Applications of Raman Spectroscopy in Forensic Science. I: Principles, Comparison to Infrared Spectroscopy, and Instrumentation.
TABLE OF CONTENTS INTRODUCTION 112 I. FEATURES OF RAMAN SPECTROSCOPY 113 II. THEORY 113 A. Rayleigh Scattering 114 B. Raman Scattering 115 C. Resonance Raman Scattering 118 D. Surface-Enhanced Raman Scattering (SERS) 120 E. Laser-Induced Fluorescence 121 III. COMPARISONS OF INFRARED AND RAMAN SPECTRAL DATA 121 A. Display Formats 121 B. Infrared Microscopy vs. Raman Microscopy 122 C. Organic Compounds 123 D. Inorganic Compounds 124 E. Composite Matrices: Paint 127 IV. INSTRUMENTATION 128 A. Source Lasers 128 B. Filters 128 C. Sampling Optical Configurations 129 D. Monochromators 130 E. Detectors 132 F. Raman Microscopes 132 ACKNOWLEDGMENTS 134 REFERENCES 134 ABOUT THE AUTHORS 135
The Raman effect, which involves inelastic scattering of light, was predicted theoretically by Smekal in 1923  and was observed experimentally by Raman and Krishnan in 1928 . It is an extremely weak process; Rayleigh (elastic) scattering, which occurs at the same time, is typically five to seven orders of magnitude greater in intensity. Early custom-built Raman spectrometers, which relied on mercury arc lamps as excitation sources, prisms as dispersive elements, and photographic films as detectors, required many hours of data collection. The first commercial Raman spectrometer was introduced in 1953 and it employed a photomultiplier tube instead of photographic film . This allowed for a more quantitative measurement of Raman-scattered light, but at the cost of the multiplexing feature of a photographic film, as data would now have to be collected by sequentially scanning through the desired spectral range. Diffraction gratings replaced prisms in the late 1950s, but a system of two or three gratings in tandem was required to remove stray light from the very intense Rayleigh-scattered peak.
One of the most significant developments in Raman instrument technology occurred in the 1960s when visible lasers replaced mercury arc lamps. Not only were highly monochromatic sources having greater power levels now feasible using coherent radiation, but use of a wider range of wavelengths to seek more favorable excitation conditions became available. The increased sensitivity afforded by the laser also led to the development of the first Raman microscope in 1973 . However, even with laser excitation and photomultiplier detectors, Raman spectra were still acquired by scanning through the entire spectral range. When diode array detectors began to be used in the 1980s, their multiplexing ability allowed simultaneous collection of scattered light over a wide spectral range, permitting shorter collection times.
Other significant instrument developments that occurred in the 1980s were the advent of efficient rejection filters to remove Rayleigh-scattered light, and the introduction of near-infrared lasers. Interferometry, as employed in a Fourier transform infrared (FT-IR) spectrometer [11,27], cannot be used to disperse (mathematically) Raman-scattered light if a Rayleigh line is present, as the contribution of the latter to an interferogram would completely overwhelm the very feeble contributions from Raman bands. Using these new filters, however, FT-Raman spectrometers became a viable means to use near-infrared lasers for excitation, and the first Fourier transform instrument was developed in 1985 .
Charge-coupled devices (CCD), introduced in 1991, were a further step in improved detection capabilities; their quantum efficiencies may approach 100% for certain wavelengths. The first FT-Raman instruments were standalone devices, but in 2002 Raman modules for use with FT-IR spectrometers were introduced . Improvements in Rayleigh-line rejection filters also occurred in terms of both their efficiencies and their abilities to gather scattered light with lower Raman shifts. Employing these specialty filters, data could now be collected to within a few wavenumbers of the Rayleigh line.
In view of these innovations and advances in instrumentation, associated improvements in software, and greatly expanded spectral libraries, Raman spectroscopy is now--in every sense of the word--a fully mature analytical technique on par with its counterpart, infrared spectroscopy. Following the introduction of dependable, inexpensive FT-IR spectrometers and associated accessories in the 1980s, infrared spectroscopy experienced a veritable quantum leap in use in the forensic science laboratory . This was particularly true for the analysis of various types of trace evidence. In contrast, forensic scientists have been slower to embrace Raman spectroscopy, owing in part to the stepwise nature and lengthy timelines of the various instrument developments. The widely held perception that many samples are not amenable to analysis because of fluorescence undoubtedly contributed to this. However, a more important factor may be insufficient understanding and appreciation of the unique capabilities of Raman spectroscopy by many forensic scientists, who may view this technique (incorrectly) as simply an indirect means to obtain an infrared spectrum.
Part I of this article is therefore devoted to addressing this latter issue, by presenting examples demonstrating how Raman spectral data can clarify, confirm, augment, or complement information provided by infrared spectroscopy, as well as by other methods. In particular, it illustrates how in certain cases, Raman spectroscopy can be the only viable means to obtain definitive structural information about some analytes present in very low concentrations. The reasons for these differences can be understood through the mechanism by which a Raman spectrum is produced. Some basic principles of Raman scattering are thus presented, along with a description of the instruments used to collect Raman data.
Based on the foundation established in the present work, analysis conditions intended to avoid or minimize fluorescence are described in Part II . A particularly important aspect of any forensic science examination of evidence--but especially those involving comparative analyses--is data interpretation. Interpretation of Raman spectral data is thus also addressed in Part II, along with a comprehensive review of the published literature describing applications of Raman spectroscopy for the analysis of a wide variety of evidentiary materials.
I. FEATURES OF RAMAN SPECTROSCOPY
There are many features of Raman spectroscopy that make it well suited for the analysis of evidence. This method requires little or no sample preparation and is nondestructive (with a few exceptions). Data for most samples can be obtained in minutes or seconds. Spectra can be obtained in situ for samples in transparent glass or plastic containers, and because water is a weak Raman-scatterer, spectra of analytes in aqueous solutions can be readily obtained. Difficult-to-access surfaces can be subject to analysis using fiber-optic probes, and with the recent introduction of hand-held portable Raman spectrometers , applications outside the laboratory are now possible, including use at crime scenes.
Using a Raman microscope, the maximum lateral (x-y direction) spatial resolution is roughly 60% of the wavelength of the laser used , as this spot size is ultimately limited by diffraction. The minimum spot size using a 513-nm (0.513-[micro]m) laser, for example, is roughly 0.3 [micro]m. Depth profiling is also possible using a confocal Raman microscope, although the z direction (depth) spatial resolution is usually lower and is dependent on the nature of the sample . For inhomogeneous materials, both lateral and depth maps can be constructed based on specific Raman-scattered peaks.
There are now a host of different excitation laser wavelengths available with various commercial Raman instruments, ranging from the ultraviolet to the near-infrared. Although it is unlikely that any one laboratory is fortunate enough to have access to all of these, having at least two lasers with wavelengths near the ends of the visible region (400-550 nm and either 780 or 785 nm), is highly desirable. It is also very useful to have access to both a dispersive spectrometer and an FT-Raman instrument. In particular, one would like to have as much flexibility as possible to try to avoid or minimize the archenemy of Raman spectroscopy: laser-induced fluorescence.
Before describing how the Raman effect arises, it is necessary to review briefly some basic principles of physics. Light and other electromagnetic radiation have both wave and particle aspects to their behavior. The wave aspect views light as composed of oscillating electric and magnetic fields, perpendicular to each other, which propagate through a vacuum with a velocity c, the speed of light. The particle aspect of light considers it to be composed of photons, and one may view a photon as the unit of light that is absorbed or emitted by an individual atom or molecule. The energy of a photon is given by E = h[upsilon], where h is Planck's constant and [upsilon] (the Greek letter nu) is the frequency of the light. The frequency of light is inversely proportional to its wavelength, [lambda], by [upsilon] = c/[lambda].
The dipole moment of a molecule, [mu], is a measure of the separation of charges in the molecule and it is a vector quantity (indicated by the bold type)--that is, it has both a direction and an intensity value. Some molecules, such as methane and carbon dioxide, do not have a permanent dipole moment whereas others, such as water, do. However, all atoms and molecules can be induced to have a dipole moment through the application of an electric field, as this serves to separate the charges in the atom or the molecule. An oscillating dipole creates electromagnetic waves, which have their greatest intensity propagating perpendicular to the direction of the dipole. There is no propagation of such waves along the axis of the dipole.
The ease with which a dipole moment is induced in an atom or molecule by an electric field is known as its polarizability, [alpha], which like dipole moment, is a measurable quantity and is an inherent property of an atom or a molecule. Unlike the dipole moment, however, it is not a vector quantity (and hence, [alpha] is not indicated in boldface). All atoms and molecules can be induced to have a dipole moment as noted, so they all have nonzero values for [alpha]. For all atoms and a few very symmetric molecules (such as methane), there is only one value of [alpha], but for all other molecules, [alpha] varies because it depends on the orientation of the electric field relative to the geometry of the molecule. For a given vibrational mode of a molecule, the dipole moment and polarizability may or may not change during the course of the vibrational motion--very important concepts that are discussed later.
A. Rayleigh Scattering
Rayleigh scattering occurs from particles that have dimensions much smaller than the wavelengths of light that are scattered. This criterion is easily met for atoms and most molecules interacting with visible or near-infrared radiation: atoms and most molecules have dimensions on the order of angstroms or tens of angstroms and the shortest wavelength of visible light is roughly 4000 angstroms. Rayleigh scattering may be conceptualized as resulting from a perturbation of the electron cloud of an atom or molecule by the oscillating electric field of light. The electrons of the cloud then oscillate in the same direction and with the same frequency as this field, and this oscillating dipole radiates light.
The induced dipole moment caused by the oscillating electric field is given the symbol [P.sub.ind] (distinct from [mu], which is the inherent dipole moment of a molecule), and its strength is proportional to both the polarizability of the atom or molecule, [alpha], and to the strength of the applied electric field, E, so [P.sub.ind] = [alpha] E (for molecules for which the polarizability varies, this is [alpha] for the direction of this field). The electric field of light varies sinusoidally with time, t, with a frequency, [[nu].sub.light], so E = [E.sub.o] cos (2[pi][[nu].sub.light]t) where [E.sub.o] is the maximum intensity of the electric field of light. Combining the two, the induced dipole moment [P.sub.ind] becomes
[P.sub.ind] = [alpha] E = [alpha] [E.sub.o] cos (2[pi][[nu].sub.light]t) (1)
This equation is simply a quantitative expression of what was described above, namely, that the oscillating electric field of light induces an oscillating dipole having the same direction and frequency as the light. This oscillation of the electron cloud then radiates and constitutes Rayleigh scattering.
