Encyclopedia of Spectroscopy and Spectrometry (3-Volume Set with Online Version)

(QF\FORSHGLDRI6SHFWURVFRS\DQG6SHFWURPHWU\7KUHH9ROXPH6HW 9ROXPH

*HRUJH7UDQWHU (GLWRU *OD[R:HOOFRPH0HGLFLQHV5HVHDUFK%HFNHQKDP8. -RKQ+ROPHV (GLWRU 8QLYHUVLW\RI2WWDZD2QWDULR&DQDGD -RKQ/LQGRQ (GLWRULQ&KLHI ,PSHULDO&ROOHJHRI6FLHQFH7HFKQRORJ\DQG0HGLFLQH/RQGRQ 8.

)HDWXUHV %HQHILWV &RQWDLQVQHDUO\DUWLFOHVFRYHULQJWKHZKROHDUHDRI VSHFWURVFRS\DQGUHODWHGDUHDV ł3URYLGHVWKHHVVHQWLDOIDFWVDQGEDFNJURXQGIRUHDFKWRSLF ł,QFOXGHVUHDGLQJOLVWVLQHDFKDUWLFOHWRKHOSWKHUHDGHUDFFHVV PRUHGHWDLOHGLQIRUPDWLRQ ł$OORZVHDV\DFFHVVWRUHTXLUHGLQIRUPDWLRQWKURXJK DOSKDEHWLFDOOLVWLQJVH[WHQVLYHFURVVUHIHUHQFLQJWRUHODWHG DUWLFOHVDQGDGHWDLOHGVXEMHFWLQGH[LQHDFKYROXPH ł,QFOXGHVQXPHURXVILJXUHVDQGWDEOHVWKDWLOOXVWUDWHWKHWH[W ł&RQWDLQVFRORUSODWHVHFWLRQVLQHDFKYROXPH ł,QFOXGHVLQLWLDODFFHVVWRWKHRQOLQHYHUVLRQZLWKH[WHQVLYH K\SHUWH[WOLQNLQJDQGDGYDQFHGVHDUFKWRROV2QJRLQJDFFHVVLV PDLQWDLQHGIRUDPLQLPXPDQQXDOIHH ł

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

5HDGHUVKLS 3URIHVVLRQDO VSHFWURVFRSLVWVZRUNLQJ LQDFDGHPLDRULQGXVWU\ HJLQ SKDUPDFHXWLFDO FKHPLFDODQG HQJLQHHULQJLQGXVWULHV UHOHYDQWOLEUDULHVDQG LQGLYLGXDOUHVHDUFK JURXSV&'520 YHUVLRQHVSHFLDOO\ VKRXOGEHXVHIXOWR WHDFKHUV XQGHUJUDGXDWHVDVD VRXUFHRIFRXUVH PDWHULDODQGWR SRVWJUDGXDWHVIRU SUDFWLFDOWLSVDQGLGHDV IRUQHZZD\VRIWDFNOLQJ SUREOHPV ,6%1 %RRN+DUGEDFN 0HDVXUHPHQWV; PP 3DJHV ,PSULQW$FDGHPLF 3UHVV 3XEOLFDWLRQ'DWH -DQXDU\ 3ULFH £695.00

PREFACE vii

This encyclopedia provides, we believe, a comprehensive and up-to-date explanation of the most important spectroscopic and related techniques together with their applications. The Encyclopedia of Spectroscopy and Spectrometry is a cumbersome title but is necessary to avoid misleading readers who would comment that a simplified title such as the "Encyclopedia of Spectroscopy" was a misnomer because it included articles on subjects other than spectroscopy. Early in the planning stage, the editors realized that the boundaries of spectroscopy are blurred. Even the expanded title is not strictly accurate because we have also deliberately included other articles which broaden the content by being concerned with techniques which provide localized information and images. Consequently, we have tried to take a wider ranging view on what to include by thinking about the topics that a professional spectroscopist would conveniently expect to find in such a work as this. For example, many professionals use spectroscopic techniques, such as nuclear magnetic resonance, in conjunction with chromatographic separations and also make use of mass spectrometry as a key method for molecular structure determination. Thus, to have an encyclopedia of spectroscopy without mass spectrometry would leave a large gap. Therefore, mass spectrometry has been included. Likewise, the thought of excluding magnetic resonance imaging (MRI) seemed decidedly odd. The technique has much overlap with magnetic resonance spectroscopy, it uses very similar equipment and the experimental techniques and theory have much in common. Indeed, today, there are a number of experiments which produce multidimensional data sets of which one dimension might be spectroscopic and the others are image planes. Again the subject has been included. This led to the general principle that we should include a number of so-called spatially-resolved methods. Some of these, like MRI, are very closely allied to spectroscopy but others such as diffraction experiments or scanning probe microscopy are less so, but have features in common and are frequently used in close conjunction with spectroscopy. The more peripheral subjects have, by design, not been treated in the same level of detail as the core topics. We have tried to provide an overview of as many as possible techniques and applications which are allied to spectroscopy and spectrometry or are used in association with them. We have endeavoured to ensure that the core subjects have been treated in substantial depth. No doubt there are omissions and if the reader feels we got it wrong, the editors take the blame. The encyclopedia is organized conventionally in alphabetic order of the articles but we recognize that many readers would like to see articles grouped by spectroscopic area. We have achieved this by providing separate contents lists, one listing the articles in an intuitive alphabetical form, and the other grouping the articles within specialities such as mass spectrometry, atomic spectroscopy, magnetic resonance, etc. In addition each article is flagged as either a "Theory", "Methods and Instrumentation" or "Applications" article. However, inevitably, there will be some overlap of all of these categories in some articles. In order to emphasize the substantial overlap which exists among the spectroscopic and spectrometric approaches, a list has been included at the end of each article suggesting other articles in this encyclopedia which are related and which may provide relevant information for the reader. Each article also comes with a "Further Reading" section which provides a source of books and major reviews on the topic of the article and in some cases also provides details of seminal research papers. There are a number of colour plates in each volume as we consider that the use of colour can add greatly to the information content in many cases, for example for imaging studies. We have also included extensive Appendices of tables of useful reference data and a contact list of manufacturers of relevant equipment. We have attracted a wide range of authors for these articles and many are world recognized authorities in their fields. Some of the subjects covered are relatively static, and their articles provide a distillation of the established knowledge, whilst others are very fast moving areas and for these we have aimed at presenting up-to-date summaries. In addition, we have included a number of entries which are retrospective in nature, being historical reviews of particular types of spectroscopy. As with any work of this magnitude some of the articles which we desired and commissioned to include did not make it for various reasons. A selection of these will appear in a separate section in the on-line version of the encyclopedia, which will be available to all purchasers of the print version and will have extensive hypertext links and advanced search tools. In this print version there are 281 articles contributed by more than 500 authors from 24 countries. We have persuaded authors from Australia, Belgium, Canada, Denmark, Finland, France, Germany, Hungary, India,

viii PREFACE

Israel, Italy, Japan, Mexico, New Zealand, Norway, Peru, Russia, South Africa, Spain, Sweden, Switzerland, The Netherlands, the UK and the USA to contribute. The encyclopedia is aimed at a professional scientific readership, for both spectroscopists and non-spectroscopists. We intend that the articles provide authoritative information for experts within a field, enable spectroscopists working in one particular field to understand the scope and limitations of other spectroscopic areas and allow scientists who may not primarily be spectroscopists to grasp what the various techniques comprise in considering whether they would be applicable in their own research. In other words we tried to provide something for everone, but hope that in doing so, we have not made it too simple for the expert or too obscure for the non-specialist. We leave the reader to judge. John Lindon John Holmes George Tranter

Editor-in-Chief John C. Lindon, Biological Chemistry, Division of Biomedical Sciences, Imperial College of Science, Technology and Medicine, Sir Alexander Fleming Building, South Kensington, London SW7 2AZ UK

Editors George E. Tranter, Glaxo Wellcome Medicines Research, Physical Sciences Research Unit, Gunnells Wood Road, Stevenage, Hertfordshire SG1 2NY, UK John L. Holmes, University of Ottawa, Department of Chemistry, PO Box 450, Stn 4, Ottawa, Canada KIN 6N5

Editorial Advisory Board Laurence D. Barron, Department of Chemistry, University of Glasgow, Glasgow G12 8QQ, UK Andries P. Bruins, University Centre for Pharmacy, State University, A Deusinglaan 1, Groningen 9713 AV, Netherlands C.L. Chakrabarti, Chemistry Department, Carlton University, Ottawa, Ontario K1S 5B6, Canada J. Corset, Centre National de la Recherche Scientifique, Laboratoire de Spectrochimie Infrarouge et Raman, 2 Rue Henri-Dunant, 94320 Thiais, France David J. Craik, Centre for Drug Design & Development, University of Queensland, Brisbane 4072, Queensland, Australia James W. Emsley, Department of Chemistry, University of Southampton, Highfield, Southampton SO17 1BJ UK A.S. Gilbert, 19 West Oak, Beckenham, Kent BR3 5EZ, UK P.J. Hendra, Department of Chemistry, University of Southampton, Highfield, Southampton SO9 5NH, UK James A. Holcombe, Department of Chemistry, University of Texas, Austin, Texas 7871-1167, USA Harry Kroto, Department of Chemistry, University of Sussex, Falmer, East Sussex BN1 9QJ, UK Reiko Kuroda, Department of Life Sciences, Graduate School of Arts and Science, The University of Tokyo, Komaba, Tokyo 153, Japan N.M.M. Nibbering, Institute of Mass Spectrometry, University of Amsterdam, Nieuwe Achtergracht 129, 1018 WS Amsterdam, The Netherlands Ian C.P. Smith, National Research Council of Canada, Institute of Biodiagnostics, Winnipeg, Manitoba MB R3B 1Y6, Canada

S.J.B. Tendler, Department of Pharmaceutical Sciences, University of Nottingham, University Park, Notttingham NG7 2RD, UK Georges H. Wagnie" re, Physikalisch-Chemisches Institut, der Universitat Winterhurerstrasse 190 CH-8057 . Zarich, . Zarich, Switzerland . D.J. Watkin, Chemical Crystallography Laboratory, University of Oxford, 9 Parks Road, Oxford OX1 3PD, UK

ACKNOWLEDGEMENTS

Without a whole host of dedicated people, this encyclopedia would never have come to completion. In these few words I, on behalf of my co-editors, can hope to mention the contributions of only some of those hard working individuals. Without the active co-operation of the hundreds of scientists who acted as authors for the articles, this encyclopedia would not have been born. We are very grateful to them for endeavouring to write material suitable for an encyclopedia rather than a research paper, which has produced such high-quality entries. We know that all of the people who contributed articles are very busy scientists, many being leaders in their fields, and we thank them. We, as editors, have been ably supported by the members of the Editorial Advisory Board. They made many valuable suggestions for content and authorship in the early planning stages and provided a strong first line of scientific review after the completed articles were received. This encyclopedia covers such a wide range of scientific topics and types of technology that the very varied expertise of the Editorial Advisory Board was particularly necessary. Next, this work would not have been possible without the vision of Carey Chapman at Academic Press who approached me about 4 years ago with the excellent idea for such an encyclopedia. Four years later, am I still so sure of the usefulness of the encyclopedia? Of course I am, despite the hard work and I am further bolstered by the thought that I might not ever have to see another e-mail from Academic Press. For their work during the commissioning stage and for handling the receipt of manuscripts and dealing with all the authorship problems, we are truly indebted to Lorraine Parry, Colin McNeil and Laura O'Neill who never failed to be considerate, courteous and helpful even under the strongest pressure. I suspect that they are now probably quite expert in spectroscopy. In addition we need to thank Sutapas Bhattacharya who oversaw the project through the production stages and we acknowledge the hard work put in by the copy-editors, the picture researcher and all the other production staff coping with very tight deadlines. Finally, on a personal note, I should like to acknowledge the close co-operation I have received from my co-editors George Tranter and John Holmes. I think that we made a good team, even if I say it myself. John Lindon Imperial College of Science, Technology and Medicine London 22 April 1999

Article Titles

Authors, Pages

A Art Works Studied Using IR and Raman Spectroscopy

Howell G M Edwards, Pages 2-17

Atmospheric Pressure Ionization in Mass Spectrometry

W. M. A. Niessen, Pages 18-24

Atomic Absorption, Methods and Instrumentation Atomic Absorption, Theory Atomic Emission, Methods and Instrumentation Atomic Fluorescence, Methods and Instrumentation Atomic Spectroscopy, Historical Perspective ATR and Reflectance IR Spectroscopy, Applications

Steve J Hill and Andy S Fisher, Pages 24-32 Albert Kh Gilmutdinov, Pages 33-42 Sandra L Bonchin, Grace K Zoorob and Joseph A Caruso, Pages 42-50 Steve J Hill and Andy S Fisher, Pages 50-55 C L Chakrabarti, Pages 56-58 U P Fringeli, Pages 58-75

B Biochemical Applications of Fluorescence Spectroscopy Biochemical Applications of Mass Spectrometry Biochemical Applications of Raman Spectroscopy Biofluids Studied By NMR Biomacromolecular Applications of Circular Dichroism and ORD Biomacromolecular Applications of UVVisible Absorption Spectroscopy Biomedical Applications of Atomic Spectroscopy

Jason B Shear, Pages 77-84 Victor E Vandell and Patrick A Limbach, Pages 84-87 Peter Hildebrandt and Sophie Lecomte, Pages 88-97 John C Lindon and Jeremy K Nicholson, Pages 98-116 Norma J Greenfield, Pages 117-130 Alison Rodger and Karen Sanders, Pages 130-139 Andrew Taylor, Pages 139-147

C 13

C NMR, Methods

C NMR, Parameter Survey

Calibration and Reference Systems (Regulatory Authorities) Carbohydrates Studied By NMR Cells Studied By NMR Chemical Applications of EPR Chemical Exchange Effects in NMR

Cecil Dybowski, Alicia Glatfelter and H N Cheng, Pages 149-158 R Duncan Farrant and John C Lindon, Pages 159-165 C Burgess, Pages 166-171 Charles T Weller, Pages 172-180 Fátima Cruz and Sebastián Cerdán, Pages 180-189 Christopher C Rowlands and Damien M Murphy, Pages 190-198 Alex D Bain, Pages 198-207

Chemical Ionization in Mass Spectrometry

Alex G Harrison, Pages 207-215

Chemical Reactions Studied By Electronic Spectroscopy

Salman R Salman, Pages 216-222

Chemical Shift and Relaxation Reagents in NMR

Silvio Aime, Mauro Botta, Mauro Fasano and Enzo Terreno, Pages 223-231

Chemical Structure Information from Mass Spectrometry

Kurt Varmuza, Pages 232-243

Chiroptical Spectroscopy, Emission Theory

James P Riehl, Pages 243-249

Chiroptical Spectroscopy, General Theory Chiroptical Spectroscopy, Oriented Molecules and Anisotropic Systems

Hans-Georg Kuball, Tatiana Höfer and Stefan Kiesewalter, Pages 250-266 Hans-Georg Kuball and Tatiana Höfer, Pages 267-281

Chromatography-IR, Applications

George Jalsovszky, Pages 282-287

Chromatography-IR, Methods and Instrumentation

Robert L White, Pages 288-293

Chromatography-MS, Methods

W W A Niessen, Pages 293-300

Chromatography-NMR, Applications CIDNP Applications

J P Shockcor, Pages 301-310 Tatyana V Leshina, Alexander I Kruppa and Marc B Taraban, Pages 311-318

Circularly Polarized Luminescence and Fluorescence Detected Circular Dichroism Cluster Ions Measured Using Mass Spectrometry Colorimetry, Theory Computational Methods and Chemometrics in Near-IR Spectroscopy Contrast Mechanisms in MRI Cosmochemical Applications Using Mass Spectrometry

Christine L Maupin and James P Riehl, Pages 319-326 O Echt and T D Märk, Pages 327-336 Alison Gilchrist and Jim Nobbs, Pages 337-343 Paul Geladi and Eigil Dåbakk, Pages 343-349 I R Young, Pages 349-358 J R De Laeter, Pages 359-367

D Diffusion Studied Using NMR Spectroscopy

Peter Stilbs, Pages 369-375

Drug Metabolism Studied Using NMR Spectroscopy

Myriam Malet-Martino and Robert Martino, Pages 375-388

Dyes and Indicators, Use of UV-Visible Absorption Spectroscopy

Volker Buss and Lutz Eggers, Pages 388-396

E Electromagnetic Radiation

David L Andrews, Pages 397-401

Electronic Components, Applications of Atomic Spectroscopy

John C Lindon, Pages 401-402

Ellipsometry

G E Jellison, Jr, Pages 402-411

Enantiomeric Purity Studied Using NMR Environmental and Agricultural Applications of Atomic Spectroscopy Environmental Applications of Electronic Spectroscopy EPR Imaging

Thomas J Wenzel, Pages 411-421 Michael Thompson and Michael H Ramsey, Pages 422-429 John W Farley, William C Brumley and DeLyle Eastwood, Pages 430-437 L H Sutcliffe, Pages 437-445

EPR Spectroscopy, Theory EPR, Methods Exciton Coupling

Christopher C Rowlands and Damien M Murphy, Pages 445-456 Richard Cammack, Pages 457-469 Nina Berova, Nobuyuki Harada and Koji Nakanishi, Pages 470-488

F 19

F NMR, Applications, Solution State

Far-IR Spectroscopy, Applications Fast Atom Bombardment Ionization in Mass Spectrometry Fibre Optic Probes in Optical Spectroscopy, Clinical Applications Fibres and Films Studied Using X-Ray Diffraction Field Ionization Kinetics in Mass Spectrometry Flame and Temperature Measurement Using Vibrational Spectroscopy Fluorescence and Emission Spectroscopy, Theory Fluorescence Microscopy, Applications Fluorescence Polarization and Anisotropy Fluorescent Molecular Probes Food and Dairy Products, Applications of Atomic Spectroscopy

Claudio Pettinari and Giovanni Rafaiani, Pages 489-498 James R Durig, Pages 498-504 Magda Claeys and Jan Claereboudt, Pages 505-512 Urs Utzinger and Rebecca R Richards-Kortum, Pages 512-528 Watson Fuller and Arumugam Mahendrasingam, Pages 529-539 Nico M M Nibbering, Pages 539-548 Kevin L McNesby, Pages 548-559 James A Holcombe, Pages 560-565 Fred Rost, Pages 565-570 G E Tranter, Pages 571-573 F Braut-Boucher and M Aubery, Pages 573-582 N J Miller-Ihli and S A Baker, Pages 583-592

Food Science, Applications of Mass Spectrometry

John P G Wilkins, Pages 592-593

Food Science, Applications of NMR Spectroscopy

Brian Hills, Pages 593-601

Forensic Science, Applications of Atomic Spectroscopy

John C Lindon, Pages 602-603

Forensic Science, Applications of IR Spectroscopy Forensic Science, Applications of Mass Spectrometry

Núria Ferrer, Pages 603-615 Rodger L Foltz, Dennis J Crouch and David M Andrenyak, Pages 615-621

Forestry and Wood Products, Applications of Atomic Spectroscopy

Cathy Hayes, Pages 621-631

Fourier Transformation and Sampling Theory

Raúl Curbelo, Pages 632-636

Fragmentation in Mass Spectrometry

Hans-Friedrich Grützmacher, Pages 637-648

FT-Raman Spectroscopy, Applications

R H Brody, E A Carter, H. G. M. Edwards and A M Pollard, Pages 649-657

G Gas Phase Applications of NMR Spectroscopy Geology and Mineralogy, Applications of Atomic Spectroscopy Glow Discharge Mass Spectrometry, Methods

Nancy S True, Pages 660-667 John C Lindon, Page 668 Annemie Bogaerts, Pages 669-676

H Halogen NMR Spectroscopy (Excluding F)

Heteronuclear NMR Applications (As, Sb, Bi) Heteronuclear NMR Applications (B, Al, Ga, In, Tl)

Frank G Riddell, Pages 677-684 Claudio Pettinari, Fabio Marchetti and Giovanni Rafaiani, Pages 685-690 Janusz Lewiski, Pages 691-703

Heteronuclear NMR Applications (Ge, Sn, Pb)

Claudio Pettinari, Pages 704-717

Heteronuclear NMR Applications (La–Hg)

Trevor G Appleton, Pages 718-722

Heteronuclear NMR Applications (O, S, Se and Te)

Ioannis P Gerothanassis, Pages 722-729

Heteronuclear NMR Applications (Sc–Zn)

Dieter Rehder, Pages 731-740

Heteronuclear NMR Applications (Y–Cd) High Energy Ion Beam Analysis High Pressure Studies Using NMR Spectroscopy

Erkki Kolehmainen, Pages 740-750 Geoff W Grime, Pages 750-760 Jiri Jonas, Pages 760-771

High Resolution Electron Energy Loss Spectroscopy, Applications

Horst Conrad and Martin E Kordesch, Pages 772-783

High Resolution IR Spectroscopy (Gas Phase) Instrumentation

Jyrki K Kauppinen and Jari O Partanen, Pages 784-794

High Resolution IR Spectroscopy (Gas Phase), Applications

E Canè and A Trombetti, Pages 794-801

High Resolution Solid State NMR, 13C

Etsuko Katoh and Isao Ando, Pages 802-813

High Resolution Solid State NMR, 1H, 19F

Anne S Ulrich, Pages 813-825

Hole Burning Spectroscopy, Methods

Josef Friedrich, Pages 826-836

Hydrogen Bonding and Other Physicochemical Interactions Studied By IR and Raman Spectroscopy Hyphenated Techniques, Applications of in Mass Spectrometry

A S Gilbert, Pages 837-843 W M A Niessen, Pages 843-849

I In Vivo NMR, Applications, 31P In Vivo NMR, Applications, Other Nuclei In Vivo NMR, Methods Induced Circular Dichroism Inductively Coupled Plasma Mass Spectrometry, Methods Industrial Applications of IR and Raman Spectroscopy Inelastic Neutron Scattering, Applications

Ruth M Dixon and Peter Styles, Pages 851-857 Jimmy D Bell, E Louise Thomas and K Kumar Changani, Pages 857-865 John C Lindon, Pages 866-868 Kymberley Vickery and Bengt Nordén, Pages 869-874 Diane Beauchemin, Pages 875-880 A S Gilbert and R W Lancaster, Pages 881-893 Stewart F Parker, Pages 894-905

Inelastic Neutron Scattering, Instrumentation Inorganic Chemistry, Applications of Mass Spectrometry

Stewart F Parker, Pages 905-915 Lev N Sidorov, Pages 915-923

Inorganic Compounds and Minerals Studied Using X-ray Diffraction

Gilberto Artioli, Pages 924-933

Inorganic Condensed Matter, Applications of Luminescence Spectroscopy

Keith Holliday, Pages 933-943

Interstellar Molecules, Spectroscopy of

A G G M Tielens, Pages 943-953

Ion Collision Theory

Anil K Shukla and Jean H Futrell, Pages 954-963

Ion Dissociation Kinetics, Mass Spectrometry

Bernard Leyh, Pages 963-971

Ion Energetics in Mass Spectrometry

John Holmes, Pages 971-976

Ion Imaging Using Mass Spectrometry Ion Molecule Reactions in Mass Spectrometry

Albert J R Heck, Pages 976-983 Diethard K Böhme, Pages 984-990

Ion Structures in Mass Spectrometry

Peter C Burgers and Johan K Terlouw, Pages 990-1000

Ion Trap Mass Spectrometers

Raymond E March, Pages 1000-1009

Ionization Theory IR and Raman Spectroscopy of Inorganic, Coordination and Organometallic Compounds IR Spectral Group Frequencies of Organic Compounds

C Lifshitz and T D Märk, Pages 1010-1021 Claudio Pettinari and Carlo Santini, Pages 1021-1034 A S Gilbert, Pages 1035-1048

IR Spectrometers

R A Spragg, Pages 1048-1057

IR Spectroscopy Sample Preparation Methods

R A Spragg, Pages 1058-1066

IR Spectroscopy, Theory Isotope Ratio Studies Using Mass Spectrometry Isotopic Labelling in Mass Spectrometry

Derek Steele, Pages 1066-1071 Michael E Wieser and Willi A Brand, Pages 1072-1086 Thomas Hellman Morton, Pages 1086-1096

L Labelling Studies in Biochemistry Using NMR Laboratory Information Management Systems (LIMS) Laser Applications in Electronic Spectroscopy Laser Induced Optoacoustic Spectroscopy

Timothy R Fennell and Susan C J Sumner, Pages 1097-1105 David R McLaughlin and Antony J Williams, Pages 1105-1113 Wolfgang Demtröder, Pages 1113-1123 Thomas Gensch, Cristiano Viappiani and Silvia E Braslavsky, Pages 1124-1132

Laser Magnetic Resonance

A I Chichinin, Pages 1133-1140

Laser Spectroscopy Theory

Luc Van Vaeck and Freddy Adams, Pages 1141-1152

Laser Spectroscopy Theory

David L Andrews, Pages 1153-1158

Light Sources and Optics Linear Dichroism, Applications

R Magnusson, Pages 1158-1168 Erik W Thulstrup, Jacek Waluk and Jens Spanget-Larsen, Pages 1169-1175

Linear Dichroism, Instrumentation

Erik W Thulstrup, Jens Spanget-Larsen and Jacek Waluk, Pages 1176-1178

Liquid Crystals and Liquid Crystal Solutions Studied By NMR

Lucia Calucci and Carlo Alberto Veracini, Pages 1179-1186

Luminescence Theory

Mohammad A Omary and Howard H Patterson, Pages 1186-1207

M Macromolecule–Ligand Interactions Studied By NMR Magnetic Circular Dichroism, Theory Magnetic Field Gradients in HighResolution NMR Magnetic Resonance, Historical Perspective Mass Spectrometry, Historical Perspective

J Feeney, Pages 1209-1216 Laura A Andersson, Pages 1217-1224 Ralph E Hurd, Pages 1224-1232 J W Emsley and J Feeney, Pages 1232-1240 Allan Maccoll†, Pages 1241-1248

Materials Science Applications of X-ray Diffraction

Åke Kvick, Pages 1248-1257

Matrix Isolation Studies By IR and Raman Spectroscopies

Lester Andrews, Pages 1257-1261

Medical Applications of Mass Spectrometry

Orval A Mamer, Pages 1262-1271

Medical Science Applications of IR Membranes Studied By NMR Spectroscopy Metastable Ions Microwave and Radiowave Spectroscopy, Applications

Michael Jackson and Henry H Mantsch, Pages 1271-1281 A Watts and S J Opella, Pages 1281-1291 John L Holmes, Pages 1291-1297 G Wlodarczak, Pages 1297-1307

Microwave Spectrometers

Marlin D Harmony, Pages 1308-1314

Mössbauer Spectrometers

Guennadi N Belozerski, Pages 1315-1323

Mössbauer Spectroscopy, Applications

Guennadi N Belozerski, Pages 1324-1334

Mössbauer Spectroscopy, Theory

Guennadi N Belozerski, Pages 1335-1343

MRI Applications, Biological MRI Applications, Clinical MRI Applications, Clinical Flow Studies MRI Instrumentation MRI of Oil/Water in Rocks MRI Theory MRI Using Stray Fields MS-MS and MSn Multiphoton Excitation in Mass Spectrometry

David G Reid, Paul D Hockings and Paul G M Mullins, Pages 1344-1354 Martin O Leach, Pages 1354-1364 Y Berthezène, Pages 1365-1372 Paul D Hockings, John F Hare and David G Reid, Pages 1372-1380 Geneviève Guillot, Pages 1380-1387 Ian R Young, Pages 1388-1396 Edward W Randall, Pages 1396-1403 W. M. A. Niessen, Pages 1404-1410 Ulrich Boesl, Pages 1411-1424

Multiphoton Spectroscopy, Applications Multivariate Statistical Methods Muon Spin Resonance Spectroscopy, Applications

Michael N R Ashfold and Colin M Western, Pages 1424-1433 R L Somorjai, Pages 1433-1439 Ivan D Reid and Emil Roduner, Pages 1439-1450

N Near-IR Spectrometers Negative Ion Mass Spectrometry, Methods Neutralization–Reionization in Mass Spectrometry Neutron Diffraction, Instrumentation Neutron Diffraction, Theory Nitrogen NMR NMR Data Processing NMR in Anisotropic Systems, Theory NMR Microscopy NMR of Solids NMR Principles NMR Pulse Sequences NMR Relaxation Rates NMR Spectrometers NMR Spectroscopy of Alkali Metal Nuclei in Solution

R Anthony Shaw and Henry H Mantsch, Pages 1451-1461 Suresh Dua and John H Bowie, Pages 1461-1469 Chrys Wesdemiotis, Pages 1469-1479 A C Hannon, Pages 1479-1492 Alex C Hannon, Pages 1493-1503 G A Webb, Pages 1504-1514 Gareth A Morris, Pages 1514-1521 J W Emsley, Pages 1521-1527 Paul T Callaghan, Pages 1528-1537 Jacek Klinowski, Pages 1537-1544 P J Hore, Pages 1545-1553 William F Reynolds, Pages 1554-1567 Ronald Y Dong, Pages 1568-1575 John C Lindon, Pages 1576-1583 Frank G Riddell, Pages 1584-1593

Nonlinear Optical Properties

Georges H Wagnière and Stanisaw Wozniak, Pages 1594-1608

Nonlinear Raman Spectroscopy, Applications

W Kiefer, Pages 1609-1623

Nonlinear Raman Spectroscopy, Instruments

Peter C Chen, Pages 1624-1631

Nonlinear Raman Spectroscopy, Theory Nuclear Overhauser Effect

J Santos Gómez, Pages 1631-1642 Anil Kumar and R Christy Rani Grace, Pages 1643-1653

Nuclear Quadrupole Resonance, Applications

Oleg Kh Poleshchuk and Jolanta N Latosiska, Pages 1653-1662

Nuclear Quadrupole Resonance, Instrumentation

Taras N Rudakov, Pages 1663-1671

Nuclear Quadrupole Resonance, Theory Nucleic Acids and Nucleotides Studied Using Mass Spectrometry Nucleic Acids Studied Using NMR

Janez Seliger, Pages 1672-1680 Tracey A Simmons, Kari B Green-Church and Patrick A Limbach, Pages 1681-1688 John C Lindon, Pages 1688-1689

O Optical Frequency Conversion Optical Spectroscopy, Linear Polarization Theory

Christos Flytzanis, Pages 1691-1701 Josef Michl, Pages 1701-1712

ORD and Polarimetry Instruments

Harry G Brittain, Pages 1712-1718

Organic Chemistry Applications of Fluorescence Spectroscopy

Stephen G Schulman, Qiao Qing Di and John Juchum, Pages 1718-1725

Organometallics Studied Using Mass Spectrometry

Dmitri V Zagorevskii, Pages 1726-1733

P 31

P NMR

David G Gorenstein and Bruce A Luxon, Pages 1735-1744

Parameters in NMR Spectroscopy, Theory of

G A Webb, Pages 1745-1753

Peptides and Proteins Studied Using Mass Spectrometry

Michael A Baldwin, Pages 1753-1763

Perfused Organs Studied Using NMR Spectroscopy

John C Docherty, Pages 1763-1770

PET, Methods and Instrumentation

T J Spinks, Pages 1771-1782

PET, Theory

T J Spinks, Pages 1782-1791

Pharmaceutical Applications of Atomic Spectroscopy

Nancy S Lewen and Martha M Schenkenberger, Pages 1791-1800

Photoacoustic Spectroscopy, Applications

Markus W Sigrist, Pages 1800-1809

Photoacoustic Spectroscopy, Methods and Instrumentation

Markus W Sigrist, Pages 1810-1814

Photoacoustic Spectroscopy, Theory Photoelectron Spectrometers Photoelectron Spectroscopy Photoelectron–Photoion Coincidence Methods in Mass Spectrometry (PEPICO)

András Miklós, Stefan Schäfer and Peter Hess, Pages 1815-1822 László Szepes and György Tarczay, Pages 1822-1830 John Holmes, Page 1831 Tomas Baer, Pages 1831-1839

Photoionization and Photodissociation Methods in Mass Spectrometry

John C Traeger, Pages 1840-1847

Plasma Desorption Ionization in Mass Spectrometry

Ronald D Macfarlane, Pages 1848-1857

Polymer Applications of IR and Raman Spectroscopy

C M Snively and J L Koenig, Pages 1858-1864

Powder X-Ray Diffraction, Applications

Daniel Louër, Pages 1865-1875

Product Operator Formalism in NMR Proteins Studied Using NMR Spectroscopy

Timothy J Norwood, Pages 1875-1884 Paul N Sanderson, Pages 1885-1893

Proton Affinities Proton Microprobe (Method and Background) Pyrolysis Mass Spectrometry, Methods

Edward P L Hunter and Sharon G Lias, Pages 1893-1901 Geoff W Grime, Pages 1901-1905 Jacek P Dworzanski and Henk L C Meuzelaar, Pages 1906-1919

Q Quadrupoles, Use of in Mass Spectrometry Quantitative Analysis

P H Dawson and D J Douglas, Pages 1921-1930 T Frost, Pages 1931-1936

R Radiofrequency Field Gradients in NMR, Theory Raman and Infrared Microspectroscopy Raman Optical Activity, Applications Raman Optical Activity, Spectrometers Raman Optical Activity, Theory Raman Spectrometers Rayleigh Scattering and Raman Spectroscopy, Theory Relaxometers

Daniel Canet, Pages 1937-1944 Pina Colarusso, Linda H Kidder, Ira W Levin and E Neil Lewis, Pages 1945-1954 Günter Georg Hoffmann, Pages 1955-1965 Werner Hug, Pages 1966-1976 Laurence A Nafie, Pages 1976-1985 Bernhard Schrader, Pages 1986-1992 David L Andrews, Pages 1993-2000 Ralf-Oliver Seitter and Rainer Kimmich, Pages 2000-2008

Rigid Solids Studied Using MRI

David G Cory, Pages 2009-2017

Rotational Spectroscopy, Theory

Iain R McNab, Pages 2017-2028

S Scanning Probe Microscopes Scanning Probe Microscopy, Applications Scanning Probe Microscopy, Theory Scattering and Particle Sizing, Applications Scattering Theory Sector Mass Spectrometers 29

Si NMR

SIFT Applications in Mass Spectrometry Small Molecule Applications of X-Ray Diffraction Solid State NMR, Methods Solid-State NMR Using Quadrupolar Nuclei Solid-State NMR, Rotational Resonance Solvent Suppression Methods in NMR Spectroscopy Sonically Induced NMR Methods

J G Kushmerick and P S Weiss, Pages 2043-2051 C J Roberts, M C Davies, S J B Tendler and P M Williams, Pages 2051-2059 A J Fisher, Pages 2060-2066 F Ross Hallett, Pages 2067-2074 Michael Kotlarchyk, Pages 2074-2084 R Bateman, Pages 2085-2092 Heinrich C Marsmann, Pages 2031-2042 David Smith and Patrik panl, Pages 2092-2105 Andrei S Batsanov, Pages 2106-2115 J W Zwanziger and H W Spiess, Pages 2128-2136 Alejandro C Olivieri, Pages 2116-2127 David L Bryce and Roderick E Wasylishen, Pages 2136-2144 Maili Liu and Xi-an Mao, Pages 2145-2152 John Homer, Pages 2152-2159

SPECT, Methods and Instrumentation

John C Lindon, Pages 2159-2161

Spectroelectrochemistry, Applications

R J Mortimer, Pages 2161-2174

Spectroelectrochemistry, Methods and Instrumentation

Roger J Mortimer, Pages 2174-2181

Spectroscopy of Ions Spin Trapping and Spin Labelling Studied Using EPR Spectroscopy

John P Maier, Pages 2182-2189 Carmen M Arroyo, Pages 2189-2198

Stars, Spectroscopy of Statistical Theory of Mass Spectra Stereochemistry Studied Using Mass Spectrometry Structural Chemistry Using NMR Spectroscopy, Inorganic Molecules

A G G M Tielens, Pages 2199-2204 J C Lorquet, Pages 2204-2211 Asher Mandelbaum, Pages 2211-2223 G E Hawkes, Pages 2224-2233

Structural Chemistry Using NMR Spectroscopy, Organic Molecules

Cynthia K McClure, Pages 2234-2245

Structural Chemistry Using NMR Spectroscopy, Peptides

Martin Huenges and Horst Kessler, Pages 2246-2260

Structural Chemistry Using NMR Spectroscopy, Pharmaceuticals

Alexandros Makriyannis and Spiro Pavlopoulos, Pages 2261-2271

Structure Refinement (Solid State Diffraction)

Dieter Schwarzenbach and Howard D Flack, Pages 2271-2278

Surface Induced Dissociation in Mass Spectrometry Surface Plasmon Resonance, Applications Surface Plasmon Resonance, Instrumentation

S A Miller and S L Bernasek, Pages 2279-2294 Zdzislaw Salamon and Gordon Tollin, Pages 2294-2302 R P H Kooyman, Pages 2302-2310

Surface Plasmon Resonance, Theory

Zdzislaw Salamon and Gordon Tollin, Pages 2311-2319

Surface Studies By IR Spectroscopy

Norman Sheppard, Pages 2320-2328

Surface-Enhanced Raman Scattering (SERS), Applications

W E Smith and C Rodger, Pages 2329-2334

Symmetry in Spectroscopy, Effects of

S F A Kettle, Pages 2335-2339

T Tensor Representations Thermospray Ionization in Mass Spectrometry Time of Flight Mass Spectrometers

Peter Herzig and Rainer Dirl, Pages 2342-2353 W M A Niessen, Pages 2353-2360 K G Standing and W Ens, Pages 2360-2365

Tritium NMR, Applications Two-Dimensional NMR, Methods

John R Jones, Pages 2366-2369 Peter L Rinaldi, Pages 2370-2381

U UV-Visible Absorption and Fluorescence Spectrometers

G E Tranter, Pages 2383-2389

V Vibrational CD Spectrometers Vibrational CD, Applications Vibrational CD, Theory Vibrational, Rotational and Raman Spectroscopy, Historical Perspective

Laurence A Nafie, Pages 2391-2402 Günter Georg Hoffmann, Pages 2403-2414 Philip J Stephens, Pages 2415-2421 A S Gilbert, Pages 2422-2432

X X-ray Absorption Spectrometers

Grant Bunker, Pages 2447-2453

X-ray Emission Spectroscopy, Applications

George N Dolenko, Oleg Kh Poleshchuk and Jolanta N Latoiska, Pages 2455-2462

X-ray Emission Spectroscopy, Methods

George N Dolenko, Oleg Kh Poleshchuk and Jolanta N Latoiska, Pages 2463-2467

X-ray Fluorescence Spectrometers

Utz Kramar, Pages 2467-2477

X-ray Fluorescence Spectroscopy, Applications

Christina Streli, P Wobrauschek and P Kregsamer, Pages 2478-2487

X-ray Spectroscopy, Theory Xenon NMR Spectroscopy

Prasad A Naik, Pages 2487-2498 Jukka Jokisaari, Pages 2435-2446

Z Zeeman and Stark Methods in Spectroscopy, Applications

Ichita Endo and Masataka Linuma, Pages 2501-2504

Zeeman and Stark Methods in Spectroscopy, Instrumentation

Ichita Endo and Masataka Linuma, Pages 2505-2509

Zero Kinetic Energy Photoelectron Spectroscopy, Applications

K Müller-Dethlefs and Mark Ford, Pages 2509-2519

Zero Kinetic Energy Photoelectron Spectroscopy, Theory

K Müller-Dethlefs and Mark Ford, Pages 2519-2526

APPENDICES Appendix 1. Periodic Table of Elements Appendix 2. Tables of SI and Related Units Appendix 3. Wavelength Scale Appendix 4. Colour, Wave Length, Frequency, Wave Number and Energy of Light Appendix 5. Magnetic Susceptibilities at 25°C

Page 2528 Pages 2529-2530 Page 2531 Page 2532 Page 2532

Appendix 6. Electronic Configuration of Elements

Pages 2533-2534

Appendix 7. Properties of some Important Solvents

Pages 2535-2536

Appendix 8. Important Acronyms in Organic Chemistry

Pages 2537-2538

Appendix 9. Equilibrium Constants at 25°c/concentration Units for Solutions

Page 2539

Appendix 10. Acronyms and Abbreviations in Quantum Chemistry and Related Fields

Page 2540

Appendix 11. Standard Potentials in Aqueous Solutions

Pages 2541-2544

Appendix 12. Typical UV Absorptions of Unconjugated Chromophores

Page 2545

Appendix 13. Typical UV Absorption Maxima of Substituted Benzenes

Page 2546

Appendix 14. Typical UV Absorption Maxima of Aromatic and Heteroaromatic Compounds Appendix 15. Common Isotopes for Mössbauer Spectroscopy Appendix 16. NMR Frequency Table

Page 2546 Page 2547 Pages 2548-2551

Appendix 17. 19F and 31P NMR Chemical Shifts

Page 2552

Appendix 18. Chemical Shift Ranges and Standards for Selected Nuclei

Pages 2552-2553

Appendix 19. Abbreviations and Acronyms used in Magnetic Resonance

Pages 2553-2556

Appendix 20. Symbols Used in Magnetic Resonance

Pages 2556-2557

Appendix 21. EPR/ENDOR Frequency Table

Pages 2557-2560

Appendix 22. Some Useful Conversion Factors in EPR

Page 2560

Appendix 23. Mass Spectrometry: Atomic Weights. Appendix 24. Conversion Table of Transmittance and Absorbanceunits

Pages 2561-2563 Page 2564

Appendix 25. Conversion Table of Energy and Wavelength Units

Pages 2565-2566

Appendix 26. Optical Components used in FT-IR-Spectroscopy

Page 2567

Appendix 27. Infrared and Raman Tables

Pages 2568-2571

Appendix 28. Selected Force Constants and Bond Orders (According To Siebert) of Organic and Inorganic Compounds Appendix 29. Fundamental Physical Constants Appendix 30. List Of Suppliers

Pages 2572-2573 Page 2574 Pages 2575-2581

Subject Classification

Atomic Spectroscopy Historical Overview Atomic Spectroscopy, Historical Perspective

C L Chakrabarti

Pages 56-58

Theory Atomic Absorption, Theory Fluorescence and Emission Spectroscopy, Theory

Albert Kh Gilmutdinov

Pages 33-42

James A Holcombe

Pages 560-565

Methods and Instrumentation Atomic Absorption, Methods and Instrumentation

Steve J Hill and Andy S Fisher

Sandra L Bonchin, Atomic Emission, Methods Grace K Zoorob and and Instrumentation Joseph A Caruso Atomic Fluorescence, Steve J Hill and Andy S Methods and Fisher Instrumentation

Pages 24-32

Pages 42-50

Pages 50-55

Applications Biomedical Applications of Atomic Spectroscopy Electronic Components, Applications of Atomic Spectroscopy Environmental and Agricultural Applications of Atomic Spectroscopy

Andrew Taylor

Pages 139-147

John C Lindon

Pages 401-402

Michael Thompson and Michael H Ramsey

Pages 422-429

Food and Dairy Products, Applications of Atomic Spectroscopy Forensic Science, Applications of Atomic Spectroscopy Forestry and Wood Products, Applications of Atomic Spectroscopy Geology and Mineralogy, Applications of Atomic Spectroscopy Pharmaceutical Applications of Atomic Spectroscopy

N J Miller-Ihli and S A Baker

Pages 583-592

John C Lindon

Pages 602-603

Cathy Hayes

Pages 621-631

John C Lindon

Page 668

Nancy S Lewen and Martha M Schenkenberger

Pages 1791-1800

Electronic Spectroscopy Theory Chiroptical Spectroscopy, Emission Theory

James P Riehl

Pages 243-249

Chiroptical Spectroscopy, General Theory

Hans-Georg Kuball, Tatiana Höfer and Stefan Kiesewalter

Pages 250-266

Chiroptical Spectroscopy, Oriented Molecules and Anisotropic Systems

Hans-Georg Kuball and Tatiana Höfer

Pages 267-281

Colorimetry, Theory

Alison Gilchrist and Jim Nobbs

Pages 337-343

G E Tranter

Pages 571-573

David L Andrews

Pages 1153-1158

Mohammad A Omary and Howard H Patterson

Pages 1186-1207

Fluorescence Polarization and Anisotropy Laser Spectroscopy Theory Luminescence Theory

Magnetic Circular Dichroism, Theory Nonlinear Optical Properties Optical Spectroscopy, Linear Polarization Theory Photoacoustic Spectroscopy, Theory Scattering Theory

Laura A Andersson

Pages 1217-1224

Georges H Wagnière and Stanisaw Wozniak

Pages 1594-1608

Josef Michl

Pages 1701-1712

András Miklós, Stefan Schäfer and Peter Hess

Pages 1815-1822

Michael Kotlarchyk

Pages 2074-2084

Theory and Applications Exciton Coupling

Nina Berova, Nobuyuki Harada and Koji Nakanishi

Pages 470-488

Methods and Instrumentation Fluorescent Molecular Probes Linear Dichroism, Instrumentation Optical Frequency Conversion ORD and Polarimetry Instruments Photoacoustic Spectroscopy, Methods and Instrumentation Spectroelectrochemistry, Methods and Instrumentation UV-Visible Absorption and Fluorescence Spectrometers Zeeman and Stark Methods in Spectroscopy, Instrumentation

F Braut-Boucher and M Aubery Erik W Thulstrup, Jens Spanget-Larsen and Jacek Waluk

Pages 573-582 Pages 1176-1178

Christos Flytzanis

Pages 1691-1701

Harry G Brittain

Pages 1712-1718

Markus W Sigrist

Pages 1810-1814

Roger J Mortimer

Pages 2174-2181

G E Tranter

Pages 2383-2389

Ichita Endo and Masataka Linuma

Pages 2505-2509

Applications Biochemical Applications of Fluorescence Spectroscopy Biomacromolecular Applications of Circular Dichroism and ORD Biomacromolecular Applications of UV-Visible Absorption Spectroscopy Chemical Reactions Studied By Electronic Spectroscopy Circularly Polarized Luminescence and Fluorescence Detected Circular Dichroism Dyes and Indicators, Use of UV-Visible Absorption Spectroscopy

Jason B Shear

Pages 77-84

Norma J Greenfield

Pages 117-130

Alison Rodger and Karen Sanders

Pages 130-139

Salman R Salman

Pages 216-222

Christine L Maupin and James P Riehl

Pages 319-326

Volker Buss and Lutz Eggers

Pages 388-396

Ellipsometry

G E Jellison, Jr

Pages 402-411

Environmental Applications of Electronic Spectroscopy Fibre Optic Probes in Optical Spectroscopy, Clinical Applications Fluorescence Microscopy, Applications Induced Circular Dichroism Inorganic Condensed Matter, Applications of Luminescence Spectroscopy Interstellar Molecules, Spectroscopy of

John W Farley, William C Brumley and DeLyle Eastwood Urs Utzinger and Rebecca R RichardsKortum

Pages 430-437

Pages 512-528

Fred Rost

Pages 565-570

Kymberley Vickery and Bengt Nordén

Pages 869-874

Keith Holliday

Pages 933-943

A G G M Tielens

Pages 943-953

Laser Applications in Electronic Spectroscopy

Wolfgang Demtröder

Thomas Gensch, Laser Induced Cristiano Viappiani and Optoacoustic Spectroscopy Silvia E Braslavsky Erik W Thulstrup, Jacek Linear Dichroism, Waluk and Jens Applications Spanget-Larsen Multiphoton Michael N R Ashfold Spectroscopy, and Colin M Western Applications Stephen G Schulman, Organic Chemistry Qiao Qing Di and John Applications of Fluorescence Spectroscopy Juchum Photoacoustic Markus W Sigrist Spectroscopy, Applications Scattering and Particle F Ross Hallett Sizing, Applications Spectroelectrochemistry, R J Mortimer Applications

Pages 1113-1123 Pages 1124-1132

Pages 1169-1175

Pages 1424-1433

Pages 1718-1725

Pages 1800-1809 Pages 2067-2074 Pages 2161-2174

Stars, Spectroscopy of

A G G M Tielens

Pages 2199-2204

Zeeman and Stark Methods in Spectroscopy, Applications

Ichita Endo and Masataka Linuma

Pages 2501-2504

Fundamentals of Spectroscopy Theory Electromagnetic Radiation David L Andrews Fourier Transformation and Sampling Theory Symmetry in Spectroscopy, Effects of

Pages 397-401

Raúl Curbelo

Pages 632-636

S F A Kettle

Pages 2335-2339

Peter Herzig and Rainer Dirl

Tensor Representations

Pages 2342-2353

Methods and Instrumentation Calibration and Reference C Burgess Systems (Regulatory Authorities) Laboratory Information David R McLaughlin Management Systems and Antony J Williams (LIMS)

Pages 166-171

Pages 1105-1113

Light Sources and Optics

R Magnusson

Pages 1158-1168

Multivariate Statistical Methods

R L Somorjai

Pages 1433-1439

Quantitative Analysis

T Frost

Pages 1931-1936

High Energy Spectroscopy Theory Mössbauer Spectroscopy, Theory Neutron Diffraction, Theory Photoelectron Spectroscopy X-ray Spectroscopy, Theory Zero Kinetic Energy Photoelectron Spectroscopy, Theory

Guennadi N Belozerski

Pages 1335-1343

Alex C Hannon

Pages 1493-1503

John Holmes

Page 1831

Prasad A Naik

Pages 2487-2498

K Müller-Dethlefs and Mark Ford

Pages 2519-2526

Methods and Instrumentation High Energy Ion Beam Analysis

Geoff W Grime

Pages 750-760

Hole Burning Spectroscopy, Methods Inelastic Neutron Scattering, Instrumentation

Josef Friedrich

Pages 826-836

Stewart F Parker

Pages 905-915

Guennadi N Belozerski

Pages 1315-1323

A C Hannon

Pages 1479-1492

László Szepes and György Tarczay

Pages 1822-1830

Geoff W Grime

Pages 1901-1905

Dieter Schwarzenbach and Howard D Flack

Pages 2271-2278

Grant Bunker

Pages 2447-2453

X-ray Emission Spectroscopy, Methods

George N Dolenko, Oleg Kh Poleshchuk and Jolanta N Latoiska

Pages 2463-2467

X-ray Fluorescence Spectrometers

Utz Kramar

Pages 2467-2477

Mössbauer Spectrometers Neutron Diffraction, Instrumentation Photoelectron Spectrometers Proton Microprobe (Method and Background) Structure Refinement (Solid State Diffraction) X-ray Absorption Spectrometers

Applications Fibres and Films Studied Using X-Ray Diffraction Inelastic Neutron Scattering, Applications Inorganic Compounds and Minerals Studied Using Xray Diffraction Materials Science Applications of X-ray Diffraction Mössbauer Spectroscopy, Applications

Watson Fuller and Arumugam Mahendrasingam

Pages 529-539

Stewart F Parker

Pages 894-905

Gilberto Artioli

Pages 924-933

Åke Kvick

Pages 1248-1257

Guennadi N Belozerski

Pages 1324-1334

Powder X-Ray Diffraction, Applications Small Molecule Applications of X-Ray Diffraction X-ray Emission Spectroscopy, Applications X-ray Fluorescence Spectroscopy, Applications Zero Kinetic Energy Photoelectron Spectroscopy, Applications

Daniel Louër

Pages 1865-1875

Andrei S Batsanov

Pages 2106-2115

George N Dolenko, Oleg Kh Poleshchuk and Jolanta N Latoiska Christina Streli, P Wobrauschek and P Kregsamer K Müller-Dethlefs and Mark Ford

Pages 2455-2462

Pages 2478-2487

Pages 2509-2519

Magnetic Resonance Historical Overview Magnetic Resonance, Historical Perspective

J W Emsley and J Feeney

Pages 1232-1240

Theory Chemical Exchange Effects in NMR Contrast Mechanisms in MRI

Alex D Bain

Pages 198-207

I R Young

Pages 349-358

EPR Spectroscopy, Theory

Christopher C Rowlands and Damien M Murphy

Pages 445-456

Magnetic Field Gradients in High-Resolution NMR

Ralph E Hurd

Pages 1224-1232

MRI Theory

Ian R Young

Pages 1388-1396

NMR in Anisotropic Systems, Theory

J W Emsley

Pages 1521-1527

NMR Principles

P J Hore

Pages 1545-1553

NMR Pulse Sequences

William F Reynolds

Pages 1554-1567

NMR Relaxation Rates

Ronald Y Dong

Pages 1568-1575

Nuclear Overhauser Effect

Anil Kumar and R Christy Rani Grace

Pages 1643-1653

Janez Seliger

Pages 1672-1680

G A Webb

Pages 1745-1753

Timothy J Norwood

Pages 1875-1884

Daniel Canet

Pages 1937-1944

Nuclear Quadrupole Resonance, Theory Parameters in NMR Spectroscopy, Theory of Product Operator Formalism in NMR Radiofrequency Field Gradients in NMR, Theory

Methods and Instrumentation Cecil Dybowski, Alicia Glatfelter and H N Cheng

Pages 149-158

EPR, Methods

Richard Cammack

Pages 457-469

In Vivo NMR, Methods

John C Lindon

Pages 866-868

C NMR, Methods

Laser Magnetic Resonance A I Chichinin

Pages 1133-1140

MRI Instrumentation

Paul D Hockings, John F Hare and David G Reid

Pages 1372-1380

NMR Data Processing

Gareth A Morris

Pages 1514-1521

NMR Microscopy

Paul T Callaghan

Pages 1528-1537

NMR Spectrometers

John C Lindon

Pages 1576-1583

Nuclear Quadrupole Resonance, Instrumentation

Taras N Rudakov

Pages 1663-1671

Ralf-Oliver Seitter and Rainer Kimmich J W Zwanziger and H W Solid State NMR, Methods Spiess Solvent Suppression Maili Liu and Xi-an Mao Methods in NMR Spectroscopy Sonically Induced NMR John Homer Methods Two-Dimensional NMR, Peter L Rinaldi Methods Relaxometers

Pages 2000-2008 Pages 2128-2136 Pages 2145-2152 Pages 2152-2159 Pages 2370-2381

Applications Biofluids Studied By NMR 13

C NMR, Parameter Survey

Carbohydrates Studied By NMR Cells Studied By NMR Chemical Applications of EPR Chemical Shift and Relaxation Reagents in NMR

John C Lindon and Jeremy K Nicholson R Duncan Farrant and John C Lindon Charles T Weller Fátima Cruz and Sebastián Cerdán Christopher C Rowlands and Damien M Murphy Silvio Aime, Mauro Botta, Mauro Fasano and Enzo Terreno

Pages 98-116 Pages 159-165 Pages 172-180 Pages 180-189 Pages 190-198

Pages 223-231

Chromatography-NMR, Applications

J P Shockcor

Pages 301-310

CIDNP Applications

Tatyana V Leshina, Alexander I Kruppa and Marc B Taraban

Pages 311-318

Peter Stilbs

Pages 369-375

Myriam Malet-Martino and Robert Martino

Pages 375-388

Thomas J Wenzel

Pages 411-421

Diffusion Studied Using NMR Spectroscopy Drug Metabolism Studied Using NMR Spectroscopy Enantiomeric Purity Studied Using NMR

EPR Imaging

L H Sutcliffe

Pages 437-445

Claudio Pettinari and Giovanni Rafaiani

Pages 489-498

Brian Hills

Pages 593-601

Nancy S True

Pages 660-667

Frank G Riddell

Pages 677-684

Claudio Pettinari, Fabio Marchetti and Giovanni Rafaiani

Pages 685-690

Janusz Lewiski

Pages 691-703

Claudio Pettinari

Pages 704-717

Trevor G Appleton

Pages 718-722

Ioannis P Gerothanassis

Pages 722-729

Dieter Rehder

Pages 731-740

Erkki Kolehmainen

Pages 740-750

Jiri Jonas

Pages 760-771

Etsuko Katoh and Isao Ando

Pages 802-813

Anne S Ulrich

Pages 813-825

F NMR, Applications, Solution State Food Science, Applications of NMR Spectroscopy Gas Phase Applications of NMR Spectroscopy Halogen NMR Spectroscopy (Excluding 19F) Heteronuclear NMR Applications (As, Sb, Bi) Heteronuclear NMR Applications (B, Al, Ga, In, Tl) Heteronuclear NMR Applications (Ge, Sn, Pb) Heteronuclear NMR Applications (La–Hg) Heteronuclear NMR Applications (O, S, Se and Te) Heteronuclear NMR Applications (Sc–Zn) Heteronuclear NMR Applications (Y–Cd) High Pressure Studies Using NMR Spectroscopy High Resolution Solid State NMR, 13C High Resolution Solid State NMR, 1H, 19F In Vivo NMR, Applications, 31P In Vivo NMR, Applications, Other Nuclei Labelling Studies in Biochemistry Using NMR

Ruth M Dixon and Peter Styles Jimmy D Bell, E Louise Thomas and K Kumar Changani Timothy R Fennell and Susan C J Sumner

Pages 851-857 Pages 857-865 Pages 1097-1105

Liquid Crystals and Liquid Crystal Solutions Studied By NMR Macromolecule–Ligand Interactions Studied By NMR Membranes Studied By NMR Spectroscopy

Lucia Calucci and Carlo Alberto Veracini

Pages 1179-1186

J Feeney

Pages 1209-1216

A Watts and S J Opella

Pages 1281-1291

MRI Applications, Biological

David G Reid, Paul D Hockings and Paul G M Mullins

Pages 1344-1354

MRI Applications, Clinical

Martin O Leach

Pages 1354-1364

MRI Applications, Clinical Flow Y Berthezène Studies

Pages 1365-1372

MRI of Oil/Water in Rocks

Geneviève Guillot

Pages 1380-1387

MRI Using Stray Fields

Edward W Randall

Pages 1396-1403

Muon Spin Resonance Spectroscopy, Applications

Ivan D Reid and Emil Roduner

Pages 1439-1450

Nitrogen NMR

G A Webb

Pages 1504-1514

NMR of Solids

Jacek Klinowski

Pages 1537-1544

NMR Spectroscopy of Alkali Metal Nuclei in Solution

Frank G Riddell

Pages 1584-1593

Oleg Kh Poleshchuk Nuclear Quadrupole Resonance, and Jolanta N Applications Latosiska Nucleic Acids Studied Using John C Lindon NMR David G Gorenstein 31 P NMR and Bruce A Luxon Perfused Organs Studied Using John C Docherty NMR Spectroscopy Proteins Studied Using NMR Paul N Sanderson Spectroscopy

Pages 1653-1662 Pages 1688-1689 Pages 1735-1744 Pages 1763-1770 Pages 1885-1893

Rigid Solids Studied Using MRI

David G Cory

Pages 2009-2017

Heinrich C Marsmann

Pages 2031-2042

Alejandro C Olivieri

Pages 2116-2127

David L Bryce and Roderick E Wasylishen

Pages 2136-2144

Carmen M Arroyo

Pages 2189-2198

G E Hawkes

Pages 2224-2233

Cynthia K McClure

Pages 2234-2245

Si NMR

Solid-State NMR Using Quadrupolar Nuclei Solid-State NMR, Rotational Resonance Spin Trapping and Spin Labelling Studied Using EPR Spectroscopy Structural Chemistry Using NMR Spectroscopy, Inorganic Molecules Structural Chemistry Using NMR Spectroscopy, Organic Molecules Structural Chemistry Using NMR Spectroscopy, Peptides Structural Chemistry Using NMR Spectroscopy, Pharmaceuticals

Martin Huenges and Horst Kessler Alexandros Makriyannis and Spiro Pavlopoulos

Pages 2246-2260 Pages 2261-2271

Tritium NMR, Applications

John R Jones

Pages 2366-2369

Xenon NMR Spectroscopy

Jukka Jokisaari

Pages 2435-2446

Mass Spectrometry Historical Overview Mass Spectrometry, Historical Perspective

Allan Maccoll†

Pages 1241-1248

Theory Fragmentation in Mass Spectrometry

Hans-Friedrich Grützmacher

Pages 637-648

Ion Collision Theory Ion Dissociation Kinetics, Mass Spectrometry Ion Energetics in Mass Spectrometry Ion Structures in Mass Spectrometry Ionization Theory

Anil K Shukla and Jean H Futrell

Pages 954-963

Bernard Leyh

Pages 963-971

John Holmes

Pages 971-976

Peter C Burgers and Johan K Terlouw C Lifshitz and T D Märk

Pages 990-1000 Pages 1010-1021

Metastable Ions

John L Holmes

Pages 1291-1297

Proton Affinities

Edward P L Hunter and Sharon G Lias

Pages 1893-1901

Statistical Theory of Mass Spectra

J C Lorquet

Pages 2204-2211

Methods and Instrumentation Atmospheric Pressure Ionization in Mass Spectrometry Chemical Ionization in Mass Spectrometry Chemical Structure Information from Mass Spectrometry Chromatography-MS, Methods Fast Atom Bombardment Ionization in Mass Spectrometry Field Ionization Kinetics in Mass Spectrometry Glow Discharge Mass Spectrometry, Methods

W. M. A. Niessen

Pages 18-24

Alex G Harrison

Pages 207-215

Kurt Varmuza

Pages 232-243

W W A Niessen

Pages 293-300

Magda Claeys and Jan Claereboudt

Pages 505-512

Nico M M Nibbering

Pages 539-548

Annemie Bogaerts

Pages 669-676

Inductively Coupled Plasma Mass Spectrometry, Methods Ion Molecule Reactions in Mass Spectrometry Ion Trap Mass Spectrometers Laser Spectroscopy Theory MS-MS and MSn Multiphoton Excitation in Mass Spectrometry Negative Ion Mass Spectrometry, Methods Neutralization– Reionization in Mass Spectrometry Photoelectron–Photoion Coincidence Methods in Mass Spectrometry (PEPICO) Photoionization and Photodissociation Methods in Mass Spectrometry Plasma Desorption Ionization in Mass Spectrometry Pyrolysis Mass Spectrometry, Methods Quadrupoles, Use of in Mass Spectrometry Sector Mass Spectrometers

Diane Beauchemin

Pages 875-880

Diethard K Böhme

Pages 984-990

Raymond E March

Pages 1000-1009

Luc Van Vaeck and Freddy Adams

Pages 1141-1152

W. M. A. Niessen

Pages 1404-1410

Ulrich Boesl

Pages 1411-1424

Suresh Dua and John H Bowie

Pages 1461-1469

Chrys Wesdemiotis

Pages 1469-1479

Tomas Baer

Pages 1831-1839

John C Traeger

Pages 1840-1847

Ronald D Macfarlane

Pages 1848-1857

Jacek P Dworzanski and Henk L C Meuzelaar P H Dawson and D J Douglas

Pages 1906-1919 Pages 1921-1930

R Bateman

Pages 2085-2092

Spectroscopy of Ions

John P Maier

Pages 2182-2189

Surface Induced Dissociation in Mass Spectrometry

S A Miller and S L Bernasek

Pages 2279-2294

Thermospray Ionization in W M A Niessen Mass Spectrometry K G Standing and W Time of Flight Mass Ens Spectrometers

Pages 2353-2360 Pages 2360-2365

Applications Biochemical Applications of Mass Spectrometry Cluster Ions Measured Using Mass Spectrometry Cosmochemical Applications Using Mass Spectrometry Food Science, Applications of Mass Spectrometry Forensic Science, Applications of Mass Spectrometry Hyphenated Techniques, Applications of in Mass Spectrometry Inorganic Chemistry, Applications of Mass Spectrometry Ion Imaging Using Mass Spectrometry Isotope Ratio Studies Using Mass Spectrometry Isotopic Labelling in Mass Spectrometry Medical Applications of Mass Spectrometry Nucleic Acids and Nucleotides Studied Using Mass Spectrometry

Victor E Vandell and Patrick A Limbach

Pages 84-87

O Echt and T D Märk

Pages 327-336

J R De Laeter

Pages 359-367

John P G Wilkins

Pages 592-593

Rodger L Foltz, Dennis J Crouch and David M Andrenyak

Pages 615-621

W M A Niessen

Pages 843-849

Lev N Sidorov

Pages 915-923

Albert J R Heck

Pages 976-983

Michael E Wieser and Willi A Brand Thomas Hellman Morton

Pages 1072-1086 Pages 1086-1096

Orval A Mamer

Pages 1262-1271

Tracey A Simmons, Kari B Green-Church and Patrick A Limbach

Pages 1681-1688

Organometallics Studied Using Mass Spectrometry Peptides and Proteins Studied Using Mass Spectrometry SIFT Applications in Mass Spectrometry Stereochemistry Studied Using Mass Spectrometry

Dmitri V Zagorevskii

Pages 1726-1733

Michael A Baldwin

Pages 1753-1763

David Smith and Patrik panl

Pages 2092-2105

Asher Mandelbaum

Pages 2211-2223

Spatially Resolved Spectroscopic Analysis Theory Neutron Diffraction, Theory

Alex C Hannon

Pages 1493-1503

PET, Theory

T J Spinks

Pages 1782-1791

A J Fisher

Pages 2060-2066

Zdzislaw Salamon and Gordon Tollin

Pages 2311-2319

Scanning Probe Microscopy, Theory Surface Plasmon Resonance, Theory

Methods and Instrumentation Neutron Diffraction, Instrumentation PET, Methods and Instrumentation Scanning Probe Microscopes SPECT, Methods and Instrumentation Structure Refinement (Solid State Diffraction) Surface Plasmon Resonance, Instrumentation

A C Hannon

Pages 1479-1492

T J Spinks

Pages 1771-1782

J G Kushmerick and P S Weiss

Pages 2043-2051

John C Lindon

Pages 2159-2161

Dieter Schwarzenbach and Howard D Flack

Pages 2271-2278

R P H Kooyman

Pages 2302-2310

Watson Fuller and Arumugam Mahendrasingam

Pages 529-539

Stewart F Parker

Pages 894-905

Gilberto Artioli

Pages 924-933

Åke Kvick

Pages 1248-1257

Guennadi N Belozerski

Pages 1324-1334

C J Roberts, M C Davies, S J B Tendler and P M Williams Zdzislaw Salamon and Gordon Tollin

Pages 2051-2059 Pages 2294-2302

Vibrational, Rotational and Raman Spectroscopies Historical Overview Vibrational, Rotational and Raman Spectroscopy, Historical Perspective

A S Gilbert

Pages 2422-2432

Theory IR Spectroscopy, Theory Nonlinear Raman Spectroscopy, Theory Photoacoustic Spectroscopy, Theory

Derek Steele

Pages 1066-1071

J Santos Gómez

Pages 1631-1642

András Miklós, Stefan Schäfer and Peter Hess

Pages 1815-1822

Raman Optical Activity, Theory Rayleigh Scattering and Raman Spectroscopy, Theory Rotational Spectroscopy, Theory

Laurence A Nafie

Pages 1976-1985

David L Andrews

Pages 1993-2000

Iain R McNab

Pages 2017-2028

Vibrational CD, Theory

Philip J Stephens

Pages 2415-2421

Methods and Instrumentation Chromatography-IR, Methods and Instrumentation Computational Methods and Chemometrics in Near-IR Spectroscopy High Resolution IR Spectroscopy (Gas Phase) Instrumentation

Robert L White

Pages 288-293

Paul Geladi and Eigil Dåbakk

Pages 343-349

Jyrki K Kauppinen and Jari O Partanen

Pages 784-794

IR Spectrometers

R A Spragg

Pages 1048-1057

IR Spectroscopy Sample Preparation Methods

R A Spragg

Pages 1058-1066

Microwave Spectrometers

Marlin D Harmony

Pages 1308-1314

Near-IR Spectrometers

R Anthony Shaw and Henry H Mantsch

Pages 1451-1461

Nonlinear Raman Peter C Chen Spectroscopy, Instruments Pina Colarusso, Linda Raman and Infrared H Kidder, Ira W Levin Microspectroscopy and E Neil Lewis Raman Optical Activity, Werner Hug Spectrometers Raman Spectrometers

Bernhard Schrader

Pages 1624-1631 Pages 1945-1954 Pages 1966-1976 Pages 1986-1992

Vibrational CD Spectrometers

Laurence A Nafie

Pages 2391-2402

Applications Art Works Studied Using IR and Raman Spectroscopy ATR and Reflectance IR Spectroscopy, Applications Biochemical Applications of Raman Spectroscopy Chromatography-IR, Applications Far-IR Spectroscopy, Applications Flame and Temperature Measurement Using Vibrational Spectroscopy Forensic Science, Applications of IR Spectroscopy FT-Raman Spectroscopy, Applications High Resolution Electron Energy Loss Spectroscopy, Applications High Resolution IR Spectroscopy (Gas Phase) Instrumentation Hydrogen Bonding and Other Physicochemical Interactions Studied By IR and Raman Spectroscopy Industrial Applications of IR and Raman Spectroscopy

Howell G M Edwards

Pages 2-17

U P Fringeli

Pages 58-75

Peter Hildebrandt and Sophie Lecomte

Pages 88-97

George Jalsovszky

Pages 282-287

James R Durig

Pages 498-504

Kevin L McNesby

Pages 548-559

Núria Ferrer

Pages 603-615

R H Brody, E A Carter, H. G. M. Edwards and A M Pollard

Pages 649-657

Horst Conrad and Martin E Kordesch

Pages 772-783

Jyrki K Kauppinen and Jari O Partanen

Pages 784-794

A S Gilbert

Pages 837-843

A S Gilbert and R W Lancaster

Pages 881-893

IR and Raman Spectroscopy of Inorganic, Coordination and Organometallic Compounds IR Spectral Group Frequencies of Organic Compounds Matrix Isolation Studies By IR and Raman Spectroscopies Medical Science Applications of IR Microwave and Radiowave Spectroscopy, Applications Nonlinear Raman Spectroscopy, Applications Photoacoustic Spectroscopy, Applications Polymer Applications of IR and Raman Spectroscopy Raman Optical Activity, Applications Surface Studies By IR Spectroscopy Surface-Enhanced Raman Scattering (SERS), Applications Vibrational CD, Applications

Claudio Pettinari and Carlo Santini

Pages 1021-1034

A S Gilbert

Pages 1035-1048

Lester Andrews

Pages 1257-1261

Michael Jackson and Henry H Mantsch

Pages 1271-1281

G Wlodarczak

Pages 1297-1307

W Kiefer

Pages 1609-1623

Markus W Sigrist

Pages 1800-1809

C M Snively and J L Koenig

Pages 1858-1864

Günter Georg Hoffmann

Pages 1955-1965

Norman Sheppard

Pages 2320-2328

W E Smith and C Rodger

Pages 2329-2334

Günter Georg Hoffmann

Pages 2403-2414

Plate 1

Plate 1 Roman die, ca. AD 300, from archaeological excavations at Frocester Villa, Gloucester, UK. Raman spectroscopy has suggested the origin of the die as sperm whale ivory. See Art Works Studied using IR and Raman Spectroscopy.

Plate 2

Plate 2 Selection of ornamental jewellery consisting of three bangles assumed to be ivory but which were shown spectroscopically to be composed of modern resins, and a genuine ivory necklace.See Art Works Studied using IR and Raman Spectroscopy.

Plate 3

Plate 3 ‘Ivory’ cat, which was identiﬁed spectroscopically as a modern limitation composed of poly(methyl methacrylate) and polystyrene resins with added calcite to give the texture and density of ivory. Reproduced with permission from Edwards HGM and Farwell DW, Ivory and simulated ivory artefacts: Fourier-transform Raman diagnostic study, Spectrochimica Acta, Part A, 51: 2073–2081 r 1995, Elsevier Science B. V. See Art Works Studied using IR and Raman Spectroscopy.

Plate 4

! with the Raman spectra of pigments contained therein. The combination of vermilion Plate 4 The historiated initials (a) ‘P’ and (b) ‘S’ from the Icelandic Jok and red ochre in the ‘P’ should be noted. The spectrum of bone white has been obtained from the letter ‘H’(not shown). Reproduced with permission from Bent SP, Clark RH, Daniels MAM, Proter CA and Withnay R. Identiﬁcation by Raman microscopy and visible reﬂectance spectroscopy of pigments on an Icelandic manuscript, (1995) Studies in Conservation 40: 31–40. See Art Works Studied using IR and Raman Spectroscopy.

Plate 5a

Plate 5a Elaborately historiated initial ‘R’ in sixteenth-century German choir book, with Raman microscopy spectra of selected pigmented regions. See Art Works Studied using IR and Raman Spectroscopy.

Plate 5b

Plate 5b Portion of top of column of historiated initial ‘R’ shown in 5a the individual pigment-grains can be clearly seen under the x100 magniﬁcation and can be separately identiﬁed using Raman microscopy. 5a and b, reproduced with permission from Clark RJH (1995) Raman microscopy: application to the identiﬁcation of pigments on medieval manuscripts. Chemical Society Reviews 187-196. r The Royal Society of Chemistry. See Art Works Studied using IR and Raman Spectroscopy.

Plate 6

Plate 6 Holy Sepulchre Chapel, Winchester Cathedral. Wall painting of ca. 1175–85 on the east well depicting the Deposition, Entombment, Maries at the Sepulchre and the Harrowing of Hell. Reproduced with permission from Edwards HGM, Brooke C and Tait JKF, An FT-Raman spectroscopic study of pigments of medieval English wall paintings, Journal of Raman Spectroscopy 28: 95–98, r 1997 John Wiley and Sons Ltd. See Art Works Studied using IR and Raman Spectroscopy.

Plate 7

Plate 7 A two-dimensional 1H-NMR spectrum of human urine, demonstrating the complexity of such mixtures. This was measured using the J-resolved pulse sequence and results in contour plot of spectrum intensity as a function of two frequency axes. The horizontal axis represents the usual 1H NMR chemical shift range and the vertical axis cover the 1H1H spin coupling range. Each 1H NMR spin-coupled multiplet is rotated such that the overlap between closely spaced signals is minimised, thus aiding interpretation. See Biofluids Studied by NMR. Reproduced with permission from John Lindon.

Plate 8

Plates 8 a and b Polarised light micrograph of liquid crystals. See Chiroptical Spectroscopy, Oriented Molecules and Anisotropic Systems; Liquid Crystals and Liquid Crystal Solutions Studied by NMR. Reproduced with permission from Science Photo Library.

Plate 9

Plate 9 Light micrograph of a liquid crystal display (LCD) of the type used to represent numerical ﬁgures. Although a liquid crystal (LC) can ﬂow like a ﬂuid its molecular arrangement exhibits some order. A LCD has a ﬁlm of LC sandwiched between crossed polarizers & set on top of a mirror. In the ‘off’ state light is able to traverse both polarizers to reach the mirror because it gets rotated through 90 degrees by the LC. In the ‘on’ state an electric ﬁeld applied across the LC alters its molecular alignment & hence its polarizing properties; light cannot traverse both polarizers to reach the mirror & the display appears black. See Chiroptical Spectroscopy, Oriented Molecules and Anisotropic Systems; Liquid Crystals and Liquid Crystal Solutions Studied by NMR. Reproduced with permission from Science Photo Library.

Plate 10

Plate 10 Polarised light micrograph of nematic liquid crystals.See Chiroptical Spectroscopy, Oriented Molecules and Anisotropic Systems; Liquid Crystals and Liquid Crystal Solutions Studied by NMR. Reproduced with permission from Science Photo Library.

Plate 11

Plate 11 The molecular composition of massive star-forming region in the galaxy. The false color background shows the heart of a massive star-forming region in the constellation of Orion obtained by the Hubble Space Telescope. The red colour is the emission of the molecular hydrogen excited by shocks driven by the powerful wind of a newly formed star in the center of the image. The bluish color is scattered light from this young star reﬂected in surrounding dust in its stellar nursery. The white spectrum shows the results of ground-based line survey of its region and reveals the presence of molecules such as methanol, formaldehyde, carbon monoxide, sulphur monoxide, and sulphur dioxide. See Cosmochemical Applications Using Mass Spectrometry; Interstellar Molecules, Spectroscopy of. Reproduced with permision from A.G.G.M. Tielens.

Plate 12

Plate 12 The composition of ices in low-mass star forming regions. This is a false color image of an infrared image obtained by the European ISO satelite (Infrared Space Observatory). The colors indicate emission by warm dust and gas heated by nearby stars in the star-forming molecular cloud Rho Ophiucus. Two bright objects just below the center are newly formed solar-type stars. The spectrum displayed on top was taken by a spectrograph on board ISO towards the indicated source. This spectrum reveals the presence of various simple molecules such as water, carbon monoxide in ice-form in these regions. The composition of these ices is very similar to those of comets in the solar system and eventually, these ices are expected to coagulate together forming comets in the budding planetary system around this newly formed star. See Cosmochemical Applications Using Mass Spectrometry; Interstellar Molecules, Spectroscopy of. Reproduced with permission from A.G.G.M. Tielens.

Plate 13

Plate 13 Large molecules in the ejecta from a dying star. When solar-type stars grow old they shed much of their mass in the form of a wind much of that in the form of complex molecules. Eventually, the star becomes very hot and sets this ejected material aglow, forming so-called planetary nebulae. The material ejected during this stage will mix with other material present in space and eventually new stars and planetary systems like our own will form these stellar ashes. The false color image shows the emission of molecular hydrogen (red) and atomic hydrogen (white) in the planetary nebula NGC 7027 obtained by the ISO satelite (Infrared Space Observatory). This spectrum reveals the presence of large complex molecules called polycyclic aromatic hydrocarbons. These molecules are formed in a process akin to that of sooting ﬂames in terrestrial envirnoments such as car engines. See Cosmochemical Applications Using Mass Spectrometry; Interstellar Molecules, Spectroscopy of. Reproduced with permission from A.G.G.M. Tielens.

Plate 14

Plate 14 A molecule of the pigment quinacridone. The coloured surfaces represent key molecular properties predicated using computational chemistry. See Dyes and Indicators, Use of UV-Visible Absorption Spectroscopy. Reproduced with permission from Molecular Simulations (www.msi.com).

Plate 15

Plate 15 Sir Frederick William Herschel (1738–1822) German-British astronomer, and disocerer of Uranus, 1781. Reproduced with permission from Mary Evans Picture Library.

Plate 16

Plate 16 Sir Chandrasehara Venkata Raman. Indian (1888–1970) physcicist, Nobel Laureate in physics 1930 and discoverer of the Raman effect, a scattering of light by a sample which results in a small fraction of the light being of a different frequency. This phenomenon has applications in analytical chemistry and molecular structure. See Electromagnetic Radiation; Vibrational, Rotational and Raman Spectroscopy, Historical Perspective. Reproduced with permission from Mary Evans Picture Library.

Plate 17

Plate 17 Imaging of a rat kidney perfused with 0.5mM triayl-methyl (TAM) radical. A few representative slices (2424 mm2), obtained from the 3D spatial image are shown. A1-A6 are vertucak slices (0.75 mm apart), B1-B3 are traverse slices (0.75 mm apart). The images show the structure of cannula (a), renal artery (b), cortex (c), and calysis (d). See EPR Imaging. Reproduced with permission from Peraiannan Kuppusamy/EPR Laboratories Johns Hopkins University School of Medicine.

Plate 18

Plate 59 Three-dimensional images (25 25 25 mm3) of an ischaemic rat heart infused with a glucose char suspension. (A) Full view of the heart. (B) A longitudinal cut out showing the internal structure of the heart. C is the cannula; Ao the aorta; PA the pulmonary artery; LM the left main coronary artery; LAD the left anterior descending artery; and LV the left ventricular cavity. See EPR Imaging. Reproduced with permission from Periannan Kuppusamy/ERP Laboratories Johns Hopkins University School of Medicine.

Plate 19

Plate 19 Small angle X-ray ﬁbre diffraction pattern recorded from Plaice ﬁsh muscle. See Fibres and Films Studied Using X-Ray Diffraction. Reproduced with permission from W. Fuller and A. Mahendrasingam.

Plate 20

Plate 20 An example of a ﬂuorescent molecular probe, Endogenous alkaline phosphatase activity in the zebraﬁsh brain was localized with an endogenous phosphatase detection kit. Enzymatic cleavage of the phosphatase substrate yields a bright yellow-green ﬂuorescent precipitate at the site of enzyme activity. See Fluorescent Molecular Probes. Reproduced with permission from Greg Cox, Molecular Probes, Inc.

Plate 21

Plate 21 A ﬂuorescent molecular probe. Fixed and permeabilized osteosarcoma cells were simultaneously stained with the ﬂuorescent lectins, Alexa 488 concanavalin A (Con A) and Alexa 594 wheat germ agglutinin (WGA). Con A selectively binds a-mannopyranosyl and a-glucopyranosyl residues, whereas WGA selectively binds sialic acid counterstained with blue-ﬂuorescent Hoechst 33342 nucleic acid stain. See Fluorescent Molecular Probes, Inc.

Plate 22

Plate 22 Radiocarbon dating. View of a linear accelerator used as part of an accelerator mass spectrometer (AMS). This device is capable of counting the relatively few carbon-14 atoms in a radioactive sample. The proportion of carbon-14 to carbon-12 atoms in the sample may be used to determine the radiocarbon age of an organic object. This is then adjusted by various corrections to give the true age. See Isotope Ratio Studies Using Mass Spectrometry. Reproduced with permission from Science Photo Library.

Plate 23

Plate 23 Raman microscope image of a pharmaceutical powder showing the relative importance of three compound, A, B and C. See Industrial Application of IR and Raman Spectroscopy. Courtesy of A.S. Gilbert and R.W. Lancaster.

Plate 24

Plate 24 3-D reconstruction of a plant cell undergoing meiotic cell division, created from two-photon cross-sectioned images. See Laser Spectroscopy Theory; Light sources and Optics. Reproduced with permission from Spectra-Physics r W. Zipfel and C. Conley/Cornell University.

Plate 25

Plate 25 Thorax and abdomen. Coloured Magnetic Resonance Imaging (MRI) scan of the thorax and abdomen of a woman aged 58 years, seen in posterior view. The arms are seen on either side, with the waist and hips at lower frame. The thorax contains the blue lung ﬁelds (upper frame) with bones of the thoracic spines at upper centre. In the abdomen are lobes of the liver (green, below the lungs) and a pair of kidneys (green, below the liver lobes). Bands of muscle can be seen around the lumbar spine (lower centre) and the shoulder (at top, magenta). MRI uses probes of radiowave energy in the presence of a magnetic ﬁeld to create slices. See MRI Applications, Clinical; MRI Instrumentation; MRI Theory; Contrast mechanisms in MRI. Reproduced with permission from Science Photo Library.

Plate 26

Plate 26 False-colour magnetic resonance image (MRI) of a whole human body, a woman, taken in coronal (frontal) section. Various parts of the body are prominent: bone appears in black, wiht the hip and shoulder joints clearly deﬁned. The lungs are also black. Vertebrae forming the spinal column are also obvious (in yellow and white). Part of the spinal cord is visible (in gold) beneath the base of the brain, the hemispherical structure of which is revealed in the section. This whole body image is the product of a number of MRI scans made along the length of the body. The complete image was rendered from data relating to the various sections stored on the scanner’s computer. See MRI Applications, Clinical; MRI Instrumentation; MRI Theory; Contrast mechanisms in MRI. Reproduced with permission from Science Photo Library.

Plate 27

Plate 27 Portrait of Francis Aston (1877–1945) British physicist and Nobel Laureate. After WW1, Aston helped Thompson in his studies of the deﬂection of ions in magnetic ﬁelds. He went on to improve Thompson’s apparatus, designing it so as to make all atoms of a given mass fall on the same part of a photographic plate. Working with neon, he found that two lines were isotopes. He repeated this with chlorine with similar results. The device, called the mass spectrometer, showed that most stable elements had isotopes. His work earned him the 1922 Nobel Prize for Chemistry, and introduced a powerful new analytical tool to science. See Mass Spectrometry, Historical Perspective. Reproduced with permision from Science Photo Library.

Plate 28

Plate 28 Magnetic Resonance Image of Human female brain. See MRI Applications, Clinical. Reproduced with permission from r Medipics/Dan McCoy/ Rainbow.

Plate 29

Plate 29 Structure of membrane proteins. Membrane peptides often self assemble into controlled oligomeric forms to make molecular selective channels whose structure can be very difﬁcult to resolve at the molecular level by most methods, although solid state NMR methods can make a contribution to their functional and structural description. Here the pentameric funnel-like bundle of the M2-peptide from the nicotinic acetly choline receptor has been resolved from 15N NMR studies of oriented M2 peptides in lipid bilayers. The funnel has a wide mouth at the N-terminal, intracellular side of the pore. The pore lining residues has also been modelled and distances between residues in the channel estimated. The a-carbon backbone is in cyan, acidic residues in red and basic residues in blue, polar residues in yellow and lipophilic residues in purple. (Figure adapted from Opella et al., (1999) Nature St. Biology, 6: 374–379). See Membranes Studied by NMR Spectroscopy.

Plate 30

Plate 30 Biological membranes which enclose all living cells, and are also present within the cells of higher life forms, are very complex and heterogeneous in their chemical make-up. Between 20 and 40% of components encoded by the genome in any one living cell eventually end up in the membranes of the cell. Any one typical membrane may be composed of over 100 different bilayer lipids and sterols, several hundreds of different proteins, some of which have polysaccharides attached to them, and the whole assembly may be supported by a protein scaffold underlying the membrane on the cytoplasmic side. The whole molecular complex is in dynamic equilibrium, with lipids and sterols rotating fast (about 109 times per second) around their long axis, and diffusing laterally (covering about 108 cm2 in one second), whilst at the same time maintaining a regular and relatively ordered structure, on average. Such dynamic information has come from a range of spectroscopic methods, including NMR, in which it is usual to study just one, or a limited number of such components in a simpliﬁed model system, and try and understand them and their interactions with a limited number of other components. (Figure by Ove Broo Sorensen of the Technical University of Copenhagen, Denmark). See Membranes Studied by NMR Spectroscopy.

Plate 31

Plate 31 Richard Ernst, Nobel Laureate in Chemistry 1991. Notes for this contribution to NMR spectroscopy and MRI. See Magnetic Resonance, Historical Perspective. Reproduced with permission from The Nobel Foundation.

Plate 32

Plate 32 Felix Bloch who was awarded the Nobel prize for physics in 1952 jointly with Edward Purcell. They led two independent research groups which, in 1945, ﬁrst detected the nuclear magnetic resonance phenomenon in bulk matter. See Magnetic Resonance, Historical Perspective. Reproduced with permission from The Nobel Foundation.

Plate 33

Plate 33 The ﬁrst fully functioning electrospray mass spectrometer built at Yale by Masamichi Yamashita in 1983. Reproduced with permission from John B. Fen.

Plate 34

Plate 34 A schematic representation of a typical electrospray mass spectrometer in which the mass analyser is a quadrupole mass ﬁlter. Reproduced with permission from John B. Fen.

Plate 35

Plate 35 A Bell Lab ZAB high ﬁeld mass spectrometer. Reproduced with permission from r Medipics/Dan McCoy/Rainbow.

Plate 36

( The insert is the actualpattern which can be Plate 36 The left-hand picture illustrates a powder diffraction pattern of S2N2 at a wavelength of 0.325 A. integrated and displayed as a function of d[ d* = 2 sin q/l]. The right-hand picture shows the result if one single S2N2 crystal is rotated in the X-ray beam. The size of the spots illustrates the difference in diffracted intensities from the separated Bragg reﬂections. See Material Science Applications of X-ray Diffraction. Courtesy of Svensson and Kvick, 1998.

Plate 37

Plate 37 Schematic cross-section through a typical superconducting clinical MR scanner. Within the cryostat (light blue) are the superconducting coils of the primary magnet (red) and active shield (green). In the bore of the magnet there are passive shim rods (grey), active shim coils (orange), gradient set (blue), whole body RF coil (black) and patient bed. The tractable diameter is generally half the magnet bore diameter. See MRI Instrumentation. Courtesy of P.D. Hockings, J.F. Hare and D.G. Reid.

Plate 38

Plate 38 a and b Black and white & colourscale diffusion weighted spin echo magnetic resonance images of postmortem human cervical spinal cord using a micro-imaging probe at 600 MHz observation frequency. See NMR Microscopy. Reproduced with permission from Doty Scientiﬁc Inc.

Plate 39

Plate 39 Small angle X-ray ﬁbre diffraction pattern recorded form of the DNA double-helix. See Nucleic Acids and Nucleotides Studied Using Mass Spectroscopy; Nucleic Acids Studied by NMR. Courtesy of A. Mahendrasigam and W. Fuller.

Plate 40

Plate 40 The anisotrophy of chemical shift, dipolar and quadrupolar coupling usually broadens NMR spectra of very large, slowly tumbling complexes making a detailed structural analysis a difﬁcult task. If however the sample is set spinning at an angle of 54.7 (MASS = magic angle sample spinning) with respect to the applied magnetic ﬁeld, all anisotropies are scaled down and eventually collapse at sufﬁciently high speeds resulting in isotropic spectra. In a special version of MASS called MAOSS, oriented membranes are set spinning at moderate speeds (1–4 kHz) at the magic angle, and now the narrow spinning side bands, which arise from not-averaged anisotropic interactions, give information about the orientation of the group being observed (here a CD3 group in deuterium NMR of a prosthetic group in a large protein embedded in membrane) at much higher sensitivity and resolution than could be obtained from a conventional spectrum of a statically positioned sample. Similar experimental condition for other nuclei (13C, 15N), permit both the magnitude and direction of the dipolar couplings to be extracted, to give high resolution structural details within a large membrane complex, (courtesy of Dr C. Glaubitz; adapted from Glaubitz and Watts, (1998) J. Mag. Res. 130; 305–316). See NMR in Anisotropic Systems, Theory; Solid State NMR, Methods.

Plate 41a

Plate 41a A superconducting NMR magnet operating at 18.8T for 1H NMR observation at 800 MHz demonstrating the size of these state-of-the-art magnets. Reproduced with permission from Bruker Instruments Inc., Billerica, MA, USA.

Plate 41b

Plate 41b A modern high-resolution NMR spectrometer. A superconducting magnet is shown at the rear, in this case providing a ﬁeld of 18.8T corresponding to a 1H observation frequency of 800 MHz. Behind the operator is the single console containing the RF and other electronics and the temperature-control unit. The whole instrument is computer controlled by the workstation shown at the right. Reproduced with permission from Bruker Instruments Inc., Billerica, MA, USA. See NMR Spectrometers.

Plate 42

Plate 42 PET functional images of glucose metabolic rate (MRGl) (right), MRI images (magnetic resonance images, left) showing anatomical detail, and coregistered (overlaid) PET/MRI (centre). See PET, Methods and Instrumentation.

