Quality Assurance in Environmental Monitoring Instrumental Methods
Edited by G. Subramanian
Weinheim - New York Base1 Cambridge - Tokyo a
This Page Intentionally Left Blank
Quality Assurance in Environmental Monitoring Edited by G. Subramanian
This Page Intentionally Left Blank
Quality Assurance in Environmental Monitoring Instrumental Methods
Edited by G. Subramanian
Weinheim - New York Base1 Cambridge - Tokyo a
Other Important Titles for Quality Assurance:
Quevauviller, Ph. (ed.) Quality Assurance in Environmental Monitoring Sampling and Sample Pretreatment 1995. 306 pages. Hardcover. DM 178,-. ISBN 3-527-28724-8
W. Funk, V Dammann, G. Donnevert Quality Assurance in Analytical Chemistry 1995.238 pages with 87 figures. Hardcover. DM 125,-. ISBN 3-527-28668-3
0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim (Federal Republic of Germany), 1995
Distribution: VCH, P. 0. Box 101161, D-69451 Weinheim (Federal Republic of Germany) Switzerland: VCH, P. 0. Box, CH-4020 Basel (Switzerland) United Kingdom and Ireland: VCH, 8 Wellington Court, Cambridge CBI 1HZ (United Kingdom) USA and Canada: VCH, 220 East 23rd Street, New York, NY 10010-4606 (USA) Japan: VCH, Eikow Building, 10-9 Hongo I-chome, Bunkyo-ku, Tokyo 113 (Japan) ISBN 3-527-28682-9
Ganapathy Subramanian 60 B Jubilee Koad Littlebourne Canterbury Kent CT3 lTP, UK
This book was carefully produced. Nevertheless. authors, editor and publisher do not warraut the 111formation contained therein to be free of errors. Readcrs are advised to keep in mind that statements. data. illustrations, procedural details or other items may inadvertently be inaccurate.
Published jointly by VCH Verlagsgescllschaft, Weinheim (Fedcral Republic of Germany) VCH Publishers, New York. NY (USA)
Editorial Director: Dr. Don Emeraon, Dr. Steffen Pauly Production Manager: Claudia Gross1
Library of Congress Card No. applied for
British Library Cataloguing-in-Publication Data: A catalogue record for this book i s available from the British Library
Die Deutsche Bibliothek - CIP-Einheitsaufnahine Quality assurance in environmental monitoring : instrumental methods I ed. by G . Subramanian. Weinheim ; New York ; Basel ; Cambridge ; Tokyo : VCH, 1995 ISBN 3-527-28682-9 NE: Subramanian, Ganapathy [Hrsg.]
0VCH Verlagsgesellschaft mbH, D-69451 Weinheim (Federal Republic of Germany), 1995 Printed on acid-free and low-chlorine paper All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form - b y photoprinting, microfilm, or any other means - nor transmitted or translated into a machine language without written permission from the publishers. Registcred names, tradcmarks, etc. used in this book, even when not specifically marked as such. are not to be considered unprotected by law. Composition: K + V Fotosatz GmbH, D-64743 Beerfelden Printing: betz-druck GmbH, D-64291 Darmstadt Bookbinding: Wilhelm Osswald + Co., D-67433 Neustadt Printed in the Federal Republic of Germany
Preface
Environmental Monitoring has focused great attention in many laboratories around the world. This is due to increased levels of legislation being enacted by various authorities such as environmental committees in Europe (EEC), the environmental protection agency (EPA) in United States. Both of these, and the importance of public health awareness, place a great responsibility on all organisations to monitor the condition of the environment. This major task and the continuing changes in regulations have created a need for advanced instrumentation technology and analytical methods which are capable of meeting the current and future requirements. This book presents a balanced overview of selected instrumental developments and their application in environmental monitoring. Each chapter is essentially selfcontained for those who wish to select a particular field of interest. The authors reflect the considerable concern over the environmental analysis of a variety of substances, describing in detail the technology of the instrumental principles and their successful application in monitoring pollutants. The coverage has been planned to report the current status of the subject to a wide range of readers who are actively involved in the field of environmental control. Graduates, postgraduates, chemical and biochemical engineers, biologists, researchers, industrial scientists, and consultants who require detailed information should all benefit from this book. I am indebted to the international group of contributors who have shared their experience and knowledge. Each chapter contains a balanced overview of the chosen topic. Chapter 1 gives an account of solid phase extraction in sample purification, its importance and application. Superfluid critical extraction in environmental analysis is discussed in Chapter 2. Chapter 3 discuss the validation and environmental analysis by Atomic Absorption Spectrometry as applied to trace metals in environment. The development of Inductively Coupled Plasma-Optical Emmission spectrometry in environmental analysis is discussed in Chapter 4. Volatile Organic Chemical monitoring and the applications of GC-MS are covered in Chapters 5 and 6. CES in environmental monitoring is reviewed in Chapter 7. Development, design, and application of Field Flow analysis is discussed in Chapter 8. Chapter 9 presents the application of software in environmental auditing and quality control. I wish to express my sincere thanks to Dr. Don Emerson and all the staff at VCH for their help in publishing this book. Canterbury, Kent October, 1995
G. Subramanian
Contents
1
The Use of Solid Phase Extraction for Environmental Samples
Dean Rood 1.1 1.2 1.3 1.3.1 1.3.2 1.3.3 1.3.4 1.4 1.5 1.5.1 1.5.2 1.5.3 1.6 1.6.1 1.6.2 1.6.3 1.7 1.7.1 1.7.2 1.7.3 1.7.4 1.8 1.8.1 1.8.2 1.8.3 1.8.4 1.9 1.10 1.11 1.12
The Importance of Sample Preparation ......................... Introduction to Solid Phase Extraction ......................... SPE Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Syringe Barrel or Cartridges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Syringe Filter or Sep-paks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Disks ...................................................... Choice of Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Using SPE Cartridges and Disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SPE Sorbents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Normal Phase Sorbents ...................................... Reverse Phase Sorbents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ion Exchange Sorbents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sorbent and Solvent Relationships ............................. Normal Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reverse Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ion Exchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Selecting the Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conditioning Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Loading Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rinsing Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elution Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent Volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent Miscibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent Volatility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent Flow Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Selecting Cartridge Size ...................................... Method Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Matrix Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analysis Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 2 2 2 4 5 6 6 7 7 7 9 10 10 11 11 12 13 13 14 14 14 14 15 15 15
16 16 18 19
VIlI
Contents
1.13 1.14
Method Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
Current Status of Supercritical Fluid Extraction in Environmental Analysis
19 20
Joseph M . Levy and Athos C. Rosselli 2.1 2.2 2.3 2.3.1 2.3.2 2.3.3 2.3.4 2.3.5 2.4
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . What is Supercritical Fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applicable Environmental Analytes and Matrices . . . . . . . . . . . . . . . . Polynuclear Aromatic Hydrocarbons and Polychlorinated Biphenyls Total Petroleum Hydrocarbons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SFE of Wet Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pesticides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dibenzofurans/Dioxins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Validation and Quality Control with Atomic Absorption Spectrometry for Environmental Monitoring
25 26 31 32 41 45 48 52 52
Ian L . Shuttler 3.1 3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 3.1.6 3.2 3.2.1 3.2.1 .1 3.2.1.2 3.2.2 3.2.2.1 3.2.2.2 3.2.2.3 3.2.2.4 3.2.2.5 3.2.2.6 3.2.3 3.2.3.1 3.2.3.2
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Use of Atomic Absorption Spectrometry in Environmental Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Need for Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Importance of Consistent Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . Standardized/Reference Methods or Quality Control? . . . . . . . . . . . . The Degree of Analytical Quality Control ...................... Quality Control Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Method Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Analytical Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preparation of Calibration/Standard Solutions . . . . . . . . . . . . . . . . . . Use of Characteristic Concentration/Mass ...................... Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Importance of Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Influence of the Blank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Type of Calibration Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Linear or Non-Linear-Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Calibration by the Method of Analyte Additions . . . . . . . . . . . . . . . . Calibration Quality Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Establishment of Performance Characteristics . . . . . . . . . . . . . . . . . . . Assessment and Influence of Contamination .................... Estimation of Detection Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55 55 56 57 57 58 59 60 60 61
63 64 64 65 66 67 68 68 70 70 71
Contents
3.2.3.3 3.2.3.4 3.2.3.5 3.2.3.6 3.3 3.3.1 3.3.2 3.3.2.1 3.3.3 3.3.3.1 3.3.4 3.4 3.5
Recovery Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Precision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison with Alternative TechniquedMethods . . . . . . . . . . . . . . . Analysis of Certified Reference Materials . . . . . . . . . . . . . . . . . . . . . . . Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Frequency of Analysis and Choice of IQC Materials . . . . . . . . . . . . . Preparation of In-house IQC Materials . . . . . . . . . . . . . . . . . . . . . . . . . Establishment of IQC Target Values and Limits . . . . . . . . . . . . . . . . . Use of Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Defining a Quality Control Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . Systematic and Random Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . External Quality Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4
Application of ICP-OES Techniques in Environmental QC
IX 73 73 77 78 79 79 81 82 82 83 85 89 90
Terry C. Dymott 4.1 4.2 4.2.1 4.2.1.1 4.2.1.2 4.3 4.3.1 4.3.1 .1 4.3.1.2 4.3.2 4.3.3 4.3.4 4.4 4.5 4.6 4.6.1 4.6.2 4.6.3 4.6.4 4.6.5 4.6.6 4.7 4.7.1 4.7.2 4.7.3 4.8 4.8.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theory of the ICP-OES Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Atomic Spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principles of Atomic Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plasma Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumental Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spectrometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polychromators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Monochromators ............................................ Plasma Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Detectors and Signal Processing ............................... Monitoring of Environmental Pollution ......................... Pre-analysis Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Sample Collection Process ................................ Sample Treatment Before Analysis ............................. Instrument and Method Detection Limits ....................... Pre-concentration Techniques .................................. Analytical Conditions ........................................ Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Airborne Particulate Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analysis Conditions and Results ............................... Analysis of Soils, Sludges and Sediments ....................... Sample Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95 96 96 96 98 100 101 101 102 103 104 106 108 110 112 112 113 113 114 116 117 118 120 121 122 123 123
x
Contents
4.8.2 4.8.3 4.9 4.9.1 4.9.2 4.9.3
Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analytical Conditions and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plant and Biological Sample Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analytical Conditions and Results .............................
5
Practical Aspects of Monitoring Volatile Organics in Air
123 124 125 127 128 129
Elizabeth A . Woolfenden 5.1 5.2 5.2.1 5.2.2 5.3 5.3.1 5.3.2 5.3.2.1 5.3.2.2 5.3.2.3 5.3.2.4 5.3.3 5.4 5.4.1 5.4.2 5.4.3 5.5 5.5.1 5.5.2 5.5.3 5.6 5.6.1 5.6.1.1 5.6.1.2 5.6.1.3 5.6.1.4 5.6.1.5 5.6.2 5.6.3 5.6.4 5.6.4.1 5.6.4.2 5.6.4.3 5.6.5
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transfer of Analytes to the Capillary GC System . . . . . . . . . . . . . . . . Solvent Extraction or Thermal Desorption . . . . . . . . . . . . . . . . . . . . . . Focusing Trap Design ........................ ............ Air Sampling Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Whole Air Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Air Sampling Using Sorbent Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Minimizing Artifact Interference for Sorbent Tubes . . . .... Pumped Air Sampling Onto Sorbent Tubes ..................... Automating Pumped Tube Monitoring .......................... Diffusive Sampling Onto Sorbent Tubes ........................ On-Line Air Stream Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . Moisture Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Permeable Membrane Dryers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Desiccant Dryers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dry Purging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analytical Instrumentation and System Calibration . . . . . . . . . . . . . . Automation and Calibration of Whole-Air Container Analysis . . . . Automation and Calibration of Sorbent Tube Analysis . . . . . . . . . . . Calibrating Automatic On-Line Air Stream Analysis . . . . . . . . . . . . . VOC Air Monitoring Applications ............................. Workplace Air Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Benefits of Diffusive Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pumped Monitoring Applications in Workplace Air . . . . . . . . . On-Line Air Applications in Workplace Air Monitoring . . . . . . . . . . Standard Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data Interpretation ........... ............................ Industrial Emission Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mobile Emission Sources, ie, Vehicle Exhaust Testing . . . . . . . . . . . . . Urban Air Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ozone Precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Air Toxics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Roadside Air Concentrations of PAHs ......................... Indoor Air Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
133 135 136 137 140 140 141 142 147 147 147 151 151 152 152 152 154 155 156 157 158 159 159 160 163 163 163 164 166 167 167 172 174 175
Contents
5.6.5.1 Building Ventilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.5.2 Building Materials Emission Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.5.3 Monitoring Polychlorinated Biphenyls (PCBs) . . . . . . . . . . . . . . . . . . Atmosphere Research Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 Appendix A . Diffusive Uptake Rates on Perkin-Elmer Sorbent Tubes . . . . . Appendix B. Calibration Calculation Methods . . . . . . . . . . . . . . . . . . . . . . . . .
6
XI 175 177 177 178 180 183 187
Quality Control and Quality Assurance Aspects of Gas Chromatography-Mass Spectrometry for Environmental Analysis
Ray E. Clement and Carolyn J. Koester 6.1 6.2 6.3 6.3.1 6.3.2 6.3.3 6.3.4 6.4 6.4.1 6.4.2 6.4.3 6.5 6.6
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Use of GC-MS in Environmental Investigations . . . . . . . . . . . . QC/QA Aspects of GC-MS Instrument Operation . . . . . . . . . . . . . . Quality Control of GC-MS Instrumentation .................... Gas Chromatograph Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mass Spectrometer Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Description of Various MS Instruments and Capabilities . . . . . . . . . QC Considerations for Qualitative and Quantitative Analysis ..... Quality Control for Qualitative GC-MS Analysis . . . . . . . . . . . . . . . Quality Control for Quantitative GC-MS Analysis . . . . . . . . . . . . . . Isotope Dilution GC-MS Analysis ............................ QC/QA Considerations for Using GC-MS in Contracted Work . . . Summary and Conclusions ...................................
7
Application of Capillary Electrophoresis for Environmental Analysis
193 194 196 196 197 197 201 202 203 205 207 208 210
Saul M. Parry and Colin I;: Simpson 7.1 7.1.1 7.2 7.2.1 7.2.2 7.3 7.3.1 7.4 7.4.1 7.5 7.6 7.6.1 7.7
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Equipment Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Capillary Zone Electrophoresis ............................... Fundamentals of CZE Theory ................................ Electroosmotic Flow (EOF) ....... ........................... Additive Based CE Separations ............................... Micellar Electrokinetic Chromatography ....................... Isotachophoresis (ITP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ITP in the Presence of an Electroosmotic Flow . . . . . . . . . . . . . . . . . Bi-Directional ITP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Introduction Methods for CE ......................... Hydro-Dynamic and Electrokinetic Sample Introduction . . . . . . . . . Novel Sample Introduction Methods ..........................
213 213 213 214 217 218 218 220 220 224 226 227 227
XI1
Contents
7.7.1 7.7.2 7.7.3 7.7.4 7.8 7.8.1 7.8.2 7.8.3 7.8.4 7.8.5 7.8.6 7.8.7 7.8.8 7.8.9 7.9 7.9.1 7.9.2 7.9.3 7.9.4 7.9.5 7.9.6 7.9.7 7.9.8 7.10 7.10.1 7.10.2 7.10.3 7.10.4 7.1 1 7.11.1 7.11.2 7.1 1.3 7.1 1.4 7.12
Electrical Sample Splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rotary Valve Injector . . . . . . . . . . . . .................... Slider Valve Sample Introduction . . . . . . . . . ............ Membrane Sample Introduction . . . . . . . . . . . . . . . . . . . . Methods for Increased Sample Loading . . . . . . . . . . . . . . . . . . . . . . . Two-Dimensional Electrophoresis ........................ ITP-ITP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ITP-CZE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Coupled Column ITP-CZE .............................. Choice of Electrolyte System ITP-CZE . . . . . . . . . . . . . . . . . . . . On-Column ITP-CZE: Transient ITP and Sample Stacking . . . . . . Field Amplified Sample Injection (FASI) for CZE . . . . . . . . . . . . . . ITP-CZE With Applied Backpressure . . . . . . . . . . . . . . . . . . . . . . . . . Alternative Approaches to On-Line Preconcentration . . . . . . . . . . . . Detection Methods for CE ..................... .. Indirect Optical Detection Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . Indirect Optical Detection for ITP . . . . . . . . . . . . . . . . . . . . . . . . . . . . Indirect UV Detection for CZE . . . . . . . . . . . . . . ............. Optimisation for Indirect UV Detection in CZE . . ... Optimisation of Injection Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrolyte System Optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Addition of Organic Solvents to Electrolyte System . . . . . . Addition of Complexing Reagents to Electrolyte System . . . . . . . . . Conductivity Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contactless Conductivity Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . In-Contact Conductivity Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . Combined Conductivity and UV Detection . . . . . . . . . . . . . . . . . . . . . Suppressed Conductivity Detection ............................ Mass Spectrometric Detection for CE ......................... The CE-MS Interface .......................... CZE-MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ITP-MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ITP-CZE-MS ........................... ................ Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
Advances in Flow Analysis .Instrumentation and its Application in Environmental Analysis
227 228 229 230 231 231 231 231 232 234 236 239 240 240 241 241 242 242 243 243 243 245 246 247 248 248 250 250 252 254 255 258 259 260
Richard J Berman 8.1 8.2 8.2.1 8.2.2 8.2.3
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flow Analysis Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gas-Segmented Continuous-Flow Analysis (CFA) . . . . . . . . . . . . . . . Flow Injection Analysis (FIA) ................................ Sequential Injection Analysis (SIA) ...........................
267 269 269 272 274
Conrents
8.2.4 Comparison of Flow Analysis Techniques . . . . . . . . . . . . . . . . . . . . . . 8.2.4.1 Simple Chemistries: Fast Reactions and Few Reagents . . . . . . . . . . . 8.2.4.2 Complex Chemistries: Slow Reactions, Many Reagents. and Heating Stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4.3 Sample Dilution 8.2.4.4 Sample Preconcentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4.5 Low Level Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4.6 Sample Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4.7 Precision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4.8 On-Line Distillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4.9 Reagent Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4.10 Waste Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.5 Comparison of Flow Analysis with Other Techniques . . . . . . . . . . . 8.3 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.1 Pumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2 Injection Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 8.3.3 Samplers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.4 Analytical Manifold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.5 Heat Baths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.6 On-Line Distillation Baths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.7 Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.7.1 Improvements in Existing Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.7.2 Adaptation of New Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.8 Data Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Flow Analysis Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.1 Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.2 Dilution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.3 Derivatization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.4 Stopped-Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.5 Solvent Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.6 Gas Diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.7 Dialysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.8 Preconcentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.9 Removal of Interferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.10 Digestion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.1 1 Combined CFA with FIA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.12 Coupling with Other Instrumentation ......................... 8.4.13 Multicomponent Determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5 Environmental Application of Flow Analysis . . . . . . . . . . . . . . . . . . . 8.5.1 Inorganic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5.1.1 Alkali and Alkaline-Earth Metals ............................. 8.5.1.2 Transition Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5.1.3 Nonmetals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5.1.4 Actinide Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5.1.5 Anions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5.1.6 Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
XI11 275 275 27 5 275 276 276 276 276 277 277 277 278 278 278 279 279 279 280 280 280 280 282 282 283 283 283 284 285 285 287 287 288 290 290 291 292 293 293 294 294 295 297 297 298 300
XIV
Contents
8.5.2 8.6 8.6.1 8.6.2 8.6.3 8.6.4 8.7
Organic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quality Assurance Aspects of Flow Analysis . . . . . . . . . . . . . . . . . . . Calibration Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Initial and Continuing Calibration Verification . . . . . . . . . . . . . . . . . Calibration Blank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spikes and Duplicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
Application of Software in Environmental Auditing and Quality Control
300 301 302 302 302 302 303
Edward So0 and Miles Jack 9.1 9.2 9.2.1 9.2.2 9.3 9.4 9.4.1 9.4.2 9.4.3 9.4.4 9.4.5 9.4.6
Environmental Auditing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Mechanics of Auditing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Why Audit? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Implementing an Audit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Use of Computers in Auditing . . . . . . . . . . . . . . . . . . . . . . . . . . . Coursafe .The Computer Based Auditing Program . . . . . . . . . . . . The Heart of Coursafe ...................................... Using Coursafe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evaluation of Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multi-Audit Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Implementing Coursafe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Some Novel Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
309 311 311 312 314 315 315 316 318 321 321 324
325
Contributors
Richard J. Berman Wacker Siltronic Corporation P. 0. Box 83180 Portland, OR 97283-0180 USA (Chapter 8)
Joseph M. Levy ACCTA Inc. 5216 Karrington Drive Gibsonia Drive, PA 15044 USA (Chapter 2)
Ray E. Clement Ontario Ministry of Environment and Energy 125 Resources Road Ontario, Canada M9P 3V6 (Chapter 6 )
Saul M. Parry Department of Chemistry Birbeck College 29 Gordon Square London WC 1H OPP United Kingdom (Chapter 7)
Terry C. Dymott AT1 Unicam York Street Cambridge CB1, 2PX United Kingdom (Chapter 4 ) Miles Jack Edward Alandale Associates Ltd. P.O. Box 20 Nuneaton Warwickshire CVlO 8RW United Kingdom (Chapter 9 ) Carolyn J. Koester Lawrence Livermore National Laboratory P.O. Box 808 Livermore, CA 94550 USA (Chapter 6 )
Dean Rood J & W Scientific 91 Blue Ravine Road Folsom, CA 95630 USA (Chapter I ) Athos C. Rosselli Suprex Corporation 125 William Pitt Way Pittsburg, PA 15238 USA (Chapter 2) Ian L. Shuttler Bodenseewerk Perkin Elmer GmbH Postfach 101761 D-88647 Uberlingen Germany (Chapter 3 )
XVI
Contributors
Colin F. Simpson Department of Chemistry Birbeck College Gordon House 29 Gordon Square London WC 1H OPP United Kingdom (Chapter 7) Edward So0 Edward Alandale Associates Ltd. P.O. Box 20 Nuneaton Warwickshire CVIO 8RW United Kingdom (Chapter 9)
Elizabeth A. Woolfenden Perkin-Elmer 761 Main Avenue Norwalk, CT 06859-0001 USA (Chapter 5 )
1 The Use of Solid Phase Extraction for Environmental Samples Dean Rood
1.1 The Importance of Sample Preparation Nearly every sample needs some type of preparation before analysis. It is unusual when a sample collected from the environment can be injected directly into a chromatograph without detrimental consequences. The preparation may be necessary to remove or reduce sample components other than those of interest. These sample components may interfere with the analysis procedure and may reduce the ability to identify correctly and quantify the potential target analytes in the sample. In most cases, the sample is completely unsuitable for analysis without some type of preparation. Even if a sample was suitable for direct analysis, often the analyte concentrations are too low for detection. Sample preparation methods often result in substantial reductions in the sample volume, thus increasing the analyte concentration so that they can be more easily detected. There are numerous sample preparation (also called extraction) methods for environmental samples. Environmental samples range from drinking water to solids such as sludge and soils. Sample preparation methods may be to accommodate a wide range of sample types. Column chromatography and liquid - liquid extractions are among the oldest and most frequently used preparation methods. One technique called solid phase extraction (SPE) is suitable for many environmental samples. SPE requires relatively simple equipment, can process multiple samples simultaneously, provides superior sample clean up to most liquid - liquid extraction methods, substantially reduces solvent consumption, and reduces the time required to prepare samples for analysis. There are also robot SPE systems available that provide unattended and reliable sample preparation. The cost of materials per sample is equal, or in some cases lower, than for a corresponding liquid-liquid extraction. In nearly every aspect, SPE provides superior sample preparation to liquid - liquid extraction methods.
2
I The Use of Solid Phase Exfraction f o r Environmental Samples
1.2 Introduction to Solid Phase Extraction Solid phase extraction (SPE) involves passing a sample that is dissolved in a solvent through a bed of small, adsorbent particles. This material is usually packed into small tubes resembling miniature liquid chromatography columns. Depending on the design of the tube, vacuum or pressure is used to force solvent containing the sample through the adsorbent. The adsorbent material, called the sorbent, will retain select compounds in the sample. Some or all of the retained compounds then can be washed from the sorbent using an appropriate solvent. This solvent is collected for analysis or, if necessary, additional sample clean up. The properties of the solvents and sorbent will determine the amount of compound retention and the ease of compound removal from the sorbent. By careful selection of the sorbent and solvents, isolation of the analytes of interest along with a minimal amount of other sample components can be achieved. By removing as many of the other sample components as possible, detection and identification of the target analytes are much easier and accurate. If the solvent containing the extracted sample is evaporated and the remaining residue redissolved in a smaller volume than the original sample, a concentration of the sample occurs. This allows the detection of very low levels of analytes not possible by direct analysis of the original sample. Also, a change in the sample solvent is possible which may be beneficial for the analysis procedure.
1.3 SPE Formats There are several different designs for SPE devices. Each design has its advantages which are related to the sample type and volume, and the number of samples to be simultaneously extracted.
1.3.1 Syringe Barrel or Cartridges The syringe barrel type of SPE tubes are referred to as cartridges (Fig. 1-1). The cartridge bodies are usually made of serological grade polypropylene and terminates in a male h e r tip. Some barrels have a flared opening to accommodate robot gripper arms or large solvent volumes. Solvent reservoirs can be attached to the cartridges using coupling adapters to increase their volume. Frits are used to hold the adsorbent in place and to act as a particulate filter. Most frits are made from polyethylene and have 20 pm pores. Samples with large amounts of particulate material may block the frit and impede the flow of solvents through the cartridge. Glass or PTFE cartridge bodies and PTFE or stainless steel frits are available when very low contaminant
1.3 SPE Formats
Cartridge
Sep-pak
3
Fig. 1-1. Cartridge and sep-pak SPE tubes.
levels are desired. The cost for these materials are substantially higher than for plastic cartridges and frits. The most common sorbents are silica based materials. The particle shape can be irregular or spherical with irregular being the most common. The average particle diameter is 40 pm with 100- 120 A pores; however, larger particle sizes are not uncommon. Other adsorbents such as Florisil, Alumina and resins are also available as SPE sorbents. Solvent flow through a SPE cartridge is most often controlled by vaccum. A single cartridge can be processed using a side arm flask apparatus (Fig. 1-2). Multiple cartridges can be simultaneously processed by using a vacuum manifold (Fig. 1-3). There are several different manifold designs, but they all function in the same basic manner and are primarily distinguished by their number of cartridge positions. Racks are used to hold collection tubes beneath each cartridge. A variety of racks are available to accommodate many different types and sizes of collection tubes.
Fig. 1-2. Side arm flask apparatus for SPE cartridges.
4
I The Use of Solid Phase Extraction for Environmental Samples
onnect to vacuum source
Fig. 1-3. Vacuum manifold for SPE cartridges.
1.3.2 Syringe Filter or Sep-paks The syringe filter type of SPE tubes are most commonly referred to as sep-paks (Fig. 1-1). The same type of materials are used for the frits and tube body as the cartridge format. This type of SPE tube is designed to be placed on the end of a 5 - 50 mL h e r tip syringe (Fig. 1-4). Solvent is added to the syringe barrel and forced through the SPE sep-pak by the syringe plunger. This design is very convenient if only a few samples need to be processed, the SPE method is very simple, or a minimal amount of equipment is available. It is much easier to process a large number of samples with cartridge SPE tubes and a vacuum manifold. The sep-pak tube has to be removed from the end of the syringe or vented using an in-line stopcock before the syringe plunger can be safely removed from the syringe. For complex SPE methods, a large number of different solvents are used. This may require a large number of inconvenient manipulations of the syringe and SPE sep-pak tube.
Fig. 1-4. Sep-pak tube and solvent syringe.
1.3 SPE Formats
5
1.3.3 Disks The newest SPE format is the flexible disk. This format was introduced by 3 M and is referred by its brand name of EmporeTM disks. The 5 - 10 pm sorbent particles are intertwined with very fine threads of PTFE resulting in a disk about 0.5 mm thick and 47 - 70 mm in diameter. The disks resemble solvent filters. The disks are placed in a typical solvent filter apparatus and solvent is forced through the disk by vacuum (Fig. 1-5). A test tube is placed in the filter flask to collect the final extract. Multiple samples can be simultaneously extracted with a manifold setup (Fig. 1-6).
Fig. 1-5. SPE disk and filter apparatus (photo courtesy of 3M).
Fig. 1-6. Vacuum manifold for SPE disks (photo courtesy of 3M).
6
I The Use of Solid Phase Extraction for Environmental Samales
The thin sorbent bed and large surface area allows very rapid solvent flow rates through the disk. One liter of water can be passed through a disk in less than 10 min. Cartridge and syringe filter SPE tubes have a maximum flow rate of 5 - 10 mL min-'; faster rates result in poor sample extraction. Also large sample volumes are difficult to handle due to the limited volume and slower flow rates of cartridges and syringe filter SPE tubes.
1.3.4 Choice of Format The choice of SPE format will primarily depend on the volume and type of samples to be extracted. Small to medium volume samples are easier to handle with the cartridge or sep-pak type of SPE tube. A large number of samples is more easily handled with cartridge SPE tubes and a vacuum manifold; the sep-pak type of tube is best suited for a low number of samples or very simple methods. For large volume samples (ie, >50 mL) the disk format is best due to its high flow characteristics. Excluding the sorbent, PTFE is the only material in the disk, thus impurities from the disk are virtually non-existent. Impurities from cartridge or sep-paks tubes and frits can sometimes be detected at low levels. Disk are susceptible to plugging by particulates in the sample. If the sample contains a high amount of solids, filter paper can be added to the top of the disk to minimize clogging problems. The cost of cartridge and sep-pak tubes are very similar, and are primarily related to the size and the amount of sorbent. Disks cost about 2-4 times more than cartridges or sep-paks.
1.4 Using SPE Cartridges and Disks The methods and steps are the same for all of the SPE formats. There are four steps to most methods with each step fulfilling a specific function. Each step is distinguished by the solvent, and the solvents used will depend on the characteristics of the sorbent and sample.
:o >
Condition
Load
Fig. 1-7. Four steps of SPE methods.
Rinse
Elute
1.5 SPE Sorbents
7
The four SPE method steps are conditioning, loading, rinsing and elution (Fig. 1-7). Conditioning creates a sorbent environment compatible with the sample and removes impurities from the cartridge. Loading is when the sample is added to the cartridge and forced through the sorbent; the analytes and other compounds in the sample are retained by the sorbent. Rinsing removes (elutes) some of the unwanted sample compounds without removing any of the analytes. Elution is the removal of the analytes from the sorbent with the minimal amount of other sample compounds. The sorbent and compounds will directly influence which solvents are used for each step. The relationships must be understood before methods can be successfully developed or modified.
1.5 SPE Sorbents The most common SPE sorbents can be placed into three classes - normal phase, reverse phase and ion exchange. The structure of the sorbent determines its classification and characteristics; however, some sorbents may actually belong to more than one class and exhibit several distinct characteristics. The most common sorbents are based on silica. With the exception of unmodified silica, functional groups are bonded to the surface of the silica particle to alter its retentive properties. The structure of the bonded groups will determine the classification in which the sorbent belongs. There are several non-silica based sorbents in common use. Florisil and Alumina are the best known examples of these types of sorbents.
1.5.1 Normal Phase Sorbents Normal phase sorbents have polar functional groups bonded to the silica surface; unmodified silica, Florisil and Alumina are also included in this group (Fig.1-8). Due to this polar character, these sorbents have a stronger affinity for polar compounds than for non-polar compounds. Thus, a normal phase sorbent will retain polar compounds more strongly than non-polar compounds. The structure of the compounds and the sorbent will influence the strength of the retention. In general, normal phase sorbents are used to extract polar compounds, typically those with hydroxyl or amine groups. Phenols and nitrosoamines are examples of compounds strongly retained by normal phase sorbents.
1.5.2 Reverse Phase Sorbents Reverse phase sorbents have non-polar functional groups bonded to the silica particles (Fig. 1-9). These sorbents have a stronger affinity for non-polar compounds
8
1 The Use of Solid Phase Extraction f o r Environmental Samples
Sorbent
Structure
Si
Silica
-Si-OH
CN
Cyano
-Si-CH,CH,CH,CN
NH2
Amino
-Si-CH,CH2CH,NH,
20H
Diol
-Si-CH2CH,CH,0CH,-CH-CH
I I
OH OH Fig. 1-8. Normal phase sorbents.
than for polar compounds. Thus, a reverse phase sorbent will retain non-polar compounds more strongly than polar compounds. Like normal phase sorbents, the structure of the compounds and sorbent will influence the strength of the retention. In general, reverse phase sorbents are used to extract non-polar compounds, typically Sorbent
Structure
C18
Octadecyl
-Si-C,,H,,
C8
Octyl
-Si-C,H,,
C4
Butyl
-Si-C,H,
C2
Ethyl
-Si-CH2CH,
C1
Methyl
-Si-CH,
Ph
Phenyl
-si-Q
CH
Cyclohexyl
S -J(-i
Fig. 1-9. Reverse phase sorbents.
1.5 SPE Sorbents
9
those with substantial hydrocarbon character. Polynuclear aromatic hydrocarbons (PAH) and many pesticides are examples of compounds strongly retained by reverse phase sorbents.
1.5.3 Ion Exchange Sorbents Ion exchange sorbents have either a cationic or anionic functional group bonded to the silica particle (Fig. 1-10). When in the ionized form, the phase will retain compounds of the opposite charge. Any compounds with the same charge as the sorbent or that are neutral will not be retained by the sorbent. Thus, sample components with the opposite charge as the sorbent will be retained. There are two groups of ion exchange phases. Cation exchange phases will retain positively charged compounds (cations). Anion exchange phases will retain negatively charged compounds (anions). In general, ion exchange sorbents are used to extract ionic compounds typically those with a carboxylic acid or amine functional group. Triazine and phenoxyacid herbicides are example of suitable compounds for ion exchange sorbents. Sorbent
Structure
SCX
-Si-CH,CH,
Benzenesulfanic acid
' Q \
-SO;H'
(strong cation exchange)
PRS
Propylsulfonic acid
-Si-CH,CH,CH,-SOy Na'
CBA
Carboxylic acid
-Si-CH,CH,COO-H'
SAX
Quaternary amine
-Si-CH,CH,CH,N'(CH,CH3),CI
(strong anion exchange)
DEA
Diethylaminopropyl
-Si-CH,CH,CH,N(CH,CHJ,
PSA
Primarylsecondary arnine
-Si-CH2CH,CH,NCH,CH,NH2
NH2
Amino
-Si-CH,CH,CH,NH,
Fig. 1-10. Ion exchange sorbents.
10
I The Use of Solid Phase Extraction f o r Environmental Samples
1.6 Sorbent and Solvent Relationships For normal and reverse phase sorbents, compound retention and elution will be controlled by the solvent selection. The polarity of the solvent will determine the strength of the solvent. A strong solvent can be simply defined as a solvent that elutes a compound from the sorbent in a smaller volume than a corresponding weaker solvent. Thus, retention and elution of compounds is controlled by the polarity of the solvent passing through the sorbent. Ion exchange phases do not depend only on solvent polarity for retention and elution. Ion exchange phases depend primarily on the pH and the ionic species in solution for retention and elution.
1.6.1 Normal Phase Table 3 - 1 lists solvents in order of increasing solvent strength for normal phase sorbents. Note that solvent strength increases as solvent polarity increases. For example, a smaller volume of methanol is needed to elute a compound completely from a normal phase sorbent than chloroform or that more retention is obtained with chloroform than methanol. Sometimes a solvent may be too strong while the next weaker solvent is too weak for a particular step. Solvent mixtures are often used to overcome this problem and to finely control retention and elution. Unfortunately, it is difficult to predict the exact amount of retention or elution change upon changing solvents or mixtures. The general direction of the change is predictable, but not the absolute amount of retention or elution change. Each compound and sorbent is affected differently by the change in solvent.
Table 1-1. Solvent strength for normal phase sorbents.
Table 1-2. Solvent strength for reverse phase sorbents.
WEAKEST
WEAKEST
hexane iso-octane toluene chloroform dichloromethane tetrahydrofuran (THF) ethyl ether ethyl acetate acetone acetonitrile iso-propyl alcohol STRONGEST methanol
water methanol iso-propyl alcohol acetonitrile acetone ethyl acetate ethyl ether tetrahydrofuran (THF) dichloromethane chloroform toluene iso-octane STRONGEST hexane
1.6 Sorbent and Solvent Relationships
11
1.6.2 Reverse Phase Table 1-2 lists solvents in order of increasing solvent strength for reverse phase sorbents. Note that solvent strength increases as solvent polarity decreases. This solvent strength relationship is the opposite of normal phase sorbents. The solvents used with reverse phase sorbents are usually limited to water, methanol, iso-propanol and acetonitrile. Less polar solvents are often too strong for most reverse phase sorbents. As for normal phase, exact predictions about retention and elution changes with solvent changes are difficult.
1.6.3 Ion Exchange There are three major factors that determines retention and elution for ion exchange sorbents. They are solvent and sample pH, ionic strength of the solvents and counter-ion identity. Regardless of the type of ion exchange phase, there is always a cation and anion pair. One is the sorbent and the other oppositely charged species is the analyte. For analyte retention to occur, the sample must be at least 2 pH units below the p K , of the cation (to maintain a positive charge) and 2 pH units above the p K , of the anion (to maintain a negative charge). At a pH/pK, difference of less than 2, retention will suffer because the analyte or sorbent will be partially neutral instead of being completely charged. This factor also requires that the p K , difference between the analyte and sorbent must be 4 or greater. Elution of a retained analyte can be controlled by the pH of the elution solvent. If the solvent pH is at least 2 pH units above the p K , of the cation, significant elution will occur since the cation becomes neutral. If the solvent pH is at least 2 p H units below the p K, of the anion, significant elution will also occur since the anion becomes neutral. As long as the sorbent or analyte is neutral, elution of the analyte from the sorbent will occur. The pK,s for the most common ion exchange phases are listed in Table 1-3. Ionic strength is a measure of the total concentration of ionic species in solution. The retention and elution of analytes are a function of the solvent concentration of other ionic species with the same charge. These ions (called counter-ions) compete with the analytes for the limited number of binding sites on the sorbent. Low ionic strength solvents enhance retention since the analytes do not have to compete with a large number of counter-ions for the binding sites. High ionic strength solvents often inhibit retention since the high number of counter-ions may occupy most of the binding sites. High ionic strength solvents cause often elution of the analytes from the sorbent since the large number of counter-ions may displace the analytes from the sorbent. Counter-ion strength is a measure of the affinity of the counter-ion for the sorbent. High strength counter-ions have a greater ability to compete and bind to the sorbent than a weaker counter-ion. Relative counter-ion strengths are listed in
12
I The Use of Solid Phase Extraction for Environmental Samples
Table 1-3. Ion exchange phase pK,s.
Phase
PK, a
SCX PRS CBA NH2 PSA DEA SAX
2.2 -2.7 2.5 - 3.0 4.5 - 5.0 9.5 - 10.0 10.0 - 10.5 10.7 - 11.2‘ 10.5 - 11 .o Always charged
a
’,
pK,s vary slightly by manufacturer, thus a range is given.
’Primary amine.
Secondary amine.
Table 1-4. If there are ions present in the sorbent conditioning and load solvents, they should be weak counter-ions. The analyte has the best chance of displacing a weak counter-ion from the sorbent, thus enhancing retention. Solvents containing strong counter-ions will cause elution and poor retention since the counter-ions will easily displace the analyte from the sorbent.
Table 1-4. Counter-ion strengths.
Cation
Relative strength
Anion
Li, H Na NH.4 K, Mg, Mn, Fe Cu(l), Zn, Co, Cd Ca cum Pb, Ag Ba
0.5 1.5 2.0 2.5 3 .O 4.5 6.0 8 s 10.0
OH, F, propionate Acetate, formate HP,, HCO CI, NO, HSO,, CN NO3 CIO, Citrate Benzene sulfonate
Relative strength 0.1 0.2 0.4 1 .o 1 .5 4.0 4.5 9.5 10.0
For each ion category, the highest strength counter-ion was normalized to 10, thus the values are relative.
1.7 Selecting the Solvents The solvents for any SPE method depends on the analytes, sample and sorbent. A small amount of experimentation is usually needed to select the best solvents; however, it is often fairly easy to narrow the selections to a few possibilities.
1.7 Selecting the Solvents
13
1.7.1 Conditioning Solvents There are two purposes to conditioning. One is to remove any impurities from the SPE cartridge. The other is to create a sorbent environment to obtain adequate retention of the analytes. Usually two solvents are required to accomplish these tasks. The first solvent (initial) is used to clean the cartridge of impurities and contaminants. The second solvent (final) is to create a favorable environment for analyte retention. All SPE cartridges have some impurities due to manufacturing, handling and packaging. These impurities need to be removed prior to sample extraction since many of them may contaminate the extracted sample. Failure to remove the impurities may result in a final chromatogram with extra peaks that interfere with the analyte peaks. The use of a strong solvent as the initial conditioning solvent will remove nearly all of the potential impurities prior to sample loading. This initial conditioning solvent should be as strong or stronger than the elution solvent. In addition to cleaning the cartridge, the initial conditioning solvent serves another function. Most reverse phase sorbents are very hydrophobic, and they need some organic solvent to solvate or wet their surfaces. Without this layer of organic solvent molecules, poor extraction and difficulties with passing water through the sorbent may occur. The final conditioning solvent should be no stronger than the sample solvent. If too strong (eg, polarity, pH, ionic strength, etc.) of a final conditioning solvent is used, reduced analyte recoveries will usually result. No problems will usually be encountered if a solvent weaker than the sample solvent is used. Ideally the final solvent should be as close to the sample solvent as possible. If the sample solvent is buffered or pH adjusted, the final conditioning solvent should have similar properties. For example, if a water sample was adjusted to pH 2 with HCl, the final conditioning solvent should be water at pH 2 adjusted with HC1. Be aware of the ionic strength and counter-ion constraints for ion exchange sorbents. Conditioning solvent volumes are usually 1-2 mL per 100 mg of sorbent.
1.7.2 Loading Solvents The purpose of the load step is to retain the analytes on the sorbent. For retention to occur, the load solvent must be a weak one for the sorbent being used. In most cases, the load solvent is the solvent that comprises the sample. The load solvent can also be the solvent used to extract the analytes from a sample matrix. Sometimes the sample solvent may be too strong for use with the desired sorbent. Dilution of the sample solvent with a very weak, miscible solvent will decrease the overall strength of the original sample solvent. The strength of the resulting solvent mixture may now be weak enough to obtain adequate retention of the analytes by the sorbent. Weaker solvents will provide the best retention of the analytes. A very large volume (0.5- 1 L) of load solvent can be used as long as elution (breakthrough) of the analytes does not occur. For large volumes (> 100 mL) of aqueous samples and re-
14
I The Use of Solid Phase Extraction f o r Environmental Samples
verse phase sorbents in the cartridge format, it is recommended to add about 0.5% methanol (v/v) to the sample to keep the sorbent solvated.
1.7.3 Rinsing Solvents Once the analyte has been retained, the sorbent is usually rinsed to wash off some of the unwanted sample components. A solvent that is slightly stronger, or the same strength as the load solvent, is used. The rinse solvent should be just strong enough to remove as many of the unwanted sample components without eluting any of the analytes. Compounds not as strongly retained as the analytes will be eluted from the sorbent, thus they will not elute with the analytes during the elution step. While a rinse step is not essential, cleaner samples will usually be obtained. A rinse step also ensures that all of the sample comes in contact with the sorbent by rinsing any of the sample that may still be on the walls of the cartridge. Some methods may use more than one rinse solvent. Rinse solvent volumes are usually 0.5-0.8 mL per 100 mg of sorbent.
1.7.4 Elution Solvents Analytes are removed from the sorbent by use of an elution solvent which is collected. The best results are obtained when a solvent volume of 0.5 -0.8 mL per 100 mg of sorbent is used. Using too strong a solvent will result in the unnecessary elution of sample components. This will result in extracted sample containing more interferences than necessary. The use of the optimal solvent strength would leave these samples components retained on the sorbent while eluting all of the analytes. Using a very small volume of a strong elution solvent results in erratic or incomplete elution of the analytes. Using too weak an elution solvent results in excessive solvent volumes or incomplete elution of the analytes. After collecting the elution solvent, it may be handled in several ways. If the analyte concentration is high enough, it can injected directly into a chromatograph. Usually it is concentrated and the dry extract dissolved in another solvent for injection. For complex samples or analytes at very low concentrations, further clean up may be necessary.
1.8 Solvent Considerations 1.8.1 Solvent Volume Many SPE methods to not have a large margin of error in the solvent volumes; however, solvent volumes are often rounded to whole milliliter (mL) units for ease of
1.8 Solvent Considerations
15
measurement. It is important to be consistent and accurate in the volume measurements, especially for the rinse and elution solvents. Estimating volumes can lead to reproducibility problems. One of the benefits of SPE is the low solvent volumes required. Using large solvent volumes creates excess solvent waste and may compromise the results.
1.8.2 Solvent Miscibility The solvent comprising a mixture must be miscible. In addition, each solvent that is passed through the sorbent should be miscible with the prior solvent. If there is an immiscible solvent present, the next solvent may not properly interact with the sorbent. Poor recovery or insufficient sample clean up may occur. If the use of miscible solvents are not possible, drying the solvent bed between solvent additions may be necessary. Allowing nitrogen or air to pass through the sorbent for 10- 15 min will usually sufficiently dry the sorbent bed. Centrifuging the SPE cartridge at 1000- 1500 rpm for 5 min will also dry the sorbent bed.
1.8.3 Solvent Volatility Solvents that have very low or high boiling points can have a subtle impact on the overall method. Volatile solvents evaporate very rapidly, thus elution solvents can be quickly and easily concentrated. If a volatile solvent such as dichloromethane is used in a mixture, it may evaporate from the storage vessel much faster than the other, less volatile solvent. This will alter the solvent mixture and most likely negatively affect the extraction. Higher boiling solvents are not prone to this problem. The problem with higher boiling solvents is that they require long times and high temperatures to evaporate, thus concentration of the elution solvent may be slow.
1.8.4 Solvent Flow Rate In general, a slow flow of solvent through the sorbent provides better results. For the cartridge format, flow rates of 3-10mLmin-' are normal while flows of 10- 100 mL min-' for disks are common. The kinetics of ion exchange are slower than those for polar and non-polar interactions. Slower solvent flow rates are recommended to ensure sufficient time for the ion exchange interactions to occur. Flow rates should be balanced against the time required to pass all of the solvent through the sorbent. This is especially relevant for large volume samples. Even at 10mLmin-', a 200mL sample would require 20min to load. Increasing the flow rate to 25 mL min-' would reduce the time to 8 min. A sacrifice in analyte retention and recovery may occur, but this may be an acceptable situation when balanced against the savings in time.
16
I The Use o f Solid Phase Extraction for Environmental Samales
For most SPE method, using a consistent flow rate is most important. Large variations in flow rate, especially for large solvent volumes, will result in erratic recoveries and sample clean up. The small variations in flow rate due to slight differences in individual cartridges are not large enough to have a significant impact on the results.
1.9 Selecting Cartridge Size The size of a SPE cartridge is dictated by the amount of sorbent needed to retain the analytes. SPE cartridges are available in a variety of sorbent masses with typical sizes of 100-1000mg in 1 - 6 m L tube volumes. For normal and reverse phase sorbents, the sample capacity is 1-370of the sorbent mass. For example, a 100 mg sorbent can retain 1- 3 mg of sample. This mass does not include the solvent portion of the sample. For ion exchange sorbents, the sample capacities are much lower. Most important is the ionic strength of the sample; highly ionic samples require large ion exchange SPE cartridges. In general, there is about 30 times less capacity with ion exchange sorbents than for normal and reverse phase sorbents; however, this value can range over a large area. It is often difficult to determine the non-solvent mass of the sample, and estimates are usually necessary. In general, the larger the sample volume and complexity (dirtiness), the larger mass of sorbent needed. The mass of the sorbent, and to some extent the diameter of the tube or disk, have the most influence on capacity. In addition to increasing sample capacity, increasing diameter also increases the flow rate of the cartridge or disk.
1.10 Method Development The best sorbent and solvents for the extraction of select analytes from a sample matrix are usually not immediately obvious. Some selection can be made based on the structure of the analytes and sorbents; however, developing the best method usually requires some experimentation. Haphazard selection of sorbents and solvents will usually be very inefficient. A logical plan will be very beneficial in developing a satisfactory method. The easiest means to select solvents and evaluate sorbents is by use of an elution profile. A known amount of the analytes are loaded onto the sorbent using a solvent as similar to the sample solvent as possible. Then a series of suitable solvents or solvent mixtures of known volume and continuously increasing strength are added to the SPE cartridge and collected. These solvent fractions are analyzed using an appropriate analysis technique. The amount of the analytes in each fraction is calculated. A curve or table can be generated that expresses the absolute or relative amount of each analyte in each solvent fraction. From this curve or table, information about the retention and elution of the analytes can be obtained and a SPE method can be easily developed.
1.10 Method Development
17
An example elution profile for a 500 mg reverse phase sorbent would follow this type of sequence:
5 mL water (these solvents are not collected). - Load: 3 mL water containing a known amount of the analytes. - Rinse/elution: 3 mL water, 3 mL each 10% methanol, 20% methanol, 30% methanol etc., up to 3 mL 100% methanol. - Conditioning: 5 mL methanol then
By plotting the amount of each analyte versus the solvent fractions, a visible representation of retention and elution is obtained (Fig. 1-1 1). This can be done for several different sorbents or solvent mixtures to determine which one, or possibly more than one, is suitable. There are several pieces of useful information that can be obtained from the profile in Fig. 1-11. It shows that there was no analyte detected in the load solvent, thus the load solvent was weak enough for proper loading. If any analyte was detected, a weaker solvent or different sorbent would need to be tried. Some elution occurred in 30% methanol, thus the rinse solvent should be 20% methanol since the rinse solvent should be the strongest solvent that does not elute any of the analyte. All of the analyte was eluted by 70% methanol. Usually a slightly stronger solvent is selected as the elution solvent, thus 80% methanol would be selected for the example in Fig. 1-11. After all of the preliminary solvents have been selected, the various solvent steps need to be combined into a single method. A blank (solvent) sample containing a known amount of the analytes is extracted using the combined method steps. The elution solvent is collected and analyzed to determine the recovery. It is recommended also to try solvents slightly stronger and weaker than those determined by the elution profile. Often these small adjustments can result in large improvements in the final method. After satisfactory analyte recovery is obtained for a blank sample, a
100
90 U
2
80
a,
70 0
?
60
10
0
-
c
10
20
T
0
I
30
40
50
YO Methanol Fig. 1-11. Example elution profile.
60
70
80
90
100
18
1 The Use of Solid Phase Extraction for Environmental Samples
control sample (contains no analytes) needs to be extracted. This is to determine if sufficient sample clean up is being obtained with the sorbent and solvent selections. If there are any interferences in the chromatrogram, either adjustments to the solvent composition or volume will be necessary, or another sorbent will have to be selected. Sometimes a small amount of analyte recovery is sacrificed to obtain sufficient sample clean up. Finally, sample spikes will need to be run. A known amount of the analytes are added to a control sample. The recovery and clean up of the sample spikes will be the final test of the method. Recovery values with standard deviations at several analyte concentrations will verify whether the method is valid. It is not unusual for recoveries to be slightly lower or higher in spike samples than for spike blank samples. If satisfactory sample clean up and recoveries are obtained, the method is complete.
1.11 Matrix Considerations The sample matrix can have a large impact on the extraction method and results. A different method is often required for extracting the same analyte from different matrices. Soils, plant tissues and water all have to be treated differently. Even within a single type of matrix there still can be large differences. For example, water samples can range from drinking water to waste water, and soils can vary tremendously from sample to sample. Many liquid samples can be added directly to a suitable SPE cartridge. If the sample solvent is too strong for the sorbent, the sample can often be diluted with a much weaker, miscible solvent to decrease the overall strength. Samples containing large amounts of particulate matter may need to be filtered to minimize any plugging of the cartridge. Solid samples have to be extracted with a suitable solvent before addition to a SPE cartridge. The solvent needs to extract the analytes efficiently from the matrix and still be compatible with the SPE sorbent. In most cases, the extraction solvent will need to be diluted with a weaker solvent. For example, a solid sample is extracted with methanol, which is too strong for a reverse phase sorbent. To make the sample compatible, an aliquot of the methanol extract is diluted with 9 times the volume of water to obtain a 10% methanol mixture. This 10% methanol extract can be added to the reverse phase sorbent and satisfactory retention for the analytes is obtained. If the original solid sample was extracted with 10% methanol, very poor extraction of the analytes from the solid matrix would probably occur. If one sorbent does not provide adequate sample clean up, changing to a different sorbent that is still suitable for the analytes may be a solution. It is fairly common that more than one sorbent will provide satisfactory retention of the analytes. The different characteristics of each sorbent will result in a different amount and type of matrix components being retained. The interferences causing the problem may no longer be in the final extract, but the analytes are still present at high levels. For complex samples such as waste water and plant extracts, or for very low analyte concentrations, more than one SPE cartridge may be necessary for sufficient
1.13 Method Considerations
19
clean up. The two cartridges need to contain a different class of sorbent. For example, a reverse phase sorbent should be paired with a normal or ion exchange sorbent and not a different reverse phase sorbent.
1.12 Analysis Considerations The analysis method of the final extract will have a large impact on the type of sorbent and method. Depending on the analysis technique and detector, very different chromatograms can be obtained from the same sample. Greater sample clean up is usually needed when using more sensitive detectors or less selective detectors. In gas chromatograhy, the use of FIDs and MS in the full scan mode requires more sample clean up than when using ECDs, NPDs, FPDs and MS in the SIM mode. HPLC detectors are usually fairly selective, thus they require less sample clean up than many techniques. Chromatographic systems with high resolution capabilities sometimes require less sample clean up. These systems can resolve a large number of sample components from each other much better than a lower resolution system. Thus, the analytes can be more easily isolated from the other sample components in the chromatogram. Regardless of the analysis technique, better sample clean up will result in much less system maintenance and downtime. Hardware and columns will not get fouled by sample residues as often and will require less cleaning and replacement. The most common cause of system and column failure is due to contamination by sample residues.
1.13 Method Considerations Important details are not always provided with published SPE methods. Sometimes these details are the difference between successful and unsuccessful extractions. There are general considerations that apply nearly to every method; however, there are some exceptions. Fortunately, these exceptions are usually noted in the methods. Even though they may not be explicitly stated in the method, the following factors need to be considered or observed. Sometimes, a small gap between the upper frit and sorbent is observed. Unless it is very large, this gap will not have a negative impact on the extraction. If desired, after levelling the sorbent by gently tapping the cartridge on a hard surface, the frit can be pushed down to fill the gap. Sorbents of the same description from different manufacturers may not be exactly equivalent. Due to differences in manufacturing processes, starting materials, and product specifications, sorbents of the same description are often slightly different. In many cases, the differences will not affect an extraction to any noticeable extent. There are some cases when even slight differences can substantially change the results of the extraction. Small adjustments to a
20
1 The Use of Solid Phase Extraction for Environmental Samales
method may be necessary if the method was developed on a different brand of SPE cartridge. It is important that once the conditioning step has started, air should not be allowed to pass through the sorbent. Once the initial conditioning solvent is added, the sorbent must remain wet during conditioning. If the sorbent goes dry, start the conditioning process again at the beginning since channelling or improperly conditioned portions of sorbent may result. This may cause poor analyte recoveries or sample clean up. Unless extremely large excesses are used, using more than the recommended conditioning solvent volumes will not cause any problems. Make sure to meet or exceed the minimum solvent volumes for conditioning. Once the sample has been added to the cartridge, the sorbent can go to dryness and not cause any problems. It is fairly common to see sorbent drying steps in SPE methods. The purpose of the drying is usually to remove traces of previous solvents before adding the next solvent. An immiscible or difficult to evaporate solvent is usually removed from the sorbent to improve the results. Many ion exchange methods use aqueous buffers to obtain an elution solvent of the proper p H or ionic strength. If the sample needs to be concentrated after elution, the buffer and water usually needs to be removed since they are difficult to evaporate and may be harmful to a gas chromatograph. This is usually accomplished by using a simple liquid-liquid extraction of the elution solvent. Depending on the analyte, water immiscible solvents such as dichloromethane, ethyl acetate, hexane or toluene are normally used to extract from the aqueous media. Drying of the organic solvent extract with sodium sulfate is often recommended to remove any traces of water. Some methods recommend that a sample not be evaporated to complete dryness. This is to minimize the loss of volatile analytes that may evaporate upon reduction of the solvent volume. Also, some analytes may decompose or adsorb on the surface of glass tubes and vials if heated when not in the presence of a solvent. To quantitate samples accurately, the volume of final sample solvent must be known. If a sample is not evaporated to dryness, a graduated collection tube is usually necessary. Solvent is added to the graduated tube until a known volume is reached. For completely dry samples, this is not a requirement since a known volume of solvent can be added to the tube. The p H of some samples may vary over a wide range or are already close to the desired pH value. When using a ion exchange method for these types of samples, measure the p H of the sample before adding any pH modifiers. Not only will this help estimate how much of a pH adjustment is necessary, it may also indicate whether the sample needs to be made more alkaline or acidic.
1.14 Example Methods One of the best sources of SPE methods is previously published references (see Bibliography). These include journal citations and methods supplied by SPE cartridge manufacturers. In many cases, existing methods can be adapted to suit most needs.
1.14 Example Methods
21
A number of methods are presented as examples of typical SPE methods. The exact conditions will be dependent on the brand of SPE cartridge and sample matrix. Thus, these methods are intended only as examples.
Pesticides in Water (organochlorine, organophosphorus, carbmates, triazines and urea herbicides) Cartridge: Condition: Load: Rinse: Elute: Analysis:
C 18, 6 mL x 500 mg 5 mL ethy acetate, 5 mL methanol then 5 mL water Add 100 mL sample at a flow rate of no more than 10 mL min-' 3 mL water. Dry 3 mL ethyl acetate. Collect Concentrate to dryness. Add 200 pL ethyl acetate. Inject 1-2 pL using GC/FID, GC/ECD or GC/NPD.
Phenoxyacid Herbicides in Water Cartridge: Condition: Sample: Load: Rinse: Elute: Analysis:
SAX, 6 mL x 500 mg 5 mL methanol then 5 mL 0.01 M KOH Adjust 20 mL water to pH 7.0t0.1 with KOH or acetic acid Add sample to the cartridge at a flow rate of no more than 5 mL min-' 5 mL 0.01 M KOH. Dry. 5 mL dichloromethane. Dry 5 mL methanoV0.1 Yo citric acid (50/50 v/v). Collect Extract elution solvent with 3 x 2 mL dichloromethane. Combine the extracts and concentrate to dryness. Methylate for analysis using GC/FID or GC/ECD
Triazines in Soil Cartridge: SCX, 6 mL x 500 mg Condition: 5 mL acetonitrile then 5 mL 1% acetic acid Sample: Add 50 mL acetonitrile to 50 g of soil. Shake for 30 min. Filter extract. Add 25 mL of 170acetic acid to 5 mL of extract Load: Add sample to the cartridge at a flow rate of no more than 5 mL min-' Rinse: 2 mL 1 Yo acetic acid, 2 mL acetonitrile, 2 mL water then 2 mL 0.1 M K2HP04 Elute: 3 mL acetronitrile/O.l M K2HP04 (50/50 v/v). Collect Analysis: Extract elution solvent with 3 x 2 mL dichloromethane. Combine the extracts and concentrate to dryness. Redissolve in HPLC mobile phase and inject 10-20 pL using HPLC/UV, or redissolve in methanol and inject 1-2 pL using GC/NPD
Triazines in Plants Cartridge: SCX, 6 mL x 500 mg Condition: 5 mL dichloromethane Sample: Add 20 mL dichloromethane/acetone (4/1 v/v) and 5 g anhydrous sodium sulfate to 5 g of shredded plant. Shake for 1 min. Filter extract
22
Load: Rinse: Elute: Analysis:
1 The Use of Solid Phase Extraction for Environmental Samples
Add 3 mL of extract dropwise to the cartridge 3 mL acetonitrile then 3 mL water. Dry 3 mL methanol. Collect Concentrate to dryness. Redissolve in 200 pL methanol and inject 1-2 pL using GC/FID or GC/NPD
PAHs in Soil Cartridge: C18, 6 m L x 1000mg Condition: 5 mL dichloromethane, 5 mL methanol then 5 mL water Sample: Add 50 mL acetonitrile to 50 g of soil. Shake for 30 min. Filter extract. Add 36 mL of water to 4 mL of extract Add diluted sample to the cartridge at a flow rate of no more than Load: IOmLmin-’ Rinse: 5 mL 50% methanol. Dry Elute: 6 mL dichloromethane Analysis: Concentrate elution solvent to 50- 100 pL. Do not dry completely. Add dichloromethane to make final volume 200 pL. Inject 1-2 pL using GC/FID or G U M S
PCBs in Transformer Oil Cartridges: Florisil, 6 mL x 1000 mg attached to top of Silica, 6 mL x 1000 mg using a coupling adaptor Condition: Add 0.5 mL hexane to Florisil cartridge to wet the frit Add 0.2-0.5 g of oil directly to Florisil cartridge Load: Rinse: none Elute: 5 x 2 mL hexane. Collect Analysis: Mix sample and inject 1-2 pL using GC/ECD
EPA Method 525: Semivolatiles in Drinking Water C 18 EmporeTM, 47 mm Disk: Condition: 5 mL dichloromethane and allow disk to soak for 1 min. 5 mL methanol and allow disk to soak for 1 min. Leave -5 mm above disk. 5 mL water and leave 5 mm above disk Add 5 mL methanol to 100- 1000 mL water sample Sample: Load: Add sample to reservoir Rinse: none Elute: 5 mL ethyl acetate and allow disk to soak for 1 min after 2-3 mL has pass through the disk. 2 x 5 mL dichloromethane. Collect all fractions Analysis: Dry combined eluants with sodium sulfate. Remove sodium sulfate. Concentrate to no less than 0.5 mL. Inject 1-2 pL using GC/MS
-
Bibliography
23
Bibliography Allender, W. J. (1989) Determination of Chlorophenoxy Herbicides in Plant Materials Exposed to Spray Drift, J Chromatogr Sci 27, 193. Andrews, J. S., Good, T. J. (1988) Trace Enrichment of Pesticides Using Bonded-Phase Sorbents, Amer Lab, Apr 12. Chladek, E., Marano, R. S. (1984) Use of Bonded Phase Silica Sorbents for the Sampling of Priority Pollutants in Wastewaters, J Chromafogr Sci, 22, 3 13. Dimson, P. (1990) Isolation of Phenol and Substituted Phenols Using a Cyclohexyl Bonded-Phase Extraction Column with HPLC Analysis, LC, I , 236. Gardner, A.M., White, K. D. (1990) Polychlorinated Dibenzofurans in the Edible Portion of Selected Fish, Chemosphere 21, 2 15. Junk, G. A., Richard, J. J. (1988) Organics in Water: Solid Phase Extraction on a Small Scale, Anal Chem 60, 451. Loconto PR (I 992) Uitra-trace Determination of Chlorophenoxyacid Herbicides in Drinking Water, Amer Environ Lab 2/92, 32. Lopez-Ada, V., Milanes, J, Dodhiwala, N. S., Beckert, W. F. (1989) Cleanup of Environmental Sample Extracts Using Florisil Solid-Phase Extraction Cartridges, J Chromafogr Sci 27, 209. Marble, L.K., Delfino, J. J. (1988). Extraction and Solid-Phase Cleanup Methods for Pesticides in Sediment and Fish, Amer Lab Nov 23. Ozretich, R. J., Schroeder, W. P. (1986) Determination of Selected Neutral Priority Organic Pollutants in Marine Sediment, Tissue, and Reference Materials Utilizing Bonded-Phase Sorbents. Anal Chem 58, 2041. Pedersen, B. A., Higgins, G. M. (1988) A Novel Cleanup Technique for Organochlorine Pesticides and PCBs in a Complex Organic Matrix, LC-GC 6, 1016. Rostad, C. E., Pereira, W. E., Ratcliff, S.M. (1984) Bonded-Phase Extraction Column Isolation of Organic Compounds in Groundwater at a Hazardous Waste Site, Anal Chem 56, 2856. Wachob, G.D. (1984) Solid-Phase Extraction of Triazine Herbicides from Soil Samples, LC, 2, 756.
This Page Intentionally Left Blank
2 Current Status of Supercritical Fluid Extraction in Environmental Analysis Joseph M. Levy and Athos C. Rosselli
2.1 Introduction Before most chemical analyses can be performed, some type of extraction is usually required to remove suspected environmental contaminants from their particular environment matrix (ie, soil) and often this is a time-consuming and costly step. One of the most common extraction methods used in the US Environmental Protection Agency is SW 846 which often requires long extraction times (up to 24 h for Soxhlet methods), large quantities of toxic organic solvents, (ie, Freon, methylene chloride) and lengthy solvent concentration or clean-up procedures. Historically, as shown in Fig. 2-1, sample preparation involves as much as 75% of the total analysis time. Although much information has been published on SFE [I- 151, little guidance is available for analysts who want to incorporate this powerful sample preparation technology into quantitative analytical methods. This includes all segments of the extraction process (ie, analyte removal, analyte transfer, and analyte collection). The fact that SFE with carbon dioxide (CO,) has been touted as the solution to the organic solvent waste disposal problem, and that SFE is fast makes this methodology attractive for the analytical environmental laboratory of the 21st century. The successful collection of extracted analytes after SFE affords tremendous analytical flexibility in terms of the ability to collect into media that are suitable for subsequent chromatographic or spectroscopic characterization. In some
75%
Sample Analysis 5%
Fig. 2-1. Analytical sample cycle.
26
2 Current Status of Supercritical Fluid Extraction in Environmental Analysis
cases this can be done without any additional liquid solvent where the supercritical CO, vaporizes and only collected analytes remain.
2.2 What is a Supercritical Fluid? Supercritical fluid extraction (SFE) employs as a solvent a highly compressed gas that is near to or above its critical temperature and critical pressure. Figure 2-2 is a phase diagram of carbon dioxide (CO,) showing its triple point, critical temperature (31.4"C), and a critical pressure (73.8 atmospheres). At the critical temperature of a substance, the vapor and liquid phases have identical densities. A gas cannot be liquefied when it is above its critical temperature, regardless of the pressure. Above the critical temperature, the substance exists distinctly as a supercritical fluid. Carbon dioxide is the most widely used supercritical fluid. Supercritical CO, is a superior solvent over conventional liquid hydrocarbons, such as hexane or methylene chloride, for extraction, because it is chemically inert, non-toxic, non-flammable, non-corrosive, relatively inexpensive and environmentally friendly. Carbon dioxide has also been shown to possess excellent solubilizing characteristics for many target analytes of environmental, petroleum, chemical, polymer, food, pharmaceutical, and biological interest. Supercritical fluids possess favorable physical properties that are intermediate between those of the gas and liquid states, as outlined in Table 2-1. These physical properties (high diffusion coefficients, low viscosities, high densities and zero surface tension) result in the rapid and efficient mass transfer of target analytes from the sample matrix into the extraction solvent. Liquids have higher surface tension, which often inhibit the penetration of liquid solvent molecules into the pores of a heterogeneous sample matrix (ie, soils). For this reason, Soxhlet extractions of soils for example,
Fig. 2-2. Phase diagram of carbon dioxide.
2.2 What is a Supercritical Fluid?
27
Table 2-1. Comparison of gas, liquid and supercritical fluid properties. Gas
Liquid
Supercritical fluid
Low density Low viscosity High diffusivity
High density High viscosity Low diffusivity High surface tension
High density Low viscosity Intermediate diffusivity Zero surface tension
Solvent Molecules -b
Target Analytes
Liquid State
I Supercritical Fluid State Fig. 2-3. GraF ical comparison of liquid and supercritical fluid penetration of a heterogeneous matrix.
often need to be performed for many hours (8-24 h) to allow for the physical penetration of a liquid solvent (ie, methylene chloride) into a typical porous matrix (ie, soil). Supercritical fluids, on the other hand, have no surface tensions limitations which allow for a rapid penetration (minutes and not hours) of a sample matrix to achieve equal to or very often greater than the extraction efficiencies that a liquid solvent could achieve. This is depicted graphically in Fig. 2-3. More specifically, these physical properties of a supercritical fluid are important because:
1. The solubility of an analyte is directly related to supercritical fluid density. Thus any given analyte has a solubility in a supercritical fluid that is similar to the liquid phase solubility and much higher than the gas phase solubility (when all are at the same temperature). 2. Solute diffusion coefficients are greater in a supercritical fluid than in the liquid phase. When compared with liquids, higher diffusivity causes rapid extraction as
28
2 Current Status of Supercritical Fluid Extraction
iii
Environmental Analysis
the analytes have the ability to diffuse rapidly from the matrix into the extraction solvent. 3. Supercritical fluids have viscosities that are gas-like and much lower than liquid phase viscosities. This causes higher mass transfer of an analyte in a supercritical fluid than in a liquid and allows the solvent to penetrate easily the pores of a matrix. 4. The physical and chemical properties of supercritical fluids, such as solvation, diffusion and viscosity are a function of density. Thus, by changing the fluid’s density, it is possible to vary the fluid’s properties from gas-like to liquid-like without crossing the gadliquid, equilibrium line. Therefore, by changing physical parameters in SFE, chemical interactions can be directly affected. This cannot be done with any conventional liquid solvent. SFE is becoming more widely accepted for the extraction of analytes due to the ability to achieve rapid, sensitive and reproducible extractions of a variety of target analytes from many different types of heterogeneous matrices. For environmental applications, this also has significant impact on the reduction and/or elimination of toxic chemical exposure in the work place, since C 0 2 is a non-toxic and non-flammable solvent. Moreover, the need to purchase and then dispose of organic liquid solvents, such as Freon are dramatically reduced. Since supercritical CO, is a tuneable solvent, its solvating power can be adjusted to extract only the target analytes that are of analytical interest. For soils, for example, this allows for the elimination of subsequent clean-up steps since the extracts are inherently cleaner compared with Soxhlet (methylene chloride) extractions, ultimately enhancing analytical reliabilities. Supercritical CO, does not have sufficient solvent strength at the typical working pressures (ie, 80- 680 atm) in SFE to extract quantitatively analytes that are polar in nature. Generally, C 0 2 is an excellent extraction media for non-polar species such as alkanes and terpenes and is reasonably good for moderately polar species such as polynuclear aromatic hydrocarbons, polychlorinated biphenyls, aldehydes, esters, alcohols, organochlorine pesticides, and fats [ 151. The extractability and selectivity of supercritical fluids can be distinctly enhanced by the addition of miscible organic compounds or modifiers to the primary supercritical fluid. Even though a supercritical fluid has a variable solvent power, there is a maximum value of that solvent power. Adding modifiers, as low as 0.5 percent in concentration, increases the solubilizing power of that supercritical fluid mixture. Both the maximum solubilizing power and the solvent selectivity of the fluid mixture are determined by the chemical identity and concentration of the modifier. Some modifiers that have been used in SFE include methanol, propylene carbonate, acetonitrile, ethanol, formic acid, and tributyl phosphate [8, 161. Mechanistically, Fig. 2-4 depicts what has come to be referred to as the Hawthorne SFE triangle [17]. The predominance of the specific segment of this triangle will be dependent upon the type of sample matrix as well as the nature of the target analytes. For environmental applications, especially regarding soils, extraction rates are kinetically driven, with a dependence on the desorption or partitioning of the target analyte from the soil matrix as well as the binding energy associated with the
2.2 What is a Supercritical Fluid?
29
Solubility Modifiers and Temperature
Chromatographic
Partitioning Kinetics
-
Binding Energy
Matrix Characteristics
Fig. 2-4. SFE mechanism (Hawthorne's Triangle) [17].
matrix active sites. For this reason, the appropriate modifier or extraction temperature serves to augment target analyte extraction efficiencies or decrease extraction times. Practically, a generalized component breakdown is shown in Fig. 2-5 of a typical SFE instrument. Depending on the manufacturer, each of these components will vary in terms of differentiating features and capabilities. Generally, the pump (ie, VariPump T'-Suprex Corporation, Pittsburgh, PA) is used to supply the supercritical C 0 2 to the system and to pressurize the sample matrix that is contained in stainless steel extraction vessels. A separate modifier pump is used to add modifier dynamically during the extraction run. The extraction vessels can either be placed into the instrument manually or fed sequentially for automated SFE instruments (ie, AutoPre~-44~~-Suprex Corporation, Pittsburgh, PA). For environmental applications where representative sampling and homogeneity is a concern, sample matrix weights as high as 50g can be accommodated. Decompression from the elevated SFE pressures is accomplished in the restrictor which controls the flow of supercritical C 0 2 through the sample matrix and allows the target analytes to fall out of solution after the pressure has been released. This is normally controlled automatically (ie, VariFlowTM-SuprexCorporation, Pittsburgh, PA) where the C 0 2 flow is monitored and is continuously updated through-
Modifier
f
Fig. 2-5. Generalized SFE component breakdown.
30
2 Current Status of Supercritical Fluid Extraction in Environrnentul Analysis
out the duration of the SFE run. After decompression, the extraction effluent is channeled into an appropriate solid phase trap where the target analytes are adsorbed and immobilized for the extent of the dynamic extraction run. The target analytes are then desorbed off of this trap with an appropriate liquid solvent directly into a suitable septum-sealed vial (ie, GC autosampler vial). SFE consists of two steps: (1) the actual extraction of the target analytes from the matrix (desorption and removal) and (2) the collection of the extracted analytes. The collection step typically includes the depressurization of the supercritical fluid with analytes to an ambient pressure condition so that the analytes can be ‘collected’ for further analysis by some analytical technique. The SFE recoveries reported in the literature really consist of both steps: the extraction and the collection. Much of the early work in analytical SFE did not consider the importance of the collection step when, in fact, some of the low extraction efficiency values reported could really have been the result of not poor extraction efficiency but low collection recovery. Methods which have accomplished the collection of extracted analytes after the depressurization of supercritical fluid solvents are either in the on-line mode [ 14, 18, 191 in which the supercritical fluid with dissolved analytes is depressurized into a chromatographic system such as a gas chromatograph (GC) or a supercritical fluid chromatograph (SFC) system, or the off-line mode [6, 141 in which the supercritical fluid with dissolved analytes is depressurized into an intermediate collection system (ie, empty or solvent-filled vial or solid phase trap). Each of these methods of collection has its own advantages and disadvantages. On-line SFE collection requires minimal sample handling between the extraction and chromatographic steps. Moreover, analyte sensitivity enhancement obtained by on-line interfacing could be significant (ie, 1000-fold or greater). Also, for this reason, sample loadability of the extraction vessel is relatively small, because of the capacity of the analytical chromatographic column and detector. In comparison, off-line SFE collection requires an intermediate holding stage for the extracted analytes and ultimately involves analyte dilution. Usually the sample size can be much larger ( > 5 - 10 g) and the extraction can be achieved with a faster flow rate (0.5 to 7 mL min-’ compressed) as compared with on-line SFE. Moreover, different variables need to be considered during off-line collection such as restrictor flow inhibition, supercritical fluid flow during depressurization, collection and desorption (solid phase trap) temperature, choice of rinse solvent for desorption, and/or the type of sorbent or packing material for immobilization of the extracted analytes. Different methods of off-line collection have been employed to obtain higher collection efficiencies depending on the application [6, 20-231. The methods of off-line collection in SFE are shown graphically in Fig. 2-6. For environmental applications, each of these four modes of off-line collection has been employed. These methods of off-line collection in SFE are as follows: 1. Collection on a dry surface. The supercritical fluid with extracted analytes is decompressed directly into an empty bottle or some type of closed system. The closed system can either be at ambient pressure or pressurized above ambient. 2. Collection into an organic, liquid solvent. The decompression takes place directly into a vial filled with a liquid solvent such as methylene chloride.
2.3 Applicable Environmental Analytes and Matrices
31
3. Collection onto a solid phase bed. Decompression onto glass beads, unibeads, octadecyl silica, or some other inert supporthorbent in a closed system. 4. Collection into a cryogenically cooled adsorbent trap. Decompression into a cold solid phase trap that could be packed with an appropriate adsorbent cooled from ambient to as low as -50°C. This approach is especially applicable to wide boiling range analytes (ie, polynuclear aromatic hydrocarbons).
SFE Effluent
Empty Vial
Solvent Filled Vial
I
Adsorbent Trap
4
d Cryogenically Cooled Adsorbent Trap
Fig. 2-6. Graphical representation of the modes of off-line collection in SFE.
2.3 Applicable Environmental Analytes and Matrices SFE has been successfully applied for the extraction of environmental target analytes in soil, sand, silt, clay, fly ash, sludge, petroleum waster, river sediment, urban dust, diesel exhaust particulate, tar pitch, fish, grain and vegetable matrices. The target analytes that have been successfully extracted include the following general categories: 0 0 0 0 0 0
Polynuclear aromatic hydrocarbons (PAH) Total petroleum hydrocarbons (TPH) Pesticides Polychlorinated biphenyls (PCBs) Polychlorinated dibenzofurans (PCDFs) Polychlorinated dibenzo-p-dioxins (PCDDs)
Examples of the utilization of SFE for these specific compound classes will be discussed with current examples from various researchers in this field.
32
2 Current Status of Supercritical Fluid Extraction in Environmental Analysis
2.3.1 Polynuclear Aromatic Hydrocarbons and Polychlorinated Biphenyls Polynuclear aromatic hydrocarbons, of which a number have mutagenic and carcinogenic properties [24], occur in all environmental matrices because they are emitted from numerous natural anthropogenic sources (eg, traffic, industrial processes). Due to this fact, their rapid and precise determination is very important. While conventional Soxhlet extraction of PAH from soil is time-consuming and requires large amounts of organic liquid solvents, comparable results may be received with SFE in less than 1 h using C 0 2 . Some soils, particularly those with high carbon content, make extractions more difficult. Furthermore, the PAH solubility in supercritical CO, decreases with increasing number of fused aromatic rings. Due to these facts, SFE has been modified with an organic solvent (modifier) to reduce the affinity of the analytes for sorptive sites of the sample matrix and increase the PAH solubility in supercritical CO,. As SFE matures as a sample preparation technique, optimized methods will be developed for specific sample matrices and target analytes like PAH. In method development strategies, various stages of the entire SFE process need to be investigated. This includes three basic stages: pre-extraction strategies, extraction strategies and collection strategies. In pre-extraction strategies, various sample manipulation techniques such as grinding, freeze-milling or adsorbent addition can be employed to make the sample matrix more appropriate (more surface area or immobilized water) for SFE. Extraction strategies involve optimizing pressures, temperatures, sample size, durations, statiddynamic modes and modifier type and concentration. Collection strategies include off-line or on-line modes, restrictor flow rates, collection temperatures, desorption temperatures, adsorbent type, and type of wash solvent. Moreover, these strategies need to be optimized for heterogeneous sample matrices to achieve the ultimate goal of the highest possible efficiency in the shortest time with high precision. One of the first SFE operational parameters that has been investigated by researchers for PAH extraction was extraction pressure. Conventionally, most researchers tend to refer to this parameter first when developing a SFE method because of its relation to controlling the solubilizing characteristics of supercritical C 0 2 . Figure 2-7 shows a typical GC/MS total ion chromatogram of an off-line SFE effluent of PAH contaminated soil in work done by Levy et al. [7]. For this soil, the SFE conditions were 75 "C, 40 min (5 min statid35 min dynamic), with a compressed flow of 2.5 mL min-' for a 600 mg sample. The effluent was collected in a solid phase trap (silanized glass beads) at - 50 "C. Table 2-2 [7] shows the G U M S PAH results for this soil at pressures of 250 atm, 350 atm and 450 atm (keeping all other extraction and collection strategies constant). The concentration levels were determined using calibrated external standards and were compared with the acceptance range for each PAH as outlined by EPA method 8270. The highest pressure of 450 atm provided the best agreement with the EPA acceptance range for all of the PAH from two to five fused aromatic rings. The lower molecular weight PAHs (ie, two and three fused aromatic rings) were in fact within the acceptance range for all three of the
2.3 Applicable Environmental Analytes and Matrices 50.0-
7
8
1. 2. 3.
45.0
4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
40.0 35.0
30.0
33
Naphthalene Acenaphthylene Acenaphthene Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benzo(a)Anthracene Chrysene Benzo(b)Fluoranthene Benzo(k)Fluoranthene Benzo(a)Pyrene lndeno (1,2,3-~d)Pyrene) Dibenzo(a,h)Anthracene Benzo(ghi)Perylene
25.0 20.0
15.0 10.0
In :t
5.0
10.0
15.0
20.0 251.0 Time (min)
I
I
30.0
35.0
I
40.0
45.0
Fig. 2-7. Off-line SFE-GC/MS characterization of PAH contaminated soil (total ion chromatogram). SFE conditions.
extraction pressures. To fall within the acceptance ranges, the higher molecular weight PAH (ie, four and five fused aromatic rings) required higher solubilizing powers. This particular soil sample was contaminated with PAHs at higher levels, which were not apparently tightly associated with the matrix surface. Using the optimized extraction pressure, Levy et al. also determined the levels of precision for the various PAH for 12 replicate runs which are listed in Tabel 2-3 [7]. The relative standard deviation (RSD) varied from 5.0 to 13.4%. It is also important to note that the determined RSDs represent total system precision including the SFE sample preparation, the off-line cryogenic solid phase PAH trapping, and the G U M S analytical determination. No additional sample clean-up or work-up was required after SFE and before the GC injection. The extraction temperature in SFE has also been investigated with regard to enhancing extraction efficiencies for PAH as well as polychlorinated biphenyls (PCBs) in soils. Langenfeld et al. [25] examined the effect of temperature and pressure on the SFE of PAH and PCBs from 3 samples types: PCBs in river sediment (NIST SRM 1939), PAH in urban dust (NIST SRM 1649) and EPA certified PAH contaminated soil from Fisher Scientific (lot AQ 103). The samples had water contents of 3, 4 and 5 % respectively, and organics content of 10, 38 and 8070, respectively. Extracted amounts were 0.5 g for the soil, 0.3 g for the dust and 0.4 g for the sediment. A 0.5-mL extraction vessel was used for all extractions. The dynamic extraction dura-
34
2 Current Status of SuDercritical Fluid Extraction in Environmental Anrilvsis
Table 2-2. Off-line SFE/GC-MS of PAH contaminated soil-effect of extraction pressui-e 17). Compound
Naphthalene Acenaphthylene Acenaphthene Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benzo(a)anthracene Chrysene Benzo(b, k)fluoranthene Benzo(a)pyrene
EPA method 8270 acceptance range (ppm)
24.2 - 40.6 14.1 - 23.5 527 - 131 414-510 1210- 1966 313 - 411 1060- 1500 144 - 1322 214-290 211 -323 130- 114 80.1 - 114.3
Concentration levels (pprn) 250 (atrn)
350 (atm)
450 (atm)
23 20 566 445 1682 351 1028 103 74 14 <1.0 <1.0
23 21 60 1 47 1 I918 439 1459 1153 235 25 1 107 64
25 22 614 458 191 I 400 1571 1269 284 314 155 89
___
Table 2-3. Off-line SFE/GC-MS precision of PAH contaminated soil [7]. Compound
Acceptance range (PPm)
Average recovery (PPm)
To RSD a
Naphthalene Acenaphthene Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benzo(a)anthracene Chrysene Benzo(b, k)fluoranthene Benzo(a)pyrene
24.2 - 40.6 521 - 131 414-510 1210- 1966 313-411 1060- 1500 144 - 1322 214- 290 211 - 323 130-114 80.1 - 114.3
25+1 614+ 35.1 458 + 29.6 1911 +98.l 400 + 10.0 1411 + 91.1 1017 + 144.6 264+ 15.2 292- 15.1 149+ 16.5 85 + 7.8
5.0 6.0 6.6 5.4 2.6 6.9 13.4 5.8 5.2 11.1 9.3
a
Based upon 12 replicates.
tion was fixed at 40min, and the extraction fluid flow rate was maintained at 0.7 -0.9 mL of condensed fluid flow by using a linear fused silica restrictor. Extracted analytes were collected in acetone for the sediment and the methylene chloride for the other matrices. Collection efficiency was verified by performing extractions of spikes from sand, which gave recoveries of > 95%. For all of the samples types, no further clean-up of the extracts was performed after the extraction, and the only additional step performed was the addition of internal standard to the extract vials. RSDs were < 3% for the soils and 5 - 15% for the dust and soil, compared with 16- 19% for the certified values.
2.3 Applicable Environmental Analytes and Matrices
35
Table 2-4. Supercritical fluid extraction efficiencies of polychlorinated biphenyls from river sediment (SRM 1939) at different temperatures and pressures [25]. -
pCB congenera
Certified concentration levels (pg g - ' (VoRSD))'
Percent recovery (TO R S D ) ~ 650 arm, 200"C, 0.67 gmL-Id
350 atm, 200"C, 0.45 gmL-Id
150 atm, 200"C, 0.19 gmL-Id
650 atm, 50"C, 102 grnL-ld
350 atm, 5OoC, 0.91 gmL It'
67 (9) 86 (2) 81 (2) 104 (2) 90 (2) 116 (2) 113 (2) 125 (2) 147 (2) 144 (2) 142 (2) 125 (2)
77 (2) 90 (2) 85 (2) 108 (2) 96 (2) 121 (2) 124 (3) 136 (3) 161 (2) 151 (3) 163 (3) 144 (3)
60 (1) 78 ( I ) 80 ( I ) 103 (2) 80 (1) 120 (1) 104 (1) 118 (1) 151 (1) 137 (3) 146 (4) 112 (2)
34 (1) 39 (1) 53 ( I ) 66 ( I ) 50 ( I ) 87 (1) 77 (1) 93 (1) 118 ( I ) 99 (2) 116 ( I ) 98 ( I )
33 (3) 38 (2) 52 (1) 64 ( 1 ) 49 (2) 87 (1) 75 (2) 90 (1) 116 ( I ) 95 (3) 114 ( 1 ) 96 ( I )
~
2,2',5 2,4,4' 2,2',5,5' 2,3',3,5' 2,3',4,4' 2,2',4,5,5' 2,3',4,4',5' 2,2',3,4,4',5' 2,2',3,4',5,5',6' 2,2'3,3',4,4' 2,2',3,3',4,4',5,5' 2,2',3,3',4,4',5' a
3.46 2.21 4.48 1.07 0.93 0.82 0.51 0.57 0.18 0.10 0.16 0.11
(3) (5) (I) (11) (1) (I) (2) (2) (6) (10) (6) (9)
Each individual PCB congener is identified by its chlorine substitution pattern. Percent recoveries versus NIST-certified concentrations. Percent relative standard deviations are based on triplicate 40-min SFE extractions performed at each conditions. Concentrations reported by NIST based on two sequential 16-h Soxhlet extractions. Denotes supercritical fluid density calculated using the ISCO 'SF-Solver' and validated experimentally by the method in Ref. [26].