What Equation (1) does not address is how Rayleigh scattering varies with the frequency of the incident light. The classical physics model for this frequency dependence assumes that the electrons of the atom or molecules are in an equilibrium-like state when they are not subject to an electric field. Because the electrons are in a bound state, they exhibit a resistance to an applied electric field rather than responding to it freely. For the small displacements that they undergo in response to an electric field, the electrons behave as if they were attached to a spring. As such, there is a "natural" frequency associated with the spring, which corresponds to the frequency of light needed to promote the electron to an excited electronic state. For most electrons, this corresponds to light in the ultraviolet region. The oscillating electric field of light drives the electrons, so the closer the frequency of light is to this natural frequency, the greater is the induced dipole moment.
The exact dependence of the intensity of Rayleigh scattering on frequency is [[nu].sup.4], that is, it is proportional the fourth power of the frequency. The frequency dependence of Raman scattering is also governed by this term. Light intensity can be expressed as either the number of photons per unit time (counts) or as power (watts). The latter measure includes the energy of the photon, which is proportional to its frequency (E = h[upsilon]). Two light beams comprising an equal number of photons of blue and red colors would register equal counts, but the former has a greater power. The [[nu].sup.4] expression refers to power and is commonly used in theoretical calculations. However, the detectors used in current Raman spectrometers measure counts and not power. Consequently, the intensity of scattered light as measured by a spectrometer varies as [[nu].sup.3], not [[nu].sup.4], but the latter will continue to be cited as the frequency dependence of scattering. The calculations that were made in this article to illustrate scattering-intensity differences based on frequency used [[nu].sup.3], as they represent values that would be observed using Raman instruments.
Probably the most obvious manifestation of the [[nu].sup.4] factor occurs for Rayleigh scattering of sunlight by the molecules (and a small amount of argon) of the atmosphere. Because blue light is scattering much more than red, the sky appears blue.
B. Raman Scattering
1. Classical Physics Description
For the classical physics description of Raman scattering, a diatomic molecule will be used as a model. It is assumed that this molecule is a harmonic oscillator, that is, the force between the two atoms is proportional to the displacement of the atoms from their equilibrium position . There is only one vibrational mode for a diatomic molecule and the course of the vibrational motion can be described by the interatomic distance. However, it is more convenient to denote it by a parameter, Q, where Q is the displacement from the equilibrium interatomic distance. Q is positive when the molecule stretches beyond its equilibrium interatomic distance and is negative when it is compressed. The maximum displacement is denoted as [Q.sub.0]. The vibrational motion as a function of time t can thus be described by
Q = [Q.sub.0] cos (2[pi][[nu].sub.vib]t) (2)
where [[nu].sub.vib] is the vibrational frequency of the molecule. As will be discussed in more detail later, the polarizability of a diatomic molecule (whether homonuclear or heteronuclear) always changes during a vibration. For small displacements from the equilibrium interatomic distance, the polarizability [alpha] as a function of Q can be approximated by
[alpha] = [[alpha].sub.0] + ([partial derivative][alpha]/[partial derivative]Q) Q (3)
where [[alpha].sub.0] is the polarizability of the molecule at its equilibrium position and the partial derivative ([partial derivative][alpha]/[partial derivative]Q) is the change in the polarizability as a function of the vibrational displacement (that is, how fast the polarizability is changing as the molecule stretches or contracts). Substituting Q of Equation (2) into Equation (3), the variation of the polarizability as the molecule vibrates becomes
[alpha] = [[alpha].sub.0] + ([partial derivative][alpha]/[partial derivative]Q) [Q.sub.0] cos (2[pi][[nu].sub.vib]t) (4)
When light is incident on an atom or a molecule, it will induce a dipole moment [P.sub.ind] = [alpha] E as indicated by Equation 1. For a diatomic molecule, [alpha] changes during the course of the vibration (Equation 4), so the equation for the induced dipole moment for a diatomic molecule as a function of time is
[P.sub.ind] = [alpha] E = [alpha] [E.sub.o] cos (2[pi][[nu].sub.light]t) = [[[alpha].sub.0] + ([partial derivative][alpha]/[partial derivative]Q) [Q.sub.0] cos (2[pi][[nu].sub.vib]t)] x [E.sub.o] cos (2[pi][[nu].sub.light]t) (5A)
Rearranging terms, this becomes
[P.sub.ind] = [[alpha].sub.0] [E.sub.o] cos (2[pi][[nu].sub.light]t) + ([partial derivative][alpha]/[partial derivative]Q) [Q.sub.0] cos (2[pi][[nu].sub.light]t) x [E.sub.o] cos (2[pi][[nu].sub.vib]t) (5B)
Using the trigonometric identity cos x cos y = 1/2 cos (x - y) + 1/2 cos (x + y), this may then be expressed as the sum of three parts
[P.sub.ind] = [[alpha].sub.0] [E.sub.o] cos (2[pi][[nu].sub.light]t) + 1/2 [Q.sub.0] [E.sub.o] ([partial derivative][alpha]/[partial derivative]Q) cos (2[pi] [[[nu].sub.light] - [[nu].sub.vib]]]t) + 1/2 [Q.sub.0] [E.sub.o] ([partial derivative][alpha]/[partial derivative]Q) cos (2[pi] [[[nu].sub.light] + [[nu].sub.vib]] t) (5C)
Each of the three terms represents a dipole oscillating at a different frequency. The first term is just the equation for Rayleigh scattering, as this dipole is oscillating at the same frequency as the incident light, [[nu].sub.light]. However, the two additional terms indicate that there are two other dipoles oscillating with frequencies equal to the sum and the difference of the frequencies of the incident light and the vibrational frequency [[nu].sub.vib]. These oscillations therefore produce inelastic scattering, and the second term that has a frequency less than that of the light ([[nu].sub.light] - [[nu].sub.vib]) produces Stokes Raman scattering. The dipole oscillating at a higher frequency ([[nu].sub.light]+[[nu].sub.vib]) gives anti-Stokes Raman scattering.
This model is qualitatively correct in three respects: (a) it predicts the occurrence of both types of Raman scattering; (b) it specifies that the polarizability of the molecule must change during the vibrational motion (that is, [partial derivative][alpha]/[partial derivative]Q must be nonzero) for inelastic scattering to occur; and (c) it indicates that the amount of inelastic scattering is dependent on the intensity of the incident light represented by [E.sub.o]. However, it also suggests that Stokes and anti-Stokes scattering should have equal intensities, which is incorrect.
This model also does not address the orientations of the induced oscillating dipoles arising from the vibration of the molecule. For Rayleigh scattering, the induced dipole is oscillating in the same direction as the polarization of the incident light but for Raman scattering this is rarely the case. Except for a few vibrations of highly symmetric molecules (such as methane or sulfur hexafluoride) that preserve its symmetry, the direction of induced dipoles for Raman active vibrations of all other molecules is in general not parallel to the polarization of the incident light. The specific direction depends on both the nature of the vibrational mode and the orientation of the molecule relative to the polarization of the incident light. Oscillating dipoles radiate more light perpendicular to the direction of oscillation, with decreasing amounts as one moves toward the axis, and there is no light radiated in both directions along the axis. There is thus a directionality to Raman scattering that can vary with each vibrational mode, but this is not usually apparent because the molecules in most samples are randomly oriented. However, it does become very apparent when highly ordered samples are excited with polarized light, as will be discussed in detail in Part II  of this article.
2. Quantum Mechanical Description
The classical physics models of Rayleigh and Raman scattering provide the best means to conceptualize these two events, as induced oscillating dipoles are not difficult to visualize physically. The quantum mechanical descriptions for these two scattering processes, while accurate both qualitatively and quantitatively, mostly lack this facility, as they are based on transition probabilities involving quantum mechanical wave functions and operators. The concept of Raman scattering as described by quantum mechanics is usually presented graphically using a diagram.
a. Scattering Energy Diagram
One very significant difference between the classical physics and quantum mechanical descriptions of atoms and molecules is the quantization of energies. The energies of the various vibrational modes of a molecule are approximately equally spaced , although for both infrared absorption and Raman scattering, only the first excited vibrational level is normally of importance. Figure 1 is a schematic depiction of Rayleigh and Raman scattering from a molecule that has three vibrational modes with frequencies of 3000, 1500, and 500 [cm.sup.-1]. The visible laser used for this illustration has a wavelength of 500 nm (green light having an energy of 20000 [cm.sup.-1]) and the laser photons incident on this molecule are represented by the upward pointing arrows. Because these photons are not absorbed, the dashed horizontal lines of this figure represent what are known as virtual energy states, that is, they are not actual discrete quantum mechanical energy levels of the molecule. Rayleigh scattering is represented by the central pair of arrows having the same lengths. The three pairs of arrows to the left of this represent Stokes Raman scattering, and the three to the right depict anti-Stokes Raman scattering. The energies of the scattered peaks are listed to the right of each downward pointing arrow, and these are what are actually measured by the spectrometer. The spectrum resulting from these seven scattering events is shown on the bottom portion of the figure, with each scattered peak depicted below its corresponding downward arrow. The abscissa of scattering spectra always displays Raman shifts (in wavenumbers) and not absolute frequencies.
To illustrate the differences between Stokes and anti-Stokes Raman spectral data, relative intensities for the three Stokes-scattered peaks of this hypothetical molecule were assigned the values shown. Based on these, corresponding intensities for the anti-Stokes peaks were calculated. Excluding the frequency dependence, the probability of a Stokes and anti-Stokes transition involving the same vibration mode is identical. Although the [[nu].sup.4] factor (which is also derived by quantum mechanics) favors anti-Stokes scattering, for the vast majority of cases, this is more than offset by the fact that at room temperature, most molecules are not in these excited vibrational states. The fraction of molecules in the ground state and in each of the three depicted states was calculated (using the Boltzmann factor), and they are listed; the total is not 100% because some of the molecules are also in overtone levels of the 500 [cm.sup.-1] mode, as well as combination levels such as 500 + 1500 [cm.sup.-1].
