Chromatography: Concepts and Contrasts

CHROMATOGRAPHY Concepts and Contrasts SECOND EDITION James M. Miller Drew University Madison, New Jersey @KEiC*ENCE A ...

Author: Miller J.M.

480 downloads 4287 Views 43MB 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

DOWNLOAD PDF

CHROMATOGRAPHY Concepts and Contrasts SECOND EDITION

James M. Miller Drew University Madison, New Jersey

@KEiC*ENCE A JOHN WILEY & SONS, INC., PUBLICATION

This Page Intentionally Left Blank

CHROMATOGRAPHY

This Page Intentionally Left Blank

CHROMATOGRAPHY Concepts and Contrasts SECOND EDITION

James M. Miller Drew University Madison, New Jersey

@KEiC*ENCE A JOHN WILEY & SONS, INC., PUBLICATION

Copyright 0 2005 by John Wiley & Sons, Inc. All rights reserved Published by John Wiley & Sons, Inc., Hoboken, New Jersey. Published simultaneously in Canada. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400, fax 978-646-8600, or on the web at w.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

For general information on our other products and services please contact our Customer Care Department within the U.S. at 877-762-2974, outside the U.S. at 317-572-3993 or fax 317-572-4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print, however, may not be available in electronic format. Library of Congress Cataloging-in-PublicationData is available. Miller, James M., 1933Chromatography: concepts and contrasts / James M. Miller.-- 2nd ed. p. cm. Includes bibliographical reference and index. ISBN 0-471-47207-7 1. Chromatographic analysis. Title. QD79.C4M55 2005 543'.8--dc22 2004003024 Printed in the United States of America 109 8 7 6 5 4 3 2 1

CONTENTS Preface to Second Edition

xiii

Acknowledgments

xv

Preface to the First Edition Symbols, Abbreviations, and Acronyms 1

Impact of Industrial and Government Regulatory Practices on Analytical Chromatography

xvii xix

1

1.1 Locus of Chromatography in Chemical Industry / 3 1.2 Governmental Organizations / 4 National Institute of Standards and Technology (NIST) / 5 Food and Drug Administration (FDA) / 6 Environmental Protection Agency (EPA) / 11 Other organizations / 12 1.3 Nongovernmental Agencies / 12 Association of Analytical Communities International (AOAC) / 14 ASTM International / 14 International Organization for Standardization (ISO) / 16 International Union of Pure and Applied Chemistry (IUPAC) / 17 1.4 Standards, Calibration, and NIST / 17 1.5 USP and Other Pharmacopoeias / 19 International Conference on Harmonization Guidlines / 21 1.6 International Guidelines for Analytical Laboratories / 22 Sources of Guidelines / 23 Method of Development, Validation, and Transfer / 24 V

Vi

CONTENTS

Instrument Qualification / 29 21 CFR Part 11: Electronic Records and Electronic Signatures / 30 1.7 Final Comments / 30 References / 31 2

Introduction to Chromatography

35

2.1 Brief History / 37 2.2 Definitions and Classsifications / 39 Separation / 39 Chromatography / 41 Chromatographic Symbols / 45 Normal Distribution / 51 Other Terms / 53 2.3 Summary / 64 References / 64 3

Band Broadening and Kinetics

67

3.1 Configurations of the Stationary Phase / 67 Characteristics of Column Packings / 68 Column Volume Characteristics / 73 3.2 Rate Theory / 75 Original Van Deemter Theory / 75 Other Rate Equations / 79 Van Deemter Plots / 83 Practical Consequences of the Rate Theory / 84 Reduced Rate Equation for HPLC (and GC) / 88 Redefinition of H / 90 References / 91 4

Physical Forces and Interactions

4.1 Intermolecular and Interionic Forces / 93 Sorption Isotherms / 94 Types of Forces / 96 4.2 Size Exclusion-Molecular Sieving / 100 4.3 Some Models / 102 Hildebrand’s Solubility Parameter / 104 Snyder’s Solvent Parameter / 104 References / 107

93

CONTENTS

5 Optimization and the Achievement of Separation

5.1 5.2 5.2 5.4

vii

109

Kinetics and Zone Broadening / 109 Thermodynamics and Zone Migration / 111 Achievement of Separation / 111 Optimization of Separations / 112 References / 115

6 Comparisons Between Chromatographic Modes

117

6.1 Gas Chromatography Compared to Liquid Chromatography / 117 Gases Compared to Liquids / 118 Gas Compressibility in GC / 119 Permeability / 121 Efficiency and Speed / 122 6.2 Supercritical Fluids and Supercritical Fluid Chromatography / 124 Mobile-Phase Properties / 127 Stationary-Phase Properties / 130 Instrumentation / 130 Density Programming / 130 Applications / 131 Summary / 134 6.3 Reduced Parameters / 134 6.4 Columnar and Planar Configurations / 135 Peaks Compared to Bands / 135 Retardation Factors / 136 References / 139 7 Gas Chromatography

Early History, Theories, and Classifications / 141 Kovats Index / 144 Rohrschneider/McReynolds Constants / 146 Methods for Selecting Stationary Phases / 147 7.2 Instrumentation for Capillary GC / 148 Mobile Phase / 148 Injection Ports and Valves / 150 Open Tubular Columns / 153 Detectors / 157 7.3 Instrumentation for Packed-Column G C / 165 Packed-Column Injection Ports / 165 Packed Columns and Stationary Phases / 166 7.1

141

Viii

CONTENTS

7.4 Stationary Phase / 168 Typical Liquids / 169 7.5 Temperature Effects / 170 Programmed Temperature GC / 172 7.6 Special Topics / 177 Fast G C / 177 Programmed Temperature Vaporizers and Large-Volume Injections / 177 Gas Chromatography Analysis of Nonvolatiles / 178 Inorganic G C / 178 Simulated Distillation / 179 7.7 Summary and Evaluation / 179 References / 180 8 Liquid Chromatography in Columns

8.1 Introductory Classifications / 184 Phase Polarities: Normal Versus Reversed / 184 Liquid-Solid (LSC) Versus Liquid-Liquid (LLC) Chromatography / 185 Low-Pressure Versus High-pressure Liquid Chromatography / 186 Isocratic Versus Gradient Elution / 188 Stationary Phase / 190 Mobile Phase / 191 8.2 Classification of HPLC Modes / 194 Liquid-Solid Chromatography / 194 Bonded-Phase Chromatography / 198 Ion Exchange Chromatography / 213 Size Exclusion Chromatography / 226 8.3 Instrumentation / 234 Pumping Systems and the Mobile Phase / 234 Sampling / 238 Columns / 240 Detectors / 244 8.4 Reversed-Phase Method Development and Optimization / 249 Overlapping Resolution Mapping / 249 Phase Selection Diagrams / 254 Buffer Effects / 256 Gradient Elution / 256 Computer Methods / 258 Troubleshooting and Tips / 259

183

CONTENTS

iX

8.5 RP-HPLC Alternatives for the Pharmaceutical Industry / 261 8.6 Preparative LC / 262 Sample Cleanup / 263 Low-Pressure Preparative LC / 263 High-pressure Preparative LC / 26.5 8.7 Special Topics / 266 Performance Qualification Kit / 266 Pseudophase or Micellar LC / 267 Other Topics / 268 8.8 Summary and Evaluation / 268 References / 270 9 Quantitation: Detectors and Methods

277

9.1 Detectors / 278 Classification of Detectors / 278 Detector Characteristics / 285 Summary of Detector Terms / 295 9.2 Data Acquisition and Processing / 296 Peak Height versus Peak Area / 296 Measurement of Areas / 297 Sources of Error / 299 9.3 Quantitative Analysis / 299 Standards and Calibration / 299 Classification of Methods / 300 Concluding Comments / 305 References / 306 10 Chromatography with Mass Spectral Detection (GC/MS and LC/MS)

10.1 Basics of Mass Spectrometry / 310 Ion Sources / 311 Analyzers / 318 Other MS Topics / 322 10.2 Gas Chromatography/Mass Spectrometry / 323 10.3 Liquid Chromatography/Mass Spectrometry / 324 10.4 Other Hyphenated Methods / 326 Liquid Chromatography/Nuclear Magnetic Resonance / 327 10.5 Summary / 328 References / 328

309

X

CONTENTS

11

Liquid Chromatography on Plane Surfaces

11.1 Paper Chromatography / 331 11.2 Thin-Layer Chromatography / 333 Manual TLC / 333 Instrumental TLC / 341 11.3 Other Topics / 347 Preparative TLC / 347 Multidimensional TLC / 347 11.4 Literature Summary and Applications / 348 References / 350 12 Qualitative Analysis

12.1 Retention Parameters / 353 Relative Retention Parameters / 355 Two-Column Plots / 355 12.2 Other Methods of Qualitative Analysis / 357 Chemical Methods / 358 Instrumental Methods / 358 References / 362 13 Capillary Electrophoresis and Capillary Electrochromatography

13.1 Principles of Electrophoresis / 366 Fundamental Equations / 366 Weak Acids and Bases / 367 Complex Formation / 369 Electrolyte Concentration / 369 Electroosmosis / 370 Diffusion / 370 Adsorption / 371 13.2 Zone Electrophoresis / 371 13.3 Capillary Electrophoresis / 372 Capillary Zone Electrophoresis / 376 Micellar Electrokinetic Chromatography / 380 13.4 Capillary Electrochromatography / 380 References / 383 14

Sample Preparation

14.1 Extraction / 390 Simple Liquid-Liquid Extraction / 391 Liquid-Solid Extraction / 402

331

353

365

387

CONTENTS

Xi

Assisted Liquid-Solid Extraction / 402 Supercritical Fluid Extraction / 404 Solid-Phase Extraction / 405 Solid-Phase Microextraction / 407 Stir-Bar Sorptive Extraction / 410 Other Extraction Methods / 410 Headspace Methods / 410 14.2 Dialysis / 411 14.3 Derivatization / 412 Methods / 413 Examples / 415 References / 417 15 Special Applications

423

15.1 Multidimensional Chromatography / 423 Gas Chromatography/Gas Chromatography / 425 Liquid Chromatography/Gas Chromatography / 427 Liquid Chromatography/Liquid Chromatography / 431 Other Combinations / 432 15.2 Biological Applications / 433 Affinity Chromatography / 434 Proteins, Peptides, and Proteomics / 438 Other Large Biomolecules / 440 Hydrophobic Interaction Chromatography / 442 15.3 Chiral Separations / 443 Stereochemical Nomenclature and Conventions / 443 Disastereoismeric Separations / 445 Chiral Stationary Phases / 446 Chiral Mobile Phases / 453 Additional Chiral Methods / 453 15.4 Other Topics / 454 Other Chromatographic Methods / 454 Field-Flow Fractionation / 455 Microfludic Devices for Chromatography and Electrophoresis (Lab-on-a-Chip) / 455 References / 456

16 Selection of a Method

16.1 Methods of Attack / 463

463

Xii

CONTENTS

16.2 The Internet / 464 Literature Searching / 464 Other Sources of Information on the Internet / 465 16.3 Experimental Approach / 466 Volatile Samples / 468 Nonvolatile Samples / 468 16.4 Summary / 471 References / 471 Some Internet Web Sites of Interest to Chromatographers / 472 Appendix A ICH Glossary

475

Appendix B

479

Index

485

PREFACE TO THE SECOND EDITION

This new, second edition has the same rationale, organization, objectives, and academic level as the first edition. It is based on the premise that chromatography (the concepts) should be described and taught in a unified manner that emphasizes the similarities and differences (the contrasts). The academic level remains at the third year undergraduate, and the subject of chromatography is built on fundamentals well-known to third-year students, topics like liquid-liquid extraction, ion exchange processes, and equilibrium theory. The text attempts to show the connections between these principles, already familiar to the students, and chromatography. Even though chromatography has become a mature science, the material from the first edition was badly out of date. Many new developments made in the intervening 15 years needed to be added. The process of updating has also resulted in an expansion, in part by inclusion of more information about topics like multidimensional chromatography, chiral separations, mass spectrometric detection, and biological applications. Liquid chromatography (LC) in columns is still the largest topic with coverage of normal and reversed phase systems, bonded phases and monolithic stationary phases, ion exchange, size exclusion, and affinity chromatography. The emergence of capillary electrochromatography (CEC) from the combination of HPLC and capillary zone electrophoresis has necessitated a short introduction to electrophoresis. Further broadening of the coverage has resulted from a new chapter on sample preparation which includes many forms of extraction including solid phase extraction (SPE) and solid phase microextraction (SPME), among others. The overall result is a book that covers most of the separation methods used in the analytical laboratory today, so it could be useful in courses covering the wider field of separation science, not just chromatography. Another addition is a unique new chapter on industrial and governmental regulatory practices. It introduces students to agencies like FDA, EPA, and NIST as well as the USP, ASTM, AOAC, ISO, ICH, and IUPAC. The xiii

XiV

PREFACE TO THE SECOND EDITION

information should be helpful in building a bridge between academe and the industrial/governmental workplace. Subjects defined and discussed include method development, method validation, instrument qualification, standard operating procedures (SOPS), and system suitability. This chapter has been placed first in the text, but for many, it should not be read first, at least not all of it. It has too many abbreviations and new topics. Rather it should serve as a reference chapter, being consulted many times in the study of chromatography, and in the end, it may serve as a good summary of current chromatographic practice. This book contains many acronyms and abbreviations that will tend to overwhelm students as they start to use the book. Unfortunately, that is the real-world experience, and rather than refrain from using them, I have attempted to minimize confusion by providing a complete listing beginning on page xix, accompanied by a comprehensive index. Also, these symbols and abbreviations are repeated often throughout the text for those who choose to read chapters out of the sequential order in which they are presented. Complete coverage of chromatography is attempted, but space did not permit some topics to be described in depth; depth is provided in the form of references which can lead the reader to more information. The book includes nearly a thousand references and many Internet Web sites. By taking advantage of these additional sources of information, upperlevcl courses should find the text suitable for their use. It can serve as a framework and outline which is readily expanded and supplemented. In the following Acknowledgments, I have named some colleagues to whom I am especially grateful, but I would be remiss if I did not salute all of those who have taught, mentored, and supported me during my career. Thanks to all of you.

JAMESM. MILLER Madison, New Jersey September 2004

ACKNOWLEDGMENTS There are many persons to whom I am indebted for their help in producing this publication. I can’t name all of them, but I want to single out a few. The staff at Wiley has been very cooperative and helpful, especially Heather Bergman. My special thanks go to a select few colleagues who reviewed this edition in every stage of development and made significant contributions to its content and accuracy. I am especially proud that three of them are former students and.alumni of Drew University. All are listed here and have my enthusiastic appreciation. They deserve the credit for the quality; any errors there may be in this monograph are mine, not theirs. Judy P. Boehlert Boehlert Associates, LLC Park Ridge, New Jersey Member of USP’s Council of Experts, 2000-2005 Jonathan Crowther Ortho-Clinical Diagnostics Raritan, New Jersey John G. Hoogerheide Pfizer Analytical R & D, Pfizer, Inc. Kalamazoo, Michigan David Locke Department of Chemistry Queens College, CUNY Flushing, New York Ron Majors Agilent Technologies Wilmington, Delaware Harold McNair Department of Chemistry Virginia Tech Blacksburg, Virginia

xvi

ACKNOWLEDGMENTS

Lee Polite Axion Analytical Laboratories Chicago, Illinois Vincent T. Remcho Department of Chemistry Oregon State University Corvallis, Oregon Joseph Sherma Department of Chemistry Lafayette College Easton, Pennsylvania

J.M.M

PREFACE TO THE FIRST EDITION

Chromatography has become the premier technique for separations and analyses. The three most important types of chromatography are gas chromatography (GC), liquid chromatography in columns (LC), and thin-layer chromatography (TLC). Because of the large amount of information available about each of them, these individual techniques are often treated separately in monographs. However, they do share a common theoretical base, and the most efficient education of a novice would be a unified study. Furthermore, a scientist working with one of the techniques may need to switch to another one, and that could be more easily accomplished if he/she were acquainted with the theories common to the two techniques. This monograph attempts a unified approach to chromatography and emphasizes the similarities and differences between the major divisions-GC, LC, and TLC. Thus the title is Chromatography: Concepts and Contrasts. In addition to the advantages mentioned above, the unified approach permits the use of one set of terms and symbols that should make learning easier. Another consequence of this approach is that the chapters covering the three main topics (GC, LC, TLC) do not stand alone; rather, they build on the introductory chapters, which cover the common principles. Hopefully, the book is short enough that the novice can afford the time to begin at the beginning and read through to the chapter that covers the information on the topic he/she is seeking. The unified approach was taken in my earlier book, Separation Methods in Chemical Analysis, and some of the material in that book has been repeated. This monograph is more narrow in coverage, of course, and has been updated with the latest information on chromatography. It is an elementary introduction to the topic, but one that is quite comprehensive. In most cases, references are provided for further information, and this book should serve as a good reference text. It includes many practical operating hints. These are presented as evidence of the applicability of the theory and not as hints without rationales. Modern LC is the newest of the three techniques and the xvii

XViii

PREFACE TO THE FIRST EDITION

one requiring the most description, so Chapter 9 on LC is the longest one. Chapter 11 describes some special techniques, including chiral separations and derivatization. The final chapter, Chapter 12, provides some suggestions for selecting a chromatographic method. Although the book is introductory and elementary, some background in chemistry will be needed in order to understand the chemical systems described. The better the reader’s background in the theory of chemistry, the more meaningful will be the sections that deal with chromatographic theory. A person with a minimal background will probably want to skip lightly over some of the theory, while a chemist may want to delve into the theory more deeply by reading some of the references provided. Only the most basic and necessary information is presented here in order to keep this book as simple as possible. The book is intended to appeal to a wide audience-academic and industrial scientists, students in both formal coursework and informal private study, and chemists and nonchemists. The academic use is best suited for the undergraduate level, but with additional assignments from the references, it could probably be used at the lower graduate level. In industry, it could be used by scientists entering any of the fields of chromatography, or it could serve as an updating for those in the field. I want to express my appreciation to a large number of people, beginning with the many scientists on whose work I have drawn in the preparation of this monograph. Many of them have taken time to communicate with me privately and I appreciate their kind efforts. Several colleagues have read parts of the manuscript and offered useful suggestions, and I am deeply indebted to them. Thanks also to my students, at Drew and elsewhere, who suffered with early versions of the manuscript and offered helpful criticisms. Special thanks for sabbatical support go to Drew University and Sandoz, Inc.

JAMES M. MILLER Madison, New Jersey November 1987

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

A A, As AC ACN ACS A D ALSSA ANDA AND1 AOAC APHA API ASEGO ASTM AWWA BP, BPC C CBER

ccc CD CDER CDC CDS CE CEC

Area (Peak area) Cross-sectional area (of a column or tube) Surface area of an adsorbent Affinity chromatography; sometimes also HPAC Acetonitrile American Chemical Society Analog-to-digital Analytical and Life Science Systems Association (formerly Analytical Instrument Association) Abbreviated new drug application Analytical Data Exchange Association of Analytical Communities International American Public Health Association Atmospheric pressure ionization (MS) Accelerated solvent extraction ASTM International, formerly American Society for Testing & Materials American Water Works Association Bonded phase; bonded phase chromatography Concentration, usually molar, M Center for Biologics Evaluation and Research (Division of FDA) Countercurrent chromatography Cyclodextrin Center for Drug Evaluation and Research (Division of FDA) Centers for Disease Control and Prevention Chromatography data system Capillary electrophoresis Capillary electrochromatography xix

XX

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

CFR CGMP: cGMP CI CITAC CRM CRV CSP CSTL CVM CZE D d

4

DAD df dP DL DQ E P Ea ECD EI EKC ELSD EOF EPA EPC ESI F

f

FDA FFF

Code of Federal Regulations Current good manufacturing practice (GMP is used in this book for CGMP and GMP) Chemical ionization (MS) Cooperation on International Traceability in Analytical Chemistry Certified reference material Critical value Chiral stationary phase Chemical Science and Technology Laboratory (Division of NIST) Center for Veterinary Medicine (Division of FDA) Capillary zone electrophoresis Diffusion coefficient; for example D,, diff. coeff. for a solute in a liquid or minimum detectability Distance between two peak maxima (in resolution equation) Diameter of a column (inside) Diode array detector Film thickness Diameter of a particle (of stationary phase or solid support) Detection limit Design qualification Peak height or electric field strength (voltage) Free energy at constant V and T Surface energy of an adsorbent (see Snyder solvent parameter) Electron capture detector or electrochemical detector Electron impact ionization (MS) Electrokinetic chromatography Evaporative light scattering detector Electroosmotic flow Environmental Protection Agency Electronic pressure control (GC) Electrospray ionization (MS) Flow rate: F, at outlet; F, corrected outlet flow; F, average corrected outlet flow Detector response factor Food and Drug Administration Field flow fractionation

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

FFPC FID FPD FR FTIR GAMP GC GGP GLC GLP GMP GPO GSC GXP ,G 27 H HIC HILIC HPLC HPTLC HS h I I IC ICH i.d. IEC IIC IPC

IQ IR IS IS0 ISRP IUPAC j

K, Ke,

xxi

Forced flow planar chromatography (TLC) Flame ionization detector Flame photometric detector Federal Register Fourier transform infrared spectroscopy Good automated manufacturing practice Gas chromatography Good guidance practices (of the FDA) Gas-liquid chromatography Good laboratory practice Good manufacturing practice Government printing office Gas-solid chromatography Any or all of the good practices Gibbs free energy at constant P and T Enthalpy Plate height or peak dispersivity; formerly HETP Hydrophobic interaction chromatography Hydrophilic interaction chromatography High performance liquid chromatography High performance thin layer chromatography Headspace; headspace sampling Reduced plate height Retention index of Kovats; also sometimes A1 Ionic strength Ion chromatography International Conference on Harmonization Inside diameter (of a column or tube) Ion exchange chromatography Ion interaction chromatography Ion pair chromatography; often called IIC, ion interaction chromatography Installation qualification Infrared spectroscopy Internal standard International Organization for Standardization Internal surface reversed phase supports International Union of Pure and Applied Chemistry Pressure correction factor (in GC) Acid ionization constant Constant; equilibrium constant

xxii

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

Kc KP k L 1 LC LIMS LLC LLE LLME LOD LOQ LSC LSE LVI MAE MDQ MDV MEKC MIP MN MP MRA MS MSD MW MWCO N N Neff Niim

NDA NF NIOSH NIST NMR NPD NPLC

Distribution constant Partition constant Retention factor Length, for example, column length Distance from one end of a capillary to the detector window Liquid chromatography; sometimes used to refer to low pressure LC Laboratory information management system Liquid-liquid chromatography Liquid-liquid extraction Liquid-liquid microextraction; also called LPME, SDME, and SME Limit of detection Limit of quantitation Liquid-solid chromatography Liquid-solid extraction Large volume injection (GC) Microwave assisted extraction Minimum detectable quantity Minimum detectable value Micellar electrokinetic chromatography Molecular imprint polymers Number average molecular weight Mobile phase Mutual recognition arrangement Mass spectrometry Mass selective detector (MS) Molecular weight (mass) Molecular weight cut off Plate number Noise (detector) Effective plate number Plate number limit New drug application National Formulary National Institute of Occupational Safety and Health National Institute of Standards and Technology Nuclear magnetic resonance (spectrometry) Nitrogen-phosphorus detector Normal phase liquid chromatography

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

NTIS 0.d. ODS OECD

00s

OPLC OQ ORD ORM OSHA OT

ov

P P' P

P O

PAGE PAH PC PDMS PF PFE PDA PH PK PLOT PPb PPm PQ psi PTGC PTV

PW

12

4

QA QC QL YZ

R

xxiii

National Technical Information Service Outside diameter (of column or tube) Octadecylsilyl Organization for Economic Cooperation and Development Out of specification Overpressurized layer chromatography (TLC) Operational qualification Office of research and development Overlapping resolution mapping Occupational Safety and Health Administration Open tubular (type of column) Ohio Valley Speciality Chemical (line of G C stat. phases) Pressure; Pi, inlet pressure; Po, outlet pressure Polarity index Partial pressure Vapor pressure, at equilibrium Polyacrylamide gel electrophoresis Polycyclic aromatic hydrocarbon Paper chromatography or personal computer Polydimethylsiloxane (polymer class) Pharmacopeial Forum Pressurized fluid extraction, also called accelerated solvent extraction (ASE) Photodiode array (detector) Acidity scale; - log a H+ - logK Porous-layer open tubular (column) Parts per billion Parts per million Performance qualification Pounds per square inch (pressure) Programmed temperature gas chromatography Programmed temperature vaporization Vapor pressure of water Fraction extracted Configuration factor (in rate equation) Quality assurance Quality control Quantitation Limit Gas constant Retardation factor (columns)

XXiV

r RAM R&D RF R M

RRF

RS

rc RI RM RPLC RSD RTIL S S" SBSE SCOT SDS SEC SFC SFE SIM SN SOP SP SPE SPME SRM

ss

STP T T' t TCD tD

THF TF TFA TIC TLC

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

Radius Restricted access media Research & Development Retardation factor (TLC) Martin retention parameter Relative response factor (for detectors) Resolution Radius of a column (inside) Refractive index (detector) Reference material Reversed phase liquid chromatography Relative standard deviation Room temperature ionic liquids Sensitivity (of a detector) Adsorption energy of a solute (see Snyder solvent parameter) Stir bar sorptive extraction Support-coated open tubular (column) Sodium dodecylsulfate Size exclusion chromatography Supercritical fluid chromatography Supercritical fluid extraction Single ion monitoring (MS) Separation number (see TZ) Standard operating procedure Stationary phase Solid phase extraction Solid phase microextraction Standard reference material System suitability Special technical publication (of ASTM) Temperature Significant temperature (in PTGC) Time Thermal conductivity detector Dwell time (LC) Tetrahydrofuran Tailing factor Trifluoroacetic acid Total ion chromatogram (MS) Thin-layer chromatography

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

tM

TOF tR

tk

TSI TZ

U

uc

USA USDA USP USP/NF

uv

UV/Vis V V

v,

VD

VM

VN

VR VR

v;

VS K S

VWD U UEOF 'ion

W

wb

wh Wi

WCOT WEF X Z

XXV

Hold-up time Time of flight (mass spectrometer) Retention time Adjusted retention time Thermospray ionization (MS) Trennzahl number Velocity of the mobile phase; u F = velocity of MP front Unified chromatography United States of America United States Department of Agriculture United States Pharmacopeia United States Pharmacopeia/National Formulary Ultraviolet; ultraviolet absorption detector Ultraviolet/visible (spectrophometry) Volume or voltage Molal volume Adsorbent surface volume (see Snyder solvent parameter) Dwell volume (LC) Volume of the mobile phase; hold-up volume; V,, volume of MP gas; V,, volume of MP liquid (also volume of SP liquid) Net retention volume Retention volume Adjusted retention volume Corrected retention volume (GC) Volume of the stationary phase; VL, volume of SP liquid; Vs, volume of SP solid Volume of solid support Variable wavelength detector Velocity of an analyte Electroosmotic flow velocity Electrophoretic velocity of an ionized analyte Weight; W, is the weight of an unknown; etc. Width of a peak at the base Width of a peak at half-height Width of a peak at the inflection point Wall-coated open tubular (column) Water Environment Federation Mole fraction Charge on an ion

XXVi

a

P

Y

YA

6 &

&O

77

e

K L

h P

pTAS Pion

v

P 0-

a2 r cp

a)

* w

5

SYMBOLS, ABBREVIATIONS, AND ACRONYMS

Separation factor Phase volume ratio Surface tension Activity coefficient for analyte A Hildebrand solubility parameter Porosity Snyder solvent parameter Viscosity Contact angle (in TLC) Permeability Dielectric constant Packing characteristic (in rate equation) or Wavelength (of a UV/Vis spectrometer) Dipole moment Micro total analysis system Electrophoretic ion mobility Reduced velocity Polarizability Standard deviation or quarter-zone width Variance Time constant (detector) Volume fraction (Hildebrand solubility parameter) Flow resistance parameter Obstruction factor (rate equation) Packing factor (rate equation) Zeta potential

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTlC ES ON ANALYTICAL CHROMATOGRAPHY Chromatography is a proven method for separating complex samples into their constituent parts, and it is undoubtedly the most important procedure for isolating and purifying chemicals. Using data from the first half of 2003, Ryan estimated that nearly 5% of all chemical research in 2003 would involve chromatography. In addition, most chromatographic instrumentation is equipped with detectors, making chromatographs true instruments, devices capable of making measurements. Consequently, this monograph will deal not only with the principles of chromatography but also with the practice of quantitative analysis. It is this latter subject that has been greatly influenced by both industry and the federal government because of the need for standards and standardization that go hand-in-hand with governmental regulation. In the modern world, these issues extend to foreign countries as well and have given rise to international organizations and guidances/regulations that need to be recognized by chromatographers worldwide. Since much important information is available on the Internet, all scientists need to be knowledgeable about its retrieval and its impact on their work. In addition, much effort is being made internationally to provide a cooperative and harmonized approach to analysis and analytical method development. Although this book is

'

Chrornutogruphy: Concepts und Contrusts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller 1

2

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

written from the perspective of chromatographers in the United States, the principles are applicable internationally, and scientists would be well advised to recognize that fact and become aware of the developments outside their own countries. Fortunately, the fundamental principles of chromatography and analytical chemistry in general are the same in academia, industry, and government, of course. Their common objective is to perform laboratory tests and procedures that are based on sound scientific principles. However, some industries operate under more stringent controls than others. For example, the pharmaceutical industry in the United States is regulated by the Food and Drug Administration (FDA), which enforces federal regulations known as the Current Good Manufacturing Practices (CGMPs)?. These regulations were promulgated to ensure the safety and efficacy of drugs by setting forth minimum standards for manufacturing and testing. The GMPs are not prescriptive and, therefore, they have been supplemented by FDA guidance documents that provide more specific details on complying with the regulations. These guidances provide insight for the practice of good chromatography in all venues where analytical chemistry is performed, in the United States and abroad. While it is true that European and Asian counterparts are similarly regulated by their respective agencies, the fundamental analytical principles are the same and are becoming internationally codified. Because these special regulations and guidances are often omitted from academic courses,' this chapter is presented to guide informed readers as they proceed to industrial and governmental employment. It also serves as a general introduction to quantitative analysis practices in chromatography by presenting and summarizing some basics of chromatographic measurement. This chapter examines: The organization of analytical chemists in a typical industrial corporation The organization and regulatory agencies of the U.S. government and of nongovernmental agencies The effect of FDA regulation on the pharmaceutical practices in the laboratory Some international guidelines for analytical chemistry in general and analytical chromatography in particular.

*The official title of the FDA regulations includes the word Current so the abbreviation should be CGMP. Howevcr, some authors use ii lowercasc c and call them the cGMPs, and others shorten the name to just GMP. For simplicity, we will use G M P in most cases.

1.1

1.1

LOCUS OF CHROMATOGRAPHY IN CHEMICAL INDUSTRY

3

LOCUS OF CHROMATOGRAPHY IN CHEMICAL INDUSTRY

Chemical companies and related industries such as pharmaceutical companies and the petroleum industry more than ever need to have laboratories devoted to analysis methods and characterization, including in most cases a section well trained in chromatography. Those that produce and sell chemicals have a laboratory function called quality control (QC) that monitors the quality of incoming raw materials, evaluates in-process intermediates, and tests the purity of final products. Assurance of the quality of manufactured products, referred to as quality assurance (QA) and carried out in conjunction with manufacturing, is a related function. Both functions may be combined, and the laboratory may be called a QC/QA laboratory. This laboratory usually performs both qualitative (identity) analyses and quantitative analyses. The latter are often performed by gas chromatography (GC) or liquid chromatography (LC). Usually, these laboratories are situated close to, or within, the manufacturing site. Typical of many companies hiring B.S. chemists, large pharmaceutical firms hire recent bachelors chemists as analytical chemists into their QC laboratories.’ Depending on the size of the company, another laboratory may be responsible for developing the methods for the QC laboratories. This function may be in the Research and Development (R&D) Department. The chromatographers in this laboratory are usually responsible for keeping up with the latest developments in chromatography and searching for and evaluating new improved methods of analysis, as well as developing methods for the QC laboratory. Instrument companies manufacturing chromatographs may also have their own instrumental R & D groups that often provide technical support. Generally, R & D groups are staffed by degree chemists at several levels with some Ph.D.s at the highest levels. Another analytical need is for a group to perform general analytical services to support the chemical activities of the company (synthesis, pilot plant, product support, etc.). These services most often include chromatography, spectroscopy, and microanalytical (elemental) analysis. Often this is a separate group of scientists and engineers and may include a small group of experts that advises and consults with technicians in the other areas who do their own analytical work. Separate groups may exist to support the sales and marketing department or the patent and law department, for analysis of competitors samples or evaluation of patent infringement, for example. Within a chemical corporation, these various laboratories are responsible for providing accurate and reliable analytical methodology. The interrelated elements required for this process are shown in Figure 1.1. Each part is important, and some of them will be discussed further in this chapter:

4

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

Figure 1. l . Interrelated elements that ensure reliability of data. Reprinted with permission from J. Miller and J. Crowther (eds), Analyticu[ Chemistry in u G M P Enuironrnent, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

standards, instrument qualification, and method development and validation. In general, government laboratories are organized similarly. Some of them are of particular interest to analysts because of the functions they perform, including the regulation of industrial practice. 1.2 GOVERNMENTAL ORGANIZATIONS

Table 1.1 lists some U.S. government laboratories and agencies that are of interest to chromatographers. Those that are part of a governmental department are listed by department in order to show the governmental organization. The ones of greatest interest to chromatographers, and the ones discussed in greatest detail in this chapter, are the National Institute of Standards and Technology (NIST), the Food and Drug Administration (FDA),

1.2

Table 1.1

GOVERNMENTAL ORGANIZATIONS

5

U.S. Government Laboratories and Scientific Agencies and Departments

Departments

Agriculture (USDA). Over 100 research labs nationwide. Commerce. Includes the National Institute of Standards and Technology (NIST), formerly the National Bureau of Standards (NBS). Energy (DOE). Sixteen laboratories including the famous ones at Argonne, Brookhaven, Los Alamos, and Oak Ridge. Health and Human Services (HHS). Includes the Centers for Disease Control and Prevention (CDC) and its division, the National Institute for Occupational Safety and Health (NIOSH); the Food and Drug Administration (FDA); and the National Institutes of Health (NIH). Justice. Includes the Drug Enforcement Agency (DEA) and the Federal Bureau of Investigation (FBI). Labor. Includes the Occupational Safety and Health Administration (OSHA). Interior. Includes the U S . Geological Survey (USGS). Treasury. Includes the Bureau of Alcohol, Tobacco and Firearms (ATF). Other Environmental Protection Agency (EPA) National Science Foundation (NSF)

and the Environmental Protection Agency (EPA) because they are most involved in standards, standardization, method development, and federal regulation. The U.S. government web site (www.firstgou.gou) can be used to locate additional information on government agencies and federal regulations. National Institute of Standards and Technology (NIST)

The mission of NIST (formerly the National Bureau of Standards, NBS) is “to develop and promote measurement, standards, and technology to enhance productivity, facilitate trade, and improve quality of life.774It was founded in 1901, making it the oldest physical science research laboratory of the federal government.’ Unlike the FDA and the EPA, it is not a regulatory agency and does not establish or enforce mandatory standards; rather, NIST develops measurement methods, instrumentation, and measurement standards for government and industry.6 The main NIST laboratory is outside Washington, D.C., in Gaithersburg, Maryland, and the second one is in Boulder, Colorado. One of the eight laboratory divisions, the Chemical Science and Technology Laboratory

6

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

(CSTL), includes an Analytical Chemistry section that is divided into five groups. One of them is the Organic Analytical Methods group where separation methods, including most of chromatography, is located. CSTL performs services like those described above for R & D departments; it “conducts research in measurement science and develops the chemical, biochemical, and chemical engineering measurements, data, models, and reference standards” for the United state^.^ Reference standards are particularly important in analytical chemistry, and a later section of this chapter is devoted to that topic. The Analytical Chemistry section of the CSTL is responsible for 850 of the 1350 NIST standards, called standard reference materials or SRMS.‘ On the occasion of its attaining the age of 100, the NIST published a booklet chronicling the first century of SRMs.’ Some chromatographic examples of SRMs are:

869a for LC selectivity 870 for LC performance 877 for LC chiral selectivity 1543 for gas chromatography/mass spectrometry (GC/MS) performance.

As an illustration of the nature of SRMs, 869a is a mixture of three polycyclic aromatic hydrocarbons (PAHs) in acetonitrile, useful for characterizing LC column selectivity for the separation of PAHs. The NIST also provides a wide range of publications and databases. Called the NIST Virtual Library, they can be accessed online at nul.nist.gou. Worldwide coordination and cooperation between the individual standardization agencies is also a task of NIST. Globally recognized measurements and standards are being developed through the efforts of many national metrological institutes worldwide, through the signing of a Mutual Recognition Arrangement (MRA) whereby 50 national standards laboratories have agreed to participate in formal interlaboratory comparisons.8 The responsibility for this effort in the United States is carried mainly by the Analytical Chemistry section of the NIST. Food and Drug Administration (FDA)

The FDA is a regulatory agency formed as a result of the government’s Food, Drug and Cosmetic Act (FD & C act) in 1938. Simply stated, its mission is “to promote and protect the public health by helping safe and effective products reach the market in a timely way, and monitoring products for continued safety after they are in use.”9 In addition to food and drugs, the FDA regulates cosmetics, medical devices (such as pacemakers), biologics (such as

1.2

GOVERNMENTAL ORGANIZATIONS

7

vaccines), animal feed and drugs, and radiation-emitting products (such as cell phones). Its web site" contains a wealth of information, some of which is indicated in the flowchart in Figure 1.2. The focus in this chapter will be on drugs. There are about 10,300 FDA-approved drugs in the United States today,' and the division of the FDA responsible for most of them is the Center for Drug Evaluation and Research (CDER). The Center for Biologics Evaluation and Research (CBER) is responsible for biologicals, and the Center for Veterinary Medicine (CVM) regulates veterinary drug products. The CDER reviews applications for new drugs (NDAs) and generic products (ANDAs) and oversees the quality and manufacturing of drugs by participating in on-site inspections with the office of regulatory affairs (ORA). The regulations it enforces are federal laws called Good Laboratory Practice (GLP) and Good Manufacturing Practice (GMP), or alternatively, Current Good Manufacturing Practice (CGMP) as noted earlier.' One might think that chromatographers would be most concerned with GLPs, but that is not the case. It is primarily the GMPs that provide the regulations applied by laboratories to give assurance that the manufactured products meet specifications. GLPs mostly concern the conduct of nonclinical laboratory (toxicology) studies, while GCPs address Good Clinical Practices. All of these regulations are sometimes lumped together and referred to as GXPs when not referring to a specific regulation. The necessity to conform to the applicable GXPs has had major effects on the operation of analytical laboratories in the pharmaceutical industry; many of these basic business principles outlined in the GMPs have been adopted by others in the wider analytical community. A major requirement regarding analytical methods is that they must be validated. Method validation is the process of acquiring data and documentation to prove that a specific method will produce reliable data with a high degree of assurance and is therefore acceptable for its intended purpose. The measures for evaluating a quantitative method, such as a high-pressure liquid chromatographic (HPLC) analysis, include accuracy, precision, specificity, linearity, range, limit of detection (LOD), limit of quantitation (LOQ), robustness, and sensitivity [added later by an International Conference on Harmonisation+ (ICH) guideline]. An equally important requirement is that instrumentation used in the testing method and during validation activities must also meet stringent controls referred to as instrument qualifications. As a matter of clarification, in

"'

See page 2. 'Spelling harmonisation with an s is the British version. In this text, when a European group or agency is being rcfcrcnced, the British spelling will he used.

Figure 1.2. Partial flowchart of the FDA web site. Reprinted with permission from LC-GC Europe, Vol. 16(1), January 2003, p. 40. LC-GC Europe is a copyrighted publication of Advanstar Communications, Inc. All rights reserved.

12

GOVERNMENTAL ORGANIZATIONS

9

general, instruments are qudlified and processes (methods) are validated. Further details on these subjects are deferred until later in this chapter. There are other key compliance issues in addition to method validation and instrument qualification including ' I : Management systems Operating procedures Personnel training Data accountability Facility adequacy and compliance Certification documentation Surely this list represents requirements that one would expect to address when attempting to improve one's laboratory practices. Although detailed discussion of all of these topics is beyond the scope of this monograph, some of the most important issues are addressed; additional information can be found in the published literature.". The GMPs are published by the National Archives and Records Administration and the Government Printing Office (GPO) in the Code of Federal Regulations (CFR), which is a codification of the general and permanent rules published by the executive departments and agencies of the federal government. It can be accessed online from the FDA web site or directly at www.~~oaccess.h.ou/c~/index.html. The CFR is divided into 50 titles, which represent broad areas subject to federal regulation; the GMPs are in Title 21, as listed in Table 1.2. Before final publication and becoming law, new proposed regulations are first published for review in the Federal Register (FR), also accessible from the FDA site, as well as directly at

Table 1.2 Sectionss of CFR Title 21. GXPs

Part I 1 Parts 50, 54, 56 Part 58 Part 1 10 Part 210 Part 21 1 Part 600 Part 610 Part 820

Electronic records; electronic signaturcs GCP for clinical laboratories GLP for nonclinical laboratories CGMP in manufacturing, packing, or holding human food CGMP in manufacturing, processing, packing, or holding of drugs CGMP for finished pharmaceuticals (GMPs) Biological products General biological products standards Quality system regulation

10

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

www.gpoaccess.gov/ f r / index.htm1. Free online access to these federal regulations is a service only recently made available, and one that should be widely exploited. Surprisingly, the FDA does not prescribe methods. In the pharmaceutical industry in the United States, the methods most often used are those published by the U S . Pharmacopeial Convention (USP/NF), which is not a federal agency or publication. The USP publishes methods, many of which have been approved by the FDA, but the FDA does not submit methods to USP. Companies are free to use their own methods as long as they are as good as, or better than, the USP/NF methods. Consequently, method development is done by companies and universities and is based on the regulations published by the FDA. The FDA (in particular, the CDER) and the USP/NF work very closely together; the FDA reviews and comments on USP information and standards. To be in compliance with FDA regulations, a pharmaceutical laboratory has to provide data and documentation to show that its methods meet the requirements published in the USP/NF. The GMP regulations in Title 21 of the CFR are general, and they are not specific enough to be enforced without further elaboration. Discussion of, and comments about, regulations are often presented online at the FDA site, and compliance issues are also addressed in the USP publication, Pharmacopeial Forum (PF). The FDA publishes CGMP Notes quarterly. These notes are intended to clarify issues and answer questions related to interpretation of the GMP regulations. They are not regulations and are primarily for internal FDA use; they can be accessed at www.fda.gou/cder/dmpq/ cgmpnotes.htm. The FDA publishes guidelines or guidance documents in an attempt to clarify the intent of the regulations it intends to enforce. In 1997, in an attempt to be more specific about the intent and meaning of the term guidance, the FDA published in the FR a notice on guidance documents.’” In effect, it created a new category of GXPs, Good Guidance Practices (GGPs), setting forth its policies and procedures for developing, issuing, and using guidance documents. The notice states; “Guidance documents do not themselves establish legally enforceable rights or responsibilities and are not legally binding on the public or the agency. Rather, they explain how the agency believes the statutes and regulations apply to certain regulated acti~ities.”’~ The majority of future guidance documents will be labeled either (1) compliance guidance, (2) guidance for industry, or ( 3 ) guidance for FDA reviewers and staff. Type 2 documents are of primary interest to chromatographers in the laboratory and are the ones receiving the most attention in this chapter. An example of guidance documents on method validation is Validation of Chromatographic Methods’4 issued in 1994 (typical of guidance documents

1.2 GOVERNMENTAL ORGANIZATIONS

11

issued before the 1997 statement and originally published in the FR). Other examples of draft guidance documents include Anulyticul Procedures for Methods Vulidution" issued August, 2000, another concerning out-of-specification (00s) results,'" and o n e on residual solvents." They can be downloaded from the F D A C D E R site at w w w . ~ ~ f i g o u / c ~ e r / g u i d u n c e / i n d e x . h t m and also from C B E R at www.fda.gou/ cher /puhlications.htm. Other FDA information of interest to pharmaceutical chromatographers can be found in reference 10. Furthermore, since many American pharmaceutical companies market their drugs outside the United States, they then need to meet the requirements of foreign regulatory agencies as well as those of the FDA. In fact, a worldwide effort to achieve common requirements and international cooperation has been ongoing since the early 1990s and is called the International Conference on Harmonisation (ICH). A later section in this chapter is devoted to the U S P / N F and international pharmacopoeias. Environmental Protection Agency (EPA)

T h e E P A has been working for over 30 years to protect human health and to safeguard the natural environment (air, water, and land) in the United States. It develops and enforces environmental regulations through 10 regional offices and 17 laboratories. The regulations are codified in Title 40 of the CFR, which can be accessed online from the E P A web site: www.epa.gou. T h e EPA Office of Research and Development ( O R D ) has publications in the following areas: general, air, EMPACT, multimedia, pollution prevention, risk, risk assessment guidelines, S T A R grant research, waste, and water.Ih Laboratory methods of interest to chromatographers can be found in the eight Laboratory Analytical Chemistry Methods Manuals, covering the topics listed in Table 1.3. They were originally published by the former Table 1.3 EPA Laboratory Analytical Methods Manuals

I . Methods for the Determination of Organic Compounds in Drinking Water;

EPA-600/4-88/039 2. Supplement I of Organics Manual; EPA-600/4-90/020 3. Supplement I1 of Organics Manual; EPA-600/R-92/ 129 4. Supplement 111 of Organics Manual; EPA-6000/R-95/ 131 5. Methods for the Determination of Inorganic Substances in Environmental Samples; EPA-600/R-93/ 100 6. Methods for the Determination of Metals in Environmental Samples; EPA-600/4-9 1/ O l O 7. Supplement I of Metals Manual; EPA-60O/R-94/ 111 8. Methods for the Determination of Chemical Substances in Marine & Estuarine Environmental Samples; EPA-600/R-92/121

12

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

Environmental Monitoring Systems Laboratory in Cincinnati between 1988 and 1995. The scope of this project can be seen from the fact that the 1988 Manual on Methods for the Determination of Organic Compounds in Drinking Water contains 13 methods cross-indexed to over 200 analytes. The individual methods are listed at the web site www.epa.gou/nerlcwww /methmans.html and can be purchased from the National Technical Information Service (NTIS).~~ Other Organizations

Two other agencies listed in Tab'le 1.1 have issued standard chromatographic methods of analysis. They are the National Institute of Occupational Safety and Health (NIOSH) and the Occupational Safety and Health Administration (OSHA). From their names one would expect them to be in the same department, but NIOSH is a subsection of Centers for Disease Control and Prevention (CDC) in the Department of Health and Human Services (HHS), and OSHA is in the Department of Labor. Like EPA they have published many methods, some covering the same chemicals. Both agencies methods are available in print and online.20.21 1.3

NONGOVERNMENTAL AGENCIES

The most relevant nongovernmental agencies and societies are listed in Table 1.4 along with their web sites and some of the relevant activities in which they engage. The latter include activities such as specification of standards and standardization (S), development and recommendation of analysis methods (M), recommended regulations (R), recommendations of nomenclature and definitions (N), and international activities promoting harmonization and cooperation (I). Many of them also publish reports and journals; some are active only in the United States while others (often identified by their names) are international. Four have been chosen for extensive commentary in this section: Association of Analytical Communities International (AOAC), American Society for Testing and Materials (ASTM) International, International Organization for Standardization (ISO), and International Union of Pure and Applied Chemistry (IUPAC). Before discussing them, a few comments will be made about a few of the others, but many of the 25 agencies listed in Table 1.4 cannot be included in this brief section. Internet URLs (universal resource locators) are given for the purpose of obtaining additional information about them. The American Chemical Society (ACS) is probably well known to all readers. Its activities, in the context of this discussion, are exemplified by its publication of specifications for reagent chemicals.22 They are the specifications for the quality grade of over 400 chemicals referred to as ACS reagents.

1.3

NONGOVERNMENTAL AGENCIES

13

Table 1.4. Nongovernmental Agencies and Societies

Name

Web Site

Activity"

Home

I . American Chemical Society (ACS)

u'iw.iic5.org

2. American National Standards Institute (ANSI )

IIWW.~~I~.\I.OI~

N, S I, s

Washington, D.C.

3. American Public Health Association (APHA )

www.upha.org

M

Washington, D.C.

4. American Water Works Awiciation (AWWA )

www.izwwu.org

M

Denver, CO

5. Analytical Kr Life Science System Association (ALSSA ), formerly Analytical Instrument Aisoc. ( A I A )

h. Aw)ciation 0 1 Analytical Communities

(AOAC ) Intcrnational

n'ww. uouc.org

S

Alexandria, VA

I, M. K, S

s

7. American Society for Quality (ASQ, formerly ASQC )

1.

8. ASTM Internationiil

1, M, N, S

9. Cooperation o n International Traceability in

Analytical Chcmi\try (CITAC )

1,

Washington, D.C.

s

Gaithersburg, MD Milwaukee. WS W. Con\hohocken, PA

Gcel, Belgium

I

Wohurn, MA

I I . Eurachcm

1. M. S

Portugsl

12. Instrumentation. Systems, and Automation Society (ISA)

1. s

Research Triangle Park. NC

13. Intcrnatiiinal Atomic Energy Crininii\sion 14. Intcrnational Conference on Harmonisation (ICH)

1. M. S

Viennct

15. International Electmtechnic;il (IEC)

I

Gcneva

A. 1, S

Netherland\

1. S

Geneva

I

Tampa. FL

19. 1nternation;il Union of Pure 'ind Applicd

1. N , S

Chemi\try ( I U P A C ) 20. In\titutc for Reference Material\ and Mcawrcmcnts. EC (IRMM )

Kc\earch Triangle Park, NC

1. s

Gcel, Belgium

21. National Ccinfcrcncc of St;ind;ird\ Lnhoratorics (NCSL ) International

1. N. S

Boulder, CO

?7

1, N, K, S

Paris

23. Product Quality Research h t i t u t e (PORI )

1. R

Arlington, VA

24. US. Ph;irmacopcial Conventi(in (LISP)

M, N, S M

Alexandria. VA

10. Collaborative Electronic Notebook System\

Aswciation (CENSA )

Conimissim

16. Intcrn;itional

I.aboratory Accreditation Coopcration (II.AC)

17. International Orgeniration lor Standardizaticin ( I S 0 ) It(.

International Society for Pharn1;iccutical Engineering

Organi\ation for Economic Cooperation and Development ( O E C D )

25. Water Environment Federation ( W E F )

-

1. M. N. R. S Geneva

Rockville, MD

Activities: I, international harmonization and cooperation; M, publication and/or development of methods; N, nomenclature and dcfinitions; R, regulations/regulatory; S, standards, standardization, and protocols.

14

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

This book also contains useful definitions and procedures for analytical chemistry, including those on chromatography. An online demonstration of the web edition can be accessed at http://pubs.acs.org/reagents/index.html. Updates to the ninth edition are also posted there. Another useful monograph gives standard methods of analysis for water and wastewater.2’ It has been published in a collaborative effort of the American Public Health Association (APHA), the American Water Works Association (AWWA), and the ’Water Environment Federation (WEF) and contains over 350 separate test methods. Furthermore, the EPA has just given regulatory approval to this latest edition, making it an official manual for EPA methods. A joint venture headed by the Analytical Instrument Association (AIA, now renamed ALSSA, Analytical and Life Science Systems Association) has produced protocols for chromatographic data interchange. Called AND1 protocols (for analytical data interchange), they are intended to increase laboratory efficiency and productivity by facilitating the integration and use of data from multiple vendors’ ii~struments.‘~ Nine chromatographic instrument companies are currently participating, and ASTM has adopted these protocols.2s Further information and references can be found in references 24 and 25.

Association of Analytical Comimunities International (AOAC) The AOAC started out in 1884 as the Association of Official Agricultural Chemists under the U.S. Department of Agriculture (USDA), then became the Association of Official Analyl.ical Chemists, and since 1991 is the Association of Analytical Communities; the abbreviation for all of these names has remained AOAC, but the breadth of the organization has increased. It has always been one of the leading organizations producing standards for analytical chemists. Its book of “official methods” is in its 17th edition” and contains 2800 tested analytical methods. It is the single most comprehensive collection of validated analytical methods available anywhere. The AOAC International has written three method validation programs and administers many contracts with other agencies and organizations, such as the FDA and the USDA. Its activities are no longer limited to regulatory functions as was implied when thle term “official” was part of its name, and increasing emphasis is being placed on international collaboration and cooperation. Further details are available at its web site (see Table 1.4).

ASTM International The ASTM’s original full name, American Society for Testing and Materials, reveals that it was originally estatilished to develop and publish standard test

1.3

NONGOVERNMENTAL AGENCIES

15

Table 1.5 Section Contents of the Annual Book of ASTM Standards

No. 1

2 3 4 5 6 7 8 9 10 11 12 13 14 15

Topic Iron and Steel Products Nonferrous Metal Products Metals Test Methods and Analytical Procedures Construction Petroleum Products, Lubricants, and Fossil Fuels Paints, Related Coatings, and Aromatics Textiles Plastics Rubber Electrical Insulation and Electronics Water and Environmental Technology Nuclear, Solar, and Geothermal Energy Medical Devices and Services General Methods and Instrumentation General Products, Chemical Specialties, and End Use Products

methods primarily for America. Now, it is significantly engaged in worldwide issues, as appropriately reflected in its change of name to ASTM International. ASTM is not a regulatory agency, and its methods are voluntary and are arrived at by consensus among groups of interested scientists. Over 30,000 individuals from 100 nations are members of ASTM International, evidence that it is no longer restricted to America. T h e methods have been published annually for many years, and currently cover over 11,000 standards in 15 sections contained in more than 70 volumes.” Table 1.5 lists the 15 sections, most of which are related to manufactured products. T h e contents can be searched online by title and by subject at the ASTM web site, but the text is only available for a fee, in print o r online. Typical methods of interest to chromatographers include D6420-99 Standard Test Method for Determination of Gaseous Organic Compounds by Direct Interface Gas Chromatography-Mass Spectrometry (Vol. 11.03) and D6156-97 Standard Practice for Use of Reversed-Phase High Performance Liquid Chromatographic Systems (Vol. 11.02). T h e ASTM International has also published 6 sets of data (designated DS), 44 manuals (designated MNL), 1434 special technical publications (designated STP), and 5 journals. Some examples of STPs are: STP 577 Calculation of Physical Properties of Petroleum Products from G a s Chromatographic Analysis STP 1161 Leak Detection in Underground Storage Tanks

16

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

STP 1223 Standardization an,d Harmonization Terminology: Theory and Practice Most of its work is done by t’echnical committees that have broad representation to assure wide consensus. While most are on specific materials, the E-committees cover miscellaneous subjects, and E-19 is the one on chromatography. International Organization for Standardization (ISO)

The I S 0 is a nongovernmental, international organization established in 1947. The name, I S 0 is obviously not an acronym for its name. In English, the prefix is0 is used to denote “same”; in fact, is0 is derived from the Greek isos, meaning “equal.” Being a truly international organization from the start, I S 0 chose as its name an acronym that is internationally recognized as denoting same or equal or standlard, regardless of the language of the user. The ISO’s work results in international agreements, over 13,500 of which have been published as international standards. Standards are classified among 40 different fields. Number 19 is Testing and number 71 is Chemical Technology. Analytical Chemistry is number 71.040, and subsection 50 (71.040.50) includes chromatographic analyses. A key word search of the online catalog turned up 80 methods using chromatography. Some examples are: I S 0 7609:1985 Essential Oils-Analysis by Gas Chromatography on Capillary Columns; I S 0 14718:1998 Animal Feeding Stuffs-Determination of Aflatoxin B1 Content of Mixed Feeding Stuffs; a Method Using High-Performancce Liquid Chromatography. A general standard of particular interest to chromatographers is ISO/IEC 17025, General Requirements for the Competence of Testing and Calibration Laboratories, published in 1999.2x The 30-page document is not available online, but it can be ordered from the web site. Somewhat like the GMPs of the FDA, it deals with general areas such as terms and definitions, management requirements, and technica I requirements. It addresses issues including quality systems, personnel, internal audits, method validation, sampling, standards, equipment, and data handling. Finally, I S 0 9000 is a family of standards dealing with quality management systems. They are generic standards, meaning that they can apply to any organization that wishes to enhance customer satisfaction by meeting customer needs and regulatory requirements. Most chromatography manufacturers and suppliers have conformed to these standards and have issued I S 0 9000 certificates. Dealing with these companies should be better as a result. It should be noted, however, that I S 0 does not carry out the certification and does not issue certificates.

1.4 STANDARDS, CALIBRATION, AND NlST

17

The American National Standards Institute (ANSI, see Table 1.4) is the official U.S. representative to ISO. It is a private, nonprofit organization that administers and coordinates the U S . voluntary standardization and conformity assessment system.'" It has done so for over 80 years and currently has approximately 1000 members, representing industrial companies, organizations, government agencies, and other institutions. It is located in Washington, D.C. More information is readily available at its web site (see Table 1.4). It should be noted that ASTM International is also a cooperating agency with I S 0 and has sponsored or co-sponsored many of its committees. International Union of Pure and Applied Chemistry (IUPAC)

The IUPAC is the granddaddy of international chemical bodies. Since its beginning in 1919 it has served worldwide as the primary agency fostering harmonization among chemical groups, industrial and academic. It has long been recognized as the authority on chemical nomenclature and terminology, atomic weights, and standardized methods for measurements. Of its eight divisions, analytical chemistry is division 5. The IUPAC publishes three journals including Pure and Applied Chemistly and has an online newsletter, Chemical Education International (see Table 1.4 for web site). Its published books include a series on solubility data,j" a periodic handbook," and a compendium on analytical nomenclature.32 The latter can he accessed online; Chapter 9 is on Separations. The IUPAC recommendations on chromatographic nomenclature were originally published in 199333and resolved many conflicting symbols and terms. Its recommendations will be used throughout this book and further discussion is presented in Chapter 2. Some current projects of IUPAC include harmonization of international quality assurance schemes for analytical laboratories and studying the definitions of asymmetrical chromatographic peaks.34 1.4

STANDARDS, CALIBRATION, AND N E T

One of the important steps in any analytical method involves calibration with appropriate standards. Prior discussion in this chapter has included some information on this topic and has given an indication of the number of organizations interested in it. This section will attempt to summarize the main aspects of the calibration process and collect in one place the contributions of the various organizations. The term standardization can be used in a number of different situations. For example, the attempts by various organizations to agree on a specification, a method, or a definition are all examples of the process of standardization. Alternatively, an example of standardization in the laboratory is the

18

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

process of comparing the strength of a solution against a standard. The preparation of such a “standard” may require its purchase from a supplier of standards, or it may be the process of comparing a newly prepared solution against a certified standard, thus producing a “secondary standard” or “working standard.” Let us examine the laboratory standardization process as it is commonly practiced by chromatographers. First is the process of agreeing on a method to be used for a particular analysis. Earlier we discussed lhose agencies that are concerned with the process of arriving at approved, standard methods. There are more than 400 organizations in the United States alone dedicated to this purpose.” Their standards can be mandatory or voluntary. Most stringent are the mandatory standards that are enforced on a regulated industry by the government. If one works for a pharmaceutical company, for example, the method to be used will be the one approved by the FDA. Once approved by the FDA, pharmaceutical companies, and other interested parties, usually submit their methods to the USP/NF for publication, although this may not occur for several years after FDA approval and they are not obligated to do so. Some standards are arrived at by consensus from among t h e constituents who will use it, usually on a voluntary ba:sis. ANSI is closest to being the centralizing voice for standards development in the United States.35 As stated earlier, ANSI is also the U.S. representative to ISO. Others we have discussed are so indicated in Tables 1.1 and 1.4. The standards just discussed can be referred to as standard test methods, but there is also another type: standard recommended practices.” The latter are generalized procedures, not specific instructions. They are recommended practices for various types of analysis and relevant test methods. Next is the process of standardizing the instrumentation in one’s own lab; this process is called instrument qualification and it will be discussed in a later section. Finally, there is the process of standardizing the method of analysis, the process we called method validatron. Extensive discussion of this procedure is given later in this chapter. Two short monographs prepared by ASTM International provide further details about standardization.‘, The standards used in standardization can be obtained from a number of sources. The official governmen1,al source, NIST, calls its standards SRMs, Standard Reference Materials, as noted earlier. A recent publication of a conversation with the current chief of the Analytical Chemistry Division of NIST contains interesting material about SRMS.’ Other agencies (including IS01 use other names such as Reference Materials (RM) and Certified Reference Materials (CRM).36 The official source for FDA methods is the USP whose list of standards are called Reference Standards (RS) now available online as well as the print version of USP/NF,

’’

1.5

USP AND OTHER PHARMACOPEIAS

19

General Chapter 11. Many analytical chemistry texts simply refer to primary standards and probably mean NIST standards. But, obviously, there are many different names used to identify suitable standards, and the one used usually depends on the context in which it is being used. Several chromatography supply houses sell standards for USP, EPA, and other standard method$’ some are traceable to NIST. In most cases standards are chemicals, and as such they must be of known purity and stable. Some may require oven drying prior to use. Some may have retest dates beyond which recertification is required. The general practice in the pharmaceutical industry is that expiration dates are final, but expiration dates are not usually attached to standards. See reference 11 for further information and reference 38 for an international guide for laboratories and accreditation bodies. For many new methods, no standards are available, of course. In that case, attempts are made to purify available chemicals as much as possible; they are then analyzed by more than one method to ensure proof of purity. If the identity of a chromatographic peak is totally unknown, as is often the case with small impurity peaks, obviously no standard can be prepared until an identification can be made and a method of synthesis is worked out. Much effort is required in this case, and often analyses must be performed without qualified standards. Quantitativc analysis in such cases is often performed by the method called area normalization (see Chapter 9) that may be very inaccurate.

1.5

USP AND OTHER PHARMACOPEIAS

Earlier discussion of the FDA’s regulation of the pharmaceutical industry discussed the role played by the USP/NF. This section will provide more information about the USP and its activities toward harmonization with pharmacopeias of other nations. The symbol USP can be used in two different ways. It can refer to the organization, the U S . Pharmacopeial Convention or to its major publication,” the USP/NF. Usually there is no confusion if the acronym is used, but if the usage is not clear, the full name will be written out. The U.S. Pharmacopeial Convention (see Table 1.4) was formed in 1820 and began publication of its pharmacopoeia. The latter served as a guide to drugs for physicians and pharmacists but had no legal status until the passage of the first Pure Food and Drug Act in 1906. It was combined with a similar publication, the National Fomulary (NF) of the American Pharmaceutical Association in 1974 when both were named as the official U.S. compendia

20

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

Table 1.6

Contents of the USP/NF

USP

NF

Introduction General Notices (GN) Official Monographs General Chapters" 1 1 Reference Standards 201 TLC Identification Test 467 Organic Volatile Impurities 621 Chromatography 726 Electrophoresis 727 Capillary Electrophoresis 1078 GMPs 1 196 Pharmaceutical Harmonization (Information from the Pharmaccutical Discussion Group) 1225 Validation of Cornpendial Methods 1251 Weighing on an Analytical Balance Reagents Refercncc Tables

Preface Admissions (submissions since last edition) General Notices (GN) Combined Index

Nutritional Supplements "Chapters Icss than 1000 arc general requircmcnts for tests and assays; those above 1000 are only informational.

under the Federal Food, Drug and Cosmetic Act.'" The USP was designated to cover drug substances and dosage forms, and thc NF, pharmaceutical ingredients. Both have been published together in one volume for many years and the latest version (reference 39) for the year 2004 is USP 27 and NF 22. It is available in print as a single volume, or on CD, or online. Currently, it is revised and reprinted every year. Two supplements are published each year between the annual revisions to keep it up-to-date. Table 1.6 shows partial contents of the USP/NF. The General Notices (GN) contain basic information about the volume that should be read, especially by chemists in the pharmaceutical industry. The Official Monographs make up the largest part of the USP; it is an alphabetical listing of USP drugs with USP monographs. The general chapters section includes assays, tests, and determinations including Chapter 621 on chromatography. The contents of the chromatography section (about a dozen pages) are listed in Table 1.7. System suitability will be defined in the next section and the terms and symbols will be discussed in Chapter 2. Another publication of USP is the Pharmacopeial Forztm (PF), a bimonthly journal that previews upcoming changes in the USP. It provides the opportu-

1.5 USP AND OTHER PHARMACOPEIAS

21

Table 1.7 List of Contents of Chromatography Chapter 621 of USP

Introduction Paper Chromatography (PC) Descending Chromatography Ascending Chromatography Thin-Layer Chromatography (TLC) Continuous Development TLC Column Chromatography (low pressure) Column Adsorption Chromatography Column Partition Chromatography Gas Chromatography High-pressure Liquid chromatography (HPLC) Size Exclusion Chromatography (SEC) Interpretation of Chromatograms (N,R , , a ) System Suitability Glossary of Symbols Chromatographic Reagents Packings Phases Supports

nity to comment on proposed changes before they become official. Another publication, Chromatographic Reagents, is a reference to brand names of column reagents listed in U S P / N F and PF. Other countries have their own pharmacopoeias, of course, and as international trade has developed, it has become necessary to coordinate and “harmonize” these various publications. The other major pharmacopoeias are the European (EP), British (BP), and the Japanese (JP).’1.4”Three of the four (BP was not included) have formed a new organization, the Internationa! Conference on Harmonisation (ICH) in the early 1990s. T h e position of the FDA regarding this effort was reported in the Federal Register in 1995,41 thereby beginning a commitment to participate in, and cooperate with, the ICH. International Conference on Harmonisation Guidelines

T h e topics addressed by the ICH are ( I ) quality, Q; (2) safety, S; (3) efficacy, E; and (4)other multidisciplinary, M. Some of the quality topics of interest are listed in Table 1.8, which includes the Federal Register citation where they were published by the FDA. Proposed new guidelines such as these go through five steps: step 2 opens the proposal for comments and step 4 is the

22

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

Table 1.8 Selected ICH Quality Topics

Reference Q2: Analytical Validation Q2A: Text on Validation of Analytical Procedures Fed. Reg. 1995, 60, 11260 Fed. Reg. 1997, 62,27463-27467 Q2B: Methodology

Q3: Impurities Fed. Q3A(R): Impurities in New Drug Substances Fed. Q3B(R): Impurities in New Drug F’roducts Fed. Q3C: Impurities: Residual Solvents Q4: Pharmacopoeias Q6: Specifications Fed. Q6A: Chemical Substances Q7: GMP Q7A: GMP for Active Pharmaceutical Ingredients Fed.

Reg. 2003, 68,6924-6925 Reg. 2003, 6N, 64628-64629 Reg. 1997, 62, 67377

Reg. 2000, 65, 83041-83063 Reg. 2001, 66, 49028-49029

final draft. When step 2 or 4 has been reached, the FDA publishes the newly proposed guidances in the FR. Step 4 guidance documents are available for use on the date they are published in the FR, which is ICH step 5. It is easy to follow this progression because the F R is available online, as are most FDA and ICH documents. An example of the progression of a document through the ICH approval process is the ICH Harmonised Tripartite Guideline: “Text on Validation of Quantitative Procedures, Q2A,” which was published (step 2) by the FDA in the F R on March 1, 1994 (58 FR 9750). Comments were accepted until May 16 of that year, and then ICH published it on October 27, 1994 (available from the ICH web site). The FDA published this document subsequently in the FR on March 1, 1995, Vol. 60, pages 11259 to 11262 (available from the gpoaccess web site). This final FDA document is also listed in the guidance documents section of the FDA web site at www.fda.gou/cder/guidance/ ichq2a.pdf. It is this guideline thdt contains the ICH glossary on validation, including the definitions of basic terms such as precision and accuracy. That part of the document can be found in Appendix A of this book. As of April 2000, the FDA has changed its policies somewhat; it now publishes in the F R only a notice that an ICH guidance is in step 5. The complete text of the actual guidance document is made available by the FDA, in print and at the web sites mentioned earlier. 1.6 INTERNATIONAL GUIDELINES FOR ANALYTICAL LABORATORIES

Many of the citations given thus far in this chapter covering the concept of good laboratory practice are rather general and have had the FDA GMPs as

1.6

INTERNATIONAL GUlDLlNES FOR ANALYTICAL LABORATORIES

23

the central focus. The issues raised are important for analytical chromatographers. More important, however, are operating guidelines for chromatographic practice that have been written as a result of the general recommendations. The GMP regulations are not very specific. It is the guidances that provide the details needed for the laboratory. The GMPs may say that analysts need to be trained, but what we need to know is how, when, and by whom. Another example is the validation document discussed above, which includes the definition of limit of detection but not an equation for its calculation. The calculations can be found in ICH Guideline document Q2B, and in this section we will consider it and other sources of specific recommended practices for chromatographers. However, at this time there is no single internationally agreed upon set of guidelines, so we need to consider the several international efforts at harmonization and regulation. Most guidelines agree that the following steps are necessary to comply with the FDA regulations for drugs: 1. Identification of the analyte: qualitative analysis 2. Method development: quantitative analysis. Composed of an assay method for the major component and the determination of impurities and/or degrandants 3. Method validation 4. Method transfer (when necessary) 5. Stability testing Steps 2 and 3 will be discussed after we take a look at the available guidelines. Sources of Guidelines

This chapter has highlighted the guidelines and guidances from the U.S. FDA and its related agencies. However, other organizations have also been active in producing international guidelines. In July 1993, representatives of IUPAC and I S 0 met in Washington, D.C., for the purpose of developing common concepts and t e r m i n ~ l o g y .The ~ ~ IUPAC recommendations resulting from that meeting were published in 1995 and have been reprinted in 199943 in an issue of Analytical Chimica Acta devoted completely to new recommendations related to ~ a l i d a t i o n . ~ ~ The ISO’s publication on requirements for competence of testing and calibration was mentioned earlier,24 but it contains only general guidelines much like the FDA’s GMPs and the general ICH documents. For specific information one must consult I S 0 guides or other collaborative documents such as those discussed in the next section.

24

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

Of importance in international circles is the CITAC/Eurachem guide to quality in analytical chemistry.4’ This 57-page document can be downloaded from the CITAC web site and is an updated version of a joint effort between CITAC and Eurachem, newly revised in 2002 to incorporate I S 0 17025. Another of their joint publications concerns analytical measurements and statistic^.^' It too is a valuable document (120 pages) and can be downloaded from the Internet. Eurachem’s guide to method ~alidation,~’ while intended for European analysts, is relevant for analysts in the United States, too. Another international document that resulted from a 1996 meeting co-sponsored by IUPAC, AOAC International, and I S 0 concerns recovery informat i ~ n . ~One ’ final source that must be included is the Organization for Economic Cooperation and Development (OECD) series on Principles of Good Laboratory Practice and Compliance Monitoring.4y It too can be downloaded from the OECD web site. Method Development, Validation, and Transfer

Many aspects of method validation have already been mentioned; it is the most important part of assuring that an analytical test method is suitable for its intended purpose. In this section we will look at the processes of method development and validation that proceed together as a new method is designed. Following that general discussion, some of the specifics regarding chromatographic method development and validation will be discussed. Although there are now many publications on method validation, one that many find useful is Green’s “ A Practical Guide to Analytical Method Validation.”’” After a brief introduction, he proceeds through the recommended steps starting with “establish minimum criteria.” Most of the subsequent steps describe the procedures for meeting the criteria established in the ICH document on validation, Q2A (see Appendix A). Some of them are discussed below. Notice of the official FDA guidance document regarding method validation for new drug applications (NDAs) was first published in August 2000.” The highlights have been presented in a short study by FDA chemists.’2 Useful discussion of validation is contained in the series of articles entitled Validation Viewpoint by Krull and Swartz in LC-GC, and in their book on the For example, they discuss specificity in their June 2001 columns4 and validation of impurity methods in two later columns.s’~s6Another recent work discusses validation following the I S 0 protocols and provides a good comparative discussion and useful I S 0 references.” Also, the USP General Chapter 1225 covers method validation in a way that generic drug manufacturers find useful. Originally published in the 1980s, the current version incorporates much material from the two ICH guidelines, but there are some differences.

1.6 INTERNATIONAL GUlDLlNES FOR ANALYTICAL LABORATORIES

25

Figure 1.3. A pharmaceutical method development flowchart. Reprinted with permission from J. Miller and J . Crowther (eds), Analytical Chernkfiy in a G M P Enuironrnent, Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

26

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

Figure 1.4. A validation process for an HPLC assay/purity method. Reprinted with permission from J. Miller and J. Crowther (eds), Annk/icul Chemistry in n G M P Enuironmenr, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

1.6

INTERNATIONAL GUlDLlNES FOR ANALYTICAL LABORATORIES

27

An extensive diagram of the validation process is shown in Figure 1.3 (from reference ll), which has been nick-named V-TR’AP, which stands for a developmental approval process that is intended to yield methods that are validatable, transferable, robust, reliable, accurate, and precise-another list of five criteria similar to the ICH list. Figure 1.4 is a schematic of a validation process specifically for an HPLC assay or purity method. Much more information can be found in the original reference.” For additional information on HPLC validation see references 56-58. Some of the specific FDA guidelines taken from the ICH document on Validation of Analytical Procedures” are as follows:

Range “If assay and purity are performed together as one test, and only a 100% standard is used, linearity should cover the range from the quantitation limit (QL) or from 50% of the specification of each impurity, whichever is greater, to 120% of the assay specification.” Accuracy “Accuracy should be assessed using a minimum of 9 determinations over a minimum of 3 concentration levels covering the specified range (e.g., 3 concentrations/3 replicates each).” Precision “Repeatability should be assessed using: (a) a minimum of 9 determinations covering the specified range for the procedure (3 concentrations/3 replicates each) or (b) a minimum of 6 determinations at 100% of the test concentration.” Detection limit ( D L ) “A signal-to-noise ratio (S/N) between 3 or 2:l is generally acceptable. The DL may be expressed as v

D L = ~ . ~ F

(1.1)

wherecr = the standard deviation of the response and S = the slope of the calibration curve.” Quanfitution limit ( Q L ) “A typical signal-to-noise ratio is 1O:l. The Q L may be expressed as cr

QL=107

( 14

Further details are available in reference 59 and in several brief but informative h3 Obviously, this list does not include all the definitions in the ICH document.”’ For example, there is no specific recommendation for robustness; rather, reference is made to ICH documents Q2A and Q2B. In such situations, it is up to the individual organization or company to write its own specification, usually in the form of a standard operating procedure, or SOP. Published reports can be consulted for information about the experiences of

28

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

other laboratories, such as these on robustness of HPLC methods.h4.hsOther laboratory practices such as the frequency of calibration of laboratory balances should be specified in SOPs in accordance with the general principles of the GXPs. The FDA is willing to accept a reasonable specification or practice in cases where there are no specifics in its guidances. Written SOPs are required, and evidence must be available to show that they have been followed. The ICH guidance document also says that all chromatographic analytical procedures should include system suitability (SS) testing and criteria. This term is not in the ICH glossary in the appendix and deserves further discussion. As the name implies, SS is the process of demonstrating that a (chromatographic) system is functioning properly (is suitable) and is ready for use. The USP lists the SS criteria for GC and LC as: the precision (relative standard deviation, RSD) from five injections if the RSD is 2.0% or less, six injections if the RSD is greater than 2.0%; the resolution R,; and the tailing factor, T . More extensive recommendations are given in the CDER Guidance on Validation of Chromatographic Methods,Is which states that: 1. Capacity factor (retention factor) should be greater than 2. 2. RSD of I 1% for n 2 5 is desirable. 3. R , L 2. 4.T I 2. 5 . Plate number L 2000. (These terms are defined in Chapter 2). We have already noted many times that there is no universal set of guidelines and recommendations, and this is also true of those we have just presented. Those of the FDA and the ICH are representative and are probably the most often used and quoted in the United States. Two others that should be read are the ISOZhand the CITAC/Eurachem guides.45 Two new alternative definitions of detection limit and quantitation limit that should be noted are currently under revision by the EPA."" They are slightly different from the ICH/FDA recommendations, and no mention of the latter is included in this new EPA proposal. To quote from the recent EPA document:66 EPA focused its assessment on four sets of concepts that are widely referenced and generally reflect the diversity of concepts advanced to date. These include (1) The EPA MDL [minimum detection limit] and ML [minimum level of quantitation] used under the CWA [Clean Water Act] programs, (2) the Interlaboratory Detection Estimate (IDE) and Interlaboratory Quantitation Estimate (IQE) adopted by ASTM International,

1.6

29

INTERNATIONAL GUlDLlNES FOR ANALYTICAL LABORATORIES

( 3 ) the Limit of Detection (LOD) and Limit of Quantitation (LOQ) adopted by the ACS, and (4) the Critical Value (CRV), Minimum Detectable Value (MDV) and Limit of Quantitation (LOQ) adopted by the IUPAC and the 1SO.

Although the ACS, IUPAC, and I S 0 concepts are functionally similar to EPA’s MDL and ML, these organizations have not developed detailed procedures for calculating detection and quantitation values. Only the EPA and ASTM concepts are supported by detailed procedures for calculating detection and quantitation values. Without such procedural details, the ACS, IUPAC, and I S 0 concepts are unlikely to be useful for establishing detection and quantitation limits in analytical methods for use in CWA programs. Therefore, the discussion below addresses the EPA and ASTM concepts only. The proposed EPA definition of minimum detection limit (MDL) is

MDL=sxt

(1.3)

where s is the standard deviation of the results and t is the Students t value from statistical tables for 99% confidence level and ( n - 1) degrees of freedom. Similarly, the definition of minimum level of quantitation (ML is

ML = 10s

(1.4)

but because the standard deviation, s, may not be readily available, the ML is often calculated from the MDL. Assuming a sample size of 7 (the minimum recommended by EPA), the MDL becomes

MDL = 3.143 X s

(1.5)

and

ML=

( l o MDL)

3.143

=

3.18 x MDL

(1.6)

For larger number of samples, the constant multiplied by the MDL will increase slightly; for example, for n = 10, the multiplier is 3.54. Unfortunately, the symbols and the equations of the EPA and ICH are slightly different so there are no universal international standards for these two parameters, but they are close and one can hope for greater harmonization in the years to come. instrument Qualification

Instrument qualification is the process of making sure an instrument is performing properly. Usually it is accomplished in four stages called (by GAMP 4”): design qualification (DQ), installation qualification (IQ), which

30

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGUMTORY PRACTICES

may be performed by the instrument manufacturer, operational qualification (OQ), and performance qualification (PQ). A certified standard, such as an SRM from NIST, should be used where applicable. Details are on the International Society for Pharmaceutical Engineering (ISPE) web site." Quite a few brief articles have been written on instrument qualification,"". 7" including one specifically on performance qualification of LC system^.^' Once an instrument passes these tests, it is ready for use. In addition, during its use, the instrument should be properly maintained, and at some later time be recalibrated. An SOP should be written to designate the time intervals and the procedures for accomplishing calibration as well as any preventative maintenance that may be required. In some cases, the PQ can serve as the basis for the recalibration procedure. 21 CFR Part 11: Electronic Records and Electronic Signatures

Previously, chromatographic raw data could be easily defined as a piece of chart paper containing a particular chromatogram. The definition is much more complex when the data are digitally recorded in a computer data file. The regulation that is concerned with these issues is referred to as 21 CFR Part 11. That name specifies that the regulation can be found in Part 11 of Scction 21 of the CFR. It deals with the tracking of computerized data, including keeping records of those with access to it and tracking the changes they have made in processing the data. This topic is receiving considerable attention at present in an attempt to arrive at final FDA regulations. Further information about this subject has been published in many journals (see, e.g., references 72 and 73) and the latest FDA (draft) guidance document was published in February 2003.7J It contains the references to the five previous guidance documents on this subject and withdraws the previous guidance and the Compliance Policy Guide 7153.17. A new guidance was issued on September 4, 2003, announcing that the FDA intends to exercise discretion in enforcing some requirements of Part 11 while it reexamines the reg~lation.~' 1.7

FINAL COMMENTS

Although this discussion has focused on the FDA regulations, it should serve to introduce chromatographers to the complexities of the current industrial practices and recommendations. More discussion is included in Chapter 2 (definitions and symbols) and Chapter 9 (quantitative analysis). Additional information about practices in the pharmaceutical industry can be found in references 11 and 12.

REFERENCES

31

REFERENCES 1. J. F. Ryan, Todays Chemist at Work. 2003 12(8), 7. 2. J. M. Miller, A m . Lab. 2000, 32(20), 13-19. 3. C. M. Henry, Chem. Eng. News, 2003, 81(1), 51-57. 4. NlST web site: www.nist.gou/public-uffairs. 5. W. Schulz, Chem. Eng. News 2001, 79(March 51, 29-33. 6. A S T M ; Standardization Basics, ASTM International, West Conshohocken, PA,

1979. 7. S. D. Rasberry, Standard Reference Materials-The First Century, N E T Special publication 260-150,National Institute of Standards and Technology, Gaithersburg, MD, 2003. 8. R. Montgomery, J. Sauenvein, and J. Rumble, Jr., A m . Lab. 2003, 35(1), 26-30. 9. FDA web site: www.fdu.gou. 10. J. M. Miller, LC-GC Europe. 2003, 16(1),38-41. 11. J. M. Miller and J. B. Crowther (eds): Analytical Chemistry in a GMP Enuironment, Wiley, New York, 2000. 12. S. Ahuja and S. Scypinski, Handbook of Modern Pharmaceutical Analysis, Academic, San Diego, 2001. 13. Fed. Reg. 1997, 62,8961. 14. Center for Drug Evaluation and Research, Validation of Chromatographic Methods, Food and Drug Administration, Rockville, MD, 1994. 15. Center for Drug Evaluation and Research, Guidance for Industry, Analytical Procedures and Methods Validation, Food and Drug Administration, Rockville, MD, 2000. 16. Center for Drug Evaluation and Research, Guidance for Industry, Inuestigating Out of Specqtcation ( 0 0 s ) T a t Results for Pharmuceutical Production, Food and Drug Administration, Rockville, MD, 1998. See also M. Swartz and I. Krull, LC-GC No. Am. 2004, 22, 132-136. 17. Center for Drug Evaluation and Research, Guidance for Industry, Q3C Impurities: Residual Soluents, Food and Drug Administration, Rockville, MD, 1997.Originally published in Federal Register 1997, 62,67377. 18. Ordcring information for O R D publications can be found at the web site: www.epa.go u / ord / htm / ordpu bs.h tm . 19. National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161.Phone: 800-553-6847. 20. M. E. Cassinelli and P. F. O'Connor (eds), NIOSH Munual of Analytical Methods, 4th ed. with 3 supplements, NIOSH, Washington, D.C., 1994,publication 94-113. www.cdc.gov /niosh /nmam / nmammenu.htm1. 21. OSHA, Evaluation Guidelines for Air Sampling Methods Utilizing Chromatographic Analysis, OSHA Salt Lake Technical Center, Salt Lake City, UT. Web site: www.osha.gou/dts /sltc /methods/index.html. 22. American Chemical Society, ACS Specijications, Reagent Chemicals, 9th ed., Washington, D.C. 2000.

32

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

23. L. S. Clesceri, A. E. Greenberg, and A. D. Easton (cds), Standard Methods for the Examination of Water and Wastewater, 20th ed., American Public Health Association, Washington, D.C., 1998.

24 See, e.g., http: // www.censa.org / htnil/ wehlinks / standards-relevant-to-ccnsa/ andi-standards. html. 25. T. A. Rooney, Today’s Chemist ut Work 1998, 7(5), 15-17. Also available at the ACS web site: puhs.acs.org/hotartcl/tcaw/98/stan. html. 26. W. Honvitz (ed), Official Methods of Analysis of’AOAC International, 17th ed., 1st rev. (in two volumes), AOAC International, Gaithersburg, MD, 2000. 27. ASTM, Annual Book of A S T M Standards, ASTM International, West Conshohocken, PA, 2002. 28. ISO/IEC 17025:1999, Generul Requirements for the Competence of Testing and Calihrution Laboratories, ISO, Geneva, 1999. 29. www.iso.ch. Search this site for “about ISO” information and online catalog. 30. Solubility Data Series, 78 volumes; Vol. 66 and beyond are published in journals. IUPAC, Research Triangle Park, NC, 1979-2002. 31. IUPAC, IUPAC Handbook, 2002-03, IUPAC, Research Triangle Park, NC, 2002. 32. J. Inczedy, T. Lengyel, and A. M. Ure, Comperidium of Analytical Nomencluture, 3rd ed., Blackwell Scientific, London, 1998. 33. L. S. Ettre, Pure Appl. Chem. 1993, 65, 819-872. See also L. S. Ettre, LC-GC. 1993, 11, 502. 34. J. A. Jonsson, LC-GC 2002, 20, 920. 35. ASTM, Handbook of Standurdizarion, ASTM International, West Conshohockcn, PA, 2003. Can he downloaded from the ASTM web site. 36. I S 0 2000-01/ 1000, The Role of Rqference Materials; Achieving Quulity in Analyticul Chemistry, International Organization for Standardization, Geneva, 2000. (Paper copy may bc requested from I S 0 web site). 37. See, for example, Rcstek Corporation, Bellefonte, PA, www.re~~tekcorp.com. 38. B. King, The Selection and Use of Reference Materials, EEE/RM/O62rev3, Eurachem guide, Portugal, 2002. (Available from Eurachcm web site). 39. United Statcs Pharmacopoeia Convention, US Pharmacopoeia 27/Nationul Formulary 22, U.S. Pharmacopoeia, Rockville, MD, 2004. 40. C. W. Koehler, Mod. Drug Disc.,2002, 5(1 I), 53-57. 41. Federal Register 1995, 60(196), Notices, 53078. 42. L. A. Curie, Anal. Chim. Acta 1999, 391, 103. 43. L. A. Currie, Anal. Chim. Acta 1999, 391, 105-126. 44. Anal. Chim. Actu 1999, 391(2), 102-252. 45. CITAC/Eurachem Guide, Guide to Quality in Analytical Chemistry, Cooperation on International Traceability in Analytical Chemistry, Geel, Belgium, 2002. 46. CITAC/Eurachem Guide, Quuntibing Uncertainty in Analytical Measurement, 2nd ed., Cooperation on International Traceability in Analytical Chemistry, Geel, Belgium, 2000.

REFERENCES

33

47. Eurachcm Guide, The Fitness for Purpose of Analtical Methods, Eurachem, Portugal, English Edition 1.0, 1998. (See web site in Table 1.4).

48. IUPAC, Harmonised Guidelines for the Use of Recovery Information in Analytical Measurement, Technical Report of Orlando Conference, 1996. (Available at Eurachem web site). 49. OECD Principles of Good Laboratory Practice, Organisation for Economic Cooperation and Development, Paris, 1998. 50. J. M. Green, Anal. Chem. 1996, 68, 305A-309A. 5 1. Federal Register 2000, 65, 52776-52777. 52. R. Brown, M. Caphart, P. Faustino, R. Frankewich, J. Gibbs, E. Leutzinger. G. Lunn, L. Ng, R. Rajagopalan, Y. Chiu, and E. Sheinin, LC-GC No. A m . 2001, 19, 74-79.

53. M. E. Swartz and I. S. Krull, Analytical Method Deuelopment and Vulidution, Marcel Dekker, New York, 1997. 54. I. Krull and M. Swartz, LC-GC No. Am. 2001, 19, 604-614. Earlier papers: ibid 1997 15, 534-540; 199 16, 1084-1090; 1999 17, 244-247. 55. J. D. Orr, 1. S. Krull, and M. E. Swartz, LC-GC No. Am. 2003, 21, 626-633. 56. J. D. Orr, I. S. Krull, and M. E. Swartz, LC-GC No. Am. 2003, 21, 1146-1152. 57. LGarcia, M. C. Ortiz, L. Sarabia, C. Vilches, and E. Gredilla, J. Chromatogr. A 2003, 992, 11-27. 58. M. E. Swartz and I. S. Krull, P h u m . Tech. 1998, 22(3), 104-119. 59. W. B. Furman, T. P. Layloff, and R. F. Tetzlaff, J. AOAC Int. 1994, 77, 1314- 1318.

60. 61. 62. 63. 64. 65. 66. 67. 68.

69. 70. 71. 72. 73. 74.

G. K. Webster and C. L. Basel, LC-GC N o . Ani. 2003, 21, 286-294. Federul Register 1996, 61(46), 9315-9319. I. Krull and M. Swartz, LC-GC 1997, 15, 534-540. I. Krull and M. Swartz, LC-GC 1998, 16, Y22-924. P. A. Peters and T. C. Paino, Phurnz. Technol., Anal. Valid. Suppl. 1999, 23, X-14. G. K. Webster, H. Li, W. J. Sanders, C. L. Basel, and G. Huang, J . Chromatogr. Sci. 2001, .?9, 273-279. Federul Register 2003, 68(48), 1 1770- 1 1790. www.phumuce~iticalonline.com/ecommcenter~~/i~s~~e.htt~il. Good Automated Manufacturing Practice, CAMP Forum, 2001, Version 4.0. www.phurmaceicticalonlitic.com G. Karct, R & D Mag. 1998, 40(8), 28-31. K. W. Sigvardson, J. A. Manalo, R. W. Roller, F. Salesi, and D. Wasserman, Pharm. Tech. 2001, 25( lo), 102-108. G. Hall and J. W. D o h , LC-GC Nor. An7. 2002, 20, 842-848. T. Mew and C. Corn, Am. Lab. News Ed. 2002, .?5(1), 17-18. 0 . Lopez, Pharm. Tech. 2002, 26(2), 36-46 and ( 3 ) , 48-62. Part 11, Electronic Records; Electronic Signatures-Scope and Application, FDA Guidunce fiir Industry, Food and Drug Administration, Washington, D.C., Feb.

2003. 75. W. Goebel, P h u m . Technol. Supp. IT 2003, 27, 8.

34

IMPACT OF INDUSTRIAL AND GOVERNMENTAL REGULATORY PRACTICES

GENERAL REFERENCES ON QUALITY IN THE ANALYTICAL LABORATORY Baiulescu, G. E., Stefan, R.-I., and Aboul-Enein, H. Y., Quality and Reliahilit)] in Analytical Chemistry, CRC Press, Boca Raton, FL, 2001. Chung, C. C., Lee, Y. C., Lam, H., and Zhang, X-M, (eds), Phurmuceutical and Medicinal Chemistty y: Analyticul Method Vulidatiori & Instrument Calihrution, Wiley, Hoboken, NJ, 2004. Garfield, F. M., Klesta, E., and Hirsch, J. Quulity Assrirunce Principles for Analytical Luhorutories, 3rd ed., AOAC International, Gaithersburg, MD, 2000. Kenkel, J., A Primer on Quulity in the Analytical Lahoratoq: Lewis, Boca Raton, FL, 2000. Prichard, F. E. (Coordinating author), Qualiy in the Analytical Chemistry Luhorutory, ACOL Series, Wiley, New York, 1997. Ratliff, T. A,, The Lahorutory Qriulity Assiirunce System: A Manuul of Qiiality Procedrires and Forms, 3rd ed., Wiley, Hoboken, NJ, 2003. Taylor, J. K., Quality Assurunce of C'lzernicul Meusiirements, Lewis, Boca Raton, FL, 1987.

Selected Web Sites See also Table 1.4 and reference 10. 1. Code of Federal Regulations (CFR)

2. Environmental Protection Agency (EPA) 3. Food and Drug Administration (FDA) 4. Federal Register (FR)

5. National Institute for Occupational Safety and Health (NIOSH) 6. National Institute of Standards & Technology (NIST) 7. Occupational Safety and Health Administration (OSHA) 8. U.S. government

www.gpoaccess.gov / cfr / index.htm1 www.epa.gou www.jda.gov www.gpoaccess.gov / f r / index.html www.cdc.gou / niosh www.nist.gov www. osha.gov www.firstgov.gov

INTRODUCTION TO CHROMATOGRAPHY Separation methods are an important part of analysis, and chromatography has developed into the premier analytical separation technique. As an example of the highly efficient separations routinely possible, consider the analysis of semivolatile pollutants, including the benzo-fluoranthenes, by gas chromatography using EPA method 8270. Figure 2.1 shows the separation and identification of 98 peaks in about 40 min. Chromatography’s rapid development can be attributed to its relative simplicity and the successful application of theory to practice. Furthermore, when equipped with sensitive detectors, chromatographs are capable of performing highly accurate quantitative analyses. This chapter, along with the next three, introduces the basic principles common to all types of chromatography. Later chapters will cover each of the individual types of chromatography. This unified presentation of the basics seems to be the best way to approach this topic: The concepts are presented first, followed by discussions contrasting the various types of chromatography (beginning in Chapter 6). Some comparisons with other methods of separation are also included. Giddings was the main advocate of the unified approach, as explained in his pioneering book,’ Dynamics of Chromatography. It has had a major impact on the development of the field. Later, he extended his material to all

Chrornutogruphy: Concepts und Contrasts, Second Edition. ISBN 0-471-47207-7 8 2005 John Wiley & Sons, Inc.

By James M. Miller

35

36

INTRODUCTION TO CHROMATOGRAPHY

Figure 2.1. Typical gas chromatographic separation showing the high efficiency of this method for the separation of semivolatilcs by EPA Method 8270. Conditions: 30 m RtxB-SSil column, 35°C programmed t o 33o"C, splitlcss. MS detector. Figurc used with pcrmission of Restek Corporation, Bellefontc, PA.

separation methods,' giving special attention to chromatography. These books are quite theoretical and will not be easy reading for beginners in chromatography, but in 1967 Giddings published a short summary at a lower level? H e has shown convincingly that there is a common theoretical framework across all chromatographic methods and, indeed, throughout all separation methods. Because these principles are very important, they have had a major impact on the organization and presentation of this book.

2.1

BRIEFHISTORY

37

A unified approach has also been followed by Karger et aL4 in their classic Introduction to Separation Science. It is a more advanced book than this one and, although older, can be consulted for additional information. To some chromatographers the term unified chromatography or unified &id chromatography describes a single chromatographic system that could be used for all separations, a sort of universal chromatograph. This suggestion, first made in 1989,s continues to have a significant following, but it is not the meaning of the term as used in this monograph. A special symposium of 12 papers presented at the 1998 meeting of the American Chemical Society (ACS) was devoted to the former meaning, and the papers were published together in the ACS Symposium Series.' It is the best single source of information about the other unified chromatography view, suggesting that the most promising approach to unification is in the field of supercritical fluid chromatography (SFC), which is discussed in Chapter 6.

2.1

A BRIEF HISTORY

It can be argued that modern, high-performance chromatography began with the publication of Martin and James's article on gas chromatography in 19.52.' It is certainly true that their publication on the use of a gas as a mobile phase in the separation of volatile fatty acids initiated the research that has resulted in the widespread use and popularity of chromatography. Although chromatography entered a new phase in the early 1950s, the Russian botanist Tswett is generally referred to as the father of chromatography. His work, originally presented in 1903 and then published in 1906, described the separation of plant pigments by column liquid chromatography, defined the terms, and demonstrated the technique so well that it was used in its original form for about 40 years. The original article is of significant historical interest and serves as a fitting introduction to a discussion of the concepts of chromatography; fortunately, it has been translated into English and is readily available.x Further information about Tswett and his work has been published by Ettre." In the period between 1906 and 1952 there were some developments of importance. For example, the techniques of plane chromatography were developed. Earliest was the use of paper as a plane support, but when thin layers of silica gel were introduced as an alternative in the late 1950s, the field of thin-layer chromatography (TLC) was born and became so popular that it largely replaced the older technique. Column chromatography developments accelerated in the 1940s. Martin and Synge published their Nobel Prize-winning article in which they introduced liquid-liquid (or partition)

38

INTRODUCTION TO CHROMATOGRAPHY

chromatography and the accompanying theory that became known as plate theory.’” It is instructive to note that in their important article, the authors suggested that gas chromatography would be an interesting technique to explore, but no one did so until Martin himself returned to it 10 years later.’ Plate theory was further explored by Craig, who published a paper entitled “Partition Chromatography and Countercurrent Distribution” in 19.50.’’ His countercurrent distribution apparatus gained some popularity but was soon replaced by liquid chromatography for most applications. Chapter 14 contains a further discussion about countercurrent liquid-liquid extraction. Glueckaufs publication in 19.5.5 of an article” titled “The Plate Concept in Column Separations” further indicates the importance of plate theory to early chromatographic development. As an alternative to plate theory, the so-called rate theory came into prominence about the same time. A very important article on rate theory was published by the Dutch workers van Deemter, Zuidenveg, and Klinkenberg.” They described the chromatographic process in packed G C columns in terms of kinetics and elucidated the diffusion and mass transfer processes in gas-liquid chromatography (GLC). A few years later, Giddings published a series of article^'^ on this topic culminating in his 1965 book,’ and the rate theory has since become the backbone of chromatographic theory. The then-new technique, gas chromatography, or GC, wa5 found to be simple and fast and capable of producing separations of volatile materials that were impossible by distillation. Furthermore, theory was found to be rather accurate in predicting optimal operating conditions, and they could be quickly tested. The field exploded! New separations led to new ideas to be tested and vice versa. GC quickly matured. It was natural to attempt to apply the successful results from G C to the older technique of LC, liquid chromatography. Some of the credit for that transfer of technology also belongs to Giddings. In 1963 he published an article entitled “LC with Operating Conditions Analogous to Those of GC,”” setting off a revolution in LC that led to its achieving a level of efficiency comparable to that achieved in GC. Because these new conditions for operating LC columns required high pressure, HPLC was used to describe high-pressure LC. The use of high pressure also produced the expected high-performance separations, so HPLC also denotes high-performance LC. In either case, HPLC is usually used to distinguish between the new, modern mode of operation as opposed to the old Tswett method. As noted earlier, USP uses LC to denote low-pressure or gravity feed LC and HPLC to denote high-pressure LC. Although the ASTM and others have recommended that HPLC not be used,I6 that recommendation has largely been ignored by chromatographers.

2.2

DEFINITIONS AND CLASSIFICATIONS

39

This is only a small glimpse of the historical development of chromatography; it is a fascinating story, and many, more complete accounts have been published by Ettre.”-” Many consider 2003 to mark the end of the first century of chromatography, and the Journal of Chromatography chose that year for the publication of its 1000th volume containing many useful reviews of recent chromatographic developments.”

2.2

DEFINITIONS AND CLASSIFICATIONS

Separation Although chromatography is the most important separation method, there are others; some of them compete with it and some complement it. A brief discussion about separations in general will provide a background for making these comparisons and contrasts. Some of the others are described more fully in Chapter 14. Separation is a familiar word, and it is unlikely that any confusion will arise from its use in reference to chemical analysis. Nevertheless, there is some merit in considering a precise, general definition such as that suggested by Rony:” Separation i s the hypothetical condition where there i s complete isolation, by rn separate macroscopic regions, of each of the rn chemical components which comprise a mixture. In other words, the goal of any separation process i s to isolate the rn chemical components, in their pure forms, into rn separate vessels, such as glass vials or polyethylene bottles.

The adjective hypothetical is used for two reasons. In the first place, it is theoretically impossible to accomplish the complete separation of the components of a mixture. Consider, for example, the process of analyte transfer in a liquid-liquid extraction. If a single extraction removes most of the analyte from phase 1 into phase 2, say 9096, each successive step will also remove 90% of the analyte that remained after the previous step. Thus 10% of the analyte from the previous step remains in the first phase and subsequent steps can remove only a fraction (0.9) of it. Clearly, all of the analyte cannot be removed; of course, when the fraction removed approaches 1, say 0.9999, we consider that virtually all of it has been removed. The second reason for using the term hypothetical is that the separated components often are not actually isolated into vessels but rather are detected and their presence recorded (on chart paper or in a computer data file). Most separation methods employ two phases. In chromatography, they are called the mobile phase and the stationary phase, as discussed in the next

40

INTRODUCTION TO CHROMATOGRAPHY

Table 2.1 Properties of Some Separation Methods

Name Gas-liquid chromatography (GLC) Gas-solid chromatography (GSC) Liquid-liquid chromatography (LLC) Liquid-solid chromatography (LSC) Liquid-Liquid extraction (LLE) Supercritical fluid extraction (SFE) Solid-phase extraction (SPE) Solid-phase microextraction (SPME) Zone electrophoresis (CZE) Capillary electrochromatography (CEC) Field-flow fractionation (FFF) Distillation Precipitation Zone melting Dialysis Filtration Size exclusion chromatography (SEC)

Number of Phases

Type of Process

2 2 2 2 2 2 2 2 I 2 1 2 2 2 2 2 2

Partition (absorption) Adsorption Partition Adsorption Partition Partition Adsorption Adsorption and partition Electromigration Partition and electromigration Electromigration Change of state Change of state Change of state Sieving (size exclusion) Sieving Size exclusion

section. In liquid-liquid extraction, they are called the extractant and the raffinate. T h e extractant in a multistage liquid-liquid extraction has a function similar to t h e mobile phase in liquid chromatography, and comparisons of these two techniques will help in understanding both of them. In distillation, the second phase (a vapor) is formed from the first phase (a liquid) as the latter is heated. This process differs from extraction and chromatography where the second phase is added. Obviously, the principles of distillation are different from those for chromatography. However, as has been noted, much of the development of GC was carried out by chemists who were accustomed to making their separations by distillation, so they labeled the chromatographic process with distillation terms such as theoretical plutes. In general, this was not helpful and the terminology has been largely revised. Table 2.1 lists some common separation methods providing a convenient reference for contrasting them with chromatography. A few separation methods employ only o n e phase; examples are field-flow fractionation and some types of electrophoresis. In these cases, a second force such as a voltage gradient is required to effect a separation. In fact, the list of processes in the third column of the table can be classified into three different categories, which will help organize variations in methods such as secondary forces. O n e can consider that all separation techniques depend upon a differential distribution of the analytes in a sample. T h e types of differential distribution a r e based on one of three different processes-different states,

2.2

phases, or environments-as

DEFINITIONS AND CLASSIFICATIONS

41

follows:

1. Different chemical states of the same matter, also called phase equilibria or changes in state. For example, the distribution between liquid and vapor states is called distillation. Other separation methods in this category are sublimation, crystallization, zone melting (refining), and precipitation. 2. Different chemical phases, also called phase distribution equilibria. This category includes most of the chromatographic processes listed in Table 2.1 as absorption (partition) or adsorption as well as liquid-liquid extraction (LLE), solid-phase extraction (SPE), supercritical fluid extraction (SFE), and solid-phase microextraction (SPME). 3. Different chemical environments or different locations. These separations are based on differential rates of migration of analytes under the influence of a field. It is a nonequilibrium process, unlike types 1 and 2, and the analytes end up in different places within the field. Some examples are: Electric field: electrophoresis (of ions) and field-flow fractionation (FFF) Gravitational field: centrifugation and filtration Thermal field: thermal diffusion Membrane (semipermeable): dialysis, osmosis, and ultrafiltration Chromatography

In chromatography, one phase is held immobile or stationary and the other one (the mobile phase) is passed over it, as noted above. The designations GC and LC indicate the physical state of the mobile phase, a gas or a liquid, respectively. Subclassifications can be made by naming the mobile followed by the stationary phases; thus we have gas-solid (GS), gas-liquid (GL), liquid-liquid (LL), and liquid-solid (LS) chromatography. As shown in Figure 2.2, supercritical fluids can also be used as mobile phases, and these techniques have been named supercritical fluid chromatography (SFC). Strictly speaking, in current practice, many stationary phases nominally listed as liquids are really chemically bonded to (or chemically polymerized and cross-linked to) a solid support [as in bonded-phase (BP) HPLC] or to the column wall (as in capillary GC). Consequently, the terms GLC and LLC are not used much, but rather open tubular (OT) GC and BP-HPLC. Other naming systems have also become popular, and Figure 2.2 shows a complete classification scheme for liquid-solid chromatography (LSC), listing the popular names and abbreviations. Included in the classification scheme are the two configurations of the chromatographic bed, a column and a

Liauid

Solid

I u m n

/ \ Adsorbent MoI sieve

Configuration Column Colu\

Stationary phase

t

Liquid

Figure 2.2. Classification of chromatographic techniques.

Supercritical Fluid WC)

Column

Solid

v

Liquid

Supercritical fluid

I

Cap Electrochrom. (CEC;

tlectrophoresis

2.2

DEFINITIONS AND CLASSIFICATIONS

43

planar surface. Also, the combination of two different techniques, bondedphase LC (BPC) and capillary zone electrophoresis (CZE), has produced capillary clectrochromatography (CEC), shown at the far right in Figure 2.2. A further classification of the GC methods on the left of the figure, according to the type of column, not the type of phase, is included in Chapter 7. The IUPAC definition of chromatography is: Chromatography is a physical method of separation in which the components to be separated are distributed between two phases, one of which is stationary (stationary phase) while the othcr (thc mobilc phase) moves in a definite direction. Elution chromatography is a procedure in which the mobile phase is continually passed through or along the chromatographic bed and the sample is fed into the system in a definite slug."'

