ISBN: 0-8247-0509-2 This book is printed on acid-free paper. Headquarters Marcel Dekker, Inc. 270 Madison Avenue, New Y...
57 downloads
1099 Views
4MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
ISBN: 0-8247-0509-2 This book is printed on acid-free paper. Headquarters Marcel Dekker, Inc. 270 Madison Avenue, New York, NY 10016 tel: 212-696-9000; fax: 212-685-4540 Eastern Hemisphere Distribution Marcel Dekker AG Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland tel: 41-61-261-8482; fax: 41-61-261-8896 World Wide Web http:/ /www.dekker.com The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special Sales/Professional Marketing at the headquarters address above. Copyright 2001 by Marcel Dekker, Inc.
All Rights Reserved.
Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage and retrieval system, without permission in writing from the publisher. Current printing (last digit): 10 9 8 7 6 5 4 3 2 1 PRINTED IN THE UNITED STATES OF AMERICA
Contributors
Marie-Isabel Aguilar, Ph.D. Department of Biochemistry and Molecular Biology, Monash University, Clayton, Victoria, Australia Victor G. Berezkin, Dr.Sc. Chromatography Laboratory, A. V. Topchiev Institute of Petrochemical Synthesis, Russian Academy of Sciences, Moscow, Russia Phyllis R. Brown, Ph.D. Department of Chemistry, University of Rhode Island, Kingston, Rhode Island Robert G. Carlson, Ph.D. Department of Chemistry, University of Kansas, Lawrence, Kansas Judy L. Carmody, Ph.D. Waters Corporation, Milford, Massachusetts Yung-Fong Cheng, Ph.D. Waters Corporation, Milford, Massachusetts ˚ ke Jo¨nsson, Ph.D. Department of Analytical Chemistry, Jan A Lund University, Lund, Sweden iii
iv / Contributors Hideko Kanazawa, Ph.D. Department of Physical Pharmaceutical Chemistry, Kyoritsu College of Pharmacy, Tokyo, Japan Francesca Lanza, Ph.D. Institut fu¨r Anorganische Chemie und Analytische Chemie, Johannes Gutenberg Universita¨t Mainz, Mainz, Germany Tzong-Hsien Lee, Ph.D.* Department of Biochemistry and Molecular Biology, Monash University, Clayton, Victoria, Australia Ziling Lu, Ph.D. Waters Corporation, Milford, Massachusetts Susan M. Lunte, Ph.D. Department of Pharmaceutical Chemistry, University of Kansas, Lawrence, Kansas Lennart Mathiasson, Ph.D. Department of Analytical Chemistry, Lund University, Lund, Sweden Yoshikazu Matsushima, Ph.D. Department of Physical Pharmaceutical Chemistry, Kyoritsu College of Pharmacy, Tokyo, Japan Uwe D. Neue, Ph.D. Waters Corporation, Milford, Massachusetts Teruo Okano, Ph.D. Institute of Biomedical Engineering, Tokyo Women’s Medical University, Tokyo, Japan Charles H. Phoebe, Ph.D. Waters Corporation, Milford, Massachusetts Ute Pyell, Ph.D. Department of Chemistry, Philipps-Universita¨t Marburg, Marburg, Germany Christina S. Robb, M.S. Department of Chemistry, University of Rhode Island, Kingston, Rhode Island
* Current affiliation: Liver Research Unit, Chang Gung Memorial Hospital, Taipei, Taiwan, Republic of China
Contributors / v Mark J. Rose, Ph.D.* Department of Pharmaceutical Chemistry, The University of Kansas, Lawrence, Kansas Corrado Sarzanini Department of Analytical Chemistry, University of Turin, Turin, Italy Bo¨rje Sellergren, Ph.D. Institut fu¨r Anorganische Chemie und Analytische Chemie, Johannes Gutenberg Universita¨t Mainz, Mainz, Germany John F. Stobaugh, Ph.D. Department of Pharmaceutical Chemistry, University of Kansas, Lawrence, Kansas Thomas E. Wheat, Ph.D. Waters Corporation, Milford, Massachusetts
*Current affiliation: Department of Drug Metabolism, Merck Research Laboratories, West Point, Pennsylvania
Contents
Contributors to Volume 41 Contents of Other Volumes 1.
iii xiii
Fundamentals of Capillary Electrochromatography
1
Ute Pyell I. II. III. IV. V. VI. VII.
Introduction Basic Concepts Theory Instrumental Developments Mobile-Phase Considerations Applications Concluding Remarks References
2. Membrane Extraction Techniques for Sample Preparation 53 ˚ ke Jo¨nsson and Lennart Mathiasson Jan A I. Sample Preparation Techniques II. Membrane Extraction Techniques vii
viii
/ Contents III. IV. V. VI.
3.
Interfacing Membrane Extraction and Separation What Can Be Achieved by Membrane Extraction? Fields of Application of Membrane Extraction Conclusions References
Design of Rapid Gradient Methods for the Analysis of Combinatorial Chemistry Libraries and the Preparation of Pure Compounds 93 Uwe D. Neue, Judy L. Carmody, Yung-Fong Cheng, Ziling Lu, Charles H. Phoebe, and Thomas E. Wheat I. II. III. IV.
4.
Introduction Theory Practice of Fast Chromatography Conclusion Symbols References
Molecular Imprinted Extraction Materials for Highly Selective Sample Clean-Up and Analyte Enrichment 137 Francesca Lanza and Bo¨rje Sellergren I. II. III. IV. V. VI. VII. VIII. IX.
Introduction Multipurpose SPE Phases High-Affinity SPE Phases Molecularly Imprinted Solid-Phase Extraction (MISPE) Previous MISPE Protocols The Development of New MISPE Protocols Polymer Synthesis-Related Factors Methods for Synthesis and Screening of Large Groups of MIPs Template Bleeding—An Unresolved Issue in MISPE Protocols
Contents / ix X. Conclusion References
5.
Biomembrane Chromatography: Application to Purification and Biomolecule–Membrane Interactions 175 Tzong-Hsien Lee and Marie-Isabel Aguilar I. Introduction II. Biomembrane-Modified Soft-Gel Chromatographic Supports III. Biomembrane-Modified Silica-Based Chromatographic Supports IV. Applications V. Conclusions and Future Directions References
6.
Transformation of Analytes for Electrochemical Detection: A Review of Chemical and Physical Approaches 203 Mark J. Rose, Susan M. Lunte, Robert G. Carlson, and John F. Stobaugh I. II. III. IV.
7.
Abbreviations Introduction Transformation Methods and Summary Tables NDTE References
High-Performance Liquid Chromatography: Trace Metal Determination and Speciation 249 Corrado Sarzanini I. Introduction II. Sample Handling III. Chromatographic Modes
x
/ Contents IV. Metal Speciation References
8.
Temperature-Responsive Chromatography
311
Hideko Kanazawa, Yoshikazu Matsushima, and Teruo Okano I. II. III. IV. V. VI. VII. VIII.
9.
Introduction Temperature-Responsive Polymers Temperature-Responsive Stationary Phases Temperature-Responsive Chromatography: Tunable Separation Temperature Effects on Retention Temperature Gradient: The Method Is Replacing Solvent Gradient Application to the Separation of Peptides and Proteins Conclusions References
Carrier Gas in Capillary Gas–Liquid Chromatography 337 Victor G. Berezkin I. Introduction II. Absolute Retention and Its Dependence on the Nature and Pressure of the Carrier Gas III. Relative Retention and Its Dependence on the Nature and Pressure of the Carrier Gas IV. Steam Capillary Gas–Liquid Chromatography and the Influence of Water Vapor on Relative Retention V. Influence of the Nature and Pressure of the Carrier Gas on the Separation VI. Effect of Carrier-Gas Solubility in SLP on Relative Retention
Contents / xi VII. Conclusion References
10. Catechins in Tea: Chemistry and Analysis
379
Christina S. Robb and Phyllis R. Brown I. II. III. IV. V. VI.
Index
Introduction The Chemical Composition of Tea Extracts Catechins Chemical Reactions of Catechins in Tea HPLC Analyses of Catechins Conclusion References
411
Contents of Other Volumes
Volumes 1–10
out of print
Volume 11 Quantitative Analysis by Gas Chromatography Josef Nova´k Polyamide Layer Chromatography Kung-Tsung Wang, Yau-Tang Lin, and Iris S. Y. Wang Specifically Adsorbing Silica Gels H. Bartels and P. Prijs Nondestructive Detection Methods in Paper and Thin-Layer Chromatography G. C. Barrett
Volume 12 The Use of High-Pressure Liquid Chromatography in Pharmacology and Toxicology Phyllis R. Brown Chromatographic Separation and Molecular-Weight Distributions of Cellulose and Its Derivatives Leon Segal Practical Methods of High-Speed Liquid Chromatography Gary J. Fallick Measurement of Diffusion Coefficients by Gas-Chromatography Broadening Techniques: A Review Virgil R. Maynard and Eli Grushka Gas-Chromatography Analysis of Polychlorinated Diphenyls and Other Nonpesticide Organic Pollutants Joseph Sherma High-Performance Electrometer Systems for Gas Chromatography Douglas H. Smith Steam Carrier Gas–Solid Chromatography Akira Nonaka
xiii
xiv / Contents of Other Volumes Volume 13
out of print
Volume 14 Nutrition: An Inviting Field to High-Pressure Liquid Chromatography Andrew J. Clifford Polyelectrolyte Effects in Gel Chromatography Bengt Stenlund Chemically Bonded Phases in Chromatography Imrich Sebestian and Istva´n Hala´sz Physicochemical Measurements Using Chromatography David C. Locke Gas–Liquid Chromatography in Drug Analysis W. J. A. VandenHeuvel and A. G. Zacchei The Investigation of Complex Association by Gas Chromatography and Related Chromatographic and Electrophoretic Methods C. L. de Ligny Gas–Liquid–Solid Chromatography Antonio De Corcia and Arnaldo Liberti Retention Indices in Gas Chromatography J. K. Haken
Volume 15 Detection of Bacterial Metabolites in Spent Culture Media and Body Fluids by Electron Capture Gas–Liquid Chromatography John B. Brooks Signal and Resolution Enhancement Techniques in Chromatography Raymond Annino The Analysis of Organic Water Pollutants by Gas Chromatography and Gas Chromatography–Mass Spectrometry Ronald A. Hites Hydrodynamic Chromatography and Flow-Induced Separations Hamish Small The Determination of Anticonvulsants in Biological Sampes by Use of HighPressure Liquid Chromatography Reginald F. Adams The Use of Microparticulate Reversed-Phase Packing in High-Pressure Liquid Chromatography of Compounds of Biological Interest John A. Montgomery, Thomas P. Johnston, H. Jeanette Thomas, James R. Piper, and Carroll Temple Jr. Gas–Chromatographic Analysis of the Soil Atmosphere K. A. Smith Kinematics of Gel Permeation Chromatography A. C. Ouano Some Clinical and Pharmacological Applications of High-Speed Liquid Chromatography J. Arly Nelson
Volume 16
out of print
Volume 17 Progress in Photometric Methods of Quantitative Evaluation in TLO V. Pollak Ion-Exchange Packings for HPLC Separations: Care and Use Fredric M. Rabel
Contents of Other Volumes / xv Micropacked Columns in Gas Chromatography: An Evaluation C. A. Cramers and J. A. Rijks Reversed-Phase Gas Chromatography and Emulsifier Characterization J. K. Haken Template Chromatography Herbert Schott and Ernst Bayer Recent Usage of Liquid Crystal Stationary Phases in Gas Chromatography George M. Janini Current State of the Art in the Analysis of Catechomalines Ante´ M. Krstulovic
Volume 18 The Characterization of Long-Chain Fatty Acids and Their Derivatives by Chromatography Marcel S. F. Lie Ken Jie Ion-Pair Chromatography on Normal- and Reversed-Phase Systems Milton T. W. Hearn Current State of the Art in HPLC Analyses of Free Nucleotides, Nucleosides, and Bases in Bological Fluids Phyllis R. Brown, Ante´ M. Krstulovic, and Richard A. Hartwick Resolution of Racemates by Ligand-Exchange Chromatography Vadim A. Danakov The Analysis of Marijuana Cannabinoids and Their Metabolites in Biological Media by GC and/or GC-MS Techniques Benjamin J. Gudzinowicz, Michael J. Gudzinowicz, Joanne Hologgitas, and James L. Driscoll
Volume 19 Roles of High-Performance Liquid Chromatography in Nuclear Medicine Steven How-Yan Wong Calibration of Separation Systems in Gel Permeation Chromatography for Polymer Characterization Josef Janc˘a Isomer-Specific Assay of 2,4-D Herbicide Products by HPLC: Regulaboratory Methodology Timothy S. Stevens Hydrophobic Interaction Chromatography Stellan Hjerte´n Liquid Chromatography with Programmed Composition of the Mobile Phase Pavel Jandera and Jaroslav Chura´cˇek Chromatographic Separation of Aldosterone and Its Metabolites David J. Morris and Ritsuko Tsai
Volume 20 High-Performance Liquid Chromatography and Its Application to Protein Chemistry Milton T. W. Hearn Chromatography of Vitamin D3 and Metabolites K. Thomas Koshy High-Performance Liquid Chromatography: Applications in a Children’s Hospital Steven J. Soldin
xvi / Contents of Other Volumes The Silica Gel Surface and Its Interactions with Solvent and Solute in Liquid Chromatography R. P. W. Scott New Developments in Capillary Columns for Gas Chromatography Walter Jennings Analysis of Fundamental Obstacles to the Size Exclusion Chromatography of Polymers of Ultrahigh Molecular Weight J. Calvin Giddings
Volume 21 High-Performance Liquid Chromatography/Mass Spectrometry (HPLC/MS) David E. Grimes High-Performance Liquid Affinity Chromatography Per-Olof Larsson, Magnus Glad, Lennart Hansson, Mats-Olle Ma˚nsson, Sten Ohlson, and Klaus Mosbach Dynamic Anion-Exchange Chromatography Roger H. A. Sorel and Abram Holshoff Capillary Columns in Liquid Chromatography Daido Ishii and Toyohide Takeuchi Droplet Counter-Current Chromatography Kurt Hostettmann Chromatographic Determination of Copolymer Composition Sadao Mori High-Performance Liquid Chromatography of K Vitamins and Their Antagonists Martin J. Shearer Problems of Quantitation in Trace Analysis by Gas Chromatography Josef Nova´k
Volume 22 High-Performance Liquid Chromatography and Mass Spectrometry of Neuropeptides in Biologic Tissue Dominic M. Desiderio High-Performance Liquid Chromatography of Amino Acids: Ion-Exchange and Reversed-Phase Strategies Robert F. Pfeifer and Dennis W. Hill Resolution of Racemates by High-Performance Liquid Chromatography Vadium A. Davankov, Alexander A. Kurganov, and Alexander S. Bochkov High-Performance Liquid Chromatography of Metal Complexes Hans Veening and Bennett R. Willeford Chromatography of Carotenoids and Retinoids Richard F. Taylor High Performance Liquid Chromatography Zbyslaw J. Petryka Small-Bore Columns in Liquid Chromatography Raymond P. W. Scott
Volume 23 Laser Spectroscopic Methods for Detection in Liquid Chromatography Edward S. Yeung Low-Temperature High-Performance Liquid Chromatography for Separation of Thermally Labile Species David E. Henderson and Daniel J. O’Connor Kinetic Analysis of Enzymatic Reactions Using High-Performance Liquid Chromatography Donald L. Sloan
Contents of Other Volumes / xvii Heparin-Sepharose Affinity Chromatography Akhlaq A. Farooqui and Lloyd A. Horrocks New Developments in Capillary Columns for Gas Chromatography Walter Jennings
Volume 24 Some Basic Statistical Methods for Chromatographic Data Karen Kafadar and Keith R. Eberhardt Multifactor Optimization of HPLC Conditions Stanley N. Deming, Julie G. Bower, and Keith D. Bower Statistical and Graphical Methods of Isocratic Solvent Selection for Optimal Separation in Liquid Chromatography Haleem J. Issaq Electrochemical Detectors for Liquid Chromatography Ante M. Krstulovic´, Henri Colin, and Georges A. Guiochon Reversed-Flow Gas Chromatography Applied to Physicochemical Measurements Nicholas A. Katsanos and George Karaiskakis Development of High-Speed Countercurrent Chromatography Yoichiro Ito Determination of the Solubility of Gases in Liquids by Gas–Liquid Chromatography Jon F. Parcher, Monica L. Bell, and Ping L. Jin Multiple Detection in Gas Chromatography Ira S. Krull, Michael E. Swartz, and John N. Driscoll
Volume 25 Estimation of Physicochemical Properties of Organic Solutes Using HPLC Retention Parameters Theo L. Hafkenscheid and Eric Tomlinson Mobile Phase Optimization in RPLC by an Iterative Regression Design Leo de Galan and Hugo A. H. Billiet Solvent Elimination Techniques for HPLC/FT-IR Peter R. Griffiths and Christine M. Conroy Investigations of Selectivity in RPLC of Polycyclic Aromatic Hydrocarbons Lane C. Sander and Stephen A. Wise Liquid Chromatographic Analysis of the Oxo Acids of Phosphorus Roswitha S. Ramsey Liquid Chromatography of Carbohydrates Toshihiko Hanai HPLC Analysis of Oxypurines and Related Compounds Katsuyuki Nakano HPLC of Glycosphingolipids and Phospholipids Robert H. McCluer, M. David Ullman, and Firoze B. Jungalwala
Volume 26 RPLC Retention of Sulfur and Compounds Containing Divalent Sulfur Hermann J. Mo¨ckel
xviii / Contents of Other Volumes The Application of Fleuric Devices to Gas Chromatographic Instrumentation Raymond Annino High Performance Hydrophobic Interaction Chromatography Yoshio Kato HPLC for Therapeutic Drug Monitoring and Determination of Toxicity Ian D. Watson Element Selective Plasma Emission Detectors for Gas Chromatography A. H. Mohamad and J. A. Caruso The Use of Retention Data from Capillary GC for Qualitative Analysis: Current Aspects Lars G. Blomberg Retention Indices in Reversed-Phase HPLC Roger M. Smith HPLC of Neurotransmitters and Their Metabolites Emilio Gelpi
Volume 27 Physicochemical and Analytical Aspects of the Adsorption Phenomena Involved in GLC Victor G. Berezkin HPLC in Endocrinology Richard L. Patience and Elizabeth S. Penny Chiral Stationary Phases for the Direct LC Separation of Enantiomers William H. Pirkle and Thomas C. Pochapsky The Use of Modified Silica Gels in TLC and HPTLC Willi Jost and Heinz E. Hauck Micellar Liquid Chromatography John G. Dorsey Derivation in Liquid Chromatography Kazuhiro Imai Analytical High-Performance Affinity Chromatography Georgio Fassina and Irwin M. Chaiken Characterization of Unsaturated Aliphatic Compounds by GC/Mass Spectrometry Lawrence R. Hogge and Jocelyn G. Millar
Volume 28 Theoretical Aspects of Quantitative Affinity Chromatography: An Overview Alain Jaulmes and Claire Vidal-Madjar Column Switching in Gas Chromatography Donald E. Willis The Use and Properties of Mixed Stationary Phases in Gas Chromatography Gareth J. Price On-Line Small-Bore-Chromatography for Neurochemical Analysis in the Brain William H. Church and Joseph B. Justice, Jr. The Use of Dynamically Modified Silica in HPLC as an Alternative to Chemically Bonded Materials Per Helboe, Steen Honore´ Hansen, and Mogens Thomsen Gas Chromatographic Analysis of Plasma Lipids Arnis Kuksis and John J. Myher HPLC of Penicillin Antibiotics Michel Margosis
Contents of Other Volumes / xix Volume 29 Capillary Electrophoresis Ross A. Willingford and Andrew G. Ewing Multidimensional Chromatography in Biotechnology Daniel F. Samain High-Performance Immunoaffinity Chromatography Terry M. Phillips Protein Purification by Multidimensional Chromatography Stephen A. Berkowitz Fluorescence Derivitization in High-Performance Liquid Chromatography Yosuke Ohkura and Hitoshi Nohta
Volume 30 Mobile and Stationary Phases for Supercritical Fluid Chromatography Peter J. Schoenmakers and Louis G. M. Uunk Polymer-Based Packing Materials for Reversed-Phase Liquid Chromatography Nobuo Tanaka and Mikio Araki Retention Behavior of Large Polycyclic Aromatic Hydrocarbons in Reversed-Phase Liquid Chromatography Kiyokatsu Jinno Miniaturization in High-Performance Liquid Chromatography Masashi Goto, Toyohide Takeuchi, and Daido Ishii Sources of Errors in the Densitometric Evaluation of Thin-Layer Separations with Special Regard to Nonlinear Problems Victor A. Pollak Electronic Scanning for the Densitometric Analysis of Flat-Bed Separations Viktor A. Pollak
Volume 31 Fundamentals of Nonlinear Chromatography: Prediction of Experimental Profiles and Band Separation Anita M. Katti and Georges A. Guiochon Problems in Aqueous Size Exclusion Chromatography Paul L. Dubin Chromatography on Thin Layers Impregnated with Organic Stationary Phases Jiri Gasparic Countercurrent Chromatography for the Purification of Peptides Martha Knight Boronate Affinity Chromatography Ram P. Singhal and S. Shyamali M. DeSilva Chromatographic Methods for Determining Carcinogenic Benz(c)-acridine Noboru Motohashi, Kunihiro Kamata, and Roger Meyer
Volume 32 Porous Graphitic Carbon in Biomedical Applications Chang-Kee Lim Tryptic Mapping by Reversed Phase Liquid Chromatography Michael W. Dong Determination of Dissolved Gases in Water by Gas Chromatography Kevin Robards, Vincent R. Kelly, and Emilios Patsalides Separation of Polar Lipid Classes into Their Molecular Species Components by Planar and Column Liquid Chromatography V. P. Pchelkin and A. G. Vereshchagin
xx
/ Contents of Other Volumes
The Use of Chromatography in Forensic Science Jack Hubball HPLC of Explosives Materials John B. F. Lloyd
Volume 33 Planar Chips Technology of Separation Systems: A Developing Perspective in Chemical Monitoring Andreas Manz, D. Jed Harrison, Elizabeth Verpoorte, and H. Michael Widmer Molecular Biochromatography: An Approach to the Liquid Chromatographic Determination of Ligand-Biopolymer Interactions Irving W. Wainer and Terence A. G. Noctor Expert Systems in Chromatography Thierry Hamoir and D. Luc Massart Information Potential of Chromatographic Data for Pharmacological Classification and Drug Design Roman Kaliszan Fusion Reaction Chromatography: A Powerful Analytical Technique for Condensation Polymers John K. Haken The Role of Enantioselective Liquid Chromatographic Separations Using Chiral Stationary Phases in Pharmaceutical Analysis Shulamit Levin and Saleh Abu-Lafi
Volume 34 High-Performance Capillary Electrophoresis of Human Serum and Plasma Proteins Oscar W. Reif, Ralf Lausch, and Ruth Freitag Analysis of Natural Products by Gas Chromatography/Matrix Isolation/Infrared Spectrometry W. M. Coleman III and Bert M. Gordon Statistical Theories of Peak Overlap in Chromatography Joe M. Davis Capillary Electrophoresis of Carbohydrates Ziad El Rassi Environmental Applications of Supercritical Fluid Chromatography Leah J. Mulcahey, Christine L. Rankin, and Mary Ellen P. McNally HPLC of Homologous Series of Simple Organic Anions and Cations Norman E. Hoffman Uncertainty Structure, Information Theory, and Optimization of Quantitative Analysis in Separation Science Yuzuru Hayashi and Rieko Matsuda
Volume 35 Optical Detectors for Capillary Electrophoresis Edward S. Yeung Capillary Electrophoresis Coupled with Mass Spectrometry Kenneth B. Tomer, Leesa J. Deterding, and Carol E. Parker Approaches for the Optimization of Experimental Parameters in Capillary Zone Electrophoresis Haleem J. Issaq, George M. Janini, King C. Chan, and Ziad El Rassi
Contents of Other Volumes / xxi Crawling Out of the Chiral Pool: The Evolution of Pirkle-Type Chiral Stationary Phases Christopher J. Welch Pharmaceutical Analysis by Capillary Electrophoresis Sam F. Y. Li, Choon Lan Ng, and Chye Pend Ong Chromatographic Characterization of Gasolines Richard E. Pauls Reversed-Phase Ion-Pair and Ion-Interaction Chromatography M. C. Gennaro Error Sources in the Determination of Chromatographic Peak Size Ratios Veronika R. Meyer
Volume 36 Use of Multivariate Mathematical Methods for the Evaluation of Retention Data Matrices Tibor Cserha´ti and Esther Forga´cs Separation of Fullerenes by Liquid Chromatography: Molecular Recognition Mechanism in Liquid Chromatographic Separation Kiyokatsu Jinno and Yoshihiro Saito Emerging Technologies for Sequencing Antisense Oligonucleotides: Capillary Electrophoresis and Mass Spectrometry Aharon S. Cohen, David L. Smisek, and Bing H. Wang Capillary Electrophoretic Analysis of Glycoproteins and Glycoprotein-Derived Oligosaccharides Robert P. Oda, Benjamin J. Madden, and James P. Landers Analysis of Drugs of Abuse in Biological Fluids by Liquid Chromatography Steven R. Binder Electrochemical Detection of Biomolecules in Liquid Chromatography and Capillary Electrophoresis Jian-Ge Chen, Steven J. Woltman, and Steven G. Weber The Development and Application of Coupled HPLC-NMR Spectroscopy John C. Lindon, Jeremy K. Nicholson, and Ian D. Wilson Microdialysis Sampling for Pharmacological Studies: HPLC and CE Analysis Susan M. Lunte and Craig E. Lunte
Volume 37 Assessment of Chromatographic Peak Purity Muhammad A. Sharaf Fluorescence Detectors in HPLC Maria Brak Smalley and Linda B. McGown Carbon-Based Packing Materials for Liquid Chromatography: Structure, Performance, and Retention Mechanisms John H. Knox and Paul Ross Carbon-Based Packing Materials for Liquid Chromatography: Applications Paul Ross and John H. Knox Directly Coupled (On-Line) SFE-GC: Instrumentation and Applications Mark D. Burford, Steven B. Hawthorne, and Keith D. Bartle
xxii
/ Contents of Other Volumes
Sample Preparation for Gas Chromatography with Solid-Phase Extraction and Solid-Phase Microextraction Zelda E. Penton Capillary Electrophoresis of Proteins Tim Wehr, Robert Rodriguez-Diaz, and Cheng-Ming Liu Chiral Micelle Polymers for Chiral Separations in Capillary Electrophoresis Crystal C. Williams, Shahab A. Shamsi, and Isiah M. Warner Analysis of Derivatized Peptides Using High-Performance Liquid Chromatography and Capillary Electrophoresis Kathryn M. De Antonis and Phyllis R. Brown
Volume 38 Band Spreading in Chromatography: A Personal View John H. Knox The Stochastic Theory of Chromatography Francesco Dondi, Alberto Cavazzini, and Maurizio Remelli Solvating Gas Chromatography Using Packed Capillary Columns Yufeng Shen and Milton L. Lee The Linear-Solvent-Strength Model of Gradient Elution L. R. Snyder and J. W. Dolan High-Performance Liquid Chromatography-Pulsed Electrochemical Detection for the Analysis of Antibiotics William R. LaCourse and Catherine O. Dasenbrock Theory of Capillary Zone Electrophoresis H. Poppe Separation of DNA by Capillary Electrophoresis Andra´s Guttman and Kathi J. Ulfelder
Volume 39 Theory of Field Flow Fractionation Michel Martin Particle Simulation Methods in Separation Science Mark R. Schure Mathematical Analysis of Multicomponent Chromatograms Attila Felinger Determination of Association Constants by Chromatography and Electrophoresis Daniel W. Armstrong Method Development and Selectivity Optimization in High-Performance Liquid Chromatography H. A. H. Billet and G. Rippel Chemical Equilibria in Ion Chromatography: Theory and Applications Pe´ter Hajo´s, Otto´ Horva´th, and Gabriella Re´ve´sz Fundamentals and Simulated Moving Bed Chromatography Under Linear Conditions Guoming Zhong and Georges Guiochon
Volume 40 Fundamental Interpretation of the Peak Profiles in Linear Reversed-Phase Liquid Chromatography Kanji Miyabe and Georges Guiochon Dispersion in Micellar Electrokinetic Chromatography Joe M. Davis
Contents of Other Volumes / xxiii In Search of a Chromatographic Model for Biopartitioning Colin F. Poole, Salwa K. Poole, and Ajith D. Gunatilleka Advances in Physico-chemical Measurements Using Inverse Gas Chromatography Nicholas A. Katsanos and Fani Roubani-Kalantzopoulou Fundamental Aspects of Aerosol-Based Light-Scattering Detectors for Separations John A. Koropchak, Salma Sadain, Xiaohui Yang, Lars-Erik Magnusson, Mari Heybroek, and Michael P. Anisimov New Developments in Liquid-Chromatographic Stationary Phases Toshihiko Hanai Non-Silica-Based Supports in Liquid Chromatography of Bioactive Compounds Esther Forga´cs and Tibor Cserha´ti Overview of the Surface Modification Techniques for the Capillary Electrophoresis of Proteins Marie-Claude Millot and Claire Vidal-Madjar Continuous Bed for Conventional Column and Capillary Column Chromatography Jia-li Liao Countercurrent Chromatography: Fundamentally a Preparative Tool Alain Berthod and Beatrice Billardello Analysis of Oligonucleotides by ESI-MS Dieter L. Deforce and Elfriede G. Van den Eeckhout Determination of Herbicides in Water Using HPLC–MS Techniques G. D’Ascenzo, F. Bruno, A. Gentili, S. Marchese, and D. Perret Effect of Adsorption Phenomena on Retention Values in Capillary Gas–Liquid Chromatography Victor G. Berezkin
1 Fundamentals of Capillary Electrochromatography Ute Pyell Philipps-Universita¨t Marburg, Marburg, Germany
I. INTRODUCTION II. BASIC CONCEPTS A. Instrumental Setup B. Propulsion of Mobile Phase
2 3 3 5
III.
THEORY A. Migration principles B. Efficiency C. Comparison to Liquid Chromatography D. Extracolumn Band Broadening E. Thermal Effects F. Bubble Formation
10 10 13 18 19 24 27
IV.
INSTRUMENTAL DEVELOPMENTS A. Column Technology B. Detection C. Gradient Elution
29 29 32 36
V.
MOBILE-PHASE CONSIDERATIONS
38
APPLICATIONS
42
VI.
1
2 / Pyell VII. CONCLUDING REMARKS REFERENCES
43 44
I. INTRODUCTION Capillary electrochromatography (CEC) can be considered as a hybrid of capillary electrophoresis (CE) and high-performance liquid chromatography (HPLC). From another point of view, CEC is a liquid chromatographic method in which the mobile phase is electroosmotically driven through the chromatographic bed. The interest in CEC stems from the underlying idea that the unique properties of the electroosmotic flow (EOF) make it possible to realize highly efficient liquid chromatographic separation systems that can overcome the peak capacity limitations of current HPLC. HPLC is a method with much lower chromatographic efficiency and peak capacity than open-tubular gas chromatography (OTGC), which can be regarded as the method of choice for those analytes that can be transferred into the gaseous phase. For those analytes that are not sufficiently volatile, however, OTGC is not an applicable alternative to HPLC. The need to improve the chromatographic efficiency and the peak capacity for nonvolatile analytes makes the development of alternatives to classical HPLC one of the major fields of activity in contemporary chromatographic research. The following techniques can be regarded as more efficient alternatives to HPLC: (1) supercritical fluid chromatography (SFC) [1], (2) capillary electrophoresis (CE), (3) micellar electrokinetic chromatography (MEKC) [2], and (4) capillary electrochromatography (CEC). Of these techniques, CEC is the closest to HPLC because the stationary and mobile phases can be identical in HPLC and CEC. In the case of ionic, acidic, or basic analytes, CEC offers (in analogy to MEKC) the possibility to combine chromatographic separation with separation due to electrophoresis, making it possible to separate strongly basic, moderately basic, weakly basic, neutral, weakly acidic, moderately acidic, and strongly acidic compounds in one run [3]. According to Colo´n et al. [4], the term electrochromatography was introduced by Berraz [5] to describe a form of paper electrophoresis. In 1952 Mould and Synge [6,7] demonstrated the use of electroosmotic flow in a thin-layer chromatographic system for the separation of polysaccharides in collodion membrane strips. Already in 1974, Pretorius et al. [8] realized the possible advantages of a liquid
Fundamentals of Capillary Electrochromatography / 3 chromatographic technique with an electroosmotically driven mobile phase. They introduced high-speed thin-layer chromatography (HSTLC) as a variant of thin-layer chromatography (TLC) with an electroosmotically driven mobile phase and much shorter run times than were obtained with traditional techniques. They also applied the principle of an electroosmotically driven mobile phase to liquid column chromatography with promising results. In the course of the development of capillary electrophoresis [9,10], the idea of electroosmotic pumping of mobile phase through a packed bed was taken up again. In their landmark paper in 1981, Jorgenson and Lukacs [11] showed for the first time the possibility of performing reversed-phase chromatography in packed capillaries employing virtually the same apparatus as developed for CE. Since these pioneering works, development of CEC has started and several names have been coined for this type of capillary separation technique: liquid chromatography with electroosmotic flow [12], electroendosmotically driven liquid chromatography [13], electrically driven liquid chromatography [14], electroendosmotic capillary chromatography [15], electrokinetic chromatography with packed capillaries [16], and capillary electrochromatography [17,18]. The last term is now generally accepted and is the one most widely used referring to a separation technique in which the separation is achieved by retentive interaction of the solutes to be separated with an immobilized second phase (the stationary phase), while a liquid mobile phase is employed that is driven through the chromatographic bed predominantly via the effect of electroosmosis.
II. BASIC CONCEPTS A. Instrumental Setup CEC can be performed with identical experimental setup as used for CE and MEKC except for the need to use a capillary that contains an immobilized second phase which is responsible for the chromatographic separation. Figure 1 depicts a block diagram of an apparatus used for CEC. In most cases the mobile phase is a solution of buffering salts in a mixture of an organic modifier with water. With these very polar mobile phases, reversed-phase separations are achieved. No pump is needed. Injection of the sample is performed directly into the high-voltage end of the separation capillary, mostly by electrokinetic injection. In this case, the vessel with mobile phase
4 / Pyell
Fig. 1 Schematic representation of CEC equipment.
is replaced by a vessel containing the sample and a programmed trapezium-shaped voltage ramp is applied for a few seconds. With electrokinetic injection the sample is dragged into the capillary by the effect of electroosmosis. In the case of neutral analytes, no discrimination of sample constituents takes place. The simplicity of electrokinetic injection compared to the difficulty of injecting a few nanoliters via a mechanical injection device into a stream of pressurized mobile phase constitutes one important advantage of CEC over micro-HPLC. Various types of separation capillaries have been used so far. The preparation of separation capillaries by packing with reversedphase silica gels (virtually exclusively octadecyl silica gel) has received the most attention in the early studies devoted to CEC. With packed capillaries, frits are necessary at both ends of the packed bed to keep the stationary phase migrating out of the capillary. However, the manufacture of inlet and outlet frits to retain the stationary phase is not a trivial task and is associated with many problems that will be discussed in a later section. Consequently, there is increased interest in the production of fritless separation capillaries. One approach to avoid the use of frits is open-tubular columns that are coated on the inner surface with a layer of stationary phase. Other approaches comprise the synthesis of soft gels anchored covalently
Fundamentals of Capillary Electrochromatography / 5 to the capillary walls or the synthesis of mechanically immobilized rigid gels. In order to avoid intolerable extracolumn band broadening, detection has to be performed in a segment of the capillary, transforming this segment into the detection cell of the detector. Independent of the type of separation capillary used, the detection cell can be placed in an open section of the capillary behind the chromatographic bed (on-column detection), or detection is directly in a segment of the chromatographic bed itself (in-column detection). In Fig. 2 the differences between these two detection modes are illustrated taking a packed capillary as an example.
B. Propulsion of Mobile Phase Ideally, in CEC the mobile phase is driven through the chromatographic bed exclusively by the effect of electroosmosis. Electroosmosis is an electrokinetic effect originating from the electrophoretic movement of the diffuse layer of the electric double layer that is formed at the interface between the liquid and the solid. In CE, electroosmosis is generated only at the electric double layer formed
Fig. 2 Difference between (a) in-column and (b) on-column detection in a packed capillary.
6 / Pyell at the inner capillary wall. In CEC with packed or gel-filled capillaries, however, the inner surface generated by the porous plug inside the capillary is much larger than the inner surface of the capillary wall (see Fig. 3). Consequently, properties of the inner surface of the porous plug, and not properties of the inner surface of the capillary wall, determine the obtainable electroosmotic velocity. These considerations have been convincingly corroborated by experiments of Dittmann and Rozing [19], who have packed coated and noncoated capillaries with identical stationary phase and found virtually no dependence of the determined electroosmotic mobility on the existence or nonexistence of the coating, although in open-tubular capillaries the electroosmotic velocity was effectively suppressed by the coating (see Fig. 4). The main driving force behind the interest of pioneering workers in CEC has been the fact that the electroosmotically driven flow in a porous plug (i.e., a capillary packed with particles) is virtually independent of the mean channel diameter. Smoluchowski has shown that Eq. (1) retains its validity if an open capillary is replaced by a porous plug [20].
Fig. 3 Generation of electroosmotic flow in a packed capillary.
Fundamentals of Capillary Electrochromatography / 7
Fig. 4 Comparison of chromatograms obtained for the separation of neutral solutes with a packed (top trace) polyvinylalcohol-coated fused-silica capillary or a (mirror trace) bare fused-silica capillary. Capillary, 335 (250) mm ⫻ 100 µm; stationary phase, 3-µm porous octadecyl silica gel (CEC Hypersil C18); mobile phase, 80% acetonitril, 20% aqueous buffer (25 mmol/L TRISHCl, pH 8); temperature ⫽ 20°C; 1 ⫽ thiourea, 2 ⫽ ethyl paraben, 3 ⫽ propyl paraben, 4 ⫽ butyl paraben, 5 ⫽ pentyl paraben, 6 ⫽ naphthalene, 7 ⫽ hexyl paraben, 8 ⫽ fluorene, 9 ⫽ anthracene, 10 ⫽ anthracene, 11 ⫽ fluoranthene. (Reprinted with permission from Ref. 19.)
V εD ε 0 ζ ⫽ i ηλ 0
(1)
where V ⫽ volume of liquid displaced per unit time, i ⫽ electric current strength, εD ⫽ dielectric constant of the bulk liquid, ε 0 ⫽ electric permittivity of vacuum, ζ ⫽ electrokinetic potential (zeta potential), η ⫽ viscosity of the bulk liquid, and λ 0 ⫽ specific electric conductivity of the liquid. Equation (1) can be used to describe the electroosmotic flow through a porous plug, provided that the flow is laminar, the local
8 / Pyell radius of curvature of the particles and the size of the pores are large compared to the thickness of the double layer, and the effects of surface conductance are negligible. Porous plugs may be made of particles of quite irregular geometry, nearly spherical particles, fibers, bundles of capillaries, or porous rods [21]. However, in CEC the velocity of the mobile phase is generally not given as flow rate (volume/ time) but as linear velocity (distance/time) determined via a nonretarded marker. By most workers, it is assumed that this linear velocity is identical or at least proportional to the electroosmotic velocity veo. veo ⫽
εD ε 0 ζ ⋅E η
(2)
where E ⫽ local electric field strength. Provided the size of the pores is large compared to the thickness of the electric double layer, the linear velocity is virtually independent of the mean channel diameter and directly proportional to the local electric field strength. This property of the electroosmotic flow makes it possible to employ separation columns in CEC that cannot be used (because of their high streaming resistance) in pressuredriven LC. Knox and Grant [13] examined the experimental conditions under which veo is independent of the mean channel diameter. Assuming that the mean channel diameter must be larger than 10 times the thickness of the electric double layer δ, they predicted for slurry-packed capillaries that the particle diameter dp has to be at least 40 times the thickness of the electric double layer δ and for drawn-packed capillaries dp ⱖ 20δ. The thickness of the electric double layer δ is a function of the electric permittivity ε of the liquid phase and of the ionic strength I ⫽ 1/2 ∑i z2i ci [13]: δ⫽
√
εkT 1000 NA e2 ∑i z2i ci
(3)
where k ⫽ Boltzmann constant, T ⫽ temperature (in kelvin), NA ⫽ Avogadro constant, e ⫽ charge of electron, zi ⫽ valence number, and ci ⫽ molar concentration. Employing Eq. (3), Knox and Grant [13] calculated that for a mobile phase (solution of a completely dissociated salt in water) con-
Fundamentals of Capillary Electrochromatography / 9 taining a 1:1 electrolyte at a concentration of 1 mmol/L, δ ⫽ 10 nm; and for a slurry-packed capillary the minimum particle diameter is 0.4 µm. In a later paper [18], they corroborated these predictions by experimental results employing capillaries packed with silica gels and octadecyl silica gels of particle diameters ranging from 1.5 to 50 µm. In their studies, the electroosmotic velocity was virtually independent of the particle diameter. With a stationary phase (beads of 0.5-µm diameter) synthesized by means of a modified Sto¨ber process, Lu¨dtke et al. [22] proved the applicability of packings with such fine material in CEC. The electroosmotic mobility observed with these packings is not significantly reduced compared to packings with material of larger mean particle diameter. Regarding the plate-height equation for a packed column (pressure-driven mobile phase) [23–25], H ⫽ 2λp dp ⫹
2γ DM (1 ⫹ 6k ⫹ 11k 2 )d 2p 8kd 2F ⫹ v v ⫹ v 24(1 ⫹ k) 2 Dm π 2 (1 ⫹ k) 2 Ds
(4)
where H ⫽ height of a theoretical plate, λp ⫽ packing factor, dp ⫽ mean particle diameter, γ ⫽ labyrinth factor, v ⫽ linear velocity of mobile phase, DM ⫽ diffusion coefficient in the mobile phase, k ⫽ retention factor of solute, dF ⫽ effective film thickness of the stationary phase, and DS ⫽ diffusion coefficient in the stationary phase, the tremendous impact of the mean particle diameter on the chromatographic efficiency obtainable with a given column is obvious. While the constraint for independence of the electroosmotic velocity of the mean channel diameter is fulfilled for the interparticle pores, in the case of porous packing materials with mean pore diameters of 6–10 nm it is assumed that there is virtually no electroosmotic flow in the intraparticular pores because of double-layer overlap. Another important feature of the electroosmotic flow is its flat streaming profile (compared to the parabolic streaming profile generated by pressure difference-induced laminar flow) reducing band broadening due to mass transfer resistance in the mobile phase, and reducing the A term of the plate-height equation. The different streaming profiles generated in the electro-driven and the pressuredriven modes are depicted in Fig. 5. This figure also illustrates the reduction of the A term of the plate-height equation through independence of flow velocity of the channel diameter.
10 / Pyell
Fig. 5 Flow profiles generated by (a) hydrodynamic flow, (b) electroosmosis in two capillaries (1) and (2) of different inner diameter (velocities of liquid segments are represented by arrows).
III. THEORY A.
Migration Principles
In CEC the linear velocity of the mobile phase is determined via a ‘‘nonretarded marker.’’ No studies are available concerning the accuracy of those measurements. With reversed-phase packing material, thiourea is commonly used as ‘‘nonretarded marker.’’ Generally, the so-determined linear velocity is treated as an equivalent of the electroosmotic velocity veo. However, in the case of on-column detection (open section of the capillary behind the chromatographic bed, see Fig. 2), differences in the linear flow velocity in the chromatographic bed and in the open section have to be taken into account. In the chromatographic bed the mobile phase takes up only a fraction ϕM of the inner capillary volume. Hence, the velocity of the mobile phase in the open section equals veo in the chromatographic bed multiplied by ϕM. The corrected linear velocity in the chromatographic bed can be calculated using Eq. 5 [26]. veo (corrected) ⫽
Lb ⫹ Lo /ϕM t0
(5)
where Lb ⫽ length of the chromatographic bed, Lo ⫽ length of the open section between the end of the chromatographic bed and the
Fundamentals of Capillary Electrochromatography / 11 detection window, and t 0 ⫽ migration time of the nonretarded marker. In the case of noncharged analytes, the separation mechanism is identical to that in liquid chromatography. The chromatographic retention factor k for a retarded solute can be calculated from the migration time of a nonretarded marker and the migration time of the retarded solute. This retention factor is identical to K ⋅ (Vs /Vm) (where K ⫽ concentration distribution coefficient, Vs /Vm ⫽ volume ratio in the chromatographic bed: volume of stationary phase/volume of mobile phase). In HPLC it is commonly assumed that the fraction of time spent outside the chromatographic bed with respect to the total migration time tm can be neglected. In CEC with oncolumn detection, however, this approximation might not be valid, especially in the case of short packed sections of the separation capillary [22]. In this case, the retention factor must be corrected using Eq. (6) [26]. k(corrected) ⫽
veo (corr.) ⋅ tm Lo ⫺ ⫺1 Lb LbϕM
(6)
where tm ⫽ migration time of the retarded solute. Euerby et al. [27] have shown that for noncharged solutes there is a negative linear relationship between the logarithm of the retention factor and the volume fraction of acetonitrile in the mobile phase (see Fig. 6). This relationship is known for liquid chromatographic separations. Taking into account that the velocity of the mobile phase is dependent on properties of the mobile phase [see Eq. (2)], optimization strategies in CEC for the separation of noncharged solutes can in part go back to those developed for HPLC. In the case of ionized solutes, however, the situation is completely different. In this case, the separation is effected by both electrophoretic and chromatographic principles. Considering acids or bases as solutes and reversed-phase material as stationary phase, the following simplifying assumptions can be made: Only the noncharged species contributes to the interaction of the solute with the nonpolar stationary phase. The migration velocity in the mobile phase can be regarded as superposition of electroosmosis and electrophoresis, while the electric field strength is constant along the separation capillary.
12 / Pyell
Fig. 6 Dependence of the logarithm of the retention factor (ln k) for neutral solutes on the volume fraction of acetonitrile in the mobile phase. Capillary, 250 mm ⫻ 50 µm; stationary phase, 3-µm porous octadecyl silica gel (Spherisorb ODS1); mobile phase, acetonitrile/water/aqueous buffer (50 mmol/L TRIS-HCl, pH 7.8). (Reprinted with permission from Ref. 27.)
If these assumptions are fulfilled, the migration velocity vm of a solute zone in the chromatographic bed is given by [28] vm ⫽
veo ⫹ vep (1 ⫹ k)
(7)
where vep ⫽ electrophoretic velocity, vep ⫽ µep ⋅ F (µep ⫽ effective electrophoretic mobility, F ⫽ electric field strength). According to Eq. (7), solutes can be separated by CEC that do differ not in k but in µep . It is important to note that in case of ionized solutes the true retention factor cannot be determined in a single experiment from to and tm . Equation (6) is no loger valid. The effective electrophoretic mobility has to be determined in a separate experiment in an open-tubular capillary using the mobile phase of the chromatographic experiment as separation buffer. Equation (7) implies that in the case of ionizible solutes, optimization strategies have to be completely different from those devel-
Fundamentals of Capillary Electrochromatography / 13 oped for HPLC or CE. However, the large number of influencing factors offers many parameters for selectivity tuning [3]. Rathore and Horva´th [29] have highlighted that in the case of so-called duplex columns having an open segment between the end of the chromatographic bed and the detection window, the observed selectivity for ionized solutes depends on the length ratio (length of chromatographic bed/length of open section), as in the open section only electrophoresis contributes to the separation, while in the chromatographic bed both electrophoretic and chromatographic effects affect the separation of zones. It has to be emphasized that the validity of Eq. (7) depends on the constraint that all simplifying assumptions listed above are met. This constraint is not fulfilled if the chromatographic retention mechanism is ion exchange (for example, separation of cations on a cation exchanger). Recently, Sta˚hlberg [30] has given a theory for zone migration in CEC for this generalized case. This theory is based on a solution of the mass balance equation. He showed that in the general case, the mixing of chromatographic and electrophoretic effects gives rise to strong nonlinear effects. Indeed, experiments with ion-exchange phases as stationary phases in CEC have brought unexpected results [31].
B. Efficiency The first works in CEC done with capillaries packed with octadecyl silica gel concentrated on progress in efficiency compared to HPLC via the use of very fine packing material (dp ⱕ 3 µm). These experiments have been made with neutral analytes in order to avoid electrophoretic effects. The maximum length of the packed bed is given not by the mechanical permeability of the packing but by the maximum voltage that can be applied between the ends of the capillary and by the electroosmotic mobility µeo. In order to exploit the full potential of the separation system, a minimum linear velocity of the mobile phase has to be obtained. This minimum linear velocity corresponds to the optimum velocity of the plate-height curve. If the minimum linear velocity vmin is 1.5 mm/s, the maximum voltage Umax ⫽ 30 kV and µeo ⫽ 0.25 cm2 /skV, then the maximum total length Ltot,max of the separation capillary is 500 mm with vmin ⫽ µeo (Umax /Ltot,max ). Knox and Grant [13] postulated that in CEC with packed capillaries the reduced plate height can be significantly smaller than 2 because of the flat mobile-phase streaming profile. These theoretical
14 / Pyell predictions have been corroborated by experimental results of Knox and Grant [18], Smith and Evans [32], and Robson et al. [33], who reported reduced plate heights h around 1 for retarded solutes. Some workers have demonstrated that with the same column (under identical conditions), a significantly better efficiency was obtained in the CEC mode than in the pressure-driven mode [18,34–37]. It can be concluded from these studies that with the so-called pseudo-electrochromatography [38] or pressurized-flow electrochromatography [35], a technique that uses a mixture of pressure-induced and electrokinetically induced flow, the chromatographic efficiency will be worse than with systems that use only electroosmosis for propulsion of the mobile phase. Bruin et al. [39] have shown that the efficiency of electro-driven open-tubular liquid chromatography (ED-OTLC) is by a factor of ca. 2 better than that of pressure-driven open-tubular liquid chromatography (PD-OTLC). In the case of open-tubular columns the improvement in efficiency is due only to the impact of the mobile-phase flow profile on the Cm term of the plate-height equation. The overall plate-height equation is valid in ED-OTLC and PDOTLC [39]: H⫽
2Dm d 2c v d 2f v ⫹ f (k) m ⋅ ⫹ f (k) s ⋅ v Dm Ds
(8)
where H ⫽ plate height, Dm or Ds ⫽ diffusion coefficient in the mobile or the stationary phase, respectively, v ⫽ linear velocity, dc ⫽ inner diameter of the column, df ⫽ thickness of the stationary-phase layer, f (k) m or f (k) s ⫽ function of k, and k ⫽ retention factor. However, there is a difference in f (k) m due to differences in the mobile-phase flow profile (see Fig. 5): ED-OTLC:
f (k) ED m ⫽
k2 16(1 ⫹ k)2
PD-OTLC:
f (k) PD m ⫽
1 ⫹ 6k ⫹ 11k 2 96 (1 ⫹ k) 2
(9) (10)
Bruin et al: [39] highlight that in ED-OTLC, capillaries with wider inner diameter (up to 25 µm) can be used than in PD-OTLC without loss of efficiency, having advantages of detection, loadability, and column preparation. Figure 7 shows the influence of the column diameter and the type of mobile-phase propulsion on the plate
Fundamentals of Capillary Electrochromatography / 15
Fig. 7 Theoretical plate-height curves for a pressure-driven (PD) system (dashed lines) and for an electroosmotic-driven (ED) system (solid lines) for different inner diameters of the open-tubular capillaries (k ⫽ 1, Dm ⫽ 10⫺9 m 2 /s). (Reprinted with permission from Ref. 28.)
height for a given retention factor (k ⫽ 1) according to Eq. (8) neglecting the resistance to mass transfer in the stationary phase [28]. With capillaries packed with octadecyl silica gel, acetonitrile– aqueous buffer mobile phases, and retarded noncharged analytes, extremely high efficiencies have been obtained in the isocratic elution mode at moderate holdup times. Seifar et al. [40] investigated the impact of the particle size on the performance. The mean particle diameter ranged from 1.5 to 50 µm. In accordance with theory, the plate height decreases with decreasing particle size. With porous particles the reduced plate height was about 2, but with the nonporous 1.5-µm particles, reduced plate heights of 1.3 were observed. On a 240-mm-long column, about 120,000 plates were generated (holdup time ca. 2.4 min, ca. 500,000 plates/m) [41]. The authors
16 / Pyell state that in order to realize the observed linear velocity under pressure, about 1900 bar is required. With columns packed with 1.5-µm nonporous octadecyl silica gel, Bailey and Yan [42] were able to separate a series of 14 nitroaromatic and nitramine explosive compounds in under 7 min (see Fig. 8), featuring efficiencies of ⬎500,000 plates/m. Dadoo et al. [43] used columns packed electrokinetically with 1.5-µm nonporous octadecyl silica gel to achieve rapid separations with high efficiencies. A sample containing 16 polycyclic aro-
Fig. 8 CEC separation of explosives. Capillary, 340 (210) mm ⫻ 75 µm; stationary phase, 1.5-µm nonporous octadecyl silica gel; mobile phase, 20% methanol, 80% aqueous buffer (10 mmol/L MES, 5 mmol/L SDS); HMX ⫽ octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine, RDX ⫽ hexahydro-1,3,5trinitro-1,3,5-triazine, DNB ⫽ 1,3-dinitrobenzene, TNB ⫽ 1,3,5-trinitrobenzene, NB ⫽ nitrobenzene, TNT ⫽ 2,4,6-trinitrotoluene, 2,4-DNT ⫽ 2,4-dinitrotoluene, tetryl ⫽ methyl-2,4,6-trinitrophenylnitramine, 2,6-DNT ⫽ 2,6-dinitrotoluene, 2-Am-DNT ⫽ 2-amino-4,6-dinitrotoluene, 2-NT ⫽ 2-nitrotoluene, 4-NT ⫽ 4-nitrotoluene, 4-Am-DNT ⫽ 4-amino-2,6-dinitrotoluene, 3-NT ⫽ 3-nitrotoluene. (Reprinted with permission from Ref. 42.)
Fundamentals of Capillary Electrochromatography / 17 matic hydrocarbons (PAHs) (classified as priority pollutants by the U.S. Environmental Protection Agency) was isocratically separated in under 10 min. With detection in a packed section of the capillary, ⬎700,000 plates/m were obtained. Lu¨dtke et al. [22] were the first to employ packings of submicrometer material in CEC. However, in their measurements the apparent reduced plate heights (not corrected for the nonpacked section of the capillary) are relatively high [h(minimum) ⫽ 3.5–4]. One possible explanation for this result is the fact that the B term of the plate-height equation [see Eq. 4] is not dependent on the particle diameter. Knox [44] has predicted that, using submicrometer material, all contributions from the A and C terms of the plate-height equation become insignificant compared to the band-broadening process, due to axial diffusion. While the use of spherical nonporous material with mean particle diameter ⬍1.8 µm with very narrow size distribution is one approach to obtain extremely high efficiency in CEC, the use of widepore material and pore flow effects is another interesting approach. Li and Remcho [45] report results obtained with octadecyl silica gels with mean pore diameters ranging from 6 to 400 nm. The authors assume that with large-pore material, perfusive transport of the mobile phase through the pores is possible, reducing significantly the plate height of the chromatographic system. The experimental results, indeed, support this theory. They also showed that the reduced plate height (mean particle diameter 7 µm) is dependent not only on the mean pore diameter but also on the ionic strength of the mobile phase, indicating ‘‘that the thickness of the electrochemical double layer obviously plays a significant role in perfusive transport through narrow channels.’’ Recently, Stol et al. [46] and SantallaGarcia et al. [47] reported, for columns packed with octadecyl silica gel (mean particle diameter 7 µm) with mean pore diameter of 400 nm, remarkably high efficiencies with mobile phases of moderate ionic strength. Up to 550,000 plates/m and reduced plate heights down to 0.26 were generated for retarded solutes. The column is acting ‘‘as if [one particle] is composed of several small particles.’’ The utility of reduced plate heights as a measure for column quality in CEC on wide-pore stationary phases can be questioned. With widepore packing material of large particle size, it is possible to obtain efficiencies that can be obtained with nonporous material only with particles of considerably smaller size.
18 / Pyell In 1995 Smith and Evans [31] observed extremely sharp zones (formally more than 8 million theoretical plates per meter) for basic antidepressants separated by CEC employing a strong cation exchanger as stationary phase and a mobile phase of pH ⫽ 3.5 (see Fig. 9). It is evident that a so-far-unknown zone-sharpening effect is involved. According to Sta˚hlberg [30], this zone sharpening might be due to a combination of electric field and adsorption effects. However, it is known that the extremely sharp zones reported in [31] are concentration dependent and often not sufficiently reproducible for analytical purposes. It is interesting to note that abnormally high efficiencies have also been observed under reversed-phase conditions employing an octadecyl silica gel as stationary phase [48].
C.
Comparison to Liquid Chromatography
Results presented by many workers [41–47] demonstrate that efficiencies between 500,000 and 700,000 plates per meter are attainable in isocratic CEC with packed capillaries. The number of plates
Fig. 9 CEC separation of basic antidepressants. Capillary, 300 (200) mm ⫻ 100 µm; stationary phase, 3-µm sulfonic acid propyl silica gel; mobile phase, 70% acetonitrile, 30% aqueous buffer (50 mmol/L NaH2PO4, pH ⫽ 3.5); 1 ⫽ bendroflumethiazide, 2 ⫽ nortriptyline, 3 ⫽ clomipramine, 4 ⫽ methdilazine. (Reprinted with permission from Ref. 31.)
Fundamentals of Capillary Electrochromatography / 19 in a real column, however, is significantly lower, due to limitations in the maximum applied voltage and the electroosmotic mobility limiting the maximum length of the packed capillary. In spite of these limitations, the attainable efficiency in CEC is about one order of magnitude higher than that in HPLC. Assuming a column resistance factor of 750 and a pressure limit of 400 bar, Dittmann et al. [49] calculated the maximum plate number achievable at the optimum velocity under the constraints mentioned above (see Table 1). According to their calculations, 5–10 times higher plate numbers per column are achievable using CEC compared with those obtained using capillary HPLC. It has to be emphasized that increasing the electroosmotic mobility is an important parameter to increase the maximum plate number for a real column in CEC.
D. Extracolumn Band Broadening In CEC the following extracolumn band-broadening contributions have to be taken into consideration: sample injection, detection, and data processing. With instrumentation designed for capillary electrophoresis, no significant band broadening due to detection and data processing is expected for CEC. Because of on-column injection and on-column (or in-column) detection, there is no band broadening due to transfer lines. In 1966 Sternberg [50] investigated in detail extracolumn contributions to chromatographic band broadening. He showed that the second moment (variance) for the total distribution can be derived from the second moments for the partial distributions. He also discussed the second-moment distributions for various input functions.
Table 1 Achievable Efficiency in Capillary HPLC and CEC Capillary HPLC Particle size (µm) 5 3 1.5
CEC
Length (cm)
Plates per column
Length (cm)
Plates per column
50 25 10
55,000 45,000 30,000
50 50 50
115,000 170,000 250,000
Source: Reprinted with permission from Ref. 49.
20 / Pyell The input function is the distribution function of the analyte concentration in the sample zone due to the injection process. Due to the use of injection valves with constant injected sample volume and pumps with constant volume flow, in HPLC the maximum sample volume given in volume units is of interest in dependence on the peak volume. These magnitudes are rather inconvenient in capillary electrochromatography. In CEC the sample is in most cases injected directly onto the column via the controlled application of a voltage (electrokinetic injection), and the velocity of the mobile phase is given not as flow rate (volume/time) but as linear velocity (distance/time) determined via a nonretarded marker. Assuming that the contribution due to detection and data processing can be neglected, the tolerable injection plug length for a column of given length and efficiency can be easily calculated [51]. The partial second moment (variance, σ2I for the injection plug (rectangular distribution) can be determined from the injection plug length L I [50]. σ2I ⫽
L 2I 12
(11)
The partial second moment (variance, σ2C ) for the distribution due exclusively to the column is given by σ2C ⫽
L2 N
(12)
where L ⫽ length of the column to the detection window and N ⫽ plate number. The total second moment (variance of the recorded peak, σ 2tot) is the sum of the partial second moments. It has to be emphasized that the variances are given in length units: σ 2tot ⫽ σ 2I ⫹ σ 2C
(13)
With electrokinetic injection, LI can be determined employing Eq. (14), taking into account zone compression during the injection process due to enrichment of the solutes in the stationary phase: LI ⫽ tI µeo UI Ltot⫺1 (1 ⫹ ks ) ⫺1
(14)
where tI ⫽ injection time, µeo ⫽ electroosmotic mobility, UI ⫽ injection voltage, Ltot ⫽ total column length, and ks ⫽ retention factor for the solute of interest with the sample solvent as mobile phase.
Fundamentals of Capillary Electrochromatography / 21 According to Pyell et al. [51], in performing CEC with packed capillaries there is a bias between the theoretical and experimental data that can be understood as a deviation of the input function from the assumed rectangular distribution function. This deviation has to be taken into account by an experimental factor FI. According to the data material presented in [51], FI equals 1.5. Equation (15) gives the maximum tolerable injection plug length, Lmax, for which the decision criterion (an increase of wh of about 5% due to extracolumn band broadening) is fulfilled: L L ⫽ 0.7 ⋅ ⋅ F I √N √N
Lmax ⫽ 1.1 ⋅
(15)
With this set of equations tolerable injection parameters can be calculated for electrokinetic injection. For a fixed injection time Eq. (16) is valid, predicting the suitable injection voltage; while for a fixed injection voltage Eq. (17) is valid, predicting the suitable injection time. Umax ⫽ 0.7 ⋅ tmax ⫽ 0.7 ⋅
L ⋅ Ltot (1 ⫹ ks ) µeo ⋅ tI ⋅ √N
L ⋅ Ltot (1 ⫹ ks ) µeo ⋅ UI ⋅ √N
(16) (17)
In Table 2, optimum injection times are given for CEC with packed capillaries assuming that ks ⫽ 0. The calculations show that with standard injection procedures (electrokinetic injection, tI ⫽ 5 s, UI ⫽ 5 kV) in CEC with packed capillaries of a total length ⱕ250 mm, the criterion for the tolerable extracolumn band broadening (an increase in wh by 5%) might not be fulfilled if the starting zone is not focused during the injection process. In Fig. 10 the dependence of the peak width at half-height in length units on the injected plug length calculated according to Eq. (14) (ks ⫽ 0) for alkyl- and arylbenzoates as test solutes is presented for a packed column [N(maximum) ⫽ 43,000–52,000, length to the detector ⫽ 258 mm]. The experimentally determined dependence of the peak width at half-height on the injected plug length corresponds to the curve predicted by theory. At low injected plug length (LI ⱕ 1 mm), wh is independent of LI , while at high volume overload wh is linearly increased with LI .
Table 2 Predicted Optimum Injection Time tmax Dependent on the Mean Particle Diameter dp of the Packing with Electrokinetic Injection (UI ⫽ 3 kV) assuming following conditions: h ⫽ 2, L ⫽ 200 mm, Ltot ⫽ 250 mm, µeo ⫽ 0.25 cm2 /s kV dp (µm) 10.0 5.0 3.0 2.0 1.5 1.0 0.5
N
tmax (s)
10,000 20,000 33,000 50,000 67,000 100,000 200,000
4.7 3.3 2.6 2.1 1.8 1.5 1.0
Source: Reprinted with permission from Ref. 51.
Fig. 10 Dependence of the peak width at half-height on the injected plug length (electrokinetic injection) for various alkyl benzoates. Capillary, 310 (258) mm ⫻ 180 µm; stationary phase, 1.5-µm porous octadecyl silica gel; mobile phase, 80% acetonitrile, 20% aqueous buffer (2 mmol/L phosphate, pH 7.3); ■ ⫽ methyl benzoate, 䊉 ⫽ ethyl benzoate, 䉱 ⫽ benzyl benzoate, 䉲 ⫽ p-tolyl benzoate, ◆ ⫽ iso-pentyl benzoate, ✚ ⫽ phenyl benzoate, ✖ ⫽ butyl benzoate. (Reprinted with permission from Ref. 51.)
Fundamentals of Capillary Electrochromatography / 23 With in-column detection (see Fig. 2) and reversal of the mobilephase flow by voltage switching, an analyte zone can be made pass several times through the detection area in the backward and forward directions. This principle has been used by Rebscher and Pyell [52] to estimate by experiment the extracolumn contribution to peak broadening in CEC with a packed capillary. By this method the length of the capillary column can be virtually increased by sections that are decided by the migration velocity of the analyte zone and the switching times. The variance of a peak σ 2 is proportional to the migrated distance. The migrated distance xm has to be calculated from the migration time and the migration velocity. A plot of σ2 versus the migrated distance xm gives a straight line (see Fig. 11). The peak variance σ 2E due to extracolumn band-broadening processes corresponds to the y intercept of the extrapolated straight line. The slope of the regression line permits one to calculate the true efficiency of the column [see Eq. 18], corresponding to the efficiency after elimination of instrumental band-broadening effects: N⫽
L m
(18)
Fig. 11 Illustration of a method to determine experimentally the peak variance σ 2E due to extracolumn band-broadening effects by voltage switching and in-column detection. The peak variance σ 2 is plotted versus the migrated distance xm.
24 / Pyell where N ⫽ plate number, L ⫽ length of the column to the detection window, and m ⫽ the slope of the regression line. Rebscher and Pyell [52] found that with the packed capillary used in their study, the instrumental band-broadening contribution was excessive, largely reducing the usable efficiency of the chromatographic system. There is indication that final frits might be the main source of the observed excessive band-broadening effects. Some workers report higher achievable chromatographic efficiencies with in-column detection compared to on-column detection [18,26,53], observing that the internal frit at the end of the chromatographic bed can be a source of substantial dispersion of analyte zones as they migrate through the frit. With a separation capillary packed with nonporous 1.5-µm octadecyl silica gel, Dadoo et al. [43] obtained extremely high efficiencies only with in-column detection. They report a drop in efficiency from 600,000–750,000 plates/m for retarded solutes to 300,000–400,000 plates/m when shifting the detection area from just before the outlet frit to 0.5–1 mm after the outlet frit.
E.
Thermal Effects
According to the considerations in Sections III.B and III.C, CEC has clear advantages over HPLC in terms of the obtainable efficiency for noncharged solutes. However, thermal effects have been neglected so far. Because of the electric propulsion of the mobile phase, Joule heat is generated in the separation capillary during the chromatographic run. The released power in the capillary corresponds to the applied voltage multiplied by the measured electric current strength. This generated heat is conducted first through the walls of the tube and then through the surrounding medium. As a result there will be a temperature difference between the capillary and the surrounding medium and there will be a temperature variation across the bore of the separation capillary. The temperature difference between the capillary and the surrounding medium does not lead to band broadening, but the temperature variation across the bore of the separation capillary does. According to Knox [44], in CEC the parabolic temperature profile which exists accross the capillary bore may cause variations in migration rates due to (1) changes in mobile-phase viscosity and (2) changes in the retention factors. In order to take this additional peak-broadening effect into account, an additional term has to be added to Eq. (4) or (8).
Fundamentals of Capillary Electrochromatography / 25 Grushka et al. [54] investigated in detail the effect of temperature gradients on the efficiency of capillary electrophoresis separations. They assumed that the actual velocity profile in the separation capillary is parabolic, due to the generated heat. Using the parabolic model, they developed an expression that relates the plate height to the capillary radius, the applied electric field strength, and the ionic strength of the mobile phase. This newly derived plate-height equation suggests that (1) at high electric field strength there is a significant peak broadening due to thermal effects, (2) the separation capillary should be thermostated, (3) thermostating at elevated temperature may have advantages, and (4) decreasing the concentration of the buffering salts in the mobile phase allows the use of wider capillaries. They state that ‘‘wider capillaries, in turn, would permit more convenient operation and large sample volumes.’’ Wider capillaries allow one to improve detection limits in combination with oncolumn or in-column spectrometric detection. The upper limit of allowable capillary inner diameter is a function of the applied field strength, the specific conductance of the mobile phase, and the diffusion coefficients of the solutes. Knox [44] calculated for capillary electroseparation methods the boundary conditions under which the plate-height contribution of the thermal effects is less than 10% the plate-height contribution from axial diffusion. It is interesting to note that, according to Knox, with parameters commonly given in CEC (concentration of buffering salt in mobile phase ⫽ 1 mmol/L3, applied field strength ⫽ 100 kV/ m), the maximum allowed capillary inner diameter is 150 µm. Capillaries used in CEC usually have an inner diameter of 100 µm. For comparison, in CE and MEKC, capillaries mostly have inner diameters of 50–75 µm. Because of the high content of organic modifier and the low concentration of buffering salts in mobile phases employed for CEC, the electric current that is measured during a chromatographic run in CEC is one to two orders of magnitude lower than in other capillary electroseparative techniques. Wright et al. [55] even demonstrated for CEC the applicability of mobile phases that do not contain added buffer salts. A separation of 11 polycyclic aromatic hydrocarbons was performed in under 13 min by CEC with an acetonitrile/water mobile phase. Whitaker and Sepaniak [56] separated fullerenes by CEC with a packed column and acetonitrile/ tetrahydrofuran or acetonitrile/methylene chloride as mobile phase. Working with mobile phases of extremely low specific electric con-
26 / Pyell ductivity might even make it possible to work in CEC with microbore columns and to perform highly efficient semipreparative separations. Although the advantages of wider separation capillaries are known, this topic has rarely been addressed in experimental studies. Yan et al. [57] and Vissers et al. [58] demonstrated the possibility of performing CEC with capillaries of 320-µm inner diameter packed with 5-µm octadecyl silica gel. While the possibility of using packed capillaries with an inner diameter ⬎⬎100 µm has been clearly demonstrated, very few data are available on the dependence (or independence) of the achievable chromatographic efficiency on the capillary inner diameter (I.D.). Rebscher and Pyell [59] estimated the influence of the capillary inner diameter on the achievable chromatographic efficiency independent of the influence of the quality of the frit and other randomly distributed parameters by packing several capillaries (inner diameters 75, 100, and 150 µm) with 3-µm octadecyl silica gel according to the same protocol. The comparison of the plate heights determined did not exhibit significantly lower plate heights for the capillaries with I.D. ⫽ 75 µm than for the capillaries with I.D. ⫽ 150 µm. Similar results have been obtained by Steiner et al. [60]. There might be, however, a significant drop in efficiency, if the inner diameter of the capillary is larger than 150 µm. Steiner et al. report a loss in efficiency of 50–60% when increasing the inner diameter from 100 to 180 µm. Banholczer and Pyell [61] report that the retention factors for several noncharged solutes determined for a packed capillary (inner diameter ⫽ 100 µm) by CEC with thiourea as marker of the holdup time were independent of the concentration of buffering salts (NaH 2 PO 4 ⫹ Na2 HPO 4 ) in the mobile phase (see Fig. 12). The phosphate concentration was varied in a range from 0.01 to 7 mmol/L. If there was a significant temperature rise inside the capillary due to Joule heating, a decrease in retention factors with increasing concentration of the buffering salts would be expected. The absence of any dependence of k on the ionic strength of the mobile phase verifies the assumption that in CEC with commonly given parameters the contribution of temperature gradients inside the capillary to the broadening of zones is negligible. Further investigations are needed to elucidate the influence of capillary dimensions, electric field strength, and specific electric conductivity of the chromatographic
Fundamentals of Capillary Electrochromatography / 27
Fig. 12 Retention factors k for several noncharged solutes versus buffer salt (phosphate) concentration in the mobile phase. Capillary, 422 (376) mm ⫻ 100 µm; stationary phase, 3-µm porous octadecyl silica gel (Nucleosil 100-3 ODS); mobile phase, 80% acetonitrile, 20% aqueous buffer (total phosphate concentration ⫽ 0.01–7 mmol/L, pH 7.2); solutes: ■ ⫽ methyl benzoate, 䊉 ⫽ ethyl benzoate, 䉱 ⫽ acenaphthylene, 䉲 ⫽ acenaphthene, ◆ ⫽ pyrene. (Reprinted with permission from Ref. 61.)
bed (also, the stationary phase might contribute to the conductance of electric current) on retention data and efficiency.
F. Bubble Formation In the early days of CEC, bubble formation during a chromatographic run was the most important practical problem that hampered the widespread use of this technique. In one of their pioneering works on CEC, Knox and Grant [18] have stated: ‘‘Drying out [of the packed capillaries] was particularly liable to occur with the wider capillaries and with the higher concentrations of electrolyte, indicating that self-heating was the primary cause.’’ They recommended thermostating of columns at temperatures close to ambient or op-
28 / Pyell erating the whole column under pressure as the most effective preventive measures. There are indications, however, that self-heating is not the primary cause of the formation of bubbles in the packing during a chromatographic run. The first indication is that the electric current strength that is measured during separation in CEC is one to two magnitudes lower than that is measured under routine CE or MEKC conditions. The results of Banholczer and Pyell [61] (see Fig. 12) strongly support the assumption that in CEC under standard conditions there is only a very small temperature rise inside the capillary. Rebscher and Pyell [52] observed that the formation of bubbles invariably started with a semipacked column at the border between the packed and the unpacked sections of the capillary. They interpreted this observation as follows: ‘‘The following segment has a higher electroosmotic mobility than the preceding segment. Consequently, the second non-packed segment has the effect of a pump reducing the pressure in the packed section below the prevailing pressure.’’ This interpretation is corroborated in a theoretical study by Rathore and Horva´th [62]. Rathore and Horva´th studied the interface of the packed and open segments of a semipacked CEC column and discontinuities associated. They show that in order to satisfy the mass conservation law, in most cases a ‘‘flow-equalizing intersegmental pressure,’’ which is different from the pressure at the two ends of the column, develops at the interface of the packed and open segments. A second reason for bubble formation is associated with frits. Rathore and Horva´th [62] highlight that silica frits most likely have zeta potentials different from those of the bulk packings, and the discontinuity of zeta potential can lead to the development of flowequalizing intersegmental pressure at such frits with concomitant bubble formation. Rebscher [63] observed that a completely packed capillary that invariably starts bubble formation at the inlet end during a chromatographic run can be used without problems, if the direction of the electroosmotic flow is reversed. Rebscher and Pyell [59] observed that if a prepared column has a frit with low (mechanical) streaming permeability, bubble formation during a chromatographic run is very likely to occur directly after the frit. This problem could be completely circumvented if the frit with low permeability was replaced by a second frit. These observations suggest that not only differences in the zeta potential between frit and bulk packing
Fundamentals of Capillary Electrochromatography / 29 but also an extremely low (mechanical) streaming permeability of the frit can result in bubble formation. Carney et al. [64] investigated factors affecting bubble formation. They concluded that bubble formation in CEC is a function of the length and nature of the frit. According to Carney et al., one possible solution of this problem is the recoating of octadecylsilane onto the silica frit, minimizing differences in the zeta potential between the frit and the bulk packing material. The necessity to operate the whole separation capillary under pressure in order to avoid bubble formation is controversially discussed in the literature [33,55,65,66]. Van den Bosch et al. [67] emphasized that operating the whole column under pressure is not necessary with a robust separation capillary. Considerations above suggest that for the design of a robust separation capillary suitable for CEC at ambient pressure, it is important that the separation capillary be as homogeneous as possible over the full length in zeta potential and in the cross-sectional area of the inner volume of the capillary accessible to the mobile phase. It is important to note that these two constraints are idealy fulfilled with fritless capillaries filled completely with stationary phase. In this case, spectrometric detection has to be performed in the chromatographic bed (in-column detection).
IV.
INSTRUMENTAL DEVELOPMENTS
A. Column Technology Work in CEC has started with capillaries slurry-packed with octadecyl silica gel [11,13,15–17,32,35,38,52,68–70]. In order to keep the packing in place, frits had to be prepared at both ends of the packing. Only Knox and Grant [13] have also worked with drawn-packed capillaries that did not need frits to stabilize the packing. Tsuda et al. [71], Bruin et al. [39], and Pfeffer and Yeung [72] were the first to perform CEC with open-tubular capillaries. CEC is still an emerging technique. Improvement of instrumentation and column technology will be of prime importance for the further development of this technique and its use in validated methods. Neither the classical packed capillary with end frits nor open-tubular capillaries seem to be the ideal separation column for CEC. Consequently, column technology in CEC has become a vigorous area of research in the last years. Advances in column technology have been recently reviewed [4,65,66,73–76] and will be addressed only briefly.
30 / Pyell With traditional packing procedures, frits are necessary to stabilize the chromatographic bed (see Fig. 2). Several methods have been reported for the production of frits in CEC: (1) reaction of sodium silicate solution with formamide to form a porous silica plug [77]; (2) sintering of a plug of native silica gel wetted with an aqueous solution of potassium silicate [78]; (3) sintering of a plug of native silica gel wetted with pure water [26]; or (4) sintering of a portion of the chromatographic packing itself with a heated filament after having flushed the column with water [32,68]. The slurry of the packing material is prepared either in an organic solvent or in supercritical carbon dioxide [33]. Mostly the packing material is transported into the column with the help of an external pump. Reynolds et al. [79] and Fermier and Colo´n [80] used columns packed by centripetal forces. Electrokinetic packing was described by Yan [81]. In CEC the stationary phase not only provides interaction sites for the solutes, it also plays the dominant role in the generation of the electroosmotic flow, hence the propagation of the mobile phase through the chromatographic bed. Consequently, the design of stationary phases suited for CEC not only has to keep in mind the retentive properties of this material but also the electrokinetic properties. Work in CEC started with stationary phases designed for HPLC. With these phases it might be possible to transfer directly a method that has been developed for HPLC onto CEC. The selectivity of the separation system may be altered by varying the stationary phase. Those stationary phases that have been used in the reversed-phase mode in HPLC should be applicable in CEC provided that their surface properties permit the generation of a sufficiently high electroosmotic velocity [82,83]. Some workers have tested successfully chiral stationary phases designed for HPLC in the CEC mode [84–90]. One important disadvantage of conventional silica-based materials, when used as stationary phases in CEC, is the dependence of the electroosmotic mobility on the pH of the mobile phase. However, some workers demonstrated the possibility of performing rapid analysis in CEC with capillaries packed with commercial octadecyl silica gel and mobile phases buffered at pH 2.5 [91]. Assuming that with a strong cation exchanger the electroosmotic velocity will be high even at low pH, Smith and Evans [31] tested the separation of basic drugs by CEC at low pH with a strong cation exchanger as stationary phase, with unexpected results (see Fig. 9). The ion exchanger was
Fundamentals of Capillary Electrochromatography / 31 made specifically for CEC, based on a specially prepared silica gel (dp ⫽ 3 µm) onto which is bonded a propyl sulfonate group. In order to make it possible to perform reversed-phase CEC with relatively high electroosmotic velocity quasi-independent of the pH, Zhang and El Rassi [92] introduced a novel silica-based multilayered stationary phase. This stationary phase comprises a relatively hydrophilic and charged sublayer attached covalently to the silica support and a retentative top layer of octadecyl groups bound chemically to the sublayer. Mixed-mode phases have also been employed [19,93,94]. Huang et al. [95] have chosen another approach to perform CEC with mobile phases of low pH. They have used a mixed-mode phase that contains octadecyl and dialkylamino groups (reversed-phase/ anion-exchange phase). The amino groups determine at low pH the charge density at the surface and the direction of the electroosmotic flow is reversed with respect to the direction found with bare silica gel. Huang et al. have demonstrated that this phase is ideally suited for the separation of (positively charged) peptides. Another approach is the synthesis of an octadecyl silica gel with light surface coverage of alkyl groups [96]. This stationary phase was designed to allow a relatively high electroosmotic velocity due to a high concentration of nonreacted silanol groups at the surface. The properties of final frits keeping the stationary phase in place determine the mechanical stability of the packed columns and their liability to bubble formation. Frits are discussed as major sources of extracolumn band broadening. These problems make it very desirable to construct fritless capillary columns for CEC. The production of fritless capillary columns for CEC has been the object of very active research in the last years. Highly swollen cross-linked hydrophobic hydrogels or rigid monoliths have been prepared in fusedsilica capillaries by in-situ synthesis, tapers have been used as an alternative to frits [97,98], and monoliths have been prepared by converting conventionally packed columns into ones with monolithic structure. Fujimoto et al. [99] realized continuous beds by radical copolymerization of N-isopropylacrylamide and 2-acrylamido-2-methylpropanesulfonic acid with N,N′-methylenebisacrylamide as crosslinking agent in capillaries pretreated with (γ-methacryloxypropyl)trimethoxysilane (inner diameter 50–75 µm). Other gel-based monoliths as stationary phases for CEC were synthesized by Liao
32 / Pyell et al. [100], Ericson et al. [101], and Palm and Novotny [102]. Chiral separations have been achieved in CEC with continuous beds via the preparation of monolithic molecularly imprinted flow-through polymers [103–106]. Peters et al. [107–109] presented a technique for the one-step preparation of ‘‘molded’’ rigid polymer monoliths in nonpretreated fused-silica capillaries (inner diameter 100–150 µm). Mayer et al. [110] report the production of fritless capillaries (tapered end at the inlet side and neither frit nor taper at the outlet side); packed with 1.5-µm nonporous octadecyl silica gel. Asiaie et al. [111] succeeded in sintering packings of capillary columns. In the sintering process the compacted powder is converted into a monolith in which the particles are joined to each other by grain boundaries. Another possibility for the stabilization of packings and thus eliminating the need for frits is the embedding of packed particles in a rigid sol-gel or a silicate matrix [112–114]. From a technical point of view, open-tubular columns are the simplest approach to fritless and robust columns in CEC. Column technology for open-tubular CEC was reviewed in 1997 by Colo´n et al. [4] and Pesek and Matyska [75]. More recent developments are the fabrication of an organic–inorganic hybrid material by the solgel method as a thin porous film attached to the inner wall of fusedsilica capillaries [115,116] and the preparation of polymer-coated 25-µm-I.D. capillaries by in-capillary copolymerization of N-tertbutylacrylamide with a charged monomer, after the pretreatment of the capillary inner surface with a bifunctional reagent [117]. Although open-tubular (OT) capillaries have some advantages over packed or filled capillaries, such as no liability to bubble formation and easy rinsing without the need to apply high pressure, OT capillaries are rarely employed in CEC. This might be due to the relatively small inner diameters of such capillaries having disadvantages in combination with optical detection methods. However, CEC with OT capillaries has a high miniaturization potential that has been used by Kutter et al. [118] for open-channel electrochromatography in combination with solvent programming using a microchip device.
B.
Detection
Basically, in CEC the same detection techniques are applicable as in capillary electrophoresis: photometric detection, fluorimetric detection [119,120], or amperometric detection, for example. The cou-
Fundamentals of Capillary Electrochromatography / 33 pling of CEC with mass spectrometry has been successfully realized by several workers, mainly by employing electrospray ionization [34,38,95,98,116,121–127]. Also, the on-line coupling of CEC with nuclear magnetic resonance spectroscopy has been recently reported [128–130]. Guo et al. [131] coupled pressurized CEC with a condensation nucleation light-scattering detector using an electrospray interface. They demonstrated that condensation nucleation light-scattering detection is a sensitive and universal detection method for CEC. Qi et al. [132] highlighted the potential of thermooptical absorbance detection as an attractive alternative to classical absorbance detection, interfacing a CEC system with an ultraviolet laser-based thermooptical absorbance detector. High detection limits because of small detection volumes or low mass load can still be regarded as the Achilles’ heel of CEC, if not only the solution of a separation problem but also the achievement of low detection limits is needed to solve a specific analytical problem. It is obvious that further development of selective and sensitive detection techniques for CEC will be very important to broaden the application scope of this potentially highly efficient separation method. There is one important difference between capillary electrophoresis and capillary electrochromatography in terms of optical detection. With both separation techniques, detection has to be performed in a section of the separation capillary in order to avoid excessive extracolumn band broadening (there are rare cases where capillaries are coupled [133]). Hence a section of the separation capillary has to be transformed into the detection cell. As in most cases, fusedsilica capillaries are used, and the transparency of these capillaries in the VIS/UV region is given. However, the outer protecting polyimide layer has to be removed for a short section of the capillary to form the on-column detection cell, making this section of the capillary brittle and fragile. The difference between CEC and CE in terms of optical detection is the fact that with CEC detection can be performed either in the chromatographic bed itself containing the stationary and the mobile phase (in-column detection, ICD) or in an open section of the separation capillary containing only the eluate (on-column detection, OCD). The terminology differing between these two detection modes was suggested in 1987 by Verzele and Dewaele [134]. In Fig. 2 the realization of these two detection modes is illustrated using a packed capillary as an example.
34 / Pyell The inherent disadvantage of on-column detection is the appearance of a discontinuity at the interface between the filled and the open sections. Rathore and Horva´th [62] have shown that these discontinuities may result in intersegmental pressure differences changing the nature of the mobile-phase flow from purely electroosmotically driven to partly pressure driven. In extreme cases even bubble formation may result. Typically, completely filled capillaries are more robust than partly filled capillaries. Analyte zones migrating through the interface between the filled and the open section can be substantially broadened (see Section III.D). In spite of these disadvantages, on-column detection is mostly applied. In the case of in-column detection, the presence of two phases in the detection window results in diffuse scattering of light at the irregular interfaces between the two phases. Consequently, in the detection volume along the optical path of the detector there is (nonspecific) intensity loss of the incident beam due to diffuse scattering, and also if the stationary phase does not absorb at the detection wavelength. The extent of intensity loss is dependent on the difference in diffraction index between the two phases. In the ideal case (match of diffraction indices of stationary and mobile phase) the detection window is transparent, becoming opaque or even nontransparent in the case of extreme differences in the diffraction index. Consequently, in-column photometric detection is possible only if the stationary phase does not absorb at the detection wavelength and if there are only moderate differences between the refraction indices of the two phases present in the detection volume. These constraints are not always fulfilled. Banholczer and Pyell [135] compared in-column with on-column photometric (UV) detection performed with fused-silica capillaries of 180-µm inner diameter, packed with 3-µm octadecyl silica gel. They determined the influence of the detection mode on linearity of the calibration function, precision, and detection limit. In their experiments, the baseline noise for in-column detection (ICD) is about twice that for on-column detection (OCD). This negative impact on the detection limit is mitigated by a signal enhancement in the case of ICD. This signal enhancement is due to enrichment of the solute in the stationary phase and can be quantitatively described with Eq. (19): SI /SO ⫽ (1 ⫹ k) ϕM
(19)
Fundamentals of Capillary Electrochromatography / 35 where SI ⫽ sensitivity for in-column detection, SO ⫽ sensitivity for on-column detection, k ⫽ retention factor, and ϕM ⫽ volume fraction of mobile phase in the chromatographic bed. The validity of this equation was demonstrated experimentally [135]. Figure 13 compares chromatograms of the same sample obtained with ICD or OCD. All other experimental conditions except the location of the detection window were kept constant. The appar-
Fig. 13 Comparison of chromatograms obtained with (a) in-column and (b) on-column photometric detection (λ ⫽ 230 nm). Separation of alkyl benzoates (identical samples). (a) Capillary, 325 (270) mm ⫻ 180-µm I.D., (b) capillary, 400 (350) mm ⫻ 180-µm I.D.; stationary phase, 3-µm porous octadecyl silica gel; mobile phase, 80% acetonitrile, 20% aqueous buffer (2 mmol/L phosphate, pH 7.3) 1 ⫽ thiourea, 2 ⫽ methyl benzoate, 3 ⫽ ethyl benzoate, 4 ⫽ phenyl benzoate, 5 ⫽ benzyl benzoate, 6 ⫽ p-tolyl benzoate, 7 ⫽ butyl benzoate, 8 ⫽ iso-pentyl benzoate. (Reprinted with permission from Ref. 135.)
36 / Pyell ent change in peak-height ratios is due to the dependence of the described enhancement effect on the retention factor. Equation (19) was derived in analogy to the considerations of Guthrie and Jorgenson [136]. Rebscher and Pyell [26] have shown that Eq. (19) is also valid for fluorescence detection in packed capillaries, if variations in the fluorescence quantum yield due to environmental effects are neglected. Taking Eq. (19) and the increase in baseline noise into account, the relative limit of detection (LOD) for a method employing ICD (in comparison to a method employing OCD) can be calculated as LOD(ICD) ⫽
LOD(OCD) ⋅ F(noise) (1 ⫹ k) ⋅ ϕM
(20)
where F(noise) ⫽ factor by which baseline noise in ICD is increased compared to baseline noise in OCD. The validity of Eq. (20) has been verified experimentally for photometric detection [135]. With F(noise) ⫽ 2 and ϕM ⫽ 0.7, the LOD with ICD is lower than with OCD provided k ⬎ 2. Consequently, there may be an improvement of the LOD for late-eluted solutes when employing ICD instead of OCD. Improvement of detection limits for late-eluted solutes by the use of detection in a packed section of the separation capillary [137] or by packed flow cells [138] was also reported for microcolumn liquid chromatography.
C.
Gradient Elution
Gradient elution improves the peak capacity of a chromatographic system via zone compression and is indispensable for analyzing very complex samples. In CEC the following gradients may be applied: gradient of the composition of the mobile phase, temperature gradient, or voltage gradient. Although in CEC temperature gradients (temperature between the melting and the boiling point of the mobile phase) can be easily realized because of the capillary dimensions of the separation column, this approach has not been reported so far, to the best of the author’s knowledge. The impact of a temperature gradient on retention factors of the solutes, however, will be much smaller than that of a gradient of the composition of the mobile phase. In the case of using commercial automated CE equipment for CEC, the realization of stepwise gradients of the composition can be
Fundamentals of Capillary Electrochromatography / 37 performed easily and reproducibly by changing the (buffer) vial at the inlet side at a predefined time interval [139,140]. Linear gradients of the composition of the mobile phase have mostly been realized with help of a gradient-delivering HPLC pump. Behnke and Bayer [68] were the first to develop pressurized gradient CEC. If a pressure-induced additional flow is to be avoided, the head of the separation capillary ‘‘merely dips into the stream of mobile phase passing by.’’ Sample is injected via a normal type or microinjection valve [127,141–144]. Also, automatization via the use of an (HPLC) autosampler is reported [37,125,145]. In Fig. 14 the separa-
Fig. 14 CEC separation of phenylthiohydantoin amino acids with gradient elution. Capillary, 207 (127) mm ⫻ 50 µm; stationary phase, 3.5-µm porous octadecyl silica gel (Zorbax ODS); mobile phase, acetonitrile, aqueous buffer (5 mmol/L phosphate, pH ⫽ 7.55) gradient; voltage 10 kV; photometric detection at 210 nm; solutes in order of elution: formamide, PTH-asparagine, PTH-glutamine, PTH-threonine, PTH-glycine, PTH-alanine, PTH-tyrosine, PTH-valine, PTH-proline, PTH-tryptophan, PTH-phenylalanine, PTH-isoleucine, PTH-leucine. (Reprinted with permission from Ref. 141.)
38 / Pyell tion of phenylthiohydantoin amino acids achieved via gradient CEC (linear gradient formed by two HPLC pumps) is shown. Yan et al [119] developed an experimental setup for gradient CEC with electrokinetic generation of the gradient of the composition of the mobile phase. Two high-voltage power supplies are used to generate two electroosmotic flows in two channels (connected to two mobile-phase reservoirs) that are merged in front of the column head. The voltage of the two high-voltage power supplies is controlled by a computer. The ratio of the electroosmotic flow rate between the two channels delivering the mobile phase is gradually changed, thus generating a gradient of the composition at the mixing tee. Other workers used a miniaturized titration device for gradient CEC [116,146]. Also, voltage programming is reported as an alternative to mobile-phase composition gradient programming [147]. It has to be emphasized, however, that this technique is of limited applicability, as no variation in the retention factor is achieved by voltage gradients.
V.