Because anti-Stokes Raman scattering only occurs for molecules in excited states, it is easy to understand why most of their peak intensities are so low. The exception to this occurs for the low-frequency vibrational modes, which do have appreciable populations in excited states at room temperature. As a result, Stokes and anti-Stokes Raman spectra for the lowest Raman shifts (wavenumbers below 200 [cm.sup.-1] or so) are mostly mirror images of each other. Because the "full" vibrational spectrum of an analyte is desired for most analytical work, however, Stokes rather than anti-Stokes scattering is used for most types of Raman spectroscopy.
b. Selection Rules
Depending on the symmetry of a molecule, not all its vibrational modes will necessarily give rise to Raman scattering. This is determined by selection rules and those vibrational modes that do produce such scattering are known as Raman active, whereas those that do not are referred to as Raman inactive. The two selection rules that govern Raman scattering (for both Stokes and anti-Stokes scattering) are: (a) for a given vibrational mode to be Raman active, there must be a change in the polarizability of the molecule during the course of the vibration; this was derived for a diatomic molecule using a classical physics approach, but it also arises from quantum mechanics for all molecular systems; and (b) the vibrational quantum number can only change by one; for Stokes scattering this involves a change from the ground state to the first excited vibrational state (+1 change), whereas for anti-Stokes, the reverse occurs (-1 change). Raman scattering events involving overtone and combination vibrational levels are forbidden for a harmonic oscillator, but because the actual forces between atoms in molecular systems are not strictly harmonic, very weak Raman scattering peaks involving these levels may occur.
c. Intensities of Raman-Scattered Peaks
For a vibrational mode to be infrared active, it must involve a change in dipole moment. In many cases, it is not difficult to determine if the dipole moment will change based on the nature of the vibrational motion and the structure of the molecule. Stretching vibrations of isolated polar bonds, for example, will always involve a dipole moment change. For symmetric molecules, such as carbon dioxide, the symmetric stretch of the two polar C=O bonds does not involve a net dipole moment change, but the antisymmetric C=O stretch does (Figure 2). Infrared absorption intensities also correlate strongly with the extent of dipole moment changes.
For Raman spectroscopy, a change in polarizability can also be predicted from the nature of the vibrational mode in certain cases (group theory applied to a molecule provides infrared and Raman activities for each of its vibrational modes, but this treatment is beyond the scope of this article). The units of polarizability are volume ([cm.sup.3]), and a change in polarizability can be viewed as a change in the volume of the molecular cloud of the molecule during a vibration. The stretch of a diatomic molecule, for example, will always involve such a change. The vibrational modes of carbon dioxide are depicted schematically in Figure 2 together with cylinders representing the molecular cloud of this molecule. The symmetric C=O stretch involves a volume change, as illustrated by the elongation of the cylinder, whereas the antisymmetric C=O stretch does not, as this vibration may be visualized as the cylinder moving back and forth without changing its volume (although the center of mass does not move during this motion). Likewise, the bending vibration (which occurs in two perpendicular planes each having the same frequency) also does not entail a volume change. The symmetric stretch is Raman active, but the antisymmetric stretch and bending vibration are not.
In addition to volume changes, polarizability changes also are determined by the extent to which the electrons of a molecule are bound to their nuclei, as well as the number of electrons in the system. The electrons that are more loosely bound, such as the [pi] electrons of double bonds and aromatic compounds, tend to be more polarizable, and polarizability also generally tends to increase with atomic number. Changes in polarizability may thus be greater for such systems when molecular volume changes occur.
Polarizability changes govern not only whether a vibrational mode is Raman active, but also correlate with the intensity of a Raman-scattered peak. The most intense peak in the Raman spectrum of benzene (near 1000 [cm.sup.-1] in Figure 3), for example, is the symmetric C-C stretch. All six carbon bonds are stretching in phase and the actual molecular motion consists of the entire ring expanding and contracting without changing its hexagonal shape. This mode is thus known as "ring breathing" and it involves a large volume change of this aromatic compound, which includes six [pi] electrons. The second-most intense peak is the symmetric C-H stretch (near 3100 [cm.sup.-1]), which involves a similar type of motion involving the six hydrogen atoms moving in phase. The most intense peaks in the Raman spectra of other aromatic compounds often include aromatic ring expansion vibrations.
For cis-3,4-dichlorocyclobutene (Figure 4), the most intense Raman peak is a C=C stretch near 1550 [cm.sup.-1]. Another intense scattering peak (near 700 [cm.sup.-1]) results from the symmetric C-Cl stretch, which not only involves a relatively large volume change, but also involves two halogen atoms.
For infrared spectroscopy, absorption coefficients provide a measure of the probability that a particular vibrational transition will occur, and they determine the strengths of absorption bands. For Raman spectroscopy the corresponding measure of the probability of a specific scattering event is called a scattering cross-section. Because of the [[nu].sup.4] dependence of scattering, the values of scattering cross-sections vary with frequency.
C. Resonance Raman Scattering
In certain cases where the wavelength of the excitation laser is at or near an absorption band maximum of an analyte, a very significant enhancement of Raman scattering intensities can occur. As most of the excitation wavelengths used for Raman spectroscopy are in the visible region, such analytes are mostly pigments and dyes--that is, colored compounds. The quantum mechanical transition probability that describes this process consists of a term in the numerator and a frequency-difference factor in the denominator. The latter is just the difference between the frequency of the laser and the resonant absorption frequency (which usually corresponds to the maximum of the absorption band, as these are quite broad for condensed phase samples) plus a constant term (which keeps the denominator from becoming zero when "resonance" occurs). Although the denominator might suggest that enhancement should occur for all absorbing compounds when the laser frequency matches the absorption maximum, this is not the case. A significant enhancement is only observed when the numerator is large, and it represents a transition from the ground electronic and vibrational states of the molecule to an excited electronic and vibrational state. The value of this term depends on how the geometry of the molecule changes when going from the ground to the excited electronic state, and the extent to which the vibrational wave functions of the ground and excited states overlap. This term is similar to one that describes the probability of an actual absorption of light, known as the Franck-Condon factor, although, in fact, it is still a scattering event that occurs and not one of absorption followed by emission. When resonance Raman scattering occurs, the closer the laser wavelength is to the maximum of the absorption band of the pigment or dye, the greater is the enhancement effect, reflective of the denominator term.
Because the Franck-Condon numerator term becomes large when there is significant overlap between vibrational wavefunctions, resonance enhancement only occurs for some of the vibrational modes of a molecule; these are vibrations that are associated with the chromophore of the pigment or dye. A good example of this occurs for hemoglobin and other hemoproteins that give resonance Raman spectra . These spectra consist of only certain vibrations of the heme chromophore (composed of a porphyrin ring) and not those of the proteins.
Another consequence of the overlap factor of the numerator term is that when different laser wavelengths occurring within the absorption band are used for excitation, variable Raman spectra may occur. This can be seen from the two spectra of a green phthalocyanine pigment (C.I. Pigment Green 7) obtained using laser excitations wavelengths of 514.5 nm (Figure 5 A) and 632.8 nm (Figure 5B). The phthalocyanine ring is similar to that of porphyrin and phthalocyanine pigments also produce resonance-enhanced Raman spectra . The visible absorption spectrum of C.I. Pigment Green 7 is depicted in Figure 5C and the positions of the two laser wavelengths used for excitation are indicated by arrows. Green pigments have visible absorption bands in both the violet and red spectral regions, representing transitions to two different excited electronic states. The low-frequency "tail" of the violet absorption might extend to 514.5 nm (Figure 5C), so it is possible that Figure 5A actually represents a composite of two resonance enhancement effects.
Significant absorption of the laser light occurs for molecules producing resonance Raman spectra, as the same Franck-Condon factor that produces scattering enhancement also results in strong absorptions. Prominent fluorescence may thus occur in such cases; normally, fluorescence is a much more efficient process than Raman scattering and at best, Raman scattered peaks might be manifested as very weak blips on a strong broad fluorescence background. With resonance enhancement, which can produce enhancement factors greater than [10.sup.6], Raman scattering peaks can become comparable or greater in intensity to that of the fluorescence background.
The true enhancement factor of resonance Raman scattering is never realized in actual practice, as there is always a strong opposing element that limits it. Because the analyte absorbs strongly, this results in a significantly reduced effective volume from which scattered light can be collected. Not only is the incident laser light strongly absorbed, but so are the Raman-scattered peaks (self-absorption). For colored matrices, this reduced volume results in a notable decrease in the ability to detect other components of the matrix that are not subject to resonance enhancement. This is the reason "pure" spectra of certain pigments or dyes producing enhancement may result when materials containing them are analyzed, even when there are other components present in much greater concentrations.
When there are two pigments used in a colored material and both produce resonance enhancement, it may be possible in certain cases to obtain essentially isolated spectra of each using two excitation wavelengths. This is illustrated by Raman data for a green spray paint obtained using laser excitation wavelengths of 514.5 and 632.8 nm (Figure 6), resulting in spectra of C.I. Pigment Yellow 74 (a monoarylide pigment) and the previously mentioned C.I. Pigment Green 7 (Figure 5B), respectively . Yellow pigments absorb in the low-wavelength region encompassing the colors violet, blue, and a portion of green, so the 514.5-nm laser line overlaps the absorption band of C.I. Pigment Yellow 74. As noted, the 632.8-nm line produces resonance enhancement for C.I. Pigment Green 7 (Figure 5).
D. Surface-Enhanced Raman Scattering (SERS)
Unlike conventional Raman spectroscopy, which is an in situ analysis, surface-enhanced Raman scattering or spectroscopy (SERS) requires some sample preparation, as analytes must be in intimate contact with an appropriate metal surface. This method was discovered by accident in 1974 when Fleischmann et al. in 1974 noticed a significant enhancement (~[10.sup.5]-[10.sup.6]) of the Raman-scattering intensity of pyridine adsorbed onto a silver electrode . Although this effect has been known for several decades, the exact mechanism of enhancement is still under active investigation. It is generally agreed, however, that laser-induced oscillations of the electrons of the metal substrate (known as plasmons) play a major role. Because laser radiation is coherent, these are concerted oscillations that can result in very intense electric fields being generated at the surface of the substrate. This effect is greatest when the frequency of the laser matches the resonant frequency of the substrate electrons, which depends on the metal and the sizes and morphologies of surface imperfections. Two other factors have been proposed as contributing to this effect, including a resonance involving a charge-transfer state created between the electrons of the adsorbed molecules and the conduction band of the metal substrate, and resonances occurring within the molecules themselves [17,18].
