213) Stumm, W.; Morgan, J. J. Aquatic Chemistry; Wiley: New York, N. Y., 1996. (214) Mandal, R.; Sekaly, A. L. R.; Murimboh, J.; Hassan, N. M.; Chakrabarti, C. L.; Back, M. H.; Gregoire, D. C.; Schroeder, W. H. Anal. Chim. Acta 1999, 395, 323-334. (215) Marinsky, J. A.; Anspach, W. M. J. Phys. Chem. 1975, 79, 439-444. (216) Reed, B. E.; Matsumoto, M. R. Sep. Sci. Technol. 1993, 28, 2179-2195. (217) Yang, Z.; Zhang, Y.; Markland, P.; Yang, V. C. Journal of Biomedical Materials Research 2002, 62, 14-21. (218) Bhadani, S. N.; Prasad, Y. K. Makromol. Chem. 1977, 178, 1841-1851. (219) Seymour, R. B.; Carraher, C. E., Jr. Polymer Chemistry. An Introduction; Dekker: New York, N. Y., 1981. (220) Yan, L.; Qian, F.; Zhu, Q. Polymer International 2001, 50, 1370-1374. 217 References (221) Yazdani-Pedram, M.; Retuert, J.; Quijada, R. Macromolecular Chemistry and Physics 2000, 201, 923-930. (222) Silverstein, R. M.; Webster, F. X. Spectrometric identification of organic compounds; John Wiley & Sons, Inc.: New York, 1997. (223) Denizli, A.; Salih, B.; Piskin, E. J. Chromatogr., A 1997, 773, 169-178. (224) Kantipuly, C.; Katragadda, S.; Chow, A.; Gesser, H. D. Talanta 1990, 37, 491-517. (225) Egawa, H.; Nakayama, M.; Nonaka, T.; Sugihara, E. J. Appl. Polym. Sci. 1987, 33, 1993-2005. (226) Choi, S.-H.; Han Jeong, Y.; Jeong Ryoo, J.; Lee, K.-P. Radiation Physics and Chemistry 2001, 60, 503-511. (227) Wen, O. H.; Kuroda, S.-i.; Kubota, H. European Polymer Journal 2001, 37, 807-813. (228) Saito, K.; Yamada, S.; Furusaki, S.; Sugo, T.; Okamoto, J. Journal of Membrane Science 1987, 34, 307-315. (229) Fischer, H. J.; Lieser, K. H. Angew. Makromol. Chem. 1993, 208, 133-150. (230) Denizli, A.; Tanyolac, D.; Salih, B.; Aydinlar, E.; Ozdural, A.; Piskin, E. J. Membr. Sci. 1997, 137, 1-8. (231) Brajter, K.; Slonawska, K. Anal. Chim. Acta 1986, 185, 271-277. (232) Schmitt, D. H.; Fritz, J. S. Talanta 1968, 15, 515-524. (233) Kabay, N.; Demircioglu, M.; Yayli, S.; Yuksel, M.; Saglam, M.; Levison, P. R. Sep. Sci. Technol. 1999, 34, 41-51. (234) Dorfner, K. Ion Exchangers; Ann Arbor Science Publishers Inc.: Michigan, 1972. (235) Stumm, W.; Sigg, L.; Sulzberger, B. The Role of Coordination at the Surface of Aquatic Particles in Chemical and Biological Regulation of Aquatic Systems; Lewis Publishers, USA, 1994. (236) Volchek, K.; Krentsel, E.; Zhilin, Y.; Shtereva, G.; Dytnersky, Y. Journal of Membrane Science 1993, 79, 253-272. (237) Kichik, V. A.; Yagodin, G. A.; Kuleshov, N. F.; Svitsov, A. A. At. Energ. 1985, 58, 272-273. (238) Pickering, W. F. Chemical speciation in the environment; Blackie Academic & Proffessional: London, 1995. (239) Apte, S. C.; Gardner, M. J.; Hunt, D. T. E. Environ. Technol. Lett. 1989, 10, 201 212. 218 References (240) Benes, P.; Steinnes, E. Water Res. 1974, 8, 947-953. (241) Juang, R.-S.; Liang, J.-F. Journal of Membrane Science 1993, 82, 163-174. (242) Samadfam, M.; Jintoku, T.; Sato, S.; Ohashi, H. Journal of Nuclear Science and Technology 1998, 35, 579-583. (243) Gregor, H. P.; Luttinger, L. B.; Loebl, E. M. J. Phys. Chem. 1955, 59, 34-39. (244) Leyte, J. C.; Zuiderweg, L. H.; Van Reisen, M. J. Phys. Chem. 1968, 72, 1127-1132. (245) Yamaoka, K.; Masujima, T. Bull. Chem. Soc. Jpn. 1979, 52, 1819-1827. (246) Lange, N. A. Lange's Handbook of Chemistry; McGraw-Hill Book Company: New York, 1985. (247) Kotin, L.; Nagasawa, M. J. Am. Chem. Soc. 1961, 83, 1026-1028. (248) Lapanje, S.; Rice, S. A. J. Am. Chem. Soc. 1961, 83, 496-497. (249) Pesavento, M.; Biesuz, R. Analytica Chimica Acta 1997, 346, 381-391. (250) Li, Y.-H.; Gregory, S. Geochim. Cosmochim. Acta 1974, 38, 703-714. (251) Rivas, B. L.; Moreno-Villoslada, I. J. Appl. Polym. Sci. 1998, 70, 219-225. (252) Buchberger, W.; Haddad, P. R.; Alexander, P. W. J. Chromatogr. 