MOBILE-PHASE CONSIDERATIONS
Not only surface properties of the stationary phase but also bulk properties of the mobile phase determine the electroosmotic mobility in the chromatographic bed. Hence, in CEC, optimization of the composition of the mobile phase must consider not only retention of solutes and selectivity of the chromatographic system (as in HPLC) but also observed electroosmotic mobility and achieved chromatographic efficiency. It is evident that strategies for the optimization of the mobile phase in CEC must differ from those that have been developed so far for HPLC, although the underlying retention mechanism for noncharged solutes is basically the same. Parameters determining the electroosmotic velocity in packings of porous particles are not completely understood [148]. Currently, users of CEC must rely on phenomenological recordings of dependencies of chromatographic parameters such as velocity of the mobile phase or efficiency on properties of the mobile phase. Regarding Eq. (2), the tentative assumption can be made that in case of a given stationary phase, constant electric field strength, and constant ionic strength in the mobile phase, the observable
Fundamentals of Capillary Electrochromatography / 39 linear velocity of the mobile phase is proportional to the ratio of the dielectric constant εD to viscosity η of the bulk liquid. Banholczer and Pyell [61] determined the linear mobile phase velocity with one packed capillary for various organic solvent–aqueous buffer mixtures (keeping the ionic strength constant) and correlated this magnitude to εD /η of the corresponding organic solvent–water mixture (see Fig. 15). Although there is a trend, a direct correlation of εD /η to the observed mobile phase velocity is not possible. The deviation from the regression line indicates possibly that the electrokinetic potential ζ cannot be regarded as independent of the mobilephase composition, although the ionic strength has been kept constant. Highest electroosmotic mobilities can be obtained in CEC with acetonitrile–aqueous buffer mixtures, making this mixture currently the mobile phase of choice in CEC. Similar results have
Fig. 15 Electrophoretic mobility µ eo plotted versus ratio dielectric constant εD to viscosity η of the corresponding solvent–water mixture. Capillary, 422 (376) mm ⫻ 100 µm; stationary phase, 3-µm porous octadecyl silica gel (Nucleosil 100-3 ODS); mobile phase, organic constituent/phosphate buffer (c(phosphate) in mobile phase ⫽ 1 mmol/L, pH ⫽ 7.2). (Reprinted with permission from Ref. 61.)
40 / Pyell been obtained by Dittmann and Rozing [70] when changing from acetonitrile to methanol or THF as organic modifier in the mobile phase. In an open capillary the electroosmotic mobility µeo is related to the thickness δ of the double layer adjacent to the wall surface [149]: µ eo ⫽
σ⋅δ η
(21)
where σ ⫽ surface charge density and η ⫽ viscosity of the bulk liquid. According to Eq. (3), δ is a function of the ionic strength I ⫽ 1/2 ∑ i z2i ci .Hence, in open-tubular capillaries for a given electrolyte (only the concentration of the buffer component is varied), the electroosmotic mobility µeo is inversely proportional to the square root of the ionic strength. The same relationship should be expected for capillaries filled with a porous plug. Indeed, Choudhary and Horva´th [150] found for a packed capillary a linear relationship between µ eo and the square root of the concentration of added NaCl (c ⫽ 20–60 mmol/L). Also, Dittmann and Rozing [19] found for a packed capillary a decrease of µ eo with increasing buffer concentration (c ⫽ 2– 20 mmol/L). In CEC, however, the concentration of the buffer component in the mobile phase is often restricted to low concentrations (c ⱕ 1 mmol/L) because of the limited solubility of the buffer component in the solvent mixtures employed as mobile phases. Wan [151] determined µeo dependent on the concentration of sodium phosphate (c ⫽ 10⫺3 –10⫺5 mol/L) in the mobile phase with a capillary packed with octadecylsilica gel. The author obtained maximum µeo at an intermediate concentration of sodium phosphate. Similar results have been obtained by Knox and Grant [18], who found maximum µeo at an intermediate concentration of the mobile-phase component NaH2 PO4 at 10⫺3 mol/L. In order to measure electroosmotic mobilities that are not distorted by the influence of the open section of the separation capillary, Banholczer and Pyell [61] determined µeo in capillaries completely packed with octadecylsilica gel employing acetonitrile/aqueous phosphate buffer as mobile phase, varying the phosphate concentration [c(NaH 2PO 4 ) ⫹ c(Na2HPO4 )] in a range from 0.01 to 7 mmol/L.
Fundamentals of Capillary Electrochromatography / 41 The experiment was repeated under identical conditions with a second column in order to give a rough estimate of the column-to-column variability. In Fig. 16 the observed electroosmotic mobility is plotted against the logarithm of the phosphate concentration in the mobile phase. The results obtained for the two capillaries are depicted in the same figure. For the two capillaries, µ eo passes through a maximum at a phosphate concentration of 0.4–4 mmol/L. These studies show that for porous plugs there is a deviation from what is described for OT capillaries. The results obtained for packed capillaries suggest that in CEC the ionic strength in the mobile phase is an important parameter. In order to obtain maximum velocity of the mobile phase, a sum concentration of buffering salts of ca. 1 mmol/L is appropriate.
Fig. 16 Electroosmotic mobility versus buffer salt (phosphate) concentration: ■ ⫽ column 1, L ⫽ 442 (390) mm ⫻ 100-µm I.D., mean of three measurements; ◆ ⫽ column 2, L ⫽ 422 (376) mm ⫻ 100 µm I.D., mean of five measurements, error bars ⫽ standard deviation; stationary phase, 3-µm porous octadecyl silica gel (Nucleosil 100-3 ODS); mobile phase, 80% acetonitrile, 20% aqueous buffer (total phosphate concentration ⫽ 0.01–7 mmol/ L, pH 7.2); separation voltage 25 kV, electrokinetic injection 5 s at 5 kV, photometric in-column detection, λ ⫽ 230 nm, marker of holdup time thiourea. (Reprinted with permission from Ref. 61.)
42 / Pyell
VI.
APPLICATIONS
Initially CEC was applied mainly to the analysis of samples of pharmaceutical interest Dulay et al. [152] demonstrated the routine application of CEC with a commercial CE instrument, separating a mixture of neutral compounds with a packed capillary. Hundreds of consecutive runs were performed over a period of weeks. They concluded that ‘‘CEC separations can be achieved in a reliable and routine manner.’’ The potential to apply CEC in pharmaceutical analysis has been investigated by Euerby et al. [27] in the analysis of a wide range of structurally diverse pharmaceutical compounds. They state that ‘‘The repeatability of retention of CEC using reverse-phase materials is excellent and is a distinct advantage over CE using bare capillaries. The repeatability of peak area and height is surprisingly good compared to conventional CE and is comparable to many commercially available HPLC systems.’’ Applications of CEC (including chiral analyses) were reviewed in 1998 in an excellent paper by Cikalo et al. [66]. Recently, various workers have been able to show that CEC can be used successfully to analyze compounds in various biological matrices, e.g., urine or plasma [145,153]. Sandra et al. [154] reported the analysis of triglycerides in vegetable oils. Dermaux et al. [155] employed CEC to analyze free fatty acids and fatty acid phenacyl esters originating from vegetable oils and margarine. They compared data obtained with CEC and micro-HPLC. CEC was found to be much superior in terms of efficiency and speed of analysis. Saeed et al. [91] report the application of CEC to the quantitative analysis of individual mono- and dihydroxy phenols in tobacco smoke. According to Saeed et al., the method presented provides the analysis of real samples in significantly shorter times than achieved by current GC and HPLC methods. Dadoo et al. [43] presented the isocratic separation of 16 polycyclic aromatic hydrocarbons (classified as priority pollutants by the U.S. Environmental Protection Agency) by CEC with a packed capillary in under 10 min. Bailey and Yan [42] realized the complete separation of a series of 14 nitroaromatic and nitramine explosive compounds under isocratic conditions in under 7 min (see Fig. 8). Currently, there is much interest in the evaluation of the potential of CEC for the analysis of biomolecules. CEC has good compati-
Fundamentals of Capillary Electrochromatography / 43 bility with mass spectrometry. In addition, electrospray ionization in mass spectrometry has evolved into a powerful tool for the analysis of biomolecules. CEC/MS might fill the place of HPLC/MS when high separation efficiency is needed. So far, published applications of CEC/electrospray ionization mass spectrometry, CEC/ion trap storage/reflectron time-of-flight mass spectrometry, CEC/electrospray ionization time-of-flight mass spectrometry, or CEC with photometric (UV) detection in the field of bioanalysis include: separation of phenylthiohydantoin amino acids [98,111,141,156] (see Fig. 14), separation of peptide and protein mixtures [94,95,116,157], analysis of peptide and protein digests [124,126,127,158], separation of isomeric polycyclic aromatic hydrocarbon–deoxyribonucleoside adduct mixtures [140], separation of derivatized mono- and oligosaccharides [96], separation of purine and pyrimidine bases and their nucleosides [159], and separation of oligonucleotides [68].
VII. CONCLUDING REMARKS Capillary electrochromatography is currently a field of very active research. This hybrid of liquid chromatography and capillary electrophoresis has proven to be superior over HPLC in terms of efficiency and offers in the case of ionizable solutes selectivities that are different from HPLC and CE. Not all phenomena that have been observed by those applying CEC are fully understood, e.g., band shapes found for ionic compounds separated on an ion exchanger. However, practical problems that hampered rapid development of this method in the early days of CEC seem to have been overcome. If theoretical predictions that the inner diameter of separation capillaries in CEC can be considerably larger than in CE without loss in efficiency prove to be well founded, detection limits of methods using CEC will be largely improved, and CEC would gain an additional advantage over CE. In order to construct robust, highly efficient, and reliable separation columns for CEC, fritless capillaries are desired. From a theoretical point of view, capillaries filled homogeneously with a bed of stationary phase are best suited to meet the needs of CEC. Encouraging results obtained with CEC are in accordance with predictions based on separation science theory. These results justify current efforts made in the synthesis of new stationary phases, in instrumental developments, in the design of coupling devices for the
44 / Pyell coupling of CEC with powerful spectrometric techniques (e.g., mass spectrometry or nuclear magnetic resonance), and in the development of new detection techniques meeting better the requirements for detection in a section of the capillary than those techniques that have been developed for HPLC. A sound theoretical understanding of parameters influencing a separation achieved by CEC will be very important in order to exploit the full potential of this new separation technique.
REFERENCES 1. R. M. Smith, J. Chromatogr. A, 856: 83 (1999). 2. J. P. Quirino and S. Terabe, J. Chromatogr. A, 856: 465 (1999). 3. I. S. Lurie, T. S. Conver, and V. L. Ford, Anal. Chem., 70: 4563 (1998). 4. L. A. Colo´n, K. J. Reynolds, R. Alicea-Maldonado, and A. M. Fermier, Electrophoresis, 18: 2162 (1997). 5. G. Berraz, Anales Asoc. Quı´m. Argentina, 31: 96 (1943). 6. D. L. Mould and R. L. M. Synge, Analyst, 77: 964 (1952). 7. D. L. Mould and R. L. M. Synge, Biochem. J., 58: 571 (1954). 8. V. Pretorius, B. J. Hopkins, and J. D. Schieke, J. Chromatogr., 99: 23 (1974). 9. J. W. Jorgenson and K. D. Lukacs, Anal. Chem., 53: 1298 (1981). 10. J. W. Jorgenson and K. D. Lukacs, Science, 222: 266 (1983). 11. J. W. Jorgenson and K. D. Lukacs, J. Chromatogr., 218: 209 (1981). 12. M. Martin and G. Guichon, Anal. Chem., 56: 614 (1984). 13. J. H. Knox and I. H. Grant, Chromatographia, 24: 135 (1987). 14. G. J. M. Bruin, P. P. H. Tock, J. C. Kraak and H. Poppe, J. Chromatogr., 517: 557 (1990). 15. T. Eimer, diploma thesis, University of Mainz, Germany, 1992. 16. H. Yamamoto, J. Baumann, and F. Erni, J. Chromatogr., 593: 313 (1992). 17. T. Tsuda, Anal. Chem., 59: 521 (1987). 18. J. H. Knox and I. H. Grant, Chromatographia, 32: 317 (1991). 19. M. M. Dittmann and G. P. Rozing, J. Microcol. Sep., 9: 399 (1997).
Fundamentals of Capillary Electrochromatography / 45 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.
34. 35. 36. 37. 38. 39. 40.
J. Th. G. Overbeek, in H. R. Kruyt (Ed.), Colloid Science Vol. 1, Elsevier, Amsterdam, 1952, p. 202. R. J. Hunter, Zeta Potential in Colloid Science, Academic Press, London, 1981. S. Lu¨dtke, T. Adam, and K. K. Unger, J. Chromatogr. A, 786: 229 (1997). J. J. Van Deemter, F. J. Zuiderweg, and A. Klinkenberg, Chem. Eng. Sci., 5: 271 (1956). M. J. E. Golay, in D. H. Desty (Ed.), Gas Chromatography, Butterworths, London, UK, 1958, p. 36. R. P. W. Scott, Liquid Chromatography Column Theory, Wiley, Chichester, UK, 1992. H. Rebscher and U. Pyell, J. Chromatogr. A, 737: 171 (1996). M. R. Euerby, D. Gilligan, C. M. Johnson, S. C. P. Roulin, P. Myers, and K. D. Bartle, J. Microcol. Sep., 9: 373 (1997). A. L. Crego, A. Gonza´lez, and M. L. Marina, Crit. Rev. Anal. Chem., 26: 261 (1996). A. S. Rathore and Cs. Horva´th, Anal. Chem., 70: 3271 (1998). J. Sta˚hlberg, Anal. Chem., 69: 3812 (1997). N. W. Smith and M. B. Evans, Chromatographia, 41: 197 (1995). N. W. Smith and M. B. Evans, Chromatographia, 38: 649 (1994). M. M. Robson, S. Roulin, S. M. Shariff, M. W. Raynor, K. D. Bartle, A. A. Clifford, P. Myers, M. R. Euerby, and C. M. Johnson, Chromatographia, 43: 313 (1996). D. B. Gordon, G. A. Lord, and D. S. Jones, Rapid. Commun. Mass Spec., 8: 544 (1994). T. Eimer, K. K. Unger, and T. Tsuda, Frezenius J. Anal. Chem., 352: 649 (1995). C. Ericson, J.-L. Liao, K. Nakazato, and S. Hjerte´n, J. Chromatogr. A, 767: 33 (1997). J. N. Alexander, J. B. Poli, and K. E. Markides, Anal. Chem., 71: 2398 (1999). E. R. Verheij, U. R. Tjaden, W. M. A. Niessen, and J. van der Greef, J. Chromatogr., 554: 339 (1991). G. J. M. Bruin, P. P. H. Tock, J. C. Kraak, and H. Poppe, J. Chromatogr., 517: 557 (1990). R. M. Seifar, S. Heemstra, W. Th. Kok, J. C. Kraak, and H. Poppe, Biomed. Chromatogr., 12: 140 (1998).
46 / Pyell 41. 42. 43. 44. 45. 46. 47.
48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.
61. 62.
R. M. Seifar, W. Th. Kok, J. C. Kraak, and H. Poppe, Chromatographia, 46: 131 (1997). C. G. Bailey and C. Yan, Anal. Chem., 70: 3275 (1998). R. Dadoo, R. N. Zare, C. Yan, and D. S. Anex, Anal. Chem., 70: 4787 (1998). J. H. Knox, Chromatographia, 26: 329 (1988). D. Li and V. T. Remcho, J. Microcol. Sep., 9: 389 (1997). R. Stol, W. Th. Kok, and H. Poppe, J. Chromatogr. A, 853: 45 (1999). T. Santalla-Garcia, L. de Ribera-Martin, R. Stol, W. Th. Kok, and H. Poppe, 23rd Int. Symp. on High Performance Liquid Phase Separations and Related Techniques, Granada, Spain, May 1999. F. Moffatt, P. A. Coper, and K. M. Jessop, Anal. Chem., 71: 1119 (1999). M. M. Dittmann, K. Wienand, F. Beck, and G. P. Rozing, LCGC, 13: 800 (1995). J. C. Sternberg, Adv. Chromatogr., 2: 205 (1966). U. Pyell, H. Rebscher, and A. Banholczer, J. Chromatogr. A, 779: 155 (1997). H. Rebscher and U. Pyell, Chromatographia, 38: 737 (1994). C. Yan, R. Dadoo, H. Zhao, R. N. Zare, and D. J. Rakestraw, Anal. Chem., 67: 2026 (1995). E. Grushka, R. M. McCormick, and J. J. Kirkland, Anal. Chem., 61: 241 (1989). P. B. Wright, A. S. Lister, and J. G. Dorsey, Anal. Chem., 69: 3251 (1997). K. W. Whitaker and M. J. Sepaniak, Electrophoresis, 15: 1341 (1994). C. Yan, D. Schaufelberger, and F. Erni, J. Chromatogr. A, 670: 15 (1994). J. P. C. Vissers, H. A. Claessens, and P. Coufal, J. High Resol. Chromatogr., 18: 540 (1995). H. Rebscher and U. Pyell, Chromatographia, 42: 171 (1996). F. Steiner, B. Scherer, and H. Engelhardt, 2nd Int. Symp. on Capillary Electrochromatography, San Francisco, CA, August 1999. A. Banholczer and U. Pyell, J. Chromatogr. A, 869: 363 (2000). A. S. Rathore and Cs. Horva´th, Anal. Chem., 70: 3069 (1998).
Fundamentals of Capillary Electrochromatography / 47 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84.
H. Rebscher, thesis, University of Marburg, Germany, 1996. R. A. Carney, M. M. Robson, K. D. Bartle, and P. Myers, J. High Resol. Chromatogr., 22: 29 (1999). L. A. Colo´n, Y. Guo, and A. Fermier, Anal. Chem., 69: 461A (1997). M. G. Cikalo, K. D. Bartle, M. M. Robson, P. Myers, and M. R. Eurby, Analyst, 123: 87R (1998). S. E. van den Bosch, S. Heemstra, J. C. Kraak, and H. Poppe, J. Chromatogr. A, 755: 165 (1996). B. Behnke and E. Bayer, J. Chromatogr. A, 680: 93 (1994). R. J. Boughtflower, T. Underwood, and C. J. Paterson, Chromatographia, 40: 329 (1995). M. M. Dittmann and G. P. Rozing, J. Chromatogr. A, 744: 63 (1996). T. Tsuda, K. Nomura, and G. Nakagawa, J. Chromatogr., 248: 241 (1982). W. D. Pfeffer and E. S. Yeung, Anal. Chem., 62: 2178 (1990). M. M. Robson, M. G. Cikalo, P. Myers, M. R. Eurby, and K. D. Bartle, J. Microcol. Sep., 9: 357 (1997). K. D. Altria, N. W. Smith, and C. H. Turnbull, Chromatographia, 46: 664 (1997). J. J. Pesek and M. T. Matyska, Electrophoresis, 18: 2228 (1997). U. Pyell, J. Chromatogr. A, 892: 257 (2000). H. J. Cortes, T. S. Pfeiffer, B. C. Richter, and T. S. Stevens, J. High Resol. Chromatogr., 10: 446 (1987). R. T. Kennedy and J. W. Jorgenson, Anal. Chem., 61: 1128 (1989). K. J. Reynolds, T. D. Maloney, A. M. Fermier, and L. A. Colo´n, Analyst, 123: 1493 (1998). A. M. Fermier and L. A. Colo´n, J. Microcol. Sep., 10: 439 (1998). C. Yan, Electrokinetic Packing of Capillary Columns, U.S. Patent 5,453,163 (1995). M. R. Eurby, C. M. Johnson, S. F. Smyth, N. Gillott, D. A. Barrett, and P. N. Shaw, J. Microcol. Sep., 11: 305 (1999). X. Cahours, Ph. Morin, and M. Dreux, J. Chromatogr. A, 845: 203 (1999). S. Li and D. K. Lloyd, Anal. Chem., 65: 3684 (1993).
48 / Pyell 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106.
S. Li and D. K. Lloyd, J. Chromatogr. A, 666: 321 (1994). F. Lelie`vre, C. Yan, R. N. Zare, and P. Gareil, J. Chromatogr. A, 723: 145 (1996). A. Dermaux, F. Lynen, and P. Sandra, J. High Resol. Chromatogr., 21: 575 (1998). M. La¨mmerhofer and W. Linder, J. Chromatogr. A, 829: 115 (1998). Y. Deng, J. Zhang, T. Tsuda, P. H. Yu, A. A. Boulton, and R. M. Cassidy, Anal. Chem., 70: 4586 (1998). C. Wolf, P. L. Spence, W. H. Pirkle, E. M. Derrico, D. M. Cavender, and G. P. Rozing, J. Chromatogr. A, 782: 175 (1997). M. Saeed, M. Depala, D. H. Craston, and I. G. M. Anderson, Chromatographia, 49: 391 (1999). M. Zhang, and Z. El Rassi, Electrophoresis, 19: 2068 (1998). Th. Adam, S. Lu¨dtke, and K. K. Unger, Chromatographia, 49: S-49 (1999). K. Walhagen, K. K. Unger, A. M. Olsson, and M. T. W. Hearn, J. Chromatogr. A, 853: 263 (1999). P. Huang, X. Jin, Y. Chen, J. R. Srinavasan, and D. M. Lubman, Anal. Chem., 71: 1786 (1999). C. Yang and Z. El Rassi, Electrophoresis, 19: 2061 (1998). G. A. Lord, D. B. Gordon, P. Myers, and B. W. King, J. Chromatogr. A, 768: 9 (1997). G. Choudhary, Cs. Horva´th, and J. F. Banks, J. Chromatogr. A, 828: 469 (1998). Ch. Fujimoto, Y. Fujise, and E. Matsuzawa, Anal. Chem., 68: 2753 (1996). J.-L. Liao, N. Chen, Ch. Ericson, and S. Hjerte´n, Anal. Chem., 68: 3468 (1996). Ch. Ericson, J.-L. Liao, K. Nakazato, and S. Hjerte´n, J. Chromatogr. A, 767: 33 (1997). A. Palm and M. V. Novotny, Anal. Chem., 69: 4499 (1997). L. Schweitz, L. I. Andersson, and S. Nielsson, Anal. Chem., 69: 1179 (1997). L. Schweitz, L. I. Andersson, and S. Nielsson, J. Chromatogr. A, 792: 401 (1997). L. Schweitz, L. I. Andersson, and S. Nielsson, J. Chromatogr. A, 817: 5 (1998). L. Schweitz, L. I. Andersson, and S. Nielsson, Chromatographia, 49: S-93 (1999).
Fundamentals of Capillary Electrochromatography / 49 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128.
E. C. Peters, M. Petro, F. Svec, and J. M. J. Fre´chet, Anal. Chem., 69: 3646 (1997). E. C. Peters, M. Petro, F. Svec, and J. M. J. Fre´chet, Anal. Chem., 70: 2288 (1998). E. C. Peters, M. Petro, F. Svec, and J. M. J. Fre´chet, Anal. Chem., 70: 2296 (1998). M. Mayer, E. Rapp, C. Marck, and G. J. M. Bruin, Electrophoresis, 20: 43 (1999). R. Asiaie, X. Huang, D. Farnan, and Cs. Horva´th, J. Chromatogr. A, 806: 251 (1998). M. T. Dulay, R. P. Kulkarni, and R. N. Zare, Anal. Chem., 70: 5103 (1998). Q. Tang, B. Xin, and M. L. Lee, J. Chromatogr. A, 837: 35 (1999). G. Chirica and V. T. Remcho, Electrophoresis, 20: 50 (1999). Y. Guo and L. A. Colo´n, Anal. Chem., 67: 2511 (1995). J.-T. Wu, P. Huang, M. X. Li, M. G. Qian, and D. M. Lubman, Anal. Chem., 69: 320 (1997). Z. J. Tan and V. T. Remcho, Anal. Chem., 69: 581 (1997). J. P. Kutter, S. C. Jacobson, N. Matsubara, and J. M. Ramsey, Anal. Chem., 70: 3291 (1998). C. Yan, R. Dadoo, R. N. Zare, D. J. Rakestraw, and D. S. Anex, Anal. Chem., 68: 2726 (1996). B. He, J. Ji, and F. E. Regnier, J. Chromatogr. A, 853: 257 (1999). S. E. G. Dekkers, U. R. Tjaden, and J. van der Greef, J. Chromatogr. A, 712: 201 (1995). K. Schmeer, B. Behnke, and E. Bayer, Anal. Chem., 67: 3656 (1995). S. L. Lane, R. Boughtflower, C. Paterson, and M. Morris, Rapid Commun. Mass Spectrom., 10: 733 (1996). J. T. Wu, P. Huang, M. X. Li, and D. M. Lubman, Anal. Chem., 69: 2908 (1997). M. R. Taylor and P. Teale, J. Chromatogr. A, 768: 89 (1997). P. Huang, J. T. Wu, and D. M. Lubman, Anal. Chem., 70: 3003 (1998). A. Apffel, H. Yin, W. S. Hancock, D. McManigill, J. Frenz, and S. L. Wu, J. Chromatogr. A, 832: 149 (1999). K. Pusecker, J. Schewitz, P. Gfo¨rer, L. H. Tseng, K. Albert, and E. Bayer, Anal. Chem., 70: 3280 (1998).
50 / Pyell 129.
130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144.
145. 146. 147. 148. 149.
J. Schewitz, P. Gfo¨rer, K. Pusecker, L. H. Tseng, K. Albert, E. Bayer, I. D. Wilson, N. J. Bailey, G. B. Scarfe, J. K. Nicholson, and J. C. Lindon, Analyst, 123: 2835 (1998). P. Gfo¨rer, J. Schewitz, K. Pusecker, L. H. Tseng, K. Albert, and E. Bayer, Electrophoresis, 20: 3 (1999). W. Guo, J. A. Koropchak, and C. Yan, J. Chromatogr. A, 849: 587 (1999). M. Qi, X. F. Li, C. Stathakis, and N. J. Dovichi, J. Chromatogr. A, 853: 131 (1999). B. Behnke, E. Grom, and E. Bayer, J. Chromatogr. A, 716: 207 (1995). M. Verzele and C. Dewaele, J. High Resol. Chromatogr., 10: 280 (1987). A. Banholczer and U. Pyell, J. Microcol. Sep., 10: 321 (1998). E. J. Guthrie and J. W. Jorgenson, Anal. Chem., 56: 483 (1984). H. Chen and C. Horva´th, Anal. Meth. Instrument, 2: 122 (1995). T. Enami and N. Nagae, Analusis, 26: M-31 (1998). M. R. Euerby, D. Gilligan, Ch. M. Johnson, and K. D. Bartle, Analyst, 122: 1087 (1997). J. Ding, J. Szeliga, A. Dipple, and P. Vouros, J. Chromatogr. A, 781: 327 (1997). Ch. G. Huber, G. Choudhary, and Cs. Horva´th, Anal. Chem., 69: 4429 (1997). A. S. Lister, C. A. Rimmer, and J. G. Dorsey, J. Chromatogr. A, 828: 105 (1998). Ch. Ericson and S. Hjerte´n, Anal. Chem., 71: 1621 (1999). A. Banholczer and U. Pyell, 23rd Int. Symp. on High Performance Liquid Phase Separations and Related Techniques, Granada, Spain, May 1999. M. R. Taylor, Ph. Teale, S. A. Westwood, and D. Perrett, Anal. Chem., 69: 2554 (1997). Y. Zhang, W. Shi, L. Zhang, and H. Zou, J. Chromatogr. A, 802: 59 (1998). B. Xin and M. L. Lee, J. Microcol. Sep., 11: 271 (1999). A. S. Rathore and Cs. Horva´th, J. Chromatogr. A, 781: 185 (1997). P. D. Grossman, in P. D. Grossman and J. C. Colburn (Eds.),
Fundamentals of Capillary Electrochromatography / 51 Capillary Electrophoresis, Theory and Practice, Academic Press, San Diego, CA, 1992, p. 3. 150. G. Choudhary and C. Horva´th, J. Chromatogr. A, 781: 161 (1997). 151. Q. H. Wan, J. Chromatogr. A, 782: 181 (1997). 152. M. T. Dulay, C. Yan, D. J. Rakestraw, and R. N. Zare, J. Chromatogr. A, 725: 361 (1996). 153. C. J. Paterson, R. J. Boughtflower, D. Higton, and E. Palmer, Chromatographia, 48: 599 (1997). 154. P. Sandra, A. Dermaux, V. Ferraz, M. M. Dittmann, and G. Rozing, J. Microcol. Sep., 9: 409 (1997). 155. A. Dermaux, P. Sandra, and V. Ferraz, Electrophoresis, 20: 74 (1999). 156. R. M. Seifar, J. C. Kraak, H. Poppe, and W. T. Kok, J. Chromatogr. A, 832: 133 (1999). 157. W. Xu and F. E. Regnier, J. Chromatogr. A, 853: 243 (1999). 158. B. Behnke and J. W. Metzger, Electrophoresis, 20: 80 (1999). 159. M. Zhang and Z. El Rassi, Electrophoresis, 20: 31 (1999).
2 Membrane Extraction Techniques for Sample Preparation Jan A˚ke Jo¨nsson and Lennart Mathiasson Lund University, Lund, Sweden
I. SAMPLE PREPARATION TECHNIQUES II. MEMBRANE EXTRACTION TECHNIQUES A. Supported Liquid Membrane Extraction (SLM) B. Microporous Membrane Liquid–Liquid Extraction (MMLLE) C. Polymeric Membrane Extraction (PME) D. Membrane Extraction with a Sorbent Interface (MESI) III. INTERFACING MEMBRANE EXTRACTION AND SEPARATION A. Interfacing with Liquid Chromatography B. Interfacing with Gas Chromatography C. Interfacing with Capillary Electrophoresis IV. WHAT CAN BE ACHIEVED BY MEMBRANE EXTRACTION? A. Cleanup and Selectivity B. Enrichment
54 56 57 62 63 63 65 65 68 70 71 72 76
53
54 / Jo¨nsson and Mathiasson C. Automation and Unattended Operation D. Solvent Consumption V. FIELDS OF APPLICATION OF MEMBRANE EXTRACTION A. Analytes B. Matrices VI. CONCLUSIONS ACKNOWLEDGMENTS REFERENCES
80 82 83 83 84 85 85 86
I. SAMPLE PREPARATION TECHNIQUES In the field of chromatographic analysis, sample preparation is a somewhat neglected aspect, even if it has received increased attention during the last years. It is frequently necessary to separate the analytes from a matrix containing various high-molecular disturbing materials, usually insufficiently characterized. This task is usually called cleanup, and it is one of the main objectives of sample preparation. Well-known examples where cleanup is mandatory are drug analysis in biological liquids such as blood or urine, and trace organic environmental analysis, where humic substances influence the detection of the analytes. Another aspect of sample preparation is enrichment of the analyte—i.e., to increase its concentration over the matrix background in order to decrease the detection limits. Obviously, these objectives are intimately connected. A third objective, which is becoming increasingly important, is the possibility for automation of the entire analytical process. Apart from the higher sample throughput possible, automation permits the use of closed analytical systems, leading to better control of contamination in trace analysis. According to a survey study [1], more than 60% of the total time spent on chemical analysis is used for sample preparation, while only 7% is used for the final analysis per se, using various chromatographic or spectroscopic techniques. It is obvious that there is a great potential for saving by automating the sample preparation process. The most commonly used techniques for sample preparation in organic analysis are liquid–liquid extraction (LLE) and solid-phase extraction (SPE). LLE is the classical technique for sample preparation, and it has been described in several recent reviews [2,3]. This
Membrane Extraction for Sample Preparation / 55 technique was the dominating one a few years ago [4] and is probably still so. For example, in the U.S. Environmental Protection Agency (EPA) protocols for environmental analysis, most methods still prescribe LLE. This technique offers large potential for tuning the extraction by chemical means, such as incorporating different specific reagents. LLE also has a high tolerance for interfering compounds, and it provides physical separation of the extracted analyte from the sample. With LLE, it is possible in many applications to achieve efficiently both cleanup and enrichment. However, there are also some serious drawbacks, such as large consumption of solvents, difficulty of automation and on-line connection to analytical instruments, as well as often troublesome formations of emulsions. SPE is the more modern of these techniques, and it is amply described in several recent books [5,6]. By means of a solid sorbent, available in many chemical and physical forms, analytes are extracted from the sample phase (normally aqueous) and subsequently eluted with an organic solvent. There are a number of ways to automate SPE in high-performance liquid chromatography (HPLC), either by using dedicated instruments, so-called SPE workstations, or by precolumn techniques [6]. The related technique of solid-phase microextraction (SPME) [7,8] is easily automated, especially in connection with gas chromatography. Some drawbacks with the SPE techniques are insufficient retention of very polar compounds, limited selectivity, and high costs associated with disposable sorbent materials. In addition, even if the amount of solvent needed for SPE is decreased compared to LLE, it might still be significant. Some aspects of SPE and LLE are easily compared in the EPA sample preparation protocols. For example, in a generic EPA SPE procedure (Method 3535) [9], 85 mL of organic solvent is needed for extraction of 1 L of water sample (30 mL for elution, 55 mL for washings and conditioning). Method 3510, which is the corresponding LLE method, is applicable to many more analytes. It requires 180 mL of organic solvent, which is more, but not dramatically so. In both these methods, the volume of the extract is reduced to 1–10 mL before analysis. As alternatives to the SPE and LLE techniques, different membrane techniques have been suggested, as recently reviewed by van de Merbel [10]. A simple division is between porous and nonporous membranes. Porous membranes are used in dialysis to separate lowmolecular analytes from high-molecular matrix components, and
56 / Jo¨nsson and Mathiasson cleanup is achieved in that way. There is no discrimination between different small molecules. There is no enrichment possible, rather a dilution. Variations of porous membrane techniques are microdialysis [11,12], which is used extensively in neuroscience research for in-vivo sampling, and electrodialysis [10], in which an electric field is employed to selectively transport charged compounds over a dialysis membrane. In addition, a number of micro- and nanofiltration techniques belong to the field of porous membrane techniques. By a combination of dialysis with SPE, both cleanup and enrichment can be performed. This is the ASTED process [13,14], the basis for a commercial automated instrument (Gilson S.A., Villeiers-le-Bel, France). Typical applications involve drug analysis in blood plasma. This somewhat neglected technique was recently reviewed [10]. Porous membrane techniques are strictly not extraction techniques, and they are not further considered here. In this review, the focus is on membrane extraction techniques utilizing nonporous membranes. A nonporous membrane is a liquid or a solid (e.g., polymeric) phase that is placed between two other phases, usually liquid but sometimes gaseous. One of these phases is the sample to be processed, the donor (or feed ) phase. On the other side of the membrane is the acceptor (or strip) phase, in which the extracted analytes are collected and transferred to the analytical instrument. With this arrangement, the versatile chemistry of LLE can be employed (and extended) in a format that is amenable to automation. This can provide unsurpassed cleanup efficiency and enrichment factors, in most cases with insignificant or no use of organic solvents. The following description will be focused on these techniques, their connection to chromatography and to capillary electrophoresis, as well as their applications to biological and environmental analysis.
II.
MEMBRANE EXTRACTION TECHNIQUES
A number of nonporous membrane techniques have been described for sample preparation in analytical chemistry. The main versions are summarized in Table 1 and described in the following sections. In Fig. 1, typical units for membrane extraction are shown. Usually they are constructed of two blocks of inert material with a machined groove in each. When the blocks are clamped together with a membrane between, a flow-through channel is formed on each side of the membrane; the upper one is the donor channel, and the lower
Membrane Extraction for Sample Preparation / 57 Table 1 Types of Membrane Extraction Phases (Donor/membrane/ acceptor)
Refs.a
Name
Abbreviation
Supported liquid membrane extraction Microporous membrane liquid–liquid extraction Polymeric membrane extraction
SLM
Aq/org/aq
[15,16]
MMLLE
Aq/org/org (org/org/ aq)
[16,17]
PME
Aq/polymer/aq or org/ polymer/aq or aq/ polymer/org Gas/polymer/gas or liquid/polymer/gas
[18]
Membrane Extraction with a Sorbent Interface a
MESI
[19,8]
First ref. for analytical sample preparation, recent review
one is the acceptor channel. For sample preparation, channel volumes are in the range 10–1000 µL. An extraction unit based on a hollow fiber membrane is also shown in the figure. Here the acceptor phase is inside the fiber lumen and the donor channel is the annular volume between the outside of the fiber and the inside of a surrounding tube or cylindrical hole. This type of unit can be made with channel volumes as small as 1 µL [20,21]. With proper modification, the membrane units shown in Fig. 1 are in principle applicable to all versions of membrane extraction.
A. Supported Liquid Membrane Extraction (SLM) Supported liquid membrane extraction has been used for some time for industrial separations, mainly for extraction of metal ions from, e.g., wastewater [24,25]. The use of SLM for sample preparation in analytical chemistry was proposed by Audunsson [15], and the field has been the subject of several earlier reviews [10,16,22,26–28]. In SLM, the membrane consists of an organic solvent, which is held by capillary forces in the pores of a hydrophobic porous membrane. Typical solvents are long-chain hydrocarbons such as nundekane or kerosene and more polar compounds such as di-hexyl ether, tri-octyl phosphate, and others. Various additives can increase the efficiency of extraction considerably, as described below.
58 / Jo¨nsson and Mathiasson
(a)
(b) Fig. 1 (a) Membrane unit with 1-mL channel volume. From Ref. 22 with permission from Elsevier Science. (A ⫽ blocks of inert material, B ⫽ membrane). (b) Membrane unit with 10-µL channel volume. From Ref. 23 with permission. Copyright 1994 American Chemical Society. (c) Hollow fiber membrane unit with 1.3-µL acceptor channel (lumen) volume. From Ref. 21 with permission. Copyright 1997 American Chemical Society. (1 ⫽ O-rings, 2 ⫽ polypropylene hollow fiber, 3 ⫽ fused silica capillaries, 4 ⫽ male nuts).
Membrane Extraction for Sample Preparation / 59
(c)
In Fig. 2, the basic principle for SLM extraction of basic compounds, e.g., amines, is presented. The sample pH is adjusted to a high enough value for the amines to be uncharged, and the acceptor channel on the other side of the membrane is filled with a stagnant acidic buffer. When the sample is pumped through the donor channel, the uncharged amines (B) are extracted into the organic membrane phase. After having diffused through the membrane, an amine molecule will be immediately protonated at the membrane–acceptor interface and therefore prevented from reentering the membrane. This is referred to as trapping, and it results in a transport of amine molecules from the donor to the acceptor phase, which after the extraction is transferred to an analytical instrument, either manually
Fig. 2 Schematic description of the SLM principle. For details, see text.
60 / Jo¨nsson and Mathiasson or on-line by a flow system. As the extract is aqueous, the technique is best compatible with reversed-phase liquid chromatography or ion chromatography. If the trapping is virtually complete, practically all analyte in the acceptor will be in the form of ammonium ions. Therefore, the concentration gradient (which controls the mass transfer rate) of the diffusing species (the free amine) will be practically unaffected by the total concentration of amine in the acceptor phase. This leads to a potentially high degree of concentration enrichment (several hundred or thousand times, depending on volumes and time), when more and more sample is pumped through the donor channel and collected in the acceptor channel. It is obvious that acidic compounds (HA) will be completely excluded from the membrane as they are already charged in the alkaline donor phase. This also holds for permanently charged compounds. Neutral compounds (N) may be extracted, but not enriched, as the concentration in the acceptor phase will never exceed that in the donor. Macromolecules, such as proteins, will typically be charged and therefore rejected. Finally, the extraction rate of uncharged macromolecules will be very low due to their low diffusion coefficients. Thus, under the conditions mentioned, the SLM extraction will be highly selective for small, basic compounds. There are many possibilities to tune the chemistry in the three phases to enrich different classes of compounds. Acidic compounds may be extracted in a similar way as amines by reversing the pH conditions in Fig. 2. Adding reagents, e.g., ion-pairing or chelating, to the donor phase permits the design of SLM extraction systems for various permanently charged compounds and metal ions. Selectivity and mass transfer can be further enhanced by incorporating various carrier molecules or ion complexes in the membrane phase. Various trapping reagents in the acceptor phase can be used to prevent analytes being extracted back into the membrane. It is also possible to add soluble antibodies to the acceptor phase, so analytes can be selectively trapped as antigen–antibody complexes, leading to ultimate selectivity [29]. This principle can also be used for enrichment of permanently neutral species. Summarizing the principles of SLM extraction: neutral, extractable species should be formed in the donor phase (or at the donor– membrane interface); these species should be transported through the membrane and in the acceptor phase become transformed to an-
Membrane Extraction for Sample Preparation / 61 other, nonextractable species. Chemically this is similar to liquid– liquid extraction into an organic solvent, followed by a backextraction into a second aqueous phase. A thorough treatment of the theory and principles of the SLM process has been presented [30], and some highlights are the following. The extraction efficiency E (fraction of analyte molecules that are recovered in the acceptor) is a function of many parameters. These include the magnitude of the partition coefficient of the analyte between the aqueous phases and the organic (membrane) phase, trapping conditions in the acceptor, flow rate of the donor, characteristics of the membrane, and dimensions of the membrane. The influence of the partition coefficient K is somewhat complex. At low values of K, E is small, as the analyte is insufficiently extracted into the organic membrane and the mass transfer is limited by the diffusion transfer through the membrane. At intermediate values, the mass transfer is limited by the transport properties in the flowing donor phase, and in this region the most efficient extraction is obtained. At very high values of K, i.e., for very hydrophobic compounds, the stripping of analyte into the acceptor phase becomes the limiting factor, and the observed extraction efficiency will decrease, as relatively large amounts of analyte will be left in the membrane. In a recent study [31], it was found that the most efficient extraction is obtained when the octanol–water partition coefficient as a rough estimate is around 103. The trapping conditions in the acceptor are also important for the extraction efficiency. If the trapping is not complete, the extraction efficiency will decrease with time, leading to less precise quantitation. This was detailed in a recent paper [32]. The extraction efficiency is highest for very low donor flow rates, and decreases as the flow rate increases. On the other hand, increasing the donor flow rate also increases the amount of analyte that is introduced into the extraction system, and the net result often is an increase in the amount of accumulated analyte in the acceptor during a given time. Thus, given enough sample volume, high flow rates lead to lower detection limits. With very small sample volumes, a low flow rate might be needed in order to squeeze as much analyte as possible from the sample. Thus, it is obvious that it is not necessary to strive for the maximum value of E, and that this parameter should not be confused with recovery. For good quantitative perfor-
62 / Jo¨nsson and Mathiasson mance, the important issue is to find conditions that lead to reproducible values of E, and the value of this parameter will be included in the calibration. This is analogous to, e.g., mass spectrometry, where the ionization efficiency usually will be far below 100%, still providing satisfactory quantitation.
B.
Microporous Membrane Liquid–Liquid Extraction (MMLLE)
As mentioned above, the SLM technique will not work well for highly hydrophobic compounds, especially if they cannot be trapped efficiently. Examples of such compounds are hydrocarbons, PCBs, etc., which can easily be extracted from water into an organic solvent, but cannot be extracted out into an aqueous acceptor as required by the SLM principle. For such applications, the technique of MMLLE is more suitable. Here the acceptor phase is an organic solvent and the same solvent forms the liquid membrane by filling the pores in the porous hydrophobic membrane [16]. In Fig. 3, the principle of MMLLE is sketched. Chemically, this is analogous to conventional liquid–liquid extraction, but performed in a flow system, easily automated and interfaced to analytical instruments. As the extract ends up in an organic phase, the technique is most easily interfaced to gas chromatography (GC) or to normal-phase liquid chromatography (NP-HPLC). In principle, the membrane could also be hydrophilic, which would lead to aqueous phase in the membrane pores. As far as we know, this has not yet been tried for analytical purposes.
Fig. 3 Schematic description of MMLLE. For details, see text.
Membrane Extraction for Sample Preparation / 63 LLE in a flow system (in the form of flow injection analysis) has been described many times, as reviewed by Valca´rcel [33], but then the organic and aqueous phases are mixed in the same flow channel and later separated. The practical problems with the phase separation seem to have prevented this technique being widely used. In MMLLE, the phases are never mixed and all mass transfer between the phases takes place at the membrane surface. The extraction efficiency is limited by the partition coefficient. If it is very high, it is possible to work with a stagnant acceptor and still get considerable enrichment into a very small extract volume. With smaller partition coefficients, it might be necessary to arrange a slow flow of the acceptor phase to remove the extracted analyte and maintain the diffusion through the membrane.
C. Polymeric Membrane Extraction (PME) One of the drawbacks of SLM extraction is the possible instability of the liquid membrane, even if this problem usually seems to be overrated, at least for analytical laboratory applications, where a membrane lifetime of several days to weeks is sufficient. This is discussed further in Section IV.C. In any case, by exchanging the SLM with a polymeric membrane, usually a silicon rubber membrane, the membrane lifetime is considerably increased. However, with a fixed composition of the membrane, the possibilities for chemical tuning (e.g., the application of carriers) of the separation process is greatly reduced. This is probably most serious for relatively polar analytes, where the hydrophobicity of the membrane has to be reduced. Also, as diffusion coefficients in polymers are lower than in liquids, the mass transfer is slower, leading to slower extractions. On the other hand, as the membrane is virtually insoluble, any combination of aqueous and organic liquids can be used. Melcher [18,34] has developed this principle, both with a trapping acceptor (as described above for SLM), and with a solvent in the acceptor channel, leading to a situation similar to MMLLE.
D. Membrane Extraction with a Sorbent Interface (MESI) The above techniques are all characterized by liquid donor and acceptor phases. For easy interfacing with gas chromatography a gaseous acceptor phase is the most convenient. The MESI technique
64 / Jo¨nsson and Mathiasson realizes this principle [8,19]. It can be applied to either gaseous or aqueous samples, and the equipment consists of a membrane module with a (usually) silicone rubber hollow fiber, into which the analytes are extracted from the surrounding liquid or gaseous sample. The technique works best for volatile and relatively nonpolar compounds. Inside the fiber flows a gas, which transports the analyte molecules away from the membrane into a cooled sorbent tube where they are trapped. By heating the sorbent trap, the analytes are desorbed and transferred to GC analysis. Incidentally, the MESI principle can be seen as a gas-phase analogy to the ASTED principle for liquid chromatography; see above. In Fig. 4, a typical MESI setup is shown. All components are connected in-line so that the carrier gas for the GC passes through the membrane fiber and the sorbent trap, resulting in a completely integrated instrument setup. Sampling can also be made off-line with the extraction module and sorbent trap in, e.g., field sampling, and the sorbent trap can later be connected to the GC and desorbed in a separate step. Matz [35] recently presented and compared this and a few other variants.
Fig. 4 Components of the MESI system coupled to a gas chromatograph. (From Ref. 8, with permission.)
Membrane Extraction for Sample Preparation / 65
III. INTERFACING MEMBRANE EXTRACTION AND SEPARATION Some of the membrane techniques can be performed in a manual and off-line way, not connected directly to a chromatographic or electrophoresis analysis instrument. A good example is a manual version of SLM extraction (named LLLME) [36] utilizing disposable units for extraction of amphetamine from blood plasma samples and subsequent off-line analysis by capillary electrophoresis. For environmental applications [37,38], the extracts from large SLM units are sometimes collected manually and injected into the HPLC using an autosampler. This permits sampling outside the lab and duplicate chromatographic analyses. In addition, connections to flow-injection analysis systems (FIA) can be mentioned, where the uses of gas-diffusion membranes (aq/ gas/aq), as well as dialysis cells, are widespread [33]. Early realizations of MMLLE [39] and PME [40] also involved FIA systems. More recent examples are an SLM-FIA setup for Cr(VI) [41] and an MMLLE-FIA application involving a chemometric evaluation of diode array spectra after an aq/org extraction was recently presented [42]. However, the membrane extraction techniques are well suited for automatic connection to chromatographic instruments, and this principle leads to many advantages, so it will now be discussed in some detail.
A. Interfacing with Liquid Chromatography Several of the membrane extraction techniques, such as SLM and PME, typically lead to an aqueous extract. Such extracts can most straightforwardly be analyzed with reverse-phase liquid chromatography (RP-HPLC). For this purpose, flow systems can easily be built up around peristaltic pumps and pneumatic valves, controlled by electronic timers, integrators, or computer systems. A typical such system is shown in Fig. 5. It was originally set up for SLM extraction of chlorinated phenols from natural waters [43], but was used also for other environmental applications [44,45]. The sample is acidified and pumped with a peristaltic pump (1) through the donor channel. During the extraction, the alkaline acceptor is kept stagnant. By switching a valve (5) after the extraction,
66 / Jo¨nsson and Mathiasson
Fig. 5 Schematic diagram of flow system for SLM extraction of chlorinated phenols. (From Ref. 43, with permission from Vieweg Publishing.)
it is transported further on, neutralized, and moved to a precolumn (9), where the analytes are adsorbed and focused. After switching the injection valve (8), the analytes are transferred to the analytical column. Reagent water for washing the donor channel between the samples can be introduced by means of a valve (2). There are also provisions to rinse the precolumn with acid between runs. This type of system has been used mainly in environmental applications for extraction of relatively large amounts of natural water with large membrane units (channel volumes 1 mL). The precolumn ascertains that all extracted analytes in one extract are analyzed in one chromatographic run. Alternatively, a smaller membrane unit can be used, so that the entire extract (or a major part of it) can be contained in the injection loop and thus injected directly into the liquid chromatograph without a precolumn [46,47]. The same approach was used by Melcher [18] and by researchers from Salamanca, Spain, in a series of papers dealing with MMLLE in org/aq or org/org configurations [48–50]. For biological samples of about 1 mL, the peristaltic pump/ solenoid valve approach is not suitable. More accurate instrumentation can be constructed using an autosampler and syringe pumps. As an example, a fully automatic system is shown in Fig. 6. It was built around an ‘‘intelligent sample processor’’ Model 231 (Gilson)
Membrane Extraction for Sample Preparation / 67
Fig. 6 Experimental setup for SLM-HPLC determination biomolecules in blood plasma or urine. (Adapted from Ref. 23 with permission. Copyright 1994 American Chemical Society.)
and originally constructed for the determination of basic drugs in blood plasma [23]. The same principle was used also in other studies [47,51,52], and this is probably the most versatile setup for automated SLM-HPLC applications. The plasma samples are held in vials in an autosampler rack. Immediately before extraction, the alkaline donor buffer is added by means of the syringe pump and robotic needle. The alkalized sample is pressed through the donor channel in the membrane unit. After the extraction is completed, the contents (10 µL) of the acceptor channel are transferred by means of a second syringe pump into the injection loop and subsequently injected into the chromatographic column. The operations of the sample preparation system are synchronized with the chromatographic computer system, so one sample is extracted during the chromatographic run of the previous sample. Thus, the sample throughout is determined by the length of the chromatogram, typically ca. 15 min. A miniaturized SLM extraction cell, comprising a hollow fiber membrane and having 1.3 µL acceptor volume inside the fiber lumen, was developed and satisfactorily connected to packed-column capillary liquid chromatography by using a miniaturized loop injector [20].
68 / Jo¨nsson and Mathiasson Often the extract after MMLLE and PME is organic, and in those cases, interfacing to NP-HPLC is more suitable. This can be realized with an autosampler in essentially the same way as described above [53].
B.
Interfacing with Gas Chromatography
The interfacing between a membrane technique and capillary GC is an integral feature in MESI for volatile compounds. The extract is in the form of a gas and the analytes are trapped on a sorbent column and thermally desorbed directly into the GC column; see Fig. 4. Slightly different technical realizations of this idea have been presented [8,35,54,55], differing mainly in the type of heating and the physical arrangement of the membrane. For liquid extracts, GC with capillary columns poses quite stringent demands on the sample: it should be essentially water-free, and the volume is quite restricted. This means that SLM and other techniques leading to aqueous extracts are less suitable for on-line connection to GC. With packed columns, however, which can tolerate injection of aqueous samples, automated connections were early realized [56–58]. By exchanging the aqueous solvent for hexane in a solvent–exchange interface, as described by Vreuls and co-workers [59], it was possible to interface an SLM extraction system to capillary GC [51]. The resulting system was quite complex, although completely automated extraction and analysis of local anesthetics in blood plasma was achieved with good performance. Membrane extraction techniques such as MMLLE and PME with organic acceptors can be interfaced to GC relatively easily. This can be realized by means of large-volume injection methodology, a topic that recently was thoroughly reviewed [60]. As a simplification of the above-mentioned SLM-GC system [51], an MMLLE-GC system for the same application was constructed, resulting in a system that was considerably more easily operated and more rugged [17]. This system is shown in Fig. 7. The extraction part is similar to that in Fig. 6: the donor phase is pumped with the syringe pump (1), the samples in the autosampler (2) are alkalized one by one and passed through the donor side (3) of the membrane unit. With the solvent pump (5), acceptor solvent (hexane) is slowly pumped through the acceptor channel (4) and subsequently to a 400-µL loop (7) connected to an injection valve (8). Provision for nitrogen purging (5) was needed to decrease carryover
Membrane Extraction for Sample Preparation / 69
Fig. 7 Experimental setup for MMLLE followed by capillary gas chromatography for determination of local anesthetics in blood plasma. (Adapted from Ref. 17 with permission. Copyright 1998 American Chemical Society.)
effects. After extraction, the contents of the loop were transferred to the GC, equipped with a retention gap (10), a retaining precolumn (11), a solvent vapor exit (14), and a capillary column (12). The GC system was set up as described by Grob and Stoll [61]. By miniaturization, the connection MMLLE-GC can be further simplified. A device called Extraction Syringe (ESy) has recently been described [62]. See Fig. 8. The main part is a single hydrophobic hollow fiber (1), mounted in the center of a Kel-F piece (A; 3) with a drilled hole. A stainless steel needle (B, 2) is extruding from the end of the fiber. The lumen of the fiber contains the organic acceptor phase, with a volume of a few microliters (5). The sample to be extracted is pumped (6) around the fiber and the analytes are partitioned into the organic solvent. The instrument is placed directly on
70 / Jo¨nsson and Mathiasson
Fig. 8 Schematic picture of the Extraction Syringe. For symbols, see text. (Adapted from Ref. 62 with permission of the Royal Society of Chemistry.)
top of a gas chromatograph for automated extraction and injection onto the GC by means of a pneumatic piston (C), so the operation mimics the operation of an autosampler injection. With a few microliters of injected volume, no special arrangements have to be made with the gas chromatograph; a conventional splitless injection is appropriate. Although this type of MMLLE-GC connection is yet little tested, it seems that this principle has significant advantages over other principles for connection of LLE to gas chromatography, and as such it should have a large application potential.
C.
Interfacing with Capillary Electrophoresis
As the sample volume in capillary electrophoresis (CE) has to be very small, typically in the nanoliter region, on-line connection of sample pretreatment devices is difficult. By means of various socalled stacking procedures, several microliters can be introduced and the analytes compressed in the beginning of the separation capillary. This simplifies the task of in-line connection, but additional aspects such as high voltages hazards still make the topic difficult and inconvenient. Two recent reviews [63,64] describe this in detail.
Membrane Extraction for Sample Preparation / 71 There are not many examples of direct connection of nonporous membrane techniques on-line with CE, but a number of dialysis applications have been presented [10]. Bao and Dasgupta [65] connected a short piece of hollow fiber membrane in-line with the capillary. This was made with both porous membranes for dialysis and gas-phase transfer and with a polymeric (silicone rubber) membrane. In the latter case, acceptor trapping of phenols in an alkaline buffer was accomplished in the same way as described for SLM above, and significant enrichment was obtained. As the membrane was short in comparison with the separation capillary, no stacking was required and the plate number of the separation was not appreciably influenced. There seems not to be any follow-up of this work, which seems rather promising, although it is easy to predict a number of practical problems. Another principle for membrane–CE interfacing is to utilize a separate micromembrane cell (Fig. 1c), equipped with a hollow fiber membrane with a small volume [21]. This approach provides a more stable system and necessitates a stacking procedure, giving an increased overall concentration enrichment. The setup was not entirely automated, as it required a manual connection of the capillary to the outlet of the membrane unit, although this in principle could be handled with a suitable valve.
IV.
WHAT CAN BE ACHIEVED BY MEMBRANE EXTRACTION?
In this section, the main advantages with membrane extraction in sample preparation for chromatography and related techniques will be discussed. Compared to other techniques, the membrane extraction techniques provide advantages in various ways. Membrane extraction probably offers the highest degree of selectivity and cleanup from complicated matrixes of all known techniques, and it is possible to achieve very high enrichment factors with preservation of the selectivity. Automation and on-line connection to instruments for chromatographic final analysis can be readily made and, compared to most other techniques for sample preparation, the use of organic solvent is much reduced, in most cases essentially to zero. This is true for a large number of compound classes, in various matrices and concentration ranges.
72 / Jo¨nsson and Mathiasson
A.
Cleanup and Selectivity
All types of nonporous membrane extraction procedures will in principle lead to a high degree of cleanup, especially between small and large molecules. The analytes to be analyzed must dissolve into the membrane, pass through it, and redissolve in the acceptor phase. In many cases, the conditions of extraction can be set so that this chain of events is possible only for a strictly limited range of compounds. The possibilities to achieve this tuning are best for the SLM technique, where selective reactions in all three phases can be utilized for this purpose, but also for the other membrane extraction techniques, there are a number of such possibilities. A simple example is the already-mentioned principle for specific extraction of basic compounds from an alkaline sample (donor) to an acidic acceptor. To increase the selectivity further, the pH difference between the donor and acceptor can be optimized, so only amines of a limited range of pK a are extracted. As an example, if the pH is 11 and 2 in the donor and acceptor, respectively, both aliphatic and aromatic amines with pK a in the approximate range 4–11 will be extracted and enriched (5–10 with optimal efficiency). On the other hand, with an acceptor pH of 5, only amines with pK a in the range 7–11, i.e., preferentially aliphatic amines, will be enriched, as the more weakly basic aromatic amines will not be trapped efficiently. By selecting the polarity of the membrane liquid to mach the polarity of the analytes, the selectivity can be further increased. With certain additives in the membrane phase, the extraction properties can be changed radically. An example of this is shown in Fig. 9, where the extraction of carboxylic acids of different polarities is strongly influenced by the contents of TOPO (tri-octyl phosphine oxide) in the membrane [66]. The most polar acid, lactic acid, was not extracted at all without TOPO, but the extraction was significantly improved by the additive. Butanoic acid, however, was well extracted without TOPO, and essentially unaffected by its concentration. There are a number of other possible additives that have been used in SLM extraction for enhancing the extraction efficiency of different classes of compounds, such as chelating or complexing reagents, crown ethers, ion-pair formers, artificial receptors, etc. Some of them are mentioned in the applications section below.
Membrane Extraction for Sample Preparation / 73
Fig. 9 Influence of TOPO content in di-n-hexyl ether on extraction efficiency E. (Adapted from Ref. 66, with permission from Elsevier Science.)
Both in biomedical analysis and in environmental analysis, an important objective for sample preparation is to remove highmolecular-weight material. In the case of biomedical analysis, such material is usually proteins, and in environmental applications it is mainly humic substances. The membrane extraction techniques are all very efficient in this respect; such high-molecular-weight compounds are often charged and therefore not extracted into organic liquids. Even if they are noncharged, the transport is so slow that their extraction is negligible. Therefore, membrane extraction leads to very clean extracts in different applications. In Fig. 10 [23] it is shown how some basic drugs, extracted from blood plasma with the instrumentation shown in Fig. 6, gave HPLC chromatograms that can hardly be distinguished from extractions from pure aqueous buffer solution. However, the response in blood plasma was significantly lower. That effect was traced to binding of the drug to plasma proteins, and preliminary studies of the possibility to evaluate protein binding effects including their kinetics have been started. Another example (Fig. 11) [52] concerns the extraction of polar drug metabolites from urine, where the chromatograms again are virtually indistinguishable from water solutions. Here, however, the
74 / Jo¨nsson and Mathiasson
Fig. 10 (a) Chromatograms of Amperozide (I), its metabolite (II) and homolog (III) with the subsequent blank after enrichment from blood plasma. (b) Corresponding chromatograms after enrichment from an aqueous buffer solution. Concentrations 4µg/mL of I and II, 8 µg/mL of III. (From Ref. 23, with permission. Copyright 1994 American Chemical Society.)
Membrane Extraction for Sample Preparation / 75
Fig. 11 Chromatograms of a water solution (a) and a urine sample (b), both spiked with 3-OH-PPX (1; 1.0 µM ), 4-OH-Ropivacain (2;0.80 µM ), 3-OHRopivacain (3; 0.83 µM ), PPX (4; 1.0 µM ), Iso-PPX (5; 0.84 µM ), and Ropivacaine (6; 0.90 µM ). (From Ref. 52, with permission of Elsevier Science.)
76 / Jo¨nsson and Mathiasson response is the same in urine and water, as there is no protein binding involved. Figure 12 compares the cleanup possibility of SLM extraction and SPE for triazine herbicides in spiked natural water [67]. The SPE chromatogram (a) shows a characteristic ‘‘humic hump,’’ with the analytes influenced by matrix peaks of nearly the same order of magnitude. The chromatogram after SLM extraction (b), with analyte concentrations that are twice lower, shows neither the ‘‘hump’’ nor any disturbing matrix peals, thereby demonstrating a higher degree of cleanup resulting in lower detection limits. The above examples all refer to SLM extraction. Although there are fewer degrees of freedom in PME and MMLLE, here also good selectivity can be obtained. In Fig. 13 [50] is shown chromatograms of sunflower oils, unspiked and spiked with triazine herbicides, after an org/aq MMLLE extraction into an acidic acceptor. The membrane device was on-line connected to RP-HPLC as described above.
B.
Enrichment
The different membrane extraction techniques behave differently when it comes to concentration enrichment factors. In aq/org types of extraction (MMLLE and PME), the maximum concentration enrichment factor is limited to the value of the partition coefficient between the donor and the acceptor phases. In those techniques, appreciable extraction factors are possible only when the partition coefficient is large; the same situation as for ordinary LLE. This does not preclude that considerable enrichment factors can be obtained when the conditions are favorable. Extraction enrichment factors of about 250 times were obtained in MMLLE extraction of cationic surfactants in natural water [53]. Also, in an aq/org PME-GC determination of chlorinated hydrocarbons and other compounds [34], extraction factors up to 200 times were obtained. In SLM, on the other hand, the enrichment factors are not limited by the partition coefficient, but from the trapping conditions in the acceptor phase, as was recently detailed [32]. With some simplifications, the maximum enrichment factor Ee(max) for SLM extraction of a basic compound as described in Fig. 2 and the accompanying text, depends on the acceptor pH (pHA ) by log Ee(max) ⫽ pK a ⫺ pH A
Membrane Extraction for Sample Preparation / 77
(a)
(b) Fig. 12 Chromatograms (LC-UV) of methoxy-s-triazine herbicides: (a) SPE extraction of spiked river water (1.0 µg/L of each analyte); (b) SLM extraction of spiked river water (0.5 µg/L of each analyte). Peak designation: 1, Simetone; 2, Atratone; 3, Secbumetone; 4, Terbumetone. (Reproduced from Ref. 67 with permission from Elsevier Science.)
78 / Jo¨nsson and Mathiasson
Fig. 13 Chromatograms of sunflower oil (a) before and (b) after spiking with 1.0 ppm of each triazine: (1) atrazine, (2) ametryne, (3) prometryne, (4) terbutryne. (From Ref. 50, with permission of the authors and Elsevier Science.)
Membrane Extraction for Sample Preparation / 79
Fig. 14 Enrichment factors of aniline (1), 3-chloro-4-methylaniline (2), 3,5dichloroaniline (3), and 3-methyl-5-nitroaniline (4), all 0.1 mg/L. Acceptor: 0.1 M sulfuric acid (pH ⬇ 1). (From Ref. 32, with permission. Copyright 1998 American Chemical Society.)
The dissociation constant of the analyte is pK a . Thus, with reasonably strong bases, it is easy to achieve large values for the maximum enrichment factor. On the other hand, as detailed below, to really obtain high enrichment factors, a high sample/extract volume rate is necessary. This is illustrated in Fig. 14, showing the attainment of the maximum enrichment factor for four aniline derivatives with pHA ⫽ 1. For aniline itself (1), pK a ⫽ 4.6 leads to a maximum enrichment factor of about 4000 times. This is apparently not reached until after long extraction times (the experiment was ended after 25 h of extraction and 6 L of sample, giving a final enrichment factor of about 2000 times and still increasing). On the other hand, for the weakly basic 3,5-dichloroaniline (3), pK a ⫽ 2.5, giving the maximum enrichment factor of only about 32 times, which is quickly reached after a short time of extraction. The situation can be improved by increasing the acid concentration in the acceptor, as shown in the cited work. Seen in another way, the enrichment factor obtained is given by the following relation: Ee ⫽ E ⋅
VS VA
80 / Jo¨nsson and Mathiasson where VS is the volume of the extracted sample and VA is the volume of the extract, in SLM the volume of the acceptor channel. From this equation, it is seen that even if E approaches 1, the enrichment factor is never larger than the volume ratio. The strength of SLM in this context is that it can provide relatively high extraction efficiencies at the same time as the extract volume is kept small. For nontrapped techniques such as MMLLE, it might not be possible to achieve a large E with a stagnant acceptor (if the partition coefficient is not enough large) and therefore the acceptor must be pumped, leading to larger VA and smaller Ee . The same, to a higher degree, is true for dialysis. This limitation in MMLLE is overcome by introducing a secondary focusing step. For capillary GC applications this can be a retention gap [17], and for LC the solvent strength can be selected so a column focusing effect is obtained [53]. The same limitation is overcome in MESI (and in the ASTED dialysis approach) using a solid-phase column where the analytes are trapped. For these cases, the VA in the equation above could be considered as the desorption volume of the solid-phase trap. Obviously, selectivity is a prerequisite for enrichment; it is pointless to enrich disturbing compounds also. In Fig. 15 is shown a result from a combined SLM-HPLC-CE application [68], where Bambuterol, a basic drug, was extracted from blood plasma using SLM. Further, it was introduced into a micro-HPLC column, from where a heart cut was transferred to the CE and finally analyzed using the double-stacking procedure and enantiomeric separation. The overall concentration enrichment factor is here about 40,000 times and the detection limit in blood plasma is ca. 0.15 nM for each of the Bambuterol enantiomers with simple UV detection in CE. The main origin of the high enrichment factor is here not the SLM step, but in analyte focusing on the HPLC column and in the stacking, but it would be impossible without the high degree of selectivity provided by the SLM step.
C.
Automation and Unattended Operation
As the membrane extraction process can be most conveniently performed in flow systems, it is easy to devise arrangements employing pumps, autosamplers, solenoid and rotary valves, etc., with computer control that can provide more or less automated operation. In Section III and Fig. 4, a number of examples of such systems are shown.
Membrane Extraction for Sample Preparation / 81
Fig. 15 Electropherograms showing plasma containing 10µM physiostigmine as a protease inhibitor (a) and plasma additionally containing 0.5 nM of each Bambuterol enantionmer. Peaks labeled A are the Bambuterol enantionmers and the peak labeled B is the physiostigmin. (Adapted from Ref. 68, with permission of ICS Technical Publications, Inc.)
For unattended operation of automated systems, the high selectivity and cleanup possible with membrane extraction is an advantage, as it prolongs the usable lifetime of columns, etc. On the other hand, membrane stability could be a limiting factor. For the PME, MMLLE, and MESI techniques, this is not a big problem, as the membranes used are polymeric and durable. Sometimes, fouling by
82 / Jo¨nsson and Mathiasson dirty samples can be seen, but it is possible to devise washing schemes in automated membrane extraction to diminish the problem. In any case, a smooth membrane surface is less amenable to fouling than, e.g., an SPE column, and it is easier to wash. It was noted [20] that the pore size of the membrane could have an important impact on the membrane fouling in MMLLE of blood plasma samples. For SLM extraction, the membrane stability is less obvious, and the issue is often raised. Here, an organic solvent is held in the pores of a hydrophobic porous membrane placed between two aqueous flowing streams. Obviously, this demands that the solvent used is nonsoluble in water and that the capillary forces that hold the liquid in the pores are enough strong to withstand inevitable pressure differences over the membrane. Practically, these potential problems are not crucial. No serious problems with pressure differences have been ever seen with SLM setups, and simple calculations show that pressures of several bar are necessary to ‘‘blow’’ out the organic phase from the pores in typical cases. The solubility of nonpolar solvents in water is very small, and a solvent such as n-undecane, which has been used extensively, forms membranes that are stable for months. Some problems may be encountered when more polar membranes are needed. A medium-polar solvent that has been used extensively is di-n-hexyl ether, which is stable in SLM membranes nearly as well as n-undecane. The inclusion of various additives might compromise the stability, and the matter calls for careful attention. Additives as hydrophobic as possible (maybe modified with alkyl chains) are advantageous. The hydrogen-binding additive TOPO, as mentioned above, can be readily used. Membrane preparations with 10% TOPO in di-n-hexyl ether are stable for weeks [52]. The material of the liquid membrane support seems to influence the stability somewhat, the most commonly used PTFE membranes being slightly better than polypropylene membranes. The regeneration of the SLM is made in a few minutes by simply soaking the membrane support in the desired liquid, wiping, and reinstalling the SLM in the membrane holder. For hollow fiber membranes, in-situ regeneration has been shown to work well [20].
D. Solvent Consumption Compared to alternative sample preparation techniques, membrane extraction demands very little solvent. This is a significant advan-
Membrane Extraction for Sample Preparation / 83 tage, as the cost of high-purity solvent is high, both for purchasing and for destruction. Even more important, the environmental implications of these solvents are considerable, both for the laboratory workers and for the outer environment. This is especially true for chlorinated solvents, where different types of restrictions and bans are discussed and partly already implemented in certain countries. MESI and PME with aqueous acceptor do not require any solvent, and SLM extraction requires only negligible amounts of high-boiling organic liquid in the membrane. Only for MMLLE and PME with an organic acceptor are small amounts of conventional organic solvents needed. For the MMLLE described above for analysis of local anesthetics in blood plasma [17], less than 1 mL of hexane was used for each sample and for the ESy operation only ca. 20 µL was needed [62]. Thus, the membrane extraction techniques compare favorably with the alternative techniques in terms of solvent consumption.
V. FIELDS OF APPLICATION OF MEMBRANE EXTRACTION The membrane extraction techniques are applicable to a variety of analytical problems. Below, a bibliography is presented which is assumed to be relatively complete for the analytical use of nonporous membrane extraction techniques in various applications.
A. Analytes With SLM extraction, acidic or basic compounds can be extracted with simple procedures involving a pH difference between donor and acceptor (see Section II.A). There are many applications with acidic compounds, mainly in environmental applications, for example, phenoxy acids [69–71], sulfonyl urea herbicides [44,72], phenolic compounds [43,73,74], salicylic acid, and carboxylic acids [46,47,66,75– 77]. Also, a number of basic compounds have been extracted with SLM extraction, such as aliphatic amines [15,56–58,78], triazine herbicides [32,37,38,67,79], aniline derivatives [80], as well as various basic drugs and drug metabolites [20,21,23,36,51,52,68,81–83]. For compounds other than simple acids or bases, alternative chemical extraction schemes have been suggested involving ion pairing, chelation, complex formation, and immunological recognition. This has been applied to amino acids [84–89], metal ions [90–93], and anionic surfactants [45].
84 / Jo¨nsson and Mathiasson There are also a few applications of polymeric (silicon rubber) membrane extraction of phenols [18,49,94,95] and salicylic acid [96] using an aqueous trapping basic acceptor, a principle that is very similar to SLM extraction With MMLLE and PME, leading to an organic extract, other types of analytes can be extracted. This applies mainly to nonionizable compounds such as toluene, chlorobenzenes, and naphthalene [18,34,97], but also to vitamin E [48], triazine herbicides [42], cationic surfactants [53], and organotin compounds [93] Finally, with MESI, where the extract is gaseous, only more or less volatile compounds are extracted. This technique has been applied to solvents such as benzene, toluene, ethylbenzene, chlorobenzene, xylenes, and similar compounds [35,54,55,98–100].
B.
Matrices
As the membrane extraction techniques are intrinsically very selective, the main application areas for these techniques are relatively complicated matrices. In the biological field, a number of applications of blood plasma analysis have been shown. In this case, it is in many cases possible to achieve in one step a very high degree of cleanup, so the analytes of interest are more or less completely separated from the matrix [17,20,21,23,36,51,57,68,74,81,82,96]. Other matrices of biological origin were urine [36,52,56,83,101] and manure [75]. Another important field of application for membrane extraction is environmental analysis. A large number of extractions from natural water (lake or river water) as well as wastewater have been presented [37,38,42–45,53–55,67,69,72,79,80,91,92,102]. In a few cases, a field version of SLM extraction has been used to perform onsite time-integrating sampling of herbicides in a river [70,71]. Some studies of trace organics in nutrient solutions for hydroponic cultivation of tomatoes [73] and in soil solutions [47,76] were performed. Both outdoor [46,58,78] and indoor [77] air have been sampled using impinger and washbottle techniques, and the extracts were extracted and enriched by membrane extraction. Air, both ambient and in headspace over a liquid or solid sample, has also been extracted with the MESI technique [54,98,99]. Finally, a number of industrial and food matrices were extracted, such as process water [18,34,97], oils [49,50,94,95], and butter [48].
Membrane Extraction for Sample Preparation / 85
VI.
CONCLUSIONS
The selection of a membrane extraction technique for sample preparation, as well as selection among different such techniques, depends mainly on the properties of the analytes and the matrix, and also on the requirements for detection limits, automation, etc. Generally, as described above, the membrane techniques offer in many cases superior cleanup and enrichment while allowing good possibilities for automation and in-line connection to the final analytical instruments if required. For suitable analytes, mainly acidic and basic compounds, but also other ionizable compounds, the SLM extraction provides the highest selectivity and enrichment possibilities. With uncharged compounds and very nonpolar compounds, MMLLE (or ESy) would be a better choice. The use of PME would be motivated for cases where exceptional membrane stability is required, and this approach can be used for all combinations of aqueous and organic phases, and therefore utilize both SLM chemistry with a trapping acceptor and MMLLE where a conventional LLE chemistry is applied. For volatile analytes and direct connection to gas chromatography, MESI is the preferred alternative. It can be applied to both gaseous and aqueous samples as well as to the headspace over a liquid or solid. In a near future, the ESy technique may become an important competitor where the need for an enrichment column is eliminated.
ACKNOWLEDGMENTS This work was over the years supported financially by grants from the Swedish Natural Science Research Council (NFR), the Swedish Environmental Protection Agency (SNV), the Swedish Council for Forestry and Agricultural Research (SJFR), the Swedish Institute, the Swedish International Development Co-operation Agency (SIDA), the Crafoord Foundation, and the European Community (DG XII). In addition, the companies Pharmacia AB, Astra Draco AB, and Astra Pain Control AB have contributed with funds and interesting applications. A number of graduate and undergraduate students as well as guest researchers have made important contributions, which are highly appreciated.
86 / Jo¨nsson and Mathiasson
REFERENCES 1. W. Pipkin, Am. Lab., Nov: 40D (1990). 2. J. R. Dean, Extraction Methods for Environmental Analysis, Wiley, Chichester, UK, 1998. 3. A. J. Holden, in A. J. Handley (Ed.), Extraction Methods in Organic Analysis, Sheffield Academic Press, Sheffield, UK, 1999, p. 5. 4. R. E. Majors, LC-GC Int., 10: 93 (1993). 5. J. S. Fritz, Analytical Solid-Phase Extraction, Wiley, New York, 1999. 6. E. M. Thurman and M. S. Mills, Solid-Phase Extraction, Principles and Practice, Wiley, New York, 1998. 7. J. Pawliszyn, Solid-Phase Microextraction, Theory and Practice, Wiley, New York, 1997. 8. Y. Luo and J. Pawliszyn, in A. J. Handley (Ed.), Extraction Methods in Organic Analysis, Sheffield Academic Press, Sheffield, UK, 1999, p. 75. 9. SW-846, Test Methods for Evaluating Solid Waste: Physical/ Chemical Methods, 3rd ed., U.S. Government Printing Office, Washington, DC, 1986 (updates 1992, 1993, 1995). 10. N. C. van de Merbel, J. Chromatogr. A., 856: 55 (1999). 11. S. M. Lunte and C. E. Lunte, in P. Brown and E. Grushka (Eds.), Advances in Chromatography, Vol. 36, Marcel Dekker, New York, 1995, p 383. 12. T. E. Robinson and J. B. Justice, Microdialysis in the Neurosciences, Elsevier, Amsterdam, 1991. 13. D. C. Turnell, and J. D. H. Cooper, J. Chromatogr., 395: 613 (1987). 14. D. C. Turnell and J. D. H. Cooper, J. Autom. Chem., 8: 151 (1986). 15. G. A. Audunsson, Anal. Chem., 58: 2714 (1986). ˚ . Jo¨nsson and L. Mathiasson, Trends Anal. Chem., 18: 318 16. J. A (1999). ˚ . Jo¨nsson, and L. Mathiasson, Anal. Chem., 70: 17. Y. Shen, J. A 946 (1998). 18. R. G. Melcher and S. A. Bouyoucos, Process Control Qual., 1: 63 (1990). 19. K. F. Pratt and J. Pawliszyn, Anal. Chem., 64: 2101 (1992).
Membrane Extraction for Sample Preparation / 87 20. 21. 22. 23. 24.
25. 26. 27.
28.
29. 30. 31. 32. 33.
34. 35. 36.
˚. E. Thordarson, S. Palmarsdottir, L. Mathiasson, and J. A ¨ Jonsson, Anal. Chem., 68: 2559 (1996). ˚ . Jo¨nsson, S. Palmarsdottir, E. Thordarson, L.-E. Edholm, J. A and L. Mathiasson, Anal. Chem., 69: 1732 (1997). ˚ . Jo¨nsson and L. Mathiasson, Trends Anal. Chem., 11: 106 J. A (1992). ˚ . Jo¨nsson, L. Mathiasson, and B. Lindega˚nrd, H. Bjo¨rk, J. A A.-M. Olsson, Anal. Chem., 66: 4490 (1994). R. A. Bartsch and J. D. Way, Chemical Separations with Liquid Membranes, ACS Symp. Ser. 642, American Chemical Society, Washington, DC, 1996. A. M. Sastre, A. Kumar, J. P. Shukla, and R. K. Singh, Sep. Purific. Meth., 27: 213 (1998). ˚ . Jo¨nsson and L. Mathiasson, Trends Anal. Chem., 18: 325 J. A (1999). ˚ . Jo¨nsson and L. Mathiasson, in S. P. Parker (Ed.), J. A McGraw-Hill 1998 Yearbook of Science and Technology, McGraw-Hill, New York, 1997, p. 222. H. Lingeman, in E. Reid, H. M. Hill, and I. D. Wilson (Eds.), Drug Development Assay Approaches Including Molecular Imprinting and Biomarkers, Vol. 25, The Royal Society of Chemistry, Cambridge, UK, 1998. ˚ . Jo¨nsson, Anal. Chem., E. Thordarson, J. Emne´us, and J. A 75: 5280 (2000). ˚ . Jo¨nsson, P. Lo¨vkvist, G. Audunsson, and G. Nilve´, Anal. J. A Chim. Acta, 227: 9 (1993). ˚ . Jo¨nsson, Anal. Chim. L. Chimuka, L. Mathiasson, and J. A Acta, 416: 77 (2000). L. Chimuka, N. Megersa, J. Norberg, L. Mathiasson, and J. ˚ . Jo¨nsson, Anal. Chem., 70: 3906 (1998). A M. Valca´rcel and M. D. Luque de Castro, Non-chromatographic Continuous Separation Techniques, The Royal Society of Chemistry, Cambridge, UK, 1991. P. L. Morabito and R. G. Melcher, Process Control Qual., 3: 35 (1992). G. Matz, G. Kibelka, J. Dahl, and F. Lenneman, J. Chromatogr. A., 830: 365 (1999). S. Pedersen-Bjergaard and K. E. Rasmussen, Anal. Chem., 71: 2650 (1999).
88 / Jo¨nsson and Mathiasson 37. 38. 39. 40. 41. 42.
43. 44. 45. 46. 47. 48.
49. 50.
51. 52.
53. 54. 55.
˚ . Jo¨nsson, Analyst, 123: 225 (1998). N. Megersa and J. A ˚ . Jo¨nsson, Int. J. Environ. L. Chimuka, M. M. Nindi, and J. A Anal. Chem., 68: 429 (1997). Y. Sahlestro¨m and B. Karlberg, Anal. Chim. Acta, 179: 315 (1986). R. G. Melcher, Anal. Chim. Acta, 214: 299 (1988). K. Ndung’u, N.-K. Djane, F. Malcus, and L. Mathiasson, Analyst, 124: 1367 (1999). R. Carabias Martı´nez, E. Rodrı´guez Gonzalo, M. P. Santiago Toribio, and J. Herna´ndez Me´ndez, Anal. Chim. Acta, 321: 147 (1996). ˚ . Jo¨nsson, ChromatoM. Knutsson, L. Mathiasson, and J. A graphia, 42: 165 (1996). ˚ . Jo¨nsson, J. Chromatogr. A., G. Nilve´, M. Knutsson, and J. A 668: 75 (1994). ˚ . Jo¨nsson, and L. Mathiasson, T. Miliotis, M. Knutsson, J. A Int. J. Environ. Anal. Chem., 64: 35 (1996). ˚ . Jo¨nsson, J. Chromatogr., 655: L. Gro¨nberg, Y. Shen, and J. A 207 (1993). ˚ . Jo¨nsson, J. ChroY. Shen, V. Obuseng, L. Gro¨nberg, and J. A matogr. A., 725: 189 (1996). M. M. Delgado Zamarreno, A. Sa´nchez Pe´rez, M. Bustamante Rangel, and J. Herna´ndez Me´ndez, Anal. Chim. Acta, 386: 99 (1999). E. Ferna´ndez Laespada, J. L. Pe´rez Pavo´n, and B. Moreno Cordero, J. Chromatogr. A., 852: 395 (1999). R. Carabias Martı´nez, E. Rodrı´guez Gonzalo, E. Herna´ndez Ferna´ndez, and J. Herna´ndez Me´ndez, Anal. Chim. Acta, 304: 323 (1995). ˚ . Jo¨nsson, J. Microcol. Sep., Y. Shen, L. Mathiasson, and J. A 10: 107 (1998). ˚ . Jo¨nsson, M. Andersson, C. Melander, J. Norberg, E. J. A Thordarson, and L. Mathiasson, J. Chromatogr. A., 870: 151 (2000). ˚ . Jo¨nsson, J. Norberg, E. Thordarson, L. Mathiasson, and J. A J. Chromatogr. A., 869: 523 (2000). S. Mitra, L. Zhang, N. Zhu, and X. Guo, J. Microcol. Sep., 8: 21 (1996). B. Hauser and P. Popp, J. High Resol. Chromatogr., 22: 205 (1999).
Membrane Extraction for Sample Preparation / 89 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74.
75. 76. 77.
G. A. Audunsson, Anal. Chem., 60: 1340 (1988). ˚ . Jo¨nsson, and L. Mathiasson, J. ChroB. Lindega˚rd, J. A matogr., 573: 191 (1992). ˚ . Jo¨nsson, ChromatoL. Gro¨nberg, P. Lo¨vkvist, and J. A graphia, 33: 77 (1992). J. J. Vreuls, R. T. Ghijsen, G. de Jong, and U. A. T. Brinkman, J. Chromatogr., 625: 237 (1992). E. C. Goosens, D. de Jong, G. de Jong, and U. A. T. Brinkman, Chromatographia, 47: 313 (1998). K. Grob and J. J.-M. Stoll, J. HRC & CC, 9: 518 (1986). J. Norberg and E. Thordarson, Analyst, 125: 673 (2000). R. Kuldvee and M. Kaljurand, Crit. Rev. Anal. Chem., 29: 29 (1999). J. R. Veraart, H. Lingeman, and U. A. T. Brinkman, J. Chromatogr. A, 856: 483 (1998). L. Bao and P. K. Dasgupta, Anal. Chem., 64: 991 (1992). ˚ . Jo¨nsson, Anal. Chim. Acta, Y. Shen, L. Gro¨nberg, and J. A 292: 31 (1994). ˚ . Jo¨nsson, J. Chromatogr. A, N. Megersa, T. Solomon, and J.A 830: 203 (1999). ˚ . Jo¨nsson, and L.-E. EdS. Palmarsdottir, L. Mathiasson, J.A holm, J. Capill. Electropher, 3: 255 (1996). ˚ . Jo¨nsson, J. Chromatogr, G. Nilve´, G. Audunsson, and J. A 471: 151 (1989). L. Mathiasson, G. Nilve´, and B. Ule´n, Int. J. Environ. Anal. Chem., 45: 117 (1991). ˚ . Jo¨nsson, J. M. Knutsson, G. Nilve´, L. Mathiasson, and J. A Agric. Food Chem., 40: 2413 (1992). G. Nilve´ and R. Stebbins, Chromatographia, 32: 269 (1991). ˚ . Jo¨nsson, and P. M. Knutsson, J. Lundh, L. Mathiasson, J. A Sundin, Anal. Lett., 29: 1619 (1996). ˚ . Jo¨nsson, L. Mathiasson, E. BureJ. Norberg, J. Emne´us, J. A stedt, M. Knutsson, and G. Marko-Varga, J. Chromatogr. B, 701: 39 (1997). L. Mathiasson, M. Knutsson, G. Bremle, and L. Ma˚rtensson, Swedish J. Agric. Res., 21: 147 (1991). ˚ . Jo¨nsson, and G. Tyler, Soil Biol. BioY. Shen, L. Stro¨m, J. A chem., 28: 1163 (1996). ˚ . Jo¨nsson, L. Ma˚rtensson, M. Magnusson, Y. Shen, and J. A Agric., Ecosyst. Environ., 75: 101 (1999).
90 / Jo¨nsson and Mathiasson 78. 79. 80. 81.
82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97.
˚ . Jo¨nsson, Chemosphere, 24: L. Gro¨nberg, P. Lo¨vkvist, and J. A 1533 (1992). J. Trocewicz, J. Chromatogr. A, 725: 121 (1996). ˚ . Zander, and J. A ˚ . Jo¨nsson, Chromatographia, J. Norberg, A 46: 483 (1997). S. Palmarsdottir, B. Lindega˚rd, P. Deininger, L.-E. Edholm, ˚ . Jo¨nsson, J. Capill. Electrophor, 2: L. Mathiasson, and J. A 185 (1995). ˚ . Jo¨nsson, and L.-E. EdS. Palmarsdottir, L. Mathiasson, J. A holm, J. Chromatogr. B, 688: 127 (1997). J. Trocewicz, Z. Suprynowicz, and J. Markowicz, J. Chromatogr. B, 685: 129 (1996). M. Bryjak, P. Wieczorek, P. Kafarski, and B. Lejczak, J. Membr. Sci., 37: 287 (1988). ˚ . Jo¨nsson, and L. Mathiasson, Anal. Chim. P. Wieczorek, J. A Acta, 337: 183 (1997). ˚ . Jo¨nsson, and L. Mathiasson, Anal. Chim. P. Wieczorek, J. A Acta, 346: 191 (1997). ˚ . Jo¨nsson, P. Dzygiel, P. Wieczorek, L. Mathiasson, and J. A Anal. Lett., 31: 1261 (1998). ˚ . Jo¨nsson, M. Milewska, and P. P. Dzygiel, P. Wieczorek, J. A Kafarski, Tetrahedron, 55: 9923 (1999). ˚ . Jo¨nsson, and M. Valiente, Anal. J. A. Calzado, C. Palet, J. A Chim. Acta, 417: 159 (2000). ˚ . Jo¨nsson, and L. M. Papantoni, N.-K. Djane, K. Ndung’u, J. A Mathiasson, Analyst, 120: 1471 (1995). N.-K. Djane, K. Ndung’u, F. Malcus, G. Johansson, and L. Mathiasson, Fresenius J. Anal. Chem., 358: 822 (1997). K. Ndung’u, N.-K. Djane, and L. Mathiasson, J. Chromatogr. A, 826: 103 (1998). K. Ndung’u and L. Mathiasson, Anal. Chim. Acta, 404: 319 (2000). T. Garcı´a Sanchez, J. L. Pe´rez Pavo´n, and B. Moreno Cordero, J. Chromatogr. A, 766: 61 (1997). E. Rodrı´gues Gonzalo, J. L. Pe´rez Pavo´n, J. Ruzicka, G. D. Christian, and D. C. Olson, Anal. Chim. Acta, 259: 37 (1992). Q. Chang and M. E. Meyerhoff, Anal. Chim. Acta, 186: 81 (1986). R. G. Melcher and P. L. Morabito, Anal. Chem., 62: 2183 (1990).
Membrane Extraction for Sample Preparation / 91 98. 99. 100. 101. 102.
S. Mitra, N. Zhu, X. Zhang, and B. Kebbekus, J. Chromatogr. A, 736: 165 (1996). M. J. Yang, S. Harms, Y. Z. Luo, and J. Pawliszyn, Anal. Chem., 66: 1339 (1994). R. Kostiainen, T. Kotiaho, R. A. Ketola, and V. Virkki, Chromatographia, 41: 34 (1995). N.-K. Djane, I. A. Bergdahl, K. Ndung’u, A. Schu¨tz, G. Johansson, and L. Mathiasson, Analyst, 122: 1073 (1997). ˚ . Jo¨nsson, J. M. Knutsson, G. Nilve´, L. Mathiasson, and J. A Chromatogr. A, 754: 197 (1996).
3 Design of Rapid Gradient Methods for the Analysis of Combinatorial Chemistry Libraries and the Preparation of Pure Compounds Uwe D. Neue, Judy L. Carmody, Yung-Fong Cheng, Ziling Lu, Charles H. Phoebe, and Thomas E. Wheat Waters Corporation, Milford, Massachusetts
I. INTRODUCTION II. THEORY A. Fundamentals B. Gradient Resolution as a Function of Column Length and Particle Size III. PRACTICE OF FAST CHROMATOGRAPHY A. General Considerations B. Fast Analytical Separations C. Fast Preparative Separations IV. CONCLUSION SYMBOLS REFERENCES
93 95 95 108 114 115 122 124 133 134 135
I. INTRODUCTION Automated synthesis has emerged as the dominant tool in the search for new lead compounds in the pharmaceutical industry [1]. The 93
94 / Neue et al. speed with which new entities are synthesized challenges the techniques used to analyze the products of the synthesis. Fast structure identification tools such as mass spectrometry (MS) can rapidly identify the success of a synthesis, but do not provide sufficient information on the purity of a new entity. On the other hand, only a rapid separation technique such as high-performance liquid chromatography (HPLC) can be used to characterize the purity and homogeneity of a sample. Classical high-resolution HPLC gradient methods are rather slow techniques: 15- to 30-min run times are not unusual. However, such a slow method is not compatible with a high-speed synthesis scheme, which may produce 1000 new chemical entities per day. This has forced a rethinking of the approach [2], and the throughput of HPLC methods has increased substantially. Nevertheless, a thorough analysis of the possibilities and constraints of fast HPLC gradients is still lacking. The basis of our understanding of gradient chromatography, especially reversed-phase gradient chromatography, was laid out by Snyder and co-workers in a series of papers in the late 1970s (Refs. 3 and 4 and references therein). Particularly the appendixes of Ref. 3 can be recommended to those who are interested in the theoretical background of gradient elution. An understanding of the principles of gradient elution has been applied by El Fallah [5] to the rapid development of isocratic methods. However, all of the early work on gradient elution focused on improvements in the separation conditions. Only recently has the emphasis shifted to the speed of a separation, due to the high demands of combinatorial chemistry on fast generic analytical methods. Preparative chromatography [6] is commonly carried out in isocratic mode. Preparative chromatography under gradient conditions has received only limited attention in the literature [7–10]. Under gradient conditions, the elution strength and the mass loading are varied simultaneously, which can result in complex peak shapes, especially if the sample contains significant amounts of more than one component. A few cases have been treated thoroughly in Ref. 7, which is recommended as a reference for those dealing with fast preparative gradients. In this chapter, we will first cover the principles of reversedphase gradient separations, and how truly optimized generic meth-
Design of Rapid Gradient Methods
/ 95
ods can be selected depending on the column and the desired run times. This discussion will include some results of computer simulations that cover the common ranges of column lengths, particle sizes, and running conditions. This is accompanied by a discussion of instrument parameters, especially the gradient delay volume and extracolumn effects. Application examples demonstrate the principles. Finally, consideration is given to the scaling of the gradient separation to larger columns for preparative purposes. Optimized sample loading schemes are demonstrated that result in simplified and automatable preparative gradient elution.