In these discussions, M will be used to denote the mobile phase, and S the stationary phase, and elution will be the type of process unless otherwise indicated. As the mobile phase passes over and through the stationary phase, the components of the mixture ideally equilibrate o r differentially partition between the two phases, resulting in different migration rates through the system. Alternatively, we could say that the various components of the mixture are retarded in their passage through the system in proportion to their interaction with (sorption on) the sorbent bed. At any given time, a particular analyte molecule is either in the mobile phase, moving along at its velocity, o r in the stationary phase and not moving at all in the downstream direction. T h e sorption-desorption process occurs many times as the molecule moves through the bed, and the time required to do so depends mainly on the proportion of time it is sorbed and held immobile. A separation is effected if the various components emerge from the bed at different times, which are called retention times. T h e process is depicted in Figure 2.3. The status of the separation is shown at five different times-five snapshots of the separation. In the first one, the mixture of analytes A and B is introduced to the bed in as narrow a zone as possible. T h e mobile phase, flowing from left to right, carries them along the bed. Since all the molecules of a particular analyte d o not encounter the exact same local environment in the stationary and mobile phases, the peak for that analyte will have a finite width in the chromatogram. Note that the width of the peaks increases with the length of time they remain in the bed. Because analyte A has a greater affinity for the mobile phase, it spends more time in the mobile phase and travels faster and elutes before analyte B. Thus A and B become separated. Another, less popular form of chromatography is controlled by displacement. The sample is pushed through the system by displacing it from the

44

INTRODUCTION TO CHROMATOGRAPHY

Direction of mobile-phase flow I

I

Detector

_t

Chromatogram

I

I

Concentration of solute in mobile phase

Concentration of solute in stationary phase B

A

0)

.-E

+ B

0

_ j

1

Fraction of bed length Figure 2.3. Schematic reprcscntation o f the chromatographic process.

stationary bed with other sample components and a strong mobile phase. Displacement chromatography, sometimes used in LSC and in preparative LC,’” has been briefly discussed recently,” but it will not be considered further in this text. The chromatographic process we have defined is also known as zonal or hatch chromatography because the sample is applied to the system all at once in one narrow zone. By contrast, the sample can be applied continuously during a run; this process is called fiontaf analysis. One application to immunoaffinity chromatography has been published recently,32 but it will not be discussed further here because of its limited use. The mode of interaction between the sample components and the two phases can be classified into two types, although many separation processes are combinations of both. If the sample is attracted to the surfaces of the phases, commonly to the surface of a solid stationary phase, the process is called adsorption. Alternatively, if the sample diffuses into the interior of the stationary phase-for example, into the bulk of a stationary liquidchromatographers call the process partition. Actually, absorption seems to be a better name for this process because we can then speak of sorption as the

2.2

ABsorption

DEFINITIONS AND CLASSIFICATIONS

45

ADsorption

Figure 2.4. Ditfercnce between absorption and adwrption

general process and add the prefixes a h or ad when we want to be more specific. For this reason, the terms absorption and adsorption will be used in this monograph even though purtitiorz is the term recommended by the IUPAC. A comic illustration of the terms is shown in Figure 2.4.. Chromatographic Symbols

In 1993, the IUPAC in cooperation with many other agencies published a codified list of chromatographic terms, symbols, and definitions.’” The most important ones are listed in Table 2.2, along with some that were previously used and should now be discontinued. Since the IUPAC list has the widest use and greatest acceptance, these terms and symbols will be used in this book (although, as mentioned, we prefer absorption to partition). Unfortunately, not all international groups have adopted the IUPAC list, and the U S . FDA (and the USP) is among them, thus causing some confusion to remain among pharmaceutical chromatographers. Included along the right edge of Figure 2.3 is a plot of time versus detector signal. This is the common way of presenting chromatographic data and is called a chromatogram. Two sample components o r peaks are shown in Figure 2.3, and it is clear that they have been separated, which is the objective of chromatography. For purposes of discussion, a simpler chromatogram is given in Figure 2.5, which shows only one sample component plus a small peak representing a nonretained component (which may not always appear). This figure will be used to illustrate some of the chromatographic definitions and symbols. Most chromatographs are operated with a constant flow ( F ) of mobile phase unless the flow is intentionally being changed or programmed. Conse-

46

INTRODUCTION TO CHROMATOGRAPHY

Table 2.2

Chromatographic Terms and Symbols

Symbol and Name Recommended by the IUPAC'"

Other Symbols and Names in Use

Distribution constant

Partition coefficient; distribution coefficient Capacity factor; capacity ratio; partition ratio Theoretical plate number; No. of theoretical plates Height equivalent to one Theoretical plate Retention ratio

Retention factor Plate number Plate height Retardation factor (in columns) Peak resolution Separation factor Total retention time Total retention volume Hold-up volume

I

I

1

I

I

I

I

I I

I

I-

Selectivity; solvent efficiency Elution time Elution volume Volume of the mobile phase; void volume; dead volume

I

I

k I

I

I I

I

I

G.

F

Time or volume of mobile phase

+

Figure 2.5. Typical chromatogram.

quently, the x axis of the chromatogram can be labeled as time ( t > or as volume ( V )since

V=txF Thus, I/ represents the volume of mobile phase that flowed during a specified operating time t . If we wish to designate the time required for a component to elute from the chromatographic system, the retention time, the

2.2

DEFINITIONS AND CLASSIFICATIONS

47

symbols t and V are designated with a subscript R:

In making an actual measurement of retention time or retention volume from a chromatogram produced by a recording device, one can measure the distance on the chromatogram from the start to the maximum of the peak of interest. Thus, VR and t,, can be represented as the distance from 0 to B as indicated in Figure 2.5. Distance on a chromatogram can be converted to time by multiplying it by the recording speed and then to volume by multiplying the time by the flow rate. When constant flow is assumed, retention time and volume can be used interchangeably, and both are shown on the abscissa in Figure 2.5. Part of the time an analyte spends in the chromatographic system is the time required to go through the interstitial space in the column (assuming other volumes in connections and the like are negligible), and part is caused by the time it spends in the stationary phase, not moving downstream. Thus, the total time o r volume can be broken down into two parts: V,

=

+

VM K , V ,

(2.3)

where VM represents the mobile-phase volume, Vs is the stationary-phase volume, and K,. is a partition coefficient called (by IUPAC) the distribution constant. Strictly speaking, this equation is only valid when V, represents a stationary liquid phase, which, we have already noted, is not often the case in most columns in use today. Also, in the past, V , has sometimes been called the dead volume in the system, but the use of that term is being discouraged because of its ambiguity.33 Rigorous derivations of this equation have been p ~ b l i s h e d , ’ . ~but . ~ ~the significance of the equation should be obvious without it; that is, the total volume of mobile phase (V,) required to elute an analyte is composed of two parts: the interspacial volume in the column, which is occupied by mobile phase and through which every analyte must pass (V,), and the mobile phase which flows while the analyte is held immobile (equal to K,.V,). The latter contribution (the length of time the analyte is immobile) is in turn determined by the amount (volume, V s ) of stationary phase and the tendency of the analyte to sorb in the stationary phase as measured by its distribution constant K c . To recap, the equation shows that there are only two things a given analyte molecule can do: move down the column with the mobile phase or sorb in or on the stationary phase. Before looking at the distribution constant more closely, note the small peak at A in Figure 2.5. It represents the time required for an analyte to pass through the system without being sorbed, and thus it measures the mobile-

48

INTRODUCTION TO CHROMATOGRAPHY

phase volume VM (again, assuming negligible volumes from extra-column connections). The IUPAC has selected the name hold-up volume for V,, defined as “the volume of the mobile phase (MP) required to elute the unretained compound from the chromatographic column and reported at column temperature and ambient pres~ure.’”~ The analogous time parameter is hold-up time, t,, “the time required for the MP to pass through the chromatographic column.”33 Because the original terms2’ were found to be misleading or superfluous, the IUPAC published these more precise, new definitions, and included some new terms. Although not discussed in this text, the comments about the concept of hold-up volume in the update3’ have helped to clarify it. In GC, air or methane is often used as the unretained component; in HPLC, there is no single simple marker, but a slight shift in the baseline is sometimes observed, depending on the solvent and the detector. The exact measurement of V,, especially in HPLC, can be difficult, but we will not digress to discuss that topic. The retention volume (or time) that has the mobile-phase volume subtracted out is of interest for theoretical work:

It is called the adjusted retention volume (or time), and it is designated with a superscript prime. It is also shown in Figure 2.5. The distribution constant K , is defined as the equilibrium concentration of an analyte (A) in the stationary phase divided by its equilibrium concentration in the mobile phase:

It is this ratio that controls the rate of migration of A. That is, as analyte A proceeds through the system at a given temperature, it partitions between the two phases and is retained in the system in proportion to its affinity for the stationary phase. The peaks in Figure 2.3 are drawn to illustrate this partitioning by showing an analyte zone as two peaks, one in the mobile phase (above each line) and one in the stationary phase (below each line). Analyte A is shown moving faster down the column than analyte B because molecules of A spend less time in the stationary phase, as indicated by the smaller peak in that phase; that is, A has a smaller partition coefficient K , than B. The use of a classical equilibrium constant, K,, in chromatographic theory implies that the system can be assumed to operate at equilibrium, even though chromatography is clearly a dynamic system. Ideally, the operating

2.2

DEFINITIONS AND CLASSIFICATIONS

49

parameters are such that the system is not far from equilibrium and the use of an equilibrium constant (distribution constant) is valid. The only place true equilibrium exists is near the apex of the chromatographic peak. Appendix B includes a calculation of distribution constants from HPLC data and compares them with true equilibrium values from a similar extraction system. The closeness of the values further serves to substantiate the equilibrium assumption. Equation (2.5) is somewhat simplified; the true thermodynamic partition coefficient would be the quotient of analyte actiuities, not concentrations. Furthermore, the equation assumes that analyte A is present in only one form (one molecular structure or ion). When this is not realized in practice, a more complex equilibrium constant must be used. Another assumption that is usually not stated is that the analytes do not interact with each other; that is, molecules of analyte A pass through the chromatographic system as though no other analytes were present. This assumption is reasonable because of the low concentrations at which analytes are present and because they are increasingly separated from each other as they pass through the system. In making use of the distribution constant in chromatography, it is useful to break it down into its two parts -p, the phase volume ratio, and k , the retention factor: K, =pk (2.6) where p = - VM (2.7) VS and (mass of A)s k= (mass of A) The retention factor used to be called the capacity factor or parfition ratio, and k' was also used as its symbol. Unfortunately, many authors continue to use the old terms and symbols. By combining Eqs. (2.4), (2.6), and (2.7), Eq. (2.9) can be derived. It serves as another definition of k , one that can easily be measured (Fig. 2.5):

Another chromatographic parameter is the retardation factor, R. It is the relative average speed u of an analyte through a chromatographic bed compared to the average mobile phase speed or velocity u :

R = -U U

(2.10)

50

INTRODUCTION TO CHROMATOGRAPHY

It will always be equal to or less than 1, and it expresses the fractional rate at which an analyte is moving. It also represents the fraction of molecules of a given analyte in the mobile phase at any given time and, alternatively, the fraction of time an average analyte molecule spends in the mobile phase as it travels through the system. For planar chromatography (such as TLC), the retardation factor has the same meaning, but it is calculated differently and its symbol is R , rather than R. More information about planar methods is given in Chapters 6 and 11. Each of the velocities in Eq. (2.10) can be defined and measured according to the length L of the chromatographic system: (2.11) and u=-

L tM

(2.12)

A combination of these three equations shows that (2.13) Furthermore, substituting Eq. (2.13) into Eq. (2.3) (2.14) -

~

1 1+k

(2.15)

Equation (2.15) shows the relationship between the two important chromatographic parameters, k and R. Most chromatographers use k to express the extent of retention of an analyte on a given column, but R could also be used. As described above, R has more physical relevance to the behavior of an analyte, and it has the advantage of allowing comparisons between columns and planar forms of chromatography (see Chapter 6). Table 2.3 contains some relative values of k and R. For example, a retention factor k of 3 , which is common in chromatographic practice, corresponds to a retardation factor R of 0.25. Thus, we can say that this analyte spent 25% of its time in the mobile phase while it passed through the column and 75% in the stationary phase. To get another perspective on the meaning of retention factor, we can compare chromatography with liquid-liquid extraction (LLE), which is perhaps more familiar to the novice chromatographer. The retardation factor in

2.2

Table 2.3

1 2 3 4

9 49

DEFINITIONS AND CLASSIFICATIONS

51

Relationships between k and R

0.25 0.44 0.56 0.64 0.81 0.96

0.5 0.33 0.25 0.20 0.10 0.02

0.25 0.22 0.19 0.16 0.09 0.0196

chromatography, R, is similar in concept to the fraction extructed in LLE. So, in the example above, only 25% of the analyte in question is in the mobile phase [“extracted” by mobile phase (MP)] at any given time, and most of the analyte (75%) is still in the stationary phase (SP). More information about LLE can be found in Chapter 14. Normal Distribution

The shapes of all the peaks in the chromatograms shown thus far are symmetrical, approximating the normal or Gaussian shape. This is the ideal shape, but it is not always achieved in practice. Theoretically, the Gaussian shape is closely approached if the analyte has undergone a sufficiently large number of sorptions and desorptions, as is the case for most peaks with retention factors of about 1 or greater. In fact, an asymmetrical peak is usually evidence that some undesirable interaction is taking place in the system and that remedial action should be undertaken to find it and change it. For all the theoretical discussions in this book, the Gaussian shape will be assumed, thus allowing us to draw some conclusions about the peaks and to use some standard nomenclature and symbols to describe chromatographic peaks. Remember that a peak represents the frequency distribution of all molecules of a particular analyte as they move as a group through the system or as they are detected as they exit from the system. The “average” molecule is found at the center of the distribution, at the position of the peak maximum, and it is this average position that is used to characterize the particular analyte. The familiar equation for the normal distribution is

Y

=

1 -P& [

1 x-x

-

[

j]

77

2

(2.16)

where y is the dependent variable, x is the independent variable, X is the average of a large number of x’s, and (T is the standard deviation. When this

52

INTRODUCTION TO CHROMATOGRAPHY

i

I

a

\

\

Figure 2.6. Normal distribution.

equation is used to represent a chromatographic zone or band, y represents concentration, x represents a retention parameter like retention volume, and a represents a peak width parameter. In the analysis of peaks resulting from a separation, it will be most useful to express the variable x in units of standard deviation. Hence, for our purposes, Eq. (2.16) can be written as (2.17) This equation is plotted in Figure 2.6. Also shown as broken lines are tangents to the points of inflection that occur at 60.7% of the peak height. Where they intersect the baseline, they cut off the distance w,, known as the peak width at the baseline. It can be seen in Figure 2.6, that w b has a value of 4a ( k 2 a ) . Consequently, cr, the standard deviation, is also called the quarterpeak width (at the base). Note also that the width at 60.7% of the peak height (the inflection point) is 2a. At 50% of the peak height the peak width is 2 . 3 5 4 ~The . latter is called the peak width at halfheight, w,, .

DEFINITIONS AND CLASSIFICATIONS

2.2

I G & -p

-rn

s .n

E &

c.

6

I

I

53

I ‘

R

-

I

I

I

I

I

x

Time

+

Figure 2.7. Figure used to define N , the plate number ( x

=

nonretained component).

Other Terms

Plate Number The most common measure of the efficiency of a chromatographic system is the plate number N.Because the concept originated from the analogy with distillation, it was originally called the number of theoretical plates contained in the chromatographic column (system). This is not a useful analogy because a chromatographic column does not contain “plates,” but the terminology has remained, and the original definition has persisted:

(2.18) The parameters t , and (T number N is dimensionless express (T in terms of peak presented in the last section. so N becomes

must be measured in the same units, so the (see Fig. 2.7). For Gaussian peaks, we can width since we know the relationships as just For example, the base width w,,is equal to 4 u ,

(2.19) corresponds to 2.354a, and N equals Similarly, w,,

N = 5.54

2) 2

(2.20)

Refer to Figure 2.6 for a summary of these relationships. If the peak is not symmetrical, different values will be calculated for N because the width measurements will not follow the predicted Gaussian

54

INTRODUCTION TO CHROMATOGRAPHY

distribution. In general, for asymmetrical peaks, N increases the higher up on the peak the width is measured. Most computers will make the computation at any of several positions on a peak, so one must be very cautious in assigning an N value to an asymmetric peak. Another method of calculating the plate number uses statistical moments. A summary of moment analysis is: Zeroth moment measures peak area. First moment measures peak mean and hence t , . Second moment measures peak variance (and a ) and hence peak width. Third moment measures peak asymmetry (skew). Higher moments can also be calculated, but these four are the most important. From them, one can also calculate skew and excess. The only practical way to make these measurements is with a personal computer (PC) and appropriate data collection software. Further details are provided in several works.z5 " The plate number is a measure of the relative peak broadening (w)that has occurred while the analyte passed through the system (in time t,). As we will see, peaks broaden (w increases) as the retention time increases, which is the reason peak width alone is not sufficient to specify the efficiency of the system. Furthermore, according to theory, N should increase slightly as k increases, but the effect is usually not large; consequently, column efficiencies ( N ) are usually quoted for a given column, without regard to the particular analyte and its k value, even though this practice is really not accurate. A related measure of system efficiency is the cffcctiue plate number, N,,,: ~

(2.21) in which the adjusted retention time is used instead of the retention time (see Fig. 2.5). We have already noted that the adjusted retention time has more theoretical significance than the retention time, and consequently the effective plate number is often a better parameter for use in comparing chromatographic columns, particularly comparing packed columns to open tubular columns. By combining equations already presented, it can be shown that the effective plate number is related to N by (2.22)

2.2

DEFINITIONS AND CLASSIFICATIONS

55

The effective plate number will increase as k increases (see Table 2.3 for some values), and it will approach N at high k values where V , is no longer of significant size compared to V,. As already mentioned, the number of theoretical plates is a term also used in distillation, but it is important to note that comparisons of efficiencies of the two techniques cannot be made by comparing plate numbers. It takes more chromatographic plates to achieve a given separation by G C than it does to achieve the same separation by di~tillation.~ Plate Height The plate number depends on the length of the column, making comparisons among columns difficult unless all are of the same length. Another related parameter that removes this dependence is the plate height H :

H = -L

N

(2.23)

where L is the total column length, so H can be thought of as the length of column that contains one plate. Clearly, N and H are inverse to each other, and H is a measure of efficiency that is independent of total length. It has the units of length, usually centimeters or millimeters, and, like N, it originated in distillation. In fact, it is sometimes referred to by the full name used in distillation-HETP, or height equivalent to a theoretical plate. But, again, it should be emphasized that the column does not contain plates, and the use of these two terms that resulted from their historical development should be discontinued. Plate hcight is onc of the fundamental chromatographic parameters and should also be defined from that perspective. That discussion is deferred until Chapter 3. Asymmetry and Tailing Several measures of asymmetry, such as the one shown in Figure 2.8, have been devised by chromatographers for asymmetric peaks. One is called the aJymmetric ratio or tailing factor (TF);

TF=

b a

-

(2.24)

where a and b are measured at 10% of the height of the peak as shown. A symmetrical peak will have a value of 1 and tailed peaks a value of greater than 1. A fronted peak (one with a leading edge and a > b ) will have a tailing factor of less than 1. Since it is not common practice to calculate a fronting factor, it is recommended that TF be used to cover both situations and that the limits on TF include both upper and lower values.

56

INTRODUCTION TO CHROMATOGRAPHY

Figure 2.8. Figure defining asymmetric ratio or tailing factor.

Since this definition is not included in the IUPAC list of terms and symbols, some confusion has arisen because another, different definition is also in common use. It is the one recommended by USP:

T = Wo.os/2f

(2.25)

where w,,,05 is the width of the peak at 5% of the height and 2f is twice the front part of the peak as shown in Figure 2.9. The measurements of peak widths or partial peak widths for Eqs. (2.24) and (2.25) are made to the actual peak lines, not to tangents drawn to the points of inflection as was done for the definition of N in Eq. (2.19). Note that the two definitions [Eqs. (2.24) and (2.2511 differ in two ways: the ratios (definitions) differ and the position of measurement on the peak differs. The two values always differ (the 10% value being greater). Most computer software used with chromatographic systems is capable of making both of these calculations. Therefore, one must be careful to designate which one is to be used in a given laboratory. Failure to do so could result in the wrong definition being used for the calculation, and an erroneous value being used for a system suitability test resulting in an incorrect conclusion about the test.

2.2

DEFINITIONS AND CLASSIFICATIONS

57

peak maximum/ T = - w0.05 2f

Figure 2.9. Figure defining USP tailing factor. Reprinted with permission from the U.S. Pharmacopeia 24,"ational Formularly 19. All rights reserved. Copyright 1999. The United States Pharmacopcial Convention, Inc.

Several authors"".') have expressed the opinion that the measurement at 10% of the peak height is better that the measurement at 5%, but no agreement has yet been arrived at. An IUPAC task force has the issue under consideration,4' and a recent study has proposed a new method for determining peak asymmetry.4z Calculation of peak number, N, for asymmetric peaks is also a concern. Foley and Dorsey2' discussed the various figures of merit for ideal and skewed peaks and proposed a new empirical equation:

(2.26) where TF is the tailing factor at 10% of the peak height [Eq. (2.24); equal to b / a ] , and wO,,,, is the peak width at 10% of the peak height and equal to h + a. Others have subsequently confirmed this equation to be much better than Eq. (2.19) for the calculation of asymmetric peaks.43 The best method is the use of statistical moments43 that were discussed earlier (see reference 35). Unfortunately, suppliers of LC columns (in 1984) did not use these superior methods in calculating the plate numbers of columns they sold.43

58

INTRODUCTION TO CHROMATOGRAPHY

rd7

Figure 2.1 0. Two nearly resolved peaks illustrating the definition of resolution.

Resolution A better measure of the efficiency of a chromatographic system is resolution, R,y.It defines the degree of separation of two analytes or peaks:

where d is the distance between the peak maxima and w h is the width of each peak at the base, as shown in Figure 2.10. The larger the resolution, the better the separation; a value of 1.0 is shown in the figure representing about 98% resolution. If one chooses to measure the peak width at half height, w h , Eq. (2.26) becomes (2.28) Recently, the USP has stated that it will accept either equation [(2.27) or (2.28)] whereas only Eq. (2.27) was acceptable before. A resolution value of 1.0 can be easily estimated if it is assumed that the widths of the two peaks were equal as expected according to theory. If w, = w B , R , = 2d/2w = d / w . Since the tangents to the peaks (dotted lines) are just touching, and since w = 4a for each peak, d must also equal 4a ( 2 a from A plus 2a from B) and R , = 4a/4a= 1 . Thus, for estimating resolution, Eq. (2.29) is often satisfactory: (2.29) since the peak widths are usually the same. A resolution of about 1.5 is necessary for complete separation, and values less than 1.0 represent poorer separations than that shown in Figure 2.10.

2.2

RS

1/1

1/4

DEFINITIONS AND CLASSIFICATIONS

59

1/16

Figure 2.1 1. Comparison o f resolution valucs for peaks of equal and nonequal heights. Reproduced from the Journal of Chrornatogruphic Science by pcrmission of Prcston Publications, Inc.

Strictly speaking, Eq. (2.27) is only valid when both peaks have the same height, as is shown in Figure 2.10. A practical approach for calculating the resolution of nonequal peaks has been provided by Snyder.44 His work includes computer-drawn pairs of peaks for a variety of resolutions and peak height ratios from 1 : 1 to 128: 1. Some of them are shown in Figure 2.11. Snyder's recommendation is to compare a given separation with those he has provided and to choose the best match, assigning that value of resolution to the chromatogram in question. This practice has become accepted by many, although others choose to ignore the effect on resolution by analytes of unequal concentration. Other possibilities have been suggested by Ca~-le,~' Kargerd6 (for peaks with different widths), and Bly4' (for size exclusion chromatograms). Resolution is also affected by peak tailing (asymmetry), and, alternatively, a poorly resolved pair of peaks of unequal height can appear as one tailed peak. Dolan has addressed these problems in his series on LC troubleshooting.4x.4')

60

INTRODUCTION TO CHROMATOGRAPHY

Time

Figure 2.12. A hypothetical separation illustrating the concept of peak capacity.

Table 2.4

Calculated Peak Capacities for Different Chromatographic Techniques

Peak Capacity Plate No.

GC

LC

SEC

100 400 1000 2500 10000

11 21

7

3

13

33

20

51 101

31

11

61

21

5 7

Source: J. C. Giddings, Anal. Clzern. 1967, 39, 1027. Copyright 1967, American Chemical Society. Reprinted with permission.

Peak Capacity A final measure of column efficiency is the peak capacity,

or the number of peaks that can be resolved ( R , = 1) by a given system in a given time ( t R or VR). Giddings'" introduced it in 1967 and used it to compare the potential separating capabilities of the various chromatographic modes. Figure 2.12 shows a series of analytes represented by triangles and a peak capacity of six, and Table 2.4 lists the values he obtained for three chromatographic systems-GC, LC, and SEC (size exclusion chromatograPhY1. A more recent work reports a comparative study of equivalent capillary columns in the three techniques, GC, LC, and capillary electrochromatography (CEC)." This work contains a good summary of the developments in measuring peak capacity following Giddings' original work. The conclusions from an empirical study of 10 alkylbenzenes on several capillary columns of about 36,000 plates are summarized in Table 2.5. The GC columns were

2.2

DEFINITIONS AND CLASSIFICATIONS

61

Table 2.5 Comparison of Peak Capacities of GC Capillary Columns and HPLC and CEC Packed Columns

Technique GC

Peak Capacity 248

N

Technique

39,000

-

Columns packed with 3.0-pm particles CEC HPLC

137 102

Peak Capacity

Columns packed with 1.S-pm particles CEC HPLC

58,000 29,000

139 148

Source: From reference 5 1.

open tubular [ L = 10 and 13 m; inner diameter (i.d.) = 200 pm] and the HPLC and CEC columns were packed with 1.5- and 3.0-pm particles. As expected, the GC column had the highest peak capacity, and CEC was slightly higher than HPLC using the 3-pm particles; with the smaller particles, the peak capacities were about the same. For more conclusions regarding ways to increase peak capacity, see the original study. A similar concept was suggested by Kaiser,” who calls his parameter a separation number (SN), or Trennzahl (TZ) in German. It is the number of analytes that can be resolved between two consecutive members of the paraffin homologous series x and x 1. It can be calculated by Eq. (2.30):

+

TZ =

‘R(x+ I ) wh(x+ I )

-

‘R(x)

+ Wh(.r)

-1

(2.30)

This term is very similar to resolution, which is equal to AtR/1.7w,, when defined with similar symbols. Substituting it into Eq. (2.301, we get T Z = 0.85R, - 1

(2.31)

which can be rearranged to

R , = 1.177TZ

+ 1.177

(2.32)

Thus, baseline resolution of 1.5 is equivalent to a TZ value of 0.275, and a TZ value of 1.0 requires an R , of 2.35. The TZ numbers can be used for programmed temperature operation in GC and gradient elution in LC, conditions under which plate numbers are not considered to be valid. Regardless of the operating conditions, a TZ number of 12, for example, means that 12 analytes eluting adjacent to each other ( R , = 1) can be resolved between two consecutive paraffins in that region of the chromatogram. Figure 2.13, taken from Freeman,s3 compares

62

INTRODUCTION TO CHROMATOGRAPHY

50.000

-

'

3 .-

35.000

-b

30,000

0

.-mu

8 c

-8

s m

E

i

15

25.000 20.000

TZ

10

15.000

5

10,000 5.000 I c8

I

%

I

1

I

I

ClO

cll

c12

c13

Carbon number

Figure 2.13. Plate number and separation number vs. carbon number for 10-m x 0 .2s--mm OT columnI at 100°C. 0 Copyright 1981 Hewlett-Packard. Rcproduced with permission.

N,

and TZ for some paraffins on an open tubular G C column. For further discussion of peak capacity, see the study by Ettre.s4

Separation Factor Separations can be effected by differences in partition coefficients as well as by the efficiencies of the columns in which they are run. Thus, another useful thermodynamic measure of the separability of two analytes is the ratio of their distribution constants on a given column. This ratio is called the separation factor, a :

(2.33) It is defined so that its numerical value is > 1.0, which would mean that analyte B is retained longer on the system than analyte A. Chapter 5 contains more discussion of the three parameters, a , N,and k .

2.2

Summary of Important Chromatographic Equations and Definitions

Table 2.6

2. K , . = k p

5. V ,

=

VM + KCVs

6. VN = K ( V s

12. R(l - R ) = -

k

(k

+

H 16. h = dP

19. T Z =

DEFINITIONS AND CLASSIFICATIONS

tK(r+l) -tR(i)

wh(.r+l) + W h ( r )

-1

63

64

INTRODUCTION TO CHROMATOGRAPHY

2.3 SUMMARY A large number of terms, symbols, and equations were given in this chapter. The equations are gathered together in Table 2.6, along with a few others that will be introduced in later chapters. As commonly used, some symbols are slightly different for GC and LC, but this should not diminish the value of the table. Appendix B contains a list of symbols and acronyms used in this text. Appendix B also contains two chromatograms and all the operating conditions needed to make calculations of the most important parameters. It is recommended that these calculations be performed in order to gain a better understanding of their meaning. A review published by MeyerjS can also be consulted for a variety of LC calculations. REFERENCES J. C. Giddings, Dynamics of Chroniutogru~)h),,Part I , Dekker, New York, 1965. J. C. Giddings, Unified Sepururion Science, Wiley-Interscience, New York, 1991. J. C. Giddings, J . Chem. Edcrc. 1967, 44, 704-709. B. L. Karger, L. R. Snyder, and C. Horvath, A n Ititrodirction to Sepurution Science. Wiley, New York, 1973. 5. D. Ishii and T. Takeuchi, J . Chromutogr. Sci. 1989, 27, 71-74. 6. J. F. Parcher and T. L. Chester (eds), Unified Chromatography, ACS Symposium Series 748, American Chemical Society, Washington, D.C., 2000. 7. A. T. James and A. J. P. Martin, Biochem. J . 1952, 50, 679. 8. H. H. Strain and J. Sherma, J . Chem. Edcic. 1967, 44, 238. 9. L. S. Ettrc, LC-GC No. Am. 2003, 21. 458-467. 10. A. J. P. Martin and R. L. M. Synge, Biochem. J . 1941, 35, 1358. 11. L. C. Craig, Anal. Chem. 1950, 22, 1346-1352. F Q V U ~SOC. . 195.5, 51, 34-44. 12. E. Glueckauf, TTUIZS. 13. J. J. van Dcemter, F. J. Zuidenveg, and A. Klinkenberg, Chcm. Eng. Sci. 1956, 5, 271. 14. J. C. Giddings, J . Chein. Phys. 1959, 31, 1462. 15. J. C. Giddings, Anal. Chem. 1963, 35, 2215. 16. ASTM, Standurd Pructice jor Liqirid Chrom~togr~~pliy Terms urid Relationships, ANSI/ASTM E 682-79, American Society for Testing and Materials, Philadelphia, 1979. 17. L. S. Ettre and A. Zlatkis (eds), 75 Years of Chromutogruphy-A Historical Dialogue, Vol. 17, J. Chromatogr. Library, Elsevier, Anistcrdam, 1979. 18. L. S. Ettre, Anal. Chemz. 1971, 43(14), 20A-31A. 19. L. S. Ettre, J . Chromatogr. 1975. 112, 1-26. 20. L. S. Ettrc and C. Horvath, Anul. Chem. 1975, 47, 422A-444A. 1. 2. 3. 4.

REFERENCES

65

21. L. S. Ettre, Am. Lab. 1978, 10(10), 85-91, and (111, 120-127. 22. L. S. Ettre, Anal. Chem. 1985, 57, 1419A-1438A; 23. L. S. Ettre, Am. Lab. 1992, 24(1) 48C-485, and (181, 15-23. 24. L. S. Ettre, LC-GC Nu. Am. 2001, 19, 506-512. 25. L. S. Ettre, J . Chromatogr. Sci. 2002, 40, 458-472. 26. L. S. Ettre, LC-GC No. Am. 2002, 20, 128-140 and 452-463. 27. J . Chromatogr. A 2003, 1000(1,2).

28. P. R. Rony, Separ. Sci. 1968, 3, 239. 29. L. S. Ettre, Pure Appl. Chem. 1993, 65, 819-872. See also, L. S. Ettre, LC-GC 1993, 11, 502. 30. A. Felinger and G. Guioehon, J . Chromatog. A 1998, 796, 59-74. 31. H. Kalasz, J . Chrornutogr. Sci. 2003, 41, 281-283. 32. H. Luo, L. Chen, Z. Li, Z. Ding, and X. Xu, Anal. Chem. 2003, 75, 3994-3998. 33. J. A. G. Dominguez, J. C. Diez-Masa, and V. A. Davankov, Pure Appl. Chem. 2001, 73, 969-992. 34. S. Dal Nogare and R. S. Juvet Jr., Gas-Liquid Chromatography, Interscience, New York, 1962, pp. 56-58. 35. E. Grushka, M. N. Myers, P. D. Schettler, and J. C. Giddings, Anal. Chem. 1969, 41, 889-892. 36. D. A. Dezaro, T. R. Floyd, T. V. Raglione, and R. A. Hartwick, Chromatogr. Forum 1986, 1(1), 34. 37. J. P. Foley and J. G. Dorsey, J . Chrumutugr. Sci. 1984, 22, 40. 38. W. W. Yau, Anal. Chem. 1977, 49, 395. 39. J. P. Foley and J. G . Dorsey, Anal. Chem. 1983, 55, 730-737. 40. L. S. Ettre, LC-GC Nu. Am. 2003, 21, 12. 41. J. A. Jonsson, LC-GC No. Am. 2002, 20, 920. 42. Z. Papai and T. L. Pap, J . Chromatogr. A 2002, 953, 31-38. 43. B. A. Bidlingmeyer and F. V. Warren, Jr., Anal. Chem. 1984, 56, 1583A. 44. L. R. Snyder, J . Chrornatogr. Sci. 197, 10, 200. 45. G. C. Carle, Anal. Chem. 1972, 44, 1905. 46. B. L. Karger, J . Gus Chrumutugr. 1967, 5, 161. 47. D. D. Bly, .I. Polymer Sci. Purt C , 1968, 21. 13. 48. J. W. D o h , LC-GC No. Am. 2002, 20, 430-436. 49. J. W. Dolan, LC-GC No. Am. 2002, 20, 594-598.

50. J. C. Giddings, Anal. Chem. 1967, 39, 1027. 51. J. C. Medina, N. Wu, and M. L. Lee, A n d . Chem. 2001, 73, 1301-1306. 52. R. Kaiser, Chromatogruphie in der Gasphase, 2nd ed., Vol. 2, Bibliographisches Institut, Mannheim, West Germany, 1066, pp. 47-48. 53. R. R. Freeman (ed), High Resolution Gas Chromatography, 2nd ed., HewlettPackard, 1981. 54. L. S. Ettre, Chroniutugraplzia 1975, 8, 291. V. R. Meyer, J . Chromutogr. 1985, 334, 197.

66

INTRODUCTION TO CHROMATOGRAPHY

SELECTED BIBLIOGRAPHY Heftmann, E. (ed), Chromatogruphy: Fundumentals and Applications of Chromutographic and Electrophoretic Methods, 2 vols., 4th ed., Elsevier, Amsterdam, 1983. Part A has individual chapters on fundamentals and techniques, and Part B has 15 chapters on separate applications. Karger, B. L., Snyder, L. R. and Horvath, C., An Introduction to Separation Sciencc, Wiley-Interscience, New York, 1973. Miller, J. M., Sepumtion Methods irz Chemical Analysis, Wiley-Interscience, New York, 1975.

BAND BROADENING AND KINETICS In describing the chromatographic process in the preceding chapter, two fundamental questions were raised: Why are analytes caused to separate from each other? Why do the individual peaks broaden in the process? The first question can be answered by applying classical thermodynamic principles, and the concept of equilibrium partitioning was introduced. That discussion will be continued in the next chapter. The broadening question can be answered by applying classical kinetic principles, which is the topic of this chapter. After both questions have been addressed, some final discussion and conclusions will be presented in Chapter 5. First, however, let us look at the stationary phases used in Chromatography.

3.1 CONFIGURATIONS OF THE STATIONARY PHASE The stationary phase can be a liquid or a solid. If it is a liquid, it can be coated directly on the inside walls of a capillary tube (column), or it can be coated on an inert solid support and be handled like a solid. In effect, then, there are three stationary-phase configurations: In the first type, a solid (with or without stationary liquid) is packed into a column; in the second type, a solid (with or without stationary liquid) is coated on the surface of a flat, plane material such as glass [thin-layer chromatography (TLC)], and in the Chrornatogruphy: Concepts and Contrusts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

67

68

BAND BROADENING AND KINETICS

third type, a liquid (or a solid) is coated on the inside wall of an open tube (OT). Since most of this book is devoted to columnar chromatography, not TLC, the rest of this section will be limited to considerations of columns. Columns can be open tubes or can be filled with packing. OT columns find most use in GC and in capillary zone electrophoresis (CZE). Most popular are columns made of fused silica that have the stationary phase (SP) bonded to the inside wall, often with polymerization and cross-linking. Further discussion of the nature of OT columns will be deferred, and the following section will be concerned with packed columns, which are most common in HPLC and packed-column GC. Characteristics of Column Packings

Most column packings are siliceous solids but there are many different types and manufacturing processes. Other packings are organic polymers. The characteristics of the packings that are important for chromatographic use include: particle size and distribution, pore size, surface area and shape, surface energy, pH compatibility, and rigidity under pressure. Throughout this book, these characteristics will be discussed where relevant so that a complete picture will be eventually obtained. To begin this discussion, let us consider some of the properties of silica packings and compare them with nonsiliceous materials. 1. Packings can be spherical or irregular in shape. In general, the spherical ones seem best for HPLC because they pack more tightly and uniformly. Figure 3.la shows a cross section of a column packed with irregular particles commonly used in GC, and Figure 3.2 illustrates a spherical silica packing of the type commonly used in HPLC. 2. Packings can be porous (like some silica gels) or nonporous (like glass beads). Because the porous ones have larger surface areas, they are usually preferred. The larger the surface area, the more sites for sorption and the likelihood of improved performance. The pores can be of various sizes, and usually a uniform pore size is best. Figure 3.1h shows a higher magnification of a common GC packing called Chromosorb P, illustrating its porous nature. Pore sizes become critical when they are of molecular magnitude because small sizes can exclude large molecules from the interior pores. HPLC packings for biomolecules require pore sizes 2 30 nm. Regulation of pore size can produce separations according to size, which is the basis of size exclusion chromatography (SEC). 3. Packings can be naturally occurring minerals or synthetically manufactured. The packing in Figure 3.1 is manufactured from a mineral called

3.1

CONFIGURATIONS OF THE STATIONARY PHASE

x4,700

69

1 rm

Figure 3.1. ( ( I ) Packed column showing magnification of pore rtructure. ( h ) High magnification ( X 4700) of Chroniosorh P. iiscd in GC.

Figure 3.2. Spherical silica beads, used in HPLC (Sphcrisorb ODS).

70

BAND BROADENING AND KINETICS

-CH-CH~-CH,-CH-CH,-CHI

I

(CH~--CH~;-

Figure 3.3. Representation of cross-linked polystyrene-divinylbenzene polymer structure

diatomaceous earth. It has a pore size of 0.4-2 p m and a surface area of 4 m2g-'. Typical silica packings for HPLC are manufactured by gelation of sodium silicate, yielding a material with an empirical formula of SiO,.H,O. Their properties can be varied in this so-called sol-gel process, and pore sizes can vary from 5 to 400 nm (referred to as microporous) and surface areas from 10 to 500 m'g-'. 4. Packings can be made from materials other than silica. Many chemicals, both inorganic and organic, have been in use for many years; examples are alumina, magnesium silicate, agarose, porous graphitic carbon, hydroxyapatite, and, more recently, zirconia, ZrO, Synthetic porous organic polymers such as polymethylmethacrylate or copolymers made from divinylbenzene and styrene are very popular and find use in GC and LC. A representation of the cross-linked structure is shown in Figure 3.3. The extent of cross-linking, controlled by the amount of divinylbenzene used, determines the pore size and rigidity of the polymer. Pore sizes can vary from 4 to 200 nm, similar to the silica gels. A different synthesis method, developed in the late 1950s, produces polymers whose pore size is not determined by the cross-linking but by the size of porogens used in the process. These polymers have become known as macroporous polymers because of their larger pore sizes, which can range up to 40 p m . Reference 2 contains many photomicrographs showing the pores of typical commercial products that can be made from organic polymers such as polyacrylic acid in addition to styrene/divinylbenzene.

.'

3.1

CONFIGURATIONS OF THE STATIONARY PHASE

71

Figure 3.4. ( a ) Macroporous and ( h ) mcsoporous 5tructure of monolithic silica column. SEM pictures courtcsy of Merck KGaA, Darmstadt, Germany. Printed with pcrmission.

5. Packings are usually microparticulates, but new ones are monolithic, a term taken from the Greek for “consisting of a single piece.” The idea of forming a single continuous monolith or rod inside a column was suggested as far back as 1970, and the first HPLC columns were reported in the late 1990s. The first successful commercial columns were introduced by Merck KGaA (Darmstadt, Germany) in 2001. Manufactured by a sol-gel process into a single stable rod,3,4 the monolithic columns so produced are enclosed in a plastic PEEK (poly ether ether ketone) sheath. They have both macroporous (typically 2 p m ) and mesoporous (typically 13 nm) structures as shown in Figure 3.4, with surface areas of 300 m’g-’. Compared to conventional microparticulate columns used in HPLC, the monolithic columns exhibit a lower pressure drop and can be operated with high efficiency at higher flow rates. These silica columns are designed for HPLC5 and CEC, and are based on the methods described in publications from Minakuchi and co-workers.” Earlier (noncommercial) organic (styrene-divinylbenzene copolymers) monoliths were not ideal for HPLC because they swelled in organic solvents and had poor mechanical strength, but new Swift columns, now available from

72

BAND BROADENING AND KINETICS

0/"

\

)i0',;H

/H-.... H ,

/H

J\/+\A\ J\/qi\oA\ ,\ I

0

0

0

'0

0

/O\

0

0

0

0

/H\

0

0

0/H

/"

.D\oAo I./O

0

Figure 3.5. Representation of some possible functional groups on the surface of silica

ISCO' are reported to be better suited to HPLC.' The Swift medium is a polymer of styrenes, methacrylates, and bifunctional monomers that can be used in the pH range of 1-14. A product review in 2004 listed six manufacturers of monolithic columns." They seem highly likely to replace microparticulate columns for some HPLC applications, and they are discussed in more detail in Chapter 8. 6. The surfaces of these solid packings can be modified to produce different chemical properties. The styrene-divinylbenzene polymers can be reacted to produce active functional groups on their surfaces for use in ion exchange chromatography (IEC), For example, when reacted with sulfuric acid, sulfonic acid groups are formed that can act as cation exchangers. The surfaces of the silica packings contain hydroxy groups that give them a heterogeneous energy surface that is usually undesirable for chromatographic use. Some typical silica surfaces are shown in Figure 3.5. To make t h e surfaces more homogeneous for chromatographic use, several treatments have been devised. The natural surface has become known as type A silica; purifying it to remove inorganic metals produces a silica known as type B silica. Other, proprietary treatments produce surface Si-H groups rather than Si-OH groups, and it is known as type C silica."' The unbonded type C silica does not retain water like type A, so it is a good choice for normal-phase HPLC (see Chapter 8 for classifications of HPLC modes). It can also be used with organic MPs in normal-phase mode as well as reversed-phase mode. Several bonded-phase materials are commercially available for the more popular form of reversed-phase HPLC. Another improvement on raw silica is the production of particles that are hybrids of silica and organic (organosiloxane) elements. Waters" has marketed such a hybrid, which is reported to have a better pH stability and improved performance for basic compounds in HPLC. Alternatively, to remove the most active sites and make the surface more uniform (deactivating it), silica is often silanized with deactivating reagents

3.1

%-OH

1 CI$-C,g

-

2 Silica

CONFIGURATIONS OF THE STATIONARY PHASE

73

OH t

I

Si.-O-Si-C18

+ 3 HCI

H20

support pamcle

Covalent band

Figure 3.6. Typical silanizing reactions used to producc bonded phases for HPLC

such as dimethyldichlorosilane for use in G C (see Chapter 7). O r it can be reacted with reagents such as trichlorooctadecylsilane, as shown in Figure 3.6 to put organic alkyl (R) groups on the surface for use in HPLC. The resulting support has a monomolecular layer of C,,H,,, an octadecylsilyl, or ODs, group which is hydrophobic. Other groups can be attached to the silica surface using similar reactions, and the supports so produced have become known as bonded phases, and the methods using them are called bondedphase chromatography (BPC). These bonded phases have become most popular and are discussed further in Chapter 8.

Column Volume Characteristics

To provide a basis for further discussions about band broadening, we need to consider in more detail the physical situation that exists when these packings are coated with liquids. For purposes of this discussion, bonded phases can also be considered to be solids with thin layers of stationary phase (SP) bonded to their surfaces. In either case, the SP (a liquid or thc bonded layer) should be thin and uniform on the surface like that shown in Figure 3.7. When liquids are coated o n solid supports, pools of stationary liquid phase can be formed in the pores o f the particles (Fig. 3.80) or between the particles (Fig. 3.8h). Such nonuniform films are undesirable, which is thc main reason packed GLC columns have been largely replaced by wall-coated open tubular (WCOT) columns with thin, uniform films.

Figure 3.7. Uniform distribution of liquid o n solid support.

74

BAND BROADENING AND KINETICS

Figure 3.8. Nonuniform distribution of liquid on solid support

The total volume in the column, V , , is made up of three parts:

v,,

V,= + V\ + VM where V,, = volume occupied by the d i d support V j = volume occupied by the stationary phase VM = volume occupied by the mobile phase

(3.1)

If the solid is nonporous, VM is the space between the particles (the irzterpurticle volume), but, if it is porous, VM includes both the interparticle volume and the internal volume of the particles. (In some instances, the size of the pores in the solid may be too small to admit the analyte molecules, and thus the internal portions of the solid may not be accessible to the sample, but that is a complication we will ignore at present.) These two types of particle give rise to two definitions of porosity. The total porosity, sT,is the fraction of void space in the column:

Even in a column that is packed very tightly with small particles, this porosity can be as high as 0.80-0.84 for a porous material; that is, a tightly packed bed of porous particles is still about 80% empty! The other measure of porosity, c , , is only that fraction of volume between the particles: interparticle volume E, = (3.3) V.,. It varies with the tightness of the packing and has a typical value of 0.40-0.45. Nonporous, or solid-core, solids have only one porosity, of course.

3.2 RATETHEORY

3.2

75

RATE THEORY

T h e brief historical development in the last chapter noted that the early theoretical studies described chromatography in terms similar to distillation o r extraction and were known as the plute theoiy. Useful as it may have been in the development of chromatography, the plate theory is of little value in modern chromatography and has been replaced by the rate theoty. Any of the early books on gas chromatography can be consulted for a discussion of the plate theory, and a discussion of countercurrent liquid-liquid extraction in Chapter 14 includes some of the same principles that can be applied to the chromatographic plate theory. Giddings" has written a definitive historical summary of the concurrent development of the plate and rate theories. Original Van Deemter Theory

The most influential study using the kinetic approach was published by van Deemter and co-workers'' in 1956. It identified three effects that contribute to band broadening in packed G C columns: eddy diffusion (the A term), longitudinal molecular diffusion (the B term), and mass transfer in the stationary liquid phase (the C term). The band broadening was expressed in terms of the plate height, H , as a function of the average linear gas velocity, u . In its simple form, the uun Deemter eqciution is

H=A

+ BL1

-

+Cu

(3.4)

Since plate height is inversely proportional to plate number, a small value indicates a narrow peak-the desirable condition. Thus, each of the three constants, A , B , and C should be minimized in order to maximize column efficiency. In Eq. (3.4), the speed of the mobile phase was expressed as its linear velocity, u , rather than its flow rate, F . T h e two are proportional for a given column, of course, since

F

= u(

A,)

(3.5)

where A , is the cross-sectional area of the column.:': Chromatographers a r e often imprecise when discussing MP flow and may use velocity instead, o r vice versa. T h e linear velocity can be measured chromatographically from the length of the column, L , and the retention time for a nonretained solute, t,:

"For an open tubular column, A , = m y z , where I' is the radius of thc column; for packed columns, the cross-sectional area exposed to the mobile phase is less due to the packing.

76

BAND BROADENING AND KINETICS

rer Time

Figure 3.9. Illustration of eddy diffusion o r multipath effect. Reprinted with permission from H. McNair and J. Miller, Basic Gas Clirvmutvgruphy, John Wiley & Sons. Copyright 1998; this material is used by permission of John Wiley & Sons, Inc.

Typical units for the velocity are centimeters per second. The flow rate is a volume term, proportional to velocity, and its units are commonly milliliters per minute. Returning to the rate equation, let us examine the three terms as proposed by van Deemter. The A term is called the eddy diffusion or multipath term, the B term is caused by molecular diffusion, and the C term concerns mass transfer of the solute. Eddy Diffusion

As originally proposed by van Deemter et al.,” the A term dealt with eddy diffusion in packed columns, as shown in Figure 3.9. The diffusion paths of three molecules are shown in the figure. All three start at the same initial position, but they find differing paths through the packed bed and arrive at the end of the column having traveled different distances. Because the flow rate of the mobile phase is constant, they arrive at different times and are separated from each other. The one taking the most direct path, path 3 in the figure, ends up in the front part of the peak, for example. Thus, for a large number of molecules, the eddy diffusion process or the multipath effect results in band broadening as shown. The A term in the van Dcemter equation is A

=

2Ad,

(3.7)

where d, is the diameter of the particles packed in the column and A is a packing factor. To minimize A , small particles should be used, and they should be tightly packed. In practice, the lower limit on the particle size is determined by the pressure drop across the column and the ability to pack very small particles uniformly. Small ranges in mesh size also promote better packing (minimal A).

3.2 RATETHEORY

77

Distance along z-axis

Figure 3.10. Z o n e widening due to diffusion. Three times are shown with

I,

> t? > t ,

Molecular Diffusion The B term of Eq. (3.4) accounts for the well-known molecular diffusion. The equation governing molecular diffusion is

where D , is the diffusion coefficient for the solute in the mobile phase and is called a obstruction factor that allows for the nature of packed beds. Open tubes (as in capillary GC) are not packed, and in that case the B term would not include an obstruction factor. Figure 3.10 illustrates how a zone of molecules diffuses from the region of high concentration to that of lower concentration with time. Going from time t , to t? to i 3 , the zone spreads and its maximum is decreased, resulting in zone broadening as the analyte proceeds through the column. The equation tells us that a small value for the diffusion coefficient is desirable so that diffusion is minimized, yielding a small value for B and for H . In general, a low diffusion coefficient can be achieved in GC by using carrier gases with larger masses such as nitrogen or argon; in HPLC, viscous liquids would promote low diffusion. However, in both cases these choices are not desirable for other reasons, so the choice of mobile phase is seldom based on the B term. In the van Deemter equation, this term is divided by the linear velocity, so a large velocity or flow rate will also minimize the contribution of the B term to the overall peak broadening. That is, a high velocity will decrease the time a solute spends in the column and thus decrease the time available for molecular diffusion. Since high velocities are desirable to minimize the time

+

78

BAND BROADENING AND KINETICS

I

Figure 3.1 1. Illustration o f zone spreading due to mass transfer ( K = 2.0).

of an analysis, the B term is usually not an important contributor to the overall band spreading ( H ) , and it is of little interest in selecting chromatographic parameters. Mass Transfer The C term in the van Deemter equation concerns the transfer of solute into and out of the stationary phase, that is, sorption and desorption. This mass transfer can be described by reference to Figure 3.11. In both parts of the figure, the upper peak represents the distribution of a solute in the mobile phase and the lower peak the distribution in the stationary phase. A distribution constant of 2 is used in this example, so the lower peak has twice the area of the upper one. At equilibrium, the solute achieves relative distributions like those shown in Figure 3.1 1a , but an instant later the mobile phase moves the upper curve downstream giving rise to the situation shown in Figure 3.11h. The solute molecules in the stationary phase are stationary; the solute molecules in the mobile phase have moved

3.2 RATETHEORY

79

ahead of those in the stationary phase, thus broadening the overall zone of molecules. The solute molecules that have moved ahead must now partition into the stationary phase, and those that were in the stationary phase must now equilibrate with the mobile phase, as shown by the arrows. The faster they can make this transfer, the less will be the band broadening. The C term in the van Deemter equation is

where d , is the average film thickness of the liquid stationary phase and 0, is the diffusion coefficient of the solute in the stationary phase. To minimize the contribution of this term, the film thickness should be small and the diffusion coefficient large. Rapid diffusion through thin films allows the solute molecules to stay closer together. Thin films can be achieved by coating small amounts of liquid on the stationary support or the column walls, but stationary liquid phases are seldom chosen for their diffusion coefficients. Minimization of the C term results when mass transfer into and out of the stationary liquid is as fast as possible. An analogy would be to consider a person jumping into and out of a swimming pool; if the solvent is water and shallow, the process can be done quickly; if the solvent is molasses (high viscosity) and deep, the process is slow. If the stationary phase is a solid, modifications in the C term are necessary to relate it to the appropriate adsorption-desorption kinetics. Again, the faster the kinetics, the closer the process is to equilibrium, and the less is the band broadening. The other part of the C term is the ratio k / ( l + k ) ' . This ratio' is minimized at large values of k , but very little decrease occurs beyond a k value of about 20. (See Table 2.3). Since large values of retention factor result in long analysis times, little advantage is gained by k values larger than 20. Additional discussion can be found in Chapter 6. Other Rate Equations

The C term in the van Deemter equation concerned mass transfer in only the stationary phase because mass transfer in the mobile phase was not thought to be significant since the equation was derived for GLC and mass transfer is fast in gases. However, when the van Deemter equation was applied to 'The term R(l - R ) is equal to k / ( l +k)' and is sometimes substituted for it in the rate equation.

80

BAND BROADENING AND KINETICS

Fused Silica

Liauid Phase

Figure 3.12. Illustration of mass transfer in the mobile phase. Reprinted with permission from H. McNair and J. Miller, Busic Gus Chrornutopplzy, John Wiley & Sons. Copyright 1998; this material is used by permission of John Wiley & Sons, Inc.

HPLC and thin-film capillary GC columns, it became necessary to include C terms for mass transfer in both the stationary and the mobile phases. Ideally, fast solute sorption and desorption will keep the solute molecules close together and keep the band broadening to a minimum. The so-called extended van Deemter equation includes two C terms, a C, term for mass transfer in the stationary phase and a C , term for mass transfer in the mobile phase. The single mass transfer term included in the original van Deemter equation was for mass transfer in the stationary phase. A simplified version of the extended equation is B H = A + - +U ( C , + C , ) u

(3.10)

This equation has found general acceptance although some modifications have been proposed and are discussed in the next section. First, we should describe mass transfer in the mobile phase. Mass transfer in the mobile phase can be visualized by reference to Figure 3.12, which shows the profile of a solute zone resulting from nonturbulent flow through a capillary tube, as originally postulated by Poiseuille in 1844. There is a range of mobile-phase velocities across the diameter of the column because of frictional resistance at the wall. One individual solute molecule may travel the column length, spending more time in the slow moving regions and arrive at the column exit after another molecule that finds itself in the higher velocity regions, on average. This dispersion is mitigated by lateral diffusion, and the faster the lateral diffusion, the better the mixing between flow streams, and the less the zone broadening. The C , term used for open

3.2

RATETHEORY

81

Expected width

Actual width

Flow

Figure 3.13. Comparison of actual zone width with the width expected. Reprinted with permission from J. Miller, Separation Methods in Chenzicul Anulysis, John Wiley Kr Sons. Copyright 197.5; this material is used by permission of John Wiley K! Sons, Inc.

tubular columns invented by GolayJ4and called the Golay equation is

C,

=

+ 1lk')d; 96( 1 + k)%,

(1 + 6 k

(3.11)

where d, is the diameter of the column. Small-diameter columns minimize broadening because the mass transfer distances are relatively small. Likewise, large diffusion coefficients promote mixing and decrease broadening. In Figure 3.13, the actual zone width is shown to be larger than the width that would be expected without the effect of broadening attributed to the C,-term. Additional modifications to the original van Deemter equation have been proposed by other workers. For example, one can argue that eddy diffusion (the A term) is part of mobile-phase mass transfer (the C , term) or is coupled with it. Giddingsls has thoroughly discussed mass transfer and has concluded that a coupled term combining eddy diffusion and mass transfer is the best way to describe mobile-phase effects. Hawkes" has summarized the development of various versions of the van Deemter equation from the original one for GLC through later ones for HPLC. He has expressed his preference for a version in which the C , term includes a number of generalized parameters including the original A term, which is no longer a separate term; in Eq. (3.12) fn denotes that H is a

82

BAND BROADENING AND KINETICS

function of the parameters listed in the parenthesis;

H

B

= -

U

+ C,U + C,

u

(3.13)

In 1999, Knox” published an interesting work summarizing his interpretation of the current status of dispersion for packed columns in chromatography. H e concluded that mobile-phase effects are the most important ones, thus giving a new view of the A term in the rate equation. This is not of much consequence in GC where OT columns have become the most popular, but his study did affect the development of HPLC packed columns. More details are included in Chapter 8.

Figure 3.14. Typical van Deemter plot. Reprinted with permission from Lee, Yang, and Bartle, Open Tubular Colurnrz Gus Chromatography, John Wiley ti Sons. Copyright 1984; this material is used by permission of John Wiley & Sons, Inc.

3.2 RATETHEORY

83

Van Deemter Plots

When the rate equation is plotted ( H vs. u ) , the so-called uan Deemterplot takes the shape shown in Figure 3.14. As one would expect from an equation in which one term is multiplied by velocity while another is divided by it, there is a minimum in the curve-an optimum velocity that provides the highest efficiency and smallest plate height. It is logical t o assume that chromatography would be carried out at the optimum velocity represented by the minimum in the curve since it yields the least peak broadening. However, if the velocity can be increased, the analysis time will be decreased. Consequently, chromatographers have devoted their time to manipulating the van Deemter equation to get the best performance in the shortest analysis time. By examining the relative importance of the individual terms to the overall equation in Figure 3.14, one sees that the upward slope as velocity is increased comes about from the increasing contribution of the C term. Therefore, most attention has been focused on minimizing it, a topic that will be covered shortly. There is no single van Deemter plot for all chromatographic systems and Figure 3.14 most closely resembles a plot for a G C column. Figure 3.15 is slightly different and is more typical of plots for HPLC columns. In it, Huber and HulsmanlX have labeled the various contributions of the four terms from the extended van Deemter equation, showing that the C , term is the most important one at high velocities for the columns they were using.

u (mrn/sec)

Figure 3.15. Typical plot o f the rate equation for LC. Reprinted from J. F. K. Huber, J . Clzrorncitog. Sci. 1969, 7, 85 by permission of Preston Publications, Inc.

84

BAND BROADENING AND KINETICS

Rate theory is a theoretical concept, but it has practical applications. It is possible to obtain a van Deemter plot for one’s column in order to evaluate it and the operating conditions. A solute is chosen and run at a variety of flow rates, being sure to allow sufficient time for flow equilibration after each change. The plate number N is calculated from each chromatogram and is then used to calculate the plate height H . The plate height values are plotted versus linear velocity or flow rate. From this plot, one can compare the optimum velocity (or flow) with the operating velocity being used in practice to determine if a change in velocity could be made to produce higher efficiencies. Practical Consequences of the Rate Theory

Let us conclude this discussion by focusing on only two rate equations-one for packed columns and one for open tubular GC. The first is represented by the extended van Deemter equation: (3.14)

and the second by the Golay equation14: 20, H=-+

qkd: ii (l+k)2D,

+ (1 + 6k + llk2)d,2ii 96( 1 + k ) 2 D ,

(3.15)

Packed Column Gas Chromatography When the rate equation is applied to GC, allowance must be made for the fact that the mobile phase (a gas) is compressible and the linear velocity u is not a constant but increases throughout the column. This effect will be discussed further in Chapter 6. Of the four terms in the rate equation, the most important one is the C , term. It tells us that our packed column should have thin films on the solid support; d, should be small. This is achieved by using supports with large surface areas and small amounts of stationary phase (low percentage of liquid-phase). The lower limit for liquid-phase percentage is that point at which the solid support is not fully covered. This limit is reached at about 0.3% (by weight) for a typical diatomaceous earth solid support. Since d , is in the C, term and is of most importance at the large gas flow rates that are preferred in GC, it is probably the most important consideration in designing a G C column. Other parameters in the Cs term are D,, the diffusion coefficient in the liquid phase, q, and the ratio k / ( l k)’. The diffusion coefficient should be

+

3.2 RATETHEORY

85

large, but often this choice cannot be exercised because the liquid phase is chosen for selectivity reasons. Previously, it had been thought that higher diffusion coefficients would be found in stationary phases of low viscosity. This is true only for small stationary-phase molecules; for polymers of the type used in GC, the diffusion coefficient is virtually independent of viscosity.” Of course, these polymers cannot be used below their glass temperatures, so low-viscosity polymers may be required for low-temperature GC column operation. The configuration factor q is determined by the type of bed and is f for uniform liquid films preferred in GLC. The k ratio should be relatively large, as we have seen. Mass transfer in the mobile phase is a significant contributor to zone broadening for packed columns, and Knox” has stated that it is the most important factor. T o minimize its effects, columns should be tightly and uniformly packed with small particles.

Wall-Coated Open Tubular (WCOT) GC Columns For O T columns, the thin liquid film is deposited directly on the walls of the column rather than on the solid support, and its characteristics are best described by the Golay equation. Typical OT columns have an inside diameter of 0.25 mm and a film thickness of 0.25 p m . Usually, the stationary phases are not simple liquid films but rather cross-linked polymers, often bonded to the silica column surface. The effect of carrier gas is shown in Figure 3.16, which gives typical van Deemter plots for the common carrier gases helium, hydrogen, and nitrogen. As predicted by the B term of the rate equation, nitrogen, which has a higher molecular weight and smaller diffusion coefficients, is more efficient and gives the smallest dispersivity H at its optimum velocity. This advantage is achieved at a relatively low flow velocity and is offset by the better performance of helium and hydrogen at higher velocities where the B term is not important, as shown in the figure. For fast analyses, helium and hydrogen are definitely preferred. It is the C, term that describes the band broadening at high velocities, and samples have larger diffusion coefficients in the light gases, hydrogen and helium. Also, smaller diameter columns result in shorter lateral diffusion distances and better mixing, keeping bands narrow. Hydrogen has been gaining in popularity for use with OT columns because it has several advantages. Being the lightest gas, its optimum velocity is highest, around 35-40 cm/s, permitting fast analyses. Its curve is the flattest, so increases in mobile-phase velocity will result in only small increases in H . Its viscosity is the lowest of the three gases, permitting its use at lower inlet pressures. In addition, hydrogen is less expensive than helium, especially for the higher purities. However, its use requires that the column oven be

86

BAND BROADENING AND KINETICS

c 10

20

30

40

50

60

70

80

90

Average linear velocity (cdsec)

Figure 3.16. Comparison of diffcrent carrier gas efficiencies in OT GC. 0 Rewurch und Deuelopment.

monitored for leaks and good ventilation be maintained in the laboratory, for safety's sake. In the United States most laboratories use helium because it is nearly as good as hydrogen, and it is safe. However, in many other countries where He is more expensive, safety precautions are taken and hydrogen is used. Packed-Column High-Performance Liquid Chromatography (HPLC) For HPLC, optimum operating conditions determined by the C terms are similar to those for GC: d, and d, should be small and D , should be large. Indeed, high-performance columns have resulted from the use of very small particles and thin films provided by bonded phases. Dong and Gant2" have compared the performance levels of 3-, 5-, and 10-pm columns. Their data is shown in Table 3.1 and in Figure 3.17. The B term is seldom of concern at the MP velocities used in HPLC. Other minor factors in the C, term were discussed earlier. Figures 3.15 and 3.17 showed that the rate equation can be considered to be a straight line if a short segment of the plot is considered. Snyder2' has shown that the straight line can be approximated by

(3.16)

3.2 RATETHEORY

Table 3.1

87

Performance Levels of lO-pm, 5-pm, and 3-pm Columns?"

Particle Diameter, d , ( pm)

Typical Lengths (cm)

Plates/Column

10 5 3

25-30 12.5-2s 3.3- 15

10,000- 12,000 10,000-20,000 4,950-22,500

Plates/Meter 40,000 80,000 1s0,000

Plates/Second" 50-100 200-300 300-600

"For pcaks at low k' using a low-dispersion instrument

'"I

60

-

50 -

Flow Rate (mL/min) Figure 3.17. Effect of particle size of the stationary phase on efficiency: ( a ) 10 p m , ( h ) 5 p m , 3 pni. Rcprinted with pcrmission from LC-GC, Vol. 2(4), Nov. 1984, p. 294. LC-C;C is a copyrighted publication of Advanstar Communications Inc. All rights rcscrved. (L.)

for particles larger than 10 p m where D is a constant:

D

=

18(d,)''.'

(3.17)

However, for smaller particles, down to 3 pm, he and others have found that a more general equation" is needed. It is a simple three-term equation, but clearly not an equation for a straight line:

H

= /@.33

B +-+cu U

(3.18)

Snyder's studyz3can be consulted for recommendations on selecting the best experimental conditions for small particle columns. He considers decreasing

88

BAND BROADENING AND KINETICS

the column pressure (and flow rate), increasing the column length, and combinations of these. His scheme uses a reduced form of Eq. (3.171, which needs to be defined before we can consider further developments in the evolution of the rate equation for HPLC. Reduced Rate Equation for HPLC (and GC)

As chromatographers attempted to fit the rate equation to different types of chromatography, they found the basic shape the same, but t h e actual van Deemter plots looked very different. Therefore, they sought a reduced form of the equation, which would be dimensionless and would yield a universal curve that would make comparisons between techniques easier. The variables that differ greatly between GC and LC for example are the diffusion coefficients and the particle sizes, so these were the ones chosen to be normalized. Giddings was the first to propose a reduced equation.24 The two parameters in the rate equation that have been redefined are the column dispersivity H and the linear velocity u . The reduced parameters, which are dimensionless, are (3.19) and (3.20) For O T columns, d p is replaced by the column diameter d,. The rate equation given in these reduced parameters is B h=y+(CS+CM)v

(3.21)

Figure 3.18 shows plots of the reduced rate equation (on a log-log basis) taken from Knox and Saleem.2s G C is compared with LC for two stationary phases and several nonretained analytes. While all the points do not fall on one smooth curve, the plot clearly shows the similarities between G C and LC plots when reduced parameters are used. The reduced parameters are also helpful in evaluating column performance. The best columns have a reduced plate height of 2-5-a number that can be thought of as representing the number of particles between sorptions -and 2 is a practical minimum. The reduced velocity represents the ratio between the flow velocity and the diffusion rate over one particle diameter; typical values should be in the range of 3-20.

3.2 RATETHEORY

89

2.0

1.o -c

3

-I

0.0

1.0

0.0

20

Log Y

3.0

Figure 3.18. Comparison of plots of reduced rate equations for G C (on left) and LC (on right). C H R = Chromosorb G; CSP = duPont CSP beads. Reproduced from the Journul of Chromutogruphic Science by permission of Preston Publications, Inc.

Bristow and Knox2' have proposed a reduced rate equation for HPLC by fitting their data to the equation

(3.22) which is the reduced form of Eq. (3.18). As shown in Figure 3.19, they find that a good HPLC column fits Eq. (3.23): h

.

2

= - + ,,0.33

+ 0.1

.

(3.23)

It has become an accepted model for LC columns and is referred to as the Knox equation. However, upon reconsideration of the published HPLC data since the 1960s, and in an attempt to identify the most important terms in the Knox equation, Knox2' has come to the conclusion that the best fitting equation is one modeled after Giddings,12.2x having a coupled term:

h=-+

[- + -

I (Dun)

I-'.(:.

(3.24)

90

BAND BROADENING AND KINETICS

h=2 / w

-+

+ 0.1~

-1

0

1

2

3

log v Figure 3.19. Reduced rate equation plot. Reprinted from reference 24 with permission

He has provided values forA, B , C , D , and n for nine different types of HPLC columns, including the new monoliths, and concludes that most band dispersion occurs in the mobile phase, not in the stationary phase, as previously thought. If this is so, further improvements in HPLC column performance can be expected from better packings or new packing structures, following t h e trend established by the new monolithic columns. In 1998, Knox published a personal view of the concept of band spreading in chromatography*’. It contains interesting insights into the developments presented in this chapter and is helpful in putting all of it into perspective. Redefinition of H

What is H anyway? The original interpretation, taken from distillation theory, was height equivalent to a theoretical plate, or HETP. We have seen that this concept was inadequate, and the preceding discussion of the van Deemter equation has presented it as a measure of the extent of spreading of an analyte zone as it passes through a column. Thus, a more appropriate term might be one that has been used several times in this discussion, column dispeusiuity. In fact, another, independent approach to the theory of chromatography defines H as

H = -u 2 L

(3.25)

where u 2 is the variance or distribution of analyte molecules about their mean in the analyte zone, as we have seen earlier, and L refers to the length

REFERENCES

91

(or distance) of movement of the analyte zone. Thus, H is truly the dispersion of an analyte per unit length migrated (or the retention time in columns). It is a superior concept that has been derived by Giddings from the random-walk model of chromatographic theory. Further elaboration of Eq. (3.25) can be found in Chapter 6, where column methods and planar methods are contrasted. For the present, it should be noted that L does not necessarily refer to column length, even though that is the definition usually used in this text. In fact, Giddings has used the equation in the form:

H

=

d/x

(3.26)

where X is used to denote the distance traversed by a zone in a given time.”’ REFERENCES 1. C. J. Dunlap, C. V. McNeff, D. Stoll, and P. W. Carr, Anal. Chem. 2001, 73, 598A-607A. See also www.zirchrom.com. 2. J. R. Benson, A m . Lab. 2003, 35(10), 44-52. 3. www. merck. de /english /Jeruices/chromuto~rai~hie /hplc/chromolith /index html. 4. A. M. Siouffi, J. Chromatogr. A 2003, 1000, 801-818 (contains 138 references).

5. D. Lubda, K. Cabrera, W. Kraas, C. Schaefer, and D. Cunningham, LC-GC No. Am. 2001, 19, 1186-1191. 6. H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka, and N. Takaka, Anal. Chem. 199, 68, 3498-3501. 7. www.swiftcolumns.com. 8. R. Stevenson, A m . Luh. News Ed. 2002, 34(1 I), 6 and Am. Lab. News Ed. 2002, 3 4 2 I), 4. 9. S. Miller, Anal. Chem. 2004, 76, YYA-101A. 10. L. Brown, B. Ciccone, J. J. Pesek, and M. T. Matyska, Am. Lab. 2003, 35(24), 23-29. 1 1. www.wuter.y.com. 12. J. C. Giddings, Dynamics of Chromatography, Part 1, Dekker, New York, 1965, pp. 13-25. 13. J. J. van Deemter, F. J. Zuidenveg, and A. Klinkenberg, Chem. Eng. Sci. 1956, 5, 271. 14. M. J. E. Golay, in Gas Chromatogruphy 1958, D. H. Desty (ed), Butterworths, London, 1958, p. 36. 15. J. C. Giddings, Dynamics of Chromatography, Part 1, Dekker, New York, 1965, pp. 40-65. 16. S. J. Hawkes, J . Chem. Educ. 1983, 60, 393-398. 17. J. H. Knox, J. Chromutogr. A 1999, 831, 3.

92

BAND BROADENING AND KINETICS

J. F. K. Huber and J. A. R. J. Hulsman, Anal. Chem. 1967, 38, 305. S. J. Hawkes, Anal. Chem. 1986, 58, 1886. M. W. Dong and J. R. Gant, LC-GC 1984, 2, 295. L. R. Snyder, J . Chrornatogr. Sci. 1969, 7, 352. E. Grushka, L. R. Snyder, and J. Knox, J . Chromatogr. Sci. 1975, 12, 25. L. R. Snyder, J . Chromatogr. Sci. 1977, 15, 441. J. C. Giddings, J . Chrornatogr. 1964, 13, 301. J. H. Knox and M. Saleem, J . Chrornatogr. Sci. 1969, 7 , 745. P. A. Bristow and J. H. Knox, Chromutogruphia 1977, 10, 279. J. H. Knox, J . Chrornatogr. A 2002, 960, 7-18. J. C. Giddings, Unified Separation Science, Wiley, New York, 1991. J. H. Knox, Band Spreading in Chromatography: A Personal View, in Aduances in Chromatography, Vol. 38, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 1998, Chapter 1. 30. J. C. Giddings, Unified Separation Science, Wiley, New York, 1991, p. 97.

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

PHYSICAL FORCES AND INTERACTIONS The process of selecting a chromatographic system for a given separation will be facilitated if we have some understanding of the forces that are operating in the system and effecting the separation. The magnitude of these forces is reflected in the numerical values of the equilibrium partition coefficients (distribution constants), which were shown in Chapter 2 to be a part of the fundamental theory of chromatography. In this chapter, we will first consider some fundamentals regarding intermolecular forces, and then we will see how these concepts have been applied to chromatographic separations. To begin, we have to admit that a discussion of intermolecular and interionic forces has to be simple because of the limited extent of our knowledge about them. The subject is very complex and our understanding is inadequate. In the end we must rely heavily on empirical relationships. This is particularly true in HPLC and Chapter 8 will continue and extend this discussion.

4.1

INTERMOLECULAR AND INTERIONIC FORCES

The terms adsorption and absorption (or purtition) were introduced in Chapter 2 along with the definition of the distribution constant K c . Now we want

Chrornutogruphy: Concepts and Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

93

94

PHYSICAL FORCES AND INTERACTIONS

(a)

-

/

/

/

(b)

(C)

‘0

to explore the forces that give rise to a particular partition coefficient and also see what effect they have on peak shape. Sorption Isotherms

If the distribution constant, K , , is a true constant, a plot of [A], versus [A], should be a straight line, as shown in Figure 4 . 1 ~ Or, . more specifically, an operational statement would be: In the range of analyte concentration where K , is a constant, the amount of analyte (its concentration) sorbed is directly proportional to the amount (concentration) in the mobile phase. We say that the system is linear, and we refer to such a graph as an isotherm, since the data are presented at a constant temperature. The chromatographic peak resulting from such a relationship is symmetrical and has the desired Gaussian distribution we have discussed earlier, as shown in Figure 4 . 2 ~ However, . at sufficiently high concentrations of analyte, deviations from linearity are

Time -+

Figure 4.2. Peak shapes corresponding to the isotherms in Figure 4.1: ( a ) linear, ( h ) Langmuir, and (c) anti-Langmuir.

4.1

INTERMOLECULAR AND INTERIONIC FORCES

95

usually observed, as shown by the deviation from the dashed line in Figure 4.1 a . Small analyte concentrations are preferred in chromatography to promote linearity. In systems that operate mainly by adsorption, it is common to find isotherms that deviate from linearity. The resulting isotherm, known as a Langmuir isotherm (Fig. 4.1 b), arises at high analyte concentrations when all of the adsorption sites on the SP have been covered up by adsorbed analyte, and none are available to the remaining analyte. This type of behavior is common in GS and LS chromatography and can result from sorption on a surface that has heterogeneous energy sites. The first analyte molecules adsorb on the most active sites, and additional molecules (at the higher concentration levels) see a surface that is much less active, and they do not adsorb as much as the earlier molecules did, resulting in nonlinearity. Sites capable of hydrogen bonding with the analyte are often the cause of this effect. The peak shape that results from a Langmuir isotherm is shown in Figure 4.2b; the skewed shape is referred to as tailing. The opposite type of isotherm (anti-Langmuir) is shown in Figure 4.lc and its resulting peak shape (called fronting) in Figure 4 . 2 ~ In . this situation, the analyte molecules, which are the first to adsorb, facilitate the sorption of additional molecules. For example, in LC when the planar molecule phenol adsorbs on alumina, the phenyl rings extend out from the surface, as shown in Figure 4.3. Additional molecules will not be attracted by the alumina surface that attracted the first molecules but by the phenyl rings, which will facilitate increased solubility. In GC this effect occurs when the temperature is too low or the system is overloaded (too large a sample). Another conscquence of nonlinear sorption is that the retention time of a given analyte will be a function of its size, as shown in Figure 4.4. In the case of tailing (Langmuir isotherms), the retention time decreases as the sample size increases, and the opposite is true for fronting (anti-Langmuir) peaks. Clearly this is a problem if one wants to make identifications based on retention times or if one’s computer is looking in a specified time “window” for a given peak. The correct retention time in Figure 4.4 is the one for the peak labeled f.Tailing also reduces resolution and complicates quantitation, so there are many reasons for eliminating it.

Q 8 Q 0

I

0

Figure 4.3. Representation o f the adsorption of phenol on alumina.

96

PHYSICAL FORCES AND INTERACTIONS

I @ -l Ra)

I I

I

4-

I

I

I I I

I

I

-

I I

I

I

(klf I I I

I

Time

----t

I

Figure 4.4. Effect of sample size o n retention time for an asymmetric peak

A fourth isotherm arises from chemisorption: a chemical reaction rather than a physical force. There are examples, especially in GC, where a basic chemical such as an amine reacts completely with the acidic surface of the stationary phase and is never seen at all. Many repeated injections are necessary to react all of the acidic sites before any amine can be eluted from the column. Types of Forces

Before examining the types of forces present in chromatographic systems, let us consider some parameters that are commonly used to reflect the magnitude of intermolecular forces in general. A typical list includes: 1. Ionization potential 2. Electron affinity 3. Electronegativity 4. Molar volume 5. Ionic radius 6. Ionic potential 7. Dipole moment 8. Dielectric constant 9. Polarizability 10. Boiling point and vapor pressure 11. Solubility 12. Activity coefficients 13. Hildebrand solubility parameter

4.1

INTERMOLECULAR AND INTERIONIC FORCES

97

In most cases these parameters define properties of a given chemical (molecules or ions) as it exists alone (neat) or perhaps in aqueous solution. For example, the high boiling point (bp) of the chemical, water, is commonly used as an illustration of the effect of intermolecular hydrogen bonding, which drastically increases the forces between individual water molecules making them harder to separate (volatilize). The bp is an example of a measurement of the forces between like molecules; but, in chromatography, we are more interested in the forces between unlike molecules, as in a solution with solute and solvent. This is a more difficult task and, as noted earlier, we are often forced to rely on empirical relationships. Another simplification we use is to categorize molecules as belonging to of one of two types, polar and nonpolar. Often we do this because we have no better way of expressing the forces we are trying to describe. In discussions of solutions, we also use nonspecific descriptions of intermolecular forces, such as “like dissolves like.” Even though they are simple, these expressions are used frequently in discussions of chromatographic intermolecular forces. lonic lnteractions Ions are favored in aqueous solutions and are important mainly in LC, although some GC work has been reported with molten ionic salts (now called room temperature ionic liquids, RTIL) for stationary liquid phases,’ and two new ones have just been described‘ indicating a new interest in these unique liquids. Forces between ions of like charge are repulsive and between ions of different charge are attractive, according to Coulomb’s law. They are relatively long-range and strong forces. The main illustration of ionic forces in LC is in ion exchange where the sample is at least partly ionized and the stationary phase contains ionic sites. More information o n ion exchange and its application in HPLC is included in Chapter 8. Ions are also attracted or repulsed by the polar ends of a dipolar molecule. These forces can be referred to as ion-dipole forces. In addition to ion exchange chromatography, they are also important in aqueous LC systems where a polar stationary phase can interact with ionic solutes. Sometimes, as in reversed-phase HPLC, ions are not retained sufficiently by the stationary phase and counterions are added to the mobile phase to form nonionic ion pairs, thereby altering the ionic environment in the system Van der Waals Forces Three types of weaker forces have been identified and together are referred to as uan der Wualsforces. They are listed in Table 4.1 according to the type of interaction: dipole-dipole, dipoleeinduced dipole, and induced dipoleeinduced dipole. They are more often called by the names listed in the second column or by the names of the investigators who first described them, as listed in the third column.

98

PHYSICAL FORCES AND INTERACTIONS

Table 4.1

Classification of Nonionic Intermolecular van der Waals Forces

Interaction Dipolc-dipole Dipole-induced dipole Induced dipole-induced dipole

Name

Investigator

Orientation

Keesom (1912) Debye (1920) London (1 930)

Induction

Dispersion

Orientation forces are easy to understand by analogy with magnetic forces. T h e opposite poles attract even though random molecular motion keeps the attraction small. Similarly, the induction forces can be viewed as similar to a magnet attracting nonmagnetic iron. The induction of a dipole depends o n the polarizability of the nonpolar molecule; large molecules with easily deformed electronic clouds have large polarizabilities. Dispersion forces cannot be explained by the magnetic analogy nor by conventional electrostatics. They are weak forces that exist even in monoatomic gases, which would not be expected to have interatomic attractions because they are symmetrical and nonpolar. It is believed that at any particular instant this symmetry is somewhat distorted d u e to the motion (and position) of the electrons of a given atom, which produces a momentary polarity. This momentary polarity can attract and be attracted by a similar polarity in a neighboring atom o r molecule in such a way as to produce a net attraction. As we have already seen, such inductions depend on the polarizability of the molecule o r atom. Dispersion forces will always be possible between molecules, but they are the only forces between nonpolar hydrocarbons such as the alkanes. For this reason alkanes are often chosen as the ideal molecules for study o r for use as standards. A n example of a chromato-

Table 4.2

Relative Magnitude of Intermolecular Forces

Polarizability Substance He Ar Xe H2

co

HCI HBr HI NHS

H20

p(A3’)

0.21 1.63 4.0 0.81 1.YY 2.63 3.5 5.4 2.24 1.48

Dipole Moment (debyes)

Molecular (mL/mol)

0 0 0 0 0.12 1.03 0.7 0.38 1 .50 1.84

26.0 23.5 27.2 24.7 26.7 23.4 29.6 35.7 20.7 18.0

volume

Energy (kcal/pair; A X 10”) Keesom Debye London 0 0 0 0 0.008 44 15 0.84 200 450

0 0 0 0 0.13 13 10 4 24 24

3.6 165 650 27 160 265 440 880 165 110

4.1

INTERMOLECULAR AND INTERIONIC FORCES

99

graphic separation in which the only forces are dispersion forces would be the GC separation of alkanes on squalane, a branched paraffin. Table 4.2 contains some typical values for these three van der Wads forces for a few atoms and simple molecules. Beginning with the noble gases that are nonpolar and have no induction or orientation forces, we can see that the polarizability increases as the size increases and that the dispersion forces also increase. The symmetrical molecule hydrogen has no permanent dipole and only a small dispersion force between its molecules. The other molecules in Table 4.2 have permanent dipoles and exhibit all three forces. Note that in the series of hydrogen halides, the dipole moment decreases as the size and polarizability increase; hence orientation and induction forces decrease as dispersion forces increase in the series. Hydrogen Bonding Hydrogen bonds''' are formed between molecules containing a hydrogen atom bonded to an electronegative atom such as oxygen or nitrogen. Such is the case in alcohols, amines, and water, all of which can both donate and receive a hydrogen atom, forming hydrogen bonds. Other molecules such as ethers, aldehydes, ketones, and esters can only accept protons-they have no hydrogens to donate; they can form hydrogen bonds only with hydrogen donors such as alcohols. A common exception is chloroform, which is a weak proton donor. As we have already seen, the strength of hydrogen bonds makes them very important in separation processes and may lead to nonsymmetrical chromatographic peak shapes. For example, the surfaces of many supports used in chromatography contain hydroxyl groups that can form hydrogen bonds and give the surface an undesirable heterogeneity that can cause tailing in GC. Later we will discuss the attempts that have been made to remove these groups by derivatization. In LC one of the popular modes uses mixtures of water and organics as the mobile-phase. Two of the most commonly used organics are methanol and acetonitrile (ACN). While their behavior as mobile-phase components in aqueous solution is usually similar, it is worth noting that methanol can hydrogen bond and ACN cannot. In fact, it has been shown that mixtures of methanol and water produce a ternary mixture in addition to the expected binary mixture.j In the region between 40 and 80% methanol, t h e mixture behaves like a ternary system with the third component being a "The name hydrogen bond is used because these "forces" are strong enough (about 5 kcal/mol) t o be considered wcak bonds-that is, chemical reaction products. The distinction between a force and a bond is not sharp, of coursc, and hydrogen bonds arc classified here with the other

forces.

100

PHYSICAL FORCES AND INTERACTIONS

water-methanol complex that can be as much as 60% of the total. Undoubtedly, this complex forms because of the strong hydrogen bonding between methanol and water. Mixtures of ACN and water also show some association, but not nearly as much, 8% or less. Thus, hydrogen bonding helps explain some of the differences between methanol and ACN when used with water in HPLC. Charge Transfer Finally, there is a group of specific interactions in which two molecules or ions combine by transferring an electron from one to the other. The process is called ~harge-transfer~ and a charge-transfer complex is formed from the attractive forces produced. It too has some characteristics of a bond, exhibiting an internal redox exchange. One of the most common examples in chromatography is the complex formed between Ag’ ions and olefins, which has been used to separate olefins from paraffins.’ A thorough discussion of the general principles of charge-transfer complexes and their uses in GC has been published.”

4.2

SIZE EXCLUSION-MOLECULAR

SIEVING

Although not a “force,” sieuing is another mechanism by which separations can be achieved in chromatography. Probably sieving is not t h e best term to use, but it does denote that separations are made on the basis of the sizes of the sample molecules. In fact, in their most common form, chromatographic separations based on size are achieved by controlling the size of the pores in the stationary phase so that some (small) molecules will be able to enter the pores while others (the large ones) cannot. The large molecules are excluded, which is why this process is correctly called size exclusion and the technique, size exclusion chromatography (SEC). Molecules of intermediate size will be partially excluded from the pores and can be separated from each other based on the fractional extent of their exclusion. In GC this process has been called molecular sieue chromatography, although the mechanism of separation probably involves mainly adsorption as well as some size exclusion. It is used to separate fixed gases such as hydrogen, oxygen, nitrogen, methane, carbon monoxide. ethane, carbon dioxide, and ethylene. The sieves are natural zeolites or synthetic materials of which the alkali metal aluminosilicates are typical. Newer molecular sieves have been especially prepared for GC from carbon. For example, Supelco’ markets its carbon molecular sieves as Carbosieve S-I1 and Carboxen 1000 for separations of permanent gases such as the one shown in Figure 4.5.

SIZE EXCLUSION-MOLECULAR

4.2

SIEVING

101

0, 12.5%)

(2.5%) I

0

I

2

I

I

6

4

I

8

I

10

I

12

Min.

I

14

I

16

I

18

I

20

I

22

I

24

Figure 4.5. Programmed temperature G C separation of gases o n Carbosieve S-I1 column. SS; col. temp.: hold 7 min at 35"C, then to 225°C at 32"C/min. 100/120 Carhosieve S-11, 10' X Flow ratc: 30 mL/min He. Reprinted with permission from the catalog of Supelco, Inc., Bellefonte, PA.

t''

In HPLC the main application has been the characterization of polymers using synthetically prepared stationary phases of varying pore sizes. Current and future applications of SEC are more likely to be to biochemical polymers. SEC was formerly known by as many as 11 different names, including gel filtration and gel permeation chromatography.' Some confusion may result from the use of the latter terms, which are no longer recommended. It must be remembered that some of the solids used as stationary phases in this technique also contain some polar functional groups, and adsorption by analytes does occur in many cases. Thus, the mechanism of their action is often more complex than a simple size exclusion. Elimination of such secondary forces is usually desirable and can sometimes be accomplished by using polar mobile phases. Another separation mechanism that depends primarily on the size of the solute molecules involves the formation of inclusion compounds, clathrates or adducts.' Stable complexes are formed whereby one (called the guest) is trapped inside the other (called the host). Van der Waals forces are present and help stabilize the complexes, and in some cases hydrogen bonds are involved in forming the cages. These methods are not widely used, but they can produce some exceptional separations. A general discussion can be found in references 10 and 11, and some typical applications are in references 12 (GC), 13 (TLC), and 14 (LC).

102

4.3

PHYSICAL FORCES AND INTERACTIONS

SOME MODELS

Applying what we know about intermolecular forces to chromatographic separations is difficult and complex. Some simplification can be achieved by considering GC and LC separately. In GC, the mobile phase is inert, and solutes are sorbed (and desorbed) only in the stationary phase. To achieve a separation they must be retained by the stationary phase to differing degrees. In general, then, the stationary phase is chosen so that sample components will be attracted to it; that is, the stationary phase should interact with the sample, and the modes of interaction are primarily the forces we have been discussing. In simplest terms, we can be guided by the organic chemists’ slogan “like dissolves like,” which means that the sample will interact with the stationary phase to the greatest extent if it is like the stationary phase. For example, one might choose a polyglycol stationary phase to separate alcohols; both the stationary phase and the analytes have hydroxy groups that can form hydrogen bonds as well as van der Waals attractions and produce different retention times for the analytes, thus separating them. There are many examples where this simple generalization holds, as will become obvious as the examples in this book are examined. However, a stationary phase of opposite polarity might be best to separate analytes that differ little in functionality. For example, the relatively nonpolar xylene isomers are polarizable and can be best separated o n a polar column that effects the separation by maximizing the slight differences in polarity among the isomers (see Fig. 4.6). Of course, there are many sample mixtures that contain analytes with a variety of functional groups and polarities, and the choice of a stationary phase is more complicated. The other force operating in GC is the volatility or vapor pressure of the analytes. At any given temperature, the various analytes will desorb (volatilize) from the stationary phase in proportion to their vapor pressures. Once volatilized and in the mobile phase, they will move down the column with the carrier gas, contributing to the desired separation if their vapor pressure is different. The fundamental relationship governing a solute’s vapor pressure is Henry’s law, which states that a solute’s partial vapor pressure is directly proportional to its mole fraction in the solution. This condition applies only over a short range at the lowest mole fractions, which is the situation ideally present in GC. The situation is somewhat different in LC where the mobile phase is a liquid and analytes are attracted to it as well as to the stationary phase. A competition between the phases results, as characterized by the relevant distribution constants. If the stationary phase is a liquid, the situation is similar to a liquid-liquid extraction; if it is a solid, the competition is

4.3 SOMEMODELS

103

AROMATIC POLLUTANTS IN WATER DB-WAX 30 meters x 0.25mrn 1.D. 0.25 micron film 0.5 !rl on-column injection 40' isothermal H, carrier @ 12 p.s.i. Chart speed: lcrn/min.

1. Benzene 2. Toluene 3. Ethylbenzene 4. p-Xylene 5. m-Xylene 6. o-Xylene

4

6

I----

I-

10 min.

Figure 4.6. Separation of xylenc isomers o n a polar column. DB-Wax. Courtesy of J & W Scientific.

between absorption into the mobile phase and adsorption onto the stationary phase. For the popular bonded phases used in HPLC, the nature of the sorption is more complex and will be discussed in Chapter 8. Chapters 7 and 8 contain more information about the choice of stationary and mobile phases, but many separation systems are chosen from experience or by trial and error when theory is inadequate to handle a complex system. In the following section we will explore some of the attempts that have been made to describe the nature of the system from a theoretical point of view. Remember that, ideally, our chromatographic systems can be represented as solutions (GLC, LLC) and/or as surface adsorptions (GSC, LSC). Bonded

104

PHYSICAL FORCES AND INTERACTIONS

phases are probably more like the latter than the former. We will begin our discussion with a consideration of the nature of solutions. Hildebrand's Solubility Parameter

An ideal solution is one in which the interactions between solute and solvent molecules are the same as those for the pure solvent and pure solute. It follows Raoult's law and has no heat of mixing, no entropy of mixing, and no change in volume on mixing. Very few solutions fit this model, so other, less ideal models have been proposed. Hildebrand has defined a regular solutionls as one in which deviations from ideality are attributed only to the enthalpy of mixing; the intermolecular forces are limited to dispersion forces. The equation that defines his model is

where 4 , is the volume fraction of the solvent, V, is the molar volume of solute A, yA is the activity coefficient of solute A, and 6, and 6, are the so-called solubility parameters for the solvent and the solute, respectively. This new parameter, the Hildebrand solubility parameter, is a measure of the internal pressure of a substance, defined as

v

where A is the molar evaporation energy and is the molal volume of the liquid. Clearly, the solubility parameter is a measure of the forces between molecules; it is the square root of the energy of vaporization per milliliter. Table 4.3 lists some solvents according to increasing polarity and includes solubility parameters. Chromatographers have tried to use the solubility parameter to predict chromatographic retention in GClh and in LC." As originally defined, it is applicable only to nonpolar molecules, but modifications have been suggested to adapt it to fit more complicated systems.". I 9 Snyder's Solvent Parameter

In contrast with the absorption (partition) model just described, Snyder2' has provided a model of adsorption in LSC. From his very exhaustive and comprehensive study has come an equation that summarizes the interactions in LSC: log K

,!,I + E , ( S " --As&(])

= log

(4.3)

4.3 SOMEMODELS

Table 4.3

105

Polarity Index and Values for Molecular Interactions for Some Solvents

Solvent n-Hexane i-Octane CCI 4 Toluene Benzene Ethyl ether Methylene chloride i-Propanol n-Propanol Tetrahydrofuran Chloroform Ethanol Ethyl acetate 2-Butanone (MEK) Dioxane Acetone Methanol Acetonitrile Nitromethane Water

P' 0.1 0.1 1.6 2.4 2.7 2.8 3.1 3.9 14.0 4.0 4.1 4.3 4.4 4.7 4.8 5.1 5.1 5.8 6.0 10.2

x,

0.25 0.23 0.53 0.29 0.55 0.54 0.38 0.25 0.52 0.34 0.35 0.36 0.35 0.48 0.31 0.28 0.37

Xd

0.28 0.32 0.13 0.18 0.19 0.19 0.20 0.41 0.19 0.23 0.22 0.24 0.23 0.22 0.27 0.31 0.37

X"

0.47 0.45 0.34 0.53 0.27 0.27 0.42 0.33 0.29 0.43 0.43 0.40 0.42 0.3 1 0.42 0.40 0.25

Group

8 8 1 5 2 2 3

Close to 8 2 6 6 6 6 2 6 7 8

&'I

6

0.00 0.01 0.14 0.22 0.25 0.29 0.32 0.63 0.63 0.35 0.31 0.68 0.45 0.39 0.43 0.43 0.73 0.50 0.49

7.3 7.0 8.6 8.9 9.2 7.4 9.7 11.5 11.5 9.1 9.3 12.7 9.6 9.3 10.0 9.9 14.4 11.7 12.6 21

Large ~

Source: Reproduced from the Journul of' Chrornutogruphic Science by permission of Preston Publications, Inc. Abstracted from reference 21.

where K is the partition coefficient for a given solute, V, is the adsorbent suqace volume o r the volume of a n adsorbed monolayer of mobile-phase liquid, E , is the average surface activity of the solid, S o is the adsorption energy of a solute, A , is the area of solid required by the adsorbed solute, and .z" is the Snyder eluent strength parameter. In comparing the effects of two different mobile-phase liquids on a given solute and a given solid stationary phase, it is easily shown that the relative partition coefficients are related to each other according to Eq. (4.4):

T h e ratio of retention volumes in the two solvents depends on their difference in eluent strength (&:) - &!'), since E, and A , are constant. Thus, if o n e has tabulated values for the eluent strength parameter s o , o n e can predict the effect on retention of changing s o . This concept forms o n e basis for the selection of mobile phases in LSC. Snyder has provided a large number of E ' ) values for solutes run on alumina, and some are included in Table 4.3. By definition, pentane has a

106

PHYSICAL FORCES AND INTERACTIONS

value of zero, and increasing values indicate increasing solvent strength. The larger the value of Eli, the less strongly retained is a given solute from that solvent. When E ' ) values are applied to other adsorbents, they should be multiplied by the relative E,, values suggested by Snyder: = 0 . 7 7 ~ "(alumina)

E"

(silica)

E"

(magnesia)

E0

(Florisil) = 0 . 5 2 ~ "(alumina)

=

0 . 5 8 ~ "(alumina)

(4.5)

(4.6)

(4.7)

While the Hildebrand and Snyder parameters are useful in predicting chromatographic behavior, they are not measures of the exact same concept, and their values are not completely consistent. There are other ways to investigate intermolecular forces. One fairly successful way has been to use a few selected solutes as probes. By observing the degree of retention of these probes in a given chromatographic system, one can get a measure of the forces that caused the probe to be retained. For example, if a hydrocarbon is run in a given chromatographic system, its retention will be determined mainly by dispersion forces and perhaps induction forces, depending on the system. Other, more polar probes can be used to measure other forces. This approach has been used in both GC and LC; one example that originated in LC measurements by Snyder has produced another set of solvent strength parameters. It is called the polarity index, P'. Snyder's" procedure for characterizing LC mobile phases was to use Rohrschneider's data and calculate partition coefficients corrected for dispersion interactions and molecular weight effects. As originally defined in Chapter 2, the partition coefficient K,. is Kc =

conc. of probe in stationary phase conc. of probe in mobile phase

To correct for dispersion and molecular weight effects, a new partition coefficient K" was defined as log K i

=

log K ;

-

log K ,

(4.9)

where K ; = KgV, and K , is a K ; value for a hypothetical n-alkane with the same molar volume as the probe. Assuming that there are three additive molecular interactions (proton donation, proton acceptance, and strong dipole), the corrected partition coefficients for each effect would add up to a total polarity index P':

P'

=

log( KF)d

+ log( KF), + log( K i ) , ,

where subscripts d, e, and n refer, respectively, to the three forces.

(4.10)

REFERENCES

107

This polarity index measures the intermolecular attraction between a solute and a solvent, whereas the Hildebrand solubility parameter is defined for pure solvent. For example, ether is not very polar and has a Hildebrand value of 7.4-about the same as hexane, which has a value of 7.3. However, ether can accept protons in the form of hydrogen bonds to its nonbonding electron pairs, and consequently its polarity index is 2.8 compared to 0.1 for hexane. Comparisons of the polarity index and the Hildebrand solubility parameter can be seen in Table 4.3. Next, Snyder chose three chemicals, one to reflect each of the three forces he had selected. To measure proton donation (acidity) h e chose ethanol as the probe; for proton acceptance (basicity), dioxane; and for dipolar attraction, nitromethane. For example, the selectivity parameter for measuring proton donation, xcl,is defined as:

(4.11) The other two parameters are defined similarly; the sum of the three parameters is thus normalized to 1. Values for some common solvents are listed in Table 4.3 (along with the Hildebrand solubility parameters and the Snyder solvent strength parameters). This discussion is continued in the following chapters: G C in Chapter 7 and HPLC in Chapter 8. T h e bottom line is that solvent/stationary phase selection is often largely guesswork.

REFERENCES 1. F. Pacholec and C. F. Poolc, Chromatogrqihia 1983, 17, 370-374. 2. J. L. Anderson and D. W. Armstrong, Anal. Clzem. 2003, 75, 485 1-4858. 3. R. P. W. Scott, Silica Gel and Bonded Phases: Their Production, Properties, arid Ul,e in LC, Wiley, New York, 1993, pp. 234-236. 4. S. P. McGlynn, Cheni. Keu. 1958, 58, I 1 13. 5. B. L. Karger, Anal. Cliern. 1967, 39(8), 24A. 6. R. J. Laub and R. L. Pecsok, J. Chromatogr. 1975, 113, 47. 7. Supelco, 595 North Harrison Road, Bellefonte, PA 16823-0048. 8. D. M. W. Anderson, F. C. M. Dea, and A. Hendrie, Talunta 1971, 18, 365. 9. M. Baron, in Physical Methods in Chemical Anuhsis, Vol. 4, W. G. Berl (ed), Academic, New York, 1961, p. 223. 10. E. C. Makin, in New Deue1opnierit.s in Sepmztion Methods, E. Grushka (ed), Dekker, New York, 1976. 11. E. Smolkova-Keulemansova, J. Clm~niatogr.1980, 184, 347-361; 1982, 251, 17.

108

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

19. 20. 21.

PHYSICAL FORCES AND INTERACTIONS

A. C. Bhattacharyya and A. Bhattacharjee, Anal. Chem. 1969, 41, 2055. V. M. Bhatnagar and A. Liberti, J . Chromatogr. 1965, 18, 177. J. Zukowski, D. Sybilska, and J. Jurczak, Anal. Chem. 1985, 57, 221552219,

J. H. Hildebrand and R. L. Scott, Regular Solutions, Prentice-Hall, Englewood Cliffs, NJ, 1962. L. Rohrschneider, J . Gas Chromatop. 1968, 6 , 5. P. J. Schoenmakers, H. A. H. Billiet, and L. deGalan, Chromatographia 1982, l S , 205. R. A. Keller, B. L. Karger, and L. R. Snyder, in Gas Chromatography, N. Stock and S. G. Perry (eds), Institute of Petroleum, London, 1970. D. E. Martire and D. C. Locke, Anal. Chem. 1971, 43, 68. L. R. Snyder, Principles of Adsorption Chromatography, Dekker, New York, 1968. L. R. Snyder, J . Chromatop. Sci. 1978, 16, 223.

5 OPTIMIZATION AND THE ACHIEVEMENT OF SEPARATION In the previous chapters we have examined the two factors that must be considered to understand how separations occur. O n e is the kinetic factor, which describes how analyte molecules spread into an increasingly wide zone during their transport through the chromatographic bed. T h e other is the thermodynamic factor, which indicates the interactions between analyte and the chromatographic phases resulting in differential sorption o r retention in the bed. In this chapter we will combine these two factors and see how a separation is achieved. O n e might summarize the process by saying, thermodynamics effects the separation and kinetics aflects the separation. Anyway, for simplicity, the discussion will be limited to column chromatography. 5.1

KINETICS AND ZONE BROADENING

T h e best definition of a measure of relative zone broadening was given in Chapter 3 as:

which expresses the analyte dispersivity per length of column. In that chapter some caution was expressed regarding the interpretation of L ; it is not Chromatogruphy: Concepts and Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley K! Sons, Inc.

By James M. Millei

109

110

OPTIMIZATION AND THE ACHIEVEMENT OF SEPARATION

necessarily the length of the column. If it were, all peaks eluting from a column would have the same peak width (4a)since u=

JHL

and H and L are constants. Perhaps it is obvious that the meaning of Eq. (5.1) is that peak width would depend on retention time t,, since retention time expresses the total time (or number of opportunities) available for zone broadening. This interpretation of L can be shown by combining a few basic equations. The definition of H is

H = -L N

(5.3)

where L is the column length. Since N is defined as

-I:(=.

7

(5.4)

and if t , and a are in the same units (time), the substitution of Eq. (5.4) into Eq. (5.3) gives

To equate Eq. (5.5) with Eq. (5.1), two different definitions of H , one has to assume that L = t,. Thus, in column chromatography, the interpretation of L is that it represents the time an analyte is in transit, the retention time, and the relationship that is most useful is v=

JHtK

from which we conclude that peak width is proportional to the square root of the retention time. This relationship correctly describes the increase in peak width that is observed as the retention time increases. If a given analyte is run on a longer column under the same operating conditions, its retention time will increase in proportion to the column length. Consequently, we can also state that peak width is proportional to the square root of the column length. Based on that fact alone, one might conclude that separations would be difficult for analytes with long retention times on long columns since the peaks would be too wide. However, we need to examine the consequences of thermodynamics on zone migration before coming to a conclusion.

ACHIEVEMENT OF SEPARATION

5.3

5.2

11 1

THERMODYNAMICS AND ZONE MIGRATION

From the thermodynamic relationships presented in Chapter 2, we know that t - - '=M t , ( l + k ) K-

R

=i,[l

$1

+K

For a given column (constant t , and /3 1, the retention time of an analyte will depend upon its distribution constant K , ; specifically, retention time will increase linearly with K , . If we now consider two analytes, A and B, which we wish to separate, we can know their relative retention times from their respective partition coefficients. Thus, on a given column, their peak maxima will be separated by a distance d : d= =

[%I(

(t,<)IJ - ( f R ) *

K,

-

K.4 )

(5.9) (5.10)

O n a longer column their retention times will be increased and so will their distance of separation; that is, the separation distance is directly proportional to the column length:

daL

(5.11)

This increasing distance of separation between peaks makes it possible to get a separation. 5.3 ACHIEVEMENT OF SEPARATION

We have seen that peaks get wider in proportion to the square rout of column length but that two peaks are separated in direct proportion to the column length. Since the resolution R , between these two peaks is given approximately by

Rs

=

wd

(5.12)

we can substitute the above relationships to get

L R,a-=&

&

(5.13)

That is, resolution is proportional to the square root of the column length.

112

OPTIMIZATION AND THE ACHIEVEMENT OF SEPARATION

z

L Figure 5.1. Achievement of separation. Adapted from J. C. Giddings, Dynamics of' Chromulogruphy, Part 1, Marcel Dekker, New York, 1965, p. 33. Courtesy of Marcel Dekker.

This relationship can also be shown graphically (Fig. S.l), where the distance d and the peak width 4 a must be in the same units. At some column length indicated by the dashed line, d will exceed 4a,the peak width, and the two analytes will be separated according to the definition of resolution [Eq. (S.l2)]. Our conclusion is that, as long as two analytes have some difference in their partition coefficients, it must be theoretically possible to make a column long enough to separate them. In practice, of course, this is not usually the easiest way to get a separation. Long columns have high pressure drops that could be impractical. The broader subject of column permeability or separation impedance is introduced in the next chapter and continues in Chapter 8 because it is of greater relevance in HPLC than in GC. Often, the bottom line is that the most important practical limitation is pressure drop. 5.4

OPTIMIZATION OF SEPARATIONS

What are the best ways to optimize separations? One way to answer this question is to consider a popular form of the resolution equation that is originally attributed to Purnell.'

