Infrared and Raman Spectroscopy in Forensic Science

John Chalmers would like to, yet again, apologise to his wife Shelley for her enduring the role of being a book editor's partner, despite having promised previously not to take such a task on again; maybe this will be the last!

Howell Edwards dedicates this book to his wife Gillian and daughter Katharine who have supported him throughout and to his research supervisor, Dr Leonard Woodward at the University of Oxford, who first stimulated what proved to be his lifelong interest in Raman spectroscopy.

Mike Hargreaves would like to thank his partner Jen, family and his fellow editors for their patience and understanding and for sometimes failing to juggle everything.

John M. Chalmers

Howell G.M. Edwards

Michael D. Hargreaves

July 2011

About the Editors

John M. Chalmers CChem FRSC

John Chalmers early retired at the end of 1999 from the United Kingdom chemical company ICI plc; John spent 34 years working with vibrational spectroscopy techniques while employed within research departments of ICI; he retired as a Business Research Associate in the Molecular Spectroscopy Team, Science Support Group, ICI Technology, Wilton Research Centre, UK. In 1994 John was the recipient of the Williams–Wright Award presented by The Coblentz Society for outstanding contributions in the Field of Industrial Infrared Spectroscopy. In 2000, John became a self-employed consultant (VS Consulting) specialising in vibrational spectroscopy; he also took up part-time positions as a Senior Research Fellow and then a Special Lecturer for a period of about 10 years within the School of Chemistry at the University of Nottingham. John has also been a visiting lecturer to the School of Chemical Sciences, University of East Anglia, from 1992–2000. Among his spectroscopic society activities, John is a Past President (2008) of the Society for Applied Spectroscopy (SAS), having served SAS previously as both a Governing Board Member and an International Delegate; John was Chair of the UK Infrared and Raman Discussion Group (IRDG) for nine years (1995–2003).

John has published over 50 peer-reviewed technical papers in scientific journals; he has also had published over 20 book chapters or reference articles. He has co-authored one book (Industrial Analysis with Vibrational Spectroscopy, with Geoffrey Dent, 1997, Royal Society of Chemistry, Cambridge); John has also edited or co-edited several books, including the highly acclaimed reference work the five-volume Handbook of Vibrational Spectroscopy (co-edited with Professor Peter Griffiths, 2002, published by John Wiley & Sons, Ltd, Chichester). Edited or co-edited book titles published include: Spectroscopy in Process Analysis (2000, Sheffield Academic Press, Sheffield), Raman Spectroscopy in Archaeology and Art History (with H.G.M. Edwards, 2005, Royal Society of Chemistry, Cambridge), Molecular Characterization and Analysis of Polymers (with Robert J. Meier, 2008, published by Elsevier, Amsterdam); and the books with titles published by John Wiley & Sons, Ltd., Chichester are: Vibrational Spectroscopy of Polymers: Principles and Practice (with Neil J. Everall and Peter R. Griffiths, 2007), Applications of Vibrational Spectroscopy in Pharmaceutical Research and Development (with Don E. Pivonka and Peter R. Griffiths, 2007), Vibrational Spectroscopy for Medical Diagnosis (with Max Diem and Peter R. Griffiths, 2008), Applications of Vibrational Spectroscopy in Food Science (two-volume set, with Eunice C.Y. Li-Chan and Peter R. Griffiths, 2010). John is also currently the Article Editor for Spectroscopy Europe.

Professor H.G.M. Edwards M.A., B.Sc., D.Phil., C.Chem. FRSC, Emeritus Professor of Molecular Spectroscopy

Howell Edwards studied Chemistry at Jesus College, University of Oxford, and carried out research for his DPhil at Oxford on chemical applications of Raman spectroscopy under the supervision of Dr. Leonard Woodward. Following a Research Fellowship at Jesus College in the University of Cambridge he took a lectureship in Structural and Inorganic Chemistry at the University of Bradford where he became Reader and then Professor of Molecular Spectroscopy and Head of the Chemical and Forensic Sciences Division. In 2003, he received the Sir Harold Thompson Award from Elsevier Science for his international contributions to vibrational spectroscopy. He is the recipient of the Emanuel Boricky Medal for 2008/2009 from Charles University, Prague, for distinguished international contributions to geochemistry and mineralogical analysis. In his research career he has published over 1000 papers on Raman spectroscopy and its applications and is the co-editor of three books: A Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line (with I.R. Lewis, 2001, Marcel Dekker, New York), Selected Topics in Raman Spectroscopic Applications: Geology, Biomaterials and Art (with F. Rull Perez, P. Vandenabeele and D.C. Smith, 2007, Publidisa Valladolid), and Raman Spectroscopy in Archaeology and Art History (with J.M. Chalmers, 2005, RSC Publishing, Cambridge). He is the recipient of the 2011 Charles Mann Award of the international Federation of Analytical Spectroscopic Societies (FACSS) for distinguished work in applications of Raman spectroscopy. Professor Edwards is a member of the Editorial Boards of J. Raman Spectroscopy, J. Molecular Structure, Spectrochimica Acta, Vibrational Spectroscopy, Drug Targeting and Analysis and Asian J. Spectroscopy. He is Associate Editor of the International Journal of Astrobiology.

Professor Edwards has wide-ranging interests in the applications of Raman spectroscopy to the characterisation of materials in forensic, art historical, polymer, pharmaceutical and archaeological contexts, the characterisation of contraband biomaterials (ivories and drugs of abuse), and spectroscopic molecular signatures relating to the biological survival of cyanobacteria in putative Martian terrestrial analogues. He is international lead coordinator of the Science Team on the RLS Raman instrument with the NASA/European Space Agency on the ExoMars project for the construction and terrestrial evaluation of a miniature Raman spectrometer adopted for a planetary robotic lander for surface and subsurface exploration and search for life on Mars. The Raman spectroscopic characterisation of contraband biomaterials, including the evaluation of portable Raman spectrometers for the field acquisition of data on ivories and drugs of abuse of forensic relevance, has been carried out with support from the Engineering and Physical Sciences Research Council and sponsored by security and law enforcement agencies.

Michael D. Hargreaves MChem, PhD, CSci, CChem, MRSC

Michael Hargreaves studied chemistry at the University of Newcastle upon Tyne, United Kingdom, and carried out a PhD under the supervision of Professor Mike George and Associate Professor Barrie Kellam at the University of Nottingham, UK, on reaction monitoring using FT-IR and Raman spectroscopy.

He undertook two postdoctoral positions with Professor Howell Edwards, at Bradford University, UK, the first on portable Raman spectroscopy for identification/screening of biomaterials, drugs of abuse and explosives, the second on evaluation and development of the RLS Raman detector and geological library with European Space Agency/NASA on the ExoMars project.

After this Mike joined industry, working for Cobalt Light Systems, with Professor Pavel Matousek, commercialising SORS and transmission Raman spectroscopy. He left to join Ahura Scientific in the Application Development Group; Ahura Scientific was subsequently acquired by Thermo Fisher Scientific, where Mike remains within the Portable Analytical Instruments Group.

Michael Hargreaves has authored or co-authored over 30 publications, covering the application of vibrational spectroscopy to the fields of drugs of abuse, explosives, pharmaceutics, geology, biomaterials and works of art.