Plate 43

Plate 43 X-ray photoelectron spectrometer. This is an analytical instrument, used mainly in metallurgy. It consists of an electron source, a sample stage and Xray detectors. A steam of electrons is accelerated toward and focussed on the sample. When an electron strikes an atom in the target, it may cause the ejection of an electron in the atomic shell. If this is replaced from an electron in a higher-energy shell, a photon of a speciﬁc wavelength is emitted, normally in the X-ray region. This may be detected and analysed, giving an indication of the identity and quantity of given elements in the sample. See Photoelectron Spectrometers. Reproduced with permission from Science Photo Library.

Plate 44

Plate 44 Common cold virus protein: computer graphics representation of protein comprising 1 of the 60 faces of the icosahe-dral capsid (casing) of rhinovirus14, a virus causing the common cold. This image is of the atomic backbone of the protein. Rhinovirus belongs to the picornavirus (small RNA viruses) group, which also includes the enteroviruses and the agent of foot and mouth disease in cattle. They are all relatively small (24–35 nanometres diameter), nonenveloped, with their genetic information held in the form of RNA. Rhinovirus infects the nose and throat, and is spread in droplets from talking, coughing and sneezing. See Proteins Studied Using NMR Spectroscopy. Reproduced with permission from Science Photo Library.

Plate 45

Plate 45 A researcher at the BOC Group Technical Centre, Murray Hill, New Jersey, USA, using an advanced Electron Spectroscopy For Chemical Analysis (ESCA) unit. ESCA provides qualitative and quantitative analysis of the chemistry of elements present in the outermost layers of solid materials. The unit is used in the development of thin-ﬁlm coatings, medical sensors, molecular sieves and catalysts. See Photoelectron Spectrometers; Photoelectron Spectroscopy. Reproduced with permission from Science Photo Library.

Plate 46

Plate 46 Photoacoustic Multi-gas Monitor. See Photoacoustic Spectroscopy, Methods and Instrumentation. Reproduced with permission from INNOVA AirTech Instruments.

Plate 47

Plate 47 The molecular structure of Bovine Seminal Ribonuclease, as determined using X-ray diffraction. The disulphide bridges are shown in yellow. The structure was drawn form the protein database entry drawn. See Proteinss Studied Using NMR Spectroscopy. Supplied by Dr G.G. Hoffmann.

Plate 48

Plate 48 A photoacoustic spectroscopy (PAS) measuring cell, shown with the germanium window half removed. The gas analysis is based on the same principle as conventional infra-red (IR) monitoring except that here the amount of IR light absorbed is measured directly by determining the amount of sound energy emitted upon the absorption. See Photoacoustic Spectroscopy, Methods and Instrumentation. Reproduced with permission from INNOVA AirTech Instruments.

Plate 49

Plate 49 Transaxial planes through the brains of three subjects representing glucose utilisation obtained by application of the radiotracer ﬂuoro-deoxy-glucose (FDG) PET tomography. The top of the images are the front of the brain and the person is looked upon from below. The values of the images are colour coded according to a linear scale red being the highest value. The middle panel shows the distribution in a healthy person. The left panel is from a person with Huntington’s disease illustrating a loss of tracer uptake at the centre as a consequence of local nervous tissue loss. The right panel is from a person who had similar features of disordered movement (cholera), but based on different brain pathology. The image in this person shows however a clear increase in tracer uptake in the central regions (called: the striatum) of the brain. See PET, Methods and Instrumentation. Reproduced with permission from Prof. Leenders.

Plate 50

Plate 50 3-D computerized reconstruction of FDG uptake measured by PET in a healthy volunteer. Part of the right upper and frontal brain is ‘‘cut out’’. The left panel shows the tracer uptake in a rest condition. The right panel shows the uptake in the same person but now during intake of a hallucinogen. It can be seen that the frontal cortex region in the brain has a higher (more red) glucose uptake in the activated condition. See PET, Methods and Instrumentation. Reproduced with permission from Prof. Leenders.

Plate 51

Plate 51 Transaxial planes through the human brain after administration of the of the radiotracer ﬂuoro-dopa indicating dopa decarboxylase capacity as measured by PET. The left panel shows a healthy volunteer whilst right panel shows a patient with Parkinson’s disease. It can be seen that the tracer uptake measured with PET in the patient is markedly reduced within the regions of the basal ganglia. See PET, Methods and Instrumentation. Reproduced with permission from Prof. Leenders.

Plate 52

Plate 52 Laser optical bench system used as an excitation source for a Raman spectrometer. Raman spectroscopy provides essentially the same sort of information about molecular structure and dimensions as do infrared and microwave spectroscopy. The laser is an ideal Raman source: it provides a narrow, highly mono-chromatic bearn of radiation which may be focused accurately into small smaple. See Raman Spectrometers; FT-Raman Spectroscopy, Applications; Industrial Applications of IR and Raman Spectroscopy; Polymer Application of IR and Raman Spectroscopy; IR and Raman Spectroscopy of Inorganic, Coordination and Organometallic Compounds. Reproduced with permission from Science Photo Library.

Plate 53

Plate 53 Direct Raman imaging, using the b-carotene line at 1552cm 1, of live corpus luteum cells. This provides a rapid insight into the distribution of different molecules within the cell (old cell below, young above). See Raman and IR Microspectroscopy. Reproduced with permission from Renishaw pic/Prof. D. N. Batchelder et al, Dept. Physics and Astronomy, University of Leeds.

Plate 54

Plate 54 Gallium Nitride Impurity Doping Images. See Raman and IR Microspectroscopy. Reproduced with permission from Renishaw pic.

Plate 55

Plate 55 A computer generated molecular model of the cardiac drug digoxin showing the molecular structure as derived by different techniques. The green structure is that determined in the solid state using X-ray diffraction. The red structure is calculated using a molecular dynamics approach. The cyan and magenta structures are the two possible structures determined in solution using NMR spectroscopy. See Structural Chemistry Using NMR Spectroscopy, Pharmaceuticals. Reproduced with permission from John Lindon.

Plate 56

Plate 56 A composite of two images of the GaAs(110) surface. The orange features obtained at positive sample bias are the Ga atoms, while green features obtained at a negative sample bias are the As atoms. Feenstra RM, unpublished results. See Scanning Probe Microscopes.

Plate 57

( area and are plotted as three-dimensional Plate 57 Three STM images of a Ni3 cluster adsorbed on a MoS2 basal plane at 4 K. All three images show a 60A representations with the same aspect ratio and with the same angle of view. The images were acquired with sample biases of: +2 V (upper), +1.4 V (middle), and 2 V (lower). Reproduced with the permission of the American Chemical Society from Kushmerick JG and Weiss PS (1998) Journal of Physical Chemistry B102: 10094–10097. See Scanning Probe Microscopes.

Plate 58

Plate 58 Vibrational spectroscopic imaging of C2H2 and C2D2. (A) Constant current STM image of a C2H2 molecule (left) and C2D2 molecule (right). The d2/d ( with 1 nA V2 images of the same area recorded with a bias voltage of (B) 358 mV, (C) 266 mV, and (D) 311 mV, with a 10 mV modulation. All images are 48 A DC tunnelling current. Reproduced with the permission of the American Association for the Advancement of Science from Stipe BC et al. (1998) Science 280: 1732–1735. See Scanning Probe Microscopes.

Plate 59

Plate 59 Fluorescence emission near-ﬁeld scanning optical microscope image (1.3 mm 1.3 mm) of a photosynthetic membrane fragment. Reproduced with permission of the American Chemical Society from Dunn RC et al. (1994) Journal of Physical Chemistry 98: 30943098. See Scanning Probe Microscopes.

Plate 60

Plate 60 Sir Isaac Newton. He is shown using a prism to decompose light. See Electromagnetic Radiation. Reproduced with permission from Mary Evans Picture Library.

2 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

A Antimony NMR, Applications See Heteronuclear NMR Applications (As, Sb, Bi).

Arsenic NMR, Applications See Heteronuclear NMR Applications (As, Sb, Bi).

Introduction The scientific study of works of art and the materials that have been used in their creation has received impetus with the development of nondestructive, microsampling analytical techniques and an increasing awareness on the part of art historians, museum conservators and scientists of the importance of characterization for the attribution of the historical period and genuiness of an artefact. Useful information about ancient technologies and methods used in the construction of works of art is now forthcoming; in particular, spectroscopists are realizing the challenge that is being provided by the analytical characterization of ancient materials. In this article, the applications of vibrational spectroscopic techniques to art are addressed and examples are taken to illustrate the following: identification of pigments, nondestructively, in preserved illuminated manuscripts, paintings and watercolours; characterization of genuine and fake artefacts, e.g. ivories; rock art and frescoes effects of environmental and climatic degradation on exposed artwork; provision of information about archaeological and historical trade routes from the spectroscopic analysis of dyes, pigments and resins;

VIBRATIONAL ROTATIONAL & RAMAN SPECTROSCOPIES Applications identification of biomaterials in carved artwork, e.g. horn, hoof and tortoiseshell; information of critical importance to art restorers and museum conservation scientists ill-restored items and the preservation of deteriorating material.

Comparison of infrared and Raman spectroscopies Although both Fourier transform-infrared (FT-IR) and Raman spectroscopies have been applied to the study of art and museum artefacts, it is possible to identify several indicators that will dictate the preferential use of either technique. Infrared studies have been reported from a much earlier date than Raman studies; hence there now exists much more comprehensive infrared database for pigments and natural or synthetic materials. Specific points also need to be considered relating to the sampling and composition of the specimens being studied. The presence of highly fluorescent coatings or impurities in ancient specimens dictated the use of infrared spectroscopy when the alternative was visible excitation for Raman spectroscopic studies; fluorescence often swamped the lower-intensity Raman signals from such samples.

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 3

However, the smaller scattering intensity of water and hydroxyl groups in the Raman effect, and their strong absorption in the infrared, generally favours the use of the Raman technique for the characterization of ancient, hydrated biomaterials such as linens and cottons. In the area of paintings and manuscripts, Raman spectroscopy has had extensive application in recent years because of the accessibility of the low-wavenumber regions of the vibrational spectrum (< 500 cm1), which are vitally important for the characterization of inorganic and mineral pigments. This region is observed only with difficulty in the infrared, and then with the use of special instruments with far-infrared capability. The shape and reflectivity of surfaces are also important for specimens that cannot be subjected to mechanical or chemical pretreatment the taking of small samples by drilling, scraping or excision is often prohibited. Curved surfaces are notoriously poor for infrared examination and specimens are better analysed after being subjected to a flattening or crushing process. The use of fibreoptic probes for remote analysis is now advocated for many infrared and Raman applications, and art objects are no exception. However, the interpretation of the spectral data from remote-probe scanning experiments is dependent on accurate subtraction of probe background; this is particularly relevant for features below 1000 cm 1, e.g. artists pigments. However, the obvious advantage of having the capability of taking the Raman or infrared spectrometer into a museum environment for in situ examination of an artefact or painting has dictated the construction of several portable units that might well provide a new dimension to these applications. Finally, the advent of FT-Raman spectroscopy with near-infrared excitation from a Nd3+/YAG laser at 1064 nm has given a new dimension to art applications, especially for the study of naturally fluorescent biomaterials such as horn, hoof, tortoiseshell and ivory. Improvements in the generation of Raman spectra and their detection using CCD-visible Raman spectroscopy are now also providing valuable new ways of analysing often difficult museum specimens. The role of Raman microscopy (and also infrared microscopy) in the characterization of micro-sized specimens will be illustrated in this article. Generally, it is now possible to achieve goodquality Raman microscopy spectra from sample areas as small as 12 µm using visible excitation (~ 0.5 mm has been claimed in some instances) and about 515 µm using infrared and Raman FT techniques. The latest advances in confocal microscopy using visible excitation and CCD detection now

means that it is possible to depth-profile suitable specimens with a resolution of about 1 µm or better; this is important for the analysis of coatings applied by artists to paintings.

Applications of IR and Raman spectroscopies Plastics

Since the first commercial plastic, parkesine, in 1862, a huge range of functional articles have been produced in these media; many advantages were appreciated, including the ability to incorporate pigments and to mould large objects. Museums now have exhibitions devoted to plastic articles, but in recent years the problems in their conservation and storage and the control of degradation in damaged plastic exhibits has become acute. In particular, the acid ester plastics such as cellulose nitrate and cellulose acetate are prone to degradation, which has often resulted in major destruction of the articles concerned. Examples include early motion picture films, Victorian substitute tortoiseshell articles, Bakelite and childrens toys, such as dolls. FT-IR and Raman spectroscopy have been instrumental in discovering the methods of degradation and the source of possible triggers; suggestions have thus been made for the proper storage and conservation of plastic articles for future generations, such as dolls from the 1940s, space suits from the 1960s and art nouveau objects from the 1920s. Glass

Degradation processes in medieval stained glass windows have been studied using FT-IR spectroscopy and the pigments applied to early specimens have been characterized using the FT-Raman technique. Similar methods have been used to study early enamels and cloisonné specimens. Faience

Faience was produced in Egypt over 5000 years ago and may truly be considered as the first hightechnology ceramic. The faience body, formed from sand with additives such as lime and natron (a hydrated sodium carbonate found naturally as a mineral in dried lake beds), was shaped and modelled, and a glaze was applied and fired. Sixteen pieces of Egyptian faience from Tell-elAmarna have been analysed using Raman microscopy and the red and yellow pigments identified as red ochre and lead antimonate yellow, respectively.

4 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

The preservation and restoration of objects made from biomaterials is particularly challenging as their degradation products are complex and diverse. Examples of problems facing museum conservators include art objects made from ivory, horn, natural resins and textiles; these objects have often become fragile and restoration is of the utmost significance and importance. Often, earlier restorative procedures, which may have been incompletely documented, are no longer satisfactory and the results of chemical deterioration of applied restoratives under the influence of solar radiation and humidity changes in storage are sadly too often plain to see. There are several good examples in the recent literature of the successful application of vibrational infrared and Raman spectroscopy to the characterization of art objects composed of biomaterials. Ivory, a generic name for the exoskeletal dental growths of certain mammalian species, has been appreciated as an art medium for thousands of years; it is soft enough to be carved and polished, yet hard enough to resist superficial weathering damage. However, the identification and attribution of archaeological ivories to species of elephant, sperm whale, narwhal, hog and hippopotamus, for example, is often fraught with difficulty, particularly where the ivory object is small and may only be a

part of a larger specimen. The observation of Schreger lines and surface morphology is extremely difficult in such cases. Very recently, Raman spectroscopy has been used to provide a suggested protocol for ivory identification and characterization and has been used successfully to determine the animal origin (sperm whale) of a Roman die excavated at Frocester Roman villa (third century AD) in the UK (Figure 1). There have also been several examples of the use of Raman spectroscopy to identify genuine and fake ivory articles. Figures 2 and 3 shows some bangles, an Egyptian necklace and a carved cat, all of which were assumed to be genuine ivory dating from the seventeenth century or later; in some cases, the articles were found to be modern imitations. The case of the carved cat is very significant, as the Raman spectra show the presence of calcite that has been added to a polymer composite of poly(methyl methacrylate) and polystyrene to simulate the density of true elephant ivory. This specimen could not therefore be three hundred years old as these polymers have only been known in the last fifty years! The FT-Raman spectra of true ivories from different mammal sources are shown in Figure 4 and of the fake specimens in Figure 5; the spectroscopic differences are clearly discernible and provide a means of identification between fake and real specimens. Scrimshaw is a special name for carved whalebone and teeth of the sperm whale; many objects in museum collections date from the eighteenth-century production of carved decorative artwork in this material by whaling sailors for their families ashore. Genuine scrimshaw is now extremely valuable and items have been created to deceive the unwary.

Figure 1 Roman die, ca. AD 300, from archaeological excavations at Frocester Villa, Gloucester, UK. Raman spectroscopy has suggested the origin of the die as sperm whale ivory. (See Colour Plate 1).

Figure 2 Selection of ornamental jewellery consisting of three bangles assumed to be ivory but which were shown spectroscopically to be composed of modern resins, and a genuine ivory necklace. (See Colour Plate 2).

A major problem in the characterization of the blue, white and green pigments resulted from a swamping of the pigment Raman signals by the silica in the intact glazes. Results for the red and yellow specimens were obtained from fractured samples. Biomaterials

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 5

Figure 3 ‘Ivory’ cat, which was identified spectroscopically as a modern limitation composed of poly(methyl methacrylate) and polystyrene resins with added calcite to give the texture and density of ivory. Reproduced with permission from Edwards HGM and Farwell DW, Ivory and simulated ivory artefacts: Fourier-transform Raman diagnostic study, Spectrochimica Acta, Part A, 51: 2073–2081 © 1995, Elsevier Science B. V. (See Colour Plate 3).

Spectroscopy has played a role in the characterization of genuine scrimshaw. Figure 6 shows a stackplot of Raman spectra of carved solid and hollow sperm whale-tooth scrimshaw specimens, a staybusk,

and a spill vase/quill pen holder; from these spectra the whale teeth and staybusk are confirmed to be genuine, but the spill vase/quill pen holder is identified as a modern fake made from polymer resin. Similarly, infrared and Raman spectroscopic studies of ancient textiles are being undertaken to derive information about the possible processes of their degradation under various burial environments, e.g. Egyptian mummy wrappings, silk battle banners, Roman woollen clothing and artistic linens from funerary depositions. A classic, topical and ongoing controversy covers the spectroscopic studies associated with the Shroud of Turin and the nature of the pigmented areas on the ancient linen. Ancient technologies and cultures often used naturally occurring biomaterials as decorative pigments and as functional repairing agents on pottery and glass. Recent studies have centred on the provision of a Raman spectroscopic database for native waxes and resins, which has been used for the nondestructive evaluation of archaeological items, e.g. Dragons blood, and it is possible to identify different sources of this material from the spectra. FT-IR spectroscopy, too, has been successful in the characterization of conifer resins used in ancient amphorae and has been particularly useful for attribution of the Baltic geographical origins of amber jewellery through its succinic acid content and for the detection of modern fakes made from phenol formaldehyde resins that are often passed off as examples of ancient ethnic jewellery. The identification of dammar and mastic resins that have been used as spirit-soluble varnishes on

Figure 4 FT-Raman spectra of true ivory; 1064 nm excitation, 500 spectral scans accumulated, 4 cm–1 spectral resolution: (a) sperm whale ivory, (b) elephant ivory, (c) walrus ivory.

6 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

Figure 5 FT-Raman spectra of fake ivory specimens; conditions as for Figure 4: (a) carved Victorian bangle, (b) large bangle, (c) small bangle, (d) cat. The absence of the characteristic proteinaceous features in true ivory near 1650 and 1450 cm–1 and the strong phosphate mode near 960 cm–1 should be noted. Also, the presence of the aromatic ring bands at 3060, 1600 and 1000 cm–1 in (b) and (d) indicate a polystyrene resin content, while the carbonyl stretching band at 1725 cm–1 in all fake specimens indicates the presence of poly(methyl methacrylate). In the cat specimen, the band at 1086 cm–1 uniquely identifies a calcite additive in the specimens of imitation ivory studied. Reproduced with permission from Edwards HGM and Farwell DW, Ivory and simulated ivory artefacts: Fourier-transform Raman diagnostic study, Spectrochimica Acta, Part A, 51: 2073–2081 © 1995, Elsevier Science B.V.

Figure 6 FT-Raman stack-plot spectra of scrimshaw specimens: (a) hollow sperm whale tooth, (b) solid sperm whale tooth, (c) whalebone staybusk and (d) spill vase/quill pen holder. Minor spectroscopic differences confirm the whalebone origin of the staybusk. The modern resin composition of the spill vase/quill pen holder is also unambiguously identified from the aromatic ring stretching bands at 3060 cm−1 and 1600 cm−1. Reproduced with permission from Edwards HGM, Farwell DW, Sedder T and Tait JKF, Scrimshaw: real or fake? An FT-Raman diagnostic study, Journal of Raman Spectroscopy, 26: 623–628 © 1995, John Wiley and Sons Ltd.

paintings and on early photographic prints, and the conservation of the latter, in particular, is very important because of the embrittlement of the substrate due to exposure to light and humidity changes.

Raman spectroscopy has been used to identify wax coatings on early photographs from the American Civil War (ca. 1865) that are showing evidence of deterioration in museum collections.

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 7

Paintings

The use of infrared and Raman spectroscopies for the characterization of paint pigments depends on two critical factors: first, the ability to record goodquality spectra nondestructively from small paint flakes or chips, and, second, the provision of a database of mineral, natural and synthetic pigments on which the basis of a comparison and attribution can be made. Some examples of the success of vibrational spectroscopic methods will now be given to illustrate the potential of these techniques for pigment analysis, and the conservation of degraded media. Several databases now exist for the comparison of unknown pigments with contemporary specimens. A most useful basis for the attribution of specimens is provided by synthetic pigments, for which the dates of first manufacture or use are well established. For example, titanium(IV) oxide (TiO2) is the most important white pigment in use today; its two major naturally occurring forms are rutile and anatase, which have been in use as paint media since 1923 and 1947, respectively. Their identification by vibrational spectroscopy is unambiguous; hence, if either is found on a disputed Renaissance painting, it could indicate a forgery or at least a recently restored work. It is interesting that the distinction between these two forms of TiO2 is not achievable with X-ray fluorescence techniques normally used for pigment analysis in art. A rather different example is provided by the work of Guys, who painted social life in France between 1843 and 1860. Seventy samples taken from 43 of Guys works revealed a heavy dependence on newly synthesized and experimental pigments at that time, including Prussian blue, Cobalt blue and French ultramarine, sometimes in admixture and unique to an artist of his time. Restoration, therefore, needs careful attention to unusual combinations of experimental pigments. A similar study emerges from the FT-IR study of Morisot's work, Bois de Boulogne; this also showed the presence of novel combinations of pigments, but ones that were more stable than others that were in use at that time. Analysis of paint flakes from the Virgin and Child, a suspected fifteenth-century work, revealed the presence of expected organic binders and varnishes but also suggested that a very heavy reworking had taken place more recently, and that the work could be a forgery. On the Jónsbók Icelandic manuscript, six pigments were identified by Raman microscopy, only bone white being indigenous to Iceland; the others, vermilion, orpiment, realgar, red ochre and azurite would all have been imported.

The most important naturally occurring and synthetic inorganic pigments used over periods of time and which feature in paintings from medieval to contemporary ages are given in Tables 1 to 7 along with the year of manufacture or documented use; the tables are constructed according to colour or type, viz. blue, black, brown/orange, green, red, white and yellow. These tables form the basis of dating of paintings and manuscripts by vibrational spectroscopy. The stability of inorganic mineral pigments relative to fugative organic dyes was realized in mediaeval times. The principal dyes used by medieval dyers were indigo from woad for blue, alizarin and purpurin from madder for red, and luteolin from weld or crocetin from saffron for yellow; some had been used long before the Middle Ages and weld was known in the Stone Age. Organic pigments have also been used on manuscripts, notably saffron, weld, indigo, woad, Tyrian purple, madder and carmine. Many natural products were extracted from lichens in the past and used to dye textiles. Notable among these were orchil for purple and crottle for brown. Other dye plants can yield greens, browns and blacks (in the last case, for example, marble gall from oak Quercus trees with added iron sulfate). Organic pigments are prone to both fluorescence and photochemical degradation. Moreover, they often scatter only very weakly, perhaps owing to the fact that they may have been made into Colourfast lakes with a mordant such as alum; hence they are difficult to identify uniquely on a manuscript owing to the lack of concentration of pigment at the sampling point. The better known organic dyes and pigments are listed in Table 8. The Raman and infrared spectra of organic pigments and dyes of relevance to artwork have now been recorded, including those of modern dyes such as methyl blue (a synthetic triarylmethane dye), methyl violet and perylene reds. Perhaps the greatest advances in recent years in the field of nondestructive pigment identification in art works have come from historiated manuscripts that have been studied using Raman microscopy. In situ analyses of the brightly coloured initials and borders of the fourteenth-century Icelandic Jónsbók, Chinese manuscripts, illuminated Latin texts and early medieval Bibles have demonstrated the power of the technique. From samples that often represented only a 20 µm fragment that had fallen into the bindings of early manuscripts, FT-IR and Raman microscopy have made considerable advances in the knowledge of colour hue technology in medieval times; for example, the different blues in an historiated initial could be attributed to a finer particle size

8 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

Table 1

Blue inorganic pigments

Pigment Azurite Cerulean blue Cobalt blue Egyptian blue Lazurite (from lapis lazuli) Manganese blue Phthalocyanine blue (Winsor blue) Posnjakite Prussian blue Smalt Verdigris a

Formula 2CuCO3⋅Cu(OH)2 CoO⋅nSnO2 CoO⋅Al2O3 CaCuSi4O10 Na8[Al6Si6O24]Sn

Date / Sourcea Mineral 1821 1775 3rd millennium BC/Mineral Mineral/1828

Ba(MnO4)2+BaSO4 Cu(C32H16N6) CuSO4·3Cu(OH)2·H2O Fe4[Fe(CN)6]3·14–16H2O CoO·nSiO2(+K2O+Al2O3) Cu(O2CCH3)2·2Cu(OH)2

1907 1936 Mineral 1704 ca. 1500 Mineral

The pigment is either specified to be a mineral and/or the date of its manufacture is listed.

Table 2

Black inorganic pigments

Pigment Ivory black a Lamp black Magnetite Mineral black Vine black a

Chemical name Basic copper(II) carbonate Cobalt(II) stannate Cobalt(II)-doped alumina glass Calcium copper(II) silicate Sulfur radical anions in a sodium aluminosilicate matrix Barium manganate(VII) sulfate Copper(II) phthalocyanine Basic copper(II) sulfate Iron(III) hexacyanoferrate(II) Cobalt(II) silicate Basic copper(II)

Chemical name Calcium phosphate + carbon Amorphous carbon Iron(II,III) oxide Aluminium silicate + carbon (30%) Carbon

Formula Ca3(PO4)2 + C + MgSo4 C Fe3O4 Al2O3·nSiO2 + C C

Date / Source 4th century BC ? ~ 3000 BC Mineral Mineral Roman

Bone black is similar to ivory black.

Table 3

Brown/orange inorganic pigments

Pigment Cadmium orange Ochre (goethite) Sienna (burnt) a Bone black is similar to ivory black. Table 4

Chemical name Cadmium selenosulfide Iron(III) oxide hydrate Iron(III) oxide

Formula

Date / Source

Cd(S,Se) or CdS (>5 µm) Fe2O3·H2O + clay, etc. Fe2O3 + clay, etc.

Late 19th century Mineral Antiquity?

Green inorganic pigments

Pigment Atacamite Chromium oxide Cobalt green Emerald green Green earth – a mix of celadonite and glauconite

Chemical name Basic copper(II) chloride Chromium(III) oxide Cobalt(II) zincate Copper(II) arsenoacetate Hydrous aluminosilicate of magnesium, iron and potassium

Malachite Permanent green deep Phthalocyanine green Pseudo-malachite Verdigris (basic)

Basic copper(II) carbonate Hydrated chromium(III) oxide + barium sulfate Copper(II) chlorophthalocyanine Basic copper(II) phosphate Basic copper(II) acetate

Cu(C32H15ClN8) Cu3(PO)4)2·2Cu(OH)2 Cu(C2(C2H3O2)2·2Cu(OH)2

Viridian

Hydrated chromium(III) oxide

Cr2O3·2H2O

Table 5

Date / Source Mineral Early 19th century 1780 1814 Mineral

Mineral Latter half of 19th century 1938 Mineral Mineral and synthetic (BC) 1838 (?1850)

Red inorganic pigments

Pigment Cadmium red Chrome red Litharge Realgar Red lead (minimum) Red ochre Vermilion (cinnabar)a a

Formula CuCl2·3Cu(OH)2 Cr2O3 CoO·nZnO Cu(C2H3O2)2·3Cu(AsO2)2 Variations on K[(AlIII⋅FeIII)(FeII⋅MgII)] (AlSi3⋅Si4)O10(OH)2 CuCO3·Cu(OH) Cr2O3·2H2O+BaSO4

Chemical name Cadmium selenide Basic lead(II) chromate Lead(II) oxide Arsenic(Ii) sulfide Lead(II,IV) oxide Iron(III) oxide + clay + silica Mercury(II) sulfide

Limited lightfastness (→ black form).

Formula CdSe PbCrO4·Pb(OH)2 PbO As2S2 Pb3O4 Fe2O3·H2O + clay + silica HgS

Date / Source ca. 1910 Early 19th century Antiquity Mineral Antiquity Mineral Mineral and synthetic (13th century)

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 9

Table 6

White inorganic pigments

Pigment

Chemical name

Formula

Date / Source

Anatase Barytes Bone white Chalk (whiting) Gypsum Kaolin Lead white Lithopone Rutile Zinc white

Titanium(IV) oxide Barium sulfate Calcium phosphate Calcium carbonate Calcium sulfate Layer aluminosilicate Lead(II) carbonate (basic) Zinc sulfide and barium sulfate Titanium(I) oxide Zinc oxide

TiO2 BaSO4 Ca3(PO4)2 CaCO3 CaSO4·2H2O Al2(OH)4Si2O5 2PbCO3·Pb(OH)2 ZnS + BaSO4 TiO2 ZnO

1923 Mineral Antiquity Mineral Mineral Mineral Mineral and synthetic (500–1500 BC) 1874 1947 1834

Table 7

Yellow inorganic pigments

Pigment

Chemical name

Formula

Date / Source

Barium yellow Cadmium yellow

Barium chromate Cadmium sulfide

BaCrO4 CdS

Chrome yellow Cobalt yellow (aureolin) Lead antimonate yellow Lead tin yellow

Lead(II) chromate Potassium cobaltinitrite Lead(II) antimonate Lead(II) stannate

Massicot Ochre Orpiment Strontium yellow Zinc yellow

Lead(II) oxide Geothite + clay + silica Arsenic(II) sulfide Strontium chromate Zinc chromate

PbCrO4 or PbCrO4·2PbSO4 K3[Co(NO2)6] Pb2Sb2O7 or Pb3(SbO4)2 [1] Pb2SnO4 [1] PbSn0.76Si0.24O3 PbO Fe2O3·H2O + clay + silica As2S3 SrCrO4 ZnCrO4

Early 19th century Mineral (greenockite) + synthetic ca. 1845 1809 1861 Antiquity Antiquity ? Antiquity ? Antiquity Mineral (and synthetic) Mineral Early 19th century Early 19th century

Table 8

Organic pigments and dyes

Colour

Pigment

Formula / Composition

Origin (Date)

Blue Black Brown

Indigo Bitumen Sepia Van Dyck brown

Indigotin C16H10N2O2 Mixture of hydrocarbons Melanin Humic acids

Plant leaf (BC), synthetic (1878) (BC) Ink of cuttlefish (ca. 1880) Lignite containing manganese (16th century?)

Green

Sap green

Allomelanins Organic dye

Purple

Tyrian purple

Red

Carmine Madder

Yellow

Permanent red Gamboge Hansa yellow Indian yellow

Quercitron Saffron Weld

6,6′-Dibromoindigotin, C16H8Br2N2O2 Carminic acid, C22H20O13 Kermesic acid, C16H10O8 Alizarin, C14H8O4 Purpurin, C14H8O5 Various azo dyes α- and β-Gambogic acids C38H44O8 and C29H36O6 Various azo dyes Magnesium salt of euxanthic acid MgC19H16O11·5H2O Quercitrin C21H20O11 Crocetin C20H24O4 Luteolin C15H10O6

Buckthorn berry (14th century?), coaltar dye Marine mollusc (1400 BC), synthetic (1903) Scale insect, cochineal (Aztec) Scale insect, kermes (antiquity) Madder root (3000 BC) Synthetic alizarin (1868) Synthetic (after 1856) Gum-resin (before 1640) Synthetic (1900) Cow urine (15th century)

Inner bark of Quercus oak Crocus flower stigma (antiquity) Plant foliage (Stone Age)

Tables 1–8 are reproduced with some modifications from Clark RJH, Raman microscopy: application to the identification of pigments on medieval manuscripts, Chemical Reviews, 187–196. 1995 © The Royal Society of Chemistry.

10 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

and not to a dilution with other materials or colours. Useful information has also been provided about additives and binding media. On some paintings and manuscripts, an inappropriate blend of adjacent colours or mixtures of pigments and binders was achieved; Raman microscopy has identified two examples of these in cadmium sulfide and copper arsenoacetate (which yields black copper sulfide) and egg tempera with lead white (which yields black lead sulfide). The term inappropriate here refers to the instability of pigments and pigment mixtures resulting in a chemical reaction over periods of time and through aerial or substratal influences. Where the pigment is coloured, the choice of exciting line for Raman spectroscopy is extremely important because absorption of the scattered light by the sample may affect the spectrum. In such cases, the exciting line is chosen to fall outside the contour of the electronic absorption bands of the pigment. Vermilion (HgS) and red ochre (Fe2O3) give poor-quality Raman spectra using green excitation but give strong Raman spectra when excited with red radiation. An interesting example of the effects due to absorption of laser radiation is provided by the Raman spectra of red lead (Pb3O4) obtained using 514.5 and 632.8 nm excitation (Figure 7). In each case, well-defined Raman spectra are obtained, but only with 632.8 nm excitation is the spectrum that of the genuine material; that obtained with 514.5 nm excitation matches that of massicot (PbO). These observations may be explained because red lead can be converted into massicot by heat. Red lead absorbs 514.5 nm radiation strongly, leading to localized heating that results in conversion of the irradiated particle or particles into massicot. Since 632.8 nm radiation is not absorbed by the red lead, there is minimal local heating with this exciting line and thus no decomposition. For coloured pigments, the absorption of exciting radiation can be used to advantage to produce enhanced Raman scattering through the resonance Raman effect. This is particularly useful for weakly scattering species, and selection of excitation within the contour of an electronic absorption band of a chromophore can produce a spectral intensity several orders of magnitude larger than that obtained in conventional Raman spectra. A classic example of this is provided by the deep blue mineral lapis lazuli, Na8[Al6Si6O24]Sn, where the species S2− and S3− are responsible for the colour. The species S3−, although present to an extent of < 1% in the pigment, gives such an intense Raman spectrum with green-red excitation that no bands due to the host lattice are observed. Other techniques fail to discriminate the presence of S3− in the aluminosilicate lattice.

Figure 7 Illustration of the effect of wavelength of laser excitation on the Raman spectra of a pigment, red lead (Pb3O4), excited with (a) 514.5nm and (b) 632.8nm radiation. The genuine spectrum is (b); the spectrum excited by green radiation in (a) corresponds to massicot (PbO), converted from Pb3O4 by localized heating in the laser beam. Reproduced with permission from Bert SP, Clark RJH and Withnall R, Non-destructive pigment analysis of artefacts by Raman microscopy, Endeavour, New Series, 16: 66–73 © 1992, Elsevier Science.

The Raman spectrum is sensitive to both composition and crystal form, as is demonstrated by titanium(IV) dioxide, the most important white pigment in use today. White pigments normally present considerable problems in pigment identification, comprising sulfates, carbonates and phosphates, which are easily distinguished in Raman spectroscopy (Figure 8). The symmetric vibration of the anion gives rise to an intense band at ~1000 cm1 for sulfates, ~1050 cm1 for carbonates and ~960 cm1 for phosphates. The exact wavenumbers of these bands are also sensitive to the cation (1050 cm1 for PbCO3 and 1085 cm1 for CaCO3). This factor becomes extremely important in database construction since artistic vocabulary generally describes the colour rather than a precise mineral origin, e.g. cadmium yellow, although strictly CdS has also been designated for organic substitutes of similar hue. Old recipes for obtaining pigment colours are often vague and employ unidentifiable materials; the same chemical compound or mixture can even have different names according to geographical locality or historical period. For example, three contemporary samples designated Naples Yellow, assumed to be

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 11

Table 9 Bible

The Raman bands of inorganic pigments in the Lucka

Pigment

Raman bands (cm−1)

Azurite

248w, 404vw, 770m, 838vw, 1098m, 1424w, 1578w

Lapis lazuli 259w, 549vw, 807w, 1096m, 1355vw, 1641w White lead 1054s Malachite

225w, 274w, 355w, 437m, 516vw, 540w, 724w, 757vw, 1064w, 110w, 1372w, 1498m

Orpiment

136m, 154m, 179w, 203m, 293s, 311s, 355s, 384m, 587vw

Realgar

124vw, 143m, 166w, 172w, 183s, 193s, 214w, 222s, 329w, 345m, 355s, 370w, 376w

Red lead

121vs, 152m, 223w, 232w, 313w, 391w, 477w, 549s

Vermilion

254s, 281w, 344m

Figure 8 Raman spectra of white pigments; differentiation between (a) lead white (PbCO3) (b) chalk (CaCO3) and (c) bone white (Ca3(PO4)2). Reproduced with permission from Best SP, Clark RJH and Withnall R. Nondestructive pigment analysis of artefacts by Raman microscopy, Endeavour, New Series 16: 66–73 © 1992, Elsevier Science.

Pb2 (SbO4)2 a lead antimonate have been shown by Raman spectroscopy to be a lead antimony oxide, Pb2Sb2O6, another oxide, Pb2Sb2O3, and the third sample mainly Pb2(CrO4)2, but also containing PbCO3. None of the samples tested actually proved to be Naples Yellow of the assumed formula!

Medieval manuscripts Most pigments in medieval manuscript illuminations are inorganic. They are coated with a small quantity of binding material and deposited on parchment that has previously been rubbed with pumice. The small size of the illuminations limits the number of spectroscopic samplings. The blue colour in a set of six French manuscripts from the twelfth century was studied in different kinds of initial historiated (with a descriptive scene), decorated, or simply coloured which showed variable hues of pure blue, grey and sometimes violaceous. Raman microspectra obtained from all of the samples show common features: they are dominated by a strong band at 548 cm −1, with

weaker bands at 260, 585 and 1098 cm −1. This group of four bands is characteristic of ultramarine blue and reveals the presence of only this pigment. In the blue-grey samples, microscopic observation showed tiny black particles that were identified by their Raman spectra as graphite. Ultramarine has been identified in a manuscript commentary on Ezekiel, ca. AD 1000, in the abbey of St Germain, Auxerre, France. The pigment ultramarine is first mentioned in written sources in the thirteenth century. This discovery established for the first time that the pigment was in fact introduced into Europe almost two centuries earlier. The manuscript was to reveal more fascinating details. Part of the dedication image at the beginning of the manuscript is a figure representing the Abbot, Heldric, kneeling before St Germain, who died about 448 and is buried in the abbey. From the spectrum of the pigments, it was established that the Abbot's blue garments were painted with indigo (woad) while only the patron saint's more glorious garment was covered with the rare and expensive lapis lazuli, thus emphasizing his importance. The letter I in the Lucka Bible at the beginning of the Book of Genesis displays seven scenes representing the seven days of Creation. The varied palette (Table 9) has been well characterized using Raman microscopy (Figure 9). The Skard copy of the Jónsbók Icelandic Law Code dates from ca. 1360 and is one of the most outstanding examples of an Icelandic medieval manuscript. The pigments have been analysed by Raman microscopy; three historiated initials and about 15 illuminated initials with associated background paintings and embellishments have been examined. Six pigments were identified unambiguously by

12 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

Figure 9 Raman spectra of the historiated letter ‘I’ from the Book of Genesis, Lucka Bible, from which the data and identification of the mineral pigments in Table 9 have been derived. Reproduced with permission of Clark RJH, Raman microscopy: application to the identification of pigments on medieval manuscripts, Chemical Reviews, 187–196 © 1995, The Royal Society of Chemistry.

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 13

Figure 10 The historiated initials (A) ‘P’ and (B) ‘S’ from the Icelandic Jónsbók with the Raman spectra of pigments contained therein. The combination of vermilion and red ochre in the ‘P’ should be noted. The spectrum of bone white has been obtained from the letter ‘H’ (not shown). Reproduced with permission from Best SP, Clark RJH, Daniels MAM, Porter CA and Withnall R (1995). Identification by Raman microscopy and visible reflectance spectroscopy of pigments on an Icelandic manuscript. Studies in Conservation 40: 31–40. (See Colour Plate 4.)

14 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

Figure 11 Elaborately historiated initial ‘R’ in sixteenth-century German choir book, with Raman microscopy spectra of selected pigmented regions. Reproduced with permission of Clark RJH (1995) Raman microscopy: application to the identification of pigments on medieval manuscripts. Chemical Society Reviews 187–196 © The Royal Society of Chemistry. (See Colour Plate 5a).

Raman microscopy: vermilion, orpiment, realgar, red ochre, azurite and bone white. Neither red lead nor lead white, pigments commonly used in Northern Europe, was identified on the manuscript: this raises questions as to the availability of these pigments in Iceland, to which they are not native, and of the effectiveness of the trading routes. Examples of two historiated initials and the Raman spectra of the pigments used are shown in Figure 10. An elaborately historiated initial R on a sixteenth-century German choir book has been studied by Raman microscopy and no fewer than eight pigments have been identified, namely, azurite, lead tin yellow, malachite, vermilion, white lead, red lead, carbon and massicot (Figure 11). Raman microscopy demonstrated that in this case the two shades of blue on the garments of the right-hand figure, although appearing to be due to different pigments, arise from the same pigment, azurite, the illuminator having used less azurite and proportionately more binder to produce the lighter

shade without choosing to add any white pigment, such as lead white, to gain the same effect. The azurite used in the deeper blue robe arises from coarse grains of pigment (~30 µm diameter) whereas the lighter blue used in the undergarment arises from fine grains (~3 µm diameter). The effect of the particle size on the depth of colour of a powder is a common feature of powdered materials, whose colour is determined by diffuse reflectance. As the particle size is reduced, the average depth to which radiation penetrates before being scattered is also reduced and hence the depth of colour is reduced. The angels wing and podium are painted in malachite, which is a basic copper carbonate similar to azurite. Both azurite and malachite have a similar provenance in that they are both associated with secondary copper ore deposits, but their spectra are very different. It is of particular interest that the dark grey colour of the pillar top is not obtained via a single pigment but by colour subtraction of a mixture of at least

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 15

seven pigments: white lead, carbon, azurite, as well as small amounts of vermilion, red lead, massicot and lead tin yellow type I. The mixture is evident at ×100 magnification by Raman microscopy, whereby each individual pigment grain may be identified (Figure 12).

Wall paintings The application of FT-Raman spectroscopy to the study of red pigments from English early medieval wall paintings in the Chapter House of Sherborne Abbey and the Holy Sepulchre Chapel in Winchester Cathedral (Figure 13) has been reported. The complexity of the spectra can be observed in Figure 14, where the pigment and substratal features are clearly differentiated. The operations of the two monastic houses were very different, as the Sherborne Abbey pigment consisted of pure vermilion, whereas that of Winchester Abbey was a 3:1 mixture of red ochre and vermilion. In both cases, spectral features due to sandstone and marble could be seen, but no identifying feature could be ascribed to a plaster substrate. The biodeterioration of the Renaissance frescoes in the Palazzo Farnese, Caprarola, Italy, arising from aggressive colonization by the lichen Dirina massiliensis forma sorediata has been studied using Raman microscopy. Over 80% of the masterpieces painted by Zuccari in 1560 have now been destroyed. The role of the lichen metabolic products and the ability

Figure 12 Portion of top of column of historiated initial ‘R’ shown in Figure 11; the individual pigment grains can be clearly seen under the ×100 magnification and can be separately identified using Raman microscopy. Reproduced with permission of Clark RJH (1995) Raman microscopy: application to the identification of pigments on medieval manuscripts. Chemical Society Reviews 187–196 © The Royal Society of Chemistry. (See Colour Plate 5b).

Figure 13 Holy Sepulchre Chapel, Winchester Cathedral. Wall painting of ca. 1175–85 on the east well depicting the Deposition, Entombment, Maries at the Sepulchre and the Harrowing of Hell. Reproduced with permission from Edwards HGM, Brooke C and Tait JKF, An FT-Raman spectroscopic study of pigments or medieval English wall paintings, Journal of Raman Spectroscopy 28: 95–98, © 1997 John Wiley and Sons Ltd. (See Colour Plate 6).

of primitive organisms to survive hostile environments generated by large concentrations of heavy metals such as mercury, lead and antimony are now being understood as a result of these studies, which will assist conservators in future projects. The red pigment on the Carlovingian frescoes in the crypt of the Abbey of St Germain at Auxerre shows up to 13 layers of paint identified by Raman microscopy, and vermilion and iron(II) oxide have both been found. The North American palaeo-Indian shelters at Seminole Canyon, at the confluence of the Rio Grande and Devils Rivers, have provided the Raman spectra of the oldest cave paintings yet studied. From a limited palette, these 3500-year-old pictographs have yielded red, white and black colours identified

16 ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY

Figure 14 FT-Raman spectra of twelfth-century wall paintings in (A) Sherborne Abbey ((a) red pigment, (b) vermilion, (c) unpainted stone, (d) calcite) and (B) Winchester Cathedral ((a) red pigment, (b) vermilion, (c) red ochre). Reproduced with permission from Edwards HGM, Brooke C and Tait JKF, An FTRaman spectroscopic study of pigments on medieval English wall paintings, Journal of Raman Spectroscopy 28: 95–98, © 1997 John Wiley and Sons Ltd.

as red ochre, calcium oxalate monohydrate (whewellite) and manganese(IV) oxide, respectively. In addition, the black pigment showed traces of an organic additive recently identified from DNA profiling as bison or deer bone marrow, presumably added in a symbolic ritual. FT-IR spectroscopy, too, has had considerable success in the identification of pigments and associated materials in paintings and manuscripts. In particular, the use of infrared microspectroscopy has proved to be an essential prerequisite for the scientific examination of complex pigment mixtures. Generally, however, inorganic mineral pigments are not well characterized in the infrared because of their low band wavenumbers; hydroxyl groups in hydrated minerals cause problems in absorption and

the substrates themselves, e.g. cellulose, linen or starch, often give rise to broad backgrounds. IR spectroscopy, therefore, is much more applicable to organic pigments, binders and mixtures. One project dealing with medieval manuscripts has had as its objective the application of smallparticle-analysis techniques to the study of pigments in medieval Armenian and Byzantine manuscripts. At the University of Chicago, 10 decorated manuscripts were sampled (Table 10). These manuscripts represent a broad chronological span ranging from the tenth century to the post-Byzantine era (sixteenth century or later). Interpretation of IR spectra was complicated by the fact that the pigment samples often contained several components in addition to the pigment. These additives often served as binding media thickening agents or extenders, for example. Because the amount of pigment is small relative to the amounts of these nonpigment components, the spectral bands of the pigment are often difficult to distinguish from interfering spectral bands of nonpigment components. Chromatographic separation prior to spectroscopic analysis was not feasible because of the insolubility and limited quantity of the samples. Spectral subtraction has been used in many cases as a means to separate the components. The infrared spectrum of a yellow pigment sample from manuscript (MS) 46 (Haskell gospels) shows several bands that can be attributed to egg yolk, a typical medium used in mediaeval times, as well as calcium carbonate and kaolin. The absorption bands at 3382, 3298 and 1600 cm 1 suggested that a primary amine is present. Putrescine (1,4-diaminobutane), an amine found in decaying proteinaceous matter, has similar spectral features. Because parchment (being of animal skin origin) would be subject to putrefaction if not properly preserved, this amine could be a product of bacteriological degradation.

Table 10 University of Chicago Special Collections manuscripts analysed by FT-IR spectroscopy

Manuscript number 972 1054

Name Archaic Mark Elfleda Bond Goodspeed Gospels

965

Rockefeller-McCormick New Testament

131

Chrysanthus gospels

232

Greek gospels

Haskell gospels

129

Nicolaus gospels

727

Georgius gospels

879

Lectionary of Constantine the reader

948

Lectionary of St Menas the wonder worker

ART WORKS STUDIED USING IR AND RAMAN SPECTROSCOPY 17

Although the pigment was not identified in this case, the information that egg tempera is present as the binding medium is valuable; the binding medium in MS 965 is hide glue. Pigment components have also been identified from their IR spectra. The infrared spectrum obtained from a purple pigment removed from MS 965 showed bands due to ultramarine blue and white lead. A red pigment (cochineal?) must also be present with the blue to produce the purple colour. In MS 972 (Archaic Mark), the presence of Prussian blue from the infrared spectrum indicates possible restoration because Prussian blue, a synthetic dye, was not available until 500 years later. Another example of the use of FT-IR spectroscopy in the authentication of paintings is provided by the analysis of the Virgin and Child, a fifteenth-century Italian painting on panel. There were already stylistic doubts and, although IR spectroscopy confirmed the presence of egg tempera as a binding medium, the blue pigment was not azurite or lapis lazuli. A prominent infrared absorption at 2091 cm1 identified the blue pigment on the Virgin's robe as Prussian blue, unknown before 1704. Hence, the painting has either been very heavily reworked or is possibly a nineteenth-century copy of an earlier work. See also: Colorimetry, Theory; Dyes and Indicators, Use of UV-Visible Absorption Spectroscopy; Fourier Transformation and Sampling Theory; Rayleigh Scattering and Raman Spectroscopy, Theory; Vibrational

CD Spectrometers; Vibrational CD, Applications; Vibrational CD, Theory.

Further reading Adelantado JVG, Carbo MTD, Martinez VP and Reig BF (1996) FTIR spectroscopy and the analytical study of works of art for purposes of diagnosis and conservation. Analytica Chimica Acta 330: 207215. Best SP, Clark RJH and Withnall R (1992) Non-destructive pigment analysis of artefacts by Raman microscopy. Endeavour 16: 6673. Clark RJH (1995) Raman microscopy: application to the identification of pigments on mediaeval manuscripts. Chemical Society Reviews 187196. Coupry C and Brissaud D (1996) Applications in art, jewellery and forensic science. In Corset J and Turrel G. (eds) Raman Microscopy: Developments and Applications, pp 421453. London: Academic Press. Edwards HGM and Farwell DW (1995) FT-Raman spectroscopic study of ivory: a nondestructive diagnostic technique. Spectrochimica Acta Part A 51: 20732079. Hoepfner G, Newton T, Peters DC and Shearer JC (1983) FTIR in the service of art conservation. Analytical Chemistry 55: 874A880A. Katon JE, Lang PI, Mathews TF, Nelson RS and Orna MV (1989) Applications of IR microspectroscopy to art historical questions about mediaeval manuscripts. Advances in Chemistry Series 220: 265288. Learner T (1996) The use of FTIR in the conservation of 12th century paintings. Spectroscopy Europe 1419. Mills JS and White R (1994) The Organic Chemistry of Museum Objects, 2nd edn. London: ButterworthHeinemann.

Astronomy, Applications of Spectroscopy See Interstellar Molecules, Spectroscopy of; Stars, Spectroscopy of.

18 ATMOSPHERIC PRESSURE IONIZATION IN MASS SPECTROMETRY

Introduction Ion sources for atmospheric-pressure ionization (API) in mass spectrometry (MS) were first described in 1958 by Knewstubb and Sugden. The work of the research group of Horning and Carroll in the late 1960s and early 1970s resulted in a system commercially available from Franklin GNO Corp. These types of instruments were mainly used in the study of ionmolecule reactions in gases in the atmosphere. Furthermore, in 1974 this type of instrument was applied with an atmospheric-pressure corona discharge ion source in the coupling of liquid chromatography to MS (LC-MS). Despite the promising results, the technique did not attract much attention at that time. Subsequently, several other API instruments were described and built. The most successful commercial instrument was the Sciex TAGA (trace atmospheric gas analyser), while API systems were also available from Extranuclear and Hitachi. In the late 1980s, the TAGA API-MS system was reengineered by Sciex for online LC-MS and was called the API-III. This system was applied by a growing group of scientists, especially within pharmaceutical industries. In the early 1980s, Yamashita and Fenn fundamentally investigated the potential of electrospray ionization in an API source. This research led to another approach to LC-MS using an API system. In 1988, Fenn and co-workers demonstrated that with this electrospray system it was possible to perform the MS analysis of high molecular-mass proteins via multiple charging. As a result of this observation, API-MS went through a period of tremendous growth, the end of which is certainly not yet here. In the mid 1980s, similar API technology was also applied in the coupling of inductively coupled plasmas (ICP) to MS. At present, there are three major application areas of API-MS: air or gas analysis, online LC-MS, and ICP-MS. In this paper, technical and instrumental aspects of

MASS SPECTROMETRY Methods & Instrumentation API-MS are discussed and a number of typical applications are briefly reviewed.

API ion source design An API ion source generally consists of four parts: 1. the sample introduction device, the design of which is highly dependent on the type of application, 2. the actual ion source region, i.e. the region where the ions are generated, 3. an ion sampling aperture, where the ions are sampled from atmospheric pressure into the vacuum system of the MS, and 4. an ion transfer system, where the pressure difference between the atmospheric-pressure source and the high-vacuum mass analyser is bridged and the ions entering through the ion sampling aperture are preferentially transferred and focused into the mass analyser region. The proper design of parts 3 and 4 is of utmost importance, as it determines to what extent the very high ionization efficiency that can be achieved in an API source actually becomes available to the analytical applications. The inevitable ion losses in the sampling and transfer regions should be kept to a minimum.

Ion sampling aperture and ion transfer system An API source system consists of a region where the ions are generated. This region can have a relatively large volume of several litres, although in practice only the ions that come close to the ion sampling aperture are efficiently sampled into the MS. Therefore, in many systems the ions are generated just in front of the ion sampling aperture. A general scheme of an API source is shown in Figure 1. In the early API systems, the ion sampling aperture consisted of a 25200 µm i.d. pinhole in a metal plate. The size of the pinhole is limited by the pumping capacity of the vacuum system of the MS.

ATMOSPHERIC PRESSURE IONIZATION IN MASS SPECTROMETRY 19

Figure 1 Schematic diagram of a typical API source, as for instance used in LC-MS.

In the first prototypes, the pinhole inner diameter was only 25 µm, while in the Sciex TAGA a larger pinhole could be applied, because of the high pumping efficiency of the cryogenic vacuum-pumping system available in this instrument. The sampling of ions through an orifice in a plate or in the tip of a cone is used in many commercial API systems. However, two alternative ion sampling devices based on the transport of ions through capillary tubes have been developed and commercialized. The first, designed by Fenn and co-workers and subsequently commercialized by Analytica of Branford, incorporates a 100150 mm × 0.5 mm ID glass capillary. Both ends of the glass capillary are coated with metal, e.g. silver, gold or platinum, in order to define the potential at both tips. The use of an insulating glass capillary enables independent control over the potentials applied in the atmospheric-pressure ion source region and in the ion acceleration region inside the vacuum. The second alternative to the ion sampling orifice is a heated stainless steel capillary, initially proposed by the group of Chait and subsequently commercialized by Finnigan. In the strong cooling that takes place upon the expansion of a gas into a vacuum chamber, as is taking place in an electrospray source, the ions act as condensation nuclei for water and solvent molecules. As a result, clusters of analyte ions and numerous solvent molecules are formed. This cluster formation can be counteracted or avoided in two ways: either by preventing the solvent ions from reaching the ion sampling orifice, or by increasing the temperature of the gas, vapour and ion mixture entering the vacuum. By flushing the area just in front of the sampling orifice with a stream of dry nitrogen, the

solvent vapour can be swept away, while the ions are forced to pass through this nitrogen curtain by the application of an electric field between the curtain plate and the orifice plate in Figure 1. An additional benefit of a curtain gas is the prevention of other neutral contaminants, such as particulate material, from entering the ion sampling orifice or capillary. This is important for the ruggedness required for day-to-day use of LC-MS, e.g. in the analysis of large series of biological samples. A curtain or countercurrent gas is applied in the sources built by Sciex, Analytica of Branford and Hewlett-Packard. Alternatively, the temperature of (part of) the ion source or of the ion sampling capillary can be increased to avoid the formation of clusters between analyte ions and solvent molecules. This obviates the need for a curtain gas, but also leaves the ion sampling device unprotected from contamination by particulate material. Ions are sampled from the atmospheric-pressure region and together with nitrogen, which is applied as nebulizing gas as well as curtain gas, entering the vacuum system. The low-pressure side of the ion sampling pinhole or the low-pressure end of the ion sampling capillary acts as a nozzle, where the mixture of gas, solvent vapour and ions expands. The expanding jet is subsequently sampled by means of a skimmer into a second vacuum stage. In most systems, the region between the nozzle and the skimmer, which is pumped by a high-capacity rotary pump, contains an additional lens, e.g. a ring or tube electrode, to preferentially have the ions sampled by the skimmer. From a gas dynamics point of view, it might be important to position the skimmer just in front of the Mach disk of the expansion. However, because ions are preferentially sampled, the positioning of the skimmer relative to the Mach disk does not appear to be very important. In some systems, nozzle and skimmer are actually not precisely aligned in order to reduce the number of neutral molecules in the highly directed flow from the nozzle entering the region with higher vacuum. Optimum ion transfer is achieved by means of a tube lens between nozzle and skimmer. At the lower pressure side of the skimmer a second expansion step takes place. In older systems, the ions entering through the skimmer were extracted, transferred and focused into the mass analyser region by means of a set of conventional flat lenses. Later, it was determined that a more efficient ion transfer and focusing can be achieved by means of an RF-only quadrupole, hexapole or octapole device (cf. Figure 1). These types of ion transfer and focusing devices are especially in use in systems for LC-MS. In other systems, as used for ICP-MS for instance, ion

20 ATMOSPHERIC PRESSURE IONIZATION IN MASS SPECTROMETRY

Figure 2

Schematic diagram of a typical API source for ICP-MS.

focusing is achieved by means of a combination of Einzel lenses and a Bessel box (cf. Figure 2). The source designs described so far are used in combination with a (triple) quadrupole instrument. Modifications of this design are required in the coupling to other types of mass analysers, e.g. the implementation of a gate or ion pulse electrode for use in combination with ion-trap and time-of-flight mass analysers.

Sample introduction devices A wide variety of sample introduction devices are available for gas analysis by API-MS. In the study of atmospheric processes, direct sampling of ionic species and clusters in various atmospheric layers may be performed. For these in situ studies, air is directly sampled into the system, eventually through specially designed sampling tubes. Alternatively, gases may be sampled and mass analysed after ionization by electrons from a corona discharge (similar to APCI). For use in combination with LC-MS and other liquid-introduction techniques, two main types of sample introduction devices are available, i.e. one for electrospray ionization and one for APCI. The devices for APCI consist of a heated nebulizer, which is a combination of a concentric pneumatic nebulizer and a heated quartz tube for droplet evaporation. Additional auxiliary gas is used to flush the evaporating droplets through the vaporizer into the actual ionization region. A schematic diagram of such a device is shown in Figure 3. Reactant ions for APCI are generated in a point-to-plane corona discharge in the atmospheric-pressure ion source. The discharge needle is operated at 35 kV and provides a corona emission current of a few µA.

For electrospray ionization, initially, 100 µm i.d. stainless-steel capillaries, i.e. hypodermic needles, were used for sample introduction. With such a device, the flow-rate is limited to ∼10 µL min−1, which is too high for many biochemical applications and too low for effective LC-MS coupling. Therefore, the initial system was modified. For biochemical applications, the dimensions of the introduction capillary were decreased to a few µm i.d. capillaries, mostly a glass or fused silica. This approach is called nanoelectrospray as it allows the continuous introduction of between 5 and 100 nLmin−1 of a liquid into the API source. Nano-electrospray is especially useful for applications where the sample amount is limited. In this way, it is possible to perform a series of MS experiments with a minute amount of sample, e.g. 30 min of various MS experiments with only 1 µL of sample. For LC-MS applications, the electrospray nebulization process was assisted by pneumatic nebulization, i.e. a nebulization gas at a high linear velocity is forced around the electrospray needle. This approach, introduced by Sciex as ionspray in 1986, was subsequently adopted by other instrument manufacturers and is now the most widely applied

Figure 3 Schematic diagram of a sample introduction device for APCI: a heated nebulizer.

ATMOSPHERIC PRESSURE IONIZATION IN MASS SPECTROMETRY 21

approach to LC-MS via electrospray interfacing. A schematic diagram of a typical sample introduction device for LC-MS via pneumatically-assisted electrospray is shown in Figure 4. Significant research has taken place in the past few years concerning the optimum geometry of the spray device in the atmospheric-pressure ion source. While initially the spray device was positioned axially relative to the ion sampling orifice, an approach which is still in use, other instrument manufacturers use a different setup, e.g. the spray device positioned orthogonally to the ion sampling orifice, or at an angle of 45°, eventually with a perpendicular heated gas probe to assist in droplet evaporation. The two main perspectives in this type of research are the ability to introduce higher liquid flow-rate while still avoiding the risk of clogging the ion sampling orifice, and the ability to introduce samples containing significant amounts of nonvolatile material, e.g. samples from biological fluids after minimum sample pretreatment, or LC mobile phases containing phosphate buffers. While the advantages of the alternative positioning of the spray device is well documented for biological fluids, the evidence for the possibilities of the use of phosphate buffers is less convincingly demonstrated. For use in ICP-MS, ions are sampled directly from the inductively coupled plasma. Special precautions are required related to the high temperatures in the plasma. A schematic diagram of an ICP-MS system is shown in Figure 2. In most cases, the liquid sample, which can be introduced directly from a vial, of a liquid-phase separation technique such as LC is nebulized into a separate spray chamber. The aerosol, containing the smaller droplets, is transported by argon to the torch, where the ICP is generated and sustained. In this so-called ICP flame, the analytes are atomized and ionized. The ions are directly sampled from the flame by the ion sampling aperture for mass analysis.

Figure 4 Schematic diagram of a pneumatically-assisted electrospray sample introduction device.

Ionization in API conditions Ionization in an API source can take place in a variety of ways, depending on the type of applications. The various way are briefly reviewed here. Gas-phase ionization

In initial applications of API sources for gas analysis and monitoring of atmospheric ionic processes, either direct sampling of the ions present in flames or in the atmosphere was applied, or ions were generated by means of electrons from an electron-emitting 63Ni foil or by means of a corona discharge. Currently, the corona discharge is the most widely applied approach for generating ions, in both gas analysis and APCI in LC-MS applications. In an API source equipped with a 63Ni foil, energetic electrons ionize the nitrogen present in a series of consecutive steps:

Similarly, other gases may be ionized. In the presence of traces of water in the ion source, these ionic species react with the water molecules:

The main ionic species generated, i.e. and H3O+.(H2O)n, may enter in ionmolecule reactions, leading to charge exchange of proton transfer reactions, respectively. This is valid for positive-ion acquisition, while negative ions may be generated either by electron-capture processes with electrons or via proton-transfer ion-molecule reactions starting from OH−.(H2O)n for instance. The same reactant ions are formed in positive-ion mode in a point-to-plane corona discharge. Approximately 39 kV is applied at the corona discharge, creating a corona current of 15 µA, depending on the point-to-plane distance and the composition of the vapours in the ion source. Corona discharges are very complicated processes that involve avalanches of ionization reactions by ionmolecule and electronmolecule collisions together with quenching

22 ATMOSPHERIC PRESSURE IONIZATION IN MASS SPECTROMETRY

processes, as discussed in detail by Meek and Craggs. In the positive-ion mode, the ionization of analyte molecules follows the general rules of chemical ionization, i.e. proton transfer may take place when the proton affinity of the analyte molecule exceeds that of the reagent gas, leading to protonated molecules:

while, in addition, various adduct ions and cluster ions with solvent molecules may be generated. In the negative-ion mode, either electron-capture products M− • or deprotonated molecules [MH]− may be generated, although the electron-capture products may further react in ionmolecule reactions to form different ionic species. The positive-ion mass spectra obtained from APCI and conventional medium-pressure CI might show differences due to the fact that the mass spectrum in medium-pressure CI is more determined by the reaction kinetics, while in APCI, due to the longer residences times of ions in the sources as well as the more frequent ionmolecule collisions, the mass spectrum more reflects the thermodynamic equilibrium conditions. Liquid-based ionization

The liquid-based ionization technique in API-MS is applicable in electrospray interfacing as well as with a heated nebulizer introduction under certain conditions, i.e. without switching the corona discharge electrode on, and for certain compounds. Here, the discussion is restricted to electrospray interfacing. In electrospray ionization and interfacing, a high potential, typically 3 kV, is applied to a liquid emerging from a capillary. This causes the liquid to break into fine threads emerging from the so-called Taylor cone formed at the capillary tip. The threads subsequently disintegrate into small droplets. The electrohydrodynamic or Rayleigh disintegration results from autorepulsion of the electrostatically charged surface which exceeds the cohesive forces of surface tension (the Rayleigh limit). In this electrospray nebulization process uniform droplets in the 1-µm diameter range are formed. These charged droplets shrink by solvent evaporation, whereas electrohydrodynamic droplet disintegration again takes place as soon as the Coulomb repulsion of the surface charges exceeds the surface tension forces. The offspring droplets generated in the field-induced droplet disintegration process carry only 2% of the mass of the parent droplet and 15% of its charge.

Analyte ions in these offspring droplets, present as preformed ions, desorb or evaporate into the gas phase and become available for mass analysis. The ion evaporation process, described first by Iribarne and Thomson and generally considered as the most important ionization mechanism in electrospray ionization, is a process which thermodynamically corresponds to bringing a solvated ion from the surface of a charged droplet to infinity. Below a certain size/ charge ratio for a droplet, the Gibbs free energy of solvation of an ionic species present in the droplet will exceed the energy required to bring this species as a solvated ion from infinity to the surface of the droplet. Under these conditions, ion evaporation is possible. According to Iribarne and Thomson, ion evaporation will take place instead of electrohydrodynamic disintegration when the critical ion evaporation droplet radius exceeds the Rayleigh limit. Electrospray ionization is especially effective for analytes that are present as preformed ions in solution, e.g. organic acids or bases that are present as deprotonated or protonated molecules, respectively, at suitable pH values, quaternary ammonium compounds, etc. For biomacromolecules like proteins, which exist in solution as a distribution of ionic species carrying different numbers of charge, electrospray ionization results in a spectrum containing an ion envelope of multiply charged ions (see Figure 5). To a first approximation, this mass spectrum reflects the charge-state distribution of the protein in solution. As the ion separation in a mass analyser is based on mass-to-charge ratio, the multiple charging allows the mass analysis of large molecules within the limited m/z-range of, for instance, a quadrupole mass spectrometer. The observation of multiple charging of proteins in electrospray ionization attracted much attention: all major instrument manufacturers introduced an API system to be used in combination with an electrospray interface. At the same time, it was found that electrospray ionization is not only suitable for the mass analysis of large molecules, but is also a very efficient ionization technique of small polar or ionic molecules, e.g. drugs and their metabolites. In the past few years, hundreds of dedicated API-MS systems equipped with electrospray and APCI interfaces have found their way into many different laboratories, especially within pharmaceutical companies and biochemistry/biotechnology laboratories. Plasma-based ionization

A plasma is an ionized gas that is macroscopically neutral, i.e. an equal number of positive and negative particles are present. In ICP-MS, the gas used to

ATMOSPHERIC PRESSURE IONIZATION IN MASS SPECTROMETRY 23

Figure 5 Ion envelope of multiply charged ions, obtained in the electrospray ionization of haemoglobin. In the spectrum, the m/z and charge state of the ions are indicated. The spectrum consists of two series: one due to a Hb-α chain (Mr 15126 Da) and one due to a Hb-βA chain (Mr 15 867 Da).

generate the plasma is argon, which has a high ionization energy (15.76 eV). The hot argon plasma has the capability to atomize most samples, and excite and ionize most elements of the periodic table. The plasma is generated in a torch. A high-frequency generator, typically operated at 27 or 40 MHz and with a power of 12 kW, is used to produce a highfrequency field through an induction coil, which is positioned at the outside of the torch. In this way, the argon plasma is produced, i.e. the argon is ionized and the plasma is sustained. Liquid samples are nebulized and the aerosol is injected into the plasma for atomization and ionization. The ionized atoms generated in this way are sampled by the ion sampling aperture and mass analysed.

Applications Gas analysis

There are a number of applications of API-MS in gas analysis, e.g. the study of environmental air contamination, human breath analysis, the study of (ionic) processes in the atmosphere, the detection of trace impurities in gases that are important in microelectronic manufacturing processes, or the identification of ions in flames. The power of this technique is perhaps best illustrated by the ability to achieve detection limits in the ppt range for oxygen, water, methane and carbon dioxide in high-purity gases like hydrogen, nitrogen, argon and helium. A proper characterization of the trace impurities in these gases is of utmost importance in semiconductor processing. Using a mobile API-MS system, such as the Sciex TAGA, real-time environmental monitoring of TNT, industrial emissions, PCBs and other environmental contaminants in the ground-level troposphere is possible.

LC-MS

The use of API-MS for the online LC-MS coupling is the largest commercial field of application of API-MS. From the early 1990s, when the potential of API-MS for LC-MS, and especially its robustness and user-friendliness, were substantially demonstrated, an astonishingly large number of API-MS systems for LC-MS were purchased. Particularly in various stages of drug development within pharmaceutical industries, the use of API-based LC-MS is preferred over the more conventional LC-UV system. In general, better reliability, confirmation, selectivity and sensitivity are achieved in LC-MS systems. In this way, high-throughput quantitative bioanalysis is possible. In addition, API-based LC-MS systems using electrospray or atmospheric-pressure chemical ionization (APCI) interfacing and ionization are applied for the quantitative and qualitative analysis of a wide variety of polar compounds in numerous matrices in a number of other fields, e.g. pesticides, herbicides and their metabolites and degradation products in environmental samples, natural products in plant extracts and cell cultures, peptides, proteins and other biomacromolecules in biochemical and biotechnological applications. ICP-MS

ICP-MS is used in practically every discipline where inorganic analytical support is required. This includes environmental, geological, biological, medical, nuclear, metallurgical (semiconductor industry) and nutritional studies. An important advantage of ICPMS in quantitative analysis, required in many fields of application, is the substantial gain in precision and accuracy that can be achieved by the use of isotopedilution mass spectrometry, where stable isotopes or

24 ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION

long-lived radio-isotopes of the elements to be analysed are added as internal standards, e.g. the use of 111Cd in the analysis of 114Cd. In addition, ICP-MS can be extremely useful in speciation. In this respect, there is a growing interest in the use of ICP-MS directly coupled to LC or capillary electrophoresis for speciation and analysis of elements of toxicological interest, like As, Se, Pb and Hg. See also: Chemical Ionization in Mass Spectrometry; Chromatography-MS, Methods; Cosmochemical Applications Using Mass Spectrometry; Inductively Coupled Plasma Mass Spectrometry, Methods; Inorganic Chemistry, Applications of Mass Spectrometry.

Further reading Bruins AP, Covey TR and Henion JD (1987) Ion spray interface for combined liquid chromatography/atmospheric pressure ionization mass spectrometry. Analytical Chemistry 59: 26422646. Carroll DI, Dzidic I, Horning EC and Stillwell RN (1981) Atmospheric pressure ionization mass spectrometry. Applied Spectroscopic Review 17: 337406. Chowdhury SK, Katta V and Chait BT (1990) An electrospray-ionization mass spectrometer with new features. Rapid Communications of Mass Spectrometry 4: 8187. Cole RB (ed) (1997) Electrospray Ionization Mass Spectrometry. Chichester, UK: Wiley. Fenn JB, Mann M, Meng CK, Wong SF and Whitehouse CM (1989) Electrospray ionization for mass spectrometry of large biomolecules. Science 246: 6471.

Hieftje GM and Norman LA (1992) Plasma source mass spectrometry. International Journal of Mass Spectrometry and Ion Processes 118/119: 519573. Iribarne JV and Thomson BA (1976) On the evaporation of small ions from charged droplets. Journal of Chemical Physics 64: 2287 and (1979) Field induced ion evaporation from liquid surfaces at atmospheric pressure. Journal of Chemical Physics 71: 4451. Knewstubb PF and Sugden TM (1958) Mass-spectrometric observation of ions in flames. Nature 181: 474475. Knewstubb PF and Sugden TM (1958) Mass-spectrometric observation of ions in hydrocarbon flames. Nature 181: 1261. Meek JM and Craggs JD (1978) Electrical Breakdown of Gases. Chichester: Wiley. Meng CK, Mann M and Fenn JB (1988) Proceedings of the 36th ASMS Conference on Mass Spectrometry and Allied Topics, June 510, San Francisco, CA, pp. 771772. Niessen WMA (1998) Liquid Chromatography Mass Specctrometry, 2nd edn. New York: Marcel Dekker. Niessen WMA (1998) Advances in instrumentation in liquid chromatography mass spectrometry and related liquid-introduction techniques. Journal of Chromatography A 794: 407435. Vela NP, Olson LK and Caruso JA, Elemental speciation with plasma mass spectrometry. Analytical Chemistry 65: 585A597A. Yamashita M and Fenn JB (1984) Electrospray ion source. Another variation on the free-jet theme. Journal of Physical Chemistry 88: 44514459. Yamashita M and Fenn JB (1984) Negative ion production with the electrospray ion source. Journal of Physical Chemistry 88: 46714675.

Introduction Atomic absorption spectroscopy has become one of the most frequently used tools in analytical chemistry. This is because for the determination of most metals and metalloids the technique offers sufficient sensitivity for many applications and is relatively interference free. There are two basic atom cells (a means of turning the sample, usually a liquid, into free atoms) used in atomic absorption spectroscopy:

ATOMIC SPECTROSCOPY Methods & Instrumentation (i) the flame and (ii) electrothermal heating of a sample cell. It is generally acknowledged that if sufficient analyte is present in the sample, then it should be determined using a flame technique because this has the added advantages of being rapid (assuming only a few elements need be determined) and, in comparison with alternative techniques, very simple to use. Electrothermal atomic absorption spectroscopy (ETAAS) requires more operator skill and is less rapid, but yields substantially superior limits of

ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION 25

detection when compared with flame atomic absorption spectroscopy (FAAS). This section describes some of the methods and instrumentation that have been developed for both flame and electrothermal techniques of atomic absorption spectroscopy.

Instrumentation The basic principle of both FAAS and ETAAS is that sample is introduced into the atom cell, where it is desolvated and then atomized. The analyte atoms so formed then quantitatively absorb light in a way that is proportional to the concentration of the atoms of the analyte in the cell. The light, which is at a specific wavelength, is then isolated from other wavelengths that may be emitted by the atom cell and then detected. Thus, much of the instrumentation used for electrothermal and flame absorption spectroscopy is identical. Both techniques require a similar light source, background correction system, line isolation device (monochromator or polychromator), detector (photomultiplier or charge coupled device) and readout system. Each of these components is discussed below, together with details of the individual atom cells (flame and the electrothermal atomizer) and sample introduction systems.

Light source The fundamental requirement of the light source is to provide a narrow line profile with little background. It should also have a stable and reproducible output with sufficient intensity to ensure that a high signal-to-noise ratio is obtained. Two basic types of light source are used for atomic absorption, of which the hollow-cathode lamp (HCL) is the more commonly used. This is a lamp in which the cathode is coated with the analyte metal of interest. Within the lamp, inert filler gas (neon or argon) is ionized by an electric current and these ions are then attracted by the cathode. The inert gas ions bombard the cathode and in so doing excite the metal ions coated on it. It is this excitation of the metal that produces the emission of radiation with wavelengths characteristic of the analyte. Hollow-cathode lamps are available for most metallic elements. A schematic diagram of a hollow-cathode lamp is shown in Figure 1. Electrodeless discharge lamps are used less frequently than the hollow-cathode lamps except for analytes such as arsenic and selenium. These lamps may be excited using either microwave energy (although these tend to be less stable) or radiofrequency energy. The radiofrequency-excited lamps are less

Figure 1 Schematic representation of a hollow-cathode lamp. From Ebdon L (1998) Introduction to Analytical Atomic Spectrometry. Reproduced by permission of John Wiley & Sons Limited.

intense than the microwave-excited ones, but are still 5100 times more intense than a standard hollowcathode lamp. In the electrodeless discharge lamp, a bulb contains the element of interest (or one of its salts) in an argon atmosphere. The radiofrequency energy ionizes the argon and this in turn excites the analyte element, causing it to produce its characteristic spectrum. For analytes such as arsenic and selenium, these lamps give a better signal-to-noise ratio than hollow-cathode lamps and have a longer useful lifetime. A schematic diagram of an electrodeless discharge lamp is shown in Figure 2.

Background correction systems There are basically three types of automatic background correction system available for atomic absorption, although manual methods such as the use of nearby non-atomic absorbing lines to estimate background absorbance may also be used. The three main automatic methods are the deuterium or hydrogen lamp, the Zeeman effect, and SmithHieftje background correction. The deuterium lamp produces a continuum of radiation, some of which will be absorbed by molecular species within the atom cell. The amount of atomic absorption observed using the deuterium lamp is negligible and hence the atomic absorption signal is obtained by subtracting the absorbance from the continuum lamp from the total analyte absorbance from the hollow cathode lamp. The deuterium or hydrogen lamp is of most value at wavelengths in the UV region (< 350 nm). The Zeeman effect background correction system is more versatile. It relies on a strong magnetic field operating at approximately

26 ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION

Figure 2 Electrodeless discharge lamp. From Ebdon L (1998) Introduction to Analytical Atomic Spectrometry. Reproduced by permission of John Wiley & Sons Limited.

1 T and 5060 Hz, placed either around the light source or, more commonly, around the atom cell to split the signal into a number of components. The π component is at the normal analyte wavelength; the σ components are typically 0.01 nm (depending on the strength of the magnetic field) either side of the π component and hence lie outside the atomic absorption profile. Background can be corrected by subtracting the absorbance with the magnetic field on, from the signal with the magnet off. It should be noted that this is a simplified description of the process, as different elements have different splitting patterns and the magnetic field may be applied longitudinally or transversely, which may also produce different splitting patterns. When utilizing the SmithHieftje system, the hollow-cathode lamp is boosted periodically to a much higher current, causing the lamp to self-absorb. In this state no atomic absorption occurs in the atom cell but the molecular absorption still remains. Background correction is achieved automatically by subtracting the signal obtained at high current from that obtained using the normal current. The process is particularly efficient at removing interferences such as that caused by phosphate or selenium determinations, although the high currents used may shorten the lifetime of the lamp. In modern instruments, modulation of the source is achieved electronically. This enables discrimination between absorption and emission signals. Previously, a rotating sector, often referred to as a chopper, placed between the source and the atom cell was used.

more complex and offers far fewer advantages for atomic absorption when compared with doublebeam spectrometers in molecular spectroscopy. This is because the reference beam does not pass through the atom cell. Despite this, double-beam instruments can compensate for source drift and for warm-up and source noise.

Line isolation devices To ensure that only light of a wavelength specific to the analyte of interest is being measured, a line isolation device is required. Until recently, the line isolation device used for atomic absorption was a

Single-beam and double-beam instruments The vast majority of instruments used for atomic absorption measurements have a single-beam configuration, using the optical layout shown in Figure 3A. The double-beam arrangement (Figure 3B) is far

Figure 3 Schematics of (A) single-beam and (B) double-beam spectrometers. From Ebdon L (1998) Introduction to Analytical Atomic Spectrometry. Reproduced by permission of John Wiley & Sons Limited.

ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION 27

monochromator. There are numerous types of monochromator, but modern instruments typically use one of three designs: Ebert, CzernyTurner or Littrow configuration. These are shown schematically in Figure 4. Light of all wavelengths enters the monochromator through an entrance slit and is then split into specific wavelengths using either a prism, or more commonly, a diffraction grating. By altering the position of this dispersing element, light of only the desired wavelength passes through the exit slit to the detector. Since the method of atomic absorption is so specific, very highly resolving monochromators are not required. Thus, the focal length of an atomic absorption monochromator is often 0.25 or 0.5 m compared with a minimum of 0.75 m required for conventional optical emission spectroscopy.

A monochromator enables only one wavelength to be interrogated at any instant; this was something of a weakness of atomic absorption spectroscopy. New technology, however, has enabled the development of multielement spectrometers that use a bank of between 4 and 6 hollow-cathode lamps and an echelle-style polychromator employing orders of 100 or more and are thus capable of considerable dispersion. In addition, such instruments have no exit slit and so a huge number of wavelengths may be focused onto an array detector such as a charge-coupled device. A schematic diagram of such an instrument is shown in Figure 5. Research continues into producing a continuum light source that would enable 30 40 analytes to be determined simultaneously.

Detection systems Traditionally, detection of the light isolated by the monochromator has been accomplished using a photomultiplier tube (PMT). Several configurations exist, e.g. end-on and side-on, and the construction may be of different materials, to increase the efficiency at different wavelengths; but basically they all work in a similar way. Light enters the multiplier through a quartz window and impacts with a photocathode that is usually made from one of a number of alloys (e.g. CsSb, NaKSbCs or GaAs), which then emits electrons. These electrons are then accelerated down a series of dynodes, each being at a more positive potential than the previous one. As the electrons impact with successive dynodes, further electrons are ejected and hence a cascade effect occurs. In this way a single photon may cause the ejection of 106 electrons. The number of electrons is

Figure 4 Schematics of different monochromator types: (A) Ebert, (B) Czerny–Turner, (C) Littrow. From Ebdon L (1998) Introduction to Analytical Atomic Spectrometry. Reproduced by permission of John Wiley & Sons Limited.

Figure 5

Echelle-based polychromator.

28 ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION

then measured at the anode and the resulting current is proportional to the radiation reaching the PMT. As described previously, a number of different materials may be used to coat the photocathode in the PMT, and each has a different response curve. Some, for example the CsSb tube, have very little sensitivity above 600 nm, but others, for example GaAs, may be used up to almost 900 nm. The response profiles of some commonly used photomultipliers are shown in Figure 6. Conventional spectrometers use a detector that is capable of measuring only one signal. Multielement instruments use state-of-the-art diode arrays or charge-coupled devices that can measure numerous spatially separated signals and can therefore simultaneously determine the signals arising from a bank of hollow-cathode lamps. A charge-coupled device may be considered to be similar to an electronic photographic plate. The device consists of several hundred linear photodetector arrays on a silicon chip with dimensions of typically 13 × 18 mm. The line isolation device (often an echelle grating) separates the analytical wavelengths and these may then be detected on different regions of the array. A detailed description of how the array works is beyond the scope of this text.

Figure 6 Response curves for several commercial photomultiplier tubes. From Ebdon L (1998) Introduction to Analytical Atomic Spectrometry. Reproduced by permission of John Wiley & Sons Limited.

Instrument control and output devices Manual controls and dials have slowly disappeared, with the large majority of modern instruments being controlled by computer. In addition to controlling the instrumental parameters, the computer may be used to program autosamplers, save experimental parameters and data, calibrate with known standards, plot calibration curves, use scale expansion facilities (if necessary) and produce statistical data from the results. A continuous graphics mode enables the signal to be monitored over a period of time. This is especially useful when flow injection, chromatography or hydride-generation transients are to be detected and integrated. Older instrumentation relied on a chart recorder for the same purpose.

Sample introduction and atom cells For flame systems, the sample is introduced via a nebulizer and spray chamber assembly. Sample is drawn up the intake capillary by the Venturi effect. The liquid sample column comes into contact with the fast moving flame gases and is shattered into small droplets. In many systems it is further smashed into still smaller droplets by an impact bead. The flame gases then carry the aerosol (the nebular) through a spray chamber containing a series of baffles or flow spoilers. These act as a droplet size filter, passing the larger droplets to waste (8590% of the sample) and allowing the finer droplets to be transported to the flame. For electrothermal atomizers, the sample is placed into the atom cell either using a hand-held micropipette or by an autosampler. Since most instruments are supplied with an autosampler and the precision associated with their use is superior to that of the micropipette, most laboratories make use of this method of sample introduction. As described previously, there are two basic types of atom cell; flame and electrothermal tube, but both can have numerous modifications. The flame cell most commonly used is the 10 cm length airacetylene burner. This provides a flame that is at approximately 2300°C (although the actual temperature will depend on the fuel/air ratio). The flame may be used in a fuel-lean mode, which produces a hot, oxidizing blue flame; in a fuel-rich mode, which produces a cooler, reducing yellow flame; or in a stoichiometric mode whose properties are between the two. Different analytes give different sensitivities depending on the flame type. Chromium, for instance, gives better sensitivity in a reducing flame, whereas for magnesium it is better to use a lean, oxidizing flame. Different areas of the flame have different temperatures and different chemical properties. The

ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION 29

efficiency of atomization of analytes will therefore depend critically on the area within the flame. Hence, it is important that the light beam from the HCL passes through the region of the flame where atomization is optimal. Optimization of the burner height is therefore necessary to ensure that maximum sensitivity is achieved. Although an airacetylene flame is sufficient for the atomization of the majority of analytes, it is not sufficiently hot or reducing to atomize analytes that form refractory oxides (e.g. Al, Mo, Si and Ti). For analytes such as these a nitrous oxideacetylene flame of length 5 cm is used. This flame is hotter (2900°C) and contains typically 2.53 times as much fuel as the air acetylene flame. Other flame types, e.g. airpropane and airhydrogen also exist, although they are less commonly used. The airhydrogen flame (2200°C) is almost invisible and has the advantage of being transparent at lower wavelengths offering improved noise characteristics for elements such as lead or tin at around 220 nm. The inhomogeneity of flame chemistry and the need for burner height optimization hold true for all flame types. One common modification to a flame cell is to place a quartz tube on the burner head so that the light beam passes through the length of the tube. This method is especially useful when using sample introduction by hydride generation or by gas chromatography. The tube is heated either by a flame or electrically using a wire winding. The analyte, which enters the tube via a hole or slit cut into the side, is atomized within the tube. The atoms then leave the tube but, because of their retention in the tube, spend longer in the light beam than in normal flame systems. An increase in sensitivity is therefore obtained. Another small modification of this system is that, instead of a quartz tube, a quartz T-piece may be used. However, the function is the same. Electrothermal atom cells have changed radically since their inception in the late 1950s. The majority of electrothermal devices have been based on graphite tubes that are heated electrically (resistively) from either end. Modifications such as the West Rod Atomizer (a carbon filament) were also devised but were later abandoned. Tubes and filaments made from highly refractory metals such as tungsten and tantalum have also been made, but they tend to become brittle and distorted after extended use and have poor resistance to some acids. Their use continues, however, in some laboratories that need to determine carbide-forming elements. For example, silicon reacts with the graphite tube to form silicon carbide, which is both very refractory and very stable. The silicon is therefore not atomized and is lost analytically. Use of a metal vaporizer prevents this.

In electrothermal atomization the sample is introduced into the tube, which is then heated in a series of steps at increasing temperature. The sample is dried at a temperature just above the boiling point of the solvent (but not so hot as to cause frothing and spitting of the sample); ashed (charred at an intermediate temperature to remove as much of the concomitant matrix and potential interferences as possible without losing any analyte); and then atomized at a high temperature. During the atomization stage, the atoms leave the graphite surface and enter the light beam, where they absorb the incident radiation. The ashing and atomizing temperatures used will depend on the analyte of interest; for example, some analytes such as lead are relatively volatile and so cannot be ashed at temperatures above 450°C, otherwise volatile salts such as chlorides will be lost. Other analytes, for example, magnesium, are less volatile and can be ashed at temperatures close to 1000°C without analyte loss. Such elements require a much higher atomization temperature. The sensitivity of electrothermal atomization AAS is greater than for flame AAS because the atoms are formed within the confines of a tube and hence spend longer in the light beam. Also, since the sample is placed within the atom cell, 100% of it is available for analysis compared with the 1015% available in flame systems. A comparison of characteristic concentrations (concentration that gives an absorbance of 0.0044) obtained for flame and electrothermal techniques for many analytes is shown in Table 1. It must be stressed that the figures for ETAAS will depend critically on the injection volume and so values tend to be given in absolute terms (i.e. a weight). Overall, the Massmann design of electrothermal atomizer in which the tube is heated from either end is still the most common (Figure 7B), but more recently transversely heated tubes have been developed (Figure 7A). The longitudinally heated tubes have a temperature gradient along the tube, with the central portion being several hundred degrees hotter than the ends. This can lead to condensation of analyte at the cooler ends and subsequent reatomization from the hot graphite surface. Several atomization peaks may therefore result. The transversely heated tubes do not have a temperature gradient and therefore do not suffer from this problem.

Interferences Flame techniques are regarded as being relatively free from interferences, but some distinct classes of interference do exist. These include a few spectral interferences (e.g. Eu at 324.753 nm on Cu at 324.754 nm), ionization interferences and chemical

30 ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION

Table 1 Comparison of characteristic concentrations for flame and electrothermal AAS

Analyte

300

800 a,5b

100

8 500

600

200

1 300

2 200a,0.1b

2.5

25 220

48 000

7 400

280

100

300

350a,4b

Electrothermal AAS (pg)

Flame AAS (µg L–1)

0.4

1 500

120

400

100

300

1 400

300

110 000

40 000

750

700

Under normal flame conditions. With vapour generation.

interferences. Ionization interference is a vapourphase interference (in the past often termed cation enhancement) that occurs when the sample contains large amounts of an easily ionized element. The presence of large concentrations of easily ionized elements will lead to a large concentration of electrons in the flame. These electrons prevent the ionization of the analyte and hence lead to higher atomic absorption signals. If the easily ionized element is not present in the standards, the analyte may be partially

Figure 7 Electrothermal tubes available commercially: (A) transversely heated graphite atomizer (THGA), (B) longitudinally heated Massmann atomizer. From Ebdon L (1998) Introduction to Analytical Atomic Spectrometry. Reproduced by permission of John Wiley & Sons Limited.

ionized (and lost analytically) and hence serious overestimates of the true concentration will be obtained. This type of interference may be overcome by adding an excess of easily ionized element to all standards and samples. Chemical interferences may exist in several different forms: Formation of less volatile compounds, e.g. when phosphate is present during the determination of calcium. Calcium phosphate is refractory and hence atomization will be retarded in comparison with calcium in the standards. Formation of more volatile compounds, e.g. chlorides. Occlusion into refractory compounds. Small amounts of analyte may become trapped in a refractory substance and hence not be atomized efficiently. Occlusion into volatile compounds. Some compounds sublime explosively and hence atomization may be enhanced.

ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION 31

Most of these interferences may be overcome by using a hotter flame (e.g. nitrous oxideacetylene); by adding a chelating reagent, e.g. EDTA, to complex preferentially with the analyte; by adding a releasing agent, e.g. lanthanum, that will combine preferentially with phosphate; or by optimizing the flame conditions and viewing height. Electrothermal AAS was renowned for being highly prone to interferences. However, modern methods and instrumentation have decreased this problem substantially. Interferences include memory effects, chemical interferences (loss of analyte as a volatile salt, carbide formation, condensation and recombination), background absorption (smoke), and physical interferences such as those resulting from placing the sample on a different part of the tube. The stabilized temperature platform furnace (STPF) concept has gone a long way to eliminating the problem. The concept is that a matrix modifier is required for interference-free determinations. This is a material that either decreases the volatility of the analyte, enabling higher ash temperatures to be achieved and hence boiling away more interference (a common example is a mix of palladium and magnesium nitrates) or increases the volatility of the matrix (for example, the introduction of air burns away many interferences leaving the analytes in the atomizer). The matrix modifier may be placed in the autosampler, and added to all samples and standards. Other requirements for the STPF concept include a rapid heating rate during atomization, integrated signals (rather than peak height), a powerful background correction system (e.g. Zeeman), fast electronics to measure the transient signal, and isothermal operation. Isothermal operation means that the analyte is vaporized from the graphite surface into hot gas. In normal electrothermal AAS, the analyte leaves the hot tube wall and enters the cooler gas phase. Under these circumstances it may recombine with some other species to form a compound and thus be lost analytically. To overcome this, platforms that are only loosely in contact with the tube walls have been developed. The analytes in this case are atomized not by the resistively heated graphite tube but by the surrounding gas. The transversely heated tubes described earlier have a platform as an integral part of the tube. A modification to this technique, called probe atomization, has also been developed. Here the sample is placed on a probe and is then dried and ashed in the normal way. The probe is then removed from the tube, which is heated to the atomization temperature. The probe is then re-introduced into the hot environment. Most analytical techniques for use with furnace work now utilize the advantages of the STPF

concept. Other interferences, e.g. formation of carbides, may be partially overcome by treating the atomizer with a carbide-forming element, for example by soaking in tantalum solution. The principle is that the tantalum occupies the active sites on the surface, thereby preventing the analyte from forming a carbide.

Methods Numerous methods have been described for use with AAS, but the majority require the sample to be in a liquid form. Some ETAAS systems allow the analysis of solids directly (e.g. by weighing small amounts onto sampling boats that may be slotted into specialized tubes), but usually, if solids are to be analysed, acid decomposition methods are required. An alternative is the analysis of slurries, using finely ground material (< 10 µm) dispersed in a solvent. Manual agitation of the slurry ensures homogeneity of the sample, enabling a representative aliquot to be introduced. Some autosamplers come equipped with an ultrasonic agitator that will perform the same task. This method of analysis is frequently used in ETAAS. If samples containing high dissolved (or suspended) solids are to be analysed by FAAS, a flow injection technique (in which discrete aliquots of sample are introduced) may be used. This has also been referred to as gulp sampling. This prevents salting up of the nebulizer and blockage of the burner slot, which are obviously undesirable effects that have an adverse effect on sensitivity and signal stability. Flow injection techniques may also be used to preconcentrate analytes. If the sample flows through a column containing an ion-exchange resin, the analytes will be retained. Elution with a small volume of acid may yield very high preconcentration factors. This technique has been used for both FAAS and, more recently, for ETAAS, yielding limits of detection far superior to those achieved under normal conditions. A typical flow injection manifold suitable for this type of application is shown in Figure 8. A more traditional method of preconcentration is solvent extraction. Flame AAS may be used as a detector for analytes in organic solvents (sensitivity

Figure 8 Typical flow injection manifold for matrix analyte preconcentration or matrix elimination.

32 ATOMIC ABSORPTION, METHODS AND INSTRUMENTATION

may even be improved because of improved nebulization efficiency), but occasional blocking of the burner slot by carbon may occur. This may be removed by gentle rubbing with a noncombustible material (such as a spatula). Organic solvents may also be introduced to ETAAS, but the dry-stage temperature has to be modified to prevent sample loss by spitting. Hydride generation is a common method for the detection of metalloids such as As, Bi, Ge, Pb, Sb, Se, Sn and Te, although other vapours, e.g. Hg or alkylated Cd, may also be determined. This technique improves the sensitivity of the analysis substantially. Since the sample is in the gas phase, the sample transport efficiency is close to 100%. The hydrides atomize readily in the flame, although this approach is usually used in conjunction with a quartz T-piece in the atom cell. Methods have been developed that trap the hydrides on the surface of a graphite tube for use with ETAAS. This leads to preconcentration and further improvements in detection limit. Chromatography has been coupled to AAS to effect speciation analysis. Here different chemical forms of an analyte are separated (either by highperformance liquid chromatography or, if they are sufficiently volatile, by gas chromatography) prior to introduction to the atomic absorption instrument. Gas chromatography is usually coupled directly with a T-piece interface arrangement, but HPLC couplings introduce the sample through the conventional nebulizer/spray chamber assembly. A small postcolumn air bleed may be necessary to compensate for the differences in flow rate between the chromatograph (12 mL min−1) and the uptake rate of the nebulizer (510 mL min−1). Chromatography is not frequently coupled with ETAAS because the atom cell is not well suited to continuous monitoring. The limited linear range offered by the AAS technique may be partially overcome by using alternative, less sensitive lines. This is a useful technique that avoids the requirement to dilute samples. Care must be taken, however, to ensure that viscosity effects do not cause a difference in nebulization efficiency between samples and standards. Alternatively, burner head rotation may be used. This shortens the path in the flame through which the light beam travels and hence decreases sensitivity. However, there is often an increase in noise as a result of using this approach.

Conclusions The sensitivity offered by the various techniques utilizing AAS for the determination of metals and

metalloids is often sufficient for many applications. Direct aspiration of samples in solution into a flame atomic absorption instrument provides the analyst with rapid acquisition of data with good precision. The technique is remarkably free from interferences and those that do exist may usually be overcome by judicious choice of operating conditions. The instrumentation involved is relatively simple and, for flame work, analyses may easily be performed by nonexpert operators. If increased sensitivity is required, the analyst has the option of preconcentrating the analyte using one of a number of methods prior to introduction to flame AAS, or alternatively may use electrothermal AAS. The latter technique, however, does require more experience if reliable results are to be obtained. A potential disadvantage of using atomic absorption is that in the past instruments have been capable of determining only one analyte at a time. Modern instrumentation has improved this number to as many as six, although it must be stressed that compromise conditions (e.g. in ETAAS temperature programs) must be used. This may lead to a decrease in overall sensitivity compared with single-element determinations. However, the ongoing development of multielement AAS will certainly boost the use of this already popular and versatile technique. See also: Atomic Absorption, Theory; Atomic Fluorescence, Methods and Instrumentation; Atomic Spectroscopy, Historical Perspective; Light Sources and Optics.

Further reading Dedina J and Tsalev DL (1995) Hydride Generation Atomic Absorption Spectrometry. Chichester: Wiley. Ebdon L, Evans EH, Fisher A and Hill SJ (1998) An Introduction to Analytical Atomic Spectrometry. Chichester: Wiley. Harnly JM (1996) Instrumentation for simultaneous multielement atomic absorption spectrometry with graphite furnace atomization. Fresenius Journal of Analytical Chemistry 355: 501509. Haswell SJ (ed) (1991) Atomic Absorption Spectrometry, Theory, Design and Applications, Analytical Spectroscopy Library. Amsterdam: Elsevier. Hill SJ, Dawson JB, Price WJ, Shuttler IL, Smith CMM and Tyson JF (1998) Atomic spectrometry update advances in atomic absorption and fluorescence spectrometry and related techniques. Journal of Analytical Atomic Spectrometry 13: 131R170R. Vandecasteele C and Block CB (1993) Modern Methods for Trace Element Determination. Chichester: Wiley.

ATOMIC ABSORPTION, THEORY 33

Atomic Absorption, Theory Albert Kh Gilmutdinov, Kazan State University, Russia

ATOMIC SPECTROSCOPY Theory

Atomic absorption spectroscopy (AAS) is a technique for quantitative determination of metals and metalloids by conversion of a sample to atomic vapour and measurement of absorption at a wavelength specific to the element of interest. Owing to high sensitivity and selectivity, the technique is widely used for fundamental studies in physics and physical chemistry: measurements of oscillator strengths, diffusion coefficients of gas phase species, partial pressure of vapours, rate constants of homogeneous and heterogeneous reactions, etc. The widest application of AAS, however, is in analytical chemistry. Nowadays it is one of the most popular techniques for trace analysis of over 65 elements in practically all types of samples (environmental, biological, industrial, etc.). The vast majority of substances to be analysed by AAS are in the condensed phase. At the same time, atomic absorption, like any other atomic spectrometry technique, can only detect free atoms that are in the gas phase. Thus, as analyte initially present in solution at a concentration c(liq) must first be transferred into the gas phase via the atomization process to produce N(g) free atoms in an atomizer volume. The analyte atoms are then detected by absorption of radiation from a primary source at a wavelength characteristic of the element resulting in an absorption signal, A. Thus, the general scheme of AAS can be presented as follows:

The primary goal of the theory of AAS is to establish the relationship between the measured analytical signal, atomic absorbance, A, and the analyte concentration, c, in the sample. Theoretical description of the two stages involves entirely different sciences: theory of atomization is based on thermodynamics, kinetics and molecular physics, while description of absorbances is based on optics and spectroscopy. Below, the two stages will be considered separately.

Atom production In the first step of an AA determination the element of interest must be atomized. The ideal atomizer would

provide complete atomization of the element irrespective of the sample matrix producing a well-defined and reproducible absorption layer of the analyte atoms. Excitation processes, however, should be minimal so that analyte and background emission noise is small. For achieving the best detection limits, the analyte vapour should not be highly diluted by the atomizer gas. Various types of electric discharges, laser radiation, electron bombardment and inductively coupled plasmas were tested as atomizers. Although not ideal, high-temperature flames and electrothermal atomizers have gained acceptance as the most common technique of atom production in AAS. Flame atomization

The longest practiced technique for converting a sample into atoms is the spraying of a sample solution into a combustion flame. The most popular flames used in AAS are the airC2H2 flame providing a maximum temperature of about 2500 K and the N2OC2H2 flame with a maximum temperature exceeding 3000 K. Atomization occurs because of the high enthalpy and temperature of the flame, and through chemical effects. Owing the generation of cyanogen radicals, which are known to be an efficient scavenger for oxygen, the nitrous oxide-acetylene flame provides not only a hotter but also a more reducing environment for atom production: the lack of oxygen moves equilibria such as MeO o Me + O to the right. Therefore a N2OC2H2 flame provides greater atomization efficiencies and thus better detection limits for refractory elements, such as Al, Ta Ti, Zr, Si, V and the rare earths. Atom production in flames is an extremely complex process consisting of many stages and involving numerous sides processes. A quantitative theory of analyte atomization in flames is absent. Qualitatively, the process can be described as follows: the solution droplets sprayed into the flame are first dried, the resulting solid microparticles become molten and vaporize or thermally decompose to produce gaseous molecules that are finally dissociated into free atoms. The atoms may further be excited and ionized and form new compounds. These processes are dependent upon the temperature and the reducing power of the flame and occur within a few milliseconds the time required by the sample to pass through the

34 ATOMIC ABSORPTION, THEORY

flame. The processes taking place in flames are shown in more detail in Figure 1. Electrothermal atomization

In the majority of cases this occurs in a small graphite tube where the sample to be analysed is introduced via a small aperture in the upper tube wall. Metals such as tantalum or tungsten are also used to construct electrothermal atomizers as well as nontubular geometry for the atomizer (graphite cups, rods, filaments, etc.). After a 550 µL droplet of the sample solution has been deposited into the graphite furnace, it is heated resistively through a series of preprogrammed temperature stages to provide: (i) drying of the droplet to evaporate the solvent (∼ 370 K for about 30 s); (ii) thermal pretreatment (pyrolysis) to remove volatiles without loss of analyte (6001500 K for 45 s) (this allows selective

volatilization of matrix components that are purged from the atomizer by a flow of inert gas); (iii) analyte atomization (15003000 K achieved rapidly at a heating rate of about 2000 K s1). A major difference of electrothermal atomization from atomization in flames is that some matrix components are removed at the pyrolysis step and the atomization takes place within a confined geometry in an inert gas atmosphere. During atomization, the purge gas flow is shut off so that analyte atoms remain in the probing beam as long as possible. The number of analyte atoms in the atomizer volume N(t), generated during the atomization stage is described by the convolution:

Figure 1 Schematic representation of the processes taking place in a flame. The bold arrows show the pathways for analyte atom production and thin arrows show side processes. Reproduced with permission from Welz B (1985) Atomic Absorption Spectrometry, 2nd edn, p. 167. Weinheim: VCH.

ATOMIC ABSORPTION, THEORY 35

Here the supply function S(t′ ), gives the rate (atoms s−1) of the primary generation of analyte atoms from the deposited sample, whereas the removal function, R(t,t′ ) characterizes the subsequent transfer of the vaporized atoms in the atomizer volume. In the simplest case of vaporization of a monolayer of atoms from the atomizer surface the supply function is given as:

where N0 is the total number of atoms in the deposited sample, ν and E are the frequency factor and the activation energy of the vaporization process, respectively, R is the gas constant and T(t) is the timedependent atomizer temperature. A typical example of the supply function in electrothermal AAS is given in Figure 2. The removal function R(t,t′) gives the probability of an analyte atom, vaporized at the time, t′, to be present in the atomizer volume at a later instant, t > t′. Early in time, the probability is equal to unity and decreases with increasing time because of loss processes. Generally, the loss processes include concentration and thermodiffusion, and convection. The knowledge of the removal function allows calculation of the mean residence time, W, of analyte in the atomizer volume. This parameter is defined as W = R(t) dt and gives the life-time of evaporated atoms in the atomizer volume. When the analyte loss is governed only by concentration diffusion through the tube ends, the residence time can be estimated by the following relationship:

where L is the tube length, a is the linear dimension of the area occupied by the sample on the atomizer surface, D is the analyte diffusion coefficient. Typically, the analyte residence time in tube graphite atomizers is a few tenths of a second, which is about 100 times longer than in a flame. The long residence time, along with the high degree of analyte atomization caused by the reducing environment, leads to a 10- to 1000-fold increase in graphite furnace sensitivity compared with flame AAS. The typical change in the total number of analyte atoms, N(t), in the tube electrothermal atomizer obtained by convolution [1] of the supply function

Figure 2 Supply function S(t ) of silver atoms computed for the change in the atomizer temperature T(t ) (E = 280 kJ mol−1, Q = 1013 s−1); the change in the total number of silver atoms within the atomizer volume N(t ).

with the diffusion removal function is given in Figure 2. The actual atom production and dissipation processes in electrothermal atomizer are much more complex than that described above. In many cases the analyte is atomized as a result of a set of multistage processes taking place simultaneously. The basic processes that occur with analyte in a graphite furnace are shown schematically in Figure 3 and listed in Table 1. The last column of the Table shows the elements for which the respective processes have been documented. In addition to the above widely used atomization techniques, mention must also be made of the hydride generation technique. The method is based on conversion of a hydride forming element (As, Bi, Ge, Sb, Se, Sn, Te) in an acidified sample to volatile hydride and transport of the released hydride to an atomizer (hydrogen diffuse flame, graphite furnace, heated quartz tube) where they are atomized to give free analyte atoms. Sodium borohydride is almost exclusively used as an agent for conversion of analyte to hydride:

where El, is the element of interest and m may or may not equal n. The advantage of sample volatilization as a gaseous hydride lies in the analytes preconcentration and separation from the sample matrix. This results in an enhanced sensitivity and in a significant

36 ATOMIC ABSORPTION, THEORY

Table 1 Selection of processes involving analyte in graphite tube atomizers

Ref. in Figure 3

Process a

Elements

Me (s,l) → Me (g)

MeO (s,l) → Me (g) MeO (s,l) + C(s) → Me (g) Me C (s,l) → Me (g)

1 1 1

MeOH (s,l) → Me (g) MeO (s,l) → MeO (g) Me (s,l) → Me2 (g) Me (ad) → Me (g) Me (g) → Me (s,l)

1 1′ 1′ 3 4

Me (g) + O (g) → MeO (g) MeO (g) → Me (g)

5 5′

MeO (g) + C(s) → Me (g) + CO Me (g) + C(s) → MeC (g)

Ag, Au, Co, Cu, Ni, Pb, Pd Al, Cd, In, Pb Mn, Mg Al, Ba, Ca, Cu, Mo, Sr, V Rb Al, As, Ga, In, Tl Se, Co As, Cu Ag, Au, Cu, Mg, Mn, Pd Al, Si, Sn Al, As, Cd, Ga, In, Mn, Pb, Zn Al, Ga, In, Tl

6′

s, l, g, ad stand for solid, liquid, gaseous and adsorbed species, respectively.

reduction of interferences during atomization. The main disadvantages of the method are interferences

Table 2

Element Ag

1.5

150

Be Bi

Flame AA 45

by substances in the solution that reduce the efficiency of hydride generation. Relative detection limits achieved by the described atomization techniques are presented in Table 2.

Selected relative detection limits (µg L–1) for different atomization techniquesa

Figure 3 Schematic representation of the basic physical and chemical processes taking place in a tube electrothermal atomizer. Solid arrows denote pathways of free analyte atoms, dotted arrows show the pathways of the analytes that are bound into molecules. Primary generation of the analyte vapour from the site of sample deposition as an atomic (1) or a molecular (1′) species. Irreversible loss of analyte from the furnace through its ends (2) and through the sample dosing hole (2′) by diffusion and convection. Physical adsorption/desorption at the graphite surface (3). Gas phase condensation (4) at the cooler parts of the atomizer. Gas phase reactions (5) that bind free analyte atoms into stable molecules or those (5′) that increase the free atom density. Heterogeneous reactions of analyte vapour with the atomizer walls: includes both production (6) and loss (6′) of free atoms at the furnace wall.

1000 15 1.5 30

GF AA

Hydride AA

0.02

0.1 0.2

Element Mo

0.03

0.15 20

Flame AA 1.5 45

0.3

75 000

GF AA 0.08 0.02 0.3 130

0.35

0.06

0.008

0.8

0.25

0.008

100

0.15

0.045

100

0.3

0.03

0.2

0.03

1.5

0.8

0.03

Hydride AA

0.035

0.01

2 0.03 1

1.5

0.1

150

0.1

0.025

Ge Hg

300 300

10 0.6

Te Ti

30 75

0.4 0.35

900

0.15

0.008

0.8

0.06

0.15

0.004

1 0.009

1.5

0.03

0.1 0.1

Detection limits are based on 98% confidence level (3 standard deviations). The values for the graphite furnace technique are referred to sample aliquots of 50 µL. Adapted with permission from The Guide to Techniques and Applications of Atomic Spectroscopy (1997) p 5. Norwalk, CT, USA: Perkin-Elmer.

ATOMIC ABSORPTION, THEORY 37

Atomic absorption AAS is based on the interaction of the probing radiation beam from a primary source with gas phase analyte atoms. Two different types of primary source are used in AAS: line sources emitting narrow spectral lines of the element to be analysed and a continuum source (normally high-pressure xenon arc lamp) that produces radiation from below 180 nm to over 800 nm. In principle, the continuum source is best suited for AAS. It has a unique capability for multielement determinations, and provides extended analytical range and inherent background correction. However, the lack of intensity found below 280 nm still limits the use of the source in research laboratories. In conventional AAS, the line radiation of the element of interest is generated in a hollow cathode lamp (HCL) or an electrodeless discharge lamp (EDL). The spectral profile, J(O), of the analytical line emitted by the source interacts with the absorption profile, k(O), of the analyte atoms in the atomizer (Figure 4). It is seen that radiation absorption is strongly dependent on the shape and relative position of the two spectral profiles. In order to describe quantitatively absorption, both emission and absorption profiles must first be well defined.

effect, interatomic collisions and hyperfine splitting. Other types of broadening (natural broadening, Stark broadening, etc.) are negligible. The Doppler broadening originates from the thermal agitation of emitting and absorbing atoms that results in a Gaussian-shaped spectral line (curve G in Figure 5). The full width, ∆νD, at one-half of the maximum intensity (FWHM) of the Doppler-broadened atomic line is given by:

Three major broadening mechanisms determine emission and absorption profiles in AAS: the Doppler

where ∆νD is in cm−1, λ in nm, T in K and M in g mol−1. The width is inversely proportional to the wavelength of the transition, to the square root of the absolute temperature, T, and to the reciprocal of the square root of the molecular weight of the analyte, M. The absorption line widths predicted from this equation range from 1 pm to 10 pm for flame and electrothermal atomizers. The Doppler widths of the lines emitted by HCLs and EDLs are 23 times lower than those for absorption profiles in atomizers. The second major broadening mechanism in AAS is collision or pressure broadening which originates from deactivation of the excited state due to interatomic collisions. There are two types of collisions. Collisions between two analyte atoms lead to resonance broadening which is negligible in analytical

Figure 4 Schematic diagram of the emission J(O) and absorption k(O) spectral profiles typical in AAS.

Figure 5 Normalized theoretical Gaussian (G), Lorentzian (L) and Voigt (V) line profiles versus dimensionless frequency. The FWHMs and areas of the Gaussian and Lorentzian peaks are equal; the Voigt profile is the convolution of the (G) and (L) profiles.

Emission and absorption line profiles

38 ATOMIC ABSORPTION, THEORY

AAS because the analyte concentration is small. However, Lorentz broadening, which involves collisions between analyte atoms and foreign species (Ar atoms in electrothermal atomizers) called perturbers, are significant. Compared to a Doppler spectral profile, the Lorentzian profile is broader, has a lower peak height and is described by a Lorentzian function f(curve L in Figure 5) with the FWHM given by:

Here Np is the number density of perturbers (cm−3), V is the collisional cross-section (cm2), MA and Mp is the molecular weight (g mol−1) of the analyte and the perturbers, respectively. The collision width predicted by Equation [5] for absorption profiles is about the same order of magnitude as the Doppler line width (110 pm). The ∆QL values for lines emitted by the HCLs and EDLs are much lower than those for the absorption profiles in atomizers because of low pressure in the primary sources. Therefore this part of the broadening is normally neglected in the sources. However, at high optical densities of an absorbing layer when the wings of the emission line play an important role, the Lorentzian component in the broadening of the emission line is also becoming important. In addition to broadening the lines, collisions also cause a small shift in the line centre (towards the longer wavelengths in an Ar environment) and an asymmetry in the far line wings. The shift is proportional to the collision width and theory predicts a value of 2.76 for the width-to-shift ratio (it is shown schematically in Figure 4). The total spectral profile of an atomic line is given by the convolution of the Gaussian profile (due to Doppler broadening) and the Lorentzian profile (due to pressure broadening). The result is the Voigt function (curve V in Figure 5), H(ω,a):

(∆νL/∆νD) is the damping constant where a = determining the shape of the Voigt profile: as the damping constant increases, the profile broadens and its peak height decreases. In the extreme cases of a = 0 and a = ∞, the Voigt function goes to the purely Gaussian function and purely Lorentzian function, respectively. Normally, a dimensionless frequency, Z, normalized by the Doppler width ∆νD

of the absorption profile is used for computations: Z = (ν/∆νD)2 . Within about 1% accuracy, the FWHM of the Voigt profile can be estimated from the following empirical equation:

Finally, the third broadening mechanism is hyperfine splitting which is caused by isotope shifts and nuclear spin splitting. Generally a spectral line consists of n hyperfine components with relative intensities, bj, (Σ bi = 1) that are located at a distance ∆ωi from the component with the minimum frequency. Each component is Doppler and collision broadened and described by the Voigt function [6]. Thus in the general case the emission J(ω) and absorption k(ω) profiles are presented as follows:

where α is the ratio of the Doppler width of the emission line to the Doppler width of the absorption line. The equation takes into account that the emission profile, J(ω), is approximately 1/ α times narrower than the absorption profile, k(ω), in the scale of dimensionless frequency ω and that the absorption profile, is pressure-shifted relative to the emission profile to a value ∆ωS. For many analytical lines used in AAS the hyperfine splitting is greater than Doppler and pressure broadening and determines the total FWHM. In graphite furnace AAS, the damping constant, a, varies from 0.17 (Be resonance line, 234.9 nm) to 3.27 (Cs, 852.1 nm) for absorption lines and its value for emission lines, ae, in the primary sources varies from 0.01 to 0.05. A relatively small value of the a-parameter means that the emission profile is primarily determined by the Doppler broadening. Figure 6 shows the emission and absorption spectral profiles of the lead 283.3 nm resonance line computed for the conditions of graphite furnace AAS using Equation [8]. The line consists of five hyperfine components that are clearly seen in the emission profile. Because of higher temperature and much higher pressure, the absorption profile is wider and shifted towards the longer wavelengths.

ATOMIC ABSORPTION, THEORY 39

Calibration curve shapes

It has been demonstrated that in AAS the atomization efficiency is close to 100% for the majority of elements in aqueous solutions. Assuming complete atomization, the shape of calibration curves in AAS is determined by the dependence of detected atomic absorbance on the number of absorbing atoms. The

relationship A = f(N) is called the concentration curve and its description is one of the primary goals of theoretical AAS. The phenomenon of radiation absorption had already been studied quantitatively early in the 18th century, mainly on liquids and crystals. The investigations resulted in the BeerLambert law relating the radiant flux, )tr, transmitted through an absorbing layer to the optical properties of the layer. In modern formulation, the law applied to a spatially uniform absorbing layer of length l and a uniform parallel probing beam is written as follows:

Absorbance, A, which is the analytical signal in AAS, is defined as the logarithm of the ratio of the incident radiant flux, Φ0, to the radiant flux, Φtr, that is transmitted through the absorbing layer of the analyte atoms. Using the BeerLambert law [9] for expressing the transmitted radiant flux, the general relationship for the absorbance recorded by an AA spectrometer can be expressed as follows:

Here, b and h are the width and the height of the monochromator entrance slit, l is the length of the absorbing layer; the limits of −O* to O* are the spectral bandwidth isolated by the monochromator. It was assumed that radiation attenuation occurs only because of atomic absorption, i.e. H(O) = 0. The absorption coefficient, k(O;x,y,z), depends on the number density of absorbing atoms in the ground state, no (cm−3), and the spectral profile, k(O), of the absorption line, and is presented as:

where )0(O) is the incident radiant flux incident at a wavelength O, P(O) is a wavelength-dependent attenuation coefficient which is the sum of absorption, k(O), and scattering, H(O), coefficients: P(O) = k(O) + H(O). Curve Jtr(O) in Figure 6 shows the spectral profile of the lead 283.3 nm resonance line that is transmitted through a uniform layer of 3 × 1012 cm−2 lead atoms. It was assumed that the spatially uniform incident radiation with the spectral profile J(O) is absorbed by the absorption profile k(O), without scattering, following Equation [9]. Generally, a non-uniform radiation with intensity distribution J(x,y) over the beam cross section is used to probe a non-uniform layer of absorbing species with the number density n(x,y,z). With non-uniform probing beams, a radiant flux passing through a surface S within a spectral bandwidth {−O*, O*} is expressed in terms of intensities as follows:

where J(O) is a function describing the spectral composition of the radiation.

Figure 6 Emission J(O ) and absorption k(O) spectral profiles of the 283.3 nm lead resonance line computed for the conditions of graphite furnace AAS (ae = 0.01, a = 1.24; T = 2100 K). Jtr(O) is the emission profile that is transmitted through a uniform layer of 3 × 1012 cm−2 lead atoms.

40 ATOMIC ABSORPTION, THEORY

Here e and m are the charge and mass of the electron, respectively, c is the velocity of light, f is the oscillator strength of the spectral transition, ∆νD is the Doppler width of the absorption line. The coefficient c(O, T) denotes the combination of values that depend only on the spectral features of the analysis lines. The general relationship [10] is simplified significantly if the following assumptions are made: (i) radial distributions of the probing beam and the analyte atoms are uniform, i.e. J(x,y) = const, n(x,y,z) = n(z); (ii) the atomizer is spatially isothermal, i.e. T(x,y,z) = const and (iii) the emission line is a single and infinitely narrow line at O0 so that it is described by Diracs delta-function: J(O) = J0 δ(O−O0). Substitution of these simplifications into Equation [10] gives:

where c(O0,T) is a constant for given transition and for a given temperature, N(cm2) = n(z)dz is the number of absorbing atoms per unit cross-sectional area along the radiation beam; in the case of uniform analyte distributions, this value is proportional to the total number of analyte atoms in the atomizer. Thus, if the above assumptions are valid, the recorded absorbance is proportional to the total number of absorbing atoms in the atomizer and the concentration curve A = f(N) is a straight line. This is a basis for AAS to be an analytical technique, because the recorded signal, A, is directly proportional to the unknown number, N, of analyte atoms. Such a concentration curve computed for the case of a lead resonance line is presented by curve 1 in Figure 7. The actual situation, however, differs substantially from that presented by simple Equation [12]. It follows from general relationship [10] that absorbance is dependent on three groups of factors: (1) the spectral characteristics of the analysis line (dependence of J and k on O); (2) the cross-sectional distribution of intensity in the incident beam, J(x,y); and (3) spatial distribution of the analyte and temperature within the atomizer (dependence of k on x,y,z). Thus detected absorbance, A, depends not only on the number, N, of absorbing atoms but also on the spectral features of the analysis lines used to probe the absorbing layer and the spatial distribution of analyte atoms and radiation intensity in the probing beam:

Figure 7 The evolution of the 283.3 nm lead concentration curve as the spectral and spatial features are successively taken into account. Emission line is: single and infinitely narrow (1), single and broadened (2); all the spectral features of the line are taken into account including hyperfine splitting and the pressure shift (3); (4) as (3) plus absorbing layer is assumed to be nonuniform in the radial cross section with the analyte number density at the atomizer bottom 4 times greater than that in the upper part of the absorbing layer.

The effect of spectral features of the analysis line and spatial nonuniformities of the absorbing layer is illustrated in Figure 7. Curve 2 is the lead concentration curve when the resonance line is assumed to be a single but broadened one; accounting for all the spectral features of the emission and absorption profiles, including hyperfine structure and pressure shift, results in curve 3. It can be seen that broadening of the analysis line lead to a bending of the concentration curve and decreasing of its slope, the major effect being the hyperfine broadening. Additional accounting for the fact that the analyte is distributed nonuniformly in the atomizer cross section leads to further curvature and a decrease in the slope of concentration curve (curve 4). Figure 7 illustrates clearly the importance of the spectral and spatial features in relationship [10]. Recent research shows, however, that these undesirable dependencies can be avoided if transmitted intensities are measured with spectral, spatial and temporal resolutions. The multidimensional absorbance detected in such a way is dependent only on the number of absorbing atoms irrespective of their spatial distribution and spectral features of the primary source.

ATOMIC ABSORPTION, THEORY 41

Interferences

An interference effect is a change of the analytical signal by the sample matrix as compared to the reference standard, typically an acidified aqueous solution. There are two types of interference in spectrochemical analysis, spectral and nonspectral. Spectral interference originates from incomplete isolation of the radiation emitted or absorbed by the analyte from other radiation detected by the spectrometer. Any reasons other than specific atomic absorption that cause radiation attenuation in equation [9] result in spectral interferences. They arise from: absorption of radiation by overlapping molecular bands or atomic lines of concomitants; scattering of source radiation by nonvolatilized microparticles (smoke, salt particles, condensed microdrops); foreign line absorption if the corresponding radiation happens to be emitted by the radiation source, in addition to the analysis line, within the spectral bandwidth of the monochromator. The first two sources of spectral interferences are most common and described as background attenuation or background absorption. In general, they are more pronounced at shorter wavelengths. Most background attenuation can be distinguished from the analyte absorption because the element only absorbs in the very narrow spectral region while the background attenuation is less specific and extends over a considerably broader wavelength band. Instrumental methods of correction for background attenuation effects include compensation using a continuum source and the Zeeman effect. For interferences other than spectral, the analyte itself is directly affected. The nonspectral interferences are best classified according to the stage at which the particular interference occurs, i.e. solute-volatilization and vapour-phase interferences. A nonspectral interference is found when the analyte exhibits a different sensitivity in the presence of sample concomitants as compared to the analyte in a reference solution. The difference in the signal may be due to: analyte loss during the thermal pretreatment stage in the electrothermal atomizer: analyte reaction with concomitants in the condensed phase to form compounds that are atomized to a lesser extent, analyte ionization or change the degree of ionization caused by concomitants. Reactions of the analyte with the tube material (graphite) or the purge gas are not normally considered to be interferences because they influence the analyte in the sample and in the standard to the same degree. A nonspectral interference is in many instances best detected by the use of the analyte addition technique. An interference exists if the slope of

the additions curve is different from that of the calibration curve. For most systems, nonspectral interferences in electrothermal AAS can be effectively eliminated by adding a proper matrix modifier (Pd-Mg(NO3)2 is the most common modifier) that delays analyte volatilization until the atomizer temperature is sufficiently high and steady for efficient atomization, and buffers the gas phase composition.

List of symbols a = damping constant; ae = damping constant for emission lines; a = the linear dimension of the area occupied by the sample on the atomizer surface; A = absorbance; b = width of monochromator entrance slit; c(liq) = concentration of analyte in solution; c = velocity of light; D = analyte diffusion coefficient; e = charge of electron; E = activation energy; f = oscillator strength; h = height of monochromator entrance slit; J(λ) = spectral profile; k(λ) = absorption profile; l = length of absorbing layer; L = tube length; m = mass of electron; M = molecular weight of the analyte; MA = molecular weight of analyte; Mp = molecular weight of perturbers; no = number density of absorbing atoms in the ground state; N = number of analyte atoms; N(g) = number of free atoms in an atomizer volume; N0 = total number of atoms in the deposited sample; NP = number density of perturbers; N(t) = number of analyte atoms in the atomizer volume; R = gas constant; R(t,t ′) = removal function; S(t ′) = supply function; t = time analyte atom present in atomizer volume; t′ = time of vaporization; T = absolute temperature; T(t) = time-dependent atomizer temperature; D = ratio of the Doppler width of the emission to the absorption line; ∆νD = width of Doppler-broadened atomic line; ∆νL = width of Lorentzian-broadened atomic line; H(O) = scattering coefficient; O = wavelength; O = wavelength; µ(O) = wavelengthdependent attenuation coefficient; ν = frequency factor; V = collisional cross section; W = mean residence time; Φ = transmitted radiant flux; Φ0 = incident radiant flux; Z = frequency (dimensionless). See also: Atomic Absorption, Methods and Instrumentation; Atomic Fluorescence, Methods and Instrumentation; Fluorescence and Emission Spectroscopy, Theory.

Further reading Alkemade CThJ, Hollander T, Snelleman W and Zeegers PJTh (1982) Metal Vapours in Flames. Oxford: Pergamon Press.

42 ATOMIC EMISSION, METHODS AND INSTRUMENTATION

Chang SB and Chakrabarti CL (1985) Factors affecting atomization in graphite furnace atomic absorption spectrometry. Progress in Analytical Atomic Spectroscopy 8: 83191. Ebdon L (1982) An Introduction to Atomic Absorption Spectroscopy. London: Heyden. Haswell SJ (Ed.) (1991) Atomic Absorption Spectrometry. Amsterdam: Elsevier. Holcombe JA and Rayson GD (1983) Analyte distribution and reactions within a graphite furnace atomizer. Progress in Analytical Atomic Spectroscopy 6: 225251. Lvov BV (1970) Atomic Absorption Spectrochemical Analysis. London: Adam Hilger.

Mitchell ACG and Zemansky MW (1961) Resonance Radiation and Excited Atoms. Cambridge: Cambridge University Press. Slavin W (1991) Graphite Furnace AAS A Source Book, 2nd edn. Norwalk, CT: Perkin-Elmer Corporation. Styris DL and Redfield DA (1993) Perspectives on mechanisms of electrothermal atomization. Spectrochimica Acta Reviews 15: 71123. Welz B (1985) Atomic Absorption Spectrometry, 2nd edn. Weinheim: VCH. Winefordner JD (ed) (1976) Spectroscopic Methods for Elements. New York: Wiley-Interscience.

Atomic Emission, Methods and Instrumentation Sandra L Bonchin, Los Alamos National Laboratory, NM, USA Grace K Zoorob, Biosouth Research Laboratories, Inc., Harahan, LA, USA Joseph A Caruso, University of Cincinnati, OH, USA

ATOMIC SPECTROSCOPY Methods & Instrumentation

Atomic emission spectroscopy is one of the most useful and commonly used techniques for analyses of metals and nonmetals providing rapid, sensitive results for analytes in a wide variety of sample matrices. Elements in a sample are excited during their residence in an analytical plasma, and the light emitted from these excited atoms and ions is then collected, separated and detected to produce an emission spectrum. The instrumental components which comprise an atomic emission system include (1) an excitation source, (2) a spectrometer, (3) a detector, and (4) some form of signal and data processing. The methods discussed will include (1) sample introduction, (2) line selection, and (3) spectral interferences and correction techniques.

Atomic emission sources The atomic emission source provides for sample vaporization, dissociation, and excitation. The ideal excitation source will allow the excitation of all lines of interest for the elements in the sample, and do this reproducibly over enough time to encompass full elemental excitation. Excitation sources include but are not limited to (1) inductively coupled plasma (ICP), (2) direct current plasmas (DCP), (3) microwave induced plasmas (MIP), and (4) capacitively coupled

microwave plasmas (CMP). Glow discharges are utilized for direct solids analyses, but will not be discussed here. An analytical plasma is a high-energy, slightly ionized gas (about 0.01 to 0.1% ionized). Inductively coupled plasmas

The most commonly used ion source for plasma spectrometry, the ICP, is produced by flowing an inert gas, typically argon, through a water-cooled induction coil which has a high-frequency field (typically 27 MHz) running through it (Figure 1). The alternating current in the coil has associated with it a changing magnetic field, which induces a changing electric field. The flowing gas is seeded with electrons by means of a Tesla coil. These electrons undergo acceleration by the electric field, and gain the energy necessary to excite and ionize the gaseous atoms by collision. This produces the plasma, self-sustaining as long as the RF and gas flows continue. Sample particles entering the plasma undergo desolvation, dissociation, atomization, and excitation. The ICP has sufficiently long residence times and high enough temperatures so that the sample solvent is completely vaporized, and the analyte reduced to free atoms, which undergo excitation. This excitation results in the emission of light at specific frequencies for elements in the sample, which is

ATOMIC EMISSION, METHODS AND INSTRUMENTATION 43

Figure 1

Inductively coupled plasma and torch schematic.

proportional to their concentration. An emission spectrometer separates the frequencies of light into discrete wavelengths, which are unique to given elements in the sample. Direct current plasmas

In a direct current plasma, a dc current passing between two electrodes heats the plasma gas, again typically argon, and produces a discharge. The most common version is the 3-electrode system (Figure 2). This system has argon flowing around two graphite anodes and a tungsten cathode to produce the plasma. The sample is introduced between the anodes. Vaporization, atomization, ionization, and excitation occur. This technique is more tolerant of samples containing a high proportion of solids than the ICP method. However, it is less efficient due to lower plasma temperatures, and the electrodes need to be replaced frequently.

can be produced at atmospheric pressure, if the design of the cavity allows. Helium plasmas typically require reduced pressure unless a Beenakker cavity is employed (Figure 3). MIPs use lower power levels (hundreds of watts) than those required by the DCP and ICP. However, due to the decreased size of the MIP, the power densities produced are comparable. Argon has an ionization potential of 15.75 eV, which is not sufficient in the argon ICP to produce the excitation ionization energy needed to ionize some elements efficiently, such as the nonmetals. Helium plasmas are more efficient for these problem elements, as the ionization potential of helium is 24.6 eV. The MIP is not very adept at handling liquid samples, due to the low powers employed. However, at higher powers above 500 watts similar performance to the ICP may be obtained. The general lack of availability of these sources as part of commercially available instrumentation, has limited their application and use. Capacitively coupled microwave plasmas

A capacitively coupled microwave plasma is formed using a magnetron to produce microwave energy at 2.45 GHz. This is brought to a hollow coaxial electrode via coaxial waveguides. The microwave power is capacitively coupled into the plasma gas, usually argon or nitrogen, via the electrode. This is an atmospheric pressure source with a small plasma volume. Power levels can range from 10 to 1000

Microwave induced plasmas

A microwave induced plasma (MIP) is an electrodeless discharge generated in a glass or quartz capillary discharge tube, often in a resonant cavity. These tubes generally have an inner diameter of the order of a few millimetres, and the plasma gas is an inert gas, such as helium or argon. The resonant cavity, which is hollow and of the order of a few centimetres diameter, allows coupling of the microwave power into the plasma gas flowing through the capillary discharge tube. The microwave power supply operates at a frequency of 2.45 GHz. Microwave plasmas

Figure 2

Direct current plasma schematic.

44 ATOMIC EMISSION, METHODS AND INSTRUMENTATION

Figure 3 Microwave induced plasma schematic illustrating the ‘Beenakker Cavity’ design.

watts, with frequencies from 200 kHz to 30 MHz. Unlike other MIPs, the CMP is able to handle greater amounts of solvent. However, the CMP has been found to be easily contaminated due to the microwave energy being conducted through a coaxial waveguide to the electrode, which must be replaced regularly.

Spectrometers The atomic emission source will produce unstable, excited atoms or ions, which spontaneously return to a lower energy state. The emission spectrum is produced when a photon of energy is generated during this transition. The basic assumption is that the emitted energy is proportional to the concentration of atoms or ions in the sample. The measurement of this energy is performed using the optics of the spectrometer to isolate the characteristic elemental emission wavelengths, and to separate this radiation from the plasma background. The spectrometer will need a high resolving power to be able to separate the lines of interest from adjacent spectral lines (at least 0.1 nm). The spectrometer consists of (1) a dispersive element, such as a grating, and (2) an image transfer assembly, which contains the entrance and exit slits, and mirrors or lenses. The grating provides dispersion of the wavelength range of interest over a given angular range. Some commonly used grating spectrometers include (1) the PaschenRunge spectrometer, which is used in both sequential and simultaneous instruments and has the advantage of extensive wavelength coverage, (2) the echelle grating spectrometer, with excellent dispersion and resolution with a small footprint, and (3) the Ebert and CzernyTurner spectrometers, which are

similar except that the latter has two mirrors to the Eberts one. The slit allows a narrow line of light to be isolated. The number of lines will represent the various wavelengths emitted by the plasma, with each line corresponding to the image produced by the spectrometer slit, and each wavelength corresponding to a specific element. There are two slits used: (1) the primary slit, which is where light enters the spectrometer, and (2) the secondary slit, where light exits, producing a line isolated from the rest of the spectrum. The imaging system consists either of lenses or of concave mirrors. One of the most useful aspects of atomic emission spectrometry is its capability for multi-element analysis. This can be achieved using either the sequential monochromator, where elements in a sample are quickly read one at a time, or the polychromator, which allows the simultaneous measurement of the elements of interest. Monochromators

Multi-element determinations using a monochromator must be sequential, as the monochromator can only observe one line at a time due to it only having one secondary slit. The slit can be set to scan the wavelengths, which is slower than in a simultaneous instrument, but allows for the selection of a wider range of wavelengths. A scanning monochromator user a movable grating to find known individual spectral lines at fixed positions, typically a Czerny Turner configuration. Although this allows the observance of only one individual wavelength at a time, it also allows all wavelengths in the range of the spectrometer to be observed. Polychromators

Polychromators have a permanently fixed secondary slit for certain individual wavelengths (individual elements). Each slit will have its own detector. This allows for truly simultaneous analyses of selected elements, resulting in much quicker analyses than the monochromator for elements that have a slit installed. However, the selection of possible analysis lines is limited. A simplified diagram of the beam path in a polychromator is shown in Figure 4. A polychromator can focus the emission lines on the circumference of the Rowland circle by using concave gratings; typically the PaschenRunge configuration is used. In summary, the polychromator has a throughput advantage while the sequential scanning monochromator has flexibility as its advantage.

ATOMIC EMISSION, METHODS AND INSTRUMENTATION 45

charge from a full pixel to an adjacent one. CCDs are best used in very sensitive, low light level applications. The CID does not suffer from blooming due to its method of nondestructively measuring the photon-generated charge. The CID allows the monitoring of any wavelength between 165 and 800 nm, whereas the range of the CCD is between 170 and 780 nm.

Methods Sample introduction

Figure 4

Beam path through a polychromator.

Detectors Photomultiplier tubes

A photomultiplier tube (PMT) consists of a photosensitive cathode, several dynodes and a collection anode. The dynodes are responsible for the increase in signal by electron multiplication. PMTs see the elemental line intensity per unit time proportionally with current and have wide dynamic ranges. A PMT, housed in a suitable mechanical movement, can scan the range of wavelengths in a spectrum sequentially, which involves longer analysis times. Alternatively, there can be a series of PMTs, each collecting the signal at discrete wavelengths at its assigned exit slit. Unfortunately, it always takes one PMT for each wavelength of light observed, whether or not the PMT is fixed or roving. Modern instruments have taken advantage of the advent of multichannel solidstate detectors to provide more flexibility in multielement analyses. Charge-coupled and charge-injection devices

The charge-coupled and charge-injection devices are solid-state sensors with integrated silicon circuits. They are similar in that they both collect and store charge generated by the light from the emissions in metal-oxide semiconductor (MOS) capacitors. The amount of charge generated in a charge transfer detector is measured either by moving the charge from the detector element where it is collected to a charge-sensing amplifier (CCD), or by moving it within the detector element and measuring the voltage change induced by this movement (CID). The CCD is susceptible to blooming in the presence of too much light, since there can be an overflow of

Sample introduction into the plasma is a critical part of the analytical process in atomic emission spectroscopy (AES). Since the ICP is the most commonly used source, the sample introduction schemes described below will focus more on it than the other sources mentioned previously. Sample is carried into the plasma at the head of a torch by an inert gas, typically argon, flowing in the centre tube at 0.31.5 L min−1. The sample may be an aerosol, a thermally or spark generated vapour, or a fine powder. Other approaches may also be taken to facilitate the way the analyte reaches the plasma. These procedures include hydride generation and electrothermal vaporization. Torches One of the torch configurations more commonly used with the ICP is shown in Figure 1. The plasma torch consists of three concentric quartz tubes through which streams of argon flow. The nebulizer gas, which carries the analyte into the plasma, flows in the centre tube. The auxiliary gas flows around the centre tube and adjusts the position of the plasma relative to the torch. The coolant gas streams tangentially through the outer tube, serving to cool the inside walls and centre of the torch, and stabilize the plasma. Nebulizers The great majority of analyses in ICP AES are carried out on liquid samples. The most convenient method for liquids to be introduced into the gas stream is as an aerosol from a nebulizer. The aerosol may be formed by the action of a high-speed jet across the tip of a small orifice or by the use of an ultrasonic transducer. A spray chamber is usually placed after the nebulizer to remove some of the larger droplets produced, and thereby improves the stability of the spectral emission. The most commonly used nebulizer designs are pneumatic and ultrasonic, although other types (electrostatic, jet impact, and mono-dispersive generators) have been described. The selection of the appropriate nebulizer depends upon the characteristics of the

46 ATOMIC EMISSION, METHODS AND INSTRUMENTATION

sample: mainly density, viscosity, organic content, total dissolved solids, and total sample volume. Additionally, the performance of a particular nebulizer can be described using several attributes such as droplet size distribution, efficiency, stability, response time, tendency to clog, and memory effects. Concentric nebulizers Solution sample introduction has been associated with pneumatic nebulizers (concentric and cross-flow) almost universally for routine analysis, due to their simplicity and low cost. However, they provide low analyte transport efficiency (< 5%) to the plasma, and may be prone to clogging. The most widely used ICP nebulizer is the onepiece Meinhard concentric nebulizer (Figure 5A). It is a general-purpose nebulizer with low tolerance for total dissolved solids, used for applications requiring low nebulizer gas flows. It operates in the free running mode whereby the solutions are drawn up by the pressure drop generated as the nebulizer gas passes through the orifice. The viscosity of the solution and the vertical distance through which the liquid is lifted affect the rate of liquid transfer. Although the concentric nebulizer is easy to use, the transport efficiency is low, owing to the wide range of droplet sizes produced. In addition, nebulizer blockage may occur due to the presence of suspended solids becoming lodged in the narrow, central sample uptake capillary. Filtering or centrifuging the sample may minimize the risk of blockage. If a sample with a high dissolved-solids content dries in the nebulizer, it may also cause nebulizer blockage or interfere with the operation of the nebulizer by decreasing the signal. Frequent cleaning of the nebulizer solves this problem. Cross-flow nebulizers The cross-flow nebulizer shows much of the same general behaviour of the concentric nebulizers. The cross-flow nebulizer is less prone to blockage and salting effects, although they may still occur as the sample solution passes through a capillary. The cross-flow nebulizer operates when a horizontal jet of gas passes across the top of a vertical tube (Figure 5B). The reduced pressure that is generated draws the liquid up the tube, where, at the top, it is disrupted into a cloud of fine droplets. Babington type nebulizer The Babington nebulizer is designed to allow a film of liquid containing sample to flow over the surface of a sphere (Figure 5C). A gas, which is forced through an aperture beneath the film, produces the aerosol. This design features the liquid sample flowing freely over a small aperture, rather than passing through a fine capillary, and is

Figure 5 Schematic of the (A) concentric nebulizer, (B) crossflow nebulizer and (C) Babington nebulizer.

therefore more tolerant of high dissolved-solids. This kind of nebulizer can be used to introduce slurries into the system since the delivery of the sample is not constrained by a capillary. However, the Babington-type nebulizer shows extensive memory effects since the solution is allowed to wet the entire face of the sphere. A modification of the design is also in use, whereby the liquid sample passes through a V-groove and a gas is introduced from a small hole in the bottom of the groove. The main advantage of the Vgroove nebulizer is its resistance to blockage. However, the design produces aerosol less efficiently, and it is of coarser size distribution than that produced by other concentric nebulizers. Frit-type nebulizer The concentric and cross-flow nebulizers are inefficient at the 1 mL min−1 flow rate at which they usually operate, producing only about 1% of droplets of the correct size to pass into the plasma. An alternative design is the frit nebulizer, which produces droplets with a mean size of 1 µm. The glass frit nebulizer is depicted schematically in Figure 6. The sample flows over the fritted glass disc as the nebulizing gas is passed through the frit. This

ATOMIC EMISSION, METHODS AND INSTRUMENTATION 47

Figure 7 Ultrasonic nebulizer schematic. Reprint permission from Ingle JD and Crouch SR (1988) Spectrochemical Analysis. p. 194, Figure 7-6. Englewood Cliffs, NJ: Prentice Hall.

Figure 6 Glass frit nebulizer schematic. Reprint permission from Ingle JD and Crouch SR (1988) Spectrochemical Analysis. p. 193, Figure 7-5. Englewood Cliffs, NJ: Prentice Hall.

design provides an excellent fine aerosol, but at the expense of the fritted disc clogging over time. The transport efficiency is very high compared to the pneumatic-type nebulizers but the operation is at flows of 100 µL min−1 or less. Ultrasonic nebulizer The ultrasonic nebulizer also produces fine aerosol and has a high sample-transport efficiency (Figure 7). The aerosol is generated when the solution is fed to the surface of a piezoelectric transducer which operates at a frequency between 0.2 and 10 MHz. Vibrations of the transducer cause the solution to break into small droplets which are then transported to the plasma. The ultrasonic nebulizer offers highly efficient aerosol production independent of the carrier-gas flow rate. With this nebulizer, more analyte can be transported into the plasma at a lower nebulizer gas flow rate than that seen with pneumatic nebulizers. This increases the sensitivity and improves detection limits since samples have a longer residence time in the plasma. However, the high efficiency also allows more water to enter the plasma, producing a cooling effect, which may decrease analyte ionization. Hence these nebulizers require desolvation to be realistically utilized. Apart from the analytical shortcomings of the ultrasonic nebulizer, the high cost of the commercial systems may be prohibitive.

Electrothermal vaporization Electrothermal vaporization can introduce liquid, solid, and gaseous (by trapping) samples to the plasma with high analyte transport efficiency (2080%) and improved sensitivity over pneumatic nebulization. A small amount (5100 µL) of sample is deposited into an electrically conductive vaporization cell (Figure 8). Initially, a low current is applied to the cell, which causes resistive heating to occur and dries the sample. A high current is then passed for a short time (typically 5 s) to completely vaporize the sample. An optional ashing stage may be used to remove some of the matrix prior to the analyte vaporization stage. A stream of argon gas is passed through the unit and carries the sample vapour to the plasma. A variable current supply is required for controlling cell heating, and the sequence of heating steps is carried out using an electronic control system. Electrothermal vaporization allows sample matrix components, including the solvent, to be separated from the analytes of interest through judicious selection of temperature programming steps. This may reduce oxide formation and the number of spectroscopic interferences. Additionally, samples with high salt content and limited volume can be analysed with electrothermal vaporization. Hydride generation The problems associated with liquid sample introduction, and the inefficiencies in analyte transport can be overcome by presenting the sample in a gaseous form to the plasma. Introducing the analyte in a gaseous form eliminates the use of a nebulizer and a spray chamber. It also provides nearly 100% analyte transport efficiency, which subsequently improves the detection limits. Additional

48 ATOMIC EMISSION, METHODS AND INSTRUMENTATION

Figure 9 Hydride generation apparatus. Reprint permission from Ingle JD and Crouch SR (1988) Spectrochemical Analysis. p. 280, Figure 10-6. Englewood Cliffs, NJ: Prentice Hall.

Figure 8 furnace.

Illustration of electrothermal vaporization graphite

benefits include no blockage problems, the possibility of matrix separation and analyte preconcentration, and the absence of water, which reduces the levels of many polyatomic ion species. The hydrides of the elements arsenic, bismuth, germanium, lead, antimony, selenium, and tin are easily formed in acidic solutions and are gaseous at room temperature. The most frequently used reaction is:

where E is the hydride forming element of interest. Figure 9 shows an apparatus for hydride generation. It allows the mixing of the sample and the reducing agent, followed by the transport of the volatile analyte species (hydride) by the carrier gas to the plasma. The gaseous sample introduction line is connected directly to the central tube of the plasma torch, eliminating the need for the conventional nebulizer and spray chamber. Additional equipment to handle gas mixing and dilution at controlled flow rates may be required. Direct solids introduction Direct insertion of solid samples into the ICP can be used for metal powders, salts, and geological samples. The sample is placed on a wire loop or cup made from graphite,

molybdenum, tantalum, or tungsten. The probe is then moved along the axis of the torch and closer to the plasma where drying takes place. At the completion of this stage, the device is propelled rapidly into the core of the plasma and measurements of the analytes can be taken. Line selection

The choice of which line to use for a given sample type is a difficult one, as most elements have many lines available for analysis. Not only should the intensity of the line be considered, but also whether the line is free of spectral interferences from both the plasma and other sample constituents. Additionally, the nature of the emission, atomic or ionic, should be considered. The emission could originate from an excited neutral atom, which is termed an atomic transition, or an excited ionized form of the element, which is called an ionic transition. While analysis of elements undergoing transitions of the ionic type tend to experience fewer consequences due to changes in the plasma operating conditions, there are elements that produce no ionic lines, such as aluminium and boron. There are very good reference books which list tables of spectral lines and are included in the Further reading section. Often modern instrumentation has sufficiently intelligent software to assist the operator with line selection. Spectral interferences and correction techniques

A spectrometer with good resolution will help greatly in the separation of adjacent lines from the

ATOMIC EMISSION, METHODS AND INSTRUMENTATION 49

Figure 10 Interelement effects of plutonium on calcium (top) and aluminium (bottom).

spectral line of interest. The effect of partial overlap can be minimized in some cases with a high-resolution spectrometer, and in other cases, can be overcome using correction techniques. When the interference is from the plasma emission background, there are background correction options available with most commercial instrumentation. The region adjacent to the line of interest can be monitored and subtracted from the overall intensity of the line. If direct spectral overlap is present, and there are no alternative suitable lines, the interelement equivalent concentration (IEC) correction technique can be employed. This is the intensity observed at an analyte wavelength in the presence of 1000 mg L−1 of an interfering species. It is expressed mathematically as:

Figure 11 Interelement equivalent concentration correction factors for several interfering elements on calcium (top) and aluminium (bottom) as calculated by a program written in-house by Gerth DJ.

some instrument manufacturers software will automatically perform the necessary corrections. Examples of the effect of 1000 ppm plutonium as the interfering element on calcium and aluminium are shown in Figure 10. This shows the subarrays obtained for separate solutions of plutonium, calcium, and aluminium. Graphical representations of interfering elements on calcium and aluminium are shown in Figure 11. These graphs are produced by a program written at Los Alamos National Laboratory by DJ Gerth, and illustrate which interfering elements have the strongest effect on the analytes. Those elements with the strongest deviation from the average are those needing the interelement correction.

List of symbols where the correction is in milligrams of analyte per litre of solution, and Iinit is the intensity read at the analyte wavelength in the presence of 1000 mg L−1 of the interfering species, Ia is the intensity the instrument will produce for a certain analyte concentration at the analyte wavelength, and Ca is the concentration in mg L−1 used to give Ia. This correction can be applied to all the analytes in a method and

Ca = analyte concentration used to give Ia; Ia = intensity at certain analyte concentration and Iinit = intensity in presence of wavelength; 1000 mg L−1 of the interfering species. See also: Electronic Components, Applications of Atomic Spectroscopy; Fluorescence and Emission Spectroscopy, Theory; Inductively Coupled Plasma Mass Spectrometry, Methods.

50 ATOMIC FLUORESCENCE, METHODS AND INSTRUMENTATION

Further reading Boumans PWJM (1987) Inductively Coupled Plasma Emission Spectroscopy Part 1, Chemical Analysis Series, Vol. 90. Chichester: Wiley. Edelson MC, DeKalb EL, Winge RK and Fassel VA (1986) Atlas of Atomic Spectral Lines of Plutonium Emitted by an Inductively Coupled Plasma. Ames Laboratory, US Department of Energy. Ingle JD and Crouch SR (1988) Spectrochemical Analysis. Englewood Cliffs, NJ: Prentice-Hall. Montaser A and Golightly DW (eds) (1992) Inductively Coupled Plasmas in Analytical Atomic Spectrometry Weinheim: VCH. Slickers K (1993) Automatic Atomic Emission Spectroscopy. Brühlsche Universitätsdruckerei.

Sneddon J (ed) (1990) Sample Introduction in Atomic Spectroscopy, Analytical Spectroscopy Library, #4. Amsterdam: Elsevier. Sweedler JV, Ratzlaff KL and Denton MB (eds) (1994) Charge-Transfer Devices in Spectroscopy. Weinheim: VCH. Thompson M and Walsh JN (1989) Handbook of Inductively Coupled Plasma Spectrometry. Glasgow: Blackie. Varma A (1991) Handbook of Inductively Coupled Plasma Atomic Emission Spectroscopy. Boca Raton, FL: CRC Press. Winge RK, Fassel VA, Peterson VJ and Floyd MA (1985) Inductively Coupled Plasma-Atomic Emission Spectroscopy, An Atlas of Spectral Information. Amsterdam: Elsevier.

Atomic Fluorescence, Methods and Instrumentation Steve J Hill and Andy S Fisher, University of Plymouth, UK

ATOMIC SPECTROSCOPY Methods & Instrumentation

Introduction

Light sources

Atomic fluorescence spectroscopy (AFS) has been used for elemental analysis for several decades. It has better sensitivity than many atomic absorption techniques and offers a substantially longer linear range. However, despite these advantages, it has not gained the widespread usage of atomic absorption or emission techniques. In recent years, the use of AFS has been boosted by the production of specialist equipment that is capable of determining individual analytes at very low concentrations (at the ng L −1 level). The analytes have tended to be introduced in a gaseous form and hence sample transport efficiency to the atom cell is very high. This article describes the instrumentation and methods available for atomic fluorescence spectroscopy, although it should be emphasized that much of the instrumentation associated with this technique is often very similar to that used for atomic absorption spectroscopy (AAS). A schematic diagram of the different parts of an AFS instrumentation is shown in Figure 1. It can be seen that the light source, atom cell, line isolation device, detector and readout system used for AFS are very similar to those used in AAS, although there may be subtle differences and the components may be less sophisticated, as described below.

As can be seen from Figure 1, light sources for AFS are placed at right angles to the detector. This is to ensure that incidental radiation from the source is not detected. Since a more intense source will lead to greater sensitivity, in AFS a standard hollow-cathode lamp (HCL) is often insufficient for the majority of applications. A range of alternatives has therefore been developed. An electrodeless discharge lamp (EDL) is suitable because it has an intensity 2002000 times greater than that of a hollow-cathode lamp. Boosted-discharge hollow-cathode lamps (BDHCL)

Figure 1

A schematic representation of an AFS instrument.

ATOMIC FLUORESCENCE, METHODS AND INSTRUMENTATION 51

are similar to standard hollow-cathode lamps but can be operated at greatly increased current. The light output is consequently several times more intense. A similar effect may be obtained by pulsing a standard hollow-cathode lamp to 1001000 mA, although this will reduce the lifetime of the lamp. Continuum sources such as 150500 W xenon lamps have also been used, but there are several problems associated with their use. The most obvious drawback is the relatively poor intensity offered at each individual wavelength by the black body irradiator. The problem is particularly bad in the ultraviolet region of the spectrum. Scatter has also been reported as being problematic. Vapour discharge lamps have been available for some elements (e.g. Cd, Ga, Hg, In, K, Na and Tl) since the inception of commercial instrumentation, although such devices are limited to the volatile elements and are therefore not universal. Lasers have the greatest output intensity and would therefore be expected to yield the highest sensitivity. Unfortunately, they are expensive and have, in the past, been produced to give light of very specific wavelengths, which may not correspond to the frequency required to excite the analyte atoms. The development of the tunable dye laser has helped to overcome this problem. Frequency doubling (i.e., halving the wavelength of the light output) has enabled wavelengths well into the ultraviolet region to be obtained. A nitrogen laser pump can be used to obtain a wavelength of 220 nm while a Nd:YAG (neodymium:yttrium aluminium garnet) laser allows 180 nm to be reached. An added advantage of highintensity lasers is that they may cause saturation fluorescence (population inversion), which nullifies the effects of quenching and self-absorption. Another source that has been used to initiate fluorescence is an inductively coupled plasma. This may be used either as a continuum source (if the actual fireball is used), or as a line source (if the tail-flame is used). The use of an ICP as a light source to initiate fluorescence is not yet a routine technique. As with AAS, modulation of the light source enables an optimal signal-to-noise ratio to be obtained. The frequency of modulation depends on the light source, but may be up to 10 kHz for an EDL.

Atom cell Several atom cells have been used for AFS. The majority of applications have used a circular flame atom cell (sometimes with a mirror placed to collect as much light as possible), but more recently inductively coupled plasmas and electrothermal atomizers

have also been used. There is a risk of self-absorption at high analyte concentrations, i.e. light fluoresced by some atoms will be absorbed by others in different parts of the flame. A circular flame has a smaller path length compared with a slot burner, hence its popularity. A variety of fuels have been used for the flames. Airacetylene, nitrous oxide acetylene and argonhydrogen diffusion flames have all been used successfully. Quenching (the nonradiative loss of energy) increases with increasing flame temperature and so the cooler airacetylene and hydrogen flames are often used. Conversely, the hot nitrous oxideacetylene flame aids the atomization of refractory compounds, The quenching crosssection of the colliding particles also effects the amount of quenching. The relative quenching cross-section of the different flame gases increases in the order argon (negligible) < hydrogen (low) < oxygen (high). It can therefore be seen that for easily atomized analytes an argonhydrogen diffusion flame offers at the best properties for atomic fluorescence. Inductively coupled plasmas atomize refractory analytes very efficiently because of their high temperature (8000 K) and therefore have a very high fluorescent yield. Care must be taken, however, to ensure that ionization of the analyte species does not occur. An added advantage is that the plasma is made up of argon and is therefore optically thin (i.e. it has a low quenching cross-section). The disadvantages of using a plasma as the atom cell are that it is more expensive and the instrumentation is more complex. When a plasma is used as an atom cell, the light source may be placed in one of two orientations. It may pass light through the side of the plasma although more recently it has been used axially, i.e. the source passes light through the entire length of the plasma. Since the second configuration has a longer path length, more analyte atoms may be excited and hence improved detection limits are obtained. The disadvantage with this configuration is that the light source must be protected in some way from the extreme temperatures of the plasma. This may be achieved by using either a water-cooled condenser containing a quartz window or by using a protective gas flow to protect the lens from the plasma tail flame. If mirrors or some other collection device are placed around the plasma, then all fluoresced light may be collected and directed to the detector. It has been noted that the optimum viewing height for fluorescence measurements using an ICP as the atom cell occurs substantially higher above the load coil than for emission measurements. Observation above 3 cm is routine and making measurements from the tail flame is common to ensure that a signal above the background and analyte emission noise is obtained.

52 ATOMIC FLUORESCENCE, METHODS AND INSTRUMENTATION

Similarly, a decrease in plasma power has also been found to be beneficial. Electrothermal atomizers for AFS share many of the advantages associated with their use for AAS. The argon inert gas that prevents oxidation of the graphite tube ensures that minimal quenching occurs, although a number of other interferences may occur. By and large, the interferences are very similar to those experienced in traditional electrothermal AAS, such as carbide formation (for some analytes), scatter by particulate matter, and losses during thermal pretreatment. A more comprehensive overview of interferences may be found in the sections on AAS. The use of matrix modifiers to assist in the separation of the analyte from the matrix is still often necessary. Where a laser is used as light source, a technique termed laser-induced fluorescence electrothermal atomization (LIF-ETA) is possible. A schematic diagram of a typical LIF-ETA system is shown in Figure 2A. The light source passes through a pierced mirror and into the atomizer. Light fluoresced from the analytes within the atomizer then reflect off the mirror and onto the detector. An alternative configuration is shown in Figure 2B, in which the laser irradiates the sample through the injection port. The two configurations are known as longitudinal and transverse geometries, respectively. The technique is

capable of extremely low detection limits (below 1 pg g−1) and has therefore found use in the nuclear, biomedical, electronics and semiconductor industries. A heated atom cell in not required in the case of mercury determinations. Mercury produces an atomic vapour at room temperature and therefore does not require a flame to cause atomization. Instead, the mercury vapour may simply be transported to an atom cell at room temperature. The radiation from the light source excites the mercury atoms in the normal way.

Line isolation devices A conventional monochromator (either Ebert, CzernyTurner, Littrow or Echelle) may be used (see AAS sections for details), but some of the more basic instrumentation uses interference filters. These are optical filters that remove large bands of radiation in a nondispersive way. A dispersion element such as a prism or a grating is therefore not required. Only a relatively narrow band of radiation is allowed to pass to the detector. The disadvantage with such devices is that they are not particularly efficient and hence much of the fluoresced light is lost. An alternative development is the multi-reflectance filter. This is shown diagramatically in Figure 3, and has the

Figure 2 (A) Schematic diagram of longitudinal laser-induced fluorescence with electrothermal atomization. (B) Transverse laserinduced fluorescence. From Ebdon L (1998) Introduction to Analytical Atomic Spectrometry. Reproduced by permission of John Wiley & Sons Limited.

ATOMIC FLUORESCENCE, METHODS AND INSTRUMENTATION 53

advantages of being 80% efficient at transmitting the wavelengths of interest while virtually eliminating background noise. Such a device therefore has a superior performance to other wavelength filters. Interference and multi-reflectance filters may be used when the analyte of interest has been separated from the matrix and all the concomitant elements. In such a case the analyte atoms are the only elements that will fluoresce when light from a monochromatic light source is used to excite them. Powerful dispersion gratings are therefore not required because, theoretically, light specific to only one element is produced. A schematic diagram of a very basic instrument that uses a multi-reflectance filter is shown in Figure 4. If other analytes are present in the atom cell, then either a monochromator is required to separate the other wavelengths (produced by emission), or, the source must be modulated so that distinction between emission and fluorescence signals may be made.

Detectors The light isolated by the line isolation device (monochromator, interference filter or multi-reflectance filter) is normally detected using a photomultiplier tube (PMT). A description of how these devices work may be found in the sections on AAS. It should be noted that some basic instruments (e.g. as shown in Figure 4), often use a solar-blind photomultiplier tube. This is a PMT that detects radiation in the UV region but does not detect light in the visible region of the spectrum.

Readout systems and data handling The readout and data-handling systems used for AFS are similar to those found for AAS instrumentation. The signal from the photomultiplier tube passes through an amplifier and then on to the readout device. This may be a digital display, a chart recorder, an integrator, or a computer with software dedicated to the instrument. The last alternative is the most commonly used in modern instrumentation. The software is likely to be able to control calibration, plot graphs, perform quality assurance tests and calculate means and standard deviations of the results. One advantage of using a computer to control the analysis is that the use of an autosampler becomes a possibility. This enables unattended operation of the instrumentation.

Applications

Figure 3

A multi-reflectance filter.

Figure 4 Schematic of a very basic nondispersive AF spectrometer.

As described previously, although techniques to utilize AFS were developed several decades ago, until recently they were not widely used. However, a new generation of simple AFS instruments have been developed to specifically detect the vapour-forming elements, such as those that form hydrides (As and Se) and mercury, which forms an atomic vapour. All of these analytes have a primary line below 260 nm and, since the analytes may be readily separated from the bulk matrix and concomitant elements, dispersion is not necessary. Instead these basic instruments originally used a simple interference filter, although these have now been superseded by the more efficient multi-reflectance filter. The theory behind hydride generation is beyond the scope of this text, but in brief, if a sample is mixed with a reducing agent, such as sodium tetrahydroborate, then some elements, e.g. As, Bi, Ge, Pb, Se, Sb, Sn and Te, will form hydrides. It must be noted that the efficiency of hydride formation depends on the oxidation state of the analyte. Arsenic(III)

54 ATOMIC FLUORESCENCE, METHODS AND INSTRUMENTATION

forms a hydride more efficiently than As(V), and Se(IV) forms a hydride whereas Se(VI) does not. It is therefore necessary to ensure that all the analyte atoms are in the same oxidation state; this is normally achieved using one of a variety of reducing agents. Once formed, the hydride leaves the reaction cell and is flushed towards the AFS instrument using an inert gas (usually argon). During this process, hydrogen is formed as a breakdown product of tetrahydroborate, and can be swept along with the hydride in the argon to the atom cell, where it can be used as the fuel for the hydrogenargon diffusion flame. More hydrogen may be added as necessary from an external source. Since a hydrogenargon diffusion flame is used, the amount of water entering the flame must be minimal (otherwise quenching and flame instability result). The hydride and water vapour are therefore separated, often by passing through a membrane drier tube. This is a tube containing a semi-permeable hydroscopic membrane surrounded by a counterflow of dry gas. Water vapour in the hydride passes through the membrane and is removed by the countergas flow. The dry hydride continues through the tube to be flushed toward the AFS instrument by the flow of argon. Mercury may be determined using very similar instrumentation, but it may also be introduced in a number of other ways. Since mercury is volatile, solid samples may be placed on a small platform and then heated. The sample is thus vaporized and the vapours are passed through a trap (sand coated with gold). The mercury is trapped onto the gold while the smoke and matrix constituents pass through and are hence separated from the analyte. Heating of the gold trap to 900°C releases the mercury, which may then be swept in a flow of argon to the AFS instrument. This technique has the advantage that the mercury may be preconcentrated on the trap, since several repeat samples may be vaporized and the mercury collected before it is subsequently removed from the trap. The use of the gold trap preconcentrator is equally applicable to the analysis of liquid samples. When a gold trap preconcentrator is used, the limit of detection for mercury is substantially below 1 ng L−1, which is superior to that obtained using atomic absorption, emission or inductively coupled plasmamass spectrometric (ICP-MS) techniques. The instrumentation described above may also be used for speciation studies. In such cases a separation technique such as high-performance liquid chromatography (HPLC) or gas chromatography (GC) is coupled with the AFS instrument. The chromatography is used to separate different analyte species and some on-line chemistry is subsequently required to convert the species into a chemical form suitable for

vapour generation. Although the chemistry behind these methods of speciation is beyond the scope of this text, it should be emphasized that the method used should be powerful enough to convert even very stable species, such as arsenobetaine or selenomethionine into atomic arsenic and selenium. The additional use of photolysis may therefore be necessary. Similarly, pyrolysis may be necessary to convert different mercury species into atomic mercury since this is the only species that will fluoresce. As described previously, vapour introduction approaches are by far the most common application of atomic fluorescence. Despite this, mention of other methods should be made. If a conventional nebulizer and spray chamber assembly (see AAS section) is used, it is possible to introduce liquid samples directly to the atom cell. In circumstances such as these, it is necessary to use more robust airacetylene or nitrous oxideacetylene flame, or perhaps an ICP. The use of an ICP as an atom cell for AFS measurements has led to the development of a number of different techniques, e.g. ASIA, an acronym for atomiser, source, inductively coupled plasmas in AFS. This technique uses a high-powered ICP as a source and a lowpowered ICP for the atom cell. It has been found that ICP-AFS yields linear calibrations over 46 orders of magnitude and is more sensitive than ICP-AES. Several techniques have been developed that use an electrothermal atomizer as an atom cell. Laserinduced fluorescence (LIF) has been discussed previously, but other techniques such as laser-excited atomic fluorescence spectroscopy (LEAFS) also exist, although not used routinely.

Interferences Atomic fluorescence has the advantage of being less prone to spectral interferences than either AES or AAS. Molecular fluorescence is less of a problem than molecular absorption is in AAS. Scatter from the light source and quenching from the gaseous species in the atom cell are often the major sources of interference. For many applications where the analyte is separated from the matrix (e.g. vapour generation) chemical interferences may exist; for example, the presence of high concentrations of some transition metals may interfere in the hydride formation process. This will inevitably lead to errors in the measurement unless preventative steps are taken.

Detection limits A discussion of the detection limits for the different approaches utilizing atomic fluorescence is difficult

ATOMIC FLUORESCENCE, METHODS AND INSTRUMENTATION 55

because they depend so heavily on the type of source used. Similarly, samples introduced in the vapour phase tend to yield substantially improved limits of detection (LODs) when compared with those introduced as a liquid. In general, if an EDL is used as a source and a circular flame as an atom cell, LODs at the µg L−1 level are obtainable. Improvement will be made if a laser is used as a source. If the sample is introduced as a vapour, LODs at the ng L−1 level are obtained. The use of an ICP as an atom cell leads to LODs at the µg L−1 level if a normal HCL is used as the source, improving by approximately an order of magnitude if a boosted discharge HCL is used. If an electrothermal atomizer is used in conjunction with a laser, the best LODs of all are obtained.

Conclusions Atomic fluorescence spectroscopy is a popular technique for those analytes that readily form vapours, and specialized instrumentation is now available for individual elements. Such instruments are simple to operate, easily automated, and offer good sensitivity and freedom from interferences. The use of other flame-based fluorescence techniques has waned considerably over the years, but research continues

into the use of electrothermal atomizers and inductively coupled plasmas as atom cells, reflecting the superior limits of detection that are potentially available. The availability of tuneable lasers may also encourage the resurgence of AFS as a routine analytical tool. See also: Atomic Absorption, Methods and Instrumentation; Light Sources and Optics; UV-Visible Absorption and Fluorescence Spectrometers.

Further reading Greenfield S (1994) Inductively coupled plasmas in atomic fluorescence spectrometry. A review. Journal of Analytical Atomic Spectrometry 9: 565592. Greenfield S (1995) Atomic fluorescence spectrometry progress and future prospects. Trends in Analytical Chemistry 14: 435442. Hill SJ, Dawson JB, Price WJ, Shuttler IL, Smith CMM and Tyson JF (1998) Advances in atomic absorption and fluorescence spectrometry and related techniques. Journal of Analytical Atomic Spectrometry 13: 131R 170R. Kirkbright GF and Sargent M (1974) Atomic Absorption and Fluorescence Spectroscopy. London: Academic Press.

Atomic Force Microscopes See

Scanning Probe Microscopes.

Atomic Force Microscopy, Theory See

Scanning Probe Microscopy, Theory.

56 ATOMIC SPECTROSCOPY, HISTORICAL PERSPECTIVE

Professor B. V. Lvov, the inventor of the very powerful analytical technique graphite furnace atomic absorption spectroscopy, has rightly pointed out in the Introduction of his definitive book entitled Atomic Absorption Spectrochemical Analysis that, The discovery of atomic absorption and the history of research into it are integral parts of the entire history of spectroscopy and spectrochemical analysis. Indeed, the early history of atomic spectroscopy, as far as spectrochemical analysis was concerned, consisted of the development of emission spectrochemical analysis, which was usually dependent on Fraunhofer lines (which are atomic absorption lines) for wavelength calibration. Following Newtons study of the spectrum of the sun in 1666, there was a period of almost 136 years entirely concerned with emission spectroscopy. Only in 1802 did Wollaston report the presence of dark bands in the continuum emission spectrum of the sun and, after a more detailed study by Fraunhofer (1814), Brewster (1820) was able to ascribe them to absorption of radiation within the suns atmosphere. Another 40 years passed before Kirchhoff and Bunsen showed that one of these dark bands in the emission spectrum of the sun corresponded exactly to the yellow emission band obtained when sodium vapour is heated in a flame. This led Kirchhoff to deduce that the Fraunhofer lines in the solar spectrum are absorption lines of elements whose flame emission spectra would contain lines at exactly the same position in the spectrum. Their work enabled Kirchhoff to develop the fundamental relationship between emission and absorption spectra: any species that can be excited to emit radiation at a particular wavelength will also absorb radiation at that wavelength. Thus Kirchhoff not only laid the foundations of atomic absorption methods of chemical analysis but also gave a striking example of their power. Indeed, it is difficult to imagine a more convincing and dramatic demonstration. Bunsen and Kirchhoff are thus rightfully considered to be founders of spectrochemical analysis. However, it is surprising that almost a century after the work of Bunsen and Kirchhoff the potentialities of atomic absorption measurements remained unexplored and unsuspected. Why?

ATOMIC SPECTROSCOPY Historical Overview As Alan Walsh has presented in his perceptive analysis of the reasons, it seems likely that one reason for neglecting atomic absorption methods was that Bunsen and Kirchhoffs work was restricted to visual observation of the spectra. In such visual methods the sensitivity of emission methods was probably better than that of absorption. Not only was the photographic recording of the spectra more tedious, but also the theory seemed to indicate they would only prove useful for quantitative analysis if observed under very high resolution. But possibly a more fundamental reason for neglecting atomic absorption is related to Kirchhoffs law (1859), which states that the ratio of the emissive power E and absorptive power A of a body depends only on the temperature of the body and not on its nature. Otherwise radiative equilibrium could not exist within a cavity containing substances of different kinds. The law is usually expressed as

where K(λT) is a function of wavelength and temperature. This law is perfectly correct but unfortunately much and possibly most of the subsequent teaching concerning it has been misleading. In most textbooks and lessons on this subject the enunciation of the law is followed by a statement to the effect that this means good radiators are good absorbers, poor radiators are poor absorbers. This statement is patently absurd without any reference to temperature or wavelength. This erroneous conclusion has been made, presumably, because Kirchhoffs constant, K, as he so clearly pointed out, is a function of wavelength and temperature. He was unable to find an analytical expression for it and it was not deduced until 1900 by Kirchhoffs successor at the University of Berlin, Max Planck, as Plancks distribution function. Many spectroscopists were misled by the widely held but misleading assumptions regarding the implications of Kirchhoffs law. What is even more surprising is that numerous spectroscopists wrote papers on atomic

ATOMIC SPECTROSCOPY, HISTORICAL PERSPECTIVE 57

emission methods and referred to the problems caused by the effects of self-absorption and self-reversal, and yet failed to make the small-step connection between atomic emission and atomic absorption! It was left to Alan Walsh, whose research experience in atomic emission spectroscopy and molecular absorption spectroscopy virtually compelled him to see the obvious connection. Walsh expressed this experience in his inimitable way in the following words: It appears to be true that having an idea is not necessarily the result of some mental leap: it is often the result of merely being able, for one sublime moment, to avoid being stupid!. Walsh in 1953 and Alkemade and Milatz in 1955 independently published papers indicating the substantial advantage of atomic absorption methods over emission methods for quantitative spectrochemical analysis. Alkemade and Milatz considered only the selectivity in atomic absorption methods, but Walsh discussed general problems of development of absolute methods of analysis. During his early experiments with flame atomizerburners, Walsh encountered the problems which arise when measuring atomic absorption using a continuum source, which might have been responsible for the neglect of atomic absorption methods. The use of a sharp-line source such as a hollow-cathode lamp not only solved this problem, it also provided the atomic absorption method with one of its important advantages, that is, the ease and certainty with which one can isolate the analysis line. The concept of putting the resolution in the source also permitted the use of a simple monochromator since the function of the monochromator, in such a case, is only to isolate the analysis line from the neighbouring lines and the background. Graphite furnace atomic absorption spectroscopy (GFAAS) has been widely used as a spectrochemical trace-analytical technique during the last 30 years. Lvov pioneered most of the theoretical and experimental developments in GFAAS and provided a masterly treatment of the possibilities of GFAAS in a paper entitled Electrothermal atomizationthe way towards absolute methods of atomic absorption analysis. RE Sturgeon, who has been responsible for much of the development of GFAAS, presented a critical analysis of what it does and what it cannot do. AKh Gilmutdinov, who did much of the theoretical development of GFAAS in the 1990s, also elucidated the complex processes and reactions that occur in an electrothermal atomizer by digital imaging of atomization processes using a charge-coupled device camera in my research laboratories. Atomic spectroscopy in the three variations that are most commonly used in spectrochemical analysis, atomic absorption, atomic emission and atomic

fluorescence, are all mature techniques, with their particular areas of strengths and weaknesses now well recognized. Many a battle has been fought and won to establish the superiority of one or the other technique over the rival techniques. It is sometimes claimed that Inductively Coupled Plasma mass spectrometry (ICP-MS), which has the winning combination of high-temperature ionization of elements in a plasma with the detection of the ions by mass spectrometry, is the ultimate trace-analytical technique that will triumph over the rival techniques. However, such claims are based on the limited applications of these techniques by practitioners whose allegiance to their own techniques gives more credit to their loyalty than to their scientific objectivity, as elaborated in the following paragraph. Some have also predicted the imminent demise of graphite furnace atomic absorption spectroscopy as a traceanalytical technique. Such a prediction, based as it is on inadequate comprehension of the enormous complexities and extreme diversities of real-life situations requiring a variety of trace-analytical techniques which possesses some special capabilities not possessed by many analytical techniques, is destined to be false, as is shown below. In my research laboratories in recent years, my 10 PhD students and three adjunct research professors have been doing research on the chemical speciation of potentially toxic elements in the aquatic, the atmospheric and the terrestrial environment. Because freshwaters and soils are systems that are usually far removed from equilibrium, our research is mostly directed to the development of kinetic schemes of chemical speciation which require routine use of inductively coupled plasma mass spectrometry and graphite furnace atomic absorption spectroscopy to measure the kinetics of metal uptake from aqueous environmental samples. As such, we routinely make hundreds of determinations of trace and ultratrace elements every week. For these determinations, we have in our research laboratories the latest models of two ICP-MS and several GFAAS systems. We have made objective studies of the performance characteristics of ICP-MS and GFAAS for kinetic studies of real-life, aqueous, environmental samples over a period covering 6 years from 1993 to 1998, and have come to the following inescapable conclusions. The analytical sensitivities of both ICP-MS and GFAAS for toxic metals are comparable. For kinetic runs on aqueous, environmental samples, ICP-MS has the great advantage of providing continuous sampling with numerous data points having adequate time resolution, whereas GFAAS provides discrete sampling with fewer data points having inadequate time resolution. However, in the most important and decisive

58 ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS

criterion of sample compatibility with the technique, ICP-MS fails completely on account of its inability to handle any solutions containing significant amounts of solute materials and/or corrosive chemicals such as hydrofluoric acid and strong inorganic acids and bases, whereas GFAAS can readily handle such samples. Since the aqueous, environmental samples for kinetic runs are always pretreated with solutes in the form of acidbase buffers, and since such solutions cannot be diluted in order to make them acceptable to ICP-MS, it is not possible to use ICP-MS for such kinetic studies without fouling the interior of the ICP-MS equipment with encrustations of the solute materials, which do serious damage. The alternative then is GFAAS, in spite of all its limitations. Because of the easy compatibility of GFAAS with the difficult sample type described above, GFAAS will continue to be used, as it is used now, until another new analytical technique which has all the advantages of ICP-MS without its fatal deficiencies mentioned above, is invented, developed and tested using real-life samples, i.e. for a long time.

See also: Atomic Absorption, Theory; Atomic Emission, Methods and Instrumentation; Atomic Fluorescence, Methods and Instrumentation; Inductively Coupled Plasma Mass Spectrometry, Methods.

Further reading Chakrabarti CL, Gilmutdinov AKh and Hutton JC (1993). Digital imaging of atomization processes in electrothermal atomizer for atomic absorption spectrometry. Analytical Chemistry 65: 716723. Lvov BV (1970) Atomic Absorption Spectrochemical Analysis. London: Adam Hilger. Lvov BV (1978) Electrothermal atomizationthe way toward absolute methods of atomic absorption analysis. Spectrochimica Acta, Part B 33: 153193. Sturgeon R (1986) Graphite furnace atomic absorption spectrometry: fact and fiction. Fresenius Zeitschrift für Analytische Chemie 324: 807818. Walsh A (1980) Atomic absorption spectroscopysome personal reflections and speculations. Spectrochimica Actya, Part B 35: 639642.

In Memory of N Jim Harrick.

Introduction Chemical reactions that occur at gassolid and liquidsolid interfaces are of central importance to a variety of research and technological areas, including biomembranes, drug design, drugmembrane interaction, biosensors, chemical sensors, heterogeneous catalysis, thin film growth, semiconductor processing, corrosion and lubrication. Many methods are used for interface studies, ranging from most simple ones like the octanolwater two-phase system for mimicking the partition of a drug between a biomembrane and the surrounding water, to most specialized and expensive techniques such as low-energy electron diffraction (LEED), Auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS/ ESCA) and ion scattering spectroscopy (ISS). Among this palette of techniques, optical reflection spectroscopy in the mid- and near-IR range

VIBRATIONAL, ROTATIONAL & RAMAN SPECTROSCOPIES Applications occupies an important complementary position. The basic equipment consists of a commercial IR spectrometer and a suitable reflection accessory that usually fits into the sample compartment of the spectrometer. Many reflection techniques permit in situ applications, and if applied in the mid IR, result in quantitative and structural information on a molecular level. Moreover, IR reflection spectroscopy features a very high performance-to-price ratio. There is a wide range of different spectroscopic reflection techniques. First one should distinguish between internal (total) and external reflection. Attenuated total reflection (ATR) belongs to the first group. It makes use of the evanescent wave existing at the interface of the IR waveguide and the sample. Commercial ATR attachments differ mainly in shape and mounting of the internal reflection element (IRE) in the light path. Most IREs enable multiple internal reflections, a prerequisite for monolayer and

ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS 59

sub-monolayer spectroscopy, and a referred to as MIRE. A variety of external reflection techniques are in use. In specular reflection (SR) the radiation reflected from the front surface of a bulk sample is collected. SR is often measured at or near normal incidence. Reflected spectral energy depends on the absorption behaviour of the sample. In regions of strong absorption the reflected energy is enhanced with respect to non-absorbing spectral regions, moreover, the reflection spectrum is usually very different from a corresponding absorption (AB) spectrum obtained by a transmission experiment. AB spectra may, however, be calculated from SR spectra by means of the KramersKronig transformation (KKT). Corresponding software for SR data processing is supplied with most commercial IR instruments. While specular reflectance is measured at or near normal incidence, IR reflectionabsorption spectroscopy (IRRAS) works from about 10° to grazing incidence. In this case the sample is placed on a reflecting substrate, usually a metal. The portion of reflected light from the sample surface is generally small compared with the energy reflected off the metal surface. Therefore, IRRAS data and transmission (T) data are analogous. From Fresnels equations (see below), it follows that parallel (||) and perpendicular (A) polarized electromagnetic waves undergo different phase shifts upon reflection. This phase shift is 180° for A-polarized light at non-absorbing interfaces. As a consequence, incoming and reflected beams cancel at the interface (node). On the other hand, at large angles of incidence ||-polarized incident light results in an enhanced electric field component perpendicular to the reflecting interface (z-axis). For thin samples, i.e. sample thickness (d) much smaller than the quarter wavelength (O/ 4) of the reflected light, RA spectra report only partial information on orientation. It should be noted, however, that for a complete orientation analysis spectra obtained with light polarized in the plane of the interface (x,y-plane) is also necessary. ATR fulfils this requirement, in contrast to RA. Diffuse reflectance (DR) is successfully applied to obtain IR spectra of rough (scattering) or dull surfaces, i.e. of media intractable by other reflection techniques. The interpretation of DR spectra, however, is sometimes handicapped by the fact that they may be a mixture of AB and SR spectra. DR spectroscopy is a sensitive tool, especially when used with an IR Fourier transform (FT) spectrometer (DRIFT). Elucidation of structureactivity relationships is the aim of many applications of reflection spectroscopy to thin layers at interfaces. In this context, polarization measurements are of considerable

importance, since molecules at interfaces exhibit often induced ordering. Low signal intensities are common in different kinds of interface spectroscopy, especially when the sample consists of a monolayer or even submonolayer as usual in heterogeneous catalysis and substrate biomembrane interaction. Although modern FTIR spectrometers exhibit very high stability, signal-tonoise (S/N) ratio enhancement by data accumulation is limited by environmental and instrumental instabilities. The fact that most commercial FT instruments are operated in a single-beam mode is disadvantageous in this respect, because the longer an experiment lasts, the greater is the time lag between sample and reference data, which facilitates the intrusion of instabilities. Several optional extras are available in order to reduce the time lag between acquisition of sample and reference spectra. One possibility is the conversion of the single-beam instrument into a pseudo-double-beam instrument by means of a shuttle which moves alternately the sample and reference into the IR beam. Such attachments were first developed for transmission experiments, and were later adapted for ATR measurements. In the latter case the sample and the reference are placed on top of one another on the same trapezoidal MIRE. A parallel beam of half the height of the MIRE is directed alternately through the upper and lower half of the MIRE by computer-controlled vertical displacement of the ATR cuvette. This method is referred to as single-beam sample reference (SBSR) technique and is described in more detail below. Polarization modulation (PM) in combination with IRRAS is a further possibility to enhance instrumental stability and background compensation when working at grazing incidence to a thin sample on a metal substrate. PM at about 50 kHz is achieved by means of a photoelastic modulator (PEM). Since under these experimental conditions the sample will only absorb light in the ||-polarized half-wave, the Apolarized half-wave of the signal is representative of the background, i.e. of the reference. Subtraction is performed by lock-in technique within each PM cycle, i.e. 50 000 times per second. As a consequence, environmental and instrumental contributions are largely compensated. Finally, it should be noted that a more general application of modulation spectroscopy can be used to obtain selective information on an excitable sample. Modulated excitation (ME) spectroscopy can always be applied with samples allowing periodic stimulation via a periodic variation of any external thermodynamic parameter, e.g. temperature, pressure, concentration, electric field, light flux. ME causes a

60 ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS

periodic variation of the absorbance at those wavelengths that are typical for the molecules involved in the stimulated process. Phase-sensitive detection (PSD) by digital lock-in technique adapted for FTIR instruments permits spectral registration of the modulated, i.e. affected, part of the sample. A typical feature of ME with PSD is the comparison of sample and reference within each period of stimulation. Within this time interval environmental and instrumental parameters are usually stable so that a very good baseline is achieved. Moreover, if one or more relaxation times Wi of the kinetic response of the stimulated sample fulfil the condition 0.1 < ZWi < 10, where Z denotes the angular frequency of stimulation, significant phase lags Ii between stimulation and sample responses will occur which are related to the reaction scheme and the rate constants of the stimulated process.