(5.14) The three variables are N (column efficiency), a (column selectivity), and

5.4

OPTIMIZATION OF SEPARATIONS

113

k , , which is the retention factor for the second of the two analytes being separated. (Several similar equations are also in common use, and the relationships among them have been discussed by Said.?) Since good resolution is indicated by a large value for R,, the three terms, i%,( a - ] ) / a ,and k / ( l + k ) should be maximized. However, time is a fourth variable that also needs to be considered. That discussion will follow our consideration of Eq. (5.14). The plate number can easily be increased by increasing the length of the column; we have just seen how that concept works. However, the resolution is increased only by the square root of the increase in length at the expense of a linear increase in time. For this reason alone, longer columns are not usually the preferred method for improving resolution. Alternatively, N can be increased by preparing a better column according to the rate equation. The most important parameters are stationary film thickness and particle diameter, both of which should be kept small as described in Chapter 3. The next term, containing a , the selectivity or separation factor, will be maximized if a is maximized. In effect this means that the separation system, the stationary and mobile phases, should be chosen to show the greatest selectivity between analytes A and B; our knowledge of intermolecular forces can help us in selecting the best system. Typical values of a range from just over 1 to nearly 2 (which would be an easy separation). Thus the values of the term ( a - 1)/a could vary from 0.001 up to 1, a range of lo', indicating that this term is the most powerful of the three. Finally, to maximize the last term, kB/(l + k , ) , the retention factor k should also be maximized. Since the two analytes are next to each other, k , is not much different from k,, and it is sufficiently accurate to merely indicate that k (either one) should increase. However, a few calculations will show that little improvement in resolution is gained as k is increased above about 10. For example, if k is increased from 1 to 50, the increase in this term is from 0.5 to about 1. The improvement in resolution is not nearly as significant as that for a , and any increase in k denotes a proportionate increase in time. In choosing the optimum value for k , one should consider not only the resolution but also the time. We can get a time-dependent equation by substituting Eq. (5.15) into Eq. (5.14): t,

=

N ( k + l)H -

U

(5.15)

to give (5.16)

114

OPTIMIZATION AND THE ACHIEVEMENT OF SEPARATION

0

2

4

6 k

0

10

12

Figure 5.2. Time-optimized partition ratio.

To find the minimum time for a given resolution, the k term is plotted versus k , as shown in Figure 5.2. The minimum is a k value of around 2, but most chromatographers have found that k values in the range of 1.5-5 represent an optimal range for HPLC. The only disadvantage in higher k values is the additional time required for analysis, and since G C is fast, many G C runs are made at higher k values. In practice, the k value is adjusted by varying the temperature or by varying the phase volume ratio ( p ) by changing the amount of stationary phase. To increase k at constant temperature, 0 must be decreased. Unfortunately, for many columns, p is not known and changing the amount of SP is not possible. Also, in HPLC, small changes in the composition of the mobile phase can be used to change k . A recent work by Chester3 reviews the literature on the methods for choosing experimental conditions for HPLC and provides a computer-based, multivariate optimization procedure based on dozens, if not hundreds, of trial solutions, the equivalent of months of laboratory work. They demonstrated how complex is the process of optimization of HPLC, even for isocratic separations when only 3 or 4 parameters are varied. Equation (5.14) can be rearranged and used to calculate the number of plates required for a given separation:

(5.17) Table 5.1 gives a number of typical values of N calculated with Eq. (5.17). It can be seen how quickly N decreases as k is increased up to about 5 , and how a small change in a from 1.05 to 1.10 decreases the plates needed by a factor of about 4.

1 15

REFERENCES

Table 5.1 Plate Number Required to Achieve a Given Resolution (in thousands, rounded to 3 significant figures)

Retention Factor, k , , 0.1 0.2 0.5 I .0 1.5 2.0 5.0 10.0 20.0 X

R, N =

1.05

1,920 572 143 63.5 44.1 35.7 22.9 19.2 17.5 15.9

=

R,

1.5 CY =

1.10

527 157 39.2 17.4 12.1 9.80 6.27 5.27 4.80 4.36

cr= 1.05

854 254 63.5 28.2 19.6 15.9 10.2 8.54 7.78 7.06

=

1.0 N =

1.10

234 69.7 17.4 7.74 5.38 4.36 2.79 2.34 2.13 1.94

One final variation of Eq. (5.14) can be derived if it is assumed that all peaks in a given separation have the same plate number:

(5.18) Resolution is maximized when the retention factors are small and the difference between them is large. This equation is not as useful as Eq. (5.14) because it does not contain CY which is the most important parameter, as we have seen. REFERENCES 1. J. H. Purnell, J . Chem. Soc., 1960, 1268. 2. A. S. Said, HRC CC, J . High Resolut. Chromutogr. Chromutogr. Commun. 1979, 2, 193- 194. 3. T. L. Chester, J . Chromutogr. A 2003, 1016, 181-193.

This Page Intentionally Left Blank

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES Before proceeding with discussions of the specifics of individual chromatographic techniques, it will be helpful to take a look at some of the differences among various chromatographic modes. The emphasis in this monograph has been on similarities among modes, but there are significant differences that should be examined. To start, the most obvious comparison to make is between GC and LC, whose differences can largely be attributed to the differences between gases and liquids. However, that discussion will naturally lead to another phase that has been largely ignored thus far in this book; it is the supercritical state. Supercritical fluid chromatography (SFC) was included in Figure 2.2, where it is appropriately positioned between GC and LC. It was also mentioned in Chapter 2 as the technique called by some unified chromatography. In this chapter, SFC will be briefly discussed after an introduction to supercritical fluids. Finally, this chapter will end by contrasting LC in columns and LC on planar surfaces.

GAS CHROMATOGRAPHY COMPARED TO LIQUID CHROMATOGRAPHY

6.1

Our comparison of GC and LC begins by examining the parameters that are most important in each technique. The two most important parameters in Chromatogruphy: Concepts and Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 J o h n Wiley & Sons, Inc.

By J a m s M. Miller

117

118

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

G C are (1) t h e nature of the stationary phase (SP) and (2) temperature. In LC they are (1) the nature of the stationary phase, (2) the nature of the mobile phase (MP), and (3) to some extent temperature. While temperature is important in both techniques, it is more critical in G C because GC is limited to volatile samples (those that boil below about 500°C or have a vapor pressure of several kilopascals* or more) that are also thermally stable. This limitation is not imposed on LC, which consequently has wider applicability. Also, LC would seem to have wider applicability because both the MP and SP interact with the analytes, whereas for G C only the SP does so. Giddings has written extensively on this subject and has provided a thorough discussion in his book.’ First of all, what he refers to as the “density of intermolecular attraction” is much lower for gases than for liquids-by about lo4. This means that the mobile phase in GC does not interact with analytes and does not appreciably cause them to desorb from the stationary phase. It simply carries them down the column when they are in the vapor state. This lack of intermolecular attraction is normally desired in G C where the gases used (He and N?)are chosen for their inertness. In LC, the mobile-phase liquids compete actively with the stationary phase to attract the analytes. To get this enhanced selectivity in GC, some workers have tried adding condensable gases to their mobile phases, similar to the approach taken in SFC. However, the main effect has been a partial deactivation of the stationary phase’ rather than the desired change in mobile-phase composition. (See also Chapter 7.) For further comparisons between GC and LC, we need to examine the differences between gases and liquids. Gases Compared to Liquids

A further comparison of the two major types of chromatography, GC and LC, reveals that the major differences can be attributed to the differences in properties between gases and liquids, such as: 1. The diffusivity of analytes in liquids is less in than in gases by a factor of about This was noted in Chapter 3. The slower diffusion in liquids results in slower speeds of analysis in LC and in a significant contribution by the mass-transfer term C, in causing zone broadening. 2. The viscosity of liquids is greater by a factor of about l o 2 . One consequence is the higher operating pressures required for HPLC. *The standard unit of pressure in S1 is the pascal, Pa. For those familiar with old units, the following conversions are given: 1 bar = 100 kPa; I atm = 101.3 kPa; 1 torr = 133Pa; 1 psi = 6.9 kPa.

61

GAS CHROMATOGRAPHY COMPARED TO LIQUID CHROMATOGRAPHY

1.o

1.o

0.8

i \

.-

0.6 0.4 0.2

0

119

-

0.8

:&: -

2

0.6

3

0.4

4 5

I

I

I

I

I

I

I

0.2 I

I

0

3. The surface tension of liquids is greater by about lo4. (Gases can be considered to have no surface tension.) The surface tension of liquids makes possible ascending TLC and PC due to the driving force of capillarity. 4. The density of liquids is greater by about lo3. 5. Liquids are only slightly compressible, and gases are readily compressed. This has a major impact on GC because the mobile-phase carrier gas is compressed at the head of the column, resulting in a variable flow rate across t h e length of the column. We need to take a closer look at this phenomenon and see what the consequences are for computing GC flow rates

Gas Compressibility in GC

Figure 6.1 shows the effect of the pressure gradient in the column on carrier gas flow. The carrier gas is under higher pressure at the column inlet (Pi) than at the column outlet ( P o ) ,which is usually at atmospheric pressure. The various curves show that the changes in flow rate are greater at the higher ratios of inlet-to-outlet pressure (P,/P,>. Thus the least change in flow is encountered when the column pressure drop is least, as is the case with open tubular columns. Usually the flow rate is measured at the exit at P,,, where the rate is at its maximum. To determine the average flow in the column, this maximum value

120

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

Table 6.1

1.1 1.2 1.3 1.4 1.5 1.6 1.7

Some Pressure Correction Factors ( j )

0.95 0.91 0.86 0.83 0.79 0.76 0.72

1.8 1.9 2.0 2.2 2.5 3.0 4.0

0.70 0.67 0.64 0.60 0.54 0.46 0.36

must be multiplied by a correction factor called the compressibility, j :

Table 6.1 lists some values for j, and it can be seen that this is a significant factor. An update to the original IUPAC terms and definitions notes that there several different j values in use.3 The one defined in Eq. (6.1) is the original James and Martin factor, and it is the only one we will use.* If the symbol F, is used to denote the exit flow at column temperature, then the average flow is

F, =jF, Similarly, Po Average pressure = P- = 7 J

Average linear velocity = ii = j u Corrected retention volume = V i =jVR =jFctR

(6.3)

(6.4)

(6.5)

(6.6)

As indicated in Eq. (6.5), when the compressibility factor is applied to the retention volume, the adjective corrected is used and a superscript zero is added to the symbol. The same convention applies to all gas volumes, such as the mobile-phase volume, which can be denoted by a subscript M ( V A =jV,) or a subscript G for gas (V: =jV,). The IUPAC report3 also suggests that j only be applied to volume measurements because it is meaningless when *The IUPAC’ recommends that the compressibility be designatcd as j : , but we will use only the symbol j in this text.

6.1

GAS CHROMATOGRAPHY COMPARED TO LIQUID CHROMATOGRAPHY

121

applied to time measurements. Thus, the corrected retention volume is correctly defined in Eq. (6.51, but no equation is given for corrected retention time because there is no such term. The flow at the outlet, F,, can be measured at ambient temperature with a soap film flow meter. In that case, several corrections must be applied to get the desired flow F,*.

where T, and T, are the ambient temperature and the column temperature, respectively, in degrees Kelvin; P is the atmospheric pressure, and p w is the vapor pressure of water at T,-both pressures are in the same units. In some cases in this monograph the symbol F is used with no special designations to refer to a nonspecific flow rate. When a GC retention volume is both corrected (for average flow) and adjusted (by subtracting out V,), a new name and symbol are used:

jV'

R

=

v; v" -

G

=

VN

(6.8)

where V , is called the net retention volume. Since an analogous situation does not exist in LC, it is possible that some confusion may arise over the use of different terms. Thus, we originally used the equation

VA = KV,

(6.9)

in our definitions in Chapter 2, and that was satisfactory for LC, but we should use V,

=

KV,

(6.10)

for GC. Often Eq. (6.10) and the term net retention volume are also used for LC, just to reduce confusion and make one equation serve both. No confusion should arise if the equations are read in context. Permeability

Permeability is a characteristic of columns that differs between G C and LC columns. When a column has a low permeability, high pressures are required for operation, and flow rates may be limited to lower values. Differences between liquids and gases are mainly responsible for low permeability in packed HPLC columns, compared to packed GC columns. Permeability K is defined as (6.11)

122

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

where 7 is the viscosity, L is the column length, cl is the average mobile-phase velocity, and A P is the pressure drop across the column. The average mobile-phase velocity can be measured as L / t M , and substituting this value into Eq. (6.11) gives an equation from which permeability can be experimentally measured: (6.12) However, this equation does not include one of the most important LC variables, the particle diameter d,. Consequently, another parameter has been defined4 that does include it. It is called the flow resistance parameter, @:

(6.13) where @ is a dimensionless parameter, and in the calculation of @ it is important to use the proper units; those most commonly used by chromatographers are A P in bar ( l o 5 Pa), tM in seconds, d , in micrometers, 7 in millipascal seconds (mPa s), and L in millimeters. For further information including some sample calculations, see Meyer.' The most permeable columns are t h e most desirable, and they will havc the smallest values of @. Typically HPLC columns will have values ranging from about 500 for spherical microparticles to 1000 for irregular microparticulates. The flow resistance parameter is included in the discussion regarding the evaluation and comparison of HPLC columns that ends Chapter 8. Bristow and Knox4 also defined some other parameters for characterizing HPLC columns, discussing in detail the desirability of using them to standardize test conditions for HPLC. In another study, Knox6 compares the theoretical aspects controlling the performance of small-bore (1 -mm) packed, capillary packed (50 to 200-pm), and capillary OT (10 to 50-pm) columns. That study also contains many references. Efficiency and Speed

Is either G C or LC inherently faster or more efficient? Giddings' attempted to answer that question by deriving the relationship (6.14)

6.1

GAS CHROMATOGRAPHY COMPARED TO LIQUID CHROMATOGRAPHY

123

Log n

Figure 6.2. Comparison of G C and LC efficiencies and analysis times. Reprinted with permission from J . C. Giddings, Anal. Cliern. 1965, 37, 60. Copyright 1965, American Chemical Society.

which compares the limit on the plate numbers for the two techniques. Substituting the appropriate viscosity and diffusivity ratios (discussed earlier) and allowing for the compressibility of gases, he found that

(6.15) where A P is in atmospheres. In effect, if GC and HPLC were run at the same pressures, HPLC would be more efficient by a factor of 10'; or if G C were run at a pressure 10' times as large as HPLC, it would be as efficient as HPLC. The problem with the first conclusion is that the time required to achieve the large plate numbers in HPLC would be very long, as shown in Figure 6.2. (Note that the figure is a log-log plot and the time axis gets large very quickly!) The problem with the second conclusion is that it has been found difficult to run GCs at the high pressures required by the equation. A similar study by Knox and Saleem' to find the optimum speed and resolution in column chromatography came to similar conclusions. One of their conclusions was that, although HPLC should be capable of a greater separating power, the HPLC column would have to be 800 m long and the separation would take 25 years.

124

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

In conclusion, theory has not provided us with much guidance, and the choice between GC and LC is seldom made on the basis of large differences in efficiencies. In most cases, a sample that is amenable to GC-that is, one that is volatile and stable-can be run faster and more easily by G C than by HPLC despite the prediction that HPLC should be more efficient. But, of course, those organic compounds that do not meet these criteria can be analyzed with HPLC, which may also offer other advantages such as selectivity, detectivity, and minimal sample preparation.

6.2 SUPERCRITICAL FLUIDS AND SUPERCRITICAL FLUID CHROMATOGRAPHY

There is a region in a gas-liquid phase diagram where a substance can exist with properties characteristic of both the gas and liquid states. This state is referred to as the supercriticalfluid state since it exists in the region beyond the critical points for the substance. Visualize a transparent container half filled with liquid and sealed. As its temperature is increased, its pressure will increase until the critical point is reached. At that point, the meniscus between the two phases disappears and the material in the tube is in the supercritical phase. A supercritical fluid is neither a gas nor a liquid but a truly unique phase with properties intermediate between the two. The properties of supercritical fluids fall between those of gases and liquids, as shown in Table 6.2. A typical phase diagram is shown in Figure 6.3 for carbon dioxide, indicating the critical point. In the mid-l960s, as high pressures became common in HPLC, it was only natural to see if the supercritical region had any special chromatographic properties. It did, and the field of supercritical fluid chromatography was born. As one would expect, SFC is a hybrid of G C and LC. Thus, the mobile phase in SFC has a viscosity only slightly greater than the gases in G C while showing much more affinity for analytes because of its higher density. On the other hand, a supercritical fluid has a higher diffusion coefficient than a

Table 6.2 Comparison of Typical Values of Some Important Chromatographic Parameters

Approximate Value

Parameter

GC

SFC

LC

Diffusion coefficient (cm*/s) Density (g/mL) Viscosity (g/cm. s)

lo-' 10-3 10-4

10-4 0.8 5 x 10-4

10-5 1 10-2

6.2

SUPERCRITICAL FLUIDS AND SUPERCRITICAL FLUID CHROMATOGRAPHY

125

7.4

0.54

-57

Temperature (“C)

31

Figure 6.3. Phase diagram of carbon dioxide

liquid, making SFC more efficient than LC. In some respects, SFC has the best characteristics of both GC and LC. Some chromatographers would say that SFC represents the ideal compromise between GC and HPLC, calling it unified chromatography (UC). Their definition of UC is that it is a technique performed with control of the column outlet pressure and the column temperature, regardless of the fluid state of the mobile phase.’ This definition does not limit UC to SFC but rather acknowledges that there is a middle region that bridges GC and HPLC. Martire9 has published a generalized theoretical treatment applicable to all three techniques-GC, LC, and (packed column) SFC-and Chester ‘(I has discussed UC from the perspective of the mobile phase. He argues that it is not desirable to distinguish SFC as a special technique but rather as one part of a continuum of mobile-phase conditions. A symposium at the 1988 ACS meeting in Boston was devoted to UC and the 12 papers presented there have been published together as a testimony to the concept of UC.” As will be seen, carbon dioxide is the mobile phase most often used in UC, and consequently the range of applications of UC is quite limited. Carbon dioxide-based binary mobile phases have been used to remedy this limitation, and they have been called the “tunable” solvents that characterize UC.’* This study I 2 also suggests that “probably the most significant contribution of the idealized concept of UC is to reinforce the vital theoretical and conceptual links between analytical and physical chemistry.” Surely, this is an interesting topic to pursue, but the remainder of this section will be devoted exclusively to a brief consideration of SFC as classically defined.

126

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

The development of SFC up to 1983 has been briefly described by Gere," showing the evolution in concept and the early applications. The first commercial instrument was offered for sale in that year, giving considerable impetus to the field, and the number of publications in SFC rose quickly in the 1980s. By 1984, at least one 1aborato1-y'~ felt that the technique had become "routine." Supercritical fluid chromatography instruments are of two types, depending on the type of column used. If packed columns are used, the instrument is more like a liquid chromatograph with a reciprocating pump and a pressurized ultraviolet (UV) detector. If OT columns are used, the instrument is more like a gas chromatograph with a syringe pump and an FID (flame ionization detector). In both types, a pressure restriction is needed somewhere after the column to keep the pressure above the critical value. In some cases it comes after the detector, which is then operated much as it is in HPLC, except that it has to have the capability of withstanding the higher pressure. If it comes before the Table 6.3 Properties of Possible Mobile Phases for SFC

Compound Nitrous oxide

Carbon dioxide

Sulfur dioxide Sulfur hexafluoride Ammonia Wate

Methanol

Ethanol Isopropanol

Ethane n-Propane n-Butane n-Pentane n-Hexane n-Heptane 2,3-Dimethylbutane Benzene Diethyl ether Methyl ethyl ether Dichlorodifluoromethane Dichlorofluoromethane Trichlorofluoromethane Dichlorotetrafluoroethane

Atm bp ("C)

T, ("C)

89 78.5" - 10 - 63.8" - 33.4 100 64.7 78.4 82.5 - 88 - 44.5 - 0.5 36.3 69.0 98.4 58.0 80.1 34.6 7.6 29.8 8.9 23.7 3.5

36.5 31.3 157.5 45.6 132.3 374.4 240.5 243.4 235.3 32.4 96.8 152.0 196.6 234.2 267.0 226.8 288.9 193.6 164.7 1 11.7 178.5 196.6 146.1

-

-

~

Critical Point Data P, (MPa) d, (g/mL)

Source: From Gouw and Jentoft, Adu. Chromatogr. N.Y. 1975, 13, 1 "Sublimation point.

7.23 7.38 7.86 3.76 11.3 23.0 7.99 6.38 4.76 4.89 4.25 3.80 3.37 3.00 2.74 3.14 4.89 3.68 4.40 3.99 5.17 4.22 3.60

0.457 0.448 0.524 0.752 0.24 0.344 0.272 0.276 0.273 0.203 0.220 0.228 0.232 0.234 0.235 0.241 0.302 0.267 0.272 0.558 0.522 0.554 0.582

6.2

SUPERCRITICAL FLUIDS AND SUPERCRITICAL FLUID CHROMATOGRAPHY

127

detector, some problems have arisen as the depressurized effluent enters the detector. Noise is generated and is believed to originate with the formation of small nonvolatile particles; those can also plug the transfer lines. The initially high expectations for SFC were not met, and SFC did not replace HPLC. As the field has shaken down, packed-column methods have become more popular, and the instrumentation generally resembles that of HPLC. There is some feeling that SFC is currently making a comeback,Is led by the pharmaceutical industry. Mobile-Phase Properties

Some chemicals that could be used in SFC are listed in Table 6.3. The one that has been used most commonly is carbon dioxide, and it will be the focus of this short introduction. Figure 6.4 shows the pressure-volume phase diagram for CO, at various temperatures. The critical values (P, = 7.4 MPa,

Volume (mL)

Figure 6.4. Pressure-volurnc diagram for carbon dioxide

128

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

Pressure, bars

740 400 370 300 220 150

74 50 36

I

0.25

I

I

I

0.50

0.75

1.00

Density, g per mL

I

I t

1.25

Figure 6.5. Pressure-density diagram for carbon dioxide. Reprinted with permission from Chem. Eng. News 1982, 60(12), 46. Copyright 1982, American Chemical Society.

= 96 mL, and T, = 31°C) intersect at the point marked X. Liquid exists in the lined space at the left of the diagram, gas and liquid are in equilibrium in the space cut off by the dashed line, supercritical fluid exists above the critical temperature, and gas exists at the right. Remember that the critical temperature is that temperature above which a gas cannot be liquefied n o matter how high the pressure. If a chemical such as CO, is above its critical temperature and the pressure is raised, the density of the fluid increases, as shown in Figure 6.5. The operating region for SFC is in t h e center of the figure, above the gas-liquid equilibrium region. Most systems are run at constant temperature; if the pressure is increased during a run, the density will increase according to the isotherms shown. Computer software control of the pump can linearize density changes to achieve linear density programming. The viscosity of a supercritical fluid is constant at constant density, regardless of the temperature and pressure. It increases with density, but it is still lower than liquid viscosities by a factor of 10 or more. As a consequence, pressure drops across SFC columns are smaller than across HPLC columns, requiring less pressure for a given flow and making high velocities realistic. OT columns can be operated at pressures only slightly above the critical value of 7 MPa (73 atm or 1100 psi), and even packed columns show pressure

6.2

SUPERCRITICAL FLUIDS AND SUPERCRITICAL FLUID CHROMATOGRAPHY

129

GLYCERIDES (OLEATES)

no MeOH

1% MeOH

2"* di ( 2 )

b/

0

1

I

I

1

2

3 la)

I

Irnin.

--

0

1

2

(bj

Figure 6.6. Effect of methanol as a mobile-phase modifier on the SFC separation of mono-, di-, and triglycerides of oleic acid. Reprinted from Am. Luh. 1984, 16(5), 19. Copyright 1984 by International Scientific Communications, Inc.

drops as low as 1.5 MPa, resulting in inlet pressures around 8.5 MPa (1300 psi). These values are readily achieved with available pumps and are not the dangerously high values originally anticipated for SFC. Carbon dioxide is not polar (about the same as hexane), and this characteristic has been a limitation in running polar analytes. Table 6.3 shows that there are not very many polar chemicals that could be considered for practical use in the supercritical region. A popular alternative is to add methanol or acetonitrile as a modifier to the carbon dioxide. Figure 6.6 shows the effect of adding 1% methanol to the CO,. in the separation of glycerides; peak shapes are improved and the retention times are decreased. The pros and cons of using C0,-based binary mobile phases has recently been summarized." For many applications using packed columns, they perform like normal-phase systems in HPLC and thus serve to complement reversed-phase HPLC for separations of nonpolar analytes.

130

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

Stationary-Phase Properties

Column stability requires that the stationary phases be bonded to the column wall (OT) or the solid support (packed). Early open tubular G C columns were not stable. Currently available HPLC packings and cross-linked open tubular GC columns are usually satisfactory, but some columns are expressly made for SFC.I6 High efficiencies have been obtained by using long packed columns or several conventional HPLC columns in series. Berger and Wilson achieved a plate number of 260,000 by using 10 columns in series.17 Using long columns or several in series is possible in SFC because of their low pressure drop, unlike the situation for HPLC where the viscosities of thc MP are much higher. Some problems have been experienced due to swelling of the stationary phase by the supercritical fluid. This is especially serious for columns that have high stationary loads, and it adversely affects the column pcrformance.lx Attempts continue to be made to find phase systems that can be used to analyze polar compounds. Since carbon dioxide is the principle MP, and it is not polar and not easily modified, the alternative is to make the SP less polar. A summary of the different approaches to this problem has been published,'" along with a discussion of the overall problem. Instrumentation

Instruments designed for use with packed columns have become the more popular type.Is Essentially they resemble HPLC instruments and use UV as the primary detection mode. Gradient elution SFC is possible when packed columns are used. Five instrument manufacturers are included in a 2002 product review,15 all using packed columns, and a fifth manufacturer can supply an instrument for both packed and OT columns. All of them also permit the use of preparative or semipreparative columns. Outlet pressures range from 90 to 600 bar and flow rates from 0.1 to 50 mL/min. Although they are expensive, costing around $60,000, most can be interchangeably used for HPLC and can be considered to have dual capability. The same cannot be said for HPLC instruments.

"'

Density Programming

Compared with the two parameters in GC and three in LC listed at the start of this section, SFC has a fourth variable, density. Since analyte solubility increases with increased MP density (molecular closeness), density programming produces effects similar to programmed temperature in GC. It is especially beneficial in polymer separations. Density programming is facili-

6.2

SUPERCRITICAL FLUIDS AND SUPERCRITICAL FLUID CHROMATOGRAPHY

0.4

\

m

t

0.2

0.1

i

131

Program: Asymptotic Density

1

1

-

o i

0

1

TIME- hours

Fluid:

2

Supercritical n-Pentane

Column:

at 210°C lorn x 100pm, i.d. SB-

Detector:

Phenyl-50 UV Absorbance

Figure 6.7. Example o f asymptotic density programming in SFC. Polystyrene oligomcr mixture with avg. MW of 2000. Reproduced from the .Jmmd o / C h r n m t r t o ~ ~Science ~ ~ ~ ~by~ jpermission ~. o f Preston Publications, Inc.

tated if the column has a low pressure drop, which favors the use of OT columns. Pulseless pumps are also desirable. Asymptotic density programming, rather than linear programming, produces equal spacings between homologs, as shown in Figure 6.7. Applications

Supercritical fluid chromatography is best suited to separate materials that are not volatile enough or thermally stable enough for GC and are not easily handled by HPLC because of problems like suitable detectors. Thus highmolecular-weight materials that do not absorb UV and cannot be used with UV detectors in HPLC are commonly run by SFC with FID detection. One example is shown in Figure 6.8; the oligomers in a polymeric detergent mixture are separated using pressure programming. Coal tar is a mixture that needs the high efficiency of OT columns and approaches the upper temperature limit of GC. Figure 6.9 compares a separation of coal tar components by GC and SFC. As expected, the GC is

132

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

COLUMN: 06-5 (0.20 MICRON FILM) QM X 50 MICRON MOBILE PHASE: CARBON DIOXIDE 175OC PROGRAMMED FROM 110 ATM TO 2 3 0 ATM AT 30 ATM MINUTE-$. FROM 2 3 0 ATM TO 340 ATM AT 6.5 ATM MINUTE-1, AFTER 4 MINUTE ISOBARIC PERIOD.

I

I

110 230

110

I

I

0 10

I

340

PRESSURE (ATM) I

20

1

30

TIME (MIN.)

Figure 6.8. SFC separation of an ethoxylatcd amine using pressure programming. Courtesy of Suprex.

performed with programmed temperature and the SFC with programmed density, as indicated in the figure. Berger2’ has reviewed the separations of polar solutes using packed columns and has pointed out the practical advantages of packed-column SFC in supporting combinatorial chemistry.’2 Many chiral columns used in HPLC have been found to work well for SFC, some even better than for HPLC. Chiral separations are treated in Chapter 15, but four studies are included in volume 785 of the 1997 Journal of Chromatography, which is devoted entirely to SFC and supercritical fluid extraction (SFE).” Preparative separations are also possible and most instruments can accommodate the larger columns required. Supercritical fluid extraction is another application of the use of supercritical fluids. It is discussed in Chapter 14.

6.2

SUPERCRITICAL FLUIDS AND SUPERCRITICAL FLUID CHROMATOGRAPHY

c

L 0

eb

3

a9 ’@ I

ii,

TIME (min)

TEMP

F

133

50

250

(.C)

9 SFC

8

0

i

0.225

d TlME (min)

DENSITY (ghL)

0.70

120 I

(bl Figure 6.9. Comparison of a coal tar separation by ( a ) G C and ( h ) SFC. Both separations performed on SE-54 OT columns with 0.25-pm film thicknesses. G C column 20 m X 300 p m , temperature programmed from 40 to 265°C. SFC column 34 m X 50 p m , density programmed. Reprinted with permission from J . C. Feldsted and M. L. Lee, Anal. Chem. 1984, 56, 619A. Copyright 1984, American Chemical Society.

134

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

Table 6.4 Advantages and Disadvantages of SFC

Advantages

Disadvantages

Faster than HPLC Higher efficiency than HPLC Lower pressure drop allows longer columns or multiple columns Operates between G C and LC and is therefore complementary to them Additional flexibility provided by density programming

Slower than G C Few choices for mobile phases Expensive instrumentation

Summary Smith24 has written a very comprehensive review of SFC covering 349 references. Also, Analytical Chemistry has included SFC in its biennial reviews since 1990. Most of them included SFC and SFE” but the last three, in 2000,2h 2002,27 and 2004,2x the coverage has included SFC and UC, but not SFE. The advantages and disadvantages of SFC are listed in Table 6.4. Just as supercritical fluids have properties intermediate between gases and liquids, so SFC is intermediate between GC and HPLC. As stated earlier, it appears that the technique is having a revival, but only time will tell. 6.3

REDUCED PARAMETERS

Reduced parameters were introduced in Chapter 3 to facilitate the discussion of the rate equation. They are (6.16) and (6.17) Their use normalizes the variability of particle size and diffusion rates when comparing GC and HLPC, and provides a uniform reduced rate equation:

h

=

B

; + ( c, + c, ) 1,

(6.18)

Consequently, a plot of Eq. (6.18) gives a unified graph, allowing a rational comparison despite the major differences between GC and HPLC. See Chapter 3 for more details.

6.4

6.4

COLUMNAR AND PLANAR CONFIGURATIONS

135

COLUMNAR AND PLANAR CONFIGURATIONS

In GC, the stationary phase is always packed in a column in order to contain the mobile phase, a gas. A similar configuration is common in LC, but since the mobile phase is a liquid, it is also possible to use another configuration-one in which the stationary phase is spread on a flat, planar surface. This configuration is called either paper chromatography (PC) or thin-layer chromatography (TLC), depending upon the stationary phase. In TLC, the stationary phase is coated on a supporting planar surface, which can be glass, plastic, or metal. Many of the stationary phases used in columnar LC can also be used in TLC, so it is interesting to compare these two techniques. As a background for that comparison, we first need to define a few other terms and symbols. Peaks Compared to Bands Up to this point it may have appeared that the terms hand, peak, and zone have been used synonymously, but let us take a closer look. As commonly used, all three terms describe the distribution of analyte molecules in space (a concentration profile), but hand represents this distribution while the analyte is still in (or on) the system, while peak refers to a distribution of analyte that has eluted from the system. The term zone is more general and includes both bands and peaks; it is used in those cases when we d o not wish to be more specific. In the context of our current discussion, this means that the zones in TLC and PC are called bands and those in column LC and in GC are called peaks. An attempt to visualize the difference between bands and peaks is given in Figure 6.10, which shows the separation of two analytes, A and B by TLC in ( a ) and by column LC in (6). The chromatogram in ( a > is obtained by densitometering the plate shown above it, and the chromatogram in ( b ) is from a conventional online detector. The partition coefficient for A, K,, is greater than the partition coefficient for B, K,. Consequently, B migrates faster than A, and as shown in Figure 6.10a, B has moved farther down the bed than A; in Figure 6.10b, B is shown eluting from the bed before A. As noted in the figure, chromatograms of the two situations show the zones in reverse order and width. Thus, zone broadening does not depend on the length L of the bed, as suggested by the definition of column dispersivity introduced in Chapter 3:

H = -v L L

(6.19)

Rather, band dispersivity is a function of the distance migrated, L,, in the

136

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

(a) Bands

I

I

I I 1

wA

> wB

i I

B

I

I Time

+

(b) Peaks

Figure 6.10. Illustration of the differencc in presentation of bands and peaks; K , > K , j .

case of TLC, (6.20) and peak dispersivity is a function of the time spent in a column, the retention time t,, a=

JHtR

(6.21)

Also, for the bands in Figure 6.10, aB> a,, while for peaks the opposite is true. While most of this text deals with column techniques and peaks, the distinction between peaks and bands needs to be maintained for clarity.

Retardation Factors There is another difference between column techniques and planar techniques in LC. It is the way that the retardation factor is expressed. In

6.4

I

I II

137

Solvent front

Start

1

COLUMNAR AND PLANAR CONFIGURATIONS

Nonretained

4 I

component

start

1

R,c= la) Band

Ib) Peak

Figure 6.11. Comparison of equations for calculating retardation factors for columns and plane surfaces. Reprinted courtesy of Gow-Mac Instrument Co.

Chapter 2, the retardation factor R was defined for column techniques as

( 6.22) and the definition is illustrated in Figure 6.116. In TLC and PC, the symbol used is R , and the definition is

RF

=

distance migrated by an analyte distance migrated by the solvent front

(6.23)

Figure 6.11a shows that the migration distances are measured from the analyte starting spot. Clearly, both R and R , can be easily calculated from their respective chromatograms, making them very useful measures for describing chromatographic results. Recall also (Table 2.6) that k = (1 R ) / R and that k is proportional to the distribution constant K c , the basic thermodynamic variable in chromatographic theory. So, a comparison of HPLC and TLC for systems with equivalent stationary and mobile phases can also be made by comparing their respective k values, each calculated with the appropriate retardation factor, R or R r . Chapter 11 contains more information about TLC and PC, where it is noted that there are several reasons why the planar techniques could have retardation factors different from the column methods. For example, many times the flow is not controlled in TLC, and consequently it is not constant. As a result, we measure “distances” on the planar surfaces rather than ~

138

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

Start

Finish

I

I

.1

1

(a)

I

B 2

11

0.016 Abs

II5

Time (mid

Figure 6.12. Comparison o f the separation o f azo dyes by ( a ) TLC and ( h ) LSC. Conditions for TLC: silica gel F-254; solvent 10% CH?CI, in hexanc. Conditions for LSC: 15-cm x 2.4-mm column o f MicroPak SI-10; solvent same as TLC; flow rate, 132 mL/h; pressure, 350 psi. See Table 6.5 for identification of dyes. Reprinted with permission from R. E. Majors, A i d . Chern. 1973, 45, 755. Copyright 1973, American Chemical Society.

retention times or volumes, and the calculated R and R , values may not be exactly the same. To see how the two methods compare, some data published by Majors’9 are given in Figure 6.12, which shows the TLC and column separations of six azo dyes, listed in Table 6.5. The similarity between the two separations is obvious from the figure, but it is better evaluated by comparing the R and R , values in Table 6.5. The R , values are taken from reference 29, but the R values had to be estimated, since no t,,,, was given. The two columns of numbers clearly show the good agreement between these two parameters and suggest that TLC data should be useful in designing column HPLC methods. It is instructive to note again that the column LC peaks in Figure 6.12 are in reverse order and reverse width compared to the TLC bands.

REFERENCES

Table 6.5

139

Dyes Separated in Figure 6.12

Number

Structure

1

TLC R ,

LSC K“

0.69

0.55

0.2 1

0.2 1

0.064

0.097

2

@N=N*NEt2

3

4

5

0, N

N =N

-@NEt,

6 “Estirnatcd from Figure 6.12h.

Also, there are some operational differences that cause differences between HPLC and TLC. Even if the exact same stationary phase is coated o n

a TLC plate and packed in a column, the TLC material usually contains an additional binder to hold the stationary phase on the plate. This binder will most likely slightly alter the characteristics of the stationary phase and result in differences between the R and R,- values. And, finally, in normal TLC procedures the stationary bed is usually dry when the chromatographic elution process is begun, whereas it is usually wet with mobile phase in the column processes. This difference may also contribute to differences between retention parameters. REFERENCES 1. J. C. Giddings, Qnarnics of Chrornatogruphy, Part I, Dekker, New York, 1965, pp. 293-301. 2. J. F. Parcher, J . Chromatogr. Sci. 1983, 21, 346. 3. V. A. Davankov, PureAppl. Chem. 2001, 73, 982-992. 4. P. A. Bristow and J. H. Knox, Chrornatographia 1977, 10, 279.

140

COMPARISONS BETWEEN CHROMATOGRAPHIC MODES

5. V. R. Meyer, Practical High-Perfonnance Liquid Chromatography, 2nd ed., Wiley, Chichester, England, 1993. 6. J. H. Knox, J . Chromatogr. Sci. 1980, 18, 453. 7. J. H. Knox and M. Saleem, J. Chromatogr. Sci. 1969, 7, 614-622. 8. D. Ishii and T. Takeuchi, J. Chromatogr. Sci. 1989, 27, 71-74. 9. D. E. Martire, J. Chromatogr. 1989, 461, 165-175. 10. T. C. Chester, Anal. Chem. 1997, 69, 165A-169A. 11. J. F. Parcher and T. L. Chester (ed), Unified Chromatography, ACS Symposium Series 748, American Chemical Society, Washington, D.C., 2000. 12. P. S. Wells, S. Zhou, and J. F. Parcher, Anal. Chem. 2003, 75, 18A-24A. 13. D. R. Gere, Science, 1983, 222, 253. 14. M. G. Rawdon and T. A. Norris, A m . Lab. 1984, 16(5), 17. 15. C. M. Harris, Anal. Chem. 2002, 74, 87A-91A. 16. Berger Instruments, Inc.; w w w .beugersfc.com. 17. T. A. Berger and W. H. Wilson, Anal. Chem. 1993, 65, 1451. 18. S. R. Springston, P. David, J. Steger, and M. Novotny, Anal. Chem. 1986, 58, 997. 19. E. Ibanez and F. J. Senorans, J. Biochem. Biophys. Methods 2000, 43, 25-43. 20. Selerity Technologies; www.seleiity.com. 21. T. A. Berger, J. Chromatogr. A 1997, 785, 3-34. 22. T. A. Berger in Unified Chromatography, ACS Symposium Series 748, J. F. Parcher and T. L. Chester (ed), American Chemical Society, Washington, D.C., 2000, pp. 203-233. 23. J. Chromatogr. A 1997, 785, 1-104. 24. R. M. Smith, J. Chromatogr. A 1999, 856, 83-115. 25. T. L. Chester, J. D. Pinkston, and D. E. Raynie, Anal. Chem. 1998, 70, 301R-319R. 26. T. L. Chester and J. D. Pinkston, Anal. Chem. 2000, 72, 129R-l3SR. 27. T. L. Chester and J. D. Pinkston, Anal. Chem. 2002, 74, 2801-2811. 28. T. L. Chester and J. D. Pinkston, Anal. Chem. 2004, 76, 4606-4613. 29. R. E. Majors, Anal. Chem. 1973, 45, 755.

7 GAS CHROMATOGRAPHY The publication in 1952 of James and Martin's study' on gas chromatography (GC) began the era of modern, high-performance chromatography. Since that time, GC has become the premier technique for the separation and analysis of volatile compounds,* and gas chromatographs have been the most widely used analytical instrument in the world,' although HPLC is becoming more widely used. Clearly, GC is a major analytical method and it is complemented by the other major form of Chromatography, HPLC, which is capable of handling the nonvolatiles not suited to GC. The currently accepted status between them is that GC can be used up to 350°C corresponding to an upper molecular weight limit of 600 D," and HPLC is used for higher molecular weight compounds. However, high-temperature GC work has been done up to 450"C, and there are some alternative ways of handling nonvolatiles by G C as discussed at the end of this chapter. 7.1

EARLY HISTORY, THEORIES, AND CLASSIFICATIONS

Before GC became popular in the late 1950s, the only way to separate volatile materials was by distillation, which separates materials based on differences in vapor pressure or boiling point. G C is similar, but its separations also depend on the nature of the stationary phase, which gives it much Chromutogruphy: Concepis und Contrusis, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

141

142

GAS CHROMATOGRAPHY

more versatility than distillation. Imagine the pleasure and surprise of separations chemists who could now separate materials with close boiling points, such as benzene and cyclohexane. And it was easy, fast, and not too expensive; in addition, they did not have to worry about azeotropes. The benzene-cyclohexane separation provides an excellent example of the separating power of GC and a basic model for the theoretical concepts behind GC. The boiling points of benzene and cyclohexane are nearly the same, 80.1 and 81.4"C, respectively, so a large separation by G C will have to be effected by differences in the intermolecular interactions between the stationary phase and each of these two analytes. Both of them are nonpolar hydrocarbons, but benzene has a .rr-electron cloud, which makes it more susceptible to induction effects and dispersion attractions (Chapter 4). Therefore, we should choose a stationary liquid phase that would accentuate this difference-a polar one. Also, using the "like-dissolves-like" rule, we might choose an aromatic compound that would interact more with benzene than with cyclohexane. One possible liquid phase that meets these criteria is dinonylphthalate, and it can be used to separate benzene and cyclohexane. The relative retention has been found to be 1.6, which represents a very good separations Cyclohexane (which has the higher boiling point) is eluted before benzene, illustrating that it is the intermolecular interactions that effected the separation, not the boiling points. One way of expressing the extent of the interaction between these analytes and the liquid phase is by activity coefficients. According to Raoult's law, the partial pressure of a solute such as cyclohexane in a solvent such as dinonylphthalate is given by P c y = xcyYcy Ph:

where pcy is the partial pressure of cyclohexane, A',, is its mole fraction, ycy is its activity coefficient, and pfy is the vapor pressure of pure cyclohexane. The relative retention of the two solutes is also equal to the ratio of their partial pressures, and since their mole fractions and pure vapor pressures would be about equal,

Activity coefficients can be calculated from G C data according to Eq. (7.3):

1.7 X 10s Y = I/,p"MW

(7.3)

7.1

EARLY HISTORY, THEORIES, AND CLASSIFICATIONS

143

5

where MW is the molecular weight of the stationary liquid and is the specific retention volume (the net retention volume at 0°C per gram of liquid phase).* In this example, the activity coefficients at 326 K were found to be 0.52 for benzene and 0.82 for cyclohexane, respectively,s and their ratio is 1.6, in agreement with Eq. (7.2). The more the activity coefficient deviates from 1, the greater is the interaction between the solute and the stationary phase, showing that benzene has the greater interaction and the longer retention time. Santiusteh has concluded that activity coefficients give information about solute-solvent interactions that is as valuable as any of the other current theories (to be discussed later). For a brief summary of the development of solute-solvent interactions and the justification of his conclusion, consult the many references in his study. The choice of dinonylphthalate as the stationary phase (SP) for this separation is one of the two important choices that govern the success of the method. The other one (which was not dominant in this example) is the column temperature (see Chapter 6), which determines the vapor pressure and hence the volatility of the analytes. The control of temperature in GC will be discussed later; some of the background theories proposed to describe and predict interactions between the SP and the analytes will be reviewed next. They are interwoven with some history and some classifications. The earliest columns were packed with solids or with solid supports coated with liquids. In accordance with our naming convention, these two types of chromatography are called gas-solid chromatography (GSC) and gas-liquid chromatography (GLC), respectively. Of the two, the latter is more commonly used, and most of this chapter concerns GLC. In 1957 Golay published his ideas for using columns that were not packed but were open tubes.’ These tubes had to have small inside diameters; they were capillary columns, but the name open tubular ( O T ) columns is more descriptive and preferred when comparing them to packed columns. These two types of column necessitate slightly different chromatographic instruments, and most of this chapter will be devoted to O T columns and instruments. Figure 7.1 shows a classification of GC modes that reflects the various options. In O T columns, the SP can be coated on, or bonded to, the inside wall, and these columns are called wall-coated OT columns (WCOT). The other types of O T columns will be discussed later. In GC, the mobile phase, usually helium, is inert and does not participate in the chemistry of the chromatographic process but only serves to carry the sample through the column. Hence, the theory of G C is concerned only with the interactions between the SP and the analytes. The basic types of forces ’The IUPAC has recommended that this tcrm, thc specific retention volume, be discontinued.’

144

GAS CHROMATOGRAPHY

Figure 7.1. Classification of GC methods. Reprinted with permission from J. Miller and J. Crowther (eds), Anulyticul Chemistry in u GMP Enoironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

and interactions were covered in Chapter 4. Starting with that base, the following discussion shows how GC theory has developed in an attempt to explain in more elaborate terms the effectiveness of the SP in effecting separations like the one just illustrated, benzene and cyclohexane. Kovats Index One approach for describing the nature of an SP was to classify its properties such as polarity or relative interactions with analytes (dispersion, induction, orientation). To express the relative retention behavior of analytes, Kovats’ proposed that the n-paraffins be used as relative standards. H e defined a retention index I in which the n-paraffins are assigned reference values 100 times their respective carbon numbers. Thus hexane has an index of 600, heptane 700, and so on. Before discussing his proposal further, we need to examine the relationship between retention volume and the members of a homologous series such as the paraffins. An isothermal GC run of paraffins on a given column will result in a chromatogram like that shown in Figure 7.2, where the retention volumes increase logarithmically with carbon number. This is the expected behavior since we know that log K

= -

(7.4)

where A X i s the enthalpy of vaporization, which can be estimated by Trouton’s rule as

7.1

EARLY HISTORY, THEORIES, AND CLASSIFICATIONS

145

Retention volume

Figure 7.2. Typical chromatogram for a homologous series of compounds.

I

s?

-

0) 0

1

1.0 0.8 0.6

0.4 0.2 1 2 3 4 5 ~~~h~ carbon atoms

Figure 7.3. Plot o f carbon number vcrws log net retention volume for a homologous series.

Since the boiling points Thol,of the members of a homologous series increase in a regular fashion, so should A , Y ; K should increase logarithmically, as should VN since VN = W,.When log VN is plotted versus the carbon number, as shown in Figure 7.3, a straight line results for most homologous series.' If any other compound is run under the same conditions, its index can be read from the graph using the log of its net retention volume. By definition, the members of any homologous series should differ from each other by 100 units just as the standards do. Alternatively, the index can be calculated as follows:

'Since t k constant.

i5

proportional to VN, it can be used on the y axis if flow rate and pressure drop are

146

GAS CHROMATOGRAPHY

where u stands for the unknown, x for the n-paraffin with x carbons and eluting just before the unknown, and x + 1 for the n-paraffin with x + 1 carbons and eluting just after the unknown. The retention index is one method for reporting G C data that was very popular but is not used much today. This relationship is not always exact,"' and the index is somewhat temperature dependent," but McReynoldsl' has published a self-consistent book of indices for 350 compounds on 77 stationary phases at 2 temperatures. Other homologous series have also been used as standards in specific industries. Later we will see that programmed temperature G C results in a regular, linear relationship between retention time and carbon number. Under those circumstances logs should not be used in Eq. (7.6) and in the retention index plot. The increase in temperature decreases the partition coefficients and effectively removes the logarithmic dependence of 1. Rohrschneider / McReynolds Constants

Chapter 4 ended by noting that one way of estimating the polarity of a stationary phase was to use selected probes as the samples and see how they were retained. The extent of interaction between the stationary phase and a given sample will be reflected in the adjusted retention time (or retention factor,k or retention index, 1 ) of each analyte. Thus, by choosing as solute probes those chemicals that are thought to have particularly strong selective interactions, one can get a measure of the relative magnitude of that interaction from its retention index. Rohrschneider'3 was the first to suggest a list of probes and a method of organizing the data. His original study is in German, but Supina and RoseI4 have described it in English. The five solute probes are listed in Table 7.1.

Table 7.1 Compounds Used for Liquid-Phase Characterization

Designation U

b C

d e

Probes Used by Rohrschneider McReynolds Benzene Ethanol 2-Butanone (MEK) Nitromethane Pyridine

Benzene n-Butanol 2-Pentanone Nitropropane Pyridine 2-Methyl-2-pentanol Iodobutane 2-Octyne 1,4-Dioxane cis-Hydrindane

7.1

EARLY HISTORY, THEORIES, AND CLASSIFICATIONS

147

For the stationary liquid phases, squalane was chosen as the least polar and assigned Rohrschneider constants of zero. To get the polarity constants for the other stationary phases, each solute in Table 7.1 was run on squalane and on the liquid phase of interest at 100°C and 20% liquid loading. The retention index I was determined for each analyte, and the difference between the two values on the two phases ( A I ) was obtained. The A Z summed for all five probes is given as:

AI

= ax

+ by + cz + d u + es

(7.7)

where a , h, c, d , and e represent the five solutes and x, y , z, u , and s represent the five Rohrschneider constants for that liquid phase. Thus a has a value of 100 for benzene and a value of 0 for the other four solutes; b is 100 for ethanol and 0 for the others, and so on. Each of the x, y , z values equals A I divided by 100, and the AZ value in Eq. (7.7) is actually the sum of the five individual A I values. In 1970 McReynolds" went one step further. He reasoned that 10 probes would be better than 5 and that some of the original 5 should be replaced by higher homologs. His probes are also listed in Table 7.1. Also, he did not divide his values by 100, so his constants are 100 times those of Rohrschneider; these larger values are the ones usually quoted currently. In fact, it has turned out that McReynolds was wrong-10 probes were not better than 5. Therefore, most compilations of McReynolds constants list only 5 , 6, or 7 values. In fact, later attempts to expand on this method have tried to reduce the number of constants to 4. Methods for Selecting Stationary Phases

Choosing a stationary phase based on McReynolds' constants has not proven to be effective, and suppliers of columns no longer give these constants in their catalogs, as they once did. Modern approaches to GC stationary phase classifications are based on the cavity model of solvation,"" but it has little practical value. In fact, the whole process of finding the best SP for a given sample is not as important as it was when packed columns were the only ones in use. Today's O T columns are so efficient, and auxiliary techniques such as multidimensional operation and mass spectral (MS) detection so effective, that they have removed the necessity to spend time finding the best SP. There remains, of course, the challenge to describe the chromatographic process in theoretical and quasi-theoretical terms for GC and for HPLC, where the situation is more complex. For a summary of past and current efforts, see Chapter 2 in Pooles comprehensive book,'' and Chapter 8 on HPLC in this book.

148

GAS CHROMATOGRAPHY

0

Injector/Splitter

Syringe

Septum P”W

Variable

,.. + sprir ~

Detector FID

.

Air H2

Make-up He

He

carrier Figure 7.4. Schematic of a typical gas chromatograph. Reprinted with permission from H. McNair and J. Miller, Basic Gas Chromatography, John Wiley & Sons. Copyright 1998; this material is used by permission of John Wiley & Sons, Inc.

7.2

INSTRUMENTATION FOR CAPILLARY GC

The essential parts of a gas chromatograph shown in Figure 7.4 are carrier gas, flow or pressure regulator, injection port/splitter, column, and detector. Usually there are three separate heated zones, for the injector, the column, and the detector. Connections between these heated zones must also be kept hot enough to prevent condensation of analytes in them. Chromatographs designed for OT columns have an injection port that allows for sample splitting and a provision for some additional “makeup” gas for the detector, as shown. Mobile Phase The most popular carrier gases are nitrogen, helium, and hydrogen. They must be very pure, and they are chosen for their inertness since, as we have seen, their only purpose is to carry the analyte vapors through the column. Helium is the most popular in the United States because of its higher efficiency at faster flow rates, as shown in the typical van Deemter plot in Figure 7.5. In this example, the optimum velocity is around 20-30 cm/s, but little efficiency is lost (and thus time is gained) at higher velocities, so most

7.2

-

1.2

-

1.0

-

E

.a-

u( I

.6-

E

+

INSTRUMENTATION FOR CAPILLARY GC

C 1 I at 175 o k = 4.95

149

c

Glass W.C.O.T. ov.101

25m x 0.25mm

c;

.4

-

-2 1

10

20

30

40

50

60

10

00

90

Average Linear Velocity (cmlsec) Figure 7.5. Effect of carrier gas on van Deemtcr curve. Reprinted from R. R. Freeman (ed), High Resolution Gas Chromatogruphy, 2nd ed., Hewlett-Packard Company, Wilmington, DE, 1981. Copyright Agilent, Inc. Reproduced with permission.

chromatographers run their O T columns around twice the optimum velocity. See Chapter 3 for a discussion of the rate equation and further discussion about the choice of carrier gas and the optimization of OT columns. Since helium is expensive and in limited supply in some parts of the world, hydrogen is increasing in popularity, but its use requires additional safety precautions to prevent explosions. In Figure 7.5 the advantageous low plate height and improved mass-transfer characteristics (flat curve at high velocities) when using hydrogen are evident. However, these advantages are not as impressive for programmed temperature GC (discussed later in this chapter) as they are for isothermal GC shown in Figure 7.5, so helium remains the gas of choice for most laboratories in the United States. Sometimes the choice of carrier is dictated by the detector. Table 7.2 lists the carrier gases used with Table 7.2

Preferred Carrier Gases for Three Detectors

Detector Flame ionization detector (FID) Thermal conductivity detector (TCD) Electron capture detector (ECD

Carrier Gas Helium Helium, hydrogen Very dry nitrogen

150

GAS CHROMATOGRAPHY

three common detectors. The electron capture detector (ECD) has the most unique requirements, and the thermal conductivity detector (TCD) works best with gases of high thermal conductivity. Over the years there has been some interest in using carrier gases that are not inert but have some effect on the partitioning process, as was mentioned in Chapter 6. Water is the one used most often, and the technique is sometimes called steam GC. This topic is included in Berezkin's review of carrier gases in open tubular GLC," and it also includes reports of other secondary effects of column pressure and nonideal gas behavior. The pressure drop across an O T column is only 10-100 kPa.* It is regulated with one or more valves in order to get the desired flow rate (linear velocity). We have already seen that a constant flow rate is desirable so that retention times will not vary, and flow-sensitive detectors will not become nonlinear (see Chapter 9). In the technique of programmed temperature (PT) GC discussed later, the viscosity of the carrier gas increases during a run, so flow rate will decrease during a PT run if it is being operated at constant pressure. On the other hand, the split injectors used with OT columns require constant pressure regulation, so older chromatographs are run at constant pressure, and the decrease in carrier gas velocity during the run is tolerated and partially compensated by starting the run at a higher pressure. Modern instruments often have electronic pressure control (EPC), so constant flow can be achieved by choosing a mode of constant linear velocity whereby the pressure is increased during the PT run. Injection Ports and Valves

Sample introduction is most often accomplished with a microsyringe through a self-sealing rubber septum, as shown in Figure 7.6. Gas sample valves are more reproducible and are preferred for gases. In all cases, the objective is to get the sample into the column rapidly and in as small a volume as possible. This is usually accomplished by flash vaporization of the sample in a heated liner that also has the capability for splitting the sample into two parts. A typical design for a split/splitless injector is shown in Figure 7.7 where it is configured for a split injection. Split injection is the oldest, simplest, and easiest injection technique. The procedure involves injecting about 1 p L of the sample by a standard syringe into a heated injection port that contains a deactivated glass liner. A plug of glass wool or other inert material in the vaporization region can be used to promote complete volatilization. The 'The standard unit of pressure in SI is the pascal, Pa. For those familiar with old units, the following conversions are given: 1 har = 100 kPa; 1 atm = 101.3 kPa; 1 torr = 133 Pa; 1 psi = 6.0 kPa.

7.2

INSTRUMENTATION FOR CAPILLARY GC

J

Syringe

/+

J

Column or injection liner

1

151

6 1

f-

Carrier gas

Figure 7.6. Simplified injection port design.

RATIO VALVE

SEPTUM PURGE

1

BUFFER

HEATED INJECTOR BLOCK

VOLUME

1

GRAPH FERRU

1

-CARRIER GASIN

SPLIT POINT

1

7I

-SPRING -STAINLESS STEEL FERRULE

K S P L I T TIP ASSEMBLY COLUMN OVEN

Figure 7.7. Inlet designed for OT columns and sample splitting. Courtesy of Varian Instruments.

152

GAS CHROMATOGRAPHY

sample is rapidly vaporized, and only a fraction, usually O.l-lO%, of the vapor enters the column (see Fig 7.7). The rest of the vaporized sample and a large flow of carrier gas passes out through a split or purge valve. The buffer volume allows the sample to pass the point of column splitting before reaching the needle valve, which could change the split ratio. There are several advantages to split injections. The technique is simple because the operator has only to control the split ratio by opening or closing the split (purge) valve. The amount of sample introduced to the column is very small (and easily controlled) as required by capillary columns that have small amounts of stationary phase and therefore can accommodate only small sample sizes. Also, the flow rate up to the split point is fast (the sum of both column and vent flow rates), and the result is high resolution separations. Another advantage is that “neat” samples can be introduced, usually by using a larger split ratio, so diluting the sample is not required. A final advantage is that “dirty” samples can be introduced by putting a plug of deactivated glass wool in the inlet liner to trap nonvolatile compounds. One disadvantage is that trace analysis is limited since only a fraction of the sample enters the column. Consequently, splitless or on-column injection techniques are recommended for trace analysis. A second disadvantage is that the splitting process sometimes discriminates against high-molecularweight solutes in the sample so that the sample entering the column is not representative of the sample injected. For these reasons, another vaporization mode, splitless injection, is sometimes used. Splitless injection uses the same hardware as split injection (Fig. 7.7), but the split valve is initially closed. The sample is diluted in a volatile solvent (such as hexane or methanol) and 1-5 FL is injected in the heated injection port. The sample is vaporized and slowly (flow rate of about 1 mL/min) carried onto a cold column where both sample and solvent are condensed. After about 4.5 s, the split valve is opened (flow rate of about SO mL/min), and any residual vapors left in the injection port are rapidly swept out of the system. Septum purge, a very small flow of carrier gas sweeping the septum and then being vented, is essential with splitless injections. The column is then temperature programmed, and initially only the volatile solvent is vaporized and carried through the column. While this is happening, the sample analytes are being refocused into a narrow band in the residual solvent. At some higher temperature, these analytes are vaporized and chromatographed as described later (in the programmed temperature section of this chapter.) High resolution of these higher boiling analytes is observed. The big advantage of splitless injection is the improved sensitivity over the split mode. Typically 20- to SO-fold more sample enters the column and the result is improved trace analysis for environmental, pharmaceutical, or biomedical samples. However, splitless has several disadvantages. It is time-

7.2

INSTRUMENTATION FOR CAPILLARY GC

153

Table 7.3 Comparison of Split and Splitless Injection Modes

Liner Liner diameter Injection method Sample size Sample/solvent ratio

Split Mode

Splitless Mode (Grob)

May contain frit or glass wool

Open

4 mm

Simple injection 0.1-10 p L 1 : 1-1:lOOO

2 mm Several steps (see text) 1-5 p L 1 : 103-1 : 10"

consuming; you must start with a cold column; and you must temperature program. You must also dilute the sample with a volatile solvent and optimize both the initial column temperature and the time of opening the split valve. Finally, splitless injection is not well-suited for volatile compounds. For good chromatography the first peaks of interest must have boiling points 30°C higher than the solvent. A comparison of the split and splitless methods is given in Table 7.3. For more information on split/splitless injections, see reference 19. Three other types of capillary inlets are direct injection, on-column, and cold on-column. On-column means inserting the precisely aligned needle into the capillary column, usually a 0.53-mm i.d. megabore and making injections inside the column. Direct injection involves injecting a small sample (usually 1 p L or smaller) into a glass liner where the vapors are carried directly to the column. Both of these techniques require thick-film capillaries and widediameter columns with faster than normal flow rates ( - 10 mL/min). Even with these precautions the resolution is not as good as with split or splitless injection. The advantages can be better trace analysis and good quantitation. Both high resolution and good quantitation result from cold on-column injections. A liquid sample is injected into either a cold inlet liner or a cold column. The cold injector is rapidly heated and the sample vaporized and carried through the column. Minimal sample decomposition is observed. For thermolabile compounds, cold on-column is the best injection technique. Syringe injection is readily automated with commercially available equipment. The procedure is analogous to manual injection and has better precision. Hinshaw 2" has listed some Open Tubular Columns

Open tubular columns are usually made of fused silica since inertness is of prime importance. The column is either kept at a constant temperature (isothermal GC) or programmed during the run (PTGC). It is a common misconception that the column temperature should exceed the boiling point

154

GAS CHROMATOGRAPHY

of the sample in order to keep the analytes in the vapor phase. Actually the column will produce better separations if the temperature is below the samples boiling point, thereby increasing its interaction with the stationary phase. The situation desired inside the column can be described in comparison to the water vapor in our environment; there is plenty of vapor well below the boiling point (as we know from those days with high humidity), but we must be above the “dew point” of the analyte or else it will “rain” in the column. The smaller the amount of stationary phase, the lower the temperature at which we can operate, so OT columns are usually run at lower temperatures than packed columns. In their original and most simple form, O T columns contain a thin film of stationary liquid on their inside walls. Hence they are referred to as wall coated open tubular, or WCOT, columns. Compared to packed columns, they have low pressure drops and small amounts of stationary phase (and relatively large p). Originally they were made of stainless steel, and the early technology has been described by Ettre.” A revolution occurred in the 1960s and 1970s when the column material changed first to glass22 and then to

DB-17 0.25 micron film 30 meters x 25mm I.D. 1 wl split injection 40cmIsec. H, carrier Atten: 2f2 Chart speed lcmlmin. llO°C isothermal

1 I ,I 6 ’

20-

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

~

decane undecane I-octanol dodecane 2.6-dimethylphenol 1-decanol 2.6-dimethylanaline naphthalene methyl decanoate methyl undecanoate

* 10

-1

Figure 7.8. Standard test mixture for OT columns run on a 30-m X 0.25-mm i.d. (inside diameter) column of DB-17 with 0.25-fim film at 110°C. Courtesy of J & W Scientific.

7.2

INSTRUMENTATION FOR CAPILLARY GC

155

fused silica.” Because of the superior performance (inertness) and flexibility of fused silica columns, they have become the most popular type. Figure 7.8 is typical of their performance. A more severe sample test mixture, including acids and bases, is shown in Figure 7.9. Note the asymmetry of peak 6 and the TZ numbers for the members of the homologous series of esters-peaks 9, 11, and 12. For a period of time, O T columns were available having characteristics intermediate between those of WCOT and packed columns. There were two types, but they were similar. Support-coated open tubular (SCOT) columns had a thin layer of solid support coated on the inside wall of a capillary tube of larger diameter than that used for WCOT columns. This layer was coated with stationary liquid similar to packed columns. Porous layer open tubular (PLOT) columns were similar but made differently; for example, the solid support was added while the capillary tube was being drawn. SCOT columns are no longer popular because larger diameter WCOT fused-silica columns (often referred to as wide-bore or rnegu-bore) are as good, more stable (no layer to flake off), and easier to use. A few PLOT columns are commercially available and are increasing in popularity, mainly for GSC applications. Subsequent discussion in this chapter will center on WCOT columns. Coating the inside of a capillary tube requires pretreatment of the silica so that the liquid will wet the surface uniformly and stick to it.24” Stable stationary phases are attained by bonding the liquid to the silica and/or cross-linking it, using a variety of methods26 including y irradiation.” The cross-linked bonded phase OT columns are very stable and can even be washed with solvents for cleaning. The outside of fused-silica columns must be coated with a high-temperature polyimide to extend their life and keep them flexible. The thickness of the film of liquid, d , , can be controlled, and a variety of thicknesses are commercially available. Thicker films are desirable for volatile samples and larger samples, but they are somewhat less efficient. When d , is known, the phase volume ratio /3 can be calculated:

where r, is the inner radius of the OT column. Thus, for OT columns, the fundamental partition coefficient K , can be calculated for a given analyte since the partition ratio k is easily determined from the chromatogram obtained, and K,.=kp

(7.9)

It is not easy to obtain partition coefficients for packed columns because the volume of the stationary phase is more difficult to calculate.

r

I

-

2

.._----

--

1

3

7. 2.6-dimethylaniline 8. n-dodecane 9. methyl decanoate 10. dicyclohexyl amine 11. methyl undecanoate 12. methyl dadecanoate

Figure 7.9. “Grob” test mixture run on 30-m DB-5OT column. Courtesy of J & W Scientific.

Journal of Chrometography, 156 (1978) 1-20.

1. 2.3-butanediol 2. n-decane 3. 1-octanol 4. nonanal 5. 2b-dimethylp enol 6. 2-ethylhexanoic acid

DB-5 .25 micron film Standardired Quality Test According to Grob, et al’ 30m x 0.25mm fused silica column

Conditions: 80 sec a t 40°C (hydrogen) 40°C to 140% programmed at 1.67°C/min. Attenuation: (1 x 4) Chart: 0.75cm/min.

t,,,

1

1

7.2

INSTRUMENTATION FOR CAPILLARY GC

157

Table 7.4 Comparison of Fused-Silica WCOT Columns

Inside diameter Film thickness ( d , ) Phase volume ratio ( p ) Column length Flow Hmin

Neff

Typical sample size

Conventional

Wide Bore

0.25 mm 0.25 p m 250-1000 15-60 m 1 mL/min 0.3 mm 3000/m 50 ng

0.53 mm 1-5 p m 25-200 15-30 m 5-15 mL/min 0.6 mm 1200/m 15 P1.g

To increase the capacity of WCOT columns, wider diameters and heavier loadings are used. These columns have replaced the SCOT columns and are known as wide-bore or mega-bore columns. The two most common sizes of WCOT columns are compared in Table 7.4, but several other sizes are commercially available, and an excellent discussion of the various types has been given by Duffy.2x The wide-bore columns are best for low-boiling mixtures and will accommodate larger samples; the conventional columns give the highest efficiencies and may permit the use of shorter lengths and thus shorter analysis times. Ettre” has summarized the effect of column diameter and liquid film thickness on OT column performance. To protect the analytical OT column from deterioration and contamination from dirty samples, a precolumn is often inserted between the injection port and the analytical column. This column, also called the retention gap, is not intended to retain analytes, so it is usually uncoated, but deactivated, fused silica. It should not cause appreciable z o n e broadening but rather should help focus the analytes as they enter the analytical column. It is especially helpful for large solvent injections (discussed later in this chapter), and if it is of wide-bore size (0.53 mm), it can facilitate on-column injection. Detectors

Virtually every conceivable means of detecting gases and vapors has been exploited in designing G C detectors, and over 100 have been described. Most of them were invented expressly for GC. Of the two most popular ones-the thermal conductivity detector (TCD) and the flame ionization detector (FID) -the FID is usually required for OT chromatographs because of its greater sensitivity. Both are classified (according to the criteria in Chapter 9) and compared in Table 7.5. Flame ionization Detector The FID is a small oxygen-hydrogen flame in which the sample is burned, producing some ions in the process. These ions are then collected, constituting a small current, which is amplified and sent to

158

GAS CHROMATOGRAPHY

Table 7.5 Comparison of TCD and FID

Classifications

TCD

FID

Concentration Bulk property Universal

Mass flow rate Specific property Slightly selective; only organics Destructive

Nondestructive Characteristics Sensitivity Detectivity (MDQ) Minimum sample size Linearity

5000 mV . mL/mg lo-'' g/mL 10-* g lo4

lo-? c / g

g/s lo-"' g 10"

Figure 7.10. Schematic of FID. Courtesy of Perkin-Elmer.

a data system. A typical FID design, (Fig. 7.10) shows the column effluent mixed with hydrogen and led to a small burner tip that is swept by a high flow of air for complete combustion. An igniter is necessary for remote lighting of the flame. The collector electrode is biased about + 300 V relative to the flame tip, and the signal current is amplified. The exact mechanism of flame ionization is still not known, but the ionization efficiency, while low, is

5

2

1

!I

-

80-

-

8

f

90 -

-

100-

cIn

f

0

-E

.-xE

h

-

b. 14 mumin N2

I

I I 20

I

I

30

I

1

40

I

I

50

I

I

60

I

I

70

Figure 7.1 1. Typical response curve for FID as a function of hydrogen flow. Courtesy of Perkin-Elmer.

I 10

d. 75 mL/min N2

c. 40 mL/min N2

a. 5 mL/min Nz

Air: 370 mL/min

/ d

I

160

GAS CHROMATOGRAPHY

Carrier gas (N2): 24 mL/min Hydrogen:39 mL/min Sample: 3.5 x 10-7g carbon

100

200

300

400

500

600

700

Air flow rate, (mL/min)

Figure 7.12. Typical response curve for FID as a function of airflow. Courtesy of Perkin-Elmer.

one of the highest available in a GC detector and sufficient to give a good sensitivity and linearity. The flow rates of hydrogen and air must be optimized for a particular detector design (and to a lesser extent, analyte). Typical values, as shown in Figures 7.11 and 7.12, are a hydrogen flow of 30-40 mL/min and an air flow of 300-400 mL/min. For O T columns, a makeup flow of carrier gas is necessary to bring the total up to about 30 mL/min, since O T flow rates are of the order of only 1 mL/min. As noted earlier, hydrogen is becoming increasingly popular as the carrier gas of choice for O T columns; this requires changes in gas flows and has prompted new FID design^.^" The FID is nearly universal, detecting all organic compounds, but not those listed in Table 7.6. Water is a compound that often produces badly tailed peaks on GC columns, so it may be advantageous that the FID does not detect it. However, the FID cannot be used for analysis of fixed gases. Response factors must be determined for good quantitative analysis3' (see Chapter 9). Table 7.6 Some Compounds Not Detected by Flame Ionization Detector

He Ar Kr Ne Xe

cs,

cos

SiC1, SiHCl SiF,

7.2

INSTRUMENTATION FOR CAPILLARY GC

161

Top view

-Outlet,

-Inlet,

above

below

Side v i m

f

Outlet

I-

Inlet

outlet

1 Iniet

Figure 7.13.Typical TCD, designed for four concentric hot wircs. Courtesy of the Cow-Mac Instrument Co.

In summary, the FID is the detector of choice for organic analysis because of its sensitivity, which allows it to be used with O T columns. It has a good stability and linearity, but it requires additional gases for its operation. While it is not totally universal, it does detect all organic compounds. Thermal Conductivity Detector The TCD cell is a metal block in which cavities have been drilled to accommodate the transducers, which can be either thermistors or resistance wires (so-called hot wires). Thermistors are most sensitive at low temperatures and find limited use. The wire filaments can be supported on holders or be mounted concentrically in a cylindrical cavity. The latter arrangement permits the volume of the cavity to be minimized, which is highly desirable. A typical design, shown in Figure 7.13, has four cavities; two is the minimum-one each for the reference and the

162

GAS CHROMATOGRAPHY Current controi

I Attenuator

Figure 7.14.Simplified Wheatstone bridge circuit for TCD. Four elements: R , and K 2 for reference, and S,and S? for example.

sample flows. A special low-volume thermal conductivity cell has been produced by Agilent (formerly Hewlett-Packard); to keep the volume small, only one cavity is used and the two gas streams (reference and sample) are passed through it alternately. The resistance wires, made of tungsten or a tungsten-rhenium alloy, are heated with a direct current (DC) source to a temperature above the block temperature and lose heat to it at a rate dependent upon the thermal conductivity of the gas in the cavity. Thus, the temperature, and hence the resistance, of the hot wire depends upon the thermal conductivity of the gas in the cavity. The wires are incorporated into a Wheatstone bridge circuit (Fig. 7.14) and produce a voltage imbalance when an analyte passes through one side of the cell. The wires can be heated at constant voltage or constant current or be maintained at a constant temperature by varying the current or voltage. Keeping the filament temperature constant requires a more complex circuit, and the output signal is derived from the electrical changes that are necessary to bring the bridge back to null, rather than from the direct voltage imbalance. Sensitivity is increased by heating the filament to a higher temperature with the power supply, and it is a function of the difference in temperature ( A T ) between the filament and the block. If a TCD is operated in air, the filaments are quickly oxidized and burn out. Even small leaks in the chromatographic system will result in gradual destruction of the hot wires.

7.2

INSTRUMENTATION FOR CAPILLARY GC

163

Table 7.7 Thermal Conductivities and TCD Response Values for Selected Compounds

Compound

Thermal Conductivity"

RMR"

Carrier Gases

Argon Carbon dioxide Helium Hydrogen Nitrogen

12.5 12.7 100.0

128.0 18.0

Samples

Ethane n-Butane n-Nonane i-Butane Cyclohexane Benzene Acetone Ethanol Chloroform Methyl iodide Ethyl acetate

51 85

17.5

13.5

177 82

10.8 14.0 10.1

1 I4 100 86

9.9 9.6 12.7

72 108 96

6.0

4.6 9.9

"Relative t o He = 100. 'Relative molar response in helium. Standard: benzcne

111 =

100.

The carrier gas must have a thermal conductivity (TC) that is very different from the analytes to be detected, so that the most commonly used gases are helium o r hydrogen. Some relative thermal conductivities a r e given in Table 7.7. With a high T C carrier gas such as He, the filament runs relatively cool; when a sample enters the sample cavity, that filament gets hotter, its resistance goes up, and a signal is produced. Other mechanisms contribute to the loss of heat from the filament, and response values cannot be calculated from thermal conductivities alone. For quantitative analysis, response values must be determined; they are included in Table 7.7. Remember that the T C D is a concentration detector and the peak areas it produces are flow dependent. Low flow rates have higher analyte concentrations and therefore better sensitivities. Although the T C D is only moderately sensitive, it is universal, simple, rugged, and inexpensive. Other Detectors T h e electron capture detector (ECD) was invented by Lovelock32 in 1961 and is probably the third most used detector. As its name implies, it is selective for materials that capture electrons-halogenand

164

GAS CHROMATOGRAPHY

nitrogen-containing compounds such as pesticides and unsaturated compounds such as the polynuclear aromatics. It is an ionization detector, but unlike the FID it is a concentration type and a bulk property type of detector. As such it is an exception to our generalization that bulk property detectors are not very sensitive. The ECD has a high standing current caused by the ionization of the carrier gas by a radioactive source. It requires very pure nitrogen as a carrier gas and a radioactive source. Formerly, tritium was used as the excitation source, but ""i is more common now because it has a higher temperature limit of 350°C. The electrons caused by the ionization produce a large current

Table 7.8 Other GC Detectors

Name

Operating Principle

Selective for"

Ionization-Type Detect0r.s

Alkali flame ionization (AFID) o r nitrogen/phos (NPD)

Photoionization (PID) Discharge ionization (DID) Helium ionization (HID)

Original alkali salt vapors cause chemical ionization in flame; now rubidium silicate heated electrically U V lamp causes photoionization High-voltage discharge He carrier gas; ionization by tritium

Aromatics Universal but used for gases Universal, but used for gases

Emission Type Detectom

Flame photometric (FPD) Atomic plasma emission

Flame excitation causes emission Plasma excitation causes emission

s, p Metals; S. P, X

Other Detectors

Hall electrolytic conductivity (HECD) Chemiluminescence: Thermal energy analyzer (TEA) Sulfur chemiluminescence Radioactivity (RAD) " X = halogen.

Catalytic reaction to form HX, H,S, NH,; measure conductance Catalytic pyrolysis Flameless thermal fragmentation p or y detectors

Nitro- and nitrosoS Radioactive

7.3

' I

INSTRUMENTATION FOR PACKED-COLUMN GC

TtD

I

ECD-?) NPD

NPD

FID

1

m,t,,,t, NPlD

1

I

I

__t__

I

I

I 1

I I I 4

165

I I

I I

I

I

I

1-1

HECD

10-15

(fa

I

I

I

I

I

TEA-*

10-12 (Pg)

I

10-9

(n&

I

10-6 (rp)

10-3

(ms)

Amount of sample (g, log scale)

Figure 7.15. Comparison of typical working ranges for common G C detectors.

output from the detector, but the presence of an electron-capturing analyte decreases this current as the electrons are absorbed. This absorption of electrons follows an equation similar to Beer's law. g/mL for ideal The ECD has a minimum detectivity of about analytes and a linearity of three or four orders of magnitude. It is very easily contaminated and somewhat troublesome to operate. The characteristics of some other detectors are summarized in Table 7.8. The popular mass spectrometric detector (MS or MSD) is discussed separately and more extensively in Chapter 10. A comparison of the working ranges of the most common detectors is shown in Figure 7.15. For more details on the detectors see the books by David,33 Dress1er,j4 Sieveq3' and Hill and McMinn," and the chapter by Colon and Baird." 7.3

INSTRUMENTATION FOR PACKED-COLUMN GC

Packed-column instruments differ from those for O T columns in the injection port design and the columns themselves. These two topics will be discussed briefly, and some reasons will be given for choosing packed columns over O T columns. Packed-Column Injection Ports

There are two injection configurations for packed-column instruments, both based on the design shown in Figure 7.6. The important criterion is the

166

GAS CHROMATOGRAPHY

position of the column packing relative to the end of the needle of the injection syringe. If the packing is close to the end of the needle, the sample will be deposited on the column and is called on-column injection. Most of the sample will be sorbed on the column without requiring prior vaporization. If the packing is a few centimeters from the end of the needle, a liquid sample will be vaporized in the volume of the injection port, which is heated to facilitate that process; this is called flash uaporization, analogous to the process in split/splitless vaporization for O T columns. When purchasing a commercial column, it is necessary to specify the instrument being used and the mode desired, so the column supplied will have the proper geometry. Unlike the split/splitless injector for OT, the process for packed columns can be carried out at constant carrier gas flow. On-column injection is used most often because it is more efficient and is usually operated at a lower inlet temperature that does not require the sample to come in contact with a hot surface, which might act catalytically to cause decomposition. Packed Columns and Stationary Phases

Packed columns are typically 2-20 f t in length and +-:-inch in outside diameter. Stainless steel is most common, but glass is also used. The packing is either a solid adsorbent (for gas-solid chromatography, GSC) or a solid support on which a liquid (the stationary phase) is coated (for gas-liquid chromatography, GLC). In either case, the particles should be small and uniform in size as indicated by their mesh range. Typical ranges are 80-100 or 100-120 mesh. The latter range includes particles with diameters from 125 Table 7.9 Comparison of Packed and WCOT Columns

inch packed

Outside diameter Inside diameter df

P

Column length

Flow Nto, Neff Hmin

Advantages

3.2 mm 2.2 mm 5 Pm 15-30 1-2 m 20 mL/min 6000 2000/m 0.5 mm

Lower cost Easier to make Easier to use Larger samples Better for fixed gases

WCOT

0.40 mm 0.25 mm 0.25 p m 250- 1000 15-60 m 1 mL/min 180,000 3000/m 0.3 mm

Higher efficiency Faster More inert Fewer columns needed Better for complex mixtures

7.3

INSTRUMENTATION FOR PACKED-COLUMN GC

167

I'

r

0

.

.

6

12

18

24

Figure 7.16. Determination of water in solvent$, using Porapak Q porous polymer at 220°C. Courtesy of Varian Instruments.

to 149 p m . Table 7.9 summarizes the differences between packed and O T columns. The active solids used for GSC include classic materials such as silica gel, alumina, and charcoal. The zeolite molecular sieves are famous for their ability to separate oxygen, argon, and nitrogen as well as some other fixed gases. Similar materials have been produced commercially especially for GC. They include other molecular sieves such as Carbosieve S, which was illustrated in Figure 4.5 for separating fixed gases including oxygen and nitrogen. Other solid SPs are organic polymers that have been modified to get desirable GC properties. One series called Porapak is ideal for water samples, and a typical separation is shown in Figure 7.16. Packed-Column Applications It is generally agreed that O T columns are superior for most GC separations, but there are some applications where packed columns are used. Some of them are

Separations that were developed in the early days of G C before O T columns became so widely available. Separations of large sample sizes as would be required for preparative uses. Packed columns have much more stationary phase in them than do OT columns, so they would be the columns chosen for preparative work. Separations requiring high selectivity. Packed columns have the flexibility to tailor-make the stationary phase so that it will have the capability to separate a specific pair of difficult-to-separate compounds. One common example is the xylene isomers, which were first successfully separated on a packed column containing the clay Bentone 34.'8 Similarly, when silver salts are incorporated into a liquid phase such as a glycol, special selectivity for olefins is obtained.") Laub and Purnel14"

168

GAS CHROMATOGRAPHY

have described the techniques for combining several liquid phases in the same column (or coupled columns) to get the best ratio for a given separation. Separations that are best effected by GSC such as the separation of oxygen and nitrogen. Some O T columns, the so-called PLOT columns, are commercially available, but they are few in number. Formerly, packed columns were thought to be more rugged than their OT counterparts, but that is probably no longer true since O T columns are now bonded and cross-linked.

7.4

STATIONARY PHASE

The selection of a stationary phase (SP) is one of the most important decisions in setting up a method. Because GC is so easy to perform, the process of selection has often been made on a trial-and-error basis and hundreds of liquids have been employed. The theoretical approach, while very complex, has been successful in aiding in the choice of liquid and has permitted the reduction in number of necessary liquids. The stationary liquid phase is coated on a solid support and packed into a column (packed-column GC) or it is coated on the wall of an open tube (WCOT), as has been discussed. The higher efficiency of OT columns has reduced the necessity for many selective liquids, and the number of O T columns necessary to analyze for “all” types of analytes is smaller than for packed columns. A persistent problem with liquid phases is their upper temperature limits. As G C is used at higher and higher temperatures in an attempt to extend its usefulness to higher and higher boiling analytes, the stationary-phase vapor pressure gets higher and higher and is evidenced as column bleed, an upward sloping baseline. Thus, a major objective has been to find liquids with increasing boiling points-polymers of high molecular weight, special new polymers, and bonded phases. Reducing bleeding is the subject of a recent study4’ that contains an extended discussion of t h e topic and an example of column bonding reactions. We have already discussed the special crosslinking reactions and surface preparations that are used to achieve stability of liquid phases on O T columns. Liquid phases also have lower temperature limits, represented by their melting points or glass transition temperatures. In most cases these temperatures are low enough so that they are well below the normal working GC temperatures. However, there are some exceptions, so the minimum should also be checked before use. Also, some special applications use subambient

7.4

STATIONARY PHASE

169

Figure 7.17. Basic structural unit o f silicone polymers used as stationary phases in GC.

temperatures at the beginning of a programmed temperature run, and these are often below the recommended minimum temperature of the stationary phase. Typical Liquids Silicone polymers, differing in the extent to which they contain polar functional groups, have become the most popular class of liquid phases. The silicone polymers have the backbone structure shown in Figure 7.17 in which all the alkyl groups are shown as methyl. These polydimethylsiloxane polymers are abbreviated PDMS. Replacement of the methyl groups with others of higher polarity yields polymers of increasing polarity. One company, Ohio Valley (OV) Speciality Chemical, produces a line of these products, and Table 7.10 contains a list of their products by OV number. Those indicated by an asterisk are available in a special formulation on OT columns, and most of them contain 1% vinyl groups to improve stability. One popular phase not included is a 5% PDMS phase that contains 5% phenyl and 95% methyl side groups, and is sold as DB-5. Several manufacturers’ phases equivalent to the PDMS series of phases are given in Table 7.11. Another popular series of polar phases is the polyethyleneglycols (PEGS); of this series the Carbowax* line is one of those commercially available. The structure of this polymer is OH-(CH,-CH,

O),,-H

-

and the approximate average weight is given as a numerical value in the naming of the particular polymer. The highest boiling member available in this series is Carbowax 20M, which has an average molecular weight of 20,000. It can be used from 60 to 225°C in packed columns and up to 280°C in special OT bonded configurations. Sol-gel technology has been used to prepare PEG coatings on OT columns3* yielding higher temperature limits (up to 320°C). An organic-inorganic hybrid sol-gel is chemically bonded to the fused-silica capillary wall producing a highly stable SP. The PDMS and PEG series of stationary phases on O T columns are usually sufficient for most applications. The choice of a particular manufac-

170

GAS CHROMATOGRAPHY

Table 7.10 Characteristics of Silicone Polymers by OV Number

Name ov-1 o v - 101 OV-3 OV-7

ov-1 1 OV-17 OV-61 OV-73 ov-22 OV-25 OV- 105 ov-202 ov-210 OV-215 OV-225 OV-275

OV-330 OV-35 1 OV- 1701

Type Dimethylsilicone gum Dimethylsilicone Phenylmethyldimethylsilicone, 10% phenyl Phenylmethyldimethylsilicone, 20% phenyl Phenylmethyldimethylsilicone, 35% phenyl Phenylmethylsilicone, 50% phenyl Diphenyldimethylsilicone Diphenyldimethylsilicone gum Phenylmethyldiphenylsilicone Phenylmethyldiphenylsilicone Cyanopropylmethyldimethylsilicone Trifluoropropylmethylsilicone Trifluoropropylmethylsilicone Trifluoropropylmethylsilicone Cyanopropylmethylphenylmethylsilicone Dicyanoallylsilicone Silicone carbowax copolymer Polyglycolnitroterephthalic Dimethylphenylcyano substituted polymer

Temp. Min.

Temp. Limit ("C)

Toluene Toluene Acetone

100 20 20

325-375 325-375 325-375

Acetone

20

325-375

Acteone

0

325-375

Ace tone

20

350-375

Acetone Toluene Acetone Acetone Ace tone

20 20 20 20 20

325-375 325-350 350-375 350-375 275-300

Chloroform Chloroform Chloroform Acetone

0 20 20 20

250-275 275-350 250-275 250-300

Acetone Acetone Chloroform Acetone

20 30 50 20

250-275 250-275 250-270 300-325

Solvent

("0

turer from among those with varying proprietary bonding techniques is usually based on price, stability, bleed rate, and efficiency, as well as the manufacturers reputation. There are also a few PLOT columns providing for GSC applications. But there are in excess df 150 different SP choices for packed-column GC in a typical supply house catalog, reflecting the diversity indicated earlier. An example would be the polar-phase diethyleneglycol succinate, DEGS, which has a temperature range from 20 to 200°C and is used for separating fatty acid methyl esters. A new class of SP is the high-stability ionic liquids.43

7.5 TEMPERATURE EFFECTS Temperature is one of the two most important variables in GC. Retention times decrease as temperature increases because partition coefficients are

7.5 TEMPERATURE EFFECTS

Table 7.1 1

Equivalent Silicone Polymer Liquid Phasesa

Other Designations

Ohio Valley Number OV-1, 101 OV-7 ov-11 OV-17 ov-210, 202,215 ov-225 OV-35 1

OV-1701

171

SP-2100 RSL-I50 SE-54

SE-30

DB- I

SPB-I

SF-96

SE-52

DB-5

RSL-200

SP-2250 SP-2401

QF-I

DB- 17 DB-210

SPB-5 SPB-20 SPB-35 RSL-300

SP-1000 SP-2300 SP-2340 SP-2100

(FFAP) Silar 5C Silar 10C UC W982

DB-225 AT-I000

DC-200

RSL-500

cs-10 DB- 1701

'Equivalent Rohr~chneider/McReynolds values.

temperature dependent in accordance with the Clausius-Clapeyron equation: A,;./ (7.10) log p ' ) = - 2.3 , f l T + constant where A V ' is the enthalpy of vaporization and is assumed to be constant over the range of temperatures investigated. The analytes vapor pressure p" decreases with increasing temperature, resulting in a decrease in partition coefficient and in retention volume. Verification of this relationship in GC is provided by Figure 7.18, in which the log of the net retention volume is plotted versus 1/T in accordance with Eq. (7.10). The slope of each line is proportional to that analytes enthalpy of vaporization, and the fact that straight lines are obtained indicates that the enthalpy is constant, as assumed. To a first approximation, the lines in Figure 7.18 are parallel, indicating that the enthalpies of vaporization for these compounds are nearly the same. A closer inspection reveals that the lines diverge slightly at low temperatures. From this observation we can draw the generalization that GC separations are usually better at lower temperatures. But examine the two analytes, n-octane and rn-fluorotoluene; their lines cross at 140°C. At this temperature, they cannot be separated; at a lower temperature the toluene elutes first; at a higher temperature the reverse is true. It is not common for elution orders to reverse, but it does happen in a few cases. The effect of temperature on column efficiency is quite complex,44 and no generalizations can be drawn. Usually it is minor and of considerably less

172

GAS CHROMATOGRAPHY

Figure 7.18. Temperature dependencc of retcntion volume. Reprinted with permission from W. E. Harris and H. W. Habgood, Tulunta 1964, I/, 11s. Copyright 1964, Pergamon Journals, Ltd.

importance than the effect on column thermodynamics. The latter is of such importance that temperature programming has become very important and is a standard feature of all research instruments. Programmed Temperature GC Programmed temperature GC (PTGC) is the process of increasing the column temperature during a run. As we have just seen, the increasing temperature will cause the partition coefficients of the analytes still on the column to decrease, and they will move faster through the column, yielding decreasing retention times. The effect can be seen in Figure 7.19. Some major differences between the two runs can be seen and are typical of PTGC. For a homologous series, the adjusted retention times are logarithmic under isothermal conditions, as we saw earlier, but they are linear when

-

L..I" .

173

TEMPERATURE EFFECTS

7.5

c13

'14

, . ) [

,

J

L

,

,

[

,

,

,

,

I

,

Retention time (min)

50 to 25QC at @C/min

t

0

I

I

I

I

4

8

12

16

I

20

I

I

I

24

20

32

Retention time (min)

I

36

(b)

Figure 7.19. A comparison o f ( ( I ) isothermal and ( h ) temperature-programmcd separation of n-paraffins. Reprinted with permission from H. McNair and J. Miller, Basic C;us Chromatography, John Wilcy & Sons. Copyright 1998: this material is uscd by pcrmission of John Wiley 24 Sons, Inc.

temperature programmed. The programmed run was begun at a lower temperature (50°C) than the one used for the isothermal run (150°C), which facilitated the separation of the low-boiling paraffins. It ended at a higher temperature (250°C), which increased the number of paraffins detected and extended the range to C2,versus C,, for the isothermal run. The peak widths are about equal in PTGC, while some fronting is evidenced in the higher boilers in the isothermal run. Since the peak widths d o not increase in PTGC, the heights of the late-eluting analytes will be increased (peak areas are constant), providing better detectivity. The advantages and disadvantages of PTGC are summarized in Table 7.12. Programmed temperature operation is good for screening new samples. A maximum amount of information about the sample composition is ob-

174

GAS CHROMATOGRAPHY

Table 7.12

Advantages and Disadvantages of PTGC

Advantages

Disadvantages

1. Better separation for wide boiling mixtures 2. Constant peak width and shape 3. Decreases time of analysis 4. Sample introduction less critical 5 No loss of quantitative accuracy 6. Lower limits of detection

1. Has additional instrument requirements 2 Ghost peaks may occur 3. Extra time required for cool down 4 Fewer stationary phases can be used

5

Subject to baseline drift and noise due to column bleeding

tained in minimum time. Usually one can tell when the entire sample has been eluted, which is often a difficult judgment to make with isothermal operation. The theory of PTGC has been thoroughly treated by Harris and Habgood4' and by M i k k e l ~ e n .The ~ ~ following discussion has been taken from a simple but adequate treatment by G i d d i n g ~ . ~ ' The dependence of retention volume on temperature was illustrated in Figure 7.18. Let us determine approximately what temperature increase is necessary to cut a given retention volume in half; that is: (7.1 1) or

where AT is the difference between the two temperatures T , and T,, and T is t h e average of the two temperatures. Taking the log and rearranging, we get

AT=

0.693 $4 T A'W'

(7.13)

Assuming Trouton's rule that A,WYT,, = 21 and a boiling temperature of 227°C (500 K) for a typical sample,

AT=

0.693 x 2 x (SOO)2 21 (500)

~

30"c

(7.14)

As an approximation, then, an increase in temperature of 30°C will cut the adjusted retention volume in half. (This rule of thumb is also useful for isothermal operation, of course).

7.5

85 100

130

160

190

TEMPERATURE EFFECTS

220

175

250 265

Temperature ("C)

Figure 7.20. Step-function approximation to programmed temperature GC. Reprinted from reference 47 courtesy of the Chemical Education Division, American Chemical Society.

The effect of temperature programming o n the migration of a typical analyte through a column is shown in Figure 7.20, where the 30°C value is used to generate the step function. The retardation factor R will double for every 30" increase since the rate of movement of the analyte through the column is equal to (Rxu).Final elution from the column is arbitrarily taken as occurring at 265"C, as shown in the figure. In actuality, the movement of the analyte through the column would proceed by the smooth curve also shown in the figure since the temperature programming would be gradual and not stepwise, as assumed by our model. If 1 is taken as the distance the analyte moved through the column in the last 30" increment, then one-half is the distance it moved in the previous (next-to-last) 30", one-quarter in the 30" before that, and so on. The sum of this series (i,f,i, and so on) approaches 21, which must equal the total length L of the column. Hence the analyte traveled through the last half of the column in the last 30"; it started moving slowly and speeded up as the

176

GAS CHROMATOGRAPHY

temperature increased. The operation of PTGC can be envisioned as follows: The sample is injected onto the end of a relatively cold column, and its components largely remain there; as the temperature increases, the analytes "boil off" and move down the column at increasing rates until they elute. It is for these reasons that the injection technique is not critical in PTGC and that all peaks have about the same peak widths since they spend about the same amount of time actively partitioning down the column. However, it must be remembered, as shown in Figure 7.18, that a given net retention volume can change with temperature relative to other analytes, so it is possible that the order of elution of a PT run could differ from that in an isothermal run. One analyte might overtake another during the programming process.4x For a variety of reasons, isothermal operation is often preferred in the workplace. If an initial screening is done by PTGC, one might wish to know what isothermal temperature would be the single best one to use. Giddings has called this isothermal temperature the significant temperature, T'. Using reasoning based on the 30°C value, he has found that

T'

=

T, - 45

(7.15)

where Tf is the final temperature (in "C), the temperature at which the analyte of interest eluted in the PT run. Two other important variables are the programming rate and t h e column length. In general, one does not vary the length but uses a rather short column (and lower temperatures). The rate is often chosen to be fast enough to save time but slow enough to get adequate separations, which results in rates between 5 and lO"/min. However, for OT columns, there is considerable evidence that slow rates (around 2.5"/min) are refera able.^" A simple optimization process has been publishedi"; it is complicated by a peak reversal, which is one unexpected feature of PTGC as was mentioned earlier. Some chromatographs are provided with ovens that can be operated below ambient temperature, thus extending the range of temperature programming. Examples can be found in the extensive review of cryogenic GC by Brettell and Grob." Programmed temperature GC can be used to help achieve retention time locking (RTL), which is the process whereby all analytes maintain the same retention times regardless of the chromatographic system. One method for finding the optimum program rate is called method translation, which considers the void time, t,, to be the fundamental time unit in GC. All heating rates are expressed in units of void time. This process has been described" and applied to the achievement of fast analyses" (treated later in this chapter). Agilent has offered software for this purpose since 199tLS4An application to toxicological screening for drugs in blood provides more rationale and details."

7.6

SPECIAL TOPICS

177

7.6 SPECIAL TOPICS

Those special topics of greatest interest can be found in other chapters. Included are GC/MS (Chapter lo), sampling methods including head space and SPME, and derivatization (Chapter 14), and multidimensional G C and chiral G C (Chapter 15). A few other topics are included here. Fast GC

Compared to other methods of analysis, G C is relatively fast, but analysts are always interested in making a process as fast as possible in order to save time. The steps necessary to achieve fast G C analyses are well known and many have been recommended in this chapter. However, since fused-silica OT columns have become popular and G C has achieved the status of a mature technique, chromatographers have made a special effort to minimize the time needed for an analysis. Their efforts have been summarized in two reviews.", Briefly, they are:

'"

Using OT capillaries of very small inner diameter such as 100 p m . Using very thin films of stationary phase. Using very fast program rates, often optimized by method translation.s3 Instruments capable of achieving these rates have become commercially available to support this requirement. Using hydrogen as the carrier gas, often, but not always at high flow rates. Using a detector with a fast time constant and operated at a vacuum. Again, this requires a suitable instrument. Fortunately, the MS detector is available. One example of the application of these general principles to the analysis of fats and oils5' provides specific details. Also, the chapter by Sacks" provides an up-to-date summary. Programmed Temperature Vaporizers and Large-Volume Injections

The injection of large samples into a G C is another procedure with an obvious advantage-better detectability. However, the popular OT columns are not capable of handling large samples, so some special techniques are necessary to make large-volume injections (LVI) possible. The most impor-

178

GAS CHROMATOGRAPHY

tant ones59 are: Cold on-column (COC) Programmed temperature vaporization (PTV) Loop-type injection Vapor-overflow A combination of the latter two, called AT-column injection") Cold on-column and PTV are the two most widely used techniques. PTV requires a special injection port that has the capability to heat up and cool down very rapidly. Gas Chromatography Analysis of Nonvolatiles Since nonvolatiles can be analyzed by LC, those methods are often preferred. But, there are a few alternatives that permit some of them, such as sugars and amino acids, to be run by GC. One possibility is to make volatile derivatives, and that topic, as has already been mentioned, is discussed in Chapter 13. Other possibilities are pyrolysis and inverse GC. Pyrolysis GC is used mainly for identifying polymers from the pattern of peaks that is obtained.6' Instead of injecting a sample, the polymer is put in a furnace capable of being heated very rapidly to a sufficiently high temperature to cause its thermal decomposition. The decomposition products are chromatographed by PTGC and are reproducible enough to provide peak patterns that can be compared with those from known polymers to provide qualitative identifications. Inverse G C is used to produce data that are the opposite or inverse of normal G C methods. Since its objective is to get information about large nonvolatile molecules that cannot be run by normal GC, the sample, composed of large molecules (often polymers or fibers) is used as the stationary phase and is then subjected to investigation with small molecules that serve as probes. In effect, the roles of solute and solvent are reversed. An application to the study of surface and bulk properties of pharmaceutical materials has been published,62 and further details on this technique are included in Condor and Young's book." Inorganic GC Most inorganic compounds are not volatile enough for analysis by GC, with the exception of some fixed gases such as CO, and SO,. Consequently, inorganic G C is usually treated as a separate topic that is concerned with the

7.7

SUMMARY AND EVALUATION

179

formation of volatile derivatives (see Chapter 14) and the use of detectors having elemental specificities like those included in Table 7.8. A thorough review has been written by who has also contributed a chapter to an inorganic chromatography book.”’ A comprehensive review by Bachmannbb includes a large section on inorganic compounds. Simulated Distillation

When GC replaced distillation as a method of analyzing petroleum products, the GC equivalent of distillation was formulated by groups such as ASTM. Appropriately enough, the methods are called simulated d i ~ t i l l a t i o n . ~ ~ 7.7 SUMMARY AND EVALUATION

Throughout this chapter recommendations have been made for optimizing GC performance and some additional discussion was included in Chapter 3. One final recommendation is a series of three studies on this topic.68-70 Updates on GC can be found in the biennial reviews in Analytical Chemistry, the last one of which was published in 2004.” Modern practices of GC are covered in the new edition of Grob and Barry’s book,” which also includes a chapter on optimization by Hinshaw. See also the list of selected references at the end of this chapter. Applications to petroleum products, the clinical and pharmaceutical industries, environmental investigations, and forensic analysis are covered in Grob and Barry’s book.” Table 7.13 concludes this chapter with a summary of the advantages and disadvantages of GC. In summary, if a s a m p l e is volatile e n o u g h f o r analysis by GC, use GC.

Table 7.13 Evaluation of GC

Advantages

Disadvantages

1. Efficient, selective, and widely applicable 2. Easily combined with MS 3. Fast 4. Inexpensive and simple

1. Samples must be volatile

5. Easily quantitated and automated 6. Reuires only a small sample (milligram range) 7 Nondestructive detectors available

2. Not suitable for thermally labile samples 3. Difficult for large samples (preparative scale) 4. Only fair for qualitative analysis

180

GAS CHROMATOGRAPHY

REFERENCES 1. A. T. James and A. J. P. Martin, Biochem. J . 1952, 50, 679.

2. H. M. McNair and J. M. Miller, Basic Gas Chromatography, Wiley, New York, 1998. 3. H. McNair, LC-GC 1992, 10, 239. 4. H. P. M. van Lieshout, H-G. Janssen, and C. A. Cramers, A m . Lab. 199, 27(12, August), 38-44 5. S. Kenworthy, J. Miller, and D. E. Martire, J . Chem. Educ. 1963, 40, 541. 6. J. M. Santiuste, 1. Chrornatogr. Sci. 2003, 41, 215-222. 7. J. A. G. Dominguez, J. C. Diez-Masa, and V. A. Davankov, Pure Appl. Chem L . 2001, 73, 969-992. 8. M. J. E. Golay, in Gas Chromatography, V. J. Coates, H. J. Noebels, and 1. S. Fagerson (eds), Academic, New York, 1958, pp. 1-13. 9. E. S. Kovats, Helv. Chim. Acta 1958, 41, 1915. 10. G. D. Mitra and N. C. Saha, J . Chromatogr. Sci. 1970, 8, 95. 11. R. A. Hively and R. E. Hinton, J . Gas Chromatogr. 1968, 6 , 203; L. S. Ettre and K. Billeb, J . Chromatogr. 1967, 30, I ; N. C. Saha and G. D. Mitra, J . Chromatogr. Sci. 1970, 8 , 84; P. G. Robinson and A. L. O d d , J . Chromatogr. 1971, 57, 11. 12. W. 0 . McReynolds, Gas Chromatographic Retention Data, Preston Technical Abstracts, Evanston, IL, 1966. 13. L. Rohrschneider, .I. Chromatogr. 1966, 22, 6. 14. W. R. Supina and L. P. Rose, J . Chromutogr. Sc. 1970, 8, 214. IS. W. 0. McReynolds, J. Chrornatogr. Sci. 1970, 8 , 685. 16. C. F. Poole and S. K. Poole, J . Chromutogr. A 2002, 965, 263-299. 17. C. F. Poole, The Essence of Chromatography, Elsevier, Amsterdam, 2003. 18. V. G. Berezkin, in Advances in Chromatography, Vol. 41, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 2001, Chapter 9. 19. K. Grob, Classical Split and Splitless Injection iri Capillaty Gas Chromatography, 3rd ed., Heuthig, Heidelberg, 1993. 20. J. V. Hinshaw, LC-GC No. Am. 2000, 18, 1234-1241. 21. L. S. Ettre, Open Tubular Columns, Plenum, New York, 1965; Open Tubular Columns, Pub. # GCD-35, Perkin-Elmer Corp., Nonvalk, CT, 1973. 22. W. Jennings, Gas Chromatography with Glass Capillary Columns, 2d ed., Academic, New York, 1980. 23. M. L. Lee, F. J. Yang, and K. D. Bartle, Open Tubular Column Gas ChromatoAvaphy; Theoty and Practice, Wiley, New York, 1984. 24. See, for example, M. L. Lee, R. C. Kong, C. L. Woolley, and J. S. Bradshaw, J . Chromatogr. Sci. 1984, 22, 136. 25. R. R. Freeman (ed), High Resolution Gas Chromatography, 2nd ed., HewlettPackard, Avondale, PA, 1981. 26. B. J. Tarbet, J. S. Bradshaw, K. E. Markides, M. L. Lee, and B. A. Jones, LC-GC Mag. 1988, 6, 232-248. 27. J. A. Hubball, P. R. Di Mauro, E. F. Barry, E. A. Lyons, and W. A. George, J . Chromatogr. Sci. 1984, 22, 185.

REFERENCES

181

28. M. L. Duffy, Am. Lab. 1985, I7(10), 94. 29. L. S. Ettre, Chromatogruphia 1984, 18, 477. 30. R. K. Simon, Jr., J . Chromatogr. Sci. 1985, 23, 313. 31. H. Y. Tong and F. W. Karasek, Anal. Chem. 1984, 56, 2124. 32. J. E. Lovelock, Anal. Chem. 1961, 33, 162. 33. D. J. David, Gas Chromatographic Detectors, Wiley-Interscience, New York, 1974. 34. M. Dressler, Selectiue Gas Chromatogruphic Defectors, Elsevier, Amsterdam, 1986. 35. R. E. Sievers (ed), Selectiue Detectors: Enuironmental, Industrial, and Biomedical Applications, Wiley, New York, 1995. 36. H. H. Hill and D. G. McMinn (ed), Detectorsfor Capillary Chromatography, Wiley, New York, 1992. 37. L. A. Colon and L. J. Baird, in Modern Practice of Gas Chromatography, 4th ed, R. L. Grob and E. F. Barry (eds), Wiley, Hoboken, NJ, 2004. 38. S. F. Spencer, Anal. Chem. 1963, 35, 592. 39. J. H. Purnell, in Gas Chromatograph~-1966, A. B. Littlewood (ed), Institute Petroleum, London, 1967, p. 3. 40. R. J. Laub and J. H. Purnell, J . Chromatogr. 1975, 112, 71; R. J. Laub, J. H. Purnell, and P. S. Williams, J . Chromutogr. 1977, 134, 249. 41. J. de Zeeuw, L. Joziasse, J. Peene, P. Heijnsdijk, and D. Zwiep, Am. Lab. 2003, 35( 16), 30-37. 42. C. Shende, A. Kabir, E. Townsend, and A. Malik, .4nal. Chem. 2003, 75, 35 18-3530. 43. J. L. Anderson and D. W. Armstrong, Anal. Chem. 2003, 75, 4851-4858. 44. W. E. Harris and H. W. Habgood, Talanta 1964, / I , 1 15. 45. W. E. Harris and H. W. Habgood, Programmed Temperature Gus Chromatography, Wiley, New York, 1966. 46. L. Mikkelsen, Adu. Chromutogr. N . Y . 1966, 2, 337. 47. J. C. Giddings, J . Chem. Educ. 1962, 39, 569. 48. L. Bingchcng, L. Bingchang, and B. Koppenheofer, Anal. Chem. 1988, 60, 2 135 -2 137. 49. L. A. Jones, S. L. Kirby, C. L. Garganta, T. M. Gerig, and J. D. Mulik, Anal. Chem. 1983, 55, 1354. SO. J. V. Hinshaw, LC-GC 1991, 9, 470-473. 51. T. A. Brettell and R. L. Grob, Am. Lab. 1985, 17(10), 19; and ( l l ) , 50. 52. M. Blumberg and M. S. Klee, Anal. Chem. 1998, 76, 3828-3839. 53. M. S. Klee and L. M. Blumberg, 1. ChromatobT. Sci. 2002, 40, 234-247. 54. Agilent, GC Method Translation Freeware, Agilent Technologies Inc., Wilmington, DE, 1998. Can be downloaded from www.agilent.com. 55. I. Rasanen, I. Kontinen, J. Nokua, I. Ojanpera, and E. Vuori, J . Chromatogr. B 2003, 788, 243-250. 56. C . A. Cramers, H-G. Janssen, M. M. van Deursen, and P. A. Leclercq, J . Chromatogr. A 1999, 856, 315-329. 57. L. Mondello, P. Q. Tranchida, R. Costa, A. Casilli, P. Dugo, A. Cotroneo, and G. Dugo, J . Sep. Sci. 2003, 26, 1467-1473.