II. THEORY A. Fundamentals Gradient methods are typically developed and optimized using purely empirical tools. However, the basic principle of a gradient separation is as amenable to a theoretical analysis as is the principle of isocratic separations. In the following section we will outline the rules for the optimization of generic gradient separations, without regard of the analytes to be separated. The net result of this discussion is the selection of truly optimized chromatographic parameters, such as linear velocity and gradient slope, depending on the choice of column length and particle size and under the constraint of a maximum analysis time. The principles for reversed-phase chromatography outlined here can be applied without difficulty to other chromatographic techniques such as normal-phase chromatography or ion exchange. The most useful tool for the assessment of the separation power of a chromatographic technique is the resolution equation. In addition, most chromatographers are familiar and comfortable with the use of the resolution equation in isocratic chromatography. Therefore, we will use the resolution between adjacent peaks as our primary measure of the capability of a gradient. For a correct application of the resolution equation to the case of gradient chromatography, we have to step back to the basic definition of resolution, Rs: Rs ⫽
∆t R w
(1)
96 / Neue et al. Here ∆t R is the retention time difference between adjacent peaks, and w is the width of a peak, in time units. In gradient chromatography, one can assume without difficulty that the width of the two neighboring peaks is identical. The resolution equation can also be written in a form which uses the retention factor: Rs ⫽
k g 2 –k g 1 wk
(2)
Here k g is the retention factor of an analyte under gradient conditions, and w k is the peak width in the same nondimensional form as the retention factor, i.e., it is normalized by the retention time of an unretained peak t 0: wk ⫽
w t0
(3)
The peak width depends on the column plate count N and the retention factor at the point of elution k e , that is, at the column exit: wk ⫽
4 √N
⋅ (k e ⫹ 1)
(4)
The retention factor at the point of elution, k e , is the retention factor that the analyte would have under isocratic elution conditions with the solvent composition at the column exit at the point of elution. With Eqs. (2) and (4), the resolution in gradient chromatography can be expressed as a function of the column plate count and parameters that depend on the gradient: Rs ⫽
√N k g 2 ⫺ kg 1 ⋅ 4 ke ⫹ 1
(5)
This is the fundamental resolution equation for the subsequent discussion of gradient chromatography. The retention factor under gradient conditions k g and the actual retention factor at the point of elution k e can be calculated for different types of gradient chromatography, from reversed-phase chromatography to normal-phase chromatography to ion-exchange. Thus an assessment of resolution in any kind of gradient chromatography is possible. For a linear reversed-phase gradient, the dependence of k g and k e on the parameters of the gradient has been established [3,4]. The
Design of Rapid Gradient Methods
/ 97
underlying assumption is a linear relationship between the logarithm of the isocratic retention factor and the percent organic solvent in the mobile phase. The following equation expresses the relationship between the retention factor under gradient conditions k g and the gradient and sample parameters: kg ⫽
冢
冣
t0 1 ⋅ ln B ⋅ ∆c ⋅ ⋅ k 0 ⫹ 1 B ⋅ ∆c ⋅ (t 0 /t g ) tg
(6)
Here B is the slope of the relationship of ln(k) to the solvent composition, ∆c is the change in organic composition over the gradient, t g is the gradient duration, and k 0 is the retention factor at the start of the gradient. The parameter G ⫽ B ⋅ ∆c ⋅
t0 tg
(7)
is a dimensionless descriptor of the gradient steepness [11], which can be used for an abbreviated description of the properties of a gradient. However, we will use the complete notation in the following analysis, since we ultimately would like to link gradient parameters with column efficiency parameters. The next equation shows the relationship between the actual retention factor at the point of elution k e and the gradient conditions: ke ⫽
k0 B ⋅ ∆c ⋅ (t 0 /t g ) ⋅ k0 ⫹ 1
(8)
Now note that our interest is in the average change of resolution as a function of the gradient conditions, not in the resolution of a specific pair of compounds. Therefore, we will assume that the values in the equation take on typical values for the type of compounds of interest. For example, the slope B of the relationship of ln(k) to the solvent composition can be assumed to take typical values for small molecules, or for larger-molecular-weight compounds such as peptides or proteins. If we assume a typical value of B for all compounds of interests, we also imply that the compound pair of interest has a constant relative retention, independent of the solvent composition. Another interpretation of this is the notion of the resolution between two average compounds. This is completely in line with the purpose of this derivation, which is intended to demonstrate the rules for the optimization of generic gradient separations.
98 / Neue et al. Writing the resolution equation with this thought in mind, we arrive at: [B ⋅ ∆c ⋅ (t /t ) ⋅ k ⫹ 1] 冦 [B ⋅ ∆c ⋅ (t /t ) ⋅ k ⫹ 1] 冧 √N ⋅ Rs ⫽ 4 冦[B[B⋅ ∆c⋅ ∆c⋅ (t⋅ (t/t )/t⋅ )k⋅ k⫹] 1]冧 ⫹ B ⋅ ∆c (t /t ) ln
0
g
0
g
2,0
0
g
1,0
0
0
g
(9)
0
g
0
Finally, if B ∗ c ∗ t 0 ∗ k 0 is large compared to 1, the equation simplifies to: Rs ⫽
√N 1 ⋅ ln(α) ⋅ 4 B ⋅ ∆c ⋅ (t 0 /t g ) ⫹ 1
(10)
where α is the ratio of the retention factors. With the abbreviation for the gradient parameters of Eq. (7), this equation becomes: Rs ⫽
1 √N ⋅ ln(α) ⋅ 4 G⫹1
(11)
Equation (11) is a close analog of the resolution equation for isocratic chromatography. The first part of the equation contains the column efficiency, the second part the relative retention α, and the third part contains the parameters of the gradient, i.e., the dimensionless gradient steepness parameter. In the equivalent resolution equation for isocratic chromatography, the third part of the equation contains the retention factor. In the subsequent discussion of the effect of gradient parameters on the performance of a gradient separation, we will use resolution as shown in Eq. (10) as our measure. Nevertheless, other parameters could be used for the same purpose. Fundamentally, the peak capacity of a gradient is the simplest measurement of the resolving power under gradient conditions. All one needs to do is measure the typical peak width of a typical analyte, or the average peak width of an average analyte, and divide the gradient duration by this number. While it is highly practical, we will not use this parameter in most of the following discussions. Resolution is a concept that is closer to the heart of the chromatographer. Therefore we will use resolution as the parameter for the discussion of optimal gradient conditions.
Design of Rapid Gradient Methods
/ 99
Nevertheless, due to the simplicity of the measurement, we will also derive the peak capacity in a gradient, using the same thought process as outlined above. The peak capacity P is the number of peaks with a width w that can be separated in a given gradient time t g: P⫽1⫹
tg w
(12)
This approach assumes that the peak width in the gradient is constant throughout the gradient. As above, the peak width depends on the gradient as described by Eq. (4), and the retention factor at the point of elution is represented by Eq. (8). As above, we assume that the parameter B, the slope of the relationship between ln(k) and the solvent composition, does not depend on the analyte. Thus we assume that it takes on a typical value for the type of analytes under consideration, for example, small molecules, peptides, or proteins. If we substitute Eq. (4) and then Eq. (8) into Eq. (12), we obtain after some rearrangement: P⫽1⫹
B ⋅ ∆c ⋅ (t0 /tg) ⋅ k0 ⫹ 1 √N tg ⋅ ⋅ 4 t0 k0 ⫹ B ⋅ ∆c ⋅ (t0 /tg) ⋅ k0 ⫹ 1
(13)
As in the previous calculation, for large k 0 this equation simplifies to: P⫽1⫹
√N B ⋅ ∆c ⋅ 4 B ⋅ ∆c ⋅ (t0 /tg) ⫹ 1
(14)
This is a fundamental description of the peak capacity of a reversedphase gradient for a given type of analytes. For slow gradients, i.e., with a large ratio of t g /t 0 , the peak capacity approaches a constant value, determined by the column plate count N, the typical slope of the relationship between the logarithm of the retention factor and the solvent composition B, and the difference in solvent composition from the beginning to the end of the gradient: P⫽1⫹
√N ⋅ B ⋅ ∆c 4
(15)
100
/ Neue et al.
In addition, for a gradient from 100% aqueous to 100% organic, this equation further simplifies to: P⫽1⫹
√N ⋅B 4
(16)
This simple equation describes the maximum peak capacity of a reversed-phase gradient from water to an organic solvent. Most chromatographers are more familiar with the concept of the resolution between peaks than with the concept of peak capacity. Therefore, we will use resolution as the parameter in the subsequent discussion. Nevertheless, it should be pointed out that a deliberation of the peak capacity of a chromatogram and the resolution between a peak pair of fixed resolution are completely analogous. We have calculated the results of Eq. (9) for a range of retention factors k 0 at the beginning of a gradient as a function of the normalized gradient slope, ∆c ∗ t 0 /t g. A gradient slope of 1 means that the solvent composition changes from 0% to 100% organic in one column volume, and a slope of 0.01 describes the same change in solvent composition over 100 column volumes. The value for the compounddependent parameter B was assumed to be 9 [5], which might be typical for the types of analytes in small-molecule combinatorial chemistry. The results are plotted in Fig. 1. For a gradient slope covering 4 orders of magnitude and the retention factor at the beginning of the gradient covering a range from 10 to 100,000, the dependence of the resolution on the gradient slope follows the same pattern. In addition, the pattern can be described well by the limiting Eq. (10). Consequently, one can use Eq. (10) to demonstrate the effects of gradient parameters on the resolving power of the gradient. In addition, one can explore the combined effects of plate count and gradient parameters on gradient resolution. Figure 1 deserves a few additional comments. For steep gradients, the resolution increases in direct proportion with a decrease in the gradient slope. This increase continues until the gradient reaches approximately 30 column volumes for a gradient from 0% to 100% organic solvent. A gradient flatter than about 100 column volumes for the same change in composition (slope ⫽ 0.01) barely increases resolution. Therefore, this figure gives us the first guidelines for the design of an optimized gradient procedure for a general analytical problem such as the one posed by the analysis of combinatorial libraries.
Design of Rapid Gradient Methods / 101
Fig. 1 Plot of gradient resolution versus gradient slope for a range of retention factors at the beginning of the gradient, ranging from 10 to 100,000. The curves follow the same pattern and can be approximated by a single curve.
Since Eq. (10) will be central to our subsequent discussion of resolution in gradient chromatography, it is worthwhile to explore it a little further. We will use a complex sample mixture to illustrate the points. The sample is a mixture of peptides obtained from the tryptic digestion of bovine serum albumin. Due to the larger molecular weight of peptides, the measures of separation power versus gradient slope are shifted toward longer run times, but the overall pattern remains the same as for small molecules. Let us look first at the chromatograms in Fig. 2. The same mixture of peptides was injected, and the gradient duration was varied from 15 min to 25, 50, 75, and 150 min. We examined the chromatograms and found 78 peaks for the 15-min gradient, and 98 peaks, 114 peaks, 137 peaks, and 162 peaks, respectively, for the longer gradients. The longer the gradient duration is, the more peaks we find, and the better is the resolution.
102
/ Neue et al.
Fig. 2 Comparison of gradient resolution as a function of gradient time for a complex mixture of compounds. As the gradient duration increases, the number of peaks recognized in this mixture increases.
Design of Rapid Gradient Methods / 103 In addition, we can plot the number of peaks versus the gradient slope in the same way as we had plotted resolution in Fig. 1. As we can see in Fig. 3, the resulting pattern follows the same pattern as the plot in Fig. 1. As the gradient slope becomes flatter, the number of peaks increases with a linear dependence on the logarithm of the gradient slope. Then, as the gradient slope becomes still flatter, one still observes an increase in the number of peaks, but the increase is smaller than in the previous cases. The number of peaks observed follows the expected pattern of the resolution equation (10) or the peak capacity equation (14). However, we also need to realize that as we vary the gradient slope, the actual elution pattern is changing with the gradient slope. A change in the gradient slope affects not only the overall resolution power of the gradient, but also the elution sequence. This is not the case in the next example. In this case, we change the column length, and we vary the gradient duration in proportion to the column length. Under these circumstances, the gradient slope (∆c ∗ t 0 /t g as defined above) remains constant. Therefore the elution pattern remains the same, and the gain in resolution with column length is due exclusively to the increase in plate count. This is shown in Fig. 4. The same composition change was executed over 5 min for
Fig. 3 Plot of the number of peaks shown in Fig. 2 versus the gradient slope. The pattern follows the expectation as shown in Fig. 1.
104
/ Neue et al.
Fig. 4 Resolution pattern obtained when the gradient volume is scaled in proportion to the column volume. Under these circumstances, the elution pattern does not change, but the resolution improves with increasing column length. This is especially apparent when examining the resolution pattern marked by the boxes.
the 5-cm column, over 15 min for the 15-cm column, and over 25 min for the 25-cm column. Two sections in the chromatograms are marked to demonstrate clearly the improvement in the resolution pattern due to the increase in the plate count. The last two examples illustrate elements of column performance using a complex sample mixture. The identical criteria apply to the simple sample mixtures of combinatorial chemistry, where one desires to separate the desired compound from raw materials and byproducts. Thus we can use the same tools demonstrated until now to analyze the performance of columns and gradient conditions for the rapid analyses required in combinatorial chemistry. However, the question posed by the needs of combinatorial chemistry
Design of Rapid Gradient Methods / 105 extend beyond what has been discussed so far. One would like to know what are the chromatographic conditions that result in the best resolving power if the constraint is the total run time of the analysis. In the following, we will address this question first for a single column. In the subsequent paragraph, we will then examine different columns and particle sizes. The column selected for the starting discussion is a column of 5 cm length packed with 5-µm particles. We can vary freely the flow rate (or linear velocity) and the gradient duration, and we are interested in the resolving power of the column as a function of both parameters. For the complete analysis, Eq. (10) needs to be expanded to incorporate the dependence of the column plate count on the gradient running conditions: N⫽
L L ⫽ H a ⋅ dp ⫹ b ⋅ (DM /u) ⫹ c ⋅ (d 2p /DM ) ⋅ u
(17)
where L is the column length, H is the height equivalent to a theoretical plate, d p is the particle size, and a, b,and c are constants. D M is the diffusion coefficient of the sample, and an appropriate typical value for the samples of interest needs to be selected; u is the linear velocity, which can be expressed as L/t 0. We can incorporate this definition of the plate count into the resolution equation: Rs ⫽
1 4 ⋅ √a (dp /L) ⫹ b(DM ⋅ t0 /L 2 ) ⫹ c(d 2p /DM ⋅ t0 ) 1 ⋅ ln(α) ⋅ B ⋅ ∆c (t0 /tg ) ⫹ 1
(18)
This equation contains all the parameters necessary for a complete analysis of the optimal gradient conditions as a function of the column choice. The gradient parameter B and the diffusion coefficient D M depend on the type of analyte. Typical values can be selected for small molecules, peptides, or even large-molecular-weight analytes. The column length L, the particle size d p , the column dead time t 0 , and the gradient slope (or simply the gradient time t g) are all parameters whose influence on the resolving power of the gradient separation we would like to explore. In the discussions in the next paragraph, we will examine the influence of particle size and column length in greater detail. As a tool, we will use a three-dimensional plot of resolution versus column
106
/ Neue et al.
dead time and gradient duration. Figure 5 contains such a plot for the 5-cm, 5-µm column selected as our first example. The column dead time is depicted in seconds on the x axis, the gradient duration, also in seconds is the y axis, and the resolution is shown in arbitrary units on the z axis. As expected, the resolution increases with increasing gradient time. For every gradient time, the resolution exhibits a maximum as a function of the flow rate. For very slow gradients, this maximum coincides with the minimum of the plate height-versus-velocity curve. For short gradients, the maximum resolution occurs at a higher flow rate. This is due to the interplay between the slope of the gradient, which decreases as t 0 decreases at constant gradient time t g , and the loss in plate count at higher velocities. Near maximum resolution is reached with a gradient of nearly 3 h (10,800 s), but quite good resolution is obtained already at a gradient run time of 10 min (600 s), at a column dead time of about 25 s. For extremely short gradients, with a gradient duration
Fig. 5 Gradient resolution as a function of gradient duration and column dead time for a 5-cm, 5-µm column. This graph is very suitable to examine column capabilities under gradient conditions.
Design of Rapid Gradient Methods / 107 of 2 min or less, the best results are obtained with a column dead time of only about 7 s. Of course, the overall resolving power of such a short gradient is significantly inferior to that of the longer gradients. A few additional comments are necessary. On any diagonal line from the front to the back of the graph the ratio of the gradient duration to the column dead time remains constant. A constant gradient profile means that the elution pattern of the separation remains constant as well. This is the single-column equivalent of Fig. 4: only the height equivalent to a theoretical plate (HETP) and the column plate count change under these circumstances. Of course, the resolution increases as the gradient duration increases, until the minimum of the plate height-versus-velocity curve is reached. An example of this is shown in Fig. 6.
Fig. 6 Comparison of gradient chromatograms for a simple mixture of small-molecular-weight analytes as a function of the gradient time, but under conditions where the gradient volume remains constant. The elution pattern does not change, but the resolution increases as the gradient duration increases.
108
B.
/ Neue et al.
Gradient Resolution as a Function of Column Length and Particle Size
In this section we will further pursue the thought process started at the end of the last paragraph. We will explore the resolving capabilities of columns of different length and of different particle sizes. This discussion will include the constraints on speed imposed by the column back-pressure. Finally, we will explore the simultaneous scaling of column length and particle size. Experience from isocratic chromatography teaches us that under these circumstances the maximum resolving power as well as the fastest speed of analysis remain constant [12]. Let us first explore the effect of different column length on gradient resolution, keeping the particle size constant. Figure 7 shows a diagram of resolution versus column dead time and gradient duration for a 15-cm, 5-µm column. For simplicity of visual comparison, the length of the resolution axis and the gradient duration axis re-
Fig. 7 Gradient resolution as a function of gradient duration and column dead time for a 15-cm, 5-µm column. For comparison with Fig. 5.
Design of Rapid Gradient Methods / 109 main the same as in Fig. 5, which shows the same plot for a 5-cm, 5-µm column. However, the length and the scale of the column dead time axis are shifted. This is due to the fact that we imposed an upper pressure limit of about 40 MPa on the column operating conditions in both graphs. The consequence of this is a ninefold shift in the shortest column dead time possible when the column length at constant particle diameter changes by a factor of 3. Due to the pressure constraint, the 15-cm column also does not reach the maximum resolution at very short run times. Only for gradient run times over 5 min (300 s) can a maximum in the resolution at constant gradient duration be reached. On the other hand, for very slow gradients the 15-cm column reaches a higher resolution than the 15-cm column. Under these conditions, the performance of both columns approaches what we know about isocratic chromatography. The difference in the performance capability of the 5-cm and the 15-cm columns is shown in Fig. 8. Here we plot the maximum resolution obtained at every gradient duration versus the gradient duration. Such a plot allows a direct comparison of the maximum capability of each column for a given expenditure in analysis time. For very fast gradients, the resolution obtained with the short column is similar to the resolution obtained with the longer column. The advantage of the higher resolution obtainable with the longer column becomes
Fig. 8 Comparison of the optimal gradient resolution for a 5-cm and a 15cm column packed with 5-µm particles.
110
/ Neue et al.
significant only for slower gradients with a gradient duration longer than about 1000 s or 15 min. On the other hand, the column backpressure of the shorter column is lower, which improves the column stability. Consequently, the shorter column is a better choice for very fast separations. Let us next explore the influence of particle size for columns of equal length. The first parameter to consider is the column backpressure. At constant column length, the back-pressure increases inversely proportionally to the square of the particle size. This means that a column with a smaller particle size will reach the backpressure limit of the instrumentation at a lower linear velocity than a column with a larger particle size. However, the column with the smaller particle size is expected to exhibit the higher plate count. This improvement in plate count should also be discernible in gradient chromatography. A comparison of 4.6 mm ⫻ 50 mm columns packed with two different particle sizes is shown in Fig. 9. The upper graph shows the performance of the 5-µm column, the lower graph the performance of the 3.5-µm column. Only the data for steep gradients are shown, with a maximum gradient duration of 1200 s or 20 min. As expected, the maximum resolution is obtained with the slowest gradient, at a column dead time of about 38 s for the 5-µm column and about 27 s for the 3.5-µm column. The resolution maximum obtained with the 5-µm column is slightly less than 80% of the resolution maximum of the 3.5-µm column. We have imposed a pressure limit of about 25 MPa for both graphs. This means that the 5-µm column can be used at a shorter column dead time than the 3.5-µm column, as indicated in the last paragraph. However, even for the 3.5-µm column, the performance maximum at a given gradient time occurs at higher pressures only for gradient times below 150 s. On the other hand, the maximum resolution for a 75-s gradient is obtained with the 5µm column well within the pressure limits and is identical to the performance of the 3.5-µm column at the pressure limit. Consequently, for target analysis times of about 5 min, the 3.5-µm column is preferred over the 5-µm column, but for very short analysis times under 2.5 min the pressure limitation of the HPLC instrument needs to be taken into account. However, there is another approach to shorten analysis time with less impact of the pressure limitation of the instrument: we can change the column length and the particle size simultaneously.
Design of Rapid Gradient Methods / 111
Fig. 9 Comparison of the resolving power for fast gradients of two columns of equal length but different particle sizes. Column length 5 cm, particle sizes 5 µm (upper graph) and 3.5 µm (lower graph).
112
/ Neue et al.
Under these circumstances, the pressure required to reach a particular t 0 is the same for all columns [12]. Conversely, the shortest analysis time achievable at a given pressure limitation is independent of the choice of the particle size. This is shown in Fig. 10. We have plotted the maximum performance at each gradient duration for a 10-cm, 10-µm column, a 5-cm, 5-µm column, a 3-cm, 3-µm column, a 2-cm, 2-µm column, and a 1-cm, 1-µm column. The shortest run time imposed by the pressure limit is 75 s. If one waits long enough, the maximum resolving power of all columns reaches the same value. However, for realistic gradient run times the shorter columns packed with the smaller particles always result in a higher performance than the longer columns with the larger particles. For example, for a 100-s gradient the resolution values are 3.0 for the 10-cm, 10-µm column, 8.0 for the 3-cm, 3-µm column, and 13.2 for the
Fig. 10 Comparison of the maximum resolution capabilities of several columns with the same ratio of column length to particle diameter.
Fig. 11 Resolution as a function of column length for a 3.5-min gradient and particle sizes of 5 µm (top) and 3 µm (bottom).
114
/ Neue et al.
1-cm, 1-µm column, a nearly 4.5-fold improvement. In addition, the resolution achievable with the 1-cm, 1-µm column under realistic gradient run times approaches the maximum resolution attainable under any conditions with this column length/particle size ratio. Of course, the practical use of such a column configuration requires a minimization of extracolumn effects, from extracolumn band spreading to detector parameters such as time constant, sampling rates, and data bunching. On the other hand, good fast results can be obtained with 2-cm, 2-µm, or 3-cm, 3-µm columns with little modification to existing instrumentation. As a final subject of this discussion, we can ask ourselves what the best column configuration is to achieve the highest resolution in a particular run time. We will examine this for two particle sizes, 5 µm and 3 µm. In Fig. 11, the resolution is plotted versus t 0 and column length for both particle sizes and a 3.5-min gradient. A pressure constraint of 40 MPa is imposed, which truncates the graphs at short run times and long column lengths. The graphs show maxima for t 0 as a function of column length, and an improvement in performance with column length limited by the column back-pressure. For the 5-µm column, the best performance is achieved with a 10-cm column and a column dead time of about 15 s. Better performance is achieved with 3-µm columns, with an optimum length of 5 cm and an optimum t 0 around 10 s. This choice of columns and operating conditions is in agreement with the experience of practitioners (e.g., Ref. 2).
III. PRACTICE OF FAST CHROMATOGRAPHY In the last section we discussed a range of parameters that help us in the selection of the best column for fast gradient separations. The discussions focussed on column length and particle size. Often, the pressure capability of the HPLC instrument posed a limit on the column choices. However, these are not the only parameters to consider. Capabilities of the pumping system such as flow rate range or gradient delay volume, detector capabilities such as sampling rates and time constants, together with system capabilities such as the overall system band spreading, can influence the result of a separation and should be optimized together with the choice of column length, particle size, and gradient parameters.
Design of Rapid Gradient Methods / 115
A. General Considerations Let us first consider the choice of the organic solvent for reversedphase chromatography. Generally, three solvents are commonly used as organic modifiers in reversed-phase chromatography: methanol, acetonitrile, and tetrahydrofuran (THF). Since a specific separation is not required, we should select the best solvent based on generic properties. The most important factor for fast separations is the viscosity of the mobile phase, since both the pressure drop across the column and the peak width are functions of the viscosity. Methanol and THF exhibit maxima of the dependence of viscosity on the composition of their mixtures with water. Acetonitrile exhibits only a very small maximum for a mixture of about 10% acetonitrile in water, as well as a much reduced viscosity relative to mixtures of methanol or THF with water [12]. The reduction in pressure is about 50% when acetonitrile is used instead of methanol, and the average column plate count is about two times higher with acetonitrile than with methanol as organic modifier. Therefore, acetonitrile is the most preferred solvent for fast reversed-phase separations. Let us consider now the effect of particle size on a gradient separation. The results obtained for a comparison of the performance of a 5-cm, 5-µm columns to a 5-cm, 3.5-µm column are shown in Fig. 12. As in Fig. 9, the resolution is plotted on the vertical axis and the gradient time, in minutes, on one of the horizontal axes. On the horizontal axis in the front of the graph, we have plotted the flow rate instead of the column dead time. This is analogous to the theoretical comparison of the same column dimensions in Fig. 9. For comparison to Fig. 9, a column dead time of about 30 s corresponds to a flow rate of about 1 mL/min. The shapes of both curves follow generally the expected pattern. The resolution at each gradient time passes through a maximum at intermediate flow rates and increases with the duration of the gradient. At the slowest gradient time, the 3.5-µm column exhibits a higher resolution than the 5-µm column. At 1 mL/min and a 16-min gradient, the resolution value obtained with the 3.5-µm column was 18.8, while the same value was 16.2 for the 5-µm column. It also needs to be pointed out that the 5-µm column was used up to a flow rate of 8 mL/min, while the pressure limit of this study was reached already at 5 mL/min for the 3.5-µm column. Both columns exhibit a clear performance maximum be-
116
/ Neue et al.
Design of Rapid Gradient Methods / 117 tween 1 and 2 mL/min, well below the pressure limit, in agreement with the expectation as shown in Fig. 9. However, if we compare the resolution pattern of both graphs for short gradients, we see a difference. Theory predicts a shift in the maximum resolution toward higher flow rates as the gradient time is shortened. This is not seen in the experimental results in Fig. 12. The reason for this difference between theory and experimental results is the setting of detector parameters such as sampling rate and detector time constant. If the sampling rate is too low or the detector time constant is too high, the expected shift in optimal performance to higher flow rates with shorter gradient run times will not be observed. Of course, smaller time constants and higher sampling rates increase the baseline noise in the chromatogram. However, under the circumstances of the analysis of samples created in a combinatorial synthesis scheme, we are not forced to maximize detector performance, and a drop in sensitivity by even a substantial factor is without consequences. Faster sampling rates increase the demands on the storage capability of the HPLC system, but this problem can be addressed with a regular maintenance schedule. Generally the sampling rate should be adjusted to yield at least 20 data points for the narrowest peak of interest. For fast gradients, the peak width might be as small as 1 s or less, requiring a sampling rate of 20 data points per second or faster. Also, the detector time constant or the equivalent settings for a digital filter should be of the order of 0.1 s or faster. The requirements on system band spreading stem from the choice of column volume, which in turn may be dictated by other parameters such as the flow rate requirement of the MS detector. Typically, precolumn band spreading can be ignored due to the fact that gradients are used. Therefore, injectors with the capacity of large injection volumina can be used even with small columns. How-
Fig. 12 Measured resolution as a function of flow rate and gradient duration. Columns: 4.6 mm ⫻ 50 mm Symmetry C 18. Top, 5 µm; bottom, 3.5 µm. Samples: methyl 3-amino-thiophenecarboxylate and 4-aminobenzophenone [18]. Temperature: 30°C. Mobile phases: A, 0.1% TFA in water; B, 0.1% TFA in acetonitrile. Gradient from 0% B to 60% B in the indicated time and at the indicated flow rate.
118
/ Neue et al.
ever, the samples in combinatorial chemistry often originate in a reaction carried out in an organic solvent or are redissolved in a general-purpose organic solvent such as dimethyl sulfoxide (DMSO). The injection of a large amount of organic solvent onto a column in 100% water can result in severe peak distortion or even elution of a large fraction of the sample in the column void volume. If the analytes are dissolved in DMSO, a dilution of the sample with the aqueous buffer (5: 1 or even 10 :1) together with a larger injection volume gets around this problem. Good results have been reported with this technique [15]. For preparative applications, one would like to dissolve the maximum amount of sample. Often, DMSO is the best solvent, but an injection of a large volume of DMSO creates peak distortions. In the applications section, we describe a technique that solves this problem. The band spreading of the detector cell is not a problem if columns with an internal diameter of 4.6 mm are used. At a flow rate of 4 mL/min, a peak with a width of 1 s has a peak volume of about 66 µL. A standard 8-µL cell is adequate for this condition. However, the detector flow cell should be considered carefully for 2-mm columns. Since sensitivity usually is not a problem, detector flow cells with a smaller cell volume should be considered. Ideally, the gradient is generated directly in front of the column. However, such a scenario is possible only with older dual-pump HPLC systems. Modern systems use a single pump, with a gradient mixer before the pump heads. Thus there is a delay before the gradient reaches the top of the column. At the end of the gradient, the solvent composition is returned to initial conditions via a step function. This step is slightly smeared out due to the influence of the system as well as the column. Finally, the column needs to be reequilibrated to initial conditions before the next sample can be injected. All these steps are shown in Fig. 13. The dashed line shows the gradient as it is programmed into the method file. The gradient starts at time 0. At the end of the gradient duration a brief isocratic washing step is incorporated. Then the solvent composition is brought back to initial conditions (arrow 1), and the column is reequilibrated with the solvent composition at the beginning of the gradient (arrow 2). The solid line is the gradient profile at the top of the column, and arrow 3 indicates the delay of the gradient due to the system delay volume. In the optimization of a gradient, it is
Design of Rapid Gradient Methods / 119
Fig. 13 Elements of a gradient separation. See text for explanation.
important to know the system delay volume. It can be measured by running a gradient without a column from methanol to methanol with a small amount of an ultraviolet absorber, for example, 0.1% uracil. Arrow 4 indicates the end of the gradient at the column top. While a sharp step has been programmed, the actual profile of this is somewhat smeared out due to the gradient delay volume of the system. Arrow 5 marks the actual gradient profile emerging from the column. It is delayed by the gradient delay volume and by the column volume. The column reequilibration has been studied by Li [13]. For a good reequilibration of the column with the starting mobile phase, Li recommends a purge at the end of the gradient consisting of 3 times the system delay volume plus 5 times the column volume before the next injection is made. This amounts to about 6 mL for a modern HPLC system equipped with a 4.6-mm ⫻ 50-mm column. It should be pointed out that these reequilibration times can be used most effectively if the next injection is made just before the start of the next gradient reaches the column inlet. The time saving equivalent to the purge of 1 gradient delay volume can be quite substantial.
120
/ Neue et al.
Several schemes have been used for a further reduction in run time, albeit with additional expense. For example, in a scenario where rapid gradients are performed with MS detection, a significant expense of the instrument is the MS detector. Under these circumstances, two gradient HPLC systems can be used to feed the detector: while the detector monitors the effluent from the gradient separation occurring on one instrument, the other instrument is equilibrating the column with the starting mobile phase. At the end of the run, the gradient is started on the second HPLC system and its effluent is monitored by the detector, while the first system is reequilibrating the column with the mobile phase at the start of the gradient. This scenario maximizes the time savings, since the detector monitors an active analysis at all times. Even more sophisticated scenarios with a single APCI-MS/MS system monitoring the effluent from two HPLC instruments have been reported [14]. The same gradient was executed by both HPLC systems, and the column effluent was combined to the MS/MS system. Good results were obtained with plasma samples. In a simpler column-switching scheme, the reequilibration of a column with the starting mobile phase is aided with a second pump that delivers only the starting mobile phase of the gradient. In this case, the columns are switched after the gradient pump has delivered the 3 system volumes of mobile phase A as recommended by Li, and the second column is reequilibrated with mobile phase A while the gradient is executed on the first column and vice versa. In this scenario, the additional expense is just an additional pump and switching valves, but the time savings comprise only the reequilibration time of the column with mobile phase A. If columns with a small diameter are used, e.g., 1 mm I.D., the purge of the system delay volume becomes the dominant factor in the reequilibration of the column with the starting mobile phase. A proper purge of the system at flow rates compatible with a 1-mm column may take as long as 40 min. In such a case, a switching valve can disconnect the column after completion of the gradient run, and the system can be purged back to the starting mobile phase at high flow rate. This step can be completed in less than a minute, and only then is the column connected back into the mobile-phase stream for reequilibration of the column proper. Unlike samples derived from body fluids, such as plasma, samples from combinatorial synthesis are usually clean. Nevertheless,
Design of Rapid Gradient Methods / 121 protection of the column from extraneous debris increases its lifetime. Well-designed guard columns do not affect the quality of a separation and can be used to increase column life. Figure 14 shows such an example. After over 2300 injections, a column protected with a guard column showed a significant deterioration of the peak shape. Then the guard column was replaced with a new guard column, and the high quality of the separation was restored. This is not an unusual example, but rather typical for the performance of high-quality guard columns. Once the typical lifetime of a guard column has been established for the routine analysis typical for analysis of combinatorial chemistry samples, the replacement of a guard column in regular intervals can be scheduled to keep the instrument operational around the clock. Properly executed, such an approach can avoid column downtime and the need to rerun hundreds of samples in high-speed analysis. In addition, an increase in column life may not be a major factor when using an analytical column, but the cost
Fig. 14 Improvement of column lifetime with the use of a guard column. A deterioration of the column performance was observed at injection 2350. The guard column was replaced, and good performance was achieved again thereafter, as shown for injection 3575. Column: 3.5-µm Symmetry C 18 ,2.1 mm ⫻ 30 mm with a 3.5-µm Symmetry Sentry Guard column, 2.1 mm ⫻ 10 mm. Sample dissolved in DMSO.
122
/ Neue et al.
of preparative columns definitely justifies the use of guard columns. It should be pointed out that the best protection of the main column is achieved if the packing in the guard column is identical to the packing in the analytical column. This also guarantees that there are no unusual peak distortions, as could occur when a different packing is used for the guard column.
B.
Fast Analytical Separations
In this section we will cover some practical aspects of fast analytical chromatography. Generally, we would like to find a fast generic separation scheme without regard to the actual separation. The discussion in the theoretical section has taught us that: 1. One should compare columns of about equal ratio of particle size to column length. 2. Faster gradients require shorter columns with a smaller particle size. 3. We should expect a substantial improvement in performance for fast gradients when comparing a shorter column with a smaller particle size to a longer column with a larger particle size. 4. This improvement is more pronounced for very rapid gradients. For the purpose of this study, we have measured the peak capacity of gradient separations as a function of the column choice and the operating conditions. The peak capacity can be measured readily by determining the average peak width for typical peaks of interest and dividing the gradient duration by the average peak width. In this study, we are comparing specifically two columns of roughly equal ratio of column length to particle size: a 5-cm ⫻ 4.6-mm, 5µm column and a 2-cm ⫻ 4.6-mm, 2.5-µm column, both containing XTerra MS C 18 packing. The internal diameter of both columns was 4.6 mm. The gradient duration was varied from 1 to 16 min, and the flow rate was varied from 0.5 to 5.6 mL/min. The results are shown in Table 1. For both columns, longer gradient times result in an improvement in the peak capacity. We obtained a maximum peak capacity of about 140 peaks for the 5-µm column at a flow rate of around 2– 3 mL/min, corresponding to a column dead time around 15 s. For
Design of Rapid Gradient Methods / 123 Table 1 Comparison of the Peak Capacity of a 2-cm, 2.5-µm XTerra MS C 18 Column and a 5-cm, 5-µm XTerra MS C 18 Column XTerra MS C18, 2.5 µm, 2 cm ⫻ 4.6 mm Flow Rate (mL/min)
Gradient duration (min)
0.5
0.7
1.0
1.4
2.0
2.8
4.0
5.3
1 2 4 8 16
22 34 59 97
16 28 43 75 109
25 46 72 109 134
36 60 86 120 149
47 74 101 129 162
56 79 105 133 170
63 85 108 142 177
68 100 122 163 194
XTerra MS C 18 ,5 µm, 5 cm ⫻ 4.6 mm Flow Rate (mL/min)
Gradient duration (min)
0.5
0.7
1.0
1.4
2.0
2.8
4.0
5.6
1 2 4 8 16
19 28 49 96
13 22 35 58 98
16 37 64 85 119
24 46 68 100 132
32 50 75 107 132
36 57 82 114 139
40 60 85 114 139
42 70 85 105 132
the 2.5-µm column, we obtained a maximum peak capacity of nearly 200 peaks at a flow rate around 5 mL/min. This corresponds to a column dead time of only about 4 s. For very fast 1-min separations, we obtained a peak capacity of over 40 peaks for the 5-µm column and a peak capacity of 68 peaks for the 2.5-µm column at a flow rate of 5.3 mL/min. For the 2.5-µm column, this flow rate coincided with the pressure limit of the HPLC instrument. At both fast gradients and slower gradients, the shorter column with the smaller particle size gives better resolution at high flow rates. However, the improvement in performance is larger with the faster gradient. For slow 16-min gradients, the improvement is of the order of 25% (peak capacity 177 versus 139). For fast gradients, use of the column with the smaller particle size results in a roughly 60% improvement in resolution at fast flow (peak capacity 68 versus 52). While this is somewhat less than the theoretical prediction shown in Fig. 10 of the theoretical section, it is still quite substantial.
124
/ Neue et al.
Fig. 15 Rapid gradient analysis of a typical sample from combinatorial synthesis. Column: 2-cm ⫻ 4.6-mm 2.5-µm XTerra MS C 18. Flow rate, 4.0 mL/ min; gradient duration, 1 min; gradient from 5% to 95% acetonitrile with a constant concentration of 0.1% TFA; injection volume, 15 µL of diluted sample.
A peak capacity of nearly 70 peaks is quite enough for the typical separation problems encountered in combinatorial chemistry. An example of such a rapid separation is shown in Fig. 15. It shows the analysis of a typical sample from a combinatorial library, using a 2-cm ⫻ 4.6-mm XTerra MS C 18 column packed with 2.5-µm particles. The analysis time was only 1 min, but the sharpness of the peak is good. The purpose of the analysis was a check of the purity of the sample. One can see that there are two side products, but they are present at a very low concentration. The small impurity peak emerging just in front of the main peak exemplifies the excellent resolving power of this very rapid gradient. The use of 2-cm, 2.5-µm columns allows the analysis of over 1000 samples per day.
C.
Fast Preparative Separations
It is often necessary to purify a quantity of a compound for further studies. For new lead candidates, the purification of 10–100 mg is
Design of Rapid Gradient Methods / 125 sufficient for metabolic screening, binding assays, and related early testing protocols. Preparative chromatography is a convenient and rapid way to achieve this goal. With proper scaling and using the identical packing for analytical and preparative chromatography, the preparation of a sufficient quantity of a pure lead compound can be accomplished without difficulty. A typical target requirement is a high recovery of the compound of interest at a purity exceeding 90%. Often, a fast routine procedure for preparative chromatography is desirable due to the large number of samples that need to be purified. It is not uncommon that 50–100 samples need to be processed every day with a single preparative instrument. Automated separation schemes guided by MS detection of the fraction with the expected molecular weight are preferred. The best and simplest approach for accomplishing a preparative separation is to scale the chromatographic conditions of an analytical separation. This is best done by scaling the gradient volume V g in direct proportion to the column volume V c , while maintaining the mobile-phase compositions at the beginning and at the end of the gradient: Vg,p Vc,p ⫽ Vg,a Vc,a
(19)
The subscripts p and a refer to preparative and analytical columns, respectively. If the column back-pressure and the instrument capability allow it, the most convenient way to scale the chromatographic conditions is to increase the flow rate in proportion to the column volume and leave the gradient table the same, but this is not a must. If the preparative run needs to be accomplished at a lower speed due to instrument constraints, one should still keep the ratio of the gradient volume to the column volume constant to achieve the same elution pattern. If we do this and use the same packing, the elution profile on the preparative column is identical to the elution profile achieved with the analytical column, with the possible exception of elution patterns early in the chromatogram. The reason for a difference in the emergence of early-eluting compounds is the fact that the gradient delay volume of the analytical instrument forces an initial isocratic section into the chromatographic run. Conversely, the gradient delay volume of preparative instruments is often negligible
126
/ Neue et al.
relative to the column volume, or at least proportionally smaller than the gradient delay volume of analytical instruments. If we would like to keep the exact same elution profile for the preparative chromatography as for the analytical chromatography, we need to add a gradient delay volume to the preparative run. This means that we add an isocratic time segment to the beginning of the preparative run, which shifts the gradient start and end in accordance with this adjustment. While this adds to the gradient run time, the identical elution pattern relative to the analytical chromatography simplifies the setup and the interpretation of the preparative chromatography. All compounds in a library can be subjected to the same elution pattern, without a need to rethink the strategy of the preparative separation. If the same instrument is used for the preparative and the analytical run, the delay time for the preparative run can be calculated using the following equation:
td ⫽
冢
Vc,p Vd ⋅ ⫺1 Fp Vc,a
冣
(20)
where V d is the delay volume of the instrument, and F p is the flow rate used for the preparative column. The delay volume of an instrument is best measured by running a step gradient to a solvent spiked with a UV absorber (e.g., methanol to methanol with 0.1% uracil) without a column attached. The delay volume is the volume needed to reach 50% of the resulting step recorded by the detector. If different instruments are used for the preparative and the analytical separations, the gradient delay volumes of both instruments need to be known to make the correct adjustment of the gradient delay for the preparative runs. The most important practical issue of a generic preparative gradient is the dissolution of the sample. On one hand, the injection volume of sample dissolved in the beginning mobile phase of the gradient is fairly unconstrained, due to the reconcentration of the sample at the top of the column. However, the solubility of the sample in the highly aqueous mobile phase at the start of the gradient is often limited, and an excessively large injection volume limits the sample throughput. On the other hand, if we inject the sample with a large amount of organic solvent present, peak distortions prior to
Design of Rapid Gradient Methods / 127 a true mass overload can be observed. A better solution needs to be found. If a large number of samples needs to be processed automatically, it is best to accomplish this using a combination of a preparative HPLC instrument with a mass spectrometer as the specific detector (Fig. 16). The MS detector can be set up to look for the expected molecular weight of the preparation and direct the collection of the fraction that contains this molecular weight [16,17]. A photodiode array detector (PDA) is used to monitor the entire preparative run. In the setup shown in Fig. 16, we used a Waters preparative HPLC system consisting of a Waters Delta 600 Multisolvent Delivery System with extended flow heads, a Model 2700 Sample Manager, a Model 996 Photodiode Array Detector, and a Micromass ZMD 4000 MS Detector controlled via FractionLynx software. The column used was a 19-mm ⫻ 50-mm, 5-µm Symmetry C 18 column, and the flow
Fig. 16 FractionLynx system for the routine chromatographic preparation of purified samples from combinatorial syntheses. The system consists of a Waters Delta 600 Preparative HPLC Gradient System, a Waters Reagent Manager and 2700 Sample Manager, a Waters 515 pump for the sample dilution, a 19-mm ⫻ 50-mm Waters Symmetry C 18 column, a Waters 996 Photodiode Array Detector, and a Micromass ZMD mass spectrometer. A Gilson FC 204 Fraction Collector is used to collect the purified sample. A stream splitter (LC Packings) is located between the column, the fraction collector and the detectors. The split ratio is 10,000:1, which means that only a very small fraction of the sample is directed toward the mass spectrometer. See text for description of the setup shown in dashed lines.
128
/ Neue et al.
rate was 30 mL/min. The split ratio of the preparative flow to the detector line was 10,000: 1, and the make-up flow to the detector was 1 mL/min. This instrument setup serves to ensure the highest yield in the collected fraction while avoiding an overload of the MS and PDA detectors from the highly concentrated sample. Three solvent lines were used on the Delta 600 Multisolvent Delivery System: line A with water, line B with acetonitrile, and line C with 1% formic acid (v/v) in water. A constant delivery of 10% of the flow from line C eliminated the need to premix water and acetonitrile with formic acid to achieve a constant concentration of 0.1% formic acid throughout the gradient. The gradient starts typically with a 0.5-min hold at 5% acetonitrile, then a rapid 5-min linear gradient to 90% acetonitrile is executed. After the gradient, the column is allowed to reequilibrate with the starting mobile phase for 4 min. Thus the total run time is less than 10 min from run to run. Figure 17a shows a loadability study as a total ion chromatogram of three sulfa drugs dissolved in a mixture of 50% DMSO, 30% methanol, and 20% water. The compounds were sulfanilamide, sulfathiazine, and sulfasoxazole. The load was increased from 5 to 30 to 100 mg per peak (15, 90, 300 mg total load), with an injection volume of 75, 450, and 1500 µL, respectively. Good results up to a load of 120 mg (40 mg per peak, not shown) were obtained with this approach. At the highest load, as shown in the single ion trace for the last peak (Fig. 17b), resolution is worse than expected because of peak splitting. In addition, a portion of the analyte elutes in the void volume. The first signs of this phenomenon can be seen already at the intermediate load (30 mg per peak), with the strong fronting of the sulfazoxazole peak and the emergence of a shoulder that coincides with the elution time of sulfathiazine.
Fig. 17 Loadability study with a set of sulfa drugs dissolved in a mixture of 50% DMSO, 30% methanol, and 20% water. Column: 19-mm ⫻ 50-mm 5-µm Symmetry C 18. Equipment setup as shown in Fig. 16. Injection, 75, 450, 1500 µL; load, 5, 30, 100 mg/peak or 15, 90, and 300 mg total load. Samples: 1, sulfanilamide; 2, sulfathiazole; 3, sulfisoxazole. Gradients: A, water; B, acetonitrile; C, 1% formic acid in water, 0.5-min hold at 85:5:10 A:B:C, then linear gradient to 90:10 B:C in 5 min. Flow rate, 30 mL/min. (a) Total ion chromatogram. (b) Single-ion chromatogram of peak 3 (Sulfisoxazole) at m/z 290, the last peak in Fig. 17a.
Design of Rapid Gradient Methods / 129
(a)
(b)
130
/ Neue et al.
However, it can be demonstrated without difficulty that the loadability limit observed in this experiment is not due to the mass load of the sample, but due to the effect of the sample diluent. In a second experiment shown in Fig. 18, we demonstrated that an injection of 1.5 mL of a sample resulted in the same peak distortion independent of the mass load. In the chromatograms shown in Fig. 18, the amount of sample injected was varied from a total load of only 15 mg of sample to 90 mg and 300 mg. The same peak distortion is observed. This means that the peak distortion is not a function of the sample load, but rather is caused by the large fraction of organic solvent used to dissolve the sample. The large concentration of organic solvent is incompatible with the initial gradient conditions and causes an undesired migration of the analytes during injection. Consequently, it would be better to prepare the sample in the mobile phase used at the beginning of the gradient or even in water. However, the low solubility of many samples in a solvent mixture with such a high water content as well as the need for a generic sample dissolution protocol sets limits to this approach.
Fig. 18 Comparison of the chromatogram of a small mass (15 mg), intermediate (90 mg), and large mass (300 mg) in a large sample volume (1.5 mL) of 50% DMSO, 30% methanol, and 20% water, demonstrating solvent overload. Single-ion chromatogram at m/z 290, representing peak 3, sulfasoxazole. Samples and conditions as in Fig. 17.
Design of Rapid Gradient Methods / 131 An alternative approach reduces this problem (Fig. 16, dashed lines). The sample is dissolved in 100% DMSO and is carried to the column with the stream of 100% acetonitrile, flowing at a flow rate of 5% of the total gradient flow rate, i.e., at 1.5 mL/min. The Delta 600 gradient system is now delivering a gradient from 0% to 85% organic at a flow rate of 28.5 mL/min. In this setup, the sample is thus diluted with the 100% aqueous 0.1% formic acid stream on-line just in front of the column. If indeed a precipitation of the sample occurs, it does so at the column top, where the retention of the sample is desired anyway. The extremely short residence time in the mixing tee and the 2-cm ⫻ 0.1-cm connection tubing prevents clogging of the tubing from a potential precipitation of the sample. This automatic dilution scheme significantly increases the sample load that can be injected onto a column. Figure 19 shows the results of a high-mass and high-volume injection using this automatic dilution scheme in comparison to the
Fig. 19 Comparison of direct injection and automatic dilution at high sample load. Samples: 1, sulfanilamide; 2, sulfathiazole; 3, sulfisoxazole. The top chromatogram resulted from the direct injection of 2000 µL of the sulfa drug mixture dissolved in 100% DMSO; the bottom chromatogram was obtained with the automatic dilution scheme described in the text. Sample load 267 mg per peak, for a total sample load of 800 mg. Shown are the total-ion chromatograms.
132
/ Neue et al.
classical approach. The analytes were the same as in Fig. 17. They were dissolved in 100% DMSO, at a concentration of 133 mg/mL each, for a total concentration of 400 mg/mL. The chromatograms were obtained with a 2000-µL sample, resulting in an injection of 267 mg of each analyte, or 800 mg in total. In the upper chromatogram in Fig. 19, the system and the chromatographic conditions were similar to those shown in Fig. 17 and 18 except that the initial hold was extended from 0.5 to 5 min. This extension of the initial hold time allowed for a direct comparison to the automatic dilution scheme, but did not result in an improvement in the separation relative to the initial experience with the shorter hold shown in Fig. 17 and 18. With automatic dilution of the sample at the head of the column, as shown in the lower chromatogram of Fig. 19, peak shape and resolution is drastically improved. A significant broadening of the peaks was observed only at loads of more than 133 mg per analyte or 400 mg in total. This broadening is due to true column overload rather than the phenomena observed due to the injection of the sample in a high concentration of organic solvent. As a result, significantly higher loads were achieved using this approach compared
Fig. 20 Single-ion chromatogram at m/z 290. Other conditions as in Fig. 19. The top trace shows the early elution and co-elution of peak 3 with the other compounds. The bottom trace shows the clean elution of peak 3 obtained with the automatic dilution scheme described in the text.
Design of Rapid Gradient Methods / 133 to the first approach. The difference in the mass loadability is demonstrated even more strongly in Fig. 20. Here we compare the singleion chromatograms at m/z 290 for the classical approach of sample injection in the upper trace to the automatic dilution approach shown in the lower trace. On the top, we observe the peak distortion of peak 3. A comparison of this trace with the upper chromatogram of Fig. 19 shows the co-elution of peak 3 with the other compounds in the chromatogram. On the bottom, we see that the early breakthrough of peak 3 has been completely eliminated, and that only a standard peak broadening due to normal mass overload is observed. Consequently, the final mass loadability of the packing for these compounds was better than 50 mg per peak per gram of packing, demonstrating the superior loadability of the packing under gradient conditions [6]. The setup described should therefore be suitable for the rapid automatic processing of samples at loads of about 400 mg using just a 19-mm ⫻ 5-cm column. The same approach allows the purification of 1 g on a 30-mm ⫻ 5-cm column.
IV.
CONCLUSION
In this study we examined the optimal conditions for gradient chromatography under the constraint of a maximum analysis time. In isocratic chromatography, the optimal chromatographic condition is always at or slightly above the minimum of the van Deemter curve. For gradient chromatography under the constraint of a maximum analysis time, this is not true. Not unexpectedly, we find that new optima emerge at higher linear velocities, and that these optima are functions of the gradient duration. For very slow gradients, the optimum linear velocity approaches the optimum linear velocity of isocratic separations. For very rapid gradients, high linear velocities result in better separations at a fixed gradient run time. For a 5-cm, 5-µm column, the optimum occurs around a linear velocity of 0.5 cm/s for an 8-min gradient and around a linear velocity of about 0.8 cm/s for a 1-min gradient. Even faster separations are possible with smaller particle sizes. For an 8-min gradient, column dead times around 3 s are optimal, with linear velocities around 0.8 cm/s for a 2-cm, 2.5-µm column. This column also gives improved results for a 1-min gradient compared to a 5-µm column, but the true performance optimum cannot be reached within the pressure capability of existing instrumentation. The measured peak capacity of just under
134
/ Neue et al.
70 peaks for a 1-min gradient is sufficient for the demands of combinatorial analysis. Of course, such results can be obtained only with well-optimized instrumentation. The detector time constant and the sampling rate need to be optimized together with the gradient and the column choice. We have discussed instrument parameters in great detail. Gradient separations also provide improved results for preparative separations. Properly optimized, the column loadability is superior in gradient separations to that in isocratic separations. An automatic dilution scheme has been demonstrated that solves the problems associated with poor solubility of the sample in water. Good results have been obtained with loads up to 50 mg of sample per 1 g of packing. Automated, mass directed purification schemes can be employed that result in 400 mg of purified sample with a gradient run time of 10 min and a total cycle time of 14 min.
SYMBOLS α ∆c ∆t R a b B c dp DM Fp G H k k0 ke kg
relative retention factor change in organic composition over the gradient retention time difference constant of van Deemter equation constant of van Deemter equation slope of the relationship of ln(k) with the solvent composition constant of van Deemter equation particle size diffusion coefficient preparative flow rate dimensionless gradient steepness parameter height equivalent to a theoretical plate (isocratic) retention factor (isocratic) retention factor at the start of the gradient actual retention factor at point of elution retention factor under gradient conditions
Design of Rapid Gradient Methods / 135 L N P Rs td tg t0 u Vc Vd Vg w wk
column length plate count peak capacity resolution gradient delay time gradient duration retention time of an unretained peak linear velocity column volume instrument delay volume gradient volume peak width peak width in dimensionless units, w k ⫽ w/t 0
REFERENCES 1. J. N. Kyranos and J. C. Hogan, Jr., Anal. Chem. News & Features, June 1: 389A (1998). 2. W. K. Goetzinger and J. N. Kyranos, Am. Lab., 27 April (1998). 3. L. R. Snyder, J. W. Dolan, and J. R. Gant, J. Chromatogr., 165: 3 (1979). 4. J. W. Dolan, J. R. Gant, and L. R. Snyder, J. Chromatogr., 165: 31 (1979). 5. M. Z. El Fallah, in U. D. Neue (Ed.) HPLC Columns, Theory, Technology and Practice,Wiley-VCH, New York, 1997. 6. G. Guiochon, S. G. Shirazi, and A. M. Katti, Fundamentals of Preparative and Nonlinear Chromatography, Academic Press, 1994. 7. F. D. Antia and Cs. Horvath, J. Chromatogr., 484: 1 (1989). 8. M. Z. El Fallah and G. Guiochon, Anal. Chem., 63: 859 (1991). 9. A. Felinger and G. Guiochon, J. Chromatogr. A, 724: 27 (1996). 10. A. Felinger, G. Guiochon, Biotechnol. Prog. 12 (1996), 638. 11. L. R. Snyder, in Cs. Horvath (Ed.), HPLC—Advances and Perspectives, Vol. 1, Academic Press, New York, 1980, p. 208. 12. U. D. Neue, HPLC Columns, Theory, Technology and Practice,Wiley-VCH, New York, 1997. 13. J. Li and J. Morawski, LC/GC, 16(5): 468 (1998).
136
/ Neue et al.
14.
W. A. Korfmacher, J. Veals, K. Dunn-Meynell, X. Zhang, G. Tucker, K. A. Cox, and C.-C. Lin, Rapid. Commun. Mass Spectrom,13: 1991 (1999). W. K. Goetzinger, private communication (1999). L. Zeng and D. B. Kassel, Anal. Chem., 70: 4380 (1998). L. Zeng, L. Burton, K. Yung, B. Shushan, and D. B. Kassel, J. Chromatogr. A,798: 3 (1998). H. N. Weller, M. G. Young, S. J. Michalczyk, G. H. Reitnauer, R. S. Cooley, P. C. Rahn, D. J. Loyd, D. Fiore, and S. J. Fischman, Molec. Diversity, 3(1): 61 (1997).
15. 16. 17. 18.
4 Molecularly Imprinted Extraction Materials for Highly Selective Sample Cleanup and Analyte Enrichment Francesca Lanza and Bo¨rje Sellergren Institut fu¨r Anorganische Chemie und Analytische Chemie, Johannes Gutenberg Universita¨t Mainz, Mainz, Germany
I. INTRODUCTION II. MULTIPURPOSE SPE PHASES
138 139
III.
HIGH-AFFINITY SPE PHASES
141
IV.
MOLECULARLY IMPRINTED SOLID-PHASE EXTRACTION (MISPE)
142
PREVIOUS MISPE PROTOCOLS A. Off-line Protocols B. On-line Protocols
146 150 160
THE DEVELOPMENT OF NEW MISPE PROTOCOLS
163
VII. POLYMER SYNTHESIS-RELATED FACTORS
164
V.
VI.
VIII. METHODS FOR SYNTHESIS AND SCREENING OF LARGE GROUPS OF MIPs 165 IX.
TEMPLATE BLEEDING—AN UNRESOLVED ISSUE IN MISPE PROTOCOLS 168
X. CONCLUSIONS
169 137
138
/ Lanza and Sellergren ACKNOWLEDGMENTS REFERENCES
170 170
I. INTRODUCTION Analytical methods for rapid determination of chemicals, biomolecules, or cells are of growing interest in various fields such as environmental control, drug development, health protection, forensics, and biotechnology [1–4]. The steadily increasing number of organic environmental pollutants (pesticides, industrial waste, and degradation products) is a major health hazard and has therefore created a need for efficient environmental monitoring relying on low-cost, rapid, and automated methods of analysis. The analytical challenge associated herewith is obvious from the large number of European Community-sponsored projects on this subject. An equally important challenge is the rapid analysis of low levels of drugs, narcotics, and their metabolites in humans as well as the quality control of food, foodstuffs, and pharmaceuticals. The development of efficient analytical methods in these fields will have a direct impact on the quality and efficiency of health care, forensic activities, and food industries.
Fig. 1 Solid-phase extraction (SPE) for enrichment and cleanup of complex samples prior to analyte (a) quantification. Steps: (1) conditioning of the sorbent, (2) percolation of the water sample, (3) washing step, and (4) elution step.
Molecularly Imprinted Extraction Materials / 139 The analysis of target molecules in complex mixtures often requires pretreatment steps [5]. First, if the analyte is present in low concentration, it needs to be concentrated in order to be detected by standard analytical techniques. Second, if it is present in a complex mixture of similar compounds, a cleanup step is required. Solidphase extraction (SPE) has become a widely used technique for sample pretreatment because it is easily automated, flexible, and environmentally friendly (Fig. 1) [6–8]. It consists of percolating a known volume of a liquid sample through a solid sorbent (in the form of a cartridge, column, or a disk) under carefully chosen conditions favoring the preferential absorption of the analyte over the matrix components. In the most common technique, uncharged analytes are absorbed on a hydrophobic sorbent. The analyte of interest is then recovered from the sorbent by elution in a small volume (smaller than the applied sample volume) of an appropriate solvent mixture. High enrichments as well as efficient sample cleanup can thus be obtained in one step.
II. MULTIPURPOSE SPE PHASES The attainable enrichment and cleanup in SPE depend primarily on the selectivity and affinity of the sorbent for the selected target analyte or analytes, the sample load capacity for the analytes, and the rate of mass transfer to and from the binding sites, the latter affecting the minimum desorption volume and thus the enrichment that can be obtained. Other factors of importance are the reproducibility of the recovery yields and the stability and reusability of the sorbent when on-line procedures are desired. For hydrophobic analytes, satisfactory results are usually obtained using standard reversed-phase sorbents. Thus hydrophobized silica (C8, C18), styrene–divinyl benzene co-polymers (PS-DVB), and graphitized carbon black (GCB) are the conventional sorbent materials used in SPE (Fig. 2) [8]. A high sample load capacity together with the wide range of trapped analytes are the interesting features of reversed-phase materials (mainly C18), whereas their poor selectivity, the narrow pH stability range, and the limited breakthrough volumes for hydrophilic analytes are their main disadvantages. In the case of hydrophilic analytes, enrichment and cleanup are usually more difficult to obtain. This leads to disturbances in the subsequent chromatographic analysis. Therefore sorbents that can be tailor-made to bind hydrophilic
140
/ Lanza and Sellergren
Fig. 2 Common multipurpose sorbents used in SPE.
analytes have been developed. For instance, highly porous mixedmode sorbents containing both hydrophobic and hydrophilic polar moieties have been designed (see Fig. 2) [9]. These have been shown to retain highly polar pesticides and metabolites such as glyphosate and hydroxylated atrazine metabolites that are not well retained on high-capacity reversed-phase materials. The mixed-mode sorbents also exhibit high selectivity and sensitivity for extracting a wide range of acidic, basic, and neutral drugs from complex matrices such as blood, urine, and human plasma. These phases are not suited, however, for on-line operation. In these cases restricted-access materials (RAM) have found widespread use by allowing direct repeat injections of untreated blood plasma or urine samples [10–12]. Through pore-size selective surface modification techniques the outer surface of the support particles can be made hydrophilic and nonsticky to plasma proteins, whereas the inner-surface, nonaccessible-to-plasma proteins are made hydrophobic or equipped with other functional groups (Fig. 3). Thus these phases are able to efficiently
Molecularly Imprinted Extraction Materials / 141
Fig. 3 Restricted-access materials (RAM) developed for direct injection of biological fluids.
remove the high-molecular-weight matrix components and quantitatively retain the drugs of interest and have therefore been used for the on-line enrichment and sample cleanup of biofluids. In this field, liquid–liquid extraction is often still the recommended method for sample preparation, although the need for high-throughput screening in the drug monitoring step has made solid-phase extraction, often in the 96-well block format, the method of choice [13].
III. HIGH-AFFINITY SPE PHASES Complex sample matrixes in combination with the need for lower detection limits has increased the demand for phases that can bind an analyte or a group of analytes both strongly and selectively. Biological molecular recognition elements (enzymes, antibodies, or cells) can be used for this purpose. For instance, immunoaffinity (IA)-based enrichment techniques can be used for specific determination of a particular analyte or a class of analytes [2,8]. As with most affinity techniques that rely on biological recognition elements, method robustness and reproducibility, receptor immobilization technique, and the time and cost involved in providing the receptor
142
/ Lanza and Sellergren
for a new target analyte are factors that may be associated with problems or difficulties. Therefore the development of synthetic phases that can offer similar recognition properties seems to be motivated. One promising way to introduce selectivity in chemical analysis is to use molecularly imprinted polymers [14–16]. These can in favorable cases recognize small molecules with affinities and selectivites exceeding that of antibody–antigen and have due to their robustness, capacity, and reproducibility potential as reusable adsorbents in assays or sample pretreatment.
IV.
MOLECULARLY IMPRINTED SOLID-PHASE EXTRACTION (MISPE)
Highly selective sorbents toward a large number of analytes of environmental and pharmaceutical interest can be prepared by molecular imprinting (Fig. 4) [17]. This technique consists of adding the analyte, or a structurally related analog, as template during the preparation of the sorbent. After the removal of the template, binding sites complementary to the analyte are obtained directly in the
Fig. 4 Principle of molecular imprinting.
Molecularly Imprinted Extraction Materials / 143 bulk material. The sorbents so prepared are able to selectively rebind the analyte from complex matrices. The most versatile approach to the synthesis of molecularly imprinted sorbents is based on self-assembly of the template and a complementary functionalized monomer (e.g., methacrylic acid, MAA) prior to polymerization (Fig. 5). Thus the template remains associated with the growing polymer during synthesis and, by adding a large portion of a cross-linking monomer (e.g., ethylenglycoldimethacrylate), sites complementary to the template are formed that remain stable after template removal. The sorbent so obtained is then ready for use and, provided that suitable conditions for template rebinding are known, can be implemented directly in the sample pretreatment procedure. MIPs have often been compared to immunosorbents due to their high affinity and selectivity for several important target analytes (Table 1) [18–20]. In fact, they often ex-
Fig. 5 Versatile imprinting protocol using methacrylic acid (MAA) as functional monomer and ethyldimethacrylate (EDMA) as cross-linking monomer in the imprinting of nicotine.
144 / Lanza and Sellergren
Table 1 Association Constants and Site Densities Calculated from Competitive Radioligand Binding Assays Using MIPs Association constant, Ka (M⫺1) Template
Assay solvent
Binding-site density n (µmol/g)
Class 1
Class 2
Class 1
Class 2
Ref.
Theophylline
MeCN-HOAca
2.9 ⫻ 106
1.5 ⫻ 104
0.016
1.3
[18]
Diazepam
Tolueneheptaneb
5.6 ⫻ 107
4.3 ⫻ 105c
0.0062
0.17c
[18]
Enkephaline
MeCN-HOAca Aq. pH 4.5d
7.7 ⫻ 106 1.0 ⫻ 107
2.3 ⫻ 104 8.3 ⫻ 104c
0.017 0.004
Morphine
Toluene-HOAce Aq. pH 6d
1.1 ⫻ 107 8.3 ⫻ 105
1.1 ⫻ 105 4.2 ⫻ 104
1.2 0.8
39 6.9
[20] [20]
S-Propranolol
Toluene-HOACe Aq. pH 6d
2.5 ⫻ 107 2.5 ⫻ 108
4.4 ⫻ 104 2.4 ⫻ 105
2.0 0.63
38 28
[63] [63]
Acetonitrile/acetic acid: 99/1. Toluene/heptane: 3/1 (v/v). c A third class of binding sites was also present. d Sodium citrate buffer (20 mM ) containing 10% or 2% ethanol. e Toluene/acetic acid: 98/2 (v/v). b
[20] [20]
Molecularly Imprinted Extraction Materials / 145
a
1.0 0.72c
146
/ Lanza and Sellergren
hibit striking similarities to antibodies, mainly with respect to crossreactivity and binding constants when a comparison is made with the corresponding immobilized antibody. Equally apparent are the differences between the two types of sorbents. Immunoaffinity phases are sometimes difficult to regenerate and reuse, and their stability is limited. On the contrary, imprints can be used under harsh conditions (high temperatures and in nonaqueous and extreme pH solutions) and regenerated and reused with little loss of performance, even after prolonged use [21–23]. They can be easily prepared, whereas the preparation and purification of antibodies is time-consuming and involves several steps. The phases are also different with respect to the optimum solvent of operation. Antibodies bind antigens with high affinity in water, in contrast to most MIPs, which exhibit optimum recognition in organic solvents [17]. Finally, an important difference between imprinted sorbents and immunoaffinity phases lies in the reproducibility of their preparation. Whereas molecularly imprinted materials can be prepared with small batch-to-batch variations, the biological immunoresponse is not reproducible and the resulting antibodies are therefore different from batch to batch.
V.
PREVIOUS MISPE PROTOCOLS
Several examples of the successful application of imprints to solidphase extraction have been described in the literature (Table 2). In most cases MIPs have been used in a cartridge format in the offline mode to clean up and enrich samples prior to various detection techniques. In some cases the protocols have only been preliminarly tested to show the feasibility of the approach [24–27]; in others they have been partly validated [28–30]. The retentivity and selectivity of the sorbents are first assessed by HPLC-UV. The results of these studies reveal whether aqueous samples can be applied directly to the column, as in the extraction of hydrophobic analytes from aqueous media, e.g., biological or environmental samples, or whether the analyte has to be transfered to an organic solvent of lower polarity prior to the MISPE, as for analytes that are not sufficiently retained in aqueous mobile phases. In the former case the adsorption step often occurs in the reversed-phase LC mode of operation and is usually nonselective (Fig. 6). Here the selectivity is introduced by selective wash procedures which serve to remove nonselectively bound matrix components [25,31–33]. The analyte is then recovered by
Molecularly Imprinted Extraction Materials / 147 eluting with solvents of increasing eluotropic strength. In the latter case, however, the adsorption occurs similar to the LC straightphase mode of operation and can be highly selective [28–30,34]. In MISPE the washing and elution conditions need to be carefully optimized in terms of pH, ionic strength, and solvent composition in order to exploit fully the MIPs’ ability to recognize the template. In cases where recognition is driven by electrostatic interactions, best results are often seen when using the porogenic solvent as wash solvent (solvent memory effect) [35]. This causes compatibility problems, since many of the porogens are not miscible with water. This problem can be solved by drying the MIP cartridges before the washing step [25,36]. Unfortunately, the drying step is difficult to combine with on-line procedures, which instead rely on the use of water-miscible solvents in the wash step. Alternatively, a coupled column system may be adopted in which the analytes are first adsorbed nonspecifically on a multipurpose SPE column, e.g., a hydrophobic sorbent (C18 reversed-phase or restricted-access sorbent) and then transferred to a dedicated MIP sorbent in an organic medium. This approach is suitable for on-line applications, as recently demonstrated by Boos et al. [37] and Bjarnason et al. [38], and has the advantage of allowing the direct injection of biological samples (Fig. 7). In addition, it leaves the imprinted sorbent wetted by organic solvents, minimizing swelling and shrinking of the material, conditioning times, and bleeding effects. Swelling–shrinkage of the polymers leads to changes in site accessibility and can cause entrapment of the analytes, which in turn leads to low recoveries. In this context the pulsed elution method recently described by Mullet and Lai also does not require solvent switching and would thus allow a higher sample throughput [34]. Once a selective extraction has been achieved, the analyte needs to be efficiently desorbed from the phase in the smallest volume. This will lead to a large enrichment and a high recovery of purified analyte. Ideally, the purity and concentration are high enough to allow direct quantification in the eluate [24,34], possibly in combination with more specific detection techniques such as DAD or MS. Most of the imprinted sorbents used in MISPE were prepared using MAA and EDMA as monomers. In these cases elution of more weakly bound analytes such as triazines [25,36,38], 7-hydroxycoumarin (2) [32], or theophylline (12) [34] (see Tables 2 and 3) can be achieved using methanol or water as elution solvent. For more
148
/ Lanza and Sellergren
Table 2 Sample Application, Washing and Elution Conditions in Previous MISPE Protocols
# Analyte
MIP format
Recognition assessment
A Pentamidine
Column
HPLC
Off-line SD
B 7-Hydroxycoumarin
Cartridge
—
Off-line SD
C Sameridine
Batch
Radioligand Off-line SA binding experiment
D Simazine
Cartridge
HPLC
Off-line SD
E Atrazine
Cartridge
Batch rebinding
Off-line SA
F Nicotine
Cartridge
HPLC
Off-line SA
G Propranolol
Cartridge
Cumulative Off-line SD recovery curves
H Darifenacin
Cartridge
HPLC
Protocol
Cotinine
Myosmine
Off-line SA
Molecularly Imprinted Extraction Materials / 149
Sample Urine
Urine
Sample modification prior to application Dilution with buffer pH 5 (1/1), then dilute with MeCN (3/7) pH adjustment (6.00)
Washing step
Elution step
Detection technique
Reference
MeCN/buffer pH 9 (1), pH 5 (2): 7/3
MeCN/Buffer pH 3: 7/3
UV
[24]
Water
MeOH
CZE
[32]
Plasma
Extraction in heptane
Heptane/EtOH (1/1)
Heptane/ EtOH/ NaOH (5 N ) 170/13/17
GC-FID
[29]
Water
None
MeOH
HPLC-UV
[25]
Beef liver extracts
Extraction in chloroform
Dichloromethane (previous drying of the cartridge) Chloroform
MeCN/HOAc: 9/1 (v/v)
HPLC-UV ELISA
[30]
Chewing gum
Dissolution in ethyl acetate/ 25% ammonia
MeCN/H2O/ TFA (97.3/ 2.5/0.2)
HPLC-UV
[28]
Plasma bile, and urine
None
MeOH/H2O 1% TFA or TEA
Scintillation counting HPLC-UV
[33]
Plasma
Addition of CH3CN 50% (deproteinization)
CH3CN with 0.1% TFA
HPLC-UV Scintillation counting
[26]
Acetonitrile (previous drying of the cartridge)
—
50% and then 70% CH3CN
(Continued)
150
/ Lanza and Sellergren
Table 2 Continued
# Analyte
MIP format
I
Cartridge
Tamoxifen
Recognition assessment
Protocol
—
Off-line SD
J Theophylline and analogues
Column and Batch remicrobinding column
On-line SA
K Triazines
Column
HPLC
On-line
L Bentazone
Cartridge
HPLC
Off-line SD
SD ⫽ selective desorption. SA ⫽ selective adsorption. CZE ⫽ capillary zone electrophoresis. TFA ⫽ trifluoracetic acid. TEA ⫽ triethylamine. ELISA ⫽ enzyme-linked sorbent assay.
strongly bound analytes such as stronger nitrogen Bro¨nsted bases, efficient elution has been achieved using eluents of the same base solvent but with the addition of small amounts of acids (e.g., acetic acid, trifluoroacetic acid) or base (e.g., triethylamine) [26–28,31,39].
A.
Off-line Protocols
Sample Cleanup of Biological Fluids The first MISPE protocol reported in the literature [24] involved the use of an imprinted dispersion polymer in a column format, for the selective enrichment of pentamidine (1) in urine. In this case the high selectivity of the polymer allowed the analyte to be detected directly in the eluate without the need for any further chromato-
Molecularly Imprinted Extraction Materials / 151
Sample
Sample modification prior to application
Washing step
Elution step
Detection technique
Reference
Plasma and urine
None
H2O, MeCN
MeCN/HOAc 4/1 (v/v)
HPLC-UV
[31]
Serum
Extraction in chloroform
Pulses of MeCN
Pulse of MeOH
UV
[34]
Humic acid, apple extract, urine Water
Adsorption on C18, desorption with MeCN
MeCN
Water plug
HPLC-UV
[38]
pH adjustment (5.0)
Buffer (50 mM, pH 5, 50 mM NaCl), MeCN
MeOH/HOAc 99/1 (v/v)
HPLC-UV
[27]
graphic separation (entry A in Tables 2 and 3). In view of the high selectivity of MIPs, this is a viable approach which brings the benefits of shorter analysis times and simpler instrumentation. In most cases, however, MISPE has been used before a chromatographic separation step. The MIP has been applied in a batchwise extraction [29] or in columns or cartridges [26–28,30–32,39]. The former format was used by Andersson et al. for the analysis of the analgesic drug sameridine (3) in human plasma (entry C in Tables 2 and 3) [29]. Quantification limits below nanomolar concentrations were required. In order to circumvent problems related to bleeding of nonextracted template (see below), this group used a compound (14) that is structurally related to the analyte as template. The structural similarity is unlikely to seriously impair the binding of the analyte and thus the detection limit can be significantly lowered. The accuracy and intra-assay precision of the complete MISPE protocol based on the batchwise preconcentration of sameridine from human plasma were at least as good as those ob-
152
/ Lanza and Sellergren
Fig. 6 Reversed-phase and normal-phase modes of operation in the off-line application of MIPs in SPE.
served for the routine liquid–liquid extraction method. The advantage of the MISPE protocol was the high selectivity, resulting in less interference in the subsequent GC analysis and potentially higher sensitivities by applying larger sample volumes. The evaluation of a MIP for the solid-phase extraction of propranolol (9) (entry G in Tables 2 and 3) from biological fluids was addressed by Martin et al. [33,39,40], who compared the cumulative recovery curves obtained with different solvents of increasing eluotropic strength containing various modifiers. It appeared that the conditions chosen for the elution of the retained propanolol were extremely important to ensure a selective extraction. In fact, the imprinted material nonspecifically adsorbed a range of molecules struc-
Fig. 7 RAM or C18–MIP coupled column system for prefractionation based on size and polarity and transfer of analyte to a solvent for enhanced recognition of the target analyte.
Molecularly Imprinted Extraction Materials / 153
Table 3
Details of the MIP Synthesis and Method Validation Results for the Protocols in Table 2 Conc. ranged
60°Cf
2400
30 (urine)
10–60 nM
10 (90) j
CHCl3
60°C
400
0.25 (urine)
60–300 µM
90
C
Toluene
UV, RT
5
8–120 nM
100
D Simazine
CHCl3
50°Cg
2000
600 mg (plasma) 500 (water)
460 nM
91
E Atrazine
CHCl3
60°C
250
20 (CHCl3 extract)
0.005–0.5 ppm
89 (93)k
F Nicotine
CH2Cl2
UV, 10°C
270
0.5 (EtOAc extract)
250 µM
ca. 100
0.5 (plasma)
0.040–10 µM
ca. 100
Porogen
A Pentamidine
i-Propanol/ water
B 7-Hydroxycoumarin
Polym. techn.a
Recoverye (%)
Cotinine
Myosmine
G Propranolol
Toluene
30
/ Lanza and Sellergren
Vsamplec (mL)
Template
154
mMIPb (mg)
#
THF/EtOAc
UV, 4°C
100
1 (dil. plasma)
47–235 µM
ca. 100
I
Tamoxifen
MeCN
60°C
500
0.5 (urine)
1.3 µM
102
J
Theophylline
CHCl3
60°C
h
0.020 (CHCl3 ext)
13–5200 µM
—
K Simazine
CH2Cl2
UV, 4°C
i
300 nM
—
L Bentazone
CHCl3
60°C
200 (humic acid water) 50 (water)
42 µM
96
500
Polymers were prepared using MAA (A–K) or 4-vinylpyridine (4VPY)/MAA (1/1) (L) as functional monomers and EDMA as cross-linking monomer in the presence of the template and various solvents (porogens) as shown schematically in Fig. 4. EtOAc ⫽ ethylacetate, THF ⫽ tetrahydrofuran, RT ⫽ room temperature. a Polymerization technique. Thermochemical initiation at elevated temperatures or UV-photochemical initiation at low temperature. b Weight of dry MIP packed in the SPE cartridge or column. c Volume and type of sample added to the MIP phase. d Typical concentration range used in the extraction tests. e Exact or approximate recoveries reported. f Polymer obtained as dispersable agglomerates of micrometer-sized particles. g Polymer prepared in water suspension. h Column dimensions: 80 ⫻ 4 mm. i Column dimensions: 150 ⫻ 4.6 mm. j Recovery by loading and washing in pH 5 buffer. In parentheses is the corresponding value after loading and washing in pH 7 buffer. k Recovery obtained by HPLC quantification. In parentheses is the corresponding value when quantification was done by ELISA.
Molecularly Imprinted Extraction Materials / 155
H Darifenacin
156
/ Lanza and Sellergren
turally similar and dissimilar to propranolol. However, a careful choice of eluent allowed the nonspecifically bound analytes to be washed off first and propranolol to be recovered quantitatively in a second step. Interestingly, comparing the strong modifiers trifluoroacetic acid (TFA) and triethylamine (TEA), only the latter resulted in selective cleanup. Also, matching the wash solvent with the porogenic solvent seemed to result in higher selectivity and recovery. Stevenson et al. compared MISPE with immunoaffinity extraction for the quantification of tamoxifen (11) in urine and plasma (entry I in Tables 2 and 3) [31,41]. This group reported considerable problems with template bleeding, i.e., continuous release of small amounts of nonextracted template, particularly upon solvent changes. A number of different wash procedures were tested, but none gave satisfactory results that would allow reliable quantification of low levels of the drug. Although the materials could be used for the selective extraction at higher concentration levels and the manufacturing of the materials was simple, the immunoaffinity technique was favored since these phases allowed the drug to be selectively extracted from biological fluids, at low concentration levels and with high accuracy. Similar problems were reported by Venn et al. in the application of MISPE in the quantification of the developmental drug darifenacin (10) in blood (entry H in Tables 2 and 3) [26,42]. A comparison of MIPs using different functional monomers and porogens with respect to chromatographic selectivity for the drug was first performed. Using MAA as the functional monomer, the resulting MIP could discriminate efficiently between the drug enantiomers and was therefore employed in an off-line SPE procedure. The blood was deproteinized by addition of acetonitrile (50%) and the sample applied directly to the MISPE cartridge. By washing the cartridge with acetonitrile, the drug was efficiently retained on the MIP, whereas early breakthrough was seen on the nonimprinted control cartridge. The selectivity was further investigated by comparing the extraction of a number of structurally related analogs and substructures. Thus the materials performed well in terms of selectivity, capacity, and robustness. Nevertheless, applications for trace- or ultra-trace-level analysis was again precluded due to the excessive template bleeding. Extraction of Analytes of Environmental Concern Triazines belong to a group of widespread herbicides that have become major pollutants of soil and ground waters [43,44]. Therefore
Molecularly Imprinted Extraction Materials / 157 analytical methods are needed to allow them to be monitored at concentrations below micrograms per liter. The triazines are available in large numbers with small structural differences and with different known basicities and hydrophobicities (Table 4). This class of compounds is therefore well suited as a model system in molecular imprinting [45]. Previous results indicate that highly selective MIPs can be prepared against triazines [45–48]. The strongest interactions are expected to be a cooperative hydrogen bond between the nitrogen para to the chlorine (or thiomethyl) substituent and an exocyclic amino group of the triazine and the carboxylic acid group of MAA [49]. This interaction is also accompanied by secondary interactions which can be the association of a second acid group to the triazine ring system or shape complementarity arising from the polymer backbone. Thus the polymers effectively discriminate between chloro- and S-triazines. For the chlorotriazines, highest selectivity is seen using MAA as monomer; whereas for the S-triazines, trifluoromethyl acrylic acid is the monomer of choice [50]. The use of MISPE for the cleanup of beef liver extracts for the quantification of atrazine (5) by HPLC-UV or ELISA was evaluated by Muldoon and Stanker (entry E in Tables 2 and 3) [30]. In both cases the application of MISPE resulted in improvements in analyte quantification. In particular, for the HPLC method the use of MISPE
Table 4 Structures of Triazines with Various Bro¨nsted Basicities and Hydrophobicities Chlorotriazine
Triazine Terbutylazine Atrazine DIA DEA Simazine Propazine
Thiotriazine
R1
R2
Triazine
R1
R2
Ethyl Ethyl Ethyl H Ethyl i-Prop
tert-Butyl i-Prop H i-Prop Ethyl i-Prop
Ametryn Prometryn
Ethyl i-Prop
i-Prop i-Prop
158
/ Lanza and Sellergren
improved the accuracy and precision and lowered the limit of detection. In the case of ELISA, better accuracy was achieved in the determination, although the precision was similar. With the analyte present at parts-per-billion levels, the reliability of either determination method would have been marginal without the MISPE step. The application of MIPs prepared using triazines as templates to the solid-phase extraction of water samples requires drying the cartridge after the sample application in order to remove water traces which would disrupt the interactions between the analytes and the sorbent. In the protocol described by Matsui et al. [25], the aqueous sample was applied to the MIP cartridge, which was then carefully dried prior to a selective dichloromethane wash (entry D in Tables 2 and 3). HPLC-UV analysis of wastes and extracts showed that all the impurities were washed off without significant elution of simazine (4) and that the analyte was recovered quantitatively in the eluate. Recently, the group of Barcelo performed a detailed investigation on the off-line application of chlorotriazine-imprinted MIPs in the analysis of real environmental water samples [36]. Two imprinted polymers were synthesized using either dichloromethane or toluene as porogen and terbutylazine as template and were implemented to the solid-phase extraction of six chlorotriazines in natural water and sediment samples. All extracted samples were analyzed by liquid chromatography/diode array detection (LC/DAD). Several washing solvents, as well as different volumes, were tested to remove the matrix components adsorbed nonspecifically on the sorbents. This cleanup step was shown to be of prime importance to the successful extraction of the pesticides from the aqueous samples. Interestingly, the wash solvent giving the highest level of cleanup and recovery was the same as that used as porogen. Optimal analytical conditions were obtained using the MIP imprinted with dichloromethane as porogen, with 2 mL of dichloromethane used in the washing step and the preconcentrated analytes eluted with 8 mL of methanol. The recoveries were higher than 80% for all the chlorotriazines except propazine (53%) when 50- or 100-mL groundwater samples, spiked at the 1-µg/L level, were analyzed (Fig. 8). The limits of detection varied from 0.05 to 0.2 µg/L when preconcentrating a volume of 100 mL of groundwater sample. Natural sediment samples from the Ebre Delta area (Tarragona, Spain) containing atrazine and deethylatrazine were Soxhlet extracted and analyzed by
Molecularly Imprinted Extraction Materials / 159
Fig. 8 LC/DAD chromatogram at 220 nm obtained after preconcentration of 100 mL of groundwater sample spiked at 1 µg/L through (a) a blank cartridge prepared using a nonimprinted control polymer and (b) a cartridge prepared using a polymer imprinted with terbutylazine. Peaks: 1 ⫽ deisopropylatrazine, 2 ⫽ deethylatrazine, 3 ⫽ simazine, 4 ⫽ atrazine, 5 ⫽ propazine, 6 ⫽ terbutylazine, and I.S. (internal standard) ⫽ diuron.
160
/ Lanza and Sellergren
the methodology developed in this work. No significant interference from the sample matrix was noticed, thus indicating good selectivity of the sorbents used. The above studies show that the imprinted sorbents can be used in two modes: in the reversed-phase mode (sample application) and in the affinity mode (dichloromethane washing). Otherwise there are only few reports on the use of MIPs for the pretreatment of samples of environmental concern. Recently an MIP imprinted using the herbicide bentazone as template was synthesized and evaluated by Baggiani et al. [27] in frontal liquid chromatography (entry L in Tables 2 and 3). The material was then used as sorbent for the solid-phase extraction and a preliminary procedure for the enrichment of the analyte from water samples was tested. Good recoveries (91–96%) and concentration factors of 3.2–15.2 were found.
B.
On-line Protocols
Extraction of Nicotine In the development of a cleanup step for nicotine and structurally related compounds in nicotine-containing chewing gum, an on-line MISPE procedure was developed (Fig. 9) and the resulting method compared with an SPE run on a nonimprinted blank column (entry F in Tables 2 and 3) [28]. The sample pretreatment would in this case be particularly suited for MISPE, since the analytes together
Fig. 9 On-line extraction of nicotine chewing-gum extracts.
Molecularly Imprinted Extraction Materials / 161 with matrix components were present in a nonpolar solvent, used to dissolve the chewing-gum matrix. Nicotine MIPs were prepared by the standard protocol using MAA as the functional monomer. The resulting polymers were crushed, sieved, and packed into standard HPLC columns for evaluation. Using acetonitrile as wash solvent, nicotine could be eluted as a sharp peak from the imprinted sorbent by adding 0.2% TFA. This led to quantitative recoveries in a large concentration interval. The elution conditions were then optimized on-line. The chewing-gum extract was added to the column and the column was washed with acetonitrile, followed by elution of the analytes. As seen, there is a pronounced difference between the MIP and the blank column in terms of breakthrough volumes and recovery of the analytes in the elution step. Whereas three of the four analytes broke through prior to the elution step on the blank column, only the most hydrophobic analyte—β-nicotyrine—broke through on the MIP (Fig. 10). Transferring the procedure to an off-line mode, quantification by reversed-phase HPLC showed that all analytes except β-nicotyrine were recovered quantitatively. Due to the enhanced stability of the nicotine MIP, achieved by thermal annealing, and the reusability of the MIP, the procedure may be suited for automation. MISPE with Pulsed Elution A method for the analysis of theophylline in blood serum based on molecularly imprinted solid-phase extraction with pulsed elution (MISPE-PE) was developed by Mullett and Lai (entry J in Tables 2 and 3) [34]. The method made use of an MIP column for the on-line enrichment of theophylline and injection of a small plug of methanol to produce a rapid pulsed desorption of the analyte. The sample was applied in chloroform, since in this solvent a complete retention of the analyte was observed. The matrix constituents (e.g., lipids) and the potential interferences (e.g., other drugs) which were not recognized by the binding sites were rapidly eluted with chloroform. Then a rapid and quantitative recovery of theophylline was accomplished in a pulsed format through injection of 20 µL of methanol. In this way the eluting analyte could be detected directly by UV. The pulsed elution approach was further applied to micro-SPE columns (0.8 mm I.D., 80 mm long), which offer the advantage of lower consumption of solvents and faster analysis times [51]. With the application of 20-µL solvent pulses it was demonstrated that the nonspecifically bound analytes could be removed and nicotine could be recovered
162
/ Lanza and Sellergren
Fig. 10 Adsorption and elution of 100 µL of a placebo chewing-gum extract spiked with nicotine (0.5 mg/mL) and β-nicotyrine, cotinine, and myosmine (0.05 mg/mL). The step elution was as follows: MeCN, 0–8 min; MeCN ⫹ 0.2% TFA ⫹ 2.5% H2O, 8–12 min; MeCN, 12–15 min. The peak identity was established by separate injections and spectral analysis using a diode array detector. Rest refers to still-not-eluted β-nicotyrine, cotinine, and myosmine. In MISPE, only β-nicotyrine could be detected in the adsorption and wash steps.
quantitatively over a linear dynamic range of 1–1000 µg/L and with a detection limit of 1.8 µg/mL. On-line Coupled Column Approaches The drying step required in many off-line procedures is not compatible with on-line operations. Moreover, if the MIP columns are to be reused, fouling caused by macromolecular matrix components precludes direct injection of biological fluids. In these cases an attractive alternative is to use a precolumn designed to transfer low-molecularweight analytes into an organic solvent while simultaneously excluding biological macromolecules. This is possible using a coupled
Molecularly Imprinted Extraction Materials / 163 column system where the analytes are first adsorbed nonspecifically on a multipurpose SPE column, e.g., a hydrophobic sorbent (C18 reversed-phase or restricted-access sorbent), and then transferred to a dedicated MIP sorbent in an organic medium. The approach is suitable for on-line applications, as recently demonstrated by Boos et al. [37] and Bjarnason et al. [38], and has the advantage of allowing the direct injection of biological samples (Fig. 7). Furthermore, the column switching allows the pretreatment step to be separated from the analysis so that one sample can be pretreated while another is being analyzed, leading to a higher sample throughput.
VI.
THE DEVELOPMENT OF NEW MISPE PROTOCOLS
Four different steps are involved in the development of a new solidphase extraction protocol based on molecular imprint: synthesis of the material, assessment of its recognition properties, development of an extraction protocol, and final validation of the protocol. First, a suitable combination of monomers and a porogenic solvent have to be found on which the template can be successfully imprinted [17]. The thermodynamic stability of the monomer–template assemblies is the key factor in this step. One or more functionalized vinyl monomers are sought which are capable of forming strong noncovalent interactions with the functional groups of the template. The monomer–template assemblies need to be stable enough for them to be transformed to specific binding sites during the polymerization. The structure and morphology of the polymer matrix are strongly influenced by the type and amount of cross-linker and porogen [52]. These factors affect in turn the sorption–desorption kinetics and the stability of the material and therefore require careful optimization as well. Once a suitable material has been found, a chromatographic characterization step is required in which the recognition properties of the sorbent are evaluated in different mobile phases. In order to predict suitable extraction protocols, the sorbents should thus be characterized in terms of affinity, selectivity, and sample load capacity (breakthrough volumes) for the analytes. Except for establishing recognition based on the retention times of the template and analogs at one or two different concentrations, frontal analysis or batch rebinding at several concentrations yields thermodynamic information in terms of binding energy and mass transfer kinetics [22,53,54].
164
/ Lanza and Sellergren
The assessment can also be done by an off-line SPE procedure in which the cumulative recovery curves of the sorbents are evaluated in different solvents and pHs. A suitability test for propranololimprinted sorbents, to be used in SPE of plasma and urine samples, was developed by Olsen et al. [40]. This was based on the solvent composition required to give a recovery of either 20% or 50% of the extracted material (ES20 or ES50 values) on the blank and the imprinted cartridge. The method was developed in order to allow a rapid estimate of the specific and nonspecific contributions to overall binding. Finally, a validation step for the overall MISPE procedure is mandatory to allow the use of the method itself in place of the regulatory methods. The issues to be considered here are the accuracy and precision of the method based on the MISPE protocol, its limit of quantitation, selectivity, and ruggedness. In particular, the interand intra-assay precision need to be checked with real samples and with certified reference materials and methods [29,30].
VII. POLYMER SYNTHESIS-RELATED FACTORS Each step in the imprinting procedure (from the synthesis to the validation) needs to be thoroughly investigated and optimized when a new MIP is synthesized against a given template. In particular, the selection of appropriate functional monomers capable of forming stable complexes with the template is critical for the success af any MISPE protocol. The type and number of functional groups present in the template provide some guidelines for the choice of the functional monomers. Ionic or hydrogen-bonding interactions are usually the first interactions to be exploited in targeting the template. Thus nitrogen bases have been successfully imprinted by using carboxylic acids such as MAA as functional monomers (see Table 3). Here the strength of binding, disregarding cyclic hydrogen-bonded structures, increases with the acidity of the functional monomer and the basicity of the template [55]. Likewise, basic functional monomers (2- and 4vinyl pyridine, diethylaminoethyl methacrylate, aminoethyl methacrylate, allylamine, N-vinyl pyrrolidinone, 4-(5)-vinylimidazole, polymerizable amidines) have been used to imprint acidic templates and acidic functional monomers (methacrylic acid, trifluoromethacrylic acid, itaconic acid, p-vinylbenzoic acid) have been used to imprint basic templates [17].