Theory of reflectance spectroscopy

resulting in

As concluded from Equation [3], perpendicular polarized incident light undergoes a phase shift of 180° upon reflection, i.e. there is a node at the reflecting interface resulting in zero electric field strength at this point. On the other hand, parallel polarized components remain in-phase. However, this conclusion holds no longer in the case of absorbing media. The corresponding equations for the ratio t between transmitted and incident electric fields are

For a comprehensive description of the theory of reflectance the reader is referred to the Further reading section. In this article, theory will only be presented when necessary for a general understanding. Fresnels equations

The theory of reflection and transmission of an electromagnetic wave by a plane boundary was first derived by Fresnel. The geometry of specular reflection and transmission is depicted in Figure 1. The incident (i) plane wave consists of the parallel (i) and perpendicular polarized (A) electric field components Ei i and EiA, respectively. The corresponding components of the reflected (r) and refracted (transmitted t) field components are denoted by Eri, Er⊥, Et i, and EtA. Fresnels equations relate the reflected and transmitted components to the corresponding incident components. For a nonabsorbing medium, i.e. the absorption indices N1 and N2 equal to zero, one obtains for the ratio r between reflected and incident electric field

where i and A denote parallel and perpendicular polarization, according to Figure 1. Equation [1] may be modified by introducing Snells law of refraction:

Figure 1 Specular reflection and transmission. The angles of incidence (i ), reflection (r ) and refraction (t ) are denoted by Ti, Tr and Tt, respectively. The corresponding electric field components are denoted by E. They are split into orthogonal portions, one parallel to the plane of incidence (x,z-plane) and the other perpendicular to this plane (parallel to y-axis). Accordingly, electric fields are referred to as parallel (i) and perpendicular (A) polarized, n1, n2, N1 and N2 denote the refractive and absorption indices in the two media.

ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS 61

In order to modify Equations [1], [3] and [4] for absorbing media one has to introduce the complex refractive index. For an incoming plane-wave it is given by

where n is the refractive index and N is the absorption index. As a consequence, r and t become complex and the resulting phase shifts differ from 0° and 180° as mentioned above.

The factor , which reduces to n21 for nonabsorbing media, results from the change of dielectrica (see Equation [8]), and cos Tt/ cos Ti takes account of the different cross-sections of the beam in media 1 and 2, respectively. rr* and tt*, become r2 and t2 for nonabsorbing media. In this case, Equations [1], [2] and [8] result in

Energy flux density

The flux density of an electromagnetic wave is described by the Poynting vector. For the case of the plane wave field one obtains for the time average in the direction of propagation where n21 denotes the ratio of refractive indices of media 2 and 1, respectively (see Figure 1). For normal incidence, i.e. Ti 0, Equation [10] reduces to where P is the permeability, i.e. the product of the permeability of vacuum P0 and the relative permeability Pr which is unity for nonconductive materials, and H denotes the permittivity which is the product of the permittivity of vacuum H0 and the relative permittivity (dielectric constant) Hr which is complex for absorbing media, according to

Introducing the absolute value of Equation [7] into Equation [6] results in, for nonconducting media,

Reflectance U and transmittance W are defined as the ratios of the corresponding energy fluxes J. According to Equation [8] they are proportional to the square of the electric field, i.e.

In order to obtain the reflectance of an absorbing medium one may introduce Equation [5] into Equation [10] or [11]. The result for normal incidence is

It should be noted that in many applications medium 1 is air or a nonabsorbing crystal, i.e. N1 0. It follows from Equation [12] that the reflectance increases with increasing absorption index of medium 2 ( N2). In the limiting case of N2 → f one obtains U → 1, i.e. a perfect mirror. Expressions for the more complicated case of oblique incidence to absorbing media have been derived (see Further reading). The KramersKronig relations

For normal incidence (Ti = 0) and nonabsorbing medium 1 (N1 = 0, see Figure 1), one obtains from Equations [1] and [5] the following expression for the complex ratio (Ti = 0) between reflected and

62 ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS

incident electric fields:

where K and I are the amplitude and phase of (Ti = 0). They are functions of the wavenumber and related to each other by the KramersKronig equations:

Experimentally, K( ) can be determined, since it is related to the reflectance at normal incidence by , i.e. ; see Equation [12]. The Drude model may be used to extrapolate the measurement to = 0 and = f. From K( ) and I( ) the components of the refractive index can be calculated according to

It follows from Equations [2] and [16] for Ti > Tc that sin Tt = sin Ti /sin Tc > 1, resulting in a complex value for the corresponding cosine:

The ratio r between internally reflected and incident electric field components is then obtained by introducing Equation [17] into Equation [1], resulting in

The corresponding equations for medium 2 are obtained by introducing Equation [17] into Equation [4], resulting in

Finally, it should be noted that incident electric fields undergo phase shifts in the ATR mode even if medium 2 is nonabsorbing. It follows from Equation [18] that For a detailed discussion, see the Further reading section. Internal reflection spectroscopy (ATR)

Internal reflection can only occur when the angle of the refracted beam Tt is larger than the angle of incidence Ti. This means, according to Snells law (Equation [2]), that the refractive index of medium 2 must be smaller than that of medium 1 (n2 < n1). This is contrary to the situation in Figure 1 where n2 > n1 was assumed. The region of total reflection begins when Tt reaches 90°, i.e. at the critical angle of incidence Tc. It follows from Equation [2] that

where Gi and GA are the phase shifts per internal reflection (no absorption) of i-polarized and Apolarized incident light. Since the phase shifts and amplitudes are polarization dependent, linearly polarized incident light is elliptically polarized after an internal reflection. This phenomenon, however, does not hinder polarization measurement in the ATR mode.

ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS 63

Applications Diffuse reflectance

The geometry of a diffuse reflection experiment is shown in Figure 2. The incident beam (I) is collimated to the sample S by means of the ellipsoidal mirror Mi. Two reflection mechanisms must be considered, specular reflection, Rs, and diffuse reflection, Rd. The former occurs at the surface and is governed by the Fresnel equations (Equations [1], [3] and [1012]). As a consequence of anomalous dispersion, specular reflected light exhibits S-shaped intensity changes at the wavelengths of sample absorption. In contrast, diffuse reflected light exhibits absorption bands at frequencies observed also with transmitted light, but with intensities deviating significantly from those measured in a transmission experiment. The intensity of the diffuse reflection spectrum may be described by the BouguerLambert law (Eqn [21]), the analogous expression to the LambertBeer law in transmission spectroscopy.

where d is the mean penetrated layer thickness, i.e. the depth of light penetration into the surface layer which results in an intensity decrease by a factor of 1/e, and D denotes the napierian absorption coefficient. Diffuse reflectance infrared Fourier transform spectroscopy (DRIFT) has become a frequently used technique to obtain IR spectra from materials intractable by transmission spectroscopy. A number of high-performance reflection accessories are available from different manufacturers (see below), allowing

Figure 2 Diffuse reflection experiment. Mi, Mr ellipsoidal mirrors for incident and reflected light; S sample; I, Rd, Rs incident diffuse, and specular reflected beams, respectively. In the magnified circle a possible ray tracing through a surface particle is shown, demonstrating the formation of mixed diffuse and specular reflected light. d is the mean penetrated layer thickness according to the Bouguer–Lambert law (Equation [21]).

the detection of quantities down to the nanogram region. Nevertheless, DRIFT spectroscopy is confronted with two intrinsic problems: (i) the superposition of diffuse and specular reflected light (see Figure 2), which may lead to distorted line shapes, and (ii) the dependence of the mean penetration depth d on the absorption coefficient. d is found to be inversely proportional to the absorption coefficient D, thus leading to a certain leveling of the band intensities. The disturbance by specular reflection may be reduced considerably by technical means (trapping) on the reflection attachment. The resulting diffuse reflection spectrum then has to be corrected in order to correspond to the absorbance of a transmission spectrum. This mathematical procedure is generally performed according to the KubelkaMunk theory. For a comprehensive and critical discussion of this theory the reader is referred to the Further reading section. Specular reflection spectroscopy (SRS)

In specular reflectance, only light reflected off the front surface is collected (see Figure 2). The reflected energy is generally small ( n2. The segmental order parameter Sseg is frequently used to characterize molecular ordering, e.g. To

68 ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS

describe the fluctuation of a functional group in a molecule via the polarized absorption bands of a typical group vibration. For uniaxial orientation the order parameter is defined according to

Thus, if the ATR geometry and the optical constants of the system are known, then Sseg may be determined measuring the dichroic ratio R of a given absorption band, followed by the calculation of the mean square cosine ¢cos2 D² and inserting this value into Equation [34]. A typical example is discussed later. For more details and examples of application the reader is referred to the Further reading section.

single-beam (SB) mode. As a consequence a singlechannel reference spectrum has to be stored for later conversion of single-channel sample spectra into transmittance and absorbance spectra. This technique suffers inaccuracy owing to drifts resulting from the instrument, the sample or atmospheric absorption. In order to eliminate these unwanted effects to a great extent, a new ATR attachment has been constructed, converting a single-beam instrument into a pseudo-double-beam instrument. The principle features of this attachment are depicted in Figure 7. As usual, a convergent IR beam enters the sample compartment. However, the focal point is now displaced from the centre of the sample compartment by means of the planar mirrors M1 and M2 to the new position F. The off-axis

Determination of surface concentration The effective thickness as indicated by Equations [26][28] holds for isotropic samples. Modification for oriented samples results

From Equation [25] and the relation dei = dex + dez and deA = dey one obtains for the surface concentration *.

where Ai and AA denote the absorbance measured with parallel and perpendicular polarized incident light, respectively, H is the molar absorption coefficient, Q denotes the number of equal functional groups per molecule and N is the number of active internal reflections. It should be noted that Equation [36] holds for peak absorbance and integrated absorbance, provided that the corresponding molar absorption coefficients are used. Special experimental ATR techniques

Single-beam sample reference (SBSR) technique Most FTIR spectrometers are working in the

Figure 7 Single-beam sample reference (SBSR) ATR attachment. (A) The focus in the sample compartment is displaced to the position F by the planar mirrors M1 and M2. The off-axis parabolic mirror M3 produces a parallel beam with a diameter of 1 cm, i.e. half of the height of the MIRE. The cylindrical mirror M4 focuses the light to the entrance face of the MIRE. M5, which has the same shape as M4, reconverts to parallel light directing it via the planar mirror M6 through the polarizer POL and it is then being focused to the detector DET by the off-axis parabolic mirror M7. (B) Alternating change from sample to reference is performed by computer-controlled lifting and lowering of the ATR cell body. Reproduced with permission of the American Institute of Physics from Fringeli UP et al. (1998) AIP Conference Proceedings 430: 729–747.

ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS 69

parabolic mirror M3 performs a conversion of the divergent beam into a parallel beam with fourfold reduced cross-section. This beam is focused to the entrance face of a trapezoidal MIRE by a cylindrical mirror M4. Therefore, the ray propagation in the MIRE is still parallel to the direction of light propagation (x-axis), enabling subdivision of the two reflective faces (x,y-planes) of the MIRE alongside at half-height. One half of the MIRE is then used for the sample (S) and the other one for the reference (R). Both S and R were encapsulated by flowthrough cuvettes, independently accessible by liquids or gases. This principle is referred to as the Single beam sample reference (SBSR) technique. SBSR absorbance spectra are calculated from sample and reference single-channel spectra which have been measured with very short mutual time delay. A most favourable benefit of SBSR technique is that no waisted time for purging is required before starting a measurement after closing the sample compartment. Moreover, the whole sequence of single-channel spectra in the S and R channels is also available, allowing reconstruction of the history of each channel at any time by conventional data handling. Modulated excitation (ME) spectroscopy and 2D IR spectroscopy Change of any external thermodynamic parameter generally exerts a specific influence on the state of a chemical system. The system response will be relaxation from the initial state (e.g. an equilibrium) to the final state (a new equilibrium state or a stationary state). In the case of a periodic change (modulation) of the parameter, the system response will also be periodic with angular frequency Z, relaxing from the initial state to a stationary state. All absorption bands of the spectrum which result from stimulated molecules or parts of them will be labelled by the frequency Z. As a consequence, it is possible to separate the modulated response of a system from the stationary response, resulting from parts of the system that were not affected by modulated excitation (ME) as well as from the background. Moreover, if the kinetics of the stimulated process is in the same time range as the period of external excitation, phase lags and damped amplitudes will result. Both depend characteristically on the stimulation frequency, and therefore one can derive relevant information on the reaction scheme and the kinetics of the stimulated process (see also caption to Figure 8). A variety of ME experiments have been reported. (i) Temperature ME of poly-L-lysine was used to study induced periodic secondary structural changes as well as the sequence of transients. (ii) The classical ATR setup (see Figure 6) facilitates the application of electric fields to immobilized thin films, such as

biomembrane assemblies or to bulk materials such as liquid crystals, since a Ge ATR plate, supporting the membrane, may be used as one electrode, and the back-wall of the cuvette as counter electrode. (iii) Hydration modulation was used to detect hydration sites of model membranes, and (iv) ME by UV radiation permitted kinetic studies of photoinduced chemical reactions. (v) ME by chemical substrates is a further versatile method to study chemically induced conformational changes of a sample immobilized to the MIRE. For that purpose, two computer-controlled pumps are used for periodic exchange the liquid (water) environment of the sample in a flow-through cell. An example demonstrating the sensitivity and high quality of background compensation of ME techniques is presented later. The principles of ME spectroscopy are depicted schematically in Figure 8. 2D FTIR spectroscopy Absorption bands in a set of modulation spectra that exhibit equal phase shifts with respect to the external stimulation are considered to be correlated. 2D correlation analysis is a statistical graphical means to visualize such a correlation in a 2D plot. Consequently, phase-resolved modulation spectra are data of a higher level and unambiguously allow a more direct and accurate evaluation. 2D plots look attractive, but, one should be aware that the information content is lower than that of the underlying modulation spectra, first because band overlapping may result in inadequate phase information, and second because 2D spectra are affected much more by baseline errors than the original modulation spectra. A comprehensive discussion can be found in the Further reading section. Sensitivity of ATR spectroscopy

Sensitivity of stationary ATR measurements Commercial multiple internal reflection elements MIRE permit up to 50 internal reflections. This is generally enough for thin-layer spectroscopy in the nanometre or even subnanometre region. As an example, Figure 9 shows a dipalmitoylphosphatidic acid monolayer, i.e. a lipid monolayer of about 2 nm thickness, which has been transferred from the airwater interface to a germanium MIRE by means of the LangmuirBlodgett technique. The dominant bands in Figure 9 result from the stretching vibrations of 28 CH2 groups of the two saturated hydrocarbon chains of the DPPA molecule. Looking at three resolved weaker bands gives an impression of the absorbance to be expected from a monomolecular coverage by functional groups of medium or weak molar absorption. The first is the terminal methyl group of the hydrocarbon chains. The

70 ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS

Figure 8 Schematic setup for modulated excitation (ME) experiments. A periodic excitation is exerted on the sample with frequency Z. The sample response S(t), as sensed by IR radiation, contains the frequency Z and higher harmonics at wavelengths that are significant for those parts of the sample that have been affected by the stimulation. Selective detection of the periodic sample responses is performed by phase-sensitive detection (PSD), resulting in the DC output and An of fundamental Z(n 1) and their harmonics nZ (n = 2, 3, }.), as well as the phase shifts In between the nth harmonic and the stimulation. This phase shift is indicative of the kinetics of the stimulated process and of the underlying chemical reaction scheme. Since the PSD output An(n = 1, 2, },n; frequency nZ) is proportional to cos(InIn.PSD), absorption bands featuring the same phase shift In are considered to be correlated, i.e. to be representative of a population consisting of distinct molecules or molecular parts. In.PSD is the operator-controlled PSD phase setting. Because of the cosine dependence, different populations will have their absorbance maxima at different In.PSD settings, thus allowing selective detection. Moreover, since in the case that 0.1 < ZWi < 10 (Wi denotes the i th relaxation time of the system), In becomes Z dependent, In In(Z). The spectral information can then be spread in the In.PSD– Z plane, resulting in a significant enhancement of resolution with respect to standard difference spectroscopy and time-resolved spectroscopy.

Figure 9 Parallel (i) and perpendicular (A) polarized ATR absorbance spectra of a dipalmitoylphosphatidic acid (DPPA) monolayer transferred at 30 mN m1 from the aqueous subphase (104 M CaCl2) to a germanium multiple internal reflection element (MIRE). Spectra were obtained from the dry monolayer in contact with dry air. A surface concentration of * 3.93 u 1010 mol cm2 was calculated by means of Equation [36] using the dichroic ratio of the symmetric CH2 stretching vibration at 2850 cm1 with respect to a linear baseline (B), resulting in R ATR [Qs(CH2)] 0.923. Angle of incidence Ti = 45°; number of equal functional groups Q = 28; number of active internal reflections N 39.

antisymmetric stretching vibration, Qas(CH3) absorbs at ∼2960 cm1. As concluded from Figure 9, this monolayer results in a peak absorbance of about 6 mAU. A weaker band is observed near 1420 cm 1 and may be assigned to the bending vibration of the D-methylene groups of the hydrocarbon chains,

G(D-CH2). Thus an approximate monolayer of D-CH2 groups results in an absorbance of only about 1 mAU. Third, a monolayer of phosphate head groups results in more intense absorption bands because of the larger transition dipolemoment of the polar group. The corresponding absorbances of PO3 stretching

ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS 71

vibrations in the range 10001250 cm1 are within 5 and 10 mAU. It is concluded that conventional ATR measurements may allow significant access to bands of about 0.20.5 mAU, which corresponds to 20 50% of a monolayer of weak absorbers. Quantitative analysis of stationary ATR spectra The DPPA monolayer spectra shown in Figure 9 are now used to demonstrate the ease of application of the formalism for quantitative analysis of ATR spectra presented earlier. Dichroic ratio of symmetric CH2 stretching The dichroic Ratio according to Equation [32] was calculated from the integrated absorbances of the symmetric CH2 stretching bands, Qs(CH2), using linear baselines as marked in Figure 9 with lower and upper limits at 2828 and 2871 cm1, respectively. The corresponding integrals were found to be = 0.381 cm1, and = 0.413 cm1 ATR resulting in R = 0.923. This is the relevant experimental quantity. Mean orientation of hydrocarbon chains Uniaxial orientation, i.e. isotropic distribution of DPPA around the z-axis, is assumed. The mean square cosine of the angle between the transition dipole moments of Qs(CH2) of the whole population of CH2 groups of the molecule (28 groups in hydrocarbon chains, 1 in the glycerol part, slightly shifted in frequency) can be calculated from Equation [32], resulting in Equation [37].

The squares of relative electric field components at the interface (z = 0) in medium 2 as calculated from Equation [29] for Ti = 45°, n1 = 4.0 (germanium), n2 = 1.5 (DPPA monolayer) and n3 = 1.0 (dry air) result in E = 1.991, E = 2.133, and E = 0.450. It follows that E = E + E = 2.441 and E = E = 2.133. The dichroic ratio for an isotropic film under these conditions would result in, according to Equation [33], R = 1.144. Explicit equations for relative electric components calculated by means of Harricks weak absorber approximation can be found in the Further reading section. Introducing the experimental value of RATR and the calculated squares of relative electric field components into Equation [37], one obtains for the mean square cosine of the angle between the transition moment of Qs(CH2) and the z-axis 〈cos2 D〉 = 0.025. This value should not be negative because its minimum is zero; however, since it is

small, we consider it to be within experimental and predominately systematic errors. Therefore, we set 〈cos2 D〉 = 0, resulting in D = 90°. This result requires that all methylene groups of the hydrocarbon chains assume an all-trans conformation and, moreover, all hydrocarbon chains are aligned normal to the MIRE, i.e. parallel to the z-axis (tilt angle 0°). The exact wavenumber of the symmetric stretching vibration of the CH2 group in glycerol is not known. However, overlapping with Qs(CH2) of the hydrocarbon chains is probable. Consequently, the bisectrice of the glycerol CH2 group may also be concluded to be predominately parallel to the x,y-plane. Mean order parameter of CH2 groups The mean segmental order parameter resulting from Equation [34] is found to be Sseg>Qs(CH2)] = . This value is representative of a perfectly ordered molecular entity with isotropic arrangement of transition dipole moments around the z-axis and perfect parallel alignment to the interface (x,y-plane). It should be noted that for 〈cos2 D〉 = 1, i.e. transition moments perfectly aligned normal to the interface (z-axis), Equation [34] results in the upper limit Sseg = 1. Lipids in natural biomembranes consist of a considerable amount of unsaturated hydrocarbon chains. Since double bonds cause unavoidably gauche defects in elongated hydrocarbon chains, which leads to a reduced chain ordering, Sseg>Qs(CH2)] is increased, reaching zero for an isotropic chain arrangement, since 〈cos2 D〉 = 1/3 in this case. It should be noted that the determination of order parameters of individual methylene groups in the hydrocarbon chains requires generally selective deuteration. In this respect, comprehensive deuterium NMR work should be mentioned (see Further reading section). A more general case of sample geometry is that of a transition moment being inclined by an angle 4 with respect to the molecular axis a and isotropically distributed around a. Furthermore, the molecular axis a forms an angle J with respect to the tilt axis t, and is isotropically distributed around it, and finally, the axis t forming a tilt angle G with the z-axis and is isotropically distributed around it. In this case, the segmental order parameter, e.g. Sseg[Qs(CH2)], may be expressed as superposition of three uniaxial orientations according to

72 ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS

The angles G, J and 4 may be distinct or fluctuating (partly or all), describing a microcrystalline ultrastructure (MCU) and a liquid crystalline ultrastructure (LCU), respectively. S J is referred to as the molecular order parameter Smol. Applying Equation [38] to the DPPA monolayer under discussion, one obtains: SG = 1, SJ = 1, S4 = , meaning no tilt (G = 0), molecular axis (hydrocarbon chain) normal to the interface (J = 0), and transition dipole moment normal to the molecular axis (4 = 90). Surface concentration and area per molecule The surface concentration may be calculated using Equation [36]. The following additional information is required: (i) the integrated molar absorption coefficient related to a linear baseline from 2828 to 2871 cm1 (see Figure 9) was = 5.7 × 105 cm mol1, (ii) the real thickness of the layer was assumed to be d = 2.5 nm, (iii) the number of equal functional groups Q = 28, and (iv) the effective thicknesses de for parallel or perpendicular polarized incident light, which were calculated from Equations [27] and [35], resulted in de,i = 3.97 nm and de,A = 4.30 nm. The mean surface concentration was found to be 3.93 × 1010 mol cm2, corresponding to a molecular cross-section of 0.427 nm2 per molecule (42.3 Å2 per molecule). This value leads to the conclusion that the two hydrocarbon chains of a DPPA molecule predominantly determine the area per molecule, since the cross-section of an elongated hydrocarbon chain is 2021 Å 2. Conclusions Quantitative analysis, including orientation measurements, has been shown to be straightforward when the formalism based on Harricks weak absorber approximation is applied. For thin adsorbed layers, such as the DPPA monolayer under discussion, the results are fairly good. Application to bulk materials may introduce systematic errors as discussed above. If the weak absorber approximation is still to be applied, one should take care to work with an angle of incidence which is at least 15° larger than the critical angle, in order to avoid significant band distortions. In many cases it is possible to use quantitative data from transmission experiments to check the validity of the formalism applied to ATR data. A general critical aspect concerning the baseline selection should be mentioned. A linear tangential baseline has been used for quantitative analysis of the symmetric CH2 stretching vibration of DPPA (see Figure 9). Obviously the correct baseline is lower, i.e. the integrated absorbances used for analysis are systematically too small. The reason for this procedure is

only to permit good reproducibility. While the determination of the dichroic ratio is indifferent with respect to the choice of the baseline, it is mandatory to use integrated or peak molar absorption coefficients which have been determined under the same conditions. Even then deviations in the range of several per cent may occur among different operators. Finally, it should be noted that ATR spectroscopy allows very good background compensation, when adequate equipment is used. Sensitivity of modulated excitation (ME) ATR spectroscopy An impression of the sensitivity of stationary measurements was given the last but one section. A limit of 0.2 mAU is suggested. This limit is beaten by one order or magnitude when the ME technique is applied. As mentioned above, the sample must fulfil the condition of a reversible stimulation by a periodically altered external thermodynamic parameter. Here, the excellent sensitivity and instrumental stability will be demonstrated, for example, with a chemical modulation experiment performed in liquid water, a very strong absorber in the 3400 and 1640 cm1 region. In order to study the influence of immobilized charges on a lipid model membrane, an arachidic acid (ArAc) bilayer was prepared on a germanium MIRE by means of the LangmuirBlodgett (LB) technique. The MIRE was transferred in the hydrated state from the LB trough into a flow-through cell and kept in permanent contact with an aqueous buffer solution. Since the carboxylic acid groups of the second monolayer were facing the aqueous phase, the degree of protonation could be controlled via the environmental pH. A periodic pH modulation between pH 3 and 10 induced a periodic protonation and deprotonation of the carboxylic acid group. It should be noted that the first ArAc LB layer was attached by head to the Ge MIRE. Obviously, this binding was so special that typical absorption bands of the carboxylic acid groups were not visible in the spectrum. Therefore, one may assume that the head group signals shown in Figure 10 result predominately from the outer monolayer of ArAc. The stationary spectral intensity is comparable to that of the DPPA monolayer shown in Figure 9. Moreover, one should note that the experiments were performed in H2O, where in the 1640 and 3400 cm1 regions there is very low spectral energy available, favouring perturbations by incomplete background compensation. In this context, only the sensitivity and selectivity of ME techniques will be discussed. A comprehensive presentation and analysis of polarized pH modulation spectra will be given elsewhere.

ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS 73

Sensitivity of ME-spectroscopy Taking the pHmodulated spectrum shown in Figure 10 as a typical example, one may estimate the sensitivity by comparing the most intense ME spectra (IPSD = 60°/IPSD = 90°, the maximum is expected at IPSD ≈ 75°, consequently, IPSD ≈ 165° should result intensity zero) with the lowest intensity ME spectrum at IPSD = 0°. In order to check the S/N ratio, the IPSD = 0° spectrum was expanded 25 times in the CH2 wagging region and is plotted as a dashed inset in Figure 10. The ordinate scaling factor for the zoomed spectrum is 4.0 × 105. Comparing it with the other ME spectra (scaling factor 1.0 × 103) one can conclude that bands as weak as 1.0 × 105 AU are still detectable. Selectivity of pH ME The highest selectivity of ME spectroscopy is achieved if the stimulation frequency Z and the kinetics of the stimulated process are matched, i.e. if 0.1 < ZWi < 10, where Wi denotes the ith relaxation time of the system. Wi is a function of the rate constants involved in the stimulated process. Under these conditions, significant Z-dependent phase shifts are expected, resulting in Ii = arctan(ZWi) for a linear system. Consequently, a molecular or confor-

mational population represented by the relaxation time Wi exhibits maximum absorbance in the ME spectrum at a PSD phase setting Ii = IPSD, thus allowing selective detection and kinetic analysis by means of phase-resolved ME spectra. Moreover, Z acts as an additional experimental degree of freedom in this context, since information on selectivity and kinetics can be spread in the Z/IPSD plane which is more selective than the unidirectional information resulting from conventional relaxation measurements. In the actual case of pH modulation exerted on a monolayer of ArAc, there is no phase resolution observed, owing to the long modulation period of Wm = 16 min, i.e. no kinetic information is available. However, unambiguous discrimination between the protonated and deprotonated populations is possible. Only one characteristic example will be given here. The most prominent band from the protonated state is the C=O stretching vibration Q(COOH) of the carboxylic acid group near 1700 cm1. All other bands in the ME spectrum that have the same phase belong to the protonated population, whereas the remaining bands featuring opposite sign are members of the deprotonated population. Consequently, if no

Figure 10 pH-modulated excitation (ME) of an arachidic acid (ArAc) bilayer attached to a germanium multiple internal reflection element (MIRE). ME was performed by pumping alternatively two buffer solutions (100 mM NaCl, pH 3 and 100 mM NaCl, pH 10) through the ATR cuvette with a modulation period of W = 16 min. T 10°C. Upper trace A0; stationary spectrum of a protonated ArAc layer for comparison with modulation spectra. Traces A1; modulation spectra at PSD phase settings IPSD 0, 30,}, 180°. The 180° spectrum corresponds to the 0° spectrum with opposite sign, because the PSD output is proportional to cos (IIPSD), see also Figure 8. I denotes the phase difference between a given band and the stimulation. Owing to the long period of Wm 16 min, the observed bands in the modulation spectra exhibit only two resolved I values, which are 180° apart, as a consequence of the fact that the chemical relaxation time of protonation/deprotonation of ArAc is much shorter than the stimulation period. In order to demonstrate the excellent S/N ratio, the ordinate of the weakest modulation spectrum has been expanded in the CH2 wagging region by a factor of 25, i.e. the ordinate scaling factor for the dashed spectrum results in 4.0 u 105 (see text).

74 ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS

phase resolution is achieved, ME spectra reduce to difference spectra, which, however, have a considerably better background and instability compensation than conventional difference spectra, since corresponding sample and reference spectra are measured and evaluated/accumulated within each period of stimulation. Consider now the wagging region γw(CH2) of the spectra shown in Figure 10. The wagging motion γw(CH2) is described as in-phase displacement of both H atoms through the HCH plane of a methylene group, where the C atom remains predominately in place. In an all-trans hydrocarbon chain the transition dipole moment of γw(CH2) is expected to be parallel to the chain direction. Deviations may occur, however, from coupling with a polar end group. In the stationary absorbance spectrum A0, one can observe nine weak bands between about 1180 and 1320 cm 1. This sequence results from concerted wagging vibrations of all methylene groups in a hydrocarbon chain with an all-trans conformation. According to IR selection rules one has to expect n/2 IR-active vibrations for an even number n of CH2 groups in an all-trans conformation. ArAc has 18 CH2 groups per chain, resulting in the above-mentioned sequence of nine bands in accordance with theory. Since these bands are found to be in phase with ν(COOH), one can conclude that deprotonation of COOH is paralleled by reversible disordering of the chain structure, most probably by introducing gauche defects. Finally, it should be mentioned that γw(CH2) belongs to the group of weak absorption bands. One can conclude, therefore, that ME IR ATR spectroscopy allows significant quantitative studies on a molecular level with submonolayer quantities of weak absorbers.

Manufacturers of reflection accessories Standard equipment for reflection spectroscopy

ASI Sense IR Technologies, 15 Great Pasture Road, Danbury, CT 06810, USA. Bruker Optics, Wikingerstrasse 13, D-76189 Karlsruhe, Germany. Graseby Specac Inc., 301 Commerce Drive, Fairfield, CT 06432, USA. Harrick Scientific Corporation, 88 Broadway, Ossining, NY 10562, USA. International Crystal Laboratories, 11 Erie Street, Garfield, NJ 07026, USA. Spectra-Tech, Inc., Warrington WA3 7BH, UK.

Special equipment for SBSR-ATR and ME-ATR spectroscopy

Optispec, Rigistrasse 5, CH-8173 Neerach, Switzerland.

List of symbols A = absorbance (decadic); d = sample thickness; de = effective thickness (Harrick); dp = penetration depth; d = mean penetrated layer thickness; E = electric field; Ex = electric field component in xdirection; Ey = electric field component in y-direction; Ez = electric field component in z-direction; J = energy flux (time average); mx = x-component of the unit vector in the direction of M; my = ycomponent of the unit vector in the direction of M; mz = z-component of the unit vector in the direction of M; M = transition dipole moment; n = refractive index; = complex refractive index; N = number of active internal reflections; r = ratio of reflected to incident field; RATR = dichroic ratio related to ATR R = dichroic R = reflectance; spectra; ratio; SJ = Smol = molecular order parameter; ST = tilt order parameter; ST = order parameter with respect to the molecular axis; Sseg = segmental order parameter; S = Poynting vector (time average); t = ratio of transmitted to incident field; T = transmittance; z = distance from surface; i, ⊥ = parallel and perpendicular polarized light, respectively;

= conjugate complex; D = absorption coefficient; D = angle between transition dipole moment and zaxis; G = phase shift; H = permittivity; H = molar absorption coefficient; H0 = permittivity of vacuum; Hr = relative permittivity (dielectric constant); K = amplitude; Tc= critical angle; Ti = angle of incidence; Tr = angle of reflection; Tt = angle of refraction (transmission); N = absorption index; O= wavelength; P = permeability; P0 = permeability of vacuum; Pr = relative permeability; Q = number of equal functional groups per molecule; = wavenumber; U = reflectance; W = transmittance; W = relaxation time; * = surface concentration; I = phase lag; Z = angular frequency. See also: Electromagnetic Radiation; Industrial Applications of IR and Raman Spectroscopy; Polymer Applications of IR and Raman Spectroscopy; Raman and IR Microspectroscopy; Surface Studies by IR spectroscopy.

Further reading Baurecht D and Fringeli UP (2000) Surface charge induced conformational changes in an arachidic acid LangmuirBlodgett bilayer observed by pH-modulated excitation FTIR ATR spectroscopy. Langmuir (in preparation).

ATR AND REFLECTANCE IR SPECTROSCOPY, APPLICATIONS 75

Beccard B and Hapanowicz R (1995) Polarization Modulation FT-IR spectroscopy. Nicolet Application Note AN-9592. Madison, WI: Nicolet. Blaudez D, Turlet J-M, Dufourcq D, Bard D, Buffeteau T and Desbat B (1996) Investigation at the air-water interface using polarization modulation IR spectroscopy. Journal of the Chemical Society, Faraday Transactions 92: 525530. Born M and Wolf E (1983) Principles of Optics, Chapter I. Oxford: Pergamon Press. Fringeli UP (1992) In situ infrared attenuated total reflection membrane spectroscopy. In: Mirabella FM (ed.) Internal Reflection Spectroscopy, Theory and Applications, Chapter 10, pp 255324. New York: Marcel Dekker. Fringeli UP (1997) Simulatneous phase-sensitive digital detection process for time-resolved, quasi-simultaneously captured data arrays of a periodically stimulated system. PCT International Patent Application, WO97/ 08598. Fringeli UP, Goette J, Reiter G, Siam M and Baurecht D (1998) Structural investigation of oriented membrane assemblies by FTIR-ATR spectroscopy. In: deHaseth JA (ed.) Fourier Transform Spectroscopy; 11th International Conference, AIP Conference Proceedings 430, pp. 729747. Woodbury, New York: American Institute of Physics. Galant J, Desbat B, Vaknin D and Salesse Ch (1998) Polarization-modulated infrared spectroscopy and X-ray reflectivity of photosystem II core complex at the gaswater interface. Biophysical Journal 75: 28882899. Greenler RG (1966) Infrared study of adsorbed molecules on metal surfaces by reflection techniques. Journal of Chemical Physics 44: 310315. Hansen WH (1973) Internal reflection spectroscopy in electrochemistry. In: Delahay P and Tobias ChW (eds) Advances in Electrochemistry and Electrochemical Engineering, Vol 9, Muller RH (ed.) Optical Techniques in Electrochemistry, pp 160. New York: John Wiley & Sons. Hapke B (1993) Theory of Reflectance and Emittance Spectroscopy. New York: Cambridge Univeristy Press. Harrick NJ (1967) Internal Reflection Spectroscopy, New York: Interscience; 2nd edn (1979) Ossining, NY: Harrick, Scientific.

Hoffmann FM (1983) Infrared reflection-absorption spectroscopy of adsorbed molecules. Surface Science Reports 3: 107192. Kortüm G (1969) Reflectance Spectroscopy. New York: Springer. Mendelsohn R, Brauner JW and Gericke A (1995) External infrared reflection absorption spectroscopy of monolayer films at the air-water interface. Annual Review of Physical Chemistry 46: 305334. Mirabella FM (ed) (1992) Internal Reflection Spectroscopy, Theory and Application. New York: Marcel Dekker. Müller M, Buchet R and Fringeli UP (1996). 2D-FTIR ATR spectroscopy of thermo-induced periodic secondary structural changes of poly-(L)-lysine: A crosscorrelation analysis of phase-resolved temperature modulation spectra. Journal of Physical Chemistry 100: 1081010825. Picard F, Buffeteau T, Desbat B, Auger M and Pézolet M (1999) Quantitative orientation measurements in thin films by attenuated total reflection spectroscopy. Biophysical Journal 76: 539551. Seelig J and Seelig A (1980) Lipid confromation in model membranes and biological membranes. Quarterly Review of Biophysics 13: 1961. Tamm L and Tatulian S (1997) Infrared spectroscopy of proteins and peptides in lipid bilayers. Quarterly Review of Biophysics 30: 365429. Urban MW (1996) Attenuated Total Reflectance Spectroscopy of Polymers. Washington, DC: American Chemical Society. Wendlandt WWM and Hecht HG (1996) Reflectance Spectroscopy, New York: Interscience. Wenzl P, Fringeli M, Goette J and Fringeli UP (1994) Supported phospholipid bilayers prepared by the LB/Vesicle Method: A Fourier transform infrared attenuated total reflection spectroscopic study on structure and stability. Langmuir 10: 42534264. Wooten F (1972) Optical Properties of Solids. New York: Academic Press. Zachman G (1995) A rapid and dependable identification system for black polymeric materials. Journal of Molecular Structure 348: 453456. Zbinden R (1964) Infrared Spectroscopy of High Polymers, New York: Academic Press.

BIOCHEMICAL APPLICATIONS OF FLUORESCENCE SPECTROSCOPY 77

Introduction A remarkable range of fluorescence spectroscopy techniques has been developed in the last forty years to characterize fundamental properties of biological systems. The rapid progress over this period has led to strategies for monitoring events that transpire on the femtosecond time scale, for discerning features smaller than the Rayleigh resolution limit of light, and for detecting individual molecules. More routinely, fluorescence spectroscopy has provided invaluable insights into the structure, function, and concentrations of macromolecules, small molecules, lipids, and inorganic ions. Moreover, fluorescence analysis of living systems has begun to reveal directly how these biochemicals function and interact in situ. The initial section of this article presents an overview of the measurement formats used in fluorescence studies of biochemical systems. In the sections that follow, the application of diverse spectroscopic techniques to the study of biomolecular systems is examined, considering both the properties and uses of intrinsic biological fluorophores and also of fluorescent probes that have been designed to report on the presence or condition of biochemicals. In the Conclusion, several fluorescence approaches that may offer new capabilities in future biochemical studies are discussed. A fully comprehensive treatment of the diverse applications of fluorescence in biological studies is not possible in this article. For example, fluorescence studies of biological porphyrins a vast field of spectroscopy is mentioned in the most cursory fashion. Readers are encouraged to refer to other articles as listed later, for additional information, as well as to the sources in the Further reading list at the end of this article.

ELECTRONIC SPECTROSCOPY Applications

Overview of measurement formats In situ measurements

In general, when information is sought regarding the spatial distribution of chemicals within cells or tissue, some variant of fluorescence microscopy is used often in combination with transmission or Nomarski imaging. From an optical standpoint, fluorescence microscopy measurements either use a wide-field geometry, in which an entire field-of-view is illuminated simultaneously, or rely on raster scanning of a small resolution element to sequentially generate a fluorescence image. In the first approach, an arc lamp is commonly used as the excitation source, and large-format images can be acquired at video rate using an array detector. In scanning fluorescence microscopy, a laser typically provides the excitation light, which is focused critically to diffraction spot sizes as small as ∼ 0.2 µm using a high numerical aperture (NA) microscope objective. The sequential nature of image formation in scanning microscopy makes this procedure relatively slow. Although large format two-dimensional (2D) images typically require half a second or more to acquire, small 2D images or line scans can be produced much more rapidly. A fundamental advantage of the two scanning techniques most commonly used confocal and multiphoton microscopies is an ability to acquire 3D images of various biological specimens by scanning the laser focal spot in a given plane, then shifting the fine focus of the microscope to repeat the process in a new plane. In confocal microscopy, outof-plane fluorescence is prevented from reaching a single-element detector by placing a small aperture in a plane conjugate to the focal spot; multiphoton

78 BIOCHEMICAL APPLICATIONS OF FLUORESCENCE SPECTROSCOPY

microscopy achieves a similar degree of axial resolution through an inherent 3D localization of the excitation focal volume. Measurements with chemical separations

Though microscopy can be used to quantify simultaneously up to several chemical species by acquiring detailed spectroscopic information, when analysis of many, potentially unidentified, cellular components is desired, a separation procedure typically is used in combination with fluorescence detection. Commonly used techniques include high-pressure liquid chromatography (HPLC), capillary electrophoresis (CE) and related capillary chromatography techniques, and gel electrophoresis. In HPLC and CE, trace analysis frequently is performed by labelling nonfluorescent components either pre- or post-separation using fluorogenic reagents. Capillary techniques frequently offer several advantages over HPLC, including faster and more efficient separations, lower sample volume requirements, and alleviation of the need for highpressure equipment. In DNA gel electrophoresis, fluorescent intercalating dyes can be used to stain oligonucleotide bands in gels after electrophoresis. For DNA sequencing in agarose gels, the Sanger method can be used to generate fluorescent fragments by labelling primers with fluorophores whose emission spectra encode a specific terminating dideoxyribo-nucleotide. Analysis of proteins using SDSpolyacrylamide slab gel electrophoresis or various capillary techniques can be accomplished by fluorescently labelling analytes before separation. Cuvette, flow, and chip measurements

A large number of studies on isolated biochemical systems can be performed in cuvettes or deep-well slides. In particular, characterizations of macromolecular structure frequently are conducted using thermostatted cells with capabilities for stoppedflow exchange of various solution modifiers (e.g. ligands, denaturants) on a millisecond time scale. In many cases, instruments used for such studies also can perform nanosecond or microsecond timeresolved measurements of emission intensity or polarization in addition to standard steady-state detection after polarization. In many instances, it is desirable to rapidly count or isolate a subpopulation of cells exhibiting a particular chemical characteristic. In flow cytometry, this property is linked to a corresponding difference in cellular fluorescence (e.g. by selectively labelling a relevant gene product), which provides a means for measuring relatively large numbers of cells in a flowing stream using automated instrumentation.

Fluorescence-activated cell sorting (FACS) uses a flow cytometer in combination with a shunting system to isolate cells exhibiting characteristic fluorescence. The presence of various oligonucleotides, peptides, and small molecules in solution can be analysed using solid state chip based devices, in which receptor molecules (often antibodies or complementary oligonucleotides) immobilized on the chip surface bind the analyte(s) of interest. As a result of analyte binding, a characteristic fluorescence signal is generated from the receptor, analyte, or some other interacting species. In ELISA (enzyme-linked immunosorbent assay), the binding of analyte is ultimately linked to the enzymatic production of a fluorescent solutionphase molecule.

Analyses based on intrinsic biological fluorescence Ideally, all biological chemicals would carry highly specific spectroscopic signatures that define not only their identities but also every aspect of their physical states and environments, and they would yield optical signals intense enough to reveal the smallest changes relevant to a particular experiment. Biological analyses, of course, are not so straightforward, and generally provide very limited information in the absence of exogenous probe chemicals. Nevertheless, there are several classes of intrinsically fluorescent biological molecules, and a host of elegant spectroscopic techniques are used to investigate the properties of these species. The aromatic amino acids tryptophan (Trp), tyrosine (Tyr) and phenylalanine (Phe) have strong deep-UV absorption bands (Oex < 230 nm) corresponding to S0 → S2 transitions, but commonly are excited to the S1 state in fluorescence studies (Oex ≈ 260280 nm) to minimize photoreaction and enhance fluorescence quantum yields (Φf). At these longer wavelengths, Trp has the largest molar extinction coefficient (Hmax ≈ 5600) and quantum yield (Φf ≈ 0.2) of the three amino acids; for Phe, the values of H and Φf are so poor that this species is rarely useful in fluorescence studies. When subjected to UV irradiation, proteins with both Trp and Tyr (Figure 1) typically exhibit emission spectra whose shape is characteristic of Trp residues (Omax ≈ 350 nm) because of nonradiative energy transfer from Tyr to Trp. Despite the fact that Trp and Tyr are not highly fluorescent, and undergo intersystem crossing and myriad photoreactions with large quantum yields, these species can be useful probes of protein structure. Emission from both Trp and Tyr is

BIOCHEMICAL APPLICATIONS OF FLUORESCENCE SPECTROSCOPY 79

Figure 1

Fluorescent amino acids.

quenched to different extents by a variety of intramolecular polar moieties (including carboxylic acids, amino groups, and imidazole groups), bound cofactors (e.g. NAD+), prosthetic groups (e.g. heme) and small molecules in solution, making determination of protein structural changes possible through measurement of excited-state lifetimes or steadystate fluorescence intensities. Unlike Tyr, Trp fluorescence often is found to be higher in proteins than in free solution. In molecules containing multiple fluorescent residues, the overall modulation of fluorescence associated with protein conformational changes can be small due to an averaging of advantageous and deleterious quantum yield changes. For this reason, time-resolved measurements of multiexponential excited state decay times sometimes can provide information unavailable from steady-state intensity measurements. The Stokes shift for Trp also can depend significantly on environment. When Trp is sequestered in the hydrophobic core of a protein, little or no reorientation of solvent takes place after excitation of the chromophore, and emission is blue-shifted relative to a Trp residue on the exterior of a protein. Extreme examples include azurin, in which Trp appears to be completely isolated from H2O (Omax < 310 nm), and denatured parvalbumin, in which the Trp emission spectrum is essentially identical to free Trp. For proteins that contain Tyr but not Trp, the shapes of the emission spectra closely match free Tyr (Omax ≈ 305 nm), regardless of local chemical environment. A number of macromolecular diffusion and conformational properties can be studied using fluorescence anisotropy, fluorescence correlation spectroscopy (FCS), and fluorescence recovery after photobleaching (FRAP). These techniques most commonly are applied to proteins labelled with highly fluorescent probes, but can exploit intrinsic fluorescence in some instances. In fluorescence anisotropy studies, polarized light is used to selectively excite molecules whose transition dipole moments are aligned with the electric field vector. Steady-state

measurements of fluorescence anisotropy can be accomplished using a continuous excitation source; provided that anisotropy decays monoexponentially after excitation, information can be obtained on properties such as rotational diffusion times in membranes and the existence of ligandhost interactions. Time-resolved measurements, in which anisotropy is measured as a function of time following pulsed excitation, can provide more detailed information on phenomena such as anisotropic rotational diffusion, and can be used to study complex processes such as electron transfer in light-harvesting chlorophyll complexes. FCS is a steady-state technique for measuring the concentration and diffusion coefficient of a fluorescent molecule based on the magnitude of signal fluctuations as molecules diffuse through a finite probe volume. Diffusion coefficients can also be determined with FRAP, a technique in which fluorescent molecules within the probe volume are photobleached with a high-intensity pulse of light. After the bleaching event, the kinetics of fluorescence recovery (representing the diffusion of fresh molecules into the probe volume) are monitored with low-intensity irradiation. Depending on the circumstances, FCS or FRAP can be useful for measuring the local viscosity of cellular or isolated biochemical environments, for monitoring ligand/host interactions, and for determining changes in diffusion coefficients associated with macromolecular conformational changes. Identification and measurement of UV fluorescent proteins in complex biological samples requires mixtures to be fractionated into individual components, often with CE or HPLC. CE with on-column UV fluorescence detection has been shown to be useful in measuring attomole quantities of protein in individual red blood cells. Peptide mapping can be accomplished using HPLC with post-column UV fluorescence detection by purifying individual proteins, then subjecting various fractions to proteolytic digests before separating the resulting peptide segments with chromatography. Metabolic derivatives of Trp and Tyr also produce significant UV fluorescence. In some secretory cells, Trp is converted to the indolamine neurotransmitter, serotonin, which in turn can be processed into the pineal hormone, melatonin. Metabolic conversion of Tyr yields the catecholamines (dopamine, norepinephrine, and epinephrine). Although the cellular concentration of transmitters is typically low, their concentrations within secretory granules can approach 1 molar. The fluorescence properties of these neurotransmitters are analogous to the parent amino acids, with indolamines exhibiting stronger emission and at longer excitation and emission wavelengths than the catecholamines.

80 BIOCHEMICAL APPLICATIONS OF FLUORESCENCE SPECTROSCOPY

Intrinsic UV fluorescence has been used to perform in situ measurements of serotonin. Loss of serotonin from astrocytes can be monitored using wide-field UV fluorescence microscopy after various chemical treatments. In addition, three-photon scanning fluorescence microscopy has been shown to be useful for tracking secretion of serotonin from granules in cultured cells in response to cross-linking of cell surface IgE receptors. Fluorescence from indolamines and catecholamines separated with CE has provided a means to measure these species in the low- to mid-attomole range. In some instances, protein fluorescence does not derive directly from the aromatic amino acids. Two such proteins native to the jellyfish genus Aequorea aequorin and the green fluorescent protein (GFP) have been particularly useful tools in biochemical studies. Aequorin emits light (Omax ≈ 470 nm) from the bound luminophore coelenterazine as the result of a Ca2+-promoted oxidation reaction, and hence has been useful as an intracellular Ca2+ probe both through microinjection and recombinant gene expression. The discovery and cloning of GFP has attracted much attention recently in part because of its utility as a reporter for various gene products when incorporated into fusion proteins. In GFP, the chromophoric unit has been identified as an imidazolone anion formed by cyclization and oxidation of tripeptide sequence (-Ser-Tyr-Gly-); fluorescence of this species is characterized by excitation maxima at ∼ 400 and ∼ 475 nm and peak emission at ∼ 510 nm. The fluorescence from GFP is intense enough to detect individual molecules of this species immobilized in aqueous gels. Both aequorin and GFP can be genetically targetted to accumulate in specific organelles to measure localized properties, such as [Ca2+] and protein diffusion coefficients in mitochondria. A variety of protein cofactors have significant intrinsic fluorescence, including the reduced nicotinamide nucleotides (NADH and NADPH) and the oxidized flavins (Figure 2). The nicotinamides are excited in the near-UV and emit maximally at ∼ 470 nm; flavins exhibit both near-UV and visible excitation maxima, and have a peak emission at ∼ 515 nm. In both groups of cofactors, the adenine group partially quench fluorescence through collisions; in flavins, adenine also can quench fluorescence by forming an intramolecular complex with the fluorescent ring system. The significance of adenine quenching is underscored by the fact that flavin mononucleotide (FMN), a highly fluorescent cofactor that lacks adenine, has a fluorescence quantum yield tenfold greater than that of flavin adenine dinucleotide (FAD). Flavins that are covalently or non-

Figure 2

Three fluorescent cofactors.

covalently bound to proteins display widely varying differences in fluorescence quantum yields, although flavoproteins often are relatively nonfluorescent. In contrast, NADH bound to proteins sometimes has substantially greater fluorescence than free NADH because of a decreased ability of adenine to quench the nicotinamide group. In dehydrogenases, for example, crystallographic studies indicate that nicotinamide cofactors are bound in an extended fashion, with the adenine and nicotinamide groups bound to different pairs of β-sheets. Moreover, conformational changes in lactate dehydrogenase induced by effector molecules can be monitored as changes in NADH fluorescence. As a rule, the redox cofactors like the fluorescent amino acids have spectroscopic properties that are sensitive to environment, and can be used to probe conformational changes in proteins. NADH and oxidized flavin (flavoprotein) fluorescence can be used as an indicator of the redox states of cells challenged with a variety of chemical effectors, and has been used to correlate the oxidative status of cells in situ and in vivo to functions such as electrical activity in brain tissue. The fluorescence of NADH also has been used to characterize the in vitro activity of individual lactate dehydrogenase molecules, which reduce the nonfluorescent NAD+

BIOCHEMICAL APPLICATIONS OF FLUORESCENCE SPECTROSCOPY 81

during oxidative production of pyruvate. This general strategy, in which single-enzyme molecules are characterized through the generation of a fluorescent product, also can be applied to reactions using engineered (nonbiological) fluorogenic substrates. Most nucleic acids have poor fluorescence properties because of extremely short excited state lifetimes, and are most often studied using nonbiological intercalating or groove-binding dyes. One notable exception is the Y base, which has been used as an indicator of tRNA folding induced by divalent cation binding. In addition, some nucleotides exhibit moderate fluorescence under highly acidic conditions, and can be measured in attomole to femtomole amounts using capillary electrophoresis.

Analyses using nonbiological probes Although fluorimetric analysis of biological systems is experimentally simpler and less disruptive to inherent chemical properties when the use of synthetic fluorescent molecules can be avoided, in many instances the intrinsic properties of biological chemicals do not provide sufficient means for characterizing or quantifying desired properties. In the case of protein analysis, for example, many species lack Trp or Tyr residues altogether; in protein molecules that do contain these amino acids, the fluorescence characteristics may be inadequate to yield the needed information. Numerous probes have been developed for investigating biological molecules in vitro and in living cells, with greatly differing strategies for uncovering desired information. No common photophysical or photochemical characteristics can be ascribed to this entire assortment of nonbiological fluorophores. Many probes have been engineered to have very large absorption cross sections and fluorescence quantum yields, and to be resistant to photoreaction. The probes based on laser dyes such as fluorescein (Hmax > 50 000; Φf > 0.9) or one of the rhodamines are good examples of this class of molecules, and often are incorporated in a number of the calcium probes, fluorescent antibodies (see below) and reagents used to label proteins for conformational studies involving fluorescence anisotropy, FCS and FRAP. In addition, by conjugating two spectroscopically distinct probes to different sites on a molecule, Förster energy transfer between the fluorophores can be used to measure intramolecular distances. Despite the advantages of these highly fluorescent molecules, probes having less optimal fluorescence properties are deliberately selected for some applications often as a compromise to achieve a greater change in optical characteristics under different chemical

environments, or to avoid toxicity to living cells. For example, a dye whose fluorescence properties depend sensitively on the polarity and rigidity of its local environment may be more useful in many instances than a rhodamine for characterizing protein conformation. One such compound is NBD chloride, an amine-reactive reagent whose product has a markedly greater Φf value when occupying a hydrophobic site. Reactive fluorogenic reagents (e.g. NBD chloride, fluorescamine and CBQCA) are essentially nonfluorescent until they react with analyte molecules, and are commonly used to tag biological species that share a particular reactive group (e.g. a thiol or amine). This low-specificity labelling approach is extremely useful for identification and measurement of many low-concentration species in a mixture when some means exist for fractionating the multiple components (e.g. HPLC). Because of the low level of reagent fluorescence, a large excess of these species typically can be present in the reaction mixture without interfering with analysis. When characterizing samples containing many biomolecules, a probe that reacts with a particular chemical moiety sometimes does not provide adequate specificity, even when used with a highly efficient separation procedure. Analysis of the protein distribution in cellular specimens is perhaps the quintessential high-specificity requirement. Because of the value of such measurements in characterizing cellular composition, immunohistochemistry is one of the most ubiquitous biological applications of fluorescence. In routine determinations of gene products, cultured cellular samples or sectioned tissue are chemically fixed and subjected to sequential application of primary and secondary antibodies, followed by imaging with fluorescence microscopy. Primary antibodies can be raised against a tremendous diversity of cellular species, ranging from membrane proteins to small diffusable species such as neurotransmitters. Several species can be analysed in a single specimen by covalently attaching dyes with different excitation/emission spectra to different secondary antibodies. Most commonly, fixed and labelled samples are analysed using either wide-field or confocal laser scanning microscopy. When one seeks to analyse a solution sample for the presence of a known antigen, ELISA can be used. In one format, an immobilized antibody binds the antigen, removing it from solution. A second antibody which is conjugated to an enzyme capable of converting a nonfluorescent substrate into a fluorescent product then is applied, and binds to the immobilized antigen. After a washing step, the enzymatic reaction is initiated. In this way, large signal amplifications

82 BIOCHEMICAL APPLICATIONS OF FLUORESCENCE SPECTROSCOPY

provide extremely sensitive assays for analytes such as hormones and drug metabolites. Although immunological techniques can be useful for measuring analyte concentrations, uncertainties in cross-reactivity and matrix effects on the antibody-antigen conjugation efficiency often limit this approach to semiquantitative determinations. Diffusable cytosolic species (e.g. second messengers) can be measured in living cells using highly specific fluorescent probes. The calcium-sensitive dyes, a broad class of molecules whose excitation or emission properties are dependent on chelation of Ca2+, are ubiquitous in cellular biology studies. For such compounds, it is important that affinity for Ca2+ is much greater than for other cationic species that may be present in the cytosol at much higher concentrations (e.g. Mg2+). In general, cytosolic free Ca2+ concentrations vary over the approximate range 100 to 1000 nM, although much higher levels are sometimes reached transiently. Common examples of Ca2+-sensitive dyes include fluo-3 (100-fold intensity increase when bound to Ca2+), fura-2 (excitation Omax changes when bound to Ca2+), and indo-1 (emission Omax changes when bound to Ca2+) (Figure 3). For these species, the Ca2+ dissociation constant has been designed to approximately match free cytosolic levels, thus maximizing the modulation in fluorescence for a given change in [Ca2+]. Like tryptophan, indo-1 has an indole-based chromophore, but in this case the S-system is further delocalized to electronically interact with carboxylate groups responsible for Ca2+ chelation. In many instances, fura-2 and indo-1 may offer advantages over fluo-3, as ratiometric measurements of excitation or emission intensities at two different wavelengths can help avoid errors associated with uncertainties in intracellular dye concentrations. Calcium probes have played an invaluable role in identifying a range of cytosolic phenomena, such as [Ca2+] spiking after hormone binding to cell surface receptors and Ca2+ waves in oocytes following a fertilization event. Probes exist for other important cytosolic effectors, such as cyclic AMP (cAMP), which is monitored by the level of Förster energy transfer between fluorescein and rhodamine attached to a cAMP-binding enzyme. Although it is possible to microinject dyes for probing cytosolic milieus through patch or intracellular pipettes, it is more common to incubate many cultured cells in a medium containing low molecular mass dyes in an esterified form (e.g. acetoxymethyl, or AM, ester). This parent species is hydrophobic, and thus can diffuse freely through cell membranes. Once the compound is localized to the cytosol, nonspecific esterases hydrolyse the hydrophobic group, generating a charged form of the probe that

Figure 3

Two common calcium probes.

can no longer diffuse through the membrane. Large numbers of cells also can be loaded rapidly using electroporation. Structures in living cells often can be characterized using probes with specificity for classes of molecules rather than for particular target analytes. Hydrophobic or amphipathic dyes, for example, can be used to gauge cell membrane fluidity and diffusion, as well as exocytosis of secretory vesicles and subsequent membrane recycling. Examples of this class of dyes are 8-anilinonaphathalene sulfonate (ANS), which displays significant fluorescence only when bound to membranes (or protected hydrophobic regions of proteins), and FM-143, a compound used extensively to study regulated secretion from neurons. Some membrane dyes have been engineered (or serendipitously found) to alter their fluorescence properties based on the electric field across the phospholipid bilayer of cell or organelle membranes. Typically, such potentiometric dyes either respond with fast kinetics to electric field changes but display small modulations in fluorescence (e.g. the styrylpyridinium

BIOCHEMICAL APPLICATIONS OF FLUORESCENCE SPECTROSCOPY 83

probes), or undergo large fluorescence changes but only very slowly. Because large, rapid changes in fluorescence generally are not obtained with existing probes, it is difficult to perform voltage measurements on individual neurons and other excitable cells. Other fluorogenic dyes, such as DAPI, ethidium bromide and the cyanine dyes, bind to DNA and/or RNA and undergo changes in fluorescence intensity of up to 1000-fold. Depending on the probe, these compounds can be used to image fixed cells or track mitotic events. Combinations of probes with different specificities for single-stranded, double-stranded and ribosomal RNA have been used to characterize the quantity and conformation of nucleic acids throughout the cell cycle.

Conclusion In the forty years since the intrinsic fluorescence of amino acids was first characterized, fluorescence spectroscopy has developed into one of the most versatile and powerful tools for investigating biochemical systems. In addition to the enormous range of applications that now exist, a variety of emerging analysis strategies provide evidence of a continued phase of rapid growth for fluorescence applications. Developments in instrumentation laser sources, optics, and detectors have made possible the characterization of individual enzyme molecules and highly fluorescent proteins such as GFP and β-phycoerythrin, a light-harvesting protein. Solid-state femtosecond sources recently have made it feasible to generate multiphoton-excited fluorescence of biological molecules in solution using relatively low integrated irradiation, and have provided reproducibility necessary for experiments involving living specimens. Various technologies that rely on fluorescence for high sensitivity analysis have been made possible through advances in microfabrication and robotics. Examples include chip-based microarray sensors for detecting a multitude of ligands simultaneously, and chips containing arrays of microscopic electrophoresis channels for performing high-throughput, rapid separations of DNA fragments. Other fluorescence techniques exploit the evanescent, or near-field, properties of light to excite fluorescence in highly restricted regions of space, thereby improving measurement sensitivity or the spatial resolution of fluorescence imaging. Near-field scanning optical microscopy (NSOM) can image structures smaller than the diffraction limit of light using fluorescence generated by an evanescent field that escapes from the aperture at the end of a drawn,

metallized fibre. The resolution obtained with this approach is determined by the dimensions of the aperture, which is usually limited by losses in throughput as the fibre tip diameter is reduced. Thus far, only a few biological applications of this powerful technique have been reported. New developments in the chemistry of fluorescent probes undoubtedly will open new applications for fluorescence in biochemistry. A growing number of companies and academic researchers are devising new biological probes, and in some cases are using nontraditional strategies such as combinatorial chemistry to generate hosts with high specificity and large binding constants for desired ligands. Genetic manipulation of cultured cells offers particularly exciting opportunities for in situ biochemical measurements. Technology now exists for using an enzyme as a reporter for neurotransmitter-activated transcription in individual mammalian cells, with sensitivities capable of measuring fewer than 50 gene product molecules. In an initial demonstration of this strategy, large amplifications of gene product expression were obtained by localizing an engineered substrate to the cytosol whose fluorescence properties change when it is degraded by the reporter enzyme. Through the judicious use of molecular and cellular biology, chemistry, and fluorescence spectroscopy, eventually it may be feasible to track the production and breakdown of individual molecules in living cells.

List of symbols H = molar extinction coefficient; O = wavelength; ) = quantum yield. See also: Fluorescence Microscopy, Applications; Fluorescence Polarization and Anisotropy; Fluorescent Molecular Probes; Inorganic Condensed Matter, Applications of Luminescence Spectroscopy; Luminescence, Theory; Organic Chemistry Applications of Fluorescence Spectroscopy; UV-Visible Absorption and Fluorescence Spectrometers; X-ray Fluorescence Spectrometers; X-ray Fluorescence Spectroscopy, Applications.

Further reading Cantor CR and Schimmel PR (1980) Biophysical Chemistry, Parts IIII. New York: Freeman. Chalfie M, Tu Y, Euskirchen G, Ward WW and Prasher DC (1994) Green fluorescent protein as a marker for gene expression. Science 263: 802805.

84 BIOCHEMICAL APPLICATIONS OF MASS SPECTROMETRY

Cobbold PH and Rink TJ (1987) Fluorescence and bioluminescence measurement of cytoplasmic free calcium. Biochemical Journal 248: 313328. Craig DB, Arriaga E, Wong JCY, Lu H and Dovichi NJ (1998) Life and death of a single enzyme molecule. Analytical Chemistry 70: 39A43A. Creed D (1984) The photophysics and photochemistry of the near-UV absorbing amino acids I. Tryptophan and its simple derivatives. Photochem. Photobiol. 39: 537562. Creed D (1984) The photophysics and photochemistry of the near-UV absorbing amino acids II. Tyrosine and its simple derivatives. Photochem. Photobiol. 39: 563 575. Everse J, Anderson B and You K-S (1982) The Pyridine Nucleotide Coenzymes. New York: Academic Press. Freifelder D (1982) Physical Biochemistry, 2nd edn. New York: W.H. Freeman. Hoagland RP (1996) Handbook of Fluorescent Probes and Research Chemicals, 6th edn. Eugene, OR: Molecular Probes. Lakowicz JR (ed) (1991) Topics in Fluorescence Spectroscopy, Vol 1, Techniques. New York: Plenum Press. Lakowicz JR (ed) (1992) Topics in Fluorescence Spectroscopy, Vol 3, Biochemical Applications. New York: Plenum Press.

Lakowicz JR (1983) Principles of Fluorescence Spectroscopy. New York: Plenum Press. Lillard SJ, Yeung ES, Lautamo RMA and Mao DT (1995) Separation of hemoglobin variants in single human erythrocytes by capillary electrophoresis with laserinduced native fluorescence detection. J. Chromatogr. A 718: 397404. Permyakov EA (1993) Luminescent Spectroscopy of Proteins. Boca Raton, FA: CRC Press. Rutter GA, Burnett P, Rizzuto R et al (1996) Subcellular imaging of intramitochondrial Ca2+ with recombinant targeted aequorin: significance for the regulation of pyruvate dehydrogenase activity. Proceedings of the National Academy of Science, USA 93: 54895494. Stryer L (1995) Biochemistry, 4th edn. New York: W.H. Freeman. Teale FWJ and Weber G (1957) Ultraviolet fluorescence of the aromatic amino acids. Biochemical Journal 65: 476482. Tsien RY (1994) Fluorescence imaging creates a window on the cell. Chemical Engineering News 72: 3444. Zlokarnik G, Negulescu PA, Knapp TE et al (1998) Quantitation of transcription and clonal selection of single living cells with beta-lactamase as reporter. Science 279: 8488.

Mass spectrometry is a powerful tool for the characterization of various biomolecules including proteins, nucleic acids and carbohydrates. The advantages of mass spectrometry are high sensitivity, high mass accuracy, and more importantly, structural information. Historically, biomolecules have proven difficult to characterize using mass spectrometry. Problems often arise with impure samples, low ion abundance for analysis due to inefficient ionization processes, and low mass accuracy for higher molecular weight compounds. Recent advances in the development of electrospray ionization (ESI) and matrix-assisted laser desorption/ionization (MALDI) now permit the analysis of biomolecules with high sensitivity and good mass accuracy. These improvements now allow the use of mass spectrometry for the identification of unknown structures and are sui for applications focused on acquiring sequence information on the samples of interest.

MASS SPECTROMETRY Applications This article provides a brief introduction to the various applications of mass spectrometry to biomolecule analysis. More detailed discussions of particular applications of mass spectrometry to such analyses can be found in other articles in this encyclopedia.

Ionization and mass analysis Mass spectrometric measurements of biomolecules involve the determination of molecular mass, or of the masses of various components of the original molecule, which can then be related to various structural properties, such as sequence. In either case, the crucial step lies in the conversion of liquid- or solid-phase solutions of the analytes into gaseous ions. Common ionization sources used in mass spectrometry for biomolecule analysis experiments are: fast-atom bombardment (FAB), ESI and MALDI. FAB was

BIOCHEMICAL APPLICATIONS OF MASS SPECTROMETRY 85

historically the ionization method of choice, but has been largely replaced by ESI or MALDI at the present time. ESI- and MALDI-based methods have particularly benefited the analysis of biomolecules because these techniques accomplish the otherwise experimentally difficult task of producing gas-phase ions from solution species that are both thermally labile and polar. Table 1 summarizes the different instrument configurations used in the analysis of various classes of biomolecules. MALDI has extremely high sensitivities with reports of detection limits at the sub-femtomolar level. A higher efficiency for protonated molecular ion production with MALDI has been observed relative to FAB. MALDI typically yields intact protonated molecular ions with minimal fragmentation and is therefore commonly referred to as a soft ionization process. MALDI sources are generally coupled to time-of-flight (TOF) mass analysers. TOF mass analysers are characterized by high upper mass limits with reduced resolution at the higher masses. The production of high molecular weight ions and subsequent analysis of these ions makes the MALDITOF combination a powerful tool for biomolecule analysis. ESI can be considered a complementary method to MALDI. As with MALDI, electrospray ionization of biomolecules yields protonated or cationized molecular ions with little or no fragmentation, and it is also referred to as a soft ionization source. A particular advantage of ESI compared to MALDI is that the analyte is sampled from the solution phase. Under these conditions, ESI is readily coupled to high performance liquid chromatography (HPLC) or capillary electrophoresis (CE) separation systems. Such combinations permit online LC-MS or CE-MS experiments. Table 1 Summary of mass spectrometry techniques used in biomolecule analysis

Analyte

Ionization source

Proteins/peptides

FAB MALDI ESI

Oligonucleotides

FAB MALDI ESI FAB MALDI ESI FAB MALDI ESI

Oligosaccharides

Lipids

Mass analyser Quadrupole, sector TOF, FTICR Quadrupole, sector, FTICR, TOF Quadrupole, sector TOF, FTICR Sector, FTICR, TOF Quadrupole, sector TOF Quadrupole, sector, TOF Quadrupole TOF Quadrupole, FTICR, TOF

The overriding feature of an ESI-generated mass spectrum is the appearance of multiply charged ions. Multiple charging results from the loss or addition of multiple hydrogen ions or metal ions (e.g. potassium or sodium) to the biomolecule. The multiple charging effect is advantageous because it allows for the analysis of high molecular weight biomolecules using mass analysers with low m/z limits. The disadvantage of multiple charging is that a spectrum has to be deconvoluted and thus spectral interpretation can be complicated, especially during the analysis of mixtures.

Molecular weight determinations of biomolecules Relative mass is an intrinsic molecular property which, when measured with high accuracy, becomes a unique and unusually effective parameter for characterization of synthetic or natural biomolecules. Mass spectrometry based methods can be broadly applied not only to unmodified synthetic biomolecules, but also to modified synthetic and natural biomolecules (e.g. glycosylated proteins). The level of mass accuracy one obtains during the measurement will depend on the capabilities of the mass analyser used. Quadrupole and TOF instruments yield lower mass accuracies than sector or Fourier transform ion cyclotron resonance (FTICR) instruments. High mass accuracy is not only necessary for qualitative analysis of biomolecules present in a sample, but is necessary to provide unambiguous peak identification in a mass spectrum. The primary challenge to accurate molecular weight measurements of biomolecules is reduction or complete removal of salt adducts. In the majority of situations, the buffers used to prepare or isolate the analyte of interest contains Group 1 or Group 2 metal salts. These metal salts can potentially interfere with the accurate mass analysis of the analyte due to the gas-phase adduction of one to several metal cations to the analyte. Optimal results are obtained only after substantive (and in some cases, exhaustive) removal of these contaminants. Recent developments for sample purification involve the use of solution additives or online purification cartridges which reduce the presence of interfering salts while retaining the ability to characterize minimal amounts of analyte. Measurement of molecular mass of biomolecules is now a suitable replacement for prior methods based on the use of gel electrophoresis. Molecular weight measurement for all classes of biomolecules is a relatively routine procedure, and the results obtained are typically of greater accuracy than those previously

86 BIOCHEMICAL APPLICATIONS OF MASS SPECTROMETRY

obtained by gel electrophoresis. A common application of mass spectrometry and molecular weight measurement is for the identification of unknown proteins. The experimentally obtained molecular weight value can be searched against the available protein databases (e.g. SwissProt or PIR) to potentially identify the protein. This application becomes particularly useful for the identification of unknown proteins when used in conjunction with enzymatically generated peptide fragments (see below).

Determination of the primary sequence of biomolecules One of the most popular and productive uses of mass spectrometry for biomolecule analysis is the sequence or structure determination through analysis of smaller constituents of the original molecule. Two different approaches for generating the smaller, sequence informative constituents from the original molecule are available. An indirect approach is to generate sequence-specific information by solution-based chemical or enzymatic reactions. Chemical or enzymatic digestion of an intact biomolecule results in a number of smaller constituents that are amenable to mass spectrometric analysis. The so-called direct approach involves fragmentation of the analyte in the gas phase. Fragmentation can be induced by the desorption or ionization process, or can be induced by collisions with neutral target molecules or by collisions with surfaces. The analysis of fragment ions initiated by gas-phase dissociation resulting from collisions with neutral molecules or surfaces are broadly referred to as tandem mass spectrometry (MS/MS) experiments. With both approaches, the sequence of the original molecule is obtained by interpreting the resulting fragments. Enzymatic approaches

Enzymatic digestion followed by mass spectrometric analysis of the resulting products is a popular and powerful approach to sequence determination. A number of enzymes are available which either are specific for a particular substituent of the biomolecule or are nonspecific but sequentially hydrolyse the analyte of interest. Trypsin is an enzyme commonly used to digest proteins into smaller peptides. Trypsin selectively cleaves proteins at the C-terminal side of lysine and arginine amino acid residues. Digestion of a protein using trypsin will generate a tryptic digest whose components are amenable to mass spectrometric analysis. In many cases, the masses of the tryptic fragments can be measured more accurately than the mass of the original molecule, thereby

improving the identification of unknown proteins. As mentioned earlier, the masses of tryptic peptides, used in conjunction with molecular weight measurements of intact proteins, can be used to search known protein databases for efficient and accurate identification of unknown proteins. Sequential digestion of biomolecules is an alternative approach to determining sequence information. The utility of any mass spectrometric sequencing method that relies on consecutive backbone cleavages depends on the formation of a mass ladder. The sequence information is obtained by determining the mass difference between successive peaks in the mass spectrum. For example, phosphodiesterases are enzymes which sequentially hydrolyse the phosphodiester linkage between oligonucleotides and nucleic acids. In the case of oligodeoxynucleotides, the expected mass differences between successive peaks will correspond to the loss of: dC = 289.5, dT = 304.26, dA = 313.27, and dG = 329.27 Da. Mass ladder methods have a distinct advantage for sequence determination, because it is the difference in two mass measurements that results in the desired information. A drawback to this approach is the limited size of the analyte that is amenable to sequential digestion. Tandem mass spectrometry approaches

The gas-phase approach to determining structural information about biomolecules is through tandem mass spectrometry (MS/MS). Tandem mass spectrometry involves isolation of the ion of interest (commonly referred to as the parent ion in MS/MS) and then dissociating this ion via collisions with neutrals or surfaces to produce fragment ions (commonly referred to as product ions in MS/MS) from which the primary sequence of the molecular ion of interest can be determined. Generally, the fragmentation process in MS/MS experiments generates product ions which contain sequence specific information from throughout the molecular ion of interest. Tandem mass spectrometry is applicable to all types of biomolecules but typically requires specialized mass spectrometry instrumentation for implementation. The most common tandem mass spectrometer is a triple quadrupole instrument, wherein the first and third quadrupoles are used as mass analysers and the middle quadrupole is used as a collision chamber. MS/MS can be performed with double focusing sector instruments and in some cases with specialized TOF mass analysers. Quadrupole ion traps and FTICR mass spectrometers are ideally suited for MS/MS experiments, and due to the operational characteristics of these mass analysers

BIOCHEMICAL APPLICATIONS OF MASS SPECTROMETRY 87

additional MS/MS experiments can be performed, allowing for MSn studies of biomolecules.

Higher order gas-phase structure and non-covalent interactions studied using mass spectrometry Similar to studies performed using NMR, hydrogen/ deuterium (H/D) exchange experiments on biomolecules are feasible with mass spectrometry. Labile hydrogens can be exchanged in solution prior to mass spectrometric analysis, or alternatively H/D exchange experiments can be performed in the gas phase. The former studies have been used to localize sites of H/D exchange on biomolecules of interest and are used for mechanistic studies of biomolecule function. The latter studies typically involve ESI-MS and permit investigations into the gas-phase conformation of biomolecules in the absence of the solvent. The majority of H/D exchange experiments have focused on peptide and protein analysis, although the method is amenable to other classes of biomolecules. A particular advantage of ESI-MS for biomolecule analysis is realized by generating the analyte ions from solution conditions that retain the secondary, tertiary and even quaternary structure of the biomolecules. Noncovalent binding of biomolecules has been observed in the ESI mass spectrum and, when operated under the appropriate conditions, the mass spectral data are a direct probe of the solution-phase biomolecule assembly. Protein assemblies, proteinnucleic acid complexes, duplex DNA and other noncovalently bound biomolecule assemblies have been studied using mass spectrometry.

See also: Chemical Structure Information from Mass Spectrometry; Chromatography–MS, Methods; Fast Atom Bombardment Ionization in Mass Spectrometry; Fragmentation in Mass Spectrometry; Hyphenated Techniques, Applications of in Mass Spectrometry; Medical Applications of Mass Spectrometry; MS-MS and MSn; Nucleic Acids and Nucleotides Studied Using Mass Spectrometry; Peptides and Proteins Studied Using Mass Spectrometry; Quadrupoles, Use of in Mass Spectrometry; Sector Mass Spectrometers; Surface Induced Dissociation in Mass Spectrometry; Time of Flight Mass Spectrometers.

Further reading Fenn JB, Mann M, Meng CK, Wong SF and Whitehouse CM (1989) Electrospray ionization for mass spectrometry of large biomolecules. Science 246: 6470. Karas M, Bahr U and Hillenkamp F (1989) UV laser matrix desorption/ionization mass spectrometry of proteins in the 100,000 Dalton range. International Journal of Mass Spectrometry and Ion Processes 92: 231242. Loo JA (1995) Bioanalytical mass spectrometry: many flavors to choose. Bioconjugate Chemistry 6: 644665. McCloskey JA (ed) (1990) Methods in Enzymology, Vol. 193 San Diego: Academic Press, Inc. Senko MW and McLafferty FW (1994) Mass spectrometry of macromolecules: Has its time now come? Annual Review of Biophysics and Biomolecular Structure 23: 763785. Smith RD, Loo JA, Edmonds CG, Barinaga CJ and Udseth HR (1990) New developments in biochemical mass spectrometry: Electrospray ionization. Analytical Chemistry 62: 882899.

88 BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY

Biochemical Applications of Raman Spectroscopy Peter Hildebrandt, Max-Planck-Institut für Strahlenchemie, Mülheim, Germany Sophie Lecomte, CNRS-Université Paris VI, Thiais, France

VIBRATIONAL, ROTATIONAL & RAMAN SPECTROSCOPIES Applications

Until the late 1960s, Raman spectroscopy was largely restricted to small inorganic and organic molecules. Concomitant to the substantial improvement of excitation sources and detection systems beginning in the early 1970s, biological molecules, which in most cases are weak Raman scatterers, became more and more accessible to this technique. Now, Raman spectroscopy has been developed to a versatile tool for studying biological systems on various levels of complexity, i.e. ranging from small amino acids to biopolymers and even to living cells. Raman spectra like IR spectra include detailed information about the molecular structure and, hence, may provide a key for elucidating structure function relationships. Unfortunately, the ability to extract structural data from the spectra is much less advanced than, for example, in NMR spectroscopy, since a sound vibrational analysis based on normal mode calculations is not straightforward. Empirical force fields only provide meaningful results in the case of small and/or symmetric molecules for which a large set of isotopomers are available. Alternatively, quantum chemical methods can be employed to calculate the force constants. Despite the recent progress in hard- and software development, this approach is as yet restricted to prosthetic groups and building blocks of biopolymers. Thus, in most cases the interpretation of the Raman spectra of biomolecules is based on empirical relationships derived from comparative studies of model compounds. The large number of normal modes of biomolecules is also associated with an experimental difficulty inasmuch as individual Raman-active modes may be closely spaced so that the observed peaks include several unresolved bands. This is particularly true for those modes which originate from chemically identical building blocks of the biopolymers such as the amide vibrations of proteins. In these cases, however, the analysis of these bands provides valuable information about the structure of the ensemble of these building blocks, e.g. the secondary structure of proteins. The selectivity of Raman spectroscopy can be substantially increased by the choice of an appropriate excitation wavelength. When the excitation energy

approaches that of an electronic transition of a molecule, exclusively the Raman intensities of the modes originating from this chromophore are enhanced by several orders of magnitude (resonance Raman, RR). In this way, it is possible to probe the vibrational band pattern of this specific part of the macromolecule regardless of the size of the remaining optically transparent matrix.

Specific experimental considerations Under non-resonant conditions, quantum yields for the Raman effect are in the order of 109, implying that sample concentrations in the millimolar range are required for obtaining Raman spectra of satisfactory quality. Resonance enhancement substantially improves the sensitivity, but even RR intensities are too weak to obtain spectra of samples which exhibit a strong fluorescence or which contain strongly fluorescent impurities. In those cases, Fourier-transform near-infrared Raman spectroscopy may be a useful alternative approach. Possible thermal denaturation or photochemical processes which may be induced by laser irradiation may impose serious constraints on Raman (RR) spectroscopic experiments of biomolecules. Such effects can be reduced by using appropriate sample arrangements such as flowing or rotating devices, or by lowering the temperature. In any case, the photon flux through a volume element of a sample should be kept as small as possible to ensure the integrity of the biomolecule during the Raman experiment. These requirements are best fulfilled using CW-lasers as excitation sources. A second parameter which is essential for measuring the spectra of sensitive biomaterial is a short duration of the Raman experiment. Scanning monochromatic detection systems (photon counting devices) may require several hours to obtain Raman spectra of sufficient quality. Polychromatic detection systems can strongly reduce the total signal accumulation time and, in the case of CCD detectors, yield a spectral resolution and signal-to-noise ratio comparable to scanning systems.

BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY 89

Structure of biomolecules Raman spectroscopic studies which are directed to gain information about the structure of biomolecules are generally performed in the stationary mode. Such experiments have been carried out with all kinds of biopolymers and their building blocks. However, up to now, the level of spectral analysis and interpretation is such that reliable structural information can be extracted from the spectra only for proteins, nucleic acids and membranes. Proteins and amino acids

Non-resonant Raman spectra of proteins display bands originating from the protein backbone, i.e. the amide modes and the amino acid side chains, in particular, the aromatic amino acids tryptophan (Trp), tyrosine (Tyr) and phenylalanine (Phe). The strongest amide band in the non-resonant Raman spectra originates from the amide I mode which is essentially a pure C=O stretching of the peptide bond. Its frequency depends on the type of hydrogen bonding and dipoledipole interactions associated with the peptide bond. As these interactions are different for the various elements of secondary structure of proteins, systematic Raman and IR spectroscopic studies of peptides and proteins with known three-dimensional structure allow the determination of amide I frequencies which are characteristic not only for the main secondary structure elements α-helix (1657 cm1) and β-sheet (1626 and 1635 cm1) but also for turns (1666 cm1) and β-sheet/turns (1679 cm1). For irregular or random coil peptide bonds the amide I is found at 1644 cm1. These findings constitute the basis for the secondary structure determination of proteins and peptides of unknown structure. In the most general way, the measured amide I envelope is fitted by a set of these bands with constant frequencies and half widths. The relative intensities, which are the only adjustable parameters, represent a measure for the relative contributions of the various secondary structure elements. This approach is also frequently used in IR spectroscopy which offers the advantage of a better signal-to-noise ratio but is restricted to measurements in D2O due to the strong absorption of water. With an increased set of reference proteins, the method has gained a considerable accuracy which can well compete with CD measurements. Complementary information about the secondary structure can be obtained by using RR spectroscopy. Upon excitation in resonance with the π → π* transitions of the peptide bonds at ∼ 190 nm, the amide modes II, III and IIc, which include the CN stretching

vibrations, are preferentially enhanced so that they appear with substantially higher intensity than the amide I mode (C=O stretching) which in turn gains intensity only via a higher lying electronic transition at ~160 nm. Like the amide I mode, the amide modes II, III and IIc are at different frequencies in the various structural elements. More important for secondary structure determination, however, are the RR intensities of these modes. The qualitatively different dipole dipole interactions in α-helical and β-sheet structures lead to a respective decrease (hypochromic effect) and increase (hyperchromic effect) of the oscillator strength which in turn affects the Raman cross-sections of these amide modes. Thus, it was possible to derive (linear) relations between the RR intensities of these modes (determined relative to an internal standard) and the secondary structure. Due to the high energy of the deep UV laser pulses required as the excitation source, special attention has to be paid to avoiding photodegradation processes of the biomolecule. Such effects are also a restriction on probing the aromatic amino acid side chains of proteins using excitation lines between 200 and 240 nm. Under these conditions, several ring vibrational modes are selectively enhanced via the La (Phe, Tyr) and Ba,b transitions (Trp). The frequencies and relative intensities of these modes can be used to study the local environment of these amino acids, i.e. to obtain information about this specific part of the tertiary structure. In particular, the modes of Tyr and Trp may provide more detailed information about the local protein structure. The p-hydroxy substituent of Tyr is generally involved in hydrogen bonding interactions with hydrogen bond acceptors or donors. These interactions are sensitively reflected by the modes ν7a, ν7a′ and ν9a since their frequencies show linear relations with the solvent donor number. Monitoring these modes in proteins allows conclusions about the kind and the strength of hydrogen bonding interactions of Tyr. In the extreme case, when Tyr is deprotonated, the entire vibrational band pattern drastically changes and, due to the redshift of the electronic transition, the resonance enhancement is altered. Also Trp can undergo hydrogen bonding interactions via the indole NH group. In that case, a band at ∼ 880 cm1 (ν10a, W17) is an appropriate marker in the RR spectra of proteins. The second important spectral marker of Trp, a Fermi doublet at ∼ 1345 cm1, exhibits a moderate resonance enhancement and can only be identified in the spectra of relatively small proteins. This band is an indicator for the hydrophobicity of the local environment of the indole ring. Based on systematic studies of Trp and model compounds, an empirical

90 BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY

relationship has been derived between the torsional angle of the C2C3CβCα entity and the frequency of the band at ∼ 1550 cm1 (W3). This marker band, however, does not appear to be applicable for analysing the Trp structure in proteins inasmuch as it appears in a spectrally crowded region. Although a priori the size of the protein is not a limitation for UV RR spectroscopic analysis of aromatic amino acids, increasing numbers of Phe, Tyr and Trp residues strongly complicate the interpretation of the spectra. Nevertheless, working with high quality spectra it is possible to probe subtle structural changes of a single amino acid even in relatively large proteins as demonstrated in the studies on haemoglobin by Spiro and co-workers. This protein consists of two (αβ dimers, each of them including 15 Phe, 6 Tyr and 3 Trp residues. It had been assumed that the T → R transition of haemoglobin was associated with a change of hydrogen bonding interactions of one Tyr (out of 12) which in fact could be confirmed by UV RR (difference) spectra. The application of Raman spectroscopy for probing amino acid side-chain structures in proteins is not restricted to excitation in the deep UV. Some ring vibrations of the aromatic amino acid side chains can also be identified in the non-resonance Raman spectra. These are, for instance, the Fermi doublet of Trp at ∼ 1345 cm 1 and a band pair of Tyr at 850 and 830 cm 1, assigned to the fundamental mode ν1 and the overtone 2ν16a, which together form a Fermi doublet as well. This doublet, which is only weakly resonance enhanced upon UV excitation, appears with considerable intensity under non-resonance conditions and is known to be a sensitive marker for hydrogen bonding interactions as derived from systematic studies on model compounds. In fact, the intensity distribution of this doublet can be used to monitor changes of hydrogen bonding interactions of Tyr in proteins. Among the aliphatic amino acid side chains, cysteine (Cys) and methionine (Met) give rise to particular strong Raman bands which appear in less cogent spectral regions. The CS stretching is found between 620 and 750 cm 1 depending on the conformation of the side chain, e.g. trans or gauche conformation of Met. In addition, the CS stretching vibrations of disulfide bridges, which constitute an important stabilizing factor for the tertiary structure of proteins, are observed at 630650, 700 and 720 cm 1 when the trans substituent X of the XCα CβSS entity is a hydrogen, nitrogen and carbon atom, respectively. However, the most characteristic feature of disulfide bridges is the SS stretching (500 525 cm 1). There is a large body of experimental data for proteins and model compounds and evidence has

been provided for a linear correlation between the frequency and the CSSC dihedral angle. Chromoproteins

Chromoproteins are characterized by an electronic absorption band in the near-UV, visible or near-IR spectral range. These bands may arise from π → π* transitions of prosthetic groups or from charge-transfer transitions of specifically bound transition metal ions. Thus, chromoproteins which may serve as electron transferring proteins, enzymes or photoreceptors, are particularly attractive systems to be studied by RR spectroscopy since an appropriate choice of the excitation wavelength readily leads to a selective enhancement of the Raman bands of the chromophoric site. Moreover, these chromophores generally constitute the active sites of these biomolecules so that RR spectroscopic studies are of utmost importance for elucidating structurefunction relationships. Photoreceptors may use light either for energy conversion (photosynthetic pigments, bacterial rhodopsins) or as a source of information to trigger a physiological response (rhodopsins, phytochromes). Typical chromophores are retinal Schiff bases which are covalently attached to the apoprotein [(bacterial) rhodopsins] and cyclic or linear tetrapyrroles (photosynthetic pigments, phytochromes). In the case of (bacterial) rhodopsins and phytochromes, light absorption initiates a photochemical reaction which is followed by a series of thermal relaxation processes. Thus, RR spectroscopic studies of these biomolecules encounter the difficulty that the exciting laser beam induces the reaction sequence. In order to avoid accumulation of various intermediate states, low temperature spectroscopy is frequently employed so that thermally activated reaction steps can be blocked. Alternatively, time-resolved RR spectroscopy may be an appropriate approach (e.g. bacteriorhodopsin) which, in addition, provides information about the chromophoreprotein dynamics. Primary interest of RR spectroscopic investigations of the visual pigment rhodopsin was directed to the analysis of the photochemical reaction. Singlebeam experiments carried out at 77 K yielded a photostationary equilibrium of the parent state and the primary photoproduct. Using an additional laser beam of a wavelength in resonance with the electronic transition of the product, it was possible to shift the equilibrium largely to the parent state. Thus, subtracting the dual-beam- from the singlebeam-excited spectrum yielded a pure spectrum of the parent state. The interpretation of these spectra was greatly facilitated by the large body of

BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY 91

experimental data of retinal model compounds including a variety of isotopomers as well as of the detailed investigations of bacteriorhodopsin (see below). Moreover, normal mode analyses based on empirical force field calculations support a reliable vibrational assignment allowing for the determination of the configurational and conformational state of the retinal chromophore. In this way, it could be shown that the primary photoprocess of the parent state in which the retinal is in the 11-cis configuration leads to a distorted all-trans configuration. Unusually strong intensities of the CH out-of-plane vibrations of the retinal chain were taken as a measure of the torsion of the CC single bonds of the chain. These intensities decrease in the intermediates formed during the subsequent thermal reactions indicating a relaxation of the chromophore geometry and the immediate protein environment. In the final step, the Schiff base, i.e. the covalent linkage to the apoprotein, was deprotonated as indicated by the shift of the CN stretching vibration and its insensitivity towards H/D exchange. Furthermore, analysis of the RR spectra in terms of retinalprotein interactions contributed essentially to the understanding of colour regulation mechanism in vision. In addition to its photolability, the plant photoreceptor phytochrome, whose chromophore is a linear tetrapyrrole, exhibits a relatively strong fluorescence which severely hinders RR spectroscopic studies under rigorous resonance conditions. Thus, Mathies and co-workers employed shifted-excitation Raman difference spectroscopy in which two spectra, measured with excitation lines separated by 10 cm1, were subtracted from each other. In this way, the fluorescence background was removed yielding only difference signals which were used to calculate the absolute RR spectrum. Similar spectra were obtained by an alternative approach using Fourier-transform Raman spectroscopy at low temperature (77 K). In that case, the excitation line in the near-infrared is shifted from the electronic transition of the chromophore to circumvent fluorescence (and unwanted photochemical processes) but yields a sufficient preresonance enhancement so that the spectra are dominated by the RR bands of the tetrapyrrole. In addition to the spectra of the parent states, those of the intermediates of the photoinduced reaction cycle were obtained by controlled illumination and increase of the temperature. As shown in Figure 1, the spectra of the various states of phytochrome (of the Pr → Pfr transformation) reveal substantial differences reflecting the changes of the chromophore configuration, conformation and interactions with the surrounding protein. In contrast to retinal proteins, however, analysis of the vibrational spectra of linear

tetrapyrroles is much less advanced so that up to now only a little structural information can be extracted from the spectra. Progress in spectral interpretation can be expected by extending the studies to recombinant phytochromes including chemically modified and isotopically labelled chromophores as well as by normal analyses based on quantum chemical force field calculations. For RR spectroscopic studies of photosynthetic pigments, the intrinsic chromophore fluorescence is again an annoyance. In addition to the techniques employed in phytochrome studies, chlorophyll-containing biomolecules allow excitation in resonance with the second (non-fluorescing) electronically excited state (∼ 350400 nm). In the RR spectra of chlorophylls several bands have been identified which are known to be sensitive markers for the coordination state of the central Mg ion. These bands as well as those originating from the C=O stretchings of the formyl and keto substituents provide information about chlorophyllchlorophyll and chlorophyllprotein interactions. This information was particularly helpful for elucidating the structural motifs of the chromophore assemblies in light-harvesting pigments. RR spectroscopic studies of genetically engineered protein variants of reaction centres have contributed to the understanding of those parameters controlling redox potentials and electrontransfer processes. In contrast to photoreceptors, haem proteins are readily accessible for RR spectroscopy inasmuch as they exhibit strong electronic transitions in the visible and near UV. Moreover, they are photostable and do not reveal any interfering fluorescence. There are numerous review articles on the RR spectroscopic results obtained for various representatives of this versatile class of chromoproteins which includes redox enzymes, electron transferring proteins and oxygen transport proteins. Interpretation of the RR spectra is backed by extensive empirical material accumulated from studies on metalloporphyrins in which spectral and structural properties were correlated. Thus, the RR spectra of haem proteins can be interpreted in terms of the kind of the porphyrin, the coordination sphere of the central iron, distortions of the haem geometry, and haemprotein interactions via the porphyrin substituents and the axial ligands. As an example, Figure 2 shows the RR spectra of cytochrome c552 in the oxidized and reduced state, displaying the RR bands in the region of those marker bands whose frequencies are characteristic for the oxidation, spin and coordination state of the haem iron. In addition to those investigations which aim to characterize the structure of the haem pocket, many

92 BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY

Figure 1 Fourier-transform RR spectra (1064 nm) of various states of phytochrome trapped at low temperature. (A) Pr; (B) Lumi-R; (C) Meta-Ra; (D) Meta-Rc; (E) Pfr.

BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY 93

Figure 2 RR spectra (413 nm) of cytochrome c552 of Thermus thermophilus in the reduced state (A) and the oxidized state (B).

studies were directed to solve specific problems related to the function of the haem protein under consideration. Detailed studies were carried out on the cytochrome ccytochrome c oxidase redox couple to understand the interprotein electron transfer mechanism. Specific emphasis was laid on the analysis of the protein complexes formed prior to the electron transfer in order to assess the effect of proteinprotein interaction on the structure of the individual haem sites. In these experiments, dual channel Raman spectroscopy, which permits the quasi-simultaneous measurement of two spectra, is particularly useful for the detection of subtle spectral changes. In this way, RR spectroscopy provided evidence for a considerable conformational flexibility of both proteins which may be of functional relevance. Many studies of haem enzymes (e.g. cytochrome P-450)

were directed to elucidate substrate-binding and substrate-specificity on a molecular level. An interesting approach for probing molecular details of the haem pocket is the use of carbon monoxide which may occupy a vacant coordination site of the haem. The vibrations of the FeCO entity can be analysed in terms of the orientation of the CO ligand with respect to the haem plane, which, in turn, depends on the molecular interactions in the haem pocket. Such studies are particularly relevant for oxygen binding proteins inasmuch as CO is an appropriate analogue for molecular oxygen. Whereas those chromoproteins discussed so far include prosthetic groups with delocalized π-electron systems, the chromophoric site in metalloproteins is constituted by a transition metal ion coordinated by various amino acid side chains. Excitation in resonance with charge-transfer transitions leads to a predominant enhancement of the metalligand vibrations and, to some extent, of internal ligand vibrations. Thus, the corresponding RR spectra are less complex although the interpretation is not necessarily unambiguous and straightforward in each case. Among these proteins, ironsulfur proteins, blue-copper proteins and non-haem iron proteins have been studied extensively by RR spectroscopy. Ironsulfur proteins reveal a characteristic vibrational band pattern in the low frequency region which, in many cases, can be used to identify the type of the ironsulfur cluster based on a comparison with the spectra of model compounds and related proteins of known structure. However, such studies are restricted to the oxidized form since the reduced complexes do not exhibit a charge-transfer transition in the visible region. This is also true for the blue-copper proteins for which the RR spectrum is dominated by modes including the CuS(Cys) stretching coordinate as well as internal modes of the Cys ligand. A substantial improvement of the understanding of these spectra was achieved by using protein variants with isotopically labelled amino acids. The key issue for RR spectroscopists dealing with non-haem iron proteins is to determine the geometry of the oxygen-bridged binuclear complexes as well as to identify the nature of the ligands. In these studies, the RR spectra of 18O-labelled samples significantly supports the interpretation. Nucleic acids

Raman bands of nucleic acids originate from inplane vibrations of the nucleic acid bases (adenine, guanine, cytosine, thymine and uracil) and from the furanose-phosphate backbone. In general, Raman spectra of DNA or RNA reveal structural

94 BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY

information about base stacking and interbase hydrogen bonding interactions. Whereas base stacking is reflected solely by intensity variations of purine and pyrimidine bands, changes of interbase hydrogen interactions, for instance, due to helix formation or conformational transitions of the helix, are indicated by both intensity changes and frequency shifts. Raman spectroscopy offers the possibility to determine the conformation of the nucleic acid inasmuch as there are marker bands indicative for the A-, Band Z-forms of DNA. The most characteristic marker bands originate from the backbone since the various helical structures differ with respect to the sugarphosphate conformation. Based on the comparison of various DNAs of known crystal structure, it was found that the phosphodiester symmetric stretching vibration is at 811 ± 3 cm1 in A-DNA whereas it is upshifted to 835 ± 5 cm1 in the B-form. Furthermore, some ring vibrations of nucleic acid bases, particularly of guanine and cytosine, also serve to distinguish between the various DNA structures (Table 1). For quantitative analysis of the DNA structure, the relative intensities of these specific marker bands can be used to assess the percentage of A-, B- and Z-conformation using the conformation-insensitive band at 1100 cm1 (symmetric stretching of the PO2 group) as a reference. In general, structural changes of DNA in biological processes such as transcription, replication or DNA packaging, are not global but are restricted only to a few nucleotides along the chain. Such changes can induce local melting of the secondary structure, reorientations and/or disruptions of the base stacking. Melting of RNA and DNA double helices always leads to the disappearance of the 814 cm1 and 835 cm1 bands, indicating a decrease in furanose conformational order and an increase of the backbone chain flexibility. Precise thermal melting profiles of both RNA and DNA double helical complexes have been determined based on the intensity variations of these bands. A careful analysis of the Raman spectra of nucleic acids allows the detection of subtle alterations of the helical organization. This has been demonstrated in studies of DNA packaging, which involves formation of DNAprotein complexes with a series of histone and non-histone proteins. A backbone conformation of the B-type family was consistently observed for the nucleosome DNA complexes irrespective of the histone and nonhistone content. Furthermore, complex formation is reflected by the intensity decrease of the adenine and guanine ring modes at 1490 and 1580 cm1, respectively. Whereas the intensity attenuation of the 1580 cm1 band was attributed to the binding of histone proteins in the small grooves of double helical

Table 1 Conformation-sensitive non-resonant Raman bands of nucleic acid

Frequency (cm–1)

Description

805–815

phosphodiester symmetric stretching: C-3′-endo conformation (A-DNA)

835 (weak)

phosphodiester symmetric stretching: C-2′-endo conformation (B-DNA)

870–880

phosphodiester symmetric stretching: C-DNA

682

guanine ring breathing: C-2′-endo-anti (B-DNA)

665

guanine ring breathing: C-3′ endo-anti (A-DNA)

625

guanine ring breathing: C-3′ endo-syn (Z-DNA)

1260 (weak)

cytosine band: B-DNA

1265 (strong)

cytosine band: Z-DNA

1318 (moderate)

guanine band: B-DNA

1318 (very strong)

cytosine band: Z-DNA

1334 (moderate)

guanine band: B-DNA

1355 (moderate)

guanine band: Z-DNA

1362 (moderate)

guanine band: B-DNA

1418 (weak)

guanine band: Z-DNA

1420 (moderate)

guanine band: B-DNA

1426 (weak)

guanine band: Z-DNA

B-form DNA, the spectral changes of the 1490 cm1 band is related to the binding of non-histone proteins in the large grooves. Raman spectroscopy has also been applied to the analysis of the BZ transformation in synthetic polymers induced by drug binding. For example, it was shown that binding of trans-dichlorodiamine platinium to poly(dG-dC)·poly(dG-dC) appeared to inhibit the right-to-left handed transition whereas the antitumour drug cis-dichlorodiamine platinium was found to reduce the salt requirement for adopting a left-handed modified Z-like conformation and rendered the transition essentially non-cooperative. A Raman spectroscopic method to obtain information about dynamic aspects of nucleic acid secondary structure monitors the deuterium exchange of the C-8 purine hydrogen. The exchange

BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY 95

rates can be determined from the time-dependent intensity increase of the bands characteristic for the deuterated bases. As the exchange process depends on the chemical environment and the solvent accessibility for the individual bases, these data allow differentiation of helical structures. Pyrimidine and purine bases of nucleic acids have electronic absorptions in the range between 200 and 280 nm. Using excitation lines in this spectral region, RR spectra of nucleic acids exclusively display bands of in-plane ring modes of the bases without any interference of bands from the backbone. However, a selective enhancement of modes originating from purine or pyrimidine bases is not possible so that the practical use of UV RR spectroscopy for structural investigations of nucleic acids is limited. On the other hand, RR spectroscopy can be employed to probe drugDNA interactions if the electronic transition of the drug is shifted towards the near-UV and visible region. In this way, intercalation of adriamycin by DNA was studied using different excitation lines to probe either the DNA Raman spectrum (364 nm) or the RR spectrum of the bound drug (457 nm). Membranes

Natural biological membranes represent multicomponent systems as they include a large variety of different lipids, proteins and carbohydrates. This heterogeneity is a serious obstacle to the interpretation of Raman spectra. Thus, the majority of spectroscopic studies have focused on liposomes which are taken as models for biological membranes. Studying the fluidity and phase transition of the lipid hydrocarbon chain, a sensitive marker band originates from the longitudinal acoustic mode (LAM) at 100200 cm1, which corresponds to an accordionlike stretching of the entire molecule along its long axis. The frequency of this mode depends on the all-trans chain length and is sensitive to gauche rotations of the chain. The LAM is quite strong in pure hydrocarbons. However, it is difficult to observe due to its proximity to the laser line so that in the Raman spectra of phospholipid dispersions, it could be detected only in a few cases, i.e. at low water content and low temperature. The skeletal optical modes between 1050 and 1150 cm1 and the ratio of the CH stretching bands at 2850 and 2885 cm1, were used to quantify membrane in order to detect membrane order perturbations induced by the incorporation of small hydrophobic molecules in the bilayer. The intensity at 1130 cm1 is thought to be a measure for the number of all-trans sections involving more than three carbon atoms. For phosphatidylcholine vesicles, it was found that the intensity of this band

relative to either the 1660 cm1 band or the CN choline stretching at 718 cm1 could be used to monitor both phase transitions and gradual melting of the lipids. Changes of a stiffening of hydrocarbon chains, e.g. by binding of divalent cations to the phosphate head groups of phosphatidylcholines, is inferred from alterations of the intensity ratio of the 1064 and 1089 cm1 bands. Interactions of membranes with larger molecules are difficult to probe by Raman spectroscopy unless selectivity can be increased by taking advantage of the RR effect. This has been demonstrated in studies on drugmembrane interactions by tuning the excitation line in resonance with the electronic transition of the drug molecule, e.g. amphotericin.

Dynamics of biomolecules Operating in the time-resolved domain, RR spectroscopy can be employed to probe the dynamics of biological systems. Hence, such studies can provide simultaneously kinetic and structural data. In many cases, time-resolved RR spectroscopy, albeit technically demanding, may represent the method of choice for elucidating the molecular mechanisms of biological reactions. Photoinduced processes

Bacterial retinal proteins such as bacteriorhodopsin and halorhodopsin act as light-driven ion pumps. The ion gradient that is generated across the membrane is then converted into chemical energy (i.e. synthesis of adenosine triphosphate). The ion translocation is linked to a photoinduced reaction cycle of the retinal chromophore. The photophysical and kinetic properties as well as the stability of this class of proteins make them an ideal system for timeresolved RR spectroscopic studies. Moreover, these proteins represent suitable objects for developing and optimizing time-resolved spectroscopic techniques. The main advantage is the reversibility of the photoinduced reaction cycle, i.e. the system comes back to the parent state in a few ms after the primary photochemical event. Thus, time-resolved RR spectra can be accumulated continuously since the fresh sample condition is readily established. This condition ensures that the protein is always in the same state when irradiated by the exciting laser beam. There are two approaches for time-resolved RR spectroscopy of these retinal proteins. Using cw-excitation a time-resolution down to 100 ns can be achieved by rapidly moving the sample through the laser focus (rotating cell, capillar flow system). Pulsed laser excitation can provide a time-resolution even in the subpicosecond range depending on the

96 BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY

pulse width of the laser. In both methods, intermediate states are probed in dual-beam experiments with the probe beam irradiating the sample after a delay time Gwith respect to the photolysis beam which initiates the reaction cycle. The systematic variation of G allows the determination of the photocycle kinetics which, along with the structural data derived from the RR spectra, can provide a comprehensive picture of the dynamics of the protein. Figure 3 shows a selection of time-resolved RR of halorhodopsin. The bands displayed in these spectra are diagnostic for the retinal configuration (C=C stretching) and confirm that the photoreaction includes the isomerization from the all-trans to the 13-cis configuration. Redox processes

Cytochrome c oxidase, a membrane-bound enzyme in the respiratory chain of aerobic organisms, reduc-

es oxygen to water. This process which takes place at the binuclear metal centre constituted by a haem a3 and a Cu ion runs via several intermediate states with life times in the micro- and millisecond range. Technical improvements now make it possible to monitor this reaction sequence by time-resolved RR spectroscopy using rapid flow systems. The starting point of the experiments is the thermally stable carbon monoxide complex of the reduced enzyme in oxygen-saturated solution. This complex is photodecomposed by a laser beam so that oxygen can bind to the catalytic centre constituting the time zero of the reaction sequence. The various intermediates are probed by a second laser beam irradiating the (photolysed) sample volume after a delay time. The oxygen-sensitive vibrational modes which can readily be identified based on the 18O/16O isotopic shifts give insight into the nature of the various intermediates and, hence, into the molecular mechanism of the oxygen reduction. A particularly interesting technical approach for these studies is based on an artificial cardiovascular device designed to maintain a continuous enzymatic reaction. Thus, it is possible to accumulate the RR signals during a sufficiently long period of time using a minimum amount of sample.

Living cells and biomedical applications

Figure 3 Time-resolved RR spectra of halorhodopsin from Natronobacterium pharaonis obtained in single (probe only, 514 nm) and dual beam experiments with variable delay times δ of the probe laser (514 nm) relative to the pump laser (600 nm).

Recent progress in laser technology and signal detection has substantially increased the sensitivity of Raman spectroscopy. This is a prerequisite for the development of Raman microspectroscopy and Raman imaging which has opened a new field for the application of Raman spectroscopy in biological and medical research. A particularly powerful method has been introduced by Greve and co-workers who combined Raman spectroscopy with confocal microscopy which allows the measurement of spatially resolved Raman spectra with a lateral resolution of less than 0.5 µm. The technique makes use of sensitive CCD detection systems by probing either the intensity of a marker band in two (spatial) dimensions (global imaging) or a complete spectrum in one dimension (linescan imaging). In these experiments, low energy excitation (> 600 nm) has to be employed and the power of the tightly focused laser beam must be reduced as far as possible to avoid photoinduced degradation of the biological objects. In this way, it is possible to study single living cells and to obtain Raman spectra of the cytoplasm and the nucleus separately. The spectral information allows determination of the DNA conformation, the relative content of the individual nucleic acid bases and the

BIOCHEMICAL APPLICATIONS OF RAMAN SPECTROSCOPY 97

DNA/protein ratio. Furthermore, less abundant biomolecules such as carotenoids or haem enzymes or non-fluorescent drugs (e.g. cobalt octacarboxyphthalocyanine) can be detected by taking advantage of the RR effect. These studies are particularly promising for elucidating metabolic processes in cells. This method has also been applied to medical problems related to disease diagnosis (e.g. artherosclerotic plaque) or optimization of medical treatment (e.g. bone implants). However, medical applications of Raman spectroscopy do not necessarily require the combination with microscopy. Numerous studies have indicated that based on the analysis of Raman spectra by statistical methods it is possible to differentiate between normal and pathological tissues. Despite the substantial technical improvements, the intrinsically low sensitivity of Raman spectroscopy constitutes a limit for general applicability. Fluorescent samples may impose an additional constraint, which, however, can be overcome by near-infrared (1064 nm) Fourier-transform Raman spectroscopy and microscopy. In summary, it appears that for special medical applications Raman spectroscopic techniques may become a powerful diagnostic tool in clinical situations. See also: Biofluids Studied By NMR; Chiroptical Spectroscopy, General Theory; Forensic Science, Applications of IR Spectroscopy; Industrial Applications of lR and Raman Spectroscopy; IR Spectrometers; Medical Science Applications of IR; Membranes Studied By NMR Spectroscopy; Nucleic Acids and Nucleotides Studied Using Mass Spectrometry; Nucleic Acids Studied By NMR; Peptides and Proteins Studied Using Mass Spectrometry; Raman Spectrometers; Surface-Enhanced Raman Scattering (SERS), Applications.

Further reading Althaus T, Eisfeld W, Lohrmann R and Stockburger M (1995) Application of Raman spectroscopy to retinal proteins. Israel Journal of Chemistry 35: 227251. Andel F, Lagarias JC and Mathies RA (1996) Resonance Raman analysis of chromophore structure in the lumiR photoproduct of phytochrome. Biochemistry 35: 1599716008. Andrew CR, Han J, den Blaauwen T et al (1997) Cysteine ligand vibrations are responsible for the complex resonance Raman spectrum of azurin. Journal of Bioinorganic Chemistry 2: 98107. Callender R, Deng H and Gilmanshin R (1998) Raman difference studies of protein structure and folding, enzymatic catalysis and ligand binding. Journal of Raman Spectroscopy 29: 1521. Carey PR (1998) Raman spectroscopy in enzymology: the first 25 years. Journal of Raman Spectroscopy 29:714.

Clark RJH and Hester RE (eds) (1986) Spectroscopy of biological systems. Advances in Spectroscopy 13. Clark RJH and Hester RE (eds) (1993) Biomolecular spectroscopy, part A. Advances in Spectroscopy 20. Clark RJH and Hester RE (eds) (1993) Biomolecular spectroscopy, part B. Advances in Spectroscopy 21. Hildebrandt P (1992) Resonance Raman spectroscopy of cytochrome P-450. In: Ruckpaul K and Rein H (eds) Frontiers in Biotransformation, Vol. 7, pp 166215. Berlin/Weinheim: Akademie-Verlag/VCH. Hildebrandt P (1995) Resonance Raman spectroscopy of cytochrome c. In: Scott RA and Mauk AG (eds) Cytochrome c. A Multidisciplinary Approach, pp 285314. Mill Valley: University Science. Kneip C, Mozley D, Hildebrandt P, Gärtner W, Braslavsky SE and Schaffner K (1997) Effect of chromophore exchange on the resonance Raman spectra of recombinant phytochromes. FEBS Letters 414: 2326. Kochendoerfer GG, Wang Z, Oprian DD and Mathies RA (1997) Resonance Raman examination of wavelength regulation mechanism in human visual pigments. Biochemistry 36: 65776587. Manoharan R, Wang Y and Feld MS (1996) Histochemical analysis of biological tissues using Raman spectroscopy. Spectrochimica Acta A 52: 215249. Otto C, de Grauw CJ, Duindam JJ, Sijtsema NM and Greve J (1997) Applications of micro-Raman imaging in biomedical research. Journal of Raman Spectroscopy 28: 143150. Spiro TG (ed) (1987) Biological Application of Raman Spectroscopy, Vol 1: Raman spectra and the conformations of biological macromolecules. New York: John Wiley & Sons. Spiro TG (ed) (1987) Biological Application of Raman Spectroscopy, Vol 2: Resonance Raman spectra of polyenes and aromatics. New York: John Wiley & Sons. Spiro TG (ed) (1988) Biological Application of Raman Spectroscopy, Vol 3: Resonance Raman spectra of haem and metalloproteins. New York: John Wiley & Sons. Tensmeyer LG and Kauffmann EW (1996) Protein structure as revealed by nonresonance Raman spectroscopy. In: Havel HA (ed) Spectroscopic Methods for Determining Protein Structure in Solution, pp 6995. New York: VCH. Thamann TJ (1996) Probing local protein structure with ultraviolet resonance Raman spectroscopy. In: Havel HA (ed) Spectroscopic Methods for Determining Protein Structure in Solution, pp 96134. New York: VCH. Twardowski J and Anzenbacher P (1994) Raman and IR Spectroscopy in Biology and Biochemistry. New York: Ellis Horwood. Verma SP and Wallach DFH (1984) Raman spectroscopy of lipids and biomembranes. In: Chapman D (ed) Biomembrane Structure and Function, pp 167198. Weinheim: Verlag Chemie. Zhao X and Spiro TG (1998) Ultraviolet resonance Raman spectroscopy of hemoglobin with 200 and 212 nm excitation: H-bonds of tyrosines and prolines. Journal of Raman Spectroscopy 29: 4955.

98 BIOFLUIDS STUDIED BY NMR

Biofluids Studied By NMR John C Lindon and Jeremy K Nicholson, Imperial College of Science, Technology and Medicine, London, UK

MAGNETIC RESONANCE Applications

Introduction Investigation of biofluid composition provides insight into the status of a living organism in that the composition of a particular fluid carries biochemical information on many of the modes and severity of organ dysfunction. One of the most successful approaches to biofluid analysis has been the application of NMR spectroscopy. The complete assignment of the 1H NMR spectrum of most biofluids is not possible (even by use of 900 MHz NMR spectroscopy) owing to the enormous complexity of the matrix. However, the assignment problems vary considerably between biofluid types. For instance, seminal fluid and blood plasma are highly regulated with respect to metabolite composition and concentrations and the majority of the NMR signals have been assigned at 600 and 750 MHz for normal human individuals. Urine composition is much more variable because its composition is normally adjusted by the body in order to maintain homoeostasis and hence complete analysis is much more difficult. There is also enormous variation in the concentration range of NMR-detectable metabolites in urine samples. With every new increase in available spectrometer frequency the number of resonances that can be resolved in a biofluid increases and although this has the effect of solving some assignment problems, it also poses new ones. Furthermore, problems of spectral interpretation arise due to compartmentation and binding of small molecules in the organized macromolecular domains that exist in some biofluids such as blood plasma and bile. All biological fluids have their own characteristic physicochemical properties and a summary of some of these is given in Table 1 for normal biofluids. These partly dictate the types of NMR experiment that may be employed to extract the biochemical information from each fluid type. An illustration of the complexity of biofluid spectra, and hence the need for ultrahigh field measurements, is given in Figure 1 which shows 800 MHz 1H NMR spectra of normal human urine, bile and blood plasma. It is clear that at even the present level of technology in NMR, it is not yet possible to detect many

important biochemical substances, e.g. hormones, in body fluids because of problems with sensitivity, dispersion and dynamic range and this area of research will continue to be technology limited. With this in mind, it would seem prudent to interpret quantitative 1H NMR measurements of intact biological materials and assignment of resonances in 1D spectra with considerable caution even when measured at ultrahigh field.

Resonance assignment in NMR spectra of biofluids Usually in order to assign 1H NMR spectra of biofluids, comparison is made with spectra of authentic materials and by standard addition to the biofluid sample. Additional confirmation of assignments is usually sought from the application of 2-dimensional (2D) NMR methods, particularly COSY and TOCSY and, increasingly, inverse-detected heteronuclear correlation methods such as HMQC and HSQC. In addition, the application of the 2D J-resolved (JRES) pulse sequence is useful for spreading out the coupling patterns from the multitude of small molecules in a biofluid. Even 2D correlation NMR spectra of complex biofluids can show much overlap of crosspeaks and further editing is often desirable. Thus simplification of 1D and 2D NMR spectra of biofluids can also be achieved using (i) spinecho methods particularly for fluids containing high levels of macromolecules (T2 editing), (ii) editing based on T1 relaxation time differences, (iii) editing based on diffusion coefficient differences, (iv) multiple quantum filtering. One major advantage of using NMR spectroscopy to study complex biomixtures is that measurements can often be made with minimal sample preparation (usually with only the addition of 510% D 2O) and a detailed analytical profile can be obtained on the whole biological sample. This in turn requires good methods for suppressing solvent resonances. Detailed 1H NMR spectroscopic data for a wide range of low molecular weight metabolites found in biofluids are given in Table 2.

BIOFLUIDS STUDIED BY NMR 99

Table 1

Normal biofluids and their physicochemical properties

Biofluid

Function

Water content a

Viscosity

Protein content b Lipid content b

Peak overlap c

Urine

Excretion Homoeostasis

+++

Bile

Excretion Digestion

+++

Blood plasma

Transport Homoeostasis Mechanical

+++

Whole blood d

Transport Oxygenation

+++

Cerebrospinal fluid

Transport Homoeostasis Mechanical

+++

Milk

Nutrition

+++

Saliva

Excretion Digestion

+++

Gastric juice

Digestion

+++

Pancreatic juice

Digestion

+++

Seminal fluid

Support for spermatozoa

+++

Prostatic fluid

Support for spermatozoa

Seminal vesicle fluid

Support for spermatozoa

+++

Amniotic fluid

Protection of fetus

+++

Follicular fluid

Reproduction

+++

Synovial fluid

Joint protection

+++

Aqueous humour

Eye function

+++

+++, ++, + indicate high, medium, low degree of constraint for NMR studies a Relative water intensity when compared with concentrations of metabolities of interest. b In biofluids with high protein or lipid contents, spin-echo spectra must normally be employed to eliminate broad resonances. c A subjective indication of spectral crowding (at 600 MHz) due to abundance of endogenous metabolites with a wide range of shifts. d Presence of cells and process of cell sedimentation gives rise to magnetic field inhomogeneity problems. e Not a limiting factor.

NMR studies of dynamic interactions Although NMR spectroscopy of biofluids is now well established for probing a wide range of biochemical problems, there are still many poorly understood physicochemical phenomena occurring in biofluids, particularly the subtle interactions occurring between small molecules and macromolecules or between organized multiphasic compartments. The understanding of these dynamic processes is of considerable importance if the full diagnostic potential of biofluid NMR spectroscopy is to be realized. Typically, it is now possible to study enzymatic reactions, chemical reactions and biofluid instability, microbiological activity in biofluids, macromolecular binding of small molecules, membrane-based compartmentation, metal complexation and chemical exchange processes.

NMR spectroscopy of blood plasma and whole blood NMR peak assignments

The physicochemical complexity of plasma shows up in its 1H NMR spectra by the range of linewidths of the signals. This means that a number of different multiple pulse NMR experiments and/or physicochemical interventions must be applied to extract useful biochemical information. Numerous high resolution 1H NMR studies have been performed on the biochemistry of blood and its various cellular components and plasma. The physical properties of whole blood pose serious limitations on direct NMR investigations, but packed erythrocytes yield more useful information on cell biochemistry. Well resolved spectra are given by plasma, and 1H NMR

100 BIOFLUIDS STUDIED BY NMR

Figure 1 800 MHz 1H NMR spectra of control human biofluids; (A) urine; (B) gall bladder bile and (C) blood plasma. Reproduced with permission of Academic Press from Lindon JC, Nicholson JK and Everett JR (1999) NMR spectroscopy of biofluids. Annual Reports on NMR Spectroscopy 38: in the press.

measurements on blood serum and plasma can provide much useful biochemical information on both low molecular weight metabolites and macromolecular structure and organization. In blood plasma, the 1H NMR peaks of metabolites, proteins, lipids and lipoproteins are heavily overlapped even at 800 MHz (Figure 1). Most blood plasma samples are quite viscous and this gives rise to relatively short T1 relaxation times for small molecules and this allows relatively short pulse repetition cycles without signal saturation. The spectral profile can be simplified by use of spinecho experiments with an appropriate T2 relaxation delay to allow signals from broad macromolecular components and compounds bound to proteins to be attenuated. By the early 1980s many metabolites had been detected in normal blood plasma although assignments were, in general, based on the observation of only one or two resonances for each metabolite. In addition, peaks from certain macromolecules

such as D1-acid glycoprotein (N-acetyl neuraminic acid and related sialic acid fragments) have been assigned and used diagnostically, in particular their N-acetyl groups which give rise to relatively sharp resonances presumably due to less restricted molecular motion. The signals from some lipid and lipoprotein components, e.g. very low density lipoprotein (VLDL), low density lipoprotein (LDL), high density lipoprotein (HDL) and chylomicrons, have also been partially characterized. For normal plasma at pH 7, the largest peak in the spectral region to high frequency of water is that of the D-anomeric H1 resonance of glucose at G5.223 (which provides a useful internal chemical shift reference). In spectra of normal human and animal plasma, there are few resonances in the chemical shift range to high frequency of G5.3 when measured in the pH range 3 to 8.5. However, on acidification of the plasma to pH < 2.5, resonances from histidine and phenylalanine become detectable. Experiments

BIOFLUIDS STUDIED BY NMR 101

with model solutions suggested that serum albumin has a high capacity for binding aromatic amino acids and histidine at neutral pH and this is responsible for their NMR-invisibility in normal human blood plasma. Serum albumin also binds a large number of other species of both endogenous and xenobiotic origin. Even commercial purified bovine serum albumin (BSA) can be shown to contain a significant amount of bound citrate and acetate which become NMR-detectable in BSA solutions at pH 2. Acidification of human plasma also renders citrate NMR-detectable in spinecho spectra as it becomes mobilized from the protein binding sites. Through the increased spectral dispersion available from the use of 600, 750 and 800 MHz 1H NMR measurements and through the use of a variety of 2D methods, the assignment of resonances in blood plasma spectra in normal individuals is now extensive (see Table 2). Blood plasma also has intrinsic enzymatic activities although many of these are not stable (particularly if the sample is not frozen immediately on collection). It has been noted that under certain pathological conditions, such as those following liver or kidney damage, enzymes that are present at elevated levels in the plasma because of leakage from the damaged tissue can cause NMR-detectable alterations to spinecho spectra of plasma. Molecular diffusion coefficients are parameters that are not related directly to NMR spectral intensities under normal conditions. However, molecular diffusion can cause NMR signal intensity changes when pulsed field gradients are applied during the NMR experiment. A number of pulse sequence developments have meant that measurement of diffusion coefficients is relatively routine. The editing of 1H NMR spectra of biofluids based on diffusion alone or on a combination of spin relaxation and diffusion has been demonstrated recently. This approach is complementary to the editing of 1H NMR spectra based on differences in T1 and T2. New methods for editing TOCSY NMR spectra of biofluids have been proposed based on differences in molecular diffusion coefficients and this has been termed Diffusion Edited TOCSY (DETOCSY). Much study has been devoted to the problem of lipoprotein analysis in blood plasma using 1H NMR spectroscopy. This has been comprehensively reviewed recently by Ala-Korpela. Lipoproteins are complex particles that transport molecules normally insoluble in water. They are spherical with a core region of triglyceride and cholesterol ester lipids surrounded by phospholipids in which are embedded various proteins known as apolipoproteins. In addition, free cholesterol is found in both the core and surface regions. The lipoproteins are in a dynamic

equilibrium with metabolic changes going on in vivo. Lipoproteins are usually classified into five main groups, chylomicrons, very low density lipoprotein (VLDL), low density lipoprotein (LDL), intermediate density lipoprotein (IDL) and high density lipoprotein (HDL) based on physical separation using centrifugation. Based on the measurement of 1H NMR spectra of the individual fractions and using lineshape fitting programs, it has been possible to identify the chemical shifts of the CH2 and CH3 groups of the fatty acyl side chains. Quantification can be carried out using either time-domain or frequencydomain NMR data. The usefulness of 1H NMR spectra for lipoprotein analysis and 31P NMR spectroscopy for phospholipid analysis in blood plasma has been explored. More recently a neural network software approach has been used to provide rapid lipoprotein analyses. NMR spectra of blood plasma in pathological states

A good deal of excitement was generated by a publication which reported that 1H NMR spectroscopy of human blood plasma could be used to discriminate between patients with malignant tumours and other groups, namely normals, patients with non-tumour disease and a group of patients with certain benign tumours. This paper stimulated many other research groups around the world to investigate this approach to cancer detection. The test as originally published involved the measurement of the averaged line width at half height of the two composite signals at G1.2 and G0.8 in the single pulse 1H NMR spectrum of human blood plasma. On comparing the averaged signal width, W, for normal subjects with those for patients with a variety of diseases and with those for pregnant women, it was reported that several statistically significant differences existed between the groups. It was proposed that the lowering of W observed in the malignant group was due to an increase in the T2* of the lipoproteins. A mass of literature now exists on this subject and it was immediately apparent that the perfect 100% sensitivity and specificity reported in the early publications has never been repeated by any group. In addition to the presence of cancer, a number of other factors have been found to cause changes in the linewidth index, W. These include diet, age and sex, pregnancy and trauma as well as hyperlipidaemia. In all cases the observed changes in W are caused by alterations in the plasma lipoprotein composition, especially the VLDL/HDL ratio. The success of heart transplantation has improved recently with the chronic use of cyclosporin A to

102 BIOFLUIDS STUDIED BY NMR

Table 2

H NMR assignments and chemical shifts for metabolities found in biofluids

Metabolite

Assignment

G(1H)

Multiplicity

G(13C)

D-Hydroxyisovalerate

CH3

0.81

C,U

D-Hydroxy-n-butyrate

CH3

0.90

C,U

n-Butyrate

CH3

0.90

D-Hydroxy-n-valerate

CH3

0.92

Isoleucine

G-CH3

0.94

A,C,P,S,U

Leucine

G-CH3

0.96

A,C,P,U

Leucine

G-CH3

0.97

A,C,P,U

D-Hydroxyisovalerate

CH3

0.97

Valine

CH3

0.99

19.6

C,P,U

Isoleucine

E-CH3

1.01

14.6

A,C,P,S,U

Valine

CH3

1.04

A,C,P,U

Ethanol

CH3

1.11

C,P,U

Propionate

CH3

1.12

Isobutyrate

CH3

1.13

E-Hydroxybutyrate

CH3

1.20

C,P,S,U

Isoleucine

J-CH2

1.26

A,C,P,S,U

Fucose

CH3

1.31

Lactate

CH3

1.33

Threonine

CH3

1.34

A,C,P,S,U

D-Hydroxyisobutyrate

CH3

1.36

D-Hydroxy-n-valerate

J-CH2

1.37

Alanine

CH3

1.48

Lysine

J-CH2

1.48

C,P,S,U

Isoleucine

J-CH2

1.48

C,P,S,U

n-Butyrate

E-CH2

1.56

P,U

Adipate

CH2

1.56

P,U

Citrulline

J-CH2

1.58

C,U

D-Hydroxy-n-valerate

E-CH2

1.64

D-Hydroxybutyrate

CH2

1.70

Arginine

J-CH2

1.70

Leucine

CH2

1.71

Lysine

G-CH2

1.73

C,P,S,U

Ornithine

J-CH2

1.81

C,P,S,U

Citrulline

E-CH2

1.88

C,P,S,U

N-Acetylglutamate

E-CH2

1.89

C,P,S,U

J-Amino-n-butyrate

E-CH2

1.91

C,P,S,U

Lysine

E-CH2

1.91

Arginine

E-CH2

1.93

C,P,S,U

Ornithine

E-CH2

1.95

A,S,U

Biofluid a

C,U

P 20.9

16.8

40.7

30.3

A,C,P,S,U

C,P,S,U

Acetate

CH3

1.95

A,C,P,S,U

Isoleucine

E-CH

1.98

C,P,U

Acetamide

CH3

2.01

N-Acetyl groups(glycoproteins)

CH3

2.02

D-Hydroxyisovalerate

E-CH

2.02

U 23.0

P,U U

BIOFLUIDS STUDIED BY NMR 103

Table 2

continued

Metabolite

Assignment

G(1H)

Multiplicity

G(13C)

N-Acetylaspartate

CH3

2.03

Proline

J-CH2

2.01

C,S,U

N-Acetyl groups(glycoproteins)

CH3

2.05

P,U

Biofluid a C,U

Proline

E-CH2

2.07

C,S,U

N-Acetylglutamate

CH3

2.04

N-Acetylglutamate

E-CH2

2.06

Glutamate

E-CH2

2.10 b

Glutamine

E-CH2

2.14b

C,P,S,U

Methionine

S-CH3

2.14

A,C,P,S,U

n-Butyrate

D-CH2

2.16

C,P,S,U

Methionine

E-CH2

2.16

A,C,P,S,U

Adipate

CH2COOH

2.22

N-Acetylglutamate

J-CH2

2.23

Acetone

CH3

2.23

P,U

U 30.1

A,C,P,S,U

Valine

E-CH

2.28

C,P,S,U

Acetoacetate

CH3

2.29

C,P,S,U

J-Amino-n-butyrate

D-CH2

2.30

E-Hydroxybutyrate

CH2

2.31

ABX

C,P,S,U

Proline

E-CH2

2.35

A,C,S,U

Glutamate

J-CH2

2.36

Oxalacetate

CH2

2.38

Pyruvate

CH3

2.38

C,P,U

Malate

CH2

2.39

E-Hydroxybutyrate

CH2

2.41

ABX

C,P,S,U

Succinate

CH2

2.43

C,P,S,U

Carnitine

CH2(COOH)

2.44

C,P,S,U

D-Ketoglutarate

J-CH2

2.45

C,P,S,U

Glutamine

J-CH2

2.46b

34.5

C,U C,P,U

31.9

C,P,S,U

Glutamate

J-CH2

2.50

A,C,P,S,U

N-Acetylaspartate

CH2

2.51

ABX

C,U

Methylamine

CH3

2.54

Citrate

CH2

2.67

A,C,P,S,U

Methionine

S-CH2

2.65

C,P,S,U

Aspartate

E-CH2

2.68c

ABX

C,P,S,U

Malate

CH2

2.69

ABX

N-Acetylaspartate

CH2

2.70

ABX

C,U

Dimethylamine

CH3

2.72

C,P,S,U

Sarcosine

CH3

2.74

Dimethylglycine

CH3

2.78

P,U

Citrate

CH2

2.80

A,C,P,S,U

Aspartate

E-CH2

2.82

ABX

C,P,S,U

Methylguanidine

CH3

2.83

Asparagine

E-CH2

2.86

C,P,S,U

Trimethylamine

CH3

2.88

104 BIOFLUIDS STUDIED BY NMR

Table 2

continued

Metabolite

Assignment

G(1H)

Multiplicity

G(13C)

Asparagine

E-CH2

2.96

C,P,S,U

D-Ketoglutarate

E-CH2

3.01

C,P,S,U

J-Amino-n-butyrate

J-CH2

3.02

Lysine

H-CH2

3.03

Cysteine

CH2

3.04

C,U

Creatine

CH3

3.04

A,C,P,S,U

Phosphocreatine

CH3

3.05

Creatinine

CH3

3.05

A,C,P,S,U

Tyrosine

CH2

3.06

ABX

C,P,S,U

Ornithine

G-CH2

3.06

C,S,U

Cysteine

CH2

3.12

ABX

C,U

Malonate

CH2

3.13

Phenylalanine

E-CH2

3.13

C,P,S,U

Histidine

E-CH2

3.14

ABX

C,P,S,U

Citrulline

D-CH2

3.15

C,U

cis-Aconitate

CH2

3.17

Tyrosine

CH2

3.20

ABX

Choline

N(CH3)3

3.21

Phosphorylethanolamine

NCH2

3.23

E-Glucose

C-H2

3.24

A,C,P,S,U

Histidine

E-CH2

3.25

ABX

Arginine

G-CH2

3.25

Trimethylamine-N-oxide

N(CH3)3

3.27

C,P,S,U

Taurine

CH2SO3

3.25

A,C,P,S,U

Betaine

N(CH3)3

3.27

C,P,S,U

Phenylalanine

E-CH2

3.28

C,P,S,U

Myo-inositol

3.28

Tryptophan

CH2

3.31

ABX

P,S,U

Glycerophosphorylcholine

N(CH3)3

3.35

Proline

G-CH2

3.33

E-Glucose

C-H4

3.40

70.6 70.6

Biofluid a

C 40.3

C,P,S,U

C,P,S,U 55.0

C,P,S,U

C,P,S,U 41.3

C,S,U

A,P,U A,C,P,S,U

D-Glucose

C-H4

3.41

Proline

G-CH2

3.42

A,P,U

Carnitine

NCH2

3.43

C,P,S,U

Taurine

NCH2

3.43

A,C,P,S,U

Acetoacetate

CH2

3.45

C,P,S,U

E-Glucose

C-H5

3.47

ddd

trans-Aconitate

CH2

3.47

76.7

A,C,P,S,U

A,C,P,S,U U

D-Glucose

C-H3

3.49

Tryptophan

CH2

3.49

ABX

P,U

Choline

NCH2

3.52

C,P,S,U

Glycerophosphorylcholine

NCH2

3.52

D-Glucose

C-H2

3.53

72.3

A,C,P,S,U

Glycerol

CH2

3.56

ABX

63.5

C,P

Myo-inositol

H1/H3

3.56

76.7

A,C,P,S,U

BIOFLUIDS STUDIED BY NMR 105

Table 2

continued

Metabolite

Assignment

G(1H)

Multiplicity

G(13C)

Glycined

CH2

3.57

C,P,S,U

Threonine

D-CH

3.59

Fructose (E-furanose)

3.59

Fructose (E-furanose)

3.59

Fructose (E-pyranose)

3.60

Sarcosine

CH2

3.61

C,P,S,U

Ethanol

CH2

3.61

Valine

D-CH

3.62

Biofluid a

C,P,S,U 64.2

C,P,S,U

Myo-inositol

H4/H6

3.63

Fructose (E-pyranose)

H6′

3.63

Fructose (E-pyranose)

3.64

Glycerol

CH2

3.65

ABX

Isoleucine

D-CH

3.68

A,C,P,S,U

63.5

C,P

Fructose (E-furanose)

H6′

3.70

Fructose (E-furanose)

3.70

D-Glucose

C-H3

3.71

73.6

C,P,S,U

E-Glucose

C-H6′

3.72

61.6

A,C,P,S,U

Leucine

D-CH

3.73

55.1

A,C,P,S,U

D-Glucose

C-H6′

3.74

61.4

A,C,P,S,U

Ascorbate

CH2

3.74

Ascorbate

CH2

3.76

Citrulline

D-CH

3.76

C,U

Lysine

D-CH

3.76

Glutamine

D-CH

3.77

Glutamate

D-CH

3.77

A,C,P,S,U

Arginine

D-CH

3.77

C,U

C,P,S,U 55.4

72.6

C,P,S,U

Glycerol

3.79

ABX

Alanine

3.79

C,P A,C,P,S,U

Ornithine

D-CH

3.79

C,P,S,U

Guanidoacetate

CH2

3.80

Mannitol

CH3

3.82

Fructose (E-furanose)

3.82

Fructose (E-pyranose)

3.84

D-Glucose

C-H6

3.84

61.4

A,C,P,S,U

D-Glucose

C-H5

3.84

ddd

72.3

A,C,P,S,U

Fructose (E-furanose)

3.85

D-Hydroxyisovalerate

D-CH

3.85

Serine

D-CH

3.85

ABX

Methionine

D-CH

3.86

E-Glucose

C-H6

3.90

Fructose (E-pyranose)

3.90

Betaine

CH2

3.90

C,P,S,U

Aspartate

D-CH

3.91

ABX

C,U

4-Aminohippurate

CH2

3.93

Creatine

CH2

3.93

A,C,P,S,U 61.6

A,C,P,S,U

106 BIOFLUIDS STUDIED BY NMR

Table 2

continued Assignment

G(1H)

Glycolate

CH2

3.94

Tyrosine

3.94

ABX

C,P,S,U

Phosphocreatine

CH2

3.95

Serine

E-CH2

3.95

ABX

Hippurate

CH2

3.97

C,P,S,U

Histidine

D-CH

3.99

ABX

C,P,S,U

D-Hydroxybutyrate

4.00

ABX

Asparagine

D-CH

4.00

ABX

C,S,U

Cysteine

4.00

ABX

Serine

E-CH2

4.00

ABX

C,U

Phosphorylethanolamine

OCH2

4.00

Phenylalanine

D-CH

4.00

C,P,S,U

Fructose (E-pyranose)

4.01

Metabolite

Multiplicity

G(13C)

Biofluid a

Ascorbate

4.03

Fructose (E-pyranose)

4.01

D-Hydroxy-n-valerate

D-CH

4.05

Creatinine

CH2

4.06

A,C,P,S,U

Tryptophan

4.06

ABX

S,U

Myo-Inositol

4.06

C,S,U

Choline

OCH2

4.07

C,P,S,U

Lactate

4.12

Fructose (E-furanose)

4.13

Proline

D-CH

4.14

A,C,P,S,U

E-Hydroxybutyrate

4.16

ABX

C,P,S,U

Threonine

E-CH

4.26

ABX

A,C,P,S,U

69.2

A,C,P,S,U S

Malate

4.31

N-methylnicotinamide

CH3

4.48

Glycerophosphorylcholine

NCH2

3.52

N-Acetylaspartate

4.40

ABX

C,U

E-Galactose

C-H1

4.52

Ascorbate

4.52

E-Galactose

C-H1

4.53

E-Glucose

C-H1

4.64

A,C,P,S,U

Water

H2 O

4.79

all fluids

Phospho(enol)pyruvate

5.19

D-Glucose

C-H1

5.23

92.9

A,C,P,S,U

Allantoate

5.26

Phospho(enol)pyruvate

5.37

Allantoin

5.40

Urea

NH2

5.78

P,U

Uridine

5.80

C,S,U

Uridine

H1c

5.82

C,S,U

cis-Aconitate

5.92

Urocanate

CH(COOH)

6.40

Fumarate

6.53

BIOFLUIDS STUDIED BY NMR 107

Table 2

continued Assignment

G(1H)

trans-Aconitate

6.62

4-Aminohippurate

C-H3/H5

6.87

3,4-Dihydroxymandelate

cyclic H

6.87

3,4-Dihydroxymandelate

cyclic H

6.90

Tyrosine

C-H3/H5

6.91

d′

3,4-Dihydroxymandelate

cyclic H

6.94

3-Methylhistidine

C-H4

7.01

Metabolite

Multiplicity

G(13C)

116.7

Biofluid a

C,P,S,U

Histidine

C-H4

7.08

C,P,S,U

1-Methylhistidine

C-H4

7.05

Tyrosine

C-H2/H6

7.20

C,P,S,U

Tryptophan

C-H5/H6

7.21

S,U

Indoxyl sulphate

C-H5

7.20

Indoxyl sulphate

C-H6

7.28

C,U

Tryptophan

C-H5/H6

7.29

S,U

Urocanate

CH(ring)

7.31

Tryptophan

C-H2

7.33

S,U

Phenylalanine

C-H2/H6

7.33

C,P,S,U

Phenylalanine

C-H4

7.38

C,P,S,U

Urocanate

C-H5

7.41

Phenylalanine

C-H3/H5

7.43

C,P,S,U

Nicotinate

cyclic H

7.53

Hippurate

C-H3/H5

7.55

Tryptophan

C-H7

7.55

S,U

3-Methylhistidine

C-H2

7.61

Hippurate

C-H4

7.64

4-Aminohippurate

C-H2/H6

7.68

Tryptophan

C-H4

7.74

S,U

1-Methylhistidine

C-H2

7.77

Uridine

C-H6

7.81

C,S,U

Histidine

C-H2

7.83

C,P,S,U

Hippurate

C-H2/H6

7.84

Urocanate

cyclic C-H3

7.89

N-methylnicotinamide

C-H5

8.19

Nicotinate

cyclic H

8.26

Formate

8.46

C,P,S,U

Nicotinate

cyclic H

8.62

N-methylnicotinamide

C-H4

8.90

Nicotinate

cyclic H

8.95

N-methylnicotinamide

C-H6

8.97

N-methylnicotinamide

C-H2

9.28

b c d e

Main biofluid in which compounds are found or have been observed; A, C, P, S and U refer to observation in amniotic fluid, CSF, Serum, Seminal fluid and urine, respectively. Varies according to the presence of divalent metal ions (especially Ca, Mg and Zn). Highly variable over physiological pH range. Significant shift differences due to the formation of fast exchanging complexes with divalent metal ions in biofluids. Not normally observed in biofluids because of low stability.

108 BIOFLUIDS STUDIED BY NMR

suppress rejection, and the better treatment of acute rejection episodes when they do occur. However, the early detection of acute cardiac graft rejection still relies on invasive and iterative right ventricular endomyocardial biopsies. This led to a new application of the measurement of the linewidth index, W, for the assessment of heart graft rejection after transplantation. Patients in moderate and severe rejection showed W values that were significantly different from those of the light rejection patients. However, as occurred in the test for cancer, the wide overlap between the values observed for each group meant that the W value alone could not be used to classify the patients into the four rejection grades. It has also been reported that the areas of two glycoprotein signals in the spinecho 1H NMR spectra of blood plasma from heart transplant patients correlated with a standard echocardiography parameter used to monitor rejection. Although a good correlation was found in 5 patients and an acceptable correlation in 3, only a poor correlation was seen in 5 further patients. It was thought that infections and inflammatory states unrelated to rejection interfered with the correlation. In human beings, diabetes is a relatively common condition which can have serious, complex and farreaching effects if not treated. It is characterized by polyuria, weight loss in spite of increased appetite, high plasma and urinary levels of glucose, metabolic acidosis, ketosis and coma. The muscles and other tissues become starved of glucose, whilst highly elevated levels of glucose are found in the urine and plasma. Based on NMR spectra, there are marked elevations in the plasma levels of the ketone bodies and glucose, post-insulin withdrawal. In general, the NMR results were in good agreement with conventional assay results. The CH3 and CH2 resonances of the lipoproteins VLDL and chylomicrons also decreased significantly in intensity relative to the CH3 signal of HDL and LDL, indicating the rapid metabolism of the mobile pool of triglycerides in VLDL and chylomicrons. The 1H NMR spectra of the blood plasma from patients with chronic renal failure during dialysis, patients in the early stages of renal failure and normals have also been analysed. For patients on acetate dialysis, the method clearly showed how the acetate was accumulated and metabolized during the course of the dialysis, as well as allowing changes in the relative concentrations of endogenous plasma components to be monitored. A subsequent 1H, 13C and 14N NMR study of the plasma and urine from chronic renal failure patients showed that the plasma levels of trimethylamine-N-oxide (TMAO) correlated with those of urea and creatinine, suggesting

that the presence of TMAO is closely related to the degree of renal failure. The uraemic syndrome is associated with a complex set of biochemical and pathophysiological changes that remain poorly understood. The first application of 750 MHz NMR for studying the biochemical composition of plasma has been reported from patients on haemodialysis (HD) and peritoneal dialysis (PD). Increased plasma levels of low MW metabolites including methylhistidine, glycerol, choline, TMAO, dimethylamine and formic acid were found. The concentrations of these metabolites, and ratios to others, varied with the type of dialysis therapy. For example, the biochemical composition of plasma from patients on PD was remarkably consistent whereas pre- and post-HD significant fluctuations in the levels of TMAO, glucose, lactate, glycerol, formate and lipoproteins (VLDL, LDL and HDL) were observed. Elevated concentrations of glycoprotein fragments were also observed and may relate to the presence of high levels of N-acetyl glucosaminidase (NAG), and other glycoprotein cleaving enzymes, in the plasma originating from damaged kidney cells. 1H NMR spectroscopy of whole blood and red blood cells

Single pulse 1H NMR measurements on whole blood give very little biochemical information owing to the presence of a broad envelope of resonances from haemoglobin and plasma proteins. Spinecho spectroscopy of whole blood can give rise to moderately well resolved signals from plasma metabolites and those present inside erythrocytes, notably glutathione, but whole blood spectra are not easily reproducible because of erythrocyte sedimentation which progressively (within a few minutes) degrades the sample field homogeneity during the course of data collection. Furthermore, there are substantial intracellular/extracellular field gradients, which give a major contribution to the T2 relaxation processes for the nuclei of molecules diffusing through those gradients, and very weak spectra are obtained. Spin echo NMR measurements on packed erythrocyte samples do give rise to well resolved signals from intracellular metabolites and a variety of transport and cellular biochemical functions can be followed by this method. One major parameter that can be obtained from NMR is the intracellular pH, and this has been measured for erythrocytes by 1H NMR spectroscopy using a suitable 1H NMR pH indicator that has an NMR chemical shift which varies with pH in the

BIOFLUIDS STUDIED BY NMR 109

range desired. Candidates include the C2-H protons from histidines in haemoglobin, but an exogenous compound can be added as an indicator. The transport of substances between the inside and outside of red cells can be monitored using NMR if the resonances from the two environments have different chemical shifts or intensities. Also, resonances outside the cell can be selectively broadened by the addition of paramagnetic species that do not cross the red cell membrane. Those used include the ferric complex of desferrioxamine, dysprosium-DTPA and the copper-cyclohexanediaminetetraacetic acid complex. To measure the rate of influx of a compound into red cells, the compound is added with the paramagnetic agent to a red cell suspension and the intensity of the resonance from the intracellular component is monitored as a function of time. This approach has been used to study the transport of glycerol, alanine lactate, choline and glycylglycine. The time scale that can be addressed covers the range of ms to hours. Disease processes studied include the use of NMR spectroscopy of erythrocytes to investigate the effect of lithium treatment in manic depressive patients showing an increase in choline. In heavy metal poisoning a considerable proportion of the metal is found in the blood and the binding of heavy metals inside erythrocytes has received much attention. Rabenstein has reviewed much of the work in the area of red blood cell NMR spectroscopy. More recently, diffusion coefficient measurement has been used to study the binding between diphosphoglycerate and haemoglobin inside intact red blood cells.

NMR spectroscopy of human and animal urine Sample preparation and NMR assignments

The composition and physical chemistry of urine is highly variable both between species and within species according to lifestyle. A wide range of organic acids and bases, simple sugars and polysaccharides, heterocycles, polyols, low molecular weight proteins and polypeptides are present together with inorganic species. Many of these moieties also interact, forming complexes some of which are amenable to NMR study, that may undergo chemical exchange reactions on a variety of different time scales. The ionic strength of urine varies considerably and may be high enough to adversely affect the tuning and matching of the RF circuits of a spectrometer probe, particularly at high field strengths. The presence of high concentrations of protein in the urine, e.g. due

to renal glomerular or tubular damage, can result in the broadening of resonances from low MW compounds that may bind to these urinary proteins. Urinary pH values may vary from 5 to 8, according to the physiological condition in the individual, but usually lie between 6.5 and 7.5. Urine samples should, therefore, be frozen as soon as possible after collection if NMR measurements cannot be made immediately. When experiments involve collections from laboratory animals housed in metabolic cages, urine samples should be collected into receptacles that are either cooled with dry ice or have a small amount of sodium azide present as a bacteriocide. However, both these procedures may inhibit or destroy urinary enzymes that may frequently be assayed by conventional biochemical methods for assessment of kidney tubular integrity in toxicological experiments. Urinary pH values are unstable because of the progressive and variable precipitation of calcium phosphates which may be present in the urine close to their solubility limits. One solution to this problem is the addition of 100200 mM phosphate buffer in the D2O added for the lock signal followed by centrifugation to remove precipitated salts. This has the effect of normalizing the pH to a range of 6.77.6 which is stable for many hours during which NMR measurements can be made. Few metabolites (except for histidine and citrate) show major chemical shift variations over this pH range. The vast majority of urinary metabolites have 1H T1 relaxation times of 1 to 4 s. Thus, the general rule of applying a 5 × T1 relaxation delay between successive 90° transients to obtain > 99% relaxation and hence quantitative accuracy cannot be applied routinely. Instead, by using 3045° pulse angles and leaving a total T1 relaxation delay of 5 s between successive pulses, spectra with good signal-to-noise ratios can usually be obtained for most metabolites in 510 minutes at high field with a generally low level of quantitative signal distortion. It has become standard practice to replace simple solvent resonance presaturation with pulse methods, which either induce such saturation or which leave the solvent water resonance unexcited, and one popular method involves using simply the first increment of the 2dimensional NOESY pulse sequence. A vast number of metabolites may appear in urine samples and problems due to signal overlap can occur in single pulse experiments. The magnitude of the signal assignment problem in urine is also apparent at ultrahigh field. There are probably >5000 resolved lines in single pulse 750 or 800 MHz 1H NMR spectra of normal human urine (Figure 1), but there is still extensive peak overlap in certain chemical shift ranges and even at

110 BIOFLUIDS STUDIED BY NMR

800 MHz many signals remain to be assigned. The dispersion gain at 800 MHz even over 600 MHz spectra is particularly apparent in 2- (or 3-) dimensional experiments that can be used to aid signal assignment and simplify overlapped spectra. Many of the endogenous compounds that have been detected in NMR spectra of human and animal urines are shown in Table 2. The biochemical composition of urine varies considerably from species to species and also with age as almost all species have age-related changes in renal function. Rats and other rodents have much higher levels of taurine, citrate, succinate, 2-oxoglutarate and allantoin than humans. Rat urine (and that of other rodents) is generally much more concentrated than human urine, and so NMR signal-to-noise ratios may be better for many metabolites. All animals have physiological processes that are modulated by biological rhythms. These include excretory processes and the urinary composition of an animal may vary considerably according to the time it is collected. Given these types of variation it is obviously of paramount importance to have closely matched (and in many cases timed) control samples where toxicological or disease processes are being studied. Dietary composition also affects the urinary metabolite profiles of man and animals, and it is important to distinguish these from disease-related processes in clinical or toxicological studies. For example, persons consuming large quantities of meat/poultry before urine collection may have NMR detectable levels of carnosine and anserine in their urine, consumption of cherries is associated with elevated urinary fructose and consumption of shellfish and fish are associated with high levels of betaine and trimethylamine in the urine. The 500 MHz and 600 MHz 1H NMR spectra of neonate urines are characterized by strong signals from amino acids, organic acids, amines, sugars and polyols. In particular, high concentrations of myo-inositol have been detected in both pre- and full-term urines. This is an important intracellular organic osmolyte in renal cells, present in high concentrations in the renal inner medulla. NMR spectra of urine in disease

Given the serious outcome of many inborn errors of metabolism, if not diagnosed and treated at an early stage, there has been a search for a rapid, sensitive and general method for the detection and diagnosis of inborn errors of metabolism in neonates. Conventional methods including specific enzyme assays and gas chromatography/mass spectrometry are sensitive but time consuming, involve considerable sample preparation and are not general. However, NMR

spectroscopy of biofluids has been shown to be a very powerful and general method for the detection of inborn errors of metabolism and many disorders have been studied by the use of biofluid NMR. A good example is provided by the spectra from individuals with a deficiency in methylacetoacetyl CoA thiolase (MACT). The conversion of 2-methylacetoacetyl CoA into acetyl-CoA and propionyl-CoA is inhibited and an accumulation of abnormal catabolic products is observed in the urine. Both 1D and 2D 1H NMR spectroscopy have been used to investigate the urinary metabolites of patients with this disorder. The urine spectra from the patients clearly showed the presence of both 2-methyl-3-hydroxybutyrate and tiglylglycine, which is characteristic for MACT deficiency due to the build up of metabolites close to the position of enzyme deficiency. A similar success was observed for studies of branched chain ketoaciduria in which the second stage of the catabolism of leucine, valine and isoleucine involves an oxidative decarboxylation. In patients with branched chain ketoaciduria, this step is blocked for all three of these amino acids. The urine of these patients takes on the odour of maple syrup and hence this condition is also known as maple syrup urine disease. 1H NMR spectroscopy was used to study the urines of patients with such branched chain ketoaciduria. The spectra showed several abnormal metabolites including the amino acids leucine, isoleucine and valine and their corresponding transamination products. It was noted that 2-hydroxyisovalerate levels were very high in the urines of all the patients studied and that, as in other inborn errors of metabolism, the levels of urinary glycine were elevated. High field 1H NMR spectroscopy has shown itself to be a very powerful tool with which to diagnose inborn errors of metabolism and to monitor the clinical response of patients to therapeutic interventions. Characteristic patterns of abnormal metabolites are observed in the patients urine for each disease. The main advantages of using 1H NMR spectroscopy in this area are its speed, lack of sample preparation and the provision of non-selective detection for all the abnormal metabolites in the biofluid regardless of their structural type, providing only that they are present above the detection limit of the NMR experiment. There are a number of studies of other disease processes using NMR spectroscopy of urine. These include glomerularnephritis, kidney transplant patients over a period of 14 days post-operation to investigate the biochemical effects of rejection, phenol poisoning and metabolic acidoses caused by alcohol ingestion.

BIOFLUIDS STUDIED BY NMR 111

Evaluation of toxic effects of xenobiotics using NMR spectroscopy of urine

The successful application of 1H NMR spectroscopy of biofluids to study a variety of metabolic diseases and toxic processes has now been well established and many novel metabolic markers of organ-specific toxicity have been discovered. The method is based on the fact that the biochemical composition of a biofluid is altered when organ damage occurs. This is particularly true for NMR spectra of urine in situations where damage has occurred to the kidney or liver. It has been shown that specific and identifiable changes can be observed which distinguish the organ that is the site of a toxic lesion. Also it is possible to focus in on particular parts of an organ such as the cortex of the kidney and even in favourable cases to very localized parts of the cortex. Finally it is possible to deduce the biochemical mechanism of the xenobiotic toxicity, based on a biochemical interpretation of the changes in the urine. A wide range of toxins has now been investigated including the kidney cortical toxins mercury chloride, p-aminophenol, ifosfamide, the kidney medullary toxins propylene imine and 2-bromoethanamine hydrochloride and the liver toxins hydrazine, allyl alcohol, thioacetamide and carbon tetrachloride. The testicular toxin cadmium chloride has been investigated in detail and the aldose reductase inhibitor HOE-843 has also been studied. Use of combined NMR spectroscopypattern recognition (PR) to evaluate biochemical changes in urine

The first studies using PR to classify biofluid samples used a simple scoring system to describe the levels of 18 endogenous metabolites measured by 1H NMR spectroscopy of urine from rats which were either in a control group or had received a specific organ toxin which affected the liver, the testes, the renal cortex or the renal medulla. The data were used to construct non-linear maps and various types of principal components (PC) scores plots in 2 or 3 dimensions. The samples were divided into two subsets, a training set and a test set. This study showed that samples corresponding to different organ toxins mapped into distinctly different regions and that in particular the renal cortical and renal medullary toxins showed good separation (Figure 2). In addition, the NMR-PR approach has been used to investigate the time course of metabolic urinary changes induced by two renal toxins. In this case, toxic lesions were induced in rats by a single acute dose of the renal cortical toxin mercury(II) chloride and the medullary toxin 2-bromoethanamine. The

Figure 2 A plot of PC1 versus PC2 for a set of rat urine samples using simple scored levels of 16 metabolites as the descriptor set. A cluster of renal cortical toxins is seen at the left, whilst the testicular and liver toxins are to the right. At the bottom the two renal papillary toxins (BEA and PI) appear to cluster with liver toxins (ALY and ANIT) but are well separated in the third principal component. Reproduced with permission of Academic Press from Lindon JC, Nicholson JK and Everett JR (1999) NMR spectroscopy of biofluids. Annual Reports on NMR Spectroscopy 38: 1–88.

onset, progression and recovery of the lesions were also followed using histopathology to provide a definitive classification of the toxic state relating to each urine sample. The concentrations of 20 endogenous urinary metabolites were measured at eight time points after dosing. A number of ways of presenting the data were investigated and it was possible to construct PC scores plots which showed metabolic trajectories which were quite distinct for the two toxins. These showed that the points on the plot can be related to the development of, and recovery from, the lesions. These trajectories allowed the time points at which there were maximal metabolic differences to be determined and provided visualization of the treated groups of animals. A toxicological assessment approach based on neural network software has been tested by analysing the toxin-induced changes in endogenous biochemicals in urine as measured using 1H NMR spectroscopy to ascertain whether the methods provide a robust approach which could lead to automatic toxin classification. The neural network approach to sample classification was in general predictive of the sample class and once the network is trained, the prediction of new samples is rapid and automatic. The principal disadvantage, common to most neural network studies, in that it is difficult to ascertain from the network which of the original sample descriptors are responsible for the classification.

112 BIOFLUIDS STUDIED BY NMR

Recently, studies have been published using pattern recognition to predict and classify drug toxicity effects including lesions in the liver and kidney and using supervized methods as an approach to an expert system. In this case the NMR spectra were segmented into integrated regions automatically, thereby providing standard intensity descriptors for input to the PR. The usefulness of the NMR-PR approach to the classification of urine samples from patients with inborn errors of metabolism has also been attempted. Urine samples from adult patients with inborn errors of metabolism comprising cystinuria, oxalic aciduria, Fanconi syndrome, porphyria and 5oxoprolinuria were studied. PCA mapping produced a reasonable separation of control urines from the clusters derived from the different types of inborn error. The rapid and unambiguous distinction of the various clinical situations that can occur following a kidney transplant remains a problem. For example, it is necessary to distinguish rejection, which requires increased immunosuppressive therapy such as cyclosporin A (CyA), from tubular necrosis, which can be caused by too high levels of CyA. At present, kidney function is generally characterized using the level of blood plasma creatinine as a monitor. However the balance between rejection and tubular necrosis remains a difficult distinction and the only clear diagnosis is an invasive kidney biopsy. A study of patients after renal transplantation has shown that NMR can reveal new markers of abnormality and in particular that high levels of trimethylamine-N-oxide (TMAO) in the urine are correlated with rejection episodes. However, the use of TMAO as a single diagnostic marker of rejection is probably not sufficiently discriminatory for clinical use because of interpatient variability. It is likely that TMAO is only one of a number of potential markers of rejection and that a combination of other metabolites present in urine will provide a new differential diagnosis of rejection and CyA toxicity. A combination of high field 1H NMR urinalysis, automatic data reduction and PR methods has been applied to investigate this. NMR-PR was seen to separate the samples into three distinct groups with the greatest separation being between the CyA toxicity group and the rejection group. It was also notable that, in general, the greater the degree of allograft dysfunction (as assessed by conventional means) the further the samples plotted away from the control group. Identification of the compounds that gave rise to the significant descriptors has not been conclusively proven. However, close examination of spectra from samples from the various classes using both 600 MHz and

750 MHz 1H NMR spectroscopy suggested that these were 3-hydroxyisovaleric acid, N-acetylated glycoproteins, TMAO, hippuric acid and a molecule related to N-methylnicotinamide.

NMR spectroscopy of cerebrospinal fluid (CSF) Because of the low protein content of normal human CSF, it is possible to use single pulse 1H NMR experiments to obtain biochemical information without recourse to the spinecho techniques often required for studies on blood and plasma. However, where serious cerebral damage has occurred, or in the presence of an acute infection, spectra may become dominated by broad resonances from proteins. Problems may also arise from protein binding and metal binding of some metabolites, with a consequential broadening of their proton resonances. There are a number of 1H NMR studies of CSF in the literature. Also, it has been shown that high quality 2D 1H NMR spectra can be obtained from human CSF, and that changes in NMR patterns can be roughly related to disease states in the donor. Examination of a number of ex vivo control samples showed high consistency in the aliphatic region of the 1D 1H NMR spectra of CSF and many assignments could be made by inspection. Diseases studied using NMR spectra of CSF include lumbar disk herniations, cerebral tumours, drug overdose, diabetes, hepatic encephalitis, multiple sclerosis, AIDS dementia complex, Parkinsons disease, CreutzfeldJakob disease, GuillainBarre syndrome and vitamin B12 deficiency. One study has reported 500 and 600 MHz 1H NMR data on the post mortem CSF from Alzheimers disease (AD) patients and controls. The main differences between the spectra of the two groups were found to be in the region G 2.42.9, and principal components analysis showed that separation of the two groups was possible based mainly on lower citrate levels in the AD patients. Non-matching in patient age and the time interval between death and autopsy caused a reduction in the inter-group differences but they were still significant (p < 0.05).

NMR spectroscopy of seminal fluids The 750 MHz 1H NMR spectra of seminal fluid obtained from a healthy individual is complex but many of the signals have been assigned (see Table 2). Fresh undiluted seminal fluid gives rise to NMR spectra with very broad and poorly resolved signals due to the presence of high concentrations of peptides (which are cleaved to amino acids by

BIOFLUIDS STUDIED BY NMR 113

endogenous peptidase activity) and the high viscosity of the matrix. The complexity of the biochemical composition of seminal fluids together with their reactivity poses a number of assignment and quantitation problems. Some metabolite signals in seminal fluids also appear to have anomalous chemical shifts when compared to simple standard solutions measured at the same pH or even when compared to other body fluids. The application of 2D NMR methods has been shown be very useful in the case of seminal fluids. An investigation of dynamic molecular processes that occur in seminal fluid has been undertaken using 1H NMR spectroscopy. Reactions that could be followed included hydrolysis of phosphorylcholine and nucleotides and zinc complexation. In view of the importance of artificial insemination in farming, it is somewhat surprising that so few studies of animal seminal fluid have been reported. However there are studies on boar seminal plasma giving details of resonance assignments. The comparison of 1H NMR spectra of seminal fluid from normal controls with those from patients with vasal aplasia (obstruction of the vas deferens leading to blockage of the seminal vesicles) and those with non-obstructive infertility has been reported. The 1H NMR spectra of the seminal fluid from patients with non-obstructive infertility were similar to those of normal subjects. However, the 1H NMR spectra of the seminal fluid from patients with vasal aplasia were grossly different from those of normal subjects and corresponded closely to those of prostatic secretions from normals, owing to the lack of seminal vesicle secretion into the fluid. In the 1H NMR spectra of the vasal aplasia patients, signals from amino acids were either absent or present at very low levels. Similarly, choline is at a low level or absent in the seminal fluid from vasal aplasia patients, as it derives (indirectly) from the seminal vesicle component. Significant differences were observed between the normal and vasal aplasia patient groups for the molar ratios of citratecholine and sperminecholine. Other studies of infertility include a procedure providing automatic diagnosis based on NMR spectroscopy and work on azoospermic subjects and prostate cancer. In addition 31P NMR spectroscopy has also been used to distinguish semen from healthy and infertile men.

are present in mixed micelles with phospholipids and cholesterol (Figure 1). They are broad as a result of short spinspin (T2) relaxation times reflecting constrained molecular motions within micellar particles. On lyophilization and reconstitution with water, the molecular mobility of a number of biliary metabolites changes significantly because of disruption of the micellar compartments. In particular the T2 relaxation times of the aliphatic side chains of lipid moieties are increased in lyophilized bile suggesting greater mobility of these molecules. Increases in signal intensities that occur on lyophilization reflect changes in compartmentation of molecules, which is related to the disruption/reorganization of the biliary micellar compartments. Signals from E-hydroxybutyrate, valine and other branched chain amino acids do not contribute significantly to the 1H NMR spectra of non-lyophilized bile, but resonances from these components are clearly resolved after lyophilization. The sharper signals in bile give rise to well resolved 2D COSY spectra that allow a comprehensive assignment of the bile salt signals. Variable temperature 1H NMR studies of human bile show that considerable dynamic structural information is available, particularly at high fields. The micellar cholesteryl esters that are abundant in bile appear to show liquid crystal behaviour, and it is possible to use NMR measurements to map the phase diagram for the complex biliary matrix. A number of studies have used both 1H and 13C NMR spectroscopy of bile to aid characterization of its composition and structure. Thus, 13C spectra of bile from fish exposed to petroleum have been studied. 31P NMR spectra of human bile have also been investigated. 1H NMR spectroscopy of bile has been used to investigate the micellar cholesterol content and lipids, and both 1H and 31P NMR have been used to study the distribution of lecithin and cholesterol. The use of 1H NMR spectroscopy of bile as a means of monitoring liver function has been proposed. The 31P NMR spectra of the bile from patients with primary biliary cirrhosis of the liver and from clinically healthy men have been compared. Also the level of lactate in bile from patients with hepatobiliary diseases including cancer has been investigated using 1H NMR spectroscopy.

NMR spectroscopy of bile

NMR spectroscopy of miscellaneous body fluids

NMR spectroscopy of bile and dynamic interactions of metabolites

The single pulse NMR spectra of bile are dominated by broad resonances that arise from bile acids that

Amniotic and follicular fluids

The first study of human amniotic fluid using 1H NMR spectroscopy detected 18 small molecule

114 BIOFLUIDS STUDIED BY NMR

metabolites including glucose, leucine, isoleucine, lactate and creatinine. Following this, other studies used a combination of 1D and 2D COSY spectroscopy to assign resonances, assess NMR methods of quantitation and investigate the effects of freezing and thawing. In addition NMR results have been correlated with other clinical chemical analyses. A total of 70 samples were measured using 1H NMR at 600 MHz at different stages of gestation and with different clinical complications and significant correlations between the NMR spectral changes and maternal pre-eclampsia and foetal open spina bifida were observed. The effects of various pathological conditions in pregnancy have been investigated using 1H NMR spectroscopy of human amniotic fluid. Several studies of amniotic fluid using 31P NMR spectroscopy have also been carried out, principally to analyse the phospholipid content. One study of the metabolic profiling of ovarian follicular fluids from sheep, pigs and cows has been reported. Milk

Surprisingly, very little has been published on the NMR spectroscopy of milk given that it is both a biofluid and a food substance. The studies generally focus on milk as a food and Belton has also reviewed the information content of NMR spectra of milk. Very recently, 19F NMR spectroscopy has been used to detect trifluoroacetic acid in milk. Synovial fluid 1H

NMR has been used to measure the levels of a variety of endogenous components in the synovial fluid (SF) aspirated from the knees of patients with osteoarthritis (OA), rheumatoid arthritis (RA) and traumatic effusions (TE). The spinecho NMR spectrum of synovial fluid shows signals from a large number of endogenous components. Many potential markers of inflammation could not be monitored because of their low concentrations or because of their slow tumbling (e.g. hyaluronic acid, a linear polysaccharide that imparts a high viscosity to synovial fluid). The low molecular weight endogenous components showed a wide patient-to-patient variability and showed no statistically significant correlation with disease state. However, correlations were reported between the disease states and the synovial fluid levels of the N-acetyl signals from acute phase glycoproteins. Correlations between the disease state and the levels and type of triglyceride in the synovial fluid have also been reported. The 1H NMR spectra of the SF of a female patient with seronegative erosive RA and of another

female patient with sarcoidosis and independent inflammatory OA were followed over the course of several months and standard clinical tests were performed on paired blood serum samples taken at the same time. It was found that the SF levels of triglyceride CH3, CH2 and CH, glycoprotein N-acetyl signals and creatinine all correlated well with one another, and with standard clinical measures of inflammation. The correlation of disease state with creatinine level is of particular interest, and the altered triglyceride composition and concentration in OA was suggested as a potential marker for the disease in SF. Spin-echo 1H NMR spectroscopy has been used to detect the production of formate and a low molecular weight, N-acetyl-containing oligosaccharide, derived from the oxygen radical-mediated depolymerization of hyaluronate, in the SF of patients with RA, during exercise of the inflamed joint. Gamma radiolysis of rheumatoid SF and of aqueous hyaluronate solutions was also shown to produce formate and the oligosaccharide species. It has been proposed that the hyaluronate-derived oligosaccharide and formate could be novel markers of reactive oxygen radical injury during hypoxic reperfusion injury in the inflamed rheumatoid joint. 13C NMR can be used to monitor the synovial fluids from patients with arthritis. In contrast to 1H NMR studies, signals are seen from hyaluronic acid, the main determinant of the viscoelasticity of the synovial fluid, even though the molecular weight is in the region 500 to 1600 kD. 13C NMR spectra of synovial fluids from patients with RA, OA, TE and cadaver controls have been compared with one another and with spectra of authentic hyaluronic acid, both before and after the incubation of the latter with hyaluronidase, an enzyme which depolymerizes the hyaluronic acid. Depolymerization of the hyaluronic acid was accompanied by a decrease in the half-band widths of its 13C resonances. The synovial fluid NMR spectra from the patients with rheumatoid arthritis had sharper signals for the C-1 and C-1 ′ carbons of hyaluronic acid than those from the osteoarthritic patients, which in turn exhibited sharper signals than those from the cadavers or the joint trauma patients. Thus the degree of polymerization of hyaluronic acid was deduced to decrease in the order: controls/joint trauma patients > osteoarthritic patients > rheumatoid arthritis patients. Since it is known that the consequence of hyaluronate depolymerization may be articular cartilage damage, it was concluded that 13C NMR spectroscopy may be a valuable method for studying these clinical relevant biophysical changes in synovial fluid.

BIOFLUIDS STUDIED BY NMR 115

Aqueous humour and vitreous humour

The first study by NMR spectroscopy on aqueous humour was on 9 samples taken during surgery for other conditions, and NMR spectra were measured at 400 MHz. A number of metabolites were detected, including acetate, acetoacetate, alanine, ascorbate, citrate, creatine, formate, glucose, glutamine or glutamate, E-hydroxybutyrate, lactate, threonine and valine. Following this, there have been a number of other studies. These include 1H NMR spectra from aqueous humour of rabbits and cod fish, 31P NMR spectra of aqueous and vitreous humour from pigs and 23Na NMR spectra of vitreous humour. Finally, the penetration of dexamethasone phosphate into the aqueous humour has been followed using 1H and 19F NMR spectroscopy. Saliva

Only limited studies using NMR spectroscopy of saliva have been reported. 1H NMR spectroscopy of human saliva was used in a forensic study and following this another group reported that only parotid gland saliva gave a well resolved 1H NMR spectrum showing significant circadian effects. No age- or sexrelated differences were observed for saliva from healthy subjects but marked differences were observed in cases of sialodentitis. Finally, the biochemical effects of an oral mouthwash preparation have been studied using 1H NMR spectroscopy. Digestive fluids

The analysis of pancreatic juice and small bowel secretions using 1H NMR spectroscopy has been reported. Pathological cyst fluid

A 1H NMR study of the fluid from the cysts of 6 patients with autosomal dominant polycystic kidney disease (ADPKD) has been reported. ADPKD in adults is characterized by the slow progressive growth of cysts in the kidney, and when these cysts reach a large size they can significantly distort the kidney and disrupt both the blood supply and renal function. ADPKD is one of the commonest causes for renal transplantation in adults. The 1H NMR spectra revealed a number of unusual features and showed the cyst fluids to be distinct from both blood plasma and urine. Unusually, the fluids from all patients contained high levels of ethanol, which was not related to consumption of alcoholic beverages or drug preparations. In general there was little variation in the composition of the cyst fluids as revealed by 1H NMR, although the protein signal intensity

did vary somewhat. It was hypothesized that this constancy of composition reflected the chronic nature of the accumulation of the cyst fluid and a long turnover time of the cyst components, which thus has the effect of averaging the compositions. The unique biochemical composition of the cyst fluids was ascribed to abnormal transport processes occurring across the cyst epithelial wall.

Drug metabolites in biofluids Several studies have used NMR spectroscopy to determine the number and identity of drug metabolites in bile. These include the use of 19F NMR spectroscopy to study doxifluridine catabolites in human bile, 19F and 13C NMR spectroscopy of perfluorinated fatty acids in rat bile and 13C NMR for monitoring the formation of formaldehyde from demethylation of antipyrine. 1H NMR spectroscopy of rat bile has been used to monitor the excretion of paracetamol metabolites. A combination of 1H and 1H13C 2D NMR methods has allowed identification of 4cyano-N,N-dimethylaniline, cefoperazone and benzyl chloride in rat bile. NMR spectroscopy of urine combined with labelling with stable isotopes such as 13C and 2H has been used to monitor the silent process of deaceteylation and subsequent reacetylation (futile deacteylation) in the rat as this has implications for the toxicity of paracetamol and phenacetin. The metabolites of 5fluorouracil in plasma and urine have been analysed using 19F NMR spectroscopy in patients receiving chemotherapy. 2H NMR has been employed to study the pharmacokinetics of benzoic acid in relation to liver function. The metabolism of hydrazine in rats has been probed using 15N NMR spectroscopy of urine. Often it is advantageous to carry out a partial extraction of metabolites using a solid phase separation cartridge. In this case, the urine containing the drug metabolites is loaded on to a C18 cartridge that is then washed with acidified water to remove very polar endogenous components. The metabolites of interest can be then be selectively washed off using methanol or water/methanol mixtures. The direct coupling of HPLC with NMR spectroscopy has required a number of technological developments to make it a feasible routine technique and now on-line NMR detection of HPLC fractions is a useful adjunct to the armoury of analytical methods. To date, HPLC-NMR spectroscopy has been applied to the profiling and identification of the metabolites of a number of drugs and xenobiotics present in biofluids such as plasma and urine and in bile samples from rats and humans. In general the

116 BIOFLUIDS STUDIED BY NMR

simple stop-flow approach has predominated in these studies. The HPLC-NMR method is most useful for compounds, and their metabolites, with suitable NMR reporter groups. This includes the use of 19F NMR spectroscopy for molecules containing fluorine and diagnostic 1H NMR resonances in regions of the NMR spectrum where solvent signals do not cause interference. For example, the pattern of NMR resonances from glucuronide conjugates is particularly diagnostic. HPLC-NMR of biofluids has been employed to identify the unusual endogenous metabolites found in rat urine after administration of compounds that induce liver enzymes leading to elevated urinary levels of a number of carbohydrates and related molecules. A wide range of 1H and 19F HPLC-NMR studies has been carried out on human and animal biofluids, predominantly urine and bile, to characterize the metabolites of exploratory drugs and marketed pharmaceuticals. In some cases, the biofluid has been subjected to solid phase extraction to remove many of the very polar materials, which facilitates the HPLC separation by avoiding overloading the column. These studies include the identification and structuring of metabolites of antipyrine, ibuprofen, flurbiprofen, paracetamol, the anti-HIV compound GW524W91, iloperidone and tolfenamic acid. Other studies using HPLC-NMR on bile include the identification of a metabolite of 7-(4-chlorobenzyl)7,8,13,13a-tetrahydroberberine chloride in rat bile and metabolites of a drug LY335979, under development to prevent multidrug resistance to cancer chemotherapy. In addition, 31P HPLC-NMR has been used to study metabolites of ifosfamide, and 2H HPLC-NMR has been employed to detect metabolites of dimethylformamide-d7. The further coupling of HPLC-NMR with mass spectroscopy has also been used to analyse drug metabolites in urine for the human excretion of paracetamol metabolites. This approach has also been used to identify metabolites of substituted anilines in rat urine. Capillary electrophoresis (CE), and related techniques, is a relatively new technique which uses a length of fused silica capillary with an optical window to enable detection, a detector (UV, fluorescence or mass spectrometry), a high voltage source, two electrode assemblies and buffer solutions in suitable reservoirs. The technique has been shown to provide very high separation efficiencies, with hundreds of thousands of theoretical plates achievable.

However, the small injection volume available to CE (a few nl) means that high sensitivity can only be achieved if concentrations of the analyte in the sample are high. Nevertheless, some results have been reported using CE-NMR and the method has been applied to the identification of paracetamol metabolites in human urine. Finally the application of capillary electrochromatography (CEC) directly coupled to NMR has been explored using the same paracetamol metabolite samples. See also: Cells Studied By NMR; Chromatography– NMR, Applications; Diffusion Studied Using NMR Spectroscopy; Drug Metabolism Studied Using NMR Spectroscopy; In vivo NMR, Applications, Other Nuclei; In vivo NMR, Applications, 31P; In vivo NMR, Methods; Perfused Organs Studied Using NMR Spectroscopy; Proteins Studied Using NMR Spectroscopy; Solvent Suppression Methods in NMR Spectroscopy; Structural Chemistry Using NMR Spectroscopy, Pharmaceuticals.

Further reading Ala-Korpela M (1997) 1H NMR spectroscopy of human blood plasma. Progress in NMR Spectroscopy 27: 475554. Gartland KPR, Beddell CR, Lindon JC and Nicholson JK (1991) The application of pattern recognition methods to the analysis and classification of toxicological data derived from proton NMR spectroscopy of urine. Molecular Pharmacology 39: 629642. Lindon JC, Nicholson JK and Everett JR (1999) NMR spectroscopy of biofluids. In Webb GA (ed) Annual Reports on NMR Spectroscopy. Vol 38, in press. London, Academic Press. Lindon JC, Nicholson JK, Sidelman UG and Wilson ID (1997) Directly-coupled HPLCNMR and its application to drug metabolism. Drug Metabolism Reviews 29: 705746. Lindon JC, Nicholson JK and Wilson ID (1996) Direct coupling of chromatographic separations to NMR spectroscopy. Progress in NMR Spectroscopy 29: 149. Nicholson JK and Wilson ID (1989) High resolution proton magnetic resonance spectroscopy of biofluids. Progress in NMR Spectroscopy 21: 449501. Nicholson JK, Foxall PJD, Spraul M, Farrant RD and Lindon JC (1995) 750 MHz 1H and 1H13C NMR spectroscopy of human blood plasma. Analytical Chemistry 67: 793811. Rabenstein DL (1984) 1H NMR methods for the non-invasive study of metabolism and other processes involving small molecules in intact erythrocytes. Journal of Biochemical and Biophysical Methods 9: 277306.

BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD 117

Most biological macromolecules, including proteins, nucleic acids and carbohydrates, are built of repeating units that are assembled into highly asymmetric structures. The specific conformations of the macromolecules cause the optical transitions of the chromophores in their repeating units (e.g. peptides, nucleotides, glycosides) to align with each other so that their electronic and magnetic transitions interact. The interactions may result in hypo or hyperchromism and in splitting of the transitions into multiple transitions. In addition the aligned transitions may exhibit a high degree of circular dichroism (CD) and optical rotatory dispersion (ORD). Because the ORD and CD spectra of macromolecules depend on their conformation, they are excellent probes for following changes in structure as a function of temperature, denaturants or ligand binding. In addition, non-optically active chromophores sometimes bind to macromolecules in highly asymmetric fashions, and generate large extrinsic CD and ORD bands in the region of absorption. These extrinsic bands can be used to follow interactions of the macromolecules with the chromophores, or to probe structural transitions of the macromolecules. The following article discusses how ORD and CD are used to study the conformation, folding and interactions of proteins, nucleic acids and carbohydrates. Specific examples are given to illustrate each application.