John M. Chalmers

Howell G.M. Edwards

Michael D. Hargreaves

July 2010

List of Contributors

W. James Armstrong, Forensic Science Northern Ireland, Carrickfergus, UK

Edward G. Bartick, Retired: FBI Laboratory – Counterterrorism and Forensic Science Research Unit, Current: Director of the Forensic Science Program, Department of Chemistry and Biochemistry, Suffolk University, Boston, USA

Steven E.J. Bell, School of Chemistry and Chemical Engineering, Queen's University, Belfast, UK

Victoria L. Brewster, Laboratory for Bioanalytical Spectroscopy, School of Chemistry, Manchester Interdisciplinary Biocentre University of Manchester, Manchester, UK

Christopher D. Brown, Thermo Fisher Scientific Portable Optical Analyzers, Thermo Fisher Scientific, Wilmington, USA

Kevin Buckley, Central Laser Facility, Science and Technology Facilities Council, Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, Didcot OX11 0QX, UK; and UCL Institute of Orthopaedics and Musculoskeletal Science, Stanmore Campus, Royal National Orthopaedic Hospital, Stanmore, UK

Lucia Burgio, Science Section, Conservation Department, Victoria and Albert Museum, London, UK

Andrew D. Burnett, School of Electronic and Electrical Engineering, University of Leeds, Leeds, UK

Mary W. Carrabba, Department of Chemistry, Southern Oregon University, 1250 Siskiyou Boulevard, Ashland, USA

John M. Chalmers, VS Consulting, Stokesley, UK

Philippe Colomban, Laboratoire de Dynamique, Interactions et Réactivité – UMR7075, CNRS, Université Pierre-et-Marie-Curie, 4, Place Jussieu, 75005 Paris, France

John E. Cunningham, School of Electronic and Electrical Engineering, University of Leeds, Leeds, UK

A. Giles Davies, School of Electronic and Electrical Engineering, University of Leeds, Leeds, UK

Paul Dean, School of Electronic and Electrical Engineering, University of Leeds, Leeds, UK

A. Deneckere, Ghent University, Department of Analytical Chemistry, Krijgslaan, Ghent, Belgium

Howell G.M. Edwards, Chemical and Forensic Sciences, School of Life Sciences, University of Bradford, Bradford, UK

Marina Epelboym, European Gem Lab- EGL USA, 580 Fifth Avenue, New York, USA

Karen Faulds, Department of Pure and Applied Chemistry, University of Strathclyde, Glasgow, UK

Peter M. Fredericks, Queensland University of Technology, Brisbane, Australia

Craig Gardner, Thermo Fisher Scientific Portable Optical Analyzers, Thermo Fisher Scientific, Wilmington, USA

Royston Goodacre, Laboratory for Bioanalytical Spectroscopy, School of Chemistry, Manchester Interdisciplinary Biocentre University of Manchester, 131 Princess Street, Manchester, UK

Robert L. Green, Thermo Fisher Scientific Portable Optical Analyzers, Thermo Fisher Scientific, Wilmington, USA

Peter R. Griffiths, University of Idaho, Department of Chemistry, Renfrew Hall, Moscow, USA

A. Guedes, Centro de Geologia e Departamento de Geociências, Ambiente e Ordenamento do Território da Faculdade de Ciências, Universidade do Porto, Porto, Portugal

Michael D. Hargreaves, Thermo Scientific Portable Optical Analyzers, Thermo Fisher Scientific, Wilmington, USA

Wayne Jalenak, Thermo Fisher Scientific Portable Optical Analyzers, Thermo Fisher Scientific, Wilmington, USA

Jan Jehli ka, Institute of Geochemistry, Mineralogy and Mineral Resources, Faculty of Science, Charles University in Prague, Prague, Czech Republic

Kathryn S. Kalasinsky, Armed Forces Institute of Pathology, Washington D.C., USA

Lore Kiefert, Guebelin Gem Laboratory, Maihofstrasse, Luzern, Switzerland

Kaho Kwok, Department of Industrial and Physical Pharmacy, Purdue University, West Lafayette, USA

Ian R. Lewis, Kaiser Optical Systems, Inc., Ann Arbor, USA

Mary L. Lewis, I. R. Lewis, Kaiser Optical Systems, Inc., Ann Arbor, USA

Edmund H. Linfield, School of Electronic and Electrical Engineering, University of Leeds, Leeds, UK

Juan Manuel Madariaga, Department of Analytical Chemistry, University of the Basque Country, Bilbao, Spain

Pavel Matousek, Central Laser Facility, Science and Technology Facilities Council, Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, Didcot, UK

L. Moens, Ghent University, Department of Analytical Chemistry, Ghent, Belgium

Ute Münchberg, Institute of Physical Chemistry, Friedrich-Schiller-University Jena, Helmholtzweg Jena, Germany

Hpone-Phyo Kan-Nyunt, GIA Laboratory Bangkok, U-Chu-Liang Building, Bangkok, Thailand

Andrew J. O'Neil, School of Pharmacy and Chemistry, Kingston University, Kingston Upon Thames, UK

Vincent Otieno-Alego, Forensic and Data Centres, Australian Federal Police, Australia

Banu Özen, Department of Food Engineering, zmir Institute of Technology, Urla, zmir, Turkey

Susan Paralusz, Consulting Gemologist, North Brunswick, New Jersey, USA

Jürgen Popp, Institute of Physical Chemistry, Friedrich-Schiller-University Jena, Helmholtzweg Jena, Germany; and Institute of Photonic Technology e. V. (IPHT), Jena, Germany

A.C. Prieto, Departamento de Física de la Materia Condensada, Cristalografía y Mineralogía, Universidad de Valladolid, Spain

A. Reip, Wolfson Centre for Materials Processing, Brunel University, Kingston Lane, Uxbridge, UK

Paola Ricciardi, National Gallery of Art, 2000B South Club Drive, Landover, USA

Petra Rösch, Institute of Physical Chemistry, Friedrich-Schiller-University Jena, Jena, Germany

J. Silver, Wolfson Centre for Materials Processing, Brunel University, Kingston Lane, Uxbridge, UK

W. Ewen Smith, Department of Pure and Applied Chemistry, University of Strathclyde, Glasgow, UK; and Renishaw Diagnostics Ltd, Nova Technology Park, Glasgow, UK

Naomi Speers, Forensic and Data Centres, Australian Federal Police, Australia

S. James Speers, Forensic Science Northern Ireland, Carrickfergus, UK

Samantha P. Stewart, School of Chemistry and Chemical Engineering, Queen's University, Belfast, UK

Stephan Stöckel, Institute of Physical Chemistry, Friedrich-Schiller-University Jena, Jena, Germany

Lynne S. Taylor, Department of Industrial and Physical Pharmacy, Purdue University, West Lafayette, USA

Figen Tokatli, Department of Food Engineering, zmir Institute of Technology, Urla, zmir, Turkey

P. Vandenabeele, Ghent University, Department of Archaeology, Ghent, Belgium

R. Withnall, Wolfson Centre for Materials Processing, Brunel University, Kingston Lane, Uxbridge, UK

Mark R. Witkowski, FDA Forensic Chemistry Center, Trace Examination Section, USA

Preface

For many years the practices of infrared and Raman spectroscopy were confined largely to dedicated academic, industrial or national research laboratories. Major technical advances over the past 10–20 years have afforded a significant broadening of the applicability of these vibrational spectroscopy techniques as a whole.