The best recoveries of 12 PCB congeners from river sediment were obtained at 350atm and 200°C at a density of 0.45 g m L - ' as shown in Table2-4 (25). Recoveries under these conditions were much better than what they were at 650atm and 50"C, even though the density of the latter case was significantly higher (1.02 g mL-'). The recoveries at 350 atm and 200°C ranged from 77 to 163% for an average of 121070, while at 650 atm and 50 "C they ranged from 34 to 1 19%, with the average being 78% recovery. The recoveries of 8PAHs from urban dust particulates was slightly better at 650 atm and 200°C than at 350 atm and 200°C as shown in Table 2-5 [25], but either conditions gave recoveries that were much better than at either pressure and 50 "C. The average recoveries were 96% at 650 atm and 200 "C, and only 31 Yo at 650 atm and 50°C. The effect of temperature was not so prominent in the case of the extraction of PAH, pentachlorophenol and heteroatomic PAH from the contaminated soil sample, with little difference noted between 50 " and 200 "C. There was, likewise, little difference between 350 and 650 atm, although recoveries did drop off at 150 atm. In most cases the SFE recoveries were greater than the 'certified' values which were obtained by conventional extraction methods, but no great significance was attached to this fact, since the authors noted that the certified values had quite large ranges, as in
36
2 Current Status of Supercritical Fluid Extraction in Environniental Atiulysis
Table 2-5. Supercritical fluid extraction efficiencies of P A H , heteroatom-containing P A H , and pentachlorophenol from high contaminated soil (EPA certified) at differen1 tempcraturej a n d pressures [25]. PAH
Naphthalene 2-Methylnaphthalene Acenaphthene Dibenzofuran Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benz(a)anthracene Chrysene Benzo(k)fluoranthene Benzo(a)pyrene Pentachlorophenol a
Certified concentration levels (pg g - ' (%RsD))~
24 (119) 57 (37) 16 (68) 306 (25) 476 (21) 1450 (39) 425 (16) 1307 (30) 961 (45) 249 (23) 311 (20) 156 (26) 98 (27) 884 (78)
Percent recovery (To RSD)" 650 atm, 20ooc, 0.67 gmL-"
350 atm, 2oo0c, 0.45 gmL-"
150 atm, 200°c, 0.19 gmL-"
207 (21) 152 (4) 135 (12) 147 (10) 127 (14) 121 (10) 97 (14) 144 (6) 132 (12) 97(ll) 103 (17) 119 (8) 98 (6) 309 ( 5 )
196 (8) 143 (4) 136 (4) 133 (4) 117 (6) 126 (4) 106 (5) 137 (6) 126 (5) 96 (4) 97 (4) 103 (4) 88 (13) 280 (2)
152 (36) 107 ( 5 ) 100 (4) 94 ( I ) 96 (1) 97 (5) 90 (4) 78 (11) 70 (10) 18(11) 18 (11)
95 (2) 47 (24) 136 (8)
650 atm, soot, 102 gmL
165 (10) 146 (10) 134 (15) 119 ( I 17) 125 (12) 135 (8) 107 (16) 146 (9) 137 (7) 98 (15) 98 (12) 85 (4) 65 (8) 229 (4)
350 atm,
jooc, 0.91 gmL
"
149 144 133 124
(I) (4) (2) (4) 115 (10) 127 (6) 106 (13) 142 (6) 135 (7) 107(12) 104 (6) 104 (2) 78 (23) 254 (10)
Percent recoveries versus certified concentrations. Percent relative standard deviations are based on triplicate 40-min SFE extractions performed at each conditions. Certified concentrations based on EPA methods SW846, 3540 and 3550. Denotes supercritical fluid density calculated using the ISCO SF-Solver and validated experimentally by the method in Ref. [26].
'
the case of naphthalene and pentachlorophenol which had values of 2 4 t 2 8 ~g g and 884 f 692 bg g - respectively. In summation, recoveries were more dependent on matrix type than other factors, and the effect of temperature cannot be explained on the basis of volatility alone, since recoveries were better for the less volatile PCB congeners than for the more volatile ones. Langenfeld et al. 1261 suggested that the best explanation was that the main effect was one of desorption rather than solubility, as long as there was enough solvent strength to dissolve the analytes, ie, above 150 atm and 200 "C for the PAH from dust, for example. The recoveries obtained thus with pure COz were better than or equal to those obtained previously with C 0 2 modified with 10% methanol at 400 atm and 100"C or with CHClF2 under similar conditions. Wenclawiak et al. [27] used SFE for the extraction of PAH (phenanthrene, pyrene, benzo(k)fluoranthene) from 3 soils of unknown origin. Sample size was 8 g, and extraction duration was 60 min, with the flow controlled by a linear fused silica restrictor. Extractions were performed at 390 atm and 100 "C. Recoveries of these three PAHs using SFE were described as comparable with those obtained using Soxhlet extraction for the three soils tested. Meyer et al. [28] extracted two to six fused aro-
',
2.3 Applicable Environmental Analytes and Matrices
37
matic ring PAHs from a high carbon content (50%) soil with toluene modified C 0 2 yielding comparable results with Soxhlet. The authors also noted that the conventional Soxhlet extraction, evaporation and clean-up procedure required 27 h compared with 3 h for SFE. Dankers et al. [29] used SFE modified for the high-speed and efficient extraction of PAH from polluted soil samples. Prior to SFE, a small amount of dichloromethane was added to a chemically dried and cryogenically ground soil sample. The SFE extract was collected within 15 -20 min and the PAHs were determined by HPLC equipped with fluorescence and UV detectors. Within-day and day-to-day reproducibilities were comparable with those obtained after a 4-h sample preparation including liquid - liquid extraction. Recoveries of samples spiked with PAHs were of the order of 100%. All sixteen individual (two to six fused rings) PAHs showed comparable recoveries, even within the extraction time of 15-20 min. This was significant because obtaining a rapid extraction and high recoveries that were independent of the PAH molecular mass has been a problem in many SFE experiments. The extraction of benzo(a)pyrene, a six ring PAH, has been reported to require 2 h of exhaustive extraction (density 0.76 g L-' 350 atm 90 "C) to achieve an extraction recovery of 90% [30]. The addition of methanol to CO, can improve the recoveries of some PAH. However, the larger compounds have only been partially extracted [3 l]. The extraction recovery of the high-molecular PAH can be improved by extending the extraction time, but this may not be acceptable when high-speed extractions are desirable. The addition of methylene chloride as a static modifier (prior to the extraction) in very small volumes (2 mL), was shown to give nearly 100% recoveries of the PAH that were independent of their molecular mass, and within a very short extraction time (15-20 min). In this case, the author suggested that the modifier penetrated the soil particles and solubilized the soil aggregates. In this way contact between the particles and the supercritical CO, were strongly increased. David et al. [32] reported the extraction of PCB from spiked and naturally incurred soil samples. Spike recoveries were generally good as long as SFE density Table2-6. SFE of PCBs from spiked sediment [32]. Extraction conditions Temperature ("C)
40 40 40
50 50 50 60 60 60
PCBs recovery (Yo)
Pressure (atm)
Density (gmL-')
Time (min)
100
0.628 0.781 0.781 0.385 0.700 0.700 0.290 0.605 0.724
30 30 15 30 30 15 30 30 30
150 150 100 150 150 200 150 200
28
52
101
153
138
180
Mean
91 106 33 91 80 38 89
74 104 44 90 82 37 98
85 97 33 46 102 78 31 66 100
I07 102 31 47 109 77 40 64 98
90 87 25 39 109 63 26 65 102
89 98 31 43 109 80 22 65
89 99 30 42 102 77 32 65 98
100
38
2 Current Status of Supercritical Fluid Extraction in Environmentai Analysis
Table 2-7. SFE of PCBs from a certified sewage sludge sample [32].
Extraction conditions
PCBs recovery (To)
Temperature ("C)
Pressure (atm)
Density (gmL-')
Time (min)
28
52
101
153
138
180
40 60 60 60 80
200 200 300 300 200
0.840 0.724 0.830 0.830 0.595
30 30
39 57 55 49 56
52 62 37 69 30
19 26 22 28 19
26 36 45 40 31
9 10 13
I I 10 10 5
a
30a 30b
30
11
7
Mean
25 33 30 35 26
100 pL methanol modifier was added. l 0 0 ~ L0.1 mol L - ' NH,C1 was added.
Table2-8. Quantitative results for optimized SFE of PCBs in certified soils [32].
PCB congener
26 28 52 101 118 138 153 180
SRM 1939
CRM 392
Certified value (Ng g - 9
Certified value (Pgg-7
4.20 9.07 4.48 0.82 0.51 0.57 NM 0.16
3.22 6.95a 4.84 0.90 0.50 0.61 NM 0.13
77 77 108 110 98 106 NM 88
NM
loo*
10 79-t 9 124+11 9 1 k 12 NM 288+ 18 313k24
Recovery (VO) NM 100 106 1 I4 65 NM 283 295
NM 100 I34 85 61 NM 98 94
NM, not measured. a Sum of PCB 28 and PCB 31
was greater than 0.7 g mL-'. Typical extraction conditions were 4O-5O0C, 100- 300 atm, 20 min dynamically for recovery data shown in Tables 2-6 to 2-8). These conditions were then applied to a certified sewage sludge (CRM 392), which yielded poor recoveries even at densities of 0.83 -0.89 g mL-' and temperatures up to 80°C. Recoveries were lower for the more highly chlorinated congeners (ie, 138-180), even with the addition of modifier to the vessel. The authors next attempted to optimize these conditions and found that at constant density, temperature had little effect. A temperature of 60°C was found to provide the best efficiencies. At constant density and flow, increasing density did not help much, with 0.75 g mL-' being optimum since, at higher densities, other components co-extracted and interfered with the effluent analysis. The biggest difference was observed with a change in flow rate. The initial work utilized a flow of only 10 mL min-' of gas-
2.3 Applicable Environmental Anafyfes and Matrices
r
Inject
39
sea-gull eggs [32]
eous flow, while the later work used expanded gas flows of 500 mL min-' . The effect of sample size was also noted, with quantitative recoveries obtained after 50 min for 100-mg samples while l-g samples were not completely extracted even after 120 min, perhaps due to fluid saturation. Good recoveries (88- 110%) were obtained for PCB congeners 52, 101, 118, 138 and 180 from a certified sediment sample SRM (1939) as listed in Table 2-8. Somewhat lower recoveries (77%) were obtained for congeners 26 and 28, perhaps because of saturation due to the high levels of these components present. Acceptable recoveries were obtained from the sewage sludge for congeners 28, 153 and 180, with no further cleanup required. For 101 and 118, recoveries were 85% and 67%, respectively. Recoveries were somewhat improved by the addition of methanol of the extraction vessel, but this resulted in the co-extraction of the matrix interferences, thereby complicating the subsequent analysis. The authors also extracted PCB from sea-gull eggs under conditions that minimized the amount of fat that was co-extracted as shown in Fig. 2-8 [32]. Moreover, PCB congeners were also extracted from milk powder and compared favorably with the classical method, except in the case of PCB 52. For some soil matrices, the use of pure supercritical C 0 2 has been inadequate to achieve quantitative yields of the entire PAH range (up to 6 fused aromatic rings). The addition of modifiers, such as methanol, toluene and methylene chloride, as described earlier, have augmented PAH recoveries to values that are in agreement with Soxhlet recoveries. Another approach has been to explore the effectiveness of in-situ chemical derivatization during SFE for the determination of reactive compounds [33, 341. Hills and Hill have reported the use of in-situ derivatization using modifiers to enhance recoveries of PAH from materials (NIST SRM 1649 and NRCC HS-3) [35]. The effect of the derivatization was to help displace the analytes from the
40
2 Current Status of Supercrifical Fluid Extraction in Environmental Analysk
Table2-9. Comparison of SFE and analysis of PAH in harbor sediments (HS-3) (kg analyte per g sediment) [35].
Phenanthrene Fluoranthrene Pyrene Chrysene
SFE-Tri-Sil a
SD
SFEb
SFE-10% MeOH
Certified value
SD
89.2 59.5 64.5 41.3
27. I 8.1 20 13.2
22.2 17.2 10.8 4.3
52 45 2.5 10
85 60 39 14.1
20 9 9 2.0
SD, standard deviation. a Average of triplicate GC-FID determinations of four extractions. Sample size 2 g at 350 atm, 60°C for 20 min. Uncertainties were not reported. Table 2-10. Comparison of SFE and analysis of PAHs in urban dust (SRM 1649) (pg analyte per g dust) [35].
Phenanthrene Fluoranthrene Pyrene Chrysene
SFE-Tri-Sil
SD
SFEa
Certified value
SD
9.97 6.88 6.97 5.83
1.62 1.60 4.36 2.61
0.63 6.30 4.5 0.96
4.5 7.1 6.5 3.6
0.3 0.7 0.6 0.2
SD, standard deviation. a Sample size 2 g at 350 atm, 60 "C for 20 min. Uncertainties were not reported.
matrix rather than increase their solubilities in supercritical CO,. SRM 1649 was an urban dust matrix containing 5 PAH, while HS-3 (National Research Council of Canada) was a harbor sediment containing 16 PAH and other organics. In this work only 4 PAH were quantitated: phenanthrene, fluoranthene, pyrene and chrysene. The derivatizing agent was a 2: 1 v/v mixture of hexamethyldisilane (HMDS) and trimethylchlorosilane (TMCS), which was added to the extraction vessel just prior to a static extraction step (ie, static modifier addition). This static step was continued for 15 min and was followed by a 15-min dynamic step. This was repeated 3 times, with the addition of the derivatizing agent added just prior to each static step. The total time for the complete extraction of 0.1 to 0.6g sample was approximately 4 h using an extraction pressure of 300 atm at 80°C. PAH recoveries as listed in Tables 2-9 and 2-10 were either equivalent to or better than the certified values. These recoveries ranged from 99-292% for the harbor sediment, and 97-222070 for the urban dust matrix. The use of the derivatizing agent also improved recoveries over what had been reported for the use of pure CO,, or CO, modified with 10070methanol. With the development of solid phase extraction (SPE), some investigations have been performed with SPE being used in conjunction with SFE. Ho and Tang [36],
2.3 Applicable Environmental Analytes and Matrices
41
for example, used a factorial (2 A 3) optimization of SFE of PAH from C-18 (octadecyl) SPE cartridges. The cartridges were cut-down and placed directly in the extraction vessel. PAH analyte solutions were deposited directly on the surface of the adsorbent prior to SFE. Levels were typically 25 pg per component with further simplex optimization carried out at 0 pg per component. For the initial optimization, pressure was varied from 100 to 300atm, temperature from 40" to 80"C, and time from 10 to 40min. Six PAH (2 and 3 fused aromatic rings) were recovered quantitatively at 200 atm, 60°C and 25 min. This work showed that the pressure was the most important factor, with extraction time being the next most important parameter. For this set of experiments, extraction temperature was found to be the least important. The above conditions were then further optimized using a variable size simplex optimization for PAH from 3 to 6 fused aromatic rings. This further optimization produced average recoveries of 85% for the 11 PAH spiked at the 10 pg level for extraction pressures from 350-400 atm temperatures > 40 " C ,and extraction durations of 25 - 35 min. Recoveries for chrysene, benzo(a)anthracene, benzo(a)pyrene, and dibenz(a,h)anthracene were further enhanced over the 95% level by the addition of methanol (300- 500 pL) directly to the extraction vessel.
2.3.2 Total Petroleum Hydrocarbons The US Environmental Protection Agency announced that the first SFE-based method (draft method 3560) is expected to be approved for routine extractions of solid samples [37]. The method was designed for the determination of total petroleum hydrocarbons (TPH) using 340-atm supercritical C 0 2 extractions, with collection of the extracted analytes in perchloroethylene, and quantitative determinations of the TPH concentrations by infrared spectrometry [38]. Due to the need for simple and portable methods of analysis in the field to support survey and remediation efforts, a natural extension of these reports was to place the SFE and the infrared spectrometer physically at the contaminated sites. In this regard, Hawthorne el al. [39] has performed correlation experiments comparing SFE with conventional Soxhlet extraction for T P H in soils. Using infrared spectrometric analysis, field measurements were performed. The nature of SFE made on-site sample preparation attractive since minimal glassware and liquid solvents were required. Table 2-1 1 lists the TPH concentrations determined from 30-min SFE runs performed at field sites, in a laboratory and by conventional Soxhlet extractions [39]. Total time between arrival at the sampling site and the beginning of the first SFE run was 15 -25 min and the TPH results were in agreement with the laboratory SFE results. This application demonstrated how SFE could be used to effectively provide immediate sample turnarounds in the field which could eventually be used to screen soil samples and reduce the total sample set sent our for laboratory characterization. Lopez-Avila et al. examined off-line SFE of petroleum hydrocarbons from various soils using Freon 113 and tetrachlorethylene (PCE) as a collection solvent [38]. A silica gel clean-up procedure was used to separate T P H from oil/grease, since TPH is defined as oil/grease minus the polar compounds removed by silica gel, with
42
2 Current Status of Supercritical Fluid Extraction in Environmental Analysis
Table 2-11. Comparison of SFE-IR determinations performed in the field and the laboratory with Soxhlet extraction [39]. Site (contamination) a
'
TPH concentration in pg g - (To RSD)b Field SFE
Farm (gasoline) Farm (diesel) Oil dump (motor oil) Fuel storage (light crude) a
6440 32 300 70800 8080
(5) (1 0)
(3) (3)
Lab SFE
Lab Soxhlet
6240 (5) 31 100 (3) 77600 (13) 8400 (13)
5480 (3) 38500 ( 1 I ) 83600 (9) 9700 (8)
Identification of the type of hydrocarbon contamination was based on GC-FID and GC-MS analysis of the samle extracts. Total pretroleum hydrocarbon (TPH) concentration as determined in each extract by IR. All SFE extraction wer performed in quadruplicate for 30 rnin and all Soxhlet extractions were performed in triplicate in 4 h.
oil/grease being defined as the material which can be extracted from an acidified soil sample by Freon 113. The total time for the analysis was approximately 40min, which included 30 min for the extraction and 10 min for the silica gel clean-up. Method accuracy was listed as 80% or greater, precision at &20% and method detection limit was 10 pg g-' of soil for a 3-g sample. This is in sharp contrast to the conventional TPH method which takes 4 h or greater, uses large amounts of Freon 113, and does not work for hydrocarbons lighter than # 2 diesel fuel. Test matrices were clay soil spiked with kerosene, light gas oil and heavy gas oil, PAH contaminated soil from Fisher Scientific (SRS 103-100), TPH-2 soil from Environmental Resource Associates, and 2 spiked clays and I native soil from NET. The clay soil spiked by the authors was spiked directly onto the surface of the soil about 15 min prior to extraction (15 rnin time being necessary to permit evaporation of the spiking solvent). For the contaminated soil SRS 103-100, for example, extractions were performed at 150, 250 and 350 atm, and at 55 O and 90°C. The best recoveries were obtained at 350 atm, 55°C and 60min. This gave recoveries of 98% of those obtainable by Soxhlet. Bicking et al. used an experimental design approach for SFE of TPH/Oil/Grease in soil [40]. This was the first report of a full experimental design approach for TPH in soil. Extracts were collected in an off-line vial containing Freon along with some 2-mm diameter glass beads. Celite was spiked with 775 and 9000 ppm of hexadecane and chlorobenzene respectively, by adding a spike solution to a Freodcelite slurry. The Freon was removed under vacuum at temperatures < 30 "C. Sample sizes (5-mL vessel) were 2 g for the Celite, and 5 - 6 g for the soils. A two factor central composite design was used to observe the effect of temperature and pressure on: a) recovery of hexadecane from Celite, b) density of the C 0 2 , c) volume (mL) of C 0 2 used and d) mass of C 0 2 used. This yielded recoveries in excess of 90% under all conditions except when the pressure was higher than 250 atm or the temperature was higher that 100°C. Density by itself was not considered to be a factor, since the authors noted that it was possible to obtain the same density with a variety of temperatures and pressures. The work also included the analysis of two native soil samples with high
43
2.3 Applicable Environmental Analytes and Matrices
Table 2-12. Comparison of extraction procedures for determination of hydrocarbons in a highcontamination-level soil sample [40]. Method
Concentration found (mg kg-')a Method 413.2'
Method 413.1 Soxhlet Shaker SFE (Freon)e SFE (dry)e
10500 (197, 9370 (225, 10200 (2210, 8400 (161,
Method
Percent of Soxhlet value
Soxhlet Shaker SFE (Freon)e SFE (dry)e a
'
n = 4) n = 4)
n
=
4)
n = 3)'
14100 (3720, 13 100 (519, 14500 (1 060, 8450 (1 200,
Method 418.1'
n = 4)
4) n = 4) n = 4)
n
=
12000 1 1 030 1 1 200 6700
(128, (436, (830, ( I 300,
Method 413.1
Method 413.2
Method 41 8.1
89.2 97.1 80.0
92.9 103 59.9
94.2 93.3 55.8
4) 4) n = 4) n = 4)
n n
= =
Standard deviation and number of replicates are given in parentheses. Sample was mixed with Na,SO, before extraction. Oil and grease, gravimetric method. Oil and grease, infrared method. Total petroleum hydrocarbons, infrared method. SFE conditions: dynamic extraction using CO, at 55 "C and 290 atm. Vials were not heated to constant mass before weighing.
and low levels of analytes as shown in Tables 2-12 and 2-13 [40]. The high level soil was extracted first at 290 atm and 55 "C and the results of SFE using Freon as a trapping solvent was compared with trapping in a dry vial, conventional Soxhlet extraction, and extraction by shaking with Freon. The SFE with Freon trapping gave results that were 93.3% to 103% of the Soxhlet value, depending on which analytical method was used. SFE with dry vial collection gave recoveries in the 55.8% to 80% range, while shaking with Freon gave recoveries of 89.2% to 94.2% of the Soxhlet values. The low level soil was also extracted under the same conditions, and SFE using a dry vial gave values that were 80% of the Soxhlet numbers, while collection in Freon gave lower recoveries, perhaps due to the loss of more polar compounds due to the fact that no drying agent was used and the more polar compounds were more strongly held by the water present than by the supercritical C02. Gere et al. provided an example of the significant reduction in organic liquid solvent consumption associated with the SFE of TPH in soil application [41]. In the preparation of multi-level (five) T P H calibration standards, iso-octane consumption was reduced by a factor of 100. Moreover, Gere's group discussed the benefits of automating SFE, including the preparation of calibration standards and the addition of internal standards. For calibration standards, conventional T P H sample preparation technology uses large volumes of solvents even though only small volumes of final solutions are required. Automated SFE eliminated wasted costs in solvent pur-
44
2 Current Sfatus of Supercritical Fluid Extraction in Environmental Analysis
Table 2-13. Comparison of extraction procedures for determination of hydrocarbons in a low-contamination-level soil sample [40]. Method
Concentration found (mg kg-
I)"
Method 413.1
Method 413.2'
Soxhlet Shaker SFE (Freon)e SFE (dry)e
2310 (164, n = 4) 1170 (216, n = 4)f
2560 2880 1360 2160
Method
Percent of Soxhlet value
Soxhlet Shaker SFE (Freon)e SFE (dry)e
(317, (905, (58, (177,
n = 3) n = 4) n = 4) n = 4)
Method 418.1'' 1130 (175, n 1560 (472, n 884 (50, n 975 (92,
4) 4) = 4) = 4) =
=
Method 413.1
Method 413.2
Method 418.1
90.2 45.7
112 53.1 84.4
138 78.2 86.3
~~~~~~~~
a
Standard deviation and number of replicates are given in parentheses. Sample was mixed with Na,SO, before extraction. Oil and grease, gravimetric method. Oil and grease, infrared method. Total petroleum hydrocarbons, infrared method. SFE conditions: dynamic extraction using CO, at 5 5 "C and 290 atm. Vials were not heated to constant mass before weighing.
Table 2-14. Automated determination of total petroleum hydrocarbons in wet sediment [42]. Replicate
Concentration (070)"
1 11 16 20 24 28 30 36 40 42
1.08 1.09 1.40 1.41 1.12 1.34 1.08 1.05 1.15 1.11
SFE operating conditions: 450 atm, 120"C, 35 min, VariflowTM at 3.0mLmin-' and lOO"C, C18hnibead trap at 100°C. a Target value: 1.1 Vo based upon GC-FID cumulative peak area measurements.
2.3 Applicable Environmentul Analytes and Matrices
45
chase and disposal. Moreover, Levy et a/. [42] has utilized sequentially automated (up to 44 vessel runs one at a time in a sequence) SFE to reproducibly extract TPH from wet sediment ( - 50 water). Total TPH concentrations as determined by GCFID are shown for up to 42 replicates in Table2-14 [42]. Using a VariflowT" variable restrictor at 100 "C and hydromatrix (diatomaceous earth) addition to the 1-g sediment samples, the excessive water was accommodated yielding average recoveries of 99.1 070 with 5.1 Vo RSD for the replicates.
2.3.3 SFE of Wet Soils Many of the naturally incurred soil samples that require characterization in environmental laboratories contain varying amounts of moisture. The presence of even small amounts of water (> 1070) can cause flow inhibition in SFE due to the freezing of water at the point of decompression after undergoing expansive cooling after depressurizing (Joule-Thompson effect). Moreover, maintaining extraction flow rates is difficult with very wet soils ( >20% water) due to compaction of the soil into the outlet of the extraction vessel frit even before the SFE stream reaches the restrictor. To avoid these problems, a number of techniques have been employed. Pre-extracting the soil sample at lower extraction pressures have not been effective and have still resulted in flow plugs. Dehydration techniques (pre-heating or air drying) have been time consuming and could promote the loss of volatile or semi-volatile analytes. Adsorbents have been used successfully to retain water such as hydromatrix (diatomaceous earth), magnesium sulfate, sodium sulfate, calcium chloride, molecular sieve and silica inside the extraction vessel. Burford et a!. studied the ability of several drying agents to eliminate restrictor plugging 1431. This included the use of three sample matrices (petroleum waste sludge, biosludge, soil) that were known to cause flow inhibition problems in SFE. Figure 2-9 graphically shows the results of the GC-FJD characterization of the extracts after SFE of a wet petroleum sludge comparing the following matrix treatments: (a) placing the sample on a bed of magnesium sulfate, (b) mixing the sample with magnesium sulfate, (c) air-drying the sample prior to SFE, and (d) oven-drying the sample prior to SFE [43]. Of the 21 drying agents tested, 1 1 reagents (hydromatrix, molecular sieves, Xanthan gum, carboxymethyl cellulose, magnesium sulfate, alumina, and Florid) successfully prevented water from plugging a capillary restrictor during extraction with 400 atm fo C 0 2 at 60°C. However, none of these drying agents irreversibly retained the water. Instead, the reagents reduced the rate at which the water was removed from the vessel. Increasing the extraction temperature to 150°C or adding a polar modifier (eg, methanol) to the C 0 2 greatly decreased the amount of water retained by the drying agents, whereas the addition of an aromatic modifier (eg, toluene) resulted in no such decrease. Even though none of the drying agents retained significant amounts of polar and non-polar test analytes in the presence of water, all the reagents except hydromatrix caused specific losses of one or more polar analytes when no water was present in the extraction vessel. Using the most effective drying agents (hydromatrix, molecular sieves 3 A and
46
2 Current Status of Supercritical Fluid Extraction in Environmental Anaiysis
A
10
i0
3b
Retention time (min)
40
Fig. 2-9. GC/FID analysis of the extracts after SFE of a wet petroleum sludge: (a) wet sample placed on a bed of anhydrous magnesium sulfate, (b) wet sample mixed with anhydrous magnesium sulfate, (c) sample air dried at room temperature for 18 h, (d) sample oven dried at 105°C for 1 h [43].
2.3 Applicable Environmentul Analytes and Matrices
47
CH,CI,
/
I t
+
inject
I
h
Sample pretreatment: None Variflow: 3 mlimin 100 "C Trap temperature: 100 "C Solid phase trap: C18/glass beads 35 min Duration: TPH concentration: 330 porn
Fig. 2-10. GC-FID characterization of automated off-line SFE of TPH contaminated soil containing 43% water [44].
5A, and anhydrous and monohydrated magnesium sulfate) a 1 : 1 wt/wt drying agenthample ratio prior to SFE was sufficient to prevent water from plugging the restrictor when used with moderately wet samples (20 wt% water), but a 4 : 1 reagent/sample ratio was required with very wet samples (90 wt% water). For samples containing volatile analytes, mixing the sample with exothermic drying agents (eg, magnesium sulfate) caused losses of volatile analytes from the heat of hydration as shown in Fig. 2-9. Moreover, new developments in restrictor technology have further increased the likelihood of extracting wet soils. Specifically, utilizing the VariflowTMautomatically variable restrictor with or without adsorbents (hydromatrix), soils with T P H contamination and 20 to 50% water content have been efficiently extracted using the a sequentially automated SFE [44]. Analytical gas chromatographic results highlighting T P H contaminated native wet soil extraction is shown in Fig. 2-10 [44]. This result demonstrates the ability to load the wet soil as received in the extraction vessel without any sample pretreatment when utilizing variable restriction. Thermally, both the restrictor and the solid phase trap were maintained at 100°C to keep the water emitting from the soil as a vapor after decompression. The excess water emitting from the soil was collected in an additional wash vial.
48
2 Current Status of Supercritical Fluid Extraction in Environnienrul Atialym
2.3.4 Pesticides As with all types of pesticides applied to the environment, determination of their residuals in plant, soil, animal life, and water is of great interest because of the harmful nature of their residuals. In the determination of such residuals, there are two methods that could be used to enhance trace analyte determinations in the matrices listed above. The first would be to isolate the analytes from the matrix by some type of sample preparation procedure, ie, extraction. Another would be to use a selective detector that is sensitive only to the analyte(s) of interest. In some cases, both of these techniques coupled together may be necessary because of matrix interferences. Some herbicides such as sulfonyl urea species are commonly used as pre- and post-emergence herbicides in order to control weed growth in cereal crops as well as in noncropland [45]. Howard and Taylor [45] investigated the use of SFE of sulfonylurea herbicides from aqueous matrices at the level of 50ppb with Empore discs. They obtained 97.2% (7.8% RSD) and 93.6% (5.5% RSD) recoveries for sulfometuron methyl (Oust) and chlorosulfuron (Glean) using supercritical C 0 2 modified with 2% methanol. The first part of their research was in trapping study where the analytes were recovered from Celite to determine the best off-line trapping conditions. The second part consisted of the analytes being recovered from Empore discs after being deposited on the discs from a 100- 1000 mL water sample. These discs were used as ordinary filter discs, but were also impregnated with C-8 (octyl), C-18 (octadecyl), etc, on a Teflon 'mat', which served to trap dissolved analytes from aqueous samples. Their use has been approved by the EPA for the analysis of phthalates and adipate esters, 2,3,7,8-TCDD, semi-volatile pesticides, organic chemicals, and PAH [45]. In conventional usage, liquid solvents are used to desorb the analytes from the disc. In this work, the analytes were desorbed by SFE. The best trapping conditions were found to be 15 "C for collection and 60 "C for desorption on 100-pm diameter stainless steel balls. Higher collection temperatures reduced recoveries, and the use of 30-pm C-18 instead of the stainless balls required excessive amounts of solvent for the desorption. The most effective desorption solvent was found to be acetonitrile. The attempt to recover the two analytes from Celite using pure supercritical C 0 2 was also made, with poor results. The pure C 0 2 extracted only 42.0% of the sulfometuron methyl and 49.8% of the chlorosulfuron compared with 92.5% and 94.8070 recoveries when 2% methanol modified CO, was used. When the Empore disc was used, C-8 was found to give better results than C-18, because the C-18 interacted more strongly with the analytes, making them more difficult to extract after adsorption on the disc. Moreover, lowering the pH to 3 improved the deposition of the analytes on the disc markedly. Unfortunately, this also made their recovery from the disc more difficult. This last obstacle was overcome by washing the disc with 2 mL of water at pH 9 just after the bulk of the water sample was run through the disc and while it was still wet. This resulted in the recoveries listed earlier. Compared with conventional techniques using the discs with liquid solvents to remove the analytes, the use of SFE reduced the use of solvents by 75% and yielded extracts with a much higher concentration of analytes, thereby eliminating an additional concentration step.
2.3 Applicable Environmental Analytes and Matrices
49
Snyder et a/. [46] extracted 12 organochlorine and organophosphate pesticides from soils using supercritical CO, and methanol modified CO,. The organochlorine pesticides used in this study were tetracholorometaxylene (TCMX), endrin, endrin aldehyde, p,p'-DDT, mirex, and decachlorobiphenyl (DCB). The organophosphate pesticides used in this study were dichlorvos, ronnel, diaxinon, parathion (ethyl), methidathion, and tetrachlorvinphos. These pesticides represented a range in polarity, molecular weight, volatility, and heteroatom composition. They varied density and pressure at constant temperature, modifier addition, temperature at constant density and the volume of supercritical fluid or time at constant flow. Extracts were collected off-line in 5 mL of methyl t-butyl ether. Concentrations ranged from 51 ng g-' for ronnel to 1040 ng g-' for diazinon. The pesticide target analytes were spiked onto the surfaces of various soil types just prior to SFE. The soil matrices consisted of sand, furnaced topsoil, river sediment, clay, and topsoil. Varying the density at constant temperature did not provide quantitative recoveries of either the organochlorine or organophosphate pesticides until the supercritical CO, was modified with 3% methanol, in which case recoveries of >90% were reported. This was true in all cases where the density was maintained above 0.5 g mL-' and the pressure above 100 atm. Varying the temperature at constant density required a pressure range of 112 to 400 atm, and had no effect on recoveries, except that the recovery of dichlorovos dropped at high extraction temperature (probably due to volatility). Varying the volume of CO, likewise had little effect, although it was noted that a minimum of 3 -4 vessel volumes were required for quantitative recoveries. The moisture content was found to give optimum recoveries when it was maintained between 5 and 10%. Recoveries decreased on either side of those values. For these spiked matrices, the authors found no benefit to using a static extraction prior to a dynamic one. The effect of pH was also examined, and it was found to help for some pesticides and hurt for others, so that the two effects cancelled out. This could be useful if only one class (acidic or basic) of pesticide were to be quantified. In a previous study, Synder et a/. compared SFE with the classical sonication and Soxhlet extractions in recovering these pesticides from fortified and native contaminated soils. Statistical evaluations were made of the precision and accuracy of SFE and the classical extraction methods [47]. Levy et al. used modifiers in the SFE of pesticides from naturally incurred soils [81. Figure 2-1 1 shows the GC electron capture detection (ECD) characterizations of pesticides in soil after off-line collection in a solid phase trap (glass beads) which was washed with methanol after dynamic extraction. Sample sizes for extraction that were used were 2 or 3 g (3-mL vessel) with flow rates of 1.5 to 2.0 mL min-' (compressed). The pesticide SFE recoveries were referenced to methanol liquid/solid extractions. During the dynamic extraction mode, modifier was also collected in the collection vial. However, no loss of pesticide analytes was seen since the collection vials were septum sealed. The pesticide concentration levels (percent recoveries) were determined by external standards and were all measured in triplicate. Modifier addition was accomplished using a separate modifier pump. The lower collection temperature ( - 5 0 ° C ) was necessary to achieve an average recovery of 95070 as opposed to a higher collection temperature of 40 "C (with only 34% average pesticide recovery). The lower collection temperature was necessary because of the relative volatilities of
50
2 Current Status of Supercritical Fluid Extraction in Envimntnental Anal-vsir
750.0-
700.0-
Yo Recovery
650.0-
Methanol: Tributylphosphate
600.0 -
95% 96% I
1
550.0500.0-
450.0400.0 350.0I
I
Inject
I/ 1 1 5
Heptachlor
10
Aldrin
1
15
20
I
1
25
30
Retention time (min)
Fig. 2-11. Off-line SFE-GC-ECD of pesticides in soil: methanol versus tributylphosphate [8]
the earlier eluting pesticides. With the use of a defferent modifier, namely tributylphosphate, (C4H&P04, similar SFE efficiencies were obtained [S]. As a new modifier (compared with the more common use of methanol) tributylphosphate, which is a complexing agent, appeared to be more effective than methanol in augmenting pesticide extraction efficiencies from the soil. For example, it was interesting to note that only 0.5% tributylphosphate compared with 5 % methanol was required to obtain similar recoveries for the target pesticides. Standard methodologies for the determination of 2,4-D compounds (chlorophenoxy acid herbicides) in soil include extraction with an organic solvent, hydrolysis of the esters to the acid and derivatization of 2,4-D acid to the corresponding methyl ester with diazomethane SFE coupled with immunochemical methods offer an alternative means for the rapid determination of 2,4-D compounds in soil [48]. In principle, a soil sample containing the 2,4-D compound is extracted either with supercritical C 0 2 or C 0 2 modified with an organic solvent such as methanol. The extracted material is then collected in methanol or phosphate buffer and analyzed by enzyme-linked immunosorbent assay (ELISA). Lopez-Avila et al. developed a method that combined SFE with an ELISA technique for determination of the 2,4-D acid, 2,4-D salt and 2,4-D ester in soil and cotton material [48]. The particular ELISA used in this study employed paramagnetic particles that were coated with an anti-
2.3 Applicable Environmental Analytes and Matrices
51
Table 2-15. Percent recoveries of 2,4-D acid from spiked topsoil, clay soil and cotton material by SFE with CO, containing 5% methanola [481. Matrix
Percent recovery ('70RSD)
-~
Low spikeb
High spike'
Topsoil Clay soil cotton
(9.9) 103 83.3 (21.1) 94.7 (18.0)
100 (6.4) 106 (15.4) 90.0 (9.7)
a
The extractions were performed at 400atm, 80"C, for 30min (dynamic). The extracted material was collected in 2 mL methanol, solution adjusted to 5 mL with methanol, and then diluted 10-fold with reagent water prior to ELISA. The calibration was performed with the 2,4-D acid standard. The numbers shown are the averages of three determinations. The low-spike level is at 100 ng g-' for soil samples and 500 ng g - ' for cotton material. The high-spike level is a 200 ng g-I for soil samples and 1000 ng gfor cotton material.
body preparation selective for 2,4-D compounds. For the ELISA, the methanol extract obtained by SFE was diluted 10-fold with reagent water and then transferred to a disposable test tube, followed by the addition of the enzyme conjugate of 2,4-D and the paramagnetic particles. After the incubation period, a magnetic field was applied to separate the paramagnetic particles with the bound 2,4-D and labeled 2,5-D analog. The presence of 2,4-D was then detected by adding the substrate (hydrogen peroxide) and the chromogen (3,3',5,5'-tetramethylbenzidine), which resulted in a colored product. The color intensity, measured at 450 nm was inversely proportional to the concentration of 2,4-D in the sample. The lowest concentration of the 2,4-D acid that was determined by the ELISA was 1 ng mL-'. This translated to 10 ng g-' in soil when a 5-g soil sample was extracted. Using 5 percent methanol modified C02, the percent RSDs reported in TabIe 2-15 were comparable (6 - 22 To) with those achieved by using GC. Lopez-Avila et al. [49] investigated the SFE of chlorophenoxy acid herbicides from soil samples with supercritical C 0 2 as extractant and tetrabutylammonium hydroxide and methyl iodide as derivatization agents. The extraction was carried out at 400 atm and 80 "C for 15 min static, followed by 15 min dynamic, at a C 0 2 flow rate of approximately 1.5 mL min-' (compressed). The use of other derivatization agents (trimethylphenylammonium hydroxide, benzyltrimethylammonium chloride, and benzyltriethylammonium chloride) proved to be less effective than the tetrabutylammonium hydroxide/methyliodide combination. Attempts to extract other compounds currently listed in the EPA SW-846 Method 8151 using supercritical C 0 2 and tetrabutylammonium hydroxide/methyl iodide were unsuccessful, either because these compounds did not derivatize or because decomposition occurred (apparently in the injection port of the GC). An in-situ derivatization and SFE of
52
2 Current Sfafus of Supercrifical Fluid Extruction in Environmental AnulyJir
the chlorophenoxy acids with pentafluorobenzyl bromide/trithylamine reagent also proved to be feasible for qualitative determination of these compounds. Hawthorne et al. also reported the extraction and methylation of 2,4-D and dicamba from sediment samples using supercritical fluid derivatization/extraction n ith trimethylphenylammonium hydroxide (TMPA) or BF,/methanol as derivatizing agents [34].
2.3.5 DibenzofurandDioxins Since chlorinated dibenzofurans have been detected in the emissions from municipal incinerators, it is important to enhance the capability of analytical methodologies for their fast and more reliable extraction and clean-up from fly ash matrices. Onuska et al. [50]developed a multi-residue SFE procedure for chlorinated dibenzofurans (PCDFs) from fly ash samples. The results were compared with those obtained by Soxhlet extraction. Extracts from the two procedures were analyzed by G U M S . Statistical analysis of the results confirmed that SFE provides data with a relative standard deviation of less than 10% while Soxhlet extraction data showed a much greater spread. In another study, Onuska described a SFE technique as a part of the determination of 2,3,7,8-tetrachlorodibenzo-p-dioxin in sediments [5 11 and all polychlorinated dibenzo-p-dioxins (PCDDs) congeners in fly ash [52].Nitrous oxide plus 2% methanol or 5% toluene were used as the modifiers in the supercritical fluid. It was demonstrated that the SFE of polychlorinated dibenzo-p-dioxins from fly ash matrices was governed by a variety of interrelated parameters, including the affinity of the solutes for the matrix, the solubility and the vapor pressure of the analytes and the diffusivity coefficients of the analytes in the supercritical fluid.
2.4 Conclusions Out of the numerous application areas for using SFE as a sample preparation technique, the extraction of environmental pollutants out of soils is perhaps one of the most straightforward and potentially the most capable of rapid and efficient extractions. This is partially due to the inherent porous nature and loose texture of soils and the fact that supercritical fluid technology has evolved to the extent that instrumental enhancements have provide increased capabilities for researchers to use in method development. SFE technology today provides environmental chemists with the opportunity to utilize new and automated sample preparation techniques to open up virtually new frontiers. This includes the use of environmentally friendly, politically acceptable solvents (ie, COz), the decrease of sample turnaround times, the elimination of toxic chemical exposure, and the elimination of solvent disposal costs. Moreover, SFE can improve the overall reliability of analytical results and establish
References
53
new standards for efficient monitoring of the environment. The numerous examples presented in this compilation attest to the legitimacy of the use of SFE in logical preference to conventional Soxhlet extractions that are currently being done. With the key development of variable restriction and automation, SFE has reached a new pinnacle for economic application of this technique. Further developments in SFE will involve the evolution of automated method development protocols, the automated interfacing on-line to gas chromatography, and the investigation of modifiers (primary and secondary) and solid phase trapping for selectivity enhancement.
Acknowledgements The devoted and perserverant efforts of Carol Hamilton in typing this manuscript and the photocopying efforts of Marcie DeFazio are gratefully acknowledged.
References [ t ] Hawthorne, S.B., Miller, D. J., J Chrornatogr Sci 1986, 24, 258-264. [2] Hawthorne, S.B., Krieger, M.S., Miller, D. J., Anal Chem 1989, 61, 736-740. (31 Hawthorne, S.B., Krieger, M.S., Miller, D. J., Anal Chem 1988, 60, 472-477. [4] Levy, J. M., Rosselli, A. C., Chromatographia 1989, 28, issue 1 I / 12. (51 Wright, B. W., Frye, S. R., McMinn, D. G., Smith, R.D., Anal Chem 1987, 59, 640-644. [6] Levy, J. M., Houck, R. K., American Laboratory 1993, 36R-36Y. [7] Levy, J. M., Dolata, L. A,, Ravey, R. M., J Chromatogr Sci 1993, 31, 349-352. [8] Levy, J. M., Dolata, L.A., Ravey, R. M. Storozynsky, E., Holowczak, K.A., J High Resolution Chromatogr 1993, 16, 368 - 37 1. [9] Ashraf-Khorassani, M., Houck, R. K., Levy, J. M., J Chromatogr Sci 1992, 30, 361 -366. [lo] Hawthorne, S.B., Miller, D. J., Krieger, M.S., J Chromatogr Sci 1989, 27, 347-354. [I I ] Nielen, M. W. F., Sanderson, J.T., Frei, R. W., Brinkman, U. A.T., J Chromafogr 1989, 474, 388- 395. [I21 Hawthorne, S.B., Miller, D. J., Langenfeld, J. J., J Chromatogr Sci 1990, 28, 2-8. [I31 Sugiyama, K., Saito, M., Hondo, T., Senda, M., J Chromatogr 1985, 332, 107- 116. [I41 Hawthorne, S.B., Anal Chem 1990, 62 (ll), 633A-642A. [I51 Levy, J. M., American Laboratory 1991, 23-32. [I 61 Levy, J. M., Ritchey, W. M., J Chromatogr Sci 1986, 24, 242. 1171 Hawthorne, S. B., Presented at the F i f h International SFE/SFC Symposium, January 11-14, 1994, Baltimore, MD. [I 81 Levy, J. M., Ritchey, W. M., J High Resolution chromatogr 1987, 10, 493. [ t 9 ] Levy, J.M., Rosselli, A.C., Storozynsky, E., Ravey, R.M., Dolata, L.A. Ashraf-Khorassani, M, LC- GC Magazine 1992, 10 (5), 386- 391. [20] Onuska, E I., Terry, K. A., J High Resolution Chromatography 1989, 12, 357 - 361. [21] Wheeler, J.R., McNally, M.E., J Chromafogr Sci 1989, 27, 534-539. [22] Wright, B. W., Wright, C. W., Gale, R. W., Smith, R.D., Anal Chem 1987, 59, 38-44. [23] Miles, W. W., Quimby, B. D., American Laboratory 1990, 28 J - 28K. [24] Lee, M. L, Novotony, M., Bartle, K. D., Analytical Chemistry of Polycyclic Aromatic Compounds, New York: Academic Press, 1981.
54
2 Current Status of Supercritical Fluid Extraction in Environtnental Anulysis Langenfeld, J. J., Hawthorne, S.B., Miller, D. J., Pawliszyn, J., Anal Chern 1993, 65 (4), 338-344. Langenfeld, J. J., Hawthorne, S.B., Miller, D. J., Tehrani, J., Anal Chein 1992, 64, 2263. Wenclawiak, B., Rathmann, C., Teuber, A., Fres J Anal Chem 1992, 344, 491-500. Meyer, A., Kleiblhmer, W., Cammann, K., J High Resolution Chrotnatrogr 1993, 16, 491 -494. Dankers, J., Groevenboom, M., Scholtis, L. H.A., Van der Heiden, C., J Chrornutogr 1993, 641, 357 - 362. Myer, L. J. D., Damian, J. H., Leischeski, P. B., Tehrani, J., in: Supercritical Fluid Technology - Theoretical and Applied Approaches to Analytical Chemistry (ACS Sytnposiuni S e r i a No. 488): Bright, F. V., McNally, M. I. P. (eds.) Washington DC: American Chemical Society, 1992; VOI. 16, pp. 221-236. Hawthorne, S.B., Miller, D. J., Anal Chem 1987, 57, 1705. David, F., Verschuere, M., Sandar, P., Fres J Anal Chem 1992, 344, 479-485. Hills, J. W., Hill, H.H., Maeda, T., Anal Chem 1991, 63, 2153. Hawthorne, S.B., Miller D. J., Nivens, D.E., White, D.C., Anal Chem 1992, 64, 405. Hills, J. W., Hill, H. H., J Chromatogr Sci 1993, 31, 6- 12. Ho, J. S., Tang, P. H., J Chromatogr Sci 1992, 30, 344-350. Lesnick, B. (USEPA - Office of Solid Waste), EPA-SFF Work Study Group Mceting, 1992. Lopez-Avila, V., Benedicto, J., Dodhiwala, N. S., Young, R., Beckert, W. F., J Chrornatogr Sci 1992, 30, 335-343. Hawthorne, S. B., Miller, D. J., Hegvik, K. M., J Chromatogr Sci 1993, 31, 26-30. Bicking, M.K. L., Hayes, T. G., Kiley, J. C., Deming, S. N., J Chrotnatogr Sci 1993, 31, 170- 176. Gere, D. R., Knipe, C. R., Castell, P., Hedrick, J., Randall Frank, L, G., Schulenberg-Schell, H., Schuster, R., Doherty, J Chrotnatogr Sci 1993, 31, 246-258. Levy, J. M., Ravey, R. M., unpublished results, 1994. Burford, M.D., Hawthorne, S.B., Miller, D. J., J Chromatogr 1993, 657, 41 3-427. Levy, J. M., Ravey, R. M., unpublished results, 1994. Howard, A.L., Taylor, L.T., J Chromatogr Sci 1992, 30, 374-382 Synder, J.L., Grob, R.L., McNally, M.E., Oostdyk, T.S., J Chromatogr Sci 1993, 31, 183 - 191. Synder, J. L., Grob, R. L, McNally, M.E., Oostdyk, T. S., Anal Chen7 1992, 64, 1940- 1946. Lopez-Avila, V., Charan, C., Van Emon, J., Environmental Lab 1993, Aug/Sept. 30-34. Lopez-Avila, V., Dodhiwala, N. S., Beckert, W. F., J Agric Food Chem 1993, 41, 2038 2044. Onuska, F. I., Terry, K.A., Wilkson, R. J., JHigh Resolution Chromatogr 1993, 16, 407-41 2. Onuska, F. I., Terry, K. A,, J High Resolution Chromatogr 1989, 12, 357 - 361. Onuska, F. I., Terry, K.A., J High Resolution Chromatogr 1991, 14, 829-834. -
3
Validation and Quality Control with Atomic Absorption Spectrometry for Environmental Monitoring Ian L. Shuttler
3.1 Introduction 3.1.1 Use of Atomic Absorption Spectrometry in Environmental Monitoring As a technique for the determination of trace elements, atomic absorption spectrometry (AAS) has, since its introduction in 1955, become a well-established technique for the assessment of trace element levels and is widely applied to the analysis of waters, foods, biological, industrial and other environmental samples. It is interesting to note that the first publication concerning an application of AAS described the use of the technique for the determination of magnesium in plant material, soil extracts, drainage water, blood serum and milk [ 11. This clearly indicates, that from the very beginning, and during the intervening 34 years, AAS has been significant in advancing our knowledges of the r6les trace elements play in the environment and their assessment. Trace elements at various concentrations are known to have significant effects both beneficial and detrimental in many biological systems. Their occurrence in air, water and soil derives from both natural sources and human activity. Their presence and concentration in the environment has been, and appears will remain so for the forseeable future, a subject of considerable concern. The levels of trace elements found in the environment are important, since they are non-biodegradable and are retained in the ecosystem indefinitely. There are approximately 30 naturally occurring elements which can be considered as trace elements. All trace elements are potentially toxic. For essential trace elements there is a balance between deficiency and excess which ensures the continued health of the organism. Similarly, the presence of toxic elements can have a deleterious effect on health. It is important in many areas that both the normal or background levels of essential, non-essential and toxic trace elements are monitored. The possibility that contamination of air, water, foods, etc., with trace elements, arising from modern agricultural and industrial processes, may have deleterious effects on the long-term health and welfare of human populations has drawn increasing attention to the need for monitoring environmental levels of trace elements in many different types of samples. Environmental surveys are im-
56
3 Validation and Oualitv Conirof with Atomic A bsoration S~ec/t-utneit.t
portant in establishing the extent of possible environmental pollution and often rcquire rapid, high sample through-put analytical methods with low detection limits and long-term precision and stability.
3.1.2 The Need for Quality Control Atomic absorption spectrometry fulfils many of these criteria and countless methods have been published on the determination of trace elements in environmental samples. However, the assessment of trace element levels in the environment rests not solely on the selection of a sensitive analytical method but also on the long-term performance and stability of the selected method. Increasingly, the concepts of method validation and quality control are becoming more important in the many laboratories that are concerned with environmental monitoring. As Mesley et al. [2] point out in their review on analytical quality assurance, however, ‘Few published papers have quality assurance of analytical data as their main topic’, and this situation must be rectified. There is a need to ensure that the analytical data provided by laboratories is precise (reproducible) and accurate (free from bias), only in this way can reliable decisions based on the analytical data be made and public and political confidence in both the extent and control of trace element levels in the environment be maintained. To illustrate the importance of this point, the data in Table 3-1 should be considered. Table 3-1. Changes in the accepted reference normal concentrations for urine chromium and whole blood lead with respect to time. Date
Accepted reference normal concentrations Urine Cr (nmol L-’)
1970 1974 1981 1982
Blood Pb (Fmol L - ’ ) 0.55
138 0.43 3
It would appear that between 1974 and 1982 there was approximately a 50-fold decrease in urine Cr concentration. In contrast, whole blood Pb values remained relatively stable between 1980 and 1982, yet during this time there was a real decrease in environmental lead pollution. The fact is that since the early 1970s, internal quality control (IQC) and external quality assurance (EQA) schemes have been available to ensure the quality of whole blood Pb determinations. These have enabled methodology and contamination to be assessed and improved. However, these were not available for Cr and it is clear that many of the early publications reported grossly contaminated values. This is not an isolated example as the data presented by Versieck [3] illustrate. The comment that ‘Trace element analysis, not uniquely but to
3.1 Inrroduciioti
57
a unique degree, is afflicted by problems both at the sample collection and measurement stage. Quality control must therefore cover all phases of the analytical process’ is as true today as it was in 1984. The use of validated methods and quality control is essential if the trust in analytical data is to be maintained. One of the main responsibilities of any analytical scientist is to provide data which are reliable and can be used in decision making. If these data are to be of any use, they must be sufficiently accurate for the intended purpose. Consequently the analytical scientist and the recipient (customer/end user) of the data must have some means of ensuring whether or not these can be relied upon and the degree of variation which can be expected. Quality control is the process used to ensure that the accuracy of analytical data are acceptable and of adequate precision for the use to which they will be put.
3.1.3 The Importance of Consistent Data Many environmental surveys involve the analysis of the thousands of samples and not only is it often impossible for any individual laboratory to cope with such a workload, but neither can such a volume of samples be processed in one day! Environmental surveys often take years to complete, and involve many different laboratories. Consequently, it is important that laboratories can show that they are producing reliable data not only within the laboratory, but also between different laboratories performing the same analysis, with respect to time. The use and maintenance of quality control procedures within and between laboratories allows such stability to be achieved and performance over time improved. Taylor [4] has shown for a group of laboratories concerned with whole blood lead determinations, the use of a common protocol and materials for quality control enabled the mean inter-laboratory relative standard deviations (RSDs) at concentrations between 0 and 2.0 Fmol L-‘ to fall from >20% to < 10% over a 12-year period and for the performance to be maintained. While the importance of achieving such stability and reliability in analytical measurements is still not fully appreciated by many analytical scientists, the economic costs of making false decisions based on inappropriate data can be immense. Griepink [5] has estimated that in Germany in 1984 the losses owing to poor performance by analytical laboratories were of the order of 1 billion DM.
3.1.4 Standardized/Reference Methods or Quality Control? There have been many attempts to encourage the use of standardized methods, which are methods having a defined protocol and shown to meet certain defined requirements. However, even published standard methods cannot always be relied upon, as many such methods have not been subject to rigorous validation. In addition, it is the author’s opinion that the use and insistence on standardized methods can retard further research and progress in analytical methodology. There have been sugges-
58
3 Validation and Quality Control with Atomic Absorption Specrrotnerry
tions and attempts to standardize electrothermal atomization AAS (ETAAS) methods, but as many of these require the use of particular instrumentation, they are restrictive. As standardized methods often require considerable effort in examining the proposed method in a collaborative trial and these take time to organize, it can often happen that by the time the standardized method is agreed upon and publi5hed, it is superseded by improved technology and instrumentation. Delves [6] has shown how the use of an agreed IQC protocol involving the uw of common IQC samples and participation in an EQA programme can improve the inter-laboratory performance both nationally and internationally. The use of such IQC schemes seems a far more practical approach to improving laboratory performance using reference materials as common IQCs rather than standardized or reference methods. The use of such schemes allows laboratories to use whatever analytical procedures they choose, as long as the results meet the quality criteria agreed upon. This does require an initial period when the analytical performance of participating laboratories is evaluated by the use of reference materials and the analysis of collected samples should not start until satisfactory analytical performance has been established. This time period can be fairly long as discussed by Braithwaite and Girling [7], where a European Community programme to monitor blood lead in the population was delayed for 2 years, which was the time required to establish that the large number of participating laboratories could meet the rigid quality control criteria.
3.1.5 The Degree of Analytical Quality Control Every analytical measurement is subject to a degree of uncertainty, that can, at best, only be reduced to an acceptable level, and a direct relationship exists between the accuracy of an analytical measurement and the time and effort required to achieve this. In any analytical procedure it is important to establish how accurate and precise the results of an analysis must be. This has a direct influence on how tight or loose the analytical quality control needs to be to ensure reliable data. It is important to bear in mind the fact that the indiscriminate pursuit of the ultimate in accuracy and precision is a waste of time and effort if: (1) it is not needed and (2) if the selected method/technology is not capable of achieving it. As an example of these requirements, Delves [8] has discussed the proposed target reproducibilities, with respect to blood lead measurements, for different applications: (a) detection of lead poisoning, concentration range 2.0- 5.0 pmol L-I, reproducibility desired k 10%; (b) industrial hygiene control, concentration range 2.0- 5.0 pmol L - ’, reproducibility desired k 10%; (c) environmental pollution studies, concentration range 0.4 -2.0 pmol L - I , reproducibility desired f5%. Clearly, for all of these applications accurate results are required but the need for tighter quality control in the lower concentration range is necessary to ensure reliable environmental monitoring. It is interesting to note that 15 years later there is still the
3.1 Introduction
59
need for other authors to emphasize this point [9]. Conversely the determination of normal levels of Cr in serum is an extremely difficult determination owing to the very low concentration present [lo], of the order of 0.1 -0.2 pg L - ’ . This is close to the 0.1 pg L - ’ detection limit of ETAAS. Chappuis et al. [I I ] using ETAAS, found a Cr concentration of 0.09 pg L-’ with a precision of 25%, for a second-generatjon Serum reference material. This is a result that is in perfect agreement with the statistical relationship between precision and concentration as the detection limit is reached and what many workers in this field would consider an excellent achievement. Yet, the authors argue that the method has poor reproducibility owing to the poor ( ! ) sensitivity of ETAAS. It is clearly important to realise the limitations of AAS as an analytical technique and to strive to achieve levels of performance that are not only acceptable but also theoretically and statistically attainable. It is ridiculous to expect any more of AAS instrumentation when it is at the limits of it’s theoretical performance. If better results are required then an alternative method with improved detection power should be selected.
3.1.6 Quality Control Principles The basic principles of quality control have been discussed in many books and papers [12-141 and it is not the purpose of this chapter to develop all these ideas and the underlying theory in detail. This chapter will discuss those techniques that the author has found to have been of use in routine applications and are easily applied on a day-to-day basis in the busy routine laboratory. Simple and effective techniques are to be preferred as these will be readily applied by the people actually carrying out the analysis, at the time the analysis is performed, rather than applied post-analysis when it could be too late. It is generally considered that the more important features of an analytical quality assurance programme include: 1. the use of validated methods; 2. properly maintained and calibrated equipment; 3. the use of reference materials to calibrate methods; 4. effective IQC; 5 . participation in inter-laboratory check sample schemes; 6. independent audits of quality control procedures; 7. external assessment by accrediation or other compliance schemes; 8. properly trained staff. The performance of an analytical laboratory is directly related to the use of a validated method and it should be understood that a ‘validated’ method is here taken to mean that the performance characteristics of the method, eg, precision and detection limit, have been established, shown to meet certain specific requirements, and are being maintained. It is clear that in many cases the important r6le that the analytical method plays in the quality performance of analytical laboratories has not been fully appreciated.
60
3 Validation and Quality Control with Atomic Absorption Specrrornerry
There is considerable confusion within the literature and scientists working in this field as to the meaning of the terms ‘quality control’ and ‘quality assurance’. Both terms are often widely used to mean the same thing. In the excellent book by Taylor [I21 (which is highly recommended to those interested in exploring this field furlher) the two terms are defined concisely as ‘Quality control is the mechanism established to control errors, while quality assessment is the mechanism to verify that the jyrtem is operating within acceptable limits’. Similarly, the Analytical Methods Committee of the Royal Society of Chemistry [I 51 defines quality control as those ‘practical activities’ used to control errors and quality assessment as ‘the objective testing of the performance of a laboratory by an external agency’. Within the context of these definitions it is important to realise that IQC procedures fall within the former category and EQA schemes are associated with the latter. The techniques discussed, are in principle, no different to those that can be used with any analytical technique or method, but are here considered with particular reference to AAS. In those places where it is appropriate, reference or attention will be drawn to which of these techniques can be found implemented in the current (mid 1993) versions of AAS instrument manufactures software. In addition, readers should not be dismayed that many of the examples and references quoted derive from the clinical field. This area has been active for a longer period with respect to method validation and quality control with AAS than more general environmental fields. However, the principles remain the same and it is to be hoped that the high standards and examples set by the clinical community will eventually be more widely adopted.
3.2 Method Validation 3.2.1 Basic Analytical Principles Before any routine quality control procedures can be defined, a suitable analytical method has to be selected for the required determination and the performance characteristics established. Table 3-2 shows some of the performance characteristics that are required of a routine, validated analytical method. From these performance characteristics, the selection of the acceptable limits of error and required test procedures can be established. Once a method has been established the routine validation of the various parameters such as calibration, precision and recovery measurements can be applied. These ensure on a day-to-day basis that major errors are controlled, such as calibration drift and poor precision.
3.2 Method Vulidution
61
Table 3-2. Typical performance characteristics of a validated analytical method. 1. Established during method development
and initial analytical validation
2. Monitored during routine use to allow assessment and control of analytical accuracy and precision
Assessment of contamination - Recovery measurements - Estimation of detection limit - Comparison with altcrnativc techniques/ methods Exchange of samples with other experienced laboratories - Assessment of within-batch precision - Assessment of between-batch precision - Analysis of certified reference materials - Regular assessment of contamination - Internal quality control, IQC - Continuous assessment of between-batch precision Participation in external quality asSewnent schemes, EQA, (if available) -
3.2.1.1 Preparation of Calibration/Standard Solutions In practical everyday use, AAS is not an absolute technique, and there is a need to calibrate the instrumentation. This is performed by preparing a range of calibration solutions, covering the analytical range required, containing known amounts of the analyte and taking each solution through the AAS analysis procedure with a well-defined protocol. Calibration is one of the most important steps in AAS and acceptable precision and accuracy can only be obtained when a good calibration procedure is used. However, the validity of the calibration relies upon the solutions used in its preparation. The question should be asked: how good are the calibration solutions? While modern AAS instruments will perform many calibration procedures automatically, ie, ETAAS autosamplers can prepare a complete calibration graph from one stock calibration solution, what are the influences on the results from the preparation of the stock calibration solution that the instrument uses? It is clear from the author’s experience, that in the last 10 years the users of AAS seem to have forgotten the care and attention required to prepare accurate calibration solutions. It is generally considered that good volumetric practice can achieve RSDs of about 0.1%, however, the increasing use of automatic micro-pipettes seems to be encouraging a total belief in these expensive devices and few laboratories seem to be aware that these devices do also need to be calibrated. A calibration can only be as good as the calibration standards that are used to produce it. The example in Table 3-3, illustrates this point. It is now common practice for many laboratories to make dilutions from commercial stock standards by taking an aliquot of 100 pL and diluting this to 100 mL. While this is a quick and easy way to arrive at a reasonable solution concentration (I000 pg L-’)for further dilution for ETAAS, Table 3-3 shows that on further diluThough tion the final concentration could be anywhere between 49 and 51 pg L-’.
62
3 Validation and Quality Control with Atomic Absorption Spectroinrtry
Table 3-3. Possible sources of error in the preparation of calibration standard 5olutionc (a) Require a solution with a nominal concentration of 100 pg L - I , prepared from dilution 01' a 100 pL aliquot of an original stock solution (1000 pg mL-' ) to 100 mL. Pipette error Flask error Pipette volume taken (pL) Pipette tolerance (70) Analyte mass in pipette volume (pg) Nominal flask volume (mL) Flask tolerance (mL) Actual flask volume (mL)
k0.15 100.15
Actual final concentration (pg L - ')
988.5
- ve
+ ve 100 1 99
- ve
+ vc
- ve
+ ve
99
I01
99.85
100.15
+ LC --\c
101
100
991.5
1008.5
99.85
101I . j
(b) From solutions above, require to prepare a solution with a nominal concentration of 50 pg L - ' from a 1000 p L aliquot diluted to 20 mL.
Pipette error Flask error Pipette volume taken (pL) Pipette tolerance (Dio) Analyte mass in pipette volume (pg) Nominal flask volume (mL) Flask tolerance (mL) Actual flask volume (mL)
I000 1 0.979 20 k 0.06 20.012
Actual final concentration (pg L - ' )
48.9
- ve
+ ve
+ ve + ve
+ ve
- ve
0.982
1.008
1.012
19.988
20.0 I 2
19.988
49.1
50.4
50.6
- ve
-be
Grade B flasks used. Tolerance values taken from DIN 12664 standard. Pipette tolerances as found from experience during routine laboratory pipette calibration procedures.
this error of t 1 wg L-' may not seem extreme it does indicate that a source of inaccuracy can be introduced before any instrumental measurements are made. What has not been taken into consideration in this example is the error in the actual stock solution. Careful reading of the manufactures labels will show that these solutions also have an error associated with them. As few laboratories seem to consider that Grade 'A' glassware is necessary for AAS and that in many cases washing machines (operating at high water temperatures) are used to clean glass/plastic ware, these errors can easily be higher. Most calibration routines assume that there is no error associated with the x-axis values, consequently it is foolish to hope for an accuracy of +_ 1 vg L-' in the final results when, without proper precautions, the preparation of the calibration solutions is not even that accurate.
3.2 Method Validation
63
The moral of this discussion is that: I . top quality glass/plastic ware should be used and the same flasks reserved for a
particular determination to reduce the risk of random error; 2. pipettes should be regularly calibrated and the results documented; 3. to avoid these errors solutions can be prepred on a mass basis by weighing, and 4. sources of error can be introduced into AAS measurements before the instrument has even been switched on! For further elaboration on this topic, Zehr and Maryott 1161 have compared the accuracy and precision of different dilution procedures for AAS measurements.
3.2.1.2 Use of Characteristic Concentration/Mass
In any method validation/quality control protocol it is important that the instrumental peformance is also monitored and that the performance is in line with what the manufacturer has specified. Nearly all AA instrument manufacturers supply tables and handbooks (often referred to as ‘cookbooks’) specifying the instruments performance characteristics. These are generally referred to as characteristic concentration (in the case of flame AAS) or characteristic mass (for ETAAS) and are the concentration or mass of analyte that will give a signal of 0.0044A, or QA, where A , is the peak absorbance and QA the integrated absorbance. The characteristic mass, mo for ETAAS is calculated from the following equation: 0.0044 rn, = _ _ x m Aint
where Aintis the blank corrected integrated absorbance signal and m the mass of analyte in pg injected into the atomizer. The characteristic concentration and/or mass are important performance characteristics indicating the instrument’s sensitivity and should be monitored on a routine basis with a calibration standard, or sample with an accurately known concentration giving a reasonable signal in the linear response range of the instrument. The characteristic concentration/mass provide an indication that the instrument is performing as expected and has been correctly optimized. Even so, day-to-day variations will be found in these values and -t20% is considered acceptable, especially for ETAAS work. Slavin et al. [I71 have discussed in greater detail the use of characteristic masses for instrument QC. However, carefully validated and controlled methods can produce very stable data. Shuttler [I81 found the characteristic mass for the determination of lead in blood to be 14- 17 pg per 0.0044 QA over a three year period, and a round-robin study of 16 laboratories to assess a new platform design for ETAAS found remarkably stable characteristic mass values [ 191. Gross variations in the characteristic concentration or mass should be investigated and could be an indication of a poorly prepared standard, incorrect experimentalhnstrument parameters or a malfunctioning instrument.
64
3 Validation and Quality Control wirh Atomic Absorprion S~~ectrotiielry
3.2.2 Calibration 3.2.2.1 The Importance of Calibration The role of calibration with respect to AAS and its effects on the overall accuracy and precision of the analytical procedure are often overlooked. Any calibration procedure affects the overall analysis in terms of speed, accuracy and precision. Thc number of calibration points, the recalibration frequency, how well matched the samples and standards are and how closely the fitted curve matches the true curve are all factors that will affect the accuracy. With any validated method, the calibration procedure must be defined and strictly adhered to. In general, an AAS instrument is calibrated by taking a series of standard reference (ie, calibration) solutions, each containing an accurately known level of analyte and measuring the instrument response of each solution using exactly the same protocol. The instrument response for each standard solution is plotted on the y-axis of the calibration graph against the concentration on the x-axis. An appropriate straight line or curve is then drawn through these points and the resulting calibration curve is then used to determine analyte levels in the unknown samples, using exactly the same procedure as applied to the standard solutions. There are a number of important points to be considered in preparing and displaying calibration graphs many of which have been covered in a series of articles by Miller [20-231. It is essential that the calibration standards cover the whole range of concentrations required in the subsequent analyses as the concentrations of samples are determined by interpolation and not by extrapolation. (The special case of extrapolation with the method of analyte additions will be discussed later.) This author recommends a minimum of five calibration standards and a blank for the preparation of a calibration curve. This provides sufficient information for some statistical evaluation of the data and enough points for most software algorithms to produce an acceptable mathematical fit if the calibration is non-linear. The stability of the calibration procedure has to be considered. If the instrument response is, to a large extent, independent of small changes in operating parameters then the calibration does not need to be performed frequently. On the other hand, if the instrument response is highly dependent upon operating conditions, frequent calibration is necessary. A flame AAS instrument can require frequent re-calibration during the working day if variables such as the stability of the hollow cathode lamp, fuel to oxidant ratio, wavelength stability and nebulizer and burner chamber efficiency change. An ETAAS instrument may also need re-calibration as the graphite tube surface ages and the analytical signals change, especially with peak height measurements. While modern AAS instruments are very reliable and stable, the stability of the calibration for a particular determination can only be established by experience and experiment. The calibration strategies for FAAS have been discussed by Tyson [24] following a survey of members of the Atomic Spectroscopy Group of the Royal Society of Chemistry, Analytical Division [25]. It would appear that the most frequently used procedure is the use of five calibration standards including a blank, with 50% of the users applying a computer-based method of curve-fitting, of which the majority were
65
3.2 Method Validation
the instruments built-in computer algorithm. It is highly likely that since the publication of this survey, the proportion of users applying the built-in functions has increased, as virtually every modern AAS instrument provides this capability. A number of authors have compared the curve-fitting algorithms used by the major manufacturers [26-291 and concluded that in general, all provide acceptable results. However, there will always be errors involved no matter what algorithm is used, although, it is important to realise that these errors will be worse, the fewer the calibration points that are used. Even though calibration is often considered a non-productive part of an instruments operating period and there is a tendency to reduce the number of calibration points to a minimum, this trend will have a direct influence on the quality of results produced.
3.2.2.2 Influence of the Blank The importance of the blank signal, hence contamination, and the influence this has on calibration is often overlooked in trace element analyses by AAS. It is vitally important that the degree and magnitude of contamination to which the procedure is affected is assessed. The auto-zero function on many modern AAS instruments appears to be a very much abused feature and is primarily designed to set the instrument null point, not to deduct large blank signals from all measurements. The data in Table 3-4 illustrate what can happen if the purpose of this function is neither fully understood nor used properly. The mean blank value will be automatically deducted
Table 3-4. Influence of blank signal magnitude on sample signal SD and precision.
Measurement
Mean signal
SD (%RSD)
Blank Sample Corrected signal
0.010 0.020 0.010
0.0010 (10%) 0.0016 (8%)
Blank Sample signal Corrected signal
0.100 0.120 0.020
0.0020 (2070) 0.0024 (2070)
Example 3
Blank signal Sample signal Corrected signal
0.005 0.020 0.015
0.0010 (20%) 0.0016 (8070)
Example 4
Blank signal Sample signal Corrected signal
0.005 0.050 0.045
0.0006 (12%) 0.0010 (2010)
Blank signal Sample signal Corrected signal
0.002 0.079 0.077
0.0003 (15%) 0.0003 (0.4%)
Example 1
Example 2
Example 5
Actual (ie, real) SD
Actual VoRSD
0.0019
18.9
0.0031
15.6
0.001 9
12.6
0.0012
2.6
0.0004
0.6
ETAAS measurements performed on Perkin-Elmer Model 41 00 ZL instrument.
66
3 Validation and Quality Control with Aromic Absorprion Speciroinciry
from the individual sample values before the sample statistics are calculated. This assumes that the blank value has no uncertainty. This is plainly untrue and the final SD should be calculated from the propagation of errors equation: SD = J(k,s,)2+(k2s2)2+ . . . (k,s,)*
(2)
where s, is the standard deviation of the ith result and k a constant. The data in Table 3-4 show that the actual SD will be larger than that calculated in situations where the blank and sample solutions have similar magnitudes. Deducting large blank values to measure small sample signals will lead to large errors in the determinations as examples 1 and 2 in Table 3-4 show. Only in those situations (examples 4 and 5 ) where the blank signal is much smaller than the sample signal does the allowance for the blank correction introduce only a small error. Consequently the blank signal should be as small as possible to avoid increasing the uncertainty in the net sample signal value. It should not be forgotten that these values are net signal responses and do not include any errors introduced from the calibration. To allow a statistically sound interpretation of the calibration curve it is crucially important to include the value for a ‘blank’ sample in the calibration curve. The blank contains no deliberately added analyte, but does contain the same solvent, reagents, etc., as the samples and is subject to exactly the same analytical procedure. The absorbance signal for the ‘blank’ will often not be zero, and it is subject to errors just as all the other points on the calibration curve are. Therefore, in principle, i t is wrong to subtract the blank value from the other standard values before preparing the calibration curve. However, this is what virtually all commercially available AAS software packages and instruments do. In the case where the ‘blank’ value is almost the same or near to the level of the instrument zero point then little error will be introduced, but larger blanks can cause problems. With the way that modern AAS instruments are used it is critical that the blank value used for subtraction is the correct value. This once again re-emphasizes the need for continuous assessment of contamination and an awareness of what a ‘normal’ blank value should be. If an incorrect ‘blank’ value is used then a systematic bias will be introduced into the calibration. This type of bias can be assessed with those software packages which allow a calibration procedure with a non-zero intercept to be used. 3.2.2.3 Type of Calibration Curve Once a series of calibration solutions have been run through the analysis procedure and the signals collected, it is necessary to select the correct means of preparing the calibration curve. This is an important question as the choice could heavily influence the subsequent sample results. The selection is not made any easier when one considers that many modern AAS instruments allow the selection of a number of possible options which generally fit a mathematical model to the experimental data. These can include: non-linear, linear, method of analyte additions to each sample, method of analyte additions as calibration, bracketing calibration, etc., in addition to which the user could be presented with the possibility to use these calibration
3.2 Method Validation
67
options with or without the curves forced through the calibration axes origin. The choice is often overwhelming and the selection seems in many cases to be decided by trial and error. In some situations, regulatory authorities can require or stipulate a certain type of calibration procedure in which case the room for manoeuvre is limited.
3.2.2.4 Linear or Non-Linear-Calibration The most convenient calibration function is linear, goes through the origin and is applicable over a wide dynamic range. In AAS, to a first approximation, the absorbance is directly proportional to the total number of atoms per unit volume. It is clear that the number density of atoms in the light path is a complex function of atomizer conditions and sample matrix composition. In general, the calibration curves are linear at low absorbance values but exhibit an increased curvature towards the concentration axis at higher concentrations. This non-linearity can be due to differences in the source and absorption line profiles, stray light, emission from the atomizers, instrumental parameters, ie., different slits, and interferences from the sample matrix. In practice, the ideal situation of a linear calibration curve may not be found. With PC controlled instruments a visual display is often provided of the calibration curve and as the calibration is of such vital importance to the eventual results, time and care should be invested in its examination. Modern software generally allows the user to perform an interactive examination of the calibration by editing calibration points, re-analyzing selected standards and changing to a different calibration algorithm, etc. Care should be taken in using these features as deleting four points from a five point calibration curve will produce a perfect calibration line but may not produce correspondingly good analytical data! Examining the calibration curve allows a decision to be made on whether the calibration is linear or non-linear, but it does not allow an assessment of which is the best straight line or curve through the points and the errors and confidence limits for the slope and intercept of the line. The choice of whether to use a linear or nonlinear calibration curve is generally up to the individual analyst. There is, historically, a reluctance to use non-linear calibration curves but a number of authors [27, 30, 311 have shown that there is no reason to limit the analytical dynamic range to the linear portion of the calibration curve. What determines the analytical dynamic range is the range where acceptable precision is obtained.
Correlation coefficients. The correlation coefficient, I-, is often used to test the linearity of a calibration graph, though too much importance is frequently attached to this statistic without a proper understanding of how it should be interpreted. A high value of Y is no guarantee that a straight line, rather than a curve, is appropriate for a particular calibration graph and for this reason the calibration graph should always be examined visually. Several publications [21, 321 have illustrated extremely well the difficulty of interpreting Y and its statistical significance and these arguments should be considered before undue weight is attached to the magnitude of the correlation coefficient in any method validation scheme.
68
3 Vulidation and Quality Control with Atomic Absorption Spectrometry
3.2.2.5 Calibration by the Method of Analyte Additions The use of a calibration curve constructed from aqueous based calibration standards is the simplest and quickest calibration technique, though it implies that samples and calibration standards have an identical analytical behavior in all stages of the detcrmination. With a few exceptions, flame AAS shows this type of behavior. In the early days of ETAAS it became clear that there were severe interferences and that the above statement did not apply. Consequently the method of analyte additions (‘standard additions’) was widely employed in an attempt to compensate for these interferences by ‘matrix matching’ using the sample itself. The method of analyte additions is an extrapolation calibration technique whereby the calibration standards are added to the sample itself and the x-axis of the calibration graph represents the level of added analyte. The initial point of the calibration (x= 0) is the signal produced by the sample with no analyte added. Extrapolating the calibration line through the points to where y = 0 gives the concentration of analyte in the original sample. Subsequent work over the last decade has lead to the introduction of the stabilized temperature platform furnace (STPF) concept [33] and Zeeman-effect background correction to overcome these problems. There is now a vast body of literature available which demonstrates that these techniques overcome many of the interferences and that accurate and precise sample measurements in complex matrices can be achieved against calibrations established with simple aqueous calibration standards. However, there appears to be considerable persistence and blind acceptance to the use of the method of analyte additions as a calibration technique. The use of the method of analyte additions as a method of quantification has significant disadvantages [34- 371. The method is time consuming: requiring a separate calibration to be established in each sample, prone to poor precision as the errors in both the slope of the calibration line and the intercept have to be included in the concentration result, and it is essential that the extrapolated calibration line is linear. To minimize these errors, at least five to six calibration points should be measured and the additions to the sample should, if possible, cover a wide range and produce a similar or greater signal to that present in the original sample. Hence for the optimum application of the method of analyte additions the approximate concentration of the analyte in the sample needs to be known so that additions with the correct concentration can be used to minimize errors in the regression and subsequent extrapolation to the concentration axis. A number of these points and typical method of additions errors are illustrated in Fig. 3-1.
3.2.2.6 Calibration Quality Coefficients The assessment and control of calibrations is a topic which continues to generate a tremendous amount of literature and discussion. The increasing use of automation in AAS instruments means that many systems run overnight, prepare calibrations automatically and often determine a number of elements sequentially. In an effort to detect ‘good’ or ‘bad’ calibrations the use of a calibration quality coefficient has
3.2 Method Validation
69
A
/
Large errors due to poor choice of additions and large extrapolation
B
T
Absorbance signal
1
Concentration
Minimized errors in extrapolation
Fig. 3-1. Typical problems associated with the method of analyte additions. (4)Illustrates the problems associated with insufficient additions, too small a difference between the additions and a large extrapolation. This produces increased errors and uncertainty in the final result. (B) Indicates how errors with the method of analyte additions can be minimized by performing sufficient additions of the correct magnitude to ensure reasonable confidence in the extrapolated line.
been proposed by a number of authors [38-401. The calibration quality coefficient (CQC) [41] is defined as:
where y i is the measurement at each data point, pi is the value predicted by the least squares line and N is the number of points not including zero. (The term CQC is used here as opposed to the term QC proposed in the original papers to avoid confusion with the overall use of QC in this chapter to mean quality control.)
70
3 Validation and Quality Control with Atoniic Absorption Spectrometry
While the use of such a coefficient is not as yet available in any commercially supplied software package for an AAS instrument, it can only be a matter of time. The use of such a CQC allows the instrument to decide automatically between a ‘good’ and ‘bad’ calibration and choose an alternative calibration method such as a robust regression method [39]. However, while these techniques and the subsequent iterations that they require [40], will undoubtedly improve the calibration, they do require a considerable increase in the number of calculations required. In addition, it should not be overlooked that these techniques merely assess the calibration. A series of poorly prepared calibration solutions can still give acceptable analytical sensitivity, perfect calibration and calibration quality coefficients yet produce inaccurate results. The final independent check on the calibration must be the analysis of quality control materials.
3.2.3 Establishment of Performance Characteristics 3.2.3.1 Assessment and Influence of Contamination The influence of the blank on the analytical results has been discussed above. In environmental analysis it is important to consider that the blank used for the samples could be totally different to that used for the establishment of the calibration. Some current AAS software packages allow such flexibility. The arguments above (Section 3.2.2.2) also hold for the sample or reagent blank, the value should be as low as possible. In many cases a sample preparation or reagent blank is unavoidable and for this reason, as part of the validation process, regular checks should be made on the blanks and the information documented. During every analysis at least one and preferably more sample blanks should be prepared and run through the whole procedure. The vessels and laboratory ware used to collect and prepare samples should be regularly assessed. It may well be that a contamination contribution cannot be avoided, however, the level of this contamination is important. Shuttler [18] showed that the contamination due to the sampling procedure for the determination of whole blood manganese was of the order of 8 nmol L-’. As the normal reference range for whole blood manganese [42] is of the order of 73 -210 nmol L-’ and patients receiving supplementation to their TPN fluids can have levels varying from 100- 600 nmol L - ’ then a possible contamination of 8 nmol L - ’ is not serious when considered against these levels. While with the chosen sampling technique, the ‘true’ value will possibly not be reported, the result obtained will be sufficiently accurate for clinical purposes and enable either a deficiency or excess of manganese from that considered normal to be easily distinguished. However, regular checks are necessary to ensure that the degree of contamination remains constant or else an undetected bias can be introduced into the results over a period of time time. For a more comprehensive discussion of contamination and the means to control it, the book by Mitchell and Zief [43] is highly recommended.
3.2 Method Validation
71
3.2.3.2 Estimation of Detection Limits The detection limit (DL) is another relevant performance characteristic of a validated analytical method and again it is a parameter which should be routinely measured. A knowledge of the DL enables the lower limits of the analytical measurements to be established and whether the method is in fact applicable to the required determination and level. The definition, calculation and implications of DLs have been widely discussed in the literature [44-471 and like the subject of calibration is one that is open to widely differing opinions. This author recommends the acceptance of the International Union of Pure and Applied Chemistry (IUPAC) recommendations [48] along with the proviso that for any quoted DL the method of determination and calculation should be stated. This would help to alleviate the considerable confusion within the literature and the widely differing values often found for identical methods and instrumentation. In addition, an appreciation of the statistics concerned with calculating a DL and interpreting the value would help to reduce the over-emphasis on DLs and their overall importance. For this reason the following discussion is perhaps longer than necessary to illustrate some of the problems associated with determining DLs. The detection limit is generally accepted as ‘that concentration which gives an instrument signal significantly different from the field blank’. Despite the relative simplicity of this sentence there is some confusion regarding its interpretation. This stems from the fact that the definition poses two questions, namely: what is signifcantly different and what is a field blank? IUPAC has recommended [48] that the limit of detection, defined in terms of either concentration (c), or amount (qL), be related to the smallest measure of response (xL)that can be detected with reasonable certainty in a given analytical procedure, where: X L = X,
+k S B
(4)
and X , is the mean of the blank measurements, S, is the standard deviation of the blank measurements and k is a numerical constant. The detection limit is given by:
where b is the sensitivity of the method, this being given by the slope of the calibration graph, if the graph is linear throughout. The constant k is at the discretion of the analyst, though IUPAC ‘strongly recommends’ a value of k = 3 and in the main, this value has generally been accepted by the analytical community. For a greater discussion of the statistical significance of the value chosen for k , Refs. [42-451 are recommended. From the point of view of this chapter, the important point to consider is how are detection limits estimated? There are two important parameters which have to be assessed, xL and s , . The blank signal x, can be obtained as the average of several ab-
12
3 Validation and Quality Control with Atomic A bsorplion Spectronietty
____
sorbance readings of a ‘field’ blank. A ‘field’ blank is defined as a sample containing the diluent, reagents and sample matrix, examined under identical conditions to thc samples but containing no analyte. Unfortunately, a ‘field’ blank is often unobtainable, especially in the trace element environmental field where many samples may contain endogenous (often very low) levels of the analyte being determined. The instrumental detection limits (IDL) can be estimated from reagent blanks and will give an indication of the instrumental performance, but have little in common with real analytical situations. The simple estimation of an IDL for AAS can be performed in the following manner. At least ten replicate measurements are made of the blank solution, using the same instrumental and experimental conditions that are used for the analysis of samples. In the case of ETAAS this is normally a solution of 0.2% v/v H N 0 3 , using a sample volume of 20 pL, pipetted from ten separate sample cups. A standard solution which lies within the linear range is also measured to enable the characteristic concentration or mass to be calculated and the IDL to be converted from an instrument response into mass or concentration. All the blank absorbance readings are included in the calculations, including negative readings and the IDL calculated from the following formula: DL
=
3xsBxmo 0.0044
It is important that the limitations of IDLs are understood. Instrument detection limits can be artificially low for the following reasons: they do not include the additional variance from subtraction of the reagent blank; the solution is matrix free; the blank solution is not a ‘field’ blank, ie, it does not bear any resemblance to a ‘real’ sample; and the measurement is made over a short time scale, which minimizes the variation owing to any possible baseline drift. In practice, IDLs should be estimated more than once, and over a period of time to allow variations in the instrument performance to be included. From a consideration of the statistics involved with estimating detection limits, readers should not be surprised to find a large variation in detection limits! Another important point is ensuring that the blank solution or reagent blank contains no analyte. It is patently false to estimate a detection limit with a blank which gives a reasonable analyte signal. This will give a much lower standard deviation than the base line noise signals and hence a low instrument detection limit from Eqn. ( 6 ) above. In the absence of a ‘field’ blank, for instance with the analysis of waste water or urine for chromium, it is almost impossible to find a sample which contains no chromium. In this case, the detection limit for the purposes of validating the method can be estimated from the intercept of the calibration graph, a, which can be used as a measure of x,, and sy,x will provide an estimate of S, from the statistics calculated from a number of linear calibration graphs prepared over low concentrations [49]. An alternative approach is to analyze a number of real samples with low concentrations and estimate the standard deviation at zero analyte concentration by extrapolating the regression line of concentration of analyte versus standard deviation to the y-axis.