Molecularly Imprinted Extraction Materials / 165 The selection of the possible functional monomers can further be based on the extensive literature from the area of host–guest chemistry [56], the binding mode to biological receptors [2], affinity chromatography [57], or crystal engineering [58]. Reports can be found describing stability and stochiometry of complexes between structural analogs of the template and functional monomers, and these reports can thus serve as initial guidelines in the monomer selection. In a number of cases, combinations of two or more functional monomers giving terpolymers or higher polymers have resulted in polymers showing better recognition properties than the corresponding co-polymers [59,60]. However, these systems are complicated by monomer–monomer association which competes strongly with the template–monomer association unless one of the monomers has a particular preference for the template. So far the success of any imprinting protocol has depended more on chemical intuition than on a rational approach. However, for general use of the imprinting technique, rational synthesis and evaluation protocols need to be established. A key step in this development process is the rapid identification and optimization of the main factors affecting the recognition properties and the morphology of the polymers. These factors are the type and concentration of functional monomers, cross-linking monomer, free-radical initiator, and porogen. Further factors related to the polymerization conditions are the pressure and temperature of polymerization. Owing to the timeconsuming procedures generally adopted for the preparation of MIPs, the assessment of the recognition properties of large numbers of MIPs so prepared is not realistic on a practical time scale. Instead, a system allowing parallel synthesis of large groups of MIPs and insitu screening of the recognition properties is desirable. One solution consists of scaling down the batch size, allowing the materials to be synthesized in vials or wells suited for direct analysis of the recognition properties.
VIII. METHODS FOR SYNTHESIS AND SCREENING OF LARGE GROUPS OF MIPs A scaled-down version of the established monolith procedure was recently developed by Takeuchi et al. [50] and by Lanza and Sellergren [61], in which molecularly imprinted polymers were pre-
166
/ Lanza and Sellergren
pared on the bottom surfaces of chromatographic autosampling vials and assessed directly by batch rebinding tests. In both cases, triazines were used as model templates and the functional monomers were used as screening parameters. In the work of Takeuchi et al., the relative amounts of functional monomers (MAA and TFM) were varied to optimize the recognition properties of MIPs prepared using ametryn and atrazine as templates, whereas six different functional monomers were screened by Lanza and Sellergren using terbutylazine (TER) as template. In both cases the polymers were prepared by photoinitiation following previous protocols with ethyldimethacrylate as a cross-linking monomer and dichloromethane as solvent. The standard batch size was scaled down ca. 100 times to a total volume of 100 µL. As seen in Fig. 11, the screening was performed in two steps. The primary assessment is based on quantitative HPLC or UV-absorbance analysis of the amount of template released from the polymer in a given solvent. Among the six functional monomers tested, methyl methacrylate, 4-vinylpyridine, and N-vinyl-2-pyrrolidone led to rapid and
Fig. 11 Method for small-scale synthesis and in-situ analysis of large series of MIPs.
Molecularly Imprinted Extraction Materials / 167 quantitative extraction, whereas methacrylic acid and trifluoromethylacrylic acid led to polymers that retained the template the most. A modest increase in the recovery was obtained after additional washing in dichloromethane but, as expected, only after adding strongly competing solvents such as acetic acid were high overall recoveries of the template obtained. The selection criteria in the first screen were based on the amount of released template in a given solvent, i.e., acetonitrile or dichloromethane, the latter being the same solvent as the one used during polymerization. As discussed above, this solvent is commonly the solvent in which the polymers exhibit the most pronounced recognition [35]. Thus, in a particular batch, if a quantitative release is seen in this step the resulting polymer cannot be expected to rebind a significant amount of the template and may thus be discarded. In this step the polymers prepared using MAA and TFM are the only polymers that can be expected to show enhanced rebinding of the template in this solvent. After having established useful functional monomers, a secondary screening for selectivity was performed. In this screening, the rebinding of the template to the MIPs was investigated in parallel to the rebinding to the corresponding blank, nonimprinted polymers [61]. Alternatively, an internal standard, structurally related to the template, may be added and the differential binding investigated [50]. Supporting the validity of the suggested screening approach, the MAA and TFM MIPs exhibited enhanced rebinding and selectivity. A scaled-up version of the MAA polymer was assessed as stationary phase in chromatography. As seen in Table 5, a number of structur-
Table 5 Capacity Factors of Triazines on Terbutylazine MIP and Blank Polymers
168
/ Lanza and Sellergren
ally related triazines were strongly retained on the corresponding column. The retention of atrazine (ATRA) was also significantly higher than the retention obtained using a polymer imprinted with atrazine. This shows that template analogs can be used to generate binding sites showing high affinity and selectivity for a particular target compound. Thus the described screening protocol allows rapid identification of the factors of importance for the generation of binding sites of high affinity and selectivity for a particular target compound.
IX. TEMPLATE BLEEDING—AN UNRESOLVED ISSUE IN MISPE PROTOCOLS One problem in molecular imprinting concerns the small amount of template that remains strongly bound to the polymer after extraction. This usually amounts to more than 1% of the amount of template given to the monomer mixture and remains bound even after careful washing of the polymer [19,21,29,31]. This may not constitute a problem in preparative separations or catalysis, but when the materials are used for sample preparation prior to analytical quantification of low levels of analytes, bleeding of this fraction will cause false results [29,31]. In view of the previously determined concentrations (Table 3), this essentially limits the use of MISPE to samples with parts per thousand or parts per million as low concentrations [42]. In spite of careful washing, slow leakage of template often occurs upon exchange of solvents. This problem may be reduced by thermal posttreatment of the materials [28] or prevented by the use of a template analog as template [29]. Obviously, more effective wash procedures may also lead to less bleeding [31]. An oftenoverlooked factor is the amount of sorbent that is required for high recovery (see Table 3). Previous examples have shown that the capacity of the sorbents often is far higher than required, i.e., only a fraction of the sites are actually used in the extraction step [42]. By adding just enough sorbent to give complete occupancy of the templated sites upon extraction, the effective bleeding will be reduced. The question is then how the template is incorporated in the polymer. Some templates may react with free radicals and become covalently incorporated in the matrix. In order to establish whether this is occurring, model reactions between the template, the initia-
Molecularly Imprinted Extraction Materials / 169 tor, and a small amount of monovinyl monomer should be carried out. No leakage can occur by this mechanism, since the template would be covalently bound to the polymer. Leakage is possible, however, if the template is physically entrapped in the densely crosslinked regions of the nuclei [28]. Swelling and shrinking of the polymer may then cause slow bleeding of this fraction. Here it seems that thermal treatments at temperatures exceeding the glass transition temperature [22], but below temperatures where the sites are deformed, alleviates some of these problems. In addition, the problem is often reduced by avoiding solvent switches or drying of the sorbent. It is important to carefully determine the recovery of the template and to verify the identity of the recovered template. The recovery can be estimated either by determining the amount of template in the extracts or the amount of template in the polymer [21,62]. For this purpose, solution 1H-NMR, HPLC, GC, elemental analysis, and scintillation counting have been used. By NMR the identity of the extracted fraction is easily verified as well as unreacted monomer remaining in the polymer. Quantification can be done by comparing the integrals of the template with that of an internal standard. More accurate quantification of the amount of template in the extract can be obtained by chromatography [21]. Furthermore, elemental analysis on a heteroatom unique for the template can be used to determine the recovery. Higher sensitivities are obtained using trace analytical techniques. High sensitivities are also obtained using radioactively labeled templates. The amount of template in the polymer before and after extraction is then compared by scintillation counting on the polymer, or by total combustion of the polymer followed by counting of the isotopes [62].
X. CONCLUSIONS MIP-related applications in analytical chemistry are presently experiencing rapid growth. Most intensely studied are the use of MIPs as recognition elements in chemical sensors or in assays and MIPs as sorbents in solid-phase extraction (MISPE) to achieve a selective cleanup of biological samples with high analyte enrichment factors. The development of MIP-based sensors, apart from the synthesis of the recognition element with adequate recognition and kinetic properties, is associated with a number of technical difficulties, i.e., the
170
/ Lanza and Sellergren
coupling of the binding event to a measurable signal and the type of signal transducer, the sensor stability, and the sensor reusability. In this regard the MISPE applications are more straightforward in view of the more limited requirements. Thus for a new target analyte an MIP exhibiting adequate selectivity, affinity, and mass transfer properties in the recognition of the analyte are the primary requirements. Eventual problems associated with template bleeding also need to be addressed. Furthermore, the time needed to synthesize these phases will be decisive for whether they will be successful in competing with alternative biological recognition elements such as antibodies. However, when these requirements are fulfilled, the MIPs can be used directly in extraction columns or cartridges in combination with existing instrumentation. High target affinity and selectivity translates to lower detection limits, shorter analysis times, and cheaper instrumentation in the resulting method, factors that are likely to lead to widespread use of these phases in the near future.
ACKNOWLEDGMENTS The authors are greatful to the European Commission for financing part of the herein reported work within the program for Training and Mobility for Researchers (TMR), contract number FMRX CT 980173, with the following participants: Damia Barcelo´, CID-CSIC, Barcelona, Spain; Werner Blau, Trinity College Dublin, Ireland; Karl-Siegfried Boos, Maximilians-Universita¨t Mu¨nchen, Germany; Kees Ensing, University of Groningen, The Netherlands; George Horvai, Technical University of Budapest, Hungary; Lars Karlsson, Astra Ha¨ssle AB, Mo¨lndal, Sweden; David Sherrington, University of Strathclyde, UK.
REFERENCES 1. M. Vanderlaan, L. H. Stanker, B. E. Watkins, and D. W. Roberts, Immunoassays for Trace Chemical Analysis; Monitoring Toxic Chemicals in Humans, Food, and the Environment, American Chemical Society, Washington, DC, 1991. 2. J. O. Nelson, A. E. Karu, and R. B. Wong, Immunoanalysis of Agrochemicals, ACS Symp. Ser. 586, American Chemical Society, Washington, DC, 1995.
Molecularly Imprinted Extraction Materials / 171 3. T. E. Mallouk and D. J. Harrison, Interfacial Design and Chemical Sensing, ACS Symp. Ser. 561, American Chemical Society, Washington, DC, 1994. 4. E. Reid, H. M. Hill, and I. D. Wilson, Drug Development Assay Approaches, Including Molecular Imprinting and Biomarkers, Methodol. Surv. Bioanal. Drugs 25, Royal Soc. Chem., Cambridge, UK, 1998. 5. C. F. Poole and S. K. Poole, Chromatography Today, Elsevier, Amsterdam, 1991. 6. I. Liska, J. Krupcik, and P. A. Leclercq, J. High Resol. Chromatogr., 12: 577 (1989). 7. L. A. Berrueta, B. Gallo, and F. Vicente, Chromatographia, 40: 474 (1995). 8. M.-C. Hennion, J. Chromatogr. A, 856: 3–54 (1999). 9. M. S. Mills, E. M. Thurman, and M. J. Pedersen, J. Chromatogr., 629: 11 (1993). 10. K.-S. Boos and A. Rudolphi, LC GC, 15: 602 (1997). 11. J. A. Perry, J. Liq. Chromatogr., 13: 1047 (1990). 12. A. Rudolphi and K.-S. Boos, LC GC, 15: 814 (1997). 13. S. Pleasance and R. A. Biddlecombe, in E. Reid, H. M. Hill, and I. D. Wilson (Eds.), Drug Development Assay Approaches Including Molecular Imprinting and Biomarkers, Vol. 25, Royal Soc. Chem., Cambridge, UK (1998), p. 205. 14. R. A. Bartsch and M. Maeda, Molecular and Ionic Recognition with Imprinted Polymers, ACS Symp. Series 703, Oxford University Press, Washington, DC, 1998. 15. D. Kriz, O. Ramstro¨m, and K. Mosbach, Anal. Chem., 69: 345A (1997). 16. D. Barcelo´, Special Issue on Molecular Imprints and Related Approaches for Solid-Phase Extraction and Sensors in Chemical Analysis, Trends Anal. Chem., 18 (1999). 17. B. Sellergren, Trends Anal. Chem., 18: 164 (1999). 18. G. Vlatakis, L. I. Andersson, R. Mueller, and K. Mosbach, Nature, 361(6413): 645 (1993). 19. K. J. Shea, D. A. Spivak, and B. Sellergren, J. Am. Chem. Soc., 115: 3368 (1993). 20. L. I. Andersson, R. Mu¨ller, G. Vlatakis, and K. Mosbach, Proc. Natl. Acad. Sci. (USA), 92: 4788 (1995). 21. B. Sellergren and K. J. Shea, J. Chromatogr., 635: 31 (1993). 22. Y. Chen, M. Kele, P. Sajonz, B. Sellergren, and G. Guiochon, Anal. Chem., 71: 928 (1999).
172
/ Lanza and Sellergren
23. 24. 25.
D. Kriz and K. Mosbach, Anal. Chim. Acta, 300: 71 (1995). B. Sellergren, Anal. Chem., 66: 1578 (1994). J. Matsui, M. Okada, M. Tsuruoka, and T. Takeuchi, Anal. Commun., 34: 85 (1997). R. F. Venn and R. J. Goody, Chromatographia, 50: 407 (1999). C. Baggiani, G. Giraudi, C. Giovannoli, A. Vanni, and F. Trotta, Anal. Commun., 36: 263 (1999). A. Zander, P. Findlay, T. Renner, B. Sellergren, and A. Swietlow, Anal. Chem., 70: 3304 (1998). L. I. Andersson, A. Paprica, and T. Arvidsson, Chromatographia, 46: 57 (1997). M. T. Muldoon and L. H. Stanker, Anal. Chem., 69: 803 (1997). B. A. Rashid, R. J. Briggs, J. N. Hay, and D. Stevenson, Anal. Commun., 34: 303 (1997). M. Walshe, J. Howarth, M. T. Kelly, R. O’Kennedy, and M. R. Smyth, J. Pharm. Biomed. Anal., 16: 319 (1997). P. Martin, I. D. Wilson, D. E. Morgan, G. R. Jones, and K. Jones, Anal. Commun., 34: 45 (1997). W. M. Mullett and E. P. C. Lai, Anal. Chem., 70: 3636 (1998). D. Spivak, M. A. Gilmore, and K. J. Shea, J. Am. Chem. Soc., 119: 4388 (1997). I. Ferrer, F. Lanza, A. Tolokan, V. Horvath, B. Sellergren, G. Horvai, and D. Barcelo, Anal. Chem., 72: 3934 (2000). C. T. Fleischer, K.-S. Boos, F. Lanza, B. Sellergren, 23rd Int. Symp. on High Performance Liquid Phase Separations and Related Techniques, Granada, Spain, 1999. B. Bjarnason, L. Chimuka, and O. Ramstro¨m, Anal. Chem., 71: 2152 (1999). P. Martin, I. D. Wilson, G. R. Jones, and K. Jones, Methodol. Surv. Bioanal. Drugs, 25: 21 (1998). J. Olsen, P. Martin, I. D. Wilson, and G. R. Jones, Analyst, 124(4): 467 (1999). D. Stevenson, Trends Anal. Chem., 18: 154 (1999). R. F. Venn and R. J. Goody, Methodol. Surv. Bioanal. Drugs, 25: 13 (1998). V. Pichon, L. Chen, N. Durand, F. Le Goffic, and M.-C. Hennion, J. Chromatogr. A, 725: 107 (1996). D. Barcelo´, S. Lacorte, and J. L. Marty, Trends Anal. Chem., 14: 334 (1995). C. Dauwe and B. Sellergren, J. Chromatogr. A, 753: 191 (1996).
26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37.
38. 39. 40. 41. 42. 43. 44. 45.
Molecularly Imprinted Extraction Materials / 173 46. M. Siemann, L. I. Andersson, and K. Mosbach, J. Agric. Food Chem., 44: 141 (1996). 47. M. T. Muldoon and L. H. Stanker, J. Agric. Food Chem., 43: 1424 (1995). 48. J. Matsui, Y. Miyoshi, O. Doblhoff-Dier, and T. Takeuchi, Anal. Chem., 67: 4404 (1995). 49. G. J. Welhouse and W. F. Bleam, Environ. Sci. Technol., 27: 500 (1993). 50. T. Takeuchi, D. Fukuma, and J. Matsui, Anal. Chem., 71: 285 (1999). 51. W. M. Mullett, E. P. C. Lai, and B. Sellergren, Anal. Commun., 36: 217 (1999). 52. B. Sellergren, Makromol. Chem., 190: 2703 (1989). 53. P. Sajonz, M. Kele, G. Zhong, B. Sellergren, and G. Guiochon, J. Chromatogr., 810: 1 (1998). 54. C. Baggiani, F. Trotta, G. Giraudi, G. Moraglio, and A. Vanni, J. Chromatogr. A, 786: 23 (1997). 55. G. Albrecht and G. Zundel, Z. Naturforsch., 39a: 986 (1984). 56. J.-M. Lehn, Pure Appl. Chem., 66: 1961 (1994). 57. G. Street, Highly Selective Separations in Biotechnology, Blackie, Glasgow, 1994. 58. M. Munakata, L. P. Wu, and T. Kuroda-Sowa, Bull. Chem. Soc. Jpn., 70: 1727 (1997). 59. M. Kempe, L. Fischer, and K. Mosbach, J. Mol. Recognit., 6: 25 (1993). 60. O. Ramstro¨m, L. I. Andersson, and K. Mosbach, J. Org. Chem., 58: 7562 (1994). 61. F. Lanza and B. Sellergren, Anal. Chem., 71: 2092 (1999). 62. B. Sellergren, B. Ekberg, and K. Mosbach, J. Chromatogr., 347: 1 (1985). 63. L. I. Andersson, Anal. Chem., 68: 111 (1996).
5 Biomembrane Chromatography: Application to Purification and Biomolecule–Membrane Interactions Tzong-Hsien Lee* and Marie-Isabel Aguilar Department of Biochemistry and Molecular Biology, Monash University, Victoria, Australia
I. INTRODUCTION II. BIOMEMBRANE-MODIFIED SOFT-GEL CHROMATOGRAPHIC SUPPORTS A. Entrapment of Whole Cells B. Immobilization of Liposomes and Proteoliposomes III. BIOMEMBRANE-MODIFIED SILICA-BASED CHROMATOGRAPHIC SUPPORTS A. Covalently Attached Phospholipid Monolayers B. Adsorbed Monolayers and Bilayers IV. APPLICATIONS A. Protein Purification B. Bioaffinity Measurements C. Drug Partition Measurements D. Peptide– and Protein–Membrane Interactions
176 177 177 178 182 183 186 187 187 188 189 190
*Current affiliation: Liver Research Unit, Chang Gung Memorial Hospital, Taipei, Taiwan, Republic of China.
175
176
/ Lee and Aguilar
V. VI.
CONCLUSIONS AND FUTURE DIRECTIONS REFERENCES
196 197
I. INTRODUCTION High-performance liquid chromatography (HPLC) is now firmly established as the premier technique for the analysis and purification of a wide range of molecules. In particular, HPLC in its various modes has become the central technique in the characterization of peptides and proteins and has therefore played a critical role in the rapid advances in the biological and biomedical sciences over the last 10 years [1]. The enormous success of HPLC can be attributed to a number of inherent features associated with reproducibility, ease of selectivity manipulation, and generally high recoveries. The most significant feature is the excellent resolution that can be achieved under a wide range of conditions for very closely related molecules as well as structurally quite distinct molecules. This arises from the fact that all interactive modes of chromatography are based on recognition forces which can be subtly manipulated through changes in the elution conditions that are specific for the particular mode of chromatography. All biological processes depend on specific interactions between molecules, and affinity chromatography exploits these specific interactions to allow the purification of a biomolecule on the basis of its biological function or individual chemical structure. In contrast, reversed-phase (RP) HPLC, ion-exchange, and hydrophobic interaction chromatography separate peptides and proteins on the basis of differences in surface hydrophobicity or surface charge. These techniques therefore allow the separation of complex mixtures, while affinity chromatography normally results in the purification of one or a small number of closely related components of a mixture. The development of quantitative models to describe the interaction of biomolecules with immobilized chromatographic ligands has also allowed HPLC to be used as a physicochemical tool for studying the molecular basis of peptide and protein surface interactions. In the case of bioaffinity chromatography, these models allow the quantitative analysis of the thermodynamics and kinetics of biomolecular interactions to be assessed [2]. In the case of RP-HPLC, several studies have demonstrated the potential of this technique in the charac-
Biomembrane Chromatography / 177 terization of the dynamic behavior of peptides and proteins at lipidlike surfaces [3–8]. More recently, a number of novel stationary phases have been prepared in which the solid-phase material has been either covalently or noncovalently attached to a phospholipid ligand [9–12]. The interaction of peptides and proteins with cell membranes plays a critical role in the regulation of several biological phenomena and represents a very important class of biorecognition processes which have historically presented enormous problems in mimicking chromatographically. For example, it has been particularly challenging to purify membrane proteins. This chapter thus reviews the recent developments in biomembrane chromatography which combine the high-resolution potential of RP-HPLC and the specific interactive properties of affinity chromatography and the application of these new materials in protein purification and biophysical studies of membrane interactions.
II. BIOMEMBRANE-MODIFIED SOFT-GEL CHROMATOGRAPHIC SUPPORTS A range of chromatographic gel materials have been used to noncovalently attach liposomes or cell membrane vesicles to a solid support. Techniques developed for the successful immobilization of the phospholipids include (1) freeze–thawing a mixture of the desired biomembrane vesicle and the dried gel beads which induces fusion and entrapment, (2) dialysis-preparation of liposomes with the beads, or (3) adsorption of the liposomes to gels which were covalently modified with hydrophobic ligands.
A. Entrapment of Whole Cells The study of the structure–function relationships of biomembranes and their interaction with various molecules is ideally performed using biological membranes isolated directly from cells of living organisms. The modification of chromatographic supports with whole cells therefore represents one important class of biomembrane system in which the conformation, orientation, and assembly of membrane proteins and receptors can be maintained in their native state and therefore their functional activity can be studied. Gel particles prepared by cross-linking between piperazine diacrylamide and
178
/ Lee and Aguilar
methacrylamide in the presence of N-allyldimethylamine have been used to immobilize human erythrocytes [13]. The continuous gel beds were first dispersed mechanically to form irregular particles consisting of cavities or channels for the immobilization of red blood cells. The negatively charged red cells were adsorbed to the positively charged gel surface via ionic interaction. Other weak interactions also contributed to the immobilization, as indicated by a moderate amount of cells which also bound to the gel particles in the absence of the cationic ligands. This material was used to study the activity of the glucose transporter Glutl within a cell membrane and the measured dissociation constants for glucose binding were in agreement with previously published values, thereby demonstrating the potential of this approach for studying the function of membrane processes.
B.
Immobilization of Liposomes and Proteoliposomes
Due to the heterogeneous and complex components found in natural membranes, valuable complementary information on the physical properties and functional roles of individual components in membranes and the effects of the membrane on the structure and activity of peptides/proteins can be obtained through the use of model membrane systems containing the specific lipid/protein of interest. The preparation of specific model biomembrane chromatographic supports thus involves several steps, namely, (1) isolation of lipid molecules from the cell source of interest or the chemical synthesis of a given lipid, (2) construction of an appropriate model system containing specific lipid molecules, and (3) subsequent incorporation of a particular peptide/protein. Such a model system not only provides a chemically defined environment but the physical properties can also be controlled, for example, by temperature, pH, and ionic strength. Liposomes and proteoliposomes have been immobilized into various types of soft gel beads in a similar way to the entrapment of whole cells into soft gel supports [9], as illustrated in Fig. 1. The gel matrices used for the immobilization of liposomes and proteoliposomes require several considerations: (1) the size distribution of the gel beads, (2) the pore size and surface area of the gel matrices, and (3) the stability of the gel support. The relatively large agarose (Sepharose) and allyldextran-bisacrylamide (Sephacryl) gel beads were used in the early development of this class of biomembrane
Biomembrane Chromatography / 179
Fig. 1 Schematic illustration of the steric entrapment of liposomes or proteoliposomes into a cross-linked gel matrix.
chromatography [14]. However, their broad size distribution and large pores limited the use of high flow rate, and poor resolution was obtained during the separation process. In contrast, Superdex 200 PG and TSK 6000 PW provide a superior sorbent as a result of their relatively homogeneous size distribution and larger surface area. In addition, the relative rigidity of these gel beads allows higher flow rates to be employed, resulting in superior separation [15]. The methodology used to immobilize liposomes/proteoliposomes involved the steric entrapment in gel beads of Sepharose 6B and Sephacryl S-1000 [14]. The nontrapped lipid vesicles are then removed by chromatographic procedures and by centrifugation. The amount of entrapped liposomes can be increased by increasing the initial lipid concentration. The amount of entrapped liposomes also depends on the size of prepared unilamellar lipid vesicles, whereby
180
/ Lee and Aguilar
higher amounts of vesicles could be entrapped in gel beads when large unilamelar vesicles with average diameter of 230 nm were used. However, this preparative method resulted in a material which exhibited short-term stability. Freeze–thawing a mixture of suspended biomembranes and dried gel beads was developed as an alternative way to induce fusion and entrapment of liposomes in gel beads, such as Sephacryl S-1000 and Superdex 200 [16,17]. The amount of immobilized liposomes was dependent on the temperature of the freeze–thaw cycle and the dimensions of the tube. Other methods involve freeze-drying the liposomes together with gel beads followed by the rehydration of the dried liposomes–gel beads or reverse-phase evaporation of organic solvent in an aqueous emulsion of lipids with gel beads [16] or by dialysis-preparation of (proteo)liposomes in beads [14,18]. As shown in Fig. 2A, noncovalent adsorption of liposomes has also been achieved on the hydrophobic alkyl chain such as C4, C8, and C18 derivatized gel beads such as Sephacryl S-1000 [19,20]. The alkyl derivatized Sephacryl S-1000 exhibited higher ligand density than the Sepharose adsorbents. Small, medium, and large vesicles of radii of approximately 20, 50, and 100 nm were immobilized onto
Fig. 2 Schematic illustration of the noncovalent immobilisation of liposomes onto a surface through (A) hydrophobic binding to covalently bound alkyl chains, or (B) specific interaction between the functional head groups of the phospholipids and activation groups on the surface of the solid support.
Biomembrane Chromatography / 181 octylsulfide-Sephacryl S-1000 in amounts corresponding to 110, 40, and 20µmol of phospholipids per milliliter of gel [19]. Although high lipid loading was found with the use of small unilamellar vesicles, partial loss of vesicle integrity was found during the adsorption process, as is evident from the release of calcein entrapped in the vesicles during the adsorption process. From the determination of entrapped calcein, radii of 20 and 50 nm were estimated for the adsorbed small and medium vesicles, respectively. Phospholipid analogs functionalized with carboxyl groups at the ω end of the acyl chain have also been immobilized onto modified substrates, such as AH-Sepharose 6B and AH-Sepharose 4B [21– 23]. However, the lipid density after the covalent immobilization onto the soft gel matrices was low and unlikely to form a bonded monolayer. Thus, this type of immobilized lipids on soft gel provided an affinity matrix for the purification of various lipases and enzymes involved in the biosynthesis of lipids. Modification of the lipid head groups with various functional groups can be used to immobilize the lipid vesicles either noncovalently or covalently onto the activated surface of gel beads (Fig. 2B). The binding of biotin to avidin or streptavidin is one of the alternative approaches to adsorb liposomes onto solid surfaces noncovalently. In this method, the introduction of biotin into the head group of phospholipids provides an anchor group to gel beads modified with avidin or streptoavidin [24]. Liposomes bearing photoreactive lipids have been bound to gel beads by the freeze–thaw/entrapment method followed by irradiation. The resulting liposomes were more stable than the entrapped liposomes during chromatography of basic proteins under gradient elution conditions [16]. The covalent immobilization of liposomes onto the surfaces of gel beads has also been achieved though the incorporation of amino-containing lipids, such as phosphatidylethanolamine into phosphatidylcholine-based liposomes. In this approach, the surfaces of gel beads were activated with amine-reactive groups, such as chloroformate, CNBr, hydroxysuccinimide, and tresyl groups [25]. However, liposomes without the amino-containing lipids, such as phosphatidylcholine and phosphatidylglycerol, can also be covalently immobilized onto the chloroformate-activated gel through a phosphoester linkage between the phosphate moiety of the phospholipid and the carbonate moiety of the active group by nucleophilic phosphate catalysis. The covalently immobilized liposomes were highly stable against osmotic effects
182
/ Lee and Aguilar
such as the presence of guanidinium chloride at concentrations up to 5 M. However, the covalently immobilized liposomes were permeable to the entrapped hydrophilic molecules upon storage as a result of membrane defects caused by the change in packing geometry of the coupled lipids in the membranes. Nevertheless, this immobilization method provided a simple, economical, and less time-consuming process than the preparation of the avidin-biotin immobilized liposomes.
III. BIOMEMBRANE-MODIFIED SILICA-BASED CHROMATOGRAPHIC SUPPORTS The development and use of microporous silica has been central to the enormous success of HPLC. This has been a result of the ability to prepare rigid and highly regular spherical silica-based chromatographic particles of well-defined particle size and porosity. A range of procedures has been developed for the immobilization of modified phospholipids onto silica to produce either monolayer or bilayer systems as shown in Fig. 3. RP-HPLC (Fig. 3A) in particular has been shown to be a sensitive analytical tool for the study of peptide and protein conformations at aqueous/lipid interfaces [3–8,26,27]. The hydrocarbon groups of RP-HPLC materials have been shown to induce secondary structures, which results in variations in the retention behavior of sequence-related peptides [3,4,28]. The sensitivity of the RP-HPLC systems lies in the fact that peptides and proteins interact with these surfaces in an orientation-specific manner via a specific hydrophobic contact area. Changes in this hydrophobic contact area can occur as a result of conformational or orientational changes, which, in turn, alter the binding properties of the peptides. Thus, the retention time of peptide and protein solutes represent a physical parameter which is highly sensitive to the conformational status of the peptides and proteins upon interaction with the hydrophobic surface. As such, RP-HPLC provided the first membrane mimetic system that was used to investigate the binding behavior of peptides and proteins with a hydrophobic lipid-like surface. Silica has now also been used to prepare a wide range of immobilized membrane systems which have been applied to a wide range of analytical studies associated with peptide– and protein–membrane interactions using the analytical procedures recently developed for RPHPLC.
Biomembrane Chromatography / 183
Fig. 3 Schematic representation of different biomembrane chromatographic systems. The silica particles can be modified with (A) long alkyl chains (i.e., reversed-phase HPLC), (B) immobilized lipid monolayers, (C) adsorbed lipid bilayers, and (D) hybrid lipid bilayers.
A. Covalently Attached Phospholipid Monolayers Membrane lipids have been covalently attached onto silica-based chromatographic supports via the reactive groups coupled at the end of the acyl chain. A wide range of phospholipids modified with reactive groups at the end of acyl chains is shown in Fig. 4. The immobilization of these types of modified phospholipids onto the surface of chromatographic supports potentially mimics the physicochemical properties of biomembrane surfaces, since the head groups are exposed to the aqueous environment. The reaction of carboxyl groups attached at the ω end of phospholipids with an aminosilane on the silica surface resulted in the formation of an immobilized lipid monolayer on the surface of chromatographic supports [29] (Fig. 3B). In addition to the use of car-
184
/ Lee and Aguilar
Fig. 4 Various synthetic phospholipids used for the preparation of biomembrane chromatographic supports. The functional moiety introduced at the ω end of the sn-1, sn-2, or both acyl chains include (A) thiol, (B) carboxyl, and (E-H) amino groups. The lipids are immobilized onto activated chromatographic supports through either one or two chains.
Biomembrane Chromatography / 185 boxyl groups as the functionalized moiety, the synthesis of a new class of immobilizable glycerophospholipids containing amino groups at the ω terminus of each of the sn-1 and sn-2 acyl chains has also been reported [11,12,30]. In addition to the immobilizable phosphatidylcholine, other lipid analogs, such as phosphatidylglycerol, phosphatidylethanolamine, phosphatidylserine, and phosphatidic acids with different charges, size, and hydration properties have also been synthesized [31,32]. The methods used for the preparation of the immobilized glycerophospholipid monolayer generally consist of three sequential steps (1) the activation step, in which the silica is pretreated and subsequently reacted with an organosilane with defined functionality whereby it is covalently bound to the silica; (2) the coupling step, in which the activated silica is subjected to reaction with the biomimetic ligands under gentle conditions to couple the ligands; and (3) the end-blocking step, in which the residual activating groups from the coupling steps are removed by means of appropriate reactions. Thus, immobilization of lipid analogs either as a single species or a mixture in different proportions onto the activated spherical chromatographic supports therefore provide a series of model biomembranes which can closely mimic the lipid compositions found in naturally occurring membranes. Lipid monolayers covalently immobilized onto silica exhibit long-term stability under aqueous conditions and also in the presence of organic solvent [11,12,29]. In addition, the salt type, ionic strength, pH, and organic solvent composition in the mobile phase can be optimized to provide baseline separation of membrane–protein mixtures [33]. However, perfusion of an immobilized biomembrane column with 0.1% TFA/H 2O, 0.1% TFA/acetonitrile, 35 mM citric acid, 35 mM NH 4H 2PO 4 , and 35 mM NH 4 Cl resulted in approximately 2% of the immobilized lipid molecules leaching from the column. However, as the residual reactive groups on the surfaces are blocked with small chemical compounds, the leaching of lipids from the column is greatly diminished under slightly acidic conditions and when stored in detergent solution for long periods of time. Although the lipid density on the silica particles is similar to that found in natural biomembranes, an important disadvantage associated with the immobilized lipid monolayer is the lack of lipid molecular dynamics, such as lateral diffusion, flip-flop, and axial displacement, due to the covalent linkage of the lipid molecules to the
186
/ Lee and Aguilar
silica surface. However, the lack of lipid dynamics associated with the fluid biomembrane surfaces is compensated by the increased stability derived from the covalent immobilization of lipids onto solid surfaces. Thus, the immobilized phospholipid monolayers on the chromatographic supports have unique applications compared to the immobilized liposomes. In particular, organic solvents and detergents can be included in the mobile phase during the chromatographic separations and interactive studies. The interaction of solutes with the immobilized biomembrane surfaces exhibits a mixed-mode mechanism including the partition of solutes into the hydrocarbon region, adsorption to the polar headgroups and glycerol backbone, and ionic interaction with the charged headgroups. Compared to the reversed-phase chromatographic surfaces, large molecules predominantly adsorb via hydrophobic interaction to the alkyl chains immobilized to silica. In addition, shorter retention times were found for solutes interacting with the immobilized lipid monolayer than with reversed-phase chromatographic materials [11,12,33]
B.
Adsorbed Monolayers and Bilayers
Studies have also demonstrated that phospholipid bilayers adsorbed onto solid supports (Fig. 3C) can exhibit a similar behavior with respect to lateral diffusion and showed a phase transition similar to that of classical model membrane systems [34–36]. The adsorption of phospholipids onto polystyrene–divinylbenzene beads formed a bilayer-like structure that was employed as a model membrane system for the assessment of membrane function [37]. The dynamic properties of the lipid molecules were retained in a single bilayer of either a single species or a binary mixture of phospholipids adsorbed ˚ , diameter 10 ⫾ 1.5 µm) [38]. onto silica particles (pore size 4000 A Transition endotherms measured by differential scanning microcalorimetry revealed that the lipid bilayer systems that consisted of two lipid species on the chromatographic support exhibited a rather broad and complex feature consistent with partial demixing of two lipids in the gel phase. In contrast, the endotherm for adsorbed bilayers with a single lipid species showed specific heat changes consistent with the presence of a single phospholipid species. In addition, the transition temperatures for the single and binary lipid bilayers were also different. These unique membrane-like thermo-
Biomembrane Chromatography / 187 dynamic properties were then exploited to provide a selective purification system of the peripheral membrane protein, myristoylated alanine-rich C kinase substrate (MARCKS)-related protein [39]. This bilayer-adsorbed chromatographic system thus provided charge-selective protein separation under biocompatible conditions solely by changes in the column temperature without changes in ionic strength, which allowed the native conformation of the protein to be maintained during the purification. In addition to the adsorbed bilayer, the perfusion of phospholipids dissolved in organic solvent (85% isopropanol) into reversedphase chromatographic supports resulted in a hybrid membrane bilayer in which a monolayer of lipids is adsorbed noncovalently onto the hydrophobic alkyl chains (Fig. 4D) [40]. The amount of adsorbed phospholipids was found to be similar to that of lipid vesicles, thereby confirming the formation of a bilayer-like structure. This system was stable in solvents containing less that 35% acetonitrile and was used to study peptide–lipid binding properties.
IV.
APPLICATIONS
The applications in the use of the immobilized lipid systems described in the previous sections can be divided into four main classes. The first category involves the use of these systems for the improved isolation of membrane proteins. The remaining applications relate to various biophysical studies of drug–membrane partitioning, protein–ligand interactions, and the binding of peptides and proteins with the model membrane surface.
A. Protein Purification While the majority of studies describing the synthesis of immobilized lipid systems have been published in the last 10 years, the potential of these systems in the purification of proteins was initially recognized over 20 years ago. In these early studies, affinity columns prepared by the immobilization of phospholipid derivatives onto Sepharose 4B were used to purify phospholipase A 2 (PLA 2 ) [21], phosphatidylglycerol phosphate synthetase [41], and phosphatidylcholine exchange protein [22]. Protein kinase C–phorbol ester receptor was also purified on a polyacrylamide-immobilized phosphatidyl-
188
/ Lee and Aguilar
serine [42]. The use of soft gels has continued more recently with the development of an affinity column for PLA 2 based on the immobilization of acylaminophospholipid analogs onto Sepharose 6B [23]. This column also demonstrated the specificity of interaction between the protein and the ligand, since PLA 2 bound only to the (S) analog and not to the (R) analog. The silica-based immobilized artificial membrane columns developed by Pigeon and co-workers have been applied to the purification of cytochrome P450 isozymes [43], cholesterol-binding protein [44], membrane-bound N-acylphosphatidylethanolamine synthase [45], and PLA 2 [33,46]. In general, it has been observed that these columns exhibit intermediate retention between that observed in reversed-phase chromatography and hydrophobic interaction chromatography indicative of an intermediate hydrophobic character which is potentially advantageous in terms of recovering protein in a biologically active form.
B.
Bioaffinity Measurements
The use of affinity chromatography in the analysis of biomolecular interactions has provided significant insight into a number of biorecognition processes. The immobilized membrane materials also provide an experimental system in which membrane proteins can be stabilized for use in protein–ligand or enzyme–substrate interactions. For example, the enzyme urease was covalently attached to phospholipid bound silica, tungsten, and fluoropolymer surfaces [47,48]. It was found that the enzyme exhibited enhanced stability on these materials compared to when it was immobilized onto an alkyl-modified silica surface. These results indicated that the immobilized lipid surfaces represented a useful mimic of the membrane surface with potential for on-line monitoring of urease. The differential hydrolysis of immobilized phosphatidylcholines by PLA 2 and phospholipase C was also shown [49]. Small unilamellar vesicles composed of phosphatidylcholine and phosphatidylethanolamine derivatives have been covalently coupled to Superdex 200 beads and used to study protein–membrane interactions and the refolding of bovine carbonic anhydrase [50]. The results indicated that the degree of retention of partially denatured proteins on the immobilized liposome column correlated with the solute hydrophobicity determined by aqueous two-phase partitioning.
Biomembrane Chromatography / 189 In addition, chromatographic refolding of unfolded carbonic anhydrase was also achieved, with up to 83% of enzymatic activity recovered after passage through the column. Silica-based immobilized artificial membranes have also been used to immobilize the enzymes chymotrypsin and trypsin [51]. The enzymes were trapped in the hydrophobic cavities rather than attached covalently, and these studies demonstrated the potential use of these systems in the analysis of enzyme–substrate and enzyme–inhibitor interactions. The sterically immobilized liposome chromatographic supports developed by Lundahl and co-workers have been used to study the functional properties of the glucose transporter protein Glut 1 derived from human red blood cells [13,16,52–54]. They demonstrated a difference in retention times for D- and L-glucose and were able to study the effect of pH on the activity of Glut 1 which provided insight into the mechanism by which glucose is transported in terms of the likely amino acids involved [52]. Additional mechanistic information was also obtained by studying the affinities of Glut 1 inhibitors forskolin and cytochalasin B as a function of pH [53] and temperature [54] using frontal affinity chromatography. The activity of Glut 1 in reconstituted proteoliposomes was also compared to the activity obtained when human red cells were immobilized directly into the gel particles [13,53]. The results showed that dissociation constants for the glucose-displaceable binding of forskolin more closely compared with published values on the red blood cells than on the reconstituted liposomes and demonstrated the potential of these systems for studying bioaffinity interactions [55–57].
C. Drug Partition Measurements Immobilized membrane surfaces also have enormous potential to provide quantitative information on solute–membrane interactions. Drug bioactivity, toxicity, and distribution all depend on drug absorption, which often depends on the partition of the drug into a membrane. Determination of the membrane partition coefficient, Km , is therefore a necessary step in drug design and development in terms of the overall physicochemical characterization of drug compounds. Immobilized membrane surfaces have thus been particularly useful in the quantitative analysis of drug–membrane partition coefficients [58–64], since chromatographic retention times can
190
/ Lee and Aguilar
be readily obtained and used to derive solute–membrane Km values according to k′ ⫽ ΦK m
(1)
where Φ is the phase ratio of the chromatographic material and is the ratio of Vs /Vm , where Vs is the volume of the interphase created by the immobilized lipid ligands and Vm is the total volume of the solvent within the column. A number of studies have reported the measurement of membrane partition coefficients for a wide range of classes of drugs. These include phenethylamine derivatives [60], α-adrenoreceptor agonists [58], a homologous set of cephalosporins [59], β-blockers [61,65], imidazoline and imidazolidine derivatives [61], nonsteroidal anti-inflammatory drugs [66], a series of 17 structurally diverse drugs [67], a set of acidic, basic, and neutral drugs from various structural classes [68], bile salts [69], a series of highly lipophilic drugs [70], neutral and ionized acids and bases [71], and a set of HIV protease inhibitors [72]. A series of 15 basic drugs was also used to investigate the partition properties of unilamellar liposomes covalently immobilized to a TSK G6000PW gel [73]. In almost all cases, the membrane partition coefficients were compared with partition data derived from either the corresponding octanol–water partition lipophilicity parameters, which has been the more conventional method for predicting drug adsorption properties [61,66,67], liposome partitioning [61], or an in-vitro drug adsorption assay [59]. The degree of correlation between the chromatographically determined values and the octanol–water coefficients varied depending on the study, which may be related to the degree of structural homology within each set of test solutes. In contrast, high correlations between the immobilized membrane coefficients and the liposome partition values or the in-vitro assays were observed in all cases. These studies therefore demonstrate the ability of these immobilized chromatographic materials to be used in the prediction of bioavailability and hence in drug design and development.
D. Peptide– and Protein–Membrane Interactions Peptide– and protein–lipid interactions play a critical role in the regulation of several biological processes, including the insertion
Biomembrane Chromatography / 191 and folding of membrane proteins, the translocation of polypeptides through membranes, and the cytolytic action of antimicrobial peptides. In contrast to the small, low-molecular-weight compounds such as those described in the previous section, which do not undergo any structural changes upon binding to the membrane, the interaction of peptides and proteins with membranes is generally associated with the induction of specific secondary structure. The ability to measure the free energy of interaction of peptides and proteins with membrane surfaces as a function of conformation of both the peptide or protein and the lipid is therefore an important part of characterizing the molecular basis of peptide– and protein–membrane interactions. The immobilized phospholipid monolayers therefore can act as a membrane mimic and be used to study the binding properties of peptides and proteins during interaction with membranes. The noncovalent immobilized artificial membrane surface has been used to study the interaction of a set of helical antibacterial magainin-2-amide peptides [40]. The results demonstrated that the retention of each peptide correlated well with lipid affinities determined by circular dichroism and further illustrates the potential of immobilized lipid systems for studying peptide–lipid interactions. The immobilized model membrane prepared through the covalent attachment of an amino-modified phosphatidylcholine to activated silica has been used to study the interaction of a series of peptide hormones, bombesin, β-endorphin, and glucagon, with the model membrane surface [11]. Dynamic elution techniques were used to measure the influence of temperature and methanol on the binding of these peptides. For all three peptides, nonlinear retention plots of ln k′ versus temperature were observed between 5 and 55°C (Fig. 5). These results contrasted with the simple linear plots observed for the small unstructured molecules N-acetyltryptophanamide and diphenylalanine. Similar results were also obtained with the cytolytic peptide melittin and a series of truncated melittin analogs in which the positively charged C-terminal residues were removed [12,30]. In addition, subtle differences between the analogs indicated that the peptides adopted different conformations and orientation in the lipid monolayer. An important feature of interactive chromatographic systems is the ability to derive thermodynamic information which can give fur-
192
/ Lee and Aguilar
Fig. 5 Plots of log k′ versus percent methanol for (A) diphenylalanine and (C) bombesin and plots of log k′ versus temperature for (B) diphenylalanine and (D) bombesin following elution from the immobilized phosphatidylcholine monolayer at temperatures between 5 and 65°C [(■) 5°C, (䊐) 15°C, (䉱) 25°C, (䉭) 35°C, (䊉) 45°C, (䊊) 55°C, (◆) 65°C] and at the following methanol concentrations: (■) 0%, (䊐) 10%, (䉱) 20%, (䉭) 30%, (䊉) 40% (䊊) 50%. (Adapted with permission from Ref. 11.)
ther insight into the interactive process. The k′ value can be related to the peptide–lipid association constant, Ka , through the expression Ka ⫽
Cs ⫽ k′Φ Cm
(2)
where Cs and Cm are the concentration of the solute in the stationary phase and the mobile phase, respectively, and Φ is the phase ratio equal to Vs /Vm . More significant, the standard unitary free energy of interaction with the immobilized lipid surface, ∆G 0assoc can then be obtained from
Biomembrane Chromatography / 193 ∆G 0assoc ⫽ ⫺RT ln K a
(3)
Hence, peptide retention can be related directly to the overall standard free energy of interaction ∆G 0assoc according to ln k′ ⫽ ln Φ ⫺
∆G 0assoc RT
(4)
The experimentally determined k′ values therefore represent a measure of the total change in free energy associated with the binding of a peptide to the lipid surface. Using the dynamic elution techniques, the dependence of ln k′ values on temperature can then be used to derive ∆H 0assoc and ∆S 0assoc according to ln k′ ⫽ ⫺
∆H 0assoc ∆S 0assoc ⫹ ⫹ ln Φ RT R
(5)
When enthalpy and entropy changes are independent of temperature, plots of ln k′ versus 1/T, i.e., van’t Hoff plots, are linear, and values of ∆H 0assoc ∆S 0assoc can be determined according to Eq. (5). In the case of nonlinear van’t Hoff plots, which indicates a strong dependence of ∆H 0assoc and ∆S 0assoc on temperature, a second-order quadratic expression can be used [74,75] to evaluate thermodynamic data as follows: ln k′ ⫽ a ⫹
b c ⫹ 2 ⫹ ln Φ T T
(6)
where ∆H 0 assoc is given by ∆H 0assoc ⫽ ⫺ R
2c d ln k′ ⫽ ⫺R(b ⫹ ) d(1/T) T
(7)
and ∆S 0assoc is equal to ∆S 0assoc ⫽ R(a ⫺
c ⫺ ln Φ) T2
(8)
Thus, elution techniques can be used to derive important thermodynamic data associated with peptide–lipid interactions as a function of both peptide and lipid structure. Calculation of ∆H 0assoc and ∆S 0assoc allows the relative contribution of each parameter to the overall ∆G 0assoc to be assessed from comparison of the magnitude of the ∆H 0asoc value with the T∆S 0assoc term [76].
194
/ Lee and Aguilar
In particular, it is generally accepted that classical hydrophobic interactions are entropy-driven at low temperatures and become increasingly enthalpy-driven at high temperatures [77]. However, nonclassical hydrophobic effects have been observed for the interaction of small molecules and peptides with lipid bilayers in which the hydrophobic interactions are enthalpy-driven at room temperature [78,79]. Calculation of the thermodynamic parameters of free energy, enthalpy, and entropy of interaction for bombesin, β-endorphin, and glucagon revealed that the binding of all peptides was enthalpy-driven, that is, they are mediated by nonclassical hydrophobic interactions [11], a result that is not readily obtained from other techniques used to measure peptide–lipid binding. The ability to derive ∆H 0assoc and ∆S 0assoc values for the interaction of peptides and proteins with lipids by chromatographic techniques therefore provides important information on the driving forces underlying peptide–and protein–membrane interactions. The dynamic behavior of the interaction of peptides with the immobilized phospholipid surface can be further characterized through analysis of the elution peak widths [4,5,80]. For low-molecularweight, conformationally rigid molecules that interact through a single binding site and with a unique orientation in the lipid monolayer, the experimentally observed elution peak will appear as a sharp Gaussian–(or near-Gaussian)-shaped peak. However, there are two major factors that can lead to atypical band broadening in the elution profile, namely, changes in peptide conformation and changes in the degree of penetration of the peptide into the lipid monolayer. Significant band broadening and asymmetric peak splitting was observed (Fig. 6) for bombesin, β-endorphin, and glucagon [11] and melittin [12], which is consistent with the formation of multiple conformational species during the interaction with the model membrane layer. Overall, these increases in bandwidth correlated with the observed fluctuations in retention behavior for these molecules, indicating that the conformational changes which give rise to the changes in retention times occur on a time scale which is equivalent to the separation time. Moreover, the results demonstrate that subtle differences in both structure and orientation of the peptide at the lipid surface can be monitored by these elution techniques. A large number of membrane-active peptides bind more strongly to anionic phospholipids, and the ability to monitor the relative affinity of peptides for different lipids is critical to establishing the
Biomembrane Chromatography / 195
Fig. 6 Elution profiles for (a) N-acetyltryptophanamide, (b) bombesin, (c) β-endorphin, and (d) glucagon obtained at 35°C with an immobilized phosphatidylcholine monolayer at different methanol concentrations. See text for other conditions. (Reprinted with permission from Ref. 11.)
molecular basis of specificity in such areas an antimicrobial peptide action [81]. Immobilized phospholipid monolayers have been prepared using the anionic lipids phosphatidylglycerol (PG) and phosphatidic acid (PA), and these model membranes were used to study the membrane interaction of melittin. Interaction studies clearly indicated a higher retention of melittin for PG and PA compared to PC [82]. These results demonstrated that the negative charge on the lipid head groups significantly affect the binding affinity of melittin to the immobilized monolayer. Such results are in agreement with previous findings in which higher affinity of melittin to the anionic membrane lipids was related to the depth of the peptides into the regions comprised of backbone glycerol and ester carbonyl groups.
196
/ Lee and Aguilar
The electrostatic interaction between the basic residues at the peptide C terminus and the negatively charged phosphate head groups further increased the affinity of melittin for the lipid molecules. Overall, these biophysical studies using chromatographic techniques have clearly demonstrated that these biomimetic monolayers provide a stable and sensitive system with which to explore the molecular mechanism of peptide conformational changes during membrane interactions.
V.
CONCLUSIONS AND FUTURE DIRECTIONS
The pivotal role of HPLC techniques in the enormous advances in biomedical sciences over the last 30 years continues unabated. In this chapter, the recent advances in novel immobilization chemistries to produce covalently and noncovalently attached phospholipids have been reviewed. The combination of these new materials with modern analytical techniques has provided a wide range of immobilized model membrane systems which can be used to provide new procedures for protein purification but also represent a new experimental approach to studying the biophysical basis of peptide– and protein–lipid interactions. The sensitivity of the immobilized lipid systems, particularly the covalently attached silica-based materials, lies in the fact that peptides and proteins interact with these surfaces in an orientationspecific manner via a specific contact regions as shown in Fig. 7. Changes in this contact area can occur as a result of conformational, orientational, or insertional changes, which, in turn, alter the binding properties of the solutes. Thus, the retention times and the experimental bandwidth of peptides and proteins are physical parameters which are highly sensitive to the conformational and insertional status of the peptides upon interaction with the immobilized phospholipid surface. Future developments in the immobilization of mixed phospholipid monolayers and phospholipid bilayers will expand the potential of biomembrane chromatography in analytical and biophysical applications. Biomembrane chromatographic techniques for the analysis of peptide–lipid interactions represents a powerful approach to studying the influence of differences in peptide and lipid structure on the folding/unfolding transitions of peptides during membrane interactions. Biomembrane chromatography is thus an ideal bio-
Biomembrane Chromatography / 197
Fig. 7 Schematic illustration of the interaction of melittin with an immobilized phosphatidylcholine monolayer in aqueous conditions at temperatures where (A) the lipid and/or melittin do not undergo conformational interconversions, and (B) interconversions are present. The corresponding elution profiles observed for melittin are also shown. (Reprinted with permission from Ref. 81.)
affinity system which is able to mimic biological membrane environments and provide a powerful complementary technique to emerging biosensor systems which monitor biomolecular interactions in a membrane environment.
REFERENCES 1. Meth. Enzymol., 270 and 271 (1996), entire volumes. 2. A. Jaulmes and C. Vidal-Madjar, in C. J. Giddings, E. Grushka, and P. R. Brown, (Eds.), Advances in Chromatography, Vol. 28, Marcel Dekker, New York, 1989, p. 1. 3. E. Lazoura, I. Maidonis, E. Bayer, M. T. W. Hearn, and M. I. Aguilar, Biophys J., 72: 238 (1997).
198
/ Lee and Aguilar
4. T.-H. Lee, P. E. Thompson, M. T. W. Hearn, and M. I. Aguilar, J. Peptide Res., 49: 394 (1997). 5. D. L. Steer, P. E. Thompson, S. E. Blondelle, R. A. Houghten, and M. I. Aguilar, J. Peptide Res., 51: 401 (1998). 6. R. S. Hodges, B.-Y. Zhu, N. E. Zhou, and C. T. Mant, J. Chromatogr., 676: 3 (1994). 7. S. E. Blondelle and R. A. Houghten, Biochemistry, 31: 12688 (1992). 8. E. Krausse, M. Beyermann, M. Dathe, S. Rothemund, and M. Bienert, Anal. Chem., 34: 252 (1995). 9. A. Lundqvist and P. Lundahl, J. Chromatogr. A, 699: 209 (1997). 10. C. Pigeon, C. Marcus, and F. Alvarez, in J. W. Kelly and T. O. Baldwin (Eds.), Applications of Enzyme Biotechnology, Plenum Press, New York, 1991, p. 201. 11. H. Mozsolits, T.-H. Lee, H.-J. Wirth, P. Perlmutter, and M.-I. Aguilar, Biophys. J., 77: 1428 (1999). 12. T.-H. Lee, D. Rivett, J. Werkmeister, D. Hewish, and M.-I. Aguilar, Lett. Peptide Sci., 5: 1 (1999). 13. C.-M. Zeng, Y. Zhang, L. Lu, E. Brekkan, A. Lundqvist, and P. Lundahl, Biochim. Biophys. Acta, 1325: 91 (1997). 14. M. Wallsten, Q. Yang, and P. Lundahl, Biochim. Biophys. Acta, 982: 47 (1989). 15. E. Brekkan, L. Lu, and P. Lundahl, J. Chromatogr. A, 711: 33 (1995). 16. Q. Yang and P. Lundahl, Anal. Biochem., 218: 210 (1994). 17. A. Lundqvist, G. Ocklind, L. Haneskog, and P. Lundahl, J. Mol. Recognit., 11: 52 (1998). 18. Q. Yang, M. Wallsten, and P. Lundahl, J. Chromatogr., 506: 379 (1990). 19. Q. Yang, M. Wallsten, and P. Lundahl, Biochim. Biophys. Acta., 938: 243 (1988). 20. Y. Zhang, C. M. Zeng, Y. M. Li, S. Hjerten, and P. Lundahl, J. Chromatogr. A, 749: 13 (1996). 21. C. O. Rock and F. Snyder, J. Biol. Chem., 250: 6564, (1975). 22. L. I. Barsukov, C. W. Dam, L. D. Bergelson, G. I. Muzja, and K. W. A. Wirtz, Biochim. Biophys. Acta., 513: 198 (1978). 23. R. Dijkman, S. H. W. Beiboer, and H. M. Verheij, Biochim. Biophys. Acta., 1347: 1 (1997).
Biomembrane Chromatography / 199 24. Q. Yang, X. Y. Liu, S. Ajiki, M. Hara, P. Lundahl, and J. Miyake, J. Chromatogr. B, 707: 131 (1998). 25. Q. Yang, X.-Y. Liu, M. Yoshimoto, R. Kuboi, and J. Miyake, Anal. Biochem., 268: 354 (1999). 26. K. L. Richards, M. I. Aguilar, and M. T. W. Hearn, J. Chromatogr., 676: 17 (1994). 27. D. Clayton, P. Holt, V. Kronina, R. Boysen, A. W. Purcell, M. T. W. Hearn, and M. I. Aguilar, Anal. Chem., 70: 5010 (1999). 28. S. E. Blondelle, K. Buttner, and R. A. Houghten, J. Chromatogr., 625: 199 (1992). 29. C. Pigeon and U. V. Venkataram, Anal. Biochem., 176: 36 (1989). 30. T.-H. Lee, H.-J. Wirth, P. Perlmutter, D. Rivett, and M.-I. Aguilar, in S. Bajusz and F. Hudecz (Eds.), Peptides 1998, Acade´miai Kiado´, Budapest, 1999, p. 782. 31. X. Qiu, S. Ong, C. Bernal, D. Rhee, and C. Pidgeon, C. J. Org. Chem., 59: 537 (1994). 32. T.-H. Lee, H.-J. Wirth, A. W. A. Clayton, W. H. Sawyer, P. Perlmutter, and M. I. Aguilar, submitted (2000). 33. C. Pidgeon, S. J. Cai, and C. Bernal, J. Chromatogr. A, 721: 213 (1996). 34. T. Ko¨chy and T. M. Bayerl, Phys. Rev. E, 47: 2109 (1993). 35. C. Naumann, T. Brumm, and T. M. Bayerl, Biophys. J., 63: 1314 (1992). 36. S. J. Johnson, T. M. Bayerl, D. C. McDermott, G. W. Adam, A. R. Rennie, R. K. Thomas, and E. Sackmann, Biophys. J., 59: 289 (1991). 37. G. S. Retzinger, S. C. Meredith, S. H. Lau, E. T. Kaiser, and F. J. Kezdy, Anal. Biochem., 150: 131 (1985). 38. A. Loidl-Stahlhofen, S. Kaufmann, T. Braunschweig, and T. Bayerl, Nature Biotechnol., 14: 999 (1996). 39. A. A. P. Schmitz, E. Schleiff, C. Ro¨hring, A. Loidl-Stahlhofen, and G. Verge`res, Anal. Biochem., 268: 343 (1999). 40. E. Krausse, M. Dathe, T. Wieprecht, and M. Beinert, J. Chromatogr. A., 849: 125 (1999). 41. T. J. Larson, T. Hirabayashi, and W. Dowhan, Biochemistry, 15: 974 (1976). 42. T. Uchida and C. R. Filburn, J. Biol. Chem., 259: 12311, (1984).
200
/ Lee and Aguilar
43.
C. Pidgeon, J. Stevens, S. Otto, C. Jefcoate, and C. Marcus, Anal. Biochem., 194: 163 (1991). H. Thurnhofer, J. Schnabel, M. Betz, G. Lipka, C. Pigeon, and H. Hauser, Biochim. Biophys. Acta., 1064: 275 (1991). S. Cai, R. McAndrew, B. P. Leonard, K. D. Chapman, and C. Pigeon, J. Chromatogr. A, 696: 49 (1995). C. Bernal and C. Pidgeon, C. J. Chromatogr. A, 731: 139 (1996). K. M. R. Kallury, W. E. Lee, and M. Thompson, Anal. Chem., 64: 1062 (1992). K. M. R. Kallury, W. E. Lee, and M. Thompson, Anal. Chem., 65: 2459 (1993). J. M. Delfino, J. Flori´n-Christensen, M. Flori´n-Christensen, and F. M. Richards, Biochem. Biophys. Res. Commun., 205: 113 (1994). M. Yoshimoto, R. Kuboi, Q. Yang, and J. Miyake, J. Chromatogr. B, 712: 59 (1998). W.-K. Chui and I. Wainer, Anal. Biochem., 201: 237 (1992). L. Lu, E. Brekkan, L. Haneskog, Q. Yang, and P. Lundahl, Biochim. Biophys. Acta., 1150: 135 (1993). L. Lu, A. Lundqvist, C.-M. Zeng, C. Lagerquist, and P. Lundahl, J. Chromatogr. A, 776: 81 (1997). A. Lundqvist and P. Lundahl, J. Chromatogr. A, 776: 87 (1997). L. Haneskog, A. Lundqvist, and P. Lundahl, J. Mol. Recognit., 11: 58 (1998). L. Haneskog, C. M. Zeng, A. Lundqvist, and P. Lundahl, Biochim. Biophys. Acta, 137: 1 (1998). Y. Zhang, S. Aimoto, L. Lu, Q. Yang, and P. Lundahl, Anal. Biochem., 229: 291 (1995). C. Pidgeon, S. Ong, H. Choi, and H. Liu, Anal. Chem., 66: 2701 (1994). C. Pidgeon, S. Ong, H. Liu, X. Qiu, M. Pidgeon, A. H. Dantzig, J. Munroe, W. J. Hornback, J. S. Kasher, L. Glunz, and T. Szczerba, J. Med. Chem., 38: 590 (1995). S. Ong, H. Liu, X. Qui, G. Bhat, and C. Pidgeon, Anal. Chem., 67: 755 (1995). S. Ong, H. Liu, and C. Pidgeon, J. Chromatogr. A, 728: 113 (1996). F. Barbato, B. Cappello, A. Miro, M. I. La Rotonda, and F. Quaglia, Farmaco, 53: 655 (1998).
44. 45. 46. 47. 48. 49.
50. 51. 52. 53. 54. 55. 56. 57. 58. 59.
60. 61. 62.
Biomembrane Chromatography / 201 63. G. W. Caldwell, J. A. Masucci, M. Evangelisto, and R. White, R. J. Chromatogr. A, 800: 161 (1998). 64. Y. Zhang, C. M. Zeng, Y. M. Li, S. Hjerten, and P. Lundahl, J. Chromatogr. A, 749: 13 (1996). 65. J. A. Masucci, G. W. Caldwell, and J. P. Foley, J. Chromatogr. A, 810: 95 (1998). 66. F. Barbato, M. I. La Rotonda, and F. Quaglia, J. Pharm. Sci., 86: 225 (1997). 67. P. Barton, A. M. Davis, D. J. McCarthy, and P. J. H. Webborn, J. Pharm. Sci., 86: 1034 (1997). 68. T. Salminen, A. Pulli, and J. Taskinen, J. Pharm. Biomed. Anal., 15: 469 (1997). 69. D. E. Cohen and M. R. Leonard, J. Lipid Res., 36: 2251 (1995). 70. F. Beigi and P. Lundahl, J. Chromatogr. A, 852: 313 (1999). 71. B. Ottiger and H. Wunderli-Allenspach, Pharm. Res., 16: 643 (1999). 72. B. H. Stewart, F. Y. Chung, B. Tait, C. J. Blankley, and O. H. Chan, Pharm. Res., 15: 1401 (1998). 73. Q. Yang, X. Y. Liu, K. Umetani, N. Kamo, and J. Miyake, Biochim. Biophys. Acta., 1417: 122 (1999). 74. D. A. Haidacher, A. Vailaya, and Cs. Horvath, Proc. Natl. Acad. Sci. (USA), 93: 2290 (1996). 75. L. A. Cole, J. G. Dorsey, and K. A. Dill, Anal. Chem., 64: 1324 (1992). 76. S. Ong and C. Pigeon, Anal. Chem., 67: 2119 (1995). 77. C. Tanford, The Hydrophobic Effect: Formation of Micelles and Biological Membranes, Wiley, New York, 1980. 78. G. Beschiaschvili and J. Seelig, Biochemistry, 31: 10044 (1992). 79. J. Seelig and P. Ganz, Biochemistry, 30: 9354 (1991). 80. A. W. Purcell, M. I. Aguilar, and M. T. W. Hearn, Anal. Chem., 65: 3038 (1993). 81. S. E. Blondelle, K Lohner, and M. I. Aguilar, Biochim. Biophys. Acta., 1462: 89 (1999). 82. T.-H. Lee, A. W. A. Clayton, W. H. Sawyer, and M. I. Aguilar, submitted.
6 Transformation of Analytes for Electrochemical Detection: A Review of Chemical and Physical Approaches Mark J. Rose,*,1,3 Susan M. Lunte,1,2,3 Robert G. Carlson,2,3 and John F. Stobaugh 1,2,3 1 Department of Pharmaceutical Chemistry, 2 Department of Chemistry, and 3 The Center for Bioanalytical Research, The University of Kansas, Lawrence, Kansas
I. ABBREVIATIONS II. INTRODUCTION A. Microseparation Techniques B. Detection Challenges in Microseparation Techniques C. Electrochemical Detection and Microseparation Techniques III. TRANSFORMATION METHODS AND SUMMARY TABLES A. Primary Amines B. Secondary Amines C. Peptides D. Thiols E. Alcohols
204 206 206 206 207 208 209 217 220 221 225
*Current affiliation: Department of Drug Metabolism, Merck Research Laboratories, West Point, Pennsylvania.
203
204
/ Rose et al.
F. Keto Compounds G. Carboxylic Acids H. Metals and Ions I. Miscellaneous Analytes IV. NDTE ACKNOWLEDGMENTS REFERENCES
I. ABBREVIATIONS AP APM ASC BBQ BDF BR/Cu(II) CA Ce(IV) Ox CLOD DEDC DFB/HPLChν-EC DHBH DM-PITC DMA/EDC
DMPM DNPHD DPBE DNPT DTC EX FA FBA FCAC
4-Aminoantipyrine N-(4-Anilinophenyl) maleimide o-Acetylsalicyloyl chloride 3,5-di-tert-Butyl-1,2-benzoquinone 3-Bromo-1,1′-dimethyl-ferrocene Biuret reagent [Cu(II)/tartrate] 2-Cyanoacetamide Ce(IV) oxidation Concentration limit of detection Diethyldithiocarbamate 2,4-Dinitrofluorobenzene and HPLChν-EC 2,5-Dihydroxybenzohydrazide p-N, N-Dimethylaminopheny-isothiocyanate 2,4-Dimethoxyaniline HCl and l-ethyl3(3-dimethylaminopropyl)carbodiimide N-(4-Dimethylaminophenyl)maleimide 2,4-Dinitrophenylhydrazine 1-(2,5-Dihydroxyphenyl)-2bromoethanone 3,6-Dinitrophthalic anhydride Dithiocarbamate Ethyl xanthate 3-Ferrocenyl acyl azide Ferroceneboronic acid Ferrocenyl acid chloride
227 229 232 234 237 240 240
Transformation of Analytes for Electrochemical Detection / 205 FEA/EDC
FEITC FITC FM FSC HEDC HPLC-hνEC IEMN/Pt
Me/CDS MLOD MP NBD-Cl N-HSP N-SEP NDA/CN⫺ NPHD NQS OPA OPA/BT OPA/MCE OPA/Sul PAP PAP/BMI PCD PHD PITC PNSA QL SA SAC
1-Ferrocenylethylamine and l-ethyl3(3-dimethylaminopropyl)carbodiimide 2-Ferrocenylethylisothiocyanate Ferrocenylisothiocyanate N-(Ferrocenyl) maleimide Ferrocenesulfonylchloride bis(2-Hydroxyethyl)dithiocarbamate HPLC and postcolumn UV irradiation with electrochemical detection 2-[2-(Isocyanate)ethyl]-3-methyl-1,4naphthoquinone and platinum reduction Metal/carbon disulfide Mass limit of detection 2-Methoxyphenol 7-Chloro-4-nitrobenzo-2-oxa-1,3diazole N-Hydroxysuccinimide phenol N-Succinimidyl 3-ferrocenylpropionate Naphthalene-2,3-dicarboxaldehyde/ CN⫺ 4-Nitrophenylhydrazine β-Naphthoquinone-4-sulfonate o-Phthaldehyde (OPA) o-Phthaldehyde (OPA)/tert-butylthiol o-Phthaldehyde/β-mercaptoethanol o-Phthaldehyde (OPA)/sulfite p-Aminophenol p-Aminophenol and 2-bromo-1-methylpyridinium iodide Pyrrolidinecarbodithoate Phenylhydrazine Phenylisothiocyanate 1-(2-Pyridylazo)-2-naphthol-6-sulfonic acid 8-Quinolinol Salicyaldehyde Salicylic acid chloride
206
/ Rose et al. TD-hν TNBSA TTAA Zn/HEDC
Titanium dioxide postcolumn photolysis 2,4,6-Trinitrobenzene-sulfonic acid (R,R)-O,O′-di-p-toluoyl tartartic acid anhydride Zinc bis[(2-hydroxyethyl)dithiocarbamate]
II.
INTRODUCTION
A.
Microseparation Techniques
The miniaturization of separation systems has received much attention in the scientific literature over the past 10–15 years. Progress has been made in providing reliable commercial microbore HPLC equipment and capillary electrophoresis (CE) instrumentation, and novel technology is currently being developed for microchip electrophoresis [1–11] and capillary electrochromatography (CEC) in packed columns [12–16]. An obvious advantage of using miniaturized separation technology is its small sample requirement, which has made possible studies such as the separation of the contents of single cells [17–21] or the on-line separation of in-vivo microdialysis samples [22]. Physical attributes such as efficient heat dissipation and flow dynamics are advantages of miniaturization and make it possible to perform efficient separations in the high electric fields encountered in CE and CEC.
B.
Detection Challenges in Microseparation Techniques
Over the past decade, there has been a continued trend in the miniaturization of various analytical techniques, especially in the area of liquid-phase separations; however, advances in the separation component of an instrument place serve demands on the associated detection module. The challenge comes from the fact that detector dimensions in miniaturized systems need to be proportionally downsized. For optically based detectors this results in a shortened path length and diminished sensitivity. When fluorescence detection is a viable option, a partial answer to this issue lies in the use of lasers as the excitation source. In general, laser-induced fluorescence (LIF) detection has provided for high-sensitivity detection of analytes in small-volume detection cells and has found application on numerous
Transformation of Analytes for Electrochemical Detection / 207 occasions in CE [19,23–28]. Despite such successes laser-based detectors are still limited by a lack of generality due to the few choices of available excitation wavelengths and the limited number of compounds that can be excited at these wavelengths. For home-made LIF detectors, other limitations often are related to operator training, and the general tediousness of setting up and optimally operating a laser-based detector on a daily basis. Many of these aspects are not of issue with commercially available systems that have recently become available. Unfortunately, from a practical standpoint, there is still a high initial start-up cost and limited lifetime for many of the lasers. This situation may change dramatically in the next decade as more solid-state lasers are developed that may provide for excitation at wavelengths less than 400 nm. Turning to another approach, mass spectroscopy (MS) clearly offers another solution to high-sensitivity detection in small volumes. Significant improvements in commercial instrumentation has occurred over the past decade, and now it can be stated that MS has become a reliable quantitative tool for the bioanalytical chemist. As present, perhaps the greatest limitation of MS is the prohibitively large start-up costs and the high maintenance costs.
C. Electrochemical Detection and Microseparation Techniques Another viable solution for trace determination in small volumes is electrochemical (EC) detection. Electrochemical detection has been applied successfully to microcolumn separations, including its use in both microbore HPLC, microbore open-tubular liquid chromatography, and capillary electrophoresis, and these applications have been reviewed [29]. The application of electrochemical detection in the low-volume flow cells of miniturized separation systems is potentially very large, due to the inherently sensitive nature of EC detection and the fact that the technique actually benefits from miniaturization. Generally, signal-to-noise ratios become more favorable with miniaturization in electrochemical detection [30], whereas these ratios become less favorable when using optical detection methods. Two detrimental aspects to the use of electrochemical detection are negative interactions of the analyte and matrix with the electrode surface, and, as is the case for fluorescence, the relatively limited percentage of analytes possessing electrochemical activity. The
208
/ Rose et al.
former manifests itself in electrode fouling and decreased detector response, which can be alleviated by the use of internal standards and/or mild detection potentials, where electrode surface fouling interactions are negligible. The lack of reasonable electrochemical detectability can be addressed by transformation of the molecule to an electrochemically active form through chemical derivatization or physical means. Previously reported transformations of this type were reviewed in 1985 [31]. The present review includes an overview of those approaches, plus additional techniques that have been developed and described in the past decade.
III. TRANSFORMATION METHODS AND SUMMARY TABLES The included tables represent a comprehensive collection of the reported chemical and physical methods used for the transformation of poorly detectable compounds to electrochemically detected analytes. The individual tables and associated discussion are categorized by the type of compound transformed and include precolumn and postcolumn methods. Details included in the tables are the reagent or method name, specific analyte(s) determined, reported limits of detection (S/N ⫽ 3), reaction conditions required to obtain this detection limit, and references. Limits of detection for all compounds are rounded to one significant figure and given as ‘‘concentration limits of detection’’ (CLOD) rather than the sometimes utilized ‘‘mass limits of detection’’ (MLOD). The application of microbore HPLC, capillary electrophoresis, and capillary electrochromatography, with their inherently low mass loading capabilities, has made it useful to quote limits of detection in terms of concentration injected onto the column rather than mass injected. MLOD can be extremely small in the case of the three aforementioned techniques and not fairly reflect the transformation methods utility in allowing for detection at trace levels. In some cases during this review, a conversion to CLOD requiring the injection volume was made from a MLOD reported by the author. When no injection volume was specified in the reported experimental procedure, it was assumed to be 20 µL when an analytical HPLC column (4.6-mm I.D.) was used. For reports containing data for multiple compounds, the CLOD listed is for the compound having the greatest sensitivity. The notation ‘‘‡’’ indicates that the quoted detection limit was obtained on a pure synthetic
Transformation of Analytes for Electrochemical Detection / 209 standard of the derivative or a sample transformed at high concentration and diluted, both situations which might not be representative of detection limits in the transformation of real samples. Reaction conditions include time and temperature as reported by the author. The notation ‘‘§’’ in reaction conditions indicates that an extraction, precipitation, or evaporation was performed as part of the derivatization procedure either to remove excess derivatization reagent or to concentrate the derivatized analyte, steps which may negatively influence precision and limit accessibility to smallvolume samples. Finally, the potentials listed in the tables are versus a Ag/AgCl reference electrode or have been converted to allow this comparision.
A. Primary Amines The greatest proportion of functional group transformations for electrochemical detection has been the derivatization of primary amines. This category includes primary alkylamines, amino acids, and peptides. Due to their unique structure, the derivatization of peptides is covered in a separate section, but as most peptides possess one or more primary amine functionalities, they can usually be derivatized with the reagents listed in this category. The most commonly applied reagent system for primary amines is o-phthaldehyde (OPA), which was reported as a fluorogenic derivatization reagent for primary amines in 1971 [85]. This compound was later discovered to additionally possess electrogenic characteristics [35,51,86]. The condensation reaction of the 1,2-dialdehyde of OPA with a primary amine to form the electroactive 1,2-disubstituted isoindole (Reaction 1) is unique among derivatization reagents in that OPA is hydrolytically stable in water, undergoes a rapid reaction at room temperature, and is specific for the primary amine functionality. The drawback in utilizing OPA is the limited stability of
Reaction 1
210
/ Rose et al.
the isoindole condensation products. The instability of the isoindole formed in this reaction has been linked to the steric nature of the amine being derivatized [87], the type of thiol, and the excess of OPA in the reaction mixture [88]. It was determined using radiolabeled phenylalanine and tyrosine that only 50% of the total derivatized product injected onto a reverse-phase high-performance liquid chromatography (HPLC) column actually reaches the detector intact [34], although such results must be carefully interpreted because isoindole product stability is highly dependent on the various reaction conditions and parameters. The substitution of sulfite for the traditional thiol co-reagent was observed to enhance isoindole stability, but in the case of electrochemical detection, optimum detection potentials were raised approximately ⫹200 mV for these more negatively charged isoindoles [59]. Most reactions of primary amines with OPA are complete within 1–2 min at room temperature, and the derivatized analytes, because of their instability, are normally analyzed immediately upon formation. CLODs in the low-nanomolar range can be obtained using oxidative detection at ⫹500 to ⫹850 mV (see Table 1). More recently, OPA has shown great utility in the determination of amino acids in biological samples of interest to neuroscientists. These studies, including those obtained by microdialysis, generate samples in the microliter size range, which often require special handling techniques, automated sample transfer, and the use of microbore chromatography. In one report, a comparison was made of five different OPA/MCE systems for the determination of aspartate and glutamate in microdialysis samples (5–10 µL) [38]. The best method in the report, which included the use of microbore chromatography and on-line microdialysis–derivatization–HPLC, allowed nanomolar CLOD ranges using a detection potential of ⫹700 mV. Other investigators have determined amino acids in rat brain tissue [55] and amino acid neurotransmitters in the cat visual cortex [37], with reported detection limits in the high and middle nanomolar range, respectively, when using OPA/BT. Optimized OPA/BT derivatization conditions, when used in conjunction with microbore chromatography, enabled other investigators to obtain a high-picomolar CLOD range for γ-aminobutyric acid with detection at ⫹800 mV [53]. Part of the optimization aspects included the use of glycine and iodoacetamide as scavengers for excess OPA and tert-butylthiol, respectively. A similar reagent for the electrogenic derivatization of primary
Primary Amines
Reagent OPA/MCE
Analyte Amino acids Glutamate Aspartate γ-Aminobutyric acid (GABA) 5-Aminolevulinic acid l-Thyroxine
Spermine Spermidine Putrescine Cadaverine Taurine Acetyl homotaurine Histamine Histamine Tryptamine Tyramine Phenylethylamine Putrescine 1,6-Diaminohexane Tryptophan S-Carboxymethyl-L-cysteine metabolites
E (mV)
CLOD
Reaction conditions
Refs.
4–25°C, 1–3 min
[32–37]
⫹600 ⫹750 ⫹700
to 50–3 nM 40 nM
4°C, 1 min
[38]
⫹700
r.t., 1–2.5 min
[39–41]
⫹450 ⫹730 ⫹650
5 µM–50 nM 50 nM 60 nM 10 nM
r.t., 1 min r.t., 30 s r.t., 2–10 min
[42] [43] [44]
⫹700
30 nM 10–30 nM
0°C, 6 min r.t., 20 s–1 min
[45] [46,47]
r.t., 5 min N/A
[48] [49]
r.t., 2 min
[50]
⫹500 2 nM ⫹1 to 100 nM ⫹1200*
⫹500
6 µM
Transformation of Analytes for Electrochemical Detection / 211
Table 1
212
Table 1
Continued
OPA/MCE OPA/BT
OPA/BT
Analyte Amphetamine Heptaminol Norephedrine Phenethylamine 2-Heptylamine γ-Aminobutyric acid(GABA) Amino acids
OPA/Sul
OPA NDA/CN ⫺
Baclofen γ-Aminobutyric acid(GABA) Alkylamines Alanine Arginine Glutamic acid Sarine Tyrosine Amino acids Glutathione Glutathione Amino acids Bisphosphonate-alendronate
E (mV) ⫹850
⫹600 ⫹800 ⫹700 ⫹750 ⫹700 ⫹700 ⫹850 ⫹600 ⫹850 0 ⫹960*
⫹850 ⫹750 ⫹800 ⫹900 ⫹650
CLOD 200 nM
40 nM 300 pM 6 nM 10 nM 200 nM 3 ng/mL 2.5 nM 50 nM N/A to 10 nM
300 pM 10 nM ‡ 2 µM 5 nM ‡ 10 nM
Reaction conditions
Refs.
r.t., 30 min
[51]
r.t., 3 min r.t., 8 min r.t., 6 min r.t., 2 min r.t., 2 min r.t., 1 min 25°C, 25 min 37°C, 15 min r.t., 1 min 20°C, 0–25 h
[52] [53] [54] [37] [55] [56] [57] [58] [59] [60]
r.t., r.t., r.t., r.t., r.t.,
[61] [62] [63] [64] [65]
10 min 1 min postcolumn 30 min 15 min
/ Rose et al.
Reagent
PITC
DM-PITC DFB/HPLC-hν-EC FITC FEITC N-HSP
MP N-SFP HPLC-hν-EC
TD-hν SA
γ-Aminobutyric acid (GABA) Alkylamines Amino acids 3-Methylhistidine Amino acids Glycine Amino acids Amino acids Amino alcohols Amino acids
⫺850 ⫺650 ⫺850 ⫹800 ⫹1100 ⫹1100 ⫹850 ⫹900
200 nM 50 nM 20 nM ‡ 1 nM ‡ 1 µM 1 µM 200 nM 30 nM
⫹500
Histamine Histamine Nτ-Methylhistamine Ethyl phenylalanate Phenethylamine Putrescine Aromatic, sulfur-containing amino acids Phenylalanine Phenylalanine
⫹560 ⫹560
Hydrazine 1,1-dimethylhydrazine
⫹800 ⫹400 ⫹450 ⫹800 ⫹800 ⫹400/ ⫹1000 ⫹1000
[66] [67] [68] [69] [70,71] [72] [73] [74]
3 nM
r.t., 30 min § r.t., 30 min § r.t., 60 min § 35°C, 10 min § r.t., 5 min § r.t., 5 min § 50°C, 40 min § r.t., 15 min § postcolumn 70°C, 30 min §
N/A 1 nM ‡
r.t., 30 s § r.t., 30 s §
[76] [77]
50 nM ‡ to 10 nM ‡
60°C, 30 min § r.t., 60 min
[78] [79,80]
400 nM
r.t. postcolumn
[81]
600 nM 30 pM/ 100 pM 800 nM
r.t. postcolumn r.t. postcolumn
[82] [83]
70°C, 60 min
[84]
§ Requires an extraction, a precipitation, or an evaporation as part of the derivatization procedure. ‡ CLOD of a pure standard or an analyte derivatized at a higher concentration and diluted. * Electrode array.
[75]
Transformation of Analytes for Electrochemical Detection / 213
TNBSA
214
/ Rose et al.
amines is naphthalenedialdehyde (NDA). Like OPA, NDA was also developed as a fluorogenic derivatization reagent [89–91] and then shown to be electrogenic. NDA reacts in a similar fashion to OPA, requiring a nucleophile such as cyanide (CN⫺ to complete the conversion of the starting ortho-dialdehyde to the N-substituted 1-cyanobenz[ f]isoindole (CBI) product (Reaction 2). In sharp contrast to
Reaction 2 the derivatives formed using OPA/thiol, the NDA derivatization products are relatively stable. NDA has been applied to preseparation derivatization of amines for electrochemical detection in several reported applications [64,65,92,93]. For analytes bearing multiple primary amine sites of similar reactivity, multiple derivatization occurs and results in a greatly diminished detectability due to fluorescence quenching [94]; however, the use of electrochemical detection avoids this problem, and in such situations there is actually an enhancement of analyte detectabilty as more sites are modified [95]. Analytes labeled with NDA have been detected down to low-nanomolar CLODs at a detection potential of ⫹700 mV. OPA and NDA are unique as amine derivatization reagents in that they are electrogenic, i.e., the reagents are inherently electrochemically inactive under the conditions that their dertivatization products are easily detectable. Despite the disadvantages to using a derivatization reagent that is equally detectable in its unreacted form, there have been reported a large number of successfully applied nonelectrogenic derivatization reagents. The primary approach has been to identify a functional group transformation whereby an electrochemically detectable moiety can be attached to the amine. Good success has been obtained through the use of appropriately substituted ferrocenes, with the ferrocene system serving as an electrochemical label. The N-hyroxysuccinimde-activated carboxylic acid derivative [75] (N-SFP, Reaction 3) and isothiocyanate-func-
Transformation of Analytes for Electrochemical Detection / 215 tionalized ferrocence [80] (FITC, Reaction 4) have both been described
Reaction 3
Reaction 4 as analytical derivatization reagents. The ferrocene system functions well as an electrochemically detectable moiety due to its mild detection potential (⫹400 to ⫹500 mV) and its electrochemical reversibility, yielding low-nanomolar detection limits. Similarly, use has been made of activated carboxylic acid (N-hydroxysuccimide) derivatives of phenol (N-HSP) [76,77] and 2-methoxyphenol (MP) [78] as electrochemical labels that result in CLODs similar to those obtained with ferrocene, but in each case requiring higher detection potentials. Other interesting approaches have concerned the use of phenyl isothiocyanates (PITC), which form thiocarbamates with primary and secondary amines (Reaction 5 and Reaction 6). Amino acid derivatives thus formed have been detected at low-nanomolar concentra-
Reaction 5
216
/ Rose et al.
Reaction 6 tions at potentials of ⫹800 to ⫹1100 mV. This approach has been successfully used for the determination of amino acids in human blood and urine [69–72]. The substitution of a dimethylamino group at the 4-position of PITC to form p-N,N-dimethylaminophenylisothiocyanate (DM-PITC) results in a reagent forming amino acid derivatives that are more easily detected than their PITC counterparts [73]. Using this approach, amino acids were detected at the highnanomolar concentration range using a potential of ⫹850 mV [73]. Reductive approaches for amino acid analysis have been dominated by the use of nitro-substituted aromatics as the electrophores. Trinitrobenzene sulfonic acids (TNBSA) were used to form the trinitrobenzene derivatives of GABA, other amino acids, and alkyl amines (Reaction 7) [66–68]. The derivatives can be detected in the high- to mid-nanomolar range using reductive potentials of ⫺600 to
Reaction 7 ⫺850 mV. Typically, these methods require additional sample preparation steps to remove excess reagent. Another approach using a nitrobenzene as an electrophore is the combination of this derivatization with HPLC-photolysis-electrochemical detection (HPLChν-EC) to produce an oxidatively detectable species. Amino acids and amino alcohols were determined through the use of precolumn derivatization with 2,4-dinitrofluorobenzene followed by HPLC-
Transformation of Analytes for Electrochemical Detection / 217 hν-EC (DFB/HPLC-hν-EC) [74]. The derivatized products were detected at mid-nanomolar concentrations using a detection potential of ⫹900 mV. Aromatic and sulfur-containing amino acids were determined using HPLC-hν-EC with a CLOD in the high-nanomolar range [81] using a detection potential of ⫹800 mV. In a separate report by the same authors, phenylalanine was determined in human urine with limits of detection in the high-nanomolar range [82]. Phenylalanine has no inherent electrochemical activity below approximately ⫹1200 mV. Postcolumn irradiation (approx. 2 min residence time) allowed oxidative detection to be accomplished at a detection potential more compatible to the use of glassy carbon electrodes (⫹800 mV) in aqueous solvent systems. Building on this work, postcolumn photolytic derivatization was used in combination with a titanium dioxide catalyst (TD-hν) for the determination of phenylalanine with limits of detection in the high- to mid-picomolar range using detection potentials of ⫹400 to ⫹1000 mV, respectively [83]. The addition of the titanium dioxide catalyst to the reactor decreased the limit of detection by one order of magnitude when compared with the same system without the catalyst.
B. Secondary Amines Transformation of secondary amines is similar to that performed on primary amines. Since OPA and NDA are specific for primary amines, the main derivatization strategy has been that of labeling the amine using a chemically reactive electrophore. Ferrocenesulfonylchloride (FSC) and ferrocenecarboxylic acid chloride (FAC) have been used in a fashion similar to that shown for primary amines to label poorly detectable secondary amines, along with amino acids and peptides (see Section III.C) [104]. Another acid chloride-based labeling reagent, o-acetylsalicyloyl chloride (ASC), can be used in two steps to label a secondary amine with an ortho-phenol electrophore (Reaction 8) [97]. As with all phenol electrophores, the deriva-
Reaction 8
218
/ Rose et al.
tives are detected at fairly high oxidation potentials (⫹900 mV). The fluorescence-imparting reagent, 7-chloro-4-nitrobenzo-2-oxa-1,3diazole (NBD-Cl), reacts preferentially with secondary amines and also yields electrochemically detectable derivatives (Reaction 9). NBD-Cl was used to derivatize hydroxyproline in purified type I col-
Reaction 9 lagen and other biological samples [98], and the resulting derivatives, stable for a few hours when refrigerated, resulted in a CLOD in the mid-nanomolar range using a detection potential of ⫹640 mV (see Table 2). Phenylisothiocyanate (PITC) was used to form electroactive derivatives of ephedrine, pseudoephedrine, and norephedrine in a reaction similar to that shown for primary amines (see Reaction 5). Isolated standards were detected into the low-micromolar range using a detection potential of ⫹1300 mV [99]. Likewise, more detectable products were formed through an isothiocyanate (see Reaction 6) using 4-dimethylaminophenyl isocyanate (DM-PITC) to derivatize arylhydroxamines from liver homogenates [101]. The derivatization to form hydroxyureas served both to increase detectability and to stabilize the arylhydroxylamines. The derivatives thus formed were stable for 14 h at 0°C and detectable into the mid-nanomolar range at ⫹500 mV.
Reaction 10
Secondary Amines
Reagent
Analyte
E (mV)
CLOD
Reaction conditions
Refs.
NQS ASC NBD-Cl PITC
2° Amines 2° Amines Hydroxyproline Ephedrine Pseudoephedrine Norephedrine Renin inhibitor (Ro 42-5892) Arylhydroxylamines 2° Amines 2° Amines
0.0 ⫹900 ⫹640 ⫹1300
3 nM ‡ 600 nM 30 nM 3 µM
60°C, 40 min § r.t., 60 min 60°C, 5 min r.t., 10 min §
[96] [97] [98] [99]
⫹550 ⫹500 ⫹1200 ⫹650
500 pM 10 nM 100 nM 100 nM
80°C, 25 min 0°C, 1 min § 70°C, 50 min r.t., 1 min §
[100] [101] [102] [103]
DFB/HPLC-hν-EC DAPIC SAC Me/CDS
§ Requires an extraction, a precipitation, or an evaporation as part of the derivatization procedure. ‡ CLOD of a pure standard or an analyte derivatized at a higher concentration and diluted.
Transformation of Analytes for Electrochemical Detection / 219
Table 2
220
/ Rose et al.
A novel method, based on the precolumn formation of copper or nickel dithiocarbamate complexes (Me-CDS, Reaction 10) [103], was applied to the detection of secondary amino compounds. The analytical products formed were detected in high-nanomolar concentrations at ⫹650 mV. In a successful combination of derivatization and postcolumn photolysis, the renin inhibitor Ro 42-5892, which contains a substituted imidazole, was labeled using 2,4-dinitrofluorobenzene and allowed to undergo postcolumn irradiation (DFB/HPLC-hν-EC, Reaction 11), resulting in detection limits into the high-picomolar
Reaction 11 range [100]. The derivatization (80°C, 25 min) was performed on extracts of human plasma using an automated sample processor and was rugged enough to be applied to 250 clinical samples. The use of reducible labels for secondary amine derivatization includes the β-naphthoquinone-4-sulfonate (NQS), which reacts (60°C, 40 min) to form the 4-amino-β-naphthoquinone (Reaction 12) [96]. The quinone-labeled analyte can then be detected at low-nanomolar concentrations at reduction potential of 0.0 mV.
Reaction 12
C.
Peptides
As mentioned in Section III.A for primary and secondary amines, peptides can be derivatized by targeting primary amine functionalities. Both OPA [105] and NDA [92,93] have been used in this fashion
Transformation of Analytes for Electrochemical Detection / 221 to label peptides for electrochemical detection. The carboxylic acid anhydride 3,6-dinitrophthalic anhydride (DNPT) was used successfully to form peptide derivatives for reductive electrochemical detection [112]. DNPT-derivatized Val-Val was detectable into the midnanomolar range when using a detection potential of ⫺240 mV. Interestingly, even though there are four primary amine groups (one α-amino site and three ε-amino groups), this reagent appeared to form only a single product when used to derivatize Lys-Lys-Lys. This suggests that one site is susbstantially more selective, which seems unlikely, or that a complete per-derivatization occurred (see Table 3). A second approach taken to peptide transformation has been the application of the century-old biuret reaction [Cu(II)], which causes the formation of oxidatively detectable copper complexes. This approach has been performed precolumn for large peptides [110], in a postcolumn format for small peptides [108,109], and as an on-column reagent system for capillary electrophoresis [111]. The products are detectable at ⫹800 to ⫹900 mV at the low-nanomolar level using HPLC and in the high-nanomolar range when using capillary electrophoresis. The utility of HPLC and postcolumn UV irradiation with electrochemical detection (HPLC-hν-EC) was demonstrated by the determination of the 6-residue cyclic peptide ‘‘520’’ in human plasma [113]. Although this peptide is described as containing no electroactive amino acids, it does contain a thiol ether functionality, which becomes electroactive upon photolysis. High-nanomolar detection limits were obtained using this approach when detecting the irradiated analytes using a detection potential of ⫹1000 mV.