1991, 558, 181 186. (253) McKnight, D. M.; Aiken, G. R. Ecological Studies 1998, 133, 9-39. (254) Perminova, I. V.; Grechishcheva, N. Y.; Petrosyan, V. S. Environmental Science and Technology 1999, 33, 3781-3787. (255) Stevenson, F. J. Humus Chemistry: Genesis, Composition and Reactions; Wiley: New York, 1994. (256) MacCarthy, P. Adv. Chem. Ser. 1989, 219, xvii-xxx. (257) Livens, F. R. Environ. Pollut. 1991, 70, 183-208. (258) Cruz, B. H.; Diaz-Cruz, J. M.; Arino, C.; Esteban, M. Electroanalysis 2000, 12, 1130 1137. (259) Brezonic, P. L. Chemical Kinetics and Process Dynamics in Aquatic Systems; Lewis Publish: Boca Raton, 1994. (260) Ochiai, M. Journal of Chromatography A 1980, 194, 224-227. (261) Rendleman, J. A., Jr. Adv. Carbohydr. Chem. 1966, 21, 209-271. (262) Bourne, E. J.; Nery, R.; Weigel, H. Chem. & Ind. (London) 1959, 998-999. (263) Tajmir-Riahi, H. A. Carbohydr. Res. 1988, 183, 35-46. 219 References (264) Briggs, J.; Finch, P.; Matulewicz, M. C.; Weigel, H. Carbohydr. Res. 1981, 97, 181 188. (265) Turner, A. H.; Houston, J. H. Industrial Applications of Surfactants III. (The Proceedings of a Symposium Organized by the North West Region of the Industrial Division of the Royal Society of Chemistry, University of Salford, 16-18th September 1991.) [In: Spec. Publ. - R. Soc. Chem., 1992; 107]; Royal Soc. Chem.: Cambridge, UK, 1992. (266) Crisp, P. T.; Eckert, J. M.; Gibson, N. A. J. Chem. Educ. 1983, 60, 236-238. (267) Pettit, G.; Pettit, L. D. IUPAC Stability Constants Database; Academic Software, 1999. (268) Relan, P. S.; Khanna, S. S.; Chand, T.; Kumari, R. J. Indian Soc. Soil Sci. 1986, 34, 250-256. (269) Morris, D. F. C.; Short, E. L. Journal of the Chemical Society, Abstracts 1962, 2672 2675. (270) Lazarev, A. N.; Makashev, Y. A.; Mironov, V. E.; Lobov, B. I. Zhurnal Fizicheskoi Khimii 1975, 49, 2258-2262. (271) Ashurst, K. G.; Hancock, R. D. Journal of the Chemical Society, Dalton Transactions: Inorganic Chemistry (1972-1999) 1977, 1701-1707. (272) Zelic, M.; Branica, M. Electroanalysis 1992, 4, 701-706. (273) Fedorov, V. A.; Robov, A. M.; Plekhanov, V. P.; Kudruk, V. V.; Kuznechikhina, M. A.; Chernikova, G. E. Zhurnal Neorganicheskoi Khimii 1974, 19, 1225-1229. (274) Fedorov, V. A.; Chernikova, G. E.; Kuznechikhina, M. A.; Kuznetsova, T. I. Zhurnal Neorganicheskoi Khimii 1975, 20, 2912-2915. (275) Schleich; R., R. Measuring Copper Release from Antifouling Paints Using Diffusive Gradients in Thin Films (DGT), Honours Thesis; Griffith University: Gold Coast, Australia, 2001. (276) Marty, J.; Persin, M.; Sarrazin, J. Journal of Membrane Science 2000, 167, 291-297. (277) Aiken, G. Humic Substances in Soil, Sediment, and Water; Geochemistry, Isolation, and Characterization; John Wiley and Sons: New York, N. Y, 1985. (278) Ross, A. R. S.; Ikonomou, M. G.; Orians, K. J. Analytica Chimica Acta 2000, 411, 91-102. (279) Bourne, E. J.; Searle, F.; Weigel, H. Carbohydrate Research 1971, 16, 185-187. 220 References (280) Jordan, R. N.; Yonge, D. R.; Hathhorn, W. E. Journal of Contaminant Hydrology 1997, 29, 59-80. (281) Hsu, J.-H.; Lo, S.-L. Journal of Environmental Quality 2000, 29, 447-453. (282) Tusseau-Vuillemin, M.-H.; Gilbin, R.; Taillefert, M. Environmental Science and Technology 2003, 37, 1645-1652. (283) DeVries, C. R.; Wang, F. Environmental Science and Technology 2003, 37, 792-797. (284) Odzak, N.; Kistler, D.; Xue, H.; Sigg, L. Aquatic Sciences 2002, 64, 292-299. (285) Garmo, O. A.; Royset, O.; Steinnes, E.; Flaten, T. P. Analytical Chemistry, ACS ASAP. (286) Bitter, J. G. A. Transport Mechanisms in Membrane Separation Processes; Plenum Press: New York, 1991. (287) Wallace, R. A.; Gable, R. J. J. Appl. Polym. Sci. 1973, 17, 3549-3552. 221