The most common metals used as substrates include small colloidal particles of silver and gold. Optimal enhancement effects depend on the appropriate choice of metal substrate, laser energy, and analyte, so the exact scattering intensity gain varies. In ideal cases, enhancement factors considerably greater than those possible with resonance Raman spectroscopy have been achieved.
Like resonance Raman spectroscopy, certain vibrational modes are affected more than others using SERS, and such spectra may exhibit considerable differences compared to conventional Raman data. This arises, in part, because analyte molecules adsorbed directly onto the metal surface may experience considerable interactions with this substrate. Not only can this affect some of the vibrational modes, but it can also change the symmetry of the molecule, which has a strong bearing on which vibrational modes can produce Raman scattering. As a result, Raman-scattering peaks that are not normally observed using conventional methods have been observed using SERS, while some normally observed are not.
As a first step toward validation of this method for the routine examination of evidence, Muehlethaler et al.  studied the repeatability and reproducibility of SERS applied to the analysis of three molecules of forensic science interest, crystal violet, methamphetamine, and 2,4,6-trinitrotoluene (TNT). The goal of this study was to provide data to support the use of this method for qualitative analyses, and the variability factors arising from measurement conditions, sampling, colloids aliquots, colloids batches, and different laboratories, were quantified. Within a given laboratory, the largest source of variation arose from measurement considerations and fluctuations in peak intensities due to Brownian motion of particles, solvent evaporation, and concentration. However, the largest variability (up to ~70% relative standard deviation) occurred between different laboratories and instruments. Despite this, the authors found that SERS is useful for identification of these three analytes, although it is currently not amenable for quantitative analyses.
Muehlethaler et al. have also reviewed applications of SERS in forensic science , and more information about the principles of SERS is presented by LeRu and Etchegoin . Schlucker provides a comprehensive discussion of SERS applications, particularly in the life sciences .
E. Laser-Induced Fluorescence
Fluorescence (also known as photoluminescence) occurs when a molecule absorbs enough laser energy to cause a transition to an excited electronic state, with the molecule then returning to the ground electronic state by emission of radiation. The vast majority of this fluorescent emission has less energy than the laser. Figure 7A depicts the first few energy levels of a particular vibrational mode for both the ground and excited electronic states of a molecule. The absorption of a laser photon will more than likely promote the molecule to an excited vibrational state, and the molecule very quickly decays to the ground vibrational level of this excited electronic state (in one or more steps). This occurs via a nonradiative transition and the lost energy is manifested as heat. From this level, the molecule then emits a photon when it returns to the ground electronic state, where it can end up in one of a number of possible vibrational states (Figure 7A). The transitions to various vibrational states are possible because there are no selection rules governing vibrational energy-level changes when they accompany electronic transitions. As a number of vibrational modes occur for most molecules, the vibrational energy levels for both the ground and excited electronic states constitute what is essentially a continuum (Figure 7B; in reality, the spacings between adjacent vibrational energy levels decrease with increasing energies). Fluorescence for most molecules is thus seen as a very broad band rather than as a series of discrete individual peaks. For some strongly absorbing analytes, nearly every photon that is absorbed is reemitted as fluorescence, so it a very efficient process compared to Raman scattering.
III. COMPARISONS OF INFRARED AND RAMAN SPECTRAL DATA
Most forensic scientists are probably more familiar with infrared spectroscopy than Raman spectroscopy, but as noted previously, the latter is becoming more widely used for the analysis of evidence. Depending on the type of evidence being examined, forensic scientists might elect to use Raman spectroscopy in lieu of infrared spectroscopy, as a method to confirm or to clarify infrared data, or as a means to seek information not obtained by infrared spectroscopy. The role that Raman spectroscopy is to assume in an overall analysis scheme is therefore dictated by the purpose of the examination, the nature of the evidence, and the degree to which the analyst wishes to characterize the evidence. Acomparison of the capabilities, merits, and limitations of these two related spectroscopic methods for the examination of different categories of analytes is thus discussed in this section.
A. Display Formats
Infrared absorption and Stokes Raman scattering both involve processes whereby molecules gain a single unit of quantized vibrational energy, and the abscissa of a Raman spectrum displays shifts rather than absolute frequencies. Consequently, infrared and Raman spectra may appear--superficially at least--to be very similar. However, even for molecules for which the same vibrational transitions are observed using both methods, they are far from mirror images (or if comparing Raman data to infrared spectra depicted in absorbance, they are far from superimposable). For some analytes, it might be very difficult to believe that the two spectra were produced by the same sample. The differences occur because the two processes are governed by separate selection rules, and there may be little or no correlation between the extents to which the dipole moment and the polarizability change for a particular vibrational mode.
The abscissa scale of infrared data is usually (but not always) shown with frequency, in wavenumbers, decreasing from left to right.A split display, which doubles the width of the abscissa scale below 2000 [cm.sup.-1], is often (but not always) used. There is even less uniformity in regard to Raman data, as displays with frequencies decreasing from left to right and right to left are both used frequently, in either split or linear scales. This, despite a 1997 recommendation by the International Union of Pure and Applied Chemistry (IUPAC) that Raman spectra be depicted with frequencies decreasing from left to right , in line with infrared spectral displays. This lack of uniformity is unfortunate, as it may hamper spectral comparisons and interpretations. Although characteristic Raman peaks can still be identified regardless of the format used, analysts may find it more difficult to develop pattern recognition skills for Raman spectra because of this lack of a convention in data presentation.
Transmission infrared data may be presented in either the percent transmittance (logarithmic) format, or in absorbance, which is a linear scale. The former is usually used for analytical work because it permits the weaker absorptions of the sample to be more readily observed, providing a better characterization of the sample. Raman spectra depict scattered peak intensities using a linear scale and, as noted, the intensity represents photon counts (counts per second or total counts) and not power. When the most intense Raman peak is depicted full scale, the very weak features, such as overtone and combination bands, will usually not be observed.
B. Infrared Microscopy vs. Raman Microscopy
1. Features and Capabilities
For both of these microsampling methods, the minimum analysis areas are limited by diffraction, but because longer wavelengths are involved with infrared spectroscopy, the maximum spatial resolution for this method is on the order of 20 [micro]m (the wavelength corresponding to 500 wavenumbers). As noted for Raman microscopy, significantly greater spatial resolutions are possible because focused beams of coherent visible or near-infrared light are used for excitation. For the analysis of successive layers of a cross-sectioned laminate, multilayered paint, or other inhomogeneous material, spectra free from contributions of adjacent layers or areas can thus be more readily obtained using Raman microscopy.
For transmission analyses using infrared microscopy, sample preparation is usually required to obtain thin slices, whereas examinations using an attenuated total reflectance (ATR) objective  are in situ, similar to a Raman analysis. One significant difference between the capabilities of the two methods is the ability to perform depth profiling using confocal Raman microscopes.
For certain types of evidence, the wider spectral range afforded by Raman microscopy can also be helpful. Infrared microscopes typically employ narrow-band mercury cadmium telluride (MCT) detectors , which have low-frequency cutoff regions near 700 [cm.sup.-1]. In contrast, typical Rayleigh-line filters used for Raman instruments allow Raman shift data to be collected to between 200 and 50 [cm.sup.-1].
2. Example: Textile Fibers
Infrared and Raman microscopes are the accessories of choice for the identification and comparison of fibers, considering their dimensions and morphologies. Fibers thus serve as a useful matrix for comparing these two techniques.
To obtain optimum infrared transmission data, fibers should be flattened to produce a uniform thickness, not only because some are quite thick, but also because their circular or irregular cross-sections would otherwise act as a lens to defocus the analysis beam. However, the pressure applied during flattening can alter the crystallinity of the fiber polymer, which may affect the resulting infrared absorptions. Spectra free of pressure-induced effects can thus be more readily obtained using Raman microscopy, and the original morphology of the fiber is unaltered by this analysis.
For bicomponent fibers, Raman microscopy also has advantages. The two main configurations used for such fibers are side-by-side and sheath/core, and for the former, the much smaller spot size of Raman microscopy permits spectra of each component to be obtained with less likelihood of contributions from the adjacent component. For the sheath/core configuration, either depth profiling through the fiber surface or analysis of the sheath and core of a cross-sectioned surface can be used to obtain data for each component. In contrast, spectra free of absorptions of the second component are more difficult to obtain in situ using infrared microscopy, particularly for the sheath/core configuration.
Raman microscope data for a bicomponent fiber comprising a polyethylene sheath and a polyester core are depicted in Figure 8 (red spectra). Confocal microscopy was used for this analysis with the excitation beam directed perpendicular to the fiber length . The beam was focused on the surface of the fiber to obtain the spectrum of the sheath (top spectrum), then data were collected from a spot closer to the center of fiber for the core analysis (bottom spectrum). There are only minor contributions of Raman peaks of the polyester core observed in the spectrum of the sheath (most evident for the two strongest peaks of polyester). One possible method to avoid even this small contribution would be to collect spectra from the surface of a cross-section of the fiber.
Another advantage of Raman microscopy is illustrated by the presence of a weak peak--below 200 [cm.sup.-1] in the Raman spectrum of the polyester core (Figure 8, bottom in red). This peak is from anatase (Figure 9D). Small amounts of titanium dioxide (anatase or rutile) are used in some fibers as delustering agents, but the far-infrared absorptions of anatase (Figure 9A) or rutile (Figure 9B) cannot be observed in fiber spectra using infrared microscopy.
A comparison of infrared (transmission) and Raman microscope data for two very similar types of fibers (nylon 6 and nylon 6,6) are shown in Figure 10. Both infrared and Raman data allow the differentiation between the two nylons, although this distinction may be more apparent for the Raman spectra (including the differences occurring in the light-blue highlighted regions). However, for fiber examiners used to interpreting infrared spectra, the prominent absorptions of Amide I and Amide II are probably more useful features for recognition of the type of fiber, as they are not as conspicuous in the Raman spectra. For discriminating between similar fibers (and other materials), there is thus merit in using both methods.