3.2 Method Validation
73
To estimate detection limits a considerable amount of time and effort can often be invested. However, the detection limit is best regarded as a useful, but only approximate, guide to instrument performance and the investment of great amounts of time in an attempt to estimate it precisely is rather a waste and in terms of quality control the time could be better spent on more important issues. For a practical, in the main, non-statistical approach to the discussion of detection limits, the article by Thompson can be recommended 1471.
3.2.3.3 Recovery Measurements In many instances recovery measurements are made to improve confidence in an analytical procedure. Recovery or ‘spike’ measurements are made by measuring the sample then a second aliquot of the sample again, to which has been added a known amount of the analyte. The recovery is generally expressed as a percentage from the following equation: %Recovery
=
(sample + addition) - sample x 100 concentration added
(7)
There appears to be a general misconception that recovery measurements can be used to assess the accuracy of an analytical procedure. Recovery measurements are again dependent upon the calibration and in many instances (especially in ETAAS) the same standard(s) used in establishing the calibration curve are used for the recovery measurements. To be of any value, recovery measurements should be made on samples with similar analyte concentrations and the additions should be of a similar concentration to that present in the sample. This presupposes that the sample plus addition concentration lies within the range of the calibration curve. Even if an error has been made in the preparation of the calibration curve, recovery measurements will not detect that error. Recovery measurements enable an assessment to be made of whether the method is free from interferences or the degree thereof, but give no indication of analytical accuracy. They are useful to enable the detection of samples which may have an unusual matrix interference and require further examination, perhaps with an alternative method.
3.2.3.4 Precision It is important for any validated analytical method that an estimate of the random error present in the procedure/method is known. Random errors are identified when replicate measurements of a single analyte in the same sample are made, there will be a distribution of results either side of the mean value and it is the extent of this distribution that we wish to estimate, is it broad or narrow? Answers to these questions can be obtained from an assessment of the precision of a method. Large random errors will produce poor precision and vice versa. While the precision of mea-
74
3 Validation and Quality Control will? Atomic Absorption Spwlroriirtry
surements is easily calculated from the standard deviation, it is more common to use the alternative, percent relative standard deviation (VoRSD) defined as: %RSD =
S ~
x 100
~nlearl
where s is the standard deviation and xmeanthe arithmetical mean, when discussing the precision of analytical measurements. The %RSD is a relative error, ie, an error estimate divided by an estimate of the absolute value of the measured quantity. Thi5 is useful in comparing the precisions of results which may have different magnitudes. In considering the relationship between standard deviation and relative standard deviation it is important to realise that as concentration increases so does standard deviation yet relative standard deviation decreases. In addition, it is important that the measurements used to calculate precision values are in terms of concentration rather than absorbance signals. It has been demonstrated [27 - 301 that at high concentrations the precision in the measured absorbance signal is usually no worse in the upper, non-linear portion of the calibration graph than in the lower, linear part. However, owing to the non-linearity of the analytical curve, the precision in concentration becomes worse. In considering the precision of an analytical method there is an important distinction to be made between the repeatability (within-batch or within-run precision) and reproducibility (between-batch or between-run precision). The repeatability precition is considered to refer to measurements made on the same batch of samples during a short time interval with the same instrument and calibration curve generally, these measurements are also made by the same operator. In contrast, reproducibility measurements, here defined with respect to validating an analytical method in a laboratory as opposed to collaborative trials, refers to measurements made on the same samples, generally with the same instrument, but over a longer time scale, several days or weeks and with freshly prepared calibrations and (if possible) different operators. Both of these should be assessed and estimated. An often overlooked use of the precision of an analytical method is that this can be used to define the unalylical dynamic range.
Repeatability. The repeatability of an analytical method is assessed during validation by taking replicate measurements on the same sample and calculating the VoRSDs. While this seems a simple definition, in practice, care should be taken to define how the measurements are made to enable the correct interpretation 150, 511. There is a great temptation with AAS instruments to measure the within-batch precision from one sample cup/vessel with repeated measurements. This really only assesses the precision of the instrument and its associated sampling system. With such a measurement the random errors introduced during sample preparation are ignored and lost. It is interesting to note that with a classical volumetric procedure most analysts would prepare two separate digestions/solutions of the sample followed by titration of a minimum of three replicate aliquots from each sample solution and then calculate the mean and VoRSD from the combined results. Sadly, with modern AAS ana-
3.2 Method Validation
75
lyses it would appear that users are assuming that replicate measurements via the autosampler from a single sample digestion/aliquot allows an assessment of repeatability. It does, but only of how precise the instrument and its autosampler are! While this is important information it is not the whole story and we are really interested in the total within-batch precision incorporating the sample preparation errors which should not be ignored, but often are. Unfortunately modern AAS instruments using autosamplers are only able to perform repeat measurements and calculate the respective statistics from replicates made on the same sample cup. The calculation of real within-batch precision requires more effort on the part of the user, but the ability to transfer analytical data directly from many of the AAS instrument control software packages into proprietory spreadsheet programs means that much of the drudgery associated with this type of data manipulation can be considerably reduced. In the interpretation of repeatability there is a need to assess how many replicate measurements are required to be made. It is clear that for a fixed concentration, increasing the number of replicates will improve the precision but in general a compromise has to be made between the time available for the analysis and the precision required. Also the concentration will play a role as Fig. 3-2 shows. The automated testing of the within-batch precision is easily accomplished with several of the current AA instrument control software packages available, but their use requires a knowledge of the performance characteristics of the method such that realistic values can be used. These tests are not designed to re-analyze every sample until the very best precision is obtained (which statistically can eventually happen!) but to detect those samples which occasionally give poor precision and repeat the analysis a specified number of times to allow a realistic assessment of the precision before continuing. Typical examples of the use of this test are when one reading from a group of four replicates is an outlier hence giving a poor (ie, high) precision. The analysis would be repeated automatically. Sometimes memory or carry-over effects can cause poor precision in ETAAS analyses, again the sample analysis can be automatically repeated. The danger with any test such as this is that the sample could be poor, for example, containing undetected colloidal or solid particles, and a good precision unattainable. For this reason it is important that such a test allows the number of repeat determinations to be specified by the analyst before the system proceeds with the next sample. The second danger with such a test concerns the concentration or signal magnitude at which the required precision is measured. Figure 3-2 shows that as the detection limit of the method is approached then the precision increases towards 30% RSD or greater. Hence it is somewhat ludicrous to expect good (ie, low) precisions from samples with analyte concentrations near the detection limit. An additional test which is specified by the United States Environmental Protection Agency (US EPA) [52]is that of performing duplicate sample analyses, generally on one sample in each batch of samples of a particular type. This requires the separate analysis of two aliquots of the same sample in the same analysis run. The relative percentage difference (RPD) in the results is calculated from the following equation:
RPD =
x 100 (S+D)/2
(9)
76
3 Yalidation and Quality Control with Atomic Absorption S~ectroinerrj~
3c
s 2c
.-c
.-m
a >, U
\
-.*-. ---*--.-...__.....___I-.-~ A
1.o 2.0 3.0 Concentration of lead pmol L-’
4.0
Fig. 3-2. Within- and between-batch prccision for blood lead analyses. A Within-batch precision and B between-batch precision. (Figure taken from Ref. [ 6 5 ] and reproduced by permission of the Royal Society of Chemistry.)
where S is the first sample value (ie, original) and D the second sample value (duplicate). Once again this test has been automated in some of the commercial AA software packages, all the user has to do is designate the position of the two duplicates in the autosampler tray and the calculations are automatically performed. However, it is important to ensure that this test can be automatically performed for all AAS techniques not just for ETAAS.
Reproducibility. With respect to AAS methods in environmental analysis, the repeatability is generally established by analyzing the same samples on a number of separate occasions using exactly the same analytical procedure. In many laboratories this will involve using the same instrument but it is important that for each separate batch of analyses that the instrument is calibrated and it is often a good idea that different users perform the complete analysis procedure. The repeatability is then calculated from the mean values obtained from the separate measurements. The way that the repeatability is calculated should be defined and documented because there is a difference between the ‘pure’ between-batch precision and the ‘total’ precision
1501. For both the within- and between-batch precisions it is important that these are assessed over a wide concentration range so that an appreciation of the variation of precision with respect to concentration for the method can be established. A useful way of examining such data is by means of the Ringborn plot [26, 271, an example of which is shown in Fig. 3-2. For AAS measurements, such plots are modified from
3.2 Method Validation
77
those originally proposed by Ringbom [26] for UV-visible spectrophotometric measurements, to show precision on the ordinate and concentration on the abscissa axes. Both the within- and between-batch data should be shown on such a figure, and the reproducibility and repeatability periodically reassessed to ensure that the method is still operating within the limits set during validation. However, analysts should be aware that the precision accounts only for random error and does not indicate systematic errors, consequently an analytical result can be very precise yet yield the wrong result. In addition, while the above discussion has only considered precision with respect to analytical errors, it is important that sampling errors are also considered. Most textbooks consider this issue, but for a practical illustration, Thompson and Maguire [53] discuss how robust analysis of variance can be used to estimate the sampling and analytical errors in surveys of trace metals in soil surveys by FAAS.
Analytical Dynamic Range. Another advantage of calculating and plotting the repeatability and reproducibility data, as shown in Fig. 3-2, is that this allows an assessment of the analytical dynamic range. The analytical dynamic range is defined as the analytically useful range of the method defined in terms of the precision required over a defined range [27, 301. Van Dalen and de Galan [27] have shown that precise results can be obtained with calibration graphs extending well into the nonlinear range. An awareness of the concentration range over which a certain precision can be reliably obtained is important as this allows intelligent control limits to be set. With respect to environmental monitoring, the interest is generally concentrated at the lower end of the concentration range. Consequently, as the data in Fig. 3-2 show, if the majority of samples could be expected to fall within the range 0.5 - 1.0 vmol L-' then a precision better than 5 % is unlikely on a routine basis. If an improved precision is required then changes should be made to the method, such as decreasing the dilution and shifting the concentration range of the measurements to within the range where such precision can be expected.
3.2.3.5 Comparison with Alternative TechniquedMethods In establishing the validity of an analytical method the importance of comparing the results obtained from the same samples either analyzed in the same laboratory or in another, experienced laboratory but using a different method should not be overlooked. There are two approaches to this situation. In the first case an established method in a laboratory can be replaced owing to a change in instrumentation, for instance, a previously validated solvent extraction FAAS method can be replaced with an ETAAS instrument. It is essential that before changing to the new method that a comparison is performed to ensure that the new method meets the established performance characteristics and that the same samples are analyzed by both methods to ensure that there is no significant bias between the two methods. The second situation is where a method could have been developed and partly validated in a new laboratory and the analysts wish to assess how well the method is performing before joining an assessment scheme or starting to accept samples for analysis. Here the exchange of samples with an experienced laboratory can enable an inde-
78
3 Vulidution and QC with AAS
pendent check on the performance of the method. It has been the author’s cxpet I ence that many laboratories will provide a few aliquots/portions of pieviously analyzed routine samples to help other laboratories gain confidence in their analytical procedures. In both these situations there is a need to make reliable comparisons bel\+een the data and one of the simplest is to prepare a least squares regrecrion on the data. However, as has been discussed by several authors [54, 551 care should be taken in comparing the data given by such a regression. Similarly, statistical tests such as t h e Students t-test, to compare the means of the results, and the F-test, if the standaid deviation data are available, can be used. If a sufficient number of samples have been exchanged then a cumulative frequency distribution of the differences can be plotted. This shows both the degree of agreement and whether there is any bias between the two sets of results.
3.2.3.6 Analysis of Certified Reference Materials As part of any method validation, certified reference materials should be analyzed. In the environmental field there are many such materials available and a number of reviews and catalogues are available [56- 581. A certified reference material ( C R M ) is defined as a reference material, one or more of whose property values are certified by a technically valid procedure, accompanied by, or traceable to a certificate or other documentation which is issued by a certifying body [2]. The results obtained for the analysis of CRMs are, unlike recovery measurements, etc., independent of the methods calibration procedure and indicate the methods accuracy. Ideally the reference material should have the same or nearly the same matrix as the samples to be analyzed, which it should also match with respect to the levels of trace elements present. Unfortunately, this is often not the case and given the vast range of different sample types which may be encountered in environmental analyses with AAS, it is often a problem to find a suitable CRM. However, this should not deter the analytical scientist from having a range of certified reference materials available to examine on a regular basis. It is far better to have examined a range of CRMs, even if they do not match the intended samples exactly, eg, analysing a range of CRM foodstuffs such as dried milk powder and bovine liver even if the actual samples are biological tissues, and assessing the accuracy of the method in use rather than not examining any CRMs at all. Owing to the intense collaborative effort required to produce CRMS they are not inexpensive, even so their use is to be encouraged and it is important that they are examined on a regular basis. It is important that the CRM(s) are analyzed more than once, including a digestion procedure if that is required, and treated exactly the same as a normal sample, to enable a meaningful assessment of the method to be achieved and realistic statistics to be calculated. The analysis of a CRM once during method validation does not mean that the method is still giving acceptable results after a year in use. The analysis of CRMs is but one of the tools available in the armoury of AAS methods in environmental analysis to ensure reliable analytical data are being produced and should be performed on a regular basis.
3.3 Quality Control
79
3.3 Quality Control While all of the method validation procedures, as described above, help to ensure that the method is producing acceptable results with respect to the calibration and generate the necessary statistics to allow an assessment of the various performance characteristics of the method, these procedures provide no indication of the long term accuracy of the methods or freedom from bias. It is vitally important that an independent means of controlling and assessing the method is applied, not on an ad hoc basis but regularly. The regular analysis of IQC materials and graphical evaluation of the data (by means of a suitable control chart) as it is produced enables the method to be assessed easily over time and within the laboratory by the individuals who are actually performing the analyses [59]. The repeatability and precision of the results can be seen at a glance. For this purpose reference or quality control materials should be used. Yeoman [60] introduced the principle of using reference materials with well defined limits of acceptance for analyte concentrations, as IQC samples. Only those results for samples which are associated with concurrently analyzed IQC specimens that are within acceptable limits are accepted as valid. The use of this approach has produced excellent inter-laboratory agreement both nationally and internationally for environmental lead monitoring [4, 6, 7, 181. In the past, schemes such as that discussed by Delves [6] and others [4, 71 have had to be applied retrospectively and manually to most AA techniques, especially ETAAS. The methods of IQC discussed below allow a within-laboratory independent assessment of accuracy and bias, which in conjunction with participation in EQA schemes will enable the analyst to improve the overall quality assurance of the laboratory. In addition, the performance of the instruments and methods can be verified on a continuous basis, thereby ensuring the production of reliable analytical data.
3.3.1 Frequency of Analysis and Choice of IQC Materials It may seem somewhat perverse to discuss the frequency of analysis and choice of IQC materials together, but the two are closely related. The frequency with which IQC materials are analyzed with the samples depends upon a number of factors such as:
1. The degree of analytical control required - information regarding this will be defined by the requirements of the analysis and the validation process and indicate whether IQC control is required every 5, 10 or 20 samples, etc. 2. Regulatory requirements - eg, the US EPA requires that the laboratory control sample (LCS) must be analyzed with each group of samples in a sample delivery group or batch of digested samples, whichever is more frequent. 3. Availability of IQC material - depends upon cost, stability, etc. The degree of analytical control to minimize errors and bias should have priority, but if the sole IQC material suitable is only available in a limited amount then a com-
80
3 Validation and Quality Contra/ with Atomic A bsorpption Speclrmieiry
promise has to be found. In general, an analysis frequency of 2 - 3 IQC sample, bracketing between 5 and 10 samples is considered by this author to be acceptable for AAS analyses. Naturally the stability of the analysis procedure assessed during the method validation procedure will influence this choice. If a large variation in the analyses, eg, 80- 120%, is acceptable then a less frequent IQC repetition rate could be necessary. Though control within the limits of 95- 105% could need a more frequent IQC re-analysis rate. The availability of an IQC material is important. If a long-term study is planned, requiring analyses over a year and sufficient IQC material is only available for 6 months then either a compromise has to be made or an alternative IQC material found. For an intelligent assessment of how often the IQC materials have to be analyzed the analyst requires a consideration of type I and type I1 statistical errors. The analyst is involved in making a decision, namely, are these IQC samples within the degree of analytical control required, hence the reliability of the actual sample results is established. The problem that can occur is one where the IQC results are unacceptable yet the sample results are satisfactory (ie, type I error, null hypothesis rejected even though it is true), this can occur when the IQC acceptance limits are set too tight for the capabilities of the analysis, and the converse where the IQC results are acceptable yet the sample results are wrong (ie, type I1 error, null hypothesis accepted even when it is false) for instance where the IQCs are analyzed too infrequently and variable within-batch drift is not detected. The two types of error are inter-dependent and in this situation depend upon the acceptable limits applied to the IQCs and the frequency with which they are measured. Both of these types of error can be minimized by increasing the frequency of IQC analysis and setting limits based on the method validation performance characteristics. From this author’s experience, where accurate, reliable and stable data are required for environmental surveys it is better to accept the risk of increased type I errors, by setting tight limits and living with the increased numbers of samples to be re-analyzed than the risk of generating poor quality data. The amount of IQC material required will depend heavily on the type of analysis, the sample aliquot required for the determination and the frequency with which the IQC test is performed. As an example, the data in Table 3-5 show some typical requirements for flame and electrothermal atomization applications. Both of these examples in Table 3-5 show the large amounts of IQC materials required just for one year of operation and demonstrates the quandry many laboratories find themselves in, in trying to establish an IQC protocol and find suitable materials. The lead in blood example, assuming 1.5 - 2 mL of blood per typical commercial sample vial equates to approximately 416 vials per year and does not allow for any breakages or losses owing to contamination. In addition, most laboratories concerned with such an analysis will be running 2 or 3 IQCs (ie, low, medium and high concentrations) and so the amount of material required becomes immense. As the commercially available CRMs and RMs are expensive and only available in limited amounts it is clear that these materials should preferably be used to validate methods and while they should be examined on a regular basis, this does not imply on a daily basis!
3.3 Quality Control
81
Table 3-5. IQC materials required for an average flame and electrothermal atomization protocol. Determination
Original sample aliquot required Number of samples per day IQC repeat frequency Amount of IQC required: per daya per week per month' per yeard a
Assuming analysis. Assuming Assuming Assuming
Flame
ETA
Fe in waste water
Pb in whole blood
10mL 10 10 30 mL 150 mL 600 mL 31.2L
50 WL 50 5 0.6 pL 3 mL 12mL 624 mL
IQC analyzed after initial calibration and calibration and IQCs repeated at end of 5 day working week. 4 working weeks per month. 52 working weeks per year.
3.3.2 Preparation of In-house IQC Materials To overcome the difficulty of finding IQC materials it should not be overlooked that in-house IQCs can be easily produced. In many instances samples are supplied to laboratories in large amounts, eg, liter amounts of waste waters, when only 10- 100 mL may be required for all the analyses. These samples which in most laboratories are disposed of after a certain period of time can be collected and used to prepare in-house IQC materials. In general it is not hard to divide the samples into low, medium and high concentration ranges. The samples can be pooled, well mixed and aliquoted out into suitable sample sizes ie, what is required on a daily or weekly basis, and stored under suitable conditions. It has to be admitted that such a procedure works well for liquid samples and that solids are more difficult owing to the problems of ensuring sample homogeneity. However, this problem can be partly overcome by reserving all digestion solutions after the completion of analysis and pooling, mixing and aliquoting as for liquid samples. This is not a perfect solution as these IQC samples will not have been through the same digestion procedure at the same time as the actual samples. However, these solutions can be used to provide regular quality control of the final AAS measurement and calibration procedure. With any in-house prepared, IQC materials it is important that the materials selected are assessed for safety, stability and that suitable storage conditions/facilities are available. The safety issue should not be overlooked, especially with biological samples. While commercially available materials are often from non-human sources and treated to ensure sterility, this may not be the case with in-house materials and all users of the samples should be aware of the handling precautions necessary to ensure safety in the laboratory.
82
3 Vulidution and Quality Control with Atomic Absorption Spectrometry
The stability of the in-house materials can only be assessed over a period of time by repeated analyses in parallel with CRMs over several months. During this time the IQC materials should be assessed not only for the stability of the analytical results, but also with respect to physical stability, there should be no indication of precipitation or decomposition. In general, stability of most sample types is improved by storage at low temperatures and in the dark. The influence of the sample containers should be considered, as it is not uncommon for some trace elements to be leached from sample container walls and it is a wise precaution to prepare a batch of blank sample containers containing water or dilute acid to be run with the proposed IQC samples during the course of the assessment procedure.
3.3.2.1 Establishment of IQC Target Values and Limits Assuming that the IQC samples are stable, then during this time sufficient data will have been collected to allow the calculation of the relevant statistics such as target values and allowable k ranges. Careful design of the experimental procedure and application of statistical techniques such as ANOVA can allow estimates of the within-batch and between-batch variability to be calculated and will show whether the day-to-day factors are important [61]. It is clear that if there are large day-to-day variations then these should be used to produce the acceptance limits. In addition, these results can be interpreted with the earlier values found during the initial method validation work. If everything is functioning correctly then the values obtained for these parameters should be similar. It is difficult to give guidance on the number of measurements that should be performed on the IQCs to establish the limits, however, for statistical reasons at least 20 measurements should be collected and preferably more [61]. The preparation of a reliable in-house IQC material represents a considerable investment in time and effort, but is necessary if good analytical control is to be established. Once the procedure is running, then it is a simple matter to ensure that subsequent IQC batches are prepared, assessed and target values established before the earlier batches are exhausted. This will ensure continuity in the quality control protocol.
3.3.3 Use of Quality Control As has been discussed earlier, IQC procedures are used to control analytical accuracy and precision. The regular use of IQC samples with target values which are close or equal to the true value and tight working limits, allows the consistent reporting of accurate and valid analytical results. However, there is an issue with respect to IQC that is often overlooked and needs to be considered if an IQC protocol is to be successfully implemented. It is vitally important that the significance of quality control is fully understood by the laboratory management and that all those involved in the analytical work are aware of this. All those concerned should have a pride and involvement in producing good quality data.
3.3 Quality Control
83
The laboratory workforce must understand why the IQC protocol is important and be motivated to comply with the protocol and be proud of achieving good performance. Without the psychological motivation of the workforce performing the analyses and operating the AA instruments, it is unlikely that any IQC protocol can be applied successfully.
3.3.3.1 Defining a Quality Control Procedure A major step to ensuring compliance with an IQC protocol is to ensure that it is simple to operate and with a defined procedure. Such a procedure could be defined as shown in Fig. 3-3. The sample results will only be acceptable if the values for the IQC samples before and after, are within the defined working ranges. This is relatively simple, but it is essential that either the instrument user or the instrument control software reacts to the results immediately. In addition, clearly defined protocols have to be established to deal with those situations when a fail action occurs. It is far better that monitoring of the IQC results is carried out on-line, and simultaneously with the analysis as this allows a quick reaction to problems which may be developing, rather than post-analysis when errors could be detected after the event. A post-analysis review of the data can lead to frustration as large batches of samples could have to be repeated, but in some situations, and especially with older instrumentation where results may not be presented immediately in concentration terms, this situation cannot be avoided. The simplest type of IQC protocol is a manual procedure, for example with flame AAS, where one IQC is analyzed every x samples and the target value and k ranges are fixed to the front of the instrument. As the analysis proceeds, and the results are displayed, the user can immediately see whether the IQC results are acceptable or not. The actual rules for acceptance or rejection must be defined and known to all users. These can take the form of: ‘If the IQC value falls outside the +- range, the IQC must be repeated; if the IQC value is still outside the +- range then the analysis must be stopped and the procedure investigated’. This is the simplest type of Pass/Fail control and can now be found automated in some of the AA control software packages commercially available. This approach can also be used with Shewart or control charts [59, 611, though this can be rather cumbersome in trying to plot data on a graph and manually carry out the analysis at the same time! An example control chart for the data presented in Table 3-6 is shown in Fig. 3-4. Control charts are generally constructed to include warning limits, generally set at lt2s and action limits at f3s and these limits have to be established before starting the quality control scheme. These limits are often established from the mean and standard deviation values of all the pooled data or preferably from a consideration of the results from an ANOVA scheme as developed in Table 3-6. The use of such charts allows the detection of time-dependent systematic errors, ie, errors that influence the accuracy. In general, such charts are used to monitor the daily or batch performance of an analytical method. The best means of continuous monitoring of IQC results during the course of an analysis, whereby some individual 20 IQC results a day could be produced (and possibly at different concentrations) is an area that
84
3 Validation and Quality Control with Atomic Absorption Spectrometry
L=l Start analysis
Calibrate AA instrument
Perform laboratory specified corrective
Calibration satisfactory?
---+
Analyze IQC(s)
a
No----+
Perform laboratory specified IQC failure action
c=l Yes
Analyze samples
Fig. 3-3. Example of a typical internal quality control protocol.
85
3.3 Quality Control
Table 3-6. Example of analysis of variance (ANOVA) to determine significance of within-day or between-day variance for establishing IQC limits.
Number per day
Days 1
2
3
4
5
6
1
8
25.9 26.1 25.1 26.3 25.1 26.5 25.1
26.1 26.1 25.1 21.1 21.1 21.1 26.9
26.1 21.1 26.5 26.5 26.1 28.4 25.1
26.1 26.5 28.2 28.0 21.1 21.4 21.8
21.1 26.3 21.8 26.9 28.0 21.6 27.4
26.9 26.3 26.9 21.1 26.1 26.1 25.1
26.3 26.5 21.6 21.1 25.9 21.1 21.1
21.8 26.5 26.5 21.1 26.1 21.4 26.1
26.0
26.1
26.1
21.3
21.3
26.4
26.8
26.8
181.9
186.1
181.0
191.1
191.1
185.1
181.6
181.5
1 2 3 4 5 6
I Means for days Sum
Means for run position
Grand total (T) Grand mean (T/mn) Correction term (T2/mn) Source of variation
213.5 21 1.4 214.9 216.1 212.1 211.6 212.4
Number of columns (rn) Number of rows (n)
1498.0 26.15 40011.5
Sum of squares
26.1 26.4 26.9 21.0 26.5 21.2 26.6
Sum
Degrees of freedom
Mean square
8 7
F
-
Between day Between run Residual Total
9.06 3.92 15.12 28.10
I
1.29 0.65 0.36 0.51
6 42 55
3.60 1.82
Values from table of percentage 0.05 0.025 points of the F-distribution 2.25 2.62 F(7.42) 2.34 2.14 F(6.42) From these calculations the between-day variation is found to be more significant than the withinrun variation. Hence the between-day variation should be used t o calculate the control chart limits. This gives 26.95k0.86 for the i 2 s warning limits and _t 1.29 for the +3s action limits.
does not yet appear to have been fully explored. Some of the current AA instrument software control packages available do allow some form of QC charting to be performed on-line, however, these tend to be graphical within-run charts, and do not allow sophisticated setting of variable limits, nor any control based on an examination of the trend in IQC results.
3.3.4 Systematic and Random Errors The problem of monitoring one IQC sample per batch or per day in an analytical procedure is that this does not provide continuous control of the procedure, especial-
86
3 Validation and Quality Control with A t o m i c Absorption Spectrometry
2m 27.5 = Y
C
27.0
m 5
26.5
.-c0 C
@
0
26.0
0 25.5
- _______._._________.____._____.__.__.__.____.__._
ly in situations where accurate and precise results are required which are stable with respect to time. Here the frequency of repeat IQC analyses must be increased. While modern AAS instruments are very stable, drift can occur, for example owing to aging of the graphite tube in ETAAS, or evaporation from sample cups during an overnight analysis run. Frequent checks can detect the onset of such changes, and allow corrective action to be taken, eg, replacement of graphite tube or a re-calibration procedure. Up to now the discussion regarding IQC has concentrated on the assumption that only one IQC material, at one concentration is being repeatedly analyzed during the course of the analysis. While such a procedure is better than no QC it does not provide control over the whole analytical range of measurements and can only correct for gross systematic errors. Consider the three calibration graphs (A, B and C) shown in Fig. 3-5. In general, most laboratories choose an IQC material with a concentration near the mid-point of the calibration range. In example A, such a choice will allow the detection of a gross shift in the calibration function as shown in calibration curve B. However, the more subtle change of a rotational shift in the calibration function as shown in calibration C will not be detected by the use of one IQC material and results at the low and high concentration range will suffer from + ve and ve bias, respectively in the results. Similarly there is a need to assess, when one IQC fails, where the problem is: is it due to random error or has a systematic change taken ~
Fig. 3-5. Example calibration curves indicating the errors that can go undetected when only one IQC material is used. (A) Original calibration curve, IQC sample results are within the target range. (B) Subsequent calibration curve with a systematic biaddrift. Change in calibration detected by the incorrect IQC results. Analysis stopped and reason for change investigated. (C) Subsequent calibration with a rotational bias. Change in calibration not detected with an IQC sample located at or near the mid point of the calibration. Results for low concentration samples will be biased high and high concentration samples biased low with respect to the ‘true’ values.
3.3 Quality Control
t
Absorbance signal
-
Concentration 4
IQC Target range
Oriainal ca1ib;ation
t
I B
Absorbance signal
.IQC Target vafue outside of limits
4
Original calibration
n
t
Concentration
I"
/
Absorbance signal
T m p l e biased low
Fig. 3-5
U
IQC Target value satisfactory
Concentration
+
87
88
3 Validation and Quality Control with Atomic Absorption S p e c f r o m e ~ y
A
30-
............................................................... 24.
__
$
12.
.
.
.
.
.
.
.
.
.
.
.
.
1
5
9
13
17
21
25
29
33
37
41
45
...............................................................
C 11. 10 -I 1
5
9
13
17
21
25
29
33
37
41
45
IQC analysis sequence number
Fig. 3-6. IQC control charts showing application of multiple-rule control according to Ref. [62]. (A) High control; (B) low control. Sequence 1 - 10 shows the application of the rule, indicates a slight systematic bias. Sequence 20, high control outside I,, level hence run rejected. Sequence 33 - 34, both controls outside the 2,, range indicating severe systematic bias, run rejected. ( - - -) _+ I s limit; (- . -. - .) +2s limit; (-----) +3s limit.
place? Increasing the number of IQC samples, for instance, to two, low and high, or three, low, medium and high, covering the range of interest allows improved control and the detection of situations depicted in calibration graph C more easily. However, the interpretation of IQC resultskharts when more than one IQC result is obtained becomes more complex. Westgard et al. 1621 have proposed a multi-rule Shewart chart for QC employing 2 IQC samples, low and high. With these rules, which can be interpreted from the two respective Shewart charts as shown in Fig. 3-6, random and systematic errors can be detected and differentiated by application of multiple rules which are applied when any IQC falls outside the +-2s range (warning limit). These rules can be summarized as:
Random error
l b One IQC result outside +3s range R4s Difference between the two IQCs exceeds 4s
3.4 External Quality Assessment
89
Systematic error 22s Two consecutive results on an IQC are outside the same limit, ie, +2s or -2s
4,, 10,
Four consecutive IQC results are outside the same limit of +Is or -Is Ten consecutive IQC results are on the same side of the mean
This multiple rule system is easily applied to within-batch continuous monitoring of IQC and allows the ready detection of within-batch bias. This multi-rule approach has been further extended by Delves [63], to accommodate three IQCs, reduce the need for continuous charting and allow on-line manual control for the determination of whole blood lead levels by micro-sampling flame AAS [64] for the purpose of long-term environmental surveys. This procedure was also applied post-run to an ETAAS method [65] for the same purposes. This multi-rule approach to quality control has been tested for over 9 years with respect to blood lead analyses and has produced some outstanding analytical data [18, 66-68]. From the point of view of analytial quality control the advantages of multiple IQC approach enables far tighter control to be achieved, and subtle changes in analytical bias to be detected almost immediately. Current AA instrument software control packages, while sometimes allowing the use of multiple IQCs (up to eight from one manufacturer) operating on a Pass/Fail basis, based on user entered upper/lower limits are only considered individually, and though some of the failure actions can be IQC specific, rather than general to all, none of the packages allow a more sophisticated and intelligent control through variable limits and rules.
3.4 External Quality Assessment While the use of a validated analytical method and an IQC protocol will ensure accurate and precise results within a laboratory, participation in a proficiency testing, or external quality assessment scheme provides an independent and continuous means of ensuring that the analytical quality control within the laboratory is effective. One of the main aims of external proficiency testing is to encourage the proper use of quality control and to act as an external reference and guard against bias. For environmental monitoring it is essential that the data provided by the many different laboratories are comparable and participation in an EQA scheme by all the laboratories involved should be considered essential. A broad overview and assessment of the aims and objectives of proficiency testing has been given by the Analytical Methods Committe of the Royal Society of Chemistry [69]. While the participation in EQA schemes in highly recommended, and in some instances is required by the regulatory authorities, ie, the National External Quality Assurance Scheme (NEQAS) for blood lead determinations in the UK [70], EPA in USA [52], there are many areas of environmental importance where such schemes are not operating. Within the UK there are established schemes covering water analysis, (Aquacheck [71]), the Yorkshire Water Scheme [72] which distributes natural wa-
90
3 Validation and Quality Control with Atomic Absorption Speciroinetry
ters, sewage, effluents, sediments and sewage sludges, trace elements in biological fluids [73] and the UK NEQAS, lead and cadmium in whole blood [70]. Some of these schemes are not limited to UK participants and there are international schemes organized by various United Nations agencies. Most of these schemes are based upon the regular distribution of individual samples from a uniform bulk material to all the participating laboratories for analysis. Results are returned to the organizers where the data are collated, analyzed statistically and reports issued to the participants. A practical example of this type of information is given by Christensen et al. [74] for the Danish external quality assurance scheme for the determination of lead in blood. The provision of such information allows participarting laboratories to assess their long-term performance in considerable detail. Some schemes assess the performance of the participating laboratories by means of a scoring mechanism which produces an index which allows ranking of the laboratories [69]. It is important that EQA takes place on a regular basis, to ensure that bias or errors do not go undetected. A minimum recommendation is for quarterly tests to be performed [69] but a more frequent distribution is to be preferred. The UK NEQAS scheme distributes a single sample every two weeks and while this may seem excessive the data produced by two laboratories analyzing a minimum of 15 EQA samples per month produced far better inter-laboratory agreement [18] than was found by another laboratory [75] for an intra-laboratory comparison whereby approximately 20 EQA samples were analyzed per year. It is also important that EQA samples are not treated in any way different to those of ordinary samples that come into the laboratory on a daily basis. This is to ensure that the performance established by the EQA results represents the actual real-life performance of the laboratory. In practice this is often difficult to achieve and if it is not possible to insert the samples ‘blind’ within the analytical run then one has to rely on the trust and honesty of the laboratory staff. The advantage of participating in an EQA scheme is that it provides on ongoing check of the laboratory performance and can in some instances detect changes, even though the within-laboratory IQC is satisfactory. Delves [76] has shown a practical example of this whereby the UK NEQAS score for his laboratory showed a distinct drift upwards and then down again which could be correlated with the absence of the technician who routinely performed the measurements. During this period a post-graduate research assistant inexperienced with the method took over the analysis and the U K NEQAS score illustrated the learning curve. While at no time was the laboratory out of control, the scoring system of this scheme showed quite clearly the effect and illustrated the benefits to be gained in assessing a laboratory’s performance via EQA.
3.5 Conclusions The last decade has seen tremendous advances in AAS instrumentation and especially with respect to ETAAS, the technique is now firmly established as the reference technique for trace element analyses in the environmental field. While many methods
References
91
have been developed, it would appear that few laboratories have spent much time developing quality control strategies specific for AAS. Few published papers concerned with the development of AAS methods list many of the performance characteristics discussed here and even fewer show quality control data. Many of the papers referenced here originate from the clinical chemistry field, primarily because workers in these areas seem to be more aware of the need for reliable analytical data. The simple application of IQC in many laboratories would dramatically improve the quality of analytical data being produced. While many laboratories are concerned with becoming accredited to many of the accreditation schemes and the techniques embodied in Good Laboratory Practise (GLP) are moving slowly out of the biochemistry and drug related testing laboratories into inorganic laboratories, however, many of these schemes do not include QC procedures and especially EQA schemes. The success stories of IQC and EQA that have been published have been referred to in the discussions above and it has recently been shown [77]in a survey that the rigour of a laboratory’s quality control procedure is directly related to that laboratory’s subsequent performance in an external assurance scheme. Those laboratories with the strongest quality control procedures scored significantly better in the external assurance scheme and the results gave a strong endorsement of routine QC as an important determinant of good analytical performance. Method validation and quality control requires a considerable investement in time, dedication and money. This investment is essential if the practice of analytical chemistry is to retain the confidence of public and political organizations and ensure that informed decision making takes place based on reliable analytical data.
Acknowledgements The advice, discussions and constructive comments from Dr H. Trevor Delves, University of Southampton and Drs William Barnett and Cindy Anderau, Perkin-Elmer Corporation, USA during the preparation of this chapter are greatfully acknowledged. The assistance of Dr Peter Marais, Technikon Pretoria, South Africa in helping to find the original literature reference to Ringbom plots was also appreciated.
References Allan, J.E., Analyst 1958, 83, 466. Mesley, R. J., Pocklington, W.D., Walker, R.F., Analyst 1991, 116, 975-990. Versieck, J., Trace Elements in Medicine 1984, I , 2. Taylor, A., Fresenius Z Anal Chem 1988, 332, 132-135. Griepink, B., Fresenius Z Anal Chem 1984, 317, 210. Delves, H.T., Anal Proc 1984, 21, 391 -394. Braithwaite, R.A., Girling, A. J., Fresenius Z Anal Chem 1988, 332, 104-709. Delves, H.T., J Anal Toxic01 1977, 1, 262. Parsons, P., Slavin, W., Spectrochim Acta Part B 1993, 48, 925-939.
92
3 Validation and Quality Control with Atomic Absorption Spectrometry
[lo] Ottaway, J. M., Fell, G. S., Pure Appl Chem 1986, 58, 1707. [ 1 11 Chappuis, P., Poupon, J., Deschamps, J.-F., Guillausseau, P.-J., Rousselet, F., Biologicul 7Yac.r Element Research 1992, 32, 85-91. [I21 Taylor, J. K., Quality Assurance of Chemical Measurements, Chelsea, MI: Lewis, 1987. [ 131 Dux, J. P., Handbook of Quality Assurance for the Analytical Chemistry Laborutory, New York: Van Nostrand Reinhold, 1986. [I 41 Garfield, F. M., Quality Assurance Principles for Analytical Labomtories, Arlington, VA: Association of Official Analytical Chemists, 1984. [I51 Analytical Methods Committee, Analyst 1989, 325, 95. [I61 Zehr, B.D., Maryott, M.A., Spectrochim Acta Part B 1993, 48, 1275-1280. [17] Slavin, W., Manning, D.C., Carnrick, G.R., Talanta 1989, 36, 171 - 178. [I81 Shuttler, I.L., PhD thesis, University of Southampton, UK, 1988. I191 Shuttler, I. L., Schlemmer, G., Carnrick, G. R., Slavin, W., Spectrochim Acta Part B 1991, 46, 583 -602. [20] Miller, J. N., Spectroscopy International 1991, 3/3, 42 -44. [21] Miller, J. N., Spectroscopy International 1991, 3/4, 41 -43. [22] Miller, J.N., Spectroscopy International 1991, 3/5, 43 -46. [23] Miller, J. N., Spectroscopy International 1991, 3/7, 45 -47. [24] Tyson, J. F., Anal Proc 1988, 25, 4-6. [25] Tyson, J.F., Bysouth, S. R., Anal Proc 1987, 24, 83 -87. [26] Ringbom, A., Z Anal Chem 1939, 115, 332-338. [27] van Dalen, H. P. J., de Galan, L., Analysf 1981, 106, 695-701. [28] Miller-Ihli, N. J., O’Haver, T. C., Harnly, J. M., Spectrochim Acta Part B 1984, 39, 1603. [29] Bysouth, S. R., Tyson, J.F., J Anal A t Spectrom 1986, I , 85-87. [30] Butler, L.R.P., Spectrochim Acta Part B 1983, 38, 913-919. [31] Barnett, W.B., Spectrochim Acta Part B 1984, 39, 829-836. [32] Analytical Methods Committee, Analyst 1988, 113, 1469- 1471. [33] Slavin, W., Manning, D.C., Carnrick, G.R., A t Spectrosc 1981, 2, 137- 145. [34] Welz, B., Fresenius Z Anal Chem 1986, 325, 95- 101. [35] Miller, J. N., Spectroscopy Europe 1992, 4/6, 26-27. [36] Gardner, M. J., Gunn, A.M., Fresenius Z Anal Chem 1988, 330, 103- 106. [37] Gardner, M. J., Gunn, A.M., Fresenius Z Anal Chem 1986, 325, 263-266. [38] Knegt, J., Stork, G., Fresenius Z Anal Chem 1974, 270, 97. [39] de Galan, L., van Dalen, H.P. J., Kornblum, G.R., Analyst 1985, 110, 323. [40] Hu, Y., Smeyers-Verbeke, J., Massart, D. L., J Anal At Spectrom 1989, 4, 605. [41] Massart, D. J. L., European Patent Application, 91200880.2, 1991. [42] Delves, H. T., University of Southampton, personal communication. [43] Mitchell, J. W., Zief, M., Contamination Control in Trace Elemenf Analysis, in: Elwing, P. J., Kolthoff, I.M., Chemical Analysis Vol. 47, John Wiley & Sons, New York, 1976. [44] Analytical Methods Committee, Analyst 1987, 112, 199. [45] Stevenson, C. L., Winefordner, J. D., Appl Spectrosc 1991, 45, 1217. [46] Miller, J.N., Analyst 1991, 116, 3. [47] Thompson, M., Anal Proc 1987, 24, 355-357. [48] “Nomenclature, Symbols, Units and Their Usage in Spectrochemical Analysis - II!’, Spectrochim Acta Part B 1978, 33, 242. [49] Miller, J.C., Miller, J.N., Analyst 1988, 113, 1351-1356. [50] Bookbinder, M. J., Panosian, K. J., Clin Chem 1986, 32, 1734- 1737. [51] Thompson, M., Analyst 1988, 113, 1579-1587. [52] USEPA Contract Laboratory Program, Statement of Work for Inorganics Analysis, Document No. ILM02.0 including Revision ILM02.1, September 1991, Exhibit E, Section V, p. E-21. [53] Thompson, M., Maguire, M., Analyst 1993, 118, 1107- 11 10. [54] Thompson, M., Analyst 1982, 107, 1169- I1 80. [55] Bookbinder, M. J., Panosian, K. J., Clin Chem 1987, 33, 1170- 1176.
References
93
[56] Rasberry, S.D., Anal Proc 1990, 27, 296-297. National Institute for Standards and Technology Standard Reference Materials Catalog, NIST Special Publication 260, Gaithersburg, MD. Commission of the European Communities, BCR Reference Materials Catalogue, Cummunity Bureau of Reference, Brussels. Gardner, J.A., Coleman, S., Farrow, S.G., Anal Proc 1993, 30, 292-295. Yeoman, W. B., in: Analytical Techniques for Heavy Metals in Biological Fluids. Facchctti, S. (ed.) Amsterdam: Elsevier, 1982, pp. 273 -284. Massart, D.L., Vandeginste, B.G.M., Deming, S.N., Michotte, Y., Kaufman, L., Chemometrics: A Textbook, Amsterdam: Elsevier, 1988, pp. 59-74. Westgard, J.O., Barry, P. L., Hunt, M. R., Clin Chem 1981, 27, 493-501. Delves, H. T., in preparation. Delves, H. T., Analyst 1970, 95, 431 -438. Shuttler, I. L., Delves, H. T., Analyst 1986, 111, 651 -656. Department of the Environment, UK Blood Lead Monitoring Programme 1984 -1987, Results f o r 1984, Pollution Report No 22, London: HMSO, 1986. Department of the Environment, UK Blood Lead Monitoring Programme 1984 - 1987, Results for 1985, Pollution Report No 24, London: HMSO, 1987. Quinn, M. T., Delves, H. T., Human Toxicol 1987, 6, 459-474. Analytical Methods Committee, Analyst 1992, 117, 97- 104. Bullock, D. G., Smith, N. J., Whithehead, T. P., Clin Chem 1986, 32, 1884- 1889. Aquacheck Scheme, Water Research Centre, Medmenham, SL7 2HD, UK. Yorkshire Water AQC Scheme, Leeds LS51 5AA, UK. Trace Elements External Quality Assessment Scheme, Department of Clinical Biochemistry and Clinical Nutrition, Robens Institute, University of Surrey, Guilford, Surrey GU2 5XH, UK. Christensen, S. L., Anglov, J.T. B., Christensen, J. M., Olsen, E., Poulsen, 0.M., Fresenius J Anal Chem 1993, 345, 343-350. Miller, D.T., Paschal, D.C., Gunter, E. W., Stroud, P.E., D’Angelo, J., Analyst 1987, /12, 1701 - 1704. Delves, H.T., Ann Cfin Biochem 1987, 24, 529-552. Thompson, M., Lowthian, P. J., Analyst 1993, 118, 1495- 1500.
This Page Intentionally Left Blank
4 Application of ICP- OES Techniques in Environmental QC Terry C. Dymott
4.1 Introduction In many analytical laboratories throughout the world, environmental analysis has become the number one priority. The increased levels of legislation now being enacted by various authorities, such as the EEC Committees in Europe and the Environmental Protection Agency (EPA) in the USA, place a responsibility on all organisations to monitor the condition of their environment and to measure their impact upon it. This, in turn, has meant that a major requirement has been created for instrumentation and analytical methodology capable of implementing the provisions dictated by the new laws. One of the main instrumental techniques which has assumed a leading role in the analysis of environmental materials is ICP-OES (Inductively Coupled Plasma - Optical Emission Spectrometry). This leading role can be attributed to the combination of beneficial features which make it particularly suited to this type of analysis. 1. ICP-OES is a fast, multi-element technique which permits the analysis of around 70 different chemical elements at the trace, minor and major concentration levels. 2. ICP-OES has adequate sensitivity for most of the commonly determined trace metals (for example Cu, Cr, Ni and Zn), superior performance to many other techniques for the more refractory elements (such as Ti, V, W) and allows the determination of some non-metals (including P, B and S). 3. In theory, samples may be introduced as liquids, gases or solids but in practice most samples are handled in liquid form, either as aqueous or acidic solutions. Measuring gaseous samples, for the stable hydride gas-forming elements like As, Se and Sb, is an important subsidiary technique. 4. ICP-OES suffers from few chemical interferences, although spectral interferences may be a problem. This makes calibration an easy process and the superior calibration linearity of this technique means that extended concentration ranges may be covered by a single calibration graph.
This chapter presents a description of the ICP-OES technique in terms of its theoretical background, a description of the instrumentation used, together with a review of its application to the analysis of typical environmental samples. It is not possible in the space available to present a full and comprehensive review of such
96
4 Application of ICP-OES
Techniques in Environmental QC
applications but the intention is to provide a balanced overview of the scope of ICP- OES analysis in this important area.
4.2 Theory of the ICP-OES Technique 4.2.1 Atomic Spectrometry Optical emission spectrometry is one of the techniques classified under the general heading of Atomic Spectrometry. The principles behind the technique are based on certain properties of the atomic form of matter.
4.2.1.1 Principles of Atomic Emission Modern theories state that an atom consists of a positively charged nucleus surrounded by a number of negatively charged electrons which are in rapid motion around the nucleus. For each electron in each atom there is a discrete set of energy levels (or orbitals) that the electron can occupy. The energy levels in an atom of element A are quite different to those which are permitted in another element B. Examples are shown in Fig. 4-1. The important fact is that atoms of the same element have similar energy levels whereas those of different elements each have their own unique set of energy levels. For an unexcited atom at normal temperatures, each atom exists in its lowest energy state, ie, its so-called ground state. To excite the atom, one or more of the loosely held outer (valency) electrons can be raised temporarily to the first, or a higher, energy level by the absorption of energy. If sufficient energy is applied, the electron is removed from the atom, which is then said to be ionised. This energy can be supplied by photons of light (as in Atomic Absorption) or by collisions due to heat. An atom in its excited state is unstable and remains there for an extremely short time. The return of the atom to the ground state occurs spontaneously and is accompanied by the emission of energy in the form of radiation. This radiation occurs at a wavelength (or wavelengths) which may be calculated from Eqs. (1) and ( 2 ) .
AE=Ej-E,= hv where Ej = the energy at the highest level E, = the energy at the lowest level h = Planck’s constant v = the frequency of the radiation then
A=- C V
4.2 Theory of the ICP-OES Technique
97
Energy
t
I
I Ground state
@ 4% A
B
Fig. 4-1. Schematic representation of the electron orbital structure for two different atoms, A and B, together with the discrete energy transitions permitted for their valency electrons.
where c = the speed of light and A = wavelength of the radiation. It is important to note that the emission of radiation does not necessarily occur at a single wavelength. An electron may return to the ground state in a single step or by a series of steps, corresponding to a number of intermediate energy levels. The greater the number of levels involved, the greater the number of spectral lines emitted and the more complex the spectrum. Energy level diagrams, as described by Mavrodineanu and Boiteaux [I], can be used to identify the permitted transitions and hence predict the wavelengths involved. The example shown in Fig. 4-2 (see p. 98) is for Mg. The wavelength of the emitted light is inversely proportional to the energy difference between the energy levels involved in the transition. In principle the larger the energy difference, the shorter the wavelength of the emitted light. For example, in the Mg case the 4.346 eV transition between the 3 s and the 3 p orbitals results in a spectral line at 285.21 nm. The 2.711 eV transition results in a line at 457.11 nm and so on. The important fact is that each element has a unique set of energy states and, therefore, each element also has a unique set of wavelengths which it can emit. By measuring the wavelength of a spectral line we can determine which element is present in a particular sample. In addition, the amount of light emitted is proportional to the number of the emitting atoms, hence we can measure how much of an element there is in a sample. This is the basis of Atomic Spectrometry.
98
4 Application of ICP- OES Techniques in Environmental QC
1St ionisation potential 7.646 ev _ 7.175 ev
ev
. _ _ _ _ _ _ _ _ _ _ - _ _ _
7 6.431 ev 6
7
5 4
3
2
‘2.711 ev 285.212 nrn 457.115 nrn
1
Ground state
Fig. 4-2. Energy level diagram for the Magnesium atom showing the permitted transitions, with the energy values (in electron volts) and the resulting wavelength of the emitted light.
4.2.1.2 Plasma Source Emission intensity is a function of a number of factors which include the temperature of the atoms. The Boltzmann relationship indicates that there is an exponential dependence on temperature and that emission will only be significant in high temperature sources. One such source is a plasma. By definition a plasma is simply a very hot gas in which a significant fraction of the atoms or molecules are ionised. This ionised nature means that a plasma gas reacts in a number of ways when subjected to electromagnetic influences. One such interaction is that a plasma gas surrounded by a time-varying magnetic field is inductively coupled, ie, electrical currents are induced in the ionised medium. It is these currents that cause resistive heating of the plasma gas so that it becomes self-sustaining. The hot plasma plume ionises a proportion of the cold incoming gas so that it conducts an electrical current and so on. The source arrangement commonly used consists of a quartz tube surrounded by a multi-turn copper induction coil powered by a radio-frequency (RF) generator. This is shown in Fig.4-3. Argon gas flows through the quartz tube to form the plasma. To initiate the plasma, a high voltage Tesla discharge into the gas is used to form a small number of ionised particles. When they reach the magnetic field, the conductance of the gas increases rapidly to the point where a large quantity of energy is coupled into the gas. After a few seconds a stable plasma will form within the induction coil area. It is prevented from touching the walls of the quartz tube by a thin screen of cool gas. Further adjustments to the applied power, argon gas flow rate, RF frequency and the design of the quartz torch are used to optimise the position, shape and type of plasma discharge. A toroidal plasma is required for analytical use. When a conductor is heated by RF currents, the current density is largest at the periphery and reduces
4.3 Instrumentul Systems RF induction
99
Electric
Argon gas (partially ionised by a Tesla discharge)
Fig. 4-3. Typical plasma source arrangement consisting of a quartz tube surrounded by a 3-turn RF induction coil. Indicated are the magnetic field generated and the induced electrical (eddy) currents in the flowing argon gas which heat the gas to form the plasma. Observation
/+-----
6.000 "C
25 ,?heg i ht
6,200 "C 6,500 "C 6.800 "C
15 mm
8,000 "C
@
10.000 "C
0 43
Sample aerosol
Fig. 4-4. Cross-section of a toroidal plasma with an indication of the temperature profile at various points in the plasma plume. The height of the typical viewing zone is indicated.
100
4 Applicafioti of ICP- OES Techniques in Environrnenfai QC
__
exponentially towards the centre. This means that the axial zone in the centre is cooler and it is relatively easy to inject a gas stream containing a sample aerosol through the centre without disturbing the plasma stability. The sample is heated by the surrounding plasma, experiencing very high temperatures for relatively long periods of time. Figure 4-4 is a cross section of a typical toroidal plasma, indicating the probable temperatures at particular points. These high temperatures are capable of exciting large numbers of elements and produce spectra of some complexity.
4.3 Instrumental Systems The schematic diagram in Fig. 4-5 outlines the basic components which make up an ICP-OES system. It may be considered to consist of four separate modules: 1. The spectrometer, which optically collects light from the plasma and separates the spectral lines produced by the different elements in the sample. 2. The plasma source, which excites the atoms in a sample to produce an emission spectrum which is characteristic of the elements in that sample. 3. The sample handling system, where a sample solution is converted into an aerosol capable of being excited by the source. 4. The detector and signal processing, where the light intensity is measured at each wavelength and ultimately converted to a digital result. Plasma
Argon in
To torch system Electronics * - control
-
Signal processing
A
4.3 Instrumental Systems
101
4.3.1 Spectrometers There are two types of spectrometer configuration used in ICP-OES: polychromators (providing simultaneous analysis) and monochromators (which are sequential in operation).
4.3.1.1 Polychromators Polychromators (or direct reading spectrometers) have been used for many years in emission work and consist of the following basic components: a focusing lens to transfer plasma light to the polychromator, an entrance (or primary) slit to select the light to be measured, a concave diffraction grating to separate the incident light into its component wavelengths, a series of exit (or secondary) slits to isolate the required spectral lines and a detector behind each exit slit to measure the intensity of the line. The most common design is the Paschen-Runge type, shown in Fig. 4-6, where the entrance slit, grating and exit slits are fixed on the Rowland circle. Typical focal lengths are 1 m and they are mounted in a light-tight and temperature controlled box for stability. The main advantage of the polychromator is speed of operation, typically up to 64 analytical lines can be measured in a period of 1 inin. The main disadvantage is that the lines to be measured are fixed at the time of manufacture, so that there is no subsequent freedom of choice.
Secondary
circle
Fig. 4-6. Optical layout of a polychromator of the Paschen-Runge design. Part of the light from the plasma source is selected by the primary slit, dispersed by the fixed grating and specific wavelengths fall on a number of exit slit/photomultiplier pairs set at the focal plane.
202
4 Application of ICP-OES Techniques in Environmental QC
4.3.1.2 Monochromators Monochromators measure wavelengths sequentially and are mainly of two types, the scanning Czerny-Turner design or the fixed echelle design. Figure 4-7 shows the Czerny-Turner design which consists of a focusing lens, a concave mirror which collimates the incident light on to a rotating plane grating, a second concave mirror to focus the dispersed light on to the fixed exit slit and detector. A single wavelength is selected at a time by rotating the grating so that the spectrum is scanned across the exit slit until the appropriate point is reached. Multi-element analysis is achieved by rotating the grating to the wavelengths one after the other. Due to the mechanical tolerances and thermal instability, the grating cannot go directly to an exact wavelength. This type of monochromator drives to a point near the desired wavelength and then scans very slowly across a scan window, taking intensity readings for typically 9 steps. The peak intensity is then extrapolated from the data using mathematical peak-fitting routines. A modern alternative to the Czerny-Turner design is the echelle monochromator shown in Fig. 4-8. Here a grating with coarse rulings disperses the spectrum into multiple overlapping orders which are then separated by a prism. These separated orders then fall on a 2-dimensional array of exit slits. A detector mounted in a carriage then moves over the array, stopping at the appropriate points to record the intensity for the selected wavelengths. The advantage of a sequential monochromator is the freedom to choose any wavelength for analysis, unlike the fixed selection in a polychromator. The main disadvantage is that the analysis takes longer to execute. There are also combined spectrometers which have both a polychromator and a monochromator in a single system.
/
Photomultiplier
Primary slit g ,ran itg
Moving
Secondaryslit Mirrors
Fig. 4-7. Optical layout of a monochromator of the Czerny-Turner or Ebert design. Selected light from the plasma is dispersed by the grating which must be rotated so that any required wavelength passes through the single exit slit and is measured by the single photomultiplier.
4.3 Instrumental Systems
103
2-dimensional
Fig. 4-8. Optical layout of a sequential echelle monochromator, utilising prism cross-dispersion and a moving photomultiplier detector. A 2-dimensional slit array allows any wavelength to be selected.
Most spectrometers cover the wavelength range of 190 nm to 800 nm but gas-purged or vacuum systems can go down to 165 nm to access lines for non-metals like S and P.
4.3.2 Plasma Sources RF generators supply the high frequency current to the induction coil which forms and sustains the plasma. A frequency of 27.12 MHz was often used in older systems but most modern generators use a frequency of 40.68 MHz because this can give better detection limits. Input power to the argon plasma can range from 0.7 kW (for simple aqueous or acidic solutions) up to 2.0 kW (when the solutions contain organic solvents), but in most cases a compromise value of 1.O kW is satisfactory. The watercooled induction coil typically has two to five turns. To obtain the best precision of results it is important to control the power to the plasma, as small changes in power can cause large changes in signal intensity. There are two main generator control methods: crystal control (where a fixed frequency is set by a piezo-electric crystal and changes in power are compensated for by a matching network) and free-running (where the power is held constant by a feedback circuit and the frequency is allowed to vary within wide limits). Both vacuum tubes and solid-state electronics are employed to provide the oscillating current supply. The plasma itself is formed in a torch, a typical example of the Fassel type is shown in Fig. 4-9. It consists of three concentric quartz tubes and the design and dimensions
104
4 Application of ICP- OES Techniques in Environmental QC
are critical in obtaining the best analytical performance. The outer tube has a diameter of about 20 mm and extends above the inner two tubes by around 20 mm. A middle tube (also known as the tulip because of its shape) is separated from the outer by a 1 mm gap and the main coolant (or plasma gas) flows through this gap with a spiral motion at rates of between 10 and 20 L min-'. The inner (or injector) tube has a 1.5 mm diameter hole in its tip through which the nebuliser gas, containing the sample aerosol, flows at around 1 L min-'. It is vital that this gas flow punches through the base of the plasma so that the sample aerosol passes up the central channel, rather than around the outside of the plasma. An additional gas flow (the auxiliary flow) passes between the inner and middle tubes. This is used to alter the plasma position in the torch and is typically employed when analysing organic solvents. Torches may be made in one piece or as three separate tubes (a demountable torch) in a base assembly. They may have quartz injector tubes (used for all solutions except those containing hydrofluoric acid) or ceramic versions for use with hydrofluoric acid.
' T
30 mrn
140
n -Coolant ,Coolant (plasma) argon gas ,Auxiliary
argon gas
Sample aerosol plus
nebulising gas
Fig. 4-9. A Fassel design of one-piece torch, indicating typical dimensions (in mm) and the purpose of the various gas flows. A ball-joint permits easy connection to the spraychamber.
4.3.3 Sample Handling Most samples are introduced to the plasma in liquid form, although gaseous hydrides can be utilised for some elements. Sample solutions are converted to an aerosol of fine droplets by a nebulising device such as a pneumatic nebuliser. Two
4.3 Instrumental Systems
-solution
+-40
rnrn 4 -'
105
in
t
Argon gas in
Meinhard glass
Cross-flow
Fig. 4-10. Pneumatic nebuliser types - Meinhard glass concentric type is self-feeding because low pressure area in gas stream causes a partial vacuum which pulls solution up from the container. The cross-flow type requires a pumped supply of sample which is sheared into droplets as liquid emerges from the capillary into the gas stream.
commonly used types are the concentric glass and crossflow nebulisers, shown in Fig. 4-10. Pneumatic nebulisers rely on a high velocity gas flow through a capillary to create a low pressure region into which solution is drawn (in the case of the concentric glass) or pumped (in the case of the cross-flow). The gas flow shears the liquid into a mist of droplets of varying diameter, typically from 0.1 to 100 pm, of which only the fraction below 10 pm can be successfully atomised and excited. These nebulisers give good sensitivity with many sample types but are prone to blockage when nebulising solutions containing high levels of dissolved solids. Another type is the Babington, or V-groove, nebuliser shown in Fig. 4-1 1. In this nebuliser the solution flows down a V-shaped groove over a narrow gas orifice to produce the aerosol. As the solution does not pass through a narrow capillary, this nebuliser is far less likely to block up and has become the preferred choice for solutions containing high solids and also particulates (ie, slurries). The grid nebuliser is a modified form of the Babington designed to damp out pulsations due to the peristaltic pump which contribute to the noise of the system. Once the aerosol is produced it passes into a spray chamber positioned below the torch. Its purpose is to remove the large droplets above 10 pm which cannot be utilised in the plasma. A typical design is the Scott type shown in Fig.4-12. Here the larger droplets fall out under gravity or collide with the chamber walls and then pass out of a drain. The percent of nebulised sample that reaches the plasma is typically less than 5. Normal spray chambers are made of glass but plastic versions are required when hydrofluoric acid solutions are to be analysed.
106
4 Application of ICP-OES Techniques in Environmental QC
Sample
Plastic body
Plact
/ Sample solution in ___)
Argon gas in Drilled , ' sapphire V-groove nebuliser
Grid nebuliser
Fig.4-11. V-groove or Babington type of nebuliser used for solutions containing high solids or particulate material such as slurries. Liquid falls under gravity across a high pressure gas jet and is sheared into droplets. Grid nebuliser modification has improved nebuliser noise characteristics.
For the ultimate in sensitivity, there is the ultrasonic nebuliser shown in Fig. 4-1 3. Here the solution is pumped over the surface of a piezo-electric transducer vibrating at 1.4 MHz. Standing waves in the liquid film cause a dense aerosol of small droplets to be ejected. These are passed through a heater (to vaporise the excess water) and then a condenser (to remove the water) to avoid overloading the plasma. These nebulisers are expensive and more prone to memory effects than the other types. They also tend to cause blockages at the narrow orifice in the torch injector tube with high solids solutions. Gas phase sample introduction is 100% efficient but can only be used in a limited number of cases. Elements such as As, Se, Sb, Bi, Sn, Te and Pb all form stable gaseous hydrides when acidic solutions react with sodium borohydride. The resulting hydrides are swept into the plasma with a carrier gas and yield detection limits significantly better than normal nebulisation.
4.3.4 Detectors and Signal Processing The intensity of the light at a selected wavelength is measured by a detector where the photon flux is converted into an electrical signal. This signal is then processed, normally by a system computer. Most detectors used in polychroniators and many of the sequential monochromators are photomultipliers although advanced array detectors are now becoming more popular, most often in combination with echelle monochromators. For a more complete review of the theoretical and instrumenta-
4.3 Instrumental Systems
107
Small (10pm) droplets pass to torch
Initial aerosol of widely varying size distribution
large droplets to fall out of aerosol flow Large droplets pumped to waste
Fig. 4-12. Scott-type spraychamber is a double pass design, employing a mixture of gravity and impaction with the end of the chamber to remove large (unusable) droplets. Drain is continuously pumped to reduce pressure fluctuations and improve plasma stability.
Coolant out t
Coolant in
Desolvated aerosol to plasma
Drain
Fig. 4-13. Schematic diagram of a CETAC ultrasonic neubuliser. A film of liquid pumped on to the air-cooled piezoelectric crystal is shattered into a dense aerosol which must be desolvated to prevent overloading the plasma. This is achieved by heating the aerosol to vaporise the liquid which is then removed in a cold condenser, leaving an essentially dry aerosol to pass into the plasma.
lo8
4 Application of ICP- OES Techniques in Environmental QC
tion background, the reader is recommended to read the books by Boumans 121 and Montaser and Golightly [3].
4.4 Monitoring of Environmental Pollution Environmental pollution is an unfortunate by-product of many human activities, such as industrial production, transportation and waste disposal from urban sites. The effects of these activities impact on all three main areas of our environment; the land, the water and the air. For example, mining activity can add pollution to the air (in the form of dust), to the water (in the form of washings from ore processing) and the land (in the form of mining waste heaps). Metals production and the incineration of urban waste products introduce fumes and particulate material into the air and, subsequently, the land. Many industrial processes and agriculture result in polluted waste waters and toxic leachates which flow into the surface water layers. Environmental pollution is monitored in a number of ways and this has resulted in a wide variety of sample types being supplied for analysis. This list shows the main classifications are:
-
Biological - serum, urine, tissue Waters - sea, fresh, waste Plant materials Soils and sediments Airborne particulates
As a result of this variety of sample types, a number of different sampling and preparation techniques have to be used. At this point it is necessary to define which chemical elements are considered to be toxic, which are held to be essential trace metals and which are known to be essen-
Table 4-1. Classification of elements in relation to their biological and environmental significance.
Essential major elements C N 0 P S C1 Na K Ca Mg
Essential trace elements
F I Se V Cr Mn Fe Co Ni Cu Zn Mo Si Sn As Toxic elementsa
Li Be Ba F CI Br As Sb Bi Pb Sn TI V Cr Mn Fe Co Ni Cu Zn Cd Hg a
A number of these elements may only be toxic at high concentration levels, eg, Se.
4.4 Monitoring of Environmental Pollution
109
tial major constituents in the human body. The generally accepted classes of elements are shown in Table 4-1. From time to time this list is subject to modification in the light of experience, for instance selenium is now classed as an essential trace element rather than a toxic element. In contrast, aluminium is now under suspicion as a toxic element instead of its previous status as a non-toxic one. Pollution legislation generally specifies certain concentration levels as being representative of pollution and these levels may vary in the different sample types under investigation. Any analytical technique considered for use in environmental analysis must possess a measurement sensitivity at least equal to, but preferably significantly better than, these levels before it can be said to be suitable. Table4-2 indicates the applicability of ICP-OES for the main chemical elements and in the main sample groupings characteristic of environmental applications. From this list it can be seen that ICP-OES has some difficulty in measuring the legal limits for toxic metals such as As, Cd and Hg in certain sample types. The use of Hydride techniques can
Table 4-2. Suitability of ICP-OES for environmental types of analysis (based on typical ICP-OES performance and without pre-concentration).
Element
A1 As B Be Ca Cd Cr cu Fe K Hg Mg Mn Mo Na Ni P Pb Se Si Sr Ti TI V Zn
++
Waters
Soils and sediments
Plant materials
Biological materials
+
++ + + +
+ + + + ++ + + ++ ++ ++ + ++ + + ++
0
+ + + + ++ + + ++ ++ ++ + ++ + + ++ + ++ + + + ++ +
0 0
0 0
0
+
0
++
0 0
+ ++ + 0 ++ + + ++ 0 + 0 0
+ +
0 0
0
++ ideal performance;
++ + + ++ ++ ++
0
++
++ 0
++ 0
++ +
+ +
++
++
+ suitable;
++
0 unsuitable.
0
++
+ + + ++ 0 0 0
++
1 10
4 Application of ICP- OES Techniques in Environmental QC
improve the level of performance but this tends to reduce the multi-element capability of ICP down to a few, often single, elements at a time. It is also the reason that pre-concentration by ion exchange, co-precipitation and solvent extraction is important in water and biological analysis. Where the ultimate in detection limits is required, alternative techniques such as GFAAS (Graphite Furnace Atomic Absorption Spectrometry) and ICP-MS (ICP-Mass Spectrometry) have a major role to play in environmental analysis. There are a number of characteristics which make ICP-OES suitable for environmental analyses. 1. True multi-element analysis, normally up to 60 elements at a time can be measured with an ICP instrument. The alternative Flame AAS technique is only really suitable for around 6 - 10 elements per sample. Many legislated analyses require that more than 15 elements need to be analysed. 2. Speed of analysis, when large numbers of elements must be determined ICP-OES is much faster than either Flame or Graphite furnace-AAS. For large centralised laboratories, the work load is often in excess of 200 samples per day. This number of samples cannot be handled by AAS, especially the much slower Graphite furnace technique. 3. The more refractory elements like boron and tungsten are all more sensitive by ICP- OES. 4. Sulphur and semi-metals such as phosphorus can be determined directly by ICP-OES. 5. ICP-OES has a very wide dynamic calibration range which means that it can measure at the high Yo level and the low pg mL-' (ppm) level with a single preparation of the sample. 6 . ICP- OES suffers far fewer chemical interferences than AAS so that sample preparation is often simpler.
4.5 Pre-analysis Considerations At this stage it is appropriate to consider some of the main factors which must be borne in mind when entering the sampling phase of an analytical procedure 141. This is the most important phase of an analysis, and often the one most poorly carried out. It must be emphasised that any mistakes made at this stage can ruin the excellence of all the stages that follow. There are two main considerations which are absolutely vital to the success of any analysis. 1. The sample taken and prepared for the analysis must be representative of the whole sample, unless only small inclusions are to be analysed. 2. The sample containers must be clean.
4.5 Pre-analysis Considerations
11 I
As an example of the first point, consider the sampling of natural waters where the following factors should be considered: the influence of tides in sea-water sampling; the proximity of tributaries when sampling river and lake waters. that the discharge of effluents into a river may occur at random times during a day or week; the presence of cold and warm layers in lakes; that still tapwater can contain erroneously high metal concentrations which are not typical of a flowing sample; the disturbance of sediments in a river or lake bottom caused by the action of taking the sample; that rainwater composition varies with the weather, season and even the time of day. Other examples from the biological and plant material field illustrate the same principle. When using human hair as an indicator of exposure to toxic materials, it must be remembered that the concentration levels in normal hair increase the closer one moves to the root of the hair. Similar differences can occur in plant materials where element concentration levels in old wood will be totally different to those in the new growing tip [ 5 ] . With regard to the second point, sample containers can produce negative or positive errors in trace metal analysis by either of two actions. They can add contaminants by leaching or surface desorption of material already present on the container walls. This may be residues of a previous sample or from materials used in manufacturing the sample container itself. Zinc compounds used as mould release agents for plastics are a notorious source of contamination in new ‘clean’ containers. Surface adsorption effects can reduce the apparent concentration of elements in a sample solution. This is particularly true of certain plastic materials, such as low density polyethylene, and glass. It is vital that laboratory containers, whether glass or plastic (this includes polyethylene, polypropylene or fluorocarbon) should always be washed first with a detergent solution that does not contain high levels of metal compounds. It is important to remember that certain detergents can contain high levels of metals, such as zinc and sodium, which remain bound to container walls even after flushing with clean distilled or deionised water. Glassware is often cleaned with a chromic acid wash because it is a powerful cleaning agent for removing organic material. However, the chromium persists in glassware for a long time, even after thorough washing with water. After the detergent wash, the containers should then be rinsed thoroughly in distilled or deionised water, followed by rinsing with a (1 + 1) solution of nitric acid. The acid is used to strip any metals adhering to the walls of the container. Another distilled water wash should be followed by a (1 + 1) solution of hydrochloric acid. A final distilled wash is made before the container is hot-air dried. Containers to be used for detection limit work should be kept separate from those to be used for general purposes to minimise the potential for cross-contamination. Bags and sacks used to collect solid samples, like soils and plant materials, must also be checked for the presence of contaminants before use [ 5 ] .
1 12
4 Application of ICP- OES Techniques in Environmental QC
4.6 Water Analysis A search of the many different environmental analyses routinely carried out around the world reveals that the most common analysis concerns water - in various states of purity from drinking water through to the many waste waters. Due to the important part water plays in our lives, there are currently more water samples analysed than any other single type. This is a very convenient sample to analyse by ICP- OES because it is already in liquid form, often requiring the minimum of preparation before analysis. Interference effects from the major matrix elements, like calcium, magnesium, sodium, etc., are minimal or easily corrected. However, for the purer forms, such as drinking water, and because of the low natural levels of abundance of many elements in natural waters, some form of pre-concentration may often be required. The high dissolved solids levels in seawaters and some waste waters can cause problems with blockage in nebulisers and torch injectors. This means that dilution and/or separation from the matrix elements can be necessary. For those intending to carry out trace metal analysis in waters, there are many reference books on the subject but the most influential text is probably the Method 200.7 Determination of Metals and Trace Elements in Water and Wastes by Inductively Coupled Plasma - Atomic Emission Spectrometry, produced by the US Environmental Protection Agency [6]. This has become the basis for many other official methods in Europe and elsewhere, including the methodology described here [ 7 , 8 ] .
4.6.1 The Sample Collection Process Before a sample is collected, the kind of data required for the final report must be defined so that the most appropriate procedures can be applied [9, lo]. For instance, in water analysis, is it the soluble metals or only the suspended solids or a total analysis that is required? In the case where only the soluble metals are to be measured, the following procedure is used. The water must be filtered through a 0.45 pm membrane filter as soon as possible after collection. It is normal to use the first 50- 100 mL of sample to rinse the apparatus, discarding this part of the filtrate. The required sample volume is then collected in an acid-cleaned vessel, preferably an appropriate plastic bottle. Borosilicate glass may be used but cause unreliable results to be obtained for boron and silicon due to contamination. Acidification with (1 + 1) nitric acid to pH 2 or less is used to stabilise the metal content. Normally 3 mL of (1 + 1) nitric acid per litre of sample should be sufficient. If it is not possible to preserve the sample on collection, it should be acidified immediately upon receipt in the laboratory and not processed for at least 16 h. If the suspended solids content is required, basically the same initial procedure is used. The difference is that the filter containing the suspended solids is retained and stored in a suitable container. No preservation is needed.
4.6 Water Analysis
1 13
For a total analysis, the whole sample is acidified with ( I + 1 ) nitric acid to pH 2 or less, preferably at the time of collection. The sample is not filtered and should be kept for at least 4 days, with periodic shaking of the vessel, before processing. A relatively new technique shows great promise for water analysis, that of on-line concentration. The basis of the method involves the preparation of small mini-columns containing an ion-exchange material. These are taken to the sampling site where known volumes of the sample are drawn through the column, trapping the required elements. The columns are sealed and taken back to the laboratory where the trapped metals are eluted using a small volume of eluent. Concentration factors of 10-20 times can be achieved, an added benefit which improves the ultimate detection limits attainable. Columns can be stored for several weeks before carrying out the analysis [ 1 1 - 151.
4.6.2 Sample Treatment Before Analysis For soluble metals analysis, the filtered and acid-preserved samples can often be analysed as received. If a precipitate forms during storage, it must be re-dissolved by adding acid and/or heating. It is normal practice to analyse samples against matrix matched standards. The suspended solids held on the filter paper must be solubilised before analysis. This is carried out in the following manner. Place the filter membrane in a suitable clean beaker and add 4 mL concentrated nitric acid. A cover is placed over the beaker and it is heated gently to dissolve the membrane. The temperature is raised to digest the material, continuing until the contents are reduced to low volume. After cooling, another 3 mL of concentrated nitric acid is added and the beaker covered again. Heating is continued until digestion is completed and the contents are then reduced to about 2 mL. The contents are cooled and 10 mL of (1 + 1 ) hydrochloric acid plus 15 mL of distilled water are added for every 100 mL in the final sample volume. The metals are re-solubilised by heating for 15 min, then the contents are cooled and washed into a volumetric flask. Any remaining insoluble material is filtered if necessary. Finally the contents of the flask are made up to volume. It is important that a blank is put through the complete procedure, whichever one is used, so that a method or laboratory blank can be established.
4.6.3 Instrument and Method Detection Limits Water analysis is carried out and measured against a number of criteria, which tend to be different in different parts of the world. There is presently no universal list of limits but the most commonly used criteria for natural waters are those published by the EPA in America. Table 4-3 shows normal concentration levels for river and sea waters, together with rain, and typical detection limits obtained using ICP instruments.
114
4 Application of ICP-OES Techniques in Environmental QC
Table 4-3. Comparison of typical ICP-OES detection limits with median concentration values reported for river water, sea water and rain. All figures in ng m L - ' . Element
River water
Sea water
A1 As B Be Ca Cd Cr cu Fe K Hg Mg Mn Mo Na Ni
100
10 3 4.5 x 103 5x 1 0 - ~ 4~ 105
P Pb Se Sr V Zn
10 0.4 1.7 x 104 0.05 1 3 100 2.2x 103 0.07 4x lo3 15 1.5 6 x 10; 10 20 3 0.2 50 0.9 20
0.1
0.05 3 10 4 x lo5 0.03 1.6 x lo6 2 10 I x lo7 2 70 0.03 0.4 8 x lo3 2 5
Rain water
ICP d.1.
1 x 103
40 20 (1 *)
5 -
2~ 103 0.5 7 30 1 x 103 500 0.01 400 60 2~ 103 10 50
10 200
1 0.1 30 2 3 2 10
60 5 (1*) 1
2 4 20 8 10 25 80 (2*) 0.5 2 5
* I C P performance figures with hydride generation.
The same situation applies for drinking water but Table 4-4 shows the figures currently used within Europe [I61 at this time. Note that the figures in brackets are for ICP performance when using the Hydride technique, not the standard pneumatic nebulisation method. It is obvious that some elements (Na, K, Mg, Ca, Sr, Ba, Fe, B, S and Si) can be determined directly with pneumatic nebulisation, others will require the use of hydride techniques [ 17, 181 or ultrasonic nebulisation [ 19, 201 to achieve adequate sensitivity while the remainder will need to be concentrated.
4.6.4 Pre-concentration Techniques The simplest form of concentration is evaporation but this is usually limited to drinking and fresh waters [21, 221. Evaporation also concentrates the solids contents of solutions and this causes sea, estuarine and most waste waters to exceed the limit for aspiration with normal pneumatic nebulisers. Babington-type nebulisers can handle these waters but this merely transfers the blockage problem to the torch injector.
4.6 Water Analysis
1 15
Table 4-4. Comparison of typical ICP-OES detection limits with Maximum Allowable Concentration (MAC) values for drinking water, with guide values where appropriate [16]. All figures in ng mL-'. Element
Guide value
MAC value
ICP d.1.
As Ba Cd Cr cu
50 100 5 50 3000a
20 (1 *) 2 2 3 2
Fe Hg Mn Ni P Pb Sb Se
200
lOOb 1
50 50 5000 50 10 10
10
5 (1 *)
2 8 10
25 50 (1 *) 80 (2*)
This value is for water held static in pipework. This value is for running water from pipework. * ICP performance figures with hydride generation.
a
Evaporation can provide a l o x -2Ox concentration factor and is quick and simple to carry out. Contamination is rarely a problem but care must be taken to avoid too vigorous a boiling action and subsequent loss of sample. Solutions must not be allowed to boil dry, otherwise loss of volatile analytes may occur. A non-selective ion exchange method has been described 1231 which consists of adding a mixed-bed exchange resin to the freshly collected sample in a plastic bottle. Extraction is carried out by shaking the bottle for a period of time. The resin is recovered and treated with a small volume of acid to retrieve the metals. Volatile elements are not lost but a major problem here is the purity of the original exchange resin. Another method involves wet digestion of the resin itself after the extraction procedure 1241. Solvent extraction has been used for multi-element concentration of trace metals in waters [25]. Using dithiocarbamate extractions, concentration factors of up to 500 have been achieved [26]. The solvent used must have a low aqueous solubility but more importantly be compatible with aspirating into the plasma. For this reason, back-extraction into a low volume of dilute acid is preferable when non-suitable solvents, like chloroform, have been used [27]. The use of selective ion-exchange resins has been referred to in Section 4.6.1 and this technique using Chelex-100 is very popular 128 - 321, especially when analysing sea-waters. Concentration factors of 25 have been measured and > 99.9% removal of the Na, K, Ca and Mg matrix is achieved. However, even with this factor elements such as Cd, Co and Pb could not be determined directly. Other compounds have been used to yield even higher factors, silica-immobilised 8-hydroxyquinoline has
1 16
4 Application of ICP- OES Techniques in Environmental QC
given concentration factors of up to 200, for example [33 -351. It has been the arrival of flow injection techniques that has revolutionised pre-concentration [36 - 391 and automated accessories are now available [40, 411. Typically the number of elements that can be extracted simultaneously is between 5 and 10. Co-precipitation is another method for simultaneously concentrating a large number of trace elements in water analysis but the concentration factors are typically only 10 to 20 [42-451. This technique does not seem to have gained widespread acceptance, especially as other methods have improved in range and concentration factors.
4.6.5 Analytical Conditions The analytical conditions chosen for water analyses have been comprehensively described in the EPA Method 200.7. In most cases, it is possible to utilise the most sensitive lines of the analytes and Table 4-5 lists the EPA selection. Spectral interferences are not a major problem with drinking and many natural waters because the individual element concentrations are low. However, in waste waters the problem can be more severe as it is possible that major pollutants may be present in a 1000 fold excess. Under these conditions, choice of another line where no spectral interferences exist may be an option but this can give rise to sensitivity problems. Interelement correction is then the only alternative - this involves measuring the effect of a single element standard of the interfering element (at a concentration which is typically found in real samples) at the analyte wavelength. The resulting correction factor is then applied to the raw data used during the calculation of results. It should be noted
Table 4-5. Recommended analytical wavelengths for water analysis. (Taken from the USEPA Water Method 200.7 [6]). Element
Wavelength (nm)
Element
Wavelength (nm)
Ag A1 As B Ba Be Ca Cd
328.068 308.215 193.696 249.678 493.409 313.042 31 5.887 226.502 228.616 205.552 324.754 259.940 194.227 766.491 670.784
Mg Mn Mo Na Ni P Pb Sb Se Si Sn Sr TI V Zn
279.079 257.610 203.844 588.995 23 1.604 214.914 220.353 206.833 196.090 251.611 189.980 421.552 190.864 292.402 213.856
co Cr Cu Fe Hg K Li
4.6 Water Analysis
1 17
that all correction factors introduce an uncertainty into the results and this will degrade precisions and detection limits. In addition, Kocherlakota [46] has reported that at least two of the IEC factors listed in the EPA Water Method 200.7 (that for Ca on the Mg 270.079 nm line and for Mg on the Ca 317.933 nm line) are incorrect. They appear to be the result of contamination in the standards used to derive the factors. This illustrates an important principle - one should not use published IECs obtained on a different instrument to one's own, without first checking that they still apply. Ideally it is better to employ a high resolution spectrometer (ie, <0.01 nm), where lines are better separated than with medium/low spectrometers (0.02 nm or worse), to minimise spectral problems. Another problem encountered, especially with sea water, is an elevation of the background argon continuum by recombination continua due to Ca and Mg (at 300 nm and 200 - 250 nm respectively). This is normally corrected by using off-peak estimates of the background intensity which are then subtracted from the overall intensity to produce the corrected analyte signal. Calibration is usually straightforward and in most cases (due the low concentration levels involved) a two-point calibration is normally satisfactory. Only where concentration levels are extremely high, eg, in waste waters or for Na, Ca and Mg in seawaters, is it necessary to use curve fitting procedures. Calibration standards generally contain analytes at 10 pg mL-' or less so that multi-element standards can be prepared. Table 4-6 (see p. 118) lists the standard mixtures recommended in EPA Method 200.7 which are commonly used. These mixtures are chosen to avoid precipitation problems but care must be exercised that excessive contaminants are not introduced from the single element standards used to make them [46]. All blanks and calibration standards should be matched with the samples in terms of their acid content to avoid nebulisation differences, which can otherwise lead to inaccurate results.
4.6.6 Quality Control The EPA CLP (Contract Lab Program) QC protocol is very comprehensive and has become the model for many other analytical procedures, even those not associated with water analysis. After the instrument has been calibrated, it must be checked with an ICV (Initial Calibration Verification) solution and calibration blank. These are repeated after every 10 samples and at the end of the analysis. Each analyte result in the ICV must be within 95 - 105% of its expected value or the instrument must be recalibrated and all samples since the last successful check re-run. The values for the blank must be less than the CRDL (Contract Required Detection Limit) shown in Table 4-7 (see p. 119). Other check standards for interelement correction are run periodically and it is important that a QC Reference sample is run at least once per analysis as a check on the accuracy [47,48]. There are now a number of SRMs (Standard Reference Materials) which are suitable for water analysis, including the NRC (National Research Council, Canada) series CASS-2, NASS-3 and SLEW-1 for saline waters and SLRS-2 for river water. Finally typical results for two EPA Quality Control standards are shown in Table 4-8 (see p. 119) to indicate the sort of performance that is possible using ICP-OES.
1 18
4 Application of ICP- OES Techniques in Environineninl OC
Table 4-6. Recommended analyte combinations to be used in multielement calibration standards. (Taken from USEPA Water Method 200.7 [ 6 ] ) . Calibration 2
Calibration 1 Analyte
Ag As B Ba Ca Cd cu Mn Sb Se
Concentration (ILg mL - '1
Analyte
0.5 10.0 2.0 1 .0 10.0 2.0 2.0 2.0 5.0 5.0
K Li Mo Na Sr
Concentration (WLg
mL-9
20.0 5 .0 10.0 10.0 1 .0
Calibration 3 Analyte
Concentration
(am L - 9 co V P
2.0 2.0 10.0
Calibration 5
Calibration 4 Analyte
Concent ration (w mL-'1
Analyte
Concentration (ILg m L - 9
A1 Cr Hg SiO, Sn Zn
10.0 5.0 2.0 10.0 4.0 5 .O
Be Fe Mg Ni Pb TI
1 .0 10.0 10.0 2.0
10.0
5.0
4.7 Airborne Particulate Analysis The application of ICP-OES to the environmental aspects of air quality have mostly been associated with the analysis of airborne particulates. Cas-phase pollutant5 have been analysed after trapping in reactive solutions but this type of analysis i, much less common. There are many sources of airborne particulate, such as coal-fired power stations, waste incineration, metals smelting and refining to name but a few. Increased use of urban waste incineration has caused a much higher awareness of the problems caused
4.7 Airborne Particulate Analysi.5
1 I9
Table 4-7. CRDLs (Contract Required Detection Limit) values required under the USEPA CLP (Contract Laboratory Program) for water analysis. (Taken from the USEPA Water Method 200.7 [6]). Element
Ag A1 As Ba Be Ca Cd co Cr cu Fe a
CRDL (PLg L - ' )
Element
10 200 loa 200 5 5000 5 50 10 25 100
K
CRDL (Pg L
'1
5000 5000 15 5000 40 3= 60 a 5a loa 50 20
Mg Mn Na Ni Pb Sb Se TI V Zn
Generally below ICP-OES detection limits, except with hydride methods or ultrasonic nebulisation.