D. Thiols Thiol groups (pKa 10–11) are more acidic than their corresponding alcohols (pKa 9.5–17), making them an easier target for derivatization reactions. Derivatization of thiols serves to enhance detectability and prevents their facile oxidation to the corresponding disulfides (see Table 4). A general review for the analysis of thiols, which includes methods for electrochemical detection, has been published [119]. The primary approach to thiol derivatization is the coupling of an electrophore to a thiol-reactive linking group. Electrophore coupling approaches have included N-(4-dimethylaminophenyl)maleimide (DMPM) [117], anilinophenylmaleimide (APM) [114,115], and N-(ferrocenyl)maleimide (FM) [118]. Addition of the thiol group to
222
Peptides Analyte
Cu(II)
Sulfur-containing amino acids, γglutamyl di- and tripeptides Desmosine Isodesmosin Amino acids Di-, tri-, quad-, pentapeptides and amino acids Dipeptides
DNPT FSC FAC HPLC-hν-EC
Tyrosine peptides Peptides Oligopeptides Peptides Peptides Peptides, amino acids, alkylamines Cyclic peptide ‘‘520’’
OPA/MCE NDA/CN⫺
E (mV)
CLOD
Reaction conditions
Refs.
⫹700
5 nM
25°C, 2 min
[105]
⫹750
5 nM
r.t., 30 min
[92,93]
N/A
N/A
r.t., 60 min
[95]
⫹700 to N/A ⫹1200 ⫹800 2 nM ⫹800 10 nM ⫹800 N/A ⫹900 700 nM ⫺240 50 nM ‡ ⫹700 30 nM ‡
r.t. postcolumn
[106,107]
postcolumn r.t. r.t. postcolumn 60°C, 20 min r.t. on-column 50°C, 0–24 h r.t., 30 min
[108] [109] [110] [111] [112] [104]
⫹1000
r.t. postcolumn
[113]
600 nM
‡ CLOD of a pure standard, or an analyte derivatized at a higher concentration and diluted.
/ Rose et al.
Table 3
Thiols
Reagent APM
Analyte N-Acetyl-L-cysteine glutathione
CLOD
Reaction conditions
Refs.
⫹1000/⫹700
6 nM ‡
0°C, 90 min §
[114]
⫹1000
10 nM
0°C, 30 min § r.t., 30 min
[115]
⫹1200 ⫹900 ⫺100 (Pd ref.)
100 nM 50 nM ‡ 3 nM
r.t., 30 min 0°C, 30 min§ 0°C, 30 min§
[116] [117] [118]
E (mV)
l-Cysteine d-Penicillamine
BBQ DMPM FM
2-(2-hydroxyphenyl)-3-(3mercaptopropionyl)-4thiazolidinecardoxylic acid Glutathione (GSH) Captopril N-Acetyl-L-cysteine, glutathione
§ Requires an extraction, a precipitation, or an evaporation as part of the derivatization procedure. ‡ CLOD of a pure standard or an analyte derivatized at a higher concentration and diluted.
Transformation of Analytes for Electrochemical Detection / 223
Table 4
224
/ Rose et al.
reactive maleimides typically occurs in 10–30 min at 0°C. Due to the ease of reagent detection, these procedures usually require an extraction to remove excess reagent. The products are detectable at ⫹900 to ⫹1000 mV in mid- to low-nanomolar levels. The product initially formed with APM and cysteine or penicillamine undergoes a subsequent cyclization upon mild heating (Reaction 13). The re-
Reaction 13 sulting electrophore has a detection potential of ⫹700 mV. Derivatives formed using the maleimide pathway with ferrocene as the electrophore (FM) can be detected into the low-nanomolar range at about ⫹500 mV. Another approach for the detection of thiol-containing substances is the reaction of thiols with 3,5-di-tert-butyl-1,2-benzoquinone (BBQ, Reaction 14) [116]. The resulting product was quantitated at high-nanomolar concentrations using a detection potential of ⫹1200 mV. Low-micromolar concentrations of cysteine have also been determined using postcolumn UV irradiation at ⫹650 mV (see Section III.A) [81]. The authors were unsure as to the species produced by irradiation that was oxidatively sensitive. A similar ap-
Reaction 14
Transformation of Analytes for Electrochemical Detection / 225 proach was used to quantitate sulfur-containing antibiotics to nanomolar concentrations at ⫹1100 mV (see Section III.I) [120].
E. Alcohols Alcohols are substantially less nucleophilic than the corresponding thiols, due to a combination of polarizability and electronegativity effects, but approximately the same reactive groups apply for linking these molecules to an electrochemical label. Due to limited nucleophilicity, alcohols derivatizations are normally performed in nonaqueous solvents. This allows the use of highly reactive linking groups, such as carboxylic acid chlorides or the corresponding active esters, isocyanates, and acid anhydrides, since such reagents are relatively stable in the nonaqueous, nonnucleophilic solvents typically used for these derivatizations (see Table 5). Alcohols including steroids have been derivatized using ferrocene as the electrophore, with either the carboxylic acid chloride (FCAC) or isocyanate (FITC) as the linking group [122]. Either reagent decomposes rapidly upon exposure to atmospheric moisture. In this same report, substitution of a carboxylic acid azide group for the isocyanate served to form a more easily handled ferrocene derivatization reagent (3-ferrocenyl acyl azide, FA). The carboxylic acid azide group reacts with alcohols through an isocyanate intermeTable 5 Alcohols Reagent SAC FCAC FITC FPA IEMN/Pt AP TTAA FBA Pd
Analyte
E (mV)
CLOD
Reaction conditions
Refs.
Anabolic steroids Hydroxysteroids
⫹1000 ⫹400
50 nM 70°C, 60 min§ 30 nM ‡ 80°C, 15 min§
[121] [122]
Hydroxysteroids Phenols ⫾Delmopinol Brassinosteroids
⫹700 ⫺200 ⫹1100 ⫹600
2 nM 40 nM 5 nM 3 nM
[123] [124] [125] [126]
Pd
100°C, 15 min r.t., 2 min§ 40°C, 60 min 70°C, 10 min
Palladium reference electrode. § Requires an extraction, a precipitation, or an evaporation as part of the derivatization procedure. ‡ CLOD of a pure standard or an analyte derivatized at a higher concentration and diluted.
226
/ Rose et al.
diate generated by heat or light to form urethanes (Reaction 15). The
Reaction 15 resulting ferrocene-labeled products were detectable at ⫹400 mV at low-nanomolar levels. In a separate study, brassinosteroids were labeled at the 2,3-hydroxy position using a condensation reaction with ferroceneboronic acid (FBA, Reaction 16) [126]. The resulting derivatives were detected at low-nanomolar concentrations using a potential of ⫹600 mV.
Reaction 16 Other activated electrophores, including salicylic acid chloride (SAC), have been used to form phenol-labeled anabolic steroids detectable in the mid-nanomolar range using ⫹1200 mV [121]. An extended alkane alcohol, (⫾)-delmopinol, was labeled using the chiral derivatizing reagent (R,R)-di-p-toluoyl tartaric acid anhydride (TTAA) [125]. Low-nanomolar concentrations of the resulting derivative were quantitated at ⫹1100 mV. A novel approach involves the use of 2-[2-(isocyanate)ethyl]-3-methyl-1,4-naphthoquinone (IEMN, Reaction 17) to form naphthoquinone-labeled derivatives [123]. The
Reaction 17
Transformation of Analytes for Electrochemical Detection / 227 derivatives are then reduced using a postcolumn platinum catalyst (Pt) and detected at ⫹700 mV at concentrations in the low-nanomolar range. The sole reported transformation of alcohols reviewed here for reductive detection involves the use of 4-aminoantipyrine (AP) for the transformation of phenols to the corresponding quinoneimines (Reaction 18) [124]. The derivatives can be detected at ⫺200 mV at
Reaction 18 the mid-nanomolar concentration range. This mild reduction potential is in contrast to the ⫹1000 mV normally required to quantitate the underivatized phenol using oxidative electrochemical detection. The authors reported that deoxygenation is not required due to the relatively low reduction potential.
F. Keto Compounds Ketone determination for electrochemical detection has been most frequently accomplished by derivatization with analogs of phenylhydrazine (Reaction 19). This approach has been for the determination
Reaction 19 of 17-keto steroids present in human blood [131,132], and for the determination of 2-oxoglutarate in chick osteoblast cultures [127] (See Table 6). The approach appears to be generally applicable for most aldehydes and ketones [130]. The resulting derivatives have been detected in the low-nanomolar range at potentials ranging from
228
Keto Compounds
Reagent
Analyte
PHD DNPHD
2-Oxoglutarate Formaldehyde Acetaldehyde Acetone Acrolein Aliphatic aldehydes Aldehydes and acetone 17-Ketosteroids
DHBH PHD NPHD DNPHD
/ Rose et al.
Table 6
CLOD
Reaction conditions
⫹800 ⫺750
200 nM 90 nM ‡
r.t., 60 min r.t., precipitation
[127] [128]
⫹300 ⫹1100 ⫹800
30 nM ‡ 50 nM ‡ 40 nM
r.t., 60 min r.t., precipitation 60°C, 30 min§
[129] [130] [131,132]
E (mV)
§ Requires an extraction, a precipitation, or an evaporation as part of the derivatization procedure. ‡ CLOD of a pure standard or an analyte derivatized at a higher concentration and diluted.
Refs.
Transformation of Analytes for Electrochemical Detection / 229 ⫹800 to ⫹1100 mV. Hydrazine reactions with keto groups have also been utilized for reductive electrochemical detection using dinitrophenylhydrazine (DNPHD) [128]. Derivatives formed in this fashion have been detected in the mid-nanomolar range at a detection potential of ⫺750 mV.
G. Carboxylic Acids A challenge in the electrochemical labeling of carboxylic acids is the development of reagents and reaction conditions for the efficient transformation of this functional group that does not degrade the analyte. The procedures used to add an electrophore to the carboxylic acid functional group generally are similar to the coupling reactions that are widely used in peptide synthesis (see Table 7). Two reviews covering the general transformation of carboxylic acids, including methods for electrochemical detection, have been published recently [141,142]. The solvents involved in these derivatizations are usually nonaqueous (i.e., tetrahydrofuran or pyridine) and allow the use of a diversity of electrophores. Substituted hydroquinone was used as an electrophore to label benzoic acid and various β-lactam penicillins by reaction with 1-(2,5dihydroxyphenyl)-2-bromoethanone (DPBE, Reaction 20) [133]. The resulting hydroquinone-labeled benzoic acid was detectable in the
Reaction 20 high-nanomolar range at ⫹600 mV. Using the same reagent, valproic acid was determined in the mid-nanomolar range using a detection potential of ⫹450 mV [134]. Due to the volatility and propensity for evaporative loss of valproic acid during determination, the derivatization reaction rendered the analyte less volatile, which allowed for its concentration by solvent evaporation. The derivative, being of increased hydrophobicity, also displayed improved chromatographic properties when separated by reverse-phase HPLC. Chenodeoxycholic acid was determined in a similar fashion by the same authors in the mid-nanomolar range using a detection poten-
230
Carboxylic Acids
Reagent DPBE
BDF PAP/BMI
FEA/EDC
DMA/EDC HPLC-hν-EC, Fe (III)
Analyte β-Lactam penicillins Valproic acid salts Chenodeoxycholic acid Fatty acids Stearic acid Chenodeoxycholic acid Prostaglandins Chiral carboxylic acids Steroid Gluccuronides Prostaglandins Oxalate
E (mV)
CLOD
Reaction conditions
Refs.
⫹600 ⫹450 ⫹600 ⫹600 ⫹700
100 nM 70 nM ‡ 200 nM ‡ 30 nM ‡ 90 nM ‡
45°C, 75°C, 75°C, 80°C, 60°C,
⫹450 ⫹450
30 nM ‡ 30 nM ‡
4°C, 120 min 37°C, 120 min§
[138] [139]
⫹1100 ⫹900
10 nM 30 µM
37°C, 60 min § r.t. postcolumn
[140] [145]
120 min 45 min 120 min 60 min 30 min
§ Requires an extraction, a precipitation, or an evaporation as part of the derivatization procedure. ‡ CLOD of a pure standard or an analyte derivatized at a higher concentration and diluted.
[133] [134] [135] [136] [137]
/ Rose et al.
Table 7
Transformation of Analytes for Electrochemical Detection / 231 tial of ⫹600 mV [135]. The derivatizations for both valproic acid and chenodeoxycholic acids were performed at 70°C for 45 and 120 min, respectively. 3-Bromo-1,1-dimethyl ferrocene (BDF) was used similarly for the attachment of the ferrocene electrophore for the detection of fatty acids in human serum [136]. Chiral carboxylic acids [138] and steroid gluccuronides, present in the urine of pregnant women [139], were labeled with ferrocene using a water-soluble carbodiimide coupling procedure (FEA/EDC, Reaction 21). The resulting ferrocene derivatives were detectable at ⫹450 mV in the mid-nanomolar range.
Reaction 21 In a separate study, it was found that p-aminophenol, when combined with 2-bromo-1-methylpyridium iodide (PAP/BMI), could be used to derivatize fatty acids, bile acids, and prostaglandins to form p-hydroxyanilides (Reaction 22) [137]. These products were detect-
Reaction 22 able in the mid-nanomolar range at ⫹700 mV. The method was applied to the determination of fatty acids in guinea pig plasma and bile acids in human bile. A dicarboxylic acid, oxalic acid, was determined using a combination of Fe(III) complexation followed by UV irradiation to form a reduced species [145]. The oxalate containing sample is mixed postcolumn with a solution of Fe(III) to form the iron complex, which, after irradiation, undergoes reduction to Fe(II), which can be detected using oxidative electrochemical detection (Reaction 23).
232
/ Rose et al.
Through this method, oxalate was determined to mid-micromolar concentrations using ⫹900 mV.
Reaction 23
H. Metals and Ions Transformations of metals and ions for electrochemical detection are accomplished by either postcolumn UV irradiation or complexation to form a reduced species for oxidative detection (see Table 8). Postcolumn UV irradiation (HPLC-hν-EC) has been utilized to determine nitrate, based on the conversion to nitrite using UV irradiation after separation with ion chromatography [143]. The nitrite form of the ion can then be detected using electrochemical detection. Nitrate samples were determined in the high-nanomolar range using ⫹1100 mV with this method; however, the authors determined that the detection limits using UV (220 nm) were slightly superior. Postcolumn photoreduction techniques were utilized for the determination of dichromate, chromate, perchlorate, and thiocyanate [144]. CLODs were obtained at the low-micromolar range using a detection potential of ⫹1150 mV. The dithiocarbamate ligand forms stable complexes with a range of metal ions and can be used both as a precolumn transformation and as an in-situ complexing agent when included in the running buffer (Reaction 24). Precolumn derivatization can be beneficial due
Reaction 24 to the increased hydrophobicity of the metal dithio complexes that are formed, which facilitates separation by reverse-phase HPLC. Dithiocarbamate complexation has been used in separate reports to quantitate copper [152], nickel, cobalt and chromium [150], lead, cadmium, and mercury [153], with low-micromolar to low-nanomolar detection limits using detection potentials of ⫹700 to ⫹1200 mV. The results obtained by this approach were in agreement with those based on atomic adsorption spectroscopy. Dithiocarbamate complex-
Metals and Ions
Reagent HPLC-hν-EC
EX PNSA HEDC Zn/HEDC DTC DEDC PCD QL
Analyte Nitrates Dichromate Chromate Perchlorate Thiocyanate Co(II) Co(II) Cd(II), Co(II), Pb(II), Ni(II), Cu(II) Cd(II), Co(II), Ni(II), Cu(II) Cu(II), Ni(II), Co(II), Cr(VI), Cr(III) Pb(II), Cd(IP), Hg(II), Co(II), Ni(II), Cu(II) Al(III), Cu(II), Fe(III), Mn(III) Al(III), Fe(III), Mn(III)
E (mV)
Reaction conditions
CLOD
Refs.
⫹1100 ⫹1150
800 nM 1 µM
0°C postcolumn 0°C postcolumn
[143] [144]
1400 Pulse 250 ⫹1000
4 nM
r.t.
[146]
50 nM 700 nM
r.t. on-column r.t. on-column
[147] [148]
⫹900
200 nM
18°C on-column
[149]
⫹700 to 200–300 nM ⫹1000 ⫹700 to 200 nM ⫹1200 ⫺500 5 µM
r.t. on-column and precolumn r.t. precolumn
[150–152]
r.t., on-column
[154]
⫺400
50°C, 10 min
[155]
200 nM
[153]
Transformation of Analytes for Electrochemical Detection / 233
Table 8
234
/ Rose et al.
ation was also proven to be adaptable to automation [151], and it was found that the use of zinc bis-[(2-hydroxyethyl)-dithiocarbamate] was preferred over salts of diethyldithiocarbamate or pyrrolidinecarbodithioate, due to better stability in aqueous media [149]. Complexing agents which have been applied to reductive electrochemical detection of metals include 8-quinolinol (QL), and 1-(2-pyridylazo)-2-naphthol-6-sulfonic acid (PNSA). The heterocyclic phenol, 8-quinolinol, has been used in the in-situ complexation of aluminum, copper, iron, and manganese [154,155]. In a successful study using QL, metals were detected at low-nanomolar levels using reductive electrochemistry at ⫺400 mV. PNSA was used in the electrochemical detection of cobalt and had the advantage of offering stable complexes reducing at relatively mild potentials. Cobalt(II) was determined at mid-nanomolar levels using ⫺250 mV as the detection potential.
I. Miscellaneous Analytes A certain number of derivatization/transformations do not fit into specific functional group categories or involve more than one location on the molecule. One such general technique was the application of HPLC-photolysis-electrochemical detection (HPLC-hν-EC), which was found to be highly suited to the analysis of a wide spectrum of organic compounds (see Table 9). In one study, HPLC-hν-EC was used to increase the electrochemical detectability of drugs of abuse when determined from human urine and serum. In this way, phenobarbital and cocaine were determined at mid-nanomolar concentrations using a detection potential of ⫹1000 mV [156]. The transformation of the analyte into an oxidizable species was theorized to occur through multiple routes depending on structure, but a high degree of success was noted for molecules containing phenyl substituents, pyrimidine, nitrone, and nitro functionalities. HPLC-hν-EC was also found to have applicability to the determination of organoiodides [157,159]. The samples were separated using reverse-phase LC and then irradiated with a low-pressure mercury discharge lamp through a knitted tubular reactor to form iodide, which can be determined in the mid-nanomolar range using a detection potential of ⫹1100 mV. It was found in subsequent studies that this system is highly dependent on solvent composition, which influences the stability of the iodide [158]. Similar assays were performed on organobromides and organochlorides with a number
Miscellaneous Analytes
Reagent HPLC-hν-EC
Enzyme reactor Pt red. Ethanol PAP CA Ce(IV)
Analyte Phenobarbital Cocaine Penicillins Cefoperazone Organoiodides Organobromides Organochlorides Choline Acetylcholine Idebenone Aromatic isocyanates Toluene diisocyanate Carbohydrates Naphthalene Phenanthrene Anthracene
E (mV)
CLOD
Reaction conditions
Refs.
⫹1000
40 nM
r.t. postcolumn
[156]
⫹1100
80 nM
r.t. postcolumn
[120]
r.t. postcolumn r.t. postcolumn
[157,158] [159]
25–37°C postcolumn
[160–167]
r.t. postcolumn r.t., § r.t., § 100°C postcolumn r.t., 1 min§
[168] [169] [170] [171] [172]
⫹1000 400 pM ⫹250 to 2 nM ⫹350(Ag) ⫹480 to 10–200 nM ⫹650(Pt) ⫹700 200 pM ⫹950 300 pM ⫹1250 30 nM ⫹400 1 µM ⫺400 40 nM
§ Requires an extraction, a precipitation, or an evaporation as part of the derivatization procedure. (Ag) Silver working electrode. (Pt) Platinum working electrode.
Transformation of Analytes for Electrochemical Detection / 235
Table 9
236
/ Rose et al.
of experimental parameters being examined, including the detection potential, electrode composition, UV irradiation, and flow rate. The lowest experimentally determined levels for a compound such as bromobenzene were in the mid-nanomolar range when detected at ⫹250 mV. Multiple applications of the quantitation of acetyl choline and choline using a postcolumn enzymatic reaction to form an electrochemically detectable analyte have been reported [160–166]. In this transformation, which has been applied primarily to the determination of choline from rat neuronal tissue, acetylcholine is enzymatically degraded to choline using a postcolumn reactor containing acetylcholinesterase, which is then subsequently degraded to hydrogen peroxide by choline oxidase, which can be detected at a potential of ⫹500 mV and a platinum working electrode (Reaction 25). Limits
Reaction 25 of detection reported for this system were in the mid-nanomolar range. The analytical systems were stable over hundreds of samples. The oxidized form of 1,4-hydroquinone, 1,4-benzoquinone, presents an analyte that can be selectively reduced to the 1,4-dihydroxy form using a platinum catalyst, which can be subsequently oxidized at a mild potential. Idebenone was determined in rat serum and brain tissues by reverse-phase HPLC separation and detection after platinum postcolumn reduction (Reaction 26) [168]. The resulting substituted hydroquinone was determined at high-picomolar concentrations using a detection potential of ⫹700 mV.
Reaction 26 The determination of atmospheric isocyanate is of interest and can be accomplished using electrochemical methods. Isocyanates can
Transformation of Analytes for Electrochemical Detection / 237 be transformed into polyurethanes and are widely encountered in the chemical industry. The reaction of phenylisocyanate with a nucleophile forms the same aryl urethane electrophore used for the derivatization of amines (see Reaction 5). When the phenyl isocyanate is the derivatization target, use of an amine nucleophile will enable the detection of the derivative, due either to the transformation to the aryl urethane or to an additional electrophore introduced by nucleophilic substitution. Aromatic diisocyanates were determined using alkaline ethanol to form the corresponding urethanes [169]. Using an enrichment column, the derivatives were detectable into the high-picomolar range at a detection potential of ⫹950 mV. In a separate study p-aminophenol was judiciously chosen as the nucleophile for the same type of reaction (Reaction 27) [170]. The resulting dia-
Reaction 27 minophenol derivative was detectable to low-micromolar concentrations but at a significantly reduced potential of ⫹600 mV.
IV.
NDTE
The application of miniaturized separation systems in these laboratories has led to an interest in developing electrochemical derivatization methods for poorly detectable amines. One very promising reagent system is p-nitrophenyl 2,5-dihydroxyphenylacetate, bistetrahydropyranyl ether (NDTE) [173]. NDTE reacts in a two-step reaction to form stable hydroquinone-labeled analytes from molecules containing one or more primary or secondary amine groups (Reaction 28). The reagent structure is such that it allows the delivery of a very oxidatively sensitive hydroquinone label in basic pH, aqueous environments.
238
/ Rose et al.
Reaction 28
Fig. 1 Cyclic voltammogram of N-ethylbenzylamine derivatized using NDTE (black) superimposed on the cyclic voltammogram of hydroquinone (gray) analyzed under the same conditions. Cyclic voltammograms were generated in 1:1 acetonitrile phosphate buffer (0.1 M, pH 7.0) from ⫺500 mV to ⫹1000 mV to ⫺500 mV (versus Ag/AgCl, [Clv] ⫽ 3 M ) on analyte (1 mM concentrations) at a scan rate of 100 mV/s on glassy carbon.
Transformation of Analytes for Electrochemical Detection / 239 Hydroquinone-labeled analytes formed using NDTE possess electrochemical detection characteristics similar to native hydroquinone. Figure 1 shows a cyclic voltammogram for NDTE-derivatized N-benzylethylamine compared to a voltammogram for native hydroquinone analyzed in the same solvent. The oxidation peak for the labeled compound occurs at a significantly milder relative potential (⫹151 mV) than for hydroquinone (⫹286 mV) at pH 7. Figure 2 shows the RP-HPLC-EC analysis of a pure standard of NDTE-derivatized N-benzylethylamine using a standard thin-layer electrochemical cell. The hydroquinone-labeled analyte is detectable down to 5 nM in a 20-µL injection (MLOD of 100 fmol) when detected using ⫹200 mV. Figure 3 shows series oxidative/reductive chromatograms for a single isocratic RP-HPLC-EC analysis of three poorly detectable esterified amino acids: L-glutamine t-butyl ester, L-methionine methyl ester, and L-isoleucine methyl ester, each at a prederivatization concentration of 50 ⫻ 10⫺6 M in 75/25 acetonitrile/borate buffer (pH 9.1). The two-step derivatization was accomplished in a total of 2 h at room temperature using NDTE. Derivatization of analytes using NDTE has yielded mid-nanomolar detection limits for numerous analytes including primary and secondary alkyl amines [173], amino acids, and peptides [174], which have been subsequently quantitated using RP-HPLC-EC, microbore HPLC-EC, and CE-EC. Continuing research is focusing on the development of improved reagents of this type and their application in
Fig. 2 RP-LC-EC determination (⫹200 mV versus Ag/AgCl, [Clv] ⫽ 3 M ) of a pure standard of NDTE-derivatized N-ethylbenzylamine (5 ⫻ 10⫺9 M, 20-µL injection).
240
/ Rose et al.
Fig. 3 RP-LC-EC determination using series oxidation (⫹300 mV) and reduction (⫺300 mV) performed on an NDTE-derivatized single solution mixture of: (a) L-glutamine t-butyl ester (50 ⫻ 10⫺6 M ); (b) L-methionine methyl ester (50 ⫻ 10⫺6 M ); (c) L-isoleucine methyl ester (80 ⫻ 10⫺6 M ). The twostep NDTE derivatization was accomplished in a total of 2 h at room temperature.
bioanalytical problems using various microsampling and microseparation technologies.
ACKNOWLEDGMENTS This research was supported by a National Cancer Institute Training Grant (CA-09242), The Pharmaceutical Research and Manufacturers Association, Proctor & Gamble, and The Center for Bioanalytical Research (CBAR) at The University of Kansas.
REFERENCES 1. C. S. Effenhauser, A. Manz, and H. M. Widmer, Anal. Chem., 65: 2637 (1993). 2. C. F. Effenhauser, A. Paulus, A. Manz, and H. M. Widmer, Anal. Chem., 66: 2949 (1994). 3. Z. H. Fan and D. J. Harrison, Anal. Chem., 66: 177 (1994).
Transformation of Analytes for Electrochemical Detection / 241 4. D. J. Harrison, A. Manz, Z. Fan, H. Lu¨di, and M. Widmer, Anal. Chem., 64: 1926 (1992). 5. S. C. Jacobson, R. Hergenro¨der, L. B. Koutny, and J. M. Ramsey, Anal. Chem., 66: 1114 (1994). 6. S. C. Jacobson, R. Hergenro¨der, L. B. Koutny, R. J. Warmack, and J. M. Ramsey, Anal. Chem., 66: 1107 (1994). 7. S. C. Jacobson, R. Hergenro¨der, J. Alvin W. Moore, and J. M. Ramsey, Anal. Chem., 66: 4127 (1994). 8. S. C. Jacobson, L. B. Koutny, R. Hergenro¨der, A. W. Moore, and J. M. Ramsey, Anal. Chem., 66: 3472 (1994). 9. L. B. Koutny, D. Schmalzing, T. A. Taylor, and M. Fuchs, Anal. Chem., 68: 18 (1996). 10. K. Seiler, D. J. Harrison, and A. Manz, Anal. Chem., 65: 1481 (1993). 11. K. Seiler, Z. H. Fan, K. Fluri, and D. J. Harrison, Anal. Chem., 66: 3485 (1994). 12. S. J. Lane, R. Boughtflower, C. Paterson, and T. Underwood, Res. Commun. Mass Spectrosc., 9: 1283 (1995). 13. F. Lelie´vre, C. Yan, R. N. Zare, and P. Gareil, J. Chromatogr. A, 723: 145 (1996). 14. G. A. Lord, D. B. Gordon, L. W. Tetler, and C. M. Carr, J. Chromatogr. A, 700: 27 (1995). 15. K. Schmeer, B. Behnke, and E. Bayer, Anal. Chem., 67: 3656 (1995). 16. N. W. Smith and M. B. Evans, Chromatographia, 38: 649 (1994). 17. P. S. Cahill and R. M. Whiteman, Anal. Chem., 67: 2599 (1995). 18. K. Pihel, S. Hsieh, J. W. Jorgenson, and R. M. Wightman, Anal. Chem., 67: 4514 (1995). 19. D. S. Gilman and A. G. Ewing, Anal. Chem., 67: 58 (1995). 20. R. T. Kennedy, M. D. Oates, B. R. Cooper, B. Nickerson, and J. W. Jorgenson, Science, 246: 57 (1989). 21. O. Orwar, H. A. Fishman, N. E. Ziv, R. H. Scheller, and R. N. Zare, Anal. Chem., 67: 4261 (1995). 22. S. M. Lunte and T. J. O’Shea, Electrophoresis, 15: 79 (1994). 23. F. Robert, L. Bert, L. Denoroy, and B. Renaud, Anal. Chem., 67: 1838 (1995). 24. A. J. G. Mank and E. S. Yeung, J. Chromatogr. A, 708: 309 (1995).
242 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45.
/ Rose et al. S. A. Shippy, J. A. Jankowski, and J. V. Sweedler, Anal. Chim. Acta, 307: 163 (1995). E. A. Arriaga, Y. Zhang, and N. J. Dovichi, Anal. Chim. Acta, 299: 319 (1995). A. Engstro¨m, P. E. Andersson, and B. Josefsson, Anal. Chem., 67: 3018 (1995). G. Nouadje, H. Rubie, E. Chatelut, P. Canal, M. Nertz, P. Puig, and F. Couderc, J. Chromatogr. A, 717: 293 (1995). A. G. Ewing, J. M. Mesaros, and P. F. Gavin, Anal. Chem., 66: 527A (1994). D. M. Morgan and S. G. Weber, Anal. Chem., 56: 2560 (1984). I. S. Krull, C. M. Selavka, C. Duda, and W. Jacobs, J. Liq. Chromatogr., 8: 2845 (1985). L. Canevari, R. Vieira, M. Aldegunde, and F. Dagani, Anal. Biochem., 205: 137 (1992). S. Murai, H. Saito, E. Abe, Y. Masuda, and T. Itoh, J. Neurochem. Trans., 87: 145 (1992). J. A. Hoskins, S. B. Holliday, and F. F. Davies, J. Chromatogr., 375: 129 (1986). L. A. Allison, G. S. Mayer, and R. E. Shoup, Anal. Chem., 56: 1089 (1984). V. Rizzo, A. Anesi, L. Montalbetti, G. Bellantoni, R. Trotti, and G. V. M. d’Eril, J. Chromatogr. A, 729: 181 (1996). Y. Qu, L. Arckens, E. Vandenbussche, S. Geeraerts, and F. Vandesande, J. Chromatogr. A, 798: 19 (1998). J. Kehr, J. Chromatogr. B, 708: 27 (1998). S. Murai, H. Nagahama, H. Saito, H. Miyate, Y. Masuda, and T. Itoh, J. Pharm. Meth., 21: 115 (1989). S. Murai, H. Saito, H. Nagahama, H. Miyate, Y. Masuda, and T. Itoh, J. Chromatogr., 497: 363 (1989). S. M. Lasley, R. D. Greenland, and I. A. Michaelson, Life Sci., 35: 1921 (1984). C. A. Costa, G. C. Trivelato, M. Demasi, and E. J. H. Bechara, J. Chromatogr. B, 695: 245 (1997). G. Lovell and P. H. Corran, J. Chromatogr., 525: 287 (1990). E. Morier-Teissier, K. Drieu, and R. Rips, J. Liq. Chromatogr., 11: 1627 (1988). P. G. Zambonin, A. Guerrieri, T. Rotunno, and F. Palmisano, Anal. Chim. Acta, 251: 101 (1991).
Transformation of Analytes for Electrochemical Detection / 243 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68.
S. Murai, H. Saito, Y. Masuda, and T. Itoh, J. Pharm. Meth., 23: 195 (1990). C. Chabenet, P. Ladure, D. Blanc-Continsouza, F. Boismare, and P. Boucly, J. Chromatogr., 414: 417 (1987). T. B. Jensen and P. D. Marley, J. Chromatogr. B, 670: 199 (1995). G. Achilli, G. P. Cellerino, and G. M. d’Eril, J. Chromatogr. A, 661: 201 (1994). A. D. Woolfson, J. S. Millership, and E. F. I. A. Karim, Analyst, 112: 1421 (1987). P. Leroy, A. Nicolas, and A. Moreau, J. Chromatogr., 282: 561 (1983). A. B. Naini, E. Vontzalidou, and L. J. Coˆte´, Clin. Chem., 39(2): 247 (1993). B. W. Boyd and R. T. Kennedy, Analyst, 123: 2119 (1998). I. Smolders, S. Sarre, Y. Michotte, and G. Ebinger, J. Neurosci. Meth., 57: 47 (1995). J. S. Soblosky, L. L. Colgin, C. M. Parish, J. F. Davidson, and M. E. Carey, J. Chromatogr. B, 712: 31 (1998). L. Millerioux, M. Brault, V. Gualano, and A. Mignot, J. Chromatogr. A, 729: 309 (1996). S. Smith and T. Sharp, J. Chromatogr. B, 652: 228 (1994). J. B. Phillips and B. M. Cox, J. Neurosci. Meth., 75: 207 (1997). W. A. Jacobs, J. Chromatogr., 392: 435 (1987). G. Turia´k and L. Volicer, J. Chromatogr., 668: 323 (1994). H. L. Rowley, K. F. Martin, and C. A. Marsden, J. Neurosci. Meth., 57: 93 (1995). E. Morier-Teissier, N. Mestdagh, J.-L. Bernier, and J.-P. He´nichart, J. Liq. Chromatogr., 16: 573 (1993). W. Buchberger and K. Winsauer, Anal. Chim. Acta, 196: 251 (1987). M. D. Oates and J. W. Jorgenson, Anal. Chem., 61: 432 (1989). W. F. Kline and B. K. Matuszewski, J. Chromatogr., 583: 183 (1992). W. L. Caudill, G. P. Houck, and R. M. Wightman, J. Chromatogr., 227: 331 (1982). W. L. Caudill and R. M. Wightman, Anal. Chim. Acta, 141: 269 (1982). W. A. Jacobs and P. T. Kissinger, J. Liq. Chromatogr., 5: 881 (1982).
244 69.
/ Rose et al.
P. Betto, G. Ricciarello, S. Pichini, L. D. Strologo, and G. Rizzoni, J. Chromatogr., 584: 256 (1992). 70. R. A. Sherwood, A. C. Titheradge, and D. A. Richards, J. Chromatogr., 528: 293 (1990). 71. R. A. Sherwood, J. Neurosci. Meth., 34: 17 (1990). 72. R. A. Sherwood, E. M. Bayliss, and O. Chappatte, Clin. Chim. Acta, 203: 275 (1991). 73. T. J. Mahachi, R. M. Carlson, and D. P. Poe, J. Chromatogr., 298: 279 (1984). 74. M.-Y. Chang, L.-R. Chen, X.-D. Ding, C. M. Selevaka, and I. S. Krull, J. Chrom. Sci., 25: 460 (1987). 75. K. Shimada, Y. Kawai, T. Oe, and T. Nambara, J. Liq. Chromatogr., 12: 359 (1989). 76. K. A. Jacobson, T. Marshall, K. Mine, K. L. Kirk, and M. Linnoila, FEBS Lett., 188: 307 (1985). 77. K. Mine, K. A. Jacobson, K. L. Kirk, Y. Kitajima, and M. Linnoila, Anal. Biochem., 152: 127 (1986). 78. K. Shimada, M. Tanaka, and T. Nambara, Chem. Pharm. Bull., 27: 2259 (1979). 79. K. Shimada, T. Oe, M. Tanaka, and T. Nambara, J. Chromatogr., 487: 247 (1989). 80. M. Tanaka, K. Shimada, and T. Nambara, J. Chromatogr., 292: 410 (1984). 81. L. Dou and I. S. Krull, Anal. Chem., 62: 2599 (1990). 82. L. Dou and I. S. Krull, J. Pharm. Biomed. Anal., 8: 493 (1990). 83. A. D. Kaufman, P. T. Kissinger, and J. E. Jones, Anal. Chim. Acta, 356: 177 (1997). 84. P. E. Kester and N. D. Danielson, Chromatographia, 18: 125 (1984). 85. M. Roth, Anal. Chem., 43: 880 (1971). 86. M. Joseph and P. H. Davies, J. Chromatogr., 277: 125 (1983). 87. J. F. Stobaugh, A. J. Repta, L. A. Sternson, and K. W. Garren, Anal. Biochem., 135: 495 (1983). 88. W. A. Jacobs, M. W. Leburg, and E. J. Madaj, Anal. Biochem., 156: 334 (1986). 89. P. deMontigny, J. F. Stobaugh, R. S. Givens, R. G. Carlson, K. Srinivasachar, L. A. Sternson, and T. Higuchi, Anal. Chem., 59: 1096 (1987). 90. B. K. Matuszewski, R. S. Givens, K. Srinivasachar, R. G. Carlson, and T. Higuchi, Anal. Chem., 59: 1102 (1987).
Transformation of Analytes for Electrochemical Detection / 245 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114.
R. G. Carlson, K. Srinivasachar, R. S. Givens, and B. K. Matuszewski, J. Org. Chem., 51: 3978 (1986). S. M. Lunte, T. Mohabbat, O. S. Wong, and T. Kuwana, Anal. Biochem., 178: 202 (1989). S. M. Lunte and O. S. Wong, LC GC, 7: 908 (1989). F. Kristjansson, A. Thakur, and J. F. Stobaugh, Anal. Chim. Acta, 262: 209 (1992). M. A. Nussbaum, J. E. Przedwiechi, D. U. Staerk, S. M. Lunte, and C. M. Riley, Anal. Chem., 64: 1259 (1992). Y. Nakahara, A. Ishigami, and Y. Takeda, J. Chromatogr., 489: 371 (1989). R. M. Smith, A. A. Ghani, D. G. Haverty, G. S. Bament, A. Y. Chamsi, and A. G. Fogg, J. Chromatogr., 455: 349 (1988). R. W. Welch, I. Acworth, and M. Levine, Anal. Biochem., 210(1): 199 (1993). J. R. G. Lomba, E. P. Gil, P. C. Moreno, R. M. G.-M. Carra, and A. S. Misiego, J. Electroanal. Chem., 410: 87 (1996). J. Leube and G. Fischer, J. Chromatogr. B, 665: 373 (1995). D. G. Musson and L. A. Sternson, J. Chromatogr., 188: 159 (1980). R. Wintersteiger, M. H. Barary, F. A. El-Yazbi, S. M. Sabry, and A.-A. M. Wahbi, Anal. Chim. Acta, 306: 273 (1995). P. Leroy and A. Nicolas, J. Chromatogr., 317: 513 (1984). R. L. Cox, T. W. Schneider, and M. D. Koppang, Anal. Chim. Acta, 262: 145 (1992). O. Orwar, S. Folestad, S. Einarsson, P. Andine´, and M. Sandberg, J. Chromatogr., 566: 39 (1991). S. G. Weber, H. Tsai, and M. Sandberg, J. Chromatogr., 638: 1 (1993). H. Tsai and S. Weber, J. Chromatogr., 515: 451 (1990). H. Tsai and S. G. Weber, Anal. Chem., 64: 2897 (1992). A. M. Warner and S. G. Weber, Anal. Chem., 61: 2664 (1989). H. Tsai and S. G. Weber, J. Chromatogr., 542: 345 (1991). M. Deacon, T. J. O’Shea, S. M. Lunte, and M. R. Smyth, J. Chromatogr. A, 652: 377 (1993). J. L. Meek, J. Chromatogr., 266: 401 (1983). L. Chen, J. Mazzeo, I. S. Krull, and S.-L. Wu, J. Pharm. Biomed. Anal., 11: 999 (1993). K. Shimada, M. Tanaka, and T. Nambara, Anal. Chim. Acta, 147: 375 (1983).
246 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135.
/ Rose et al. K. Shimada, M. Tanaka, T. Nambara, and Y. Imai, J. Pharm, Sci., 73: 119 (1984). Y. Imai, S. Ito, and K. Fujita, J. Chromatogr., 420: 404 (1987). K. Shimada, M. Tanaka, T. Nambara, Y. Imai, K. Abe, and K. Yoshinaga, J. Chromatogr., 227: 445 (1982). K. Shimada, T. Oe, and T. Nambara, J. Chromatogr., 419: 17 (1987). K. Shimada and K. Mitamura, J. Chromatogr. B., 659: 227 (1994). C. M. Selavka I. S. Krull, and K. Bratin, J. Pharm. Biomed. Anal., 4: 83 (1986). R. Wintersteiger and M. J. Sepulveda, Anal. Chim. Acta, 273: 383 (1993). K. Shimada, S. Orii, M. Tanaka, and T. Nambara, J. Chromatogr., 352: 329 (1986). M. Nakajima, H. Wakabayashi, S. Yamato, and K. Shimada, J. Chromatogr., 641: 176 (1993). C.-Y. Li and M. W. Kemp, J. Chromatogr., 455: 241 (1988). G. Egginger, E. Blaschke, W. Lindner, and A.-M. Olsson, J. Chromatogr., 666: 275 (1994). K. Gamoh, H. Sawamoto, S. Kakatsuto, Y. Watabe, and H. Arimoto, J. Chromatogr., 515: 227 (1990). K. K. Kaysinger, J. William M. Pierce, and D. E. Nerland, Anal. Biochem., 222: 81 (1994). W. A. Jacobs and P. T. Kissinger, J. Liq. Chromatogr., 5: 669 (1982). E. Bousquet, S. Tirendi, O. Prezzavento, and F. Tateo, J. Liq. Chromatogr., 18: 1933 (1995). G. Chiavari and C. Bergamini, J. Chromatogr., 318: 427 (1985). K. Shimada, M. Tanaka, and T. Nambara, Anal. Lett., 13: 1129 (1980). K. Shimada, M. Tanaka, and T. Nambara, J. Chromatogr., 307: 23 (1984). R. K. Munns, J. E. Roybal, W. Shimoda, and J. A. Hurlbut, J. Chromatogr., 442: 209 (1988). E. Bousquet, V. Cavrini, R. Gatti, and A. Spadaro, J. Liq. Chromatog. Rel. Technol., 21: 2873 (1998). E. Bousquet, N. A. Santagati, and S. Tirendi, J. Liq. Chromatogr. Rel. Technol., 20: 757 (1997).
Transformation of Analytes for Electrochemical Detection / 247 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161.
K. Shimada, C. Sakayori, and T. Nambara, J. Liq. Chromatogr., 10: 2177 (1987). S. Ikenoya, O. Hiroshima, M. Ohmae, and K. Kawabe, Chem. Pharm. Bull., 28: 2941 (1980). K. Shimada, E. Haniuda, T. Oe, and T. Nambara, J. Liq. Chromatogr., 10: 3161 (1987). K. Shimada, E. Nagashima, S. Orii, and T. Nambara, J. Pharm. Biomed. Anal., 5: 361 (1987). J. Knospe, D. Steinhilber, T. Hermann, and H. J. Roth, J. Chromatogr., 442: 444 (1988). T. Toyo’oka, J. Chromatogr. B, 671: 91 (1995). Y. Yasaka and M. Tanaka, J. Chromatogr. B, 659: 139 (1994). M. Lookabaugh and I. S. Krull, J. Chromatogr., 452: 295 (1988). L. Dou and I. S. Krull, J. Chromatogr., 499: 685 (1990). L. E. Leon, A. Ri´os, M. D. L. deCastro, and M. Valca´rcel, Anal. Chim. Acta, 234: 227 (1990). Y. Nagaosa and T. Mizuyuki, J. Liq. Chromatogr., 18: 3139 (1995). J. Zhou, L. Zhang, and E. Wang, J. Chromatogr., 619: 103 (1993). H. Ge and G. G. Wallace, Anal. Chem., 60: 830 (1988). A. M. Bond and T. P. Majewski, Anal. Chem., 61: 1494 (1989). A. M. Bond and G. G. Wallace, Anal. Chem., 54: 1706 (1982). A. M. Bond and G. G. Wallace, Anal. Chem., 55: 718 (1983). A. M. Bond and G. G. Wallace, Anal. Chem., 53: 1209 (1981). A. M. Bond and G. G. Wallace, Anal. Chem., 56: 2085 (1984). A. M. Bond and Y. Nagaosa, Anal. Chim. Acta, 178: 197 (1985). Y. Nagaosa, H. Kawabe, and A. M. Bond, Anal. Chem., 63: 28 (1991). C. M. Selavaka, I. S. Krull, and I. S. Lurie, J. Chromatogr. Sci., 23: 499 (1985). C. M. Selavka and I. S. Krull, Anal. Chem., 59: 2699 (1987). C. M. Selavka and I. S. Krull, Anal. Chem., 59: 2704 (1987). C. M. Selavka, K.-S. Jiao, I. S. Krull, P. Sheih, W. Yu, and M. Wolf, Anal. Chem., 60: 250 (1988). N. Tyrefors and P. G. Gillberg, J. Chromatogr., 423: 85 (1987). G. Damsma, B. H. C. Westerlink, and A. S. Horn, J. Neurochem., 45: 1649 (1985).
248 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174.
/ Rose et al. K. Fujimori and K. Yamamoto, J. Chromatogr., 414: 167 (1987). P. E. Potter, J. L. Meek, and N. H. Neff, J. Neurochem., 41: 188 (1983). T. Yao and M. Sato, Anal. Chim. Acta, 172: 371 (1985). N. Kaneda, M. Asano, and T. Nagatsu, J. Chromatogr., 360: 211 (1986). C. Eva, M. Hadjiconstantinou, N. H. Neff, and J. L. Meek, Anal. Biochem., 143: 320 (1984). T.-R. Tsai, T.-M. Cham, K.-C. Chen, C.-F. Chen, and T.-H. Tsai, J. Chromatogr. B, 678: 151 (1996). H. Wakabayashi, M. Nakajima, S. Yamato, and K. Shimada, J. Chromatogr., 573: 154 (1992). M. Dalene, L. Mathiasson, G. Skarping, C. Sango¨, and J. F. Sandstro¨m, J. Chromatogr., 435: 469 (1988). S. D. Meyer and D. E. Tallman, Anal. Chim. Acta, 146: 227 (1983). S. Honda, T. Konishi, and S. Suzuki, J. Chromatogr., 299: 245 (1984). J. R. Mazzeo, I. S. Krull, and P. T. Kissinger, J. Chromatogr., 550: 585 (1990). M. J. Rose, S. M. Lunte, R. G. Carlson, and J. F. Stobaugh, Anal. Chem., 71: 2221 (1999). M. J. Rose, J. M. Rose, S. M. Lunte, K. L. Audus, R. G. Carlson, and J. F. Stobaugh, Anal. Chim. Acta, 394: 299 (1999).
7 High-Performance Liquid Chromatography: Trace Metal Determination and Speciation Corrado Sarzanini Department of Analytical Chemistry, University of Turin, Turin, Italy
I. INTRODUCTION II. SAMPLE HANDLING A. Matrix Removal and Analyte Preconcentration B. Derivatization III. CHROMATOGRAPHIC MODES A. Normal and Reversed-Phase Chromatography B. Ion Chromatography C. Ion-Interaction Chromatography D. Chelation Ion Chromatography IV. METAL SPECIATION A. Arsenic B. Selenium C. Lead D. Mercury E. Tin F. Chromium REFERENCES
250 250 251 254 255 255 260 272 280 283 284 288 290 291 293 295 298
249
250
/ Sarzanini
I. INTRODUCTION Trace metal determination and speciation have received particular attention in the last years, mainly with respect to environmental samples and industrial products analysis. Liquid chromatography (LC) has become one of the main powerful analytical tools for the analysis of complex matrices (e.g., foods, new materials, pharmaceutical and environmental samples) and speciation studies, in the field of metal analysis. LC has stimulated studies on new more selective materials, enabling difficult separations, and for the improvement of detection sensitivity and selectivity. Spectrophotometric or electrochemical detection has been coupled with postcolumn reactions, but a new approach, hyphenation, concerning the coupling of unconventional detectors with LC, has become the emerging field of research. Some general reviews or comments with particular regard to ion chromatography, complexation ion chromatography, as well as advances in detection techniques summarize LC potentialities and will be introduced hereafter. In this chapter, with respect to the new philosophy of simple removal of interferents and/or analyte preconcentration before separation and specific detections of metal ions, a brief introduction to sample handling is presented. Classical and more recent applications of liquid chromatography for the analysis of metal species are presented, and the different approaches are discussed with particular reference to ion-exchange, ion-pair, and chelation separation mechanisms.
II.
SAMPLE HANDLING
The main practical problems in real sample analysis are related to sample handling and include sample collection, dissolution, cleanup, trace enrichment, and matrix elimination. Sampling and sample storage will not be considered here, but a few words must be devoted to sample treatments, due to the need for matrix removal, sample preconcentration, and/or derivatization. Sample cleanup can be performed off-line, prior to chromatographic analysis, or on-line, incorporated into the chromatographic hardware. Among the cleanup procedures, those using ion-exchange mechanisms can be used to reduce the alkalinity or the acidity of a sample [1] by using a highcapacity cation-exchange resin in the H ⫹ form or a high-capacity
HPLC: Trace Metal Determination
/ 251
anion-exchange resin in the OH ⫺ form, respectively. Alternatively, electrochemical devices can be used [2,3]. Similar procedures can be designed to suit different sample types, by varying the form of the resin used to achieve an alternative chemical modification of the sample, e.g., reversed phase (see below) for removing nonpolar compounds. The cleanup resins are often commercially available as disposable cartridges, offering a rapid and versatile sample pretreatment.
A. Matrix Removal and Analyte Preconcentration Different approaches can be followed [4] in order to remove the matrix and preconcentrate the analytes: separations can be static or dynamic. The chemical techniques used in preconcentration must provide analyte isolation as well as enrichment through minimal sample manipulation in order to avoid contamination and to obtain low sample blanks. Such procedures are usually defined off-line or on-line, involving sample treatments which are performed independently of or in direct connection with the analytical instrumentation. Among the main methods—evaporation, precipitation, co-precipitation, flotation, extraction (homogeneous or heterogeneous), sorption (adsorption, chelation on functionalized supports, ion-exchange), electrochemical methods, and special techniques (e.g., hydride generation)—particular attention has been devoted to the broadly named liquid–liquid or liquid–solid extractions (solid-phase extraction, SPE), probably the most effective of the techniques employed nowadays. From a practical point of view, the liquid–liquid extraction can be performed following two procedures: ionizable compounds can be extracted into organic solvents as ‘‘neutral’’ ion pairs in the presence of an appropriate ion-pairing agent; or hydrophylic compounds, which are difficult to remove from the aqueous phase, can be extracted by forming a hydrophobic complex with an appropriate complexing agent and the reaction-may be used for a selective isolation. The most widely employed extractants for metal ions form neutral chelates which have greater affinity for organic solvents than for the aqueous phase; the metal determination following this approach can be performed by direct analysis of the organic phase. The liquid–liquid extraction methods may be used for selective or group separation of trace elements, collected in the extract, or for matrix removal. These methods have the advantage of simplicity and rapidity, but the low concentration factors achieved are one of
252
/ Sarzanini
their major drawbacks. The extraction of compounds from a solution onto a solid phase (SPE) using silica, alumina, celite, charcoal, polymeric, or ion-exchange resins has long been practiced. Silica gels or polymers have been bound with a wide variety of functional groups (e.g., alkyl, phenyl, amino, ciano, diol, alkylsulfonate, and quaternary ammonium groups), and more recently with chelating functional groups (e.g., carboxylic, quinolines, iminodiacetic) to provide a specific interaction with analytes. No attempts are made hereafter to define foams, resins, or sorbents of inorganic or organic nature. It must be mentioned that if silica-based supports are used, the residual silanol sites provide, for the analyte, a second type of binding sites, so it is advisable to use as supporting material in the column either a silica C18 end-capped type or a polymeric-based C18 derivatized support. For SPE the acting mechanisms, for the retention of metal ions and their species, coupled with the different kinds of materials, are: Adsorption of neutral species as such or obtained via interaction with hydrophobic groups formed by adding a proper ligand able to originate neutral complexes with the analytes or adding an ion-interaction reagent able to originate ion pairs, with hydrophobic properties, with metal ions or their charged complexes Chelation obtained by reaction between metal ions and coordinating groups grafted on the solid Ion-exchange performed in two ways: cation exchange, which involves a direct exchange of the metal ion with the counterion of the resin and its retention; or anion exchange, which requires the preliminary addition of a ligand (usually sulfonated or phosphonated) able to form a negatively charged complex with the analyte From a practical point of view, liquid–solid extraction is commonly performed by driving the sample through a column (microcolumn, cartridge, microtube, or other kind of container) packed with the proper stationary phase, namely, a preconcentrator. When complexation is required in on-line systems, the sample of interest is injected into the manifold and merged with the chelating agent at a confluence point downstream. After passing through a mixing coil, which allows enough time for the chelate to form, the metal chelate
HPLC: Trace Metal Determination
/ 253
is preconcentrated on the column. This procedure is superior to the direct addition of the chelating agent to the sample, since in this case a purification cartridge can be added before mixing sample and ligand. To overcome the problems arising from off-line complexation and preconcentration (e.g., sample poisoning, extraction procedures), great effort has been devoted to develop on-line preconcentration procedures and, for on-line high-performance liquid chromatography (HPLC) applications the preconcentrator could be inserted instead of the injection loop. Obviously, in consideration of the mechanism and materials selected, the parameters to be kept under control will be proper pH for complex formation (precipitation must be avoided); ligand or ion-interaction reagent concentration and solubility (sufficient to originate complex or ion pair but not so high as to compete for resin sites); ionic strength (mainly for ion-exchange competition); and, from a practical point of view, elution velocity of the sample must be the maximum allowed by absorbtion, chelation, or ion-exchange kinetics utilized for metal-ion retention. In all cases, attention must be paid to ensuring the compatibility of the preconcentration step (the strength of the retention of analytes onto the preconcentrator) with the composition of the eluent used for the subsequent recovery of the analytes and their separation. Strictly related to the above-mentioned procedures, based on SPE, multidimensional liquid chromatographic techniques prove to be successful in matrix removal and trace separation and involve the use of two or more columns. The multidimensional technique is also known as ‘‘heart cut’’ column switching. The term ‘‘column switching’’ includes all techniques by which the direction of the flow of the mobile phase is changed by valves so that the effluent (or a portion of it) from the primary column is passed to a secondary column for a defined period of time. The column-switching technique has also been used in ion chromatography (IC) to solve the column overloading problems encountered when difficult samples have to be analyzed. In this case, matrix elimination is effected mechanically [5] rather than chemically via sample pretreatment. The technique can employ two four-way valves inserted before and after a precolumn (e.g., guard column, separator column, or an entirely different type of column from the separator column, such as a reversed-phase column). By configuring the valves in such a way that the bulk of the matrix is diverted to waste and only a heart cut of the analyte of interest is transferred to the separator column, the system is effec-
254
/ Sarzanini
Fig. 1 Schematic diagram of column switching. Initial valve position: the sample flows through the column CS12A and then into the CS10. Valve switched: the eluent goes into CS10 and then into CS12A. [Reprinted from M. A. Rey, J. M. Riviello, and C. A. Pohl, Column switching for difficult cation separations, J. Chromatogr. A, 789: 149–155 (1997), with permission from Elsevier Science.]
tive in either eliminating or at least simplifying sample preparation [6] (see Fig. 1). It must be emphasized that a great number of papers, referring to column switching or multidimensional liquid chromatography, concern methods that do not actually use two chromatographic columns, but rather use simply a short column (preconcentrator) coupled with an analytical column.
B.
Derivatization
Derivatization is an alternative approach to solve the mentioned problems. Derivatization is a chemical reaction able to modify the nature of the analytes or of the matrix, and can be usefully employed both pre- and postcolumn. In the first case the metal species are usually reacted, e.g., with a chelating agent that enables their separation by means of the selected mode. In the latter case the reaction is used for enabling analyte detection or lowering the detection limits; that is, e.g., for metal ions detectable with poor sensitivity by conductometric detection, a reaction with metallochromic ligands
HPLC: Trace Metal Determination
/ 255
may enable their UV-visible detection with higher sensitivity. In this case a postcolumn reactor (PCR) ‘‘reaction coil,’’ where the reaction takes place, is inserted between the analytical column and the detector and reached by the eluent and the reagent solution. The main requirements are (1) reduced dead volumes due to tubes, connections, and reactor; and (2) the reaction of derivatization must be very fast, in order to reduce the time spent inside the reactor and to avoid peak diffusion. A particular case of ‘‘derivatization’’ is represented by the suppressed conductometric detection in IC. To enhance the detection sensitivity, the contribution to the conducibility due to the eluent is ‘‘suppressed’’ through a chemical reaction (the main components of the eluent are transformed into water, weak acids, or weak bases). To define a border for the term ‘‘derivatization,’’ some other more specific approaches, e.g., hydride generation or ion generation, followed in the new techniques involving the coupling (hyphenation) of HPLC with ‘‘unconventional’’ detectors (e.g., atomic spectroscopy, mass spectrometry, inductively coupled plasma-mass spectrometry), will be discussed hereafter.
III. CHROMATOGRAPHIC MODES Several modes of HPLC can be performed for the analysis of metal species, including normal-phase chromatography (NPC), reversedphase chromatography (RPC), ion-pair or ion-interaction chromatography (IPC, IIC), ion chromatography (IC) (cation and anion exchange, ion exclusion), chelation ion chromatography (CIC), and their couplings—multidimensional and multimode chromatography. The basic principles, retention mechanisms (processes involving solute interactions in both the mobile and stationary phases), materials, and applications are summarized hereafter.
A. Normal and Reversed-Phase Chromatography Both these two modes of HPLC are used for the separation of metal chelates and organometallics, so each will be discussed below. Normal-phase chromatography (NPC) is characterized by the use of an inorganic adsorbent or chemically bonded stationary phase with polar functional groups (e.g., silica, cyanoalkyl-silica) and a nonaqueous mobile phase (one or more polar solvents diluted to the
256
/ Sarzanini
desired eluting power). The mechanism acting in the NPC is a liquid–liquid partition, and both retention and selectivity are dramatically influenced by the presence of polar additives (water) in the mobile phase. Reversed-phase chromatography (RPC) employs nonpolar solids of high surface area (usually alkyl-bonded silica packing, e.g., C 8 or C 18 groups grafted to the silica surface) as stationary phase and an aqueous–organic solvent mixture as mobile phase. Retention in RPC occurs by nonspecific hydrophobic interactions of the solute with the stationary phase. Secondary chemical equilibria (see below) optimize separation selectivity by varying the mobile-phase composition. The solute retention is attributed to both adsorption and partition phenomena, while neutral and ionic solutes can be separated simultaneously. The current approach for normal-phase chromatography (NPC) is based on the formation of metal chelates (e.g., diacetylbisthiobenzhydrazones, dithizone, diethyldithiocarbamate), their extraction from the sample, their injection and elution, performed with organic solvent mixtures of n-heptane/benzene, toluene, diethyl ether/acetonitrile or similar. A typical reaction for heavy-metal ions is with ammonium tetramethylenedithiocarbamate (ATDC) [7], which gives neutral complexes. These kinds of chelates can be separated in the same manner as neutral compounds on normal-phase silica columns (e.g., 1% propanol in hexane). The main problems encountered in NPC are connected with the thermodynamic stability of the metal chelates and solubility, the need for extraction, and the risk of pollution of the sample when traces must be detected. Among the mentioned mechanisms, normal-phase chromatography is not at present the most widely used. RPC has been widely used for the separation of neutral or weakly charged metal complexes, but the more extensive applications are based on ion-pairing mechanism, so related procedures will be detailed hereafter. Analysis for trace metals is carried out by the formation of metal chelates with separation by RPC on C18 columns and the use of organic-based mobile phases. Dithiocarbamates are the most frequently reported complexing agents, due to the strong chelating ability of their sulfur groups and their ability to form nearly water-insoluble metal salts with all metals except sodium and other alkali and alkaline earth metals. Dilli et al. [8] completed a comprehensive study on reversed-phase HPLC behavior of diethyl-
HPLC: Trace Metal Determination
/ 257
dithiocarbamate (DEDTC) complexes of Cu, Co, Cr, Ni, and Hg with a variety of columns and mobile phases. In this case DEDTC complexes were preformed off-column (60°C, 15 min), extracted into chloroform, and finally dissolved in CH 3OH and injected for the separation onto a C18-column (µBondapak, Waters). The study showed that the ligand must also be present in the mobile phase for low concentration of chelates, to avoid their dissociation. In addition, the better suitability of a water–methanol eluent with respect to a methanol–acetonitrile–water mixture has been demonstrated. In this way one avoids interference due to the involvement of acetonitrile in ternary complexes formation. A similar procedure was developed for Co and Ni DEDTC-chelate separation on dimethyloctadecylsilyl-bonded silica. The preconcentration of chelates by a double extraction (diethyl ether followed by methanol) coupled with UV detection allowed the attainment of 5 and 50 µg/L detection limits for Co and Ni, respectively [9]. More recently a detailed study of the RP-LC behavior of metals (Zn, As, Fe, Cd, Pb, Ni, Cu, Hg, Co, and Cr) complexed by reaction with a homologous series of five dialkyldithiocarbamate ligands, of general formula R 2NCS 2⫺, has been reported [10]. A single C18-column (µBondapak, Waters) and either methanol–water or methanol–acetonitrile– water mobile phase proved suitable for the separation of metals with the exception of As, Fe, and Zn. The ternary system was superior for higher homologs. Studies to overcome the main drawback of an extraction step, due to the low solubility of DEDTC complexes, were also made, and a thiosemicarbazone was synthesized that was able to form soluble complexes [11]. 2-Acetylpyridine-4-ethyl-3-thiosemicarbazone complexes of Co, Cd, Fe, Ga, Ni, In, and Zn were separated on a PS-DVB column and the mechanism was found to be a mixed-mode one (RP and IIC) since Co, Fe, and In complexes are positively charged and the best separation was achieved by adding NaClO 4 to the CH 3 –H 2O mixture. The problem arising from the presence of more than one coordination form for the corresponding central ion for nitrogen–oxygen coordinated complexes (which produces a general peak broadening) has been tentatively solved by Ming et al. [12]. Their work uses a precolumn and on-column derivatization for the separation of V(V), Co, Fe, and Ni through the formation of binary and ternary peroxo complexes with 4-(2-pyridylazo)-resorcinol (PAR) and H 2O 2. The study showed that, with methanol–water eluent and a RP col-
258
/ Sarzanini
umn, V(V)–PAR binary complexes yield two peaks which are converted into a V–PAR–peroxo complex single peak if H 2O 2 is added to the sample within PAR before elution. 2-(5-bromopyridylazo)-5diethylaminophenol (5-Br-PADAP) is another 2-pyridylazo complexing reagent that is highly sensitive and selective. Studies for the separation and determination of metal ions [Cu, Co, Fe, Ni, V(V), Pd] as 5-Br-PADAP chelates by RP-HPLC showed that only the retention of Co(III)-5-Br-PADAP complex is affected by varying the concentration of surfactant added to the eluent [13]. In addition, a stronger interaction resulted for tetrabutylammonium (TBA) with respect to cetyltrimethylammonium (CTA) and cetylpyridinium (CP) cations. The alkyl group, such as that of CTA or TBA, may interact with the C18 chain on the stationary phase by molecular interaction so that the charged part of the surfactant is exposed on the surface, increasing its polarity so that the Co chelate is eluted earlier. Following this approach, a selective preconcentration method with a cationexchange resin for RP-HPLC of the Co-5-Br-PADAP complex was recently developed [14]. Co complex, in aqueous solution, is readily oxidized to the Co(III)-5-Br-PADAP inert cationic complex, which is retained on a sulfonated XAD-4 resin. Co is detected spectrophotometrically (588 nm) after elution onto a C18 analytical column (Capcell SG-120) with a methanol–water eluent added of ethylenediaminetetraacetic acid (EDTA) and TBA and without 5-Br-PADAP. The absence of 5-Br-PADAP favors the dissociation e.g., of Cu and Zn chelates, and other metal ions eluted later, such as Fe and Ni, do not interfere. The detection limit for Co in water samples is reported to be 5.9 ng/L. Various azo dyes have also been considered for the chromatographic separation of metal chelates on a reversed-phase RP-18 column, and the study focused on the separation and determination of V(V) at trace levels in natural waters [15]. The originality of this investigation is due to the optimization of the RP column selectivity by introducing a tetraalkylammonium salt into the system. The metal chalates considered are neutral or cationic and ion-paired complexes are not involved, whereas other metal ions (e.g., Fe, Al) do not interfere in the determination. 8-Quinolinol (HQ) is another extensively used ligand for the separation of metal ions by HPLC. For this ligand the methods are also based on metal-ion complexation, usually by heating the sample in the presence of HQ, one or two extraction steps with an eluent-com-
HPLC: Trace Metal Determination
/ 259
patible solvent, and injection of complexes into the chromatographic system. As an example, the simultaneous determination of Mo(VI), V(V), Cu(II), and Fe(III) at the parts-per-billion level in sea water can be mentioned [16]. To overcome the problems arising from offline complexation and preconcentration (e.g., sample poisoning, time consuming) a column-switching technique has been proposed [17]. In the mentioned study two compatible eluents of different eluotropic strength were selected, one (CH 3 CN/H 2O) to concentrate the metal– HQ complexes onto a precolumn (Nucleosil C18) and the second (CH 3CN/H 2O/HQ) to elute the analytes from the precolumn onto the analytical column (C18). The linear dynamic range is from 5 ppb to 10 ppm for Al and from 40 ppb to 5 ppm for Cu and Fe. LC has been shown to be a powerful technique for the determination of trace noble metals, and among the different ligands employed, thiazolylazo reagents play a remarkable role in RP-HPLC. Studies on the retention of Pd, Pt, Rh, and Ru chelates of 1-(2-thiazolylazo)-2-naphthol (TAN) enabled the determination of Pd and Ru [18]. Basova et al. [19] separated the 4-(2-thiazolylazo)resorcinol (TAR) complexes of Rh, Ru, Cu, and Co on an ODS column with acetonitrile. Separation and determination of Rh, Ru, and Os chelates with TAR was also reported [20]. Saraswati et al. [21] used TAR as a chelating reagent in the reversed-phase HPLC separation of transition metals from rare earth elements in low-alloy steels by increasing the concentration of the eluent (octane–1-sulfonate–tartaric acid). For noble metals more satisfactory results have been obtained using a new thiazolylazo reagent, 2-(6-methyl-2-benzothiazolylazo)-5-diethylaminophenol (MBTAE) [22]. MBTAE complexes (Pt, Os, Ni, Co and Rh, Pd or Ir, Ru) were separated and determined by RP-HPLC using both C18 and C8 columns with methanol–n-butanol–water or methanol–water eluents, respectively [23,24]. Several papers focusing on the analysis of the isomers of metal complexes have been published, and recent advances in HPLC have made available columns in which the active solid phase is chiral, for the separation of enantiomers. Akama et al. [25] achieved the identification of two geometric isomers of the chromium(III) complex with 1-phenyl-3-methyl-4-benzoyl-5-pyrazolone, through HPLC continuous-flow, fast—atom bombardment, mass spectrometry. Acetonitrile–water and n-hexane–tetrahydrofuran as the mobile phase for an ODS and a silica gel column were used. Carreri et al. [26] have shown the NP and RP HPLC potentiality in a comprehensive
260
/ Sarzanini
study on seven iron–carbonyl compounds substituted with alkynes, R 2C 2. Several stationary phases, including silica, CN, C18, and phenyl columns, coupled with positive-ion and negative-ion chemical ionization mass spectra of the compounds obtained with particlebeam mass spectrometry (MS), were considered for the separation and the structural elucidation of organometallic compounds. Finally, a comprehensive review has recently been published [27] on normalphase (NP) and reversed-phase (RP) HPLC modes for the analysis of organometallic compounds and metal coordination compounds, in the synthesis and reaction of metal complexes, in studies of the kinetics and mechanism of the complexation of the metal complexes, and in bioinorganic chemistry.
B.
Ion Chromatography
The term ion chromatography (IC) referred in the past to ionexchange chromatography only. Ion-exchange chromatography is basically a simple process based on a reversible interchange of ions between a solution and a solid, inorganic or polymeric insoluble material, containing fixed ions and exchangeable counterions. Analytes are separated on the basis of their different relative affinities for the ionic centers of the stationary phase. With respect to the use of the conductometric detection, two ion chromatographic modes are defined: suppressed and nonsuppressed. In the first mode, before detection, the eluate is driven through the suppressor unit, where the background conductivity is greatly reduced, so that the sensitivity with which sample ions can be detected is increased. Nonsuppressed ion chromatography is performed without the use of a suppressor unit, with ion-exchange resins of low exchange capacity and very dilute eluents, so that the background conductivity is quite low. In the cation-exchange technique the metal ions are normally reacted with an anion of a weak acid to reduce their charge density in the eluent solution before entering the separation column, where they are separated owing to their respective affinities toward the active sites of the separating resin. Ligands are also required to avoid precipitation when an acidic eluent is not suitable for the selected columns. Ion chromatographic separation by anionic exchange of metal ions involves their presence as negatively charged complexes, which can be obtained in two ways. The first way, off-line, is through their formation before the chromatographic separation (precolumn complexation—complexes must be stable enough to
HPLC: Trace Metal Determination
/ 261
avoid decomposition during separation or ligand must be added to the eluent). The second way, on-line, is based on the complexation in the chromatographic column itself, by adding the proper ligand to the eluent. Ion chromatographic separation of metal ions based on anionic exchange offers the potentiality of different selectivity, reduced problems for metal-ion hydrolysis, and application to complex sample matrices. Therefore, ion-exchange chromatography is based on electrostatic interactions between the ions to be exchanged, but other reactions may occur, e.g., hydrophobic reactions between the sample and nonionic regions of the stationary phase, or additional reagents are intentionally introduced in order to achieve or optimize the separation. This means that in IC, in addition to the ion-exchange reaction, the secondary chemical equilibria (SCE) play a relevant role [28]. Figure 2 shows a schematic illustration of the equilibria existing on a cation- or an anion-exchanger between a solute cation (M 2⫹ ), an added ligand (H 2L), and an eluent cation (enH ⫹ ). In the first case, see Fig. 2a, the complexation due to the deprotonated ligand moderates the cation exchange through a pulling effect. Figure 2b represents the equilibria involved for the retention of an anionic complex, where the excess of the ligand itself (deprotonated) or another anion added to the eluent, competes for the fixed sites. Finally, Fig. 2c shows the pushing and pulling coupled effects of the ligand and of the competing cation, where complexation reduces the availability of the free M 2⫹ for the exchange. It must be underlined that in all cases the eluent pH is a determining parameter, since it is acting both on the ligand dissociation and on the cation protonation. Metal ions can be retained on silica-based ion-exchange columns and silica itself can act as both anion and cation exchanger. Ion exchangers produced by chemically bonding ion-exchange groups to a silica backbone were the main type of early column packings, pellicular ion exchangers formed by coating a silica core with a polymeric ion-exchange material. Polymer-coated silica cation exchangers for IC can be synthesized by depositing varying film thicknesses of a prepolymer onto porous silica; immobilization is achieved by in-situ cross-linking reactions using radical starters or radiation. Poly(butadiene-maleic acid) (PDMA) is the preferred polymer. An advantage of silica-based materials is the low probability of secondary interactions between solute ions and the silica substrate and higher column capacities over synthetic materials. On the other hand, serious
262
/ Sarzanini
Fig. 2 Secondary chemical equilibria in cation- (a, c) or anion- (b) exchange chromatography. Metal ion M 2⫹ as such or as an anion complex: (a) cation exchange, the ligand (L 2⫺ ) exerts the pulling effect on the cation analyte (M 2⫹ ) through the complexation; (b) anion exchange, the deprotonated ligand (L 2⫺ ) competes (pushing effect) with the anion complex (ML 2 2⫺ ); (c) cation exchange, pulling and pushing effects of the ligand and of a competing cation on the retention of M 2⫹.
HPLC: Trace Metal Determination
/ 263
drawbacks exist with silica-based materials. Both low and high pH values must be avoided: below pH 2, covalent bonds linking the ionexchange functionality become unstable and the functional groups are cleaved; while above pH 8, silica matrix may be dissolved. When IC was first introduced [29], column packing for cation determinations consisted of surface-sulfonated 25-µm polystyrene beads cross-linked with 2% divinylbenzene, and they were used with packed-bed suppressors. Due to the large differences in selectivity of alkaline earths toward alkali cations, HCl was needed to elute the monovalent cations, while stronger divalent eluent components (e.g., m-phenylenediamine) were required for the elution of divalent cations. Due to the long column equilibration time between the two eluent systems, it has proven difficult to provide simple isocratic elution to allow the separation of both classes of cations in a reasonable period of time, so that two different columns were dedicated for the two classes of cations. Nevertheless, chromatograms were characterized by quite long total run times and by poor peak efficiency. Heavy metals have also been separated on these kinds of materials with diluted strong acids and/or organic complexing agents. The development of new methods of synthesis for ion chromatography (latex-coated columns, IonPac CS3, Dionex Co.), together with the replacement of m-phenylenediamine by the zwitterion 2,3-diaminopropionic acid monochloride [30], made it possible to simultaneously analyze both the alkali and alkaline earth elements, in the presence of ammonium, in one column, and to improve peak efficiencies, although analysis still required a long time and baseline resolution was not completely achieved. A different kind of column containing carboxylate functionalities, instead of the traditional sulfonate ones with a low selectivity for the hydronium ion, the so-called Schomburg column, was introduced by Kolla et al. [31]. This stationary phase, based on a poly(butadiene-maleic acid) (PBMA) co-polymer silica gel coated, coupled with eluents containing slightly acidic complexing agents (e.g., tartaric acid), was used to separate Li ⫹, Na ⫹, NH 4 ⫹, K ⫹, Mg 2⫹, Ca 2⫹, Sr 2⫹, and Ba 2⫹ in nonsuppressed IC. The unique selectivity of the carboxylic groups, and the competition of the chelating agent in the eluent with the cation-exchange sites for divalent cations only, provided good separation and short analysis times, the only limitation being the operative pH range imposed by the silica material. The performance of PBMA silica columns for separation of alkali, alkaline earth, and heavy- or transition-metal ions
264
/ Sarzanini
has been studied in the presence of different organic complexing agents (α-hydroxyisobutyric, tartaric, citric, oxalic, pyridine-2,6dicarboxylic acid, and EDTA) and applied to determine cations in water, food, ore, sole brine, and other real samples [32,33]. Common reversed stationary phases, permanently coated with suitable hydrophobic agents such as alkylsulfonates or alkylsulfates, have also been used for the separation of transition and heavy metals [34,35]. A great improvement in metal-ion chromatography was obtained with new synthetic packing materials basically characterized by an ethylvinylbenzene substrate cross-linked with 55% divinylbenzene surrounded by solvent compatible functional groups (sulfonic, carboxylic/phosphonic, carboxylic/phosphonic/crown ether), (IonPac Columns, Dionex Corporation, Sunnyvale, CA, USA). The above-mentioned polymers in some cases (e.g., IonPac CS5 and CS5A) bear both quaternary ammonium and sulfonate functional groups, in a pellicular layer located on the core of the beads. The presence of both cation- and anion-exchange groups enables more sophisticated separations: (1) Sun et al. [36] speculated on this property and separated Ge and Sn, by pure ion exchange, as cation and anion, respectively. Metals were determined simultaneously on a IonPac CS5 by working at the proper pH (⬍1.5), ensuring the presence of HGeO 3 ⫺ and Sn 2⫹ species; (2) Motellier et al. [37] developed a liquid chromatographic method with an on-line preconcentration step (IDA chelating column) using pyridine-2,6-dicarboxylic acid (PDCA), which forms negatively charged complexes with metals (Co, Cu, Ni, Mn, Zn), and a CS5 analytical column. They showed that the exchange mechanism for Ca on the analytical column is much different from that of transition metals (pure ion exchange of positively charged complexes). The method proved to be suitable for waters of low ionic strength. This kind of columns (IonPac CS 5A) allowed the simultaneous determination of heavy and transition metals with a gradient program (based on oxalic acid and sodium chloride eluents) coupled with: (1) spectrophotometric detection after postcolumn reaction with 4-(2-pyridylazo)resorcinol PAR, (2) 2-[(5-bromo-2pyridyl)azo]5-diethyl-aminophenolo. This procedure was used for biochemical samples (technique validated by five standard biochemical references) [38] and nitrate/phosphate fertilizer solutions (detection limits 1–30 ng) [39], respectively. The main difficulty which had to be overcome was the greatly
HPLC: Trace Metal Determination
/ 265
different selectivity of the ion-exchange resins, normally used to separate monovalent and divalent cations, respectively. These resins were characterized by a much larger selectivity for divalent cations than for monovalent cations, thus divalent cations have greater retention times than monovalent ones. As an alternative way to simultaneously determine mono and divalent cations, a column-switching (dual-column) technique was reported [40]. Samples are eluted through two separate columns, each optimized for the chromatography of one group of cations. Separation is accomplished by switching the eluent flow paths from one column to the other. This method has several advantages compared to gradient techniques (e.g., shorter run times, no equilibration period between runs). Nevertheless, this technique requires sophisticated instrumentation, and sensitivity for divalent cations is poor. Column-switching techniques between latex-based IonPac CG3 and CS10 columns have been used by Betti et al. [41] to determine alkali and alkaline earth elements in sea water. Transition and heavy metals have been determined in sea-water samples by multidimensional liquid chromatography [42]. The separation was achieved by coupling a dynamically coated sorbent column, a preconcentrator, and a chelating column. The metal concentrations (Cu, Ni, Co, Mg, Ca, Sr, Fe and Pb, Zn, Mn, Cd) in sea-water samples were 5.0–50 µg/L. As discussed in depth, samples containing alkali metals, alkaline earth cations, and the ammonium ion are difficult to analyze. Environmental samples, at low levels of ammonium in matrices with a high concentration of sodium, are a typical case. This is due mainly to the similar selectivities of ammonium and sodium ions for the common stationary phases containing sulfonate or carboxylate cation-exchange functional groups. This problem has been solved by a column-switching technique which enables the determination of trace concentrations of the common inorganic cations (Li, Na, K, Mg, Ca) and ammonium in the presence of large concentrations of either sodium or ammonium [43]. Figure 3 shows, step by step, the isocratic elution (eluent: methanesulfonic acid) of monovalent and divalent cations during a column switching between a carboxylated (IonPAc CS 12A) and a sulfonated column at decreased cation-exchange site density (IonPAc CS 10), which provides the needed selectivity. Considering the experimental data available for the separation of these cations, the best separations were achieved through the use of polymer-based columns containing carboxylic (IonPac CS12,
266
/ Sarzanini
Fig. 3 Step-by-step isocratic elution of monovalent and divalent cations by column switching. After elution of monovalent cation from CS12A to CS10 column, the valves are switched and the order of columns is reversed. [Reprinted from M. A. Rey, J. M. Riviello, and C. A. Pohl, Column switching for difficult cation separations, J. Chromatogr., 789: 149–155 (1997), with permission from Elsevier Science.]
CS14) [44,45] or carboxylic and phosphonic (IonPac CS12A) [46,47], or carboxylic/phosphonic/crown ether (IonPac CS15) [48] functionalities. Figure 4 clearly shows the different selectivity achieved with IonPac CS 12A (a) and IonPac CS 15 (b), due to the presence of an 18-crown-6 ether group (permanently attached on the macroporous substrate beads of the later column). Since the development of new materials for mono- and divalent cations, expert systems for the planning of appropriate dilutions, suitable detector output ranges, and standard additions were developed for alkali and alkaline earth metals in mineral waters [49]. Moreover, applications of the optimized systems (CS12 and methansulfonic eluents) for the determination of Na ⫹, K ⫹, Mg 2⫹, and Ca 2⫹ in melted-snow samples from alpine sites [50] and for the determination of elements in Antarctic samples [51], have been shown. The high capacity which characterizes some of these stationary phases provided the means to develop a method based on high-volume direct injection for trace-level determination [52]. Satisfactory results for the determination of Na ⫹, K ⫹, Mg 2⫹, and
HPLC: Trace Metal Determination
/ 267
Fig. 4 Comparison of selectivities achieved by carboxylic/phosphonic groups (IonPac CS 12A) and carboxylic/phosphonic/crown ether groups (IonPac CS15). (a) Column; IonPac CG 12A and CS 12A; eluent, 15 mM sulfuric acid; flow rate, 1.0 mL/min; peaks, 1 ⫽ Li ⫹, 2 ⫽ Na ⫹, 3 ⫽ NH 4 ⫹, 4 ⫽ K ⫹, 5 ⫽ Mg 2⫹, 6 ⫽ Ca 2⫹. (b) Column, IonPac CG 15 and CS 15; eluent, 15 mM methanesulfonic acid ⫹ 7.5 mM hydroxylamine ⫹ 5% acetonitrile (40°C); flow rate, 1.2 mL/min; peaks, 1 ⫽ Li ⫹, 2 ⫽ Na ⫹, 3 ⫽ NH 4 ⫹, 4 ⫽ Mg 2⫹, 5 ⫽ Ca 2⫹, 6 ⫽ K ⫹. [Reprinted from M. A. Rey, C. A. Pohl, J. J. Jagodzinski, E. Q. Kaiser, and J. M. Riviello, A new approach to dealing with high-tolow concentration ratios of sodium and ammonium ions in ion chromatography, J. Chromatogr., 804: 201–209 (1998), with permission from Elsevier Science.]
268
/ Sarzanini
Ca 2⫹ in a well water samples were also obtained using a IC Pak CM/ D (Waters) column with an eluent containing citric acid and pyridine-2,6-dicarboxylic acid (PDCA) [53]. Tartaric and dipicolinic acids have also been used for mono- and divalent cations [54]. Ohta et al. [55] studied the cation-exchange properties of a commercially available unmodified silica gel (Develosil 30-5) in the acidic region. They obtained the simultaneous separation and determination of Na ⫹, NH 4⫹, K ⫹, Mg 2⫹, and Ca 2⫹ in environmental water samples by coupling nitric acid with a selective complexation (PDCA) of divalent cations. They also demonstrate the main cause of cation exchange attributed to aluminum present as an impurity in the silica gel [56]. More recently, the modification of a silica gel by a coating method with Al enabled the cation-exchange separation of mono- and divalent cations [57]. A synthesized weak cation exchanger (co-polymerization of vinyl groups covalently bound onto silica surfaces with acrylic acid) has been used as stationary phase and the pure ion-exchange mechanism is modified by adding 18-crown-6 ether and acetonitrile to the mobile phase [58] for alkali and alkali earth cation separation. Highly sensitive and excellent separation for these cations, in various natural samples, was achieved (run time 15 min) using an unmodified silica gel and 1 mM oxalic acid/3 mM 18-crown-6 eluent [59]. It seems of interest to mention that Dumont and Fritz [60] showed that increase in the separation factors and change in the elution order of alkali metal cations could be obtained using nonaqueous solvents with macroporous cation-exchange resins (low-capacity PS-DVB sulfonated resin). This behavior is due to the change in solvated ionic radii for Li and Na, so that higher resolution is obtained in the separation of ions that usually elute close together (Li/Na, K/NH 4⫹). The use of lithium ion and neutral pH or methanol and lithium chloride as the eluent gave very low detection sensitivities and poor separation of mono- and divalent cations on porous silica [61–63]. More recently, the separation of six-inorganic monovalent cations was achieved on a porous silica gel microcolumn with a 30% (v/v) acetonitrile and benzyltrimethylammonium (BTMA) chloride in the mobile phase [64]. This result could be explained through a reduction of the hydrophobic interaction between cations and BETMA introduced onto the surface of silica through cationexchange and hydrophobic interactions. The advantage of the
HPLC: Trace Metal Determination
/ 269
method is that BTMA cation both competes with analytes for the exchange sites and is used for their indirect UV detection. Cation exchange performed on silica using weakly or neutral eluents gives very good sensitivities for cations. Ion chromatographic separation of metal ions based on anionic exchange offers the potential of different selectivities, reduced problems for metal-ion hydrolysis, and can be applied to complex sample matrices. Notwithstanding the fact that many organic acids (from mono-, di-, tricarboxylic acids to chelating agents such as α-hydroxyisobutyric acid, tartaric, citric, oxalic, pyridine-2,6-dicarboxylic acid, 1,2-diaminocyclohexanetetraacetic acid, diethylenetriaminopentaacetic acid) have been used for the simultaneous ion chromatography of anions, alkali, alkaline earth, and heavy metals [65–67], ethylenediamminetetraacetic acid (EDTA) plays a fundamental role. Since EDTA forms, at the proper pH, negatively charged complexes with divalent or trivalent metal ions, the possibility of simultaneous separation of anions from metal ions is also feasible, as well as the speciation of metal ions. In these procedures, such complexes can be obtained both off-line and on-line. Some examples of applications of an EDTA eluent have been reported for the determination of anions and divalent cations in natural and pharmaceutical samples [68]. Both UV and conductivity detection were used, working at pH values of 6–8. Experiments were also performed with binary eluent systems comprising EDTA as complexing agent. An UV absorption reagent was used to enhance detection limits for Ca 2⫹ and Mg 2⫹ [69,70]. For sea-water samples [71], silica-based anion-exchange analytical columns enhanced sensitivity and enabled detection limits from 20 µg/L for Mg 2⫹ to 0.4 mg/L for Ca 2⫹ with UV and conductivity detection and eluent pH at 4.8. Komarova et al. [72] studied the ion chromatographic behavior of anionic EDTA complexes of vanadium (V) and (IV). As mentioned above, the ion chromatographic determination of metal–EDTA complexes can be performed with anion-exchange columns [68,71,73–76]. In this way, anions and metals can be separated as anionic complexes in a single run. In addition, this approach enables specific determinations; e.g., Se(VI) was determined (detection limit 4.8 µg/L) in real samples by suppressed anion chromatography in the presence of anions and heavy-metal ions [74]. Alternatively, the separation of the metal–EDTA complexes can be carried out with a cation-exchange column [66]. In this case the retention
270
/ Sarzanini
mechanism of analytes involves the cation exchange of free metal ions (e.g., Cu, Fe, Zn, Ni, Pb, Mn, alkali, and alkaline earth metals) which are present at low pH values. Detailed studies on the behavior of metal–EDTA complexes in cation chromatography as a function, e.g., of pH and ionic strength, provides data that can be used to develop selective models for understanding the contributions of different charged or neutral species on the retention mechanism involved. A selective method for preconcentrating and determining Pb at trace levels, 0.5 µg/L detection limit, was developed based on these considerations [77]. Recently, the simultaneous determination of common inorganic anions, magnesium and calcium ions, in various environmental waters has been achieved by indirect UV-photometric detection ion chromatography coupling trimellitic acid–EDTA eluent with an anion-exchange column (TSKguardgel QAE-SW) [78]. Theoretical approaches to the retention of anionic metal complexes have been developed for anion exchange [75,76,79] and for cation exchange [77,80]. Ion chromatography, as well as reversed-phase liquid chromatography, is used increasingly for the separation of lanthanide metals, but classical ion-exchange techniques do not always provide satisfactory separation. An approach to rare earth separation has been proposed by Strelow et al. [81], who evaluated the distribution coefficients for yttrium and some lanthanides between AG 50W-X4 resin and hydroxyethylenediaminetriacetate in monochloroacetate buffer solution, the results have been applied to the quantitative separation of yttrium and neodimium from samarium and heavier lanthanides. In these conditions, the control of pH is essential because the peak positions are very sensitive to change in this parameter. Silica gelbased sulfonated-form cation-exchange gel and sulfonated polystyrene gel columns were used with lactic (LA) and α-hydroxyisobutyric (HIBA) acids by Kawabata et al. [82]. A gradient elution with HIBA or LA enabled them to overcome interferences due to polyatomic ions and isobars in ICP-MS detection of rare earths; e.g., GdO ⫹ and GdOH ⫹ overlap all the isotopes of Yb and Lu, and LaH ⫹ hinders the free isotope of Ce. The method gave a complete separation of all rare earths within 1–5 ng/L detection limits, and the method has been successfully applied to the analysis of Tm, Yb, and Lu impurities in a Gd matrix and Ce in pure La 2O 3 [82]. The Dowex 50 ⫻ 8 cation-exchange resin has been used by
HPLC: Trace Metal Determination
/ 271
Farin˜as et al.[83] as a tool to separate rare earth elements, as a group, in geological materials from the matrix elements, using increasing HCl concentrations (from 2 to 6 M ). The determination of lanthanides is finally performed by ICP spectrometry. The separation of each rare earth element can be performed either with a cation- or anion-exchange mechanism, according to the eluent composition and the properties of the stationary phase. Cation exchange of lanthanides has been performed in pellicular, latexagglomerated columns (IonPac CS3 [84] and CS10 [85]) in the presence of appropriate chelating agents such as HIBA. Lanthanides form complexes with HIBA that lower the affinity of the lanthanide for the cation-exchange resin. The elements are eluted according to the stability of the complexes formed, with Lu (the most stable complex with HIBA) eluting first and La (the weaker complex with HIBA) eluting later. In the anion-exchange mechanism, lanthanides have been predominantly separated using a mixed-bed column (IonPac CS5 and, more recently, CS5A [86–89]) containing both anion- and cation-exchange sites. Different complexing agents can be used in a gradient mode, as single components or in a mixture, to obtain a baseline resolution among each rare earth element. The first nine elements (La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, and Dy) have been determined as impurities in a YbF 3 matrix, used in the production of optical fibers, by coupling oxalate and diglycolate as ligands in the eluent with a mixed-mode ion-exchange packing (IonPac CS5) [88]. Through the optimization of the method [89] the co-elution of Ho and Er was avoided, enabling the determination of 11 elements in the presence of excess of Yb in the matrix (see Fig. 5). Detail was also given on the different contributions to the retention mechanism (anion or cation exchange) when such a mixed-bed column is used, showing the secondary chemical equilibria involved [89] (see Fig. 6). Finally, an example of a very limited number of papers dealing with ion-exclusion chromatography applications to metal determinations is given. Tanaka et al. [90] have developed an anion-exclusion method in nonsuppressed IC of mono- and divalent cations, establishing a simple technique to be applied in acid rain, river, underground, lake, and forest soil waters. The simultaneous separation of Na ⫹, NH 4 ⫹, K ⫹, and Mg 2⫹, Ca 2⫹ has been obtained with an anionexclusion chromatographic column packed with polymethacrylatebased weakly acidic cation-exchange resin in the H ⫹ form (TSKgel
272
/ Sarzanini
Fig. 5 Chromatogram of a YbF 3 sample spiked with lanthanides, 5 mg/L each. Stationary phase, IonPac CS5: Eluent gradient: t ⫽ 0 min, 80 mM oxalic acid; t ⫽ 15 min, 26 mM oxalic acid, 23 mM diglycolic acid; t ⫽ 22 min, 80 mM oxalic acid. Detection at 520 nm after postcolum derivatization with PAR. (From Ref. 89.)
OA-PAK-A 300 ⫻ 7.8 mm). An eluent containing 0.75 mM sulfuric acid, 2 mM tartaric acid, and 7.5% methanol allowed the separation of mono- and divalent cations in about 25 min.
C.
Ion-Interaction Chromatography
Ion-interaction chromatography (IIC), also called soap chromatography, ion-pair chromatography, and dynamic ion-exchange chromatography, is a typical example of a chromatographic process based on secondary equilibria. Two approaches are usually followed to perform an IIC separation of metal ions: 1. Eluents containing an ion-interaction reagent (IIR) (e.g., alkylammonium salts, alkyl sulfates, or alkylsulfonates) are used and stationary phases (conventional RP or polymers) are dynamically modified into low-capacity ion exchangers.
HPLC: Trace Metal Determination
/ 273
Fig. 6 Equilibria involved in lantanides (M 3⫹ ) separation in the presence of ligands (L 2⫺) in a mixed-bed column (IonPac CS5). (From Ref. 89.)
Elution of cations is achieved by their complexation and ion pairing of the negatively or positively charged complexes with IIR. During the separation, the retention of neutral analytes, analytes having the same or opposite charge with respect to the IIR, will be unaffected, decreased, or increased. Organic modifiers, and in some cases the complexing agent, are also added to the eluent. 2. Common reversed-phase stationary phases are permanently coated with suitable hydrophobic agents such as alkylsulfonates or alkylsulfates with a sufficiently long alkyl group. In this case stationary phases are preloaded with the proper, very lipophilic, IIR. The mechanism of elution is governed by the mobile phase in two ways: (a) eluents containing a strong driving cation and a small amount of complexing agent (‘‘push–pull’’ method, e.g., mobile phase containing ethylenediammonium cation and tartaric acid); (b) eluents containing a very weak driving cation and higher concentrations of complexing agent (‘‘pure pull’’ mechanism).