C. Organic Compounds
1. Molecules Having Identical Infrared and Raman Activities
For molecules that lack symmetry or have few elements of symmetry, all of the vibrational modes are both infrared and Raman active. An example of this is cis-3,4-dichlorocyclobutene, which has a plane of symmetry, and all 24 of its vibrational modes are both infrared and Raman active. Low temperature (~15[degrees] Kelvin) data of polycrystalline films were used for this illustration , as the spectral lines for both methods are quite narrow, facilitating a comparison (Figure 4).
Even from a cursory examination of these two spectra, it is evident that they are quite different, and in many cases, the relative peak intensities exhibit opposite tendencies for the two methods: vibrations with strong infrared absorptions often produce weak Raman peaks, and vice versa. This occurs because symmetric vibrations (including localized vibrations that are symmetric even if the molecule as a whole does not have any symmetry elements) often involve little or no dipole moment changes. In contrast, they may involve significant molecular cloud volume changes. An example of this is the C=C stretch (near 1550 [cm.sup.-1] in Figure 4), which produces only a small dipole moment change, and hence a weak infrared absorption. As noted previously, this vibration produces the most intense peak in the Raman spectrum. An exception to this general intensity trend occurs for the symmetric C-Cl stretch. This vibration (near 600 [cm.sup.-1] in Figure 4), produces both a strong dipole moment change and a significant volume change of the molecular cloud, hence both infrared and Raman peaks are prominent.
Because infrared absorptions are depicted using a logarithmic scale, numerous very weak overtone/combination absorptions may be seen in Figure 4, but these are not evident in the Raman spectrum as depicted. They do occur, but an ordinate expansion would be required to detect them (assuming they are not lost in the noise).
The three Raman peaks seen between 200 and 150 [cm.sup.-1] illustrate the advantage of observing vibrations closer to the Rayleigh line, as these low-frequency C-Cl bending vibrations occur below the region accessible to even extended-range infrared spectrometers. The peaks that are observed below 100 [cm.sup.-1] are not vibrations of the molecule, but rather, crystal lattice vibrations (known as phonons) where the vibrating unit is the entire molecule. They serve to illustrate the type of Raman data that can be obtained for crystalline substances (mostly organic compounds) using filters that allow data to be collected to within a few wavenumbers of the Rayleigh line.
2. Mutual Exclusion
Molecules having high degrees of symmetry will usually have some vibrations that are either infrared inactive, Raman inactive, or both, and they often have quite simple spectra. In addition, infrared and Raman spectra of such molecules will not usually consist of the same transitions, as was the case for cis-3,4-dichlorocyclobutene. For molecules that have a center of symmetry, in fact, there is mutual exclusion: vibrations producing infrared absorptions will not give Raman scattering peaks, and vice versa (and there may also be vibrational modes that are neither infrared nor Raman active). An example of a molecule that has a center of symmetry is carbon dioxide (Figure 2), which was discussed previously. The symmetric stretch is Raman active but does not produce a dipole moment change, whereas the antisymmetric stretch and the bending mode are both Raman inactive but do produce dipole moment changes.
A second example of a molecule with a center of symmetry is benzene (Figure 3), and as noted, the two most intense Raman peaks involve symmetric expansions of the entire ring geometry, but these do not involve dipole moment changes. Likewise, the other vibrations producing Raman peaks are infrared inactive, and none of the infrared absorptions has a counterpart in the Raman spectrum. Unlike carbon dioxide, however, benzene has several vibrational modes that are neither infrared nor Raman active.
D. Inorganic Compounds
Raman scattering peaks are generally narrower than infrared absorption bands and this difference is usually much more pronounced for inorganic compounds and salts, as their infrared spectra often include broad or very broad absorptions. The narrow breadths of Raman peaks are a useful feature for the differentiation and identification of some inorganic compounds and salts, and they are especially helpful when interpreting spectra of composites and mixtures containing such components. In addition, characteristic vibrations of some inorganic compounds and salts occur in the lower-frequency regions . For infrared spectroscopy, an extended-range spectrometer, which allows data to be collected to 250 [cm.sup.-1], is required to observe absorptions in this region, which is inaccessible using an infrared microscope. As illustrated by the Raman spectrum of the core of the bicomponent fiber containing anatase (Figure 8, bottom), however, low-frequency vibrations can be quite important for identification of some inorganic constituents.
The differences in the breadths of absorption and scattering peaks for some inorganic compounds and salts can be seen by comparing infrared and Raman data for anatase (Figures 9A and 9C), rutile (Figures 9B and 9D), three nitrate salts (Figure 11), Chrome Yellow (an inorganic pigment composed primarily of lead chromate, Figures 12A and 12D), bismuth vanadate (Figures 13A and 13E), and diamond (Figures 14A and 14B).
Infrared (Figure 13B) and Raman (Figure 13C) spectra of an automotive basecoat  that contains bismuth vanadate and rutile illustrate the advantage of having narrower Raman peaks. The main bismuth vanadate infrared absorption (Figure 13A) overlaps the high-frequency portion of the very broad rutile band (see Figure 13B, which is the infrared spectrum of a white automotive paint that contains only rutile). When both pigments are present, the bismuth vanadate absorption is manifested as what appears to be a broadened rutile feature (Figure 13B) rather than a discrete peak. Consequently, the presence of bismuth vanadate can be overlooked based solely on infrared data. However, the Raman peaks of both bismuth vanadate (Figure 13E) and rutile (Figure 13D) are seen clearly as distinct features in the spectrum of the basecoat (Figure 13C).
Infrared spectra of a series of inorganic salts involving a covalently bonded cation (e.g., N[H.sub.4.sup.+]) or anion (e.g., N[O.sub.3.sup.-]) combined with various monoatomic counter ions are often very similar . This occurs because the absorptions that are observed arise solely from vibrational modes of the covalently bonded ion. However, the monoatomic counter ion influences the crystal structure of the salt, which in turn can affect the vibrations of the covalent ion. Often, the greatest similarities in infrared spectra for such series occur when adjacent members of a particular chemical group (e.g., sodium and potassium) are the counter ions, or when the counter ions have the same charges and sizes. Infrared spectra of sodium nitrate and potassium nitrate (Figure 11), for example, are more similar to each other than they are to that of barium nitrate.
The Raman spectra of these three salts are also similar, but they include peaks that are more reflective of the properties of the counter ions. Lattice vibrations (phonons) involving motions of the cations and anions of salts occur at low frequencies, and if Raman active, they may be observed in some spectra; the Raman peaks observed below 200 [cm.sup.-1] in Figure 11 are lattice vibrations. Because they reflect motions of the monoatomic cations, the differences between the frequencies of these lattice vibrations (in percent) are often greater than those of the other Raman features.
The infrared absorptions of barium nitrate are readily distinguished from those of the other two nitrates, whereas the Raman spectra of the three are more similar qualitatively. As noted for the nylon fibers, there is thus merit in using both infrared and Raman methods in tandem when a differentiation is sought between some closely related analytes.
A crystal of diamond is, in effect, a single (very hard) molecule, as each carbon atom is covalently bonded to four others (Figure 14F). Because of the very high symmetry of diamond, it has only one Raman active mode, a lattice vibration (phonon) involving carbon-stretching motions. The very broad infrared absorptions of diamond (Figure 14A) are overtone/combination vibrations of such motions and they are actually very weak bands, as they were taken of a diamond window 1 mm (1000 [micro]m) thick. The thickness of a typical infrared sample is 1 to 10 [micro]m, so the diamond window is between 100 and 1000 times as thick as a normal sample.
Silicon and germanium have lattice structures similar to that of diamond, and their Raman spectra are depicted in Figures 14c and 14d, respectively, as a comparison. Consistent with the principles of vibrating systems , the heavier atoms of silicon and germanium have corresponding lower vibrational frequencies.
Raman spectra of a series of carbonates (Figure 15) also illustrate the advantages of the narrow breadths of Raman scattering peaks for distinguishing between salts of the same anion. In this case, a distinction between two polymorphic forms of calcium carbonate (CaC[O.sub.3]), calcite and aragonite, is also demonstrated.
Three examples of inorganic compounds for which Raman spectroscopy is clearly the preferred choice between the two vibrational methods are illustrated in Figure 16. Because of their high symmetries, elemental sulfur ([S.sub.8]) and red phosphorus are both weak infrared absorbers (Figures 16A and 16C). Quite thick samples are required to produce significant infrared peaks, resulting in spectra having pronounced sloping baselines. However, the Raman peaks of both (Figures 16B and 16D) are readily observed and sulfur is a very strong Raman scatterer.
Although it has only a single vibration, elemental iodine ([I.sub.2]) has a center of symmetry so it is a trivial example of mutual exclusion. The stretching vibration does not produce a dipole moment change and it is infrared inactive (Figure 16E). As noted, all diatomic molecules exhibit a polarizability change as they vibrate because this results in an increase in the volume of the molecular cloud, and iodine produces a low frequency Raman peak (Figure 16F).
E. Composite Matrices: Paint
Paint is one of the more complex materials routinely encountered as trace evidence because it may consist of a composite of additives, binders, light-absorbing pigments, and pigments used to produce opacity or other optical effects. With the exception of binders, these can include both inorganic and organic compounds having a very wide range of concentrations. Because of differences in infrared absorption and Raman scattering properties, infrared and Raman spectra of paints (and some other composite materials) may each provide distinct information about the constituents.
Inorganic pigments are usually stronger Raman scatterers than polymeric organic binders, owing in part to the greater number of electrons of their constituent atoms compared to organic compounds. Consequently, peaks of inorganic pigments may dominate Raman spectra of paints containing them. This is illustrated by infrared (Figure 12B) and Raman (Figure 12C) data for a yellow nonmetallic automotive finish that contains Chrome Yellow. Prominent absorptions of both an acrylic melamine binder (which also contains styrene) and Chrome Yellow (Figure 12A) are observed in the infrared spectra, whereas the Raman spectrum is essentially that of the pigment alone. Only a very weak Raman peak at 1003 [cm.sup.-1] from styrene is observed in this spectrum.
The infrared spectrum of a red nonmetallic automotive finish that contains Molybdate Orange, an inorganic pigment related to Chrome Yellow, is shown in Figure 17A (the Molybdate Orange absorption is labeled with its frequency). The Raman spectrum of this finish is shown in Figure 17B , and although this paint contains a relatively small amount of Molybdate Orange, the Raman spectrum also consists primarily of pigment peaks (Figure 17C). However, these peaks appear to be weak in that they are superimposed on a strong fluorescence background.