Table 4-8. Results obtained for two EPA Water Quality Control samples using ICP-OES. Values are reported as mg L I. Sample 1 Element
Wavelength (nm)
Measured value
Certified range
Ba Ca K Mg Mo Na
455.404 315.887 766.490 279.553 281.615 589.592
1.20 0.52 0.24 0.1 1 1.68 0.07
0.88 - 1.31 0.42 - 0.58 0.14 - 0.26 0.08-0.16 1.24 - 2.56 0.07 - 0. I 9
Sample 2 Element
Wavelength (nm)
Measured value
Certified range
Ba Ca K Mg Mo Na
455.404 31 5.887 766.490 279.553 281.615 589.592
10.1 7.49 1.88 0.55 11.3 1.46
9.4- 11.4 6.45 - 8.57 1.14-1.90 0.40 - 0.60 9.90- 13.8 1.30-1.78
120
4 Application of ICP- OES Techniques in Environnienral QC
to the environment, on one hand solid wastes are reduced by up to 9Ou;o but on the other hand new waste products like dust and fumes are created. I t is usual for incineration plants to carry out extensive COSHH (Control of Substances Hazardous to Health) testing under the terms of their licences. This involves on-site testing of toxic fumes and off-site monitoring of waters and vegetation.
4.7.1 Sample Collection Samples are commonly collected by sucking known volumes of air through a glass fibre or membrane filter. The volume of air used depends on the contamination Icrel and the elements being measured. Table 4-9 shows the TLV (Threshold Limit Value\), typical recommended collection volumes and detection limits for a number of metals. A number of portable, battery-powered samplers are available for personal monitoring at the workplace. This type of sampler is also used for sampling at siter rf here no mains electricity is available. Two types of filter are used but it is preferable to use the membrane type because they have lower, and more constant, levels of contaminants and their pore size ir more uniform [49]. Simple digestion procedures using oxidising acid mixtures can dissolve the filter and its contents. Typical filters used in personal samplers are 0.8 pm pore size and 25 mm in diameter. They can pass about 15 - 20 L min- air at a pressure differential as low as 70 kPa. These high flowrates mean shorter sampling times can be employed or superior detection limits obtained with reasonable collection times.
'
Table 4-9.Industrial Health TLV (Threshold Limit Values) for metal fumes. Also shown are llie typical volumes of air sampled to permit the listed detection limits to be obtained using ICP-OES analysis.a (Reproduced with permission from Atomic Spectroscopy 1988, 9 (5), 154). -~~~~
Metal
TLV
(ms m-3) A1 Be Cd cu Cr Fe
5 .O 0.002 0.05 0.2 0.5 5 .O
Mn Ni Pb Ti Zn
1 1
a
.o .o
0.15
5.0 5.0
Air volume sampled (L)
Detection limit in the (mg m-')
180 500 140 90 90 1 50 22 90 180 100 100
0.006 0.0004 0.004 0.01 I 0.01 I 0.007 0.014 0.005 0.015 0.01 0.005
ail
TLV values are average exposures for an 8-h day or a 40-h working week, defined by the American Conference of Governmental Industrial Hygienists in 1977.
4.7 Airborne Particulate Analysis
121
An alternative for use at large installations, such as power stations or incinerator chimneys, is the cascade impactor or electrostatic filter. Two types of sample are, therefore, generally available - particles held on a membrane filter or loose powder taken from a cascade impactor or the like. Solid samples cannot, in general, be analysed directly by ICP- OES, although experimental systems do exist, so that dissolution of the samples is necessary.
4.7.2 Sample Preparation Where samples are mainly metals, as in welding fume, and collected using personal samplers, the usual dissolution procedure is to use simple acid digestion. The metals involved define the procedure used. For Cr, Cu, Fe, Mn, Ni, Zn, Ag, Cd, Pb, Au, B, Ba, Bi, Ca, Mg, Sr and V the filters are simply digested in 10mL concentrated acid. For Al, Be, Co, Mo and Ti the filters are digested in 10 mL of a (1 + 1) mixture of nitric and hydrochloric acids. After digestion the samples are heated at 140 "C to near dryness and then re-solubilised to a final volume of 10 mL with 0.1 To v/v nitric acid [50]. For samples containing high ash contents, dissolution schemes similar to those developed for the analysis of siliceous materials are generally used. These can be acidbased (generally involving hydrofluoric acid or mixtures of this with boric acid) or lithium metaborate fusions. Acid-based schemes are preferred because they minimise the dissolved solids content of the final solution and permit the use of pneumatic nebulisers without fear of blockages [51, 521. Increasingly the use of microwave digestion is becoming the normal technique [53, 541. By using a sealed vessel and heating it in a microwave oven, it is possible to shorten the dissolution time considerably. Typically a sample treated by the traditional pressure digestion method requires about 50 min of heating whereas the microwave procedure takes about 6 min. The EPA has now certified some methods with microwave dissolution [55]. Commercial microwave digestion systems vary in their complexity and capability, but are designed to cope with the acidic environment and employ over-pressure safety devices. Originally household units were often used but they rapidly corrode due to the lack of resistant plastic liners and fume extraction accessories. A typical dissolution scheme for a total analysis is as follows: Transfer 0.5 g of powdered sample (ground to pass a 100 mesh 0,15 mm sieve) to the digestion vessel and add 7 mL H F and 4 mL aqua regia. Seal the vessel and heat for 4 min at 100% power (650 W) followed by 4 min at 75 Yo power. Cool to room temperature, open vessels and add 50 mL of 2% m/v boric acid to complex any remaining HF. Heat for a further 5 min at 100070power to redissolve any insoluble fluorides. Cool, transfer to a volumetric flask and make up to volume. The advantage of complexing any excess H F in this way is that the normal quartz torch and spraychamber may be used for analysis without risk of corrosive attack. If any uncomplexed H F remains, a plastic or inert spraychamber and a torch fitted with a ceramic injector will need to be used.
122
4 Application of ICP-OES
Techniques in Environnienfril QC
4.7.3 Analysis Conditions and Results The results shown in Table 4-10 are for an SRM (NIST 1648 Urban particulate) using aqua regia dissolution of the sample. ICP-OES is ideally suited to this type of analp sis, where concentration levels for the various elements may vary widely from U h to yg L-’. Its superior calibration range means that in most cases only a single sample solution need be prepared, without the dilutions necessary in AAS (flame or furnace). A full method blank is vital and standards should be matrix-matched. Spectral interferences may be a problem where very high concentrations of transition metals like W, Fe, etc., are involved. Useful SRMs for use as method checks are NIST SRM 1648 Urban Particulate, BCR 176 Incineration Ash and a series of Coal Fly Ash NIST SRM, 1633, 2689, 2690 and 2691.
Table4-10. Results obtained for the NIST SRM 1648 Urban Par-
ticulate using ICP-OES. Sample dissolution by acid digestion. Values are reported as weight Vo or Fg g - ’ as appropriate, with standard deviations for the certified material. (Reproduced with permission from Atomic Spectroscopy 1988, 9 (6), 191). Element
Measured valuea
Certified range”
~~
A1 Ca Fe Mg Pb Zn a
3.60 0.6 3.32 0.1 0.58 0.434
3.40
? 0.1 1
3.91 k0.10 0.8 0.655 k 0.008 0.476 k0.014
All results in weight 9‘0
-
-
~
~
~
~
~
Element
Measured value
Certified range
Cd
11.3 603 666 84 112
75i 7 609 i 21 860 82i 3 140+ 3
cu Mn Ni V
All results in Fg g-’
~
4.8 Analysis of Soils, Sludges and Sediments
123
4.8 Analysis of Soils, Sludges and Sediments When analysing soil samples and sediments, it is often only the available metals (ie, those elements that can be leached out by percolating water) that are to be analysed. Elements which cannot be solubilised by the sort of acids used are most unlikely to become toxic hazards in the normal course of events. A serious source of long-term pollution results from the practice of disposing of urban solid wastes by burying them in landfill sites or by the use of sewage sludges as fertilisers for agricultural purposes. Toxic elements in these materials are leached out slowly with time by the action of acidic rainwater percolating down through the layers. The end result can be severe contamination of underlying soil and groundwater, with the consequent risks to the purity of drinking water supplies extracted from aquifers [56]. The metals found in the groundwater depend on a number of factors. The ultimate aim of the sample preparation is to simulate the natural leaching action of rain to give an estimate of the available metals content. This is normally done by taking a known weight of waste and leaching it with a dilute acid, typically 0.5 M acetic acid or aqua regia [57]. Alkaline extractants, such as NH4HC03-DTPA [58], have also been used.
4.8.1 Sample Collection Soil and sediment analysis are good examples of the problems of taking truly representative samples. Sufficient material must be taken to be indicative of the whole site being sampled and then it must be sub-divided to provide the final representative sample. For most cases the aim should be to provide at least 500 g of dried material. Samples should be placed in thick gauge polythene bags which have not be used for any other purpose. Paper bags should not be used as they can contain significant quantities of B and Zn as contaminants [ 5 ] .
4.8.2 Sample Preparation Soil samples are dried to constant weight at about 110°C. It may take as long as 2 days to achieve this. The dried sample is ground in a metal-free mortar and pestle to pass a 100 mesh (0,15 mm) sieve. Each sample powder is shaken thoroughly before a known weight of sample is taken. For a total analysis, conventional pressure dissolution [59- 631 or microwave attack (as described in Section 4.7.2) are used [64-671. Leaching is generally carried out shaking several grammes of air-dried soil with 200 mL of 0.5 M acetic acid. After the initial gassing has subsided, shake for a further 30 min and then filter through a Whatman No. 40 filter or its equivalent.
124
4 Application of ICP- OES Techniques in Environmental QC
Leaching can also be carried out using aqua regia attack. Typically 2 g of sample is digested with 30mL aqua regia by boiling to low volume. After cooling, 25 mL concentrated nitric acid is added and boiled to dryness in order to complete the destruction of any organic material present. The residue is dissolved in distilled water and analysed against standards which have been through a similar treatment.
4.8.3 Analytical Conditions and Results The results shown in Table 4-1 1 were taken from the analysis of a waste sample and the underlying soil levels, as measured by ICP-OES after acetic acid leach. This demonstrates the linearity advantage of the ICP- OES technique because all the elements were determined in a single solution without the need for further dilution. The second set of results in Table 4-12 are for a river sediment using the aqua regia treatment. These show the level of precision attainable with ICP-OES, typically 1 To or better except for the low level molybdenum result. Sediments are more difficult samples to analyse because they generally contain high levels of calcium, aluminium and magnesium (responsible for large and variable background shifts) and iron (potentially a large number of spectral interferences at trace element levels). Background correction and spectral line choice are, therefore, important factors here. Guidelines are laid down by various authorities which specify the maximum permitted levels of toxic elements in soils and sludges, when used for agricultural purposes. Current EEC and UK Department of the Environment levels are shown in Table 4-13. There are a considerable number of SRMs available for checking soils and sediment analysis, together with a small number of sewage sample. Table 4-14 (see p. 126) lists the samples and suppliers. Table 4-11. Results obtained by ICP-OES for a 0.5 M acetic acid leach of soil samples taken from the surface layers of a landfill site and from the underlying soil base material. All results in wg g - ' . (Reproduced with permission from Spectrochim Acta 1988, 4 3 B , 378). Element
Landfill material
Soil base material
A1 As Cd Co Cr cu Fe Mn Ni Pb V Zn
12000 0.1 0.8 8.4 37 430 36 000 3 40 27 140 5.5 420
8 800 18 0.3 3.8 18 79 14000 200 24 110 4.9 95
4.9 Plant and Biological Sample Analysis
125
Table 4-12. Results obtained using ICP-OES for a river sediment reference sample digested with aqua regia. All values are shown as mgL-'.
Element
Wavelength (nm)
Known value
Measured value
A1 Ba Ca Cd Cr cu Fe Mn
308.215 455.404 315.887 214.438 205.552 324.7 54 259.940 257.610 202.030 23 1.604 213.618 220.353 334.941 309.31 1 206.200
47.0 2.50 79. I 0.90 55.4 3.91 44 1 1.87 0.15 0.53 137 6.42 3.50 0.97 10.6
49.4 2.92 75.7 0.88 57.3 3.98 444 1.84 0.19 0.56 135 6.38 3.42 1.10 10.8
Mo Ni P Pb Ti V Zn
RSD To
0.2 0.5 0.6 0.3 0.9 0.3 0.3 0.1 10
2.4 0.6 0.6 2.3 0.7 0.7
Table4-13. Current EEC and UK Department of the Environment Guidelines for the maximum permitted levels of 10 regulated elements in sewage sludges used for agricultural purposes. All values as mg kg-'. Levels for soils are shown for comparison.
Element
Limit value in sludge
Limit value in soils
As Cd Cr cu Hg Mo Ni Pb
1000 16 300 750
Se Zn
2500
50 3 150 80" 1 4 50" 300 3 200 a
" Values
dependent on pH.
4.9 Plant and Biological Sample Analysis Plant materials are often used as indicators of both airborne and sub-soil contamination [ 5 ] . Samples may be root crops, like turnip or beetroot, or leaves of plants, such as lettuce or cabbage. In addition, biological samples like blood, urine and hair are
126
4 Application of ICP-OES
Techniques in Environniental QC
Table 4-14. Certified reference materials for soils, sludges and sediment analysis.
BCR (Community Bureau of Reference) /8I/ CRM CRM CRM CRM CRM CRM CRM
141 Calcareous Loam Soil 143 Sewage Sludge - amended Soil 144 Sewage Sludge - domestic 146 Sewage Sludge - industrial 277 Estuarine Sediment 280 Lake Sediment 320 River Sediment
NIST (National Institute of Standards & Technology) 1821 SRM 1646 Estuarine sediment SRM 2704 Buffalo River Sediment
NRCC (National Research Council, Canada) /83J MESS-I Estuarine Sediment BCSS-1 Coastal Marine Sediment PACS-1 Harbour Marine Sediment NIES (National Institute fo r Environmental Studies) 1841 50-02 Pond Sediment
IAEA (International Atomic Energy Agency) /85] SOIL-7 Soil SL-1 Lake Sediment SDM2/TM Marine Sediment
NRCCRM (National Research Centre, China) /86/ (283-01 River Sediment C83-02 Tibet Soil C73-01 Stream Sediment C83-03 Polluted Farmland C73-02 Stream Sediment C73-03 Stream Sediment C73-04 Stream Sediment C73-05 Stream Sediment C73-06 Stream Sediment C73-07 Stream Sediment C73-08 Stream Sediment C73-09 Stream Sediment C73-10 Stream Sediment C73-I I Stream Sediment C73-12 Stream Sediment C74-01 C74-02 C74-03 C74-04 C74-05 C74-06 C74-07 C74-08
Soil Soil Soil Soil Soil Soil Soil Soil
4.9 Plant and Biological Sample Analysis
127
Table 4-15. Normal levels of major and trace elements in plant and biological materials. Element
A1 As B Ba Bi Ca Cd
co Cr
cu Fe Hg K Li Mg Mn Na Ni P Pb Se Si Ti V Zn a
Biological materials a
Plant materials
Urine
Leaf
Serum 2-6
60 15 - 120 709-919 89 20
92 - 109'
2.3 - 85 6.3 - 110
0.5 - 4.6
60
0.64 - 9 98 0.7 - 1.5 17-300 180 4-114
3 - 300
3.8- 13.5
Root
0.04 - 0.3 800 - 1250 870 - 1870 160-211 8 I9 - 27.5 ' 0.4 - 1 .O 3130-3370' 0.6 - 5.0 1 15 - 163' 16-130
346- I019
241 - 578
0.008 - 0.041
0.005 - 0.033
<2-5 5.6-21
<2- <4 5.1 - 9
103 - 402 1.0- 8.1 174- 1626
98 269 1.1 -4.5 404 - 1552 ~
0.005 - <0.06
0.004 - 0.039
1.9 - 7.5
1.6 - 4.2
2.5 - 10' 10
2-20 193 - 2040
800 - 1200
Concentrations as ng mL-l. Concentrations as pg g-'. Values as pg mL-l.
analysed as a means of estimating workers exposure to toxic materials [68-711. Many trace metals in waters are concentrated by fish, shellfish particularly, and plankton and these organisms are often used as an indirect method of measuring water quality [72, 731. Table 4-15 indicates typical concentrations for major and trace elements in plant and biological materials.
4.9.1 Sample Collection The actual sampling procedures for plant materials will vary considerably, depending on whether it is a root or foliage sample. Again, taking a representative sample is important. Foliage should be cut, not plucked or torn, at least 3 cm above the soil level to avoid contamination. A composite sample from at least 25 plants should be
128
4 Application of ICP-OES Techniques in Environmental QC
taken and it should not weigh less than 1 kg. Materials like roots should be washed free of soil and about 200 g of sample taken. These samples are air-dried at a temperature below 80°C and ground to pass a 17 mesh (1 mm) sieve [ 5 ] . Collecting samples of a biological nature is fraught with many problems, mainly due to contamination from the surgical instruments used to obtain them. These problems have been well-documented [74] and the appropriate precautions necessary are now well-known.
4.9.2 Sample Preparation Sample preparation of biological materials consists of an ashing step, to destroy the large carbon content, and a dissolution step to produce a solution suitable for nebulisation. Sometimes the two procedures are carried out at the same time, as in the wet-ashing technique. There are two main methods employed for this task, dry-ashing (where the material is subjected to elevated temperatures in muffle furnaces or low temperatures with oxygen ashers) and wet-ashing (where oxidising acid mixtures and elevated temperatures are mainly used). Much has been written on this subject but the well-known texts from Gorsuch [75] and Bock [76] are recommended to the reader. Dry-ashing has a number of advantages: it does not require expensive equipment or hazardous reagents, can deal with large batches without continuous attention and can be relatively quick. It also has a number of disadvantages: with volatile elements (such as Hg, As, Cd and Pb) variable losses occur due to volatilisation. Migration of elements to the crucible walls and the formation of insoluble residues are two other problems encountered with simple dry-ashing. The latter two problems can be overcome or minimised by the use of ashing aids. These are normally the nitrates of calcium or magnesium which are mixed with the sample before ashing begins. They can assist in the retention of the volatile species, ensure the ash remains soluble after the ashing process and help in the oxidation process. A typical scheme for the dry ashing of plant materials is as follows [77]: Transfer 2 g of dried sample to a platinum or silica crucible and place in a cool muffle furnace. Increase the temperature to 450°C and hold this temperature until a small whitish-grey ash residue remains. This may take up to 24 h in some cases. When finished, the crucible is removed from the furnace and allowed to cool (covered with a watch glass to prevent airborne contamination). Add 10 mL of 6M hydrochloric acid to the crucible, taking care to avoid losses due to foaming. Remove the watch glass and wash this into the crucible and then evaporate the contents to dryness, slowly and without boiling. Moisten the residue with 2 mL of concentrated hydrochloric acid, cover with a watch glass and boil gently for 2 min. Add 10 mL of deionised or distilled water and boil again. Cool and transfer the contents to a 50 mL volumetric flask, along with washings from the watch glass. Make up to volume then filter through a Whatman 541 filter paper, reject the first few mL of filtrate and then retain the rest for analysis.
4.9 Plant rind Biological Sample Analysis
129
Dry-ashing with a low temperature oxygen plasma asher has been used in recent years but is generally not as convenient. Oxidation only proceeds on the surface of the sample so that it must be agitated frequently to expose fresh material and it does not work well with large particles [78]. Wet-ashing is used much more often, because of its versatility, but care must be exercised due to the hazardous nature of some of the procedures. Generally there are two main procedures: (1) those based on nitric and sulphuric acid mixtures and (2) those based on nitric and perchloric acid mixtures. Both methods have their advantages but the perchloric acid route is the more hazardous. The basic mechanism involves an initial mild oxidation with nitric acid, followed by a more vigorous reaction with hot sulphuric or perchloric acid. Care must be taken that the exothermic reaction does not get out of control and these procedures should not be used with samples containing high fat or oil contents. The recommended techniques published by the Analytical Committees of the Royal Society of Chemistry (UK) [79, 801 provide excellent advice on this subject. A typical procedure is as follows: Transfer 0.2 g of dried material to a 100 mL long-necked Kjeldahl flask followed by the addition of 1 mL concentrated sulphuric acid, 5 mL concentrated nitric acid and 1 mL concentrated perchloric acid. Heat gently until the initial reaction subsides, then gradually increase the heat until the mixture boils. Carry on digesting the material until white fumes of sulphuric acid appear and then for a further 15 min. Cool and transfer to a 100 mL volumetric flask and make up to volume with distilled or deionised water. The advantage of this particular procedure is that the residues are kept moist by the sulphuric acid until the perchloric acid has evaporated, thus minimising the risk of an explosion. Final solutions may not be colourless, indicating some residual organic material remains. This is not normally a problem unless sufficient remains to cause changes to solution viscosity and/or surface tension. However, if some form of pre-concentration exercise is to follow the digestion, such solutions must be further treated. In addition, grass samples (and some other herbage samples) often will not completely solubilise because a precipitate of fine silica particles is produced. These have to be removed by centrifuging or filtration before analysis in order to prevent nebuliser blockage. High purity acids are required to reduce the risk of contaminating the samples during the dissolution stage. Care must be taken when wet-ashing samples containing high alkali metal concentrations with perchloric acid, as they can form relatively insoluble perchlorates. Similarly sulphuric acid can form insoluble sulphates with high concentrations of Ca, Ba, Sr and Pb. Finally, volatile elements such as As, Se, Hg, B and others can be lost during the high temperature phases of digestion and the appropriate precautions must be taken.
4.9.3 Analytical Conditions and Results These types of sample place a premium on the nebuliser and spray chamber, this being the area most at risk due to the high dissolved solids levels and the high acid con-
130
4 Applicution of ICP- OES Techniques in Environmental QC
Table 4-16. Results of the determination of As and Se by Hydride/ICP-OES in plant5 and biological materials. Sample
Element (units)
Pine needles NIST 1575
As
(u 8 - 9
0.21
0.19
Spinach NIST 1570
As (Pi4 g - 9
0.15
0.13
Wheat flour NIST 1567
Se
1.1
0.95
(r-lsg - ' )
Certified value
Mcawrcd calue
Blood serum
Se (ng m L - ' )
-
I10
Blood serum + 100 ng mL-'
Se (ng m L - ' )
-
205 reco\ ery)
centrations resulting from the digestion procedures. In addition, acids like sulphuric acid cause marked increases in solution viscosity and, therefore, the calibration standards must be matrix-matched to the samples to obtain accurate results. Internal standards may be required to remove variable viscosity effects if it is impossible to matrix-match adequately each of the solutions. The presence of high levels of Ca and A1 can cause large background shifts which must be corrected for using one or two background points. Provided that the dissolved solids levels are kept below 270,it is likely that ultrasonic nebulisation can help improve sensitivity for many of the marginal concentration levels found with biological samples. For example, normal concentration levels for Ca, Na, Mg, P, Ba, Cu, Fe, Zn and Si in serum can be determined without preconcentration at a precision of about 1%. Others such as Al, Cd, Li, Pb, Sr, Ti and Zr are close to the detection limit and give precisions in the 5 - 10% range. Hydride analysis is also a very useful technique for the analysis of biological materials and results are shown in Table 4-1 6 . There are a range of SRMs for this type of analysis, well-known examples being the NIST SRMs 1573 Tomato leaves, 1575 Pine needles and 1577 Bovine liver.
References [ l ] Mavrodineanu, R., Boiteux, H., Flume Spectroscopy, New York: John Wiley and Sons, 1965. [2] Boumans, P. W. J. M., Inductively Coupled Plasma - Emission Spectroscopy, Part I , Nev
York: John Wiley and Sons, 1987. [3] Montaser, A., Golightly, D. W., Inductively Coupled Plasmas in Ancilytical Atomic Spectrometry, New York: VCH Publishers Inc., 1987. [4] Molek Zwetolitz, J., Environ Lab 1990, 2 (2), 32, 34-37. [ 5 ] Berrow, M. L., Anal Proc (London) 1988, 25 (4), 1 1 6- 1 1 8.
References
13 1
[6] Method 200.7 Determination of Metals and Trace Elements in Water and Wastes by ICP-OES. ICP Information Newsletter 1992, 18 (3), 147 169. [7] Annual Book of ASTM Standards, Volume 11.01, 1992. [8] Hoffmann, H.-J., GZT Fachz Lab 1988, 32 (3), 180- 182, 184. [9] British Standard, BS6068: Section 6.5: 1991. [ 101 British Standard, BS6068: Section 6.6: 1991. [ I I] McLeod, C. W., 1992 Winter Conference on Plasma Spectrocheinistry, San Diego, (A, January 1992, Abstracts, Amherst MA: ICP Information Newsletter, 1992. [I21 Hartenstein, S. D., Christian, G. D., Ruzicka, J., Can J Spectrosc 1985, 30 (6), 144- 148. [I31 Caroli, S., Alimonti, A,, Petrucci, F., Horvath, Z., Anal Chim Acta 1991, 248 ( I ) , 241 -249. [I41 Wuensch G., Knobloch, S., Luck, J., Bloedorn, W., Spectrochim Acta Part B 1992, 47B (l), 199 - 202. [I51 Porta, V., Sarzanini, C., Abollino, O., Mentasti, E., Carlini, E., J Anal A f Specfrom 1992, 7 (2), 19 - 22. [I61 Off J Eur Comm, 1980, L 229, 11 -29. 1171 Thompson, M., Pahlavanpour, B., Thorne, L.T., Water Research 1981, IS, 407-41 1 . [I81 Cave, M. R., Green, K.A., J Anal At Spectrom 1989, 4 (2), 223-225. [I91 Nygaard, D.D., Bulman, F., Spectroscopy (Int) 1990, 2 (21, 44-47. [20] Anderson, J., A t Spectrosc 1992, 13 (2), 99- 104. [21] Thompson, M., Ramsey, M., Pahlavanpour, B., Analyst 1982, 107, 1330- 1334. [22] Gorlach, U., Boutron, C.F., Anal Chim Acta 1990, 236 (2), 391 -398. [23] Baucom, E. I., Ferguson, R. B., Wallace R. M., Symposium of Hydrogeochem. and Streamsediment Reconnaissance f o r Uranium in the US, US Dept of Energy, Grand Junction, Colorado. [24] Dupont, V., Auger, Y., Jeandel, C., Wartel, M., Anal Chem 1991, 63 (S), 520-522. [25] Cresser, M., Solvent extraction in Flame Spectroscopic Analysis, London: Butterworths, 1978. [26] McLeod, C. W., Otsuki, A., Okamoto, K., Haraguchi, H., Fuwa, K., Analyst 1981, 106, 41 9-428. [27] Sugimae, A., Anal Chim Acta 1980, 121, 331 -336. [28] Van Berkel, W. W., Overbosch, A. W., Feenstra, G., Maessen, F. J. M. J., J Anal Af Spectroni, 1988, 3 (I), 249-251. [29] Cheng, C., Akagi, T., Haraguchi, H., Anal Chim Acta 1987, 198, 173- 181. [30] Vermeiren, K., Vandecasteele, C., Dams, R., Analyst 1990, 115, 17-22. [31] Berman, S.S., McLaren, J.W., Willie, S.N., Anal Chem 1980, 52, 488-492. [32] Sturgeon, R. E., Berman, S. S., Desaulniers, J.A. H., Mykytiuk, A. P., McLaren, J. W., Russell, D. S., Anal Chem 1980, 52, 1585 - 1588. [33] Watanabe, H., Goto, K., Taguchi, S., McLaren, J. W., Berman, S. S., Russell, D. S., Anal Ckem 1981, 53, 738-739. [34] Sturgeon, R. E., Berman, S. S., Willie, S. N.,Tn/anta 1982, 29, 167. [35] Porta, V., Sarzanini, C., Mentasti, E., Mikrochim Acta 1989, III (3-6), 247-255. [36] Hartenstein, S.D., Ruzicka, J., Christian, G.D., Anal Chem 1985, 57, 21 -25. [37] Hirata, S., Umezaki, Y., Ikeda, M., Ann/ Chem 1986, 58, 2602-2606. [38] Pereiro Garcia, M. R., Diaz Garcia, M. E., Sanz Medel, A., J Anal A t Spectrom 1987, 2 (lo), 699 - 703. [39] McLeod, C. W., Zhang, Y., Cook, I., Cox, A., Date, A. R., Cheung, Y. Y., JRes Natl Bur Stand (US), 1988, 93 (3), 462-464. [40] Knapp, G., Muller, K., Strunz, M., Wegscheider, W., J Anal A t Spectrom 1987, 2 (9), 61 1-614. [41] Prakash, N., Csanady, G., Michaelis, M. R. A,, Knapp, G., Mikrochim Acta 1989, 111 (3-6), 257-265. [42] Hiraide, M., Ito, T., Baba, M., Kawaguchi, H., Mizuike, A,, Anal Chem 1980, 52, 804-807. [43] Shan, X., Tie, J., Xie, G., J Anal A t Spectrom 1988, 3 (I), 259-263. -
1 32 [44] [45] (461 [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] 1591 [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [SO]
[8 11 [82] 1831 [84] [85] [86]
4 ICP- OES Techniques in Environmental QC
Quigley, M. N., Vernon, F., Anal Proc (London) 1991, 28, 175 - 176. Anderson, P., Ingri, J., Water Res 1991, 25 (51, 617-620. Kocherlakota, N., Analyst (London) 1992, 117, 401 -406. Byrne, A.R., Analyst (London) 1992, 117, 251-258. Maier, E.A., Anal Proc (London) 1990, 27 (lo), 269-270. Dukla, J. J., Risby, T.H., Anal Chem 1976, 48, 640A-653A. Thomas, T. C., Jehl, L. J., A t Spectrosc 1988, 9 (51, 154- 156. Karstensen, K.H., Lund, W., J A n a l A t Spectrom 1989, 4 (4), 357-359. Infante, R. N., Acosta, 1. L., A t Spectrosc 1988, 9 (6), 191 - 194. Bettinelli, M., Baroni, U., A t Spectrosc 1988, 9 (5), 157-165. Paudyn, A. M., Smith, R. G., J Anal A t Spectrosc 1990, 5 (61, 523 - 529. Tatro, ME., Spectrosc Int 1990, 2 (5), 16, 19-20. Castagnoli, O., Zavattiero, E., Musmeci, L., Alimonti, A., Caroli, S., Spectrosc Europe 1992, 4 (5), 10-16. Blank, F. T., BLGG, Holland. Personal communication. Boon, D.Y., Soltanpour, P.N., Commun Soil Sci Plant Anal 1991, 22 IS), 369-378. Williams, M.C., Stallings, E.A., Foreman, T.M., Gladney, E.S., At Spectrosc 1988, 9 14), 110- 114. Hawke, D.J., Lloyd, A., Analyst (London) 1988, 113, 413-417. McQuaker, N.R., Kluckner, P.D., Chang, G.N., Anal Chem 1979, 51, 888-895. Kimbrough, D. E., Wakakuwa, J. R., Environ Sci Techno1 1989, 23 (7), 898-900. Sellers, C., J Assoc OffAnal Chem 1990, 73 (3), 476-478. Kammin, W.R., Brandt, M. J., Spectrosc (Eugene Oreg) 1989, 4 (3), 49-50, 52, 5 5 . Bettinelli, M., Baroni, U., Pastorelli, N., Anal Chim Acfa 1989, 225 (l), 159- 174. Millward, C.G., Kluckner, P. D., J Anal A t Spectrom 1989, 4 (81, 709-713. Nieuwenhuize, J., Poky-Vos, C. H., Van den Akker, A. H., Van Delft, W., Anulyst (London) 1991, 116, 347-351. Caroli, S., Spectrochim Acta 1988, 43B (4/5), 371 -380. Caroli, S., Alimonti, A,, Delle Femmine, P., Petrucci, F., Senofonte, O., Violante, N., Menditto, A,, Morisi, G., Menotti, A., Falconieri, P., J Anal A t Spectrom 1992, 7 (9), 859-864. Schramel, P., Spectrochim Acta 1988, 43B (a), 881 -896 Lo, F.B., Arai, D.K., A m Ind Hyg Assoc J 1989, 50 (5), 245-251. Grant, W.A., Ellis, P. C., J Anal A t Spectrom 1988, 3 (6), 815-820. Nakashima, S., Sturgeon, R.E., Willie, S.N., Berman, S.S., Analysf (London) 1988, 113 (I), 159- 163. Versiek, J., Speecke, A,, Hoste, J., Barbier, F., Clin Chem 1973, 19, 473. Gorsuch, T. T., The Destruction of Organic Matter, Oxford: Pergamon Press, 1970. Bock, R., Handbook of Decomposition Methods in Analytical Chemistry, Glasgow: Blackie, 1979. M. A. F. F. Technical Bulletin 27, The Analysis of Agricultural Muferials, London: HMSO, 973. Munter, R. C., Grande, R.A., Ahn, P.C., ICP Info Newsletter 1979, 5 (7), 368-383. Analytical Methods Committee (Royal Society of Chemistry, UK), Analysf (London) 1959, 84, 214. Analytical Methods Committee (Royal Society of Chemistry, UK), Analysr (London) 1976, 101, 62-66. Community Bureau of Reference (BCR) Commission of European Communities, Directorate General of Science, Research and Development, 200 rue de la Loi, B-1049 Brussels, Belgium. Office of Standard Reference Materials, Building 202, Room 204, National Institute of Standards and Technology, Gaithersburg, MD 20878-9950, USA. National Research Council of Canada, Institute of Environmental Chemistry, Marine Analytical Chemistry Standards Program, Montreal Road, Ottawa, Ontario KIA OR6, Canada. National Institute for Environmental Studies Japan Environment Agency, Yatabe-machi, Tsukuba, Ibaraki 305, Japan. International Atomic Energy Agency, Waggramerstrasse 5, PO Box 100, A- 1400 Vienna, Austria. Office of CRMs, Center for CRMs, National Institute of Metrology, Heping Li, Beijing China.
5 Practical Aspects of Monitoring Volatile Organics in Air Elizabeth A. Woolfenden
5.1 Introduction Environmental air pollution largely remains a series of unanswered questions: 0 0 0 0
What pollutants are present? At what concentration? Where are they coming from? What harm do they do?
Reliable answers to these questions can only be obtained by improving the quality and quantity of air monitoring data collected. It is only with more, reliable atmospheric concentration information that the impact of pollution in any given atmosphere can be correctly evaluated and cost-effectively controlled. With regard to volatile organic compounds (VOCs) in air, the specific measurements required include: (1) emission profiles from each potential pollution source, (2) detailed information on background (natural) levels and, (3) accurate VOC concentrations in the atmosphere concerned. However, monitoring VOCs in air is not a simple analytical procedure and requires the following factors to be taken into consideration:
of VOCs are often unknown within a factor of 10 and range from 100 ppm at emission points and in some workplace atmospheres to low ppt levels in rural air. - Most atmospheres contain a complex mixture of hundreds, possibly thousands, of different species. - Organic compounds have a wide range of toxicity. - The anthropogencity or toxicity of most organic compounds is currently very poorly understood. - In many cases, trace level toxic target analytes are present in the same atmosphere as relatively high concentrations of less toxic components. - Atmospheric concentrations
There has already been a rapid growth in organic air pollution monitoring in response to increased legislative and public pressure specifically affecting workplace air, industrial/mobile emissions and urban air quality. Many of the methods used to date have employed simple on-line detectors or continuous spectroscopic tech-
134
5 Practiccrl Aspects of Monifortng Volatile Organics in Air
niques. Such methods are typically restricted to the measurement of components ‘11 concentrations above 1 ppm. They are also limited to either producing cn\elopc 111formation or to the monitoring of a small number of target analytes (typically lcss than 10) that must be present in the atmosphere at concentrations above thoje 01 other interfering VOCs. Low cost, ‘direct reading’ methods such as indicaLor tubcs are also restricted to relatively high concentration samples and can only bc used for one individual species or class of compound. None of these approaches take sufficient account of the complexity of atrnospheric samples and provide little practical information on the nature and extent of an) potential environmental hazard. Given the number and toxicity range of VOCs which may be present in an atmosphere, most new standard methods require speciated monitoring, ie, monitoring the concentration of each individual organic compound [ I , 21. Such speciated information can only be obtained using non-continuous, high resolution capillary gas chromatography (GC), ideally in conjunction with mars spectrometer (MS) detection.
Table 5-1. Mass of an analyte ‘x’collected from I , 2 or 10 L air samples at different atrnosphet ic concentrations (assuming ‘x’ has a molar weight of 100 g). -
Sample type
Typical concentration
Stack emission
0.1
Mobile emission (2 m from tail pipe)
Mass collected in 1 L
Mass collected in 2 L
Mass collected in I O L
0.4 pg - 4 mg
0.8pg-8rng
4 p g - 40 mg
2.5 - 250 ppb
10- l000ng
20 ng - 2 pg
l00ng- IOug
Fenceline/ severe urban
lO-250ppb
40-1000ng
80 ng - 2 pg
0.4- 10 Kg
Indoor air
1 - lOOppb
4 - 400 ng
8 - 800 ng
40ng-4pg
Average human exposure to benzene
- 3 PPb
11 ng
22 ng
110ng
Normal urban
1-10ppb
4-40ng
8-80ng
40 - 400 ng
Normal rural
0.1 - 1 ppb
0.4 - 4 ng
0.8 - 8 ng
4-40ng
Forested
0.25 -2.5 ppb
1 -10ng
2-20ng
20 - 200 ng
Everest/K2
0.025 - 7.5 ppb
0.1 -30ng
0.2 - 60 ng
1 - 300 ng
Arctic (on an ultra clean day)
15 - 50 ppt (BTX)
60 - 200 pg
0. I2 - 0.4 ng
0.6-2ng
~
- 1000ppm
~
5.2 Transfer of Analytes to the Capillary GC System
135
5.2 Transfer of Analytes to the Capillary GC System VOC air samples are almost never sufficiently concentrated to be directly introduced into a capillary GC system via a standard gas sampling valve (GSV) (see Table 5-1). However, a wide range of GC compatible air sampling techniques are now available to the analyst. These include: -
Pumped sorbent tubes. Whole air containers such as passivated canisters or Tedlar bags. Diffusive samplers. Semi-continuous on-line air stream sampling.
The respective advantages and limitations of these different sampling methods will be reviewed in a later section, however, whichever method is employed some form of focusing step is required before capillary GC analysis (Fig. 5-1). Thermal desorption (TD) or solvent extraction (SE) are then used to transfer the focused VOCs to the analytical system. Pumped sorbent tubes
On-line air streams
Passivated canisters
Diffusive sampling
# Rapid Thermal Desorption
High resolution capillary GC analysis
Fig. 5-1. Focusing VOC air pollutants prior to capillary GC analysis.
136
5 Practical Aspects of Monitoring Volatile Organics it7 Air
5.2.1 Solvent Extraction or Thermal Desorption Solvent extraction has the advantage that, provided conventional solvents rather t h a n supercritical fluids are used, little apparatus is required over and above a conventional capillary GC/GC-MS. However, several limitations exist in terms of operator illtensive sample preparation, poor desorption efficiencies particularly for polar compounds (typically 20-40%), low sensitivity, use of hazardous solvents and unpredictable effects of water [3]. Two of these disadvantages, ie, the requirement for extensive manual sample preparation and operator contact with hazardous solvents can be minimized by using equilibrium headspace techniques to transfer analytes to the GC (Fig. 5-2) [4].However, headspace does not improve detection limits (still in the order of 1 ppm) and desorption efficiencies may still be low.
Fig. 5-2. Extraction and analysis of workplace air samples using equilibrium headspace-GC.
Poor detection limits are an inherent disadvantage of solvent extraction procedures. At least 1-2 mL of solvent is required to extract a sorbent trap. Of this only 1-2 pL is injected into the GC resulting in an immediate dilution factor of lo‘. Other effects which limit the sensitivity of solvent extraction methods include interference from solvent artifacts and from the solvent itself. The alternative approach, ie, thermal desorption, uses heat and a flo% of inert gas to transfer analytes from the focusing device to the capillary analytical column. It offers significant advantages in terms of detection limits and simplifies automation. Desorption efficiencies above 99% are routinely and reliably achieved and 100°6 of the desorbed components can be transferred to the analytical system if required. N o manual sample preparation steps are involved and no hazardous solvent is introduced. The only limitation of thermal desorption is that it is a ‘one shot’ technique, ie, once desorbed into the analytical system no sample remains for repeat analysi5. However, provided strict sample protection methods are implemented on the instrumentation used (see Section 5 . 9 , this is rarely a problem in practice. Duplicate iamples may also be readily collected if necessary. In fact the reliability of modern conimercial instrumentation now means that less than 1 per thousand tube samples are lost during automatic thermal desorption-GC analysis [ 5 ] .
5.2 Transfer of Analytes to the Capillary GC System
137
The cost efficiency and detection limit advantages of thermal desorption thus make it the preferred GC sample introduction method for almost every speciated VOC air monitoring application.
5.2.2 Focusing Trap Design Much of the success of VOC air monitoring methods depends upon the performance of the focusing device. Principal requirements include quantitative retention of target analytes during the focusing stage and quantitative and rapid transfer of compounds to the capillary GC at the start of the analysis. There are several alternative focusing trap designs. Principle options include: 1. Trapping on a ‘large’ sorbent bed at ambient temperature. 2. Cryofocusing on capillary tubing or glass beads (temperatures typically below - 150 “C). 3. Trapping on a ‘small’, moderately cool sorbent bed.
Examining each of these options in turn, the ‘large’ sorbent bed trap (typically 200- 1000 mg adsorbent) offers advantages in terms of operational simplicity (no liquid cyrogen required). However, it does not offer quantitative retention of ultra volatile analytes such as C2 hydrocarbons and heats slowly during the desorption stage to produce broad component bands (Fig. 5-3). Capillary or glass bead cryofocusing systems consume vast quantities of liquid cryogen (up to 6.5 L h-’ in some cases) [6] and may plug with ice during the analysis of humid samples. As thermal desorption is a dynamic process relying on a con-
4 mrn ID tube to 2 mm ID tube to column
4 mm ID tube to column
Ethane Ethylene
I
I
I
I
I
I
0
2
4
0
2 min
4
min
Fig. 5-3. Single stage desorption produces broad component bands.
138
5 Practical Aspects of Monitoring Volatile Organics in Air
Fig. 5-4. Design of a cold trap containing small (10- 100 mg) quantities of adsorbent.
stant flow of inert gas, restriction or blockage of the gas flow will result in incomplete desorption and unreliable results. A small sorbent cold trap (10- 100 mg adsorbent) (Fig. 5-4) provides an ideal compromise. The moderately low temperatures required for quantitative retention of ultra volatile species such as ethane, ethylene and acetylene can be achieved with electrical (Peltier) cooling to -30°C (Fig. 5-5) thus eliminating the requirement for liquid cryogens. Such traps can also be heated sufficiently fast (40°C per second,
C2 Hydrocarbons Component recovery v volume sampled Sampling time: 40 min 16 to 19 ng each component
400000
3
350000
300000{ 5
1
I
2500001
c
3 0
200000~
a
a
-
Ethylene (75' RH) 1 D--fl Acetylene (75 RH)
100000~ J
50000{
Flow rate (mumin) 15
Q-
"I'
"I'
"I'
';
"I'
"I'
" I
25 "(I
35
"I'
"I'
$ 1
5.2 Transfer of Analytes to the Capillary GC System
139
8 7 --
6
--
5 -4
--
3
--
2 --
I --
o ! 0
I
20
40
60
i 80
Heating Rate (degCfsec)
Fig. 5-6. Rapid (40 "C per second) desorption of a small sorbent cold trap produces capillary compatible 1 -2 s wide peaks.
ie, 2400 "C per min) to allow rapid transfer of analytes to the analytical column without additional cryofocusing and without compromising capillary GC performance (Fig. 5-6). Provided the 'outlet' end of such a trap is restricted as shown in Fig. 5-4, narrow capillary peaks are still produced with desorption flow rates down to 2-2.5 mL min-'. This allows splitless operation with 0.32 mm and 0.53 mm ID capillary columns. The design of a commercially available Peltier cooled trap is shown in Fig. 5-7. This trap is compatible with all air sample types.
Fig. 5-7. Design of a commercially available Peltier cooled trap.
140
5 Practical Aspects of Monitoring Volatile Organics in Air
5.3 Air Sampling Methods 5.3.1 Whole Air Sampling Whole air sampling procedures, using passivated stainless steel canisters (Fig. 5-S), Tedlar bags or some other suitable container, have been developed primarily for apolar species ranging in volatility from acetylene to trichlorobenzene [7]. Passivated stainless steel canisters in particular, have been extensively evaluated in the US for ‘Air Toxics’ (US EPA Method TO-14) and C2- C9hydrocarbons. During analysis an aliquot of canister air (typically 400mL) is drawn into the focusing trap using a pump and mass flow controller. Canisters and bags offer advantages for sampling very volatile components which cannot be quantitatively retained on conventional sorbents at ambient temperature and also benefit from low (-zero) artifact interference provided the containers are well cleaned. However, canisters are expensive to buy and transport and can be difficult to clean once contaminated. Specific conditions in terms of temperature, humidity and pressure are also required for the air inside the canister if the full range of target analytes is to be monitored without loss of high boilers [S]. Collecting grab samples of air using canisters is simply achieved by opening a valve on an evacuated canister and reclosing it again once the pressure inside approaches atmospheric. However, the collection of time weighted average samples requires relatively complex apparatus (Fig. 5-9) which can be difficult to assemble and operate in the field. Most canister methods (eg, US EPA Method TO-14) also specify that only a relatively small air sample volume should be transferred from the canister to the GC analytical systems (ie, only 300-500mL) and this limits sensitivity.
Fig. 5-8. Passivated stainless steel canister used for VOC air sampling.
141
5.3 Air Sampling Methods To AC
I Insulated enclosure
K
Inlet
Vacuumlpressure gauge Timer
- Inlet manifold
I
Valv
- 1.6 meters
I I Metal bellows I I
(- 5 ft)
typepump 'b'b forpressurized 1 I sampling I I ,-II
Ground level Vent
Mass flow meter
4
@+P
II
Iv
Mass flow
'41
J Canister
I '
I
I
Fig. 5-9. Apparatus required for time weighted average sampling using a canister. (Diagram taken from USEPA Method TO-I4 [7]).
5.3.2 Air Sampling Using Sorbent Tubes Sorbent tubes (Fig. 5-10) provide a much lower cost, more practical and more versatile alternative to canisters or other containers for most VOC air monitoring applications. They are not suitable for collecting ultra volatile components such as C2
142
5 Practical Aspects of Monitoring Volatile Organics in Air
Fig. 5-10. Sorbent tubes for VOC air monitoring.
hydrocarbons at ambient temperature, but are compatible with both apolar and polar compounds [9] and offer quantitative retention of organics ranging in volatility from C3 hydrocarbons. Some inorganic vapors such as N 2 0 and CS2 can also be quantitatively collected at ambient temperatures. Other advantages include a) the collection of multi-liter air samples if required b) that the sample tubes are clean and ready for immediate reuse after thermal desorption- GC analysis and c) that sample tubes can usually be reused at least 100 times before the sorbent needs replacing [lo]. One limitation of sample tubes is that artifacts from the sorbent material may interfere with component analysis unless stringent conditioning and storage procedures are applied. 5.3.2.1 Minimizing Artifact Interference for Sorbent Tubes Sorbent Selection. Most artifact interference can be eliminated by selecting an appropriate sorbent or series of sorbents for the particular target analyte range (Table 5-2). It is worth nothing at this point that perhaps the most popular tube packing material - TenaxTMTA - is in fact a very weak adsorbent only suitable for components less volatile than benzene. Guidance on sorbent selection is given in Table 5-2 and is also available in published form [ 1 I ] , but as a rough guide the retention volume of an analyte on a given sorbent should exceed 100 L g-' for optimum performance. If a suitable strong sorbent is selected, the mass of sorbent required in the sample tube can be kept below 1 g. This is a critical issue - many early sorbent
5.3 Air Sampling Methods
143
Table 5-2. Sorbent selection guidelines. Sample tube sorbent
Approximately analyte volatility range
Maximum temperature ("C)
Specific surface area (m2g - 9
Example analytes
Glass fiber filters
n-C,, to n-C3,
>4OO0C
CarbotrapCTM CarbopackCTM
n-C, to n-C,,
>400 "C
12
Alkyl benzenes, PAHs, PCBs
Tenax TATM
bp 100°C to 400°C n-C, to n-C30
>350°C
35
Aromatics except benzene, Apolar components (bp > 100 "C) and less volatile polar components (bp> 150°C)
Tenax GR
bp 100°C to 450°C n-C, to n-C30
>350°C
35
Alkylbenzenes, PAHs, PCBs and as above for Tenax TA
CarbotrapTM CarbopackBTM
(n-CJn-C, to n-C14)
400 "C
100
Wide range of VOCs including, ketones, alcohols, and aldehydes (bp > 75 "C) and all apolar compounds within the volatility range specified. Plus per fluorocarbon tracer gases
ChromosorbTM 102
bp 50 "C to 200 "C
250 "C
350
Suits a wide range of VOCs including oxygenated compounds and haloforms less volatile than methylene chloride
Chromosorb 106
bp 50°C t o 200°C
250°C
7 50
Suits a wide range of VOCs including hydrocarbons from n-C, to n-C,2. Also good for volatile oxygenated compounds
Porapak Q
bp 50°C to 200°C n-C, to n-C,2+
250 "C
550
Suits a wide range of VOCs including oxygenated compounds
Particulate emission
144
5 Practical Aspects of Monitoring Volatile Organics in Air
Table 5-2 (continued) Sample tube sorbent
Approximately analyte volatility range
Maximum temperature ("C)
Specific surface area (m2g - 7
Example analytes
Porapak N
bp 50°C to 150°C n-C, to n-C,
180°C
300
Specifically selected for volatile nitriles: acrylonitrile, acetonitrile and propionitrile. Also good for pyridine, volatile alcohols, from EtOH, MEK, etc.
Spherocarb
-30°C to 150°C C, to n-C,
> 400 "C
1200
Good for very volatile compounds such as VCM, ethylene oxide, CS,, CH,CI,. Also good for volatile polars eg MeOH, EtOH and acetone
Carbosieve SIII
-60°C to 80°C
>4OO0C
800
Good for ultra volatile compounds such a (CJ, C, and C, hydrocarbons
Molecular sieve
-60°C to 80°C
350°C
Charcoal
-80°C to 50°C
>400"C
Used specifically for 1,3-butadiene and nitrous oxide
> 1000
Rarely used for thermal desorption because metal content may catalyze analyte degradation. Use, with care for ultra volatile C,, C,, C, hydrocarbons
CarbotrapCTM, CarbopackTM, CarbotrapTM, CarbopackBTM, and Carbosieve SillTM, are all trademarks of Supelco Inc., USA. TenaxTM,is a trademark of Enka Research Institute. ChromosorbTM,is a trademark of Manville Corp.
5.3 Air Sampling Methods
145
tube monitoring studies suffered from excessive artifact formation and erroneous results principally because wide bore (5/8 inch diameter) tubes (volume approximately 20 mL) were used containing large masses of sorbent. This resulted in the following difficulties: 0 0
0
0
Incomplete conditioning of freshly packed tubes. Incomplete purge of aidoxygen prior to desorption. (Approximately 200 mL carrier gas is required to sweep air from a 20mL tube. Any air left in the sample tube during thermal desorption will oxidize the sorbent and analytes producing artifacts). Uneven tube heating during desorption which caused some parts of the sorbent bed to reach excessive temperatures and start to degrade while other parts did not reach the required temperature for complete desorption. Errors during air sampling at pump flow rates < 50 mL min-' due to ingress of VOCs via diffusion. (On a 5/8 inch diameter tube with the sorbent surface 15 mm from the tube inlet, diffusion of VOCs will occur at a rate equivalent to a flow of approximately 3 - 5 mL min-'. This can cause a significant error if low pump flow rates are used).
All these concerns are eliminated or minimized if sample tube diameters are restricted to 1/4 inch (6 mm). Standard 1/4 inch (6 mm) diameter by 3.5 inch (89 mm) long tubes with a volume of only 3 mL, contain only 200 mg- 1 g sorbent depending on sorbent density. These dimensions facilitate both stringent conditioning of freshly packed tubes and uniform heating during desorption. Air is also completely purged with only 30 mL of carrier gas prior to tube heating and diffusive uptake rates are in the order of only 0.5mLmin-' thus allowing the use of sampling pump flow rates as low as 5-1OmLmin-'. In fact, sorbent bed length is a much more important factor than sorbent mass in determining the overall analyte retention capability of a sample tube. If a more retentive sample tube is required for a given VOC monitoring application, it is much more practical and effective to change to a stronger sorbent than move to a larger tube. For example, Safe Sampling Volume (SSV) data presented in Ref. [2] for standard 1/4 inch by 3.5 inch tubes, show hexane with an SSV of only 3.2 L using Tenax TA, but with an SSV of 2000000L on an identical sample tube packed with Spherocarb. Such vast improvements in analyte retention are only feasable by correctly selecting an adsorbent of appropriate strength and could never be achieved by increasing tube size. lhbe conditioning and storage. Effective tube conditioning and storage procedures must be applied especially for trace level monitoring. It is always advisable to use more stringent conditions of desorption temperature, desorption time and carrier gas flow for the tube conditioning procedure than those used for sample analysis. Sample tubes should also always be securely capped with Swagelok-type screw cap fittings with PTFE ferrules and kept in clean atmospheres for long term storage and transportation [12]. Large sealed glass jars filled with pure nitrogen make ideal storage and transport containers.
146
5 Practical Aspects of Monitoring Volatile Organics in Air
Q)
-c Q)
>,
5 Q)
-rE0
If these precautions are applied, trace level monitoring is possible. Figure 5-1 1 shows the analysis of a tube containing 1 pg tetrachloroethylene using thermal desorption-capillary GC and electron capture detection (ECD). The level shown equates to less than 0.1 ppt tetrachloroethylene in 10 L air. Figure 5-12 shows the analysis of a 1 L air sample containing 100 ppt bischloromethylether (BCME) detected using a mass spectrometer detector operating in full scan mode. The mass of analytes collected from different air volumes, as shown in Table 5-1, illustrates the exceptional detection limits which can be achieved using pumped air sampling onto sorbent tubes and thermal desorption-GC analysis.
0
E
zm
c,
Q
z
x
1
Fig. 5-11. Analysis of a sorbent tube containing 1 pg tetrachloroethylene.
Fig. 5-12. Analysis of 1 L air containing 100 ppt bischloromethyl ether.
5.3 Air Sampling Methods
147
5.3.2.2 Pumped Air Sampling Onto Sorbent Tubes
Air is usually pumped into one or more sorbent tubes using a personal monitoring pump. Several replicate samples may be collected in parallel using constant flow type pumps in instances where a repeat analysis capability is essential. The direction of the gas flow is always reversed during thermal desorption so that higher boiling compounds are backflushed easily from the sampling end of the tube. If the target analytes cover a wide volatility range, tubes may be packed with a series of sorbents of increasing strength. Alternatively a train of tubes may be coupled together in series. Examples of applications requiring multi-sorbent trapping media include total gasoline vapor monitoring [13], seawaterlsediment analysis [ 14, 151 and air toxic analysis [16]. These are all described in more detail in a later section. Pump flow rates above 5 - 10 mL min-' (1/4-inch diameter tubes) should be used in order to minimize errors due to ingress of VOCs via diffusion. Flow rates in excess of 200 mL min-I are also not recommended for standard 1/Cinch sample tubes unless for short term (eg 10 min) monitoring [17]. [NB. High sampling flow rates can be used longer term for high boiling materials such as PCBs and PAHs in air (see Section 5.6)]. 5.3.2.3 Automating Pumped Tube Monitoring
Pumped tube sampling procedures can now be automated using commercial sequential tube sampling (STS) devices (Fig. 5-13). The unit shown is a portable device developed by Perkin-Elmer Corp. in collaboration with the US EPA Atmospheric Research and Exposure Assessment Laboratory (AREAL) under a Federal Technology Transfer Agreement. This particular unit is operated via 12 V battery or mains electricity and is housed in a weatherproof box. It is compatible with most standard monitoring pumps (constant flow type preferred) and keeps the pump fully charged during operation. Sampling times may be selected between 1 min and 100 h per tube. Sequential samplers like the Model STS 25 may be used to sample the surrounding air or sample gas streams (gas stream flow rates 400 mL min-'). Sample tubes are fitted with diffusion limiting caps while on the unit carousel. This prevents ingress of air pollutants outside the sampling period (Fig. 5-14). 5.3.2.4 Diffusive Sampling Onto Sorbent Tubes
Some thermal desorption sample tubes offer the option of diffusive as well as pumped monitoring (Fig. 5-15 ) . Tube-form diffusive samplers overcome the air speed restrictions of early badge type diffusive monitor designs and can be used with air speeds down to 3 - 5 cm s-l [18] ie, they can be used both for personal exposure measurements and static monitoring. Diffusive monitors are particularly useful for collecting time weighted average personal exposure data during workplace air monitoring studies (see Section 5.6). They eliminate the requirement for cumbersome personal monitoring pumps, can be worn close to the breathing zone (Fig. 5-16) and
148
5 Practical Aspects of Monitoring Volatile Organics in Air
Fig. 5-13. Model STS 25 Sequential Tube Sampler manufactured by Perkin-Elmer.
Fig, 5-14. Sample tube fitted with a diffusion limiting cap for minimizing VOC ingress while in the STS 25.
5.3 Air Sampling Methods Diffusion membrane (if fitted)
gauze
149
Stainless steel tube Adsorbent \
/
gauzes
Fig. 5-15, 1/4 inch sample tube configured for diffusive monitoring.
Fig. 5-16. Diffusive monitors are light, unobtrusive and easy to wear near the breathing zone.
have negligible impact on a worker’s normal behavior. Replicate samples can also be collected with little impact on monitoring cost if required. Diffusive (passive) sampling is now a well accepted and well validated technique [19, 201 with international quality assurance schemes available to test method performance. According to Fick’s law of diffusion, compounds will diffuse at a fixed rate from the outer surface of the monitor (ambient concentration) to the adsorbent (zero concentration) (Fig. 5-17). The sorbent must be strong enough to prevent back diffusion of the analyte and, as a general rule, guidance given for sorbent selection for pumped tube monitoring also applies to diffusive sampling. The uptake rate (U)of an analyte is directly proportional to the surface area of the monitor ( A )and inversely proportional to the diffusion path length (2).It is also related to the diffusion coefficient of the component in air (0) (Eqs. (1) and (2))
150
5 Practical Aspects of Monitoring Volarile Organics in Air
Fig. 5-17. Principles of diffusive monitoring.
Uptake rate (Urn(mL min-') =
Uptake rate Up (ng ppm-' min-') =
6OxDxA
z 60 x D x A x MW 24.25 x Z
where MW is the molecular weight of the analyte. Given an 'ideal' adsorbent, the uptake rate will remain constant for a given analyte provided the dimensions of the monitor remain constant. For the tubes shown (Figs. 5-10 and 5-15); A is 0.193 cm2 and Z is 1.59 cm. These are now internationally accepted as standard diffusive tube-type monitor dimensions. A considerable number of uptake rates have already been published for thermal desorption tube-type diffusive monitors (211. A summary of these is presented in Appendix A. International validation protocols for diffusive methods, requiring both laboratory and field experiments, have also been established [22 -241 and reliable methods now exist for accurately calculating/predicting effective uptake rates from ideal uptake rates using retention volume data [25, 261. As the tube monitor dimensions have been rigidly fixed, uptake rates may be accepted as standard and applied universally once determined. Diffusive sampling is most suited to monitoring individual organic compounds or a narrow volatility range of components as only one sorbent can be used at any one time. This said, several different monitors may be worn or placed simultaneously if required. The relatively slow sampling rate of 1/binch tube-type monitors (equivalent to 0.5 mL min-I) restricts detection limits to around 1- 10 ppb (FID detection) for normal 8-h sampling periods. Long term (3 - 28 day) diffusive sampling times are currently under evaluation [27] and may be used for sub-ppb detection limits.
-
5.4 Moisture Management
151
5.3.3 On-Line Air Stream Sampling Semi-continuous on-line air sampling and GC analysis is a relatively new technique with applications ranging from urban air quality testing to occupational hygiene (see Section 5.6). The procedure involves a volume of air being pumped directly into the focusing trap via an inert mass flow controller. Air is usually sampled for the greatest possible percentage of the cycle time - for example key urban air monitoring methods demand minimum 40-min sampling per hourly cycle [ 11. After sample collection, the trap heats rapidly, transfers the sample to the GC analytical column and initiates the GC separation. As soon as the trap recools the system is ready to collect the next sample. The chromatographic analysis of the previous sample continues while the next sample is collected.
5.4 Moisture Management Moisture management need not be a complex issue with regard to air monitoring applications. Many sorbents, for example Tenax, Chromosorbs, Porapaks and Carbotraps/packs, are essentially hydrophobic and retain very little water, even when sampling at high humidity (> 90% RH). As a rule, tubes packed with such sorbents and held at ambient temperatures will not retain more than 1 - 2 mg of water whatever humidity air is being sampled. No additional sample drying is generally required. This range of hydrophobic adsorbents is applicable to the vast majority of organic components (see Table 5-2) ie, all those components with volatility < n-C4 [9]. However, if stronger, less hydrophobic sorbents such as charcoal, the CarboxenTMseries, Spherocarb or Carbosieve SIIITMare required or if whole-air containers or on-line air sampling methods are used, some water will be retained and passed through to the focusing device. In these cases, steps must be taken to eliminate water from the system before trap desorption and chromatographic analysis. Provided a small, electrically-cooled, sorbent-packed focusing device is used, such as that described above (see Section 5.2.2). There are several alternative approaches to water removal which can be used. Key options include: Permeable membrane dryers used between the sample focusing trap (moist suitable for whole-aidon-line air samples). - Desiccant dryers (used as with the membrane dryers). - Dry purging, ie, selecting suitable focusing trap parameters (in terms of sorbent type/mass, trapping temperature and dry gas flow) such that water is eliminated while components of interest are quantitatively retained. Dry purging is applicable to all air sample types. -
Each of these methods has a role within the broad range of VOC air monitoring applications and the advantages and limitations of each are reviewed below.
152
5 Practical Aspecfs of Moniforing Voiafiie Organics in Air
5.4.1 Permeable Membrane Dryers Membrane driers such as the NafionTM dryer from Perma Pure Inc. (8 Executive Drive, PO Box 2105, Toms River, NJ 08754, USA) have been extensively evaluated for use with hydrocarbons, halocarbons and aromatics 1281. They are widely used for ozone precursor monitoring and air toxics methods (see Section 5.6). Their main limitation is that polar analytes are removed with the water. They are also reported to produce minor artifacts which may become significant when monitoring trace level (ppt) pollutants in rural air. However, they do effectively remove water from high humidity samples and their selectivity f o r apolar analytes can be an advantage if only those components are of interest. Membrane dryers are applicable to a wide range of apolar analytes including C2 hydrocarbons as their use is independent of focusing trap temperatures.
5.4.2 Desiccant Dryers Desiccant dryers, containing potassium carbonate or some other salt, can also be used between the air sample and the focusing trap to remove water from the air as it passes onto the trap. Desiccants have not been as extensively evaluated as membrane dryers and their characteristics are not fully understood with regard to which organics are retained with water and which pass straight through. However, they are compatible with a wider polarity range of VOCs than membrane dryers and do quantitatively retain and release large quantities of water. Desiccant temperatures must be carefully controlled during the water removal process and the dryers generally require regeneration between each sample during automatic analysis.
5.4.3 Dry Purging Dry purge procedures are compatible with every air sample type and can be used for all but the most volatile analytes. Once the humid air sample has been transferred onto the focusing trap, a volume of dry gas is purged through the trap to eliminate water. This is of course only applicable to analyte - sorbent combinations which offer quantitative retention of the analyte and the simultaneous removal of most of the water. If applicable, dry purge methods offer the ideal water removal procedure as no additional hardware is required in the air stream thus minimizing the risk of either artifact formation or unpredictable sample losses. With the wide range of sorbents now available commercially, most VOC target analytes can be handled this way. For example, it is possible to use dry purging for almost every component that is quantitatively monitored using ambient sorbent tube monitoring. US EPA Method TO-14 (Air
5.4 Moisture Management
153
Staae 1: Sample Transfer to the Focusing Trap
itage 2: Dry Purging
Purge Gas Plus Water
GC Analytical Column
-
Stage 3: Focusing Trap Desorption Analyte Transfer to the GC Analytical Column ----f
GC Analytical Column
Fig. 5-18. Sequence of operation for dry purge mode.
Toxic) target analytes which range in volatility from freons 12 and 14 can also be quantitatively retained on a packed focusing trap set at temperatures above zero, while water is purged to vent [29,30]. Dry purging of a sorbent tube type air sample is simply carried out be selecting an appropriate focusing trap temperature ( > 0 "C) and extending the tube desorption time until most of the water has transferred from the tube to the trap and from the
154
5 Practical Aspects of Monitoring Volatile Organic in Air
TIC excluding 44 Maximum Signal 3940789 Max for Run 3940789 100.
13
I
75
50
25
c
m
0
.58
6.41
10.26
14.12
17.97
21.82
25.68
29.53
33.38
37.23
41.08
Time mln 1. 2. 3. 4.
Freon 12 Freon 114/rnethanol Acetonitrile Freon 11 5. Acrylonitrile 6. Dichlommethane
7. Methyl-tert-butylether
8. Methyl ethyl ketone 9. n-Butanone 10. Benzene 11. Toluene 12. Ethylbenzene 13. o-Xylene 14. 4-Ethyltoluene
15. 16. 17. 18. 19.
rn-Dichlorobenzene p-Dichlorobenzene 0-Dichlorobenzene 1,2,4-Trichlorobenzene Hexachlorobutadiene
Fig. 5-19. Simultaneous thermal desorption-GC-MS analysis of TO-14 air toxics target analytes plus polar VOCs using dry purge.
trap to vent. Care must be taken when selecting the trap sorbent and dry purge conditions that analytes are not lost with the water. If the humid air sample is transferred onto the trap from a whole-air sample or air stream, an extra step is introduced into the normal sequence of operation (Fig. 5-18). The main advantage of dry purge mode is its compatibility with wide polarity range VOCs. Compounds such as alcohols, ketones, aldehydes, esters, ethers and glycols can all be quantitatively retained and separated from water using this procedure (Fig. 5-19). It is not suitable for Cz hydrocarbons as these compounds require subzero trapping temperatures. Methods requiring ethane, ethylene and acetylene analysis in humid air are therefore restricted to desiccant or membrane dryers.
5.5 Analytical Instrumentation and System Calibration Although manual instrumentation performs a useful function during method development and research phases of an operation, most industrial and routine contract laboratories now require system automation to maximize productivity and minimize
5.5 Analytical Instrumentation and System Calibration
15 5
cost per analysis. One key factor in effective automation is the elimination of high cost, operator intensive consumables such as liquid nitrogen. Field studies using systems with liquid nitrogen-cooled focusing devices have been shown to incur excessive operating costs, demand considerable operator attention and suffer poor reliability (up to 18% down time due to liquid cryogen related problems alone) 161. Cryogen free systems are thus generally preferred for automatic thermal desorption-GC analysis. However, several other aspects of system automation and calibration also require careful consideration if good reliability is to be maintained. Factors affecting reliability will vary depending upon the type of automatic system required.
5.5.1 Automation and Calibration of Whole-Air Container Analysis The process of sampling air into whole-air containers is rarely automated because of the logistical complexity of sampling into a series of large (typically 6L) canisters or Tedlar bags. Automation is usually limited to custom-made grab sampling systems, incorporating a series of remote controlled on-off valves to open and close evacuated canisters at predefined times. The analytical process can be automated however, using multiport valves to switch between different air samples. Key requirements for such ‘whole-air’ autosamplers include calibrant canisters, blanks and gasphase internal standard introduction. Ideally the entire calibration process should be fully automated with a gas-phase internal standard introduced during every sample, blank and canister (external) standard introduction cycle. A typical sequence of events is as follows: Step 1: Sample flow path (including focusing trap) fully leak tested before the sampling sequence. Step 2: First sample introduced into the focusing trap using a controlled flow and fixed sampling time. A fixed volume of gas-phase internal standard is introduced into the sample air stream at some stage early in the sample introduction process. Step 3: Maintain the gas flow direction and purge the focusing trap to remove air (and water if applicable) using inert (carrier) gas. Step 4: Desorb the cold trap to initiate the GC run. Reverse the flow of carrier gas through the cold trap during desorption in order to backflush components from the cold trap onto the capillary column. Keep the trap hot for sufficient time to ensure complete desorption of the components of interest. Step 5: Allow the trap to cool and start introducing the next sample as soon after the trap has reached its trapping temperature as is required (ie, at the appropriate cycle time). The introduction procedure for the next sample can be started while the GC analysis of the previous sample continues.
156
5 Practical Aspects of Monitoring Volatile Organics in Air
At least one of the canisters in the rack should contain a nominal or accurate gas standard of all target analytes to allow retention time confirmation and/or accurate external standard calibration for each component of interest. Another should contain a blank or zero-air sample if possible. NB. If the components of interest are all chemically similar (for example all are hydrocarbons), a carbon-number type calibration may be used (see Appendix B).
5.5.2 Automation and Calibration of Sorbent Tube Analysis Automation of tube sampling procedures using a sequential tube sampler, is possible for both air/gas streams and ambient air quality measurements. One such unit (Fig. 5-13) is described in Section 5.3.2.3. In view of the ‘one-shot’ nature of thermal desorption, great care must be taken when automating tube analysis. Tubes on the autosampler must be sealed both before and after analysis to prevent loss of analytes of interest and ingress of laboratory air contaminants. Automatic tube desorption units thus require tubes to be cappedkealed with fittings that are inert and can be automatically removed and replaced (or opened and closed) by the instrument. This ensures that sample integrity is preserved before analysis and that tubes are kept clean and immediately ready for re-use after analysis. Each sample tube must be thoroughly leak tested without heat or carrier gas flow before analysis. The leak test should be suitably thorough (for example, less than 5% pressure drop after 3 0 s under pressurized conditions) and any failures should be stored in system memory and communicated to the data handling device to keep all system components in step. Tubes which fail the test should be resealed and preserved intact, ie, without being desorbed.
rrier gas in Injection using a conventional .GCsyringe
I
-
spring
50-100mUmin Carrier gas flow
Fig. 5-20. Introduction of liquid standard onto a sample tube via a flash vaporizing injec:tor.
5.5 Analytical Instrumentation and System Calibration
/ Appropriate adsorbent
157
Sample deposited here using a clean syringe
Fig. 5-21. Direct introduction of a liquid standard onto a sample tube.
Air must be purged from the sorbent tube before desorption to eliminate risk of analyte and/or sorbent oxidation. Essential system status functions, including temperature settings, carrier gas pressure and ionization gases (if applicable), should be continually checked throughout the analytical sequence to ensure no sample is desorbed without the analytical system being ‘ready’. Systems which do not include ionization gas checks as part of the standard system ‘ready’ status checks can be fitted with commercial ‘Flame-out’ accessories if required. Automatic tube desorption is normally calibrated via external standard introduction. Recommended procedures are illustrated as Methods A and B in Figs. 5-20 and 5-21 respectively. Method A is preferred for air monitoring applications as standard components are introduced onto the sorbent in the vapor phase, in a way which most closely represents normal air sampling. Some commercial automatic thermal desorption systems also offer the facility of automatic gas phase internal standard introduction via a gas sampling valve (GSV). Such systems allow fixed volumes of standard gas to be introduced automatically onto the back of a sample tube after the leak check and purge stages and before tube desorption. Both internal and external standard calibration calculations are described in Appendix B. Thermal desorption tubes are also available as certified standards [31]. These reference tubes have been extensively evaluated and tested for long term (2 year) storage stability. Certified standards are not generally used for routine calibration purposes, but are used for analytical quality assurance testing. It is generally strongly recommended that laboratories carrying out routine analyses of any kind participate in an appropriate quality assurance scheme and several such schemes already exist for VOC air monitoring [32]. Results submitted by laboratories participating in QA schemes are typically graded into 3 or more classes. Although the results for each individual laboratory are not usually made public, a summary of the overall performance of all those participating is usually published in tabular form (see Table 5-3).
5.5.3 Calibrating Automatic On-Line Air Stream Analysis On-line air stream sampling systems are usually calibrated in the same way as wholeair sample systems, ie, via intermittent sampling of a canister air standard. The canis-
158
5 Practical Aspects of Monitoring Volatile Organics in Air
Table5-3. Tabulated results from an example round of the UK quality assurance scheme: Workplace Analysis Scheme for Proficiency (WASP). Analyte
Participants (number)
Category 1 (To)
2
78 I1 19
22 16 23
41 46 52
2.5
25 24 24
24 17 25
56 66 50
20 17 25
48 56 56
31 32 15
33 41 42
29 21 43
23 26 26 22
52 54 50 60 54 51
22 21 27
43
26 19 23 22 36 28
667
24
49
21
(070)
Membrane Cadmium Chromium Lead
31 38
Glass-Fibre Cadmium Chromium Lead
Charcoal (solvent extraction) Benzene Toluene Xylene
Tenux (thermal desorption) Benzene Toluene Xylene 1, I ,1-Trichloroethene Trichloroethene Tetrachloroet hane Total
44
18
10 21
ter air standard may be ‘nominal’ (ie, contain only approximate concentrations of each analyte) and used for retention time confirmation only or may be an accurate standard for external standard calibration purposes. Appropriate calibration calculations are described in Appendix B.
5.6 VOC Air Monitoring Applications Armed with the appropriate sampling, analytical and calibration technology, it is now possible to address the wide range of VOC air monitoring applications. These generally fall into one of four categories: -
Workplace air monitoring, ie, occupational hygiene.
- Industrial/mobile source emission testing.
5.6 VOC Air Monitoring Applications
159
Air quality assessment - indoor air - ambient air - urban air - Specific research programs such as biogenic emission testing and pollution transport. -
A review of each of the major application areas is presented below with examples.
5.6.1 Workplace Air Monitoring New workplace health and safety legislation and codes of practice have continued to reduce the exposure limit levels of hazardous substances in workplace air. Many of the new regulations have also increased employer liability, ie, in some countries it is now no longer the responsibility of the enforcement agency to prove negligence, but rather the responsibility of the employer to prove compliance [33, 341. The new regulations have also tended to move away from the collection of general atmospheric concentration data and tend to favor personal exposure monitoring so that the health record of a worker can be related directly to his or her individual exposure level. Personal exposure measurements are collected using monitors on each individual worker placed as close as possible to the breathing zone. These monitoring methods are preferred because wide variations in worker behavior (both between individuals and from day-to-day with a single individual) make it impossible to extrapolate general atmospheric concentration data to workers’ personal exposure values. In fact personal exposure measurements collected from a large group of workers all ‘doing the same job’ or from one individual doing a repetitive task day after day can vary over nearly two orders of magnitude (Fig. 5-22). This effectively means that large numbers of samples must be collected before a reliable average exposure level can be calculated. One or two measurements are not sufficient. This in turn means that monitoring costs must be minimized before it is feasible for industry to collect meaningful information on personal exposure levels as required by law.
5.6.1.1 Benefits of Diffusive Sampling
Diffusive monitoring is an ideal option for workplace air monitoring - if applicable (see above). The simple, light monitors can be worn close to the breathing zone and provide a low cost method for collecting large numbers of samples. The monitors are not intrusive and have negligible impact on an individual’s normal work pattern. Diffusion is primarily used for monitoring individual components or small groups of similar compounds for example ethylene oxide, anaesthetic gases in operating theatres [35], benzene toluene and xylene (BTX) measurements, etc.
160
5 Practical Aspects of Monitoring Volatile Organics in Air
Probability of result occurring for a given atmosphere
Skewed distribution usually found in workplace air monitoring
5
10
15
20
25
30
units
Measured air concentration or exposure level
Fig. 5-22. Distribution of personal exposure measurements collected from a group of individuals all ‘doing the same job’.
5.6.1.2 Pumped Monitoring Applications in Workplace Air Pumped tube sampling is still prefered for wide boiling range samples, for example total gasoline vapor monitoring (n-C, to n-Clo), where two or more sorbents are required in series to quantitatively retain and desorb the entire analyte volatility range (Fig. 5-23) [131.
0-
5-
L=
n-propane
n-butane
10
-
7
2-methylpentane
15
-
20
-
25
-
30
-
35
-
40
-
45
-
50
-
Data from Charcoal tube
Data from Chromosorb 106 tube
ethyl methyl benzene trimethyl benzene
r
Fig. 5-23. Thermal desorption-GC analysis of total gasoline vapor.
5.6 VOC Air Monitoring Applications
16 1
Trace level analyses of highly toxic components in workplace air generally also require pumped monitoring because of the low detection limits required (for example BCME with a workplace limit level of 1 ppb) [36]. Pumped monitoring can be used to sample relatively large volumes of air (> 10 L) and sub-ppb detection limits are then possible with a wide range of detectors. Pumped monitoring is also preferred for ‘unknown’ atmospheres where the nature and concentration of organic air pollutants has not been tested. Overall, pumped sampling onto sorbent tubes is still the most widely used of all air monitoring techniques, as it is compatible with the widest range of VOC pollutants and offers excellent detection limits. In fact, pumped tube sampling methods are so effective at collecting and concentrating organic volatiles that sample overload can become a significant practical consideration. For example, if sampling toluene and pentane at atmospheric concentrations of around 1 ppm, using conventional 50mL min-’ pump flow rates, then -92 pg toluene and -72 pg pentane will be collected over a typical 8 h shift (see Table 5-1). In capillary GC terms, these masses are vast. It is also woth noting that atmospheric concentrations of 1 ppm are low by workplace air standards - even larger analyte masses are collected in many cases. High resolution capillary GC analytical columns (0.2 and 0.3 mm ID) in fact perform best with analyte masses in the range 10-200 ng. Sophisticated sample splitting options are thus required on automatic thermal desorption instrumentation if
Stage 1 -Primary ( tube) desorption ‘Inlet split‘
‘Desorb’ flow
4
GC detector
inlet W
GC analytical column
Stage 2 - Secondary (trap) desorption ‘Outlet split’ GC
Carrier inlet
GC analytical column
Fig. 5-24. ‘Double splitting’ for high concentration samples.
162
5 Practical Aspects of Monitoring Volatile Organics in Air
the systems are to be sufficiently versatile to handle both trace level (ppt) and high level (ppm) atmospheric concentrations. Double splitting devices are available on some commercial desorbers (Fig. 5-24). These introduce an optional split point both before and after the focusing trap. Provided the split flows of such systems can be full adjusted using needle valves and selected on or off using appropriate solenoid or ‘shut-off’ valves, such a splitter arrangement can be used to provide split ratios anywhere in the range zero to 30000: 1. High split ratios are required for the analysis of concentrated atmospheric samples (Fig. 5-25) and also facilitate use of sensitive selective detectors such as the electron capture detector (ECD). High split ratios are also useful during direct desorption of VOCs from solid and liquid samples. Having a split on the inlet to the cold trap also provides a means of retaining an aliquot of sample if multiple thermal desorption-GC analyses are required on a single sample. Such retrapping of components from the prefocusing trap split vent onto a second sample tube can only be carried out manually and is therefore only recommended for critical samples where, for some reason, a duplicate was not taken. Nevertheless, this method has been shown to allow quantitative recollection of large percentages of sample if required [37]. Trace level analyses of highly toxic components in the workplace generally require pumped monitoring because of the low detection limits required. Pumped monitoring can be used to sample relatively large volumes of air and sub-ppb detections limits are then possible with a wide range of detectors (Figs. 5-1 1 and 5-12; Table 5- 1).
z
Fig. 5-25. Thermal desorption-capillary GC analysis of 4.3 mg of benzene and toluene using a total split ratio of 27000: 1. Note: 4.3 mg is approximately the mass of each component that would be collected from 25 L of air containing 50 ppm benzene and toluene.
5.6 VOC Air Monitoring Applications
163
5.6.1.3 On-Line Air Applications in Workplace Air Monitoring
Workplace air concentrations of very toxic components such as BCME are sometimes also monitored on-line to provide semi-continuous ‘real-time’ concentration data to complement personal exposure measurements. This allows rapid response should an emergency situation arise. 5.6.1.4 Standard Methodology
Historically, workplace air monitoring methods have been published as a series of individual standard methods for example the NIOSH Handbook of Standard Hazardous Methods (US), the UK Health and Safety Executive series of Methods for the Determination of Hazardous Substances (MDHS methods) and German DIN methods. The advantage of such methods is that they give clear, step-by-step analytical procedures for the most common workplace air pollutants. However, a major limitation of this type of standard is that the technology specified is effectively frozen in time at the point of method publication. The numbers of methods involved usually means that the time and cost of reviewing the standard procedures is prohibitively expensive. A more enlightened approach, now being taken by Health and Safety organisations world-wide, is therefore to specify method performance criteria and method validation protocols [22-24, 381 which allow methodology to develop and evolve within fixed performance limits. With regard to method performance criteria, the European Standards Organization (CEN), the USEPA and NIOSH/OSHA insist on very similar factors: the accuracy of the sampling and analysis method selected must fall within 30% or k 25% respectively and must have been tested against an alternative fully validated procedure. 5.6.1.5 Data Interpretation
Another important aspect of workplace air monitoring is interpretation of the data collected. A log graph of atmospheric concentration versus probability of result occurrence for 3 different workplace atmospheres is shown schematically in (Fig. 5-26). The first atmosphere shown has an average personal exposure measurement of 9.99 ppm - just below the hypothetical 10 ppm threshold limit value (TLV) of that component. The second has an average exposure of 3.33 ppm and the third 1 ppm. Questions: 0
Do all of these results demonstrate compliance with the limit value?
0
What should the occupational hygienist/factory manager’s response be to each of these situations?
In fact, the first case (average TLV 9.99 ppm) is an example of clear non-compliance. If the average exposure measurement falls on or just below the limit value, half of
164
5 Practical Aspects of Monitoring Volatile Organics in Air
Probability of result occurring for a given atmosphere
T
Calculated average concentration atmosphere 1
l/lOth limit level Log of measured air concentrationor exposure level
Calculated average concentration atmosphere 2
Il3rd limit level
Calculated average concentration atmosphere 3
limit level
Fig.5-26. Log plots of personal exposure measurements collected at 3 different locations in a hypothetical factory.
the workers monitored must, statistically, have been exposed to levels above the limit value. This is obviously not a satisfactory situation and the response of the occupational hygienist/factory manager must be to reduce exposure levels by some means. By the same token, an average exposure measurement of 1/3rd the limit value would, typically, statistically mean that 95 070 of the individual results collected were below the limit level, ie, one in twenty workers would, on average, still be exposed to levels above the limit value in such an atmosphere. This is generally considered to be a case of borderline compliance and typically demands one of two responses. 1. Bring levels down if at all possible. 2. If exposure levels cannot be significantly reduced without incurring excessive cost then a very regular routine monitoring regime (eg, weekly) must be maintained in order to ensure levels do not start to rise.
The third example, with an average exposure of l/lOth the limit value, demonstrates good compliance. An individual working in such an atmosphere would, statistically, have negligible risk of being exposed to concentrations above the limit. If no changes were made to the industrial operation being carried out in that environment, these low levels would mean that routine monitoring would normally only be required on an infrequent (eg, annual) basis.
5.6.2 Industrial Emission Testing International directivesAegislation and World Health Organisation (WHO) guildelines affecting industrial emissions have led to more stringent controls on VOC emissions from industrial sites. Some of the new international regulations impact emissions to air, water and soil in a combined approach called ‘Integrated Pollution
5.6 VOC Air Monitoring Applications
165
Control' (IPC). Industrial sites affected by emission control legislation will be granted operating licenses only if they are shown to be in compliance with strict pollution emission limits. Pollution emission from a given industrial site may be calculated using mass balance estimations, ie, comparing the mass of raw materials going in versus the mass of finished goods produced. However, an increasing level of VOC monitoring will also be required to confirm and supplement this calculated data. Testing the emissions from a given source usually requires both stackhent monitoring and perimeter fence line measurements. Thus both concentrated and trace level data may be required. The wide range of different air monitoring applications involved may also demand a variety of different air sampling strategies, ie, sorbent tubes, whole air containers and on-line semi-continuous measurements. However, no personal exposure data is required and most sampling methods can thus be automated. The wide polarity and volatility range of VOCs emitted by many pollution sources usually means that sorbent tubes provide the most flexible sampling medium for this work. However, such devices can only be used as part of a sample 'train' for stack emission testing, incorporating particle filters and condensation units, etc., before the sorbent tube. The sampling stage can often be automated by pumping a flow of stackhent gas (typically 400 mL min-' to 5 L min-') to a sequential sampling unit. Most commercial sequential sampling units allow sampling times to be selected from a few minutes to several hours per tube. The example unit described above (see Sections 3.2.3) has capacity for up to 25 sorbent tubes. Fence-line monitoring can also be automated using appropriately configured sequential tube samplers. As the type of industrial stackhent or mobile emission source may vary considerably from example to example it is usually recommended that professional advice be sought regarding optimum positioning of the sampling point and concerning the specific hardware components which should be included in the sample train. The potential advantages of semi-continuous on-line monitoring and analysis are obvious in terms of offering real time data for both perimeter fence and stack emission testing. However, the costs associated with placing an entire TD - GC analytical system at several different monitoring locations is usually prohibitive unless some particularly hazardous analyte is involved. It is of course also possible to monitor 24 h day-' (ie, round-the-clock) using sequential tube sampling and automated, offline centralized analysis. Perhaps the most flexible, cost effective arrangement is to have a single versatile analytical system containing a focusing device which is compatible with a full range of air sampling strategies. This can then be used for direct on-line air stream analysis part of the time and automatic off-line tube or canister analysis at other times. Canisters or other whole air containers are rarely used for industrial emission testing unless ultra volatile components are of interest (NB, components such as C, hydrocarbons are not quantitatively retained in conventional sorbent tubes) or unless specified by a particular standard method. Long term (2 week) diffusive tube sampling methods, currently under evaluation for perimeter fence line measurements [39], may also provide a cost effective alternative to pumped tube, on-line or canister methods in the future.