Instruments used to be large, complicated to operate, with even the simplest experiment often challenging to set up and run. Advances in technology have resulted in smaller, easier to use instrumentation that is much more user-friendly. Demands and needs from users for increased portability of scientific instrumentation have produced spectrometers and interferometers of small dimensions and of sufficient quality such that handheld Raman and Fourier transform infrared (FT-IR) instruments have been realized over the past few years, opening up much wider application of Raman and FT-IR spectroscopy to forensic science applications, particularly for adoption into field usage.

This book is intended to introduce a novice or established spectroscopic practitioner of analytical chemistry to the technical elements of Raman and infrared spectroscopy as applied to forensic science, outlining several proven and potential applications within this field. It is not intended to describe advanced topics such as non-linear Raman or time-resolved vibrational spectroscopy, but rather to address the applications of Raman and IR spectroscopy to the different fields of forensic work, from explosives to narcotics and from bio-agents to works of art.

The early chapters introduce the reader to the principles of forensic science and how Raman and IR spectroscopy can be applied. Chapter 2 introduces the basics of vibrational spectroscopy and the instrumentation that may be found routinely, ranging from bench-top through portable to handheld systems. To complement this, Chapter 3 discusses sampling techniques and considerations of analysis to aid in the non-destructive analysis of samples.

The following sections of the book are split into overviews and case-study chapters comprising topics covering the following areas: crime scene, counter terrorism/homeland security, drugs of abuse, archaeology/mineralogy and consumer products, including pharmaceutics. Each chapter is written by internationally respected scientists. This broad selection of topics is complemented by relevant application examples, highlighting how IR, Raman and terahertz (THz) spectroscopy can be applied to these fields. To complement this, each chapter is referenced so that users can read up on and investigate areas that interest them.

Commercial Raman, near-IR, mid-IR and THz spectrometers differ widely in their applicability, configuration and performance. No one system can be applied to all possible applications; specific manufacturers are mentioned within the text to identify a particular approach, configuration or application. Where manufacturers are mentioned, this does not infer an endorsement, but it may be useful to the reader to understand the special design or application objectives and requirements.

It is the editors' and contributors' hope that those just developing an interest in the application of infrared and Raman spectroscopy to forensic analysis and that those who practice it already will find this book useful not only as a source of new information, but also as a reference work. Furthermore, we hope that it will inspire readers to delve deeper into the applications of vibrational spectroscopy that have not yet been explored in this rapidly expanding field.

Notes on convention (or lack of them): it is usual practice to plot IR spectra from high wavenumber (on the left) to low wavenumber (on the right); this convention is held throughout the book. Raman shifts are often shown plotted either way, that is, low shift (on the left) to high shift (on the right) or vice versa. It has not been possible to ensure all the spectral plots have been standardised in this way, particularly those that have been reproduced from other publications, so readers are directed to check the individual plots. In addition, the Raman shift axis only shows the Stokes-shifted bands, unless stated otherwise. Mostly Raman shifts are noted in the unit cm−1, rather than the more correct form of Δ cm−1.

John M. Chalmers

Howell G.M. Edwards

Michael D. Hargreaves

August 2011

Section I

Introduction

Chapter 1

Introduction and Scope

John M. Chalmers¹, Howell G.M. Edwards² and Michael D. Hargreaves³

¹VS Consulting, Stokesley, UK

²Chemical and Forensic Science, University of Bradford, Bradford, UK

3Thermo Scientific Portable Optical Analyzers, Thermo Fisher Scientific, Wilmington, Mass., USA

1.1 Historical Prologue

Forensic science can be defined as the application of scientific principles to the public domain in courts of law, which were held by the Romans in the public forum.

Although evidence of the unlawful killing of a human being was presented in public fora from quite early times, such as the post mortem examination of the body of Julius Caesar after his assassination, which revealed 23 stab wounds but only one of which was judged to be fatal, and poisoning in particular, where the appearance of organ degradation gave rise to the conclusions that toxic materials had been ingested, these pronouncements were in the realm of the prototype medical examiners and pathologists and not chemical analysts [1].

The first chemical analysis of an historical artefact that can be viewed as forensic in its approach was reported in the literature by Sir Humphry Davy in 1815. The development of the Marsh test for arsenic poisoning in 1836 was a landmark event that launched the birth of analytical forensic science. This was followed quite rapidly by the public fascination for scientific analysis applied to crime as appeared in the Victorian gothic novel in detective stories such as Armadale [2], authored by Wilkie Collins in 1866, and culminating in the adventures of Sherlock Holmes, whose creator Sir Arthur Conan Doyle introduced in A Study in Scarlet [3] in December 1887, just one year before the notorious Jack the Ripper brought terror to the East End of London, England. Subsequently, some 56 of Conan Doyle's short stories, published in the popular Strand Magazine between 1891 and 1927, commencing with A Scandal in Bohemia, established the Sherlock Holmes genre to an appreciative public; an unsuccessful attempt by Conan Doyle to terminate the Holmes character in a fatal meeting with his arch-enemy Professor Moriarty at the Reichenbach Falls in 1893 resulted in intense public outrage; such was the growing public perception of the scientific approach to crime solution at that time, and the detective re-appeared to his public once more in The Hound of the Baskervilles in 1901.

The seemingly voracious appetite of readers for the scientific detection of crime in the mid-nineteenth century is illustrated in Armadale by the attempt by Miss Gwilt to murder her fiancé, the eponymous Armadale of the novel, using a chemical reaction between an unspecified liquid in a purple flask supplied by a mysterious admirer and the generation of an odourless, tasteless and undetectable gas whilst he was sleeping. The author had acknowledged appropriately the assistance and advice of an un-named professional chemist in the preface to his novel, thereby lending a veneer of respectability and credibility to the background science contained in the text! The activities of Sherlock Holmes and his analytical skills and observations pervade the Conan Doyle stories and hint at the prophetic accomplishments of Conan Doyle, said to be based upon his University mentor, Dr Joseph Bell, that were significantly in excess of the extant knowledge in the late 1880s.

It is, therefore, perhaps not surprising that the first recorded acceptance of forensic chemical analysis used in a court of law to secure a conviction occurred as late as 1912 in France when Emile Gourbin, who had a seemingly good, watertight alibi, was faced with evidence of his poisoning of his lover, Marie Latelle, using contaminated poudre de riz, a customised cosmetic preparation that was fashionable at that time. The scientific analyst at the centre of this landmark prosecution was Edmond Locard, who demonstrated that particles of material under the fingernails of the accused matched the composition of the cosmetic preparation purchased from a local pharmacist but with the addition of bismuth. It is interesting that some years later in 1927, this same Edmond Locard, then a professor at the University of Lyon, proposed his now famous and fundamental eponymous Exchange Principle that is at the basis of modern forensic science: that "every vigorous contact leaves a trace".

There quickly followed the establishment of analytical laboratories internationally dedicated to forensic science, early examples of which in France was that of Edmond Locard in 1910 and in the United States by August Vollmer in 1924. The first Chair of Legal Medicine in a University, so establishing an academic forensic protocol, was established in Harvard in 1932.

A classic case of murder by poisoning that first escaped detection in the United Kingdom but which relied heavily upon chemical analysis to secure prosecution, was that of Major Herbert Armstrong, who systematically poisoned his wife using arsenic in 1921; only when he tried to repeat the exercise to remove a business rival did Major Armstrong receive his just desserts – a close relative of the business rival was the local pharmacist, who recollected that Armstrong had purchased large quantities of arsenic over the previous year. Exhumation of the Major's wife revealed to the pathologist a large amount of arsenic in her remains – as exemplified by the adoption of the Marsh test; Armstrong was convicted of her murder and hanged in 1922 [4, 5].