Spectra of the yellow nonmetallic basecoat that contains bismuth vanadate and rutile (discussed previously, Figure 13) also serve to illustrate the differences between infrared and Raman data, as inorganic pigment peaks likewise dominate the Raman spectrum. However, in this case somewhat stronger binder features are observed because the Raman data were obtained of an intact finish system that included the clearcoat. Clearcoat layers are typically much thicker than basecoats and the excitation beam passed through the clearcoat of this finish system prior to reaching the basecoat (see Figure 13F).
Many organic pigments may also be stronger Raman scatterers than polymeric organic binders because their chemical structures usually include fused aromatic rings and conjugated unsaturated bonds, and as noted, such moieties are easily-polarized. Automotive finishes are characterized by their vivid colors, and high concentrations of organic pigments may be used in some of them. Consequently, Raman spectra of such finishes may also be dominated by pigment features. For organic pigments that produce resonance-enhanced spectra, this is certainly the case, even when very low concentrations of such pigments are used. In such cases, pigment absorptions might not be observed in infrared spectra and the only definitive indication of the presence of these pigments is from Raman data.
In addition to paint, other evidentiary materials that might be colored, such as fibers, inks, plastics, paper currency, and cosmetics, could also contain colorants that produce resonance-enhanced Raman spectra. Infrared and Raman spectra for these are also likely to produce distinct information, as described in Part II .
All Raman instruments comprise five major optical components: an excitation source, illumination optics, collection optics, a wavelength dispersing/measuring unit (monochromator), and a detector. Spectrometers are classified as either dispersive or FT-Raman based on the mechanism used to separate or measure the various wavelengths of scattered light. As their names imply, dispersive instruments use a grating to physically disperse this band into its component wavelengths, whereas an FT-Raman instrument uses interferometry and a Fourier transform mathematical process to measure each wavelength. Both types are multiplexing instruments--that is, they simultaneously measure the intensities of the separated or sorted wavelengths without having to measure each individually, but they do so by different means.
A. Source Lasers
The requirements of an ideal Raman spectrometer light source would include a high degree of monochromaticity (that is, a very narrow bandwidth), a stable output with regard to both wavelength and intensity, and a highly collimated beam. Depending on the type of instrument and intended applications, a relatively high power level (up to several hundred milliwatts) may also be required. Because the shift in energy of Raman-scattered light for a particular vibrational transition is always independent of the excitation wavelength, a source having a broad bandwidth would produce broad scattered peaks. For the same reason, a source that produces an output with a wavelength that shifts during an analysis would result in either broad or multiple scattered peaks.
These ideal requirements are met by some, but not all, commercial lasers. A truly ideal source would also be inexpensive, but most lasers used for Raman spectroscopy do not meet this criterion. Analysts might wonder why laser pointers, which sell for a few dollars and which have outputs of a few milliwatts, cannot be used for some of their Raman microscopy applications that require only a few milliwatts of power. However, the output of these inexpensive lasers have either broad bandwidths, multiple output lines, or laser wavelengths that can shift (known as mode hopping) with environmental influences, so they are not used for Raman spectroscopy.
Lasers are classified as gas, ion, or solid-state depending on the physical state of the substance emitting coherent radiation. The main gas laser used for Raman spectroscopy is the helium neon laser, which emits a red 632.8-nm line (this laser is also used in the interferometer of FT-Raman instruments as a calibration device). The most common ion or plasma laser used for Raman spectroscopy is the argon ion laser, which emits blue (457.9 and 488 nm) or green (514.5 nm) radiation. Most lasers used as excitation sources are solid-state lasers. The wavelengths of some of the visible and near-infrared lasers used on Raman instruments include 405 (solid-state), 457 (solid-state), 457.9 (argon ion), 473 (solid-state), 488 (argon ion), 514.5 (argon ion), 532 (Nd:YAG solid-state, frequency doubled), 594 (solid-state), 632.8 (helium-neon gas), 638 (solid-state), 660 (solid-state), 685 (solid-state), 780 (solid-state), 785 (solid-state), 830 (solid-state), 980 (solid-state), and 1064 nm (Nd:YAG solid-state).
Some of the lasers used as sources produce additional lasing lines apart from the one intended for excitation. They are still used because the extraneous emissions are much weaker than the primary laser line and they can be removed easily. This is done using filters known as bandpass filters because they transmit only a narrow band of wavelengths, as illustrated by Figure 18C.
A second type of filter, which is used in all Raman instruments, is the Rayleigh-line rejection filter. Because Rayleigh-scattered light may be six or seven orders of magnitude greater than the signals of interest, early Raman instruments relied on two or three grating systems used in tandem to remove stray Rayleigh light. However, with the advent of efficient rejection filters, only a single grating is usually now used. Rejection filters are classified as either edge filters or notch filters, and the basis of this terminology can be seen from the transmission curves of the two types depicted in Figures 18A and 18B, respectively. Either type can be used for conventional Raman spectroscopy, but notch filters are required for those specialized applications where both Stokes and anti-Stokes scattered light are to be analyzed.
As noted, FT-Raman instruments must use a Rayleigh-line rejection filter in order to operate, and the transmission of an FT-Raman filter used in the 1990s is depicted in Figure 19B. Newer generations of rejection filters (Figures 18A and 18B) are more efficient in terms of the amounts of light that they transmit and remove. The oscillations observed in the transmission curves of Figures 18 and 19 are fringes caused by the laminated structures of these filters, which function through interference of light waves.
Filtering functions in a Raman spectrometer may also be performed by use of dichroic beamsplitters. Beamsplitters are used in Raman microscopes to allow coincident paths for two beams or rays traveling in opposite directions. They serve to introduce and remove a desired beam or ray from such a path, and by using a dichroic beamsplitter, they can do so in a more selective manner. As their name implies, dichroic beamsplitters allow a certain range of wavelengths to be transmitted, but allow a different range to be reflected.
C. Sampling Optical Configurations
Most Raman instruments are configured to collect 180[degrees] backscattered light, with the sample situated either vertically or horizontally. The latter is more convenient for most samples, particularly powders, as they can be analyzed by simply placing them on a flat surface or stage instead of in a cell. Schematics for the illumination and collection optics used for horizontal and vertical samples are shown in Figures 20 and 21, respectively. The former is a schematic of a Raman microscope, but it serves to illustrate horizontal sampling where the excitation beam and the backscattered light are coincident for a portion of the optical train. The same lens is used to focus the laser beam and to collect and collimate scattered light, which emanates from a cone having an apex located at the focal point of the lens. A beamsplitter is used to allow coincident paths for the illumination beam and the backscattered light. The diagonal plate of Figure 20 labeled "Injection/Rejection Filter" represents this device, but it is a dichroic beamsplitter. It reflects the laser beam so that it can be focused onto the sample, but it also prevents this same wavelength of light (from Rayleigh scattering) from passing through it and reaching the spectrograph. This dichroic beamsplitter is thus functioning as a mirror for the laser wavelength, but as a window for the longer wavelengths of Raman-scattered light.
For the diagram of a FT-Raman instrument (Figure 21), this filtering function is performed by the 1064-nm Rayleigh-line rejection edge filter. A 1064-nm bandpass filter to remove extraneous lasing lines is also depicted.
For samples that are mounted vertically, a similar configuration using coincident paths for the illumination and collection beams may be employed. Alternatively, vertical sampling may occur as illustrated in Figure 21, where a lens is used to focus the illumination beam and a parabolic mirror with a small hole to allow passage of this beam is used to collect and collimate scattered light.
1. Czerny-Turner Spectrograph
A common monochromator used for dispersive Raman instruments is the Czerny-Turner spectrograph (strictly speaking, spectrographs include a detection system, whereas monochromators refer only to the dispersing unit). A two-dimensional schematic of a Czerny-Turner spectrograph is depicted in Figure 20, and Figure 22 provides a three-dimensional perspective of this device. The diffraction gratings used for Raman spectroscopy are holographic gratings, which derive their name from their method of production. These gratings produce significantly less stray light than conventional grooved gratings--a very important consideration in view of the intensity of the Rayleigh line.
The spectral resolution provided by the Czerny-Turner spectrograph depends on the spacing between the grooves of the grating, the distance between the focusing mirror and the detector, and the size of the pixels on the detector array. The latter two are constants for a given spectrometer, but many commercial Raman instruments allow users to choose from different gratings. They are defined by the number of grooves per millimeter (g/mm) and common ones used for Raman spectroscopy include 600, 1200, and 1800 g/mm. Spectral resolution increases with the number of grooves. When multiple gratings are available, they may be mounted on a turret (Figure 22). The turret is rotated to change gratings, or when a different spectral range is to be analyzed using the same grating.
Spectral resolution will also change if different excitation wavelengths are used with the same grating. Diffraction gratings separate light based on interference effects, which are determined by wavelengths and not frequencies. A grating produces a swath of light consisting of a roughly constant range of wavelengths, but Raman shifts cover a range of wavenumbers and not wavelengths. Consequently, the same swath may cover only portions of a desired Raman shift range, depending on the wavelength of the excitation laser. This is illustrated by Figure 23, which depicts Raman data collected for a sample using the same grating, but with excitation wavelengths of 473, 632.8, and 785 nm. The linear CCD array detector of this illustration spans approximately 110 nm, which for the 473-nm excitation corresponds to a Raman shift range of 0-3988 [cm.sup.-1]. Assuming 1024 linear pixels for this CCD array, this gives an average spectral dispersion of 3.9 [cm.sup.-1] per pixel. However, for the 785-nm excitation, this span of 110 nm covers only a portion of the 0-3100 [cm.sup.-1] Raman shift region, and for complete coverage, three separate acquisitions are required. With no overlap, the three would encompass Raman shift ranges of 0-1566 [cm.sup.-1], 1566-2789 [cm.sup.-1], and 2789-3770 [cm.sup.-1], and the average spectral dispersions for these would be 1.5, 1.2, and 0.96 [cm.sup.-1] per pixel, respectively.
2. Michelson Interferometer
Interferometers do not use dispersing elements and sorting of scattered light on a FT-Raman instrument is performed mathematically. For Raman spectroscopy modules used on FT-IR instruments, the same optics and software are used for both analyses. This section describes very briefly how a Michelson interferometer performs this function, but for the full details, the reader is referred elsewhere [11,27].