166
5 Practical AsDects of Monitoring Volatile Organics in Air
5.6.3 Mobile Emission Sources, ie, Vehicle Exhaust Testing Vehicle exhaust emissions from gasoline (petrol) engines have conventionally been analyzed using Tedlar bags to collect the sample and a GSV to introduce a small aliquot of the exhaust vapors from the bag into a capillary GC analytical system. However, a wider volatility, polarity and concentration range of analytes can be monitored if sorbent tubes and thermal desorption-GC methods are used to collect and analyze the samples (Fig. 5-27). Tube sampling of vehicle exhaust emissions can be automated using commercial stream selection modules and sequential tube samplers. Diesel particulate emissions from diesel engines are conventionally collected on glass fiber filters, desorbed with solvent and injected into short capillary GC analytical columns. Both diesel and lubricating oil fractions are usually of interest (boiling range n-Clo to n-C40) and although manual thermal desorption methods have been
Fig. 5-27. Car exhaust emission ( - 8 ppm VOCs) collected on a sorbent tube and analyzed using a dual capillary column, pressure switch system with dual FID. AUTOAO5 shows the analysis of C, to C , volatile components using an Alumina PLOT column. AUTOBOS shows the analysis of C, to C , , components on a methyl silicone column. (See Section 5.6.4.1 for more details on the chromatographic system used).
5.6 VOC Air Monitorina Amdications
167
successfully developed for extracting the analytes from portions of the glass fiber filters the procedure is hard to automate with such high boiling components. Automation is possible with quantitative recovery up to about n-C36 however, these analytes are a ‘borderline’ case for thermal desorption. If the sample contains ~~, fluid extraction or lubricating oil components less volatile than ~ I - Csupercritical conventional solvent extraction methods are more appropriate.
5.6.4 Urban Air Quality Current concern regarding poor air quality in major cities worldwide has led to increased investment in urban air monitoring programs by many national governments. Until recently these programs have been primarily restricted to monitoring inorganic gases, ie, NO,, SO, and ozone. However, recent research has shown that VOCs can contribute significantly to urban air pollution. Unsaturated hydrocarbons are specifically thought to contribute to the formation of ozone and photo-chemical smogs at ground level in urban and industrialized areas [9]. Speciated VOC monitoring applications in the urban air quality field tend to fall into two important, overlapping categories: 1. Hydrocarbons ranging in volatility from Cz to n-CIo. These are conventionally referred to as ‘Ozone Precursors’. 2. Halogenated and aromatic hydrocarbons ranging in volatility from freon to trichlorobenzene, ie, ‘Air Toxics’. 5.6.4.1 Ozone Precursors
The 1990 US Clean Air Act Amendment now requires that ambient air be monitored for ozone and its precursors in key urban areas which fail new air quality standards (Fig. 5-28). Similar recommendations have also been made in Europe following the 1992 Ozone Directive [40] and the UNECE (United Nations Economic Commission for Europe) protocol on controlling VOC emissions. Relatively complex lists of C2-Clohydrocarbon target analytes are available both from US EPA and European legislators (Tables 5-4 and 5-5) and are generally required to be monitored on-line in automated, unattended field monitoring stations [ 11. Multiple monitoring stations (sometimes known as Photochemical Air Monitoring Stations (PAMs)) are planned for many of the worst affected cities worldwide and will be located according to prevailing wind direction (Fig. 5-29). The ideal field-based on-line ozone precursor analyzer provides quantitative trapping and stable chromatographic separations of complex target analyte mixtures without using any liquid cryogen. Liquid coolants are prohibitively expensive and logistically unsuitable for use in unattended field monitoring stations. Cryogen free focusing devices are commercially available for trapping the VOCs as has been discussed above, however, such complex target analyte mixtures also present a significant chromatographic challenge to the analytical system.
168
5 Practical Aspects of Monitoring Volatile Organics in Air
Areas Designated Nonattainment for Ozone
U
Fig. 5-28. US areas failing new air quality standards.
Fig. 5-29. Location of ozone precursor monitoring stations in and around an urban area.
5.6 VOC Air Monitoring Applications
169
Table 5-4. US EPA Target VOC ozone precursors. Ethene Ethylene Ethane Propylene Propane iso-Butane n-Butane trans-2-Butene I-Butene iso-Butene cis-2-Butene Cyclopentane iso-Pentane n-Pentane
2-Methyl-2-butene Cyclopentene trans-2-Pentene 3-Methyl-t-pentene t -Pentene cis-2-Pentene 2,2-Dimethylbutane 3-Methylpentane 2-Meth ylpentane
2,3-Dimethylbutane Isoprene 4-Methyl- t -pentene 2-Methyl-I-pentene n-Hexane
trans-2-Hexene cis-2-Hexene Methylcyclopentane 2,4-Dimethylpentane Benzene Cyclohexane 2-Methylhexane 2,3-DimethyIpent ane 3-Methylhexane 2,2,4-Trimethylpentane n-Heptane Methylcyclohexane 2,3,4-Trimethylpentane Toluene
2-Methylheptane 3-Methylheptane n-Octane Ethyl benzene m-Xylene p -X y 1ene Styrene o-Xylene n-Nonane iso-Propylbenzene n-Propylbenzene 1,3,5-Trimethylbenzene 1,2,4-Trimethylbenzene
Table 5-5. Preliminary European list of target VOC ozone precursors. Ethane Ethylene Acetylene Propane Propene n-Butane i-But ane
1 -Butene trans-2-Butene cis-2-Butene n-Pentane i-Pentane trans-2-pent am
cis-2-Pentene
Isoprene n-Hexane 2-Methylpentane 3-Methylpentane n - Heptane Benzene Toluene
Ethyl benzene o-Xylene m-Xylene 1,2,4-Trimethylbenzene 1,3 ,STrimethylbenzene
Conventional, single column capillary separations of ozone precursors have either been carried out using apolar columns with a subambient starting point, or using an Alumina PLOTTMcolumn. Neither approach is satisfactory: the former requires cryogenic fluid to cool the GC oven and is still unable to resolve the C, components - the latter offers good resolution of the most volatile species, but retains components less volatile than xylene. This can lead to loss of high boiling components of interest and severe problems of contamination and chromatographic instability. These limitations may be overcome by using a pressure switching facility to separate the analytes on two columns (Fig. 5-30). Chromatographic conditions are given in Table 5-6. Using this approach components desorbed from the focusing trap are transferred to the initial methyl silicone capillary column. Early eluting peaks which are poorly resolved on this column are allowed to pass onto the Alumina PLOT column for more effective separation. No cryogenic fluid is required. When less volatile, well resolved components start to elute from the methyl silicone capillary the effluent from this first column is switched through uncoated, deactivated fused silica to a second FID using a Dean’s pneumatic switch [41]. The switch occurs at around 13- 14 min. Carrier gas continues to pass through the Alumina PLOT column to the
170
5 Practical Aspects of Monitoring Volatile Organics in Air
V 50 m, 0.22 rnrn id., lprn phase
0.32 mrn i d . AI203
Na2S04PLOT column
,Uncoated, deactivated fused silica
6 6 FID 1
FID 2
Fig. 5-30. Dual capillary column, pressure switch system for the complete chromatographic resolution of C, to n-C,, hydrocarbon ozone precursors.
first FID after the switch, thus both chromatographic analyses proceed in parallel from then on. The excellent resolution obtained from such a system is shown in Fig. 5-31.As no valve is involved in switching the direction of effluent flow from the first column, the switch occurs instantaneously with no apparent baseline disturbance. The two column, multidimensional separation described offers considerable advantages to on-line analysis. Key points include: No cryogenic GC oven cooling required. Complete protection of the sensitive PLOT column from high boiling contamination. Long term stability of chromatographic performance and retention time precision. Optimum chromatographic resolution.
Table 5-6. GC conditions for dual column, pressure switch analysis of VOC ozone precursors. ~~
GC oven
45 "C isothermal for 15 rnin 5°C per rnin to 170°C
15 "C per rnin to 200 "C 200°C for 5 min Pressure switch time Dual FID detection Helium
- 13 min Carrier gas at 48 psi (331 kPa) Mid-point gas at 23 psi (159 kPa)
171
5.6 VOC Air Monitoring Applications
GC starting temperature 45°C Run time. 40 minutes
.'9
Column 1 AI,O, Na,SO, PLOT
I
8
m
=
ge
7
5n 2
y_u
5
I Column 2: Methylsilicone
12 mV
10
8
6
0
4
8
12
16
20
24
28
32
36
40
44
5
Time (minutes)
Fig. 5-31. Dual column, pressure switch analysis of C, to n-C,, hydrocarbon ozone precursors.
The two column approach also means that liquid cryogen is totally eliminated from the focusing and chromatographic processes. This further enhances the stability and reliability of the analytical system and simplifies peak recognition during post run data processing. Operator intervention should only be required for changing GC gases, replenishing the calibrant gas and manipulating batches of data. Air is usually sampled for up to 40 - 45 min per hour and detection limits (FID) of 10- 100 ppt are routinely achieved. Such semi-continuous on-line analysis of ozone precursors and other speciated VOCs is now widely and routinely used in both outdoor and indoor environments to monitor the diurnal variation in pollutant concentrations (Fig. 5-32). Data generated at unattended field monitoring stations is conventionally transferred to a central computer system via modem. Many countries including the UK, France, Ger-
172
5 Pracfical Aspects of Monitoring Volatile Organics in Air
I
Hours Fig. 5-32. Diurnal variation of toluene in indoor air over 250 h.
many, USA etc., thus operate networks of ambient air quality monitoring stations generating real time information. 5.6.4.2 Air Toxics
The USA has set the pace for ambient air toxic monitoring with US EPA Method TO-I4 [7] which focuses on apolar halogenated and aromatic hydrocarbons (Table5-7) and Title 3 of the 1990 US Clean-Air Act Amendment which includes polar and apolar VOCs. Extensive evaluation work relating to the application of passivated stainless steel canisters for sampling apolar TO-I4 target analytes has been carried out by the US Table 5-7. US EPA TO-14 target compounds. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
Freon 12 Methyl chloride Freon 114 Vinyl chloride (1,3-Butadiene) Methyl bromide Ethyl chloride Freon 11 Vinylidene chloride Dichloromethane 3-Chloropropene Freon 113 1,l-Dichloromethane
15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. cis-l,2-Dichloroethylene 28.
Chloroform 1,2-Dichloroethane l,l,l-Trichloroethane Benzene Carbon Tetrachloride 1,2-DichIoropropene Trichloroethylene cis-l,3-Dichloropropene trans-I ,3-Dichloropropene 1,1,2-TrichIoroethylene Toluene 1,2-Dibrornoethane Tetrachloroethylene Chlorobenzene
29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42.
Ethylbenzene m and p-Xylene Styrene 1,I ,2,2-Tetrachloroethane o-Xylene 4-Ethyl toluene 1,3,5-Trimethyl benzene 1,2,4-Trimethyl benzene Benzyl chloride m-Dichlorobenzene p-Dichlorobenzene o-Dichlorobenzene 1,2,4-Trichlorobenzene Hexachlorobutadiene
5.6 VOC Air Monitoring Applications
173
Fig. 5-33. Sorbent tube configurations for monitoring wide boiling range samples.
EPA in recent years. However, outside the USA these compounds are more commonly monitored using sorbent tubes [9,42,43]. Recent interest in polar VOCs (esters, ketones, glycol ethers, alcohols, etc.) which cannot be collected in canisters, has also increased the general acceptance of sorbent tube sampling methods for air toxics. Most air toxic samples are best collected by pumping 1 - 10 L air volumes through a tube or tubes containing a series of sorbents of increasing sorbent strength. Detection limits in the order of low ppt are comfortably achieved in most cases. Typical sorbent combinations are illustrated in Fig. 5-33 [13, 16). Safe long term storage has been reported for both low level samples and blank tubes (Figs. 5-34 and 5-35). This
I
/
Ion Range 34 to 200 amu
1
Ion Range 34 to 200 amu
! 9
4
6
8
10
12
14
16
18 20
22
T h e (rnln)
24
26
28
30
32
34
36
38
40
Fig. 5-34. The analysis of two Carbotrap C, Carbotrap and Carbosieve SIII tubes: (A) Used to collect a 2 L Himalayan air sample and (B) opened and immediately reclosed at monitoring site. Total storage time for both tubes between conditioning and analysis was 2 months.
-
174
5 Practical Aspects of Monitoring Volatile Organics in Air
- Sample Tube Capped with Swagelok Type Caps and PTFE Ferrules
- Artifacts are at the 0.1 ng Level
I
1
Fig. 5-35. Blank Tenax GR tube after 6 months storage at room temperature using 1 /4-inch brass Swagelok-type caps and PTFE ferrules.
is supported by data demonstrating safe storage of 800 ng level samples for up to 27 months [44]. The thermal desorption-GC-MS analysis of 10 ppb polar and apolar air toxics from a sorbent tube is shown in Fig. 5-19. The key advantages of tubes versus canisters for this particular analyte range include: Lower capital and transportation costs. No complex sampling apparatus required. - Wide analyte polarity and volatility range. - Improved detection limits. - No sampler cleaning required before reuse. -
-
If the full range of air toxic analytes is to be monitored, strong, water retentive sorbents will be required in the tubes. Similarly if either on-line or canister air sampling is used water will also be collected. In each of these cases some form of sample drying step will be required before capillary GC-MS analysis. As all air toxic target analytes are quantitatively retained at temperatures above 0 "C by the type of sorbent filled focusing trap described above (see Section 5.2.2), dry purging procedures can be used to eliminate water. This technique is compatible with polar and apolar components and can be used with all air sampling methods (see Section 5.4). NB. Mass spectrometer detection is invariably preferred for air toxics analysis. This essentially precludes field monitoring.
5.6.4.3 Roadside Air Concentrations of PAHs As a final example of an urban air monitoring application, thermal desorptioncapillary GC-MS analysis methods have now also been developed for PAHs ranging from naphthalene to 5 ring components in roadside air [45]. PAHs are present in the exhaust emissions of most vehicles. Air samples are typically collected using high pump flow rates (500 mL min-l) through tubes packed with Tenax GR or another weak, thermally stable sorbent. High desorption temperatures and fast gas flow rates are required to extract the analytes.
5.6 VOC Air Monitoring Applications
175
5.6.5 Indoor Air Quality Research into ‘Sick Building syndrome’ (SBS), general human exposure and improved building design requires detailed information on air quality. In this instance, however, concentration data for VOCs in the environment means very little without supplementary information relating to building ventilation rates, outdoor air concentrations and possible emission sources, eg, building materials, furnishings, domestic products, etc. Special VOC air monitoring techniques are used in each of these individual research areas. Figure 5-36 shows the analysis of a sample of nursery school air collected onto a sorbent tube using diffusive monitoring. Sorbent tubes are invariably used in the indoor environment because of their portabilty and cost efficiency. Most monitoring work is still carried out using pumped monitoring although recent studies using long term (3 -29 day) diffusive sampling of indoor air have demonstrated the usefulness of this technique [27]. The simplicity of diffusive sampling has considerable advantages when monitoring inside schools, homes and offices, etc. 188
TOT
600
CHRO,
5:-
1200 10:m
1809
15:oB
24-88 2R. 00
3006 25:08
Fig. 5-36.Indoor air at a Swedish preschool.
5.6.5.1 Building Ventilation Building ventilation rates are usually assessed using SF, or perfluorocarbon (PFC) tracer gases. Sources of these compounds are typically placed in different areas/ rooms of the building under test and the changing atmospheric concentration over time allows air movement to be measured. Although SF, was most widely used in early ventilation research most modern programs now use PFCs which are more readily adsorbed and monitored using conventional thermal desorption-capillary GC and ECD detection [46]. Several different PFCs are also available for use and this has the additional benefit of allowing different compounds to be placed in different parts of the building (Fig. 5-37). The interchange of air between rooms can then be assessed more accurately. The capillary GC analysis of a standard containing three of the most popular PFCs is shown in Fig. 5-38. Diffusive monitoring methods have been developed for these compounds using Carbotrap tubes and have been used for field studies of air quality in domestic homes and offices [47].
’
176
5 Practical Aspecfs of Monitoring Volatile Organics in Air
First floor
Ground floor 1 uving r m
PDfCP
PMCP PMCH PDCB
= Perfluoromethylcyclopentane
= Perfluoromethylcyclohexane = Perfluorodimethylcyclobultane
Fig. 5-37. Measuring ventilation and interchange of air between rooms in a domestic dwelling. Reproduced with the kind permission of Dr. Henk Bloemen, RIVM.
Pertluoromethylcyclopentane
PerfluoromethyIcyclohexane
Perfluorodimethylcyclobutane
1
”
1
J.
50 m, 0-32 mm Alumina Plot column GC over: 50°C for 1 minute 3.5”CImin to 180°C 180°C for 7 minutes ECD detection
1 8
1
,
1,
5.6 VOC Air Monitoring Applications
177
5.6.5.2 Building Materials Emission Testing
Emission testing for indoor air pollution sources has focused on the development and application of reliable sampling devices for the materials concerned [48,49]. One example of a commercially available, portable sampling device which can be used for both field and laboratory experiments is the FLEC (Field and Laboratory Emission Cell) [48]. A fixed flow rate of pure air is passed through the cell over the material under test and thence through a sorbent tube which concentrates organic vapors emitted by the sample. The FLEC was developed by the Danish National Institute for Occupational Health and has been extensively used for building materials research [50,51]. An example of the emission profile of linoleum sampled using the FLEC is shown in Fig. 5-39.
50
1
U C headspace over wet linoleum surface "
I
Butyrk acid
"
Hexanoic acid
Fig. 5-39. Analysis of damaged linoleum flooring using the Field and Laboratory Emission Cell (FLEC).
5.6.5.3 Monitoring Polychlorinated Biphenyls (PCBs)
The analysis of trace level polychlorinated biphenyls (PCBs) is another important air quality application in certain indoor environments. Several years ago, PCBs were extensively used in the manufacture of sealing compounds used in building construction - ie, around door frames, etc. Most PCBs are very high boiling compounds, but some, including PCBs 28, 52, 101 and 153 are sufficiently volatile to be present in the vapor phase in significant atmospheric concentrations. Using a method developed at LFUG in Mainz, Germany, air samples can be pumped through Tenax GR tubes at rates exceeding 1 L per minute for several hours. Thermal desorptioncapillary GC-MS analysis is then used to measure PCB concentrations [52]. Typical analytical results are shown in Fig. 5-40. Method performance is shown in Fig. 5-41. Extensive testing has shown that target analytes do not break through the sorbent bed during the sampling process.
178
5 Practical Aspects of Monitoring Volatile Organics in Air
PCB-52
PCE-28
Fig. 5-40. Part of the Total Ion Chromatogram of a concentrated real air sample (total PCB concentration 7 ng L-I) collected near a sealing compound. loo x
Box
wx
40%
20%
ox
t
t
PCB-KWgm PCB Conpner
Fig.5-41. Percent recovery from a Tenax GC tube spiked with 50ng of each PCB congener.
5.6.6 Atmosphere Research Applications Current atmospheric research programs requiring speciated VOC concentration data include long range pollution transport studies (The Cooperative Programme for Monitoring and Evaluation of the Long Range Transmission of Air Pollutants in Europe (EMEP) and The Cooperative Program for Monitoring and Evaluation of the Long Range Transmission of Air Pollutants in America (EMAP)), air-sea water VOC concentration equilibria [42] and biogenic emissions, ie, natural VOC emissions from plants and compost [53- 551.
5.6 VOC Air Monitoring Applications
400 3:21
800 6:41
179
1200
1O:Ol
13:21
16:41
Fig. 5-42. Thermal desorption-GC-MS analysis of a monoterpene standard from a Tenax tube. Results presented courtesy of Dr. Kristensson, University of Stockholm, Sweden
Biogenic emissions studies do present a significant analytical challenge as the compounds of interest - primarily monoterpenes and isoprene - are relatively labile and require careful, generally sorbent tube based, trapping techniques. Conventionally, spherocarb adsorbent is used for isoprene and Tenax [53] or Carbotrap [56] are used for monoterpenes. MS detection is invariably preferred and detection limits for 3 L air samples typically range from 20 to 50 ppt using full scan mass spectrometer detection. Pumped sampling methods are preferred. A standard chromatogram of monoterpenes desorbed from Tenax is shown in Fig. 5-42. Research into VOC equilibria between air and surface sea water benefit from the fact that both air and sea water samples can be analyzed using essentially the same methodology. VOCs from water (or sediment) samples are simply purged with inert gas onto sorbent tubes and then subjected to the same analytical procedures as air samples [14, 15,421. Detection limits as low as 5 ppt VOC in water have been reported using such methods. As international programs for pollution transport research, such as EMEP and EMAP, are focusing on only a limited number of apolar target VOCs, including C2 hydrocarbons, these samples are primarily collected using canisters. However, most of these research projects are long term (10 year programs) with only a very limited number of samples being collected and analyzed on a weekly basis. The logistical limitations of canister sample collection and transportation are thus less significant in this case. This concludes the overview of speciated VOC air monitoring methods and applications. Many other air monitoring applications examples could have been included and related fields including flavor and fragrance analysis could have been discuss-
180
5 Practical Aspects of Monitoring Volatile Organics in Air
ed, but it is hoped that the methodology described adequately covers the most important aspects of speciated VOC air monitoring.
5.7 Concluding Remarks The Challenge
Answers to the questions listed at the start of this chapter remain to be discussed:
- What pollutants are present? -
At what concentration?
- Where are they coming from? -
What harm do they do?
A vast number of VOC concentration measurements and pollution source emission profiles are required before the natural balance of chemicals in the atmosphere is understood and before the extent and potential hazard of manmade pollution can be correctly assessed. The extent of our current ignorance in this field and the size of the analytical challenge it presents is somewhat daunting. However every individual accurate speciated VOC measurement made and reported contributes to improving our current understanding and will help to clarify what if any air pollution remedial action is required on a global scale.
Post Script: The author should not comment regarding the likelihood of necessary remedial action being taken once the required steps have been identified. The challenge faced by politicians at that stage may well outweigh that faced by environmental science at the current time!
References [I] US EPA, Office of Air Quality Planning and Standards, Enhanced ozone monitoring network design and siting criteria guidance document, EPA-450/4-91-033, 1991. [2] UK Health and Safety Executive, Methods for the Determination of Hazardous Substances No. 72, Volatile organic compounds in air. Laboratory method using porous polymer adsorption tubes and thermal desorption with GC analysis, 1992. [3] Callan, B., Walsh K., Dowding P., Chemistry and Znduslry 1993, April, 250-252. [4] Canela, A.M., Weaver, T. R., Determination of alkylated benzenes in air by headspace GC, Presented at the Air and Waste Management Assoc. AGM, Vancouver, 1991. [5] Brown, R. H., UK Health and Safety Executive, Private communication. [6] US EPA study of ambient air quality in Atlanta, Georgia, 1990. [7] US EPA Compendium Method TO-14, The determination of VOCs in ambient air using Summa@passivated canister sampling and GC analysis, 1988.
References
18 1
[8] Coutant, R. W., McClenny, W. A., Competitive adsorption effects and stability of VOC and PVOC in canisters, Proceedings of 1991 US EPA/AWMA International Symposium: Measurement of Toxic and Related Air Pollutants, Vol I , pp382-388. [9] Ciccioli, P., Cecinato, A., Brancaleoni, E., Frattoni, M., Liberti A., J High Res Chromatogr 1992, 15, 75-84. [lo] Leighton, P., Diffusive Monitor 1988, May, 1. [l 11 Perkin-Elmer Thermal Desorption Data Sheet 10, A guide to adsorbent selection. [I21 Ciccioli, P., Brancaleoni, E., Cecinato, A., Sparapini, R., Frattoni, M., J Chromatogr 1993, 643, 55-69. [I31 Coker, D. T., Van der Hoed, N., Sauders, K. J., Tindle, P. E., Ann Occup Hyg 1989,33, 15-26. [I41 Bianchi, A., Varney, M.S., Phillips, J., J Chromatogr 1989, 467, 111-129. [I51 Bianchi, A., Varney, M.S., Phillips, J., J Chromatogr 1991, 652, 413-450. [16] Perkin-Elmer Thermal Desorption Application Note 39, Monitoring US EPA Method TO-14 air toxics using automatic sorbent tube sampling and thermal desorption. [I71 UK Health and Safety Executive, Methods for the Determination of Hazardous Substances No. 40, Toluene in air. Laboratory method using pumped, porous polymer adsorbent tubes, thermal desorption and GC. [18] Wright M. D., Diffusive uptake rates for the Perkin-Elmer tube-BCR air sampling intercomparisons at VITO (Mol, Belgium), February 1991-April 1992, UK HSE Report IR/L/IA/93/3, March 1993. [I91 UK Health and Safety Executive, Methods for the Determination of Hazardous Substances No. 66, Mixed hydrocarbons (C, to Clo)in air. Laboratory method using porous polymer diffusion samples, thermal desorption and GC. [20] UK Health and Safety Executive, Methods for the Determination of Hazardous Substances No. 50, Benzene in air. Laboratory method using porus polymer diffusive samples, thermal desorption and GC. [21] CARIWorking Group 5 , The Diffusive Monitor 1993, October, 6. [22] Provisional European Norm PrEN838, Air quality - workplace atmospheres - standard practice for validation of diffusive samplers for the determination of gases or vapours. [23] UK Health and Safety Executive, Methods for the Determination of Hazardous Substances No. 27, Protocol for assessing the performance of a diffusive sampler. [24] Cassinelli, M. E., Hull, R. D., Crable, J. V., Teass, A. W., US National Institute of Occupational Safety and Health (NIOSH), Protocol for the evaluation of passive samplers, Diffusive Sampling: An alternative approach to workplace air sampling, CEC Pub1 No. 10555EN, 1987. [25] Van den Hoed, N., Van Asselen, O.L. J., Ann Occup Hyg 1991, 35 (3), 273-285. [26] Nordstrand, E., Kristensson, J., American Industrial Hyg Assoc J 1994, 55 (lo), 935-941. [27] Brown,V.M., Crump, D.R., Gardiner, D., Yu, C.W.F., Environmental Techno1 1993, 14, 77 1- 777 [28] McClenny, W.A., Pleil, J.D., Oliver, K.D., J A i r Pollut Control Assoc 1987, 37, 244-248. [29] Tipler, A., Water management in capillary GC air monitoring systems, Presented at AWMA Conference: Measurement of Toxic and Related Air Pollutants, May 1994. [30] Tipler, A., Dang, R., Hoberecht, H., A system for the determination of trace level polar and non-polar toxic organic compounds in ambient air, Presented at AWMA Conference: Measurement of Toxic and Related Air Pollutants, May 1994. [3I] Vandendriessche, S., Griepink, B., The certification of benzene, toluene and m-xylene sorbed on Tenax TA in tubes: CRM-112, CEC, BCR, EUR12308 EN, 1989. [32] UK Health and Safety Executive, Workplace Analysis Scheme for Proficiency, 1988. [33] Proposed European Council Directive on the Protection of the Health and Safety of Workers from the Risks Related to Chemical Agents at Work. [34] UK Statutory Instrument No. 1657 on Health and Safety, The Control of Substances Hazardous to Health Regulations, 1988.
182
5 Practical Aspects of Monitoring Volatile Organics in Air
[35] Gray, W. M., O’Sullivan, J., Houldsworth, H. B., Musgrave, N., The use of diffusive samplers to monitor occupational exposure to waste anaesthetic gases, CEC Pub1 No. 10555EN, Brussels-Luxembourg, 1987. [36] Calvin, R.P., House, M., Environ Tech Letters 1989, 9, 563-570. [37] Kristensson, J., The Diffusive Monitor, 1988, 1, 3. [38] Provisional European Norm PrEN482, General requirements for the performance of procedures for workplace measurements. [39] Blakeley, K. C., Atmospheric monitoring of aromatic hydrocarbons in the community around BP Chemicals at Baglan Bay from January to December 1991, BP Branch Report No. 124380. [40] European Council Directive on Air Pollution by Ozone, 1992. [41] Broadway, G. M., An automated system for used, without liquid cryogen, for the determination of VOCs in ambient air, presented at 14th Intl Symposium on Capillary Chromatography, Baltimore, USA, 1992. [42] Bianchi, A. P., Varney, M. S., J Chromatogr 1993, 643, 11- 23. [43] Figge, K., Dommrose, A.M., Fres Z Anal Chem 1988, 332, 606-611. 1441 Lindquist, F., Bakkern, H., Stability of chlorinated hydrocarbons on Tenax, TNO Report No. R90/268 sponsored by the Community Bureau of Reference of the European Commission. [45] Rilly, I., Monitoring Poly Aromatic Hydrocarbons (PAHs) in roadside air using automated thermal desoprtion - GC-MS analysis, UK Transport and Road Research Laboratories. Report in preparations. [46] Littler, J., Prior, J., Martin, C., Automation extension and use of the PCL multi tracer gas technique for measuring interzonal air flows in buildings, Report RIB/1985/718, Research in Buildings Group, Polytechnic of Central London. [47] Bloemen, H. J. Th., Balvers, T. T. M., Verhoeff, A. P., van Wijnen, J. H., van der Torn, P., Knol, E., Ventilation rate and exchange of air in dwellings: Development of a test method and pilot study, National Institute for Public Health and Environmental Hygiene, The Netherlands, Report, 1992. [48] Wolkoff, P., Clausen, P.A., Nielsen, P.A., Gustafsson, H., Johnson, B., Rasmusen, E., Healthy Buildings 1991, 160- 165. [49] Gunnarsen, L., Nielsen, P.A., Wolkoff, P., Design and characterization of the CLIMPAQ Chamber for Laboratory Investigations of Materials, Pollution and Air Quality, Proceedings of 6th Intl Conference on Indoor Air Quality and Climate, Helsinki, 1993. [50] Jensen, B., Wolkoff, P., Clausen, P. A., Wilkins, C.K., Characterization of Linoleum: Parts 1 and 2, Proceedings of 6th Intl Conference on Indoor Air Quality and Climate, Helsinki, 1993. [51] Wolkoff, P., Clausen, P. A., Nielsen, P. A., Gunnarsen, L., Documentation of FLEC - Identification of emission processes from carpet, linoleum, paint and sealant, Proceedings of 6th Intl Conference on Indoor Air Quality and Climate, Helsinki, 1993. [52] Perkin-Elmer Corporation, Thermal Desorption Application Note No. 36, Sampling and analysis of PCBs in indoor air using thermal desorption, capillary GC and MS detection. [53] Janson, R., Kristensson, J., Sampling and analysis of atmospheric monoterpenes, Report CN-79, Dept. of Meteorology, Stockholm University, 1991. [54] Robertson, G. W., Griffiths, D. W., Macfarlane-Smith, W., Butcher, R. D., Phytochemical Analysis 1993, 4, 152-157. [55] Ciccioli, P., Brancaleoni, E., Frattoni, M., Cecinato, A., Brachetti, A., Atmospheric Environment 1993, 27A (12), 1891- 1901. (561 Ciccioli, P., Brancaleoni, E., Cecinato, A., DiPalo, C., Brachetti, A., Liberti, A., J Chromatogr 1986, 351, 433 -449.
183
Appendix A
Appendix A Diffusive Uptake Rates on Perkin-Elmer Sorbent Tubes Hydrocarbons ~
Compound
Sorbent
Confidence level
Uptake rate (ng ppm-' min-'1
1,3-Butadiene n-Pentane
Mol. sieve 13x Chrom. 106 Carbopack Ba Chrom. 106 Tenax TA Tenax TA Porapak Q Tenax GR Chrom. 106a Chrom. 106 Tenax TAa Carbotrap B Tenax TAa Tenax TA Tenax GR Chrom. 106 Carbopack B Chrom. 106 Tenax TAa Tenax TAa Tenax TA Chrom. 106 Tenax GR Tenax TAa Tenax GR Chrom. 106 Porapak Q Tenax TAa Chrom. 106 Chrom. 106 Tenax TAa Chrom. 106 Tenax TAa Chrom. 106 Tenax TAa Tenax TA
Fully validated Fully validated Lab. trials Fully validated Fully validated Lab. tested Fully validated Lab. tested Lab. tested Fully validated Fully validated Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Fully validated Fully validated Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Calculated Fully validated Lab. tested Fully validated Fully validated Calculated (ideal) Calculated (ideal) Calculated (ideal) Calculated (ideal) Lab. tested
1.3 1.46 1.77 1.77 1.3 (8 h) 0.64 (29 days) 1.37 1.81 1.72 1.95 1.77 1.94 1.67 (8 h) 1.2 (29 day) 2.12 1.94 2.06 2.13 2.00 1.82 (8 h) 1.91 (29 day) 2.10 2.48 2.0 2.43 1.9 2.38 2.0 2.15 2.40 2.12 2.26 2.26 2.37 2.37 2.65 (29 day)
n-Hexane Benzene
n-Heptane Toluene
n-Octane Xylene
Ethyl benzene
Styrene n-Nonane iso-Propyl benzene Trimethylbenzene
184
5 Practical Aspects of Monitoring Volatile Organics in Air
Compound
Sorbent
Confidence level
Uptake rate (ng ppm-' min-')
n-Decane
Tenax TA Tenax TA Tenax TA Porapak Q
Fully validated Lab. tested Lab. tested Calculated
2.3 (10min-8 h) 2.96 (29 day) 3.38 (29 day) 2.5
n-Undecane Cumene
Halogenated Hydrocarbons Compound
Sorbent
Confidence level
Uptake rate (ng ppm-' min-')
Methyl chloride Vinyl chloride 1,l-Dichloroethene Trichlorotrifluoroethane Chlorotrifluoromethane Dichloromthane
Spherocarb Spherocarb Spherocarb Chrom. 102
Lab. Lab. Lab. Lab.
tested tested tested tested
1.3 2.0 2.5 3.5
Chrom. 102
Lab. tested
1.8
Chrom. 106 Chrom. 102 Chrom. 102 Chrom. 102 Tenax TA Tenax TA Tenax TA Chrom. 106 Tenax GRa Chrom. 102 Tenax GRa Chrom. 102 Chrom. 106 Chrom. 102a Chrom. 106 Chrom. 102 Tenax GRa Chrom. 106 Tenax TAa Chrom. 102 Chrom. 106 Spherocarb
Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Calculated Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Calculated
1.5 1.56 1.9 3.6 2.59 2.29 2.20 2.45 2.1 8 2.35 3.72 2.87 2.5 2.3 2.3 2.3 2.92 3.1 2.8 2.6 2.45 3.O
1,2-Dichloroethane Fluothane Halothane Enflurane Isoflurane Bromoethane Trichloromethane (chloroform) Tetrachloromethane (carbon tetrachloride) Trichloroethene l,l,l-Trichloroethane Tetrachloroethane Epichlorohydrin Vinyl bromide
Appendix A
185
Compound
Sorbent
Confidence level
Uptake rate (ng ppm-' min-')
1,2-Dibromoethane Perfluorodimethylcyclobutane Perfluoromethylcyclopentane Perfluoromethylcyclohexane
Chrom. 106 Carbotrap
Lab. tested
5.0 (15+l)mL-'h-'
Carbotrap
Lab. tested
(15k1) mL-'h-'
Carbotrap
Lab. tested
(15k1) mL-'h-'
Esters and Glycol Ethers Compound
Sorbent
Confidence level
Uptake rate (ng ppm-' min-')
Ethyl acetate
Chrom. 106a Tenax TA Tenax TA Porapak Q Tenax TA Porapak Q Tenax TA Porapak Q Chrom. 106 Tenax TAa Chrom. 106 Tenax TAa Chrom. 106a Tenax TA Tenax TA
Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Fully validated Fully validated Fully validated Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Lab. tested Calculated
1.77 1.43 2.26 (8 h) 2.0 2.6 (8 h) 1.5 1.8 2.8 2.3 2.1 2.1 1.9 1.85 1.52 2.1 (10 min-8 h)
Chrom. 106
Lab. tested
2.1
n-Butylacetate Methyl methacrylate Butyl acrylate 2-Methoxy ethanol 2-Ethoxy ethanol 2-Methoxyethylacetate Et hoxy et hylacetate Butoxy ethanol Methoxy propanol Methyl ethoxy ethanol Methyl oxitol
186
5 Practical Aspects of Monitoring Volatile Organics in Air
Aldehydes and Ketones Compound
Sorbent
Confidence level
Uptake rate (ng ppm - min - '1
Acetone Methyl isobutylketone (2-methyl-4-pentanone) Cyclohexanone Furfural
Spherocarb Tenax TA Chrom. 106a Tenax TA Tenax TA
Calculated Lab. tested Lab. tested Calculated Fully validated
1.9 1.71 2.01 2.3 2.5
'
Miscellaneous ~~~~~~~~~~~~~
Compound
Sorbent
Confidence level
Uptake rate (ng ppm-' min-l)
Propan-2-01 Acrylonitrile Acetonitrile
Carbondisulfide Nitrous oxide
Spherocarb Porapak N Porapak N Porapak N Porapak N Porapak N Spherocarb Mol. sieve 5A
Calculated Fully validated Fully validated Fully validated Fully validated Fully validated Fully validated Lab. tested
Ethylene oxide 1,CDioxane
Spherocarb Spherocarb
Lab. tested Calculated (ideal)
2.0 1.35 1.0 (2 h) 0.8 (8 h) 1.4 (2 h) 1.3 (8 h) 2.6 (10 min - 8 h) 1.25 (10 min- 8 h) 1.6
Propionitrile
a
3.O
Preferred sorbent. A nickel disk, rather than the conventional stainless steel gauze, was used to support the Carbotrap sorbent material during method validation for these perfluorocarbon tracer gases. The uptake rates may not be applicable to samples using conventional steel gauzes.
Appendix B
187
Appendix B Calibration Calculation Methods This appendix describes recommended calibration calculation procedures for a range of VOC air monitoring methods. 1. Whole-air sample analysis and semi-continuous on-line air stream analysis. Calibration using an accurate canister standard (a) Introduce several replicates of a series of canister air standards covering the required concentration range. Use the same air introduction conditions in terms of flow, humidity, pressure, sampling time, etc., as for the samples themselves. (b) Calibrate the data handling system in units of concentration. It should not be necessary to enter the air volume used into the data handling method as long as this will remain constant from run to run. (c) Repeat the calibration exercise with a single canister air standard introduction at regular intervals. (d) If the instrumentation used includes a gas sampling valve (GSV) for introduction of gas phase internal standards, internal standard calibration mode should be selected on the data handling system and a fixed aliquot of internal standard gas should be introduced via the GSV during the introduction of each air sample and canister air standard. The system will then adjust the sample concentration results reported for each run depending on the size of the internal standard peak. In other words, if the size of the internal standard peak drops by lo%, the data handling will assume that the system response has fallen off for some reason and will take this into account when reporting the sample component concentrations. If no internal standard introduction facity exists, an external standard calculation should be used. Note 1: Deuterated hydrocarbons make ideal standards for VOC air monitoring methods if MS detectors are used. Note 2: The concentration of internal standard gas selected should be such that if the mass of analyte contained in the GSV loop were present in the volume of air normally sampled, the concentrations in the air would have been representative of field conditions. For example, 0.5 mL injections of 0.1 - 10 ppm standards would be ideal for calibrating 500 mL air samples ranging in concentrations from 0.1 - 10 ppb.
188
5 Practical Aspects of Monitoring Volatile Organics in Air
2. Whole air sample analysis and semi-continuous on-line air stream analysis. Calibration via response-factor-independent or carbon-number type calculation (a) This method is appropriate for target analytes which are all chemically similar or which give a similar response on the GC detector used. (NB. Mass spectrometer detectors typically give similar response factors for a wide range of analytes.) (b) A nominal whole-air canister standard containing all target analytes must first be used to establish target component retention times. (c) Accurate canister standards containing 2 or more of the target analytes in air should then be introduced using the same volume as for the sample. At least 3 replicates should be used at each concentration level. (d) If all the compounds of interest and the internal standard have response factors (RFs) which are identical or very similar, the RF determined by the multilevel calibration described above (c) can then be entered for all of the target analytes. (e) For chemically similar compounds such as hydrocarbons the response of a GC flame ionization detector (FID) is roughly proportional to the number of carbons in the molecule. An RF can therefore be calculated per carbon by multiplying the actual RF determined for the standard (as described in (d) above) by the number of carbons in the standard compound. For example, the RF of a propane standard (3 carbons per molecule) would be multiplied by 3 to give a carbon RF, hexane (6 carbons per molecule) would be multiplied by 6 to give the carbon RF and so on. The carbon RF can then be applied to each target analyte via the GC data handling method and analysis results will be reported in carbon concentration units, ie, ppbC, ppmC, etc. These results can then be converted to ‘true’ analyte concentrations by dividing the carbon concentration unit figures with the numbers of carbons in the target analytes concerned. For example, butane at 12 ppbC would be 3 ppb, toluene at 80.5 ppbC would be 11.5 ppb, etc. Alternatively the carbon RF can be divided directly by the number of carbons in each analyte to give a nominal RF for each component. These nominal RF values can then be entered into the data handling method and will produce ‘true’ concentration data directly in the chromatographic reports. (f) Whichever approach is used [ie, (d) or (e)], the ‘sample amount’ factor in the data handling method can again be ignored provided the volume of air introduced remains constant for each sample and canister air standard.
(g) See point [d] from Appendix B, Part 1.
Appendix B
189
3. Calibration of tube samples
(a) External standards are almost always introduced onto sorbent tubes via liquid injection. It is therefore necessary to calculate what concentration liquid standards are required to inject the masses of analyte which would be retained during pumped or diffusive air sampling. For example, if monitoring an atmosphere containing around 10 ppb benzene (MW 78) by pumping 10 L of air onto a Carbotrap tube, the mass of benzene which would be retained can be calculated using Boyle's Law, as follows: 78 g (1 mole) of benzene vapor occupies -25 L at room temperature.
Therefore, 10 L of pure benzene vapor would contain
- 3 1.2 g benzene.
Therefore, 10 L of air with benzene at 1ppm would contain -3 1.2 pg of benzene. Therefore, 10 L of air with benzene at 10 ppb would contain 312 ng of benzene. A liquid standard should thus be prepared which contains approximately 3 12 ng of benzene per standard injection. Two more concentrated and two more dilute solutions should also be prepared to give a series of 5 liquid standards for multilevel calibration. Alternatively if monitoring a 2 ppm C2C14 atmosphere using 8 h (480 min) diffusion with C2C14 having an uptake rate (U) of 3.1 ngppm-'min-' on Chromosorb 106, the following calculation is then required:
U= ie, 3.1
=
ng ppm x min ng retained 2 ppm x 480 min
therefore the mass of C2C14retained ng = 3.1 ~ 2 x 4 8 = 0 2976 ng A liquid standard should thus be prepared which contains approximately 2976 ng of C2C14per standard injection. Again a series of 5 different concentra-
tion standards should be prepared for multi level calibration purposes. (b) When preparing the standards themselves use the density of the analyte to calculate what volume analyte is required per unit volume standard solution. For example, benzene has a density of 0.879 g mL-' (ie, -4.4 mg 5 pL-'). Therefore if a 5 pL injection of liquid standard solution was to contain 312 ng benzene an approximate dilution of 15000: 1 would be required, ie, a solution in the order of 0.00667% v/v benzene in suitable clean solvent would need to be prepared. Practically speaking this could be carried out most simply by accurately measuring 7 pL of pure benzene into 100 mL pure solvent (ie, 6.153 mg benzene per 100 mL solution).
190
5 Practical Aspects of Monitoring Volatile Organics in Air
A 5 pL injection of this standard solution would contain exactly 307.6 ng. To calculate what atmospheric concentration this mass of analyte would represent in 1OL of air, the following calculation is required: Atmospheric concentration 307.6 ng per 10 L air To convert this to ppm units use the equation atmospheric conc (ppm) =
ie,
mass (pg)x 24.45 air volume x analyte MW (8)
0.3076 x 24.45 10x78
Therefore the atmospheric concentration would have been 0.00964 ppm, ie, 9.64 ppb. Considering the second example of diffusively sampling CZCl4: C2C14 has a density of 1.625 g mL-' (8.125 mg 5 pL-'). If a 5 pL standard solution was to contain -2976 ng of analyte, a dilution of 1 : 2730 would be required, ie, a 0.0366% v/v solution would be required.
-
This could be prepared by accurately introducing 37 pL C2CI, into 100 mL pure solvent (ie, 60.125 mg C2CI4per 100 mL solution). A 5 pL injection of this standard solution would contain exactly 3006.25 ng. To calculate what atmospheric concentration this mass of analyte would represent in air, the following calculation is required: If 3006.25 ng C2C14 was collected diffusively on a sample tube over an 8 h (480 min) exposure time with an uptake rate of 3.1 ng ppm-' min-' then the atmospheric concentration would have been: 3006.25 ng ie, ___ = 2.02ppm conc (ppm) = ___ U x min 31 x480 (c) Once the series of liquid standards has been prepared by introducing different volumes of analyte into clean solvent, and once the calculations of the atmospheric concentrations that these represent have been made, at least three replicate injections should be made at each level.
(d) The calculated atmospheric concentrations represented by the series of liquid standard injections should be entered into the data handling method as 'standard amount'. An appropriate RF value will then be calculated by the system to allow chromatographic results to be printed directly in atmospheric concentration units. Once this multilevel calibration has been completed, routine system checks can simply be made using a single mid-concentration range standard.
Appendix B
191
(e) If it is possible that the sampling conditions may change during monitoring, for example, the exact pumped volume of air or diffusive sampling time may vary from sample to sample, then the data handling method will require the sampling conditions used in step (b) above (ie, 10 L or 480 min in the examples given) to be entered as ‘sample amount’ and as the multiplication factor during calibration. The actual sample amounts (ie, liters of air pumped or minutes diffusively sampled) should then be entered appropriately for each individual sample. GC automation software will usually allow the ‘sample amount’ parameter to be entered for each sample in an automation set. This then compensates for variations in sample size and allows results to be automatically reported in atmospheric concentration units. (f) Some thermal desorbers allow a small fixed volume gas phase internal standard
to be introduced via a GSV onto the back of each tube prior to desorption. A few commercial systems allow this operation to be carried out fully automatically. As in point (d) of Appendix B, Part 1, the mass of internal standard introduced onto each tube should be calculated and equated to an atmospheric concentration, ie, as if that mass had been collected from the volume of air sampled. If this facility is available , ‘internal standard’ calibration should be used and the ‘standard amount’ should be entered into the data handling method in atmospheric concentration units. If the facility is not available, use external standard calibration.
This Page Intentionally Left Blank
6 Quality Control and Quality Assurance Aspects of Gas Chromatography-Mass Spectrometry for Environmental Analysis Ray E. Clement and Carolyn J. Koester
6.1 Introduction The study of the sources, effects, and transport and fate of organic pollutants in the environment has been the subject of intense scrutiny in recent years. Since some of these compounds are considered to be toxic at very low concentrations, analysis methods are required that are capable of achieving accurate, precise determinations at lower than parts-per-billion (ppb; lo-’ g/g) concentrations. In some cases, partsper-quadrillion (ppq; g/g) concentrations must be determined. As environmental legislation in many countries is becoming stricter, in some cases leading to severe fines for companies and even possible jail terms for individuals, it is imperative that the data used to prove pollution events and which are used to identify the persons or companies responsible be defensible in court. Thus, the quality control (QC) and quality assurance (QA) procedures followed are as important as the specific methodologies employed. For the determination of trace concentrations of organics in environmental Samples, the only real method of choice is gas chromatography- mass spectrometry (GCMS). Other techniques are better at providing definitive identification of molecules, but not at the trace concentrations generally found in environmental samples, and not in complex mixtures of hundreds of different compounds. GC-MS is an effective technique for environmental analysis applications because of the tremendous separating power of modern fused silica wall-coated open-tubular (capillary) columns (FSOT), and because mass spectrometers are available that can detect from picogram to femtogram quantities of organic molecules. The technique of GC can only be used to determine those molecules that can be put into the vapor state (about 15-20070 of known organic molecules), but fortunately the majority of the most important organic pollutants fall into this category. Much work is being performed to develop liquid chromatography- mass spectrometry (LC - MS) methods for those substances not easily amenable to GC analysis, but only GC-MS methods will be considered here. As described below, there are many uses of GC-MS in the QC/QA activities required for environmental investigations. Emphasis here is given to the bench-level QC steps needed to achieve acceptable accuracy and precision for qualitative and quantitative analysis, but some discussion is presented on the quality assurance
194
6 Quality Control and Quality Assurance Aspects of GC-MS
aspects of GC-MS analysis, and of the uses and benefits of GC-MS for other aspects of environmental analysis.
6.2 The Use of GC-MS in Environmental Investigations Various aspects of QC and QA to be considered for any GC-MS-based environmental analysis method are shown in Fig. 6-1. With every stage of the analytical process there are many possible QC/QA actions that can be taken. These actions can be associated with the overall study and subsequent data interpretation (global QC/QA), can be specific elements of individual analytical processes (method-specific QC), or are actions to ensure correct instrumental operation (instrument-specific QC). The level of QC and QA needed for specific analytical work depends on the objectives of the study. One can argue that, ideally, the QC/QA operations taken to assure ‘correctness’ must be performed regardless of the nature of the investigation, and therefore are non-negotiable. This is impractical, because some QC operations - such as the analysis of many replicates - may significantly increase the cost of the investigation. Thus, few replicates may be made during a general environmental survey de-
Fig. 6-1. Flowchart showing the general steps in environmental analysis, with emphasis on the principal QC and QA concerns at each stage of the analysis.
6.3 QC/QA Aspects of GC-MS Instrument Operation
195
signed to locate possible contamination ‘hot-spots’, which will then be studied in detail, while many more replicate samples will be analyzed in a study to gather evidence in support of litigation activities. Environmental investigations can be classified into monitoring, survey, litigation, and research categories, as shown in Fig. 6-2, and the QC and QA aspects of using GC-MS must be considered on an individual basis depending on the objectives of each investigation. Notwithstanding the above considerations, many of the requirements for benchlevel QC for GC-MS operation are pretty much the same regardless of the final objectives of the study. In fact, the principal consideration for the GC-MS operator will be whether a quantitative method for specific target compounds, or a qualitative ‘scan’ to identify the compounds present, will be performed. As we are further removed from specific bench-level activities, the use of GC-MS in an overall quality management plan depends more upon the study objectives and cost. For example, the principal quality concerns associated with sampling are whether the sample is representative of the environmental compartment under investigation, whether the specific sample to be analyzed is homogenous, and whether sample preservation during shipping and storage is adequate to ensure contamination, losses, or transformation of analytes do not occur. GC-MS can be used to analyze sample containers before they are used for sampling to ensure they are free of contamination. By GCMS analysis of several aliquots from each sample container, sample homogeneity
t
monitoring
t survey t t
litigation research
representative homogeneous t sample preservation t t
.
representative sub-aliquot extraction & cleanup t blanks
t calibration
t
response factors GC separation
correct analyte identity precision & accuracy t analyte recovery t detection limits t
Fig. 6-2. Flowchart showing the steps to be followed to ensure proper operation of a GC-MS system before its use for environmental analysis.
196
6 Quality Control and Quality Assurance Aspects of GC-MS
may be assessed. Spiked travelling blanks may be analyzed by GC-MS to help assess possible analyte losses during shipping and storage. A complete quality management program would include all of these elements; however, the cost of analysis could be substantially increased. Also, while such QC activities may be important to the overall program, their absence does not prevent the GC-MS operator from achieving excellent precision and accuracy for the determination of analyte concentrations in individual samples.
6.3 QC/QA Aspects of GC-MS Instrument Operation 6.3.1 Quality Control of GC-MS Instrumentation The first step in using GC-MS in quality control is to establish that the instrumental system which is used to detect the analyte of interest is functioning optimally, In order to generate believable data, the GC-MS must be capable of providing adequate chromatographic resolution of complex mixtures, producing reproducible mass spectra, obtaining adequate detection limits for the analytes of interest, and providing acceptable stability of response to target analytes. Since GC-MS is composed of two independent units, both the GC and the MS must be tested before the optimal operation of the GC-MS system can be established. The importance of realizing that two separate units are operating jointly in a GC-MS system is often emphasized by MS service engineers, who frequently complain that 90% of the problems that the user attributes to the MS are, in fact, GC problems. In practice, one is often able to document acceptable GC-MS conditions by tuning the MS and calibrating the mass axis, and then analyzing a multi-component standard which is composed of the analytes in an appropriate solvent. This instrument calibration should be performed at least once each day that samples are analyzed. A multi-component standard of analytes should also be analyzed daily. The results of these daily calibrations should be monitored with time, possibly by plotting analyte response factors in a control chart. A gradual drop in response factors indicates that the GC-MS system sensitivity is falling, and corrective action such as changing the GC column or cleaning the MS ion source may be indicated. Figure 6-2 is a flowchart that describes this process. The GC-MS system performance is generally acceptable for environmental trace analysis applications if the data generated meet the following criteria: 1. Symmetrical capillary column GC peak shape is obtained, and GC peak widths of 1- 10 s (dependent on the column, analyte, and retention time) are observed. 2. GC peaks for all major components are well resolved. 3. GC retention times of analytes are consistent with previous data collected under similar conditions (within approximately f3 s).
6.3 Q U Q A Aspects of GC-MS Instrument Operation
197
4. Instrumental response, as measured by the signal to noise ratio produced by the detector response to the analyte or by the detector response in area counts per mass of analyte introduced into the GC-MS, is consistent with previously collected data. 5. Mass spectra produced by the GC-MS must match closely with the corresponding standard library mass spectra (relative abundance of major mass peaks should be better than f 1 0 % of comparable mass peaks in library mass spectrum).
If the above criteria are not satisfied it can be difficult to determine the causes, as problems can be associated with the GC, MS, or the GC-MS interface. The most difficult problems to troubleshoot occur when poor GC-MS performance results from multiple factors associated with both the GC and the MS simultaneously. Some guidelines for troubleshooting GC-MS performance problems are given below.
6.3.2 Gas Chromatograph Performance If the MS calibration results were typical (for example, no significant changes in MS operating parameters such as extraction lens voltages) and GC-MS data obtained for a known standard were not acceptable, a problem with the GC is indicated. Table 6-1 describes common GC problems which we have encountered and their solutions. More information on GC troubleshooting is available in Refs. [l-51. Often, the quick fixes of changing the injector’s septum, changing the injection port liner, and clipping 0.5 to 1 m from the head of the column to remove non-volatile sample components and active sites from the column head will restore GC-MS sensitivity. If high levels of background noise are observed, baking the column at the maximum temperature at which it is operated (not to exceed the isothermal operating temperature of the column) often eliminates the problem. It is often advisable to analyze a carefully designed (and commercially available) test mixture to assess column performance. Standard test mixtures usually contain analytes which have a wide range of functional groups; the peak shapes and retention times of these analytes are helpful in diagnosing GC problems. Table 6-2 lists compound types found in test mixtures and indicates problems which can be diagnosed by these compound classes. All vendors of GC columns should supply the chromatogram produced when a standard column test mixture was analyzed. Most also include information about the GC program and gas flow rates used during the collection of the data and, also, analyte retention times, partition ratios, and peak widths. By comparing current and past data, column performance can be assessed.
6.3.3 Mass Spectrometer Performance The performance of the MS is typically assessed during the start-up and mass calibration procedures. Although each type of MS has a unique start-up protocol, most mass calibration procedures follow a common set of steps. First, vacuum integ-
198
6 Quality Control and Quality Assurance Aspects of GC-MS
Table 6-1. Commonly encountered GC problems and their solutions. Problem
Principal causes
Solution
Peak tailing
Adsorptive sites Polarity mismatch of the stationary phase, solute or solvent Poor injection technique Some compounds, ie, alcohols, amines, and carboxylic acids tail on most columns GC or column contamination Broken column Plugged syringe Improper injection technique Plugged syringe Leak in injector Improper injection technique Contaminated injection port liner/column Injector leak Incorrect carrier gas flow Contaminated injection port liner/column Septum bleed
Replace liner/column Change solvents
Baseline drift (upward) Irregular peak shape
Peaks reduced in size
Loss of resolution Irreproducible retention times Ghost peak
Check recommended method Use pH modified stationary phase or derivatize compounds
Clean injection port/condition (or change, if necessary) column Clip or replace column Cleadchange syringe Consult GC manual for proper technique Cleadchange syringe Find and fix leak Consult GC manual for proper technique Change/clean liner; clip/change column Find and fix leak Measure and correct Cleadchange liner; solvent rinse column Use higher temperature septum
Table 6-2. Compound classes used in mixtures developed to test GC column performance. Compound type
Comment
Only interact with stationary phase; standards to which all other peaks are compared; can be used to calculate retention indices and the numbers of theoretical plates; poor peak shapes suggest gas flow problems, poor injection technique, column contamination, or damaged stationary phase Interacts with hydrogen-bonding species; thus, indicates column Alcohols activity; tailing peaks indicate active sites in liner or column contamination Usually phenols are used; will tail if column too basic, indicating Acids irreversible adsorption Usually anilines are used; will tail if column too acidic, indicating Bases irreversible adsorption Fatty acid methyl esters Used to calculate number of theoretical plates Used to calculate number of theoretical plates Polynuclear aromatic hydrocarbons Hydrocarbons
6.3 QC/QA Aspects of GC-MS Instrument Operation
199
Table 6-3. Compounds used in GC-MS tuning and calibration. Compound
Molecular weight (amu)
Useful mass range (amu)
Major ions used in tuning (relative abundances)
Perfluorotributylamine (FC-43)
67 1
50 - 503
Perfluorotripentylamine (FC-70)
821
50 - 602
Perfluorotrihexylamine (FC-71)
97 1
50 - 102
Perfluorophenanthrene (FC-53 11)
624
55 - 624
69(100); 131 (48); 219(51); 264(13); 502(3) 69(100);, 181(14); 269(20); 314(3); 564(1); 602(0.4) 69(100); 169(25); 231(13); 364(5); 614(1); 702(0.4) 69(100); 132(47); 243(8); 455(18); 5 5 5 ( 5 ) ; 624(1)
The mass ranges given indicate the mass ranges which are suitable for tuning in EI mode 161. Major ions used in tuning and their relative abundances (in parentheses) are also given.
rity is verified and a check is done to make sure that all heated zones (injector, transfer line, etc.) are at their set temperatures. Then, a mass calibration compound is used to set mass peak widths and relative abundances according to accepted standards for the specific calibration compound used. Several calibration compounds are available for use in GC-MS applications (Table 6-3). The mass spectrum of the ideal calibration compound has many masses, encompassing a wide mass range, with which the mass axis of the MS may be calibrated. For example, perfluorotributylamine (PFTBA) is one of the most widely used calibration compounds. PFTBA has a well established spectrum; m/z 69 is its base peak, m/z 219 has an abundance of approximately 50% relative to the base peak, and m/z 502 has an abundance of approximately 2.5% relative to the base peak. Another important function of the calibration compound is to check mass resolution. Mass resolution can be checked by determining how well-resolved a 12C-peak is from its '3C-isotope peak; for example, mass resolution is often checked by comparing the degree of separation of the m/z 502 and m/z 503 mass peaks of the PFTBA calibration compound. If the MS can accurately reproduce the mass spectrum of PFTBA, it should also produce valid mass spectra for unknown compounds. A second check of the MS is sometimes performed by examination of the mass spectra of methyl stearate or decafluorotriphenylphosphine. Calibration of the MS is the first step in identifying components of complex samples; this ensures that the MS is producing spectra that are consistent with those found in standard data bases of reference mass spectra that are used to identify unknown sample components. Once it has been established that the MS can produce the correct mass spectrum, and that the mass axis is calibrated, an initial assessment of the MS sensitivity may be made by its response to the calibration compound, which should be consistent from day to day. This gives an initial indication that the MS is operating with sufficient sensitivity to meet the detection limits required for trace analysis methods. However, the overall GC-MS sensitivity and instrumental detection limits can only
200
6 Quality Control and Quality Assurance Aspects of GC-MS
Table 6-4. Documenting GC-MS performance. Problem
Principal causes
Solution
Insufficient vacuum 0 Roughing pumps < 0.5 mbar 0 High vacuum pumps >10-7-10-Smbar
Inadequate pump operation Air leak
Change pump oil or repair Pump If water: N, ratio > 2 : 1, air leak is indicated. Leak check with Ar
Tune not consistent with previous day’s
Loose connections on source parts /lenses Dirty source
Check for electrical continuity Clean source
Poor mass peak shapes and widths 0 Peak width at half height <0.5 amu 0 Asymmetrical peak
Dirty source Dirty quadrupoles or quadrupole prefilters
Clean source Clean quadrupoles or prefilters
Mass axis calibration unacceptable
Insufficient tuning compound introduced into source Electronic or quadrupole tuning problem
Add tuning compound to reservoir
Insufficient tuning compound introduced into source Dirty source or quadrupoles Mistuned
Add tuning compound to reservoir
Spectrum of tuning compound unacceptable
Repair electronic system
Clean source/quadrupoles Retune instrument
~~~~~~~~~~~~~~~
Each day of instrument operation, it should be verified that the above problems, in the order listed, are not occurring.
be determined by analysis of standards of the analytes at concentrations no greater than 1- 10 times the detection limits required for the specific applicaton. If the above performance checks (Table 6-4) indicate the GC-MS system is not operating within specifications, and standard corrective procedures are not effective, then the GC-MS operator may attempt to reproduce the system’s original installation specifications. If none of these activities are successful, then electronic problems may exist that must be addressed by qualified service personnel. For additional information about the theory and application of GC-MS, readers are referred to other sources [6,7]. Although it is beyond the scope of this chapter to discuss details of the calibration and quality control procedures for GC-MS operation using ionization modes other than EI, brief mention will be made of chemical ionization (CI). CI is a soft ionization technique, which typically does not produce the extensive fragmentation that is encountered with EI. Thus, CI is often used to determine molecular ions, which is important information for the identification of unknown compounds. However,
6.3 QC/QA Aspects of GC-MS Instrument Operation
201
routine methods using CI are difficult to implement because of the problems associated with reproducibly obtaining CI mass spectra. The fragmentation patterns observed for CI mass spectra are much more sensitive than corresponding EI mass spectra to operating variables such as source temperature and reagent gas pressure in the MS ion source. It is especially difficult to accurately reproduce CI gas pressure in different mass spectrometers because of differences in ion source designs and in how source pressures are measured. Therefore, it is crucial to check that the calibration compound yields the expected fragmentation pattern in CI mode. Each laboratory using CI should be encouraged to develop calibration procedures which work for the analysis at hand and which are adhered to stringently by the MS operators. The QC procedures followed must be sufficient to ensure that the GC-MS system is operating with specifications known to result in acceptable instrument sensitivity and which will result in the generation of reproducible mass spectra. Known analytes should produce expected responses as defined by past performance of the GC-MS. Specifically, it can be documented that the GC-MS is operating properly if wellresolved, symmetrical GC peaks with widths of 1 - 10 s are obtained for known analytes and that these analytes may be detected when present at concentrations just above their limit of detection.
6.3.4 Description of Various MS Instruments and Capabilities The specific type of GC-MS system used determines the limitations in the data that can be generated for the analysis of trace analytes in environmental samples. It is important to know how to calibrate the GC-MS system for a specific application and how to correctly interpret the data generated by different types of instrumental systems. There are many possible configurations of GC-MS systems, but here we will comment on the three types most commonly used for environmental analysis applications. Quadrupole mass spectrometers are especially well suited to perform qualitative analysis applications. They can be operated at the fast scan rates required to obtain full mass spectra from the narrow-width GC peaks expected from capillary columns (full mass range scanned in about one second). The most important limitation of the quadrupole MS is that it can only be operated at about nominal mass resolution (ie, mass 502 can be resolved from mass 503, but mass 502.0 could not be resolved from an ion at mass 502.1). In practical terms, this means that a quadrupole MS could not distinguish between the molecular ions of N2 and C2H, (both nominal mass 28). Quadrupole MS are also well-suited for quantitative analysis applications that require the use of selected ion monitoring, because of the ease of switching between the various masses selected. Although quadrupole mass spectrometers are not as selective as the other types discussed below, their selectivity can be dramatically improved by combining two sets of quadrupole analyzers in tandem in the same instrument. The resulting GC-MS-MS technique can be very effective for use in ultratrace quantitative analysis applications that require a high degree of selectivity. The use of this instrument for determination of the chlorinated dioxins has been re-
202
6 Quality Control and Quality Assurance Aspects of GC-MS
ported (81, and the complex calibration procedures required for this work have been described [9]. Double-focusing, high resolution mass spectrometers (HRMS) employ a combination of magnetic and electrostatic sectors to permit accurate mass determinations, provided that the proper calibration procedures are followed. The mass resolution (R)between two barely separated peaks of masses ( m )and (m+ A m ) is given by: R = m / A m . Modern HRMS instruments can readily achieve resolutions of 10000 to 15000 for environmental GC-MS analysis applications. By comparison, the resolution of a quadrupole MS for the separation of mass 500 from 501 is only 500. With a resolution of 10000, an ion of mass 500.00 could be distinguished from another ion of mass 500.05. Thus, a much greater selectivity for specific analytes can be achieved by using HRMS compared with single quadrupole MS. The disadvantages of HRMS instruments compared with quadrupole MS instruments are slower scan speeds, greater cost, and much greater complexity of operation (including calibration, troubleshooting and servicing, and basic instrument operation). A relatively new type of commercial MS is the Ion Trap (ITD). The ITD can be used for the same applications as the quadrupole MS: Although currently available ITD instruments provide only nominal mass resolution - much like the quadrupole MS - modified ion traps have been developed that have the capability for high resolution operation [lo]. Due to its simple design and low cost, use of the ITD will likely become much more common for environmental GC-MS applications; in fact, the ITD has been approved for use with several US EPA methods. The ITMS must be operated with a different set of cautions than either the quadrupole or HRMS systems. If too much analyte is allowed to enter the ITMS, space-charge effects could occur that may distort the mass spectrum obtained compared to the standard EI mass spectra found in common data bases. The use of automatic gain control in recent models of the ITMS has largely eliminated this problem, but users should still be cautious of concentration effects. A properly applied QC program with calibration procedures designed to detect concentration effects should ensure that results are comparable with those obtained from other types of GC-MS systems.
6.4 QC Considerations for Qualitative and Quantitative Analysis If the GC-MS instrumentation is functioning within acceptable limits as described in the previous section, then there are still many considerations to satisfy before environmental samples can be successfully analyzed. The approach taken for a specific application depends principally upon whether quantitative or qualitative analysis is to be performed. For quantitative target compound determinations, the technique of isotope dilution mass spectrometry is generally followed (when appropriate standards are available). For each of these types of GC-MS applications, there are special QC steps that must be considered to ensure that the objectives of the analysis are achieved.
6.4 QC Considerations for Qualitative and Quantitative Analysis
203
6.4.1 Quality Control for Qualitative GC-MS Analysis Although the majority of environmental analyses that employ GC-MS instrumentation are quantitative target compound determinations, the qualitative identification of organic compounds by their mass spectra and retention times is an equally important application. Quality control criteria are somewhat lacking for qualitative identifications - in part because it is difficult to assign numerical values to the accuracy of the match. In other words, a compound structure assigned to a mass spectrum is either correct or it is not. Yet, there are degrees of closeness of the match that can be assigned; for example, the actual identity of the unknown may be an isomer of the assigned structure, or it may be another member of the same compound class, or may simply have some common structural features. Some discussion on this topic has appeared in the literature [ l l - 141, and the general principles for compound identification by GC-MS have been described in recent books [6,7, 151. For compound identifications by using GC-MS data, computerized systems for matching the unknown mass spectrum with reference mass spectra in a database are generally used. Manual identifications by detailed interpretation of a mass spectrum may also be performed, but such work is very difficult and requires many years experience - and extensive knowledge of organic chemistry as well as mass spectrometry. The discussion here is limited to compound identification by computer search methods. Also, only positive-ion electron ionization (EI) will be considered, since most environmental GC-MS analysis work is performed by using positive EI mass spectra, and extensive computer libraries are only available for positive-ion EI mass spectra. For a positive identification, a standard of the compound chosen by the computer search must be analyzed by using the same conditions as the unknown, and both the mass spectrum and retention time (or index) of the standard must match the unknown. To perform such confirmations for every compound identified during analysis of an environmental sample is generally not practical. However, users may develop retention time (or index) libraries for common environmental contaminants - most of which will be also represented in commercial libraries of mass spectra; if a good match is found between an unknown and a reference compound in both the mass spectral and retention libraries, then the assumption of positive identification is generally correct. The separate analysis of standards for definitive confirmation only needs to be performed in special cases. Quality concerns with qualitative GC-MS analysis often occur when identifications from computer searches are reported without any qualifying information. To help users interpret the results of computerized search identifications, a classification system was developed by Taguchi et al. [16]. This system uses four categories for identification, depending on the analytical information available. These categories are: 1. Confirmed Identification - An identical match is observed between the mass spectrum and retention time of the unknown analyte, compared with those of an authentic standard of the identified compound, within experimental error of the measurement system.
204
6 Quality Control and Quality Assurance Aspects of GC-MS
2. Provisional Identification - A high level of confidence is placed upon this identification, based on the available GC-MS data. However, an authentic standard has not been analyzed by using the same conditions as the unknown. This category includes compounds identified solely by matching unknown and reference mass spectra (ie, no retention data).
3. Compound Class - Sufficient information does not exist for a coumpound identification. However, the mass spectral data are indicative of a specific molecular structure or compound class for the analyte. This category includes the case where no reference mass spectrum is found to closely match the unknown, and sufficient additional information does not exist to make a Provisional identification. Compound class identifications may be achieved by manual interpretation of the mass spectrum. 4. Unknown - Mass spectra cannot be identified without additional information. No reference mass spectra are found to closely match the unknown mass spectrum, and no specific structural features of a compound class are evident. Based on the above classification system, most environmental sample compound identifications - performed by using GC-MS data - probably fall into the Provisional category. Some general guidelines for ensuring adequate quality for Provisional compound identifications are the following: 1. All computer search systems calculate some type of match factor to indicate the degree of closeness between unknown and reference mass spectra. For provisionally-identified mass spectra, the match factor value must be at least 80% of the value that would be calculated for a perfect match. Users should remember that these match factors are not related to a statistical confidence level for accuracy of identification. 2. Reference spectra that satisfy the above criterion must be inspected visually by the analyst. All principal mass peaks (>20% abundance relative to the base peak) and characteristic isotope clusters must correspond to those of the unknown mass spectrum. The difference in relative abundances between corresponding peaks in the reference and unknown spectra should not be greater than about 20%. The base peaks in the reference and unknown spectra (that are not attributed to background or a co-eluting substance) must be the same. 3. Derivatizations or any other chemical treatment of the sample should be avoided, except for special applications where it is desired to identify certain specific compound classes. 4. Analysis of an appropriate blank sample is mandatory; many contaminants identified in samples may originate from field, sample container, reagent, or other laboratory contamination. 5. Approximate minimum mass spectrometer scan ranges for compound identification are 35 - 300 amu (volatile organics), or 40-510 amu (semi-volatile organics). Standard mass spectrometer calibration procedures must be followed. 6. The data base used for computerized searching must be at least of the size and quality of the most recent version of the mass spectral data base available from
6.4 QC Considerations for Qualitative and Quantitative Analysis
205
the National Institute for Standards and Technology (NIH/EPA/NIST Mass Spectral Data Base).
6.4.2 Quality Control for Quantitative GC-MS Analysis There are two GC-MS techniques commonly used for the quantitative determination of target compounds. In mass chromatography (MC), full mass spectra from continuous scanning of the GC effluent are stored on computer disk. After the analysis, the abundances of selected masses can be extracted from each stored mass spectrum, and plotted versus mass spectrum scan number of retention time. Compounds that have the selected masses in their mass spectra will show up as peaks in the mass chromatogram, whereas compounds that eluted from the GC and which have mass spectra that do not contain the selected masses will not appear in the mass chromatogram. The peak heights or areas of the peaks in the mass chromatogram can be used for quantitation of the selected analytes. In selected ion monitoring (SIM), the masses characteristic of the target compounds to be quantified are chosen before GC-MS analysis, and only those masses are monitored. The advantage of SIM over MC is that from 10 to 100 times lower detection levels can be achieved. The disadvantage of SIM is that only the chosen masses and retention times are available for confirmation of analyte identity, whereas in MC any mass within the set scan range of the mass spectrometer can be plotted. If a poor choice of masses for a SIM analysis is made, then the sample extract will have to be re-analyzed with different masses. For the lowest detection limits in well characterized sample types, SIM is generally employed. If determination of analytes near the method detection limit is not a concern, then a MC analysis offers greater flexibility and additional analytical information. Many of the quality control procedures used for both a SIM and MC determination are similar. The principal steps taken to ensure good precision and accuracy for quantitative determinations are similar to the steps generally taken for any environmental analysis method. These general steps as illustrated in Fig. 6-1 were discussed earlier. Specific actions that are important to GC-MS-based methods are discussed below: 1. Isotope Dilution - The technique of isotope dilution GC-MS is one of the most important and powerful quality control tools used in quantitative environmental GC-MS analysis. It is discussed in the next section.
2. Sampling and Sample Preparation - It is important to quality check the freedom from contamination of all sampling equipment, glassware, and reagents by using GC-MS detection at least at the sensitivity that will be employed for the final sample analysis. For example, to check the cleanliness of sampling equipment, solvent rinses fo all apparatus may be checked for contamination by GC analysis using a flame ionization detector. This would not be sufficient to determine whether low-level contamination existed that may appear during analysis of sample extracts by using the more sensitive and selective GC-MS SIM technique.
206
6 Quality Control and Quality Assurance Aspects of GC-MS
3. Choice of Masses to Monitor - For SIM quantitation, the masses characteristic of the target analytes are chosen before sample analysis. They should be the most abundant masses in the mass spectra of the analytes, that are not present in the mass spectra of known interferences. The masses chosen should also be those at highest mass in the mass spectrum, because lower-mass ions are more prone to interferences. It is generally not acceptable to choose only one characteristic mass for each analyte, because for almost any nominal mass chosen there are many compounds that contain that mass in their mass spectra. Therefore, two or three masses characteristic of each analyte to be determined are generally chosen. If high resolution mass spectrometry is employed, then two characteristic masses for each analyte are generally sufficient - one mass is used for quantitation and the second is used for confirmation of analyte identity. The ratio of the abundances or peak heights of the masses chosen should be the same as the ratio of their corresponding relative abundances in the mass spectrum of the analyte. Deviations from these ratios greater than 20- 30% of the ratios observed in the mass spectrum may indicate an interference is present, or that the GC-MS system needs to be recalibrated. However, there are several factors that affect the observed ratios, and large deviations are possible under some conditions. Generally, under routine analysis conditions when the GC-MS is functioning normally, one should not be alarmed unless the ratios change from those usually observed. If low resolution mass spectrometry is employed, then three or more masses are used. Choosing additional masses must be balanced against lower sensitivity. 4. Confirmation of Identity
- For a MC analysis, the full mass spectrum of the analyte may be available for confirmation of analyte identity. When combined with the known retention time or retention index of the analyte, the correctness of the identification is seldom questioned. The full mass spectrum may not be available because the analyte concentration is too low, or the analyte mass spectrum may be difficult to separate from the mass spectrum of co-eluting substances. It may be possible to separate the mass spectra of co-eluting substances by using reverse-search computer techniques [7]. For a SIM determination, confirmation of identity is from the retention time of the analyte, and from the correct ratios of the masses monitored. Generally, ultra-trace organic analysis methods that require the use of SIM analysis also include a selective sample cleanup designed to remove known interferences. If potential interferences exist that have similar retention times to the analytes and also have mass spectra with masses in common with some (or all) of the masses used for the SIM experiment, then additional masses can be added that are characteristic of the interferences. The criteria for analyte identification would then include the requirement that no response for the interference be observed at the same retention time as the analyte.
6.4 QC Considerations for Qualitative and Quantitative Analysis
207
6.4.3 Isotope Dilution GC-MS Analysis One of the principal advantages of using GC-MS for environmental trace analysis is that the technique of isotope dilution can be employed to monitor the efficiency of analytical procedures on individual samples. In the isotope dilution technique, a known amount of a stable isotope-enriched analog of the analyte is added to each sample at some stage of the analytical procedure, and the isotopically-labeled standard is measured in the final GC-MS analysis to determine its recovery through the entire or a portion of the analytical process. How this standard is used depends upon when it is added to the sample. For example, if added to the sample test portion before any extraction or other treatment steps, then the stable isotope analogue of the analyte may be used to determine analyte recovery through the entire analytical procedure. Ideally, the percent recovery of the isotopically labeled standard can be used to correct the calculated concentration of the analyte for losses during analysis. Stable isotope-enriched standards may also be used to determine losses through each stage of the analytical procedure; for example, extraction efficiency, recovery through sample cleanup, and losses at the GC injection port may be determined. For environmental analysis, the most common isotopically-enriched standards used are deuterated (ie, dlo-phenanthrene), 13C-labeled,or "C1-labeled (ie, '3C,2-2,3,7,8tetrachlorodibenzo-p-dioxin,or '3C12-2,3,7,8-TCDD).The assumption is that since the stable isotope-labeled standard is chemically and physically almost identical to its corresponding native analyte, then it should behave almost identically to the analyte through the various analytical procedures, and therefore provides a most effective monitor of the quality of the analysis. Since the standard is added to the surface of the test portion, it may not be incorporated into the sample matrix identically to the corresponding analyte, and therefore high recovery of the standard may be achieved in some cases where the analyte could not be extracted effectively. In such a situation, one would be incorrect to assume that high standard recovery indicated high analyte recovery. As a validated analytical method should include experiments to verify analyte extraction efficiency, this problem should occur rarely. GC-MS-based methods are ideal candidates for the isotope dilution technique, because although the istope-labeled analog mimics the analyte almost identically throughout the analysis, its molecular weight will be greater than the analyte, and both the standard and analyte can be easily distinguished from each other by use of a mass spectrometer detector. For example, l3Cl2-TCDDis used in the determination of 2,3,7,8-TCDD in environmental samples at parts-per-trillion or lower concentrations. The stable isotope-labeled TCDD has a molecular weight 12 amu greater than that of the native TCDD, and therefore is easily determined separately from the analyte - even if the analyte and isotopically-labeled standard co-elute from the GC column. Some I3C atoms are present in the native TCDD as well, but the natural abundance of I3C is so low that this does not present a problem. It is important in practice to check the purity of all stable isotope-labeled standards used, because it is possible that some of the native (unlabeled) compound is present. In such cases, addition of the standard to samples would result in a high bias for the GC-MS analysis results.
208
6 Quality Control and Quality Assurance Aspects of GC-MS
6.5 QC/QA Considerations for Using GC-MS in Contracted Work In the previous sections, we have emphasized the bench-level actions that must be taken to ensure analytical results of acceptable quality are obtained when employing GC-MS-based methods for the analysis of environmental samples. In many environmental monitoring studies, analytical work is contracted to private laboratories. The organizations who pay for this work may not possess experts who can interpret the results supplied by the contract laboratories with respect to data quality, and therefore may not be able to determine whether the data are of sufficient quality to meet the objectives of the study. This problem is especially difficult in the environmental analytical field, because there are very few real-matrix environmental reference materials available that can be used to check method performance. Therefore, contracted work should be designed to incorporate data quality checks. Most of these quality checks will increase the cost of the analytical work, so how many of them are used will depend upon the objectives of the work and the end-use of the data generated. Quality elements that can be incorporated into environmental analytical work that employs GC-MS-based methods include the following: 1. Use of Stable Isotope Standards - Ensure that the contractor uses methods based on the isotope-dilution technique whenever possible. The contractor must check the purity of all standards used, and must be able to trace the sources and history of use of all standards.
2. Dedicated Equipment for Ultra-Trace Work - For some work carried out at trace concentrations (ie, parts-per-trillion determination of 2,3,7,8TCDD), it is very difficult to keep the analytical system clean enough to prevent cross-contamination. In such cases, it may be necessary to purchase dedicated glassware, GC syringes, columns, etc., to be used only for the ultra-trace work. For example, glassware used for the analysis of hazardous waste should never be used for the determination of trace contaminants in drinking water. 3. Detailed Report - Results received from contracted work should never consist of simply a table of analyte concentrations. A description (or reference to) all methodology used, with emphasis on quality control protocols followed and a discusssion of precision, accuracy, and sources of error, should be supplied by the contract laboratory. 4. Blanks Analyzed - Every competent laboratory will analyze method blanks with
each batch of samples. Depending on the nature of the specific study, other types of blanks may be requested, such as travelling/storage blanks, glasswareheagent blanks, spiked blanks (especially when samples are stored for extended periods), etc. 5. Blind Replicates - A competent laboratory should be able to reproduce its ana-
lytical results for standard sample types. If known replicates are submitted for analysis, the comparison may show an artificially high degree of reproducibility,
6.5 QC/QA Considerations for Using GC-MS in Confmcted Work
209
because the contract laboratory may treat the replicates with special care. Also, if aware that two samples are replicates, the laboratory may repeat the analyses if results do not correspond. Unfortunately, none of the other analyses will likely be repeated. If a sample is split and submitted as two different samples with two unique sample codes, the analytical laboratory will not be able to determine which samples are replicated (provided the samples submitted are all similar in appearance). If sample storage and analyte stability are not problems, then a large amount of one or a few samples can be homogenized and periodically submitted for analysis during the course of a large study to provide a periodic monitor on laboratory performance. For some applications, certified real-matrix reference materials may be submitted as samples; however, in the environmental analysis field very few such reference materials are available, and the contract laboratory is probably very familiar with them. 6 . Archived Samples - The above actions may provide a good indication of the precision and reproducibility of a contract laboratory, but accuracy is very difficult to determine. This problem is made especially difficult because so few real-matrix environmental reference materials are available. Spiked standards may be supplied to the laboratory - as unknowns - for analysis, but standards are generally much easier to analyze than real samples, and thus laboratory performance may be rated artificially high. One way around this problem is to use previously analyzed samples - ones that have been analyzed a number of times and by a number of different laboratories - and for which consensus concentration values are known. These ‘pseudo-reference’ materials can be valuable tools in determining contract laboratory performance. Of course, sufficient care must be taken to ensure the analyte concentration remains constant during storage. 7. Use of Multiple Contractors - For major projects that require many analyses, it can be beneficial to employ two contract laboratories. One would provide the bulk of the analytical work, while a second could analyze a small fraction of the samples as duplicates (5 - 10%). If good agreement between the two laboratories is not observed, then serious analytical problems may be indicated. For the determination of analytes not near the detection limits for the standard GC-MS-based analytical method employed, the average between-laboratory agreement should probably be 25% or better.
8. Detection Limit Validation - There seems to be as many ways of determining method detection limits as there are laboratories. The real low-level capability of a laboratory for a specific method and sample matrix must be determined by the analysis of real samples, but this is difficult when such ideal samples or appropriate reference materials do not exist. Laboratories may be submitted blanks spiked with analytes near the detection limit (within 5 - 10 times the stated detection limit), or archived samples known to have analyte concentrations near the detection limit. 9. Use of Alternative GC-MS Technique - For GC-MS-based trace environmental analyses, it is generally not possible to compare results by using alternative analysis techniques. However, several mass spectrometric techniques are available for
210
6 Quality Control and Quality Assurance Aspects of GC-MS
this purpose. If the principal technique employed is low resolution quadrupole mass spectrometry, then some samples should be analyzed by using high resolution mass spectrometry (GC-HRMS) or tandem mass spectrometry (GCMS-MS) HRMS and MS-MS techniques seem to be complementary for this purpose [17, 181. Alternate ionization techniques can also be used in some cases (positive and negative-ion chemical ionization), but this requires special expertise not available to many contract laboratories. 10. Expert Assistance - In addition to employing a contrct laboratory to analyze
environmental samples, it may be prudent to contract the services of an analytical expert - one not associated with the analytical laboratory - to oversee the work and to examine the results submitted, specifically with regard to the quality control elements of the work, and the precision and accuracy of the data. If inhouse trace analysis expertise does not exist, and the results of the work are to be used for sensitive issues such as regulatory compliance or high-profile public issues, then such assistance is highly recommended. Of course, the best time to employ such an expert is at the design stage of the analytical work, so the appropriate sampling, analysis, and QC protocols can be chosen. 11. Archival Data/Sample Storage - Contracted laboratories may not keep all of the unused sample portions and sample extracts, or raw data stored on computer, for very long after the final analytical data have been reported. It is important that some storage time for samples and data be negotiated before work has begun, so a reasonable time remains to re-check any sample data that are contentious or are of suspected poor quality. Since this may require additional computer storage media and physical storage space, including refrigerators, extended sample and data storage may add to the cost of the work. However, the cost is much greater if the whole sampling program must be repeated.
6.6 Summary and Conclusions Quality control aspects of the use of various GC-MS techniques for environmental analysis applications have been emphasized. The types of GC-MS systems used and the principal types of data generated were also described, because understanding the basic principles is essential to design effective QC measures that ensure data of defined quality can be generated. Coverage was not meant to be comprehensive, but the most important factors in the design of a QC program were discussed. It is especially important that those associated with the use and interpretation of trace environmental data as generated by GC-MS-based methods are familiar with these principles, because the QC measures above are essential to obtaining data of defined quality.
References
21 1
References [l] Supelco, Inc., HPLC Troubleshooting Guide, Bellefonte, PA, 1991. [2] Message, G. M., Practical Aspects of Gas Chromatography/Mass Spectrometry, New York:
Wiley Interscience, 1984. [3] Schomburg, G., Gas Chromatography: A Practical Course, New York: VCH Publishers, 1990. [4] Poole, C. F., Poole, S. K., Chromatography Today, New York: Elsevier Science Publishers, 1991. [5] Ravindranath, B., Principles and Practice of Chromatography, New York: John Wiley and Sons, 1989. [6] Watson, J.T., Introduction to Mass Spectrometry, 2nd edn, New York: Raven Press, 1985. [7] Karasek, F. W., Clement, R. E., Basic Gas Chromatography - Mass Spectrometry: Principles and Echniques, Amsterdam: Elsevier Publishers, 1 988. [8] Reiner, E. J., Schellcnberg, D.H., Taguchi, V.Y., Environ Sci Technol 1991, 25, 110-117. (91 Schcllenberg, D. H., Bobbie, B.A., Reiner, E. J., Taguchi, V. Y., Rapid Commun Mass Spectrom 1987, I , 111 - 113. [lo] Londry, F.A., Wells, G. J., March, R.E., Rapid Commun Mass Spectrom 1993, 7, 43-45. [ l l ] Christrnan, R.F., Environ Sci Technol 1982, 16, 143A. [I21 Christman, R.F., Environ Sci Technol 1982, 16, 594A. [13] Letters to the Editor, Environ Sci Echnol 1982, 16, 595A. [14] Christman, R.F., Environ Sci Technol 1984, 18, 203A. [15] McLafferty, F. W., Interpretation of Mass Spectra, 3rd edn, Mill Valley, CA: University Science Books, 1980. [16] Taguchi, V. Y., Wang, D. T., Jenkins, S. W. D., Kleins, R. J., Clement, R. E., Canadian J AppZ Spectrosc 1992, 37, 145- 148. [17] McCurvin, D.M.A, Clement, R.E., Taguchi, V. Y., Rciner, E. J., Schellenberg, D. H., Bobbie, B.A., Chemosphere 1989, 19, 205 -212. [18] Clement, R. E., Tosine, H. M., Mass Spectrom Rev 1988, 7, 593 - 636.