182

GAS CHROMATOGRAPHY

58. R. D. Sacks, in Modern Practice of Gas Chromatography 4th ed., R. L. Grob and E. F. Barry (eds), Wiley, Hoboken, NJ, 2004. 59. W. Engewald, J. Teske, and J. Efer, J . Chromatogr. A 1999, 842, 143-161. 60. S. de Koning, M. Kurano, H-G. Janssen, and U. Th. Brinkman, J . Chromatogr. A 2004, 1023, 165-174. 61. V. G. Berezkin, V. R. Alishoyev, and I. B. Nemirovskaya, Gus Chromatography of Polymers, Elsevier, Amsterdam (reprinted) 1983; R. W. May, E. F. Pearson, and D. Scothern, Pyrolysis GC, Chemical Society, London, 1977; J. 0 . Walker and C. J. Wolf, J . Chromatogr. Sci. 1970, 8, 513; J. C. Hu, Adu. Chromatogr. N. Y., 1984, 23, 149. 62. J. Domingue, D. Burnett, and F. Thielmann, Am. Lab. 2003, 35(14), 32-37. 63. J. Condor and C. Young, Physiochemical Measurement by Gas Chromatography, Wiley, Chichester, UK, 1979. 64. P. C. Uden, J . Chromatogr. 1984, 313, 3-31. 65. P. C. Uden, in Inorganic Chromatographic Analysis, Vol. 78, Chemical Analysis Series, J. C. MacDonald (ed), Wiley, New York, 1985, Chapter 5. 66. K. Bachmann, Tulanta, 1982, 29, 1. 67. L. G. Chorn, J . Chromatogr. Sci., 1984, 22, 17. 68. J. V. Hinshaw, LC-GC No. Am. 2000, 18, 296-304. 69. J. V. Hinshaw, LC-GC No. Am. 2000, 18, 610-616. 70. J. V. Hinshaw, LC-GC No. Am. 2000, 18, 1040-1047. 71. G. A. Eiceman, J. Gardea-Torresdey, E. Overton, K. Carney, and F. Dorman, Anal. Chem. 2004, 76, 3387-3394. 72. R. L. Grob and E. F. Barry (eds), Modern Practice of Gas Chromatography, 4th ed. Wiley, Hoboken, NJ, 2004.

SELECTED BIBLIOGRAPHY Ettre, L. S., and Hinshaw, J. V., Basic Relationships of Gus Chromatography, Advanstar Communications, Cleveland, OH, 1993. Grob, R. L. and Barry, E. F., (eds), Modern Practice of Gus Chromatography, 4th ed., Wiley, Hoboken, NJ, 2004. Hinshaw, J. V., Monthly column on “GC Connections,” LC-GC No. Am. Advanstar Communications, Cleveland, OH, 1993. www.chromatogruphyonline.com. McNair, H. M. and Miller, J. M., Basic Gus Chromatography, 1st ed., Wiley, Hoboken, NJ, 1998. Poole, F., The Essence of Chromatography, Elsevier, Amsterdam, 2003.

LIQUID CHROMATOGRAPHY IN COLUMNS

The original form of chromatography introduced by Tswett in 1903 was LC carried out in columns, but its development was slow until after GC was introduced. When the principles and theories of G C were applied to LC, it too developed quickly. From the late 1960s until thc present, the pace of research in LC has accelerated, and many further improvements and modifications are to be expected in the future. This is especially true in the biomedical area since LC is much more amenable to the analysis of large biomolecules than is GC; Chapter 15 contains more on this topic. Versatility is much greater in LC than in GC because both phases, stationary and mobile, affect the separation and because a wide range of phases (stationary and mobile) can be used in LC. Stationary phases include many types of bonded phases with a wide range of polarities, as well as materials with ion exchange and sieving (size exclusion) properties. The configuration of the chromatographic bed can be a column, covered in this chapter, or a planar surface as in thin-layer chromatography (TLC), covered in Chapter 11. Most of the discussion in this chapter will be on HPLC, high-performance LC, using sophisticated instrumentation to handle a wide spectrum of samples.

Chromatography: Concepts and Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

183

184

LIQUID CHROMATOGRAPHY IN COLUMNS

Figure 8.1. Classification of liquid chromatographic techniques. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical ChemiJtq in u G M P Enuironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

8.1

INTRODUCTORY CLASS1FlCAT1ONS

Figure 8.1 shows a classification of chromatographic methods, highlighting the column HPLC techniques. The abbreviations used will be defined as each of the techniques is discussed in this chapter. The first three methods (LSC, BPC, and IEC) are examples of the classic elution chromatographic process that has been discussed thus far in this book. The fourth (SEC) is based on a mechanical sieving process and not on equilibrium thermodynamics. Likewise, the fifth (affinity) is a special case; specific strong intermolecular attractions operate in the sorption stage and a major change in MP composition is necessary for the desorption stage. Since affinity chromatography finds greatest use with biomolecules, it is covered in Chapter 15. Note that parts of this classification scheme differ slightly from the one presented in Chapter 2 (Fig. 2.2). Each classification was selected to highlight particular features, and this chapter is organized according to Figure 8.1 To begin, let us look at some additional comparisons between modes of LC. Phase Polarities: Normal Versus Reversed

Tswett’s original work was with polar stationary phases (SP), and consequently most of the original work in HPLC in the 1960s was also with polar SPs such as silica gel. Polar SPs are used with nonpolar mobile phases (MP). Originally, this was the normal mode of operation. Later, chromatographers

8.1

INTRODUCTORY CLASSIFICATIONS

185

introduced columns with reversed polarities: the SP was nonpolar and the M P was polar. Over time, this reuersed-phase mode became the more popular one. This may be somewhat confusing to a novice chromatographer; the normal-phase mode is not the one that is most common or normal, reversed phase is. Liquid-Solid (LSC) Versus Liquid-Liquid (LLC) Chromatography In LSC, the SP is a solid, and polar silica gel is most common. Thus, most LSC is operated in the normal mode, as just discussed. In LLC, a liquid SP must be coated on a solid support material as is commonly done in packedcolumn GC. That liquid can be either polar or nonpolar, but the MP with which it is used must have the opposite polarity (be immiscible) so that the SP is not washed off the support, and it must be saturated with the SP liquid. One interesting LLC system that was formerly popular was a ternary system such as water/ethanol/isooctane, which separates into two immiscible phases; one can be held immobile as the SP and the other is used as the MP. However, most LLC systems, including that used by Martin and Synge in their Nobel prize-winning study on LLC, have a polar SP and a less polar MP, the combination called normal phase. Reversed-phase LLC systems are not as stable, and LLC is not used much any more, so it was not included in Figure 8.1 and will not be discussed further. It has been replaced by the very popular bonded-phase (BPC) mode, which will be discussed in greater detail later.

Figure 8.2.Low-pressure LC schematic. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chemistry in a GMP Environment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

186

LIQUID CHROMATOGRAPHY IN COLUMNS

Table 8.1 Advantages of HPLC

Advantage High speed High resolution High sensitivity High accuracy Automated systems

Explanation Analysis times measured in minutes or seconds Columns tightly packed with small, uniform particles Parts-per-million (ppm) to sub-parts-per-billion (ppb) detection limits High precision sampling devices and good standards yield accurate analyses Unattended operation, from sample preparation to report generation

Low-Pressure Versus High-pressure Liquid Chromatography

Low-pressure, open-column (or gravity flow) chromatography was the original approach developed by Mikhail Tswett in 1903. This approach is still used today, primarily in organic and biological laboratories, for the purification of samples that require only modest resolution. The technique usually employs a large-diameter column ( > 1 cm i.d.1 packed at ambient pressures with relatively large particles ( > 37 pm), and little other instrumentation except perhaps a fraction collector. The result of this approach is a loosely packed (low-resolution) column with low-pressure requirements (see Fig. 8.2). A more m o d e r n version of low-pressure operation is called flash chromatography; it will be described in more detail later. Also, such columns are often used for preparative work, which will also be discussed later.

Figure 8.3. HPLC schematic. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chemistry in a GMP Enuironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

INTRODUCTORYCLASSIFICATIONS

8.1 COLUMN : 3 cm 3pM ODS

187

NITROBENZENE

5

0

10

TIME ( SECS )

Figure 8.4. Example of high-speed HPLC. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chemistry in a GMP Enuironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

High-performance liquid chromatography has the additional advantages listed in Table 8.1. HPLC columns are tightly packed ( > 6000 psi) with small and uniformly sized particles ( 5 10 pm). This column design results in high-resolution separations but adds demanding instrumental requirements to the system. The tightly packed column demands a high pressure (up to 5000 psi) and requires a high-pressure pump and a specialized injection device that isolates the sample (at ambient pressure) from the high-pressure flow. After the column, an in-line detector is introduced, allowing for the continuous monitoring of the eluents. Finally, the components are connected with low-volume tubing and connectors in order to minimize the dispersion of the sample as it travels through the system. A schematic of a complete

OMe OMe

0

OM,

0

OMe

OMe

* M OMc

0

OMc

0

1

2 3 4 Time [mini

5

Figure 8.5. Example of high-resolution HPLC. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chemistry in a GMP Enuironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

188

LIQUID CHROMATOGRAPHY IN COLUMNS

b 0

Aflatoxins In peanut butter; Fluorescence Detection

1

10

20

1. Aflatoxin B 1

5 ppb

2. Aflatoxin G 1

1 ppb

3. Aflatoxin B 2

3 ppb

4. Aflatoxin G 2

1 ppb

30minutes

Figure 8.6. Example of high-sensitivity HPLC. Reprinted with permission from J. Miller and J. Crowther (eds), Anulytical Chemisiy in u G M P Etzvironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wilcy & Sons, Inc.

instrument is shown in Figure 8.3. Most of this chapter is concerned with the HPLC form of LC. Clearly, HPLC has important advantages over the low-pressure method, including the possibility of easily performing a reproducible analysis with a gradient. Figures 8.4-8.6 clearly illustrate the high speed, resolution, and sensitivity, respectively. T h e main disadvantage at the present time is the lack of a detector that is both universal and sensitive. as listed in Table 8.2. lsocratic Versus Gradient Elution

Isocratic elution refers to the technique of using constant solvent composition throughout the chromatographic analysis. During gradient elution, the mobile phase is changed from a “weak” to a “strong” solvent during the analysis. Gradients are generally chosen for samples with large numbers of components o r those in a dirty or unknown matrix. Table 8.2

Disadvantagesof HPLC

Disadvantage Expensive instrumentation Experience required Lack of universal and sensitive detectors Expensive supplies Not good for qualitative analysis

Explanation Typical HPLC systems cost $30K to $50K Complex chemistry and instrumentation Maybe mass spectrometery will meet this need soon Columns, fittings, and other consumables are expensive Retention time is characteristic but not a positive ID; spectroscopic methods needed for confirmation

8.1

INTRODUCTORY CLASSIFICATIONS

189

Figure 8.7. Typical isocratic separation. Reprinted with permission from J. Miller and J. Crowther (eds), Anulyricul Chemistry in u G M P Enuironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

If a sample contains analytes that have widely divergent affinities for the column, a gradient is useful in shortening the analysis time and improving the shape of the peaks, effects similar to the use of programmed temperature in GC. Figure 8.7 illustrates a typical isocratic analysis of a complex mixture. The first few peaks elute too close to the void volume of the column, suggesting that the mobile phase is too strong for these compounds. Also, the last peaks are short and broad with very long retention times, indicating that the mobile phase is too weak for these compounds. The solution to these problems is to begin with a weaker solvent and gradually increase the solvent strength throughout the course of the analysis. This is the definition of a gradient. Figure 8.8 shows that with the gradient (and a sample similar to the one in Fig. 8.7), the resolution of the early eluting peaks is improved, and the widths of the later peaks have been decreased while their heights have increased. The overall gradient separation yields more consistent peak widths, improved sensitivity, and shorter analysis times than would be possible for the corresponding isocratic separation. The components of the gradient for the M P can be binary, tertiary, or quaternary. The starting composition should be weak, which means that it

-

Figure 8.8. Typical gradient elution separation. Reprinted with permission from J. Miller and J. Crowther (eds), Anulyticul Chemistry in u G M P Environment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

190

LIQUID CHROMATOGRAPHY IN COLUMNS

should be very different in polarity from stationary phase. In reversed-phase HPLC, the SP is nonpolar or hydrophobic, so the starting MP should be very polar (usually water plus an organic modifier). During the gradient run, the MP is made “stronger” by increasing the proportion of the less polar component [usually acetonitrile, methanol, or tetrahydrofuran (THF)]. The opposite is required for a normal-phase gradient; namely, the proportion of the polar solvent is increased during the run. Polarity is probably not the best concept to use for chromatographic descriptions of the nature of phases. Polarity per se is poorly defined and refers to properties of pure solvents, whereas in HPLC we are concerned with mixtures of solvents. What we really want is a solvent strength parameter, and that discussion, begun in Chapter 4, will be continued throughout this chapter. Nevertheless, chromatographers find it convenient to refer to polarity as though it were the correct and best parameter for describing chromatographic systems. Some additional aspects of gradient elution will be discussed in more detail later in this chapter. Stationary Phase

Chapter 3 included a general introduction to chromatographic stationary phases (SP). According to our earlier classification, the stationary phase can be a solid, a liquid, or a chemically bonded phase. In the latter two cases, the phase must be coated on, or bonded to, particles of a solid support, usually a porous material or the newer monolithic support structures. For HPLC, only a few materials have found widespread use as stationary solid supports; they are silica, synthetic polymers such as the styrene-divinylbenzene copolymer, and some polysaccharides. The most common types and uses are given in Table 8.3. High-performance LC theory predicts that small porous packings and thin SP layers are preferred (see Chapter 3). However, as the particle diameter is decreased and the efficiency increased, the pressure requirements also increase, except when the column length is decreased. As a consequence, the

Table 8.3 Characteristics of Stationary Phases

Chemical Composition Silica Zirconia Styreneedivinylbenzene Polysaccharides Other polymers

Used for

Limitations

LSC, LLC, BPC (NP and RP), SEC, IEC IEC, BPC (NP and RP) BPC, IEC SEC, IEC SEC

Soluble at pH 2 8

Compressible

8.1

INTRODUCTORY CLASSIFICATIONS

191

smallest particle sizes in routine use are about 3 pm, and very thin films have been achieved by bonding the SP to the solid support, making a monomolecular layer or a thin cross-linked polymer, as described in Chapter 3. An alternative method for achieving the effect of a small particle diameter is to coat a thin layer of porous solid on a solid core (such as glass). Such materials were widely used in the early days of HPLC and are usually called pellicular supports (from the Greek word for skin). Today, nanometer-thick layers of porous silica are coated on 3- to 5-pm solid particles. Pellicular solids are easier to pack than microporous solids, but they are less stable, have smaller capacities, and are more expensive. Since good packing methods are now known for the microporous solids, the pellicular solids have become much less popular. They are sometimes used in guard columns (discussed later in this chapter). Over the years a variety of methods for packing columns has been tried. To produce high-efficiency columns with small particles, the best method uses a wet slurry packing at high pressure. Originally a balanced-density solvent was recommended' to keep the particles suspended, but this required the use of expensive and toxic liquids such as tetrabromoethane. More recent work has demonstrated that good columns can be packed by using common solvents such as acetone or methanol. For many laboratories, the time required to learn how to slurry pack cannot be justified for the number of columns needed, and commercial columns are purchased. However, the procedures are well documented and the necessary equipment is available, so labs can pack their own columns. Monolithic columns represent a new approach to the preparation of ideal columns. They are prepared in situ and then enclosed in a sheath of plastic PEEK as described in Chapter 3. The high-pressure packing process is not needed, nor are end fittings needed to retain the packing. Applications with these columns are discussed in the bonded-phase section of this chapter. Conventional columns packed with 10-pm particles, are 150 mm long X 4.6 mm (id.) and made of straight stainless steel tubes, although heavy wall glass is also possible. When smaller-diameter packings are used, the columns are typically shorter: for example, 5-pm packing is generally used in 100- to 150-mm columns and 3-pm packing in SO- to 150-mm columns. Some workers have gotten better performance with steel columns that have a polished inner surface, but a recent study2 disputes this conclusion. More information about columns is included in the instrumentation section of this chapter.

Mobile Phase The selection of a mobile phase (MP) should be based on intermolecular forces, as discussed in Chapter 4 and elaborated for G C stationary phases in

192

LIQUID CHROMATOGRAPHY IN COLUMNS

Nitromethane Figure 8.9. Selectivity grouping of solvents in Tablc 8.4. Reproduced from the Journul of’ Chromatogruphic Scierzce by permission of Preston Publications, Inc.

Chapter 7. The original approach by Rohrschneider, followed by McReynolds, was to investigate the nature of GC stationary phases by using a few common chemicals as probes, whose retention on a given liquid reflected the extent of their interaction with the stationary phase. By choosing probes with selective interactions, Rohrschneider and McReynolds could determine a set of numbers that characterize the liquids (stationary phases) under study. Snyder’ applied this concept to LC solvents and defined a polarity index, P’, based on the behavior of three solvents, dioxane, ethanol, and nitromethane, as described in Chapter 4. A three-coordinate plot of his three selectivity parameters is shown in Figure 8.9. Nearly all the solvents studied fit into one of eight groups represented by circles in the figure; only triethylamine (1) and chloroform (8) do not fall into the expected regions. The constituents of each group are listed in Table 8.4. While the categorization is good, it is not perfect; for example, group 8 contains aromatic hydrocarbons and nitro-compounds that would not seem to have similar attractive forces. Consequently, this approach was later refined

8.1 INTRODUCTORY CLASSIFICATIONS

193

Table 8.4 Classification of Solvents in Figure 8.9

Group

Solvents

1

Aliphatic ethers, tetramethylguanidine, hexamethyl phosphoric acid amide, (trialkylamines)" Aliphatic alcohols Pyridine derivatives, tetrahydrofuran, amides (except formamide), glycol ethers, sulfoxides Glycols, benzyl alcohol, acetic acid, formamide Methylene chloride, ethylene chloride (a)' Tricresyl phosphate, aliphatic ketones and esters, polyethers, dioxane (b)' Sulfones, nitriles, propylene carbonate Aromatic hydrocarbons, halo-substituted aromatic hydrocarbons. nitro compounds, aromatic ethers Fluoroalkanols, rn-cresol, water (chloroform)'

2 3 4 5 6 7

8

"Somewhat more basic than other group 1 solvents. 'This group is rather broad and can be subdivided as indicated into groups 6a and 6b; however, normally there is no point to this in practical usage of the present scheme. ' Somewhat less basic than other group 8 solvents.

based on the Kamlet-Taft solvatochromic parameter scheme to produce a better diagram.4 Figure 8.10 is the revised diagram for aliphatic solvents, and it has been found useful in reversed-phase (RPLC) applications. To use the selectivity parameters in selecting solvent mixtures for LC, Snyder3 assumes additivity in proportion to the volume fractions of the solvents:

The solvents chosen, A and B, should be from two of the eight solvent groups and be as different as possible. For normal-phase LC, the solvents usually chosen are a nonpolar organic hydrocarbon modified with a polar organic such as an alcohol or ester. For reversed-phase LC, they are water modified with a polar organic such as methanol , acetonitrile, or THF. When three solvents are better than two, the ideal ternary mixture for normal-phase LC should have three solvents chosen from the three apices of the triangle. Snyder' recommended ethyl ether from group 1, methylene chloride from group 5 , and chloroform, which is usually included in group 8, because these three are widely separated on the plot (Fig. 8.10). Ternary solvents are not common in RPLC but have been used to good advantage when necessary.

194

LIQUID CHROMATOGRAPHY IN COLUMNS

BASIC

YRC0i I//

ACIDIC

Fluoroalcohols

CHCI,

CH,CI,

DIPOLAR

Figure 8.10. A replot of Figure X.9 for aliphatic solvent classes. Reprinted from L. Snyder, P. Carr, and S. Rutan, “Solvatochromically Based Solvent-Selectivity Triangle,” .IChronzatogr. . A, 1993, 656, 537-547, Copyright 1993, with permission from Elsevier.

8.2

CLASSIFICATION OF HPLC MODES

Liquid-Solid Chromatography

The original work by Tswett was carried out with CaCO,, a solid stationary phase (LSC), and most of the early LC work used silica gel as the stationary phase. Such a system was defined earlier as normal-phase LC (NPLC). The mode of action in LSC is adsorption, but the process is quite complicated because molecules of the mobile phase compete with analyte molecules for the active sites on the solid surface, and silica is energetically heterogeneous. Any water present in the system will be strongly attracted to the silica surface, and there is evidence that there can be two or three layers of water adsorbed on silica (chemisorption). The most strongly adsorbed water layer cannot be removed with dry solvents, but the other layers can be. To get silica completely dry requires heating to temperatures above 200°C. Because of its importance in LSC, silica was thoroughly studied, and further details can be found in a number of published works.5,6 Snyder’ has thoroughly studied adsorption chromatography, and some of his results were summarized in Chapter 3. He recommends covering 50-100% of the stationary solid surface with a monolayer of water. This requires up to

8.2 CLASSIFICATION OF HPLC MODES

195

Table 8.5 Solvent Polarities Listed by Solubility Parameters 6

Compound n-Pentane

n-Hexane Diethyl ether Cyclohexane n-Propyl chloride Carbon tetrachloride m-Xylene Ethyl acetate Benzene

Chloroform Dibutylphthalate Propyl nitrile Acetone Isopropanol Acetonitrile Methanol Polyacrylonitrile

Water

6

7.1 7.3 7.4 8.2 8.5 8.6 8.8 8.9 9.2 9.2 9.3 9.96 10.0 11.4 12.1 14.5 15.4 23.4

E('

(Alumina) 0.00 0.0 1 0.38 0.04 0.30 0.18 0.58 0.32 0.40 0.56 0.82 0.65 0.95

0.04 g of water per 100 m2 of surface, or about 4-1596 water added to the stationary phase. This process is not as easy as it sounds, and a long time is required for the system to come to equilibrium. This necessity to control the solid surface activity, coupled with the pH limitation (useful range is between 2 and 8) has contributed to a significant decrease in the use of LSC for analytical separations, although it is still popular for preparative LC. The mobile phase in LSC is chosen for the following reasons: (1) proper strength or polarity, ( 2 ) low viscosity, ( 3 ) compatibility with detector, and (4) volatility, if the analytes are to be recovered by evaporation of the mobile phase. If silica gel is the stationary phase, the main component of the mobile phase should be nonpolar. Typical solvents are listed in Table 8.5 in increasing order of the Hildebrand solubility parameter 6, so the ones near the top of the table are the ones to use. Also included in the table is the Snyder solvent parameter E " . Choosing the optimum mobile-phase polarity is the most difficult task in LSC. By using a mixture of liquids, the polarity can be adjusted to the necessary level. A simple approach uses the Snyder solvent parameter, as shown in Figures 8.11 and 8.12. In Figure 8.11, Yost and Conlon8 show how the Snyder parameter varies with composition for mixtures of liquids in heptane. Theoretically, mixtures that have the same E " value should produce equal retention volumes. A test of this concept is shown in Figure 8.12, where acrylamide was run on a silica gel column with the four mixtures indicated,

196

LIQUID CHROMATOGRAPHY IN COLUMNS

c

I

I

I

I

I

Solvent B

THF or acetonitrile Ethyl acetate or acetone

E0

Methylene chloride or chloroform lsopropyl ether or butyl chloride

CCL,

0

20

40

60

80

100

Percentage of

B

Figure 8.11. Polarity E ” of mixed solvents as a function of composition. Solvent A is hcptanc. Courtesy of Perkin-Elmer.

each having an F ” value of 0.53 achieved by mixing a second solvent with chloroform. The only surprise is the chromatogram obtained with the mixture of ethanol and chloroform; it produced undesirable peak broadening for unknown reasons. Another, and more complete, listing of E ” values for mixed solvents is shown in Figure 8.13, taken from the work of Saunders.’ The dashed line at a value of 0.3 is given as an example of the use of these data. This line cuts through six different solvent mixtures, all of which should, in theory, produce the same results for a given sample. Using a similar procedure, or just trial and error, a mobile phase can be selected. The optimum situation exists when the analytes in the sample elute with retention factors in the range of about 1-5. One rule of thumb” is that an increase of 0.05 units in 6‘’ will decrease the retention factor by one-half to one-quarter. The previous examples were for isocratic operation. For gradient elution, the mobile phase should become more “polar.” In the simplest case, increasing amounts of a polar solvent are added to a nonpolar solvent such as hexme.* Some recommended gradients are given in Tables 8.6 and 8.7. *Contrary to popular opinion, tz-hexanc is surprisingly toxic. Its long-term exposure limit is as low as 20 ppm. A safer solvent is 2-methylpentane, i-hexane, which has a limit of SO0 ppm.

8.2

t

12 rnin

95% Chloroform

5% Methanol

CLASSIFICATION OF HPLC MODES

22 rnin

197

4k94% Chloroform 6% Ethanol

L11

16 minInj.

75% Chloroform 25% Acetonitrile

13 rnin

75% Chloroform 25% THF

Figure 8.12. Comparisons o f chromatograms of acrylamide with four mobile phases, each with a t;" value of 0.53. Column: 50 cm x 3 mm Sil-X. Flow: 1 mL/min. Courtcsy of Perkin-Elmer.

Snyder and Saunders in their thorough discussion,"' have recommended that an optimum solvent program should have a change in F " of 0.04 unit per column volume of solvent. The chromatogram shown in Figure 8.14 uses the series 2 gradient from Table 8.7 and represents the separation of an extreme mixture ranging from the nonpolar hydrocarbon squalane to the polar sugar glucose. Note, however, the irregular peak shapes, some of which can probably be attributed to displacement effects that can arise in drastic stepwise gradient runs. Even greater versatility can be achieved if three solvents are mixed, but the theory and the instrumentation are more complex. Several methods for selecting ternary mixtures will be described later. The advantage of a gradient run is similar to that achieved using programmed temperature in GC in that it allows one to adjust the retention factors of the analytes to get the best separations. The big disadvantage is the

198

0 I

LIQUID CHROMATOGRAPHY IN COLUMNS

.05 "

'

.10 .I5

Want stren@h, 'G silica)

.20

25

.30

.35

I

,

,

I

010 30 sol00

d %

0

.40

p

I

*

I t

.55

.60

I

I

.65

.70

.75

in iml

5 112 3 5 t

.50

45

10

I

1

30

5

I l

I

I

50100

I

1

I

1%ACNiniRCl

2

1

10

5

3

I

l

2030501

I

l

I

I

l

l

% MeOH in iml

1 0 60100

-

i.LLLd%El,@mMC

01235

10

1

1 1 1 1 1

0

05 I

I

40 5060100 IIII~I%ACNI~MC

1

I

2

3

I

I

5

I

10 I

203050100 I

IW IsoproWlchbnde MC Whylenechlonde E t g EthylEther ACN Acetanmile Md)H Methanol

0

I

5

1

I

%M&HI~MC

I

I

3050100 I I 1 1 %MeonInEt20

0 51020305070100 Hx Haane

I

I

111

6080

'6 ACN in €120

2

3

I

I

5 I

0 1 3 5 10

10 I

50 70 100

v

% MeOH in ACN

Figure 8.13. Solvent strength ( 6 ' ' ) of some mixed solvents (on silica). Reproduced from the Journd of Chromatographic Science by permission of Preston Publications, Inc.

time required to reequilibrate the LC column; it is not as simple as cooling a G C oven. Table 8.7 includes a recommended list of solvents to be used to recondition the column back to its original (polar) status. About 10 column volumes of each solvent are required. Many separations formerly done by LSC are now done with polar bonded phases like cyano and amino, and these are discussed in a subsequent section. The main exceptions are preparative LC and TLC. Bonded-Phase Chromatography

The original bonded phases were made for G C since it was anticipated that chemical bonding would prevent the stationary phase from bleeding at high temperatures. The technology for reacting organic moieties with solid surfaces was already established because of the silylation reactions used to deactivate GC solid supports (see Chapter 7). The original G C bonded phases are susceptible to decomposition from trace amounts of water or

8.2

CLASSIFICATION OF HPLC MODES

199

Table 8.6 Three Eluotropic Series for Silica Columns”

Eluotropic Series’ 8‘’

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60

1

2

3

Pentane 4.2% PrCl in pentane 10% PrCl in pentane 21% PrCl in pentane 4% Ether in pentane 11% Ether in pentane 23% Ether in pentane 56% Ether in pentane 2% Methanol in ether 4% Methanol in ether 8% Methanol in ether 20% Methanol in ether 50% Methanol in ether

Pentane 3% CH,CI, in pentane 7% CH,Cl,in pentane 14% CH,C12 in pentane 26% CH,CI, in pentane 50% CH,CI, in pentane 82% CH,CI, in pentane 3% Acetonitrile in benzene 11% Acetonitrile in benzene 31% Acetonitrile in benzene Acetonitrile

Pentane 4% Benzene in pentane 1 1% Benzene in pentane 26% Benzene in pentane 4% EtOAc in pentane 11% EtOAc in pentane 23% EtOAc in pentane 58% EtOAc in pentane

Source: After Snyder, Modern Practice of Liquid Chromatography, Kirkland (ed.), Wiley-Intcrscience, New York, 1971. “All percentages are by volume. hAbbreviations: PrCI, isopropyl chloride; EtOAc, ethyl acetate. Table 8.7 Solvents for Incremental Gradient Elution

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

Series 2 1. n-Heptane

n-Heptane Carbon tetrachloride Heptyl chloride Trichloroethane n-Butyl acetate n-Propyl acetate Ethyl acetate Methyl acetate Ethyl methyl ketone Acetone n-Propanol Isopropanol Ethanol Methanol Water

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

I

Carbon tetrachloride Chloroform Ethylene dichloride 2-Nitropropane Nitromethane Propyl acetate Methyl acetate Acetone Ethanol Methanol Water

Mixtures

Column-Recondilioning Solvents 1. 2. 3. 4. 5.

Ethanol Acetone Ethyl acetate Trichloroethane Heptane

Sources: From R. P. W. Scott and P. Kucera, J . Chromatogr. Sci., 1973, 11, 83, by permission of Preston Publications, h e . ; From R. P. W. Scott and P. Kucera, Anal. Chem. 1973, 45, 749.

200

LIQUID CHROMATOGRAPHY IN COLUMNS

Likely peak identity

1 Squalane

2

3

4 5 6 7 8 9 10 11 12 13 14

Anthracene Methyl stearate Benzophenone Chloroaniline Nitroaniline p-Dinitroknzene p-Nitrophenol Dihydrocholesterol Catechol Phenacetin Adenine Phenolphthalein EEDQ

21

15 16 17 18 19

Quinine Acetylsalicylic acid Benzoic acid t-BOC leucine t-BOC plycine 20 Alanine 21 Glucose 1

13 1

20

19

18 17

4A

Retention volume

Figure 8.14. LSC gradient elution chromatogram. Column: 50-crn X S-mm i.d. Bio-Sil A. Flow: 0.5 mL/min. Sample: 10 mg in 50 k L . Rcprinted with permission from R. P. W. Scott and P. Kucera, Anal. Chern. 1973, 45, 749. Copyright 1973, American Chemical Society.

oxygen" and are therefore no longer used. Newer GC bonded phases, used on OT columns, are cross-linked polymers as described in Chapter 7, and they are very stable, efficient, and popular. Similar, but somewhat different, bonded phases have been found to be excellent for HPLC. For a historical review of the first 15 years of development of bonded phases, see the study by Gilpin.'' At the beginning of this chapter it was noted that bonded phases have become the most popular stationary phases for HPLC, particularly those used in reversed-phase LC (RPLC). The following discussion of the chemical reactions involved in producing bonded phases will be limited to these nonpolar stationary phases. Unless otherwise indicated, the examples of BPC in this section will all be of the reversed-phase type. However, it must be remembered that not all BPC is RPLC, and not all RPs are bonded. Preparation of Bonded Phases Most bonded phases (BPs) are formed by reacting chlorosilanes with silica that has reactive silanol groups on its surface. Figure 8.15 depicts some probable functional groups on the silica surface, and Figure 8.16 shows some typical deactivating reactions (or silanizations). For reversed-phase supports, the R group, sometimes called a ligate, is an alkyl chain that in practice can be up to 18 carbons long. If the reacting silane has only one reactive chloro group, the reaction is simple and replaces the active hydrogen with a R group such as dimethyloctadecylsilyl.

8.2

201

CLASSIFICATION OF HPLC MODES

0/"

\

\0'/H I

/s"\

0

,H-

0

*...

,H

A\ A\

0

/O\

J\Ai\oA\

0

/H\ 0

/H

0

"0

0

0

0

0/H

,\ A\07\

0

/H

I./O

0

0

Figure 8.15. Representation of some possible functional groups on the surface of silica.

Bonded phases produced with this reaction are preferred for theoretical study because they are well-defined and regular; they are called rnonofunctional phases, or brush phases, since it is thought by some investigators that the C,, chains extend out from the surface like brushes. If the silane has more than one reactive chloro group, the reaction can be more complex, as shown. After hydrolysis with water, some cross-linking can occur, resulting in an undefined polymer on the silica surface. The reaction shown in Figure 8.16 is only one of several possibilities. The surfaces of the polymeric phases are more complex than the monomeric, and the surface of either type is affected by the solvents in contact with it.'3 Unreacted chloro groups can be hydrolyzed to hydroxyl groups and then be reacted again with trimethylchlorosilane to eliminate as many hydroxyl

0

\ / ,Si--O / 0 O \

H

+

/

/

/

-+

/H

0

CH, \ I ,Si-O-Si-R

I

CHS

0

0 \

/H

\ /H , Si-0

0

CI-Si-R

/

$-" ,S--0

,Si4

I

0

0

O\

CH,

+

\

c1

I I c1

CI-Si-R

-+

,?-0

/ 0

\

OH

I

OH

To\/R \ ,Si-0

0

\

2-0-Si-R

7Si-o-Si-R OH I 0/

0

\ Ho-Si -R

/

1

7 '

OH

Figure 8.16. Some silanizing reactions t o produce bonded phases for LC.

202

LIQUID CHROMATOGRAPHY IN COLUMNS

Table 8.8 Common Ligates for Bonded Supports Attached to Silica through a Silyl Linkage (see text)

RPLC

Octadecyl (C,,H,,, ODs) Octyl (C,H,,) Propyl (C,H,)

Phenyl Methoxy

NPLC Amino (C3H,NH2) Amino (C, H ,N(CH,),) Diamino (C,H,NCH, H,NH 2 ) Cyano (C,H,CN) Glycophase (C, H ,OCH ,CHCH,OH)

I

OH

groups as possible. This last silanization step is referred to as end cupping, which removes about half of the previously unreacted, residual silanol groups on the surface of the silica where the larger octadecylsilyl (ODS) reagent was not able to penetrate. Supports that have been end capped are usually found to have different selectivities from those that are not end capped, although end capping is only about 40% effective. Preventing analytes from contacting residual silanol groups on the SP is an important objective in the preparation of BP supports. These days, most BPs use a diisopropyl-octadecylsilyl group, which, because of its size, can stearically prevent contact with embedded SiOH groups. Additionally, purified silica gel (silica gel type B; see Chapter 3 ) can be used, or a polar spacer can be bonded to the silica. Bonded phases of other polarities can be prepared by substituting other groups for the octadecyl ligate. Some of the most common ones are listed in Table 8.8. The octyl ligate is also popular for RPLC, but very short chains show significant polarity. The ligates in the table for NPLC are polar in accordance with our definition. Of the nonpolar ligates, octadecyl or ODS is the most popular and most stable. In general, as the chain length is reduced from 18 carbons to 8 to 2, the degree of hydrophobicity decreases. Characterization of Bonded Phases A variety of measurements, such as total carbon analysis and BET (named after the inventors Brunauer, Emmett, and Teller) surface area, have been used to determine the effectiveness of the silylation reactions and the coverage of the silica surface. From these measurements the percentage of carbon on the surface can be calculated and used to designate the degree of surface coverage. Values ranging from 2 to 30% carbon have been reported. Pyrolysis GC, electron spectroscopy for chemical analysis (ESCA), fluorescence, nuclear magnetic resonance (NMR), Fourier transform infrared (FTIR), and Raman spectroscopy have all been used to characterize the surface of bonded layers. One study,14 by analyzing the decomposition

8.2

CLASSIFICATION OF HPLC MODES

203

Table 8.9 Analysis of Commercial ODS Phases

Amount of Liqate (g/g silica)

Manufacturer (particle size)

Type"

ODS

End Capping"

Du Pont (12 p m ) LiChrosorb (10 pm)

M D

6.19 5.50

ND ND

Beckman Ultrasphere IP Adsorbo-sphere HS (7 p m )

M M M M

Nucleosil (10 p m ) Alltech ( 5 pm)

Vydak 201-HS Versapack (10 p m ) Techsil (5 pm)

T T

T

4.15 8.5 1 6.71 3.84

3.24 8.91 2.01 2.05 4.62 11.82 2.38

Source: Adapted from S. D. Fazio et al., A w l . Chern. 19x5, 57, 1559, with permission. Copyright 1985, American Chemical Society. "Typc: M = monochloro; D = dichloro; T = trichloro. hTrimethyl.

products produced by the reaction of bonded reversed phases with HF, was able to determine the type of reaction (monofunctional or polyfunctional), the extent of end capping, and the distribution of alkyl group chain lengths. Some of the results of the study are summarized in Table 8.9. Such results help explain the chromatographic differences between bonded phases manufactured by different companies. It is generally agreed that there are two categories of adsorption sites on bonded chromatographic silica, one that forms strong attractions and the other weaker ones. This is important because this mixed-mode sorption is believed to be a major cause of tailing. The weak site probably corresponds to the bonded sites and the strong one to unbonded (free) silanols. Unbonded silica also shows two levels of sites; the stronger ones are the ones silanized in the bonding process. Ludes et al." in their investigation of desorption rates of C,-bonded silica found rates similar to that of other workers, indicating that both fused silica and silica gel have chemically similar adsorption sites. The other reason for the characterization studies is to find explanations for the retention mechanisms by which SPs work, and ultimately to devise a rational process for making and choosing BP columns. It is generally agreed that the mechanism of sorption in BPs can be either of two types, adsorption or absorption, called partition by most chromatographers (see discussion in Chapter 2). The so-called soluophobic theory of RPLC was elaborated originally by Horvath and Melander." Their model assumes that the nonpolar bonded phase acts more like a solid than a liquid and attracts analytes primarily by adsorption. The binding of an analyte to the surface reduces the

204

LIQUID CHROMATOGRAPHY IN COLUMNS

surface area of the analyte exposed to the mobile phase, and it can be considered to be sorbed partially because of this solvent effect; that is, the analyte is sorbed because it is solvophobic. Sorption increases as the surface tension of the mobile phase increases. Lockel’ concluded that bonded phases acted more like modified solids than thin liquid films, but that they were sufficiently different from solids to require a different theoretical treatment. He too would suggest that adsorption rather than absorption is the more likely mechanism of interaction. By contrast, Sentell and DorseyIX have data supporting the alternative mechanism, absorption. A study of taxanes separated on a variety of reversed-phase systems by Dolfinger and Lockeiq has led them to the following conclusion about the nature of the stationary phase, for example, a C , , bonded material, in a MP of aqueous methanol. They believe that the surface of the SP contains an excess of sorbed methanol, so the effective SP is a mixture of C,, and methanol, not just C II(. Therefore, the hydrocarbon part of an analyte can interact with the SP through nondispersive forces with the C (adsorption), while any polar functional groups on the analyte can absorb on the methanol. This duality is what gives RPLC its breadth of applications beyond those that would be expected from a system based on the simple polar-nonpolar descriptions given earlier in this chapter. While it is not practical to try to summarize all of this work in a book such as this, the early instrumental investigations of the bonded phase are included in the reviews by Gilpin.”,’” A thorough review of the retention process in RPLC was published in the Journal of Chromatography A in 1993.21 Twenty-eight review studies are included, covering theory (thermodynamics), SP structure, and empirical correlations and predictions. As would be expected, differences of opinion exist between the various research groups, and even this comprehensive report is not the final word on reversed phases. A short summary of retention mechanisms of BPLC was published the following year,’2 and Scott has compiled information on silica gel and bonded phases into a single volume.” Snyder and his colleagues are concluding a series of seven studies24-’(’ in which they have described the development of an empirical model for characterizing reversed-phase column selectivity. They have chosen five column-dependent parameters: hydrophobicity, steric selectivity, hydrogen bond activity, hydrogen bond basicity, and cation exchange/ion interaction behavior. They have examined 154 different columns including type A and type B silicas, and those with end capping and embedded polar groups. They conclude that their approach has resulted in a valid and reasonably complete de5cription of RPLC column selectivity. Raman studies reported by Pemberton et al.”-74 include some interesting molecular pictures based on their results. Figure 8.17, taken from their third

8.2

TFCl8SF/MeOH

((1)

CLASSIFICATION OF HPLC MODES

205

TFC18SF/l -BuOH

(b)

Figure 8.17. Molecular pictures showing the intcractions between ( m ) methanol and ( h ) hutanol and a C18 stationary bonded phase. Reprinted with permission from reference 33. Copyright 2003 American Chemical Society.

’’

study on the effects of self-associating solvents, dramatically shows the different associations of two solvents, methanol and 1-butanol with a trifunctional C , , SP. The primary interaction for methanol is adsorption, whereas butanol shows a significant absorption as well. In order for analytes to penetrate into the bonded phase, the mobile phase used must “wet” the surface of the bonded phase. An SP that is thoroughly covered with ODS is quite hydrophobic and may not be wetted by a totally aqueous mobile phase, thus preventing the analytes from contacting the stationary phase. As shown in Figure 8.17, monomeric ODS groups normally stand up like a “brush,” but in pure water they collapse and fold into a collapsed structure. Nevertheless, some work has been reported with pure water as the MP,”.” but usually an organic modifier such as methanol or acetonitrile is added to the MP to achieve wetting of the SP, or the SP contains embedded polar groups. Solvents used for MPs should always be of HPLC grade. HPLC grade means that the solvent has been purified (by ion exchange and RPLC), UV absorbers have been removed, and it has been filtered through a 0.45-pm filter. The requirement for high-quality solvents extends to water, of course, and special high-purity grades are available commercially, as are systems for

206

LIQUID CHROMATOGRAPHY IN COLUMNS

in-house preparation of purified water. Separations are often better with an acetonitrile mixture (mainly because of an absence of H-bonding problems as described earlier), but methanol is less toxic, cheaper, and more easily disposed of. If the sample contains weak acids or bases, buffers in the millimolar range are added for stability and improved peak shapes The expected order of elution from an RP column would be from polar (hydrophilic) to less polar (hydrophobic). Further guidelines for choosing a mobile-phase mixture are given later in this chapter. Remember, for gradient elution in RPLC, as the mobile phase gets stronger, it gets less polar; hence, the proportion of the organic solvent is increased during the run. Since the ligates are chemically bonded to the surface of the silica and will not wash off, it is no longer necessary to choose a mobile phase with a polarity opposite to that of the bonded phase. That is to say, a reversed-phase support like ODS could be used with a relatively nonpolar organic mobile phase, thus establishing a system that does not fit into either categoryreversed phase or normal phase. Some workers have given such a system the name nonaqueous reuersed phase, or NARP. Advantages and Disadvantages of Bonded Phases Bonded phases are stable, as evidenced by their popularity. Compared to LSC, they come to equilibrium much faster, they do not show irreversible sorption and produce less tailing, they can be used with a wide variety of solvents and trace amounts of water are not likely to be critical, and they come in a range of polarities. The most important disadvantage is that many commonly used silica-based bonded phases must be used within the pH range of 2-7.5, although type B silica is reported to be useful between 2 and 10. Also, some care must be taken to prevent the polar functional groups such as amine and nitrile from reacting with sample components and/or being oxidized. Amino columns are known for irreversible adsorption of keto groups in many cases. The pH limitation has stimulated the production of bonded phases with the C,, ligates attached to an organic polymer rather than to silica. These supports, which have the C,, group on the phenyl ring of polystyrenedivinylbenzene, can be used in the pH range from 0 to 14. The polymer is highly cross-linked to give it the rigidity needed for high-pressure operation. It does not seem to exhibit the interfering T interactions common to the polymer alone. In general, the polymer-based materials produce separations similar to those on silica-based supports, but they are somewhat less efficient with lower plate numbers and poorer resolution. Another disadvantage of RPLC is the fact that it has been found virtually impossible to produce two identical ODs-bonded reversed-phase columns. This reality makes it difficult to transfer a method to another lab or even to replace a worn-out column with an equivalent one-even from the same

8.2

CLASSIFICATION OF HPLC MODES

207

manufacturer. Regulated laboratories using standard methods may find it impossible to get reproducibility when using new columns, although columnto-column reproducibility is much less of a problem than it used to be. The problem is lessened by adequate robustness testing in method development, using multiple columns to ensure that the method is usable across columns or even manufacturers. However, a lot of attention has been paid to characterizing column packings and comparing columns that should be equivalent. The development of standard operation procedures for the manufacture of bonded ODS packings has been r e ~ o r t e d . ~The ’ result has been the production of a standard reference column (designated BCR-722), which has been accepted by the Institute for Reference Materials and Measurements (IRMM) of the European Commission. Equivalent Bonded Columns Since no two columns can be considered identical, the best alternative is to find two that are equivalent, that is, two that can be used to produce nearly the same separations. The column parameters that must be used to match columns are particle size, pore size, surface area, carbon load, type of bonding (monomeric or polymeric), ligand density, and end capping. A complete chapter on selecting an equivalent column based on these characteristics has been written by Hartwick38 from the perspective of a pharmaceutical chromatographer. He concludes that choosing an equivalent column is one of the hardest jobs a chromatographer has. In search of a simpler parameter, Ying and Dorsey3’ discovered that the phase volume ratio ( P I is the single most important one in determining retentivity. They recommend it over the carbon load value often provided by column packing manufacturers or the ligand density, which must be calculated. Another helpful publication on this topic is distributed by Mac Mod.4” In it, 60 commercial columns are compared on the basis of hydrophobicity, polarity, column efficiency, peak shape (for basic compounds), silanol activity, and metal activity. Ion Interaction Chromatography (IIC) This discussion is limited to the most common type of IIC-reversed-phase mode using bonded alkyl ligates (such as C,, and C,) as the stationary phase. Its main advantage over regular reversed-phase LC is that it facilitates the analysis of samples that contain both ions and molecular species. In regular reversed-phase operation, ionic species are not retained much if at all; pH is used to control the retention factors by controlling the degree of ionization of weak acid/base analytes. If a sample contains analytes that vary widely in p K values, only a few of them can be separated at a given pH, and

208

LIQUID CHROMATOGRAPHY IN COLUMNS

the others (as ions) will elute unseparated close t o the dead volume (nonretained) peak. Several groups of workers suggested the addition of counterions to the mobile phase to form neutral ion pairs with the analyte ions, thus causing them to be attracted to the stationary phase, retained, and separated. Schill and his group were working in ion pair extractions and adapted their procedures to chromatography,4' which became known as extraction chromatography. Knox" and Laird4' used the detergent cetyltrimethylammonium bromide in the mobile phase and dubbed their technique soup chromutography. O t h e r workers entered the field, which became confused by the diversity of names by which it was called and by the debate that arose over the mechanism of action. In summary, two mechanisms have been proposed to explain the results. O n e assumes that the ion pairs are formed in the mobile phase and behave as nonionic moities similar to other polar molecules in RPLC. Thus, the technique is often referred to as ion pair chromatography (IPC). T h e other mechanism rests on the belief that the counterions selectively sorb in the stationary phase and attract and retard the analyte ions by a n ion exchange mechanism; this mechanism requires that the ion pair reagent have a hydrophobic end that would be attracted to the alkyl chain o n the bonded phase and a n ionic site on the other end. Cantwell and c o - ~ o r k e r s ~ ' ~ ~ ~ proposed a double-layer model for these systems and conclude that either ion exchange o r adsorption can be the dominant process. Probably both mechanisms are partially correct, and the predominant mechanism may depend on the operating conditions. In any case the following discussion will not attempt to distinguish between the mechanisms o r justify either one; further theoretical discussion can be found in the study by Cecchi et. aLJS T h e counterions commonly used in IIC a r e listed in Table 8.10. Waters Associates introduced a line of reagents called PIC reagents, and these Table 8.10 Some IIC (Ion-Pair) Reagents

For Basic Samples

For Acidic Samples

Methanesulfonic acid Tetrabutylammonium hydroxide Pentanesulfonate: Na salt (PIC reagent B-5) Tetraethylammonium hydroxide Hexanesulfonate: Na salt (PIC reagent B-6) Tetrabutylammonium phosphate (PIC reagent A) Heptanesulfonate: Na salt (PIC reagent B-7) Hexadecyltrimethylammonium bromide Octanesulfonate: Na salt (PIC reagent B-8) Trihexylamine 2-Naphthalenesulfonate: Na salt Triheptylamine Dodecylsulfate: Na salt Trioctylamine Dioctylsulfosuccinate: Na salt Citrate Picrate Perchlorate

209

CLASSIFICATION OF HPLC MODES

8.2

PACKING:

pBONDAPAK/C,, PICTMSEPARATION

COLUMN:

4 mrnx30 cm

SOLVENT:

Methanol/Water (50/501 PIC Reagent 8-7 1 2 3 4 5 6

Maleic Acid Phenylephrine Phenylpropanolamine Phenacetin Naphazoline Pyrilamine

2

6

I

I F

LY z 7

~

U

t

0

I

5 Time (min)

I

10

Figure 8.18. Separation of a mixture of antihistamines and decongestants by ion pair chromatography. Courtesy of Millipore Corp., Waters Chromatography Division.

210

LIQUID CHROMATOGRAPHY IN COLUMNS

B.

A.

4

2

I

0

f

I

20

TIME (min)

40

1

56

1

v 1

2

3

4

-0

Figure 8.19. Effect of adding ion pair reagent in an RPLC separation of anesthetics. Samples: (1) benzocaine (peak 2 in chromatogram h, ( 2 ) lidocainc (peak 1 in chromatogram h),

( 3 ) tetracaine; (4)etidocaine. Column: Radial Pak C,8. Mobile phase: 60% acetonitrile in water; pH 3; 1%#triethylamine added in chromatogram h. Reprinted from the Joiinzul of Chromatogruphic Science by permission of Preston Publications, Inc.

names are included in the table. In using the IIC reagents, the charge on the reagent must be opposite to the charge on the analytes. The amines that are listed are not ionic, but they are used in acidic situations, where the protonated (cationic) forms exist. Increasing the concentration of IIC reagent will increase analyte partition ratios up to a point and then level out. Typical concentrations are 0.005-0.05 M, although higher concentrations are sometimes required. As IIC is currently practiced, a bonded reversed-phase separation is usually the first system tried. If some ions cannot be retained, an IIC reagent is added. A typical separation is shown in Figure 8.18. Alternatively, the use of an ion-pairing reagent can improve a chromatogram by decreasing the retention times and improving the peak shapes, as shown by Bidlingmeyer4' (Fig. 8.19). In general, IIC is a useful technique as an alternative to ion exchange chromatography (IEC) for the separation of ions or mixtures of ions and molecules. It does have some disadvantages: the ionic solutions are often corrosive and result in short column life, some of them also absorb in the UV and limit the use of the UV detector, and the older silica-based supports are limited to pH values below about 7.5. In addition, the mobile phases should not be left standing overnight. They should either be replaced by water or

8.2

CLASSIFICATION OF HPLC MODES

21 1

pumped continuously at a slow rate. Further details about IIC can be found in the review and book by Hearn.47 Suggestions for column care have been given by Rat~el.~’. 4y Hydrophobic Interaction Chromatography Another type of bonded phase is intended to be used with biopolymers. In addition to the larger pore size that is required ( 2 30 nm), these materials are intended to be used under milder conditions that will not denature the samples. When only water is used as the mobile phase, without any organic modifier, proteins will not be denatured and their hydrophobic characteristics will enable them to be separated on a nonpolar bonded phase. Many nonpolar bonded phases are too hydrophobic to be used with water alone as discussed earlier, so new bonded phases have been made to accommodate these requirements. This type of chromatography is called hydrophobic interaction chromatography (HIC).”’ Supports are prepared by bonding C,,, C,, C,, or phenyl ligates on Sepharose or on a silica that has been previously bonded with a polyamide coating. Since the hydrophobic interaction between bioanalytes and the stationary phase is increased at high ionic strengths, gradient elutions can be achieved with a decreasing ionic strength gradient (the opposite of the effect in ion exchange, as we will see). For a brief summary, see the study by Cooke et aL5’ Further discussion of HIC, including applications, can be found in Chapter 15. Internal Surface Reversed-Phase Supports In 1985, Hagestam and Pinkerton” published a report on a new type of stationary phase they had synthesized. They called it an internal surface reuer~sed-phase(ISRP) support, and it is also known as a Pinkerton column. The patented idea has been licensed exclusively to Regis Chemical Company.

Serum proteins (cannot enter small pores)

Non-adsorptive surface ISrequired

Internal Surface Reversed Phase Stationary phase found only inside pores, where drugs separate

Figure 8.20. Representation o f internal surface reversed-phase (ISRP) type support. Courtesy of Regis Chemical.

212

LIQUID CHROMATOGRAPHY IN COLUMNS

The concept behind the invention is shown in Figure 8.20. It operates with two mechanisms-size exclusion and reversed-phase bonded sorption. The purpose is to facilitate the injection of plasma samples without prior clean up to remove proteins that normally clog a reversed-phase column. The pore size of surfaces containing the stationary phase is small enough that the large protein molecules cannot enter and the outer surfaces to which they are exposed do not retain them at all, so they are eluted off the column quickly. The smaller analyte molecules penetrate the pores where they experience interactions with the bonded phase and are selectively retained and chromatographed. Monolithic Bonded Phases Monolithic silica columns, introduced in Chapter 3 (including references), are offered commercially by Merck KGaA under the name Chromolith. They represent a new option for achieving high-resolution separations,53 and are unlike conventional methodology that recommends the use of smaller and smaller particle sizes, with the resulting drawback of increasing column pressure. The monolithic columns contain two sizes of pores, macro (2 p m ) and meso (13 nm), somewhat like the ISRP columns just described. They have low pressure drops and can be used at fast flow rates. The pores are interconnected (reticulated), so all surfaces are available to small molecules (up to 13 nm diameter). Monolithic columns are expected to replace packed microparticle columns for some analyses because of their high flow rates at reasonable pressures. Both the silica and the organic polymer types are available with bonded phases, including the popular C,, used for RPLC. For example, Figure 8.21,

1500

-2 ._

Q r/)

1000 1250

Ju"

4

**

4

**

1 0

System

4

2

4 6 Flow rate (rnL/min)

8

10

Figure 8.21. Backpressure as a function of flow rate for an HPLC system, with and without a Chromolith column. Reproduced from theJournal qf Chrornutogruphic Science by permission of Preston Publications, A Division of Preston Industries, Inc.

8.2 CLASSIFICATION OF HPLC MODES

213

400

m A

u

200

0

L

0.0

I

0.2

I

0.4

I

0.6

1

0.8

I

1.o

I

1.2

Time (min)

Figure 8.22. Chromatogram of 21 seven-component mixture analyzed under fast conditions: Flow, 8.0 mL/min; gradient from 10/90 ACN/water to 30/70 in 0.1 min, held at 30/70 for 0.3 min, increased t o 05,’s and hcld for 0.6 min, and then rcturncd to starting conditions at 1.01 min. Pcak idcntifications: ( I ) benzamide. (2) N-methylhenzamide, ( 3 ) benzyl alcohol, (4) acetophenonc, (5) ethyl paraben, (6) propyl paraben, (7) biphenyl. Reproduced from the .lournu1 of Chrornutogruphic Scicnce by permission of Preston Publications, A Division of Preston Industries. Inc.

taken from a recent study by Smith and M ~ N a i rshows ~ ~ the pressure requirements compared to the HPLC system without a column at differing flow rates. A very fast analysis of a seven-component mixture in just over 1 min was achieved at 8 mL/min on a 5-cm Merck column (Fig. 8.22). The authors also conclude that the monolithic columns have selectivities comparable to particulate columns, making it possible to transfer methods between the two types, limited, of course, by the variations that currently exist between manufacturers. MacCalley ” also found the monolithic columns to be very efficient, with flat van Deemter plots, but basic compounds tailed significantly at pH 7. Some improvement was effected by using higher temperatures. Further evaluation of monolithic columns will doubtless occur in the near future. At present, their future looks bright. Zirconia-Bonded Phases The temperature limitations of silica packings led Carr and co-workers”, ” to examine zirconia as a possible alternative. They developed a polymer-induced colloidal aggregation method to produce porous zirconia microspheres with large (30-nm) pores that can be used at high temperatures. Uncoated zirconia can be used for NPLC and IEC, and when coated (cross-linked) with polybutadiene (PBD) an RPLC phase is produced. A review of the development of zirconia phasess8 lists eight different formulations, of which the PBD-ZrO, has become the most popular. A recent analysis and critical comparison of the RPLC and IEC contribu-

214

LIQUID CHROMATOGRAPHY IN COLUMNS

tions to retention on the PBD-ZrO, support and the traditional ODS silica support has provided much information on the nature of these materials.”1 ZirChrom Separations, Inc.‘” was formed in 1995 to market zirconia columns and Supelco6’ has also entered the field recently.’* Two reviews summarize the literature on ultrastable metal-oxide-based stationary phases for HPLC.6”64 Reference 62 also provides the latest information on HPLC column innovations as does Majors’ column in LC-GC” and his review of HPLC column packing technology.66 Normal-Phase Bonded Phases The nitrile and amine bonded phases are most popular for normal-phase operation. Nonpolar liquids such as the hydrocarbons and chlorocarbons are used as the mobile phases, although as noted before, even alcohols and aqueous mixtures can be used. The nitrile phase produces separations somewhat similar to silica gel in LSC, but it has the advantage of attaining equilibrium faster than the LSC systems. Normalphase HPLC is well suited for separating positional isomers. The need for methods complementary to the popular RPLC, and the problems with the use of silica as described in the section on LSC, provide a stimulus for continued searching for newer polar SPs. One such example is the proposal to use polymer beads made from glycidyl methacrylate-co-ethylene dimetha~rylate.~’ Other suggestions will undoubtedly follow. Ion Exchange Chromatography

The name ion exchange aptly describes the process used chromatographically to separate ions and some polar molecules. Because ion formation is favored in aqueous solutions, the mobile phase in IEC is aqueous, usually buffered to a particular pH. Ionic exchange sites are immobilized on the stationary phase, represented in the following equation as R-.

The B+ ion represents the analyte being separated from other cations, such as C + and D+. The cations will be separated from each other if the resin (stationary phase) has a selective affinity for the various analyte cations. The cation A + , which was the cation already on the resin, must not be too strongly held by the resin or exchange will not occur. It must be a component of the mobile phase, and its concentration in the mobile phase can be used to control the partition ratios of the analyte ions according to the equilibrium in Eq. (8.1). The previous example was an exchange of cations and is an example of cation exchange chromatography. If, on t h e other hand, the resin contains cationic sites, it is capable of exchanging anions, and the process is called

4.

E

0

50

025-

75

Leucine

I

I

I

I

I

A

Histidine

I

I

I

Lysine

I

I

Arginine

I A Effluent cc.

Figure 8.23. Classic ion exchange separation of amino acids using step-gradients on a 100-cm Dowex 50 column. Reprinted with permission from reference 68. Copyright 1951, American Society of Biological Chemists, Inc.

pi

100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 - 525 550 575 600- -625 . ~ 650 675 pH 3 41.375" pH 9.2,25" pH 11.0,25" pH 4.25,50" PH 4 25.75" 6 7,250PH 8 3025"

Threonine

216

LIQUID CHROMATOGRAPHY IN COLUMNS

anion exchange chromatography. In both cases there may be some adsorption of analytes on the resin itself, thus complicating the mechanism, but these secondary effects will be ignored in this discussion. Classical IEC Ion exchange resins have been available for a long time as rather large beads used primarily at low pressures for rather crude separations. As early as 1951 Moore and Stein”x published their classic work showing an ion exchange separation of amino acids. One of their separations is shown in Figure 8.23; it is a gradient analysis, and it requires a long analysis time. Instruments designed specifically for such analyses were called amino acid analyzers; they were probably the first LC instruments, although they were not generally recognized as such. Today a renewed interest in IEC has been generated by developments in other HPLC techniques. However, the old resins and classical procedures are still used, so this section will be used to present the fundamentals of IEC and old methodology. The next section will describe the major advances resulting in high-performance IEC. In the preceding discussion, it was established that the following items were important in IEC:

1. The selectivity of the resin for various ions 2. The particular ionic form (counterion) of the resin at the start of the analysis 3. The concentration of this counterion in the mobile phase 4. The pH of the mobile phase These items will be considered for the classical ion exchange resins. Resins The stationary phase or resin commonly used in IEC is the styrene-divinylbenzene copolymer, which was shown in Figure 3.3. Divinylbenzene provides the cross-linking, and its percentage is usually specified in the resin specifications. Functional groups are put on the phenyl rings to provide the ionic sites; the most common ones are listed in Table 8.11, which includes both cation and anion types. The labels “strongly” and “weakly” refer to the acid/base strengths of the functional groups (acids for the cation resins and bases for the anion resins). The weak exchangers can only be used 8 for the anion over limited pH ranges: 2 6 for the cation resins and I resins. The number of ionic sites put on the resin will determine the capacity of the resin; typical values are 2-5 mequiv/g of resin. Sample sizes must be small enough not to exceed these limits. The degree of cross-linking is usually between 4 and 16%. Highly crosslinked resins are harder, more brittle, less permeable, less susceptible to

8.2

CLASSIFICATION OF HPLC MODES

217

Table 8.1 1 Common Types of Ion Exchange Resins

Cation

Functional group

Anion

Strongly Acidic

Weakly Acidic

-SOI-H+

-CO,-H+

Strongly Basic

Weakly Basic

CH,

R

I

-N+ -CH,CI-

I

R

CH,

Trade name Dowex Duolite Amerlite Permut it

sow

c-20 IR-120 Q-100

I I

-N+ -HCI

-

1

CC-3 IRC-50 Q-210

A-I01 IRA-400

3 A-2 IR-45 S-300

s-100

swelling and changing of bed volume, and more selective. A compromise around 8% is usually required to get sufficient permeability without excessive variations in bed volume with changing MP conditions. Many other substrates have been used for IEC in addition to the polystyrenes. Some are natural products such as cellulose, often cross-linked for added rigidity, and others are manufactured, such as polydextran gels. Most of them can only be used at low pressures, like the polystyrenes just discussed. Operating Conditions Table 8.12 shows typical selectivities of a cation resin as a function of cross-linking. For the monovalent cations given, Li is the least strongly sorbed, and the other values are calculated relative to it.

Table 8.12

Relative Partition Coefficients for Some Cations

Cross-linking Cation

4%

8%I

10%

Li H Na NH, K Rb cs Ag TI

1.00 1.30 1.49 1.75 2.09 2.22 2.37 4.00 5.20

1.00 1.26 1.88 2.22 2.63 2.89 2.91 7.36 9.66

1.00 1.45 2.23 3.07 4.15 4.19 4.1.5 19.4 22.2

Source: Mallinckrodt Chemical Co.

218

LIQUID CHROMATOGRAPHY IN COLUMNS

Elution order is controlled by (1) ionic charge, (2) hydrated radius (especially for cations), and (3) polarizability. For a typical resin, the order of increasing bonding to the resin among divalent cations is UO, < Mg < Zn < Co < Cu < Cd < Ni < Ca < Sr < Pb < Ba For anions on an anion exchange resin, a typical order is

F- < O H - < acetate < formate < C1- < SCN

< Br- < chromate < nitrate < 1- < oxalate < sulfate < citrate These resins are not selective enough to permit the separation of ions that are close to one another in the listing. For example, Li, Na, and K cannot be separated. However, a common cation separation, originating from the Manhattan Project, that has become popular6’ uses the formation of coordination complexes to convert cations to anions via their chloro complexes:

FeCIi- is formed in 12 M HC1, and the resulting anion is strongly held on an anion exchange column. Several divalent cations that form these chloro complexes to lesser degrees can be separated from iron and from each other: Ni, Mn, Co, Cu, and Zn. The sample is made up in 12 M HCI to convert the metals to the chloro complexes. Ni(I1) does not form an anionic complex and can be washed off the column while the other metals are retained. Using a stepwise gradient elution with decreasing concentrations of HCI, the other chloro complexes are destroyed, one by one, reverting to cations that are not retained and are washed off the column. This separation depends on the chloro complex equilibria as well as on chromatographic equilibria. Perhaps it should not be classed as a chromatographic separation because it does not meet the usual chromatographic requirement that the species put on the column (the sample) is the same one that is eluted. In this case, anions are put on the column but cations are eluted. But let us return to conventional IEC and consider the order of anion selectivities. It can be seen that an anion resin in the citrate form would be useless-citrate is the anion most strongly held (at a neutral pH), and it would not be appreciably displaced by any other anion in the list. To convert a citrate or sulfate resin to another form requires extensive washing with a mobile phase highly concentrated in the other ion in order to effect the displacement equilibrium. It is even difficult to convert completely a chloride resin to its hydroxide form. Thus, in purchasing a resin, it is important to

8.2

219

CLASSIFICATION OF HPLC MODES

(b)

3 1

2

0

5

3

10

15

I

I

5 10 Retention time (Minutes)

0

I

15

I

20

I

25

Figure 8.24. Effect of ion exchange counterion on separation speed. Column: Zipax SCX. ( u ) 0.1 M K H 2 P 0 , , p H 4.4; ( h ) 0.1 M NaHzPO,, pH 4.4. Reprinted from R. C. Williams, D. R. Baker, and J. A. Schmit, J . Qzromutogr-. Sci. 1973, I / , 619 by pcrmission of Preston

Publications, Inc.

note the counterion with which it is sold. Commonly anion resins are sold in the chloride form and cation resins in the hydrogen form. Similarly, the counterion in the mobile phase affects the retention factors. Figure 8.24 shows an example of a separation of organic acids at p H 4.4 in which the only variable was the cation counterion -K+ in Figure 8 . 2 4 ~and N a + in Figure 8.24b. Because K + is more strongly held than Na', and it competes with the analytes for ionic sites on the resin, the overall retention times in Figure 8 . 2 4 ~are shorter. A secondary effect in this case is the alteration in the elution order. As with other HPLC processes, the slow diffusion in the mobile liquid phase is a source of zone broadening. Diffusion inside an ion exchange resin in the presence of fairly high ionic strengths is especially slow. Traditionally IEC must be run very slowly to allow time for exchange to occur. T h e rates are increased at increased temperature, but classic separations are usually run at ambient temperature. Where an instrument is used and temperature can be controlled, the range of 40-60°C is usually recommended. When organic acids o r bases are separated by IEC, the p H can be used t o affect their degree of ionization and hence their retention factors. This is most effective within t- 1-2 p H units of their p K values.

220

LIQUID CHROMATOGRAPHY IN COLUMNS

Gradient elution is also helpful; for acids, the pH can be increased during a run, hastening the elution of the stronger acids. Alternatively or in addition, the ionic strength can be increased to promote faster elution. Such a gradient system was used in the analysis of amino acids mentioned earlier (Fig. 8.23). Modern IEC Improved stationary phases7" similar to those developed for the other types of HPLC have led to improved separations by IEC. The old resins described above were followed by pellicular resins that were much more efficient and incompressible but had lower capacities. As is the case in the other forms of HPLC, they have been largely replaced by small nonporous particles, silica and polymeric, that have the ionic groups directly on the particle or attached to a ligate or polymer on the particle surface. Staby and his co-workers have published a series of three studies in which they compare and evaluate strong anion and strong cation ion exchange resin^.^' Their studies also provide useful references to the current literature on available resins. These developments were spurred by new resins, produced by Dow Chemical in 1975 and licensed to a new company, Dionex, that was founded to exploit them for inorganic analy~is.'~ Dow also sold them patent rights for a microconductivity detector and a counterion suppressor column, a second column used to remove counterions in the mobile phase, which otherwise would have a high conductivity. Dowex used the term ion chromatography (ZC) to distinguish it from other ion exchange (IE) techniques. Although the name ion chromatography has continued to be associated with this form of IEC, and the manufacturers who make this equipment contend that it is a special technique, the separations performed by IC are ion exchange separations.

-''

/on Chromatography Aside from the new ion exchangers developed for IC, the major difference between it and other IE techniques is the use of a patented second column to reduce the ionic strength in the mobile phase so that the analyte ions can be detected by a patented microconductivity detector.75 Originally, this was achieved by using, as part of the mobile phase, a counterion that could be converted to a molecular form by an ion exchange process in the second, suppressor, column.76 For example, in an anion analysis using carbonate as the counterion, nonionic carbonic acid is formed when the original cations (K' or Na+) are exchanged for H C in the suppressor column, which is a strong cation exchanger. The acidity of the mobile phase helps prevent the ionization of the carbonic acid without

8.2 CLASSIFICATION OF HPLC MODES

Strong Base Anion Exchanger in the COP F O “the separator“

-~

b+C0,2-

i

221

NaN03

“the suppressor“

Figure 8.25. Representation o f eluent suppression scheme for ion chromatographic separation of nitrate and sulfate. Reprinted with permission from H. Small, Anal. Chem. 1983, 55, 23SA. Copyright 1983, American Chemical Society.

222

LIQUID CHROMATOGRAPHY IN COLUMNS

Eluant Flowpath in MicmMembrane Suppressors

Membrane

\

Membrane

Ibl Figure 8.26. Alternative forms of suppressor column for ion chromatography: ( u ) Hollow fiber and ( h ) membrane sandwich. Reprinted by permission of Dionex Corporation, Sunnyvale, CA.

affecting the analytes that are conjugate bases of strong acids-ions such as nitrate, chloride, and bromide. Figure 8.25 shows a diagrammatic representation of the separation of nitrate and sulfate ions. The use of a suppressor column is not without problems. Eventually, the resin becomes exhausted and needs to be regenerated, which is inconvenient. Also, the slightly ionized carbonic acid produces a small continuous baseline conductance signal, so that the vacancy peak from the sample injection is detected and produces a negative peak that can interfere with other analytes

8.2

CLASSIFICATION OF HPLC MODES

223

eluting at that time. The use of a second column also results in some zone broadening, which decreases the overall efficiency of the analysis, but not seriously. The resin capacity problem was initially resolved using continuous, flowing suppressor streams that contact the chromatographic stream through a porous membrane. Hollow-fiber ion exchange tubing (Du Pont’s Nafion) packed with plastic beads to decrease the internal volume and zone spreading has been used.77 Figure 8.26 shows the action of two designs: ( a > a hollow fiber and ( b ) a sandwich. The vacancy peak (carbonate dip) has been eliminated as a problem by using faster, more efficient IE columns78 so that the vacancy peak occurs early in the chromatogram. The desire to use a universal detector like a conductance detector is the reason for the use of a suppressor column. Alternative detection systems have been used without a suppressor; some ions absorb in the UV and can be used with non-UV-absorbing counterions in the mobile phase; some work has

I

Li

+

I”’

a-

I No*- m.’-

-

0

2 4 6 Minutes

8

Figure 8.27. Separation of common anions by ion chromatography. Reprinted by pennission of Dionex Corporation, Sunnyvale, CA.

Figure 8.28. Separation of monovalent cations by ion chromatography. Reprinted by permission of Dionex Corporation, Sunnyvale, CA.

224

LIQUID CHROMATOGRAPHY IN COLUMNS

0

4

8

12

16

tR (min.)

Figure 8.29. Separation of ninc transition mctal cations hy ion chromatography. Kcprintcd by permission o f Dionex Corporation, Sunnyvalc, CA.

been done with the universal refractive index detector; and, finally, indirect detection (discussed more fully later in this chapter) has been applied in which the mobile phase contains a UV absorber such as the phthalate ion, and the analyte peaks are detected as negative peaks by a UV detector.'" Most important, the improved technology has produced excellent separations of ions that could not be directly separated by IEC, as was noted earlier. Some inorganic separations are shown in Figures 8.27-8.29. Organic ions can also be separated, of course, as shown in Figure 8.30. More details about the development of IC can be found in the study by Haddad'". In an effort to further improve the methodology, Dionex" has produced a new generation of instruments based on the concept called "reagent-free IC." The name originates from the fact that hydroxide ion is generated within the instrument by electrolysis of water. Solutions of hydroxide ion are used as the mobile phases, including gradients of increasing KOH concentration.** Ion suppression is achieved by reaction of the hydroxide with acid from the electrolysis to produce water, and conductivity can be used as the detector.

8.2

Column: Eluent 1:

Eluent 2: Eluent3: Flow Rate.

DeBctlon: injecSon Voi.:

0

lonPacs AS1 1 D.I. Water 5.0 rnM Sodium hydrodde 100 mM Sodium hydroxide 2.0 Wrnin. S w d oonductivity, AutoSuppressionmode 10 &L

I

131

I

5

CLASSIFICATION OF HPLC MODES

Peaks 1 IsqacQylmefAylphosphate 2. QUlMte 3. Fiuonde 4. Autlate

5. Pmponate 6. Fwmate

Peaks:

5 mgit 3 1 5 5 5

78. 9. 10.

Memyisulfonate Pyruvate Chlorite Valerate 11. Manochlomacstate 12. Bmmate 13. C h M e

5 mglL

5

5 5

5 5 2

5 5

?4

3 3 3

'i 91

Minutes

225

Cahate Maionate Maieate Sulfate Oxalate Kebrnaionate 26. Tungstate 21. Phthalate 28. Phosphate 29. Chmmate 30. Cilrate 31. Tmsrbalyiate 32. iM&W 33 cis-Aconilate 34. bans.Acawtate MA ;%

20. 21. 22 23. 24 25.

I

lo

i

15

5

5 5

5 5 5 10 10 10 10 10

10 10

10

Figure 8.30. Ion exchange separation of anions. Reprinted by permission o f Dionex Corporation, Sunnyvale, CA.

''

Everything is self-contained, and no external reagents are required.8', Figure 8.30 shows an anion separation using this new system. Because the system is capable of high-speed analysis, an RP monolithic column has been successfully tested with it for separation of inorganic anionsx5 Other recent applications have been reported from the annual IE symposium" in September, 2002. Ligand Exchange Chromatography A cation exchange resin can also be used to separate analytes that form coordination complexes with the metal attached to the resin. For example, a cation resin in the Cu(I1) or Zn(I1) form can be used to separate amino acids by using ammonia as a competing ligand. Some typical examples are given in references 87 and 88. Ligand exchange methods include separations achieved using ion pair conditions, and they can also be used for chiral separations (Chapter 15). Summary The discussion of modern IEC has included a number of peripheral techniques, but the basic ion exchange concept is the same as that described earlier. The major variables of pH, temperature, and counterion are used to advantage. In addition, an organic solvent is sometimes added to the mobile phase since it can have a significant effect on the separation. The newer, fast, highly efficient resins have smaller capacities than the older polystyrene resins. Values range from about 1 to 500 pequiv/g of

226

LIQUID CHROMATOGRAPHY IN COLUMNS

resin, which is about one-tenth to one-thousandth of the old resins. From the examples presented it can be seen that IEC has developed into a widely versatile and useful complement to the other forms of HPLC. For example, the separation of amino acids is vastly improved over the original work just 30 years earlier (Fig. 8.23). IEC is also advantageous for anion determinations, which, especially for mixtures, can be superior to colorimetric methods or ion-selective electrode measurements. The determination of cations is also easily handled by IEC, but atomic absorption spectroscopy (AAS) and inductively coupled plasma (ICP) methods are more common. Some nonchromatographic applications of ion exchangers include: Deionization of water Deionization of raw sugar solutions Softening of water Preparation of standard solutions, for example, carbonate-free NaOH from resin in OH- form and standard solution of Na2S0, Rmoval of interferences, for example, PO:- and F- AAS or ion-selective electrodes (ISE); trace metals Determination of total equivalents of dissolved salt using resin in H + form Dissolution of precipitates, for example, BaSO, with resin in H f form Concentration of trace elements for AAS, voltammetry, and so forth Acid catalysis, for example, 2-methylpropene + MeOH -+ MTBE Microstandards Size Exclusion Chromatography

Size exclusion chromatography (SEC) is different from all the other HPLC methods because separations are based on a physical sieving process, not on chemical attractions and interactions. The mobile phase merely acts as a solvent and eluent for the sample. Size exclusion chromatography has its origin in two separate groups of workers and was called by two separate names. Beginning in 1959 the name gel filtration chromatography came into use by those using dextran gels to separate biochemical polymers by using aqueous mobile phases. A few years later, the name gel pemeation chromatography was coined by polymer chemists separating synthetic organic polymers on polystyrene gels by using nonaqueous mobile phases. Other names were also used, but now it seems preferable to use the term size exclusion to include both types, and the use of older terms should be minimized. Size exclusion chromatography is a method for separating molecules based on their hydrodynamic volume in solution by using stationary phases with

8.2

n5

n4

I

I

I

CLASSIFICATION OF HPLC MODES

I

1

-

'

Total exclusion \

\ \

.-g In

227

\

n3

-

L

(D

3

P W U

.2

n'

-

Figure 8.31. Illustration of permeation and exclusion limits in SEC.

pore sizes capable of discriminating among the analytes in a sample. Its primary application has been to the analysis of polymers, and it is commonly used to determine molecular weight distributions of polymers. Theory T o a first approximation, the theory is very simple. A sample containing analytes with a variety of sizes is introduced to the SEC column, which is packed with a stationary phase that is inert except that it has carefully controlled pore sizes. The largest molecules will be unable to penetrate the pores, will be excluded, and will elute at the time for nonretained materials (total exclusion). Slightly smaller molecules will be able to penetrate some of the pores and will be slightly retained. Smaller and smaller analytes will penetrate an increasing fraction of the pore volume in the packed bed, be increasingly retained, and thus be separated from each other. Eventually an analyte size is reached that can penetrate all the pores, and all molecules of that size and smaller will elute together, representing the limit of the columns separating power (total permeation). This behavior is represented in Figure 8.31, which shows the total exclusion and total permeation limits. Note that the x axis is logarithmic and that the large molecules elute first, followed by the smaller ones. The basic chromatographic equation

v,< v, + KVS =

228

LIQUID CHROMATOGRAPHY IN COLUMNS

takes on new meaning for SEC. The volume of the stationary phase, V s , is really the volume of the pores of the stationary phase, and the partition coefficient K has a new meaning. It measures the extent to which a given analyte has penetrated the pores and has a value of zero for total exclusion and unity for total permeation. This restriction on the possible values of K severely limits the range of analytes that can be separated on one SEC column and verifies the prediction in Chapter 2 that SEC has a maximum peak capacity of about 20. This corresponds to a range in molecular weights of about 1.5 decades, as shown in Figure 8.31. In actuality, some adsorption of analytes on the stationary phase may occur, and partition coefficients greater than unity are observed. Also, the theory is much more complex than just presented. Bly" has classified the process into three mechanisms: steric exclusion, restricted diffusion, and thermodynamic considerations, and the process has been thoroughly studied. The rate equation is also different for SEC. For polymers, the longer retained peaks have smaller peak dispersivities, H , than early peaks, in direct contrast to normal HPLC expectations. In part this is due to the fact that the smaller molecules that elute last have higher diffusion coefficients and therefore less mass transfer zone spreading. Stationary Phases The stationary phases used in SEC are solids ranging from synthetic polymers to glasses. The earliest ones were soft gels, which cannot be used at high pressures. Consequently, the development of new phases has paralleled somewhat the development of more rigid resins and gels in IEC, as noted earlier. Some phases and their trade names are given in Table 8.13. The rigid phases, preferred for fast analysis, are chosen based on their pore sizes (and range of applicability), their compatibility with the mobile phase to be used, and their inertness. Since most are made from silica or glass, they are restricted to pH values less than about 7.5. If adsorption is a problem, the phases can be deactivated; the silylation reactions discussed earlier are popular. Rigid packings with smaller pores have been developed to permit the separation of smaller molecules with molecular weights in the range of 100-1000. They are fast and capable of baseline resolution of small homologs. Since any single-column packing has a limited range, it is often necessary to perform SEC with multiple columns in series. This procedure extends the molecular weight range over which separations can be effected. Many scientists use a set of three mixed-bed columns, each of which has a blend of two or more pore sizes."" As a consequence, the time of analysis is increased, so

8.2

CLASSIFICATION OF HPLC MODES

229

Table 8.13 CommerciallyAvailable Packings for SEC

Principal Trade Name" (Supplier)

Chemical Type/Particle Size

Flow Limits: Typical Linear Flow Rateh Efficiency Pressure Drop' (plates/ft)'

Polymeric Rigid TSK PW (Toyo Soda) 'Spherogel PW (Altex) *Bio-Gcl Tnk (Bio-Rad) Shodex Ionpa (Showa Dcnko) "Shodex Aqueous (Perkin-Elmer) 'Spheron (Lachema) (Koch-Light) (Knaucr KG)

Cross-linked hydroxylated polyether 11-15 p m

I50 cm/h

3500

200 psi Sulfonated cross-linked polystyrene 10 p m

170 cm/h

Cross-linked poly (2(2-hydroxyethyl-methacrylate) 20-40 p m

100 cm/h

5000

500 psi 800 500 psi

Po!ymeric, compre.ssihle

-Toyopearl (Toyo Soda) ' Fractogel

(E. Merck)' Sephadex (Pharmacia)

'Sephacryl (Phmnacia)'

Ultrogel AcA (LKB)

'

Bio-Gel A (Bio-Rad)

Cross-linked hydroxylated polyether 30-50 p m

25 cm/h

800

Cross-linked dextran 40- 120 p m (dry) Sephadex, sequentially reacted with ally1 chloride and mcthylenc bisacrylamide

10 c m / h

200

20 cm/h

500

Agarose, postpolymerized with methylene bisacrylamide 25-55 p m (dry)

25 cm/h

14 psi 500

Agarose 40-80 p m ; 80- 150 p m ; 150-3 (dry)

10 cm/h

2 psi 200 2 psi

Siliceous, Utzderiuutized

' Li Chrospher (E. Merck) Du Pont SEC Zorbax S E (9- 11 p m ) Zorbax PSM (5-7 p m ) 'Spherosil (Rhone-Progil)

Porous silica 10 p m Porous silica 5 - 1 1 p m Porous silica 40- 110 p m

I800 em/ h 1000 psi 4000 cm/h

7000 2500

-

300

-

'CPG (Elcctronucleonics)

Porous glass 55, 100, 150 p m (all *30%)

300 cm/h

250

~

7003

230

LIQUID CHROMATOGRAPHY IN COLUMNS

Table 8.13 (Continued)

Principal Trade Name" (Supplier)

Chemical Type/Particle Size

Flow Limits: Linear Flow Rateb Pressure Drop'

Typical Efficicncy (plates/ft)"

400 cm/h

5000

2000 psi 400 cm/h

3500

2000 psi 400 cm/h

3500

2000 psi 400 cm/ h

3500

2000 psi 300 cm/h

5000

800 psi 500 cm/h

3500

1200 psi 300 cm/h

200

Siliceous, Derivatized ' LiChrosorh Diol

(E. Merck)

1,2-dihydroxypropyl silica 5 wm

SynChropak (Synchrom)

1,2-dihydroxypropyI silica 10 p m

Aquapore (Brownlee)

I ,2-dihydroxypropyl silica

Aquachrom (Chromatix)

1,2-dihydroxypropyI silica 10 p n

TSK SW (Toyo Soda)

Derivatized silica treatment unknown 8-12 g

pBondagel (Waters)

Derivatized silica treatment unknown 8-12 Fm ("Polyether phase")

'GlycophaseG (Pierce) *Glycerol G P C (Electronucleonics)

1,2-dihydroxypropyl glass 55, 100 ( *20) p n

10 p m

-

Source: Reprinted from P. L. Dubin, A m . Lab., 1983, 15(1), 62. Copyright 1983 by International Scientific Communications. '*Commercial name. 'Substrate sold as bulk packing; optimal data from literature. 'Limitation on linear flow rate as imposed by loss of efficiency, with or without gel compression. ' Pressure drop corresponding to flow rate in footnote h , or maximum encountered in normal operation (30-cm column). "Obtained with low MW nonionic solute.

attempts are being made to find ways to minimize the total column length and still retain accurate and precise molecular weights. Some suggestions can be found in reference 90. Mobile Phases There are few requirements the mobile liquid must meet: it should (1) be a good solvent for the sample; (2) have a low viscosity, such as any HPLC solvent; (3) wet the stationary phase and, if possible, help deactivate it; and (4)be compatible with the detector being used. For soft gels, it must also cause the gel to swell. In general, polar liquids such as water, THF, and chloroform are used. Applications Size exclusion chromatography is the easiest LC technique to understand and apply. Choosing the stationary and mobile phases is relatively easy, and the separation achieved on the selected system is predictable.

8.2

CLASSIFICATION OF HPLC MODES

231

4

1

1125

1400

Retention volume (mi)4

Figure 8.32. SEC separation of oligomers in a commercial surfactant (Triton X-45). From K. M. Bombaugh, W. A. Dark, and R. F. Levangie, Separ. Sci. 1968, 3, 375; courtesy of Marcel Dekker.

It is used mainly to separate large molecules, synthetic polymers, and biopolymers with molecular weights between 2000 and 2,000,000. Figure 8.32 shows a typical separation of a surfactant in which the individual oligomers have been resolved. Cazes” listed 100 types of compounds that have been studied by SEC. To determine molecular weights, a calibration curve like that in Figure 8.33 is first prepared. This is a semilog plot like the general one shown earlier (Fig. 8.31) and is linear from C, to C,,, including both hydrocarbons and t riglycerides.

F

I

1

I

I

1

Standards Symbol Hydrocarbons o Triglycerides 0

-

c z m

.-

f

-

-$ 0 L

P

loo

---

--

or

1

1100

I

1200

I

I

1300 1400 Retention volume ’ V (mf)

I 1500

Figure 8.33. Typical SEC molecular weight calibration curve. From K. M. Bombaugh, W. A. Dark, and R. F. Levangie, Separ. Sci. 1968, 3, 375; courtesy of Marcel Dekker.

232

LIQUID CHROMATOGRAPHY IN COLUMNS

An unresolved SEC curve, composed of only one large single peak, can be used to determine the molecular weight distribution of a polymer. Higher resolution is neither necessary nor desirable because the desired information can be obtained from the peak shape (cumulative area vs. retention), as described by CazesY2 The SEC method has become an important part of polymer analysis, and further details can be found in polymer books as well as monographs on SEC."' Analytical Chemistry has reviewed SEC separately from its biennial HPLC reviews through l99ELy4 These reviews are a good source of the latest information on the topic. Another recent reviewo5 covers the use of SEC to study branching in polymers and also reports data using a multiangle lightscattering detector. Determination of Molecular Weight and Molecular Weight Distribution of Polymers Using SEC The molecular weight-log retention volume plot can be used to determine the molecular weight and molecular weight distribution of a polymer sample if monodisperse calibration standards are available for that polymer."' Such standards are available for polystyrene and polyethylene oxide. For a polydigperse sample of one of these polymers, the SEC chromatogram generated using a column of the appropriate pore size and a differential refractive index (RI) detector will show a broad peak. The broadening is not the result of chromatographic spreading, but rather is an envelope reflecting the large number o f unresolved oligomers in the sample. Commercially available SEC software divides the peak into a number (say 100) of slices. Each slice has a retention volume, VR,; and if the slice is sufficiently narrow, it can be assumed to be effectively monodisperse and to have a molecular weight M , which can be read off the calibration plot, to give molecular weight of the polymer as a function of V,. The software also determines the area A , of each slice. Since the RI peak area of each slice is proportional to the weight of oligomer in the slice, the software determines the weight of polymer as a function of VR, and integrates this to produce the cumulative weight percent. Finally, the software makes the cross-plot of cumulative weight percent vs. log M , which is the molecular weight distribution. These calculations can, of course, also be done manually. By definition, the number-average molecular weight, MN, is given by

where n , is the number of moles of polymer of molecular weight M,. MN is conventionally determined experimentally from measurement of colligative properties or end-group analysis. Polymer properties are more dependent on

8.2

CLASSIFICATION OF HPLC MODES

233

the larger oligomers than the smaller ones in a polydisperse polymer, so a weight-average molecular weight is also useful. By definition, the weightaverage molecular weight, MW, is given by M W = cw,MI = ~ n , M 1 2 / M ~ In ,

(8.5)

where w,is the weight fraction of oligomers of molecular weight MI. M W is conventionally obtained from light-scattering methods. The molecular weight averages can be obtained from SEC-determined M I and A , data, since and

so and

(8.9) and

(8.10) The dispersity (or polydispersity) of a polymer, D , is defined as

D

= MW/MN

(8.11)

For D < 1.1, a polymer can be considered effectively monodisperse. In SEC, retention is governed by the size of a polymer molecule in solution. This is related to the polymer molecule's molecular weight, M , but a more fundamental parameter to use is the hydrodynamic volume, V,, of the polymer in the mobile-phase solution; Vh actually refers to a hypothetical solvated sphere that is subject to the same frictional forces in solution as would a real polymer molecule; Vl, is related to M by Vh

=

M[771

(8.12)

where [773 is the intrinsic viscosity of the solution,

This is used to construct the universal calibration plot, which is the plot of log Vh vs. VK and is based on polymer size in solution. T h e plot is generated

234

LIQUID CHROMATOGRAPHY IN COLUMNS

using a polymer for which monodisperse standards of known intrinsic viscosity are available, for example, polystyrenes. For other polymers for which no such standards are available, an RI detector can be supplemented with an online flow viscometer detector to generate the data needed to use the universal calibration plot to determine average molecular weights and molecular weight distributions. 8.3

INSTRUMENTATION

A schematic diagram of the apparatus used for HPLC was shown in Figure 8.3. Not all of the parts shown are required for a simple apparatus, but commercial instruments often include additional options such as four-solvent capability, autosamplers, column thermostats, degassers, and guard columns. Updates about HPLC instruments can be found in the biennial reviews in Anulyticul Chemistiy""; these reviews are separate from those on LC theory and methodologyy7. Pumping Systems and the Mobile Phase Before discussing the actual pumps, let us consider the mobile phase and its requirements. The mobile liquid must be pure (free of particulates and UV absorbers), and special chromatographic grades are available for most common solvents, including water. Some solvents are routinely stabilized with small amounts of chemicals that can significantly alter their solvent properties for HPLC use and may absorb in the UV. For example, chloroform is often stabilized with ethanol or pentene. Particulate matter finds its way even into pure solvents, requiring the use of a filter in the intake line from the solvent reservoir. Precolumn filters or guard columns are also recommended. These disposable items help protect expensive columns from impurities. If they are at all flammable, HPLC liquids are often fire hazards. Others are toxic. Care must be exercised in their use; consult standard safety manuals and HPLC solvent guides such as Sadek.y9 Like most ethers, T H F and dioxane tend to form explosive peroxides. They should not be stored for long periods of time, and care must be exercised if they are purified. Before attempting any distillation, consult a reliable reference such as the ones by Riddick and Bunger"' and Sadek". Oxygen is soluble in many solvents, particularly water, and it should be removed prior to use because it tends to be released at the low-pressure end of the apparatus, the detector, where it can cause noise or even a complete loss of signal. Many commercial instruments include the degassing capability in the solvent reservoir for convenience. Some possibilities are the following,

8.3

INSTRUMENTATION

235

used individually or in combination: vacuum, vacuum through a gas-permeable fluoropolymer membrane, heating, sonication, and sparging with helium. The last is probably the easiest to include in an instrument, but the most efficient is the combination of sonication under vacuum.

Pumps The high-pressure pump required for HPLC is a critical part of the apparatus, so a lot of attention has been directed toward the development of good pumps. A versatile pump should be able to deliver flows from a fraction of a milliliter up to at least 10 mL/min with a precision of about 1% up to a maximum pressure around 35 MPa. Two types can be distinguished: constant flow and constant pressure. Constant tlow is most desirable for the concentration-type detectors and for reproducible retention times, but these pumps often produce noise in the detector because of the cyclic nature of their action. Remember that a blockage in the LC system will cause a constant

To Column

I

1 Solvent Reservoir

Figure 8.34. Schematic of dual-headed reciprocating piston pump, parallel design. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chemi,ytty in a G M P Enuironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons. Inc.

236

LIQUID CHROMATOGRAPHY IN COLUMNS

flow pump to produce increasingly high pressures, necessitating protection with a pressure relief valve. Most HPLC pumps are based upon the reciprocating piston design. Although single-headed reciprocating piston pumps are still available, the majority of pumps employ one of two versions of a dual-headed reciprocating piston design (Fig. 8.34). The first style of dual-headed design places the two pump heads in parallel and operate 180" out of phase, allowing one pump

Figure 8.35. Schematic of dual-headed reciprocating piston pump, series design. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chernisfiy in a GMP Enuironrnent, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wilcy & Sons, Inc.

8.3

INSTRUMENTATION

237

head to deliver high-pressure MP while the other pump head refills with MP. When the pump outputs are combined, their individual pulsations compensate for each other, in effect damping the variations in flow and producing a constant rate. Another popular pump design is the series dual reciprocating piston pump (Fig. 8.35). In this design the pump heads are assembled in series, rather than parallel. The first piston is typically twice the volume of the second piston, allowing the first piston to deliver high-pressure flow onto the column while simultaneously refilling the second piston. When the first piston is empty, the second piston assumes the function of delivering MP while the first piston refills. Once again this design results in a pump delivering nearly pulse-free flow. Small pulsations are difficult to eliminate due to the differences in pressure between the flowing stream (high pressure) and the contents of the pump head at the beginning of its outward stroke. Even though liquids are only slightly compressible, this pressure difference is enough to cause a slight change in pressure each time the pump changes from piston 1 to piston 2. These pulsations are further minimized through the use of mechanical or electronic pulse dampers. They are noticeable only under the highest sensitivity settings of the HPLC system. Most pumping systems can accommodate three or four solvents. A proportional metering valve can be used to produce accurate mixtures of two, three, or all four. This capability is necessary for gradient elution. Gradient Devices Gradient devices can be divided into two types: high pressure and low pressure (gravity flow). In a typical low-pressure system, various proportions of up to three solvents are mixed before they enter the high-pressure pump. The proportions are regulated by valves that are in turn controlled electronically with microprocessors. Some means must be provided to mix the liquids thoroughly before they enter the pump, usually a low-volume mixing chamber. High-pressure systems provide one high-pressure pump for each solvent; the output from each pump is regulated by an electronic control, and the combined streams are mixed and sent to the sample valve. Although this system is more costly because of the need for additional pumps, it produces more precise mixtures at the extremes of a gradient. In all gradient devices, the tubing volumes must be kept small so the gradient formed is quickly delivered to the column. Prior degassing is required because the mixing process often results in the liberation of heat, which can result in gas evolution since the solubility of dissolved gases is lower at the higher temperature. The heating caused by the enthalpy of

238

LIQUID CHROMATOGRAPHY IN COLUMNS

t

20

40

60

% Modifier

ao

100

Figure 8.36. Variation in viscosity of water mixtures with increasing amounts of modifiers. Reprinted with permission from S. van der Waal, Chromutogruphiu 1985, 20, 274.

mixing in itself can be a problem unless temperature control is provided. Another concern in designing gradient systems is the change in viscosity that accompanies mixed solvents. Figure 8.36 shows the nonlinear behavior of mixtures of some common solvents with water. Constant-flow pumping systems are required to handle these gradients. The development of a RP-HPLC gradient system is covered later in this chapter. Sampling

Not included in Figure 8.3 are a precolumn and an in-line filter before the sample valve. These are not required, but the filter is highly recommended.

8.3

(a)

INSTRUMENTATION

239

(6)

Figure 8.37. Six-port sampling valve for HPLC. ( a ) Fill sample loop. ( h ) Inject sample. Reprinted with permission from Scot in Modern Practice of Liquid Chromatography, J . Kirkland (ed), John Wiley & Sons. Copyright 1971; this material is used by permission of John Wiley & Sons. Inc.

Otherwise, the column will act as an effective particle filter and, as it gets clogged, the operating pressure will increase. Clogged filters can be replaced and will protect expensive columns. A precolumn may be necessary to precondition the solvent before the sample is introduced. When the stationary phase is silica or silica based, a precolumn of silica is often used to saturate the mobile phase with silica and thus decrease the dissolution of silica at the top of the column. This is especially recommended when working near the upper pH limit of silica, which is between pH 7 and 8. High-pressure valves are most often used for sample introduction in HPLC. A typical valve, shown in Figure 8.37 has six ports and two positions, one for loading and one for sample introduction to the column. Interchangeable loops with different calibrated volumes are attached to the valve and are filled with a syringe. They range in volume from 10 p L to several hundred microliters. For smaller volumes, special valves have been designed with internal loops and low-volume connections. Some valves have large loops that are intended to be only partially filed and then back-flushed for use. Many details need to be considered for good sampling, and the study by Dolan""' is highly recommended. The sample should be dissolved in the initial mobile phase if possible. If it is not very soluble, other solvents can be used, but they should be as similar to the mobile phase as possible, and equal to or less than the MP in eluent strength. Use of a sample solvent that is too strong can produce split peaks,

240

LIQUID CHROMATOGRAPHY IN COLUMNS

especially for early eluting analytes. Also, blanks should be run to see if there are any effects of this new solvent on the chromatogram. Another alternative is to use larger injections of more dilute samples. The upper limit is about $ of VM, which would be about 100 p L for a conventional analytical column (4.6 mm X 250 mm). Columns

An introduction to column packings was begun in Chapter 3 and continued earlier in the respective sections of this chapter. This section deals with column sizes and other operational details. Column Dimensions The conventional columns we have been discussing have inside diameters of 4.6 mm, lengths of 3-25 cm, and are packed with porous microparticles, 3 or 5 or 10 p m in diameter. Slightly wider or narrower columns are not much different in efficiency, and 4.6 mm is convenient because it is the inner diameter of commercially available +-inch stainless steel. Since efficiency can be increased by improving the density of the packing, several commercial units use special means for compressing the column after it is packed. Other columns have been designed for use with soft gels that tend to swell and shrink during use. They have movable end pieces that can be adjusted to conform to the bed volume. Fast LC Some very short columns that are only 3-5 cm long and are packed with very small particles ( 3 p m ) have already been mentioned. Sometimes referred to as " 3 X 3" columns, they are less costly and give good separations with minimum consumption of mobile phase. The term fast or high-speed HPLC has been applied to their use at high flow rates of 3-4 mL/min. Figure 8.4 showed one such example. In general, separations that need 4000 or fewer plates can be separated with these columns in a minimum of time. Dead volumes in the flow path must be kept to a minimum, and special detectors may be needed; some of the practical aspects of fast HPLC have been described by Dong and Gant.""

Microbore Columns As the name implies, these column types have small inside diameters. The sizes vary, and some columns are packed while others are open tubes (OT). A summary of the major types is presented in Table 8.14. The most common commercial columns are 1 mm i.d. and are packed. In his publication on the theoretical aspects of packed and OT HPLC columns, K ~ o x " )concluded ~ that O T columns should be capable of producing the highest plate numbers (in excess of several hundred thousand), but

8.3 INSTRUMENTATION

Table 8.14

241

Column Characteristics

Conventional, packed Microbore, packed OT, packed" OT, not packed

i.d.

Length

Flow Rate

4.6 mm 0.2-1 mm 40-80 p m 15-50 p m

3-25 cm IO-IO'cm 1-100 m 1-100 m

1 mL/min 1-20 pL/min 0.5-2 pL/min < 1 pL/min

Packing DiameteI

5 Pm 5 Pm 10-30 p n

"Packing may be modified becausc columns are prepared by filling a glass capillary and then drawing it out, cxposing thc packing to high temperature and cmbcdding it in the glass wall.

they are currently limited by injection and detection capabilities. The same conclusion was reached by Reese and Scott."" An evaluation of slurry-packed OT columns (20-50 p m ) has been published."I4 The advantages of microcolumns are: (1) high efficiencies, (2) increased detectivity, primarily with mass flow detectors, ( 3 ) decreased solvent consumption and lower costs, (4) compatibility with online detectors like MS, and ( 5 ) the possibility of using exotic mobile phases or reagents and special new detectors. The separation shown in Figure 8.38 illustrates several of the

-

6

5

4 3 2 1 MINUTES

Figure 8.38. High-speed microbore separation of biogenic amines. Reprinted with permission from E. J. Caliguri, P. Capella, L. Bottari, and I. N. Mefford, A n d . Chern. 1985, 57, 2423. Copyright 1985, American Chemical Society.

242

LIQUID CHROMATOGRAPHY IN COLUMNS

desirable features just listed: the minimum plate height was only 16 p m ; regular k-inch ss tubing was used (1.2 mm id.) for columns that were slurry packed; the flow rate was 0.2 mL/min (optimum was 0.1 mL/min), which required a small amount of solvent; analysis time was less than 6 min; detection limits were about 1 pg; and a specially constructed amperometric detector was used.i05 The disadvantages are similar to those just discussed for fast columns: all "dead" volumes in tubing and connectors as well as the detector volume must be kept sufficiently small to prevent extracolumn zone broadening. It is also possible that special pumps, sample valves, and detectors may be required. The interest in microbore columns has resulted in the publication of three books"'h~1"8 and a pamphlet as well as several review articles.'"'.

'"

Other Column Features Column permeability is an important concept that was covered in Chapter 6. Three other topics, external zone broadening, guard columns, and temperature control, are covered in this section.

External Zone Broadening If attention is not paid to the extracolumn volumes in connecting tubes, injection devices, and detectors, the efficiency of a separation may be decreased by zone broadening in these regions. This is especially important for the small and microbore columns just described when they are used at low flow rates.Ii2 If we express the zone broadening in terms of the peak width, the equationt that governs the total effect is (8.15) where w represents the width of the peak at its base. The width of a peak due to the apparatus, wapp,can be determined by removing the column from the apparatus, joining together the injector and the detector, and running a one-component "sample." The width of the resulting peak can be measured, but fast detector and readout response times are required. A typical value should be 40 p L . To estimate the effect of this extracolumn broadening on the overall (total) peak width, let us assume that a good column will produce peaks with widths as narrow as 100 pL. Then, W&

= ( l o o ) ' + (40)'= 11,600

(8.16)

and

w,,~= 108 p L 'For such equations, variances are additive, not standard deviations.

(8.17)

8.3

INSTRUMENTATION

243

Thus, the extra broadening of 40 p L has increased the peak width by 8 p L or 8%, which is acceptable. Normal liquid chromatographs use small-bore &-inch tubing ( < 0.25 mm i.d.), special connectors, minimum tube lengths, and detectors with volumes around 8-12 pL. For the low-flow situations described earlier, the inner diameter of the tubing is more critical and special detectors may be required. Guard Columns To help protect analytical columns from degradation by dirty samples, it has become common to include a “guard” column between the injector and analytical column. These columns are short and are intended to be replaced frequently, after they become contaminated. Often they are dry packed with a pellicular support that is chemically similar to the one being used in the analytical column. Although they contribute a little zone broadening to the system, the effect is small and worth the convenience of protecting the main column. Temperature Control In most of the early instruments, the columns were exposed to ambient conditions, and no provision was provided for thermostating them. More recently it has become clear that there are advantages to controlling the temperature, and some analyses are markedly improved at elevated temperatures. For example, higher temperatures provide faster kinetics for better mass transfer, lower solvent viscosities for higher flow rates, and decreased adsorption, often the cause of peak tailing. Therefore, it is somewhat surprising that investigators have not reported consistent improvement in resolution for RPLC.”.’ Nevertheless, to increase flexibility, newer instruments provide this capability, and temperature control is highly recommended to control aberrant effects.” Thermostatic jackets are commercially available for retrofitting on older instruments. Some work has also been done with temperature programming, similar to the technique used in GC’14,but that has not become common. Investigations of high-temperature and high-speed HPLC have continued, facilitated by the use of zirconia columns that can tolerate the higher temperatures. They have produced some useful conclusions that should lead to improved performance at elevated temperatures. For example, Thompson and Carr“s have shown that narrow-bore (2.1 mm id.), highly retentive columns should be used and that the detector time constant (see Chapter 9) is the limiting factor in their optimization of short columns. They believe that “conventional wisdom” is at best incomplete and at worst incorrect. Stevenson,11hin his report on the HPLC 2002 meeting, includes a summary of the latest developments presented there. Another new approach to the use of high temperature has been to use two columns in series and to optimize their use through thermal tuning.”’ Clearly, this is an area that will continue to be explored.