The use of infrared spectroscopy to determine molecular structure has its roots firmly established in the nineteenth century, since the discovery of the infrared region of the electromagnetic spectrum in 1800 by Sir William Herschel. But the Raman effect was first observed experimentally only in 1928 by Sir Chandrasekhar Raman, following a theoretical prediction by Smekal in 1923, which resulted in the Nobel Prize for Physics for Raman in 1930. Lord Rayleigh, commenting on the observation of the Raman effect, judged this to be one of the four most important discoveries in physics of all time.

At first, the relative ease of recording photographically the wavenumber-shifted radiation of the weak Raman effect compared favourably with the point by point plotting of moving coil galvanometer signals used in infrared spectroscopy and gave an impetus to Raman spectroscopy in molecular structure analysis that surpassed the infrared investigations. However, it became quickly apparent that the onset of fluorescence emission swamped the weaker Raman data, often saturating the photographic emulsions used in the spectrographic recording. For many years this disadvantage was paramount in Raman spectroscopy and it was only the advent of tuneable laser excitation and novel methods of detection coupled with computerised data acquisition that offered possibilities to circumvent it. Hence, although mid-infrared spectroscopy started to be applied to forensic analysis from the 1950s, Raman spectroscopy was only similarly used from the 1990s; in both cases, however, the coupling of a microscope to the analytical spectrometer was a necessary advancement. The advent of portable and handheld spectrometers has further advanced this application space, meaning analysts can analyse in situ artefacts of interest.

In the past, the greatest stumbling block to the application of both infrared spectroscopy and Raman spectroscopy to forensic structural analysis and molecular characterisation was the quantity of material that was required for analysis and the further requirement that in most cases the preparation of the specimen for the optical illumination processes necessitated the destruction, mechanically or chemically, of the sample itself. This was paralleled in chemical analysis, in that even as early as 1815 Davy [6] recognised that his experiments on the archaeological decorative wall painting artefacts from the recently excavated Pompeii archaeological site resulted in the complete destruction of the samples presented for analysis. Even 100 years later, Eccles and Rackham [7], in their comprehensive studies of porcelains in the British Museum and Victoria and Albert (V&A) Museum collections in the United Kingdom required the donation of multiple items from tea and dinner services which were sacrificed in the determination of factory body chemical compositions using wet chemical analysis. It was, therefore, a very important turning point in the mid-twentieth century when it was realised that advances in spectroscopic technology now made available for the first time the possibility of acquiring chemical molecular data from valuable specimens that was truly non-destructive of the sample [8–10].

In the 1970s great strides forward were made when optical microscopes were coupled with spectroscopes to provide chemical identification data from spatially minimal regions of samples. The first infrared microspectrometer appeared in the mid-1960s and the first Raman microspectrometer was announced from the laboratory of Michel Delhaye and Paul Dhamelincourt in the University of Lille, France, in 1976; this was termed a molecular optical laser examiner (MOLE) which was quickly applied to the investigation of several interdisciplinary problems, including some fragments from oil paintings. The first papers in this area, which could perhaps be classified as ground-breakers for the later application of Raman microscopy in forensic science, then appeared from the laboratories of the Natural History Museum of Paris under the direction of Bernard Guineau, who analysed the inorganic pigments from mediaeval manuscripts in museum collections [11].

1.2 The Application of Infrared Spectroscopy and Raman Spectroscopy in Forensic Science

Both techniques share a microspectroscopic capability for the recording of data from particles in the nanogram to picogram range, which is paramount for the interrogation of specimens non-destructively; little if any sample preparation is required, which means that vibrational spectroscopy is often regarded as a first-pass analytical technique for the screening and identification of suspect materials which then may require some further analytical data from more destructive operations.

Naturally, infrared spectroscopy and Raman spectroscopy have particular advantages and disadvantages, which sometimes dictate that one or other technique is preferred for special applications: for example, the presence of water or hydrated chemical species in specimens can hamper the mid-infrared analysis and the ready inaccessibility with mid-infrared spectrometers of low wavenumber features below 400 cm–1 can severely limit the characterisation of drug polymorphs and heavy metal inorganics. The operational dependence of the Raman effect upon molecular polarisability rather than the dipole moment for the infrared means that polar groups such as –OH and C=O are better seen in the infrared spectrum, whereas homopolar unsaturation involving C=C groups and N=N is better evidenced in the Raman spectrum. The degradation of keratotic materials such as skin, hair and nail associated with human remains in burial environments is best followed through the –S–S– modes near 500 cm–1 in the Raman spectrum as this feature does not appear at all in the infrared spectrum.

A major factor in Raman spectroscopy applications to materials in a forensic context is the ability to overcome or circumvent fluorescence emission and this needed the advent of laser excitation at longer wavelengths from the visible into the near-infrared region of the electromagnetic spectrum, typically at 785, 830 or 1064 nm. Modern state of the art vibrational spectroscopic laboratories involved in forensic analysis therefore have several laser sources available for adoption in this respect, especially where samples are highly coloured, such as pigments and dyes.

Field–use capability of instrumentation is a desirable development for the adoption of miniaturised infrared and Raman spectrometers at crime scenes and for the examination of large or very fragile objects and artefacts. In this context, the penetration of packaging and the interrogation of specimens through transparent or semitransparent containers is also possible through shrewd selection of the radiation wavelength, and the possibilities of the terahertz (THz) region of the spectrum is affording much interest in this respect. An important factor here is the incorporation of database recognition packages within the chosen instrument to identify materials that are of relevance to forensic examination, such as drugs of abuse, explosives, chemical warfare agents and their chemical precursors, which may be correlated with drugs factories and synthetic bomb-making crime scenes. The use of such instrumentation by non-expert security forces and agents by the adoption of selection algorithms is also a real challenge for specialised spectroscopists.

Of special interest is the recording of seized specimens of suspect materials that can be examined whilst still contained in their evidential bags sealed at source and which need not be sampled or opened in the analytical laboratory; the data from such analyses carried out under these conditions can circumvent any doubts raised about the integrity of the preservation of the evidential material between the source and the analytical laboratory This type of analysis has a distinct parallel in the scientific examination of artworks, which for operational reasons cannot be removed from their transparent covers or holders; in a forensic art study [12] of the Armada Jewel made for Queen Elizabeth I by Nicholas Hilliard, a prestigious court limner, in 1588, and now in the V&A Museum in London, the Raman spectroscopic characterisation of the pigments used was achieved by interrogation of the painting through its rock crystal cover plate, with some rather surprising results.

Finally, the so-called molecular fingerprint that is provided from the mid-infrared spectrum or Raman spectrum must be well-characterised and robust: for example, the question arises as to how many vibrational spectroscopic features are necessary to define a particular compound unequivocally – this is not easy to assess and sometimes it is relatively easy to differentiate between chemically similar materials and not so in other cases. In pigment characterisation, for example, the two forms of lead (II) oxide are readily differentiated in the low wavenumber region using Raman spectroscopy; in the geological field, anatase can readily be differentiated form rutile and brookite, yet all are titanium (IV) oxides; and the polymorphs of calcium carbonate, calcite and aragonite are easily discriminated by both mid-infrared and Raman spectroscopy. In each case, the detection of more than one vibrational spectroscopic feature is essential for correct identification of the specimen. In the area of drugs analysis, cocaine hydrochloride and freebase cocaine (crack cocaine) can be differentiated as can caffeine base and caffeine hydrate. The power of these techniques thus rests in an appreciation of the necessity for recording spectra of the best quality consistent with speed and rapid identification that is often a de rigueur requirement of the end user.