This Page Intentionally Left Blank
7 Application of Capillary Electrophoresis for Environmental Analysis Saul M. Parry and Colin F. Simpson
7.1 Introduction Capillary electrophoresis (CE) is the generic name for a family of electrophoretic separation techniques undertaken in narrow bore capillaries. This review will concentrate on the use of isotachophoresis (ITP), capillary zone electrophoresis (CZE) and micellar electrokinetic chromatography (MEKC) and their application to environmental analysis. The fundamental aspects of the application of CE for environmental analysis are covered by using real examples to help explain the theory. Included are details on the separation parameters employed; sample introduction; detection methods and hyphenated techniques. As far as possible, examples include sufficient detail to permit the separation to be repeated.
7.1.1 Equipment Requirements The basic apparatus for CZE is represented schematically in Fig.7-I A, showing the separation capillary (formed of silica or a polymer, eg. PTFE) filled with buffer, located between two reservoirs. A detector is placed to monitor the separated analytes in-capillary at a defined point along its total length. High-voltage is applied at the inlet reservoir, and the outlet reservoir grounded. Samples are introduced at the capillary inlet. The apparatus is termed an ‘open system’. In ITP, the capillary is usually shielded from the electrolyte reservoir(s) using a cellulose acetate membrane to prevent hydrodynamic flow of electrolyte (Fig. 7-1 B). The apparatus is termed a ‘closed system’.
7.2 Capillary Zone Electrophoresis In capillary zone electrophoresis (CZE), separations are generally performed in a continuous background electrolyte (BGE), with a constant DC voltage applied. Samples are introduced at the inlet side of the capillary and separated due to differences
214
7 Application of Capillary Electrophoresis for Environmental Analysis Data acquistion
I
Terminator Capillary inlet
Capillary outlet
I , Buffer-reservoir Electrolyte buffer
I High voltage
y
r;;b$l
Septum
-
Drain
I 1
J
Buffer-reservoir
Electrolyte buffer
Fig.7-1. (A) The basic apparatus required for CZE. Note that the electrolyte reservoirs are positioned at equal height and the liquid levels are equal to minimise hydrodynamic flow. (B) The basic apparatus for ITP. Note the use of a membrane to minimise hydrodynamic flow. Platinum electrodes are used in both apparatus for application of the applied potential. (Reproduced with permission [l, 21).
I
w1 j
1
-
Thermostat
Membrane Electrode
in the mobilities of the analyte ions under the influence of the applied high-potential. This applied potential also can produce a flow of electrolyte, known as electroosmosis (EOF). The popularity of CZE arose from the early work by Jorgenson and Lukacs [3,4], where the use of 75 pm ID capillary and up to 30000 V enabled rapid, high efficiency separations ( > 400000 plates) to be achieved on a tryptic digest of heat denatured chicken ovalbumin. Figure 7-2 shows the rapid, high-resolution separation of a sample containing a number of anions that may be obtained using CZE, demonstrating the potential application of the technique to environmental samples.
7.2.1 Fundamentals of CZE Theory When an electrical potential ( U )is applied to a capillary (filled with BGE) of length L, the electric field strength ( E ) is given by E = U/L V cm-'. Injected ions will migrate at a velocity vi in the applied field according to:
where pi is the mobility of the ion i. It is important to recognise that mobility is a signed quantity, positive for anions and negative for cations. The differing mobilities
7.2 CaDillarv Zone Electroohoresis
2, nitrite (4); 3, nitrate (4). Electrolyte = 25 mM chloride, 0.5 mM OFM Anion-BT (chloride form) (Waters Chromatography), pH 8: HT = 20 k V 214 nm detection: 40 cm x 0.075 mm ID capillary. INJ = 5 kV, 45 s. Sample spiked with 40 ~ L Na M octanesulphonate prior to analysis. (Reproduced with
1
z
,
21 5
..
of sample components i result in their different migration velocities thus, the time of migration of component i is given by:
In the case of weak electrolytes it is often the effective mobility of the ion which is most useful. The ionic mobility relates to the mobility of the completely dissociated analyte, whilst the effective mobility (pFfl)is related to the degree of ionisation, a.
The net mobility (pnet)can be calculated from the electropherogram using the following equation:
where pepis the electrophoretic mobility and peOfis the EOF ‘mobility’ (see below). L is the total capillary length, 1 the separation distance, and t, the migration time of the solute for wholly dissociated analytes, but for partially dissociated analytes pep should be replaced by illeff. The EOF may be conveniently calculated by measuring the time it takes for an uncharged species to be transported from the point of application to the detector (to), using the following calculation: &,f
= l.L/V*to
(5)
The electrophoretic mobility may be calculated from the following equation:
An interesting effect can be observed when analysing fully ionised species in the presence of an EOF, in that the net mobility will parallel the EOF over a wide pH range.
216
7 Application of Capillary Electrophoresis for Environmental Analysis Paraquat
(B)
(A)
liquat
Paraquat
0.0016
0.0005
E
liquat 2-Aminopyridine (1.5.
U L n N
I
I
+ rn 01
Y
c
f 0 VI 1
n
a
I-
c -
1
Fig. 7-3.Electropherograms of two herbicides analysed by CZE. BGE = 0.10 M phosphate, pH 3.5 in (A); pH7.0 in (B); 80cmxO.05Om ID capillary (50cm to detector). UV detection (254nm). HT = 15 kV. INJ = 15 kV, 5 s. (Reproduced with permission [6]).
This phenomenon is shown in Fig. 7-3 for the separation of paraquat and diquat standards. Adsorption of analytes to the capillary wall is a real problem in some separations, leading to zone broadening and irreproducible migration times. High ionic strength buffers may be used to solve this problem, as can the inclusion of a substance in the BGE with preferential adsorption. The selectivity in CZE may be changed a number of ways: changing the BGE pH (including dynamic shifts in pH 17- 151); the addition of organic solvents; complexation with a selected additive; the type of detector used (universal or highly specific); sample introduction: many of these aspects are covered in the section on indirect UV detection. Other parameters that may be altered to effect a separation include: the physico-chemical nature of the BGE; the voltage applied during the separation; and the temperature of the separation. The effect of applying a potential across the capillary is to generate Joule heating. Changing the temperature of the capillary and contents has a crucial effect in all electrophoretic separations because the mobility of ions increases 2% for every 1 "C rise in temperature [15]. Moreover, the nature of the BGE may change if excessive Joule heating occurs (eg, viscosity and pH changes). Thus, it may be essential to thermostat the capillary in some fashion to ensure reproducible results. The use of efficient thermostatting also enables the possibility of a temperature programmed separation 1161. If thermostatting is not available, the use of constant current rather than constant voltage for the separation is to be recommended. The use of constant current can minimise Joule heating effects, as it is the voltage that swings rather than the current.
21 7
7.2 Capillary Zone Electrophoresis
7.2.2 Electroosmotic Flow (EOF) When a solid is in contact with an ionic solution, ions are present on the surface and, to preserve electrical neutrality, ions must also be present in the solution. Thus an electrical double layer is formed. This double layer may be viewed as being composed of a relatively immobile layer of negative ions on the surface of the solid (silica in CZE)and a diffuse layer of positive counter-ions extending into the liquid (Fig. 7-4). Under the influence of an applied electric field, a portion of the diffuse layer moves because of the electric forces acting on the ionic charge present. A shear plane is set up some distance from the solid surface (although this distance is very small, a few nanometers). Since the ions are solvated, their migration causes a net movement of the solvated ions and hence a flow of liquid because of the associated nature of aqueous systems, towards the ground or negative electrode. This flow is called electroendosmosis (electroosmosis (EOF))and the magnitude and direction of the flow is a function of the nature of the solid and the ionic solution in contact with it, both of which can moderate the flow velocity and its direction. It is important to recognise that the 'mobility' of EOF is also a signed quantity and when calculating the magnitude of the net mobility of a species the sign of these two quantities, pepand peOfmust be taken into account. A constant flow rate is reached when the force exterted on the ions (and hence the bulk of the liquid) is exactly balanced by the frictional drag exerted by the viscosity of the liquid. The potential at the shear plane is called the zeta potential (0.Hence a flow of liquid occurs towards the appropriate electrode depending on the charge induced on the walls which is a function of the nature of the carrier electrolyte, the
8
a M
"v Q
Surface with fixed charges
Stern layer of
adsorbed ions
w
Ions move in bulk liquid
Q
Diffuse layer of mobile ions
Fig. 7-4. (A) The electrical double layer. (B) Schematic representation of the EOF in CZE. (Reproduced with permission 117 - 191).
2 18
7 Application of Capillary Electrophoresis f o r Environmental Analysis
nature of the surface, additives present and importantly, pH. The EOF possesses a virtually flat plug-flow profile in sharp contrast to the parabolic flow profile of LC, and partially accounts for the very high efficiencies achieved in CE.
7.3 Additive Based CE Separations 7.3.1 Micellar Electrokinetic Chromatography Micellar electrokinetic chromatography (MEKC or MECC) and the use of complexation CE are both methods to change the selectivity in CE. The theory of MEKC will not be covered here, thus readers should refer to the following reviews of MEKC [ZO-231, covering a wide range of applications. Terabe el al. [24] first demonstrated the use of surfactant micelles for the separation of neutral analytes in CE. The surfactant is added to the BGE above its critical micelle concentration (CMC) value, forming micelles. The principle involves the distribution between the solvent and the core of the micelle of the analyte. Differing hydrophobicities determine this ratio, allowing the separation of otherwise inseparable analytes due to the different attractions to the core. The order of migration is those very polar analytes with no micelle interaction elute first at time to, followed by those with increasing interaction (tR),to finally those analytes completely held within the micelle at time tMc. The period between the first and last peak as described is known as the retention window. Figure 7-5 shows the MEKC separation of eleven priority phenols.
4 2 1
0
12
24 Min
36
48
Fig. 7-5. MEKC separation of eleven priority phenols. BGE = 0.05 M SDS in 0.005 M phosphate /0.01 M borate: pH 6.6: 1 mx0.180 mm ID capillary. HT = 10 kV @ 130 PA: 254 nm UV detection: 1000 ppm of each phenol in mixture. Peaks: (1) methanol; (2) 4-nitrophenol; (3) 2,4-dichlorophenol; (4) 2-nitrophenol; ( 5 ) 2-chlorophenol; (6) 2,4,6-trichIorophenol; (7) pentachlorophenol; (8) 2-methyl-4,6-dinitrophenol; (9) 2,4-dinitrophenol; (1 1 ) 4-chloro-3methylphenol; (12) 2,4-dimethylphenol; (1 3) sudan 111. (Reproduced with permission 1251).
7.3 Additive Based CE Separations
219
0
EM.E
c
E" z40 -
.-c 4-
f20o(
01
I
I
I
I
I
I
8
9
10
11
12
PH
I
0
I
I
100 200 300 Borate concentration
I
400
1 1
4
3
II)
c
a
R a
2
21 2 &-
Fig. 7-6. (A) Effect of BGE pH on the retention window in MEKC. BGE = 50 miv octyl glucoside, 400 miv borate: 80 cm x 0.050 mm ID capillary. HT = 15 kV. INJ = hydrodynamic (5 s at 20 cm). UV detection (210 nm). (B) Effect of borate concentration on the retention window. Experimental conditions as in (A), except BGE = pH 10. ( C ) BGE = pH 9. Other experi-
Cai and Rassi [26] used novel micelles for the separation of four triazine herbicides and some polyaromatic hydrocarbons (PAHs). The micelles had adjustable surface charges, achieved by the complexation between an octylglucoside surfactant and alkaline borate. The retention window could be easily fine tuned over a wide range by varying the pH and/or borate concentration of the BGE. Figure 7-6 shows the effect of a changing BGE pH upon the retention window, and the subsequent separation of four triazines. Detection limits were in the low p~ range (3 - 9 pm) using hydro-dynamic injection. The inclusion of complexing substances in the BGE has been widely used to increase the selectivity in both CZE and ITP. Imasaka et al. [27] used cyclodextrin modified MEKC for the determination of anthracene derivatives. The use of a helium - cadmium laser for fluorimetric detection enabled the determination of 9,lO-dimethylanthracene down to 7 x low9M.
220
7 Application of Capillary Electrophoresis for Environmental Analysis
7.4 Isotachophoresis (ITP) Isotachophoresis (ITP) has been widely used for the analysis of environmental samples, and is ideally suited for this purpose owing to the concentration effect on dilute species. Excellent reviews of ITP can be found in the following references [28-411, and the books by Everaerts et al. [42] and Bocek et al. [43]. Isotachophoretic separations use discontinuous buffer systems; one buffer termed the leading electrolyte (LE), possesses a higher mobility than all analyte ions of interest, followed by another buffer, termed the terminating electrolyte (TE), whose mobility is only slightly lower than the lowest mobility analyte ion. The sample is injected at the interface between these two electrolytes, and on application of the field (constant DC current 0- 500 PA) the sample and the electrolytes start to migrate and disengage into discrete contiguous zones in time. In this state the individual zones will eventually all migrate through the capillary at constant velocity in a condition resembling a stack of coins. The moment the separation reaches this ‘steady-state’, zone lengths do not change (applying a closed system without an EOF). The concentration of all the components in their zones is approximately equal to the LE concentration, according to the Kohlrausch regulating function (see below) at the steady state: ci(x)z. w ( x ) = C 2= constant (x) (7)
-
i
mi
where the summation is for all analytes i, present at point x in the capillary. For a given x, w ( x ) has a constant value pre-determined by the initial conditions prior to application of the potential. Therefore, all zones are at essentially the same concentration. Two isotachopherograms are shown in Fig. 7-7, representing the typical output from the main detectors used in ITP. A wide variety of optimisation strategies have been developed for ITP, based upon simple, but accurate computer simulations of the separation parameters [46- 551. The ease in computer simulations of ITP separation should eventually minimise optimisation time, however at present, no PC based optimisation procedures are commercially available.
7.4.1 ITP in the Presence of an Electroosmotic Flow At the present time interest in ITP has been centred on the applicability of ‘open systems’, for quantitative analysis [56 - 601. This has resulted from the predominance of open-systems on the market. Beckers et al. [61] have descirbed four possible modes for ITP in the presence of an EOF (Fig. 7-8): (a) the cationic and anionic; and (b) the reverse cationic and reverse anionic. In actuality, further modes may be found to be feasible: eg. coating of the capillary wall to reverse the EOF can enable anionic separations to occur in parallel with the EOF.
1.4 Isofachophoresis (ITP)
221
7 i
JA Fig. 7-7. Typical output in ITP. (A) Photometric determination of Cr(”). (1) tap water sample; (2) spiked tap water in (1). LE = l O m ~C1-, pH 3.5 (J-alanine) with 0.1% w/v hydroxyethylcellulose (HEC). TE = acetate. M); (2) methanesulphonate M); Peaks: (I) naphthalene-1,3,6-trisulphonicacid (NTS) (3) picrate (2xlO-’~). 30pL sample loading. (2) As (1) except: sample spiked with Cd”) at a 2x M. Picrate = M. LOD = 4-5 ppb. (B) Conductimetric detection of some heavy metals using ITP. LE = 0.02 M Na or K, and hydroxyisobutyric acid (HIBA) to pH 5 TE = H + c. 0.005 M (acetate). LOD = < 5 ppb for Cu. Low ppb for all analytes using cation exchanger (Chelex 100) with high high affinity for heavy metals. (Reproduced with permission [44, 451).
-~I
+ T
S
L
II
AM - T
S
L
ii +
CM
(C)
RAM-
(Dl RCM
ITP
EOF
ITP
EOF
I
+
c - c
,
I I
I
Fig. 7-8. Schematic representation of the four modes in ITP with an EOF. (A) CM = cationic mode; (B) AM = anionic; (C) RAM = the reversed anionic; (D) RCM = t h e reversed cationic mode. For further explanation, see text. (Reproduced with permission
[611).
222
7 Application of Capillary Electrophoresis for Environmental Analysis
Histidine
-0.01
'
0
'
'
'
.
'
5
'
'
.
t (min)
'
'
10
'
'
'
'
'
15
Fig. 7-9. The velocity of the EOF in the cationic mode. The time interval between peaks decreases as the capillary fills with TE, ie, the EOF increases. LE = 0.01 M KOH, pH 5 adjusted with acetic acid. TE = acetic acid, pH 3.5. The histidine peak represents the transition between the LE (L) and TE (T). UV detection (254 nm). HT = 4 PA. (Reproduced with permission [61]).
The velocity of the EOF is of real importance in ITP, as the length of detected zones will be a function of the EOF. During an ITP run the capillary gradually becomes filled with a different electrolyte, thus the velocity of the EOF may continuously change [61]. A change in EOF will result in non-quantitative results as zone lengths may be continuously changing, therefore any change must be minimised. The relative change in EOF may be monitored by alternately injecting an EOF marker dissolved in LE and TE, (Fig. 7-9) or the voltage drop over the capillary at constant applied current. A constant EOF results in a linear relationship between voltage drop and time. From Fig. 7-9 it can be seen that the EOF varies quite strongly in open ITP systems. Moreover, the degree of the change is also dependent upon the TE used for the analysis. Ackermans et al. [62] showed that the use of a TE with a relatively high mobility at a pH where a high ionic concentration is present (pH
p K a , a strong variation in EOF arises due to the high voltage gradient across the capillary. Thus, a TE with a very low effective mobility and/or low ionic strength should be avoided to minimise EOF fluctuations [62]. The sample solution may also effect quantitative analysis in ITP. Increased sample sizes result in EOF increase, and this effect is particularly noticeable for samples of lower ionic strength than the LE. It is clear that using ITP in the open mode presents limitations on the range of electrolytes that may be usefully employed. Thus, one must control or eliminate the EOF to work quantitatively in ITP. For many years this factor has been recognised, and effective control of the EOF can be quite simply achieved by the addition of surface-active compounds in the electrolyte and sample solutions. Methylhydroxycellulose (MHEC) has been widely used for this purpose, and a variety of other coatings are possible (eg, surfactants). The effect of addition of MHEC on the voltage drop during an ITP analysis is clearly shown in Fig. 7-1 1. The availability of commercially coated capillaries is now also increasing, and some workers have used proprietory GC coatings (eg, DB-17) to eliminate the EOF.
7.4 Isotachophoresis (ITP)
(4
v A
223
T
L
T'
"i
B
L
C
L
i
T
I
13
26 39 Time lminl
52
65
Y
10
20
30
Time lmin)
a
U
m
; n
a I
13
14
15 16 Time (minl
17
18
Fig.7-10. (A) Dependence of the EOF upon the TE used for ITP. Isotachopherograms for a LE = sodium nicotinate, pH 5.4. TE = A , 0.01 M histidine, pH 5 (histidine); B, 0.01 M histidine, pH 6.7 (MES); C, 0.01 M y-aminobutyric acid (GABA), pH 3.5 (formate) and 0,0.01 M GABA, pH 5 (nicotinate). It can be clearly seen that differing TEs can cause strong variations in the EOF velocity, resulting in different times of analysis. (B) Experimental conditions as (A), except to all solutions 0.05% of MHEC was added. The suppression of the EOF is clearly demonstrated. Reproduced with permission [62]).
The use of surface-active substances diminishes the change in the EOF; however for quantitative analyses ideally an internal standard may also be required [62] although the choice of internal standard may be difficult when analysing complex samples with many unknowns. A possible method to overcome the changes in the EOF during an analysis might be to directly control the EOF externally [63 - 651. The use of an externally applied electric field to the outside of the capillary allows one to control the zeta potential at the capillary inner-surface to a defined value, and also monitor and change that potential during a run. The general applicability of open ITP systems is still under investigation at present, but is likely to develop rapidly as the number of practical applications develop because the fundamental theory of ITP is essentially the same in both open and closed systems [61]. The use of open ITP also facilitates hyphenated techniques such as
224
>
7 Application of Capillary Electrophoresis for Environmental Analysis
20
5 15 01
m
$10
> 5
13
26 39 Time (mint
52
65
C D A
B
0
13
26 39 Time Imin)
52
65
Fig. 7-11.Relationship between the voltage drop and time of analysis. (A) For the electrolyte system in Fig. 7-10A. (B) For the electrolyte system in Fig. 7-10B. The addition of MHEC leads to a linear relationship between voltage drop and time. (Reproduced with permission 1621).
ITP-MS, and is expected to establish itself alongside the currently more popular electrophoretic technique, CZE.
7.5 Bi-Directional ITP Hirokawa et al. [66] have recently demonstrated a bi-directional isotachophoretic system, allowing both cations and anions to be separated simultaneously. The choice of electrolytes is actually very simple, in that the LE for anions must also be the TE for cations, and vice versa, ie, the pH buffering counterions from the leading anion also function as the terminating cation and the buffering counterions from the leading cation function as the terminating anion. It is the difference between the pH of the anolyte and catholyte, ie, solutions surrounding the anode and cathode respectively, that is restrictive in bi-directional ITP [66].
7.5 Bi-Directional ITP
e Leading anion
(+I
pH buffer + II Terminating cation +
Leading cation + e p H b ffer + Terminating anion
225
(-1
17.
Example 10mM CI20mM /3-alanine (pH =3.6)
t
(+I
(+)
+ CI-
/3-ala++ pH = 3.60
+
Acetate/3-ala+ pH = 4.38
10mM K+ + Acetic acid (pH = 4.8)
(-1
t 20mM
/3-ata+Acetate pH -3.83
K+ + +Acetate-
pH = 4.80
(-)
The system configuration for bidirectional ITP is shown in Fig. 7-12, with an example of an electrolyte system included. The composition of the electrolytes is selected to ensure good buffering ability and should fulfil the following rules [66]:
and
where PKQA and PKQC are the pKa values of the anolyte and catholyte buffering ions, and pH,A and pHLc are the pH of the leading (pH,) anolyte and catholyte, respectively. The pKa values of the buffer electrolyte and the pH, of the bi-directional system should also satisfy the following conditions:
and
When Eqs. (8) and (9) satisfy PHLA =pKQAand pHLc =pKQc, Eqs. (10) and (11) can be rewritten: 0.5 I PHE -pHL,
I1.5
(12)
Equation (12) expresses the desirable range of the pH difference between the catholyte and anolyte: ie, the pH of the catholyte should be slightly higher than that
226
7 Application of Capillary Electrophoresisfor Environmental Analysis
t
c
C
5 1min
-
.-c0 E
t ,
Ethanolamine
c
0
a
Time -+
t
Time
Fig. 7-13.The potential gradient record of an equimolar mixture of three cations and three anions. Experimental conditions as Fig. 7-12, except: HT = 100 PA. (Reproduced with permission (661).
of the anolyte [66]. The potential gradient record of a bi-directional ITP separation is shown in Fig. 7-13.
7.6 Sample Introduction Methods for CE An ideal injection loads a representative aliquot, reproducibly, in such a manner that the inherent efficiency of electrophoretic separations is not adversely affected. Injection of sample has been accomplished in many ways, some of which are specific to the particular electrophoretic technique. The first generation of ITP apparatus used syringe injection directly into the separation channel. This was possible because sample loading in ITP is of the order of microliters. Attempting the same procedure in CZE would be unlikely to yield satisfactory results, primarily because sample loading is of the order of nanoliters! Thus, the method of sample introduction is critical in CZE and will often determine the efficiency and sensitivity obtainable [67]. It has been shown that: (1) the maximum number of theoretical plates obtainable is proportional to the square of the ratio of the column length ( L ) to initial zone length ( E ) , ie, the applied sample volume must be minimised for maximum efficiency
and: (2) that the injection of a high concentration sample will result in perturbation of the local potential gradient and promote Joule heating. The resulting mismatch between sample and BGE conductivity leads to dispersion and possible thermal degradation of labile species.
7.7 Novel Sample Introduction Methods
227
7.6.1 Hydro-Dynamic and Electrokinetic Sample Introduction The most common forms of sample introduction are the hydro-dynamic and electrokinetic methods, and for maximum reproducibility both should be computer controlled. Both methods are included in virtually all commercial apparatus. It has been noted that simply by placing the capillary into a sample introduces a variable quantity of sample, largely due to bufferhample density differences [68]. This appears to be a ubiquitous problem. Hydro-dynamic injection involves placing the capillary inlet into the sample solution and raising them relative to the capillary outlet. Sample enters the capillary via gravity flow. Hydro-static injections utilise either a pressure applied to the sample vial or by a vacuum induced at the capillary outlet. For simplicity only the term hydro-dynamic will be used. Both methods enable precise injections, however care must be taken that pressurehacuum lines, etc., are well maintained to ensure long term reproducibility with these methods. Electrokinetic injections are the most simple to accomplish and control in CZE, whereby the inlet reservoir is replaced with a sample reservoir and high voltage applied. A sample bias may be introduced resulting from the combination of both electrophoretic mobility and electroendosmotic flow. Thus, high mobility ions will be applied in larger quantities than lower mobility ions. Ions with a high mobility opposite to the EOF may not even enter the capillary. The effect of differential conductivity between sample and buffer may worsen this bias, although this can be used to increase the selectivity of the separation. The widespread acceptance of these injection methods results from their simplicity and ease of construction. For precise quantitative analysis they represent an inadequate solution. an ideal injection in CZE would place a well defined ‘plug’ of sample into the capillary, without mixing with the BGE. The exact volume should be known as small errors play a larger role when handling nanoliters of solution. A number of elegant devices have been demonstrated [69-751, however most have not been constructed for use with capillaries of less than 100 pm ID.
7.7 Novel Sample Introduction Methods 7.7.1 Electrical Sample Splitting Bocek et al. [69] and Everaerts et al. [70] have both used an electrical sample splitter for sample introduction in CZE and ITP. The principal and design of their splitters are shown in Fig. 7-14. The splitting ratio is determined by the ratio of current between the different capillaries. The 200 pm ID capillaries used in Bocek’s design was an integral part of the whole apparatus, formed using casting technology, while Everaerts used a 250 pm ID PTFE separation capillary. Most CZE applications now
228
7 Application of Capillary Electrophoresis for Environmental Analysis
a
%I
II/'
1
AI
b
-C
85
SO
A2
3
!I
eal2
Fig. 7-14. Principle and design of the splitter. The current, I , , drives the original sample plug, n,, through the dosing capillary. The current is then split into I,, driving an aliquot of sample, n,, into the analytical capillary whilst Z, drives the remaining sample, n, = n, - n 2 , to drain. Everaerts et al. used a double-T splitter of the same principle. (Reproduced with permission [69,70]).
use <100pm ID silica capillaries, thus a new design would be required for these methods to become popular.
7.7.2 Rotary Valve Injector Tsuda et al. [71] fabricated a rotary type injector similar to those used in liquid chromatography. The schematic design and reproducibility data are shown in Fig. 7-1 5. The use of fine ceramics and tetrafluoroethylene resin ensured good electrical insulation, enabling injections to be made with the high-voltage applied. FEP tubing ( 0 . 2 ~195 mm) served as the capillary. Sample left in channel 6 prior to injection may decompose over time due to the application of the high electrical field strength. It should be possible to construct such a device for capillaries with IDS of 100 pm or less, the main difficulty will be in fabricating a disc sufficiently thin and strong to enable the loading of, say, 100nL or less.
7.7 Novel Sample Introduction Methods
6
229
1
e--L 6
3 2 1 2 3 8(a)
C
8(b)
Fig. 7-15. Schematic design of a rotary injector made of fine ceramics. (1) rotor; (2) stator; (3) plate for setting rotor and stator; (4) central pin; ( 5 ) extra tubing between the injector and reservoir; (6) tubing for sample introduction; (7) analytical capillary; (8) knob made of stainless steel covered by silicone tubing. Positions (a) and (b) are load and inject respectively. (Reproduced with permission [71]).
7.7.3 Slider Valve Sample Introduction Jarofke [72] used a slider valve for sample introduction in ITP (Fig. 7-16). The sampling device was constructed using PTFE for the outer housing and plexiglass for the slider itself. The device enabled precise placement of the sample plug between the LE and TE. Sample was introduced through the lower sample channel, with the slider in the 'down' position. Using compressed air applied to the lower piston, the sample is then switched to the separation channel and the potential applied. The device allows simultaneous flushing of the capillary with the chosen electrolytes whilst filling the sample channel prior to introduction. The slider valve is certainly a promising method for introducing a defined volume of sample into capillaries, however only one other design has been published [73]. Both designs use > 250 pm ID PTFE capillaries for the separation. A slider valve designed for < 100 pm ID silica capillaries has yet to be published, but is certainly desirable.
Drain (TI 4
Terminal electroly
Capillary tube
VA'Glider U
'
byringe
Piston
t Compressed air
Fig. 7-16. Schematic design of the slider valve for ITP with the sample just introduced into the analytical capillary. Pressure on the top piston allows sample to be applied while rinsing the capillary. (Reproduced with permission [72]).
230
7 Application of Capillary Electrophoresis for Environmental Analysis
7.7.4 Membrane Sample Introduction Dasgupta and Bao 1741 developed a novel membrane interface for sample injection. The design of the injector and some electropherograms are shown in Fig. 7-17. The required membrane is attached to the capillary and a protective jacket sheaths it. Sample flows through the gap and is introduced via diffusion/permeation through the membrane. Using porous membranes sampling of trace gases could be accomplished. Silicone rubber membranes require the use of a desorption wash to minimise peak tailing resulting from the 'dead zone' of the reservoir and membrane interior. Preconcentration of the sample may also be achieved by increasing injection times.
P
P
(4 M
A
(B)
*
* A
(C)
t
A'
A
E
M
A'
M
A'
A B
C
0
A'
Fig. 7-17. Membrane sample introduction. Connecting different membranes to silica capillaries: A and A' = silica capillaries 0.075 mm IDx0.375 mm: M = membrane. (A) m = 0.10 mm ID porous polypropylene membrane, exposed for 1.5-2mm length. P = PEVA tube segment, 0.25 mm ID. (B) Silicone membrane (originally 0.25 mm IDX0.45 mm OD) into which the capillaries are forced. A and A are separated by 0.1-0.2mm. (C) The sampling interface with jacket installed. (D) B = sealant (typically PVC tube); C = micro-tee and D = PTFE tubular jacket. (Reproduced with permission [74]).
Wallingford and Ewing [75] have demonstrated the use of a 'micro-injector' for sample introduction, using a dual-barrelled device. The micro-injector possessed a minature high-voltage capillary electrode alongside the tapered CE capillary, enabling injection of sample from single animal cells. The method used for sample introduction is clearly critical in CE, and advances are really required to facilitate the introduction of a known absolute volume of sample to the capillary. Many of the proposed methods show real promise, however it is the problem of construction rather than design that hampers progress. Nevertheless, new methods of sample introduction will remain of interest.
7.8 Methods for Increased Sample Loading
23 1
7.8 Methods for Increased Sample Loading To enhance the separation/sample capacity of CE often requires some form of sample pretreatment for clean-up purposes/preconcentration. Minimal sample preparation is prefered if sample integrity is to be maintained, thus also reducing the total time required per analysis. However, when the sample contains the analyte of interest at very low levels, some form of selective preconcentration will usually be required to enable quantitation, especially with complex enivronmental samples containing disparate levels of ions.
7.8.1 Two-Dimensional Electrophoresis Two-dimensional electrophoretic separations have been shown to be effective for online concentration and clean-up of complex matrices. Ornstein [76] and Davies [77] first demonstrated the principle with disc gel electrophoresis, using ITP in the first dimension followed by a zone electrophoretic separation in the second.
7.8.2 ITP- ITP Mikkers et al. [78] developed coupled column ITP for the analysis of low concentrations of analytes in complex mixtures. Two PTFE capillaries were used; the first ‘preseparation’ tube was of relatively large bore to allow increased sample volumes and separation current. Its purpose was to preconcentrate the sample prior to transfer to the secondary, analytical capillary. The two capillaries were coupled using a T-piece, termed the bifurcation block, perpendicular to the separation capillaries and connected to a secondary electrode (Fig. 7-18). The block is used to remove the components of no interest by the correct switching of current between the electrodes 10 and 16. Coupled-column ITP also permits the use of different electrolytes for each separation stage, allowing increased selectivity and sensitivity in the analytical capillary. The coupled-column system was later produced commercially [79] and has found widespread use in isotachophoretic analyses.
7.8.3 ITP-CZE The requirements for coupling ITP-CZE are now fairly well understood. A number of publications have examined various aspects of the coupling, including those relating to the different instrumental configurations, ie, coupled column and on-column [80-871.
232
7 Application of Capillary Electrophoresis for Environmental Analysis
Fig. 7-18. Apparatus for coupled column ITP. (1) TE reservoir; (2, 8, 11, 18) FTFE lined valve; (3) injection block; (4) connection to drain; ( 5 ) pre-separation capillary (0.8 mm ID); (6) conductimeter; (7) T-piece; (9) counterelectrode compartment with membrane; (10) counter-electrode filled with doubly distilled water; (12) analytical capillary (0.2mm ID); (13) UV detector; (14) conductimeter; (15) counter-electrode compartment with membrane; (16) counter-electrode filled with doubly distilled water; (17) syphon tap; (19) power supply, X = distance from telltale conductimeter (6), to bifurcation point. (Reproduced with permission [78]).
7.8.4 Coupled Column ITP- CZE Coupled-columns have also found a use in ITP- CZE; allowing large volumes of dilute samples to be preconcentrated in the ITP step, and transferred to the CZE capillary in an almost ideal sharp initial zone for CZE. The selective analyte transfer enables removal of macro components not of analytical interest which, in addition, could lead to overloading the capillary leading to irreproducible migration times. A schematic view of ITP-CZE instrumentation is shown in Fig. 7-19. The comparison of both ITP-ITP and ITP-CZE for the determination of a broad spectrum coccidiocidic in feedstuffs [89] demonstrated the utility of CZE in the analytical capillary. Using ITP- ITP in trace analysis can lead to very short zones that are difficult to quantitate and identify. As can be seen from Fig. 7-20, the ITP- CZE analysis provided excellent resolution of the separand (HalofuginoneHFG). The excess of small mobile ions were removed from the sample train prior to analysis in the CZE stage. High sensitivities were achieved ( l o p 8M) using UV detection, with good reproducibility (RSD 1Yo). The timing and method of sample transfer from the preseparation to analytical capillary are critical for reproducible and quantitative analysis. Following Mikker’s
7.8 Methods for Increased Sample Loading
L Fused-sika(
I
I
Terminating electrolyte
p,&
-Sample loop (12pL)
capillary Capillary joint
233
Small volume injection (3pLl
Septum injector Conductivity detector
-
Comparator
Fig. 7-19.Schematic diagram of the coupled column ITP-CZE apparatus. The comparator enables the automated switching of the potential between the LE and BGE reservoirs by comparing the conductimeter output with a preset reference voltage (U,) ie, sample transfer between capillaries is automated. P = potentiometer. (Reproduced with permission [88]).
1550
1500
1450
1400
1350
700
900
1100
1300
w
Fig. 7-20. The comparison of coupled column (A) ITP-ITP and (B) ITP-CZE. Note that the HFG zone is not fully resolved in (A), due to a zone with a similar mobility in the ITP stack. A zone electrophoretic separation in the analytical capillary (second dimension) provides excellent resolution of HFG from the previously co-migrating analyte. (Reproduced with permission [89]).
[78] approach transfer occurs through switching the current between the bifurcation block and the end of the analytical capillary, ie, the process is essentially an electrokinetic injection. Van der Greef et al. [90, 911 used hydro-dynamic injection with electro-kinetic injection, whereby an injection of TE into the TE reservoir (closed system) when the sample train reached the pre-separation detector (positive displacement), or by lowering the CZE outlet for a set time. In all of the above methods the ITP stage may be fine-tuned to minimise/maximise the number of sample compo-
234
7 Application of Capillary Electrophoresis for Environmental Analysis
nents migrating between the leading and terminating electrolytes; thus, ITP preconcentration alone can be used to increase selectivity. The coupled columns technique requires specialised apparatus and the majority of commercial instruments are not configured for this application. Adaptation is possible but, some capillary ‘cartridge’ based systems may not allow access for coupled columns.
7.8.5 Choice of Electrolyte Systems for ITP-CZE When attempting ITP- CZE in the coupled column arrangement the combination of the working electrolytes must be considered. Krivankova et al. [92] used computer simulations of the processes occurring during and after the transfer of analytes between capillaries. Three basic arrangements for electrolyte combination have been proposed (Fig. 7-21). The simplest coupling may be achieved using the TE for the BGE in CZE (Fig. 7-21 A). Sample is injected between the ITP electrolytes, with current applied between the leading and terminating electrodes. At the bifurcation block the LE is removed to the secondary electrode, and the sample train is directed to the CZE capillary by switching the current to the end of the analytical capillary. CZE then proceeds in the terminating electrolyte. In practice this method may be difficult to reproduce, and it is often advantageous to transfer a certain volume of the leading electrolyte for all combinations of electrolytes (usually defined by the distance from the first detector to the capillary to ensure trapping of zones): This results in focussing of zones after the high voltage switching, minimising dispersion due to mixing during transfer, flushing, etc. The second combination (Fig. 7-21 B) uses the LE for the CZE analysis. The ITP separation proceeds as before. However once the analyte zones have entered the CZE
System
0
(c)
ITP
0
0
Fig. 7-21. Schematic representation of electrolyte systems used in coupling ITP with CZE. The left panel shows the situation in the preseparation capillary at the end of the ITP stage. The right panel shows the situation in the analytical capillary at the beginning of the CZE separation for (A) the terminating electrolyte (T) as BGE (T-S-T); (B) the LE (L) as BGE (LS-L), and (C) a different electrolyte completely as the BGE (BGE-S-BGE). S = sample. (Reproduced with permission [92]).
7.8 Methods .for Increased Sample Loading
235
ZE
ITP
d
170 50 Time is)
Fig. 7-22. On-line sample clean-up using ITP prior to CZE. Discrete spacers are added to the sample to obtain defined fractions of the sample. These fractions may then be directly transferred to the analytical capillary. (A) The fraction marked (a) from the isotachopherogram is transferred to the CZE capillary. Only the main part of the chloride zone and all sample constituents migrating zone electrophoretically in the TE were removed. (B) The fraction transferred was defined by the spacing constituent S, and by the terminating zone. Sample clean-up is not high in this instance. (C) Only the sample zones migrating between the two spacers were transferred to the CZE capillary, enabling 95% of the sample matrix constituents to be removed prior to CZE. (Reproduced with permission [93]).
capillary, the high voltage is switched off and the TE flushed out with LE. The use of an applied voltage during rinsing to minimise sample back-flushing may be necessary [90]. The final electrolyte combination uses a BGE for CZE different to both the ITP electrolytes (BGE - S - BGE). The procedure follows that of the LE - S - LE system except, following injection into the CZE (containing BGE) capillary, the preseparation capillary is flushed with the BGE (Fig. 7-21 C). The utility of each method is largely sample dependent, however the use of a different BGE for CZE allows a more orthogonal approach to the separation, ie, a true two-dimensional separation [81]. Kaniansky et al. [93] have demonstrated the use of ITP-CZE for sample pretreatment in the analysis of two model compounds artificially added to urine (Fig. 7-22). Neither compound is known to be present in urine. Although not directly related to environmental analysis, the urine matrix demonstrates the possibilities for the analysis of a wide range of complex media. Analyte zones were bracketed by spacers in the ITP analysis allowing efficient transfer of specific sample fractions to the CZE capillary. Kaniansky et al. [93] improved the sample clean-up by minimising the number of analyte zones in the ITP stack and minimising the ITP to CZE electrolyte similarity, ie, in the first dimension (ITP) the separation is performed according to pK, values
236
7 Appficafion of Capillary Electrophoresisfor Environmenful Analysis
(minimum of analyte zones in stack, controlled by dissociation), whilst in the second dimension resolution was mainly governed by differences in ionic mobilities. The combination enabled a 150ppb concentration detection limit for the sulphanilate ion in urine, and 5 ppb in model mixtures (25 pL injection) [93]. Coupled column ITP- CZE enables one to undertake difficult separations and obtain good resolution between components that neither technique alone would easily accomplish. Despite this, the method requires specialised, but easily constructed, instrumentation. In another approach a single capillary is used for the ITP- CZE analysis (see below).
7.8.6 On-Column ITP- CZE: Transient ITP and Sample Stacking Mikkers et al. [94], in some of the first applications of CZE, used the stacking method for sample application. This method involves dissolving the sample in a solvent with lower conductivity than the electrolyte (eg, water) and the sample is applied using hydro-dynamic or electrokinetic injection. Upon application of the high voltage, sample components stack due to the lower conductivity of the sample solution with respect to BGE, which results in a higher potential gradient across the applied sample. This results in a very narrow zone of analyte being formed, leading to higher system efficiencies (Fig. 7-23). The on-column approach may also be performed in a number of other ways, depending on sample composition and the BGE (Fig. 7-24). For samples or electrolytes
1
3
3
1
I
0.01 A 254nm
L
12
(A)
8 4 Time imin)
o1
1
:
1
2
8 4 Time imin)
0
(6)
Fig.7-23. The effect of sample conductivity on the CZE separation of three anionic analytes. (A) Constituents dissolved in water; (B) constituents dissolved in BGE. (Reproduced with permission [94]).
7.8 Methods for Increased Sample Loading
I
fl
f2
'x'o-2A--
I
L+S
c (B)
237
L S
2AO
2.60 2.80 3.00 Time (mid
3.20
Fig. 7-25 Fig. 7-24. Different modes of transient ITP preconcentration in on-column ITP-CZE. (A) The BGE co-ion, has a higher mobility than sample components (S), serving as the LE (L). A suitable TE (T) can be used for preconcentration which is then replaced with the BGE (C) reservoir at time t , , after which the separation proceeds in the CZE mode. A suitable terminating ion may also be added to the sample prior to analysis. (B) The BGE co-ions' mobility is sufficiently low to serve as the TE, thus a suitable leading ion may be added to the sample, or a sample constituent may function as the LE. (C) Both previous modes may be combined and this method will always occur when using method (A) to analyse a sample containing salts of highly mobile ions. (Reproduced with permission 1881). Fig. 7-25. The effect of the addition of a terminating ion to the sample in transient ITP-CZE. Octanesulphonate was added in varying amounts to the sample solution. BGE = l O m ~Na chromate, pH 8 (dil. H,SO,). 0.5 mM NICE Pak OFM anion-BT (Waters Chrom. used for EOF reversal). HT = - 15 kV. INJ = 30 s at 5 kV. Peaks: (1) chloride (35 ppb); (2) Sulphate (48 ppb); (3) nitrate (62 ppb); fluoride (19 ppb). (Reproduced with permission [95]).
with ions of high mobility in the direction of the EOF, an internal terminating ion can be added to the sample prior to electrokinetic injection (Fig. 7-25), or the electrolyte co-ion may be chosen so that its mobility is lower than that of the analytes. On application of high voltage the sample components concentrate between the leading and terminating components (Fig. 7-26).
238
7 Application of Capillary Electrophoresis f o r Environmental Analysis IAl Stage one: capillary and electrode
(6) Stage two: beginning o f electromigration
placed i n t o a sample
across the concentration boundary
T o anode
Sa
1 xox (C)
OO
xx[
Concentration boundary
Stage three: isotachophoretic steady s t a t e
I X
Concentration
x
x ~ I boundary
(Dl Stage four: capillary and electrode placed back i n t o the carrier electroiyte
To anode Capil
. Concentration boundary
Concentration boundary
Fig. 7-26. The four stages of sample introduction using transient ITP preconcentration. In this case the BGE serves as the LE. analyte anion, 0 carrier electrolyte anion, x sample matrix anion. (Reproduced with permission [96]).
Transient ITP-CZE may be accomplished by applying the sample, followed by transferring the capillary inlet to a suitable terminating electrolyte. Application of voltage results in the stacking of sample analytes between the LE and TE. After a predetermined time the capillary inlet is then placed back into the BGE (LE), which then penetrates the stack, causing zone destacking due to the shift in pH. Separation then proceeds essentially under CZE conditions. When the sample or BGE contains no suitable leading ion, or certain high mobility zones are required to be out-of-stack, the addition of a leading ion to the sample is required (Fig. 7-24). The BGE (if its mobility is low enough) or a sample component may be used as the terminating ion. Many samples will contain ions that can fulfil both requirements; this will often be the case where samples contain salts of highly mobile ions. Karger et al. [88] evaluated both coupled column and on-column ITP-CZE for analysing a standard protein mixture. A 50-fold preconcentration of sample could be achieved using the on-column coupling, while the coupled column arrangement achieved a gain in detectability of at least three orders of magnitude. The combina-
7.8 Methods for Increased Sample Loading
239
tion of a very high sample loading (c. 12 pL);preconcentration; sample clean-up and selected ion analysis using judicious current switching ensure that the coupled column technique will always outperform on-column transient ITP preconcentration. The major limitation of the technique is the lack of commercial availability. Coupled column ITP- CZE apparatus will undoubtedly appear on the market at some point, as occurred for ITP- ITP, and will provide a significant step forward for the determination of trace analytes by electrophoretic separations.
7.8.7 Field Amplified Sample Injection (FASI) for CZE Chien and Burgi [97 - 1021 have used field-amplified polarity-switching injections (FASI) in CZE to increase sample loading. Positive and negative ion concentrations may be enhanced either together or individually in a single injection. The method is schematically shown in Fig. 7-27. FASI may not be suitable for some commercial apparatus as rapid polarity switching of the power supply is required. Several orders of magnitude in signal enhancement is achievable.
Samole vial
Sample vial
+I
I
I
3I jI I
lOOmM buffer
H2°
I
i&$l
<
100mM buffer
-
LW Capillary column
' 4,
Fig. 7-27. A schematic representation of FASI. (A) A short plug of water is introduced at the inlet prior to introduction of sample (dissolved in water). The positive ions are injected first using a positive polarity potential. (B) The negative ions are then injected using a negative polarity potential. (C) After sample injection, the capillary inlet is transferred to the BGE reservoir and the polarity is switched back to positive. The CZE separation now starts. (Reproduced with permission [loll).
7 24Q
7 Application of Capillary Electrophoresis for Environmental A n
approaches to on-line sample pretreatment include the use of ch graphic micro-precolumns [ 104- 1061 and the onloff-line coupling w‘0Qato, [106-108]. ~ 1 of 1 the methods described may be combined with off-line k pretreatment (eg, solid phase extraction), to improve further the absolute detectio; SaQPI
7.9 Detection Methods for CE
241
limits. It is expected that the use of selective sample pretreatment will greatly extend the utility of electrophoretic separations.
7.9 Detection Methods for CE Detection of analytes in CE has been achieved using a variety of methods: electrochemical [109]; radiometric [ 1lo]; fluorescence [l 1I]; axial-beam absorbance [112]; multi-channel Raman spectroscopic [I 131; conductivity [114]; mass spectrometric [115]. In-column UV-vis detection is by far the most common mode, and all commercial systems use UV detection as standard. Fluorescence detection is available with some instrumentation. Optical detection methods are limited in their sensitivity due to the minimal pathlength of the capillary ID, and the possibilities for optimisation of the detector cell arrangement have been evaluated [116]. Various capillary cells (Z-shaped [I 171; extended pathlength [118]; multi-reflectance cells [119]; focussing lenses [120]) have all been used to optimise detection. Zare et al. [121] increased the optical pathlength for UV detection by using rectangular capillaries equipped with a specialised injection system [122]. A gain in detectability of nearly 20 was achieved when using 1 mmx 50 pm capillary (detection across 1 mm) in comparison with a 50 pm circular capillary. The use of alternative detection methods have also demonstrated sensitive detection of environmental pullutants. Electrochemcial detection has been successfully used for the determination of chlorophenols in industrial waste waters [I231 and some metal standards [124]. Electropherograms for both separations are shown in Fig. 7-29. The use of direct UV absorption detection in CE is obvious, but not all analytes possess a sufficiently strong UV absorbing chromophore. It is possible in these circumstances to detect at 185- 200 nm when many analytes will have some UV absorption. However, the background may be unacceptably high at this low wavelength and restricts the choice of BGE it is possible to use, especially the addition of organic solvents. Due to this, indirect methods of detection have been developed.
7.9.1 Indirect Optical Detection Methods The predominance of UV detectors in commerical instruments has led to indirect UV methods being developed with excellent results. Part per trillion sensitivities have been achieved for small, high mobility ions using sample stacking. The simplicity of indirect UV detection will no doubt ensure its popularity, extending the capabilities of present commercial CE systems to allow almost universal analyte detection. Jandick and Bonn [125] have extensively reviewed indirect optical detection for CZE in their recent book.
242
7 Application of Capillary Electrophoresis for Environmental Analysis
- 4.65 P % 4.55 L
U 3
4.45
I
0
c
4
8
12 16 Time ( M i d
20
4.35
24
I
,
,
200
!
,
400 Time (s)
,
600
,
)
800
Fig. 7-29. Electrochemical detection of inorganic and organic pollutants in CZE separations. (A) Electropherogram of an industrial waste water sample. BGE = NaH,PO, (0.045 M) - sodium borate (0.015 M), pH 8. C = 65 crnx0.025 mm ID. HT = 20 kV. INJ = 30 s at 20 kV. Peaks: A = 2-~hlorophenol(50 ng mL-I); B = 2,4-dichlorophenol; C = 2,6-dichlorophenol; D = ophenylphenol; E = 2,3,4,6-tetrachlorophenol;F = 4,5,6-trichloroguiacol; G = pentachlorophenol; H = phenol; I = 4-chlorophenol; J = catechol; K = 2,4,6-trichlorophenol. LOD = 0.001 ng m L - ' . (Reproduced with permission [123]). (B) Electropherogram of four transition metals. BGE = 0.005 mol L - N,N-dimethylbenzylamine (DBA), 0.0065 rnol L-' HIBA: pH 4.9 (acetic acid). C = 60 cmx0.025 mm. HT = 30 kV. INJ = 2 s at 1OkV. 1.10-6molL-' Pb2+, and 1 ~ 1 0 - ~ m o l L for - ' other ions. L O D = 2 ~ 1 0 - ~ r n r n o l L ~ ' for Pb2+. (Reproduced with permission [124]).
'
7.9.2 Indirect Optical Detection for ITP Indirect optical detection using UV [126] and fluorescence [ 1271 was first demonstrated with ITP. The use of a counter-ion in the leading electrolyte whose absorption was pH dependent enables a stepwise detector response, due to the stepwise pH gradient of the contiguous zones in ITP. Clearly, the response is maximised when the counter-ion's molar absorptivity varies significantly with small pH shifts. A secondary effect may occur if a strongly absorbing co-counter-ion is added to the LE, in that the step-wise decrease in counter-ion concentration in successive zones may be detected. In low p H anionic ITP analysis, quinine (pH 3 - 5 ) and creatinine (pH 4- 5.5) may be used for both UV and fluorescence detection, whilst for cationic analyses the sulphanilate ion (pH 3 - 5 ) and barbital ion (pH 7 - 8.5) can be used [126- 1271.
7.9.3 Indirect UV Detection for CZE The use of indirect UV detection in CZE has centred on the analysis of low molecular weight analytes usually undertaken by ion chromatography [128- 1341. Bocek et al. [I351 described the fundamental principals of this method, whereby the electrolyte co-ion shows strong absorption. Detection is indirect because the sample zones
7.9 Detection Methods for CE
243
displace the co-ion, resulting in a reduction in absorbance, ie, negative peaks. Maximum sensitivity and efficiency are obtained using low conductance samples and matching the effective mobilities between sample and electrolyte ions. The match between mobilities is critical for good peak shape and thus sensitivity.
7.9.4 Optimisation for Indirect UV Detection in CZE A number of parameters can be optimised using indirect UV detection: the voltage applied for injection (if used) and analysis; the elctrolyte pH and concentration: the use of complexation to maximise resolution between similar mobility ions: the use and choice of organic modifiers to the electrolyte: concentration of the EOF modifier (if required); and, of course, the absorbing co-ion and choice of detector wavelength. It is generally desirable to have the analytes mobility and the EOF move in the same direction the minimise analysis times. A few examples of possible ways to optimise separations and a discussion on some problems that might be encountered are considered below.
7.9.5 Optimisation of Injection Method A comparison of hydro-dynamic and electrokinetic injection demonstrates the selectivity and enrichment that may be gained from a well defined sample stacking regime [136]. Figure 7-30 shows the comparison of the two injection methods for an estuarine water sample. An approximate ten-fold gain in sensitivity was achieved using electrokinetic sample introduction. Nitrite however could not be detected using electrokinetic injection as the sulphate peak overlapped. Direct UV detection enabled the determination of nitrite.
7.9.6 Electrolyte System Optimisation The mobility of analytes and the EOF velocity are a function of the ionic strength and pH of the electrolyte. The addition of some materials and organic solvents can also affect these parameters. Figure 7-31 shows the drastic effect a change in electrolyte pH can have on the separation. By raising the pH the NH; ion becomes less protonated, and its effective mobility decreases. An increased EOF is also observed, as the silanol @K, 7-8) groups on the capillary wall become more ionised [137]. Optimisation of the electrolyte pH can be easily achieved by analysing the sample under three of four significantly different pHs and interpolating for best resolution, thus providing a relatively simple method for fine-tuning separations. Harrold et al. [138] demonstrated the use of an electrolyte whose mobility was strongly pH dependent. The analysis of some highly mobile anions (pH 7.7) (eg, thiosulphate) and less
244
7 Application of Capillary Electrophoresis for Environmental Analysis
2.20j. 2.40
, ,
,
, ,
2.60
.,
, ,
.,
, , ,
,
,
. . , . .. , . , , . ,
2.80
3.00 3.20 3.40 3.60 Migration time bin)
2.80
3.00 3.20 3.40 Migration time (min)
(A)
3.80
, ,
,...,
4.00
4.20
1
- 2.40 x
2.35
5 2.30 n
a 2.25 2.20
(6)
...3.60
3.80
400
Fig. 7-30. A comparison between hydrodynamic and electrokinetic injection. (A) Electropherogram resulting from a hydrostatic injection of an estuarine water sample. BGE=SmM chromate with 0 . 5 m ~ . CIA-Pak OFM ANION-BT (Waters) at p H 8 . C = 60 cmx0.075 mm ID. HT = 20 kV. INJ = 30 s. Indirect UV detection at 254 nm. Peaks: (1) chloride (12.4pg mL-'); (2) sulphate-S (1.1 pg mL-I); (3) nitrate-N (0.4 pg m L - ' ) ; (4) carbonate. (B) Electrophergram of the same sample using electrokinetic injection. Conditions as for a, except: INJ = 30s at 3 kV. Peaks: (4) fluoride (0.031 pg mL-'); ( 5 ) formate; (6) phosphate-P (0.013 pg mL-'); (7) carbonate. LQD = low ( < 10) ng mL-'. (Reproduced with permission [136]).
mobile haloacetic and alkylsulphonic acids (pH 3.5) were achieved using pyromellitic acid and detection at 250 nm. Selectivity was changed by varying the electrolyte pH (Fig. 7-32). The ionic strength of the BGE is also important to optimise. The ionic strength should be minimised to reduce Joule heating which affects both analyte mobility and diffusion rates. However the subsequent instability of the co-ion absorbance increases noise and drift. A decrease in the BGE ionic strength must be balanced by the con-
7.9 Detection Methods for CE 4
4
2.0 Time imin)
1.6
1.8
(4
245
2.2
1.2
1.4 1.6 Time (min)
(B)
Fig. 7-31. The effect of BGE pH on the separation of the Group IA metal ions. BGE = 5 mM morpholinoethanesulphonate (MES). C = 52 cmx0.075 mm. HT = 25 kV. INJ = 10 cm for 30 s. Peaks: (1) potassium; (2) ammonium; (3) sodium; (4)lithium. (Reproduced with permission [137j).
r
(
3.0
&
.
,
I
~
4.5
)
~
4.0
~
1 I
1
4.5 Time (min)
1 I
1
5.0
,
~ 1
1
I~
5.5
~
~,
~
1
1
1
1
6.0
Fig. 7-32. Separation of alkylsulphonic acids by CZE with indirect UV detection (250 nm). BGE = 2.5 mM pyromellitic acid, 0.75 mM hexamethonium hydroxide, pH 3.5 (NaOH). C = 50 cmx0.050 mm. INJ = 100 mm for 20 s. Peaks (all 2.0 mg L-'): (1) methanesulphonate; (2) ethanesulphonate; (3) 1-propanesulphonate; (4)1-butane sulphonate; ( 5 ) 1-pentanesulphonate; (6) l-hexanesulphonate: (7) 1-heptanesulphonate; (8) 1-octanesulphonate. (Reproduced with permission [138]).
current decrease in peak efficiency and sensitivity, ie, the ionic strength must be sufficiently high to allow a separation to be achieved with sufficient sensitivity.
7.9.7 Addition of Organic Solvents to Electrolyte System In ITP a common method to separate ions with very close mobilities is to add an organic solvent to the electrolyte. This changes a number of the separation parameters: eg, pK,s, electrolyte conductivity, viscosity and solvation effects. Figure 7-33 shows how the addition of methanol to the electrolyte can effect the separation. Commonly used organic modifiers include acetone, acetonitrile, ethanol; methanol is a good modifier in a large number of cases. Haddad and Buchberger [139] examined the effects of five common solvents (up to 30%) on th separation of ten inorganic anions. From their results no predictions
246
7 Application of Capillary Electrophoresis for Environmental Analysis
I
2.8
3.0
5.5
3.4
3.2
3.6 Migration time ( m i d
6.0
6.5
7.0
7.5
3.8
4.0
8.0
Migration time (min)
Fig. 7-33.The effect of the addition of methanol to the BGE on the separation of some inorganic anions. Peaks: (1) thiosulphate; (2) bromide; (3) chloride; (4) iodide; ( 5 ) sulphate; (6) nitrite; (7) nitrate; (8) chlorate; (9) thiocyanate; (10) fluoride. The electrolyte in (B) contains 30% methanol in the BGE. (Reproduced with permission [139]).
could be made on the effect of organic modifiers but changes in migration times of the analytes were observed which affected the order of the separation.
7.9.8 Addition of Complexing Reagents to Electrolyte System The inclusion of a reagent whose complexing characteristics are dynamic and rapid enables a separation to be achieved due to the different degrees of complex formation, eg, the separation of the alkali, alkali earth and lanthanides is difficult to achieve due to their very small relative mobility differences, and BGE pH changes show very little effect on apparent mobility. Complexing additives in the BGE can solve this problem. Chen and Cassidy [I401 used hydroxyisobutyric acid (HIBA) as the complexing agent for the separation of 26 metal ions (Fig.7-34). N,N-Dimethylbenzylamine served as the absorbing co-ion, whose concentration was adjusted until the background absorption reached the maximum of the detector linear range (0.05 AU with 50/75 pm ID capillaries in this instance), which provided the maximum linear calibration range [140]. The separation of 27 metals by the selective complexation with lactate ion has been demonstrated [ 1411 with detection limits in the range 0.05 -0.5 pg/mL (50- 500 ppb).
7.10 Conductivity Detection
247
0.0528 0.0526 0.0524 -
-a 0,0522 3 W
$
n L
0.0520
-
ia 0.0518 0.0516
-
-
12
14
0.0514 0.0512
p
2
3
4
5
6
7
8
9
1
0
Retention time (min)
Fig. 7-34. Electropherogram for the separation of 26 metal ions by CZE with indirect UV detection (214 nm). BGE = 6 mmol L-I DBA and 4.2 mmol L-' HIBA, 0.2 mmol L-' Triton X-100, pH 5 (acetic acid). C = 60 cmx0.075 mm ID. HT = 30 kV. INJ = 6 s at 9.8 cm. LOD = c. 1 vg mL-' for the lanthanides, 0.6 pg mL-' for the transition and alkali earth ions and c. 0.1 -0.8 pg mL-' for alkali metal ions. Peaks: (1) K + ; (2) Ba+; (3) Ca'; (4) Na'; ( 5 ) Mg2+; (6) Mn2+; (7) Fez; (8) Co2+; (9) Ni2+; (10) Zn2+;(11) Li+; (12) La3+; (13) Ce3+;(14) Pr3+;(15) Nd3+;(16) Sm3+;(17) Eu3+;(18) Gd3+; (19) Cu2+; (20) Tb3+; (21) Dy3+; (22) Ho3+; (23) Er3+; (24) Tm3+; (25) Yb3+ (26) Lu3+. (Reproduced with permission [140]).
7.10 Conductivity Detection Conductivity detection has been widely used in CE. The use of conductivity detectors (conductimeters) were predominant in ITP, as the stepwise detector output allows both quantitative (step length) and qualitative (step height) data to be extracted from a single separation. A number of conductimeter designs for CZE have been published, with commercial designs now available from Dionex and ATi. An analyte zone's conductivity is typically measured by two electrodes in contact with the solution, the detection cell forming part of the separation capillary. The resulting passage of current by the analyte conductivity is then processed using an electronic device and recorded. Minaturisation may be achieved without loss in sensitivity, unlike optical detection. The galvanic isolation of the detector (using an isolation transformer) is required to prevent current leakage to ground through the detector, and thus irreproducibility,
248
7 Application of Capillary Electrophoresis for Environmental Anulysis
7.10.1 Contactless Conductivity Detection An alternative approach for ITP has been to use an alternating high-frequency signal (1 MHz) applied externally to the capillary. The distribution of the electromagnetic field being determined by the conductivity of the isotachophoretic zone [142]. Vacik et al.’s [I431 design is shown in Fig. 7-35. Electronic thermostatting of temperature dependent detector parts led to good reproducibility and stability of the detector response (differing by 0.5% after 30 minutes). The minimum measurable zone length was 0.3 mm, with detection limits similar to the contact version utilising a 0.001 M LE.
Cross section
A- B - 2 3
Fig. 7-35. Schematic diagram of the contactless conductimeter. (A) Arrangement of the capillary and electrodes (capacitive cell). (1) copper enamelled wires (0.2 mm diameter); ( 2 ) and (3) the shielding, consisting of a lower metal plate with milled flutes for the electrodes (sealed with polystyrene adhesive) and a hole for the capillary and a top plate to screw onto the lower. (B) Principle of the detector. (1) generator; ( 2 ) capacity cell; (3) receiver; (4) recorder. (Reproduced with permission [143]).
A contactless detector has certain advantages over its counterpart, as long term stability is unaffected by possible electrode processes occurring and dead volumes are avoided. Since Vacik et af.’s work no interest has been shown in the contactless conductimeter. Such a design may be of use in fabricating detectors for capillaries < 100 pm ID, particularly in commercial instruments.
7.10.2 In-Contact Conductivity Detection Mikker ef al. first demonstrated the use of contact conductivity detection in CZE [144], using equipment designed for ITP without EOF. Foret et al. [145] demonstrated off-column detection of chlorine, sulphate and nitrate with detection limits at M.
Huang et al. [146] used two platinum electrodes (25 pm OD) inserted into two laser drilled holes (40 pm OD) exactly opposite each other. The platinum electrodes
249
7.10 Conductivity Detection
1'__ '1i' " K+
+ Ym
-0
U 0 I
I
0
1
I
2
3
4 5 6 Time (min)
7
8
Fig. 7-36. (A) Diagram of the on-column conductimeter for CZE and (B) an electropherogram of the separation of four cations, R b + , K + , Na' and Li+ (all at 2x lo-' M). BGE = 20 mM MES, pH 6 (histidine). C = 60cmx0.075 mm ID. HT = 15 kV. INJ = 5 s at 5 kV. LOD = lo-' M for Li'. Refer to text for explanation. (Reproduced with permission [146]).
were temporarily held in place with heated poly(ethy1ene) glycol (PEG). Once solidified, the PEG was removed from the outside of the capillary. An epoxy resin produced a permanent seal. Wires were then soldered to the electrodes, with the entire cell being placed in a plexigalss jacket (Fig. 7-36). The AC conductimeter followed Everaerts et al. 's [42] design principles but with a reduction in detector noise. Huang et al. [147,148] also demonstrated the analysis of low molecular weight carboxylic acids using the same detector design, noting the relationship between analyte ionic mobility and peak area. The use of an internal standard, at known concentration, enabled quantitation of all components without the need for numerous calibration plots. A reduction in BGE concentration, whilst maintaining the BGE to analyte ratio, increased sensitivity whilst maintaining resolution, until instrumental M. and contamination factors became dominant. Detection limits were An end-column conductimeter of simpler design has also been constructed [149]. Construction is similar to previous designs, except the sensing electrode (50 pm OD platinum wire) was fed up through the capillary to a 40 pm laser drilled hole approximately 7 mm up from the end of the capillary. The electrode is then sealed in the end of the capillary using epoxide and attached to a fine, insulated, lead wire connected to the previous ac circuitry. A protective jacket with a c. 1 mm hole is aligned with the hole in the side of the CZE capillary, allowing electrolyte to flow out when the device was placed in a earthed buffer reservoir (Fig. 7-37).
250
7 Application of Capillary Electrophoresis for Environmental Analysis
Fig. 7-37.Construction and design of the end-column conductivity detector. (A) Alignment of the electrode with the eluant hole in the capillary wall; (B) as in A but with electrical connectors; (C) horizontal cross section. The protective jacket is not shown. Measurements are made between the sensing electrode and ground electrode. (Reproduced with permission [149]).
7.10.3 Combined Conductivity and UV Detection The use of both UV and conductivity detection during the same analysis offers many advantages, some of which are obvious from Fig. 7-38. Firstly, individual detectors may register zones the other is unable to detect. Dual detection also enables the determination of mobilities [42, 147, 1481, a useful tool in quantitative conductimetric analysis using the internal standard method. The UV response may therefore be calibrated on an absolute basis [149].
7.10.4 Suppressed Conductivity Detection The most recent form of conductimeter has used tubular ion-exchange membranes to permit suppressed detection, much as in Ion Chromatography, using regenerant solutions. Different membranes are required for anionic and cationic analysis. For anion determination, sodium counter-ions are replaced with more mobile hydrogen ions, protonating the buffer. As the weak acid formed is less dissociated than the
7.10 Conductivity Detection
I
0
25 1
5
(A’
1
2
3
4 5 Time ( m i d
6
7
8
Fig. 7-38. Combined (A) UV (254 nm) and (B) conductimetric detection for the CZE separation of (1) K + ; (2) N a + ; (3) Li+; (4) cyclohexamine; (5) pyridoxamine; (6) H,O; (7) dansyl-isoleucine. The simultaneous recording of zones by both detectors can aid in the correct identification of peaks. BGE = 5 mM MES pH 6 (histidine). HT = 15 kV. (Reproduced with permission [149]).
salt, background conductivity is reduced. Cationic suppression replaces the strong acid in the BGE with a strong base, neutralising the weak base. Dasgupta and Bao’s [ 1501 SUCCESSconductimeter demonstrated the advantage of the presence of a water dip in their electropherograms. The dip half-width divided by migration time is proportional to the applied sample volume (Fig. 7-39). One drawback of SUCCESS detection was a noticeable loss of efficiency with early migrating peaks. The cause was attributed to the membrane itself, and not the conductimeter. Whether the effect was a result of the individual membrane, or due to the abrupt change in surface chemistry (silica to membrane) was not ascertained. Avdalovic et al. [151] demonstrated a range of applications using suppressed detection (Fig. 7-40). Suppression of a 2 mM borate buffer reduced background conductance from 300 W Scm-’ to 2 - 3 pS cm-’. Minimum detection limits for comM ( c10 ppb) without the use mon inorganic anions were in the range (2- 10)x of sample preconcentration. This is, to date, the most sensitive conductimeter produced for CZE. Conductivity detection has already proved its utility in isotachophoresis, enabling sensitive determination at relatively cheap cost. The almost universal detection of analyte zones by conductivity is of real advantage in CE, and thus conductimeters may be expected to appear for CZE on a commercial basis in time. A primary problem will be that in the design of a ‘user-friendly’ detector, the mass production of a reproducible apparatus may be difficult owing to the minimal capillary internal diameters employed in CZE. Once a solution is found it is likely that conductivity detection will become as widespread as UV detectors presently available.
252
7 Application of Capillary Electrophoresis for Environmental Analysis
c
PI
M
P2
Cl
U I
I
I
I
I
4
6
8
10
12
Time (rnin)
0
50 100 150 200 250 Area of water dip (AUI
300
Fig. 7-39. (A) Suppressor designs. (a) Wetted Nafion segment (N) into which the capillaries are forced; (b) radiation grafted membrane (M) inserted into PVC tubes and butt-joined to the capillaries; (c) suppressor (b) coupled to detector: (E I , E2) wire electrodes; (P3) PVC tube sleeve; ( G ) epoxy adhesive; (R) reservoir. (B) Electropherogram of four anions near the detection limits: 50 yg L-' of each ion except fluoride (10 pg L-I). (C) Plot of the relationship between area of the water dip: anaiyte peak area. It is quite clear that the relationship is linear. (Reproduced with permission [1501).
7.11 Mass Spectrometric Detection for CE In common with other separation techniques (eg, LC), the identity of the separated components cannot be easily determined using conventional detectors and corresponding migration times. Thus, as in chromatography, the on-line combination of CE with the mass spectrometer (MS) is clearly desirable. Niessen et al. [ 1521 have recently reviewed the current and possible future aspects for the coupling of CE with mass spectrometric detection.
7.11 Mass Spectrometric Detection for CE
253
Cation or anion
Detector electro
2 I
6
6
9
Fig. 7-40. (A) Schematic design of the suppressor. (B) Electropherogram of three halides, with partial resolution of isotopes. Peaks: (1) 37C1-; (2) 35C1-;(3) iodide; (4) 79Br- and “Br-. The inset shows 95% pure 37C1- and natural abundance chloride. (C) Electropherogram of 13 anions (10 WM each). Peaks (ppm): (1) bicarbonate; (2) chlorite (0.67); (3) fluoride (0.19); (4) phosphate (0.95); (5) chlorate (0.83); (6) perchlorate (0.99); (7) nitrite (0.62); (8) nitrate (0.46); (9) sulphate (0.96); (10) chloride (0.35); (1 1) iodide (1.27); (12) bromide (0.80); (13) chromate (1.16). (Reproduced with permission 11511).
254
7 Application of Capillary Electrophoresis for Environmental Analysis
7.11.1 The CE-MS Interface An interface is required between the electrophoretic separation method and the mass spectrometer in order to facillitate the formation of gas phase ions from the liquid phase emanating from the capillary terminus, which may then be detected by the mass spectrometer. Interfaces using a wide range of ionisation modes and mass spectrometer designs have been coupled in on-line CE - MS, however most applications have used electrospray ionisation (ESI) as the interface to produce gas phase ions for MS detection. Continuous flow fast atom bombardment (CF-FAB) [153] is the second most used interface; however the coupling is difficult to achieve. Two method of capillary arrangement in the CE- MS interface have been proposed: (1) positioning the CE capillary outlet so that the eluant enters the source directly from the tip; and (2) a coupling based on a 'liquid junction' [154], whereby the high-voltage is effectively de-coupled by a small (50 pm) gap between the analytical capillary and a transfer capillary entering the source. This method utilises gravity flow or pumps to drive the CE zones through the transfer column; this transfer inevitably introduced some dispersion of zones (Fig. 7-41). A stable electrical connection
(A) Original CE/MS interface Electrospray electrical lead
(6)Sheath flow (coaxial) interface CE buffer
I
SF,
Electrospray
.
\
.1
Vapour-deposited
* CE capillary
Teflon holder
Electrospray Stainless steel tubing
(C) Liquid junction interface Buffer reservoir (makeup solvent) n
n
SF,
death liquid
(D) Sheathless interface Electrical contact
Electrosorav
-- - - -- --
CE-buffer
NZ
Etched gold coated capillary
Fig. 7-41. Schematic illustration of the CE-MS interface designs. (A) The original design using a metallised capillary terminus. (B) The sheath flow (coaxial) interface used by most research groups. (C) The liquid junction coupling. (D) A new sheathless interface designed to optimise MS sensitivity. (Reproduced with permission [157]).
7.11 Mass Spectrometric Detection for CE
255
between the CE capillary and the ionisation mode is critical to ensure adequate ‘grounding’ of the driving potential and reproducible operation.
7.11.2 CZE - MS The first interface (Fig. 7-41) used a capillary with a silver coated tip for CZE- ESI - MS. The metallic terminus served both as the CZE ‘ground’ and the ESI source voltage, ie, the CZE terminus was held at about 3 - 6 kV, defining the electric field gradient along the CZE capillary. Problems were encountered as the deposited silver was soon removed and the emanating spray was somewhat unstable. Despite this Smith et al. [I 551 demonstrated that CZE-MS was viable. Smith et al. [156] next used a sheath-flow interface to overcome these problems (Fig. 7-41). The co-axial flow of a conducting liquid in the outer capillary ensures correct electrical contact is maintained by the flow of the sheath liquid, held at the ESI voltage, over the analytical capillary. This results in a more stable spray current, at the expense of increased chemical noise. Until recently, essentially the same sheath-flow interface design as that described by Smith et al. [156] has been used by the majority of research groups and instrument manufacturers. Attention is now returning back to a sheathless design [157] (Fig. 7-41) in an effort to optimise MS sensitivity. Excellent sensitivity has been achieved using < 20 pm ID capillaries with CZE-ESI -MS (detection limits approximately 5 attomoles [158]). The use of both liquid junction and coaxial interfaces for CZE-ESI-MS [159] have been compared. Several antibiotic classes used in the aquaculture industry and some marine toxins were used as model analytes (Fig. 7-42). CZE-MS and CZE- MS - MS determination enabled the confirmation of these substances in shellfish extracts at low ppm levels. In virtually all respects the co-axial approach offered a more robust and reproducible interface. Quantitation was also facilitated using the coaxial approach by flow injection of standards in the sheath flow. Garcia and Henion used gel-filled capillaries to examine aromatic sulphonates [ 1061 (Fig.7-43) using a liquid junction pneumatically assisted CZE - ESI - MS interface. Relatively high concentrations of urea used in the separation were effectively held up in the gel, unaffecting MS performance. In order to achieve satisfactory ion currents for MS detection the gel capillary had to be overloaded. The same workers also determined 8 model sulphonylurea herbicides using a 35 cm silica capillary with the same liquid junction interface [161] (Fig. 7-43). Although good sensitivities were achieved, CZE-MS could not confirm the presence of these herbicides at real environmental levels. Brumley [ 1621 investigated the use of CZE and CZE - CF- FAB - MS for the analysis of aromatic sulphonic acids in a leachate from a hazardous waste site. Eight acids were separated by CZE-UV (Fig. 7-44), within a time window that could be used for qualitative description of zones eluting in this period. Brumley found that large organic constituents led to decreased mobility, and thus shorter migration times as the EOF was counter to the anion mobility. Electron-withdrawing groups (eg, C1,
256
7 Application of Capillary Electrophoresis f o r Environmental Analysis Liquid-junction interface configuration CE-column
Nebulizer gas inlet Transfer capillary HT-lead
i
C E-c o Iumn
Make-up inlet for coaxial configuration (from syrunge pump via injector)
0
2
4
6 Time (mini
8
10
12
Fig. 7-42. (A) Schematic diagram of a fully-articulated ESI-MS interface showing the liquid junction and coaxial interfaces. (B) Coaxial CE-MS separation of four toxins. (C) Liquid junction CE-MS separation of the same toxins as in (B). Note that in (C) resolution is poor and the background chemical noise is much worse. (Reproduced with permission [I 591).
G
4
l It I
0 0.0
.
_
.
1.0
2.0
.
11
3.0 17
5.0
4.0 22
28
Time (mi111
6.0 34
8.0
7.0 40
45
Time (min)/Scan
(C)
(21 IM-HI-
Total selected ion current
100
00 - SOi
Benzene sulfonate Ill
@@ so; I-
so;
11) [M-H)m/z 157
= m/z 207
Naphthalene sulfonate I21 l-amino-4Naphthalene sulfonate (3)
04
0
10
20
30
40
50
Time lmin)
Fig. 7-43. Fast CE-MS. (A) The apparatus for rapid analysis. A , HT lead; B, Pressure inlet for injection and flushing of capillary; C, Battery power supply for movement of reservoirs; 0, Exit of CE capillary; E, transfer capillary; F, N, inlet for MS; G, liquid junction reservoir. H , HT for ESI; I, CE capillary (35 cmx0.075 mm), J, safety catch for disabling power supply. (B) CZE-MS analysis of eight sulphonylurea compounds. (1) bensulfuron methyl; (2) sulphometuron methyl; (3) tribenuron methyl; (4) nicosulphuron; (5) chlorimuron ethyl; (6) thifensulphuron methyl; (7) metsulphuron methyl; (8) chlorsulphuron. The inset is the UV trace. (C) Gel-CZE-MS of three aromatic sulphonates using the device in (A). (Reproduced with permission [160, 1611).
258
7 Application of Capillary Electrophoresis .for Environmental Analysis 4-chlorobenzenesulfonate
(A)
0.04
0.04
-
a l
u m c n L 0
n
0.02
0
II
D ,
.
,
.
10 Tirne'(rnin)
20
22.5
Scan number Fig. 7-44. CZE and CZE-CF-FAB-MS of Stringfellow leachate. (A) Electropherogram of the leachate using borate-boric acid BGE at pH 8.3. (B) Negative ion mass electropherogram of m / z 191 from the leachate obtained using the same buffer. (Reproduced with permission [162]).
NO,) and small organic groups (eg, methyl) resulted in faster mobilities and concurrently longer migration times. Both observations are in accordance with electrophoretic theory, whereby mobility is a function, amongst other factors, of the analytes shape, and charge to size (mass) ratio. The electropherogram of a leachate [I621 is shown in Fig. 7-44.The 4-chlorobenzenesulphonic acid peak was confirmed by comparison with CZE migration times; CE- FAB - MS, and CZE- CF- FAB - MS. The smaller peaks in the UV trace could not be detected by CZE-MS, and a loss of efficiency arose with CZE- MS.
7.11.3 ITP- MS A theme in CZE-MS has been that detection limits are much higher than one requires for real world solutions. This is not surprising considering the nanoliter sample loadings. In response Udseth et al. [I631 developed ITP-ESI-MS using their
259
7.11 Mass Spectrometric Detection for CE
R I E CITP-MS
Ml-zLdA-
44
45
Time (min)
45
46
47
48
49
50
51
Time ( m i d
Fig.7-45. A comparison of (A) ITP-MS and (B) CZE-MS in the analysis of four quaternary phosphonium ions. The ITP analysis clearly shows better resolution of the ions, and represents an almost ideal flat topped peak for MS detection. Further, the limits of detection using ITP-MS are at least two orders of magnitude better than those for CZE-MS. (Reproduced with permission [1631).
coaxial interface. Solutions M of quaternary phosphonium ions gave S/N ratios of around lo2, with detection limits of approximately lo-" M. The comparison of ITP- MS and CZE - MS is given in Fig. 7-45. One very useful attribute of MS detection for ITP is the simplicity of peak identification once data is acquired and the determination of mixed zones is more straight forward than in UV or conductivity detection. Furthermore, the use of 100 pm ID capillaries, 2 m in length allowed a very long electrokinetic injection (5 min at 23 kV for a lop9M sample) as, up to a point, maximum sample loading in ITP is defined by the volume of LE between the injection and detection points.
7.11.4 ITP- CZE -MS The coupling of ITP- CZE- MS also provides increased sensitivity and has been used in both the coupled column [164] and transient on-column [165] ITP-CZE mode. Transient coupling enables at least 100-fold improvement in sensitivity, whilst coupled column ITP- CZE provided a 200-fold gain over that of CZE - MS alone.
260
7 Application of Capillary Electrophoresis f o r Environmenfal Analysis
It is likely that whenever a complex matrix containing lower than ppm concentrations of the separand is to be analysed, on-line sample preconcentration will be required. CE - MS will undoubtedly become more popular and easier to interface during the next few years.
7.12 Summary and Conclusions Capillary electrophoresis is a relatively new technique for many people, but has developed rapidly. The very high efficiency of CE lends itself to the analysis of the components of complex matrices, and has thus received attention from some environmental scientists. The minimal solvent use in CE is also ideally suited for the environmental scientist, minimising costs and the production of waste solvent. This factor alone will interest many scientists who analyse hazardous samples routinely (eg, radioactive) . The commercial apparatus available on the market over the last few years has undoubtably brought CE to the attention of scientists, however the relatively inflexible design of most systems hinders progress in CE. Recently, modular systems have appeared and these are expected to widen the use of CE generally; just as was the case for LC. The combination of CE with very sensitive detection methods (eg, MS) provides a resolving power equal to capillary gas chromatography- mass spectrometry (GC-MS), and enables the analysis of a wide range of analytes not amenable to GC - MS. Thus, the commercial availability of hyphenated techniques and different detector systems with CE would enhance the utility of CE for many. Despite the efficiency of CE, the actual detection limits are often in the low ppm range, and thus well above required legal limits. On-line ITP preconcentration has been shown to provide a solution here. The lack of commercial two-dimensional CE apparatus poses a problem for the acceptance of CE in environmental analysis, although in time coupled column apparatus can be expected to appear on the market. Quantitation in CE is feasible, although the injection method will often dictate the accuracy and precision of the separation. In this respect, further research will be required to produce an injection system that provides a known volume under changing conditions. A range of normalisation procedures have been reported [166- 1691 to improve quantitation, and should be used. At the very least, the use of an internal standard is recommended. The prospects for CE in environmental analysis appear to be extremely promising. At present many organisations involved with environmental analysis are evaluating CE, and a method for the analysis of anions in aquatic samples by CZE has been submitted to the ASTM committee D-19 on Water and subcommittee D 19.05 on Inorganics in Water (USEPA committees) for review [170]. The method is applicable to a variety of aqueous samples, and demonstrates that CZE is a useful tool for the environmental scientist. ITP has long been used for environmental analysis, and provides a powerful preconcentration method for dilute analytes. The fundamental problem in ITP is the
References
261
method of detection, and in particular the data output format. The very narrow contiguous zones of ITP require very high-resolution detectors with a fast response time. Mass spectrometric detection may be the almost ideal detection method for ITP, particularly if a Time of Flight (TOF) MS were to be used. Kaniansky et d.[171] have recently demonstrated that the on-line column coupling of CITP-CZE can provide ultra-trace determination of paraquat and diquat in aqueous samples. Limits of detection were found to be mol/L using a 90 pL injection and detection at 3 10 nm. Adsorption losses were significant at these levels and were eliminated by spiking samples and cleaning the nalgene containers with diethylenetriamine. It is interesting to note that the CZE stage employed a 250 micron i. d. capillary, providing upto 545000 theoretical plates per meter for 25 pL sample of paraquat (10.6~ mol/L). CE certainly has a real future in the analysis of environmental media. The ability to resolve closely related compounds is excellent, and the possibilities of using CE for the analysis of the speciation of metallic compounds should interest many scientists. However, CE is still a relatively young technique with a very different separation mechanism to that of chromatography. It is probably this difference that many scientists find off-putting when first contemplating the technique. Many lengthy books have recently been published about CE in general, and more are being published every month covering a wide range of application areas. It is most unlikely that CE will now dissappear from the armoury of the analyst, and more interest is being increasingly shown in the analysis of samples other than biological. It is hoped that this chapter in some way contributes to the increased use of CE in environmental laboratories, not just for the analysis of trace pollutants but also as a method for studying interactions in the environment.
References [I] Herb, R., Dulffer, T., Hermrann, H., Kobold, U., Chromatographia 1990, 30 (11/12 Dec), 675-685. [2] Everaerts, F. M., Verheggen, Th. P. E. M., J Chromatogr 1974, 91, 837-851. [3] Jorgenson, J.W., Lukacs, K., J Chromatogr 1981, 218, 209-216. [4] Jorgenson, J. W., Lukacs, K., Anal Chem 1981, 53, 1298- 1302. [5] Jones, W.R., Jandick, P., J Chromatogr 1992, 608, 385-393. [6] Cai, J., Rassi, Z.EL, JLiq Chrom 1992, 15 (6+7), 1193-1200. [7] Phosichal, J., Deml, M., Gebauer, P., Bocek, P., J Chromatogr 1989, 470, 43-55. [8] Sustacek, V., Foret, F., Bocek, P., J Chromatogr 1989, 480, 271 -276. [9] Bocek, P., Deml, M., Poshical, J., Suder, J., J Chromatogr 1989, 470, 309-312. [ l o ] Bocek, P., Deml, M., Pospichal, J., J Chromatogr 1990, 500, 673-680. [Ill Sudor, J., Pospichal, J., Deml, M., Bocek, P., J Chromatogr 1991, 545, 331-336. [I21 Foret, F., Fanali, S., Bocek, P., J Chromatogr 1990, 516, 219-222. [I31 Tsuda, T., Anal Chem 1992, 64, 386-390. [I41 Chang, H., Yeung, E.S., J Chromatogr 1992, 608, 65-72. [I51 Purghart, V., Games, D.E., J Chromatogr 1992, 605, 139-142. [I61 Whang, C., Yeung, E.S., Anal Chem 1992, 64, 502-506.