244

LIQUID CHROMATOGRAPHY IN COLUMNS

Table 8.15

Common HPLC Detectors

Spectroscopic

Ultraviolet absorption (UV/Vis) Fixed-wavelength UV Variable-wavelength UV Photodiode array (PDA), also called diode array detector (DAD) Fluorescence Mass spectrometer (MS); see Chapter 10

Electrical

Electrochemical detector (ECD); amperometric Conductivity detector

Other

Refractive index (RI) Evaporative light-scattering detector (ELSD)

Detectors

Most HPLC detectors are based o n very common phenomena, and are, with few exceptions, modifications of existing technology and not based o n new principles. This was not the case with GC detectors, most of which were invented specifically for GC. Ultraviolet is the most popular detector; its use is limited mainly because it is not universal (see Chapter 9). T h e RI detector is universal, but it is not as sensitive o r as stable and cannot be used with gradient elution. It is used mainly in SEC. Mass spectroscopy is increasing in popularity and is the subject of Chapter 10, but it is expensive. T h e evaporative light-scattering detector (ELSD) is the newest o n e adapted for HPLC, but the search continues for a sensitive, universal L C detector to complement existing detectors. Table 8.15 lists the common detectors to be discussed here. T h e principles of operation of the common H P L C detectors will not be discussed in this chapter since they are thoroughly treated in books on instrumental methods and in other reference books."8 Detector classifications can be found in Chapter 9, along with a general discussion of the need to design flow cells that have small volumes and fast time constants. T h e geometry of the flow channel is also critical in achieving a maximum path length without undue sensitivity t o flow oscillations, turbulent flow, o r refractive index changes (in the case of UV). Ultraviolet Detectors

FixedWavelength UV Detector T h e simplest version of the U V detector is the fixed-wavelength detector. This detector employs a U V light source-typically a low-pressure mercury vapor lamp. T h e mercury lamp

8.3

INSTRUMENTATION

245

provides several distinct lines of UV radiation, with the 253.7-nm wavelength (commonly referred to as 254 nm) being the most intense. The radiation is then passed through a filter to remove the extraneous wavelengths. Variable-Wavelength UV Detector (VWD) Due to the fact that not all compounds absorb at 254 nm, variable-wavelength detectors are used to allow the option of choosing the wavelength. This is accomplished by adding a monochrometer to the detector design. The monochrometer starts with a continuum UV source, such as a deuterium lamp, producing a broad band of radiation from 190 to > 800 nm. The light is then bounced off a grating, which separates the light into a spectrum of its various wavelengths. The grating is placed on a movable platform, allowing the user to choose any wavelength from the spectrum. Photodiode Array Detector (PDA) The PDA takes the UV detector one step further than the VWD by allowing the user access to all of the wavelengths simultaneously "'. This is accomplished by starting with a continuum source as in the VWD and passing the entire spectrum of light through the detector cell. The light is then bounced off a grating as in the VWD. In this case, the grating does not move, and the single detector is replaced with a multiple array of individual detectors (photodiodes). These detectors are arranged on a single chip referred to as a photodiode array. A schematic is

Figure 8.39.Schematic of a photodiode array UV detector. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chemistry in a GMP Enuironment, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

246

LIQUID CHROMATOGRAPHY IN COLUMNS

shown in Figure 8.39, and further information, including a list of commercial instruments, has been published. 12(' An advantage of this detector arises from the inclusion of a personal computer (PC) to sort out signals, so that the user is given access to the entire UV spectrum all of the time. A variety of tasks can be performed, such as peak purity (comparing UV spectra at various points along the peak), compound confirmation (adding spectral information to the retention time), and reprocessing any single wavelength as a chromatogram. Special threedimensional (isometric) and two-dimensional contour plots can be drawn from stored data. Two wavelengths can be ratioed to each other to aid in the detection of unresolved analytes and to characterize analytes qualitatively. Overlapping peaks can be deconvoluted using various mathematical procedures. lndirect Detection One way to use a UV detector for analytes that do not absorb is to include a UV absorber as part of the mobile phase. When the non-UV-absorbing analytes enter the detector cell, they cause a decrease in the baseline absorbance, appear as negative peaks, and can be used for routine analysis. This effect and similar ones are sometimes responsible for unwanted, unexpected peaks in chromatograms, and then they are called ghost peaks or pseudopeaks. A recent review of this entire subject 12' calls them system peaks, and several other studies treat the same topic,122including one on the theory of the origin of such peaks in ion pair chromatography.'*' Refractive lndex (RI) Commercial LC refractometers are based on one of two designs, deflection or Fresnel. As with most bulk property detectors, both types require reference and sample cells. The RI detector requires good temperature control- k 0.001"C for maximum sensitivity of refractive index units. Normally the temperature is controlled to only +0.01", and less than maximum sensitivity is attained. Even this degree of regulation also requires a length of small-bore tubing on the inlet side to bring the eluent to detector temperature. As discussed earlier, the volume in this tubing must be minimal to prevent extracolumn peak broadening. When first turned on, it may take several hours for an RI detector to stablize. In general, the RI detector is not very sensitive, and it cannot be used in gradient elution. Both positive and negative peaks can appear, depending on their refractive indices relative to the mobile phase, and this is considered somewhat disadvantageous. However, it is universal in response, which accounts for its popularity, especially in SEC and in preparative LC.

8.3

INSTRUMENTATION

247

Fluorescence Detector Typical of fluorescence in general, the conventional fluorescence detector'24 used in HPLC is more selective than the UV detector and can be 100-1000 times more sensitive. It is this sensitivity that accounts for its popularity and its use following postcolumn reactors. The excitation sources used in LC instruments are as varied as those used in conventional fluorometers-from mercury vapor to xenon to quartz halogen to deuterium-but the most interesting designs have used lasers. Lasers can be focused on very small areas, even the end or a transparent cross section of a capillary column, yielding effective cell volumes as small as 1 nL. These laser-based detectors are called laser-induced fluorescence or LIF detectors. They give increased sensitivity (up to lo9 times greater than UV) and are very compatible with microbore columns.'25 Additional information and references on LIF detectors is included in Chapter 13. Since few compounds fluoresce, applications of this detector are limited. Electrochemical Detectors The only electrochemical detector (ECD) in current use is amperometric. However, some workers have used the term coulometric for detectors that operate at a high current efficiency, and others have used the term pofarogruphic when the electrode is mercury. The term LCEC is commonly used to represent LC with electrochemical detection. While electrochemical detectors are extremely popular and useful for some analyses, some analysts find them difficult to use due to electrode fouling and maintenance. However, since the detectors are used in aqueous solutions typical of the popular reversed-phase mode, their use continues to expand. They are selective and sensitive, and their very small cell volumes make them ideal for microbore columns. One common application is the determination of catacholamines at the low parts per billion (ppb) range. Evaporative Light-Scattering Detector The search for a universal HPLC detector has resulted in greater use of the evaporative light-scattering detector (ELSD), and at least six instruments are now commercially available.'2h The basic principle, as the name implies, is the evaporation of the rnobilephase solvent, leaving the analytes as solid residues that will scatter light in a suitable beam. Evaporation is preceded by nebulization as shown in Figure 8.40 Further details can be found in references 127-129. The ELSD is not really universal since many volatile analytes will be lost, and some that are only slightly volatile will be partially lost and cannot be quantitated. Hence its use is greater for higher molecular weight analytes. Detectivity and linearity can be problematic. Still, this detector is not spectroscopic, and it complements the popular UV detectors. It is a destructive detector and cannot aid in the identification of analytes, unlike the mass

248

LIQUID CHROMATOGRAPHY IN COLUMNS

Figure 8.40. Schematic of typical evaporative light-scattering detector. Reprintcd courtesy o f Alltech, Inc.

spectrometer.'2q A recent study reports a comparison of the ELSD with another specialty detector, the chemiluminescent nitrogen detector, for nonUV-absorbing compounds.'-"' Some troubleshooting tips and techniques have been published by Young and Dolan.'3' Other Detectors The conductivity detector is increasing in popularity because it is usually the detector used in ion chromatographs. The special problems associated with its use were discussed in the section on IC. Sub-ppb detection limits and six orders of magnitude linearity have been reported.13' Many other detectors have been used in HPLC, including atomic absorption, chemiluminescence, conductivity (including photoconductivity), density, dielectric constant, electron spin resonance (ESR), flame emission, plasma emission, NMR, optical activity, phosphorescence, photoacoustic, radioactivity, Raman, spectroelectrochemical, streaming current (electrokinetic), thermal energy analyzer (TEA), thermal lens calorimetry, and viscometry, plus mass spectrometry, which is discussed in Chapter 10.

8.4

REVERSED-PHASEMETHOD DEVELOPMENT AND OPTIMIZATION

249

Postcolumn Reactors The use of a reactor at the outlet of the column can be used to produce a species that can be measured by one of the standard detectors, such as UV/visible, fluorescence, or electrochemical. Probably the earliest example of the use of postcolumn reactions was in the determination of amino acids by colorimetry using ninhydrin as the reactant. See the section on derivatization in Chapter 15, as well as the study in Analytical Chemistry,"' or the book edited by K r ~ l l for ' ~ ~further details. REVERSED-PHASE METHOD DEVELOPMENT AND 0PTIMlZATl0N

8.4

This section will attempt to outline steps that can be taken to develop and troubleshoot reversed-phase separation methods. Considerations leading to the selection of HPLC as the method of choice are discussed in Chapter 16. As previously discussed, RPLC is usually amenable to semipolar and polar analytes and/or ions. Figure 8.41 summarizes the selection steps in choosing an HPLC system. The development and validation of methods was introduced in Chapter 1. The parameters that comprise validation (at least for the International Conference on Harmonisation, ICH) are listed in Appendix A. The choice of stationary phase would most likely be a reversed-phase material of 3-pm diameter and short column length. The criteria for selection of the mobile phase were developed starting in Chapter 4 and continued earlier in this chapter, where it was concluded that the solvents of choice are water, methanol (MeOH), acetonitrile (ACN), and THF. Following are some methods for selecting the best mixture of these solvents for RPLC. Although some of these methods have been largely replaced by computer programs, they are still useful in optimizing separations. In actual fact, many separation schemes are selected based on prior work in the literature or, for simple mixtures, by the application of the information presented thus far. Overlapping Resolution Mapping

In 1980 Glajch and c o - w ~ r k e r s applied '~~ the strategy developed by Snyder in conjunction with a mixture-design statistical technique. They arrived at a new method of data analysis that they called overlapping resolution mapping (ORM). The steps in the method are: (1) selection of the three ideal solvents la Snyder; ( 2 ) chromatographing the sample mixture up to 10 times using different mixtures of the three solvents chosen statistically; (3) preparation of resolution maps for each pair of analytes in the mixture; and (4)overlapping

250

LIQUID CHROMATOGRAPHY IN COLUMNS

HexareSolbble

Normal-Phase Adsorption ~

Normal-Phase Bonded OrganicSoluble __

Metbanol MethanoliH20 Solube

THF-Soluble MW < 2000

Nonionic

Reversed-Phase Bonded

Gel Permeation GPC

~

3

Peversed Phase

Reversed-Phase l o i PairtngiSuppression

Ionic I

lo? Exchange

Peptides Protew Estimate

M W of Sample

Reversed Phase

~

OrganicSoluble

Unknown MW Range

Gel Permeation Chromatography WC)

I MW < 2000

Gel Filtration Aqueous ~

GFSiSEC

-L

I

Ion Exciange

Pberornenex

Known MW Raige

r pH2-75 pH>75 CationExchange AnionExchange

OH 2-7.5 Reversed ?hase pH > 7.5 LlydroDhobic Interaction (PIC)

- AffivnityiB oaffinlty Figure 8.41. Some guidelines for selecting HPLC columns. Courtesy of Phenomenex, Inc. Copyright Phenornenex Inc.

8.4

REVERSED-PHASEMETHOD DEVELOPMENT AND OPTIMIZATION

251

these maps to obtain an ORM that designates the ideal mixture(s1 to be used for optimal separation. The three solvents chosen for RPLC were methanol from group 2, ACN from group 6, and T H F from group 3. These three solvents are not as widely separated from each other in the diagram as those for normal phase are, but they are sufficiently different to produce good separations. The (polar) carrier solvent used was water, and the stationary phase was a C, bonded phase. This procedure will be illustrated with a reversed-phase separation of nine substituted naphthalenes. First, the concentration of each of the three solvents in water must be optimized to produce reasonable retention times. As we have seen, the ideal situation exists when the partition ratios are in the range of about 1.5-5 (slightly higher and lower values can also be considered). For this example, the ideal concentrations were: Mobile phase A : 63% methanol in water Mobile phase B : 52% ACN in water Mobile phase C: 39% THF in water The incorporation of these 3 MPs into a statistical design is shown in Figure 8.42. The 10 mixtures shown include the 3 MPs (points 1, 2, and 3 at

A

Figure 8.42. Statistical design of mixtures of ternary mobile phases. Reprinted with permission from reference 135.

252

R

LIQUID CHROMATOGRAPHY IN COLUMNS

k=O

1

2

3

4

5

6

7

8

9

11

2.

Figure 8.43. Solvent selectivity data summary for cight mobile phases. Reprinted with pcrmission from reference 135.

the apices), .50/.50 mixtures of pairs of solvents (points 4, 5 and 6), a 33/33/33 mixture of all 3 (point 7), and 16/16/67 mixtures of all 3 (points 8, 9, and 10). The data for the first 7 runs in this study are shown in Figure 8.43, where it can be seen that some unexpected variations occur, but there are no reversals in peak elution order. Next, a computer-generated resolution map is made for each (adjacent) pair of peaks using the resolution values from the runs (only 7 in this example out of the possible 10). The contour map for peaks 8 and 9 is shown in Figure 8.44, which also gives the 7 resolution values found in the 7 runs. An arbitrary minimal resolution (e.g., 1.5) is chosen, and the portion of the map giving a resolution equal to or greater than that value is shown as clear space in Figure 8.4.5. Three pairs of peaks (4/5, 5 / 6 , and 7/81 are resolved by all MPs and can be disregarded. The other 5 maps are overlapped to see what areas of good resolution they have in common (Fig. 8.46). Any MP mixture falling in the clear region labeled A will give a resolution of at least 1.5 for all 8 pairs of peaks. The point marked X in the figure is an optimal mixture chosen by another method described in the same study."' If the elution order changes, additional maps must be prepared to consider the original order plus all the additional new pairs, but the ORM method can handle such a situation. For a complicated mixture, the proce-

8.4

REVERSED-PHASEMETHOD DEVELOPMENT AND OPTIMIZATION

253

Figure 8.44. Resolution map for peaks 8 and 9. Reprinted with permission from reference 135.

Figure 8.45. Simplified ( R , > 1.5) resolution maps for all eight pairs of peaks. Reprinted with permission from reference 135.

254

LIQUID CHROMATOGRAPHY IN COLUMNS

__

MeOH

Peak pairs

Pair

2-3 '

ACN Figure 8.46. Overlapping resolution map (ORM) combining the data in Figure 8.45. Reprinted with permission from reference 135.

dure may become too laborious or the individual components may not be available for preparing a standard test mixture. Perhaps some easy-to-separate components can be eliminated from the screening, and selective detectors can be used to establish the identity of eluting components to detect changes in elution order. Phase Selection Diagrams

A number of other methods have been used to optimize ternary solvent systems, many of them'36-'3x similar to the window diagram used in GC. Only one more will be described briefly here. The authors, Schoenmakers et a1.,l3' have prepared what they call phase selection diagrams for several reversed-phase LC separations. The solvents used for their mixtures are the same ones recommended by Snyder: methanol, ACN, and THF. First, they make a gradient run with one of the solvents (usually methanol) in water. From that run they choose the best isocratic conditions and run the sample with it. Then they choose a binary mixture of one of the other solvents in water, prepare a mixture that is isoeluotropic (same solvent strength, using Snyders solvent strength parameter), and rerun the sample. In

8.4

REVERSED-PHASEMETHOD DEVELOPMENT AND OPTIMIZATION

0

Volume fraction THF

-

255

0.32

Figure 8.47. Typical phase selection diagram. Reprinted with permission from reference 139.

one example, the first binary mixture was 50% methanol, 50% water, and the other mixture was 32% T H F and 68% water. The In k values for these two runs are plotted on the y axis of Figure 8.47. It can be seen that the elution order in the first case was 1 (and 2 unresolved), 3 , 4, 5, 6 and in the second was 1, 3, 2, 6, 5 , 4. The two points for each peak are joined by straight lines (dashed in the figure), and the phase selection diagram is drawn by calculation of an optimization criterion, R,, defined as:

n

n-1

nRs=

i= I

ki+ 1 - ki k i + l+ k , + 2

(8.18)

where n is the number of peaks in the chromatogram. Equation (8.18) can be derived from Eq. (5.18) in Chapter 5 , and it is plotted on an arbitrary scale on the right in the figure. It predicts that, for this example, the best separation will occur at its maximum value, which occurs at a (ternary) mixture of 10% methanol, 25% THF, and 65% water, as shown in the figure. In this case, the predicted separation does occur. If it does not, a new revised diagram is prepared using the data from the three runs, and a new optimum

256

LIQUID CHROMATOGRAPHY IN COLUMNS

mixture is predicted. Thus, with only a few trials, a satisfactory separation can be achieved. Buffer Effects

Although the procedures just described did not include any reference to pH, it is customary and often critical that pH be controlled with a buffer in the MP, especially if any of the analytes have acidic or basic functionalities. For RP, the pH is usually adjusted so that the analytes are in their molecular forms; that is, the pH selected should be at least two pH units away from the p K , of any analyte, if possible. That is not to say that a separation might not be improved by an adjustment of the pH that would cause some analytes to be partially ionized and therefore elute earlier than (and be separated from) the molecular analytes. Such an experiment is included in a training program described by Miller.'4" However, the retention factors of the ionized analytes will vary greatly with small changes in pH, making it difficult to develop a method that is reproducible and robust. This situation is complicated by the fact that pH as commonly measured with a glass electrode will not have meaning in mobile phases containing appreciable amounts of organic modifiers. The definition of pH is based on measurements with glass electrodes in aqueous solutions, and the significance of their pH readings depends on their calibration. More details and additional considerations of the influence of organic modifiers o n RP-HPLC systems have been nicely summarizcd by Sykora and co-workers.'" Table 8.16 lists some buffers commonly used in RP-HPLC. If the stationary phase is silica based, it must be remembered that the pH range will probably need to be restricted to the range from 2 to 8. The concentration of buffer is not critical and is usually 10-50 mM. Discussion of buffers and t h e optimization process has been thoroughly explained,'", '43 and new equations and developments continue to be p u b l i ~ h e d . ' ~ ~ Gradient Elution As explained earlier in this chapter, gradient elution is one alternative way to

run an HPLC analysis when an isocratic method is found to be inadequate. In a gradient run, the elution strength of the MP is made stronger during the process, which in the case of RP means that the percentage of organic modifier is increased during the run. In the simplest case, two solvents are prepared, called A and B, with B designating the stronger one. The analysis is begun with solvent A, and then increasing amounts of B are blended into A during the run. It is highly recommended that neither A nor B be pure solvents, but that at least some of the other solvent is included in each of the two, A and B, to facilitate mixing.

8.4

REVERSED-PHASEMETHOD DEVELOPMENTAND OPTIMIZATION

257

Table 8.16 Buffers Used in HPLC

Buffer

PK,

Buffer Range"

Trifluoracetic acid Phosphoric acid/mono or di-K phosphate

>> 2

1.5-2.5 < 3.1 6.2-8.2 11.3-13.3

Citric acid/tri-K citrate Formic acid/K-formate Acetic acid/K-acetate Mono-/di-K carbonate Bis-tris propane'. HCI/ Bis-tris propane Trisd. HCl/tris Ammonium chloride/ammonia 1-Methylpiperiden . HCI/ 1-Methylpiperdine Triethlyamine . HC1/ triethylamine

2.1 7.2 12.3 3.1 4.7 5.4 3.8 4.8 6.4 10.3 6.8 9.0 8.3 9.2

10.1 11.0

2.1-6.4 2.8-4.8 3.8-5.8 5.4-7.4' 9.3-11.3 5.8-7.8 8.0- 10.0 7.3-9.3 8.2-0.2

9.1-11.1 10.0-1 2.0

uv Cutoff !' 210 nm (0.1%)

< 200 nm (0.1 %)

< 200 nm (10 mM) 230 nm (10 mM)

210 nm 210 nm < 200 nm < 200nm 215 nm 225 nm 205 nm 200 nm

(10 mM) (10 mM) (10 mM) (10 mM) (10 rnM)

(10 mM) (10 mM) (10 mM)

215 nm (10 mM)

< 200 nm (10 mM)

"pH range allowed with this huffer (conservative estimate) 'JAhsorhance < from ref. 7. 'Requires addition of an acid (e.g., acetic or phosphoric). "Tris (hydroxymethyl) aminomethane. "1.3-his [Tris(hydroxymethly) methylaminoll propane. Source: Reference 142, p. 299.

Gradients can be linear or nonlinear, and the gradient system should be capable of producing a variety of gradient shapes. Linear gradients are common, but the Snyder solvent parameter produced will not likely be linear, as was shown earlier in Figure 8.8. To get a linear polarity gradient with the systems shown in Figure 8.8 would require a concave gradient. Choosing a gradient is often done by trial and error using an initial rate of 2% per minute. The optimization process has been well described.'4' -I4' Often, method development is begun with a gradient run, much as a programmed temperature run was described in Chapter 7 for G C method development and scouting runs. Using a wide range in solvent polarities, a wide range in analyte retention factors can be covered in a short time and one can usually assume that all analytes have been eluted shortly after the run reaches 100%. Several runs might produce all the information necessary to set up a suitable method.I4' In any gradient instrument, there is a delay between the time the gradient is started and the time the gradient reaches the column. If we call this the dwell time, t,, there is a related dwell volume, VD, depending on the flow

258

LIQUID CHROMATOGRAPHY IN COLUMNS

rate, F , of MP: t,

=

VD/F

(8.19)

The dwell volume is the total volume in the tubing, connections, and sampling valve/loop between the gradient former and the head of the column. Typically, it is several milliliters; consequently, changes in composition may take several minutes to reach the head of the column. If the sample is injected when the gradient is started, the sample arrives at the column before the gradient does. Until the gradient arrives, isocratic conditions prevail in the column. Therefore, it is desirable to determine the dwell volume in the method development process. A simple procedure has been described."*. "" It is a characteristic of the separation that must be taken into account in the method development process. It will affect the retention times and may affect resolution. When a method is transferred to a different instrument, the dwell volume is likely to be different, affecting the chromatographic results. An increase in the dwell volume will cause the gradient to arrive at the head of the column at a later time so the retention times will be longer and the method appears to be out of specification. Including the dwell volume in the method specifications is not likely to be helpful since the receiving laboratory will probably not be able to change the dwell volume in its instrument. A better approach is for the method developer to use a very low dead volume system for development and incorporate an initial isocratic portion into the gradient. The receiving lab can then use its instruments dwell time and modify the length of the isocratic portion. Computer Methods

The methods just described are labor intensive, often involving some trial and error. However, computerized programs are commercially available that can be used to simulate HPLC runs based on HPLC theory and thus save time and reagents. The procedures for using these programs have been thoroughly presented,"' and Table 8.17 lists some of them. Probably the most widely used program is DryLab, which was described in 1986 for isocratic runs''* and later for gradient elution runs.'s3 To use this software, Table 8.17

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

Some Commercial Method Development Software

DryLab, LC Resources: www.lcresources.com PESOS, Perkin-Elmer: www.lus.perkineltner.com ChromSmart MD, Intelligent Laboratory Solutions, Inc: www.chromsmurt.com HPLC Optimization, ChemSW: www.chemsw.com ACD/Method Dev. Suite ACD/ChromManager, ACD Labs: www.ucdlabs.com ChromSword, Merck KGaA: w w w .hii.hitachi.com/LC%2Ochromsword.htm

8.4

REVERSED-PHASEMETHOD DEVELOPMENT AND OPTIMIZATION

259

two or three runs are made in the laboratory, and the results are entered into the program. The software then produces diagrams and predictions for future runs. Another popular program is PESOS.'S4 Unlike DryLab, it does not require any survey runs prior to using the software. Many other valuable studies on computer-assisted method development in chromatography have been published together in an issue of the Journal of Chromatography AI5' and also as a separate volume.'s6 A total of 42 studies are included covering four categories: optimization, expert systems, special techniques, and applications. Two studies are on Dry Lab and specific applications include several for the pharmaceutical industry, for IEC, and for TLC. A three-step strategy that combines HPLC method development with Plackett-Burman experimental design has been published."' Troubleshooting and Tips

It is impossible to create an exhaustive list of all symptoms and problems that can occur in HPLC. The most comprehensive compilation is the book by DoIan and Snyder,"' and Meyer has depicted common errors in pictures.'" Many vendors supply their own brief troubleshooting guides (see e.g. Phenomenex'"' and LC-GC Magazine'6' ). Dolan writes a regular troubleshooting column in LC-GC that provides many useful tips on HPLC. There are, however, a number of common causes that account for a majority of the HPLC symptoms. Here is a list of those symptoms and the associated causes. HPLC Pump-Source of Most Common HPLC Problems Symptom Periodic pressure fluctuation Cause 1 Leaky check valve For the check valve to work correctly, the internal parts (ball and seat) must be able to form a tight seal. If foreign objects (air or dirt) are introduced into the check valve, it will leak. The leaks do not produce any visible signs such as drops of liquid, only regular or periodic pressure perturbations. Solution 1 After making sure that there is no source of air or dirt entering the pump, open the purge valve and purge the pump at a high flow rate. It may help to gently tap the check valves with a small wrench or screw driver. This will help to dislodge any trapped air bubbles. If this does not fix the problem, remove the check valve and either clean it through sonication in dilute nitric acid or replace it with a new check valve. Cause 2 Leaky pump seals If the pump seals are leaking badly, liquid will be visible coming from the back of the pump heads. The seals have a limited lifetime and need to be

260

LIQUID CHROMATOGRAPHY IN COLUMNS

replaced as a part of the routine preventative maintenance of the HPLC system. Although seals may last over 12 months, it is a good idea to replace them at regular intervals (every 6 months). Follow the manufacturers direction for replacing the seals, along with any associated backup seals, O-rings, or wear retainers. Symptom High pressure There is a clog somewhere in the system. Disassemble the HPLC components starting with the last component-usually the detector. Keep removing components until the high-pressure problems go away, then focus the troubleshooting efforts on that piece of equipment. Probable Cause 1 Clog in one of the following filters: 1. Any filter in the pump (some pumps have a filter built into the purge valve) 2. High-pressure in-line filter (comes after the injector) 3. Inlet frit of the guard column 4. Inlet frit of the analytical column Solution 1 Clean the filter by sonication for 10 min in dilute nitric acid (6 M), or replace the filter. Wash thoroughly with water after nitric acid cleaning. Cause 2 Clogged Tubing Due to MP Buffer Precipitation Check the entire injector and the tubing leading from the injector to the column. If you have been using buffer and have recently introduced a high concentration of organic solvent, the buffer may have precipitated in the lines. Solution 2 Wash the system with pure water. Hot water will dissolve precipitates best. Start at a very low flow rate (0.1 mL/min) until the pressure decreases. Cause 3 The sample has precipitated in the lines Solution 3 Wash the system with a high concentration of the sample solvent. Then make several large-volume injections of that same solvent. This should dissolve any precipitated solvent. For now on, dissolve the sample in the MP to avoid precipitation. Symptom Low pressure Cause 1 Leak Solution 1 Concentrate on the high-pressure part of the HPLC. This includes everything from the pump heads to the top of the column. Find the leak (drips or puddle) and tighten the fitting(s). Cause 2 No solvent This one happens to the best of u s . . .you ran out of solvent! Solution 2 Refill the solvent container, then purge the lines and pump with the fresh MP.

8.5

RP-HPLC AND ALTERNATIVES FOR THE PHARMACEUTICAL INDUSTRY

261

Symptom Poor peak area or height reproducibility Cause Problem with the autosampler Solution Concentrate your efforts on the injection system. There may be a clog in the syringe needle or a leaky line between the syringe (or metering device) and the sample vial. Replace the needle or tighten the leaky tubing. Symptom Noisy baseline Confirmation Step Shut off the pump in order to rule it out as a cause. If the noise remains after the pump has been shut off, it is probably a detector problem or a bubble lodged in the detector cell. Cause 1 Dirty detector cell Solution 1 Clean the cell with dilute (6 M) nitric acid. This is most easily accomplished with a 10-mL syringe and an adapter that allows the syringe to connect directly to the cell inlet. Inject 5 mL into the cell and allow it to stand for 10 min. Follow that with an additional 5 mL and rinse the cell with water. DO NOT PUMP THIS SOLUTION THROUGH YOUR COLUMN. Cause 2 Old detector lamp (especially for drifting baseline) Solution 2 Replace the lamp. Tips To aid in preventing problems, the following tips are offered: Tip 1. Use guard columns and precolumn filters to protect your columns. Tip 2. Use HPLC grade solvents, including water. Tip 3. Filter all solvents through 0.45-pm pore filters. Tip 4. Flush RP columns (and all lines and connections) with at least 10 column volumes of pure water if a buffer has been used. Then, for storage, pump 10 volumes of methanol through the column followed by a new solvent mixture. An 80/20 mixture of methanol and water, respectively, is satisfactory; the mixture must have at least 10% organic to prevent collapse of the bonded phase. For cleaning and regeneration steps, see Majors."' Tip 5. Dissolve the sample in mobile phase if at all possible. If another solvent mixture must be used, be sure it is miscible with the MP and that the sample does not precipitate out in the MP. Tip 6. Read Dolans Top-10 list.'h4 8.5 RP-HPLC AND ALTERNATIVES FOR THE PHARMACEUTICAL INDUSTRY

Many of the compounds encountered in the pharmaceutical industry are very polar and basic and not well separated by RP. The nonpolar SP does not retain the samples adequately to separate them, and free silanol groups in the silica of the SP cause unwanted secondary interactions and possible ion

262

LIQUID CHROMATOGRAPHY IN COLUMNS

exchange. A recent review of the literature summarizes techniques that have been suggested to overcome these problems.'65 The study by Yang et aLSy should also be consulted. An alternative to RPLC that was suggested in 1990 for separating polar solutes has been dubbed hydrophilic interaction chromatography, HILIC, and it is finding use in the pharmaceutical area. The abbreviation HILIC was suggested by Alpert'" instead of HIC because the latter has been used to denote hydrophobic interaction chromatography (discussed earlier in this chapter). HILIC is in effect an example of normal-phase (NP) chromatography, not RP, but the mechanism is probably mixed and not the conventional LSC described earlier. The stationary phases most often used are bonded phases in which the groups attached to the silica are hydrophilic (polar). Alpert used a poly(2-hydroxyethyl aspartamide) bonded to silica gel and others have used amides, carbamides, cyclodextrins, amino functionality, and the more conventional cyano bonded groups. The mobile phases are more like the ones used in RPLC, mixtures of water and acetonitrile, for example. The HILIC mechanism is more effective at the higher ( 270%) concentrations of ACN. Evidence that the separations are mainly NP comes from the elution order of the samples run: The polar solutes are retained the longest, which is the opposite order from RPLC. Conventional NPLC with nonpolar mobile phases has not been successful because the biological and pharmaceutical samples being run are not soluble in the nonpolar mobile phase. There is evidence, however, that the mechanism also involves the immobilization of some mobile phase, enriched in water, on the SP, and that an absorption into that phase occurs as well. Strege has used HILIC for a number of applications: with mass spec (MS) detection for the analysis of polar compounds for natural product drug disc0ve1-y'~~;for chiral separation of polar drugs'68; and a mixed-mode system combining HILIC with anion and cation exchange chromatography and MS detection for the separation of natural product extracts.'69 HILIC-MS has also been recently used to separate small polar compounds in food analy~is.'~"

8.6 PREPARATIVE LC

Liquid chromatography is often used to clean up samples to remove unwanted matrix interferences. Similarly, the isolation of a newly synthesized compound from its reaction mixture or a newly identified chemical from a natural product can be achieved with large columns that allow the separation of large amounts of material. This section discusses some of the factors to be

8.6

PREPARATIVE LC

263

considered in these types of separations, which are usually known as preparative, or prep, methods. Most of the discussion will be on low-pressure methods, but the principles also apply to high-pressure operation. Many publications on prep LC have been written, including those by Bidlingmeyer. '72 There are no definite divisions between low and high pressure and between analytical and preparative sample sizes. By low-pressure LC, we mean operation under gravity flow up to several hundred kilopascals of pressure. Prep LC will be loosely defined as sample sizes ranging from a few hundred milligrams up to several grams. Typically prep separations are performed with sample sizes 2 1 mg of sample per gram of column packing. 17'3

Sample Cleanup A number of manufacturers now supply short LC columns to be used under gravity flow or under the flow induced by vacuum, a centrifuge, or a syringe. Some of the trade names used are Sample Enrichment and Purification (SEP-PAC), PreSep Extraction, Bond-Elute, and solid-phase extraction. The columns are a few inches long and can be packed with any of the stationary phases used in LC, although silica and bonded phases are most common. The separations take a few minutes, and the eluent is collected for further analysis. The inexpensive cartridges are disposed of after use. More information is included in Chapter 14. Low-Pressure Prep LC The types of prep separations included in this section vary widely; they are not instrumental methods, although some pumps, detectors, injectors, and fraction collectors may be used. They are simple, similar to the original work by Tswett, and sometimes called classical LC. The study by Crane et al.173is typical. Table 8.18 Solids Used as Stationary Phases in LSC (in Order of Increasing Activity)

1. 2. 3. 4. 5. 6. 7. 8.

Sucrose Starch Inulin Magnesium citrate Talc Sodium carbonate Potassium carbonate Calcium carbonate

9. Calcium phosphate 10. Magnesium carbonate 1 1. Magnesium oxide 12. Silica gel 13. Magnesium silicate (Florid) 14. Alumina 15. Charcoal, activated 16. Fullers earth (kaolin-type clay)

264

LIQUID CHROMATOGRAPHY IN COLUMNS

Table 8.19 Typical Operating Conditions for Low-Pressure Preparative LC

Column i.d. Column length Weight of column packing Particle size Operating pressure Flow rate Linear velocity Plate number Sample size Sample volume

1-4 cm 20- 100 cm 5-500 g > 40pm

10-300 kPa

2-20 mL/min 0.1-1.0 mm/s 200-2000 0.1-10 g 0 . 5 2 0 mL

The stationary phase most often used is silica, but other solids are also listed in Table 8.18. Newer bonded phases can also be used, but they are more expensive. Relatively large particle sizes are needed to keep the pressure drop low and facilitate easy dry packing. The range of particle sizes is often large for the sake of economy if high efficiency is not needed. The columns are often made of glass and consequently can be used only at low pressures. Compared to analytical columns, they are short and fat with inside diameters around 1-4 cm and lengths of 20-100 cm. In all prep columns, higher efficiency is attained if the bottom of the column is rounded, and that configuration is easily achieved with glass. The volume of the column, and hence its sample capacity, goes up with the square of the column diameter. The amount of stationary phasc needed to pack the column increases by the same proportion, which is the reason for using low-cost packings. Some typical operating conditions are given in Table 8.19. The mobile phase should have the requirements discussed in the LSC section, but in particular it should be inexpensive and volatile to facilitate sample recovery. If silica gel is the stationary phase, the amount of water in the system becomes critical, and care must be exercised since inexpensive solvents may be wet. Inexpensive solvents may also contain less volatile impurities that will be concentrated when the solvent is evaporated, causing contamination of the isolated sample. The mobile phase can be allowed to flow under the force of gravity, a low-pressure pump can be used, or compressed gas can be used to pressurize a solvent reservoir. The linear velocity should be about one-third of that used in analytical columns. The sample can be applied to the top of the column with a microsyringe or pipet using the stop-flow method, or by use of an inexpensive, low-pressure valve. The eluent is usually collected in separate tubes using an automated fraction collector. Inexpensive UV detectors with large solvent volumes are available, or flow cells can be fitted to conventional UV/visible instruments.

8.6

PREPARATIVE LC

265

Since the samples run in prep LC are often very crude, it is to be expected that the column packing will become contaminated with chemicals that did not elute. Cleaning can be accomplished with polar solvents such as propanol or ethyl acetate (for normal-phase systems), or the contaminated top part of the packing can be scraped out and replaced with new packing. The latter alternative may be cheaper, considering the cost of solvents and waste disposal. Often TLC is used for fast screening of mobile phases since the stationary phases used in prep LC are also available for TLC. The TLC R , value should be equal to or less than 0.3. In prep work, the separation is usually optimized so that the column can be run with a sample overload and still produce adequate separations. Overloading will cause the plate number and partition ratio to decrease, and chromatograms produced this way are not pretty to look at. Flash Chromatography In 1978, Still et al.’74 published a “quick-and-dirty” prep method that has become very popular and even has its own name-flush chromatography. They used short, fat columns (1-5 cm i.d.x 18 in.), 40- to 65-pm silica, and low-viscosity solvents (e.g., ethyl acetate/ petroleum ether mixtures). The exact solvent mixture is selected by screening the sample via TLC to give an R , value of about 0.35. However, they found that when the proportion of the polar component was small, it was better to use about half as much of it for the column separation as was used for TLC. Basically, the procedure is as follows. The column is partially filled (5-6 inches) with the silica gel, and solvent is added to fill it. Compressed air is used to flush the solvent quickly through the packing, compressing it and removing all air from it. The sample is then added to the top of the column, and the column is again filled with solvent. The pressure is adjusted to get a flow of 2 inches/min, and the separation is accomplished in 5-10 min. Because the MP is forced through the column with pressurized air, the separation is much faster than gravity flow separations. While this method has been found useful, it is time consuming, so it is not surprising that some automated systems are now being marketed.’75 (See, e.g., ISCO’7” and Argonaut 177 ). A typical application is the use of automated flash chromatography for high-throughput screening for drug discovery.’7X Dry Column LC Another variation of low-pressure LC is performed on a column of dry packing”” similar to conventional TLC. The column is a nylon tube. When the mobile phase reaches the end of the column, flow is stopped, so the sample is not eluted. Rather, the column is sliced, and the components are recovered from the packing after their locations are established by fluorescence or staining reagents. This column method can handle larger samples than prep TLC.

266

LIQUID CHROMATOGRAPHY IN COLUMNS

High-pressure Prep LC

For difficult separations, high-performance columns are required, which means small particle columns, pumps, injection valves, detectors, and data systems.'x" The use of an instrument will also improve the speed of analysis and offers the possibility of automatic operation. A recent poll found that nearly 75% of the prep LC was done with high performance (HPLC) instruments.'" One possibility is to scale up an analytical instrument for prep work. High-efficiency prep columns are available for that purpose, but only column diameters two or three times analytical size are usable on analytical instruments. The limitations are the pump and the detector. A column diameter three times as large as the analytical column will require nine times as much flow to maintain the same linear velocity, and this flow rate may approach the limit of the pump. The detector must be capable of handling these higher flows and sample sizes without overloading. As a compromise, a semiprep configuration is often used, and the sample is run repeatedly to get sufficient sample isolated. Thirteen commercial preparative units have been compared,'x2 one of which can be operated as a supercritical fluid prep chromatograph. These instruments can also accommodate larger columns and hence larger samples. Capacities up to 750 g and flows of 5 L/min are claimed. Columns are expensive and overload easily. In addition to normal slurry packing techniques, radial and axial compression devices'83 can be used to improve performance. Another method of operating a prep HPLC is the use of a method called continuous annular chromatography (CAC). The sample is continuously (rather than batch) fed onto the top of a cylindrical annular column that is rotated during the run. Mobile-phase flow is by gravity and the fractions can be collected at the bottom. A recent review contains 55 reference^."^ The refractive index detector is popular in prep units because gradients are not often used, it is universal, and it is less sensitive. UV detectors are often too sensitive, and short path length cells are needed. Alternatively, the UV detector can be detuned to a less sensitive wavelength, or the effluent stream can be split, sending only part of it through the detector The maximum sample size that can be injected without a loss in resolution can be calculated with Eq. (8.9)"':

r

Since the dead volume VM is rather large in prep columns, the size of the sample volume can also be quite large and may require a pump for filling the

87

SPECIAL TOPICS

267

sample loops in the sample valve. For maximum efficiency, the sample should be spread out laterally over the entire column, and a study of this effect has recommended special inlet configurations.'" Since 1986, nine international symposia have been held on preparative and industrial chromatography. The papers from these meetings have been published in the JoLirnal of Chromutogruphy, the most recent being volume 1006. 8.7 SPECIAL TOPICS

There are a few topics to be introduced here even though space does not permit them to be thoroughly discussed. References are provided for further information. Performance Qualification Kit

The requirement in the regulated pharmaceutical industry for HPLC instrument qualification was presented in Chapter 1. To make the process easier and to ensure compliance, a test kit has been offered commercially.1XXIt consists of a set of NIST-traceable solutions, a test column, software, and test protocols. It can be used to qualify both isocratic and gradient systems. Pseudophase or Micellar LC

Aqueous micellar solutions were first used as mobile phases in LC in 1980. A micellar solution is one that contains a surfactant at a concentration above the critical micelle concentration, about lo-' M. The nature of micelles and their use in liquid chromatography has been described.lx" For HPLC, micelles have been used with nonpolar bonded phases and are typical of reversed-phase separations as we have discussed them. They have also been called pseudophase HPLC or soap chromatography. At present they seem to be unique enough to warrant such special names, but it may turn out that they are just special forms of reversed-phase LC. The surfactants that have been used are the anionic dodecylsulfate, cationic quarternaries such as hexadecyltrimethylammonium ion, and nonionics such as Triton X-100 and Brij-35.'") The solvent is water or water-organic mixtures. The special properties of the mobile phases are claimed to provide unique selectivities and have been shown to be capable of handling direct injection of blood serum samples for therapeutic drug monitoring."' Gradients can be run, and no equilibration is necessary to return them to the original conditions, thus saving time. See the introduction to reference'"* for a literature summary.

268

LIQUID CHROMATOGRAPHY IN COLUMNS

Other Topics

A few special instrument configurations are worthy of a brief note. In cases where a single pass through a column has not produced adequate resolution, the eluent from the column can be passed through the column again. This process has become known as recycle chromatography and has been most often used in size exclusion and prep chromatography. As a minimum, a valve is required to divert the eluent back to the pump where it replaces the mobile phase. Often, however, this process also includes some heartcutting, especially in prep LC. Another special application is the miniaturization of HPLC, so-called microchip HPLC. This topic is included in Chapter 1.5. 8.8 SUMMARY AND EVALUATION

An interesting summary of the practical aspects of HPLC theory was published by KnoxIy3back in 1977. A more recent and extensive work is the book by M e ~ e r . Both ' ~ ~ of these publications discuss the quality of HPLC columns and HPLC separations in terms of some testing procedures that can be performed. The discussion of these parameters was begun in Chapter 6, where the flow resistance parameter @ was defined: (8.21)

A dimensionless parameter, it typically has a value between 500 and 1000; values higher than that indicate a probable blockage in the system. Another dimensionless parameter of interest is the separation impedance, E , a low value for which indicates a good column:

E=h2@

( 8.22) (8.23)

A theoretical lower limit for E is about 2000 and values higher than l o 5 describe columns that can no longer be called high perjiormance.l6" In evaluating an HPLC column, it is recommended that calculations of h , v , @, and E be made from a typical set of data.'66 One such set of calculations has been published'"i and is summarized in Table 8.20. Four commercial columns were compared using the column parameters we have defined and described: N , H , h , and a. Although variations exist among the columns, the values presented provide convenient comparisons against which

8.8

269

SUMMARY AND EVALUATION

Table 8.20 Evaluation of Four Commercial 15-cm C,, Columns at a Reduced Flow of Approximately 5

Column

Parameters Y

F" (mL/min) k , toluene Nh

H(pm) h (D h P (bar) E

A

B

C

D

4.9 0.50 3.8 13,200 11.3 1.9 1090 34 3900

5.2 0.60 3.3 11,900 12.6 2.5 630 36 400

4.1 0.60 2.1 12,100 12.4 2.5 690 35 4300

4.8 0.55 2.5 10,600 14.1 3.5 5 10 52 6300

Source: From P a d s and McCoy."' Reproduced from the Joiiniul of' Chrornuiugruphic Science by permission of Preston Publications, Inc. "Volumetric flow rate. "Calculated employing width at half height.

one can judge his/her own columns performance. It is highly recommended that any new or repacked column be tested to evaluate its performance, and that it be periodically reevaluated to determine when it needs to be repacked or discarded. Reference 195 can be consulted for additional details on column evaluation. Table 8.21 contains a summary of the advantages and disadvantages of HPLC in columns. Although HPLC is compared to GC, the two techniques should be considered to be complementary, not competitive-each has its own special advantages and applications. A concise retrospective on HPLC was published by Snyder'"" in 2000, and it makes a fitting summary for this chapter. The biennial reviews in Analytical Chemistry are good sources of information; see, for example, reference 192. Table 8.21

Evaluation of HPLC

Advantages I . Applicable to nonvolatilc samples 2. Efficient, selective, and widely applicable 3. Quantitation is possible 4. Requires only a small sample 5. Nondestructive

Disadvantages I . Often slower than GC 2. Experimentally more complex than GC 3. Less compatible with MS 4. No detector equivalent to FID used in GC

270

LIQUID CHROMATOGRAPHY IN COLUMNS

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.

R. E. Majors, Anal. Chem. 1972, 44, 1722. J. M. Anderson, J . Chromatogr. Sci. 1984, 22, 343. L. R. Snyder, J . Chromatogr. Sci. 1978, 16, 223. L. R. Snyder, P. W. Carr, and S. C. Rutan, J . Chromatogr. A 1993, 656, 537-547. K. K. Unger, Porous Silica, Elsevier, New York, 1979. R. P. W. Scott, Adu. Chromatogr. N. Y. 1982, 20, 167. L. R. Snyder, Principles of Adsorption Chromatography, Dekker, New York, 1968. R. W. Yost and R. D. Conlan, Chromatogr. Newsletter (Perkin-Elmer),1972, 1(1), 5. D. L. Saunders, J . Chromatogr. Sci. 1977, 15, 372. L. R. Snyder and D. L. Saunders, J . Chromatogr. Sci. 1969, 7 , 195. J. N. Driscoll and I. S. Krull, Am. Lab. 1983, 15(5),42. R. K. Gilpin, J . Chromatogr. Sci. 1984, 22, 371. C. H. Lochmuller and M. T. Kersey, Anal. Chem. 1988, 60, 1910-1914. S. D. Fazio, S. A. Tomellini, H. Shih-Hsien, J. B. Crowther, T. V. Raglione, F. R. Floyd, and R. A. Hartwick, Anal. Chem. 1985, 57, 1559. M. D. Ludes, S. R. Anthony, and M. J. Wirth, Anal. Chem. 2003, 75, 3073-3078. C. Horvath and W. Melander, J . Chromatogr. Sci. 1977, 15, 393. D. C . Locke, J . Chromatogr. Sci. 1974, 12, 433. K. B. Sentell and J. G. Dorsey, Anal. Chem. 1989, 62, 930-934 R. Dolfinger and D. C. Locke, Anal. Chem. 2003, 75, 135551364, R. K. Gilpin, Anal. Chem. 1985, 57, 1465A. J . Chromatogr. A 1993, 656, 1-615. J. G. Dorsey and W. T. Cooper, Anal. Chem. 1994, 66, 857A-867A. R. P. W. Scott, Silica Gel and Bonded Phases, Wiley, New York, 1993. N. S. Wilson, M. D. Nelson, J. W. Dolan, L. R. Snyder, R. G. Wolcott, and P. W. Carr, J . Chromatogr. A 2002, 961, 171. N. S. Wilson, M. D. Nelson, J. W. Dolan, L. R. Snyder, and P. W. Carr, J . Chromatogr. A 2002, 961, 195. N. S. Wilson, M. D. Nelson, J. W. Dolan, L. R. Snyder, P. W. Carr, and L. C. Sander, J . Chromatogr. A 2002, 961, 217. J. J. Gilroy, J. W. Dolan, and L. R. Snyder, J . Chromatogr. A 2002, 961, 757. J. J. Gilroy, J. W. Dolan, P. W. Carr, and L. R. Snyder, J. Chromatogr. A 2004, 1026, 77. N. S. Wilson, J. J. Gilroy, J. W. Dolan, and L. R. Snyder, J . Chromatog. A 2002, 961, 91-100. J. J. Gilroy, J. W. Dolan, P. W. Carr, and L. R. Snyder, J . Chromatogr. A 2003, 1000, 757-778. J. E. Pemberton, M. Ho, C. J. Orendorff, and M. W. Ducey, J . Chromatogr. A 2001, 913, 243-252. M. W. Ducey, Jr., C. J. Orendorff, J. E. Pemberton, and L. C. Sander, Anal. Chem. 2002, 74, 5585-5592.

REFERENCES

271

33. C. J. Orendorff, M. W. Ducey, Jr., J. E. Pemberton, and L. C. Sander, Anal. Chem. 2003, 75, 3028-3036. 34. C. J. Orendorff, M. W. Ducey, Jr., J. E. Pemberton, and L. C. Sander, Anal. Chem. 200, 75, 3037-3043. 35. N. Nagae, T. Enami, and S. Doshi, LC-GC No. Am. 2002, 20, 964-972 36. M. D. Foster and R. E. Synovec, Anal. Chem. 1996, 68, 2838-2844. 37. C. F. von Hohenesche, V. Ehwald, and K. K. Unger, J . Chromatogr A 2004, 1025, 177-187. 38. R. Hartwick, in Analytical Chemistry in a GMP Environment, L. M. Miller and L. C. Crowther (ed), Wiley, New York, 2000, Chapter 10. 39. P. T. Ying and J. G. Dorsey, Talanta 1991, 38, 237-243. 40. Comparison Guide to C18 Reversed Phase HPLC Columns, 2nd Ed., Mac-Mod Analytical, Chadds Ford, PA, 2003; www.muc-mod.com. 41. S. Eksborg, P. 0. Lagerstrom, R. Modin, and G. Schill, J . Chromatogr. 1973, 83, 99; S. Eksborg and G. Schill, Anal. Chem. 1973, 45, 2092. 42. J. H. Knox and G. R. Laird, J . Chromatogr. 1976, 122, 17. 43. H. Liu and F. F. Cantwell, Anal. Chem. 1991, 63, 2032-2037. 44. J. G. Chen, S. Weber, L. L. Glavina, and F. F. Cantwell, J . Chromatogr. A 1993, 656, 549-576. 45. T. Cecchi, F. Pucciarelli, and P. Passamonti, Anal. Chem. 2001, 73, 2632-2639. 46. B. A. Bidlingmeyer, J . Chromatogr. Sci. 1980, 18, 525. 47. M. T. W. Hearn, Adu. Chromatogr. N . Y. 1980, 18, 59; M. T. W. Hearn (ed.), Ion-Pair Chromatography, Dekker, New York, 1985. 48. F. M. Rabel, Adv. Chromatogr. N . Y. 1979, 17, 53. 49. F. M. Rabel, J . Chromatogr. Sci. 1980, 18, 394. 50. S. Hjerten, Adv. Chromatogr. N . Y . 1981, 19, 59. 51. N. Cooke, P. Shieh, and N. Miller, LC-GC 1989, 7, 954-958. 52. I. H. Hagestam and T. C. Pinkerton, Anal. Chem. 1985, 57, 1757. 53. N. Tanaka, H. Kobayashi, K. Nakanishi, H. Minakuchi, and N. Ishizuka, Anal. Chem. 2001, 73, 420A-429A. 54. J. H. Smith and H. M. McNair, J . Chromatogr. Sci. 2003, 41, 209-214. 55. D. V. McCalley, J . Chromatogr. A 2002, 965, 51-64. 56. M. P. Rigney, T. P. Weber, and P. W. Carr, J . Chromatogr. 1989, 484, 273-291 57. M. P. Rigney, E. F. Fukenbush, and P. W. Carr, J . Chromatogr. 1990, 499, 464. 58. C. J. Dunlap, C. V. McNeff, D. Stoll, and P. C. Carr, Anal. Chem. 2001, 73, 598A-607A. 59. X. Yang, J. Dai, and P. W. Carr, J . Chromatogr. A 2003, 99, 13-31. 60. www.zirchrom.com. 6 1. w w w .sigmaaldnch.com. 62. R. Stevenson, Am. Lab. News Ed. 2003, 35(11), 6. 63. J. Nawrocki, C. Dunlap, A. McCormick, and P. W. Carr, J . Chromatogr. A 2004, 1028, 1-30. 64. J. Nawrocki, C. Dunlap, J. Li, J. Zhao, C. V. McNeff, A. McCormick, and P. W. Carr, J . Chromatogr. A 2004, 1028, 31-62.

272

65. 66. 67. 68. 69. 70. 71. 72. 73.

74. 75. 76. 77. 78. 79. 80. 8 1. 82. 83. 84.

85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99.

LIQUID CHROMATOGRAPHY IN COLUMNS

R. E. Majors, LC-GC No. Am. 2003, 21, 240-2519, R. E. Majors, Am. Lab. 2003, 35(20), 46-54. M. Petro, F. Svec, and J. M. J. Frechet, Anal. Chem. 1997, 69, 3131-3139. S. Moore and W. H. Stein, J . Biol. Chem. 1951, 192, 663. K. A. Kraus and G. E. Moore, J . A m . Chem. Soc. 1953 75, 1460. F. M. Rabel, Adu. Chromatogr. N . Y . 1979, 17, 53. A. Staby, I. H. Jensen, and I. Mollerup, J . Chromutogr.A 2000, 897, 99. A. Staby and I. H. Jensen, J . Chromatogr. A 2001, 908, 149. A. Staby, M-B. Sand, R. G. Hansen, J. H. Jacobsen, L. A. Anderson, M. Gerstenberg, U. K. Bruus, and I. H. Jensen, J . Chromatogr. A 2004, 1034, 85-97. R. M. Baum, Chem. Eng. New3 1985, May 20, p. 9. H. Small, Anal. Chem. 1983, 55, 235A. H. Small, T. S. Stevens, and W. C. Bauman, Anal. Chem. 1975, 47, 1801. T. S. Stevens, G. L. Jewett, and R. A. Bredeweg, Anal. Chem. 1982, 54, 1206. T. S. Stevens and M. A. Langhorst, Anal. Chem. 1982, 54, 950. H. Small and T. E. Miller, Anal. Chem. 1982, 54, 462. P. R. Haddad, Anal. Chem. 2001, 73, 266A-273A. www.dionex.com . Y. Liu, N. Advalovic, C. Pohl, H. Dhillon, and R. Kiser, Am. Lab. 1998, 30(22), 48~-54~. M. W. Kinderman, Am. Lab. 2003, 35(3), 26-31. J. S. Rohrer, Toduys ChemiJt at Work 2003, 12(5), 17-19; www.tcuwonline.org. P. Hatsis and C. A. Lucy, Anal. Chem. 2003, 75, 995-1001. R. Stevenson, Am. Lab. News Ed. 2003, 35(4), 42-46. J. D. Navratil and H. F. Walton, A m . Lab. 1976 8(1), 69. V. A. Davankov and A. V. Semechkin, J . Chromatogr. 1977, 141, 313. D. D. Bly, Science 1970, 168, 527. R. Nielson, P. Alden, and J. Yang, A m , Lab. 2004, 36(3), 34-44. J. Cazes, J . Chem. Educ. 1970, 47, A461 and A505. J. Cazes, J . Chem. Educ. 1966, 43, A567 and A625. W. W. Yau, J. J. Kirkland, and D. D. Bly, Modern Size Exclusion Chromatography, Wiley, New York, 1979. H. G. Barth, B. E. Boyes, and C. Jackson, Anal. Chem. 1998, 70, 251R-278R. S. Podzimek, Am. Lab. 2002, 34(1), 38-45. W. R. LaCourse, Anal. Chem. 2002, 74, 2813-2832. The latest review is J . G. Dorsey, W. T. Cooper, B. A. Siles, J. P. Foley, and H. G. Barth, Anal. Chem. 1998, 70, 591R-633R. J. A. Riddick and W. B. Bunger, in Technique of Chemistiy, 3d ed., Vol. 2, Organic Solvents, A. Weissberger (ed.), Wiley, New York, 1970. Sadek, P. C., The HPLC Solvent Guide, 2nd ed., Wiley-Interscience, New York, 2002.

REFERENCES

100. 101. 102. 103. 104. 105. 106. 107. 108.

109. 110. 111. 112.

273

J. W. Dolan, LC-GC 1985, 3, 1050. M. W. Dong and M. R. Cant, LC-GC 1984, 2, 294. J. H. b o x , J . Chromatogr. Sci. 1980, 18, 453-461. C. E. Reesc and R. P. W. Scott, J . Chromutogr. Sci. 1980, 18, 479-486. R. T. Kennedy and J. W. Jorgenson, Anal. Chem. 1989, 61, 1128-1 135. E. J. Caliguri, P. Capella, L. Bottari, and I. N. Mefford, Anal. Chem. 1985, 57, 2423. P. Kucera (ed), Microcolumn High-Performance Liquid Chromatography, Elsevier, Amsterdam, 1984. R. P. W. Scott (ed), Small Bore Liquid Chromatography Columns, Wiley, New York, 1984. M. Novotny and D. Ishii (eds), Microcolumn Separation Methods, Elsevier, Amsterdam, 1985. F. Steiner, Applications of Narrow-Bore Columns in HPLC, Hewlett-Packard, Palo Alto, CA, 1991, pub. #12-5091-2736E from German affiliate. R . P. W. Scott, Ado. Chromatogr. N . Y . 1983, 22, 247. M. Novotny, LC-GC 1985, 3, 876. R . E. Majors, L C - G C N o . A m . 2003, 21, 1124-1133.

113. F. V. Warren, Jr. and B. A. Bidlingmeyer, Anal. Chem. 1988, 60, 2824-2827. 114. See W. R. Biggs and J. C. Fetzer, Anal. Chem. 1989, 61, 236-240 and references therein. 115. J. D. Thompson and P. W. Carr, Anal. Chem. 2002, 74, 4150-4159. 116. R. Stevenson, A m . Lab. 2002, 34(22), 13-19. 117. Y. Mao and P. C. Carr, LC-GC No. Am. 2003, 21, 150-167. 118. See, e.g., D. Parriot (ed), A Practical Guide to HPLC Detection, Academic, San Diego, 1993. 119. D. G. Jones, Anal. Chem. 1985, 57, 1057A and 1207A. 120. J. R. Riordon, Anal. Chem. 2000, 72, 483A-487A. 121. S . Levin and E. Grushka, Anal. Chem. 1986, 58, 1602. 122. J. W. Dolan, LC-GC 1988, 6, 112-116. 123. J. Stahlberg and M. Almgren, Anal. Chem. 1989, 61, 1109-1112. 124. M. B. Smalley and L. B. McCowan, “Fluorescence Detectors in HPLC,” in Advances in Chromatography, Vol. 37, P . R. Brown and E. Grushka (ed), Marcel Dekker, New York, 1997. 125. J. W. Jorgenson and J. deWit, in Detectors for Capillary Chromatography, H . H . Hill and D. G. McMinn (ed), Wiley, New York, 1992, Chapter 15. 126. M. J. Felton, Anal. Chem. 2002, 74, 631A-634A. 127. J. A. Koropchak, L-E Magnusson, M. Heybroek, S. Sadain, X. Yang, and M. P. Anisimov, in Advances in Chromatography, Vol. 40, P. R. Brown and E. Grushka (ed), Marcel Dekker, New York, 2000, Chapter 5. 128. A. Kuch and R. Saari-Nordhous, A m . Lab. 2001, 33,(6), 61. 129. C. S. Young and J. W. Dolan, LC-GC No. A m . 2003, 21, 120-128.

274

LIQUID CHROMATOGRAPHY IN COLUMNS

130. M. C. Allgeier, M. A. Nussbaum, and D. S. Risky, LC-GC No. A m . 2003, 21, 376-381. 131. C. S. Young and J. W. D o h , LC-GC No. Am. 2004, 22, 244-250. 132. L. N. Polite, H. M. McNair, and R. D. Rocklin, J . Liq. Chromatogr. 1987, 10, 829-838. 133. R. W. Frei, H. Jansen, and U. A. Th. Brinkman, Anal. Chem. 1985, 57, 1529A. 134. I. S. Krull (ed), Reaction Detection in Liquid Chromatography, Dekker, New York, 1986. 135. J. L. Glajch, J. J. Kirkland, K. M. Squire, and J. M. Minor, J . Chromatogr. 1980, 199, 57. 136. R. J. Laub, Am. Lab. 1981, 13(3), 47. 137. S. N. Deming and M. L. H. Turoff, Anal. Chem. 1978, 50, 546. 138. B. Sachok, R. C. Kong, and S. N. Deming, J . Chrornatogr. 1980, 199, 317. 139. P. J. Schoenmakers, A. C. J. H. Drouen, H. A. H. Billiet, and L. de Galan, Chromatographia 1982, 15, 688. 140. J. M. Miller, Am. Lab. 2000, 32(20), 13-19. 141. D. Sykora, E. Tesarova, and D. W. Armstrong, LC-GC No. A m . 2002, 20, 974-981. 142. L. R. Snyder, J. J. Kirkland, and J. L. Glajch, Practical HPLC Method Deuelopment, 2nd ed., Wiley, New York, 1997, Chapters 6 and 7. 143. J. W. D o h , LC-GC No. Am. 1999, 17,478; 2000, 18, 18-32, 118-125,286-294, 376-382, 478-487. 144. See, e g , P. Nikitas and A. Pappa-Louisi, J . Chromatogr. A 2002, 971, 47-60. 145. P. Jandera and J. Churacek, Gradient Elution in Column Liquid Chromatography, Elsevier, Amsterdam, 1985. 146. C. Liteanu and S. Gocan, Gradient Elution Chromatography, Wiley, New York, 1974. 147. L. R. Snyder, J. W. Dolan, and J. R. Cant, J . Chrornutogr. 1979, 165, 3 and 31. 148. L. R. Snyder, J. J. Kirkland, and J. L. Glajch, Practical HPLC Method Development, 2nd ed., Wiley, New York, 1997, Chapter 8. 149. J. W. Dolan, LC-GC No. A m . 2000, 18, 478. 150. L. R. Snyder and J. W. D o h , LC-GC 1990, 7, 524-537. 151. L. R. Snyder, J. J. Kirkland, and J. L. Glajch, Practical HPLC Method Deuelopment, 2nd ed., Wiley, New York, 1997, Chapter 10. 152. L. R. Snyder, J. W. D o h , and M. P. Rigney, LC-GC 1986,4, 921. 153. J. W. D o h , and L. R. Snyder, LC-GC 1987,5, 970; and J. W. D o h , and L. R. Snyder, Am. Lab. 1990, 22(8), 50. 154. J. R. Cant, F. L. Vandemark, and A. F. Poile, A m . Lab. 1990, 22(8), 15. 155. J . Chromatogr. A 1989, 485, 1-675. 156. J. L. Glajch and L. R. Snyder (ed), Computer-Assisted Development for High Peflorrnance Liquid Chromatography, Elsevier, Amsterdam, 1990. 157. W. Li and H. T. Rasmussen, J . Chromatogr. A 2003, 1016, 165-180. 158. J. W. DoIan and L. R. Snyder, TroubleshootingLC Systems, Humana, Clifton, NJ, 1989.

REFERENCES

275

159. V. R. Meyer, Pitfalls and Errors of HPLC in Pictures, Huthig, Heidelberg, Germany, 1997. 160. www.phenemoniw.com. 161. L C-GC North America, www .chromatographyonline.com. 162. J. W. D o h , LC-GC No. Am., monthly feature column. 163. R. E. Majors, LC-GC No. Am. 2003, 21, 19-26. 164. J. W. D o h , LC-GC No. Am. 2003, 21, 262-266. 165. R. J. M. Vervoort, A. J. J. Debets, H. A. Classens, C. A. Cramers, and G. J. de Jong, J . Chromatogr. A 2000, 897, 1-22 (119 references). 166. A. J. Alpert, J . Chromatogr. 1990, 499. 177-196. 167. M. A. Strege, Anal. Chem. 1998, 70, 2439-2445 168. D. S. Risley and M. A. Strege, Anal. Chem. 2000, 72, 1736-1739. 169. M. A. Strege, S. Stevenson, and S. M. Lawrence, Anal. Chem. 2000, 72, 4629-4633. 170. H. Schlichtherle-Cerny, M. Affolter, and C. Cerny, Anal. Chem. 2003, 75, 2349-2354. 171. B. A. Bidlingmeyer, Preparatiue Liquid Chromatography, Elsevier, Amsterdam, 1987. 172. B. A. Bidlingmeyer, Practical HPLC me tho do log^ and Applications, Wiley, New York, 1992, pp. 269-283. 173. L. J. Crane, M. Zief, and J. Horvath, Am. Lab. 1981, 13(5), 128. 174. W. C. Still, M. Kahn, and A. Mitra, J . Org. Chem. 1978, 43, 2923. 175. R. C. Willis, Today’s Chemist at Work 2004, 13(3), 51-52. 176. www.isco.com. 177. www.argotech.com 178. G. R. Eldridge, H. C. Vervoort, C. M. Lee, P. A. Cremin, C. T. Williams, S. M. Hart, M. G. Goering, M. O’Neil-Johnson, and L. Zeng, Anal. Chem. 2002, 74, 3963-3971. 179. B. P. Engelbrecht and K . A. Weinberger, Am. Lab. 1977, 9 ( 5 ) , 71. 180. See, e g , K. J. Duff and R. C. Ludwig, “The Packing and Evaluation of Small Particle Preparative Columns,” A m . Lab. 1994, 26,(5 April), 32KK-32MM or publication T495004, Supelco, Bellefonte,. PA. 181. LC-GC No. Am. 2003, 21, 14. 182. S. Miller, Anal. Chem. 2003, 75, 477A-479A. 183. C. Laub, J . Chromatogr. A 2003, 992, 41-45-151 184. F. Hilbrig and R. Freitag, J . Chromatogr. B 2003, 790, 1-15. 185. R. P. W. Scott and P. Kucera, J . Chromatogr. 1976, 119, 467. 186. A. W. J. De Jong, H. Poppe, and J. C. Kraak, J . Chromatogr. 1978, 148, 127. 187. J . Chromatogr. 2003, 1006, 1-293; see also J . Chromatogr. 1988, 484; 1991, 556 and 557; 1992, 590(1); 1994, 658(2). 188. Microsolv Technology Corp., Long Branch, NJ; www.microsolutech. 189. J. G. Dorsey, in Advances in Chromatography, Vol 27, P. R. Brown and E. Grushka (ed), Marcel Dekker, New York, 1987, Chapter 5.

276

LIQUID CHROMATOGRAPHY IN COLUMNS

190. M. F. Borgerding and W. L. Hinzc, Anal. Chem. 1985, 57, 2183. 191. F. J. DeLuccia, M. Arunyanart, and L. J. Kline Love, Anal. Chem. 1985, 57, 1564. 192. W. R. Lacourse, Anal. Chem. 2002, 74, 2813-2832. 193. J. H. Knox, J . Chromatogr. Sci. 1977, 15, 352-364. 194. V. R. Meyer, Practical High-Pe$ormance Liquid Chromatography, 2nd ed., Wiley, Chichester, England, 1993. 195. R. E. Pads and R. W. McCoy, J . Chromatogr. Sci. 1986, 24, 66. 196. L. R. Snyder, Anal. Chem. 2000, 72, 412A-420A.

BIBLIOGRAPHY Bidlingmeyer, B. A., Practical HPLC h'ethodoloby and Applications, Wiley, New York, 1992. Dolan, J. W., and Snyder, L. R., Troubleshooting LC Systems, Humana, Clifton, NJ, 1989 Snyder, L. R., Kirkland, J. J., and Glajch, J. L., Practical HPLC Method Development, 2nd ed. Wiley, New York, 1997. Sadek, P. C., The HPLC Solvent Guide, 2nd ed., Wiley-Interscience, New York, 2002.

QUANTITATION: DETECTORS AND METHODS The principles of quantitative analysis are basically the same for all analytical methods, and they are dependent on the correct application of a variety of interdependent factors such as: Obtaining a representative sample Reducing errors by careful sample handling Using reference standards properly Correctly applying the required test method Using appropriate instrumentation Maintaining the instrumentation Accurately and correctly handling data and calculations In a regulated industry, the following additional factors will have to be considered: Using validated test methods Using qualified (IQ, OQ, PQ) instrumentation (see Chapter 1) Using calibrated instrumentation Performing system suitability testing

Chromatography: Concepts and Contrusts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Millet 277

278

QUANTITATION: DETECTORS AND METHODS

Since the acquisition of data is usually achieved with a detector that is part of the chromatographic system, it is important to understand general detector characteristics so their use can be optimized. This chapter begins with a general discussion of detectors, building on the specific information included in the discussions of GC and LC, Chapters 7 and 8, respectively. It is followed by a section on data acquisition and then methods of quantitative analysis. Relevant material on method validation, quantitative analysis, and terms and symbols is included in Chapter 1, and it should be consulted for background information. 9.1

DETECTORS

This section on detectors classifies and compares them and presents their common characteristics. Also included is a general discussion of the method validation process. Classification of Detectors

Chromatographic detectors share many similar properties, yet they can be distinguished by their unique characteristics. While there are many ways detectors can be classified, five types are especially important in chromatography: Concentration versus mass flow rate Bulk property versus specific property Selective versus universal Destructive versus nondestructive Analog versus digital Each will be discussed briefly; the first one is probably the most significant. Concentration Versus Mass Flow Rate This classification system distinguishes between those detectors that measure the concentration of the analyte in the mobile phase compared to those that directly measure the absolute amount of analyte irrespective of the volume of mobile phase. One consequence of this difference is that peak areas and peak heights are affected differently by changes in mobile phase (MP) flow rate. The most common chromatographic detectors are classified in Table 9.1. The basic properties of most of them can be found in their respective chapters; the mass spectrometric (MS) detector is described in more detail in Chapter 10.

9.1

DETECTORS

279

Table 9.1 Classification of Chromatographic Detectors

Detector

Selectivity: Concentration ( C ) Bulk Property ( B ) Universal (U) or or Somewhat (SS) Detection Mass Flow Rate (M) Solute Property 6 ) Very (VS) Limit (ng)

HPLC

uv/vrs

Fluorescence RI MS Conductometric Amperometric GC FID TCD ECD MS

C C C

MU

C C

M C C M

s s

B

s

B

s s

B S

s

ss

VS U Uh

ss ss

ss

U VS Uh

0.1 0.001 1000 0.01 10 0.00 1 0.0 1 1 0.0001 0.01

"See text. hClassification is for total ion mode; see Chapter 10

To understand the difference between these two detector types, consider the effect on a UV signal if the flow is completely stopped. The detector cell remains filled with a given concentration of analyte, and its absorbance continues to be measured at a constant level. The resulting signal would look something like that shown in Figure 9 . 1 ~This . is typical of a concentrationtype detector. However, for a mass flow rate detector like the FID in which the signal arises from a burning of the sample, a complete stop in the flow rate will cause the delivery of the analyte to the detector to stop and the signal will drop to zero. When the flow is resumed, the analyte will be returned to the detector and the signal will resume as shown in Figure 9.lb. This is typical of a mass flow rate detector. Now consider a less dramatic change in flow-say from 1 mL/min to 0.5 mL/min. Suppose the peak detected at the higher flow had a relative height of 1.0 and a relative area of 1.0 for both types of detector. At the lower flow, the area of the concentration detector would be doubled, but its height would remain at 1.0. For the mass flow rate type, the lower flow would result in an unchanged area, and, since the peak would be broader, the height would be reduced to 0.5. These relationships are shown in Figure 9.2, and we conclude that the concentration detector produces areas that are flow dependent, but the mass flow rate type does not. On the other hand, the mass flow rate type produces a signal (peak height) that is flow dependent, but the concentration type does not.

280

QUANTITATION: DETECTORS AND METHODS Flow stopped

Flow resumed

I

I

Concentration type

Figure 9.1. Effect of stopping the flow on two types of detector: concentration and mass flow rate.

Detector

Flow rate

type

1 mL/rnin

0.5mL/min

Figure 9.2. Effect of flow rate on peak sizes for two types of detector: concentration and mass flow rate.

This difference in performance has three consequences. First, the flow rate has different effects on these two detector types. For the concentration type, an increase in flow causes a decrease in peak area, but the peak height remains the same. For the mass flow rate type, an increase in flow causes an increase in peak height but the peak areas remain unchanged. For most chromatographic analyses the flow rate is kept constant, and there are

9.1

281

DETECTORS

instances in programmed temperature gas chromatography (PTGC) where the flow decreases during programming, resulting in peak heights that vary and adversely affect their use for quantitative analysis. Second, it is difficult to compare the sensitivities of these two types of detector because their signals have different units; the better comparison is between minimum detectable quantity (see later discussion), which can have the units of mass for both types. And finally, valid comparisons between detector types can only be valid when one specifies the flow rate, the concentration, and the particular analyte. Most of the detectors in Table 9.1 are the concentration type. The exceptions are the FID and the MS; the latter is classified as a mass flow rate detector for both GC and HPLC. However, when the MS detector is used in HPLC with electrospray ionization at low flow rates (sometimes called nanospray conditions), its signal unexpectedly has the characteristics of a concentration-type detector as just described; that is, the areas of the peaks increase at a low flow rate. Apparently, the increasing efficiency of ionization offsets the lower number of analyte molecules present in the eluent, making it behave as a concentration-dependent detector.' Use is made of this fact in some HPLC runs where critical peaks are run through the MS detector at a reduced flow rate, regulated by a flow splitter, a technique called peak parking.' An example is shown in Figure 9.3 for an HPLC analysis of a

I

0

'

"

'

I

5

~

'

~

'

I

10

~

15

~

~

20

"

~

~

'

'

25

Time (min)

Figure 9.3. An HPLC example of peak parking. Reprinted from American Laboratory, 2003

35(24), 14. Copyright 2003 by International Scientific Communications, Inc.

I

'

~

282

QUANTITATION: DETECTORS AND METHODS

cytochrome c digest, where the initial flow of 300 nL/min is reduced to 50 nL/min for 4 min beginning at about 18 min. The additional analysis time in the MS also permits the accumulation of more scans, which improves the signal-to-noise ratio and provides data for further experiments such as MS/MS. The Separation Solutions column in American Laboratory in March, 2003, carried an interesting discussion about flow rate changes and quantitation.2 It provides answers to some specific questions about detector performance, and it serves to elaborate further this discussion and to emphasize the importance of this particular detector classification system. The data from most electrophoretic separations, but not always including capillary electrochromatography (CEC), are different from chromatographic data in one major respect. Each analyte migrates at a different rate, and the slower they migrate the broader their peaks become and the later they pass the detector. In chromatography, the total retention volume is made up of two contributions and the rate of analyte migration in the mobile phase is constant for all of them; it is their sorption in the stationary phase that effects their separation and determines when they pass the detector (i.e., elute). In electrophoresis, the later, wider peaks (which will have constant peak heights) will have larger peak areas than they should because they are moving more slowly through the detector. To correct for this effect, the area can be divided by the migration time.3 Bulk Property versus Specific Property Detectors referred to as bulk property types are constantly measuring a particular property that is exhibited by both the mobile phase and the analyte. For example, a refractive index (RI) detector used in HPLC is constantly measuring the RI of the MP. When an analyte appears in the detector, there is a change in RI, and it is detected by the transducer and recorded. This detector measures the same bulk property of the MP and the MP containing the analyte (the solution of analyte in MP). Often this type of detector has two cells, a sample and a reference cell; the output signal is t h e difference, or the ratio, between them. These detectors are inherently less sensitive because they only measure changes in a given property. In addition, they cannot be used in gradient elution HPLC because the baseline will drift as the bulk property (e.g., the RI) of the MP changes with MP composition. By comparison, a specific property type produces no signal (or perhaps only a very small signal) when there is no sample present. The appearance of a sample in the detector introduces a new type of signal and thus produces a relatively large signal (compared to zero signal for the baseline). This is the reason for using an MP with no UV absorption at the wavelength being

9.1

DETECTORS

283

monitored, as the solvent for a UV detector,in HPLC. This type of detector is called an analyte property or solute property detector, and it is inherently more sensitive than the bulk property type. Fortunately, most chromatographic detectors are the solute property type, as listed in Table 9.1. Selective versus Universal This detector category refers to the number or percentage of analytes that can be detected by a given system. A universal detector such as the RI detector theoretically detects all solutes, while the selective type responds to particular types or classes of compounds. There are differing degrees of selectivity; the UV detector is not very selective and detects many organic compounds while the fluorescence detector is much more selective and detects only those species that fluorescence in solution. Three degrees of selectivity are given in Table 9.1: universal (U), somewhat selective (SS) or very selective (VS). The IUPAC has defined two terms to indicate the degree of ~electivity.~ A selective detector is defined as: “ A detector which responds to a related group of sample components in the column effluent” and a specific detector is “a detector which responds to a single component or to a limited number of components having similar chemical characteristics.” This distinction between selective and specific is generally the one used in science, and the term specific could be substituted for very selective in Table 9.1. However, the classification system under discussion in this section is only intended to distinguish between selective and universal. Both types of detector have advantages. The universal detectors are used when one wants to be sure all eluted solutes are detected. This is important for qualitative screening of new samples whose composition is not known, for polymer analysis using SEC, and when performing quantitative analysis by the area normalization method (to be discussed later in this chapter). On the other hand, a selective detector that has enhanced sensitivity for only a certain group of compounds can provide trace analysis for that group even in the presence of other compounds of higher concentration. It can simplify a complex chromatogram by detecting only a few of the compounds present, selectively “ignoring” the rest. A common example is the analysis of pesticide residues in environmental samples whose matrix is very complex, but only the electronegative (halogenated) analytes are detected by the use of an electron capture detector (ECD) in a GC analysis. Other examples include the choice of wavelength of a UV detector to detect selectively only aromatic compounds; the use of an MS detector in the selected ion monitoring mode whereby it will detect ions with only one given mass (see Chapter lo); or a fluorescence detector that adds selectivity in both excitation and emission wavelengths and can be more selective than a UV detector. An example of

284

QUANTITATION: DETECTORS AND METHODS

11

3

1

6

h 7

8

10

-

I

0

14

12

pH 3.2

pH 3.6

I

10

I

20

A

* F-

I 30 min

Figure 9.4.Dual-detector chromatograms showing the selectivity of the fluorescence detector. Reprinted from Gross and Gruter, “Quantitation of mutagenic/carcinogenic heterocyclic aromatic amines in food products,” J . Chrornatogr. 1992, 592 278. Copyright 1992, with permission from Elsevier.

the latter is shown in Figure 9.4 in which it can be seen that the fluorescence detector did not detect peaks 2, 3, 4, 6, 7, and 8 but was more sensitive (than the UV detector) for most other peaks, especially peak 10.’ Destructive versus Nondestructive Nondestructive-type detectors are necessary if the separated analytes are to be reclaimed for further analysis, as, for example, when identifications are to be performed using auxiliary instruments. One way to utilize destructive detectors in this situation is to split the effluent stream and send only part of it to the detector, collecting the rest for further analysis. Analog versus Digitar Most detectors produce analog (continuous) signals that must be digitized before they can be manipulated by a digital computer. The main exceptions are the radioactive detector and the mass spectrometer.

9.1

DETECTORS

285

Detector Characteristics

The most important characteristic of a detector is the signal it produces, but three other important ones are noise, time constant, and cell volume. These will be discussed first to provide a background for the discussion about the signal.

Noise Noise is made up of random signals produced by a detector in the absence of a sample. In chromatography, it can be seen on the baseline and is also superimposed on the analyte peaks. Usually it is given in the same units as the normal detector signal-millivolts, microamps, absorbance units, and so on. Ideally, there should not be any noise, and detectors can be operated at a reduced sensitivity where noise is not evident. However, at higher levels of sensitivity noise arises from random fluctuations from the electronic components from which the amplifiers are made, from stray signals in the environment, from contamination, and from instrument malfunctions or problems such as chromatographic leaks or entrapped air bubbles in HPLC pumps. Circuit design can eliminate some noise attributed to electronic components and connections; shielding and grounding can isolate the detector from the environment; and sample pretreatment and proper instrument operation and maintenance can eliminate some noise from contamina-

+Time Figure 9.5. Example for the measurement of noise level and drift for a TCD. Copyright ASTM. Reprinted with permission.

286

QUANTITATION: DETECTORS AND METHODS

tion and malfunctions. Some additional suggestions for reducing noise can be found in reference 6. The definition of noise used by ASTM is depicted in Figure 9.5. The two parallel lines drawn between the peak-to-peak maxima and minima enclose the noise, given in millivolts in this example. In addition, the figure shows a long term noise or drift occurring over a period of 30 min. If at all possible, the sources of the noise and drift should be found and eliminated or minimized because they restrict the minimum signal that can be detected. The ratio of the signal to the noise is a convenient characteristic of detector performance. It conveys more information about the lower limit of detection than does the noise alone. Commonly the smallest signal that can be attributed to an analyte is one whose signal-to-noise ratio, or S/N, is 2 or more. As originally presented in Chapter 1, many organizations have used the S/N ratio to define two terms, the limit of detection (LOD), which is given a value of 3, and the limit of quantitation (LOQ), which is given the

S/N=lO

Figure 9.6. Illustration of two signal-to-noise (S/N) ratios: 10 and 3. Reprinted with permission from V. Meyer, Pitfalls & Errors of HPLC in Pictures, Wiley-VCH, Weinheim, Germany, 1997. Copyright John Wiley & Sons, Inc.

9.1

DETECTORS

287

I

0

20

40

60

Time(=)

Figure 9.7. Example for the measurement of response time of a TCD. Copyright ASTM. Reprinted with permission.

value of 10. These two levels are depicted in Figure 9.6, and one can appreciate that a S/N of 2 or more would be required to assure that a real signal is present. Time Constant The time constant, 7,is a measure of the speed of response of a detector. Specifically, it is the time (in seconds or milliseconds) a detector takes to respond to 63.2% of a sudden change in signal, as shown in Figure 9.7. The full response (actually 98% of full response) takes four time constants and is referred to as the response time. Unfortunately, some workers define response time as 2.2 time constants (not 4.0), corresponding to 90% of full-scale deflection (not 98%); others define a rise time as the time for the signal to rise from 10 to 90%. To further confuse the situation, some use the terms time constant and response time interchangeably. Either time constant or response time should be specified for a detector in every written procedure. However, because some instrument manufacturers have chosen to use other names such as filter or rise time without defining their relationship to time constant, some trial and error may be necessary to determine the optimum setting for a given instrument and to equate it to instruments from other vendors. In general, small time constants are most desirable, even though the noise level will be more evident.

288

QUANTITATION: DETECTORS AND METHODS

Figure 9.8. Effcct of detector time constant on peak characteristics;

T,

< T ? < T? < T~

Figure 9.8 shows the effect of increasingly longer time constants on the shape of chromatographic peaks. The deleterious effects are the changes in retention time (peak position in the chromatogram) and on peak width, both of which get larger as the time constant increases. The area, however, is unaffected, so quantitative measurements based on area will remain accurate while only those based on peak height will be in error. Consequently, special attention should be paid to peak height methods to ensure optimization of the time constant and achieve necessary method robustness. More dramatic is the effect of time constant on a chromatographic run. Figure 9.9 compares a five-component reversed-phase HPLC separation performed using time constants of 0.05 s ( a ) and 5 s ( h ) . The larger time constant produces wider peaks with decreased peak heights, decreased plate numbers, and decreased resolution-all undesirable. Equation (9.1) can be used to calculate an approximate maximum allowable time constant,’ based on a peak broadening of not more than 1%:

The parameters used for the calculation are measures of peak width or those related to it, namely the width at base (w,),the quarter-peak width or standard deviation ( a ) ,and the plate number ( N ) . Note that column length is not important except to the extent it affects these other parameters; thus, plate height ( H ) is not relevant to this calculation. The peak used for this calculation should be the first one (or an early one) in a chromatogram because it will be the narrowest and the most critical one (in either isocratic HPLC or isothermal GC).

9.1

DETECTORS

289

N

4 W m

7

1

P

I

z w

Figure 9.9. Effect of time constant o n a chromatographic separation. ( a ) T = 0.05 s, ( b ) T = 5 s. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chemistry in a G M P Enuironrnent, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

Either of the forms of Eq. (9.1) can be used for the calculation. For example, a peak with a retention time of 10 min on a good HPLC column of conventional dimensions (4.6 mm i.d. and 25 cm length) and a plate number of 20,000 would require a time constant of 0.4 s or less: 7=

0.1[

10 x 60

420,000

j

= 0.4

s

290

QUANTITATION: DETECTORS AND METHODS

This same peak might have a peak width (at base) of 2.0 mm, measured on a chart at a speed of 1 cm/min. Its quarter-band width, cr, would be 3 s, requiring a time constant of 0.3 s or less. These calculations lead us to conclude that the detector time constant for use in HPLC should be on the order of 0.5 s or less. Another commonly used rule-of-thumb' is that the time constant should be less than 10% of the width of the peak at half height, w , , . Many laboratories do not operate their HPLC detectors in this range. In fact, it is probably true that too many laboratories have no specification on the time constant in any of their methods! Problems often result during method transfer when time constants are not specified, and the receiving laboratory is using a detector whose default value for time constant is too large. Peaks from capillary GC columns are likely to be even narrower than HPLC peaks, and they require even smaller time constants. Finally, one needs to be aware that the overall time constant for the entire system is limited by the largest response value for any of the individual components: detector, amplifier, or data system. Large time constants do have the advantage of smoothing the short-term noise from a detector. This effect is sometimes called damping. The temptation to decrease instrument noise and improve S / N by increasing the time constant must be avoided. Valuable information can be lost when the data system does not faithfully record all the available information, including noise. Cell Volume Operation of chromatographic detectors is optimized when their internal volumes are small, since band broadening is minimized. However, for concentration-type detectors, the magnitude of that volume has special importance. Suppose the cell volume of a concentration detector is so large that the entire sample could be contained in one cell volume. The shape of the resulting peak would be badly broadened and distorted. Estimates can be made of ideal cell volume requirements since the width of a peak can be expressed in volume units (the base width, 4a,where the x axis is in milliliter or microliter units). A narrow peak from a capillary G C column might have a width as small as 1 s, representing a volume of 0.017 mL (17 p L ) at a flow rate of 1 mL/min. An ideal detector for this situation should have a significantly smaller volume, say 2 p L . When this is not possible, extra (makeup) mobile phase can be added to the column effluent to sweep the sample through the detector more quickly. This remedy will be helpful for mass flow rate detectors but less so for concentration detectors. In the latter case, the makeup MP dilutes the sample, lowering the concentration as well as the resulting signal-not a satisfactory solution in many cases.

9.1

DETECTORS

291

Consequently, concentrationtype detectors must have very small volumes if they are to be used successfully when the chromatographic peaks are very narrow. However, too small a cell may result in decreased detectivity Another consideration related to cell volume and peak width is the number of data points that are needed to provide a true representation of a chromatographic peak. Narrow peaks require fast data-sampling times as well as small cell volumes (and time constants). Like large cell volumes, too few data points will distort peak shapes, producing peaks that are not symmetrical and Gaussian. This topic is discussed later in this chapter. Signal The detector output o r signal is of special interest when an analyte is being detected. T h e magnitude of this signal (peak height o r peak area) is proportional to the amount of analyte and is the basis for quantitative analysis. For chromatography, the peak area is most often used. T h e signal specifications to be discussed are sensitivity, minimum detectability (LOD and LOQ), and linearity (linear range and dynamic range). Sensitivity The detector sensitivity S is equal to the signal output per unit concentration o r per unit mass of an analyte, depending on the detector classification. T h e units of sensitivity are based on area measurements of the peaks and differ for the two main detector classifications, concentration and mass flow rate." For a concentration-type detector, the sensitivity is calculated per unit concentrution of the analyte in the mobile gas phase:

where A is the integrated peak area (in millivolts/minute), E is the peak height (in millivolts), C is t h e concentration of the analyte (in milligrams/milliliter), W is the mass of the analyte present (in milligrams), and F, is the MP flow rate in milliliters/minute. T h e resulting dimensions of sensitivity for a concentration detector are millivolts per milliliter/milligram. For a mass flow rate type of detector, the sensitivity is calculated per unit muss of the analyte in the mobile phase:

s = AW - = -EM

(9.4)

where M is the mass flow rate of the analyte entering the detector (in milligrams/second), W is the mass of the analyte (in milligrams), the peak area ( A ) is in ampere-seconds, and the peak height ( E ) is in amperes. In this case, the dimensions of sensitivity are ampere-seconds/milligram o r

292

QUANTITATION: DETECTORS AND METHODS

Minimum detectability = 1X

mg/mL

/

/

I

/

/

I

I I

/

4Dynamic range4-

I

I Upper limit of dynamic range

I

I

Concentration of n-butane in the carrier gas at the detector, (mg/mL)

Figure 9.10. Example of a plot to determine the dynamic range of a TCD. Copyright ASTM. Reprinted with permission.

coulomb/milligram. As noted earlier, the differences in the units of sensitivity between the two types of detector makes comparisons of the sensitivities difficult. Figure 9.10 shows a plot of detector signal versus concentration for a thermal conductivity detector (TCD), a concentration-type GC detector. The slope of this line is the detector sensitivity S according to Eq. (9.3). A more sensitive detector would have a greater slope. It should be noted that the term sensitiuity is often used to express the degree to which a detector can detect small amounts of an analyte. Strictly speaking, that usage is incorrect; the proper term is detectiuity or detection limit (discussed later). While it may be true that a detector with a higher sensitivity will also exhibit a lower detectivity, these two terms represent different concepts and should not be interchanged. Because the range of sample concentrations often extends over several orders of magnitude, this plot is often made on a log-log basis to cover a wider range on a single graph. For example, a UV detector, having a large linear range, can enable 0.1 % impurities to be quantitated simultaneously with the assay of the active ingredient (at approximately 100%). (However, the definition of sensitivity as the slope of the line is based on a linear-linear plot.) As shown at the upper end of the graph (Fig. 9.101, linearity is lost and eventually the signal fails to increase with increased concentration. These phenomena will be discussed further in the section on linearity.

9.1

DETECTORS

293

Minimum Detectability or Limit of Detection (LOD) The lowest point on Figure 9.10, representing the lower limit that can be detected, has been called by a variety of names such as minimum detectable quantity (MDQ), limit of detection (LOD), detection limit (DL), and detectivity. The ICH definition of LOD is “the lowest amount of an analyte in a sample which can be detected but not necessarily quantified as an exact value”’ The IUPAC report on chromatography“) has defined the minimum detectability D , as: D = -2 N (9.5) S

where N is the noise level and S is the sensitivity as just defined. Note that the numerator is multiplied by 2 in accordance with the definition discussed earlier that a detectable signal should be at least twice the noise level, although for many purposes the LOD is taken as three times the S/N ratio. Typical units of detectability are milligram/milliliter for a concentration-type detector and milligrams/second for a mass flow rate type. Chapter 1 contained several computational formulas for LOD or detection limit, DL. The ICH formula” for DL is DL=-

3.3a S

where (T is the residual standard deviation of the data, and S is the sensitivity. There are several similar ways of deriving an estimate of the standard deviation in Eq. (9.6). It can be based on the RSD (relative standard deviation) of the blanks or on the calibration curve, which should contain analyte concentrations in the range of the DL. In the latter case, it can be estimated from the residual standard deviation of the regression line or from the standard deviation of y intercepts of regression lines.” For typical chromatographic data, the regression approach is probably better and obviates the need to run many (about 10) blanks, although it may be desirable to run a separate calibration curve just for LD determination.” It has been shown that the weighted least squares (WLS, as opposed to the ordinary least squares, OLS) method of regression has distinct merit over the classical approach.13 It illustrates that the RSD at the lower end of the curve, near the DL, is less than at the higher end, justifying the use of the regression method. Further discussion of statistics is beyond the scope of this text, but there is abundant 1iterat~re.l~ -I8 An alternative approach is to measure the noise and express the DL as 2 or 3 times the noise and the quantitation level (QL) at 10 times the noise. Or, in the regulated industries, analysts may set their DLs considerably above the levels they know they can detect based on their laboratory experiences, just to ensure that they will comply with regulations.

294

QUANTITATION: DETECTORS AND METHODS

If the LOD is multiplied by the peak width of the analyte peak being measured, using appropriate units, the value that results has the units of milligrams and represents the minimum mass that can be detected chromatographically, for both concentration and mass flow rate detectors. Some call this value the minimum detectable quantity, or MDQ. It is a convenient measure for comparing detection limits between detectors of different types, as suggested earlier. A related term is the limit of quantitation (LOQ), defined as “the lowest amount of an analyte in a sample which can be determined as an exact value.”“’ It should be greater than the LOD and is widely taken as 6 times the signal-to-noise ratio. Alternatively, the ACS guidelines on environmental analysis’’ specify that the LOD should be 3 times the S/N and the LOQ 10 times the S/N. The definitions of the USP are similar and also state that the LOQ should be no less than 2 times the LOD.20 Other agencies may have other guidelines, but all are concerned with the same need to specify detection and quantitation limits, and the relationship between them. Equations and calculations for detectivity used by ICH and EPA were given in Chapter 1.

Linear Range and Dynamic Range The straight line in Figure 9.10 curved off and became nonlinear at high concentrations. It becomes necessary to establish the upper limit of linearity to measure the linear range, which is a fundamental parameter in the method validation process. Since Figure 9.10 is often plotted on a log-log scale, the deviations from linearity are minimized and the curve is not a good one to use to show deviations. A better plot is one of sensitivity versus concentration, as shown in Figure 9.11. Here the analyte concentration can be on a log scale to get a large range while the y axis (sensitivity) can be linear. According to the ASTM specification, the upper limit of linearity is the analyte concentration, corresponding to a sensitivity equal to 95% of the maximum measured sensitivity. The upper dashed line in the figure is drawn through the point representing the maximum sensitivity, and the lower dashed line is 0.95 of that value. Having established both ends of the linear range, the minimum detectivity and the upper limit, the linear range is defined as their quotient: Linear range

=

Upper limit Lower limit

(9.7)

Since both terms are measured in the same units, the linear range is dimensionless. Obviously, a large value is desired for this parameter. Linear range should not be confused with dynamic range, which can be seen on Figure 9.10 as terminating at the point at which the curve levels off

t

E“

\

E

-I

+---------:--k

-

I I I

m

I

5 E v

i

.-> ...t-

rsa e n

------------

--

A

o

b

o

-

sma,

v

0.95S, I

I

i L ; Minimum detectability

range

I

I

I I

Upper limit of linearity

I

and shows no more increase in signal with increasing concentration. The upper limit of the dynamic range will be higher than the upper limit of the linear range, and it represents the upper concentration at which the detector can be used. Summary of Detector Terms

A number of terms defined in the previous sections are also parameters to be measured in method validation. Detection limit, quantitation limit, linearity, range, and specificity are five of the nine terms defined in the ICH guidelines.’ These definitions have not been included in the discussion just concluded, but they are listed in Appendix A, which serves as a useful summary of this section. Robustness is another ICH term, and one that is sometimes confused with ruggedness, which is not in the ICH list. Robustness is “a measure of [a procedure’s] capacity to remain unaffected by small, but deliberate variations in method parameters and provides an indication of its reliability during normal usage.”‘ Some chromatographic parameters that may be deliberately varied are flow rate, temperature, different columns (from different suppliers or lots), and (for HPLC) different MP compositions or pH. On the other hand, ruggedness is the degree of reproducibility of test results obtained by the analysis of the same samples under a variety of normal test conditions such as different laboratories,

296

QUANTITATION: DETECTORS AND METHODS

different analysts, different instruments, different lots of reagents. . . Ruggedness is a measure of reproducibility of test results under the variation in conditions normally expected from laboratory to laboratory and from analyst to analyst.2o

That definition is very close to the ICH definition of reproducibility, a measure of precision, and is clearly different from robustness. Unfortunately, the 2002 IUPAC harmonized guidelines includes in its description of ruggedness the following: “The ruggedness of a method is tested by deliberately introducing small changes to the procedure and examining the effect on the r e s ~ l t s . ” ’Their ~ examples of factors that could be addressed are nearly the same as those for robustness given by ICH as listed above. One could hope that this discrepancy will be resolved in a timely fashion and that true harmonization will eventually be established. In the meantime, current usage in the United States favors the ICH definitions.’” Having introduced the concept of precision, note that it is also one of the nine ICH terms; the final two are accuracy and analytical procedure. Consult Appendix A for a complete list of ICH terms. Definitions and computational formulas of some of these terms can be found in chapters and book on statistics, including undergraduate textbooks on quantitative analysis and analytical chemistry, such as the one by Christian.” 9.2

DATA ACQUISITION AND PROCESSING

A number of issues have to be addressed in setting up, acquiring, and processing of data from the detectors that have been described. Choices have to be made regarding the hardware and software to be used. Whereas strip chart recorders and integrators have been used in the past, modern chromatography data systems (CDS) are computer-based. In the regulated industries, this trend is being driven by the GMP requirements of 21 CFR Part 11 as described in Chapter 1. Seven different data stations and associated software have been summarized and compared.” They acquire analog data from chromatographic detectors, digitize and integrate it, and calculate and report the results. In some laboratories, the CDS is integrated into a laboratory information management system (LIMS), which can also handle data from nonchromatographic instruments. Further information about CDSs and LIMS can be found in McDowalls chapter.’3 Peak Height versus Peak Area

The choice between the use of peak height and peak area is one that may not even be considered, since a CDS provides peak areas with a high degree of accuracy. Area measurements are generally considered to be the fundamen-

9.2

DATA ACQUISITION AND PROCESSING

297

tal measure of the quantity of analyte. They represent the integration of the entire sample in the peak, not just the peak height, the average value. However, if the mobile phase flow rate is not constant, concentration-type detectors show greater errors in area measurements than in peak height measurements, for the reasons given earlier in this chapter. This would be a factor to consider for TC detectors used in PTGC or UV detectors used in HPLC. In summary, peak area is the preferred measurement especially if the detector time constant is too large or there are run-to-run changes in chromatographic conditions that can cause changes in peak height or width (but not area). Some examples of the latter are retention factor, temperature, or sample introduction method. However, peak height measurements are less affected by overlapping peaks, noise, and sloping baselines, so peak heights are often preferred for these situation^.^^

Measurement of Areas The process of measuring peak areas consists of integrating the area under the peaks. Electronic or computer algorithms are a necessity for high-precision work; in addition, they make it possible to handle quantitative measurements where the chromatography is poor. There are several parameters that need to be set to optimize the integration process and assure high accuracy. They are sampling rate, peak width, and peak threshold. The CDS default values are seldom satisfactory and may result in errors. Therefore, the chromatographer should be somewhat familiar with the actual process carried out in the integration process and understand the meaning of these parameters and how to set them. A brief introduction is given here, but Dyson2j can be consulted for a detailed discussion, and Ouchi has provided useful summaries in his column The Data File in LC-GC Mugmine." The sampling rate should provide at least 10 points per peak2'; 20 is better*'; more are unnecessary and simply consume memory space. Therefore, narrow capillary GC peaks should be sampled between 5 and 20 times per second, while once per second may be sufficient for HPLC peaks." If the sampling rate is too small, the digital peak will not provide an accurate representation of the analog data. The raw data may be bunched to get fewer, averaged values2" that are used to detect the start and the end of a peak. In many CDS, this value is set by the peak width parameter. Normally, the peak width selected is the width of an early peak, which is usually the narrowest. If the peaks get much wider during an isothermal G C run or an isocratic HPLC run, the peak width parameter can be set to change during the run.

298

QUANTITATION: DETECTORS AND METHODS

The threshold parameter defines the size of the signal that exceeds the noise level and is used to determine the start and end of a peak. The optimum setting will be low enough to trigger the start of a peak when the signal just barely exceeds the noise but high enough to prevent the noise from being mistaken for a peak. That is, the threshold value should be low for small, wide peaks and higher for large, sharp peaks. The CDS can be used to select an optimum value by monitoring the baseline noise over a short period of time and then selecting an appropriate value. Area measurement results from the accumulation of data between the start and end signals generated according to the above principles. The raw data are used, and the sum is designated as the peak area, usually in millivolts/second or microvolts/second. Note that several figures in this chapter have been labeled “signal” on the y axis, which would denote peak height. However, peak area is equally valid. The highest single value in the accumulated data is designated as the peak height. Narrow peaks are the ones that are most likely to be poorly inte-

Peak sensed. Integration begins.

t

Perpendicular from valley

Baseline drift sensed. Integration not begun

Peak skimmed

Baseline changes

Figure 9.12. Examplcs of some methods of integrating areas of chromatograms with a computcr data station.

9.3

QUANTITATIVE ANALYSIS

299

grated; they are usually the early peaks and those that are obtained by fast chromatography. Some hints for handling these difficult peaks have been published.”’ Sources of Error

Some characteristics of chromatographic data that can cause errors in the integration process are noise, unresolved peaks, and sloping baselines. Noise3’ is often reduced by a process called smoothing, which has become a cornerstone of modern regression methodology.32An introduction to smoothing and a proposal for a new alternative to the popular Savitzky-Golay filter has been published by Eilers.” Katz” and Meyer 2x.33 have published books that include information on errors. A few examples are given here regarding unresolved peaks and baseline problems. The area under unresolved peaks can be separated by dropping a perpendicular from the valley between them as shown in Figure 9.12. Alternatively, a small peak on the tailing edge of a larger peak can be skimmed 08. Baseline changes can be compensated for by forcing new baselines to be used. Proper measurement of peak areas requires considerable attention to such measurement options. When properly done, the error associated with the measurement should be well under 1%. 9.3 QUANTITATIVE ANALYSIS

First of all, a qualitative analysis must prcccdc a quantitative analysis. Perhaps it need not be a complete qualitative analysis, but it is helpful to know as much as possible about the sample. Standards must be available for the analytes to be determined. The sample must be representative, and it must be stable. A good chromatographic separation is highly desirable, and the detection system must have the ideal specifications we have discussed. Decisions have to be made about the number of analytes to be determined in the sample, the level of precision that will be needed, and so on. Standards and Calibration

All quantitative analyses and validations need to be based on standards. Usually, the standards are highly characterized chemicals, often with the same identity as the analytes to be run. Chemical standards must be pure; many are certified by an appropriate agency. The USP lists about 1000 reference standards, and the NIST has over 1200 standard reference materials (SRMs), although few are of use in chromatography. Additional information about standards and their handling was presented in Chapter l .

300

QUANTITATION: DETECTORS AND METHODS

The calibration procedure may vary somewhat depending on the instrument and the method, but basically the process is intended to establish a relationship between the calibration standard and the output signal from the instrument. A linear relationship is most desirable, and this discussion will be limited to that case. However, even if the calibration is not linear, quantitative analysis is still possible, but additional precautions must be taken. Two types of calibration are common, depending on the number of standards run-single point and multiple point. If a single standard is used, prior investigation is necessary to establish that a linear relationship does in fact exist. The single-point calibration calculation is based on the assumption that the linear relationship passes through the origin, so it is necessary to show that a blank produces a zero signal. This type of Calibration is sometimes referred to as the two-point method, even though the second (zero) point might not be confirmed for each analysis. Classification of Methods

Five methods of quantitative analysis will be discussed briefly, proceeding from the most simple and least accurate to those that are capable of higher accuracy. Area Normalization Area normalization is a procedure often used in chromatography when pure standards are not available or when impurity peaks have not even been identified. As the name implies, it is really a calculation of area percent, which is assumed to be equal to the weight percent. If X is the unknown analyte,

where A , is the area of X and the denominator is the sum of all the areas. For this method to be accurate, the following criteria must be met: 1. All analytes must be eluted. 2. All analytes must be detected. 3. All analytes must have the same sensitivity. It is not very likely that all three conditions will be met, but this method is very simple and is often the only choice if some analytes have not been identified or are not available in pure form for standards. It is usually only semiquantitative at best because of the potential differences in response

9.3

QUANTITATIVE ANALYSIS

301

factors for the different compounds in a given sample. When performing analyses by normalization without standards, data can be reported as area percent rather than weight percent to emphasize that they may be semiquanti tat ive. Area Normalization with Response Factors If standards are available, the third limitation can be removed by running the standards to obtain correction factors or so-called relative response factors f , or RRF. One substance (it can be an analyte in the sample) is chosen as the reference, and its response factor f , is given an arbitrary value such as 100. Mixtures, by weight, are made of the reference and the other analytes, and they are chromatographed. The areas of the two peaks-A, and A , for the reference and the unknown, respectively-are measured, and the relative response factor of the unknown, f , , is calculated:

Wr/WR is the weight ratio of the unknown to the reference. Relative response factors of some common compounds have been published for the two most common GC detector^,'^ and a variety of other relative response factors can be found in the literature. However, for the highest accuracy, one should determine his/her own factors. An alternative method is to obtain a calibration curve for each analyte and for the reference. The relative response factor is the ratio of the slope of the analyte curve relative to that of the reference.3s When the unknown sample is run, each area is measured and multiplied by its factor. Then, the percentage is calculated as before: (9.10)

External Standard This method can be performed graphically using multiple-point calibration. Known amounts of the analyte of interest are chromatographed, the areas are measured, and a calibration curve like Figure 9.13 is plotted. The multiple standards used in this method are different concentrations of the same standard. They might be dilutions from a single standard preparation, although an error in preparing the single standard will cause all of the diluted standards to be in error. The signals produced by the standards are plotted versus their concentrations, and a straight line is fitted to the data by the linear regression process.

302

QUANTITATION: DETECTORS AND METHODS

Figure 9.13. Typical HPLC calibration curve. Correlation coefficient = 0.99999. Reprinted with permission from J. Miller and J. Crowther (eds), Analytical Chcrnisty in a GMP Enuironrnent, John Wiley & Sons. Copyright 2000; this material is used by permission of John Wiley & Sons, Inc.