This book comprises overview chapters and case study chapters written by experts and practitioners who have a wealth of experience in the application of infrared, Raman and THz vibrational spectroscopic techniques to forensic analysis in which several of the points made above are investigated and exemplified; several outstanding challenges remain that need the collaboration of vibrational spectroscopists, forensic practitioners and front-line security forces in the advancement of technologies in the fight against crime, contraband trafficking and international terrorism.

References

1. W.J. Tilstone (2006) Forensic Science: An Encyclopedia of History, Methods and Techniques, ABC-CLIO, Santa Barbara, Calif.

2. Armadale (1866) Wilkie Collins, Smith, Elder, London.

3. A Study in Scarlet, in Beeton's Christmas Annual (1887) November issue, Ward Lock, London.

4. J.W. Nicholson (1992) Arsenic – the enigmatic element, Education in Chemistry, July, pp. 101–103.

5. J.H.H. Gaute and R. Odell (1989) The New Murderer's Who's Who, Harrap, London.

6. H. Davy (1815) Phil. Trans. Roy. Soc., 105, 97.

7. H.A. Eccles and B. Rackham (1922) British Museum, London.

8. H.G.M. Edwards and J.M. Chalmers (2005) Raman Spectroscopy in Archaeology and Art History, Royal Society of Chemistry Publishing, Cambridge.

9. E. Ciliberto and G. Spoto (eds) (2000) Modern Analytical Methods in Art and Archaeology, Chemical Analysis Series, 155, J. Wiley & Sons, Ltd, Chichester.

10. P. Vandenabeele, H.G.M. Edwards and L. Moens (2007) Chemical Reviews, 107, 675.

11. B. Guineau, M. Lorblanchet, B. Gratuze, L. Dulin, P. Roger, R. Akrich and F. Muller (2001) Archaeometry, 43, 211.

12. A. Derbyshire and R. Withnall (1999) Journal of Raman Spectroscopy, 30, 185.

Chapter 2

Vibrational Spectroscopy Techniques: Basics and Instrumentation

John M. Chalmers¹, Howell G.M. Edwards² and Michael D. Hargreaves³

¹VS Consulting, Stokesley, UK

²Chemical and Forensic Sciences, University of Bradford, Bradford, UK

³Thermo Scientific Portable Optical Analyzers, Thermo Fisher Scientific, Wilmington, Mass., USA

2.1 Introduction

This chapter is a consideration of some of the basic principles, differences and commonalities, and instrumentation underlying applications of the vibrational spectroscopy techniques of infrared (IR) spectroscopy, terahertz (THz) spectroscopy and Raman spectroscopy to forensic sciences and related areas of study. Chapter 3 builds on this and focuses on the sample preparation and presentation methods associated with the practice of these techniques and issues as they relate to recording vibrational spectra.

In books dealing with vibrational spectroscopy, it has been a common practice to separate out into sections IR and Raman spectroscopy; however, in this chapter to help emphasise their commonalities and differences, we have endeavoured, as much as is convenient, to mingle and discuss them simultaneously. It is not intended that this chapter be viewed as an in depth treatise of IR, Raman and THz spectroscopy, but rather as a useful short introduction to what follows in the subsequent chapters.

2.2 Vibrational Spectroscopy Techniques

2.2.1 The Basics and Some Comparisons

In the context of forensic science, IR, Raman and THz spectroscopy can be considered as tools for the identification, interrogation or detection of materials, such as a piece of evidence, via their molecular structure characterised from their vibrational spectra; the spectra may be interpreted in terms of the functional groups present (or absent), pattern-matched to a library database of reference spectra, or quantified or classified from relative band intensities.

2.2.1.1 Wavelength/Wavenumber Ranges and Selection Rules

IR and THz radiation form part of the energy continuum of the electromagnetic spectrum. Their frequencies lie between the lower energy visible wave region and the higher energy microwave region. As is very well documented elsewhere, for example References [1–6], IR and THz absorption spectra are related to the interaction of this electromagnetic radiation with molecular vibrations. Raman spectra arise from the inelastic scattering of monochromatic radiation as a consequence of its interaction with molecular vibrations. Raman spectroscopy principles have also been very well documented elsewhere, for example References [2, 4, 5, 7–11]. A prerequisite for a molecular vibration to be IR active, that is, give rise to an absorption band, is that there must be a change of dipole moment during a normal mode of vibration of the molecule; for a vibration to be Raman active requires that there is a change in polarizability during the normal mode of vibration. These conditions convey the selection rule difference between IR and Raman spectroscopy.

A very simplistic schematic comparison between mid-IR and Raman spectroscopy techniques is shown in Figure 2.1. It depicts that monochromatic radiation (of wavelength λ) striking a sample is scattered in all directions; much of this scatter is elastic (Rayleigh scattering), some of it (a small proportion, ca. 1 part in 10⁶ or less) is changed in wavelength, either increased (λ + λv) or decreased (λ − λv) – the Raman shift, by an amount corresponding to the sample's Raman active vibrational mode wavelength λv; the schematic of a mid-IR transmission set-up depicts that radiation from a broadband source passing though a sample is attenuated (absorbed) by the IR active vibrational mode of wavelength λv. The Raman shift may be either positive (Stokes) or negative (anti-Stokes) relative to the unchanged wavelength; in conventional Raman spectroscopy it is the Stokes-shifted Raman bands that are measured, since these have the higher intensity and involve transitions from lower to upper energy vibrational levels; anti-Stokes shifts occur as a consequence of a few molecules existing in an excited vibrational energy level undergoing vibrational transition to a lower vibrational energy level [4, 7–10].

Figure 2.1 Simplified schematic comparison between Raman scatter and mid-IR absorption (transmission measurement), see text for details. (Please refer to the colour plate section.)

IR active normal modes of vibration of organic molecules give rise to absorption bands in the mid-IR region; this region is defined as the 2.5–25.0 μm wavelength region. Since, the wavelength, λ, in a vacuum is related to the frequency, ν, by:

(2.1) equation

where c is the speed of light in a vacuum, then this wavelength region is equivalent to 4000–400 cm−1. The reciprocal centimetre, cm−1, is the unit of wavenumber, , that is, the number of waves in a unit length. Thus:

(2.2) equation

Today, almost all mid-IR spectra are recorded and displayed in an abscissa linear wavenumber format from 4000 to ca. 400 cm−1. Raman spectra are often similarly displayed, although, because they occur as a consequence of inelastic scatter from a single wavelength source, strictly they should be displayed on their abscissa scale as a Raman shift, which is denoted by Δcm−1, and, similarly to mid-IR spectroscopy, Raman active normal modes of vibration of organic molecules occur in the range from 4000 to ca. 400 Δcm−1. Consequently, mid-IR and Raman spectra are considered a complementary pair of fingerprinting tools that can be used to characterise the molecular structure of a sample, and as a consequence invaluable tools for many forensic studies. In practice, Raman spectra are often recorded and may be displayed readily to a much lower wavenumber, without the need for changing the optical components of the spectrometer, see later. However, near-, mid- and far-IR (and THz) spectrometers may use different designs or utilise components of differing optical properties. These are also discussed briefly later.