262
7 Application of Capillary Electrophoresis f o r Environmental Analysis
[I71 Foret, F., Bocek, P., Capillary Electrophoresis, in: Advances in Electrophoresis Vol. 3: Radola, B. J. e f al. (eds.) Weinheim: VCH, 1989. [18] Knox, J. H., Grant, I.H., Chromatographia 1987, 24, 135- 143. [ 191 Knox, J. H., Chromatographia 1988, 26, 329 - 337. [20] Snopek, J., Jelinek, I., Smolkova-Keulemansova, E., J Chromatogr 1988, 452, 571 -590. [21] Powell, A.C., Sepaniak, M. J., J Microcolumn Separations, 1990, 2, 278-284. [22] Gariel, P., Chromatographia 1990, 20 (3/4 August), 195-200. [23] Janini, GM., Issaq, H. J., J Liq Chrom 1992, 15(6+ 7), 927 -960. [24] Terabe, S., Otsuka, K., Ando, T., Anal Chem 1985, 57, 834-841. [25] Ong, C. P., Lee, H. K., Li, S. F. Y., J Chromafogr 1991, 542, 473 -482. [26] Cai, J., Rassi, Z., J Chromatogr 1992, 608, 3 1-45. [27] Irnasaka, T., Nishitani, K., Ishibashi, N., Anal Chim Acta 1992, 256, 3-7. [28] Everaerts, F. M., Geurts, M., Mikkers, F. E. P., Verheggen, Th. P. E. M., J Chromatogr 1976, 119, 129- 155. [29] Arlinger, L., J Chromatogr 1974, 91, 785-794. [30] Nee, T. W., J Chromatogr 1974, 93, 7- 15. [31] Bocek, P., Dernyl, M., Janak, J., J Chromatogr 1974, 91, 829-835. [32] Nee, T. W., J Chromatogr 1975, 105, 231 -249. [33] Moore, G. T., J Chromatogr 1975, 106, 1 - 16. [34] Svoboda, M., Vacik, J., J Chromatogr 1976, 119, 539-547. (351 Bocek, P., Dernl, M., Kaplanova, B., Janak, J., J Chromatogr 1978, 160, 1-9. [36] Delmotte, P, J Chromatogr 1979, 165, 87- 101. [37] Mikkers, F.E.P., Everaerts, EM., Peek, J.A., J Chromatogr 1979, 168, 293-315, 317-332. [38] Gebauer, P, Bocek, P., J Chromatogr 1983, 267, 49-65. [39] Bocek, P., Foret, F., J Chromatogr 1984, 313, 189-222. [40] Krivakova, L, Foret, F., Gebauer, P., Bocek, P., J Chromatogr 1987, 390, 3 - 16. [41] Hirokawa, T., Nakahara, K., Kiso, Y., J Chromatogr 1989, 4700, 21 -41. [42] Everaerts, F. M., Beckers, J. L., Verheggen, Th. P. E. M., Isotachophoresis: Theory, Instrumentation and Applications. Journal of Chromatography Library, Vol. 6, Amsterdam: Elsevier, 1976. [43] Bocek, P., Deml, M., Gebauer, P., Dolnik, V., Analytical ZTP, Weinheim: VCH, 1988. [44] Kaniansky, D., Zelensky, I., Zelenska, V., J Chromatogr 1987, 390, 11 1 - 120. [45] Everaerts, F. M., Mikkers, F. E. P., Verheggen, Th. P. E. M., J Chromatogr 1985,320, 263 -268. [46] Hirokawa, T., Yokota, Y., Kiso, Y., J Chromatogr 1991, 545, 267-281. [47] Gebauer, P., Bocek, P., J Chromatogr 1985, 320, 49-65. [48] Dolnik, V., Deml, M., Gebauer, P., Bocek, P., J Chromatogr 1991, 545, 249-266. [49] Hirokawa, T., Nishino, M., Aoki, N., Kiso, Y., Sawamoto, Y., Yagi, T., Akiyama, Y., J Chromatogr 1983, 271, D1 -D106. [50] Mosher, R.A., Thormann, W., Bier, M., J Chromatogr 1985, 320, 23-32. [51] Hirokawa, T., Nakahara, K., Kiso, Y, J Chromatogr 1987, 408, 27-41. [52] Mosset, D., Gariel, P., Desbarres, J., Rosset, R., J Chromatogr 1987, 390, 69-86. [53] Dose, E., Guiochon, G. Anal Chem 1991, 63, 1063- 1072. [54] Hirokawa, T., Kiso, Y., Anal Sci 1992, 8 (DEC), 737-748. [55] Caslavska, J., Kaufmann, T., Gebauer, P., Thormann, W., J Chromatogr 1993, 638, 205 - 21 4. [56] Thormann, W., J Chromatogr 1990, 516, 211 -217. [57] Gebauer, P., Thormann, W., J Chrornatogr 1991, 545, 299-305. [58] Hjerten, S., Kiessling-Johansson, M., J Chrornatogr 1991, 550, 81 1 - 822. [59] Gebauer, P., Thormann, W., J Chromatogr 1991, 558, 423 -429. [60] Reijenga, J. C., Aben, G. V.A., Verheggen, Th. P. E. M., Everaerts, F. M., J Chromatogr 1983, 260, 241 -254. [61] Beckers, J.L., Everaerts, EM., Ackermans, M.T., J Chromatogr 1991, 537, 429-442. [62] Ackermans, M.T., Everaerts, EM., Beckers, J. L., J Chromatogr 1991, 545, 283-297. [63] Lee, C.S., Blanchard, W.C., Wu, C., Anal Chem 1990, 62, 1550- 1552.
References
263
Hayes, M. A., Ewing, A. G., Anal Chem 1992, 64, 512-516. Chang, H., Yeung, E.S., Anal Chem 1993, 65, 650-652. Hirokawa, T., Watanabe, K., Yokota, Y., Kiso, Y., J Chromatogr 1993, 633, 251 -259. Delinger, S. L., Davis, J. M., Anal Chem 1992, 64, 1947- 1959. Grushka, E., McCormick, R.M., J Chromatogr 1989, 471, 421 -428. Deml, M., Foret, F., Bocek, P., J Chromatogr 1985, 320, 159-165. Verheggen, Th. P. E. M., Beckers, J. L., Everaerts, F. M., J Chromatogr 1988, 452, 615 -622. Tsuda, T., Mizuno, T., Akiyama, J., Anal Chem 1987, 59, 799-800. Jarofke, R., J Chromatogr 1987, 390, 161 - 167. McDonnel, T., Pawliszyn, J., J Chromatogr 1991, 559, 489-497. Bao, L., Dasgupta, P.K., Anal Chem 1987, 64, 991 -996. Wallingford, R. A., Ewing, A. G., Anal Chem 1987, 59, 678-681. Ornstein, L., Ann NY Acad Sci 1964, 121, 321. Davies, D., Ann N Y Acad Sci 1964, 121, 404. Mikkers, F. E. P., Verheggen, Th. P. E. M., Everaerts, F. M., J Chromatogr 1979, 169, 21 -38. CS ITP Analyzer. Institute of Radioecology and Applied Nuclear Techniques, Labeca-Villa, Spisska Nova Ves (Slovakia) 1982. [80] Foret, F., Sustacek, V., Bocek, P., J Microcolumn Separations 1990, 2, 229-233. [81] Kanniansky, D., Marak, J., J Chrornatogr 1990, 498, 191 -204. [82] Beckers, J.L., Everaerts, EM., J Chromatogr 1990, 508, 3-17, 19-26. [83] Beckers, J.L., J Chromatogr 1993, 641, 363-373. [84] Schwer, C., Gas, B., Lottspeich, F., Kenndler, E., Anal Chem 1993, 65, 2108-2115. 1851 Vinther, A., Soeberg, H., J Chromatogr 1991, 559, 3-26. [86] Gebauer, P., Thormann, M., Bocek, P., J Chromatogr 1992, 608, 47-57. [87] Beckers, J.L., Ackermans, M.T., J Chromatogr 1993, 629, 371-378. [88] Foret, F., Szoko, E., Karger, B.L., J Chromatogr 1992, 608, 3-12. [89] Krivankova, L., Foret, F., Bocek, P., J Chromatogr 1991, 545, 307-313. [90] Stegehuis, D.S., Tjaden, U.R., Van der Greef, J., J Chromatogr 1991, 538, 393-402. [91] Stegehuis, D.S., Tjaden, U.R., Van der Greef, J., J Chromatogr 1992, 591, 341 -349. [92] Krivankova, L., Gebauer, P., Thorrnann, W., Mosher, R.A., Bocek, P., J Chromafogr 1993, 638, 119- 135. [93] Kaniansky, D., Marak, J., Madajova, V., Simunicova, E., JChromatogr 1993, 638, 137- 146. [94] Mikkers, F. M., Everaerts, EM., Verheggen, Th. P.E.M., J Chromatogr 1979, 169, 11 -20. [95] Bondoux, G., Jandik, P., Jones, W. R., J Chromatogr 1992, 602, 79-88. [96] Jandik, P., Jones, W. R., J Chromatogr 1991, 546, 431 -443. [97] Burgi, D.S., Chien, R., Anal Chem 1991, 63, 2042-2047. [98] Chien, R., Burgi, D.S., Anal Chem 1992, 64, 1046- 1050. [99] Chien, R., Helmer, J.C., Anal Chem 1991, 63, 1354-1361. [IOO] Chien, R., Anal Chem 1991, 63, 2866-2869. [loll Chien, R., Burgi, D.S., J Chromatogr 1991, 559, 153-161. [lo21 Chien, R., Burgi, D.S., Anal Chem 1992, 64, 489A-496A. [lo31 Reinhood, N. J., Tjaden, U. R., Van der Greef, J., J Chromatogr 1993; 641, 155- 162. [lo41 Cai, J., Rassi EL, Z., J L i q Chrom 1992, 15 (6+7) 1179-1192. [I051 Scwartz, M.E., Merion, M., J Chromatogr 1993, 632, 209-213. [lo61 Debets, A. J. J., Mazereeuw, M., Voogt, W.H., Van Iperen, D. J., Lingeman, H., Hupe, K.P., Brinkman, U. A. Th., J Chromatogr 1992, 608, 151 - 158. [lo71 Yarnamoto, H., Manabe, T.,Okuyama, T., J Chromatogr 1990, 515, 659-666. [lo81 Hutta, M., Marak, J., Kaniansky, D., J Chromatogr 1990, 509, 271 -282. [lo91 Wallingford, R.A., Ewing, A.G., Anal Chem 1988, 60, 1972-1975. [110] Pentoney, S., Zare, R.N., Quint, J., Anal Chem 1989, 61, 1642-1647. [ill] De Bokx, P.K., Gillissen, E. E. A., Van de Weijer, P., Bekkers, M. H. J., Van Bommel, C. H. M., J Chromatogr 1992, 598, 115 - 121. [I 121 Taylor, LA., Yeung, E. S., J Chromatogr 1991, 550, 831 -837. [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79]
264 [113] [I 141 [115] [116]
7 Application of Capillary Electrophoresis for Environmental Analysis
Chen, C., Morris, MD., J Chromatogr 1991, 540, 355-363. Huang, X., Zare, R. N., Sloss, S., Ewing, A. G., Anal Chem 1991, 63, 189- 192. Moseley, M., Deterding, L., Tomer, K. B., Jorgenson, J., J Chromatogr 1989, 480, 197 - 209. Bruin, G. J.M., Stegeman, G., Van Asten, S. C., Xu, X., Kraak, J. C., Poppe, H., J Chromatogr 1991, 539, 163-181. [117] Chervet, J.P., Van Soest, R.E.J., Ursem, M., J Chromatogr 1991, 543, 439-449. [118] Grant, I.H., Steuer, W., J Microcolumn Separations 1990, 2, 74-79. [119] Wang, T., Aiken, J. H., Huie, C. W., Hartwick, R.A., Anal Chem 1990, 63, 1372-1376. [I201 Tsuda, T., Kobayashi, Y.,J Chromatogr 1990, 515, 357-361. [121] Tsuda, T., Sweedler, J.V., Zare, R.N., Anal Chem 1990, 62, 2149-2152. [I221 Zare, R. N., Tsuda, T., J Chromatogr 1991, 559, 103- 1 1 0. [123] Gaitonde, C.D., Pathak, P.V., J Chromatogr 1990, 514, 389-393. [I241 Lu, W., Cassidy, R.M., Baranski, A. S., J Chromatogr 1993, 640, 433 -440. [125] Jandick, P., Bonn, G., CE of Small Molecules and Ions, Weinheim: VCH, 2993. [126] Reijenga, J.C., Lemmens, A. A. G., Verheggen, Th. P. E. M., Everaerts, F. M., J Chromatogr 1985, 320, 67-73. [127] Reijenda, J.C., Verheggen, Th.P.E.M., Everaerts, F.M., J Chromatogr 1984, 283, 99- 1 1 1 . [I281 Wang, T. W., Hartwick, R. A., J Chromatogr 1992, 607, 119- 125. [129] Nielen, M. W. F., J Chromatogr 1991, 588, 321 - 326. I1301 Sustacek, V., Foret, F., Bocek, P., J Chromatogr 1991, 545, 239-248. [I311 Ackermans, M. T., Everaerts, EM., Beckers, J.L, J Chrornatogr 1991, 549, 345 -355. [132] Bruin, G. J. M., Van Asten, A. C., Xu, X., Poppe, H., J Chromatogr 1992, 608, 97- 107. [I331 Ma, Y.,Zhang, R., J Chromatogr 1992, 625, 341 -348. [134] Wang, T., Hartwick, R. A., J Chromatogr 1992, 5689, 307 - 3 13. [135] Foret, F., Fanali, S., Ossicini, L, Bocek, P., J Chromatogr 1989, 470, 299-308. [136] Jackson, P.E., Haddad, P. R., J Chromatogr 1993, 640, 481 -487. [I371 Weston, A., Brown, P. R., Jandick, P., Jones, W.R., Heckenberg, A. L., J Chromatogr 1992, 593, 289-295. [138] Harrold, M.P., Wojtusik, M. J., Riviello, J., Henson, P., JChromatogr 1993, 640, 463-471. [I391 Buchberger, W., Haddad, P.R., J Chromatogr 1992, 608, 59-64. [140] Chen, M., Cassidy, R.M., J Chromatogr 1993, 640, 425-431. [141] Shi, Y.,Fritz, J.S., J Chromatogr 1993, 640, 473-479. [142] Gas, B., Demjanenko, M., Vacik, J., J Chromatogr 1980, 192, 253-257. [143] Vacik, J., Zuska, J., Muselasova, I., J Chromatogr 1985, 320, 233-240. [I441 Mikkers, F. E.P., Thesis. Eindhoven University, 1980. [145] Foret, F., Deml, M., Kahle, V., Bocek, P., Electrophoresis 1986, 7, 430. (1461 Huang, X., Pang, T. K. J., Gordon, M. J., Zare, R. N., Anal Chem 1987, 59, 2747-2749. I1471 Huang, X., Gordon, M. J., Zare, R. N., J Chromatogr 1989, 480, 285-288. [148] Huang, X., Luckey, J.A., Gordon, M. J., Zare, R.N., Anal Chem 1989, 61, 766-770. [I491 Zare, R.N., Huang, X., Anal Chem 1991, 63,2193-2196. [LSO] Dasgupta, P.K., Boa, L., Anal Chem 1993, 65, 1003-1011. [151] Avdalovic, N., Pohl, C.A., Rocklin, R.D., Stillian, J.R., Anal Chem 1993, 65, 1470-1475. [I521 Niessen, W.M.A., Tjaden, U.R., Van der Greef, J., J Chromatogr 1993, 636, 3 - 19. [153] Caprioli, R.M. (ed.) Continuous Flow Fast Atom Bombardment MS, New York: Wiley, 1990. [154] Lee, E.D., Muck, W., Henion, J.D., Covey, T. R., Biom, Env Mass Spec 1989, 18, 844-850. (1551 Olivares, J.A., Nguyen, N.T., Yonker, C.R., Smith, R.D., Anal Chem 1987, 59, 1230- 1234. [156] Smith, R.D., Baringa, C., Udseth, H. R., Anal Chem 1988, 60, 1948- 1950. [157] Smith R.D., Wahl, J.H., Goodlet, D.R., Hofstadler, S.A., Anal Chem 1993, 65, 574A- 584A. [I581 Wahl, J.H., Goodlet, D.R., Udseth, H.R., Smith, R.D., Anal Chem 1992, 64, 3194-3196. [I591 Pleasance, S., Thiobault, P., Kelly, J., J Chromatogr 1992, 591, 325-339. [160] Henion, J.D., Garcia, F., Anal Chem 1992, 64, 985-990.
References Garcia, F., Henion, J. D., J Chromatogr 1992, 606, 237 -247. Brumley, W. C., J Chromatogr 1992, 603, 261 -272. Udseth, H. R., Loo, J.A., Smith, R.D., Anal Chem 1989, 61, 228-232. Reinhoud, N. J., Tinke, A. P., Tjaden, U. R., Niessen, W. M.A., Van der Greef, J., J Chromatogr 1992, 627, 263-271. Thompson, T. J., Foret, F., Vouros, P., Karger, B. L., Anal Chem 1993, 65, 900-906. Guiochon, G., Anal Chem 1991, 63, 1154- 1160. Beckers, J. L., Everaerts, F. M., Ackermans, M.T., J Chromatogr 1991, 537, 407-412. Ackermans, M. T., Everaerts, F. M., Beckers, J. L., J Chromatogr 1991, 549, 345 -350. Yeung, E. S., Lee, T.T., Anal Chem 1991, 63, 2842-2853. Romano, J. P., Kroll, J., J Chromatogr 1993, 640, 403-412. Kaniansky, D., Ivanyl, F., Onuska, F.I., Anal Chem 1994, 66, 1817-1824.
265
This Page Intentionally Left Blank
8 Advances in Flow Analysis Instrumentation and its Application in Environmental Analysis Richard J. Berman
8.1 Introduction Flow analysis techniques have made a significant contribution to improving analytical procedures and laboratory efficiency for over 35 years. The principal driving force for the widespread use and acceptance of flow analysis has been the universal desire to automate manual wet chemical analysis methods. Once a manual procedure is automated, improvements in sample throughput and reproducibility of results are gained. At the same time, it is normally found that automation results in decreases in reagent and sample use, human exposure to hazardous chemicals, volume of hazardous waste generated, and cost per sample analyzed. In addition, the compatibility of flow analysis instrumentation with laboratory quality assurance procedures has made flow analyzers valuable workhorses for laboratories striving to create and maintain economical quality control programs. Although flow analysis has traditionally been associated with large-scale multichannel laboratory analyzers, there has also been much interest in the development of on-site and on-line process analyzers. In the future, such analyzers will play a major role in allowing for real-time monitoring for pollution prevention and for on-site field testing. Flow analysis is generally accepted as covering the techniques known as gas-segmented continuous-flow analysis (CFA) [l] and flow injection analysis (FIA) [2]. Both techniques share several basic principles which are illustrated in Fig. 8-1 : 1. Sample introduction into a carrier stream. 2. Small-bore conduits. 3. Sample transport to a flow-through detector, often with additional reagent addition@) and/or treatment step(s) prior to arriving at the detector. 4. Detector response is related to the concentration of analyte tested. 5. Analyte concentration in an unknown sample is calculated using a calibration curve created by running a series of standard solutions of analyte through the flow analyzer.
Over the years, there have been thousands of flow analysis applications developed for measuring analytes in agricultural, biochemical, clinical, environmental, food, industrial, and pharmaceutical samples. In environmental analysis, the greatest use
268
8 Advances in Flow Analysis - Instrumentation and its Application Sample
.Air-----
(CFA only)
Reagent/
1
T
"
Analytical
I
To waste
Fig. 8-1. Schematic diagram of a flow analysis system.
of flow analyzers continues to be in routine analysis of water quality parameters such as nitrate, orthophosphate, and ammonium. Analyzers are commercially available from several vendors who typically offer a wide variety of applications for drinking water, waste water, sea water, surface water, ground water, soil and plant tissue analysis. In the realm of the routine use of flow analyzers, factors such as number of channels, sample throughput, detection limit, dynamic range, ease-of-use, and changeover time between chemistries are important issues. In addition, the practical consideration of the cost savings gained by automating is often at the heart of the decision to purchase a flow analyzers for routine testing. At other end of the flow analysis spectrum is both the tremendous body of literature published on CFA [3 - 81 and FIA [2,9, 101, and current research in the development of new instrumentation, techniques, and applications [l I]. Over the past decade, many concepts have been demonstrated and to a certain extent made available commercially. Examples of advances include in-line preconcentration [ 121, multi-component analysis [ 131, miniaturization [14], and hyphenation with various analytical detectors such as atomic spectrometers [15, 161. While most routine and experimental flow analysis techniques can be classified as CFA or FIA, there have also been techniques developed which do not conform to the traditional definitions. For example, sequential injection analysis (SIA) [ 17 - 191 is a technique where sample is introduced, mixed with reagents, and even calibrated in a novel way using a single pumping channel and a selector valve. Many other hybrid procedures have been developed, such as combined CFA with FIA [20-221 and combined chromatography with FIA [23 - 251 or CFA [26].The hybridization of analytical techniques and advances in instrumentation currently occurring often makes it difficult to define clearly a procedure as belonging exclusively to either CFA or FIA. It has also created a situation where the traditional arguments concerning the advantages of one technique over the other are becoming less clear as technology improves. Flow analysis is now evolving to where the complementary nature of the various injection and flow analysis techniques are creating a situation where combinations of these techniques are being made to create solutions to analytical problems. Since the overall scope of flow analysis is very large, this chapter will be limited to a review of basic flow analysis principles, recent advances in instrumentation and environmental application development, and quality assurance aspects of flow analysis.
8.2 Flow Analysis Review
269
8.2 Flow Analysis Review 8.2.1 Gas-Segmented Continuous-Flow Analysis (CFA) The first two decades of flow analysis were dominated by a technique known as gassegmented continuous-flow analysis (CFA). The procedure was invented by Skeggs [I] and successfully commercialized by Technicon Corporation. Initially, Technicon's AutoAnalyzerTMwas primarily used for clinical applications. Later on, industrial and environmental applications were developed, many of which have become approved by industrial and governmental regulatory authorities such as the US Environmental Protection Agency (US EPA), American Society for Testing and Materials (ASTM), and Deutsches Institut fur Normung (DIN). Extensive development of CFA instrumentation and applications occurred through the 1960s and 1970s as can be attested to by the nearly 8000 papers concerning CFA published by 1975 [27]. However, since the introduction of FIA in the mid-1970s there has been very little published on CFA while there has been a dramatic rise in FIA literature. One explanation for the shift in development effort may be due to the fact that by the mid-1970s CFA had become a mature technology with thousands of applications already developed. However, the versatility of CFA can be seen even today by the fact that it is still commonly used in laboratories around the world. The basic components of a gas-segmented continous-flow analyzer are schematically shown in Fig. 8-2. In general, sample, gas, and reagents are pumped into an analytical cartridge via pumptubes in a perstaltic pump. The analytical cartridge is where the reagents, segmenting gas, and sample are added and processed according to the automated chemical analysis procedure. In the simplest form, the cartridge
Bubble gating A:
--
Reagent Waste
s Fig. 8-2. Schematic diagram of a gas-segmented continuous-flow analysis system. ?ivo options for handling bubbles in detection are shown: electronic bubble gate or debubbler.
270
8 Advances in Flow Analysis - Instrumentation and its Application
consists of an assortment of mixing tees and coils made of either glass or polymeric tubing. In order for the gas segments to flow smoothly through the conduits, it is necessary to add a small amount of surfactant to one of the reagents so the inner conduit walls are wetted. Depending on the application being automated, the cartridge could also include components such as heat baths, separation modules for dialysis, gas diffusion, or liquid-liquid phase separation, in-line columns for sample treatment (ion-exchange, preconcentration, oxidationheduction reactors), UV digestion coils, and on-line distillation heads. The components of the original AutoAnalyzerTMhad inner diameters (ID) on the order of 3 mm. The second generation design, the AutoAnalyzer IITM, had 2 mm inner diameters. Third generation hardware with 1 mm ID components was not available for general use until the RFA-300 (Perstorp Analytical Environmental) was introduced in 1984 and the TRAACS-800 (Bran+Luebbe) in 1986. The reduction in dimension resulted in reduced reagent use and increased sample throughput [28]. Depending on the application, the carrier can be either the sample flow from the sampler, a diluent, or a reagent. The gas is added either continuously or emitted in short bursts using an air bar or solenoid valve to phase bubble addition with pump pulsations. Phasing was introduced to help minimize proportioning errors that result in increased noise at the steady state signal which, in turn, result in poor precision. Air is most commonly used as the gas for segmentation, although other gases are used for chemistries which are sensitive to interferences contained in air, such as the use of N2 in the analysis of nitrate. The purpose of the addition of a gas is to segment the flow thereby reducing sample dispersion (longitudinal diffusion) as it travels through the reaction coils. The air segments also act to cause turbulent flow within each liquid segment, especially when flowing through mixing coils with a small radius of curvature. Sample is introduced from an autosampler by the alternate aspiration of wash solution and sample solution. In this manner, alternate zones of sample and wash are merged with the segmented carrier stream in a tee fitting. One common misconception of CFA is that each individual liquid segment between two gas bubbles contains a single sample. Rather, as the sample/wash flow from the sampler reaches the segmented carrier, it is added over the course of many individual carrier segments. Within the liquid segments there are three main zones, each of which contributes to the observed peak response at the detector. On the front end there are several liquid segments with heterogeneous compositions of sample to reagents. This is caused by sample dispersion into the wash solution in the tubing from the sampler or from liquid segments which had both sample and wash added to them during the transition of zones. The same is true on the back end of the sample zone, where now the issue of dispersion is more pronounced and wash out from the analytical cartridge is highly influential on the peak response return to baseline. The initial goal of CFA was to create a wide zone in the center segments which had a homogeneous composition of sample to reagent volumes. This creates a situation where the reagent and sample composition in each of these segments is identical so that by the time the flow reaches the detector, there is a steady state response over a period of time where the peak response is measured. The steady-state response simply represents a situation when there is enough sample introduced so that dispersion effects on the front and back
8.2 Flow Analysis Review
271
ends of the sample zone are small compared with a central zone where there are multiple segments of homogeneous mixtures of sample, reagents, and product. There are several reasons why it was believed that a steady state response was essential for reproducible and accurate analyses. First, there was a perception that CFA was simply an extension of beaker chemistry and and that it was necessary to have a homogeneous reaction mixture to have an accurate analysis. Another reason a wide steadystate peak response was necessary was because of the crude peak finding procedures used on the 1960s and 1970s era Technicon AutoAnalyzerTMsystems. Chart recorders were used to record data and since the recorder was able to mark a peak once per minute and since the error in pumping was severe, it was necessary to have a broad target zone at steady-state where the recorder marked the peak maximum. In modern CFA instrumentation, the broad steady-state peaks are not necessary because data acquisition software programs include sophisticated peak finding algorithms. This has led to increased sample throughout in CFA and peak shapes that appear similar to FIA peak responses. The understanding of the meaning of steadystate has been another misconception of CFA. In CFA, the reaction does not need to go to completion in order for the analysis to be made at the detector. In fact, reactions may be less than 50% complete by the time they reach the detector, but that is acceptable since each sample introduced into the CFA system should be treated in the same way with respect to mixing and residence time. In this way, several concepts introduced and proven by FIA, such as controlled dispersion and the reproducibility of residence times in flow manifolds are also important considerations in CFA. When sample processing in the analytical cartridge is complete the flow reaches a detector where either the gas bubble is removed with a debubbler or it is allowed to pass through the flowcell with the bubbles intact. In the case of a photometric detector, gas bubbles in the flowcell result in a response to the refractive index variation on going from liquid to air. Several instruments have been designed which use an electronic bubble gate to filter out the periodic appearance of gas bubbles in the flowcell. The rationale behind bubble gating is well founded since it is well known that debubblers act as sources of peak broadening and can therefore reduce sample throughout by requiring increased wash times. Operationally, however, it has been found that analyzers without phasing and bubblegating are easier to operate and maintain even though those features have been shown to improve reproducibility. Nevertheless, analyzers with continuous addition of air and debubblers for gas bubble removal have been proven to give good reproducibility and sample throughputs. An example of a common CFA method is the USEPA approved automated procedure for the measurement of orthophosphate in water [29]. The schematic flow diagram is shown in Fig. 8-3 and typical performance parameters for the method are listed in Table 8-1. Sample is added to an air-segmented carrier stream which contains surfactant. A mixed reagent is then added where orthophosphate reacts with molybdenum (VI) and antimony (111) under acidic conditions to form an antimonyphosphomolybdate complex. This complex is subsequently reduced with ascorbic acid to form a blue-colored complex. The reaction mixture passes through a heating coil thermostated at 37 "C to assist in color development. The reaction mixture then flows to a flow through spectrophotometric detector where the absorbance is measured at 660 nm.
272
8 Advances in Flow Analysis - Instrumentation and its Application
Debubble to waste
37'c
.
Heat bath'
Detector
J
I
Fig. 8-3. Schematic flow diagram for orthophosphate by air-segmented continuous-flow analysis.
Table 8-1. Performance for orthophosphate method by CFA. ~~
~~
Range (pg P per liter)
10- 1000
Analysis rate (samples per hour) Sample time (s) Wash time (s) Detection wavelength (nm) Method detection limit (pg P per liter) Precision at 200 pg P per liter (VoRSD) Precision at 800 pg per liter (VoRSD) Precision at 1 pg per liter (VoRSD) Precision at 15 pg per liter (VoRSD) Sample to sample carryover (To) Color reagent consumption (mL per test) Sample consumption (mL per test) Result for check calibrant (pg P per liter) Check calibrant value (pg P per liter) Correlation coefficient (rxy)
80 20 25 660
-
1 0.71 0.29
0.38 0.41 0.53 780 80 1 0.9999
0.25 - 20 55
30 35 660 0.1 1
-
1.05 1.83 0.63 0.27 0.92 12.3 12.7 0.9996
Courtesy of Perstorp Analytical Environmental.
8.2.2 Flow Injection Analysis (FIA) Since its introduction in 1975 by Ruzicka and Hansen [30], flow injection analysis (FIA) has proven to be a versatile technique in flow analysis. Over 3000 research papers and several monographs [2, 10, 311 have been published on FIA in automated analytical chemistry. The book Flow Injection Analysis [2] by Ruzicka and Hansen is an excellent resource for detail on the theory of FIA and includes an extensive bibliography of FIA applications and techniques. There have been several recent reviews of FIA [32-381 and of applications in environmental analysis [I 1,391.
8.2 Ffow Analysis Review
Carrier
273
Waste Detector Sample loop
Reagent U
Fig. 8-4. Schematic flow diagram of a flow injection analysis system.
A schematic diagram of the basic components of a flow injection analysis system is shown in Fig. 8-4.In the simplest form of FIA, the sample is injected into a nonsegmented carrierheagent stream and transported to a detector. The basic principles of FIA are sample injection, controlled sample dispersion, and reproducible timing in the system. A discrete volume of sample is first loaded into an injection loop and then the valve is switched to the inject position. The carrier stream passes through the sample loop forcing the sample into the analytical manifold. As in CFA, there are a variety of sample treatment steps that can occur before the sample zone reaches the detector. As the sample is transported to the detector it is allowed to disperse into the carrierheagent stream creating a gradient in composition of sample, reagent, and product over the sample zone. Reagents are often added in mixing tees so as to create an even distribution of reagent over the entire sample zone. Since the process of controlled dispersion is entirely reproducible from one sample injection Table 8-2. Techniques and detectors used in FIA.
Techniques
Detectors
Transport Dilution Derivatization Stopped-flow Solvent extraction Gas diffusion Dialysis Preconcentration Digestion Ion-exchange Reaction rate measurement Precipitation Solid-phase extraction
Atomic spectrometry ( U S , ICP-AES) Molecular spectrometry (UV, Visible, FTIR) Amperometry Voltammetry Fluorescence Chemiluminescence Mass spectrometry Conductivity Potentiometry (ISE)
274
8 Advances in Flow Analysis - Instrumentation and its Application
Fig. 8-5. Schematic flow diagram for the method for nitrate+nitrite by flow injection analysis. Nitrite alone is determined by removing the cadmium reduction column.
to the next, it is possible to use the detector response to calibrate with known standard solutions of analyte. The introduction of the use of injection valves and controllable pumps has resulted in a wide variety of techniques to be developed with FIA. Table 8-2 lists many of the techniques and detectors now used with FIA systems. An example of a FIA application is the analysis of nitrate in water (Fig. 8-5). Sample is injected into a buffered carrier stream and reduced to nitrite by cadmium metal packed in a column 1401.A mixed reagent is then added to the flowing stream where nitrite is diazotized with sulfanilamide and coupled with N-(I-naphthy1)ethylenediamine to form an azo dye that is detected spectrophotometrically at 543 nm in an absorbance detector.
8.2.3 Sequential Injection Analysis (SIA) Sequential injection analysis (SIA) is a new flow analysis procedure that was developed with the intention of creating a more rugged flow injection system for chemical sensing and continuous process monitoring using a single piston pump [16- 191. In sequential injection analysis (Fig. 8-6), a selector valve is at the heart of the system. Under precise computer control, the selector valve alternately chooses sample and reagent and stacks them into a tubular conduit. The sample and reagent zones are then injected into a reaction coil. The selector valve is then switched to a carrier stream which carries the two zones to a detector. In order to increase sample and reagent zone penetration the carrier flow direction can be reversed several times until adequate mixing has occurred. Although the technique was originally intended for process analysis, it is beginning to find utility in laboratory analysis with the use of a persistaltic pump [41].
8.2 Flow Analysis Review
Sample
RC
215
Detector
Fig. 8-6. Schematic diagram of a sequential injection analysis system. All lines are connected to a central selector valve, SV. Sample and reagents are stacked into the holding coil, HC, and are then mixed in the reaction coil, RC, by repeated flow reversals. Finally, the product is delivered to the detector.
8.2.4 Comparison of Flow Analysis Techniques Over the years there has been a lot of discussion on the advantages of either CFA or FIA over the other technique [28,42,43]. In reality, both techniques are very similar in principles, hardware components, and in most cases will give similar results for a given chemistry. There are, however, situations where there is a definite benefit to using one of the techniques. The following discussion will describe how CFA, FIA, and SIA perform with respect to common flow analysis techniques. 8.2.4.1 Simple Chemistries: Fast Reactions and Few Reagents CFA and FIA perform similarly with respect to analysis rate, reproducibility, and detection limit. The residence time for a sample in these systems is typically less than 2 min. SIA also works well with these chemistries although the analysis rate is slower.
8.2.4.2 Complex Chemistries: Slow Reactions, Many Reagents, and Heating Stages Both CFA and FIA will work with complex chemistries. However, there is a point at which the increased residence time in such systems results in increased sample dispersion in FIA so that detection limit suffers. In CFA, gas-segmentation minimizes sample dispersion so complex reactions and long residence times have little affect on the peak shape and height in the detector. SIA is capable of such systems, but there may be limitations in the number of ports in the selector valve and analysis rate. 8.2.4.3 Sample Dilution There are several dilution techniques available in flow analysis. It is also common for a flow analyzer to be used as a front end solely for the purpose of sample prepa-
276
8 Advances in Flow Analysis
-
Instrumentation and its Application
ration prior to detection on another analyzer such as an atomic absorption spectrometer or a HPLC. The dilution factors obtainable by FIA using the injection valve are higher, more precise, and less complicated than those typically achieved in CFA. Dilution schemes in SIA have been described and are typically a result of precise control of the sample volume injected and dispersed.
8.2.4.4 Sample Preconcentration The precise control of injection valves and pump flow in FIA has created the opportunity for a wide range of preconcentration techniques. Preconcentration has been demonstrated using liquid-liquid extraction, stopped-flow gas diffusion, and packed columns [2,12]. In general, FIA is more compatible with these techniques than CFA. SIA should also work well with packed preconcentration columns, since after a sample has been loaded onto a column, an eluent could be chosen using the selector valve to remove the concentrated sample from the column.
8.2.4.5 Low Level Detection For most of the common chemistries it is possible to achieve similar detection limits whether CFA or FIA is used. However, more complicated chemistries which require longer residence times will have better detection limits with CFA since sample dispersion is minimized and color development can be maximized. Sensitivity in FIA can be improved with stopped flow techniques. Another important factor is the quality of the detector used.
8.2.4.6 Sample Throughput Both FIA and CFA are capable of high volume sample processing. Analysis rates for common chemistries are typically 60 to 100 samples per hour, although very simple chemistries have been demonstrated at the 200-300 samples per hour rate. Analysis rates in SIA are slower than in FIA and CFA. This is because the reagents are stacked into a conduit along with the sample and mixing only occurs through dispersion and flow reversal processes. This slows down the rate at which the entire sample zone comes in contact with all reagents and result in longer analysis times.
8.2.4.7 Precision Both CFA and FIA have been shown to give similar peformance with respect to precision. Precision in both techniques improves as the analysis rate is decreased because there is less carryover from one sample to the next. There is often a compromise that must be made between acceptable reproducibility for a method and sample throughput rate. Precision is typically less than 1% relative standard deviation (RSD) for
8.2 Flow Analysis Review
277
most chemistries. However, precision at both the high (20% RSD) and low (< 0.2% RSD) extremes may be acceptable or necessary depending on the application and industry using the information. SIA should give good precision since it is a highly controlled technique. 8.2.4.8 On-Line Distillation
On-line flash distillation techniques have been used in CFA for many years in the total cyanide, phenol, and fluoride applications. In the procedure, the sample is segmented, acidified, exposed to ultraviolet light (total cyanide only), and heated to 140 "C + where it is volatilized into a miniature distillation head. The volatile acidified form of the analyte is then condensed in a cooled condenser and resampled back into the analytical cartridge where the reagents are added to form the detectable product. Since the residence time in such system is around 5 min, FIA is not practical because of sample dispersion which leads to poor detection limits. Since these analytes need to be measured in the low part-per-billion (microgram-per-liter) range, it is necessary to use CFA. SIA would have a problem handling the resample aspect of an on-line distillation stage. 8.2.4.9 Reagent Consumption FIA usually operates at higher flow rates than CFA. This usually results in faster processing of a given sample so the initial result comes out sooner. Reagent consumption for both techniques is still much less than what is used in the manual procedures. A typical manual procedure may require 100mL of reagents and sample whereas automated flow methods typically use less than 2 mL of reagents per sample. SIA has been demonstrated to offer the advantage of even less reagent consumption because only the amount of reagent necessary to process each sample is introduced into the reaction manifold. 8.2.4.10 Waste Generation As pressure to reduce laboratory waste increases, automated flow analysis procedures will continue to offer the advantage of greatly reduced volumes when compared with manual methods. Both CFA and FIA produce similar volumes of waste solution. SIA has the potential of producing even less waste since only the minimal volume of reactive reagents need be introduced into the reaction tube. This compares with CFA and FIA where all reagents are continuously pumped through the system whether a sample is being processed or not.
278
8 Advances in Flow Analysis - Instrumentation and its Application
8.2.5 Comparison of Flow Analysis with Other Techniques Most of the analytes measured with flow analysis procedures can also be determined with other analytical instruments, such as gas, liquid and ion chromatrographs, and atomic spectrometers. FIA actually shares much in common with liquid chromatography (LC). FIA can be thought of as LC without a column. Flow analysis differs from LC in that there is generally no chromatographic separation of components before detection. Instead, analyte-specific detection is achieved by addition of reagents that form a detectable product with the analyte. Advantages of flow analyzers are that they tend of be simpler to operate, give faster results, have chemistries that are highly specific for each analyte, and have higher sample throughputs. Advantages of other instruments include the fact that they may not use hazardous wet chemical reagents and that they can give multi-component information for a single sample. Many laboratories have evaluated the alternative for automated analysis by comparing performance of flow analyzers with other instruments. For example, ion chromatography (IC), FIA, and CFA were evaluated for the determination of nitrate in natural surface water [44]. The precision and accuracy of the three techniques was similar, however, the detection limit using FIA was found to be worse (0.03 versus 0.006 mg NO; per liter) and the analysis rate for IC was only 10 samples per hour compared with 60 samples per hour for FIA and CFA. While the throughput for IC seems much worse, it is actually not so bad since it was also possible to determine sulfate and chloride from the same injection. CFA and IC were also evaluated for the determination of chloride in surface and ground water samples [45]. Discrete analyzers, also known as batch analyzers, are also being used for environmental sample analysis. One water company laboratory made a comparison of CFA, FIA and discrete analysis for the analysis of routine water quality parameters [46]. Batch injection analysis was also recently compared with FIA [47]. The two techniques were found to share many of the same features, although the batch technique does not require a flow manifold.
8.3 Instrumentation 8.3.1 Pumps Peristaltic pumps are most commonly used in flow analyzers. While there are several well known problems with peristaltic pumps, such as pumptube wear, reagent compatibility with pumptube materials, and the introduction of low frequency noise, it has been difficult to come up with a more practical and economical means of liquid propulsion. Syringe pumps, for example, are excellent at delivering nearly pulseless flows, which in turn improves detection limits and precision, however, the cost per pumping channel is prohibitive for routine equipment where as many as 8 flows per chemistry need to be pumped.
8.3 Instrumentation
279
A sinusoidal pump, based upon a simple cam-driven computer-controlled piston pump, was introduced for SIA [17, 481. It offers the advantages of no pumptubing and pulseless flow, however the flow must be linearized and generally, results in lower sample throughput compared to FIA and CFA. Pressurized vessels have been used to propel flow in process analyzers [49, 501. This type of propulsion is ideally suited for on-line or remote analyzers where it is important to minimize the number of moving parts that could fail or require maintenance. However, it is often difficult to balance pressures when there are many reagents. Electroosmotic flow has also been used for pumping in capillaries [51] and micromachined channels [52].
8.3.2 Injection Devices Six-port two-way rotary injection valves and eight-port slider valves continue to be commonly used in FIA. Additionally, it is also possible to introduce a sample by use of two three-port solenoid valves, and by hydrodynamic injection [2] with two indepently controlled pumps.
8.3.3 Samplers Random access samplers have become popular for use with flow analyzers. They offer several benefits for modern laboratories interested in following quality control procedures and minimizing labor costs. First, it is possible to load containers filled with a check standard into a single position and specify in the software both the location of that container and the frequency with which that check standard should be analyzed. This saves the operator the trouble of pouring a separate cup and placing it in each position where a check standard is to be run. Also, with the use of a dilutor, it is possible for the analyzer to find and re-run off-scale samples after they have been automatically diluted. This saves the operator the trouble of making manual dilutions.
8.3.4 Analytical Manifold Glass reaction coils have traditionally been used in CFA in both macro- and microcontinuous flow systems. Polyethylene and ethylvinyl acetate have also been used for reaction coils. Using polymeric tubing in CFA requires the addition of more surfactant than with glass, but offers the advantage of unbreakable coils. Teflon and polyethylene tubes are most commonly used in FIA. Mixing occurs either with the use of individual ‘T’fittings or in manifolds with integrated ‘T’ fittings. The most
280
8 Advances in Flow Analysis - Instrumentation and its Application
common means of connecting coils and 'T' is with 1/4-28 threaded nuts. Both flanged and ferrule seals have been used. There continues to be a trend towards miniaturization that started with integrated microconduits [2] and is currently being implemented with microlithographic techniques in micromachined analytical manifolds [ 141.
8.3.5 Heat Baths Heating is now commonly accomplished by wrapping polymeric tubing around a heated metallic block. The material chosen for the coil is dependent on the maximum temperature reached and the chemical composition of the reaction. The use of polymeric heating coils represents an improvement over previous CFA heat baths that contained breakable glass coils.
8.3.6 On-Line Distillation Baths The on-line distillation procedure in CFA requires the acidfied sample to be heated to 140- 160"C.In the past, a glass coil immersed in hot mineral oil was used to control the temperature. There were several problems with such a bath including the long time required for stabilization, cool down, and safety and disposal issues. A new dry distillation bath was introduced that consists of a coil of PEEK tubing wrapped around a metallic spool that sits on a heating cone [20]. This new design has been shown to perform similarly to the previous design for total cyanide, phenol and fluoride.
8.3.7 Detectors There are two general ways in which there have been advances in detectors used with flow analysis: (a) Improvements in existing detectors, and (b) adaptation of new detectors.
8.3.7.1 Improvements in Existing Detectors The main ways in which existing detectors can be improved have to do with achieving lower detection limits and elimination of matrix effects. For spectrophotometric detectors, lower detection limits were traditionally obtained by increasing the amount of sample introduced into the system and by increasing the flowcell pathlength. While detector technology for flow analyzers had not advanced much over the years, there had been more effort put into the development of high performance low-noise
8.3 Instrumentation
28 1
monochromator-based UV-Vis detectors for HPLC. Such detectors were incorporated into a flow analyzer that operated in both FIA and CFA modes [20]. The detectors have absorbance range scales down to 0.0005 absorbance units full scale (AUFS) which gives an indication of the absorbance they are capable of distinguishing. The reality of flow analyzers, however, with their hydraulically noisy flows and small background absorbances has meant that the very lowest range scales are not practical. Nevertheless, the use of these detectors has proven to be very successful in reaching lower detection limits than have been commonly achieved. The other benefit of such detectors is that it is not necessary to change the volume of sample injected, pumptube sizes, or flowcell when switching from one concentration range to another. With everything in the system held constant except the range scale on the detector, different concentration ranges for nitrite were measured. It was found that with the absorbance scale set to 1 AUFS the range was 0.1 to 10 mg NO;-N/per liter, whereas if the absorbance scale was set to 0.005 AUFS the dynamic range was from 0.001 to 0.100mg NOT-N per liter [20]. For the most part, detectors with analog outputs have been used with flow analyzers and the signals are either recorded on a chart recorder or they are converted to a digital signal for processing. The limitation with chart recorders and software programs that emulate chart recorders is that there is normally a dynamic range limit of a factor of 100-200 in concentration. This is because a sensitivity range scale must be set on the detector which then corresponds to the detector output voltage, such as 0.5 AUFS = 0 to 1 V Full Scale. If the absorbance is greater than 0.5 AU, the recorder or software will not be able to measure the response since it will be off-scale. To get around this limitation several techniques have been developed to improve the dynamic range at both the low and high ends of concentration. For example, optical dilution is a technique where sequential flowcells of varying lengths are used to give different dynamic ranges. A technique known as auto-dilution is accomplished with a random access sampler and a syringe pump that meters in precise volumes of sample and diluent to give dilution of off-scale samples. Both of the preceding techniques and numerous other procedures were developed to extend the range over which flow analyzers operate without having to change the system and recalibrate. Observing that further improvements in HPLC detectors were gained with the use of high-resolution digital detectors, a new photometric detector with a 24-bit analogto-digital (A-D) converter was developed for use with a flow analyzer (Perstorp Analytical Environmental). For example, the 24-bit detector was compared with the monochromator-based detector for the spectrophotometric measurement of cyanide using pyridine-barbituric acid and chloramine-T. The monochromator detector typically has a detection limit of 0.0005 mg CN per liter and a working range of 0.005 to 0.5 mg CN per liter. With the 24-bit detector, the detection limit is 0.0002 mg CN per liter and the dynamic range is 0.005 to 5 mg CN per liter, a factor of 10 higher in dynamic range. The reason for the improvement is due to the increase in resolution from 4096 resolution units with typical 12-bit A-D convertors to over 16 million resolution units with the 24-bit A-D convertor. This ability to cover at least three-orders-of-magnitude and still achieve low limits of detection eliminates the need to run separate ranges for a single chemistry and minimize the need for as many manual and mechanical auto-dilutions.
282
8 Advances in Flow Analysis
-
Instrumentation and its Application
The elimination of matrix effects (refractive index, turbidity, color) has been the subject of other detector developments. In turbid or colored samples, the practice of dialysis continues to be the best technique for clean-up prior to detection. Matrix matching by using a wash or carrier stream of the same matrix as the sample has also been extensively used to create a flow of consistent matrix composition through the detector. However, due to the variable salinity in estuary and sea water, matrix matching techniques are very difficult. In addition, it is difficult to create interference-free artifical or low-nutrient seawater. Therefore, there have been several detector designs developed specifically with the needs of seawater analysis in mind. Double-beam techniques have been developed in which light is passed through the flowcell and is then split into two beams. The sample beam passes through a bandpass filter of an absorbing analytical wavelength while the reference beam passes through a bandpass filter of a non-absorbing wavelength. Both signals are detected at photodiodes and the ratio compared. In this manner, any effect due to refractive index is cancelled. Another way to compensate for refractive index variations is to use a large diameter flowcell. The large diameter effectively makes the signal insensitive to refractive index by exposing a larger active area of the photodiode detector cell [53]. Flow analysis has been coupled with atomic spectrometry for many years and there continue to be improvements in sample introduction devices [54]. 8.3.7.2 Adaptation of New Detectors
From the outset of flow analysis, it was recognized that a variety of detectors could be coupled with the front end sample processing system. This feature has enhanced the capabilities and performance of dozens of optical and electrochemical detectors. Recent application of new detectors in flow analysis include the use of photodiode array detectors for multicomponent detection [55 - 581, a multichannel amperometric detector [59],ion-selective field-effect transistors [60, 611, electrochemical gas sensors [62], and piezoelectric crystals [63, 641.
8.3.8 Data Handling Modern flow analyzers are interfaced by sophisticated software programs, many of them now being offered in a Microsoft WindowsTMformat. This is quite an improvement over chart recorders, which are still widely used with many AutoAnalyzerTMsystems in the field. Modern software programs include quality control features such as automatic insertion of check standards. Results are presented in spreadsheet format and can be exported to other laboratory information management systems. Icon-based system design using programs such as Visual BasicTMrunning under Microsoft WindowsTMhas been demonstrated for a dual-channel sequential injection analyzer [65]. System set-up is made on the computer screen, where generic
8.4 Flow Analysis Techniques
283
hardware is made application specific by insertion of icons representing components. The hardware layout is then activated automatically without the need for the operator to physically change anything by hand. Finally, advances in information technology are also being applied to flow analysis, such as multimedia programs on CDROM for interactive training and troubleshooting.
8.4 Flow Analysis Techniques This section will discuss important flow analysis techniques which are relevant to environmental analysis. A more detailed review on the fundamentals of flow analysis techniques along with many cited applications can be found in Refs. [2] and [3]. Reference [66] gives a good overview of important issues in CFA and FIA. The tremendous popularity of flow analysis has been due to its combination of simple instrumentation with a wide range of techniques and detectors. These techniques can generally be considered to be procedures for reproducible sample treatment (sample transport, dilution, derivatization, stopped-flow, solvent extraction, gas diffusion, dialysis, preconcentration, removal of interferences, and digestion) prior to coupling with a variety of detectors (single- and multicomponent determination). Each flow analysis procedure is made up of the appropriate combination of techniques and detector. Several techniques may be used in a given application and each technique can serve one of several roles depending on the application. For example, gas diffusion can serve both to separate an analyte from interfering species and to concentrate it into a smaller volume zone when operating in the stopped-flow mode.
8.4.1 'Ikansport The simplest use of a flow analyzer is for reproducible sample transport to an analyte-specific detector. Limited dispersion systems are useful when the original composition of the sample is to be measured, such as in some flame atomic absorption applications [67], and potentiometric measurements of pH [68].
8.4.2 Dilution Due to the limited dynamic range of most detectors in flow analysis, it has been necessary for there to be development of both on-line dilution and preconcentration schemes. On-line dilution can be set up so that either all samples are diluted or only those that result in off-scale responses. If all samples are to be diluted, dialysis and dilution loops have traditionally been used with CFA. Dialysis typically gives a dilution factor of 50, dilution loops up to a factor of 100, and combined they give up
284
8 Advances in Flow Analysis - Instrumentation and its Application
to a factor of 5000. FIA dilution techniques have proven to be both simpler and more precise. All samples can be diluted by simply loading a small volume of sample into the injection loop and injecting the sample into a high flow rate carrier. Such a sample zone can be further diluted by filling a portion of a second injection valve with the diluted sample and injecting it into the analytical manifold. Gradient dilutions involve simply quantitating over a small segment of the tail of the sample zone as it flows through the detector. If only those samples that are offscale are to be diluted there are a few techniques for diluting those samples. One way is to have the analyzer automatically and precisely meter a fixed volume of sample and a fixed volume of diluent into an empty container. The diluted sample can then be analyzed on the flow analyzer and the results are displayed with consideration of the dilution factor. A new technique has been introduced in FIA in which a high precision pump is used to meter the sample into only a small portion of the sample loop. The loop is then injected into the carrier stream and is analyzed [69]. The advantages of this technique are that it is simpler than other dilution procedures, it involves only one valve, and allows a continuous range of dilution factors to be obtained.
8.4.3 Derivatization The most common technique in flow analysis is treatment of the sample zone with reagents to form a detectable product with the analyte of interest. Single line FIA with a derivatizing reagent carrier has been demonstrated for simple chemistries with fast reaction kinetics. The sample is injected into the reagent and product is formed as the sample disperses into the reagent stream. More commonly, multiline systems are used in FIA and CFA to bring sample and reagents together in a series of mixing ‘T’s. Flow analysis has also been used for both precolumn and postcolumn derivatization in liquid and ion chromatography [25, 701. For example, the trace metals Co, Ni, Cu, Zn, Cd, and Pb were determined by ion chromatography followed by postcolumn reaction with 4-(2-pyridlazo)resorcinol with the absorbance measured at 500nm [71]. Another technique, known as reversed FIA, involves injection of reagents into a carrier stream consisting of continously aspirated sample. As in conventional FIA, the detector response is a transient peak that results from reagent dispersion into the sample carrier. This technique was originally developed for the measurement of orthophosphate in seawater using continuous sample aspiration and reagent injection [72]. It was further developed into a procedure where various reagents were injected into a sample carrier stream both sequentially [73] and in parallel [74] to give multicomponent information when using a single detector at a single wavelength. There are several characteristics of reversed FIA which make it a useful technique for continuous environmental monitoring and process control. These include the fact that there is usually an abundance of sample, reagent consumption is reduced, and it is possible to obtain multicomponent information using spatially resolved injec-
8.4 Flow Analysis Techniques
285
tions of various reagents and a single detector (single wavelength of diode array detection for spectrophotometry).
8.4.4 Stopped-Flow Stopped-flow is a technique used mainly in FIA in which one of the flows in stopped for a period of time during each analysis [2, 751. The flow is stopped either by having a pumping channel stop or by switching valves in such a way that a certain zone of fluid is stopped. There are several uses of stopped-flow techniques. First, stopping the flow allows for increased reaction times with no additional dispersion except for a small amount of molecular diffusion. This is helpful when the reaction kinetics for color formation are slow, or when the sample zone is being pretreated for a long period of time. For example, in the case of ultraviolet light-induced decomposition steps, it is possible to stop the flow of the sample zone while it is illuminated by the UV light in order to obtain higher decomposition rates while minimizing dispersion. Stopped-flow can also be used with solvent extraction, gas diffusion, or dialysis to preconcentrate an analyte into a static receiving solution. The receiving solution is held constant on one side of a semi-permeable membrane that is mounted in a separation cell while the sample zone passes on the other side of the membrane. After the sample zone has passed through and on to waste, the receiving stream begins to flow and proceeds to the detector. Finally, kinetic information can be obtained for a reaction by stopping the flow while the sample zone is in the detector flowcell. When the flow is stopped the rate of product formation can be measured by plotting the detector signal versus time. Since it is the change in signal over time that is being measured rather than an absolute signal, it is possible to measure the concentration of an analyte in a sample containing a background signal. The technique has had success in biotechnological applications and could prove useful in environmental analysis in samples with a matrix which has a background absorbance.
8.4.5 Solvent Extraction Flow analysis has traditionally been dependent on development of analyte-specific reagents or detectors to obtain selectivity. Chromatography, in contrast, has traditionally been based on separation of the analyte of interest from other sample constituents followed by detection at a non-specific detector, such as a conductivity detector. Over the years, the advantages of both approaches have been extensively borrowed from each other so that it is now common to see chromatography with postcolumn derivatization for analyte-specific detection and flow analysis with some degree of separation. Separations can be achieved in flow analysis with solvent extraction, gas diffusion, dialysis, and solid-phase extraction. Flow analysis has proven to be an excellent mechanism to automate solvent extractions [2,76,77]. Many of the sampe processes that are currently carried out on a
286
8 Advances in Flow Analysis - Instrumenfation and its Application
Fig. 8-7. Schematic diagram of solvent extraction in flow injection analysis. Separation of the organic and aqueous zones occurs in the phase separator, PS. This can be achieved either through gravity separation or membrane separation.
macroscale in separatory funnels to extract an analyte from one immiscible liquid to another, can be carried out on a microscale in flow conduits. The avantages of automated extraction include a reduction in solvent use, waste generated, and human exposure to solvents since all liquids are confined to the flow conduits. Liquid-liquid extractions in CFA have traditionally been carried out without air segmentation because the organic solvent provides the segmentation. In this way, CFA and FIA are very similar, differing only in the means of sample introduction. Typically, in FIA with solvent extraction the sample is injected into a aqueous carrier stream to which is often added a derivatizing reagent and a salting-out agent (Fig. 8-7). As the organic reagent is added to this stream a reproducible pattern of organic segments is introduced. Extraction of the analyte into the organic phase occurs in an extraction coil which is a tightly coiled tube in which the extractable species in the aqueous phase separate into the organic phase. The organic phase is then separated from the aqueous phase in either a ‘T’ separator by differences in density, or in a membrane phase separator equipped with a hydrophobic membrane which allows passage of only the organic phase. Finally, the organic phase containing the extracted analyte continues to the flow through detector cell where the analyte is deteced. Anionic surfactants in river water were measured using spectrophotometric detection of the cationic dye-surfactant ion pair that extracted into the organic phase [78]. Cation surfactants in water were determined indirectly by extraction of their ion pairs with thiocyanocobalt(I1) into 4-methylpentan-2-one-followedby phase separation and detection by flame AAS [79]. Although liquid-liquid extractions can be considered primarily useful for separation of an analyte from its matrix, it can also serve to concentrate that analyte prior to analysis. If the volume of organic phase relative to the aqueous phase is small, the effective concentration of analyte will become greater as it is extracted. In addition, aqueous-aqueous extractions have been achieved through the use of liquid membranes consisting of a hydrophobic membrane pretreated with a solvent support [SO]. Preconcentration can be achieved with such membranes by stopping the flow of the receiver stream while continuing to pass the donor stream over the membrane. When the receiver stream flow is started again it carries with it a concentrated zone of analyte to the detector.
8.4 Flow Analysis Techniques
287
8.4.6 Gas Diffusion Certain analytes can be chemically converted to their volatile derivatives which can then be separated by gas diffusion across a hydrophobic membrane such as Teflon. Typically, the sample is injected into an aqueous carrier stream to which is then added a reagent which converts the analyte to its volatile form. The donor stream then flows past a hydrophobic membrane which simultaneously has a recipient stream passing on the other side. Gas transfer is made essentially irreversible by use of a recipient stream that either converts the volatile form of the analyte to the non-volatile form or passes it into a gaseous recipient stream. The recipient stream is then either treated with a derivatizing agent or sent directly to a detector. Since only gaseous species pass through the membrane, both matrix color and ionic interferences are eliminated prior to detection. Both planar and tubular membrane separators have been used. Examples include conversion of carbonate and bicarbonate to carbon dioxide [81,82], ammonium to ammonia [83], and cyanide to hydrocyanic acid [84]. In addition metals (As, Hg, Se, Sb, Te, Bi, and Sn) which form gaseous hydrides can be separated across a membrane into a gaseous carrier stream and carried to an atomic spectrometer [85 - 871, although gas-liquid separators without membranes are commonly used today [88]. Gas diffusion has been found to be more efficient in FIA because reduced channel dimensions in the gas diffusion cell improves gas transfer efficiency. However, for applications with long residence times which may warrant air-segmentation to improve detection limit, gas diffusion with segmented donor and recipient streams has also been used. Gas diffusion can also be combined with stopped-flow to provide a preconcentration technique. If the recipient stream is stopped and the donor stream allowed to pass, a preconcentration effect occurs. For example, trace level NOz from ambient air was preconcentrated from a segmented donor stream into a stopped unsegmented recipient stream [89]. When the flow was started again, the recipient stream carried the NOz to a nitrate selective electrode. Gas diffusion can also be used to eliminate salinity or refractive index errors in the measurement of ammonia in seawater. The sample is combined with a basic reagent to adjust the pH to 12. Ammonia diffuses across a hydrophobic membrane into a recipient stream. Colorimetric detection of ammonia follows using an acid-base indicator or more specifically by reaction with alkaline phenol and hypochlorite.
8.4.7 Dialysis The purposes for dialysis in environmental analysis include separation of the analyte from unwanted constituents in the original sample matrix (removal of color, particulates, turbidity), and precise dilution. Dialysis occurs in the same type of membrane separators used for solvent extraction and gas diffusion although the hydrophobic membrane used for those separations is replaced with a hydrophilic membrane such as cellulose acetate. Dialysis is routinely used for front-end sample treatment in the analysis of soil and plant extracts which typically need color removal and dilution prior to analysis.
288
8 Advances in Flow Analysis - Instrumentation and its Application
8.4.8 Preconcentration Preconcentration is an area of increasing interest within flow analysis, particularly for lowering detection limits in environmental samples. For example, in fields such as marine chemistry, the concentration level at which a chemical species is environmentally important may be below conventional detection limits [90]. In most cases, sample preconcentration also serves to reduce interferences since the analyte is separated from its original matrix. There are four general techniques used for preconcentration: solvent extraction, solid-phase extraction, gas diffusion, and precipitation. The dual role of these techniques in preconcentration and interference removal results in the fact that there is an overlap in the use of these techniques, which can generally be called separations. While all separation steps do not necessarily result in a preconcentration effect, it is true that all preconcentrations are to some degree a separation step. While solvent extraction and gas diffusion can be used for preconcentration, they are principally used for separation. The most convenient and popular technique for preconcentration is solid-phase extraction. The use of solid-phase reactors for preconcentration and other techniques is reviewed in several references [2, 12,911. This technique has been developed primarily with FIA and there is very little in the literature on solid-phase preconcentration with CFA. The technique involves passage of a large volume of sample through a small packed reactor containing a solid reagent which preferentially binds the analyte. This is followed by passage of a suitable eluent which releases the concentrated analyte to the analytical manifold where it is processed prior to detection. Preconcentration factors up to 100 can be obtained with this technique. The packed column can conveniently be used in place of a sample loop in the injection valve of a flow injection analyzer (Fig. 8-8). In the ‘load’ position, sample is ‘loaded’ through the packed column. The enrichment factor is dependent on the volume of sample solution pumped through the column and on the volume of eluent used for release. Care must also be taken so the amount of analyte does not overload the capacity of the column. In the ‘inject’ position, the valve is switched so that the eluent passes through the column and the analyte is ‘eluted’ to the analytical manifold. Solid-phase extraction in FIA is similar to trace enrichment in chromatography [92] with a few notable differences. First, since relatively large particles are used in the column (50-200 pm) the sample flow rate is fast and low pressure pumps can be used. The increased flow rate results in faster operation and better sensitivity. Also, the absence of a chromatographic column allows for more possibilities in type of eluent and preconcentration column material since there are not compatibility problems [24]. Solid-phase extraction has principally been used as a preconcentration and matrix removal technique for trace metals prior to detection by atomic absorption [93] or inductively coupled plasma spectrometry [94]. For example, Cr(II1) in seawater was measured by flame atomic absorption spectrometry following preconcentration on a poly(hydroxamic acid) resin [95]. Preconcentration has also been used for other environmentally critical analytes, followed by spectrophotometric detection with suitable reagents. For example, 0.1 pg P per liter of dissolved reactive phosphorus in natural waters was measured using acidic ammonium molybdate and tin(I1) chloride reagents following sample preconcentration on a Bio-Rad AG
8.4 Flow Analysis Techniques
Sample
Load position
(c)
289
Sample
Inject position
Fig. 8-8. Preconcentration using a packed column in FIA (a). In the load position (b), sample is passed through the column where analyte is concentrated. In the inject position (c), carrier elutes the analyte to the analytical cartridge for detection. The direction of elution is often the reverse of the direction of loading to prevent zone spreading.
1-X8anion-exchange resin column [96]. Preconcentration by sorbent extraction has been found to be enhanced by detecting product directly in the column by flow injection ‘optosensing’ [97]. The advantages of this technique are that lower detection limits can be reached since there is no dilution during an elution step, and kinetic information can be used to discriminate interferents. For example, orthophosphate was mixed with heptamolybdate and then concentrated on a Cis column. Ascorbic acid is then pumped across the sorbed bed, the flow is stopped, and the signal is measured. The valve is then switched to the elute position so that the product is removed from the column. Precipitation has also been used for on-line preconcentration and matrix removal in FIA. In this technique, the sample is treated with a reagent which forms a precipitate with the analyte to be measured. The precipitate is trapped on a stainless steel or nylon membrane filter and is subsequently solubilized by a reagent and then detected. For example, lead was preconcentrated and separated from its original matrix by precipitation following treatment with an ammonia solution in a reversed FIA sys-
290
8 Advances in Flow Analysis - Instrumentation and its Application
tem [98]. The precipitate is collected on a filter and then eluted after nitric acid solution is injected into the system.
8.4.9 Removal of Interferences Many of the sample pretreatment techniques described above serve in some way to remove interferences. Several of the techniques, such as solvent extraction and gas diffusion, actually physically separate the analyte of interest from the original sample matrix. Solid-phase extraction also acts to isolate the analyte from the rest of the original sample matrix, however, since sorbents are not specific for each analyte, they will collect classes of species and not just the analyte. Solid-phase reactors have also been used exclusively for removal of interferences in both CFA [3] and FIA. For example, nickel interference in the determination of arsenic by flame atomic absorption spectrometry (AAS) has been minimized by the on-line removal of the nickel on a strong cation-exchange column [99].
8.4.10 Digestion Automated on-line digestion procedures are powerful techniques within flow analysis. These procedures involve treatment of the sample with a suitable digestion reagent followed by digestion by means of exposure to either high temperature, ultraviolet light, or microwave energy. These procedures offer the advantage of automating digestion procedures which are typically carried out off-line by high temperature refluxing or distillation procedures. In addition, coupling the on-line digestion with automatic detection saves the trouble of having a two-step process. An example of a high temperature digestion procedure is the CFA method for the determination of total inorganic phosphate. When this step is combined with a flow through UV digestor, total phosphorus can be obtained since the UV radiation decomposes the organic-phosphorus bonds to release orthophosphate. Other on-line UV digestion procedures include the measurement of total soluble nitrogen [loo], total organic carbon [loll, and total cyanide [102]. The total cyanide procedure was recently improved by development of a technique for measuring total cyanide without interference from thiocyanate [ 103, 1041. Traditionally, the automated UV procedure resulted in artificially high total cyanide concentrations because UV digestion decomposed thiocyanate to cyanide while the manual distillation procedure did not. To solve the problem, a lamp with a longer wavelength in the UV spectrum was used in the UV digestor to remove the shorter wavelengths of UV light that were decomposing the thiocyanate. In addition, the system was simplified by use of isothermal distillation (gas diffusion) to separate the acidified sample solution that contained HCN into a recipient stream containing NaOH [84, 1041. The number of applications taking advantage of photolysis in flow analysis is very limited when considering the great number of organic compounds determined with HPLC after photolytic conversion [105].
8.4 Flow Analysis Techniques
291
Microwave digestion has been used primarily as an off-line digestion technique followed by automated flow analysis. The main difficulties in coupling microwave digestion with flow analysis have been problems with matching temperature and pressure differences between sample in the microwave digestor and a conventional flow system. Throughput is also slow since the residence time in the digestor is long. Recently, on-line microwave digestion has been demonstrated for sample pretreatment prior to hydride generation and cold vapor atomic absorption spectrometry [ 1061. Additionally, an automated system for the determination of total phosphorus in waters by on-line microwave digestion followed by amperometric detection was described [107]. Also, chemical oxygen demand has been determined in well and river waters by FIA with oxidation in a microwave oven [log]. Direct introduction of solid samples into continuous-flow systems by use of a leaching agent and ultrasonic irradiation has been reported [109]. The procedure was applied to the determination iron in plant materials by acid dissolution followed by complex formation with 1,lO-phenanthroline.
8.4.11 Combined CFA with FIA Gas-segmentation offers the benefit of reducing sample dispersion which allows for applications which require long residence times with little affect on sensitivity. Meanwhile, FIA offers the benefits of convenient and precise sample injection volumes, various techniques with precise control of flows (stopped-flow, preconcentration), and fast display of results. However, gas-segmentation has rarely been used in FIA despite the obvious potential benefits offered by segmentation. Clearly, it is possible to combine the use of the injection valve for sample introduction with an analytical manifold based upon gas-segmented flow to obtain a hybrid FIA-CFA system which benefits from both techniques [21, 1101. As long as the gas-segmentation pattern is regular and there is an ability to measure peak area, such hybrid systems provide good sensitivity along with good precision. One recent study involved the use of an air-segmented carrier stream for coupled to an ICP-MS [I 111. In another system, hybridized FIA-CFA was used for the measurement of both free and total cyanide [I041 with a field portable flow analyzer (Fig. 8-9). Due to the low detection requirements and long residence time in the UV digestor it was necessary to use airsegmented flow to reduce dispersion. At the same time, it was not practical for there to be an autosampler for sample introduction. Therefore, an injection valve was used for sample introduction into a carrier stream which was then acidified and segmented. The flow passes through a UV digestor (total cyanide only) and then through a gas diffusion cell where the HCN passes into a non-segmented alkaline receiving solution which flows to the detector. Without air segmentation the peak height was found to be 30% of the peak height with segmentation. Sample injection has also been used in a gas-segmented continuous flow hydride generator [112]. Another application is the determination of chemical oxygen demand in aqueous environmental samples by segmented flow injection analysis [113]. Potassium permanganate
292
8 Advances in Flow Analysis - Instrumentation and its Application Gas diffusion
eQ-*waste
-
Air Base
Membrane
----- - QO-
i
Pump
Amperometric detector
Fig. 8-9. Schematic diagram of a hybrid CFA/FIA system for free and total cyanide. For the determination of free cyanide the UV lamp is off or is by-passed.
was used as both the oxidant and spectrophotometric reagent, the temperature was thermostated at 95 "C, and the residence time was 8 min.
8.4.12 Coupling with Other Instrumentation Flow systems have been combined with a wide range of analytical instruments to provide for either convenient automated sample preparation of derivatization procedures. Within the field of environmental analysis, the use of flow analysis techniques coupled with atomic spectrometers continues to be an area of expanded development. While gas-segmented continuous flow methods have been used with atomic absorption and atomic emission spectrometers for many years [114, 1151, FIA has recently come into its own as a versatile technique for sample preparation prior to atomic absorption spectrometry (AAS) [15, 1161. The principle advantages of FIA in AAS include: (a) the ability to use small volume samples in addition to other dilution techniques; (b) the capability of reducing the loading of undesirable matrices on nebulizers and torches; (c) rapid sample throughput; and (d) the ability to carry out on-line chemical processes, such as hydride generation and standard additions. These features allow for extended ranges in AAS and for the analysis of samples in matrices with high dissolved solid content or chlorinated organic solvents. A flow injection system is now commercially available (Perkin-Elmer) that was designed and optimized for use with atomic absorption instrumentation [117, 1181. FIA has been used to automate many of the sample preparation steps in flame atomic absorption spectrometry such as on-line reagent addition, matrix removal, hydride generation, and preconcentration [15, 1191. Other 'hyphenated' flow analysis systems that have been demonstrated include FIA-electrothermal AAS [ 1201, FIA-graphite furnace AAS [95], cold vapor atomic fluorescence spectrometry [121], FIA-TCP-AES [222], FIA-ICP-MS [123, 1241, FIA-GC [125], FIA-IC [25, 72), FIA-LC [25, 721, CFA-LC [26], and FIA-FTIR [126].
8.5 Environmental Application of Flow Analysis
293
8.4.13 Multicomponent Determination Although the majority of flow analysis procedures involve measurement of a single analyte using a single detection channel, there are a variety of techniques available for obtaining multicomponent determination. There are two main techniques for multicomponent analysis in FIA: use of multivariate detectors, and sequential detection. Multivariate detectors are detectors in which there is an array of response information over which the processed sample zone is measured. The information is then specifically dispersed to unique detection channels, such as in an ICP atomic emission spectrometer, or multivariate statistics is used to interpret a sample response which may be a combination of individual responses of each component. Such is the case when using a photodiode-array spectrophotometer in combination with a derivitizing reagent which forms product with more than one species to be detected in a sample, or the electrochemical response in an array of electrochemical detectors. After a calibration has been made of the pure components through the detection system it is possible to analyze the unknown sample and obtain concentration values for the individual components provided there are enough response differences. One example of this approach is the simultaneous determination of cobalt and nickel with photodiode array detection [127]. The method exploits the fact that the complexes between the metal ions and 4-(2-pyridylazo)-resorcinolare slightly different and be interpreted using the partial least squares regression technique with outlier detection. Fast scanning and pulsed voltammetric procedures also give multicomponent information. For example, Cd(I1) and Pb(I1) were determined using pulsed voltammetric analysis in a CFA system with stopped-flow analysis in the detector flowcell [128].
Multicomponent analysis by sequential analysis can be implemented in several ways. First, using single detector, various reagents can be injected into a reversed FIA system. The responses to the various products can be measured assuming there is temporal resolution between the two products at the single detector. Another implementation involves splitting the stream after sample injection into several arms, each of which is treated in such a way that multiple species can be determined.
8.5 Environmental Application of Flow Analysis While most common environmental applications are available from the commercial flow analyzer suppliers, there are several good sources of information regarding specific environmental applications developed on flow analyzers. These can be found in the extensive bibliographies of applications that have been published for FIA [2, 9, 101and CFA [3 - 81. Since most environmental applications involving flow analysis are with water samples, a very good resource for recent developments is the Water Analysis Review published every other year in Analytical Chemistry [11]. This section will offer examples of recent application development in environmental analysis.
294
8 Advances in Flow Analysis
-
Instrumentation and its Application
The use of automated flow analysis techniques for the analysis of seawater has been the subject of several separate reviews [129-1311. A recent publication describes improved methods for the determination of nutrients in seawater using a microbore continuous flow analyzer with higher analysis rates and reduced interference of magnesium on ammonia [132].
8.5.1 Inorganic Analysis 8.5.1.1 Alkali and Alkaline-Earth Metals
In addition to the flow analysis bibliographies described above, the analysis of cations by FIA has been reviewed [133]. Calcium and Magnesium. There have been many electrochemical, spectrophotometric and flame AA spectrometric flow analysis procedures reported for the analysis of calcium and magnesium separately. For example, magnesium was determined indirectly by amperometric measurement of excess Eriochrome Black T after formation of the magnesium complex at pH 11.5 [134]. Several simultaneous procedures have also been developed including FIA with photodiode array detection. For example, calcium (1 - 10 mg Ca per liter) and magnesium (1 -20 mg Mg per liter) were measured simultaneously using 4-(pyridyl-2-azo)resorcinol complexing agent [58]. Potassium and Sodium. In addition to the automated flame AAS procedures that have been used for years, it is possible to extract the insoluble crown ether complexes of the alkali metals into an organic phase for spectrophotometric detection 1135-1371. For example, a FIA application for potassium was developed where a water sample was injected into an ethanolic solution of dipicrylamine to which was then added a toluene solution containing dibenzo-18-crown-6 [ 1351. After solvent extraction and phase separation the aborbance of organic layer was measured at 420 nm. Ammonium. Ammonium is one of the most common species routinely tested with flow analysis equipment. However, as with other critical analytes, the method used is entirely dependent on the local standard procedure allowed, the sample matrix, and the sensitivity requirements. For example, while the US EPA requires the indophenol blue method for pollution discharge monitoring, several other countries allow a gas diffusion method which often involves less hazardous reagents. Recent published work on ammonium analysis includes the measurements of ammonium to 50 pg L-' in natural waters using Nessler reagent after preconcentration on a cation exchange column [138]. A fully automated FIA system for the continuous monitoring of ammonium in fishfarming seawater based upon the salicylate method with a 3.5% NaCl carrier solution was developed [139]. Ammonium in seawater has also been determined by CFA with gas diffusion followed by derivatization with phenol, nitroferricyanide, and hypochlorite to give indophenol blue which is measured at
8.5 Environmental Application of Flow Analysis
295
640 nm. Gas diffusion eliminates problems with salinity or refractive index errors and gas segmentation allows for the sensitive and specific reagent chemistry and heating stage to take place. The detection limit with this method was 0.05 WM NH2-N per liter [140]. Gas diffusion of ammonia from water samples provided additional selectivity for potentiometric and conductimetric detection in flow injection analysis [141]. A flow injection system for continuous monitoring of ammonia in freshwater streams was developed based upon gas diffusion followed by potentiometric detection in a solid-state tubular flow-through ammonium electrode [ 1421. Beryllium. An FIA method for beryllium with chrome azurol S as the chromogenic agent was reported [143].
8.5.1.2 Transition Metals
Chromium. Cr(II1) in seawater was determined by flow injection flame AAS after it was concentrated on a poly(hydroxamic acid) resin [144]. In another procedure, Cr(V1) and Cr(II1) were complexed with sodium diethyldithiocarbamate, concentrated on a C,*-bonded SiOz reversed-phase sorbent column. Both Cr(V1) and total chromium were then determined by flow injection electrothermal AAS [1451.Cr(II1) and Cr(V1) were also separated on activated alumina followed by flame AAS [146]. Manganese. Flow injection methods for the determination of low levels of manganese in seawater were developed with chemiluminescence detection after preconcentration on a chelating resin [58, 1221. A spectrophotometric method involved the preconcentration of Mn(I1) onto 8-hydroxyquinolineimmobilized onto a vinyl polymer gel followed by elution with acid and detection of the malachite green formed from the reaction of leucomalachite green and potassium periodate with Mn(I1) acting as a catalyst [ 1471. Manganese was also determined with stopped-flow catalytic spectrophotometry [ 1481 and polarography [1491.
Iron. Iron(II1) was determined in seawater by FIA with preconcentration on a 8-quinolinol immobilized chelating resin column to improve sensitivity and remove interferences from other heavy metal ions, such as Mn(II), Cr(III), Co(II), and Cu(I1) [150]. After elution with dilute hydrochloric it is mixed with luminol solution, aqueous ammonium, and hydrogen peroxide, successively. The iron concentration was obtained from the chemiluminescence intensity and the detection limit was 0.05 nmol L-' when using an 18-mL sample. Cobalt. Cobalt was complexed with sodium diethyldithiocarbamate which was then preconcentrated onto a CI8-bonded silica reversed-phase column [ 1511. A 28-mL sample was collected over 10 min, eluted, and detected using electrothermal AAS. Using seawater samples, a detection limit of 1.7ng Co per liter was obtained. Chemiluminescence methods have also been developed which include a preconcentration step as a first step [152].
296
8 Advances in Flow Analysis - Instrumentation and its Application
Copper. Several flow injection solvent extraction methods have been reported that involve either spectrophotometric [I 531 or ICP-AES detection [I 541. Zinc. The zinc(II)-5,7-dibromo-8-quinolinol complex was extracted into the diethylether phase of a flow injection solvent extraction system [I%]. After phase separation the fluorescence of the organic layer was measured and the detection limit for zinc analysis was 3 pg Zn per liter. Mercury. There are a number of methods for the determination of mercury by the cold vapor AAS procedure, however until recently they were not very successful for total mercury which includes organomercury species. A new procedure involves online oxidation with sulfuric acid and potassium persulfate [156]. In another report, both CFA and FIA were coupled to a cold vapor AA spectrometer for Hg analysis [157]. Decomposition of organic mercury compounds using the combination of Cu(I1) and Sn(I1) in alkaline solution was used and the detection limit using CFA was 0.08 pg Hg per liter and 0.5 pg Hg per liter with FIA [ISS]. This is another example of the improved sensitivity obtained by adding segmentation. Also, a continuous sampling and monitoring system was developed for analyzing mercury in wastewaters. The samples are digested, reduced, and extracted into 0.5 mm tubing, prior to determination by cold vapor AAS. The detection limit was 0.15 pg Hg per liter. Finally, Hg(I1) was determined after solvent extraction (n-butyl acetate) and fluorescence excitation at 452 nm and detection at 476 nm [159]. Aluminum. A portable continuous FIA monitor of aluminum which makes an analysis every 30min has been developed [160]. In the system, Fe(II1) interference is reduced by reduction to Fe(I1) with 1,lO-phenanthroline followed by addition of pyrocatechol violet and spectrophotometric measurement at 580 nm. Speciation of aluminum species was achieved based on determination of reaction rates with oxine [1611. After mixing, the aluminum-trioxinate complex and residual oxine were extracted into chloroform and the residual oxine determined spectrophotometrically. Speciation of total, reactive non-exchangeable, and reactive exchangeable aluminum has been achieved by alternate injection of sample either directly or through a cation exchange column [162]. The sample is then acidified, mixed with iron masking solution, and finally mixed with pyrocatechol violet. Direct injection gives total reactive aluminum and injection after passing through a cation exchange column gives nonexchangeable reactive aluminum, consisting of organic aluminum species. The difference between the two results is the reactive exchangeable aluminum, consisting of free aluminum species. Another method involves the sensitive spectrofluorometric FIA method measurement of the aluminum complex with 5,7-dibromo- 8-quinolinol after extraction into diethyl ether [163].
Lead. Lead has been determined by flow injection potentiometric stripping analysis [I641 and electrothermal or flame AAS [69]. Arsenic. Arsenic has primarily been determined by the arsenidhydride generation technique prior to atomic spectrometry. On-line prereduction of As(V) to As(II1) by
8.5 Environmental Application of Flow Analysis
297
addition of HCl and KI-ascorbic acid followed by hydride generation and flame AAS has been reported [165]. L-Cysteine has also been used as a reducing agent prior to hydride generation. L-Cysteine offers the advantages of much lower acid concentrations compared with KI [166], and was found to reduce interferences in both AAS [167] and DC plasma atomic emission spectrometry [168]. Antimony. Antimony is commonly determined by AAS after reduction to Sb(II1) with L-cysteine [ 1641 or HC1-KI-ascorbic acid [ 1211 and hydride generation. Speciation of Sb(II1) and Sb(V) is accomplished by complexing Sb(V) with citrate to prevent hydride formation, followed by antimony-hydride formation and detection by AAS [169]. Multiple Metals. Multiple metal ions have been determined using FIA preconcenhation techniques and flame AAS. For example, a cation exchange column packed with Si02-bonded octadecyl groups was used to collect the diethyl ammonium diethylthiocarbamate-metal (Cd, Pb, Cu) complexes [I 701. Collection occurred over a 20-s period followed by elution with methanol or ethanol into the flame. Electrothermal AAS has also been used with flow injection preconcentration techniques to determine Cd, Pb, Ni, and Cu [171]. Also, see Section 8.4.13 for a discussion of other multicomponent flow analysis techniques, such as photodiode array detection following derivitization with a non-specific spectrophotometric complexing agent. 8.5.1.3 Nonmetals
Boron. A FIA method for the determination of boron in water with in-line preconcentration on Amberlite IRA-743 and spectrophotometric detection of the azomethine-Hi-boron complex [172]. The method was applied to various natural waters and the detection limit was 1 pg B per liter with no significant interferences. Silicon. Most automated methods for silicon involve the molybdenum blue spectrophotometric procedure. Interference from molybdophosphoric acid complex was reduced with oxalic acid addition [173]. Another method involved formation of the molybdosilicate complex followed by ion-pairing with Rhodamine B [1741. Selenium. Flow injection atomic absorption spectrometry with hydride generation is commonly used for selenium. Selenium was also determined at 1 pg Se per liter after preconcentration and measurement of the selenium catalyzed reduction spectrocphotometric reagent [175]. Methods for the determination of selenium in waters and soils were recently reviewed [176]. 8.5.1.4 Actinide Elements
Thorium. Thorium was determined fluorometrically after FIA preconcentration [177].
298
8 Advances in Flow Analysis - Instrumentation and its Application
Uranium. A spectrophotometric FIA procedure for uranium was reported which involved preconcentration on Dowex 50-X8 and reaction with Arsenazo I11 in HC1 stabilized with Triton X-100 [178]. Uranium(V1) in waters was also determined fluorometrically following preconcentration on an activated silica gel [ 1791. 8.5.1.5 Anions
Sulfur Anions. Several automated spectrophotometric methods for sulfate analysis based on exchange of sulfate with a detectable species have been reported. For example, sample was passed through a column of regenerated cellulose containing particles of chloranilate. Sulfate was exchanged for the chloranilate which was detected at 530 nm [180]. Sulfate was also determined after preconcentration on a strong anion exchange column and spectrophotometric measurement of the complex with methylthymol blue and barium [181]. A FIA procedure for sulfide using Brilliant Green was reported with a range of 40 - 2000 pg S per liter [ 1821. Sulfite in brine was determined spectrophotometrically after reaction with 5,5'-dithiobis(2-nitrobenzoic acid) [183]. Nitrogen Anions. Nitrate and nitrite are most commonly determined by reduction of nitrate to nitrite followed by spectrophotometric measurement of a derivatized product. The nitrate concentration is calculated based upon the difference in results from the nitrate+nitrate analysis minus the nitrite analysis. Nitrate has also been measured by direct UV spectrophotometry with photodiode array detection and chemometric data analysis, Recent automated methods include the use of online-UV-induced photooxidation in K2S20s solution for the determination of total soluble nitrogen by FIA with direct spectrophotometric detection at 226 nm of the NO, formed [99]. Also, nitrate, nitrite, and ammonium were determined by an FIA method where nitrate and nitrite were reduced to ammonia which was detected conductometrically [184]. A simultaneous procedure for nitrite and nitrate in water using flow injection biamperometry was reported [185]. Phosphorus Anions. An FIA system was optimized for low level phosphorus determination in natural waters [96]. Using the tin(I1) chloride/acidic ammonium molybdate method and a detector consisting of a super-bright red light-emitting diode, a photodiode, and a 20-mm pathlength flowcell, phosphorus was measured to 1 v g P per liter. With addition of a preconcentration column of Bio-Rad AG 1-X8 anion-exchange resin, the detection limit became 0.1 pg P per liter. Another method used alcohol to accelerate the formation of the complex of molybdophosphoric acid and malachite green. Phosphate was determined spectrophotometrically at 650 nm and the working range was 1-5000 Fg P per liter [186]. FIA methods were also developed based upon reaction of molybdophosphate with Crystal Violet [I871 and Rhodamine B [188]. Phosphate was also determined indirectly by introducing samples into a CFA system containing Pb(I1) [189]. The decrease in free Pb(I1) was measured using a lead-selective electrode and this was related to the phosphate concentration. A FIA system with spectrophotometric detection was used to determine
8.5 Environmental Application of Flow Anafysis
299
simultaneously phosphate and silicate as their heteromolybdic acids [ 1901. Measurements are based on the color of the ion-pairs formed between the heteromolybdic acids and Rhodamine B. A flow injection system for the determination of ortho- and pyrophosphates in natural and waste waters was described [ 1911. Orthophosphate was measured down to 0.3 1.18P per liter after preconcentration on an organotin modified support, followed by elution, spectrophotometric detection of the reduced 12-molybdophosphoric acid product at 660 nm. To measure pyrophosphate, an initial on-line hydrolysis to orthophosphate occurs at 50 "C and at neutral pH with the addition of inorganic pyrophosphatase. Halides. Ion-selective electrodes have been used with flow analyzers for the measurement of halides. Recent work includes the determination of bromide in seawater with standards prepared in NaCl solutions [I921 and ISEs constructed from pressed pellets of Ag2S/AgX (X = C1, Br, or I) [193]. Chloride continues to be measured indirectly by reaction with Hg(SCN)2 followed by complexation of thiocyanate with Fe(II1). While this is the approved standard method for chloride measurement in many countries, there is a problem since the waste generated contains mercury. A non-mercurimetric chloride method was developed based upon oxidation to chlorine followed by gas-diffusion with amperometric detection [1941. This procedure was also used for the determination of bromide after oxidation to bromine [195]. Chloride was determined indirectly by precipitation as AgC1, which was then dissolved with ammonia, sodium thiosulfate, or potassium cyanide solution prior to analysis by AAS [196]. Chloride ions in natural and wastewaters was also determined by FIA using a column packed with silver chloranilate powder [197]. The chloranilate ion concentration released was directly proportional to the chloride ion concentration and was monitored at 530 nm. There were no interferences from anions and interferences from multicharged cations were eliminated by use of a cation-exchange column inserted in the upper stream of the reaction column. A FIA method for iodate and iodide in seawater based on their catalytic effect on the fading of an indicator formed from the reaction of iron(II1) thiocyanate and nitrite [I981 was described. Iodate is determined with an anion exchange column in-line, total iodine is determined without the anion exchange column, and iodide is calculated by the difference. Carbonate and Bicarbonate. Dissolved inorganic carbon was determined by FIA conductiometrically [ 1991 and with a C02-sensitive gas electrode [200]. Carbonate was determined indirectly by precipitation as CaC03, which was then dissolved with HC1 solution prior to analysis by AAS [196]. Cyanide. Several spectrophotometric FIA methods for cyanide have been reported. For example, cyanide was determined based on its reaction with sodium isonicotiafter gas diffusion through a silicone rubnate-3-methyl-1-phenyl-2-pyrazolin-5-one ber membrane [ZOI].Cyanide was also determined amperometrically after gas diffusion at a silver wire working electrode [202]. Total cyanide methods involving continuous-flow on-line distillation have recently been improved so that thiocyanate is not decomposed by the UV digestion procedure [103, 1041. Cyanide was also determined by FIA with piezoelectric crystal detection [203]. Cyanide-containing samples react
300
8 Advances in Flow Analysis - Instrumentation and its Application
with the metallic gold electrode of a piezoelectric crystal. This results in the formation of a soluble dicyano-gold complex and the loss in gold is detected as a resonant frequency change by the piezoelectric crystal. 8.5.1.6 Gases Ozone. Spectrophotometric [204] and chemiluminescence based [205] methods for the determination of ozone have been reported recently. Chlorine. Free and combined chlorine were determined in water by FIA using N,Ndiethyl-p-phenylenediamineover the 0.1 -8.0 mg of Clz per liter range [206]. Another FIA method for the determination of free chlorine in the presence of other chlorine species was reported [207]. It is based on the reaction of free chlorine with 4-nitrophenylhydrazine and N-(1-naphthyl)ethylenediamine dihydrochloride with detection at 532 nm. Sulfur Dioxide. Piezoelectric quartz crystal detectors with several coating materials sensitive to sulfur dioxide were tested for optimal response to SO, in the gas phase in a flow injection system [63].