The main components of the Michelson interferometer consist of a source, a beamsplitter, a stationary mirror, a moving mirror, and a detector (Figure 21). A Michelson interferometer generates an interferogram, a function that contains all of the information about the various wavelengths of light that are analyzed. This information is retrieved from the interferogram using a Fourier transform algorithm.
During operation, a collimated beam of scattered light is directed to the beamsplitter, and half of this light travels to the moving mirror while the other half is sent to a stationary mirror. Both beams return to the beamsplitter where they recombine and travel to the detector. When the distance between the stationary mirror and the beamsplitter and the moving mirror and the beamsplitter is the same, there will be constructive interference between all the different wavelengths composing the scattered light. A maximum amount of light reaches the detector at this point. When the moving mirror is at some other distance, however, some waves will interfere constructively, others destructively, while most will be in an intermediate state, and less light will reach the detector. This will change as the moving mirror travels back and forth, and the intensity of light that reaches the detector as a function of the position of the moving mirror produces the interferogram.
Each wavelength of scattered light will be coded as a cosine wave in the interferogram, and this wave will have an intensity proportional to that of the original light wave. The interferogram is the sum of each of these cosine waves, and a cosine Fourier transform is performed on the interferogram. This mathematical process can be viewed as a cosine wave sorter, as it extracts each cosine wave from the interferogram and plots its intensity versus its wavelength; this, however, is simply the spectrum of light that entered the interferometer. Light from a helium neon laser, which travels through the interferometer at the same time, produces its own interferogram, which is single cosine wave. Because the helium neon wavelength is known precisely, it serves to correlate the cosine waves of both interferograms to the wavelengths of the scattered light that entered the interferometer.
Interferometry has several advantages over use of a grating. For dispersive instruments, either an internal or external standard (often a silicon wafer; see Figure 14C) is used for calibration. For a FT-Raman instrument, a calibration is performed, in effect, with each scan so there is no need to run a separate standard. To change the spectral resolution on a dispersive instrument, a grating change must be made. Spectral resolution on an FT-Raman instrument is determined by the length of travel of the moving mirror, which is set by a software command. An FT-Raman spectrum also has a constant resolution throughout the entire spectral region. In addition, an interferometer permits a greater energy throughput, as both the entrance slit and the grating of the Czerny-Turner spectrograph (Figure 22) may produce some attenuation of incident light.
1. Charge-Coupled Device (CCD) Array Detectors
The multiplexing ability of an FT-Raman instrument results from the Fourier transform process, as all of the wavelengths of scattered light are sorted and measured by this one mathematical calculation. In a dispersive instrument, multiplexing is performed by a linear CCD array (Figure 22), as each pixel of this device functions as an individual detector.
A CCD array detector consists of a silicon wafer that is divided into thousands of small rectangles (pixels), each of which operates somewhat like a miniature capacitor (an electrical storage device consisting of two parallel metal plates holding opposite charges). The positive plate of the capacitor attracts any electrons that are in the conduction band of the silicon semiconductor. Normally, there are few thermal electrons in this band because the detector is cooled to below room temperature. However, when a photon having sufficient energy to promote an electron from the silicon valence band to the conduction band enters the detector, an electron-hole pair can be generated. This electron is captured and stored by the positive plate of the capacitor. Once the storage capacity of the first pixel is reached, data collection ceases and the charges are transferred through a chain of pixels. They are then measured individually as voltages, which are proportional to the number of photons that were incident upon each pixel. Typical linear CCD arrays used for Raman spectroscopy consist of either 1024 x 256 pixels or 2048 x 512 pixels.
The ability of a photodetector to convert light energy into an electrical signal is measured by its quantum efficiency, which may be defined as the percent of photons incident on a detector that produces a measurable electrical signal. The quantum efficiency of a typical CCD array detector used for Raman spectroscopy is shown in Figure 19A. The minimum photon energy required to promote a silicon valence electron to the conduction band corresponds to a wavelength of 1100 nm, while the maximum may vary between 400 and 200 nm. As expected, these are also the approximate cutoff regions for most CCD detectors (Figure 19A).
2. Indium Gallium Arsenide (InGaAs) Photodetector
An InGaAs detector is composed of an alloy of indium arsenide (InAs) and gallium arsenide (GaAs) and it is a semiconductor at room temperature. The mechanism for converting light energy into an electrical signal is similar to that of a CCD. An incident photon having sufficient energy will create an electron-hole pair and the charge produced, which is proportional to the number of photons, is measured. However, the energy difference between the valence band and the conduction band (the bandgap energy) of InGaAs is less than that of silicon. Consequently, lower-energy photons can be detected compared to the CCD detector, and the long wavelength cutoff region of the InGaAs detector is approximately 1700 nm (Figure 19A). Germanium detectors cover a spectral range similar to that of InGaAs and are somewhat more sensitive, but they require cooling with liquid nitrogen.
As illustrated by the Raman shift ranges for the various excitation wavelengths depicted on Figure 19A, an InGaAs detector must be used instead of a CCD array if 1064-nm excitation is desired. All FT-Raman instruments use this excitation wavelength plus an InGaAs or germanium detector (Figure 21). Because multiplexing is performed by the Fourier transform, InGaAs detectors used on FT-Raman spectrometers are not configured for linear array detection. However, dispersive instruments equipped with 1064-nm Nd:YAG lasers have been introduced recently, and these do use linear InGaAs array detectors.
F. Raman Microscopes
The coupling of a microscope to a Raman spectrometer is a natural extension of this instrument, which for normal operation requires a condensing lens to focus the excitation beam and an objective or mirror to collect Raman-scattered light (Figure 21). Raman microscopes collect backscattered light and a single lens serves as both the condenser and the objective. Because the laser beam is collimated, an infinity-corrected lens is used.
Common objectives used on Raman microscopes include 100x, 50x, and 10x, and for many systems, users can select from more than one objective. Changing an objective has several effects on both the excitation of samples and the efficiency with which Raman-scattered light is collected. The spot size of a focused laser beam is roughly its wavelength divided by the numerical aperture of the lens, so the maximum spatial resolution of a Raman microscope is achieved using the highest numerical aperture objective. However, along with decreasing the surface area of laser illumination this also results in a decreased sampling volume because with a higher numerical aperture, light from the edges of the beam enters the sample at more oblique angles. This alone would tend to make this a less sensitive analysis, but other factors offset this effect. The higher numerical aperture means much more scattered light can be collected because of the greater solid angle of the collection cone. In addition, the smaller image size (area of illumination) results in less scattered light being attenuated at the entrance slit of the Czerny-Turner spectrograph (Figure 22), which is also a focal plane of the optics. The higher flux of the excitation beam can also contribute to an increase in sensitivity, although at the same time, it can also be detrimental to some samples. Consequently, there is a tradeoff involved and users may need to experiment to determine which objective works best for a particular sample.
The ability to perform Raman spectroscopy depth profiling using a confocal microscope was first demonstrated by Turrell, Delhaye, and Dhamelincourt . In a confocal microscope, the scattered light collected by the objective is directed to a second infinity-corrected lens. Located at the focal point of this second lens is a small pinhole, known as the confocal hole. This pinhole serves to remove light rays that originated from sampling depths other than from the focal point of the collection lens, as illustrated in Figure 24. Light from planes located at these other depths form images on either side of the plane of the confocal hole, and the smaller this hole, the better is the depth spatial resolution. As discussed previously, only small amounts of Raman-scattered light from depths other than the selected one are evident in the confocal Raman microscope spectra of the bicomponent fiber (Figure 8).
Some Raman microscopes allow the generation of scattered-light images, also known as Raman mapping. Depending on the options available, users can generate an image based on one specific Raman peak, two or more peaks, or the entire spectrum. Rather than rastering the laser beam through the chosen area to be mapped, the stage of the microscope is moved according to preset user commands, including setting the dimensions of each pixel. Data from hundreds or even thousands of pixels can be collected and stored, and an image is created in false color representing the intensity of the chosen peak or peaks, or the whole spectrum. After completing a scan, Raman spectra from any chosen pixel can be retrieved if the user had set the collection up to include the entire spectrum. An example of a Raman spectral image taken from the surface of an Excedrin[R] tablet that contains aspirin, acetaminophen, and caffeine is shown in Figure 25, together with spectra taken from four pixels (marked with crosses). With the use of a confocal Raman microscope, a 3D image or map can also be also constructed.
The authors would like to thank: Professor Joseph W. Nibler (Oregon State University), Dr. Fran Adar (Horiba Scientific), Andrew Bowen (US Postal Inspection Service Laboratory), Dr. Ken Smith (Renishaw), Professor John Lombardi (John Jay College of Criminal Justice, CUNY), Dr. Mike Carrabba (Rogue Spectroscopy, LLC), Dr. David Tuschel (Horiba Scientific), and Dr. Chris Palenik (Microtrace LLC) for their very helpful discussions; Dr. Adar, Mr. Bowen, Dr. Sergey Mamedov (Horiba Scientific), Dr. Tuschel, Horiba Scientific, Thermo Fisher Scientific, BaySpec, Renishaw, Foster + Freeman, and Bruker Optics, Inc., for allowing them to reprint Raman data from some of their presentations; and Dr. Adar, Mr. Bowen, Jason Dunn (Washington State Crime Laboratory), Professor Lombardi, Dr. Sonja Peterson (Washington State Crime Laboratory), and Dr. Smith for taking the time to review this manuscript, or portions thereof.
[1.] Adar F: Depth resolution of the Raman microscope: Optical limitations and sample characteristics; Spectroscopy 25(3): 16; 2010.
[2.] Adar F: Fundamentals of a Raman microscope (Horiba Scientific webinar); Horiba Scientific: Edison, NJ; http://www.horiba.com/us/en/scientific/products/raman-spectroscopy/raman-academy/webinars/fundamentals-of-a-raman-microscope-webinar/ (Accessed June 12, 2018).
[3.] Adar F, Delhaye M, DaSilva E: Evolution of instrumentation for detection of the Raman effect as driven by available technologies and by developing applications; J Chem Edu 84:50; 2007.