274
/ Sarzanini
Since pH can alter the stoichiometry of the complexes and their overall charge, resolution of chelates is greatly dependent on eluent acidity. Therefore, research in this field is devoted to evaluating the nature and concentration of proper ligands and IIR as well as organic and ionic strength modifiers and eluent pH. The wide number of parameters involved and governing retention reflect the great versatility of the system to manipulate selectivity and to obtain good separations even when complicated matrices have to be separated. Nevertheless, the number of variables involved turns out to be a drawback when the effect of each mobile-phase parameter has to be considered during the modeling of the ion-interaction mechanism, which indeed has been controversial to be interpretated since the introduction of ion-interaction chromatography in the early 1980s. The ion-pair formation interaction and the dynamic ion-exchange interaction modes are illustrated schematically in Fig. 7. Wide work on IIC ion-paired metal complexes of nitroso-naphthol sulfonates, with liquid–liquid extraction and on-line derivatization, has been done by Sire´n [91]. A modification of the proposed methods consists of an on-line derivatization of metal ions; in this procedure the metals are injected into a methanol–water eluent containing a quaternary ammonium bromide (e.g., cetyltrimethylammonium, CTA) and, after the column, they are mixed with a ligand solution (1-nitroso-2-naphthol-6-sulfonate) [92]. This procedure makes the metal-ion separation governed by the kinetics of formation of complexes and ion pairs and retention onto the postcolumn mixer–reactor system. On-column DEDTC chelates formation–preconcentration and RP-HPLC separation has been used for the determination of Cd, Cu, and Ni in sea-water samples [93]. A C2-bonded silica microcolumn, loaded with a dithiocarbamate-cetyltrimethylammonium ion pair, enabled the retention of metal ions and their complexation. Elution on ODS analytical columns was optimized by adding cetyltrimethylammonium (CTA) bromide to a CH 3 CN–H 2 O mixture, owing to the relative instability of Cd–DEDTC anionic complex which was eluted as a neutral ion pair. Azo ligands have also proven suitable for trace-metal ion separation and determinations. The chromatographic behavior of 3-(5-chloro2-hydroxyphenylazo)-4,5-dihydroxy-naphthalene-2,7-disulfonic acid (Plasmocorinth B) and its metal-ion complexes has been studied in ion-pairing reactions for metal preconcentration and separation by
HPLC: Trace Metal Determination
/ 275
Fig. 7 Schematic illustration of (a) ion-pair formation interaction and (b) dynamic ion-exchange interaction for an anionic solute (anion metal complex, ML ⫺ ) and a free metal ion (M ⫹ ).
HPLC [94] in the presence of tetrabutylammonium ion. The separation of analytes, optimized with a flow-gradient elution, was successfully applied to river-water samples, enabling analyte metals to be separated from alkaline and alkaline earth elements. A detailed study on the effects of the mobile-phase composition (IIR, counterion, organic modifier, pH) on IIC separation and electrochemical detection of metal complexes with SPADNS, Acid Alizarin Violet N, and Plasmocorinth B was made [95]. The optimized procedure enabled the determination of Cu, Co, Fe, Ni, and V at microgram-perliter levels in natural waters. The suitability of a ligand for metal-ion determinations in an ion-interaction chromatographic mechanism depends on the selectivity of the ligand for the metal cations considered. An approach to select the appropriate ligand has been followed
276
/ Sarzanini
Fig. 8 Molecular structure of the ligand Calcion. (From Ref. 96.)
by Sarzanini et al. [96]. In this approach, the determination of the thermodynamic and conditional stability constants of Cu 2⫹, Ni 2⫹, Fe 3⫹, and Al 3⫹ with cyclo-tris-7-(1-azo-8-hydroxynaphthalene-3,6disulfonic acid), calcion, or calcichrome (see Fig. 8), allowed selection of the proper experimental conditions for further application in the IIC separation. In this case pH was the key parameter for controlling selectivity and thus avoiding interferences by certain metal ions. The separation became specific for Al/Fe or Ni/Cu coupled cations by working at 4.5 or 7.5 pH values, respectively (Fig. 9). Metal ions have also been separated as anionic chelates with pyridylazosulfoaminophenol derivatives [97,98]. The retention behavior of these metal chelates in IIC has been elucidated as a function of eluent composition [99] with respect to the significant differences found in methanol–water and acetonitrile–water systems as a function of the volume fraction of water [100]. More recently, 2-(5nitro-2-pyridylazo)-5-[N-n-propyl-N-(3-sulfopropyl)amino]-phenol (nitro-PAPS) has been used as a precolumn chelating reagent for the simultaneous determination of V and Co, in steel samples, by IIC with an ODS column and an acetonitrile–water eluent containing TBA (IIR) and EDTA [101]. The method enabled detection limits of 17 ppt for V and 55 ppt for Co without preconcentration and with a spectrophotometric detection. Octadecyl-bonded silica permanently coated with sodium dodecylsulfate in the presence of complexing agents was considered for the separation of transition metals [102]. In that work an ion-exchange mechanism similar to that of fixed-site exchangers is shown
HPLC: Trace Metal Determination
/ 277
Fig. 9 Determination of Al 3⫹ and Fe 3⫹ at pH 4.5 (left) and Ni 2⫹ and Cu 2⫹ at pH 7.5 (right) in a tap-water sample. Eluent 53–47% (v/v) methanol in water containing 40 mM buffer, 25 mM TBA, 25 mM NaCl, and 5.0 ⫻ 10 ⫺7 M Calcion. Tap water as such (dotted line) and spiked with 1 ppm of each metal ions (solid line). (From Ref. 96.)
to occur, and both the pushing effect of the eluting cation and the pulling effect of the complexing anion are taking place but the latter plays a dominant role in the process of elution. A significative example of the mentioned approach is the detailed study of Cassidy and Sun [103]. They compared the performance of an anion separation with a cation separation, each based on an ion-interaction system that used cetylpyridinium chloride or n-octanesulphonate to modify a reversed stationary phase. In the first case, transition metals (Mn, Co, Ni, Cu, and Zn) were eluted with an oxalate eluent. The anionexchange system provided column efficiences comparable to that for the cation system. This approach may be attractive for solving analytical problems taking into account the considerably different order of separation obtained by the two systems. A recent study on chromatographic behavior of metal ions [Fe(III), Cu, Pb, Zn, Ni, Co, and
278
/ Sarzanini
Mn] in IIC when the stationary phase is modified with various alkanesulphonates (1–10 carbon atoms in the alkyl chain) must be mentioned [104]. Reversed-phase ion-pair procedures involving EDTA have also been considered in optimizing separation and detection of metal species. Different techniques such as precolumn derivatization without complexing agent in the eluent or on-column derivatization may be less efficient and give rise to peak broadening. Ion-pair reversedphase high-performance chromatography has been investigated by coupling EDTA with tetraethylammonium (TEA) [105], tetrapropylammonium (TPA) [106], and tetrabutylammonium (TBA) [105,107,1 08] bromide ion-pairing agents. TBA resulted the best suitable ionpairing agent in all cases, and the use of EDTA in the eluent [105,108], together with high complexation constants, shifted equilibrium in favor of chelate formation reaching lower detection limits. The data obtained [105] clarify some aspects of the separation mechanism of ion-interaction chromatography for metal ions in different oxidation states and confirm that the retention of divalent and trivalent metal ions complexed with EDTA takes place through an ionexchange mechanism in which the ion exchanger is dynamically generated by the retention of the counterion in the stationary phase [108]. Studies on the separation and determination of metal ions [Cu, Co, Fe, Ni, V(V), Pd] as Br-PADAP chelates 2-(5-Br-2-pyridylazo-5diethylaminophenol) by RP-HPLC showed that only the retention of the Co(III)-5-Br-PADAP complex is affected by variation of the concentration of the surfactant added to the eluent [109]. Stronger interactions are shown to result for tettrabutylammonium (TBA) with respect to cetyltrimethylammonium (CTA) and cetylpyridinium. An alkyl group, such as that of CTA or TBA, may interact with the C 18 chain on the stationary phase by molecular interaction so that the charged part of the surfactant is exposed on the surface, increasing its polarity so that the Co chelate is eluted earlier. Vachirapatama et al. [110] developed a method based on precolumn ternary complex formation of Nb and Ta with Br-PADAP and citrate. The selectivity of the separation of metal complexes, ion-paired with TBA, was optimized by the concentration of methanol and citrate in the mobile phase. The method, applied to geological materials, is characterized by detection limits of 30 and 410 ppt for Nb and Ta, respectively.
HPLC: Trace Metal Determination
/ 279
Dynamic ion exchange allows the use of bonded microparticulate alkyl silicas as nonpolar stationary phases with aqueous buffers containing low concentrations of hydrophobic ions for the separation or rare earth elements. Lanthanide separation was pioneered by Cassidy [111] using gradient elution with sodium octanesulfonate as an ion-interaction reagent, which provides virtual ion-exchange sites by adsorption on a nonpolar C 18 stationary phase with α-hydroxyisobutyric acid as complexing eluting component. Quantitation is achieved by postcolumn reaction with Arsenazo III and optical absorbance detection at 658 nm, with typical detection limits below 1 ng for each element. Alternatively, 2-(4-pyridylazo)resorcinol (PAR) can be used as postcolumn reagent. In most cases lanthanides need to be precomplexed with proper ligands and then separated. The most widely used ligands are lactic [112], glycolic [113,114], HIBA [115–118], and nitrilotriacetic [119] acids. In certain cases, the achievement of complete resolution among all the rare earths is difficult. Kuroda et al. [113,114] showed that glycolato complexes are eluted within 20–30 min, but their method suffers from incomplete resolution between Ho and Sm, Eu, Gd, Tb, and Dy, which co-elute. The same authors showed that better resolution and separation is achieved using lactate and laurylsulfate as hydrophobic ion in the eluent [112]. Dodecylsulfate [115,120] and 1-octanesulfonate [121] have been employed in the determination of some lanthanide elements in apatite, bastnesite, and monazite samples. A coupled column chromatographic procedure based on a semipreparative reversed-phase column and a dynamic cation-exchange column has been proposed by Lucy et al. [122]. This configuration has been applied to the determination of lanthanides in uranium matrix, due to the problem of degradation of neutron economy of nuclear reactors by impurities of rare earths. A C 18 column and HIBA eluent has been used to remove uranium, while a C 18 column loaded with C 20SO 4 and a HIBA gradient enabled lanthanide determination after postcolumn reaction with arsenazo III. Several examples of coupling both isocratic and gradient reversed-phase separations with ICP [123–125] and ICP-MS [126] detections are available. Theoretical approaches to the retention of anionic metal complexes have also been developed for IIC or dynamic ion exchange [127,128], and a comparison of prediction power between theoretical
280
/ Sarzanini
and neural-network models in ion-interaction chromatography of metal complexes, having single or double charges, has also been published [129].
D. Chelation Ion Chromatography Trace-metal analysis with ion chromatography procedures becomes very difficult with samples of high ionic strength such as concentrated brines or sea waters, as the ion-exchange sites can become ‘‘swamped’’ with salt ions or, in the case of ion-pair chromatography, the pairing mechanism is modified. For these kinds of samples the selectivity of the chromatographic separation can be enhanced by the use of chelation ion chromatography. Chelation chromatography involves both the formation of coordinate bonds and an ion-exchange process, due to free or protonated chelating groups which act as ionexchange sites, in the stationary phase and, in some cases, complexation in the eluent. Chelation ion chromatography or, more correctly, high-performance chelation ion chromatography (HPCIC), is based on the use of high-performance substrates for trace-metal separation and determination. Two main approaches can be followed to obtain proper stationary phases: (1) chemical bonding of the chelating group to the substrate; or (2) coating of a substrate with a ligand which is permanently trapped onto the substrate. Kantipuly et al. [130] illustrated the extensive range of chemically bonded chelating resins available for the separation and concentration of trace metals, among many combinations of chelating ligands and supporting materials. The different kind of stationary phases, available on the market or laboratory made, have recently been reviewed [131] and they are silica- or polymer-based materials with chelating agents grafted onto the surface. Alternatively, chelating ligands can be immobilized by adsorption onto a styrene–divinylbenzene co-polymer, silica gel, or other synthetic polymer. For the separation of metal ions, these precoated columns use an aqueous mobile phase at a relatively high concentration of an inorganic salt, e.g., 0.5–1.0 M KNO 3. The most widely used resin to preconcentrate and to separate elements and groups of elements in sea water, applied for the first time almost three decades ago [132], is an iminodiacetate resin (i.e., Chelex-100). Since this resin was successfully used in batch analysis for the determination of many metals in complex matrices, it was packed in a column form and eluent forced through at 1–2 mL/min.
HPLC: Trace Metal Determination
/ 281
Experimental results showed that only partial recoveries of some metal ions could be obtained under these conditions. Despite attempts to improve metal retention on Chelex-100 resin, it was finally concluded that the low recovery of metal ions was due to a physical degradation of the resin under pressure because of the low degree of cross-linking of the polystyrene–divinylbenzene supporting polymer. Among many combinations of chelating ligands and supporting materials, Bonn et al. [133,134] successfully characterized and used stationary phases with silica-bound iminodiacetic acid (IDA) functions. The chromatographic behavior of some elements (Mg, Fe, Co, Cd, and Zn) is optimized under a mixed-mode mechanism, ion exchange and chelate formation with eluents (e.g., citric or tartaric acid), but other ions (e.g., Cu, Ni, and Pb) are irreversibly retained and strong complexing agents such as pyridine-2,6-dicarboxylic acid (PDCA) are required for the complete elution [135]. A more highly cross-linked macroporous PS-DVB has been developed [136] with iminodiacetate functional groups that allow operations at high pressure without physical degradation (MetPac CC-1 column, Dionex). At pH 5.2–5.6, polyvalent metal ions are selectively concentrated into MetPac CC-1, alkali metals and anions are not retained, and a selective elution of alkaline earth metals can be achieved using ammonium acetate. Concentrated metals and lanthanides are eluted, with the exception of chromium, with acid into a cation-exchange column, acting as interface before the analytical column, from which they are successively driven to the ion chromatographic separation with a PDCA or an oxalate complexing eluent and detection is obtained after postcolumn derivatization with PAR. Detection limits range from 0.2 to 1 µg/L (Fe, Cu, Ni, Zn, Co, Mn, Cd, Pb) [136,137]. The procedure was successfully used for metal-ion determination in sea water from the Venice lagoon, and with 60-mL sample preconcentration the detection limits for Cu, Ni, Zn, Co, and Mn were lowered to 0.05–0.1 µg/L [138]. A modification of the above-mentioned procedure allowed metal recoveries from chelating resin with the same eluent used for the ion chromatographic separation (75 mM H 2SO 4 / 100 mM HCl/100 mM KCl) and with a cation-exchange column with higher capacity. In this way detection limits of 10 and 30 ng/L were achieved for Cd and Pb, respectively, in sea-water samples (200 mL) [139]. Voloschik et al. [140] used a silica gel-based sorbent with chemically bound amidoxime functional groups for the selective de-
282
/ Sarzanini
termination of metals (Mn, Pb, Cd, Co, Ni, Zn, Cu, and Hg) in waters, since this kind of resin showed a weak affinity for Mg and Ca and this eliminated their interference in the determination. The examples described above are in some cases mentioned as chelating chromatography but actually refere more than to an actual chelating ion chromatography, to application of chelating materials to preliminary separation and/or preconcentration of metals. The development of CIC, based on ligands chemically bound to silica or polymer phases, refers to a technique applied for a long time in metal-ion preconcentration and matrix removal before spectroscopic determination based on permanent loading of sorbents by chelating agents [141–144]. Jones and co-workers widely investigated this field in order to improve detectability of both alkaline earth and heavy metals. They showed that the chelating ability of Xylenol Orange and Chrome Azurol S coated onto polystyrene–divinylbenzene neutra hydrophobic substrates, changes markedly with different substrates [145]. Studies devoted to transition-metal determination [146–148] allowed the optimization of an interesting procedure ˚ pore size, impregbased on 10-µm-particle-size PS-DVB resin, 100 A nated with Xylenol Orange, stable from pH 0.5 to 11.5 [149]. The stationary phase enables the removal of Ca and Mg during the sample on-column preconcentration at pH 6, and the separation and determination of Zn, Pb, Ni, and Cu with a step-gradient pH elution. The only drawback of the method is the coelution of Cd–Mn, but satisfactory results are obtained for sea-water samples. This is a good example of how small variations in chelating ability between dyes can be very useful for specific separations. Using HPCIC the retention order of metal ions, including earth metals, is reversed with respect to IIC, and so barium is eluted first as a sharp peak, followed by Sr, Mg, and Ca. The barium separation and determination was successfully optimized in oil-well brines, at milligram-perliter concentrations in 1600-mg/L Ca samples [150]. A neutral hydrophobic resin (PS-DVB), first impregnated with methylthymol blue (3,3′-bis[N,N-di(carboxymethyl)aminomethyl]thymol-sulfonephthalein) was coupled with an acid elution (0.5 M KNO 3 ⫹ 0.5 M lactic acid) and UV detection (PCR: PAR ⫹ ZnEDTA). Detection limits of 1 mg/L for Ba and 8 mg/L for Sr, respectively, were reached. Also, this Sr and Ba separation was optimized in milk powder analysis with a column impregnated with phthalein purple (o-cresolphthalein-3′,3″-bis-methyleneiminodiacetic acid) [151]. A recent paper de-
HPLC: Trace Metal Determination
/ 283
tails an investigation into the parameters involved in the production of a range of dye-impregnated chelating columns (10 chelating dyes, mainly on triphenylmethane– or azo-based dyes) for preconcentration and separation of alkaline earth, transition metals, and heavy metals at trace levels within an application devoted to Al determination in sea water [152]. More recently, a porous graphitic carbon reversed-phase column has been used with a mobile phase containing a selective metallochromic ligand for the separation of alkaline earth metals (o-cresolphtalein complexone) [153]. To extend the range of metals separated by dynamically coated columns, a polystyrene– divinylbenzene reversed-phase column was precoated with methylthymol blue (MTB), a metallochromic ligand which complexes metal ions through two iminodiacetic acid functional groups [154]. This kind of column proved suitable for samples at ionic strengths up to 1.0 M NaCl and was highly selective for UO 2 2⫹. Large-volume injection (2.0 mL) and an eluent step-gradient procedure enabled the determination of UO 2 2⫹ without interference, at microgram-per-liter concentration in a saline lake sample spiked with 1 mg/L Mn, Cd, Zn, Pb, Ni, and Cu. Finally, Paull et al. [155], using a mobile phase containing MTB, developed dynamic chelating chromatography, in which the metallochromic ligand is included within the mobile phase. The new system, compared to MTB-precoated chelating columns [154], allows the separation of transition and heavy metals and shows different selectivity and improved peak shapes.
IV.
METAL SPECIATION
Metal speciation refers to the identification and quantitation of the different forms of a particular element, such as organometallic, chelated, or free forms, or different oxidation states in a particular sample. Since variations in the chemical form define the toxicity or essentiality of the analyte, this information is particularly crucial in environmental and toxicological investigations. Robarts et al. [156], about 10 years ago, reviewed more than 440 papers on metal speciation by liquid chromatography, but the growing interest in this field of research depends on the possibility of coupling the separative power of LC with more sensitive and selective detection techniques. Detection and quantification of the separated species at trace levels was generally performed by atomic spectroscopic techniques. When these are coupled with a separation system, information on
284
/ Sarzanini
speciation instead of the total concentration or amount of element(s) of interest is provided. Atomic absorption spectrometry can be used as a sensitive detector after chromatographic separation [157]. Among atomic emission techniques [158,159], the majority of applications have been based on inductively coupled plasma (ICP) emission spectroscopy [160,161]. This source is more suitable for LC, since the chromatographic flow rate is compatible with the conventional (i.e., pneumatic chamber) ICP interface. When used as a detector for HPLC, ICP offers good sensitivity, a dynamic range of over five orders of magnitude, and multielement detection capabilities. However, conventional direct connection HPLC-ICP coupling can suffer from poor transport efficiency, particularly when pneumatic nebulizers are used. Such coupling also demonstrates low tolerance for many of the organic solvents commonly employed in HPLC eluents. Investigations have therefore been performed to characterize the effect of mobile-phase composition and flow rate on HPLCICP methods such as new kinds of nebulizers, e.g., a microflow ultrasonic nebulizer or direct-injection nebulization (DIN), increasing transport efficiencies to the ICP interface. Finally, ICP has been used as a source for mass spectrometry (MS), and ICP-MS has become one of the most powerful techniques for speciation analysis when coupled with separation procedures [162–165]. As mentioned, inductively coupled plasma atomic emission spectrometry (ICP-AES) and inductively coupled plasma mass spectrometry are two different and sensitive methods used in trace-metal speciation. The lower limits of detection (subnanogram to subpicogram levels), the wide linear ranges and isotope analysis capability, and the high precisions (0.1–5.0% RSD) associated with ICP-MS make it more advantageous than ICP-AES. In addition, the capability of ICP-MS for isotope ratio determinations allows isotope dilution analysis to be used. More recently, a perspective on the current technology and applications of LC-MS has been published [166,167]. Some examples on speciation methods and advances for As, Se, Pb, Hg, and Cr will be detailed hereafter.
A.
Arsenic
The molecular forms of arsenic subject to speciation analysis are anions, e.g., arsenite As(III), arsenate As(V), monomethylarsonate (MMA), and dimethylarsinate (DMA); cations, e.g., arsenobetaine (AsB), arsenocholine (AsC), and tetramethylarsonium ion (TMAs);
HPLC: Trace Metal Determination
/ 285
or uncharged compounds at neutral pH, e.g., arsenous acid. There is no general rule or trend that relates toxicity with oxidation state or chemical form and number of substituents linked to the central atom, but As(III) is more toxic than As(V). Recently, arsenic speciation in humans and food products has been reviewed [168]. A detailed study on AsB, AsC, and TMAs cations was done by Blais et al. [169]. An evaluation of direct, ion-pairing, reversedphase, and cation-exchange chromatographic procedures involving different stationary phases and eluents has been made in order to optimize atomic absorption detection through a new on-line thermochemical hydride-generation interface (THG). Strong cation-exchange chromatography was incompatible with the THG interface, solving the problem of AsC-TMAs co-elution on C18 column. The best choice was a normal-phase HPLC approach: cyanopropyl stationary phase and a methanolic eluent containing a silanol masking agent. The optimized procedure gave detection limits of 13.3, 14.5, and 7.6 ng for AsB, AsC, and TMAs cations, respectively. A simple method [170] has been developed for As(III) and As(V) speciation. An extractive chromatography has been performed by eluting samples (pH 2.5) through a stationary phase, silanized diatomite, modified with dioctyltin dichloride (C 8H 17 ) 2SnCl 2 [171]. As(V) was selectively retained and As(III) passed through the column, As(V) recoveries (85–115%) and column regeneration were obtained with 2 M HCl. The arsenic contents in the eluates were determined either by flame AAS or graphite furnace AAS; detection limits were not given, but the linearity of the method was 0.01–0.20 µg/mL. Arsenic speciation was evaluated by separation on an anion- and a CAS1 ion-exchange column and off-line hydride-generation electrothermal atomic absorption spectroscopy, for As(III), As(V), AsB, MMA, DMA, p-aminophenylarsenate ( p-APA), and AsC [172]. In this study, As(III) co-eluted with AsB. A thermal decomposition procedure was introduced to convert organoarsenicals into As(V) before reduction by sodium borohydride, and detection limits resulted comparable to HPLC-HG-AAS [173] and HPLC-ICP-MS [174] procedures. Ion chromatographic methods for elemental speciation (As, Se, and Cr) using microbore columns with direct-injection nebulization by ICPAES have been described by Gjerde et al. [175]. An anion HPLC procedure with a phosphate eluent was developed for arsenite As(III), arsenate As(V), monomethylarsonate, di-
286
/ Sarzanini
methylarsinate, arsenobetaine, and arsenocholine separation [176]. After optimization, five peaks only were well resolved, due to the overlap of As(III) and AsB. The analytes were determined by hydride-generation atomic absorption spectrometry ((HG-AAS). The arsenic content of the unresolved peak was assigned to As(III) ⫹ AsB or As(III), with or without a microwave-assisted oxidation step [177,178] before HG-AAS determination on two consecutive injections. The method applied to urine samples gave, with a cleanup procedure, detection limits within 8–15 µg/L and 2.5–5.3% standard deviations. Similar studies were also done for DMA, AsB, and AsC in deionized water, urine, and cleanup dry residue from urine samples [179]. The stability of arsenic species, in relation to food treatment procedures (seafoods and mushrooms), has been investigated with an HPLC-UV-HG-AFS (AFS, atomic fluorescence spectrometry) method [180]. A separation with a strong anion-exchange resin (BAX-10) and a gradient eluent concentration (K 2SO 4) at 60°C has also been proposed for As(III), As(V), MMA, DMA, and AsB [181], but in this case a complete separation was not achieved due to the co-elution of inorganic species. Rauret et al. [182] developed an IC-HPLC procedure for As(III), As(V), MMA, and DMA separation and ICP-AES determination by coupling the systems with the hydride-generation sample introduction technique. The procedure was improved [183] by checking two different columns (Nucleosil-5SB and Hamilton PRP X-100) and by comparing isocratic and gradient elution. The peak profile was improved by filtering the data corresponding to low concentrations with a Fourier transform. With such a procedure, detection limits between 2.7 As(III) and 11.4 As(V) µg/L were obtained. Several reports have appeared on LC coupled with ICP-MS for metal speciation determination As [174,184–187]. A direct-injection nebulizer interface work [188,189] demonstrated that reversedphase microbore columns and eluents containing ion-pairing agents could be coupled with mass spectrometric (MS) detection for arsenic speciation. Good efficiency was also obtained using interfaces based on hydride-generation manifolds. Hydride generation was used to increase the efficiency of conventional pneumatic nebulizers in LCICP-MS studies of arsenic speciation [190,191] and to avoid interference from the polyatomic species which has the same m/z value as 75 As isotope. Even by using a membrane gas separator [192], the determination was subject to the interference of the molecular ion
HPLC: Trace Metal Determination
/ 287
ArCl ⫹. The removal of 40Ar 35Cl ⫹ interference was optimized [174,193] by coupling ion chromatography and ICP-MS detectors, and lowered detection limits resulted for As(III), As(V), DMA, and MMA with the use of an He–Ar gas mixture as ionization source (0.032 and 0.080 ng for DMA and MMA, respectively). A separation of As(III), As(V), MMA, and DMA was performed with reversed-phase ion-pair liquid chromatography (IIR: TBA– phosphate) [194]. The eluate was delivered to the hydride-generation system after a prereduction with L-cysteine at 95°C in diluted nitric acid. In this way ArCl ⫹ was removed and conventional detection limits were reduced to 11–51 ng/L values. Seven molecular forms of arsenic [As(III), As(V), MMA, DMA, AsB, AsC, and tetramethyarsonium, TMAs] were separated, by anion- and cation-exchange HPLC, and detected by on-line flame atomic absorption spectrometry (AAS). The potential interference due to phosphorus on the 193.7-nm arsenic line is avoided because it is separated by an anion-exchange HPLC procedure [187]. Silicabased and polymeric cation-exchange columns were examined, and complete separation of cationic species and DMA was achieved by isocratic elution (pyridine eluent) on a bare silica column (E Si EOH groups act as cation-exchange sites). The retention of DMA at low pH was attributed to the presence of DMA ⫹ species. In a further development of the method, Larsen et al. [195] studied the determination of eight arsenic compounds in urine with anion- and cationexchange separations by coupled HPLC-ICP-MS. The sample is silica C18 cleaned and introduced into the system; four anionic [As(III), As(V), MMA, and DMA] and four cationic [AsB, AsC, TMAs, and trimethylarsine oxide (TMAO)] arsenic compounds were detected at m/z 75, since chloride is separated chromatographically and does not interfere. In the cation-exchange chromatographic system the chloride eluted with the void volume, and in the anion-exchange procedure is eluted more than 100 s later than the last analyte peak, well separated from the arsenic species. TMAO, not included in the previous work, eluted in the cation-exchange system due to protonation of the AsCO bond in the acidic mobile phase. A pH value of 10.3 in anion separation enabled the retention of arsenous acid, which is eluted with the void volume in other anion-exchange chromatographic systems [196]. The procedure [195] has been applied to investigate the speciation of arsenic in seafood samples and to elucidate the biosynthetic pathway involved in marine metabolism
288
/ Sarzanini
[197]. In this field, Harrington et al. [198] described extractions and cleanup procedures to isolate water-soluble arsenic species in marine brown algae for their determination by both GF AAS and HPLCICP-MS. More recently [199], a speciation of these compounds was obtained by micellar liquid chromatography coupled with ICP-MS detection. The method, based on a micellar mobile phase (CTAB, propanol, and borate buffer), and on a PRP-1 separation column coupled with the ICP-MS system, allowed linear dynamic ranges of three orders of magnitude and detection limits in the picogram range (90–300) and overcame the problems of chloride since it is not coeluted with any of the four arsenic species. A detailed study on the suitability of ion spray (IS) technique for arsenic speciation analysis in biological samples must be mentioned. A cation-exchange HPLC has been coupled to IS-MS(-MS) detection for analysis of organoarsenic species. Elemental and molecular, dual-mode analysis is presented using standard mixtures. Although detection limits are not low as those obtained by HPLCICP-MS, the results indicate IS-MS-MS as a complementary technique to ICP-MS for speciation analysis [200]. Papers published since 1980 on As and Se speciation in environmental matrices have recently been reviewed by Guerin et al. [201].
B.
Selenium
The thermochemical hydride-generation (THG) interface mentioned for arsenic speciation has also been used for HPLC-AAS determination of selenoniocholine (SeC) and trimethylselenonium cations (TMSe) [202]. The cation-exchange chromatography that resulted was unsuitable for the THG interface, and cations were separated on a cyanopropil stationary phase with a methanolic phase containing a silanol-masking agent. The calculated detection limits were 43.9 and 31.3 ng for SeC and TMSe, respectively. An improvement of chromatographic behavior and reduced detection limits were observed by adding trimethylsulfonium iodide to the mobile phase. Houck et al. [203] obtained, for inorganic selenium species, Se(IV) and Se(VI), detection limits of 10–20 ng/mL with ion-pairing reversed-phase (silica C18) microscale-liquid chromatography (eluent: methanol– water–TBA, flow rate 50 µL/min) and a direct-injection nebulizer (DIN) coupled with ICP-MS. More recently, Yiang et al. [204] coupled a similar ion-pair chromatographic separation with an ICP-MS detector using an ultrasonic nebulizer. By this approach the detec-
HPLC: Trace Metal Determination
/ 289
tion limits for TMSe, Se(IV), and Se(VI) were 0.17, 0.76, and 0.53 ng/mL, respectively. Se(IV) and Se(VI) species were also separated by Shum et al. [205] on an anion-exchange microcolumn (eluent: sodium carbonate–bicarbonate, flow rate 100 µL/min) with a DINICP-MS detection system. Isotope ratio measurements on chromatographically separated species of Se gave detection limits of 7–8 ng/ mL for both of the species (based on 78Se and peak-area measurements). A Nucleosil 100-SB anion-exchanger column was used with an eluent ammonium citrate for HPLC fraction collection and electrothermal atomic absorption spectrometry of Se(IV), Se(VI), and TMSe speciation [206]; the detection limits ranged from 11 to 32 ng/mL in water and urine matrices. Inorganic selenium species in aqueous samples have also been separated using an anion-exchange column with a two-step eluent-switching procedure [207]. The best separation was obtained with 25 mM K 2SO 4 eluent, switching to 200 mM after 200 s at a flow rate of 2.0 mL/min. After the separation by on-line acidification, microwave reduction, and hydride generation, the species were detected with an atomic fluorescence detector. The method provided detection limits of 0.2 and 0.3 ng/mL within 1.5% and 2.0% RSD for selenite and selenate, respectively. Speciation of eight selenium compounds has also been obtained with a strong cation-exchange column by interfacing the chromatographic system with an ICP-MS by a high-pressure hydraulic nebulizer [208]. Speciation of some organic selenium compounds has been considered in a recent review [209]. The speciation of organic selenium compounds (SeCy, SeMet, and TMSe ion) by HPLC-ICP-MS in natural samples (enriched yeast, human serum, and urine) has also been performed on a reversed-phase analytical column (Hamilton RP1) [210]. By optimizing the eluent polarity and ion-pairing agent (pentane sulfonate) with regard to TMSe cationic species, a satisfactory separation is achieved and inorganic species are eluted in the void volume. Detection was performed by ICP-MS using 82Se for quantification, and enabled to detect 0.20, 0.60, and 0.20 ng/mL for SeCy, SeMet, and TMSe, respectively, within a 0- to 500-ng/mL analytical dynamic range. Quijano et al. [211] reported the use of a mixed column for the speciation of selenocystine (SeCy), selenomethionine (SeMet), selenite, and selenate with ICP-MS. The method was improved by including two additional selenium species (TMSe and selenoethionine) using a Spherosorb 5ODS/A microcolumn and phosphate buffer
290
/ Sarzanini
eluents at pH 2.8 and 6.0, respectively. The method enabled detection limits in urine matrix of 0.5 ng/mL and precision better than 5% for the six selenium compounds [212]. A detailed study on inorganic selenium and selenoaminoacids speciation has been made by on-line coupling a high-performance liquid chromatographic–microwave–digestion system (HPLC-MW) with an AA, an ICP, or an ICP-MS spectrometer [213]. The system proposed allows complete separation of selenoaminoacids in urine samples, but the inorganic selenium peaks are overlapped; their speciation is obtained, however, by carrying out a second injection and detection with the microwave turned off. The ICP-MS detector provided the lowest detection limits, 0.16, 0.59, 0.66, and 0.19 ng/mL for total inorganic Se, selenomethionine, selenoethionine, and Se(IV), respectively. From previous experience [213,214], a detailed comparison of quadrupole ICP-MS and double-focusing sector field ICP-MS for the speciation of selenium in urine was made, also comparing different HPLC-ICP-MS interfaces (DIN, MW-HG) [215].
C.
Lead
Organolead compounds are present in the environment by biomethylation of inorganic lead and as a results of the use for a long time of tetraalkylead compounds as antiknock additives in gasoline. Robecke et al. [216] evaluated the HPLC behavior of tetramethyllead (TTML) and tetraethyllead (TTEL) on a LiChrospher 60 column with different eluent mixtures. Both acetonitrile-LiClO 4 and methanol-LiClO 4 eluents showed that a 10% aqueous solution 0.1 M LiClO 4 is sufficient to achieve separation of TTML and TTEL, but a ternary solvent (methanol–chloroform–LiClO 4) optimizes separation and reduces analyzing time (4 min). The method, applied to natural waters, gave, with normal pulse amperometric detection at a glassy carbon electrode, detection limits of 15.5 and 17 ng for TTML and TTEL, respectively. Shum et al. [188] achieved the separation of two alkyllead (TTML, TTEL) and three organomercury species (methyl-, ethyl-, and phenylmercury) using ion-pair chromatography and ICPMS detection. The separation was made by a PEEK microcolumn (5 cm ⫻ 1.6 mm I.D.) packed with C18 material and with an acetonitrile–water eluent containing ammonium pentanesulfate at 100 µL/ min flow rate. Detection limit was 0.2 pg and the method was validated with a freeze-dried urine reference material. The separation of TTML, TTEL, and triphenyl-lead (TPhL) by ion-pair HPLC and
HPLC: Trace Metal Determination
/ 291
ICP-MS detection was also obtained with a methanol–water eluent containing 4 mM sodium pentanesulfate [217]. With this approach the main problem encountered was the separation of TTEL from inorganic lead. In an attempt to optimize this separation, a gradient elution was applied [218]. During the optimization of chromatographic parameters, only inorganic and methyllead resulted, strongly affected by a change of the ion-pairing agent concentration, and 8.0 mM gave the best choice for resolution and for lowering the influence of salt concentration on the nebulizer tip and sampling orifice of the mass spectrometer. Inorganic, triethyl-, triphenyl-, and tetraethyl-lead were well resolved with a gradient methanol concentration, from 40% to 90%, over 10 min. The high concentration of organic solvent gave a loss of sensitivity only for TTEL, which was the last to elute, and the relative detection limits were 2.8, 3.5, 77.5, and 7.4 ng/ml for TEL, TPhL, TTEL, and inorganic lead, respectively. The method has been successfully applied to the quality control of the water supply of a U.S. Environmental Protection Agency (EPA) laboratory (trace metal contents) (Cincinnati, OH, USA). More recently, Brown et al. [219], on the basis of work of Al-Rashdan et al. [217,218], optimized the ion-pair HPLC separation of trimethyllead from Pb 2⫹, by a gradient program from 10:90 to 30:70 ratio of methanol to buffer (HAc/ NaAc) between 4 and 7 min. Coupling the chromatographic system and ICP-MS by a single-pass spray chamber with a concentric glass nebulizer, a 0.48-ng Pb/g detection limit was obtained for trimethyllead ions. On-column derivatization procedures have also been developed for simultaneous organic ionic lead and mercury species separation. Cammann et al. [220] developed an on-line enrichment on a RP-18 precolumn by adding methyl thioglicolate to the sample. Trimethyl-, triethyl- dimethyl-, and diethyl-lead, methyl- and ethyl-mercury were separated on an Hypersil ODS column with a mixture of methanol and citric acid buffer. Detection limits in the range 270–800 ng/L were obtained by spectrophotometric detection (UV, 235 nm).
D. Mercury A method has been developed based on the formation and separation of methyl, ethyl, phenyl, and inorganic mercury complexes with ammonium tetramethylenedithicarbamate [221]. An ODS RP-18 column was used and the elution was performed with an acetonitrile–
292
/ Sarzanini
water–APDC-buffered eluent. Detection was achieved by interfacing a glass flow cell between the chromatographic system and a cold-vapor atomic absorption spectrometer (CVAAS). The on-column procedure, in comparison with precomplexation, showed reduced detection sensitivity only for inorganic mercury, which is very high with respect to the other species. The on-line procedure was chosen and, after eluent optimization, detection limits between 0.5 and 0.015 ng/mL were achieved by coupling a sample (100 ml) preconcentration onto a C18 microcolumn. Diethyldithiocarbamate (DDC), hexamethyleneammonium (HMA)-hexamethylenedithiocarbamate (HMDC), and pyrrolidinedithiocarbamate (PDC) were tested for enrichment and separation of methyl, ethyl, methoxyethyl, ethoxyethyl, phenyl, and inorganic Hg complexes [222]. A RP C18 column was used and the best results were obtained with PDC complexes coupled with an acetonitrile–water-buffered eluent. Analytes were determined by ultraviolet, postcolumn oxidation, cold-vapor atomic absorption spectrometry (UV-PCO-CVAAS) [223]. The UV-PCO step was introduced to destroy the complexes, after elution, to increase the yield of the following reduction step for CVAAS determination. Methyl- and ethoxyethylmercury co-eluted but, since HCl pretreatment of samples decomposes the latter, it was possible to evaluate both species by analyzing treated and untreated samples. Detection limits of 5.0 µg/L, obtained with the HPLC-UV-PCO-AAS system, were lowered to 0.5 ng/L with a preconcentration (300-mL samples) of mercury chelates on a microcolumn (Hypersil-ODS RP C18). ICP-MS detection associated with ultrasonic nebulization was used for methyl-, ethyl-, and inorganic mercury after separation on a C18 reversed-phase column with a methanol–acetonitrile–2-mercaptoethanol eluent containing ammonium acetate [224]. Detection limit values (0.4–0.8 ppb) that were 10 times lower than those with LC-ICP-MS and a conventional nebulizer, and comparable to those for LC-ICP-MS with cold vapor generation, were obtained. Munaf et al. [225] proposed a preconcentration and liquid chromatographic separation of methyl, ethyl, and inorganic mercury based on microcolumns (Develosil-ODS and STR-ODS-H) with a cysteine–acetic acid eluent. The cysteine concentration was the limiting factor because low concentrations are insufficient to elute mercury species and too large concentrations hindered the mercury–cysteine complex decomposition before the cold-vapor detection. The sensitivity of the method was enhanced by coupling a preoxidation step (with
HPLC: Trace Metal Determination
/ 293
potassium peroxodisulfate and Cu catalyst) to the reaction for mercury-vapor generation; in this way a 0.1-ng Hg detection limit was obtained. An ion chromatographic separation of methyl, ethyl, and inorganic mercury as cysteine complexes was developed [226]. The eluent composition (acetic acid, sodium perchlorate, and cysteine) was optimized with respect to the separation procedure and to the reductive reaction (NaBH 4) which permits the detection of mercury with CVAAS. On-line preconcentration procedures were also investigated using both C18 and an ion-exchange microcolumn. The detection limits, for 100-mL samples, were 2, 10, and 4 ng for Hg, CH 3 Hg, and C 2 H 5Hg, respectively. Tetra-n-alkylammonium bromide ion-pairing agents and sodium halides in methanol–water mixture were investigated as mobile phases for the separation of inorganic mercury and organomercury species (methyl-, ethyl-, benzyl-, phenyl-) [227]. The effect of tetramethylammonium (TMA), tetraethylammonium (TEA), and tetrabutylammonium (TBA) ions on the capacity factors of the species investigated was examined. The retention of Hg 2⫹ was greatly dependent on the concentration of ion-pair reagent and its molecular size, and the capacity factor for Hg 2⫹ increased with an increase of both parameters. In contrast to this behavior, organic mercury species showed lower capacity-factor changes. The TBA ion-pairing agent was efficient for the separation of all the species and the addition of sodium chloride to the mobile phase gave better peak shapes and lower retention times. When UV and DCP detection was compared, the sensitivities resulted in opposite behavior and the detection limits ranged from 0.2 to 8.0 ng with UV and from 255 to 175 ng with DCP for benzylmercury and methyllmercury species. Relative high detection limits for DCP may be attributed to the high carbon content (organic solvent and ion-pair reagent) enhancing the background and low atomization efficiency due to larger ion-pair species difficult to penetrate into the DCP.
E. Tin At the begining of 1990, cyano phases (cyanopropyl-bonded silica gels) were used in normal-phase mode (eluent: hexane–acetonitrile– tetrahydrofuran) and tetraalkyl-tin compounds were separated according to their polarity under the following elution order: tetrabutyl-, tetraethyl-, tributyl chloride, tetraphenyl-, triethyl chloride, diphenyl dichloride, and diethyl dichloride [228]. The peak shapes and
294
/ Sarzanini
resolution were improved by an iodine chloride on-column pretreatment. UV detection (220 nm, time constant 50 ms) showed a complete separation in about 90 s with 6 mL/min mobile-phase flow rate. No details were given on detection limits and analytical dynamic range of detectable concentrations. Astruc et al. [229,230] did a theoretical and experimental study on on-line discontinuous detection in liquid chromatography by graphite furnace atomic absorption spectrometry (GFAAS) and its application to butyltin compounds at trace level. Separations of butyltin moieties have been obtained using a Nucleosil column with a 0.001% tropolone solution in toluene. The chromatographic procedure has been applied to tetrabutyl, tributyl-, dibutyl-, and monobutyltin speciation in water. The first two analytes are co-eluted and the last strongly retained onto the column. By coupling on-line detection by GF AAS, the detection limit for dibutyltin was 10 ng/L. A highly fluorogenic reaction between triphenyltin (TPhT) and 3-hydroxyflavone in a micellar medium (Triton X-100) has been coupled with an ion-exchange chromatographic method (column, cation exchanger Partisil10 SCX; eluent, methanol–water, 0.15 M ammonium acetate) enabled reaching a detection limit of 0.02 ng for 200-µL sample injection [231]. The method was applied to the determination of TPhT in bottom sea water after enrichment on a C18 cartridge and gave satisfactory results at the nanogram-per-liter level [232]. The cation-exchange chromatography of tributyltin and triphenyltin, usually performed with ammonium acetate eluents, has been optimized by the addition of benzyltrimethylammonium chloride (BTMA), which also allows the indirect detection of trialkyltins [233]. The method has been applied to trimethyltin and triethyltin speciation in addition to the above-mentioned compounds. The detection limits obtained (0.15–2.5 mg/L) were not so low, due to the type of detector used, but the chromatographic approach is attractive. Various types of couplings have been developed that utilize both inductively coupled plasma atomic emission spectrometry (ICPAES) and direct-current plasma atomic emission spectrometry (DCP-AES) [234] or ICP-MS [235] for detection. A reversed-phase liquid chromatography (column, C8, 3 µm, 30 ⫻ 3 mm I.D.; eluent, methanol–water 5 mM/sodium 1-pentanesulfonate) has been optimized for trimethyl-, triethyl-, tripropyl-, tributyl-, and triphenyltin separation and determination by ICP-MS equipped with an ultra-
HPLC: Trace Metal Determination
/ 295
sonic nebulizator [80]. Detection limits ranged between 2.8 and 16 pg Sn for various tin species, and the entire procedure required less than 6 min. An HPLC method (column, Kromasil-100, 5-µm, C18, 150 ⫻ 21 mm I.D.; eluent, 0.05% triethylamine in acetonitrile–acetic acid–water, 65: 10: 25) has been developed to achieve both molecular information and good sensitivity and selectivity for organo-tin species [236]. The optimized procedure enabled detection with both API and ICP mass spectrometry of dibutyl-, tributyl-, diphenyl-, and triphenyltin in sediments. Rivas et al. investigated the effect of different spray chambers in HPLC-ICP-MS [237] on the detection limits for organotin compounds. The instrumental interface, i.e., nebulizer and spray chambers, appeared to be the critical point. HPLC has been deeply discussed for the speciation of organotin compounds and the performance of the method of detection by Harrington et al. [238].
F. Chromium Chromium is a ubiquitous element, not only for its occurrence in nature but also for the numerous anthropogenic influences resulting from its widespread industrial applications. In environmental studies, its analytical determination is connected with the differences in biological and toxicological behavior of its two main oxidation states, Cr(III) and Cr(VI), Cr(III) is essential for the maintenance of the glucose tolerance factor in the human body. Cr(VI), due to its high oxidation potential and to its relatively small size, which enables its penetration through cell membranes, is toxic and carcinogenic. On the other hand, the occurrence of Cr(III) in the biotic environment as the aquo-hydroxocomplexes [Cr(H 2 O)6⫺n (OH) n](3⫺n)⫹, due to its size, makes it almost entirely excluded from penetrating cell membranes. Discussions and reference lists for chromatographic techniques with off-line separation and preconcentration, or on-line methods for chromium, can be found [239–241]. The possibility of using a complexing agent in the mobile phase for the determination of chromium speciation has been shown [242–245], including reversed-phase chromatography after the formation of neutral chelates [246,247], ion chromatography [248], and ion-pair chromatography [249–254]. Studies on chromium speciation with ion chromatographic separation of its EDTA complexes involves, in addition to common parameters, a detailed evaluation of the temperature effect. This is due to the slow formation rate of the complex between Cr(III) and EDTA.
296
/ Sarzanini
An anion chromatographic separation (polymer-based anion exchanger; eluent, EDTA–oxalic acid) has been optimized by working at 40°C [255]. Detection was performed by direct introduction of eluate into an ICP-MS system. This method enabled 80–88 ng/L detection limits for Cr(III) and Cr(VI), respectively, within a linear range from 0.5 to 5000 µg Cr per liter and simultaneous determination of chromium speciation and Mn, Fe, Ni, Cu, Mg, and Ca in water samples. Arar et al. [256] developed a method using a reversed-phase guard column NG1 coupled with an IonPac AS7 anion-exchange column. The eluent used was 250 mM ammonium sulfate and 100 mM ammonium hydroxide, as developed by Dionex. In these conditions, good sensitivity has been obtained by spectrophotometric detection [0.3 µg/L for Cr(VI) after postcolumn reaction with diphenylcarbazide]. The method has proven suitable to be applied in wastewater analysis. Postcolumn catalytic oxidation of luminol allowed the chemiluminescence detection of Cr(III) and Cr(VI) at detection limits of 0.1 and 0.3 µg/L, respectively [257], after separation by anionexchange on an IonPac AS4A column and an acidic eluent of 0.28 M KCl. The technique developed was not applied to a real environmental matrix, but to a Certified Reference Material, IAEA/W4 Simulated Fresh Water. More recently, Derbyshire et al. [258] obtained excellent resolution of the two chromium species using a single mixed-bed ion-exchange column (IonPac CS5) with continuous elution. Detection limits of 0.002 ng/mL for both Cr(III) and Cr(VI) were obtained after the optimization of postcolumn reactions [reduction of Cr(III) and oxidation of luminol]. The method gave results in very good agreement with certified values for water reference materials. Examples of different approaches for chromium speciation with anion exchange are two ion chromatographic procedures developed by Pobozy et al. [259]. In the first method (column, anion exchanger; eluent, potassium hydrogenphthalate), Cr(VI) was retained and Cr(III) was eluted in the void peak and postcolumn oxidized to Cr(VI). In this manner both Cr(III) and Cr(VI) were spectrophotometrically detected after postcolumn reaction with diphenylcarbazide (DPC). The second procedure was based on Cr(VI)–Cr(III) anion species separation after Cr(III) precomplexation with 1,2-diaminecyclohexane-N,N,N′,N′-tetraacetic acid (DCTA). Detection limits evaluated for the first and second methods were 2.5 and 4.5 ng/mL for Cr(III) and 1.8 and 1.5 ng/L for Cr(VI), respectively. Among ion-pair
HPLC: Trace Metal Determination
/ 297
applications, Posta et al. [252] optimized the Cr(VI)–Cr(III) separation on a RP C18 column (length, 5 cm) by using a TBA–acetate, ammonium acetate, phosphoric acid, and methanol-based eluent at 2.5 mL/min flow rate. The eluent composition enabled the chromium species separation, enhancing the sensitivity of detection obtained by coupling HPLC to a flame AAS by a high-pressure capillary with a hydraulic high-pressure nebulization. Detection limits of 0.03 and 0.02 mg/L for Cr(III) and Cr(VI), respectively, were reduced to 0.5 µ/L for Cr(VI) after a preconcentration step. Detection limits of 0.3–0.5 µg/L can be achieved using HPLC in combination with ICP-MS [260]. An anion-exchange column (Waters IC-Pak A, 50 ⫻ 4.6 mm, 10 µm) containing trimethylammonium functionalized groups on polymethacrylate and a cation-exchange column (Waters Guard-Pak, 5 ⫻ 3.9 mm, 5 µm) containing sulfonic groups on polybutadiene maleic anhydride silica were coupled for the simultaneous chromatographic separation of chromium species [260]. A step gradient with increasing nitric acid concentration and decreasing pH was used as the eluent. The analytical method has been optimized through the use of different Cr isotopes for data acquisition, comparing the interference of some species (e.g., chloride, chlorate, perchlorate, sulfate, sulfite, sulfide, tiosulfate, carbonate, cyanide, and organic species) at different m/z values. An m/z of 52 was chosen as the ideal isotope for Cr speciation in wastewaters. Since in all cases the main interest is in lowering detection limits and reducing analysis time, for chromium speciation great attention has been paid to interfacing ICP-MS detectors with HPLC separation modes. Despite the good detection limits, ICP-MS is prone to some interference when real samples have to be analyzed, the most serious problem being the formation of polyatomic ions, expecially below atomic mass number 80. For example, with organic liquids, owing to the presence of carbon, the abundant molecule 40Ar 12C ⫹ obscures 52Cr ⫹, the main isotope of chromium. Ion-pair chromatography with tetrabutylammonium acetate and 25% methanol on a Eurospher 100 C 18 5-µm column was applied for the separation of Cr(III) and Cr(VI) [253]. Speciation analysis was studied using hydraulic high-pressure nebulization in combination with ICP-MS. Addition of oxygen to the aerosol gas and effective desolvation were necessary prerequisites in order to apply ICP-MS as a selective and sensitive detection technique, thus reducing polyatomic interference from carbon, due to the presence of methanol.
298
/ Sarzanini
Exploiting the residual cationic-exchange capacity of a conventional anion-exchange column (IonPac AG5, 50 ⫻ 4 mm, 15 µm, Cr(VI) anions and Cr(III) cations were retained and eluted with discontinuous elution in two steps by 0.3 M nitric acid for Cr(VI) and 1.0 M nitric acid for Cr(III) [261]. Detection was achieved by coupling ICP-MS with a sample introduction technique using hydraulic highpressure nebulization, and the relative detection limits were 0.1 µg/ L for Cr(III) and 0.2 µg/L for Cr(VI). Total chromium, Cr(III) and Cr(VI), speciation was achieved by Powell et al. [262] by coupling HPLC anion chromatography (eluent, nitric acid) with direct-injection nebulization and ICP-MS. The detection limits obtained were 60 and 180 ng/L for Cr(III) and Cr(VI), respectively. A procedure developed for Cr(VI) determination [263] has been modified by Caruso et al. [264]. They used mixed-mode columns, namely, IonPac AS7, for Cr(III) and Cr(VI) separation with (NH 4) 2SO 4 eluent (pH 9.2). The Cr(III) species was stabilized with EDTA before sample analysis and detection was performed both by ICP-AES and ICP-MS equipped with a high-performance interface and a concentric nebulizer. To avoid polyatomic interference at m/z 52 from 36S16O ⫹ from the eluent, an m/z 53 value was chosen for detection. Relative detection limits were 0.40 ppb for Cr(III) and 1.0 ppb for Cr(VI) within a 4% RSD and a linear dinamic range from 3 to 600 ppb and from 5 to 1000 ppb for Cr(III) and Cr(VI), respectively, in aqueous media. An automated, on-line, two-column ion-exchange system has been proposed more recently (265). Cation and anion species of chromium are sequentially retained by eluting samples through a chelating column (Chelex 100) and an anion-exchange (AG MP-1) column. Recoveries were obtained by eluting Cr(III) with 2.0 M nitric acid and Cr(VI) with a NH 4OH/NH 4NO 3 mixture. Synthetic samples (75– 350 ng/mL Cr) gave recoveries of 91% and 100% for Cr(III) and Cr(VI), respectively, at a flame AAS detector, but anomalous results were obtained analyzing real samples, e.g., tap water. It must be pointed out that this and similar methods, more than chromatographic procedures, could be defined as flow-injection analysis (FIA).
ACKNOWLEDGMENTS I wish to thank Professor Edoardo Mentasti for his continuous support over the years and express my very special thanks to Maria
HPLC: Trace Metal Determination
/ 299
Concetta Bruzzoniti, Ph.D., in collecting some of the experimental data reported.
REFERENCES 1. W. R. Jones and P. Jandik, J. Chromatogr. Sci., 27: 449 (1989). 2. W. W. Buchberger and P. R. Haddad, J. Chromatogr., 789: 67 (1997). 3. P. R. Haddad, P. Doble, and M. Macka, J. Chromatogr. A, 856: 145 (1999). 4. C. Sarzanini, in A. Gianguzza, E. Pelizzetti, and S. Sammartano (Eds.), Marine Chemistry an Environmental Analytical Chemistry Approach, Vol. 25, Kluwer, Netherlands, 1997, p. 131. 5. S. R. Villasen˜or, Anal. Chem., 63: 1362 (1991). 6. S. R. Villasen˜or, J. Chromatogr., 602: 155 (1992). 7. S. Lehame, J. Chem. Ed., 63: 727 (1986). 8. S. Dilli, P. R. Haddad, and A. K. Htoon, J. Chromatogr., 500: 313 (1990). 9. V. Gonza´lez Rodri´guez, J. M. C. Romero, J. M. Ferna´ndez Solis, J. Pe´rez Iglesias, and H. M. Seco Largo, J. Chromatogr. A, 673: 291 (1994). 10. S. Dilli and P. Tong, Anal. Chim. Acta, 395: 101 (1999). 11. M. V. Main and J. Fritz, Talanta, 38: 253 (1991). 12. X. Y. Ming, Y. H. Wu, and G. Schwedt, Fresenius Z. Anal. Chem., 342: 556 (1992). 13. Y. Zhao and C. Fu, Anal. Chim. Acta, 230: 23 (1990). 14. N. Uehara, A. Katamine, and Y. Shijo, Analyst, 119: 1333 (1994). 15. J. Miura, Anal. Chem., 62: 1424 (1990). 16. H. Ohashi, N. Uehara, and Y. Shijo, J. Chromatogr., 539: 225 (1991). 17. E. Ryan and M. Meaney, Analyst, 117: 1435 (1992). 18. N. A. Beketova, E. M. Basova, V. M. Ivanov, and T. A. Bol’shova, Zh. Anal. Khim., 45: 2178 (1990). 19. E. M. Basova, T. A. Bol’shova, E. N. Shapovalova, and V. M. Ivanov, Zh. Anal. Khim., 45: 1947 (1990). 20. Q. P. Liu, H. S. Zhang, and J. K. Cheng, Talanta, 38: 669 (1991). 21. R. Saraswati, N. R. Desikan, and T. H. Rao, Mikrochim. Acata, 109: 253 (1992).
300 22. 23. 24. 25. 26. 27. 28.
29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.
/ Sarzanini L. Qiping, Z. Huashan, and C. Jieke, Chem. Res. Chinese Univ., 9: 18 (1993). Q. Liu, J. Liu, Y. Tong, and J. Cheng, Anal. Chim. Acta, 269: 223 (1992). L. Wuping and L. Qiping, Fresenius Z. Anal. Chem., 350: 671 (1994). Y. Akama, T. Iwadate, A. Tong, Y. Takahashi, and S. Tanaka, J. Chromatogr., A, 789: 479 (1997). M. Carreri, P. Manini, G. Predieri, and E. Sappa, J. Chromatogr. A, 789: 461 (1997). P. Wang and H. K. Lee, J. Chromatogr. A, 789: 437 (1997). P. Hajos, O. Horvath, G. Revesz, J. Peear, and C. Sarzanini, in A. Dyer, M. Hudson, and P. Williams (Eds.), Progress in Ion Exchange, The Royal Society of Chemistry, Cambridge, UK, 1991, p. 144. H. Small, T. S. Stevens, and W. C. Bauman, Anal. Chem., 47: 1801 (1975). R. E. Smith, Ion Chromatography Applications, CRC Press, Boca Raton, FL, 1988. P. Kolla, J. Ko¨hler, and G. Schomburg, Chromatographia, 23: 465 (1987). D. Yan and G. Schwedt, Fresenius J. Anal. Chem., 338: 149 (1990). L. M. Nair, R. Saari-Nordhaus, and J. M. Anderson, Jr. J. Chromatogr., 671: 43 (1994). R. M. Cassidy and S. E. Elchuk, Anal. Chem., 54: 1558 (1982). P. Janos and M. Brol, Fresenius J. Anal. Chem., 344: 545 (1992). Q. Sun, H. Wang, and S. Mou, J. Chromatogr., 708: 99 (1995). S. Motellier and H. Pitsch, J. Chromatogr., 739: 119 (1996). H. Lu, S. Mou, and J. M. Riviello, J. Chromatogr., A, 857: 343 (1999). W. Al-Shawi and R. Dahl, J. Chromatogr., 391: 35 (1999). R. D. Rocklin, M. A. Rey, J. R. Stillian, and D. L. Campbell, J. Chromatogr., Sci., 27: 474 (1989). M. Betti, G. Giovannoni, M. Onor, and P. Papoff, J. Chromatogr., 546: 259 (1991). I. N. Voloschik, M. L. Litvina, and B. A. Rudenko, J. Chromatogr., 671: 205 (1994). M. A. Rey, J. M. Riviello, and C. A. Pohl, J. Chromatogr., A, 789: 149 (1997).
HPLC: Trace Metal Determination 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67.
/ 301
D. Jensen, J. Weiss, M. A. Rey, and C. A. Pohl, J. Chromatogr., 640: 65 (1993). M. A. Rey and C. A. Pohl, 1993 Pittsburgh Conf. Atlanta, GA. C. A. Pohl, M. A. Rey, and R. J. Joyce, 1995 Pittsburgh Conf., New Orleans, LA. M. A. Rey and C. A. Pohl, J. Chromatogr. A, 739: 87 (1996). M. A. Rey, C. A. Pohl, J. J. Jagodzinski, E. Q. Kaiser, and J. M. Riviello, J. Chromatogr., 804: 201 (1998). N. Gros and B. Gorenc, J. Chromatogr. A, 697: 31 (1995). A. Do¨scher, M. Schiwikowski, and H. W. Ga¨ggeler, J. Chromatogr. A, 706: 249 (1995). K. A. Welch, W. B. Lyons, E. Graham, K. Neumann, J. M. Thomas, and D. Mikesell, J. Chromatogr., 739: 257 (1996). E. Kaiser, J. Riviello, M. Rey, J. Statler, and S. Heberling, J. Chromatogr., 739: 71 (1996). M. Pantsar-Kallio and P. K. G. Manninen, Anal. Chim. Acta, 314: 67 (1995). U. S. Hong, H. K. Kwon, H. Nam, G. S. Cha, K. H. Kwon, and K. J. Paeng, Anal. Chim. Acta, 315: 303 (1995). K. Otha, M. Sando, K. Tanaka, and P. R. Haddad, J. Chromatogr., A, 752: 167 (1996). K. Otha, H. Morikawa, K. Tanaka, Y. Uryu, B. Paull, P. R. Haddad, Anal. Chim. Acta, 359: 255 (1998). K. Otha, H. Morikawa, K. Tanaka, P. R. Haddad, J. Chromatogr. A, 804: 171 (1998). F. Steiner, C. Niederla¨nder, and H. Engelhardt, Chromatographia, 43: 117 (1996). K. Otha and K. Tanaka, Anal. Chim. Acta, 381: 265 (1999). P. J. Dumont and J. S. Fritz, J. Chromatogr., 706: 149 (1995). T. Iwachido, M. Shinomiya, and M. Zenki, Anal. Sci., 6: 277 (1990). T. Iwachido, M. K. Ikeda, and M. Zenki, Anal. Sci., 6: 593 (1990). T. Takeuchi and T. Miwa, Anal. Chim. Acta, 282: 565 (1993). E. Munaf, R. Zein, T. Takeuchi, and T. Miwa, Anal. Chim., Acta, 334: 39 (1996). D. Yan and G. Schwedt, J. Chromatogr., 516: 383 (1990). D. Yan and G. Schwedt, Fresenius Z. Anal. Chem., 338: 149 (1990). E. A. Gautier, R. T. Gettar, R. E. Servant, and D. A. Battistoni, J. Chromatogr. A, 706: 115 (1995).
302 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90.
/ Sarzanini W. Zhou, W. Liu, and D. An, J. Chromatogr., 589: 358 (1992). J. Z. Gao, H. J. Bian, and J.-G. Hou, J. Chromatogr., 657: 95 (1993). J. G. Hou, H. J. Bian, J. Z. Gao, and J. W. Kang, Chromatographia, 43: 473 (1996). A. A. leGras, Analyst, 118: 1035 (1993). T. V. Komarova, O. N. Obrezkov, and O. A. Spigun, Anal. Chim. Acta, 254: 61 (1991). W. Buchberger, P. R. Haddad, and P. W. Alexander, J. Chromatogr., 558: 181 (1991). C. Sarzanini, O. Abollino, E. Mentasti, and V. Porta, Chromatographia, 30: 293 (1990). P. Hajo´s, G. Revsez, C. Sarzanini, G. Sacchero, and E. Mentasti, J. Chromatogr., 640: 15 (1993). P. Hajo´s, G. Re´ve´sz, O. Horva´th, J. Peear, and C. Sarzanini, J. Chromatogr. Sci., 34: 291 (1996). C. Sarzanini, G. Sacchero, E. Mentasti, and P. Hajos, J. Chromatogr., 706: 141 (1995). K. Otha and K. Tanaka, Anal. Chim. Acta, 373: 189 (1998). Yamamoto, K. Hayakawa, A. Matsunaga, E. Mizukami, and M. Miyazaki, J. Chromatogr., 627: 17 (1992). P. Alumaa and J. Pentsuk, Chromatographia, 38: 566 (1994). F. W. E. Strelow and A. H. Victor, Talanta, 37: 1155 (1990). K. Kawabata, Y. Kishi, O. Kawaguchi, Y. Watanabe, and Y. Inoue, Anal. Chem., 63: 2137 (1991). J. C. Farin˜as, H. P. Cabrera, and M. T. Larrea, J. Anal. Am Spectrom., 10: 511 (1995). S. S. Heberling, J. M. Riviello, M. Shifen, and A. W. Ip, Res. Dev., Sept: 74 (1987). S. Ro¨llin, Z. Kopatjtic, B. Wernli, and B. Magyar, J. Chromatogr., A, 739: 139 (1996). A. W. Al-Shawi and R. Dahl, Anal. Chim. Acta, 333: 23 (1996). T. Williams and N. W. Barnett, Anal. Chim. Acta, 264: 297 (1992). M. C. Bruzzoniti, E. Mentasti, C. Sarzanini, M. Braglia, G. Cocito, and J. Kraus, Anal. Chim. Acta, 332: 49 (1996). M. C. Bruzzoniti, E. Mentasti, and C. Sarzanini, Anal. Chim. Acta, 353: 239 (1997). K. Tanaka, K. Otha, P. R. Haddad, and J. S. Fritz, J. Chromatogr. A, 804: 179 (1998).
HPLC: Trace Metal Determination 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114.
/ 303
H. Sire´n, Ann. Acad. Sci. Fennicae, A, II. Chemia, 233: 98 (1991). H. Sire´n and M.-L. Riekkola, J. Chromatogr., 590: 263 (1992). S. Comber, Analyst, 118: 505 (1993). C. Sarzanini, G. Sacchero, M. Aceto, O. Abollino, and E. Mentasti, J. Chromatogr., 640: 179 (1993). C. Sarzanini, O. Abollino, M. C. Bruzzoniti, and E. Mentasti, J. Chromatogr. A, 804: 241 (1998). C. Sarzanini, M. C. Bruzzoniti, O. Abollino, and E. Mentasti, J. Chromatogr. A, 847: 233 (1999). C. Ohtsuka, K. Matsuzawa, H. Wada, and G. Nakagawa, Anal. Chim. Acta, 252: 181 (1991). C. Ohtsuka, K. Matsuzawa, H. Wada, and G. Nakagawa, Anal. Chim. Acta, 256: 1 (1992). C. Ohtsuka, K. Matsuzawa, H. Wada, and G. Nakagawa, Anal. Chim. Acta, 294: 69 (1994). A. Alvarez-Zepeda, B. N. Barman, and D. E. Martire, Anal. Chem., 64: 1978 (1992). B. Chen, Q. Zhang, H. Minami, S. Inoue, and M. Uto, J. Chromatogr. Sci., 37: 306 (1999). P. Janos and M. Broul, Fresenius Z. Anal. Chem., 344: 545 (1992). R. M. Cassidy and L. Sun, J. Chromatogr., 654: 105 (1993). S. Zappoli, L. Morselli, and F. Osti, J. Chromatogr., 721: 269 (1996). M. L. Marina, P. Andre´s, and J. C. Die´z-Masa, Chromatographia, 35: 621 (1993). S. R. Villasen˜or, Anal. Chem, 63: 1362 (1991). J. F. Jen and C. S. Chen, Anal. Chim. Acta, 270: 55 (1992). G. Sacchero, O. Abollino, V. Porta, C. Sarzanini, and E. Mentasti, Chromatographia, 31: 539 (1991). Y. Zhao and C. Fu, Anal. Chim. Acta, 230: 23 (1990). N. Vachirapatama, P. Doble, Z. S. Yu, and P. R. Haddad, Chromatographia, 50: 601 (1999). R. M. Cassidy, Chem. Geol., 67: 185 (1988). M. Adachi, K. Oguma, and R. Kuroda, Chromatographia, 29: 579 (1990). R. Kuroda, T. Wada, G. Kishimoto, and K. Oguma, Chromatographia, 32: 65 (1991). K. Oguma, K. Sato, and R. Kuroda, Chromatographia, 37: 319 (1993).
304 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129.
130. 131. 132. 133. 134.
135. 136.
/ Sarzanini V. Kuban, I. Jancarova, and R. Sulova, Fresenius J. Anal. Chem., 342: 706 (1992). A. Dadone, A. Mazzucotelli, F. Baffi, and R. Frache, Ann. Chim. (Rome), 74: 553 (1984). J. M. Hwang, J. S. Shih, Y. C. Yeh, and S. C. Wu, Analyst, 106: 869 (1981). N. M. P. Moraes and H. M. Shihomatsu, J. Chromatogr. A, 679: 387 (1994). R. Kuroda, T. Wada, Y. Kokubo, and K. Oguma, Talanta, 40: 237 (1993). A. Hrdlicka, J. Havel, and M. Valiente, J. High Resol. Chromatogr., 15: 423 (1992). R. Kuroda, M. Adachi, K. Oguma, and Y. Sato, Chromatographia, 30: 263 (1990). C. A. Lucy, L. Gureli, and S. Elchuk, Anal. Chem., 65: 3320 (1993). X. J. Yang, Talanta, 41: 1807 (1994). S. Kobayashi, Y. Wakui, M. Kanesato, H. Matsunaga, and T. M. Suzuki, Anal. Chim. Acta, 262: 161 (1992). V. Kuban, I. Jancarova, V. Otruba, and V. Kanicky, Anal. Chim. Acta, 254: 21 (1991). D. S. Braverman, J. Anal. Atom. Spectrom., 7: 43 (1992). P. R. Haddad and R. C. Foley, J. Chromatogr., 500: 301 (1990). P. Janos and M. Broul, Fresenius Z. Anal. Chem., 344: 545 (1992). G. Sacchero, M. C. Bruzzoniti, C. Sarzanini, E. Mentasti, H. J. Metting, and P. M. J. Coenegracht, J. Chromatogr. A, 799: 35 (1998). C. Kantipuly, S. Katragadda, A. Chow, and H. D. Gesser, Talanta, 37: 491 (1990). P. Jones and P. N. Nesterenko, J. Chromatogr., 789: 413 (1997). J. P. Riley and D. Taylor, Anal. Chim. Acta, 40: 479 (1968). G. Bonn, S. Reiffenstuhl, and P. Jandik, J. Chromatogr., 499: 669 (1990). G. Bonn, S. Nathakarnikitkool, and P. Jandik, in P. Jandik and R. M. Cassidy (Eds.), Advances in Ion Chromatography, Vol. 1, Century International, Medfield, MA, 1991, p. 197. A. R. Timerbaev and G. K. Bonn, J. Chromatogr., 640: 195 (1993). A. Siriraks, H. M. Kingston, and J. M. Riviello, Anal. Chem., 62: 1185 (1990).
HPLC: Trace Metal Determination 137. 138. 139. 140. 141. 142. 143.
144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160.
/ 305
Dionex, Technical Note 25 (1990). R. Caprioli and S. Torcini, J. Chromatogr., 640: 365 (1993). N. Cardellicchio, S. Cavalli, and J. M. Riviello, J. Chromatogr., 640: 207 (1993). I. N. Voloschik, M. L. Litvina, and B. A. Rudenko, J. Chromatogr., 671: 51 (1994). O. Abollino, E. Mentasti, V. Porta, and C. Sarzanini, Anal. Chem., 62: 21 (1990). V. Porta, C. Sarzanini, O. Abollino, E. Mentasti, and E. Carlini, J. Anal. Atom. Spectrom, 7: 19 (1992). C. Sarzanini, O. Abollino, and E. Mentasti, in M. J. Slater (Ed.), Ion Exchange Advances, Elsevier, London, UK, 1992, p. 279. O. Abollino, C. Sarzanini, E. Mentasti, and A. Liberatori, Spectrochim. Acta, 49A: 1411 (1993). P. Jones, and G. Schwedt, J. Chromatogr., 482: 325 (1989). P. Jones, O. J. Challenger, S. J. Hill, and N. W. Barnett, Analyst, 117: 1447 (1992). O. J. Challenger, S. J. Hill, P. Jones, and N. W. Barnett, Anal. Proc., 29: 91 (1992). O. J. Challenger, S. J. Hill, and P. Jones, J. Chromatogr., 639: 197 (1993). B. Paull, M. Foulkes, and P. Jones, Analyst, 119: 937 (1994). B. Paull, M. Foulkes, and P. Jones, Anal. Proc., 31: 209 (1994). P. Jones, M. Foulkes, and B. Paull, J. Chromatogr., 673: 173 (1994). B. Paull and P. Jones, Chromatographia, 42: 528 (1996). B. Paull, P. A. Fagan, and P. R. Haddad, Anal. Commun., 33: 193 (1996). B. Paull and P. R. Haddad, Anal. Commun., 35: 13 (1998). B. Paull, P. Nesterenko, M. Nurdin, and P. R. Haddad, Anal. Commun., 35: 17 (1998). K. Robards, P. Starr, and E. Patsalides, Analyst, 116: 1247 (1991). K. G. Heumann, L. Rottmann, and J. Vogl, J. Anal. Atom. Spectrom., 9: 1351 (1994). L. Ebdon, S. Hill, and R. W. Ward, Analyst, 112: 112 (1987). A. Astruc, R. Pinel, and M. Astruc, Anal. Chim. Acta, 228: 129 (1990). F. Laborda, M. T. C. de Loos-Vollebregt, and L. de Galan, Spectrochim. Acta, 46B: 1089 (1991).
306 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183.
/ Sarzanini P. C. Uden, J. Chromatogr. A, 703: 393 (1995). S. J. Hill, M. J. Bloxham, and P. J. Worsfold, J. Anal. Atom. Spectrom., 8: 499 (1993). N. P. Vela and J. Caruso, J. Anal. Atom. Spectrom., 8: 787 (1993). K. Sutton, R. M. C. Sutton, and J. A. Caruso, J. Chromatogr. A, 789: 85 (1997). K. Sutton and J. A. Caruso, J. Chromatogr. A, 856: 243 (1999). W. M. A. Niessem and A. P. Tinke, J. Chromatogr. A, 703: 37 (1995). W. M. A. Niessem, J. Chromatogr. A, 856: 179 (1999). L. Benramdane, F. Bressolle, and J. J. Vallon, J. Chromatogr. Sci., 37: 330 (1999). J. S. Blais, G.-M. Momplaisir, and W. D. Marshall, Anal. Chem., 62: 1161 (1990). E. Russeva, I. Havezov, and A. Detcheva, Fresenius Z. Anal. Chem., 347: 320 (1993). V. M. Shkinev, B. Ya Spivakov, G. A. Vorob’eva, and Yu. A. Zolotov, Anal. Chim. Acta, 167: 145 (1985). H. Heng-bin, L. Yan-bing, M. Shi-len, and N. Zhe-ming, J. Anal. Atom. Spectrom., 8: 1085 (1993). E. Hakala and L. Pyy, J. Anal. Atom. Spectrom., 7: 191 (1992). B. S. Sheppard, J. A. Caruso, D. T. Heitkemper, and K. A. Wolnik, Analyst, 117: 971 (1992). D. T. Gjerde, D. R. Wiederin, F. G. Smith, and B. M. Mattson, J. Chromatogr., 640: 73 (1993). M. Lo´pez-Gonza´lvez, M. M. Go´mez, M. A. Palacios, and C. Ca´mara, Chromatographia, 43: 507 (1996). M. A. Lo´pez-Gonza´lvez, M. M. Go´mez, C. Ca´mara, and M. A. Palacios, Fresenius Z. J. Anal. Chem., 346: 643 (1993). M. A. Lo´pez-Gonza´lvez, M. M. Go´mez, C. Ca´mara, and M. A. Palacios, J. Anal. Atom. Spectrom., 9: 291 (1994). M. A. Palacios, M. Go´mez, C. Ca´mara, and M. A. Lo´pez, Anal. Chim. Acta, 340: 209 (1997). J. T. van Elteren and Z. Slejkovec, J. Chromatogr. A, 789: 339 (1997). K. J. Lamble and S. J. Hill, Anal. Chim. Acta, 334: 261 (1996). G. Rauret, R. Rubio, and A. Padro´, Fresenius Z. Anal. Chem., 340: 157 (1991). R. Rubio, A. Padro´, J. Alberti, and G. Rauret, Mikrochim. Acta, 109: 39 (1992).
HPLC: Trace Metal Determination 184.
185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205.
/ 307
J. W. McLaren, K. W. M. Siu, J. W. Lam, S. N. Willie, P. S. Maxwell, A. Palepu, M. Koether, and S. S. Berman, Fresenius Z. Anal. Chem., 337: 721 (1990). Al-Rashdan, N. P. Vela, J. A. Caruso, and D. T. Heitkemper, J. Anal. Atom. Spectrom., 7: 551 (1992). J. C. Van Loon and R. R. Barefott, Analyst, 117: 563 (1992). S. H. Hansen, E. H. Larsen, G. Pritzl, and C. Cornett, J. Anal. Atom. Spectrom., 7: 629 (1992). S. C. K. Shum, H.-M. Pang, and R. S. Houck, Anal. Chem., 64: 2444 (1992). S. C. K. Shum, R. Neddersen, and R. S. Houck, Analyst, 117: 577 (1992). S. Branch, W. T. Corns, L. Ebdon, S. Hill, and P. O’Neil, J. Anal. Atom. Spectrom., 6: 155 (1991). W. C. Story, J. A. Caruso, D. T. Heitkemper, and L. Perkins, J. Chromatogr. Sci., 30: 427 (1992). T. Nakahara, Spectrochim. Acta Rev., 14: 95 (1991). N. P. Vela, L. K. Olson, and J. A. Caruso, Anal. Chem., 65: 585A (1993). C. J. Hwang and S. J. Jiang, Anal. Chim. Acta, 289: 205 (1994). H. Larsen, G. Pritzl, and S. H. Hansen, J. Anal. Atom. Spectrom., 8: 557 (1993). B. S. Sheppard, W.-L. Shen, J. A. Caruso, and D. T. Heitkemper, J. Anal. Atom. Spectrom., 5: 431 (1990). H. Larsen, G. Pritzl, and S. H. Hansen, J. Anal. Atom. Spectrom., 8: 1075 (1993). C. F. Harrington, A. A. Ojo, V. W. M. Lai, K. J. Reimer, and W. R. Cullen, Appl. Organomet. Chem., 11: 931 (1997). H. Ding, J. Wang, J. G. Dorsey, and J. A. Caruso, J. Chromatogr., 694: 425 (1995). J. J. Corr and E. H. Larsen, J. Anal. Atom. Spectrom., 11: 1215 (1996). T. Guerin, A. Astruc, and M. Astruc, Talanta, 50: 1 (1999). J. S. Blais, A. Huyghues-Despointes, G. M. Momplaisir, and W. D. Marshall, J. Anal. Atom. Spectrom., 6: 225 (1991). R. S. Houck, S. C. K. Shum, and D. R. Wiederin, Anal. Chim. Acta, 250: 61 (1991). K. L. Yiang and S. J. Jiang, Anal. Chim. Acta, 307: 109 (1995). S. C. K. Shum and R. S. Houck, Anal. Chem., 65: 2972 (1993).
308 206. 207. 208. 209. 210. 211. 212. 213. 214. 215. 216. 217. 218. 219. 220. 221. 222. 223. 224. 225. 226.
/ Sarzanini F. Laborda, D. Chakraborti, J. M. Mir, and J. R. Castillo, J. Anal. Atom. Spectrom., 8: 643 (1993). L. Pitts, A. Fisher, P. Worsfold, and S. J. Hill, J. Anal. Atom. Spectrom., 10: 519 (1995). W. Goessler, D. Kuehnelt, C. Schlagenhaufen, K. Kalcher, M. Abegaz, and K. J. Irgolic, J. Chromatogr. A, 789: 233 (1997). K. Pyrzyn˜ska, Analyst, 121: 77R (1996). R. M. Olivas, O. F. X. Donard, N. Gilon, and M. Potin-Gautier, J. Anal. Atom. Spectrom., 11: 1171 (1996). M. A. Quijano, A. M. Gutie´rrez, M. C. Pe´rez-Conde, and C. Ca´mara, J. Anal. Atom. Spectrom., 11: 407 (1996). M. A. Quijano, A. M. Gutie´rrez, M. C. Pe´rez-Conde, and C. Ca´mara, Talanta, 50: 165 (1999). J. M. G. Lafuente, M. L. F. Sa´nchez, and A. Sanz-Mendel, J. Anal. Atom. Spectrom., 11: 1163 (1996). J. M. G. Lafuente, M. Dlaska, M. L. F. Sa´nchez, and A. SanzMendel, J. Anal. Atom. Spectrom., 13: 423 (1998). J. M. G. Lafuente, J. M. Marchante-Gayo´n, M. L. F. Sa´nchez, and A. Sanz-Mendel, Talanta, 50: 207 (1999). M. Robecke and K. Cammann, Fresenius Z. Anal. Chem., 341: 555 (1991). A. Al-Rashdan, D. T. Heitkemper, and J. A. Caruso, J. Chromatogr. Sci., 29: 98 (1991). A. Al-Rashdan, N. P. Vela, J. A. Caruso, and D. T. Heitkemper, J. Anal. Atom. Spectrom., 7: 551 (1992). A. A. Brown, L. Ebdon, and S. J. Hill, Anal. Chim. Acta, 286: 391 (1994). K. Cammann, M. Robecke, and J. Bettmer, Fresenius Z. Anal. Chem., 350: 30 (1994). C. Sarzanini, G. Sacchero, M. Aceto, O. Abollino, and E. Mentasti, J. Chromatogr., 626: 151 (1992). R. Falter and H. F. Scho¨ler, Fresenius Z. Anal. Chem., 353: 34 (1995). R. Falter and H. F. Scho¨ler, J. Chromatogr., 675: 253 (1994). C. W. Huang and S. J. Jiang, J. Anal. Atom. Spectrom., 8: 681 (1993). E. Munaf, H. Haraguchi, D. Ishii, T. Takeuchi, and M. Goto, Anal. Chim. Acta, 235: 399 (1990). C. Sarzanini, G. Sacchero, M. Aceto, O. Abollino, and E. Mentasti, Anal. Chim. Acta, 284: 661 (1994).
HPLC: Trace Metal Determination 227. 228. 229. 230. 231. 232. 233. 234. 235. 236. 237. 238. 239. 240. 241. 242. 243. 244. 245. 246. 247. 248. 249.
/ 309
Y. S. Ho and P. C. Uden, J. Chromatogr., 688: 107 (1994). A. Praet, C. Dewaele, L. Verdonck, and G. P. Van Der Kelen, J. Chromatogr., 507: 427 (1990). A. Astruc, M. Astruc, and R. Pinel, Anal. Chim. Acta, 228: 129 (1990). M. Astruc, A. Astruc, and R. Pinel, Mikrochim. Acta, 109: 83 (1992). R. Compan˜o´, M. Granados, C. Leal, and M. D. Prat, Anal. Chim. Acta, 302: 185 (1994). R. Compan˜o´, M. Granados, C. Leal, and M. D. Prat, Anal. Chim. Acta, 283: 272 (1993). E. Pobozy, B. Glod, J. Kaniewska, and M. Trojanowicz, J. Chromatogr., 718: 329 (1995). S. J. Hill, Anal. Proc. 29: 399 (1992). H. J. Yang, S. J. Jiang, Y. J. Yang, and C. J. Hwang, Anal. Chim. Acta, 312: 141 (1995). S. White, T. Catterick, B. Fairman, and K. Webb, J. Chromatogr. A, 794: 211 (1998). C. Rivas, L. Ebdon, and S. J. Hill, J. Anal. Atom. Spectrom., 11: 1147 (1996). C. F. Harrington, G. K. Eigendorf, and W. R. Cullen, Appl. Organometal. Chem., 10: 339 (1996). C. A. Johnson, Anal. Chim. Acta, 238: 273 (1990). M. Sperling, S. Xu, and B. Wellz, Anal. Chem., 64: 3101 (1992). M. Sperling, X. Yin, and B. Wellz, Analyst, 117: 639 (1992). J. F. Jen, and C. S. Chen, Anal. Chim. Acta, 270: 55 (1992). M. L. Marina, P. Andre´s, and J. C. Die´z-Masa, Chromatographia, 35: 621 (1993). J. F. Jen, G. L. Ou-Yiang, C. S. Chen, and S. M. Yang, Analyst, 118: 1281 (1993). G. L. Ou-Yiang, and J. F. Jen, Anal. Chim. Acta, 279: 329 (1993). X. Yao, J. Liu, J. Cheng, and Y. Zeng, Fresenius Z. Anal. Chem., 342: 702 (1992). C. M. Andrle and J. A. C. Broekaert, Fresenius Z. Anal. Chem., 346: 653 (1993). I. T. Urasa and S. H. Nam, J. Chromatogr., 27: 30 (1989). S. B. Roychowdhury and J. A. Koropchak, Anal. Chem., 62: 484 (1990).
310 250. 251. 252. 253. 254. 255. 256. 257. 258. 259. 260. 261. 262. 263. 264. 265.
/ Sarzanini M. Trojanowicz, E. Pobozy, and P. J. Worsfold, Anal. Lett., 25: 1373 (1992). A. V. Padarauskas and L. G. Kazlauskiene, Talanta, 40: 827 (1993). J. Posta, H. Berndt, S. K. Luo, and G. Schaldach, Anal. Chem., 65: 2590 (1993). N. Jakubowski, B. Jepkens, D. Stuewer, and H. Berndt, J. Anal. Atom. Spectrom., 9: 193 (1994). J. Lintschinger, K. Klacher, W. Go¨ssler, G. Ko¨lb, and M. Novic, Fresenius Z. Anal. Chem., 351: 604 (1995). Y. Inoue, T. Sakai, and H. Kumagai, J. Chromatogr., 706: 127 (1995). E. J. Arar and J. D. Pfaff, J. Chromatogr. A, 546: 335 (1991). H. G. Beere and P. Jones, Anal. Chim. Acta, 293: 237 (1994). M. Derbyshire, A. Lamberty, and P. H. E. Gardier, Anal. Chem., 71: 4203 (1999). E. Pobozy, E. Wojasinska, and M. Trojanowicz, J. Chromatogr., 736: 141 (1996). M. Pantsar-Kallio and P. K. G. Manninen, J. Chromatogr. A, 750: 89 (1996). C. Barnowski, N. Jakubowski, D. Stuewer, and J. A. C. Broekaert, J. Anal. Atom. Spectrom., 12: 1155 (1997). M. J. Powell, D. W. Boomer, and D. R. Wiederin, Anal. Chem., 67: 2474 (1995). Technical Note 26, Dionex Corp., Sunnyvale, CA, May 1990. F. A. Byrdy, L. K. Olson, N. P. Vela, and J. A. Caruso, J. Chromatogr., 712: 311 (1995). P. A. Sule and J. D. Ingle, Jr., Anal. Chim. Acta, 326: 85 (1996).
8 Temperature-Responsive Chromatography
Hideko Kanazawa and Yoshikazu Matsushima Pharmacy, Tokyo, Japan
Kyoritsu College of
Teruo Okano Institute of Biomedical Engineering, Tokyo Women’s Medical University, Tokyo, Japan
I. INTRODUCTION II. TEMPERATURE-RESPONSIVE POLYMERS III.
IV.
V. VI.
TEMPERATURE-RESPONSIVE STATIONARY PHASES A. Polymer Synthesis and Temperature-Responsive Solubility Changes
312 313 314 314
TEMPERATURE-RESPONSIVE CHROMATOGRAPHY: TUNABLE SEPARATION A. Temperature-Responsive Chromatography of Steroids B. Effects of Salt Addition on Retention C. Effect of Hydrophobic Groups of NIsopropylacrylamide Co-polymers Modified on Silica
314 318
TEMPERATURE EFFECTS ON RETENTION
322
TEMPERATURE GRADIENT: THE METHOD IS REPLACING SOLVENT GRADIENT
324
314
319
311
312
/ Kanazawa, Matsushima, and Okano
VII. APPLICATION TO THE SEPARATION OF PEPTIDES AND PROTEINS A. Effect of Temperature on Peptide Separation B. Effect of Salt Concentration in the Mobile Phase on Peptide and Protein Separations VIII. CONCLUSIONS ACKNOWLEDGMENTS REFERENCES
326 326 329 333 333 334
I. INTRODUCTION The needs of biotechnology for high-resolution purification and analytical technology have spurred new methods in chromatographic systems, in both columns and equipment. In the case of separation of peptides, proteins, other biological molecules and maintaining viable cells, it is frequently necessary to avoid the use of organic solvents in the mobile phase, as these cause sample denaturation. Total avoidance of organic solvents is also advantageous for environmental reasons. Reversed-phase chromatography (RPC) using chemically modified stationary phases is generally faster and easier than other modes of high-performance liquid chromatography (HPLC), and consequently, has achieved very wide popularity. The retention and selectivity in RPC are controlled primarily by changing the polarity of the mobile phase. Use of organic solvents is necessary to prevent excessively long retention times with conventional reversephase columns. Stimuli-responsive polymers, which change their structure and physical properties in response to external signals, comprise new materials with interesting applications in biomaterial science and technology. Such polymers responsive to changes in pH, temperature, and light have been widely utilized for drug delivery systems [1–3], cell culture substrates [4–6], and bioconjugates [7–9]. However, there have been few reports of the use of these stimulus-responsive polymers in chromatographic separations. In this chapter, our recent studies of temperature-responsive chromatography are reviewed. The method for the molecular design of temperature-responsive polymers exhibiting remarkable changes in response to temperature is discussed. Achievement of new chromatography systems in which surface properties and resulting func-
Temperature-Responsive Chromatography / 313 tion of the HPLC stationary phase are controlled by external temperature changes is reviewed. This method should be effective in biological and biomedical separations of peptides and proteins by using only aqueous mobile phases.
II. TEMPERATURE-RESPONSIVE POLYMERS There are many synthetic polymers with molecular conformations sensitive to environmental factors such as pH [10], electric field [11], chemical species [12], and temperature [13]. Temperature seems to be one of the most attractive tools in such stimuli. The various properties of aqueous polymer solutions have been studied by many investigators [14–17]. Among those, the phase transition of aqueous poly (N-isopropylacrylamide) (PIPAAm) solution is quite sensitive, reversible, and reproducible to thermal stimulation, in contrast to that of other polymers [14,16]. PIPAAm is well-known to exhibit a thermally reversible phase transition at 32°C [14]. This transition temperature is called a lower critical solution temperature (LCST). Below the LCST, the polymer chain of IPAAm is hydrated to an expanded form which is soluble in water. PIPAAm undergoes a reversible phase transition to an insoluble form and dehydrates to a compact form above the LCST. We have previously observed changes from hydrophilic to hydrophobic properties for PIPAAm-terminally grafted surfaces due to rapid changes in polymer conformation [18]. PIPAAm-grafted surfaces show hydrophilic properties at low temperatures; as temperature increases, the contact angle of water also increases, indicating increasingly hydrophobic surface properties. The temperature-responsive surface-property changes of terminal-grafted surfaces were rapid and significant. It is suggested that these features are due to conformational freedom for the PIPAAm graft chains [19], which influence polymer dehydration and hydrogen bonding with water molecules. Using this feature, we developed a new chromatography system in which hydrophilic–hydrophobic property changes for the stationaryphase surfaces respond to external temperature changes. The solute interactions with the stationary phase can be altered by changing the temperature of the mobile phase. Gewehr et al. examined gel permeation chromatography using porous glass beads modified with the IPAAm polymers [20]. Grafted PIPAAm was used to control pore size by changing column tempera-
314
/ Kanazawa, Matsushima, and Okano
ture. They performed separation of dextran with water as a mobile phase. The report was concerned mainly with temperature-responsive coil-globule conformation changes in PIPAAm molecules to control pore size. The idea had been presented by Okahata et al. [21], who succeeded in thermoselective permeation from a polymergrafted capsule membrane. However, only slight changes in solute retention had been reported in chromatography. Hosoya et al. [22,23] also reported a surface-selective modification procedure for the incorporation of PIPAAm into porous polymer beads. Galaev et al. reported temperature-induced displacement of proteins from dye-affinity columns. Poly (N-vinylcaprolactam) was used for the polymer shielding of Blue Sepharose [24].
III. TEMPERATURE-RESPONSIVE STATIONARY PHASES A.
Polymer Synthesis and Temperature-Responsive Solubility Changes
Carboxyl semitelechelic polymers of IPAAm were synthesized by radical telomerization using 3-mercaptopropionic acid (MPA) as a chain-transfer agent. The chain-transfer polymerization is easy both to control the polymer molecular weight and to introduce a functional group at one polymer chain end [25]. The carboxyl groups were activated with N-hydroxysuccinimide using dicyclohexylcarbodiimide (DCC) in dry ethyl acetate. Figure 1 summarizes the synthesis of the IPAAm polymer and the coupling to aminopropyl silica surfaces. A packing material based on the modification of silica with PIPAAm, a thermally reversible polymer, is quite sensitive to temperature changes. Below 32°C, it shows hydrophilic behavior as PIPAAm chains hydrate and swell, while above that temperature the chains shrink and the material acts a hydrophobic stationary phase.