The IR region of the electromagnetic spectrum is divided into three regions, known in increasing wavelength coverage, as the near-, mid- and far-IR regions; these cover the wavelength ranges of about 780–2500 nm (2.5 μm), 2.5–25.0 μm and 25 to about 1000 μm, respectively; these are equivalent to ca. 12 800–4000 cm−1, 4000–400 cm−1 and 400 to ca. 10 cm−1, respectively. Raman spectra typically cover the wavenumber region between ca. 400–5 Δcm−1 to ca. 4000–3800 Δcm−1, the former limit depending very much on the spectrometer design and purpose and the optical components present. To match linear wavenumber mid-IR spectra, Raman spectra are frequently displayed as extending between ca. 4000–3800 Δcm−1 and ca. 400–5 Δcm−1. While mid- and far-IR (and Raman) spectra are usually always displayed today in a linear wavenumber format, near-IR spectra are often displayed in a linear wavelength format. Traditionally, far-IR spectroscopy is understood to extend from 400 cm−1 (25 μm wavelength) to about 10 cm−1 (1 mm wavelength). The terahertz frequency region (see Chapter 5.5) lies between the IR and microwave regions of the electromagnetic spectrum, that is, from 0.3 to 10 THz. Terahertz spectroscopy usually refers to spectroscopy undertaken using optical frequencies covering the range from 0.3 to 3.0–6.0 THz, that is, about 10 cm−1 to 100–200 cm−1 (from 0.1 mm to 50 μm wavelength) [12, 13]; see also Chapter 5.5. Hence, the classical far-IR spectroscopic region encompasses much of what has become commonly known today as the THz spectroscopy region. The main discrimination has arisen because of the types of instrumentation used; see later and Chapter 5.5. Traditional far-IR spectroscopy uses instrumentation that is similar to that used for mid-IR spectroscopy, but with appropriate changes in key optical components, for example, for a Fourier transform (FT) spectrometer, the source, beamsplitter and detector [3]. THz spectroscopy came to the fore this century with the launch of commercial spectrometers based on time-domain THz spectroscopy in which the THz radiation is generated using ultrashort laser pulses [13, 14]; see later and Chapter 5.5. These spectrometers are generally much more sensitive than FT-IR spectrometers operating in the low frequency far-IR region. Figure 2.2 shows the relationship between the vibrational spectroscopy techniques, their units and occurrence within the electromagnetic spectrum.

Figure 2.2 Schematic showing relationship between ranges for vibrational spectroscopy and the electromagnetic spectrum. Note: (a) is linear in wavelength, (b) is linear in wavenumber. The Raman ranges depicted represent from the exciting line to Δ4000 cm

−1.

2.2.1.2 Sampling Considerations

While Chapter 3 focuses primarily on vibrational spectroscopy sampling techniques, it is worthwhile here mentioning briefly a few comparisons between IR and Raman spectroscopy sampling techniques. For many laboratory-based forensic-type applications, both mid-IR and Raman spectroscopy are commonly used in microscopy-based configurations in which the vibrational spectroscopy spectrometer is interfaced to an optical microscope system, see later, which enable both coincidental visual and spectroscopic examinations, either as single-point, mapping or imaging measurements. As we will read exemplified throughout chapters within this book, compared with mid-IR spectroscopy, Raman spectroscopy offers some particular practical advantages to many sampling situations imposed by forensic-type investigations and studies. Raman spectroscopy is a scattering technique that utilises a laser as a source, and as such therefore there is a minimal requirement for any sample preparation and presentation and the technique has an inherently higher spatial resolution than IR spectroscopy techniques, particularly when visible or short-wavelength near-IR lasers are used. Raman microscopy excited with a short wavelength visible laser offers the capability of examining a sample size of about 1 × 1 × 5 μm, which has clear benefits to many sample-limited forensic investigations, while, in contrast, high spatial resolution mid-IR microscopy studies are constrained by diffraction effects and therefore usually limited to a lateral sample dimension of 10 μm or greater [15]. The higher spatial resolution achievable with Raman microspectroscopy has however the potential to lead to sub-sampling issues, since the lower volume of material examined in a single point examination may not be representative of the study issue. A significant practical advantage of Raman (and near-IR) spectroscopy over mid-IR spectroscopy has been that of the comparative wavelengths of the sources used; the shorter wavelengths used with Raman (and near-IR) spectroscopy enable the use of telecommunications-type optical fibres, see later, which has led to the development of remote fibre-optic probes for applications that can, where necessary, be operated over long distances (>10 m in some instances) and which are well suited to a range of in situ/on site/in the field forensic applications. As discussed later, mid-IR fibres are much less robust than those capable of being utilised with Raman or near-IR spectrometers and also are only able to be used effectively over shorter distances. For field use, hand-held spectrometers for mid- and near-IR and Raman spectroscopy applications are now all commercially available, see later.

2.2.1.3 Sensitivity, Surfaces and Signal Enhancement Techniques

While practically Raman spectroscopy may be perceived as offering many advantages compared with mid-IR spectroscopy, conventional Raman spectroscopy is considered to be the lower sensitivity technique; so, for instance, as a generality, mid- and near-IR spectroscopy have lower limits of detection for trace components in formulations and mixtures. Where appropriate, the sensitivity of Raman spectroscopy to a chromophore may be significantly enhanced by tuning the excitation Raman laser wavelength to be in coincidence (or near so) to the absorption wavelength maximum of the chromophore in its electronic (visible or UV) spectrum, thereby generating an enhanced intensity spectrum – resonance Raman (RR) spectrum – of the chromophore. Enhancements of up to 10⁶ in Raman scattering cross-sections have been reported [4, 10]. For surface-layer specific measurements and many bulk and micro-measurements, then the internal reflection technique of mid-IR reflection spectroscopy has many applications in forensic science; the technique, commonly referred to as the ATR technique (ATR; attenuated total reflection), see Chapter 3, may be used to record a mid-IR (or THz) spectrum of a sample in optical contact with an internal reflection element (IRE); the surface layer probed is inter alia both wavelength and IRE dependent, but in the mid-IR region can vary from between about 0.3–0.5 μm at the high wavenumber end to about 1.0–3.5 μm at the low wavenumber limit. Surface enhanced Raman spectroscopy (SERS; see, for example, Chapter 6.3 by Faulds and Smith) can be used to overcome the sensitivity limitation to the detection of trace amounts of material. In SERS, a target analyte is adsorbed onto a suitable surface, commonly of Ag or Au, to create a plasmon with a resonance coincident with that of a visible or near-IR laser used to excite Raman scatter. Typical signal enhancements with SERS may be in the range of 10⁵ to 10⁶. Enhancements of the order of 10¹³ to 10¹⁵ for the chromophore of an analyte may be achieved by a combination of RR and SERS – surface enhanced resonance Raman spectroscopy (SERRS), see Chapter 6.3.