ELECTRONIC SPECTROSCOPY Applications asymmetric environments, leading to characteristic CD spectra in the near-UV, which may serve as useful probes of the tertiary structure of proteins. Analysis of the secondary structure of proteins

Many methods have been developed to extract the conformation of proteins in solution from CD and ORD data. In early studies, the ORD spectra of proteins were analysed to yield structural data, but when commercial CD spectrometers became available in the late 1960s, CD became the method of choice because CD spectra have better resolution

Proteins and polypeptides Proteins and polypeptides have two major classes of chromophores, the amide groups of the peptide backbone, which absorb light in the far UV (below 250 nm) and the aromatic amino acid side chains and disulfide bonds, which absorb light in both the near (320250 nm) and far-UV. Far-UV ORD and CD are useful for studying protein structure and folding because many conformations that are common in proteins, including D-helixes, E-pleated sheets, poly-L-proline II-like helices and turns, have characteristic spectra. Figure 1 illustrates representative CD spectra of model polypeptides with different secondary structures. In addition, the chromophores of the aromatic amino acids of proteins are often in

Figure 1 The circular dichroism of polypeptides with different conformations. (—) D-helix, poly-L-lysine in water, pH 11.1, 22°C; (---) E-sheet, poly(Lys-Leu-Lys-Leu) in 0.5 M NaF at pH 7; (·····) E-turn (Type III), N-acetyl-Pro-Gly-Leu-OH in trifluoroethanol at –60°C; ( ) random coil, Glu-Lys-Lys-Leu-Glu-Glu-Ala in 20 mM sodium phosphate, pH 2, 0°C; ( ) PE, poly-L-proline in trifluoroethanol. Reprinted from data in Greenfield NJ and Fasman GD (1969) Biochemistry 8: 4108–4116, Brahms S and Brahms J (1980) J. Mol. Biol. 138: 149–178, by permission of Academic Press, Venyaminov SY, Baikalov IA, Shen ZM, Wu C-SC and Yang JT (1993) Anal. Biochem. 214: 17–24, and Bovey FA and Hood FP (1967) Biopolymers 5: 325–326 by permission of John Wiley & Sons, Inc.

118 BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD

than ORD spectra. Basically, all of the analysis methods assume that the spectrum of a protein can be represented by a linear combination of the spectra of the secondary structural elements and a noise term. In constrained fits the sum of all the fractional weights of each structure must be equal to one. In the first attempts to elucidate protein conformation from CD spectra, the spectra of proteins were fitted by a combination of the spectra of polypeptides with known conformations using the method of least squares (multilinear regression). Once the conformation of many proteins had been determined from X-ray crystallographic analysis, the CD spectra of several well-characterized proteins were deconvoluted into reference (basis) spectra for the D-helix, antiparallel and parallel E-pleated sheets, E-turn and random conformations using multilinear-regression analysis. These extracted curves were then used in place of the polypeptide standard curves. Later, singular-value decomposition (SVD) and convex-constraint analysis (CCA) were used to extract the basis curves. The basis curves obtained from deconvoluting a set of protein CD spectra may vary greatly depending on the choice of reference proteins. This occurs because some proteins have unusual CD spectra in the far UV, due to aromatic amino acids, disulfide bridges, or rare conformations. To overcome these difficulties, some methods use selection procedures so that only proteins with spectral characteristics that are similar to those of the protein to be evaluated are used as standards. Methods that use selected reference data include ridge-regression, variable-selection and neural-network procedures. The following computer programs for analysing CD data to yield the secondary structures of proteins and polypeptides are widely available: MLR is a non-constrained and G&F, LINCOMB and ESTIMATE are constrained least-squares-analysis programs; CONTIN uses the ridge-regression technique; VARSLC and SELCON use singular-value decomposition combined with variable selection to compare the structure of unknown proteins with standard; CCA uses the convex-constraint algorithm to deconvolute sets of CD spectra into a minimum number of basis curves and K2D is a neural-network recall program. Table 1 compares the agreement between the secondary structures of 16 proteins and poly-L-glutamate, determined by analysis of the CD spectra using these computer programs, with the structures determined by X-ray crystallography. Determination of the thermodynamics of protein folding

The change in the CD spectra as a function of temperature can be used to determine the thermodynamics

of unfolding or folding of a protein because the observed change in ellipticity, Tobs, is directly proportional to the change in concentration of native and denatured forms. When the protein is fully folded Tobs = TF and when it is fully unfolded Tobs = TU. For a monomeric protein, the equilibrium constant of folding, K = folded/unfolded. If we define D as the fraction folded at a given temperature, T, then

where 'G is the free energy of folding, R is the gas constant and n is the number of molecules.

where 'H is the Vant Hoff enthalpy of folding, 'S is the entropy of folding, TM is the observed midpoint of the thermal transition and 'Cp is the change in heat capacity for the transition. At TM, K = 1; therefore 'G = 0 and 'S = 'H/TM. Rearranging these equations, we obtain:

To calculate the values of 'H and TM that best describe the folding curve, initial values of 'H, 'Cp, TM, TF and TU are estimated, and Equation [6] is fitted to the experimentally observed values of the change in ellipticity as a function of temperature, by a nonlinear least-squares curve-fitting routine such as the LevenbergMarquardt algorithm. Similar equations can be used to estimate the thermodynamics of folding of proteins and peptides that undergo folded multimer to unfolded monomer transitions. Figure 2 illustrates how changes in ellipticity as a function of temperature and concentration have been used to determine the enthalpy of folding of a peptide derived from the coiled-coil domain of the yeast GCN4 transcription factor, which undergoes

BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD 119

using Equation [2] and is plotted as a function of denaturant. The curve is extrapolated to zero denaturant to obtain the free energy of folding of the native material. Figure 3 depicts the determination of the free energy of folding of a coiled-coil DNA binding protein from a urea-denaturation study. The peptide forms a two-stranded coiled-coil D-helix when folded and is a single-stranded unordered peptide when dissociated. The association constant and free energy of folding, determined from the denaturation data (Figures 3B and 3C), agree with those determined from the change in ellipticity as a function of peptide concentration (Figure 3A).

Figure 2 The change in fraction folded of a coiled-coil leucine zipper peptide, GCN4-P1, as a function of temperature and peptide concentration. The concentrations range from 1 µM with the lowest TM to 20 µM with the highest TM. The enthalpy of unfolding was 35.0 ± 1.1 kcal mol–1 (monomer) which compared to that of 34.7 ± 0.3 kcal mol–1 measured by calorimetry. Abstracted with permission from data in Thompson KS, Vinson CR and Freire E (1993) Biochemistry 32: 5491–5496. Copyright 1993 American Chemical Society.

a two-state transition between a folded two-stranded D-helical coiled coil and a monomeric disordered state. The values for the enthalpy of the folding transition, determined from the changes in ellipticity as a function of temperature, agree with the values obtained from scanning calorimetry. In some cases thermal folding and unfolding studies are impractical. The protein may have a very high TM, or may aggregate and precipitate upon heating. In these cases, denaturants, such as urea or guanidine-HCl, may be added to a protein or peptide to induce unfolding. K, the equilibrium constant of folding, is determined from the ellipticity observed at each concentration of denaturant. It is assumed that the protein is native in the absence of denaturant, and fully unfolded when there is no further change in ellipticity upon addition of higher concentrations of perturbant. At each concentration of denaturant,

D, TF and TU have the same definitions as in thermal unfolding above and n is the number of identical subunits. [C] is the total concentration of protein monomers. The free energy of folding is evaluated at every concentration of denaturant

Analysis of conformational transitions

Some native proteins undergo conformational transitions that do not result in formation of a denatured state. Such transitions may involve changes from an D-helical to a E-pleated sheet conformation (or vice versa). Conformational changes of proteins are sometimes followed by oligomerization and/or aggregation and precipitation reactions that may have important clinical consequences. For example, transitions from a random or D-helical form to a Esheet may be involved in the pathology of diseases such as scrapie, mad cow disease and Alzheimers disease. Figure 4 illustrates how CD has been used to follow the thermal stability and conformational transitions of the scrapie amyloid (prion) protein, PrP27-30. The CD spectra of the solvent-exposed (Figure 4A) and rehydrated solid state PrP27-30, obtained under various conditions, were deconvoluted using the CCA algorithm and five common spectral components were identified (Figure 4B). Infectivity quantitatively correlated with an increasing proportion of a native, E-sheet-like secondarystructure component (Figure 4C), a decreasing amount of an D-helical component, and an increasingly ordered tertiary structure. Detection of folding intermediates

Often when a protein or peptide folds or unfolds, the process is not a two-state transition between the native and totally disordered forms; intermediate states exist. These states may be transient, and may be observed during kinetic measurements of folding or unfolding. Alternatively, they may be stable intermediates, seen when the peptide is subject to denaturing conditions, such as high temperatures, or exposure to urea, guanidine or detergents. These states may be partially folded, with well-defined tertiary structures, or may be so called molten

120 BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD

Table 1

Comparisons of methods of analysing protein conformation from circular dichroism dataa

Computer program

Standards

D-helix

E-sheet

E-turn

Wavelength range, nm

Linear regression – unconstrained fit MLR (1)

4 peptides (1)

240–178

0.91

0.13

0.43

0.21

0.07

0.16

MLR (1)

4 peptides (1)

240–200

0.92

0.14

0.74

0.16

0.23

0.16

Linear regression – constrained fit G&F (2)

Poly-L-lysine (2)

240–208

0.92

0.13

0.61

0.18

LINCOMB (3)

4 peptides (1)

240–178

0.93

0.11

0.58

0.15

0.61

0.11

LINCOMB (3)

4 peptides (1)

240–200

0.94

0.11

0.71

0.13

0.53

0.14

LINCOMB (3)

17 proteins (4)

240–178

0.94

0.09

0.62

0.14

0.21

0.13

LINCOMB (3)

17 proteins (4)

240–200

0.92

0.10

0.09

0.28

0.52

0.12

Singular-value decomposition SVD (4)

17 proteins (4)

240–178

0.98

0.05

0.68

0.12

0.22

0.10

SVD (4)

17 proteinsb (4)

240–200

0.96

0.07

0.43

0.14

0.04

0.13

Convex-constraint algorithm CCA (5)

17 proteins (4)

260–178

0.96

0.10

0.62

0.18

0.39

0.18

CCA (5)

17 proteins (4)

240–200

0.97

0.10

0.42

0.20

0.52

0.22

CONTIN (6)

17 proteins (4)

260–178

0.93

0.11

0.56

0.15

0.58

0.08

CONTIN (6)

17 proteins (4)

240–200

0.95

0.13

0.60

0.15

0.74

0.07

17 proteins (4)

260–178

0.97

0.07

0.81

0.10

0.60

0.07

Ridge regression

Variable selection VARSLC (7)

Variable selection – self consistent method SELCON (8)

17 proteins (4)

260–178

0.95

0.09

0.84

0.08

0.77

0.05

SELCON (8)

17 proteins (4)

260–190

0.94

0.09

0.73

0.09

0.84

0.05

SELCON (8)

17 proteins (4)

240–200

0.93

0.10

0.73

0.11

0.71

0.06

SELCON (8)

33 proteins (9)

260–178

0.93

0.09

0.91

0.07

0.53

0.09

SELCON (8)

33 proteins (9)

260–200

0.88

0.12

0.86

0.09

0.46

0.09

Neural-network analysis K2D (10)

18 proteins (10)

240–200

0.95

0.09

0.77

0.10

Mean

STD

Mean

STD

Mean

STD

0.36

0.27

0.20

0.16

0.22

0.08

The secondary structures of 16 proteins plus poly-L-glutamate were analysed using each computer program as described by Sreerama and Woody (8). P is the correlation coefficient between the CD-estimated and X-ray conformations and V is the meansquare error between the CD-estimated and X-ray conformations. The secondary structures were assigned by the method of Kabsch and Sander (11). When protein databases were used as standards, each protein analysed was excluded from the data set used as the references. When the V value is higher than the standard deviation (STD) of the mean value of each conformation found in the 17 samples which were analysed, the program does a relatively poor job of analysing that conformation. PGA was excluded from the calculation of P and V because the fit was obviously impossible. (1) Brahms S and Brahms J (1980) Determination of protein secondary structure in solution by vacuum ultraviolet circular dichroism. Journal of Molecular Biology 138: 149–178. Continued

BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD 121

Table 1

continued

(2) Greenfield N and Fasman GD (1969) Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry 8: 4108–4116. (3) Perczel A, Park K and Fasman GD (1992) Analysis of the circular dichroism spectrum of proteins using the convex constraint algorithm: a practical guide. Analytical Biochemistry 203: 83–93. (4) Hennessey JP and Johnson WC Jr (1981) Information content in the circular dichroism of proteins. Biochemistry 20: 1085–1094. (5) Perczel A, Hollósi M, Tusnady G and Fasman GD (1991) Convex constraint analysis: a natural deconvolution of circular dichroism curves of proteins. Protein Engineering 4: 669–679. (6) Provencher SW and Glöckner J (1981) Estimation of globular protein secondary structure from circular dichroism. Biochemistry 20: 33–37. (7) Manavalan P and Johnson WC Jr (1987) Variable selection method improves the prediction of protein secondary structure from circular dichroism spectra. Analytical Biochemistry 167: 76–85. (8) Sreerama N and Woody RW (1994) Poly(pro)II helices in globular proteins: identification and circular dichroic analysis. Biochemistry 33: 10022–10025. (9) Toumadje A, Alcorn SW and Johnson WC Jr (1992) Extending CD spectra of proteins to 168 nm improves the analysis for secondary structures. Analytical Biochemistry 200: 321–331. (10) Andrade MA, Chacón P, Merelo JJ and Morán F (1993) Evaluation of secondary structure of proteins from UV circular dichroism spectra using an unsupervised learning neural network. Protein Engineering 6: 383–390. (11) Kabsch W and Sander C (1983) Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22: 2577–2637.

Figure 3 The effects of concentration and urea on the folding of the helix-loop-helix DNA binding domains of the muscle-regulatory DNA transcription factors, MyoD. (A) The change in ellipticity at 222 nm of Myo D as a function of concentration. The data were used to calculate the constant of dimerization K dimer = 9.4 µM. (B) The change in fraction folded of MyoD as a function of urea concentration. (C) 'G of folding calculated from the data in (B). The extrapolated free energy of folding calculated from the value at 0 urea was similar to those calculated from the dimerization constant. Redrawn with permission from data in Wendt H, Thomas RM and Ellenberger T (1998) J. Biol. Chem. 273: 5735–5743. Copyright 1998. The American Society for Biochemistry & Molecular Biology.

122 BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD

Figure 4 (A) Solvent-induced conformational transition of PrP27-30 scrapie amyloid (prion) protein in the solid state on quartz glass. Films were exposed for 20 to 30 min to: SDS, sodium dodecyl sulfate, TFE, trifluorethanol; HFIP, hexafluorisopropanol; TFA, trifluoroacetic acid; FA, 99% formic acid. “f” is the spectrum with no treatment at 23°C. (B) Data on PrP27-30 obtained under various conditions, deconvoluted into five basis curves using the convex-constraint algorithm. (—) E-turn type I or III and/or 310-type helix; (·····) alpha helix; (----) E-sheet; (– – –) E-turn type II; ( ) additional chiral contribution. (C) Correlation of the solvent-induced changes in the Esheet component with infectivity. The data for dehydrated films are open circles and rehydrated films are closed circles. The data obtained with TFA and FA, where the proteins were highly helical and the infectivity was very low, are excluded from the fits. (o) dehydrated films (▲) rehydrated films. Redrawn with permission from data in Safar J, Roller PP, Gajdusek DC and Gibbs CJ Jr (1993) Protein Sci. 2: 2206–2216. Copyright 1993 The Protein Society.

globules. Molten globule is a term describing a compact state with native-like secondary structure but slowly fluctuating tertiary structure. For example, methanol induces the formation of a moltenglobule state in the globular protein, cytochrome C. Figure 5 illustrates the effect of increasing quantities of methanol on the CD spectra of cytochrome C in the near and far UV. A loss of ellipticity occurs in the near UV, that is present in the native state arising from the tertiary interactions of the aromatic chromophores (Figure 5A). The far-UV spectra (Figure 5B), however, show that the helical content of cytochrome C actually increases upon the addition of methanol, as shown by the increasing ellipticity at 222 and 208 nm.

Determination of the kinetics of protein folding

Besides the study of protein folding under equilibrium conditions, CD can be used to measure rapid events in protein folding and unfolding by attaching a rapid mixing device to a CD spectropolarimeter. Stopped-flow CD can be used both to obtain the kinetic constants of folding reactions and to define the conformation of folding intermediates. Data can be collected in the far and near UV. The far-UV data give information about the secondary structure of the protein, while the near-UV data are windows that monitor the formation of tertiary structure. Stopped-flow CD studies of highly helical twostranded coiled coils and four-helix bundle proteins

BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD 123

Figure 5 (A) Near-UV CD and (B) Far-UV CD spectra of cytochrome c (pH 4.0 in 0.5 M NaCl, 24°C) at methanol concentrations shown near the curves. The negative bands in the near-UV spectrum are attributed to the side chain of the single Trp59. They decrease and almost vanish at methanol concentration ! 40%. Abstracted with permission from data in Bychkova VE, Dujsekina AE, Klenin SI, Elisaveta IT, Uversky VN and Ptitsyn OB (1996) Biochemistry 35: 6058–6063. Copyright 1996 American Chemical Society.

have shown them to fold in essentially two-state transitions. The kinetics of folding of globular proteins, however, can be very complex. For example, the kinetics of the unfolding and refolding of bovine beta-lactoglobulin, a predominantly E-sheet protein in the native state, are illustrated in Figure 6. The kinetics of unfolding (Figure 6A, B) show a singlephase transition between the folded and denatured form, with the loss of secondary structure (Figure 6A) and tertiary structure (Figure 6B) being simultaneous. The refolding reaction, in contrast, is a complex process composed of different kinetic phases

(Figure 6C). In particular, a burst-phase intermediate is formed during the dead time of stopped-flow measurements that shows more intense ellipticity signals in the far UV than does the native state. This moreintense CD is due to the transient formation of a nonnative alpha-helical structure (Figure 6D). The CD spectrum suggests the folding intermediate exists in a molten globule state, since there are no near-UV CD bands indicative of tertiary structure. Similarly, intermediate states have been shown to exist in the refolding of diverse proteins including cytochrome C, ribonuclease HI from E. coli, dihydrofolate

124 BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD

Figure 6 (A) Unfolding curve of E-lactoglobulin ( E-LGA) measured by the CD changes at 293 nm at pH 3.2 at 4.5°C. The change was initiated by jumping the guanidine HCl (GdnHCl) concentration from 0 to 4.0 M. The vertical arrows show the change in ellipticity in deg.cm2.dmol–1 (B) Same, measured at 220 nm. (C) Kinetic refolding curve of E-LGA measured by the CD change at 221 nm. Refolding was initiating by dropping the concentration of GdnHCl from 4 to 0.4 M. (D) CD spectrum of E-LGA in the far UV at pH 3.2 and 4.5°C. (—) native state; (---) unfolded state in 4 M GdnHCl; ( ) disulfide-cleaved and carboxymethylated material in 4 M GdnHCl; (o) ellipticity of E-LGA at 0 time after refolding; (•) ellipticity of E-LGA at 255 min after refolding. Reprinted from data in Kuwajima K, Yamaya H and Sugai S (1996) J. Mol. Biol. 264: 806–822, by permission of Academic Press.

reductase, alcohol dehydrogenase, carbonic anhydrase, retinoic acid-binding protein and staphylococcal nuclease. Analysis of proteinligand interactions

Circular dichroism can be used to follow the binding of ligands to proteins, peptides, nucleic acids and carbohydrates provided one of two criteria is fulfilled. Either the ligand must bind in an asymmetric fashion that induces extrinsic optical activity in the chromophores of the bound ligand, or the binding must result in a conformational change in the macromolecule that results in a change in its intrinsic CD spectrum. In the case of equivalent binding sites, the change in ellipticity due to complex formation is directly proportional to how much ligand is bound. Thus, the change in ellipticity as a function of substrate concentration can be used to estimate the binding constants.

Extrinsic ellipticity bands (called Cotton effects), developed when chromophores bind to proteins, have been used to study proteinligand interactions in many diverse systems. Figure 7A illustrates the induction of extrinsic CD bands when retinol binds to interphotoreceptor retinol-binding protein, and Figure 7B shows the change in the ellipticity at the wavelength of maximal ellipticity as a function of ligand concentration. Extrinsic Cotton effects have been used to study the binding of many other cofactors to proteins, including the binding of nucleotides to lactic dehydrogenase, folates and antimetabolites to dihydrofolate reductase, ATP to GroEL and pyridoxal phosphate to tryptophan synthase. They have also been used to study the interactions of substrates with enzymes including tryptophan synthase and N-acetylglucosamyl transferase, to study the interactions of drugs with bovine serum albumin, and to examine the interactions of DNA binding proteins with DNA.

BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD 125

dimerization domains, and a relatively unstructured basic region which binds to the DNA. Upon DNA binding the basic region assumes a helical conformation.

Nucleic acids Nucleic acids are polymers of nucleotides that consist of purine or pyrimidine bases attached to a phosphorylated sugar. Monomeric sugars are asymmetric and they induce a small amount of ellipticity in the attached planar-symmetric base. Polymerized nucleic acids, however, are rigid and highly asymmetric. In the polymerized nucleic acids, the bases stack with one another in precise orientations depending on the conformational state. Base stacking leads to characteristic splittings of the transitions of chromophores of the bases into multiple transitions, and the rigidity leads to CD bands with high rotational strength. CD spectra of polynucleotides are sensitive both to their sequence and conformation, and can provide a good deal of structural information. Analysis of the secondary structure of polynucleotides

Figure 7 (A) Near-UV mean-residue circular-dichroism ellipticity, [T], in deg.cm2.dmol–1, of interphotoreceptor retinol-binding protein (IRBP), 0.4–0.8 mg ml–1, with and without bound retinol. (B) Circular-dichroism titration curve of apo-IRBP, 0.4 mg ml–1, with retinol. The height of the extrinsic Cotton effect at 330 nm is calculated in ellipticity per mole of IRBP. Redrawn with permission from data in Adler AJ, Evans CD and Stafford WF (1985) J. Biol. Chem. 260: 4850–4855. Copyright 1985 The American Society for Biochemistry & Molecular Biology.

The binding of ligands to proteins often results in conformational transitions. The resultant change in the intrinsic ellipticity of the backbone amides caused by the conformational change can be used to obtain the ligand-binding constants. For example, divalent cations are effectors in many biological systems, and their binding often induces large conformational changes in the target protein. Calcium-ion binding to calmodulin and troponin C induces an increase in the ellipticity of the proteins due to stabilization of D-helical regions. Many other broad classes of protein-ligand interactions result in conformational changes. For example, DNA binding to many different DNA transcription factors results in increases in the helical content of the factors. These DNA binding proteins often contain D-helicalcoiled-coil or helix-loop-helix motifs, which serve as

Homopolynucleotides Even very simple polynucleotides show multiple conformational states and complex CD and ORD spectra. Compounds as small as homodimers of ribo and deoxyribonucleotides show evidence of base stacking in solution from measurements of optical activity. For example, Figure 8A illustrates the circular dichroism properties of several riboadenylate compounds. AMP shows a single absorption band and a single negative CD band at 260 nm. Diadenylic acid has a single UV band at 260 nm, but a split CD spectrum with a positive band at 270 nm and a negative band at 250 nm. Theoretical calculations of the CD spectrum agree with a model where the bases are stacked in a parallel fashion in the dinucleotide. The stacking leads to hypochromicity in the absorption spectrum, and splitting of the transitions into two transitions, one perpendicular and one parallel to the helix axis with equal strength and opposite rotatory strength. Poly(rA) has a spectrum similar to that of diadenylate but higher in magnitude. The CD spectrum of poly(rA) changes with pH, chain length and temperature. At neutral pH, poly(rA) is unprotonated and forms a single-stranded helix. At acidic pH, poly(rA) forms dimers. Depending on the pH there are two conformational forms, one where poly(rA) is fully protonated, and one where it is half-protonated. Figure 8B shows the

126 BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD

Figure 8 (A) Circular dichroism of adenylate compounds at pH 8.5 in unbuffered 0.1 M NaF at 22°C. (—) poly (rA); ( ) rA-2’Omethyl-prAp; (-----) rA(3cp5)rA; (·····) rAMP-5'. Abstracted with permission from data in Adler AJ, Grossman L and Fasman GD (1969) Biochemistry 8: 3846–3859. Copyright 1969 American Chemical Society. (B) Rotational strength of the positive band of adenylate oligomers at 0°C as a function of the chain length, N. (u) pH 7.0; (s) pH 4.5. (C) Rotational strength of the positive band of poly(rA) as a function of temperature at ( u) pH 7.4 and (s) pH 4.5. (B) and (C) are reprinted from data in Brahms J, Michelson AM and Van Holde KE (1966) J. Mol. Biol. 15: 457–488, by permission of Academic Press.

chain-length dependence, and Figure 8C the temperature dependence, of the rotational strength of the positive-ellipticity band of the single-stranded form at pH 7.4 and the double-stranded half-protonated form at pH 4.5. Both the chain-length dependence and the unfolding of the double-stranded forms as a function of temperature, are more cooperative than that of the single-stranded form. Heteropolynucleotides Circular dichroism spectra of nucleic acids are complex, compared to those of homopolynucleotides because four different bases contribute to the CD spectra, and the spectra are sequence dependent. In addition, nucleic acids display great conformational diversity and may form single-, double- or triple-stranded helices. For example, the spectra of poly(rA), poly(rU), the dimer poly(rA)·poly(rU) and the trimer poly(rU)· poly(rA)·poly(rU) are illustrated in Figure 9.

Complex formation results in shifts in wavelength and intensity of the bands, compared to the addition of the spectra of the unmixed components. The various oligomerization states of the polynucleotides, moreover, can exist in multiple conformations. For example the A, B and Z forms of double-stranded nucleic acids have distinct CD spectra. Figure 10A illustrates the CD spectra of poly (dAdC)·poly(dGdT) in the A, B and Z conformations and Figure 10B illustrates CD spectra of a double-helical polynucleotide, poly(dGdC)· poly(dGdC) in the B form, the Z form and in a form in which the bases assume the Hoogsteen base-pairing conformation. Analysis of conformational transitions of nucleic acids

Conformational transitions of nucleic acids can be followed by examining CD and ORD spectra as a

BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD 127

whether there are folding intermediates, as in the case of proteins. Analysis of ligand binding to nucleic acids

Figure 9 The CD spectra of (——) Poly(rA); (·····) Poly(rU); (----) poly(rU)·poly(rA); ( ) poly (rU)·poly (rA)·poly (rU) at pH 7 in sodium phosphate buffer. Redrawn with permission from data in Steely T, Gray DM and Ratiff RL (1986) Nucleic Acid Research 14: 10 071–10 090. Copyright 1986 Oxford University Press.

function of temperature and denaturants, and the thermodynamic parameters of the transitions can be determined using the same equations used to determine the thermodynamics of folding of proteins. In addition, the spectra of nucleic acids, obtained under varying conditions, can be deconvoluted using the CCA and SVD methods, to determine

Circular dichroism and ORD are excellent methods to follow the binding of ligands to nucleic acids. The same equations that are used to analyse protein ligand interactions may be applied to the study of DNAligand interactions. For example, CD and ORD have been used to examine the binding of DNA to proteins, drugs and amines. Figure 11 illustrates the effect of binding of spermidine to poly(dT)·poly(dA)·poly(dT). The addition of the amine has dramatic stabilizing effects. The conformation of the polynucleotide complex undergoes sequential changes from B-DNA to triplex DNA as the concentration of spermidine is increased from 0 to 50 µM. At 60 µM spermidine, the CD spectrum of triplex DNA is comparable to that of \-DNA, with a strong positive band centred around 260 nm. A negative band is also found at 295 nm. At higher concentrations of spermidine, however, the intensity of the positive band progressively decreases. The peak intensity is found at a 1:0.3 molar ratio of DNA phosphate-spermidine. Figure 12 illustrates an example of DNA-protein binding study. The gene 32 coded protein of bacteriophage T4 is necessary for genetic recombination and DNA replication. It denatures poly(dAdT)·poly(dAdT)] and T4 DNA at temperatures far below their regular melting temperatures. The protein interacts with native DNA and a variety of synthetic polynucleotides. Illustrated here is the interaction of the protein with poly(dA). The CD spectra suggest that the protein keeps poly(dA) in a conformation that is equivalent to that of a singlestranded form in high salt. Similar results were

Figure 10 (A) The CD spectra of poly (dAdC)·poly(dGdT) in the (·····) A, (—) B and (---) Z conformations. Redrawn from data in Riazance-Lawrence JH and Johnson WC Jr (1992) Biopolymers 32: 271–276 by permission of John Wiley and Sons, Inc. (B) CD spectra of poly(dGdC)·poly(dGdC) in the (—) B-form, (- - -) Z-form and (····) in a form in which the bases assume the Hoogsteen basepairing conformation. Abstracted with permission from data of Seger-Nolten GMJ, Sijtsema NM and Otto C (1997) Biochemistry 36: 13 241–13 247. Copyright 1997 American Chemical Society.

128 BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD

Figure 11 CD spectra of poly(dT)·poly(dA)·poly(dT) in the presence of (s) 0; (●) 60; (,) 70; (j) 80; (.) 90 and ( u) 100 µM spermidine. The insert shows the effects of (s) 0, (●) 5, (.) 10, and (i 25 µM spermidine. Data were recorded at 25°C in 10 mM sodium cacodylate and 0.5 mM EDTA at pH 7.2. Redrawn with permission from data in Thomas TJ, Kulkarni GD, Greenfield NJ, Shirahata A and Thomas T. (1996) Biochem J. 319: 591–599. Copyright (1996) Portland Press.

obtained with other dimeric DNA analogues including poly(dA)·poly(dT) and poly(dT).

Carbohydrates Carbohydrates are much more difficult to study using ORD and CD than proteins and nucleic acids and much less literature is available. First, carbohydrates mainly absorb below 190 nm, a region difficult to examine using commercial CD spectrometers. Second, unlike proteins and nucleic acids, there are no common sets of spectral characteristics which easily define the conformation of carbohydrates. However, ORD and CD have been used to obtain useful information about the structure of carbohydrates. In favourable circumstances they can give information about the structure and linkages in carbohydrates. In addition, they can distinguish helical from disordered conformations and be used to monitor conformational transitions induced by salts, solvents and temperature.

spectra of monomeric sugars, researchers have been able to assign the CD bands to specific transitions, and to determine quadrant rules for the effect of structure on the sign of the transitions. For example, the CD spectra of methyl-E-galactopyranoside, carrageenan and agarose are shown in Figure 13. X-ray films of carrageenan show that the conformation is a double helix. The CD spectrum of carrageenan is consistent with the conformation seen in the X-ray analysis. There are conflicting models for the structure of agarose in gels. In one model the chains are extended and in the other they are helical. The geometry-dependent linkage contributions to the CD of agarose were determined by subtracting the monomeric CD from the CD of the polymer. Application of the quadrant rules indicated that agarose was also helical. The difference in CD sign between carrageenan and agarose arises from a translocation of the beta-D-galactose O-2 atom from one quadrant to the neighbouring quadrant of the C-5-O-5-C-1 ether chromophore of the preceding anhydro sugar residue.

Determination of carbohydrate structure and linkages

Analysing conformational transitions of carbohydrates using CD and ORD

While no simple basis spectra exist that can be used to determine the conformation of carbohydrates, analysis of the CD spectra of monomeric carbohydrates has provided empirical rules for extracting some conformational information. By comparing the

While the analysis of the CD spectrum of carbohydrates to obtain detailed structural information is complex, CD and ORD serve as simple methods for following orderdisorder transitions in carbohydrates. Because the absorption bands of

BIOMACROMOLECULAR APPLICATIONS OF CIRCULAR DICHROISM AND ORD 129

Figure 12 CD spectra of poly(dA) complexed with the gene 32 coded protein of bacteriophage T4 at (—) 1.0, (----) 29.9 and (·····) 47.7°C. The nucleotide to protein ratio is 8.5. Also illustrated ( ) is the spectrum of free poly(dA) at 1°C. Redrawn with permission from Greve J, Maestre MF, Moise H and Hosoda J (1978) Biochemistry 17: 887–893 Copyright (1978) American Chemical Society.

carbohydrates are at very low wavelength, many of the studies of conformational transitions have been done using ORD, since ORD bands extend even into the visible region and are more easily accessible. For example, Figure 14 shows how the effects of potassium-ion concentration and temperature on the order disorder transition of xanthan, an extracellular polysaccharide produced by Xanthomonas campestris, have been followed by measuring the optical rotation at 365 nm. The data were used to find the enthalpy of folding of the polymer. The data could also be fitted by ZimmBragg helix-coil transition analysis. Similar studies of conformational transitions have been performed on other carbohydrates including acetan, carrageenan, succinoglycan and hyaluronate. In addition to monitoring the folding of carbohydrates by examining changes in the intrinsic

Figure 13 CD spectrum of (—) methyl E-D-galactopyranoside, (···) carrageenan and ( ) agarose. Redrawn with permission from data in Arndt ER and Stevens ES (1993) J. Am. Chem. Soc. 115: 7849–7853, Copyright (1993) American Chemical Society, Arndt ER and Stevens ES (1994) Biopolymers 34: 1527–1534 and Stevens ES and Morris ER (1990) Carbohydr. Polym. 12: 219–224.

optical activity of the carbohydrate, structural information has been obtained by examining the CD and ORD spectra of bound dyes. For example when methylene blue binds to carrageenan, the binding induces optical activity in the dye which changes as a function of temperature. There is a progressive loss of ellipticity as a function of temperature, suggesting that the dye binds to the double-helical form of the polysaccharide, but not the coil form.

List of symbols K = equillibrium folding constant; n = number of molecules; N = chain length; P = correlation coefficient; R = gas constant; T = temperature; 'Cp = change in heat capacity; 'G = free energy of folding; 'H = Vant Hoff enthalpy of folding 'S = entropy of folding; T ellipticityV mean-square error See also: Biomacromolecular Applications of UVVisible Absorption Spectroscopy; Carbohydrates Studied By NMR; Chiroptical Spectroscopy, General

130 BIOMACROMOLECULAR APPLICATIONS OF UV-VISIBLE ABSORPTION SPECTROSCOPY

Spectrometry; Vibrational CD Spectrometers; Vibrational CD, Applications; Vibrational CD, Theory.

Further reading

Figure 2

Analytical process integrity.

This article is restricted to the calibration requirements for spectrometers within the regulatory context which from a practical standpoint means the major pharmacopoeias.

Spectroscopic coverage Current spectrometric monographs include calibration requirements for: infrared (IR), near infrared (NIR), ultraviolet and visible (UV-visible), nuclear magnetic resonance (NMR), circular dichroism and polarimetry. Monographs for atomic spectrophotometry, emission and absorption, fluorescence and X-ray fluorescence are available in the European

NIS 45, Edition 2, May 1996, Accreditation for Chemical Laboratories. Environmental Protection Agency, 2185-Good Automated Laboratory Practices, 1995. International Conference on Harmonisation Note for guidance on validation of analytical procedures: methodology, 1995, CPMP/ICH/281/95. FDA Center for Drug Evaluation and Research, Reviewer Guidance; Validation of Chromatographic Methods, November 1994. 21 CFR 211 SUBPART I LABORATORY CONTROLS §211.160 (b) (4). e.g. United States Pharmacopoeia 23, 1995, chromatography, dissolution testing, Refractive index, spectrophotometry and light scattering. e.g. BP 1998, Appendix II; Infrared spectrophotometry, Ultraviolet and Visible spectrophotometry and Pharm. Eur. V.6.19 Absorption Spectrophotometry. Australian Code of GMP for Therapeutic GoodsMedicinal Products; Appendix D, November 1991; Guidelines for Laboratory Instrumentation. DEPARTMENT OF HEALTH AND HUMAN SERVICES Food and Drug Administration [Docket No. 93N0184] Barr Laboratories, Inc.; Refusal to Approve Certain Abbreviated Applications; Opportunity for a Hearing, Vol. 58 No. 102 Friday, May 28, 1993 p 31035 (Notice) 1/1061.

172 CARBOHYDRATES STUDIED BY NMR

Introduction Carbohydrates, the most widely abundant biological molecules, are key components within a wide variety of biological phenomena. Polysaccharides such as glycogen, cellulose and starch have important structural or nutritional roles, but it is the mediation of specific recognition events by carbohydrates that has sparked detailed analyses of structure, conformation and function. The non-invasive, non-destructive nature of the technique is commonly mentioned as the most important reason for analysing carbohydrates by NMR, and is of considerable advantage when dealing with small amounts of precious material. This advantage is, however, offset by the relative insensitivity of the technique when compared with other methods such as mass spectrometry or enzymatic approaches that may be more appropriate for specific problems of sequence determination or compositional analysis. The value of NMR analysis is apparent in the additional data obtainable regarding composition, structure, conformation and mobility necessary for understanding recognition processes. The analysis of carbohydrates by NMR is characterized by a modification of existing techniques to address the particular problems of oligosaccharide spectra, namely poor dispersion, heterogeneity, a relatively high degree of interresidue mobility and a lack of conformational restraints. This article will attempt to provide an introduction to a large field of study, and readers are encouraged to look to the Further reading section for additional information.

Sample preparation In high-field NMR, the use of high quality sample tubes can have a significant impact upon the quality of the spectra; 5 mm tubes are suitable for most work, with 10 mm sample tubes often being used for 13C analysis. For biologically relevant NMR analysis, experiments are generally carried out in aqueous solution. A strong water signal prevents direct analysis of proton signals lying beneath it and, owing to

MAGNETIC RESONANCE Applications the problems of dynamic range, would reduce the sensitivity with which weak solute signals are detected. If, as is often the case, the exchangeable OH or NH protons are not of interest, then experiments are usually carried out in 2H2O. Repeated dissolution and evaporation or lyophilization is recommended to reduce the proportion of residual protons remaining. In most cases, two dissolutions into > 99.8% 2H2O followed by a final dissolution into > 99.95% 2H2O, preferably from a sealed ampoule, should be sufficient. Other precautions, such as pre-wetting pipettes, further rounds of dissolution and drying, or the use of a dry box may be used for greater sensitivity or quality. Although extremes of pH or pD (outside the range δ 58) are to be avoided, buffer solutions are only necessary in two cases: if the molecule is charged, and the charge state is relevant to the study; and in the study of acid-labile carbohydrates, such as sialylated oligosaccharides. The use of 1H2O as solvent is necessary in some cases, and requires appropriate water suppression techniques. Care must be taken to ensure that the water is free from impurities, as well as dissolved gases or paramagnetic species. A deuterated solvent to provide an appropriate lock signal should be added, usually 510% 2H2O. Oligosaccharides are not normally soluble in nonaqueous solvents, except for dimethyl sulfoxide (DMSO). Despite the biologically irrelevant environment, this permits the observation of exchangeable protons, and there is also some increase in the proton spectral dispersion. Similar precautions to those noted for the use of 2H2O should be observed, such as the use of high isotopic purity (> 99.96%) solvent, and repeated dissolution to remove exchangeable protons. To approach optimal line widths, it is often advisable to remove soluble paramagnetic components by passage through a suitable chelator. Degassing to remove dissolved oxygen is also recommended. For acceptable spectra within a reasonable time using a 500 MHz spectrometer, 10 nmol is an approximate lower limit for 1D 1H spectra; for 2D 1H spectra, amounts of 1 µmol are preferable, although spectra with as little as 100 nmol are possible with care.

CARBOHYDRATES STUDIED BY NMR 173

Sample volumes vary between 0.40.7 mL, depending on the size of the spectrometer RF coils. Temperature should be maintained at a constant level, preferably one or two degrees above room temperature for stability. Sample spinning is not necessary, except in cases where resolution is a particular problem. T1 for protons in carbohydrates is of the order of 0.1 to 0.5 s, and recycle delays of 11.5 s are usually sufficient. Referencing of samples is commonly to acetone at δ 2.225 relative to DSS (5,5-dimethylsilapentanesulfonate) at 25°C, or DMSO at 2.5 ppm.

Analysis of carbohydrate structure by NMR The problems of carbohydrate structure addressed by NMR can be divided into four parts: 1. What is the composition of the molecule: what monosaccharides are present, are they in furanose or pyranose configurations, and α or β anomeric forms? 2. What is the nature of the linkages between the monosaccharides: at what positions do the substituents occur, and in what sequence? 3. What is the conformation of the molecule: what conformation or family of conformations are populated about the O-glycosidic bonds and hydroxymethyl rotamers? 4. What are the dynamic properties of the molecule? Whilst the first and second questions can be satisfactorily answered by NMR, they are perhaps better approached in tandem with chemical or enzymatic methods, such as hydrolysis, followed by reduction and analysis of alditol acetates by GC-MS, or digestion with specific glycosidases. Such a concerted approach has been used to successfully determine the composition and configuration of many oligosaccharides and glycans. The unique advantages of NMR in the analysis of carbohydrate structure are only fully apparent in consideration of points 3 and 4. In contrast to the unbranched polymeric nature of polypeptide and nucleic acid chains, carbohydrates may be branched structures, capable of substitution at several points. The monosaccharide constituents are polymerized in nature by a non-template directed, enzymatic process; the resulting oligosaccharides are often heterogeneous, differing in detail from a consensus structure. NMR is particularly efficient at investigating the solution conformations, and the dynamic properties of such molecules.

NMR of carbohydrates

Primary analysis and assignment: 1H 1D spectra

The 1H NMR spectrum of the disaccharide galactopyranose β 14 linked to glucopyranose (Galpβ1-4Glcp, lactose, shown in Figure 1) is shown in Figure 2. Despite the relative simplicity of the molecule, with only 14 proton signals observable, the spectrum is surprisingly complex. This is due to the relatively small chemical shift dispersion of the non-exchangeable ring protons, resonating between δ 34. The small chemical shift dispersal combined with homonuclear spinspin coupling gives rise in many cases to non-first-order spectra, complicating the measurement of spin-coupling values. This makes the assignment and analysis of even quite small saccharides quite difficult, and as a consequence the use of spectrometers of field strength 400 MHz or above is recommended. Primarily owing to the electron-withdrawing effect of the ring oxygen, the anomeric (H1) protons give rise to signals lying outside this envelope of ring protons, between δ 4.255.5. Since equatorial protons experience a shift of approximately δ 0.5 relative to axial proton signals, α anomeric protons of D-sugars tend to be found between δ 4.95.5, and β protons between δ 4.34.7. The ring protons of each monosaccharide type give rise to characteristic patterns within the envelope of overlapping ring proton resonances. However, incorporation of a monosaccharide residue into an oligosaccharide will cause changes in these chemical shift patterns. For example, substitution at a given carbon causes the signal arising from the attached proton to change by δ 0.20.5. The glucose residue of the sample shown in Figure 2 is free to mutarotate between α and β forms. The signals arising from these are easily distinguished from those owing to the galactose H1 by the difference in area. The two glucose resonances have a combined area approximately equal to that of the galactose H1 resonance.

Figure 1 Structure of the disaccharide Galpβ1-4Glcp, showing the numbering of protons within the sugar rings.

174 CARBOHYDRATES STUDIED BY NMR

Figure 2 1H 1D spectrum of the disaccharide galactopyranose β1-4glucopyranose at 500 MHz and 303 K in 2 H2O. Peak assignments are given for well-resolved peaks, following the numbering of Figure 1. The unresolved proton ‘envelope’ consists of the remaining ring proton resonances. The peak marked ‘HOD’ arises from residual water within the sample.

Anomeric signals typically exhibit characteristic doublets, arising from the 3JHH H1H2 coupling, whereas ring protons show more complex multiplet patterns. These couplings are proportional to the dihedral angle between the two protons. Typical values for 3JHH in carbohydrates are: axialaxial 78 Hz, axialequatorial and equatorialequatorial 34 Hz. During the early 1980s, Vliegenthart and coworkers identified a number of structural reporter groups, outlined in Table 1. Measurement of the chemical shift values, coupling patterns and line widths of signals arising from these elements can be interpreted to aid in the assignment and interpretation of 1D 1H NMR spectra. Sugar composition can thus be identified, and signals assigned on the basis of chemical shift and characteristic coupling values. The type and number of each can be established by consideration of chemical shifts, coupling patterns and relative integrals. Owing to the poor dispersion of the majority of resonances, this is usually limited to resolved anomeric protons or other structural reporter groups. Despite the development of highly efficient experiments based upon the use of selective excitation to produce edited one-dimensional spectra with more manageable information content, full proton assignment usually

relies upon two- or, in some cases, three-dimensional techniques. Two-dimensional homonuclear analysis

The poor dispersion and strong coupling present in carbohydrate proton spectra present particular problems for the assignment process. In addition, the

Table 1 Structural reporter Vliegenthart and co-workers

groups

described

Sugar type

Proton(s)

Parameter

Information content

All

δ, J

Mannose

H2, H3

Sialic acid

Fucose

H5, CH3

Galactose

H4, H4

Amino sugars

N-acetyl CH3

Residue and linkage type Substitution within core region of branched glycan Type and configuration of linkage Type and configuration of linkage: structural environment Type and configuration of linkage Sensitive to small structural variations

CARBOHYDRATES STUDIED BY NMR 175

characteristics of a given spin system differ for each oligosaccharide and are highly dependant on structure. The use of two-dimensional experiments alleviates these problems. Coherence transfer methods The correlated spectroscopy (COSY) experiment reduces the degree of resonance overlap by separating resonances into two orthogonal proton dimensions. The 1D spectrum lies along the diagonal, with cross-peaks joining pairs of J-coupled spins. For carbohydrates, assignment can then start with the well-resolved anomeric protons, and continue by stepwise J-correlation to the remaining nuclei. This simple process is complicated in many cases by overlap and strong coupling between signals, even in two dimensions. To reduce these difficulties, high quality spectra are needed, with optimal digital resolution. Sometimes absolute value mode, as opposed to pure phase spectra, can be preferable for easy assignment, despite the theoretical disadvantages of a phase-twist line shape. Carbohydrate line widths are generally narrower than those of proteins and nucleic acids, and the use of absolute value spectra

does not significantly diminish the quality of spectra, as seen in Figure 3. The COSY spectrum requires a pseudo- echo weighting function to remove dispersive line shapes, reducing the sensitivity of the experiment. If sensitivity is an issue, then an inherently more sensitive experiment, such as the absorption mode DQ-COSY (see below) should be used. If necessary, linkage positions can be identified using a COSY spectrum of a sample dissolved in DMSO-d6. The hydroxyl protons do not exchange with solvent, and give rise to observable signals within the spectrum. Substitution removes the hydroxyl proton at that position, which can then be identified by the absence of a hydroxyl proton signal. The COSY experiment may fail to generate the expected cross-peaks, for several reasons: 1. For strongly coupled neighbours, the cross-peak lies close to the diagonal, and may not be seen. 2. Pairs of nuclei with small scalar coupling values (3JHH) will produce low intensity cross-peaks. Since the active coupling is antiphase in COSY cross-peaks, they will cancel if the value of J is less than the line width. This is particularly

Figure 3 1H–1H 2D COSY spectrum of galactopyranose β 1-4glucopyranose at 500 MHz and 303 K in 2 H2O. A full proton assignment is usually possible from a spectrum like this, and peak identities are shown where space permits. The well-resolved anomeric protons provide a useful starting point for assignment, and subsequent pairs of spin-coupled nuclei are correlated through off-diagonal peaks.

176 CARBOHYDRATES STUDIED BY NMR

noticeable in larger molecules. Common examples are: D-mannopyranose β H1H2 (J < 1 Hz) and D-galactopyranose H4H5 (J < 1 Hz). 3. Unrelated resonances may simply occur at the same position, and cannot be distinguished. Experiments incorporating isotropic mixing sequences The homonuclear HartmannHahn (HOHAHA) and the closely related total correlation spectroscopy (TOCSY) experiments are used to overcome problems of overlap and strong coupling. These experiments give correlations to all other spins within the same coupling network. The second pulse in the COSY sequence is replaced by an isotropic mixing period, such as a spinlock sequence (TOCSY) or an efficient decoupling sequence, such as MLEV-17 (HOHAHA). The spin system becomes strongly coupled and coherence transfer occurs between all nuclei within the coupling network. This illustrates an advantage over experiments such as RELAY-COSY: values of 3JHH between adjacent nuclei vary, and the use of an isotropic mixing period produces an increased efficiency of transfer. The presence of small couplings between nuclei, reducing the efficiency of transfer, still presents a problem. Resolution in this type of experiment is better than for a COSY spectrum; however, more cross-peaks are generated, producing a more complicated spectrum. This is at first sight hard to interpret; however, proton assignments in each ring can be determined by inspection of a line lying through the frequency of the anomeric proton of the sugar residue. The cross-peaks found along this line can be distinguished by reference to the COSY spectrum, and by consideration of the multiplet structure. Pyranose rings have an effectively rigid ring geometry, and it is possible to estimate the couplings between adjacent protons based upon the dihedral angle between them. Judicious variation of the length of the isotropic mixing period can also be used to give stepwise correlations along the spinsystem, since shorter periods correlate proportionately smaller fragments. One-dimensional analogues of this experiment, using selective pulses to specifically excite anomeric protons, may be quicker and more efficient, especially when the anomeric protons are well resolved. Multiple-quantum homonuclear experiment Incorporation of a multiple-quantum filter into a COSYtype sequence reduces the complexity of the spectrum, and aids in the assignment process. A DQF-COSY, incorporating a double-quantum filter transparent to two or more coupled spins, therefore passes all signals except singlets. Such a

sequence has a slightly reduced sensitivity (crosspeak intensity is approximately half that of a comparable COSY), but there is no need for the sensitivityreducing weighting functions that are necessary with COSY. One noticeable feature is the absence of a diagonal, thus cross-peaks lying close to the diagonal can be more easily identified. Incorporation of a triple-quantum filter produces a spectrum with signals due only to three or more coupled spins with two resolvable couplings. There are no singlets or doublets, including anomeric resonances, although suppression is not absolute, due to the presence of small couplings along four or more bonds. Under ideal conditions, one would expect to see resonances from only H5 and H6 protons, and this experiment is best used in conjunction with HOHAHA for full assignment. Sensitivity is much reduced relative to COSY spectra. Through-space dipolar interaction methods Although primarily used for conformational analysis, the nuclear Overhauser effect spectroscopy (NOESY) experiment is also helpful in the assignment process. The resulting spectra show correlations between protons coupled by through-space dipolar interactions. Both inter- and intraresidue NOEs help to confirm the consistency of assignments made by the techniques described above. Long-range NOEs are not often seen in carbohydrates, and the observation of NOEs between protons on adjacent residues is often evidence of the linkage positions between the two sugars. Through-space dipolar interactions are of most use in the conformational analysis of biomolecules, as described in the section on conformational analysis below. At 500 MHz, moderate-sized (more than six residues) oligosaccharides lie within the spin-diffusion limit. However, for smaller molecules, as the value of the function Z0Wc (where Z0 is the Larmor frequency of protons, and Wc is the correlation time of the molecule) approaches 1 then the value of the NOE tends towards 0. Cross-peak intensities of NOESY spectra of smaller oligosaccharides (25 residues) may thus become too small to measure accurately. In such cases, the rotating frame Overhauser effect spectroscopy (ROESY, originally referred to as CAMELSPIN) experiment is commonly used to measure NOE values. To reduce the appearance of HOHAHA-like cross-peaks, a low power spinlock field should be used, and the transmitter carrier offset to the low-field end of spectrum. The offset dependency of cross-peak intensities should also be removed by 90° pulses at either end of spinlock period.

CARBOHYDRATES STUDIED BY NMR 177

Three-dimensional experiments, often essentially hybrids of simpler sequences such as HOHAHACOSY, have been used successfully to resolve cases of particularly difficult resonance overlap. Hydroxyl protons

The quest for additional conformational information has led to the investigation of hydroxyl protons in aqueous solution. Samples are dissolved in mixed methanolwater or acetonewater solvents, and analysed in capillary NMR tubes at low (5 to 15°C) temperatures. Chemical exchange of hydroxyl protons is reduced to the point that it is possible to use them as probes of hydration and hydrogen bonding. Distance information can also be extracted from NOESY or ROESY spectra under these conditions. 13C

NMR of carbohydrates

Primary analysis and assignment

Initial investigations of the 13C spectra of carbohydrates were of natural abundance samples, although spectra of isotopically-enriched samples have become more common in recent years. 13C spectra are usually acquired with broad band proton decoupling. The resulting spectra are not very complex with, at natural abundance, a single sharp peak for each nucleus, there being no visible carbon carbon coupling. The value of the carbonproton coupling is approximately 150180 Hz, and without proton decoupling this splitting, along with further two- and three-bond couplings, can render the spectrum quite complex. Monosaccharides exhibit characteristic carbon chemical shift patterns. Anomeric carbons typically occur between δ 90110 in pyranose rings, C6 (hydroxymethyl) signals between δ 6075, and the remaining carbons resonate between δ 65110. Substitution at a given carbon causes an approximate shift of up to δ + 10. The usefulness of this shift in identifying the position of substitution is complicated by an additional δ 1 to 2 shift of the surrounding carbon signals. In addition, the local structure of the saccharide can have a considerable effect upon the chemical shift values. Residue type and substitution patterns can thus be discovered, with care, from measurements of the carbon chemical shifts of a simple sugar. Databases of 13C chemical shift information for large numbers of simple carbohydrates are available for comparison in the literature. Carbonproton and carboncarbon spin couplings A consistent estimate of the anomeric configuration

is given by the value of the one-bond 1JCH coupling; β anomers show a value of ∼ 160 Hz, and α anomers ∼ 170 Hz. Three-bond carbonproton scalar coupling values, such as 3JCCCH, or 3JCOCH, provide useful indicators of conformation. These couplings can be used to confirm the internal conformation of a saccharide residue, using appropriate Karplus curves. Across the glycosidic link, values of the couplings H1C1OCx (where x is the aglyconic carbon) and C1OCxHx are proportional to the glycosidic dihedral angles I and \ respectively (see section on conformational analysis below). Selective incorporation of 13C has been successfully used to simplify and enhance the measurement of these parameters. Linkage positions can be unambiguously determined by the identification of these couplings using experiments such as the 2D 13C1H multiple-bond correlation experiment (HMBC). This experiment correlates carbon and proton resonances by means of 3JCH couplings; 1JCH correlations are suppressed, leaving only long-range couplings. Those couplings that span the glycosidic bond can thus be identified and measured. Multidimensional, heteronuclear analysis

Enrichment with 13C allows the use of a wide range of useful experiments that greatly facilitate the investigation of carbohydrate structure. Proton detected versions of carbonproton correlation experiments, such as that shown in Figure 4, overcome the sensitivity limitations of observing carbon signals. If the proton assignments are known, assignment of carbon resonances from such spectra is a relatively simple process. The use of carbon nuclei to edit spectra into an orthogonal frequency dimension to overcome overlap has led to the development of a series of useful hybrid 3D or pseudo-3D experiments, such as HCCH-COSY or HMQC-NOESY. Editing spectra in this way has helped to assign the spectra not only of complex glycans in free solution, but also of the attached carbohydrate chains of glycoproteins, both at natural abundance as well as enriched with 13C and 15N. The use of these heteronuclei allows the carbohydrate resonances to be observed separately from those due to the protein portion of the molecule.

Conformational analysis Measurement of 3JHH couplings shows that pyranose rings do not show any large degree of internal flexibility, except for pendant groups such as the hydroxymethyl in hexopyranoses. Interpretation of these uses the Haasnoot parametrization of the Karplus equation for 3JHH couplings in HCCH

178 CARBOHYDRATES STUDIED BY NMR

Figure 4 HSQC 1H–13C correlation spectrum of galactopyranose β1-4glucopyranose at 500 MHz and 303 K in 2H2O. Resonances are labelled with the appropriate assignment for each correlation. In practice, these can be relatively easily identified by comparison with previously assigned proton or carbon chemical shifts. Reproduced by courtesy of Homans SW and Kiddle GR, University of St. Andrews.

portions of the saccharide ring:

where T is the HCCHc torsion angle, 'FL are the Huggins electronegativities of the substituents relative to protons, and [ is either +1 or 1 depending upon the orientation of the substituent. If stereo-specific assignments have been made, orientation about hydroxymethyl rotamers can be determined by similar parametrizations. Conformational variability in pyranoside oligosaccharides is mostly owing to variations about the glycosidic torsion angles I and \ and, for 1 → 6 linkages, the Z angle (Figure 5). NMR analysis of carbohydrate conformation thus concentrates upon determining the orientations about these dihedral angles. This is made difficult by the lack of conformational information available: even in the most favourable conditions, a maximum of around three NOEs are seen between each pair of residues in oligosaccharides, and observation of long-range correlations between different parts of the molecule is extremely unlikely. Interresidue correlations in NOESY or ROESY spectra can be used to determine distance information by making use of the fact that the intensity of the cross-peak is proportional to the inverse sixth

power of the distance between the two nuclei. The relatively inflexible ring geometry of pyranose rings allows intraresidue NOEs to be used to calibrate adjacent intraresidue correlations. When implementing such experiments it is important to ensure that the initial rate approximation holds. The motion of the molecule, both internal and overall, also affects the intensity of the NOE; thus interpretation of NOE values must take into account the motional model

Figure 5 A 1 → 6 linked disaccharide showing the dihedral angles I, \ and Z, about which conformational variation is greatest.

CARBOHYDRATES STUDIED BY NMR 179

used. Since it is now clear that the majority of carbohydrate structures are flexible to some degree, oscillating through an ensemble of related conformations, this can be a problem. Flexibility is understood to be particularly marked about 1 → 6 linkages owing to rotation about the additional dihedral angle, Z. The use of three-bond 3JCH couplings as conformational probes has been described, using the Tvaroska parametrization of the Karplus relationship:

where T is the HCOC torsion angle. However, in a similar situation to that described for NOE restraints, these analyses have been hampered by the extent to which these values are subject to conformational averaging. Because of these problems, angular information derived from 3JCH values is perhaps better used for evaluating the quality of conformational ensembles. Recent work has addressed the use of 13C within isotopically enriched compounds to derive additional conformational parameters, such as 13C 13C scalar couplings, and 13C 1H or 13C 13C NOEs. These approaches have yielded useful results, but have been hampered by the lack of a well-established Karplustype relationship for 13C 13C scalar coupling. Investigators have thus made considerable use of theoretical methods such as molecular mechanics to complement the limited experimental data available. Dynamics

The time-averaged nature of many NMR derived parameters, and the degree to which it affects oligosaccharide conformations has made it necessary to attempt to measure the degree of mobility within oligosaccharides. Since the relaxation of protonated carbons is almost entirely due to the directly attached proton it is possible to measure relaxation parameters from them that are not affected by variations in internuclear distance. Investigation of carbohydrate dynamics has been approached by molecular modelling in combination with the measurement of relaxation rates, including the T1, T2 of 13C and 1H, as well as 1H 1H and 13C 1H NOE values, to define the spectral density, given by

where Wc denotes the rotational correlation time.

These measurements have been interpreted by the LipariSzabo model-free approach, in which molecular motion is separated into overall and internal motions, related to the overall and internal correlation times WO and Wi:

where 1/W = 1/WO + 1/WO. S2 is a measure of the degree of internal reorientation, varying in value from 0 for isotropic reorientation to 1 for no internal motion. These measurements can then be used to derive dynamic models of carbohydrates and oligosaccharides. It is now generally accepted that carbohydrates are dynamic molecules with internal motions that occur on a time-scale faster than the overall motion of the molecules.

List of symbols J = coupling constant; J(Z) = spectral density function; T1, T2 = relaxation constants; 'FL = Huggins electronegativities; T = torsion angle; Wc, Wi and WO = rotational, internal and overall correlation times respectively; I, \ and Z = glycosidic torsion angles. See also: 13C NMR, Methods; Labelling Studies in Biochemistry Using NMR; Nuclear Overhauser Effect; Nucleic Acids Studied By NMR; Proteins Studied Using NMR Spectroscopy; Two-Dimensional NMR, Methods.

Further reading Bush CA (1996) Polysaccharides and complex oligosaccharides. In: Grant DM and Harris RK (eds) Encyclopaedia of Nuclear Magnetic Resonance, Vol 6, pp 37463750. Chichester: Wiley. Homans SW (1993) 1H NMR studies of oligosaccharides. In: Roberts GCK (ed) NMR of Macromolecules: a Practical Approach, pp 289314. Vol 134 of Rickwood D and Hames BD (series eds) Practical Approach Series. Oxford: Oxford University Press. Homans SW (1993) Conformation and dynamics of oligosaccharides in solution. Glycobiology, 3: 551 555. Hounsell EF (1995) 1H NMR in the structural and conformational analysis of oligosaccharides and glycoconjugates. Progress in Nuclear Magnetic Resonance Spectroscopy 27: 445474. Serianni AS (1992) Nuclear magnetic resonance approaches to oligosaccharide structure elucidation. In: Allen HJ

180 CELLS STUDIED BY NMR

and Kisalius EC (eds) Glycoconjugates: Composition, Structure and Function, pp 71102. New York: Marcel Dekker. Tvaroska I (1990) Dependence on saccharide conformation of the one-bond and three-bond coupling constants. Carbohydrate Research, 206: 5564. van Halbeek H (1994) NMR developments in structural studies of carbohydrates and their complexes. Current Opinion in Structural Biology 4: 697709.

van Halbeek H (1996) Carbohydrates and glycoconjugates. In: Grant DM and Harris RK (eds) Encyclopaedia of Nuclear Magnetic Resonance, Vol 2, pp 1107 1137. Chichester: Wiley. Vleigenthart JFG, Dorland L and van Halbeek H (1983) High resolution spectroscopy as a tool in the structural analysis of carbohydrates related to glycoproteins. Advances in Carbohydrate Chemistry and Biochemistry 41: 209374.

CD Spectroscopy of Biomacromolecules See

Biomacromolecular Applications of Circular Dichroism and ORD.

Introduction Nuclear magnetic resonance (NMR) methods have become available recently to study metabolism and morphology at the cellular level. The NMR methods may be classified as magnetic resonance spectroscopy (MRS), magnetic resonance imaging (MRI) and combinations of both, magnetic resonance spectroscopic imaging (MRSI). MRS provides information on the chemical composition of cells and its changes under specific circumstances, MRI yields X-ray like images of cellular anatomy and physiology and MRSI can study the spatial distribution of some metabolites within large cells or cellular aggregates. Notably, most of the nuclei participating in cellular reactions or some of their isotopes are NMR active. This allows a large variety of cellular functions to be monitored by different NMR methods. Table 1 summarizes the magnetic properties of the nuclei most commonly used in NMR studies of cells as well as the biological information which can be obtained.

MAGNETIC RESONANCE Applications

NMR studies of cells probably started in the mid 1950s analysing the dynamics of water in blood using 1H NMR. However, modern NMR studies of cellular metabolism began later with the introduction of commercially available high-field spectrometers and Fourier transform NMR techniques. Improve-ments in signal-to-noise ratios allowed, in the early 1970s, a study by 13C NMR of the metabolism of glucose in a suspension of yeast and to determine by 31P NMR the intracellular pH in erythrocyte suspensions. Since then studies of cellular metabolism by NMR methods have been used routinely in many laboratories. The development of cellular NMR is supported by some inherent advantages of the NMR method with respect to more classical approaches. First, the noninvasive character of NMR allows repetitive, noninvasive measurements of metabolic processes as they occur in their own intracellular environment. Second, the magnetic properties of nuclei like the relaxation times T1 and T2 or the homonuclear and

CELLS STUDIED BY NMR 181

Table 1

Nuclei 1

NMR properties and type of biological information provided by various nuclei

Frequency (MHz)

Spin

400

1/2

161.9

1/2

Natural Relative abundance (%) sensitivity

5.1 × 10–4

0.037

2.9 × 10–2

Water transport and metabolism

0.37

1.0 × 10–3

Nitrogen metabolism

99.63

1.0 × 10–3

Nitrogen metabolism

99.98

100

6.6 × 10–2

1.6 × 10–2

100.6

1/2

105.8

3/2

100

9.3 × 10–2

376.3

1/2

100

0.83

61.8

18.7

3/2

54.2

40.5

5/2 1/2

28.9

1.11

1.0

Information Metabolite concentration. Cell fingerprinting. Flow through some pathways. Intra and extracellular pH. Cellular volume. Microscopic imaging. MRI Concentration of phosphorylated metabolities. Bioenergetic status. Intra- and extracellular pH. Cellular volume. Phospholipid metabolism Quantitative measurements of metabolic flow through specific pathways Membrance potential. Cellular volume. Na+/H exchange Intra- and extracellular pH. Oxygenation state. Divalent metal ion concentration Lipid structure. Rates of hydration–dehydration reactions. Water flow and perfusion Membrane potential

0.016 93.1

9.7 × 10–3

heteronuclear spin coupling patterns, contain unique information on the physiological or pathological status of the cells and on the flux through specific metabolic pathways. Third, NMR methods allow the acquisition of images of the spatial distribution of water in sufficiently large single cells or in cellular aggregates, and it seems likely that this approach will be extended to other metabolites in the near future. Despite these advantages, the NMR method is not devoid of drawbacks. In particular, NMR is a relatively insensitive technique with a metabolite detection threshold of around 101 mM for in situ cells and 50 µM in extracts for moderate-field spectrometers. Thus, to be able to obtain useful NMR spectra with an adequate signal-to-noise ratio, the cell cultures need to be grown to densities similar to those found in tissues. Even if cell extracts are used, these must be prepared from a sufficiently large number of cells to generate, in the NMR tube, metabolite concentrations in excess of the lower threshold for detection. These demanding conditions require the use of specialized perifusion systems for in situ NMR studies or facilities for large-scale cell culture in work with cell extracts. In this article, we describe general procedures for cell culture compatible with NMR studies and illustrate the information that NMR methods can provide on cellular metabolism and morphology, with examples involving mainly the use of 31P, 13C and 1H NMR. Several reviews, some of them quoted in the Further reading section, have covered this topic previously.

Large-scale and high-density cell culture procedures compatible with NMR studies The use of cell cultures provides a powerful tool to understand the cellular and molecular mechanisms occurring in vivo. Different types of cell culture exist depending on the origin of the cells used in the preparation. Primary cultures are obtained after enzymatic, chemical or mechanical disaggregation of the original tissue. Cell lines, transformed cells or cell strains are obtained from primary cultures after several passages from transplantable tumours or by cellular cloning. The in vivo behaviour is more closely resembled by primary cultures, but these usually require complex preparation resulting in relatively small cellular yields. Cellular lines or strains, and cellular clones have the advantage of being homogenous cell populations that are easy to grow and maintain in large scale, but their behaviour is not as comparable to the in vivo situation. Figure 1 illustrates the general procedure used to obtain a primary culture. The tissue of interest is dissected from the appropriate organ obtained from embryonic, newborn or adult donors. The dissected tissue is then chopped and disaggregated, digested enzymatically and the resulting cells are washed and collected by sedimentation. Cell counting and assessment of viability are easily accomplished later by optical microscopy, using a haemocytometer for cell counting and the dye exclusion criteria for viability assessment. The next step consists of cell seeding.

182 CELLS STUDIED BY NMR

Figure 1

General steps in the preparation of primary cell cultures.

Figure 2 High density cell culture devices compatible with in situ NMR spectroscopy. The tubes containing the entrapped cells can normally be accommodated in commercial high-resolution probes and placed inside the magnet (indicated by N and S).

CELLS STUDIED BY NMR 183

Cells seeded normally attach to flat surfaces, grow, proliferate and differentiate until reaching confluence. Usually, NMR studies demand cell numbers in the range 107108 cells, a quantity requiring a large number of culture dishes or flasks. To further increase the surface of the culture it is possible to use rollers. These devices allow cells to grow in the inner surface of a rolling cylinder which contains a moderate amount of culture medium. Several cylinders can be maintained rolling simultaneously in a rack, increasing enormously the surface area of the culture and thus the number of confluent cells. The excellent monograph of Freshney provides a more detailed description of every cell culture procedure and optimized techniques for particular cells. There are two different kinds of experimental design in NMR studies of cells: (i) the study can be performed inside the magnet in real time, monitoring the metabolism of cells under in situ conditions; or (ii) extracts can be prepared from the cells after different incubation times and examined by NMR under high-resolution conditions (Figure 2). In the first case special precautions need to be taken to guarantee adequate nutrient supply, good oxygenation conditions and efficient removal of waste products from the NMR tube containing the cells. Different methods have been developed to fulfil these requirements. All of them involve a perifusion system adapted to the NMR tube which contains the cells immobilized or grown in different matrices. The perifusion system consists basically of a peristaltic pump, a thermostatically stabilized gas-exchange chamber or oxygenator and the NMR tube containing the cells. The peristaltic pump delivers fresh medium to the oxygenator and NMR tube and removes the used medium to a waste reservoir. The oxygenator is constructed of gas-permeable Silastic tubing equilibrated with an atmosphere of 95% O2 /5% CO2. The NMR tube contains the entrapped cell suspension. There are basically three types of cell entrapment protocol compatible with NMR studies: (i) polymer threads, (ii) microcarrier beads or (iii) hollow fibre bioreactors. Cells are entrapped in polymer threads (agarose or alginate) by mixing the cell suspension with a solution of the liquid polymer (above the gelling temperature) and extruding the mixture through a Teflon tube immersed in ice. The polymer solidifies at low temperature entrapping the cells in filaments which can be easily perifused. The advantage of this method lies in its simplicity, but on the other hand, cells are placed in a nonphysiological environment. Alternatively, anchorage dependent cells can be grown on the surface of solid or porous microcarrier beads of approximately 100 µM diameter. The beads can be made of different materials and these include

Figure 3 31P NMR spectrum (161.1 MHz) of C6 glioma cells grown in a hollow fibre bioreactor. Note the important contribution of the GPC and GPE resonances. Inset: intra- (in) and extracellular (ex) resonances of DMMP. From Gillies RJ, Galons JP, McGovern KA, Scherer PG, Lien YH, Job C, Ratcliff R et al. NMR in Biomedicine 6: 95–104. Copyright 1993 John Wiley & Sons Limited. Reproduced with permission.

collagen, polystyrene, polyacrylamide, gelatine, and Sephadex. This method has the advantage that cells can be treated as a suspension. However the density of cells may not be sufficiently high since most of the volume in the NMR tube is filled by the beads. The hollow fibre bioreactor approach consists of growing cells on the surface of capillary fibres which are gas and nutrient permeable. Cells are grown directly in a bioreactor adapted to the NMR tube where the experiment is performed, reaching higher density than with other methods. However, there are some technical difficulties in setting up this model properly and it is relatively expensive. For a detailed description of the bioreactor approach see the literature given in the Further reading section.

Applications 31P

31P

NMR of cells

NMR spectra of cells show a small number of resonances but contain crucial information on cellular biochemistry and physiology. Figure 3 depicts a representative 31P NMR spectrum obtained in situ from a culture of C6 glioma cells grown in a bioreactor. It is possible to distinguish clearly resonances from the β, α and γ phosphates of nucleotide triphosphates, phosphocreatine (PCr), phosphodiesters (glycerolphosphorylcholine, GPC, and glycerolphosphorylethanolamine, GPE), inorganic phosphate (Pi) and phosphomonoesters (PME). The resonances from the nucleotide triphosphates contain primarily (> 90%) the contribution of MgATP and are

184 CELLS STUDIED BY NMR

normally referred to as ATP resonances. The PME peak is a composite resonance containing mainly contributions from phosphocholine (PCho) and phosphoethanolamine (PE), as well as from sugar phosphates in a smaller proportion. 31P NMR spectra provide quantitative information on the relative concentrations of phosphorylated metabolites. For spectra acquired under fully relaxed conditions, the area of the 31P NMR resonance is proportional to the concentration of the metabolite. Absolute quantifications are more difficult to perform but can also be obtained if the concentration of one of the phosphorylated metabolites in the sample (normally ATP) is measured by an independent method, such as enzymatic analysis. Unfortunately, the resonances from the α and β phosphates of ADP overlap under in situ conditions with those from the α and γ phosphates of ATP, thus precluding direct quantification of ADP in situ. The concentration of ADP in situ can be estimated indirectly from the difference in area between the α ATP or γ ATP peaks and the area of the β ATP peak, which is almost exclusively derived from ATP. Probably, the main interest in 31P NMR spectroscopy lies in its ability to assess the bioenergetic status of the cell. This is determined by the relative rates of ATP-producing and -consuming processes according to Equation [1]

In most mammalian cells, ATP is produced aerobically by oxidative phosphorylation and hydrolysed to provide the energy needed to support biosynthetic activities, maintain transmembrane ion gradients and perform cellular work. The balance between production and consumption of ATP determines its steady-state level and therefore the net availability for energy. If oxygen and nutrients become limiting, ATP synthesis may proceed for a short time using PCr as a phosphate donor in reaction [2]

intracellular and extracellular pH (in Pi-containing media) from the chemical shift of the Pi resonance. This is because Pi has an apparent pKa of 6.7 and its chemical shift is pH-dependent in the physiological range. Normally the chemical shift of the Pi resonance is measured with respect to internal references like PCr or α-ATP, which are nontitratable in the physiological pH range, (pKa ≈ 4.5). The following expression can be used to measure intracellular pH from the chemical shift of Pi(δ ) measured with respect to the α-ATP resonance:

The values 11.26 and 13.38 represent the acidic and alkaline limits of the Pi titration curve chemical shifts with respect to the α-ATP resonance. Phosphonic acid derivatives, like methylene diphosphonic acid (MDPA) or dimethyl methylphosphonate (DMMP), resonating far from physiological phosphates (ca. 2030 ppm) can be used as an external reference for chemical shift and concentration calibrations. Sometimes these references cross the cell membrane and two different resonances are observed for the intracellular and extracellular compounds. In these cases it is possible to determine the relative cellular volume from the relative areas of the intra- and extracellular resonances (Figure 3 inset). 31P NMR spectra also contain resonances from crucial phospholipid metabolites. These include the phosphodiesters, glycerolphosphocholine and glycerolphosphoethanolamine and the phosphomonoesters, PCho and PE. Tumoural cells and in vivo tumours show increased PME resonances, often caused by increased PCho content. This characteristic of tumour cells has been proposed to contain diagnostic or even prognostic information on tumours. The metabolic reasons underlying the increase in PCho in tumour cells are not completely understood but seem to be caused by an increase in its synthesis rather than to an increase in the degradation of phosphatidylcholine. 13C

Under these conditions no significant change occurs in the ATP resonances but a net decrease in the intensity of PCr is observed. When the ATP-buffering capacity of the PCr-creatine system is exceeded, net ATP hydrolysis occurs, resulting in a net increase in Pi and acidification, as described by Equation [1]. Another important aspect of 31P NMR spectroscopy is that it allows the determination of

13C

NMR of cells

NMR spectroscopy allows the detection of resonances from 13C, the stable isotope of carbon with a magnetic moment. The natural abundance for 13C is only 1.1% of the total carbon and its magnetogyric ratio is approximately one-fourth of that of the proton. These two circumstances make 13C NMR a relatively insensitive technique. However, this sensitivity problem can be improved by using 13Cenriched substrates. The combination of 13C NMR detection

CELLS STUDIED BY NMR 185

and substrates selectively enriched in 13C have made it possible to follow in vivo and in vitro the activity of a large variety of metabolic pathways in cells and subcellular organelles. These include glycolysis and the pentose phosphate pathway, glycogen synthesis and degradation, gluconeogenesis, the tricarboxylic acid cycle, ketogenesis, ureogenesis and the glutamate, glutamine, γ-aminobutyric acid cycle in the brain among others. The design of 13C NMR experiments with selectively 13Cenriched substrates is similar to the classical radiolabelling experiments using radioactive 14C. A relevant difference is that 13C precursors are used in substrate amounts, while 14C substrates are normally used in tracer amounts. However, 13C NMR presents important advantages over 14C. First, the metabolism of the 13C-labelled substrate can be followed in real time, in situ and noninvasively. Second, even if tissue extracts are prepared, the detection of 13C labelling in the different carbon atoms of a metabolite does not require separation and carbonby-carbon degradation of the compound, two normal requirements in experiments with 14C. Finally, the analysis by 13C NMR of homonuclear spincoupling patterns and isotope effects allows the determination of whether two or more 13C atoms occupy contiguous positions in the same metabolite molecule, a possibility only available to 13C NMR methods. As a counterpart to these advantages, 13C NMR is significantly less sensitive than other conventional metabolic techniques like radioactive counting, mass spectrometry or optical methods. In particular, 13C NMR studies with cells in situ are difficult to find and most of the 13C NMR work has been performed with cell extracts. This allows operation under high-resolution conditions and increases significantly both the sensitivity and the resolution. Figures 4 and 5 illustrates the use of 13C NMR spectroscopy to study the metabolism of (1- 13C)glucose in primary cultures of neurons and astrocytes. A simplified scheme of the metabolism of (1- 13C)glucose in neural cells is given in Figure 4. Briefly, (113C)glucose is metabolized to (3- 13C)pyruvate through the Embden Meyerhoff glycolytic pathway. The (3- 13C)pyruvate produced can be transaminated to (3- 13C)alanine, reduced to (3- 13C)lactate or enter the tricarboxylic acid (TCA) cycle through the pyruvate dehydrogenase (PDH) or pyruvate carboxylase (PC) activities. A net increase in (3- 13C)lactate reveals increased aerobic glycolysis and is normally observed under hypoxic conditions in normal cells. If (313C)pyruvate enters the TCA cycle though PDH it produces (2-13C)acetyl-coenzyme A first, and subsequently (4- 13C)α-ketoglutarate. In contrast, if (313C)pyruvate is carboxylated to (3- 13C)oxalacetate by

PC, it will produce (2-13C)α-ketoglutarate in the next turn of the cycle. The α-ketoglutarateglutamate exchange is many times faster than the TCA cycle flux and therefore, 13C labelling in the glutamate carbon atoms reflects accurately the labelling in the α-ketoglutarate precursor. Thus, (4- 13C)glutamate or (2-13C)glutamate are produced from (1- 13C)glucose depending on the route of entry of the (3- 13C) pyruvate into the TCA cycle. Therefore, a comparison of intensities from the C4 and C2 glutamate resonances provides a qualitative estimation of the relative contribution of PDH and PC to the TCA cycle. Figure 5 shows proton decoupled 13C NMR spectra of perchloric acid extracts obtained from primary cultures of neurons (upper panel) and astrocytes (lower panel) incubated for 24 h with 5 mM (1- 13C)glucose. The most relevant resonances are the ones derived from the C3, C4 and C2 carbons of glutamate; C3, C4 and C2 carbons of glutamine (or glutathione); C3 carbon of alanine; C3 carbon of lactate; C6 carbon of N-acetylaspartic acid and C2 carbon of acetate (see Figure 5 legend for resonance assignments). The spectra of Figure 5 reveal important differences in the metabolism of (1- 13C)glucose by neurons and astrocytes. Briefly, the lactate C3 resonance is higher in the spectrum of astrocytes than in that of the neurons. The glutamate C4 resonance is significantly larger in neurons than in astrocytes, while the C2 glutamine resonance in astrocytes is higher than in neurons. Taken together these results suggest that astrocytes metabolize glucose mainly through aerobic glycolysis having an important contribution from PC, while neurons metabolize glucose mainly through the TCA cycle, using PDH as the main route. It is possible to obtain in cells quantitative determinations of flux through specific metabolic pathways or through specific steps of a pathway using 13C NMR. There are two possible strategies, both based on the principles of isotopic dilution and both requiring the use of mathematical models of metabolism. The first strategy allows the determination of fluxes by solving linear systems of equations relating flux to fractional 13C enrichments in specific carbons of precursors and products. The second strategy involves the determination of relative isotopomer populations from an analysis of the 13C13C spin coupling patterns. For a more precise description of both methods the reader is referred to the article of Portais and co-workers and Künnecke and co-workers in the Further reading section. 1H

NMR of cells

NMR has an inherently higher sensitivity than 31P NMR and thus a larger number of metabolites or 13C

186 CELLS STUDIED BY NMR

Figure 4 Metabolism of (1-13C) glucose in neural cells. HMP: hexose monophosphate shunt. TCA: tricarboxylic acid cycle; OAA: oxalacetate; α-KG: 2-oxoglutarate; GABA: γ−aminobutyric acid; Ac CoA: acetyl coenzyme A. Only one turn of the TCA is considered for clarity. Filled or open circles in TCA cycle intermediates indicate labelling from PDH or PC, respectively.

is potentially detectable. However, two limitations need to be considered: (i) the water signals from the cells need to be eliminated and (ii) the complex

homonuclear spin coupling patterns due to overlapping proton resonances have to be resolved. Both drawbacks have been overcome in part by the

CELLS STUDIED BY NMR 187

Figure 5 Proton decoupled 13C NMR spectrum (125.7 MHz) of neutralized perchloric acid extracts obtained after the incubation of primary cultures of neurons (upper trace) or astrocytes (lower trace) with 5 mM (1-13C) glucose during 24 h. 1: alanine C3, 2: lactate C3, 3: N-acetylaspartic C6, 4: acetate C2, 5: glutamine C3, 6: glutamate C3, 7: glutamine C4, 8: glutamate C4, 9: glutamine C2, 10: glutamate C2. An artificial line broadening of 1 Hz was used in both cases.

development of water suppression techniques and different one-, two- or n-dimensional editing methods. The information provided by 1H NMR spectra of cell suspensions is illustrated in Figure 6 which shows a representative 1H NMR spectrum of a suspension of rat erythrocytes acquired using a simple one-pulse sequence (Figure 6B). The most prominent resonances derive from the H3 proton of lactate, the choline protons of ergothioneine and the glutamate and glutamine resonances from glutathione. All these resonances are located on top of a much broader resonance arising from low-mobility lipids. Therefore 1H NMR spectroscopy allows lactate production to be followed easily. This is normally done using spin echo sequences which eliminate the broad lipid component because of its short T2 (Figure 6A). In addition to lactate, 1H NMR normally allows the direct quantification of creatine singlets, singlets from choline-containing compounds and many other metabolites. Sometimes, quantification of these compounds requires very high resolution conditions, high spectrometer observation frequencies and computerized deconvolution

algorithms. Indeed the 1H NMR spectrum provides a true fingerprint of the metabolites present in the cell and it appears to be characteristic for every cell type. This has been clearly shown in studies of cell typing with neural cells. Furthermore, it has been shown recently that fully transformed cells present an increase in the 1H resonance of the PCho peak, and that the ratio PCho to GPC may serve as an indicator of multiple oncogene lesions. A more recently developed application of 1H NMR to cellular studies relies on the noninvasive determination of intra- and extracellular pH. The procedure is based in the use of a probe containing a reporter proton with a chemical shift sensitive to pH in the physiological range. This method is inherently more sensitive than previous 31P NMR approaches. Figure 6A illustrates more clearly this procedure, by showing the spectrum of an erythrocyte suspension after the addition of the extrinsic pH probe imidazol1-yl acetic acid methyl ester. The imidazole protons, H2 (ca. 8.2 ppm) and H4, H5 (ca. 7.25 ppm) are clearly observed in the aromatic region of the spectrum between 6 and 9 ppm (inset to Figure 6A). Two

188 CELLS STUDIED BY NMR

Figure 7 T2-weighted spin echo microscopic 1H NMR images (360.13 MHz) of an isolated neuron from Aplysia californica before (A) and after (B) hypotonic shock. Resolution (x,y,z) is 20 µm × 20 µm × 150 µm. Artificial sea water appears as a bright background surrounding the cell. T2 differences between nucleus and cytoplasm cause these structures to appear as bright and dark, respectively. Reproduced with permission of the American Physiological Society from Hsu EW, Aiken NR, Blackband SJ (1996) American Journal of Physiology 271: C1895–C1900.

Figure 6 1H NMR spectra (360.13 MHz) of a rat erythrocyte suspension in the presence of a 1H NMR pH probe. The water resonance (not shown) was attenuated with a 2 s presaturating pulse. (A) Spin echo acquisition with 5 ms interpulse delay. Inset: expansion of the aromatic region. (B) Conventional one-pulse acquisition. Inset: Chemical structure of imidazole-1-yl acetic acid methyl ester, the pH probe used. a: TSP, b: lactate H3, c: glutamate H3,H3′ in glutathione, d: aspartate H3,H3′, e: creatine (methyl), f: trimethyl groups in choline, carnitine and ergothioneine, g: methanol, h: H2 of amino acids, i: lactate H2, k: H4 and H5 of the pH probe, l: H2 of the pH probe. H2i: intracellular probe, H2o: extracellular probe. Methanol is produced by endogenous esterases immediately after the addition of the probe.

signals for the imidazole H2 proton (H2 i and H2 o) with different chemical shifts are observed. These signals have been shown to be derived from molecules of the probe located in the intra- (H2 i) and extracellular (H2 o) space. Thus the different chemical shift of intra- and extracellular probes allows the determination of the intra- and extracellular pH values as well as the transmembrane pH gradient. Finally, impressive progress in NMR microscopy methods has allowed the acquisition of 1H NMR images of single cells. T2-weighted NMR images from single cells normally show a dark cytoplasm and relatively bright nucleus. Figure 7 provides an illustrative example of cellular NMR imaging, showing the morphology of a neuron from Aplysia californica before (Figure 7A) and after (Figure 7B) hypotonic shock. The hypotonic shock increased the relaxation times T1 and T2 of cytoplasm and nucleus,

but did not change the apparent diffusion coefficient of water. In conclusion, 1H NMR methods provide information on glycolysis, fingerprint cell typing, state of proliferation, intra- and extracellular pH and cellular micromorphology. Representative articles describing these 1H NMR applications are given in the Further reading section. Other nuclei

In addition to the most commonly used nuclei, such as 31P, 13C and 1H, other nuclei can be used in NMR studies of cellular metabolism (Table 1). The 19F nucleus has a magnetic moment slightly smaller than 1H, whereas its chemical shift range is larger by a factor of approximately 20. No contributions from tissue background signals are observed, which makes 19F a useful magnetic label for metabolic studies. Using difluoromethylalanine or fluorinated derivatives of BAPTA, 19F NMR spectroscopy has been extensively used to measure intracellular pH as well as cytosolic free Ca2+ and free Mg2+ in peripheral blood lymphocytes and other cells. The 23Na nucleus is a quadrupolar spin 3/2 nucleus with approximately 100% natural abundance. Because of its quadrupolar nature, some of the 23Na is invisible to the NMR experiment. However, it has been possible to resolve intra- and extracellular 23Na resonances in yeast cells and erythrocytes and to monitor Na+H+ exchange in the latter cells. The 17O nucleus has an unfavourable gyromagnetic ratio but a large chemical shift range, close to 1000 ppm. 17O NMR has been used extensively to study water metabolism in erythrocytes and subcellular organelles.

CELLS STUDIED BY NMR 189

Finally, 2H is one of the most recent additions to the repertoire of cellular NMR methods. 2H has a natural abundance of only 0.02% and therefore the use of 2H NMR requires the use of selectively enriched substrates. 2H has been mainly used in cells to investigate the structure in situ of choline lipids in the cell membrane.

List of symbols ADP = adenosine 5′-diphosphate; ATP = adenosine 5′-triphosphate; BAPTA = 1,2-bis(2-aminophenoxy) ethane-N,N,N′,N′-tetraacetic acid; CoA = coenzymeA; CDP = cytidine 5′-diphosphate; DMMP = dimethyl methylphosphonate; GPC = glycerolphosphocholine; GPE = glycerolphosphoethanolamine; MDPA = methylene diphosphonic acid; MRI = magnetic resonance imaging; MRS = magnetic resonance spectroscopy; MRSI = magnetic resonance spectroscopic imaging; PCho = phosphocholine; PE = phosphoethanolamine; PC = pyruvate carboxylase; PCr = phosphocreatine, PDH = pyruvate dehydrogenase; Pi = inorganic phosphate; PME = phosphomonoesters; TCA = tricarboxylic acid cycle; TSP = 2,2 ′,3,3 ′-tetradeutero trimethylsilyl propionate; T1 and T2 = relaxation times. See also: Biofluids Studied By NMR; Carbohydrates Studied By NMR; In Vivo NMR, Applications, 31P; In Vivo NMR, Methods; NMR Microscopy; Nucleic Acids Studied Using NMR; Perfused Organs Studied Using NMR Spectroscopy; Solvent Supression Methods in NMR Spectroscopy; Two-Dimensional NMR Methods.

Further reading Bhakoo KK, Williams SR, Florian CL, Land H and Noble MD (1996) Immortalization and transformation are associated with specific alterations in choline metabolism. Cancer Research 56: 46304635.

Freshney RI (1994) In: Freshney RI (ed) Culture of Animal Cells: a Manual of Basic Technique. New York: WileyLiss. Gil S, Zaderenko P, Cruz F, Cerdan S and Ballesteros P (1994) Imidazol-1-ylalkanoic acids as extrinsic 1H NMR probes for the determination of intracellular pH, extracellular pH and cell volume. BioOrganic and Medicinal Chemistry 2: 305314. Gillies RJ, Galons JP, McGovern KA, Scherer PG, Lien YH, Job C, Ratcliff R, Chapa F, Cerdan S, Dale BE (1993) Design and application of NMR-compatible bioreactor circuits for extended perifusion of high-density mammalian cell cultures. NMR in Biomedicine 6: 95104. Hsu EW, Aiken NR, Blackband SJ (1996) Nuclear magnetic resonance microscopy of single neurons under hypotonic perturbation. American Journal of Physiology 271 (Cell Physiol 40): C1895C1900. King GF and Kuchel PW (1994) Theoretical and practical aspects of NMR studies of cells. Immunomethods 4: 8597. Künnecke B, Cerdan S and Seelig J (1993) Cerebral metabolism of (1,2-13C2)glucose and (U-13C4)3-hydroxybutyrate in rat brain, as detected by 13C spectroscopy. NMR in Biomedicine 6: 264277. Lundberg P, Harmsen E, Ho C, and Vogel HJ (1990) Nuclear magnetic resonance studies of cellular metabolism. Analytical Biochemistry 191: 193222. Portais JC, Schuster R, Merle R, Canioni P (1993) Metabolic flux determination in C6 glioma cells using carbon-13 distribution upon (1-13C)glucose incubation. European Journal of Biochemistry 217: 457468. Szwergold BS (1992) NMR spectroscopy of cells. Annual Review of Physiology 54: 775798. Urenjak J, Williams SR, Gadian DG and Noble M (1993) Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. Journal of Neuroscience 13: 981989. Vogel HJ, Brodelius P, Lilja H and Lohmeier-Vogel EM (1987) Nuclear magnetic resonance studies of immobilized cells. In Mosbach K (eds.), Methods in Enzymology, vol. 135, B: pp. 512536. New York: Academic Press.