Data can be recorded on spreadsheets that can be used to perform the regression analysis to produce the best straight line and the slope, correlation coefficient (Y), and the residuals. Commonly, the correlation coefficient is used to judge the quality of the data, with 1.0 representing the ideal and 0.9999 a very good calibration. The value in Figure 9.13 is 0.99999. However, it should be noted that in one of its guidelines, the IUPAC concluded that “the correlation coefficient is misleading and inappropriate as a test for linearity and should not be used.”” It appears that this recommendation is being ignored. Standard solutions of analyte vary in concentration, so a constant volume must be introduced to the column for each. This requires a reproducible method of sample introduction; a valve is adequate, but syringe injection in GC is usually inadequate, particularly for syringes that contain sample in the needle. Errors around 10% are common; better results are obtained from autosamplers that inject at least 1 FL. Manual HPLC analyses are usually carried out with sample valves rather than syringes. The fixed-volume loops provide very constant volumes, thereby making the external standard method suitable for HPLC. Single-point methods can be used too, where a ratio is established between one standard and the unknown. Such is the case for some data systems. A calibration mixture prepared from pure standards is made by weight and chromatographed. Absolute calibration factors, equal to the mass or moles per area produced, are stored in the data system for each analyte. When the unknown mixture is run, these factors are multiplied times the respective areas of each analyte in the unknown, resulting in a value for the mass of each analyte. Since this procedure is a one-point calibration, as

9.3

QUANTITATIVE ANALYSIS

303

compared to the multipoint curve described before, it is somewhat less precise. Note also that these calibration factors are not the same as the relative response factors used in the area normalization method. lnternal Standard This method and the next one are particularly useful with techniques such as chromatography, which often are not too reproducible from day to day, and in situations where one does not want to recalibrate often. It does not require exact or consistent sample volumes or response factors since the latter are built into the method. The standard chosen for this method cannot ever be a component in a sample; a known amount of this standard is added to each sample; hence the name internal standard. It must meet several criteria:

1. It should elute near the peaks of interest. 2. But, it must be well resolved from them. 3. It should be chemically similar to the analytes of interest and not react with any sample components. 4. Like any standard, it must be available in pure form. The standard is added to the sample in the same concentration as the analyte(s) of interest and prior to any chemical procedures (extractions, derivatizations, or other reactions). If many analytes are to be determined in one sample, several internal standards may be used to meet the preceding criteria. One or more calibration mixtures are made from pure samples of the analytes, depending if a one-point calibration is desired or if a graph is to be plotted (see discussion under the external standard method). A known amount of internal standard is added to the calibration mixture(s) and to the unknown. Usually, the same amount of standard is added to each, and this is often done most conveniently by volume. All areas are measured and referenced to the area of the internal standard, either by the data system or by hand. If multiple standards are used, a calibration graph as shown in Figure 9.14 is plotted where both axes are relative to the standard. If, as shown in the figure, the same amount of internal standard is added to each calibration mixture and each unknown, the abscissa can simply represent concentration, not relative concentration. The unknown is determined from the calibration curve or from the calibration data in the data station. In either case, any variations in conditions from one run to the next are canceled out by referencing all data to the internal standard. A simple calculation example has been published”7 that will be of help to those performing this technique for the first time. In principle, this method should produce better accuracy, but some exceptions have been r e ~ o r t e d . ~ ’

304

QUANTITATION: DETECTORS AND METHODS

Concentration

Figure 9.14. Example of calibration plot using the internal jtandard method

Standard Addition Method In this method the standard is also added to the sample, but the chemical chosen as the standard is the same as the analyte of interest. It requires a highly reproducible sample volume, which is a limitation with syringe injection in GC, as noted earlier. The principle of this method is that the extra signal produced by the addition of standard is proportional to the original signal. Equations can be used to make the necessary calculations, but the principle is more easily seen graphically. Figure 9.15 shows a typical standard addition calibration plot. Note that a signal is present when no standard is added; it represents the

- 3 - 2 - 1

0

1

2

3

4

Amount of standard added Figure 9.15. Example of a calibration plot using the standard addition method.

9.3

QUANTITATIVE ANALYSIS

305

original concentration, which is to be determined. As increasing amounts of standard are added to the sample, the signal increases, producing a straightline calibration. To find the original “unknown” amount, the straight line is extrapolated until it crosses the abscissa; the absolute value on the abscissa is the original concentration. In actual practice, the situation is more complex, and a thorough summary has been provided by Bader.’8 For example, different calculations are required when the total volume is kept constant and when it varies as standard is added. Matisova and co-workers’9 have suggested that the need for a reproducible sample volume can be eliminated by combining the standard addition method with an in situ internal standard method. In the quantitative analysis of hydrocarbons in petroleum, they chose ethyl benzene as the standard for addition, but they used an unknown neighboring peak as an internal standard to which they referenced their data. This procedure eliminated the dependency on sample size and provided better quantitation than the area normalization method they were using. The standard addition method can also be used with only one standard. A ratio is formed from which the unknown concentration can be calculated:

(9.11) where A stands for the measured response and C the concentration, and the subscripts refer to the analyte (XI, the total combined analyte plus standard (TI, and the standard ( S ) . Concluding Comments

The basic principles governing all quantitative analytical determinations should be followed in performing any of the analyses discussed. For example, each sample should be run in duplicate or triplicate along with the standards and the blanks, throughout a run. Some of the regulation agencies have specific requirements, but some decisions are left to the experience of the chromatographer. The suggestions in Chapter 1 for system suitability as recommended by the FDA4” are good guidelines to follow. One possible sequence of the samples to be run is given in Table 9.2, but there are many variables to consider and this is not a universal recommendation. For further discussion on system suitability criteria, consult reference 41. Chromatographic results can be very precise, down to about 0.1% RSD in the ideal case and are typically less that 0.5%. However, for all quantitative methods the precision and accuracy decrease as the concentration of the analyte decreases.

306

QUANTITATION: DETECTORS AND METHODS

Table 9.2 One Possible Sequence for a Quantitative Analysis

Sample

Number of Injections

Blank Sample to demonstrate LOD Critical resolution mixture Reference solution Sample 1 Sample 2 Sample 10 Reference solution Sample 11

2 2 2

Reference solution Etc.

2

REFERENCES 1. D. W. Neyer, K. M. Hahnenberger, and C. G. Bailey, A m . Lab. 2003, 35(24), 11-15. 2. U. D. Neue and T. Gilby, Am. Lab. 2003, 35(5), 56-59. 3. D. N. Heiger, High Performance Capillary Electrophoresis-An Introduction, Hewlett-Packard, Waldbronn, Germany, 1992, Publication # 12-5091-6199E. 4. J. Inczedy, T. Lengyel, and A. M. Ure, Compendium of Analytical Nomenclature, 3rd ed., Blackwell, UK, 1998. Also available online at www.iupac.org/publications/analytical compendium /. 5. G. A. Gross and A. Gruter, J . Chromatogr. 1992, 592, 271. 6. G. I. Ouchi, LC-GC 1996, 4, 472-476. 7. V. R. Meyer, Pructical High-Peifonnance Liquid Chrornutograpy, 2nd ed., Wiley, West Sussex, England, 1994, pp. 309-310. 8. E. L. Johnson and R. Stevenson, Basic Liquid Chromatography, Varian, Palo Alto, CA 1978, p. 278. 9. ICH Harmonised Tripartite Guideline, Text on Validation of Analytical Procedures, Q2A, Fed. Reg. 1995, 60, 11260-11262. 10. L. S. Ettre, Pure Appl. Chem. 1993, 65, 819-872 11. ICH Harmonised Tripartite Guideline, Guideline on the Validation of Analytical Procedures: Methodology; Availability, Fed. Reg. 1997, 62, 27463-27467. 12. D. L. MacTaggart and S. 0. Fanvell, Am. Enuiron. Lab. 1998, I0(6), 4-5. 13. J. R Burdge, D. L. MacTaggart, and S. 0. Fanvell, J . Chem. Educ. 1999, 76, 434-439. 14. L. Currie, Pure Appl. Chem. 199, 67, 1699-1723. 15. I. Krull and M. Swartz, LC-GC 1998, 16, 922.

REFERENCES

307

16. L. A. Currie, Anal. Chim. Acta 1999, 391, 127-134. 17. M. Thompson, S. L. R. Eleison, and R. Wood, Pure Appl. Chem. 2002, 74, 835-855. 18. J-P. Antignac, B. Le Bizec, F. Monteau, and F. Andre, Anal. Chim. Acta 2003, 483, 325-334. 19. D. MacDougall et al., Anal. Chem. 1980, 52, 2242-2249. 20. United States Pharmacopia, USP 27/NF22, U.S. Pharmacopeial Convention, Rockville, MD, 2004. 21. G. D. Christian, Analytical Chemistry, 6th ed., Wiley, Hoboken, NJ, 2004, Chapter 3. 22. D. Noble, Anal. Chem. 1995, 67, 617A-620A. 23. R. D. McDowall, in Analytical Chemistry in a GMP Environment, J. M. Miller and J. B. Crowther (ed), Wiley, New York, 2000, Chapter 14. 24. W. Kipiniak, J. Chromatogr. Sci. 1981, 19, 332. 25. N. Dyson, Chromatographic Integration Methods, 2nd ed., Royal Society of Chemistry, Cambridge, England, 1998. 26. G. I. Ouchi, LC-GC 1991, 9, 474-477 and 628-633. 27. E. Katz (ed), Quantitative Analysis Using Chromatographic Techniques, Wiley, New York, 1987, p. 43. 28. V. R. Meyer, in Aduances in Chromatography, Vol. 35, P. R. Brown and E. Grushka (ed), Marcel Dekker, New York, 1995. 29. G. I. Ouchi, LC-GC 1995, 13, 714. 30. J. V. Hinshaw, LC-GC No. Am. 2002, 20, 34-38. 31. G. I. Ouchi, LC-GC 1996, 14, 472-476. 32. P. H. Eilers, Anal. Chem. 2003, 75, 3631-3636. 33. V. R. Meyer, Pitfalls and Errors of HPLC in Pictures, Huthig, Heidelberg, Germany, 1997. 34. W. A. Dietz, J. Gas Chromatogr. 1967, 5, 68. 35. J. B. Crowther, in Analytical Chemistry in a GMP Environment, J. M. Miller and J. B. Crowther (ed), Wiley, New York, 2000, Chapter 15. 36. J. A. Magee and A. C. Herd, J. Chem. Educ. 1999, 76, 252. 37. P. Haefelfinger, J. Chromatogr. 1981, 218, 73. 38. M. Bader, J. Chem. Educ. 1980, 57, 703. 39. E. Matisova, J. Krupcik, P. Cellar, and J. Garaj, J. Chromatogr. 1984, 303, 151. 40. FDA, Analytical Procedures and Methods Validation, Guidance for Industry document, Food and Drug Administration, Rockville, MD, 2000. (Available online at FDA site.) 41. J. W. D o h , LC-GC No. Am. 2004, 22, 430-435

This Page Intentionally Left Blank

10 CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS)

It has been mentioned several times in this book that chromatography is not ideal for qualitative analysis. As will be described in Chapter 12, the retention parameters such as retention time or retention factor do serve to characterize a given solute and can be used for making qualitative identification when compared with standards. However, for positive confirmations of the identities of unknown components in a mixture, spectral measurements are required. The most commonly used spectral method is mass spectrometry (MS). The combination of a chromatographic instrument interfaced to a mass spectrometer is often referred to as a hyphenated method and designated as GC-MS or GC/MS. Such combinations have been achieved between the mass spectrometer and the following chromatographic systems: GC, HPLC, SFC, TLC, and CEC. They are so important and commonly used that they are given special attention in this chapter. Multiple hyphenation, as in GC/MS/MS, also common and sometimes called hypernation, will be discussed later in this chapter. These combinations can be considered to be multidimensional as well, and that topic is covered in Chapter 15. An alternative way of describing these combinations is simply to consider the mass spectrometer to be the detector attached to the chromatographic system. Thus, in the discussions of detectors in Chapter 9, the mass spec-

Chromatography: Concept.5 and Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

309

310

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

trometer was included and sometimes called the MSD, or the mass selective detector, in addition to the designation MS. No significance is attached to either concept in this chapter, which is mainly concerned with GC/MS and LC/MS. At the end of the chapter, some discussion will be included on the other hyphenated MS methods and other chromatographic combinations such as those with Fourier transform infrared spectrophotometry (FTIR). First, we will look at the considerations involved in interfacing the chromatograph with the mass spectrometer and briefly summarize the current status of GC/MS and LC/MS. This will be followed by a short section introducing mass spectroscopy to chromatographers. Conventional chromatographic columns are normally operated so that the column exits are at atmospheric pressure, although it is possible to run a G C with a vacuum outlet.' By comparison, mass spectrometers must be run under high vacuum. Thus, the process of joining the two involves an interface that can take a chromatographic effluent at atmospheric pressure and reduce it to a vacuum condition (about lo-" torr). As will be seen, this process is different in GC and HPLC but has been ideally achieved in both cases. It is much easier to couple a GC effluent to an MS than an HPLC effluent, which contains a liquid mobile phase and often a buffer, both of which have to be removed or enriched or exploited (solvent-assisted ionization). Consequently, GC/MS was the first hyphenated technique to be successfully produced (19591, and it has become a relatively simple and cost-effective instrumental method. Bench-top instruments have become smaller, and mass production has allowed the costs of instruments to decrease. Both parts of the instrument as well as all data are handled by computer, and these too have improved and become less expensive. Many books are available covering GC/MS from the chromatographer's point of view.'-' The past few years have seen major changes and improvements in LC/MS. Interfacing was initially more difficult, but a major innovation was the realization that interfaces could also serve as the ionization devices. This field is now in a growth spurt and has been adapted to handle biological samples that were previously considered impossible to be run by MS. Today LC/MS plays a major role in biotechnology due to its capability to perform rapid, high-resolution separations and identifications of large biomolecules. See Chapter 15 for more information. While LC/MS itself is not new and is adequately covered in the encyclopedia volume on detectors for LC,' the current literature is full of new applications,' and books are appearing that cover the new advances and applications.',

'

10.1 BASICS OF MASS SPECTROMETRY

In a mass spectrometer, operated under high vacuum, analytes are ionized, sometimes fragmented, and then directed to a mass analyzer where they are

10.1

Chrom. Effluent

*

Ion Source

-

BASICS OF MASS SPECTROMETRY

0 e t

Analyzer

-

311

Computer

t 0

r

Fore Pump

Figure 10.1. Major parts of a mass spectrometer.

separated according to their mass-to-charge ratios, m / z . The ion current generated is plotted versus m / z ratios to produce a mass spectrum, which is characteristic of the original analyte and can be used for both qualitative and quantitative analysis. Usually the z value of the ions is + 1, so the m / z ratio actually is the same as the mass for that ion. The parts of a mass spectrometer are shown diagrammatically in Figure 10.1 and schematically for a GC/MS in Figure 10.2. The ionization chamber, analyzer, and detector are under high vacuum of 10-4-10-8 torr. Vacuum is achieved with a diffusion or turbo molecular pump, backed by a rotary fore pump. This chapter will present a cursory review of a mass spectrometers ion source and mass analyzer; detailed books on MS are available."' Ion Sources

The first stage is the ion source for the ionization of analytes. The most common methods for achieving ionization are shown in Table 10.1. Both positive and negative ions are produced in most sources, but the negative ions are the ones typically collected and used.

Electron lmpact (El) Electron impact is the oldest and most common method because GC/MS has been the most popular configuration. It is still the most popular ion source for GC/MS, but newer methods have been introduced for LC/MS. In the EI source, electrons are emitted from a heated filament (usually tungsten at 70 eV> and bombard the vaporized analyte molecules, causing ionization and the loss of an electron. In the following examples, A H will be used to represent an analyte, rather than A, because some reactions involve the loss or gain of a hydrogen. AH+e--+AH++2e-

(10.1)

ELECTRON BEAM

Figure 10.2. Schernatlc dlagram of a GC/MS with data system. Copyrlght 1992 Agilent Technologies, Inc. Reproduced with permission.

10.1 BASICS OF MASS SPECTROMETRY

313

Table 10.1 Some Common MS Ion Sources Upper Mass Limit (Daltons)

Ionization Method

Molecular ion, M or Fragmentation

Used with GC

1,000 1,000

Electron impact ionization(E1) Chemical ionization (CI) Negative (NCI) Positive (PCI)

F M

U.wi with HI'L C Particle beam Thermospray ionization (TSI) Electrospray ionization (ESI) Atmospheric pressure ionization (API) Atmospheric pressure photoionization (APPI) Atmospheric pressure chemical ionization (APCI)

10,000 2,000 200,000 10,000 10,000 10,000

Other

Fast atom bombardment (FAB) Matrix-assisted laser desorption ionization (MALDI)

10,000 500,000

F M

The ion AH ' has the m / z ratio of the molecular mass of the analyte and is called the molecular ion. Depending on its structure and the amount of energy absorbed by this ion, it may proceed through the analyzer unchanged or it may break apart into smaller fragments. Usually, many fragments are formed and EI is considered to be a hard type of ionization. Some of the fragments will be ionic and some free radical; the ions proceed to the analyzer and the radicals are pumped off in the vacuum. Figure 10.3 shows a typical EI spectrum. The peaks are scaled to the largest ion, which is called the base peak. This ionization process had been thoroughly studied, and fragmentation patterns are understood well enough to permit the identification of many organic compounds,",'* but that subject is beyond the scope of this discussion and is information that is really not needed by chromatographers who usually achieve identifications from available databases, to be discussed later.

Chemical lonization (Cl) Chemical ionization is an indirect method of forming ions. A reagent gas such as methane, isobutane, or ammonia is introduced into the reaction chamber and is ionized, similar to Eq. (10.1). These ions then collide with other reagent gas molecules, causing further ionization by the transfer of either a proton or a hydride ion. These reactions occur at a higher pressure (1-10 torr). For methane reagent gas, some

314

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

.

"';

% I00

70

60 50

10

M-CWHP

0 20

30

40

50

60

70

80 mte

90

100

110

Figure 10.3. Typical EI mass spectrum (ethyl sec-butyl ether). Reprinted from R. Silverstein Spectrometric Identification of Organic Compounds, 6th ed., Wiley. Copyright 1997, John Wiley, Inc. This material is used by permission of John Wiley & Sons, Inc.

reactions are CH,-+ CH,

(10.2)

H,

(10.3)

CH,'+ CH,

4

C H , + + CH,

-+

C,H,'+

This ionized reagent gas then collides with analyte molecules, ionizing them by similar processes such as the following: C H , + + AH 4CH, C,H,'+

AH

-+ C,H,

+ AH,'

(10.4)

+A'

(10.5)

CH,'+AH+CH,+AH+

(10.6)

The process, first patented in 1965, is less energetic than EI and produces less fragmentation. Such methods are often referred to as soft ionization as compared with EI which is hard. The molecular ion (A', AH', or A H 2 + in the above examples) is more prominent and is useful in assigning the molecular weight to an analyte species. Negative ions are also produced by other, similar reactions. CI instruments can be used to detect either the positive ions as shown in this example (called PCI) or negative ions (called NCI). Further information on chemical ionization is readily available.13 ionization Methods for HPLC The particle beam method used in HPLC is an electron impact method, similar to the EI method in GC. As can be seen in the diagram in Figure 10.4 this method requires that the sample be

10.1

BASICS OF MASS SPECTROMETRY

315

Figure 10.4. Schematic of particle beam LC/MS ionization system. Copyright 1988 Agilent Technologies, Inc. Reproduced with permission.

somewhat volatile, so it is not applicable to the large polar molecules like proteins. On the other hand, for smaller molecules, it offers the advantages that many EI reference spectra are available for library searching, and it can be readily switched between use with LC and GC. Like some other early attempts to interface LCs with MS, particle beam is not used much due to insufficient sensitivity and limited range of appli~ability.'~Other methods listed in Table 10.1 have replaced it. Rather than attempting to remove the LC mobile phase, they nebulize and vaporize it and use it in the ionization process. This solvent-assisted ionization occurs at atmospheric pressure or slightly below atmospheric, which is quite different from classic ionization, which is done at moderate vacuum (lop4 torr). The most common methods are thermospray ionization (TSI), electrospray ionization (ESI), and atmospheric pressure ionization (API). Since they are all designed to take the HPLC column effluent directly, no other interfacing is necessary; they also serve as the interfaces between the LC and the MS. A schematic of a thermospray interface is shown in Figure 10.5. The MP from the column needs to contain a volatile buffer such as ammonium

316

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

Figure 10.5. Schematic of a thermoyxay interface. Reprinted with permission from LC-GC, Vol. 14(3), March 1996, p. 236. LC-GC is a copyrighted publication of Advanstar Communications Inc. All rights reserved.

acetate so that it can participate in a chemical ionization mechanism after the MP is heated, vaporized, and sprayed into the chamber, as shown. The process is not much different from CI described above for GC/MS. Although this method has some drawbacks, it is suited for use with reversedphase chromatographic systems at typical HPLC flow rates. The ESI and API methods are more reliable and are replacing TSI for most applications. They both operate at atmospheric pressure using similar configurations like the one shown in Figure 10.6. The MP is forced through a capillary needle with nebulizing gas under a voltage gradient so that a spray is produced. In ESI, analyte ions are formed aided by an organic acid such as formic acid, which is present in the MP. This chemical ionization process is soft and produces AH,' and A - ions. These ions are transported into an intermediate pressure zone (about 1 torr) through a heated capillary and a skimmer, as shown. The interface shown in Figure 10.6 also has an octopole focusing lens system before the ions enter the quadrupole analyzer. One drawback of this method is the limitation on flow rate of 2 0.2 mL/min. Low flows can be achieved by using an effluent splitter or a microcolumn. A major advantage is its ability to form multiply charged ions that can have m / z ratios from 1 to 25. Figure 10.7 is one example, myoglobin, where m / z ratios are indicated for each of the peaks detected. The MS software includes an algorithm to make these peak assignments. Each m / z ratio is L

L

(10.7)

where MW is the molecular mass of the compound, and z is the number of

Figure 10.6. Schematic of an electrapray triple-quadupole tandem MS. Reprinted with permission from LC-GC, Vol. 14(3), March 1996, p. 236. LC-GC is a copyrighted publication of Advanstar Communications Inc. All rights reserved.

31 8

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

Figure 10.7. Positive ion electrospray ionization mass spectrum of myoglobin. Reprinted with permission from LC-GC, Vol. 12(12), Dec. 1994, p. 914. LC-GC is a copyrighted publication of Advanstar Communications Inc. All rights reserved.

protons added and hence also the total charge. For example, the peak at about 1700 has a charge z of + l o , so MW calculates out to be 16,990 or 1.70 x l o 4 to three significant figures. The MW calculated for the other peaks is the same. The book edited by Colels gives more details including applications. The API process is similar, but a corona discharge at the tip of a needle at high voltage (3-6 kV) is placed a few centimeters beyond the spray tip to produce the ionization. The analytes with the highest proton affinities are most readily ionized, making it more selective than ESI. Unwanted cluster ions are also formed, causing a high background. Methods for removing them include the insertion of a gas cloud or membrane between the ion source and the analyzer or the promotion of collision-induced dissociations (CID) in this region by adjustment of pressures and accelerating voltages." API can be used at normal HPLC flow rates (up to 2 mL/min), the same as for TSI, which is an advantage over ESI. Analyzers The second stage shown in Figure 10.1 is the mass analyzer. The ions formed in the first stage are separated in this stage. The main methods used are

10.1 BASICS OF MASS SPECTROMETRY

319

Table 10.2 Some MS Analyzers

Analyzer Type Sectors Magnetic

Electric Quadrupole Ion trap Time-of-flight (TOF) Fourier transform ion cyclotron resonance (FT-ICR)

Upper Mass Range 104

lo4 lo5 2 lo5 2 104

Resolution 10'-105

102-104 103-104 103- 104 104-10h

listed in Table 10.2. For use in chromatographic systems, analyzers need to be able to scan through a wide range of masses in a very short time so that many scans (at least 10) can be made on each chromatographic peak. The magnetic sector methods scan too slowly for GC peaks, so most GC/MS work has been done with quadrupoles and ion traps. More recently, and especially for LC work, the time-of-flight (TOF) analyzer has been rediscovered" and found to be ideal because of its faster scanning speed-several hundred spectra per second compared with 10 spectra per second for quads and ion traps. Quadrupoles and /on Traps A schematic of the quadrupole analyzer was included in the GC/MS instrument shown in Figure 10.2. The ions formed in the ion sources are separated in the quadrupole analyzer according to their m / z ratios by adjusting the DC and R F voltages impressed on the four poles shown. Only the one ion with the correct ratio is able to pass through the poles without being neutralized, arriving at the detector at the appointed time in the scan. The quadrupole analyzer is suited to use with chromatographs because of its fast scanning capability, but its upper limit is reached for peaks narrower (width at base) than 100 ms. More information can be found in instrumental textbooks or GC/MS books such as reference 4. The ion trap analyzer is based on the same principle as the quadrupole, but the geometry is entirely different (it is a monopole) as shown in Figure 10.8. Its operation is even more mysterious than the quadupole because multiple collisions between analyte and other atoms and ions occur in a single confined space (the trap). Eventually, at a given KF voltage applied to the ring electrode, ions of only one m / z ratio will be ejected through the end cap to the detector, so a scan of voltages will provide a separation of ions. More about the ion trap in a study in Analytical Chemistly18and the book edited by March and Todd.19 Time of Flight The principle of the TOF analyzer is in its name; separation is based on the time it takes for each ion to traverse a fixed distance (through

320

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS)

I

I

Filament

RF Ring Electrode

End

1 Ion Signal

A

RFVOLTAGE

I

INITIAL

LENS C

ION SIGNAL

I

D

TIME

t

MASS SPECTRALPEAKS

I

IONIZE

+

I

I

SCAN ION TRAP

IONIZE

SCAN

Figure 10.8. Schematic diagram of an ion trap detector. Traces A and B show R F and DC voltages required to produce a mass spectrum. Traces C and D show the various times involved. Rcprinted from H. Hill and D. McMinn (eds), Detectors for Cupilluty Chrornutogruplzy, John Wiley & Sons. Copyright 1992 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.

the space called a drift tube) to the detector, based on its mass (assuming equal charges of +1). All ions acquire the same kinetic energy in the ion source, so their flight time is inversely proportional to the square root of the m / z ratio. Characteristics of Analyzers Two critical characteristics of analyzers are the mass range over which they can be operated and the resolution they can provide. Mass ranges are included in Table 10.2.

10.1

n

BASICS OF MASS SPECTROMETRY

321

fl+Afl 1.00H

R

\

-

"

I

M :an

\ -.___ Afl -

__-'

0.10H 0.05H

I

(B)

Figure 10.9. Definition of resolution for mass spectrometry: ( u ) the 10%.valley definition and ( h ) the 5% height definition. Reprinted from H. Hill and D. McMinn (eds), Deiectors for Cupillmy Chrornutogruphy, John Wiley & Sons. Copyright 1992 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.

The concept of resolution in MS is similar to the one given in Chapter 2 for chromatographic systems, but the definition is different and the resolution values have a different meaning. In chromatography, a resolution of 1.5 indicates a baseline separation. In MS, a resolution of 500 indicates that an ion of mass 500 is baseline separated from the ion of mass 501. Figure 10.9 shows an example; the actual equation used for the calculation is

M R s = m

(10.8)

where M is the mass of the peak and A M is the distance between peaks as shown in the figure. The difference, A M , can be measured at 5 , 10 (shown in the figure), or 50% of the peak height, which can cause some confusion. If the difference is measured near the base of the peak, and the peaks are of the same width and just separated at that point, then A M would also

322

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

represent the width of peak M . For some analyzers, like quadrupole, A M is a constant, unlike the chromatographic systems.

Other MS Topics It has been traditional to distinguish two levels of resolution in mass spectrometry-low resolution (up to 500 or 600) and high resolution (up to los or 10‘). High resolution can be obtained by combining two or more analyzers in one instrument, called a double-focusing MS. Traditionally these have been combinations of sector instruments with a quadrupole lens between the st ages. A different type of combined MS has been called tandem MS or MS“, where n denotes the number of stages. In these instruments, one ion from the first stage is isolated and regionized, and secondary or daughter ions are formed that are then separated in the second stage. Usually the same type of analyzer is used in each stage, for example, quadrupole/quadrupole. In a triplequadrupole tandem, which has been popular since 1980, the second stage is not used to separate ions but rather as a collision cell in which the second stage of ionization occurs. Hence, tandem MS produces data like product-ion scans, which are different from conventional MS scans. Tandem MS instruments in which the first quadrupole stage is replaced by a double-focusing stage are called hybrids. A review of hybrids has recently appeared in Analytical Chemistry.”’ Nine different instruments are described. Some of these hybrids can be used for high resolution and also for daughter production. They have resolutions up to 800,000 and cost between $300,000 and $2 million. Data from a conventional mass spectrometer can be collected in two ways, total ion (TIC) and single ions, called selected-ion monitoring (SIM). In TIC (also called scan mode), all ions detected are displayed as one signal, so the output looks just like a chromatogram from any other chromatographic detector. Any peak in this TIC can be examined to see what ions it contains, and this information is used primarily to identify unknowns. By comparison, the SIM output is set to display only those ions (of a given m / z ratio) that have been chosen for collection. So, for example, if the mass of a particular analyte or analyte fragment is chosen as the single ion, the only peaks in the chromatogram will be for peaks that contain that ion. If a characteristic fragment ion was chosen, all analytes that fragment to give that ion will appear in the chromatogram. Up to six ions can be acquired using conventional MS software. SIM is more selective and sensitive than TIC. Consider the difference between the acquisition of data over a range that covers all possible ions, say 43 to 500, for a TIC, and one that collects data for only 6

10.2

GAS CHROMATOGRAPHY/MASS SPECTROSCOPY

323

ions, for a SIM. The former collects data for 76 times as many ions (457/6), so much more time is spent on each ion in SIM, increasing its sensitivity. Consequently, SIM is used primarily for quantitative analyses, often at trace levels. The actual identification of an unknown analyte can be accomplished by matching its mass spectrum from the TIC with reference spectra available in the MS software. The largest MS database, containing over 130,000 spectra, is available from NIST. Using algorithms in the software that compare the most intense MS peaks (usually 8 in number) with those in the library, matches can be made and possible identification presented on the computer screen along with an indication of the quality of the matches. Smaller, specialized libraries are also available for specific types of compounds such as drugs or EPA pollutants. Most libraries contain spectra obtained with EI ionization, so they are not suitable for making matches with spectra obtained by other ionization methods. Those ionization methods that produce mainly molecular ions (like CI) are not usually a problem, of course, and EI and CI data are complementary. Mass spectroscopy also provides the opportunity to examine a peak for its purity. Often a chromatographic peak that looks symmetrical may in fact be composed of two or more compounds, and nonsymmetrical peaks can arise from nonresolved peaks. An examination of spectra from the front of a peak, apex of the peak, and tail of the peak can reveal if different compounds are present in one peak. If differences are found, all of the scans taken within that peak can be examined to see if it is possible to extract spectra that represent mainly one compound or another, and then identification can be attempted. Another unique characteristic of MS is the possibility to use an isotope of a given compound as an internal standard (IS) in a quantitative analysis. The mass spectra will differ, but the isotopes need not be separated chromatographically since they are easily separated spectroscopically. This method is sometimes referred to as the isotope dilution method of quantitation, and some feel it is the ultimate IS.

10.2 GAS CHROMATOGRAPHY/ MASS SPECTROSCOPY

Open tubular (OT) columns up to 0.32 mm i.d .are well suited for use in a GC coupled to a MS because of their low flow rates, typically 1-2 mL/min. The type of interface called direct is used with all OT (capillary) columns. The end of the OT column is extended from the G C directly into the ion source of the MS. The GC flows are low enough, and the vacuum pumping

324

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

high enough that the necessary vacuum required by the mass spectrometer can be easily maintained. In this arrangement, the outlet of the GC column is operated at vacuum. This is not common, but it has been shown that vacuum operation can, in fact, be advantageous.' Previously, with packed columns, more elaborate interfaces were necessary to enrich the GC effluent in the analytes. Semipermeable membranes and sintered glass were used, and the most popular devices carried the names of their inventors: Watson-Biemann, Llewellyn-Littlejohn, and Ryhage." Today, GC/MS has become almost as routine as G C itself, and a GC/MS system is an essential part of most GC research labs. For the analysis of volatile samples, there is no better instrument. LC/MS is another story.

10.3 LIQUID CHROMATOGRAPHY/ MASS SPECTROSCOPY

In HPLC/MS, the ion sources also serve as the interfaces: TSI, ESI, and API are the most popular. The concept of sampling at atmospheric pressure seemed impossible only a few years ago. Table 10.3 compares these three modes of operation in more detail and Figure 10.10 shows the mass ranges for applications as high as 10'. Flow rate considerations are important for ESI operation. Recent studies describing applications of LC/MS and tandem LC/MS/MS have reported the reasons for their differing preferences of ion sources: ESI in one case2' and API in another.23These are just two examples of which there are many in the current literature. In adapting an HPLC method to LC/MS operation, several changes in MP composition need to be considered. These include: Increasing the volatility of the buffer Changing the pH Adding adducts, such as Na'

Table 10.3 A Comparison of LC/ MS Interfaces

Feature Source pressure Optimum flow rate (mL/min) Detection limit, full scan (ng)" Minimum mass range (daltons) Optimum HPLC mode

TSI

ESI

API

1-1000 mtorr

Atmospheric 1-2"

Atmospheric

1-2

5-10 120

RP

1-10 80 RP

"In the ESI interface, the column effluent in typically split 1/20th to l / l O O ' h interface; actual flows through the ESI interface are approximately 20 pL/min. 'SIM (selected Ion Monitoring) can increase scnsitivity by 1000-10,000 fold.

1-2 1-10 150

RP

ahead of the

10.3

LIQUID CHROMATOGRAPHY/MASS SPECTROSCOPY

325

Electrospray Thermospray and Atmospheric-pressure chemical ionization

Aonpolar 10'

103

102

104

105

Molecular weight

L

Figure 10.10. Application ranges of LC/MS interfacing techniques. Reprinted with permission from LC-GC, Vol. 14(3), March 1996, p. 236. LC-GC is a copyrighted publication of Advanstar Communications Inc. All rights reserved.

The MP needs to be volatile. Reversed-phase methods are usually ideal because the major components, water and methanol or ACN, meet this requirement. However, when buffers are used in RPLC, they also need to be volatile. Ammonium acetate was one example mentioned earlier, but others are listed in Table 10.4. Trifluoroacetate (TFA) often forms ion pairs with basic analytes, thereby making them neutral and decreasing their MS dete~tibi1ity.l~Mixing the column effluent with a 3 : 1 mixture of propionic acid:2-propanol will break the ion pairs prior to MS analysis, yielding a 10-50 times improvement in signal. The rate of flow of MP needs to be relatively low to mate with the MS, so columns of reduced diameter are usually preferred. Table 10.4 Volatile Ions for LC/MS Buffers

Cations Ammonium Diethylammonium (DEA) Triethylammonium (TEA)

Anions Acetate Carbonate Formate Heptafluoroacetate Trifluoroacetate (TFA)

326

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

The applications suited to LC/MS often contain many peaks, whereas HPLC has lower plate numbers than GC. The solution is often to use LC/MS”, which is more costly and has restricted the number of such instruments in use. Further discussion of LC/MS can be found in Chapter 15, which includes material on biological applications. 10.4 OTHER HYPHENATED METHODS

Other chromatographic methods that have been coupled with MS include TLC, SFC, CEC (and CZE). SFC is handled much like GC, and CEC much like HPLC. TLC is quite different, of course, and requires special techniques. Details can be found in Poole’s book.2s In general, it would be expected that SCF/MS should use interfaces such as GC/MS since the supercritical fluids become gaseous when reduced to atmospheric pressure, but the interface conditions are more severe because of the higher (critical) pressure. O T columns, because of their low pressure drops, are favored as they are in GC. The two interfaces that have been used are a direct fluid injection (DFI) and a molecular beam apparatus. DFI has been used with packed columns” and with OT columns,27 using both chemical ionization and electron impact ionization. For a more complete discussion of both interfaces, see the chapter on SFC in the ACS Symposium Series edited by Ahuja2’; the review on LC/MSZYalso contains considerable information about SFC/MS. Coupling CEC to MS uses interfaces similar to LC/MS, but CEC presents some special problems. Maintaining the voltage gradient, low flow rates, and incompatible buffers are a few of the problems. Consequently, CEC/MS has not yet become very popular. Although TLC is the subject of the next chapter, an introduction to the technique was given in Chapter 6. Since the separated analytes are contained on the thin layer (of silica, e.g.1, the methods described earlier will clearly not be applicable for interfacing TLC to MS. The reviews of Wilson3’ and Poole” summarize the literature. The MP is easily evaporated from the TLC plate and is not a problem, but getting the analytes off the plate and into a mass spectrometer requires specialized sampling techniques like those used for desorption from solids such as FAB and MALDI. Some of the interfaces being used are described in references 32-35. Until recently, one could only purchase a hyphenated instrument for either G C or LC. Now, at least one instrument company is producing an instrument that can be switched between GC/MS and LC/MS.36 Undoubtedly others will follow. A product review of benchtop GC/MS instruments provides more complete inf~rmation.~’

10.4

OTHER HYPHENATED METHODS

327

Other spectroscopic instruments that have been interfaced to chromatographs are FTIR and NMR. In the case of GC/FTIR, large samples are desired because they facilitate detection by IR, but the vapor state analytes are not easily sampled by IR. A light pipe has been designed that can be heated to keep the analytes in the vapor state and also has the following requirements of a good IR cell for use with GC: small volume, long path length, and high transmission. A typical light pipe is 50 cm x 1 mm in size and has a reflecting gold coating. Its description and details of its use including the data handling requirements have been discussed.3x The other interface is a type of matrix isolation in that the analytes are frozen on a rotating band that moves into the IR beam. A comparison of these two types has been published.3y The liquid from an HPLC is compatible with normal IR sampling and is less of a problem. However, LC mobile phases may not be transparent in the IR, and water is a particularly difficult solvent to handle. For volatile organic solvents, evaporation is possible, and the remaining nonvolatile analytes can be deposited into KBr and pressed into pellets. Further discussion can be found in r e f e r e n ~ e . ~ " Liquid chromatography/mass spectroscopy and LC/FTIR provide different data and are complementary; IR is particularly advantageous for identifying isomers that cannot be distinguished by MS. MS offers the possibility of quantitative analysis by the isotope dilution method, which makes it uniquely attractive. Mixed tandem combinations such as GC/FTIR/MS are also possible since IR is nondestructive; it is possible to combine the three instruments into one. The special requirements and some applications have been described.4' Other tandem combinations are possible, of course. One common one is HPLC/UV/MS.

Liquid Chromatography/ Nuclear Magnetic Resonance

Nuclear magnetic resonance does not have the low detectability limits of the other spectroscopies used in tandem with chromatography, so it has not been used as e~tensively.~' Also, the high magnetic field of current NMR instruments requires that the chromatograph be located at a considerable distance from the NMR probe. There are only a few reports of its use with GC, and, for HPLC, special glass flow cells are needed. For proton studies, the mobile phase should ideally have no protons, so for reversed-phase work, deuterium oxide is used. Other deuterated solvents are quite expensive. LC/NMR is finding use in the pharmaceutical industry for impurity analysis.43At least one commercial instrument offers LC/MS/NMR.

328

CHROMATOGRAPHY WITH MASS SPECTRAL DETECTION (GC/MS AND LC/MS )

10.5 SUMMARY

Mass spectrometry has been effectively coupled to both GC and HPLC instruments, making combined instruments that are very valuable for both qualitative and quantitative analysis. Recent improvements in both fields, but especially in MS and the associated computer control and handling data, have opened new opportunities, especially for the analysis of large biological molecules.

REFERENCES 1. C. A. Cramers, G. J. Scherpenzeel, and P. A. Leclercq, J . Chromatogr. 1981, 203,

207. 2. M. McMaster and C. McMaster, G C / M S : A Practical Users Guide, Wiley-VCH, New York, 1998. 3. H. J. Hubschman, Handbook of G C / M S : Fundamentals & Applications, Wiley, New York, 2001. 4. M. Oehme, Practical Introduction to GC-MS Analysis with Quadrupoles, Wiley, New York, 1999. 5. F. Kitson, B. Larsen, and C. McEwen, Gas Chromatography and Mass Spectrometry: A Practical Guide, Academic, New York, 1996. 6. J. B. Crowther, T. R. Covey, and J. D. Henion, in Detectorsfor Liquid Chromatography, E. S. Yeung (ed), Wiley, New York, 1Y86, Chapter 8. 7. W. M. A. Niessen, J . Chromatogr. A 1999, 856, 179-197. 8. B. Ardrey, Liquid Chromatography-Mass Spectrometry: An Introduction, Wiley, Hoboken, NJ, 2003. 9. W. M. A. Niessen, Liquid Chromatography-Muss Spectrometry, 2nd ed., Marcel Dekker, New York, 1999. 10. See, e.g., E. D e Hoffmann and V. Stroobant, Mass Spectrometry: Principles and Applications, 2nd ed., Wiley, New York, 2002. 1 1 . R. M. Smith, Understanding Mass Spectra: A Basic Approach, 2nd ed., Wiley, Hoboken, NJ, 2003. 12. R. M. Silverstein and F. X. Webster, Spectrometric Identification of Organic Compounds, 6th ed., Wiley, New York, 1997. 13. A. G. Harrison, Chemical Ionization Mass Spectrometry, C R C Press, Boca Raton, FL, 1983. 14. D. A. Volmer and D. L. Vollmer, LC-GC 1996, 14, 236-242. 15. R. B. Cole (ed), Electrospray Ionization Mass Spectrometry: Fundamentals, Instrumentaion, and Applications, Wiley, New York, 1997. 16. E. C. Huang, T. Wachs, J. J. Conboy, and J. D. Henion, Anal. Chem. 1990, 62, 713A-725A. 17. M. M. van Deurson, J. Vens, H. Janssem, P. A. LeClerg, and C. A. Cramers, J . Chromatogr. A 2000, 878, 205-213.

REFERENCES

329

18. S. A. McLuckey, G. J. Van Berkel, D. E. Goeringer, and G. L. Glish, Anal. Chem. 1994, 66, 689A-696A and 737A-743A. 19. R. E. March and J. F. J. Todd (eds), Practical Aspects of Ion Trap Mass Spectrometry, CRC Press, Boca Raton, FL, 1995. 20. K. Cottingham, Anal. Chem. 2003, 75, 315A-319A. 21. See, e.g., W. H. McFadden, J . Chromatogr. Sci. 1979, 17, 2. 22. L. Bonnington, E. Eljarrat, M. Guillamon, P. Eichhorn, A. Taberner, and D. Barcelo, Anal. Chem. 2003, 75, 3128-3136. 23. Y. Hsieh, K. Merkle, G. Wang, J-M. Brisson, and W. A. Korfmacher, Anal. Chem. 2003, 75, 3122-3127. 24. R. L. Cunico, K. M. Gooding, and T. Wehr, Basic HPLC and CE of Biiomolecules, Bay Bioanalytical Laboratory, Richmond, CA, 1998. (Also distributed by Varian: www.uariuninc.com). 25. C. F. Poole, The Essence of Chromatogruphy, Elsevier, Amsterdam, Netherlands, 2003. 26. J. B. Crowther and J. D. Henion, Anal. Chem. 1985, 57, 2711. 27. R. D. Smith and H. R. Udseth, Anal. Chem. 1983, 55, 2266; R. D. Smith, H. R. Udseth, and H. T. Kalinoski, Anal. Chem. 1984, 56, 2971. 28. R. D. Smith, B. W. Wright, and H. R. Udseth, in Chromatography and Separation Chemistry, S. Ahuja (ed.), American Chemical Society, Washington, D.C., 1986, pp. 260-293. 29. T. R. Covey, E. D. Lee, A. P. Bruins, and J. D. Henion, Anal. Chem. 1986, 58, 1451A. 30. I. D. Wilson, J . Chromatogr. A 1999, 856, 429-442. 31. C. F. Poolc, J . Chromatogr. A 2003, 1000, 963-984. 32. G. J. Van Berkel, A. D. Sanchez, and J. M. E. Quirke, Anal. Chem. 2002, 74, 6216-6223. 33. F-L. Hsu, C-H Chen, C-H Yuan, and J. Shiea, Anal. Chem. 2003, 75, 2493-2498. 34. L. S. Santos, R. Haddad, N. F. Hoehr, R. A. Pilli, and M. N. Eberlin, AnulChem. 2004, 76, 2144-2147. 35. I. Meisen, J. Peter-Katalinic, and J. Muthing, Anal. Chem. 2004, 76, 2248-2255. 36. T. L. Sheehan, Am. Lab. 2002, 34(18), 40. 37. R. Mukhopadhyay, Anal. Chem. 2004, 76, 213A-216A. 38. P. R. Griffiths J. A. de Haseth. and L. V. Azarraga, Anal. Chem. 1983, 55, 1361A. 39. J. F. Schreider, J. C. Demirian, and J. C. Stickler, J . Chromatogr. Sci. 1986, 24, 330. 40. R. S. McDonald, Anal. Chem. 1986, 58. 1906. 41. C. L. Wilkins, Science 1983, 222, 291. 42. K. Albert, J . Chromatogr. A 1999, 856, 199-21 I . 43. J. C. Lindon, J. K. Nicholson, and I. D. Wilson, J . Chromatogr. B 2000, 748, 233.

This Page Intentionally Left Blank

I1 LIQUID CHROMATOGRAPHY ON PLANE SURFACES There are two popular LC techniques in which the stationary phase is not packed into a column but rather exists as a planar surface: paper chromatography (PC) and thin-layer chromatography (TLC). The development of PC preceded TLC by some 10-15 years, and a large number of excellent separations were devised for it. But beginning about 1956, it was found that TLC could also be used for most of these separations and that it was faster, more reproducible, more versatile, more convenient, and more efficient (higher resolution). As a result, most laboratories have abandoned the use of PC with its large cumbersome glass chambers. Those who have not continue to use PC because of its lower cost and/or flexibility. 11.1

PAPER CHROMATOGRAPHY

The paper used for chromatography can be common filter paper (Whatman #1 is popular), but chromatographic grades are also available. The formats can vary widely from large sheets to cylinders to circular sheets as shown in Figure 11.1. The large sheets, necessary for good separations and shown in Figure l l . l a , must be developed by a descending flow of mobile phase since the solvents will not flow the required distance against gravity. Although very common at one time in the pharmaceutical industry, their use has been Chromatography: Concepts and Contrasts, Second Editioti. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

331

332

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

Figure 11.l. Developing chambers for planc chromatography: (0)descending-used with PC; ( b ) ascending-used with TLC and PC; ( c ) sandwich-used with TLC; ( d ) horizontal-uscd with paper as shown, but also adaptable for HPTLC.

discontinued. The smaller sheets shown in Figure 11.1b develop by ascending flow and need to be supported at the top or formed into stable cylinders. Also popular is the format shown in Figure 1l.ld where a circular sheet of filter paper is placed over a Petri dish with a wick cut out of it for solvent transfer. The theoretical aspects related to PC have been summarized by Stewart,' and Smith2 has provided a teachers' guide in the same volume. Because undried paper contains a significant amount of adsorbed water, PC can be classified as an LC method and the separation mechanism as absorption (partition). Other liquids can be applied to change its characteristics. For example, silicone oils, petroleum jelly, paraffin oil, and rubber latex are used as nonpolar phases. Special papers are also commercially available that contain adsorbents or ion exchange resins or are specially treated (e.g., acetylated) or are made of other fibers (e.g., glass, nylon). Paper chromatography still finds some use at the precollege level and for elementary demonstrations,3 for the reasons listed before. More details on PC can be found in the monograph by Sherma and Z ~ e i g but , ~ the rest of this chapter will be devoted to TLC.

11.2

THIN-LAYER CHROMATOGRAPHY

333

11.2 THIN-LAYER CHROMATOGRAPHY

The early work in TLC was performed manually in nearly the same way as PC. The difference was that the stationary phase (SP), usually silica gel, was coated on carriers such as glass, aluminum, or plastic instead of being a sheet of paper. Because the SP was polar, the chromatographic mode was classified as normal phase. The TLC plates were placed in glass chambers (smaller ones than those used for PC), as shown in Figure 1 l . l b and developed by an ascending mobile phase (MP). The lack of commercial plates necessitated the time-consuming and inefficient coating of one’s own plates. Today the situation is much improved; the simple, manual procedure is still used and is even the method of choice for many USP procedures. Still, for many scientists, the main use of TLC is for qualitative analysis and screening, and it is performed in a fashion that could be described as quick and dirty. On the other hand, many of the steps in the TLC procedure have been instrumented with the result that it is less labor intensive, is also capable of producing data that have better reproducibility and quantitation, and can be handled with modern data systems. These efforts began in Europe but are being implemented worldwide. So, in order to describe TLC accurately, the following material is divided into two sections: The first, called manual TLC, will include the basic concepts and procedures, and the second, called instrumental TLC, will describe some of the mechanical devices and instrumental components used to automate TLC. The early TLC literature covers the manual procedures while most modern publications include both the manual and instrumental methods. In the latter category are a study by Poole and Poole,’ an encyclopedia chapter,‘ and a very comprehensive and up-todate handbook.’ See also the Selected Bibliography at the end of this chapter. Manual TLC

The steps in TLC include the following: Sample application Development (the name used for the actual running of the chromatographic process) Visualization or detection of the separated analytes Quantitation Each of these will be described briefly. As in all chromatographic processes, the sample should occupy as small an area as possible on the bed. Solutions of the sample can be applied as spots or bands at one end of the bed using micropipets or microsyringes. In neither

334

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

case should the application of sample disturb the bed, so that for soft TLC plates, the sampling device cannot touch the surface. The harder layers on commercial plates are preferred for this reason. A template can facilitate the process. The sample spots should be dried before development, so the solvent is usually chosen for its volatility as well as its ability to dissolve the sample. Note that the solvent does not need to be the mobile phase (MP) or have any relationship to the MP, unlike the situation in column LC. To keep the spots small, sample can be applied repetitively to the same area, allowing the previous application to dry before reapplication. In general, the placement of the spots must be far enough from the end of the bed to prevent them from dipping into the solvent reservoir (1-2 cm). The prepared plate is developed in a closed, presaturated chamber usually made of glass and large enough to accommodate the conventional 20-cm X 20-cm plates. Figure 11.16 is the most common style, sometimes referred to as a normal or N chamber. A saturation pad wet with mobile phase is inserted to presaturate the chamber, especially if the mobile phase contains volatile solvents, and all chambers are covered, in an attempt to keep the vapor saturated with MP. Obviously, when the cover is removed to insert the plate, some vapors escape and the chamber is no longed fully saturated. Development time varies and is about 30 min for conventional normal-phase TLC. After the development is completed and the plate dried, the analyte bands must be located unless they are colored. A comprehensive list of detection reagents is available in Volume I1 of the Handbook of Chromatography edited by Zweig and Sherma.' If the analytes fluoresce, they can be located under a UV lamp. If not, the use of a phosphor in the TLC plate may make it possible to locate them as nonfluorescing (quenching) spots on the phosphorescing plate. For qualitative identification, the data obtained by planar techniques are reported as R , values, which were defined in Chapter 2 and discussed in Chapter 6. Alternatively, the distances migrated can be measured relative to a standard run on the same plate, which usually is more reproducible. The colors caused by selective visualizing reagents can also be used for identification. The spots can be quantitated by measuring them manually in situ. Two methods are commonly used. The first method involves a visual estimate of the quantity of analyte by comparing its spot size with the sizes of standards that bracket it. The semiquantitaive result is reported as being between two of t h e standards and provides an accuracy of about 25%. Somewhat greater accuracy can be achieved if the spot sizes are actually measured and plotted, assuming that the square root of the spot area is approximately proportional to the log of the weight of the analyte.

11.2 THIN-LAYER CHROMATOGRAPHY

335

Alternatively, spots can be scraped off the TLC plate and extracted to remove the analytes, which can then be quantitated by techniques such as UV/visible spectrophotometry. Stationary Phase The most popular thin layer is silica gel, and it is estimated that about 90% of TLC separations are performed on conventional TLC silica plates. The silica gel used in TLC has the same properties as that used in LSC in columns, and the discussion about silica in Chapters 3 and 8 is relevant. In brief, the silica has a heterogeneous energy surface and many very active silanol groups. It picks up water from the atmosphere very readily and will preferentially adsorb the most polar component in a mobile-phase mixture. Most of the discussion about LSC in Chapter 8 is also applicable to TLC. One difference is that, in TLC, binders in amounts up to 15%, are usually used to produce a stable layer and good adherence to the backing plate. Calcium sulfate (gypsum, designated with a G in the name, e.g., silica gel G ) and proprietary compounds related to polyvinyl alcohol (PVA) are most common. The PVA plates are very stable and will withstand rather rough handling. While it is not difficult to make ones own plates, most laboratories prefer commercially prepared plates, especially those made with PVA. It must not be forgotten, however, that both binders may modify the adsorption properties of the silica and can produce somewhat different separations. Binders may also be the cause of poor mass-transfer properties exhibited by many stationary phases, especially at high flow rates, thereby making the plates less efficient.’ To facilitate visualization of samples, a phosphor can be added to the plates during preparation. Most plate manufacturers use the designation F (e.g., silica gel F) to identify these plates. When they are viewed under 254-nm UV light, most analytes appear as dark spots against a phosphorescent background and can be located. Here, too, the phosphor becomes a part of the SP, but the inorganic compounds (typically Mn-activated zinc silicate) used are present in small amounts and usually do not alter the separations. In use, TLC plates are often activated by heating above 100°C for an hour or more. Plates containing organic binder (PVA) should not be heated above 150°C. Dried plates are stored in desiccators to keep them dry and clean. Unfortunately, the plates are exposed to the atmosphere during the time required to apply the sample and get them into the system. Depending on the particular plate and the humidity, large amounts of water from the atmosphere can be adsorbed on the plate during this time, changing its extent of activation; some plates can absorb more than half their equilibrium concentration of water from a room atmosphere of 50% relative humidity within 3

336

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

min. As a result, many chromatographers use commercial TLC plates directly from the box. Other chemicals used as stationary phases include alumina, cellulose and cellulose ion exchangers, polyamide, magnesium oxide, and Kieselguhr. Silver nitrate can be added to silica gel to retain olefins selectively. The chapter by Rabel in the Handbook of Thin-Layer Chrornatographylo can be consulted for further information on the stationary phases in TLC. As would be expected, when the nonpolar bonded phases were developed for column use, they were also made into TLC plates. The operation of TLC in a reversed-phase mode is not new, but the bonded materials make it more practical and extend the possible applications, including ion pair chromatography. Some difficulty can be experienced with hydrophobic bonded layers in TLC because they may result in the inability of an aqueous mobile phase to move up the plate. This phenomenon was discussed earlier in t h e section on bonded phases (Chapter 8). If the mobile phase does not wet the plate, the capillary action will be absent. For example, some C,, bonded plates require mobile phases containing less than 25% water, although there are now C,, plates (e.g., Merck RP-18W) that are partially modified and allow development with MPs up to 100% water. Theory predicted that smaller particles would improve efficiencies in LC: columns, and, when tested and found to be true, resulted in HPLC as we have seen. So, small particles (about 5 p m ) were also tried in TLC. As in column LC, these plates resulted in higher performance and the technique has been named high-performance TLC, HPTLC. However, unlike column HPLC, HPTLC has not taken over the field, and the older version of TLC also remains popular. Some characteristics of these two types of plates are included in Table 11.1. The conventional plates have plate numbers (N)of several hundred, but the newer HPTLC plates have risen to more than a thousand. They are thinner than regular plates, the development proceeds to only about half the distance on regular plates, and the separations are faster, as shown in Figure 11.2. They require smaller samples, and streaking has been found to be the best method of sample application. A third type of plate is a small one, made on microscope slides for use in fast screening. These slides are easily prepared from a slurry of silica gel in a Table 11.1

Comparison of Types of TLC Plates

Type

Thickness

Regular Microscope slides High performance Preparative

250 p m 250 p m f 150 p m 0.5-2 mm <

Particle Size

Sample Size

10-12 p m

1pL 1pL 50-500 nL 150 p L (band

10-12 p m 5-6 p m 5-40 p m

2

<

application)

337

11.2 THIN-LAYER CHROMATOGRAPHY

Separation distance (solvent migration) [mm]

Running time [mrn] 2

4

I

6

I

8

I

1

I

I

20

0

f

30

Figure 11.2. Comparison of HPTLC and conventional TLC. Reproduced with permission from the Camag Bibliography Service, November 1985, Camag Scientific, Inc., Muttenz, Switzerland.

mixture of methanol and methylene chloride, but they are also available commercially. Directions are available in older references such as Snyder ” as well as Peifer’s original work.” A fourth type of TLC plate has a thick layer for preparative work, and all four are compared in Table 11.1. Preparative TLC is discussed later in the chapter. Several special layers are commercially available. For example, one end of the plate can contain a preadsorption layer, which is a zone of solid that will not retain the sample. When the sample is spotted in this zone, it will be carried by the mobile phase to the junction with the silica layer so that it arrives there in a minimum volume as a small sample. Preadsorbent plates allow larger sample volumes to be applied diffusely and quickly while resulting in tight, band-shaped width zones. Figure 11.3 shows the cross section of a preparative plate where this preadsorption layer has been combined with a wedge shape for the active layer. The wedge shape will facilitate good separations of larger samples by allowing the analytes to

Preadsorbent

Barrier

Adsorbent

Glass Figure 11.3. Cross section of tapered prep TLC plate. Courtesy of Analtech. Patent No. 4,348,286, Sept. 7, 1982.

338

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

spread out in the third dimension, thus keeping the zone spreading minimal along the axis of migration. Mobile Phase Usually, the liquid mobile phase is a mixture of liquids chosen by consulting the published literature (see the applications section later in this chapter) and optimized by trial and error because results can be obtained so quickly and easily. The principles of liquid mobile-phase selection, begun in Chapter 4 and continued in Chapter 8, are, of course, relevant. Following Snyder's work with the selectivity triangle, reported in Chapter 8, Niredy's group proposed a procedure called Prisma for the selection of normal-phase solvents" for TLC. Further details, including the extension into reversed-phase systems have been published.6, l 4 Application of these principles is useful only for more sophisticated work such as the forced-flow systems discussed in the next section, and for manual TLC they represent overkill. For normal-phase systems on silica gel, a nonpolar liquid is modified with a more polar one, and small amounts of a third component such as acetic acid are often added to deactivate the plate slightly and decrease tailing. The amount of water is critical in determining the activity of the plate, as already indicated. Many times the chosen mobile phase is simply a mixture that works, and a typical example is a mixture of butanol-acetic acid-water. For reversed-phase systems, a polar mixture like those used in column LC are satisfactory, and combinations of water and acetonitrile or methanol are common. Remember that some hydrophobic stationary phases will not have sufficient capillarity for largely aqueous mobile phases, as was described in the last section. Manual TLC is usually run isocratically, and gradient elution will be discussed later. Because the isocratic mobile phases are usually mixtures of solvents, continued use of a mobile phase will probably result in preferential evaporation of the most volatile component, thus changing its composition. Demixing can also occur on the bed if the MP components have differing polarities since the MP is moving on a dry bed and the most polar components will be the most strongly adsorbed. For these reasons, MPs are never reused. Nature of Mobile-Phase flow The theory of chromatography has been treated in earlier chapters, but the planar methods have some special characteristics that need further discussion. In manual TLC, the flow of the MP is not controlled as it is in the column methods and in the forced-flow TLC methods discussed later. It is dependent on the surface tension y and the viscosity 7 of the MP as well as on the nature of the stationary bed. The

11.2

THIN-LAYER CHROMATOGRAPHY

339

Weight mobile liquid Weight stationan/ solid

13 Fraction of distance to front

Figure 11.4. Variation in solvent concentration along direction of flow.

prevailing model considers the bed to be composed of interconnected capillaries of varying diameter. Initially the bed is dry; the liquid is applied at one end and is drawn up the bed against the pull of gravity by capillary action. As a result, the solvent front moves faster than the bulk mobile phase. A homogeneous bed is preferred to minimize the variation in capillary flow across it. Nevertheless, one would expect that the quantity of liquid on the bed would vary, decreasing from the reservoir to the solvent front. Figure 11.4 shows this phenomenon (and it also shows that the gradient is more pronounced in PC than in TLC, reflecting the greater heterogeneity of paper compared to a thin layer of silica). The distance the MP front moves, sf,has been shown to be proportional to the square root of the migration time t : St =

fi

(11.1)

where the proportionality constant k is directly proportional to the surface tension and inversely proportional to the viscosity:

(11.2) where d, is the particle diameter and 0 is the contact angle (nearly 0 for TLC, so that cos f3 is unity).

340

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

The velocity of the solvent front, u , , is

k uf=2s,

(1 1 . 3 )

and is proportional to the surface tension of the mobile phase and inversely proportional to its viscosity and the distance the front has moved. Thus, as noted earlier, the solvent velocity is not constant and decreases the further the solvent has migrated. Eventually, of course, the velocity goes to zero (for ascending methods), which puts an upper limit on the distance the solvent can migrate and determines the maximum size for TLC plates. Not only is the flow not constant, but it cannot be easily controlled; the rate is determined by the solvent and the nature of the bed, and the result may not be optimum for good chromatography. These deficiencies can be overcome if the MP flow is controlled as described in the section on instrumental TLC. Since the mobile phase is moving through a dry bed, several other undesirable effects occur. The adsorption of the first liquid (at the front) on the stationary phase is exothermic, causing the front to have a higher temperature than the rest of the system. Since the temperature of the system is not usually controlled but is allowed to assume the ambient value, some evaporation may occur at the solvent front. If the solvent is composed of a mixture of liquids, preferential evaporation of the most volatile one will cause a slight change in the solvent composition. In fact, the adsorption of a mixed mobile phase will probably also cause some changes in composition because the most polar component will be preferentially sorbed. The situation can become so severe that solvent demixing can occur. At best, a mixed solvent mobile phase is probably not uniform across the planar bed, and some temperature differentials probably exist as well. Evaluation of Manual TLC Manual TLC is used for fast screening of mixtures and for surveying methods for possible use in HPLC. The advantages of TLC in Table 11.2 show why this is true. The advantage listed last is of special importance: All sample analytes are present and in evidence on a TLC plate. Because it often occurs that some analytes migrate very little if at all, they will remain at the point of sample application. However, in column LC, they remain at or near the head of the column and are never eluted or detected and may even invalidate a quantitative analysis (when area normalization is used). They accumulate and contaminate the head of the column. The advantage in TLC is that these analytes can be detected. In this way, TLC can serve as a complementary method to HPLC and can aid in HPLC method development.

11.2

THIN-LAYER CHROMATOGRAPHY

341

Table 11.2 Advantages and Disadvantagesof Manual TLC

Advantages Fast Simple, inexpensive apparatus High sample throughput using multiple samples per plate Variable shapes of plates and SPs Two-dimensional TLC is easy Detection and quantitation are static All analytes are on plate and detectable

Disadvantages Flow is not constant Flow cannot be controlled easily Temperature and solvent gradients can exist Not always predictable or in agreement with theory Limited plate numbers Not automated; labor intensive Less accurate quantitation

The main disadvantages of manual TLC can be eliminated by instrumentation and automation as summarized in the next section. Instrumental TLC then becomes competitive with column HPLC, and this comparison will be addressed. Instrumental TLC

Many aspects of TLC have been instrumented or automated," but, as yet, there is no such thing as a TLC instrument. Each of the steps in the process has been instrumented (sample applicator, development chamber, scanning of a developed plate, data handling) but all of them have not been combined into one instrument. Some examples of each step will be discussed. The purpose of instrumenting TLC is to make the process more reproducible and efficient and remove the tedium of manual processing. Each step in the overall process can be improved, but improvements in one aspect affect the others, so each and every step ought to be improved to maximize the effects. As is often the case in chromatography, the sampling step is the most crucial, and that is true in TLC also. Sample Application For the purposes of quantitation, not only must sample areas be kept small but the volumes of sample solutions delivered to the plate must be accurately known and controlled. Mechanical sample spotters are commercially available that can apply exact volumes of multiple samples to chosen locations. One example is the Nanomat,16 which can be used with a variety of samplers that deliver their total volume when touched to a plate, still requiring manual operation. Fully automatic samplers are also available that can be used with microsyringes to deliver exact amounts from 0.1 to 5 p L .

342

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

Band application is used in analytical and in preparative TLC by automatic devices such as the Camag Linomat.I6 This process has the same objectives and requirements as spotting, and in general, it results in better separations and is preferred for quantitative analysis. For preparative work, the bands can be applied over the entire bed (no empty spaces between spots), permitting larger amounts to be applied. Development Chambers Manually operated development chambers suffer from the fact that it is impossible to saturate the vapor space with MP vapor since the chamber lid has to be removed to insert the plate. There are automated development chambers that solve this problem by allowing some time between the insertion of the plate and its placement into the MP reservoir. After saturation is achieved, the plate can be lowered into the MP to begin the development. The development process is monitored by a charge-coupled device (CCD) sensor and can be automatically stopped at a predetermined distance or time. Other automated procedures that can be run on conventional chambers include multiple development and automated multiple development ” in which the plate is developed several times in the same direction with the same or different solvents and dried between developments. Spot concentration occurs with each development, keeping it small and improving separations. This process could be considered to be a form of step gradient analysis. Sample chambers are available that have different geometries from the N chamber discussed earlier. One type is configured like a sandwich and is called an S chamber (see Fig. 1 1 . 1 ~ )It . has less vapor space and is more easily saturated. More important, the S chamber can also be ouerpressurized allowing the flow rate of MP to be controlled, although for that application it is usually run in a horizontal orientation. It is called overpressured layer chromatography or optimum performance laminar chromatography (OPLC), and it is one example of forced-flow planar chromatography (FFPC). There are two other types of FFPC: one uses centrifugal force to cause MP flow (rotation planar chromatograpy, RPC), and the other uses a voltage gradient as in capillary electrochromatography (CEC), and it is usually called electroplanar chromatography (EPC). Using FFPC effectively removes the limitations on the ability to control the chromatographic process in TLC, which has kept manual TLC from developing into an efficient process that could rival HPLC. FFPC modes also offer the possibility of eluting the analytes from the plate and detecting them online as an alternative to detection on the plate. Or, to use the terms defined in Chapter 6, FFPC can be performed in either of two modes, one that detects analytes as bands and one that detects them 3’‘

343

11.2 THIN-LAYER CHROMATOGRAPHY

Planar Technique

TLC

I

I

Mode

Manual

Instrumental

MP Flow

Capillary Forces

Forced Flow (FFPC)

I

1

External Force

I

Detection (Visualization)

I

I

I

Offline

Pressure OPLC

Online

Offline

Centrifugal RPC

Online

Offline

Electroosmotic EPC

I

Offline

Figure 11.5. Classification of TLC operating modes.

as peaks. Figure 11.5 summarizes the different types of TLC according to these different operating parameters. The theory of FFPC is, of course, similar to that discussed in Chapter 8, but it is summarized in TLC terms in the Sherma and Fried Handbookix Brief discussions of the FFPC techniques are included in several studies.s,6. 1s. II) Of the three methods, OPLC is the most popular and it is included in this chapter. However, interest continues in both EPC2” and RPC,” the latter having been coupled online to a mass spectrometer. There is one parameter used in the planar techniques that has not yet been defined in this text. It was suggested by Martin in 1949 as a way to relate an analyte’s structure and its free energy to its chromatographic performance. It is called R , and its definition is

( 11.4) By comparing R , values of compounds that differ by only a methylene group or a functional group, a table of R , values can be compiled for individual functional groups. Since they are proportional to the free energies for these groups, they should be additive, and the total R , value for any analyte can be predicted by summing the values for the groups of which the

344

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

molecule is composed. This idea is known as the Martin equation:

(11.5) Some workers prefer to report TLC data as R M values rather than R , values. Further discussion, and comparisons with other TLC models have been published.” Overpressured Layer Chromatography Bryson and Papillard 22 have written an up-to-date and comprehensive introduction to OPLC. Briefly, the procedure is: the TLC layer is covered with a flexible sheet in a horizontal S chamber and is pressurized to remove any vapor space above the plate. It is fitted with connections to permit mobile phase to be forced across the plate at a constant and controlled rate, and it can also be configured for direct attachment to a detector to provide online operation as mentioned above.

T LC

forced flow

HPTLC forced flow

10

j 0

I

I 2

I

I I 6 8 Solvent Front Migration (crn) I

4

I

I

I 10

I

I

12

Figure 11.6. Comparison between capillary flow and forced flow TLC and HPTLC. Reprinted from Poole and Poole, “Multidimensionality in Planar Chromatography,” J . Chromatoy. A , 1995, 703, 573-612. Copyright 1995, with permission from Elsevier.

11.2 THIN-LAYER CHROMATOGRAPHY

345

The TLC layer can be circular (radial flow) or rectangular (linear flow). Although an apparatus for OPLC has been manufactured in Budapest 23 for nearly 20 years, another one from France is currently available from Bionisis in the United States.24 Excellent details are given in the chapter on OPLC in Sherma and Fried's Handbook.2s Poole and Poole2' have published van Deemter type of graphs showing comparisons between capillary flow and forced-flow performance (Fig. 11.6) for regular and high-performance (HPTLC) plates. The use of forced flow clearly lowers the plate height, especially for the HPTLC plates. This increase in efficiency was the main objective in instrumenting TLC for forced flow. In addition, analysis times can be decreased, and gradient elution can be used.27 The latest on FFPC methods is included in two recent reviews published together in the special review issue of the Journal of Chromatography A.y3228 Poole9 notes that a few laboratories use FFPL methods and that chambers capable of attaining higher pressures would be desirable, and Nyiredy2' describes all three FFPC methods, including three recent trends in OPLC multilayer OPLC, serial connection of OPLC plates, and predictions based on method comparisons. It appears that the future of FFPC is uncertain and probably depends on the availability of adequate instrumentation.

Densitometry Accurate and precise quantitation of TLC separations is greatly improved with the use of instrumental methods of which densitometry is the most popular. A typical densitometer," which could also be used for scanning electrophcrograms has the following operating characteristics: Reflectance or transmission Absorbance or fluorescence Accommodates plates up to Wavelength range: 190-800

modes measurements 20 X 20 cm nm

Multiwavelength scanning, up to 31 channels Computer controlled and data processed

Full spectra available for qualitative analysis Further details on the use of densitometers is available in the references given.', 29 Typical precision is 1-3% relative standard deviation (RSD). Recent progress in TLC detection methods, including diode-array scanners and

346

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

Table 11.3 Similarities and Differences Between TLC and HPLC

Similarities Retardation parameters ( R and R,) are similar Optimum performance is achieved at same retardation parameters: For TLC, R , = 0.25 For HPLC, k = 3 (equivalent to R = 0.25) Both conform to rate theory (when TLC flow is controlled).

Differences 1. TLC has the advantages listed in Table 11.2. 2. HPLC more easily automated. 3. HPLC shows higher performance, especially at high MP velocities. 4. Most popular modes of operation

are opposite, NP for TLC and RP for HPLC

image analyzers, is discussed by Nyiredy.28 Image analysis and in situ scanning mass spectrometry are two of the topics in Poole’s review.’

Comparisons with Column Methods In some ways the techniques of TLC and HPLC are complementary. For example, TLC is usually run in normal mode and HPLC in reversed mode. Table 11.3 lists some of their similarities and differences. Chapter 6 contained a comparison of retention data showing good agreement in retardation factors, R for HPLC and R , for TLC when both systems had identical mobile and stationary phases. It is important that the comparison be made between equivalent terms, retardation factors. Unfortunately. many workers compare HPLC retention factors ( k ) with TLC retardation factors ( R F )which , is like comparing oranges to lemons and does not provide a valid comparison. Furthermore, optimum resolution in TLC is obtained when R , is about 0.25, and it is interesting to note that an R , of 0.25 corresponds to a retention factor k of 3.0, which is in the optimal range found for column work. We can conclude that both methods provide similar, if not exactly equal, retention parameters and that their kinetic effects can both be described by the rate theory (Chapter 3). When they differ, the differences may be due to the fact that TLC proceeds on a dry bed and HPLC does not or that the TLC binder has had a strong influence on the separation. However, when the van Deemter type of curves are compared (Fig. 11.71, it can be seen that the LC column methods provide better efficiencies (lower plate height), better mass-transfer properties (smaller C term), and higher operating velocities. Add to this the fact that HPLC can be fully automated, and you have the primary reasons why HPLC is the more popular technique for most analytical applications. By contrast, TLC is most popular for its low cost, simplicity, and flexibility.

347

11.3 OTHER TOPICS

0

I 10

I 20

I

30

I

40

I

50

Reduced Velocity

Figure 11.7. Plot of reduced plate height against reduced mobile-phase velocity for a typical HPTLC plate and HPLC column. Reprinted from C. Poole and S. Poole, “Multidimensionality in Planar Chromatography,” J . Chromufogr. A , 1995, 703, 573-612. Copyright 1995, with permission from Elsevier.

11.3 OTHER TOPICS

Preparative TLC

The characteristics of preparative plates were included in Table 11.1, and one special preparative plate was depicted in Figure 11.3. Thicker layers permit the application of larger amounts of sample. The spots can be scrapped off a developed plate and extracted to recover the analytes, or the effluent flow in an online FFPC configuration can be collected and isolated. Further information and details on both capillary and FFPC methods have been summarized.”’ Multidimensional TLC

Two-dimensional TLC is very easy to carry out with a rectangular plate. The first dimension of the process is run in one direction. The plate is then dried, rotated 90”, and the second run is made with a second mobile phase. The analyte spots will be distributed across the two-dimensional space of the entire plate, hopefully better separated than is possible with a one-dimensional separation.

348

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

Figure 11.8. Composition of T L C plate with two different stationary phases. Courtesy Whatman. Inc.

Another type of plate contains two layers, both of which are active but very different in polarity. Figure 11.8 shows a plate composed of a thin layer of nonpolar bonded phase and a larger area of silica gel. These plates are used in two-dimensional TLC with two different mobile phases. The separating power of both normal- and reversed-phase LC are combined in one plate. The separation of 13 sulfonamides on such a plate is shown in Figure 11.9. Other multidimensional modes can couple TLC with other separation processes and detection systems such as MS. This topic is included in Chapter 15, but the review study of Poole and Poole26 is specifically devoted to TLC. 11.4

LITERATURE SUMMARY AND APPLICATIONS

Extensive bibliographies of the early work have been and the CAMAG Co. provides a continually updated bibliographic service.” Stahl, who first used the name TLC, has written an interesting account of his 20 years in the field34 as well as a useful laboratory handbook.35 The story of TLC has also been written by Ettre and Kalasz.36 PooIe3’ has reviewed

11.4

LITERATURE SUMMARY AND APPLICATIONS

349

1.Sulfisoxazole 2. Sulfathiazole 3. Sulfadiazine

4. Sulfaquinoxaline

5. Sulfachlorpyridazine

6. Sulfaguanadine 7. Sulfamerazine 8. Sulfabromethazine 9. Sulfadimethoxine 10. Sulfamethazine 11. Sulfaethoxypyridazine

12. Sulfanilamide 13.Sulfapyridine

SULFONAMIDES ON WHATMAN MULTI-K PLATE Plate: Multi-K;

C1a

-

strip on silica gel plate; 20 x 20 cm

Sample Data Pmparatiin: 13 Sulfonamides. 1 mg/ml in Acetone/Methanol (9O:lO) Applied Volume: Approx. 10 pI applied to CIS strip Development: 2-Dimensional (1) CIO development to 16 cm in equilibrated tank (2) Silica gel development to 14 cm in unequilibrated tank

Solvent: (1) (CISlayer): Toluene/CHiCN (8020) (2) (Silica gel layer): Ethyl acetate/MeOH/ NH4OH (85:15:0.6) Figure 11.9. Two-dimensional separation of 13 sulfonamides on Multi-K plate. Courtesy Whatman, Inc.

planar chromatography at the turn of the century. TLC has been review biennially in Analytical Chemistry since 1970 and provides a good source of information on the latest developments in planar chromatography. The latest review is in 2004.3x Virtually all types of compounds have been run by TLC, so the literature is a good place to begin in setting up a method. Reviews by type of compound (amino acids to toxins) are included in the Sherma and Fried Handbook,’ and the CAMAG bibliography servicejj provides continual updating.

350

LIQUID CHROMATOGRAPHY ON PLANE SURFACES

REFERENCES 1.

2. 3. 4.

5. 6.

7.

8. 9. 10. 11. 12. 13.

14. 15. 16. 17.

18.

19. 20. 21. 22. 23. 24. 25.

26.

G. H. Stewart, Ado. Chromatogr N . Y . 1965, I , 93-111. I. Smith, Ado. Chromatogr. N. Y. 1965, 1, 61-92. See, for example; P. G. Markow, J. Chem. Educ. 1988, 6.5, 899-900. J. Sherma and G. Zweig, Paper Chromatography and Electrophoresis, Vol. 2, Paper Chromatography, Academic Press, New York, 1971. C. F. Poole and S. K. Poole, Anal. Chem. 1989, 61, 1257A-1269A. J. Sherma, in Encyclopedia of Chromatography, Vol. 13, R. A. Myers (ed), Wiley, New York, 2000, pp. 11485-1 1498. J. Sherma and B. Fried (ed), Handbook of Thin-Layer Chromatography, 3rd ed., Marcel Dekker, New York, 2003. G. Zweig and J. Sherma (eds), Handbook of Chromatography; General Data and Principles, Vol. 11, CRC Press, Boca Raton, FL, 1972, pp. 103-189. C. F. Poole, J. Chromatogr. A 2003, 1000, 963-984. A review with 143 references. F. M. Rabel, in Handbook of Thin-Layer Chromatography, 3rd ed., J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. L. R. Snyder, Ado. Chromatogr. N. Y . 1967, 4 , 3. J. J. Peifer, Mikrochim. Act 1962, 529. K. Dallenbach-Toelke, Sz. Nyiredy, B. Meier, and 0. Sticher, J. Chromatogr. 1986, 36.5, 63. C. Cimpoiu, Optimization, in Handbook of Thin-Layer Chromatography, 3rd ed.,J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. C. F. Poole and S. K. Poole, Anal. Chem. 1994, 66, 27A-37A. CAMAG Scientific Inc. (USA), Wilmington, NC. www.camug.com/en/. W. Golkiewicz, Gradient Development in TLC, in Handbook of Thin-Layer Chromatography, 3rd ed., J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. T. Kowalska, K. Kaczmarski, and W. Prus, Theory and Mechanism of TLC, in Handbook of Thin-Layer Chromatography, 3rd ed., J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. D. Nurok, Anal. Chem. 2000, 72, 634A-641A. D. Nurok, J. M. Novotny, M. A. Carmichael, J. J. Kosiba, R. E. Santini, G. L. Hawkins, and R. W. Replogle, Anal. Chem. 2004, 76, 1690-1695. G. J. Van Berkel, J. J. Llave, M. F. De Apadoca, and M. J. Ford, Anal. Chem. 2004, 76, 479-482. N. Bryson and D. Papillard, LG-GC No. A m . and LC-CC No. A m . Supplement, Application Notebook, 2004, 22, 366-378. OPLC-NIT, Ltd, Budapest, Hungary Bionisis, Inc., Salem, C T 06420. www.bionisis.com. E. Mincsovics, K. Ferenczi-Fodor, and E. Tyihak, Overpressurized Layer Chromatography, in Handbook of Thin-Layer Chromatography, 3rd edition, J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. C. F. Poole and S. K. Poole, J. Chromatogr. A 1995, 703, 573-612.

SELECTED BIBLIOGRAPHY

351

27. W. Golkiewicz, Gradient Development in TLC, in Handbook of Thin-Layer Chromatography, 3rd ed., J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. 28. Sz. Nyiredy, J . ChromatobT. A 2003, 1000, 985-999. 29. M. Prosek and I. Vovk, Basic Principles of Optical Quantification in TLC, in Handbook of Thin-Layer Chromatography, 3rd ed., J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. 30. Sz. Nyiredy, Preparative Layer Chromatography, in Handbook of Thin-Layer Chromatography, 3rd ed., J. Sherma and B. Fried (eds), Marcel Dekker, New York, 2003. 31. K. Macek, I. M. Hais, J. Kopecky, and J. Gasparic, Bibliography of Paper Chromatography and Thin Layer Chromatography, 1961-65, supplementary volume, J. Chromatogr., Elsevier, Amsterdam, 1968. 32. D. Janchen (ed), Thin Layer Chromatography: A Cumulative Bibliography in Three Parts, 1965-1973, Camag, Muttenz, Switzerland, 1974. 33. Bibliography Service on Thin-Layer Chromatography, CAMAG Co., Wilmington, NC. www.cumag.coni/en/cbs e.htm. Volume 91 was published in September, 2003. 34. E. Stahl, J . Chromatogr. 1979, 165, 59. 35. E. Stahl, Thin Layer Chromatography, 2d ed., Springer, New York, 1969. 36. L. S. Ettre and H. Kalasz, LC-GC No. A m . 2001, 19, 712-721. 37. C. F. Poole, J. Chromatogr. A 1999, 856, 399-427. 38. J. Sherma, Anal. Chem. 2004, 76, 3251-3261.

SELECTED BIBLIOGRAPHY Cserhati, T. and Forgacs, E., Thin-Layer Chromatography, in Chromatography Fundamentals, Applications, and Troubleshooting, J. Q. Walker (ed), Preston,, Niles, IL, 1996, pp. 185-207. Fried, B., and Sherma, J., Thin-Layer Chromatography, 4th ed., Marcel Dekker, New York, 1999. Fried, B., and Sherma, J. (eds), Pructicul Thin-Layer Chromatography: A Multidisciplinavy Approach, CRC Press, Boca Raton, FL, 1996. Hahn-Deinstrop, E., Applied Thin-Layer Chromatography, Wiley-VCH, Weinheim, Germany, 2000. Journal of Planar Chromatography-Modern T L C , Research Institute for Medicinal Plants, Budakalasz, Hungary, in cooperation with Springer Hungarica, Budapest, Hungary; 6 issues/year. Nyiredy, Sz. (ed), Planar Chromatography: A Retrospective Mew for the Third Millenium, Springer, Budapest, Hungary, 2001. Sherma, J., and Fried, B. (eds), Handbook of Thin-Layer Chromatography, 3rd ed., Marcel Dekker, New York, 2003. A voluminous and comprehensive book. Touchstone, J. C. and Sherma, J. (eds.), Techniques and Applications of Thin Layer Chromatography, Wiley, New York, 1985.

This Page Intentionally Left Blank

12 Q UAL ITAT IVE ANALYSIS Both GC and LC find primary use in quantitative analysis, which was covered in Chapter 9. For qualitative analysis they are not as effective, and consequently a spectroscopic method is usually used in conjunction with chromatography for qualitative confirmation. Most commonly, mass spectrometry (MS) is the technique chosen to be coupled to a chromatograph, and it is of sufficient importance to warrant its own chapter, Chapter 10. T h e MS can be considered to be the chromatographic detector, o r the combination of two techniques can be referred to as a hyphenated system, such as GC/MS. Hyphenated systems are also included in the section on multidimensional chromatography in Chapter 15. 12.1

RETENTION PARAMETERS

Chromatographic retention parameters used for qualitative analysis include the following related parameters: retention time o r retention volume, retention factor ( k ) , retardation factor ( R ) , and the Kovats retention index ( I ) . For a defined system with invariant conditions, a given analyte will have a constant retention volume, and an unknown can be identified by comparison with retention volumes of standards. Figure 12.1 contains two chromatograms Chromutogruphy: Concepts und Contrasts, Second Editioti. By James M. Miller ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

353

354

QUALITATIVE ANALYSIS

I

7

3 4 5

8

9

I

1 II

2

I

*

04-

I

1 I I

I

18

I

24

i30

I

36 min.

I

I I

I

1

a

I 1

I

C

d

1 I

a. Methyl alcohol b. Ethyl alcohol c. n-propylalcohol d. n-butyl alcohol e. n-amyl alcohol

e

I

0

I

6

I 12

I 18

I 24

I 30

I 36 min.

I

(b) Standard Figure 12.1. Identification of unknowns by retention time using standards. Courtesy of H. McNair and E. Bonnelli, Basic Gas Chromatography. Varian, 1968.

run under the same conditions, one with an unknown (Fig. 1 2 . 1 ~ and ) one with five known alcohols (Fig. 12.lb). The five peaks in the first chromatogram are probably the five alcohols with the same respective retention times shown in the chromatogram of the standards. However, too many variables must be closely controlled to make this relationship useful except when samples are run in close time proximity on the same system; slight

12.1

RETENTION PARAMETERS

355

changes in columns, temperatures, flow rates, and other variables make it useless otherwise. Another drawback is the fact that there are over 30,000 chemical compounds in commercial use, and each one cannot have a distinctive retention volume on a given system. If a compound or a mixture is truly unknown, its chromatogram will not provide characteristic retention volumes on the basis of which identifications can be made. In short, chromatography can be used for qualitative analysis for a limited set of chemicals, but its main use is not for screening unknowns. Relative Retention Parameters

The other retention parameters listed above can be calculated from the retention volume as described in Chapter 2, with the exception of the Kovats index, I , which was introduced in Chapter 7. The Kovats index is really a relative retention parameter that is ratioed to a homologous series of hydrocarbons, and it is not used much any more. However, an EPA laboratory has reported using the Lee retention index (which uses polyaromatic hydrocarbons as reference standards)' to aid in making identifications when reviewing inconclusive mass spectral data.' The use of a single standard produces a ratio like cy, the separation factor, which was defined in Chapter 2 as (12.1) The main problem with cy is that there is no single standard to which data have been ratioed, and consequently there are no tabulations of relative retention data in the literature, as there are for the Kovats retention index (referenced to n-paraffins) or others like the Lee retention index just mentioned. Two-Column Plots

Qualitative analysis is enhanced if data are acquired on more than one system. This topic, often called multidimensional analysis, is covered in Chapter 15, but a few simple illustrations are included in this chapter. For example, in GC it is fairly common and easy to run a sample on each of two columns (often in the same column oven) that are chosen to be widely different in their polarities. The results can be plotted as net retention volumes or as Kovats index values on either linear or log scales, as shown in Figure 12.2. In either case, straight lines result for homologous series, thus aiding qualitative identifications. The principle is simple: the more data, the more reliable the analysis.

356

QUALITATIVE ANALYSIS

-5

-BP2= l W 10 ; 0

V O

// '

'

C 0

9

-

-

1 -

0 W

I

I

I

I

I

Two column plots are more effective if the two stationary phases are very different from each other. If they are of opposite polarity, then the elution order of the analytes in a given sample should be very different (opposite) on the two phases. For example, in the GC separation of a mixture of n-heptane, tetrahydrofuran, 2-butanone, and n-propanol, the elution order on a polar phase such as Carbowax is in the order just listed; on a nonpolar phase like SE-30, the order is the exact opposite. The two separations are shown in Figure 12.3; neither one follows the boiling point order. In HPLC, similar reversals can be expected between a normal-phase separation and a reversed-phase one. The elution order of analytes should be opposite (reversed) in one mode compared to the other. This effect is

12.2

OTHER METHODS OF QUALITATIVE ANALYSIS

357

2

TYPICAL CHROMATOGRAM ON CARBOWAX

(4

TYPICAL CHROMATOGRAM ON DC 200 (b)

Figure 12.3. Effect of stationary-phase polarity in a four-component separation: ( u ) Carbowax 20M (polar) and ( h ) DC-200 (nonpolar). Samples and their boiling points: (1) n-heptane (98), (2) tetrahydrofuran (64), ( 3 ) 2-butanone (80), and (4) n-propanol (97). Reprinted with permission of the COW-MAC Instrument Co., Bethlehem, PA.

expected in chromatography, but it is easy to forget it and assume that the elution order is always the same regardless of the column. 12.2 OTHER METHODS OF QUALITATIVE ANALYSIS

Virtually every technique imaginable has been examined in an attempt to Many are improve chromatography’s ability to perform qualitative ar~alyses.~

358

QUALITATIVE ANALYSIS

specific for only one sample type or only one chromatographic procedure, but some typical examples will be discussed to indicate the range of possibilities. Many are old and not used very much since MS has become so common. They have been divided into chemical methods and instrumental (detector) methods, other than MS. Chemical Methods

The chemical methods can be divided into those that are applied to the sample before it is analyzed and those that are applied after the analytes have been chromatographed. In the former category, the so-called precolumn reactions are many derivatizations. As discussed later, derivatives are usually made to facilitate their being chromatographed (volatility, detectability, and so on), but in some cases the derivatives may chromatograph differently and provide qualitative information. One example is the reactor of B e r ~ z a . ~ which was used with GC to remove functional groups prior to analysis. Similarly, some reactions can be carried out in situ in the injection port of a G C because of its high temperature, which produces increased reaction rates between the sample and added reagents. Pyrolysis is another common precolumn treatment used in G C and introduced in Chapter 7.s Nonvolatile samples are pyrolyzed very rapidly at high temperatures, and the degradation products are chromatographed. Polymers can be distinguished by the pattern of peaks obtained, without actual identification of the individual peaks. Chemical reactions are used to identify analytes after they have been separated. In TLC, it is common to spray the plate with a chemical that will selectively react with certain analytes. In GC, the column effluent can be bubbled into a solution containing a derivatizing reagent that will form a color or a precipitate' similar to the well-known spot tests. In HPLC, the column effluent can be mixed with a second reagent stream to produce a postcolumn reaction for purposes of identification or simply for detection.' Instrumental Methods

The instrumental methods are focused on the detector or an auxiliary instrument used as a supplementary detector. In both GC and HPLC there are detectors that show selectivity for particular groups of compounds or functional groups. One unusual, specific detector is the moth (alive) used with a G C to detect the presence of sex pheromones in a column effluent.8 The physical response of the moth clearly indicates which peak represents its sex hormone. Humans also sniff column effluents to identify particular odors in the flavor and fragrance industry.

12.2

OTHER METHODS OF QUALITATIVE ANALYSIS

359

MINUTES

ELECTRON CAPTURE DETECT0

-5

J-0

Figure 12.4. Dual-channel presentation of G C analy3is of gasoline sample on a packed DC-200 column. Courtesy of Perkin-Elmer Corp.

Most dual detectors are run in parallel, the column effluent being split and run through both of them simultaneously. In GC the technique is known as dual channel GC; usually, one of the detectors chosen is universal and the other is highly selective. Figure 12.4 shows the analysis of a commercial gasoline sample with dual detection by flame ionization (FID) and electron capture (ECD). The FID detects all the hydrocarbons, but the ECD is selective for the alkyl lead additives in gasoline and permits their detection without interference from the hydrocarbons. Another example9 is the separation of atmospheric hydrocarbons with detection by FID and a photoionization detector (PID), which is more sensitive for unsaturated hydrocarbons (Fig. 12.5). The identities of the peaks are given in Table 12.1 along with the PID/FID response ratios. It can be seen that the ratios can be used as additional information in assigning peak identities. In fact, Figure 12.6 shows that the hydrocarbon type (saturates, olefins, or aromatics) can be assigned from the detector ratio in many cases. A similar arrangement in HPLC is the simultaneous, parallel detection of column eluents at two wavelengths in a UV detector. In this case, the two signals are usually ratioed against each other, giving rise to a chromatogram, or ratiogram, that produces squared-off peaks for individual components. Figure 12.7 shows four chromatograms and four ratiograms of an unknown mixture run with four different mobile phases.'" The ratiogram was helpful in determining how many analytes were in the sample, four or five. Note that a ratio near 1 does not produce a peak [e.g., peak 1 in run ( b ) ] ,and that a well-resolved peak gives a squared-off peak [peak 2 in run ( b ) ] .The third

360

QUALITATIVE ANALYSIS

40 -

i

13

PI D

'

1.23

20

I

FID

-

15

14

10

h

Q, W

cn

z

&? v)

10

6

15

I1

W

22

a

5

0 I

10

20

MINUTES

30

40

Figure 12.5. Dual-channel presentation of GC analysis of air contaminants in parking lot. For identification of peaks, see Table 12.1. Reprinted with permission from W. Nutmagul, D. R. Cronn, and H. H. Hill, Jr., Anal. Chenz. 1983, 55, 2160. Copyright 1983, American Chemical Society.

12.2

361

OTHER METHODS OF QUALITATIVE ANALYSIS

Table 12.1 Identification of Peaks in Figure 12.5= ~~

Peak No. 1 2 3 4 5 6 7

8 9 10 11 12

Compound n-Butane 2,3-Dimethylbutane 2-Methylpentane 3-Methylpentane 1-Hexene 2,4-Dimethylpentane Benzene 2,3-Dimethylpentane 3-Methylhexane 2,2,4-Trimethylpentane Toluene n-Octane

Normalized Peak PID/FID No. -

4 3 3 75 15 129 40 12 28 100 22

13 14 15 16 17 I8 19 20 21 22 23

Compound

Normalized PID/FID

Ethylbenzene p- and m-Xylene o-Xylene n-Nonane Isopropylbenzene n-Propylbenzene p-Ethyltoluene 1,3,S-Trimethylbenzene o-Ethyltoluene 1,2,4-Trimethylbenzene 1,2,3-Trimethylbenzene

84 116 82 25 134 103 76 158 105 123 95

Source: Reprinted from reference 9, with permission. “Including observed PTD/FID ratios normalized to toluene

PID FID

I

1-Hexene

-

\

Isopropylbenzene 1-Octene

1-Nonene 1-Decene n-Hexane n-Heptane n.0ctane n-Nonane n-Decane

Hydrocarbon compound

Figure 12.6. Relative (PID/FID) response for 15 hydrocarbons. Reprinted with permission from W. Nutmagul, D. R. Cronn, and H. H. Hill, Jr., Anal. Chem. 1983, 55, 2160. Copyright 1983, American Chemical Society.

362

QUALITATIVE ANALYSIS

5

[65% MeOH/35% H20

I

40% THF/60% H20

50% ACN/50% H20

20% THF/25%ACN/55% H20

---(a)

Ib)

(C)

Id)

Figure 12.7. LC chromatograms and ratiograms using U V detection at two wavelengths (255 and 280 nm) with four different mobile phases. Reprinted with permission from A. C. J. H. Drouen, H. A. H. Billiet, and L. De Galan, Anul. Chem. 1984, 56, 971. Copyright 1984, American Chemical Society.

peak in ratiogram ( 6 ) is not square, indicating a possible overlapping impurity; the presence of this fifth component is confirmed in chromatogram ( d ) . The availability of multiple-wavelength UV detectors (see photodiode array detectors in Chapter 8) makes possible analyte recognition by increasing the number of independent informational degrees of freedom. A computer is needed to handle the data. Some recent examples have been given in HPLC" and SFC.12 Mathematical deconvolution of peaks containing up t o three components has been reported" without requiring prior knowledge of the identity or the spectra of the analytes. Another variation using multiple detection is the molecular weight chrom a t ~ g r a p h , 'which ~ is a GC with two gas density detectors and two different carrier gases. The data from this instrument can be used to calculate the molecular weight of each analyte in the range from 2 to over 400.

REFERENCES 1. M. L. Lee, D. L. Vassilaros, C. M. White, and M. Novotny, Anal. Chem. 1979, 51,

768. 2. W. P. Eckel and T. Kind, Anal. Chem. Acta 2003, 494, 235-243. 3. V. G. Berezkin, Chemical Methods in Gas Chromatography, Elsevier, Amsterdam, 1983; D. A. Leathard, in Advances in Chromatography, Vol. 13, J. C. Giddings (ed), Dekker, New York, 1975, p. 265; L. S. Ettre and W. H. McFadden (eds), Ancillary Techniques of Gas Chromatography, Wiley-Interscience, New York, 1969.

REFERENCES

363

4. M. Beroza, Anal. Chem. 1962, 34, 1801. 5 . V. G. Berezkin, V. R. Alishoyev, and I. B. Nernirovskaya, Gas Chromatography of Polymers, Elsevier, Amsterdam (reprinted), 1983; R. W. May, E. F. Pearson, and D. Scothern, qVrolysis GC, Chemical Society, London, 1977; J. 0. Walker and C. J. Wolf, J . Chromatogr. Sci. 1970, 8, 513. J. C. Hu, Adu. Chromutogr. N. Y . 1984, 23, 149. 6. J. T. Walsh and C. Merritt, Jr., Anal. Chem. 1960, 32, 1378. 7. R. W. Frei, H. Jansen, and U. A. Th. Brinkman, Anal. Chem. 1985, 57, 1529A. 8. B. A. Bierl, M. Beroza, and C. W. Collier, Science 1970, 270, 87. 9. W. Nutrnagul, D. R. Cronn, and H. H. Hill, Jr., Anal. Chem. 1983, 55, 2160. 10. A. C. J. H. Drouen, H. A. H. Billiet, and L. De Galan, Anal. Chem. 1984, 56, 971. 11. A. C. J. H. Drouen, H. A. H. Billiet, and L. De Galan, Anal. Chem. 1985, 57, 962. 12. K. Jinno, T. Hoshino, T. Hondo, M. Saito, and M. Senda, Anal. Chem. 1986, 58, 2696. 13. R. F. Laccy, Anal. Chem. 1986, 58, 1404. 14. C. E. Bennett, L. W. DiCave, Jr., D. G. Paul, J. A. Wegener, and L. J. Levasc, Am. Lab. 1971, 3(5), 67.

This Page Intentionally Left Blank

13 CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

Electrophoresis is not a chromatographic method, but recently these two techniques have been merged into a single technique called capillary electrochromatography (CEC). The purpose of this chapter is to present CEC and contrast it with conventional electrophoresis, capillary electrophoresis (CE), and with conventional HPLC. In the process, we will discuss some of the basics of electrophoresis to clearly differentiate electrophoresis from chromatography. And, since it is also the case that the techniques are complementary separation methods, chromatographers are likely to appreciate some background in electrophoresis. In any event, it is instructive to study the concepts of electrophoresis and contrast them with HPLC. What brought electrophoresis and HPLC together was their common use of capillary columns. The first reported use of electrophoresis in open tubular capillary columns was published in 1967,' but several publications in 1 9 7 9 ~ 1 9 8 1gave ~ a real impetus to the field, using fused-silica columns similar to those used in capillary GC. Later, when a pseudostationary phase was introduced as a secondary phase, the technique took on some characteristics of LC and became known as electrokinetic chromatography (EKC) or, if the second phase was micellular, micellar electrokinetic chromatography

Chrornutogruphy: Concepts and Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wilcy & Sons, Inc.

By James M. Miller

365

366

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

(MEKC).3 In this chapter, we will consider MEKC* to be a special type of capillary electrophoresis, rather than a chromatographic method. The move to try wall-coated open tubular columns, and packed capillaries with stationary phases of the type used in reversed-phase HPLC, launched the operational mode now called CEC, a bona fide chromatographic method. This chapter begins with the principles of electrophoresis, followed by a brief description of CE, capillary zone electrophoresis (CZE), and MEKC, and finally ends with CEC.

13.1

PRINCIPLES OF ELECTROPHORESIS

Electrophoresis is a separation method in which the driving force is a voltage gradient. Electrically charged species (ions) dissolved or suspended in an electrolyte and subjected to a voltage gradient are caused to move toward the electrode carrying the opposite charge. Separations occur when cations move in one direction and anions in the other, or when the rates of migration of similarly charged ions are different. The original work by Tiselius4 was done in a U-tube containing unsupported or free solutions separated by boundaries that were caused to move by the imposed voltage. For this invention of moving boundary electrophoresis in 1937, Tiselius was awarded a Nobel prize in 1948. Later forms of electrophoresis were carried out in columns and on supported planar media wetted with liquid, usually an aqueous solution. The planar material can be paper or silica (like TLC) or a variety of gels such as starch or polyacrylate polymer. The planar techniques are called by a variety of names, including zone electrophoresis and gel electrophoresis. The term zone was defined in Chapter 6 as an inclusive term used for both peaks and bands. Its use in electrophoresis serves to differentiate between (zonal) methods in which there is separation into zones and moving boundary electrophoresis where complete separation is not achieved.s Fundamental Equations

In zone electrophoresis a sample is placed at one edge of a planar material and caused to move under the influence of a voltage gradient. In such a field the rate of travel of an ion can be expressed as its electrophoretic ionic “MEKC is also called micellar electrokinetic capillary chromatography (MECC) but MEKC will be used in this book.

13.1

PRINCIPLES OF ELECTROPHORESIS

367

mobility ion pion: (13.1) where uion is the velocity of travel of the ion and E is the electric field strength or voltage gradient V / L in volts per centimeter. Separations are effected by differential ionic mobilities of the analytes. If an ion is assumed to be spherical and Stoke’s law is applied, its ionic mobility can also be expressed as: 2

(13.2)

Pion =

where r is the radius of the ion, z is its charge, and solvent. In terms of velocity,

7)

is the viscosity of the

(13.3) And if s / t is substituted in Eq. (13.1) for uion, the new equation can be solved for s, the distance traveled in time t :

(13.4) That is, the distance migrated in time t is directly proportional to the ionic mobility and the voltage gradient, V / L . Realistically, other factors play a role in the electrophoretic migration such as molecular diffusion, electroosmotic flow, temperature gradients, and intermolecular attractions between the analytes and the supporting media. Consequently, little use is made of the above equations beyond the general concepts of the effects of ion size and charge, voltage, and viscosity on electrophoretic migration. We conclude that: 1. 2. 3. 4. 5.

A divalent ion will migrate twice as fast as a univalent one of like size. Cations will migrate in the opposite direction to anions. Small ions will migrate faster than large ions of like charge. A low viscosity solvent favors increased mobilites. A high voltage produces faster separations.

Weak Acids and Bases In addition to ions, compounds such as weak acids or bases that are partially ionized will also move in an electrophoretic field. Since their degree of ionization is pH dependent, the pH of the solvent will be another factor

368

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

Table 13.1. Acid lonziation Constants and Electrophoretic Behavior for Histidine and Glutamic Acid

Ionization Constants Amino acid Histidine Glutamic acid

PK,

PK,

PK,

6.10

9.18 4.28

9.76

(protonated) 2.30

Isoelectric Point

Relative Electrophoretic Migration at pH 3.3

7.7 3.3

-

7.2

68

-8

+7

+72

-

9.3 +10 +77

Source: Data from reference 6.

governing their movement and separation. At a pH equal to an acid’s p K,, the acid is half ionized; and, at pH values approximately k 2 pH units of its pK,, the acid can be considered to be either fully ionized (at 2 units above the pK,) or nonionized (molecular; at 2 units below its pK,). The optimum pH at which to separate two acids, X and Y, ha5 been found to be

Another situation commonly encountered concerns compounds that have both acidic and basic properties (called amphoteric), such as the amino acids. Consider histidine and glutamic acid whose pK, values are given in Table 13.1. Obviously glutamic acid is the stronger acid, and it has two carboxylic acid groups, whereas histidine has only one. The p K , values in Table 13.1 are for the protonated form; that is, at a pH equal to p K , , the amino acid is half in the molecular form and half in the cationic form. Similarly, at a pH equal to pK,, a given amino acid is half in the molecular form and half in the anionic form. At some point between p K , and pK,, the amino acid must be 100% in the molecular form; this is known as its isoelectric point. At its isoelectric point an amino acids exists as a “zwitterion,” which has no net charge and will not migrate under an electrophoretic voltage gradient. A representative zwitterion formula is

H

I / /

R-C-C

1 1 3

H-N-H

\

0

oe

I

H ‘In equation 13.5, p is used for ionic mobility as defined in equation 13.2.

13.1 PRINCIPLES OF ELECTROPHORESIS

369

The electrophoretic data in Table 13.1 reflect these principles for the two amino acids. (Note that the migrations are relative values at each pH.) At the first pH, 3.3, the glutamic acid is very close to its isoelectric point and migrates very little; histidine is virtually completely protonated and cationic, so it moves very far toward the negative electrode. At p H 7.2, histidine is near its isoelectric point and moves very little; glutamic acid, having lost its carboxylic proton, is anionic and moves toward the positive electrode. Finally, at p H 9.3 both amino acids are negatively ionized and move toward the positive electrode. Note, however, that glutamic acid, one form of which is in the divalent form, moves farther than monoionic histidine. In this example, any of the three pH values would be satisfactory for separating these two amino acids. In actual practice, with a large number of compounds to be separated, the choice would be more restricted. Complex Formation

By forming complexes between a solute and a component of the solvent (buffer), it is possible to form ionic complexes from molecules, and to change the sign of an ion from positive (when uncomplexed) to negative (in a complex). The major example of the first type is the carbohydrates, which do not ionize readily. However, they do form ionic complexes with the borate ion in basic solution, so they can be separated by electrophoresis. Anionic complexes can be formed with several other inorganic oxyacids." Some complexes can be formed that will result in a change in the sign of the charge. Consider, for example, the complex formed between a metal ion and a ligand such as the chloride ion. At a low chloride ion concentration the metal ion will be cationic; as the concentration of chloride is increased, a point will be reached where the anionic chloro complex will predominate. The sign of the charge will depend on the ligand concentration and the stability constant, so a separation of metal ions can be based primarily on their differences in stability constants. This use of a complex-formation reaction is very similar to the procedure used in ion exchange chromatography (see Chapter 8). Electrolyte Concentration

So far we have assumed the presence of an electrolyte in the solvent. It has served as a buffer and/or complexing agent. In general, an electrolyte is necessary to carry the current and lower the resistance of the support. However, a high concentration of electrolyte may be disadvantageous because ionic mobility is inversely proportional to the square root of the ionic strength. As ionic strength increases, the mobility decreases, which is unde-

370

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

sirable. The reason for the decreased mobility is that an ion experiences a smaller local voltage gradient at a high ionic strength due to the aggregation around it of counterions, which shield it. For example, a layer of anions will surround a given cation, and a more diffuse layer of cations will probably surround the anionic layer. These layers are different from, and should not be confused with, the actual formation of anionic complexes as discussed in the last section. Particles of colloidal size (such as large proteins) also acquire a charged surface layer known as the Helmholtz double layer. Because the layers are of opposite charge, a potential, called the zeta potential, can be assumed to exist between them. This potential affects the electrophoretic mobility of colloids as given in the following equation7:

(13.6) where 5 is the zeta potential and L is the dielectric constant. The ionic mobility increases as the zeta potential increases, but a high electrolyte concentration tends to compress the double layer and decrease the zeta potential and the ionic mobility. Electroosmosis The concept of the Helmholtz double layer is useful in explaining another effect in electrophoresis-electroosmosis. The double layer arises when a solid is in contact with a liquid; the surface of the solid acquires a negative charge and attracts a double layer of cationic counterions from the solution around it. When a potential is applied across such a system, a flow of solvent results. This model describes the electrophoresis system and explains the observed flow of bulk solvent. It is called electroosmotic flow (EOF). When the solvent moves, it carries analyte ions and molecules with it. This flow is not intended to occur in electrophoresis and complicates the system, so some correction must be made. Usually this is done by observing the movement (if any) of an uncharged molecule and correcting the observed movement of the ions by an amount equal to the EOF (added to the anions and subtracted from the cations). However, as will be explained shortly, E O F is often desirable in C E and it serves to move ions indiscriminately, like the mobile phase does in chromatography. In effect, in CE, the E O F caused by the voltage gradient performs the same function as the pump needed to move the solvent in HPLC.