Ammonia. Low concentrations of ammonia in the atmosphere were determined by FIA using a spectrophotometric reagent (salicylate and dichloroisocyanurate) 12081. Different ammonia sensitive coatings were tested on piezoelectric quartz crystals in a flow injection system [63]. Hydrogen Cyanide. HCN was separated from a mixture containing HCN and H2S by two-stage membrane differentiated FIA [64]. Sample gases were passed through the outside of a membrane tube of porous poly(viny1idene difluoride), through which flowed 50 mm NaOH. After absorption of acidic gases, the pH of the solution was reduced to 9.6 by addition of a carbonate buffer. The solution then flowed along the outside of a silicone rubber membrane tube containing 25 mm NaOH in the stopped-flow mode. Under these conditions, HCN but not H2S diffused into the alkaline receptor. The HCN was determined spectrophotometrically.
8.5.2 Organic Analysis While the vast majority of flow analysis applications have been for ion analysis, depending on the field of application there are also many methods for the analysis of organic compounds. In environmental analysis, the use of flow analysis for organic analysis continues to be an area of interest as the number of tests required increases, but at the same time, the costs must be minimized. There have been two recent reviews on the current and future application of FIA for the analysis of organic compounds in seawater [I301 and freshwater [209]. On-line digestion procedures for
8.6 Quality Assurance Aspects of Flow Analysis
301
organic nitrogen, organic carbon, and organic phosphorus were described in Section 8.4.10. Aldehydes in fog water samples were determined by FIA-post column derivitization with 3-methyl-2-benzothiazolinone hydrazone following HPLC separation [210]. Also, an FIA method for formaldehyde based on its inhibition of the Brilliant Green-sulfite reaction at pH 7 was described [21I]. Phenol is commonly determined by on-line distillation followed by reaction with 4-aminoantipyrene and spectrophotometric detection. The method does not speciate the various substituted phenol compounds and is sensitive to 2 pg phenol per liter. Preliminary studies have shown that on-line preconcentration on acid resistant CI8-reversedphase sorbent followed by elution with methanol/water and reaction with 4-aminoantipyrene can give detection limits to 0.5 pg phenol per liter. A FIA method was developed for the herbicides diquat and paraquat based upon their reduction with alkaline sodium dithionite [212]. Organophosphorus and carbamate insecticides were determined by FIA [213]. A flow injection method for carbamate pesticides was also developed in which the pesticides were hydrolyzed to their corresponding phenols and reacted with p-aminophenol under alkaline conditions in presence of KI04 [214, 2151. A series of procedures for formetanate, carbaryl, propoxur, and ethiofencarb in natural waters were developed. Anionic surfactants were determined by combination with the dyes Rhodamine B and 4-[[4-(dimethylamino)phenyl]azo]-2-methylquinoline under acidic conditions followed by solvent extraction and spectrophotometric and spectrofluorometric detection [216]. Cationic surfactants, such as quaternary ammonium salts, interact with Bromocresol Purple buffered at pH 8, resulting in a decrease in the absorbance at 588 nm [217]. In addition, anionic surfactants, such as sodium dodecyl sulfate, also depressed the interaction. A flow injection procedure for the determination of commercially available fatty amine ethoxylate-based non-ionic surfactant in seawater was reported. The procedure is based on the measurement of the chemiluminescence emission resulting from oxidation of the tertiary amine group with NaOCl at pH 10.5 in the presence of Rhodamine B, which acts as a sensitizer [218].
8.6 Quality Assurance Aspects of Flow Analysis Flow analysis instruments are well suited for incorporation into analytical laboratory QA/QC programs. The increased number of analyses required within a QA/QC program result in an increased need to automate laboratory tests. Automated flow analyzers offer an economical alternative to analyze the increasing number of samples while at the same time increasing reproducibility and reducing reagent consumption and waste generated. The US EPA’s Contract Laboratory Program describes the ‘Required QC/QA Operations’ (Section V, ILMOl .O) for atomic absorption systems. This protocol has become widely used for flow analysis systems. The general guidelines of the protocol include the following.
302
8 Advances in Flow Analysis - Instrumentation and its Application
8.6.1 Calibration Standards These must be freshly prepared each time an analysis is performed and discarded after use. Prepare a blank and a minimum of three standards in the appropriate range and in the same matrix as the samples.
8.6.2 Initial and Continuing Calibration Verification Immediately after calibration and after every 10 samples, the accuracy of the instrument is verified with an independent standard. An independent standard is defined as a standard composed of the analyte from a different source than that used in the calibration standards. The independent standard concentration must be at or near the mid-range of the calibration curve. The calibration verification must reflect the analysis conditions of the samples and may not be treated differently. In practical terms, this means that a blank should not be run prior to the check standard. Independent standards are available from a variety of independent companies that offer certified reference and QC samples.
8.6.3 Calibration Blank A blank must be run immediately after every initial and continuing calibration verification. It must also be analyzed at the beginning of the run and after the last sample. In addition, matrix, sample, or digestion blanks may be incorporated into an analysis. This value may then be subtracted from the sample concentrations.
8.6.4 Spikes and Duplicates At least one spike and duplicate must be prepared from each group of samples of a similar matrix or sample type. Most software packages that operate modern flow analysis instrumentation include many of the QC protocol features. This includes the ability to define which sample cup contains the check calibrant, and alarms which alert the operator if the calibration verification falls outside of the acceptable range.
References
303
8.7 Conclusion As flow analysis approaches its 4th decade, it is clear that its use for environmental monitoring has been extensive. However, it is also clear that flow analysis has not been fully developed in terms of hardware, techniques, or applications. Significant improvements in flow analysis can be expected in three areas: (a) hybridization; (b) on-site/on-line analyzers, and (c) miniaturization. Considering the similarities in flow-based technologies, such as CFA, FIA, LC, FFF, and electrophoresis, further hybridization of techniques can be expected. Hyphenation to other analytical systems will continue to expand the use of flow analysis beyond automated wet chemistry to other areas within analytical chemistry. While most flow analyzers are used in laboratories, there will be a trend toward development of analyzers for on-site (field) or on-line continuous monitoring. On-site analyzers provide the ability of measuring samples immediately after they are collected. This is important for samples which are difficult to preserve and to obtain timely information. On-line analyzers can be used for continuous monitoring of sample streams. While this has primarily been considered from a chemical process control perspective, it can be equally valuble for monitoring environmental parameters. This monitoring can occur during a waste treatment process for monitoring and control purposes and for passive monitoring of remote locations. FIA is particularly attractive for on-site/on-line analyzers because only a small volume of sample is injected into the system and the rapid analysis time. Finally, flow analyzers have great potential for development as miniaturized chemical analyzershensors. All of the basic flow analyzer components have been designed on a miniature scale using microlithographic techniques [14]. Pumps, valves, conduits, mixers, and detectors have been developed and can be assembled of silicon wafers. Such analyzers could be used as in-situ chemical sensors or for on-site sample analysis.
References [l] Skeggs, L. T., A m J Clin Pathol 1957, 283, 3 11 - 322. [2] Ruzicka, J., Hansen, E. H., Flow Injection Analysis, 2nd edition, New York: Wiley, 1988. [3] Furman, W. B., Continuous Flow Analysis: Theory and Practice. New York: Marcel Dekker, 1976. [4] Technicon Instruments Corporation, Technicon Autoanalyzer Bibliography, 195711967, New York, 1976. [5] Technicon Instruments Corporation, Technicon Bibliography, 196711973, New York, 1974. [6] Technicon Instruments Corporation, Technicon Bibliography Supplement # I, New York, 1974. [7] Technicon Instruments Corporation, Technicon Bibliography Supplement # 2, New York, 1975. [8] Technicon Instruments Corporation, Technicon Symposia Proceedings 1963/1976, Media Inc., New York, 1977. [9] Tecator AB, Flow Injection Bibliography, HBganBs, Sweden, 1987.
304
8 Advances in Flow Analvsis - Instrumentation and its Amlication
[lo] Valcarcel, M., Luque de Castro, M. D., Flow Injection Analysis-Principles and Applications. New York: Wiley, 1987. [I 11 MacCarthy, P., Klusman, R. W., Cowling, S. W., Rice, J. A., Anal Chem 1993,65,244R-292R. [ 121 Fang, Z. L., Flow-Injection Separation and Pre-Concentration, Weinheim: VCH, 1993. [13] MacLaurin, P., Worsfold, P. J., Microchem J 1992, 45, 178- 188. [14] Manz, A., Graber, N., Widmer, H.M., Sensors and Actuators B (Chemical) 1990, BI(I-6), 244- 288. [I51 Tyson, J.F., Anal Chim Acta 1990, 234, 3-12. [16] Burguera, J. L., Flow Injection Atomic Spectroscopy, New York: Marcel Dekker, 1989. [17] Ruzicka, J., Marshall, G.D., Anal Chim Acta 1990, 237, 329-343. [18] Gubeli, T., Christian, G.D., Ruzicka, J., Anal Chem 1991, 63, 2407-2413. [19] Baron, A., Guzman, M., Ruzicka, J., Christian, G.D., Analyst (London) 1992, 117, 1839- 1844. [20] Berman, R. J., Renn, C. N., Arment, S.F., A m Lab 1991, 23, 23-32. [21] Gardner, W. S., Malczyk, J. M., Anal Chem 1983, 55, 1645- 1647. 1221 Li-Chiang, T., Xiao-Ping, S., Yi-Yun, Z., Zheng-Liang, Z., Anal Chim Acta 1990, 238, 183- 190. [23] Ruzicka, J., Chemometrics and Intelligent Lab. Systems 1990, 8, 281. [24] Ruzicka, J., Christian, G. D., Analyst 1990, 115, 475 -486. [25] Engelhardt, H., Klinkner, R., Schoendorf, G., J Chromatogr 1990, 535, 41 -53. 1261 Frei, R. W., Jansen, H., Brinkman, V. A.T., Anal Chem 1985, 5715, 529A- 1539A. [27] Swartz, M. K., in: Continuous Flow Analysis: Theory and Practice; Furman, W.B. (ed.) New York: Marcel Dekker, 1976; p. iii. (281 Patton, C. J., Crouch, S. R., Anal Chim Acta 1986, 179, 189-201. [29] Methods for Chemical Analysis of Water and Wastewater. March 1984, EPA-60014-79-020, Phosphorus, All Forms, Method 365.1, US Environmental Protection Agency, Washington, D. C. [30] Ruzicka, J., Hansen, E. H., Anal Chim Acta 1975, 78, 145. [31] Karlberg, B., Pacey, G. E., Flow Injection Analysis: A Practical Guide, Amsterdam: Elsevier, 1989. [32] Sson, J.F., Microchemical J 1992, 45, 143-162. [33] Fang, Z., Microchemical J 1992, 45, 137-142. [34] Chen, D., Laboratory Robotics and Automation 1991, 3, 125- 132. [35] Bilitewski, U., Ding, T., Schmid, R.D., Angewandte Chemie (Weinheim) 1991, 103, A105-A107, A110, A112. [36] Valcarcel, M., Luque de Castro, M.D., Fres J Anal Chem 1990, 337, 662-666. [37] Fang, Z., Sun, L., Xu, S., Anal Chim Acta 1992, 261, 557-576. [38] Zolotov, Y. A., Shpigun, L.K., Microchemical J 1992, 45, 225-231. [39] Luque de Castro, J.D., Valcarcel, M., Intern J Environ Anal Chem 1990, 38, 171-183. (401 Nydahl, F., Talanta 1976, 23, 349-357. [41] Ivaska, A., Ruzicka, J., Analyst, in press. [42] Straka, M.R., A m Environ Lab 1989, 21, 60-63. 1431 Stockwell, P. B., J Autom Chem 1990, 12, 95 - 103. [44] Burke, E.M., Suarez, EX., Hillman, D.C., Heithmar, E.M., Water Research 1989, 23, 519 - 521. [45] Patton, C. J., Alex, P. A., Conttrell, G. L., 20th Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies, Detroit, Michigan, October 1993, Abstracts. [46] Hemmings, P., Anal Proc (London) 1989, 26, 133- 134. 1471 Wang, J., Microchemical J 1992, 45, 219-224. [48] Ruzicka, J., Marshall, G.D., Christian, G. D., Anal Chem 1990, 62, 1861- 1866. [49] Van der Linden, W.E., Anal Chim Acta 1986, 176, 91-101. [50] Malcolme-Lawes, D. J., Pasquini, C., J Autom Chem 1988, 10, 25 - 30. [51] Shaorong, L., Dasgupta, P.K., Anal Chim Acta 1992, 268, 1-6.
References
305
Manz, A., Harrison, D. J., Fettinger, J. C., Verpoorte, E., Ludi, H., Widmer, H.M., Transducers '91, I991 International Conference on Solid-state Sensors and Actuators, Digest of Technical Papers (No. 91CH2817-5), New York: IEEE: 1991. Johnson, K. S., Sakamoto-Arnold, C. M., Willason, S. W., Beehler, C. L., Anal Chim Acta 1987, 201, 83. Gustavsson, A., Trends Anal Chem 1989, 8, 336-338. Ridder, C., Norgaard, L., Chemometrics and Intell Lab Systems 1992, 14, 297-303. Blanco, M., Coello, J., Itturriaga, H., Maspoch, S., Redon, M., Riba, J., Anal Chim Acta 1992, 259, 219-224. Cladera, A., Gomez, E., Estela, J.M., Cerda, V., Alvaarez-Ossorio, A., Rincon, F., Salva, F., Intern J Environ Anal Chem 1991, 45, 143- 152. Whitman, D. A., Seasholtz, M.B., Christian, G. D., Ruzicka, J., Kowalski, B. R., Anal Chem 1991, 63, 775-781. Hoogvliet, J. C., Reijn, J.M., Van Bennekom, W.P., Anal Chem 1991, 63, 2419-2423. Cobben, P. L. H.M., Egberink, R. J. M., Bomer, J., Haak, J.R., Berveld, P., Reinhoudt, D. N., Sensors and Actuators B (Chemical) 1992, B6, 304-307. Alegret, S., Bartroli, J., Jimenez-Jorquera, C., del Valle, M., Dominguez, C., Esteve, J., Bausells, J., Sensors and Actuators, B (Chemical) 1992, B7, 555 - 560. Nagashima, K., Kamaya, M., Ishii, E., Sensors and Actuators, B (Chemical) 1992, B9, 149-154. Trojanowicz, M., vel Krawczyk, T. K., Sensors and Actuators, B: (Chemical) 1992, B9, 33 - 39. vel Krawczyk, T.K., Trojanowicz, M., Anal Sciences 1992, 8, 329-335. Kester, M. D., Wade, A. P., 20th Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies, Detroit, Michigan, October 1993, Abstracts. Patton, C. J., Wade, A. P., in: AnalyticalInstrumentation Handbook: Ewing, G. W. (ed.) New York: Marcel Dekker, 1990, pp. 979-1051. Brown, M.W., Ruzicka, J., Analyst 1984, 109, 1091. Cui, F., Hansen, E. H., Ruzicka, J., Anal Chim Acta 1985, 169, 209. McIntosh, S., Fernandez, F. J., Guo, T., Instrument Solutions 1991, March Issue, 22, 24-25. Luque de Casto, M.D., Valcarcel, M., J Chromatogr 1992, 600, 183- 188. Chambaz, D., Haerdi, W., J Chromatogr 1992, 600, 203-210. Johnson, K. S., Petty, R.L., Anal Chem 1982, 54, 1185. Luque de Castro, M.D., Valcarcel, M., Trends Anal Chem 1986, 5, 71. Ruz, J., Rios, A., Luque de Castro, M. D., Valcarcel, M., Fres Z Anal Chem 1985, 322, 499. Christian, G. D., Ruzicka, J., Anal Chim Acta 1992, 261, 11 -21. Kalberg, B., Thelander, S., Anal Chim Acta 1978, 98, 1. Bergamin Filho, H., Medeiros, J. X., Reis, B. F., Zagatto, E. A. G., Anal Chim Acta 1978, 101, 17. Motomizu, S., Oshima, M., Kuroda, T., Analyst 1987, 112, 1233. Martinez-Jiminez, P., Gallego, M., Valcarcel, M., Anal Chim Acta 1988, 215, 233. Audunsson, G., Anal Chem 1986, 58, 2602. Seifler, E., Kambosos, K., Chanas, A., in: Advances in Automated Analysis, Technicon International Congress 1970, Vol I: Mount Kisco, NY: Futura, 1972; pp. 509-51 1. Van der Linden, W.E., Analyst 1983, 108, 1013. Van Son, M., Schothorst, R.C., Den Boef, G., Anal Chim Acta 1983, 155, 89. Zhu, Z., Fang, Z., Anal Chim Acta 1987, 198, 25. Pacey, G.E., Straka, M. R., Cord, J. R., Anal Chem 1986, 58, 502. Pacey, G. E., Hollowell, C. L., Miller, K. G., Straka, M. R., Gordon, G., Anal Chim Acta 1986, 179, 259. Yamamoto, M., Takada, K., Kamamaru, M., Yasuda, M., Yokoyama, S., Anal Chem 1987,59, 2446 - 2448. Hanna, C.P., %son, J.F., McIntosh, S., Anal Chem 1993, 65, 653-656. Martin, G.B., Meyerhoff, M.E., Anal Chem 1987, 59, 2345-2350.
306
8 Advances in Flow Analysis
-
Instrumentation and its Application
[90] Obato, H., Karatani, H., Nakayama, E., Anal Chem 1993, 65, 1524-1528. [91] Luque de Castro, M. D., Trends Anal Chem 1992, 11, 149- 155. [92] Frei, R. W., Zech, K. (eds.) Selective Sample Handling and Detection in High-Performance Liquid Chromatography, Amsterdam: Elsevier; 1988. [93] Welz, B., Microchem J 1992, 45, 163- 177. [94] Wang, X., Barnes, R. M., J Anal At Spectrom 1989, 4, 509. [95] Shah, A., Devi, S., Anal Chim Acta 1990, 236, 469-473. [96] Freeman, P. R., McKelvie, I.D., Hart, B.T., Cardwell, T. J., Anal Chim Acta 1990, 409-416. [97] Lacy, N., Christian, G. D., Ruzicka, J., Anal Chem 1990, 62, 1482- 1490. [98] Martinez-Jiminez, P., Gallego, M., Valcarcel, M., Analyst 1987, 112, 1233. [99] Novozamsky, I., Van Eck, R., Van der Lee, J. J., Houba, V. J.G., Ayaga, GO., At Spectrosc 1988, 9, 97. [loo] Hinkamp, S., Schwedt, G.Z., Wasser Abwasser Forsch 1991, 24, 60-65. [lo11 Edwards, R.T., McKelvie, I.D., Ferett, P. C., Hart, B. T., Bapat, J. B., Hoshy, K., Anal Chim Acta 1992, 261, 287 - 294. [lo21 Methods f o r Chemical Analysis of Water and Wastewater, March 1984, EPA-600/4-79-020, Cyanide, Total, Automated UV, Method 335.3, US Environmental Protection Agency, Washington, DC. [lo31 Meeussen, J.C.L., Keizer, M.G., Lukassen, W.D., Analyst 1992, 117, 1009-1012. [lo41 Berman, R. J., Renn, C.N., Christmann, D., A m Environ Lab 1993, 5, 32-34. [lo51 Krull, I. S., Reaction Detection in Liquid Chromatography, New York: Marcel Dekker, 1986. [lo61 Tsalev, D. L., Sperling, M., Welz, B., Analyst (London) 1992, 117, 1729- 1733. [lo71 Hinkamp, S., Schwedt, G., Anal Chim Acta 1990, 236, 345-350. [I081 Balconi, M.L., Borgarello, M., Ferraroli, R., Realini, F., Anal Chim Acta 1992, 261, 295 - 299. [lo91 Lazaro, F., Luque de Castro, M.D., Valcarcel, M., Anal Chim Acta 1991, 242, 283-289. [110] Tian, L., Sun, X., Xu, Yi., Zhi, Z., Anal Chim Acta 1990, 238, 183-190. [ 1111 Craig, J. M., Beauchemin, D., 20th Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies, Detroit, Michigan, October 1993, Abstracts. [112] Yamamoto, M., Yasuda, M., Yamamoto, Y., Anal Chem 1985, 57, 1382- 1385. [113] Tian, L.C., Wu, S.M., Anal Chim Acta 1992, 261, 301-305. [114] Isreeli, M., Pelavin, M., Kessler, G., A n n NY Acad Sci 1960, 87, 636-649. [115] Bailey, B. W., Lo, F. C., Anal Chem 1971, 43, 1525- 1526. [116] Jackson, K.W., Qiao, H., Anal Chem 1992, 64, 50R-66R. [117] Fang, Z., Welz, B., Schlemmer, G., J Anal A t Spectrom 1989, 4, 91. [118] Davey, D.E., Metz, G. J.H., J Anal A t Spectrom 1988, 3, 1988. [119] De la Calle-Guntinas, M., Madrid, Y., Camara, C., Anal Chim Acta 1991, 252, 161- 166. [120] Sperling, M., Yin, X., Welz, B., Analyst (London) 1992, 117, 629-635. [121] Tanaka, H., Yamamoto, E., Mortia, H., Shimomura, S., Anal Sci 1992, 8, 93-95. [122] Ambrose, A. J., Ebdon, L., Foulkes, M.E., Jones, P. J Anal A t Spectrom 1989, 4, 219. [123] Al-Swaidan, H. M., Lacy, N., Christian, G. D., Anal Lett 1989, 22, 2653 -2663. [I241 Thomspon, J. J., Houck, R.S., Anal Chem 1986, 58, 2541. [I251 Nord, L., Johansson, S., Brotell, H., Anal Chim Acta 1985, 175, 281 -287. [126] Morgan, D.K., Danielson, N.D., Keaton, J.E., Anal Left 1985, 18, 1979. [127] Ridder, C., Norgaard, L., Chemonometrics and Intel1 Lab Systems 1992, 14, 297-303. [128] Jayaweera, P., Ramaley, L., Anal Chem 1989, 61, 2102-2108. [129] Mee, L.D., Sci Total Environ 1986, 49, 27-87. [130] Wardencki, W., Fres J Anal Chem 1991, 340, 207-212. [131] Shpigun, L.K., Kolotyrkina, I.Ya., Zolotov, YuA., Anal Chim Acta 1992, 261, 307-314. [132] Jodo, M., Kawamoto, J., Tochimoto, M., Coverly, S.C., J A u t o m Chem 1992, 14, 164-167. [133] Van Staden, J.F., Water S A 1989, 15, 153-168. [134] Downard, A. J., Hart, J.B., Powell, H.K. J., Shuangua, X., Anal Chim Acta 1992, 269, 41 -48.
References
307
[135] Motomizu, S., Yoshida, K., Toei, K., Anal Chim Acta 1992, 261, 225-231. [136] Korenaga, T., Izawa, M., Ito, T., Takahashi, T., Bunseki Kagaku 1990, 39, 711-716. [137] Yoshida, K., Motomizu, S., J Flow Injection Anal 1990, 7, 130- 137. [138] Filha, M. M. S., Freire dos Reis, B., Bergamin Filho, H., Baccan, N., Anal Chim Acta 1992, 261, 339-343. [139] Muraki, H., Higuchi, K., Sasaki, M., Korenaga, T., Toei, K., Anal Chim Acta 1992, 261, 345 - 349. [140] Application Note: Ammonium-Nitrogen in Seawater, Perstorp Analytical Environmental, Wilsonville, Oregon, 1992. [141] Frenzel, W., Cheng-Yu, L., Fres J Anal Chem 1992, 342, 276-280. [142] Alegret, S., Alonso, J., Bartoli, J., Marinez-Fabregas, E., Analyst 1989, 114, 1443- 1447. [143] Kutsera, L., Ghaziaskar, H., Lai, E.P.C., Anal Lett 1992, 25, 2289-2304. [144] Shah, A., Devi, S., Anal Chim Acta 1990, 236, 469-473. [145] Sperling, M., Xu, S., Welz, B., Anal Chem 1992, 64, 629-635. [146] Chapin, T.P., Johnson, K.S., Coale, K.H., Anal Chim Acta 1991, 249, 469-478. [147] Resing, J.A., Mottl, M. J., Anal Chem 1992, 64, 2682-2687. [148] Nakano, S., Nozawa, M., Yanagawa, M., Kawashima, T., Anal Chim Acta 1992,26,183 - 188. [I491 Jiang, Z., Anal Chim Acta 1992, 260, 45-49. [150] Obata, H., Karatani, H., Nakayama, E., Anal Chem 1993, 65, 1524-1528. [I511 Sperling, M., Yin, X., Welz, B., JAnal At Spectrom 1991, 6, 614-622. [152] Sakamoto-Arnold, C.M., Johnson, K.S., Anal Chem 1987, 59, 1789- 1794. [153] Motomizu, S., Korechika, K., Nippon Kagaku Kalshi 1991, 6, 795-802. [154] Wang, X., Fenxi Shiyanshi 1991, 10, 7- 10. [155] Fernandez, P., Perez, C., Gutierrez, A., Camara, C., Fres JAnal Chem 1992,342, 597-600. [156] Hanna, C.P., Tyson, J.F., McIntosh, S., Anal Chem 1993, 65, 653-656. [157] Korenaga, T., Izawa, M., Noguchi, H., Takahashi, T., Bunseki Kagaku 1992, 4, 17-22. [I581 Munaf, E., Takeuchi, T., Ishii, D., Haraguchi, H., Anal Sci 1992, 342, 154-156. [159] Mariscal, M., Galban, J., Urate, M., Aznarez, J., Fres J Anal Chem 1992, 342, 157- 162. [160] Benson, R., Worsfold, P., Sweeting, F., Anal Chim Acta 1990, 238, 177-182. [161] Clarke, N., Danielsson, L., Sparen, A., Intern J Environ Anal Chem 1992, 48, 77- 100. [162] Rogeborg, E. J.S., Henrikson, A., Vatten 1985, 41, 48-53. [I631 Fernandez, P., Perez, C., Gutierrez, A., Camara, C., Talanta 1991, 38, 1387-1392. [I641 Frenzel, W., Bratter, P., Anal Chim Acta 1986, 179, 389. [165] Driehaus, W., Jekel, M., Fres J Anal Chem 1992, 343, 352-356. [ 3 661 Welz, B., Sucmanova, M., 20th Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies, Detroit, Michigan, October 1993, Abstracts. [167] Brindle, I., Alarbi, H., Karshman, S., Le, X., Zheng, S., Chen, H., Analyst (London) 1992, 117, 407 - 41 1. [168] Chen, H., Brindle, I., Le, X., Anal Chem 1992, 64, 667-672. [ 1691 Mohammed, B., Ure, A., Reglinski, J., Littlejohn, D., Chem Speciation Bioavailability 1990, 2, 117-122. [170] Fang, Z., Guo, T., Welz, B., Talanta 1991, 38, 613-619. [171] Sperling, M., Yin, X., Welz, B., J Anal At Spectrom 1991, 6, 295-300. [I721 Sekerka, I., Lechner, J.F., Anal Chim Acta 1990, 234, 199-206. [173] Von Standen, J., Pauer, J., Van Vliet, H., Kempster, P., Water SA 1990, 16, 205-210. [174] Mas, F., Estela, J., Cerda, V., Anal Chim Acta 1990, 239, 151-155. [175] Ayoama, E., Akamatsu, K., Nakagawa, T., Tanka, H., Anal Sci 1991, 7, 617-621. [176] Potin-Gautier, M., Astruc, M., Vermeulin, P., Analusis 1991, 19, m57-m62. [177] Shen, Z., Zhang, Y.,Guo, D., Huang, Y., Fenxi f f u a m e 1991, 19, 1075-1077. [1781 Perez-Pavon, J., Garcia-Pinto, C., Rodriguez-Garcia, E., Moreno-Cordero, B., Anal Chim Acta 1992, 264, 291 -296. [179] Havel, J., Vrchlabsky, M., Hohn, Z., Talanta 1992, 39, 795-799.
308
8 Advances in Flow Analysis - Instrumentation and its Application
[I801 Ueno, K., Sagara, F., Higashi, K., Yakata, K., Yoshida, I., Ishii, D., Anal Chim Acta 1992, 261, 241 -245. [181] Karlsson, M., Person, J.A., Moeller, J., Anal Chim Acfa 1991, 244, 109-113. [182] Ensafi, A.A., Anal Lett 1992, 25, 1525-1543. [183] MacLaurin, P., Parker, K. S., Townshend, A., Worsfold, P. J., Barnett, N. W., Crane, M., Anal Chim Acta 1990, 238, 171- 175. I1841 Cardoso de Faria, L., Pasquini, C., Anal Chim Acta 1991, 245, 183-190. [185] Trojanowicz, M., Mauszewski, W., Szostek, B., Michalowski, J., Anal Chim Acta 1992, 261, 391 -398. [186] Motomizu, S., Yasuda, Y., Oshima, M., Nippon Kagaku Kaishi 1991, 12, 1624- 1631. [I871 Burns, D.T., Chirnpalee, D., Chimpalee, N., Ittipornkul, S., Anal Chim Acta 1991, 254, 197- 200. [I881 Mas, F., Estel, M., Cerda, V., Water Air Soil Pollut 1990, 18, 531 -535. [I891 Hara, H., Kusu, S., Anal Chim Acta 1992, 261, 41 1-417. [190] Mas, F., Estela, J.M., Cerda, V., Intern J Environ Anal Chem 1991, 43, 71 -78. [191] Spivakov, B.Ya., Maryutina, T.A., Shpigun, L. K., Shkinev, V. M., Zolotov, Yu.A., Ruseva, E., Havezov, I., Talanta 1990, 378, 889-894. [192] Ilcheva, L., Yanakiev, R., Cammann, K., JZndian Chem SOC1991, 68, 154-157. [I931 .Najib, EM., Othman, S., Talanta 1992, 39, 1259-1267. I1941 Nikolic, S. D., Milosavljevic, E. B., Hendrix, J. L., Nelson, J. H., Analyst (London) 1991, 116, 49 - 52. [195] Nikolic, S. D., Jankovic, T. D., Milosavljevic, E. B., Hendrix, J. L., Nelson, J. H., Fres J Anal Chem 1992, 342, 98- 102. [196] Esmadi, F.T., Kharoaf, M.A., Attiyat, A.S., Talanta 1990, 37, 1123-1128. [I971 Sagara, F., Tsuji, T., Yoshida, I., Ishii, D., Ueno, K., Anal Chim Acta 1992, 270, 217-221. [198] Oguma, K., Kitada, K., Kuroda, R., Mikrochimica Acta 1993, 110, 71 -77. [199] Jardim, W.F., Guimaraes, J.R., Allen, H.E., Cienc Cult (Sao Paulo) 1991, 43, 454-456. [200] Hara, H., Okabe, Y., Kitagawa, T., Anal Chem 1992, 64, 2393-2397. [201] Kuban, V., Anal Chim Acta 1992, 259, 45-52. [202] Nikolic, S. D., Milosavljevic, E. B., Hendrix, J. L., Nelson, J. H., Analyst (London) 1992, 117, 47 - 50. [203] Bunde, R.L., Rosentreter, J. J., Microchem J 1993, 47, 148-156. [204] Onari, Y., Anal Sci 1991, 7, 305-306. [205] Chung, H.K., Bellamy, H. S., Dasgupta, P. K., Talanta 1992, 39, 593-598. [206] Gordon, G., Sweetin, D.L., Smith, K., Pacey, G.E., Talanta 1991, 116, 145-149. [207] Verrna, K. K., Jain, A., Townshend, A., Anal Chim Acta 1992, 261, 233-240. [208] Bristow, A. W., Comm Soil Sci Plant Anal 1991, 22, 1741- 1752. [209] Bao, L., Dasgupta, P. K., Anal Chem 1992, 64, 991 -996. [210] Suzaki, H., Igawa, M., Anal Sci 1991, 7, 133-134. [211] Safavi, A., Ensafi, A.A., Anal Chim Acta 1991, 252, 167-171. [212] Perez-Ruiz, T., Martinez-Lozano, C., Tomas, V., Intern J Environ Anal Chem 1991, 44, 243 -252. [213] Kurnaran, S., Tran-Minh, C., Anal Biochem 1992, 200, 187- 194. [214] Khalaf, K.D., Morales-Rubio, A., Sancenon, J., De la Guardia, M., 20th Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies, Detroit, Michigan, October 1993. Abstracts. [215] Khalaf, K.D., Sancenon, J., De la Guardia, M., Anal Chim Acta 1992, 266, 119-126. [216] Motomizu, S., Koayashi, M., Anal Chim Acta 1992, 261, 471 -475. [217] Yarnarnoto, K., Motomizu, S., Anal Chim Acfa 1991, 246, 333-339. [218] Lancaster, J.S., Worsfold, P. J., Lynes, A., Anal Chim Acta 1990, 239, 189-194.
9 Application of Software in Environmental Auditing and Quality Control Edward So0 and Miles Jack
9.1 Environmental Auditing Environmental auditing is a management tool to determine the compliance of an organisation with some form of pre-set standard. Audits of many types are now becoming common place in industry (financial, safety, quality, environmental, etc). The reason for this is because an audit can provide a true picture of the state of the organisation’s activities prior to something going wrong. This proactive approach to seeking out problems can pay financial dividends. In the present climate, more and more severe penalties can be imposed on organisations for breaches of regulations and standards. This is nowhere more true than in the case of environmental standards, where the flow of legislation is ever increasing. The environmental audit becomes an invaluable tool in planning future strategies in line with current and planned regulations. The International Chamber of Commerce defines environmental auditing as ‘Auditing is a management tool comprising a systematic, documented, periodic and objective evaluation of how well environmental organisation, management and equipment are performing with the aim of helping to safeguard the environment by: (1) facilitating management control of environmental practices (2) assessing compliance with company policies, which would include regulatory requirements’. An environmental audit, however, is more than just a procedure to prevent prosecution for breaches of environmental laws. In fact, there is no legal requirement to carry out environmental audits, or any other type of operational audit. Companies can carry out environmental audits for business reasons, because they can lead to increased business opportunities and help to eliminate costs. Obviously the moral question of safeguarding the environment also plays a part. Public organisations also carry out environmental audits in response to public pressure. There has been a recent upsurge in guidelines and standards directed at organisations of all types, designed to improve the environmental performance of all organisations whose operations may have an impact on the environment. These have ranged from schemes and guidelines introduced by trade and professional bodies to practices defined by legislation both national and international. Perhaps the best known scheme in Britain is the British Standard’s Institute’s ‘British Standard for Environmental Management Systems (BS7750)’. This standard,
310
9 Application of Software in Environmental Auditing and QC
introduced in 1992, lays down a framework to help organisations establish an effective environmental management system. It is similar in effect to BS5750, which deals with Quality Management. It takes the key elements of BS5750 and applies them to environmental management. As with BS5750, the intention is for organisations to apply for certification under the scheme, thus tying them into a commitment to implement environmental management techniques. It is to be expected that companies and organisations accredited under the scheme will be perceived by the public and the business community as being more ‘green’ than their competitors. This is what has happened to companies implementing BS5750 with regard to the quality of services offered. In the current climate of ‘green awareness’, this reaction can only be to the advantage of the concerned organisation. One of the mainstays of the standard is the concept of environmental auditing and within the standard the concept is taken further than merely ensuring that the organisation is complying with environmental legislation. It is defined as ‘A systematic evaluation to determine whether or nor environmental performance complies with planned arrangements and whether or not these arrangements are implemented effectively, and are suitable to fulfil the organisation’s environmental policy’. The audit is therefore used to determine whether the organisation’s objectives, as laid down in the environmental policy, are being met and whether specific environmental targets are being achieved. The audit then becomes much more than just a verification of compliance. It is used to determine whether specific targets are being met with the objective of continually improving standards by raising the targets as they are achieved. The audit then becomes the main management tool for assessing how environmental performance is improving. Certification under BS7750 should help organisations applying for registration under the European Communities’ ‘Eco-Audit’ scheme. This is a voluntary regulation developed by the EC to implement a community wide standard for environmental auditing. It is perhaps the first step on the road towards compulsory environmental audits for all organisations. This scheme requires organisations to register their operations by site rather than on a company wide basis. Once committed to the scheme the organisation must (1) establish an environmental protection system, (2) carry out an environmental audit of the system, (3) set objectives to improve environmental performance and (4) produce an environmental statement which must be validated by accredited environmental auditors and submitted to a competent authority for access by the public. The EC scheme sets out what should be examined in the audit, thereby setting a standard for conformity between all EC states. Registration under the scheme allows the organisation to display an ‘Eco-Audit’ logo. This will obviously provide a business advantage to organisations displaying the logo, especially in the public sector, where public opinion will favour ‘green’ companies.
9.2 The Mechanics of Auditinp
31 1
9.2 The Mechanics of Auditing 9.2.1 Why Audit? Audits can be time consuming, costly to develop and implement and are not, in themselves, productive. They require the upkeep of in-house expertise or the hiring of expensive consultants. They do not produce anything that can be marketed directly and may result in time consuming and costly recommendations. The question obviously arises of why organisations should carry out audits? The bottom line is that audits are carried out to ensure that an organisation is complying with all relevant regulations, thereby protecting themselves from prosecution. This in itself is a major incentive to carry out audits, with the rising costs of court cases and fines. Convictions for breaches of environmental law are increasing as are the number and complexity of the regulations themselves. Senior management even face imprisonment if convicted of breaking environmental regulations. The cost of a single conviction may far outweigh the cost of implementing an audit system. Another reason for carrying out audits, is if an organisation wishes to apply for registration under one of the current environmental management schemes. Auditing is specified as one of the requirements for certification. The advantages of registration should exceed the initial cost and inconvenience of setting up an auditing system. There may also be a moral or ethical reason for carrying out environmental audits. The potential for harm to the environment from modern industrial practices is immense. An environmental audit is one way of assessing the potential for harm to occur and for identifying measures to reduce the risk to the environment. However, the reasons for carrying out environmental audits go beyond merely checking compliance. There are a number of benefits to be gained from carrying out an audit. Potential benefits include: Assurance of compliance with existing laws and regulations. A check of the state of readiness for the implementation of future laws and practices. Reduction of the potential for prosecution under civil or criminal law. Comparison of achievements with competitors and other organisations. Determination of the technologies and practices that will be required in future to remain viable and competitive. Reduce the costs associated with current operations through the elimination of waste. Prevention of potentially damaging releases to the environment. Identification of new or existing technologies to improve production whilst minimising environmental damage. Verification of suppliers or contractors operations and procedures in line with relevant regulations and responsibilities. Assist in obtaining insurance and liability cover for potentially environmentally damaging operations.
312
9 Application of Software in Environmental Auditing and QC
Increase environmental awareness in employees and management. Improve public relations and confidence with neighbours, the public and shareholders. - Improve environmental knowledge and the flow of information between similar organisations. - Placement on the lists of companies specialising in ‘green investments’. - Identification of potential market opportunities in new and innovative techniques.
-
9.2.2 Implementing an Audit The actual form of the audit will depend upon the individual organisaton and the operations that will be audited. However, there must be several factors present before an effective audit can be carried out. These are: 1. Commitment - the audit should have the backing of top management. The audit team should have their full support and authority. This commitment should extend to an undertaking to implement all recommendations arising from the audit. 2. Resources - resources should be set aside for the audit team. This will include the time required to develop and implement the audit, the hiring of consultants, or training of personnel and the purchase of any hardware or software. 3. Leadership - someone must be appointed to be responsible for the audit system. This should be a top management post with the authority to approve expenditure and to present the results to the board.
These elements should be defined in the organisation’s enviromental policy. Once in place, the audit details can be decided. The audit parameters must be defined. It is possible to spend all of your time carrying out audits and getting deeper and deeper into smaller and smaller areas. This is of no benefit to the organisation, therefore the audit objectives must be set in advance. These will serve as a goal and a direction for the development and carrying out of the audit. The objectives that need to be set include: 1. Define the goals - this requires a clear definition of what the audit is to achieve. The goals need to be set by management. For example, the goal of the audit may be to ensure that the organisation is complying with all relevant regulations, or it may be required to look at all possible improvements to minimise waste. 2. Define the standards - an audit requires a standard against which performance is measured. Where standards do not exist, the audit becomes an assessment or review. The standards selected may be simply compliance with the regulations, it may be the organisation’s environmental policy, or even an imposed external standard. 3. Set the scope - in an ideal world, all areas of the organisation’s activities would be subject to environmental audits. This is the so called ‘cradle to grave’ approach. However, for most organisations this is not a practical solution. There must be
9.2 The Mechanics of Auditing
3 13
some form of targeting so that resources are not expended on areas of low risk whilst leaving high risk areas unchecked. The organisation should conduct an initial review of all their operations to determine where the areas of high risk lie. Set the priorities - again, organisations cannot normally audit everything, a list of priorities must be drawn up to allow the most important areas to be audited first. The priorities may also be set as to how the audit is carried out. The first priority must always be to ensure compliance with the law, following that the environmental policy and finally external standards. Set the timescale - an audit should be carried out to a set timescale, with a timetable set for completing the audit, issuing the report and implementing the recommendations. Once the management structure and the parameters are in place, work can begin on the audit itself. The audit team must be specified: this can consist of one person, if the audit is brief and the auditor is highly qualified and experienced, but normally a team of two or three is selected. The team should consist of both operational and expert personnel. Where possible the Site Manager should be included, to ensure cooperation from the site. If necessary, external help may be required in the form of consultants. In some cases this is a requirement for registration under one of the environmental management schemes. Another reason is that it may be more expensive to maintain in-house expertise. The audit itself must then be developed. The audit consists of a series of questions grouped into checklists. The questions are designed to determine the level of compliance with the standards against which the site is being audited. The exact structure of the audit will depend on the parameters set out earlier. The questions themselves should be objective, as one of the functions of an audit is that it can be repeated, possibly by a different team. It is best to stick to questions requiring a short yesho answer possibly with a brief explanation required from the auditor. The audit can be developed in one of two ways, either internally, or an external, third-party audit can be purchased and either modified to the needs of the organisation or applied as it stands. There are a number of third-party audits now available on the market, ranging from simple checklists to expensive computer operated systems covering all areas that may need to be audited. The advantages of developing an audit internally, are that the resulting audit is completely in line with the organisation’s environmental policy and with the company culture. The disadvantages are that the process can be costly in terms of time, and there is the possibility that the audit team may miss areas of importance, whereas a third-party audit should have been thoroughly tested and the development costs shared between the customers. One way of mitigating these problems is to hire outside expertise to assist the audit team and to use one of the audit development packages available that run on personal computers. The implementation of the audit requires the audit team to examine physically the site to be audited. The audit team must satisfy themselves as to the veracity of the answers to audit questions. A verbal assurance of compliance must never be accepted
314
9 Application of Software in Environmental Auditing and QC
when this cannot be verified. The audit team should make observations as they inspect the site for use in future recommendations. Following completion of the audit, a report should be developed and presented to management. This should lay out the results of the audit and list recommendations for improvements. If the audit is a follow-up to previously carried out audits, there should be a comparison with previous results and a note of where improvements have been made and recommendations implemented. The recommendations raised are the most important part of the audit. If the recommendations are not considered and implemented, the audit report becomes a mere historical document and there will be no improvements, in fact standards may even slip as it is seen that no action is taken as a result. This is why it is important to have management commitment to implementing recommendations. Regular audits should then be carried out to monitor the improvements and ensure that standards continue to rise.
9.3 The Use of Computers in Auditing As in all areas of modern life, computers are beginning to be introduced into auditing. As is generally the case in other areas where computers have been introduced, they have led to increases in productivity and improvements in results, but they bring their own problems. A computer cannot carry out an audit by itself, nor can it produce anything but the simplest of audit questions. Computers therefore will not replace people in the audit process. Their role is to provide the auditor with a superior work tool to increase the efficiency of the audit team and to improve the interpretation and presentation of results. Used correctly a computer can speed up the audit process and reduce the resources required to prepare, carry out and report on an audit, thus producing a time and cost saving. The real key to the use of computers in auditing is the sophistication of the software now available and the proliferation of cheap, personal computers. Until relatively recently large organisations have relied mainly on expensive, main frame computers. Their use in auditing is limited as they are fixed and there is often pressure on the time available. With the recent rise in the use of PCs, the audit team can have full time access to a relatively cheap resource, one that will be duplicated at every site they visit, or, with the growth in portable computers, taken with them. In its simplest form using a computer during an audit can mean nothing more than using software packages that are readily available in any company. For example the audit itself can be written on a word processor and the results recorded onto the computer. The results can be ascribed a numerical answer (eg, yes = 1, no = 0) and recorded on a spreadsheet, so that the results can be analysed numerically and even graphically. This approach will enhance the audit report, but is unlikely to produce any cost or time saving.
9.4 Coursafe
-
The Computer Based Auditing Program
315
More recently, there have been a number of specialist auditing software packages released onto the market. These range from inexpensive packages designed to make developing, recording and reporting on audits easier and more sophisticated, to expensive audit systems complete with questions, so that the only thing required is to load up the package and begin the audit. However, for the reasons explained earlier in this chapter, this is not always a good thing. The rest of this chapter deals with the application of one of the audit development packages currently available and how this has been used successfully in a number of organisations.
9.4 Coursafe
-
The Computer Based Auditing Program
Coursafe is a PC computer based auditing tool with the following functions: creation of audit checklists, a tool for audit inspection, evaluation of the performance of an audited area, production of various reports, observation and action lists, comparisons of audits performed at different locations and time, keeping track of outstanding items in action lists. Version 3.03 of the program runs under Microsoft Disk Operating System, version 3.x and above. The program takes up 2 megabyte of hard disk space and requires only 640 kilobytes of random access memory. This configuration is found in the majority of the personal computers on the market. It will run on computers with 8086 or faster microprocessors, the more powerful microprocessors will give better performance. Although Coursafe makes full use of colours, users with monochrome systems are also able to use the program on their machines. In addition to being used on standalone computers, Coursafe can be installed on a network. It has been successfully used on a Novel1 network at Henley College in Coventry where regular hands-on courses are taught. Coursafe was initially developed for safety professionals to perform safety auditing. As it is so flexible, its use has extended into environmental and quality auditing. Since its launch in 1990, Coursafe has been used in the following areas: petrochemicals, pharmaceuticals, public utilities, food and automotive manufacturing, offshore installations, education establishments and consultancies.
9.4.1 The Heart of Coursafe Central to the success of Coursafe is its ease in developing audit checklists, as well as filling in the checklist and evaluating performance. Ease of use constitute only a small part of a successful environment auditing system. Coursafe is only as good as the checklist. A poorly thought-out checklist will give unsatisfactory results. The
316
9 Application of Software in Environmental Auditing and QC
checklist may be developed internally by knowledgeable persons within a company or with the help of specialist consultants. Coursafe has the ability to handle up to 10000 questions in a checklist. They are arranged in 20 sections with each section containing 20 parts. Each part has up to 25 questions. There is no limit to the number of checklists Coursafe can use. For example, an environmental checklist could contain the following sections: POLICY ORGANISATION AND MANAGEMENT WASTE LIQUID EFFLUENT GASEOUS EMISSIONS NOISE MEASURES ENERGY TRANSPORT LAND AND PREMISES MONITORING AND FEEDBACKS EMERGENCY RESPONSES Within the ‘Liquid Effluent’ section the checklist should cover the following areas: ESTABLISHMENT OF SITE STANDARDS MONITORING OF DISCHARGES DOCUMENTATION COMPLIANCE WITH STATUTORY LIMITS FUTURE TRENDS Some of the questions in the section on ‘Compliance with statutory limits’ are: Does the site produce liquid waste stream(s) which idare passed into controlled waters? Has the site written consent to discharge the trade effluent into the waters? Does your site effluent contain any substances which are currently on the ‘Red List’? If Yes, list them in the ‘Notes’. Have you applied or plan to apply BATNEEC to control the emission of the ‘Red List’ substances? From the examples above, one can see that the questions can be made to be very specific to a single site and that it needs regular updating to reflect the changes in legislation.
9.4.2 Using Coursafe An auditor using Coursafe will be faced with a typical answer screen (Fig. 9-1). The question is shown in the middle of the screen. Just above it, the auditor is reminded of the section and part where the question is located. The bottom portion of the
9.4 Coursafe - The Computer Based Auditing Program
COURSAFE V3.03
Tuesday
29-12-1992
317
11:47 am
ENVIRONMENTAL AUDIT
QUESTION
3: LIQUID EFFLUENT 4: COMPLIANCE WITH STATUTORY LIMITS
3 .4.3 -.
Does your site effluent contain any substances which are currentl on the ’Red List‘? I f yes, 1st them in the ‘Notes’.
f ,4
ANSWER:
F1 : He1 Alt-X : En$
? : NIA
Y :tfes
Esc : Abandon .c-l : Accept
N : NO
c+ Select
On iile : N
P g U F PrevPgDn : Next
I
F7 : Notes F9 : Recommendation
Fig. 9-1. A typical Coursafe audit question.
screen shows the other facilities available to the auditor. Therefore infrequent users will not have to remember lists of commands when using Coursafe. The possible answer to this question may be ‘Yes’, ‘No’ or ‘Not Applicable’. The auditor enters the answer either by typing the first character of the offered option (which is ‘Y’, ‘N’, or ‘?’) or by moving the highlight (cursor) over his choice and then pressing the ‘Enter’ key. This is consistent throughout the whole program. Some questions accepts numerical answer within the range of 0 to 100% inclusive. The type of acceptable answer to a question is decided by the developer of the checklist. In Coursafe, there are two levels of help information. Pressing the ‘FI’ function key once will bring out the guidance on how to answer the question. Repeating the keystroke will display the instructions on how to use Coursafe. Extra information may be entered by the auditor using the ‘Notes’ facilities. On pressing the function key, ‘F7’, a panel will appear on the screen for entering text. For example, with the question in Fig. 9-1, the auditor could put in the names of the substances which caused concern. Also, recommendations may be entered by pressing the ‘F9’ function key. This is different from ‘Notes’ because the auditor could enter a priority level as well as text. The recommendations will form the action list. Most annual audits take some time to complete, typically spreading over a few days. Coursafe has a ‘Review progress’ screen which clearly pinpoints the areas that have not been answered (Fig. 9-2).
318
9 Application of Software in Environmental Auditing and QC
/
, COURSAFE V3.03
..........
Tuesday
29-12-1992
EDWARD ALANDALE REVIEW ENVIRONMENTAL AUDIT SECTION
1
2
1150 am
3
4
5
6
,
7
8
I
Fig. 9-2. Review progress screen.
Even if a single question contained in a part has not been answered, it will be identified on this screen. To move to that outstanding question, the auditor moves the cursor to the ‘X’ mark followed by the ‘Enter’ key on the keyboard.
9.4.3 Evaluation of Performance The compliance of an audited site can be displayed in two ways. As each question is weighted numerically, a quantitative measure is possible. The first representation can be seen in Fig. 9-3. The score of each section is shown as percentage of the maximum score of the section. This display is called cascade analysis because the score of the parts and the questions are easily accessed. For example, to reveal the breakdown of the score in section 4, the cursor needs to be moved onto the section description and followed by the ‘Enter’ key. Similarly, the scores of individual questions in a part are revealed. Figure 9-4 shows the actual score of each section and Fig. 9-5 presents the result in a graphical form. Switching the display is extremely easy - just by pressing the function keys ‘F6’ and ‘F8’. These displays can be printed out or sent to a file by pressing the ‘F2’ key. The alternative method of displaying the result is the Cross Tabulation. Figure 9-6 shows the scores of the parts in a checklist. The row shows the section and the col-
9.4 Coursafe
(
COURSAFE V3.03
Tuesday
-
29-12-1992
3 14 5 6 7 8 9
I
11:59 am
EDWARD ALANDALE ANALYSIS ENVlRONMENTAL AUDIT SECTION 1 2
3 19
The Computer Based Auditing Program
%
POLICY ORGANISATION AND MANAGEMENT WASTE
90
85 65
LlGtUfD EFiWENTI
67 70
GASEOUS EMISSIONS NOISE MEASURES ENERGY TRANSPORT LAND AND PREMISES
100 72 64 89
Percentage
81
I
I
F1 : He1 Alt-X : E n 8
Esc : Done tl : Select
t : Move up 4 : Move down
F2 : Print F9 : Prof.
F6 : TexVGraph F8 : TotaldPercent
Fig. 9-3. Cascade score analysis of audit.
(COURSAFE V3.03
Tuesday
29-1 2-1992
EDWARD ALANDALE ANALYSIS
2
ORGANISATION AND MANAGEMENT
4
LIQUID EFFLUENT GASEOUS EMISSIONS NOISE MEASURES ENERGY TRANSPORT LAND AND PREMISES
5 6 7 8 9
Fig. 9-4. Cascade analysis showing actual scores.
1:47 pm
320
9 Application of Software in Environmenlal Auditing and QC
( COURSAFE V3.03
Tuesday
29-12-1992
2:02 pm
EDWARD ALANDALE ANALYSIS ENVIRONMENTAL AUDIT SECTION
Score 1800 850 650 670 700
1 2 3
'
4
5 6
1000 720 640 890
:i
Max 2000 1000 1000 1000 1000 1000 1000 1000 1000
ll
Fig. 9-5. Cascade analysis in graphical form.
f COURSAFE V3.03
Tuesday
EDWARD ALANDALE ANALYSIS ENVIRONMENTAL AUDIT SECTION 1 POLICY c)
n n e ~ u i e A m n h i kin
L vnuruwoniiviv
niww
...
3 WASTE 4 LIQUID EFFLUENT 5 GAS _....__._.._ 6 NOISE MEASURES
____
7 ENERGY
&j$/ ..... .... 8
:?j
....: i >... ..::.::9 ............ .:... ..... ......... :. ....... .......... : I . .......... ........ &,: :.:. ..........
I
Tk&j$po~~ LAND AND PREMISES
A 7 Itf
%
90
0 100 100 100 100 100 83
100
100 i00 ioo ioo ioo 100 67 100 50 33 100 83 55 56 80 69 100
64 89
Percentage
FI : He1 'Ait-X :
En8
~~~~~~~~
1 2 3 4 5 6 100 43 100 100 100 100
+fit3 IVV
I
2:17 pm
29-12-1992
Esc : Done : Select
+ J ~
~
~
c a t + : Move
~
Fig. 9-6. Cross Tabulation presentation of result.
F2 : Print F8 : TotaldPercent
F9 : Prof.
I
9.4 Coursafe
-
The Comimter Based Auditim Proeram
321
umns correspond to the parts. Any value less than 1OOolo shows that compliance is below the best.
9.4.4 Multi-Audit Analysis After an initial round of auditing, it is advantageous to place the audit results of all the sites of an organisation side by side to obtain an overall performance indicator. The limitation of Coursafe at the moment is that only 25 columns can be displayed at a time. If an organisation has 40 sites, the analysis is carried out twice; twenty sites in each analysis. Figure 9-7 shows the layout for the annual results for 25 sites. The advantage of displaying the data in such a way is that Coursafe can ‘zoom’ into the audits and look at the score of each section, part and individual question. Figure 9-8 shows this facility performing such an analysis on an environmental audit. This is very useful for identifying weak areas in a multi-site organisation. More probably, the organisation may want results shown with some sort of grouping. Coursafe allows the user to collate the results of the sites according to the division to which they belong. For instance, instead of 25 columns (one for each site), only 5 are used (each column would correspond to a division). The value shown in each column is the mean score of the sites within the division. Coursafe allows trends to be handled easily. The annual results are shown in rows. Total quality management is moving towards integration of various measurements. In response to this many organisations are merging their safety, environmental and occupational hygiene functions. Coursafe has the facility to produce annual performance based on the summation of the measurements in each function.
MULTl AUDIT ANALYSIS
1
MODULES PERIOD : 1992
Environmentalaudit
LOCATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Total nnnnnnnnn n n n n n n n n n n n n n n n n nn
Fig. 9-7. Multi-audit analysis of 25 sites.
9.4.5 Implementing Coursafe Coursafe is designed to be an open system. It allows users to configure the system to match their organisation and culture. This flexiblity can be seen from seemingly trivial features to vital aspects of the system.
322
9 Application of Soffware in Environmental Auditing and QC
MULTl AUDIT ANALYSIS MODULES PERIOD : 1992
I I Endmental auda
LOCATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 TOM nnnnnnnnn n n n n n n n n n n n n n n n n nn
I
I I
MULTI AUDIT ANALYSIS SECTIONS PERIOD : 1992
LOCATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Total
policy nnnnnnnnn Omanlsation and rnanaaernentn n n n n n n n n nnnnnn nnnnnn nnnnnn
n n n n n
n n n n n
n n n n n n n n n n n n n n n n n n n n n n n n
n n n n n n n n n
n n n n n n n n n
n n n n n n n n n
n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n
n n n n n n n n n
n n n n n n n n n
n n n n n n n n n
nn nn nn nn nn nn nn nn nn
I
MULTl AUDIT ANALYSIS
PARTS PERIOD : 1992
LOCATION
Establishmentofsite...
es
12 nn nn nn
3456789 nnnnnnn nnnnnnn nnnnnnn
10 n n n
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 TO!A
n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n
nn nn
MULTl AUDIT ANALYSIS QUESTIONS PERIOD : 1992 12 3 Doesthesiteprcduceliq... n n n Hasthesitethewritten co... n n n DO& Your site ~fRUentcon... n n n Have you applied or plan... n n n
LOCATION 456789 nnnnnn nnnnnn nn nnn n nnnnnn
10 n n n
n
11 n n n n
~
Fig. 9-8. Zooming into the audit of 25 sites.
12 n n n n
13 n n n n
14 n n n n
15 n n n n
16 17 n n n n n n n n
18 n n n n
19 20 n n n n n n n n
21 n n n n
22 23 24 25 To$l n n n n n n n n n n n n n n n n
nn nn nn nn
I
9.4 Coursafe
-
The Computer Based Auditing Program
323
The feeling of ownership engendered by a product is crucial in determining the degree of acceptability. A product which is personal is often cherished and well used. To this end the opening screen of the Coursafe program can be tailored to display the company logo using screen colours according to the company livery. Many companies have the desire to project their ethos and culture throughout their establishment. Coursafe is geared to do just that. For example, positive reinforcement is a proven tool that is in favour with many progressive companies. Coursafe not only gives a quantifiable measurement of performance at the conclusion of an audit but also it can be set to reveal the weighting of each question of the checklist. Managers have been known to be spurred into corrective actions when they become aware of areas of non-compliance. Needless to say, for this method of management to be effective over a long period some form of incentive must be in place. The challenges to be faced are dependent on the way in which Coursafe is implemented in a company. A well resourced company would rely principally on experienced professionals to perform their audits. These auditors will firstly use Coursafe as aide memoire as they scrutinise a site. Coursafe is therefore performing like a normal checklist; ensuring that all the relevant areas are covered and not missed out. In addition, the auditors could make use of the ‘Notes’ facility in the program to record their observations and the ‘Recommendation’ facility to build their proposed action list as they see necessary. The action items are given a priority level. At the end of an audit, using a few keystrokes on the computer, the overall performance of a site can be calculated. As each question in the checklist is weighted, the site performance is given a quantitative score. The feature most valued by the auditors is the ability of Coursafe to print out a list of actions for the exit meeting, prior to leaving the site. When the auditors return to their own office, they can then transfer their audit observations and results to their wordprocessor, database and presentation packages. The final audit reports are therefore issued quickly and with the knowledge that they have the agreement of the local management. In the present stringent financial climate, experienced auditors are scarce in most companies. The ways in which companies overcome this problem are varied. Some resorted to ‘doubling up’, in others words, safety professionals are used to carry out environmental audits and vice versa. Others delegate the auditing function downwards to the line managers. The few experienced auditors are then used to verify the resulting audits. Regardless of the selected method, the auditor in the field needs effective guidance so that the audit will yield accurate and consistent results. Coursafe amply fulfils this need and beyond. When faced with a question on the checklist, the auditor can invoke the ‘Help’ function. Coursafe displays the background to that particular question, the legislation in force, the company standard and current best practices. Furthermore it can advise the auditor on what to look for and what the acceptable standard should be. Lastly, if compliance is not achieved, corrective actions may be recommended. Therefore, Coursafe can also be used as a training tool. In recent times, the consumer has grown aware of the environmental issues. Sensing this, many companies have made unjustified environmental claims to boost the sale of their products. Forthcoming legislation will make it difficult for such irresponsible conduct. Environmental statements will undergo more scrutiny and audit
324
9 Application of Software in Environmental Auditing and QC
by independent bodies will be the norm. Coursafe has the facility for verifiers to validate audit results.
9.4.6 Some Novel Applications Companies principally use Coursafe to assess the effectiveness of their own safety, environmental and quality management. Some have gone beyond this application and three are worth a mention. 1. A Coursafe user company supplies chemicals to shops found commonly on the
high street. Their excellent customer support included the provision of a Coursafe auditing package which provides training on the storage, handling and disposal of chemicals. 2. The environmental performance of a company which retails equipment fabricated by other manufacturers is dependent on the duty of care of the suppliers. To ensure its own standard is met, a Coursafe user company applies its checklist on their existing and potential suppliers. Their environmental scores are considered along with the normal factors before orders are placed. 3. Coursafe is an extremely flexible and user friendly auditing tool. Therefore it is not a surprise to hear that a manager borrowed the package to audit the swimming pools in the neighbourhood to help him decide where to send his children for lessons.
Index
AAS (atomic absorption spectrometry) 55 - comparison with ICP 110 2,4-D acid 51 additional test 75 adsorbents 45 - GC-FID characterization 45 - introduction into the ICP 104 aerosol gas (see also carrier gas) 99, 105 air analysis (see also airborne particulates) 118 air sampling 140 air toxic 172 airborne particulates 118 - analysis 122 - filters 120 - impactors 121 - sample preparation 121 - samplers 120 - sources of 118 - threshold limits 120 alternative techniques methods 77 Alumina 7 aluminum 296 amino 8f. ammonia 300 ammonium 294 analysis considerations 19 - gas chromatography 19 - HPLC 19 analysis of variance (ANOVA) 85 analysis rates 276 analyte additions 68 analytical dynamic range 77 - comparison 77 analytical quality assurance 59 animal tissue analysis 127 anion exchange phases 9 antimony 297 aqua regia, decomposition with 121, 123
argon gas flow rates in ICP 104 aromatic sulphonates 255 arsenic 296 artifact interference 142, 144f. - analysis 146 - due to ingress of VOCs via diffusion 145 - errors 145 - purge of air/oxygen 144 - tube conditioning and storage 145 athonic surfactants 301 atomic absorption spectrometry see AAS atomic emission - energy levels and diagrams 97 - ground state atoms 96 - ionised state atoms 96 - principles 96 - spectral lines 97 - wavelength(s) of emission 96 audit 309, 311 ff. - benefits 311 - development 313 - effectiveness 311 - environmental 309 - parameters 311 - report 314 - use of computers 314 audit questions 316 audit team 313 auditing software packages 315 Auto-AnalyzerTM 269 auto-dilution 281 automated tests 75 - within-batch precision 75 automatic thermal desorption 156 - leak testing 156 - sealed tubes 156 automatic tube desorption 157 auxiliary gas 104
326
Index
Babington nebuliser (see also nebulisers) 105f., I 14 backflush 155 - automation 156 background continuum 117, 224, 130 background correction, off-peak 117, 130 beryllium 295 bicarbonate 299 bi-directional ITP 224f. - calculation of LE/TE composition 225 - electrolyte system 225 biogenic emissions 178 biological samples 125 - analysis 129 - collection 128 - element concentration levels 127 - preparation 128 blank 70, 195 blank signal 65, 71 boron 297 British Standard for Environmental Management System 309 building materials emission 177 building ventilation 175 - tracer gases 175 butyl 8 C 8 octyl 8 C18 octadecyl 8 calcium 294 calibration 64, 110, 302 - accurate canister standard 187 - calculation methods 187 - EPA recommended standards 118 - GC-MS 196, 199 - importance 64 - linearity 110, 124 - procedure 64 - quality coefficients 68 - response-factor 188 - solutions 61 - tube samples 188 calibration curve 64, 66f. calibration function 67 calibration graph 64, 86 calibration standard solutions 61 f. - preparation 61f. - sources of error in preparation 62 calibration strategies 64 canister 140, 172, 179 capillary electrophoresis (CE) 213 capillary GC 134 capillary zone electrophoresis (CZE) 21 3 f.
carbon dioxide 26 carbonate 299 carboxylic acid 9 carrier gas 104 cartridges of SPE 2f., 6 - glass 2 - polypropylene 2 -PTFE 2 - size 16 - use 6 cation exchange phases 9 CE-MS interface 254 - designs 254 ceramic injector 104, 121 certified reference materials (CRM) 78 certified soils 38 certified standards 157 CFA/FIA system 292 - schematic diagram 292 characteristic concentration 63 characteristic mass 63 chelation, solvent extraction in water analysis 115 chlorinated dioxins 201 chlorine 300 chlorophenoxy acid herbicides 5 1 - derivatization agents 51 - soil 51 choice of IQC materials 79 chromatography 195 - resolution 195 chromium 295 clay soil 42 cobalt 295 co-precipitation, concentration by 116 coil, RF 98 collection 30 - GC 30 - off-line SFE 30 - on-line mode 30 - SFC 30 commitment 3 12 compliance 163, 318 - score analysis 318 2,4-D compounds 50 - soil 50 computer based auditing tool 3 15 computerized search identification 203 f. - confirmed identification 203 conditioning solvents 13 - final 13 - initial 13 - purpose 13
Index
-
strength
13 13 conductivity detection 247f., 250, 252 - contactless 248 - in-contact 248 - suppressed 250 - suppressor designs 252 consistent data 57 contactless conductimeter 248 - schematic diagram 248 contamination 65, 70 - influence 65 - in sample containers 111 - in sampling 127 control charts 83, 86, 88 coolant (plasma) gas 104 copper 296 correlation coefficient 67 - method 68 cotton 51 counter-ion 11 coursafe 315ff., 320f., 323f. - advantage 315f. - answer screen 316f. - application 324 - checklists 316 - Cross Tabulation 318, 320 - display 318 - feature 323 - functions 315 - multi-audit analysis 321 f. - questions 316 - review progress 3 17f. - score analysis 3 18ff. - use 315, 317, 323 cross-flow nebuliser (see also nebulisers) 104 crystal control (of generator frequency) 103 cyanide 299 cyano 8 cyclohexyl 8 Czerny-Tbrner mounting (see also monochromators) 102f. - volume
data interpretation 163 data quality 208 - contract laboratories 208 Dean’s pneumatic switch 168ff. dehydration techniques 45 demountable torch 104 derivatization 274 desiccant dryers 152 desorption efficiencies 136 detection limit (DL) 71, 209
327
- alternative GC-MS technique 209
validation 209 detection methods 243, 245 f. - addition of complexing reagents 246 - addition of organic solvents 245 - electrolyte system optimisation 243 - optimisation of injection method 243 detection methods for CE 241 - absorbance 241 - capillary cells 241 - electrochemical 241 - fluorescence 241 - indirect optical detection 241 - radiometric 241 - rectangular capillaries 241 detectors 99, 102, 106, 281 - array type 107 - chart recorder 281 - photomultiplier type 106 dialysis 274 diethylaminopropyl 9 diffraction gratings - concave 101 - plane 101 diffusive monitoring 159 diffusive samplers (monitors) 147ff. - principles 150 - uptake rate 148 diffusive sampling 150, 165 digestion 274, 290f. - microwave 291 - UV 290 dilution 274f. diol 8 disks of SPE 5f. dispersion 270f., 273 dry ashing 128 - advantages vs disadvantages 128 - ashing aids 128 - muffle furnace 128 - oxygen plasma 129 dry purge 152 dust analysis (see also airborne particulates) 118 -
EC scheme 310 echelle mounting (see also monochromators) 103 effect of temperature 35 - PAH, pentachlorophenol and heteroatomic PAH 35 electroosmotic flow (EOF) 217, 223 - dependence 223
328
Index
-
diffuse layer 217 electrical double layer 217 - stern layer 217 elemental significance in environmental analysis 108 elutions solvents 14 - purpose 14 - strength 14 - volume 14 end-column conductivity detector 250 - combined conductivity and UV detection 250 - construction and design 250 entrance slit 101 environmental auditing 309 environmental monitoring 55 environmental performance 324 environmental sample types 108 environmental target analytes 3 1 - pesticides 31 - polychlorinated biphenyls 3 1 - polychlorinated dibenzo-p-dioxins 3 1 - polychlorinated dibenzofurans 3 1 - polynuclear aromatic hydrocarbons 3 1 - total petroleum hydrocarbons 31 enzyme-linked immunosorbent assay (ELISA) 50 - 2,4-D acid, 2,4-D salt and 2,4-D ester in soil and cotton material 50 EPA Method 200.7, water analysis 112 EPA Method 525 22 equilibrium headspace 136 ESI-MS interface 256 - schematic diagram 256 ethyl 8 evaporation, concentration by in water analysis 114 exit slit@) 101 exposure limit levels 159 external quality assessment 89 external quality assurance (EQA) 56, 90 - samples 90 extractability 28 extraction temperature 33 - PAH 33 - polychlorinated biphenyls 33 field amplified sample injection (FASI) 239 - schematic representation 239 Field and Laboratory Emission Cell (FLEC) 177 field blank 71 filtration
- airborne particulates - water samples 112
120
Flick’s law 148 Florisil 7 flow analysis 267f., 275, 278, 280, 293, 301 - applications 293 - comparison of techniques 275ff. - detectors 280ff. - instrumentation 278 - quality assurance 301 - schematic diagram 268 flow analysis techniques 283ff., 287 f., 290f. - combined CFA with FIA 291 - derivatization 284 - dialysis 287 - digestion 290 - dilution 283 - gas diffusion 287 - preconcentration 288 - removal of interferences 290 - solvent extraction 285f. - stopped-flow 285 - transport 283 flow inhibition in SFE 45 flow injection analysis (FIA) 267, 272f., 279 - function 273 - principles 267 - schematic flow diagram 273 - valves 279 focusing 135 focusing device 137f., 155, 167 - ambient temperature trap 137 - automation 155 - cold-trapping 137 - cryofocusing 137 - electrical (Peltier) cooling 138 - liquid cryogen 137, 155 free running (generator frequency) 103 frequency of analysis 79 frits 2 - polyethylene 2 -PTFE 2 - stainless 2 gas diffusion 270, 274 gas phase internal standard 187, 191 gas-segmented continuous-flow analysis (CFA) 267, 269ff. - function 269 - goal 270 - principles 267 - schematic diagram 269 - steady state response 271
Index GC-MS 196f., 199ff., 203f. assessing column performance 197 calibration 196f., 199 chemical ionization 20Of. compound identification 203 f. computer search 203 confirmation of identity 206 GC-MS-MS 201 ion trap 202 isotope dilution 202, 205, 207 mass chromatography 205 mass spectra 197 peak shape 196 performance check 200 quadrupole 201f. qualitative analysis 202f. quantitative analysis 202f. response factors 196 scan speed 202 selected ion monitoring 205 f. sensitivity 197 troubleshooting 197 generator, radio-frequency 98, 103 - crystal controlled 103 - free running 103 - power 103 - radio frequency 103 GFASS (graphite furnace atomic absorption spectrometry) 110 grating, see diffraction grating ‘green’ companies 310 grid nebuliser (see also nebulisers) 104, 106 groundwater analysis 123 guidelines 204 - mass chromatography 205 - quantitative analysis 205 - selected ion monitoring 205
halides 299 harbor sediments 40 heavy metals 221 herbicides 216, 219, 255, 301 - diquat 216 - paraquat 216 - sulphonylurea 255 - triazines 219 high resolution mass spectrometer 202 - mass resolution 202 high-speed 37 - dichloromethane 37 - PAH 37 hybride generation 95, 106, 109, 114, 130
329
hydride forming elements 106 sodium borohydride 106 hydrochloric acid, digestion with 129 hydrofluoric acid, dissolution with 121 hydrogen cyanide 300 hydrophobic 13 hyphenation 268, 292 -
-
ICP-AES (or OES) - advantages of 110 - detection limits for water analysis 114 - suitable of 108 - technique 95 ICP-MS 110 impactors 121 indirect detection in ITP 242 indirect UV detection for CZE 242 inductive coupling 98 industrial emissions 164f. - fence-line monitoring 165 - sample train 165 - stack emission testing 165 in-house IQC materials 81 injectors 107, 114 - ceramic 104, 121 - quartz 104 in-line distillation 280 in-situ derivatization of PAH 39 instrumental detection limit (IDL) 72 interelement correction 116 internal quality control (IQC) 56, 79ff., 83, 88 - choice of materials 79ff. - control charts 88 - establishment 82 - in-house materials 81 - material 80 - preparation 81 - protocol 83f. - statistical errors 80 internal standards 130 ion exchange 274 ion exchange, in water analysis - concentration factors 115 - non-selective 115 - on-line 113 - selective 115 ion exchange phase pK,s 12 ion exchange sorbents 9 - anion 9 - cation 9 ionic species 10f. ionic strength 11
330
Index
IQC target values and limits 82 - establishment 82 iron 295 isotachophoresis (ITP) 220, 222, 224 - discontinuous buffer systems 220 - surface-active compounds 222 - leading electrolyte 220 - open systems 220 - steady state 220 - terminating electrolyte 220 - velocity of the EOF 222 - voltage drop and time of analysis 224 ITP-CZE with applied backpressure 240 - schematic representation 240 Joule heating 216 Kohlrausch regulating function 220 - Cr("') 221 leachates, in soil analysis 123 leaching solutions 123 lead 296 leadership 3 12 linear calibration 67 liquid cryogen 167, 171 liquid -liquid extraction 1 lithium metaborate fusion 121 loading solvents 13 - purpose 13 - strength 13 - volume 13 long term storage 173 magnesium 294 manganese 295 mass balance estimations 165 mass spectrometric detection for CE 252, 254f., 258f. - CE-MS interface 254 - continuous flow fast atom bombardment (CF-FAB) 254 - CZE-MS 255 - electrospray ionisation (ESI) 254 - ITP-CZE-MS 259 - ITP-MS 258 - liquid junction 254 - quantitation 255 - sheat flow (coaxial) interface 254 - sheathless design 255 matrix matching of standards 130 matrix treatments 45 - air-drying 45
- drying agents 45 - mixing with magnesium sulfate 45
-
oven-drying 45
- placing on magnesium sulfate 45
Maximum Allowable Concentrations (MAC) in water analysis 115 microwave digestion 121 membrane dryers 152 mercury 296 metal ions 221, 242, 247 methanol 10 methanol to CO,, addition 37 - PAH 37 - static modifier 37 method considerations 29f. - different 19 - dryness 19f. - gap 19 - manufacturers 19 - pH 19f. method development 16 - elution 16 - profile 16 - solvent 16 - sorbent 16 method development strategies 32 - collection 32 - extraction 32 - pre-extraction 32 method validation 56, 60, 78 methods (GC-MS) 205f. - isotope dilution 205 - masses to monitor 206 - selectal ion monitoring 206 methyl 8 micellar electrokinetic chromatography (MEKC) 218 - critical micelle concentration (CMC) 21 8 - phenols 218 migration time 215 miniature 303 miniaturization 268, 303 mobility 214f. - electrophoretic 215 modes off-line collection in SFE 31 - solid phase bed 31 modifiers 28 moisture management 151, 174 - desiccant dryers 151 - dry purging 151, 174 - permeable membrane dryers 151 monochromator detector 281
Index monochromators - Czerny-Turner 102 - echelle 103 - sequential operation 102 multi component analysis 268, 293 multiple metals 297 multi-rule 88 multi-rule Shewart chart 86, 88 national external quality assurance scheme (NEQAS) 89 - blood lead determination 89 nebulisers 104f. - Babington 105f., 114 - blockage of 112, 114 - concentric glass 104 - cross-flow 104 - droplet sizes 105 - grid 105f. - pneumatic 114, 121 - ultrasonic 106f., 114, 130 - v-groove 105f. nitrate with FIA 273f. - schematic flow diagram 274 nitric acid, digestion with 112, 121, 129 nitrogen 298 nitrosoamines 7 non-linear-calibration 67 normal phase 10 - sorbents 7 off-line collection 30f. - cryogenically cooled adsorbent trap 31 - dry surface 30 - graphical representation 3 1 - liquid solvent 30 - modes in SFE 31 - solid phase bed 31 on-column conductimeter 249 - diagram 249 - electropherogram 249 on-column ITP-CZE 236f. - different modes of transient ITP 237 - sample stacking 236 - transient ITP 236 on-line air sampling 151 on-line analyzer 303 on-line distillation 277 on-line preconcentration 240 - chromatographic micro-precolumns 240 - coupling with LC 240 organic and inorganic anions 244 - alkylsulphonic acids 244
-
carbonate 244
- chloride 244 - fluoride 244
formate 244 haloacetic acid 244 - nitrate-N 244 - phosphate-P 244 - sulphate-S 244 organochloride 49 organophosphate pesticides 49 - modifiers 49 - soils 49 orthophosphate with CFA 271f. - schematic flow diagram 272 ozone 300 ozone precursor 167 -
PAH 22, 39 - modifiers 39 PAH contaminated soil 34 - precision 34 - SFE-GC/MS characterization 33 Paschen-Runge mounting (see also polychromators) 101 PCBs 22 Peltier cooled trap 139 perchloric acid, digestion with 129 - safety in use 129 perfluorotributylamine (PFTBA) 199 performance 197 - gas chromatography 197 - mass spectrometer 197 performance characteristics 60, 70 peristaltic pump 105, 278 - problems 278 personal exposure 147, 159, 164 pesticides 9, 21 - in soil 50 pH 1Of. phenol 7, 242, 301 phenoxyacid herbicides 21 phenyl 8 phosphorus 298 photometric detector 281 photomultiplier 99 plant materials - analysis of 125 - normal concentration levels in 127 - sample collection 127 plasma 98 - axial zone 99 - temperature distribution in 99 - toroidal shape 98
331
332
Index
pollution transport studies 178 pollution types 108 polychlorinated biphenyls 177 polychromators 101 - Rowland circle 101 polynuclear aromatic hydrocarbons 9, 32 - solubility 32 potassium 294 precipitation 274 precision 73, 76 - between-batch 76 - obtainable in analyses 117, 124, 130 - within batch 76 preconcentration 268, 270, 274, 276, 288f. - packed column 288 - precipitation 289 - solid-phase extraction 288 - in water analysis 112, 114 pressurized vessels 279 primarylsecondary amine 9 process of sampling air 155 propylsulfonic acid 9 provisional identification 204 - compound class 204 pumped air sampling 147 pumped monitoring 147, 161 - pump flow rates 147 - sequential tube sampling 147 pumped tube sampling 160 qualitative analysis 202 quality assurance 60, 193, 301 quality control 56, 60, 79, 193, 203 - charting 85 - protocols, in water analysis 117 - use 82 quality control chart 86 quality control principles 59 quality control procedure 83 - definition 83 quality elements 208ff. - alternative GC-MS technique 209 - archived samples 209 - blanks analyzed 208 - blind replicates 208 - dedicated equipment 208 - detailed report 208 - detection limit validation 209 - expert assistance 210 - multiple contractors 209 - sample storage 210 - stable isotope standards 208 quality management programm 195
quantitative analysis 202 quarternary phosphonium ions 259 quartz injectors 104 random access samplers 279 random error 73, 85 - percent 74 range of elements measured in ICP 95, 109 reaction coils 279 recovery measurements 73 relative percentage difference (RPD) 75 relative standard deviation 74 repeatability 74 reproducibility 58, 74, 76 restrictor technology 47 - wet soils 47 reverse phase 11 - sorbents 7 reversed FIA 284 rinsing solvents 14 - purpose 14 - strength 14 - volume 14 river sediment 35 Rowland circle (see also polychromators) 101 sample containers 110 - cleaning of 111 sample introduction methods for CE 226ff. - electrical sample splitting 227 - electrokinetic 227 - high concentration sample 226 - hydro-dynamic 227 - membrane sample introduction 230 - micro-injector 230 - rotary valve injector 228 - slider valve 229 - ubiquitons injection 227 sample introduction systems 99, 104 sample matrix considerations 18 - complex samples 18 - impact 18 - liquid samples 18 - solid samples 18 sample preparation 1, 25 - carbon dioxide 25 - SPE 25 sample splitting 161 sample tube 149 sampling 195, 205 - archived samples 209 - homogenous 195
Index methods 140 pre-collection considerations 110 - preservation 195 - representative 195 - sample preparation 205 SAX quaternary amine 9 scanning monochromators 102 SCX benzenesulfonic acid 9 seagull eggs 39 secondary slits 101 sediment, soils and sludges analysis 123 segmented flow analysis 267 selection - SPE format 6 - SPE size 16 - SPE solvents 12 selectivity 28, 216 selenium 297 separation 170 - adventages 170 sep-paks 4 sequential injection analysis (SIA) 268, 273, 275 - schematic diagram 275 sequentially automated SFE 45 - hydromatrix 45 - TPH 45 - variable restrictor 45 sewage sludge sample 38 SFE 37 - component 29 - mechanism 29 SFE instrument 29 - automated 29 - decompression 29 - extraction vessels 29 - modifier pump 29 - pumps 29 SFE operational parameters 32 - extraction pressure 32 - PAH 32 SFE of pesticides 49 - GC electron capture detection (ECD) characterizations 49 Sick Building syndrome (SBS) 175 side arm flask apparatus 3 signal processing 99 silica 7f. silicon 297 sinusoidal pump 279 sodium 294 software programs 282 soil 51 -
-
- 2,4-D acid 51 - desorption 28
EPA certified 36 matrices 39 - partitioning 28 soil, high-contamination-level 43 - hydrocarbons 43 soil, law-contamination-level 44 - hydrocarbons 44 solid phase extraction (SPE) 2, 40, 274 solvent extraction (FIA) 136, 274, 286 - schematic diagram 286 - in water analysis 115 solvent flow rate 15 - cartridge 15 - disks 15 solvent miscibility 15 solvent power 28 solvent relationships 11 solvent strength 10, 28 - normal phase 10 - reverse phase 11 solvent volatility 15 solvent volume 14 sorbent bed 15 sorbent selection 142, 145 - retention volume 142 - Safe Sampling Volume (SSV) 145 sorbent strength 173 sorbent tube 141, 146, 173f., 183 - advantages 174 - hydrocarbons 183 sorbent tubes, uptake rates 183ff. - aldehydes ketones 186 - ester glycol ethers 185 - halogenated hydrocarbons 184 - miscellaneous 186 sorbents 3 - particle diameter 3 - shape 3 spacers 235 SPE format, selection 6 SPE method, examples 20 SPE method steps 7 - conditioning 7 - elution 7 - loading 7 - rinsing 7 SPE size, selection 16 SPE solvents, selection 12 SPE sorbents 7 SPE triangle 28 speciated monitoring 134 -
-
333
334
Index
spectrometer 99 - monochromators 102 - polychromators 101 - wavelength range of 102 speed of operation 101 spiked sediment 37 spray chamber 106 - Scott-type 107 stabilized temperature platform furnace 68 standard 207 standard deviation 66 standard methods 163 - method performance criteria 163 - method validation protocols 163 Standard Reference Materials (SRM) 117, 122, 126, 130 standardized reference methods quality control 57 stopped-flow 274 sulfonylurea herbicides 48 - aqueous matrices 48 - empore discs 48 - methanol modified 48 sulfur 298 sulfur dioxide 300 supercritical CO 28 supercritical fluids 26ff. - density 26 - diffusion coefficient 26 - physical property 27 - surface tension 26 - viscosity 26 surface adsorption 111 syringe filter 4 syringe pumps 278 systematic errors 85 TCDD 207 Tesla discharge 112 thermal desorption 136, 162 third-party audit 313 thorium 297 torch 103 - dimensions 104 - Fassel type 103 - outer tube 104 - tulip (middle) tube 104 total petroleum hydrocarbons 41 - correlation experiments 41 - field measurements 41
- infared spectrometric analysis - TPH 42
41
TPH contaminated 47 - GC-FID 47 - off-line SFE 47 TPH in soil 43 - provided of the significant reduction 43 - solvent consumption 43 TPH/Oil/Grease in soil 42 - experimental design 42 trace concentrations 193 trace elements 55 trap design 137 'Iliazines 21 tube sampling procedures 156 two column 170 two-dimensional electrophoresis 231f., 234 f. - apparatus 232 - bifurcation block 231 - coupled column ITP-CZE 232 - electrolyte systems for ITP-CZE 234 - ITP-CZE 231 - ITP-ITP 231 - sample clean-up 235 uranium 298 urban air quality 167 urban dust 40 UV-Vis detectors 281 vacuum manifold 3 variflowrM 47 - soils with TPH 47 vehicle exhaust emissions 166 VOC air monitoring applications
158
water analysis 112 weet petroleum sludge 46 - air dried 46 - GC/FID analysis 46 - magnesium sulfate 46 - oven dried 46 wet ashing, biological materials 129 wet petroleum sludge 45 wet sediment 44 - total petroleum hydrocarbons 44 within-batch bias 88 zinc 296