190 CHEMICAL APPLICATIONS OF EPR

Introduction EPR spectroscopy is a technique that affords a means for the detection and quantification of paramagnetism, i.e. the presence of unpaired electrons. It is applicable to solids, liquids or even gases. A compound or molecular fragment containing an odd number of electrons is paramagnetic. In this article we give examples to show the power and breadth of the technique. The reader is advised, however, to read the Encyclopedia article on the theoretical aspects of EPR to appreciate fully the strength of the technique and also the excellent series Specialist Periodical Reports on Electron Spin Resonance, published by the Royal Society of Chemistry, which has now reached volume 16. The technique can be applied to obtain information on the following four areas. 1. How atoms are bonded together in molecules (structure). 2. The route by which compounds are transformed from one to another, i.e. which bonds break and which are formed (mechanism). 3. The rate at which these processes occur (kinetics). 4. The molecular interactions that exist, for example, between solvent and solute (environment). Knowledge of the g values and the detailed hyperfine interactions (A values) allow us to help identify radical species, and these parameters contain information about the electron distribution within the molecule. Radicals are often present as intermediates during a reaction; consequently their identification will give information concerning the reaction mechanism and measurement of how their concentration changes with time will give kinetic data. Let us now look in greater detail at several major applications of the technique.

Organic and organometallic free radicals; structure, kinetics and mechanism There are many many examples of the power of the technique in this area and space does not permit us

MAGNETIC RESONANCE Applications to cover fully all examples. Articles by Tabner and by Rhodes fully cover the basic principles and practical aspects for the detection and evaluation of radical anions and cations (see Further reading) while more recently Davies and Gescheidt have reviewed the literature up to 1996 in the Specialist Reports referred to earlier. In the same publication, Alberti and Hudson cover the topics of organic and organometallic free radicals up to 1995. Consequently we will select just one or two examples from these areas. An early method, widely used for radical generation is based on the redox system Ti3+H2O2. This was first used by Dixon and Norman to generate HO and HOO radicals obtained by rapid mixing of aqueous solutions of TiCl3 and H2O2 with a substrate immediately before the EPR cavity. It has been suggested that the main reacting species in such a system is TiOO3+. The EPR signals obtained from such a system depend upon such factors as flow rate, temperature, pH and the ratio of the two components. Flow systems such as this have been used quite successfully to study polymerization reactions that are initiated by redox reactions. Analysis of the hyperfine interactions of the reacting species enabled the structure of the radicals to be determined. The use of such flow and/or stopped flow systems has been used quite extensively to investigate the kinetics of radical reactions. These include the use of a stopped flow EPR system to investigate the oneelectron oxidation of a range of phenols by a dimeric manganese(IV/IV) triazacyclononane complex with and without hydrogen peroxide. Another use has been the development of a technique of photolytic time-resolved EPR that allows kinetic data to be obtained with relative ease. The method involves flowing oxygen-free solutions of the materials under investigation through a flat reaction cell placed within the EPR cavity. The flow rates of the solutions could be adjusted so that signal intensities were not dependent on their cavity dwell time. The solutions could be irradiated either continuously or intermittently and thermostatted by the use of a stream of dinitrogen. Signal averaging is used to overcome signal-to-noise difficulties.

CHEMICAL APPLICATIONS OF EPR 191

Spin trapping, spin labels and spin probes in polymers and biological systems The role of free radicals in biological and medicinal systems is of considerable interest as they have been implicated in a wide range of diseases such as cancer. In such systems, the direct detection of free radicals is often not possible because of their low steady-state concentration arising from their high reactivity and transient nature. In this instance, the technique of spin trapping is used. Spin trapping involves the addition of a diamagnetic molecule, usually a nitrone or nitroso compound, which reacts with a free radical to give a stable paramagnetic spin adduct that accumulates until it becomes observable by EPR (Eqn [1]).

By determination of the parameters of the spin adduct spectrum it is often possible to identify the nature of the primary trapped radical, or at least to determine the type of radical trapped. Table 1 gives examples of the range of hyperfine coupling constants obtained for a variety of trapped species for the two most commonly used spin traps DMPO (dimethylpyridine N-oxide) and PBN (α-phenyl-N-t-butyl nitrone). Care has to be taken, however, because the coupling constants vary with solvent polarity (for example, in water the DMPO adduct of the t-butoxy radical has a(N) = 1.48 and a(H) = 1.60 mT while in toluene a(N) = 1.3 and a(H) = 0.75 mT). The spin trap must react at relatively fast trapping rates and the radical adduct must have a reasonable half-life; for example, DMPO has a ktrapping for OH of 2 × 10 9 M1 s1 and half-life of about 15 min, whereas for O2, ktrapping is 1 × 101 M1 s1 and the half-life is about 90 s. Consequently the DMPOO 2adduct is rarely seen. This difficulty has been overcome by the use of phosphorus-substituted DMPO, in particular where one of the methyl groups in the 5 position has been replaced by a diethoxyphosphoryl group. This increases the half-life of the OOH adduct considerably in both organic and aqueous solvents, with the difference being more pronounced in aqueous solvents. One other advantage of the use of phosphorus-substituted DMPO spin traps is that the β-31P hyperfine coupling constant has a much larger variation (2.55.5 mT) than the β-H hyperfine

Table 1 Hyperfine coupling constants (mT) for a variety of spin-trapped adducts

PBN Free radical

·H · Phenyl · CCl · OH O2–

DMPO H

1.57

2.5

1.53

0.82(2H)

1.60

0.44

1.39

0.175

1.54

0.27

1.50

1.43

0.225

1.45

1.16

But O

1.48

1.60

SO3–

1.47

1.60

PhS

1.29

1.415

coupling constant (0.62.5 mT), thus allowing an easier identification of the trapped radical. Table 2 gives a range of bimolecular rate constants with both PBN and DMPO for a variety of radical species. In a study of heterogeneous catalysis, the spin trapping technique has been used to prove the presence of radical species on a catalyst surface. In a study of a palladium metal catalyst supported on alumina, it was shown that hydrogen is dissociatively chemisorbed by trapping hydrogen atoms with PBN. Alkyl and aromatic free radicals were also shown to be present when other feed stocks were used. In a separate study of the photodecomposition of acetaldehyde, it was shown that high-power irradiation (600 W mercury/xenon lamp) generated a different PBN adduct from that given with irradiation under direct sunlight and that the presence of oxygen played a major role in the photochemistry. Table 2 Second-order rate constants for radical spin trapping for the traps PBN and DMPO

Spin trap PBN

T( qC)

· · Ph· HO· Me

4 × 106

RC H2

1.3 × 105

2 × 10 7

2 × 10 9

1 × 10 2

2 × 10 9

1 × 101

2.6 × 106

ROO DMPO

Rate constant (M –1s –1)

–

RC H2

192 CHEMICAL APPLICATIONS OF EPR

Since the first reported study on nitroxide spin labelling in 1965, the technique has found widespread application to biological and chemical problems. The introduction of a small free radical, the spin label, onto the molecule of interest gives environmental information at that point. It has the ability to measure very rapid molecular motion and is usually free from interference, thus making it a very powerful technique that gives useful details about dynamic processes at the molecular level. The types of free radical mostly used as spin probes or spin labels are based on nitroxides, such as the label known as TEMPOL (4-hydroxy-2,2,6,6-tetramethylpiperidinooxy). Figure 1 shows the change in the EPR spectrum of a spin-labelled system going from a fully mobile isotropic system (Figure 1C) to a fully immobilized glass type spectrum (Figure 1A). Measurement of the hyperfine parameters enables information to be obtained about the mobility or rigidity of the system under study. The spin probe technique has been used to investigate the interactions of the anionic surfactant SDS and gelatin. It was found that for simple SDS solutions the rotational correlation time of the spin probe, 16-doxylstearic acid methyl ester, increased slightly with increased surfactant concentration. In

contrast, in the presence of gelatin these properties varied markedly as a function of the stoichiometric ratio of the surfactant to gelatin concentration, with the correlation time decreasing as the hyperfine coupling constant increased with increasing surfactant concentration. Such behaviour was attributed to the characteristics of the various amino acids present in the gelatin. The use of spin labelling in polymer systems has enabled information to be obtained about polymer chain size and also their conformational state, with three main states being distinguished. The ability of some nitroxide radicals to diffuse easily into various polymer systems, coupled with the fact that their EPR spectra showing three lines caused by coupling to 14N show a high degree of rotational motion in the amorphous parts of the polymer, means that studies varying the temperature through the glass transition temperature Tg will enable changes to be measured in the rotational correlation time. Since the rotational correlation time obeys the Arrhenius law (Eqn [2]),

which is valid for T > Tg, values for the rotational activation energy of the nitroxide free radical can be measured. Since this does not depend on radical size but only on the polymer matrix, a measure of polymer mobility is obtained. There are several excellent reviews on the application of the method to studies in polymer systems and also to investigations of biological membranes. These are listed in the Further reading section.

Study of transition metal compounds

Figure 1 EPR spectra arising from a 14N (I = 1) showing (A) randomly and rigidly oriented in a frozen matrix; (B) antisotropic motion in a randomly oriented environment; (C) isotropic spectrum with no restriction on movement.

EPR spectroscopy is widely used to study the structure and geometry of d-transition metal ions (TMI) in inorganic complexes, in biological systems and in catalysts. Information obtained through analysis of the spectrum varies from simple identification of the metal centre to a thorough description on the electronic structure of the complex. At a simple level, the main interactions experienced by the TMI that consequently influence the EPR spectrum are (i) the electronic Zeeman effect and (ii) interaction between the electron and the nuclear spin. In the first case, the interaction is expressed via the g tensor which carries information on electronic structure; in the second case, the metal hyperfine interaction is expressed via the A tensor. Although the information regarding the transition metal compound can be comprehensive, we shall

CHEMICAL APPLICATIONS OF EPR 193

confine ourselves here to a simple illustration on the power of EPR for investigating the symmetry of the metal centre in the solid state. The simplest case arises for TMI with one d electron. For a metal in a tetragonally distorted octahedral field experiencing compression along the 4-fold axis (D4h), the five d orbitals have the splittings shown in Figure 2 with the nondegenerate dyz state lowest in energy. For a slight distortion, the low-lying 2-fold degenerate excited states (dxy, dxz) are close to the ground state, as depicted in Figure 2A. The small energy separation '1 results in a short relaxation time with the result that the EPR signal can only be observed at low temperatures. Greater distortions lead to longer relaxation times, so that the EPR signals become observable at increasingly higher temperatures with gA > g|| as shown in Figure 2A. In the case of elongation along the 4-fold axis, the unpaired electron now

Figure 2

occupies a doubly degenerate ground state (dxz, dyz) and no EPR signal is observed. However, if the ion experiences a trigonal distortion with a resulting point symmetry of D3d as shown in Figure 2B, then the single electron occupies the dz orbital so the order of the g-values changes or becomes inverted (i.e. g|| > gA) as depicted in Figure 2B. This simple example illustrates the effects of the ground state of the TMI on the resulting EPR spectrum. As shown above, the point symmetry at the metal can therefore be reflected in the g values, but the A values can also be influenced in a similar manner. Furthermore, depending on the point symmetry, the principal axes of g and A may or may not be coincident. The relationship between these tensors, the EPR spectral characteristics and the point symmetry of the metal are shown in Table 3. The importance of these relationships is that each type of EPR 2

Orbitals and corresponding EPR spectrum for an octahedral complex with (A) tetragonal and (B) trigonal distortion.

194 CHEMICAL APPLICATIONS OF EPR

Table 3 Relationship between g and A tensor, EPR symmetry and the point symmetry of the paramagnets

EPR symmetry

Coincidence g and A tensors of tensor axis

Isotropic

gxx = gyy = gzz Axx = Ayy = Azz

Axial

gxx = gyy ≠ gzz Axx = Ayy ≠ Azz

Rhombic

Molecular point symmetry

All coincident

Oh Td O, Th, T

All coincident

D4h, C4v, D4, D2d, D6h,C6v, D6, D3h, D3d, C3v, D3

gxx ≠ gyy ≠ gzz Axx ≠ Ayy ≠ Azz

All coincident

D2h, C2v, D2

Monoclinic gxx ≠ gyy ≠ gzz Axx ≠ Ayy ≠ Azz

One axis of g and A coincident

C2h, Cs, C2

Triclinic

gxx ≠ gyy ≠ gzz Axx ≠ Ayy ≠ Azz

Complete non- C1, C1 coincidence

Axial noncollinear

gxx ≠ gyy ≠ gzz Axx = Ayy ≠ Azz

Only gzz and Azz coincident

C3, S6, C4, S4, C4h, C6, C3h, C6h

behaviour is associated with a restricted number of point symmetries, which places constraints upon the geometrical structures of the paramagnet. For example, if the paramagnet is known to have, say, rhombic symmetry, then the associated geometry must belong to one of the point groups D2h, C2v or D2 (Table 3). It would be incorrect to assign a structure that belongs to a more symmetric arrangement, e.g. D4h which is strictly axial. Therefore, for a system with unknown structure, if the EPR symmetry can be determined (Table 3) we have invaluable structural information since the paramagnet can only belong to a restricted range of point symmetries. As an example, consider the EPR spectrum of H2[V(R,RHIDPA)2] diluted in the corresponding zirconium compound (Figure 3B). At first glance the spectrum appears to have axial symmetry, since the features associated with the x and y directions are not split. Simulation of a spectrum based purely on axial symmetry is not entirely satisfactory (Figure 3A). The introduction of a small anisotropy in both gxx, gyy and Axx, Ayy, however, produces an improved simulation (Figure 3C). Although the rhombicity is small, it is consistent with the known structure of the compound. From the crystallographic data, the highest possible point symmetry at the vanadium is C2. In more complex cases, other techniques such as variable-frequency CW-EPR, ENDOR and ESEEM are required to provide a more detailed description on the electronic structure of transition metal complexes. Nevertheless, one can start to appreciate the power of CW-EPR for studies of paramagnetic transition metal complexes.

Figure 3 Experimental and simulated X-band powder EPR spectra of H2[V(R,R-HIDPA)2] diluted in the corresponding zirconium compound, where HIDPA = hydroxyiminodipropionate. (A) Simulated spectrum with axial symmetry and g|| = 1.9195, gA = 1.9839, A|| = 17.5 mT and AA = 4.9 mT; (B) experimental spectrum; and (C) simulated spectrum with rhombic symmetry gzz = 1.9195, gxx = 1.9848, gyy = 1.9829, Azz = 17.15 mT, Axx = 4.6 mT and Ayy = 5.2 mT. Reproduced with permission of the Royal Society of Chemistry from Mabbs FE (1993) Some aspects of the electron paramagnetic resonance spectroscopy of d-transition metal compounds. Chemical Society Reviews 22: 313–324.

Surface chemistry and catalysis EPR has been widely applied to surface science and catalysis in order to examine a variety of surface paramagnetic species important in catalytic processes including adsorbed atoms, ions or molecules that may be intermediates in chemical reactions, intrinsic surface defects, transition metal ions supported on an oxide and spin labels interacting with a surface. Information regarding the important physicochemical characteristics of the surface can be gained

CHEMICAL APPLICATIONS OF EPR 195

through analysis of the EPR spectrum. Properties such as surface crystal field, surface redox properties, surface group morphology, mobility of adsorbed species, coordination of surface metal ions and identification of the active site have been studied on a variety of surfaces. In many cases, the surface itself is diamagnetic and investigations by EPR rely on indirect methods. A large amount of information regarding the state of a surface can be gained by adsorption of probe molecules (i.e. a molecule whose properties in the adsorbed state can be monitored and evaluated by EPR). These molecules may be divided roughly into two classes. The first is paramagnetic probes (or probes that become paramagnetic upon adsorption). The spectroscopic features of simple paramagnetic probes such as NO and O2 can provide detailed information on surface electrostatic fields. For example, the C values of the adsorbed superoxide anion (O2) are known to depend not only on the nominal charge of the surface cation where adsorption occurs but also on the coordination environment of the cation itself, as shown in Figure 4. Paramagnetic probes such as VCl4 or MoCl5 retain their paramagnetism during reaction with surface hydroxyl groups, allowing one to study the distribution of the surface groups themselves. The other type are diamagnetic probes that remain diamagnetic upon adsorption. These probes allow one to investigate the properties of a paramagnetic adsorption site, such as the changes in the coordination sphere of a surface transition metal ion. Metal oxides are frequently involved in catalysis, either directly or indirectly as supports. By studying the charge transfer reactions at the gassolid or liquidsolid interface, the nature, number and strength of the acidic or basic catalyst can be investigated. The oxidizing properties of the oxide can be studied by formation of the radical cations of aromatic molecules such as anthracene and perylene, while the reducing properties of the surface can be monitored through the formation of the radical anions of molecules with large electron affinity such as tetracyanoethylene (TCNE) or nitroaromatic compounds. The EPR spectra of the corresponding radical cations and anions yield both qualitative and quantitative information on the redox nature of the surface and the subject has been reviewed. Finally, a molecular approach to catalysis requires a deep understanding of the coordination processes occurring at the catalyst surface. The use of EPR to investigate the various types of coordination chemistry that occur on catalytic systems involving transition metal ions and oxide matrices have been

Figure 4 Simulated EPR spectra of the superoxide (O2–) anion adsorbed on differently charged cationic sites. The gzz values of the radical have the form gzz = ge 2λ/∆ where λ is the spin–orbit coupling constant for oxygen and ∆ is the extent of the energy splitting between the two π* orbitals, caused by the local electrostatic field of the cation. As the charge varies, ∆ will vary and different gzz values are observed (A–D). The anion is also sensitive enough to distinguish between cations with the same nominal charge which, because of different coordinative environments (e.g. 5-coordinated, 4C or 3C surface cations), exert slightly different electrostatic fields (E).

reviewed. It has been shown that the reactivity and properties of the central transition metal ion in the heterogeneous system are heavily influenced by the coordinating ligands (much more than in the homogeneous counterparts) so that a variety of TMI complexes can be created and stabilized at the surface. Changes to the EPR spectroscopic features as a function of variation in the coordination sphere of a surface-supported TMI can easily be followed. An example of this is shown in Figure 5 for a Ni+ supported catalyst. Changes in the spectroscopic features of the signal as a function of differing CO pressures illustrate the sensitivity of the technique to perturbations in the local environment. This coordination chemistry approach to EPR studies of the mechanism of catalytic reactions is illustrated in the dimerization of olefins and the oxidation of methanol on the surface of oxide catalysts.

196 CHEMICAL APPLICATIONS OF EPR

Figure 5 X-band EPR spectra of Ni+ complexes formed upon CO adsorption on reduced Ni/SiO2 catalysts under a pressure of (A) 10 torr CO followed by evacuation at 340 K, (B) 10 torr, (C) 100 torr, (D) 400 torr. Subscript a represents axial and subscript e represents equitorial positions. Reprinted with permission from Dyrek K and Che M (1997) EPR as a tool to investigate the transition metal chemistry on oxide surfaces. Chemical Reviews 97: 303–331 Copyright 1997 American Chemical Society.

Applications of pulsed techniques and high-frequency EPR The applications discussed above were performed using conventional CW (continuous wave) methods, predominantly at X-band (∼ 9 GHz) frequency. However, in recent years EPR spectroscopy has experienced an extraordinarily rapid development in two noted areas, namely pulsed techniques and highfrequency EPR. Already both developments have proved to be invaluable tools in important areas of physics, chemistry and biology, and no article on the applications of EPR spectroscopy is complete without an explanation of the powerful attributes of these techniques. The difference between high-frequency and lowfrequency EPR is usually defined by the magnet system. Superconducting or Bitter magnets are employed for high-frequency EPR (such as those operating at 94 GHz), whereas traditional electromagnets are used for lower frequencies (e.g. Q-band

(∼35 GHz) to L-band (∼1 GHz)). High-frequency EPR offers numerous advantages over the low-frequency spectrometers, including increased resolution of g factors (which is particularly important in complex or disordered systems where spectral lines are not resolved at lower fields) and increased absolute sensitivity (which is important in studies of small single crystals or any system where the number of spins is inherently low). High-frequency EPR also allows measurement of a number of integer-spin systems (such as Ni(II) species) which often have very high zero-field splittings. These systems only start to become resolved at high frequency where the energy of the photon is above the zero-field energy. The number of applications is growing rapidly, but the most compelling areas of development are in complex disordered systems prevalent in structural biology, including new insights into the detailed mechanisms associated with photosynthesis and many metalloproteins. Because EPR is only sensitive to the sites with unpaired spins, it targets only the most

CHEMICAL APPLICATIONS OF EPR 197

important bonds that largely determine the chemical and physical properties of the metalloprotein. Highfrequency EPR is also particularly appropriate for studying these complex enzymes since the active sites often involve metal ions or metal clusters with high zero-field splittings, which are EPR-silent at low frequency. Furthermore, most disordered systems are better studied at high frequencies because the improved spectral resolution allows the local structure to be inferred, especially in conjunction with other techniques such as ENDOR. Exploitation of pulsed techniques in chemical research has been hampered in the past by expensive instrumentation and the lack of sufficiently fast digital electronics. The situation has undoubtedly changed in recent years, and the various pulsed EPR methods have found increased or exclusive use in areas such as time-resolved EPR, as tools for studying molecular motion, in electronnuclear double resonance spectroscopy and in EPR imaging. Fourier transform EPR has proved particularly useful for studying the various spin polarization mechanisms involved in free radical chemistry, while the very high time resolution involved in echo-detected EPR makes it an ideal technique for measuring rates of chemical reactions and determining relaxation times. The underlying basis of pulsed EPR, like NMR, requires the generation of a spin echo by a short intense microwave pulse. However, in some cases the coupling between the electron spin and the nuclear spins can cause a modulation of the echo intensity. As this electron spin echo envelope modulation (ESEEM) can only occur if anisotropic interactions are present, it is only found in the solid state. Analysis of the ESEEM spectra allows determination of very small hyperfine couplings which are too small to be measured by CW-EPR, so that information on the number, identity and distance of weakly inter-acting nuclei may be obtained. A wide variety of applications of the nuclear modulation effect have been demonstrated, including determination of ligand structure in metalloproteins, magnetic properties of triplet states and location and coordination of surface complexes. In fact, the ESEEM method ranks with EXAFS as one of the most important tools for determining the structures of active centres in metalloproteins.

Conclusions It is intended in this article to demonstrate the diversity of applications in which CW-EPR is traditionally used, while briefly describing some current uses of more advanced EPR techniques. However, with recent technological developments in FT-EPR, pulsed

ENDOR, EPR imaging and high-frequency EPR, new and previously undreamed of prospects are likely to emerge. In many cases the spectroscopic data of interest cannot be obtained by any other method. This fact alone will ensure EPR a place as a dominant spectroscopic technique of major importance well into the future.

List of symbols a = hyperfine splitting constant, units are mTesla; a0 = isotropic hyperfine coupling constant, units are mTesla; A|| = hyperfine coupling constant parallel to a unique symmetry axis, units are mTesla; AA = hyperfine coupling constant perpendicular to a unique symmetry axis, units are mTesla; Ea = rotational activation energy, in kJ mol1; g = g factor; g|| = parallel to a unique symmetry axis; gA = perpendicular to a unique symmetry axis; W = free radical tumbling correlational time (in seconds). See also: EPR Imaging; EPR, Methods; EPR Spectroscopy, Theory; Inorganic Chemistry, Applications of Mass Spectrometry; Spin Trapping and Spin Labelling Studied Using EPR Spectroscopy; Surface Studies By IR Spectroscopy.

Further reading Bales B, Griffiths PC, Goyffon P, Howe AM and Rowlands CC (1997) EPR insights into aqueous solutions of gelatin and sodium dodecylsulphate. Journal of the Chemical Society, Perkin Transactions 2 2473 2477. Berliner LJ and Reubens J (eds) (1989) Biological Magnetic Resonance, vol. 8, Spin Labelling Theory and Applications. New York: Plenum Press. Carley AF, Edwards HA, Hancock FE, et al (1994) Application of ESR to study the hydrogenation of alkenes and benzene over a supported Pd catalyst. Journal of the Chemical Society, Faraday Transactions, 90: 3341 3346. Che M and Giamello E (1987) In: Fierro JLG (ed) Spectroscopic Characterisation of Heterogeneous Catalysts, vol. 57, Part B, p B265. Amsterdam: Elsevier Science Publishers. Dixon WT and Norman ROC (1962) Free radicals formed during the oxidation and reduction of peroxides. Nature, 196, 891. Dyrek K and Che M. (1997) Chemical Review 97: 305 331. Egerton TA, Murphy DM, Jenkins CA and Rowlands CC (1997) An EPR study of spin trapped free radical intermediates formed in the heterogeneously assisted photodecomposition of acetaldehyde. Journal of the Chemical Society, Perkin Transactions 2 24792485.

198 CHEMICAL EXCHANGE EFFECTS IN NMR

Marsh D (1994) Spin labelling in biological systems. Specialist Periodical Reports, Electron Spin Resonance 14: 166202. Möbius K (1994) EPR studies of photosynthesis. Specialist Periodical Reports, Electron Spin Resonance 14: 203245. Ranby B and Rabek JF (1977) ESR Spectroscopy in Polymer Research. Berlin: Springer-Verlag (and references therein). Rhodes CJ (1991) Organic radicals in solid matrices. Specialist Periodical Reports, Electron Spin Resonance 13a: 56103.

Schweiger A (1991) Pulsed electron spin resonance spectroscopy: basic principles, techniques and examples of applications. Angewandte Chemie, International Edition, English 30: 265292. Tabner BJ (1991) Organic radicals in solution. Specialist Periodical Reports, Electron Spin Resonance 13a: 155. Wasserman AM (1996) Spin labels and spin probes in polymers. Specialist Periodical Reports, Electron Spin Resonance 15: 112151.

MAGNETIC RESONANCE Theory

Introduction Chemical exchange, in NMR terms, means that a nucleus moves among a set of magnetic environments. The sample is macroscopically at equilibrium, but an individual nucleus exchanges among a number of sites, so the magnetic properties of the nucleus are modulated by the exchange. Chemical exchange effects were recognized early in the development of NMR: at about the same time (and in the same laboratory) as the scalar coupling. In N,Nc-dimethyl formamide and many other molecules with N,Nc-dimethyl groups (Figure 1), the two methyl groups have different chemical shifts if the molecule is rigid. At low temperature this is effectively the case, because of restricted rotation about the CN bond. However, as the sample is warmed, the two signals broaden, coalesce, and eventually start to sharpen up to a single line at the average chemical shift, as shown in Figure 1. This behaviour is usually interpreted in terms of an NMR timescale. For rotation about the CN bond, which is slow on this timescale at low temperature, the two signals are distinct and relatively sharp. At higher temperature, the rotation is faster and all that we can detect is the average resonance frequency. The broad coalescence line shape is characteristic of the intermediate timescale, and is perhaps the most familiar and obvious manifestation of chemical exchange. In this article we will concentrate on intermediate exchange, because a thorough

Figure 1 1H NMR spectra of the two N-methyl groups in 3dimethylamino-7-methyl-1,2,4-benzotriazine, as a function of temperature.

CHEMICAL EXCHANGE EFFECTS IN NMR 199

understanding of this case clarifies most other aspects of chemical exchange. Chemical exchange in all three regimes slow, intermediate and fast has an effect on the NMR spectrum. Typical values of the parameters mean that processes with rates in the range of 1104 s−1 can be studied most easily. In other words, this means activation energies for the reaction in the range of approximately 4080 kJ mol −1 (1020 kcal mol−1). This includes conformational changes in ring structures, restricted rotations about chemical bonds, ligand rearrangements in coordination complexes, and some intermolecular transfer processes. Molecules are dynamic entities, so the effects of chemical exchange are apparent in many NMR spectra. The theory of chemical exchange in simple; uncoupled spin systems have often been couched in terms of the Bloch equations, and this is the approach used in most of the literature. However, we feel that it is simpler and easier to consider the time domain. This time-domain method allows us to treat all exchanging systems slow, intermediate or fast, coupled or uncoupled in a consistent and simple way. There is also a simple physical picture of the spectroscopic transition probability that helps in this interpretation of the theory.

In chemical exchange, the two exchanging sites, A and B will have different Larmor frequencies, ZA and ZB. The exchange will carry magnetization from A to B and vice versa. If we assume equal populations in the two sites, and the rate of exchange to be k, then we can set up two coupled Bloch equations for the two sites, as in Equation [4].

The observable NMR signal is the imaginary part of the sum of the two steady-state magnetizations, MA and MB. The steady state implies that the time derivatives are zero, and a little further calculation (and neglect of T2 terms) gives the NMR spectrum of an exchanging system as in Equation [5].

The Bloch equations approach The Bloch equations for the motion of the x and y magnetizations (usually called the u- and v-mode signals), in the presence of a weak radiofrequency field, B1, are given in Equation [1].

This equation can be further extended to systems with unequal populations and more sites, using the same techniques.

Chemical exchange in the time domain In this equation, Z1 is the frequency of the RF irradiation, Z0 is the Larmor frequency of the spin, T2 is the spinspin relaxation time and Mz is the z magnetization of the spin system. We can simplify the notation somewhat by defining a complex magnetization, M, as in Equation [2].

With this definition, the Bloch equations can be written as in Equation [3].

If we create (by a pulse) the magnetization, MA and MB, at time zero, and then turn off the B1 magnetic field, Equation [4] can be simplified as in Equation [6].

200 CHEMICAL EXCHANGE EFFECTS IN NMR

In this equation, the matrix L is given by Equation [7], where we have made Z1 = (ZA + ZB)/2 and δ = (ZA − ZB)/2.

Equation [6] is a set of first-order differential equations, so its formal solution is given by Equation [8], in which exp( ) means the exponential of the matrix. In practice, we diagonalize the matrix L with a matrix of eigenvectors, U, as in Equation [9] to give a diagonal matrix, Λ, with the eigenvalues of L down the diagonal.

Equation [8] becomes Equation [10]. The exponential of a diagonal matrix is again a diagonal matrix with exponentials of the diagonal elements, as in Equation [11].

The eigenvalues of Equation [7] are given in Equation [13]

These eigenvalues are the (complex) frequencies of the lines in the spectrum, as in Equation [11]: the imaginary part gives the oscillation frequency and the real part gives the rate of decay. If k < G (slow exchange) then there are two different imaginary frequencies, which become ± G in the limit of small k (see Figure 2). In fast exchange, when k exceeds the shift difference, G, the quantity in the square root in [13] becomes positive, so the roots are pure real. This means that the spectrum is still two lines, but they are both at the average chemical shift, and have different widths. The full expressions for these line shapes are given in the Appendix. Because of the role of the eigenvectors in Equation [11], the factor (amplitude) multiplying the complex exponential is itself complex. The magnitude of the complex amplitude gives the intensity of the line and its phase gives the phase of the line (the mixture of absorption and dispersion). In slow exchange, the

As was mentioned above, the observed signal is the imaginary part of the sum of MA and MB, so Equation [11] predicts that the observed signal will be the sum of two exponentials, evolving at the frequencies λ1 and λ2. This is the free induction decay (FID). In the limit of no exchange, the two frequencies are simply ZA and ZB, as we would expect. When k is nonzero, the mathematics becomes slightly more complicated. If we ignore relaxation and exchange, then L is a Hermitian matrix with real eigenvalues and eigenvectors. However, when the exchange is important, the Hermitian character is lost and the eigenvalues and eigenvectors have both real and imaginary parts. The eigenvalues are given by the roots of the characteristic equation (Eqn [12]). Figure 2 Decomposition of the coalescence line shape into individual lines. Top: experimental spectrum from Figure 1; middle: calculated line shape to match experimental spectrum; bottom: individual lines calculated as in the Appendix.

CHEMICAL EXCHANGE EFFECTS IN NMR 201

two lines have the same real part, but the imaginary parts have opposite signs, so the phase distortion is opposite. The sum of these distorted line shapes gives the familiar coalescence spectrum, as in Figure 2. In fast exchange, the two lines are both in phase, but one line is negative. This negative line is very broad, and decreases in absolute intensity as the rate increases, leaving only the single, positive, inphase line for fast exchange. Therefore, the real and imaginary parts of the frequency, and the real and imaginary parts of the amplitude, provide the four parameters that define a line in an NMR spectrum: its intensity, its phase, its position, and its width. This gives us a time-domain picture of the chemical exchange, which can easily be converted to a spectrum by using the Fourier transform.

Systems with scalar coupling Scalar coupled systems are more complicated, but fundamentally no different from the uncoupled systems described in the time domain. If there is scalar coupling in the spectrum, the line shape becomes more complex. For instance, Figure 3 shows the line shape of the diacylpyridine ligand as part of a rhenium complex. The rhenium bonds to the pyridine nitrogen and to only one of the carbonyls, lifting the symmetry of protons 3 and 5. However, the rhenium can break this latter coordination and bond to the other carbonyl, which effectively interchanges protons 3 and 5 on the pyridine ring. During this exchange, they retain their coupling to the central proton 4, and the shift of 4 does not change. Therefore, proton 4 remains as a sharp triplet even when protons 3 and 5 are very broad. In any strongly coupled spin system, each line in the spectrum is a mixture of transitions of various nuclei, which depends on the chemical shifts and coupling constants. When a chemical exchange happens, all the spectral parameters change with it. Therefore, a magnetization (or a coherence) that was associated with a single spectral line in one site may be spread among several lines in the other site. In order to deal with these complexities, we must use the density matrix. One way of thinking of the density matrix is that it is a list of all the observables of a spin system. For example, the magnetizations of the two exchanging sites form part of the density matrix for that system. At equilibrium, the density matrix is just the z magnetization created by the static magnetic field. In a pulse Fourier transform NMR experiment, this z magnetization is flipped into the xy plane, and

Figure 3 Example of chemical exchange in a coupled spin system. The rhenium can bond to one or other of the carbonyl oxygen atoms, so that the symmetry of the molecule is lifted. Exchange of the rhenium between carbonyl groups broadens the signals of protons 3 and 5 (at 8.46 ppm and 7.82 ppm), but leaves 4 (8.38 ppm) unaffected.

divided among the individual lines in the spectrum, which then precess around the z axis. In order to calculate the spectrum of the system (or any other observable), we need to be able to follow the density matrix as a function of time. The equation of motion of the density matrix (ρ) is given in Equation [14], where H is the Hamiltonian of the spin system.

Some manipulation allows us to reformulate Equation [14] as Equation [15]. Mathematically, this means using superoperators in Liouville space L, but the details need not concern us here. The important point is that the density matrix becomes a vector of all possible observables of the system in this case we only deal with the xy magnetizations. Anything

202 CHEMICAL EXCHANGE EFFECTS IN NMR

we do to the system, pulses, free precession, relaxation, exchange, is represented by a matrix.

If we use frequency units (h/2π = 1), then the solution to Equation [15] is given in Equation [16], which is identical to Equation [8].

Relaxation or chemical exchange can be easily added in Liouville space, by including a Redfield matrix, Rso (so = superoperator), for relaxation, or a kinetic matrix, Kso, to describe exchange. Both relaxation and exchange are described very conveniently by superoperators. The complete equation of motion becomes Equation [17]. Note the similarity to the equations for two uncoupled sites, but now we derived them rigorously from the equations of motion of the density matrix.

The details of what the density matrix is at time zero, what we observe in the receiver, and how we construct the superoperators for a general spin system are all beyond the scope of this article. However, the basic idea remains: even the most complex exchange line shape is a sum of individual lines, with complex frequencies and complex intensities, as in Equation [17]. Each of those lines has a transition probability.

Physical interpretation of the transition probability The standard expression for the probability of a transition from an initial state, \, to a final state, I, induced by a perturbation Ix (the RF field along x), is given in Equation [18].

We can write the matrix element in Equation [18] in a rather more complicated way, as in Equation [19].

The reason for rewriting the equation is that the right hand side can be interpreted as the dot product of the operator Ix with the operator |I² ¢\|. We have moved to Liouville space, in which all the operators are vectors. The dot product in Liouville space is defined as the trace of the product of the operators in the original spin space. One important point is that Equation [19] is independent of basis set. The matrix element in Equation [18] becomes a simple dot product of two Liouville space vectors, as in Equation [20].

A dot product of two vectors gives the projection of one vector on the other, so Equation [20] defines the projection of |I² ¢\| onto Ix. In other words, this is the projection of the transition onto the total xy magnetization. The standard transition probability is the square of this projection. This leads to a physical interpretation of the transition probability. The spin system at equilibrium is represented by the total z magnetization, Fz. This is the sum of the z magnetizations of the sites, weighted by the equilibrium populations of each site. To distinguish the weighted sum, we give this spin operator a symbol different from Iz. A pulse flips this into the xy plane, so that immediately after the pulse, the spin system is represented by Fx. Now each transition receives a share of the total x magnetization. Its share is given by the projection of the transition (as an operator) onto the operator Fx. The transitions evolve independently as a function of time. However, we do not observe the transition directly, but rather the total xy magnetization. The detector is simply a coil of wire, and we measure the xy magnetization as a function of time to give us an FID. Fourier transforming the FID gives us a spectrum. Therefore, each transition contributes to the total signal according to its projection along Ix. The intensity of a transition is the product of how much

CHEMICAL EXCHANGE EFFECTS IN NMR 203

coherence it received from Fx at the start, times how visible it is to the receiver. Note that the Liouville matrix, iL+R+K may not be Hermitian, but it can still be diagonalized. Its eigenvalues and eigenvectors are not necessarily real, however, and the inverse of U may not be its complex-conjugate transpose. If we allow complex numbers in it, Equation [17] is a general result. Since Λ is a diagonal matrix we can expand in terms of the individual eigenvalues, Oj. We can also apply U (backwards) to Ix, and obtain Equation [21].

Each of the terms in the sum is a transition. The NMR signal is always a sum of decaying sine waves, whose frequency and decay rate are given by the imaginary and the real parts of Oj. The intensity is governed by the coefficients in Equation [21] of this term. This coefficient is the product of two terms. The first, (UIx)j, tells us how much the coherence overlaps with the receiver. The second term, (U−1 Fx)j tells us how much the coherence received from the equilibrium z magnetization. The product of these terms is the generalization of the transition probability. If we do not have relaxation, the two terms are complex conjugates, and we recover the

normal transition probability. However, the transition probability for a general system may have an imaginary part. If we cling to idea that each term in Equation [21] represents a transition, then an individual transition may be out of phase at t = 0. Immediately after the excitation pulse, the total magnetization will lie on the y axis, but the individual components may not. However, our true observable is still only the FID. Whether we want to decompose it into individual transitions of individual spins depends very much on the system we are studying.

Example of exchange of coupled spins Provided we can construct the matrices, the description of exchange is quite simple. Again, the details are beyond our scope, but an example of the Liouvillian (no exchange) for a coupled spin system is given in Equation [22] for an AB spin system, with coupling constant J. The eigenvalues of this matrix are the familiar line positions of an AB system.

If there is an exchange between A and B given by a rate k, then we set up the two blocks, as we set up the two exchanging spins in the Bloch equations. Note that we have made spin A in one block, exchange with spin B in the other. The full Liouvillian, including exchange, is given by Equation [23], in which dots replace zeroes to emphasize the form of the matrix. In this particular case, called mutual exchange, the spins simply permute themselves. We can simplify this matrix to half its original size, as in Equation [24], but

for nonmutual exchange we must retain Equation [23].

Except for its size, this is exactly the same form as in our previous cases. The spectrum can therefore be expressed as a sum of four lines, as in Figure 4.

204 CHEMICAL EXCHANGE EFFECTS IN NMR

Equation [25].

If we have two sites, A and B, then we can write the analogous equation for two sites as in Equation [26].

If the two sites exchange with rate k during the relaxation, then a spin can relax either through normal spinlattice relaxation processes, or by exchanging with the other site. Equation [26] becomes Equation [27].

Figure 4 Calculated spectra for mutual exchange in an AB spin system, as a function of exchange rate. The heavy traces are the total line shape and the lighter lines show the individual components. The bottom spectrum shows the typical static AB spectrum, which broadens and coalesces at higher rates of exchange.

Slow chemical exchange The term slow in this case means that the exchange rate is much smaller than the frequency differences in the spectrum, so the lines in the spectrum are not significantly broadened. However, the exchange rate is still comparable with the spin lattice relaxation times in the system. In this case, we can measure the rates by doing a modified spinlattice relaxation experiment, pioneered by Hoffman and Forsen. In the absence of exchange (and let us ignore dipolar relaxation), each z magnetization will relax back to equilibrium at a rate governed by its own T1 as in

This equation is very similar to Equations [6] and [7]. The basic situation is just as in intermediate exchange, except that now we are dealing with z magnetizations rather than xy. The frequencies are zero, and the matrix now has pure real eigenvalues, but the approach is the same. We also stay in the time domain, since a relaxation experiment follows the z magnetizations as a function of time. As before, the time dependence is obtained by diagonalizing the relaxationexchange matrix, and calculating the magnetizations for each time at which they are sampled. There are two main applications of slow chemical exchange: one is to determine the qualitative mechanism, and the other is to measure the rates of the processes as accurately as possible. For the first case, in which we have a spectrum in slow exchange, we need to establish the mechanism: which site is

CHEMICAL EXCHANGE EFFECTS IN NMR 205

exchanging with which. For this purpose, the homonuclear two-dimensional (2D) experiment EXSY (exchange spectroscopy) (the same pulse sequence as NOESY (2D nuclear Overhauser effect spectroscopy), but involving exchange) is by far the best technique to use. If there is no exchange, then the spectrum lies along the diagonal and there are no cross-peaks. Exchange between sites leads to a pair of symmetrical cross-peaks joining the diagonal peaks of the same site, so the mechanism is very obvious. The EXSY pulse sequence starts with two π/2 pulses separated by the incrementable delay, t1. This modulates the z magnetizations, so that the relaxation that occurs during the mixing time which follows, tm, is frequency-labelled. Finally the z magnetizations are sampled with a third π/2 pulse. Magnetization from a different site that enters via exchange will have a different frequency label. A two-dimensional Fourier transform then produces the spectrum. However, care must be taken in choosing the mixing time if there are multiple exchange processes. If the mixing time is too long, there is a substantial probability that a spin may have exchanged twice in that time, leading to spurious cross-peaks. For careful rate measurements, once the mechanism is established, it is our opinion that onedimensional (1D) methods are superior to quantitative 2D ones. Apart from the fact that 1D spectra can be integrated more easily, we also have more control over the experiment. Modern spectrometers can create almost any type of selective excitation, so that we can control the conditions at the start of the relaxa-

tion. For two sites, a nonselective inversion that inverts both sites equally will mask most of the exchange effects and the relaxation will be dominated by T1. However, if one site is inverted selectively, then that site can regain equilibrium by either T1 processes or by exchanging with the other site that was left at equilibrium. The inverted signal will relax at roughly the sum of the exchange and spinlattice relaxation rate, while the signal that was unperturbed at the start of the experiment shows a characteristic transient, as in Figure 5. For multiple sites, a wide range of initial conditions is available. Standard nonlinear least-squares methods allow us to fit these curves and derive values for the rates involved.

Fast exchange In fast exchange the spectrum has coalesced, usually to a single line, so the information in the spectrum is not so rich. For two equally-populated sites, for instance, we get a single Lorentzian line with a width proportional to G2/k. We can measure the line width (or better, T2) via a CPMG (CarrPurcell pulse sequence, MeiboomGill modification) experiment or a T1ρ measurement. However, if we do not know the frequency difference, G, we cannot obtain the rate. If the exchange rate is not too fast (< 10 4 roughly) modifications of the T2 experiments can help. Both experiments have an inherent timescale: in CPMG, it is the timing of the refocussing pulse; in T1ρ, it is the precession frequency about the spin-locking field. If the exchange is fast with respect to the experiment,

Figure 5 Inversion–recovery curves for slow exchange between two sites. Lines A and B are the results from one experiment. They show the recovery after a nonselective inversion of both sites, showing that the two sites have slightly different T1 values. C and D are obtained from a different experiment. Line C shows the recovery of a site that has been selectively inverted, and line D shows the behaviour of the line that was not perturbed in this experiment. The inverted line relaxes faster, due to a combination of spin–lattice, relaxation and exchange, and the unperturbed line shows the characteristic transient.

206 CHEMICAL EXCHANGE EFFECTS IN NMR

we measure a T2 appropriate to the coalesced spectrum. If, however, the exchange is slower than the experimental timescale (but still fast with respect to the frequency difference), the apparent T2 reflects the individual sites. As we change the timescale of the experiment, the apparent T2 changes, and so we can obtain a value for the rate itself.

Regardless of whether U is unitary, its inverse is given by Equation [30], where ∆ is the determinant of Equation [29].

Conclusions All of the effects of chemical exchange can be calculated by following the appropriate magnetizations as a function of time. For slow exchange, this is the complex coupling of relaxation and exchange that leads to transient behaviour in modified spinlattice relaxation experiments. For intermediate exchange, the rather complex line shapes can always be decomposed into a sum of individual transitions, even though the transitions are distorted in phase, intensity, position, and width by the dynamics of the system. For two uncoupled sites, this can be calculated quite easily, but for larger coupled systems the construction of the matrices may get complicated. However, the resulting lines are always governed by a transition probability, provided we extend our definition to allow probabilities with both real and imaginary parts. Working in the time domain is already familiar from Fourier transform spectroscopy, and it provides a complete and simple approach to chemical exchange.

Equation [28] then says that the signal is given by the following, regardless of slow or fast exchange.

The values of the eigenvectors have two forms, depending on whether G > k (slow exchange), or G < k (after coalescence). In the first case, the eigenvalues are given as:

and a convenient matrix of eigenvectors is given by:

Appendix Eigenvalues and eigenvectors for two-site exchange

For two-site equally populated chemical exchange, the two magnetizations at time zero are equal, so that Equation [11] can be written as in Equation [28], where the eigenvalues are complex numbers.

Since we are dealing with nonHermitian matrices, the matrix formed by the eigenvectors will not be unitary, and will have four independent complex elements. Let us simply call them a, b, c, and d, so that U is given by:

After coalescence, when the rate is greater than the frequency difference, the two transitions are both at zero frequency (i.e. the average chemical shift), but have different widths and intensities. The eigenvalues are pure real, as in Equation [34],

and the eigenvectors are similar, but reflect the fact that k2 − δ2 is now positive.

CHEMICAL IONIZATION IN MASS SPECTROMETRY 207

List of symbols B1 = radiofrequency field; Fz = total z magnetization; h = Plancks constant; Ix = perturbation RF field ; k = rate of exchange; Kso = kinetic along x ; i = superoperator matrix; L = Liouville superoperator; Lso = superoperator matrix; Mz = z magnetization of the spin system;Rso = Redfield superoperator matrix; T2 = spinspin relaxation time; T1ρ = T1 in rotating frame; u = x magnetization of spin system; ν = y magnetization of spin system; γ = magnetogyric ratio; G = chemical shift difference; λ = eigenvalue; Λ = diagonal matrix; I = final state;\ = initial state; Z1 = frequency of RF irradiation; Z0 = Larmor frequency of the spin. See also: NMR Principles; NMR Pulse Sequences; NMR Relaxation Rates; Two-Dimensional NMR, Methods.

Further reading Bain AD (1988) The superspin formalism for pulse NMR. Progress in Nuclear Magnetic Resonance Spectroscopy 20: 295315. Bain AD and Duns GJ (1996) A unified approach to dynamic nmr based on a physical interpretation of the transition probability. Canadian Journal of Chemistry 74: 819824.

Binsch G (1969) A unified theory of exchange effects on nuclear magnetic resonance lineshapes. Journal of the American Chemical Society 91: 13041309. Binsch G and Kessler H (1980) The kinetic and mechanistic evaluation of NMR spectra. Angewandte Chemie, International Edition in English 19: 411494. Gutowsky HS and Holm CH (1956) Rate processes and nuclear magnetic resonance spectra. II. Hindered internal rotation of amides. Journal of Chemical Physics 25: 12281234. Jackman LM and Cotton FA (1975) Dynamic Nuclear Magnetic Resonance Spectroscopy . New York: Academic Press. Johnson CS (1965) Chemical rate processes and magnetic resonance. Advances in Magnetic Resonance 1: 33102. Kaplan JI and Fraenkel G (1980) NMR of Chemically Exchanging Systems. New York: Academic Press. Orrell KG, Sik V and Stephenson D (1990) Quantitative investigation of molecular stereodynamics by 1D and 2D NMR methods. Progress in Nuclear Magnetic Resonance Spectroscopy 22: 141208. Perrin CL and Dwyer T (1990) Application of two-dimensional NMR to kinetics of chemical exchange. Chemical Review 90: 935967. Sandstrom J (1982) Dynamic NMR Spectroscopy . London: Academic Press. Sorensen OW, Eich GW, Levitt MH, Bodenhausen G and Ernst RR (1983) Product operator formalism for the description of NMR pulse experiments. Progress in Nuclear Magnetic Resonance Spectroscopy 20: 163 192.

Introduction In chemical ionization mass spectrometry (CIMS), ionization of the gaseous analyte occurs via gas-phase ionmolecule reactions rather than by direct electron impact (EI), photon impact or field ionization. EI ionization of a reagent gas (present in large excess) is usually followed by ionmolecule reactions involving the initially formed ions and the reagent gas neutrals to produce the chemical ionization (CI) reagent ion or reagent ion array. Collision of the reagent ion(s) with the analyte (usually present at ∼ 1% of the reagent gas

MASS SPECTROMETRY Methods & Instrumentation pressure) produces one or more ions characteristic of the analyte. These initial analyte ions may undergo fragmentation or, infrequently, react further with the reagent gas to produce a final array of ions representing the CI mass spectrum of the analyte as produced by the specific reagent gas. To a considerable extent, the usefulness of CIMS arises from the fact that a wide variety of reagent gases and, hence, reagent ions can be used to ionize the analyte; often the reagent system can be tailored to the problem to be solved. Problems amenable to solution by CI approaches include (a) molecular

208 CHEMICAL IONIZATION IN MASS SPECTROMETRY

mass determination, (b) structure elucidation and (c) identification and quantification. In many instances, CI provides information that is complementary rather than supplementary to that obtained by EI, and often both approaches are used. After a brief discussion of instrumentation, the major approaches, in both positive ion and negative ion CI, to the solution of the above problems are discussed. The focus is primarily on the use of medium-pressure mass spectrometry; CI at atmospheric pressure or at much lower pressures in ion trapping instruments are discussed elsewhere.

Instrumentation Most commonly, CI studies have been performed at total ion source pressures of 0.1 to 1 torr using conventional sector, quadrupole or time-of-flight mass spectrometers with ion sources only slightly modified from those used for EI ionization. The essential change is that the ion source is made more gas-tight by using smaller electron beam entrance and ion beam exit slits; the latter is accomplished by having either exchangeable source volumes or a moveable ion exit slit assembly to change from EI to CI. Most manufacturers now supply instruments capable of both EI and CI and incorporating enhanced pumping to maintain adequately low pressures in the remainder of the instrument when the source is operated at elevated pressures. Provision is made for introduction of the reagent gas and for introduction of the analyte by a heated inlet system, by a solids probe, by a direct exposure probe or by interfacing a gas chromatograph to the ion source. Capillary gas chromatographyCI mass spectrometry is a particularly powerful and sensitive approach to identification and quantification of the components of complex mixtures.

The enthalpy change for Equation [1] is given by

where the proton affinity of X, PA(X), is given by

If PA(M) > PA(B), the reaction is exothermic and normally occurs at the ionneutral collision rate determined by ion-induced dipole and iondipole interactions. Conversely, if the reaction is endothermic, the rate decreases exponentially by the enthalpy of activation and the sensitivity of the ionization process decreases. Most of the exothermicity of Equation [1] resides in MH+ and is available as internal energy to promote fragmentation of MH+. To maximize the probability of MH+ formation, which provides molecular mass information, one uses the least exothermic reaction possible. On the other hand, structural information is derived from the fragment ions observed in the CI mass spectrum and more exothermic protonation may be desirable to promote such fragmentation. The fragmentation of the even-electron MH+ ions usually involves elimination of even-electron stable molecules to form an even-electron fragment ion. Thus if the molecule contains a functional group Y, the fragmentation of MH+ frequently involves elimination of the stable molecule HY. For RYH+, the tendency to eliminate HY is roughly inversely related to the PA of HY, with the result that the ease of loss of Y as HY is approximately

Brønsted acid chemical ionization Gaseous Brønsted acids, BH+, react with the analyte M primarily by the proton transfer reaction

Other, usually minor, reactions that are possible include hydride ion abstraction (Eqn [2]) and charge exchange (Eqn [3])

although the stability of the resulting ion also plays a role in competing loss of different functional groups from a common MH+. Obviously, the more readily fragmentation occurs, the less likely it is that MH+ ions, giving molecular mass information, will be observed, although with suitable choice of reagent ions, MH+ ions are frequently formed when no M+ ions are observed in the EI mass spectrum. In other cases, such as hydroxylic compounds, the use of Brønsted base CI (see below) may prove useful in obtaining molecular mass information. Table 1 lists Brønsted acid reagent systems which have seen significant use, along with the major reactant ions and the proton affinities of the conjugate bases. Most organic molecules have PAs in the range

CHEMICAL IONIZATION IN MASS SPECTROMETRY 209

7601000 kJ mol−1. Thus protonation by H3+ will be strongly exothermic, while protonation in i-butane or ammonia CI will only be mildly exothermic or even endothermic. CH4 as a reagent gas is unique in that it forms two reactant species, CH5+ and C2H5+, in almost equal abundance; the polar reagent gases H2O, CH3OH and NH3 produce solvated protons with the extent of solvation depending on the partial pressure of the reagent gas. The usefulness of Brønsted acid CI in establishing molecular masses is illustrated by the comparison of the EI, CH4 CI and NH3 CI mass spectra of ephedrine in Figure 1. No molecular ion is observed in the EI mass spectra. By contrast, both the CH4 CI and NH3 CI mass spectra show abundant MH+ (m/z 166) ions, clearly establishing the molecular mass. In line with the relative PAs of Table 1, the less exothermic protonation in NH3 CI leads to less extensive fragmentation of the MH+ ion. As is often the case, CH4 CI provides not only molecular mass information (MH+) but also structurally informative fragment ions; the fragmentation of protonated ephedrine is rationalized in Scheme 1 and is seen to involve the formation of even-electron fragment ions by the elimination of small, stable, even-electron neutral molecules (the numbers in Scheme 1 indicate the m/z of the ionic fragment). In this case the CH4 CI mass spectrum provides more structural information than the EI mass spectrum. The effect of protonation exothermicity on the extent of fragmentation is shown more dramatically by the Brønsted acid CI mass spectra of the tripeptide H-Val-Pro-Leu-OH shown in Figure 2 and obtained using mass-selected reactant ions. With NH4+ as the reactant ion [PA(NH3) = 853.5 kJ mol −1] only MH+ is observed, while the more exothermic protonation by C2H5+ [PA(C2H4) = 680.3 kJ mol −1] leads to significant fragmentation of MH+, and the very exothermic protonation by N2OH+ [PA(N2O) = 580.7 kJ mol −1] leads to essentially complete fragmentation of MH+. Table 1

Brønsted acid reagent systems

PA(B) Reactant ion (BH+) (kJ mol−1) + H2 H3 423.4 N2O–H2 N2OH+ 580.7 CH5+ 550.6 CH4 C2H5+ 680.3 H+(H2O)na 696.0 H2O H+(CH3OH)na 773.6 CH3OH C4H9+ 819.6 i-C4H10 NH3 H+(NH3)na 853.5 a Degree of solvation depends on partial pressure of reagent gas; proton affinity given for monosolvated proton. Reagent gas

Figure 1 EI, CH4 CI and NH3 CI mass spectra of ephedrine (relative intensity as a function of m/z).

210 CHEMICAL IONIZATION IN MASS SPECTROMETRY

exchange to produce, initially, the odd-electron molecular ion (Eqn [6]), which may undergo further fragmentation (Eqn [7]).

Since the M+ ion formed initially is the same as that formed initially in EI ionization, the fragmentation reactions observed will be similar to those observed in EI mass spectra. The essential difference is that, while the M+ ions formed by electron ionization have a wide range of internal energies, the M+ ions formed by charge exchange have an internal energy, E(M+), given approximately by

where RE(R+), the recombination of R+, is defined by

Figure 2 Brønsted acid CI mass spectra of H-Val-Pro-Leu-OH using (A) NH4+, (B) C2H5+ and (C) N2OH+ as reagent ions. Reprinted with permission from Speir JP, Gorman GS, Cornett DS and Amster IJ (1991) Controlling the dissociation of peptide ions using laser desorption/chemical ionization Fourier transform mass spectrometry. Analytical Chemistry 63: 65–69. Copyright (1991) American Chemical Society.

Ammonia has a very high PA, with the result that NH4+ (and solvated forms thereof) will efficiently protonate only those analytes with a PA greater than about 860 kJ mol−1; in effect, this means that only nitrogen-containing analytes (such as Figures 1 and 2) are efficiently protonated. Analytes with lower PAs, which contain a lone pair of electrons, frequently give MNH4+ adducts.

and IE(M) is the adiabatic ionization energy of M. Reagent systems have been developed (Table 2) giving reactant ions with recombination energies over the range 9.316 eV. By comparison, most organic molecules have ionization energies in the range 8 12 eV, so that, by suitable choice of the charge exchange reagent, a wide range of internal energies can be imparted to the molecular ion in Equation [6]. A potential use of charge exchange ionization is to enhance differences in the mass spectra of isomeric molecules compared to EI ionization; the use of charge exchange has the advantage, compared to low-energy EI, that the sensitivity of ionization is maintained. This use of charge exchange is illustrated in Figure 3 Table 2

Charge exchange reagent systems

Reagent gas

Reactant ion (R +·) +

RE(R ·)(eV)

C6H6

N2–CS2

CS2+

CO2–C6F6

C6F6+

CO–COS

COS+

Charge exchange chemical ionization

Xe+

CO2

CO2+

In charge exchange CI, the reagent gas is chosen to produce an odd-electron species R+ on EI ionization. This ion reacts with the analyte M by charge

Kr+

N2+

15.3

Ar+

15.8, 15.9

9.3 ∼10.0 10.2 11.2 12.1, 13.4 13.8 14.0, 14.7

CHEMICAL IONIZATION IN MASS SPECTROMETRY 211

Figure 3 EI and C6F6+ charge exchange (CE) mass spectra of cis- and trans-4-methylcyclohexanol. Harrison AG and Lin MS, (1984) Stereochemical applications of mass spectrometry. 3. Energy dependence of the fragmentation of stereoisomeric methyl cyclohexanols. Organic Mass Spectrometry 19: 67–71. Copyright John Wiley & Sons Limited. Reproduced with permission.

which compares the C6F6+ charge exchange mass spectra of cis- and trans-4-methylcyclohexanol with 70 eV mass spectra. The low-energy charge exchange spectra are much simpler; more importantly, the differences in the spectra of the two isomers are much enhanced following charge exchange ionization. A particularly useful application of charge exchange ionization is for the selective ionization of specific components of mixtures. The principle of the method is illustrated in Figure 4 which shows the ionization energies of alkanes, cycloalkanes, alkylbenzenes and alkylnaphthalenes (components of petroleum products) plotted as a function of molecular mass. Also shown, as horizontal lines, are recombination energies of c-C6H12+, C6H5Cl+ and (CH3)3C6H3+. These species should ionize only those hydrocarbons with ionization below the appropriate line. Clearly, C6H5Cl+ will ionize only the benzene and naphthalene components, while (CH3)3C6H3+ will be even more selective for the naphthalenes and higher molecular mass benzenes.

Brønsted base chemical ionization Gaseous Brønsted bases, B−, usually react with organic analytes M by proton abstraction to produce

Figure 4 Ionization potentials of hydrocarbons as a function of molecular mass. Reprinted with permission from Sieck LW (1983) Determination of molecular weight distributions of aromatic components in petroleum products with chlorobenzene as reagent gas. Analytical Chemistry 55: 38–41. Copyright (1983) American Chemical Society.

[M−H]− as the initial product.

Equation [10] is normally rapid provided it is exothermic, i.e. provided ∆H (BH) > ∆H (M) or, alternatively, provided PA(B−) > PA([M − H]−), where

In general, little fragmentation of [M − H]− is observed, with the result that Brønsted base CI often provides molecular mass information through formation of the [M − H]− ion that is 1 amu lower than the molecular mass. Table 3 lists possible Brønsted base reagents along with the ∆H of the conjugate acid BH. For comparison, phenols have gas-phase acidities in the range of 13701465 kJ mol −1, carboxylic acids have acidities in the range of 13501422 kJ mol −1, alkynes in the range of 15201580 kJ mol −1 and alkylbenzenes about 1590 kJ mol −1. Also listed in the Table are the electron affinities of the neutral species

212 CHEMICAL IONIZATION IN MASS SPECTROMETRY

B; these are relevant to the tendency of B− to react by electron transfer to form M−. These electron affinities are quite high with the exception of O2− which is the only reactant likely to react by electron transfer to M. Of the bases listed OH− has seen by far the greatest use. It can be prepared by electron bombardment of a CH4N2O (∼ 90:10) mixture (at usual CI source pressures) through the reaction sequence

where the CH4 also serves to thermalize the electrons. Alternatively, the addition of CH3ONO to the methane bath gas leads to formation of CH3O−. The Cl− and Br− ions can be prepared by dissociative electron capture by halogen-containing compounds such as CH2Cl2 or CH2Br2. The F− ion is usually prepared by dissociative electron capture by NF3. The OH− ion is a strong base and will abstract a proton from most organic analytes to form [M − H]− Thus, OH− CI can be usefully employed to obtain molecular mass information. For example, the OH− CI mass spectra of ephedrine shows [M − H]− (m/z 164) as the dominant ion, and OH− CI could be used as readily as CH4 CI or NH3 CI (Figure 1) to establish the molecular mass. More importantly, Brønsted base CI can often be used to provide molecular mass information in cases where Brønsted acid CI fails. This is illustrated in Figure 5 which compares the CH4 CI and OH− CI mass spectra of 2,3-cyclopropyl-1,4-undecanediol. While the CH4 CI mass spectrum is quite uninformative (as is the EI mass spectrum), the OH− CI mass spectrum shows an abundant [M − H]− (m/z 199) ion, as well as fragmentation by sequential loss of two water molecules from [M − H]−. The hydroxyl ion can react with some RX molecules by an SN2 reaction to give X− and does react with esters, R′ COOR, in part, by addition/elimination to give R′COO− Table 3

Brønsted base reagents

B−

H (BH) (kJ mol−1)

NH2−

1690

75.3

−

EA(B) (kJ mol−1)

1636

176.6

O−·

1598

141.0

CH3O−

1594

151.5

F−

1552

328.0

O2−

1477

42.3

Cl−

1393

348.9

Br−

1356

324.7

Figure 5 CH4 CI and OH− CI mass spectra of 2,3-cyclopropyl1,4-undecanediol (relative intensity as a function of m/z).

+ ROH. The reactions of CH3O− have not been extensively studied; since it has a similar PA to that of OH− one would expect the reactions to be similar. The F− ion is a somewhat weaker base than OH− or CH3O− and thus is somewhat more selective in its proton abstraction reactions. Cl− is quite a weak base and will abstract only very acidic protons; in other cases [M + Cl]− adducts have been observed, which has proven useful in providing molecular mass information. The O− ion, which can be produced by electron bombardment of a N2N2O (∼ 9:1) mixture, is unique in that it is both a strong Brønsted base and a radical. As a result, it reacts with many organic analytes not only by proton abstraction but also by H-atom abstraction, H2+-abstraction and H-atom and alkyl group displacement. This variety of reactions is illustrated in Figure 6 for C5 ketones. The [M − H2]− · in ketones results to a significant extent by H + 2 abstraction from one α-carbon; the resultant ion undergoes further fragmentation by loss of the opposite alkyl group (see m/z 41, Figure 6, for example). Alkyl group displacement results in formation of carboxylate ions for ketones. Clearly O− Cl provides

CHEMICAL IONIZATION IN MASS SPECTROMETRY 213

Figure 6 O–• CI mass spectra of C5 ketones (Mr 86). Data from Marshall A, Tkaczyk M and Harrison AG (1991) O chemical ionization of carbonyl compounds. Journal of the American Society of Mass Spectrometry 2: 292–298.

a greater possibility of isomer distinction than OH− CI which produces only [M − H]−. However, O− CI usually has a considerably lower sensitivity than OH− CI or Brønsted acid CI.

Electron capture chemical ionization In electron capture CI, ionization of the analyte M occurs by electron capture (Eqn [14]) or by dissociative electron capture (Eqn [15]).

Both reactions are resonance processes that require electrons of near-thermal energy to occur efficiently.

Typically, a high-pressure (0.11.0 torr) buffer gas is used to thermalize the electrons (emitted from a heated filament) by inelastic scattering and positive ion ionization processes. Ideally, the buffer gas should not form stable anions, and buffer gases that have been used include He, H2, N2, CH4, NH3, CO2 and i-C4H10. Polyatomic gases such as CH4, NH3 iC4H10 or CO2 have proven to be most efficient in promoting electron capture. The first three have seen the most use, in part because they are often used in Brønsted acid CI and are readily available. The major attraction of electron capture chemical ionization (ECCI) is the prospect of enhanced sensitivity compared to other forms of CI or electron ionization. Whereas the maximum rate constants for the ionmolecule reactions involved in Brønsted acid or Brønsted base CI are in the range (24) × 10−9 cm3 molecule−1 s−1, electron capture rate constants can be as high as 10−7 cm3 molecule−1 s−1 because of the high mobility of the electron. Since the sensitivity of the ionization process depends, in part, on the rates of the ionization processes, in favourable circumstances ECCI can be up to 100 times more sensitive than other forms of ionization; this increase can be important when detection and quantification at trace levels is desired. At the same time the rate constants for electron capture are very compound-sensitive and can be very low, with the result that ECCI does show variable and, sometimes, very low sensitivity. It appears that only analytes with appreciable electron affinities (perhaps 0.5 eV) provide reasonable sensitivity in ECCI. The presence of halogen or nitro groups (particularly when attached to S-bonded systems), a conjugated carbonyl system or highly conjugated carbon systems leads to reasonably high electron affinities. Thus, as examples, dinitrophenols, dinitroanilines and polychlorinated aromatic compounds show good sensitivity in ECCI. Analytes that are not suitable candidates for electron capture can often be made so by suitable derivatization. Perfluoracyl, pentafluorobenzyl, pentafluorobenzoyl and nitrobenzyl derivatives are often used not only to increase the electron capture rate constants but also to enhance chromatographic properties. However, in contrast to electron capture gas chromatography, it is not sufficient that electron capture be facile; in addition, ions characteristic of the analyte must be formed since ions characteristic only of the derivatizing agent leaves the identity of the analyte in question. This is illustrated by the data in Table 4 for derivatization of phenols, where it can be seen that the perfluoroacyl derivatives yield ions characteristic only of the derivatizing agent, whereas the pentafluorobenzyl and pentafluorobenzoyl derivatives yield ions characteristic of the analyte. Also

214 CHEMICAL IONIZATION IN MASS SPECTROMETRY

Table 4

Derivatization of phenolsa

Relative intensity Derivative –COCF3 –COC2F5 –COC3F7 –CH2C6F5 (1)b

Other ions (relative intensity)

M–•

ArO −

0 0

0.2 0

CF3COO− (100)

C3F7CO− (100), C3F6CO− (6)

100 5

–COC6F5 (0.4)b

100

–COC6H3(CF3)2 (0.08)b

100

C2F5COO−(100), C2F4CO− (23)

C6F5−(8)

Data from Trainor TM and Vouros P (1987) Electron capture negative ion chemical ionization mass spectrometry of derivatized chlorophenols and chloroanilines. Analytical Chemistry 59: 601–610.

Relative molar response.

included in the Table are the relative sensitivities (relative molar response) for the three derivatives which give ions identifying the analyte; clearly, the pentafluorobenzyl or pentafluorobenzoyl derivatives are to be preferred in this case. In addition to variable sensitivity, two further problems that can be encountered in ECCI are nonreproducibility of the spectra and the observation of artifact peaks, i.e. ions that are not explainable simply on the basis of electron capture by the analyte. ECCI mass spectra tend to depend rather strongly on experimental conditions such as buffer gas identity and pressure, source temperature, purity of the buffer gas, amount of analyte introduced, instrument used and instrument focusing conditions. The variability of the spectra obtained is illustrated by the example in Figure 7. At 100ºC source temperature with CH4 buffer gas, nordiazepam shows M− (m/z 170) as the only significant ion, while at 150ºC source temperature Cl− is the dominant ion, and with 1 part in 2000 of oxygen in the methane, predominant formation of [M − H]− (m/z 169) is observed. This amount of oxygen could easily be in the system if there is a small air leak. The occurrence of artifact peaks is illustrated in Figure 8 which shows the ECCI spectra of tetracyanoquinodimethane with CH4 and CO2 as buffer gases. The major peaks in the spectrum with CH4 buffer gas represent artifact peaks arising from reaction of the analyte with free radicals produced on electron bombardment of methane; these altered species are subsequently ionized by electron capture. By contrast, the spectrum with CO2 buffer gas shows only M−. CO2 clearly is preferred in this case although when oxidation of the analyte occurs (such as formation of fluorenone from fluorene) the use of CO2 as a buffer gas leads to substantial artifact peaks. It appears that the most

Figure 7 Electron capture CI mass spectra of nordiazepam under different conditions (relative intensity as a function of m/z). Data from Garland WA and Miwa BJ (1983) Biomedical and Environmental Mass Spectrometry 10: 126–129.

prevalent pathway to these artifacts involves a surface-catalysed reaction of the analyte with some component of the reagent gas plasma, which produces one or more species with a higher cross section for electron capture than the original analyte. Alteration of the analyte presumably occurs when other ionization methods are used; the difference is that in other modes of CI all species are ionized with similar efficiencies, while in ECCI the altered species may be preferentially ionized. Despite these difficulties with ECCI it is often the method of choice when identification and quantification at very low concentrations and/or involving complex mixtures is necessary. In such work it is necessary to control the experimental parameters carefully since they can affect both spectra and sensitivity.

CHEMICAL IONIZATION IN MASS SPECTROMETRY 215

Figure 8 Electron capture CI mass spectra of tetracyanoethylene and tetracyanoquinodimethane using CH4 and CO2 as buffer gases. Reprinted with permission from Sears LJ and Grimsrud EP (1989) Elimination of unexpected ions in electron capture mass spectrometry using carbon dioxide buffer gas. Analytical Chemistry, 61: 2523–2528. Copyright (1989) American Chemical Society.

List of symbols etherm = electron of near-thermal energy; E = energy; m/z = mass to charge ratio ; ∆Hº = enthalpy change. See also: Ion Energetics in Mass Spectrometry; Ion Molecule Reactions in Mass Spectrometry.

Further reading Budzikiewicz H (1986) Negative chemical ionization (NCI) of organic compounds. Mass Spectrometry Reviews 5: 345380. Burrows EP (1995) Dimethyl ether chemical ionization mass spectrometry. Mass Spectrometry Reviews 14: 107115. Chapman JS (1993) Practical Organic Mass Spectrometry, 2nd edn, Chapters 3 and 4. Chichester, UK: Wiley. Creaser CS (1995) Chemical ionization in ion trap mass spectrometry. In: March RE and Todd JFJ (eds)

Practical Aspects of Ion Trap Mass Spectrometry, Vol III, Chapter 7. Boca Raton, FL: CRC Press. Harrison AG (1992) Chemical Ionization Mass Spectrometry, 2nd edn. Boca Raton, FL: CRC Press. Munson MSB and Field FH (1996) Chemical ionization mass spectrometry I. General introduction. Journal of the American Chemical Society 88: 26212630. Vairamani M, Mirza UA and Srinivas R (1990) Unusual positive ion reagents in chemical ionization mass spectrometry. Mass Spectrometry Reviews 9: 235 258. Westmore JB and Alauddin MM (1986) Ammonia chemical ionization mass spectrometry. Mass Spectrometry Reviews 5: 381465. Winkler FJ and Splitter JS (1994) Sterochemical effects in the positive- and negative-ion chemical ionization mass spectra of stereoisomeric molecules. In: Splitter JS and Ture ek F (eds) Applications of Mass Spectrometry to Organic Stereochemistry, Chapter 16. New York: VCH.

216 CHEMICAL REACTIONS STUDIED BY ELECTRONIC SPECTROSCOPY

Introduction This article covers applications of UV and visible electronic spectroscopy to studies of chemical reactivity. This encompasses both studies of reaction kinetics and of equilibria and chemical exchange phenomena. Chemical reactions can occur in different phases. Many of the reactions occur in the gas phase and the reactions which take place in this phase will be covered in the section on reaction kinetics. In solution, the volume occupied by the molecules will be larger than that in the gas phase, thus the molecules will come close together and the freedom for translational motion is restricted. There is another important difference between the solution and the gas phase, which is the great proximity between the solvent molecules and the solute molecules. The solvent can act as a heat sink and energy can pass from the solvent to the solute and this affects the vibrational modes and causes the reactants to overcome activation barriers. Solvent polarity will affect reactions in solutions containing ions. Polar solvents lower the energy required for the formation of ions. In solution many processes and reactions will take place such as diffusion-controlled reactions, reactions of hydrogen and hydroxyl ions, the formation of solvated molecules and electron transfer reactions. The rate coefficients depend on many factors such as the solvent and the formation of a solvent cage, the pressure and the ionic strength of the solution. Many reactions take place on solid surfaces, and solid materials catalyse such reactions. Surface catalysis is very important in industry, and makes important contributions to some atmospheric reactions. Molecules will be adsorbed on solid surfaces through physicosorption in which the attractive force between the reactant molecules and the surface is of the van der Waals type. Enthalpy of adsorption, ∆Had, of physicosorption is relatively small, about −40 kJ mol1. With chemical sorption the value for the enthalpy of adsorption is large because a strong bond is formed between the chemical molecule and the surface. Catalytic reactions can be classified as homogeneous catalysis or heterogeneous catalysis.

ELECTRONIC SPECTROSCOPY Applications

UV-visible spectroscopy for kinetics studies The problem of variable response factors

UV-visible spectroscopy has proved useful in biochemical analysis, environmental studies, in forensic science, drug kinetics, food quality, identification and quantification of chemical and biological substances but is limited as a tool for the investigation of molecular structure compared with infrared and nuclear magnetic resonance. However, it is quite useful for the study of reaction kinetics and equilibrium. There is a problem facing the use of a UV-visible spectrum for the calculation of the concentration of different species if two species absorb at the same region, and in this case there is a need to change the reagents to obtain different colours for different species. Also, it is important to carry out a close inspection for the two bands which have the same wavelength. An example of such a system is the absorption of potassium permanganate along with the absorption of a soluble azo dye, which appears as if it is absorbing at the same region, this being near 530 nm. A close inspection and calculations show that Omax of KMnO4 and the dye are 533 and 542 nm, respectively, while the extinction coefficients for KMnO4 and the dye are 1.01 and 1.88 u 10 14 dm3 mol−1 cm−1, respectively. Theory of UV spectroscopy

In UV-visible spectroscopy we are interested in studying the electronic energy. This energy can be calculated from the observed bands in the UV-visible region of the spectrum (195750 nm). The observed bands in this region are due to the excitation of the electrons by light of a certain wavelength, from the ground state to the excited state. This excitation of the energy of electrons between energy states is in accord with the FrankCondon principle which states that an electronic transition in a molecule takes place so rapidly compared with the vibrational motion of the nuclei that the internuclear distance can be regarded as fixed during the transition.

CHEMICAL REACTIONS STUDIED BY ELECTRONIC SPECTROSCOPY 217

The types of electronic transitions are given in Table 1. In organometallic compounds a very intense colour can arise due to the d → d, π → π* and n → π* transitions. This is true especially in transition metals connected to an organic molecule. Several factors, such as substitution (electronic and steric effect), solvent, temperature, and pH will affect electronic transitions and band intensity and shape. In the following paragraphs a brief discussion is given. Sample preparation

To obtain useful and accurate information from the UV-visible spectrum it is important to fix appropriate conditions for recording the spectrum. The studied material must be stable and not undergo decomposition by itself under the experimental conditions, which include concentration, solvent and temperature. The solute must dissolve completely in the solvent. It must not be colloidal, because this will given an error in the calculation of the different species present in solution. Also, the solution must not undergo hydrolysis or decomposition in this solvent, otherwise a new species will form which may not be soluble in the solvent. Cells for the UV region below 330 nm are made of fused silica. In the case of the visible region glass cells may be used. Cells for special measurements can be used and they are classified as sampling cells, flow cells and rectangular cells. In all measurements it is important to work with concentrations which are within the BeerLambert law. There is a need to carry out calibration so that by preparing different concentrations of the sample in the solvent and plotting the concentration against the recorded absorbance we have a straight line. It is necessary to use a concentration of the sample within this calibrated curve. The usual concentration, which is used in recording a UVvisible spectrum, is 10−310−5 M. The solute must not change its concentration due to decomposition or precipitation.

Table 1

UV-visible absorption of some chromophores

Chromophore

Compound

Electronic transition

λmax (nm)

C6 H6

Benzene

π → π*

200, 255

εa (m2 mol–1) 800

Solvent effects

It is very important to choose the appropriate solvent. The best solvent is that which is able to solubilize the solute completely, which is inert and will not interact with the solute and thus cause a perturbation of its electronic structure, and which will not absorb in the region where the sample will absorb. Table 2 gives the cutoff wavelength for a selective range of solvents. Sometimes it is important to use a solvent with a high solubilizing ability like water, ethanol, etc. Solvents must be of high purity because of the presence of impurities will affect the spectrum and some of those impurities have absorptions in the same region as that of the solute, while others will interact with the solute and this will change its structure. It is important to take into consideration the solvent effect on the solute. So if we take a solute dissolved in an inert solvent such as cyclohexane as our reference absorption, then some solvents will cause the absorption of the solute to shift more to the red region and this shift is denoted as bathochromic or redshift. Other solvents may cause the band to shift to the blue region and this is called blue-shift or hypsochromic. This phenomenon is important in UV-visible spectroscopy because it is possible to study the solutesolvent interactions and hydrogen bond strength. Some solute molecules can be used as a probe for measuring the solvent properties.

Table 2

Omax for different solvents

Solvent

Omax (nm)

Water

165

Acetonitrile (spectroscopic grade)

190

Hexane

199

Heptane

200

Isoctane

202

Diethyl ether

205

Ethanol

207

Propan-2-ol

209

Methanol

210

Cyclohexane

212

Acetonitrile (chemical grade)

213

Dioxan

216

Dichloromethane

233

Tetrahydrofuran

238

Trichloromethane

247

Tetrachloromethane

257

C C

Ethylene

π → π*

180

1300

C O

Acetone

π → π*

185

Dimethyl sulfoxide

270

n → π*

277

Dimethylformamide

271

N N

Azomethane

n → π*

347

Benzene

280

N O

Nitrosobutane n → π*

665

Pyridine

306

Propanone

331

Molar absorptivity.

218 CHEMICAL REACTIONS STUDIED BY ELECTRONIC SPECTROSCOPY

Light absorption and colour

Specialized techniques

Colour change is very important for the determination of different species present in the media whether they are the reacting species or the species formed during the reaction. From the spectra it is easy to calculate the concentration of all species and therefore to study the kinetics of the reaction.

Stopped flow method This method can be applied for the study of gases and liquids. The usual detection techniques are absorption spectrometry. High flow rates are required, to promote mixing for fast reactions.

Solution conditions and analysis

To obtain accurate data and concentrations for the different species present in solutions at specific times, it is important to obtain correct conditions for monitoring the species by the change of the absorbance of those species. It is important to determine the following conditions and their variation. Temperature and its variation It is possible to change the temperature and this will lead to a change of absorbance, thus, it is possible to calculate the activation energy of the reaction. pH This is very important for aqueous solutions. The pH can have a large affect on the UV-visible spectrum, rates of complexation or change in reaction kinetics. pH can have a pronounced effect on shifting the equilibrium in enolketo tautomerism. A good example is the effect of pH on the equilibrium in the system dichromate/chromate. Ionic strength The ionic strength of the solution is important for some reactions, especially for solutions which contain ions other than the neutral reactant molecules; interaction will occur between ions other than the reactants and this will affect the rate constant. Effect of pressure For the majority of chemical reactions it is difficult to study the pressure effect. The reaction rate is insensitive to changes in pressure and a large pressure must be applied before we observe any important change. Sometimes we need to study the effect of pressure because it provides a useful insight into the mechanism of the reaction. Solvent polarity In general, one should use solvents with low polarity so there will be no interference in the formation and decomposition of different species. On occasions, however, it is necessary to use solvents with high polarity to increase the solubility of the solutes or to affect the activation energy of a certain reaction. Other factors that are important are the order of addition of different reagents, mixing and stirring to obtain solution homogeneity and choice of the proper wavelength which will be used to monitor the change of concentration during the reaction.

Flash photolysis and pulse radiolysis In the flash photolysis technique a reactant is irradiated with an intense flash of visible or ultraviolet light. The intensity must be sufficient to produce a measurable change in chemical composition, but of short duration compared with that of the ensuing reactions, which are to be studied. Multidimensional spectroscopy and derivative spectroscopy When bands of reactants and reaction products overlap in the fundamental UV-visible absorption spectra the reaction kinetics cannot be followed by the classic UV method. In many cases, the second derivative UV-visible spectrophotometry (D-2) provides an alternative method to solve the problem. Even-order derivatives are suitable to follow kinetics because the maxima in the UV-visible derivative spectrum can be associated with the minima and a low-noise online spectra is obtained which can be computed up to the 6th order derivative and even up to the 10th order with the newly developed computers. On the other hand, the first derivative does not provide the above association; and other higher odd-order derivatives are less precise, though in practice it has proved valuable to work with spectra of the 3rd and 5th order.

Applications to chemical reactions Many applications of UV-visible spectroscopy to the study of chemical reactions have been published and some illustrative examples which have been reported in the literature during the last two years will be covered. Complex formation

The kinetics of transformation of complexes in organic media can be studied. Such studies will indicate whether the complexes will form in a one-step reaction mechanism or a multistep reaction mechanism. It is possible to determine the rate constants and the stability constant of the complex. UV-visible spectroscopy can be used to study the protonation of certain complexes and it is possible to obtain information on the position of protonation. Reduction of complexes can be studied by UV-visible spectroscopy. Reduction of trans-[Pt(CN)4X2]2−

CHEMICAL REACTIONS STUDIED BY ELECTRONIC SPECTROSCOPY 219

(X = Cl or Br) [as model compounds for antitumouractive platinum(IV) pro-drugs] to [Pt(CN)4]2− by Lmethionine, MeSR, has been studied at 25°C in the range 0 < pH < 12 (X = Cl) and 0 < pH < 6 (X = Br) by use of stopped-flow spectrophotometry. It was concluded that methionine-containing biomolecules may compete with thiol compounds for reduction of platinum pro-drugs under acidic conditions, and also in neutral solutions with low concentrations of thiolcontaining biomolecules. The formation kinetics and the thermodynamic stability of iron(III) complexes with a new tetradentate ligand, N(alpha)-salicyl-L-alaninehydroxamic acid (H-2slalh), have been investigated by the use of UV-visible and stopped-flow spectrophotometric methods. By comparing the pH-dependent absorbance change, the mechanism of the complex formation was explained. Photochemical reactions

Photochemical behaviour of full-aromatic E-carbolines, nor-harmane, harmane and harmine in chloroform medium and the corresponding UV absorption and fluorescence emission spectra have been discussed. Irreversible electron transfer from the singlet excited E-carboline molecule to the chloroform molecule in the transient excited charge transfer complex has been proposed as the primary photochemical process initiating the mechanism of secondary reactions in this system. In deoxygenated water, methanol, and ethanol, 4hydroxybenzonitrile (4-HBN) is photoisomerized into 4-hydroxybenzoisonitrile (4-HBIN), which is then hydrolysed into 4-hydroxyformanilide in acidic medium. The triplettriplet absorption of 4-HBN (Omax = 300 nm) is detected by transient absorption spectroscopy. The triplet is converted into long-lived transients absorbing in the far-UV. The analysis of the kinetics of 4-HBIN formation as a function of irradiating photon flux shows that the photoisomerization of 4-HBN is a two-stage photoprocess. According to triplet-quenching studies, the first stage proceeds via the 4-HBN triplet to yield an intermediate capable of absorbing a second UV photon, which then gives 4-HBIN in the second stage. Mechanistic considerations indicate that this intermediate is likely to be an arizine. Tautomerism and isomerization

Tautomerism is an important phenomenon and an example of this process is the equilibrium in Schiff bases. These bases can exist as an equilibrium between two species, keto enol [Scheme 1]. This process is affected by a number of factors such as

solvent polarity, temperature and pH. The kinetics of tautomerism can be studied by UV-visible spectroscopy and we will deal with a few examples.

The tautomerism between a hydroxy Schiff base and the corresponding ring-closed oxazolidine was kinetically studied in chloroform. This method indicated that this reaction is pseudo-first-order. UV examination was used to deduce the molecular species in various pH buffers. In an acid solution (e.g. pH 3.0) the Schiff base existed as the protonated Schiff base at the imine nitrogen atom, and in the alkaline region (e.g. pH 9.0) as the oxazolidine form. The ketoenol equilibrium constants of acetylacetone, ethyl acetoacetate and ethyl benzoylacetate in water at 25°C were determined by studying the influence of surfactants on their UV-visible spectra. These measured equilibrium constants were used to obtain the reactivity of the ketones towards several nitrosating agents. Atmospheric, environmental reactions, gas phase and free radical kinetics

The UV absorption cross-sections of peroxyacetyl nitrate (PAN), CH3C(O)O2NO2, have been measured as a function of temperature (298, 273 and 250 K) between 195 and 345 nm. Photolysis becomes the most important atmospheric loss process for PAN, and the OH reaction is found to unimportant throughout the troposphere. The atmospheric chemistry of benzene oxide/oxepin, a possible intermediate in the atmospheric oxidation of aromatic hydrocarbons, has been investigated by visible and UV photolysis. The results indicated that the major atmospheric sinks of benzene oxide/oxepin are the reaction with OH radicals and photolysis and under smog chamber conditions, with high NO2 concentrations, also reaction with NO3. A flash photolysis-resonance fluorescence technique was used to study the rate constant for the reaction of OH radicals dimethyl carbonate over the temperature range 252370 K. Pulse radiolysis/ transient UV absorption techniques were used to

220 CHEMICAL REACTIONS STUDIED BY ELECTRONIC SPECTROSCOPY

study the ultraviolet absorption spectra and kinetics of CH3OC(O)OCH2 and CH3OC(O)CH2O2 radicals at 296 K. The codisposal of trace metals (e.g. Co), synthetic chelates (e.g. ethylenediaminetetraacetic acid, H(4) EDTA) and water-miscible organic solvents has occurred at some contamination sites. The reactions of Co(II)-EDTA with a redox reactive naturally occurring solid, goethite, in aqueous and semiaqueous (methanolwater, acetonewater) suspensions was studied. UV-visible spectroscopy indicated that goethite catalysed oxidation of Co(II)-EDTA to Co(III)-EDTA by dissolved O2. These reactions have important implications on the fate of the redox-sensitive metal in complex, mixed waste environments. Photocatalytic degradation

A number of organic pollutants present in industrial and domestic wastewater resist biodegradation and they are poisonous even at low concentrations. A new method for removing toxicants from wastewater is based on the use of photocatalysis. It was found that titanium dioxide (TiO2) is the best photocatalyst for the detoxification of water because it is cheap, nontoxic and easy to use and handle. The energy gap in TiO2 is 3.2 eV and thus it can be activated at >400 nm. The primary oxidant responsible for most advanced oxidation processes, i.e. the use of photochemical methods, is the hydroxyl radical, which is formed by the reduction reactions of electrophiles with water or hydroxide ions. The mechanism of the formation of the hydroxyl radical is well discussed in the literature. Heterogeneous photocatalytic oxidation of organic compounds in aqueous solution is achieved by the reactive hydroxyl radical. The photocatalytic process can remove a large number of organic hazardous compounds in water. Phenols, carboxylic acids and herbicides are among a large number of pollutants, which are destroyed in air and in water using photocatalysis. Photodegradation of 1,2,3,4-tetrachlorodibenzop-dioxin (1,2,3,4-TCDD) in hexane solution was studied under controlled near-UV light exposure in the spectral region from 325 to 269 nm. Irradiation experiments carried out at a constant light energy (700 mJ) showed that the percentage of 1,2,3,4TCDD left in the solution after irradiation changed from about 55 to 75%, with a minimum of 55% at 310 nm. Further irradiation experiments carried out at two wavelengths, namely 310 and 269 nm, and light energy ranging from 0 to 4000 mJ, showed that the photodegradation reaction of the TCDD always followed pseudo-first-order kinetics.