2.2.1.4 IR and Raman Bands

As mentioned above, the normal modes of vibration of organic molecules occur in the range ca. 4000–400 cm−1 (mid-IR) or ca. 4000–400 Δcm−1 (Raman), and not every fundamental molecular vibration of a molecule gives rise to a mid-IR absorption and/or a Raman-shifted band within its characteristic spectra: it depends on the selection rule. As stated already also: to absorb mid-IR radiation there must be a change of dipole moment during the molecular vibration, while to give rise to a Raman-active band there must be a change of polarizability during the molecular vibration; the relative intensities of these depend on the magnitude of the changes. Consequently, carbonyl groups (C=O), which are polar, tend to feature as strong stretching mode (νC=O) bands within a mid-IR spectrum; while more symmetrical vibrations such as the νC=C and νS-S can be relatively very much more weakly absorbing or even non-absorbing in a mid-IR spectrum but feature strongly within a Raman spectrum.

Molecules may undergo either internal (intra-) molecular bond vibrations or between molecules (inter-) molecular vibrations (e.g., a hydrogen-bonded pair). Although, these vibrations are not strictly independent, in many cases they are essentially localised and give rise to a mid-IR absorption band or Raman shifted band within a narrow wavenumber region that is characteristic of a particular moiety. For example, if one considers a –CH2– (methylene) group, see Figure 2.3, then this is characterised by bands attributable to bond stretching and deformation vibrations; these normal modes of vibration of the methylene group give rise to bands in either or both the mid-IR and Raman spectrum of the molecule, with relative intensities that depend, respectively, on the existence (selection rule) and strength of the dipole moment change (IR) or polarizability change (Raman) during the vibration, and which, for example, occur at: ca. 2940–2915 cm−1 for the antisymmetric stretch, 2870–2840 cm−1 for the symmetric stretch and 1480–1440 cm−1 for the bend (scissor) deformation mode; (note, for convenience, we have omitted the Raman shift notation of Δ).

Figure 2.3 –CH2– group fundamental vibrations (see text for details). Arrows indicate respective motions of the atoms; + and − indicate opposite motions perpendicular to the plane of the molecular group. Reproduced with permission, from J.M Chalmers, Mid-infrared Spectroscopy: The Basics, Chapter 2, pp. 29–66 in Biomedical Applications of Synchrotron Infrared Microspectroscopy. A Practical Approach, ed. D. Moss, RSC Publishing, Cambridge (2011).

An example, using nitrocellulose, of the some of the complementarities of mid-IR and Raman spectra and sampling requirements is shown in Figure 2.4 [16]. The Raman spectrum was recorded directly from neat powder, while a thin film from solution was prepared for the mid-IR transmission measurement; the differing relative intensities between bands within the two spectra are also clearly evident.

Figure 2.4 Comparison of the vibrational spectra of nitrocellulose (13.4% N). Top: mid-IR transmission spectrum recorded from a prepared thin film. Bottom: Raman spectrum recorded from neat powder. Figure, with axes labels adapted, reproduced from Reference [16] with permission of Elsevier B.V.

Between 4000 and 1500 cm−1 (excluding overtone, combination and Fermi resonance bands) [5, 17], a mid-IR or Raman spectrum can be conveniently sub-divided into three characteristic regions. These regions are approximately: 3600–2500 cm−1, which is associated with X–H stretching vibrations, where X is C, O, N or S (P and Si, being heavier, give rise to X–H stretching vibrations between 2500 and 2100 cm−1); 2500–2000 cm−1 is the region in which triple (e.g., −C N) and cumulative double bonds (e.g., −N=C=S−) occur; between about 2000–1500 cm−1, the fundamental stretching vibrations of double bonds occur (e.g., C–O, C=C, C=N); while the region between 1500 and 400 cm−1 is known as the fingerprint region, so-called, because, for instance, quite similar molecules can give different band patterns; these band patterns are sometimes subtle in this region reflecting, for example, differences in polymorphic form, degree of crystallinity, extent of orientation, state of hydration and so on. These attributes are keys to the value and success of mid-IR and Raman spectroscopy in being able to fingerprint, identify and/or classify materials.

Absorption bands that are observed within the near-IR region are essentially the overtone and combination bands of X–H vibrations, where X is C, O and N (some second and third overtone bands of the νC=O band may also be observed). These bands are much weaker in intensity (becoming increasingly weaker with each increasing overtone progression) than the normal modes observed within the mid-IR region and consequently, in general, much greater sample volumes are required to observe near-IR spectra than mid-IR spectra; these properties have led to the development of near-IR analysis (NIRA) as a strong tool for the quality assurance, authentication and classification of many condensed-phase materials, particularly foodstuffs and related products. There is no Raman spectroscopy equivalent to the wavenumber range achieved in near-IR spectroscopy and analysis.

The bands that are observed in the far-IR and THz and Raman low wavenumber (< ca. 400 cm−1) regions of organic molecules include skeletal modes. They arise from complex vibrations; these include internal mode whole molecule vibrations or external inter-molecular modes such as H-bonding [18]; see also Chapter 5.5 by Burnett et al. Spectra within this region can be useful for distinguishing between structurally similar compounds, particularly condensed-phase crystalline materials. The low wavenumber region also contains useful information on heavy atom vibrations, which can be especially useful for identifying the presence of or distinguishing inorganic molecules. A simple and good example of the latter, shown in Figure 2.5 [19], is between two polymorphs of titanium dioxide, TiO2; see also Figure 7.1 of Chapter 7.3 by Burgio.

Figure 2.5 Raman spectra of anatase and rutile polymorphs of titanium dioxide. Adapted from Reference [19] with permission of John Wiley & Sons, Ltd.

2.2.2 Quantitative and Classification Analyses

Before closing this section on the similarities and differences between the vibrational spectroscopic techniques, we now consider briefly quantitation and classification analyses. IR absorbance and Raman scattered band intensities are, to a first approximation, both linearly proportional to the number density of species giving rise to the band. Therefore, direct comparisons of relative band intensities within an observed Raman spectrum or set of spectra is relatively straightforward; this is also the case with relative intensities within spectra recorded using the photoacoustic sampling technique in FT-IR spectroscopy, since these relate directly to absorbance, although, for spectra recorded using rapid scan FT-IR spectroscopy, particularly, one may need to carefully consider the relationship between the sample thickness and the optical modulation frequency of the spectrometer, in addition to that between the optical absorption length and the thermal diffusion length [20, 21]. IR (or THz) spectra recorded in transmission or using the ATR or mid-IR transflection sampling techniques must therefore be converted to absorbance spectra before any valid quantitative measurements can be made; modern spectrometers do this automatically and can display absorbance spectra directly; near- and mid-IR spectra recorded using the diffuse reflection sampling approach are more usually converted to absorbance-equivalent spectra using the log10(1/R) relationship or Kubelka–Munk algorithm, respectively, see below. From these IR absorbance or absorbance-equivalent IR spectra and Raman spectra, relative (height or area ratios) or normalised band intensities may be measured and used for quantitation, while sets of full spectra or selected regions thereof may be analysed using multivariate data analysis techniques (chemometrics) to develop quantitative calibrations and/or provide visual classifications to a component's presence or to discriminate between sample types within a set of similar samples.

While as stated above, relative Raman band intensities within a spectrum may be used directly in quantitative measurements, IR spectra recorded using transmission and reflection sampling techniques must be first converted to absorbance (or equivalent) before they can used for quantitation purposes. Although this is today done automatically by the spectrometer software, the process is:

a. IR transmission spectrum conversion to absorbance values and Beer's Law:

In a transmission (or ATR or transflection) measurement, if I and I0 represent the intensity of radiation passing through the sample and reaching the detector and the intensity of radiation reaching the detector without the sample being present, respectively, then the percentage of radiation transmitted, T, by the sample at a given wavenumber, , can be represented as:

(2.3) equation

[Transmittance, T, has values between 0 and 1, so the transmittance of the sample at a particular wavenumber, , is given by ].