IV.
TEMPERATURE-RESPONSIVE CHROMATOGRAPHY [26]: TUNABLE SEPARATION
A.
Temperature-Responsive Chromatography of Steroids
Using a column packed with PIPAAm-modified silica, separation of steroids was carried out by changing temperature. With increas-
Temperature-Responsive Chromatography / 315
Fig. 1 The synthesis of IPAAm polymer and the coupling to aminopropyl silica surfaces.
ing temperature, increased interaction between solutes and the PIPAAm-grafted surfaces of the stationary phases was observed. Temperature-dependent resolution of steroids was achieved using only water as a mobile phase. The separation selectivity and retention are controlled by a small change in column temperature, without any change in the eluent. The effect of column temperature on the elution behavior was examined. Figure 2A shows the elution behavior of five steroids and benzene in aqueous mobile phases over a variety of column temperatures. The retention times of the five steroids depended largely on temperature. At lower temperatures than LCST (32°C), four steroids were not resolved, as shown in Fig. 3(a). Excellent resolution of the steroids was achieved at 50°C, as shown in Fig. 3(c). Additionally, retention times of the steroids increased with increasing tempera-
316
/ Kanazawa, Matsushima, and Okano
Fig. 2 Retention factors of five steroids on the temperature-responsive column (A) with pure water as mobile phase and (B) with 1.0 M NaCl aqueous solution as mobile phase. Flow rate, 1.0 mL/min; detection, UV, 254 nm. Solutes: 䊉, benzene; 䊊, hydrocortisone; 䉱, prednisolone; 䉭, dexamethasone; ■, hydrocortisone acetate; 䊐, testosterone.
ture. The elution profile was strongly affected by temperature. Changes in the retention times for hydrophobic steroids such as testosterone were larger than those of their hydrophilic counter parts and benzene. On the temperature-responsive column, the increase in retention times with increasing temperature clearly demonFig. 3 Chromatograms of a mixture of five steroids and benzene with pure water (upper) and with 1.0 M NaCl aqueous solution (lower) as mobile phase at (a) 5, (b) 25, and (c) 50°C. Chromatographic conditions are the same as those in Fig. 2. Peaks: 1, benzene; 2, hydrocortisone (mg/mL); 3, prednisolone (0.166 mg/mL); 4, dexamethasone (0.061 mg/mL); 5, hydrocortisone acetate (0.007 mg/mL); 6, testosterone (0.027 mg/mL). Injection volume, 20 µL.
Temperature-Responsive Chromatography / 317
318
/ Kanazawa, Matsushima, and Okano
strated a reversed tendency compared with ordinary HPLC columns. In general, the adsorption of molecules on the surface and the viscosity of mobile phases were decreased, while the solubility was increased at an elevated temperature. Hence, the retention times should decrease with increasing temperature for a normal chromatographic process. However, in the PIPAAm-modified columns, the opposite behavior of retarded retention times was observed with increasing temperature, in spite of decreased back-pressure of the column. It is noteworthy that the retention of steroids shows large changes above the LCST of PIPAAm. This implies that the transition from hydrophilic surface properties to hydrophobic ones at LCST causes this anomalous retention behavior of the steroids. Partition coefficients of the substances in octanol/water system expressed as log P values are known to represent the relative order of hydrophobicity. The order of separation on a temperature-responsive polymer-modified column was the same as that of the hydrophobicities of the steroids, i.e., by log P values. The results of our studies indicate that the driving force for retention in this system is the hydrophobic interactions between the solute molecules and polymer chains on the surface. At temperatures higher than LCST, PIPAAm-grafted surfaces exhibited hydrophobic properties and the sensitivity to hydrophobicity of the solutes was remarkably increased.
B.
Effects of Salt Addition on Retention
The LCST of IPAAm polymers was determined by measuring the optical transmittance of polymer aqueous solutions. The temperature dependence for the optical transmittance of IPAAm polymer solutions of various concentrations in NaCl was investigated. In pure water the transition temperature (LCST) was observed at 32°C. The LCST decreased with increasing concentrations of NaCl, while the sharp soluble–insoluble changes were maintained. The LCST shifted remarkably to 20°C in 1 M NaCl solution (Table 1). As PIPAAm is a nonionic compound, an electrostatic interaction would not affect LCST. The lowering of LCST by the addition of salt should, therefore, be due to acceleration of dehydration, i.e., saltingout [26]. Effect of salt addition in the mobile phase was then examined. Difference in the PIPAAm phase-transition point was hardly observed between pure water and 0.1 M NaCl. However, if a mobile
Temperature-Responsive Chromatography / 319 Table 1 Effect of Salt Concentration on LCST of PIPAAm and poly (IPAAm-co-5% BMA) NaCl concentration (mol/L) 0 0.1 0.5 1.0
LCST (°C) PIPAAm
poly(IPAAm-co-BMA)
32.0 30.5 25.0 20.0
24.0 23.0 19.0 15.0
phase with a concentration higher than 1 M NaCl was used, the transition temperature decreased to 20°C. Large elution differences were noted between the lower and the higher temperature of LCST induced by the salt. The effects of column temperature and NaCl in the mobile phase on the retention profile of steroids are shown in Figs. 2 and 3. The retention times of steroids at 25°C were much longer with 1 M NaCl than with pure water as the mobile phases. In Fig. 3a some extent of resolution was obtained even at 5°C. The retention time of testosterone (peak 6) changed from 34 min to 48 min by changing the mobile phase from pure water to 1 M NaCl at 50°C. This should result from a lowering of the LCST by changing the mobile phase. The fact suggests that we can control LCST and the hydrophobicity of the surface of the stationary phases, and thus elution time, by salt concentration of mobile phases.
C. Effect of Hydrophobic Groups of NIsopropylacrylamide Co-polymers Modified on Silica Generally, the LCST should decrease with increasing polymer hydrophobicity [27]. The phase transition of PIPAAm is due to stability of the hydrophobic groups in the polymer chain in aqueous media. The LCST values for PIPAAm with carboxyl end groups was regulated by the amount of hydrophobic co-monomer, butyl methacrylate (BMA) [24]. To increase the hydrophobicity of the polymer, we have synthesized carboxyl semitelechelic co-polymers of IPAAm with BMA as hydrophobic co-monomer using radical telomerization with MPA. In contrast to the IPAAm homo-polymer, the co-polymer with BMA increased the aggregation force of PIPAAm chains. Critical temperatures for hydration–dehydration changes observed for
320
/ Kanazawa, Matsushima, and Okano
Fig. 4 Chromatograms of a mixture of five steroids and benzene with pure water on poly(IPAAm-co-BMA) modified column. Flow rate, 1.0 mL/min; detection UV, 254 nm. Peak number corresponds to the steroids indicated in Fig. 3. (Data from Ref. 28.)
IPAAm–BMA co-polymers were lower than the LCST that homogeneous PIPAAm demonstrated in aqueous media, as shown in Table 1. The influence of column temperature on the retention behavior of steroids on the co-polymer-modified stationary phase was greater than that on the homo polymer-modified one [28]. As shown in Fig. 4, retention times of the steroids increased remarkably with increasing BMA composition (percent molar ratio was indicated). The temperature-responsive elution behavior of the steroids was strongly affected by the hydrophobicity of the polymer chains grafted on the silica surface. Figure 5 shows the variation of ln k values with the temperature change of the column of IPAAm–BMA co-polymer-modified silica with water as the mobile phase. The stationary phase showed a
Temperature-Responsive Chromatography / 321
Fig. 5 Comparison of ln k in water on poly(IPAAm-co-BMA) modified column. Chromatographic conditions are the same as those in Fig. 4. Solutes: 1, hydrocortisone; 2, prednisolone; 3, dexamethasone; 4, hydrocortisone acetate; 5, testosterone.
greater affinity for hydrophobic steroids at the higher temperature (30°C) compared with those at 5°C. These observations would be due to the temperature-responsive conformational change of the IPAAm co-polymer. Because the LCST of the aqueous solution of PIPAAm– 5% BMA co-polymer was 24°C as shown in Table 1, increasing retention suggested that the phase transition of co-polymer on the surface occurred between 5 and 30°C. As described above, the column packed with PIPAAm and copolymer-modified silica showed drastic changes in retention of solutes by small changes in column temperatures. There should be interactions between the solutes and polymer chains of the surface on the stationary phase in our system. The temperature-responsive interaction between PIPAAm-modified silica and the steroids should be due to changes in the surface properties of the PIPAAm-grafted stationary phase by the reversible transition of hydrophilic–hydrophobic PIPAAm-grafted surface properties.
322
/ Kanazawa, Matsushima, and Okano
Fig. 6 Reproducibility of chromatograms of a mixture of steroids on poly (IPAAm-co-BMA) modified column at 50°C. Chromatographic conditions are the same as those in Fig. 4. Peaks: 1, benzene; 2, cortisone; 3, hydrocortisone; 4, prednisolone; 5, dexamethasone; 6, betamethasone; 7, cortisone acetate 8, hydrocortisone acetate; 9, testosterone.
Reproducible retention times of separation of the steroids were obtained on repeated runs, as shown in Fig. 6.
V.
TEMPERATURE EFFECTS ON RETENTION
In contrast to gas chromatography (GC), in which temperature changes during chromatographic runs are common practices, temperature control has never been a major subject in liquid chromatography (LC). Temperature is known to play a significant role in biomolecule and chiral separations, but its influence on the separation of small molecules in reversed-phase HPLC is much less important.
Temperature-Responsive Chromatography / 323 Almost all of the physical parameters that play important roles in LC separations are functions of temperature. The influence of temperature on retention is a function of the free-energy changes in the interaction between the analyte and the stationary phase, as described by Eq. (1): ln k ⫽ ⫺
∆H ∆S ⫹ ⫹ ln Φ RT R
(1)
where k is the retention factor for a given analyte, ∆H and ∆S are the enthalpy change and the entropy change for the retention reaction, respectively, R is the gas constant, T is the absolute temperature, and Φ is the phase ratio of the stationary and mobile phases. Enthalpy changes dominate the retention reactions in most chromatographic systems. In common normal- and reversed-phase separations, retention reactions are usually exothermic, with relatively small enthalpy changes that cause modest decreases in retention time when the temperature increases. However, much higher enthalpy changes occur in the separation of large molecules; therefore, temperature has a much larger effect on the separation. Although free-energy values can be used to predict the temperature influence on retention, the reverse is of more practical value. Calculating free-energy changes from temperature-retention functions may be useful in understanding the retention mechanisms. A van’t Hoff plot of ln k versus reciprocal temperature (1/T ) normally will show a linear regression, with the slope representing the enthalpy change involved for the retention reaction. The entropy values can be calculated from the y intercept of the van’t Hoff plot. Figure 7 shows the van’t Hoff plots for the steroids on the PIPAAm-modified column. We found curvature in the plots at LCST on the temperature-responsive polymer-modified column. Generally, these plots should be linear for conventional chromatographic processes on commercially available reversed-phase columns under conditions where retention mechanisms do not change. This curvature indicates a change in retention mechanism at that temperature and is consistent with the known phase transition of the grafted PIPAAm stationary phase. Additionally, the slope of the van’t Hoff plots on the PIPAAm column is negative, opposite to that seen for conventional chromatography. This provides additional evidence that the interaction between steroids and temperature-responsive surfaces becomes stronger at elevated temperature.
324
/ Kanazawa, Matsushima, and Okano
Fig. 7 van’t Hoff plots of the five steroids on the temperature-responsive column. Chromatographic conditions are the same as those in Fig. 2. Solutes: 䊊, hydrocortisone; 䉱, prednisolone; 䉭, dexamethasone; ■, hydrocortisone acetate; 䊐, testosterone. (Data from Ref. 26.)
Several workers have reported nonlinearity of van’t Hoff plots. Papadopoulou-Mourkidou noted a distinct break at 10°C in the van’t Hoff plots for fenvalerate optical isomers [29]. The author believed that this break indicated a change in retention mechanism at that temperature. Jinno et al. have noted curvature in the van’t Hoff plots for PAHs at temperatures between 40 and 60°C for a polymeric phase [30]. Sentell and Henderson have also noted curvature in the van’t Hoff plots for PAHs on high-density, reversed-phase materials [31]. They explain these observations as a phase transition of the bonded octadecyl chains to a more ordered and extended state.
VI.
TEMPERATURE GRADIENT: THE METHOD IS REPLACING SOLVENT GRADIENT
In isocratic elution of samples containing solutes with a wide range of polarity, it is sometimes difficult to achieve the desired resolution in a reasonable time. It may be necessary to use gradient elution
Temperature-Responsive Chromatography / 325 where volumes of an organic solvent, composition of mobile phase, or other properties of the solvent (e.g., pH or ionic strength) are changed during the separation. Little attention has been paid to programmed temperature changes during chromatographic runs in HPLC. LC can use solvent strength as a function of time (solvent gradient), which has a much greater potential for changing the retention and selectivity than does temperature change. Also, the large heat capacity of conventional LC columns considerably limits the warm-up and cool-down rates of columns. Water jackets provide excellent heat exchange and temperature stability but require a circulating, thermostatted water bath and restrict free access to the column. A cryogenic bath can be used for subambient temperature control. Successful applications of temperature programming usually involve small-inner-diameter columns because of their favorable heat exchange characteristics [32–34]. Its major effect is a reduction in retention and an improvement in peak efficiency as temperature increases. For complex samples this may be of great utility, as the column peak capacity is increased significantly. The increase in efficiency is the result of a decrease in solvent viscosity as temperature increases. This enhances the mass transfer rate between the mobile and stationary phases. Besides the constraints on instrumental design and the requirement of solvent preheating, one major disadvantage of operating at high temperatures is that faster dissolution of the silica matrix occurs, and column lifetime will inevitably become shorter. Some sample components are thermally labile, and temperature reduction can prevent sample degradation during separation. On HPLC columns packed with temperature-responsive-polymer-modified silica, gradient elution-like effect can be achieved with a single mobile phase by controlling external temperature. Figure 8 shows chromatograms of a mixture of four steroids and benzene with a step gradient by changing the column temperature [35,36]. Peaks 2 and 3 are not properly resolved at 5°C. At 30°C, peaks 2 and 3 were well resolved, but the retention time of peak 5 was increased too much. To improve the situation, we use a step gradient technique of using column temperature, shown in the lower chromatogram of Fig. 8. Baseline separation through all of the peaks of the steroids was achieved within half of the retention at 30°C. These features are due to rapid changes in polymer conformation, which are attributed to the mobility of grafted polymers.
326
/ Kanazawa, Matsushima, and Okano
Fig. 8 Chromatograms of a mixture of four steroids and benzene with a step gradient by changing column temperature. Chromatographic conditions are the same as those in Fig. 4. Peaks: 1, benzene; 2, cortisone; 3, prednisolone; 4, hydrocortisone acetate; 5, testosterone. (Based on Ref. 35.)
VII. APPLICATION TO THE SEPARATION OF PEPTIDES AND PROTEINS A.
Effect of Temperature on Peptide Separation
In recent years, HPLC techniques have became increasingly important for the separation of proteins without denaturation, which is often observed with the use of the reversed-phase mode. Many reversed-phase methods for proteins have employed C18 columns with mobile phases containing acetonitrile in low-pH buffers. Little attention has been paid to recovering biological activity. Acetonitrile almost always leads to total denaturation of proteins. Similarly, acidic environments, including the standard trifluoroacetic acid-based
Temperature-Responsive Chromatography / 327 buffers, are sometimes destructive to the activities of many enzymes. Thus these conditions should be avoided in the separation of most proteins. Because temperature-responsive chromatography (TRC) is performed in an aqueous environment that includes structure-stabilizing salts, TRC allows for the retention of biological activity. Isocratic elution by an aqueous mobile phase alone is the basis for separation of peptides and proteins [37]. The separation of a mixture of three peptides, insulin chains A and B and β-endorphin fragment 1-27, was achieved by changing the column temperature with NaCl aqueous solution as a sole eluent, as shown in Fig. 9. The three peptides were not separated at 5°C, which was lower than the LCST. As the
Fig. 9 Chromatograms of a mixture of insulin chains A, B and β-endorphin fragment. Peaks: 1, insulin chain A; 2, β-endorphin fragment 1-27; 3, insulin chain B. Column, poly(IPAAm-co-BMA)-modified silica; eluent, 0.075 M NaCl aqueous solution; flow rate, 1.0 mL/min; detection, UV, 225 nm. (Based on Ref. 37.)
328
/ Kanazawa, Matsushima, and Okano
Table 2 Amino Acid Sequence of Peptides Peptide Insulin chain A
Insulin chain B
β-Endorphin fragment 1-27 α-Endorphin β-Endorphin
α-Neoendorphin Leucine enkephalin-Lys
Amino acid sequence Gly-Ile-Val-Glu-Gln-CysSO 3-CysSO 3-AlaSer-Val-CysSO 3-Ser-Leu-Tyr-Gln-LeuGlu-Asn-Tyr-CysSO 3-Asn Phe-Val-Asn-Gln-His-Leu-CysSO 3-GlySer-His-Leu-Val-Glu-Ala-Leu-Tyr-LeuVal-CysSO 3-Gly-Glu-Arg-Gly-Phe-PheTyr-Thr-Pro-Lys-Ala Tyr-Gly-Gly-Phe-Met-Thr-Ser-Glu-LysSer-Gln-Thr-Pro-Leu-Val-Thr-Leu-PheLys-Asn-Ala-Ile-Ile-Lys-Asn-Ala-Tyr Tyr-Gly-Gly-Phe-Met-Thr-Ser-Glu-LysSer-Gln-Thr-Pro-Leu-Val-Thr Tyr-Gly-Gly-Phe-Met-Thr-Ser-Glu-LysSer-Gln-Thr-Pro-Leu-Val-Thr-Leu-PheLys-Asn-Ala-Ile-Ile-Lys-Asn-Ala-TyrLys-Lys-Gly-Glu Tyr-Gly-Gly-Phe-Leu-Arg-Lys-Tyr-Pro-Lys Tyr-Gly-Gly-Phe-Leu-Lys
column temperature was raised to 40°C, these peptides, which consist of 21–30 amino acid residues, were well resolved. The elution order of the three peptides should reflect their hydrophobic properties. The effect of temperature on capacity factors of peptides and proteins was measured on the poly(IPAAm-co-BMA)-modified silica column (Tables 2 and 3). This method was also applicable for the separation of both low-molecular-weight proteins, such as ribonuclease and chymotrypsinogen, and high-molecular-weight proteins, such as ovalbumin, catalase, and bovine serum albumin (Fig. 10). The protein retention times increase as the hydrophobicity of the polymer increases. To avoid strong column–solute interactions, the PIPAAm homo-polymer-modified column, which is less hydrophobic than PIPAAm-BMA co-polymer-modified ones, was employed in the chromatography of some large proteins. When separation at lower temperatures is necessary because of the stability of the proteins, the IPAAm co-polymer-modified column should be used. During the chromatography of peptide and protein samples, it is possible for some of the more hydrophobic components to became
Temperature-Responsive Chromatography / 329 Table 3 Effect of Column Temperature on Capacity Factor
Insulin chain A Insulin chain B β-Endorphin fragment 1-27 α-Endorphin β-Endorphin α-Neoendorphin Leucine enkephalin-Lys
5°C
30°C
50°C
⫺0.46 ⫺0.25 ⫺0.33 ⫺0.22 ⫺0.25 0.00 ⫺0.02
0.01 1.68 0.92 — — — —
0.02 4.46 2.46 0.19 0.69 0.60 0.27
Mobile phase, 0.9% NaCl aqueous solution; detection, 215 nm or 280 nm depending on sample; flow rate, 0.5 mL/min. (Based on Ref. 37.)
strongly adsorbed to the stationary phase, resulting in loss of column performance. The PIPAAm-modified columns are cleaned by washing with several volumes of cold water at 5°C, because the surface property of the stationary phase becomes hydrophillic at this temperature. This column cleaning is drastically different from those of reversed-phase columns, where a high concentration of the organic solvent is used to clean up.
B. Effects of Salt Concentration in the Mobile Phase on Peptide and Protein Separations The effects of salt concentration on the separation of peptides were investigated. Table 4 shows the effects of salt concentration and temperature on the retention times of insulin chain A, insulin chain B, and β-endorphin fragment 1-27. Retention times of insulin chain B and β-endorphin fragment 1-27 were remarkably increased as the column temperature was raised. Salt concentration also affected peptide retention time, and longer peptide retention times were observed with higher NaCl concentrations at any given temperature. Effects of salt concentration and temperature on retention time of chymotrypsinogen A are shown in Fig. 11. The retention times of the proteins were increased with increasing salt concentration. The increase of retention with an increase of the salt concentration may be explained in terms of the effect of salting-out. Ammonium sulfate shows a larger effect of salting-out and the retention times were longer than with the same concentration of NaCl. The retention time of chymotrypsinogen A was 8.47 min with 0.5 M (NH 4) 2 SO 4 as a mo-
330
/ Kanazawa, Matsushima, and Okano
Fig. 10 Effects of temperature on retention times of peptides and proteins. Eluent, 0.5 M NaCl; column, A, PIPAAm homopolymer-modified silica, B, poly(IPAAm-co-BMA)-modified silica; flow rate, 1.0 mL/min; detection, UV, 280 nm.
Temperature-Responsive Chromatography / 331 Table 4 Effects of Salt Concentration and Temperature on Retention Times of Peptides
Temperature (°C) 5 35 50
Insulin chain A
Insulin chain B
β-Endorphin Fragment 1-27
0.5 M
1.0 M
0.5 M
1.0 M
0.5 M
1.0 M
1.36 1.53 1.59
1.38 1.58 1.79
2.14 4.22 7.04
2.70 8.28 16.28
1.55 2.68 3.98
2.08 6.18 11.51
Mobile phase, NaCl aqueous solution; detection, UV, 215 nm; flow rate, 0.5 mL/min. (Based on Ref. 37.)
Fig. 11 Effects of salt concentration and temperature on the retention time of chymotrypsinogen A. Eluent, 䊊, 1.0 M NaCl; 䊐, 0.5 M NaCl; column, poly(IPAAm-co-BMA)-modified silica; flow rate, 1.0 mL/min; detection, UV, 280 nm.
332
/ Kanazawa, Matsushima, and Okano
bile phase with a flow rate of 1.0 mL/min at 40°C, while it was 3.48 min with a mobile phase of the same concentration of NaCl under the same conditions. The temperature-responsive interaction between PIPAAm-modified silica and the eluates are due to changes in the surface properties of the stationary phase by the reversible transition from a hydrophilic to a hydrophobic nature. The results shown above indicate that the present system is a kind of hydrophobic interaction chromatography (HIC). In conventional HIC systems, separations are based on the surface hydrophobic functionality of proteins and peptides. Usually they were performed using a starting mobile phase of very high ionic strength to promote hydrophobic binding and eluted using decreasing salt concentrations. On the contrary, in temperature-responsive chromatography, the separation of peptides and protein is achieved using a single mobile phase by controlling external temperature. The surface property of the stationary phase becomes hydrophobic above LCST. Hydrophobic interactions are strengthened by increasing temperatures and proteins are retained in the column longer. Thus a mobile phase can be selected which is more compatible with proteins and peptides. In TRC, salt gradient can be used in addition to temperature gradient for elution of peptides, proteins, and other biopolymers. Consequently, stationary phases based on IPAAm co-polymermodified silica may prove useful in high-performance peptide and protein separations. In such separations, retention on the supports also may involve hydrophobic interactions between protein and polymer chains. Recently, temperature-responsive chromatographic separations of amino acid phenylthiohydantoins with water as a sole mobile phase were reported [38]. The ability of the proposed temperature-responsive polymermodified stationary phase to separate the solutes without the use of organic solvent is advantageous from the point of view of keeping biological activity, for environmental reasons, and for the economical cost of the mobile phases [39]. HPLC with this new packing material is applicable for the separation of peptides and drugs with aqueous solutions as mobile phases, for the purification of peptide drugs from impurities during
Temperature-Responsive Chromatography / 333 synthesis using recombinant methods, and for the analysis of bioactive peptides in body fluids.
VIII. CONCLUSIONS As described above, PIPAAm-modified silica exhibits temperaturecontrolled hydrophilic–hydrophobic surface property changes in aqueous systems. Using a column packed with PIPAAm-modified silica, separations of steroids and peptides were carried out by changing temperature. With increasing temperature, increased interactions between solutes and PIPAAm-grafted surfaces of the stationary phases were observed. Temperature-dependent resolution of steroids was achieved using only water as a mobile phase. This system would be highly useful to control the function and properties of an HPLC stationary phase by changing only temperature in an aqueous solvent. Such temperature-responsive chromatography is a new concept in LC with a high potential and versatility. In this chapter, we described a polymer that is responsive only to temperature. New chromatography systems using temperature and/or pH-responsive polymers are in progress in our laboratory. Recently, several types of polymers have been studied to fabricate chromatography systems. Peter et al. [40] reported modification of the pore of rigid polymeric monoliths with PIPAAm chains. The grafted chains can block and control the flow through the micrometer-sized pores of the monolith. Ihara et al. [41] reported the combshaped polymer (ODA)n , as a lipid membrane analog, immobilized silica gel for separation with molecular recognition. Yakushiji et al. [42] investigated the effect of the architecture of grafted PIPAAm on a silica gel surface. Sawada and Jinno reported capillary electrophoresis using non-cross-linked poly (acrylamide-co-N-isopropylacrylamide) for the separation of structurally similar solutes [43]. These unique features of functional polymers have produced a new concept of chromatography.
ACKNOWLEDGMENTS This work was supported in part by Grant-in-Aid for Scientific Research 11680845 from the Ministry of Education, Science, Sports, and Culture of Japan.
334
/ Kanazawa, Matsushima, and Okano
REFERENCES 1. T. Okano, Y. H. Bae, and S. W. Kim, in J. Kost (ed.), Pulsed and Self-Regulated Drug Delivery, CRC Press, Boca Raton, FL, 1990, pp. 17. 2. R. Yoshida, K. Sakai, T. Okano, and Y. Sakurai, J. Biomater. Sci. Polym. Edn., 3: 243 (1992). 3. R. Yoshida, K. Sakai, T. Okano, Y. Sakurai, Y. H. Bae, and S. W. Kim, J. Biomater. Sci. Polym. Edn., 3: 155 (1991). 4. N. Yamada, T. Okano, H. Sakai, F. Karikusa, Y. Sawasaki, and Y. Sakurai, Makromol. Chem. Rapid Commun., 11: 571 (1990). 5. T. Okano, N. Yamada, H. Sakai, and Y. Sakurai, J. Biomed. Mater. Res., 27: 1243 (1993). 6. T. Okano, N. Yamada, M. Okuhara, H. Sakai, and Y. Sakurai, Biomaterials, 16: 297 (1995). 7. G. C. Chen and A. S. Hoffman, Nature, 373: 49 (1995). 8. M. Matsukata, Y. Takei, T. Aoki, K. Sanui, N. Ogata, Y. Sakurai, and T. Okano, J. Biochem., 116: 682 (1994). 9. M. Matsukata, T. Aoki, K. Sanui, N. Ogata, A. Kikuchi, Y. Sakurai, and T. Okano Bioconjugate Chem., 7: 96 (1996). 10. J. Kopecek, J. Vacik, and D. Lim, J. Polym. Sci. A-1, 9: 2801 (1971). 11. T. Tanaka, I. Nishio, S. Sun, and S. Ueno-Nishio, Science, 218: 467 (1981). 12. K. Ishihara, N. Muramoto, and I. Shinohara, J. Appl. Polym. Sci., 29: 211 (1984). 13. Y. H. Bae, T. Okano, and S. W. Kim, J. Polym. Sci. Polym. Phys., 28: 923 (1990). 14. M. Heskins, J. E. Guillet, and E. James, J. Macromol. Sci. Chem. A., 2: 1441 (1968). 15. S. Fujishige, K. Kubota, and I. Ando, J. Phys. Chem., 93: 3311, (1989). 16. S. Fujishige, Polym. J., 19: 297 (1987). 17. H. G. Schild and D. A. Tirrell, J. Phys. Chem., 94: 4352 (1990). 18. Y. G. Takei, T. Aoki, K. Sanui, N. Ogata, Y. Sakurai, and T. Okano, Macromolecules, 27: 6163 (1994). 19. R. Yoshida, K. Uchida, Y. Kaneko, K. Sakai, A. Kikuchi, Y. Sakurai, and T. Okano, Nature, 374: 240 (1995). 20. M. Gewehr, K. Nakamura, N. Ise, and H. Kitano, Makromol. Chem., 193: 249 (1992).
Temperature-Responsive Chromatography / 335 21. Y. Okahata, H. Noguchi, and T. Seki, Macromolecules, 19: 493 (1986). 22. K. Hosoya, K. Kimata, T. Araki, N. Tanaka, and J. M. J. Fre`chet, Anal. Chem., 67: 1907 (1995). 23. K. Hosoya, E. Sawada, K. Kimata, T. Araki, N. Tanaka, and J. M. J. Fre`chet, Macromolecules, 27: 3973 (1994). 24. I. Y. Galaev, C. Warrol, and B. Mattiasson, J. Chromatogr. A, 684: 37 (1994). 25. Y. G. Takei, T. Aoki, K. Sanui, N. Ogata, T. Okano, and Y. Sakurai, Bioconjugate Chem., 4: 341 (1993). 26. H. Kanazawa, K. Yamamoto, Y. Matsushima, N. Takai, A. Kikuchi, Y. Sakurai, and T. Okano, Anal. Chem., 68: 100 (1996). 27. L. D. Taylor and L. D. Cerankowski, J. Polym. Sci. Polym. Chem., 13: 2551 (1975). 28. H. Kanazawa, Y. Kashiwase, K. Yamamoto, Y. Matsushima, A. Kikuchi, Y. Sakurai, and T. Okano, Anal. Chem., 69: 823 (1997). 29. E. Papadopoulou-Mourkidou, Anal. Chem., 61: 1149 (1989). 30. K. Jinno, T. Nagoshi, N. Tanaka, M. Okamoto, J. C. Fetzer, and W. R. Biggs, J. Chromatogr., 436: 1 (1988). 31. K. B. Sentell and A. N. Henderson, Anal. Chim. Acta, 246: 139 (1991). 32. M. I. Chen, C. Horva´th, J. Chromatogr. A, 788: 51 (1997). 33. F. Houdiere, P. W. J. Fowler, and N. M. Djordjevic, Anal. Chem., 69: 2589 (1997). 34. K. Jinno, D. P. Carney, J. B. Phillips, Anal. Chem., 57: 575 (1985). 35. H. Kanazawa and Y. Matsushima, Yakugaku Zasshi, 117: 817 (1997). 36. H. Kanazawa, Y. Matsushima, T. Okano, Trends Anal. Chem., 17(7): 435 (1998). 37. H. Kanazawa, Y. Kashiwase, K. Yamamoto, Y. Matsushima, N. Takai, A. Kikuchi, Y. Sakurai, and T. Okano, J. Pharm. Biomed. Anal., 15: 1545 (1997). 38. H. Kanazawa, T. Sunamoto, Y. Matsushima, A. Kikuchi, Y. Sakurai, and T. Okano, Anal. Chem., in press. 39. K. Yamamoto, H. Kanazawa, Y. Matsushima, K. Oikawa, A. Kikuchi, and T. Okano, Environ. Sci., 7(1): 47 (2000). 40. E. C. Peters, F. Svec, and M. J. Fre`chet, Adv. Mater., 9: 630 (1997).
336
/ Kanazawa, Matsushima, and Okano
41.
H. Ihara, S. Nagaoka, H. Tanaka, S. Sakaki, and C. Hirayama, J. Liq. Chromatogr. & Rel. Technol., 19: 2967 (1996). T. Yakushiji, K. Sakai, A. Kikuchi, T. Aoyagi, Y. Sakurai, and T. Okano, Anal. Chem., 71: 1125 (1999). H. Sawada and K. Jinno, Electrophoresis, 18: 2040 (1997).
42. 43.
9 Carrier Gas in Capillary Gas–Liquid Chromatography Victor G. Berezkin A. V. Topchiev Institute of Petrochemical Synthesis, Russian Academy of Sciences, Moscow, Russia
I. INTRODUCTION II. ABSOLUTE RETENTION AND ITS DEPENDENCE ON THE NATURE AND PRESSURE OF THE CARRIER GAS A. Definition of Absolute Retention B. Dependence of Absolute Retention on the Nature and Pressure of the Carrier Gas III. RELATIVE RETENTION AND ITS DEPENDENCE ON THE NATURE AND PRESSURE OF THE CARRIER GAS A. Definition of Relative Retention B. Theoretical Aspects of the Dependence of Relative Retention on the Nature and Pressure of the Carrier Gas C. Experimental Confirmation of the Relative Retention Dependence on the Average Column Pressure and the Nature of the Carrier Gas
338
339 339 342
345 345
349
353 337
338
/ Berezkin
IV.
V. VI.
VII.
D. Correlation for Calculation of Relative Retention Values Measured on the Same Column Using Various Carrier Gases STEAM CAPILLARY GAS–LIQUID CHROMATOGRAPHY AND THE INFLUENCE OF WATER VAPOR ON RELATIVE RETENTION INFLUENCE OF THE NATURE AND PRESSURE OF THE CARRIER GAS ON SEPARATION EFFECT OF CARRIER-GAS SOLUBILITY IN STATIONARY LIQUID PHASE ON RELATIVE RETENTION CONCLUSION REFERENCES
362
364 366
369 371 372
I. INTRODUCTION In gas–liquid chromatography (GLC) the main participants in the chromatographic processes are: the compounds to be separated (sorbate); the stationary liquid phase (SLP), and the carrier gas. Most of the studies in GLC were devoted to the role of the SLP, the interaction between the SLP and the sorbate, and the determination of relative retention values, which are interpreted to be the chromatographic constants of the sorbates. In general, the role of the carrier gas was considered from a narrow point of view. Many researchers believe that the carrier gas serves to transport the solutes along the column and to broaden the chromatographic zones. Also, many chromatographers believe that relative retention values are chromatographic constants of the sorbate, which do not depend on the nature and pressure of the carrier gas. Let us quote some conventional opinions on the role of the carrier gas. The mobile phase in gas chromatography is generally considered to be inert in the sense that is doesn’t react chemically with the sample or stationary phase nor does it influence the sorption–desorption or partitioning processes that occur within the column. Thus the choice of carrier gas doesn’t influence the selectivity. It can influence resolution through its effect on column efficiency which arises from differences in solute diffusion rates [1]. The role of the mobile phase in GC is thus limited to that of carrier to transport the solute through the column [2].
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 339
Relative retention values are a function only of the partition coefficient of the given compound and that of the standard. The distortion probability of the relative retention by the experimental error is very limited. Relative retention values are of great importance, especially in gas chromatography, where phase equilibrium depends only on the properties of stationary phase [3]. Retention index for the given sorbate practically depends only on the content of stationary phase and operating temperature [4]. The above quotations, as a rule, refer to conventional analytical gas chromatography, where average column pressure does not exceed 5 atm. Recently, however, it was shown that even at these conditions the nature and pressure of the carrier gas, such as He, H 2 , N 2 , and CO 2, have a significant influence on the retention. The changes in conventional opinion on the role of the carrier gas in GLC and the development of fundamental theories of the carrier gas influence on the relative retention (as well as on the separation selectivity) is essential for a better utilization of GLC separations. The purpose of this review is to summarize the studies of effects of the carrier gas on the relative retention at conditions of routine capillary GLC, and to estimate the analytical importance of these effects. The carrier gas effects mentioned above are valid for columns of all types. However, this review is restricted to capillary columns only. The reasons for this restriction are (1) the analytical capillary chromatography is currently in common use, and (2) the measurement accuracy of the retention values is much greater for capillary columns. In this review we pay the most attention to carrier gas effects on the relative retention, since these values are in common use in chromatographic practice.
II. ABSOLUTE RETENTION AND ITS DEPENDENCE ON THE NATURE AND PRESSURE OF THE CARRIER GAS A. Definition of Absolute Retention Retention characteristics of sorbate are usually subdivided into two groups: absolute and relative retention (see, e.g., [3]). Retention times and values that are functions of the retention time are consid-
340
/ Berezkin
ered absolute retention. Parameters that are a function of the ratio of the absolute retention (including the ratio of the retention times) are referred to as relative retention. Total retention time t R is the main absolute retention characteristic [5]. Adjusted retention time t′R is the total retention time minus the holdup time t M: t′R ⫽ t R ⫺ t M
(1)
The values t′R and t R depend on the flow rate of the mobile phase. Total retention volume V R and adjusted retention volume V′R are more invariant values (not dependent on the flow rate of the mobile phase): VR ⫽ FC tR
(2)
V R′ ⫽ F C (t R ⫺ t M )
(3)
where F C is the flow rate of the mobile phase at column temperature T c measured at the column outlet, and Fa is the flow rate of the mobile phase at ambient temperature Ta measured at the column outlet (T a and T C are in kelvin). Since the carrier gas is compressible, its average flow rate at column temperature differs from that at the column outlet. Thus, the average flow rate of the mobile phase depends on the experimental conditions (see, e.g., [1–4,6–12]). The following equations can be used to determine the average column pressure Pav and the average flow rate F av of the mobile phase. P av ⫽
Po j
(4)
F av (T c , P av ) ⫽ jF (T a , P o )
冢 冣 Tc Ta
(5)
where Po is the outlet pressure of the carrier gas and j is a compressibility correction factor, taking into consideration the volumetric expansion of the carrier gas during its movement along the column. The compressibility correction factor j 23 averaged over the column length (l) is given by [7–12] j (l) ⫽ j 23 ⫽
3 (P i /Po )2 ⫺ 1 2 (P i /Po )3 ⫺ 1
where Pi is the inlet pressure of the carrier gas.
(6)
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 341
The compressibility correction factor j 34 averaged over the residence time (t) (i.e., taking into consideration the duration of sorbate residence in the column) is expressed by [13–17] j (t) ⫽ j 34 ⫽
4 (P i /Po )3 ⫺ 1 3 (P i /Po )4 ⫺ 1
(7)
It seems reasonable to use the j 34 compressibility correction factor in the measurements of the retention volume since the definition of that retention term is based on the definition of the retention time: F av (T c , P av ) ⫽ j 34 F (T a , P o )
冢 冣 Tc Ta
(8)
With Eq. (8), the net retention volume V N can be given as follows [6]: V N ⫽ (t R ⫺ t M ) F av (T c , P av )
(9)
The following ratio can be used for the specific retention volume (V g0) at column temperature. V g0 ⫽
VN ws
(10)
where w s is the weight of the active stationary phase in the column. The specific retention volume does not depend on the amount of the active sorbent (for example, stationary liquid phase) in the column. The specific retention volume is a more invariant relative retention value than the net retention values. It depends on the column length and the film thickness of the SLP. The specific retention volume deals with the sorbate’s thermodynamic distribution constant K Di between the SLP and the mobile phase. Since [1–4,7,8,12] K Di ⫽
VN Vl
(11)
where V l is the volume of the SLP in the column and since V l ⫽ w s / ρ (where ρ is the SLP density at column temperature), then K Di ⫽ Vg0 ρ
(12)
342
/ Berezkin
With Eq. (12), and analogous ratios, gas–liquid chromatography can be used as a simple and effective techniques for the determination of thermodynamic values, characterizing the sorbate equilibrium in the SLP/mobile phase system (see, e.g., [1,7,8,12,18,19]). The definition of absolute retention and its physical meaning have been discussed in the scientific literature in recent years [6,20– 27]. The techniques for precise determination of the holdup retention time are of special importance for the precise calculation of both absolute and relative retention; see, for example, refs. 29 and 30. Eliminating any solid influence is an important issue in the determination of the sorbate thermodynamic characteristics, as considered in detail in Ref. 31. Considerable gain in measurement accuracy and in the reliability of results can be obtained [32,33] by using packed columns. Therefore, studies in this direction are of great importance and should be pursued. Understanding the correct dependence of the absolute retention on the experimental conditions is essential for both the physicochemical interpretations of the absolute retention and its application for physicochemical measurements. It is necessary to understand the various sorption processes, especially the adsorption phenomena. Published theoretical and experimental studies [31] and conventional techniques for the estimation of adsorption phenomena allow the qualitative determination of equilibrium values characterizing the net equilibrium of sorbate in SLP/gas system (without disturbing or distorting the sorbate’s adsorption on the SLP/gas and solid/SLP interfaces). Also, it is necessary to take into account the dependence of the absolute retention on the pressure and the nature of the carrier gas. The carrier gas can have a significant influence on the absolute retention, as reported, for example, in Refs. 18 and 19. A detailed study of the phenomenon is presented below.
B.
Dependence of Absolute Retention on the Nature and Pressure of the Carrier Gas
Fundamentals of the dependence of the absolute retention on the pressure and nature of the carrier gas were developed in the 1960s. Everett and Stoddart [34] were the first to consider that dependence. Desty and co-workers [35], Cruickshank and co-workers [36–39],
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 343
Conder and Young [18], Laub and Pesckok [19], and Giddings and co-workers [40,41] made important contributions to the understanding of that dependence. The topic has been discussed and developed in many studies [42–55]. Conder and Young [18] have examined the dependence of the absolute retention on the pressure and the nature of the carrier gas. When generalizing their results for conventional analytical chromatography (the inlet pressure of the carrier gas does not exceed 15 atm [39]), the following equations for the specific retention volume were obtained: ln
Vgi (Pav ) ⫽ β′i ⋅ Pav Vgi (0)
log
(13)
Vgi (Pav ) ⫽ ⫺ 0.4343 β′i ⋅ Pav Vi (0)
β′i ⫽
冤 冢
(14)
冣 冥
∂ ln γi∞ 2B12 ⫺ Vi∞ ⫹λ 1⫺ RT ∂X2
β′gi ⫽
2B12 ⫺ Vi∞ RT
冤 冢
β′li ⫽ λ 1 ⫺
(15)
(16)
冣 冥
∂ ln γi∞ ∂X2
(17)
X2⫽0
Pav ⫽ Po J43 J43 ⫽
⫽ β′gi ⫹ β′li
X2⫽0
1 3 [(Pi /Po ) 4 ⫺ 1] ⫽ J34 4 [(Pi /Po ) 3 ⫺ 1]
(18) (19)
where V gi (Pav ) ⫽ V Ni (Pav )/w s , V gi (Pav ) is the specific retention volume for compound i, w s is the weight of sorbent (SLP); V gi(Pav) and V gi (0) are specific retention volume in the system SLP/gas at average column pressure of Pav and at zero pressure, respectively; Pi and Po represent the inlet and outlet pressures of carrier gas, respectively; β′i is a general coefficient characterizing the carrier gas influence on the retention; B 12 is the second (mixed) virial coefficient characterizing the interaction between the carrier gas and the sorbate in the gas phase; V i∞ is the partial molar volume of the ith compound dissolved in the SLP at infinite dilution (molar volume of liquid sorbate
344
/ Berezkin
at absolute column temperature T commonly used in the calculations instead of V i∞ [18,19]; X 2 is the molar fraction of carrier gas dissolved in the SLP (traditionally, X 2 ⬇ λP), λ is the molar solubility of the carrier gas in the SLP at the pressure of 1 atm); γ i∞ is the solute activity coefficient at infinite dilution; and R is the molar gas constant. Let us consider the physical meaning of the coefficient β′. From Eq. (13), β′ can be written as β′ ⫽
冤
冥
Vgi (Pav ) d ln Vgi (Pav ) d ln ⫽ dPav Vgi (0) dPav
(20)
It follows from Eq. (20), that the coefficient β′ is the rate of change of lnV gi (Pav) with the change in the average column pressure Pav. From Eqs. (15)–(17), β′ is the sum of two terms, one reflecting the interaction between the sorbate and carrier gas in the mobile phase and the other term the interaction in SLP. Since not much discussion of the virial coefficients of the sorbate and the carrier gases appear in most monographs on gas chromatography (or are discussed very briefly), it seems reasonable to present some brief comments on the β′ characteristic of gases (see, e.g., [56– 59]). The gas law equation for 1 mole of ideal gas (the particles of ideal gas have no volume and there are no interactions between them) is given as PV ⫽ RT
(21)
where P is the gas pressure, V is the gas volume, T is the temperature, and R is the gas constant. Even at atmospheric pressure, the deviations of Eq. (21) from the behavior of real gases is pronounced. Van der Waals suggested an equation that gives a satisfactory description of the dependence P ⫽ f (V, T) for the gas phase, and which qualitatively describes the state of the liquid phase:
冤P ⫹ 冢Va 冣冥 [V ⫺ b] ⫽ RT 2
(22)
In the above equation the constant a takes into account the attraction forces between gas molecules, and the constant b represents the volume of the molecules (strong repulsive forces of short duration). The constants a and b in Eq. (22) can be considered as the
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 345
Table 1 Constants of the Van der Waals Equation (22) for Various Gases [58] Gas
a(L2 ⋅ bar ⋅ mol⫺2 )
b(sm 3 ⋅ mol⫺1 )
He H2 N2 CO 2
0.03457 0.2476 1.408 3.640
23.70 26.61 39.13 42.67
deviations in the behavior of real gases from ideal state. These values are listed in Table 1 [58] for the gases that are commonly used in gas chromatography as carrier gases. The increase in the nonideality follows the order: He ⬍ H 2 ⬍ N 2 ⬍ CO 2. Virial state equations are used more commonly for describing the properties of real gases. As an example, take the following equation [58]: PV ⫽ 1 ⫹ BP ⫹ CP 2 ⫹ . . . RT
(23)
The values of virial coefficients B, C, etc., are determined by the collision of molecules: paired collision (B), triple that (C), and so on. The virial coefficients depend on the temperature only. In gases of low density the paired collision is the most likely to occur. Therefore, in Eq. (23), all terms after the B coefficient can be neglected. Note that in Eq. (15), the second (mixed) virial coefficient B 12 is used to characterize the interactions between sorbate and carrier gas.
III. RELATIVE RETENTION AND ITS DEPENDENCE ON THE NATURE AND PRESSURE OF THE CARRIER GAS A. Definition of Relative Retention By the relative retention we mean a dimensionless value representing a ratio of two parameters of the same dimension. The relative retention is the value of the one magnitude (numerator) expressed in terms of the other magnitude (denominator). Functions of primary relative retention (i.e., relative retention time or relative retention volume) can be referred to the relative retention as well.
346
/ Berezkin
Relative retention is used routinely in gas chromatography (see, e.g., [1–4,7–10,12]). Hence, the theoretical interpretation of the relative retention and the dependence of the relative retention on the experimental parameters are of the greatest importance. The following expression is a valid relative retention time expression for two compounds i and j in the same phase (for example, in the SLP): t′Ri tRi ⫺ tM (tRi /tM ) ⫺ 1 ⫽ ⫽ t′Rj tRj ⫺ tM (tRj /tM ) ⫺ 1
αij ⫽
(24)
where tRi′ ⫽ t Ri ⫺ t M , t′Rj ⫽ t Rj ⫺ t M, tRi′ and t′Rj are the adjusted retention times for compounds i and j, respectively, t Ri and t Rj are the retention times for solutes i and j, respectively, and t M is holdup retention time (the retention time of an unretained compound). Equation (24) gives the most commonly used relative retention expression in chromatography. It is the ratio of the residence time of two compounds in the same SLP placed in a chromatographic column. In the specific case when compound j is a standard ( j ⫽ st) Eq. (24) can be written as αist ⫽
t′Ri t′Rst
(25a)
or αi ⫽
t′Ri t′Rst
(25b)
As shown in Ref. 60, the relative retention is a specific case of the following expression:
冢
Rel ⫽ P K ⫹
ri ⫺ rm rn ⫺ rm
冣
(26)
where Rel is the relative retention; P, K are the constants for the given system of retention units; r i, r m, r n are retention values (or their functions) for the compound i and for standards m and n. Table 2 gives known relative retention expression, based on Eq. (24)–(26). In our opinion, the retention factor k i, is one of the most important relative retention parameters. The retention factor k i is expressed as
Relative Retention Values in GC for the Specific Case of Eq. (24) Parameters of Eq. (24)
Relative retention Relative retention time (volume) Retention index
P 1 100
K
ri
rM
rn
Variant of Eq. (24)
0
ti
tM
tz
Rel iz ⫽
Z
log (t i ⫺ t M)
log (t z ⫺ t M )
log (t (z⫹1) ⫺ t M )
I i ⫽ 100z ⫹ 100 ⫽ 100z ⫹100 where
Arithmetic index
100
Z
ti
tz
t z⫹1
100
Z
ti
tM
tz
⫽
t′i t′ z
log [(t i ⫺ t M )/(t z ⫺ t M )] log [(t (z⫹1) ⫺t M )/(t z ⫺ t M )] log (t′i /t′z) , log t′(z⫹1) /t′z t′z⫹1 ⬎ t′i ⱖ t′z
A i ⫽ 100z ⫹ 100 where
Relative index
ti ⫺ tM tz ⫺ tM
ti ⫺ tz t (z⫹1) ⫺ t z
,
t z⫹1 ⬎ t i ⱖ t z
R i ⫽ 100z ⫹ 100
t′i t′z
, t′z⫹1 ⬎ t′i ⱖ t′z
Carrier Gas in Capillary Gas–Liquid Chromatography
Table 2
/ 347
348
/ Berezkin ki ⫽
t′Ri tRi ⫺ tM tRi ⫽ ⫽ ⫺1 tM tM tM
(27)
In this equation the numerator reflects the residence time of compound i in the stationary phase (for example, in the SLP), while the denominator reflects the residence time of an unretained compound M( j) in the column (i.e., in the mobile phase). In the chromatographic literature, the retention factor k i is not considered a relative retention. In our opinion, this is not logical. Equations (24) and (27) are the ratios of retention times for two compounds. Thus, the retention factor should be considered as one of relative retention values. However, note that the retention factor is a particular relative retention value. Conventional relative retention values (for example, relative retention time) are ratios of residence time for two compounds in different phases. In the case of the retention factor, the numerator is the retention (residence) time for compound i in the SLP, while the denominator is the retention time of an unretained compound M in the mobile phase. Therefore, in the author’s opinion, it seems reasonable to consider the relative retention time as a one-phase relative retention and the retention factor as a polyphase (namely, two-phase) relative retention. Needless to say, one-phase relative retention and one-phase chromatography are different and unrelated terms. Note that the present approach to the term ‘‘relative retention’’ in chromatography is not the conventional one. In the subsequent discussion, all of the following terms, the retention factor k i, the relative retention time α ij, and the Kovats’ retention index I i, are considered as ‘‘relative retention’’ terms. The retention factor, since it is proportional to the partition constant (see, e.g., [1,12]), 1 (28) β is of special interest for the determination of physicochemical equilibrium values. In Eq. (28), K Di is the partition constant for the ivolatile compound in the system SLP/gas, β is the phase ratio, β ⫽ V M /V l, V M is the volume of the gas phase in the column, and V l is the SLP volume in the column. Relative retention is used commonly in modern GLC to characterize and identify the solutes [1–4]. Also, it is of interest for measurements of equilibrium physicochemical values [18,19]. ki ⫽ KDi ⋅
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 349
B. Theoretical Aspects of the Dependence of Relative Retention on the Nature and Pressure of the Carrier Gas The equations obtained in the 1960s for the dependence of the absolute retention on the carrier gas pressure and its nature reflect the physical interaction between the carrier and sorbate at the conditions of GLC. The specific retention volume is directly proportional to the adjusted retention time (see, e.g., [1,12]). Hence, using the above dependences, the following equation for the relative retention time can be obtained: βP t′i Vgi (Pav ) Vgi (0) ⋅ e ′i av Vgi (0) (βi′⫺(β′j )Pav ⫽ ⋅e ⫽ ⫽ t′j Vgj (Pav ) Vgj (0) ⋅ e βj′Pav Vgj (0) ⫽ αij (0) ⋅ e (βi′⫺(βj′ ) Pav ⫽ αij (0) ⋅ e ∆β ′ij ⋅ Pav
αij ⫽
(29)
Note that the value of ∆β′ij ⫽ (β′i ⫺ β′j ) is rather small. We can estimate ∆β′ij using the experimental data listed in the review by Young [42]. We calculated ∆β′ij values for the following cases. 1. Sorbates: 2-methylpentane and 2-methylbutane; dinonylphtalate as SLP; argon as carrier gas; temperature 298 Κ, ∆β′ij ⫽ 0.3 ⫻ 10⫺2 atm⫺1. 2. Sorbates: n-octane and 2,2,4-trimethylpentane; squalane as SLP; hydrogen as carrier gas; temperature 338 Κ, ∆β′ij ⫽ 0.9 ⫻ 10⫺2 atm⫺1. Typical average column pressures of carrier gas are between 2 and 5 atm. Thus, at Pav ⫽ 3 atm, for the above two examples, the values of (β′i ⫺ β′j )Pav are 0.009 and 0.027, respectively. We see that the values of (β′i ⫺ β′j )Pav are rather small. We can, then, expand e (βi′⫺βj′)PavMcLaurin series, and if we take only the first two terms in the series we obtain (see, e.g., [61]), e (βi′⫺β′j )Pav ⬇ 1 ⫹ (β i′ β′j )Pav
(30)
With Eq. (30) in Eq. (29) we can obtain the linear relationship reflecting the dependence of α ij on Pav: αij ⫽ αij (0)[1 ⫹ (β′i ⫺ β′j )Pav] ⫽ αij (0) ⫹ αij (0)(β′i ⫺ β′j )Pav ⫽ aij (0) ⫹ bαi Pav
(31)
350
/ Berezkin
The Kovats’ indices are commonly used in GC [1–4,7–10,12]. The Kovats’ index (I i) for compound i can be expressed as Ii ⫽ 100z ⫹ 100
log αiz log α(z⫹1) z
(32)
where α iz is the relative retention of compound i relative to n-alkane, containing z carbon atoms, and, α (z⫹1)z is the relative retention of n-alkane containing (z ⫹ 1) carbon atoms to the n-alkane with z carbon atoms. The retention times relationship is t z⫹1 ⬎ t i ⱖ t z, where t z and t (z⫹1) are the retention times of n-alkanes that contain z and (z ⫹ 1) carbon atoms, respectively, and t i is the retention time of compound i. Equation (29) and Eq. (32) give log
αiz (Pav) ⫽ 0.4343(β′i ⫺ β′z )Pav αiz (0)
(33)
log
α(z⫹1)z (P) ⫽ 0.4343(β′(z⫹1) ⫺ β′z ) Pav α(z⫹1)z (0)
(34)
Substituting Eqs. (33) and (34) into (32) gives Ii ⫽ 100z ⫹ 100
log αiz (0) ⫹ 0.4343 ∆β′iz Pav log α(z⫹1) z (0) ⫹ 0.4343 ∆β′(z⫹1) z Pav
(35)
where ∆β′iz ⫽ (β′i ⫺ β′z )
(36)
∆β′(z⫹1)z ⫽ (β′(z⫹1) ⫺ β′z )
(37)
When we expand Eq. (35) in a McLaurin series and use only the first two terms, we obtain the following relationship for I i : Ii ⬇ Ii (0) ⫺ bIi Pav
(38)
where bIi ⫽ 100
冤
冥
∆β′(z⫹1)z log αiz (0) 0.4343 ⫺ ∆β′iz log α(z⫹1) z (0) log α(z⫹1)z (0)
(39)
Equations (13) and (14) are used to obtain the dependence of the retention factor on the average column pressure. Since the β′i Pav values are small (for the listed examples the values are ⬃4 ⫻ 10⫺2 ), eβi′ Pav can be approximated by a McLaurin series to give
Carrier Gas in Capillary Gas–Liquid Chromatography Vgi (Pav ) ⫽ Vgi (0)(1 ⫹ β′i Pav )
/ 351 (40)
Since Vgi ⫽ Aki
(41)
where A⫽
β ρ
(42)
(ρ is the SLP density and β is the phase ratio), then combining Eqs. (40) and (41), the relationships for k i (Pav) can be written as ki (Pav ) ⫽ ki (0)(1 ⫹ β′i Pav ) ⫽ ki (0) ⫹ bki Pav
(43)
bki ⫽ ki (0) ⋅ β′i
(44)
From the equations obtained, it is seen that the relative retention parameters α ij, I i, and k i are linearly dependent (1) on the average column pressure of carrier gas and (2) on the nature of carrier gas used, since the coefficient β′i depends on such characteristics of the carrier gas as the second (mixed) virial coefficient and the carrier gas solubility in the SLP. The relative retention time α ij expressed as an exponential function [see Eq. (29)] can be presented as a linear equation [see Eq. (31)]. To make an estimate of approximate relationships obtained and original equations, we compare the values of the initial exponential expression of α(Pav ) [Eq. (29)] and the linear approximation α(Pav ) [Eq. (31)]. The two functions were calculated for the same average column pressures, Pav , over the range Pav ⫽ 1.0–5.5 atm. Experimental data on the relative retention time α ij (a) and the values ∆β′i ⫽ (β′i ⫺ β′j ) were measured using heavy SF 6 as the carrier gas [62]. The values α e (Pav ) and α c (Pav ) for three compounds are listed in Table 3, where subscript e indicates experimental values and c indicates calculated values. As seen from the table, α c(Pav) ⬇ α e (Pav ) with high accuracy. The listed comparison between the accurate function α e (Pav ) and the approximate that α c (Pav ) justifies changing the complex exponential function α c (Pav ) for the more simple and demonstrative linear term α c (Pav ). The important functional dependence of the relative retention on the average column pressure and on the nature of the carrier
352
Heptanol-1 Pav (atm) 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5
Decanol-1
2,6-Dimethylphenol
α 1 (P av )
α 2 (P av )
α 1 (P av )
α 2 (P av )
α 1 (P av )
α 2 (P av )
0.125 0.126 0.127 0.128 0.129 0.130 0.131 0.132 0.133 0.135
0.125 0.126 0.127 0.128 0.129 0.130 0.131 0.132 0.133 0.134
0.438 0.440 0.442 0.443 0.445 0.447 0.448 0.450 0.452 0.454
0.438 0.440 0.442 0.443 0.445 0.447 0.448 0.450 0.452 0.454
0.847 0.848 0.849 0.851 0.852 0.854 0.855 0.856 0.858 0.859
0.847 0.848 0.849 0.851 0.852 0.854 0.855 0.856 0.858 0.859
Experimental conditions: column 30 m ⫻ 0.25 mm coated with PEG-20M (d f ⫽ 0.25 µm), temperature 155°C, carrier gas SF 6 , standard n-dodecanol.
/ Berezkin
Table 3 Comparison of the Exponential Function α 1 (P av ) and Its Linear Approximation α 2 (P av ) over the Range of Average Column Pressures P av 1–5 atm
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 353
gas was obtained theoretically in our studies [62–67]. The equations indicate that the relative retention is linearly dependent on the average column pressure of the carrier gas at the conditions of conventional gas chromatography (Pav ⱕ 5 atm).
C. Experimental Confirmation of the Relative Retention Dependence on the Average Column Pressure and the Nature of the Carrier Gas Qualitative Studies The qualitative result is a necessary and important stage in any research. Here, we establish the variation of the relative retention with the average column pressure and with the nature of the carrier gas (i.e., when one carrier gas is changed for another). The discussed dependence is of special interest for capillary columns, since the resolution and measurement accuracy of the relative retention values are significantly greater than those for packed columns. For example, the data of Matisova and Kuran [68] show reproducibility of the retention index with a mean standard deviation of S ⫽ 0.03 i.u., and with the confidence interval ⫾0.10 i.u. Recently, the reproducibility in the retention index has been increased due to the increasing stability of all the parts of the gas chromatograph. Vigdergauz and co-workers [69–74] performed the first studies in GLC on the influence of the pressure and nature of the carrier gas on the relative retention. Their work examined the retention index with increasing carrier gas pressure of high pressures (tens of atmospheres). The experiments were performed over a wide range of high pressures because the researchers did not have precise equipment that allowed them to study carrier gas effects in routine chromatography (Pav ⬍ 5 atm). Twenty years later, Vigdergauz summarized the results of the studies in this field, saying that: ‘‘Unlike conventional gas chromatography, where the nature of a permanent carrier gas used and its operating mode have a significant influence on band-broadening but not on phase equilibrium, in the described variant (barochromatography) the permanent carrier gas is used at such pressures (up to several MPa) that the characteristics of the carrier gas can be described by an equation of state with the second and the third virial coefficients’’ [75]. Hence, Vigdergauz believed that at the conditions of routine chromatography (at pressures up to 5–10 atm and using
354
/ Berezkin
permanent gases), the carrier gas does not influence the retention values (and column selectivity). As an example, data by Vigdergauz and Semkin [69] show that at low pressures (up to 2.1–2.5 atm) (see Table 4 [69]), when helium is replaced by another carrier gas (argon, carbon dioxide), the relative retention (retention factor and relative retention time) barely change. Vigdergauz and Semkin [69–74] were the first to establish empirically the linear dependence of the retention index I i on the average carrier gas pressure Pav at high pressures in capillary GLC: Ii ⫽ A ⫹ BPav
(45)
where A and B are constants. The dependence was studied experimentally for hydrocarbons only (i.e., for compounds of the same class). Rijks was the first to pay attention to the variation of the retention index for some hydrocarbons with a change in the average column pressure [76]. As noted by him [76], when the average column pressure is increased from 2.31 atm to 3.78 atm (∆P ⫽ 1.47 atm), the change in the retention index ∆ I is 0.05 i.u. for hexane, 0.23 i.u. for cyclohexane, and 0.28 i.u. for benzene. Sojak and Rijks [77–80] performed systematic studies on the influence of the nature of the carrier gas on the retention index of some aromatic compounds on a column coated with squalane; see Table 5 [77]. As can be seen, when hydrogen is replaced by nitrogen, the retention index of the aromatic hydrocarbons is changed between 1.01 and 1.27 i.u. This ∆ I is much greater than the experimental error. Table 4 Influence of the Carrier Gas Nature on the Retention Factor (k) and the Relative Retention Time (α) of Hydrocarbons in Capillary GLC (calculated by the data listed in ref. 69) Helium (2.5 atm) Sorbate n-Pentane n-Hexane Methylcyclopentane Benzene
Argon (2.3 atm)
Carbon dioxide (2.1 atm)
k
α
k
α
k
α
0.68 2.16 2.68 3.32
0.31 2.14 2.68 3.35
0.66 2.14 2.68 3.35
0.31 1.0 1.25 1.57
0.68 2.18 2.73 3.35
0.31 1.0 1.25 1.54
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 355
Table 5 Variation of the Retention Index of Aromatic Hydrocarbons with Nitrogen in Place of Hydrogen as Carrier Gas [77] Retention index Compound Benzene Toluene Ethylbenzene p-Xylene o-Xylene
H2
N2
∆I ⫽ I(N 2 ) ⫺ I(H 2 )
642.07 749.30 838.94 853.26 874.41
643.34 750.49 840.12 854.27 875.56
1.27 1.19 1.18 1.01 1.15
Experimental conditions: column 100 m ⫻ 0.25 mm coated with squalane, temperature 70°C, inlet pressure 4 atm.
Table 6 [78] gives the variation of the retention index for alkylbenzenes with increasing (1–2 atm) carrier gas pressure (nitrogen) and with replacing hydrogen by nitrogen on a column coated with tricyanoethoxypropane (TCEP). We see that both the carrier gas pressure and the nature of the carrier gas have a significant influence on the retention index of alkylbenzenes. The studies by Sojak, Rijks, and co-workers [77–80] establish the experimental dependence of the retention index of hydrocarbons on the nature and pressure of the carrier gas in GLC. It was also
Table 6 Variation of Retention Index for Alkylbenzenes with the Pressure and Nature of Carrier Gas on Column Coated with TCEP [78] Alkylbenzene
∆I (1–2 atm, N 2 ) ∆P
∆I ⫽ I(H 2) ⫺ I(N 2) (1 atm)
Toluene 1,4-Xylene 1,2-Xylene 1-Methyl-4-ethylenbenzene 1,3-Diethylbenzene 1,3-Dinethyl-5-ethylbenzene n-Pentylbenzene 1,2,4,5-Tetramethylbenzene
1.0 0.7 1.2 1.0 1.1 0.9 1.4 1.3
1.7 1.3 1.2 1.1 1.1 1.2 1.1 1.1
Experimental conditions: column 50 m ⫻ 0.25 mm coated with 1,2,3-tricyanoethoxypropane (TCEP), oven temperature 95.4°C.
356
/ Berezkin
shown that the carrier gas effects on the retention index depend on the type of hydrocarbon: alkanes ⬍ alkenes ⬍ cycloparaffines ⬍ aromatic hydrocarbons. Berezkin and Korolev [81,82] showed that the behaviors established by Sojak, Rijks, and co-workers [77–80] are valid not only for hydrocarbons but also for other classes of organic compounds. The retention index for different classes of organic compounds, using two carrier gases (N 2 and He), are listed in Table 7 [81]. We see that the change in the retention index depends on the nature of the sorbate. Thus, the nature and pressure of the carrier gas have a significant influence on the retention index for various classes of organic compounds. Unfortunately, most chromatographers do not pay attention to the above dependencies, even though they contain new and important information. The conventional opinion on the relative retention chromatographic constant needs to be revised. The following additional studies of the above-mentioned functional dependence, and experimental verification of Eqs. (31), (38), and (43), stress further the necessity of revising the conventional approach to the role of the carrier gas in GLC. Quantitative Studies in the Range of Low Pressures Different carrier gases differ in the degree of their nonideality. Therefore, we would expect them to differ in their effect on relative retention. We will divide the carrier gases into two groups: ‘‘light’’ and ‘‘heavy.’’ ‘‘Light’’ carrier gases behave closer to ideal gases.
Table 7 Influence of the Carrier Gas Nature on the Retention Index of Organic Compounds of Different Classes [81] Retention index Sorbate Octanol-1 2,6-Dimethylphenol 2,6-Dimethylaniline Naphthalene Methylpelargonate
N2 1052.90 1079.86 1133.06 1155.43 1206.55
⫾ ⫾ ⫾ ⫾ ⫾
∆I ⫽ I(N 2) ⫺ I(H 2)
He 0.09 0.13 0.08 0.06 0.10
1052.59 1079.33 1132.54 1154.74 1206.56
⫾ ⫾ ⫾ ⫾ ⫾
0.09 0.11 0.08 0.06 0.04
0.31 0.53 0.52 0.69 ⫺0.01
Experimental conditions: column 25 m ⫻ 0.2 mm coated with SE-30 (d f ⫽ 0.13 µm), temperature 90°C.
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 357
‘‘Light’’ Carrier Gases Recently, we have studied the functional depenencies of the relative retention on the average column pressure for the following carrier gases: He, H 2 , N 2 , N 2 O, and CO 2. Equations (31), (38), and (43) agree well with the experimental data for organic compounds of different chemical classes, obtained on capillary columns of different diameter, length, and polarity of the SLP (nonpolar SE-30 and polar PEG-20M) [63–67,83–85]. Typical results on the dependence of the retention index on the average column pressure of helium are presented in Table 8 [66]. We see that Eq. (38) was found to fit well the experimental data even though helium is closest to an ideal gas. Our experiments were done on a high-performance, narrowbore capillary column (0.14 mm I.D.). In recent years, this type of column has received considerable attention (see, for example, [86]). With other carrier gases (hydrogen and, especially, nitrogen), the dependence of the retention index on the average column pressure (∂I i /∂Pav) ⫽ b Ii is more pronounced than for helium (see Table 9 [24,30]). It is desirable then, to use helium as carrier gas to minimize the column pressure effect. The coefficient b αij ⫽ α ij(0)(β′i ⫺ β′j ) [see Eq. (38)] can be both positive and negative. This caused by the fact that the difference (β′i ⫺ β′j ) could have both positive and negative value (see Tables 8 and 9), depending on whether β′i ⬎ β′j or β′i ⬍ β′j . The variation of the retention factor k i with the average column pressure Pav is greater for CO 2 than that for N 2. This observation can be explained by the greater deviation of CO 2 from ideal gas behavior than that of nitrogen (see Table 10 [27]). As seen from Table 10, replacing N 2 by CO 2 results in an increase in the retention factor sensitivity to the carrier gas pressure, and an increase in the carrier gas effects on the retention and separation.
Table 8 Parameters of Eq. (38) for Some Organic Compounds [66] Sorbate Octanol-1 2,6-Dimethylphenol 2,6-Dimethylaniline Naphthalene
I oi 1054.61 1090.07 1148.81 1176.40
⫾ ⫾ ⫾ ⫾
b Ii 0.03 0.11 0.10 0.07
0.0056 ⫺0.0825 ⫺0.0939 ⫺0.1032
⫾ ⫾ ⫾ ⫾
0.0022 0.0104 0.0103 0.0069
R
SD
0.87 0.98 0.98 0.99
0.022 0.093 0.092 0.061
Experimental conditions: column 75 m ⫻ 0.14 mm coated with SE-30 (d f ⫽ 0.3 µm), helium as carrier gas, temperature 120°C (from here on, R is correlation coefficient, SD is standard deviation).
358
/ Berezkin
Table 9 Comparison of b Ii Coefficients [Eq. (38)] for Various Organic Compounds on a Capillary Column Coated with SE-30 at 120°C Using He, H 2, and N 2 as Carrier Gases [24,30] b Ii Sorbate Octanol-1 2,6-Dimethylphenol 2,6-Dimethylaniline Naphthalene
He
H2
N2
0.0056 ⫺0.0825 ⫺0.0939 ⫺0.1032
0.091 0.137 0.155 0.153
0.203 0.326 0.364 0.411
Table 11 [63] presents the limiting values of the retention indexes I io for two carrier gases, N 2 and CO 2. As seen, I iN2 (0) ⬇ I iCO2 (0). Thus, the limiting values of the retention index do not depend on the nature of the ‘‘light’’ carrier gases. Table 12 shows the variation in retention index due to a 2-atm increase of the carrier gas pressure for various carrier gases (helium, hydrogen, and nitrogen). We see that nitrogen has the most influence on the retention index, while helium has the least effect. The change in the retention index of all compounds studied with two carrier gases (H 2 and N 2 ) is much greater than that attributable to experimental error. With helium as the carrier gas, ∆ I is significant only for three of the four compounds studied.
Table 10 Characteristics of the Linear Dependence of the Retention Factor [Eqs. (43) and (44)] for Various Carrier Gases [27] N2
CO 2
No.
k i(0)
b ki
R
k i(0)
b ki
R
1 2 3 4
8.364 6.782 4.767 1.479
⫺0.132 ⫺0.113 ⫺0.074 ⫺0.022
0.99 0.99 0.99 0.99
8.392 6.794 4.772 1.476
⫺0.218 ⫺0.179 ⫺0.112 ⫺0.032
0.99 0.99 0.99 0.99
Experimental conditions: column 25 m ⫻ 0.25 mm coated with PEG-20M (d f ⫽ 0.5 µm), temperature 125°C. Sorbates: 2,6-dimethylphenol (1), 2,6dimethylaniline (2), naphthalene (3), and methylpelargonate (4).
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 359
Table 11 Limiting (Invariant) Values of the Retention Index I io for Two Carrier Gases (N 2 and CO 2) and Two SLPs [63] Sorbate
I oi (N 2)
I oi (CO 2)
SLP: SE-30, 130°C, standard n-alkanes homologous series Octanol-1 2,6-Dimethylphenol 2,6-Dimethylaniline Naphthalene
1052.66 1091.03 1151.61 1182.09
⫾ ⫾ ⫾ ⫾
0.19 0.02 0.18 0.07
1052.51 1090.88 1151.40 1181.96
⫾ ⫾ ⫾ ⫾
0.15 0.05 0.01 0.05
SLP: PEG-20M, 100°C, standard n-alcohols homologous series Methylpelargonate Decanol-2 Naphthalene 2,6-Dimethylphenol
758.99 858.66 986.66 1142.12
758.50 857.74 986.53 1142.46
In our studies we examined both commonly used carrier gases (helium, hydrogen, and nitrogen) and the rarely used nitrous oxide. Table 13 [67] compares the linear dependence of the retention factor on the average column pressure for helium and for nitrous oxide. The experimental data fit the linear behavior well (Eq. [43]). Note that with nitrous oxide as carrier gas, the dk i /dP av value is 2.3–3.9 times as much as with helium. The difference in the retention characteristics when using various carrier gases can be used for the identification of the solutes. The sorbate’s nature is revealed in the variation of the second mixed Table 12 Variation of the Retention Index with Increasing Average Column Pressure for 2 atm (data from [63,65,66]) Sorbate Octanol-1 2,6-Dimethylphenol 2,6-Dimethylaniline Naphthalene
He
H2
N2
0.01 ⫺0.16 ⫺0.19 ⫺0.20
0.18 0.28 0.31 0.30
0.40 0.66 0.72 0.82
Experimental conditions: column 75 m ⫻ 0.14 mm coated with SE-30 (d f ⫽ 0.3 µm), temperature 120°C.
360
Nitrogen oxide
Helium
Compound
k i(0)
b ki
R
SD
K i(0)
b ki
R
SD
n-Decane Octanol-1 2,6-Dimethylphenol n-Undecane 2,6-Dimethylaniline Naphthalene n-Dodecane
2.317 3.285 4.733 4.334 5.734 6.592 8.208
⫺0.096 ⫺0.124 ⫺0.289 ⫺0.165 ⫺0.235 ⫺0.257 ⫺0.297
0.983 0.977 0.890 0.975 0.980 0.989 0.995
0.043 0.066 0.091 0.116 0.094 0.072
2.278 3.241 3.925 4.364 5.586 6.459 8.097
⫺0.032 ⫺0.041 ⫺0.045 ⫺0.071 ⫺0.065 ⫺0.074 ⫺0.076
0.943 0.996 0.993 1.000 0.999 0.997 0.978
0.013 0.004 0.006 0.000 0.002 0.006 0.019
Experimental conditions: column 25 m ⫻ 0.16 mm, SLP SE-30 (d f ⫽ 0.25 µm), 100°C.
/ Berezkin
Table 13 Characteristics of the Linear Dependence [Eq. (43)] Using Two Carrier Gases (Helium and Nitrogen Oxide) [67]
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 361
virial coefficient B 12 , molar volume V∞i , and d ln V∞i / dX 2 [see Eqs. (13)–(17)]. On the basis of the experimental results we can contribute as follows for ‘‘light’’ carrier gases: (1) linear equations (31), (38), and (43) reflecting the dependence of Rel i(Pav ), fitting well to the experimental data for He, H 2, N 2, CO 2 , and N 2 O as carrier gases. (2) The relative retention values commonly used in the chromatographic literature are not chromatographic constants because of their dependence on the carrier gas nature and pressure. (3) The influence of the carrier gas on the relative retention (for example, retention factor) is in the following order: He ⬍ H 2 ⬍ N 2 ⬍ CO 2 (the b ki ⫽ ∂k i /∂Pav value increases in the same direction). (4) The limiting value of the relative retention lim Rel i (Pav ) ⫽ Rel i(0) at carrier gas pressure approaching zero, Pav → 0, is a true chromatographic constant of the solute. (5) Rel iHe(0) ⬇ Rel iH2(0) ⬇ Rel iN2(0) ⬇ Rel iCO2(0) is true for all the compounds studied. Thus the limiting value Rel i(0) is a chromatographic constant because it is practically independent of the nature and pressure of the carrier gas. The results obtained above lead to the conclusion that some of the existing fundamental opinions in GLC need to be revised. ‘‘Heavy’’ Carrier Gases It is of interest to examine the influence of nonstandard, ‘‘heavy’’ carrier gases on the relative retention for the detailed study of the phenomenon. The reasons for studying ‘‘heavy’’ carrier gases (molecular weight more than 100) are as follows. (1) These gases should have the most influence on the retention (as a consequence of their nonideal nature [87]) and their application could be of practical importance. (2) Some deviations from the abovediscussed pressure and nature of the carrier gas dependencies for traditional carrier gases can be expected. (3) Finally, these gases do not appear to have been studied as carrier gases. Only a single study was published [62] on the dependence of the retention index on the pressure of a ‘‘heavy’’ carrier gas such as SF 6 (molecular weight 146). The results for the relative retention α ij are listed in Table 14 [62]. As seen, the linear dependence α ij ⫽ f (Pav ) conforms well to the experimental data. However, limiting values of the retention index using the heavy carrier gas SF 6 are different from those using He. This difference is inconsistent with the relationship Rel G1 (0) ⬇ Rel G2(0) noted earlier for the ‘‘light’’ carrier gases.
362
/ Berezkin
Table 14 Parameters of Eq. (31) for the Relative Retention Time of Organic Compounds Using SF 6 as Carrier Gas [62] Sorbate Heptanol-1 Methylpelargonate Octanol-1 Decanol-2 Nonanol-1 Decanol-1 Naphthalene 2,6-Dimethylphenol
α ij
b ij
R
0.123 0.153 0.187 0.239 0.290 0.435 0.491 0.844
⫺0.0020 ⫺0.0025 ⫺0.0020 ⫺0.0041 ⫺0.0058 ⫺0.0033 ⫺0.0024 ⫺0.0027
0.99 0.98 0.99 0.99 0.99 0.99 0.99 0.99
Experimental conditions: column 30 m ⫻ 0.25 mm coated with PEG-20M (d f ⫽ 0.25 µm), temperature 155°C, standard n-dodecanol.
In summary, the following properties of ‘‘heavy’’ carrier gas such as SF 6 are absorbed [62]. (1) The retention index is linearly dependent on the average column pressure of SF 6 (in all the cases studied the correlation coefficient is more than 0.99). (2) The limiting values of the retention index I oi when using SF 6 and He differ unlike the case of the ‘‘light’’ carrier gases studied earlier. (3) A qualitative difference in the peak shape on the column coated with polar (PEG20M) and nonpolar (SE-30) SLP is found; the peak of the sorbate is asymmetric on the column coated with the nonpolar SLP.
D. Correlation for Calculation of Relative Retention Values Measured on the Same Column Using Various Carrier Gases As shown above, the relative retention depends on the nature and pressure of the carrier gas. Thus, it is reasonable to develop a method for the calculation of the relative retention at condition of using the carrier gas G 1 and at its pressure P1 (it is assumed that the relative retention values are known for a number of compounds both at G 1 , P1 and at G 2 , P2 ).
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 363
The approach to the problem consists in the determination of the following dependence using the limited data for the mentioned conditions: Reli (P1 ,G1 ) ⫽ f [Reli (P2 ,G2 )]
(46)
where Reli (P1 ,G1 ) and Reli (P2 ,G2 ) are relative retention for compound i at average column pressure P1 (carrier gas G1 ) and P2 (carrier gas G2 ), respectively. From the experience gained in the comparative calculation of the physicochemical properties of organic compounds (see [88]), the following linear equation for the retention factor [67] can be proposed: ki (P1 ,G1 ) ⫽ Aki (P2 ,G2 ) ⫹ B
(47)
where A and B are constants. Experimental verification of Eq. (47) is found in Table 15 [67]. It is clear that Eq. (47) describes the experimental data well. Therefore, the data obtained at a given set of experimental conditions can
Table 15
Experimental Verification of the Linear Correlation (47) [67] Parameters of Eq. (47)
No. 1 2 3 4 5 6 7
Carrier gas G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2
He H2 He H2 He H2 He N2 He N2 N2 CO 2 N2 CO 2
P av (atm)
A
B
R
SD
2.67 2.56 4.52 4.15 11.85 11.3 6.38 6.76 2.67 2.58 5.26 5.67 1.78 1.68
0.0120
1.0484
0.9999
0.00426
0.0089
0.9635
1.0000
0.00196
0.0255
0.9450
0.9999
0.01655
0.2450
0.9469
0.9999
0.01488
0.0061
0.9661
0.9999
0.00545
0.0301
0.9002
0.9999
0.00477
0.0150
0.9568
1.0000
0.00343
Experimental conditions: column 75 m ⫻ 0.14 mm coated with SE-30, temperature 120°C.
364
/ Berezkin
be used for deriving the data for another set of experimental conditions, provided that limited data are known at the new condition.
IV.
STEAM CAPILLARY GAS–LIQUID CHROMATOGRAPHY AND THE INFLUENCE OF WATER VAPOR ON RELATIVE RETENTION
If we consider the carrier gas as a factor for separation improvement, it is of interest to extend the number of compounds used as mobile phases (or as an active component in a mobile phase based on an inert carrier gas). It is expected that inorganic and organic compounds having unusual properties should have a significant influence on the relative retention and on the column selectivity. Water molecules, having significant dipole moment (see, e.g., [89–92]), can interact with each other as well as with polar molecules of other compounds via hydrogen bonds with O, N, F, Cl, or S atoms in other molecules. Hence, there were attempts to develop steam chromatography (see the reviews [15,93–96]). In addition, attention has been concentrated on the use of water-containing mobile phases in GC on packed columns. In spite of the fact that capillary steam chromatography was developed many years ago [97], its advantages have not been adequately demonstrated up to the present [98–102]. Water vapor does not affect the characteristics of flame ionization detectors (FID), which is used commonly in this variant (see the review [103]). FID insensitivity to water is an advantage of watercontaining mobile phases. It is worth noting that a change in retention was found when using moist carrier gas on capillary columns coated with PEG-20M, which was obtained by bubbling dry gas through the water at ambient temperature (see Table 16 [98]). As is apparent from the table, the replacement of the dry carrier gas by the moist one has a significant influence on the retention index: for alcohols the variation is ⬃13 i.u., for aromatic compounds it is ⬃4.2–4.5 i.u. Hence, even a small quantity of polar water in the carrier gas leads to a significant variation in the relative retention. A second characteristic of moist carrier gas is that the retention factor k i increases as the water partial pressure increases [98,99], unlike the case with traditional gases (He, H 2, N 2 , CO 2 ). These results indicate the different nature of water interaction with solutes [99]. When using water vapor with columns coated with polar phase,
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 365
Table 16 Influence of Water Introduction (⬃20 mm Hg) into the Carrier Gas (He) on the Retention Index of Organic Compounds on a Capillary Column (25 m ⫻ 0.2 mm) Coated with PEG-20M (d f ⫽ 0.13 µm) at 70°C [98]
Sorbate
Dry carrier gas, I d (i.u.)
n-Propanol Toluene p-Xylene o-Xylene n-Pentanol
1024.75 1038.81 1128.29 1174.60 1235.59
⫾ ⫾ ⫾ ⫾ ⫾
0.16 0.09 0.06 0.04 0.08
Carrier gas saturated with water vapor at 22°C, I w (i.u.) 1038.49 1043.27 1132.53 1179.15 1248.53
⫾ ⫾ ⫾ ⫾ ⫾
0.20 0.22 0.18 0.31 0.48
∆I ⫽ I w ⫺ I d (i.u.) 13.74 4.46 4.24 4.55 12.94
the water dissolves in the SLP. This leads to changes in the SLP content and the SLP becomes more polar. If a nonpolar SLP, such as SE-30, (water practically does not dissolve in this phase) used in analogous experiments, the introduction of water into the carrier gas has a little influence on the retention (the variation of retention index is ⬇ 0.1 i.u.) Let us consider in detail the mechanism of water vapors influence on the relative retention. As a result of water dissolving in the polar SLP (PEG-20M), its concentration in the SLP increases directly with the water partial pressure in the mobile phase [100]. Therefore, it would, be expected that the water effect on the retention of polar compounds would increase directly with the water concentration in the mobile phase. The experimental study shows that the relative retention increases in direct proportion to the partial pressure of water vapors in the carrier gas. The following relationships are observed [100]: ki ⬇ ki lim ⫹ gki PH2O
(48)
αij ⬇ αij lim ⫹ gαi PH2O
(49)
where k i lim ⫽ limk i (at PH2O → 0) and α i lim ⫽ lim α ij (PH2O → 0), g ki and g αi are constants, and PH2O is the partial pressure of water vapors in the carrier gas (N 2 ). Table 17 [100] presents the characteristics of Eq. (49). The values of the correlation coefficient and the standard deviation show
366
/ Berezkin
Table 17 Characteristics of Eq. (31) for the Dependence of Relative Retention Time α ij on the Partial Pressure of Water Vapor in the Binary Carrier Gas (N 2 ⫹ H 2O) [100] Sorbate
α ij lim
gα
R
SD
Ethanol Decane Propanol Undecane Butanol Pentanol
0.22 0.37 0.48 0.75 1.01 2.01
0.001 ⫺0.0005 0.002 ⫺0.001 0.0044 0.087
0.94 0.97 0.94 0.98 0.94 0.97
0.34 0.008 0.54 0.017 0.107 0.143
Experimental conditions: column 30 m ⫻ 0.25 mm coated with PEG-20M (d f ⫽ 0.5 µm), standard ethylbenzene.
that the equation discussed is in good agreement with the experimental data. Capillary columns coated with cross-linked SLP, e.g., PEG-20M, are rather stable to the action of water vapor. When operated for 300 h in a flow of mixed eluents with high concentrations of water vapor at oven temperatures up to 98°C, the main characteristics of the column such as efficiency, selectivity, and capacity are practically constant [101]. Using a water-containing mobile phase with varying content of water allows for variation of relative retentions on columns coated with a polar SLP. The relative retention parameters (retention factor, relative retention time, and retention index) change linearly with increasing water partial pressure in the carrier gas. Thus, gradient elution can be used as well. The above behavior predicts a wide application of capillary steam chromatography. As an example, we note the sharp separation of the hydrocarbon part and the polar components contained in petroleum [102].
V.
INFLUENCE OF THE NATURE AND PRESSURE OF THE CARRIER GAS ON SEPARATION
The task of the analyst is to improve the resolution of the critical solute pair. The influence of the nature and pressure of the carrier gas on the resolution (R s) must be taken into account [104]. As is
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 367
known (see e.g., [31]), the separation factor is a function of selectivity S, efficiency E, and capacity C: R ⫽ SEC
(50)
where S⫽
αij ⫺ 1 αij
(51)
E⫽
√N 4
(52)
C⫽
k k⫹1
(53)
where α is the relative retention (or selectivity factor) of the two compounds forming the critical separation pair, N is the plate number of the column and k is the retention factor of the second compound in the separation pair. Note that when α ⬇ 1.0, then S ⬇ (α ij ⫺ 1). Let us consider the dependence of R s on the nature and pressure of the carrier gas assuming that the dependence of E and C on the average column pressure can be neglected and that α ij depends on the carrier gas pressure: Rs ⫽ EC[αij (Pav ) ⫺ 1]
(54)
Using Eq. (31) in Eq. (54) gives Rs ⫽ EC [αij (0) ⫺ 1] ⫹ Ecbαij Pav ⫽ Rscl (0) ⫹ Ecbαij Pav
(55)
where Rscl (0) ⫽ EC [αij (0) ⫺ 1]
(56)
It follows from Eq. (55) that the peak resolution R s depends on the nature and pressure of the carrier gas. Thus, varying column pressure and changing carrier gas can influence the separation, and a different mobile phase in GC can improve the separation selectivity and increase the R s value. Equations (55) and (56) for the peak resolution were derived in Ref. 104. As an example we will consider the influence of the binary carrier gas composition (N 2 ⫹ H 2O) on the peak resolution, and peaks
368
/ Berezkin
inversion, when the partial pressure of water in the mobile phase is changed [100]. An example of the separation of an organic mixtures, using dry and moist carrier gases, are presented in Fig. 1. Figure 1a shows the chromatogram of the organic mixture using dry carrier gas (N 2 ), and Fig. 1a′ is for moist carrier gas (PH2O ⬇ 70 mm Hg). With the
Fig. 1 Influence of water vapor on the elution order of the components of the analyzing mixture. Solutes: 1, ethanol; 2, decane; 3, propanol; 4, undecane; 5, ethylbenzene; 6, p-xylene; 7, butanol; 8, o-xylene. Experimental conditions: quartz capillary column 30 m ⫻ 0.5 mm coated with PEG-20M (film thickness 0.5 µm), 71°C, carrier gas nitrogen (a, b), nitrogen ⫹ steam (a′ ⫽ 70 mm Hg, b′ ⫽ 165 mm Hg).
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 369
Fig. 2 Chromatogram of pesticides: dicofol (1) and methoxychlor (2) using helium (a) and carbon dioxide as carrier gas (b). Experimental conditions: capillary column 12 m ⫻ 0.32 mm coated with Carbowax 20M (film thickness 0.25 µm), 260°C. Retention times, min: (a) dicofol, 8, 11; methoxychlor, 8, 26; (b) methoxychlor, 16; dicofol, 16, 32.
dry carrier gas, butanol elutes before o-xylene, while with the moist carrier gas the first peak is o-xylene, and the second is butanol. Another inversion of the elution order of alcohols and n-alkanes when replacing the dry carrier gas with a moist one is shown in Figs. 1b and 1b′. Elution order inversion and improvement of the separation can be achieved using conventional carrier gases as well. Chromatograms of a pesticides mixture (dicofol ⫹ methoxychlor) on a column coated with Carbowax 20M using helium and carbon dioxide as mobile phases are shown in Fig. 2. As seen, the replacement of helium by carbon dioxide leads to (1) elution order inversion (with CO 2 methoxychlor elutes before dicofol, and (2) improvement of the separation with CO2 as carrier gas. The nature of the carrier gas used can have a significant influence on the separation and elution order of the sorbates. Thus, the selectivity of capillary columns depends both on the SLP and the carrier gas used.
VI.
EFFECT OF CARRIER-GAS SOLUBILITY IN STATIONARY LIQUID PHASE ON RELATIVE RETENTION
The question of the influence of the processes occurring in the SLP of the relative retention and its dependence on the nature and pressure of the carrier gas at low pressures (up to 10–15 atm) has been
370
/ Berezkin
controversial for a long time. The effects of the carrier gas solubility in the SLP are usually neglected. Thus, Conder and Young wrote [18]: ‘‘for the normal carrier gases used in GLC, such as helium, nitrogen, hydrogen, argon and oxygen, the error involved in assuming insolubility at room temperature is negligible (⬍5 ⫻ 103 mm3 /mol).’’ Cruickshank et al. [39] showed that the error attributable to carrier gas solubility is small for carbon dioxide and n-octadecane SLP at 80°C. Hence, many chromatographers (see, for example, [105]), when evaluating and interpreting data of the carrier gas influence on the retention, believe that β′i ⬇ β′gi, [β′gi ⫽ (2B 12 ⫺ V ∞i )/RT] [see Eq. (15)], that is, they take into account only gas-phase processes. However, in studies we published recently [82–85], we state another point of view. To address the issue at hand it is pertinent to (1) determine the solubility of some conventional carrier gases in SLPs such as polyethylene glycol and polydimethylsiloxane and (2) estimate the role of SLP used in the carrier gas effect on the retention. The data on gas solubility in two SLPs are listed in Table 18 [106]. As is apparent from the table, the solubility of the carrier gases in the two SLPs is pronounced. The solubility order is He ⬍ N 2 ⬍ CO 2. That order is in agreement with the increase of carrier gas influence on the retention. The data obtained support the assumption of an active role of SLP on the carrier gas pressure (and nature) effects on the retention. To study the SLP influence on the relationship Rel i (Pav ) ⫽ φ(Pav ,G), the values β′ip and β′ia [see Eqs. (15) and (43)] for polar (β′ip) and nonpolar (β′ia ) SLPs were determined. When the SLP has no effect on the carrier gas, the β′ip and β′ia values should be close for the columns coated with different phases. A difference between β′ip Table 18 Solubility of Carrier Gases in SLPs at 90°C [106] Solubility, mL/L (at 760 mm Hg)
Carrier gas Helium Nitrogen Carbon dioxide
Polyethylene glycol PEG-20M
Polydimethylsiloxane PMS-100
16 32 770
71 170 890
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 371
Table 19 Dependence of β′ip (PEG-20M) and β′ia(SE-30) on the Nature of the SLP Using Nitrogen as Carrier Gas (data from [85]) Sorbate
SE-30
PEG-20M
β′ia /β′ip
Octanol-1 2,6-Dimethylphenol 2,6-Dimethylaniline Naphthalene
⫺0.029 ⫺0.029 ⫺0.029 ⫺0.030
⫺0.024 ⫺0.024 ⫺0.025 ⫺0.024
1.21 1.21 1.16 1.25
and β′ia values measured on columns coated with different SLPs indicates that at least one SLP has an influence on the carrier gas effect. A comparison of β′ip and β′ia values on two columns, 25 m ⫻ 0.16 m coated with SE-30 (β′ia) and PEG-20 (β′ip, using nitrogen at average pressure and 130°C, is shown in Table 19. As seen, the coefficients β′ip and β′ia differ noticeably from each other (16–25%). Hence, the SLP has a significant influence on the discussed phenomenon (effect carrier gas nature and pressure on retention). In Ref. 99 we studied a chromatographic system having a mixed mobile phase containing water as the active component and two SLPs of different polarity. At the experimental conditions, water is practically insoluble in one of the SLPs (silicone SE-30) but dissolves in the other SLP (PEG-20M). The relative retention on the column coated with nonpolar polydimethylsiloxane phase SE-30 is independent of whether a dry or moist helium (partial pressure of water vapor 22 mm Hg) is used as the carrier gas. On the other hand, using moist helium leads to great variations in the relative retention on the polyethylene glycol column. The observed phenomenon is related to the fact that water vapor in the carrier gas dissolves in PEG-20M, thus changing the nature of the SLP. This change leads to great variation in the column selectivity. Thus, unlike the conventional opinion, the SLP nature, through its interaction with the carrier gas, has a significant influence on the relative retention. Undoubtedly, this effect can be used advantageously in capillary chromatography.