A semitransparent TiO2 film with extraordinarily high photocatalytic activity was prepared on a glass substrate by sintering a TiO2 sol at 450°C. The photocatalytic properties of the film were investigated by measuring the photodegradative oxidation of gaseous acetaldehyde at various concentrations under strong and weak UV light irradiation conditions. The kinetics of acetaldehyde degradation as catalysed by the TiO2 film as well as by P-25 powder were analysed in terms of the LangmuirHinshelwood model. It is shown that the number of adsorption sites per unit true surface area is larger with the TiO2 film, as analysed in powder form, than with P-25 powder. Meanwhile, the first-order reaction rate constant is also much larger with the film than with P-25 powder. Moreover, under most experimental conditions, particularly with high concentrations of acetaldehyde and weak UV illumination intensity, the quantum efficiency was found to exceed 100% on an absorbedphoton basis, assuming that only photogenerated holes play a major role in the reaction. This leads to the conclusion that the photodegradative oxidation of acetaldehyde is not mediated solely by hydroxyl radicals, generated via hole capture by surface hydroxyl ions or water molecules, but also by photocatalytically generated superoxide ion, which can be generated by the reduction of adsorbed oxygen with photogenerated electrons. Solvents effects and solvolysis

Oxidations of arylalkanes by (Bu4NMnO4)-Bun have been studied, e.g. toluene, ethylbenzene, diphenylmethane, triphenylmethane, 9,10-dihydroanthracene, xanthene and fluorene. Toluene is oxidized to benzoic acid and a small amount of benzaldehyde; other substrates give oxygenated and/or dehydrogenated products. The manganese product of all of the reactions is colloidal MnO2. The kinetics of the reactions, monitored by UV-visible spectrometry, show that the initial reactions are first order in the concentrations of both (Bu4NMnO4)-Bun and substrate. No induction periods are observed. The same rate constants for toluene oxidation are observed in neat toluene and in o-dichlorobenzene solvent, within experimental errors. The presence of O2 increases the rate of (Bu4NMnO4)-Bun disappearance. The reactions of toluand dihydroanthracene exhibit primary isotope effects: The rates of oxidation of substituted toluenes show only small substituent effects. In the reactions of dihydroanthracene and fluorene, the MnO2 product is consumed in a subsequent reaction that appears to form a charge-transfer complex. The rate-limiting step in all of the reactions is hydrogen atom transfer from the substrate to a

CHEMICAL REACTIONS STUDIED BY ELECTRONIC SPECTROSCOPY 221

permanganate oxo group. The enthalpies of activation for the different substrates are directly proportional to the 'H for the hydrogen atom transfer step, as is typical of organic radical reactions. The ability of permanganate to abstract a hydrogen atom is explained on the basis of its ability to form an 80 ± 3 kcalmol−1 bond to H, as calculated from a thermochemical cycle. Polymerization

The kinetics of oxidation of a macromolecule, poly(ethylene glycol) [PEG] by ceric sulfate in sulfuric acid medium has been studied by means of UVvisible spectrophotometry. The observed difference in rates of oxidation has been explained in terms of cage formation. The oxidation of PEG proceeded without the formation of a stable intermediate complex. The order with respect to the concentrations of PEG and ceric sulfate has been found to be one and the overall order is two. The effects of acid concentration, sulfate ion concentration, ionic strength and temperature on the rate of the oxidation reaction have been studied. The thermodynamic parameters for the oxidation reaction have also been presented. Based on the experimental results, a suitable kinetic expression and a plausible mechanism have been proposed for the oxidation reaction. Reaction of trans-1,3-diphenyl-1-butene (D), the trans ethylenic dimer of styrene, with trifluoromethanesulfonic acid in dichloromethane has been performed at temperatures lower than room temperature using a stopped-flow technique with real time UV-visible spectroscopic detection. The main product of the reaction was the dimer of D. A transient absorption at 340 nm has been assigned to 1,3diphenylbutylium, a model for the polystyryl cation. Other absorptions at 349 nm and 505 nm have also been observed and were assigned to an allylic cation, 1,3-diphenyl-1-buten-3-ylium, resulting from hydride abstraction from D. This species was very stable at temperatures lower than −30°C. A general mechanism was proposed based on a kinetic study of the reactions involved. Determination of metals and molecules at low concentration

A new and simple method for measuring peroxides in a single living cell has been developed, and the generation of peroxides upon ultraviolet (UV) irradiation was measured in human and pig pidermal keratinocytes. The method was based on the fact that the nonfluorescent dye dihydrorhodamine 123 reacts in the presence of peroxides, such as H2O2, and changes into fluorescent rhodamine 123, and hence

the fluorescence intensity is proportional to the amount of reacted peroxide. The epidermal kerationcytes were loaded with the dihydrorhodamine under a fluorescence microscope and exposed to UV radiation. Taking C as the content of peroxides generated within the cell and I as the increase influence (radiation intensity × time = photons cm−2), the following empirical relationship was established: C = Cs (1 − exp−kI), where Cs is the content of peroxides at the saturation state, and k is a kinetic parameter. The dependence of the two parameters on wavelength in the range 280400 nm was studied. In human keratinocytes Cs had a peak at 310 nm and a small peak (shoulder) at 380 nm, while k increased gradually toward shorter wavelengths. In pig keratinocytes, on the other hand, k had a peak around 380 nm and a shoulder at 330 nm, while Cs remained unchanged. Aminotriazole, an inhibitor of catalase, and low temperatures increased the stationary levels of peroxide generation in pig keratinocytes upon UV irradiation, indicating that the reaction used for measuring intracellular peroxides is competitive with the intrinsic reactions in scavenging peroxides. A first simultaneous EPR and visible spectrophotometric study is reported on the interaction of the stable free radical 1,1-diphenyl-2-picrylhydrazyl (DPPH) dissolved in ethanol with thioglycolic acid (HSCH2CO2H, TGA). The results of the kinetic studies at room temperature allow us to assume 1:1 stoichiometry of the reaction between DPPH and TGA giving 1,1-diphenyl-2-picrylhydrazine (DPPH) and thioglycolic disulfide. The linear plots of EPR and UV-visible responses versus the quantity of added TGA are used to find the DPPH molar absorptivity at 520 nm to be 12350 ±3% 1mol −1 cm−1 which may be used as a criterion for the purity of the material itself. It was also found that the paramagnetic and optical properties of a 30-year old sample gave results suggesting that in the solid state DPPH is a fairly stable material. Reaction kinetics, substitution effects, structurereactivity relationships

The kinetics of the addition of arenesulfinic acids to 4-substituted 2-nitroethenylarenes was studied by means of UV spectrophotometry. The effect of 4-substituting groups in benzenesulfinic acids, and the change in reactivity of the nitroethylene system due to typical electron-donating and electron-withdrawing groups were investigated. The substituent effect on benzenesulfinic acid fits Hammetts equation, p-value at 298 K being −1.12. Kinetic studies were carried out at 288308 K, and the activating energy and the enthalpy of activation were determined.

222 CHEMICAL REACTIONS STUDIED BY ELECTRONIC SPECTROSCOPY

The kinetics of the addition of unsubstituted and substituted benzenesulfinic acids to various 2-halogenonitroethenylarenes was studied by means of UV spectroscopy. The effect of 4-substituents in benzenesulfinic acids and the change in reactivity of the nitroethylene system in 2-halogeno-2-nitroethenyl-arenes in the presence of various substituents in the benzene ring and various halogens at the double bond were investigated. Kinetic studies were made at 288308 K, and the activating energy and enthalpy of activation were determined. Effect of pH, pressure, temperature and ionic strength on reaction kinetics

The degradation kinetics of xanthate in homogeneous solution as a function of pH at 5°C, 20°C and 40°C was systematically studied by UV-visible spectrophotometric measurements. The results indicate that the degradation of ethyl xanthate is rapidly increased with decreasing pH at pH < 7. At pH 78, the maximum half-life of the xanthate appears. The degradation was faster at pH 910, but at pH > 10 the half-lives of xanthate once again increase. The investigations were also extended to different media other than pure water, such as, 0.1 M NaClO4, 0.1 M NaNO3, 0.1 M NaCl as well as in the supernatants of flotation tailings of sulfide minerals. The rate constants of xanthate degradation were calculated and presented together with half-lives and activation energies of xanthate degradation. The degradation products and reaction mechanisms are discussed based on experimental results. The kinetics of the association reaction of the phenoxy radical with NO were investigated using a flash photolysis technique coupled to UV absorption spectrometry. Experiments were performed at atmospheric pressure, and theoretical calculations showed that the rate constant is at the high-pressure limit above 50 torr (6.7 × 103 Pa) for temperatures below 400 K. Upon increasing the temperature, the reaction was found to be reversible, and the equilibrium kinetics have been studied at seven temperatures between 310 and 423 K. Validation of molecular orbital calculations

The suitability of Gaussian distribution functions to describe the shape and temperature dependence of the UV absorption continua of peroxy radicals has been investigated. The ethylperoxy radical was used as a test case. Its 298 K absorption continuum was found to be best described by a semilogarithmic Gaussian distribution function. A linear Gaussian distribution function performed less well but still

adequately described the continuous absorption. The temperature dependence of the ethylperoxy radical UV absorption continuum was also well predicted. Analogous results obtained for the methylperoxy radical support these conclusions. A theoretical comparison of the semilogarithmic and linear Gaussian distribution functions is given and a potential energy diagram of the ethylperoxy radical derived. The experimentally determined absorption cross-sections of HO2 have been reanalysed. It is shown that either the measurements at short wavelengths are in error or an unidentified electronic transition of HO2 exists. Enzymatic reactions

A one-step spectrophotometric method for monitoring nucleic acid cleavage by ribonuclease. H from E. coli and type II restriction endonucleases has been proposed. It is based on recording the increase in the UV absorbance at 260 nm during the course of enzymatic reaction. Duplexes stable under the reaction conditions were chosen as substrates for the enzymes being studied. In order to obtain duplex dissociation following their cleavage by the enzyme appropriate temperature conditions were selected. The spectrophotometric method may be applied for rapid testing of the nuclease activity in protein preparations as well as for precise quantitative analysis of nucleic acid degradation by enzymes. This method may be successfully employed in kinetic studies of nucleic acidprotein interactions. See also: Biomacromolecular Applications of UV-Visible Absorption Spectroscopy; Colorimetry, Theory; Dyes and Indicators, Use of UV-Visible Absorption Spectroscopy.

Further reading Atkins PW (1998) Physical Chemistry, 6th edn. Oxford: Oxford University Press. Barrow GM (1986) Introduction to Molecular Spectroscopy, 17th edn. London: McGraw-Hill Book Company. Denney RC and Sinclair R (1991) Visible and Ultra Violet Spectroscopy. New York: Wiley. Kempt W (1989) Organic Spectroscopy, 2nd edn. London: Macmillan Education. Ladd M (1998) Introduction to Physical Chemistry, 3rd edn. London: Cambridge University Press. Pilling MJ and Seakins PW (1996) Reaction Kinetics, 1st edn. London: Oxford Science Publication. Pérez-Bendito D and Silva M (1988) Kinetics Methods in Analytical Chemistry, 1st edn. Chichester: Ellis Horwood. Vemulapalli K (1993) Physical Chemistry. New York: Prentice-Hall International.

CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR 223

Introduction Chemical shift and relaxation reagents are substances that, when added to the systems under study, enable the spectroscopist to tackle a specific problem of spectral assignment, stereochemical determination or quantitative measurement. They are usually represented by paramagnetic compounds, and their effects may exploited by the solute and the solvent, manifested as large shifts and/or pronounced relaxation enhancement of the lines. Both effects reflect the response of the nuclear dipole to the local magnetic field of an unpaired electron in the reagent, whose magnetic moment is 658.21 times higher than that of the proton. A knowledge of the basic theory of the paramagnetic interaction is useful to pursue a better exploitation of the effects induced by the paramagnetic perturbation. In principle, the interaction of a nucleus with an unpaired electron (hyperfine interaction) has the same nature as a nucleusnucleus interaction, i.e. it is determined by a direct, through-space dipole dipole coupling (HD) and an indirect, through-bonds contact (scalar) coupling (HS):

The hyperfine shift arises from components present in both terms which are invariant with time in the external magnetic field; these contributions are indicated as dipolar or pseudocontact shifts when resulting from dipolar interactions and Fermi or contact shifts when resulting from scalar interactions, respectively. The time-dependent fluctuations in either term will result in nuclear relaxation (dipolar and scalar relaxation, respectively). Although, in theory, correct expressions for HD and HS can be derived, the elusive nature of the orbiting electrons and the number of their interactions make it very difficult to develop an accurate theory for shift and relaxation promoted by paramagnetic substances. In general one may conclude that, while measurements of the effects promoted by paramagnetic species are easy, interpretation of data is difficult.

MAGNETIC RESONANCE Applications

According to their dominant effect on the NMR spectra of surrounding nuclei, paramagnetic reagents can be classified as chemical shift or relaxation reagents. Organic free radicals and metal ions with isotropic electron density, such as Mn2+ (five electrons in five d orbitals) or Gd3+ (seven electrons in seven f orbitals) do not give rise to pseudocontact shifts and are commonly regarded as relaxation probes. On the other hand, anisotropic ions with short electron relaxation times, such as all other lanthanide ions, can be referred to as chemical shift probes. Species such as Ni2+, Fe3+ and Cr3+ give rise to both effects and defy classification.

Chemical shift reagents In 1969, Hinckley suggested that the use of paramagnetic lanthanide complexes might be very useful to simplify unresolved 1H, NMR resonances, thus allowing otherwise intractable spectra to be easily interpreted (Figure 1). Although the introduction of high-field magnets and two-dimensional methods have largely reduced the need for such qualitative use of lanthanide induced shift (LIS) measurements, the exploitation of the large local fields generated by some lanthanide ions is still an item of strong interest in several areas of NMR applications. When a substrate interacts with the lanthanide complex, it magnetically active nuclei feel the presence of the unpaired f electrons; their LIS is basically determined by two properties of the lanthanide complex, namely (i) its Lewis acid behaviour and (ii) the number of unpaired electrons. The commercially available lanthanide shift reagents (LSRs) are mainly tris(β-diketonate) complexes, having the 2,4-pentanedione chelate structure (Table 1). These neutral complexes are soluble in organic solvents and they promptly form acidbase adducts with substrates endowed with Lewis base behaviour. In fact alcohols, ethers, amines and nitriles reversibly interact with the LSR to form paramagnetic adducts whose 1H NMR shifts may be far away from the diamagnetic values. As the substrate B in fast exchange between the LSR-bound and free forms, the effect will be averaged on all the molecules of the substrate. The

224 CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR

Table 1

Lanthanide diketonate chelates used as NMR shift re-

agents

R′

t-C4H9

t - C4H9 t -C4H9

CF3 n -C3F7

n -C3F7 C2F5 CF3

n-C3F7 C2F5 n -C3F7

Acronym

Pr Eu Dy Ho Yb Eu Eu Pr Eu Eu Eu

Ln(thd)3 or Ln(dpm)3

Eu(pta)3 Eu(fod)3 Eu(tfn)3 Eu(fhd)3 Eu(dfhd)3

thd = tris(2,2,6,6-tetramethyl-3,5-heptanedionato); dpm = tris(dipivaloylmethanato); pta = tris(1,1,1-trifluoro-5,5-dimethyl-2,4-hexanedionato); tfn = tris(1,1,1,2,2,3,3,7,7,8,8,9,9,9-tetradecafluoro-4,6-nonanedionato); fhd =tris(1,1,1,2,2,6,6,7,7,7-decafluoro-3,5-heptanedionato); dfhd = tris(1,1,1,2,2,3,3,7,7,7-decafluoro-4,6-heptanedionato); fod = tris(1,1,1,2,2,3,3-heptafluoro-7,7-dimethyl-4,6-octanedionato). Figure 1 Simplification of the low-field 1H NMR spectrum of 1pentanol by Eu(thd)3. The clustered signals of the methylene chain are totally separated into clearly defined multiplets. (A) normal spectrum; (B) after addition of Eu(thd)3

resulting shifts will depend upon the strength of the interaction, the geometry of the adduct and the LSR to substrate ratio. The acid character of the lanthanide chelate and the donor and steric properties of the substrate define the strength of the acidbase adduct. Usually more stable complexes are found as the ionic radius of the metal decreases. Fluorinated diketonates increase the Lewis acidity of the lanthanide ion with respect to the protonated derivatives. This explains why Eu(fod)3 (see Table 1) is the mostly commonly used LSR.

substrate nuclei, and has then been related to the value of the hyperfine coupling constant. The dipolar or pseudocontact term, which depends on the spatial proximity of a given nucleus to the paramagnetic centre and on the anisotropy of the magnetic susceptibility tensor. In order to extract structural information from LIS data there has to be an accurate determination of the latter term. The first step in the procedure consists of the determination of the all-bound shifts ∆iLSRS. In the presence of the equilibrium between the substrate (S) and the lanthanide shift reagent (LSR)

Contributions to the lanthanide-induced shift

The shift induced by the lanthanide ion on the resonances of the organic substrate results from three contributions. A diamagnetic term corresponding to the coordination shift. On the 1H resonances it is usually upfield and rather small. The contact or Fermi term, which depends on the delocalization of unpaired electron density on the

the observed ∆Gi are proportional to the term ([LSR S]/[S])∆iLSRS. Thus ∆Gi are measured for various LSR to S ratios (U) and graphs are obtained by plotting the observed shifts ∆Gi = Giobs − Giο(where Giο is the chemical shift measured in the absence of the LSR or, better, in the presence of diamagnetic La or Lu analogue) as a function of U.

CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR 225

If an axial symmetry of the dipolar magnetic field in the LSRS adduct is assumed, the dipolar shifts are proportional to the term (3 cos2I 1)/ri3. In this case, the principal magnetic axis of the lanthanide is taken as collinear with the lanthanideligand bond. ri is the distance between a given nucleus and the paramagnetic ion and I is the angle between the ri vector and the Lnligand bond. Many systems have been successfully investigated on the basis of this simplifying assumption. The occurrence of axial symmetry is justified as a result of time averaging effects. This means that the fast rotation of the substrate along the coordination axis introduces an effective axial symmetry, where I now refers to the angles between ri and the rotation axis rather than the actual symmetry axis. In the more general case of a rhombic, nonaxial symmetry, the dipolar contribution to the (allbound) LIS is described as follows:

where ri, Ti and :i are the usual spherical coordinates of the i nucleus i the reference frame of the principal magnetic susceptibility tensor. K1 and K2 are coefficients related to the anisotropy of the magnetic susceptibility tensor. Several computer programs have been written to find the molecular geometry that best fits the set of experimental ∆i values. A decrease in temperature causes an increase in the lanthanide-induced chemical shifts, which usually provides an enhanced resolving power of the LSR. This behaviour results (i) from an increase of the concentration of the adduct and (ii) from the temperature dependence of the paramagnetic shift (the dipolar shift shows a 1/T2 dependence, whereas the contact term displays a 1/T dependence). Diketonates are insoluble in water. Diglycolates have been proposed as substitutes for studies in aqueous solutions. Another possibility deals with the use of lanthanide aquo-ions, and perchlorates and nitrates have been used in several cases: they are soluble at pH < 6 and provide good shifts for a number of anionic functions. Eventually, good results in aqueous solutions have been obtained by using EDTA chelates which work at pH values as high as 10. Chiral shift reagents

It is well established that a method to distinguish enantiomers by NMR is to convert them into

diastereoisomers by means of appropriate optically active reactants. Thus it was suggested that the use of chiral LSR (LSR*) can induce shift differences in the spectra of enantiomers:

R and S give the same NMR spectrum as they are mirror images, whereas LSR*R and LSR*S are diastereoisomers and then are characterized by different NMR properties. In the presence of fast exchange between free and bound forms, R and S will give rise to different signals as they represent average values between the free and the bound forms. The enantiomeric shift difference between the ∆G LIS for the R and S configuration is usually indicated as ∆∆G. The observed ∆∆G values depend on the equilibrium constants for the formation of the labile complexes between the enantiomers and the chiral reagent, and on the shift differences between the diastereomeric complexes. The first chiral LSR was reported by Whitesides and consisted of a camphor-based Eu3+ complex. Since then, several chiral LSR* have been proposed and many of them are commercially available (Figure 2 and Table 2). In some cases, it has been found that enantiomeric resolution has been obtained also with the use of an achiral reagent such as Eu(fod)3. This behaviour has been explained in terms of the formation of ternary 1:2 adducts which are no longer mirror images (such as LSRRR and LSRRS). In principle, another explanation may be possible on the grounds that Eu(fod)3 may be considered as a racemic mixture of enantiomers. Lanthanide shift reagents for alkali metal ions

An interesting extension of the use of LSR deals with the separation of the resonances of 7Li+, 23Na+ and 39K+ present in different biological compartments. Such NMR-active metal cations are of clinical importance, but from their routine NMR spectra it is not possible to distinguish whether they are in the intra- or the extracellular compartments. The addition of a paramagnetic complex that selectively distributes in one compartment only, can remove such signal degeneracy. In fact, the interaction with the paramagnetic species induces a shift in other resonances of the alkaline ions present in the same compartment as the LSR (Figure 3). Up to now in laboratory practice, four complexes have been

226 CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR

Table 2

Chiral shift reagents of the β-diketonato type

studied in detail and applied to answer several questions of biomedical interest: the Dy(III) complexes of PPP5- (triphosphate) and TTHA6 (triethylenetetraaminehexaacetate) and the Dy(III) and Tm(III) complexes of DOTP5 (tetraazacyclododecaneN, N′, N″, N″′ tetrakis (methylenephosphonate)). The most effective reagent so far reported is the chelate Dy(PPP)27, first introduced by Gupta and Gupta in 1982. However, this metal complex has been proved to present several disadvantages that prevent its use in living animals or humans. First of all it is rather toxic, probably due to irreversible ligand dissociation promoted in vivo, with subsequent release of the metal ion. Furthermore, the complex is involved in several protonation equilibria and shows competition with endogenous divalent cations (Mg2+, Ca2+) which reduces its resolution ability. The Dy(TTHA)3− complex is much less toxic, as a consequence, probably, of the high stability constant of Dy3+ with this multidentate ligand. However, the reduced value of the negative charge on the complex and the fact that it is mainly localized on unbound carboxylic groups, away from the paramagnetic centre, makes this metal chelate less effective in removing the signal degeneracy. This implies that higher doses of the LSR are needed for an optimum resolution. The Dy(III) and Tm(III) complexes of DOTP represent an important

Figure 2 1H NMR spectra of α-phenylethylamine in CCI4 after addition of tris[(3-tert-butylhydroxymethylene)-D-camphorato]europiumIII. Molar ratio LSR* to substrate 0.5: (A) S-form; (B) racemic mixture.

Figure 3 23Na NMR spectra at 2.1 T and 39°C of whole human blood before (A) and after (B) the addition of Dy(DOTP)5- (5 mM). Intra- and extracellular sodium resonances are resolved.

CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR 227

improvement in the search for safe and effective LSRs for metal cations of biological relevance. In fact, these metal chelates are extremely resistant to dissociation processes over a wide range of pH and present in axial symmetry and interaction sites for the cations very close to their 4-fold axis of symmetry, which ensures a maximum shifting efficiency. Indeed Dy(DOTP)5 produces a magnitude of shift analogous to that observed for Dy(PPP)27 at the same concentration. However, the efficiency of the Dy(III) chelate has been shown to be markedly quenched by stoichiometric amounts of Ca2+, and thus the Tm(III) complex, less sensitive to these effects, has found wider applications. In conclusion, it is worth commenting that, although several studies have been possible by the use of the available LSRs, much remains to be done in terms of synthetic strategy and ligand design in the search for safer and more effective compounds.

Relaxation reagents An alternative exploitation of paramagnetic complexes as auxiliary NMR reagents deals with their effect on the relaxation of substrate nuclei. Indeed, in the presence of a relaxation reagent the electron nucleus relaxation may easily become so efficient that it predominates over the intrinsic relaxation processes of a given nucleus. Such effects may be exploited either on the solute or on the solvent resonances. A number of applications based on the use of relaxation reagents have been reported, ranging from structural assignments to nuclear Overhauser effect (NOE) quenchers for correct intensity measurements, from the assessment of the binding to proteins to the separation of resonances from different compartments. A particularly important class of relaxation reagents is represented by the contrast agents for magnetic resonance imaging (MRI). The first example dealing with the use of paramagnetic species to shorten the relaxation time of the solvent nuclei deals with the very first detection of the NMR phenomenon, as Bloch used iron(III) nitrate to avoid saturation of the water proton signal. Then in the 1970s it was rather common, in organic and organometallic chemistry, to add to the substrate solution a reagent such as Cr(acetylacetonate)3 (Cr(acac)3) or Fe(acac)3 in order to shorten the T1 of 13C and 15N resonances without introducing broadening of the signal. In Figure 4 the 13C intensities of the Fe(CO)5 signal, measured at 2.1 T, in the presence of variable amounts of Cr(acac)3, are reported. At this field the T1 value of the 13CO groups is very long as it is determined by the chemical shift anisotropy term; by

Figure 4 13C NMR spectra (A) and integrated intensities (B) of a Fe(CO)5 sample with different added concentrations of Cr(acac)3. Different symbols refer to independent measurements.

increasing the magnetic field strength it becomes markedly shorter. Thus, the availability of high magnetic fields has limited to some extent the use of relaxation reagents for the purpose of an overall shortening of the relaxation times. The presence of a relaxation reagent induces a dominant relaxation path for all the resonances of a given substrate present in solution and this may be exploited for quantitative NMR determinations. In fact, this method eliminates the problems associated with a correct intensity measurement when the various resonances are endowed with different relaxation times and different NOE enhancements. Another application of relaxation reagents deals with the selective removal of a given spin-coupling pattern, i.e. the interaction with a paramagnetic reagent produces a so-called chemical decoupling mechanism. In fact, in an AX coupled system the multiplicity of each resonance may be removed when the mean lifetime of the coupled nucleus in a given state (α or β) becomes shorter than the inverse of the coupling constant JAX. Therefore, the occurrence of efficient relaxation induced by the paramagnetic reagent causes the removal of the effective coupling. Theory of paramagnetic relaxation

The quantitative description of relaxation effects induced by paramagnetic species presents theoretical problems that are far from trivial. However, it is customary to resort to the simplified approach developed earlier by Solomon and Blombergen to

228 CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR

account for the relaxation enhancement of solvent nuclei of aqueous solutions of paramagnetic metal ions. On the basis of this approach the observed solvent relaxation rate is given by the sum of three contributions:

where R is the relaxation rate in the absence of the paramagnetic compound, R represents the paramagnetic contribution due to the exchange of water molecules in the inner coordination sphere with water molecules in the bulk, and R is the diffusioncontrolled contribution from water molecules in the outer coordination sphere of the paramagnetic centre. In the absence of a coordinative interaction between the paramagnetic metal ion and the substratesolvent molecules, only the latter term contributes to the observed paramagnetic relaxation enhancement. In this case the relaxation enhancement is exclusively due to the electronnucleus dipolar interaction between the metal ion and the solvent molecules diffusing in the outer coordination sphere of the complex. This interaction is modulated by the translational diffusive motion of solute and solvent and by the electronic relaxation time. An analytical expression for this contribution has been derived by Freed and co-workers in a study of the relaxation enhancement of solvent molecules in solutions containing stable nitroxide radicals.

In the expression above, Cos is constant for a given observed nucleus, a is the distance of closest approach between the paramagnetic centre and the diffusing solvent molecules and D is the relative solutesolvent diffusion coefficient. The dependence on the electronic relaxation time is expressed by the nonLorentzian spectral density functions J(Z). As far as the inner sphere contribution is concerned, R is determined by the relaxation time T1M of the nuclei of the substratesolvent complex and by its lifetime τM, weighted by the molar ratio of the bound substratesolvent:

where [C] is the molar concentration of the paramagnetic reagent, q is the number of interacting

sites, and [S] is the concentration of the substrate solvent complex. In the case of dilute water solutions, when the relaxation enhancement is observed on the solvent nuclei, [S] = 55.54 M. The T1M value in the paramagnetic reagentsubstrate adduct is determined by (i) the dipolar interaction between the observed nucleus and the unpaired electron(s) and (ii) the contact interaction between the magnetically active nucleus and the unpaired electron density localized in the position of the nucleus itself. An analytical form is given by the SolomonBloembergen equation (here for T1; an analogous expression is derived for T2):

where S is the electron spin quantum number, JH and JS are proton and electron magnetogyric ratios, respectively, r is the distance between the metal ion and the given nucleus of the coordinated substrate, ZH and ZS are the proton and electron Larmor frequencies, respectively. τc1,2 and τe are the relevant correlation times for the time-dependent dipolar and contact interactions, respectively:

where τR is the reorientational correlation time and τS1 and τS2 are the longitudinal and transverse electron spin relaxation times. The contact term is often negligible, especially when dealing with a paramagnetic ion bound to a macromolecule because the hyperfine coupling constant A/ is in the MHz range, whereas the coefficient of the dipolar term is one order of magnitude larger. The protons relaxation enhancement (PRE) method

Upon interacting with a macromolecule the relaxation induced by paramagnetic species usually displays remarkable changes, primarily related to the increase of the molecular reorientational time τR on going from the free to the bound form. This may

CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR 229

result in a strong increase of the R term, from which it is possible to assess the affinity (and the number of binding sites) between the interacting partners. Quantitatively, in the presence of a reversible interaction between the paramagnetic species and the macromolecule, the observed enhancement depends on both the molar fraction of the macromolecular adduct χb and the solventwater relaxation in the all-bound limit R . The increase in the relaxation rate is expressed by the enhancement factor H∗:

The asterisk indicates the presence of the macromolecule in solution. The enhancement factor ranges from 1 (no interaction) to Hb= R /R1p in the allbound limit (strong excess of macromolecule).

The determination of the binding parameter KA (association constant) and n (number of independent sites characterized by a given KA value) from the mass action law is straightforward. Given the association equilibrium between the metal complex

M and the protein (enzyme) E,

where n time [E] gives the concentration of each class of sites on the macromolecule. Superscript b and subscripts f and t indicate bound, free and total, respectively. The experimental procedure consists of the determination of the enhancement factor H∗ through two distinct titrations. In the first titration, a rectangular hyperbola describing the change of H∗ as the [E]t to [M]t ratio increases is reported (Figure 5). The treatment of the obtained binding isotherm yields the value of Hb and the product nKA. In the second titration, the behaviour of H∗is monitored at fixed macromolecule concentration by changing the metal complex concentration. Results are conveniently expressed under the form of a Scatchards plot:

The value of r may easily be calculated once the all-bound enhancement Hb is known. In the same way, the free [M]f concentration is obtained from the total [M]t complex concentration.

By plotting r/[M]f vs. r a straight line is obtained whose x-axis intercept gives the n value and whose slope is equal to KA. Contrast agents for magnetic resonance imaging Figure 5 Direct (A) and inverse (B) PRE titrations: by plotting H* vs. macromolecule concentration a rectangular hyperbola is obtained, whose analysis yields the Hb value together with the nKA product. The Scatchard's plot obtained from the inverse titration allows the determination of n and KA separately.

Magnetic resonance imaging (MRI), one of the most powerful tools in modern clinical diagnosis, is based on the topological representation of NMR parameters such as proton density and transverse and

230 CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR

longitudinal relaxation times. Differences in these parameters allow impressive anatomical discrimination to be made, and make it possible to distinguish pathological from healthy tissues. The potential of MRI is further strengthened by the use of suitable contrast agents (CAS), compounds that are able to alter markedly the magnetic properties of the region where they are distributed. Among them, paramagnetic Gd3+ complexes (seven unpaired electrons) are under intense scrutiny because of their ability to enhance the proton relaxation rates of solvent water molecules (Figure 6). Their use provides the physicians with further physiological information to be added to the impressive anatomical resolution commonly obtained in the uncontrasted images. Thus, administration of Gd-based contrast agents has entered into the pool of diagnostic protocols and is particularly useful to assess organ perfusion and any abnormalities in the bloodbrain barrier or in kidney clearance. Several other applications, primarily in the field of angiography and tumour targeting, will soon be available in clinical practice. Nowadays about 35% of MRI examinations make use of contrast agents, but this percentage is predicted to increase further following the development of more effective and specific contrast media than those currently available. In order to be used as a CA for MRI a Gd3+ complex must fulfil two basic requirements: (i) it must have at least one coordinated water molecule in fast exchange (on the NMR relaxation timescale) with the solvent bulk, and (ii) the Gd3+ ion must be tightly chelated to avoid the release of the metal ion and the ligand which are both potentially harmful. Figure 7 reports the schematic structures of four ligands whose Gd3+ chelates are currently used as CAs for MRI. The overall PRE of water protons (R + R , see Theory of paramagnetic relaxation section above) referred to a 1 mM concentration of Gd3+ chelate is called millimolar relaxivity (hereafter relaxivity). At the magnetic field strength currently employed in MRI applications, the relaxivity of a Gd3+ complex is roughly proportional to its molecular weight, i.e. it is determined by the value of the molecular reorientational time τR. It follows that the search for high relaxivities has been addressed to systems endowed with slow tumbling motion in solution. This condition may be met either (i) by linking a Gd3+ chelate to high molecular weight substrates like carboxymethylcellulose, polylysine or dendrimers, etc. or (ii) by forming noncovalent adducts with serum albumin. In the latter case the Gd3+ chelates are designed in order to introduce on the surface of the ligand suitable functionalities able to strongly interact with the serum protein. In addition to providing high relaxivities (which, in turn, allows

Figure 6 Male patient, 54 year old, 67 kg with multiple hepatic metastases from a neuroendocrine tumour of the pancreas: images are taken before (A) and 90 minutes after (B) intravenous administration of 0.1 mmol kg–1 of a liver-specific Gd3+ chelate.

the administered doses of CA to be reduced) the formation of adducts between Gd3+ complexes and albumin are of interest for the design of novel angiographic experiments for which a reduced clearance time and a better compartmentalization in the circulating blood is required.

List of symbols a = distance of closest approach between the paramagnetic centre and the diffusing solvent molecules; D = relative solutesolvent diffusion coefficient; H = hyperfine shift; HS = hyperfine shift scalar coupling; HD = hyperfine shift dipole coupling; Ki = interaction constant between LSR and S ; K1 and K2 = coefficients related to anisotropy of magnetic susceptibility tensor; KA = association constant; n = number of independent sites characterized by a given KA value; q = number of interacting sites; ri = distance between nucleus and paramagnetic ion; ri, Ti and :ι = spherical coordinates of i nucleus in the reference frame of the principal magnetic

CHEMICAL SHIFT AND RELAXATION REAGENTS IN NMR 231

Figure 7 Schematic structure of the Gd3+ chelates used in clinical practice. DOTA = 1,4,7,10-tetraazacyclododecane-N,N′,N′′,N′′′tetraacetic acid; DTPA = diethylenetriaminepentaacetic acid; HPDO3A = N″′-(2-(1-hydroxypropyl))-1,4,7,10-tetraazacyclododecaneN,N′N″-triacetic acid; DTPABMA = N,N″-bis(N-methylcarbamoylmethyl)-diethylenetriamine-N,N′,N″-triacetic acid.

susceptibility tensor; R = relaxation rate in absence of paramagnetic compound; R = diffusion-controlled contribution to relaxation rate from water molecules in the outer coordination sphere of the paramagnetic centre; R = paramagnetic contribution to relaxation rate due to exchange of water molecules in the inner coordination sphere with those in the bulk; R = solventwater relaxation in the allbound limit; S = substrate; T1M = relaxation time of the protons of the water molecule in the first coordination sphere of the complex; T1 = relaxation time; χb = molar fraction of macromolecular adduct; ∆iLSRS = all-bound shift of the LSR-S complex; ∆δi = observed LSR-induced shifts; ∆∆G = enantiomeric shift difference between the ∆G LIS for R and S configuration; H* = enhancement factor; I = angle between ri vector and Lnligand bond or rotation axis; JH and JS = electron and proton magnetogyric ratios; U = LSR to S ratio; Wc1,2 and We = relevant correlation times for time-dependent dipolar and contact interactions; WM = residence lifetime of the water molecule in the first coordination sphere of the complex; τR = reorientational correlation time; WS1 and WS2 = longitudinal and transverse electron spin relaxation times; ωH and ωS = proton and electron Larmor frequencies. See also: Contrast Mechanisms in MRI; EPR Spectroscopy, Theory; MRI Applications, Biological; MRI

Applications, Clinical; MRI Applications, Clinical Flow Studies; MRI Theory; NMR Relaxation Rates; NMR Spectroscopy of Alkali Metal Nuclei in Solution; Nuclear Overhauser Effect.

Further reading Aime S, Botta M, Fasano M and Terreno E (1998) Lanthanide (III) chelates for NMR biomedical applications. Chemical Society Reviews 27: 1929. Bertini I and Luchiant C (1986) NMR of Paramagnetic Molecules in Biological Systems. Menlo Park: Benjamin/Cummings. Cockerill AF, Davies GLO, Harden RC and Rackham DM (1973) Lanthanide shift reagents for nuclear magnetic resonance spectroscopy. Chemical Reviews 73: 553 588. Martin ML, Delpuech J-J and Martin GJ (1980) Practical NMR Spectroscopy, pp 377409. London: Heyden. Peters JA, Huskens J and Raber DJ (1996) Lanthanide induced shifts and relaxation rate enhancements. Progress in NMR Spectroscopy 28: 283350. Sherry AD and Geraldes CFGC (1989) Shift reagents in NMR spectroscopy. In: Bünzli J-CG and Choppin GR (eds) Lanthanide Probes in Life, Chemical and Earth Science. Amsterdam: Elsevier. Von Ammon R and Fisher RD (1972) Shift reagents in NMR spectroscopy. Angewandte Chemie, International Edition in English 11: 675692.

232 CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY

Mass spectra of chemical compounds have a high information content. This article describes computerassisted methods for extracting information about chemical structures from low-resolution mass spectra. Comparison of the measured spectrum with the spectra of a database (library search) is the most used approach for the identification of unknowns. Different similarity criteria of mass spectra as well as strategies for the evaluation of hitlists are discussed. Mass spectra interpretation based on characteristic peaks (key ions) is critically reported. The method of mass spectra classification (recognition of substructures) has interesting capabilities for a systematic structure elucidation. This article is restricted to electron impact mass spectra of organic compounds and focuses on methods rather than on currently available software products or databases.

Introduction Mass spectrometry (MS) is the most widely used method for the identification of organic compounds in complex mixtures at the nanogram level and below. NMR and, sometimes, also infrared spectroscopy provide better structural information than MS data but these spectroscopic methods cannot (or only with a substantial loss of sensitivity) be coupled with chromatographic separation techniques. Despite the great amount of information contained in mass spectra it is difficult to extract structural data because of the complicated relationships between MS data and chemical structures. The fragmentation processes which finally result in the measured data characterize MS as a chemical method, in contrast with NMR and IR. Chemical effects are, in general, more difficult to describe and to predict than physical ones. The occurrence of rearrangement reactions means that structure identification from mass spectra is often difficult or impossible; furthermore functional groups do not always produce the same peaks. This background explains why structure elucidation in organic chemistry is mainly based on 13C NMR data, while H

MASS SPECTROMETRY Methods & Instrumentation

NMR, IR, UVVIS and MS data are considered only to a much lesser extent. The aim of spectra evaluation can be either the identification of a compound (assuming the spectrum is already known and available) or interpretation of spectral data in terms of the unknown chemical structure (when the spectrum of the unknown is not available). Identification is performed best by library search methods based on spectral similarities; a number of MS databases and software products are offered for this purpose and are routinely used. The more challenging problem is the interpretation of a mass spectrum, which is still a topic of research projects in chemometrics and computer chemistry. No comprehensive solutions are yet available and these methods are only rarely used in routine work. Four groups of different strategies have been applied to the complex problem of substructure recognition (or the recognition of more general structural properties) from spectral data. (1) Knowledge-based methods try to implement spectroscopic knowledge about spectrastructure relationships into computer programs. Because of the lack of generally applicable rules this approach was not successful in MS. However, spectroscopic knowledge has been extensively applied in other methods to guide the construction of mathematical models, and for optimizing model parameters. (2) Appropriate interpretive library search techniques can be used to obtain structural information if the unknown is not contained in the library. (3) Correlation tables containing characteristic spectral data (key ions) together with corresponding substructures have met with only limited success because a specific structural property does not always give the same spectral signals. (4) Spectral classifiers are algorithms based on multivariate classification methods or neural networks; they are constructed for an automatic recognition of structural properties from spectral data. In principle, mass spectral data are fully determined by the chemical structure of the investigated

CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY 233

compound and the experimental spectroscopic conditions:

For problems of practical interest, however, the relationship F can neither be formulated as a mathematical equation nor as an algorithm. Consequently, methods for the prediction of the mass spectrum from the chemical structure have had only very limited success. A similar difficult situation arises for the inverse problem, spectra interpretation, Figure 1. Only for selected cases can the relationship

be established and applied to mass spectra of unknowns. Some relationships f for different structural properties have been developed by the use of statistical methods or neural networks rather than by the influence of spectroscopic knowledge. Because of the above-mentioned difficulties the intervention of the human expert is still required in mass spectra interpretation. However, this situation causes a number of drawbacks. Human experts are not sufficiently available and are expensive; furthermore their success rate can hardly be quantified. Automated instruments with high throughput are able to produce huge amounts of data, and analytical interest in complex mixtures (environmental chemistry, food chemistry, combinatorial synthesis) is rapidly increasing. Thus the extensive use of computer-assisted methods for data interpretation is highly desired.

Figure 1

Characteristic mass spectra signals A number of masses in the low mass region (key ions) are considered to be characteristic for certain substructures or classes of compounds. In addition, mass differences (key differences) between the molecular ion and abundant fragment ions or between abundant fragment ions are often related to functional groups. Correlation tables containing such spectral data and the corresponding chemical structures are contained in several textbooks on MS. The use of these tables is widespread and often helpful in MS; however, the capability of this approach must not be overestimated. The following example demonstrates the potentials and drawbacks. From the NIST Mass Spectra Library a random sample containing 5000 compounds with a phenyl group (class 1) and 5000 compounds without a phenyl group (class 2) have been selected; from these data it was estimated whether the signal at m/z 77 (corresponding to the ion C6H5+) can be used to predict the presence or absence of a phenyl group in the molecule. Let n1 be the number of compounds from class 1 exhibiting a peak at m/z 77 in a given intensity interval, and n2 be the corresponding number for compounds from class 2. Assuming for an unknown compound that the a priori probability for belonging to class 1 is 0.5 the probability p(phenyl | I77) can be calculated from the signal at m/z 77 as

Table 1 shows results for this example using four intensity intervals. Depending on the intensity of the peak at m/z 77, the probability for the presence of a phenyl group decreases or increases. If a peak at m/z

Relationships between mass spectrum and chemical structure.

234 CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY

Table 1 Significance of a peak at m/z 77 for the presence of a phenyl group in a molecule

absence of a key ion in general does not indicate the absence of the corresponding compound class.

Intensity interval (%B)

p(phenyl | I77)

I≤1

497

2486

0.17

Mass spectra similarity search

1 90%B and peak height at m/z 105 > 30%B. The result is a notordered hitlist containing all spectra that exactly match the search profile. (2) A numerical similarity between two spectra is used to find the reference spectra most similar to the spectrum of the unknown (Figure 2). In this case the resulting hitlist is ordered according to the value of the similarity criterion. Of course any type of search can be combined with additional search criteria (relative molecular mass range, restrictions to the molecular formula, name fragments, etc.). Unfortunately, the capabilities for chemical structure searches are still very limited in commercially available MS databases. The performance and effectiveness of a spectral library search depends on the quality of the given spectrum, the content of the spectral library, the

The terms n1 and n2 are the number of spectra (compounds) with a peak at m/z 77 in the given intensity interval for compounds that contain a phenyl group and those that do not, respectively; p(phenyl | I77) is the a posteriori probability for the presence of a phenyl group (assuming an a priori probability of 0.5)

77 is absent or has an intensity not higher than 1% of the base peak (%B) the probability for the presence of a phenyl group is still 17%. Intensities above 60%B strongly indicate a phenyl group (91% correct assignments to class 1). However, most spectra have peak intensities between these limits and therefore do not allow a reliable answer about the presence or absence of a phenyl group. A systematic examination of about 40 key ions between m/z 30 and 105 can be summarized as follows: (1) only a small number of key ions provide significant structural information, and (2) the

Figure 2

Spectral library search.

CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY 235

Table 2 bases

Common quality problems with mass spectra in data-

Spectrum reduced to a small number of (large) peaks Limited mass range (start mass too high) Peaks present far above the relative molecular mass Poor dynamics of peak intensities Wrong isotope peak intensities Tilted peak intensities Additional peaks from impurities or instrument background Small peaks missing Wrong mass assignment Illogical mass differences Spectrum does not match with given compound Missing experimental data Missing structure data Too many replicate spectra

used similarity criterion and the implemented software technique. A fundamental restriction is given by the limited structural information present in mass spectra, for instance isomers can often not be distinguished by their mass spectra. Libraries

Any spectroscopic retrieval system in chemistry depends on an appropriate structural diversity of the reference data. Commercially available MS databases today contain about 40 000 up to almost 300 000 spectra. Some libraries contain a number of replicate spectra (collected from different sources), especially for common compounds. The compositions of existing libraries (except a few dedicated small spectra collections) are not systematic. Some compound classes are very well represented (for instance hydrocarbons) while others are not but may be of major interest to a particular user. Good software support for building user-libraries is therefore essential. Table 2 lists common problems of quality deficiencies in MS reference data.

no reference spectrum fulfills all restrictions it is possible to go back one step). The resulting list of spectra can be ordered by applying an additional similarity criterion. The method allows (and requires) the interaction of the user; consequently spectroscopic knowledge (for instance about the background peaks) can be easily considered; however, automated processing of spectra is not possible. The use of appropriate sorted index files avoids time-consuming sequential searches and makes the method very fast. An example for this method is presented in Figure 3. Spectra similarity criteria

An infinite number of different similarity measures for matching two mass spectra is possible. In reality, about a dozen similarity criteria have been described and tested; however, full details of the algorithms that are actually implemented in commercial software are seldom described. Some similarity criteria are purely mathematically oriented, others are influenced by spectroscopic ideas. A single criterion cannot fit all requirements: on the one hand, a compound which is contained in the library should be retrieved even if the spectrum deviates in some parts from the reference spectrum; on the other hand, the user expects hits with chemical structures similar to the unknown even though they exhibit different spectra. In the ideal case the similarity criterion has a practicable sensitivity to chemical structures: identical compounds measured under different spectroscopic conditions appear more

Peak search

A simple and evident method for library search of mass spectra applies user-selected restrictions for the presence of peaks. Initially, a characteristic peak in the spectrum of the unknown is selected (typically one at a high or unusual mass number and with a not too small intensity) and an intensity interval for reference spectra is defined. The peak search software answers with the number of reference spectra containing a peak at the given mass and in the given intensity interval. This procedure is continued until a reasonable number of hits is reached (in cases where

Figure 3 Peak search example using the NIST mass spectral database. The mass spectrum of a hypothetical unknown is from caffeine contaminated with a phthalate. Manual selection of relevant peaks easily allows the spectroscopist to consider probable contaminations (peaks at m/z 149, 167). The correct solution is found after the input of four peaks by excluding the typical phthalate peaks. Note the wide intensity intervals applied. Should a peak at m/z 149 be required the correct compound is not found.

236 CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY

similar than two compounds that are slightly different in their structure. The two main types of mass spectral data, peak heights and masses, have different reliability. While the presence of a peak at a certain mass number is well reproducible the peak height often varies greatly under slightly different experimental conditions. Spectra similarity criteria have to consider this aspect. For similarity searches an efficient ordering of the spectra in the library is not possible and, therefore, a rather time-consuming sequential search is necessary. Some speed-up is possible by a pre-search in which, for instance, reference spectra are quickly eliminated that do not have a minimum number of peaks in common with the spectrum of the unknown. Before calculating the spectra similarity the spectrum of the unknown is treated by cleaning procedures, for instance peaks at m/z 28, 32 and 40 (probably from N2, O2 and Ar respectively) and peaks from column bleeding are removed, and the peak heights are normalized to percents of the base peak. The selection of relevant peaks and weighting of peak heights is described below. Let UAm and RAm be the (weighted) abundances at mass m in the spectrum of the unknown (U) and the reference spectrum (R), respectively. Summation in Equations [2] to [4] is over all selected masses m. The most used similarity criterion for mass spectra is based on the correlation coefficient:

Figure 4 explains this similarity criterion: each mass is considered as a point in a coordinate system with one axis for the abundances of the unknown and the other for the reference. For identical spectra all points are located on a straight line that passes through the origin (S1 = 1). If the two spectra are different the correlation coefficient for the regression line as given in Equation [2] is a measure of the similarity of the spectra. The value of S1 ranges from 0 to 1 because the regression line is forced to pass through the origin and all abundances are 0 or positive. The important range of S1 between 0.9 and 1 can be widened by calculating S ; for practical reasons S1 or S are often multiplied by 100. A peak list with nominal (integer) masses can be considered as a point in a q-dimensional space, with q equal to the maximum relevant mass number. The coordinates of the point are given by the abundances (weighted peak heights). Because the value of q is high the q-dimensional space cannot be visualized

directly but can be handled by rather simple mathematics. A two-dimensional simplification is used in Figure 5 to demonstrate this view of MS data. Masses 43 and 58 have been selected for the coordinates; each spectrum (from the example in Figure 4) can be represented by a point or by a vector (starting at the origin). Similarity criterion S1 is equal to the cosine of the angle between two spectral vectors. Alternative similarity measures are the Euclidean distance S2:

and the city block distance S3:

The similarity criterion used in the PBM (Probability Based Matching System) software is based on socalled uniqueness values for masses and abundances. The uniqueness of a signal is equivalent to the information content (as defined in information theory) and is given by the negative logarithm of the probability of the occurrence of this signal; this means that a low probability corresponds to a high uniqueness. Probabilities for masses and intensity intervals have been estimated from the database used. Matching of spectra is performed by using selected peaks that exhibit highest uniqueness values for mass and intensity. The MS database system MassLib uses a composite similarity criterion containing the similarity criterion S1 together with measures for the number and intensity sums of peaks that are common or not in both spectra. The algorithm has been optimized to give good results for identification as well as for interpretation. In the STIRS (Self-training Interpretive and Retrieval System) system for 26 data classes, specific similarity criteria are defined using characteristic masses or mass differences. For each of these criteria the most similar reference spectra are searched with the aim of obtaining information about the presence of substructures. Weighting

Intensity scaling and mass weighting is used to consider the different significances of mass numbers and peak heights. In general, peaks in the higher mass range are more important than peaks in the lower mass range; because large peaks dominate the values of most similarity criteria, peak heights are often scaled to enhance the influence of small peaks. A

CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY 237

Figure 4 Comparison of the mass spectrum from an unknown with two reference spectra using the similarity measure S1 (based on the correlation coefficient). Axes in the two plots at the bottom: horizontal, peak height %B of unknown; vertical, peak height %B of reference. Each point corresponds to a mass number with a peak in at least one of the two compared spectra.

general transformation of intensities Im at mass m is given by Equation [5]

been obtained with ν = 2 and w = 0.5 [see Stein (1994) in the Further reading section]. Selection of revelant masses

Optimum values for v and w depend on the used similarity citerion; for S1 (Eqn [1]) good results have

The selection of masses used to calculate a spectral similarity may greatly influence the result. The user

238 CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY

Figure 6 Different selection of masses for the calculation of spectral similarity (schematic). The fit search considers only masses with peaks in the reference spectrum and therefore is insensitive to impurities in the unknown. Figure 5 Mass spectra can be considered as points or vectors in a multidimensional spectral space. For simplicity only two mass numbers (43, 58) have been selected in this example. Am, abundance (peak height in %B) at mass m; U, spectrum from unknown; R1, reference spectrum of propanal; R2, reference spectrum of acetone. Measures for spectral similarity are the Euclidean distance (d), the city block distance (∆43 + ∆58) or the cosine of angle α (equivalent to S1 in Eqn [2]).

often can decide between three modes for mass selection (Figure 6). (a) All masses for which at least one of the compared spectra has a peak with an intensity above a defined threshold are considered. High similarity values are obtained only if the spectra are almost identical; this approach therefore is suited for identity searches (sometimes also called purity searches). (b) Only masses with a peak in the reference spectrum are considered. High similarity values are obtained if the reference spectrum is almost completely contained in the spectrum of the unknown. Non-matching peaks in the unknown do not influence the result; therefore this approach is routinely applied if the unknown may contain impurities or is a mixture of compounds (fit search, reverse search). (c) Only masses with a peak in the spectrum of the unknown are considered. This method is not tolerant of additional peaks in the reference spectrum. From a high similarity value one may conclude that parts of the chemical structure of the reference compound are present in the unknown (interpretive search, forward search). Simple selection methods of peaks such as k highest peaks in the spectrum (or in given mass intervals) have been applied to save data storage and computation time but are no longer important. A special selection of masses is used in the MassLib system: a

peak at mass m is considered to be significant if its intensity is higher than the average intensities at masses m − 14 and m + 14. Figure 7 presents an example of an MS library search in which the unknown was contained in the library.

Mass spectra classification Overview

When the unknown compound is not contained in the spectral library then pure identification methods are less useful. For unknown unknowns interpretive search systems or classification methods are required to obtain structural information that can be used for constructing molecular structure candidates. Such candidates are usually created manually by applying spectroscopicchemical knowledge and intuition. A serious drawback of this strategy is that the solution is rarely complete and the procedure hardly can be documented or verified. The most important systematic approach for structure elucidation of organic compounds is still based on the DENDRAL project, in which, for the first time, artificial intelligence principles have been applied to complex chemistry problems (Table 3). The central tool is an isomer generator software capable of generating an exhaustive set of isomers from a given molecular formula. The generator also has to consider structural restrictions, which are usually obtained from spectral data. Substructures which have to be present in the unknown molecular structure are collected in the so-called goodlist while forbidden substructures are put into the badlist. Mass spectrometry can contribute to this approach in various aspects: the molecular formula can be determined from high resolution data, and structural information can be derived even from low resolution data.

CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY 239

Figure 7 Example of a library search by MassLib. A mass spectrum from testosterone has been considered as unknown and searched in a library consisting of 130 000 spectra (including duplicates). A mass spectrum from testosterone is contained in this library and has been found as the first hit (most similar spectrum). The other hits are from the same compound class and demonstrate the interpretive capabilities of this system.

The two most important computer-assisted strategies for the recognition of structural information from mass spectral data are: (a) the structures in the hitlist from a spectra similarity search are used to estimate the probability of substructures in the unknown. (b) Random samples of mass spectra are first characterized by a set of variables and then methods from multivariate statistics or neural networks are applied to develop spectral classifiers. Substructure recognition by library search

Library search systems have been developed or adjusted for the purpose of obtaining structurally similar compounds in the hitlist. In such methods an evaluation of the molecular structures of the hitlist

compounds can provide useful information about the presence or absence of certain substructures in the unknown. It is not trivial to define or select substructures that should be considered for this purpose. For the STIRS system, considerable effort has gone into the search for substructures that can be successfully classified by the implemented spectral similarity search. The MassLib system uses a predefined set of 180 binary molecular descriptors to characterize the similarity of structures. In most investigations a more or less arbitrary set of substructures, functional groups or more general structural properties (compound classes) has been considered. Self-adapting methods that automatically analyse the molecular structures in the hitlist (for instance by searching for frequent and large substructures) have not been used up to now in MS.

240 CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY

Table 3

Scheme for systematic structure elucidation

(1) Determine the molecular formula of the unknown (2) Derive structural restrictions as substructures from spectral and other experimental data and from pre-knowledge about the unknown (3) Make consistency checks of the structural restrictions (4) Add all substructures that are considered to be present in the unknown to the goodlist. Add all substructures that are considered to be absent in the unknown to the badlist (5) Generate all isomers for the given molecular formula that contain all substructures from the goodlist but none from the badlist (6) Test the generated molecular candidate structures: (a) by a comparison of predicted spectra with measured spectra (b) by a comparison of predicted properties with measured properties (c) by considering more complicated structural restrictions that could not be handled by the isomer generator (7) If the number of survived molecular candidates is too large try to create additional structural restrictions and continue at step 3

The number of occurrences of a certain substructure in the hitlist is compared with the corresponding number for the library and a probability is derived for the presence of that substructure in the unknown. This classification method is a variant of the wellknown k-nearest neighbour classification. Each mass spectrum is considered as a point in a multidimensional space; the neighbours nearest to the spectrum of the unknown correspond to the most similar reference spectra in library search. If the majority of k neighbours (k is typically between 1 and 10) contain a certain substructure then this substructure is predicted to be present in the unknown. A drawback of this approach is the high computational effort necessary for classifying an unknown because a full library search is required. The performance has been described by Stein (1995, see Further reading section) as sufficient to recommend it for routine use as a first step in structure elucidation. Multivariate classification

A mass spectral classifier is a part of a computer program that uses the peak list of a low resolution mass spectrum as input and produces information about the chemical structure as output. For such a classification procedure a number of methods are available in multivariate statistics. Many of them have already been applied to various problems in chemistry; classification of mass spectra (with the aim of recognizing chemical compound classes) was one of the pioneering works in chemometrics.

Application of most multivariate classification methods first requires a transformation of the data to be classified (the mass spectrum) into a fixed number of variables (so-called features). Besides the simple use of the peak intensities for these variables more sophisticated transformations have been applied. Problem-relevant numerical spectral features xj are defined as linear or nonlinear functions of the peak intensities; the definitions are based on spectroscopic and/or mathematical concepts. Table 4 contains a list of frequently used spectral features for mass spectra. It has been shown that such features are more closely related to chemical structures than the original peak data; an example will illustrate this. A characteristic fragment in a homologous series of compounds may not appear at the same mass but may be shifted by a multiple of 14 mass units, corresponding to (CH2)n. The often used modulo-14 summation features (also called mass periodicity spectra) consider this fact by an intensity summation in mass intervals of 14. Other spectral features are based on intensity ratios and reflect competing fragmentation pathways; autocorrelation features contain information about mass differences between abundant peaks. The stability of molecular ions and some functional groups can be characterized by features that describe the distribution of peaks across the mass range. The appropriate generation and selection of spectral features is considered to be the most essential part in the development of spectral classifiers. The development of a classifier (Figure 8) for a particular substructure requires a random sample of spectra which is selected from a spectral library. Typically some 100 spectra from compounds containing the substructure (class 1) and some 100 spectra from compounds not containing the substructure (class 2) are necessary. One part of the Table 4 Types of numerical spectral features; a set of spectral features can be used to classify a mass spectrum as to whether a certain substructure or a more general structural property is present or absent in the molecule Modulo-14 intensity sums

xj = ∑ Im + 14n

Logarithmic inten- xj = In Im /Im + ∆m sity ratios

n : 0, 1, 2,…, j, m : = 1, 2,…, 14 Im = max (Im, 1)

Autocorrelation

xj = ∑ ImIm + ∆m /∑ ImIm j, ∆m : 1, 2,…, 50 m : a defined mass range

Distribution of peaks

xj = ∑ Ieven / ∑ Im

even : all even masses m : all masses

xj = 100 / ∑ Im m : all masses Relative base peak intensity xj, spectral feature; Im, intensity in %base peak at mass m; ∆m, mass difference

CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY 241

applications of the classifier to unknowns by the scheme:

Figure 8 Classification scheme for mass spectra. The mass spectrum is transformed into a set of typically 5 to 20 variables (spectral features). A linear or nonlinear mathematical combination of the features results in a discriminant variable. The value of the discriminant variable together with the required maximum statistical risk for wrong answers determines which of the three possible answers is given: ‘class 1’ (substructure present), ‘class 2’ (substructure absent) or ‘answer not possible because error risk is too high’.

random sample is used for a training of the classifier, the other part for testing it. Classification is based on the value of a discriminant variable z which is defined either as a linear function of the q selected spectral features xj:

Simple yes/no classifiers without the possibility to refuse the classification answer are not adequate for the recognition of substructures from mass spectra. Todays performance of mass spectra classification by multivariate methods can be summarized as follows: (1) only a rather small number of substructures can be recognized with a low error rate. (2) Predictions of the absence of a substructure are usually more accurate than predictions of its presence. (3) Erroneous classifications cannot be avoided completely; therefore the intervention of a human expert and the parallel use of other spectra interpretation methods are advisable. (4) For small molecules a systematic and almost complete structure elucidation is sometimes possible by mass spectra classification and by application of the obtained structural restrictions in automatic isomer generation. An example for this approach of structure elucidation is presented in Figure 9, where ethyl 2-(2-hydroxyphenyl)acetate has been considered as unknown. The molecular formula and the mass spectrum are given. Application of software MSclass resulted in substructures that are assumed to be present (goodlist) and others that are assumed to be absent (badlist); only substructures relevant to the molecular formula are shown. Considering these structural restriction, six isomers are possible, including the correct solution. The isomer generator used was MOLGEN. The computation time on a Pentium 233 MHz is more than 20 hours for the generation of all isomers, but only 4 s when the structural restrictions of the goodlist and badlist are considered.

Conclusions or as a nonlinear function which is usually implemented as a neural network. During the training process the parameters of the classifier can be directly calculated (as for instance by multiple linear regression or partial least-squares regression) or they have to be adjusted iteratively (if a neural network is used). The aim of the training is to obtain values for z as defined by target values (for instance +1 for class 1 and −1 for class 2). The test set (which has not been used for the training) serves to estimate optimum classification thresholds z1 and z2 for

The best worldwide performance has been claimed for more than one commercial MS database system. However, more neutral observers state that automated spectra interpretation systems have a rather limited scope. Spectra library search systems are now widely used in MS laboratories and do a good job with routine problems. They are not so useful with complex problems or if the unknown is not contained in the library; however, current research promises considerable improvements of these methods in the future. The routine application of library

242 CHEMICAL STRUCTURE INFORMATION FROM MASS SPECTROMETRY

Figure 9 Systematic structure elucidation using the molecular formula of the unknown, structural restrictions from automatic mass spectra classification and exhaustive isomer generation. Ethyl 2-(2-hydroxyphenyl)acetate is the ‘unknown’.

search methods and spectra classification liberate the human spectroscopist from time-consuming, simple work or at least provides valuable preliminary suggestions for the expert.

List of symbols Am = abundance at mass m (weighted peak intensity); b = parameter of discriminant function; d = distance; Im = intensity (peak height) at mass m in %base peak; m = mass number; n = number of compounds or spectra; p = probability; q = number of variables (masses, spectral features); R = reference spectrum; S = similarity between two mass spectra; U = spectrum from unknown; x = spectral feature; z = discriminant variable.

See also: Chromatography-MS, Methods; Forensic Science, Applications of Mass Spectrometry; Fragmentation in Mass Spectrometry; Laboratory Information Management Systems (LIMS); Medical Applications of Mass Spectrometry; Pyrolysis Mass Spectrometry, Methods.

Further reading Adams MJ (1995) Chemometrics in Analytical Spectroscopy. Cambridge: The Royal Society of Chemistry. Clerc JT (1987) Automated spectra interpretation and library search systems. In Meuzelaar HLC and Isenhour TL (eds) Computer Enhanced Analytical Spectroscopy, Vol 1, pp 145162. New York: Plenum Press. Davis R and Frearson M (1987) Mass Spectrometry (Analytical Chemistry by Open Learning). Chichester: Wiley.

CHIROPTICAL SPECTROSCOPY, EMISSION THEORY 243

Drablos F (1987) Symmetric distance measures for mass spectra. Analytica Chimica Acta 201: 225239. Gasteiger J, Hanebeck W, Schulz KP, Bauerschmidt S and Höllering R (1993) Automatic analysis and simulation of mass spectra. In Wilkins CL (ed) Computer-Enhanced Analytical Spectroscopy, Vol 4 pp 97133. New York: Plenum Press. Gray NAB (1986) Computer-Assisted Structure Elucidation. NewYork: John Wiley. Massart DL, Vandeginste BGM, Buydens LMC, DeJong S, Lewi PJ and Smeyers-Verbeke J (1997) Handbook of Chemometrics and Qualimetrics: Part A. Amsterdam: Elsevier. McLafferty FW, Loh SY, and Stauffer DB (1990) Computer identification of mass spectra. In Meuzelaar HLC (ed) Computer-Enhanced Analytical Spectroscopy, Vol 2 pp 163181. New York: Plenum Press.

McLafferty FW and Turecek F (1990) Interpretation of Mass Spectra, 4th edn. Mill Valley: University Science Books. Owens KG (1992) Application of correlation analysis techniques to mass spectral data. Applied Spectroscopy Reviews 27: 149. Stein SE and Scott DR (1994) Optimization and testing of mass spectral library search algorithms for compound identification. Journal of the American Society of Mass Spectrometry 5: 859866. Varmuza K and Werther W (1996) Mass spectral classifiers for supporting systematic structure elucidation. Journal of Chemical Information and Computer Science 36: 323333.

Chemically Induced Dynamic Nuclear Polarization See CIDNP, Applications.

Chemometrics Applied to Near-IR Spectra See Computational Methods and Chemometrics in Near-IR Spectroscopy.

All interactions involving light with chiral molecules discriminate between the two possible circular polarizations (left = L and right = R). In the absence of a perturbing static electric or magnetic field, the light emitted by a molecular chromophore will be partially circularly polarized if the emitting species is chiral or optically active. In this context these two terms have identical meaning, and describe a molecular structure in which the mirror image isomers (enantiomers) are not superimposable. The experimental

ELECTRONIC SPECTROSCOPY Theory technique of analysing the extent of circular polarization in the light emitted from chiral molecules has been variously referred to as circularly polarized emission (CPE), emission circular intensity differentials (ECID), or circularly polarized luminescence (CPL). In this article we will use the acronym CPL to describe this spectroscopic technique. In addition, it will not be necessary in the discussion presented here, to use the more specific terms of circularly polarized fluorescence or circularly polarized

244 CHIROPTICAL SPECTROSCOPY, EMISSION THEORY

phosphorescence. Although CPL and circular dichroism (CD) spectroscopy, i.e. the difference in absorption of left versus right circular polarization, share many common characteristics, they differ from one another in two main ways. First, due to the Franck Condon principle, CPL probes the chiral geometry of the excited state in the same way that CD probes the ground state structure, and secondly, CPL measurements reflect molecular motions and energetics that take place between the excitation (absorption) and emission. Both of these effects have been exploited in CPL measurements, and will be a major focus of what is presented here.

Circularly polarized luminescence transition probabilities In CPL spectroscopy one is interested in measuring the difference in the emission intensity ('I) of left circularly polarized light (IL) versus right circularly polarized light (IR). By convention this difference is defined as follows

Just as in ordinary luminescence measurements, the determination of absolute emission intensities is quite difficult, so it is customary to report CPL measurements in terms of the ratio of the difference in intensity, divided by the average total luminescence intensity.

glum is referred to as the luminescence dissymmetry ratio (or factor). The extra factor of in Equation [2] is included to make the definition of glum consistent with the previous definition of the related quantity in CD, namely, gabs.

where in this equation HL and εR denote, respectively, the extinction coefficients for left and right circularly polarized light. The time dependence of the intensity of emitted light of polarization V, at a particular wavelength, O, for a specific transition n → g of a molecule with orientation : (in the laboratory frame) may be expressed

in terms of the transition probability, W as follows

where Nn (:,t) denotes the number of molecules in the emitting state | n 〉 with orientation : at time t, and fσ(O) is a normalized lineshape function. The number of molecules in the emitting state with orientation : depends on the orientation at the time of excitation (t = 0) with respect to the polarization S and direction of the exciting beam.

Kn denotes the fraction of molecules that end up in state | n 〉 that were initially prepared in the intermediate state | e 〉 by the excitation beam. This quantity is assumed to be independent of orientation. In the most general case we would also need to describe the time dependence of any conformational changes that take place between the time of excitation and emission, but for simplicity this will not be considered in the treatment presented here. The differential intensity of left minus right circularly polarized light may now be expressed as follows

where we have introduced the differential transition probability 'Wgn

The probability of emitting a right or left circularly polarized photon may be related in the usual way to molecular transition matrix elements through Fermis Golden Rule. Under the assumption that the emitted light is being detected in the laboratory 3 direction (see Figure 1), and allowing for electric dipole and magnetic dipoles in the expansion of the moleculeradiation interaction Hamiltonian we obtain the following expressions

CHIROPTICAL SPECTROSCOPY, EMISSION THEORY 245

Figure 1

Laboratory (1, 2, 3) and molecular (x, y, z) coordinate systems, and the Euler rotation matrix, (R).

where K(O3) is a proportionality constant, 1 and 2 refer to laboratory axes and the electric dipole transition moment Pgn and the imaginary magnetic dipole transition moment mgn are defined as follows.

(isotropic), the emitting state may not be, due to the photoselection of the emitting sample by the excitation beam. We will denote an orientational distribution by brackets 〈…〉. Substitution of Equations [5] and [11] into Equation [6], and allowing for an ensemble of orientations we obtain the following

The differential transition rate is, therefore,

The final connection between molecular properties and experimental observables requires knowledge of the orientational distribution of the emitting molecules with respect to the direction and polarization of the excitation light and the direction of detection, and the time dependence of this distribution.

Circularly polarized luminescence from solutions There have been very few attempts at measuring CPL from oriented samples due to the inherent problems associated with measurement of circular polarization in the presence of linear polarization. For this reason we restrict the discussion of the orientation dependence of CPL to randomly oriented molecular samples such as occur in liquid solutions. Of course, even if the ground state distribution is randomly oriented

We may formally separate the time dependence of the orientational distribution, K(:,t), from the number of molecules excited at time t = 0 by an excitation pulse as follows

N (: 0) may be calculated from knowledge of the initial distribution, and direction and polarization of the excitation beam.

where Peg is the absorption transition vector. Final expressions for the formal relationship between the time-dependent circularly polarized luminescence intensity or glum require knowledge of experimental geometry, and K(:,t).

246 CHIROPTICAL SPECTROSCOPY, EMISSION THEORY

For purposes of illustration, we examine here the most common situation of random (isotropic) ground state distributions, and 90° excitation/emission geometry with unpolarized excitation. If the emission is detected in the laboratory 3 direction, then the excitation is polarized in the 13 plane. Equation [13] may then be written as

development of the formal expression for I(t) parallels that presented above. The final result for glum(t) is obtained by evaluating all of the integrals implied in Equation [16], and the related expression for I(t). In this particular simple model, the time-dependence is only in the decay of the emitting state, and is identical for 'I and I, so that glum is time independent. This resulting general expression for the frozen limit is

and the final formal expression (in the laboratory coordinate system) is obtained by substituting this result into Equation [12].

where we have explicitly labelled the square brackets to indicate that the absorption takes place at time 0 and the emission at time t. The ensemble average over molecular orientations implied in Equation [16] may be performed if K(:, t) is known. We first examine the case in which the emitting sample is frozen, i.e. the orientational distribution of emitting molecules is identical to that prepared by the excitation beam. In this case K(:, t) = exp(−t/τ), where W denotes the emission lifetime, and, since the orientations are fixed, the subscripts on the square brackets may be eliminated. The orientational average may be performed by relating the transition moments in the molecular coordinate system to the laboratory system using the appropriate elements of the Euler matrix (see Figure 1). If one assumes, for example, that the absorption transition moment defines the molecular z-axis, then one must evaluate terms such as the following

This photoselection of emitting distributions also affects the measurement of total luminescence, and the

where we have allowed for different lineshapes for CPL and TL, and an arbitrary absorption transition direction. This result simplifies under conditions in which the absorption and emission transition directions are parallel. Furthermore, the reader is referred to the Further reading section concerning the results expected for other excitation/emission geometries. It should be noted that if the luminescence is partially linearly polarized, the measurement of CPL is problematic due to experimental artifacts. For this reason it is usually the case that the excitation/emission geometry and incident polarization are chosen so as to eliminate the possibility of linear polarization in the luminescence. The other useful limiting case is the situation in which the orientation of emitting molecules is isotropic (random). This limiting case is appropriate for small molecular systems with relatively long emission lifetimes. If the distribution prepared by the excitation beam has been completely randomized by the time of emission, then Equation [16] reduces to the following

where we have assimilated the absorption strength and other parameters associated with the excitation into the constants. The orientational averaging in this equation involves the product of only two elements of the Euler matrix. Just as in the previous discussion, the time dependence of the total luminescence and the CPL for this model are identical. The final formal

CHIROPTICAL SPECTROSCOPY, EMISSION THEORY 247

relationship for this isotropic limit is

The major difference between this result and the frozen result is the fact that, in Equation [18], the different components of the emission transition vectors contribute unequally to the observable. In principle, this would allow one to investigate the chirality of molecular transitions along specific molecular directions, but to date this has not been exploited. Equation [20] illustrates one of the operational principles of CPL (and CD) spectroscopy. In most situations one is interested in studying transitions that are formally forbidden. If the transition is allowed, the dipole strength, which is proportional to the denominator of Equation [20], is large. Since the rotatory strength represented by the numerator of Equation [20] is generally quite small, because of the magnitude of the magnetic dipole transition moment, glum (or gabs) for allowed transitions is usually very small. In CPL spectroscopy this leads to the practical result that one does not generally study highly luminescent dyes, or other strongly allowed transitions, but rather, the most interesting molecular structural information comes from the study of weakly or mildly luminescent transitions in which the inherent molecular chirality is closely connected with the emissive chromophore.

The number of R enantiomers that absorb left circularly polarized light, NL(R) is proportional to HL(R), and similarly for HR(R). Thus, we may make the following substitution

since all of the proportionality constants cancel. The number of R enantiomers that would absorb right circularly polarized light is exactly equal to the number of S enantiomers that would absorb left circularly polarized light, i.e. NR(R) = NL(S). This substitution yields the following

'NL is the differential population (R − S) of excited enantiomers that would result from left circularly polarized excitation. In general, an exact description of CPL from racemic mixtures requires consideration of the competition between racemization and emission. Under the assumption that racemization is much slower than emission, we may write the following expression for CPL from a racemic mixture under left circularly polarized excitation as follows

Circularly polarized luminescence from racemic mixtures One of the most useful applications of CPL spectroscopy has been in the study of racemic mixtures. This experiment is possible, because, even though the ground state is racemic, the emitting state can sometimes be photoprepared in an enantiomerically enriched state by use of circularly polarized excitation. If the racemization rate of the excited state is less than the emission rate, then the emission will be circularly polarized. We consider a racemic sample containing equal concentrations of R and S enantiomers. The preferential absorption of circularly polarized light is related to the CD through the absorption dissymmetry ratio gabs which was defined in Equation [3]. Rewriting this equation for the CD of the R enantiomer we obtain the following

Substituting from above we obtain the following expression

where we have explicitly labelled the excitation wavelength as O′. Examination of Equation [25] shows that the measurement of CPL from racemic mixtures depends on the product of the CD and CPL. Since the measurement of glum is generally limited to magnitudes greater than 10−4, the magnitude of the intrinsic dissymmetry ratios must usually be greater than 10−2 for this unique measurement to be feasible. This restriction has limited applications of this technique to racemic lanthanide complexes, since intraconfigurational f ↔ f transitions that obey magnetic dipole

248 CHIROPTICAL SPECTROSCOPY, EMISSION THEORY

transition rules (i.e. ∆J = 0, ±1) often are associated with dissymmetry ratios greater than 0.1.

Time-resolved circularly polarized luminescence For the simple model enantiopure systems described above, it was concluded that the time dependence of the CPL and total luminescence were identical, and, therefore, the dissymmetry ratio contained no dynamic molecular information. This, of course, would not be the case if intramolecular geometry changes, that would effect the chirality of the molecular transitions, were occurring on the same time scale as emission. However, no such examples of this type of study have yet appeared. Time-resolved CPL measurements have been useful in the study of racemic mixtures of lanthanide complexes in which racemization or excited state quenching is occurring on the same time scale as emission. It was assumed in the previous section, that the enantioenriched excited state population, which was prepared from a racemic ground state distribution by a circularly polarized excitation beam, was maintained until the time of emission. If this is not the case, then the differential excited state population is time-dependent. Neglecting orientational effects, we may derive the following expression for the time-dependence of the excited state population following an excitation pulse of left circularly polarized light

It is also possible to extract dynamic information concerning racemization rates through measurement of steady-state CPL. Again neglecting time-dependent orientational effects, one may integrate Equations [26] and [27] over long times (0 → ∞) and obtain the following result

This equation has the expected limiting behaviour. If krac > k0 then the magnitude of glum becomes negligibly small. The second application of time-resolved CPL is in the study of enantioselective (or more properly, diastereomer-selective) quenching. In these experiments an enantiopure quencher molecule is added to a racemic solution. The interaction of the chiral quencher with the individual enantiomers is such that the excited state of one of the enantiomers is quenched more rapidly than the other. This process is depicted schematically in Figure 2, where we have assumed the quencher is an R enantiomer, and kRS and kRR denote the quenching rate constants. This process may be described using the following simple kinetic model

where krac is the racemization rate and k0 is the excited state decay rate. Since the total number of emitting molecules decays as

one can express the time-dependence of the luminescence dissymmetry as follows

This technique has been used in a number of studies in which the decay of glum has been analysed to extract the racemization rate constant.

Figure 2 Schematic energy level diagram for measurement of CPL from a racemic mixture.

CHIROPTICAL SPECTROSCOPY, EMISSION THEORY 249

These reactions are modelled in terms of a diffusional step (k or k ), resulting in the formation of a bimolecular encounter complex. This is followed by competing pathways for dissociation of the complex (k or k ) or energy transfer (k or k ). Deactivation of the excited quencher is described by kP. The observed CPL is directly proportional to the difference in excited state concentrations of the two enantiomers, and this may be related to the specific rate constants introduced above. If, for example, we make the reasonable assumption that the diffusion and deactivation are independent of chirality, one can derive the following expression for 'N(t)

where

experimental observables. These are, of course, of primary importance in interpreting CPL measurements in terms of inter- and intramolecular processes, including molecular reorientation, racemization and chemical reactions which affect the net molecular chirality. These processes are all well understood, and specific information is most often obtained within an appropriate model system. The more fundamental aspects of molecular chirality associated with the molecular transition matrix elements is less well developed. Although there have been a few successes in interpreting the sign and magnitude of a CPL spectrum in terms of the identity and structure of a specific enantiomer, in this regard, it is certainly not the case that the technique is as well developed as CD spectroscopy. Even in the case of fairly highly symmetric chiral lanthanide complexes, the level of computation and development of f-f intensity theory is not yet at the stage where one can associate a priori the sign of an individual transition with the identity of the enantiomer.

List of symbols and

If the quenching reactions are far from diffusion control (i.e. kET 2). An inherently dissymmetric chromophore possesses no symmetry element of the second kind and thus is chiral by itself. In an inherently symmetric chromophore a symmetry element of the second kind exists and thus it is achiral by itself. In Figure 4 the enone chromophore is shown as inherently symmetric and inherently dissymmetric. The local symmetry of a

CHIROPTICAL SPECTROSCOPY, GENERAL THEORY 255

Figure 3 Term scheme for an absorption and the corresponding CD band with contributions of one allowed and one forbidden vibrational progression. 00 is the 0–0 transition band. 1 and x are a totally symmetric and a non-totally symmetric vibration, respectively. n = 1,2, ....

chromophore cannot be sharply defined because the chromophore itself is only a qualitative quantity. For CD bands a characterization by the size of the dissymmetry factor g is possible. For an inherently dissymmetric chromophore the dissymmetry factor g is about or larger than 102, for an inherently symmetric chromophore g < 102 (mostly ≈ 104) if the transition is electrically allowed and magnetically forbidden. g > 102 (mostly 5 × 103 to 101) if the transition is magnetically allowed and electrically forbidden, also for an inherently symmetric chromophore. What is a chromophore?

A chromophore is that part of the molecule where the absorption proceeds and where the main change of the geometry or electron density, etc. appears after

the excitation process. The area of the chromophore is not very well demarcated from the residual parts of the molecule which are not involved in the absorption process. The shading in Figure 5 demonstrates the continuous variation from the centre of the excitation to the undisturbed residue of the molecule. A carbonyl chromophore with an nS*-transition and an olefin chromophore with a SS*-transition are regarded as two distinct chromophores (Figure 5). If the two chromophores are located in a molecule next to each other, an influence on the position of the band and on the intensity of the absorption results. If the interaction is strong enough, the two chromophores have to be considered as one new chromophore. The number of chromophores in a molecule is very large because of the large number of transitions

256 CHIROPTICAL SPECTROSCOPY, GENERAL THEORY

Figure 4

The inherently symmetric (left) and inherently dissymmetric (right) enone chromophore.

Figure 5

The enone chromophore as an example for the interaction of two chromophores, i.e. the ‘en’ and ‘one’ chromophore.

localized in different parts of the molecule. Also in a group, like the carbonyl group, there is an infinite number of chromophores. One important chromophore for CD spectroscopy besides the nS*-transition is a SS*-transition, located in the carbonyl area and at wavelength lower than 150 nm. According to Snatzke a molecule can be divided into several spheres (Figure 6). The first sphere is the chromophore itself which can be either chiral or achiral. The second sphere is the neighbourhood to the chromophore, the third sphere the area following the second sphere, and so on. The criteria by which the spheres are differentiated are the distance and the interaction between the parts of the molecules like conjugation etc. Usually changes in the second sphere can affect the sign and size of the CE whereas substitutions in spheres farther away lead to smaller effects.

The quantitative description of the rotational strength Methods for the description of the optical activity are either based on an estimation of the sum over the rotational strengths of all transitions weighted by a dispersion term as in the case of frequency-dependent optical rotation (ORD; denoted as I or [M]) or of the rotational strength of one transition in the case of circular dichroism, denoted as 'H or TEqn [32]). Numerical quantum-mechanical calculations of the rotational strength RNK for a few transitions | N〉 → | K〉 of a molecule lead nowadays to results of

Figure 6 Two examples for a decomposition of a molecule into spheres according to Snatzke.

increasing reliability and thus a determination of the absolute configuration becomes possible with a CD measurement over the spectral region to which the analysed transitions belong. But this is only a development in the right direction, and further advances in computational abilities and resources are necessary. The possibility for numerical calculation is not always available in the chemists daily work today. From this point of view there is no way to renounce the qualitative concepts and semi-qualitative methods used up to now to correlate sign and absolute configuration. Therefore, the following techniques will be discussed in more detail: 1. the quantum-mechanical calculation of the rotational strength, 2. the polarizability theory, 3. the one-electron mechanism,

CHIROPTICAL SPECTROSCOPY, GENERAL THEORY 257

4. the model of independent groups (MIG) of Tinoco, 5. the exciton model (exciton chirality method). Quantum mechanical calculation of the rotational strength

In order to calculate the rotational strength, reasonably good wavefunctions for all electronic states should be available. Calculation with a number of semiempirical and ab initio techniques have been performed with more or less success. A review of used techniques has been given by Woody in Nakanishi et al. As discussed there, for the calculation of the rotational strength, the electric transition moments can be either evaluated with the dipole-velocity method ¢N | pi | K² with the linear momentum operator pi which leads to an originindependent rotational strength RNK but overestimates errors from inaccurate wavefunctions far away from the atomic centre, or with the dipolelength method ¢N | Pi | K² which leads to an origindependent rotational strength RNK and is not so sensitive to other errors of the wavefunctions. Up to now the dipole-velocity method has been the method of choice for numerical quantum-mechanical calculations. More recent analyses and calculations by Grimme do not support this choice, in contrast to the leading opinion. It seems to us that one may choose the dipole-velocity method or the dipolelength method depending on the kind of molecule and the quality of the calculated wavefunctions. Furthermore, the extent of taking into account singly and doubly excited configurations is a problem when trying to achieve good results for the direction of the electric and magnetic dipole transition moments and thus for the rotational strength. A new development can also be found in the use of the density functional theory for the calculation of RNK by Grimme. The polarizability theory

Kirkwood expressed the rotatory power in terms of the polarizabilities and anisotropies of the polarizabilities of groups in the molecule. First-order perturbation theory leads in this case to a perturbation potential expressing the interaction of two transition moment dipoles i, j at a distance Rij, where this distance is large compared to the separation of charge in the dipoles. Such terms are important if the molecule contains two or more strong absorption bands near the frequency region where the rotatory dispersion is calculated originating from allowed electric dipole transitions. But even the strongest absorption bands may contribute little to the optical activity by this mechanism if the chromophoric groups are

unfavourably disposed relative to each other. In the last few years there has been an improvement in the calculation of the optical rotation with Kirkwoods polarizability theory by using chirality functions as a possibility to find symmetry-adapted functions as shown by Haase and Ruch for asymmetric methane or allene derivatives. The thus obtained electric dipole transition moments are grouped together to polarizabilities or to polarizability tensor coordinates, respectively. The basis for the description is a second-order perturbation theory of the optical rotation. The one-electron mechanism

Historically, the one-electron mechanism was introduced by Condon and co-workers as an optical activity which originates from the excitation of a single electron in a field of suitable dissymmetry. In the MO concept, the one-electron mechanism is an excitation of one electron to a singly-excited electronic configuration of an inherently symmetric chromophore perturbed by chiral surroundings. Depending on the symmetry of the chromophore the transition is either electrically and/or magnetically dipole allowed and without the perturbation fulfills the condition that the dot product of the electric and magnetic dipole transition moments of this transition is zero. By the perturbation with groups/atoms in chiral surroundings a magnetically and/or electrically allowed dipole is induced by borrowing intensity from other transitions of the same chromophore. Often the language intensity is stolen from other transitions has been used in the literature, e.g. contributions to the rotational strength of the n → S* transitions are stolen by this mechanism from electric-dipole-allowed V → V* or S → S* transitions. There are two possibilities for a perturbation: On the one hand electronic interactions are responsible for the intensity borrowing which is often called the dynamic coupling model. On the other hand vibrational processes change the geometry and, by taking into account the dependence of 〈 〉0n and/or 〈 〉n0 on nuclear coordinates, intensity is borrowed by the transition N → K via nuclear vibrations from other allowed transitions of the chromophore. Then different vibrational progressions contribute to the UV and CD spectra as shown in Figure 3. This mechanism is often called the static coupling or vibronic coupling model. Depending on the type of transitions, a nomenclature system has been introduced by Moscowitz and co-workers and Weigang where the CD of a transition (a) without intensity borrowing is called a case I CD, (b) with intensity borrowing via one forbidden electronic dipole transition is called a case II CD, and (c) with intensity borrowing via

258 CHIROPTICAL SPECTROSCOPY, GENERAL THEORY

electrically and magnetically forbidden transitions is called a case III CD. Whereas in the first case only progressions with totally symmetric vibrations contribute to the CD, in cases II and III progressions with non-totally symmetric vibrations and vibrational CD bands of the same or different sign can also contribute to an electronic CD band as shown by Weigang and in Snatzke by Moscowitz. The model of independent groups for the calculation of the rotational strength of large molecules or polymers

Tinocos aim with the model of independent groups (MIG method) was to calculate the total rotational strength RNK for a transition |NÒ → |KÒ for a large molecule or a polymer possessing a large number of equal and different groups, i.e. chromophores for which the circular dichroism is given by

The excited state may or may not be degenerate or accidentally degenerate. In the case of an n-fold degeneracy of the state |AÒ, the excited states |LÒ, linear combinations of the degenerate states |AÒ, contribute to the rotational strength (Eqn [39]):

For the evaluation of the rotational strength the molecule is decomposed into different groups i which are at a distance ⎜ ij ⎜ = ⎜ i j ⎜ and oriented differently with respect to each other. The operator of the interaction between two groups i, j is given by the potential ij. Depending on the assumption for the interaction between the groups, different approximations for the wavefunctions of the groups are necessary. In the simplest case of completely isolated groups the wave function of the molecule can then be expressed as a product of the wavefunctions of the groups. It is assumed that the group electronic wavefunctions do not overlap, which means that there is no electron exchange between groups. Only the interaction between two groups at a time is considered. Furthermore, it is assumed that singly and doubly excited states on different groups are possible. To calculate the total rotational strength for a transition it is necessary to describe the electric and magnetic dipole transition moment operator in an adequate way. Whereas the electric moment operator is the sum of the moments of the different groups, for the magnetic moment operator two contributions have to be taken into account. Analogously to the electric moments, there is a contribution from the magnetic moment of every group. The second contribution comes from the interaction of groups. In a simplified picture a moving electron in one group leads to a magnetic moment in the other group and gives rise to a large contribution to the rotational strength, which resembles the exciton coupling and the coupled oscillator model as discussed in a later section. The rotational strength of a non-degenerate system with the assumptions given above is obtained as follows:

CHIROPTICAL SPECTROSCOPY, GENERAL THEORY 259

The coupling Visr, jtv between the groups i, j with their singly- or doubly-excited configurations r, s and t, v is given by the matrix elements

The potential energy operator ij of Equation [41] stands for the electrostatic interaction between the groups i and j. Mir, Mis, etc. are group wavefunctions, where the first index represents the group and the second index the electronic state of the group r, s, ... in i and t, v, ... in j. 〈 〉i0a and 〈 〉ia0 are the electric and the magnetic dipole transition moments of an excitation of a group i for a transition from the ground state | 0〉 to the excited state | a〉 (| 0〉 → | a〉). a0 and b0 are the energies given in wavenumbers of the transitions | 0〉 → | a〉 and | 0〉 → | b〉, respectively. The first term of Equation [40] contributes to the optical activity of a molecule if the transition | N〉 → | K〉 is electrically and magnetically allowed and the chromophore is inherently dissymmetric. This term represents the optical activity of isolated groups. In the second term the electric and magnetic dipole transition moments on different groups contribute by coupling the groups in their excited states by electrostatic interaction. The interaction of electric and magnetic dipole transition moments of different transitions of one isolated group are described with the third and the fourth term of Equation [40]. The fifth term shows the dependence on the difference of the electric dipole moment of the ground and excited state. This term is as the fourth term very small because of the division by a0 instead of a difference like . The coupling of the electric dipole transition moments of two different groups in the sixth term represents the exciton model of two groups if their excited states are non-degenerate. This is the term which is the origin for the exciton chirality method of Harada and Nakanishi if the two excited states of the coupled electric transition moments are obtained by absorption in different spectral regions. This term corresponds to the theory of coupled oscillators. The interaction between degenerate groups does not contribute to the rotational strength of a polymer here. This effect is discussed in the last section for a system of two chromophores.

decomposed into an achiral skeleton and achiral groups, so-called ligands, chirality functions have been developed which allow a quantitative description of chiral phenomena as a function of phenomenological parameters of the ligand. Additionally these functions allow the correlation of the sign of the chirality phenomenon and the absolute configuration of the molecule. A systematic analysis of such chirality functions, taking advantage of the symmetry properties of the skeleton, has been given by Ruch and Schönhofer. The ligands, the number of which is determined by the chosen binding sites and the symmetry of the skeleton, are at positions where symmetry operations applied to the skeleton permute the ligands on the different possible binding sites and in this way create new distinguishable molecules. The binding site of a ligand is correlated with an argument for the ligands in the function, i.e. the arguments O(l1), O(l2), etc. correspond to the positions 1, 2, etc. in Figure 7. In the approximation of the so-called qualitative completeness, the chirality function F(l1,l2,l3,l4) for allene derivatives with a skeleton of D2d symmetry and with four binding sites for the ligands which depend on two phenomenological parameters O(li), P(li) for each ligand li can be given by

K1 and K2 are two additional phenomenological parameters. For the optical rotation the phenomenological parameters for a large number of ligands have been determined. Ligand parameters of one and the same ligand are different if they are bound to achiral skeletons of different symmetry and different sites, i.e. a determined parameter for one special ligand can only be taken for compounds with identical skeletons. There are further restrictions for the

Rules for the determination of the absolute configuration Chirality functions

For molecules with an inherently symmetric chromophore or molecules which can be formally

Figure 7 Achiral skeleton with D2d symmetry and the four substitution positions 1 to 4 for an allene.

260 CHIROPTICAL SPECTROSCOPY, GENERAL THEORY

description of the optical rotation by chirality functions because of some experimental and theoretical reasons. One restriction is the assumption that the ligands at a binding site of the skeleton have to be invaria

Encyclopedia of Spectroscopy and Spectrometry (3-Volume Set with Online Version) - PDF Free Download (2025)