The more a sample absorbs the mid-IR radiation, the lower is the value of T. The transmittance of a pure sample of path length (thickness) l, where l is in centimetres, is expressed as:

(2.4) equation

is the linear absorbance coefficient (cm−1) at wavenumber [3].

The amount of mid-IR radiation absorbed, A, by the pure sample at is expressed as:

(2.5) equation

A is known as the absorbance. For a pure sample:

(2.6) equation

where: c is the concentration of the sample, and c × l represents the relative number of absorbing molecules in the mid-IR beam; is the absorptivity at wavenumber . Equation (2.6) is commonly referred to as Beer's Law.

b. IR diffuse reflection:

(i.) Mid-IR: the mid-IR diffuse reflection spectrum of a finely powdered sample is typically recorded diluted within a non-absorbing powdered matrix powder such as KCl, see Chapter 3; in an FT-IR spectroscopic measurement, the single-beam diffuse reflection spectrum of the neat diluent is recorded as the background spectrum, and the diffuse reflectance R∞ of the sample is then calculated as the ratio of the single-beam spectrum of the diluted sample to that of the neat diluent (see also Chapter 3).

R∞, is then usually converted to the Kubelka-Munk function, f(R∞) as:

(2.7) equation

where k′ represents the absorption coefficient, which to a first approximation may be considered as equivalent to the linear absorption coefficient of the sample, and s is the scattering coefficient of the sample.

(ii.) Near-IR: in near-IR diffuse reflection measurements, the sample diffuse reflectance, R, as with mid-IR, is typically calculated as the ratio at each wavelength of the single-beam spectrum of the sample to that of an appropriate reference, such as a ceramic or polytetrafluoroethylene, PTFE, disc. However, unlike mid-IR diffuse reflectance spectra, near-IR diffuse reflectance spectra are usually output as log10(1/R), since in practice this tends to yield more reproducible determinations [22].

Providing there are no significant concentration-dependent molecular interactions, and other parameters such as temperature and environment are essentially invariant, then a normalised Raman band intensity value or normalised mid-IR (or near-IR or THz) absorbance intensity could be used for a simple univariate calibration and measurement, such as the concentration of an analyte. Normalisation can be achieved with some IR transmission measurements by using a fixed sample pathlength, for example, such as a fixed pathlength solution transmission cell; this is not achievable with a Raman measurement. In a mixture then it may be possible to determine the concentration of a minor component by using an essentially invariant band intensity of a major component for normalisation, or alternatively, if no such option is available, a bivariate band intensity ratio method may be developed that is correlated with such as a plot against the relative mass concentrations of two components in a homogeneous mixture. While such simply based analyses may be widely utilised for well-understood simple analytical methods, it is more common today to use more robust multivariate data regression (statistical) analysis procedures such as partial least squares (PLS).

2.2.2.1 Multivariate Data Analyses

PLS is just one of a large suite of multivariate data processing methods now made use of to process, quantify and/or classify vibrational spectroscopic data, commonly referred to as chemometric techniques. (The use of these approaches became commonplace with near-IR spectral data, where today they are widely used in many authentication and quality control methods.) PLS is part of the sub-set known as a supervised method; it is an algorithm used for calibration and analysis of spectroscopic data.

Providing the linear relationship shown in Equation (2.6) holds for all molecules within the IR beam, that is there are no intermolecular interactions between molecules, then for a mixture Beer's Law can be applied in a linear and additive form [23–25]. Thus for an n-component system, the absorbance A at wavenumber can be represented by the summation:

(2.8)

and expressed as:

(2.9) equation

In Equation (2.9), the system has been normalised by the inclusion of the pathlength l term, which in any one spectrum this term should remain constant over all wavenumbers. For a quantitative calibration and prediction set of mid-IR spectra then Equation (2.9) forms the basis of the multiple linear regression (MLR) and classical least squares (CLS) algorithm approaches, in which for the calibration model the sum-squared error over all the wavenumbers of all the spectra within the calibration data set is minimised. Although very similar, MLR requires for the calibration data set independently input reference values for the component concentrations, whereas CLS requires pure spectra of each of the components within the system for developing a calibration model. If we assume that each spectrum within a data set (whether calibration, validation or prediction) has been appropriately normalised, then Equation (2.9) can be modified to:

(2.10) equation

where within any given vibrational spectroscopy specific data set I can be used to represent: normalised IR (mid- or near-) absorbance (or absorbance equivalent) intensity, normalised THz absorbance intensity, or normalised Raman scattering intensity cross-section. While both MLR and CLS have been and are still used for calibration model development for many well-defined, well-understood and relatively simple IR and Raman spectroscopic quantitative component determinations, many data sets of interest to a wide range of applications, including those within forensic science, cannot be so easily fully described and defined in terms of their components. This and the increasing need to classify spectra as well as quantify components within complex spectra has led to the wide use and development of a range of multivariate data analysis techniques, many of which are mentioned or exemplified in the applications discussed within the chapters within this book, for instance inter alia these include:

ANN (artificial neural network): Brewster and Goodacre (Food);

GA (genetic algorithm): Brewster and Goodacre (Food);

HCA (hierarchical cluster analysis): Brewster and Goodacre (Food), Kwok and Taylor (Counterfeit tablets);

PCA (principal component analysis): Brewster and Goodacre (Food), Fredericks (Fibres), Madariaga (Dyes and pigments), Ozen and Tokatli (Food), Kwok and Taylor (Counterfeit tablets), Witowski and Carrabba (Counterfeit Pharmaceutical Labels);

PCR (principal component regression): Rösch et al. (Bioagents);

PLS/PLSR (partial least squares/partial least squares regression): Brewster and Goodacre (Food), Fredericks (Fibres), Matousek (Explosives); Ozen and Tokatli (Food), Rösch et al. (Bioagents);

PLS-DA/PLS-LDA (PLS linear discriminant analysis): Madariaga (Dyes and pigments);

SIMCA (soft independent modeling of class analogy): Ozen and Tokatli (Food), Taylor (Counterfeit tablets);

SLDA (stepwise linear discriminant analysis); Brewtser and Goodacre (Food).

It is not within the remit or purpose of this book to provide a description of the modus operandi of these and other chemometric/statistical analysis routines that have been used with vibrational spectroscopic data sets or to differentiate between the benefits and limitations of each, since these are well covered expertly in many articles and books specifically intended for that purpose; examples of these are provided in References [26–29]. We will, however, endeavour (in layman's terms without the use of matrix mathematics!) to provide a concise descriptive account of the HCA, PCA, PCR and PLS methods.

HCA is an unsupervised technique (i.e., it makes no a priori assumptions about the data set) that creates a hierarchy of clusters based on a between-object (spectrum) distance matrix, which (if agglomerative) is generated from such as a nearest-neighbour distance metric in multivariate space. The output from an HCA is usually in the form of a dendrogram, which ranks (clusters) spectra in a tree-like structure according to how similar/dissimilar they are; see, for example, that for the Raman spectroscopic analysis of Cialis tablets shown as Figure 9.1.8 of Chapter 9.1 authored by Kwok and Taylor, and that for the Raman spectroscopic analysis of Bacillus