VII. CONCLUSION New concepts on the role of the carrier gas in conventional GLC were discussed in view of recent studies. Contrary to the conventional
372
/ Berezkin
opinion, it was shown that the relative retention values are linearly dependent on the average column pressure and nature of the carrier gas. According to the theory developed here and experimental studies, the limiting value of the relative retention, at carrier gas pressure approaching zero, should be considered as a true chromatographic constant of the sorbate at routine conditions of GC. We note that relative retention, previously assumed to be a constant, in fact varies with experimental conditions. The influence of the carrier gases on the retention is growing in the following order: helium ⬍ hydrogen ⬍ nitrogen ⬍ carbon dioxide. The order is in agreement with the physical properties of the carrier gases, reflecting the degree of nonideality. It was shown experimentally that the mobile phase in GC can have a significant influence on the separation factors of and on the elution order. When one carrier gas replaces another on a given column (or using a gas mixture as mobile phase or increasing carrier gas pressure), it is possible to get a separation equivalent to application of several columns coated with various SLPs. Finally, it must be noted that the development of new concepts of the role of the carrier gas in capillary gas–liquid chromatography is not yet finished. This aspect of gas chromatography has not been adequately studied to date, and further research into the effects of the carrier gas on the retention and separation is needed.
REFERENCES 1. C. F. Poole and S. A. Schuette, Contemporary Practice of Chromatography, Elsevier, Amsterdam, 1984. 2. B. Ravindranath, Principles and Practice of Chromatography, Ellis Horwood, Chichester, UK, 1989. 3. M. Krejci, J. Payurek, and R. Komers, Vypocty a veliciny v sorpcny kolonove chromatografii (Values and Their Calculation in Column Sorption Chromatography), SNTL, Praha, 1990. 4. Handbuch der Gaschromatographie, Herausgegeben von E. Leibnitz und H. G. Struppe, Akademische Verlagsgesellschaft, Leipzig, 1984. 5. L. S. Ettre, Pure Appl. Chem., 65(4): 819 (1993).
Carrier Gas in Capillary Gas–Liquid Chromatography
/ 373
6. J. F. Parcher, Chromatographia, 47(9/10): 570 (1998). 7. S. Dal Nogare and R. S. Juvet, Gas-Liquid Chromatography, Wiley-Interscience, New York, 1962. 8. A. I. M. Keulemans, Gas Chromatography, Chapman & Hall, New York, Reinhold, London, 1956. 9. M. L. Lee, F. J. Yang, and K. D. Bartle, Open Tubular Gas Chromatography, Wiley, New York, 1984. 10. J. V. Hinshaw and L. S. Ettre, Introduction to Open-Tubular Column Gas Chromatography, Advanstar Communications, Cleveland, OH, 1994. 11. A. T. James and A. J. P. Martin, Biochem. J., 50: 679 (1952). 12. G. Guiochon and C. L. Guillemin, Quantitative Gas Chromatography, Elsevier, Amsterdam, 1988. 13. D. H. Everett, Trans. Faraday Soc., 61: 1637 (1965). 14. K. A. Gol’bert and M. S. Vigdergauz, Kurs gazovoi khromatografii (Course of Gas Chromatography), Khimiya, Moscow, 1967. 15. M. S. Vigdergauz, A. V. Garusov, V. A. Ezrets, and V. I. Semkin, Gazovaya khromatografiya s neideal’nymy eluentami (Gas Chromatography with Nonperfect Eluents), Nauka, Moscow, 1980. 16. R. J. Laub, Anal. Chem., 56: 2115 (1984). 17. V. A. Davankov, L. A. Onuchak, S. Yu. Kudryashov, and Yu. I. Arutyunov, Chromatographia, 49(7/8): 449 (1999). 18. J. R. Conder and C. L. Young. Physicochemical Measurement by Gas Chromatography, Wiley, New York, 1979. 19. R. J. Laub and R. L. Pescok. Physicochemical Applications of Gas Chromatography, Wiley, New York, 1978. 20. L. M. Blumberg, Chromatographia, 41: 15 (1995). 21. V. A. Davankov, Chromatographia, 42: 111 (1996). 22. V. M. Blumberg, Chromatographia, 42: 112 (1996). 23. L. S. Ettre and J. V. Hinshaw, Chromatographia, 43: 159 (1996). 24. V. A. Davankov, Chromatographia, 44: 279 (1997). 25. Z. Wu, Chromatographia, 44: 325 (1997). 26. V. A. Davankov, Chromatographia, 44: 329 (1997). 27. L. M. Blumberg, Chromatographia, 44: 326 (1997). 28. R. Lebron-Aguilar, J. E. Quantinilla-Lopez, and J. A. GarciaDominguez, J. Chromatogr. A, 760: 219 (1997).
374 29.
/ Berezkin
J. E. Quantinilla-Lopez, R. Lebron-Aguilar, and J. A. GarciaDominguez, J. Chromatogr. A, 767: 127 (1997). 30. S. Vezzani, G. Castello, and D. Pierani, J. Chromatogr. A, 811: 85 (1998). 31. V. G. Berezkin, Gas-Liquid-Solid Chromatography, Marcel Dekker, New York, 1991. 32. R. Lebron-Aguilar, J. E. Quantinilla-Lopez, A. M. Tello, A. Fernandez-Torrez, and J. A. Garcia-Dominguez, J. Chromatogr. A, 697: 441 (1995). 33. A. M. Tello and J. A. Garcia-Dominguez, J. Chromatogr. A, 793: 383 (1998). 34. D. H. Everett and C. T. H. Stoddart, Trans. Faraday Soc., 57: 746 (1961). 35. D. H. Desty, A. Goldup, G. R. Luckhurst, and W. T. Swanton, in M. van Swaay (Ed.), Gas Chromatography 1962, Butterworths, London, 1962, p. 67. 36. A. J. B. Cruickshank, M. L. Windsor, and C. L. Young, Trans. Faraday Soc., 62: 2341 (1966). 37. A. J. B. Cruickshank, B. W. Gainey, and C. L. Young, in C. L. A. Harbourn (Ed.), Gas Chromatography 1968, Elsevier, Amsterdam, 1969, p. 76. 38. A. J. B. Cruickshank, B. W. Gainey, and C. L. Young, Trans. Faraday Soc., 64: 337 (1968). 39. A. J. B. Cruickshank, B. W. Gainey, C. P. Hicks, T. M. Letcher, R. W. Moody, and C. L. Young, Trans. Faraday Soc., 65: 1014 (1969). 40. P. D. Schettler, M. Eikelberger, and J. C. Giddings, Anal. Chem., 39: 146 (1967). 41. J. C. Giddings, M. N. Myers, and J. W. King, J. Chromatogr. Sci., 7: 276 (1969). 42. C. Young, Chromatogr. Rev., 10: 129 (1968). 43. D. H. Everett, Trans. Faraday Soc., 61: 1637 (1965). 44. B. W. Gainey and C. L. Young, Trans. Faraday Soc., 64: 349 (1968). 45. D. H. Everett, B. W. Gainey, and C. L. Young, Trans. Faraday Soc., 64: 1667 (1968). 46. R. S. Laub, Anal. Chem., 56: 2115 (1984). 47. C. P. Hicks and C. L. Young, Trans. Faraday Soc., 64: 2675 (1968). 48. R. L. Pecsok and M. L. Windsor,. Anal. Chem., 40: 1238 (1968).
Carrier Gas in Capillary Gas–Liquid Chromatography 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.
59.
60. 61. 62. 63. 64. 65. 66. 67.
68. 69. 70.
/ 375
A. J. B. Cruickshank, M. L. Windsor, and C. L. Young, Proc. Roy. Soc., 295A: 259 (1966). S. Wicar and J. Novak, J. Chromatogr., 95: 1 (1974). S. Wicar and J. Novak, J. Chromatogr., 95: 13 (1974). B. Khalfaoui and D. M. T. Newsham, J. Chromatogr. A, 688: 117 (1994). A. Voelkel and J. Fall, J. Chromatogr. A., 721: 139 (1995). R. S. Laub and R. L. Pecsok, J. Chromatogr., 98: 511 (1974). J. R. Conder and S. H. Lauger, Anal. Chem., 39: 1461 (1967). J. O. Hirschfelder, Ch. F. Curtiss, and R. B. Bird, Molecular Theory of Gases and Liquids, Wiley, New York, 1954. R. J. Laub and R. L. Pecsok, J. Chromatogr., 98: 511 (1974). V. V. Eremin, S. I. Kargov, and N. E. Kuz’menko, Real’nye gazy (Real Gases), Moscow, Chemical Dept. of Moscow State University, 1998. M. P. Vukalovich and I. I. Novikov, Uravnenie sostoyaniya real’nykh gazov (Equation of State for Real Gases), Gosenergoizdat, Moscow-Leningrad, 1948. V. G. Berezkin, J. Chromatogr., 98: 477 (1974). H. B. Dwight, Tables of Integrals and Other Mathematical Data, Macmillan, New York, 1961. I. V. Malyukova, V. G. Berezkin, C. Cramers, and H.-G. Jansen, Neftekhimiya, 37: 266 (1997) (in Russian). V. G. Berezkin, A. A. Korolev, and I. V. Malyukova, Analusis, 25: 299 (1997). V. G. Berezkin, A. A. Korolev, and I. V. Malyukova, J. High Resol. Chromatogr., 20: 333 (1997). V. G. Berezkin, A. A. Korolev, and I. V. Malyukova, J. Microcolumn Sep., 8: 389 (1996). V. G. Berezkin, A. A. Korolev, and I. V. Malyukova, Neftekhimiya, 35: 463 (1995) (in Russian). V. G. Berezkin, E. Yu. Sorokina, I. V. Malyukova, E. N. Orlov, and E. M. Antipov, J. High Resol. Chromatogr., 21: 407 (1998). E. Matisova and P. Kuran, Chromatographia, 30:(5/6) 328 (1990). M. S. Vigdergauz and V. I. Semkin, Neftekhimiya, 9: 470 (1969) (in Russian). M. S. Vigdergauz and V. I. Semkin, Zh. Fiz. khimii, 45: 470 (1972) (in Russian).
376 71. 72. 73. 74.
75.
76.
77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88.
/ Berezkin M. S. Vigdergauz and V. I. Semkin, J. Chromatogr., 58: 95 (1971). M. S. Vigdergauz and V. I. Semkin, Zh. Fiz. khimii, 45: 931 (1971) (in Russian). M. S. Vigdergauz and V. I. Semkin, Zh. fiz. khimii, 46: 691 (1972) (in Russian). V. I. Semkin, Izuchenie neidealnosti gazoobraznih smesei uglevodorodov i ih proizvodnyh medodom gazovoi khromatografii. (The study of non-ideality of gaseous mixtures of hydrocarbons and their derivatives by gas chromatography), thesis, GIAP, Moscow, 1971. K. A. Gol’dbert and M. S. Vigdergauz, Vvedenie v gazovyu khromatografiyu (Introduction to Gas Chromatography), Khimiya, Moscow, 1991, p. 75. J. A. Rijks, Characterisation of hydrocarbons by gas chromatography; means of improving accuracy, thesis, Eindhoven, Amsterdam, 1973. L. Sojak and J. A. Rijks, J. Chromatogr., 119: 505 (1976). L. Sojak, J. Janak, and J. A. Rijks, J. Chromatogr., 138: 119 (1977). L. Sojak and M. S. Vigdergauz, J. Chromatogr., 148: 159 (1978). L. Sojak, P. Farkas, J. Janak, S. Rang, and O. Eisen, J. Chromatogr., 287: 271 (1984). V. G. Berezkin and A. A. Korolev, Zavodsk. Lab. 59(6): 15 (1993) (in Russian). V. G. Berezkin and A. A. Korolev, Doklady Akademii Nauk, 340(4): 490 (1995) (in Russian). V. G. Berezkin, A. A. Korolev, I. V. Malyukova, and R. G. Mardanov, Izvestiya AN, ser. khim., 12: 2177 (1997) (in Russian). V. G. Berezkin, A. A. Korolev and I. V. Malyukova, Izvestiya AN, ser. khim., 2: 314 (1998) (in Russian). V. G. Berezkin, A. A. Korolev, I. V. Malyukova, and Zh. Analit, Khimii, 54(2): 123 (1999) (in Russian). A. van Es, High Speed Narrow Bore Capillary Gas Chromatography, Huethig, Heidelberg, 1992. J. H. Dymond and E. B. Smith, The Virial Coefficients of Pure Gases and Mixtures, Clarendon Press, Oxford, UK, 1980. M. X. Karapet’yants, Metody sravnitel’nogo analiza (Methods
Carrier Gas in Capillary Gas–Liquid Chromatography
89. 90. 91. 92. 93. 94. 95. 96. 97.
98. 99. 100. 101. 102. 103. 104. 105. 106.
/ 377
for Comparative Calculation of Physicochemical Properties), Khimiya, Moscow, 1965 (in Russian). O. Ya. Samoilov, Structure of Aqueous Electrolyte Solutions and Hydration of Ions, Consultants Bureau, New York, 1965. A. Ben-Naim, Water and Aqueous Solution, Pergamon Press, New York, 1974. V. V. Sinyukov, Voda izvestnaya i neizvestnaya (Unknown and Known Water), Znanie, Moscow, 1987 (in Russian). G. C. Pimental and A. L. McClellan, The Hydrogen Bond, Freeman, San Francisco, 1960. A. Nonaka, Adv. Chromatogr., 12: 223 (1975). B. A. Rudenko, M. A. Baidarovtseva, and M. N. Agaeva, Zh. Analit. khimii, 30: 1191 (1976) (in Russian). V. G. Berezkin, Izvestiya AN, ser. khim., 10: 1831 (1999) (in Russian). A. V. Garusov and M. S. Vigdergauz, Uspekhi khimii, 46: 928 (1977) (in Russian). V. G. Berezkin, B. A. Rudenko, E. A. Kyazimov, M. N. Agaeva, A. A. Rodionov, and A. A. Serdan, Izvestiya AN SSSR, ser. khim., 10: 2352 (1975) (in Russian). V. G. Berezkin and A. A. Korolev, Zh. Analit. khimii, 50: 1057 (1995) (in Russian). V. G. Berezkin and A. A. Korolev, Izvestiya AN, ser. khim., 11: 2558 (1998) (in Russian). V. G. Berezkin and T. P. Popova, Zh. Fiz. khimii, 73(8): 1484 (1999) (in Russian). L. G. Berezkina, V. P. Mukhina, and V. G. Berezkin, Zh. Fiz. khimii, 70: 1888 (1996) (in Russian). Guay-Chuan Chen and E. R. Rohwer, J. Chromatogr. A., 845: 43 (1999). V. G. Berezkin, Crit. Rev. Anal. Chem., 20(5): 291 (1989). V. G. Berezkin, Neftekhimiya, 37: 366 (1997) (in Russian). R. L. Grob (Ed.), Modern Practice of Gas Chromatography, Wiley, New York, 1995. V. G. Berezkin, V. V. Egorov, V. V. Kuznetsov, and A. A. Korolev, Izvestiya AN, Ser. khim., 5: 923 (1999) (in Russian).
10 Catechins in Tea: Chemistry and Analysis
Christina S. Robb and Phyllis R. Brown Department of Chemistry, University of Rhode Island, Kingston, Rhode Island
I. II. III. IV.
INTRODUCTION THE CHEMICAL COMPOSITION OF TEA EXTRACTS CATECHINS CHEMICAL REACTIONS OF CATECHINS IN TEA A. Theaflavin Formation B. Tea Cream Formation V. HPLC ANALYSES OF CATECHINS A. Sample Preparation B. HPLC Conditions C. Detection Methods VI. CONCLUSION REFERENCES
379 383 384 385 385 386 387 387 390 392 407 407
I. INTRODUCTION Tea is one of the most commonly consumed beverages in the world. It is defined as a fragrant brew prepared from the leaves of the plant 379
380
/ Robb and Brown
Camellia sinesis. Tea originated in China and Southeast Asia and has been consumed for over 2000 years [1]. In Eastern cultures tea has always been viewed as a herbal remedy as well as refreshment. Only recently have Western cultures recognized the healthenhancing properties of tea. Tea is usually grown in areas close to the Equator, which receive an annual rainfall of 50 in. per year and have an average temperature of 30°C [2]. Tea trees are evergreen plants but lie dormant during the winter season. They are maintained as shrubs during the growing season and are harvested every 8–12 days. When harvesting occurs, the apical bud of the plant and the first two leaves are removed. The official term for this part of the plant is the tea flush [2]. The distinctive color, flavor, health benefits, and aroma of different kinds of tea result from the chemical reactions that occur during the processing of the tea flush [2]. Green, oolong, and black tea are the main types of tea produced (Fig. 1). The processes used to produce these various types of tea differ in the degree of withering and
Fig. 1 The manufacturing procedure for tea. Tea flush is the term given to the part of the plant picked for manufacture. It normally consists of the apical bud and the lower two leaves. When the flush has withered and has been macerated, it is termed dhool. Dhool easily undergoes fermentation, as the polyphenol oxidase and catechins are in close proximity. (Adapted from Ref. 1. Copyright CRC Press.)
Catechins in Tea: Chemistry and Analysis / 381 fermentation that the leaves undergo [1]. Once the tea is manufactured, it is sized, graded, and evaluated for flavor and color by tea tasters. Although tea tasters are highly trained and experienced individuals, this method of grading tea is not optimal. Ideally, the degree of fermentation or withering should be identified by the changes in chemical composition. High-performance liquid chromatography (HPLC) has become the method of choice for tea analysis, and its use has revealed much about the chemical composition of all the tea constituents. Finger et al. [3] published a review on the chromatography of tea constituents in 1992. Catechins and methyl xanthines (especially caffeine) have been studied most extensively. Catechins have attracted widespread attention for their health-enhancing properties. Although there is a range of catechin molecules in tea, the term ‘‘tea catechin’’ is commonly used to describe catechin molecules in general. Catechins possess a wide range of medicinal properties, many of which stem from their strong antioxidation capacity [4–7]. They have been identified as being a preventative in the development [8] and growth of cancer cells [9–12] and cardiovascular disease [13]. The consumption of green-tea extracts (which are rich in catechins) has been linked to increased plasma antioxidant power in humans [14], reduced blood glucose [15], and improved energy expenditure and fat oxidation [16]. Catechins also have strong antimicrobial properties [17]. The ortho-dihydroxy catechol (3′,4′-OH) arrangement on the B ring of flavan-3-ols and the hydroxypyranone group (Table 1) have been identified as the chemical structures responsible for the effective antioxidation properties of the catechins. Antioxidants prevent molecules such as DNA and protein from being attacked by free radicals, which initiate the growth of cancer [4]. Green and black teas have shown greater peroxyl-radical scavenging abilities than vegetables with good antioxidation properties such as kale and broccoli [5]. Since black tea contains low amounts of catechins compared to green tea, it was proposed that theaflavins and thearubigins (catechin-based molecules) might contribute to the antioxidant capacity of black tea [18]. Chung et al. [19] found that gallated catechins have shown the greatest peroxynitrite radical-scavenging ability in vitro. In lipid systems, individual catechins show different antioxidation capacities [6]. These differences have been attributed
382
/ Robb and Brown
Table 1 Structures of Catechin-based Molecules Flavan-3-ol Catechin
Gallocatechin
Catechin gallate
Chemical name
R1
R2
[2R,3S]-2[-3,4Dihydroxyphenyl]-3,4dihydro-1[2H]-benzopyran3,5,7-triol [2R,3R]-2-[3,4,5Trihydroxyphenyl]-3,4dihydro-1[2H]-benzopyran3,5,7-triol [2S,3R]-2-[3,4Dihydroxyphenyl]-3,4Dihydro-1-[2H]Benzopyran-3,5,7-triol3[3,4,5trihydroxybenzoate]
H
OH
OH
OH
H
1,2,3 trihydroxy benzoate
to the difference in reduction potentials (see Table 2), stabilities, and relative partition coefficients of the catechins. Catechins have also been shown to suppress angiogenesis [11]. Angiogenesis is a process of blood vessel formation around tumors that leads to the invasion of healthy cells and organs by cancer. Epigallocatechingallate (EGCG) has been found to be the most effective catechin in this process. It functions by preventing the binding of cancer to urokinase. This discovery is a major advance in cancer therapy, as other urokinase inhibitors possess either weak inhibition abilities or are highly toxic [11]. Antioxidants also aid in the prevention and treatment of cardiovascular disease. Tea flavanoids have been shown to reduce platelet aggregation that leads to thrombosis [13]. It is believed that the key mechanism leading to cardiovascular disease is the oxidation of low-
Catechins in Tea: Chemistry and Analysis / 383 Table 2 Redox Potentials of Catechins
Component Epigallocatechin (EGC) Gallocatechin (GC) Epigallocatechin gallate (EGCG) Gallocatechingallate (GCG) Epicatechin (EC) Catechin (C) Epicatechin gallate (ECG)
First redox potential versus SCE (V) 0.09 0.13 0.14 0.15 0.19 0.20 0.20
Reprinted from Ref. 2. Copyright CRC Press.
density lipoproteins (LDLs). The oxidation leads to damage of the arterial wall, producing sites for fibrous plaque formation. LDL oxidation does not occur until α-tocopherol is depleted. α-Tocopherol functions as an antioxidant in human LDL. Zhu et al. [20] showed that Longjing tea catechins could regenerate α-tocopherol in vitro, thus providing protection against oxidative attack. The different responses of individual catechins in biological systems may also be related to the bioavailability of each catechin. Pharmacokinetic studies on catechins suggest that the body adsorbs them rapidly. However, EGCG appears to behave differently from epigallocatechin (EGC) and epicatechin (EC). It has a lower bioavailability and is excreted mainly through bile, whereas EGC and EC are excreted in both urine and bile. Half-lives of catechins are in the range of 3–5 h [21]. In Japan, green tea is incorporated into chewing gums and mouthwashes to prevent caries [22]. The anticariogenicity functions on several levels. As antimicrobials, catechins attack the oral bacteria [23]. However, catechins also inhibit dental plaque formation and bacterial adherence to the tooth surface [24].
II. THE CHEMICAL COMPOSITION OF TEA EXTRACTS When hot water is added to tea, a clear extract develops. The most common method of tea preparation in Western countries is to soak a tea bag in a cup of hot water. Typically, a 3-min immersion pro-
384
/ Robb and Brown
Table 3 Composition of Green and Black Tea Component 1 2 3 4 5 6 7 8 9 10 11
Compound
Green Tea
Black Tea
Catechins Theaflavins Phenolic acids and depsides Flavanols Other polyphenols Methyl xanthines Amino acids Peptides/protein Organic acids Carbohydrates Minerals/ash
30–42% 0% 2% 2% 6% 3–6% 6% 6% 2% 11% 10–13%
3–10% 2–6% 3% 1% 23% 3–6% 6% 6% 2% 11% 10–13%
Reprinted from Ref. 1. Copyright CRC Press.
duces 0.35% of water-soluble constituents referred to as extract solids [2]. Price and Spitzer [25] measured the effect of temperature in the range 25–65°C on the concentration of ultraviolet (UV)-active soluble constituents produced from black-tea leaves. The appearance with time of soluble constituents was monitored using a peak absorbance of 271 nm; a first-order rate plot was followed. The activation energy was 41 kJ, and the number of constituents increased with increasing temperature. Table 3 shows the principal water-soluble components found in green and black teas. A variety of different constituents are present. The major constituents of both green and black teas (about 40%) are polyphenolic compounds, which affect the flavor, aroma, and color of the tea [2]. Components 1–5 in Table 3 [2] are all polyphenolicbased compounds. Although other plant species synthesize polyphenolic compounds, Camellia sinesis is unique in the range of polyphenols it contains [2]. Catechins, the major polyphenol present in green tea, are 30–42% of the total constituents.
III. CATECHINS Catechins are defined as a subgroup of the flavan-3-ol family as shown in Table 2. Flavan-3-ols contain a hydroxypyranone ring system attached to a hydroxylated phenyl ring. The hydroxylated phenyl ring is commonly referred to as the B ring. Catechins contain a
Catechins in Tea: Chemistry and Analysis / 385 3,4 EOH arrangement on the B ring; this structure is also called catechol or pyrocatechin. (⫹)-Catechin [(⫹)-C] is a specific compound which is commonly found in tea. Gallo catechins (GC) are also a subgroup of flavan-3-ol; they have a 3,4,5 EOH trihydroxy-substituted B ring. The term ‘‘gallo’’ comes from its analogy to gallic acid, 1,2,3trihydroxy benzoic acid. (⫹)-C and GC are found in tea in small quantities. Gallated catechins are catechin esters formed by gallic acid. It is believed that gallocatechingallate (GCG) and catechin gallate (CG) are not native to the tea plant, but the product of a racemization or epimerization reaction caused by the drying process in the manufacturing. The epiisomers contain the R 2 group in the syn orientation, whereas nonepiisomers contain the group in the anti orientation. Catechin composition changes significantly with the age of the leaf that is harvested. The younger the tea leaves, the higher is the catechin content [2]. The presence of catechins contributes to the good taste of tea, as illustrated by the subtle flavor of green tea compared to the more astringent black tea [26]. Price and Spitzer [25] studied the rates of extraction of certain catechins from a Japanese green tea. The ungallated catechins (EC and EGC) were extracted at a faster rate than the gallated ones within the temperature range studied (50–80°C). The smaller catechins therefore are extracted faster than the larger ones. This observation supports the hypothesis that the rate-determining step of extraction is a diffusive one (through the leaf matrix to the surface). Arrhenius plots gave activation energies in the range 31–49 kJ/mol, which are in accordance with the average value found for the soluble constituents of black tea. Catechins are extracted slower than caffeine. This is surprising, as caffeine has a greater barrier to extraction, with activation energy of 59 kJ/mol, and diffusion coefficients 100 times smaller in tea than in water [27].
IV.
CHEMICAL REACTIONS OF CATECHINS IN TEA
Catechins undergo chemical reactions, both in the processing of different kinds of tea and in the preparation of the beverage.
A. Theaflavin Formation During the fermentation part of the manufacturing procedure for black tea (outlined in Fig. 1), ungallated catechins undergo conden-
386
/ Robb and Brown
Table 4 Theaflavin Reactants and Products Theaflavin product
Catechin reactant
Theaflavin Isotheaflavin Neotheaflavin Theaflavin-3-gallate
(⫺)-Epicatechin (⫺)-Epicatechin (⫹)-Catechin (⫺)-Epicatechin
Theaflavin-3′-gallate Theaflavin-3,3′-digallate
(⫺)-Epicatechingallate (⫺)-Epicatechingallate
Gallocatechin reactant (⫺)-Epigallocatechin (⫹)-Gallocatechin (⫹)-Gallocatechin (⫺)-Epigallocatechin gallate (⫺)-Epigallocatechin (⫺)-Epigallocatechin gallate
Reprinted from reference 28 with kind permission from Kluwer Academic publishers.
sation reactions with gallated catechins [2]. This reaction is initiated by polyphenol-oxidase. The products of such reactions are called theaflavins. Table 4 outlines the nomenclature of theaflavins formed from the catechins. Theaflavins, which are characterized by a benzotropolone ring system, contribute to the overall astringency (briskness) and brightness of black tea [28]. Theaflavin formation explains why green tea is high in overall catechin content and that theaflavins are absent, whereas black tea has a higher theaflavins content as shown in Table 3. Theaflavins can undergo further condensation reactions to form thearubigins.
B.
Tea Cream Formation
Caffeine reacts with catechins and catechin-based molecules in tea extracts to form tea cream. As freshly prepared tea cools, the color of the drink darkens and a red/brown cloud develops. This change is particularly noticeable in more concentrated extracts of black tea, where a color change from dark brown to milky-red occurs. The cloudiness has been termed ‘‘tea cream’’ because it resembles tea to which cream has been added [29]. Caffeine, catechins, theaflavins, and thearubigins are known to be the major constituents of tea cream. The onset of cream formation in tea is driven by the complexation of polyphenols. Tea cream is more evident in black tea than in green tea [3]. Green tea is said to form a haze rather than cream. The mechanism of tea cream formation is still being debated in the
Catechins in Tea: Chemistry and Analysis / 387 literature. A recent study by Chao and Chiang [29] on the roles of caffeine and catechins in the cream formation of a semifermented black tea, in which thearubigins and theaflavins were not present, concluded that caffeine plays the major role in cream formation. The addition of caffeine to the tea extract was found to increase tea cream formation. The amount of caffeine added was correlated to temperature for cream formation. It was also found that the addition of catechins that lack ester linkages [(⫺)-EC, (⫹)-C, and (⫺)-EGC] had less of a tendency to form a complex with caffeine than the catechin esters [(⫺)-ECG and (⫺)-EGCG]. Since the catechin esters possess extra hydroxyl groups compared to the nonesterified catechins, extra intermolecular hydrogen bonding sites are provided for caffeine. It was proposed that through the bifunctionality they possess (a galloyl group and a hydroxy-phenyl B ring), the esterified catechins could form a more compact ‘‘claw’’ structure with caffeine than the nonesterified forms. These more compact structures are believed to be less soluble than the larger structures that the nonesterified catechins form.
V. HPLC ANALYSES OF CATECHINS Reversed-phase high-performance liquid chromatography (RPLC) is the most commonly employed technique for the analysis of catechins in tea. Initially it was used to fractionate tea liquors into their respective flavanols. In 1976 Hoeffler and Coggon [30] published the first quantitative analysis of catechins from a tea extract using a C18 HPLC column, (Fig. 2). Since 1976, RPLC has been commonly used to isolate, identify, and quantify catechin levels in tea.
A. Sample Preparation Sample preparation techniques for catechin analyses vary with both the sample and the investigation. The introduction of C18 stationary phase eliminated the need for intensive sample preparation of tea extracts. Therefore, most investigations using a C18 stationary phase only filter a tea extract before injection [31–38]. The use of internal standards such as L-tryptophan [37] and narigine in methanol [31] require the addition of the standards to the water used for extraction. Khokar et al. [39] adjusted the pH of the extract to 3.2 with citric acid before filtration, and several investigators have treated tea extracts with acetonitrile before analysis [40,41]. In addi-
388
/ Robb and Brown
Fig. 2 Chromatogram of tea flavanols in 400 µg of a crude green-tea extract. Mobile phase, acetic acid–methanol–dimethylformamide–water (1:2:40: 157); 20 µL injection; 280 nm detector a.u.f.s, flow rate 2.0 mL/min; temperature 23°C; pressure 2500–3500 psi. (Reprinted from Ref. 30 with permission from Elsevier Science.)
tion, purification of tea extracts before RPLC analysis has been performed by the use of C18 Sep-Pak cartridges [42] and countercurrent chromatography [43]. Most methods optimize for the caffeine content of teas. Lin et al. [44] removed the caffeine by reacting the extract with a calcium carbonate solution. The caffeine remained in solution, whereas the catechins precipitated out. Studies have also been performed on catechins in biological fluids such as plasma [45–47], saliva [22], and urine [48]. The preparation of biological samples for catechin analysis is more involved than for tea extracts. Gamache et al. [48] analyzed urine for EGC and EC before and after enzymatic hydrolysis. Nonhydrolyzed urine was diluted with deionized water, filtered, and centrifuged before analysis. Urine was hydrolyzed using the following procedure. An equal
Catechins in Tea: Chemistry and Analysis / 389 volume of urine and hydrolysis solution (125,000 U β-glucuronidase; 250 U sulfatase in 0.1 M sodium acetate, pH 5.0 with acetic acid) was incubated at 37°C for 2 h. The samples were then diluted with water and filtered before analysis. In the analysis of plasma, animal (rats [46] and rabbits [47]) and human plasma has been used [37,45–46]. Both catechin-spiked plasma [37,45–47] and catechin-containing plasma was used [45– 47]. In some investigations catechin-spiked plasma is used as a control and real samples are also run [45–47]. Carando et al. [45] prepared plasma samples as follows. An aliquot of plasma was placed in an ice bath and protein was precipitated by the addition of acetonitrile. The mixture was then centrifuged and the supernatant filtered before analysis. If the plasma was used as a control, a specific amount of (⫹)-catechin was added to it. If the sample was from a human volunteer, the plasma was treated with ethylenediaminetetraacetic acid (EDTA) and centrifuged before undergoing the preparation procedure. Dalluge et al. [37] spiked human plasma with EGCG. The plasma was then extracted with methylene chloride and ethylacetate according to the procedure of Lee et al. [49]. A portion of the extracted sample was lyophilized, dissolved in aqueous acetonitrile, and analyzed. Ho et al. [47] precipitated plasma proteins by adding acetonitrile-containing deoxyherigine (the internal standard) to plasma. The plasma protein then underwent solid-phase adsorption onto alumina. The residue was vortex-mixed with perchloric acid solution to release the adsorbed (⫹)-catechin. Donovan et al. [50] optimized a method for the analysis of catechin and its methylated metabolites at the levels expected to be present in plasma. Tsuchiya et al. [22] applied a specific complexing reaction of borate to isolate catechins in saliva and other biological fluids. The reaction has been previously used for the analysis of catecholamines and catechol amino acids. The reaction can be applied to catechins because of the same 3,4-hydroxyl arrangement on the B ring. To each saliva diluent, 4-methylcatechol was added in hydrochloric acid for use as an internal standard, 0.5 mL of 0.15% (w/v) diphenylborate in 5% (v/v) ethanol-potassium phosphate buffer (pH 8.5) and tetran-butylammonium bromide in 20% (v/v) n-octanol/n-hexane. The diphenyl borate, which forms a negatively charged complex with the catechins, was extracted to the organic phase by forming an ion pair with the tetra-n-butyl ammonium anionic counterion. The mixture was shaken and centrifuged. The supernatant was then treated with
390
/ Robb and Brown
7.5% (v/v) trifluoroacetic acid before analysis. The catechin complex is easily dissociated in acidic conditions, and the catechins in free form transfer to the aqueous phase of the trifluoroacetic acid solution. This preparation method selectively isolated salivary catechins. The analysis of catechins in fruits and legumes also requires more sample preparation than tea extracts. Fruit flesh [51,52], fruit seeds [53], and legumes [51] have all been analyzed. In the case of the fruit and legumes, the first step of sample preparation was to chop the flesh and freeze-dry it [51,52]. The seeds were freeze-dried whole [53]. The dried material was ground into a powder and stored below 0°C. The powder undergoes a series of extractions and hydrolyses before it is ready for analysis [51–53]. Ritsch et al. [52] performed benzoylation reactions on fruit. The benzoylation reaction is performed by dissolution of the fruit powder in pyridine containing benzoyl chloride. The mixture is left in an ultrasonic apparatus for 1 h at 60°C. Methanol is then added to the mixture. The benzoates precipitate when water is added to the reaction mixture. The mixture undergoes purification on a Sep-Pak C18 cartridge and the catechin benzoates are eluted with a mixture of isooctane, diethyl ether, and acetonitrile. Catechins as their benzoates improve the detection limit by 10- to 20-fold compared to the underivatized catechins. In addition, catechins do not degrade as quickly in the form of the benzoate.
B.
HPLC Conditions
Both isocratic [30,32,38,41–44,46,47,52] and gradient [22,33– 37,39,40,45,48,50,51,53,54] solvent systems have been used for the analyses of catechins. Mobile phases are generally based on methanol [30,32,37,38,42,44,46,48,54] or acetonitrile [22,31,33–35,39, 40, 43,45,47,50–52,55] and contain acid. Methanol is better than acetonitrile for peak shape [38] (Fig. 3). Kumamoto et al. [41] studied mixed solvents of water, acetonitrile, and ethyl acetate. Formic [42,54], acetic [30,32–35,38,40,43,46,53,54], phosphoric [38,40,43], and trifluoroacetic [21,44,36,37] acids have also been used in the mobile phases. The presence of acid is necessary not only for complete separation of the catechins but also for optimal efficiency (Fig. 4). Although catechins are extremely stable at pH ⬍ 4, stability is pH dependent in the range 4–8: the lower the pH, the greater the stability. Catechins instantly degrade at pH levels above 8 [55]. Phosphate [39,45,48–50,51] and acetate buffers [31,36,37] have also
Catechins in Tea: Chemistry and Analysis / 391
Fig. 3 The effect of organic solvent in the mobile phase of catechins, caffeine, and gallic acid. C18 Kingsorb column; detection 210 nm; flow rate 1 mL/min; (a) methanol/water mobile phase (20/80); (b) acetonitrile/water mobile phase (10/90). 1 ⫽ Gallic acid, 2 ⫽ (⫹)-GC, 3 ⫽ (⫺)-EGC, 4 ⫽ (⫹)-C, 5 ⫽ Caffeine, 6 ⫽ (⫺)-EGCG, 7 ⫽ (⫺)-EC, 8 ⫽ (⫺)-GCG, 9 ⫽ (⫺)-ECG. (Reprinted from Ref. 38 with permission from Elsevier Science.)
been added to methanol/acetonitrile solvents to attain optimal separations. Although C8 columns have been used [22,35], ocatadecyl silica (ODS) or C18 columns are the most popular. Dalluge et al. [36] performed a comparison of the performance of various stationary phases for catechin analysis. Deactivated stationary phases utilizing ultrapure silica, with either monomeric or polymeric C18 surfaces, provided optimal and reproducible separations of the catechins, due to the maximal coverage of the silica support that is achieved.
392
/ Robb and Brown
Fig. 4 The effect of acid on the separation of a standard catechin mixture using a deactivated monomeric C18 column: (A) water-acetonitrile gradient containing trifluoroacetic acid; (B) water-acetonitrile gradient without trifluoroacetic acid. 1 ⫽ EGC, 2 ⫽ (⫹)-C, 3 ⫽ Caffeine, 4 ⫽ EC, 5 ⫽ EGCG, 6 ⫽ GCG, 7 ⫽ ECG. (Reprinted from Ref. 36 with permission from Elsevier Science.)
Column temperature affected catechin separations. Usually, about 30°C is used [38,39,51,53,54]. However, ambient temperatures of about 20°C [30,31,43,45] and higher temperatures of 35°C [48] and 40°C [46,50] have also been employed to get good separations. Goto et al. [40] developed a two-step linear gradient using acetonitrile, water, and phosphoric acid to separate catechins and caffeine on an ODS column. A column temperature of 40°C was used to separate caffeine and (⫹)-catechin, which co-eluted at 30°C and at 50°C [40] (Fig. 5).
C.
Detection Methods
UV Detection The RPLC detection of catechins has been performed by several techniques. Since the compounds contain UV-absorbing chromophores,
Catechins in Tea: Chemistry and Analysis / 393
Fig. 5 The effect of column temperature on catechin standards and caffeine: (A) at 50°C; (B) at 40°C; (C) at 30°C. Develosil ODS-HG column, flow-rate 1 mL/min, diode-array detection. Two step linear gradient of water-acetonitrile-phosphoric acid. 1 ⫽ GC, 2 ⫽ EGC, 3 ⫽ (⫹)-C, 4 ⫽ Caffeine, 5 ⫽ CEG, 6 ⫽ EGCG, 7 ⫽ GCG, 8 ⫽ ECG, 9 ⫽ CG. (Reprinted from Ref. 40 with permission from Elsevier Science.)
single-wavelength UV and photodiode array detection have been widely used [22,30–36,39–41,45,50–53]. The UV absorption maxima for catechins are at 210 and 270–280 nm. The UV-visible absorption maxima for catechins and proanthocyanidins (catechin oligomers) are distinctive among plant phenolics, which generally absorb between 250 and 400 nm [53]. However a lower wavelength of 231 nm has been used [40]. Figure 6 shows a catechin separation for a green-tea extract with UV detection. Catechins can also be detected by chemical reaction detection using UV detection [54]. In the presence of concentrated sulfuric acid, 4-dimethylaminocinnamaldehyde can be employed as a selective reagent. Catechins and proanthocyandins undergo a condensation reaction with the reagent. The products of this reaction show a maximum absorbance at 640 nm, whereas the other phenols, indoles,and terpenes present in plant extracts produce reaction products with different or weak absorbances. A 200- to 400-fold sensitivity increase was found for (⫺)-EC compared to other phenolic
394
/ Robb and Brown
Fig. 6 Chromatogram of a green-tea infusion. Conditions: C18 column; acetonitrile–aqueous acetate buffer (1.0 mM acetic acid, 1.0 mM sodium acetate in water, pH 4.5). The mobile phase was programmed linearly from 12% to 21% (0–18 min), then from 21% to 65% (18–40 min) acetonitrile at a flow rate of 0.7 mL/min. UV detection with absorbance at 210 nm. Peak identities: 1 ⫽ EGC, 2 ⫽ Caffeine, 3 ⫽ EC, 4 ⫽ EGCG, 5 ⫽ ECG, 6 ⫽ internal standard (naringenin). (Adapted from Ref. 31 with permission from Elsevier Science.)
compounds. An absorbance ratio of 640/620 nm was used to calculate the concentrations. The only drawback to this approach is that the use of sulfuric acid requires all parts of the equipment to be inert. Photodiode Array Detection Several investigations have used photodiode array detection for analyzing tea liquors [21,31,33–35,40,52]. It is more accurate than a single-wavelength UV detector, as it records over all UV/visible wavelengths. The data from the spectrometer, which are recorded, can be manipulated later. Spectra from samples can be overlaid with the spectra obtained from standards and from spectral libraries. Peak identification and peak purity can be obtained. Bailey et al. [33,34] performed two investigations of the nonvolatile constituents of black-tea liquor by HPLC photodiode array detection (DAD). A
Catechins in Tea: Chemistry and Analysis / 395 range of amino acids, methyl xanthines, thearubigins, and catechins were identified. Ritsch et al. [52] and Revilla et al. [53] performed analysis of catechins in fruits. In both cases, sample purification was necessary before the samples could be analyzed. In all the studies, the isomers catechin and epicatechin were identified by their retention times, as the spectra are identical. Tsuchiya et al. [22] performed the determination of catechins in human saliva using DAD. Using a gradient method, eight catechins and an internal standard were all separated in less than 15 min without interfering peaks (Fig. 7). Catechins were simultaneously and selectively determined in the concentration range 0.05–25.0 µg/mL. In replicate spiking experiments with standards, the mean recovery ranged between 86% and 99%. In both intra- and interassay, coefficients of variation were within 2.3%. It was determined that when rinsing the mouth with an aqueous solution of green-tea extract (5.0 mg/mL) containing eight catechins, each catechin was retained at µg/mL levels in saliva for up to 60 min. Fluorescence Detection Arts and Hollman [51] investigated fluorescence detection as an alternative to UV detection for analyzing catechins in fruits and legumes. It is more specific and sensitive than UV detection and reduced the sample cleanup time needed to prepare fruits and legumes for catechin analysis, as the juices could be analyzed with minimal sample preparation. Fluorescence detection at 280 nm excitation and 310 nm emission was performed. A 10-fold increase in the detection limit sensitivity was achieved for (⫹)-C and (⫺)-EC. The peak separation and baseline were also improved compared to UV. Ho et al. [47] determined (⫹)-catechin in rabbit plasma by fluorescence detection as part of a pharmacokinetic study. Excitation and emission wavelengths were 280 and 310 nm. The effect of pH and temperature on (⫹)-catechin was also investigated. Degradation of (⫹)-catechin was rapid at pH 5, and it was determined that plasma samples underwent minimal degradation when stored at 0°C and rapid degradation at 25°C. The rabbits were given a 15-mg/kg dose of (⫹)-catechin by intravenous administration. The data fitted a twocompartment model from which the pharmacokinetic parameters were obtained. The half-life for the elimination of (⫹)-catechin was 44 ⫾ 6 min. The method was found to be stable, reproducible, accurate, precise, and with high recoveries. Based on a signal-to-noise ratio of 3, the limit of detection was determined as 1 ng. The limit
396
/ Robb and Brown
Fig. 7 (A) HPLC chromatogram of saliva before mouth rinsing with green tea. (B) HPLC chromatogram of saliva 60 min after mouth rinsing with green-tea extract (5.0 mg/mL). Conditions: CLC-C8 Shim-pack column (250 ⫻ 4.6 mm I.D., 5-µm particle size). Mobile phase: A, acetonitrile–TFA–water (10:0.3:89.7); B, acetonitrile–water (30 :70, v/v) at a flow rate of 1.0 mL/min and column temperature of 50°C. Gradient conditions were 100% A at 0 min to 25% A and 75% B at 17 min in a linear gradient elution mode. Diode array detection was used, with the eluates being monitored at 274 nm. Peak identities: 1 ⫽ GC, 2 ⫽ EGC, 3 ⫽ C, 4 ⫽ EC, 5 ⫽ EGCG, 6 ⫽ GCG, 7 ⫽ ECG, 8 ⫽ CG, 9 ⫽ internal standard 4-methyl catechol. (Adapted from Ref. 22 with permission from Elsevier Science.)
of quantitation was found to be 0.02 µg/mL. At four different concentrations the coefficients of variance were found to be less than 7%. Figure 8 shows a separation of (⫹)-C and the internal standard deoxyhegenamine in a plasma sample attained by Ho et al. [47]. Carando et al. [45] also investigated fluorescence detection for determining (⫹)-C in plasma and compared it directly to UV. The limit of fluorescence detection was found to be 5 ng/mL and the limit of quantita-
Catechins in Tea: Chemistry and Analysis / 397
Fig. 8 HPLC chromatograms of (A) blank plasma, (B) spiked plasma, and (C) plasma sample taken 30 min after intravenous administration of 15 mg/ kg of (⫹)-catechin with a calculated concentration of 1.333 µg/mL. Conditions: 12% acetonitrile in a 35 mM phosphoric acid solution (pH 2.5). Cosmosil-AR C18 column (5 µm, 150 ⫻ 4.6 mm I.D.). Column temperature was ambient and the flow rate was 1.0 mL/min. Fluorescence was monitored at an excitation wavelength of 280 nm and an emission wavelength of 310 nm. Peak identities: 1 ⫽ (⫹)-catechin, 2 ⫽ internal standard (deoxyhegenamine). (Adapted from Ref. 47 with permission from Elsevier Science.)
398
/ Robb and Brown
tion 40 ng/mL. The detection limits for UV detection were 1 and 4 µg/mL. At four different concentrations, the coefficients of variation for the fluorescence method were less than 3.9% and the mean errors ranged from ⫺0.48 to 0.43% and coefficients of variation were less than 2.15%. The mean relative errors ranged from ⫺0.75% to 0.9% for UV detection. For between-run assays, the coefficients of variation (and mean relative) error for UV and fluorescence were 6.47% (⫺4.8% to 0.35%) and 2.63% (0.33% to 2.6%), respectively. Recoveries by both methods were good, 85.31% for fluorescence and 84.47% by UV. Fluorescence detection was deemed more suitable for (⫹)-C determination in human plasma, due to its lower detection limit. Blood was drawn from human volunteers 12 h after they had consumed the same flavanoid-rich foods, including fruits, vegetables, and 300 mL of red wine. Large variations (268 to 809 ng/ml) in the amount of (⫹)-C in plasma were observed. However, it can be concluded that significant amounts of (⫹)-C are present in blood circulation after consumption of a flavanoid-rich meal. A study on the temperature stability of (⫹)-C was also performed. The same rapid degradation was observed at 25°C as seen by Ho et al. [47]. In this study, it was preferred to store the plasma samples at ⫺20°C, as this reduced degradation from 0°C. Electrochemical Detection Electrochemical detection can also be used for catechin detection [41,48,50], as specific catechins have different reduction potentials (Table 2). Donovan et al. [50] compared the use of UV, fluorescence, and coulometric electrochemical detection for the analysis of (⫹)catechin, (⫺)-epicatechin, and their methylated analogs. Hackett et al. [56] showed that 60% of the catechin excreted in urine was methylated on the catechol ring. The methylation is believed to be due to the enzyme catechol-O-methyl-transferase. The optimized HPLC separation of the catechins and the analogs is shown in Fig. 9. Taxifolin was the internal standard used. The comparison of detection methods was performed for UV detection at 280 nm, electrochemical detection at 200 and 400 mV, and fluorescence detection with excitation at 280 nm and emission at 310 nm. Using UV detection, the detection limit of (⫹)-C and (⫺)-EC were 22 and 23 ng, respectively. The detection limits of the methylated analogs ranged from 20 to 23 ng. The detection limits of electrochemical detection could not be determined at 200 mV for the methylated analogs. The detection limits for (⫹)-C and (⫺)-EC and the taxifolin were 5 ng for each.
Catechins in Tea: Chemistry and Analysis / 399
Fig. 9 HPLC chromatogram of (⫹)-catechin, (⫺)-epicatechin, and their 3′– and 4′-O-methylated analogs. Conditions: Lichrosphere C18 column, 4 ⫻ 250 mm, 5-µm particle size. The mobile phase was heated to 40°C and delivered at 0.5 mL/min and consisted of A, 50 mM ammonium dihydrogen phosphate at pH 2.60 B, 20% A in acetonitrile. A multilinear gradient began using 100% solvent A, from 5 to 35 min, solvent B was increased to 21.5%; and from 35 to 70 min, solvent B was increased to 50%. Fluorescence detection was performed with excitation at 280 nm and emission at 310 nm, 70 ng on column. Peak identities: 1 ⫽ (⫹)-catechin, 2 ⫽ (⫺)-epicatechin, 3 ⫽ 3′-O-methyl-catechin, 4 ⫽ 4′-O-methyl-catechin, 5 ⫽ 3′-O-methylepicatechin, 6 ⫽ 4′-O-methylepicatechin. (Adapted from Ref. 50 with permission from Elsevier Science.)
At 400 mV the detection limits were 4–5 ng for (⫹)-C, (⫺)-EC, and taxifolin, and they ranged from 10 to 20 ng for the analogs. At 600 mV the detection limits of all the compounds was about 5 ng, but the use of such a high potential was not believed to be selective enough for such a complex matrix as plasma. Fluorescence detection gave detection limits of 3 ng for the catechins and all the methylated
400
/ Robb and Brown
analogs except the 3′-O-methyl-epicatechin, for which it was 2 ng. Fluorescence detection was determined to be the most sensitive method of detection in this study. Kumamoto et al. [41] compared the use of UV and electrochemical detection for the analysis of catechins in green tea. Separated catechins were detected at 750 mV with electrochemical detection and at 280 nm by UV detection. An ethyl acetate, acetonitrile, water mobile phase was used in conjunction with a C18 column. Electrochemical detection was determined to be 50 times as sensitive as UV detection for the analysis of EC, EGC, ECG, and EGCG. UV detection revealed the presence of compounds other than catechins, which absorbed at 280 nm in the tea extract. Caffeine was found to interfere with EGC. Since caffeine is not electrochemically active, it did not interfere with EGC when electrochemical detection was used. In a tea extract the catechins were determined within 5% error. The addition of a few percent of ethyl acetate into a water-andacetonitrile mobile phase affected the retention times of catechins and enhanced the separation of EC and EGCG. When a small amount of acetonitrile was replaced with ethyl acetate, the capacity factors of catechins increased; and when a small amount of water was replaced with ethyl acetate, the capacity factors decreased. These effects were attributed to the formation of a pseudo-stationary phase by the ethyl acetate, which enhances the retention of catechins. Catechins are more soluble in acetonitrile than water, so the use of water in the mobile phase hinders the elution of catechins, whereas the use of acetonitrile enhances it. The coulometric array detector has also been used for catechin analysis [48]. An array of coulometric sensors was integrated with the chromatography and compounds were detected by their oxidation or reduction potentials. Since individual catechins have specific redox potentials (Table 2), electrochemical detection is effective. A flow cell capable of handling the flow rate of the HPLC contained a dual electrode system. Compounds that react with the first electrode were eliminated from the cell, and compounds that are not detected at the first electrode underwent detection at the second one. A detailed voltametric signature across the array was generated. Figure 10 shows a chromatogram of hydrolyzed urine, which contains EGC and EC. Since co-eluting compounds from the HPLC may differ in oxidation or reduction state, peak purity can be assessed. Quantitation is easily performed, and classification of analogous structures
Catechins in Tea: Chemistry and Analysis / 401
Fig. 10 Chromatogram of EGC and EC in hydrolyzed urine. This chromatogram compares the response of the same channel (⫺40 mV versus Pd) for the standard (epigallocatechin 200 ng/mL, epicatechin 60 ng/mL) and samples obtained before and after green-tea consumption and are presented at a sensitivity of 2 µm. (Reprinted from Ref. 48 with permission.)
is possible. The detection limits of the technique for EC, ECG, EGC, and EGCG were 0.6, 0.1, 2.0, and 0.3 ppb, respectively. These results were found to be in accordance with previously published data [49]. Chemiluminescence Detection Polyphenolic compounds exhibit chemiluminescence when they react with hydrogen peroxide and acetaldehyde. Chemiluminescence [46] was used to detect EGCG at picomole levels in both rat and human plasma samples. It eliminated interfering peaks due to other components in the plasma, which often overlapped, with the EGCG peak. Chemiluminescence detection improved both the sensitivity and selectivity of the method but was applied only to EGCG. The major drawback of this detection method is that postcolumn derivitization must be performed, and this is time consuming. Mass Spectrometry Detection Mass spectrometry (MS) detection has also been used for catechin analysis. Thermospray [44], electrospray [37,42], and plasmaspray [35] interfaces to the liquid chromatograph have all been employed.
402
/ Robb and Brown
The thermospray interface for MS depends on the thermal generation of a spray and a further heat treatment of the spray to yield desolvated ions. Thermospray interfaces give a wide variety of ion yields with compound type. The flow rate and temperature of the inlet tube have to be optimized for each compound class. Good sensitivity is obtained for compounds which ionize efficiently. However, in a tea extract a large number of compound classes are present (which ionize at different efficiencies); thus optimization is more difficult. The use of gradient elution requires the operating conditions of the MS to be changed as the eluent is changing. Lin et al. [44] performed thermospray LC-MS-MS in the study of catechins in a tea extract. The caffeine was removed from the tea extract before analysis. Four principal catechins were identified from their [M ⫹ H]⫹ ions: (⫺)-EC, (⫺)-EGC, (⫺)-ECG, and (⫺)-EGCG. Collision-induced dissociation spectra of the [M ⫹ H]⫹ ions gave simple fragmentation patterns which could be used to characterize the molecules. Bailey et al. [35] used a plasma spray interface to examine a tea infusion. In this case, the inlet capillary tube became the cathode in a glow discharge system. A voltage difference was applied between the capillary tube and the ion source block, and a plasma of the HPLC eluent was produced. This interface provided more fragmentation than thermospray and ionized a wider range of compounds. (⫺)-EC, (⫺)-EGC, (⫺)-ECG, (⫺)-EGCG, (⫺)-C, and (⫹)-GCG were all identified. However, the assignment of (⫹)-C and EC required the use of retention-time matching, because they have the same molecular mass. It is therefore impossible to tell by mass spectral data alone which of the diastereoisomers is present. G. K. Poon [42] used an electrospray LC-MS system to analyze a tea extract but introduced the sample directly into the electrospray MS without obtaining a LC separation. The mass spectrum of catechin and its analogs all produced strong protonated molecules in the negative ionization mode. Capillary liquid chromatography has also been interfaced to electrospray MS for catechin detection [37]. Two capillaries were used. A capillary was prepared from fused silica and packed with deactivated C18 silica. A commercially available packed capillary was also tested, which contained conventional C18 packing material. A gradient elution system was employed. The initial conditions were 85% 5 mM ammonium acetate with 0.05% TFA (A), 15% 40:60
Catechins in Tea: Chemistry and Analysis / 403
Fig. 11 Selective-ion chromatograms of a green-tea extract obtained from capillary LC-MS. The chromatogram was obtained using a 506 m O.D. ⫻ 256 m I.D. ⫻ 34-cm capillary column packed with Zorbax Eclipse monomeric C18. The selected ion chromatograms shown correspond to the [M ⫹ H]⫹ ions of (A) epicatechingallate, m/z ⫽ 307; (B) 1-epigallocatechingallate, 2-gallocatechingallate, m/z ⫽ 459; (C) 1-(⫹)-catechin, 2-epicatechin, m/z ⫽ 291; (D) 1-epicatechingallate, m/z ⫽ 443. The relative detector responses are normalized to the most abundant peak in the chromatogram. (Reprinted from Ref. 38 with permission.)
(v/v) acetonitrile: methanol with 0.05% TFA (B). The gradient was 0% to 40% B over 45 min. The selective ion-monitoring mode was used to detect catechins in green-tea infusions (Fig. 11). The nondeactivated C18 packing material performed poorly. There was irreversible adsorption of catechins and poor resolution, especially under isocratic conditions. The monomeric encapped (deactivated) capillary performed much better, and the optimized method resolved EGC, C, EC, EGCG, GCG, and ECG. The detection limit was 20 ng
404
/ Robb and Brown
of each component in the mixture, which is lower than with any previous MS method. Fourier Transform Infrared Detection Recently, on-line Fourier transform infrared (FTIR) detection for HPLC was used to determine catechins in green-tea extracts [57]. FTIR has matured as a detection technique for HPLC in recent years. Somsen et al. [58] published a review of the development of LC-IR in 1999. Unlike MS, FTIR can readily distinguish between isomers. In addition it is nondestructive; both sample recovery and purification are feasible. FTIR can also aid in the structural elucidation of unknown components. Components can be easily identified both from their spectral characteristics and from spectral library matching. Robb et al. [57] used the Bourne Scientific Infrared Chromatograph (IRC) [59] to determine catechins in a green-tea extract. This HPLC-FTIR contains a solvent-elimination interface. The interface has two main parts, a heated drift tube and a zinc selenide (ZnSe) plate (Fig. 12). The effluent from the HPLC was split before the introduction to the drift tube, with 0.08 mL/min entering the drift tube. Upon entering the drift tube the effluent was nebulized ultrasonically into a fine mist. The mist traversed the drift tube under a vacuum. The solvent surrounding the solutes was removed and the solutes left the drift tube as a fine powder. Underneath the drift tube is a stage containing a ZnSe plate. The stage moves at a rate of 2 mm/min; therefore each part of the chromatogram occupies a different position upon the plate. The infrared beam scans the plate and displays the spectra in real time on a computer screen. Figure 13a shows the 1519-cm⫺1 spectrum for the green-tea extract. The chromatogram was obtained using a C8 stationary phase, a 90:10 isocratic mobile phase of 0.1% formic acid:acetonitrile, a column temperature of 30°C, and a drift tube temperature of 130°C. A flow rate of 0.2 mL/min from the HPLC was split and 0.08 mL/min entered the IRC. The spectra were recorded at 10 scans/min. Three catechins were identified. Figures 13b, 13d, and 13e show the spectra of the components that eluted at 8, 11, and 37 min. Figure 13c shows the spectrum of EGC. A clear positive match is evident for the spectrum of the component that eluted at 8 min. The spectra of the 11- and 37-min components could not be matched to the standards that were available in-house. However, structural aspects of the catechins can be elucidated from the spectra. The 11-min compo-
Catechins in Tea: Chemistry and Analysis / 405
Fig. 12 Schematic of interface for the Bourne Infrared Chromatograph.
nent has characteristics consistent with both a gallocatechin and a gallate. It is clearly not an epiisomer, as the aromatic CH wag is a single band at 790 cm⫺1. A double band in this region characterizes epiisomers. Thus it is a (⫹)-catechin. It has a strong carbonyl at around 1700 cm⫺1 and broad bands at 1090 and 1030 cm⫺1. Gallocatechin character is evident from the broad bands in the region 1280– 1373 cm⫺1 and the small band at 735 cm⫺1. A spectrum of the compound, which eluted at 37 min, is shown in Fig. 13e. The single stretch in the aromatic CH wag region identifies the compound as (⫹)-catechin not an (⫺)-epicatechin. In addition, the broad bands in the OH-deformation region identify the compound as a gallocatechin. Therefore, both of these components can be classified as be-
406
/ Robb and Brown
Fig. 13 a) Peak chromatogram of a green-tea infusion, obtained from LCFTIR. b) FTIR spectra of 8 minute component. c) FTIR spectra of EGC. d) FTIR spectra of 11 minute component. e) FTIR spectra of 37 minute component.
Catechins in Tea: Chemistry and Analysis / 407 longing to the gallocatechin family; however, they are clearly different compounds. Differences are especially obvious in the carbonyl region near 1700 cm⫺1. Additional research is required to determine the exact structure of these two components.
VI.
CONCLUSION
Catechins are the most common kind of polyphenol found in tea. The consumption of catechins has been shown to be beneficial for many aspects of health. Since tea is widely available and cheap in most parts of the world, it appears that catechin consumption is one of the easier methods to improve health. HPLC is by far the most common method of catechin analysis, regardless of the sample matrix. Sample preparation techniques are minimal for tea extracts but more complex for fruits, legumes, and biological liquids. Typically, for the HPLC analyses, acidic mobile phases are used with C18 and C8 stationary phases and column temperature is an extremely important parameter in catechin analyses. UV is the most common method of detection. However, UV detection is not sensitive enough for analyzing catechins or their methylated analogs at the low levels at which they are found in biological liquids. Using chemical reaction detection can lower the detection limit of UV, but this is time consuming. Fluorescence and electrochemical detection are more suitable for routine analysis of biological catechin-containing samples. To identify the isomeric differences between catechins, FTIR and electrochemical detection are the methods of choice. FTIR is the only detector that can spectroscopically provide clear structural identification of components. Mass spectrometry has also been applied to catechin analysis, with several different interfaces to the HPLC. The increasing interest in improvement of health by the use of antioxidants continues to make the analysis of catechins in a variety of matrixes very important. The improvement of existing technologies and the development of new assays will indeed be welcomed by the food and dietary supplement industries as well as by the biomedical world.
REFERENCES 1. D. A. Balentine, S. A. Wiseman, and L. C. M. Bouwens, Crit. Rev. Food Sci. Nutr. 37(8): 693 (1997).
408
/ Robb and Brown
2. M. E. Harbowy and D. A. Balentine, Crit. Rev. Plant Sci., 16(5): 415 (1997). 3. A. Finger, S. Kuhr, and U. H. Engelhardt, J. Chromatogr., 624: 293 (1992). 4. A. Wiseman, D. A. Balentine, and B. Frei, Crit. Rev. Food Sci. Nutr., 37(8): 705 (1997). 5. G. Cao, E. Sofic, and R. L. Prior, J. Agric. Food. Chem., 44: 3426 (1996). 6. S.-W. Huang and E. N. Frankel, J. Agric. Food. Chem., 45: 3033 (1997). 7. P. Pietta, P. Simonetti, and P. Mauri, J. Agric. Food. Chem., 46: 4487 (1998). 8. I. E. Dreosti, M. J. Wargovich, and C. S. Yang, Crit. Rev. Food Sci. Nutr., 37(8): 761 (1997). 9. M. Rouhi, Chem. Eng. News, June 9: 11 (1997). 10. Y.-L. Lin, I.-M. Juan, Y.-L. Chen, Y.-C. Liang, and J.-K. Lin, J. Agric. Food. Chem., 45: 3033 (1997). 11. Y. Cao and R. Cao, Nature, 398: 381 (1999). 12. M. Maeda-Yamamoto, H. Kawahara, N. Tahara, K. Tsuji, Y. Hara, and M. Isemura, J. Agric. Food. Chem. 47: 2350 (1999). 13. L. B. M. Tilburg, T. Mattern, J. D. Folts, U. M. Weisgerber, and M. B. Katan, Crit. Rev. Food Sci. Nutr., 37(8): 771 (1997). 14. I. F. F. Benzie, Y. T. Szeto, J. J. Strain, and B. Tomlinson, Nutr. Cancer, 34(1): 83. 15. D. Zeyuan, T. Bingyin, L. Xiaolin, H. Jinming, and C. Yifeng, J. Agric. Food. Chem., 46(10): 3875 (1998). 16. A. G. Dulloo, C. Duret, D. Rohrer, L. Girardier, N. Mensi, M. Fathi, P. Chantre, and J. Vandermander, Am. J. Clin. Nutr., 70: 1040 (1999). 17. J. M. T. Hamilton-Miller, Antimicrob. Agents Chemother., 39(11): 2375 (1995). 18. W. J. Blot, J. K. McLaughlin, and W. H. Chow, Crit. Rev. Food Sci. Nutr., 37(8): 739 (1997). 19. H. Y. Chung, T. Yokozawa, D. Y. Soung, I. S. Kye, J. K. No, and B. S. Baek, J. Agric. Food. Chem., 46: 4484 (1998). 20. Q. Y. Zhu, Y. Huang, D. Tsang, and Z.-Y. Chen, J. Agric. Food. Chem., 47: 2020 (1999). 21. P. C. H. Hollman, L. B. M. Tilburg and C. S. Yang, Crit. Rev. Food Sci. Nutr., 37(8): 719 (1997).
Catechins in Tea: Chemistry and Analysis / 409 22. H. Tsuchiya, M. Sato, H. Kato, T. Okubo, L. R. Juneja, and M. Kim, J. Chromatogr. B, 703: 235 (1997). 23. S. Sakanaka, M. Kim, M. Taniguchi, and T. Yamamoto, Agric. Biol. Chem., 53: 2307 (1989). 24. S. Sakanaka, T. Sato, M. Kim, and T. Yamamoto, Agric. Biol. Chem., 53: 2307 (1989); Agric. Biol. Chem., 54: 2925 (1990). 25. W. E. Price and J. C. Spitzer, Food Chem., 50: 19 (1994). 26. Y. R. Liang, Z. S. Liu, Y. R. Xu, and Y. L. Hu, J. Sci. Food Agric., 53: 541 (1990). 27. M. Spiro, D. Jaganyi, and M. C. Broom, Food Chem., 45: 333 (1992). 28. A. Robertson, in K. C. Wilson and M. N. Clifford (Eds.), Tea: Cultivation to Consumption, p. 555. Kluwer Academic Publishers, London (1992). 29. Y. C. Chao and B. H. Chiang, J. Sci. Food Agric., 79: 1687 (1999). 30. A. C. Hoefler and P. Coggon, J. Chromatogr., 129: 460 (1976). 31. W. E. Bronner and G. R. Beecher, J. Chromatogr. A, 805(1/2): 137 (1998). 32. M. Ding, H. Yang, and S. Xiao, J. Chromatogr. A, 849(2): 637 (1999). 33. R. G. Bailey, I. McDowell, and H. E. Nursten, J. Sci. Food Agric., 52: 509 (1990). 34. R. G. Bailey, H. E. Nursten, and I. McDowell, J. Chromatogr., 542: 115 (1991). 35. R. G. Bailey, H. E. Nursten, and I. McDowell, J. Sci. Food Agric., 66: 203 (1994). 36. J. J. Dalluge, B. C. Nelson, J. B. Thomas, and L. C. Sander, J. Chromatogr., 793: 265 (1998). 37. J. J. Dalluge, B. C. Nelson, J. B. Thomas, M. J. Welch, and L. C. Sander, Rapid Commun. Mass Spectrom., 11: 1753 (1997). 38. H. Wang, K. Helliwell, and X. You, Food Chem., 68(1): 115 (2000). 39. S. Khokar, D. Venema, P. C. H. Hollman, M. Dekker, and W. Jongen, Cancer Lett., 114: 171 (1997). 40. T. Goto, Y. Yoshida, M. Kiso, and H. Nagashima, J. Chromatogr., 749: 295 (1996). 41. M. Kumamoto, T. Sonda, K. Takedomin, and M. Tabata, Anal. Sci., 16: 139 (2000).
410
/ Robb and Brown
42. 43. 44.
G. K. Poon, J. Chromatogr. A, 794: 63 (1998). R. Amarowicz and F. Shahidi, Food Res. Int., 29(1): 71 (1996). Y. Y. Lin, K. J. Ng, and S. Yang, J. Chromatogr., 629: 389 (1993). S. Carando, P.-L. Teissedre, and J.-C. Cabanis, J. Chromatogr. B, 707: 195 (1998). K. Nakagawa and T. Miyazawa, Anal. Biochem., 248: 41 (1997). Y. Ho, Y.-L. Lee, and K.-Y. Hsu, J. Chromatogr. B, 665: 383 (1995). Tea catechins and cancer, Application Note 70-2216, ESA Inc., Chelmsford, MA. M.-J. Lee, Z.-Y. Wang, H. Li, L. Chen, Y. Sun, S. Gobbo, D. A. Balentine, and C. S. Yang, Cancer Epidemiol. Biomark. Prevent., 4: 393 (1995). J. L. Donovan, D. L. Luthria, P. Stremple, and A. L. Waterhouse, J. Chromatogr. B, 726: 277 (1999). I. C. W. Arts and P. C. H. Hollman, J. Agric. Food Chem., 46: 5156 (1998). B. Ritsch, R. Galensa, and K. Herrmann, J. Chromatogr., 448: 291 (1988). E. Revilla, M. Bourzelx, and E. Alonso, Chromatographia, 31(9/ 10): 465 (1991). D. Treutter, J. Chromatogr., 21(233): 185 (1989). Q. Y. Zhu, A. Zhang, D. Tsang, Y. Huang, and Z-Y Chen, J. Agric. Food Chem., 45: 4624 (1997). A. M. Hackett, L. A. Griffiths, A. Broillet, and M. Wermeille, Xenobiotica, 13: 279 (1983). C. S. Robb, S. Geldart, J. A. Seelenbinder, and P. R. Brown, Masters thesis. Unpublished results. G. W. Somsen, C. Gooijer, and U. A. Th. Brinkman, J. Chromatogr. A, 856: 213 (1999). S. Bourne, Am. Lab., 30(16): 17F-J (1998).
45. 46. 47. 48. 49.
50. 51. 52. 53. 54. 55. 56. 57. 58. 59.
Index
Absolute retention: definition of, 339–342 dependence on nature and pressure of the carrier gas, 342–345 Acetonitrile, 115 Acetylcholine, 235 Adsorbed phospholipid monolayers and bilayers for biomembrane chromatographic support, 186–187 Adsorption of neutral species (in retention of metal ions), 252 Alcohols, transformation for electrochemical detection, 225–227
Amines: poorly detectable, 237–240 primary, transformation for electrochemical detection, 209–217 secondary, transformation for electrochemical detection, 217–220 Amino acids, SLM extractions for, 83 Ammonium tetramethylenedithiocarbamate (ATDC), 256 Amperometric detection, 32 Analytes: of environmental concern, extraction of triazines, 156–160
411
412
/ Index
[Analytes] membrane extraction applications for, 83–84 transformation of analytes for electrochemical detection, 203–248 detection challenges in microseparation techniques, 206–207 electrochemical detection and microseparation techniques, 207– 208 microseparation techniques, 206 NDTE, 237–240 transformation methods and summary tables, 208–237 Anilinophenylmaleimide (APM), 221 Anionic surfactants, SLM extractions for, 83 Anthracene, 235 Aromatic isocyanates, 235 Arsenic, speciation analysis of, 284–288 ASTED process, 56 Atomic spectroscopy, 255
Bioaffinity measurements, immobilized lipid systems for, 188–189 Biological fluids, MISPE protocol for sample cleanup of, 150–156 Biomembrane chromatography, 175–201 applications, 187–196
[Biomembrane chromatography] bioaffinity measurements, 188–189 drug partition measurements, 189–190 peptide- and protein-membrane interactions, 190–196 protein purifications, 187–188 biomembrane-modified silica-based chromatographic supports, 182– 187 adsorbed monolayers and biolayers, 186–187 covalently attached phospholipid monolayers, 183–186 biomembrane-modified softgel chromatographic supports, 177–182 entrapment of whole gels, 177–178 immobilization of liposomes and proteoliposomes, 178–182 future directions, 196–197 Biomolecular analyses, CEC application in, 42–43 Black tea, 380 composition of, 384 Blood serum, analysis of theophylline in, 161–162 3-Bromo-1,1-dimethyl ferrocene (BDF), 231 3,5-di-tert-Butyl-1,2benzoquinone (BBQ), 224
Index / 413 Butyl methacrylate (BMA), 319–320
Cadmium, 232 Capillary electrochromatography (CEC), 1–57, 206 applications, 42–46 basic concepts, 3–9 instrumental setup, 3–5 propulsion of mobile phase, 5–9 instrumental developments, 29–38 column technology, 29–32 detection, 32–36 gradient elution, 36–38 mobile-phase considerations, 38–42 theory, 10–29 bubble formation, 27–29 comparison to liquid chromatography, 18–19 efficiency, 13–18 extracolumn band broadening, 19–24 migration principles, 10– 13 thermal effects, 24–27 Capillary electrophoresis (CE), 2 interfacing between membrane extraction technique and, 70–71 Capillary HPLC, achievable efficiency in CEC and, 19 Carbohydrates, 235 Carboxylic acids, transformation for electrochemical detection, 229–232
Carrier gas in capillary gas– liquid chromatography, 332–377 absolute retention and its dependence on nature and pressure of the carrier gas, 339– 345 definition of absolute retention, 339–342 dependence on nature and pressure of carrier gases, 342–345 effect of carrier-gas solubility in SLP on relative retention, 369, 371 influence of nature and pressure of carrier gas on the separation, 366– 369 relative retention and its dependence on nature and pressure of the carrier gas, 345–364 correlation for calculation of retention values measured on same column using various gases, 362–364 definition of relative retention, 345–348 dependence of retention on nature and pressure of the carrier gas, 349– 353 experimental confirmation of retention dependence on average column pressure and nature of carrier gas, 353–362
414
/ Index
[Carrier gas in capillary gasliquid chromatography] steam capillary gas–liquid chromatography and influence of water vapor on relative retention, 364–366 Catechins in tea, 379–410 catechins, 384–385 chemical complications of tea extracts, 383–384 chemical reactions of catechins in tea, 385–387 tea cream formation, 386–387 theoflavin formation, 385–386 HPLC analysis of catechins, 387–407 detection methods, 392– 407 HPLC conditions, 390–392 sample preparation, 387– 390 Cation-exchange technique, 260–261 Cefoperazone, 235 Chelation (in retention of metal ions), 252 Chelation ion chromatography, 280–283 Chemical composition of tea extracts, 383–384 Chemical reactions of catechins in tea, 385–389 tea cream formation, 386– 387 theoflavin formation, 385– 386 Chemiluminescence detection
for catechin analysis, 401 Choline, 235 Chromium, 232 speciation analysis of, 295– 298 Cobalt, 232 Cocaine, 235 Column length, gradient resolution as function of particle size and, 108–114 Column technology in CEC, 29–33 Concentration limits of detection (CLOD), 208, 210 Copper, 232
Derivation, 251–255 Design of rapid gradient methods, 93–136 practice of fast chromatography, 114–133 fast analytical separations, 122–124 fast preparative separations, 124–133 symbols, 134–135 theory, 95–114 fundamentals, 95–107 gradient resolution as function of column length and particle size, 108–113 Detection of catechins, 392–407 chemiluminescence detection, 401 electrochemical detection, 398–401 fluorescence detection, 395– 398
Index / 415 [Detection of catechins] Fourier transformation infrared (FTIR) detection, 404–407 mass spectrometry detection, 401–404 photodiode array detection, 394–395 UV detection, 392–394 Detection techniques applicable for CEC, 32–36 Diethyldithiocarbamate (DEDTC), 256–257 N-(4-Dimethylaminophenyl) maleimide (DMPM), 221 Dinitrophenylhydrazine (DNPHD), 229 3,6-Dinitrophthalic anhydride (DNPT), 221 Direct-injection nebulization (DIN), 284 Drug partition measurements, immobilized lipid systems for, 189–190
Efficiency of CEC, 13–18 comparison of efficiency of capillary HPLC and, 19 Electrochemical detection: for catechin analysis, 399– 401 microseparation techniques and, 207–208 Electrodialysis, 56 Electro-driven open-tubular liquid chromatography (ED-OTLC), 14–15
Electroosmotic flow (EOF), 2 Electroosmotic mobilities obtainable in CEC, 39– 41 Electrospray ionization mass spectrometry, CEC applied with, 43 Electrospray ionization timeof-flight mass spectrometry, CEC applied with, 43 Environmental applications for membrane extraction, 83, 84 Environmental pollutants, extraction of, 156–160 Epicatechin (EC), 383, 385 Epicatechin gallate (ECG), 383 Epigallocatechin gallate (EGCG), 382, 383, 389 Ethyldimethacrylate (EDMA), imprinting protocol using, 143, 147 Ethylenediaminetetraacetic acid (EDTA), 269–270 Extracolumn band broadening contribution to CEC, 19–24
Fast analytical chromatography, practice of, 114– 133 fast analytical separations, 122–124 fast preparative separations, 124–133 Ferrocenecarboxylic acid chloride (FAC), 217
416
/ Index
Ferrocenesulfonylchloride (FSC), 217 N-(Ferrocenyl)maleimide (FM), 221 Flavan-3-ol family, 383, 384 Flow-injection analysis system (FIA), 65 Fluorescence detection for catechin analysis, 395–398 Fluorometric detection, 32 Fourier transform infrared (FTIR) detection for catechin analysis, 404–407
Gallocatechin (GC), 383, 385 Gallocatechingallate (GCG), 383, 385 Gas chromatography, interfacing between a membrane extraction technique and, 68–70 Gas–liquid chromatography (GLC), carrier gas in, 332–377 absolute retention and its dependence on nature and pressure of the carrier gas, 339–345 definition of absolute retention, 339–342 dependence on nature and pressure of carrier gases, 342–345 effect of carrier-gas solubility in SLP on relative retention, 369–371 influence of nature and pressure of carrier gas on the separation, 366–369
[Gas-liquid chromatography (GLC)] relative retention and its dependence on nature and pressure of the carrier gas, 345–364 correlation for calculation of retention values measured on same column using various gases, 362–364 definition of relative retention, 345–348 dependence of retention on nature and pressure of the carrier gas, 349– 353 experimental confirmation of retention dependence on average column pressure and nature of carrier gas, 353– 362 steam capillary gas–liquid chromatography and influence of water vapor on relative retention, 364–366 Gradient CEC, 36–38 Gradient chromatography (GC), 94 See also Design of rapid gradient methods Graphite furnace atomic absorption spectrometry (GFAAS), 294 Graphitized carbon black (GCB), 139, 140 Green tea, 380, 383 composition of, 383
Index / 417 High-affinity SPE phases, 141–142 High-performance chelation ion chromatography (HPCIC), 280–283 High-performance liquid chromatography (HPLC), 2 achievable efficiency in CEC and capillary HPLC, 19 analyses of catechins by, 387–407 detection methods, 392– 407 HPLC conditions, 390– 392 sample preparation, 387– 390 -photolysis-electrochemical detection (HPLC-hvEC), 234 See also Trace metal determination and speciation with HPLC High-speed thin-layer chromatography (HSTLC), 3 HIV protease inhibitors, immobilized lipid systems for, 190 Hydrophobic interaction chromatography (HIC), 176, 333 Hydrophobized silica, 139, 140
Idebenone, 235 Inductively-coupled plasma atomic emission spectrometry (ICP-AES), 284
Inductively-coupled plasmamass spectrometry, 255, 284 Instrumental setup for CEC, 3–5 Interfacing membrane extraction and separation, 65–71 interfacing with capillary electrophoresis, 70–71 interfacing with gas chromatography, 68–70 interfacing with liquid chromatography, 65–68 Ion exchange (in retention of metal ions), 252 Ion-exchange chromatography (IC), 176, 260–272 Ion-interaction chromatography (IIC), 272–280 Ion trap storage/reflection time-of-flight mass spectrometry, CEC applied with, 43 Ions, transformation for electrochemical detection, 232–234 IPAAm polymers, 313–314 -BMA co-polymers, 319–320 effect of hydrophobic groups on N-IPAAm co-polymers, 319–322 LCST of, 318–319 synthesis of, 315 Isocratic chromatography, resolution equation for, 98 Keto compounds, transformation for electrochemical detection, 227–229
418
/ Index
Lead, 232 speciation analysis of, 290– 291 ‘‘Light’’ carrier gases, 357 Lipophilic drugs, immobilized lipid systems for, 190 Liposomes, immobilization of, 178–182 Liquid chromatography (LC), 250 comparison to CEC, 18–19 interfacing between a membrane extraction technique and, 65–68 Liquid chromatography/diode array detection (LC/ DAD), 158 Liquid–liquid extraction (LLE), 54–55, 56, 251– 252 Lower critical solution temperature (LCST), 313 of IPAAm polymers, 318– 319
Manufacturing procedure for tea, 380 Mass limits of detection (MLOD), 208 Mass spectrometry, 255 for catechin analysis, 403– 404 Matrices, membrane extraction applications for, 84 Medicinal properties of catechins, 381–383 Membrane extraction: for sample preparation, 53– 91
[Membrane extraction] advantages of membrane extraction, 71–83 fields of application of membrane extraction, 83–84 interfacing membrane extraction and separation, 65–71 techniques of membrane extraction, 56–64 techniques of sample preparation, 54–56 with a sorbent interface (MESI), 57, 63–64 interfacing with gas chromatography, 68 Mercury, 232 speciation analysis of, 291– 293 Metal ion, SLM extractions for, 83 Metals: transformation for electrochemical detection, 232–234 See also Trace metal determination and speciation with HPLC Methacrylic acid (MAA), imprinting protocol using, 143, 147 Methanol, 115 2-(6-Methyl-2-benzoiazolylazo)-diethylaminophenol (MBTAE), 259 Micellar electrokinetic chromatography (MEKC), 2, 3 Microchip electrophoresis, 206 Microdialysis, 56
Index / 419 Microporous membrane liquid– liquid extraction (MMLLE), 57, 62–63 applications for, 84 interfacing with gas chromatography, 68, 69–70 interfacing with liquid chromatography, 66, 68 Microseparation techniques, 206 detection challenges in, 206–207 electrochemical detection and, 207–208 Migration principles for CEC, 10–13 Mobile phase: of CEC, 38–41 in gas chromatography, 338–339 Modes of HPLC for analysis of metal species, 255– 283 chelation ion chromatography, 280–283 ion chromatography, 260– 272 ion-interaction chromatography, 272–280 normal and reversed phase chromatography, 255– 260 Molecularly imprinted polymers (MIPs): association constants and site densities calculations using, 144–145, 154–155 methods for synthesis and screening of, 165–168
Molecularly imprinted solidphase extraction (MISPE) materials, 137–173 development of new MISPE protocols, 163–164 high-affinity SPE phases, 141–142 methods for synthesis and screening of large groups of MIPs, 165–168 molecularly imprinted solidphase extraction (MISPE), 142–146 multipurpose SPE phases, 139–141 polymer syntheses-related factors, 164–165 previous MISPE protocols, 146–163 off-line protocols, 150–160 on-line protocols, 160–163 with pulsed elution, 161– 162 template bleeding as unresolved issue in MISPE protocols, 168–169 Multipurpose SPE phases, 139–141
Naphthalene, 235 NDTE, 237–240 Nickel, 232 Nicotine, extraction of, 160– 161 p-Nitrophenyl 2,5-dihydroxyphenylacetate, bis-tetrahydropyranyl ether. See NDTE
420
/ Index
Nonporous membrane extraction techniques, 56 Non-steroidal anti-inflammatory drugs (NSAIDS), immobilized lipid systems for, 190 Normal-phase chromatography (NPC), 255–260
Off-line MISPE protocols, 150–160 extraction of analytes of environmental concern, 156–160 sample cleanup of biological fluids, 150–156 On-line MISPE protocols, 160–163 extraction of nicotine, 160– 161 MISPE with pulsed elution, 161–162 on-line coupled column approach, 162–163 Oolong tea, 380 Open-tubular gas chromatography (OTGC), 2 Optical detection, difference between CEC and CE in, 33–34 Organobromides, 235 Organochlorides, 235 Organoiodides, 234, 235
Particle size, gradient resolution as function of column length and, 108– 114
Penicillins, 235 Peptide(s): effect of temperature on separation of, 326–329 effects of salt concentration in the mobile phase on separation of, 329–333 -membrane interactions, immobilized lipid systems for, 190–196 transformation for electrochemical detection, 220–221, 222 Pharmaceutical analyses, CEC application in, 42 Phenanthrene, 235 Phenethylamine derivatives, immobilized lipid systems for, 190 Phenobarbital, 235 Phenylisothiocyanates (PITC), 215–216, 218 Phosphatidylcholine exchange protein, purification of, 187, 188 Phosphatidylglycerol phosphate synthetase, purification of, 187, 188 Phospholipase A 2 (PLA 2), purification of, 187, 188 Phospholid monolayers, covalently attached, for biomembrane chromatography supports, 183– 186 Phospholipids: adsorbed phospholipid monolayers and bilayers, 186–187 immobilization of, 177
Index / 421 Photodiode array detection for catechin analysis, 394– 395 Photometric (UV) detection, 32 CEC applied with, 43 o-Phthaldehyde (OPA), as reagent system for primary amines, 209– 214 Pollutants of soil and underground water, extraction of triazines as, 156–160 Poly (N-isopropylacrylamide) (PIPAAm), 313–314 Polybutadiene-maleic acid (PDMA), 261–263 Polymeric membrane extraction (PME), 57, 63 applications for, 84 interfacing with gas chromatography, 68 Polymers: polymer synthesis-related factors in MISPE protocols, 164–165 polymer synthesis and temperature-responsive solubility challenges, 314 stimuli-responsive, 312 temperature-responsive, 313–314 Poorly detectable amines, 237–240 Porous membrane extraction techniques, 55–56 Pressure-driven open-tubular liquid chromatography (PD-OTLC), 14–15
Primary amines, transformation for electrochemical detection, 209–217 Propulsion of mobile phase of CEC, 5–9 Protein-membrane interactions, immobilized lipid systems for, 190–196 Proteins: effects of salt concentration in the mobile phase on separation of, 329–333 purification immobilized lipid systems for, 187– 188 Proteoliposomes, immobilization of, 178–182 1-(2-Pyridylazo)-2-naphthol-6sulfonic acid (PNSA), 234 8-Quinolinol (QL), 234, 258– 259 Rapid gradient methods, design of, 93–136 practice of fast chromatography, 114–133 fast analytical separations, 122–124 fast preparative separations, 124–133 symbols, 134–135 theory, 95–114 fundamentals, 95–107 gradient resolution as function of column length and particle size, 108–113
422
/ Index
Redox potentials of catechins, 383 Relative retention: correlation for calculation of values measured on the same column using various carrier gases, 362– 363 definition of, 345–348 dependence on average column pressure and nature of the carrier gas, 353–362 dependence on nature and pressure of the carrier gas, 349–353 effect of carrier-gas solubility in SLP on, 369– 371 influence of water vapor on, 364–366 Resolution equation, 95–98 Restricted-access materials (RAM), 140, 141 Retention: effect of salt solutions on, 318–319 temperature effects on, 322–324 See also Absolute retention; Relative retention Reversed-phase chromatography (RPC), 255–260, 312 organic modifiers in, 115 Reversed-phase gradient chromatography (RPGC), 94–95 Reversed-phase (RP) HPLC, 176–177
Salt solution: effect on retention of, 318–319 effects in mobile phase on peptide and protein separations, 329–333 Sample handling, 250–255 derivatization, 254–255 matrix removal and analyte preconcentration, 251– 254 Sample preparation using membrane extraction, 53–91 advantages of membrane extraction, 71–83 automated and unattended operation, 80–82 cleanup and selectivity, 72–76 enrichment, 76–80 solvent consumption, 82– 83 fields of application of membrane extraction, 83–84 analytes, 83–84 matrices, 84 interfacing membrane extraction and separation, 65–71 interfacing with capillary electrophoresis, 70–71 interfacing with gas chromatography, 68–70 interfacing with liquid chromatography, 65–68 membrane extraction technique, 56–64 membrane extraction with sorbent interface (MESI), 57, 63–64
Index / 423 [Sample preparation using membrane extraction] microporous membrane liquid–liquid extraction (MMLLE), 57, 62– 63 polymeric membrane extraction (PME), 57, 63 supported liquid membrane extraction (SLM), 57–62 sample preparation techniques, 54–56 Secondary amines, transformation for electrochemical detection, 217–220 Selenium, speciation analysis of, 288–290 Silica-based chromatographic supports, biomembranemodified, 182–187 adsorbed monolayers and biolayers, 186–187 covalently attached phospholipid monolayers, 183– 186 Soft-gel chromatographic supports, biomembrane-modified, 177– 182 entrapment of whole gels, 177–178 immobilization of liposomes and proteoliposomes, 178–182 Solid-phase extraction (SPE), 54, 55, 56, 139 common sorbent materials used in, 139, 140
[Solid-phase extraction (SPE)] for enrichment and cleanup of complex samples, 138, 139 mechanisms for retention of metal ions, 252 methods of, 251 Solid-phase microextraction (SPME), 55 Solvent gradient, replaced with temperature gradient on HPLC columns, 324–325 Sorbate, 338 retention characteristics of, 339–340 Stationary liquid phase (SLP), 338 effect of carrier-gas solubility in SLP on relative retention, 369–371 Steam capillary GLC, influence of water vapor on relative retention and, 364–366 Steroids, temperature-responsive chromatography of, 314–318 Stimuli-responsive polymers, 312 Styrene-divinyl benzene copolymers (PS-DVB), 139, 140 Supercritical fluid chromatography (SFC), 2 Supported liquid membrane (SLM) extraction, 57–62 applications for, 83 interfacing with liquid chromatography, 65, 67
424
/ Index
Tea, catechins in, 379–410 catechins, 384–385 chemical complications of tea extracts, 383–384 chemical reactions of catechins in tea, 385–387 tea cream formation, 386–387 theoflavin formation, 385–386 HPLC analysis of catechins, 387–407 detection methods, 392– 407 HPLC conditions, 390–392 Tea cream, formation from catechins, 386–387 Temperature-responsive chromatography (TRC), 311–336 application to separation of peptides and proteins, 326–333 effect of temperature on peptide separation, 326–329 effects of salt concentration in mobile phase on peptide and protein separations, 329–333 temperature effects on retention, 323–324 temperature gradient, 324– 325 temperature-responsive chromatography, 314–322 effect of hydrophobic groups of N-IPAAm copolymers modified on silica, 319–322
[Temperature-responsive chromatography (TRC)] effects of salt addition on retention, 318–319 TRC of steroids, 314–318 temperature-responsive stationary phases, 314 Template bleeding (unresolved issue in MISPE protocols), 168–169 Tetraethyllead (TTEL), 290– 291 Tetrahydrofuran (THF), 115 Tetramethyllead (TTML), 290–291 Theaflavin formation from catechins, 385–386 Theophylline in blood serum, analysis of, 161–162 Thermal effects of CEC, 24–27 Thiols, transformation for electrochemical detection, 221–225 Tin, speciation analysis of, 293–295 Toluene diisocyanate, 235 Trace metal determination and speciation with HPLC, 249–310 chromatographic modes, 255–283 chelation ion chromatography, 280–283 ion chromatography, 260– 272 ion-interaction chromatography, 272–280 normal and reversed phase chromatography, 255–260
Index / 425 [Trace metal determination and speciation with HPLC] metal speciation, 283–298 arsenic, 284–288 chromium, 295–298 lead, 290–291 mercury, 291–293 selenium, 288–290 tin, 293–295 sample handling, 250–255 derivation, 254–255 matrix removal and analyte preconcentration, 251–254 Transformation of analytes for electrochemical detection, 203–248 detection challenges in microseparation techniques, 206–207 electrochemical detection and microseparation techniques, 207–208 microseparation techniques, 206 NDTE, 237–240 transformation methods and summary tables, 208– 237 alcohols, 225–227 carboxylic acids, 229–232
[Transformation of analytes for electrochemical detection] keto compounds, 227–229 metals and ions, 232–234 miscellaneous analytes, 234–237 peptides, 220–221, 222 primary amines, 209–217 secondary amines, 217– 220 thiols, 221–225 Triazines as pollutants of soil and ground water, 156– 160 Tricyanoethoxypropane (TCEP), 355 Trinitrobenzene sulfonic acids (TNBSA), 216 Tri-octyl phospine oxide (TOPO), 72, 73 UV detection, for catechin analysis, 393–394 Water vapor, influence on relative retention, 364–366 Whole cell entrapment for biomembrane chromatographic support, 177– 178