Diffusion Working against separation is longitudinal molecular diffusion, which will occur during the time the solute is in the solvent (on the support). In Chapter

13 2

ZONEELECTROPHORESIS

371

6 we saw that band broadening (expressed as the quarter-band width c r > is directly proportional to the square root of the retention time or the time of diffusion. This zone broadening is another reason for keeping the analysis time short.

Adsorption Strictly speaking, adsorption is not a part of the electrophoretic process, but in practice some analytes do adsorb on the support. When this occurs, it will decrease the expected ion mobility and usually will result in a tailing band. It is to be avoided if at all possible.

13.2 ZONE ELECTROPHORESIS Zone electrophoresis (ZE) has been one of the best techniques for the separation and analysis of proteins and other complex biological mixtures. A brief discussion of some of the prominent features is presented here to serve as an introduction to capillary electrophoresis (CE). We will not elaborate on the column methods or the preparative uses of ZE. Just as paper chromatography preceded thin-layer chromatography (see Chapter 1l), so ZE on paper preceded thin-layer electrophoresis (TLE). The difference between TLC and TLE is the use of a voltage gradient (up to 100 V/cm) in the latter. Otherwise, the processes are similar. The high voltage used in TLE causes considerable heat to be generated, so cooling also has to be provided. TLC and TLE have been combined on the same plate to produce two-dimensional separations. The topic of multidimensional separations is included in Chapter 15. More useful protein separations have been possible when the planar material is composed of a gel that has pore sizes approximating those of proteins. In this case, the moving ions are separated by molecular sieving in addition to the differences in ionic mobilities, resulting in better resolution. The first gels were prepared from starch, but those most commonly used are made from N, N'-methylene-bisacrylamide. Variable pore sizes can be achieved by regulating the degree of cross-linking of the polymer. These gels are of limited stability and have to be prepared in vertical troughs just prior to use. The use of these gels is commonly called polyacrylamide gel electrophoresis or PAGE. Polyacylamide gel electrophoresis has become very popular for protein separations. It has also been combined with isoelectric focusing (IEF) to achieve a two-dimensional separation system called 2DGEL.' When performed under denaturing conditions that use sodium dodecylsulfate (SDS) as the surfactant, as is most often the case, it is called IEF-SDS PAGE. This

372

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

technique is capable of resolving over 1000 components. Reference 5 contains an evaluation and further references. At the time the Human Genome Project was started, PAGE was the technique of choice to separate the DNA (deoxyribonucleic acid) fragments produced in the sequencing process. Automated DNA sequencing instruments containing up to 96 gel lanes are used. Laser fluorescence is used as the detection method to examine each lane and the data are plotted as electropherograms, which look like chromatograms. In many labs, PAGE has been replaced by capillary gel electrophoresis (CGE) or capillary zone electrophoresis (CZE) for DNA sequencing. Commercial instruments capable of high-throughput analysis are available for this work, and other separations required for combinatorial chemistry and proteomics.” They provide 96 simultaneous parallel separations via capillary electrophoresis, drastically cutting the time required for analysis. Clearly, the future of most separations by electrophoresis will be done by capillary methods. 13.3 CAPILLARY ELECTROPHORESIS

The use of capillaries instead of planar surfaces in electrophoresis was i n recognition of the success in using fused-silica capillaries in chromatography and t h e expectation that small capillaries would be more successful in removing the joule heating that accompanies electrophoresis. These expectations have been met and, as a result, CE has become a popular complement to HPLC. CE offers the advantage that method development is much more predictable than for HPLC. Electrophoretic migration follows the simple rules just presented, whereas the molecular interactions in HPLC are much more complex. However, the very high voltages used represent a hazard that requires careful isolation and adequate safety precautions. A typical apparatus is shown in Figure 13.1. Typical capillaries are 25-75 p m i.d. and 25-100 cm in length. For electrophoresis they are filled with an aqueous buffer solution, often referred to as the operating buffer, run buffer, or background electrolyte (BGE). The detector most often used is UV absorption, and special adapters are used to accommodate the capillary tubing to accomplish direct online absorbance measurements. The fused-silica capillaries used are coated on the outside with a protective layer of polyimide that prevents UV transmission through it. For use in CE, the imide layer must be removed on a short section of the capillary near one end, and detection is accomplished through this “window.” Imide layer removal can be accomplished with heat (a small flame will do) and/or a solvent. In the most common usage, the detector is located near the cathode end and samples are drawn into the capillary at the anode end.

13.3

CAPILLARY ELECTROPHORESIS

373

Figure 13.1. Typical capillary electrophoresis apparatus. Reproduced from G. D. Christian, Anulyticul Chemistry, 6th ed., John Wiley & Sons. Copyright 2004, John Wiley & Sons. This material is used by permission of John Wiley & Sons, Inc.

As mentioned earlier, electroosmotic flow (EOF) occurs in electophoretic media and it is usually undesirable, but that is not the case in CE. The use of silica capillaries results in an EOF flow that moves toward the cathode. At and above pH 2 (the approximate pK, for silica), the inside surface of the silica capillary, which contains hydroxyl groups, will form a layer of protons (H') next to the negatively charged wall. This positively charged layer, including the solvating water molecules, is attracted to the cathode causing the EOF. The velocity of the EOF, utOF in centimeters/second is

At higher pH, more protons are formed and the EOF increases as shown in Figure 13.2 For silica capillaries that can be operated between pH 3 and 8, there is a fivefold increase in EOF over that range. However, the selection of the working pH is often based on other factors and not only the EOF desired. The time necessary for an ionic analyte to migrate the length of the capillary from the anode end to the detector, I , is (13.8) Note that 1 will be less than L to an extent depending on the geometry of the detector flow cell. For example, for a 100-cm capillary ( L ) ,I might typically be 80 cm.

374

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

3

4

5

6

7

8

PH

Figure 13.2. Effect of p H on electroosmotic flow. Reprinted with permission from K. Lukacs and J. Jorgenson, J . High Res. Chromatogr. 1985, 8, 407-41 1. Copyright Wiley-VCH, Weinheim. Germany.

For cations, both velocities will be toward the cathode, but for anions, u,,, will be toward the cathode but uion will be toward the anode. Thus, the positions of ions as they arrive at the detector will be in the order shown in Figure 13.3: cations followed by neutrals and then anions. Another important aspect of the E O F is the flow profile that results compared with the flow profile we have seen in HPLC. Figure 13.4 shows

NEUTRALS CATIONS

Small highly charged -3

I

ANIONS

Small

I

t highly charged

A TIME

Figure 13.3. Symbolic electropherogram showing order of elution of molecules and ions in capillary electrophoresis. Reproduced from D. R. Baker. Capillaly Electrophoresis, John Wiley & Sons, p. 21. Copyright 1995, John Wiley & Sons. This material is used by permission of John Wiley & Sons, Inc.

13.3 CAPILLARY ELECTROPHORESIS

Laminar Flow

(4

375

Electroosmotic Flow ( b)

Figure 13.4. Flow profiles: pressure-driven laminar flow versus electroosmotically driven. Reproduced from G. D. Christian, Analytical Chernistq, 6th ed., John Wiley & Sons, p. 634. Copyright 2004, John Wiley & Sons. This material is used by permission of John Wiley & sons,

Inc.

both of them, and it can be seen that the flat profile for EOF produces much less zone spreading. This is a very important characteristic of CE. There are five major modes of CE. They are listed in Table 13.2. Figure 13.5 provides a classification system that will help to explain how they differ. The three modes shown in the lower left of the figure are the ones composed of a single electrolyte and in the presence of EOF. Capillary zone electrophoresis is referred to as free-flow or free-solution electrophoresis since it is carried out in open tubular capillaries. Because of its simplicity and popularity, it is the technique we will consider in most detail. Later MEKC, a variation of CZE in which micelles are added to form a pseudo-second phase, will be described. Although it has been labeled as a chromatographic method, it is more like electrophoresis and sufficiently different from CEC to be discussed in this section. CGE has gel-filled capillaries and is similar to PAGE, just discussed. The other modes of CE in Table 13.2 will not be discussed further in this brief chapter, but they are described in several electrophoresis books" and a small commercial booklet.2'

-*''

Table 13.2 Major Modes of Capillary Electrophoresis

Capillary Zone electrophoresis (CZE) Also referred to as free solution or free-flow CE Micellar electrokinetic chromatography (MEKC) Contains a pseudostationary phase of micelles Capillary gel electrophoresis (CGE) Operating buffer contains a polymeric gel Capillary Isoelectric Focusing (CIEF) Separates according to isoelectric points Capillary Isotachophoresis (CITP) Employs two buffers that enclose the analyte zones between them

376

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

Electroohoresis

Two Buffer

Single Buffer

7 With EOF

CZE

MEKC

CGE

No EOF

No EOF

ClEF

ClTP

~

~

Figure 13.5. Classification of capillary electrophoresis techniques.

Capillary Zone Electrophoresis

Separations by CZE are based on the same electrophoretic principles discussed above, except that there is no support or bed to cause unwanted sorption, only a column wall. The mobile phase (MP) or BGE is usually an aqueous buffer, although a few studies have reported separations using a nonaqueous MP.22,23 In addition to the electrophoretic migration, the analytes are moved by the EOF as just described. The additional advantages of CZE are: Very high voltages (up to 35 kV) can be used to give separations 20 times as fast as planar electrophoresis. The capillary format provides good thermal conductivity and high surface area-to-volume ratio to effectively dissipate the heat built up by the use of high voltages. The electrogenerated flow profile results in less zone broadening than in HPLC. An online detector provides data necessary for quantitative analysis. An early study by Jorgenson and Lukacs2 showed that separation efficiency was proportional to applied voltage, and they were able to generate plate numbers as high as 400,000. A simple diagram of the apparatus is shown in Figure 13.1. The UV detector is most commonly used, just as in HPLC. The other common detectors are modeled after those used in HPLC: electrochemical and laser-induced fluorescence (LIF) being the two most popular ones. Two chapters in Khaledi’s book cover spectral detectors24 and electrochemical

13.3 CAPILLARY ELECTROPHORESIS

377

detectors” in CE. Because CE detectors are usually used online, making their measurements through the fused-silica column which has, in effect, a very short path length, their detectivity is not as good as in HPLC. As a result, there is more interest in the LIF detector, which inherently is capable of lower detection limits. Since no discussion of this detector was included in Chapter 8, an introduction is included here. Originally, LIF detectors were assembled by individual investigators from commercially available parts, for use in HPLC.2h,27The use of LIF in CE is quite similar and Zare’s group reported some applications using CZE and EKC.” Of the lasers used, the HeeCd laser is common; it has excitation lines at 442 and 326 nm. Usually the fluorescence emission is taken off at right angles to the excitation, minimizing the interference from scattered radiation. Several commercial instruments are now available.”’ Unfortunately, not many compounds fluoresce, so derivatization may be necessary to achieve detection by fluorescence methods. Chapter 14 contains some discussion on derivatization. The column is usually thermostated and run near ambient temperature. The polarity can be reversed by simply reversing the voltage leads for anion analysis when the EOF is too low to carry them to the detector (at low pH). New capillaries need to be conditioned. Usually this is accomplished with pretreatment with dilute NaOH ( - 0.1 M) followed by water and then run buffer. Sometimes this treatment is also required between runs. For information about the care and maintenance of capillaries, see reference 31. It should also be recognized that chemical reactions are occurring in the two buffer reservoirs, so they become contaminated with use and should be replaccd frequently. Sample introduction is more complex than in HPLC. The inlet end of the capillary is removed from the running buffer and placed in the sample solution. Transfer of a small sample (a few nanoliters) into the capillary can be accomplished in one of several ways:

’’

1. Electroosmotic injection can be used to draw the sample into the capillary. Usually a lower voltage (about 2 kV) is used for a measured time. This method works best if EOF is high so that even the anions (which are repelled by the cathode) are sampled. 2. The sample can be siphoned into the capillary by hydrostatic injection. This method can be as simple as raising the sample solution to a designated height above the operating height of the buffer reservoir for a measured time. 3. Or, in a similar fashion the sample solution can be hydrodynamically injected using a pressurized sample solution or a vacuum applied at the column outlet.

378

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

Commercial instruments include autoinjectors that function in at least one of these modes. To increase detectivity (a common deficiency), stacking of samples has been used online (in the capillary). This topic has been reviewed.32 One problem with CZE is the tendency of some cationic analytes to sorb on the negatively charged capillary wall. Unwanted adsorption was mentioned earlier as a problem with electrophoresis. This affect is detrimental to protein separations. To minimize it, protein separations are often run at low pH where the charge on the wall is minimal. Alternatively, the capillary can be coated with additives or bonded-phase capillaries can be used. Consult the Beckman booklet.33 Other disadvantages of CZE include poor LODs, poor quantitative precision, and poor electropheric migration reproducibility. For some analysts, these deficiencies, compared to RPLC, are severe enough to prevent them from making use of CZE However, CZE has been found to be especially useful for chiral separations since they can be performed by adding a chiral agent to the run buffer and thus do not require the

1 3

6

9

12

10mM HDAS-beta-CD 10mM ODAS-gamma-CD

15

Time (min) Figure 13.6. Typical chiral separations by C E showing the effcct of different chiral additives. Reprinted from T. Goel, J. Nikelly, R. Simpson, and B. Matuszewski, “Chiral Separation of Labetalol Stereoisomers in Human Plasma by Capillary Electrophoresis,” J . Chromutogr. A , 2004, 1027, 213-221, Copyright 2004, with permission from Elsevier.

13.3

CAPILLARY ELECTROPHORESIS

379

purchase of an expensive chiral stationary-phase (CSP) column as is used in HPLC. Surfactants and cyclodextrins (CDs) are among the commonly used chiral agents. A comparison of the effect of two different CDs on the separation of labetalol stereoisomers is shown in Figure 13.6.34 For further information, consult the Beckman booklet” or Chankvetadzes book.36Also, Chapter 15 includes additional discussion about chiral analysis. The theory of CZE has been comprehensively discussed by Poppe3’ and his chapter is a good source of CZE principles. Discussion of method development and validation can be found in references 38-40. Because CZE has been explored most thoroughly by pharmaceutical analysts, there are quite a few publications that will be of interest to workers in that field. Included are another booklet in the Beckman ~ e r i e s , ~a’ chapter in the Advances in Chromatography series,42 several books,4’,44 and Section 727 of the USP.4’ Reviews of the status of CZE have been published in 19994hand 2Q04.47 Capillary zone electrophoresis is not effective in separating neutral molecules. To extend the method to accommodate neutral molecules, a second phase can be added to serve as a pseudostationary phase in a chromatographic sense. One such phase is the formation of micelles and the technique is called MEKC.

Micellar Electrokinetic Chromatography

When a surfactant is added to an aqueous solution in sufficiently high concentration, called the critical micelle concentration, spherical aggregates called micelles are formed. Because the surfactant molecules have both polar and hydrophobic regions, they will aggregate with their polar ends extending into the solution and their nonpolar ends directed to the center of the sphere. This gives the micelle the appearance of a surface charge and it will migrate in an electrophoretic field. Being distinct from the aqueous solvent, they constitute a second phase, which can be considered to be a pseudostationary phase in a chromatographic sense. Analytes partition between the bulk aqueous (mobile) phase and the (pseudostationary) micellar phase, thus adding a secondary equilibrium to the separation process. The neutral analytes that have partitioned into the micelle will now be electrophoresed in proportion to the charged micelle. The chapter by Khaledi is a useful source of more information including a summary of application^,^^ and Terabe4’ has written a short summary of MEKC principles. A typical example is the mixed micellar separation of 17 corticosteroids shown in Figure 13.7.

380

I

Kl

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

1

20

I

30

I

h(rr*r)

40

I

50

Figure 13.7. Separation o f 17 corticosteroids by MEKC. Reprinted from Bumgarner and Khaledi, “Mixed Micclles of Short Chain Alkyl Surfactants and Bile Salts in Electrokinetic Chromatography: Enhanced Separation of Corticosteroids,”J. Chrornutogr. A 1996, 7.38, 275-283, Copyright 1996, with permission from Elsevier

13.4

CAPILLARY ELECTROCHROMATOGRAPHY

Capillary electrochromatography is a true chromatographic method, but one that has features of both RPLC and CZE. The apparatus most closely resembles CZE since the flow of the mobile phase (or run buffer) is driven by EOF, but a small (50-250 p m i.d.) HPLC column, like those described in Chapter 8 for HPLC, is used instead of an uncoated open tube. The advantage of the flat E O F flow profile is achieved yielding high efficiencies. But, in addition, the presence of a voltage gradient superimposes an electrophoretic effect on any charged species in the sample. This hybrid method has been called by many names, but CEC seems to have become the one most widely accepted. The 1981 study by Jorgenson and Lukacs5” is generally considered to mark the beginning of CEC. The basic description just presented for CZE is relevant for CEC; the difference is the chromatographic effects from the presence of a stationary phase. The fundamentals have been summarized by Pyell5’ who reported that efficiencies are about 10 times better for CEC than for HPLC, and that the van Deemter curves for CEC show lower plate heights ( H I and almost n o increase in H at velocities higher than the minimum H (see Fig. 13.8). Since the highest electrophoretic mobilities are obtained with acetonitrile-aqueous

13.4

CAPILLARY ELECTROCHROMATOGRAPHY

381

h 76 .

’i I

u (mm/s)

1

2

3

4

5

6

Figure 13.8. A comparison of van Deemter cumes for HPLC, CEC, and SFC. Reprinted with permission from K. Bartle, A. Clifford, P. Myers, M. Robson, K. Seale, D. Tong, D. Batchelder, and S . Cooper, “Packed Capillary Chromatography with Gas, Supercritical, and Liquid Mobile Phases,” in Unified Chromurogruphy:ACS Symposium Seria 748, J. F. Parcher and T. L. Chester (eds), American Chemical Society, Washington, DC, 2000. Copyright 2000, American Chemical Society.

buffer mixtures, this mixture has become the most popular. A good buffer concentration is about 1 mM. The choice of column type has been the subject of many investigations. Packed capillaries are the most common, using small-diameter particles as in HPLC, but all of the types currently being investigated for HPLC are also being used in CEC. A recent summary is availables2 and an entire volume of the Journal of Chromatography As3 has been devoted to CEC. Typically, packed columns have inside diameters of 100 p m or less and open tubular columns 20 p m or less. The major problem with packed columns has been the fabrication of the frits that retain the packing. Associated with this problem is the formation of bubbles that are believed to be generated by a change in EOF that occurs in the frits. Piraino and DorseyS4have published a comparison of frits, but the other alternative is to use the newer monolithic columns. See the studies of Bedair and El Rassi in Electrophore~is~~ and volume 1013 of the Journal of ChromutogruphyA,5h,57 the latter (volume 1013) including all the papers from the HPCE 2003 symposium.S8An example of the use of molecular imprint polymer sorbents in CEC is also in this volume.” Since these columns do not have frits, the above-mentioned problems can be avoided, and they are increasingly being used in CEC. A very thorough review of stationary and mobile phases in CEC, including 307

382

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

references has been published.‘” Optimization of particle dimensions has been reported to produce columns with more than 10‘ plates in a relatively short time.“ When ionic or partially ionic analytes are present in a sample, CEC functions as a unique type of chromatography because of the dual effects of chromatographic sorption and electrophoretic migration.62 In addition, there is some electromigration of analytes in the stationary phase, unlike the

1.2

-I

0.8

-

1

7

N v)

I ._

C

3

0.6 -

m f! 0.4 $ a 8

.m 0.2 ._ I

0.0 1

I

0

1

1

1

I

3

2

4

5

I

6

Retention time (min)

1.2 h

P

1.0

(4

4

0.0 0.8

1

1.o

1

1.2

1.4

I

1.6

Migration time (min) (b) Figure 13.9. Comparison of the separation of three proteins between CEC and CZE. Reprinted from D. Bandilla, and C. D. Skinner, “Protein Separation by Monolithic Capillary Electrochromatography,” J . Chromatogr. A 2003, 1004, 167-1 79, Copyright 2003, with permission from Elsevier.

REFERENCES

383

Table 13.3 Advantages and Disadvantagesof CEC

Advantages 1. 2. 3. 4. 5.

Greater efficiency Wide range of applications Faster No troublesome pumps Less expensive chiral analyses

Disadvantages 1. 2. 3. 4.

Poor detectivity (LOD) Poor quantitative precision Poor migration time precision High-voltage hazard

situation in HPLC.h3 Published examples of separations on monolithic columns include amino acids and peptidesh3 and proteimh4 Figure 13.9 compares one such separation by CEC and CZE. While the retention order remained the same in both techniques, the retention factors differ, reflecting the chromatographic effect in the CEC separation. The plate number and the resolution are both higher by CEC, but the speed of analysis is faster by CZE. Capillary electrochromatography can be operated by gradient elution as in HPLC, except that the gradients are by steps since the buffer reservoirs need to be changed from one mobile phase to another. In addition, it may be possible to also run voltage gradients and temperature gradients. The field is still new and growing. The advantages and disadvantages of CEC listed in Table 13.3 are mainly comparisons with HPLC. The main advantage is the higher efficiency that results from the ideal flow pattern obtained in electrophoresis, as discussed earlier. The combination of HPLC and CZE provides separations based on both effects for ionized materials and thus can effect separations not possible by HPLC alone. As noted earlier, the electrophoretic effects are predictable, so the combined effects should be predictable if one has separate chromatographic data. Terabe's group6' has shown some success in modeling the retention behavior in CEC from chromatographic and electrophoretic data. Of the disadvantages, the poor detectivity and precision are sufficiently severe to discourage many chromatographers from the use of CEC, although C E is finding use for chiral separations. The jury is still out. To date, only a few books have been published on CEC,", 67 plus a set of two C D S . ~h9~ .

REFERENCES 1. S. Hjerten, Chromatogr. Rev. 1967, 9, 122. 2. J. W. Jorgenson and K. D. Lukacs, Anal. Chem. 1981, 53, 1298-1302. 3. S. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya, and T. Ando, Anal. Chem. 1984, 56, 1 1 1.

384

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

4. A. Tiselius, Trans. Faraday Soc. 1937, 33, 524. 5. M. Bier,in A n Introduction to Separation Science, B. L. Karger, L. R. Snyder, and C. Horvath (eds), Wiley, New York, 1973, Chapter 17. 6. J. R.Whitaker, Electrophoresis in Stabilizing Media, Vol. 1, Academic, New York, 1967. 7. R. J. Akers, Am. Lab. 1972, 4(10), 41. 8. R. P. Tracy, High-Resolution, Two-Dimensional, Gel Electrophoresis or Proteins: Basic concepts and Recent Advances, in Chromatography and Separation Chemistry S. Ahuja (ed), ACS Symposium Series 297, American Chemical Society, Washington, D.C., 1986. 9. See, for example www,combisep.com. 10. K. D. Altria (ed), Capillary Electrophoresis Guidebook: Principles, Operation, and Applications, Humana, Totowa, NJ, 1995. 11. D. R. Baker, Capillary Electrophoresis, Wiley, New York, 1995. 12. P. Camilleri (ed), Capillary Electrophoresis: Theory and Practic, CRC Press, Boca Raton, FL, 1998. 13. P. D. Grossman and J. C. Colburn (eds), Capillary Electrophoresis: Theory arid Practice, Academic, San Diego, CA, 1992. 14. R. A. Hartwick, Introduction to Capillary Electrophoresis. CRC Press, Boca Raton, FL, 1994. 15. M. G. Khaledi (ed), High Perfiirmance Capillary Electrophoresis: Theory, Teckniques, and Applications, Wiley, New York, 1998. 16. J. P. Landers (ed), Handbook qf’ Capillary Electrophoresis, 2nd ed., CRC Press. Boca Raton, FL, 1997. 17. Z . L. Rassi, CE and CEC Reviews 2002: Aduunces in the Practice and Application oj Capillary Electrophoresis and Capillary Electrochromato~raphy,Wiley, Hoboken, NJ, 2002. 18. R. Weinberger, Practical Capillary Electrophoresis, 2nd ed., Academic, San Diego, 2000. 19. A. Weston and P. R. Brown, HPLC and C E : Principles arid Practice, Academic, San Diego, 1997. 20. R. L. Cunico, K. M Gooding, and T. Wehr, Basic HPLC and CE of Biomolecules, Bay Bioanalytical Laboratory, Richmond, CA, 1998. Also available from Varian at www.uarianinc.com. 21. Anon., Introduction to Capillary Electrophoresis, Beckman Instruments, Fullerton, CA, 1994, item 360643. See www.beckman.com. 22. H. Cottet, W. Vayaboury, D. Kirby, 0. Giani, J. Taillades, and F. Schue, A n d . Chem. 2003, 75,5554-5560. 23. M. D. Cantu, S. Hillebrand, M. E. C. Quciroz, F. M. Lancas, and E. Carrilho, J . Chromatogr. B 2004, 799, 127-132. 24. L. Cruz, S. A. Shippy, and J. V. Sweedler, in High P e f o m a n c e Capillary Electrophoresis, M. G. Khaledi (ed), Wiley, New York, 1998, Chapter 9. 25. B. R. Bryant, F. D. Swanek, and A. G. Ewing, in High P e f o m a n c e Capillary Electrophoresis, M. G. Khaledi (ed), Wiley, New York, 1998, Chapter 10. 26. J. Gluckman, D. Shelly, and M. Novotny, J . Chromatogr. 1984, 31 7, 443-453.

REFERENCES

385

27. E. L. Guthrie, J. W. Jorgenson, and P. R. Dluzneski, J . Chromutogr. Sci. 1984, 22, 171- 176. 28. E. S. Yeung, P. Weng, W. Li, and R. W. Giese, J . Chromatogr. 1992, 608, 73-77. 29. J. P. Quirino, M. T. Dulay, L. Fu, T. D. Mody, and R. N. Zare, J . Sep. Sci. 2002, 25, 819-824. 30. See, for example, www.beckman.corn; www.ce-resources.com; www.esuinc.com w w w .e min c .corn. 31. K. D. Altria, S. M. Bryant, B. J. Clark, and M. A. Kelly, LC-GC 1997, 15, 34-38. 32. Z. K. Shihabi, J . Chromatogr. A 2000, 902, 107-117. Contains 77 references. 33. H. E. Swartz, R. H. Palmieri, J. A. Nolan, and R. Brown, Introduction to Capillary Electrophoresis of Proteins and Peptides, Beckman Instruments, Fullerton, CA, 1993, item 266923. 34. T. V. Goel, J. K. Nikelly, R. C. Simpson, and B. K. Matuszewski, J . Chromatogr. A 2004, 1027, 213-221. 35. M. M. Rogan and K. D. Altria, Introduction to the Theory and Applications of C h i d Capillary Electrophoresis, Beckman Instruments, Fullerton, CA, 1995, item 726388. 36. B. Chankvetadze, Capillary Electrophoresis in Chiral Analysis, Wiley, New York, 2002. 37. H. Poppe, in Advances in Chromatography, Vol. 38, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 1998, Chapter 6. 38. R. Weinberger, Am. Lab. 2003, 3561, 54-56. 39. H. Fabre and K. D. Altria, LC-GC No. A m . 2001, 19, 498. 40. D. Heiger, R. E. Majors, and R. A. Lornbardi, LC-GC 1997, 15, 14-23. 41. K. D. Altria and M. M. Rogan, Introduction to Quantitative Applications of Capillary Electrophoresis in Pharmaceutical Analysis, Beckman Instruments, Fullerton, CA, 1994, item 538703. 42. S. F. Y. Li, C. L. Ng, and C. P. Ong, Pharmaceutical Analysis by Capillary Electrophoresis, in Advances in Chromatography, Vol. 35, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 1995. 43. G. Lunn, Capillary Electrophoresis Methods for Pharmaceutical Analysis, Wiley, New York, 2000. 44. S. M. Lunte and D. M. Radzik (editors), Pharmaceutical and Biomedical Applications of Capillary Electrophoresis, Pergamon, Amsterdam, 1996. 45. United States Pharmacopoeia Convention, US Pharmacopoeia 27/National Formulary 22, U.S. Pharmacopoeia, Rockville, MD, 2004. 46. K. D. Altria, J . Chromatogr. A 1999, 856, 443-463. 47. K. D. Altria and D. Elder, J . Chromatogr. A 2004, 1023, 1-14. 48. M. G. Khaledi, in High Performance Capillary Electrophoresis, M. G. Khaledi (ed), Wiley, New York, 1998, Chapter 3. 49. S. Terabe, Anal. Chem. 2004, 76, 240A-246A. 50. J. W. Jorgenson and K. D. Lukacs, J . Chromutogr. 1981, 218, 209-216. 51. U. Pyell, in Aduances in Chromatography, Vol. 41, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 2001, Chapter 1.

386

CAPILLARY ELECTROPHORESIS AND CAPILLARY ELECTROCHROMATOGRAPHY

52. L. A. Colon, T. D. Maloney, J. Anspach, and H. Colon, in Chromatography, Vol. 42, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 2003, Chapter 2. 53. C. Horvath (ed), J . Chromatogr. A 2000, 88, 1-513; 38 papers on CEC. 54. S. M. Piraino and J. G. Dorsey, Anal. Chem. 2003, 75, 4292-4296. 55. M. Bedair and Z. El Rassi, Electrophoresis 2002, 23, 2938-2948. 56. M. Bedair and Z. El Rassi, J . Chromatogr. A 2003, 1013, 35-45. 57. M. Bedair and Z. El Rassi, J . Chromatogr. A 2003, 1013, 47-56. 58. A. Paulus and A. Guttman (eds), Journal o f Chromatography A 2003, 1013, 1-232. 59. P. T. Vallano and V. T. Remcho, J . Chromatogr. A 2000, 887, 125-135. 60. J. Jiskra, H. A. Claessens, and C. A. Cramers, J . Sep. Sci. 2003, 26, 1305-1330. 61. R. Stol, H. Poppe, and W. Th. Kok, Anal. Chem. 2003, 75, 5246-5253. 62. D. Hoegger and R. Freitag, J . Chromatogr. A 2003, 1004, 195-208. 63. R. Stevenson, A m . Lab. 2002, 34201, 13-17. 64. D. Bandilla and C. D. Skinner, J . Chromatogr. A 2003, 1004, 167-179. 65. Z. Liu, K. Otsuka, and S. Terabe, J . Chromatogr. A 2002, 959, 241-253. 66. Z. Deyl and F. Svec (eds), Capillary Electrochemistty, Elsevier, Amsterdam, 2001. 67. I. S. Krull, R. L. Stevenson, K. Mistry, and M. E. Swartz, Capillary Electrochromatography and Pressurized Flow Capillary Electrochromutography, HNB Publishing, New York, 2000. 68. Capillary Electrophoresis: The Comprehensive, Interactive Tool for Beginners and Advanced Users of CE, CD, Pub. #5968-32348, Hewlett-Packard Co, Wilmington, DE, 1999. Now Agilent Technologies: www.agilent.com/chem. 69. Capillary Electrophoresis: Capillary Electrochromatogruyhy-techno lo^ and Applications, CD, Pub. #5968-3231E, Hewlett-Packard Co, Wilmington, DE, 1990. Now Agilent Technologies: www.agilent.com/chem.

14 SAMPLE PREPARATION Most chromatographic methods used to separate and quantify samples include additional steps in the preparation of the sample. These steps are sometimes referred to as preliminary treatments or sample cleanup. In fact, chromatography itself (usually LSC) is also used for sample cleanup as was described in Chapter 8. Other sample preparation techniques are listed in Table 14.1. Many of them are also separation methods, and some share common aspects with chromatography. For example, liquid-liquid extraction (LLE) can produce separations similar to LLC. One such comparison can be found in Appendix B. Some of the phases used in solid-phase extraction (SPE) and solid-phase microextraction (SPME) have been developed because of their successful use in HPLC or GC. Not only do sample preparation methods have a lot in common with chromatographic methods, but they are becoming an increasingly important part of chromatographic method development. Without modern improvements, including the use of liquid handling devices and robots, the preliminary treatment phase of a method can be the most time-consuming part of an analysis. The increasing practice of combinatorial chemistry has resulted in an increase in number of samples to be analyzed, and HPLC is often the method of choice. Miniaturization of extraction processes, thereby using less solvent, and the use of SPE are examples of method changes that are making

Chromatography: Concepts and Contrasts, Second Edition. By James M. Miller ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

387

388

SAMPLE PREPARATION

Table 14.1 Some Sample Preparation Techniques Used in Chromatographic Analyses

1. 2. 3. 4. 5. 6. 7.

Dissolution (with and without sonication) Precipitation (with and without filtration or centrifugation) Evaporation Chromatography Extraction (LLE, SPE) Dialysis (including electrodialysis) Derivatization

possible increased automation, improved recovery and reproducibility, and decreased time requirements. Two reviews, one by Smith' and the other by Moldoveanu,* provide a clear picture of the current situation and includes many references for further information. Majors' has recently surveyed chromatographers to find out their current uses of sample preparation techniques. His results, shown in Figure 14.1, are more detailed than the list in Table 14.1 and present the 39 most popular techniques. Column chromatography, SPE, and LLE are among the most widely used. He also confirmed that sample loads are increasing and that chromatographers are searching for ways to decrease sample prep time while also investigating newer extraction techniques for solid samples. Included in this chapter are some basic principles, typical equipment, and important references for further study, but no attempt will be made to discuss the development of a logical strategy for sample preparation. A useful strategy has been published by LC-GC4; it begins by considering the physical state of the sample, solid or liquid, and covers most of the extraction methods plus dialysis in its flowcharts. Smith' has arranged his study into three sections: analytes in solid samples, in solution, and in the gas phase. LeBlanc's review of sample preparation techniques authorized by EPA' is also organized this way. Alternatively, in his comprehensive book Mitra' has separate chapters based on the various volatilities of the sample: nonvolatile, semivolatile, and volatile. Other references in this chapter will provide examples of other strategies. This chapter is organized by method and subdivided according to the physical states of the phases used in extraction. A rather extensive introduction to LLE is given to provide a link to readers' prior experiences with extraction and to show the similarities between LLE and LC that may help them to understand the basics of chromatography. Following the extraction section, dialysis and derivatization are reviewed briefly. The sample preparation methods discussed in this chapter are clearly very important parts of most analytical methods. However, there are other steps

14.1

EXTRACTION

389

Figure 14.1. Extent of use of various sample preparation procedures in 2002. Reprinted with permission from LC-GC No. Am., Vol. 20(12), December 2002, pp. 1098-1113. LC-GC No. Am. is a copyrighted publication of Advanstar Communications Inc. All rights reserved.

390

SAMPLE PREPARATION

in the analysis process that are not discussed here that should not be forgotten since they can also be critical for a successful procedure. Some examples are sample collection (including the acquisition of a representative sample), sample transport and storage; sample preparation (including grinding, slicing, sonication, maceration, etc.), sample measurement (usually weighing), and sample dissolution. Mistakes in any of these early stages can invalidate the finest chromatographic analysis.

14.1

EXTRACTION

There are many forms of extraction in addition to the old standby LLE performed in separatory funnels. Figure 14.2 is one classification of extraction techniques suggested by P a ~ l i s z y n .It~ distinguishes between methods that are: Equilibrium versus preequilibrium Equilibrium versus steady state Flow-through versus batch Exhaustive versus nonexhaustive By preequilibrium Pawliszyn means that the contact between phases is broken before the system has reached equilibrium, making these methods

Figure 14.2. Classification of extraction techniques. Reprinted with permission from reference 6. Copyright 2003 American Chemical Society.

14.1

EXTRACTION

391

effectively nonequilibrium methods. Steady state refers to permeation techniques such as membrane extractions that operate by continuous steady-state transport of analytes through membranes. Flow-through is a contact process that could also be labeled continuous (as opposed to batch). Whichever terms are used, the classification should be helpful to the chromatographer faced with the necessity of selecting the best sample preparation steps for his/her problem. Each of the extraction methods included in Figure 14.2 will be discussed, beginning with LLE, which is presented in a classical fashion. The other methods are the ones of major interest in modern analysis for the reasons mentioned earlier. The discussions are brief but are well referenced for further information. An international symposium on advances in extraction technologies is held annually and is a good source of information on current research in the field. Eighteen papers from the March 2003 meeting have been published together, providing examples of applications and fundamental aspects of extraction.' Simple Liquid-Liquid Extraction The most common form of extraction is performed with two immiscible liquid phases. The sample is dissolved in the raffinate and is contacted with the extractant. The partition coefficient, K,, which describes the thermodynamic control over the process can be defined as:

(14.1) where [A],,,, is the concentration of the analyte A in the raffinate phase, and [AIextris t h e concentration in the extractant phase, so chosen to correlate with the practice in chromatography* and, as in chromatography,

K,=pxk

(14.2)

where p is the phase volume ratio (V,x,,/V,c,fi) and k is the partition ratio (now called the retention factor in chromatography):

(14.3) where W is the weight of the analyte in each of the respective phases. If Q is "It is not important which tcrm is in the numerator and which in the denominator, but the choice in Eq. (14.1) makes the definition similar to that in chromatography. In much of the extraction literature the reciprocal is uscd, and raffinate (thc aqueous phase) is in the denominator and extractant (the organic. nonaqueous phase) is in the numerator.

392

SAMPLE PREPARATION

used to denote the fraction of analyte extracted, its relationship to k is given by: 1 Q = fraction extracted = (14.4) (1 + k ) ~

(14.5) Since, in chromatography, the retardation factor, R , is also equal to 1/(1 then Q must be similar in concept to R (see Chapter 2). Also, 1 - Q = the fraction unextracted

=

~

k (1 + k )

+k),

(14.6)

Two solutes can be separated if their partition coefficients differ, preferably one being greater than unity (and remaining in the raffinate) and the other less than unity (and being extracted). Similar to chromatographic practice, we can define a separation factor (or quotient), a , as: (14.7) where A and B represent two analytes, usually defined so a has a magnitude greater than 1. Actually a is less useful than it may seem. The efficiency of an LLE also depends on the actual values of the partition coefficients. As a general rule, the separation is most efficient when the product of the two partition coefficients is unity.' Also, the volumes of the two phases can be varied to improve a separation. It has been shown"'." that the optimum value for the phase volume ratio, P , is (14.8) Distribution Coefficients Some solutes form dimers or higher oligomers in one or both of the solvents and others form ion pairs or complexes. When this occurs, the simple partition coefficient we have been using is not useful since all of the species need to be included in the equilibrium expression. For these situations the distribution coefficient K must be used:

K,

=

(conc. of all forms of A),;,rf (conc. of all forms of A),,,,

(14.9)

Acetic acid partitioning between water and an organic phase represents a system that includes dimerization and ionization, and it will be used to illustrate the need for distribution coefficients rather than the simpler partition coefficients. When acetic acid (HOAc) is partitioned between water

14.1

EXTRACTION

393

and benzene, two equilibria are established in addition to the simple distribution of the acid between the two phases. These are the formation of a dimer in the organic layer and the ionization in the aqueous layer. All of the reactions are summarized in the following diagram: Benzene layer Benzene layer

K'

2HOAc =(HOAc),

'I

If;p

HOAc

+ H,O

(14.10) = ' K,

H,O'+

OAc-

The dimerization equilibrium constant K, is (14.1 1) The ionization equilibrium constant K , is (14.12) The distribution coefficient, which takes into account all forms in which the acetic acid is present (acid, dimer, acetate ion), is

Substituting the expressions for K, and K , intoK,, the following relationship is obtained: (14.14) From Eq. (14.14) we can see that the partition coefficient depends on the pH of the aqueous phase and on the amount of acetic acid in the benzene phase. Thus K, depends on the total concentration of acetic acid, unlike a simple partition coefficient, which does not. Batch LLE Batch countercurrent extraction is used for selective extractions such as the separation of metal ions and complex organic mixtures. Often a one-step extraction using separatory funnels is sufficient, but sometimes a multistep extraction is required to isolate a larger percentage of the desired analyte. At least five considerations are involved in selecting the two liquid phases for an extraction.

394

SAMPLE PREPARATION

First, the two phases must be immiscible, although there is always some solubility of each phase in the other; most commonly one phase is aqueous and the other nonaqueous. Second, the phases should be presaturated with each other before use to prevent changes in volume. Third, the two phases must separate from each other quickly and without emulsion formation. Violent shaking of the two phases in the extraction vessel is not necessary and must be avoided or this action is likely to produce an emulsion. Sometimes the sample itself will serve as an emulsifying agent. To break an emulsion, try heating the separatory funnel by running hot water or steam over it. Other possibilities are centrifugation, adding neutral salts to the aqueous layer, adding an additional solvent, changing the pH slightly, altering the volume ratio ( p ) slightly, or filtering it through a hydrophobic membrane. Still another is to use a product such as Hydromatrix" or Isol~te,'~ where LLE is performed in a column prepared by dispersing one of the phases over an inert material such as diatomaceous earth. The volume ratio should be chosen to get the best separation, as indicated by Eq. (14.8). Furthermore, LLEs are more effective with several steps using a small volume of extractant rather than one large one. A fourth consideration in the workup of the sample is whether or not the sample can be recovered easily from the extractant (or raffinate). Even if recovery is not required, detection of the solute and determination of its quantity may be necessary. For example, if the measuring step were to be performed by ultraviolet absorption, the extractant should be transparent in the ultraviolet. Finally, we have to consider the rate at which the system comes to equilibrium. Fast reactions are desirable, but some substances are slow in coming to equilibrium, and sometimes the rates are different in going from raffinate to extractant and vice versa. One procedure that has been suggested to speed up extractions is the formation of a homogeneous phase by raising the temperature of the ~ y s t e r n . 'On ~ cooling, the two-phase system is reestablished and equilibrium is quickly attained. The equations given above apply to one step extractions and to each step of a multistep process. A multistep extraction can be done in either of two ways: 1. The original sample (raffinate) is extracted several times and the

extractant solutions are combined into one solution for further analysis; this is an example of cross-current extraction, 2. The extract from the first step is put into a second separatory funnel and both funnels are reextracted. This is an example of countercurrent extraction, and it results in a distribution of analytes in many funnels, producing zone broadening similar to the effect in chromatography.

14.1

EXTRACTION

395

Probably the best way to gain an understanding of zone broadening in countercurrent LLE is to consider a simple example. For this purpose we shall use a single substance rather than a mixture since we want to study zone broadening per se. For our “system” we shall use a series of common separatory funnels and two immiscible solvents that will be referred to as extractant and raffinate. For simplicity, the volumes of the two solvents will be equal in each funnel, for example, 50 m L of each. The sample will be called A, and the sample size will be taken as 100 mg. We shall assume that sample A is preferentially soluble in the raffinate and has a partition coefficient in this system of 2.0 (at some constant but unspecified temperature). Then (14.15) The countercurrent process is depicted in Figure 14.3 as a series of five separatory funnels. In the first step (zero transfer) 50 m L each of extractant and raffinate (no further sample will be added to the system). Thus all of the sample is initially in the first funnel. Funnel 1 is shaken and allowed to come to equilibrium, at which time, according to the partition coefficient, (14.16) In other words, 33.3% of the sample is extracted and 66.7% remains in the raffinate. The In the first transfer, the extractant (33.3 mg of A in 50 mL) is moved to the second funnel. A 50-mL quantity of fresh extractant is added to the first funnel (which still contains the raffinate and 66.7 mg of A) and 50 m L of fresh raffinate are added to the second funnel.’ Now both funnels 1 and 2 are shaken and allowed to come to equilibrium. The new amounts of A in each phase in each funnel are indicated in the figure. The rounding off of the numbers produce some ratios that are not exactly 2.0 : 1.O; for example, the amounts in funnel 1 in transfer 1 are 45:22. This process is repeated as shown in Figure 14.3, moving extractant on to the next funnel and adding fresh extractant to funnel 1 and fresh raffinate to the newest funnel. For a final example consider what occurs in funnel 2 in transfer 2. After transfer 1, 22.2 mg of A are in the raffinate (upper phase) in funnel 2 and will remain there for the next step, and 22.2 mg of A in the extractant are added ‘ I n Figure 14.3 the upper phase (raffinate) stays in the same funnel and is, in effect, transferred to the one below it, while the lower phase (extractant) is moved on to the next funnel. The quantity of A transferred is indicated on the arrows.

396

SAMPLE PREPARATION Funnel number r

rransfer

lumber, n,

1

2

4

3

5

Figure 14.3. Example of stepwise countercurrent liquid-liquid extraction. K , = 2.0; p = 1.0. Reprinted with permission from J. Miller, Separation Methods in Chemical Analysis, John Wiley & Sons. Copyright 1975; this material is used by permission of John Wiley & Sons, Inc.

from funnel 1. The total amount of A in funnel 1 in transfer 2 is 44.4 mg, and it is redistributed according to the partition coefficient: 29.6

( K p ) * = 2.0 = 14.8 ~

(14.17)

The appropriate numbers are entered in Figure 14.3 after rounding off. Table 14.2 summarizes the amount of A in the extractant phase in each funnel after each of four transfers (five equilibrations). The process can be carried on further, of course, but this is sufficiently far enough to observe what is happening.

14.1

EXTRACTION

397

Table 14.2 Amount of A in the Extractant Phase in Each Step of a Five-Funnel LLEa

Funnel Number

Number of

Transfers 0 1 2

3 4 " K p= 2.0;

1

2

3

4

5

33.3 22.2 14.8 9.9 6.6

0 11.1 14.8 14.8

0 0 3.1 1.4 9.9

0 0 0 1.2 3.3

0 0

13.2

0 0 0.4

Total Amt. of A in Extractant (mg) 33.3 33.3 33.3 33.3 33.3

p = 1.0; starting with 100 mg A.

We can make three important observations: 1. The sample is now distributed in five funnels, whereas it began in only one funnel; broadening has indeed occurred. 2. The distribution of the sample is nonsymmetrical. 3. The largest amount of sample is in the second funnel. Since all of the sample was in funnel 1 at the beginning, we can see that the sample is moving down the row of funnels, but slowly, since K , > 1. If additional steps are added to this model, the distribution of solute in the funnels becomes symmetrical and the plot follows the binomial expansion:

(14.18) where n , is the number of transfers. Note that the first term in Eq. (14.18) is the fraction unextracted (1 - (I and ), the second is the fraction extracted, Q. As with a one-step extraction, separations are effected when the analyte has a K , sufficiently different from the distribution coefficients for the other solutes. Since each solute will be distributed among several funnels used in the process, isolation of the analyte of interest may not be as easily achieved as originally expected. More steps may be required than is easily accomplished. Therefore, Craig et al.lS designed an apparatus to facilitate multistep extractions on a semiautomatic basis. It was commercially manufactured, having as many as 1000 funnels, and on its way to extensive adoption when HPLC was introduced and made it obsolete. If chromatography had not been developed in the 1950s, multistep LLE would have become the method of choice for many separations. Now, our interest in LLE is mainly for sample preparation, and chromatography is the method of choice for the actual separation.

398

SAMPLE PREPARATION

Multistep LLE is quite similar to LLC. If each separatory funnel in LLE were called one plate, this model would represent the plate theory that was used in the early days to describe the chromatographic process. However, the plate theory has been abandoned in favor of the rate theory (see Chapter 3), which provides more useful information about the chromatographic process. Liquid-Liquid Microextraction (LLME) As with other techniques, there have been attempts to use smaller and smaller amounts of solvent for LLE. However, it was not until a study in 1996 reported the use of an 8-pL drop that attention became focused on the use of very small drops of extractant." Very quickly a method developed using a conventional microsyringe to suspend an even smaller drop (1 or 2 pL) of solvent in a sample solution to perform a liquid-liquid extraction." The technique functions much like SPME, which is discussed later in this chapter, because the extracted sample can be taken directly to a gas chromatograph and injected. Because of its similarity to SPME, it became known as single drop microextraction (SDME), but it is also called liquid-phase microextraction (LPME) and solvent microextraction (SME) as well as LLME. The basic procedure is very simple: A I - p L drop is suspended from the tip of a 10-pL microsyringe in a small vial containing the sample. The sample solution should be stirred and can also be heated. After the extraction has proceeded to equilibrium or near equilibrium (usually several minutes), it is withdrawn into the syringe, which is then withdrawn from t h e vial, taken to a GC, and injected. The extractant has to be volatile enough to serve as a chromatographable solvent and be separated from the analytes extracted from the sample.When an internal standard is used, precision is better than 5%. A recent example of the use of SDME for the determination of endosulfans in water by GC-ECD includes discussion of details such as optimization of drop size, sorption profiles, and effects of stirring, temperature, and salt addition.19 For extraction, a 1.5-pL drop of isooctane was used. The procedure can also be used for headspace (HS) analysis, and successful automation of the HS technique has been reported.2" A review of SDME, which appeared in 2002, summarizes its development."

''

Continuous LLE In the common laboratory procedures for continuous LLE, the volume of extractant is held constant and recycled in a cross-current process. Two types of apparatus are required, depending on which phase is heavier. Figure 14.4 shows both types. The extractant is heated in the flask. vaporizes, and is condensed so that it contacts the raffinate phase as it returns to the reservoir. Just as in the stepwise process, the law of diminish-

14.1

EXTRACTION

399

Figure 14.4.Apparatus for continuos liquid-liquid extraction: ( u ) extractant lighter and ( h ) raffinate lighter. E = extractant, R = raffinate. Reprinted with permission from J. Miller, Srpurulion Merhods in Chemical AnulJris, John Wiley &: Sons.Copyright 1975; this material is used by permission of John Wiley & Sons, Inc.

ing returns suggests that the extraction should not be continued for a long time. As the concentration of solute in the extractant increases, an equilibrium value is reached beyond which further extraction produces no change.

Liquid-Liquid Extraction with Membranes A variety of membranes have been used to separate two liquids that may not be immiscible and cannot be used for normal LLE.22,23 Unlike dialysis, in which the membrane has pores whose sizes provide the basis for a separation based on sieving, these membranes are not chosen according to pore size, and many of them are

400

SAMPLE PREPARATION

Table 14.3 Types of Membrane Extractions

Name

Supported liquid membrane Microporous membrane LLE

Phase Combinations Year of First Abbreviations (Donor-Membrane-Acceptor) Reference SLM

Aqueous-organic-aqueous

1986

MMLLE

Aqueous-organic-organic or

1998

Polymeric membrane extraction

PME

Membrane extr. with sorbent interface

MESI

organic-organic-aqueous Aqueous-polymer-aqueous or organic-polymer-aqueousor aqueous-polymer-organic Gas-polymer-gas or liquid-polymer-gas

1990

1992

Source: Adapted from reference 22.

nonporous. Table 14.3 lists four common modes of use that have been developed. A simple holder has two halves, each with a channel cut into it, and then clamped together with the membrane in the middle, as shown in Figure 1 4 . 5 ~ .The membranes listed in Table 14.3 are either polymers or (porous) polymers saturated with an organic phase. The wide range of possible membranes and membrane combinations provides great selectivity. The membranes themselves are hydrophobic; some examples used in microporous membrane LLE (MMLLE)24 are PTFE (polytetrafluoroethylene) materials made by Millipore Corp. and Spectrum Medical. More recently. hollow fibers have been used as the membrane material for bioanalytical sample preparation2s and for trace-level water pollution studies.*(‘ One example of a supported liquid membrane (SLM) uses two aqueous solutions, one acidic and one basic, separated by a hydrophobic porous membrane whose pores are filled with an organic compound such as a long-chain hydrocarbon, a large ether, or an organic phosphate. Bases in the basic aqueous solution can be extracted into the acidic layer by a protonation mechanism (see Fig. 14.5b). A concise summary has been published,” and considerably more detail can be found in references 22 and 23. Similar separations can be achieved between a solution and a vapor space occupied by sweeping carrier gas. After analytes are trapped from the carrier gas, they can be desorbed and run by GC. This technique is known as membrane extraction with a sorbent interface (MESI).*’

Final Comments Liquid-liquid extraction is an old established method. Simple one-step extractions using separatory funnels are included in many analytical procedures. Usually, however, these extractions are not quantitative, and either multiple steps are required or the exhaustive, continuous

14.1

EXTRACTION

401

Figure 14.5. Membrane devices for LLE. ( a ) Basic configuration, ( h ) SLM configuration, ( c ) MMLLE . Rcprintcd with pcrmission from LC-GC No. A m . , Vol. 21(.5), May 2003, pp. 424-438. LC-GC No. Am. is a copyrightcd publication o f Advanstar Communications Inc. All rights reserved.

process must be used, both of which are very time consuming. As a consequence, newer versions of LLE such as the membrane methods are becoming popular. Another method, which uses diatomaceous earth particles to facilitate continuous extractions between aqueous biological samples and an organic extractant, has been applied to the automatic extraction of plasma samples prior to LC/MS/MS analysis2* Simultaneous distillation-extraction (SDE) is the name of another special extraction process that has been used prior to GC analysis by the flavor and fragrance industry since 1964. It uses a unique apparatus that combines extraction with steam distillation. The study of Pollien and Chaintreau*’ summarizes the developments through 1997 and describes a new apparatus for larger scale sample preparation.

402

SAMPLE PREPARATION

For other extraction information, the newer books by Sekine and Hasegawa,30 Dean ,3' and Handley32 can be consulted. Smith' and Mitre" have already been cited as good general references on sample preparation methods, including LLE. Liquid-Solid Extraction

The classic liquid-solid extraction (LSE) method is a continuous method using an apparatus called a Soxhlet extractor. A solid sample is placed in a paper thimble and the solvent is passed over it, usually by distilling it from a sample flask somewhat like the extractors shown in Figure 14.4. The sample is extracted by condensed solvent, which is recycled in the distillation. The technique is simple but time consuming. Commercial equipment 33 is available for automating the process and decreasing the time required. However, the sample is subjected to boiling solvent before being washed, cooled, and concentrated.' Other modifications have resulted in a variety of newer methods that use pressurization, ultrasonication, and microwave irradiation to improve the process. Each of these will be described briefly. Assisted Liquid-Solid Extraction

Increasing the temperature and/or the pressure of the extracting liquid are obvious ways to assist the extraction process. When these parameters are increased enough, the solvent reaches its critical point and beyond; the method is then called supercritical fluid extraction (SFE). More information about supercritical fluids is included in Chapter 6, which discusses supercritical fluid chromatography (SFC). It should be noted that many extractions have been described in which the pressure is high but below the supercritical point; in some cases the difference has not been made clear, and these extractions have been classed as SFEs. This failure to distinguish between SFE and high-pressure extraction can cause confusion. In addition, the use of carbon dioxide has become associated with SFE and SFC, to the exclusion of other solvents. Therefore, to encourage scientists to differentiate between subcritical and supercritical extractions, pressurized fluid extraction (PFE) is often used to denote subcritical extraction conditions. It should be noted that the EPA sanctions the use of the term PFE as opposed to ASE, because the latter is a trademark. The following discussion follows the classifications most often found in the literature, and SFE will be discussed later. The use of subcritical hot water was first proposed in 1995.34It was found that as water is heated, its polarity changes from very polar to only slightly polar, thus making temperature a very important extraction parameter. At

14.1

Shut-off valves

EXTRACTION

403

Preheat coil

Cooling loop

\

Cooling water Needle valve Collection vial Extraction water Collection solvent

Figure 14.6. Schematic diagram of subcritical water extraction using cooling/collection system. Reprinted with permission from reference 34. Copyright 1995 American Chemical Society.

the higher temperatures it becomes very effective for extracting nonpolar organics such as polychlorinated biphenyls (PCBs). The apparatus originally proposed for extracting PCBs from soil34 is shown in Figure 14.6. HPLC columns are used to contain the sample, and HPLC pumps provide the pressure necessary to pump the hot water through the column. The next year Dionex published a study using a similar apparatus but calling their technique by their trademark, ASE, accelerated solvent extraction.” The solvents they used were a 1 : 1 mixture of methylene chloride and acetone, a 1 : 1 mixture of hexane and acetone, and pure methylene chloride. Recoveries of volatile hydrocarbons ranged from 97 to 100%. Samples can be processed automatically in batches of up to 24 using their commercially available equipment.”’ Other manufacturers include Supelco” and Applied Separations.3x PFE has been reviewed by Richter.3y Ultrasound and microwaves have also been used to assist in the extraction process, as they have for other laboratory tasks. Microwave-assisted extraction (MAE) or microwave-assisted solvent extraction (MASE) can be accomplished in a microwave oven. The original work reported in 19864” used samples in vials in a conventional microwave oven. Short (30-s) heating times were used to prevent boiling, and the process was repeated several times if necessary. The extractions of polar compounds into a 1 : 1 mixture of methanol and water were better than those done by Soxhlet extraction and took much less time. Extractions of nonpolar fats into hexane was less dramatic, probably because nonpolar solvents do not absorb microwaves as well as polar solvents do. Later workers used sample vessels that were sealed (and pressurized) and solvents that will absorb the radiation, for example, a 1 : 1 mixture of hexane and acetone. Automated, computerized commercial unit 4 1 , 4 2 are now available and provide for stirring as well as monitoring of the temperature and pressure. A typical study4’ demonstrates how much better MAE performs with this equipment for the extraction of organic compounds from

404

SAMPLE PREPARATION

standard reference soils and sediments compared with competitive methods such as Soxhlet. The problem with the inability of low dielectric solvents to absorb microwave radiation, partially solved by using mixed solvents, has also been addressed by adding a microwave-absorbing solid absorber to the sample A review in 2000 contains 139 references,” and a general discussion of the topic was presented in 2001.4hSeveral books have also been publi~hed.~’.~’ Static sonication is also effective in aiding extraction. A dynamic method has been described for the extraction of organophosphate esters by using a cell (LC column) immersed in a temperature-regulated sonic bath and pumped with solvent by an HPLC pump.4y More recently, a flow injection manifold has been adapted for LLE without phase separation.”’ Supercritical Fluid Extraction Supercritical fluids were introduced and defined in Chapter 6. That chapter proceeded to discuss SFC, which provided the impetus for analytical chemists to try SFE in their laboratories. Previously SFE had been used only for commercial extractions such as the decaffeination of coffee. As with SFC, the solvent used in SFE has almost exclusively been carbon dioxide, and SFE suffers from the same limitation as SFC: C 0 2 is not very polar. The partial solution is the same: Modifiers such as methanol, isopropanol, or acetonitrile have been added to the CO,. Figure 14.6 shows a schematic of a typical commercial apparatus.” It can be operated in either a static or a dynamic mode. In the dynamic mode, the flow of extractant is continuous; it expands and exits the system at the top in Figure 14.7, where it

Figure 14.7. Schematic diagram of a generic SFE system. Copyright 1994 Agilent Technologics, Inc. Reproduced with permission.

14.1

EXTRACTION

405

is collected in a suitable solvent or is deposited on a sorbent material (as depicted). A review in 1998 of SFE instruments described nine laboratory models,” and a later reviewj3 compares the collection devices for analyticalscale SFE. Supercritical fluid extraction has been used to determine fats in food,’4 replacing the older Soxhlet method. Several EPA methods, including one for polynuclear aromatics (PNAs), use SFE, although its popularity for this application has been declining due to matrix effects in different types of environmental samples. On the other hand, the speed, ease of use, and low organic solvent usage have been found to be ideal for extracting potential migrants from paper and board used for food packaging.’5 Smith’s review5’ and an earlier issue of the Journal of Chromatography A” contain articles on SFE as well as SFC. There are several books that contain additional applications as well as principles of operation.”-“ Solid-Phase Extraction Solid phase extraction is used to clean up samples” by sorbing impurities on a solid phase contained in a column or tube while the analyte is eluted from the tube. Alternatively, the analytes can be sorbed on the solid, allowing the rest of the sample to pass out of the tube. Either way, the impurities are separated from the analyte. The sample is usually a liquid or a solid dissolved in a solution. The name SPE, coined in 1982, distinguishes it from LSE (see above) in which the sample is a solid being extracted with a liquid. Solid-phase extraction became popular when prepacked, disposable cartridges were marketed in the late 1970s. A typical tube is shown in Figure 14.8, and the process (impurities being retained) in Figure 14.9. The process has similarities to liquid chromatography, especially affinity chromatography, and many solid stationary phases used in HPLC find use in SPE. Majors4 found that, just as in HPLC, the reversed-phase mode is the most popular, and ODS (octadecylsilyl) was used by about 25% of the respondents,* followed by C,. A typical apparatus for manual operation is shown in Figure 14.10. Robots can be used to automate the process, and completely automated instruments are also available. One study describes a unique combination of HPLC-MS with SPE but without HPLC columns for drug analysis.6’ Because the principles of operation are similar to those for HPLC, method development for SPE is also similar to that for HPLC, and information is provided in some of the studies cited. A convenient guide has been prepared by S u p e l ~ o , and ~ ~ a number of other vendors distribute SPE bibliographies.hi -” ‘These data were normalized; workers use more than one mode, so total usage adds up to over 100%.

406

SAMPLE PREPARATION

Figure 14.8.Typical design of an SFE cartridge. Reprinted with permission from LC-GC, Vol. 4(10), October 1986, pp. 972-984. LC-GC is a copyrighted publication of Advanstar Communications Inc. All rights reserved.

Figure 14.9. Steps in the operation of SPE. = interference; I = analyte of interest. Reprinted with permission from LC-GC, Vol. 4(10), October 1986, pp. 972--84. LC-GC is a copyrighted publication of Advanstar Communications Inc. All rights reserved.

To speed up sample processing, for example, to support combinatorial chemistry, SPE solids have become available in other formats." For example, thin disks that look like those used for filtering allow faster flows than the original cartridges because of their proportionally larger surface area, making them very popular for analyzing trace organics in water. A liter of aqueous sample that takes 1-2 h to process by SPE cartridges takes only about 15 min using disks. The well plates used for microtiter applications have been

14.1

EXTRACTION

407

Syringe for loading sample SPE cartridge

Collection tube rack

Solvent wash bottle

Vacuum manifold

Vacuum flask

Figure 14.10. Apparatus for SPE. Reprinted with permission from LC-GC, Vol. 4(10), October 1986, pp. 972-984. LC-GC is a copyrighted publication of Advanstar Communications Inc. All rights rcscrvcd.

adapted for many analytical uses, including SPE. A typical well plate has an 8 X 12 configuration, making a total of 96 wells, each of which is flow-through just like the SPE cartridges they resemble, facilitating automation and increasing output. The eluent containing the analyte can be collected in a 96-well tray similar to trays used in chromatographic autosamplers. Validation of a method using 96-well plates, automated SPE, and HPLC/MS/MS for high-throughput bioanalysis is typical of the current applications.” The use of SPE has become widespread, and it continues to grow and find new uses. Batch-to-batch reproducibility appears to be the most important parameter in the selection of an SPE d e ~ i c e ,similar ~ to the problems experienced in HPLC (see Chapter 8). Other drawbacks include the high cost of disposable materials and the limited selectivity, especially for polar compounds. One novel approach to the selectivity problem is the use of molecular imprint phases, MIPS.’” Several books are available for more details7’-73 on SPE. Solid-Phase Microextraction The technique known as SPME is not just a miniaturization of the SPE apparatus just discussed. It involves the use of solid phases that will sorb analytes as SPE phases do, but these phases are attached to a supporting injection device that resembles a microsyringe, and these solid phases are immersed in the sample. Figure 14.11 shows one commercial device, and Figure 14.12 illustrates how it can be immersed in a sample. It is most often used for GC sampling since the device can be used like a microsyringe to inject samples. For use with HPLC, an “in-tube” configuration is used.74

408

SAMPLE PREPARATION

Figure 14.11. Cutaway diagram of SPME holder and fiber. Reprinted with permission o f Supelco, Bellefonte, PA 16823.

The technique, introduced in 1990 by Arthur and Pawli~zyn,’~ makes use of the types of materials used as stationary phases in capillary GC, such as polydimethylsiloxane (PMDS) or Carbowax. Instead of an open tubular column with the coating on the inside, the SPME samplers have the coating on the outside of a fused-silica fiber. When dipped into a solution containing analytes similar to those that could be separated on a GC column containing the same coating, these analytes are sorbed on the fiber and can be removed with it for injection into a GC. So, in fact, the SPME sampler is not necessarily a solid, but a sorbent (e.g., PDMS) coated on, or bonded to, a solid fiber. The sampling process and the parameters that need to be considered are similar to those described earlier for LLME: stirring, heating, time of sorption, and the like. The SPME sampler can also be used for headspace analysis, and a slightly different version is manufactured for

14.1

EXTRACTION

409

Figure 14.12. Stepwise SPME extraction procedure. Reprinted with permission of Supelco, Bellefonte. PA 36823

HPLC sampling. Desorption in a standard GC injection port is usually done splitless with a 0.75-mm i.d. liner at a high enough temperature to get quantitative recovery. GC autosamplers are available that accept SPME probes. Commercial fibers are also available with polyacrylate, Carboxen, or Carbowax coatings. For small molecules, the Carboxen fiber is often the best c h o i ~ e . ’New ~ sol-gel coatings have been reported77 that offer the possibility of improved fiber stability. The applications guide78 published by Supelco provides references to over 400 publications. The technique seems to work best for relatively nonpolar analytes extracted from water samples. A newer version of the technique has the active sorbent coated on the inside of the sample tube and is called the in-tube te~hnique.~’ Solid-phase microextraction has worked so well for so many applications that one wonders why it was not invented earlier. It is simple, relatively

410

SAMPLE PREPARATION

inexpensive, and well-adapted for use with chromatography, especially GC. One drawback is the fragility of the fibers, which have a limited usage of around 30 samples. Also, volatile analytes can be lost during transfer to the GC injection port. In addition to the voluminous literature available from S ~ p e l c o , ~Penton ' has written a chapter on SPE and SPME for GC,'" and a recent symposium, ExTech 2002, contained many papers on SPME that have now been published in the Journal of Chromatography A." Several books are also available.x2- x 4 Stir-Bar Sorptive Extraction

Another configuration for making solid-phase extractions from solutions is to put the sorptive phase (usually PDMS) directly on the stirring bar used to stir the solution. The stir bar has a much larger surface area than the SPME fiber, so it can sorb a greater amount of analyte. Since the first study in 1999,8s SBSE has become popular and a commercial product is now available." The disadvantage of the SBSE method is that it is not as easy to automate for desorption into a GC, but an autosampler is commercially available." Procedures and applications have been summarizedH7~ "; a typical application is the analysis of volatile phenols in wines.'" Other Extraction Methods

Two special extraction methods are the subject of a recent review."" They use restricted access media (RAM) and large particle supports that are finding use in online sample preparation of biological fluids. An example of a RAM is the internal surface reversed-phase supports discussed in Chapter 8, but many others are discussed in the review. Headspace Methods

If the analyte to be sampled is volatile, the process can often be facilitated by taking the sample from the headspace. Headspace (HS) denotes the vapor space above a confined solution or solid sample. Only volatile materials will be present in significant amounts in the HS, thus simplifying the matrix from which the sample is taken. Complex aqueous solutions such as blood and urine contain many potential interferences that are not volatile, so they are frequently sampled by HS methods. Since the HS samples are by definition volatiles, G C is usually the preferred method of analysis. Details concerning the analysis of volatile organic compounds (VOCs), including HS, can be found in the recent chapter by Slack et aL9'

14.2

DIALYSIS

411

The HS vapor above a sample can be sampled directly with a gas-tight syringe after a thermostated sample has been allowed to come to equilibrium. Elevated temperature is often used to promote a greater concentration of analyte in the vapor. This technique, known as static HS, can also be accomplished with an autosampler. Other variables that must be maintained for quantitative work include the pressure and the phase volume ratio, fi.y2.y' Dynamic HS sampling can be accomplished by sweeping the headspace gases through an adsorption tube to concentrate them. Known as purge and trap, this method can also be used to remove volatiles from liquids and solids. Commercial apparatuses are available and a brief guide has been published by Restek."' Many standard EPA methods use purge and trap; a recent publication evaluates it for MTBE (methyl-t-butyl ether) analysis in groundwater.y5Alternatively, the HS sample can be trapped directly on the end of a capillary column by cooling it, often called cryogenic focusing. One problem that arises with this method is the formation of a plug of ice in the column. A general discussion, including methods of water removal is included in Kolb's review of HS sampling with capillary columns."' As mentioned earlier, SPME and SDME methods are commonly used for HS sampling. In both cases, the partition coefficient between the vapor sample and the (solid or liquid) extractant will determine the quantity sorbed on the extraction device. In all likelihood, this amount will be different from that obtained by static HS sampling, and a judicious choice of extractant can be used to lower detection limits for these methods. Some recent applications of SPME are the analysis of phenols in wine,"' amphetamines in urine,"' ethanol in and volatiles in food.""' A typical HS example of SDME is t h e microcxtraction from aqueous solutions using octanol.""

14.2 DIALYSIS The two phases used in dialysis are both liquids (as in LLE), but they are separated by a semipermeable or porous membrane. As a result, the phases need not be immiscible. The porosity of the membrane is chosen so that only some solutes can pass through it, and this selective permeation is the basis of separation. The mechanism is very different from that described earlier for membrane separations. The pore sizes of dialysis membranes are usually characterized according to their molecular weight cut off (MWCO), defined as the molecular mass of the smallest compound that is at least 90% retained by the membrane."'* The phenomenon is somewhat analogous to size exclusion, except that in dialysis the small molecules go through the membrane, whereas in SEC they diffuse in and out of the pores of the stationary phase.

41 2

SAMPLE PREPARATION

Dialysis is seldom used in the analytical laboratory to make a multicomponent separation. Its main use is in concentrating, desalting, and purifying such materials as proteins, hormones, and enzymes. By proper choice of the membrane pore size, the MWCO, large molecules that might interfere with a subsequent analysis step are retained and separated from the desired analytes. Alternatively, the retained large molecules can be analyzed after low-molecular-weight components have been removed. Dialysis is simple, but it is slow and not very selective. It can be carried out in a batch mode, prior to chromatographic analysis. But for faster analysis, it can be coupled online through a six-port valve to an HPLC instrument or to a capillary electrophoresis (CE) instrument. ' 0 2 Attempts have been made to increase the speed of dialysis as well as its selectivity. One possibility is to use a membrane in the form of a hollow fiber, which offers the advantages of speed and compact size. It is even possible to use small hollow fibers for in vivo sampling, for example, in neurochemical studies. The small hollow fibers can be inserted into organs or tissues to get samples that are free of large biomolecules. More details can be found in the review of microdialysis sampling for pharmacokinetic applications.Io3 The so-called perfusate can then be directly connected to an HPLC injection valve,'"4 similar to the set up for analytical scale analysis.If the sample contains electrolytes, some of which are not able to pass through the membrane, the Donnan membrane theory predicts an unequal distribution of the diffusible electrolyte on the two sides of the membrane. Three other factors that complicate this simple mechanism are adsorption, the presence of an electrokinetic charge on the membrane, and osmosis. The membrane itself is a major factor in determining the extent of all three of them, and the solute affects the degree of adsorption and osmosis as well. If ionic analytes are to be separated, electrodes can be placed in each of the two dialysis liquid chambers, giving rise to electrodialysis in which the driving force is the electrical field as well as the size exclusion. For typical cellulose-based membranes, the voltage is usually 10 V or less. Enrichment can be much better than for dialysis alone.

14.3 DERlVATlZATlON There are many reasons for performing a chemical reaction on a sample to form derivatives. Two reasons can be identified as beneficial for chromatographic analysis: the derivatization improves either the analysis or the detectability.

14.3 DERlVATlZATlON

413

A major method for improving a GC analysis by derivatization is by forming volatile compounds from nonvolatile samples. Often the derivatives will also be more stable thermally, which is necessary for GC analysis. Examples of these reactions are given later. A wide variety of other reactions have been reported whereby the derivatized analytes are more easily separated than the original ones. One major example is the formation of diastereoisomers, which makes possible the resolution of enantiomers, as discussed in Chapter 15. Improved detectivity usually arises from the incorporation of a chromophore into the analytes. This is most often used in HPLC, where the derivatives are designed so that they will have UV absorption or will fluoresce. In GC, an analogous example is the incorporation into the analytes of functional groups that will enhance their detectivity by a selective detector such as the ECD. The purpose of forming the derivatives is to improve the limit of detection, selectivity, or both. Another example is the use of deuterated reagents to form derivatives that can be easily distinguished by their higher molecular weight when analyzed by GC/MS or LC/MS.

Methods The methods of derivatization can be divided into two sets of categories: preand postcolumn methods, and off-line and online methods. For example, the formation of volatile derivatives for GC is usually prepared off-line in separate vials before injection into the chromatograph (precolumn). There are a few exceptions where the reagents are mixed and injected together; the derivatization reaction occurs in the hot GC injection port (online). Precolumn reactions that do not go to completion will produce mixtures that are more complex than the starting sample. As a result, excess reagent is usually used to drive the reaction to completion, thus leaving an excess of the reagent in the sample. Unless a prior separation step is used, the chromatographic method must be designed to separate these additional impurities. When performed off-line, the precolumn reactions can be used with slow reactions and can be heated. Postcolumn methods usually provide detectivity and are run online. The classic example is the ninhydrin reaction with amino acids. For best results, the reaction must be fast and the mixing chamber efficient without introducing excessive dead volume. Most of the examples are in HPLC. In the United States, several companies have specialized in supplying derivatization reagents, and they provide considerable information about derivatization in their literature.'"', "I6 The examples are so numerous that

414

SAMPLE PREPARATION

Table 14.4

Guide to Derivatizationa

Functional Group

Method

Derivatives

Acids

Silylation Alkylation

RCOOSi(CH,), RCOOR'

Alcohols and phenolsunhindered and moderately hindered

Silylation

R-0

Alcohols and phenolshighly hindered

Si(CH,),

-

0

II

Acylation Alkylation

R- 0- C-PFA R-0-R

Silylation

R- 0- Si(CH,), 0

Arnines (1 and 2")

Amines (3")

II

Acylation Alkylation

R -0-CR-0-R'

Silylation

R -N-SXCH,), 0

Acylation Alkylation

R-N-C-PFA R-N-R'

Alkylation

PFB carbamate

PFA

I1

0

Amides

Amino acids

II

Silylation (a)

(a) RC-NHSi(CH,), (unstable) 0 0

Acylation (b)

(b) RC-NH-C 0

Alkylation (c)

(c) RC-NHCH,

I1

II

PFA

II

Esterification/acylation Silylation (a)

(a)

Acylation + silylation (b)

(b)

Alkylation (c)

(c)

RCHOOSi(CH

I

1,

N- SXCH 3 ) 3 RCHOOSKCH ,I,

I

N -TFA RCHCOOR'

I

NHR'

415

14.3 DERlVATlZATlON

Table 14.4 (Continued)

Functional Group

Catecholamines

Method

Acylation

+ silylation (a)

Derivatives H R- N-

HFB

(a) 0 Si(CHJ, 0 Si(CH,),

q(

H R-N-HFB

Acylation (b) (b)

Carbohydrates and sugars

(a)

Silylation (a)

Acylation (b)

OHFB OHFB OSi(CHJ,

I

-(CH

)r

OTFA

I

(b)

-(CH,),

Alkylation (c)

-

OR

(c)

-(CH Carbonyls

-

Ir

Silylation

TMS -0- N=C

Alkylation

CH,-O-N=C

Source:Courtcsy of Regis Chemical. “Ahhrcviations: TMS = trimcthyl silyl; PFA hcptafluorobutyryl.

=

pcrfluoroacyl; TFA

=

/ \ / \

tritluoracetyl; HFB

=

many books have been published on the subject “ ) 7 ~ ’ 1 ;i the following discussion will highlight only the major uses.

Examples T h e two major examples of derivatization in chromatography are the formation of volatiles for GC and the creation of UV detectivity o r fluorescence in LC. Each will be discussed briefly.

416

SAMPLE PREPARATION

Volatile Derivatives for GC The reactions to produce volatile derivatives can be classified as silylation, acylation, alkylation, and coordination complexation. Examples of the first three types are included in Table 14.4 and include the hydroxyl, amine, and carbonyl functional groups. The fourth reaction type, coordination complexation, is used with metals; typical reagents are trifluoroacetylacetone and hexafl~oroacetylacetone."~Drozd' has reviewed the field and provided over 600 references. Silylation reactions are very popular and need further description. A variety of reagents are commercially available, and most are designed to introduce the trimethylsilyl group into the analyte to make it volatile. A typical reaction is the one between his-trimethylsilylacetamide (BSA) and an alcohol:

''

R-OH

/

+ CH,-C

\

0 -SXCH 3)3

-

N-SXCH,),

R-O-Si(CH,),

/

+ CH,-C

\

OH (14.19)

N-Si(CH,),

A closely related reagent contains the trifluoroacetamide group and produces a more volatile reaction by-product (not a more volatile derivative); the reagent is bis(trimethy1silyl)-trifluoroacetamide (BSTFA). If a solvent is used, it is usually a polar one, commonly the bases DMF (dimethylformamide) and pyridine. An acid catalyst such as trimethylchlorosilane (TMCS) and heating are sometimes needed to speed up the reaction. In general, the ease of reaction follows the order:"'" alcohols 2 phenols

Table 14.5

2

carboxylic acids

2

amines 2 amides

Silylating Agents for GC

Abbreviation

Reagent

TMSIM BSTFA BSA MSTFA TMSDMA TMSDEA TMCS HMDS

N-Trimethylsilylimidazolc

N,0-bis(trimethy1silyl)trifluoroacetamide N,0-bis(trimethylsilyl)acetamide

N-methyl-N-trimethylsilyltrifluoroacetamide N,N-dimethylaminotrimethylsilane Trimethylsilyldiethylamine Trimethylchlorosilane Hexame thyldisilazane

14.3 DERlVATlZATlON

41 7

Table 14.6 Derivatives for LC

Analyte Functional Group Acids, carboxylic

Alcohols Aldehydes Amines (I" and 2")

Reagents for UV Absorption

Fluorescence

0-p-nitrobenzy-N, N'diisopropyliso-urea (PNBDI) p-Bromophenacyl bromide (PBPB) 3,5-Dinitrobenzoyl chloride (DNBC) p-Nitrobenzyloxy-amine HCI (PNBA) N-Succinimidyl-pnitrophenyl acetate (SNPA) DNBC

4-Bromomethyl7-methoxycoumarin

SNPA p-Nitrobenzyl-N-npropylamine HCI PNB DNBC DNBC

Amino acids Isocyanates Ketones Phenols Thiols

Dansyl hydrazine 7-Chloro-4-nitrobenzo2-oxa-1,3-diazole (NBD chloride) Dansyl chloride, o-phthaladehyde (OPT) OPT -

Danysl hydrazine Dansyl chloride NBD chloride

Source: Regis Chemical; used with permission.

The order of reactivity of the reagents is TMSIM 2 BSTFA 2

2

BSA 2 MSTFA

2

TMSDMA 2 TMSDEA

TMCS 2 HMDS

See Table 14.5 for the chemical names corresponding to these abbreviations. Derivatives for HPLC Detection The lack of a universal detector for HPLC and the popularity of the UV detector have caused chromatographers to seek derivatization reactions that introduce UV chromophores into sample analytes. In those instances where the derivative also fluoresces, additional sensitivity can be obtained by fluorometric detection. Table 14.6 contains a list of the most common derivatizing reagents for this purpose. More details can be found in a recent review"' and a book on the subject.' "

418

SAMPLE PREPARATION

REFERENCES 1.

2. 3. 4. 5.

6.

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

16. 17. 18. 19.

20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31.

R. M. Smith, J . Cliromatogr. A 2003, 1000, 3-27. S. C. Moldoveanu, J . Chromatogr. Sci. 2004, 42, 1-14. R. E. Majors, LC-GC Guide to Sample Preparation, LC-GC Muguzinc Advanstar Communications, Eugene, OR, 2000. R. E. Majors, LC-GC No. Am. 2002, 20, 1098-1113. G. LeBlanc, A Review of EPA Sample Preparation Techniques for Organic Compound Analysis of Liquid and Solid Samples, LC-GC No. Am. 2001, 19, 1120-1130. S. Mitra, (ed), Sample Preparation Techniques in Analyiical Cliemistry, WileyInterscience, Hoboken, NJ, 2003. J. Pawliszyn, Anal. Cheni. 2003, 75, 2543-2558. J . Chromatogr. A 2004, 1025, 1-146. L. B. Rogers, J . Chem. Educ. 1968, 45, 7. M. T. Bush and P. M. Densen, Anal. Clieni. 1948, 20, 121. E. Grushka, Sepur. Sci. 1972, 7, 293. w w w .ansysinc.com/producis/cotisum /sampleprep/lle/hulk html. w w w .argotech.com /~~roducts/unrr~.sis/spe~colunins/liq~iid. html. K. Murata, Y. Yokoyama, and S. Iheda, Atiul. Chem. 1972, 44, 805. L. C. Craig, W. Hausmann, E. H. Ahrens, and E. J. Harfenist, Anal. Cliem. 1951, 23, 1236. M. A. Jeannot and F. F. Cantwell, Anal. Chem. 1996, 68, 2236-2240. M. A. Jeannot and F. F. Cantwell, A n d . Chem. 1997, 69, 235-239. Y. He and H. K. Lee, Anal. Chem. 1997, 69, 463444640, M. C. Lopez-Blanco, S. Blanco-Cid, B. Cancho-Grande, and J. Simal-Gandara, J . Chromatogr. A 2003, 984, 245-252. D. C. Wood, J. M. Miller, and I. Christ, LC-GC No. Am. 2004, 22, 516-522. E. Psillakis and N. Kalogerakis, Trends Anal. Cliem. 2002, 21, 53-63. J. A. Jonsson and L. Mathiasson, J . Chromatogr. A 2000, 902, 205-225. J. A. Jonsson and L. Mathiasson, in Advances in Chromatography, P. R. Brown and E. Grushka (eds), Vol. 41, Marcel Dekker, New York, 2001, Chapter 2. Y. Shen, J. A. Jonsson, and L. Mathiasson, Anal. Chem. 1998, 70, 946-953. 0 . B. Jonsson, U. Nordlof, and U. L. Nilsson, Anal. Chem. 2003, 75, 3 174-3 179. D. Kou and S. Mitra, A d . Chem. 2003, 75, 6355-6360. J. A. Jonsson and L. Mathiasson, LC-GC No. Am. 2003, 21, 424-438. S. X. Peng, T. M. Branch, and S. L. King, Anal. Chem. 2001, 73, 708-714. P. Pollien and A. Chaintreau, Anal. Chem. 1997, 69, 3285. T. Sekine and Y. Hasegawa, Solvent Extraction Chemistry, Dekker, New York, 1977. J. R. Dean, Extraction Methods for Environmentul Analysis, Wiley, New York, 1998.

REFERENCES

419

32. A. J. Handley (ed), Extraction Methods in Organic Analysis, Sheffield Academic Press, Sheffield, 1998. 33. www.foss.dk/c/p/solution.s/products/defrrult.us~~. Choose Soxtec from the Products Direct menu. 34. Y. Yang, S. Bowadt, S. B. Hawthorne, and D. J. Miller, Anal. Chem. 1995, 67, 4571-4576. 35. B. E. Richter, B. A. Jones, J. L. Ezzell, N. L. Porter, N. Avdalovic, and C. Pohl, Anal. Chem. 1996, 68, 1033-1039. 36. www .dion ex. corn . 37. www. signiu-uldrich. com. 38. Applied Separations, Inc., Allentown, PA. www.ui~i~liedscparatioiIs.com. 39. B. E. Richter, LC-GC Cur. Tretds Develop. Sanzple Preparation, 1999, I7(6S), S22-S28. 40. K. Gunzler, A. Salgo, and K. J. Valco, J . Chromatogr. 1986, 371, 299-306. 41. www. cem .corn /puges/mar.w. htm . 42. www.milestonesci. com. 43. V. Lopez-Avila, R. Young, and W. F. Beckert, Anal. Cheni. 1994, 66, 1097. 44. S. Shah, R. C. Richter, and H. M. S. Kingston, LC-GC No. Am. 2002, 20, 280. 45. C. S. Eskilsson and E. Bjorklund, J . Chromatogr. A 2000, 902, 227-250. 46. R. C. Richter, D. Link, and H. M. Kingston, Anal. Chem. 2001, 73, 30A-37A. 47. H. M. Kingston and S. J. Haswell (eds), Microwave Enhmced Chemistry, American Chemical Society, Washington, D.C., 1997. 48. M. D. L. de Castro and J. L. Luque-Garcia, Acceleration and Autoniation of Solid Sample Treatment, Elsevier, Amsterdam, 2002. 49. C. Sanchez, M. Ericsson, H. Carlsson, A. Colmsjo, and E. Dyremark, J . Chromatogr. A 2002, 957, 227-234. 50. J. Ruiz-Jimenez and M. D. L. deCastro, Anal. Chim. Acta, 2003, 489, 1-1 1 and 223-232. 5 1. www.ugilent.corn . 52. B. Erickson, Anal. Chem. 1998, 70, 333A-336A. 53. C. Turner, C. S. Eskilsson, and E. Bjorklund, J . Chromatogr. A 2002, 947, 1-22. 54. J. M. Levy, V. Danielson, R. Ravey, and L. Dolata, LC-GC 1994, 12, 920. 55. C. Nerin, E. Asensio, and C. Jimenez, Anal. Chem. 2002, 74, 5831-5836. 56. R. M. Smith, J . Chromatogr. A 1999, 856, 83-115. 57. J . Chromatogr. A 1997, 785, 205-403. (Comprised of 18 papers.) 58. B. Wenclawiak (ed), Analysis with Supercritical Fluids: Extraction and Chromatography, Springer, Berlin, 1992. 59. S. A. Westwod (ed), Supercritical Fluid Extraction and Its Use in Chromatographic Sample Preparation, Blackie, London, 1993. 60. M. D. L. de Castro, M. Valcarcel, and M. T. Tena, Analytical Supercritical Fluid Extraction, Springer, Berlin, 1994.

420

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

70.

71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84.

SAMPLE PREPARATION

M. A. McHugh and V. J. Krukonis, Supercriticul Fluid Extraction: Principles and Practice, 2nd ed., Buttenvorths, London, 1994. R. E. Majors, LC-GC 1986, 4 , 972-984. G. D. Bowers, C. P. Clegg, S. C. Hughes, A. J. Harker, and S. Lambert, LC-GC 1997, 25, 48-53. Supelco, The Supelco Guide to Solid Phase Extraction, 2nd ed., Supelco Inc, Bellefonte, PA, 1988. ww w .JTBa ker. com . www. Waters.com. www.Varianinc.com. R. E. Majors, LC-GC 2001, 19, 678-687. L. Yang, R. P. Clement, B. Kantesaria, L. Reyderman, F. Beaudry, C. Grandmaison, L. DiDonato, R. Masse, and P. J. Rudewicz, 1. Chromatogr. B 2003, 792, 229-240. F. Lanza and B. Sellergren, in Advances in Chromatography, P. R. Brown and E. Grushka (eds), Vol. 41, Marcel Dekker, New York, 2001, Chapter 4. See also, V. T. Remcho and Z. J. Tan, Anal. Chem. 1999, 71, 248A-255A; F. Chapuis, V. Pichon, F. Lanza, B. Sellergren, and M.-C. Hennion, J. Chromatogr. B 2004, 804, 93-101. J. S. Fritz, Analytical Solid-Phase Extraction, Wiley-VCH, New York, 1999. N. J. K. Simpson (ed), Solid-Phase Extraction: Principles, Techniques, and Applications, Marcel Dekker, New York, 2000. E. M. Thurman and M. S. Mills, Solid-Phase Extraction: Principles and Practice, Wiley, New York, 1998. R. Eisert and J. Pawliszyn, Anal. Chem. 1997, 69, 3140-3147. C. L. Arthur and J. Pawliszyn, Anal. Chem. 1990, 62, 2145. R. Shirey, Selecting the Appropriate SPME Fiber for Your Application Needs, Pub. T499232, Supelco Inc., Bellefonte, PA, 1999. S. Bigham, J. Medlar, A. Kabir, C. Shende, A. Alli, and A. Malik, Anal. Chem. 2002, 74, 752-761. Supelco, SPME Applications Guide, Supelco, Inc., Bellefonte, PA, 1999. H. Kataoka, S. Narimatsu, H. L. Lord, and J. Pawliszyn, Anal. Chem. 1999, 71, 4237-4244. Z. E. Penton, in Aduances in Chromatography, P. R. Brown and E. Grushka (eds), Vol. 37, Marcel Dekker, New York, 1997, Chapter 6. J. Chromatogr. A 2003, 999, 1-210. J. Pawliszyn, Solid Phase Microextraction: Theory and Practice, Wiley, New York, 1997. J. Pawliszyn (ed), Applications of Solid Phase Microextraction, Springer, Berlin, 1999. S. A. S. Wercinski (ed), Solid Phase Microextraction: A Practical Guide, Marcel Dekker, New York, 1999.

REFERENCES

421

85. E. Baltussen, P. Sandra, F. Davis, and C. Cramers, J. Microcolumn Sep. 1999, I I , 737. 86. www.gerstelus.com. Select Twister. 87. L. Nardi, Am. Lab. 2002, 34(1), 30-37. 88. F. David, B. Tienpont, and P. Sandra, LC-GC No. A m . 2003, 21, 108-118. J . Chromatogr. A 2004, 1025, 263--67. 89. J. Diea, C. Dominguez, D. A. Guillen, R. Veas, and C. G. Barroso, J. Chromatgr. A 2004, 1025, 263-267. 90. S. Souverain, S. Rudaz, and J.-L. Veuthey, J . Chromatogr. B 2004, 801, 141 156. 91. G. C. Slack, N. H. Snow, and D. Kou, in Sample Preparation Techniques in Analytical Chemistry, S. Mitra (ed). Wiley-Interscience, Hoboken, NJ, 2003, Chapter 4. -

92. B. Kolb and L. S. Ettre, Static Headspace-Gas Chromatography: Theory and Practice, Wiley-VCH, New York, 1997. 93. Restek, A Ttchnical Guide for Static Headspuce Analysis lising G C , Restek Corp., Bellefonte, PA, 2000. www.restekcoT.com. 94. Restek, Optimizing the Analysis of Volatile Organic Compounds, Restek Corp., Bellefon te, PA, 2003. www. rrstekcorp. com. 95. J. D. Evans and M. R. Colsman, LC-GC No. A m . 2003, 21, 42-52. 96. B. Kolb, J. Chrornutogr. A 1999, 842, 163-205. 97. R. C. Mejias, R. N. Marin, M. deV. G. Moreno, and C. G. Barroso, J. Chromutogr. A 2003, 995, 11-20. 98. N. Raikos, K. Christopoulou, G. Theodoridis, H. Tsoukali, and D. Psaroulis, J. Chromatogr. B 2003, 789, 59-63. 99. D. Zuba, A. Parczewski, and M. Reichenbacher, J. Chromatogr. B 2002, 773, 75-82. 100. C. Bicchi, C. Cordero, E. Liberto, P. Rubiolo, and B. Sgorbini, J. Chromutogr. A 2004, 102, 217-226. 101. A. Theis, A. Waldack, S. Hansen, and M. Jeannot, Anal. Chem. 2001, 73, 5651-5654. 102. N. C. van de Mcrbel, J. Chromatogr. A 1999, 856, 55-82. A review with 115 references. 103. M. I. Davies, Anal. Chim. Actu 1999, 379, 227-249. 104. W. C. Tseng, M. H. Yang, T. P. Chen, and Y. L. Huang, Analyst 2002, 127, 560. 105. Regis, GC Deriuatization, Regis Technologies, Inc., Morton Grove, IL, 2000. www.registech.com. 106. Fluka, Silylating Agents, Fluka Chemie AG, Buchs, Switzerland, www.sigmaaldrich.com.

1995.

107. A. E. Pierce, Silylation of Organic Compounds, Pierce Chemical, Rockford, IL., 1968. www.piercenet.com.

422

SAMPLE PREPARATION

108. J. F. Lawrence and R. W. Frei, Chemical Derivutization in L i p i d Chromatogruphy, Elsevier, Amsterdam, 1976. 109. K. Blau and G. S. King (eds), Handbook of Derivutives for Chromatograply. Heyden, 1 10. D. R. Knapp, Handbook of Analytical Derivatizution Reactions, Wiley, New York, 1979. 111 J. Drozd, Chemical Derivatization in Gas Chromatography, Elsevier, Amsterdam, 1981. 112. T. Toyooka, Modern Derivatization Methods for Separation Science, Wiley, New York, 1999. 113. G. Lunn and L. C. Hellwig, Handbook of Deriuatizution Reactions for HPLC. Wiley, New York, 1998. (Also available on CD.) 114. R. W. Mosier and R. E. Sievers, Gas Chromatography of Metal Chelute.s, Pergamon, Oxford, 1965. 115. J. Drozd, J . Chromatoy. 1975, 113, 303. 116. R. W. Frei, H. Jansen, and U. A. Th. Brinkman, Anal. Chem. 1985, 57, 1529A. 117. I. S. Krull (ed), Reaction Detection in Liquid Chromatography, Dekker, New York, 1986.

15 SPECIAL APPLl CAT1ONS With John G. Hoogerheide

Three special topics are of sufficient importance to be featured in this chapter: multidimensional chromatography, biological applications, and chiral separations. Included in the biological section is a discussion of affinity chromatography that was deferred from Chapter 8. A final section covers a few other topics to provide more complete coverage of chromatography.

15.1

MULTIDIMENSIONAL CHROMATOGRAPHY

When a conventional one-dimensional separation is not adequate to resolve all the components of a sample, it may be possible to devise a multidimensional process that will yield better resolution. The concept is to spread out the separation in two dimensions rather than just one. The earliest examples are of planar methods such as TLC and thin-layer electrophoresis. One example of two-dimensional TLC was presented in Chapter 11. Another combination that was exploited early in the development of two-dimensional separations involved the combination of TLC in one direction and electrophoresis at a right angle. These are both examples of planar separations

Chrornutogruphy: Coricepts und Contrusts, Second Edition. By James M. Miller ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

423

424

SPECIAL APPLICATIONS

where visualization of the two-dimensional process is easy, but online detection is generally lacking, and quantitative analysis is difficult. Further details on multidimensional planar separations can be found in the publications of Poole and Poole’ and Nyiredy.’ The rest of this section will deal with column methods only. Some chromatographers would consider the addition of a spectroscopic detector like MS to a chromatographic separation as an example of two-dimensional method since the MS adds a second dimension of new information. Some scientists prefer to use the term hyphenation to describe the combination of chromatography with spectroscopic detectors in contrast to coupled-column techniques, but others use the two designations interchangeably. Many investigators would include both of them in the broader classification of multidimensional analysis, and some would call GC/GC/MS a three-dimensional process. Chapter 10 contains discussion of GC/MS and LC/MS, but the focus in this chapter is on the coupling of column methods. The basic theory of two-dimensional separations was presented by Giddings, first in an article3 and then in Cortes’ book4 on the subject. Although the latter covers multiple dimensions, most of the actual applications are only two-dimensional. These two dimensions need to be orthogonal (at right angles) in order to maximize the two-dimensional space created. Indeed, Giddings’ definition is based on that premise.’ The number of peaks, n , that can be resolved in a two-dimensional separation is the product of the peak capacities of the two operations: n?L)= n , x nz = n ;

(15.1)

thus showing the advantage of two-dimensional methods. The limits to peak capacity were presented in Chapter 2. Usually these two separation processes are performed sequentially, although some pairs of techniques, such as chromatography and electrophoresis, can be run simultaneously. As opposed to planar methods, in which the components are distributed in two-dimensional space, two-dimensional column separations give components distributed in two-dimensional time. Column separations have the added advantage of online detection. All multidimensional separations require orthogonal techniques to be used in each dimension. Orthogonal methods are those that use different mechanisms or modes of action in the separation. For multidimensional column methods, each column should provide a very different type of separation. For example, in GC/GC, one column could be very polar and the other nonpolar. The combination of LC and G C would be considered two-dimensional by virtue of the large difference in their separating mechanisms. A two-dimensional LC separation might combine reversed phase (RP) in one column with size exclusion chromatography (SEC) in the second, thereby attaining true

15.1

MULTIDIMENSIONAL CHROMATOGRAPHY

425

orthogonality. A second requirement for true orthogonality is that components separated in the first dimension must not be recombined in the second dimension. Another term that is often used in this field is lzeart cutting, which refers to the process of isolating a poorly separated portion (a cut) of the molecules from the first dimensional process and running it on the second dimension. Heart cutting is not a true two-dimensional process since the whole sample is not run in the second dimension. It is still a valuable procedure and one that provided the first G C examples of improved resolution by the use of two dissimilar columns. T h e two-dimensional separation can be run off-line o r online, the latter being preferred for convenience, ease of automation, and improved data handling. T h e disadvantage to the online mode is that the mobile phases may be incompatible, as in a combination of R P with normal-phase (NP) HPLC; the former mobile phase is usually aqueous and the latter nonaqueous. This problem seldom occurs in GC, where the mobile phase can be helium for either polar o r nonpolar column separations, and GC/GC is a very common two-dimensional method. In o n e configuration, a n online detector can be placed after each column so the data from both of them can be combined into a two-dimensional display. This is often accomplished with two dissimilar columns in parallel, each with its own detector. T h e most interesting and effective recent multidimensional separations are online, use only one detector located after the second of two columns in series, and are truly orthogonal. T h e rest of this section will be primarily devoted to this type of operation; optimization of the separation is usually called tuning. Tuning can be achieved in several ways, including: Using o n e column of normal lcngth and o n e of very short length Adjusting column variables such as temperature and flow rate Using innovative valving o r modulation between the columns Using cryogenic trapping Using original software to make optimal use of complex data such as that obtained by MS detection Some examples will be included in the following discussion. Gas Chromatography/ Gas Chromatography

From the previous discussion it should already be clear that GC is the technique most easily adapted for two-dimensional separations. Much of the early G C / G C work involved heart cutting, which is not a fully two-dimensional method since only part of the sample is separated on the second

426

SPECIAL APPLICATIONS

Figure 15.1. Generalized schematic of a GC/GC multidimensional system: I

= injector; M =: modulator; D = detector. Reprinted from J. Dalluge, J. Bcens, and U. Brinkman, “Comprchcnsivc Two-dimensional Gas Chromatography: a Powerful and Versatile Tool,” J . Chromtiiogr. A 2003, 1000, 69-108 Copyrighr 2003, with permission from Ekeuier.

column; nevertheless, these methods can provide better separations for the part of a sample that is the most difficult to resolve. Coupling of columns is achieved with conventional valving, or a special type called the Deans switch, invented in 1968.s A pneumatic device, it switches flows based on relative pressures in the streams and can replace a conventional valve to facilitate heart cutting. A comprehensive summary of heart cutting G C / G C , including the use of Deans switching has been written by Lewis,6 and a brief history and some applications have been reviewed.’ A recent study’ describes another new microswitching procedure. For a true two-dimensional GC separation, a cryotrap or modulator is usually required,’, as shown in Figure 15.1. A fraction from the first column is isolated with the modulator and transferred to the second column. Since only a single detector is used, the separation in the second column has to occur in less time than it takes to elute a single peak from the first column. Usually it is desirable to take three or four samples from each peak from the first column, so the run time for the second column can be only a few seconds at most. Since the second separation must be very fast, the second column is short and is operated at a high temperature and fast flow rate. The interface/modulator between the two columns can take several forms’- I * ; the two most common ones are a thermal sweeper and a cryogenic trap. Several have been patented,’, l 3 and new ones are currently being p r ~ p o s e d . ’Is~ At , least two commercial instruments are available, including a

15.1

MULTIDIMENSIONAL CHROMATOGRAPHY

427

retrofit for an existing G C “ and a complete GC/GC/TOFMS instrument.” Reviews in 1999,Ix 2002,1” 2003,20 and 200421summarized the literature and have been accompanied by a publication” from the First International Symposium on Comprehensive Multidimensional GC, which was held in March 2003. Included in the latter are reports of modulation devices and many applications. One of the most impressive applications has been the analysis of cigarette smoke. Two studies using the Zoex modulator illustrate the power of this technique. In one case,23 over 30,000 peaks were isolated using GC/GC/TOFMS. In another,24 over 1000 compounds with S/N ratios of 100 or greater were separated and identified, 430 of which had good MS matches. This work required not only modulation but also new software that could make use of the MS isotopic abundances to aid in identifications. In effect, using an MS detector to give a GC/GC/MS system results in a separation that is indeed three dimensional. The diagrams of the data obtained by orthogonal GC/GC are quite complex and are usually presented as contour diagrams; Figure 15.2 is an example. For optimal viewing, color is often used, so this black-and-white reproduction cannot convey all the information adequately. Figure 15.3 shows a small portion of the two-dimensional contour plot (Figure 15.2) illustrating how that plot correlates with the associated second-dimension chromatogram. Many other applications are included in the references cited.

Liquid Chromatography/Gas Chromatography The combination of LC and G C is usually performed in the order indicated -LC being the first dimension. The interface between the systems must be capable of transferring the relatively large amount of mobile phase (MP) from the LC into the GC. As reported in Chapter 7, the use of large sample volumes in GC is one area that is being developed currently and helps facilitate this transfer. Evaporation is another option, but care must be taken to prevent the loss of volatile analytes. The relatively volatile solvents often used in normal-phase LC can be handled, but the aqueous phases used in RP, which may also contain nonvolatile salts, are more difficult, and as a consequence there has not been much development of RPLC/GC. Figure 15.4 summarizes the various LC/GC transfer t e c h n i q ~ e s ,sepa~~ rating them into NP and RP categories. The popular techniques are the three listed: (1) retention gap, also called on-column, (2) concurrent eluent evaporation, employing a loop-type interface, and (3) hot injectors, most often with programmed temperature vaporization (PTV). To divert the large volume of gas generated by vaporizing the LC solvent, types 1 and 2 are usually

Figure 15.2. Two-dimensional plot of GC/GC/MS chromatographic data of cigarette smoke. Reprinted from J. Dalluge, L. van Stee, X. Xu, J. Williams, J. Beens, R. Vreuls, and U. Th. Brinkman, “Unravelling the Composition of Very Complex Samples by Comprehensive Gas Chromatography Coupled to Time-of-Flight Mass Spectrometry,” J . Chromafop. A 2003, 974, 169-184, Copyright 2003, with permission from Elsevier.

Figure 15.3. (A) Expanded detail of a small section of Figure 15.2. The vertical line at 583 s indicates the second-dimension chromatogram, which is shown at the left (B). (C) is the deconvoluted mass spectrum of the second-dimension peak at 0.24 s, and (D)is the corresponding library spectrum. Reprinted from J. Dalluge, L. van Stee, X. Xu, J. Williams, J. Beens, R. Vreuls, and U. Th. Brinkman, “Unravelling the Composition of Very Complex Samples by Comprehensive Gas Chromatography Coupled to Time-of-flight Mass Spectrometry,” J . Chromatogr. A 2003 974, 169-184, Copyright 2003, with permission from Elsevier.

430

1

SPECIAL APPLICATIONS

NOUMAL-PHASE ELUENTS

I

UETENTION GAP TECHNIQUE

:[TCERID

REVERSED PHASE ELUENTS

L~RECT[~
CONCURRENT ELUENT EVAPORATION

, HOT INECTOUS

Figure 15.4. Classification of LC/GC transfer techniques. Reproduced from reference 25, Copyright 2002, John Wiley & Sons. This material is used by permission of John Wiley & Sons Ltd, Chichcster. England.