[4.] Azo Optics: Raman Spectroscopy--Overview and Product Solutions for Raman Spectroscopy; Andor Technology: Belfast, UK; https://www.azooptics.com/Article.aspx?ArticleID=309 (Accessed June 12, 2018).
[5.] Bartholomew CR, McConnell AA, Smith WE: Resonance Raman excitation profile of cobalt phthalocyanine at room temperature and 10 K; J Raman Spectrosc 20:595; 1989.
[6.] Bartick EG: Infrared microscopy and its forensic applications; In Saferstein R (Ed): Forensic Science Handbook, Vol. 3, 2nd ed; Prentice Hall: Upper Saddle River, NJ; p 252; 2010.
[7.] Buzzini P, Massonnet G: A market study of green spray paints by Fourier transform infrared (FTIR) and Raman spectroscopy; Sci Justice 44:123; 2004.
[8.] Canadian Institute of Gemmology: Canadian Institute of Gemmology Newsletter May 2012; https://www.cigem.ca/news/gl-newsletter-may-2012 (Accessed June 11, 2018).
[9.] Coates J: New technologies for process analytical and quality control applications: Compact Raman; Spectroscopy 21(2):68; 2006.
[10.] Fleischmann M, Hendra PJ, McQuillan AJ: Raman spectra of pyridine adsorbed at a silver electrode; Chem Phys Lett 26:163; 1974.
[11.] Griffiths PR, De Haseth JA: Fourier Transform Infrared Spectroscopy, 2nd ed; Wiley InterScience: New York, NY; 2007.
[12.] Hendra P, Jones C, Warnes G: Fourier Transform Raman Spectroscopy; Ellis Horwood: New York, NY; p 93; 1991.
[13.] Hirschfeld T: Raman microprobe: Vibrational spectroscopy in the femtogram range; J Opt Soc Am 63:476; 1973.
[14.] Horiba Scientific: Raman spectroscopy for geological materials analysis (Application Note RA23); Horiba Scientific: Edison, NJ; 2012; http://www.horiba.com/fileadmin/uploads/Scientific/Documents/Raman/RA23-Raman_spectroscopy_for_geological_materials_analysis.pdf (Accessed June 12, 2018).
[15.] International Union of Pure and Applied Chemistry (IUPAC): Nomenclature, symbols, units, and their usage in spectrochemical analysis XVIII. Laser-based molecular spectrometry for chemical analysis--Raman scattering processes; Pure Appl Chem 69:145; 1997; https://www.degruyter.com/downloadpdf/j/pac.1997.69.issue-7/pac199769071451/pac199769071451.pdf (Accessed June 3, 2018).
[16.] LeRu EC, Etchegoin PG: Principles of Surface-Enhanced Raman Spectroscopy and Related Plamonic Effects; Elsevier: New York, NY; 2008.
[17.] Lombardi JR, Birke RL: A unified approach to surface-enhanced Raman spectroscopy; J Phys Chem 112:5605; 2008.
[18.] Lombardi JR, Birke RL: A unified view of surface-enhanced Raman scattering; Accounts Chem Research 42:734; 2009.
[19.] Muehlethaler C, Leona M, Lombardi JR: Review of surface enhanced Raman scattering applications in forensic science; Anal Chem 88:152; 2016.
[20.] Muehlethaler C, Leona M, Lombardi JR: Towards a validation of surface-enhanced Raman scattering (SERS) for use in forensic science: Repeatability and reproducibility experiments; Forensic Sci Int 268:1; 2016.
[21.] Pelletier MJ: Introduction to applied Raman spectroscopy; In Pelletier MJ (Ed): Analytical Application of Raman Spectroscopy; Blackwell Science: Malden, MA; pp 1-52; 1999.
[22.] Raman CV, Krishnan KS: A new type of secondary radiation; Nature 121:501; 1928.
[23.] Schlucker S (Ed): Surface Enhanced Raman Spectroscopy: Analytical, Biophysical and Life Science Applications; Wiley-VCH: New York, NY; 2010.
[24.] Semrock filters; Semrock: Rochester, NY; https://www.semrock.com/filters.aspx (Accessed June 12, 2018),
[25.] Smekal A: Zur Quantentheorie der Dispersion; Naturwiss 11:873; 1923.
[26.] Spiro TJ: Resonance Raman spectra of hemoproteins; Methods Enzymol 54:233; 1978.
[27.] Suzuki EM: Forensic applications of infrared spectroscopy; In Saferstein R (Ed): Forensic Science Handbook, Vol. 3, 2nd ed; Prentice Hall: Upper Saddle River, NJ; pp 75-251; 2010.
[28.] Suzuki EM: Infrared spectra of U.S. automobile original finishes (Post--1989). VIII: In situ identification of bismuth vanadate using extended range FT-IR spectroscopy, Raman spectroscopy, and X-ray fluorescence spectrometry; J Forensic Sci 59:1205; 2014.
[29.] Suzuki EM, Buzzini P: Applications of Raman spectroscopy in forensic science. II. Analysis considerations, spectral interpretation, and examination of evidence; Forensic Sci Rev; 30:155; 2018.
[30.] Suzuki EM, Carrabba M: In situ identification and analysis of automotive paint pigments using line segment excitation Raman spectroscopy: I. Inorganic topcoat pigments; J Forensic Sci 46:1053; 2001.
[31.] Suzuki EM, Nibler JW: Vibrational spectra of cis-3,4-dichlorocyclobutene; Spectrochim Acta Part A 30:15; 1974.
[32.] Thermo Fisher Scientific: Raman spectroscopy, unveiling the secrets of limestone (Application Note 52569); Thermo Fisher Scientific: Madison, WI; 2014; https://tools.thermofisher.com/content/sfs/brochures/CAD-Molecular-Raman-LimestoneCarbonates.pdf (accessed June 12, 2018).
[33.] Turrell G, Delhaye, Dhamelincourt: Characteristics of Raman microscopes; In Turrel G, Corset J (Eds): Raman Microscopy: Developments and Applications; Academic Press: London, UK; pp 27-50; 1996.
[34.] Tuschel D: Selecting an excitation wavelength for Raman spectroscopy; In Molecular Spectroscopy Workbench: The 2006 Collection; Spectroscopy: Iselin, NJ; pp 11-17; December 2016; http://images2.advanstar.com/PixelMags/spectroscopy/digitaledition/12-2016-sp.html#10 (Accessed June 20, 2018).
[35.] Weimer R, Clary J, Heintz R, Wall M: Analysis of bicomponent fibers using confocal Raman mapping; JASTEE 3:21; 2012.
E.M. Suzuki (*)
Washington State Crime Laboratory Washington State Patrol Seattle, Washington United States of America
Department of Forensic Science Sam Houston State University Huntsville, Texas United States of America
(*) Corresponding author: Dr. Edward M. Suzuki, Washington State Crime Laboratory, Washington State Patrol, Seattle, WA 98134; + 1 206 262 6020 (voice); firstname.lastname@example.org.
ABOUT THE AUTHORS
E. M. Suzuki; P. Buzzini
Edward M. Suzuki received his B.S. degree in chemistry from the University of Washington (Seattle, WA) in 1970 and his Ph.D. in chemistry (physical) from Oregon State University (Corvallis, OR) in 1975. Dr. Suzuki's doctoral dissertation involved the characterization of highly reactive chemical species trapped in low-temperature argon matrices using various spectroscopic methods, including infrared, Raman, and electron paramagnetic resonance.
Dr. Suzuki is currently a supervising forensic scientist at the Washington State Crime Laboratory (Seattle, WA). He has worked for over 38 years in the field of forensic science and has testified in over 750 criminal cases. His main research interests include applications of infrared, Raman, and X-ray fluorescence spectroscopies for the analysis of various types of evidence, and particularly, for the identification of pigments in automotive finishes. He has published over 30 research papers, primarily in the area of vibrational spectroscopy.
Dr. Suzuki has helped teach classes on forensic applications of infrared spectroscopy for the FBI Academy (Quantico, VA), IR Courses Inc. (Bowdoin College: Brunswick, ME), Eastern Washington University (Cheney, WA), the California Criminalistics Institute (Sacramento, CA), Microtrace LLC (Elgin, IL), and public forensic science laboratories in New Hampshire, Illinois, California, and Singapore. He is a fellow of the American Academy of Forensic Sciences; a member of the American Chemical Society, the Society for Applied Spectroscopy, the Coblentz Society, the American Society of Trace Evidence Examiners, and the Northwest Association of Forensic Scientists; and is certified as a fellow by the American Board of Criminalistics.
Patrick Buzzini graduated from the Institut de Police Scientifique of the School of Criminal Sciences with the University of Lausanne (Lausanne, Switzerland). In 2007, he obtained a doctoral degree in forensic science from the same institution on the application of Raman spectroscopy to criminalistics and particularly to the discrimination of dyed fibers. Dr. Buzzini is an associate professor in forensic science with the Department of Forensic Science at Sam Houston State University (Huntsville, TX).
Dr. Buzzini has more than 15 years of experience as an instructor, researcher, and caseworker in criminalistics, with emphasis in trace evidence. His research interests include the forensic applications of microscopic and spectroscopic methods (i.e., Raman spectroscopy, infrared spectroscopy, and microspectrophotometry) to various types of trace evidence and questioned documents as well as addressing problems of physical evidence interpretation. He has published over a dozen papers in the field of forensic science.
Dr. Buzzini has provided training nationally and internationally to practitioners in the field and to the legal community in the areas of trace evidence analysis and interpretation. He is a fellow of the American Academy of Forensic Sciences, a member of the Organization of Scientific Area Committees (OSAC) on Chemistry and Instrumental Analysis, a member of the American Society of Trace Evidence Examiners, and an associate member of the International Association for Identification.
REFERENCE: Suzuki EM, Buzzini P: Applications of Raman spectroscopy in forensic science. I: Principles, comparison to infrared spectroscopy, and instrumentation; Forensic Sci Rev 30:111; 2018.
|Printer friendly Cite/link Email Feedback|
|Author:||Suzuki, E.M.; Buzzini, P.|
|Publication:||Forensic Science Review|
|Article Type:||Technical report|
|Date:||Jul 1, 2018|
|Previous Article:||Professional Review and Commentary.|
|Next Article:||Applications of Raman Spectroscopy in Forensic Science. II: Analysis Considerations, Spectral Interpretation, and Examination of Evidence.|