Figure 15.5. Basic instrumentation for online coupled LC/GC. Reprinted from T. Hyotylainen and M.-L. Reikkola, “On-line Coupled Liquid Chromatography-Gas Chromatography,” J . Chromatogr. A 2003, 1000, 357-384, Copyright 2003, with permission from Elsevier.

operated with a solvent vapor exit (SVE) placed between the precolumn and the analytical column. The basic instrumentation for LC/GC is shown in Figure 15.5. A recent review provides experimental details and applications. One disadvantage to LC/GC as a coupled technique is that GC cannot be used to analyze all compounds that can be separated and detected by LC. The biggest obstacle is that many compounds that can be separated in the LC

’’

15.1

MULTIDIMENSIONAL CHROMATOGRAPHY

431

dimension simply are not volatile enough to be analyzed in the G C dimension. In such cases, a more appropriate two-dimensional separation might be LC/LC. Liquid Chromatography/Liquid Chromatography

A variety of two-dimensional LC methods have been described"; many of them show similarities to the two-dimensional G C methods described earlier, including heart cutting. In designing an LC/LC procedure, however, one has to deal with the fact that the mobile phases differ between modes. In LC, unlike in GC, the mobile phase, as well as the stationary phase, participates in the process of defining the retention characteristics of the system. Thus, the combination of two LC columns of differing separating characteristics is likely to require the use of two different mobile phases. Consequently, a truly orthogonal LC/LC system is often limited by the necessity to have compatible mobile phases. For example, NP followed by R P might present difficulties with immiscibility of the NP fraction injected on the R P column. Although that sequence has already been demonstrated by using a microbore column as the first dimension.'x Some LC combinations that are easier to combine are: IEC and RPLC S E C and RPLC SEC and NPLC Combinations of RPLC with electrophoretic methods It is not always the case, but in the interest of meeting the second requirement for a successful two-dimensional separation, that is, no recombination in the second dimension, the first dimension is generally the lower resolution of the two. Wehr'" has described a number of LC/LC systems used in proteomic studies, including several directly coupled columns and several using valve switching between two or three columns. In one combination of SEC and RPLC, it is noted that the selectivities of the two columns are similar enough to produce a separation that is not truly orthogonal and thus less than desired. Overall, this study provides a good introduction to the possibilities of LC/LC separations. Systems that cannot be easily combined include RPLC and NPLC, which would be the G C equivalent of a polar column/nonpolar column combinatior;. However, at least one group of investigators3" has compromised the definition of N P and has used a silica column (as is conventional) with an

432

SPECIAL APPLICATIONS

aqueous mobile phase (not conventional) for their NP first dimension along with an R P for the second dimension. They were very pleased with the two-dimensional separation they achieved. Evidence that such two-dimensional systems can be successful arises from subsequent publications as noted above.*# Another example using two dissimilar R P columns and selective tuning, resulted in a two-dimensional separation that the authors feel is truly orthogonaL3' The stationary phases were either ODS or amino or cyano bonded phases; the parameters adjusted for optimization included temperature, MP strength, and buffer strength. Other groups have done similar work and have included the use of monolithic silica columns.32 Another difference between GC/GC and LC/LC is the inapplicability of cryogenic modulation to LC/LC. The most common alternative is the use of conventional multiport valves. Using the same concept described for GC/GC, namely a very short second-dimension column and total retention time, and an eight-port valve and second pump, Jorgenson and Bushey 33 demonstrated a truly orthogonal separation in 1990. They successfully separated proteins using a cation exchange column (and gradient elution) in the first dimension and a SEC column for the second dimension. The total run time on the first column was around 300 min compared to 6 min on the second column. Commercial liquid chromatographs can be easily converted to two-dimensional operation by the addition of an eight-port valve and a second pump:" The main limitation is the necessity to select two columns that can be operated with compatible mobile phases. The addition of a mass spectrometer as the detector and appropriate software can provide excellent three-dimensional separations. Other Combinations

These G C and LC examples of multidimensional chromatography serve as a fitting introduction to many other combinations. The major ones are with supercritical fluid chromatography (SFC) and capillary electrophoresis (CE). Since SFC is intermediate in characteristics between GC and LC (see Chapter 61, and some consider it to be the ideal unified technique, it is a natural to combine with GC or LC or both. Also, the C E methods (CZE and CEC) are very similar to capillary RPLC and IEC and make ideal combinations with HPLC. Two-dimensional CEj4 is also a technique likely of interest to chromatographers. The book edited by Mondello et aL3' contains additional, individual chapters on SFC,j"-37 multidimensional planar chromatography," and multidimensional electrodriven separations,3y as well as six chapters on applications. Other applications can be found in the 22 papers from the symposium mentioned earlier.25

15.2

BIOLOGICAL APPLICATIONS

433

15.2 BIOLOGICAL APPLICATIONS

Separations and quantitation of biological molecules present a number of challenges to the analytical chemist. First is the complexity of the sample matrix. Blood, urine, tissue, and other biological matrices may contain a plethora of other components, including salts, proteins, nucleic acids, and carbohydrates. Some components are in solution, but others may be present as solids or gels. Working out an appropriate sample preparation is often a major challenge in the analysis of biomolecules. A second challenge is the wide range of physical properties. Some of the smaller biomolecules (MW I 500) are not much different from other small molecules, and chromatographing them follows the principles given thus far in this text. Typical of this group would be the less polar analytes such as vitamins and lipids, many of which can be readily analyzed by GC or RP or normal-phase HPLC. Other small biomolecules are quite polar, however, and separations of these compounds, commonly done by reversed-phase HPLC, require the use of stationary phases optimized for use in highly aqueous mobile phases. Some analytes are charged; others, such as amino acids, are zwitterionic. In the analysis of acidic compounds, trifluoroacetic or phosphoric acids are commonly added to suppress deprotonation. Separation of charged species requires careful control of mobile-phase pH to ensure that the analyte is in an appropriate ionization state. These charged compounds are typically separated by IEX or by RP with the use of ion-pairing reagents. An additional challenge in the HPLC analysis of small biomolecules is that many analytes have no chromophore, thus rendering UV detection difficult. Especially in the field of amino acid analysis, numerous precolumn and postcolumn derivatization chemistrics have been devised to increase sensitivitY.4".-I' Larger biomolecules include many biopolymers. Although chromatographic methods have been used to separate and measure hydrophobic proteins such as membrane proteins, most work has been done on soluble proteins and nucleic acids. Since these materials are present in aqueous biological matrices, the methods most often chosen to separate and quantitate them are those using aqueous mobile phases, such as RPLC and IEC. Recent developments in mass spectrometry (MS) have also contributed to the success of analyzing biomolecules by using the combined technique of LC/MS as described in Chapter 10. Table 15.1 lists all of the applicable methods and their major applications and indicates where they can be found in this book. Most have already been presented; the exception is affinity chromatography, which was not included in Chapter 8 but is the subject of the next section. Next, an introduction to bioanalytical chromatography will be given, organized by application rather than by the techniques listed in Table 15.1.

434

SPECIAL APPLICATIONS

Table 15.1

Chromatographic and Related Methods for Analyzing Biomolecules

Method

Introduced in

Used For

Thin-layer chromatography, TLC 2D PAGE HPLC, mainly R P Ion exchange chromatography, IEC Ion interaction chromatography, IIC Hydrophobic interaction chromatography, HIC Internal surface reversed phase, ISRP Size exclusion chromatography, SEC Affinity chromatography

Chapter 11 Chapter 13 Chapter 8 Chapter 8 Chapter 8 Chapter 8

Wide range of small molecules Proteomics Wide range Ions and ionizable molecules Ions and molecules Biopolymers

Chapter 8

Small molecules and biopolymers Biopolymers Chapter 8 This chapter Specific, strong binding (lock & key) Capillary electrochromatography, CEC Chapter 13 Wide range Capillary zone electrophoresis, C Z E Chapter 13 Wide range Micellular electrokinetic Chapter 13 Uncharged molecules chromatography, MEKC LC/MS Chapter 10 Wide range Multidimensional methods This chapter Wide range; for increased resolution

The only applications covered here are proteins, peptides, and proteomics. The b o o k edited by Katz4’ elaborates on this material, beginning with the introduction by Simpson.43 Another useful book, organized according to technique, is the one by Cunico et a1.43

Affinity Chromatography

Affinity chromatography (AC) is characterized by its use of uniquely selective stationary phases, which are comprised of ligands bonded onto a solid support. It is also known by the names biospecific interaction chromatography (BIC) and bioselective adsorption chromatography (BAC). The bonded phases in AC are somewhat similar to the other bonded phases discussed in Chapter 8, but their specificity and strength of interaction with the analyte set them apart. Affinity chromatography is used primarily with biomolecules, for two reasons. First, many biomolecules form strong and stable complexes with ligands such as receptors, substrates, or antibodies. Second, since analysis of biomolecules is often done in highly complex media, the specificity of the interaction of analyte with stationary phase allows the exclusion of nonbinding species, thereby greatly simplifying the analysis.

15.2

Table 5.2

BIOLOGICAL APPLICATIONS

435

Some Group-Specific Ligands and Their Specificitiesa

Ligand

Specificity N u t u r d Ligurit1.s

Antigen Antibody Concanavalin A, lentil lectin, whcat germ lectin Soybean trypsin inhibitor, methyl esters of various amino acids, D-amino acids Glutathione Protein A and protein G DNA, RNA, nucleosides, nucleotides

Antibody Antigen Polysaccharides, glycoprotcins, glycolipids Various proteases Glutathione-binding proteins Many immunoglobulin classes and subclasses Nuclcases, polymerases, nucleic acids

Synthetic Ligancls Phenylboronic acid Cibacron Blue F3G-A dye, derivatives of AMP, NADH, and NADPH Amino acids: lysine, arginine Ni(I1) (IMAC) Fe(III), Ga(II1) (IMAC) Molecular imprinted polymers (MIP)

Glycosylated hemoglobins, sugars, nucleic acids, and other cis-diol containing substances Certain dehydrogenases via binding at the nucleotide binding site Proteases Histidine containing proteins/pcptides Phosphorylated peptides Nonbiological analytes

“Ahhreviations not defined in text o r table: RNA, ribonucleic acid; AMP, adenosine monophosphate; NADH, reduced nicotinamide adenine dinucleotide; NADPH, reduced nicotinamide adenine dinucleotide phosphate: IMAC, immobilized metal ion affinity chromatography.

Ligands can be classified into two types, monospecific and group-specific. Monospecific ligands are those binding only one species; an example would be a monoclonal antibody to a protein such as human serum albumin. Such a monospecific ligand would be expected to bind the analyte with a large association constant while binding weakly, or not at all, to other analytes. Group-specific ligands are those binding groups of analytes; an example would be a lectin, which would bind any protein containing the appropriate carbohydrate structure. Some typical group-specific ligands and their specificities are listed in Table 15.2. Very strong complexes can be formed in AC separations to the extent that binding may appear to be an “all-or-nothing” phenomenon rather than an equilibrium between bound and unbound species. Many AC separations are therefore run in a batch elution or step-gradient mode; the steps are shown in Figure 1S.6.45 Three different types of molecules are shown as three

436

SPECIAL APPLICATIONS

Chromatography Steps

1. Sample Loading

2. Complementary Binding Begins

3. Column Wash to Remove Contaminants

4. Elution 5. Regeneration

Figure 15.6. Reprcscntation of thc process of affinity chromatography. C'ourtcsy of Phenomcncx, Inc. Copyright Phenomenex Inc.

different shapes; only the circles fit into the ligand and the others elute from the column. A change in effluent composition causes the release of the bound compounds, and they are washed off the column as a band, separated from the others. Although this process is somewhat analogous to a stepgradient LC method, these highly 5pecific interactions d o not operate under the normal chromatographic equilibrium processes, and the chromatographic theory presented earlier does not apply. Some might even say that this is not a chromatographic process because of the way it is carried out. A special category of A C uses immobilized metal ions in the stationary phase and is called immobilized metal affinity chromatography, or IMAC. The use of Ni" or Cu2+ complexed by nitriloacetic acid, for example, allows selective retention of proteins rich in histidine residues. Since these metal ligand-analyte interactions are not as strong as those in typical affinity chromatography, some chromatographic-type separations result. An exam-

15.2 BIOLOGICAL APPLICATIONS

437

ple'" from Regnier's lab reports the separation of tryptic digests. Using IMAC, a typical ovalbumin digest was separated into five peaks, each composed of a group of similar analytes. Similarly, Posewitz and Temptst4' used immobilized gallium(II1) to separate phosphopeptides from nonphosphorylated species. These highly specific columns can be as short as 1 cm and separation times as short as 1 min. The advantage to operating in column mode is that in-line detection can be used to quantitate the eluted analyte. If AC is used for sample preparation purposes, it is not necessary to carry out the separation in columns at all. Sorption onto the substrate (stationary phase) can be accomplished in a beaker, and then, after decanting the solvent, the analyte can be released by adding a new solvent. Alternatively, the AC supports can be used as the packings in solid-phase extraction (SPE). Affinity chromatography is used for quantitative and preparative purposes. Four reviews"-5' and at least six books"-" are available for further information. Volume 768 of the Journal of Chromatography B contains 20 articles on affinity-type separations and studies in chromatography and electrophoresis." Volume 49 of the Journal of Biochemical and Biophysical Methods contains 43 articles on various aspects of affinity chromatography, including theoretical, preparative, and quantitative aspects.'" A review by Lee and Lee"" covers applications of AC in proteomics and provides a list of commonly used ligands and analytes. Preparative AC applications have not been too popular because of ligand instability and the cost of the sorbents, but several review articles have presented the applications to date."'. '" New ligands are being proposed, and the needs of cornbinatorial chemistry have increased the interest in t h e technique for analytical and preparative applications. Among the synthetic ligands listed in Table 15.2 is a class called molecular imprint polymers (MIP), which were introduced in Chapter 14. Unique three-dimensional binding sites can be prepared that will be effective not only for biological molecules such as proteins but also more generally for other molecules, thereby extending the applications of AC beyond biomolecules. The process involves the synthesis of a polymer around a template molecule that will be the analyte. After the polymer is formed, the template is removed, leaving its shape to serve as the ligand on the stationary phase (SP). The historical development of MIPs and their applications as chromatographic SPs has been reviewed in Analytical Chemistry"' " and a special issue of the Journal of Chromatography B6' reports on the use of MIPs in separation science. Some of the most useful applications have been for enantiomeric separations.'' A short summary of all of these affinity chromatographic variations, including a list of companies offering commercial affinity resins, has been p~blished.~'

438

SPECIAL APPLICATIONS

Proteins, Peptides, and Proteomics

Probably the most rapidly developing area of bioanalytical chemistry involves the analysis of mixtures of proteins and peptides, many of which arise from the field of proteomics and the major effort to understand the human genome. Further consideration of proteomics is not included here, but a chapter on recent trends has been published.6x The necessity to separate complex mixtures of very large biomolecules has resulted in major advances in the use of HPLC, coupled in many cases with MS, replacing the twodimensional electrophoretic methods formerly used. Some of the basic differences between small-molecule HPLC/MS and biomolecular HPLC/MS will be discussed, followed by some recent examples of the successes of the newer techniques. Large biomolecules differ from smaller analytes in more than size. Most proteins and nucleic acids assume spatial conformations based on their secondary and tertiary structures, such as intramolecular bonding, coiling, and folding. A macromolecule in its typical conformation is said to be “native,” while perturbation of this conformation is labeled “denaturation.” While smaller analytes may assume different conformations, interconversion between conformational states is generally fast when compared to the chromatographic timescale, and separations are done on the “average” conformation. In the case of macromolecules, however, just the right conditions are often required to refold a denatured species, and denaturation can thus be nearly irreversible. Denaturation can therefore generate multiple conformations of the same molecule, with each form having different retention characteristics. Some chromatographic conditions, like high temperature ( 2 50°C), extreme pH, some solvents, and even the chromatographic process of sorption and desorption can cause denaturation. Therefore LC conditions must be well-defined and reproducible to ensure consistent separations. Originally it was thought that under typical RPLC conditions all proteins were denatured,”9 but it now appears that some peptides and proteins are more stable than others. Those analytes that exist in multiple conformational forms often elute as several peaks, some with odd shapes. Under RPLC and HIC conditions, the denatured protein generally elutes later than the native protein, but in IEC and AC the opposite is true.”9 These differences are attributable, in general, to the exposure of new amino acid residues on denaturation. In aqueous solution, proteins tend to fold with the polar amino acids exposed to the solvent and nonpolar residues in the interior of the protein. On denaturation, the nonpolar residues become exposed and can interact with the stationary phase, leading to longer retention. In the case of IEC, denaturation may lead to a very different distribution of charges than that of the native molecule.

15.2 BIOLOGICAL APPLICATIONS

439

Figure 15.7. Approximate correlation between the molecular weight (of protcins and peptides) and the stationary-phase pore diameters. Courtesy of Phenomenex, Inc. Copyright Phenomenex Inc.

The SP used to separate polymers must have sufficiently large pores to admit the analytes to the interior of the resins. Figure 15.7 shows the correlation between protein/peptide molecular weight and the pore size needed for total inclusion. Consequently, typical commercial SPs for biomolecules have pore sizes of at least 30 nm (300 A). The chemical nature of the SP (usually either silica or organic polymer) can also exhibit heterogeneous surface sites that interact with proteins to cause nonideal, tailing peak shapes in RPLC. Lowering the pH of the mobile phase (MP) to 2 (the lower limit for many silicas) will often improve peak shapes due to suppression of polar SP sites. One of the most commonly used acids is trifluoroacetic acid (TFA). This acid serves not only as a pH adjustor, but since it is partially retained by many SPs it also acts as an ion-pairing agent. Careful adjustment of TFA concentration can be used to optimize retention factors and resolution of peptides. Consequently, it is important to specify the amount of TFA added (typically 0.1-0.3% w/v) and clearly state the ratio basis (w/v or v/v).'" A typical RPLC separation" of proteins on a C,, column is shown in Figure 15.8. Note that the pore size is 30 nm and that it is a gradient run from 0 to 40% acetonitrile (ACN). Gradients are almost always used in RP

440

SPECIAL APPLICATIONS

2

0.2 -

I ‘

238TP3405 0.1

50 mm length 4.0 mL/min

-

3

4

k 0.0

1

I

I

I

I

I

Figure 15.8. Typical protein separation on 3-pm revcrsed-phase column. Peak identities: (1) ribonuclease, (2) insulin, ( 3 ) cytochrome C, (4) BSA, and (5) myoglobin.oConditions: Column: Vydac 238TP34-SO “monomeric” C-I8 phase, 3-pm particle size, 300-A pore size, 4.6 mm i.d. x SO mm; flow: 4.0 mL/min; UV detection at 215 nm. Gradient elution from 20% ACN (with 0.l%>TFA) to 456 ACN (with 0.1%’ TFA) in 4 min. Reprinted with permission from Vydcc Application Note 9902, Grace Vydac, Columbia, MD.

protein separations since isocratic conditions often fail to elute the analytes. There is a trend to use capillary columns and very low flow rates to facilitate the use of MS detectors; in these separations formic acid is preferred to TFA since TFA tends to quench electrospray ionization.” A further trend to minimize analysis times results from the need to analyze more complex mixtures like those encountered in proteomics. Wehr has published a series of three studies describing the developments in proteomics from the point of view of a chromatographer.*‘, ’I He describes the approaches that have been taken to make it possible to separate the very complex mixtures that are encountered, highlighting multidimensional methods (including multiple columns and MS detection) and affinity chromatography, both of which have been discussed in this chapter. Two reviews75.7h have also appeared, providing additional details. Twenty-three papers from the 2002 symposium on Separation of Proteins, Peptides, and Polynucleotides have been published as volume 1009 of the Journal of Chromatography A,’’ Figure 15.9 is typical of the RPLC/MS data obtained for a protein extract7’; the original is in color, which is preferable, but even the black-and-white reproduction conveys the complexity of the data.

’‘

Other Large Biomolecules

Until recently, separation of mixtures of large biomolecules has been accomplished by gel electrophoresis, but advances in HPLC have led to the prediction that chromatographic methods may become dominant.” The classes of biomolecules most likely to be run by chromatography are carbohydrates and nucleic acids. HPLC is increasingly being used in cancer research

15.2

BIOLOGICAL APPLICATIONS

441

Figure 15.9. ( a ) Three-dimcnsional plot of LC/MS chromatogram of a protein extract from a breast cancer tissue section digested with trypsin. ( h )The corresponding planar ion density map. Reprinted with permission from reference 78. Copyright 2003 American Chemical Society.

442

SPECIAL APPLICATIONS

for DNA adducts and DNA oxidative modification.8” The approach for these types of applications is similar to that just described for proteins and peptides. Compared to gel electrophoresis, HPLC offers the advantages of simplicity, the flexibility provided by gradient elution, the multidimensionality of MS detection, and ease of quantitation. The two books already ment i ~ n e d ~ ~provide .” good summaries, but development in these fields is very rapid at present. Electrophoresis is still an important method, but recent developments in biopolymer analysis have been in capillary electrophoresis (CE), including capillary electrochromatography (CEC). Some of it has been summarized in the biennial reviews in Analytical Chemistry.8’.

*’

Hydrophobic Interaction Chromatography

Hydrophobic interaction chromatography, or HIC, another chromatographic technique that was introduced in Chapter 8, is targeted for separation of biomolecules. HIC is generally run under conditions in which denaturation of the analytes is prevented or minimized, and therefore it is used extensively in preparative mode for the isolation of biologically active compounds. Largescale purifications of proteins typically employ low-pressure HIC. Analytical separations are most commonly done on small-particle columns. Separations in HIC are unique in their mode of separation. As the name implies, the technique differentiates analytes on the basis of their hydrophobicities. Stationary phases are most often based on silica, polymer, or cross-linked agarose beads bonded with hydrophobic ligands. Ligand hydrophobicity increases with alkyl chain length, so a butyl phase is less hydrophobic than an octyl phase, for example. Other alkyl and phenyl phases are used, as are more polar materials such as polyethers based on polyethylene glycol and polypropylene The HIC mobile phases are almost always entirely aqueous salt solutions. Sample injection and loading is done at very high mobile-phase salt concentrations, such as 2 M sodium sulfate. Elution is accomplished by a gradient of decreasing salt concentration. High concentrations of salt in the mobile phase enhance the binding of analytes to the stationary phase in a process called “salting out” or “salt-promoted a d ~ o r p t i o n . ”As ~ ~ the mobile-phase salt level decreases during the gradient, the analytes are desorbed, with the most weakly bound compounds eluting first. Careful selection of mobile and stationary phases is critical to success in HIC. Use of an SP that is too hydrophobic can lead to denaturation because of the harsh elution conditions required to elute the analyte. Choice of salt is also important. In general, monovalent cations have a greater salting-out

15.3 CHIRAL SEPARATIONS

443

effect than divalent cations; likewise, multivalent anions appear to contribute more to hydrophobic interactions than monovalent anions.", "' Ammonium sulfate is the salt most commonly used in HIC. Reviews are available covering the application of HIC to proteinss3.H7. " and carbohydrates.'" Additional information can be obtained from several vendor websites."", 'I' 15.3 CHIRAL SEPARATIONS

It is becoming increasingly important to distinguish between enantiomers for optically active compound^.^' For many drugs, only one optical isomer is pharmacologically active, and a total analysis that does not separate and quantitate the enantiomers is unsatisfactory. Enantiomers have identical physical properties and can rarely be separated by conventional chromatographic systems. Thus new approaches to the separation of chiral compounds have been developed. This separation challenge has been addressed by various methods, the oldest being the formation of diastereomers by using a chiral reagent that creates two chiral centers in the products. Since diastereomers have different physical properties, they can be separated with conventional nonchiral chromatographic systems. This approach was the major one used in GC when the field began to develop over 30 years ago. However, it has some disadvantages, and since not all analytes are amenable to analysis by GC, other approaches to chiral separations have been developed. These can be divided into two groups: those using chiral stationary phases (CSPs) and those using chiral mobile phases (CMPs). These three methods, diastereomer formation, CSPs and CMPs, will be discussed separately after a brief review of stereochemical nomenclature and conventions. Stereochemical Nomenclature and Conventions

Some chemical compounds, for example, those having four different groups attached to a central tetrahedral carbon, can exist in two mirror-image isomeric forms, one of which rotates plane-polarized light to the right (dextrorotatory) and the other rotates it to the left (levorotatory). One is designated by "d" and the other by "I" or by the signs ( + ) and ( - 1. They are said to have a chiral center or a chiral carbon. Such isomers, which are neither superimposable nor interconvertable are called enantiomers of one another. A 50 : 50 mixture of enantiomers is said to be rucemic. Compositions of enantiomeric mixtures other than 50: 50 are designated either by the ratio of enantiomers, (e.g., 97: 3) or the enuntiomeric purity (e.g., 97%). Some

444

SPECIAL APPLICATIONS

RS Enantiomers f

SR'

SS' researchers use the units of enuntiomerzc excess, which is the fraction of one enantiomer minus the fraction of the other (e.g., 10.97 - 0.031 = 0.94). Rather than designating enantiomers according to the direction of rotation of polarized light, which requires an experiment to be performed, chemists have devised another system based on several rules that can be applied to the arrangement of the atoms in spacc.'" The symbols R and S are used to denote the two forms. This convention is now the one most commonly used and will be followed in this section. Molecules with more than one chiral center can exist as diastereoisomer-s, or diustereomers. For a chiral molecule with two chiral centers, one center will be either R or S; for clarity, we designate the orientation of the second center as R' or S'. Four different structures can bc idcntificd, as shown in Figure 15.10. These molecules are not mirror images, and one pair of diastereomers usually differs in physical and/or chemical properties from the other pair. For example, if a chromatographic separation of these four isomers is attempted with an achiral system, two peaks can usually be obtained, one including the diastereometric forms RS' and SR' and the other the forms RR' and SS'. If this mixture is run on a chiral chromatographic system, all four compounds may be resolved. An example of such a four-component separation (by CE) was included in Chapter 13. In the special case where the diasterioisomers are more symmetrical, for example, when one half of the molecule is the mirror image of the other half, the two diasteriomeric forms are superimposable and are not optically active even though they have two chiral centers. These are called meso structures. The reason they are not optically active is that the rotation caused by the first chiral center is exactly counteracted by the equal and opposite rotation from the second center. These principles can be used to guide the chromatographer in performing chiral separations. First diasteriomeric separations will be discussed, and then CSPs and CMPs.

15.3

CHIRAL SEPARATIONS

445

Diastereomer Separations

As just described, a racemic mixture of optical isomers (designated R and S> cannot be separated chromatographically unless a chiral element is introduced into the procedure. One way is to react the mixture with a chiral reagent (designated, e.g., R'), to form reaction products that will have two chiral centers: One isomer will generate an RR' reaction product and the other the SR' combination. These two diastereomers will have different physical properties and can be separated chromatographically with nonchiral phases.

o=s=o

1

2

1 RS 2 ss

t

0

1

2

3

Time (min)

4

5

Figure 15.1 1. Separation of amphetamine-NPSP diastereomers on Supelcosil LC-Si in chloroform: heptane, 8:2. Reprinted with permission from C. R. Clark and J. M. Barksdale, A n d . Chem. 1983, 56, 958. Copyright 1983, American Chemical Society.

446

SPECIAL APPLICATIONS

Derivatization to form diastereomers requires a chiral reagent that iq 100% of one isomer. As an example, Clark and Barksdale"' separated the R and S enantiomers of amphetamine after reacting the racemic mixture with 1[(4-nitro-phenyl)sulfonyl]prolyl chloride, which was 100% of the S isomer. The RS and SS diastereometric amides were separated on a silica gel column by HPLC by using a chloroform/heptane (8/2) mobile phase. The chromatogram is shown in Figure 15.11 along with the structure of the derivative. In another example, the enantiomers of 21 amino acids were separated after they were derivatized with a chiral reagent to form diastereomers." The review by Lochmuller and Souter'" summarizes early GC work on enantiomer separations, including a comprehensive discussion of the use of liquid crystals as stationary phases. For GC two useful chiral reagents are N-trifluoroacetyl-L-prolyl chloride (TCP) and menthyl chloroformate (MCF). TCP is used for the resolution of amines and MCF for alcohols. The necessity of using an optically pure reagent is one disadvantage of this method of separation. Ideally the reagents should be 99% optically pure, but successful separations have been done with reagent purities as low as 90%. Also, the reagent must react equally with the two enantiomers in the sample. The derivatization method can be problematic in preparative work because after separation the individual diastereomers must be reacted to remove the unwanted derivatizing reagent, and then the enantiomers must be purified. For these reasons, this approach has not been the most successful or popular. It is easier to use a chiral chromatographic system, either a CSP or a CMP. The former is most often used. Chiral Stationary Phases

The use of chiral stationary phases (CSPs) is the most popular method for achieving chiral separations because chiral groups can be bonded to the surface of a stationary support using the technologies described earlier. For GC, CSPs are the only option since it has not been found possible to operate GC with chiral mobile phases. After disappointing results in separating diasterioisomers by GC, some chiral CSPs were produced about the same time as open tubular columns gained in popularity. The combination of more efficient columns and stable CSPs led to more successful chiral separations. One of the first successful phases produced was formed by the reaction of L-valine-t-butylamide with the CN group on a polysiloxane such as OV-225, making a chiral polysiloxane. Later cyclodextrin-type phases became available. A concise summary o f chiral GC separations can be found in Chapter 5 of the book by Beesley and Scott," the booklets of several laboratory supply houses,"8.'" and the comprehensive 2001 review by Schurig.'"" In his 2002 review of chiral separations,

15.3

447

CHIRAL SEPARATIONS

Column:

LIPODEX E-25 rn x 0.25rnrn ID

Conditions:

Injection volume: 1UI Carrier gas: 60 kPa Hydrogen Split: 1 : 100 Temperature: 90 C (0 min iso);4C / rnin to 190 C

Detection:

FID

Chromatogram: 2

I

T-

L

I

1

10

I

r

20 min

Figure 15.12. GC separation of the cnantiomers of some amino acid methyl esters (TFA) on Lipodcx E. Reprinted with permission from MACHEREY-NAGEL, Dueren, Germany. Identification o f peaks: (1) alaninc, (2) valine, ( 3 ) leucinc, (4) proline, ( 5 ) aspartic acid, (6) phenylalaninc. In all cases cxccpt proline, D elutes bcforc L

Ward“” indicated a recent increase in chiral GC separations. Typical of the excellent separations possible on fused-silica capillaries is the amino acid example“’* in Figure 15.12. Considerable development work on CSPs has been done for HPLC phases. Pirkle“” has discussed the early work on CSPs for HPLC, giving examples and describing how they work. An early review of chiral stationary phases“’4 divided CSPs into five groups: 1. 2. 3. 4. 5.

“Brush”-type bonded phases of the type pioneered by Pirkle Phases containing “cavities,” of which the cyclodextrins are typical Phases containing helical polymers such as cellulose Phases for ligand exchange Phases for affinity chromatography

448

SPECIAL APPLICATIONS

To these could be added phases of more recent vintage: 6. Proteins such as the albumins and ovomucoids 7. Macrocyclic antibiotics such as rifamycin, vancomycin, and teicoplanin

New CSPs continue to be reported as investigators attempt to find better and better chiral selectors.'05 Other classification systems have been suggested so that newer ones can be fit into an overall scheme. Table 1.5.3 contains an exhaustive listing with only three groups: natural, semisynthetic, and synthetic.'"' Brush-type phases have been developed primarily by Pirkle et al."" These phases are based on three sites of interaction between one enantiomer and the SP. Phases can be .rr-donors, .rr-acceptors, or a combination. One of the most successful phases was formed by reacting a silylpropylamine bonded phase with a chiral N-(3,.5-dinitrobenzoyl)aminoacid to form chiral phases of the general formula shown in Figure 15.13. The linkage (Y) between the reacting groups can be ionic or covalent, as shown; both types are now commercially available and are known as Pirkle columns. The ionic CSPs can be used only with relatively nonpolar mobile phases, so the covalent CSPs were designed for use with polar mobile phases. The amino acid moieties used in the commercial products include D-phenylglycine and L-leucine. Pirkle CSPs have been used for HPLC separations of the enantiomers of alcohols, sulfoxides, bi-P-naphthols, P-hydroxysulfides, heterocycles such as hydantoins, succinimides, and agents related to propranolol, among others. A second type of chiral separation is based on size exclusion as well as chiral interactions. For example, inclusion complexes are formed between cyclodextrin and molecules with the correct size and shape to fit inside it. Beta-cyclodextrin, for example, has a molecular weight of about 1000 and with 35 chiral centers; the model shown in Figure 1.5.14 has been proposed to show its shape and interaction with analytes. It appears that the analyte molecule must also have a functional group in the right position to hydrogen bond with a secondary hydroxyl group on the edge of the cage. Computergenerated images of some inclusion compounds illustrate these principles.")x Some of the enantiometric compounds separated by HPLC on a 10-cm P-cyclodextrin column are listed in Table 15.4.'"' Phases based on helical polymers such as cellulose and its derivatives, including carbamates and esters, provide enantiomeric separations of a wide variety of compounds. The mode of separation is a combination of inclusion complex, polar, and r-7~ interactions. Most of these columns are operated in the NP mode, but newer phases compatible with R P conditions have become available.

15.3

CHIRAL SEPARATIONS

449

Table 15.3 Main Groups of Chiral Selectors

Source Nat ural

Type Proteins

Chiral Selector Serum alhumin Orosomucoid ( a I-acid glycoprotein) Ovomucoid Cellobiohydrolasc I Avidin A Chymotrypsin A Ovotransferrin A a-, p-, and y-Cyclodcxtrins

Techniques

LC,CE. CEC. Membranes. ex tract ion LC.CE. CEC, CE GC. TLC.

cryst

Polysaccharidcs

Antibiotics

Low M , molecules

Semisynthetic

Modified oligosaccharides Modified polysaccharidcs

Ccllulosc

Starch Dcxtran Vancomycin Teicoplanin Ristocetin Avoparcin Amino acids Cholic acids/bilc salts Alkaloids Tartaric acids Derivatiscd cyclodcxtrins

LC.CE. GC S FC LC. CE, SFC cryst.

LC,CE,

CEC'. GC

Polysaccharidc carbarnates Polysaccharide esters

TLC

LC. CE.

SFC,TLC.

membr,ine\,

extrxtion

Synthetic

Modified low M , molecules

Ion-cxchangc selectors

Synthetic low IW? molecules

Pirkle-type sclcctors

Receptor molecules

Helical synthetic Polymers

Polyacrylamides Polyacrylatcs Cross-linked tartaramides

MIPS

LC. CE. CEC,SFC,

extraction LC, CE, GC. TLC, SFC. menihrancs

LC,CE, extraction, cry5t LC,CE, SFC LC LC.CE, SFC,

membranes Source: After Maier et al."'" "A = analytic scale; P = preparative scale

Scale"

450

SPECIAL APPLICATIONS

NO* R = phenyl, i-butyl, p-naphthyl, or i-propyl

H Y = NI or -O-'NH,/ \

Figure 15.13. Structure of "Pirkle" CSP.

Figure 15.14. Geometry of P-cyclodextrin. Reprinted with permission from A m . Lab. 1985, 17(5), 80. Copyright 1985 by International Scientific Communications.

Ligand exchange phases are used primarily for amino acid separations. In this technique, a chiral copper complex is immobilized on the SP. Analyte enantiomers form diastereomeric complexes with the chiral complex, forming the basis for the separation. The use of this technique has been reviewed by Davankov."" Proteins covalently bound to silica supports provide chiral separations based on a combination of polar and hydrophobic interactions. For example a,-glycoprotein (AGP) separates enantiomers of a broad range of molecular types."' Human serum albumin, bovine serum albumin, and ovomucoid columns have somewhat more limited applications. Selectivity with protein

15.3 CHIRAL SEPARATIONS

451

Table 15.4 Separation Data for Enantiomeric Compounds Using a 10-cm P-Cyclodextrin Bonded-Phase Column

k Enantiomeric Mixture D, L-Alanine P-naphthylamide D, L-Methionine -naphthylamide D, L-Alanine P-naphthyl ester Dansyl-D, L-phenylalanine Dansyl-D, L-leucine N-Benzoyl-D, L-arginine-Pnaphthylamide ( + )/( ) Mephobarbital" -

(

+ )/( - )2,3-O-isopropylidene-2,3-

ow(+) Isomer

Lor(-) Isomer

a

R,

Mobile" Phase

6.1 3.6 1.8 3.8 4.2 30.4

5.1 2.7 1 .0 3.1 3.0 28.5

1.20 1.33 1.80 1.23 1.40 1.07

2.0 2.4 2.6 1.1 2.4 0.6

50:50 50:50 50:50 55:45 50:50 50:50

16.9 11.84

14.8 10.56

1.14 1.12

1.6 1.2

20:80 48:52

dihydroxy-l,4bis(dipheny1phosphino)butane

Source: From Armatrong and De Mond."'"

"Numbers represent the volume percentage of methanol to water. "A 25-em P-cyelodextrin column was used to generate these data.

phases can be modified by pH and ionic strength. Since the business ends of these phases are proteins, however, stationary phases are subject to irreversible changes through denaturation. Too high a temperature or the use of charged mobile-phase modifiers may permanently change the selectivity of the column. Although vancomycin and teicoplanin are the most commonly used macrocyclic antibiotics for chiral LC separations as many as seven macrocyclic antibiotics have been evaluated for the enantiometric separation of dansyl amino acids.'". The mechanisms of separation are complex and are thought to include hydrogen bonding, inclusion complexes, ionic interactions, and rr-rr interactions. The macrocylic columns are versatile since they can be run in both RP and NP modes. Choosing a CSP Exactly how CSPs work is not known, and the choice of a CSP for a particular separation has been mostly by trial and error. However, some of the principles of chiral recognition have been identified for the Pirkle columns. Pirkle'('3 has observed that three simultaneous interactions between the chiral stationary phase and the analytes must occur for chiral recognition and separation. Typical interactions are those discussed in Chapter 4 and include hydrogen bonding, dipole-dipole interactions, dipole-induced dipole interactions, and rr-rr associations. The geometric arrangement of the chiral

452

SPECIAL APPLICATIONS

phase and the analyte must also be suitable, thus imposing a steric requirement. It has also been shown that the spacing between the silica surface and the chiral moiety can be important.'"3,' I 3 For example, a change in the number of carbons in the chain attached to the chiral moiety (shown in Figure 15.13 as three carbons) can reverse the order of elution of two enantiomers. Furthermore, chiral recognition is reciprocal. For example, the dinitrobenzoyl amino acid CSP can separate N-acetylated a-arylalkylamines. Therefore, a CSP made with an a-arylalkylamine functionality should be able to separate analytes containing the dinitrobenzoyl functionality. Such has been shown to be the case, and dinitrobenzoyl derivatives of amines, a-amino esters, and amino alcohols have been separated."' Choosing a CSP can be difficult; there are hundreds of columns and many manufacturers to choose from, and it is likely that more than one CSP will work for a given separation. Finding a similar separation in the literature is always one possibility, and a comparison of literature findings with one's application could provide a solution. Beesley and Scott have provided a discussion of applications organized according to type of CSP in Chapter 1 1 of their book."' Armstrong and Zhang'" havc published a list of 14 manufacturers and classified their products according to type. They noted that the macrocyclic class, which includes the cyclodextrins, glycopeptides, and crown ethers, has had the biggest impact on enantiometric separations. Choosing between RP and N P is also a decision that must be madc. Pirkle-type phases are usually operatcd as NPLC and cyclodextrins as RPLC. Solubility may be a deciding factor in this decision, but many CSPs are recommended for use in only one modc. In fact, a CSP might operate very differently in different modes. Armstrong and Zhang'" noted that Cyclobond-I-SN forms inclusion complexes in RP, .rr-complexes in NP, and only surfacc adsorption interactions in polar organic mode. Increasingly, however, manufacturers"" have attempted to produce CSPs that can be used in either NP or RP. Application literature is available from many of them."'-"" Problems can arise when a chiral separation is adapted to HPLC/MS. Many chiral methods require the use of normal phase (NP), but conventional NP mobile phases such as hexane do not support the formation of ions critical for electrospray (ESI), and some of the buffers such as phosphate are incompatible. This topic has been addressed by Desai and Armstrong"' who have produced five recommendations that can be used to make the transformation, and they found that LC/MS can be a valuable tool for chiral drug discovery. One final issue is the order of elution of the two enantiomers. Care must be taken in assigning the optical rotation to peaks based on polarimetric detection since both the magnitude and direction of optical rotation can be

15.3 CHIRAL SEPARATIONS

453

affected by changes in the mobile phase."' Not only changes in mobile-phase composition but also changes in temperature can cause elution order reversa1.I2j Chiral Mobile Phases

Much less use has been made of chiral mobile phases (CMPs). The use of CMPs in HPLC assumes that the analytes form strong associations with a chiral component in the mobile phase and that these complexes interact differently with the achiral stationary phase. Although the mechanisms are not all clear, intermolecular associations that seem to form the basis of chiral separations are hydrogen bonding, metal chelation, ligand exchange, and ion pair formation. One example is the separation of acid enantiomers by forming ion pairs with chiral bases such as quinines.'14 Most of the ligand exchange methods are included in a review by Davankov et al."" The use of chiral mobile phases is not well suited for preparative LC. Often the chiral agents cannot be removed by volatilization for easy recovery of the analytes. Also, the chiral agents in the MP often cannot be recovered for reuse, making this method an expensive one. Additional Chiral Methods

In addition to the G C and HPLC examples given, similar principles have been applied to other chromatographic systems such as TLC and SFC as well as the various C E methods. For example, a special cyclodextrin bonded phase, with an appropriate binder, has been used in TLC plates to separate isomers, including optical isomers."' In addition, many of the references given (such as number 97) include sections or chapters on capillary electrophoretic methods. Supercritical fluid chromatography is of great utility in chiral separations. A major advantage of this technique is its relatively low pressure drop over the analytical column, which allows the use of multiple columns in series to achieve higher efficiencies. These columns do not have to contain identical packing materials; use of different column trains can provide unique selectivities. In addition, SFC is ideal for preparative chiral separations since the mobile phase is volatile, easily removed, and nontoxic. Capillary electrophoresis has the advantage that chiral separations can often be effected by adding a chiral agent to the running buffer, thus negating the need to use an expensive CSP. In many laboratories, CE finds its primary use in chiral separations. A typical separation was shown in Chapter 13, which also includes references. Wang and Khaledi12' have summarized the field in their chapter.

454

SPECIAL APPLICATIONS

Analytical Chemistry added the topic Chiral Separations to its list of biennial reviews beginning in 2000"' and continuing through 2004."" A wide variety of additional information is contained in the individual chapters of the book edited by Subramanian."'

15.4 OTHER TOPICS There are a few chromatographic methods that have not been discussed anywhere in this text, and brief descriptions of them are included here. Other topics to be mentioned are field-flow fractionation (FFF) and miniature total analysis systems ( pTAS) also called lab-on-a-chip. Other Chromatographic Methods

In the mid-twentieth century, a technique sometimes called countercurrent extraction or countercurrent distribution (CCD) was being developed along with chromatography. The two techniques were designed for similar applications and shared some of the same principles. Some mention of this activity, including the work of Craig, was included in the extraction section of Chapter 14. Several of these separation methods have survived to today and continue to be developed. The major one is called countercurrent chromatography (CCC) or high-speed countercurrent chromatography (HSCCC). It is an LLC method because the stationary liquid is unsupported on a solid support. Ito et a1.'28 is credited with the invention of the CCC instrument in 1966, and in 1984 he reviewed the early development of this technique."", "" Knight has also included a brief history in her review of the use of CCC for the purification of peptides.I3' Since then several books"'. have appeared, and improvements and additional applications have been r e ~ 0 r t e d .-I3" I~~ The main application is for preparative work. The process is continuous, centrifugally driven,"' and thus it is also called centrifugal partition chromatography (CPC)I3' or centrifugal liquid-liquid chromatography. Major advantages of CCC are the absence of a solid stationary support that could cause unwanted adsorption and the availability of a large liquid volume for the absorption (partition) process. Additional advantages include the speed of the separation and the fact that the method can be used in semipreparative mode. CCC has an additional advantage for isolations: Since the separation is done in a liquid-liquid system, there is little loss of material. In LC, compounds may bind so tightly to the stationary phase that they never elute; in GC, components may not be volatile; but in CCC, all the components are recoverable from one or the other of the phases. There are a number of other processes related to CCC in that they are continuous and used primarily for preparative work. Since they are not primarily used for analytical work, they will not be discussed further.

15.4

OTHER TOPICS

455

Field-Flow Fractionation

Field-flow fraction (FFF) is included here because its inventor, Giddings, aptly referred to it as one-phase chromatography for macromolecules and Since there is only one phase, some other field must be present to effect a separation, much like the electrophoretic process. Fields that have been applied include sedimentation, flow, electrical, and thermal. A comprehensive description has been written by Martin,'''I and a recent brief report14' on its use for separations of macromolecules and cells includes some additional references.

Microfluidic Devices for Chromatography and Electrophoresis (Lab-on-a-Chip)

For some time there has been an interest in miniaturizing chromatographic systems. Since narrow columns are inherently more efficient, and the low flow rates of micro HPLC columns make them easier to interface with mass spectrometers, microfluidics has become a new area of investigation. In addition, technological advances in preparing silicon-based microchips for the microelectronics industry offered the possibility of transferring that technology to the separation sciences. Thus were born the new technology known as micro total analysis systems, and the acronym pTAS, also known as lab-on-a-chip. The easiest separation systems to engineer with these microcharacteristics are G C and those systems whose flows are driven electrochemically: electrophoresis (especially CE), CEC, MEKC, isotachophoresis, and isoelectric focusing. The historic development is included in Part 1 of the pTAS review in Analytical Chemist~y,'~' which also reviews the theory and technology. Part 2 of the review covers standard operations (sample preparation, injection, detection, etc.) and applications.'43An earlier article in the same journal also discusses basic concepts and a useful overview."' Electrophoretic separations have received considerable attention"' and are covered in a book ~ h a p t e r ' ' ~and several reviews.'"~ Applications to LC have lagged due to the difficulties of inducing flow in nonionic media, but the 2002 review of HPLCI4" contains a few references and the major issues being addressed have been presented.'"l Typical of the current status is a study discussing sample filtration, concentration, and open-channel electrochromatography.15' A review of commercially available instruments"' includes several LCs.

456

SPECIAL APPLICATIONS

REFERENCES 1.

2. 3. 4. 5. 6. 7. 8. 9.

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

17.

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

C. F. Poole and S. K. Poole, J . Chromutogr. 1995, 703, 573-612. Sz. Nyiredy, in Multidimensionul Chromatogruphy, L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Wiley, Chichester, England, 2002, Chapter 8. J. C. Giddings, Anal. Chem. 1984, 56, 1258A-1270A. J. C. Giddings, in Multidimensional Chromatography: Techniques und Applications, J. H. Cortes (ed), Marcel Dekker, New York, 1990, pp. 1-27. D. R. Deans, Chromatographia 1968, I , 18-22. D. R. Deans, J . Chromutogr. 1984, 289, 43-51. A. C. Lewis, in Multidimensionul Chromutography, L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Wiley, Chichester, England, 2002, Chapter 3. K. MacNamara, R. Leardi, and A. Hoffmann, LC-GC No. Am. 2004, 22, 166- 186. P. Marriott, M. Dunn, R. Shellie, and P. Morrison, Anul. Chem. 2003, 75, 5532-5538. E. B. Ledford, Jr., J. R. TerMaat, and C. A. Billesbach, Introduction to GC x GC, Zoex Technical Note KT 039595-1, Zoex Corp., Lincoln, NE; available at www.zoex.com/html/technote kt 030505- 1.html. J. V. Hinshaw, LC-GC No. Am. 2004, 22, 32-40. E. B. Ledford, Jr., J. B. Phillips, J. Xu, R. B. Gaines, and J. Blomberg, Am. Lab. 1996, 28(9), 22-25. P. J. Marriott, in Multidimen.sionul Chromutogruphy, L. Mondello, A. C. Lewis, and B. D. Bartlc (eds), Wilcy, Chichester, England, 2002, Chapter 4. E. B. Lcdford, Jr. and J. B. Phillips, U. S. Patent 6,007,602, Dec. 28, 1999. U. A. Th. Brinkman and R. J. J. Vreuls, J . Chromutogr. A 2003, 1019, 1-2. P. A. Bueno, Jr., and J. V. Seely, J . Chromatogr. A 2004, 1027, 3-10 Zoex, KT 2003 G C x G C System, a Two-Stage Thermal Modulator, Zoex Corp., Lincoln, NE; E. B. Ledford, Jr., J. R. TerMaat, and C. A. Billesbach, “What Is Loop Modulation‘?” Technical Note KT 030606-1, 2003, online at www.zoex.com. LECO, Pegasus GC-TOFMS System, LECO Corp., St. Joseph, MI, www.leco.com . J. B. Phillips and J. Beens, J . ChromatohT. A 1999, 856, 331. R. C. Y. Ong and P. J. Marriott, J . Chromatogr. Sci. 2002, 40, 276-291. J. Dalluge, J. Beens, and U. A. Th. Brinkman, J . Chromatogr. A 2003, 1000, 69- 108. T. Gorecki, J. Harynuk, and 0. Panic, J . Sep. Sci. 2004. 27, 359-379. J . Chromutogr. A 2003, 1019, 1-285. J. Dalluge, L. L. P. van Stcc, X.. Xu, J. Williams, J. Beens, R. J . J. Vreuls, and U. A. Th. Brinkman, J . Chromatogr. A 2002, 974, 169-184. X. Lu, J. Cai, H. Kong, M. Wu, R. Hua, M. Zhao, J. Liu, and G. Xu, Anul. Chem. 2003, 75, 4441-4451. See also X. Lu, M. Zhao, H. Kong, J. Cai, J. Wu, M. Wu, R. Hua, J. Liu, and G. Xu, J . Sep. Sci. 2004, 27, 101-109.

REFERENCES

457

25. L. Mondello, in Multidimensional Chromatography, L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Wiley, Chichester, England, 2002, Chapter 2. 26. T. Hyotylainen and M-L. Riekkola, J . Chromatogr. A 2003, 1000, 357-384. 27. C. Corradini, in Multidimen.siona1 Chromatography, L. Mondello, A. C. Lewis, and B. D. Bartlc (cds), Wiley, Chichester, England, 2002, Chapter 5. 28. P. Dugo, 0. Favoino, R. Luppino, G. Dugo, and L. Mondello, Anal. Chem. 2004, 76, 2525-2530. 29. T. Wehr, LC-GC No. Am. 2002, 20, 954. 30. R. E. Murphy, M. R. Schure, and J. P. Foley, Anal. Chenz. 1998, 70, 4353. 31. C. J. Venkatramani and Y. Zelechonok, Anal. Chem. 2003, 75, 3484-3494. 32. N. Tanaka, H. Kimura, D. Tokuda, K. Hosoya, T. Ikegami, N. Ishizuka, H. Minakuchi, K. Nakanishi, Y. Shintani, M. Furuno, and K. Cabrera, Anal. Chem. 2004, 76, 1273-1281. 33. M. M. Bushey and J. W. Jorgenson, Anal. Chem. 1990, 62, 161. 34. C. Yang, L. Zhang, H. Liu, W. Zhang, and Y. Zhang, J . Chromatogr. A 2003, 1018, 97-103. 35. L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Multidimensional Chromatography, Wiley, Chichester, England, 2002. 36. F. M. Lancas, in Multidimensional Chromatography, L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Wiley, Chichester, England, 2002, Chapter 6. 37. T. L. Chester, in Multidimensional Chronzutography, L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Wiley, Chichester, England, 2002, Chapter 7. 38. Sz. Nyivady, in Multidimensionul Chromutogruphy, L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Wiley, Chichester, England, 2002, Chapter 8. 39. M. D. Degen and V. T. Remcho, in Multidimensional Chromutography, L. Mondello, A. C. Lewis, and B. D. Bartle (eds), Wiley, Chichester, England, 2002, Chapter 9. 40. J. C. Anders, BioPharm 2002, April, 32. 41. P. HuSek and P. Simck, LC-GC' 2001, 19, 986. 42. E. D. Katz (ed), High Performance Liquid Chromutogruplzy: Principles und Methods in Biotechnology, Wiley, Chichester, England, 1996. 43. C. F. Simpson, in High Pe'erg'orrnance Liquid Chromutogruphy: Pritzciples and Methods in Biotechnology, E. D. Katz (ed), Wiley, Chichester, England, 1996, Chapter 1. 44. R. L. Cunico, K. M. Gooding, and T. Wehr, Rusk HPLC and CE of Biomolecules, Bay Bioanalytical Laboratory, Richmond, CA, 1998. Also distributed by Varian, Inc., Walnut Creek, CA; www.ourianinc.com. 45. T. P. Bradshaw, Introduction to Peptide and Protein HPLC: A U~servGuide, Phenomenex, Torrance, CA, 2000. 46. A. Amini, A. Chakraborty, and F. E. Regnier, J . Chromatogr. B 2002, 772, 35-44, 47. M. C. Posewitz and P. Tcmptst, Anal. Chem. 1999, 71, 2883. 48. R. R. Walters, Anal. Chem. 1985, 57, 1099A. 49. I. Parikh and P. Cuatrecasas, Chem. Eng. News 1985, 63341, 17.

458

SO.

51. 52.

53. 54. 55. 56. 57. 58. 59. 60. 61.

62. 63. 64. 65. 66. 67.

68.

69.

70. 71. 72. 73. 74. 75. 16. 11. 78. 79.

SPECIAL APPLICATIONS

P.-0. Larsson, M. Glad, L. Hansson, M.-0. Mansson, S. Ohlson, and K. Mosbach, Adu. Chromatogr., N. Y . 1983, 21, 41. D. S. Hage, Adu. Chromutogr., N . Y . 2003, 42, 377. C. R. Lowe and P. D. G. Dean, Afinity Chromutogruphy, Wiley, New York, 1974. J. Turkova, Afinity Chromutogruphy, Elsevier, Amsterdam, 1978. W. H. Scouten, A,finity Chromatography, Wiley, New York, 1981. I. Chaiken, M. Wilchek, and I. Parikh (eds), Aflinify Chromatogruphy arid Biological Recognition, Academic, New York, 1984. H. Schott, Affinity Chromatography: Templute Chromutography of N~icleicAcids and Proteins, Dekker, New York, 1985. P. Mohr and K. Pommerening, A,flfinity Chroniatograpl~y,Dekker, New York, 1985. J. Chromatogr. B 2002, 768, 1-214. J . Biochem. Biophys. Metli. 2001, 49, 1-743. W-C Lee and K. H. Lee, Anal. Biochem. 2004, 324, 1. N. E. Labrou, J. Chromatogr. B 2003, 790, 67-78; 93 references. V. 1. Muronetz and T. Korpela, J. Chromutogr. B 2003, 790, 53-66; 134 references. V. T. Remcho and Z. J. Tan, Anal. Chem. 1999, 71, 248A-255A. K. Haupt, Anal. Chem. 2003, 75, 376A-383A. J . Chromatogr. B 2004, 804(1), 1-254. Y. Lu, C. Li, H. Zhang, and X. Liu, Anul. Chim. Act 2003, 489, 33-43. M. S . Lesney, “Sticking with Affinity Chromatography” Modern Ding Discouety 2002, 3121, 27-29. P. G. Righetti and A. Castagna, “Recent Trends in Proteome Analysis,” in Advances in Chromatography, Vol. 42, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 2003. K. Benedek, in High Perfiormarice Liquid Chromatography: Principles and M e t h ods in Biotechnologv, E. D. Katz (ed), Wiley, Chichester, England, 1996, Chapter 9. Application Note 9804, VYDAC, Hesperia, CA, 1998; www.uydac.com. Application Note 9902, VYDAC, Hesperia, CA, 1999; www.uydac.com. J. Eshraghi and S. K. Chowdhury, Anal. Chem. 1993, 65, 3528. T. Wehr, L C / G C No. A m . 2001, 19, 102. T. Wehr, LC/GC No. Am. 2003, 21, 214. H. Wang and S. Hanash, J. Chromatogr. B 2003, 787, 11-18. E. P. Romijn, J. Krijgsveld, and A. J. R. Heck, J . Chromutogr. A 2003, 1000, 589-608; 112 references. J . Chromatogr. A 2003, 1009, 1-236. A. R. Ivanov, L. Zang, and B. L. Karger, Anal. Chem. 2003, 75, 5306-5316. D. T. Gjerde, C. P. Hanna, and D. Hornby, DNA Chromatography, Wiley-VCH, Weinheim, Germany, 2002.

REFERENCES

459

80. J. M. Wages and E. D. Katz, in High Perforniance Liquid Chromatography: Principles and Methods in Biotechnology, E. D. Katz (ed), Wiley, Chichester, England, 1996, Chapter 7. x1. S. N. Krylov and N. J. Dovichi, A n d . Chem. 2000, 72, 1 1 1 R- 128R. 82. S. HLIand N. J. Dovichi, Anal. Chem. 2002, 74, 2833-2850. 83. J. A. Queiroz, C. T. Tomaz, J. M. S. Cabral, J. Biotechnol. 2001, 87, 143. 84. J . Porath, J. Chromutogr. 1986, 376, 33 1. 85. S . Pihlman, J. Rosengren, S. Hjertn. .I. Chro/nafogr. 1977, 131, 99. 86. S . Roe in Protein Purification Methods: A Pructical Approach, E. L. V. Harris, S. Angal (eds), IRL Press, Oxford, 1989, pp. 221-232. 87. H. P. Jennissen, in Biochromutogruphy, M. A. Vijayalakshmi (ed), Taylor & Francis, London, 2002, pp. 46-71. 88. M. T. W. Hearn, C h r o m a t o p . Sci. Series 2002, 87, 99. 89. Z. El Rassi, J. Chromatogr. Library 2002, 66, 41. YO. w w w .chromatography. amershumbiosciencesmni 91. www.to.sohhiosep.com. 92. A. M. Rouhi, Chem. Eng. News 2003, 81(18), 45. 93. R. S. Cahn, C. K. Ingold and V. Prelog, Angew. Chem. 1966, 78, 413-447; Angew. Chem. Infernat. Ed. Eng. 1966, 5, 385-415; and V. Prelog and G. Helmchen, Angew. Chem. 1982, 94, 614-631, Angew. Chem. Internut. Ed. Eng. 1982, 21, 567-583. 94. C. R. Clark and J. M. Barksdale, A n d . Chem. 1984, 56, 958. 05. R. H. Buck and K. Krummen, J . Chronzutogr. 1984, 315, 279. 96. C. H. Lochmullcr and R. W. Souter, J . Chromatogr. 1975, 113, 283. 97. T. E. Beesley and R. P. W. Scott, Chiral Chromatography, Wiley, Chichester, England, 1998. 98. Restek, A Guide to the Analysis of Chirul Compounds by G C , Rcstek Corp., Bellefonte, PA, 1997; www. restekcorp.com. 99. Supelco, Chiral Cyclodextrin Cupilluty GC Columns, Supelco, Inc., Bellefonte, PA, 1995; www..sigmaaldrich.com/brund.s/supelco home.htm1. 100. V. Schurig, J . Chromatogr. A 2001, 906, 275-299; 168 references. 101. T. J. Ward, Anul. Chem. 2002, 74, 2863-2872; 19 1 references. 102. Macherey-Nagel, Dueren, Germany D004. w.Macherey-NageLch. 103. W. H. Pirkle, in Chromatography and Separation Chemistly, S. Ahuja (ed), ACS Symposium Series 297, American Chemical Society, Washington, D.C., 1986. 104. R. Dappen, H. Arm, and V. R. Meyer, J. Chromatogr. 1986, 373, 1. 105. F. Gasparrini, D. Misiti, and C. Villani, J . Chromatogr. A 2001, 906, 35-50. 106. N. M. Maier, P. Franco, and W. Linder, J . Chromatogr. A 2001, 906, 3-33. 107. W. H. Pirkle, M. H. Hyun, and B. Bank, J. Chromatogr. 1984, 316, 585. 108. D. W. Armstrong, T. J. Ward, R. D. Armstrong, and T. E. Beesley, Science 1986, 232, 1132. 109. D. W. Armstrong and W. De Mond, J. Chromutogr. Sci. 1984, 22, 411.

460

SPECIAL APPLICATIONS

110. V. A. Davankov, A. A. Kurganov, and A. S. Bochkov, Adu. Chromutogr. N . Y. 1983, 22, 71. 111. J. Hermansson, J . Chromatogr. 1983, 269, 71. 112. V. S . Sharp, M. L. Letts, D. S. Risley, and J. P. Rose, Chirality 2004, 16, 153-161. 113. W. H. Pirkle and R. S. Readnous, Anul. Chem. 1991, 63, 16-20. 114. C. Pettersson, J . Chromatogr. 1984, 316, 553. 115. D. W. Armstrong and B. Zhang, Anal. Chem. 2001, 73, 557A-561A. 116. See, for example, ProntoSil columns manufactured by Bischoff Chromatography and marketed in the USA by MacMod; www.rnac-mod.com. 117. Anonymous, Chiral Application Guide, Regis Technology, Morton Grove, IL, 2004. 118. Chrom Tech, Chiral Application Handbook, Application Note 19, ChromTech, Apple Valley, MN, 55124; www.chromtech.co.uk. 119. ASTEC, Chirobiotic Handhook, 4th ed., Advanced Separation Technologies, Inc. (ASTEC), Whippany, NJ, 2002; Cyclobond LC Handbook, 6th ed., ASTEC, 2002; Chiraldex GC Handbook, 6th ed, ASTEC, 2002; www.astecusa.co. 120. www.macherey-nagel.com. 121. M. J. Desai and D. W. Armstrong, J . Chromutogr, A 2004, 1035, 203-210. 122. C . Roussel, N. Vanthuyne, M. Serradeil-Albalat, and J-C. Vallejos, J . Chrornutogr. A 2003, 99.5, 79-85. 123. T. E. Beesley and R. P. W. Scott, Chirul Chrornutography, Wiley, Chichester, England, 1998, p. 309. 124. A. Alak and D. W. Armstrong, A n d . Chem. 1986, 58, 582. 125. F. Wang and M. G. Khaledi, “Chiral Separations by Capillary Electrophoresis,” in High Performance Capilluly Electrophoresis, M. G. Khaledi (ed), Wiley, New York, 1998. 126. T. J. Ward, A n d . Chem. 2000, 72, 4521-4528. 162 references. 127. G. Subramanian (ed), Chiral Sepuration Techniques, 2nd ed., Wiley, New York, 200 1. 128. Y. Ito, M. Weinstein, I. Aoki, R. Herada, E. Kimura, and K. Nunogaki, Nuturc 1966, 212, 985. 129. Y. Ito and W. D. Conway, Anul. Chem. 1984, 56, 534A. 130. Y. Ito, Adv. Chrornutogr. N . Y. 1984, 24, 181. 131. M. Knight, in Advancesin Cizrornutogruphj>Vol. 31, J. C. Giddings, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 1992, Chapter 4. 132. Y. Ito and W. D. Conway (cds), High Speed Counterunent C/irornutograpli, Wiley, New York, 1996. J. M. Menet and D. Thiebaut, (eds),Coiirztcrcurrcnt Chromatography, Marcel 133. Dekker, New York, 1999. 134. Y. Ito, F. Yang, P. Fitze, J. Powell, and D. Ide, J . Chromutogr. A 2003, 1017, 71-81. 135. A. Berthod, M. J. Ruiz-Angel, and S. Carda-Broch, Anul. Chem. 2003, 7.5, 5886-5894.

REFERENCES

461

136. C. M. Harris, Anal. Chem. 2003, 75, 373A. 137. A. Berthod and B. Billardello, in Advances in Chromatography Vol. 40, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 2000, Chapter 10. 138. J. Cazes, Am. Lab. 1990, 22(14), 40D. 139. J. C. Giddings, Am. Lab. 1978, 10(5), 15-31. 140. M. Martin, in Advances in Chromatography Vol. 40, P. R. Brown and E. Grushka (eds), Marcel Dekker, New York, 2000, Chapter 1, pp. 1-138. 141. R. C. Willis, Mod. Drug Disc. 2003, 6(6), 23-25. 142. D. R. Reyes, D. Iossifidis, P-A Auroux, and A. Manz, Anal. Chem. 2002, 74, 2623-2636; 342 references. 143 P-A. Auroux, D. Iossifidis, D. R. Reyes, and A. Manz, Anal. Chem. 2002, 74, 2637-2652; 372 references. 144. D. Figeys and D. Pinto, Anal. Chem. 2000, 72, 330A-335A. 145. S . Chen, PhurmaCenomics 2003, 3(8),56-63. 146. L. J. Jin, J. Ferrance, Z. Huang, and J. P. Landers, in Handbook o f Modern Pharmaceutical Analysis, S. Ahuja and S. Scypinski (eds), Academic, San Diego, 2001, Chapter 16. 147. J. Khandurina and A. Guttman, J . Chromatogr. A 2002, 943, 159-183. 130 references. 148. J. P. Landers, Anal. Chem. 2003, 75, 2919-2927; 101 references. 149. W. R. Lacourse, Anal. Chem. 2002, 74, 2813-2832. 150. C . M. Harris, Anal. Chem. 2003, 75, 65A-69A. 151. B. S. Broyles, S. C. Jacobson, and J. M. Ramsey, Anal. Chem. 2003, 75, 276 1-2767. 152. M. J. Felton, Anal. Chem. 2003, 75, 505A-508A.

This Page Intentionally Left Blank

16 SELECTION OF A METHOD This chapter is intended to provide some suggestions for the novice chromatographer who is charged with the responsibility of devising a method of chromatographic analysis for a given problem. It assumes that a decision has already been made that chromatography is the best choice for the analysis, and this discussion does not compare chromatography with other instrumental methods. It suggests a method of attack for a separation problem and compares the various forms of chromatography in choosing a method. 16.1

METHODS OF ATTACK

There are three ways to approach a separation problem: by experimentation, using the best theoretical predictions such as those contained in this monograph; by searching the literature in an organized and thorough way to learn what others have done; and by asking someone who may know. The last suggestion may appear to be facetious, but it is probably the best place to begin. If there is anyone you know who might have had some experience with the type of problem you are trying to solve, he or she may be able to give you more information in a shorter time than either of the other two modes of action. There may be someone in the same lab or a similar lab within your company or school to whom you can turn for advice. Too often, resource Cl~romatography:Conceptr and Contracts, Second Edition. ISBN 0-471-47207-7 0 2005 J o h n Wiley & Sons, Inc.

By James M. Millet 463

464

SELECTION OF A METHOD

people within an organization are not consulted and insufficient attention is paid to the transfer of information between labs and departments. There is no benefit to reinventing the wheel. There are a number of other people or organizations to whom you can turn for help. Many of them, such as the FDA and EPA, were discussed in Chapter 1 and often their publications are available online. Such publications may contain methods for the analyte of interest or for closely related compounds. The company from whom you who purchased your chromatographic equipment or from whom you plan to purchase equipment may be able to help you. A number of these companies maintain applications laboratories for this purpose. Contact your sales representative. Several journals provide reviews and bibliographic data. The popular journal, Analytical Chemistry, publishes biennial reviews covering many areas of interest to chromatographers. Currently (20021,’ these reviews are: GC, planar chromatography, supercritical fluid and unified chromatography, column LC: Equipment and instrumentation, capillary electrophoresis for the analysis of biopolymers, chiral separations, and micro total analysis systems, Part 2. The coverage and style depends on the authors for a given year or topic, and the topics chosen vary from year to year. In the alternate years the journal’s reviews cover the different areas of application, and these too are a good source of information. Distributors of chromatographic supplies and accessories often employ experts who may be able to help you. Some publish regular newsletters and product bulletins. If you need the names of such firms, check the directories published annually by Analytical Chemistry and other journals. 16.2 THE INTERNET

These days the information contained in the literature and available from chromatographic companies and supply houses, as well as a wide range of general sources, can be located via the Internet. Chapter 1 contains many helpful URLs, and many of the other chapters also contain a few relevant links. An additional list is given at the end of this chapter. One problem with URLs, of course, is that they can change without notice. This section will discuss some of the ways of obtaining information from the Internet, beginning with conventional searching of journal databases. Literature Searching Searching the literature is easier and faster since the necessary databases are available on the Internet. By far the most complete and useful database is

16.2 THE INTERNET

465

STN International,2 which is operated by the Chemical Abstracts Service, CAS. There is a fee for using this service. In a brief discussion about searching the literature for capillary electrophoresis references, Weinberger has provided some good suggestions for searching in general. He recommends the electronic database at the National Library of Medicine, P ~ b M e d While .~ it is limited to biomedical content, it can be a useful source for chromatographic references. Science Direct’ is a library service that is available to institutions for a fee. The database is restricted to a relatively few publishers, but it contains many journals useful to chromatographers. For example, a search for “chromatographic analysis residual solvents” (using the Boolean AND operator) found 22 references, most of which were relevant. Members can print out the articles found. Another service that is free for searches is available from Ingenta.‘ Its database is not nearly as extensive, but it does include the Journal of Chromatography A and B , and the Journal of Chromatographic Science, among others. To obtain a copy of an article, page charges are levied and the manuscript is faxed from a participating library. The same search as the one conducted on Science Direct found eight citations. There are a variety of general search engines available on the web such as Google’ and Yahoo.’ A repeat of the same search, performed on Google, produced 24,500 hits. Not all of them were examined, but included in the first group were a citation from the Journal of Chromatographic Science, which could be downloaded for $4 a page; and another for an 11-page bulletin published by Supelco, which could be downloaded for free. A serious and comprehensive search is probably better done using one of the scientific search engines.

Other Sources of Information on the Internet

The governmental and regulatory sites were presented in Chapter 1 and will not be repeated. Also, many companies, universities, technical societies, and private individuals have established web sites to promote and teach chromatography. They contain a variety of types of information such as guides, manuscripts, slide presentations, courses, and e-seminars, many of which can be downloaded without charge. See, for example, Katzekevich and McNair’s work on HPLC,” Irgum’s analytical chemistry “springboard” of information,“’ Regis’s booklet on stereochemistry,” or Supelco’s on SPME,” as well as Agilent’s e-seminars, which provide question-and-answer opportunities for registrants.’j Many, many more can be found by searching the URLs given at the end of the chapter.

466

SELECTION OF A METHOD

Another activity is chat rooms or e-groups. Three of interest to chromatogaphers are c h i r a l c h r ~ m , 'Chromatography ~ Forum,ls and Chrom-L." Several journal publishers will send newsletters and/or tables of contents of current journals upon request. They include: Wiley," Elsevier," EBSCOhost electronic journals,'" and the Virtual Journal of Proteomics.'" And, finally, many newsletters are available, including those from Agilent, Mac-Mod, Phenomenex, Restek, Supelco, Varian, and Wiley; see end of chapter for URLs. The latter is called SeparationsNow, and it has separate newsletters on GC, HPLC, electrophoresis, hyphenated techniques, and other techniques. In addition, its web site2' has an extensive listing of links that far exceeds those given here. Adept Scientific plc makes available, free online from the library4science store, e-books on chromatography in their Chrom-Ed Series by Scott.'* There are 11 books, most on fundamentals, but also including one on preparative chromatography. Some can be downloaded as pdf files for $4.95.

16.3

EXPERIMENTAL APPROACH

It is unlikely that the literature will fail to reveal at least some information that can provide a clue for the initial experiments. However, if that were to be the case, and if there were no information available about the sample, how would one proceed? Let us assume that a separation is needed and that it has been decided that the best possible separation would be by chromatography. Scott2' was the first to suggest that there are three objectives that need to be considered in a chromatographic analysis: speed, resolution, and quantity of sample to be separated. Since these three objectives are interrelated, improvements in one are usually achieved at the expense of the others, and their relationship is best depicted as a triangle, as shown in Figure 16.1. As a rough guide to method selection, the triangle is divided into sections showing the most likely method for achieving the desired performance with respect to the three objectives. The dividing lines are arbitrary, of course, and this approach is oversimplified, but it does give an overview of the most common uses of the various chromatographic modes. Table 16.1 provides another comparison of the various chromatographic techniques by listing the parameters that operate in each of them. Generally, as the number of operating variables increases, so does the flexibility and range of applications. While it presents an interesting contrast, it has the limitation that it does not consider other attributes such as detection and quantitation.

16.3 EXPERIMENTAL APPROACH

467

Speed

Resolution

Cuantity Figure 16.1. Method selection triangle.

Table 16.1 Parameters Available in the Various Chromatographic Modesa

GC

LC

SFC

TLC

SP T

SP T MP

SP T MP Density

SP (T) MP Shape

“Abbreviations: SP = stationary phase; MP = mobile phase; T = temperature

The most important first step in designing a particular analysis is to thoroughly investigate the nature and the history of the samplek) to be analyzed. Are the samples solids, liquids, gases, solutions, suspensions, or emulsions? Are liquid samples delivered in aqueous or nonaqueous medium? Are samples of biological origin and expected to contain salts, macromolecules, or other potential interferences? Are the analytes expected to be volatile or nonvolatile? Has any previous pretreatment been done on the samples? What are the analytes in question? Are there closely related species to be separated? Are the analytes expected to be present in trace quantities or as a major component? What kinds of instrumentation are available for these analyses? Answers to questions such as these are essential to further method selection. Before a sample is introduced into a GC, some evidence should be sought to ensure that it is completely volatile. Its source and history might provide

468

SELECTION OF A METHOD

that information, as would a microdistillation. If total volatility cannot be confirmed, chromatographic analysis should begin with LC.

Volatile Samples

If it has been determined that the sample is volatile enough to be run by GC, an OT column should be chosen to match the sample as closely as possible; that is, a nonpolar column such as DB-1 should be chosen for a nonpolar sample, and a polar column, such as Carbowax for a polar sample. A DB-5 column is a good universal screening column. For initial screening, a short length of 10-15 m is desirable, as is a thin film of about 0.25 p m . A rather fast temperature program of about 20"/min should be used, starting at a temperature below the lowest boiling component in the sample and proceeding to the highest boiler or to the upper temperature limit of the column (whichever is lower). The upper temperature should be maintained until it can be assumed that all analytes have been eluted. Depending on the quality of the separation obtained on the fast screening run, temperatures and/or stationary phases can be selected to improve the analysis.

Nonvolatile Samples

Figure 16.2 shows one flow sheet for selecting a chromatographic method. Normally, nonvolatile samples are run by HPLC, but consideration should be given to the possibility of derivatizing the sample to get volatility adequate for GC. (Consult Chapter 14 for possible derivatizations.) Other variables considered in Figure 16.2 are the number of samples and the importance of speed. The analysis of complex samples can probably be facilitated if they are cleaned up prior to analysis by using short columns and/or extractions, as described in Chapter 14. Flash chromatography (Chapter 8) has become very popular for fast, crude preliminary separations. Other preliminary treatments to be considered are drying to remove water, derivatization, pH adjustment, and preconcentration. Care must be taken in these pretreatments, however, to ensure that the analytes of interest are not removed. Fast screening is often accomplished by TLC, as indicated in Figure 16.2. The choice of stationary phase can be based on sample solubility and polarity according to Figure 16.3. Silica gel is commonly used for a test run, but a reversed-phase system can also be used. For normal-phase analysis, a polar

16.3

EXPERIMENTAL APPROACH

469

Does the sample vaporize readily?

Is the number of similar samples small?

Can a stable

volatile derivative be formed?

tQl

I I

3

I

Is speed important (for a limlted number of samples)?

I

I

I

Figure 16.2. Choice of a chromatographic method. Courtesy of Regis Chemical

mobile phase is used first, followed by less polar ones and mixtures, depending on the chromatographic results. The development of an HPLC method was introduced in Chapter 8. Figure 8.41, typical of the flow diagrams provided by many manufacturers of HPLC columns and column packings, can serve as a simplified method selection diagram. If there is any reason to suspect that the sample contains analytes with a wide range of molecular weights or any analytes with molecular weights over 2000, then SEC should be considered as a first step, as shown in the figure. Subsequent steps are based on sample solubility and sample polarity. For small molecules (MW s 2000 in the figure), the essential choice is between normal phase (for the less polar analytes), reversed phase (for the more polar analytes), and ion exchange (for the ionic analytes). Obviously, there are many choices, and the selection process is not simple.

P

-4 0

~

AQUEOUG SOLVENTS

ORGANIC SOLVENTS

SAMPLE

SOLUBlLlTV

-IONIC

-

c NONlONlC

LOW

HIGH

-

-

SAMPLE

POLARITY

PARTITION

-

PARTITION

LION EXCHANGE

-

PARTITION

PH

SAMPLE

BASIC (ANON)

(CATION)

ACIDIC

BASIC TO NEUTRAL

ACIDIC TO NEUTRAL

-- -

REVERSED PHASE-PARTITION

-

-ADSORPTION

SEPARATION MECHANISM

PEI CELLULOSE

DEAE CELLULOSE

PERMUTIT T

CM CELLULOSE

BONDED REVERSED PHASE

BONDED REVERSED PHASE

CELLULOSE MONOACETATE

BONDED REVERSED PHASE

BONDED PHASE

ALUMINUM OXIDE

40

STATIONARY PHASE

REFERENCES

471

Gradient elution can facilitate HPLC screening, just as programmed temperature does in GC.

16.4

SUMMARY

In most cases a combination of approaches is probably desirable-some discussion with colleagues, some literature searching, and some experimentation. Personal preference and availability of instrumentation are strong factors in the decision-making process, and they have not been included in this discussion. In this regard, the authors preferences undoubtedly influenced the material presented, but it is hoped, only minimally.

REFERENCES 1. Fundamental Reviews, Anal. Chem. 2002, 74, 2623-2918. (Some 2004 reviews can be found in issue 12 of volume 76.) 2. www.stn weh.cas.org. 3. R. Weinberger, Am. Lub. 2004, 36(3), 60. 4. Click on PubMed at: www.ncbi.nlm.nih.gov. 5. www. sciencedirect. corn. 6. www.ingenta .corn. 7. ww w.google.com 8. ww w.yahoo. corn. 9. http://hplc.chem.shu.edu/hplc/index.html. 10. www.unuchem.umu.se/jurnpstution.htm. 1 1. www.registech.corn /chirul/chiralreuiew. pdf. 12. www.sigmaaldrich.com Select eBookShelf. 13. www.chem .ugilent. corn. 14. listes.cru .f;/wws/arc / c h iralchrom. 15. www .chrornforum.corn. 16. groups.yahoo.com /group/chrom-I. 17. www. varianinc.com. 18. www. elsevier.corn. 19. ejournals.ehsco.com. 20. www. proteomiesuj.com. 2 1. www.separationsnow.com. 22. www.luboratorytalk.com/books/chem/chrom. 23. R. P. W. Scott, in Gas Chromatography, 1964, A. Goldup (ed), Institute of Petroleum, London, 1965, pp. 25-37.

472

SELECTION OF A METHOD

SOME INTERNET WEB SITES OF INTEREST TO CHROMATOGRAPHERS

All entries begin with http://www. 1. Agilent Technologies 2. Alltech Associates 3. Alpha Omega Technologies, Inc. 4. Altria’s site on electrophoresis 5. American Laboratory 6. Analtech, Inc. 7. Antek Instruments 8. Applied Biosystems 9. ASTEC 10. J. T. Baker 11. Beckman Coulter, Inc 12. Bio-rad Laboratories 13. Burdick and Jackson 14. Camag 15. Chiral Technologies, Inc. 16. Chromatography Research Supplies 17. ChromPack 18. ChromTech Co 19. Dionex Corp. 20. ES Industries 21. GERSTEL, Inc. 22. Gilson, Inc. 23. Gow-Mac Instrument Co. 24. Grace,Vydac 25. Hamilton Company 26. High Chrom 27. J & W Scientific

28. 29. 30. 31.

Jones Chromatography LC-GC North America LC Resources Leco Corp

ch em.agilen t .corn alltechweb.com alphaomegatech.corn ceandcec.com iscpubs .corn th inlayer. corn antekhou.com appliedbiosystems .corn astecusa.com jtba ker. corn beckmancoulter.com biorad. corn bandj.com amag.com chiraltech .corn chromes .corn Varianinc.com chromtech.co.uk dionex. corn esind.com gerstelus.com Gilson .corn gow-rnuc.com y d a c .corn hamiltoncompany. corn hichrom. co.uk chem. agilent .corn /cag /cabu/jandw. htm joneschrom. corn chrornatographyonLine.com lcresources.corn /homepage leco.com

SOME INTERNET WEB SITES OF INTEREST TO CHROMATOGRAPHERS

32. Leap Technologies 33. Mac-Mod Analytical 34. Phenomenex, Inc. 35. 0 1 analytical 36. Perkin-Elmer Corp 37. Pierce Biotechnology 38. Polymer Laboratorie 39. Regis Technologies 40. Restek Corp. 41. RS Tech Corp. 42. Scott Specialty Gases 43. Scientific Resources 44. SGE, Inc. 45. Supelco 46. ThermoElectron Corp. 47. TosoHaa 48. UpchurchScientific 49. Varian, Inc 50. VICI Valco Instruments Co Inc. 51. VYDAC 52. Waters Corp. 53. Whatman 54. Zoex Corp.

1eaptec.com mac-rnod.com phenomenex.com oico.com 1as.perkinelmer.com piercenet.corn polymerlabs.com registech .corn restekcoip.com rstechcoip.com scottgas.com sciresources.com sge .com sigmaaldrich.corn thermo.com tosohaas.com upchurch.com varianinc.corn uici.com ydac.com waters.com whatrnan.com zoex.com

473

This Page Intentionally Left Blank

APPENDIX

A

ICH GLOSSARY

ICH HARMONISED TRIPARTITE GUIDELINE Text on Validation of Analytical Procedures Q2A

This Guideline has been* developed by the appropriate ICH Expert Working Group and has been subject to consultation by the regulatory parties, in accordance with the ICH Process. At Step 4 of the Process the final draft is recommended for adoption to the regulatory bodies of the European Union, Japan, and USA. 1. ANALYTICAL PROCEDURE The analytical procedure refers to the way of performing the analysis. It should describe in detail the steps necessary to perform each analytical test. This may include but is not limited to: the sample, the reference standard and the reagents preparations, use of the apparatus, generation of the calibration curve, use of the formulas for the calculation, etc.

2. SPECIFICITY Specificity is the ability to assess unequivocally the analyte in the presence of components which may be expected to be present. Typically these might include impurities, degradants, matrix, etc. Lack of specificity of an individual analytical procedure may be compensated by other supporting analytical procedure(s1. *Source:Fed. Reg. 1995, 60, 11260-1 1262. Chromatography: Concepts und Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

475

476

APPENDIX A

ICH GLOSSARY

This definition has the following implications: Identification: to ensure the identity of an analyte. Purity tests: to ensure that all the analytical procedures performed allow an accurate statement of the content of impurities of an analyte, i.e., related substances test, heavy metals, residual solvents content, etc. Assay (content or potency): to provide an exact result which allows an accurate statement on the content or potency of the analyte in a sample.

3. ACCURACY The accuracy of an analytical procedure expresses the closeness of agreement between the value which is accepted either as a conventional true value or an accepted reference value and the value found. This is sometimes termed trueness. 4. PRECISION The precision of an analytical procedure expresses the closeness of agreement (degree of scatter) between a series of measurements obtained from multiple sampling of the same homogeneous sample under the prescribed conditions. Precision may be considered at three levels: repeatability, intermediate precision, and reproducibility. Precision should be investigated using homogeneous, authentic samples. However, if it is not possible to obtain a homogeneous sample, it may be investigated using artifically prepared samples or a sample solution. The precision of an analytical procedure is usually expressed as the variance, standard deviation, or coefficient of variation of a series of measurements.

4.1. Repeatability Repeatability expresses the precision under the same operating conditions over a short interval of time. Repeatability is also termed intra-assay precision. 4.2. Intermediate precision Intermediate precision expresses withinlaboratories variations: different days, different analysts, different equipment, etc. 4.3. Reproducibility Reproducibility expresses the precision between laboratories (collaborative studies, usually applied to standardization of methodology). 5. DETECTION LIMIT The detection limit of an individual analytical procedure is the lowest amount of analyte in a sample which can be detected but not necessarily quantitated as an exact value.

ICH HARMONISED TRIPARTITE GUlDLlNE

477

6. QUANTITATION LIMIT The quantitation limit of an individual analytical procedure is the lowest amount of analyte in a sample which can be quantitatively determined with suitable precision and accuracy. The quantitation limit is a parameter of quantitative assays for low levels of compounds in sample matrices, and is used particularly for the determination of impurities and/or degradation products. 7. LINEARITY The linearity of an analytical procedure is its ability (within a given range) to obtain test results which are directly proportional to the concentration (amount) of analyte in the sample. 8. RANGE The range of an analytical procedure is the interval between the upper and lower concentration (amounts) of analyte in the sample (including these concentrations) for which it has been demonstrated that the analytical procedure has a suitable level of precision, accuracy and linearity. 8. ROBUSTNESS The robustness of an analytical procedure is a measure of its capacity to remain unaffected by small, but deliberate variations in method parameters and provides an indication of its reliability during normal usage.

This Page Intentionally Left Blank

APPENDIX

B.l

B

TYPICAL CHROMATOGRAPHIC CALCULATIONS

Figure B.l shows a liquid chromatographic separation of acetylacetonate chelates of beryllium, chromium, ruthenium, and cobalt. The conditions of the separation were: Stationary phase: a ternary mixture of 64% ethanol, 34% water, and 1.6% isooctane. Mobile phase: a ternary mixture of 98% isooctane, 2% ethanol, and 0.08% water. V, 1.32 mL Vs 0.142 mL F 0.291 mL/min Column length L 23 cm Inside diameter of columns 2.7 mm Inlet pressure Pi 50 atm Chart speed 1 cm/min

Calculate p for this column, and for both the beryllium and the cobalt chelates calculate the following parameters: V,, V,, k , R , K , N , and H . For the unresolved mixture of chromium and ruthenium chelates, calculate the resolution, R,, assuming that the peaks are of equal height. Chrornutography: Concepts and Contrasts, Second Edition. ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

By James M. Miller

479

0.6

t-

Figure 6.1. Liquid chromatographic separation of metal chelates.

481

482

APPENDIX B

The partition coefficients for these chelates have been determined by static methods and found to be as follows: beryllium, 4.4; chromium, 16.1; ruthenium, 23.2; and cobalt, 33.5. Figure B.2 shows a gas chromatogram of toluene. The operating conditions and some other necessary data are given below: Column: 48 X 0.25 in. (outside diameter); inside diameter = 5.0 mm; packed with 12.0 g containing 20% (w/w) dinonyl phthalate on 80/100 mesh Chromosorb P Column temperature Ambient temperature Ambient pressure P,,

100°C 25°C 760 torr

Outlet flow measured at ambient conditions with a soap-bubble flowmeter Chart speed Density of Chromosorb P Density of dinonyl phthalate Vapor pressure of water at 25°C Calculated pressure correction factor J

80 mL/min 1 in./min 2.26 g/mL 1.03 g/mL 24 torr 0.65

Calculate as many chromatographic parameters as you can, including the total volume of the column. The actual calculations follow. Note that in this example the symbol V, is used to denote the mobile phase (which is a gas) rather than the general symbol VM. Similarly, V , is used rather than Vs for the stationary phase. 8.2

CALCULATIONS: GAS CHROMATOGRAM OF TOLUENE

1. Total volume of column, V,. Radius = 0.25 cm. V , = 3.14 X (0.2512X 4 X 12 X 2.54 = 23.9 mL. 2. t , = 19.45 cm X 1 min/in. X 1/2.54 in./cm = 7.66 min. 3. t k = (19.45 - 0.70) x 1/2.54 = 7.39 min, or t , = 0.70/2.54 = 0.28 and t k = 7.66 - 0.28 = 7.38 min. 4. F, = 80 X 373/298 X 736/760 = 97 mL/min. F, = 97 x 0.65 = 63 mL/min. 5. VR = 7.66 x 97 = 743 mL. 6. V; = 7.39 x 97 = 716 mL. 7. V: = 743 X 0.65 = 482 mL. 8. V , = j x VA = 716 x 0.65 = 466 mL, or V , = V i - V: = 482 - 18 = 464 mL.

8.2

CALCULATIONS: GAS CHROMATOGRAM OF TOLUENE

483

9. V: = t , x Fc = 0.28x 63 = 17.6mL. 10. V L= ~ 12 g x 0.20x 1/1.03 mL/g = 2.33mL. 11. Vss = 12 x 0.80x 1/2.26 mL/g = 4.25mL. V: = 17.6 72.8% = cT Vl,= 2.3 9.5% V,, 17.7% 24.2= total volume; compare with volume calculated in item 1.

=> -

12.

FC

-

63

A x ET x 60 - 0.182x 0.73x 60 = 7‘88 cm/sec, u = L/60tM = 122/60 x 0.28= 7.26cm/sec. =

Or

13. /3 = 17.6/2.23= 7.6. 14. K = V,/V, = 465/2.33= 199. 15. k = K / P = 199/7.6= 26.2. Also note: V i = VG(1 + k ) = 17.6(27.2)= 479 mL (compare with item 7). 16. R = V:/Vi = 17.6/482= 0.0366,or toluene spends 3.7% of its time

in the mobile phase. (1 - R) = 0.9634,or toluene spends 96.34% of its time in the stationary phase.

l-R

k = - R = 96.3/3.6= 26.3(compare with item 15). 17. N = 16(19.45/1.96)*= 16(9.93)’ = 16(99) = 1584 plates. 18. H = L / N = 122 cm/1584 = 0.077cm = 0.77mm. 19. h = H / d , = 0.77/0.163= 4.4

This Page Intentionally Left Blank

INDEX Absorption, 44-45 Accuracy, definition of, 27, 476 Activity coefficient, 142 ACS, see American Chemical Society Adjusted retention volume, 48 Adsorption, 44-45 Affinity chromatography, 434-437 ALSSA, 13, 14 American Chemical Society, 12 Analog-to-digital conversion (A/D), 284, 297-299 Analyzers, for MS, 318-321 AND1 protocols, 14 ANSI, 13, 17, 18 Anion analysis by HPLC, 223 AOAC, 13, 14 Area, measurement of, 297-299 Area normalization, 300-301 ASTM, 13, 14-16 Asymmetry of peaks, 55-57, 96 Atmospheric pressure ionization (API), 318 Background electrolyte, 372 Band, definition of, 135 Biological applications, 433-443 Bleeding, column, 168 Bonded phase chromatography (BPC), 198-214 advantages and disadvantages of, 206 ligates attached, 202 preparation of, 200-202 Buffeds): in HPLC, 255-256 in electrophoresis, 373 Calibration, 17, 299-300 Capacity factor, see Retention factor

Capillary columns, 143, 153-157 Capillary electrochromatography (CEC), 380-383 Capillary electrophoresis (CE), 372-380 capillary zone electrophoresis (CZE), 376-380 chiral separation, 378 classification of techniques, 376 micellar electrokinetic chromatography (MEKC), 380 sample introduction, 377 Carbon dioxide phase diagram, 125, 127, 128 Carbowax", 169 Carrier gases, 148-150 effect on efficiency in OT columns, 149 CDER, division of FDA, see Center for Drug Evaluation and Research Center for Drug Evaluation and Research (CDER), 7 , 8 Centrifugal partition chromatography, 454 CGMP, see Good Manufacturing Practice (GMP) Charge-transfer, 100 Chemical ionization MS, 313 Chemisorption, 96 Chiral mobile phases, 453 Chiral separations, 378, 443-454 Chiral stationary phases (CPSs), 446-453 Chromatogram, 45 Chromatographic Data Systems (CDS), 296 Chromatographic symbols, 45-51 Chromatography: classification of, 41-43, 144, 184 definition of, 39-63 history of, 37-39 Chromatographic terms, symbols, and equations, 46, 63

Chromatography: Concepts and Contrasts, Second Edition. By James M. Miller ISBN 0-471-47207-7 0 2005 John Wiley & Sons, Inc.

485

486

INDEX

CITAC, 13, 24 Clathrate, 101 Clausius-Clapeyron equation, 171 Code of Federal Regulations (CFR), 9, I1 Part 11, 30 Column: microbore, 240 open tubular (OT), 143, 153-157 packed, 68-74 wide-bore, 157 Column selectivity, see Separation factor Comparisons: CEC and CZE, 382 gases and liquids, 118-119 GC and LC, 117-124 linear flow and electroosmotic flow. 375 peak height and peak area, 296-297 peaks and bands, 135-136 retardation factors, RF and R, 136-139 Compressibility factor, 120-121 Corrected retention volume (time), 120 Countercurrent chromatography, 454-455 Craig countercurrent extraction apparatus, 397 point data, 126 Critical micellar concentration, 379 Cryogenic GC, 176 Current Good Manufacturing Practice (CGMP). See Good Manufacturing Practice (GMP) Curves: calibration, 292, 295, 302, 304 vanDeemter, 82, 83, 86, 87, 89, 90 Cyclodextrin, 448 CZE, see capillary electrophoresis Data, acquisition of, 296-299 Data system(s): Chromatography (CDS), 296 LIMS, 296 Dead volume, see hold-up volume DEGS, 170 Densitometry, in TLC, 345 Derivatization, 412-417 Detection limitk), 27, 476. See also Detectivity Detectivity, 292, 293-294 Detectors, 278-296 cell volume of, 290-29 1 characteristics of, 285-296 classification of, 278-284 minimum detectability, see Limit of detection LOD GC, 157-165 HPLC, 243-248 linear range, 292, 294 noise, 285-287

responsc factors, relative, 301 selectivity 279, 283-284 signal, 291-295 timc constant of, 287-290 Dialysis, 41 1-412 Diastereomer separations, 445-446 Development of methods, 4, 248-250 Diffusion, 77, 370 Diffusion coefficient, 77 Dispersion forces, 98 Dispersivity, 90, 109, 135 Displacement chromatography, 43 Distribution constant, 47-49, 106, 392-393. Drift, 286 Dual detectors (dual channel GC), 359 Dwell volume and time, 257-258 Eddy diffusion, 76 Effective plate number, 54 Electrochemical detectors in LC, 247 Electron capture detector (ECD), 164- I65 Electron impact, 311 Electronic pressure control, IS0 Electrophoresis: capillary electrophoresis, 372-380 principles of, 366-371 zone electrophoresis, 371 -372 Elcctroosmosis and electroosmotic flow, 370, 373-37s Electrospray ionization, 316-318 Eluotropic series, 199 Elution, gradient, see Gradient elution Elution Chromatography, 43 End capping, 202 Environmental Protection Agency, EPA, 11, 14 Equations, summary table of, 63 Equivalent column(s) in HPLC, 207 Evaporative light scattering detector (ELSD), 247-248 Extended van Deemter equation, 80 External standard, 301 -303 Extra-column zone broadening, 242-243 Extraction, 390-41 1 Flame ionization detector (FID), 157-161 Flash chromatography, 265 Fast GC, 176, 177 Fast LC, 240 FDA, see Food and Drug Administration Federal Register, 9, 22 Field flow fractionation, 455 Flow, measurement of in GC, 121 Flow resistance parameter, 122, 268 Fluorescence detector, 247

INDEX

487

Food and Drug Administration (FDA), 2, 6- 11 FR, see Federal Register Frontal analysis, 44 Fronting, see Peak, shape of

Hydrophobic interaction chromatography (HIC), 209-210, 442-443 Hydrophilic interaction chromatography (HILIC), 261

CAMP, see Good automated manufacturing practice Gas compressibility factor, 119-121 Gas-liquid chromatography (GLC), 143- 144 Gas-solid chromatography (GSC) 100, 101, 143-144, 167 Gaussian distribution, see Normal distribution GC/FTIR, 327 GC/MS, 323 Gel filtration, see Size exclusion chromatography Gel permeation, see Sizc exclusion chromatography General Notices, 20 GLP,see Good Laboratory Practice GMP, see Good Manufacturing Practice Golay equation, 81 Good automated manufacturing practice (CAMP), 29 Good guidance practices (GGP), 10 Good laboratory practice (GLP), 7 Good Manufacturing Practice (GMP), 2, 7, 9 Gradient elution, 188-190, 199, 237, 256-258, 383 Guidance documcnts, 10 GXP, 7, 10

IC, see Ion chromatography ICH, see International Conference o n Harmonization, Ideal solution, 104 Inclusion compounds, 101 Induction forces, 98 Injection port, GC, 150, 165-166 Injection valve, 239 Installation qualification (IQ), 29 Instrument qualification, see Qualification of instruments Inorganic GC, 178 Integration of peaks, 297-299 Internal standard, 303, 323 Internal surface reversed phase (ISRP) LC, 211-212 International Conference on Harmonization (ICH), 7, 11, 21-22 glossary o f terms, 475-477 International Organization for Standardization, (ISO), 16 International Union of Pure and Applied Chemistry, (IUPAC), 13, 17 Internet web sites, 13, 34, 472-473 Inverse GC, 178 Ion chromatography, 220-225 I o n exchangc chromatography, 214-225 Ion interaction chromatography (IIC), 207-21 1 Ion pair chromatography, see Ion interaction chromatography (IIC) Ion sources, MS, 311-318 Ion trap analyzer, MS, 319 Ionic mobility, 367 ISO, see International Organization for Standardization Isocratic LC, definition of, 188 Isothermal chromatography, 173 Isotherms, sorption, 94 IUPAC. 13, 17 Nomenclature, 45

H or HETP, 55. See see Dispersivity and Plate height Headspace analysis, 410-41 1 Henrys Law, 102 High performance liquid chromatography, HPLC, 186-188 column diameter, 240-242 column efficiency, 268 column equivalency, 207 column evaluation, 268-269 detectors, 243-248 instrumentation, 234-249 pumps, 234-237 temperature control, 242 troubleshooting, 259-261 High performance TLC (HPTLC), 336-337 High prcssurc LC, 186 Hildebrand’s solubility parameter, 104-105 History of chromatography, 37, 141 Hold-up volume, 48 Homologous series plots, 145 HPLC, .see High performance LC Hydrogen bonding, 99

Keesom forces, 98 Kinetics of chromatography, 67, 75-88 Kovats retention index, 144-146 Knox equation plot, 89-90 Lab-on-a-chip, 455 LdbOratOry Information Management Systems (LIMS), 296 Langmuir isotherm, 95

488

INDEX

Large volume injection, GC, 178 LC/FTIR, 327 LC/MS, 324-326 LC/NMR, 327 Ligand exchange chromatography, 225 Limit of detection (LOD), 29, 293-294 Limit of quantitation (LOQ), 294 LIMS, see Laboratory information management systems Linearity of detectors, 294-295, 477 Linear dynamic range, 294-295, 477 Liquid chromatography (LC). See also High performance liquid chromatography (HPLC) low pressure, 185, 186 normal phase, 184, 194 planar, see thin layer chromatography (TLC) preparative, 262-266 reversed phase, 184, 248-258 Liquid-liquid extraction (LLE), 393-402 Liquid-liquid microextraction, 398 Liquid phase, see Stationary phase Liquid-solid chromatography (LSC), 194-198 Liquid-solid extraction, 402-404 London forces, 98 Longitudinal molecular diffusion, 77 Low pressure LC, 185, 186 Martin equation, 343 Mass spectroscopy (MS and MSD), 310-324 analyzers, 318-321 ion sources, 3 1 1-3 18 Mass transfer, 78-79 MDQ, see Limit of detection (LOD) Mcthod development, 249-259 computer methods, 258 Method translation, 176 Method validation, see Validation Micellar LC, 267 Micellar electrokinetic chromatography (MEKC), 379 Microbore columns in LC, 240 Microfluidic devices, 455 Microliter syringe, see Syringe pTAS, 455 Minimum detectable quantity (MDQ) or Minimum detection limit, see Limit of Detection Mobile phase: GC, see Carrier gas LC, 191-194. See also Buffeds) in HPLC Mobility, see Ionic mobility Molecular diffusion, see Diffusion Molecular imprint polymers (MIP), 407, 437 Molecular sieves, 100

Molecular weight determination by SEC, 232-234 Monolithic columns, 71-72, 191, 212-213 Multidimensional chromatography, 423-432 GC/GC, 425-427 LC/GC, 427-431 LC/LC, 431-432 TLC, 347 Multipath term in rate equation, 76 National Formulary, 19 National Institute of Standards and Technology (NIST), 5 NDA, see New drug application Net retention volume, 121 New drug application (NDA), 7 NIST, see National Institute of Standards and Technology Nitrogen phosphorous detector, 164 Noise (detector), 285-287 Nomenclature, see IUPAC nomenclature Non-retained solute (pcak), 47-48 Non-volatile samples, 468-471 Normal distribution, 51-52 Normal phase LC (NPLC), 184, 194 NTIS, 12 Number of theoretical plates, .see Plate number Open tubular (OT) columns, see Column, open tubular (OT) Operational qualification (OQ), 30 Optimization of separations, 112-1 15 Orientation forces, 98 Overlapping resolution mapping, 249-254 Packcd columns, 68-74 nature of, 73-74 PAGE, 371-372 Paper chromatography, 33 1-332 Particle beam, 314-315 Particle size, effect on efficiency, 87 Partition, 44 Partition coefficient, 105. See also Distribution constant Partition ratio, see Retention factor Peak: definition of, 45, 52 shape of, 55, 94 symmetry of, 55-57 Peak capacity, 60 Peak fronting, 55 Peak height vs. peak area, 296 Peak parking, 281 Peak tailing, 55 Peak width, 52, 297

INDEX

Pellicular supports, 191 Performance qualification, 30 kit for, 267 Permeability, 121- 122 Phase selection diagrams, 254-255 Phase volume ratio, 49, I55 Photodiode array, see Diode array detector, Pinkerton (ISRP) columns, 210-211 Plate hcight, 55, 90 Plate number, 53 Plate theory, 75, 398 Polarity index, 106-107 Pore size of LC supports, 68-70

LC, 262-267 TLC, 347 Pressure correction factor, 119- 121 Pressure equivalences, 118, 150 Programmed temperature G C (PTGC), 172- 176 Programmed temperature vaporization, 177 Proteins, peptides, and proteomics, 438-440 Pseudophase LC, 267 Pumps for LC, 234-237 Pyrolysis GC, 178 Quadrupole analyzer, MS, 319 Qualification of instruments, 4, 9, 29 Quality assurance/control, 3 Quantitative analysis, 299-306 Quantitation limit, definition of, 477 Quarter-pcak width, 52 Range, see Linear dynamic range Raoults Law, 142 Rate equations, comparisons of, 79-82, 84-88 Rate theory, 75-91 Ratiograms, 350-362 Reduced paramcters, 88-90, 134 Reference standards, 6, 17-19, 299-300 Refractive index (RI) detector, 246 Regular solution, 104 Relative response factors. See Response factors Relative retention, 355 Repeatability, see Precision Rcproducibility, see Precision, Research and development ( R & D ) , 3 Resolution, 58-59 in MS, 321 Response factors (detector), 301 Response time, 287

489

Restricted access media (RAM), 210, 410, 434 Retardation factor for columnar chromatography, R , 49-51 Retardation factor for planar chromatography, RF., 136-138 Retention gap, 157 Retention index of Kovats, 144-146 Retention factor, 49 Retention time, 47 Retention time locking, 176 Retention volume, 46-49 adjusted, 48 corrected, 120 net, 121 specific, 143 Reversed phase LC, 184, 249-259 Robustness, 295, 477 Rohrschneider/McReynolds constants. 146-147, 192 Ruggedness, 295-296 Sample size, maximum for preparative LC, 266 Sampling valve, 239 Selected ion monitoring (SIM), 322 Selectivity, detector, 279, 283-284. See also Separation factor Sensitivity (detector), 291 Separation, definition of, 39 Separation factor, 62, 355 Separation impedance, 268 Separation number, 61 Septum, 150, 151 Sieving, molecular sieves, 100. See also Size cxclusion chromatography Signal-to-noise ratio (S/N), 286 Significant temperature, 176 Silanization of solid supports, 73 Silica: surface of, 72 types of, 72, 201, 204 Silicone polymer stationary phases, 169-170 Simulated distillation,l79 Size exclusion chromatography, 100, 226-234 Snyder’s solvent parameter, 104-107, 192-195 Soap chromatography, 208, 267 Solid adsorbents (stationary phases), 16 Solid phase extraction (SPE), 405-407 Solid phase microextraction (SPME), 407-410 Solid support, 67 Solubility parameter of Hildebrand, 104, 105 Solvophobic theory, 203 Sorption, see Absorption; Adsorption Sorption isotherms, 94 Specificity, 475

490

INDEX

Split injection, 150-152 Splitless injection, 151-153 Splitter, inlet for GC, 150-152 SRM, see Reference standards Standard addition, 304-305 Standard deviation (quarter-zone width), 51-53 Standard operating procedure (SOP), 27 Standardization, 17-19 Standards, chemical, see Reference standards Stationary phase: configuration of, 67-74 in GC, 85, 166-167, 168-170 in LC, 87, 190. See also Bonded phases in TLC, 335-338 Statistical moments, 54 Stir-bar sorptive extraction, 410 Supercritical fluid chromatography (SFC), 37, 124- 134 Supercritical fluids, 124-129 Supercritical fluid extraction (SFE), 404-405 Surface deactivation, 72-73, 201 Syringe, 150, 151 System peaks, 246 System suitability (SS), 21, 28, 305 Tailing factor, 55-57 Tandem MS, 322 TCD, see Thermal conductivity detector Temperature, effect in GC, 170-176 Temperature programming, see Programmed temperature G C Theoretical plate, see Plate number Thermal conductivity detector (TCD), 161-163 Thermodynamics of zone migration, 111 Thermospray ionization, 315-316 Thin layer chromatography (TLC), 333-350 classification of, 343 instrumental, 34 1-350 manual, 333-341 OPLC, 344-345 Threshold, 298 Time, retention, see Retention time Time constant (detector), 287-290 Time of flight (TOF) analyzer, MS, 319

TLC. set Thin laycr chromatography Total ion chromatogram (TIC), 322 Trennzahl number, 61 Trifluoroacetate (TFA), 325 Troubleshooting, HPLC, 259-261 Troutons rule, 174 Tswett, 37 Two-column plots, 355-357 Ultraviolet absorption detector (UV), 244-246 Unified chromatography, 37, 125 United States Pharmacopeia (USP), 10, 13, 19-21 general notices, 20 reference standards, 20 relationship to the FDA, 10 Validation of methods, 4, 7, 9, 10, 18, 24-29, 295 Valve, sampling, 238-239 Van Deemter equation, 75-79 Van Deemter plots, 83 Van der Walls forces, 97-99 Velocity: average linear, of mobile phase, 49, 120 of solute, 49 reduced, see Reduced parameters Viscosity of water mixtures, 238 Void volume, see Hold-up volume Volatile samples, 468 Volume: hold-up, 48 retention, 47 VTR'AP methodology, 27 Wall-coated OT columns , GC, 85, 143, 144, 154-157, 166 Water. G C separation of, 167 Xylene isomers, separation of, 103 Zirconia bonded phases, 213 Zone, definition of, 135 Zwitterion, 368